One-step nanoscale expansion microscopy

Abstract

The present invention relates to a microscopy method for obtaining a high-resolution image of a biological sample comprising a substance to be analyzed, wherein an image sequence is obtained by expansion microscopy comprising the steps: g. providing a biological sample, h. mixing said biological sample with a swellable material and forming bonds, preferably covalent bonds, between said substance to be analyzed and the swellable material, thereby forming a mixture, i. fragmenting said substance to be analyzed, j- expanding the mixture comprising said swellable material and said fragmented substance to be analyzed, k. staining the sample either before or after steps (a), (b), (c), or (d) and imaging the obtained mixture under an optical microscope equipped with a camera or a resonant scanner. It also refers to compounds and stabilization chambers for use in such a method.

Description

[0001] One-step nanoscale expansion microscopy

[0002] The present invention relates to a microscopy method for visualizing biological mole- cules such as proteins by using expansion microscopy.

[0003] Technical background

[0004] Fluorescence imaging is one of the most versatile and widely-used tools in biology1. Although techniques to overcome the diffraction barrier were introduced more than two decades ago, and the nominal attainable resolution kept improving, fluorescence microscopy still fails to image the morphology of single proteins or small molecular complexes, either purified or in a cellular context.

[0005] Optical microscopy has been one of the most valuable tools in biology for more than two centuries, and has been considerably enhanced by the introduction of super-res- olution microscopy, two decades ago. Nevertheless, optical imaging remains difficult to perform below 10-20 nm. Several recent works have presented localization preci- sions down to 1-2 nm, or even below, but the application of such imaging resolution to biological samples has been severely limited by two fundamental problems. First, the achievable structural resolution is determined by the labeling density, which is limited by the size of the fluorescent probes (typically 1 nanometer or larger). Sec- ond, fluorophores can interact via energy transfer at distances below 10 nm, which results in accelerated photoswitching (blinking) and photobleaching, and thus in sub- stantially lower localization probabilities.

[0006] The solution to these two problems would be to separate the fluorophores spatially by the physical expansion of the specimen, in what is termed expansion microscopy (ExM). To then reach the molecular scale, one would combine ExM with optics-based super-resolution. This has been attempted numerous times, but the resulting perfor- mance typically reached only ~10 nm. The ExM gels are dim because the fluoro- phores are diluted by the third power of the expansion factor, thus limiting optics techniques that prefer bright samples, as stimulated emission depletion (STED), or saturated structured illumination (SIM). For example, tomographic STED microscopy. In addition, the ExM gels need to be imaged in distilled water, since the ions in buff- ered solutions shield the charged moieties of the gels and diminish the expansion factor. The use of distilled water reduces the performance of techniques that rely on special buffers, as single molecule localization microscopy, SMLM (Extended Data Fig. 1).

[0007] A third class of optical super-resolution approaches is based on determining the higher-order statistical analysis of temporal fluctuations measured in a movie, e.g. super-resolution optical fluctuation imaging (SOFI) or super-resolution radial fluctua- tions (SRRF). The resolution of these approaches is inversely correlated to the dis- tance between the fluorophores and they do not require especially bright samples or special buffers, implying that they should benefit from ExM.

[0008] Summary of the invention

[0009] Surprisingly, the inventors found that one-step nanoscale expansion (ONE) micros- copy can provide a solution to the above problems by combining the 10-fold axial ex- pansion of the specimen (1000-fold by volume) with a fluorescence fluctuation analy- sis to enable the description of cultured cells, tissues, viral particles, molecular com- plexes and single proteins. At the cellular level, using immunostaining, the present technology revealed detailed nanoscale arrangements of synaptic proteins, including a quasi-regular organisation of PSD95 clusters. At the single molecule level, upon main chain fluorescent labelling, the shape of individual membrane and soluble pro- teins can be visualized as shown in the examples. Moreover, conformational changes undergone by the ~17 kDa protein calmodulin upon Ca2+binding were readily ob- servable. Also imaged and classified were molecular aggregates in cerebrospinal fluid samples from Parkinson's Disease (PD) patients, which represents a promising new development towards improved PD diagnosis. ONE microscopy is compatible with conventional microscopes and can be performed with the software of the present in- vention. This technology bridges the gap between high-resolution structural biology techniques and light microscopy, and provides a new avenue for discoveries in biol- ogy and medicine.

[0010] To test this hypothesis, X10 expansion microscopy with SRRF are combined and thereby establish a technique (one-step nanoscale expansion (ONE) microscopy (Ex- tended Data Fig. 1)). ONE was implemented using conventional confocal or epifluo- rescence microscopes and reached an imaging performance that enables imaging in- dividual protein shapes. To aid in its implementation, we generated a ONE software platform, as a plug-in for the ImageJ (Fiji) (Supplementary Fig. 1; Supplementary Software). A general shortcoming of this technique, in comparison to more estab- lished procedures as oSTORM-ExM, is that its axial resolution is equivalent to the res- olution of the confocal microscope, divided by the expansion factor, and is thus sub- stantially poorer than the XY-plane resolution.

[0011] The invention is therefore directed to:

[0012] A microscopy method for obtaining a high-resolution image of a biological sample comprising a substance to be analyzed, the method comprising: a. obtaining an image sequence of said sample by expansion microscopy, and b. optionally, analyzing said image sequence by super-resolution radial fluctua- tions (SRRF), wherein the image sequence comprises 1000 frames or more and wherein the frames are ac- quired using an objective having a numerical aperture of 1.3 or higher.

[0013] Further to a compound having formula (I)

[0014] (I), wherein n is an integer of from 1 to 4, and wherein

[0015] A is a selected from the group consisting of a double-bond containing group and a fluorophore-containing moiety (such as a fluorophore attached via a linker).

[0016] Further to a mixture comprising a substance to be analyzed, e.g. a protein, contain- ing at least one anchor group derived from the above anchoring agent and a swella- ble material comprising a swellable agent.

[0017] Further to a stabilization chamber for a gel, the stabilization chamber comprising a. a chamber holder b. a chamber cover c. a cage for holding a composition of a biological sample embedded in a swella- ble material, the cage comprising a cage patterning; characterized in that the cage patterning comprises polygonal, preferably pyramid and / or trapezoid shaped beams. Further to a method of determining the 3D structure of a biological molecule, prefer- ably a protein, the method comprising: a. providing a purified solution comprising the biological molecule, preferably the protein b. anchoring the biological molecule, preferably the protein, to a swellable agent of a swellable material c. optionally semi-drying the biological sample / protein-anchor agent on a glass coverslip until a thin film of solution remains on the glass coverslip d. homogenization of the sample by cutting peptides, preferably with proteinase K e. expanding the sample f. optionally labelling the expanded sample, preferably with an NHS-Ester with a first fluorophore g. optionally labelling of at least one specific, preferably more than two, most preferably 3-10 specific amino acids with fluorophores allowing discrimination h. performing an imaging procedure according to item 1, 12 or 14 protocol i. performing a super-resolution radial fluctuations (SRRF) analysis j. optionally creating a 2D model (2D shape recognition) or 3D model of labelled peptides using positions of the first fluorophore (of the fluorescence signals) to re- trieve the overall structure of a protein k. optionally correlating the 3D model of the labelled peptides with the positions of the specific amino acids optionally analyzing the positions of the specifically labelled amino acids to retrieve a unique sequence of those amino acids giving the fingerprint of a protein.

[0018] Further to a method of capturing a plurality of images of a biological sample placed on an optical microscope and using an image sensor in the optical path of the micro- scope, the method comprising: setting an imaging specification including an image size and a theoretical pixel size; setting a scanner specification including a number of frames taken from the biological sample and a scan frequency at which the frames are scanned, wherein the number of frames is at least 300, preferably at least 500, more preferably more than 1000 and most preferably more than 1400 frames; and operating the image sensor according to the scanner specification, wherein the method preferably further comprises performing a super-resolution radial fluctuations analysis, SRRF analysis, on each of the number of frames. Detailed description of the invention

[0019] The present invention refers to the embodiments according to the following items.

[0020] The following embodiments of the present invention can be combined with one an- other without departing from the scope of the invention. All documents cited herein are incorporated by reference.

[0021] 1. A microscopy method for obtaining a high-resolution image of a biological sample comprising a substance to be analyzed, the method comprising: a. obtaining an image sequence of said sample by expansion microscopy, and b. optionally, analyzing said image sequence by super-resolution radial fluctuations (SRRF), wherein the image sequence comprises 1000 frames or more and wherein the frames are acquired using an objective having a numerical aperture of 1.3 or higher.

[0022] 2. Microscopy method for obtaining a high-resolution image of a biological sam- ple comprising a substance to be analyzed, preferably a microscopy method according to item 1, wherein an image sequence is obtained by expansion microscopy comprising the steps: a. providing a biological sample, b. mixing said biological sample with a swellable material and forming bonds, preferably covalent bonds, between said substance to be ana- lyzed and the swellable material, thereby forming a mixture, c. fragmenting said substance to be analyzed, d. expanding the mixture comprising said swellable material and said frag- mented substance to be analyzed, e. staining the sample either before or after steps (a), (b), (c), or (d) and f. imaging the obtained mixture under an optical microscope equipped with a camera or a resonant scanner.

[0023] 3. The microscopy method according to item 1 or 2, wherein imaging is per- formed using epifluorescence, preferably wide-field epifluorescence, or con- focal imaging, wherein the numerical aperture of the objective is 1.3 or higher, preferably of from 1.4 to 1.51, further preferably 1.4 or 1.51. The microscopy method according to any one of items 1 to 3, wherein ana- lyzing the image sequence is performed by temporal radiality auto-correla- tions (TRAC), preferably for brightly labeled samples with direct labeling, and / or by temporal radiality pairwise product mean (TRRPM), preferably for dim samples The microscopy method according to any one of items 1 to 4, wherein ana- lyzing the image sequence is performed, using radiality magnification of 12 to 60, preferably 15 to 45, further preferably 20 to 40, preferably at 35. The microscopy method according to any one of items 2 to 5, wherein the substance to be analyzed is a protein and wherein step (a) comprises conju- gating proteins within the sample with a bifunctional crosslinker, wherein the bifunctional crosslinker comprises a protein -reactive chemical group and a gel-reactive chemical group. The microscopy method according to any one of items 1 to 6, wherein the substance to be analyzed is a protein and wherein step (a) comprises conju- gating proteins within the sample with more than one bifunctional cross- linker, wherein the bifunctional crosslinker comprises a protein-reactive chemical group and a gel-reactive chemical group. The microscopy method according to any one of items 2 to 7, wherein step (b) comprises embedding the sample in a swellable material wherein the substance to be analyzed within the sample, e.g. a protein, is anchored to the swellable material, preferably via an anchoring group optionally derived from an anchoring agent. The microscopy method according to any one of items 2 to 8, wherein the substance to be analyzed is a protein and wherein step (c) comprises sub- jecting the sample to digestion. The microscopy method according to any one of items 2 to 9, wherein ex- panding the mixture comprises swelling the swellable material to form an ex- panded sample. The microscopy method according to any one of items 2 to 10, wherein the mixture is a composite comprising a gel and the substance to be analyzed. The microscopy method according to any one of items 1 to 11, wherein sig- nal fluctuations are measured by imaging the mixture repeatedly, using the resonant scanner, preferably using a frequency of 4 to 30 kHz, preferably 8 to 24 kHz, preferably imaging the sample of from 1000 to 100,000 times, preferably 1150 to 50,000 times, further preferably 1250 to 10,000 times, further pref- erably 1350 to 5000 times, further preferably 1400 to 3000 times or wherein fluorescence is measured by imaging the mixture repeatedly, using an optical camera, preferably imaging the sample of from 1000 to 100,000 times, preferably 1150 to 50,000 times, further preferably 1250 to 10,000 times, further preferably 1350 to 5000 times, further preferably 1400 to 3000 times. The microscopy method according to any one of items 2 to 12, wherein the method further comprises placing the expanded mixture in a stabilization chamber, preferably wherein the expanded mixture is immobilized in a stabi- lization chamber, the stabilization chamber containing multiple polygonal structures. The microscopy method according to any one of the preceding items, wherein the acquisition of fluctuations is restricted by using stimulated emis- sion depletion (STED), preferably by using 3 dimensional (3D) STED. The microscopy method according to item 14, wherein STED is provided by a 2D doughnut laser in the x,y plane, e.g. a x,y-doughnut laser, and / or wherein 3D STED is provided by a 3D-doughnut laser, e.g. a x,y-doughnut laser in combination with a z-doughnut laser. The microscopy method according to item 14 or 15, wherein the STED or 3D STED laser comprises, radiation generating means capable of emitting a first and a second beam, the first beam being an excitation beam, the second beam being a depletion beam relative to the first beam, an optical element for focusing the first and the second beam on the object, the optical element being arranged relative to the radiation generating means for defining a common optical path (OP) for both the first and the second beam, and a phase modifying member inserted in said common optical path (OP), wherein the phase modifying member is optically arranged for leaving the wavefront of the first beam substantially unchanged, and for changing the wavefront of the second beam so as to create an undepleted region of inter- est (ROI) in the object. The microscopy method according to any one of items 2 to 16, wherein step (d) provides an expansion factor of about 2 to 30, preferably about 4 to 24, further preferably about 10 to 20 of the sample in each dimension. The microscopy method according to any one of items 2 to 17, wherein the swellable material is a gel, preferably a hydrogel, comprising N,N-dime- thylacrylamide and / or sodium acrylate, preferably N,N-dimethylacrylamide. The microscopy method according to any one of items 1 to 18, wherein ana- lyzing the image sequence by SRRF comprises dividing each pixel into a plu- rality of subpixels. The microscopy method according to item 19, wherein analyzing the image sequence comprises measuring local radial symmetries in the sequence of images. The microscopy method according to item 19 or 20, wherein measuring local radial symmetries comprises, measuring for every subpixel intensity gradient vectors for a ring of surrounding subpixels to determine a gradient of conver- gence. The microscopy method according to item 21, wherein the image sequence is processed based on the analysis of the gradient of convergence of sub-pix- els over a radiality stack. The microscopy method according to item 22, wherein analyzing the image sequence further comprises summing of from 10 to 100,000, preferably 10 to 50,000, further preferably 20 to 10,000, further preferably 100 to 5000, fur- ther preferably 500 to 2000 measured images, optionally background sub- tracted images. The microscopy method according to item 23, wherein analyzing the image sequence further comprises a temporal analysis of fluctuating fluorophores attached, preferably covalently bound to the substance to be analyzed, e.g. a protein, based on radiality. The microscopy method according to any preceding item, wherein the pro- tein is attached to the swellable agent via a serine, further optionally via an anchoring group bound to the serine. The microscopy method according to any preceding item, wherein the pro- tein is attached to the swellable agent via at least one type of anchoring group, preferably at least two types of anchoring groups, wherein a first type is attached to a serine and a second type is attached to another amino acid of the protein, preferably a lysine, a cysteine, a threonine, a phenylalanine, a histidine, or glutamine. The microscopy method according to any preceding item, using wide-field epifluorescence or confocal microscopy for obtaining an image sequence. The microscopy method according to any preceding item, wherein the method achieves an effective maximum resolution of about 0.1 nm. The microscopy method according to any preceding item, wherein the mix- ture expands essentially isotropical in at least 2 dimensions, preferably in 3 dimensions. The microscopy method according to any preceding item, wherein local radial symmetries are analyzed via higher order temporal statistics. The microscopy method according to any preceding item, wherein the pro- tein is attached to the swellable agent by use of a compound or anchoring group derived from an anchoring agent according to any one of items 36, 37, 39, 40, such as by attaching said protein with said compound via a serine. The microscopy method according to item 31, wherein the anchoring group is attached to the protein as shown by general formula (III): The microscopy method according to any preceding item, wherein the pro- tein is attached via a lysin to the swellable material via 6-((acry- loyl)amino)hexanoic acid succinimidyl ester. The microscopy method according to any preceding item, wherein a fluoro- phore is attached to the protein via a serine, e.g. by use of a compound ac- cording to any one of items 36, 38, 39, 40, and / or via an amino acid com- prising an amino terminal group, e.g. by use of an NHS-ester fluorescein. The microscopy method according to item 34, wherein the fluorophore is at- tached to the protein as shown by general formula (IV): n is an integer of from 1 to 4, and wherein

[0024] A is a selected from the group consisting of a double-bond containing group and a fluorophore-containing moiety (such as a fluorophore at- tached via a linker). Compound according to item 36, wherein A has the formula (II) Compound according to item 37, wherein A is selected from the group con- sisting of fluorescein, xanthene derivatives, cyanine derivatives, squaraine derivaties and ring-substituted squaraines, squaraine rotaxane derivetives, naphthalene derivatives, coumarine derivatives, oxadiazole derivatives, an- thracene derivatives, pyrene derivatives, oxazine derivatives, acridine deriva- tives, arylemthine derivatives, tetrapyrrole derivatives and dipyrromethene derivatives, preferably fluorescein. Compound according to item 36, having the structure of general formula (V) o Anchoring agent for interacting with a biological sample, preferably a protein, wherein a. the anchoring agent is compatible to a swellable material, preferably, wherein the anchoring agent covalently binds to said swellable mate- rial, b. the anchoring agent is suitable to be chemically, preferably covalently anchored to a functionalized or non-functionalized serine group of the biological sample, characterized in that the anchoring agent comprises a heterocycle comprising P, S, and 0, and a functional group selected from the group consisting of a double-bond containing group and a fluorophore.

[0025] Mixture comprising a substance to be analyzed, e.g. a protein, containing at least one anchor group derived from the anchoring agent according to item 40 and a swellable material comprising a swellable agent. Mixture according to item 41, wherein the mixture comprises more than one type of anchor group derived from an anchoring agent. Mixture according to item 41 or 42, wherein the mixture comprises two dif- ferent types of anchor groups, wherein a first type of anchor group binds to a serine residue of the substance to be analyzed and a second type of an- chor group binds to a residue from a different amino acid of the substance, preferably to a lysine residue. Mixture according to item 43, wherein the at least one anchor at a serine residue of the biological sample and the least one anchor at a lysine residue of the biological sample are bound to the same protein of the biological sam- ple. Mixture according to item 43 or 44, wherein the at least one anchor at a ser- ine residue is covalently bound to the hydroxy group of serine. Mixture according to any one of items 43 to 45, wherein the amine group of serine is not modified by the anchoring agent. Mixture according to any one of items 43 to 46, wherein the swellable agent is in the form of a gel, preferably a hydrogel. Mixture according to any one of items 41 to 47, for use in a microscopy method according to item 1. Stabilization chamber for a gel, the stabilization chamber comprising a. a chamber holder b. a chamber cover c. a cage for holding a composition of a biological sample embedded in a swellable material, the cage comprising a cage patterning; character- ized in that the cage patterning comprises polygonal, preferably pyra- mid and / or trapezoid shaped beams. Stabilization chamber according to item 49, comprising polygonal shaped col- umns, wherein the ratio of the width of the base of the beams to the height of the beams is from 2: 1 to 1:2. Stabilization chamber according to item 49 or 50, comprising polygonal shaped beams, wherein the height of the polygons is 0.25 to 2 mm, prefera- bly 0.3 to 1 mm, further preferably about 0.5 mm. Stabilization chamber according to any one of items 49 to 51, comprising po- lygonal shaped beams, wherein the polygon width is between 0.2 to 1 mm, 0.3 to 0.8 mm, preferably, 0.4 to 0.6 further preferably about 0.5 mm. Stabilization chamber according to any one of items 49 to 52, comprising trap- ezoid shaped beams, wherein the angle formed by the base of the polygonal and the side of the polygonal is 85° to 30°, preferably 75 to 50, further pref- erably about 70°, and the angle formed by the top of the polygonal and the side of the polygonal is 95 to 150°, preferably 105 to 130°, further preferably about 110°. Stabilization chamber according to any one of items 49 to 53, wherein the cage patterning forms imaging zones having circular, oval, rectangular or square shape, preferably rectangular or square shape. Stabilization chamber according to any one of items 49 to 54, wherein the shape of the imaging zones has a diameter of from 3 mm to 30 mm, prefera- bly from 4 to 20 mm, further preferably from 5 to 12 mm, further preferably about 7 mm. Stabilization chamber according to any one of items 49 to 55, wherein the cage patterning comprises a mesh formed by the beams, the beans of the mesh having a minimum distance of from 1 to 3 mm, preferably from 1,5 to 2,5 mm, further preferably about 2 mm. Stabilization chamber according to item 56, wherein at least 40%, at least 50%, preferably at least 60% or at least 70% of the cage patterning is a mesh. Stabilization chamber according to any one of items 54 to 57, wherein at most 60%, or at most 50%, preferably at most 40% or at most 30% of the cage patterning are imaging zones. Stabilization chamber according to any one of items 49 to 57, wherein the cage patterning comprises voids allowing the composition to contact a co- verslip provided below the cage patterning. Stabilization chamber according to any one of items 49 to 58, wherein the cage is 3D printed. Stabilization chamber according to any one of items 49 to 59, wherein the cage preferably consists of a Mg-AI alloy. Stabilization chamber according to any one of items 59 to 61, wherein the co- verslip is made of glass. Stabilization chamber according to any one of items 49 to 62, wherein the cage comprises a central imaging zone and a plurality of, preferably 4 to 8, further preferably 6, flanking imaging zones. Stabilization chamber according to any one of items 49 to 63, wherein the cage does not comprise needles or circular boundaries for holding the compo- sition. Stabilization chamber according to any one of items 49 to 64, wherein the chamber is sealed, preferably by glycerol. Stabilization chamber according to any one of items 49 to 65, wherein the chamber comprises at least one net groove. Stabilization chamber according to any one of items 49 to 66, wherein the chamber the chamber limits drift of the composition to 0.075 pm / s, preferably 0.01 pm / s, more preferably 0.001 pm / s, further preferably 0.0005 pm / s. Stabilization chamber according to any one of items 49 to 67, for use in a mi- croscopy method according to item 1. A method for providing an expanded biological sample as specimen for mi- croscopy, the method comprising a. attaching at least one anchor group derived from the anchoring agent according to item 40 to the biological sample, b. embedding the biological sample in the swellable material, c. fragmenting the biological sample embedded in the swellable material, and d. swelling the swellable material; characterized in that the at least one anchor group is capable of anchoring to at least one amino acid group in the biological sample. The method according to item 69, wherein swelling results in enlarging by at least 5x in each dimension, preferably at least lOx in each dimension, most preferably between lOx and 20x in each dimension. The method according to item 69 or 70, wherein swelling is performed by washing with double distilled water, for at least 2, preferably at least 4, fur- ther preferably at least 6 hours. The method according to item 71, wherein the washing comprises at least 2, preferably at least 3, further preferably at least five solution changes. The method according to any one of items 69 to 72, wherein step (c) is per- formed by digestion with Proteinase K. The method according to any one of items 69 to 73, wherein the biological sample is labelled after step (c) using at least 50x or lOOx excess, preferably 200x to 300x excess of NHS-ester fluorescein (#46409, ThermoFisher Scien- tific) in NaCHCh buffer after step (c). The method according to any one of items 69 to 74, wherein step (a) com- prises: a. adding a first anchoring agent, preferably 6-((acryloyl)amino)hexanoic acid succinimidyl ester, to the sample, b. incubating the sample for 0.5 to 5 hours, preferably about 2 hours, and subsequently c. adding a second anchoring agent, preferably having formula (I), to the sample, d. incubating the sample at 4 °C for 6 to 48 hours, preferably 8 to 24 hours, further preferably 10 to 16 hours. The method according to any one of items 69 to 75, wherein step (c) of the method comprises: a. heating the composite to temperature of from 30 °C to 130 °C for a du- ration of 5 hours or more b. optionally treating the composite with an enzyme capable of fragment- ing the biological sample, preferably proteinase K; further optionally simultaneously treating the composite with an enzyme capable of frag- menting the biological sample, preferably proteinase K. The method according to any one of items 69 to 76, wherein the biological sample is fragmented by incubation at 30-80°C, preferably 40-60°C, most preferably between 45 and 55°C or at 50°C. The method according to any one of items 69 to 77, wherein the biological sample is fragmented by incubation with proteinase K for at least 10 hours, or at least 14 hours, or at least 18 hours, or at least 20 hours , or at least 24 hours, preferably between 12 and 30 hours, most preferably between 18 and 24 hours. The method according to any one of items 69 to 78, wherein a proteinase having an activity of 1 to 30 U / mL, preferably 4 to 20 U / mL, further prefera- bly 6 to 10 U / mL is used in step (c). The method according to any one of items 69 to 79, wherein the proteins in the expanded sample are digested in a way, that more than 5%, preferably more than 10%, or 15%, most preferably more than 20% of the peptide bonds are broken before expansion of the sample. The method according to any one of items 69 to 80 which is part of the mi- croscopy method according to item 1. A method of determining the 3D structure of a biological molecule, preferably a protein, the method comprising: a. providing a purified solution comprising the biological molecule, prefer- ably the protein b. anchoring the biological molecule, preferably the protein, to a swellable agent of a swellable material c. optionally semi-drying the biological sample / protein-anchor agent on a glass coverslip until a thin film of solution remains on the glass co- verslip d. homogenization of the sample by cutting peptides, preferably with pro- teinase K e. expanding the sample f. optionally labeling the expanded sample, preferably with an NHS-Ester with a first fluorophore g. optionally labeling of at least one specific, preferably more than two, most preferably 3-10 specific amino acids with fluorophores allowing discrimination h. performing an imaging procedure according to item 1, 12 or 14 proto- col i. performing a super-resolution radial fluctuations (SRRF) analysis j. optionally creating a 3D model of labeled peptides using positions of the first fluorophore to retrieve the overall structure of a protein k. optionally correlating the 3D model of the labeled peptides with the po- sitions of the specific amino acids l. optionally analyzing the positions of the specifically labeled amino acids to retrieve a unique sequence of those amino acids giving the finger- print of a protein. The method of item 82 wherein step (d.) is performed by incubation with proteinase K at 30-80°C for at least 10 hours, or at least 14 hours, or at least 18 hours, or at least 20 hours, or at least 24 hours, preferably between 12 and 48 hours or 14 to 30 hours, most preferably between 18 and 24 hours. The method of item 82 or item 83, wherein step (f.) is performed by using at least 50x, or lOOx, or 150x, or 200x, or 250x, or 300x excess, preferably 200x to 300x excess of NHS-ester. The method of any one of items 82 to 84 wherein g. is performed as de- scribed in any of the items 93-100 (items below describing the image analy- sis). A method of capturing a plurality of images of a biological sample placed on an optical microscope and using an image sensor in the optical path of the microscope, the method comprising:

[0026] - setting an imaging specification including an image size and a theoreti- cal pixel size;

[0027] - setting a scanner specification including a number of frames taken from the biological sample and a scan frequency at which the frames are scanned, wherein the number of frames is at least 300, preferably at least 500, more preferably more than 1000 and most preferably more than 1400 frames; and

[0028] - operating the image sensor according to the scanner specification, wherein the method preferably further comprises

[0029] - performing a super-resolution radial fluctuations analysis, SRRF analy- sis, on each of the number of frames. The method of item 86, further comprising:

[0030] - equipping the microscope with an objective having a numerical aper- ture between 1.3 and 1.6, preferably 1.4, or 1.45 or 1.51; and / or

[0031] - providing the biological sample as a specimen for microscopy expanded according to the method of one of items 69 to 81; and / or

[0032] - labelling the biological sample with a compound according to any one of items 36, 38, 39, 40; and / or

[0033] - lighting the biological sample with a laser light, preferably light of a doughnut laser, preferably a 3D-STED doughnut laser, preferably a first doughnut laser in horizontal orientation and a second doughnut laser in a vertical orientation. The method of item 86 or 87, wherein: the images size is set between 64x64 pixels and 256x256 pixels, prefer- ably at 128x128 pixels; and / or the theoretical pixel size is set below 100 nm, preferably between 20 and 100 nm, preferably between 24 nm and 80, more preferably between 30 and 50 nm, most preferably at 48 for a high SNR and 98 for a low SNR; and / or the scan frequency is set between 2 kHz and 30 kHz, preferably be- tween2 kHz and 24 kHz, preferably the scan frequency is set to 2 kHz, or 8 kHz or 16 kHz or 24 kHz; and / or the scan frequency is set to achieve a scan duration between 5 ms and 200 ms, preferably between 12 and 40 ms, preferably between 20 to 40 ms, or 25 ms. The method of item 88, wherein the scan frequency is set to 2 kHz for two- dimensional images over time, is set to 8 kHz for two-dimensional images in- cluding multiple channels or colors over time, is set to 12 kHz for three-di- mensional images over time or two-dimensional image stacks over time, and is set to 16 to 24 kHz for three-dimensional images including multiple chan- nels or colors over time or two-dimensional image stacks including multiple channels or colors over time. The method of one of items 19 to 22, further comprising:

[0034] - placing one or more beads on the specimen stage of the microscope, wherein the one or more beads fluoresce in multiple colors;

[0035] - capturing one or more images of the beads for different colors or differ- ent wavelengths;

[0036] - determining a relative location and / or displacement of the one or more beads for each color from the captured one or more images; and

[0037] - storing chromatic aberration correction values for each color, to allow chromatic aberration correction of the scans of the biological sample. The method of one of items 86 to 90, wherein the image sensor is one of a resonant scanner, a digital camera, and a light sensitive chip, preferably a complementary metal-oxide-semiconductor, CMOS, or electron-multiplying charge-coupled device, EMCCD, and wherein, preferably, the image sensor has a pixel size in the range of 120 nm to 200 nm. A method for generating a high resolution image of a biological sample from an image stack, wherein the image stack comprises between 300 and 10.000, preferably between 500 and 10.000, more preferably between 1000 and 5000, further preferably between 1200 and 4000, more preferably be- tween 1400 and 2500 images or frames, such as at least 300 , preferably at least 500, more preferably at least 1000 and further preferably at least 1400 images or frames or most preferably 2000 images or frames the method comprising:

[0038] - performing a super-resolution radial fluctuations analysis, SRRF analy- sis, on each image of the image stack. The method of item 92, further comprising:

[0039] - performing an automatic drift correction before performing SRRF analy- sis. The method of item 93, further comprising:

[0040] - gathering between 5 to 70, or 6 to 50, or 7 to 30, preferably 8 to 20, or

[0041] 9 to 15, or 10 to 12 images or frames from the image stack, wherein performing an automatic drift correction is performed on the basis of the gathered images or frames to test a drift behavior of the sample. The method of item 93 or 94, wherein performing automatic drift correction comprises:

[0042] - if the image stack contains images of multiple channels or colors, split- ting the image stack into one image stack per channel or color;

[0043] - determine cross-correlation of each channel separately; and

[0044] - perform drift correction by:

[0045] - summing between 2 and 70, or 5 to 50, or 8 to 20, most preferably 10 to 15 images or frames, or 0,5% to 5%, preferably 1%, of the images or frames of the image stack;

[0046] - detecting the center of bright sample structures from any channel or color; and

[0047] - if movement of a bright sample structure is detected in the summed images or frames, calculating a drift

[0048] - using he drift for all channels of the stack. The method of one of items 92 to 95, wherein performing SRRF analysis comprises setting parameters for the SRRF analysis, including:

[0049] - setting a radiality magnification between 12 to 60, preferably 15 to 45, further preferably 20 to 40, preferably at 35; and / or

[0050] - setting a number of ring axes between 2 and 8, preferably at 8; and / or

[0051] - setting a temporal analysis mode for Temporal Radiality Auto-Correla- tion to an order of 2 to 30, or 4 to 20, preferably to an order of 4 to 12, most preferably 8; and / or

[0052] - activate integration of temporal correlations; and / or

[0053] - activate removal of positivity constraints; and / or

[0054] - activate gradient smoothing; and / or

[0055] - activate minimization of SRRF pattering; and / or - deriving the setting parameters for the SRRF analysis from a scanner specification included in metadata of the image stack. The method of one of items 92 to 96, further comprising:

[0056] - forming a resolved image from results of the SRRF analysis; and / or

[0057] - repeating the SRRF analysis for each channel or color, wherein forming the resolved image includes combining hyper-stack images of the SRRF analysis of each channel or focal plane. The method of item 97, further comprising:

[0058] - performing chromatic aberration correction on the resolved image, wherein, preferably, the chromatic aberration correction is performed based on chromatic aberration correction values calculated from multi-col- ored beads, and wherein corrected coordinates are then applied to SRRF- resolved images. The method of item 97 or 98, further comprising:

[0059] - reconstructing an image hyperstack of the biological sample based on the resolved image. The method of one of items 92 to 99, wherein performing the SRRF analysis is repeated for each image. Computer-readable medium comprising computer-executable instructions that, when executed by a computer or processor, configure the computer or processor to perform the method of one of items 92 to 100. Computer system comprising:

[0060] - a computer readable medium according to item 101;

[0061] - an image storage configured to store an image stack; and

[0062] - a processor configured to execute the computer-executable instructions stored on the computer readable medium. A method of capturing an image of a biological sample placed on an optical microscope and using a camera in the optical path of the microscope, the method comprising: - using SRRF analysis and setting the frame rate to achieve a scan dura- tion between 5 ms and 200 ms, preferably between 12 and 40 ms, preferably between 20 to 40 ms, or 25 ms

[0063] - capturing at least 500, at least 1000 at least 1250 at least 1500 frames for reconstruction of an SRRF image. A method of capturing an image of a biological sample placed on an optical microscope, the method comprising:

[0064] - using SRRF and setting the number of frames used for reconstruction for a single image to more than 500

[0065] - imaging of an expanded sample, wherein the proteins in the expanded sample are digested in a way, that more than 5%, preferably more than 10%, or 15%, most preferably more than 20% of the peptide bonds are broken before expansion of the sample. A microscopy method for obtaining a high-resolution image of a biological sample comprising a substance to be analyzed, the method comprising:

[0066] - obtaining an image sequence of said sample by expansion microscopy using more than one anchor to connect the sample to the swellable gel. A microscopy method for obtaining a high-resolution image of a biological sample comprising a substance to be analyzed, the method comprising:

[0067] - providing an expanded sample with an expansion factor of at least lOx

[0068] - performing SRRF imaging and analysis defined by using radiality magni- fication of 12 to 60, preferably 15 to 45, further preferably 20 to 40, preferably at 35.

[0069] I. Fixation of the sample: How to achieve a sufficient sample fixation, which al- lows taking images with a high resolution according to the present invention. a) Chamber:

[0070] 1. Size and form of beams / columns / pyramids (Angles, height, how broad, which distance)

[0071] Exemplary chamber details as shown in Figure 1

[0072] Material used for production

[0073] Chamber cover / EN AW 5083 [AIMg4,5MnO,7]

[0074] 3D-printed Gel cage I Pursa White Tough ink

[0075] Chamber holder / EN AW 5083 [AIMg4,5Mn 0,7]

[0076] Chamber dimensions

[0077] 73x28 mm outer dimensions and 40x22 mm inner dimensions with height of 4 mm.

[0078] The coverslip-mounting area is 40x22 mm with height of 0.5 mm.

[0079] 2. Examplary size of the imaging windows (ranges)

[0080] 3D-printed Gel cage (net) dimensions (LxWxH)

[0081] Gel cage dimension: 40x22x3 mm

[0082] Central imaging zone: 7x7x0.5 mm

[0083] Six-flanking imaging zones: 5x5x0.5 mm

[0084] Counter-drift squares: 2x2x0.5 mm

[0085] Net backbone-Quadrilateral polygon dimension: 0.5x0.5 mm with symmetrical 70° base angles and 110° top angles

[0086] Exemplary polygon properties (the polygons forming the cage patterning):

[0087] • Range for the height of the polygons: 0.25 to 1 mm (most preferred 0.5)

[0088] • polygon width: 0.2 to 0.8 (most preferred = 0,5)

[0089] • polygon height: 0.25 to 2 mm / 0.3 to 1mm / most preferred 0.5mm)

[0090] • angle alpha: of 80° to 30° (most preferred 70°)

[0091] • angle beta: optimum is parallel (180° - angle alpha); parallel to ground of the chamber

[0092] • A preferred minimum number of counter-drift squares is such that about 50% of the chamber should be filled with the mesh • A preferred size of the imaging zones (diameter) in any shape, wherein rec- tangle or square is preferred: 3 mm to 30 mm, 4 to 20 mm, 5 to 12 mm, 7 mm

[0093] • distance of „mesh-lines" (i.e. the size of counter-drift segments, preferably in the form of squares): 1 to 3 mm , 2 mm

[0094] • the mesh is preferably a mesh having open voids, i.e. it maybe necessary to have a completely „open" ground, since the sample / gel has to touch the glass at the bottom in order to provide best results

[0095] 3. Needles, rectangular shape or round structure as mesh may not work or are unfavourable for the following reasons:

[0096] • Needles are sub-optimal as induce breaks in the gel that end up in fragment- ing the gel in smaller pieces. Such cracks induces medium discontinuity, which affects the refractive index and light scatter properties.

[0097] • Rectangular imaging boundaries are superior for imaging to circular bounda- ries. Navigating the x,y-stage parallel to straight lines is far easier than follow- ing a circular path around the round boundaries. While in theory a round boundaries would work in limiting drifts, they pose a navigational-challenge while screening for samples.

[0098] Round structures for providing the cage patterning may work to a certain extent.

[0099] 4. Alternative solutions / structures include

[0100] Gluing the gels with acrylic-based sprays or sugar based-coating. Growing the gels on chemically-active glass surfaces, using multiple types of chemistry. However, these approaches provide inferior results as compared to the use of the mesh / cage patterning of the present invention. Also sandwiching the gels between 2 cover- glasses is inferior. Anchoring the gels to Poly-L-lysine coated coverslips can partially work in immobilizing the gels, however, gel-shrinkage was also observed, which is counter-productive, as the expansion factor was diminished.

[0101] 5. Sealing of the chamber

[0102] Glycerol (glycerin) can be coated on the internal side of the bottom chamber and a glass coverslip, preferably a #1.5H glass coverslip of 40x22 mm dimensions can be added, then the net can be mounted and a piece that can be roughly equal to the size of the net can be slid onto it. The upper part of the chamber can then be added onto the lower-chamber part+net combination and held together by the two flanking rods. Exemplary details of setting up the chamber are as follows:

[0103] 1. Cutting of a rectangular gel piece out of the main gel with a size that is slightly bigger than the 3D-printed net dimension.

[0104] 2. With a kimwipe paper, removing excess water from the gel, not much, just a bit, even if water remains in the gel.

[0105] 3. Assembling the chamber by adding the rectangular coverslip accompanied with the package. Don't forget to add grease on the metal part of the chamber before adding the rectangular coverslip, to prevent water leaking on the microscope equipment.

[0106] 4. Adding the net on top of the glass.

[0107] 5. Sliding the gel onto the net.

[0108] 6. Cutting the excess pieces of the gel out. IMPORTANT NOTICE: Please do not use a smaller piece than the net, since the gel piece should be equal to that of the net size.

[0109] 7. Leaving it rest for 10 minutes under no light condition.

[0110] 8. With forceps, lifting the net from the side and remove the collected water under the gel with kimwipe paper, be gentle not to scratch the glass surface. Using regu- lar tissues is not advised.

[0111] 9. Placing the net back onto the glass surface. You would instantaneously see the gel adhering to the glass surface.

[0112] 10. Placing the assembled ONE Chamber onto the microscope stage.

[0113] 11. Letting it rest from 5 to 10 minutes and enjoy stable gel imaging.

[0114] It should be noted that after 40-70 minutes of imaging, if gels display a rapid drift af- ter being stable, that means the gels are expelling water and started to shrink. It is advised to stop collecting data at this point as the expansion factor would be de- creasing at this moment.

[0115] 6. Other exemplary special aspects of the chamber

[0116] 1. The chamber can comprise a net groove (as shown in the picture above), this was designed for convenience, it is used when the experimenter wants to re- move the net, they can either lift it by hand from the two flanking grooves or use forceps. The first chamber model didn't have the grooves and were less convenient in handling.

[0117] 2. The chamber can comprise water collection wells. These spaces allow ex- cess water flow into these spaces outside imaging zone. b) Exemplary gel-preparation: 1. Preferred thickness and possible restriction in thickness of the gel The optimal gel thickness is between 0.4 to 0.5 cm. To achieve that, 70 to 90 μl of gelation solution for 18 mm coverslips is added.

[0118] If the gel is thinner / thicker, breaking of the gel can occur or the gel can become hard to handle.

[0119] 2. „Drying" of the Gel

[0120] The gel is gently dried, known as semi-drying. This step can be essential for samples intended to investigate protein shapes. Since the anchor can be added in 100 μl PBS solution the night before, this amount of liquid will affect the gelation reaction. The sample are semi-dried until a fine film of solution remains on the coverslip before adding the gelation solution and allowing the polymerization process. A too dry sam- ple turns white (due to the light interference pattern provided by salts in the buffer), indicating that the dried proceeded for too long, i.e. the sample became too dry.

[0121] Such a sample is too dry and the purified protein molecules aggregate and cannot be separated.

[0122] 3. Dimethylacrylamide (DMAA) as gel

[0123] DMAA as in the prior art is able to cross-link without the presence of BlS-acrylamide (BIS) cross-linker as most of expansion protocols utilize. DMAA is preferred as it is assumed that this might limit or reduce non-isotropic expansions. It is known from the literature that changing the concentration of BIS affects the expansion factor, which is is probably due to how much it cross-links the polymers together. c) Further general comments on the requirements for imaging at that high resolu- tion

[0124] Stable gels allow prolonged imaging. The chamber limits the drifts to nearly 0.0005 pm / s. This value allows 5-dimension (x,y,c,z,t) imaging with up to 2000 frames in each channel (up to 3 minutes of imaging using 12 kHz resonant scanners).

[0125] A drifting sample with up to 0.075 pm / s can be compensated for. Drifts without the stabilizing chamber can reach to more than 0.5 pm / s.

[0126] II. Chemistry I sample preparation: a) Example of Crosslink Protein to the gel

[0127] 1. AcX is a standard reagent - binding according to the present invention (with lysine / serine) Acryloyl-X (Ac-X, 6-((Acryloyl)amino)hexanoic acid succinimidyl ester) reacts with amines in proteins, particularly lysines, amine-modified nucleic acids and N-termini of peptide chains. This reaction, therefore, introduces a gel-linkable component into these cellular macromolecules.

[0128] In general, NHS-esters like Acryloyl-X, which should be suitable, are NHS-esters hav- ing a terminal acrylic function or acryloylamino function which is linked to the NHS- ester, for example linked by linear Cl-12 alkyl chain linker.

[0129] A problem with it can be the following: for any cell work, the materials are normally fixed before using Ac-X. Fixation modifies many of the amine groups, implying that Ac-X is not as effective as it would have been in the absence of fixation. This reduces its usefulness.

[0130] 2. Working principle of the Ser (serine)-Binder and advantages thereof The main advantage of a Ser-Binder is to avoid the problem of fixation limiting the effectiveness of the binder. Unlike amine groups, the hydroxyl group of serines is not affected by common fixation procedures, implying that this molecule would link to substantially more sites in molecules than Ac-X.

[0131] Ser-Binder was designed to in two variants, one molecule (reactive moiety of the binder) that have one function which is to anchor the serine residues in the gel, and a second molecule (reactive moiety of the binder) that anchors the serines but also report their position. The second Ser-binder variant carries a FAM (fluorescein dye). Both variants work by binding specifically to serine amino-acids in proteins and the reaction takes place as presented below.

[0132] However, as mentioned this reaction is difficult and for it to work a series of optimi- zations were carried out.

[0133] It should be noted that the serine targeting molecule may not dissolve well in some H2O / DMSO solution.

[0134] Therefore, the solubility and reactivity of this molecule was tested in different H2O / DMSO solution in different concentrations.

[0135] The following scheme shows the underlying reaction:

[0136] Scheme. Labelling of Ser in DMSO / H2O - Results of solvent optimization

[0137] Solvent cone Result

[0138] 10% DMSO / H2O 25 mM precipitation

[0139] 20% DMSO / H2O 25 mM precipitation

[0140] 50% DMSO / H2O 25 mM clear solution, reaction works

[0141] 10% DMSO / H2O 5 mM precipitation

[0142] 20% DMSO / H2O 5 mM (a bit) precipitation, reaction works

[0143] 50% DMSO / H2O 5 mM clear solution, reaction works

[0144] 10% DMSO / H2O 2.5 mM precipitation

[0145] 20% DMSO / H2O 2.5 mM (a bit) precipitation, reaction works

[0146] 1) In 50% DMSO / H2O solution, the serine targeting reagent dissolve and react well.

[0147] 2) In 20% DMSO / H2O solution, precipitation forms, but the labelling reac- tion still works.

[0148] 3) In 10% DMSO / H2O solution, huge precipitation forms and it looks this is not a good condition for the labelling reaction.

[0149] Serin was is abundant, i.e. amounting to ~7.3% of all amino acids in proteins in the mouse brain.

[0150] Moreover, there are not many reactive amino acids. Those that are highly reactive (amine- and carboxyl-carrying amino acids) are all present in the same protein seg- ments, exposed on the protein surface. Binding to them would not be very different than binding to amines, since the same protein parts are bound. Serine is also pre- sent in non-surface areas, so that it enables us to bind to protein parts that are not accessible to Ac-X. An additional advantage of the Ser-binder, is that it is not involved in fixation pro- cess. As mentioned above, it is not modified by fixation, therefore it does not lose any reactivity after fixation.

[0151] Exemplary structure of the Ser-binder:

[0152] Exact Mass: 389.12 Exact Mass: 738.17

[0153] 1 2 alkene-PSI FAM-PSI

[0154] Scheme. Structure of Ser-binder. Ser-binder 1: Alkene-PSI and Ser-binder 2: FAM- PSI, comes with fluorescein molecule attached to it. FAM (6-Carboxyfluorescein) is a fluorescent dye.

[0155] Furthermore, PSI having a dye or fluorescent function or dye or fluorescent function (preferably FAM) which is linked to the PSI may be suitable, for example linked by linear substituted or unsubstituted Cl-12 alkyl chain linker.

[0156] 3. Combination of AcX and Ser-binder - reaction protocol

[0157] This reaction is sequential and works sub-optimally or even fails if not carried out in the respective order:

[0158] Add 3mg / ml Ac-X anchor onto the specimens in (3 μl Ac-X in 50 μl PBS at pH 7.4 for 2 hrs at room temperature. Then add the Ser-binder 1: 30 μl of PSI + 20 μl DMSO or Ser-binder 2: 10 μl of 3 mg / 50 μl + 40 μl DMSO, and incubate overnight at 4° C. For proper function of Ser-binders, the final DMSO concentration should be around 50% when mixed with other solutions.

[0159] If the sequential reaction is not followed, this can affect the result of the reaction.

[0160] For example: 1. Adding Ser-binders before Ac-X: Ac-X may not work, or may yield poorer anchorage of samples into gels.

[0161] 2. Adding Ser-binders in less than 20% DMSO environment, this showed no anchorage improvement of Ac-X+Ser-binder. Implying that 20% DMSO or less becomes an unfavorable condition for the reaction.

[0162] 3. Incubating Ser-binders for about 4 hr or less did work, however, the incubation overnight yielded the brightest signals, implying that is reaction under these conditions might be a slow reaction.

[0163] This Ser-binder did not exist in the prior art and was surprisingly found by the inven- tors. b) Example of Anchoring I homogenization / digestion

[0164] 1. Protocol

[0165] Thaw the stock of the anchoring reagent Acryloyl-X.

[0166] Prepare a stock solution of 10 mg / ml, add 500 μl DMSO to a 5 mg Acryoloyl-X bottle. Prepare 5x 100 μl aliquots. Add 3 μl / 100 μl PBS achieving a 3 mg / ml. Add enough Acryloyl-X / PBS solution to cover the sample 100 μl for purified proteins and 300 pl for cell culture.

[0167] Seal the plate with parafilm and incubate at 4°C overnight.

[0168] 2. Necessary features of the anchoring / homogenization / digestion Proteinase K was added at a 1: 100 concentration to reach a final activity of 8 U / ml. Samples were incubated at 50°C (wet chamber) from 18 hr to 24 hrs. Substantial trial-and-error processes were followed, including varying concentrations of protein- ase K, different amounts of potential cofactors, like ions, different temperatures, dif- ferent lengths of time, and so on. c) Labelling

[0169] Generally, any NHS ester can be suitable for labelling the substance to be analyzed, e.g. a protein or a fragment thereof. Preferred NHS esters include the following:

[0170] NHS-ester fluorescein:

[0171]

[0172] NHS-ester comprising STAR635P fluorophore, wherein the fluorophore is shown in the following:

[0173] , and

[0174] Atto643 NHS-Ester.

[0175] 1. NHS-Ester ( / V-Hydroxysuccinimidesters) based labelling (e.g.: NHS-Ester STAR635 (Abberior®))

[0176] Digestion buffers were washed with sodium bicarbonate buffer at pH 8 to 8.4 three times while on gentle shaking for 5-10 minutes each. The NHS-Ester labeling was added in sodium bicarbonate buffer at pH 8 to 8.4 in huge abundance between x200 to x300 times. The gels were incubated in 8 cm Petri dishes and about 5 ml of buffer + NHS-ester dye was added. The Petri dishes were then transferred on a horizontal shaker with gentle shaking speed and incubated at room temperature for 30 to 45 minutes.

[0177] 10- to 20-fold more NHS-ester dyes is added than recommended in the literature. NHS-Ester reactivity is lost in the first few minutes upon adding water-based buffers. It is imperative to be added into the gels in the first minute upon mixing with sodium bicarbonate buffer, to ensure maximum labeling efficacy.

[0178] The fast buffer exchange is referred to the washing step of NHS-Ester labeling. The advantages are:

[0179] 1. Saving time

[0180] 2. Washing with sodium bicarbonate buffer prevent the NHS-Ester mol- ecules from aggregating and sticking randomly into the gel poly- mers.

[0181] Exemplary Images are shown in Figures 2 - 6.

[0182] II Imaging / Image analysis

[0183] Images were acquired using objectives of 1.4, 1.45 and 1.51 NA with a theoretical pixel size of 98 nm (optionally for confocal imaging, wherein 100 is according to the state of the art). For a higher resolution, the theoretical pixel size was set to 48 or 24 nm to gain more resolution and detail, which may be at the cost of slightly lower de- tection rate. Images acquired on a camera-based system had a predetermined pixel size of 100 nm. The acquisition speeds were ranging between 20 to 40 ms and 25 ms on a resonant scanner of 8 kHz and on a camera, respectively, for xyct. For hy- perstacks of xyczt acquisitions, images were acquired using 8 kHz and 12 kHz scan- ners in bidirectional mode (after the necessary alignments) to compensate for speed loss. 12 to 30 ms per frame can be selected for good outcome. 25 ms can be opti- mal. Up to 200 ms per frame can be employed. Images of 8-bit depth can be ac- quired at a line format ranging from 128x128 to 256x256. The scanning modality on a confocal microscope can be set to "minimize time interval". To maintain natural fluctuations of fluorophores, line accumulation or line averaging can be omitted dur- ing scanning. A frame count starting from 200 and up to 4000 can be acquired. A frame count of at least 1500 to 2000 can be selected for optimal computed resolu- tion.

[0184] I. Improvement of Imaging parameters: a) Characteristics to improve the image quality

[0185] 1500 to 4000 frames may be preferably. 2 kHz for x,y,t images work. 8 kHz is re- quired for x,y,c,t (multiple channels over time). 12 kHz is required for x,y,z,t (stacks in one color over time). 16 to 24 kHz are required for x,y,c,z,t (stacks in multiple col- ors over time).

[0186] The scan frequency determines the acquisition speed and depends on the experi- mental settings. Meaning the more channels to acquire, the higher the scan fre- quency is required to maintain a certain speed range for frame acquisition.

[0187] A faster than 10 ms per frame might induce issues in the computational step. A slower than 200 ms per frame yields images with less detail. b) Preferred ranges of image acquisition parameters

[0188] In a 128x128 scan format over a scan speed between 2 to 24 kHz, a photon count average from the entire 128x128 frame is optimal once reported from 4 to 40 pho- tons per pixel (depending on the density of labels and nature of specimen; i.e. de- pending on the nature of sample: For single purified protein molecules, an average of 4 photons per frame is typical, for cell cultures its about 20 photons, for densely labeled samples and tissue sections can reach around 40 photons.). c) Relation of the detection speed, image dimension, scan rate

[0189] The speed at which photons are detected should not be compromised by adding multiple imaging dimensions, as in different colors and optical sections, while record- ing the time series. An ideal scan rate ranges between 12 to 30 ms per frame of 128x128. Setting 64x64 will be faster which might be relevant for slower scanners when having 2 channel recording, then the final speed is still between the range of 12 to 30 ms / frame, i.e. speed per pixel. Faster imaging risks computation failure, as there may be frames that do not have any detected photons inside and slower would still work but sub-optimally. d) Demands for the hardware:

[0190] • 1.3, preferably 1.4 NA or higher lOOx objectives.

[0191] • Resonant scanner • Sensitive detectors (ideally with photon counting precision).

[0192] • Piezo stage for z-stacks (x,y,z,t images) e) Further improvements

[0193] Bringing the sample into one of the gel edges. As there may be an index mismatch between the glass-gel interface, the deeper the specimens are imbedded inside the gel, the more artifacts are likely to arise in the computation. The samples are brought to within the first 1 to 2 pm of the gel, by the semi-drying process or coating the coverslips with Poly-L-lysine. f) 3D STED

[0194] 3D-STED was first introduced in 2008. Here the present invention can also be com- bined with 1 or 2 doughnut lasers: a laser in x,y and a laser in z orientations. This al- lows for less photon flux being detected by the detectors. This enhances the axial resolution from 45 nm down to 3.7 nm. Tomography STED can also be combined with the methods of the present invention.

[0195] II. Image analysis - Technical Processing information for image analysis:

[0196] • Number of frames: up to 4000 frames, with a minimum of 1000, pref- erably 1500 frames, further preferably 2000.

[0197] • Channels to process: multiple-automatic

[0198] • Drift channel reference: automatic

[0199] • Radiality magnification: from 10 to 35 (35 provides more detail as shown in the scheme above)

[0200] • Ring axes: 8

[0201] • Temporal analysis mode: Temporal Radiality Auto-Correlations (TRAC)

[0202] • TRAC order: 4 or higher and up to 20, depending on the sample prepa- ration

[0203] • Distance to scale: 0

[0204] • Known distance: 1

[0205] • Expansion Factor: 13.3 (or any expansion factor used by the user)

[0206] Advanced options:

[0207] • Integrate temporal correlations: active

[0208] • Remove positivity constraint: active

[0209] • Renormalize: inactive

[0210] • Do gradient smoothing: active • Do intensity weighting: active

[0211] • Do gradient weighting: active

[0212] • PSF FWHM: 3.17

[0213] • Minimize SRRF patterning: active

[0214] The technical methodology is shown in Figures 7 - 9. a) Further important aspects of the imaging procedure

[0215] 1. Recording expanded samples for more the 500 frames, surpassing the needed 1500 frame-threshold for TRAC4 analyses (temporal autocorrelations analysis of order 4) or higher order to work. This was only possible because of the gel stabi- lizing chamber. This allows the yielding of high resolution images.

[0216] 2. The idea of labeling the individual protein peptides in purified protein samples to determine their shape was implemented for this invention and was not reported anywhere in the literature in fluorescence microscopy. b) Further important aspects to the drift correction

[0217] Treating the gels as live specimens, meaning implementing drift correction measures that corrects by trying to corelate a frame with the preceding one. By increasing the search radius of drifts and every 10thimages can be binned together to get bright enough signal for the drift correction algorithm to work. In the case of multiple chan- nels, we implemented a solution whereby the software will try to determine the brightest signal to apply drift-correction and propagate this correction onto the other channels without input from the user.

[0218] III. Exemplary Protein Structure determination

[0219] 1. Sample preparation - protocol

[0220] In order to achieve the invention, a couple of obstacles had to be figured out before the first successful images of protein shapes to be observed using light microscopy and expansion technology followed by intensity fluctuation analysis.

[0221] The protocol used is as provided below, with commentary on the troubleshooting:

[0222] 1. 1.5 to 2 μl of purified protein of interest of 1 mg / ml concentration were added into 100 μl PBS buffer of pH 7.4 and spread onto 18 mm co- verslips. Lower concentrations can become too difficult to spot and lo- cate in the expanded gels. Higher concentrations can induce protein aggregation. 2. 3 μl of Ac-X where added into the samples (as described above) and in- cubated for 2 to 3 hrs at room temperature.

[0223] 3. 10-30 μl of FAM-PSI or PSI in 50% final concentration of DMSO is added and incubated overnight. A reversed order of Ac-X / FAM-PSI or less DMSO percentage can interfere with the success rate of anchoring to the gels. In the prior art, the trend for the skilled person was lately even not to use anchors at all, rather entanglement concepts, which only spread protein molecules apart but would not expand individual protein molecule. Double anchoring as defined by the present invention was thus counter intuitive for the skilled person and not obvious from the prior art. Normally the skilled person would try to avoid too much digestion and too much anchoring.

[0224] 4. The samples were then semi-dried as described above until a thin film on coverslip remains. In the following, the gelation solution is added. It should be noted that too dry samples can destroy the protein molecules and induce aggregation, while a lot of anchoring buffer remaining af- fects the constituents of the gel, ending in weaker robustness and an altered expansion factor.

[0225] 5. The remaining expansion procedure is as described above.

[0226] 6. The labeling is as described, with excess NHS-Ester.

[0227] 7. Washing steps as described above.

[0228] 8. Expanding the gels as described above.

[0229] Adaption of the sample preparation

[0230] Fixatives (for fixation of e.g. proteins) interferes with the anchoring process and re- sults in poor shape preservation. Hence, fixation of single molecules experiments de- signed for revealing protein shapes is not essential to the present invention.

[0231] 2. Labelling with one or more colors to follow the protein chain (for example Cys- teine and Serine) (see Figures 10 and 11): a) Simultaneous labelling with NHS-Ester to make the chain visible

[0232] Simultaneous labelling as shown in Figures 10 and 11 can make the protein chain visible. b) Structure calculation from the images of double-labelled proteins In principle, the same principles as cryo-electron microscopy can be employed for structure calculation: a 3D structure can be obtained by combining the information from multiple 2D images, using a specialized software. Multiple amino acid labeling, however, will enable us to barcode the protein, thereby obtaining direct information on their identities, for which Cryo-TEM is not suitable.

[0233] Principles and validation of ONE microscopy

[0234] \NQ first attached a gel-compatible anchor (Acryloyl-X) to protein molecules, either purified or in a cellular context, and then embedded these samples into a swellable X10 gel28'29. Proteins were hydrolysed by proteinase K or by heating in alkaline buffers, leading to main chain breaks. This enables a highly-isotropic 10-fold expansion of the sample, which is achieved by distilled water incubations28'29. The alkaline peptide hy- drolysis we use, at >100°C, has been initially designed to generate free amino acids from protein mixtures (e.g. urine), for clinical investigations. It is therefore sufficient to cause the fragmenting of every protein, under our analysis conditions. Its current implementation, according to the protocol described in Methods, provides a mild frag- mentation of individual proteins, while maintaining most of the fluorophores that were present on proteins before expansion (for more details see the Supplementary Discus- sion).

[0235] We then imaged the samples using wide-field epifluorescence or confocal microscopy, acquiring series of images (movies) of hundreds to thousands of images (ideally 1500- 2000) in which the fluorescence intensity of the fluorophores fluctuates (Extended Data Fig. 2). Each pixel of a frame was then magnified into a large number of subpixels, and the local radial symmetries of the frame (which are due to the radial symmetry of the microscope's point-spread-function, PSF) were measured. This parameter, termed "radiality" was analyzed throughout the image stack, by higher-order temporal statis- tics, to provide the final, fully resolved image6'7'27.

[0236] In theory, the precision of the SRRF technique should reach values close to 10 nm6. SRRF should therefore be able to separate fluorophores found at 20 nm from each other, provided the signal-to-noise ratio is sufficiently high. We found this to be the case, using nanorulers (provided by GATTAquant30), of precisely defined size (Supple- mentary Fig. 2).

[0237] In practice, most previous implementations of SRRF have reached ~50-70 nm. This is partly due to the fact that the presence of overlapping fluorophores reduces radiality in conventional samples6'7, and partly due to the aims of the respective SRRF imple- mentations, which did not target ultimate performance in terms of resolution, and therefore did not optimize a number of parameters. First, the highest resolutions are obtained by analysing higher-order statistical correlations, whose precision is depend- ent on the number of frames acquired, as discussed not only for SRRF, but for SOFI as well26. While most publications use less than 300 frames, we found that results are optimal when using 1500-2000 frames (Supplementary Fig. 3). Working with low frame numbers reduces the achievable resolution, even when working with ExM gels20'31. Second, the signal-to-noise ratio needs to be optimized carefully32, as we also demon- strate in Supplementary Fig. 4.

[0238] These limitations are alleviated by ExM (see Supplementary Discussion for more de- tails). The distance between the fluorophores increases, enabling the study of intensity fluctuations from individual dye molecules independently. The signal-to-noise ratio also increased, even for idealized samples consisting only of fluorescently-conjugated nano- bodies in solution (Extended Data Fig. 3). This approach should therefore allow an optimal SRRF performance, which, divided by the expansion factor, should bring the resulting imaging precision to the molecular scale (Figure 12b), as long as the gel expands isotropically in all dimensions. The X10 gel, based on A£ / V-dimethylacrylamide acid (DMAA), rather than the acrylamide used in typical ExM protocols, has a more homogeneous distribution of cross-links33, thus leading to fewer errors in expansion (see34for a further discussion on gel homogeneity).

[0239] To assess the performance of ONE microscopy in a cellular context, we first analysed microtubules, a standard reference structure in super-resolution imaging techniques17. Gels were stabilized in specially-designed imaging chambers (Supplementary Fig. 5), which enabled us to image the antibody-decorated microtubules at both 10-fold and ~3.5-fold expansion (the ZOOM ExM technique35was used for the latter; Figure 12).

[0240] The microtubule sizes matched previous measurements36-37, ~60 nm in diameter, when labelled with secondary antibodies, and around 30-35 nm, when labelled with secondary nanobodies (Extended Data Fig. 4c, d).

[0241] We then evaluated a purified ALFA-tagged EGFP construct bound simultaneously by two anti-GFP nanobodies38and by an anti-ALFA nanobody39. This results in a triangular semi-flexible arrangement, which we termed a "triangulate smart ruler" (TSR, Figure 12b; Extended Data Fig. 5). The TSR aspect observed in ONE microscopy is consistent with crystal structures of nanobody-EGFP and nanobody-ALFA complexes (Figure 12b, c).

[0242] To reveal the protein molecules themselves (the 10-fold expansion eliminates the en- dogenous GFP fluorescence for example (Figure 12e)), we labelled the TSRs using NHS-ester fluorescein40-41, which is sufficiently stable, under our imaging conditions, for this type of experiment (Supplementary Fig. 7). This is possible because proteins are broken during homogenization at multiple main chain positions, and each resulting peptide has an exposed amino terminal group that can be efficiently conjugated with N-hydroxysuccinimide ester (NHS-ester) functionalized fluorophores. It is known that nanobodies are not as strongly anchored to ExM gels as other proteins, owing to their low lysine content (only 2 lysines for the ALFA nanobody), and most of their peptides are lost after homogenization17. Their fluorescein signal is therefore poorer than that of GFP (Figure 12f, Extended Data Fig. 5g). This is an unexpected bonus of nanobod- ies, because nanobody signals do not obscure those resulting from the protein of in- terest (see Supplementary Fig. 6 for a gallery of examples). In these experiments we used a smaller pixel size than in Figure 12f (0.48 nm vs. 0.98 nm), which enabled us to often observe dual fluorophores, in agreement with the fact that these nanobodies can be labelled at two positions. The distances between the two fluorophores on one nanobody are consistent with the size of nanobody molecules (see the graph in Figure 12f). Measuring the full width at half-maximum (FWHM) of the fluorescence signals resulted in an apparent particle size of ~1 nm in the different fluorescence signals, including the fluorescein channel (Extended Data Fig. 5). Values within the same range are obtained when turning to an often-used technique to quantify the resolution of individual images in super-resolution fluorescence imaging, as well as in conventional electron microscopy, the Fourier Ring Correlation (FRC) determination42'43. We applied this approach to our images, relying on the NanoJ-SQUIRREL package42which has a blockwise implementation, to provide FRC values for different regions within individual images (Extended Data Fig. 5), obtaining values within the low single-digit nanometer range.

[0243] ONE microscopy can reveal protein shapes

[0244] Considering the fact that proteins expand lOOOx in volume but fluorophores do not, we hypothesized that our NHS-ester labelling method could be optimized to enable the analysis of protein shapes by ONE microscopy. In the TSR experiments (Figure 12) we used limited NHS-ester labelling, to avoid the fluorescent labelling of the nanobodies, which also limited the GFP labelling to poorly distinguishable blobs. To observe the GFP shape better, we used optimal labelling conditions (high excess of NHS-ester flu- orescein), and then proceeded to ONE microscopy. The expected shape and size were obtained for the GFP molecules (Supplementary Fig. 8). We next applied this approach to antibody molecules, and we could observe immediately recognizable outlines for IgGs, IgAs and IgMs (Figure 13a-c, Supplementary Fig. 9). Fluorescent labels attached to secondary IgG antibodies could also be observed in the same images (Figure 13a; Supplementary Fig. 9) and also in complexes between fluorescently-conjugated pri- mary and secondary antibodies, or nanobodies (Figure 13d).

[0245] We then applied the same labelling method to a membrane protein, the full-length 03 human y-aminobutyric acid (GABAA) receptor homopentamer, a ligand-gated chloride channel44, producing imaging that resembled "front" and "side" views of the receptor, similar to its structure, as derived from crystallography and single-particle cryogenic electron microscopy (Cryo-EM) structures (Figure 13e,f; Supplementary Fig. 10). It is worth noting that particles observed by ONE microscopy are indeed single molecules, and no averaging or classification has been performed on these datasets.

[0246] We next investigated a protein of unknown structure, the ~225 kDa otoferlin, a Ca2+sensor molecule that is essential for synaptic sound encoding45. The outlines provided by ONE microscopy imaging strongly resemble the AlphaFold46prediction for this pro- tein (Figure 13g, h, Supplementary Fig. 10). Moreover, scanning in both the axial and lateral dimensions, using confocal laser scanning microscopy, enabled us to obtain 3D information on single otoferlin molecules (Figure 13i). At the opposite end of the Ca2+sensor size spectrum, we sought to visualize the small (~17 kDa) protein calmodulin, expressed as a GFP chimera. To our surprise, even for such small particles, it was possible to observe dynamic changes in their shape upon Ca2+binding (Figure 13j-n). We applied both heat denaturation and proteinase K treatments for the homogeniza- tion of calmodulin, to test whether these methods would lead to different results. The proteinase K presumably removes all amino acids that are not anchored into the gel, and is therefore more aggressive than the heat denaturation47. However, both meth- ods resulted in similar observations for calmodulin, implying that both can be used for observing the shape of purified proteins.

[0247] Protein averaging using ONE microscopy

[0248] One important challenge in ONE microscopy is to obtain deeper insight in the protein structure, beyond imaging the rough shape of an individual protein. In principle, one should be able to average the images of multiple proteins, but many difficulties are apparent, from our low resolution in the axial direction, which implies that every image is a projection of a relatively large volume, to difficulties in determining the orientation of the particles. Nevertheless, we attempted this approach, relying on the GABAA re- ceptor, in complex with GABAAR-binding nanobodies48carrying fluorophores. These nanobodies recognize the receptors in vitro, providing the expected ONE images (Sup- plementary Fig. 12). Selected views fit very well with cryo-ET images of the receptor (Supplementary Fig. 13). To generate an average receptor image, we used an automated particle detecting routine to find about 5000 particles, removing the ones with any signs of uncompen- sated drift (for drift examples, see Supplementary Fig. 14). We then inspected visually all of the remaining particles, to choose those that appeared to be in a "front view", showing a reasonably round appearance, with nanobodies placed at the edges of the receptor (1225 particles). These particles were subjected to an analysis of the peaks of fluorescence, which were then mapped into one single matrix, which represents the "averaged receptor". All sets of positions were rotated so that the nanobodies point "up" (Figure 14).

[0249] As indicated by the different examples, not all receptor particles are decorated by five nanobodies, due to various issues, ranging from incomplete nanobody binding to the loss of fluorophores during the anchoring and expansion steps. This implies that the averaging procedure induces a bias for the nanobody image, since the nanobody with the highest fluorescence will be the one placed in the "up" position. This type of aver- aging will result in a prominent labelling for the nanobodies in this position, and little else in the nanobody channel. Even if more nanobodies are always present on the same receptor, small imprecisions in their placement will make them not align properly in the average image, resulting in one bright nanobody, up, and a blur elsewhere.

[0250] The results are nonetheless satisfactory. The average receptor obtained has a pore in the middle, and has the nanobody at a reasonable position. Two main fluorophore positions appear on the nanobody, at the expected distance from each other, exactly as predicted by the double labeling of the nanobody (two fluorophores on each nano- body). To determine whether this average view is "good enough", we modeled recep- tor images, starting from the Cryo-EM coordinates of the receptors. Our models indi- cate that, in a perfect world, the blur observed in our receptor would come from a placement error, for each fluorophore, of about 2 nm (Figure 14c). However, taking into account the fact that our 1225 particles have slightly different positions and tilt angles (Figure 14d), the average receptor image is as good as it could be expected (Figure 14b).

[0251] Overall, this approach indicates that single-particle averaging is indeed possible, albeit it would benefit greatly from increased resolution in the axial dimension, which can be achieved, for example, by using ExM protocols with expansion factors beyond lOx49-50. This would enable us to determine better the position and tilt of each particle, thereby improving the results. In principle, a similar procedure could be performed using purified tubulin assembled into microtubules in vitro. Unfortunately, this is an incredibly difficult experiment, since the microtubules are unstable in vitro. If they are thoroughly fixed, using glutaralde- hyde, then they barely link to the gel, since majority of amines are modified by the fixative. If they are poorly fixed, with PFA, they link into the gel, but only in part, since a degree of microtubule depolymerization is still observed, leading to a destruction of the structure, quite visible both before and after expansion (Supplementary Fig. 15). However, we did succeed in averaging the free tubulin dimers that result from the depolymerization of the microtubules, and found that they averaged to an object that is similar to the known structure of this dimer (Supplementary Fig. 15).

[0252] We also extended this analysis to actin. To use a natural, rather than in vitro, system, we turned to cell cultures subjected to detergent extraction during fixation. This pro- cedure results in the preservation of actin filaments51, which could then be analyzed in ONE microscopy. The images reproduce the known size of the actin filaments and the distance between the actin subunits, as well as providing views of the filament pitch (Figure 15).

[0253] Visualization of synaptic proteins

[0254] We next tested the performance of ONE microscopy in cultured neurons, focusing on the synaptic transmission machinery. To test whether a simple epifluorescence micro- scope could be obtained for ONE microscopy in samples where molecular resolution is not absolutely needed, we turned to an Olympus iX83 TIRF microscope, equipped with an Andor iXon Ultra EMCCD camera. This microscope should provide a maximal reso- lution of ~3 nm, since the obtained resolution is limited by the pixel size, which reaches ~1.5 nm, when the radiality and expansion factor are taken into consideration (the expected resolution cannot be better than 2-fold the pixel size52).

[0255] Synaptic vesicles fuse to the plasma membrane to release their neurotransmitter con- tents, and their molecules are afterwards endocytosed, following different pathways53. A significant fraction of the vesicle proteins are found on the plasma membrane, form- ing the so-called "readily retrievable pool"54. It is unclear whether these proteins are grouped in vesicle-sized patches, or whether they are dispersed on the presynaptic membrane. We investigated this here, using a fluorescently-conjugated antibody di- rected against the intravesicular domain of the Ca2+sensor synaptotagmin 1 (Sytl), an essential component of the vesicles. The labeling density was sufficiently high to reveal that endocytosed vesicles have the expected circular shape (Figure 16a). Turn- ing to the readily retrievable pool, we found that the molecules were grouped in areas of size consistent with theoretical expectations for single fused vesicles, with copy numbers similar to the values expected for this molecule (7-15 per vesicle55'55), taking into account the fact that one antibody can bind up to two Sytl molecules (Figure 16b, c). Removing cholesterol from the plasma membrane, using methyl-p-cyclodextrin (MpCD), forced the dispersion of synaptotagmin molecules (Figure 16b, c), albeit it left the overall synapse organization unaffected (Supplementary Fig. 16). The numbers of antibodies were not estimated by counting the individual antibody spots, which is a difficult procedure under the lower resolution provided by the epifluorescence micro- scope, but by analyzing the fluorescence intensity of single vesicles, compared to the intensity of individual antibodies (Supplementary Fig. 17). One important consideration is whether the intensity of specific spots can be used reliably in ONE images. While the current analysis suggests this, we also performed a more formal analysis of spot fluo- rescence in single-molecule images, which are the most challenging for ONE analyses. As shown in Supplementary Fig. 18, ONE does result in signal variations, but they do not impede the differentiation of real signals from background values.

[0256] In the post-synaptic compartment, we could confirm known organization principles, including the layered aspect of the postsynaptic density (PSD), in which molecules like PSD95, Shank and Homer occupy different positions in the axial direction (Figure 16d), or the clustered distribution of postsynaptic receptors, with NMDA receptors typically observed in more central locations than AMPA receptors (Figure 16e; see also58; see Supplementary Fig. 20 for a quantification of the receptor positions). These experi- ments confirm the ease with which ONE microscopy provides multicolor super-resolu- tion in crowded cellular compartments.

[0257] PSD95 has a retiolum-like organization

[0258] The fine structure of the PSD, or even its very existence, is a matter of considerable debate. The current prevalent view is that the PSD is maintained by liquid-liquid phase separation (LLPS)59, which, intuitively, implies an amorphous organization. To test this hypothesis, we immunostained PSD95 in cultured hippocampal neurons, using a spe- cific nanobody (Figure 16f), and we switched back to confocal microscope equipped with resonance scanning, to improve image resolution. PSD95 appears to be organized in a quasiregular lattice, a conclusion that was strengthened by overlaying PSD images to obtain average views (Extended Data Fig. 6). An analysis of the distances between PSD95 spots revealed that they have a preferred spacing of ~8-9 nm, which is signif- icantly different from a random distribution (Extended Data Fig. 7). A similar result was obtained when using a Ripley curve-like analysis (Figure 16g, h; see Supplemen- tary Fig. 19 for details). To test the stability of the supramolecular PSD95 arrangements observed, we incubated the cells with 1,6-hexanediol, an alcohol that has been often used to cause the dispersion of liquid phases60. This treatment readily dispersed other components of the PSD, as Homerl and Shank2, but did not affect PSD95, which appeared to remain unchanged at the confocal imaging level (Extended Data Figure 7d). This was no longer the case when samples imaged in ONE microscopy, as the 1,6-hexanediol treatment caused the PSD95 arrangements to lose much of their reg- ularity (Figure 16f-h). Importantly, this organization of PSD95 is not an artefact due to steric hindrance induced by the nanobodies, since post-expansion labeling results in virtually identical PSD images (Supplementary Fig. 21).

[0259] These results suggest that the PSD95 positioning may only be partially, but not fully, controlled by LLPS mechanisms. We propose the term "retioluni' for this nanoscale PSD patterning, a Latin term describing small string nets with knots at regular inter- vals51. These fine details of PSD95 organization are fundamentally different from the PSD95 nanodomains observed in the past by super-resolution imaging of antibody- based immunostaining58-52. The nanodomains observed in the past are most likely a result of the limited resolution of the respective technologies, as demonstrated in Ex- tended Data Fig. 8.

[0260] While all of the results presented above on synaptic proteins were derived from neu- ronal cell cultures, we would like to point out that ONE microscopy can also be applied to tissue samples to investigate such protein arrangements, as we performed for brain slices of more than 200 pm in thickness (Extended Data Fig. 9).

[0261] Towards Parkinson's Disease diagnostics

[0262] While ONE microscopy provides substantially more details than more established su- per-resolution methods X10 ExM and STED microscopy, or even their combination (Supplementary Fig. 22), its precise resolution remains relatively difficult to estimate. To provide a measurement for different structures, we performed a number of FRC analyses, as indicated in Supplementary Fig. 23. As shown in the last panels of this figure, the average FRC value is below 1 nm when suitably small pixels are used in the ONE analysis. This observation is confirmed by our ability to measure intra-molecular distances as low as 0.5 nm, within single molecules (Supplementary Fig. 24).

[0263] Overall, while precise claims of resolution are difficult to make for this imaging scale, these observations suggest that ONE microscopy should be able to tackle real-world questions. Another pre-condition for such analyses is that the expansion is sufficiently precise (isotropic) to not modify the protein shapes drastically. This is indeed the case, as demonstrated by a large number of measurements, performed across a range of different proteins (Supplementary Fig. 25). Therefore, we next sought to address a pathology-relevant imaging challenge by ONE microscopy. Parkinson's disease (PD) is a neurodegenerative disease characterized by the accumulation of aggregates composed of several proteins, of which alpha-synu- clein (ASYN) is the most prominent53. In the cell, ASYN can exist as a monomer, or assemble into species of different sizes, in a process believed to be relevant in the context of PD. These species include soluble oligomers and fibrils (e.g.64). In patient- derived samples, such as cerebrospinal fluid (CSF), detecting disease-relevant ASYN species is thought to be a relevant marker for PD diagnostic purposes and for moni- toring disease progression, since the measurement of ASYN levels alone has proven to lack diagnostic relevance55-57.

[0264] We explored the diagnostic potential of imaging ASYN assemblies in the CSF of PD patients versus controls (Supp. Table 1). A nanobody able to bind ASYN58was used because full-length immunoglobulins only provide relatively poor labeling due to their large size (Figure 17a, Supplementary Fig. 26). Different types of ASYN assemblies could be detected (Figure 15b) and PD patients had higher levels of oligomer-like structures (Figure 17c, d).

[0265] We classified the structures observed and noticed that the "very large" assemblies (>200 nm in length, >50 nm in width) were found at similar frequency in PD patients and controls (Figure 17e). The same was observed for assemblies in the 50 to 200 nm length range (Figure 17e). However, this was not the case for the smaller, oligomer- like assemblies. Some resembled strikingly polymorphic ASYN assemblies that have been recently described by Cryo-EM59-70, while others had an annular organization as observed in the past by negative stain transmission EM or Cryo-EM71'72. All oligomer- like species were significantly more abundant in PD CSF than in control samples (Figure 17f), and their cumulative analysis, which alleviates ambiguities due to imperfect clas- sification, resulted in a good discrimination of PD patients and age-matched controls (Figure 17g, h; for overviews of more ASYN objects see Extended Data Fig. 10). We conclude that the analysis of ASYN aggregates by ONE microscopy is a promising pro- cedure for PD diagnosis.

[0266] Finally, ONE can also analyze medical samples that have been fixed for prolonged (and uncontrolled) time periods, as observed for chemically fixed blood sera from COVID-19 patients, obtained commercially (Supplementary Fig. 27). Here we observed that the patient IgGs clustered in specific regions of the viral particles, whose detailed compo- sition could be targeted in the future.

[0267] For completeness, we performed an FRC analysis of the ASYN and virus samples, pre- sented in Supplementary Fig. 28. Multi-laboratory applications of ONE microscopy

[0268] An important issue for any new technology is its wide applicability, in multiple labora- tories. To test this issue, we collaborated with academic laboratories in Homburg (Ger- many), Wurzburg (Germany) and at MIT (USA), and with the industrial laboratory of microscope developer Leica Microsystems (Mannheim, Germany). We focused on GABAA receptors and tubulin, samples that were well described in the rest of the work (Supplementary Figures 29 to 33). We were able to show that ONE can be applied in different laboratories, with some of the experiments even surpassing our original ap- plications by either using larger expansion factors (MIT laboratory, 20x expanded im- munostained microtubules in cultured cells; Supplementary Fig. 31; post-expansion stained bassoon in x20 expanded mouse brain tissue, Supplementary Fig. 33) or by using faster scanning to allow volumetric zONE imaging in two color channels (Leica Microsystems laboratory; Supplementary Fig. 29).

[0269] Discussion

[0270] Here we show that a fluorescence microscopy procedure based on a combination of X10 ExM and radial and temporal fluorescence fluctuation analysis (SRRF) can provide sufficient spatial resolution to analyze single protein shapes. ONE was implemented in five laboratories on different microscopes. Therefore, ONE microscopy makes super- resolution imaging broadly available, in a fashion that has always been a primary goal of ExM73. Moreover, no special handling, unusual fluorophores or reagents are neces- sary. The ONE data processing is relatively fast because the SRRF procedure is per- formed in minutes (see Supplementary File 1 for details). The initial immunostaining and expansion procedures take, combined, 3-4 days, while imaging individual regions of interest only takes between 35 seconds and 2 minutes, depending on the number of color channels.

[0271] The ONE axial resolution surpasses that of confocal microscopy by one order of mag- nitude, owing to the lOx expansion factor. Further improvements of axial resolution could be introduced in the future, through total internal reflection fluorescence (TIRF), lattice light-sheet microscopy74, or multi-focus microscopy7. The only major limitation we see is that ONE cannot be applied to live samples, due to the ExM procedures. Therefore, the MINFLUX concept8'11'75'76is currently the only solution for live imaging in the very high-resolution domain. Nevertheless, future developments in ONE micros- copy are likely to enable 3D structural analysis of proteins, either purified or in cells and tissue samples, at resolutions approaching electron cryo-microscopy and tomog- raphy techniques, at room temperature and at a fraction of the cost. Developments envisaged include refined anchoring chemistries, gels that are homogeneous to sub- nanometer levels, as well as imaging automation, to enable the analysis of tens of thousands of particles in a time-efficient manner.

[0272] Overall, we conclude that the ONE technology provides a simple, robust and easily applied technique, bridging the gap to X-ray crystallography and electron microscopy- based technologies.

[0273] Extended Data Figure legends

[0274] Extended Data Fig. 1. A general overview of the ONE microscopy approach, a, Biological samples are linked to gel anchors, relying on Acryloyl-X, followed by X10 gel formation and homogenization, either by proteinase K additions or by autoclaving in alkaline buffers. Full expansion is achieved by repeated washes, and is followed by mounting gel portions in a specially designed chamber. In principle, one could image the samples using different super-resolution procedures. Techniques benefitting from bright samples, as STED or SIM, suffer due to the fluorophore dilution induced by the expansion procedure. Techniques requiring special buffers (e.g. SMLM) are negatively affected by the water environment. In contrast, technologies relying on fluorophore fluctuations profit from the expansion, as the fluorophores are spatially separated and can fluctuate independently13, b, Repeated imaging is performed (up to 3000 images), in any desired imaging system (confocal, epifluorescence, etc.), to detect signal fluc- tuations, which are then computed using through a plugin (ONE platform) based on the SRRF algorithm, before assembling the final super-resolved images.

[0275] Extended Data Fig. 2. A detailed view of the ONE procedure, a, Processing a stack of diffraction-limited images with SRRF, based on the analysis of a gradient of convergence of sub-pixels over a radiality stack, results in super-resolved images with resolutions varying between 50-70 nm. b, The ONE procedure adapts the SRRF algo- rithm to expanded gels, c-f, A detailed explanation of the analysis procedure, c, A sample was fixed and expanded using a 10-fold expansion protocol (X10). The sample was then imaged using a resonant scanner on a confocal microscope. The zoomed-in view indicates one bright spot, whose size in real space is limited by diffraction to ~200-300 nm, but represents a 10-fold smaller size in the pre-expansion space (see scale bars in the middle panels). Every pixel is then subjected to a 10-fold radiality magnification and is then subjected to the procedure explained in panels d-f, which provides the final, high-resolution image (right-most panel), d, Signal fluctuations are measured by imaging the sample repeatedly, using the resonant scanner (here at 8 kHz), e, A view of the overall signals, obtained by summing 20 of the fluctuating im- ages (raw in the left-most panel, background-subtracted in the middle panel), or by summing 1000 images, f, Each image from series obtained as in panel b is subjected to a temporal analysis of fluctuating fluorophores, based on radiality magnification6, thereby providing a super-resolved image whose level of detail becomes optimal after ~1500 frames.

[0276] Extended Data Fig. 3. Expansion microscopy results in a higher signal-to- noise ratio. Expansion microscopy, which separates proteins of interest and removes much of the other cellular components (e.g. lipids, metabolites) should result in a higher signal-to-noise ratio (SNR), a, To test this, we analyzed here the simplest pos- sible sample, consisting of Star635P-conjugated nanobodies on glass coverslips, or in expanded gels, using confocal microscopy, relying on analysis using a resonant scan- ner. b, The SNR of these samples increases by 2-fold, on average, after expansion. N = 30-24, P = 0.000001, Mann-Whitney Ranksum test.

[0277] Extended Data Fig. 4. An analysis of tubulin immunostainings. a & b, An anal- ysis of tubulin, following immunostainings relying on primary antibodies detected using Star635P-conjugated secondary nanobodies. While the overall signal distribution is similar to that obtained with secondary antibodies (Figure 12), one can observe often pairs of fluorescent spots in very close vicinity (marked by dotted circles in the cross section), which probably represent the two fluorophores on each nanobody. For a for- mal analysis of this issue on different nanobodies, see Figure 12 and Extended Data Fig. 5. c, Immunostainings relying on primary antibodies followed by secondary anti- bodies (upper panel) or by secondary nanobodies (lower panel), d, The graph shows the diameter of microtubules in when using secondary antibodies (left; N = 49 micro- tubule profiles) or secondary nanobodies (right; N = 101).

[0278] Extended Data Fig. 5. In-depth analysis of GFP-nanobody complexes, a, Dot blots to validate that each nanobody was binding specifically the TSR individually. Ni- trocellulose membranes were spotted with TSRs and bovine serum albumin, as control, and the spots were revealed with the respective nanobodies, using a fluorescence scanner (GE-Healthcare Al 600). b, An overview of an image showcasing nanobodies bound to their GFP target, c, An analysis of distances from STAR635P to Cy3 nano- bodies, in normal images or after mirroring one of the fluorescence channels, as a negative controls. The close-distance interval is largely removed by mirroring. N = 40- 40 TSRs. Performing this in samples lacking the GFP, in which the nanobodies are randomly distributed, results in no differences between the normal and mirrored dis- tributions. N= 40 / 40 images, d, Overview of the TSR using only two-color nanobody labeling (same as the one used in Figure 16c, d), along with two different examples. The sample is also labelled using NHS-ester fluorescein, and a small pixel size (0.48 nm) is used, to enable the optimal visualization of the TSRs. e, An analysis of the signal-to-noise ratio of the TSRs, obtained by measuring the noise levels in the vicinity of the nanobodies. The noise levels are normalized to 1, implying that the normalized signal of the respective nanobodies now provides directly the signal-to-noise ratio. N = 20-18, 12-14, and 17-11 measurements, P < 0.0001, Mann-Whitney test, f, A Fou- rier Ring Correlation (FRC) analysis of nanobody images, g, The best and average resolutions obtained per image, in the different color channels (N = 4 to 5 analyses for each), h, To approximate the apparent resolution of the system, we drew line scans across spots and measured the full width at half maximum (FWHM) in curve fits exe- cuted on the line scans. The graph plots the FWHM of 129, 135, and 132 fluorescein, Cy3 and STAR635P line scans. The values are significantly different between the color channels, p < 0.0001, Kruskal-Wallis test. The box plot shows the median, 25thper- centile and the range of values.

[0279] Extended Data Fig. 6. Further PSD examples, a, ONE imaging of PSDs, employing a resonant scanner and a final pixel size of 1 nm, to achieve a high resolution (same procedure and resolution as in Figure 14f). b, Examples of PSD95 stainings, after treat- ment with 1,6-hexanediol (Hex), as in Figure 14f. c, We averaged the PSD95 signals for both control and Hex-treated synapses (8 PSDs imaged in top views, for each treatment). The control shows a somewhat regular pattern, while the Hex treatment seems to perturb this.

[0280] Extended Data Fig. 7. A detailed analysis of the PSD. a, The PSD was im- munostained for PSD95, Homerl and Shank2, as in Figure 14, and images were taken at different heights along the Z-axis (zONE imaging). An overlay (summed image) is shown in the left panel, along with an analysis of the proteins at different Z levels, using a colormap that describes the positions along the Z axis (right panel), b, The distance between PSD95 spots was computed from images as in panel a, and was compared to that obtained from positioning the molecules randomly within the PSD95, N = 10 synapses, Friedman test followed by Dunn-Sidak, p=0.0001. c, The lateral distance between PSD95 spots and between PSD95 and Homerl or Shank2. The min- imal distance between each PSD95 spot and a Homerl / Shank2 spot is shown (meas- ured in the lateral plane, in 2D projections of the PSD). N = 10 synapses, from 2 independent experiments. While the distance between PSD95 spots has a non-random character, as indicated in panel b, the distances to Homerl or Shank2 spots are not different from randomized distributions (Dunn-Sidak tests, p>0.1), possibly also be- cause these two molecules are immunostained using antibodies, which causes the flu- orescence signals to scatter broadly, d, Confocal microscopy analysis of the PSDs, in non-expanded samples. In control conditions all three components analyzed here (PSD95, Homerl, Shank2) are well colocalized. The addition of 3% 1,6-hexanediol (Hex) causes the dispersion of Homerl (magenta), while 10% Hex also disperses Shank2 (blue). PSD95 remains largely unaffected by Hex. e, An analysis of the average PSD95 spot profile confirms this impression, N = 10-7-10 neurons, a set from 3 inde- pendent experiments, f, We analyzed the dispersion of Homerl (left) and Shank2 (right) away from the PSD95 spots. The signal present in synapses (near the PSD95 labeling, but not within the PSD) was analyzed, to determine the % that is not corre- lating to the PSD structure. The same samples were analyzed as in panel e.

[0281] Extended Data Fig. 8. ExM-STED (ExSTED) imaging of PSDs. a, Hippocampal cultures were immunostained for PSD95 and VGIutl, and were additionally labelled with NHS-ester fluorescein, after homogenization, b, A gallery of high-zoom ExM-STED views of synapses, with a focus on PSD95. Relatively large PSD domains are visible, as in most previous works in the literature, and unlike most of our ONE images, c, To determine if this is simply an issue of resolution, we aimed to generate ExM-STED-like images with ONE microscopy, by reducing its resolution. We employed an epifluores- cence microscope (as opposed to a rapidly scanning confocal in the panels dealing with PSD95 in Figure 16), and we used the temporal radiality pairwise product mean (TRPPM) option of analysis, which broadens the resulting spots. The results are very similar to ExM-STED images, demonstrating that the modular / domain appearance of the PSD95 stainings is a result of insufficient resolution, with a retiolum being evident only at very high resolution (under optimal ONE imaging).

[0282] Extended Data Fig. 9. ONE analysis of brain slices, a, Images of a 200 pm-thick rat brain section before (left) and after (right) expansion, relying on autoclaving for homogenization47. The scale bar does not take the expansion factor into consideration. The sections were labelled by using NHS-ester fluorescein incubations, b, Epifluores- cence images of expanded brain slices, focusing on Bassoon and Homerl as pre- and postsynaptic markers, respectively, c, Similar images, taken using the ONE procedure, d, Line scans executed over the areas indicated in panels b and c. As expected, far more details can be observed in ONE than in simple epifluorescence microscopy.

[0283] Extended Data Fig. 10. A gallery of ASYN object images from 7 PD patients and 7 controls. The images were obtained following the procedure indicated in Figure 15a. See Supp. Table 1 for details on the respective patients.

[0284] Supplementary Figure legends

[0285] Supplementary Fig. 1. ONE analysis and examples, a & b, Several views of the starting interface of the ONE software package. The examples show the intuitive soft- ware choices. See also the "Readme / Help" file of the software package, c, Examples of different potential artifacts that should be avoided in ONE imaging, d, Different potential choices in how to resolve ONE images. We suggest using the temporal radi- ality pairwise product mean (TRPPM) procedure for dim samples. This reduces the obtainable resolution, but follows much better the potential sample shape. For brightly labelled samples with direct labeling, the temporal radiality auto-cumulant (TRAC4) procedure provides the best resolution and SNR, indicating the positions of the indi- vidual fluorophores. Supplementary Fig. 2. Evaluating SRRF analysis performance using DNA ori- gami nanorulers, in non-expanded samples, a, Nanorulers with single Atto647N molecules (R SM) were generated by GATTAquant30carrying fluorophores on each end of DNA structures of 80, 60, 50, 30, 20 and 10 nm in length. They were then imaged using a confocal resonant scanner, without expansion procedures. The first panel shows confocal maximal intensity projections (MIPs) for each of the rulers. The second panel shows temporal radiality averaging (TRA) analysis overviews. White boxes indi- cate the magnified regions displayed in the third panel. The fourth panel shows a temporal radiality auto-correlations of fourth order (TRAC4) analysis, overlaid with the respective confocal MIPs. The remaining panels show different ruler examples, ac- quired at different starting pixel sizes, using either a hybrid detector (HyD) or an ava- lanche photodiode detector (APD), and analyzed in different SRRF modalities. This analysis is shown in the fifth panel for 50 nm pixel size, using a HyD and analyzed using TRAC4. The sixth and seventh panels show rulers acquired at 100 and 50 nm pixel sizes, using an APD and analyzed using TRAC4. The eighth panel shows rulers that were acquired at 100 nm pixel size and were analyzed using default SRRF settings (TRA). b, Magnified overviews of selected regions (indicated by blue rectangles) from each of the ruler exemplary images to the left, c, Signal-to-noise ratio (SNR) analysis of HyD and APD detectors, N = 25 and 30 for HyD and APD, respectively. Mann Whit- ney test, p = 0.004. d, Normalized line scans across the different ruler images, as indicated in the respective panels in (a), e, Apparent FWHM of the different rulers. N = 17, 17, 18, 17, 18 and 17 for 80, 60, 50, 30, 20, and 10 R SM, respectively. A Kruskal-Wallis test was applied, followed by Dunn's post hoc test, p < 0.0001.

[0286] Supplementary Fig. 3. The effect of frame number on SRRF analysis, a & b, 80 nm rulers were imaged at 100 and 50 nm pixel sizes, and were then analyzed with the default SRRF parameter (temporal radiality average, TRA), using varying frame counts (termed F in the figure), from 100 to 2000. c & d, The same procedure was repeated using temporal radiality auto-correlations (TRAC4) for rulers of 10 to 80 nm. The frame count does not affect the TRA analysis as much as it affects TRAC4. The TRA performance, which is the parameter reported in most publications, is far poorer than that of TRAC4, when sufficient frames are analyzed.

[0287] Supplementary Fig. 4. SNR effect on SRRF performance, a, The top panel shows an overview of 30 frame-MIPs of an 80 nm ruler, followed by MIPs of the same ruler that were subjected to 2-fold, 5-fold, 10-fold and 20-fold increase in noise. Noise was added artificially, using a Matlab routine. The initial SNR was 27.84 .The second panel shows TRAC4 analyses of the data. The third and fourth panels show a magnified region from the resonant scan MIPs, and their respective TRAC4 analysis results, b, The same analysis was performed on expanded GABAAR. Note that the receptor pore disappearing at x5 fold noise in TRAC4 resolved image. The nanoruler image is cor- rupted far more strongly by a 2-fold increase in noise than that of the GABAAR, owing to the substantially higher original SNR of the receptor image (76.72).

[0288] Supplementary Fig. 5. Technical scheme of the stabilization chamber used in this work. The exact measurements and materials for the stabilization chamber are included in the figure text. The 3D-printed gel cage patterning can be organized according to the user's preferred design. Only a suggested design is included here (many others work equally well). The exact design files can be obtained from the corresponding authors, to produce this chamber in any facility.

[0289] Supplementary Fig. 6. TSR gallery, a, An example of a TSR. The first panel shows a ONE image of a TSR, the middle panel shows a cartoon model that fits the imaged TSR, and the third panel shows an overlay of the ONE image and the model, b, A gallery of TSRs (upper panels) and a best guess of cartoon models overlaid over the TSR images (lower panels).

[0290] Supplementary Fig. 7. Bleaching properties of fluorescein, Cy3, and STAR635P. a, A representation of the structures of each of the used dyes, followed by a table of their properties. The molecule structures and properties were reproduced from measurements of commercial providers:1https: / / broadpharm.com / product / bp- 23900,2https: / / broadpharm.com / product / bp-22535, and3https: / / abberior.shop / ab- berior-STAR-635P. b, Normalized bleach curves from expanded specimens at 8000 Hz, and non-expanded specimens at 8000 Hz and 200 Hz.

[0291] Supplementary Fig. 8. ONE imaging of purified eGFP molecules, a, The first panel shows a ONE overview of eGFP molecules labelled with NHS-Ester STAR635P. The second panel shows a magnified area. The third panel shows the eGFP 1EMA PDB structure. The fourth panel shows a PDB / fluorescence overlay, b, A measurement of the apparent width and length of the molecules, from line scans as the examples shown in panel a, in blue and orange. A total of 17 single molecules were measured, c, A gallery of eGFP molecules.

[0292] Supplementary Fig. 9. Further ONE examples of immunoglobulin imaging, a, An overview of a field showing IgG antibodies labelled using NHS-fluorescein (left), along with a few zoom-in images of fluorescently-conjugated secondary IgG antibodies (right; Abberior Star635P conjugation shown in blue), b, Several examples of IgG an- tibodies imaged in different positions and perspectives, c, A gallery of the expected antibody shapes, obtained by convoluting a PDB IgG structure with a ONE point- spread-function, after revolving the IgG molecules in 3D space randomly. A few en- larged views are shown, along with a multitude of small-sized views, to explain how IgG molecules should appear when they are visualized in fluorescence in random ori- entations. The typical IgG views are similar to the modeled ones, d, Fluorescence (Abberior Star635P) and Coomassie SDS-PAGE gels indicating the size distribution of antibody fragments. A mouse monoclonal primary antibody was run on the gels, along the secondary antibody imaged in panel a. The gel was first imaged under a fluores- cence (Cy5 channel) and then total proteins were revealed with Coomassie brilliant blue staining. The results suggest that numerous small fragments are expected for both primary and secondary antibodies in the ONE images, not only full antibodies, due to impurities being present in the commercial antibody samples, e-f, An overview of IgA molecules, g-h, A similar overview of IgM molecules. The antibody structures are shown using Pymol representations from PDB structures 1HZH, 1IGA, and 2RCJ.

[0293] Supplementary Fig. 10. GABAA receptor and otoferlin galleries, a, An overview of images of GABAA receptors, b, The images display GABAA receptors in different 3D positions. The positional indications are best guesses performed by an experienced investigator, c, Overview images of otoferlin (right panel), and blank buffer as a control (left panel), d, Otoferlin labelled with NHS-ester fluorescein ONE images in different 3D positions, e, Otoferlin labelled with NHS-ester STAR635P ONE images.

[0294] Supplementary Fig. 11. Calmodulin gallery, a, An overview of calmodulin ONE acquisitions in the presence and absence of calcium. This molecule was expressed and purified as a chimera containing mEGFP. The compact signal associated to the GFP molecule, as observed already in the TSR images in Figure 12, has a limited contribu- tion to the overall size of the molecule, b, Exemplary zoomed calmodulin ONE images. The asterisk denotes the best guess of GFP molecule bound to calmodulin.

[0295] Supplementary Fig. 12. GABAAR nanobody labeling, a, Confocal images of ex- panded GABAAR labelled with anti-GABAAR nanobodies (NBs) conjugated to STAR635P. b, Confocal images of expanded GABAAR mixed with anti-eGFP nanobodies, which only induce little non-specific background, c, Magnified regions of single receptor either labelled with anti- GABAAR or anti-eGFP NBs. d, A gallery of ONE images showing GABAAR in white and anti-GABAAR NBs in red.

[0296] Supplementary Fig. 13. Super-imposition of ONE microscopy images and Cryo-EM data, a, A cartoon view of the 50JM GABAAR / NB PDB. The red dots repre- sent the 2 fluorophores on each nanobody, b, Cryo-EM images of representative 2D- classes of the GABAAR / NB complexes, derived from the same samples as used for ExM. c, The first panel shows a ONE image of GABAAR / NB. The second panel shows a mag- nified region of a single receptor. The third panel shows a Cryo-EM / ONE overlay.

[0297] Supplementary Fig. 14. Drift compensation, a, A resonant confocal X10 image of otoferlin molecules at the start of a 1500-frame time series recording (first panel) and at its end (second panel). The third panel shows a maximum intensity projection of the resonant confocal scan. The blue arrow indicates the direction of drift. The fourth panel shows ONE processing without drift correction. A streak artefact is evident as a result, b, Applying drift correction, using the SRRF software, to the same acquisi- tion and maximum intensity projection yields an image (first panel) similar to the first image in (a). The second panel shows the result of the ONE processing with drift correction application. The last set of panels show magnified regions of otoferlin mol- ecules. An otoferlin AlphaFold cartoon is presented for comparison (not drawn to scale). In panel 3, the ONE image is overlaid with its counterpart from the same da- taset, processed without drift correction (blue).

[0298] Supplementary Fig. 15. ONE imaging of in vitro assembled microtubules, a, The upper panel shows in i / / Zro-assembled microtubules that are stably fixed with 1- 2.5% glutaraldehyde (GA). GA above 0.2% interfered with anchoring into the gels. The lower panel shows less stable microtubules, fixed with 8% PFA. These microtu- bules are deformed and tend to depolymerize, but this fixation does allow a reasonable degree of anchoring to the gels, b, ONE images of microtubules fixed with 8% PFA and expanded and labelled using NHS-ester fluorescein, c, A magnified region, d, An averaged analysis of side views of microtubule segments. The top panel shows an ideal image, obtained by convoluting the PDB structure of a microtubule segment (3J2U) with a fluorescent PSF, in which every amino acid is labelled fluorescently. The second panel shows a realistic model, in which sparser labelling is considered, and in which different microtubule segments are overlaid with slight tilt angle differences (up to 5°). The third panel shows an averaged processed ONE image of 175 partially depolymer- ized tubulin segments. The graph shows the respective line scans across each of the panels, e, A gallery of alpha-beta tubulin dimers that were left unpolymerized, f, The first panel shows a tubulin dimer reconstructed from 105 dimers, following the same procedure as for the GABAARS, shown in Figure 14. The second panel shows a 1TUB PDB cartoon structure, and the third panel shows a ribbon display of the molecules, for comparison.

[0299] Supplementary Fig. 16. A confocal analysis of synapses after M0CD treat- ments. a, Confocal images of hippocampal cultures immunostained for the three syn- aptic markers employed in Figure 14a-c (Sytl, vGlutl and PSD95), relying on the same staining protocol as in Figure 14a-c. b, The panels show a magnified region. The cul- ture morphology and synapse distribution are similar with or without MPCD treatments. Supplementary Fig. 17. Sytl count analysis, a, The first panel shows a ONE overview image of Sytl, VGIuTl, and PSD95 channels. The middle panel shows the Sytl channel alone, with two selected regions indicating signal from isolated antibod- ies. The two regions are magnified and displayed on top of each other in the third panel, b, The first graph shows the mean grey value of isolated spots in control and MPCD treated neurons. The second graph shows the mean grey value of vesicles. N = 22-19, 2 independent experiments, Mann-Whitney test, p = 0.6507 for the first graph and p = 0.8494 for the second graph, c, A graph showing the number of Sytl anti- bodies per vesicle. Sytl antibody numbers were estimated by dividing the vesicle in- tensity value by single AB intensity value. Mann-Whitney test; p = 0.8937.

[0300] Supplementary Fig. 18. Intensity analysis. Specific nanobodies, which detect GABAARS, are compared to non-specifically bound eGFP nanobodies and to background noise, a, A set of images that shows GABAARS bound to their respective nanobodies, GABAARS + eGFP NBs, and a blank control, b, Fluorescence intensities were analyzed across the different conditions. N = 37, 28, and 33 images for GABAAR NB, eGFP NB and blank, respectively, from 2 independent experiments.

[0301] Supplementary Fig. 19. PSD95 model, a, To complement the distance analysis presented in Extended Data Fig. 7b, we analyzed the PSD95 distribution using a spot averaging procedure similar to a Ripley curve profile. To explain this analysis in more detail, we modeled it here. The top row of panels shows PSD-like spots, placed in a perfectly regular arrangement (left), with positions varying by 20 or 50% from perfect regularity (middle), or placed randomly (right). The bottom rows of panels show aver- age spots, obtained by overlaying the areas surrounding each of the individual spots in the model arrangements from the top panels. This procedure results in arrange- ments in which the central spot is surrounded by increasingly weak spots, with virtually no regular spots around it in the right-most panel, b, Lines were drawn from the center of each spot in the bottom panels in panel a, in all directions, and were then averaged. The average line going from the center of a spot to the periphery shows a prominent peak if the arrangement is regular, since the neighboring spots are always present at a set distance, and thus provide a visible intensity peak. The less regular the arrange- ment is, the less clear the second peak becomes. It disappears completely when the spot positions are fully random.

[0302] Supplementary Fig. 20. Distribution of post-synaptic elements, a, A ONE over- view of a hippocampal culture immunostained for GluA2, GluN2b and PSD95. b, A cartoon that color-codes the analysis presented in the lower two graphs. The lateral span of the GluN2b, GluA2 and PSD95 signals is presented in green, blue and red. The second graph shows the distance between the center of the GluN2b cluster and the GluA2 periphery. This implies that the GluN2b cluster is positioned relatively close to the center of the GluA2 distribution, since the value here is very close to the half of the GluA2 span. N = 8 post-synapses analyzed, 2 experimental replicates. Kruskal- Wallis test was applied, p = 0.0049 **. Supplementary Fig. 21. ONE analysis of PSD95 labelled post-expansion. Neu- ronal cultures were fixed, expanded using the XIOht protocol and labelled after expan- sion using a PSD95 NB conjugated to Atto488. a, For comparison purposes, pre-ex- pansion labelled PSD95 images are reproduced from Figure 16f (top panel) and Ex- tended Data Fig. 6a (the other 3 panels) of the original manuscript, b, Post-expan- sion labelled PSD95 examples, c, Exemplary line scans over pre- and post-labelled PSD95 show similar cluster spacing, d, Spot FWHM and spot distances measurements showed similar values. N = 140, and 113 measurements spot FWHM, and 402, and 172 for spot distances, for pre- and post-expansion, respectively. Mann Whitney test, non-significant p > 0.05.

[0303] Supplementary Fig. 22. STED, ExSTED and ONE comparison, a, The synaptic proteins Bassoon, Homerl and PSD95 were imaged using STED, ExSTED (X10 ex- pansion combined with STED), and ONE microscopy, b, The same procedure was ap- plied to tubulin.

[0304] Supplementary Fig. 23. Detailed analysis of Fourier ring correlation, a, FRC analysis of ONE images collected with a pixel size of 0.98 nm. The first panel row shows ONE images of the different specimens. The second row shows the correspond- ing FRC maps. The third row shows ONE images overlaid over FRC maps, using a screen-blend mode. The fourth and fifth rows show magnified views, b, A graph plot- ting the minimal FRCs in nm. c, A graph plotting the average FRCs in nm. Please note that all the labelled targets reside in the "bluest" regions of the map, indicating minimal FRCs that correspond to high resolution. N = 7, 8, 12, 7, 7, 7, and 7 for eGFP, tubulin dimer, actin, calmodulin, PSD95, Homer 1, and Shank2, respectively, d & e, FRC anal- ysis of ONE images achieved with a pixel size at 0.98, 0.48, and 0,24 nm for GABAAR and otoferlin, respectively, f, The graphs shows minimal and average FRCs in nm for GABAAR ONE images. N = 8, 6, and 9 for 0.98, 0.48, and 0.24 nm images, respectively, g, The graphs shows minimal and average FRCs in nm for otoferlin ONE images N = 10, 10, and 9 for 0.98, 0.48, and 0.24 nm images, respectively. All experimental sets were performed with at least 2 replicates.

[0305] Supplementary Fig. 24. Intra-molecular measurements, a, GABAAR ONE im- ages acquired with 0.98, 0.48, and 0.24 nm pixel sizes, for the same region, b, GABAAR magnified examples from the first image in the panel above, c, One particular GABAAR molecule displayed at different resolutions, d, Equally-scaled otoferlin molecules ac- quired at different resolutions, e, ONE images of GABAAR and otoferlin at 0.24 nm overlaid with their respective PDBs. f, The graphs show 2 exemplary line scans for peptide segments in GABAAR and otoferlin. g, A graph showing peak-to-peak distances in Angstrom. N = 30 for GABAAR, and 30 for otoferlin, 10 independent experiments for GABAARS and 4 independent experiments for otoferlin.

[0306] Supplementary Fig. 25. Expansion precision evaluation, a, A direct compassion between single-molecule ONE images and their respective PDB / AlphaFold models. The purple line indicate the line scan used to measure the molecule dimension indicated in the first graph, b, The upper graph shows measurements of molecule dimensions, in nm. The horizontal purple line indicates the expected value, obtained from measure- ments of PDB structures (for all molecules except otoferlin), or AlphaFold predictions (for otoferlin). The lower graph shows the variability of these measurements, in the form of the coefficient of variance. N = 34, 17, 192, 75, 10, 10, 14, 8, and 18 for NB, eGFP, actin, tubulin dimer, GABAAR, otoferlin, IgG, IgA, and IgM; at least 2 experi- mental replicates were carried for all experiments. Paired t tests were carried out to determine whether the measured values are different from the values predicted by the PDBs; the respective p values are reported above the plots.

[0307] Supplementary Fig. 26. The nanobody imaging of ASYN objects is specific and is not easily reproduced by antibodies, a, Low-resolution images of CSF- containing samples, or blanks (clean, BSA-coated coverslips). Only a few dim spots, presumably representing single nanobodies, are seen in the blanks, b, Quantification of the signal intensity, as a sum across all image pixels. N = 7-9; Mann-Whitney test, P= 0.0002. c, Individual examples of oligomers immunolabelled with nanobodies (top) or antibodies (bottom), d, Averages of ASYN objects from individual patients, immu- nolabeled with nanobodies or antibodies, e, An analysis of the average object size in antibody-labelled samples, as in Figure 15. N = 2 patients for each condition; the graph shows mean ± range of values. Nanobodies reveal differences between patients, at object sizes of only a few nm. Antibodies have difficulties in this direction, as their large size causes a lower-fidelity labelling, and as their sizes obscure the actual sizes of small objects.

[0308] Supplementary Fig. 27. ONE analysis of SARS-CoV-2 viral particles, a, ONE overview of a sample containing SARS-CoV-2 viral particles immunostained against Spike Protein SI. b, More detailed views of two particles, indicating the Spike Protein SI and the native IgG molecules from the serum of the patients. Interestingly, a do- main-like structure is observed, which is presumably induced by the native IgGs gath- ering the spike proteins together, by the dual binding capacity of the IgG molecules. Supplementary Fig. 28. FRC Analysis of SARS-CoV-2 and ASYN aggregates, a, FRC analysis of ONE images achieved at 0.98 and 0.78 nm per pixel for SARS-CoV- 2 and ASYN, respectively. The first panel shows ONE images of the two specimens. The second panel shows the corresponding FRC maps. The third panel shows ONE images overlaid with FRC maps, using screen-blend mode. The fourth panel shows magnified overlays. Please note that the FRC scale for SARS-CoV-2 ranges between 2.0 and 2.9 nm, with red color in this dataset still indicating very good FRC values, b, The graphs plot the minimal and average FRCs in nm. N = 10, and 8 for SARS-CoV-2 and ASYN, respectively, 2 experimental replicates.

[0309] Supplementary Figures 29. ONE microscopy applied at the confocal head- quarters of Leica Microsystems and at the Center for Integrative Physiology and Molecular Medicine (CIPMM) of the Saarland University (UdS). As GABAAR were systematically investigated in this study, we chose them as a reference to evaluate the applicability of ONE technique at different laboratories using different systems, a, Using a STELLARIS 8 microscope at Leica Microsystems, we present a snapshot of a plane from a 5-dimension x,y,z,c,t image of GABAAR+NB. b, The first panel shows a depth projection of a zONE stack. The second panel shows a set of GABAARS that were magnified. The optical sectioning of the first example is displayed in the rightmost panel, c, GABAARS were also successfully imaged at CIPMM, Saarland University (UdS), as shown in full 3 full-scale overviews and in their respective magni- fied regions. Several microscopes were used at the CIPMM, which are presented in the next figure.

[0310] Supplementary Figures 30. GABAARS could be imaged with different micro- scopes. Acquisition settings were matched among different systems to the level that each system allowed. The highest achievable speeds were used for each system. This was systematically characterized (data not shown, but can be presented upon re- quest). a, Images from the first panel show GABAAR ONE images acquired from differ- ent microscopes. As the imaging systems were pushed to their speed limit, background noise was substantially higher on older models. The second panel shows ONE images with background subtraction. The third panel shows a magnified receptor example, b, Magnified regions showing the noise readings of each of the used microscopes, c, A graph showing the achievable signal-to-noise ratio (SNR), as well as the SNR normal- ized to acquisition speed. Not surprisingly, higher SNRs yielded better ONE images. N = 22, 22, 22, 23, and 23 for LSM780, LSM880, Abberior STED, SP5, and STELLAIRS 8, respectively, Kruskal-Wallis test was applied p < 0.0001 ****.

[0311] Supplementary Figures 31. ONE microscopy applied at the MIT, Cambridge, USA. Pre-expansion labelled tubulin specimens were expanded X10 and X20 and were imaged on a STELLARIS 8 system. The X20 gel recipe was modified from the ExR protocol49, as follows. The first expansion gel components are 17.25% (w / v) sodium acrylate, 5% (w / v) acrylamide, ImM BIS, lx PBS, 0.05% (w / w) APS / TEMED. The re- embedding gel is composed of 10% acrylamide, 2.5 mM BIS, water, 0.05% (w / w) APS / TEMED. The second expansion gel is composed of 17.25% (w / v) sodium acrylate, 5% (w / v) acrylamide, ImM BIS, lx PBS, 0.05% (w / w) APS / TEMED. a, An overview of tubulin from X10 and X20. b, The upper panel shows an X10 confocal image, with its respective ONE image in the lower panel, c, The first upper two panels show X20 confocal images of tubulin. Their respective ONE images are shown in the lower panel. Note the significantly dimmer images, as the expansion factor gets higher. The third panel shows a magnified region of X20 ONE, followed by an overlay of con- focal and ONE images. The graph shows normalized line scans across the tubulin width. The cyan curve shows the average line scan and the red curve is a fit using a double Gaussian formula. N = 26 line scans, d, A magnified region of X20 ONE image is shown. Another portion was magnified and displayed in a white box. The dotted pale-yellow lines are an estimation of the tubulin structure.

[0312] Supplementary Figures 32. ONE microscopy applied at the University of Wurzburg, Germany. Specimens were expanded X5.8 and were post-expansion la- belled for tubulin, before imaging using an LSM900 microscope, a, The upper panels show a confocal overview and the lower panels show the respective ONE images, b, The upper panel shows an X5.8 confocal image and its respective ONE image in the lower panel, c, The first image is a magnified region of X5.8 ONE followed by an overlay of a confocal and a ONE image. The graph shows normalized line scans across tubulin width. The cyan curve shows the average line scan and the red curve shows the re- spective fit. N = 20 line scans.

[0313] Supplementary Figures 33. Post-expansion bassoon labeling ONE im- ages. Tissue sections were expanded and then labelled against bassoon following the expansion revealing (ExR) protocol49at the MIT, Cambridge, USA. a, An ExR20 (X20 expansion) confocal overview imaged with a x40 objective, b, Three different exem- plary ONE images of bassoon using a x63 objective. The first image is a resonant scan MIP of 20 frames, followed by a ONE image, and an overlay with its respective confocal image. The white square indicates the magnified region to the right, c, Similar to (b) but using a xlOO objective.

[0314] Supplementary Tables

[0315] Supplementary Table 1. Patient details.

[0316] ID Sex Age Diagno- sis

[0317] 1407 m 82 PD

[0318] Average ages: 76.7 ± 2.3 years (PD), 72.0 ± 2.9 years (controls); no significant dif- ference (Mann-Whitney Ranksum test).

[0319] PD: Parkinson's disease. CDB: Corticobasal degeneration. PNP: Peripheral neuropa- thy. PSP: Progressive supranuclear palsy. RLS: Restless legs syndrome.

[0320] Supplementary Table 2. Image format and analysis technical information.

[0321] Figure Panel Micro- Objec- Num- Reso- Pixel Camera SRRF zONE scope tive ber of nant size settings analy- step frames scanner (nm) sis size* fre- (when quency applied) b Olympus 100x 2000 n.a. 1 25 ms ex- TRAC4

[0322] ZOOM TIRF 1.49 NA posure,

[0323] 300 EM Gain b Ex-

[0324] STED c TCS SP5 lOOx 1500 8 kHz 0.98 n.a. TRPPM,

[0325] STED 1.4 NA TRAC4 d e-f TCS SP5 lOOx 1500 8 kHz 0.48 n.a. TRAC4

[0326] STED 1.4 NA 2 a-d f TCS SP5 lOOx Up to 8 kHz 0.98 n.a. TRAC4

[0327] STED 1.4 NA 4000 h i Leica TCS 63x Up to 24 kHz 0.98, n.a TRAC4 0.05

[0328] SP8 1.41 NA 400 0.48 pm

[0329] Lightning k & m

[0330] 3 a TCS SP5 lOOx 2000 8 kHz 0.98 n.a. TRPPM, -

[0331] STED 1.4 NA TRAC4

[0332] 4 b & c

[0333] 5 a & b Olympus lOOx 2000 n.a. 1 30 ms ex- TRPPM

[0334] TIRF 1.49 NA posure,

[0335] 300 EM Gain d e Olympus lOOx 2000 n.a. 1 30 ms ex- TRPPM

[0336] TIRF 1.49 NA posure,

[0337] 300 EM Gain f

[0338] 6 a & b TCS SP5 lOOx 1500 8 kHz 0.98 n.a. TRPPM, -

[0339] STED 1.4 NA TRAC4

[0340] 1 b TCS SP5 lOOx 1500 8 kHz 0.98 n.a. TRAC4

[0341] STED 1.4 NA 2 c & f

[0342] 4 a TCS SP5 100x 1500 8 kHz 0.48 n.a. TRAC4

[0343] STED 1.4 NA b & e

[0344] 5 b, d, & TCS SP5 lOOx 1500 8 kHz 0.98, n.a. TRAC4 f STED 1.4 NA 0.48

[0345] 6 a

[0346] 7 a Leica TCS 63x 250 12 kHz 8c 0.96 n.a. TRAC4 0.05

[0347] SP8 1.41 NA 24 kHz pm

[0348] Lightning

[0349] 8 a & b c Olympus lOOx 2000 n.a. 1 35 ms ex- TRPPM

[0350] TIRF 1.49 NA posure,

[0351] 300 EM

[0352] Gain

[0353] 9 c

[0354] 10 - TCS SP5 lOOx 1500 8 kHz 0.98 n.a. TRAC4, -

[0355] STED 1.4 NA TRPPM

[0356] Sup. Panel Micro- Objec- Num- Reso- Pixel Camera SRRF fig- scope tive ber of nant size settings analy- ures frames scanner (nm) sis fre- quency

[0357] 1 d TCS SP5 lOOx 2000 8 kHz 0.98 n.a. TRA,

[0358] STED 1.4 NA TRPPM,

[0359] TRAC2 & 4

[0360] 2 a & b

[0361] 3 a TCS SP5 lOOx 2000 8 kHz 9.8 n.a. TRA

[0362] STED 1.4 NA b c TCS SP5 lOOx 2000 8 kHz 9.8 n.a. TRAC4

[0363] STED 1.4 NA d a TCS SP5 lOOx 2000 8 kHz 4.9 n.a. TRAC4

[0364] STED 1.4 NA b a TCS SP5 lOOx 2000 8 kHz 0.98 n.a. TRA

[0365] STED 1.4 NA a & c a, b, f, TCS SP5 lOOx Up to 8 kHz 0.73 n.a. TRPPM

[0366] & h STED 1.4 NA 2000 a & b c & d TCS SP5 lOOx Up to 8 kHz 0.73, n.a. TRAC4

[0367] STED 1.4 NA 3000 0.36 a & b d TCS SP5 lOOx 1500 8 kHz 0.73 n.a. TRAC4

[0368] STED 1.4 NA c a & b TCS SP5 lOOx 2000 8 kHz 0.98 n.a. TRAC4

[0369] STED 1.4 NA b, c, & e a Olympus lOOx 2000 n.a. 1 30 ms ex- TRPPM

[0370] TIRF 1.49 NA posure,

[0371] 300 EM

[0372] Gain

[0373] Olympus lOOx 2000 n.a. 1 30 ms ex- TRPPM

[0374] TIRF 1.49 NA posure, 300 EM

[0375] Gain a & b a & b Abberior 100x n.a. 2 n.a. n.a.

[0376] Ex- STED 1.4 NA

[0377] STED a & b

[0378] ONE a, d, & TCS SP5 lOOx Up to 8 kHz 0.73 n.a. TRAC4 e STED 1.4 NA 4000 a-e a TCS SP5 lOOx Up to 8 kHz 0.98 n.a. TRAC4

[0379] STED 1.4 NA 4000 c a TCS SP5 lOOx Up to 8 kHz 0.98, n.a. TRAC4

[0380] STED 1.4 NA 4000 0.73,

[0381] 0.36 a b TCS SP5 lOOx 1500 8 kHz 1.58, n.a. TRAC4

[0382] STED 1.4 NA 0.98 a a & b STELLARI lOOx Up to 16 kHz 0.73 n.a. TRAC4 0.05

[0383] S 8 1.49 NA 2000 pm c a LSM780 63x 1.4 Up to 1.2 kHz 0.67 n.a. TRAC4

[0384] NA 2000 a a Abberior 100x Up to 1.4 kHz 0.73 n.a. TRAC4

[0385] STED 1.4 NA 2000 a a STELLARI 100x Up to 16 kHz 0.73 n.a. TRAC4

[0386] S 8 1.49 NA 2000

[0387] 31 b c & d STELLARI 100x 2000 24 kHz 0.49 n.a. TAC2,

[0388] S 8 1.4 NA TRAC4

[0389] 32 a & b

[0390] 33 b & c STELLARI 100x 2000 24 kHz 0.49

[0391] S 8 1.4 NA

[0392] *Piezo stage step size (not corrected for expansion factor).

[0393] **n.a. not applicable.

[0394] Supplementary File 1.

[0395] ONE Platform plugin manual. This file consists of a series of screen views explain- ing how the one plugin functions. Please follow the instructions in the respective im- ages.

[0396] Supplementary Discussion

[0397] Resolution. As presented in the main text, the ONE resolution enhancement relates almost exclusively to the lateral (XY) plane. Resolution along the Z axis depends on the expansion factor of the gel, being equivalent to the axial resolution of the confocal microscope used, divided by the expansion factor. This results in a difference of more than 20-fold between the axial and the lateral resolution, which will have significant effects on the image quality. This situation parallels conventional transmission electron microscopy (TEM), in which the thickness of the specimen limits the axial resolution to a similar 20- to 40-fold above the lateral resolution.

[0398] This situation implies that the optimal samples for ONE imaging would have a limited number of objects within the axial imaging volume of 40-60 nm (pre-expansion; vol- ume calculated for a conventional confocal microscope and a 10-15x expansion factor). Denser structures will cause a signal overlap that will confuse the identification of in- dividual structures. The use of purified proteins, which can be diluted to the desired signal density, is an optimal application for ONE microscopy, since the dilution factor avoids the potential issues with axial resolution. The axial resolution problem is espe- cially evident for microtubules, whose thickness is not sufficiently large, under lOx expansion, to avoid imaging the entire microtubule structure within one confocal vol- ume. Therefore, the entire "tube" of the microtubule appears as a band of fluorescence in the images shown in Figure 12. When the expansion factor is raised beyond lOx, this is no longer a problem, and the sides of the microtubule become evident.

[0399] We have not encountered any issues relating to the sample density in the lateral (XY) plane: the shape of individual proteins is maintained well, and all measurements we performed provided results within the expected boundaries.

[0400] Sample anchoring into the gel. We are currently relying on NHS-ester chemistry to anchor proteins into the gels, using the well-established chemical Acryloyl-X. This molecule reacts to amine groups on lysines and on the N-termini of the proteins in the sample. As lysines make up ~5% of all amino acids in proteins, most proteins should have sufficient anchor points for accurate gel anchoring. The only problem we can envision is the fact that aldehyde fixatives also modify amine groups. If gel anchoring appears faulty in specific samples, possibly due to excessive fixation, we suggest using an epitope retrieval strategy, in which the sample is heated to 95°C in basic buffers (pH 8-9). This strategy should eliminate some of the fixative effects, and should enable accurate gel anchoring. Performing the anchoring in basic buffers, overnight, should also assist with this issue.

[0401] Homogenization. The heat-based homogenization is optimal for retaining fluoro- phores already present in the samples (pre-expansion labeling), since it breaks the proteins, but it does not proceed, in the version we optimized, to the removal of every amino acid. At the same time, it does not rely on the diffusion of an enzyme deep into tissues, so it is optimal for these preparations. In contrast, the proteinase K presumably removes all amino acids that are not anchored into the gel. This approach is optimal for single proteins, since the fluorophore positions become quite precise, being always near the anchor points. However, proteinase K diffusion in thick tissues is poor, and therefore this approach is not suited for tissue slices of over ~10 pm.

[0402] SRRF performance. The initial implementation of SRRF resulted in a 50-70 nm res- olution6, leading to the impression that this is the best achievable resolution for this technique, as it is implied by some subsequent works (e.g.32). This is not the case, as demonstrated in our work on nanorulers (Supplementary Figs. 2-4). The name SRRF serves as an umbrella term for a number of different analyses, including the temporal radial ity average (TRA) and temporal radiality auto-correlation (TRAC)6. The latter method is a higher-order statistical analysis (following the procedures initially introduced for SOFI26, whose contrast, accuracy and final resolution are substantially higher than those of the TRA method. The TRA analysis does not consider higher-order temporal correlations, which makes it comfortable to use with limited numbers of frames {e.g. 100-300 frames), thus rendering it a method of choice for live-cell SRRF5-27. TRA is heavily dependent on the distance between the fluorophores, and performs best when the different fluorescent objects are separated by more than 70% of the full width at half maximum (FWHM) of the point-spread-function (PSF32). This implies that this procedure is not intended to produce a very high resolution, unlike the TRAC analyses. These analyses do produce better resolutions, but require larger numbers of frames for optimal performance, something that does not seem to be clear in the literature, since all SRRF implementations are often performed with as few as 100 frames. Nevertheless, the optimal resolution obtained with TRAC analyses can be pushed towards 20 nm, and maybe even beyond this value, under ideal imaging con- ditions (Supplementary Figs. 2-4). We therefore conclude that SRRF should not be considered to be limited to 50-70 nm resolutions, as explained in the Supplementary Notes of the original SRRF publication6.

[0403] As for most other super-resolution approaches, the pixel size limits the resolution to a value of approximately its double52. This limitation can be overcome, as indicated in Supplementary Fig. 2, by reducing the initial pixel size. This, however, will result in a lower signal-to-noise ratio (SNR), which is an essential parameter for all fluctuation- based analyses. Even when applied to low numbers of frames, SRRF provided excellent images when the signal-to-noise ratio surpassed 10-156'32. Below these values, SRRF will perform more poorly than many other related methods, as MUSICAL or ESI32, implying that users should consider carefully the noise levels of their images, as ex- plained in Supplementary Fig. 4.

[0404] The ONE procedure is designed to alleviate two of the main problems of the TRAC analysis, the fluorophore distance and the SNR. First, the distance between the fluor- ophores increases in all dimensions, leading to their dilution by the third power of the expansion factor. Second, the SNR increases profoundly (Extended Data Fig. 3). This is an important side effect of removing all cellular materials that are not embedded into the gels.

[0405] The remaining problem, that of acquiring sufficient frames for optimal performance, depends on 1) sample stability, and 2) fluorophore bleaching. The solutions to these issues come in the form of an improved gel-holding chamber (Supplementary Fig. 5) and of rapid resonant scanning, which reduces fluorophore bleaching (Supple- mentary Fig. 7). The latter effect is known from other super-resolution fields, as STED77and is probably due to the fact that rapid scanning lowers the light dose re- ceived continuously by every fluorophore, thereby reducing the possibility of excessive excitation and damage. Materials and Methods

[0406] Nanorulers. Custom-designed linear nanorulers of varying length (80, 60, 50, 30, 20, and 10 nm), carrying one Atto647N molecule on each end, were purchased from GAT- TAquant GmbH, Grafelfing, Germany.

[0407] Conventional cell cultures. Tubulin immunostaining was performed in the U2OS cell line, obtained from Cell Lines Service (CLS, Eppelheim, Germany). The cells were grown in a humidified incubator (5% CO2, 37°C), in Dulbecco's Modified Eagle Medium (DMEM #D5671, Merck, Darmstadt, Germany), with the addition of 10% FCS (fetal calf serum, #S0615, Merck) and 4 mM glutamine (#25030-024, ThermoFisher Scien- tific, Waltham, USA), with an antibiotic mixture added at 1% (penicillin / streptomycin (ThermoFisher Scientific). For imaging purposes, cells were grown overnight on poly- L-lysine-coated coverslips (#P2658, Merck).

[0408] Hippocampal cultured neurons. Animals (Wistar rats, P0 to Pl) were treated according to the regulations of the local authority, the Lower Saxony State Office for Consumer Protection and Food Safety (Niedersachsisches Landesamt fur Verbraucherschutz und Lebensmittelsicherheit), under the license Tbtungsversuch T09 / 08. In brief, the hippo- campi were dissected from the brains, were washed with Hank's Balanced Salt Solution (HBSS, #14175-053, Invitrogen, Waltham, MA, USA), before being incubated under slow rotation in a digestion solution containing 15 U / ml papain (#LS003126, Worthing- ton, Lakewood, USA), with 1 mM CaCh (#A862982745, Merck), 0.5 mM EDTA and 0.5 mg / ml L-cysteine (#30090, Merck), in DMEM. This procedure is performed for 1 hour at 37°C, before enzyme inactivation with a buffer containing 10% FCS and 5 mg / ml bovine serum albumin (BSA, #A1391, Applichem, Darmstadt, Germany) in DMEM. The inactivation solution is replaced after 15 minutes with the growth medium, containing 10% horse serum (#S900-500, VWR International GmbH, Darmstadt, Germany), 1.8 mM glutamine and 0.6 mg / ml glucose in MEM (#51200046, ThermoFisher Scientific), which is used to wash the hippocampi repeatedly. The neurons are then isolated by trituration using a glass pipette, and are sedimented by centrifugation at 800 rpm (8 minutes). The cells are then resuspended in the same medium and are seeded on PLL- coated coverslips, for several hours, before replacing the buffer with Neurobasal-A culture medium (#10888-022, ThermoFisher Scientific), containing 0.2% B27-supple- ment (#17504-044; ThermoFisher Scientific) and 2 mM GlutaMAX (#35050-038, Ther- moFisher Scientific). The neurons are then maintained in a humidified incubator (5% CO2, 37°C) for at least 14 days before usage.

[0409] Brain slices. We dissected rat brains from P0-P1 rat pups (Wistar), as above. The brains were then fixed with 4% PFA (#30525894, Merck) in PBS, for 20 hours. The fixed brains were then placed in agarose (4% solution, #9012366, VWR Life Science, Han- nover, Germany), before cutting to the desired thickness (100-200 pm) using a vi- bratome.

[0410] Patients. Patients were in treatment at Paracelsus Elena Klinik, Kassel, Germany. They had been diagnosed with Parkinson's disease according to standard criteria78-80. Neu- rological control patients had been diagnosed with a variety of non-neurodegenerative disorders. For a detailed presentation of patients, their ages and diagnoses, see Sup- plementary Table 1. The informed consent of all of the participants was obtained at the Paracelsus Elena Klinik, following the principles of the Declaration of Helsinki.

[0411] CSF samples. CSF samples were collected at the Paracelsus Elena Klinik, Kassel, Ger- many, following identical standard operating procedures (SOPs). CSF was gained by lumbar puncture in the morning with the patients fasting and in sitting position. The CSF was processed by centrifugation at 2000 x g for 10 minutes at room temperature and aliquots of supernatant frozen within 20-30 minutes and stored at -80 °C until analysis. Samples with red blood cell count>25 / μl or indication for an inflammatory process were excluded.

[0412] Preparation of microtubule samples. We reconstituted unlabelled tubulin (T240-A, Te- buBio, Offenbach, Germany) to a concentration of 100 pM. We prepared stabilized microtubules by polymerizing through step-wise increase of the tubulin concentration. Initially, a 3 pM tubulin solution in M2B (magnesium 2X buffer: 80 mM PIPES with 1 mM EGTA and 2 mM MgCb, pH 6.8, adjusted with KOH), in the presence of 1 mM GMPCPP (NU-405L, Jena Bioscience, Jena, Germany) was prepared at 37 °C to nucle- ate short microtubule seeds. Next, the total tubulin concentration was increased to 9 pM in order to grow long microtubules. To avoid further microtubule nucleation, we added 1 pM tubulin at a time from a 42 pM stock solution and waited for 15 min between the successive steps. We centrifuged the polymerized microtubules for 10 min at 13,000 x g to remove any non-polymerized tubulin and short microtubules. We discarded the supernatant and carefully resuspended the pellet in 800 μl M2B-taxol (M2B supplemented with 10 pM taxol (T7402, Merck, Darmstadt, Germany).

[0413] Immunostaining procedures.

[0414] Tubulin immunostaining. U2OS cells were first incubated with 0.2% saponin (#47036, Sigma Aldrich), to extract lipid membranes. This procedure was performed for 1 minute in cytoskeleton buffer, consisting of 10 mM MES (#M3671, Merck), 138 mM KCI (#K42209636128, Merck), 3 mM MgCb (#M8266-100G, Sigma-Aldrich), 2 mM EGTA (Merck 324626-25GM) and 320 mM sucrose, at pH 6.1. The cells were then fixed, using 4% PFA and 0.1% Glutaraldehyde (#A3166, PanReac, Darmstadt, Germany), in the same buffer. Unreacted aldehyde groups were quenched using 0.1% NaBH4 (#71320, Sigma Aldrich now Merck), for 7 minutes in PBS, followed by a second quenching step with 0.1 M glycine (#3187, Carl Roth), for 10 minutes in PBS. The samples were blocked and simultaneously permeabilized using 2% BSA and 0.1% Triton X-100 (#9036-19-5, Sigma Aldrich), in PBS (room temperature, 30 minutes). Primary tubulin antibodies (#T6199 Sigma Aldrich, #302211 Synaptic Systems, Gottingen, Germany, #302203 Synaptic Systems, #abl8251 Abeam, Cambridge, UK) were applied for 60 minutes at room temperature, and were then washed off with permeabilization buffer, followed by an incubation of the samples with secondary antibodies (#ST635P-1001, Abberior, Gottingen, Germany). Alternatively, the primary antibodies were saturated with secondary nanobodies (#N1202-Ab635P-S and #N2402-Ab635P-S, both NanoTag Biotechnologies GmbH, Gottingen, Germany) for 30 minutes at room temperature, us- ing a ratio of 1:5 for the primary antibody: secondary nanobody, respectively. After- wards, the antibody mixture was diluted in the blocking buffer, and was applied onto the cells for 60 minutes at room temperature. Five washes with permeabilization buffer followed by three PBS washes (each one for 10 minutes), before continuing with cel- lular expansion.

[0415] Neuronal immunostainings. Neurons were fixed with 4% PFA in PBS (#D8537-500ML, TheromFisher), for at least 30 minutes, before quenching with 50 mM glycine (in PBS) for 10 minutes, and blocking / permeabilizing using 2.5% BSA (#9048-46-8, Sigma-Al- drich), 2.5% NGS, and 0.1% Triton X-100 (#1003287133, Sigma-Aldrich) in PBS (30 minutes at room temperature, unless specified elsewhere otherwise). The antibodies and / or primary nanobodies were diluted in 2.5% BSA, 2.5% NGS in PBS, and they were added to the coverslips for 60 minutes at room temperature. This was followed by washing with the permeabilization buffer (30 minutes, three buffer exchanges), and by the application of secondary antibodies or nanobodies, in the same buffer, for 45 minutes at room temperature. Specimens were then washed five times with permea- bilization buffer and a final wash with PBS was then performed (15-30 minutes, three buffer exchanges). The primary antibodies used were anti-synaptotagminl (SYT1, #105011 Synaptic Systems), anti-Homerl (#160 003, Synaptic Systems), anti-Shank2 (#162204 Synaptic Systems), anti-GluR2 (Alomone Labs, #AGC-005, Jerusalem, Is- rael), anti-GluN2b (Neuromab 75-101, California, USA), anti-MAP2 (Novus Biologicals #NB300-213), anti-vGluTl (#135304, Synaptic Systems), anti-Bassoon (#ADI-VAM- PS003-F, Enzo, New York, USA). Primary nanobodies were FluoTag-X2 anti-PSD95 (clone 1B2, #N3702, NanoTag Biotechnologies GmbH). Secondary antibodies were conjugated to Alexa 405 (#abl75674, Abeam), Alexa Fluor 488 (AF488, #706-545- 148, Dianova), Cy3 (#711-165-150, Jackson ImmunoResearch), Abberior STAR580 (AS580 #ST580-1006, Abberior), Abberior STAR635P (#2-0112-007-1, Abberior), FluoTag-X2 STAR635P #N2002-Ab635P and #N2402-Ab635P (NanoTag Biotechnolo- gies GmbH). For post-expansion immunostainings of PSD95, tubulin and bassoon, the gels were blocked with 5% NGS in PBS + 0.1% Triton X-100 (PBST) for 2 h, and were then incubated with either PSD95-nanobody, tubulin- or bassoon-antibody at concen- trations of 5-10 pg / ml in PSBT overnight at 4 °C. The gels were washed 4x30 min each. PSD95 gels were expanded by adding water, while tubulin and bassoon gels were then incubated with secondary antibodies in PBST overnight. On the next day, the gels were washed 4x30 min each in PBST and we then expanded the gels by adding water.

[0416] Live immunostaining using synaptotagmin 1 antibodies. Surface Synaptotagmin 1 (Sytl) molecules were first blocked using unconjugated 604.2 Sytl antibodies (#105311 Synaptic Systems), for 10 minutes at room temperature, in Tyrode buffer lacking Ca2+(to reduce drastically both exo- and endocytosis; the Tyrode buffer con- tained 124 mM NaCI (#K52190904041, Merck), 5 mM KCI, 2 mM CaCI2(A862982, Merck), 1 mM MgCh, 30 mM glucose, 25 mM HEPES (K45408310520, Merck), at pH 7.4). The neurons were then wash with room temperature-Tyrode buffer and incu- bated over an ice water bath and exposed to fluorescently-conjugated Sytl antibodies (#1O5311AT1, Synaptic Systems) for 40 minutes, to enable limited exo- and endocy- tosis. The neurons were then washed with ice-cold Tyrode buffer and then were fixed with 4% PFA for 20 minutes, and quenched with 50 mM glycine for 10 minutes. The samples were then blocked with 2.5% BSA in PBS for 30 minutes and vGluTl antibody was added prior to permeabilization for 1 h. Three brief washing steps with blocking buffer preceded the half-an hour permeabilization step (0.1% Triton, 2.5% BSA, 2.5% NGS in PBS), and neurons were labelled for PSD95 using the FluoTag-X2 anti-PSD95 nanobody (NanoTag Biotechnologies GmbH), as indicated above. Synapses were iden- tified as regions in which vGluTl and Sytl signals were found adjacent to the PSD95 staining.

[0417] Immunostaining of cerebro-spinai fluid (CSF) samples. Cerebro-spinal fluid probes were obtained from PD patients and controls at the Paracelsus Elena Klinik (Kassel, Germany), and were stored at -80°C before use. 20 μl amounts of CSF were placed on BSA-coated coverslips, enabling the sedimentation of multiprotein species overnight at 4° C. Fixation with 4% PFA (10 minutes, room temperature) and quenching with 50 mM glycine (10 minutes, room temperature) was followed by the application of either antibodies (Alpha-synuclein #128211 and 128002, Synaptic Systems) or Alpha-synu- clein nanobody2, SynNb268, custom produced and fluorescently-conjugated by NanoTag) for 1 h at room temperature, in 2.5% BSA in PBS buffer. For the case of antibodies, secondary Aberrior STAR635P was applied for 1 h at room temperature. Five washes with 2.5% BSA in PBS were followed by mild post-fixation with 4% PFA for 4 min, and by the expansion procedures.

[0418] Brain slice immunostaining. The fixed brain slices were first quenched using 50 mM glycine (in PBS), followed by three washes with PBS (each for 5 minutes), and blocking and permeabilization in PBS containing 2.5% BSA and 0.3% Triton X-100, for 120 minutes at room temperature. The primary antibodies used (Bassoon, #ADI-VAM- PS003-F, Enzo Life Sciences GmbH, Lorrach, Germany; Homerl, #160003, Synaptic Systems) were diluted in the same buffer (lacking Triton X-100) to 2 pg / ml and were added to the slices overnight, at 4°C. Three washes with PBS (each for 5 minutes) removed the primary antibodies, enabling the addition of secondary antibodies conju- gated with Abberior Star635P (#ST635P-1001, Abberior, Gottingen, Germany) for Bas- son identification, or with Cy3 (#711-165-152, Dianova, Hamburg, Germany) for Homerl identification. The secondary antibodies were diluted to 1 pg / ml in PBS con- taining 2.5% BSA, and were incubated for 3 hours at room temperature. The brain slices were finally subjected to five washes with PBS containing 2.5% BSA (each wash for 5 minutes), followed by two final 5-minute washes in PBS.

[0419] Immunostaining of SARS-CoV-2 particles. Intact SARS-CoV-2 samples deposited by the Center for Disease Control and Prevention were obtained through BEI resources, NIAID, NIH: isolate USA-WA1 / 2020, NR-52281 (Cat# NATSARS(COV2)-ERC, Zep- toMetrix, USA). The samples consisted of patient serum containing viral particles, fixed chemically using aldehydes, in a buffer containing BSA. An average of 9200 viral par- ticles were allowed to adsorb onto single BSA-coated coverslips overnight at 4° C. Samples were mildly fixed with 4% PFA for 4 min before immunostaining using anti SARS-CoV-2 Spike Protein SI (Cat# PA5-114447, ThermoFisher Scientific) and anti- human IgG (Fc)-Alexa 488 (Cat# 109-545-170, Jackson ImmunoResearch), as de- scribed above.

[0420] GFP-nanobody complex (TSR) generation. The monomeric (A206K) and non-fluores- cent (Y66L) EGFP (mEGFP*) was modified to have an ALFA-tag on the N-Terminus and a HaloTag on its C-terminus (ALFA-EGFP-HaloTag). This construct was expressed in a NebExpress bacterial strain, and it had an N-terminal HisTag, followed by a bdSUMO domain, which enables the specific cleavage of the HisTag39later on, after the purifi- cation procedures. Bacteria were grown at 37°C with shaking at 120 rpm in terrific broth (TB) supplemented with kanamycin. When reaching an optical density (OD) of ~3, the temperature was reduced to 30°C and bacteria were induced using 0.4 mM isopropyl p-D-l-thiogalactopyranoside (IPTG), with shaking for another ~16h. Bacteria lysates were incubated with Ni+resin (Roche complete) for 2h at 4°C. After several washing steps, the ALFA-tag-mEGFP(Y66L)-HaloTag protein was eluted by enzymatic cleavage on the column by using 0.1 pM of SENP1 protease for 15 minutes. Protein concentration was determined using Nanodrop (ThermoFisher), and purity was as- sessed by Coomassie gels. Complex formation was performed by mixing, for Ih at room temperature, in a final volume of 40 pl, the following: 25 pmol of ALFA-EGFP- HaloTag and 30 pmol of 3 different single-domain antibodies: FluoTag-Q anti-ALFA (Cat# N1505), FluoTag-X2 anti-GFP (clone 1H1, Cat# N0301) and FluoTag-X2 anti- GFP (clone 1B2), all from NanoTag Biotechnologies GmbH. The control experiments were performed by a similar procedure, without including the target protein ALFA- EGFP-HaloTag. Expression and purification of eGFP used in Supplementary Fig. 8 were performed as before81. Briefly, Neb Express E. co / / strain (New England Biolabs) was cultured in terrific broth at 37°C and induced using 0.4 mM IPTG for 16h at 30°C. Bacteria pellets were sonicated on ice in 50 mM HEPES pH 8.0, 500 mM NaCI, 5 mM MgCh, and 10% glycerol. After removing cell debris by centrifugation, the lysate was incubated for lh with complete His-Tag Purification Resin (Roche) at 4°C. After wash- ing the resin in batch mode with more than 10 column volumes (CV), eGFP was enzy- matically eluted using 0.1 pM of SUMO protease. Concentration was determined by absorbance at 280, using the molecular weight and extinction coefficient of eGFP. Pu- rified protein was diluted in 50% glycerol and stored in small aliquots at -80°C.

[0421] Polyacrylamide gel electrophoresis (PAGE). A primary mouse monoclonal antibody against synaptobrevin 2 (Cat# 104 211, Synaptic Systems) and a secondary antibody conjugated to Abberrior Star635P (Cat#ST635P-1002-500UG) were mixed with reduc- ing 2x Laemmli buffer (63 mM Tris-HCI pH 6.8, 2% SDS, 100 mM DTT, 20% glycerol) and heated for 10 minutes at 96°C. The denatured and reduced samples were then loaded in a self-cast Tris-glycine 12% polyacrylamide gel, and lOpg of total protein was loaded per lane. Electrophoresis was run at low voltage, at room temperature. The gel was briefly rinsed using distilled water and fluorescence was read on a GE- Healthcare AI-600 imager using a far-red filter (Cy5 channel). Next, the gel was sub- merged for 4 hours in Coomassie Brilliant Blue solution to stain all proteins, following by incubation with destaining solutions, before finally being imaged using the same GE-Healthcare Al 600 gel documentation system.

[0422] Dot Blot. In a stripe of nitrocellulose membrane (GE Healthcare), 5 mg of bovine serum albumin (BSA) and 1 pg of ALFA-tagged EGFP(Y66L)-HaloTag were spotted and let to dry at room temperature. Membranes were then blocked in PBS supplemented with 5% skim milk and 0.05% Tween-20 for 1 h with tilting / shaking. FluoTag-X2 anti-GFP Cy3 (clone 1B1), FluoTag-X2 anti GFP-AberriorStar635P (clone 1H1) and Fluotag-X2 anti-ALFA AbberiorStar635P (all from NanoTag) were used at 2.5 nM final concentra- tion in PBS with 5% milk and 0.05% Tween-20 for lh with gentle rocking. After 1 h incubation at room temperature and protected from light, 5 washing steps using 2 ml each were performed with PBS supplemented with 0.05% Tween-20 for a total of 30 minutes. Membranes were finally imaged using a GE-Healthcare Al 600 system.

[0423] 1,6-hexanediol treatments. This compound (#240117-50G, Aldrich) was diluted in the neuronal Neurobasal-A culture medium at 3% for 2 minutes, and 10% for 12 minutes, before fixation and further processing for immunostaining.

[0424] Purified proteins. Immunoglobulins A and M were purchased from Jackson Immu- noResearch and Immunoglobulinss G from Abberior, Gottingen, Germany (AffinityPure IgA 109-005-011, ChromePure IgM 009-000-012, and ST635P-1001, respectively) and were diluted in PBS, before expansion procedures. Otoferlin was produced according to standard procedures82, and was diluted in 20 mM HEPES, 100 mM KCI, 0.05% DDM buffer, before being used at 0.4 mg / ml concentration. For GABAA receptors a construct encoding the full-length human GABAA receptor b3 subunit (Uniprot ID P28472), with an N-terminus TwinStrep tag, was cloned into the pHR-CMV-TetO2 vector83. A lentiviral cell pool was generated in HEK293S GnTI-TetR cells as described previously84. Cells were grown in Freestyle 293 expression medium (#12338018, Gibco) supplemented with 1% fetal bovine serum (#11570506, Gibco), ImM L-Glutamine (#25030149, Gibco), 1% NEEA (Gibco #11140050) and 5 mg / ml blasticidin (Invivogen #ant-bl-5b) at 37 °C, 130 r.p.m., 8% CO2 and induced as described85. Following collection by cen- trifugation (2,000 g, 15 min), the cell pellets were resuspended in PBS, pH=8 supple- mented with 1% (v / v) mammalian protease inhibitor cocktail (Sigma-Aldrich). Cell membranes were solubilized with 1% (w / v) n-dodecyl 0-D-maltopyranoside (#D3105GM, DDM, Anatrace) for lh. The insoluble material was removed by centrifu- gation (12,500g, 15 min) and the supernatant was incubated with 300 mL Strep-Tac- tin® Superflow® resin (IBA lifesciences) while rotating slowly for 2h at 4°C. The beads were collected by centrifugation (300g, 5 min) and washed with 150mL of 0.04% (w / v) DDM, PBS pH=8. The sample was eluted in 2.5 mM Biotin, 0.02% (w / v) DDM, PBS pH=8 and used for imaging at 1 mg / ml concentration. For the purification of the GABAA receptor in complex with the £3-[33 specific nanobody (Nb25)86, Nb25 was fluores- cently labelled with STAR635P at the N- and C-termini generating Nb25-STAR635P. 20 μl of 10 pM Nb25-STAR635P was added to the sample prior to the elution step and incubated for 2h at 4°C while rotating. The excess Nb25-STAR635P was removed by washing the beads with 6 bed volumes of 0.04% (w / v) DDM, PBS pH =8, eluted with 2.5 mM Biotin, 0.02% (w / v) DDM, PBS pH=8 and used for imaging at 3mg / mL con- centration. The same procedure was applied for the negative control anti-eGFP nano- bodies. To test that Nb25-STAR635P could still bind the receptor, 2 pM of Nb25- STAR635P was added to the 03 homomeric receptor reconstituted in nanodisc as de- scribed previously87. 3.5 μl of the sample was applied to a freshly glow-discharged (PELCO easiGlow, 30 mA for 120s) 1.2 / 1.3 UltrAuFoil grid (Quantifoil), which were blotted for 2.5s and plunge-frozen using a Leica EM GP2 plunger at 14 °C and 99% humidity. Imaging was performed on a Titan Krios G2 microscope at the MRC LMB equipped with a F4 detector in electron counting mode at 300kV at a nominal magni- fication of 96,000 corresponding to a calibrated pixel size of 0.824 A. 300 movies were collected using EPU (Thermo Fisher Scientific, version 2.0-2.11) with total dose of 38 e / A2and 6.43s exposure time. The movies were motion corrected using MotionCor288. Contrast transfer function estimation was performed with CTFFIND-4.1.1389. Particle picking was performed using a retrained BoxNet2D neural network in Warp90, followed by 2D classification in cryoSPARC91. Calmodulin was purified as previously described92, and was used in calcium free buffer: 150 mM KCI, 10 mM HEPES, 5 mM EGTA, or calcium+ buffer: 150 mM KCL, 10 mM HEPES, 2 mM CaCb, at pH = 7.2, before expan- sion procedures. In brief, calmodulin 1 (mRNA reference sequence number NM_031969.2) was tagged with mEGFP and an ALFA-tag, for affinity purification pur- poses. The construct was transfected in HEK293 cells using Lipofectamine 2000 (Invi- trogen, Carlsbad, CA, USA #11668019), following the manufacturer's protocol. After expression for ~24 hours, the cells were lysed in a PBS buffer containing 1% Triton X- 100, 2 mM EDTA and a protease inhibitor cocktail. The debris was removed by centrif- ugation, and the supernatant was added to an ALFA Selector PE resin (NanoTag Bio- technologies), where it was allowed to bind for 60 minutes (4°C, under rotation). After two washes with lysis buffer and one wash with PBS (ice-cold), the bound proteins were eluted by adding the ALFA peptide. The purified protein was analyzed by Coo- massie gel imaging (published in92).

[0425] Expansion procedures. X10 expansion of cultured cells was performed using proteinase K exactly as described in the following protocol article:29. X10 expansion relying on autoclaving (XIOht47) was performed as follows. The samples were incubated with 0.3 mg / ml Acryloyl-X (SE; #A-20770, Thermo Fisher Scientific) in PBS pH 7.4, overnight, at room temperature. The samples were then subjected to three PBS washes (5 minutes each), while preparing the gel monomer solution, exactly as described29. The solution was pipetted on parafilm and was covered by upside-down coverslips contain- ing cells, or with brain slices that were then also covered with fresh coverslips. Polymerization was allowed to proceed overnight at room temperature, in a humidified chamber. Homogenization of proteins and single molecules were performed using 8 U / ml proteinase K (PK, #P4850 Sigma Aldrich now Merck) in digestion buffer (800 mM guanidine HCI, 2 mM CaCb, 0.5 % Triton X-100, in 50 mM TRIS, #8382J008706, Merck), overnight at 50°C. Homogenization of cell cultures and brain slices was done by autoclaving for 60 minutes at 110°C in disruption buffer (5% Triton-X and 1% SDS in 100 mM TRIS, pH 8.0) followed by a 90 minutes incubation for temperature to cool down to safe levels. Before autoclaving, the gels were first washed using 1 M NaCI, and were then washed at least four times in disruption buffer, for a total time of at least 120 minutes. Gel expansion was then performed by ddthO washing, for several hours, with at least five solution exchanges. Expansion was performed in 22 x 22 cm square culture dishes, carrying 400-500 ml ddl-hO. When desired, the samples were labelled using 20-fold molar excess of NHS-ester fluorescein (#46409, ThermoFisher Scientific) in NaCHCh buffer at pH = 8.3 for 1 h, before the washing procedure that induced the final expansion.

[0426] ZOOM expansion procedures. Fixed U2OS cultured cells were incubated in anchoring solution (25 mM Acrylic acid N-hydroxysuccinimide ester in 60% v / v DPBS and 40% v / v DMSO) for 60 minutes. Afterward, cells were moved to monomer solution (30% w / v Acrylamide and 0.014% w / v N-N'-methylenbisacrylamide in PBS buffer). After 60 min, the gelation process was started by adding initiators (0.5% w / v TEMED and 0.5% w / v APS) to the monomer solution. The hydrogel-cell hybrid was homogenized in de- tergent solution (200 mM SDS, 50 mM boric acid in DI water, pH titrated to 9.0), at 95 °C for 15 min, following by 24 h at 80 °C. ZOOM-processed samples were then stained using the previously mentioned anti o-tubulin antibodies (1:400 in PBST).

[0427] Microscope systems. For image acquisition, small gel fragments were cut and were placed in the imaging chamber presented in Supplementary Fig. 2. Paper tissues were used to remove any water droplets around the gels, before enabling the gels to equil- ibrate for at least 30 minutes on the microscope stage. Epifluorescence imaging was performed using an Olympus 1X83 TIRF microscope equipped with an Andor iXon Ultra 888, lOOx 1.49 NA TIRF objective, and an Olympus LAS-VC 4-channel laser illumina- tion system. Confocal imaging was performed, for most experiments, using a TCS SP5 STED microscope (Leica Microsystems, Wetzlar, Germany), using a HCX Plan Apochro- mat STED objective, 100x, 1.4 NA, oil immersion. The LAS AF imaging software (Leica) was used to operate imaging experiments. Excitation lines were 633, 561, and 488 nm, and emission was tuned using an acousto-optical tunable filter. Detection was ensured by PMT and HyD detectors. Images were taken using a resonant scanner at 8 kHz frequency. 5D-stacks for zONE were performed using a 12 kHz resonant scanner mounted on a Leica TCSSP8 Lightning confocal microscope. Samples were excited with a 40% white light laser (WLL) at wavelengths of 633, 561 and 488 nm, and acquisitions were carried out using HyD detectors in unidirectional-xyct line scans or in uni- and bi- directional xyczt line scans.

[0428] Image acquisition. Objectives of 1.4, 1.45 and 1.51 NA were used to acquire images with a theoretical pixel size of 98 nm. For a higher resolution, the theoretical pixel size was set to 48 nm, at the cost of slightly lower detection rate. Images acquired on camera-based system had a predetermined pixel size of 100 nm. The acquisition speeds were ranging between 20 to 40 ms and 25 ms on a resonant scanner of 8 kHz and on a camera, respectively, for xyct. For hyperstacks of xyczt acquisitions, images were acquired using 8 kHz and 12 kHz scanners in bidirectional mode (after the nec- essary alignments) to compensate for speed loss. Images of 8 bit depth were acquired at a line format ranging from 128x128 to 256x256. The scanning modality on a confo- cal was set to "minimize time interval" (Leica LAS software). To maintain natural fluc- tuations of fluorophores, we did not use line accumulation or line averaging during scanning. A frame count starting from 200 and up to 4000 were acquired. We recom- mend a frame count of at least 1500 to 2000 for optimal computed resolution.

[0429] Image processing. ONE image processing is enabled through a Java-written ONE Plat- form under "ONE microscopy" in Fiji. The ONE microscopy plugin utilizes open-source codes from Bioformats Java library, NanoJ-Core, NanoJ-SRRF, NanoJ-eSRRF, and Im- age Stabilizer5-7-93'94. ONE plugin supports multiple video formats of single or batch analyses in xyct. Hyperstacks with 5-dimensions xyczt format are processed with zONE module. This module allows the user to select the optical slices and channels to resolve at ultra-resolution. Upon irregularities in resolving one or more channels within one or more planes, zONE leaves a blank image, and computes the remaining planes within a stack. The image processing is fully automated and requires minimal initial user in- put. Aside from the expansion factor, preset values and analysis modalities are auto- matically provided (for more details, see Supplementary Fig. 1). The ONE plugin has a pre-installed safety protocol to skip failures in computations or uncompensated drifts, without affecting the progress of batch analysis. Data analysis, parameters and irreg- ularities are reported in log files. The ONE plugin automatically linearizes the scale, based on radiality magnification and expansion factor corrections. In addition, ONE offers the possibility to correct for chromatic aberration by processing multi-channel bead images as a template that is applied to super-resolved images of the biological samples. The correction is performed by applying the Lucas-Kanade algorithm93. For the ONE Microscopy plugin to store complex multi-dimension images from hyperstacks, we modified the Java code of the ImageJ library and adapted it locally. ONE Platform source code and plugin are available on https: / / www.rizzoli-lab.de / ONE. For best per- formance, we recommend to download a preinstalled version on Fiji available via the same link.

[0430] Image analysis and statistics. For single-object analyses, such as synaptic vesicle or antibody analyses, signal intensities and distances between objects were analyzed manually using ImageJ (Wayne Rasband and contributors, National Institutes of Health, USA). Line scans were also performed and analyzed using ImageJ. For the analysis of PSDs (Figure 13), spots were identified by thresholding band-pass filtered images, relying on empiric thresholds and band-pass filters, organized in the form of semi-automated routines in Matlab (version 2017b, The Mathworks, Inc, Natick, MA, USA). Spots were either overlaid, to determine their overall signal distributions, or their center positions were determined, to measure distances between spots (in either the same or different channels). The same procedure was used for the averaging analysis of CSF samples (Figure 15) and for the analysis of spot distances for the GFP-nanobody assemblies (Extended Data Fig. 5). FWHM values were measured after performing line scans over small but distinguishable spots, as indicated in Figure 12, followed by Gaussian fitting, using Matlab. The averaging analysis of GABAA receptors is presented in detail in the main text, and was performed using Matlab. In brief, receptors were detected automatically, as particles with intensities above an empirically-derived threshold. To remove particles with uncompensated drift, we eliminated all receptors coming from images in which a large proportion of the particles were oriented similarly. We then inspected visually all of the remaining particles, to choose those that appeared to be in a "front view", showing a reasonably round appearance, with nanobodies placed at the edges of the receptor (visible in the second color channel). All particles were centered on the intensity maxima of the respective GABAAR channel images. The particles were subjected to an analysis of the peaks of fluorescence, using a bandpass procedure, followed by identification of maxima95, the positions of the peaks were calculated to below-pixel precision and were rounded off to a pixel size of 0.384 nm (the starting pixel size was 1 nm). These positions were then mapped into one single matrix, which represents the "averaged receptor", as indicated in the main text. Av- eraging analyses of tubulin and actin were performed similarly. In brief, microtubule segments, tubulin dimers or actin strands were selected manually and were overlaid, to generate average views. For the tubulin dimers, we calculated the peaks of fluores- cence, as performed for the GABAA receptors, above. Model objects were generated, as a comparison, by convoluting the amino acid positions in the respective PDB struc- tures with empirically-derived ONE spots. All of these analyses were performed using Matlab. The signal to noise ratio (SNR) for single nanobodies was determined by meas- uring the average pixel intensities within the nanobody "spots" and away from them, and dividing the two measurements. Identically-sized circular regions-of-interest, suf- ficient to capture the nanobody spots completely, were used for both signal and back- ground (noise) regions. Plots and statistics were generated using GraphPad Prism 9.3.1 (GraphPad Software Inc., La Jolla, CA, USA) or SigmaPlot 10 (Systat Software Inc., San Jose, CA, USA), or using Matlab. Statistics details are presented in the respective figures. Figures were prepared with CorelDraw 23.5 (Corel Corporation, Ottawa, On- tario, Canada). Description of the Figures

[0431] Figure 1: Scheme - Chamber design.

[0432] Item Specifications

[0433] Figure 2: Testing the reactivity of FAM-PSI (Ser-Binder + fluorescein). The samples were labeled with FAM-PSI and NHS-Ester STAR635 (e.g. Abberior®). The labeling procedure is as follows: PFA fixed cultured neurons were incubated in buffer contain- ing 50% sodium bicarbonate and 50% DMSO + 3 mg / ml of FAM-PSI. Then the sam- ples were washed in sodium bicarbonate buffer (3x buffer exchange) and NHS-Ester labeling was carried in the presence of sodium bicarbonate buffer. The samples were washed several times on gentle shaking for 3 to 5 hours with sodium bicarbonate buffer.

[0434] Figure 3: AcX + Ser-Binder Image (+Figure Legend including protocol). Double la- beling of serine and lysine residues in expanded samples with FAM-PSI and Ac-X, re- spectively.

[0435] Figure 4: Ac-X + Ser-binder chemistry + 35 radiality magnification improves the preservation and resolving power for protein structures. Adding 3D-STED doughnut lasers (x,y and z-doughnut lasers) improves the axial resolution down to 1,1 and 3.7 nm in x,y,z directions, respectively. The new chemistry and the enhanced axial reso- lution allow to see details in the intracellular loops of the GABAA receptors, which was not possible before, due to the disorganized nature of these domains.

[0436] Figure 5: Single molecules anchored with Ac-X and PSI. PSI alone works, but works better when combined with Ac-X.

[0437] Figure 6: 48 nm pixel size yields higher detail in the single protein molecules. The smaller starting pixel size allows the detected photons to be digitized onto more pix- els. This enhances the prediction possibility of the individual fluorophore positions. Figure 7: Methodology.

[0438] Figure 8: ONE analysis and examples, a & b, Several views of the starting interface of the ONE software package. The examples show the intuitive software choices. See also the "Readme / Help" file of the software package, c, Examples of different poten- tial artifacts that should be avoided in ONE imaging, d, Different potential choices in how to resolve ONE images. We suggest using the TRPPM procedure for dim sam- ples. This reduces the obtainable resolution, but follows much better the potential molecule shape. For brightly labeled samples with direct labeling, the TRAC4 proce- dure provides the best resolution and SNR, indicating the positions of the individual fluorophores.

[0439] Description

[0440] Figure 9: Analysis software flowchart including essential and optional features.

[0441] Figure 10: ONE images of Ser-binder FAM-PSI + Ac-X anchored single protein mole- cules, imaged in two colors.

[0442] Figure 11: ONE images of Cys-binder (XXX) + Ac-X anchored single protein mole- cules, imaged in two colors.

[0443] Figure 12: ONE performance in cellular samples and in vitro, a-b, Tubulin im- munostainings (relying on primary and secondary antibodies) imaged using STED, without expansion (a), confocal after X10 expansion, ONE microscopy, ONE with ZOOM ExM (3.5-fold), and STED with X10 expansion (ExSTED, b). c, The graph depicts the line scans indicated by the dashed lines, which come from two different gels, with different expansion factors (10-fold for ONE, 3.5-fold for ZOOM-ONE). The image (AVG ONE) shows an average of 36 cross-sections, d, ONE microscopy images at different Z-axis levels, obtained by confocal scanning at different heights (zONE). e, The general scheme of the GFP (Green Fluorescent Protein)-based assemblies (generated in Pymol, using the PDB structures 6I2G and 3K1K), along with a typical ONE microscopy image, with the rough positioning of the molecules indicated by the cartoon. The three nano- bodies carry three spectrally different fluorophores. f, Two further examples, relying on a design in which NB1 and NB3 carry identical fluorophores. To detect GFP, the samples were labelled with NHS-ester fluorescein, after homogenization. Here we used a small pixel size, enabling the detection of two fluorophores connected to the nano- bodies (see Supplementary Figure 6). Line scans across the indicated fluorophores are also shown.

[0444] Figure 13: ONE analysis of single molecules, a-c, Images of isolated immunoglobulins (IgG, IgA, IgM), labelled with NHS-ester fluorescein after homogenization. The IgGs were secondary anti-mouse antibodies, carrying STAR635P (blue), d, Distances be- tween fluorescently-conjugated IgGs and fluorescently-conjugated secondary antibod- ies (top) or secondary nanobodies (bottom). In panels a-d we used a different fluctu- ation analysis, termed temporal radiality pairwise product mean, TR.PPM, unlike the temporal radiality auto-cumulant to the order of 4 (TRAC4)6approach used in most other figures. Unlike TRAC4, which aims to separate the individual fluorophores, TRPPM enhances the cohesiveness of the fluorophores decorating the single antibod- ies, resulting in cloud-like signals whose distances are easily measured (N = 20 / 19, for AB:AB / AB:NB). Right panel: distance between the two secondary nanobodies binding single IgGs (N = 9). e-f, A TRAC4 analysis of purified GABAA receptors. Line scans across specific profiles seem to detect the receptor pore, g, AlphaFold-predicted model of otoferlin structure, h, ONE examples of otoferlin images, i, Z-axis ONE imaging, indicating that two components (presumably C2A and TM domains) are relatively far from the main body of the molecule, j-k, 1CLL and 1CFD PDB structures of the Ca2+sensor calmodulin, in presence or absence of its ligand, respectively, along with ONE images, after proteinase K-based homogenization and expansion (see Supplementary Fig. 11 for more images). I, The expected elongation by ~1 nm was reproduced (p = 0.0006, Mann-Whitney test; N = 70-155). m-n, Similar analysis, after homogenization using autoclaving (p <0001, N = 66-197). The box plot shows the medians, the 25thpercentile and the range of values.

[0445] Figure 14: Averaging of GABAAR ONE images, a, A gallery of GABAAR / NB ONE images, b, An average view generated by processing 1225 receptor images, c, Recep- tors were modeled with increasing errors in the fluorophore placement. The error is indicated below above each receptor, in nanometers. The pore of the receptor is no longer visible when the fluorophores are displaced by 2 nm from their original posi- tions. This analysis assumes that all receptors we analyzed are placed in a perfectly vertical position, and that there is no error in the identification of the receptor center. Neither of these assumptions is reasonable, d, The receptors were modeled with a variable shift in their lateral positioning and in their vertical tilt angles. The angles and placement errors are indicated above and to the left, respectively. Here we assume that fluorophore placement is perfect. This analysis reveals the fact that minor errors in the receptor orientation would lead to the same type of image as we obtained in panel b.

[0446] Figure 15: ONE imaging of cytoskeletal components, in NHS-ester labelled cells, a, HeLa cells were subjected to extraction according to a previously published protocol51. Panels from top to bottom show an overview of non-expanded, extracted HeLa cells labelled with NHS-ester chemistry (lOx objective), followed by a higher magnification view and by two views of cells expanded after extraction, b, An exemplary image of a cytoskeletal filament, c, Composite actin gallery, d, A PDB cartoon, followed by a mod- eled averaged PDB, depicting a best-case scenario for ONE microscopy of actin fibrils under ideal conditions, and the achieved ONE image reconstructed from 47 different actin regions, e, f, & g, A set of measurements that were applied over actin filaments, quantifying the longitudinal peak-to-peak distance along the filament (distance be- tween actin monomers), the actin filament width, and the coefficient of variance of these two measurements, respectively. These values resemble well the known dimen- sions from actin PDBs (magenta lines). N = 198 and 130 for the first graph and second graph, respectively; 2 independent experiments.

[0447] Figure 16: ONE reveals pre- and postsynaptic structures, a-c, Synaptic vesicles were labelled live using an antibody against a luminal epitope of synaptotagmin 1 (Sytl, magenta). The vesicular glutamate transporter (vGluTl, blue) and PSD95 (gray) were immunostained using an antibody and a nanobody, respectively, a, Recently endocy- tosed vesicle exhibiting circular morphology, b, Readily retrievable pool molecules form patches containing Sytl / vGluTl (top), which are dispersed by cholesterol extraction using M0CD (bottom), c, M0CD causes molecules to spread across larger areas (left: N = 22-19, 2 independent experiments, p < 0.0044, Mann-Whitney test; right: N = 22- 22, 2 independent experiments, p = 0.8937), although the signal per vesicle (the Sytl copy number) remains unchanged, d, A visualization of PSDs (top and side views), after immunostaining PSD95 with the same nanobody used in a-c, and Shank2 and Homerl with specific antibodies. The graph indicates the axial positioning, which agrees well with the literature57. N = 11 measurements for each protein, 2 independent experiments; symbols show the medians, SEM and SD. e, Side view of a postsynapse displaying PSD95, MAP2 and two glutamate receptors (GluA2, AMPA type, and GluN2b, NMDA type), f, ONE images of PSD95 (top views), with, or without, the addition of 10% 1,6-hexanediol (Hex), g, Line scans through the PSD95 staining shown in panel f. h, An analysis of PSD95 spot profiles; N = 10-7 synapses, Friedman test followed by Dunn-Sidak testing, p = 0.0027; the error bars show the SEM. For details on the anal- ysis, see Supplementary Fig. 19.

[0448] Figure 17: A promising avenue for Parkinson's Disease (PD) diagnostics, a, Cerebro- spinal fluid probes were obtained from PD patients and controls, and 20 μl amounts were placed on BSA-coated coverslips, followed by ONE imaging, after immunolabeling ASYN using a specific nanobody68, b, A gallery of typical ASYN species observed in the CSF samples, c, Average ASYN assemblies from a PD patient and a control, d, An analysis of the spot profiles detects significant differences, with the average control object being smaller than the average PD object. All ASYN assemblies for the control and PD patients were averaged, from 3 independent experiments, Friedman test fol- lowed by Dunn-Sidak, p = 0.0237; errors show SEM. e, An analysis of the numbers of the larger assemblies in CSF samples. No significant differences, Mann-Whitney tests, p = 1, p = 0.7104. f, An analysis of the numbers of oligomers in CSF samples. All comparisons indicated significant differences, Mann-Whitney tests followed by a Ben- jamini-Hochberg multiple testing correction with FDR of 2.5%; p values = 0.0105, 0.0023, 0.0111, 0.0012, and 0.0012, in the respective order of data sets, g, Analyses of the numbers of oligomers, as a proportion of all ASYN assemblies analyzed (left), or as numbers per acquisition in (h). Both procedures discriminate fully between the PD patients and the controls. For the second procedure, the lowest PD value is 50% larger than the highest control. N = 7 PD patients and 7 controls, Mann-Whitney test, ZK0.0001 for both h and g.

[0449] Extended Data Fig. 1. A general overview of the ONE microscopy approach, a, Bio- logical samples are linked to gel anchors, relying on Acryloyl-X, followed by X10 gel formation and homogenization, either b proteinase K additions or by autoclaving in alkaline buffers. Full expansion is achieved by repeated washes, and is followed by mounting gel portions in a specially designed chamber. In principle, one could image the samples using different super-resolution procedures. Techniques benefitting from bright samples, as STED or SIM, suffer due to the fluorophore dilution induced by the expansion procedure. Techniques requiring special buffers (e.g. SMLM) are nega- tively affected by the water environment. In contrast, technologies relying on fluoro- phore fluctuations profit from the expansion, as the fluorophores are spatially sepa- rated and can fluctuate independentlyl3. b, Repeated imaging is performed (up to 3000 images), in any desired imaging system (confocal, epifluorescence, etc.), to de- tect signal fluctuations, which are then computed using through a plugin (ONE plat- form) based on the SRRF algorithm, before assembling the final super-resolved im- ages.

[0450] Extended Data Fig. 2. A detailed view of the ONE procedure, a, Processing a stack of diffraction-limited images with SRRF, based on the analysis of a gradient of con- vergence of sub-pixels over a radial ity stack, results in super-resolved images with resolutions varying between 50-70 nm. b, The ONE procedure adapts the SRRF algo- rithm to expanded gels, c-f, A detailed explanation of the analysis procedure, c, A sample was fixed and expanded using a 10-fold expansion protocol (X10). The sam- ple was then imaged using a resonant scanner on a confocal microscope. The zoomed-in view indicates one bright spot, whose size in real space is limited by dif- fraction to ~200-300 nm, but represents a 10-fold smaller size in the pre-expansion space (see scale bars in the middle panels). Every pixel is then subjected to a 10-fold radiality magnification and is then subjected to the procedure explained in panels d-f, which provides the final, high-resolution image (right-most panel), d, Signal fluctua- tions are measured by imaging the sample repeatedly, using the resonant scanner (here at 8 kHz), e, A view of the overall signals, obtained by summing 20 of the fluc- tuating images (raw in the left-most panel, background-subtracted in the middle panel), or by summing 1000 images, f, Each image from series obtained as in panel b is subjected to a temporal analysis of fluctuating fluorophores, based on radiality magnification6, thereby providing a super-resolved image whose level of detail be- comes optimal after ~1500 frames.

[0451] Extended Data Fig. 3. Expansion microscopy results in a higher signal-to-noise ra- tio. Expansion microscopy, which separates proteins of interest and removes much of the other cellular components (e.g. lipids, metabolites) should result in a higher sig- nal-to-noise ratio (SNR), a, To test this, we analyzed here the simplest possible sam- ple, consisting of Star635P-conjugated nanobodies on glass coverslips, or in ex- panded gels, using confocal microscopy, relying on analysis using a resonant scan- ner. b, The SNR of these samples increases by 2-fold, on average, after expansion. N = 30-24, P = 0.000001, Mann-Whitney Ranksum test.

[0452] Extended Data Fig. 4. An analysis of tubulin immunostainings. a & b, An analysis of tubulin, following immunostainings relying on primary antibodies detected using Star635P-conjugated secondary nanobodies. While the overall signal distribution is similar to that obtained with secondary antibodies (Figure 12), one can observe often pairs of fluorescent spots in very close vicinity (marked by dotted circles in the cross section), which probably represent the two fluorophores on each nanobody. For a formal analysis of this issue on different nanobodies, see Figure 12 and Extended Data Fig. 5. c, Immunostainings relying on primary antibodies followed by secondary antibodies (upper panel) or by secondary nanobodies (lower panel), d, The graph shows the diameter of microtubules in when using secondary antibodies (left; N = 49 microtubule profiles) or secondary nanobodies (right; N = 101).

[0453] Extended Data Fig. 5. In-depth analysis of GFP-nanobody complexes, a, Dot blots to validate that each nanobody was binding specifically the TSR individually. Nitrocel- lulose membranes were spotted with TSRs and bovine serum albumin, as control, and the spots were revealed with the respective nanobodies, using a fluorescence scanner (GE-Healthcare Al 600). b, An overview of an image showcasing nanobod- ies bound to their GFP target, c, An analysis of distances from STAR635P to Cy3 nanobodies, in normal images or after mirroring one of the fluorescence channels, as a negative controls. The close-distance interval is largely removed by mirroring. N = 40-40 TSRs. Performing this in samples lacking the GFP, in which the nanobodies are randomly distributed, results in no differences between the normal and mirrored distributions. N= 40 / 40 images, d, Overview of the TSR using only two-color nano- body labeling (same as the one used in Figures 16c, d), along with two different ex- amples. The sample is also labeled using NHS-ester fluorescein, and a small pixel size (0.48 nm) is used, to enable the optimal visualization of the TSRs. e, An analysis of the signal-to-noise ratio of the TSRs, obtained by measuring the noise levels in the vicinity of the nanobodies. The noise levels are normalized to 1, implying that the normalized signal of the respective nanobodies now provides directly the signal-to- noise ratio. N = 20-18, 12-14, and 17-11 measurements, p < 0.0001, Mann-Whitney test, f, A Fourier Ring Correlation (FRC) analysis of nanobody images, g, The best and average resolution obtained per image, in the different color channels (N = 4 to 5 analyses for each), h, To approximate the apparent resolution of the system, we drew line scans across spots and measured the full width at half maximum (FWHM) in curve fits executed on the line scans. The graph plots the FWHM of 129, 135, and 132 fluorescein, Cy3 and STAR635P line scans. The values are significantly different between the color channels, p < 0.0001, Kruskal-Wallis test. The box plot shows the median, 25th percentile and the range of values.

[0454] Extended Data Fig. 6. Further PSD examples, a, ONE imaging of PSDs, employing a resonant scanner and a final pixel size of 1 nm, achieving high resolution (same procedure and resolution as in Figure 14f). b, Examples of PSD95 stainings, after treatment with 1,6-hexanediol (Hex), as in Figure 14f. c, We averaged the PSD95 signals for both control and Hex-treated synapses (8 PSDs imaged in top views, for each treatment). The control shows a somewhat regular pattern, while the Hex treat- ment seems to perturb this.

[0455] Extended Data Fig. 7. A detailed analysis of the PSD. a, The PSD was im- munostained for PSD95, Homerl and Shank2, as in Figure 14, and images were taken at different heights along the Z-axis (zONE imaging). An overlay (summed im- age) is shown in the left panel, along with an analysis of the proteins at different Z levels, using a colormap that describes the positions along the Z axis (right panel), b, The distance between PSD95 spots was computed from images as in panel a, and was compared to that obtained from positioning the molecules randomly within the PSD95, N = 10 synapses, Friedman test followed by Dunn-Sidak, p=0.0001. c, The lateral distance between PSD95 spots and between PSD95 and Homerl or Shank2. The minimal distance between each PSD95 spot and a Homerl / Shank2 spot is shown (measured in the lateral plane, in 2D projections of the PSD). N = 10 synapses, from 2 independent experiments. While the distance between PSD95 spots has a non-ran- dom character, as indicated in panel b, the distances to Homerl or Shank2 spots are not different from randomized distributions (Dunn-Sidak tests, p > 0.1), possibly also because these two molecules are immunostained using antibodies, which causes the fluorescence signals to scatter broadly, d, Confocal microscopy analysis of the PSDs, in non-expanded samples. In control conditions all three components analyzed here (PSD95, Homerl, Shank2) are well colocalized. The addition of 3% 1,6-hexanediol (Hex) causes the dispersion of Homerl (magenta), while 10% Hex also disperses Shank2 (blue). PSD95 remains largely unaffected by Hex. e, An analysis of the aver- age PSD95 spot profile confirms this impression, N = 10-7-10 neurons, a set from 3 independent experiments, f, We analyzed the dispersion of Homerl (left) and Shank2 (right) away from the PSD95 spots. The signal present in synapses (near the PSD95 labeling, but not within the PSD) was analyzed, to determine the % that is not corre- lating to the PSD structure. The same samples were analyzed as in panel e.

[0456] Extended Data Fig. 8. ExM-STED (ExSTED) imaging of PSDs. a, Hippocampal cul- tures were immunostained for PSD95 and VGIutl, and were additionally labeled with NHS-ester fluorescein, after homogenization, b, A gallery of high-zoom ExM-STED views of synapses, with a focus on PSD95. Relatively large PSD domains are visible, as in most previous works in the literature, and unlike most of our ONE images, c, To determine if this is simply an issue of resolution, we aimed to generate ExM-STED- like images with ONE microscopy, by reducing its resolution. We employed an epiflu- orescence microscope (as opposed to a rapidly scanning confocal in the panels deal- ing with PSD95 in Figure 16), and we used the temporal radiality pairwise product mean (TRPPM) option of analysis, which broadens the resulting spots. The results are very similar to ExM-STED images, demonstrating that the modular / domain ap- pearance of the PSD95 stainings is a result of insufficient resolution, with a retiolum being evident only at very high resolution (under optimal ONE imaging).

[0457] Extended Data Fig. 9. ONE analysis of brain slices, a, Images of a 200 pm-thick rat brain section before (left) and after (right) expansion, relying on autoclaving for homogenization47. The scale bar does not take the expansion factor into considera- tion. The sections were labeled by using NHS-ester fluorescein incubations, b, Epiflu- orescence images of expanded brain slices, focusing on Bassoon and Homerl as pre- and postsynaptic markers, respectively, c, Similar images, taken using the ONE pro- cedure. d, Line scans executed over the areas indicated in panels b and c. As ex- pected, far more details can be observed in ONE than in simple epifluorescence mi- croscopy. Extended Data Fig. 10. A gallery of ASYN object images from 7 PD patients and 7 controls. The images were obtained following the procedure indicated in Figure 15a. See Supp. Table 1 for details on the respective patients.

[0458] Supplementary Fig. 1. ONE analysis and examples, a & b, Several views of the starting interface of the ONE software package. The examples show the intuitive software choices. See also the "Readme / Help" file of the software package, c, Exam- ples of different potential artifacts that should be avoided in ONE imaging, d, Differ- ent potential choices in how to resolve ONE images. We suggest using the TRPPM procedure for dim samples. This reduces the obtainable resolution, but follows much better the potential molecule shape. For brightly labelled samples with direct label- ing, the TRAC4 procedure provides the best resolution and SNR, indicating the posi- tions of the individual fluorophores.

[0459] Supplementary Fig. 2. Evaluating SRRF analysis performance using DNA origami nanorulers, in non-expanded samples, a, Nanorulers with single Atto647N molecules (R SM) were generated by GATTAquant30 carrying fluorophores on each end of DNA structures of 80, 60, 50, 30, 20 and 10 nm in length. They were then imaged using a confocal resonant scanner, without expansion procedures. The first panel shows con- focal maximal intensity projections (MIPs) for each of the rulers. The second panel shows temporal radiality averaging (TRA) analysis overviews. White boxes indicate the magnified regions displayed in the third panel. The fourth panel shows a tem- poral radiality auto-correlations of fourth order (TRAC4) analysis, overlaid with the respective confocal MIPs. The remaining panels show different ruler examples, ac- quired at different starting pixel sizes, using either a hybrid detector (HyD) or an av- alanche photodiode detector (APD), and analyzed in different SRRF modalities. This analysis is shown in the fifth panel for 50 nm pixel size, using a HyD and analyzed using TRAC4. The sixth and seventh panels show rulers acquired at 100 and 50 nm pixel sizes, using an APD and analyzed using TRAC4. The eighth panel shows rulers that were acquired at 100 nm pixel size and were analyzed using default SRRF set- tings (TRA). b, Magnified overviews of selected regions (indicated by blue rectangles) from each of the ruler exemplary images to the left, c, Signal-to-noise ratio (SNR) analysis of HyD and APD detectors, N = 25 and 30 for HyD and APD, respectively.

[0460] Mann Whitney test, p = 0.004. d, Normalized line scans across the different ruler im- ages, as indicated in the respective panels in (a), e, Apparent FWHM of the different rulers. N = 17, 17, 18, 17, 18 and 17 for 80, 60, 50, 30, 20, and 10 R SM, respec- tively. A Kruskal-Wallis test was applied, followed by Dunn's post hoc test; p < 0.0001.

[0461] Supplementary Fig. 3. The effect of frame number on SRRF analysis, a 8i b, 80 nm rulers were imaged at 100 and 50 nm pixel sizes, and were then analyzed with the default SRRF parameter (temporal radiality average, TRA), using varying frame counts (termed F in the figure), from 100 to 2000. c 8i d, The same procedure was repeated using temporal radiality auto-correlations (TRAC4) for rulers of 10 to 80 nm. The frame count does not affect the TRA analysis as much as it affects TRAC4. The TRA performance, which is the parameter reported in most publications, is far poorer than that of TRAC4, when sufficient frames are analyzed.

[0462] Supplementary Fig. 4. SNR effect on SRRF performance, a, The top panel shows an overview of 30 frame-MIPs of an 80 nm ruler, followed by MIPs of the same ruler that were subjected to 2-fold, 5-fold, 10-fold and 20-fold increase in noise. Noise was added artificially, using a Matlab routine. The initial SNR was 27.84. The second panel shows TRAC4 analyses of the data. The third and fourth panels show a magni- fied region from the resonant scan MIPs, and their respective TRAC4 analysis results, b, The same analysis was performed on expanded GABAAR . Note that the receptor pore disappearing at x5 fold noise in TRAC4 resolved image. The nanoruler image is corrupted far more strongly by a 2-fold increase in noise than that of the GABAAR , owing to the substantially higher original SNR of the receptor image (76.72).

[0463] Supplementary Fig. 5. Technical scheme of the stabilization chamber used in this work. The exact measurements and materials for the stabilization chamber are in- cluded in the figure text. The 3D-printed gel cage patterning can be organized ac- cording to the user's preferred design. Only a suggested design is included here (many others work equally well). The exact design files can be obtained from the corresponding authors, to produce this chamber in any facility.

[0464] Item Specifications

[0465] (1) Chamber cover / EN AW 5083 [AIMg4,5Mn 0,7]

[0466] (2) 3DA-printed Gel cage / Pursa White Tough ink

[0467] (3) Coversltp 40 x 22 x 0,18 mm I Menzel-GI ser

[0468] U1Cnarn'jet holder < El I AW 5083 lA-Mg 5M;i 0 7]

[0469] Supplementary Fig. 6. TSR gallery, a, An example of a TSR. The first panel shows a ONE image of a TSR, the middle panel shows a cartoon model that fits the imaged TSR, and the third panel shows an overlay of the ONE image and the model, b, A gallery of TSRs (upper panels) and a best guess of cartoon models overlaid over the TSR images (lower panels).

[0470] Supplementary Fig. 7. Bleaching properties of fluorescein, Cy3, and STAR635P. a, A representation of the structures of each of the used dyes, followed by a table of their properties. The molecule structures and properties were reproduced from measurements of commercial providers: ^ttpsV / broadpharm.com / product / bp- 23900,2https: / / broadpharm.com / product / bp-22535, and3https: / / abberior.shop / ab- berior-STAR-635P. b, Normalized bleach curves from expanded specimens at 8000 Hz, and non-expanded specimens at 8000 Hz and 200 Hz.

[0471] Supplementary Fig. 8. ONE imaging of purified eGFP molecules, a, The first panel shows a ONE overview of eGFP molecules labelled with NHS-Ester STAR635P. The second panel shows a magnified area. The third panel shows the eGFP 1EMA PDB structure. The fourth panel shows a PDB / fluorescence overlay, b, A measurement of the apparent width and length of the molecules, from line scans as the examples shown in panel a, in blue and orange. A total of 17 single molecules were measured, c, A gallery of eGFP molecules.

[0472] Supplementary Fig. 9. Further ONE examples of immunoglobulin imaging, a, An overview of a field showing IgG antibodies labelled using NHS-fluorescein (left), along with a few zoom-in images of fluorescently-conjugated secondary IgG antibod- ies (right; Abberior Star635P conjugation shown in blue), b, Several examples of IgG antibodies imaged in different positions and perspectives, c, A gallery of the ex- pected antibody shapes, obtained by convoluting a PDB IgG structure with a ONE point-spread-function, after revolving the IgG molecules in 3D space randomly. A few enlarged views are shown, along with a multitude of small-sized views, to explain how IgG molecules should appear when they are visualized in fluorescence in ran- dom orientations. The typical IgG views are similar to the modeled ones, d, Fluores- cence (Abberior Star635P) and Coomassie SDS-PAGE gels indicating the size distribu- tion of antibody fragments. A mouse monoclonal primary antibody was run on the gels, along the secondary antibody imaged in panel (a). The gel was first imaged un- der a fluorescence (Cy5 channel) and then total proteins were revealed with Coo- massie brilliant blue staining. The results suggest that numerous small fragments are expected for both primary and secondary antibodies in the ONE images, not only full antibodies, due to impurities being present in the commercial antibody samples, e-f, An overview of IgA molecules, g-h, A similar overview of IgM molecules. The anti- body structures are shown using Pymol representations from PDB structures 1HZH, 1IGA, and 2RCJ. Supplementary Fig. 10. GABAA receptor and otoferlin galleries, a, An overview of images of GABAA receptors, b, The images display GABAA receptors in different 3D positions. The positional indications are best guesses performed by an experienced investigator, c, Overview images of otoferlin (right panel), and blank buffer as a con- trol (left panel), d, Otoferlin labelled with NHS-ester fluorescein ONE images in differ- ent 3D positions, e, Otoferlin labelled with NHS-ester STAR635P ONE images.

[0473] Supplementary Fig. 11. Calmodulin gallery, a, An overview of calmodulin ONE ac- quisitions in the presence and absence of calcium. This molecule was expressed and purified as a chimera containing mEGFP. The compact signal associated to the GFP molecule, as observed already in the TSR images in Figure 12, has a limited contribu- tion to the overall size of the molecule, b, Exemplary zoomed calmodulin ONE im- ages. The asterisk denotes the best guess of GFP molecule bound to calmodulin.

[0474] Supplementary Fig. 12. GABAAR nanobody labeling, a, Confocal images of ex- panded GABAAR labelled with anti-GABAAR nanobodies (NBs) conjugated to STAR635P. b, Confocal images of expanded GABAAR mixed with anti-eGFP nanobod- ies, which only induce little non-specific background, c, Magnified regions of single receptor either labelled with anti- GABAAR or anti-eGFP NBs. d, A gallery of ONE im- ages showing GABAAR in white and anti-GABAAR NBs in red.

[0475] Supplementary Fig. 13. Super-imposition of ONE microscopy images and Cryo-EM data, a, A cartoon view of the 50JM GABAAR / NB PDB. The red dots represent the 2 fluorophores on each nanobody, b, Cryo-EM images of representative 2D-classes of the GABAAR / NB complexes, derived from the same samples as used for ExM.. c, The first panel shows a ONE image of GABAAR / NB. The second panel shows a magnified region of a single receptor. The third panel shows a Cryo-EM / ONE overlay.

[0476] Supplementary Fig. 14. Drift compensation, a, A resonant confocal X10 image of otoferlin molecules at the start of a 1500-frame time series recording (first panel) and at its end (second panel). The third panel shows a maximum intensity projection of the resonant confocal scan. The blue arrow indicates the direction of drift. The fourth panel shows ONE processing without drift correction. A streak artefact is evi- dent as a result, b, Applying drift correction, using the SRRF software, to the same acquisition and maximum intensity projection yields an image (first panel) similar to the first image in (a). The second panel shows the result of the ONE processing with drift correction application. The last set of panels show magnified regions of otoferlin molecules. An otoferlin AlphaFold cartoon is presented for comparison (not drawn to scale). In panel 3, the ONE image is overlaid with its counterpart from the same da- taset, processed without drift correction (blue). Supplementary Fig. 15. ONE imaging of in vitro assembled microtubules, a, The upper panel shows in vitro-assembled microtubules that are stably fixed with 1-2.5% glutaraldehyde (GA). GA above 0.2% interfered with anchoring into the gels. The lower panel shows less stable microtubules, fixed with 8% PFA. These microtubules are deformed and tend to depolymerize, but this fixation does allow a reasonable de- gree of anchoring to the gels, b, ONE images of microtubules fixed with 8% PFA, ex- panded and labelled using NHS-ester fluorescein, c, A magnified region, d, An aver- aged analysis of side views of microtubule segments. The top panel shows an ideal image, obtained by convoluting the PDB structure of a microtubule segment (3J2U) with a fluorescent PSF, in which every amino acid is labelled fluorescently. The sec- ond panel shows a realistic model, in which sparser labeling is considered, and in which different microtubule segments are overlaid with slight tilt angle differences (up to 5°). The third panel shows an averaged processed ONE image of 175 partially depolymerized tubulin segments. The graph shows the respective line scans across each of the panels, e, A gallery of alpha-beta tubulin dimers that were left unpol- ymerized. f, The first panel shows a tubulin dimer reconstructed from 105 dimers, following the same procedure as for the GABAAR S, shown in Figure 14. The second panel shows a 1TUB PDB cartoon structure, and the third panel shows a ribbon dis- play of the molecules, for comparison.

[0477] Supplementary Fig. 16. A confocal analysis of synapses after MPCD treatments, a, Confocal images of hippocampal cultures immunostained for the three synaptic mark- ers employed in Figure 14a-c (Sytl, vGlutl and PSD95), relying on the same staining protocol as in Figure 14a-c. b, The panels show a magnified region. The culture mor- phology and synapse distribution are similar with or without MpCD treatments.

[0478] Supplementary Fig. 17. Sytl count analysis, a, The first panel shows a ONE over- view image of Sytl, VGIuTl, and PSD95 channels. The middle panel shows the Sytl channel alone, with two selected regions indicating signal from isolated antibodies. The two regions are magnified and displayed on top of each other in the third panel, b, The first graph shows the mean grey value of isolated spots in control and MPCD treated neurons. The second graph shows the mean grey value of vesicles. N = 22- 19, 2 independent experiments, Mann-Whitney test, p = 0.6507 for the first graph and p = 0.8494 for the second graph, c, A graph showing the number of Sytl anti- bodies per vesicle. Sytl antibody numbers were estimated by dividing the vesicle in- tensity value by single AB intensity value. Mann-Whitney test; p = 0.8937.

[0479] Supplementary Fig. 18. Intensity analysis. Specific nanobodies, which detect GABAAR S, are compared to non-specifically bound eGFP nanobodies and to back- ground noise, a, A set of images that shows GABAAR S bound to their respective nanobodies, GABAAR S + eGFP NBs, and a blank control, b, Fluorescence intensities were analyzed across the different conditions. N = 37, 28, and 33 images for GABAAR NB, eGFP NB and blank, respectively, from 2 independent experiments.

[0480] Supplementary Fig. 19. PSD95 model, a, To complement the distance analysis presented in Extended Data Fig. 7b, we analyzed the PSD95 distribution using a spot averaging procedure similar to a Ripley curve profile. To explain this analysis in more detail, we modeled it here. The top row of panels shows PSD-like spots, placed in a perfectly regular arrangement (left), with positions varying by 20 or 50% from per- fect regularity (middle), or placed randomly (right). The bottom rows of panels show average spots, obtained by overlaying the areas surrounding each of the individual spots in the model arrangements from the top panels. This procedure results in ar- rangements in which the central spot is surrounded by increasingly weak spots, with virtually no regular spots around it in the right-most panel, b, Lines were drawn from the center of each spot in the bottom panels in panel (a), in all directions, and were then averaged. The average line going from the center of a spot to the periphery shows a prominent peak if the arrangement is regular, since the neighboring spots are always present at a set distance, and thus provide a visible intensity peak. The less regular the arrangement is, the less clear the second peak becomes. It disap- pears completely when the spot positions are fully random.

[0481] Supplementary Fig. 20. Distribution of post-synaptic elements, a, A ONE overview of a hippocampal culture immunostained for GluA2, GluN2b and PSD95. b, A cartoon that color-codes the analysis presented in the lower two graphs. The lateral span of the GluN2b, GluA2 and PSD95 signals is presented in green, blue and red. The sec- ond graph shows the distance between the center of the GluN2b cluster and the GluA2 periphery. This implies that the GluN2b cluster is positioned relatively close to the center of the GluA2 distribution, since the value here is very close to the half of the GluA2 span. N = 8 post-synapses analyzed, 2 experimental replicates. Kruskal- Wallis test was applied, p = 0.0049 **.

[0482] Supplementary Fig. 21. ONE analysis of PSD95 labelled post-expansion. Neuronal cultures were fixed, expanded using the XIOht protocol and labelled after expansion using a PSD95 NB conjugated to Atto488. a, For comparison purposes, pre-expansion labelled PSD95 images are reproduced from Figure 16f (top panel) and Extended Data Fig. 6a (the other 3 panels) of the original manuscript, b, Post-expansion la- belled PSD95 examples, c, Exemplary line scans over pre- and post-labelled PSD95 show similar cluster spacing, d, Spot FWHM and spot distances measurements showed similar values. N = 140, and 113 measurements spot FWHM, and 402, and 172 for spot distances, for pre- and post-expansion, respectively. Mann Whitney test, non-significant p > 0.05. Supplementary Fig. 22. STED, ExSTED and ONE comparison, a, The synaptic pro- teins Bassoon, Homerl and PSD95 were imaged using STED, ExSTED (X10 expan- sion combined with STED), and ONE microscopy, b, The same procedure was ap- plied to tubulin.

[0483] Supplementary Fig. 23. Detailed analysis of Fourier ring correlation, a, FRC analy- sis of ONE images collected with a pixel size of 0.98 nm. The first panel row shows ONE images of the different specimens. The second row shows the corresponding FRC maps. The third row shows ONE images overlaid over FRC maps, using a screen-blend mode. The fourth and fifth rows show magnified views, b, A graph plot- ting the minimal FRCs in nm. c, A graph plotting the average FRCs in nm. Please note that all the labelled targets reside in the "bluest" regions of the map, indicating minimal FRCs that correspond to high resolution. N = 7, 8, 12, 7, 7, 7, and 7 for eGFP, tubulin dimer, actin, calmodulin, PSD95, Homer 1, and Shank2, respectively, d & e, FRC analysis of ONE images achieved with a pixel size at 0.98, 0.48, and 0,24 nm for GABAAR and otoferlin, respectively, f, The graphs shows minimal and aver- age FRCs in nm for GABAAR ONE images. N = 8, 6, and 9 for 0.98, 0.48, and 0.24 nm images, respectively, g, The graphs shows minimal and average FRCs in nm for otoferlin ONE images N = 10, 10, and 9 for 0.98, 0.48, and 0.24 nm images, respec- tively. All experimental sets were performed with at least 2 replicates.

[0484] Supplementary Fig. 24. Intra-molecular measurements, a, GABAAR ONE images acquired with 0.98, 0.48, and 0.24 nm pixel sizes, for the same region, b, GABAAR magnified examples from the first image in the panel above, c, One particular GABAAR molecule displayed at different resolutions, d, Equally-scaled otoferlin mole- cules acquired at different resolutions, e, ONE images of GABAAR and otoferlin at 0.24 nm overlaid with their respective PDBs. f, The graphs show 2 exemplary line scans for peptide segments in GABAAR and otoferlin. g, A graph showing peak-to- peak distances in Angstrom. N = 30 for GABAAR , and 30 for otoferlin, 10 independ- ent experiments for GABAAR S and 4 independent experiments for otoferlin.

[0485] Supplementary Fig. 25. Expansion precision evaluation, a, A direct compassion between single-molecule ONE images and their respective PDB / AlphaFold models. The purple line indicates the line scan used to measure the molecule dimension indi- cated in the first graph, b, The upper graph shows measurements of molecule di- mensions, in nm. The horizontal purple line indicates the expected value, obtained from measurements of PDB structures (for all molecules except otoferlin), or Al- phaFold predictions (for otoferlin). The lower graph shows the variability of these measurements, in the form of the coefficient of variance. N = 34, 17, 192, 75, 10, 10, 14, 8, and 18 for NB, eGFP, actin, tubulin dimer, GABAAR , otoferlin, IgG, IgA, and IgM; at least 2 experimental replicates were carried for all experiments. Paired t tests were carried out to determine whether the measured values are different from the values predicted by the PDBs; the respective p values are reported above the plots.

[0486] Supplementary Fig. 26. The nanobody imaging of ASYN objects is specific and is not easily reproduced by antibodies, a, Low-resolution images of CSF-containing samples, or blanks (clean, BSA-coated coverslips). Only a few dim spots, presumably representing single nanobodies, are seen in the blanks, b, Quantification of the signal intensity, as a sum across all image pixels. N = 7-9; Mann-Whitney test, p = 0.0002. c, Individual examples of oligomers immunolabeled with nanobodies (top) or anti- bodies (bottom), d, Averages of ASYN objects from individual patients, immuno- labeled with nanobodies or antibodies, e, An analysis of the average object size in antibody-labelled samples, as in Figure 15. N = 2 patients for each condition; the graph shows mean ± range of values. Nanobodies reveal differences between pa- tients, at object sizes of only a few nm. Antibodies have difficulties in this direction, as their large size causes a lower-fidelity labeling, and as their sizes obscure the ac- tual sizes of small objects.

[0487] Supplementary Fig. 27. ONE analysis of SARS-CoV-2 viral particles, a, ONE over- view of a sample containing SARS-CoV-2 viral particles immunostained against Spike Protein SI. b, More detailed views of two particles, indicating the Spike Protein SI and the native IgG molecules from the serum of the patients. Interestingly, a do- main-like structure is observed, which is presumably induced by the native IgGs gathering the spike proteins together, by the dual binding capacity of the IgG mole- cules.

[0488] Supplementary Fig. 28. FRC Analysis of SARS-CoV-2 and ASYN aggregates, a, FRC analysis of ONE images achieved at 0.98 and 0.78 nm per pixel for SARS-CoV-2 and ASYN, respectively. The first panel shows ONE images of the two specimens. The second panel shows the corresponding FRC maps. The third panel shows ONE im- ages overlaid with FRC maps, using screen-blend mode. The fourth panel shows magnified overlays. Please note that the FRC scale for SARS-CoV-2 ranges between 2.0 and 2.9 nm, with red color in this dataset still indicating very good FRC values, b, The graphs plot the minimal and average FRCs in nm. N = 10, and 8 for SARS-CoV-2 and ASYN, respectively, 2 experimental replicates.

[0489] Supplementary Figures 29. ONE microscopy applied at the confocal headquarters of Leica Microsystems and at the Center for Integrative Physiology and Molecular Medicine (CIPMM) of the Saarland University (UdS). As GABAAR were systematically investigated in this study, we chose them as a reference to evaluate the applicability of ONE technique at different laboratories using different systems, a, Using a STELLARIS 8 microscope at Leica Microsystems, we present a snapshot of a plane from a 5-dimension x,y,z,c,t image of GABAAR +NB. b, The first panel shows a depth projection of a zONE stack. The second panel shows a set of GABAAR S that were magnified. The optical sectioning of the first example is displayed in the rightmost panel, c, GABAAR S were also successfully imaged at CIPMM, Saarland University (UdS), as shown in 3 full-scale overviews and in their respective magnified regions. Several microscopes were used at the CIPMM, which are presented in the next fig- ure.

[0490] Supplementary Figures 30. GABAAR S could be imaged with different micro- scopes. Acquisition settings were matched among different systems to the level that each system allowed. The highest achievable speeds were used for each system. This was systematically characterized (data not shown, but can be presented upon request), a, Images from the first panel show GABAAR ONE images acquired from different microscopes. As the imaging systems were pushed to their speed limit, background noise was substantially higher on older models. The second panel shows ONE images with background subtraction. The third panel shows a magnified recep- tor example, b, Magnified regions showing the noise readings of each of the used mi- croscopes. c, A graph showing the achievable signal-to-noise ratio (SNR), as well as the SNR normalized to acquisition speed. Not surprisingly, higher SNRs yielded better ONE images. It is worth noting that ONE images conducted on a Zeiss LSM900 sys- tem (Supplementary Fig. 32) yielded a better SNR than LSM780 and LSM880, but the data was not combined with SNR analyses of GABAAR S, as the imaged target was different. N = 22, 22, 22, 23, and 23 for LSM780, LSM880, Abberior STED, SP5, and STELLAIRS 8, respectively, Kruskal-Wallis test was applied p < 0.0001 ****.

[0491] Supplementary Figures 31. ONE microscopy applied at the MIT, Cambridge, USA. Pre-expansion labelled tubulin specimens were expanded X10 and X20 and were im- aged on a STELLARIS 8 system. The X20 gel recipe was modified from ExR proto- col49 as follows. The first expansion gel components are 17.25% (w / v) sodium acry- late, 5% (w / v) acrylamide, ImM BIS, lx PBS, 0.05% (w / w) APS / TEMED. The re-em- bedding gel is composed of 10% acrylamide, 2.5 mM BIS, water, 0.05% (w / w) APS / TEMED. The second expansion gel is composed of 17.25% (w / v) sodium acry- late, 5% (w / v) acrylamide, ImM BIS, lx PBS, 0.05% (w / w) APS / TEMED. a, An over- view of tubulin from X10 and X20. b, The upper panel shows an X10 confocal image, with its respective ONE image in the lower panel, c, The first upper two panels show X20 confocal images of tubulin. Their respective ONE images are shown in the lower panel. Note the significantly dimmer images, as the expansion factor gets higher. The third panel shows a magnified region of X20 ONE, followed by an overlay of con- focal and ONE images. The graph shows normalized line scans across the tubulin width. The cyan curve shows the average line scan and the red curve is a fit using a double Gaussian formula. N = 26 line scans, d, A magnified region of X20 ONE image is shown. Another portion was magnified and displayed in a white box. The dotted pale-yellow lines are an estimation of the tubulin structure. Supplementary Figures 32. ONE microscopy applied at the University of Wurz- burg, Germany. Specimens were expanded X5.8 and were post-expansion labelled for tubulin, before imaging using an LSM900 microscope, a, The upper panels show a confocal overview and the lower panels show the respective ONE images, b, The up- per panel shows an X5.8 confocal image and its respective ONE image in the lower panel, c, The first image is a magnified region of X5.8 ONE followed by an overlay of a confocal and a ONE image. The graph shows normalized line scans across tubulin width. The cyan curve shows the average line scan and the red curve shows the re- spective fit. N = 20 line scans.

[0492] Supplementary Figures 33. Post-expansion bassoon labeling ONE images. Tissue sections were expanded and then labelled against bassoon following the expansion revealing (ExR) protocol49at the MIT, Cambridge, USA. a, An ExR20 (X20 expansion) confocal overview imaged with a x40 objective, b, Three different exemplary ONE images of bassoon using a x63 objective. The first image is a resonant scan MIP of 20 frames, followed by a ONE image, and an overlay with its respective confocal im- age. The white square indicates the magnified region to the right, c, Similar to (b) but using a xlOO objective.

[0493] Cited References

[0494] 1 Renz, M. Fluorescence microscopy-a historical and technical perspective. Cytometry A

[0495] 83, 767-779 (2013). / / doi.org: 10.1002 / cyto.a.22295

[0496] 2 Bond, C., Santiago-Ruiz, A. N., Tang, Q. & Lakadamyali, M. Technological advances in super-resolution microscopy to study cellular processes. Mol Ceil 82, 315-332 (2022).

[0497] > 7doi.org : 10, 1016 / j.molcel.2021.12.022

[0498] 3 Liu, S., Hoess, P. & Ries, J. Super-Resolution Microscopy for Structural Cell Biology.

[0499] Annu Rev Biophys 51, 301-326 (2022). https: / / doi.org: 10.1146 / annurev-biophys- 1025 912

[0500] 4 Schermelleh, L. et al. Super-resolution microscopy demystified. Nat Cell Biol 21, 72- 84 (2019). Hl b.-s / / HOi.org:10.1038 / s41556-018-0251-8

[0501] 5 Sigal, Y. M., Zhou, R. & Zhuang, X. Visualizing and discovering cellular structures with super-resolution microscopy. Science 361, 880-887 (2018).

[0502] I f .ii.org: 10.1126 / science.aaul044

[0503] 6 Gustafsson, N. etal. Fast live-cell conventional fluorophore nanoscopy with Image.] through super-resolution radial fluctuations. Nat Commun 7, 12471 (2016). https: / / doi.org : 10, 1038 / ncommsl2471

[0504] 7 Laine, R. F. et al. High-fidelity 3D live-cell nanoscopy through data-driven enhanced super-resolution radial fluctuation. bioRxiv, 2022.2004.2007.487490 (2022). I 'H O . / / u Tc o . U, > I O / :- J22.04.07.487490

[0505] 8 Balzarotti, F. etal. Nanometer resolution imaging and tracking of fluorescent molecules with minimal photon fluxes. Science 355, 606-612 (2017).

[0506] H r i >i 1 >Lc , 0 ' 6 / science.aak9913

[0507] 9 Bates, M. etal. Optimal precision and accuracy in 4Pi-STORM using dynamic spline

[0508] PSF models. Nat Methods 19 , 603-612 (2022). oi.org: 10.1038 / s41592-022-

[0509] 01465-8 Luciano A. Masullo, A. M. S., Lucia F. Lopez, Mauricio Pilo-Pais, Guillermo P. Acuna 8i Fernando D. Stefani. An alternative to MINFLUX that enables nanometer resolution in a confocal microscope. Light: Science & Applications (2022).

[0510] Weber, M. et al. MINSTED nanoscopy enters the Angstrom localization range. bioRxiv, 2022.2003.2018.484906 (2022). htps: / / doi.or '2022.03.18.484906

[0511] Lelek, M. etal. Single-molecule localization microscopy. Nat Rev Methods Primers 1 (2021). htps: / / doi.i 8 / s43586-021-00038-x

[0512] Helmerich, D. A. etal. Sub-10 nm fluorescence imaging. bioRxiv, 2022.2002.2008.479592 (2022). https: / / doi.orq:10.1101 / 2022.02.08.479592 Chen, F., Tillberg, P. W. & Boyden, E. S. Optical imaging. Expansion microscopy.

[0513] Science 347, 543-548 (2015). https: / / doi.org : 10.1126 / science.1260088

[0514] Cahoon, C. K. et al. Superresolution expansion microscopy reveals the three- dimensional organization of the Drosophila synaptonemal complex. Proc Natl Acad Sci USA 114, E6857-E6866 (2017). https: / / doi.orq:10.1073 / pnas.l705623114 Gambaroto, D. etal. Imaging cellular ultrastructures using expansion microscopy (U- ExM). Nat Methods 16, 71-74 (2019). https: / / doi.org: 10.1038 / s41592-018-0238-l Gao, M. et al. Expansion Stimulated Emission Depletion Microscopy (ExSTED). ACS Nano 12, 4178-4185 (2018). htps: / / doi.org: 10, 1021 / acsnano.8b00776

[0515] Halpern, A. R., Alas, G. C. M., Chozinski, T. J., Paredez, A. R. 8i Vaughan, J. C. Hybrid Structured Illumination Expansion Microscopy Reveals Microbial Cytoskeleton Organization. ACS Nano 11, 12677-12686 (2017). https: / / doi.o 021 / acsnano.7b07200

[0516] Kim, D., Kim, T., Lee, J. 8i Shim, S. H. Amplified Expansion Stimulated Emission Depletion Microscopy. Chembiochem 20, 1260-1265 (2019). https: / / doi.o ! / cbic.201800775

[0517] Li, R., Chen, X., Lin, Z., Wang, Y. 8i Sun, Y. Expansion enhanced nanoscopy. Nanoscale 10, 17552-17556 (2018). https: / / doi. » / c8nr04267e

[0518] Wang, Y. etal. Combined expansion microscopy with structured illumination microscopy for analyzing protein complexes. Nat Protoc 13, 1869-1895 (2018). htps: / / doi.QH; Hi tj !H / s41596-018-002.3-8

[0519] Xu, H. et ai. Molecular organization of mammalian meiotic chromosome axis revealed by expansion STORM microscopy. Proc Natl Acad Sci USA 116, 18423-18428 (2019). htps: / / doi.org : 10.1073 / pnas.19024"

[0520] Hell, S. W. 8i Wichmann, J. Breaking the diffraction resolution limit by stimulated emission: stimulated-emission-depletion fluorescence microscopy. Opt Lett 19, 780- 782 (1994). https: / / doi.orq:10.1364 / ol.l9.000780

[0521] Gustafsson, M. G. Nonlinear structured-illumination microscopy: wide-field fluorescence imaging with theoretically unlimited resolution. Proc Natl Acad Sci USA 102, 13081-13086 (2005). htps: / / doi.org: 10.1073 / pnas.0406877102

[0522] Betzig, E. et ai. Imaging intracellular fluorescent proteins at nanometer resolution.

[0523] Science 313, 1642-1645 (2006). https: / / doi.org: 10.1126 / science.1127344

[0524] Dertinger, T., Colyer, R., Iyer, G., Weiss, S. & Enderlein, J. Fast, background -free, 3D super-resolution optical fluctuation imaging (SOFI). Proc Natl Acad Sci U S A 106, 22287-22292 (2009). htps: / / doi.< 3 / pnas.0907866106

[0525] Culley, S., Tosheva, K. L., Matos Pereira, P. & Henriques, R. SRRF: Universal live-cell super-resolution microscopy. IntJ Biochem Cell Biol 101, 74-79 (2018). https: / / doi.org: 10..1016 / j.biocel.2018.05.014

[0526] Truckenbrodt, S. et aL X10 expansion microscopy enables 25-nm resolution on conventional microscopes. EMBO Rep lV (2018). Truckenbrodt, S., Sommer, C., Rizzoli, S. 0. 8i Danzl, J. G. A practical guide to optimization in X10 expansion microscopy. Nat Protoc lA, 832-863 (2019). https: / / doi.org: 10, 1038 / s41596-018-0117-3

[0527] Raab, M. et al. Using DNA origami nanorulers as traceable distance measurement standards and nanoscopic benchmark structures. Sei Rep 3, 1780 (2018). htps: / / doi.ou , 0, '.038 / s4159 . *’ 1 ’ '05-x

[0528] Klimas, A. et at. Magnify is a universal molecular anchoring strategy for expansion microscopy. Nat Biotechno! (2V23}. htps: / / doi.c, - i’ o’ 3 / s41587-022-01546-l Torres-Garcia, E. etal. Extending resolution within a single imaging frame. Nat Commun 13, 7452 (2022). htps: / / doi.org: 10.1038 / s41467-022-34693-9

[0529] Bani H. Cipriano, S. J. B., Renu Sharma, Dominic Rumore, Wonseok Hwang, Robert M. Briber, and Srinivasa R. Raghavan. Superabsorbent Hydrogels That Are Robust and Highly Stretchable. ACS Publications, 4445-4452 (2014). https: / / doi.org: https: / / doi.org / 10.1021 / ma500882n

[0530] Gao, R. etal. K highly homogeneous polymer composed of tetrahedron-like monomers for high-isotropy expansion microscopy. Nat Nanotechnol 13, 698-707 (2021). htps: / / doi.orq: 10.1038 / s41565-021-00875-7

[0531] Park, H. E. etal. Scalable and Isotropic Expansion of Tissues with Simply Tunable Expansion Ratio. Adv Sci (Weinh) 6, 1901673 (2019). htps: / / doi.org: 10.1002 / advs.201901673

[0532] Mikhaylova, M. et al. Resolving bundled microtubules using anti-tubulin nanobodies.

[0533] Nat Commun 3, 7933 (2015). / doi.org: 10.1038 / ncomms8933

[0534] Weber, K., Rathke, P. C. & Osborn, M. Cytoplasmic microtubular images in glutaraldehyde-fixed tissue culture cells by electron microscopy and by immunofluorescence microscopy. Proc Natl Acad Sci U S A 75, 1820-1824 (1978). https: / / doi.o as.75.4.1820

[0535] Sograte-Idrissi, S. et al. Nanobody Detection of Standard Fluorescent Proteins Enables Multi-Target DNA-PAINT with High Resolution and Minimal Displacement Errors. Cells 8 (2019). htps: / / doi.org: 10.3390 / cells8010048

[0536] Gotzke, H. etal. The ALFA-tag is a highly versatile tool for nanobody-based bioscience applications. Nat Commun 10, 4403 (2019). htps: / / doi.om m i038 / s41467-019-12301-7

[0537] Damstra, H. G. J. etal. Visualizing cellular and tissue ultrastructure using Ten-fold

[0538] Robust Expansion Microscopy (TREx). Elite 11 (2022). https: / / doi.o ’554 / eLife.73775

[0539] M'Saad, 0. 8i Bewersdorf, J. Light microscopy of proteins in their ultrastructural context. Nat Commun 11, 3850 (2020). https: / / doi.org: 10.1038 / s41467-020-17523-8 Culley, S. etal. Quantitative mapping and minimization of super-resolution optical imaging artifacts. Nat Methods 15, 263-266 (2018). https: / / doi.o 038 / nmeth.4605

[0540] Almada, P. et al. Automating multimodal microscopy with NanoJ-Fluidics. Nat Commun 10, 1223 (2019). htps: / / doi.org: 10.1038 / s41467-019-09231-9

[0541] Miller, P. S. 8i Aricescu, A. R. Crystal structure of a human GABAA receptor. Nature

[0542] 512, 270-275 (2014). htps: / / doi. !8 / naturel3293

[0543] Roux, I. et al. Otoferlin, defective in a human deafness form, is essential for exocytosis at the auditory ribbon synapse. Cell 127, 277-289 (2006). https: / / doi.orq: 10,1016 / j.cell.2006.08.040

[0544] Jumper, J. etal. Highly accurate protein structure prediction with AlphaFold. Nature

[0545] 596, 583-589 (2021). htps: / / doi.> , I > 38 / s41586-021-03819-2

[0546] Saal, K.-A. et al. Heat denaturation enables multicolor X10-STED microscopy at single-digit nanometer resolution. bioRxiv, 2022.2006.2024.497479 (2022). https: / / doi.orci: 10,1101 / 2022.06.24.497479 Miller, P. S. eta / . Structural basis for GABA(A) receptor potentiation by neurosteroids. Nat Struct Mo! Biot 24, 986-992 (20171 htps: / / doi. M / nsmb.3484

[0547] Sarkar, D. et a / . Revealing nanostructures in brain tissue via protein decrowding by iterative expansion microscopy. Nat Biomed Eng 6, 1057-1073 (2022). https: / / doi.org: 10, 1038 / s41551-0> ’

[0548] Chang, J. B. eta / . Iterative expansion microscopy. Nat Methods 14, 593-599 (2017). https: / / doi.o 038 / nmeth.4261

[0549] Svitkina, T. Imaging Cytoskeleton Components by Electron Microscopy. Methods Mo!

[0550] Biol 1365, 99-118 (2016). https: / / doi.orq:10.1007 / 978-l-493>

[0551] SHANNON, C. E. Communication in the Presence of Noise. PROCEEDINGS OF THE

[0552] IRE (1949). https: / / doi.orq:10.1109 / JRPRQC.1949.232969

[0553] Gan, Q. 8i Watanabe, S. Synaptic Vesicle Endocytosis in Different Model Systems.

[0554] Front Cell Neurosci 12, 171 (2018). https: / / doi.org: 10■3389 / fncel■2018■0,' 1

[0555] Wienisch, M. & Klingauf, J. Vesicular proteins exocytosed and subsequently retrieved by compensatory endocytosis are nonidentical. Nat Neurosci 9, 1019-1027 (2006). https: / / doi.o 038 / nnl739

[0556] Mutch, S. A. et a / . Protein quantification at the single vesicle level reveals that a subset of synaptic vesicle proteins are trafficked with high precision. J Neurosci 31, 1461-1470 (2011). htps: / / doi.orq:10.1523 / JNEUROSCI.3805-10.2011

[0557] Takamori, S. et a / . Molecular anatomy of a trafficking organelle. Celll27, 831-846 (2006). htps: / / doi.org: 10.1016 / j.cell.2006.10.030

[0558] Dani, A., Huang, B., Bergan, J., Dulac, C. 8i Zhuang, X. Superresolution imaging of chemical synapses in the brain. Neuron 68, 843-856 (2010). https: / / doi.o 016 / j.neuron.2010.11.021

[0559] Helm, M. S. eta / . A large-scale nanoscopy and biochemistry analysis of postsynaptic dendritic spines. Nat Neurosci 24, 1151-1162 (2021). htps: / / doi.o 038 / s41593-021-00874-w

[0560] Hayashi, Y., Ford, L. K., Fioriti, L., McGurk, L. 8i Zhang, M. Liquid-Liquid Phase

[0561] Separation in Physiology and Pathophysiology of the Nervous System. J Neurosci 41, 834-844 (2021). https: / / doi.orq:10.1523 / JNEURQSCI.1656-20.2020

[0562] Alberti, S., Gladfelter, A. 8i Mitag, T. Considerations and Challenges in Studying Liquid-Liquid Phase Separation and Biomolecular Condensates. Cell 176, 419-434 (2019). htps: / / doi.org: 10.1016 / j.cell.2018.12.035

[0563] Rich, A. The illustrated companion to the Latin dictionary, and Greek lexicon. 552 (Longman, 1849).

[0564] Biederer, T., Kaeser, P. S. & Blanpied, T. A. Transcellular Nanoalignment of Synaptic Function. Neuron 96, 680-696 (2017). https: / / doi.o j / j.neuron.2017.10.006

[0565] Spillantini, M. G. etal. Alpha-synuclein in Lewy bodies. Nature 333, 839-840 (1997). https: / / doi.o 038 / 42166

[0566] Uversky, V. N. A protein-chameleon: conformational plasticity of alpha-synuclein, a disordered protein involved in neurodegenerative disorders. J Biomol Struct Dyn 21, 211-234 (2003). htps: / / doi.orq: 10.1080 / 07391102.2003.10506918

[0567] Bruggink, K. A., Kuiperij, H. B., Ekholm-Petersson, F. & Verbeek, M. M. Detection of elevated levels of alpha-synuclein oligomers in CSF from patients with Parkinson disease. Neurology77 , 510; author reply 510-511 (2011). htps: / / doi.o 212 / WNL0b013e318219dd92

[0568] Mollenhauer, B. etal. Direct quantification of CSF alpha-synuclein by ELISA and first cross-sectional study in patients with neurodegeneration. Exp Neurol 213, 315-325 (2008). htps: / / doi.org: 10.1016 / j.expneurol.2008.06.004

[0569] Tokuda, T. etal. Decreased alpha-synuclein in cerebrospinal fluid of aged individuals and subjects with Parkinson's disease. Biochem Biophys Res Commun 349, 162-166 (2006). htps: / / doi.i 6 / j.bbrc.2006.08.024 De Genst, E. J. et al. Structure and properties of a complex of alpha-synuclein and a single-domain camelid antibody. J Mol Biol 402, 326-343 (2010). https: / / doi.org: 10, 1016 / j.jmb.2010.07.001

[0570] Guerrero-Ferreira, R. etal. Two new polymorphic structures of human full-length alpha-synuclein fibrils solved by cryo-electron microscopy. EHte S (2019). htps: / / doi.o,554 / elife.48907

[0571] Li, B. etal. Cryo-EM of full-length alpha-synuclein reveals fibril polymorphs with a common structural kernel. Nat Comman d, 3609 (2018). htps: / / doi.o ■ , i038 / s41467-018-05971-2

[0572] Winner, B. etal. In vivo demonstration that alpha-synuclein oligomers are toxic. Proc Natl Acad Sci U S A 108, 4194-4199 (2011). htps: / / doi.org: 10, 1073 / pnas.1100976108

[0573] Zhang, H., Griggs, A., Rochet, J. C. 8i Stanciu, L. A. In vitro study of alpha-synuclein protofibrils by cryo-EM suggests a Cu(2+)-dependent aggregation pathway. Biophys J 104, 2706-2713 (2013). htps: / / c 5 / j.bpj.2013.04.050

[0574] Wassie, A. T., Zhao, Y. 8i Boyden, E. S. Expansion microscopy: principles and uses in biological research. Nat Methods 16, 33-41 (2019). htps: / / doi.org: 10.1038 / s41592- 018-0219-4

[0575] Chen, B. C. etal. Lattice light-sheet microscopy: imaging molecules to embryos at high spatiotemporal resolution. Science 346, 1257998 (2014). https: / / doi.o i / science.1257998

[0576] Eilers, Y., Ta, H., Gwosch, K. C., Balzarotti, F. 8i Hell, S. W. MINFLUX monitors rapid molecular jumps with superior spatiotemporal resolution. Proc Natl Acad Sci U S A 115, 6117-6122 (2018). htps: / / doi.org: 10.1073 / pnas.1801671i i

[0577] Gwosch, K. C. etal. MINFLUX nanoscopy delivers 3D multicolor nanometer resolution in cells. Nat Methods 17 , 217-224 (2020). htps: / / doi,org:10.1038 / s41592-019-0688- 0

[0578] Wu, Y. etal. Resonant Scanning with Large Field of View Reduces Photobleaching and Enhances Fluorescence Yield in STED Microscopy. Sci Rep 5, 14766 (2015). https: / / doi.o ! / srep 14766

[0579] Dubois, B. et al. Research criteria for the diagnosis of Alzheimer's disease: revising the NINCDS-ADRDA criteria. Lancet Neurol 6, 734-746 (2007). https: / / doi.org: 10.1016 / 51474-4422(07)70178-3

[0580] Gibb, W. R. Accuracy in the clinical diagnosis of parkinsonian syndromes. Postgrad

[0581] MedJM, 345-351 (1988). https: / / doi.orq:10.1136 / pqmj.64.751.345

[0582] McKeith, I. etal. Dementia with Lewy bodies. Lancet Neurol 3, 19-28 (2004). htps: / / doi.o 016 / sl474-4422(03)00619-7

[0583] Gerdes, C. et al. k nanobody-based fluorescent reporter reveals human alpha- synuclein in the cell cytosol. Nat Commun 11, 2729 (2020). https: / / doi.org: 10, 1038 / S41467-0: n , . 75-0

[0584] Pullara, F. etal. k general path for large-scale solubilization of cellular proteins: from membrane receptors to multiprotein complexes. Protein Expr Purif 87 , 111-119 (2013). htps: / / doi.org: 10.1016 / j.pep.2012.10.007

[0585] Aricescu, A. R., Lu, W. & Jones, E. Y. A time- and cost-efficient system for high-level protein production in mammalian cells. Acta Crystallogr D Bio! Crystallogr 62, 1243- 1250 (2006). htps: / / doi.org: 10.1107 / 50907444906029799

[0586] Elegheert, J. et al. Lentiviral transduction of mammalian cells for fast, scalable and high-level production of soluble and membrane proteins. Nat Protoc 13, 2991-3017 (2018). htps: / / doi.org: 10, 1038 / s415hp o )

[0587] Masiulis, S. et al. GABAA receptor signalling mechanisms revealed by structural pharmacology. Nature 3&>, 454-459 (2019). https: / / doi.org: 10.1038 / s41586-018- 0832-5 Uchanski, T. etai. Megabodies expand the nanobody toolkit for protein structure determination by single-particle cryo-EM. Nat Methods 13, 60-68 (2021). https: / / doi.ob , , ; i, , > !8 / s41592-02.0-01001-6 Nakane, T. etai. Single-particle cryo-EM at atomic resolution. Nature 537, 152-156 (2020). htps: / / doi.org.10.1038 / s41586-020-28 t1' Zheng, S. Q. etai. MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy. Nat Methods 14, 331-332 (2017). htps: / / doi.org: 10, 1038 / nr1 5' Rohou, A. & Grigorieff, N. CTFFIND4: Fast and accurate defocus estimation from electron micrographs. J Struct Bio / 192, 216-221 (2015). https: / / doi.c sb.2015.08.008 Tegunov, D. & Cramer, P. Real-time cryo-electron microscopy data preprocessing with Warp. Nat Methods 16, 1146-1152 (2019). hl n . 7doi.orq:10.1038 / s41592 019- 0580-y Punjani, A., Rubinstein, J. L., Fleet, D. J. 8i Brubaker, M. A. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat Methods 14, 290-296 (2017). htps: / / doi.org: 10.1038 / nmeth.4169 Perego, E. et a / , K minimalist model to measure interactions between proteins and synaptic vesicles. Sci Rep 10, 21086 (2020). https: / / doi.org: 10.1038 / s41598-020- 77887-1 Li, K. The image stabilizer plugin for Imaged, 2008). Linkert, M. etai. Metadata matters: access to image data in the real world. J Cell Bio!

[0588] 189, 777-782 (2010). .orq : 10.1083 / icb.201004104 Grier, J. C. C. a. D. G. Methods of Digital Video Microscopy for Colloidal Studies. Journal of the Colloid and Interface Science, 298-310 (1996). https: / / doi.org: https: / / doi.org / 10.1006 / jcis.1996.0217

Claims

Claims1. Microscopy method for obtaining a high-resolution image of a biological sample comprising a substance to be analyzed, wherein an image sequence is obtained by expansion microscopy comprising the steps: a. providing a biological sample, b. mixing said biological sample with a swellable material and forming bonds, preferably covalent bonds, between said substance to be ana- lyzed and the swellable material, thereby forming a mixture, c. fragmenting said substance to be analyzed, d. expanding the mixture comprising said swellable material and said frag- mented substance to be analyzed, e. staining the sample either before or after steps (a), (b), (c), or (d) and f. imaging the obtained mixture under an optical microscope equipped with a camera or a resonant scanner.

2. The microscopy method of claim 1, comprising analyzing said image sequence by super-resolution radial fluctuations (SRRF), wherein the image sequence comprises 1000 frames or more and wherein the frames are ac- quired using an objective having a numerical aperture of 1.3 or higher.

3. The microscopy method according to claim 1 or 2, wherein signal fluctuations are measured by imaging the mixture repeatedly, using the resonant scanner, preferably using a frequency of 4 to 30 kHz, preferably 8 to 24 kHz, preferably imag- ing the sample of from 1000 to 100,000 times, preferably 1150 to 50,000 times, fur- ther preferably 1250 to 10,000 times, further preferably 1350 to 5000 times, further preferably 1400 to 3000 times or wherein fluorescence is measured by imaging the mixture repeatedly, using an opti- cal camera, preferably imaging the sample of from 1000 to 100,000 times, preferably 1150 to 50,000 times, further preferably 1250 to 10,000 times, further preferably 1350 to 5000 times, further preferably 1400 to 3000 times.

4. The microscopy method according to any of claims 1 - 3, wherein analyzing the image sequence by SRRF comprises dividing each pixel into a plurality of subpix- els.

5. The method of any of claims 1-4, wherein the method comprises: a. providing a purified solution comprising the biological molecule, preferably the proteinb. anchoring the biological molecule, preferably the protein, to a swellable agent of a swellable material c. optionally semi-drying the biological sample / protein-anchor agent on a glass coverslip until a thin film of solution remains on the glass coverslip d. homogenization of the sample by cutting peptides, preferably with proteinase K e. expanding the sample f. optionally labelling the expanded sample, preferably with an NHS-Ester with a first fluorophore g. optionally labelling of at least one specific, preferably more than two, most preferably 3-10 specific amino acids with fluorophores allowing discrimination h. optionally performing an imaging procedure, preferably according to any of claims 1-4 and / or wherein the acquisition of fluctuations is restricted by using stimu- lated emission depletion (STED), preferably by using 3 dimensional (3D) STED i. performing a super-resolution radial fluctuations (SRRF) analysis j. optionally creating a 2D model (2D shape recognition) or 3D model of labelled peptides using positions of the fluorescence signals to retrieve the overall structure of a protein, k. optionally correlating the 3D model of the labelled peptides with the positions of the specific amino acids optionally analyzing the positions of the specifically labelled amino acids to retrieve a unique sequence and / or 3D coordinates of those amino acids giving the fingerprint of a protein.

6. The method of any of the preceding claims, wherein performing SRRF analysis comprises setting parameters for the SRRF analysis, including:- setting a radiality magnification between 12 to 60; and / or- setting a number of ring axes between 2 and 8; and / or- setting a temporal analysis mode for Temporal Radiality Auto-Correla- tion to an order of 2 to 30; and / or- activate integration of temporal correlations; and / or- activate removal of positivity constraints; and / or- activate gradient smoothing; and / or- activate minimization of SRRF pattering; and / or- deriving the setting parameters for the SRRF analysis from a scanner specification included in metadata of the image stack; and / or- setting the theoretical pixel size below 100 nm; and / or- setting the scan frequency to achieve a scan duration between 5 ms and 200 ms.

7. The method of any of claims 1-6, wherein the labelling is performed by using at least 50x, or lOOx, or 150x, or 200x, or 250x, or 300x excess, preferably 200x to 300x excess of NHS-ester.

8. The method of any of claims 1-7 wherein providing an expanded biological sample as specimen for microscopy comprises a. attaching at least one anchor group derived to the biological sample, b. embedding the biological sample in the swellable material, c. fragmenting the biological sample embedded in the swellable material, and d. swelling the swellable material; characterized in that the at least one anchor group is capable of anchoring to at least one amino acid group in the biological sample.

9. The method according to any of claims 1 - 8, wherein the homogenization / frag- mentation comprises:- heating the composite to temperature of from 30 °C to 130 °C for a dura- tion of 5 hours or more- optionally treating the composite with an enzyme capable of fragmenting the biological sample, preferably proteinase K; further optionally simultane- ously treating the composite with an enzyme capable of fragmenting the biological sample, preferably proteinase K10. The method according to any of claims 1 - 9, wherein the biological sample is fragmented by incubation with proteinase K for at least 10 hours.

11. The method according to any of claims 5 - 10, wherein the anchoring agent for interacting with a biological sample, preferably a protein, is characterized as follows: a. the anchoring agent is compatible to a swellable material, preferably, wherein the anchoring agent covalently binds to said swellable mate- rial, b. the anchoring agent is suitable to be chemically, preferably covalently anchored to a functionalized or non-functionalized serine group of the biological sample, characterized in thatthe anchoring agent comprises a heterocycle comprising P, S, and 0, and a func- tional group selected from the group consisting of a double-bond containing group and a fluorophore.

12. A compound having formula (I)(I), wherein n is an integer of from 1 to 4, and whereinA is a selected from the group consisting of a double-bond containing group and a fluorophore-containing moiety (such as a fluorophore attached via a linker).

13. A mixture comprising a substance to be analyzed, e.g. a protein, containing at least one anchor group derived from the anchoring agent of claim 12 and a swellable material comprising a swellable agent.

14. A stabilization chamber for a gel, the stabilization chamber comprising a. a chamber holder b. a chamber cover c. a cage for holding a composition of a biological sample embedded in a swella- ble material, the cage comprising a cage patterning; characterized in that the cage patterning comprises polygonal, preferably pyramid and / or trapezoid shaped beams.

15. A method of capturing a plurality of images of a biological sample placed on an optical microscope and using an image sensor in the optical path of the micro- scope, the method comprising: setting an imaging specification including an image size and a theoretical pixel size; setting a scanner specification including a number of frames taken from the biological sample and a scan frequency at which the frames are scanned, wherein the number of frames is at least 300, preferably at least 500, more preferably more than 1000 and most preferably more than 1400 frames; and operating the image sensor according to the scanner specification,wherein the method preferably further comprises performing a super-resolution radial fluctuations analysis, SRRF analysis, on each of the number of frames.

Images