Method for continuous single molecule localization of conventional fluorescent antibody probes

Abstract

Method for single molecule localization microscopy (SMLM) based on conventional fluorescent binder probes, including antibodies, that are continually transported at low concentration over a sample under continuous illumination and microscopic acquisition.

Description

METHOD FOR CONTINUOUS SINGLE MOLECULE LOCALIZATION OF CONVENTIONAL FLUORESCENT ANTIBODY PROBES

[0001] The invention is directed to a method for single molecule localization microscopy (SMLM) down to Angstrom-level resolution of target epitopes in a static sample utilizing continuous active transport of conventional fluorescent binder probes (e.g. antibodies, nanobodies) across the sample that can bind to specific target epitopes balanced by simultaneous illumination and acquisition to photobleaching of the bound fluorescent probes. The continuous, directional delivery of fresh fluorescent probes to the sample importantly prevents the steady increase of a local concentration of prebleached free probes that can bind to and block target epitopes (as would occur in the absence of active transport) and, additionally, permits the use of simple widefield epifluorescence microscopy acquisition. The illumination must be of sufficiently low intensity to avoid excessive photobleaching of the free fluorescent probes as they travel varying distances across the illuminated portion of the sample before binding to target epitopes, with the exact intensity level limiting the maximal extent of probe prebleaching (e.g. limiting the prebleached fraction to <1%). The low illumination intensities required for the technique can be provided by an LED, obviating the need for more expensive, high-powered laser illumination required by other SMLM techniques. Furthermore, the use of widefield illumination, as opposed to axially restricted illumination (optical sectioning) required for other SMLM techniques, permits simultaneous volumetric monitoring of both in-focus as well as out-of-focus probes out to roughly ±5 pm axial displacements from the focal plane. Finally, at the low illumination intensities required for the technique, substantially more photons can be collected from each bound fluorescent probe before it photobleaches, facilitating the widefield detection of highly out-of-focus probes with extended point spread functions (PSFs) and, even more significantly, permitting probe localizations down to few Angstrom resolution.BACKGROUND

[0002] Conventional approaches to SMLM are based on the sparse detection of fluorescent binder probes that label specific target epitopes within a fixed biologicalsample. The following three conventional approaches to SMLM were all first proposed in 2006: a. PhotoActivated Localization Microscopy (PALM)Betzig et al., Science 313, 1642 (2006) b. STochastic Optical Reconstruction Microscopy (STORM)Rust, Bates, & Zhuang, Nature Methods 3, 793 (2006) c. Point Accumulation for Imaging in Nanoscale Topography (PAINT)Sharonov and Hochstrasser, PNAS 103, 18911 (2006)

[0003] Three distinct strategies for sparse detection are employed by these approaches, namely photo-activation (PALM), chemical quenching (STORM), and transient binding (PAINT).

[0004] For PALM and STORM, target epitopes in the sample are first bound with a fluorophore-labeled antibody. For PALM, cycles of photo-activation followed by acquisition of the sparsely photo-activated probes until they are all photobleached are then performed. For STORM, after antibody staining, the sample is immersed in a quenching buffer that places the vast majority of the probe fluorophores in a dark state. A movie of the stochastic blinking of transiently active fluorophores is then continuously recorded until most or all fluorophores are irreversibly photobleached. In both cases, the target epitopes are bound with high labeling efficiency at the outset followed by detection of the sparsely distributed bright fluorophore fraction.

[0005] For PAINT, there are three principal implementations utilizing binding to super-resolve specific target epitopes in a fixed biological sample, namely i. DNA-PAINTJungmann et al., Nature Methods 11, 313 (2014) ii. Image Reconstruction by Integrating exchangeable Single-molecule localization (IRIS)Kiuchi et al., Nature Methods 12, 743 (2015) iii. universal PAINT (uPAINT) on fixed cells Gunasekara et al., Bioconjug Chem 34, 825 (2023)

[0006] For DNA-PAINT, like PALM and STORM, the first step is the binding of an antibody to the sample at high labeling efficiency. In this case, however, the antibody is conjugated not to a fluorophore but to a short oligo (e.g. 10 - 20 nucleotides in length). The sample is then immersed in a buffer containing a sufficiently low concentration of a complementary oligo conjugated to a fluorophore, known as the imager strand. Transientbinding events of the imager strand to the bound antibodies, with a binding half-life up to about 10 seconds (US 11,536,715 B2), can then be resolved over the background of the free fluorescent imager strand using optical sectioning microscopy (e.g. using total internal reflection fluorescence (TIRF) or HILO), as was also used for the initial PAINT publication (Sharonov and Hochstrasser, PNAS 103, 18911, 2006).

[0007] For IRIS, unlike PALM, STORM, and DNA-PAINT, the sample is not prelabeled with an antibody; instead, the sample is immersed in a buffer containing a sufficiently low concentration of fluorophore-labeled probes that themselves are capable of transiently binding in a direct and specific manner to the target epitopes with a binding half-life up to 3 seconds (US 10,858,688 B2). More recently, probes with a half-life up to 7.13 seconds have been demonstrated for IRIS (Zhang et al., Cell Reports Methods 2, 100301, 2022); probes with significantly longer half-life are considered suboptimal for efficient SMLM acquisition using IRIS. As for DNA-PAINT, optical sectioning (e.g. TIRF or HILO) is employed to minimize the background arising from free fluorescent probe and thereby optimally image and super-resolve individual probe binding events.

[0008] Both DNA-PAINT and IRIS are based on the monitoring of transient binding events. For both techniques, optical sectioning microscopy is considered essential to reduce the background of free probe at a high concentration of applied probe that allow for optimal and efficient SMLM acquisition. Widefield microscopy is in particular disfavored, as the illumination intensity required to capture sufficient photons before each transiently bound probe unbinds (on the roughly less than one second timescale) would, under widefield illumination, heavily bleach the out-of-focus pool of freely diffusing probes before they can even reach the sample, thereby significantly reducing the density of fluorescent binding events. To avoid this photobleaching problem, optical sectioning using TIRF microscopy is typically utilized to restrict illumination of the sample to roughly <100 nm from the glass. Simple diffusion of the probe from the sides or above the illuminated volume (defined by the 100 nm depth of the evanescent field from the glass) is then sufficient to supply the 3D illuminated volume within the sample with fresh fluorescent probe. Even if some amount of free fluorescent probe is prebleached before binding, the rapid unbinding of this dark probe (typically on the less than 1 second timescale) would not limit the maximum achievable labeling efficiency, as the same epitope can be later bound by a new fluorescent probe.

[0009] For the third PAINT-based technique of uPAINT, a biological sample (e.g. adherent cells) is immersed in a solution containing a low concentration of a conventionalfluorophore-labeled antibody. Continuous illumination and imaging of sparsely bound antibody probes leads eventually to their photobleaching, with the photobleaching of bound probes continually balanced by the diffusion and binding of free probes to unoccupied epitopes on the sample. uPAINT differs from all other PAINT techniques in that the binders, typically conventional antibodies, should not be highly transient in their binding. uPAINT was first demonstrated on live biological samples for the tracking of sparsely bound receptors with HILO microscopy (Giannone et al., Biophysical Journal 99, 1303, 2010), with the super-resolved trajectories of single receptors allowing measurement of mobility at the single molecule level. More recently, uPAINT has been extended to the super-resolved imaging of static epitopes in fixed cells (Gunasekara et al., Bioconjug Chem 34, 825, 2023). Gunasekara et al. applied a low concentration of probe to a fixed sample and then continuously illuminated and imaged the fixed sample on a TIRF microscope, allowing for acquisition of each spontaneously bound probe until it photobleached, generating high localization precision for the static epitopes across the fixed sample. In both applications of uPAINT to either live or fixed samples, passive diffusion is relied on to transport the probes to epitopes in the sample.

[0010] For uPAINT, optical sectioning is again considered essential to avoid a too high background for acquisition created by the freely diffusing fluorescent probes. Optical sectioning based on restricted illumination of the sample (e.g. using TIRF and HILO) reduces — but does not entirely eliminate — photobleaching of the freely diffusing probes as compared to widefield illumination. For super-resolution of static epitopes with uPAINT, Gunasekara et al. additionally note the advantage of collecting all photons from each binding event until the probe is photobleached, similar to PALM. However, an advantage over PALM, which is a single fluorophore technique, is the applicability of uPAINT as well to probes conjugated with multiple fluorophores (i.e. with degree of labeling, DOL > 1), with all photons from the probe collected until every fluorophore on the probe is photobleached. An additional advantage over PALM is the applicability of uPAINT to generic fluorophores and not just photoswitchable fluorophores. Despite these potential advantages, significantly higher resolution using uPAINT compared to standard SMLM approaches has not yet been demonstrated.

[0011] There are three significant limitations to the previously reported applications of uPAINT for super-resolved imaging of target epitopes in a fixed sample. First, uPAINT requires laser-based optical sectioning using TIRF or HILO, with TIRF restricting observation to the vicinity of the coverslip glass (<100 nm) and HILOrestricting observation to a tilted light sheet that is inhomogeneous across the focal plane in the sample. Second, while photobleaching of the free probes is reducedoptical restriction of the illumination in TIRF and HILO, it is not eliminated and can still be significant over the relatively long acquisition timescales required for SMLM, leading to an ever-increasing local concentration of freely diffusing yet photobleached probes (see below discussion of Figure 1). These photobleached probes can bind to unoccupied epitopes on the sample, reducing the maximum achievable labeling efficiency for uPAINT.

[0012] The conventional resolution of a microscope is defined here as either the full width at half max (FWHM) or the standard deviation, <J, of the PSF, with FWHM « 2.35<J. For a microscope setup employing a high numerical aperture objective (e.g. N.A. = 1.46), the resolution is approximately FWHMxy= 200 - 235 nm (oxy= 85 - 100 nm) laterally and approximately FWHMZ= 500 - 600 nm (<JZ= 213 - 255 nm) axially over the wavelength range 570 - 670 nm. The lateral localization accuracy achieved by the above-listed approaches to SMLM can under special circumstances range down to a few nm, but is more typically on the order of 10 - 20 nm. For SMLM, sparse detection of single fluorophore-labeled probes in the sample allows assignment of all detected photons emanating from a spatially isolated event to a single labeled epitope. For a single detected photon, the purely statistical localization accuracy (in the absence of additional camera readout noise and background contributions) would be equivalent to the conventional resolution. For the detection of N photons from a single probe, the obtainable lateral resolution, limited only by statistics, would be equivalent to the standard lateral resolution divided by the square root of the number of detected photons, or On purelystatistical grounds, therefore, at a detection wavelength of 670 nm, 10,000 photons would be the absolute minimum required to achieve a lateral resolution of ° = 1 nm(“Angstrom resolution”). For standard approaches to SMLM, the typical number of detected photons per localization event is on the order of only a few hundred to a few thousand, well below the statistical limit of 10,000 required to even statistically obtain Angstrom resolution. Camera readout noise, vibrations, drift, and background add of course additional systematic uncertainties, but these can be well-controlled on modem SMLM setups (using state-of-the-art cameras, active vibration damping, and very sparse detection) such that the achieved resolution is principally limited by the amount of detected photons per event. As stated above, photon collection closer to or over this 10,000 photon limit may theoretically be possible with the recent application of uPAINTto fixed cells (Gunasekara et al., Bioconjug Chem 34, 825, 2023), but this has not yet been proven.

[0013] Improvement in the conventional axial resolution of at best FWHMZ= 500 - 600 nm for conventional approaches to SMLM is also achievable down to roughly 50 - 60 nm, e.g. by passing the emitted light through a refractive element (e.g. a cylindrical lens) or a diffractive element (SELFI) before detection, or by using one or more optical splitting elements to detect the emission light at multiple focal planes (biplane detection). A more complicated approach that yields an even higher increase in axial resolution is 4Pi microscopy, which is based on using two objectives mounted above and below the sample. For axial resolution, as well, the achievable localization accuracy will be principally limited by the number of detected photons. Higher photon counts would therefore also improve the axial resolution, with the axial resolution scaling, as for the lateral resolution, as the inverse of the square root of the number of detected photons.

[0014] Two noteworthy advances have been introduced in recent years that extend the achievable resolution of SMLM down to the Angstrom level, namely i. MINimal fluorescence photon FLUXes microscopy (MINFLUX) Balzarotti et al., Science 355, 606 (2017) ii. Resolution Enhancement for Sequential Imaging (RESI) Reinhardt et al., Nature 617, 711 (2023)

[0015] MINFLUX is fundamentally different from the other SMLM approaches, as it is based on point detection using STimulated Emission Depletion (STED) microscopy in conjunction with STORM-based SMLM. Here, each individual sparsely activated fluorophore is “triangulated” in a series of steps by applying a minimum excitation flux that samples only the perimeter of the PSF for the fluorophore. While MINFLUX works fast enough to even track single particles in live cells down to few Angstrom resolution, it is nevertheless fundamentally limited in speed (throughput) by its reliance on point detection compared to the highly parallel, simultaneous detection of multiple probes over the entire area of the field of view acquired by a camera. The main advantage of MINFLUX is therefore for the super-resolved tracking of individual proteins down to Angstrom resolution in live cells, which lies outside the scope of the current patent.

[0016] The second Angstrom-resolution technique listed above called RESI, on the other hand, is based on a simple extension of DNA-PAINT. In conventional DNA-PAINT, individual binding events achieve a resolution of roughly 10 - 20 nm. Typical protein sizes are, however, on the order of a few nm. As many, if not most, protein targets areclustered in cells, it is difficult to confidently assign binding events, that individually attain a resolution of only down to 10 - 20 nm, to the correct target protein. RESI works by statistically addressing this “assignment” problem. For RESI, instead of a single antibody-oligo conjugate used to label a single target epitope (as for DNA-PAINT), a library of antibody-oligo conjugates is created with always the same antibody domain but different oligo sequences: oligol, oligo2, oligo3, etc.. Labeling the cells with a mixture of these conjugates then achieves a “stochastic” or statistical labeling of the sampling that is then less dense for any given bound conjugate pool. Instead of only one acquisition step based on applying a single complementary oligo imager strand to the sample (as for DNA- PAINT), multiple acquisition steps are employed in a sequential fashion for each antibody-oligo conjugate. In each acquisition step, as the density of labeling of the sample with any specific antibody-oligo is lower (by a factor equal to the number of distinct oligos), one can more confidently assign all events localizing to 10 - 20 nm of a particular location to the same target epitope. RESI has been shown to achieve few Angstrom resolution on synthetic repeated structures (DNA origami) and on biological oligomeric structures like the nucleopore complex. The main flaw of RESI is that “stochastic” or statistical labeling cannot guarantee that each antibody-oligo is spaced more than 10 - 20 nm from every other identical antibody-oligo labeling the target epitopes in the sample. For example, consider a protein that forms a homodimer and an antibody that labels these epitopes with near 100 percent efficiency. With only 10 distinct oligos used for the antibody-oligo library, roughly one out of 10 homodimers would be potentially labeled with the same antibody-oligo. Even with 100 distinct oligos, roughly one percent of homodimers would still on average be labeled with the same antibody-oligo. For a single target epitope, a serious disadvantage of RESI is the requirement for multiple distinct oligo sequences that exhibit minimal cross-talk (a significant challenge for large oligo libraries) and their laborious consecutive application to the sample in multiple rounds of DNA-PAINT acquisition. RESI is therefore poorly suited for serial Angstrom resolution of distinct targets in a single biological sample.

[0017] A novel approach to SMLM capable of achieving Angstrom resolution has very recently been proposed by the author in a patent submission (Kinkhabwala, WO 2025 / 125533 Al) entitled “A METHOD FOR SINGLE MOLECULE LOCALIZATION MICROSCOPY AT ANGSTROM RESOLUTION BASED ON CONVENTIONAL FLUOROPHORE-LABELED ANTIBODY PROBES”. Conventional Antibody Localization Microscopy, or CALM, is based on cycles of sparse binding of the samplewith a conventional fluorescent antibody (or antibody-like) probe used for immunofluorescence (IF), washing of the sample, and then illumination and image acquisition until irreversible photobleaching of the sparsely bound fluorescent probes. In CALM, as all light is collected from the probes until they are photobleached, photon counts per probe of well over the statistical threshold of roughly 10,000 photons can be obtained, allowing for single-shot Angstrom-level resolution of the isolated probes. As multiple fluorophores can be used to label a single probe (DOL > 1), even higher photon counts per probe, reaching into the millions, are potentially realizable. For CALM, separate steps of staining, washing, and imaging are applied in a repeated fashion until a targeted percentage of the epitope is labeled in the sample. CALM can be performed rapidly through use of a microfluidic pump, with the staining and washing steps pumped over a static biological sample mounted in an enclosed flow chamber with inlet and outlet. Localizations to Angstrom resolution of the probes — reaching most or all of the available targeted epitope in the sample — can be achieved on the roughly few hour timescale, depending on the number of epitopes. To avoid bleaching of the free fluorescent probe during the staining and washing steps, the illumination should be turned off. Photobleaching of free fluorescent probe could otherwise lead to binding of the now dark probe to the sample, thereby blocking, either permanently (stable CALM, sCALM, based on stable antibody binders) or temporarily (semi-stable CALM, ssCALM, based on semistable binders), specific target epitopes and limiting the achievable labeling efficiency. Once the sample has been thoroughly washed, no more free fluorescent probe should remain, allowing the illumination to be turned on and only the bound probe signal to be acquired.

[0018] CALM shares some similarities with the recent extension of uPAINT to fixed cells described above (Gunasekara et al., Bioconjug Chem 34, 825, 2023); namely, both techniques are based on the gradual application of an antibody to a sample and the acquisition of sparse binding events until probe photobleaching. However, CALM has five significant advantages: (1) Microfluidic separation of the staining, washing, and acquisition steps avoids any prebleaching of the free pool of antibody implying no limit on the achievable labeling efficiency. (2) Separation of the staining and acquisition steps further implies no need for optical sectioning by sophisticated microscopy techniques like TIRF or HILO; simple widefield microscopy is sufficient. (3) Widefield illumination and detection at high photon collection potentially enables simultaneous acquisition of “out-of- focus” probes located out to a few micron above or below the focal plane (in contrast tothe <100 nm restriction for TIRF). (4) Acquisition in the absence of a free probe background yields a statistically higher localization accuracy for the same number of detected photons. (5) The absence of free probe background additionally implies that the density of localizations in each step can be adjusted independently of the on-rate constant for the probe (in the presence of free fluorescent probe, the on-rate constant would, along with the applied concentration of the free probe, specify the amount of background for a desired density of probes simultaneously binding the sample).

[0019] Separation of the staining and washing steps from the imaging steps in CALM, however, requires a specialized setup with tight and accurate coordination of the microfluidic handling and image acquisition. An even simpler implementation of CALM that would obviate the need for such tight instrumental coordination, yet still limit the risk of the binding of photobleached probe to the sample using simple widefield microscopic acquisition, would be desirable.OBJECT OF THE INVENTION

[0020] In the current invention, a new and simpler variant of CALM, called continuous flow CALM (cfCALM), is introduced that is preferentially based on free- running transport of a very low concentration of the probe over the sample along with continuous image acquisition. For cfCALM, sparse yet dense probe localizations down to few Angstrom statistical resolution can be continuously acquired over a region of interest (ROI) within the sample and over long timescales by balancing the photobleaching of the continually illuminated bound probes with simultaneous delivery and binding of new fluorescent probes to target epitopes in the sample. As the concentration of the applied free fluorescent probe is very low, the background is also correspondingly low, allowing for robust detection of the bound fluorescent probes even on a widefield microscope setup. Unlike other SMLM approaches based on high-power laser illumination, the low illumination intensities required by cfCALM are reachable with LED illumination, further simplifying and reducing the cost of the instrumental requirements. The illumination intensity for cfCALM should not be too high to limit the risk of the binding of already photobleached probe to the sample, implying a photobleaching timescale significantly longer than the timescale of probe transport and diffusion to target epitopes in the sample. As fresh fluorescent probes continuously enter the illuminated portion of the sample from upstream, no steadily growing accumulation of freely diffusing photobleached probe in the volume including and surrounding the illumination can take place (as it does for uPAINT),with therefore no significant limit on the achievable labeling efficiency of the epitope (see Figure 1). Low, gentle illumination (e.g. provided by an LED) has the additional benefit of allowing for even higher photon collection efficiency from each probe and consequently higher localization precision.

[0021] Microfluidics has been used in the past for SMLM in various ways, with a particular early example being the microfluidic exchange of one transiently binding DNA- PAINT probe for another on the sample to enable serial acquisition of different epitope targets (Jungmann et al., US 2016 / 161472 Al). However, the realization that continuous transport of a conventional antibody probe at low concentration to a sample would allow for sparse sample labeling at arbitrarily high labeling efficiency and, additionally, would permit the use of simple widefield microscopic acquisition (e.g. enabling simultaneous volumetric acquisition of probes out to ±5 pm axial displacements) has not been previously reported and constitutes the principal novelty of the current invention.

[0022] As well, while passive diffusion-based delivery and binding of conventional antibody binders to a sample has previously been shown to enable the sparse labeling required for SMLM (specifically, the recent extension of uPAINT to fixed cells reported by Gunasekara et al., Bioconjug Chem 34, 825, 2023), there are six important distinctions of the current invention of cfCALM relying on active transport of the probe to the sample: The novel combination of (1) continuous transport and (2) low illumination intensity together permit probe localization with (3) no significant limit on labeling efficiency and (4) the use of simple widefield microscopic acquisition that, in combination with (5) higher photon collection efficiencies at low illumination intensity, enables (6) simultaneous volumetric detection of both in-focus and out-of-focus probes out to axial displacements of approximately + / - 5 pm.

[0023] Finally, in comparison to the previously reported invention of CALM (Kinkhabwala, WO 2025 / 125533 Al), cfCALM achieves similar localization accuracies but without the need for tight coordination of microfluidic control (staining and washing steps) and microscopic image acquisition. cfCALM is much simpler to implement, as both the pumping and imaging processes are performed in a completely independent and continuous fashion.BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 shows observational setups for uPAINT based on either TIRF or widefield microscopy, and for cfCALM using widefield microscopy.

[0025] Figure 2 shows a schematic of a preferred experimental setup for cfCALM.

[0026] Figure 3 : Depiction of the illuminated area at the focal plane and the ROI, as well as determination of the maximum transport distance, / ., from the upstream boundary of the illuminated area at the focal plane to the downstream boundary of the ROI. (A) Rectangular geometry. (B) More general geometry.

[0027] Figure 4 shows conventional IF vs. cfCALM super-resolution of tubulin in fixed HeLa cells.

[0028] Figure 5 shows the simultaneous detection of both in-focus and out-of- focus probes within a fixed tissue slice.DESCRIPTION OF THE INVENTION

[0029] Detailed here is a method called cfCALM for SMLM based on conventional fluorescent binder probes (e.g. antibodies, nanobodies) that are delivered to targets in a static sample using continuous transport and simultaneous illumination and microscopic image acquisition to photobleaching. cfCALM, like the earlier CALM invention, is based on construction of a snapshot image at down to Angstrom resolution of a static sample (which could, for example, be a fixed biological sample). Super-resolved tracking of proteins in live cells, as is possible with the original uPAINT method or MINFLUX, is outside the scope of the current invention.

[0030] Object of the invention is therefore a method for determining the position of a target on a static sample by providing a medium containing a probe comprising a first component capable of binding directly and specifically to the target bound to a second component comprising a fluorescent substance by the steps:A. providing the medium to the static sample such that the probe binds at least transiently to the target,B. illuminating the static sample with a light source, thereby generating emission light from the fluorescent substance of the probe,C. acquiring one or more images of a region of interest (ROI) contained within an illuminated area at a focal plane of the static sample comprising emission light,D. determining the position of the probe observed over the one or more images to estimate the position of the target on the static sample, characterized in that the binding of the probe to the target has a first binding half-life greater than 15 seconds and that the probe is actively transported over the sample such thatthe maximum transit time of the probe from the upstream boundary of the illuminated area at the focal plane to the downstream boundary of the ROI is less than 20 times of a photobleaching half-life of the fluorescent substance of the probe.

[0031] In other words, the method of the invention is directed to locating a target in a static sample using a medium containing a probe comprising a first component capable of binding directly and specifically to at least part of the target with a first binding half-life greater than 15 seconds and a second component containing a fluorescent substance, characterized by the steps: a. bringing the medium into contact with the static sample and actively transporting the probe to the static sample, b. illuminating the static sample with a light source, c. acquiring one or more images of the emission light from the fluorescent substance, d. determining the position of the probe observed over the one or more images to estimate the position of the target.

[0032] Preferable, the static sample is a biological sample like a tissue slice or cell sample immobilized on a substrate.

[0033] In the drawings the following reference numbers are used to refer to the following features. Similar reference numbers are used in the various figures to refer to components that serve a similar or identical function.001 Sample well002 Sample well wall003 Static sample004 Solution containing probe005 Coverslip glass006 Microscope objective lens007 2D coordinate system displaying one lateral coordinate, x, and the axial coordinate, z 008 TIRF illumination volume for uPAINT009 Probe photobleaching gradient arising from TIRF illumination volume for uPAINT 010 Contrast bar indicating the fraction of fluorescent probe, / 011 Upstream boundary of illumination cone for widefield microscopy012 Downstream boundary of illumination cone for widefield microscopy013 Illuminated area at the focal plane014 Upstream boundary of the illuminated area at the focal planeDownstream boundary of the illuminated area at the focal plane Probe photobleaching gradient arising from the widefield illumination cone Sample chamber Sample chamber ceiling S ampl e chamb er ini et Sample chamber outlet Direction of probe transport Fresh fluorescent free probe directly upstream of the focal plane Partly photobleached free probe directly downstream of the focal plane Plot of the fraction of fluorescent free probe, over the extent of the focal plane as a function of the lateral coordinate, x Line indicating the value of unity for the fraction of fluorescent free probe, / Line indicating the fraction of fluorescent free probe, / as a function of lateral position, x Sample chamber length Sample chamber height Sample chamber width Free fluorescent probe Antibody Fluorescent substance 3D coordinate system Region of interest (ROI) contained within the illuminated area at the focal plane Upstream boundary of the ROI Downstream boundary of the ROI Axial axis Opening angle of illumination given by the numerical aperture of the objective Free photobleached probe Photobleached fluorescent substance Bound fluorescent probe Bound photobleached probe that upon binding was already photobleached Bound photobleached probe that upon binding was still fluorescent 2D coordinate system displaying lateral axes x and j’ Maximum length, L, from the upstream boundary of the illuminated area at the focal plane to the downstream boundary of the ROI046 Length, Zi, defined along the x axis at fixed yi from the upstream boundary of the illuminated area at the focal plane to the downstream boundary of the ROI047 Length, Z2, defined along the x axis at fixed yi from the upstream boundary of the illuminated area at the focal plane to the downstream boundary of the ROI048 Length, Z3, defined along the x axis at fixed j’3 from the upstream boundary of the illuminated area at the focal plane to the downstream boundary of the ROI049 Redisplay of lengths Zi, Z2, and Z3 aligned at their left edges, with Z2 the longest 050 Equation determining the maximum length, Z, for an arbitrarily shaped ROI contained within an arbitrarily shaped illuminated area at the focal plane051 Single tubulin filament in HeLa cell using conventional IF052 Single tubulin filament imaged with cfCALM053 Boxed region containing filament of interest054 Tubulin epitope 1 (located within filament of interest)055 Tubulin epitope 2 (adjacent to filament of interest)056 Tubulin epitope 3 (located within filament of interest)057 In-focus probe058 Out-of-focus probe059 Boxed region containing an out-of-focus probe

[0034] In one embodiment, the method could be characterized by actively transporting the probe to the static sample through the use of flow, an electric field, a magnetic field, or gravity.

[0035] Directional transport of the probe over the sample is a critical aspect of the current invention, as it provides the static sample with a continuous supply of fresh fluorescent probe with no limit on the achievable labeling efficiency (in contrast to uPAINT, which relies solely on diffusion to deliver the probe to the sample and for which a photobleached pool of probe is generated over time in the vicinity of the illuminated portion of the static sample, as depicted in Figure IB). For a preferred aqueous medium, flow of the aqueous medium would satisfy the directional delivery condition of the invention and would furthermore speed the delivery of the probes to their targets in the static sample over that possible by diffusion alone. Additionally, if the probes are either charged or magnetic, they could be transported through the aqueous medium to their targets in the static sample in the respective presence of an electric or magnetic field. Anadditional mechanism for probe delivery to the sample could be sedimentation onto the sample driven by gravity.

[0036] Static settings could be used (e.g. for the transport speed and illumination intensity) in an optimal way to allow for dense yet sufficiently sparse localizations over long timescales.

[0037] For stable binders, with residence times longer than the entire illumination and imaging process, and static transport settings, the density of localizations will be highest at the beginning of the acquisition and will gradually decrease due to the approach to binding saturation of the target epitope.

[0038] For semi-stable binders, with residence times shorter than the entire illumination and imaging process, the density of localizations will depend additionally on the efficiency of replacement of photobleached bound probes with new fluorescent probes binding to the same target and will therefore depend on the exact binding half-life.

[0039] Figure 1 shows observational setups for uPAINT based on either TIRF or widefield microscopy, and for cfCALM using wi defield microscopy. (A) Initial condition for the fraction of free fluorescent probe using uPAINT and TIRF microscopy. (B) Later timepoint for the setup shown in panel A. (C) Initial condition for the fraction of free fluorescent probe using uPAINT and widefield microscopy. (D) Later timepoint for the setup shown in panel C. (E) Initial condition for the fraction of free fluorescent probe using cfCALM based on continuous transport of the fluorescent probe over the sample mounted in a microfluidic sample chamber and widefield microscopy. (F) Later timepoint for the setup shown in panel E.

[0040] In Figure 1, the fraction of freely diffusing, fluorescent probe is depicted for uPAINT applied to a fixed sample based on TIRF at the start of the illumination (panel A) and at a later timepoint (panel B). Specifically, in a sample well (001) defined by a peripheral wall (002), a static sample (003) is mounted. A solution (004) containing the probe is then brought into contact with the static sample (003). The static sample (003) is mounted on a coverslip glass (005) and observed through a microscope objective lens (006), with 2D coordinate system (007) shown. TIRF microscopy is used to illuminate the sample in the form of a thin (<100 nm) optical section consisting of an illuminated volume (008) within the sample. Initially, upon turning on the illumination, the fraction of fluorescent probe is instantaneously uniform throughout the solution (004) (panel A). Immediately thereafter, however, a photobleached gradient (009) of probe will start to form both within and away from (due to diffusion of photobleached probe) the illuminatedvolume (panel B). The contrast bar for the fraction of fluorescent probe is shown (010), with the darkest gray regions having a fluorescent probe fraction of f = 1 and lighter regions having a fluorescent probe fraction of f < 1. In panels C and D, the suboptimality of widefield microscopy for uPAINT is illustrated. Here, widefield illumination of the sample takes the form of a conical hourglass shape, with the upstream (Oi l) and downstream (012) boundaries of the cone of illumination indicated by the dashed lines. The area of illumination of the sample at the focal plane (013) is also shown, with its upstream (014) and downstream (015) boundaries along the lateral axis coordinate, x (007), indicated. At initiation of the illumination, the fraction of fluorescent probe will again be instantaneously uniform throughout the solution (004) contained within the sample well. Immediately thereafter, however, a photobleached gradient (016) of probe will start to form both within and away from (due to diffusion of photobleached probe) the illuminated volume, with a specific late time for this process depicted in panel D. The depth and extent of the photobleaching gradient for widefield microscopy (panel D) will in general be higher than for TIRF microscopy (panel B), assuming a similar laser power is delivered to the back plane of the objective in both cases. By contrast, the preferred flow- based delivery of the probes over the sample used for cfCALM would be compatible with widefield microscopy, as illustrated in panels E and F. Here, the sample is mounted within a sample chamber (017) — with flow chamber ceiling (018) indicated — and the solution containing the probe is flowed into the inlet (019) and out from the outlet (020) in the direction indicated by the arrows (021). Again, upon turning on the illumination, the fraction of fluorescent probe will be instantaneously uniform through the flow chamber (panel E). At later times, for a steady transport of the probe at fixed concentration and fixed velocity, a stable gradient for the fraction of fluorescent probe will quickly form and remain constant throughout the entire acquisition. The free probe immediately upstream (022) of the illuminated area at the focal plane (013), will have a fluorescent fraction of f = 1, as this probe has yet to be illuminated (here, flow is implicitly assumed to be much more faster than diffusion, which is typically the case). However, the free probe immediately downstream (023) of the illuminated area at the focal plane (013) will have a fluorescent fraction f < 1, with the exact value determined by the photobleaching rate of the fluorophore, the 3D geometry of the chamber and widefield illumination cone, and the flow velocity. For a relatively homogeneous detection of the probes, f should not deviate too strongly from unity over the ROI contained within the illuminated area at the focal plane (013), implying a proper combination of low illumination and sufficiently fast flow(see detailed discussion below). A representative plot of the fraction of fluorescent probe, / observed over the illuminated area at the focal plane (013) along coordinate x (024) is shown, with the value of f = 1 displayed as a dotted line (025) and / as a solid line (026).

[0041] Figure 2: Schematic of a preferred experimental setup for cfCALM. (A) Flow of free fluorescent antibody probe through a microfluidic chamber containing a static sample that is observed with an objective in an epifluorescence setting (cone of illumination emanating from the objective is indicated). (B) Photobleaching of a free fluorescent antibody probe within the cone of illumination. (C) Binding of a fluorescent antibody probe and a photobleached probe to the sample. (D) Acquisition of emission light from the bound fluorescent probe leads to its eventual bleaching. Photobleaching of bound fluorescent probes is balanced by the flow-based replenishment and binding of new fluorescent probes to the static sample.

[0042] In Figure 2, a schematic of the preferred flow-based implementation of cfCALM is given. In panel A, a microfluidic sample chamber (017), specified by its length (027), height (028), and width (029), is used to set up a directed and continuous flow of the solution containing the probe into the inlet (019) and out of the outlet (020). A static sample (003) is mounted within the sample chamber (017). The flowing medium contains free probes (030), comprised of an antibody (or antibody-like) binder (031) and a fluorescent substance (032). The static sample (003) is illuminated with and observed through a microscope objective lens (006) using widefield microscopy. The position of the sample chamber relative to the objective can be moved along the 3D axes (033). A specific ROI (034), contained within the illuminated area at the focal plane (013) and acquired by a camera, is also indicated, along with its upstream boundary (035) and downstream boundary (036). Preferably, the upstream boundary of the ROI (035) should coincide with the upstream boundary of the illumination cone (011). The axial axis of the microscope objective is shown (037) along with the opening angle of the illumination (038) determined by the objective numerical aperture. The upstream boundary (011) and downstream boundary (012) of the illumination cone are indicated. Photobleaching of a free fluorescent probe labeled in panel A as (030) is shown in panel B as (039), with the photobleached fluorescent substance depicted as a filled star (040). The binding of a free fluorescent probe, indicated as (030) in panel B, generates a bound fluorescent probe (041) in panel C. Similarly, the binding of a free photobleached probe, indicated as (039) in panel B, generates a bound photobleached probe (042) in panel C. In panel D, illumination and detection of the bound fluorescent probe, indicated as (041) in panel C, leads togeneration of a photobleached bound probe (043). Detection of many such bound fluorescent probes at low illumination intensity until they are photobleached allows for super-resolved localization down to Angstrom resolution, as shown below in Figure 4.

[0043] In Figure 3, panel A, a zoomed-in view of the illuminated area at the focal plane (013) and an ROI (034) contained within it. The transport direction (021) is also indicated along with a lateral 2D coordinate system for axes x and j’ (044). The length, L (045), determined from the upstream boundary of the illuminated area at the focal plane (014) at position xi to the downstream boundary of the ROI (036) at position X2 is indicated. As both the illuminated area at the focal plane and the ROI are rectangular, the calculation of L is independent of position y. In panel B, the determination of the longest L for a more general geometry of illumination and detection is illustrated. The transport direction (021) and lateral 2D coordinate system for axes x and j’ (044) are again specified, as are a circular illuminated area at the focal plane (013) and a polygonal ROI (034). In this more general setting, the longest length L should be defined as the maximum value, Zz, obtained along the x axis at each yt intersecting the ROI. As an example, three different values for Li are shown, specifically, Zi (046), Li (047), and / .3 (048). These lengths are redisplayed below and compared by aligning their left edges (049), with Li clearly the longest, as determined by the displayed equation (050). This is despite the fact that Zi (046) extends out to the most downstream position in the ROI and / .3 (048) starts at the most upstream position of the illuminated area at the focal plane.

[0044] Figure 4: Conventional IF vs. cfCALM super-resolution of tubulin in fixed HeLa cells. (A) Conventional IF of tubulin forming microtubules in a HeLa cell fixed with paraformaldehyde (PF A). (B) cfCALM super-resolution image down to few Angstrom resolution of a different PFA-fixed HeLa cell in a separate sample chamber. (C) Zoomedin view of the boxed region shown in B with individual tubulin targets. (D), (E), and (F) Raw images from the cfCALM movie of the individual tubulin targets shown in panel C.

[0045] In Figure 4, a comparison of images obtained with conventional IF and cfCALM super-resolution of tubulin in PFA-fixed HeLa cells is made (live cells were fixed with 4% PFA in PBS for 10 min at room temperature). In panel A, a conventional IF image (100 nm pixels) of microtubules stained with an anti-alpha-Tubulin antibody following the manufacturer’s protocol (REA1136, REAfinity™, Miltenyi Biotec). An isolated microtubule is indicated (051). In panel B, a cfCALM super-resolution image of a similarly prepared HeLa cell in a different sample chamber is shown (10 nm pixels), with a similar microtubule as in panel A also indicated (052). The sample chamber was a p-Slide VI 0.5 Glass Bottom (Ibidi), which has a height of 0.54 mm, a length of 17 mm, and a width of 3.8 mm. The antibody was used at a 1 : 100,000 dilution from the stock concentration and flowed through the sample chamber at 50 pL / min for 40 min using a microfluidic pump (Aria, Fluigent). The velocity of the fluid was therefore, V = 406 pm / s. The length, L, of the ROI was 211.3 pm. The transit time of the flow over the entireROI was then T = = 0.52 s, which was significantly shorter than the photobleachinghalf-life of APC in the focal plane of roughly 15 seconds (for the illumination intensity of 5.33 W / cm2at the focal plane using 100% power for the 630 nm LED). A boxed region (053) containing a particular microtubule filament is shown. In panel C, a zoomed-in view corresponding to the boxed region (053) of panel B is shown. Individual tubulin epitope targets are displayed as single 10-nm pixels, with three indicated, namely, tubulin epitope 1 (054) within the filament, tubulin epitope 2 (055) adjacent to the filament, and tubulin epitope 3 (056) within the filament. Raw image frames from the cfCALM movie for the zoomed-in image corresponding to panel C are displayed in panels D - F. In panel D, a representative 5 -second frame showing the pattern of fluorescence from tubulin epitope 1 (054) is shown. In total, 173217 photons were collected from tubulin epitope 1 over 45 seconds, implying a potential lateral resolution down to <J^at= 2.4 A. In panel E, a representative 5 -second frame showing the pattern of fluorescence from tubulin epitope 2 (055) is shown. In total, 98,979 photons were collected from tubulin epitope 2 over 20 seconds, implying a potential lateral resolution down to <J^at= 3.2 A. Finally, in panel F, a representative 5-second frame showing the pattern of fluorescence from tubulin epitope 3 (056) is shown. In total, 269,331 photons were collected over 50 seconds, implying a potential lateral resolution down to <J^at= 1.9 A. cfCALM images were obtained on a Zeiss Axio Observer 7 widefield microscope using a 63x (N.A. = 1.46) alpha Plan- Apochromat oil objective (Zeiss) and additionally equipped with an autofocus module (Definite Focus.2, Zeiss). cfCALM measurements were made of the displayed HeLa cell by recording 1-min fluorescence movies (5 seconds / frame, 12 frames) alternated with a single transmission image (50 ms exposure) under continuous flow of the antibody using a pressure-driven pump (Aria, Fluigent). The cfCALM super-resolution image (shown in panels B and C) was obtained by analysis of the movies using the Tanitracer code (Kiuchi et al., Nature Methods 12, 743, 2015), which registers the image frames from the movies laterally (based on interpolated shifts calculated from the intermittent transmission images), identifies isolated particles, tracks the particles over multiple frames, and finally creates an estimate of each tracked particle’s center of mass.

[0046] Figure 5 shows the simultaneous detection of both in-focus and out-of- focus probes within a fixed tissue slice. (A) Displayed is frame 545 (5-second exposure) from a 3.15 hour movie of an anti-CD81-APC antibody (REA513 REAfinity™, h, APC, #130-119-787, Miltenyi Biotec) delivered at a low concentration (1 : 10000 dilution) by continuous flow to a few pm thick tonsil slice fixed with PF A. Boxed region includes a single highly out-of-focus probe, with the next panels displaying sequential frames from the movie, specifically, frames (B) 545, (C) 546, (D) 547, (E) 548, (F) 549, and (G) 550. (H) Further zoom-in of the boxed region from panel A. Several simultaneously detected in-focus probes are indicated. (I) Gibson-Lanni model for the PSF of a particle that is 4.6 pm out of focus.

[0047] In Figure 5, the simultaneous detection of both in-focus as well as highly out-of-focus probes is demonstrated for an anti-CD81-APC antibody (DOL = 1, REA513 REAfinity™, h, APC, #130-119-787, Miltenyi Biotec) applied at a 1 : 10,000 dilution to a PFA-fixed, few micron thick tissue slice. In panel A, the full ROI for a single 5-second frame of the acquisition — specifically, the 545thframe from a 3.15 hour movie — is displayed. Visible are both in-focus probes (057) and out-of-focus probes (058) with extended PSFs. The boxed region (059) includes a single out-of-focus probe. Frames 545, 546, 547, 548, 549, and 550 of the movie are displayed for the boxed region (059) in panel A in panels B, C, D, E, F, and G, respectively. In panel H, the boxed region (059) from panel A is zoomed in further. Interspersed in this image are several in-focus probes (057) that can be clearly separated from the out-of-focus probe due to their smaller size as well as their differing temporal profiles (appearance / disappearance and fluctuating brightness) over the movie sequence shown in panels B - G. Such simultaneous detection of both infocus and out-of-focus probes is not possible with TIRF illumination (e.g. required for uPAINT on fixed cells, Gunasekara et al., Bioconjug Chem 34, 825, 2023), which illuminates the sample only over axial displacements of at most <100 nm from the glass coverslip. For the out-of-focus probe, roughly 5.4 million photons were collected over its 30 sec acquisition. In panel I, a Gibson-Lanni PSF model (Gibson & Lanni, J. Opt. Soc. Am. A 9, 154, 1992) is shown for a particle at an assumed 1 pm axial displacement from the coverslip glass and with the focal plane positioned at 5.6 pm from the coverslip glass. The Gibson-Lanni model was calculated using the PSFGenerator plugin (Kirshner et al., J. Microscopy 249, 13, 2013) for Imaged (Schneider et al., Nat. Meth. 9, 671, 2012). Other parameters used for the model were: refractive index of the objective immersion oil (1.53), refractive index of the sample immersed in water (1.333), working distance (100 pm),particle position (1 pm), wavelength (670 nm), NA (1.46), and pixelsize (103.14 nm). The PSF (Gibson-Lanni model) is highly sensitive to changes in focus of the particle, allowing for precise localization in z down to the statistical limit. Detection at two different focal planes displaced from one another at the micron scale — so-called biplane acquisition — would furthermore remove any ambiguity regarding the positive vs. negative displacement of the particle with respect to the focal plane. Taking into account the 5.4 million collected photons and assuming conventional lateral and axial resolutions of axy= 100 nm and axy= 255 nm (see above), implies a potentially achievable localization precisions down 1.1 A for the probe.

[0048] Conventional approaches to SMLM illuminate the sample with a laser delivering anywhere from 100 to 10,000 W / cm2at the focal plane within the static sample. These high intensities are, in part, necessitated by the specific SMLM technique, e.g. to capture enough photons for each transient binding event for PAINT or each blinking event for STORM. However, CALM and cfCALM do not require such fast acquisition of photons, allowing for the use of much lower intensities. As shown below, at much lower intensities, the fluorophores have significantly longer photobleaching half-lives permitting much higher photon counts per probe in the range of tens of thousands to millions (see the next paragraph as well as the single probes highlighted in Figures 4 and 5).

[0049] To demonstrate the relationship between illumination intensity and collected photons per probe before photobleaching, two different illumination intensities were used to photobleach sparsely distributed probes using a similar experimental setup to that shown for cfCALM in Figure 4 (panels B - F), but with washing of the free fluorescent probe before acquisition. Specifically, the same 100% LED power (R / 631 / 33 nm LED, Colibri 7, Zeiss) as used for the cfCALM data in Figure 4 (panels B - F) was compared with 10% power of a laser diode (400 mW 640 nm laser diode, LDI-4-488 laser unit, 89 North). The 100% LED delivered 5.33 W / cm2at the focal plane in the sample. The 10% laser diode delivered 44.4 W / cm2at the focal plane in the sample. For the analysis of the photobleaching image series, multiple probes were assessed for their total integrated intensities by drawing ROIs around clusters of from roughly 10 to 20 probes followed by subtracting off the local background contribution (assessed by a background ROI) and dividing by the camera gain of 1.67 (0.6 e- / ADU, 16-bit HDR mode, Prime BSI sCMOS camera, Photometries). For the 100% LED, 75 probes were assessed in total with an average of 444,939 photons / probe collected 1-5 A), consistent with — albeit slightly higher than — the photons / probe reported for the three selected probelocalizations shown in panels D - F of Figure 4. For the 10% laser, 58 probes were measured with an average of “only” 38,744 photons / probe collected (cr^ = 5.1 A), which is more than ten times smaller than that reported above for the 100% LED. Note that the 10% laser intensity utilized used here, and corresponding to 44.4 W / cm2, was already significantly lower than the more typical range of 100 - 10,000 W / cm2employed for standard SMLM approaches. Here, as well, all photons were collected before photobleaching, which is not the case for the most commonly used SMLM approaches based on STORM or PAINT, for which photons are only collected for the duration of a transient blinking or binding event. Low illumination intensity and collection of all photons from the probe before photobleaching of cfCALM, therefore account for its much higher photon collection efficiencies per localization event (104- 107photons) compared with standard SMLM approaches like DNA-PAINT and STORM (102- 104photons).

[0050] The use of lower intensity illumination for cfCALM therefore plays two important roles: (1) Avoidance of significant photobleaching of the free fluorescent probes before they can bind, (2) Collection of many more photons per probe (roughly 104- 107photons) to achieve statistical localization precision down to the Angstrom level.

[0051] In another embodiment, the method could be further characterized by transporting the probe such that a maximum transit time of the probe is less than 20 times a photobleaching half-life of the fluorescent substance.

[0052] As stated above for the originally proposed CALM technique (WO 2025 / 125533 Al), the acquisition of images during the staining process should generally be avoided, as the free fluorescent probe may be bleached before it binds and thereby blocks a targeted epitope, with the blocking either permanent (stable CALM, sCALM) or temporary (semi-stable CALM, ssCALM). For sCALM, this would lead to a non-detection of the particular epitope, reducing overall labeling efficiency. For ssCALM, the exact severity of the temporary blocking of an epitope on the labeling efficiency would depend on the residence time, but would nevertheless generally reduce the overall efficiency of detection of probe binding events on the sample.

[0053] However, for the simultaneous staining and imaging employed by cfCALM, the risk of blocking epitopes with prebleached probes can be largely avoided if prebleaching of the free probes en route to epitopes in the sample can be reduced to an insignificant level. The percentage of probe that is prebleached before reaching a particular epitope in the sample depends on the amount of photobleaching that occurs over the timescale beginning with the entry of the probe into the illuminated region and itsbinding to an epitope. A useful quantity to consider is the maximum transit time, Tmax, defined as the longest timescale (over the entire illumination and imaging process) for the transport of the probe over the maximum distance, L, from the upstream boundary of the illuminated area at the focal plane (014) to the downstream boundary of the ROI (036) at the focal plane (see Figure 3). For a fixed transport speed, V the maximum transit time would be Tmax= L / V. The photobleaching half-life of the fluorescent substance, 7half-life, should be determined at the center of the focal plane, i.e. at or near the imaging axial axis (037). The condition for this embodiment is then Tmax= ^ < 20 Thalf-life, implying a minimum transport speed of V > L / (20 Thalf-life) and an unbleached fraction of the probe satisfying roughly f > 2~20« 10-6everywhere within the ROI.

[0054] In the preferred application of the invention, the upstream boundary of the ROI (035) coincides with the upstream boundary of the illuminated area at the focal plane (013), with free probes entering the ROI here unbleached; the fraction of fluorescent free probe would therefore always start from unity at the upstream boundary of the ROI (035) and decrease monotonically along the flow direction, x (007). Over the entire extent of the ROI (034), the unbleached fraction, , will exceed 10-6everywhere, ensuring a sufficient detectability of the probe throughout the ROI at least for highly expressed markers in cells, which can number in the millions. For these markers, detection of only one in a million probes at the downstream boundary of the ROI (036) would still be marginally possible.

[0055] The preferred application of the invention, however, would be for more homogeneous detection, with e.g. f > 0.95 holding over the entire ROI, yielding the following tighter criterion of Tmax= ^ < 0.074 Thalf-life.

[0056] If the transport speed varies with time as 7(t), then the maximum transit time should be determined implicitly for all t0over the entire illumination and imaging process from L(t0) = 7(t)dt, with the maximum transit time corresponding tothe largest value

[0057] In the case of a variable illumination intensity, / (t), then the twenty -fold half-life, 20 Thalf-life, corresponding to (1 / 2)20« 10-6, or the “millionth” of the fluorophore (only 1 in a million probes are still fluorescent), would need to be determined for each t0as Tmillionth(t0). The condition for this embodiment can therefore be written more generally as T(t0) < Tmillionth(t0) for all t0over the entire illumination and imaging process. This criterion could be numerically checked to hold as well for thecombination of variable illumination and variable velocity, if the exact temporal variability of each is known.

[0058] In the case of spatially inhomogeneous illumination over the illuminated area at the focal plane, this would need to be additionally accounted for in determining the expected photobleaching for every possible trajectory (see Figure 3B) over the sample starting at the upstream boundary of the illuminated area at the focal plane and terminating at the downstream boundary of the ROI.

[0059] For stable binder probes, a maximum transit time substantially smaller than the photobleaching half-life would be preferred to limit photobleaching of the free fluorescent probes (see Figure IF and Figure 2, panels A - C) to less than a few percent before binding the sample. Note that for the above-described experimental example (Figure 4, panels B - E), the maximum transit time was indeed much less than the photobleaching half-life, with Tmax= 0.52 s = 0.035 Thalf-life« Thalf-life, implying furthermore that the fluorescent fraction satisfied f >= 0.98 over the ROI over the duration of the experiment.

[0060] For semi-stable binder probes, a maximum transit time up to 20 times the photobleaching half-life could be tolerated for detection over the entire ROI. However, a maximum transit time greater than 20 times the photobleaching half-life would imply that less than one in a million probes would be fluorescent when binding at the right side of the ROI, leading to an extremely inefficient acquisition of localizations in this region of the sample, even in the case of high binding saturation of the target by mostly dark semistable binder probes.

[0061] In another embodiment, the method could be further characterized by using a probe with a second component capable of binding to the first component with a second binding half-life greater than 15 seconds.

[0062] For example, a secondary fluorescent antibody binding stably to an unlabeled primary antibody, as used for conventional IF, could be used for the approach. As well, cfCALM could be used as a “slow” version of DNA-PAINT by conjugating a significantly longer oligo than used for DNA-PAINT to the antibody binder and an equally long complementing oligo for the imager strand to achieve a binding half-life significantly greater than the roughly 10 second upper limit for DNA-PAINT (Jungmann et al., US 11,536,715 B2). Here, a disadvantage of cfCALM would be fewer unique binding events (slower acquisition), but a significant advantage would be the much higher photon counts / probe for cfCALM due to the lower illumination intensity (seephotobleaching comparison of laser vs. LED illumination above) that allow for Angstrom resolution.

[0063] The oligo lengths could be comparable to those used for cyclic high-content spatial biology approaches based on similar oligo-labeled antibodies and imager strands (see e.g. Black et al., Nature Protocols 16, 3802, 2021). For these cyclic spatial biology approaches, the static sample is pre-labeled with different antibody-oligo conjugates. For each cycle, the static sample is then stained with a particular complementary fluorescent imager strand, washed, and then imaged for conventional IF. For cfCALM, each complementary fluorescent imager strand would successively be transported across the pre-labeled static sample in a continuous fashion in conjunction with simultaneous free- running image acquisition.

[0064] In another embodiment, the method could be further characterized by applying a feedback loop based on the one or more images to adjust the amount of probe in the medium, the illumination intensity of the light source, the flow velocity, the electric field strength, or the magnetic field strength.

[0065] For faster acquisition with cfCALM, a constant density of localizations could be maintained by such active feedback control based on live monitoring of the total number of localizations across the sample and immediate on-the-fly modification of the relevant above-listed experimental parameters. Such feedback control would be especially relevant for stable binders to maintain a high instantaneous density of bound fluorescent probes even in the vicinity of high binding saturation of the target.

[0066] In another embodiment, the method could be further characterized by using a static sample containing at least two targets and repeating steps a to d sequentially for different probes specific to each of the targets.

[0067] In another embodiment, the method could be further characterized by in step c distorting the emission radiation by a refractive or diffractive optical element before acquiring the one or more images.

[0068] In another embodiment, the method could be further characterized by in step c splitting the emission radiation using one or more optical elements and imaging at different focal planes.

[0069] Either of the preceding two embodiments could be used to obtain Angstrom-level localization accuracy along the axial direction (z-axis in Figures 1 and 2).

[0070] In another embodiment, the method could be further characterized by imaging the emission light using a widefield microscope.

[0071] A benefit of cfCALM over PAINT -based techniques is the significantly lower applied concentration of free fluorescent probe, generating a lower background of free fluorescent probe and permitting the use of simple widefield microscopy. There is therefore less need for removal of this background for cfCALM using the more complicated optical sectioning setups required for other PAINT -based techniques. The background of free fluorescent probe is sufficiently low and photon detection from the bound fluorescent probes sufficiently high that simple widefield microscopic detection could be employed even for sample chambers with relatively high profiles, as for example for the data shown in Figure 4 (panels B - F) for which a 540 pm tall chamber was used. Here, reduction of the chamber height would generically lead to a proportional reduction in the background intensity under widefield detection. Instead of the 540 pm chamber employed in Figure 4 (panels B - F), a chamber height as low as tens of micrometers could be employed, reducing the background by an additional order of magnitude. Even further repression of remaining background using optical sectioning approaches could nevertheless be utilized as the next embodiment describes.

[0072] In another embodiment, the method could be further characterized by imaging the emission light on a point-scanning confocal microscope, a spinning disc microscope, a programmable array microscope, a structured illumination microscope, a light-sheet microscope, a TIRF microscope, a HILO microscope, or a 4Pi microscope.

[0073] For all SMLM approaches, as well as for the originally proposed CALM, bound fluorescent probes located significantly above or below the focal plane can create a background emission light that can bias the localization of the in-focus probes. As discussed immediately above, for cfCALM, as for PAINT -based approaches, an additional background emission light is generated by the free fluorescent probes. Optical sectioning could be used to significantly reduce both sources of background emission light.

[0074] In another embodiment, the method could be further characterized by illuminating with coherent or incoherent light from the light source.

[0075] In another embodiment, the method could be further characterized by illuminating with no greater than 50 W / cm2at the focal plane.

[0076] The light intensities required for standard SMLM approaches to obtain sufficient localization accuracy require powerful coherent light generated by laser illumination. As the illumination light should be much lower for cfCALM to avoid excessive photobleaching of the free fluorescent probe before it can bind to the static sample, even LED illumination can suffice. For example, the cfCALM datasets displayedin Figure 4 (panels B - F) and Figure 5 were obtained using an LED at 100% power (5.3 W / cm2).

[0077] In another embodiment, the method could be further characterized by using a static sample containing two or more targets, using a medium containing two or more probes specific for each target, in step c imaging the emission radiation in two or more detection channels, and in step d determining the positions of the two or more probes to estimate the positions of the targets.

[0078] In another embodiment, the preceding embodiment could be further characterized by using in step a two or more light sources.

[0079] For example, two or more light sources could provide illumination specific for the two or more detection channels of the immediately preceding embodiment.

[0080] In another embodiment, while performing step c, stabilizing the lateral and / or axial positions of the static sample using feedback control based on tracking the lateral and / or axial positions of a fiducial marker could be performed. The feedback control could either be applied in a stepwise (intermittent) or continuous manner. Examples of fiducial markers would be a gold particle, the image of the static sample in a different channel (e.g. transmission image), or the average position of the ensemble of probes that are tracked across two or more frames. These different types of fiducial markers can be used alone or in combination for even greater accuracy.

[0081] In another embodiment, before step d, relative shifts of the images could additionally be determined by tracking one of the above listed fiducial markers, with these relative shifts used to improve the determination of the position of the probe in step e.

[0082] The preferred implementation of cfCALM based on free-running microfluidic pumping and continuous image acquisition together allow for a simpler and more efficient means for collecting sufficient photons (roughly 104- 107photons per localization) from single bound fluorescent probes to localize them to Angstrom precision both laterally and axially (with axial resolution achieved preferably using biplane acquisition). The requirement for lower illumination power of cfCALM even permits the use of LED illumination instead of more expensive laser illumination, further simplifying the microscope setup compared to standard SMLM approaches. As an additional benefit for cfCALM, we have observed that lower illumination power is actually beneficial, leading to a dramatically higher number of photons collected from a fluorophore before its photobleaching, implying that lower illumination powers automatically allow for higher localization precision down to the Angstrom level.

Claims

Claims1. Method for determining the position of a target on a static sample by providing a medium containing a probe comprising a first component capable of binding directly and specifically to the target bound to a second component comprising a fluorescent substance by the steps: a. providing the medium to the static sample such that the probe binds at least transiently to the target, b. illuminating the static sample with a light source, thereby generating emission light from the fluorescent substance of the probe, c. acquiring one or more images of a region of interest (ROI) contained within an illuminated area at a focal plane of the static sample comprising emission light, d. determining the position of the probe observed over the one or more images to estimate the position of the target on the static sample, characterized in that the binding of the probe to the target has a first binding half-life greater than 15 seconds and that the probe is actively transported over the sample such that the maximum transit time of the probe from the upstream boundary of the illuminated area at the focal plane to the downstream boundary of the ROI is less than 20 times of a photobleaching half-life of the fluorescent substance of the probe.

2. Method according to claim 1 characterized in that the probe is actively transported to the static sample by flow of the medium, an electric field, a magnetic field, or gravity.

3. Method according to any of claims 1 to 2 characterized in that the second component of the probe is bound to the first component with a second binding half-life greater than 15 seconds.

4. Method according to any of claims 1 to 3 characterized in providing a feedback loop based on the one or more images to adjust the amount of probe in the medium, the illumination intensity of the light source, the flow velocity, the electric field strength, or the magnetic field strength.

5. Method according to any of claims 1 to 4 characterized in providing a static sample containing at least two different targets and repeating steps a to d sequentially for different probes specific to each of the targets.

6. Method according to any of claims 1 to 5 characterized in that the emission radiation is distorted by a refractive or diffractive optical element before step c).

7. Method according to any of claims 1 to 6 characterized in that the emission radiation is split by using one or more optical elements and imaging at different focal planes.

8. Method according to any of claims 1 to 7 characterized by imaging the emission light on a widefield microscope.

9. Method according to any of claims 1 to 8 characterizedin that the static sample is illuminated with coherent light or incoherent light from the light source.

10. Method according to any of claims 1 to 9 characterized in that the static sample is illuminated with no greater than 50 W / cm2at the focal plane.

11. Method according to any of claims 1 to 10 characterized by imaging the emission light with optical sectioning using a pinhole, a spinning disc, a programmable array microscope (PAM), a structured illumination microscope, a light-sheet microscope, a TIRF microscope, or a HILO microscope.

12. Method according to any of claims 1 to 11 characterized in that the static sample contains two or more targets and using a medium containing two or more probes specific for each target, in step c and imaging the emission radiation in two or more detection channels, and determining the positions of the two or more probes to estimate the positions of the targets in step d.

13. Method according to claim 12 characterized by using in step a two or more light sources.

14. Method according to any of claims 1 to 13 characterized in that the static sample is a tissue sample immobilized on a substrate.

15. Method according to any of claims 1 to 14 characterized in that in step c the position of the static sample is stabilized by using a feedback control based on tracking the lateral position of a fiducial marker while acquiring the images.

16. Method according to any of claims 1 to 15 characterized in that in step c, the position of the static sample is stabilized using a feedback control based on tracking the axial position of the fiducial marker while acquiring the images.

17. Method according to any of claims 1 to 16 characterized in determining the relative lateral shifts of the acquired images before step d by tracking the lateral position of the fiducial marker.

18. Method according to any of claims 1 to 17 characterized in determining the relative axial shifts of the acquired images before step d by tracking the axial position of the fiducial marker.

Images