Methods for building artificial RNA organelles in living cells

Abstract

Protein and RNA-based condensates are emerging as an alternative to classical membrane-bound organelles for the task of compartmentalizing molecules and biochemical reactions. We describe methods for making RNA condensates in mammalian cells. We further show that aptamers make it possible to recruit peptides and proteins to the condensates with high specificity. Such RNA condensates can be modularly customized and offer a route toward creating systems of functional artificial organelles.

Description

[0001]METHODS FOR BUILDING ARTIFICIAL RNA ORGANELLES IN LIVING CELLS CROSS REFERENCE TO RELATED APPLICATIONS This application claims the benefit under 35 U.S.C. Section 119(e) of co- pending and commonly-assigned U.S. Provisional Patent Application No.63 / 679,560, filed August 5, 2024, entitled “METHODS FOR BUILDING ARTIFICIAL RNA ORGANELLES IN LIVING CELLS”, the contents of which is incorporated by reference herein. STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH AND DEVELOPMENT This invention was made with government support under 1938194, and 2134772 awarded by the National Science Foundation. The government has certain rights in the invention. TECHNICAL FIELD Embodiments of the disclosure concern at least the fields of cell biology and biochemistry. BACKGROUND OF THE INVENTION Membraneless compartmentalization emerges from the condensation of proteins and RNA, and is recognized as a primary mechanism through which cells dynamically control biochemical processes. By co-localizing nucleic acids, enzymes and metabolites, membraneless organelles (MLOs) such as nucleoli, Cajal bodies and stress granules are believed to regulate biogenesis, transcription, post-transcriptional modification and degradation, and generally improve cellular fitness. The emergence of biomolecular condensates has also been linked to disease, particularly to neurodegeneration. The ability to express “designer condensates” with prescribed properties would be a valuable tool to program cellular behavior and engineer synthetic cells. Remarkable examples based on engineered peptides or natural condensate forming RNA sequences, have highlighted the feasibility of this concept. The generality of these strategies, however, is hampered by challenges linked to protein engineering and the limited programmability of natural RNA repeats. There is a need in the art for methods and materials useful to form synthetic RNA condensates in living cells. SUMMARY OF THE INVENTION Artificial biomolecular condensates are emerging powerful tools to control cellular behaviors. As disclosed herein, we have developed a method to build artificial condensates within living mammalian cells through the design of modular RNA motifs formed by a single, short strand of RNA. These condensates emerge spontaneously, creating RNA-rich compartments that remain segregated from the surrounding environment. The RNA sequences include stem-loop domains that fold as the RNA is transcribed, and then condense in the nucleus and / or cytoplasm through loop-loop interactions. These sequences can be optimized and diversified, making it possible to generate distinct populations of condensates that do not mix, and can be localized to different subcellular locations such as the cytoplasm or nucleus. The RNA motifs can further be selected to recruit small molecules and proteins to the RNA-rich phase. By introducing additional RNAs that link two distinct types of condensates, embodiments of the invention can create droplets with multiple subcompartments, whose organization can be controlled by tuning the stoichiometry of different RNA sequences. These artificial condensates have many applications in the study and manipulation of molecular functions inside living mammalian cells. The invention disclosed herein has a number of embodiments. Embodiments of the invention include, for example, methods of making a synthetic RNA condensate molecule within a living cell, typically a mammalian cell. In certain embodiments of the invention, such methods comprise transducing the cell with a vector comprising a DNA sequence selected to encode a RNA condensate polynucleotide sequence; and then allowing the cell to transcribe the DNA sequence so that a synthetic RNA condensate molecule is formed within the cell. In such methods, the vector and the DNA sequence are selected to form a synthetic RNA condensate molecule that comprises 100-300 nucleotides, and is circular. In these methods, the synthetic RNA condensate molecule is designed to comprise complementary sequences that allow the RNA polynucleotide to self-assemble, typically so as to form at least 3 arm structures comprising a stem and a loop, wherein: the stem in said arm structures comprise at least 10 paired nucleotides; the loops in said arm structures comprise at least 6 nucleotides; and a first loop in the synthetic RNA condensate comprises at least 6 nucleotides selected to be complementary to at least 6 nucleotides in a second loop in a synthetic RNA condensate such that the at least 6 nucleotides in the first loop of a first synthetic RNA condensate molecule and the at least 6 molecules in a second loop of a second synthetic RNA condensate molecule couple together so that the synthetic RNA condensate molecules aggregate to form a RNA condensate in the cell. In typical embodiments of the invention the cell is a mammalian (e.g. human) cell. In certain embodiments of the invention, the synthetic RNA condensate molecules are designed to influence where in the cell the molecules form condensates. For example, in some embodiments of the invention the RNA condensate molecule is designed to form not more than 3 arm structures; to have kissing loops that form not more than 2 or 3 G:C bonds, and to form arms that comprise not more than 10, 1112 or 13 paired nucleotides such that condensates preferentially form in the cytoplasm of the cell. In other embodiments of the invention, the RNA condensate molecule is designed to form at least 3 or 4 arm structures; to have kissing loops that form more than 2-3 G:C bonds, and have arms that comprise more than 11, 12 or 13 paired nucleotides such that condensates preferentially form in the nucleus of the cell. In certain embodiments of the invention, the synthetic RNA condensate molecule comprises a RNA aptamer motif sequence; and / or the synthetic RNA condensate molecule comprises a polynucleotide sequence to which a RNA aptamer couples via base pairing. In some embodiments of the invention, the RNA aptamer motif sequence is disposed in a loop structure of the synthetic RNA condensate molecule. In certain embodiments of the invention, the synthetic RNA condensate molecules are designed so that the first loop and the second loop are disposed in a single synthetic RNA condensate; and the first loop and the second loop in the single synthetic RNA condensate cannot couple together. In some embodiments of the invention, the method forms a plurality of synthetic RNA condensates by transfecting the cell with a plurality of vectors selected to encode a plurality of RNA condensation sequences so that a plurality of synthetic RNA condensates are formed. Optionally in these methods at least one synthetic RNA condensate molecule is designed to include at least 6 nucleotides in one or more loops that are selected to be orthogonal. In some embodiments of the invention, the RNA condensate is coupled to a protein and / or a nucleic acid. Embodiments of the invention include compositions of matter comprising a mammalian cell having a synthetic RNA condensate molecule disclosed herein. One such composition of the invention comprises an RNA condensate formed in a living cell by a plurality of aggregated synthetic RNA condensate molecules made by the methods disclosed herein. Additional embodiments of the invention include methods of making an RNA condensate within a mammalian cell using the synthetic RNA condensate molecules disclosed herein. Embodiments of the invention include a living cell comprising a plurality of different RNA condensates made by the methods disclosed herein. In certain embodiments of the invention that include multiple different synthetic RNA condensate molecules, and at least 6 nucleotides in a first loop are orthogonal with loops in other synthetic RNA condensate molecules so as to form distinct condensates that do not mix. In certain embodiments of the invention, the cell is transfected with a plurality of different vectors comprising a plurality of different DNA molecules so as to form a plurality of RNA condensates within the cell that do not aggregate. Other objects, features and advantages of the present invention will become apparent to those skilled in the art from the following detailed description. It is to be understood, however, that the detailed description and specific examples, while indicating some embodiments of the present invention, are given by way of illustration and not limitation. Many changes and modifications within the scope of the present invention may be made without departing from the spirit thereof, and the invention includes all such modifications. BRIEF DESCRIPTION OF THE DRAWINGS Figure 1. Circularized RNA nanostars generate condensates in different cell lines. A, We transfected ssRNA nanostar in mammalian cells using the Tornado plasmid, which results in a circularized motif. ssRNA nanostars form condensates through their kissing loop interactions. B, Representative confocal microscopy images of RNA condensate forming in HEK293T, HeLa, and U-2 OS cells. Nuclei of condensate-expressing cells are outlined. C, Schematic of RNA nanostars recruiting MCP-mCherry through the addition of an MS2 aptamer as an additional arm. D, Split channel fluorescent images of a HEK293T cell co-transfected with MCP-mCherry and MS2-nanostar stained with DFHBI. FRAP analysis was performed on condensates highlighted with white squares and shown in E. The nucleus is outlined in the merged channel image. E, Fluorescence images (top) and pixel intensity profile (bottom) at different time points after photobleaching. Condensates were imaged before, and every 200 ms for 90 s after bleaching. Blue and orange dots indicate the mean pixel intensity of the bleached area at the corresponding time point. Error bars indicate standard error. The dark blue and red lines indicate the fitted curve from equation . Mean and standard error were calculated from 3 fields one replicate. F, Epifluorescence micrographs (top) demonstrating coalescence of droplets formed during in vitro transcription at 37°C. RNA was labeled with CY3-UTP. Time-dependent change (bottom) in aspect ratios of condensates (gray dots), computed as the major and minor axes ratio from the best-fit- ellipses. The dashed line is a fit of the exponential function . G, Epifluorescence micrographs (top) and time-dependent change in ratios of condensates (bottom) of an oval droplet formed in HEK293T no relaxation over the imaging period. Fluctuation in aspect ratios results from condensate movement. Cells were stained with Hoechst (blue) and DFHBI (green). H, Co- localization of condensates with other cellular membraneless organelles in HEK293T cells. Cells were stained with Hoechst (blue), DFHBI (green), primary antibodies and Alexa Fluor™ 647 labeled secondary antibodies (pink). Nuclei of condensate expressing-cells are outlined in the merged channel image. All cellular expression results were captured 48 hours after transfection except for the time-lapse results. Micrographs in panels B and H are z-projections of stack images captured by a confocal microscope. Micrographs in panel G are single slices of the stack images. Micrographs in panels D, E, and F are captured using an epifluorescence microscope. All experiments were replicated three times; images are representative examples. Scale bar, 5 μm. Figure 2. Nuclear and cytoplasmic condensate formation depends on nanostar design. A, Schematic depicting parameters determining nanostar concentration inside or outside the nucleus, where is the transcription rate, is the nuclear export rate, is the degradation rate, is the volume ratio correction term, defined as , is the nanostar concentration inside of the nucleus, is the inside of the cytoplasm, and ∅ indicates RNA degradation products. B, C, D, Simulations generate different concentration profiles with variedexport rate ( ), while other parameters remain constant with , , and .Phase separation is initiated, and visible condensates begin to form when theconcentration surpasses the critical concentration ( ). While the transcription rate isdetermined by the amount of plasmid entering the cell, the nuclear export rate depends on the size of the nanostar. E, G, I, K Representative z-projections of confocal microscopy images showing condensate growth over the course of 48 hours from different nanostar designs. Cells were stained with Hoechst (blue) and DFHBI (Green). For each design, the nucleus of the same cell is outlined at different time points. Inserts in E are single slice midplane images, showing all condensates localized outside of the nucleus. F, H, J, L, Dot plots showing temporal evolution of individual condensate volume inside and outside of the nucleus. Above each dot plot, we report the number of sampled cells at each time point, across multiple fields of view, from three replicates. Bar plots show the average volume of condensates inside one cell. Scale bar, 5 μm. Figure 3. Engineering condensate volume distribution and localization through nanostar design. A, C, E, Representative z-projection of confocal microscopy images showing condensates produced by RNA nanostar variants with different arm length (A), arm number (C), and kissing loop sequence (E). Cells were stained with Hoechst (blue) and DFHBI (green). The yellow box highlights the same design. The nuclei of condensate-expressing cells are outlined for convenience. B, D, F, Left: Distributions of condensate volume in the nucleus and cytoplasm on a log scale. Each dot represents a condensate. Red lines indicate the mean. Right: Number of condensates in the nucleus or cytoplasm measured in individual cells. Red lines indicate the mean. G, Bar chart comparing the total condensate volume for 50 cells, and average volume per cell. H, I, Nuclear condensate volume is exponentially distributed, while cytoplasmic condensates display power-law-like distribution for nanostars with different arm lengths (H), arms numbers and kissing loops (I). To generate cumulative distribution functions (CCDFs) of condensate volumes, condensates were individually segmented in three dimensions and rescaled by the mean condensate volume in each cell. After rescaling, data from all cells were pooled and plotted collectively. Each dot represents a condensate. The yellow box highlights the same nanostar variant. The CCDF was also compared with the expectation for an exponential distribution (dashed line). All experiments were replicated three times. Scale bar, 5 μm. Figure 4. Nanostar kissing loop sequences determine condensate miscibility. A, Schematic of nanostars labeled with Broccoli or Pepper interact through complementary but non-palindromic kissing loops. Fluorescence micrographs showing that condensates form only when both nanostars are produced. B, Scatter plots showing pixel intensities in the Broccoli and Pepper channels for each pixel within the white- highlighted regions of interest, along with the corresponding Pearson correlation coefficients. In the bottom plot, nuclear and cytoplasmic condensates exhibit distinct correlations, reflecting differences in the stoichiometry of the two nanostars. These variations are likely driven by differences in fluorescence tagging, which may affect nanostar folding and, consequently, their distribution patterns. C, Schematic of nanostars with kissing loops A, B, C, respectively labeled with Broccoli, Pepper, or Mango aptamers and fluorescence micrographs of the corresponding condensates in cells. D, Distinct condensates that do not mix (orthogonal) emerge upon co-transfection of plasmids each carrying one of the 15-nt arm nanostars with kissing loops A, B, C, respectively labeled with Broccoli, Pepper, or Mango aptamers. E, H, Confocal fluorescence micrographs of cells expressing modified nanostars tagged with Pepper (longer arms) or Mango (weaker kissing loops) to modify their subcellular location. F, I, Dot plots illustrate changes in the distribution of nuclear and cytoplasmic condensate volume in log scale. Each dot represents a condensate. Red lines indicate the mean. G, J, Bar charts comparing the average volume of nuclear and cytoplasmic condensates. K, We redesigned the cellular localization of orthogonal condensates using modified Pepper and Mango-tagged nanostars (Broccoli design unchanged). Nuclei are outlined for convenience. HEK293T Cells were stained with Hoechst, DFHBI (for Broccoli), HBC620 (for Pepper), and TO1-B (for Mango). Mango-tagged condensates were imaged after fixing cells, due to poor permeability of the cognate dye into the nucleus. Tri-color images were acquired by live cell imaging, followed by on-stage fixation and imaging of the same field of view. All experiments were replicated three times. Images are z-projections of stack images captured by a confocal microscope. Scale bar, 5 μm. Figure 5. Chimeric linker nanostars at different ratios yield condensates with programmable sub-compartmentalization. A, Schematic of two-arm and four- arm chimeric linker nanostars. B, Increasing the ratio of chimeric linker over orthogonal nanostars yields a higher level of mixing. C, D, Larger field-of-view (top) and zoomed- in (bottom) confocal micrographs of cells producing nanostars with different ratios between DNA templates ([Broccoli NS - T]:[Linker NS - T]:[Pepper NS - T]) using two-arm (C) or four-arm (D) linkers. E, F, For two-arm linkers, mixing indices and increase parallel to the linker ratio. Higher mixing indices for four-arm linkers resulted from doubled valency. G, Confocal fluorescence micrographs show four-arm linker nanostars (labeled with Broccoli aptamer) form condensates independently. HEK293T cells were stained with Hoechst, DFHBI (for Broccoli), and HBC620 (for Pepper). White arrows indicate non-homogeneous mixing. Images were acquired 48 hours after transfection. Large field-of-view images (top) in panel C are z- projections of stack images captured by a confocal microscope. Zoomed-in view (bottom) on the bottom are single slices from the stack. All experiments were replicated three times. Scale bar, 5 μm. Figure 6. Autocatalytic circularization of RNA nanostar and condensate formation. A, We adopted an expression system in which each nanostar sequence (in green) is flanked by 5’- (in orange) and 3’- (in blue) self-cleaving ribozymes designed by Litke et al10. Once transcribed, ribozymes autocatalyse and generate functional groups on the new RNA ends. The RNA molecule then becomes a substrate for an endogenous RNA ligase, and becomes circularized before ligation. B, After endogenous circularization an RNA nanostar includes three stems (in black), three identical and self-complementary kissing loops (in orange), a ribozyme fragment (in gray, at the bottom), and other adenine spacers (in gray). Phase separation is induced by inter-molecular hybridization between kissing loops. Figure 7. Circularization significantly enhances RNA accumulation. Representative images showing fluorescence from HEK293T cells with no transfection, transfected with plasmids expressing circular Broccoli, linear, or circularized nanostars presenting three 15 nucleotide arms carrying the Broccoli aptamer (15-3A-Br). Linear 15-3A-Br was expressed using the same vector, but without the 3’- and 5’- ribozyme sequence. Without circularization, fluorescence signals were barely detectable. Images are representative of three replicates. All images were taken using the same exposure time. Scale bar, 50 μm. Figure 8. Cotranscriptional formation and growth of 15nt-3A-Br condensates in vitro. Condensate formation during in vitro transcription. DNA templates are partially annealed to have a double-stranded promoter. RNA strands were transcribed in vitro at 37°C using 7.5% (v / v) T7 polymerase and transcription buffer and stained by 40 μM DFHBI. Images are representative of three replicates. Figure 9. In-gel staining of Broccoli-tagged RNA nanostar in HEK293T cells. HEK293T cells were transfected with a plasmid encoding 15-3A-Br. After 48 hours of expression, total RNA was isolated, separated by denaturing PAGE, and stained with DFHBI-1T and 1x SYBR Gold. RNA nanostars tagged with Broccoli aptamer are clearly detectable. Circularized RNA nanostars migrated faster than linear ssRNA strands, consistent with previous observations10. Figure 10. RNA nanostars form condensates in HEK293T, HeLa, and U-2 OS cells. Cells are stained with NucBlue reagent and 40μM DFHBI. Expression level is higher in HEK293T cells as the presence of T antigen facilitates plasmid replication. We tested two variants of our nanostar design, with 15 nucleotide long arms 15-3A-Br (left) and 20 nucleotide long arms 20-3A-Br (right). Images are representative of three replicates. Scale bar, 5 μm. Figure 11. Flow cytometry plots and quantification comparing untransfected cells with cells expressing circularized Broccoli or circularized Broccoli-tagged RNA nanostars. Cells expressing condensates have lower transfection efficiency (B), are larger, exhibit higher granularity (C), and lower fluorescence intensity (D) when compared to Broccoli-expressing cells. Cells were examined 24 hours post-transfection. SSC-A, side scatter amplitude; FSC-A, forward scatter amplitude. Figure 12. Shell-like structures emerge in cells expressing a variety of nanostars, regardless of their capability to form condensates. A, Expression of circularized Broccoli yield diffused fluorescence in the cytoplasm and formed shell- like structures in the nucleus. B, Replacing kissing loops with polyadenine domains disrupted condensate formation and generated shell-like structures. Changing one kissing loop was sufficient to disrupt condensate formation. C, Expression of nanostar with a non-palindromic kissing loop formed shell-like structures in the nucleus. D, Eliminating one arm from the nanostar formed exclusively nuclear shell-like structures (top). Addition of one arm formed big, bright condensates and shell-like structures in the nucleus and small puncta in the cytoplasm. E, Motifs designed to have weak nanostar-nanostar interactions formed abundant nuclear shell-like structures. F, Shell- like structures formed in the nucleus of cells expressing nanostar 15-3A-Br. Cells were stained with NucBlue reagents and 40 μM DFHBI. Images are all z projections, and are representative of three replicates. Scale bar, 5 μm. Figure 13. Nanostars including the MS2 aptamer form condensates when adopting different kissing loops with varying strength. Representative images (left) and zoomed-in views (right) of the white-squared area showing cells transfected with nanostars that include an MS2 domain and differ by kissing loop sequence (A / B / WT). The strength of kissing loop interactions increases as GC content increases from top to bottom, resulting in more nuclear retention and more aggregate-like condensates in the cytoplasm. Cells were stained with NucBlue reagents and 40 μM DFHBI. Images are representative of three replicates. Scale bar, 10 μm. Figure 14. RNA nanostars lacking the MS2 domain do not recruit mCherry. Representative images showing transfection of plasmids expressing MCP-mCherry (top) and co-transfection plasmids expressing MCP-mCherry and nanostars (bottom). MCP-mCherry expression in cells leads to a homogenous distribution of fluorescent signals across cells. Peptide concentration in the cytoplasm is higher than in the nucleus. Without MS2 aptamer, expression of nanostar forms condensates with no colocalization with MCP-mCherry. Images are representative of three replicates. Scale bar, 10 μm. Figure 15. The ratio of RNA-nanostar-expressing plasmid and mCherry- expressing plasmid determines the efficiency of mCherry recruitment. We found a high ratio of nanostar / reporter plasmid to be necessary for recruitment. Lower ratios (1:1 and 3:1) result in excessive binding of MCP-mCherry. Nanostar-MCP-mCherry complexes in the nucleus are too large to be exported, resulting in dispersed fluorescence in the nucleus. Images are representative images of three replicates. Scale bar, 5 μm. Figure 16. FRAP of nuclear condensates formed by nanostars including the Broccoli (15-3A-Br) and Pepper (15-3A-Pp) aptamers. Representative images (top) and plots of normalized fluorescence intensity showing FRAP behavior of Broccoli aptamer (A) and Pepper aptamer (B). Cells were stained with NucBlue reagents and 40 μM DFHBI or 10 nM HBC620. Blue dots indicate mean intensity at the corresponding time point. Orange error bars indicate standard error. The dark blue line indicates the fitted curve from equation . Mean and standard error were calculated from 3 of view, each belonging to one replicate. Scale bar, 2 μm. Figure 17. FRAP of condensates (15-3A-Br nanostar) produced during in vitro transcription targeting Broccoli aptamer or Cy3. Representative images (top) and plots of normalized fluorescence intensity showing FRAP behavior of Broccoli aptamer (A) and CY3-labeled RNA (B). The in vitro transcription reaction was supplied with 1% CY3-labeled UTP, and diluted 10x using 1x transcription buffer and DFHBI (final concentration 40 μM) before imaging. Images are captured in GFP channels for A and RFP channels for B. Blue dots indicate mean intensity at the corresponding time point. Orange error bars indicate standard error. The dark blue line indicates the fittedequation . The mean and standard error were to one replicate. Scale bar, 1 μm. Figure 18. Time-dependent coalescence of condensates produced during in vitro transcription of nanostar 15-3A-Br. Nanostars were transcribed using 1% CY3- UTP for labeling; condensates were monitored in a sealed chamber, and imaged on a heat stage at 37°C. Analysis results track 16 condensate fusion events measured within one field of view from one sample. Figure 19. RNA condensates altered localization of fibrillarin (Figure S15), Coilin (Figure S16), and G3BP1 (Figure S17); while demonstrating no colocalization with nuclear speckles and P bodies. Z-projected images are representative images from three replicates. Scale bar, 10 μm. Figure 20. RNA nanostars forming nuclear condensates colocalize with the nucleolus and exclude fibrillarin. A, Z-projection (top row) and single-slice confocal image (bottom row, white square in Z-projection) showing condensates formed from 15nt-A-Br colocalizing with the nucleolus. In both high (i) and low (ii) expression cells, condensates could exclude fibrillarin, creating a dark region with reduced fibrillarin intensity, which is absent in the 10nt-A-Br nanostar (see Fig.2 of the manuscript) that forms condensate exclusively in the cytoplasm (B) or non-transfected cells (C). Although most nuclear condensates demonstrated fibrillarin accumulation on their surface, they were not exclusively nucleolar, as shown in A(iii), indicated with white arrows. Representative images from three replicates. Scale bar, 10 μm. Figure 21. Cajal bodies cluster on the surface of large nuclear condensates. A, Z-projection (top row) and single-slice confocal image (bottom row, white square in Z-projection) showing Cajal bodies clustered on the surface of condensates inside of nuclei. In both fields of view (i) and (ii), Coilin localizes at the surface of nuclear condensates. B (i), RNA nanostars that form exclusively cytoplasmic condensates (10 nt arm length, see Fig. 2 of the manuscript) did not alter Coilin localization in most expression cells. However, in some cases, Coilin seemed to be enriched into the nucleolus (B (ii)). C, Cajal body in non-transfected HEK293T cells. Representative images from three replicates. Scale bar, 10 μm. Figure 22. The formation of RNA condensates can correlate with the formation of stress granules in the cytoplasm. G3BP1 colocalizes with nuclear shell-like structures likely due to structure-mediated RNA decay. A, Z-projection (top row) and single-slice confocal images (bottom rows) of cells transfected with plasmids expressing RNA nanostars. In some cells, RNA condensation correlates with stress granule formation, as exemplified in cell 1; this does not occur in all cells, as seen in cell 2. In cell 3, nanostars form shell-like structures that colocalize with G3BP1. B, Shell-like structures are consistently observed across different fields of view in cells expressing RNA nanostars. G3BP1 colocalizes with shell-like structures but not with nuclear condensates. C, Pixel intensity profile of a shell-like structure (white square in B) illustrating its layered organization in which G3BP1 surrounds nanostars. D, E, Shell-like structures recruiting G3BP1 form also in cells transfected to express circularized Broccoli RNA, but are absent in non-transfected cells. Images are representative of three replicates. Scale bar: 10 μm. Figure 23. Nuclear speckles (A) and P bodies (B) do not colocalize with RNA condensates. Z-projection (top row) and single-slice (bottom row, white square in Z-projection) confocal images of cells were fixed 48 hours post-transfection. Images are representative of three replicates. Scale bar: 10 μm. Figure 24. Split-channel images of the micrographs shown in Figure 2. Scale bar, 10 μm. Figure 25. Individual and total condensate volume of a representative cell. A, Representative z-projection of confocal microscopy images in Fig. 2 of the manuscript, showing condensate growth over the course of 48 hours from different nanostar designs. We selected and tracked a single cell in the field of view. B, Dot plots showing individual condensate volume at each time point. Dashed line represents the mean volume. C, Line plots and grouped bar plots showing the change of total volume over time. Plots in B and C are for condensate volumes collected for an individual cell tracked over time in the corresponding row of A, indicated by white arrows. Scale bar, 5 μm. Figure 26. Condensate volume quantification workflow. A, For each confocal image, regions containing condensate-expressing cells were cropped and we manually labeled nuclear and cytoplasmic condensates (i) to generate a machine- learning segmentation pipeline using Labkit, an ImageJ plugin9. (ii). B, Comparison between a region of interest containing three condensate-expressing cells (i) and its mask (ii) that had condensates inside of the nucleus (white) and outside of the nucleus (gray). C, Masks were analyzed in IMARIS to create surfaces for condensate populations inside (ii) and outside (i) of the nucleus. D, Data cleaning was performed to exclude single-voxel signals (i). Neighboring condensates that were fused were cropped manually. Statistics like condensate volume and number were automatically calculated by IMARIS and output for further analysis. Figure 27. Split-channel images of the micrographs shown in Figure 3. Scale bar, 10 μm. Figure 28. Complementary cumulative distributions of condensate volume at different time points post-transfection. Data in Fig. 2 F, H, J, L. To generate cumulative distribution functions (CCDFs) of condensate volumes, condensates were individually segmented in three dimensions and rescaled by the mean condensate volume in each cell. After rescaling, data from all cells were pooled and plotted collectively. Each dot represents a condensate. Figure 29. mCherry recruitment to condensates is maintained across nanostar kissing loops of increasing strength, however protein diffusivity decreases. A, Condensates formed from three different types of kissing loops can recruit MCP-mCherry. Plasmids expressing MCP-mCherry and RNA nanostar are transfected in a 1:9 ratio. The stoichiometry of plasmids delivered varies between individual cells, leading to diverse expression profiles. B, FRAP was performed on condensates colocalized with mCherry fluorescence, observed in panel A, including condensates located in the cytoplasm, and condensates located in the nucleus presenting a spherical or an irregular shape. Blue dots indicate mean intensity at the corresponding time point. Shaded areas indicate the standard error. The dark blue line indicates the fitted curve from equation . Mean and standard error were calculated from 3 belonging to one replicate. Scale bar, 10 μm. Figure 30. Nucleus volume quantification workflow with IMARIS. A, Example images showing the automated segmentation. We trained a machine learning model by manually labeling the nuclei and background on slices. The same model was applied to all confocal images and generated primary segmentation results as the example image showing in (i). To split touching nuclei, we applied a watershed method with estimated object size = 8 µm (ii), where circles indicate seed positions. B, Example images showing the data cleaning process. Nuclei split along the z-axis due to incomplete imaging (i) and nuclei with abnormally high pixel intensities indicating apoptosis or mitosis cells (ii) were manually excluded. Nuclei that failed to be split using the watershed algorithm were cut manually using the cutting tool in Imaris software to ensure proper segmentation of all nuclei (iii). Touching nucleus or nuclei without a distinguishable border were also removed. C, Example image showing classification results, where nuclei of condensate-expressing cells were highlighted in green and nuclei of non-expressing cells were in blue. Figure 31. Nucleus volume quantification for nanostar-expressing cells. Nuclei of cells expressing nanostars exhibit an average volume that is consistently enlarged when compared to non-expressing cells (dashed line). The expression of condensate-forming nanostars seem to cause more significant nucleus enlargement, characterized by increased mean volume and by a shift in the overall volume distribution. This may be related to increased osmotic pressure within the nucleus caused by excessive production of circularized RNA. This analysis corresponds to data shown in Fig. 3 of the manuscript, imaged 48 hours after transfection. Each dot represents the nucleus volume from a nanostar-expressing cell that is normalized to the average nucleus volume of non-expression cells in the same experiment. Red lines indicate the mean. Figure 32. Split-channel images of the micrographs shown in Figure 4. Scale bar, 5 μm. Figure 33. Tri-color images were acquired by live cell imaging followed by on-stage fixation and fixed imaging at the same field of view. A, Split channel images of condensates shown in Fig.4D. B, Split channel images of condensates shown in Fig.4K. For fixed cell imaging results, strong Mango signals in the yellow channel spillover into the green channel, resulting in weak green signals that colocalize with the yellow signal. All experiments are replicated three times. Scale bar, 20 μm. Figure 34. A, Z-projection of confocal microscopy images showing that Pepper-labeled nanostars (Fig. 4) with a WT kissing loop (GCGCGC) produce condensates that lose spherical shape, appearing as aggregates of smaller condensates. B, C, The cellular localization of the Pepper-WT nanostars does not change when compared to Pepper-B nanostars (Fig. 4). D, Z-projection of confocal microscopy images showing that Mango-tagged nanostars (Fig. 4) with kissing loop (AUAUAU) yields nuclear shells and diffusive Mango fluorescence in the cytoplasm, but not condensates were found. All experiments are conducted in triplicate. Scale bar, 5 μm. Figure 35. Split-channel images of the micrographs shown in Figure 5. Scale bar, 10 μm. Figure 36. Broccoli- and Pepper-labeled nanostars form orthogonal condensates in the absence of RNA linkers. Nanostars with 20nt long arm and orthogonal kissing loops form non-mixing condensates. Images are representative of three replicates. Scale bar, 10 μm. Figure 37. Mixing index is influenced by non-uniform mixing. A), Representative z-projection image showing a cell transfected with nanostars with kissing loops A and B, and RNA linker at 1:4:1 plasmid ratio. B), Image showing the Broccoli signal (nanostar A) at a single slice, indicating a non-uniform mixing in which we can identify a high signal region (white arrow) and low signal region (yellow arrow). Classifying this droplet as partial (with mask generated in C) or complete (with mask generated in D) mixing leading to significantly different mixing index, as shown in E). In the manuscript, all droplets from 1:4:1 ratio are defined as complete mixing. Figure 38. Sequence specific RNA recruitment into condensates. A, Schematic of two strategies for recruiting target mRNA into condensates: (1) modifying the nanostar to include a recruitment domain complementary to the target RNA, or (2) modifying the target sequence to include a kissing loop. B, Schematic illustrating condensate recruitment of circular Broccoli aptamer as a proof-of-concept for strategies 1 and 2. C, Co-transfection of plasmids expressing the modified nanostar and the Broccoli target results in condensate formation and co-localization of Broccoli and Pepper signals. D, Complementary sequences on both the nanostar and the target RNA are required for co-localization. E(i), In strategy 1, adding a recruitment loop appears to hinder condensation, which can be rescued by increasing the nanostar template ratio. E(ii), A high ratio of target RNA containing the loop leads to shell-like condensates with co-localized Broccoli and Pepper signals in the cytoplasm. F, Co-transfection of plasmids expressing the nanostar and the kissing loop-modified Broccoli target results in condensate formation and co-localization of Broccoli and Pepper signals. DETAILED DESCRIPTION OF THE INVENTION In the description of embodiments, reference may be made to the accompanying figures which form a part hereof, and in which is shown by way of illustration a specific embodiment in which the invention may be practiced. It is to be understood that other embodiments may be utilized, and structural changes may be made without departing from the scope of the present invention. Unless otherwise defined, all terms of art, notations and other scientific terms or terminology used herein are intended to have the meanings commonly understood by those of skill in the art to which this invention pertains. In some cases, terms with commonly understood meanings are defined herein for clarity and / or for ready reference, and the inclusion of such definitions herein should not necessarily be construed to represent a substantial difference over what is generally understood in the art. Many of the aspects of the techniques and procedures described or referenced herein are well understood and commonly employed by those skilled in the art. Certain aspects and embodiments of the invention discussed below are shown in associated publications, (Fabrini et al. Co-transcriptional production of programmable RNA condensates and synthetic organelles. Nat. Nanotechnol. (2024); https: / / doi.org / 10.1038 / s41565-024-01726-x; hereinafter “Fabrini et al.”, and Stewart et al. Modular RNA motifs for orthogonal phase separated compartments. Nat. Commun. (2024) 15.1: 6244; hereinafter “Stewart et al.”) the contents of each of which is incorporated herein by reference. The following text discusses various embodiments of the invention. Condensation of RNA and proteins is central to cellular functions, and the ability to program it would be valuable in synthetic biology and synthetic cell science. Described herein is a modular platform for engineering synthetic RNA condensates from tailor-made, branched RNA nanostructures that fold and assemble co- transcriptionally. Up to three orthogonal condensates can form simultaneously and selectively accumulate guest molecules. The RNA condensates can be expressed within synthetic and living cells to produce membrane-less organelles with tunable number, size, morphology and compositions, and that display the ability to selectively capture proteins. The in-situ expression of programmable RNA condensates can underpin spatial organization of functionalities in both biological and synthetic cells. As used herein, the term "RNA condensation" refers to RNA molecules that self-assemble or aggregate together to form concentrated, liquid-like compartments within a test tube, cell or the like. These RNA molecules can be modified to interact with proteins and / or small molecules, and recruit them to the condensates. Biomolecular condensates are associated with a multitude of processes in living cells, including gene expression, metabolism, and diseases1. There is growing interest in methods for producing artificial condensates with controllable composition, viscoelastic properties, and subcellular location, as a means for manipulating cellular functions2. One approach to building artificial condensates is to establish weak, non- specific interactions among engineered proteins that carry intrinsically disordered domains3,4. Similarly, RNA molecules designed to include multiple, short sequence repeats can form weak, easily reconfigurable bonds that yield artificial RNA condensates5–8. A notable disadvantage of non-specific interactions is their promiscuity, which makes it difficult to build condensates with a compositional identity, and therefore with exclusive functions. An alternative approach is to take advantage of multivalent molecules with site-specific interactions9,10. We and others recently described in vitro assembly of RNA condensate from short single strands of RNA (100-200 nucleotides-long) that operate as multivalent particles11,12. These structural motifs, termed single-stranded RNA (ssRNA) nanostars, consist of at least three tandem stem-loops that fold during transcription, functioning as “arms” (Fig. 1A). The arms interact through sequence-specific binding of their loop domains, known as kissing loops, which are typically palindromic and identical on each arm. Kissing loops can be designed to be orthogonal, so distinct nanostars can produce distinct condensates that do not mix11,12. The modular design of these nanostars allows for the inclusion of additional arms, enabling the addition of domains for the recruitment of small molecules and proteins to the dense phase, without compromising condensate formation11,12. Here we demonstrate that ssRNA nanostars can generate condensates within living mammalian cells with controlled mixing patterns, and that the interactions and localization of these condensates can be tuned by modifying the nanostar design. We produced ssRNA nanostars in HEK293T cells using the TORNADO expression system developed by Litke et al.13, which uses the U6 promoter and includes domains for spontaneous RNA circularization to extend its half-life (Fig.6 and 7, Table 1). We started with a nanostar design that carries three 15-nucleotide (nt) long arms and each with a 6-nt long kissing loop UCGCGA, shown in Fig. 1A (design 15nt-3A-Br). We included the Broccoli aptamer in one of the arms, to visualize the expressed RNA in cells by adding its fluorogenic ligand 3,5-Difluoro-4-hydroxybenzylidene imidazolinone (DFHBI) to the media14. This nanostar design generates condensates during in vitro transcription at constant temperature (Fig. 8), and we observed comparable structures forming in HEK293T, HeLa, and U-2 OS cells transfected with the nanostar-carrying plasmids (Fig.1B, Fig.9 and 10). Transfection in HEK293T cells yields a higher expression level when compared with the other two cell types, because of the presence of the SV40 large T antigen that enhances gene expression15,16. Flow cytometry control experiments confirmed that expression of Broccoli alone with the Tornado system results in lower granularity when compared to nanostar-expressing cells (Fig. 11); microscopy controls further confirm diffuse fluorescence in the cytoplasm, with few puncta in the nucleus and shell-like structures that may correspond to circularized RNA colocalized with other organelles13,17(Fig. 12A). Replacement of at least one kissing loop with a polyA sequence disrupts condensate formation, confirming that condensation is driven by loop-loop interactions (Fig.12B). The modular structure of RNA nanostars makes it possible to include additional, non-interacting arms for the recruitment of guest molecules through aptamer domains (see, e.g., Figure 38). To illustrate this, we included the MS2 aptamer18as an additional arm, with the goal of recruiting to the condensates a reporter protein carrying an MCP domain. In our experiments we co-transfected two plasmids into cells, one encoding the modified RNA nanostar and the other encoding an MCP-mCherry reporter serving as guest molecule for the condensate (Fig. 1C, Fig. 13 and 14), and we tested the colocalization of Broccoli and mCherry signals. Live-cell imaging confirmed successful mCherry recruitment into RNA condensates (Fig. 1D, Fig. 15), though mCherry remained predominantly cytoplasmic due to the absence of a nuclear localization signal (NLS). In contrast, condensates lacking the MS2 domain did not affect the spatial localization of mCherry (Fig. 14). Using fluorescence recovery after photobleaching (FRAP) we estimated the diffusion timescale for mCherry guest protein recruited to the dense phase. Given a below 1 nM for the MS2-MCP complex19, this timescale should provide a good estimate of nanostar mobility. Cytoplasmic condensates exhibit recovery with a fitted time constant of s, while nuclear condensates show negligible recovery, indicating that RNA binding proteins in the cytoplasm may improve liquidity5(Fig. 1E). In contrast, FRAP analysis of nuclearcondensates using Broccoli ( ) or Pepper ( ) 20 reveals fasterrecovery of these small molecule fluorogenic guests, respectively and (Fig.16). In vitro experiments with Broccoli, where condensates were produced during transcription at constant temperature, exhibited a recovery profile similar to that measured in vivo (Fig. 17A). Because small-molecule ligands bind their RNA targets non-covalently, the observed recovery likely reflects ligand exchange rather than RNA mobility, explaining the distinct recovery behaviors between MCP-mCherry and RNA aptamers17,21. Conversely, condensates formed in vitro using transcribed RNA labeled with Cy3-UTP showed no recovery after bleaching, indicating low mobility of the nanostars, consistent with previous findings11(Fig.17B). Condensates formed in vitro consistently underwent fusion with slow relaxation dynamics, with one representative event exhibiting a relaxation timescale of (Fig. 1F, Fig. 18). However, in living cells, we did not capture any fusion events within our observation windows (Fig. 1G). Collectively, these findings suggest that while exchange of guest molecules occurs, these RNA condensates are highly viscous. We examined the colocalization of our RNA condensates with other membraneless organelles through immunostaining of relevant proteins (Fig. 1H, Fig. 19). Immunostaining of fibrillarin indicates shuttling of RNA nanostars to the nucleolar region perhaps due to their strong secondary structure. In the majority of condensate- expressing cells, fibrillarin is recruited to the surface but it remains excluded from condensates, demonstrating selective partitioning of molecules (Fig. 20A). Similarly, coilin also aggregates on the condensates surface, suggesting recruitment of Cajal bodies (Fig. 21A). We did not find colocalization of stress granules and RNA condensates in the cytoplasm. However, cells that produce RNA condensates present an increased likelihood of stress granule formation compared with cells that don’t, suggesting a correlation between abundance of circular RNA and metabolic stress22(Fig. 22). Additionally, in some nuclei, we found that G3BP1 colocalizes with shell- like structures, but not with condensates. In these cases, G3BP1 wraps on the outside of these shells, whose formation was observed in various cells producing circular RNA strands, irrespective of their ability to phase separate (Fig. 12 and 22). This might indicate the involvement of G3BP1 in a decay pathway targeting these circular RNAs23. The shell-like structures might also bear similarity with anisosomes emerging under demixing of RNA binding–deficient TDP-43 and the protein chaperone HSP7024. We found no colocalization with P-bodies and nuclear speckles (Fig.23). Because RNA nanostars are produced in the nucleus and then exported to the cytoplasm, their local concentration is determined by the relative speed of transcription, export outside of the nucleus, and degradation. These processes crucially affect condensate formation, which occurs only if the nanostar concentration exceeds a critical value25. We illustrate the importance of production and export rates through a simple compartment model (Fig. 2A), which does not attempt to discriminate between active and passive transport, and neglects nuclear degradation. In the model our molecular subunits (RNA nanostars) are produced at a rate of inside the nucleus and then exported into the cytoplasm at a constant rate : , . rate constant in cytoplasm and is a correction term in volume ratio, defined as . The nuclear production rate of nanostars in each cell depends primarily on the amount of plasmids delivered into the cell during transfection, a factor that should be independent of small variations in the nanostar design. Similarly, the degradation rate should only depend on the nanostar concentration. In contrast, the specific structural design features of the nanostar can affect the export constant , which is governed by the size-sensitive permeability of the nuclear membrane. Consequently, the export rate is sensitive to changes in subunit design, meaning that tuning the nanostar size and intermolecular affinity offers a way to modulate nuclear export26. In particular, depending on the export rate of subunits, the critical concentration can be reached at different times in the nucleus and in the cytoplasm, or may never be reached, as shown in Fig.2B-D. To test this, we sought to affect nuclear export rate by modifying the nanostar arm length11. We designed nanostars with arm length ranging between 10 and 20 nt, and measured the RNA condensate volume over time, 6-, 12-, 24-, and 48-hours post- transfection (Fig. 2E-J, Fig. 24 and 25)27. We found that 10-nt arm nanostars (arm length 3.4 nm, molecular weight (MW) 56.8 kD) form condensates exclusively in the cytoplasm as they likely diffuse quickly through the nuclear pore complex (Fig.2E and F), demonstrating a similar behavior as simulated in Fig. 2B. These nanostars do not colocalize with nucleoli, nor with Cajal bodies (Fig.20B and 21B), likely due to their rapid export to the cytoplasm. Nanostars with 15-nt arm (arm length 5.1 nm, MW 66.3 kDa) and 20-nt arm (arm length 6.8 nm, MW 75.8 kDa), having sizes close to the nucleus’ permeability barrier, form condensates inside the nucleus first (Fig.2G and J). Nanostars with 20-nt condensed at an earlier time point as the growth rate of condensates scales with nanostar sizes, consistent with illustrative simulation results in Fig. 3C and with our related findings using in vitro DNA nanostars28. In addition to being influenced by subunit size, export rate should also be impacted by subunit interactions, which can promote the assembly of clusters before nuclear export. To test this, we changed the 15 nt arm nanostar kissing loop, termed design A (UCGCGA), to the HIV wild type kissing loop (WT, GCGCGC), the strongest possible variant29. We observed a significant increase in the nuclear nanostar ratio, indicating higher nuclear retention driven by faster complex assembly (Fig.2K and 2L, Fig.25). We then sought to further elucidate the design parameters influencing condensate localization, abundance, and morphology, by systematically varying size (arm length), valency (number of arms) and avidity (kissing loop strength) of the nanostars28. We provide a comprehensive overview of our results in Fig. 3. For each variant, we measured corresponding condensate volume, number, and cellular localization 48 hours after transfection (Fig.26 and 27). Confirming the results of our kinetic experiments, nanostars differing by arm length produce condensates with distinct localization and size (Fig. 3A and B). Nanostars with 10-nt arms produce condensates localized exclusively in the cytoplasm, while nanostars with 15-nt, 20-nt, and 25-nt long arms generate condensates both in the cytoplasm and in the nucleus, where there are fewer but larger condensates. Next, we varied the nanostar arm number between 2 and 4, and found that the nuclear condensate volume correlates with the arm number (Fig. 3C and D). A reduction of valency by replacing kissing loop sequences with the same number of adenine bases yielded similar effects as deleting arms (Fig. 12). We also found that nanostars with more arms yield fewer but larger condensates in the nucleus, and correspondingly, more small puncta in the cytoplasm. More arms likely accelerate nuclear aggregation of nanostars, thereby slowing down transport to the cytoplasm. At the same time, the increased valency enhances nucleation in the cytoplasm, leading to the formation of more small puncta. Finally, we investigated the impact of kissing loop interaction strength on condensate formation through four variants, ranked from the weakest (F variant, UAUAUA) to the strongest (WT variant, GCGCGC) (Fig. 3E and F, Table 2). The strongest variant (WT) yields large condensates in the nucleus and a higher number of smaller puncta in the cytoplasm, resembling the behavior of the 4-arm nanostar. In contrast, weak kissing loops variants E and F primarily form nuclear shells. Overall, larger and fewer condensates accumulate in the nucleus as we increase arm number and strength of the kissing loops, likely due to the rapid formation of nuclear aggregates which hinder export. When increasing nanostar arm length, a threshold for export appears to occur between 10 and 15 nt arm length (Fig.3G). Across designs, the size distribution of condensate volume differs between the nucleus and the cytoplasm. This is evident when plotting the complementary cumulative distribution functions (CCDF), shown in Fig. 3H and I30. We normalized condensate volume with respect to the average condensate volume per cell, and found that the nuclear condensates follow an exponential distribution (the normalized CCDF of an exponential distribution is a line with a slope of -1). In contrast, the volume distribution of cytoplasmic condensates is broader than an exponential, and more consistent with a power law. This difference is likely explained by faster injection of nanostars in the nucleus compared to the cytoplasm30, consistent with the simple model presented in Fig.2A. No significant deviation in the CCDF was observed across designs, suggesting comparable nucleation and coalescence time scales. The CCDF of condensate volume measured at different time points (experiments reported in Fig. 2) shows similar trends, with nuclear condensate volume scaling exponentially, and cytoplasmic condensates scaling like a power law across designs (Fig.28). Collectively, these experiments demonstrate that RNA nanostars are a robust motif to build cellular condensates, and that structural and sequence variations make it possible to influence the cellular localization and the size of condensates. We verified that variations to the nanostar design do not compromise the ability of condensates to recruit target proteins. We modified nanostars with A, B, and WT kissing loops to include an MS2 aptamer, enabling their corresponding condensates to recruit MCP- mCherry (Fig.29A). FRAP experiments revealed slower mCherry recovery for the WT kissing loop compared to A and B, indicating that stronger kissing loop interactions result in higher condensate viscosity (Fig. 29B). Finally, we observed that, across designs, cells expressing RNA nanostars exhibit an enlarged nucleus, likely as a consequence of increased local osmotic pressure (Fig.30 and 31). Diverse RNA nanostars can be created that carry different sequences in their kissing loop domains (Fig.4, Fig.32). For example, a two-nanostar system carrying the fluorogenic aptamers, Broccoli, and Pepper (produces red fluorescence upon the addition of (4-((2-hydroxyethyl)(methyl)amino)-benzylidene)-cyanophenylacetonitrile 620 (HBC620)) can generate condensates in vitro when one component carries a non- palindromic kissing loop complementary to the other11(Fig. 4A). When only one of these two nanostars was expressed in cells, we observed diffuse fluorescence in the cytoplasm, and hollow shell structures in the nucleus that are not deemed condensates as previously observed for circularized RNA13(Fig. 4B, top and middle). When both nanostars are co-expressed, condensates exhibiting both Broccoli and Pepper fluorescence appear in the cytoplasm and nucleus, as confirmed by a high Pearson correlation coefficient (PCC). Interestingly, the ratios of Broccoli to Pepper fluorescence signals differ between cytoplasmic and nuclear condensates, as shown by distinct linear relationships in the fluorescence scatter plot (Fig.4B, bottom): while the Broccoli-carrying nanostar yields condensate both in the nucleus and the cytoplasm, the Pepper-carrying nanostar yields condensates primarily in the cytoplasm. Given the same expected hybridization energy of their kissing loops, this difference is likely attributed to the aptamer which may influence nanostar folding and structure, yielding a faster nuclear export for Pepper-carrying nanostars. Leveraging the sequence-specificity of loop-loop interactions, RNA nanostars can be optimized to generate a variety of distinct, non-interacting condensates. To illustrate this we characterized a set of three nanostars, termed A, B, and C12, each carrying distinct kissing loops and distinct fluorogenic aptamers, Broccoli, Pepper, and Mango (produces yellow fluorescence upon the addition of TO1-Biotin (TO1-B)) respectively31. Each of the A, B, and C kissing loops was designed to engage in homomeric binding but not heteromeric interactions with other kissing loops12. Fig.4C shows that each nanostar generates condensates, with differences in their cellular localization: while nanostar with the Broccoli and Pepper aptamer demonstrated distributions consistent with described earlier, nanostar with kissing loop C and Mango primarily yield condensate with a less spherical morphology in the nucleus. In this case, the Mango aptamer may introduce additional loop-loop bonds that increase the overall nanostar interaction strength and promote nuclear aggregation. The simultaneous transfection of multiple plasmids each carrying a distinct nanostar with non-interacting kissing loops results in the formation of condensates that do not mix (Fig.4D, Fig.33) and maintain the subcellular localization observed when expressed individually (Fig. 4C). Taking advantage of the design guidelines identified in Fig. 2 and 3, we then sought to shift the cellular localization of Pepper-tagged condensates (from cytoplasm to nucleus) and Mango-tagged condensates (from nucleus to cytoplasm). To localize Pepper-tagged nanostars to the nucleus, we extended their arm length from 15 nt and 20 nt, and to localize Mango-tagged nanostars to the cytoplasm, we adopted weaker kissing loops, switching from design C (2 AU pairs and 4 GC pairs) to design E (4 AU pairs and 2 GC pairs). With this change, we observed Pepper condensates primarily in the nucleus (Fig. 4E). While the size of nuclear condensate remains similar, their number significantly increased resulting in a higher nuclear-to-cytoplasmic volume ratio (Fig.4F and 4G). For Pepper-carrying nanostars, enhancing kissing loop strength from A to WT showed low efficiency in relocating condensates to the nucleus (Fig.34). For Mango-carrying nanostars, weakening the kissing loop from C to E successfully relocated condensates into the cytoplasm, although a substantial fraction still formed in the nucleus (Fig.4H-J). Further weakening the kissing loop to F (6 AU pairs) eliminated condensate formation and yielded only nuclear shells (Fig 34D). Cotransfection of the new plasmid triplet resulted in orthogonal condensates with the expected changes in subcellular distribution (Fig.3K). Collectively, these results indicate that the principles governing the orthogonality, morphology, and cellular localization of condensates remain valid across variants; however, the efficiency in manipulating condensate properties is affected by details of nanostar and aptamer sequence. We finally demonstrate how to systematically co-localize two orthogonal condensates, or even completely mix the corresponding nanostars, through the production of RNA linker motifs (Fig. 5, Fig 35) previously characterized in vitro12,32. A linker is a “chimeric” nanostar in the sense that it includes kissing loops complementary to two of the orthogonal nanostar one wishes to link. We tested two types of linkers, with two or four arms (Fig.5A). The colocalization level of orthogonal nanostars depends on the ratio of the linker plasmid concentration relative to the concentration of plasmids of the orthogonal nanostars (Fig.5B, Fig.36). Fig.5C shows example images of nanostars A and B (20nt arm, labeled with Broccoli and Pepper respectively) in living cells which are transfected with plasmids at increasing two-arm linker levels. At nanostar:linker:nanostar plasmid ratios of 2:1:2, 1:1:1, 1:2:1, 1:3:1, nuclear condensates with Broccoli and Pepper fluorescence are colocalized while they remain distinct compartments producing Janus-like morphologies. At higher ratios, nanostars become completely mixed. We observed similar results for the four-arm linker (Fig. 5D), however a 1:2:1 ratio of nanostar:linker:nanostar is sufficient to produce mixed condensates in the nucleus. Because now each linker carries twice as many arms, this result is consistent when compared with the two-arm linker yielding mixed condensates at a 1:4:1 ratio. In contrast, cytoplasmic condensates exhibit the opposite trend. Compared to the orthogonal nanostars, two-arm linkers are smaller, facilitating faster nuclear export and resulting in complete mixing at lower linker ratios in the cytoplasm (Fig.5C). Meanwhile, four-arm linkers are larger than the orthogonal nanostars, leading to slower nuclear export. Consequently, a higher linker ratio is required to achieve complete mixing in cytoplasmic condensates (Fig.5D).The degree of mixing between the two phases of nuclear condensates was quantified through mixing indices and that measure the fraction of red in green, and green in red condensate12(Fig. 5E and 5F). Compared with , index showed more fluctuation due to a lower signal-to-noise ratio of Broccoli compared to Pepper (Fig.37). Notably, at the ratios of 1:4:1 and 1:5:1, we observed non-homogeneous mixing with the four- arm linker. To our surprise, by labeling the two-arm and four-arm linkers with the Broccoli aptamer, we found that four-arm linkers form condensates when expressed individually (Fig. 5G), suggesting that the non-homogeneous mixing observed in Fig. 5D (white arrows) likely resulted from the local formation of non-fluorescent labeled linker clusters, as indicated by the lower fluorescence in the microscopy images. We have demonstrated a simple strategy to build programmable, orthogonal RNA condensates in living mammalian cells from short single strands of RNA folding into star-shaped motifs, taking advantage of recent progress in DNA and RNA nanotechnology32–37. With a constructive approach, our work confirms that sequence- specific RNA interactions can not only promote condensation, but also have a structural role, control condensate properties, and determine condensate identity2,38–43. For the first time, we observed the effects of RNA sequence on tuning condensate localization independent of any cellular localization signals. An exciting application of these programmable nanostars is the sequence-specific hybridization to cellular nucleic acids without requiring a protein recognition site44, which could immediately enable control of RNA nuclear export and translation. In illustrative embodiments of the invention, the synthetic RNA condensate molecule comprises a circular polynucleotide comprising 100-300 nucleotides. In addition, the synthetic RNA condensate molecule is formed to comprise complementary polynucleotide sequences that allow the RNA polynucleotide to self- assemble so as to form 3, 4, 5 ,6, 7 or 8 arm structures, said arm structures comprising a stem and a loop. In these three-dimensional structures, the stem in the arm structures comprise at least 10 paired nucleotides (i.e. are coupled to a complementary stretch of at least 10 paired nucleotides also in the synthetic RNA condensate molecule). The loops in the arm structures typically comprise at least 6 nucleotides. In addition, in these molecules, the base of the stems (distal to the loop) in the arm structures can comprise at least 1-3 unpaired linker nucleotides that couple at least 10 paired nucleotides in a first arm structure to at least 10 paired nucleotides in a second arm structure. In addition, in these synthetic RNA condensate molecules, a first loop in the synthetic RNA condensate can comprise at least 4-6 nucleotides selected to be complementary to at least 4-6 nucleotides in a second loop (in the same or a different) synthetic RNA condensate molecule) such that the at least 4-6 nucleotides in the first loop of a first synthetic RNA condensate molecule and the at least 4-6 molecules in a second loop of a second synthetic RNA condensate molecule couple together via base pairing so that the synthetic RNA condensate molecules aggregate to form a RNA condensate. In some embodiments of the invention, the RNA condensate molecule is further coupled to a protein and / or a nucleic acid. In certain embodiments of the invention, the DNA is a linear dsDNA and / or is transcribed using a purified mRNA polymerase. Optionally in these methods, the DNA is transcribed within a human cell. In illustrative working embodiments of the invention, 6 nucleotides in a loop of the molecule comprise a HIV-type Kissing-Loop motif, however other loop interaction sequences are possible, with such sequences typically selected to have differing amount of G:C parings in the loops. In typical methods of the invention, the DNA sequence is created so that the loops are disposed in a single transcribed synthetic RNA condensate molecule, and in addition, the loops in the single synthetic RNA condensate molecule are disposed in a molecular architecture where they should not couple together in a single polynucleotide molecule (e.g. the loops are held by atomic forces distal to each other in the molecular architectural framework of the single polynucleotide molecule). In certain embodiments of these methods of making a synthetic RNA condensate molecule in a mammalian cell, the synthetic RNA condensate molecule comprises a RNA aptamer motif sequence; and / or the synthetic RNA condensate molecule comprises a polynucleotide sequence to which a RNA aptamer couples via base pairing. Optionally, for example, this RNA aptamer motif sequence is disposed in a stem, or can be included as an additional domain on the 3′ or 5′ end of the condensing RNA polynucleotide; and / or the synthetic RNA condensate molecule comprises a 3′ or 5′ single stranded overhang (typically of at least 10 or 15 residues), to which a RNA aptamer connects via base pairing. Embodiments of the invention can incorporate a variety of aptamers known in the art such as those described in the RNAapt3D and / or UTexas Aptamer Databases. RNAapt3D is a database of RNA aptamer which contains RNA sequences, their secondary / tertiary structural information, target proteins of RNA aptamers and network of RNA-protein interaction. Ryuma Sato, Koji Suzuki, Yuichi Yasuda, Atsushi Suenaga, Kazuhiko Fukui, "RNAapt3D: RNA aptamer 3D-structural modeling database", Biophys J. 2022 Sep 22;S0006- 3495(22)00773-1. The UTexas Aptamer Database includes aptamer sequences, binding affinities, and characteristics of over 1,400 aptamers, as well as a description of the original nucleic acid library, binding buffer conditions, and the DOIs of the original aptamer publications (Askari, A., et al., Nucleic Acids Research, Volume 52, Issue D1, 5 January 2024, Pages D351–D359. In certain embodiments of the invention, the methods disclosed herein are designed to form a plurality of synthetic RNA condensate molecules by transcribing a plurality of DNA sequences selected to encode a plurality of RNA condensation sequences (so that a plurality of synthetic RNA condensate molecules are formed). Related embodiments of the invention include methods of making a RNA condensate comprising transducing a mammalian cell with a vector comprising a plurality of synthetic RNA condensates designed to comprise 100-300 nucleotides and be circular. Typically in these methods, the synthetic RNA condensates comprise complementary sequences that allow the synthetic RNA condensates to self-assemble so as to form at least 3 arm structures comprising a stem and a loop, wherein: the stem in said arm structures comprise at least 10 paired nucleotides; the loops in said arm structures comprise at least 6 nucleotides; at least a first loop in said arm structures comprises at least 6 nucleotides selected to be complementary to at least 6 nucleotides in at least a second loop such that the at least 6 nucleotides in a first loop of a first synthetic RNA condensate and the at least 6 molecules in a second loop of a second synthetic RNA condensate couple together so that the synthetic RNA condensates aggregate to form a RNA condensate; wherein the plurality of synthetic RNA condensates are combined under conditions selected to facilitate condensation; and then allowing the plurality of synthetic RNA condensates to aggregate so that the RNA condensate is made in the cell. In certain of these methods, the plurality of RNA condensates comprise a homogeneous population of polynucleotides. In other embodiments of the invention, the plurality of synthetic RNA condensates comprise a heterogeneous population of polynucleotides. In some embodiments of the invention, the plurality of synthetic RNA condensates are selected to include arm and loop structures that facilitate condensation formation in the nucleus of the cell. In some embodiments of the invention, the plurality of synthetic RNA condensates are selected to include arm and loop structures that facilitate condensation formation in the cytoplasm of the cell. In certain embodiments of the invention, the plurality of synthetic RNA condensates are further coupled to proteins and / or nucleic acids. Embodiments of the invention include compositions of matter comprising a synthetic RNA condensate molecule disclosed herein. For example, one such embodiment of the invention comprises a composition of matter comprising at least one circular synthetic RNA condensate molecule made by the methods disclosed herein. Optionally these compositions further comprise a protein, for example a protein coupled to the synthetic RNA condensate molecule via a RNA aptamer. In certain compositions of the invention, the synthetic RNA condensate molecule is disposed within a synthetic cell. A related composition of the invention comprises a RNA condensate formed by a plurality of aggregated synthetic RNA condensate molecules made by the methods disclosed herein. Additional embodiments of the invention include methods of making a RNA condensate. Typically these methods comprise combining together (e.g. within a mammalian cell) a plurality of synthetic RNA condensate molecules selected to comprise 100-300 nucleotides and further comprise complementary sequences that allow the synthetic RNA condensate molecules to self-assemble so as to form 3, 4, or 5 or more arm structures comprising a stem and a loop. In such methods, the stem in the arm structures comprise at least 10 paired nucleotides (i.e. are coupled to a complementary stretch of at least 10 paired nucleotides also in the synthetic RNA condensate molecule); the loops in the arm structures comprise at least 6 nucleotides; and the base of the stems (distal to the loop) in the arm structures comprise at least 1-3 unpaired linker nucleotides that couple the at least 10 paired nucleotides in a first arm structure to the at least 10 paired nucleotides in a second arm structure. In such methods, the synthetic RNA condensate molecules are selected to comprise at least a first loop in the arm structures comprises at least 4-6 nucleotides selected to be complementary to at least 4-6 nucleotides in at least a second loop such that the at least 4-6 nucleotides in a first loop of a first synthetic RNA condensate molecule and the at least 4-6 molecules in a second loop of a second synthetic RNA condensate molecule couple together so that the synthetic RNA condensate molecules aggregate to form a RNA condensate. In these methods, the plurality of synthetic RNA condensate molecules are combined under aqueous conditions selected to facilitate condensation (e.g. physiological conditions). These methods further include allowing the plurality of synthetic RNA condensate molecules to aggregate so that the RNA condensate is made. In some embodiments of these methods, the plurality of synthetic RNA condensate molecules comprise a homogeneous population of polynucleotides. In other embodiments of these methods, the plurality of synthetic RNA condensate molecules comprise a heterogeneous population of polynucleotides. In certain embodiments of these the plurality of synthetic RNA condensate molecules are further coupled to proteins and / or nucleic acids. Numerical References Above 1. Alberti, S. & Hyman, A. A. Biomolecular condensates at the nexus of cellular stress, protein aggregation disease and ageing. Nat. Rev. Mol. Cell Biol. 22, 196– 213 (2021). 2. Dai, Y. et al. Programmable synthetic biomolecular condensates for cellular control. Nat. Chem. Biol.19, 518–528 (2023). 3. Schuster, B. S. et al. Controllable protein phase separation and modular recruitment to form responsive membraneless organelles. Nat. Commun. 9, 2985 (2018). 4. Dzuricky, M., Rogers, B. A., Shahid, A., Cremer, P. S. & Chilkoti, A. De novo engineering of intracellular condensates using artificial disordered proteins. Nat. Chem.12, 814–825 (2020). 5. Jain, A. & Vale, R. D. RNA phase transitions in repeat expansion disorders. Nature 546, 243–247 (2017). 6. Xue, Z. et al. Targeted RNA condensation in living cells via genetically encodable triplet repeat tags. Nucleic Acids Res.51, 8337–8347 (2023). 7. Ji, R. et al. RNA Condensate as a Versatile Platform for Improving Fluorogenic RNA Aptamer Properties and Cell Imaging. J Am Chem Soc 146, 4402–4411 (2024). 8. Wu, J. et al. Self-assembly of intracellular multivalent RNA complexes using dimeric Corn and Beetroot aptamers. J. Am. Chem. Soc.144, 5471–5477 (2022). 9. Banani, S. F. et al. Compositional Control of Phase-Separated Cellular Bodies. Cell 166, 651–663 (2016). 10. Li, P. et al. Phase transitions in the assembly of multivalent signalling proteins. Nature 483, 336–340 (2012). 11. Stewart, J. M. et al. Modular RNA motifs for orthogonal phase separated compartments. Nat. Commun.15, 1–13 (2024). 12. Fabrini, G. et al. Co-transcriptional production of programmable RNA condensates and synthetic organelles. Nat. Nanotechnol.1–9 (2024). 13. Litke, J. L. & Jaffrey, S. R. Highly efficient expression of circular RNA aptamers in cells using autocatalytic transcripts. Nat. Biotechnol.37, 667–675 (2019). 14. Filonov, G. S., Song, W. & Jaffrey, S. R. Spectral Tuning by a Single Nucleotide Controls the Fluorescence Properties of a Fluorogenic Aptamer. Biochemistry 58, 1560–1564 (2019). 15. Tan, E., Chin, C. S. H., Lim, Z. F. S. & Ng, S. K. HEK293 Cell Line as a Platform to Produce Recombinant Proteins and Viral Vectors. Front. Bioeng. Biotechnol.9, 796991 (2021). 16. DuBridge, R. B. et al. Analysis of Mutation in Human Cells by Using an Epstein- Barr Virus Shuttle System. Molecular and Cellular Biology (1987) doi:10.1128 / mcb.7.1.379-387.1987. 17. Bereiter, R. et al. Engineering covalent small molecule-RNA complexes in living cells. Nat Chem Biol (2025) doi:10.1038 / s41589-024-01801-3. 18. Johansson, H. E., Liljas, L. & Uhlenbeck, O. C. RNA recognition by the MS2 phage coat protein. Semin. Virol.8, 176–185 (1997). 19. Johansson, H. E. et al. A thermodynamic analysis of the sequence-specific binding of RNA by bacteriophage MS2 coat protein. Proc. Natl. Acad. Sci. U. S. A. 95, 9244–9249 (1998). 20. Chen, X. et al. Visualizing RNA dynamics in live cells with bright and stable fluorescent RNAs. Nat. Biotechnol.37, 1287–1293 (2019). 21. Li, X., Wu, J. & Jaffrey, S. R. Engineering fluorophore recycling in a fluorogenic RNA aptamer. Angew. Chem. Weinheim Bergstr. Ger.133, 24355–24363 (2021). 22. Zhou, W.-Y. et al. Circular RNA: metabolism, functions and interactions with proteins. Mol. Cancer 19, 172 (2020). 23. Fischer, J. W., Busa, V. F., Shao, Y. & Leung, A. K. L. Structure-Mediated RNA Decay by UPF1 and G3BP1. Mol Cell 78, 70–84.e6 (2020). 24. Yu, H. et al. HSP70 chaperones RNA-free TDP-43 into anisotropic intranuclear liquid spherical shells. Science 371, (2021). 25. Hyman, A. A., Weber, C. A. & Jülicher, F. Liquid-liquid phase separation in biology. Annu. Rev. Cell Dev. Biol.30, 39–58 (2014). 26. Timney, B. L. et al. Simple rules for passive diffusion through the nuclear pore complex. J Cell Biol 215, 57–76 (2016). 27. Arzt, M. et al. LABKIT: Labeling and segmentation toolkit for big image data. Front. Comput. Sci.4, (2022). 28. Agarwal, S., Osmanovic, D., Klocke, M. A. & Franco, E. The Growth Rate of DNA Condensate Droplets Increases with the Size of Participating Subunits. ACS Nano 16, 11842–11851 (2022). 29. Clever, J. L., Wong, M. L. & Parslow, T. G. Requirements for kissing-loop- mediated dimerization of human immunodeficiency virus RNA. J. Virol.70, 5902– 5908 (1996). 30. Lee, D. S. W. et al. Size distributions of intracellular condensates reflect competition between coalescence and nucleation. Nat Phys 19, 586–596 (2023). 31. Cawte, A. D., Unrau, P. J. & Rueda, D. S. Live cell imaging of single RNA molecules with fluorogenic Mango II arrays. Nat. Commun.11, 1283 (2020). 32. Jeon, B.-J., Nguyen, D. T. & Saleh, O. A. Sequence-Controlled Adhesion and Microemulsification in a Two-Phase System of DNA Liquid Droplets. J. Phys. Chem. B 124, 8888–8895 (2020). 33. Chang, G., Fygenson, D. K. & Saleh, O. A. Emulsion imaging of a DNA nanostar condensate phase diagram reveals valence and electrostatic effects. The Journal of Chemical (2022). 34. Nguyen, D. T. & Saleh, O. A. Tuning phase and aging of DNA hydrogels through molecular design. Soft Matter 13, 5421–5427 (2017). 35. Udono, H. et al. Programmable computational RNA droplets assembled via kissing-loop interaction. ACS Nano 18, 15477–15486 (2024). 36. Agarwal, S., Osmanović, D., Dizani, M., Klocke, M. & Franco, E. Dynamic control of DNA condensation. Nat. Commun.15, 1915 (2024). 37. Malouf, L. et al. Sculpting DNA-based synthetic cells through phase separation and phase-targeted activity. Chem 2023.03.17.533162 (2023) doi:10.1101 / 2023.03.17.533162. 38. Roden, C. & Gladfelter, A. S. RNA contributions to the form and function of biomolecular condensates. Nat. Rev. Mol. Cell Biol.22, 183–195 (2021). 39. Poudyal, R. R., Sieg, J. P., Portz, B., Keating, C. & Bevilacqua, P. RNA sequence and structure control assembly and function of RNA condensates. RNA 27, 1589– 1601 (2021). 40. Garcia-Jove Navarro, M. et al. RNA is a critical element for the sizing and the composition of phase-separated RNA–protein condensates. Nat. Commun. 10, 1– 13 (2019). 41. Rhine, K., Vidaurre, V. & Myong, S. RNA Droplets. Annu. Rev. Biophys.49, 247– 265 (2020). 42. Bose, M., Rankovic, B., Mahamid, J. & Ephrussi, A. An architectural role of specific RNA-RNA interactions in oskar granules. Nat Cell Biol 26, 1934–1942 (2024). 43. Langdon, E. M. et al. mRNA structure determines specificity of a polyQ-driven phase separation. Science 360, 922–927 (2018). 44. Kim, J. & Franco, E. RNA nanotechnology in synthetic biology. Curr. Opin. Biotechnol.63, 135–141 (2020). EXAMPLE MATERIALS AND METHODS FOR MAKING PROGRAMMABLE ARTIFICIAL RNA CONDENSATES IN MAMMALIAN CELLS Exemplary sequence design Nanostars were designed based on published in vitro results1using NUPACK2. For each design, 10 NUPACK trials were run, the one that generated the lowest defect score was selected. Broccoli, Pepper, Mango and MS2 aptamer sequences were taken from the literature3–6. Exemplary design of nanostar stem sequences and inclusion of aptamer domains The stem sequence used in nanostar variant 15nt-3A-Br (Fig. 1 of the manuscript) was adapted from Stewart et al. (Stem 1 design)1. We used NUPACK scripts to generate the sequence of 10 and 20 nt-long stems de novo, as summarized in Note 1 below. The 25 nt-long stem sequence was adapted from Fabrini et al7. Stem sequences were modified to insert Broccoli and Pepper aptamer as part of one of the arms, assuming that nanostar folding assists with correct aptamer folding. We adapted the length of aptamer-including stems to remain comparable with the other nanostar arms. For 10nt- / 15nt- / 20nt- long arms, we respectively used 4bp+4bp / 6bp+6bp / 8bp+8bp long domains to flank aptamers. For the 25nt-arm, 8bp+8bp domains were used to flank aptamers due to limitations in DNA synthesis. The sequence of the arm with aptamers was optimized using the NUPACK script in Note 2 below. Because Mango and MS2 aptamers include a functional loop, they cannot be inserted into an arm. For this reason, we included them as an additional arm at the 3’ end of their nanostar. We chose the 3’ end to prioritize nanostar transcription and folding relative to aptamer folding. Broccoli, Pepper, and Mango aptamers were selected to demonstrate orthogonality because 1) they have non overlapping emission spectra; 2) their fluorescence does not require aptamer dimerization, which would introduce undesired interactions between nanostars. To develop orthogonal nanostars (Fig. 4 and 5 of the manuscript), the stem of 15nt-3A-Br was modified to minimize interactions between nanostars. Using a script similar to the one in Note 1, we generated multiple 15nt-long stems. We selected two presenting the lowest defect scores and minimal interaction with the 15nt-3A-Br nanostar, and we modified them to include Pepper and Mango aptamers. Exemplary design of nanostar kissing loop sequences All kissing loops are 9 nt long and include a 6 nt interaction sequence flanked by 3 unpaired adenine residues, 2 upstream and 1 downstream of the interaction sequence (5’-AA…A-3’). The wild-type kissing loop (5’-GCGCGC) was adapted from the HIV-1 palindromic kissing loop sequence8. Orthogonal kissing loops 5’-UCGCGA, 5’-GUCGAC, and 5’-GGUACC were taken from Fabrini et al.7Kissing loops 5’- GUAUAC and 5’-UAUAUA were designed by simply replacing GC pairs with AU pairs. Non-palindromic kissing loops were adapted from the 3sβ set designed by Stewart et al1. Exemplary RNA synthesis for in vitro characterization All RNA strands for in vitro experiments were transcribed from custom DNA templates synthesized by Integrated DNA Technologies as LabReady resuspension, standard desalt purification. We annealed non-coding DNA templates with a 21-nt complement including the T7 promoter region and a 4 nt sealing domain (5’-GCGC). These templates were annealed in 1X TE / 50 mM NaCl from 90°C to RT at -1°C / min. RNA strands were individually transcribed using the AmpliScribe T7-Flash transcription kit (ASF3507, Biosearch Technologies) following the manufacturer’s protocol. RNA strands were then purified using the Monarch® Spin RNA Cleanup kit (T2040S). We estimated the concentration of purified RNA using a Nanodrop 2000c by measuring absorption at 260 nm, and the extinction coefficients were calculated using OligoAnalyzer Tool provided by the manufacturer. Exemplary plasmid development The highly stable nanostar stem-loop domains facilitate polymerase dissociation. As a result, polymerase chain reactions (PCR) of nanostar DNA templates generate products with incorrect lengths and sequences. For this reason, inserts were directly purchased from Integrated DNA Technologies as two single-stranded, 5’ phosphorylated oligonucleotides containing the sequence of interest, flanked by NotI and SacII restriction sites. The two strands were annealed in 50 mM NaCl and 1x TE buffer using a heat treatment protocol including a 5-minute heat at 90°C, followed by a slow temperature ramp at -1°C / min, and held at 20°C. The resulting products were double-stranded DNA fragments with sticky ends ready for ligation. After annealing, strands were purified with a DNA cleanup kit (NEB T1030). The DNA encoding the nanostar sequences was inserted in the pAV-U6+27-Tornado-Broccoli (Addgene 261587) plasmid. Plasmids were prepared by (1) digestion with NotI-HF (NEB R3189S) (2 μL for 20 μL reactions) at 37°C for 1 hour; (2) purification with the DNA cleanup kit; (3) digestion with SacII (NEB R0157S) (2 μL for 20 μL reactions) at 37°C for 1 hour; and (4) purification with a 0.8% 1x TAE agarose gel to select the product with the correct size. Digested backbones were finally purified using a gel extraction kit (Qiagen 28704). Digested backbone and inserts were ligated at a 1:10 molecular ratio by overnight incubation with T4 DNA ligase (NEB M0202S) at 4°C. Ligated plasmids were transformed into 50 μL DF5Hα competent cells (Thermofisher EC0112 and 18258012) following the manufacturer’s protocol. We then extracted plasmid DNA using a Miniprep kit (Qiagen 27106) following the manufacturer’s protocol. Extracted plasmids were finally sequenced by Eurofins Genomics (whole plasmid sequencing service). The plasmid expressing MCP-mCherry was purchased from Addgene (207668). Exemplary cell culture and maintenance HEK293T (ATCC® CRL-3216™), HeLa (ATCC® CCL-2™), and U-2 OS (ATCC® HTB-96™) cells were grown in Dulbecco Modified Eagle’s Medium (DMEM), high glucose, pyruvate (ThermoFisher 11995065) containing 10% Fetal Bovine Serum (FBS) and 100 U / ml Penicillin / Streptomycin (Thermo Fisher) and maintained at 37 °C with 5% CO2 in a humidified incubator. Cells used for imaging were cultured in µ-Slide 8 Well high slides (Ibidi GmbH). Exemplary transfection Seeding density was adapted across cell types to achieve ~70% confluence at transfection. Three wells were seeded as replicates for each condition tested. Lipofectamine 2000 (ThermoFisher 11668019) was used for transfecting HEK293T cells. FuGene HD (Promega E2311) was used for transfecting HeLa and U-2 OS cells as it demonstrated less cytotoxicity (additional details below). For experiments involving the expression of multiple nanostars, the total amount of plasmid DNA used in each experiment was kept constant, with an equal proportion of each nanostar variant. For experiments involving the co-delivery of plasmids expressing nanostars and MCP-mCherry, the total amount of plasmid DNA also remained constant. The plasmids encoding nanostars and MCP-mCherry were mixed and delivered in a 9:1 ratio. Transfection using Lipofectamine 2000 HEK293T cells were seeded at 5*105cells / mL, 500 µL / well into 24 well plates one day before transfection. Transfection was performed following the manufacturer's protocol, with a 500 ng final plasmid concentration and a 2 µL final Lipofectamine 2000 volume per well. The DNA-lipid complex was incubated with cells for 4-6 hours and then aspirated and changed to complete DMEM as mentioned above. Cells were reseeded into µ-Slide 8 Well high slides (Ibidi GmbH) the following day at 5*105cells / mL for imaging. Slides were pre-coated by incubating with 0.001% poly-L-lysine for at least 30 minutes, followed by washing once with PBS before the addition of cells. We incubated reseeded cells overnight and imaged them the next day. 1.6.2 Transfection using FuGene HD U-2 OS cells were seeded at 4*105 / mL, 500 µL / well into 24 well plates one day before transfection. Transfection was performed by mixing 1.65 μg of plasmid with OptiMEM to a total volume of 78 μL. The mixture was vortexed for 1-2 s and centrifuged down. Then we added 4.95 μL Fugene HD to the mixture, vortexed for 1 s, centrifuged down, and incubated at room temperature for 15 minutes. For HeLa and U-2 OS cells, a 15 μL mixture was added to each well. Media change was performed the next day before reseeding. Cells were reseeded into 8-well Ibidi slides following the same protocol as HEK293 cells using the seeding density mentioned at the beginning of this paragraph. Fluorescence microscopy and live cell imaging Live cell staining The culture medium from overnight incubation was aspirated and replaced with fresh medium supplemented with 2 drops of NucBlue Live reagent (Thermo Fisher R37605) per mL of media, along with the appropriate staining dyes according to experimental conditions. For conditions involving the Broccoli aptamer, we used 40 µM of 3,5-Difluoro-4-hydroxybenzylidene imidazolinone (DFHBI) (Lucerna, 400- 5mg); for experiments involving the Pepper aptamer, we supplied 10 nM of (4-((2- hydroxyethyl)(methyl)amino)-benzylidene)-cyanophenylacetonitrile 620 (HBC620) (MedChemExpress, HY-133520). Live cells were then incubated for at least 15 minutes at 37°C before imaging. Cells were imaged in the presence of dyes. Fixed cell staining and immunostaining Mouse anti-Coilin (Cajal body colocalization) was purchased from Abcam (ab11822, 1:1900, 1 μg / mL). Mouse anti-SC35 (nuclear speckle colocalization) was purchased from Abcam (ab11826, 1:200, 5 μg / mL). Goat anti-fibrillarin (nucleolus colocalization) was purchased from Antibodies.com (A85370, 1:200). Mouse anti- G3BP1 (stress granules colocalization) was purchased from Thermo Fisher (66486-1- IG, 1:200, 5 μg / mL). Mouse anti-DCP1A (P body colocalization) was purchased from Novus biological (H00055802-M06, 1:200). Before fixation, cell culture media were removed and cells were rinsed with PBS (Thermo Fisher 10010023). Cells were then fixed in PBS buffer (Thermo Fisher 14190144) containing 4% paraformaldehyde (Thermo Fisher 043368.9M) for 10 minutes at room temperature, and washed with the PBS buffer three times, each for 5 minutes. Next, we permeabilized cells with 0.5% Triton X-100 (Sigma Aldrich 9002- 93-1) in the PBS buffer for 10 minutes and washed them three times. For imaging condensates involving the Mango aptamer, we added PBS supplemented with NucBlue reagent (Thermo Fisher R37605) and 200 nM of TO1-Biotin (TO1-B) (ABM, G955). Cells were incubated in the buffer for 15 minutes before imaging. For experiments involving immunostaining, cells were further blocked using 3% BSA (w / v, Sigma Aldrich 9048-46-8) in the PBS buffer for 1 hour, and washed three times. Then, cells were stained with corresponding primary antibodies diluted to the above-mentioned concentrations with 3% BSA in the PBS buffer and incubated at 4°C overnight. The next day, primary antibodies were removed and cells were washed three times before the addition of secondary antibodies (Thermo Fisher A-21236, 1:1000 in 3% BSA in PBS buffer). We incubated cells in secondary antibodies for 1 hour before removing the buffer and washing them three times with PBS. For the final wash, PBS was supplemented with 40 μM DFHBI and NucBlue reagent. Cells were incubated in the buffer for 15 minutes before imaging. Microscopy FRAP and fusion experiments were performed with epifluorescence imaging using a Nikon Eclipse TI-E inverted microscope and a 60x oil immersion objective. Z- stack confocal images using a Nikon Ti microscope equipped with an NL5+ camera. Images in Figure 4D and 4K were captured using a Yokogawa CSU X1 spinning disk confocal on an inverted Zeiss stand. Hoechst (NucBlue staining) signals were detected in the UV channel (Ex 405 nm). Broccoli aptamer fluorescence was measured using the GFP channel (Ex 488 nm). Mango aptamer fluorescence was measured using the YFP channel (Ex 514 nm). Pepper aptamer, CY3, and mCherry fluorescence was detected using the RFP channel (Ex 561 nm). Finally, Alexa Fluor™ 647-labeled secondary antibody fluorescence was detected using the 647 nm channel (Ex 647 nm). Image processing Measuring the volume and number of condensates in the cytoplasm and in the nucleus Cells expressing condensates were cropped into small regions of interest (ROI). At least 50 expressing cells were selected for each nanostar structure. For each ROI, condensates were segmented using Labkit, a machine learning-based plugin for ImageJ, by manually providing labels followed by a training step9. The plugin generates masks where condensates in the nucleus and in the cytoplasm are classified into different groups, manifested as distinct pixel intensities. The masks were transformed into Imaris files (Imaris, Oxford Instruments), and reconstructed into 3D surfaces using Imaris. Condensates in the nucleus and in the cytoplasm generated separate surfaces. Any surface object with a volume smaller than 0.015 μm3(~ two voxels) was considered noise and excluded from our dataset. Adjacent surfaces connected to one object by the software algorithm were manually eliminated. Individual condensates were assigned to different cells manually by comparing reconstruction results and cropped ROIs. Surface volumes were exported for each ROI and pooled into one dataset for each nanostar variant. Calculating nuclear volume with or without nanostar expression Nuclear volumes were calculated using the surface tool in Imaris (Oxford Instruments). Before processing, we prepared a cropped ROI when large areas in the field of view lack condensate-expressing cells. Surfaces were generated using the machine learning segmentation tool. A general training algorithm was saved, and the same training classifier was applied to each image. Once machine learning segmentation is implemented, a watershed algorithm of 8 µm is applied to split touching nuclei. We removed surfaces with a high average intensity (likely apoptotic), near the border of the field of view, and volumes smaller than 450 µm (likely segmentation noise). The thresholds for intensity and volume cutoff were adjusted slightly for each image. Manual cleaning of the dataset was done by removing nuclei cut in the z-axis, apoptotic nuclei, and by splitting any touching nuclei. We removed nuclei stacked on top of each other, a frequent event in HEK293T cells. Touching nuclei were removed when Imaris failed to split touching nuclei due to errors in the cut surface. Multiple touching nuclei, or nuclei with a border difficult to distinguish, were removed as well. Cleaned datasets were further processed to identify two classes of nuclei, either expressing or non-expressing condensates.. We examined multiple fields of views in multiple samples, gathering data for at least 50 condensate-expressing cells for each nanostar structure. The volume of each nucleus was normalized by the average volume of all non-expressing nuclei for every sample group. Partitioning coefficients for linked condensates Linked condensates were detected and segmented in FIJI and Python3 referred to previous method reported by Fabrini et al7. The pipeline included preparing cropped ROI, image enhancement and mask generation, and partitioning coefficients calculation. ROI were defined from single plane confocal microscopic images with two channels (Em 488 nm and Em 561 nm, corresponding to Broccoli and Pepper channels) acquired with 60x lens and multiple FOVs per sample from three replicates. Cropped ROIs were segmented in FIJI using a macro: for each channel, denoising via a Gaussian Blue (sigma = 1.5), sliding paraboloid background subtraction (smoothing disabled, radius = 10), contrast enhancement (saturated = 0.35), and finally convert to binary masks using the Li or Otsu masks. These two segmentation methods were chosen as they tend to over- or under-segmenting. The final partitioning coefficients were calculated as the average of results generated from both methods. Fluorescence images were analyzed using a Python-based pipeline leveraging libraries such as numpy, pandas, and skimage. For each image, green and red fluorescence channels were normalized to a [0,1] scale, and masks were applied to extract mean fluorescence intensities from regions of interest. The quantities , , and , where is the average intensity of channel X within Y were computed. G stands for the Green channel and R stands for the Red channel. Due to the nature of mixing patterns, two approaches were used based on the specified mixing type: partial mixing and complete mixing. For 2 arm linkers, ratios ≥ 1:4:1 were considered as complete mixing; for 4 arm linkers, ratios ≥ 1:2:1 were considered as complete mixing due to doubled linker valency. Discussion about the effect of mixing type selection can be found at Figure 37. For partial mixing, non-overlapping regions of the masks were identified by subtracting their intersection (nonoverlapping_mask). For complete mixing, the union of the masks was used (union_mask). Mean fluorescence intensities from these regions for both channels were used to compute overlap strengths, represented as J-values, which quantify the ratio of cross-masked to self-masked intensities. was defined as and was defined as . J-values were calculated separately for by Li and Otsu methods, with the final values averaged between the two. Fluorescence recovery after photobleaching (FRAP) FRAP experiments were performed using a Nikon Eclipse TI-E inverted microscope with a temperature control unit. Temperatures were maintained at 37 °C for all experiments. For in vitro experiments, RNA strands were transcribed at 37°C using 7.5% (v / v) T7 polymerase from the AmpliScribe T7-Flash transcription kit (ASF3507, Biosearch Technologies), and a customized transcription buffer: 40mM Tris-HCl, 10 mM NaCl, 30 mM MgCl2, 2 mM spermidine, 7.5 mM each NTP, 10 mM DTT. We supplied 1% CY3-labeled UTP (ENZ-42505). After three hours, transcription samples were diluted 10 times using our customized transcription buffer, to reduce background fluorescence caused by excess CY3-UTP. Samples for FRAPing Broccoli aptamer were diluted in transcription buffer with DFHBI to a final concentration of 40 μM. In vitro samples were loaded into a house-made chamber and sealed with epoxy (Gorilla, 5- minute set) for imaging. For in vivo experiments, cells were stained using the protocol described in section 1.7.1. Condensates were bleached with a 488 nm laser for 200 ms. For in vitro samples, imaging was captured once before bleaching and every 5 seconds for 10 min after bleaching. For in vivo samples, imaging was captured once before bleaching and every 200 ms for 2 min after bleaching. Images were analyzed by extracting time-dependent average intensities within the bleached area and unbleached area. Normalization was performed to correct photobleaching caused by repetitive imaging using the equation below: where I denotes the mean pixel intensity in the bleached or unbleached area, t denotes the time point, max denotes the highest pixel intensity within the area among all time points. FRAP plots report the mean ± error bar from N=3 (one region of interest from one replicate was quantified, total three replicates). Time-dependent coalescence analysis We monitored condensate fusion events using a Nikon Eclipse TI-E inverted microscope with a temperature control unit. The temperature was maintained at 37 °C for all experiments. For in vitro experiments, RNA strands were transcribed, labeled with 1% CY3-UTP, diluted 10 times with transcription buffer, and sealed in a chamber with epoxy, following the same protocol as FRAP experiments. Samples were imaged every 5 minutes for the first 4 hours, then every 20 minutes until 10 hours. For in vivo fusion experiments, we imaged cells (in media supplemented with 40 μM DFHBI and 2 drops / mL NucBlue) every 5 minutes for 40 minutes. Data processing was performed using a script in Python3 described in our previous work1. To summarize, each fusion event was identified manually and segmented using Otsu thresholding. The binary mask was then labeled to extract the centroid position, major and minor axis lengths, and orientation. Best-fit-ellipses were generated based on the extracted data. The aspect ratio was calculated as the major-to-minor axes ratio and used for curve fitting. The time constant was calculated by fitting the Aspect Ratio vs time profiles with an exponential decay with the formula . Flow cytometry Flow cytometry experiments were performed using a BD FACSAria flow cytometer. Cells were transfected in 24 well plates, as described in section 1.6. We changed the media 24 hours after transfection, and collected cells 48 hours after transfection. For collection, we aspirated media, washed with PBS, and trypsinized the cells. After trypsinization, cells were resuspended in PBS supplemented with 10% FBS and dyes and filtered through the 40μm cell strainer (Fisher Scientific Cat.# 22363547) for flow cytometry. Hoechst was detected using a laser with Ex 405 nm and a 450 / 50 nm filter; DFHBI was detected using a laser with Ex 488 nm and a 530 / 30 nm filter. Total RNA extraction Cells were transfected in 24 well plates, as described in section 1.6. We changed the media 24 hours after transfection, and collected cells 48 hours after transfection. For collection, we aspirated media, washed with PBS, and trypsinized the cells. After trypsinization, cells were resuspended in PBS, lysed, and RNA was purified using the Monarch® Total RNA Miniprep Kit (NEB T2010S). RNA concentration was estimated using a Nanodrop 2000c by measuring absorption at 260 nm. PAGE gel electrophoresis Gel premix was prepared by adding 42 g of urea to nanopure water, the mixture was then heated until the urea completely dissolved. This mixture was allowed to cool to room temperature, and then a 40% (v / v) 19:1 acrylamide / bis-acrylamide solution was added in the appropriate volume for the desired percentage (final volume 100 mL). 8 mL of pre-mix was added in appropriate ratios with TBE and nanopure water, ammonium persulfate (APS), and tetramethylethylenediamine (TEMED) to start polymerization. Gels were cast in 8 × 8 cm, 1 mm thick disposable mini gel cassettes (Thermo Scientific, #NC2010) and allowed to polymerize for 30 minutes before electrophoresis. After curing, the gel was pre-run in a 1X TBE buffer for 30 minutes. Wells were washed carefully to remove excessive urea. Samples and low-range ssRNA ladder (NEB, N0364S) were prepared by mixing individual strands with denaturing RNA loading dye (NEB, B0363S), then heated at 70 °C for 10 minutes and immediately placed on ice. Due to the low expression of exogenous RNA in mammalian cells, 5 μg of total RNA extraction was loaded onto each well. Gels were run at room temperature at 100 V in 1X TBE unless otherwise noted. After electrophoresis, the gels were washed three times, each for 5 minutes, with nanopure water, then stained with DFHBI-1T staining buffer (10μM DFHBI-1T, 40mM HEPES, 100mM KCl, and 1mM MgCl2) for 15 minutes. After staining, gels were imaged using the Bio-Rad Gel Imaging Systems. Then, gels were washed three times, each for 5 minutes again to remove the DFHBI- 1T and stained in 1xSYBR Gold Nucleic Acid Gel Stain for 15 minutes and imaged again. Note 1 NUPACK script for 20nt-3WT stem sequence design. A similar script was used for the 10 nt long stem designs. --------------------------------------------------------------------- material = rna1999 temperature = 37 trials = 10 structure NS= U1 D20 U9 U2 D20 U9 U2 D20 U9 domain arm1= N20AAGCGCGCAN20AA domain arm2= N20AAGCGCGCAN20AA domain arm3= N20AAGCGCGCAN20 NS.seq = arm1 arm2 arm3 prevent = AAAA, CCCC, GGGG, UUUU, KKKKKK, MMMMMM, RRRRRR, SSSSSS, WWWWWW, YYYYYY Note 2 NUPACK Script for 15nt-3A-Br stem sequence optimization —--------------------------------------------------- material = rna1999 temperature = 37 trials = 10 # DU+ notation structure NS= U41 D6 (U12 D6 (U9) U17) U2 U39 # Broccoli aptamer sequence underlined domain arm1= GCGAGAGCGCUGCCCAAUCGCGAAGGGCAGCGCUCUCGCAA (SEQ ID NO: 1) domain arm2= N6ACGGUCGGGUCCN6AAUCGCGAAN6GUCGAGUAGAGUGUGGGN6AA (SEQ ID NO: 2) domain arm3= GCGUUCACACUGACCAAUCGCGAAGGUCAGUGUGAACGC (SEQ ID NO: 3) NS.seq = arm1 arm2 arm3 prevent = AAAA, CCCC, GGGG, UUUU, KKKKKK, MMMMMM, RRRRRR, SSSSSS, WWWWWW, YYYYYY Tables Caption Sequence ● Sequences GAGGGC-AUACUA RNA sequence com rise the U6+27 romoter GAGGGCCUAUUUCCCAUGAUUCCUUCAUAU G A A U U U A G C C U A C G T A A TATTAGTACAAAATACGTGACGTAGAAAGTA ATAATTTCTTGGGTAGTTTGCAGTTTTAAAAT C T A A T A G C T Kissing loop nomenclature Sequence D 5’ - A CUGUAG AA - 3’ References in Example Above 1. Stewart, J. M. et al. Modular RNA motifs for orthogonal phase separated compartments. bioRxiv 2023.10.06.561123 (2023) doi:10.1101 / 2023.10.06.561123. 2. Zadeh, J. N. et al. NUPACK: Analysis and design of nucleic acid systems. J. Comput. Chem.32, 170–173 (2011). 3. Filonov, G. S., Moon, J. D., Svensen, N. & Jaffrey, S. R. Broccoli: Rapid Selection of an RNA Mimic of Green Fluorescent Protein by Fluorescence-Based Selection and Directed Evolution. (2014) doi:10.1021 / ja508478x. 4. Chen, X. et al. Visualizing RNA dynamics in live cells with bright and stable fluorescent RNAs. Nat. Biotechnol.37, 1287–1293 (2019). 5. Cawte, A. D., Unrau, P. J. & Rueda, D. S. Live cell imaging of single RNA molecules with fluorogenic Mango II arrays. Nat. Commun.11, 1–11 (2020). 6. RNA Recognition by the MS2 Phage Coat Protein. Semin. Virol. 8, 176–185 (1997). 7. Fabrini, G. et al. Co-transcriptional production of programmable RNA condensates and synthetic organelles. bioRxiv 2023.10.06.561174 (2024) doi:10.1101 / 2023.10.06.561174. 8. Clever, J. L., Wong, M. L. & Parslow, T. G. Requirements for kissing-loop- mediated dimerization of human immunodeficiency virus RNA. J. Virol.70, 5902– 5908 (1996). 9. Arzt, M. et al. LABKIT: Labeling and segmentation toolkit for big image data. Front. Comput. Sci.4, (2022). 10. Litke, J. L. & Jaffrey, S. R. Highly efficient expression of circular RNA aptamers in cells using autocatalytic transcripts. Nat. Biotechnol.37, 667–675 (2019). All publications mentioned herein (e.g. those listed above) are incorporated by reference to disclose and describe aspects, methods and / or materials in connection with the cited publications. Many of the techniques and procedures described or referenced herein are well understood and commonly employed by those skilled in the art. Unless otherwise defined, all terms of art, notations and other scientific terms or terminology used herein are intended to have the meanings commonly understood by those of skill in the art to which this invention pertains. In some cases, terms with commonly understood meanings are defined herein for clarity and / or for ready reference, and the inclusion of such definitions herein should not necessarily be construed to represent a substantial difference over what is generally understood in the art.

Claims

CLAIMS 1. A method of making a synthetic RNA condensate molecule within a living cell, the method comprising: transducing the cell with a vector comprising a DNA sequence selected to encode a RNA condensate polynucleotide sequence; allowing the cell to transcribe the DNA sequence so that the synthetic RNA condensate molecule is formed, wherein the vector and the DNA sequence is selected to form a synthetic RNA condensate molecule that: is circular; comprises 100-300 nucleotides; comprises complementary sequences that allow the RNA polynucleotide to self- assemble so as to form at least 3 arm structures comprising a stem and a loop, wherein: the stem in said arm structures comprise at least 10 paired nucleotides; the loops in said arm structures comprise at least 6 nucleotides; and a first loop in the synthetic RNA condensate comprises at least 6 nucleotides selected to be complementary to at least 6 nucleotides in a second loop in a synthetic RNA condensate such that the at least 6 nucleotides in the first loop of a first synthetic RNA condensate molecule and the at least 6 molecules in a second loop of a second synthetic RNA condensate molecule couple together so that the synthetic RNA condensate molecules aggregate to form a RNA condensate in the cell.

2. The method of claim 1, wherein: (a) the RNA condensate molecule forms not more than 3 arm structures; the kissing loops form not more than 2 or 3 G:C bonds, and the arms comprise not more than 10, 1112 or 13 paired nucleotides such that condensates form in the cytoplasm of the cell; or (b) the RNA condensate molecule forms at least 4 arm structures; the kissing loops form more than 2-3 G:C bonds, and the arms comprise more than 11, 12 or 13 paired nucleotides such that condensates form in the nucleus of the cell.

3. The method of claim 1, wherein: the cell is a mammalian cell; the synthetic RNA condensate molecule comprises a RNA aptamer motif sequence; and / or the synthetic RNA condensate molecule comprises a polynucleotide sequence to which a RNA aptamer couples via base pairing.

3. The method of claim 3, wherein the RNA aptamer motif sequence is disposed in a loop.

4. The method of claim 1, wherein the at least 6 nucleotides in the first loop comprise a HIV-type Kissing-Loop motif.

5. The method of claim 1, wherein: the first loop and the second loop are disposed in a single synthetic RNA condensate; and the first loop and the second loop in the single synthetic RNA condensate cannot couple together.

6. The method of claim 2, wherein the mammalian cell is a human cell.

7. The method of claim 1, wherein the method forms a plurality of synthetic RNA condensates by transducing the cell with a plurality of vectors selected to encode a plurality of RNA condensation sequences so that a plurality of synthetic RNA condensates are formed.

8. The method of claim 1, wherein the wherein the at least 6 nucleotides in the first loop are orthogonal.

9. The method of claim 1, wherein the RNA condensate is coupled to a protein and / or a nucleic acid.

10. A composition of matter comprising a living cell comprising a RNA condensate made by the method of claim 1.

11. A composition of matter comprising a living cell comprising a plurality of RNA condensates made by the method of claim 7.

12. The composition of claim 11, wherein the at least 6 nucleotides in the first loop are orthogonal with non- interacting kissing loops so as to form distinct condensates that do not mix.

13. A RNA condensate formed by a plurality of aggregated synthetic RNA condensates made by the method of claim 1.

14. The RNA condensate of claim 13, wherein the cell is transduced with a plurality of different vectors comprising a plurality of different DNA molecules so as to form a plurality of RNA condensates that do not aggregate.

15. A method of making a RNA condensate comprising: (a) transducing a mammalian cell with a vector comprising a plurality of synthetic RNA condensates selected to: be circular; comprise 100-300 nucleotides; comprise complementary sequences that allow the synthetic RNA condensates to self-assemble so as to form at least 3 arm structures comprising a stem and a loop, wherein:the stem in said arm structures comprise at least 10 paired nucleotides; the loops in said arm structures comprise at least 6 nucleotides; at least a first loop in said arm structures comprises at least 6 nucleotides selected to be complementary to at least 6 nucleotides in at least a second loop such that the at least 6 nucleotides in a first loop of a first synthetic RNA condensate and the at least 6 molecules in a second loop of a second synthetic RNA condensate couple together so that the synthetic RNA condensates aggregate to form a RNA condensate; wherein the plurality of synthetic RNA condensates are combined under conditions selected to facilitate condensation; and (b) allowing the plurality of synthetic RNA condensates to aggregate so that the RNA condensate is made in the cell.

16. The method of claim 15, wherein the plurality of RNA condensates comprise a homogeneous population of polynucleotides.

17. The method of claim 15, wherein the plurality of synthetic RNA condensates comprise a heterogeneous population of polynucleotides.

18. The method of claim 15, wherein the plurality of synthetic RNA condensates are disposed within a human cell.

19. The method of claim 15, wherein the plurality of synthetic RNA condensates are selected to include arm and loop structures that facilitate condensation formation in the nucleus of the cell.

20. The method of claim 15, wherein the plurality of synthetic RNA condensates are further coupled to proteins and / or nucleic acids.

Images