Intracellular production of hydrogels and synthetic RNA granules by multivalent molecular interactions

Abstract

Some protein components of intracellular non-membrane-bound entities, such as RNA granules, are known to form hydrogels in vitro. The physico-chemical properties and functional role of these intracellular hydrogels are difficult to study, primarily due to technical challenges in probing these materials in situ. Here, we present iPOLYMER, a strategy for a rapid induction of protein-based hydrogels inside living cells that explores the chemically inducible dimerization paradigm. Biochemical and biophysical characterizations aided by computational modelling show that the polymer network formed in the cytosol resembles a physiological hydrogel-like entity that acts as a size-dependent molecular sieve. We functionalize these polymers with RNA-binding motifs that sequester polyadenine-containing nucleotides to synthetically mimic RNA granules. These results show that iPOLYMER can be used to synthetically reconstitute the nucleation of biologically functional entities, including RNA granules in intact cells.

Figure 1. Schematic illustration of iPOLYMER

(a) Rapamycin induces rapid, stable and specific binding between FKBP and FRB molecules. (b) Two series of proteins, YF_xN and CR_xM, were engineered to track their expression in cells: a yellow fluorescent protein (YFP) on up to five tandem repeats of an FKBP domain and a cyan fluorescent protein (CFP) on up to five tandem repeats on an FRB, spaced by 12 amino acid linker sequences. Mixing YF_x5 and CR_x5 (left) with rapamycin is expected to induce the formation of a hydrogel network (right). YF_XN and CR_XM contain N-repeats of FKBP and M-repeats of FRB with the same linkers, respectively.

Figure 2. In silico implementation of iPOLYMER demonstrates its feasibility for hydrogel network synthesis

(a) Four reversible reactions between monomeric FKBP, FRB and rapamycin molecules modeled in our simulations. Each binding unit in the tandem repeats of FKBP or FRB can undergo the four reactions in the presence of rapamycin. (b) Estimated probabilities that iPOLYMER will produce aggregates of a threshold size of 100 or larger for different valence numbers of the FKBP and FRB molecules. An aggregate of size 100, as defined in the Supplementary Methods, comprises 25% of the total number of FKBP and FRB molecules initially present in the simulated system. The sharp increase in the probability values indicates that efficient polymerization can be achieved when the individual valence numbers of FKBP and FRB are at least three, with the total valence number of FKBP and FRB molecules being at least six. (c) Estimated probabilities that iPOLYMER will produce aggregates of a threshold size of 100 or larger for different valence numbers of the FKBP and the FRB molecules and different numbers of rapamycin molecules, determined for each simulation by multiplying a base number of rapamycin molecules with the common valency of FKBP and FRB [e.g., base rapa # (160) x valency (4) = rapa # (640)] in order to scale the effect of peptide valency on the number of binding sites against rapamycin. The observed sharp decrease in the probability values indicates that efficient polymerization requires a sufficient concentration of rapamycin. This implies that, in addition to the valence numbers of FKBP and FRB, the concentration of the dimerizing agent is expected to directly affect phase transition.

Figure 3. iPOLYMER puncta formation in living cells

(a) Time-lapse imaging of fluorescent puncta formation in COS-7 cells at indicated times relative to the addition of rapamycin. Scale bars: 10 μm. Punctate structures enriched with CFP, YFP, and FRET signals start to emerge within 5 min after 333 nM rapamycin addition, while DMSO treated cells demonstrate lack of puncta formation. The FRET ratio fold-change was significantly greater at puncta compared to that in the cytosol, which in turn was significantly greater than in DMSO treated cells. (b) Frequency of iPOLYMER puncta formation plotted against valence numbers in FKBP and FRB constructs. F_N represents valence number of cytoYF_xN, whereas R_M represents cytoCR_xM. (c) Probability of iPOLYMER formation plotted against the total valance number N+M. In order to avoid bias, the combinations (N=1, M>1) and (N>1,M=1) were excluded from the data. Note that peptides with single valency should not lead to network formation, confirmed by the rare puncta formation in (b) for F1 or R1.

Figure 4. Biophysical analysis of iPOLYMER in living cells

(a) Top panel : Confocal fluorescence images of representative regions of cells expressing YF_x5 and CR _x5 subjected to FRAP analysis. Photobleaching was conducted under the following conditions: cytosolic region before administrating 333 nM rapamycin (cyan circle), cytosol outside the puncta (green circle), and inside the puncta (red circle). Scale bar: 10 μm. Lower panel: Fluorescent intensity transients before and after photobleaching. The colors correspond to those of the representative regions in the top panel. (b) iPOLYMER puncta allowed protein tracers to pass through. Top panel: Fluorescence intensity profile of mCherry in the cytosol in a line-scan FRAP experiment is shown in an x-t presentation. The fluorescence was photobleached at a single spot located in the middle of the puncta, as indicated by an arrow. The arrowhead indicates the time of bleaching. Middle panel: Representative normalized fluorescent intensity profiles shown for each experimental condition as a pseudo-colored image. Fluorescence recovery kinetics quantified from the data for mCherry and mCherry-β-galactosidase signals inside (puncta) or outside (outside) the puncta. Lower panel: The recovery kinetics were quantified by two parameters; mobile fraction (left graph) and half-recovery time (right graph). The values of these parameters were not significantly different inside and outside the puncta (p-value > 0.05, error bars: S.E.M.) for both tracer molecules. (c) Representative images of a mCherry-TGN38 labeled vesicle in contact with iPOLYMER punctum, indicated by the white arrow (left panel). The red crosses in the middle and right panels mark the position of the vesicle during a period of 150 s at 5 s intervals when colliding with the punctum (middle panel) and in the cytosolic region free of visible puncta (right panel). Scale bars: 5 μm.

Figure 5. Correlative EM analysis of iPOLYMER puncta in living cells in comparison with stress granules

Transmitted EM (TEM) images of iPOLYMER puncta were obtained in COS-7 cells by correlating CFP-FRB_x5 fluorescence image (top left panel (fluorescence)) with TEM image (top middle panel, shown as overlaid with correlated with fluorescence image (fluorescence + TEM), scale bar: 10 μm). The cells with apparent iPOLYMER puncta induced by 333nM rapamycin administraton were identified before EM imaging by referring to the grid pattern in the bright field image (bottom left panel (bright field), scale bar: 100 μm) High-magnification image of an iPOLYMER punctum is shown in the top right panel (iPOLYMER, scale bar: 500 nm), compared to the TEM image of an actual stress granule induced by 30 min incubation with 0.5 mM arsenite (bottom right panel (stress granule), scale bar: 500 nm). Negative control EM image of the cytosol without stress granule induction is also shown in the bottom middle panel (negative control). The iPOLYMER punctum exhibited an electron-dense granulo-fibrillar structure (black arrow in the top right panel) without any membranes surrounding it, resembling the actual stress granule (black arrows in the bottom right panel).

Figure 6. In vitro characterization of iPOLYMER

(a) Mixing 100 μM YF_x5 and 100 μM CR_x5 (left) with 500 μM rapamycin in a 1.5 mL tube instantly led to a turbid appearance (right). Size of grid: 0.5 cm. (b) Top panel: Fluorescent, FRET, and bright field (BF) microscopic images of iPOLYMER aggregates formed in vitro. Scale bar: 20 μm. Lower panel: Mixing 100 μM YF_x5 with 100 μM CR_x5 and DMSO did not form any aggregates (without rapamycin), and the same was true when mixing 100 μM YF_x5 with 500 μM rapamycin (without CR_x5). (c) Aggregates were collected by centrifuge and observed under a dissection microscope. Colored pellet was observed after centrifugation and removal of the supernatant (top panel). The pellet was isolated on a coverslip for further observation and experimentation (lower panel). The fragments were translucent with clearly defined shapes (before), and the aggregates retained their three-dimensional shape and translucent appearance after the removal, demonstrating the identity as a hydrogel. The hydrogel was mechanically deformed with a micropipette tip (deformed), and was confirmed to regain its original shape (after), which was almost indistinguishable from that before applying the deformation (before). (d) Pore size evaluation of the iPOLYMER hydrogel in vitro. Hydrogels collected by centrifuge were re-suspended with fluorescent tracers with distinct diameters, and observed under a confocal fluorescent microscope. While a D-Cy5 tracer with 4.3 nm diameter penetrated into the hydrogel (top panel), 6nm-diameter Q-dot (middle panel) and 20 nm-diameter fluorescent beads (lower panel) were clearly excluded from the hydrogel. (e) The ratio of the tracer fluorescence intensity inside the hydrogel to that of outside the gel for each tracer molecule. Error bars, S.E.M. ***, p<0.01. The observed difference of the ratios associated with D-Cy5 (4.3 nm) and Q-dot (6 nm) suggest that the hydrogel functions as a molecular sieve with an effective pore size of 4.3-6 nm.

Figure 7. Reconstituting RNA granules by using iPOLYMER as scaffold

(a) Schematic illustration of the RRM-CR_x5 construct used in RNA granule reconstitution. Three RRM domains from TIA-1 were fused to CR_x5. (b) Immuno-staining images of COS-7 cells expressing RRM-CR_x5 and cytoYF_x5 treated with rapamycin to form iPOLYMER puncta. Scale bar: 10 μm. The line scan plot from A to B in the enlarged image confirms co-localization of endogenous PABP with the functionalized iPOLYMER puncta. (c) Immuno-staining of the RRM-functionalized iPOLYMER with the universal stress granule markers G3BP-1, eIF3b, and eIF4G and corresponding line scan plots from A to B shown in each overlay image. As a negative control, ribosomal P antigen, which does not accumulate in the stress granules, was also stained. RRM-functionalized synthetic analogue of RNA granules accumulated all the three stress granule markers (7/17 cells showed accumulation of G3BP-1, 12/29 cells of eIF3b, and 8/15 cells of eIF4G), while ribosomal P antigen accumulation was not observed (0/27 cells).

Figure 8. Light-inducible iPOLYMER-LI provides reversibility and spatial control over puncta formation

(a) Designs of light-inducible iPOLYMER-LI constructs. The two peptides, SSPB and iLID, bind to each other upon blue light (488 nm) stimulation in a reversible manner (top panel). YFP-SSPB_x6 and mCherry-iLID_x6 contain six repeats of iLID and SSPB, respectively, spaced by nine amino acid linker sequences (lower panel). Due to design principles similar to YF_x5 and CR_x5, the two peptide chains are expected to form a polymer network upon light irradiation. However, unlike YF_x5 and CR_x5, the network formation is reversible. (b) Reversible puncta formation by YFP-SSPB_x6 and mCherry-iLID_x6. The cell was irradiated with 488 nm laser before each frame during the stimulus. The blue rectangle labels images observed under stimulus. Light-induced formation of the protein aggregates was readily observed within 15 min (middle panel and magnified detail). By ceasing stimulation, the aggregates were dispersed within 7.5 min (lower panel), demonstrating the reversible nature of the light-inducible version of iPOLYMER, iPOLYMER-LI. (c) Aggregates were formed repetitively at distinct locations within the same cell using the iPOLYMER-LI (left panel). The blue rectangles label images obtained under stimulation. Magnified views of two stimulated regions of interest, ROI1 and ROI2, are also shown highlighted by colors that correspond to those in the right top panel. Fluorescence intensities are shown in the right middle and right lower panels for YFP and mCherry respectively, in which the stimulation timing is highlighted by blue. During the first stimulation, only ROI1 was illuminated, whereas only ROI2 was illuminated during the second stimulation. The apparent overshoot in fluorescence intensity right after each stimulation is probably an artifact, since the fluorescence intensities at the puncta were often saturated during the stimulation. Black arrows in the plots indicate the timings of the images shown in the left. Taken together, these results show that fluorescent puncta were dynamically formed and dispersed locally, demonstrating both the reversibility and spatio-temporal control over this process.

Figure 9. RNA granule reconstitution by light-inducible iPOLYMER-LI

(a) Immuno-staining of the RRM-functionalized iPOLYMER-LI with the universal stress granule markers PABP-1, G3BP-1, eIF3b, and eIF4G and corresponding line scan plots from A to B shown in each overlay image. Immunostaining results are shown in blue in overlay images. As a negative control, ribosomal P antigen, which does not accumulate in the stress granules, was also stained. RRM-functionalized iPOLYMER-LI puncta (YFP-SSPB_x6 and mCherry-iLID_x6) accumulated all the three stress granule markers (33/65 cells showed accumulation of PABP-1, 25/40 cells of G3BP-1, 23/51 cells of eIF3b, and 52/75 cells of eIF4G), while ribosomal P antigen accumulation was not observed (0/75 cells). Scale bars: 10 μm. (b) Comparison between iPOLYMER- and iPOLYMER-LI-based stress granule analogues and the actual stress granules. Stress granule analogues were similar to its physiological counterparts in the specific recruitment of universal stress granule markers. Rapamycin-induced analogues differ from physiological stress granules in reversibility of the formation process, while iPOLYMER-LI-based analogues have overcome the discrepancy.