Cellular Function of a Biomolecular Condensate Is Determined by Its Ultrastructure

Abstract

Biomolecular condensates play key roles in the spatiotemporal regulation of cellular processes. Yet, the relationship between atomic features and condensate function remains poorly understood. We studied this relationship using the polar organizing protein Z (PopZ) as a model system, revealing how its material properties and cellular function depend on its ultrastructure. We revealed PopZ's hierarchical assembly into a filamentous condensate by integrating cryo-electron tomography, biochemistry, single-molecule techniques, and molecular dynamics simulations. The helical domain drives filamentation and condensation, while the disordered domain inhibits them. Phase-dependent conformational changes prevent interfilament contacts in the dilute phase and expose client binding sites in the dense phase. These findings establish a multiscale framework that links molecular interactions and condensate ultrastructure to macroscopic material properties that drive cellular function.

Figure 1.. Visualization of PopZ condensates and filaments by cryo-electron tomography

(A) Phase-contrast imaging and cryo-ET of in vitro PopZ condensates. (top left) Phase contrast image of PopZ condensates formed by adding 50 mM MgCl2 to 5 μM of purified PopZ in 10 mM sodium phosphate (pH 7.4) with 150 mM NaCl. Scale bar: 5 μm. The zoom-in box is provided for illustrative purposes. (top middle) Cryo-ET tomogram of 2 μM purified WT-PopZ in the same buffer, captured with 53,000-fold magnification. Scale bar: 100 nm. The purple boxes indicate magnified regions of the condensate. (top right) 3D rendering of segmented tomograms. (bottom) Six zoomed-in regions illustrating the flexibility of filaments in PopZ condensates. Yellow arrows highlight filamentous structures. Scale bar: 10 nm. (B) Filament length distribution fits an isodesmic model. Length distribution of over 8000 filaments from three independent condensates (see Fig. S4B). Columns represent the mean filament length across the three condensates; error bars denote standard deviation. The turquoise line indicates a fit to an isodesmic model. (C) Cryo-ET reveals filaments within PopZ condensates in Caulobacter crescentus. Representative cryo-ET images of the PopZ microdomain in ΔpopZ Caulobacter cells expressing mCherry-tagged WT-PopZ. Scale bar: 100 nm. Filaments can be seen in the zoomed-in panel. Scale bar: 10nm. Additional reconstructions and representative filament images are shown in Fig. S1A.

Figure 2.. The oligomerization domain is necessary and sufficient for filamentation and condensation

(A) Schematic of the PopZ construct. The construct includes an N-terminal 6xHIS fusion (brown) fused to PopZ from Caulobacter crescentus, which codes for a short helical N-terminal region, H1 (pink box), a 78 amino-acid intrinsically disordered region (IDR, gray curly line), and a C-term region comprising three predicted helices: H2, H3, H4 (blue boxes). Positions −20 and 132 (indicated by asterisks) were used to covalently attach fluorescent dyes for imaging and smFRET studies. (B) AlphaFold 2 model of the PopZ monomer shown for illustrative purposes. A TEV cleavage site, introduced between residues 99 and 100, is present in our PopZ construct only when specified. (C-D) TEV cleavage separates the H1-IDR region from the OD, leading the OD to condense and H1-IDR to be excluded from the condensates. (C) SDS-PAGE gel revealed with Coomassie staining. TEV protease was added to TEV-cleavable PopZ in vitro and incubated for one hour before pelleting down the dense phase (2000 g, 3 min). P is the pellet fraction, which contains the OD, SN is the supernatant, which contains the IDR. The IDR migrates as a larger species due to its stretches of densely packed negative charges. (D) Phase and fluorescence images 30 minutes after incubation with TEV protease. The formed droplets (phase) are composed of cleaved OD (OD-AF488) and exclude cleaved H1-IDR (H1-IDR-AF647). Scale bars: 5 μm (E) OD condensates are made up of filaments. (left) cryo-ET tomogram of 5 μM H1-IDR-TEV-OD cleaved for 3 minutes with TEV protease in 20 mM Tris-HCl pH 8.0, 0.5 mM EDTA, 2 mM DTT. Magnification: 53,000x. The purple box indicates a zoomed-in region of the condensate. Scale bar: 100 nm. (right) 3D rendering of filaments of the segmented tomograms.

Figure 3.. PopZ oligomerization into trimers, hexamers, and multiples of hexamers and the role of L6 in condensate formation, structure, and dynamics

(A) Mass photometry analysis of purified PopZ variants at 250 nM. (left) WT PopZ (gray) forms trimers and hexamers, while the P146A variant (magenta) is trapped in the trimeric state, and the ΔL6 variant (tan) is trapped in the hexameric state. (middle) Comparison of WT PopZ in the presence (turquoise) and absence (gray) of MgCl₂ in 10 mM sodium phosphate pH 7.4 + 150 mM NaCl, showing that MgCl₂ promotes higher-order oligomerization. (right) Relative populations of monomer, trimer, and hexamer across several PopZ variants, including WT (gray), ΔL6 (tan), P146A (magenta), Δloop (light gray), DH2 (green) and OD* (P146A-W151G-L152S, blue). Columns represent the mean of biological duplicates, and error bars represent the standard deviation. (B) Analytical ultracentrifugation of PopZ variants at approximately 100 μM in 10 mM sodium phosphate pH 7.4 + 150 mM NaCl shows that ΔL6 (tan) remains hexameric even at high concentrations, whereas P146A (magenta) and WT (gray) form higher-order oligomers, with WT assembling into the largest species. (C) ΔL6-PopZ forms condensates made up of hexamers. (left) Phase contrast image of ΔL6-PopZ condensates in solution. Scale bar: 5 μm. (middle) Cryo-ET tomogram of a ΔL6-PopZ condensate formed by adding 50 mM MgCl₂ to 50 μM of purified ΔL6-PopZ in 10 mM sodium phosphate pH 7.4 + 150 mM NaCl. Scale bars: 100 and 10 nm. (right) 3D rendering of the traced ΔL6-PopZ tomograms. (D) Phase contrast (top) and fluorescence microscopy (middle) show that ΔL6-PopZ condensates wet the surface of glass coverslips, whereas WT PopZ condensates (bottom) largely remain spherical over a period of four hours. Scale bars: 5 μm. (E) (left) Fluorescence recovery after photobleaching (FRAP) experiments showing the recovery profiles of AF647-labeled PopZ for different condensate compositions. Mean ± standard deviation is shown for two biological replicates, each averaged over at least nine condensates. The solid lines indicate fits to a simple exponential model. (right) Representative examples of droplets used for FRAP experiments. Scale bars: 5 μm. (F) Deletion of the L6 region impacts client protein (ChpT) diffusivity inside the condensate. (left) FRAP recovery curves of ChpT inside WT-PopZ condensates (gray) and ΔL6 condensates (tan). Mean ± standard deviation over two biological replicates, each averaged over at least twelve condensates. The solid lines indicate fits to a two-phase exponential model. (right) Representative examples of droplets used for FRAP experiments. Scale bars: 5 μm.

Figure 4.. The conformational equilibrium of PopZ is phase-dependent

(A) smFRET reveals that PopZ adopts extended and compacted conformations in the dilute phase. FRET efficiency distribution of the smFRET reporter labeled at C-20 and C132 shows two predominant conformations in the dilute phase (light gray): extended and compacted. The extended conformation coincides with the predicted behavior of a random coil (dark gray), whereas the compacted conformation indicates proximity between the N- and C-termini. (B) H1 drives the compacted conformation. (left) Schematic illustrating two experimental conditions. In the first condition, labeled FL-PopZ was mixed with an excess of unlabeled WT PopZ. For each labeled PopZ monomer (fluorophores indicated by green and red circles), two unlabeled PopZ monomers are shown, where the H1 of the unlabeled monomer competes with the H1 of the labeled monomer for a binding site on the OD (excess WT, top). In the second condition, labeled FL-PopZ was mixed with an excess of unlabeled ΔH1-PopZ (missing residues 1–23). In this case, only the H1 of the labeled PopZ monomer can bind the OD (excess DH1, bottom). (right) FRET efficiency plot shows that excess WT promotes the extended conformation (gray), similar to the denaturing condition (tan). In contrast, excess ΔH1-PopZ promotes the compacted conformation (blue). (C) H1 recruits ChpT to PopZ condensates. Fluorescence microscopy images show 1 nM ChpT labeled with AF488 recruited by FL-PopZ condensates (top), while ChpT remains diffuse in the presence of condensates of isolated OD (bottom). In both conditions, PopZ concentration was 5 μM. Scale bar: 5 μm. (D) ChpT does not interact with PopZ in the dilute phase. (left) FRET efficiency histograms indicate that ChpT at 5 μM (tan) only marginally increased the population of the extended conformation of PopZ (gray). (right) Sedimentation coefficient distribution from AUC for individual samples of PopZ (gray) and ChpT (magenta) as well as combined (tan). The combination did not indicate the formation of a new PopZ:ChpT species. (E) In the dense phase, PopZ adopts exclusively the extended conformation. (left) 2D histograms of FRET efficiency versus burst duration. Typical burst durations for single protein particles are shorter than 10 ms. Durations greater than 15 ms (turquoise box) were selected to isolate fluorescence bursts coming from within condensates. (right) Condensed PopZ exclusively adopts the extended conformation. FRET efficiency histogram of condensed and dispersed (same condition as in (A)) PopZ. (F) A model of phase-dependent interactions. In the dilute phase, H1 is predominantly bound to the OD, preventing ChpT (green) from interacting with H1. In the dense phase, H1 is released from the OD, enabling ChpT binding.

Figure 5.. Filaments are crucial for PopZ function in vivo

(A) Spotting dilution assay of Caulobacter cells expressing mCherry-PopZ or mCherry-ΔL6-PopZ from a plasmid in a NA1000 parB::cfp-parB background. Cells expressing mCherry-PopZ exhibit normal growth (top), whereas cells expressing mCherry-ΔL6-PopZ (bottom) do not grow. (B) Schematic illustrating the migration and capture of the ParB:parS focus (blue) by the PopZ microdomain (red). At the start of the cell cycle, ParB forms a complex with the parS centromere that is anchored to the PopZ condensate at the old pole. During chromosome replication, one copy of the ParB/parS complex remains anchored at the old pole via ParB interactions, while the other copy migrates to the new pole, associating with PopZ with assistance from a ParA gradient (not shown) to facilitate rapid migration^,. (C) The effect of PopZ mutants on DNA segregation. (top) Schematic illustrating the oligomerization pattern and localization of different mCherry-PopZ variants (red) expressed in WT Caulobacter cells expressing CFP-labeled ParB and endogenous WT PopZ (black). (middle) Representative time-lapse images of synchronized cells at three time points, starting from swarmer cells. The cells express mCherry-PopZ variants from a plasmid, while CFP-ParB (blue) replaces WT ParB to track DNA segregation. Check marks or X-marks indicate whether chromosome segregation proceeded properly, i.e., whether ParB foci were anchored and the old pole, replicated, migrated to the new pole, and captured at the new pole within expected time frames. (bottom) Kymographs displaying the cell body (black) and CFP-ParB (cyan) over time. Time progresses from the left to the right. The old pole is positioned at the bottom, while the new pole is at the top. Scale bar: 1 μm.

Figure 6.. Mechanistic depiction of PopZ oligomerization, filamentation, and condensation

(A) Schematic representation of the PopZ sequence highlighting its subdomains and their roles in assembling PopZ into oligomers, filaments, and condensates. (B) Illustration of the hierarchical assembly of PopZ from monomers to condensates. (i–iii) PopZ oligomers in the dilute phase (dashed gray circles). (i) AlphaFold 2 model of the PopZ monomer, with the N-terminal domain—including the client-binding helix (H1)—in pink, the IDR in gray, the H2 helical region in green, helices three and four (H3–H4) in blue, and the C-terminal six amino acids in red. (ii) AlphaFold model of the PopZ trimer, showcasing the formation of a triple coiled coil via the H3–H4 region (blue), with H1 helices (pink) interacting proximally to these coiled regions. Conformational flexibility is illustrated by depicting two H1 helices close to H3–H4 and one further away. (iii) AlphaFold hexamer model displaying interactions between two trimers (colored purple and green) mediated by inter-trimer H2 region interactions. The IDRs extend from the center of the hexamer, creating a negatively charged cloud around it. Four of the six H1 helices are shown proximal to their corresponding H3–H4 regions. (iv-vi) PopZ filament structure in the dense phase (dashed brown circles). (iv) Fitting of three hexamers (purple and green) into a density extracted from the condensate tomogram (yellow), demonstrating a tail-to-tail organization between hexamers with IDRs extending from the center of each hexamer. In this configuration, the IDRs are positioned away from H3–H4 and are involved in binding client molecules. (v) A larger segment of the tomogram (yellow) with fitted hexamers, illustrating variability in filament lengths. Unused portions of the tomogram are outlined in black. Here, only the OD of each hexamer is displayed. (vi) Portion of a tomogram of the PopZ condensate (from Fig. 1) revealing the filamentous ultrastructure of the condensate. A scale bar is shown for each circle. (C) Phase-dependent interoligomer contacts. (left) Trimer and hexamer interactions in the dilute phase, with the gray ellipse representing a cloud of negative charge due to negatively charged residues and the flexibility of the IDR. (right) In the dense phase, conformational changes opening the H1 helices remove the IDR cloud around hexamers, enabling interfilament contacts.