Liquid-Liquid Phase Separation of the DEAD-Box Cyanobacterial RNA Helicase Redox (CrhR) into Dynamic Membraneless Organelles in Synechocystis sp. Strain PCC 6803

ABSTRACT

Compartmentalization of macromolecules into discrete non-lipid-bound bodies by liquid-liquid phase separation (LLPS) is a well-characterized regulatory mechanism frequently associated with the cellular stress response in eukaryotes. In contrast, the formation and importance of similar complexes is just becoming evident in bacteria. Here, we identify LLPS as the mechanism by which the DEAD-box RNA helicase, cyanobacterial RNA helicase redox (CrhR), compartmentalizes into dynamic membraneless organelles in a temporal and spatial manner in response to abiotic stress in the cyanobacterium Synechocystis sp. strain PCC 6803. Stress conditions induced CrhR to form a single crescent localized exterior to the thylakoid membrane, indicating that this region is a crucial domain in the cyanobacterial stress response. These crescents rapidly dissipate upon alleviation of the stress conditions. Furthermore, CrhR aggregation was mediated by LLPS in an RNA-dependent reaction. We propose that dynamic CrhR condensation performs crucial roles in RNA metabolism, enabling rapid adaptation of the photosynthetic apparatus to environmental stresses. These results expand our understanding of the role that functional compartmentalization of RNA helicases and thus RNA processing in membraneless organelles by LLPS-mediated protein condensation performs in the bacterial response to environmental stress.

IMPORTANCE Oxygen-evolving photosynthetic cyanobacteria evolved ~3 billion years ago, performing fundamental roles in the biogeochemical evolution of the early Earth and continue to perform fundamental roles in nutrient cycling and primary productivity today. The phylum consists of diverse species that flourish in heterogeneous environments. A prime driver for survival is the ability to alter photosynthetic performance in response to the shifting environmental conditions these organisms continuously encounter. This study demonstrated that diverse abiotic stresses elicit dramatic changes in localization and structural organization of the RNA helicase CrhR associated with the photosynthetic thylakoid membrane. These dynamic changes, mediated by a liquid-liquid phase separation (LLPS)-mediated mechanism, reveal a novel mechanism by which cyanobacteria can compartmentalize the activity of ribonucleoprotein complexes in membraneless organelles. The results have significant consequences for understanding bacterial adaptation and survival in response to changing environmental conditions.

INTRODUCTION

The formation of subcellular compartments in which specific metabolic processes occur is crucial in all cells. Although formation of lipid-membrane-bound organelles is the primary compartmentalization mechanism in eukaryotic cells, liquid-liquid phase separation (LLPS) also results in formation of a diverse range of membraneless organelles (1–3). These assemblies concentrate associated biomolecules, allowing for enhanced spatial and temporal reactive efficiency or protection of protein or RNA components from degradation (3). LLPS-mediated polypeptide condensation into membraneless organelles is enhanced by intrinsically disordered, low-complexity domains (LCDs) that spontaneously aggregate into lipid-like droplets, a process aided by protein oligomerization (4, 5). Physiologically, LLPS frequently occurs in response to stress conditions, enhancing cell survival, and importantly, LLPS is a dynamic process, being reversible upon stress alleviation (3, 6–8). Functionally, stress conditions stabilize LLPS aggregate structure, thereby compartmentalizing and concentrating specific macromolecules and allowing for alteration of ribonucleoprotein (RNP) components and potential exchange with the surrounding environment (6).

Elucidation of how bacteria compartmentalize biological functions in the absence of lipid-bound organelles is a crucial question. Protein aggregates that appear as foci are frequently observed, commonly associated with membranes, and involve housekeeping functions that are not stress associated (9–11). Limited reports have recently provided evidence of LLPS formation of membraneless organelles in bacteria that are also associated with housekeeping processes (1, 8, 12–16). These compartments, referred to as aggresomes, are highly dynamic, RNP assemblages that provide a rapidly reversible mechanism to aid in the cellular response to stress conditions (8).

Since membraneless organelles frequently contain RNA and protein, it is not unexpected that remodelers of RNP structure, DEAD (Asp-Glu-Ala-Asp)-box RNA helicases, are prominent components of a range of cytoplasmic processing bodies in eukaryotes (17). In contrast, direct evidence supporting LLPS-mediated formation of protein condensates involving DEAD-box RNA helicases in prokaryotes is limited. Hondele et al. (13) showed that only the three Escherichia coli DEAD-box RNA helicases that contain LCDs form phase-separated droplets in vitro and foci in vivo, a process requiring ATP hydrolysis. Evidence was also provided by the demonstration that RNase E assembles LLPS condensates to form RNA degradosomes that localize within the nucleoid-filled cytoplasm in Caulobacter crescentus (12). RNase E assembly into these condensates is dependent on RNA and an intrinsically disordered C-terminal domain. In C. crescentus, the DEAD-box RNA helicase RhlB is known to interact with RNase E, thus indirectly associating RhlB with degradosome LLPS condensates (18).

DEAD-box RNA helicases belonging to superfamily 2 are ubiquitously encoded in all domains of life, where they perform roles in all aspects of RNA metabolism (19). Regulation of gene expression occurs via ATP-driven rearrangement of RNA secondary structure, resulting in alteration of RNP complexes (19). As a component of this regulation, helicases function in translation, ribosome biogenesis, and RNA degradation, frequently associated with stress responses (19–25). In addition, DEAD-box RNA helicases can also function as chaperones, binding and clamping RNA to avoid either degradation or translation, and participate in RNA shuttling between cellular compartments or specific locations in eukaryotic systems (19, 26, 27).

Cyanobacterial RNA helicase redox (CrhR), the only DEAD-box RNA helicase encoded in the model cyanobacterium Synechocystis sp. strain PCC 6803 (hereafter Synechocystis), has been extensively characterized as a component of the stress adaptation mechanism. CrhR stress regulation is complex, involving a series of auto-regulatory mechanisms (28, 29). Expression is induced by a range of abiotic stresses at 30°C and by low temperature at 20°C (28, 30–32). Stress alleviation represses induction via rapid activation of conditional proteolysis, a mechanism that requires a degron targeting sequence and dimer to monomer conversion (33, 34). Cellular fractionation and immunoelectron microscopy suggested that CrhR cellular distribution is categorized as two populations, either as a peripheral thylakoid membrane-associated protein or as a soluble polyribosome-associated factor (35). Functionally, CrhR has been associated with maintaining photosynthetic capacity as deduced from various omic analyses identifying CrhR interaction with photosynthesis-associated RNA and protein targets (36–38) and physiologically by regulation of thylakoid membrane structure and function (39, 40). A major mechanism by which cyanobacteria regulate photosynthetic performance involves adjusting the composition of the thylakoid membrane, the site of photosynthetic light harvesting (41). However, many questions remain concerning the mechanism(s) that control protein targeting and assembly to the cyanobacterial thylakoid membrane (41). Thus, despite exhaustive characterization of helicase expression regulation during abiotic stress and recovery transitions, evaluation of the potential contribution of CrhR cellular localization and structural organization to the Synechocystis stress response requires further elucidation.

It has recently been reported that in cyanobacteria, a limited number of housekeeping proteins associated with photosynthesis and cell division functions routinely self-assemble into foci; however, direct involvement of LLPS is limited (42–56). Evidence that LLPS contributes to microcompartment organization in cyanobacteria is currently limited to the in vitro self-assembly of three purified components of the carboxysome, the ubiquitous protein-bound subcellular structure in which photosynthetic carbon fixation occurs (47, 50, 53, 56). Thus, direct evidence of LLPS involvement in dynamic membraneless organelle formation requires further investigation in cyanobacteria.

Here, we utilized laser scanning confocal immunofluorescence microscopy (LSCIM), 1,6-hexanediol treatment, and in vitro condensation to demonstrate that CrhR compartmentalizes into LLPS-mediated membraneless organelles in response to abiotic stress. These complexes form randomly distributed foci under normal conditions but aggregate into larger complexes in the presence of stress. Subsequently, a number of these complexes coalesce to generate crescent-shaped structures localized to a single site on the cell periphery, outside of but not overlapping the thylakoid membrane. These complexes dynamically reorganize spatially and temporally between these two forms in response to fluctuating abiotic stress conditions. Dynamic rearrangement of cellular organization involving formation of CrhR-containing membraneless organelles provides a mechanism by which RNA metabolism and thus the composition and fate of RNP particles could rapidly be regulated. CrhR membraneless organelle localization also identifies the region between the cytoplasmic and thylakoid membranes as an important functional domain, deserving of further analysis.

RESULTS

Immunofluorescent detection of CrhR during temperature shifts.

Prior fractionation studies revealed that CrhR may be present in multiple locations within the cell; therefore, it was of interest to examine subcellular CrhR localization during temperature shifts with greater resolution using an in vivo approach. To accomplish this, we utilized optimized laser scanning confocal immunofluorescence microscopy (LSCIM), with Leica Lightning deconvolution. In addition, sample preparation was based on immunofluorescent staining improvements in E. coli (57), resulting in better antibody access to the cytoplasm, and crucially, reduced background autofluorescence from chlorophyll a (Chl a) embedded in the thylakoid membrane at all wavelengths.

Initially, we examined CrhR distribution during a time course in wild-type cells prior to, during, and after cold stress alleviation to capture potentially transient stress-induced changes in helicase cellular localization (Fig. 1). Cellular distribution and abundance were dramatically affected by a temperature shift between 30 and 20°C. At the optimal growth temperature, 30°C, CrhR was present as a reduced number of foci that were randomly dispersed through the cells (Fig. 1A, 30°C control). In distinct contrast, at 20°C CrhR was primarily clustered in a single pronounced crescent associated with the outer portion of the thylakoid membrane, although more diffuse, patchy distributions within the cytosol were also observed (Fig. 1A, 20°C cold stress). A time course of CrhR downregulation in response to cold stress recovery at 30°C revealed rapid conversion of the crescent structures to low-abundance foci (Fig. 1A, 30°C 1, 2, and 3 h), as observed in cells grown continuously at 30°C (Fig. 1A, 30°C control). A schematic representation of the observed pattern of CrhR localization during temperature shifts is shown in Fig. 1B. These visual interpretations were confirmed by quantification using the corresponding ImageJ plugins of both the red autofluorescent signal derived from thylakoid-embedded Chl a (red) to CrhR (green) ratios (Fig. 1C) and the average particle size (Fig. 1D). In both cases, the corresponding parameter increased dramatically in response to a temperature downshift and subsequently decreased to basal levels in response to stress alleviation. As a control for antibody specificity and background autofluorescence, green fluorescent signal was not observed at 20°C or 30°C in the ΔcrhR mutant background in which the entire crhR open reading frame (ORF) was deleted (see Fig. S1A and B in the supplemental material).

FIG 1.

CrhR dynamic structure and localization during temperature shift. (A) Confocal immunofluorescence microscopy (LSCIM) analysis of cells derived from a time course experiment in which wild-type Synechocystis cultures were transferred from the optimal growth temperature (30°C control) to cold stress (20°C) for 3 h followed by a return to 30°C for 3 h. Cells were fixed, permeabilized, and immunoprobed successively with polyclonal CrhR antiserum (1:400) followed by Alexa Fluor 488 anti-rabbit (1:400, Thermo Fisher). Individual channels of representative images are shown, corresponding to CrhR localization (green) and thylakoid membrane as indicated by autofluorescence of embedded Chl a (red), along with merged frames to indicate CrhR signal in context with surrounding thylakoid membranes, as detected by LSCIM. Scale bar = 1 μm. (B) Depiction of predominant cellular distribution of CrhR at each time point. Thylakoid regions are stylized as concentric red lines enveloped by the thick black line of the plasma membrane. (C and D) Quantification. ImageJ quantification of mean CrhR signal normalized to autofluorescence (C) and CrhR particle size (D) was performed on 10 cell fields spanning data from biological duplicates using default plug-ins. Error bars depict the standard deviation from the mean. Significant differences from cold-stress values were assessed by one-way ANOVA and Dunnett’s multiple-comparison test. ****, P < 0.0001.

We extended our initial evaluation of CrhR localization by determining the diametric fluorescent distribution of Chl a and/or CrhR in a two-dimensional plane using both the ImageJ plot-profile tool and the Interactive 3D Surface Plot plugin. Representative fields of wild-type cells grown at 20°C or 30°C are shown in Fig. 2A and B, respectively. Fluorescent signal intensity corresponding to CrhR (green channel) was plotted as a function of either distance coordinates versus Chl a fluorescent signal (red channel) to provide relative transverse distribution (Fig. 2C and E) or surface plot coordinates with the red channel as a visual landmark delineating cell boundaries (Fig. 2D and F) for the three representative cells grown at either 20°C or 30°C as shown in Fig. 2A and B, respectively. Both of these profiles indicated that CrhR primarily clusters into a discrete, crescent-shaped complex that is external to, but not overlapping, the thylakoid membrane Chl a distribution at 20°C (Fig. 2C and D). As evident in the surface plot, the crescent structure is composed of groups of associating aggregates, as indicated by sites of local maxima and surrounding minima (Fig. 2D). This distribution is in sharp contrast to that observed at 30°C, where CrhR is primarily localized in foci in the cytoplasmic compartment (Fig. 2E and F).

FIG 2.

Two-dimensional localization. (A to F) Wild-type Synechocystis cultures were grown at 20°C (A) and 30°C (B), and three representative cells were analyzed for transverse distribution (C and E) or surface plot profiling (D and F) depicting fluorescent distribution of CrhR (green) relative to the autofluorescence from Chl a in the thylakoid membrane (red). (C and E) Quantification of fluorescent intensity. Fluorescence in the green and red channels was calculated in ImageJ using a line traversing the three cells as shown in panels A and B or surface plot profiling of all of the cells indicated in panels A and B to provide a three-dimensional representation of the relative localization of CrhR and the thylakoid membrane. Note that the z axis of each cell surface plot was significantly enhanced at 20°C. LSCIM imaging was performed as described in Fig. 1. Scale bar = 0.5 μm.

To obtain a more detailed understanding of the structure and positioning of CrhR aggregates in three-dimensional space, a series of deconvoluted z-stacks were obtained for the time points characterized above in two dimensions. The photostability and high quantum yield of the commercial Alexa Fluor 488 dye, along with the distinct isolation of dye signal from autofluorescence using spectral gating, enabled construction of high-resolution three-dimensional (3D) images. The image analysis platform Imaris was used to compile and depict the z-stacks in 3D, as shown in Fig. 3. In congruence with the 2D analysis shown in Fig. 1 and 2, significant alterations in CrhR localization and structure were observed for cells transitioning in response to a temperature shift. In wild-type cells, CrhR primarily aggregates into a single low-temperature-induced crescent in a plane adjacent to yet external to the thylakoid membrane (Fig. 3A and B). As observed in the 2D images (Fig. 1A, 20°C merged), the CrhR crescents lack substantial overlap with the red autofluorescent Chl a signal, indicative of the lack of voxel colocalization between CrhR and the thylakoid membrane (Fig. 3A and B). In comparison, and reminiscent of Fig. 1A, 30°C, CrhR localizes in randomly dispersed spherical foci in the cytoplasm at 30°C (Fig. 3C and D).

FIG 3.

Three-dimensional localization. (A to D) Wild-type (WT) Synechocystis cultures grown at 20°C for 3 h (A and B) and 30°C (C and D) were imaged by LSCIM. z-stacks were compiled using the blended mode option in Imaris. (A and B) CrhR localization in cold-stressed cells. The bulk of the CrhR signal assembles as crescent-shaped aggregates of CrhR (green) at a single location outside the thylakoid membrane (red) at 20°C in a representative cell field (A) or magnified images (B). (C and D) CrhR localization at 30°C. In contrast, individual CrhR foci are randomly distributed throughout the cells at 30°C, as shown in a representative full field (C) and magnified images (D). (E and F) 3-D modeling. 3-D images of representative cells were constructed in Imaris using object-object modeling to visualize the relative distribution of CrhR and thylakoid membranes in three dimensions for wild-type Synechocystis grown at 20°C (E) and 30°C (F). (G) ΔcrhR. As a control, ΔcrhR cells lacking CrhR expression were grown at 20°C and imaged, and object-object modeling was performed to visualize potential nonspecific antibody background fluorescence. For object-object modeling, CrhR (green) and thylakoid membrane (red) distributions were visualized in three dimensions by overlaying the two channels using the maximum intensity projection (MIP) mode in Imaris. LSCIM imaging was performed as described in Fig. 1. Colors represent near thylakoid membrane CrhR objects (magenta) or interior CrhR objects (blue), while autofluorescence from Chl a embedded in thylakoid membranes is shown in red. (H) Quantification. Imaris quantification of the cell volume occupied by CrhR from cells grown under the indicated conditions: wild type (WT) at 30 and 20°C and ΔcrhR at 20°C. Data points shown on the scatterplot reflect the particle size averaged from a minimum of 170 modeled cells obtained from multiple independent cell fields. Error bars depict the standard deviation from the mean. ****, P < 0.0001. Scale bars = 1 μm except for panels A, C, and G (left = 3 μm) and E and F (left = 0.7 μm).

These observations were further enhanced by establishing thresholds to strictly quantify and categorize CrhR signal into one of two classifications, either proximal/overlapping or distal to the thylakoid membrane, for all three-dimensional data. To achieve this objective, object-object modeling of both the thylakoid membranes and CrhR were conducted using the Surface tool in Imaris. In essence, a broad mask initially encompassing all Synechocystis cells in each field as demarcated by the thylakoid region is computer generated and further refined through smoothing and manual curation of dividing or inactive cells. The resulting three-dimensional images shown in Fig. 3F and G depict the Chl a-derived autofluorescence indicative of the thylakoid membrane shown in red and the CrhR-derived fluorescence in close proximity to the thylakoid membrane (magenta) or distal to the thylakoid membrane (blue). At 20°C, CrhR is predominantly localized in close proximity to but outside of the thylakoid membrane in a large crescent composed of individual aggresomes (magenta) (Fig. 3E). Although a relatively minor proportion of the CrhR (blue) appears to localize away from the immediate vicinity of the aggregates (magenta), these structures still interact with the thylakoid membrane-associated complexes (Fig. 3E). Note that two crescents are observed in Synechocystis cells preparing to divide. Similar object-object modeling performed on wild-type cells grown at 30°C again indicated that CrhR formed randomly localized foci (Fig. 3F). As a control, Synechocystis cells in which the crhR ORF is completely deleted, ΔcrhR, and that thus are devoid of CrhR expression, were also imaged (Fig. 3G). As expected, CrhR-derived green fluorescence was not detected in these cells, and thus foci or aggregates were not observed. Volume quantification of CrhR objects in wild-type cells agreed with the two-dimensional quantification of signal intensity since volumes were minimal at 30°C and enhanced 10-fold at 20°C (Fig. 3H). Foci were also not detected in ΔcrhR cells, lacking CrhR expression, at 20°C (Fig. 3H). The corresponding changes in localization, shape, and structural organization of the wild-type CrhR complexes at both temperatures are further visualized in Movie S1, WT 30°C, and Movie S2, WT 20°C. The multicell fields shown in the movies clearly indicate that CrhR-containing crescents are composed of multiple aggresomes. Multiple cells in the 20°C field are in the process of dividing, and thus two crescents are visible.

CTE deletion eliminates CrhR crescent formation during cold shock.

Previous examination of a biochemically inactive, partial truncation mutant of CrhR, CrhR_TR, containing only the first RecA-like domain and thus lacking the entire C-terminal extension (CTE), demonstrated dramatic impacts on both the abundance and sedimentation profile of CrhR polypeptides (35, 39). In these studies, the CrhR_TR polypeptide was constitutively expressed at all temperatures. It was therefore of interest to subject crhR_TR cells to LSCIM imaging. CrhR_TR localization and aggregation were aberrant in crhR_TR cells since they lack prominent crescents and only contain foci that primarily exhibit uniformly distributed cytoplasmic localization, irrespective of temperature (Fig. 4A and B). Similar to the data shown in Fig. 1C, mean intensity ratio, ImageJ quantification of the mean CrhR:Chl a signal intensity ratio observed in wild-type cells shown in Fig. 3F and G, increased significantly between 30°C and 20°C (Fig. 4C, mean intensity ratio). A similar increase was also detected in the crhR_TR cells, an unexpected observation since previous Western blot analysis indicated that CrhR_TR abundance was identical at 30°C and 20°C, a level equal to that observed in wild-type cells at 20°C (28). We suggest that the enhanced CrhR:Chl a signal intensity ratio observed here is not reflective of additional CrhR_TR at 20°C but results from the previously observed decrease in Chl a concentration in crhR_TR cells at 20°C (39). Again, similar to the data shown in Fig. 1D, ImageJ quantification of wild-type cells grown at 20°C, shown in Fig. 3C, also confirmed a significant increase in particle size in wild-type cells shifted from 30 to 20°C, representative of focus to crescent transition (Fig. 4C, average particle size). In contrast, the average focus size remained consistently low at both temperatures in crhR_TR cells, resembling the particles observed in wild-type cells at the nonpermissive temperature, 30°C, as shown in Fig. 1D (Fig. 4C, average particle size). The enhanced abundance of CrhR_TR in these cells correlates with the loss of proteolytic downregulation at 30°C (compare Fig. 4A and B) (28, 33, 34). Furthermore, the observed increase in the number of foci but not crescents in crhR_TR cells, suggests that CrhR abundance is not the primary determinant of crescent formation in the absence of the CTE.

FIG 4.

C-terminally truncated CrhR exhibits localization and abundance defects. (A and B) Cultures of crhR_TR expressing CrhR_TR lacking the CTE, grown at 30°C (A) were transferred to cold stress at 20°C for 3 h (B), and LSCIM imaging was performed as described in Fig. 1. (C) Quantification. Signal quantification was performed using ImageJ; multiple comparisons among groups were conducted using Tukey’s multiple-comparison test. Error bars depict the standard deviation from the mean. *, P < 0.05; **, P < 0.01; ****, P < 0.0001.

Abiotic stress-induced formation of CrhR aggregates.

To examine the conservation of CrhR spatial and temporal structure and localization in response to additional abiotic stresses, wild-type cells were analyzed for accumulation and distribution of CrhR during exposure to low temperature, high salinity, or medium light stress. In accordance with previously reported induction of CrhR under diverse abiotic stress conditions (32), we observed CrhR induction in response to low temperature and to a slightly reduced extent to osmotic and light stress by Western blot analysis (Fig. 5A, Western). In confirmation of the Western blot results presented above, LSCIM revealed that CrhR aggregated into limited numbers of minimal-sized foci in wild-type cells at 30°C that were primarily cytoplasmically localized (Fig. 5B). LSCIM also revealed enhanced CrhR accumulation in response to low temperature, and as reported above, CrhR aggregated into crescent-shaped structures that were primarily localized externally to the thylakoid membrane at 20°C (Fig. 5C). In the absence of temperature stress at 30°C, similar increases in CrhR abundance and aggregation into crescents were observed in response to osmotic (Fig. 5D) and medium light stress (Fig. 5E), although to differing degrees. Quantification of the area covered by CrhR particles confirmed the visual analysis, indicating that larger CrhR aggregates formed in response to abiotic stress (Fig. 5F). We suggest that the observed differences in magnitude of CrhR accumulation and localization observed in osmotic- and light-stressed cells represent a gradation of response, originating from a combination of our choice of the magnitude and duration of stress exposure. Overall, these results indicate that similar compartmentalization of CrhR by condensation occurs in response to a range of abiotic stresses, previously shown to induce CrhR transcript and protein accumulation (32, 58).

FIG 5.

CrhR localization under diverse abiotic stresses. Cultures of wild-type Synechocystis were grown in the absence of stress at 30°C and exposed to either high salt (600 mM) or medium light (150 μmol photons m⁻² s⁻¹) at 30°C or cold stress at 20°C for 3 h, at which time aliquots were prepared for protein extraction and LSCIM imaging. (A) CrhR abundance. A Western blot showing CrhR abundance is consistently enhanced by the three stresses, cold (20°C), osmotic (NaCl), and medium light (ML). Simultaneous detection of PsaA was used as a protein loading control. The Coomassie brilliant blue (CBB)-stained gel corresponding to the Western blot is also shown as a control for protein loading. (B to E) LSCIM detection of CrhR (green) and Chl a (red) in cells treated as indicated was performed as described in Fig. 1. Scale bar = 3 μm. (F) Abiotic stress quantification. ImageJ quantification of the CrhR particle area was performed as described in Fig. 1, using the adaptive thresholding plugin (adaptiveThr) to first identify particles. Error bars depict the standard deviation from the mean. *, P < 0.01; **, P < 0.01; ****, P < 0.0001.

Evidence of CrhR aggregation into LLPS membraneless organelles.

Proteins that form LLPS-mediated membraneless organelles possess structural domains and amino acid sequences that contribute to their formation (59). In support of LLPS-mediated CrhR condensation, flDPnn disorder analysis (60) predicted that CrhR contains a low-complexity domain (LCD) spanning a portion of the C-terminal extension from amino acids 456 to 489 (Fig. S2A). Importantly, this domain was deleted in the crhR_TR mutant. LCDs can guide the process of LLPS by facilitating multivalent protein-protein and protein-RNA interactions (59). Functionally, flDPnn analysis further predicted the presence of a protein-protein and an RNA interaction domain within the LCD (Fig. S2A). Five amino acid sequence motifs characteristic of DEAD-box RNA helicases plus the relative localization of the predicted LCD domains in CrhR are shown in Fig. S2B. These domains are positioned C-terminal of the degron motif (34) with the RNA binding domain containing Arg (R)-Gly (G)-rich motifs (61) (Fig. S2B). The GR and the frequent Asn (N) and Gln (Q) residues associated with the CrhR C-terminal extension and the LCD constitute sequences known to be involved in LLPS-mediated membraneless compartment formation (59) (Fig. S2B). Repeating the flDPnn disorder prediction with CsdA and 25 representative CrhR-type DEAD-box RNA helicases resulted in a similar pattern of enriched disorder in the CTE (Fig. S2C), indicative of evolutionary conservation within the cyanobacterial-specific CrhR RNA helicase clade (62).

To initially explore the possible liquid or solid nature of CrhR aggregates, we treated cold-stressed cells with 1,6-hexanediol, an aliphatic solvent utilized to dissolve LLPS particles (63). Hexanediol is a standard tool for selectively studying eukaryotic stress granule kinetics, as solid complexes held together by strong protein-protein interactions are largely unaffected (63). Consistent with our hypothesis, while LSCIM indicated that CrhR cellular localization in crescents was normal at 20°C in the absence of 1,6-hexanediol (Fig. 6A, 0%), a progressive reduction in CrhR aggregation in both foci and crescents was observed by exposure to increasing 1,6-hexanediol concentrations from 5% to 10% for 30 min (Fig. 6A, 5–10%). This visual observation was confirmed by ImageJ analysis that revealed a linear decrease (r² = 0.9388) in the CrhR:Chl a ratio in response to increasing 1,6-hexanediol exposure (Fig. 6B). As a control for potential effects of 1,6-hexanediol, Western blotting indicated that CrhR abundance was not affected by exposure to 10% 1,6-hexanediol at 20°C for 30 min (Fig. 6C). The corresponding Coomassie brilliant blue (CBB)-stained gel is shown as a loading control (Fig. 6D). In addition, we ruled out the potential that the observations originated from compromised cell integrity, as viability of the 10% 1,6-hexanediol treated cells was similar to that of untreated cells (Fig. 6E). Overall, these results suggest that LLPS-mediated hydrophobic interactions, the type disrupted most effectively by 1,6-hexanediol, are a driving force for CrhR aggregation.

FIG 6.

CrhR aggregation in vivo involves LLPS. Wild-type Synechocystis cultures were grown to mid-log phase at 30°C and subjected to cold stress at 20°C for 3 h and exposed to 0, 5, or 10% 1,6-hexanediol for 30 min. (A) LSCIM localization. Localization of CrhR (green) and Chl a (red) in cold-stressed cells exposed to a concentration curve of 1,6-hexanediol from 0 to 10% for 30 min was detected by LSCIM imaging as described in Fig. 1. (B) Quantification. ImageJ quantification of the mean CrhR signal (green) normalized to Chl a autofluorescence (red) was performed as described in Fig. 1. (C) Western analysis. Western blot showing that CrhR abundance is not affected by the absence (–) or presence (+) of 1,6-hexanediol (10%) for 30 min in two biological replicate cultures, Hex-1 and Hex-2. (D) Protein loading control. The Coomassie brilliant blue (CBB)-stained gel corresponding to the Western blot shown in panel C is provided as a protein loading control. (E) Viability. Synechocystis cells treated with 10% 1,6-hexanediol and used for LSCIM imaging and Western analysis were also tested for hexanediol effect on viability. Error bars depict the standard deviation from the mean. ****, P < 0.0001. Scale bars = 3 μm.

We next sought to better define the phase separation properties of CrhR using an in vitro liquid droplet reconstitution assay, in which the buffer environment can be precisely controlled. In this assay, phase separation of purified CrhR was evidenced by the appearance of spherical droplets which are enriched in fluorescently labeled CrhR (YFP-CrhR). Guided by conditions that induced LLPS of a structurally similar RNA helicase from E. coli, DeaD (13), we utilized high-content microscopy-based analysis to examine the ability of a range of conditions to induce droplet formation using purified yellow fluorescent protein (YFP)-CrhR (Fig. 7). YFP-CrhR formed droplets readily in the absence of any addition, a result that was not substantially affected by the addition of ATP or RNA or both substrates (Fig. 7A, YFP-CrhR, YFP-CrhR + ATP, YFP-CrhR + RNA, and YFP-CrhR + ATP + RNA). Incubation under conditions that would be anticipated to remove RNA from YFP-CrhR, RNase treatment and high salt, did however, severely disrupt droplet formation (Fig. 7A, YFP-CrhR(R) and YFP-CrhR + KCl). These visual conclusions were quantified to identify and characterize circular objects according to fluorescent intensity (Fig. 7B, relative integrated intensity (A.U.)/droplet), size (Fig. 7B, average droplet size), and number of droplets (Fig. 7B, droplet number). In reference to the YFP control, YFP-CrhR alone exhibited a dramatic capacity for droplet formation, as indicated by all three quantification criteria (Fig. 7B). Although ATP or RNA did not significantly accentuate the intensity or number of droplets, both RNA alone and RNA in combination with ATP significantly enhanced the average droplet size, signifying that incorporation of RNA was required for YFP-CrhR condensate formation (Fig. 7B, average droplet size). Conversely, removal of RNA by high-salt treatment significantly decreased aggregate formation, based on all three criteria (Fig. 7B). In contrast, RNase treatment did not alter the number of droplets, suggesting that high salt more efficiently removed RNA from YFP-CrhR, resulting in a reduced ability to form droplets (Fig. 7B, droplet number).

FIG 7.

In vitro phase-separated YFP-CrhR droplet formation is positively associated with RNA. (A) Enriched YFP-CrhR droplets were fluorescently imaged using the MetaXpress high-content screening platform to simultaneously capture LLPS events in a low-salt buffer. When indicated, ATP (1.5 mM), ssRNA (polyU; 50 ng/μL), or high KCl (225 mM) were added. Reactions were performed in triplicate in 96-well plates. Images from nine independent fields of droplets were obtained and quantified for each biological replicate (n = 27). (B) Quantification. Relative integrated intensity, average droplet size, and droplet number were quantified using MetaXpress software, configured with a custom module as described in Materials and Methods. Deviation from the mean was assessed by one-way ANOVA and Tukey’s multiple-comparison test. ****, P < 0.0001. (C) The LLPS-promoting effect of different nucleic acid species was assessed using RNase-treated YFP-CrhR [YFP-CrhR(R)]. Droplet formation was monitored in the same low-salt buffer as in panel A, using RNase-treated YFP-CrhR [CrhR(R)] combined with the addition of the indicated RNA or DNA species at a final concentration of 250 ng/μL or high KCl (225 mM), as indicated. The nucleic acid additives, ssRNA (polyU) and dsRNA [poly(I·C)] and the dsDNA (sheared salmon sperm) were used directly, whereas ssDNA was produced by heat denaturation of dsDNA snap-cooled immediately prior to the assay. Boiled YFP-CrhR (5 min) served as an inactive, protein-containing control in the presence of ssRNA. (D) Quantification. Relative integrated intensity per droplet was performed using MetaXpress software, configured with a custom module, modified as described in Materials and Methods. Deviation from the mean was assessed by one-way ANOVA and Tukey’s multiple-comparison test. ***, P < 0.001. Scale bar = 10 μm.

The results presented in Fig. 7A and B indicated RNA as a driver for YFP-CrhR droplet formation. We therefore further examined YFP-CrhR aggregation in the presence of different RNA or DNA species, after initial depletion of residual YFP-CrhR-associated RNA by RNase A treatment, designated YFP-CrhR(R) (Fig. 7C). Consistent with the in silico prediction of an RNA interaction element in the CTE LCD (Fig. S2A), CrhR phase separation responded most vigorously to RNA and not DNA, exhibiting a significantly enhanced degree of aggregation in response to both single-stranded RNA (ssRNA; polyU) and double-stranded RNA [dsRNA; poly(I·C)]. Droplet formation was also shown to be RNA specific since ssDNA or dsDNA did not significantly enhance condensate formation above the intensity of the heat-denatured YFP-CrhR(R) control (Fig. 7C and D, YFP-CrhR(R)+boil+ssRNA). Despite the lack of an obvious influence of ATP in phase separation of the independent droplets in the analysis presented in Fig. 7A and B, we dissected a possible contribution of ATP binding in the absence of ATP hydrolysis by inclusion of the slowly hydrolysable ATP analog ATP-γ-S in our system (64). The intensity of YFP-CrhR aggregates in the presence of ATP-γ-S and ssRNA was indistinguishable from that of purified protein mixed with ATP and ssRNA, suggesting that RNA but not ATP hydrolysis was required for droplet formation.

DISCUSSION

Here, we provide evidence that LLPS-mediated compartmentalization of the single DEAD-box RNA helicase encoded in Synechocystis sp. PCC 6803, CrhR, into membraneless organelles is achieved by dynamic temporal and spatial fluctuations in response to diverse abiotic stresses. Laser scanning confocal immunofluorescence microscopy revealed that CrhR is localized in foci randomly distributed primarily within the cytoplasm at 30°C. Abiotic stresses induced rapid aggregation of foci into elongated aggresome structures, several of which coalesce into a crescent. Spatially, CrhR-containing aggresomes are confined to a single region of the cell, forming striking crescents that occur between the thylakoid and cytoplasmic membranes. Temporally, CrhR crescents are highly dynamic, dissolving rapidly in vivo in response to stress alleviation. Dissolution was also elicited by 1,6-hexanediol, a treatment known to dissolve liquid-like condensates (63), providing evidence that CrhR aggresome formation occurs via an LLPS-mediated condensation mechanism. Spontaneous in vitro formation of CrhR droplets further confirmed the ability of CrhR to self-assemble into large phase-separated complexes, a reaction that occurred in an RNA-dependent manner. Together, these observations are suggestive of an LLPS-mediated mechanism by which CrhR coalescences into membraneless organelles associated with the outer portion of the thylakoid membrane in response to stress conditions.

This higher-order structural change most likely involves structural and sequence attributes of the predicted CrhR LCD C-terminal domain and demonstrated dimerization of CrhR (34), both of which are known to regulate LLPS (5, 59, 65, 66). Self-aggregation of LCD-containing proteins is known to induce phase transition in vitro (65, 66), providing a mechanism explaining LLPS formation and their dynamic properties (17). This process also applies to LLPS-mediated aggregation of other DEAD-box RNA helicases in eukaryotic and prokaryotic systems (8, 13, 17, 67). The CrhR CTE is also rich in arginine-glycine motifs and in asparagine (N) and glutamine (Q) residues, sequences associated with LLPS-mediated focus formation (13, 61). Structurally, CrhR aggregation was not driven solely by CrhR abundance, as only foci, and not aggresomes or crescents, were observed in CrhR_TR cells that constitutively express elevated levels of a truncated form of CrhR, lacking the LCD containing CTE, at all temperatures (28). The abundant foci observed in CrhR_TR cells suggest that other factors can contribute to focus formation, potentially involving CrhR_TR association with RNA and/or LCD-containing protein partners. These attributes provide additional evidence that the CrhR C-terminal LCD domain can spontaneously form LLPS aggregates under stress conditions.

Maintenance of thylakoid membrane structure and function to perform the light reactions of photosynthesis is crucial for cyanobacterial survival; however, many aspects remain elusive (43). Similar to the CrhR foci observed here at 30°C, foci associated with thylakoid biogenesis or stress repair are also localized within the thylakoid membrane or between the thylakoid and cytoplasmic membranes (68–70). However, in vivo, aggresome-like structures similar to those observed here for CrhR have been reported only for pigment-protein complexes within the thylakoid membrane (46) and crescent-like structures of PilB1 formed at the cell periphery, associated with twitching motility and the direction of phototaxis movement (71). Structurally, it is interesting to speculate that the mechanism dictating CrhR crescent localization is associated with type IV pilus-mediated phototaxis motion mediated by light sensing via lensing (72).

Functionally, we therefore propose that aggresome-associated CrhR is involved with the maintenance of thylakoid membrane integrity and/or function in response to abiotic stress. This proposal is supported by our previous results showing that expression of a truncated form of CrhR, CrhR_TR that lacks the cyanobacterial specific C-terminal extension, dramatically affects thylakoid membrane structure, carboxysome abundance, photosynthetic carbon fixation and a concomitant rapid decrease in survival (39). A reduction in fitness has also been observed in response to inhibition of membraneless organelle formation by LLPS in E. coli (8). In addition, this inference is further supported by omic and physiological analysis indicating altered expression of photosynthetic components in the absence of functional CrhR (36–38, 40). Combined, these observations suggest that CrhR aggresomes maintain thylakoid membrane and/or carboxysome formation and function that dramatically affect fitness during low-temperature stress.

Biochemically, the duplex RNA unwinding and annealing activities exhibited by CrhR in vitro (73) would contribute to the concentration and alteration of RNP component composition and functionality associated with the abiotic stress response. CrhR could function in translation initiation to produce thylakoid- or cytoplasmic-membrane proteins or as a clamp to stabilize RNP particles, to either enhance or deter RNA translation or degradation. In addition, CrhR-catalyzed unwinding combined with proteolytic degradation of CrhR in response to stress alleviation could contribute to the dynamic dissolution of CrhR aggresomes (28, 33, 34, 73). A potential target of CrhR-mediated alteration of RNA/RNP structure could be the crhR transcript itself, which is a major target for CrhR binding (38). This interaction is necessary for proper auto-regulated processing of the dicistronic rimO-crhR transcript at 20°C (29, 38). The resulting CrhR RNA helicase activity could alter the ATP:ADP ratio in the aggresome, an emerging factor associated with maturation of membraneless organelle dynamics (13, 14). Merging insights from our in vivo and in vitro data, it is likely that a combination of LCD-containing CrhR, dsRNA originating from translational impairment (74), and other physical parameters such as slow diffusion rate toward the cell periphery (75) would work in concert to dictate the extent of CrhR aggregation in an RNA-dependent manner.

The results presented here show that an LLPS-mediated mechanism compartmentalizes the Synechocystis DEAD-box RNA helicase, CrhR, into membraneless organelles in response to abiotic stress. This sequestration would provide cyanobacteria with the ability to regulate RNA and RNP structure, composition, and fate, allowing alterations in thylakoid- and/or cytoplasmic-membrane composition and, thus, function. The intriguing questions regarding the composition of CrhR aggresomes and the structural and functional mechanisms generating the restricted localization and crescent formation remain to be elucidated. On a broader scale, the data suggest that LLPS may perform a more extensive role in compartmentalization of biochemical reactions in cyanobacteria than previously anticipated. These results have important implications for regulation of RNA metabolism and cellular fitness in all bacteria.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Axenic cultures of the glucose-tolerant, nonmotile strain Synechocystis sp. PCC 6803 were maintained on BG-11 agar plates at 30°C under continuous illumination (36 μmol photons m⁻² s⁻¹) (76). Cells were grown in liquid BG-11 at 30°C with continuous shaking (150 rpm) coupled with aeration provided by bubbling with humidified air and continuous illumination (36 μmol photons m⁻² s⁻¹) (76). Medium light was provided at 150 μmol photons m⁻² s⁻¹. Two crhR mutant strains were used in this study: a ΔcrhR mutant in which the entire crhR ORF was replaced with a kanamycin resistance cassette (33) and a truncation mutant, crhR_TR, in which residues R228 to Q492 were deleted by an aminoglycoside-3′-adenyltransferase (aadA) cassette (39). The ΔcrhR and crhR_TR strains were grown in BG-11 medium supplemented with kanamycin (50 μg/mL) and spectinomycin, respectively, plus streptomycin at 50 μg/mL each.

Immunofluorescent staining.

The framework for immunofluorescent staining was adapted from Trigo et al. (77) and Park et al. (57). For fixation, a 1-mL aliquot of cells in mid-exponential phase (optical density at 750 nm [OD₇₅₀], 0.3 to 0.6) was pelleted and suspended in 4% paraformaldehyde (PFA) in 1× phosphate-buffered saline (PBS) at room temperature for 30 min. Cells were washed, alcohol treated, and permeabilized according to Park et al. (57) with the following modifications: to lower autofluorescent intensity, chlorophyll was depleted with an overnight 100% methanol incubation at 4°C, followed by extensive methanol washing until the supernatant was clear. Permeabilization was achieved using a two-stage process on cells prior to immobilization, first by incubation in 1 mg/mL lysozyme for 30 min at 37°C, followed by a 10-min room temperature incubation in 0.2% Triton X-100. Samples were pelleted at 600 × g for 5 min, suspended in 50 μL of fresh 1× PBS, and allowed to dry at 30°C on Thermo Fisher Probe On Plus slides. Subsequent blocking, antibody hybridization, and washing steps were performed as described by Trigo et al. (77), except that the wash steps were performed for 10 min in 2× PBS and 0.1% Triton X-100. The primary antibody was polyclonal serum raised against CrhR in rabbit (1 in 400). The secondary antibody was an anti-rabbit Alexa Fluor 488 (1:400) conjugate (Sigma-Aldrich). Multiple cell fields were generated from two independent biological replicates from which a minimum of 170 cells were analyzed at each time point for each genotype and abiotic stress time course.

Laser scanning confocal immunofluorescence microscopy (LSCIM).

Images of fixed cells were obtained using the Leica Falcon SP8 system with 100× Plan-Apo oil objective (numerical aperture [NA], 1.4), and Leica Application Suite X (LAS X) software version 3.7.2. Detection of the green (CrhR), and red (Chl a) channels was performed by sequential scanning in lightning deconvolution mode according to the following settings: excitation by white laser light (WLL) at 488 nm and emission range of 503 to 515 nm (Alexa Fluor 488) as collected by an HyD detector or excitation at 552 nm and emission range of 680 to 755 nm (Chl a) as detected by a photomultiplier tube detector. Images corresponding to 5 to 10 cell fields across each slide were captured, corresponding to each time point or condition.

ImageJ quantification of fluorescence.

The z-stacks were analyzed four ways using raw data files in ImageJ. Initially, quantification of mean intensity (CrhR/Chl a ratio) and average particle size (μm²) was performed using the histogram and analyze particle functionalities, respectively, in the Fiji package (version 1.53C). Graphs depict analysis of medial slices from 10 independent cell fields, comprising a minimum of 300 cells. Then, to depict the relative transverse localization of red autofluorescence and green CrhR-derived fluorescence, transects were drawn to encompass Chl a (red) and the predominant CrhR (green) signals in representative cells and analyzed using ImageJ Plot Profile. All CrhR pixel values were normalized (y = 1.0) to the point of maximum measured arbitrary units. Finally, surface plot profiles were generated using the Interactive 3D Surface Plot plugin (version 2.4.1) on ovals encompassing representative cells.

Imaris movie generation.

Synechocystis cells were grown to mid-log phase at 30°C, and an aliquot was subjected to cold stress at 20°C for 3 h. LSCIM images were collected as described above. Representative cell fields were imported into Imaris version 9.7.1 (Bitplane). Automatic three-dimensional assembly of both red and green channels was conducted during the file conversion process. Chl a autofluorescence, indicative of thylakoid membrane position, is shown in red. CrhR is depicted either as the original Alexa Fluor 488 signal (green) or as Imaris-generated models, as described below. Red (Chl a) and green (CrhR) signals in all Imaris-derived images were set to the same display values of minimum 0.0 maximum 3,248 gamma 1.0, and minimum 0.0 maximum 3,271 gamma 1.0, respectively. Movies S1 and S2 depict 3-D images of cells grown at 30°C and 20°C, respectively.

Imaris 3-dimensional modeling.

The surface tool wizard in Imaris version 9.7.1 (Bitplane) was used to model both the red representing Chl a autofluorescence, thus identifying the location of the thylakoid membrane, and green (CrhR signal detected by the Alexa Fluor 488 secondary antibody) channels in 3 dimensions, as verified previously (78). Thresholds were applied to generate red channel models using absolute intensity to resolve cyanobacterial cells, with surface smoothing set to 0.2 μm (wild type) or 0.3 μm (ΔcrhR). Local variations in cell density further necessitated both automatic and user-guided segmentation steps as follows. Objects in close proximity were disconnected in the first pass by setting the seed point diameter to 1.4 μm and filtering by quality to exclude cells dissected by the field of view and, where necessary, in a second pass manually using the cut tool. Objects overlapping in the z-dimension that could not be clearly extracted were deleted. Objects identified as single objects but that constituted dividing cells were manually segmented. Lastly, models produced for cells exhibiting unusually minimal red autofluorescence that appeared nonspheroidal were excluded. Parameters for green channel modeling, corresponding to CrhR, were determined using the control ΔcrhR images to establish a baseline for expected nonspecific background in the wild type. Surface smoothing was auto-set to 0.0412 μm, and thresholding was determined using a background subtraction set to 0.155-μm sphere diameter and adjusting the local contrast value to 484 to remove background fluorescence. Nearby object splitting was achieved with a seed point diameter of 0.206 μm and filtering with a quality value of 442. The objects were then filtered to a minimal voxel size of 28 and a maximal distance of 0 μm from the surface of red models, representing the thylakoid membrane region, to exclude extraneous signal. Finally, CrhR objects were sorted into two classes: “near membrane” (magenta), at a distance no greater than 0.1 μm from the surface of the red autofluorescence model, and “interior” (blue), at distances beyond 0.1 μm from this surface. Volumes of CrhR aggregates (μm³) were obtained from Imaris Vantage View and analyzed in GraphPad Prism software version 9.2.0. Individual data points represent measurement from multiple independent fields, obtained from a minimum of two biological replicates, encompassing a minimal sample size of 200 cells for each 20°C or 30°C condition.

Cell viability after hexanediol challenge.

To assess potential 1,6-hexanediol effects on cell viability, an aliquot from each culture exposed to the highest concentration of hexanediol (10%) for 30 min at 20°C was washed and serially diluted with BG-11 and plated in triplicate on BG-11 agar plates. After incubation at 30°C for 7 days, colony counts allowed calculation of the CFU/mL.

CrhR structural disorder and putative function prediction.

The amino acid sequence of CrhR was retrieved from the UniProt database (ID: Q55804) and uploaded to the flDPnn webserver for disorder analysis (60). Raw values for residue-specific disorder propensity as well as protein, DNA, and RNA binding propensity for residues categorized as disordered were exported in csv format and plotted using GraphPad Prism software version 9.2.0.

CrhR Western blot analysis.

Western blot analysis of CrhR protein abundance was performed as previously described (76). Synechocystis cells were grown to mid-log phase and subjected to the indicated stress conditions. Cell pellets were lysed by vortexing in the presence of glass beads, to release all CrhR into the supernatant, and clarified by centrifugation. The soluble protein concentration was determined using a Bradford assay with bovine serum albumin (BSA) as the standard. For each sample, wild-type CrhR (55 kDa) was detected by separation on a 10% Tris-glycine SDS-PAGE gel, transferred to a nitrocellulose membrane, and probed with rabbit anti-CrhR antibody (1:5,000), followed by horseradish peroxidase (HRP)-conjugated rabbit antibody (1:10,000) and colorimetric detection using ECL (Bio-Rad Clarity Western ECL substrate, Mississauga, ON, Canada) recorded on a ChemiDoc MP imaging system (Bio-Rad). Coomassie brilliant blue (CBB) staining and, where indicated, simultaneous detection of PsaA (Agrisera) were used as controls for protein loading.

YFP-CrhR expression, purification, and in vitro droplet assembly analysis.

A yellow fluorescent protein fusion construct was generated by amplifying the mVenus sequence of pSHDY-P_rha-mVenus_rhaS (AddGene plasmid no. 137662) for HiFi assembly (NEB Biolabs) at the N terminus of the crhR cassette within a linearized pRSET A-crhR overexpression vector (73). Seamless integration was performed as per manufacturer recommendations to produce YFP-CrhR. The primers and plasmids used in this study are listed in Table S1. Expression and subsequent purification of mVenus-CrhR was conducted as described previously for CrhR (34). Protein purity and approximate concentration were assessed using SDS-PAGE and Coomassie staining. Size exclusion fractions were pooled to produce high-concentration (6 μM) stocks in size exclusion chromatography (SEC) storage buffer (10 mM HEPES, pH 7.3, 500 mM KCl, 1 mM dithiothreitol [DTT], and 10% glycerol). Stocks were aliquoted and snap-frozen in liquid N₂ and stored at −80°C. A 15-mg/mL stock of purified enhanced YFP (eYFP), a kind gift of Yi Shen (University of Alberta), was diluted to 6 μM in SEC storage buffer and used as a control for droplet assembly.

Characterization of CrhR in vitro droplet formation followed procedures outlined for analysis of generalized RNA helicase condensates (13, 79). All buffer reagents were prepared in RNase-free water and filtered through a 0.2-μm membrane (Sigma-Aldrich). In vitro condensation assays were performed in a low-salt buffer (30 mM HEPES, pH 7.4, 0.6 μM YFP-CrhR, 50 mM KCl, 0.4 mg/mL BSA, ssRNA [polyU; 50 ng/μL], and 20 U RNasin RNase inhibitor [Promega]) in a final volume of 75 μL. When indicated, ATP (1.5 mM), ssRNA (polyU; 50 ng/μL), or high KCl (225 mM) were added. Snap-thawed YFP-CrhR was treated with RNase A (0.5 μg/μL) at room temperature for 40 min to generate CrhR(R) prior to use. The RNA additives, ssRNA (polyU) and dsRNA (poly(I-C)), as well as the dsDNA (sheared salmon sperm) were used directly at 250 ng/μL, whereas ssDNA was produced by heat denaturation of dsDNA snap-cooled immediately prior to the assay. Boiled YFP-CrhR (5 min) served as an inactive but protein-containing control in the presence of ssRNA.

For each experiment, fresh snap-thawed YFP-CrhR was pipetted into individual wells in a 96-well plate (Thermo Fisher M33089), diluted 10-fold with appropriate premixed buffer, and allowed to settle to the bottom of the well by gravity at room temperature for 1 h. Epifluorescent images were then obtained using a MetaXpress XLS system (Molecular Devices, San Jose, CA, USA). Dense particles at the well bottom were visualized with a 60 × 0.85 NA Plan Fluor objective, FITC/ALEXA488 filter set, and sCMOS camera (2180 by 2180 pixels). The digital confocal setting was enabled to reduce interference from of out-of-focus droplets above the bottom of the well. A total of three wells, and nine sites within each well, were imaged for each condition. Representative images were converted to RGB from grayscale in Photoshop and false-colored yellow, without alteration of brightness or exposure.

Image quantification was accomplished using a custom MetaXpress module, which constituted two principal segmentation components. First, a Top Hat filter set to size 25-pixel circular shape was applied to raw images and served as the input for the second Find Blobs component. Thresholds were set to identify droplets between minimum and maximum widths of 0.4 and 30 μm, respectively, while counted objects required a fluorescent intensity greater than 600 arbitrary units above local background. Exported data were further refined by removal of identified objects with a shape factor of less than 0.5 and a size greater than 10 μm², corresponding to large, spurious, nondroplet artifacts. Size and integrated intensity values were imported into GraphPad Prism for graphing and statistical evaluation. For examination of nucleic acid species and YFP-CrhR LLPS, the shape factor export filter was removed, and the size requirement was elevated to objects less than 80 μm². Cross-comparison of treatments was conducted only for those coimaged on the same plate. Depicted means and error bars were derived from measurement of each individual droplet outlined by the threshold parameters described above, without binning by well site, to avoid the influence of uneven droplet distribution in wells.

Statistical evaluation.

For quantification of signal features derived from microscopy, raw data obtained from ImageJ, Imaris, or MetaXpress were exported to GraphPad Prism software version 9.2.0 and graphically depicted. Significant differences between groups were ascertained using one-way analysis of variance (ANOVA) with default settings and Dunnett’s multiple-comparison test comparing selected means to a control or, where indicated, Tukey’s multiple-comparison test to compare the means of multiple groups among themselves.