Bacterial strains and growth conditions
All strains, plasmids, and oligonucleotides used in this study are listed in Supplementary Tables 3–5. Recombinant C. crescentus strains were obtained by the electroporation of integrating or replicating plasmids. C. crescentus cells were grown aerobically in peptone yeast extract (PYE) rich medium or M2G minimal medium containing 0.2% (w/v) glucose at 30 °C. Recombinant E. coli strains were obtained using the standard clone method and were grown aerobically in Luria–Bertani (LB) medium at 37 °C unless otherwise stated. The plasmids were constructed by Gibson assembly or the standard PCR-based mutagenesis method and were verified by DNA sequencing. All oligonucleotides were synthesized by Sangon Biotech.
When required, antibiotics were used at the following concentrations (liquid/solid media for C. crescentus; liquid/solid media for E. coli; µg ml−1): kanamycin (5/20; 50/50), chloramphenicol (1/2; 20/20), spectinomycin (not applicable; 50/50), ampicillin (not applicable; 100/100). All reagents used in this study were purchased from Sigma-Aldrich unless otherwise stated.
Bioinformatic analyses of PodJ
The intrinsic disorder tendency of PodJ was analyzed using four independent programs: Metadisorder MD259, SPOT60, Cspritz61, and IUPred262. The scores of these programs were plotted against the PodJ sequence and assigned between 0 and 1, and a score above 0.5 indicates disorder. In the current study, an IDR was predicted with a disorder probability above 0.75 in PodJ.
TOPCONS63 and HHpred64 were used to analyze the protein topology and domain, respectively, especially in the prediction of transmembrane domains and the cellular orientations of proteins. Predicted cytoplasmic termini of these proteins were used for the construction of fluorescent fusion proteins.
Classification of Intrinsically Disordered Ensemble Regions (CIDER)65 was used to predict the charge distribution of PodJ_N.
The rapid automatic detection and alignment of repeats algorithm (RADAR)57 was used to detect the TRs in PodJ_N, and the results were manually cured.
Protein expression and purification
The proteins purified and analyzed in this study are listed in Supplementary Table 6. Recombinant E. coli BL21(DE3) cells harboring the expression plasmid were grown in LB medium at 37 °C until the OD600 reached approximately 0.6, after which cells were induced with 0.5 mM Isopropyl β-D-1-thiogalactopyranoside (IPTG) at 16 °C for 12 h. Cell pellets were collected and resuspended in lysis buffer containing 25 mM HEPES (pH 7.5), 800 mM NaCl, 10% (v/v) glycerol, and 20 mM imidazole. After sonication and centrifugation, the supernatant was incubated with Ni2+-NTA agarose resin at 4 °C for 1 h, which was pre-equilibrated with the lysis buffer. After washing with 10–20 column volumes of wash buffer containing 25 mM HEPES (pH 7.5), 800 mM NaCl, 10% (v/v) glycerol, and 30 mM imidazole, the recombinant His-tagged protein was eluted from the agarose beads with elution buffer containing 25 mM HEPES (pH 7.5), 500 mM NaCl, 10% (v/v) glycerol, and 500 mM imidazole. When needed, the eluted proteins were concentrated with a dialysis bag (Sangon Biotech, SP132574, MWCO 10,000 g mol−1) immersed in PEG-8000 powder and further dialyzed in a dialysis buffer containing 25 mM HEPES (pH 7.5), 200 mM NaCl, 10% (v/v) glycerol, and 1 mM dithiothreitol (DTT) three times at 4 °C. Concentrations of purified proteins were determined using the Bradford protein assay kit (Beyotime Biotech, P0006). All recombinant proteins were obtained with purity >80% and their SDS-PAGE analyses are shown in Supplementary Fig. 13. The dialyzed proteins were stored at −80 °C before use.
Native polyacrylamide gel electrophoresis (PAGE) analysis
To investigate the oligomeric state of PodJ, native PAGE analysis was performed. The purified PodJ_N protein was diluted using a 5× native sample loading buffer (Sangon Biotech, C506032) and separated on a 7.5% native PAGE gel (BBI, C601100). Bovine serum albumin (BSA) (BBI, A600903) was used as a control. The loading concentration of purified PodJ_N applied to each lane varied between 3.6 and 7.2 μg. The gel was run in 1× HEPES native PAGE running buffer (pH 7.5, BBI, C601110) at 80 V for 4 h at 4 °C, and stained with Coomassie brilliant blue R-250.
In vitro liquid–liquid phase separation (LLPS) assay and data analysis
For the in vitro phase separation assay, all experiments were performed in a buffer containing 25 mM HEPES (pH 7.5), 10% (v/v) glycerol, 1 mM DTT, and 200 mM NaCl, unless otherwise stated. Approximately 2 µl (5 µM) purified protein solution (PodJ_N, YFP-PodJ_N, or SpmX(∆TM)-mCherry) was loaded onto a 35-mm glass bottom dish (Cellvis, D35-20-1-N) and imaged immediately using a Nikon A1R+ confocal laser scanning microscope equipped with a 100× oil immersion objective lens. The same amount of purified YFP or mCherry protein was used as the control. The specimens were illuminated with a 488-nm laser for yellow fluorescence and with a 561-nm laser for red fluorescence. YFP fluorescence was detected using the FITC filter (Nikon, excitation filter 480/15, dichroic mirror 505, and emission filter 535/20), and mCherry (mChy) fluorescence was detected using the TRITC filter (Nikon, excitation filter 540/25, dichroic mirror 565, and emission filter 605/55). All images were acquired using the same laser power, exposure time, gain, and offset settings at ~25 °C, unless otherwise stated.
Two types of sample loading methods were used to determine the effects of proteins (SpmX, PleC, or BSA) upon the PodJ phase separation. (I), 2 µl YFP-PodJ_N protein solution is loaded onto the glass bottom. When the YFP-PodJ_N droplets are formed (~15 min), another 2 µl of tested protein solution with indicated concentration is added onto the edge of the YFP-PodJ_N sample to let them touch each other, and the images are taken meanwhile. (II), 2 µl YFP-PodJ_N protein solution is mixed directly with 2 µl tested protein solution and loaded onto the glass bottom. The images are taken immediately.
The fluorescence intensity and the size of YFP-PodJ_N liquid droplets were quantitatively analyzed using MicrobeJ66. To visualize the interfacial interaction between PodJ and SpmX, the three-dimension (3D) image stacks (0.1 µm Z steps) were captured, and the mapping analysis was performed using 3Dscript67.
Single-particle tracking and analysis
Single microspheres of YFP-PodJ_N droplets were tracked using the Fiji/ImageJ plugin MTrackJ. Microspheres on the glass surface or near the edge of the tested protein solutions were excluded from the data analysis to avoid artifacts. The feature size and minimum intensity of YFP-PodJ_N microspheres were empirically chosen so that most of the visible microspheres were detected (>80%) in a frame. For the tracks with at least 8 consecutive frames, their trajectory coordinates (x, y, t) were used to calculate the two-dimensional mean-squared displacement (MSD). The MSD(t) can be defined by Eq. (1):
where r(t) is the position of the microspheres at time t and r(0) is the initial position.
The mean MSD was used to calculate the apparent diffusion coefficient (D) using Eq. (2):
where α is the anomalous diffusion exponent obtained by linearly fitting the dataset with log(MSD) (y) and t (x). D is obtained by linearly fitting the dataset with MSD (y) and tα (x). The MSD increases linearly with t for the PodJ_N droplets alone and the PodJ_N droplets in the presence of PleC(∆TM) or BSA (α ≈ 1), indicating that these PodJ_N microspheres did not experience subdiffusion. In contrast, the PodJ_N droplets in the presence of SpmX(∆TM) have a markedly different behavior with α ≈ 0.78, showing subdiffusion motions. To obtain the D, we fit the MSD data to MSD = 4Dt0.78 for PodJ_N droplets in the presence of SpmX(∆TM), and MSD = 4Dt for other samples. Only tracks with coefficient of determination (R2) ≥ 0.8 are included in the analysis.
Partitioning analysis of molecules into the condensed phase
Partitioning analysis was used to investigate the phase distribution of molecules. The partitioning coefficient, K, was defined using Eq. (3):
where Iout and Iin are the average fluorescence intensity outside and inside the condensates, respectively. The fluorescence intensity data were obtained using the ROI manager tool of Fiji/ImageJ, and the same‐sized regions of at least 30 droplets and background from three independent experiments were selected in each sample. The partitioning coefficient is often expressed as Ln K, where a negative value indicates the proteins are likely to partitioning into the condensed phase.
Fluorescence recovery after photobleaching (FRAP)
FRAP was used to investigate the dynamic internal rearrangement and the internal-external exchange of molecules within PodJ condensates. The in vitro FRAP analysis of liquid droplets formed by YFP-PodJ_N or its variants was performed using a Nikon A1R+ confocal laser scanning microscope with a 100× oil immersion objective lens. The fluorescence signal within the selected regions of protein droplets was bleached using a 488-nm laser at 50% laser power for approximately 5 s. After photobleaching, time-lapse images were captured at a rate of 1 s for approximately 5 min. The droplets with diameters ~5 µm were selected for assays at ~25 °C.
For the in vivo FRAP analysis, C. crescentus cells (NA1000, pBXMCS2-Pxyl-yfp-podJ_N and ∆spmX, pBXMCS2-Pxyl-yfp-podJ_N) and E. coli cells containing plasmid (pCDF-YFP-PodJ, pCDF-YFP-PodJ_N, pCDF-YFP-PodJCC1–3, pCDF-YFP-PodJCC4–6, or pCDF-YFP-PodJIDR) were induced and immobilized on a 1.5% (w/v) agarose pad. Due to the relatively small sizes of C. crescentus cell poles (diameter: 0.2–1 µm) and the requirement for a relatively large bleach region (diameter: ≥0.4 µm), the PodJ proteins were expressed from high copy plasmids and driven by the PxylX promoter as described above. The fluorescence signal within the selected region was bleached using a 488-nm laser at 50% laser power for approximately 2 s. After photobleaching, time-lapse images were captured every 2 s for about 5 min at 28 °C.
For each indicated time point (t), the fluorescence intensity within the bleached region was normalized to the fluorescence intensity of a nearby, unbleached region. The normalized fluorescence intensity of pre-bleaching region was set to 100% and the normalized fluorescence intensity at each time point (It) was used to calculate the fluorescence recovery using Eq. (4):
GraphPad Prism 5.0 program was used to plot and analyze the FRAP experiments.
Characterization of subcellular accumulation of PodJ
For cell imaging studies of PodJ in C. crescentus, recombinant strains containing a single copy of sfgfp-podJ (C. crescentus NA1000, xylX::Pxyl-sfgfp-podJ; ∆podJ, xylX::Pxyl-sfgfp-podJ; ∆popZ, xylX::Pxyl-sfgfp-podJ; ∆tipN, xylX::Pxyl-sfgfp-podJ; ∆spmX, xylX::Pxyl-sfgfp-podJ) or its variant (∆podJ, xylX::Pxyl-sfgfp-podJ_N) in the chromosome were cultivated overnight at 30 °C and transferred to fresh PYE medium at a ratio of 1:10 (v/v). Cells were grown to an optical density at 600 nm of 0.7, and then induced with 0.003% (w/v) xylose for 3 h before imaging.
For cell imaging studies in E. coli BL21(DE3), recombinant cells containing relevant expression plasmids (pCDF-YFP-PodJ, pCDF-YFP-PodJ_N, pCDF-YFP-PodJCC1–3, pCDF-YFP-PodJCC4–6 or pCDF-YFP-PodJIDR) were cultivated overnight at 37 °C and inoculated in fresh LB medium at a ratio of 1:100 (v/v). Cells were grown to an OD600 of 0.4 and then induced with 0.1 mM IPTG for 2 h before imaging.
The C. crescentus and E. coli cells were immobilized on a 1.5% (w/v) agarose-PYE pad and a 1.5% (w/v) agarose-LB pad, respectively. Cells were then imaged with a Nikon Eclipse Ti2-E inverted fluorescence microscope equipped with an Andor iXon Ultra DU897 EMCCD camera, using a 100× oil immersion objective lens. The specimens were illuminated with a 488-nm laser for green/yellow fluorescence. All images were acquired at 25 °C with the same laser power, exposure time, gain, and offset settings. The fluorescence intensity of the cells was quantitatively analyzed using MicrobeJ66.
Inclusion body detection in E. coli
To rule out the possibility of dysfunctional aggregation of PodJ in cells, we co-expressed YFP-PodJ together with an inclusion body marker IbpA-mCherry44 in E. coli. The recombinant cells containing pCDF-IbpA-mCherry plasmid, or those co-transformed with pBAD-YFP-PodJ, pBAD-YFP-PodJCC1–3, or pBAD-YFP-PopZ plasmids, were cultivated overnight at 37 °C and inoculated in fresh LB medium at a ratio of 1:100 (v/v). Cells were grown to an OD600 of 0.4 and then induced with 0.5 mM IPTG and 5 mM L-arabinose for 2 h before imaging. The results demonstrated that PodJ/PopZ does not co-localize with IbpA, indicating that PodJ/PopZ accumulation did not form inclusion bodies in E. coli, whereas PodJCC1–3 did (Supplementary Fig. 5c).
To further test the possibility of aggregation of PodJ variants in cells, we co-expressed YFP-PodJ truncations together with a soluble protein marker mCherry in E. coli. The recombinant cells containing the pBAD-mCherry plasmid, or those co-transformed with pBAD-mCherry and pCDF-YFP-PodJ, pCDF-YFP-PodJ_N, pCDF-YFP-PodJCC1–3, pCDF-YFP-PodJCC4–6, or pCDF-YFP-PodJIDR plasmids, were induced, prepared, and imaged as described above. The results demonstrated that a high proportion of aggregates formed when expressing PodJCC1–3 in E. coli (Supplementary Fig. 5b).
Time-lapse imaging of PodJ during the cell cycle in C. crescentus
For PodJ imaging throughout the C. crescentus cell cycle, the C. crescentus wild-type strain containing a sole copy of sfgfp-podJ in the chromosome (NA1000 ∆podJ, xylX::Pxyl-sfgfp-podJ) was cultivated in M2G medium and induced by adding 0.003% (w/v) xylose 1 h before cell synchronization. Swarmer cells were isolated from the culture by centrifugation (20 min at 11,000 ×g, 4 °C) after mixing with 1 volume of Percoll (GE Healthcare). The synchronized swarmer cells expressing sfGFP-PodJ were then immobilized on a 1.5% (w/v) agarose-PYE pad containing 20 µg ml−1 kanamycin and 0.003% (w/v) xylose and imaged every 2 min using a Nikon Eclipse Ti2-E time-lapse imaging system over 1–2 cell divisions at room temperature (~4 h). The fluorescence intensity and cell length were quantitatively analyzed using MicrobeJ66.
Kymographs of fluorescence intensity were acquired using the built-in kymograph function of MicrobeJ66. The background signal was subtracted before the kymograph analysis, and the observation of the stalk at the cell pole of C. crescentus was considered as the old cell pole. A pre-division cell was selected as the starting point in Supplementary Fig. 1a. Another round of kymograph analysis was performed after the first cell division.
Transmission electron microscopy (TEM)
For PodJ condensate visualization in living cells, the C. crescentus NA1000 xylX::podJ strain or recombinant cells of E. coli containing YFP-PodJ expression plasmids were induced and prepared as described above. Cells were fixed with 2.5% (w/v) glutaraldehyde in 0.1 M phosphate-buffered solution (pH 7.4, PBS) overnight at 4 °C. The cells were subsequently washed with PBS buffer and dehydrated in graded ethanol or acetone solutions. After embedding in epoxide resin, 50-nm thin frozen sections were cut using a Leica UC6 ultramicrotome and mounted on carbon-coated Formvar copper TEM grids. After staining with 2% (w/v) uranyl acetate, the samples were examined using FEI Tecnai Spirit BioTWIN electron microscopy at an operating voltage of 200 kV. Images were obtained using a Gatan 832 CCD camera.
Estimation of polar PodJ concentration
Quantitative genome-wide protein measurements revealed that there were 1747 molecules of PodJ per C. crescentus cell when grown to the mid-log phase in liquid PYE medium39. According to the quantification of intracellular sfGFP-PodJ in Supplementary Fig. 1a, we assumed that ~60% of the total PodJ protein accumulated at the new cell pole of C. crescentus, which has a hemispherical shell with a radius of approximately 100 nm. The protein concentration of PodJ at the cell pole was estimated based on Eq. (5):
where C is the protein concentration of PodJ at the new cell pole, N is the PodJ protein molecule number at the cell pole, NA is the Avogadro constant (~6.022 × 1023), π is the mathematical constant (~3.14159), and R is the radius of the hemispherical shell (~100 nm). Based on this expression, C was calculated to be approximately 0.79 mM, a concentration that is about 300-fold higher than the minimum concentration of PodJ LLPS in vitro.
Screening of PodJ client proteins by co-localization experiments in E. coli
To screen for the client proteins of PodJ, E. coli BL21(DE3) was used because it does not contain any homologous polarity proteins of C. crescentus. In total, 23 cell cycle- or polarity-related proteins were selected from the C. crescentus localisome49 (Supplementary Table 1). The expression plasmids of the tested proteins were constructed using a fluorescent tag within the pBAD or pACYC vector, while those of fluorescent-tagged PodJ proteins were constructed based on the pCDF or pBAD vector (Supplementary Table 4).
To examine the subcellular localization of the tested protein, the expression plasmid was transformed into E. coli BL21(DE3) and induced by 0.1 mM IPTG for the pACYC-derived vectors, and 5 mM L-arabinose for the pBAD-derived vectors at 37 °C for 2 h. To examine the subcellular localization of the tested protein in the presence of PodJ, the expression plasmids of the tested protein and PodJ were co-transformed into E. coli BL21(DE3) and induced by the addition of 0.1 mM IPTG and 5 mM L-arabinose simultaneously. To further determine the interaction domain in PodJ, the tested protein was co-expressed with PodJ variants instead of full-length PodJ. Cells were prepared and imaged with a Nikon Eclipse Ti2-E inverted fluorescence microscope. Fluorescence of GFP/YFP/CFP was detected using the FITC filter (Nikon, excitation filter 475/35, dichroic mirror 499, and emission filter 530/43), and fluorescence of mCherry (mChy) was detected using the TRITC filter (Nikon, excitation filter 542/20, dichroic mirror 570, and emission filter 620/52) or the Texas Red filter (Nikon, excitation filter 555/35, dichroic mirror 585, and emission filter 630/70). The fluorescence intensity along the cell length was quantitatively analyzed using MicrobeJ66.
We used strict criteria to determine if a tested protein was recruited by PodJ or PodJ variants: (I) the localization pattern of the tested protein changed after co-expression with PodJ or PodJ variants; (II) the two proteins were 100% co-localized in >90% E. coli cells. Failure to meet either of these two criteria meant that the tested protein was not directly recruited by PodJ proteins, or the recruitment was uncertain in E. coli. At least 200 cells were calculated for each test set.
Time-lapse imaging of the recruitment processes of PodJ client proteins in E. coli
To verify the recruitment of client proteins (PleC, CpaE, FliG) by PodJ, we examined the dynamic profiles of client subcellular localizations with PodJ induction in E. coli. The expression plasmids of the tested client proteins and PodJ were co-transformed in E. coli BL21(DE3). Cells were first induced by 5 mM L-arabinose at 37 °C for client protein expression for 2 h. Then, the client-expressing cells were immobilized on a 1.5% (w/v) agarose-LB pad containing 0.5 mM IPTG (for YFP-PodJ induction) and 5 mM L-arabinose, and imaged every 5 min using a Nikon Eclipse Ti2-E time-lapse imaging system at 37 °C (~4 h). The induction of YFP was used as a negative control.
Assessment of the recruitment of PodJ client proteins in C. crescentus
To analyze the PodJ recruitment of PleC, CpaE, and FliG in C. crescentus, recombinant strains (NA1000, pBVMCS6-Pvan-pleC-mCherry; ∆podJ, pBVMCS6-Pvan-pleC-mCherry; ∆podJ, xylX::Pxyl-sfgfp-podJ, pBVMCS6-Pvan-pleC-mCherry for PleC; NA1000, pBVMCS6-Pvan-mCherry-cpaE; ∆podJ, pBVMCS6-Pvan-mCherry-cpaE; ∆podJ, xylX::Pxyl-sfgfp-podJ, pBVMCS6-Pvan-mCherry-cpaE for CpaE; NA1000, pBVMCS6-Pvan-mCherry-fliG; ∆podJ, pBVMCS6-Pvan-mCherry-fliG; ∆podJ, xylX::Pxyl-sfgfp-podJ, pBVMCS6-Pvan-mCherry-fliG for FliG) were constructed. Overnight cultures of recombinant cells were transferred into fresh M2G medium at a ratio of 1:10 (v/v) and grown to an OD600 of 0.7. The cells were then induced with 50 µM vanillate or 0.003% (w/v) xylose plus 50 µM vanillate at 30 °C for 3 h to express the client protein alone or the co-expression with PodJ. Cells were imaged with a Nikon Eclipse Ti2-E inverted fluorescence microscope as described above. The fluorescence intensity and the localization of fluorescent focus along the cell position were quantitatively analyzed with MicrobeJ66.
Screening of the negative regulator for PodJ subcellular localization in E. coli
To screen for the negative regulator of PodJ subcellular localization, 11 polarity proteins (Supplementary Table 2) that reside at the C. crescentus cell poles were selected from the C. crescentus localisome49 and their effects were examined with PodJ subcellular localization. The expression plasmids of these candidate proteins were constructed and expressed with or without fluorescent-tagged PodJ in E. coli BL21(DE3). Cells were prepared and imaged with a Nikon Eclipse Ti2-E inverted fluorescence microscope as described above. The fluorescence intensity along the cell length was quantitatively analyzed with MicrobeJ66.
The localization pattern of PodJ was observed after co-expression with these candidate proteins. Since PodJ alone is in a bipolar pattern in E. coli, a negative regulator was defined as the protein that inhibits or damages the bipolar localization of PodJ in E. coli.
Analysis of SpmX-PodJ interaction in C. crescentus
To verify the protein-protein interaction between PodJ and SpmX in C. crescentus, three sfGFP-PodJ expressing recombinant strains (NA1000, xylX::Pxyl-sfgfp-podJ, i.e., wild-type strain; ∆spmX, xylX::Pxyl-sfgfp-podJ, i.e., ∆spmX strain; NA1000, xylX::Pxyl-sfgfp-podJ, pBVMCS6-Pvan-mCherry-spmX, i.e., SpmX O/E strain) were constructed. The localization of PodJ with or without SpmX expression was investigated by inducing cells with 0.003% (w/v) xylose or 0.003% (w/v) xylose plus 500 µM vanillate, and imaged as described above. The fluorescence intensity along the cell length was quantitatively analyzed with MicrobeJ66.
Time-lapse imaging of the dynamic SpmX-PodJ interaction in C. crescentus
To better understand the SpmX regulation on PodJ in vivo, we examined the dynamic profile of PodJ condensates by inducing the cells with a single copy of mCherry-spmX. A sfGFP-PodJ and mCherry-SpmX co-expressing recombinant strain (NA1000, xylX::Pxyl-sfgfp-podJ, vanA::Pvan-mCherry-spmX) was constructed. For the titration of sfGFP-PodJ with mCherry-SpmX, the cells were first induced by 0.03% (w/v) xylose at 30 °C for 3 h for PodJ expression. The cells were then immobilized on a 1.5% (w/v) agarose-PYE pad containing 0.03% (w/v) xylose and different concentrations of vanillate (0, 50, 500, or 5000 µM, for mCherry-SpmX induction), and imaged every 5 min using a Nikon Eclipse Ti2-E time-lapse imaging system at 30 °C (~4 h).
For the adverse titration of mCherry-SpmX, the cells were first induced by 50 µM vanillate at 30 °C for 3 h for mCherry-SpmX expression. The cells were then immobilized on a 1.5% (w/v) agarose-PYE pad containing 50 µM vanillate and different concentrations of xylose (0, 0.003%, 0.03%, or 0.3% (w/v), for sfGFP-PodJ induction), and imaged as described above. The cell pole fluorescence intensity was quantitatively analyzed with MicrobeJ66.
Statistics and reproducibility
All experiments were repeated at least three times independently. No statistical method was used to predetermine sample size and no data were excluded from the analyses. All statistical analyses were performed using the GraphPad Prism 5.0 program. Statistically significant differences were determined using Welch’s or Student’s t-test, or a one-way or two-way analysis of variance with Tukey corrections as indicated. Data are presented as means ± standard error of the mean with the number of experimental replicates (n) provided in the figures or corresponding figure legends. P < 0.05 was considered significant.
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.