Proteins with a predicted function related to PHB in C. crescentus

A previous work confirmed the synthase activity of the predicted PhaC protein encoded by CC_1380 in C. crescentus. This study also showed that C. crescentus accumulates PHB (Qi & Rehm, 2001). As previously noted for C. crescentus and other α-proteobacteria (Rehm, 2003), the phaC gene is not in the same genome position as other PHB-related genes (Figure 1a). A bioinformatic analysis of the C. crescentus genome showed the presence of a putative phasin (PhaP, CC_2160) and a depolymerase (PhaZ, CC_0250) encoding genes in independent operons located in different positions of the genome. In a different genome region, the genes coding for the PhaA and PhaB enzymes (CC_0510 and CC_0511 respectively) are located next to a transcriptional regulator predicted to code for PhaR (CC_0509). A previously published list of transcriptional start sites (McGrath et al., 2007) indicates the presence of a transcriptional start site upstream of phaA, suggesting that phaA and phaB are transcribed from the same promoter. The gene downstream of phaB (CC_0512) consists of an uncharacterized conserved domain that is only present in bacteria closely related to C. crescentus, CC_0512 is followed by a putative transposable element. An examination of the genomic synteny (Oberto, 2013) of CC_0512 revealed that it is frequently downstream of phaB, suggesting a possible relation with PHB metabolism. However, this gene is also frequently together with genes involved in different metabolic pathways. Our bioinformatic search did not reveal the presence of any additional phasins or alternative PHB synthases.

Proteomic analysis of isolated native carbonosomes

To identify other proteins that could be associated with the C. crescentus PHB granules, we decided to isolate native carbonosomes and to identify the proteins present by proteomic analysis. Staining C. crescentus cells with Nile Red from a culture grown to an OD₆₆₀ of 0.6 in low phosphate conditions (M5GG medium), indicated the presence of several granules per cell. Fractionation in a sucrose gradient of a cell lysate of cells grown in this condition resulted in two bands (Figure 1b) that were independently recovered. The protein composition of these samples was analyzed by Tris-tricine SDS-PAGE (Figure 1c), the more evident difference between these fractions is the clear enrichment in the top fraction of a small protein with an apparent molecular weight similar to that of the predicted phasin PhaP (15 kDa), suggesting that this fraction contains the PHB granules. Another protein enriched in this fraction had a molecular weight similar to that of the predicted depolymerase PhaZ (47 kDa). Electron microscopy images showed that the better-defined top fraction was mainly composed of granules with an average diameter (shortest axis in case of irregular shape) of 267 ± 46 nm (granule size may change due to sample preparation for electron microscopy) and some membrane contamination, while the more diffuse second fraction consisted mainly of membrane debris (Figure 1d, top and bottom respectively). These results together suggest that the top fraction is enriched with PHB granules. The proteins present in the top fraction were identified by two-dimensional liquid chromatography coupled to mass spectrometry (MudPIT), this resulted in 158 possible carbonosome-associated proteins (Figure 2 and Table S1). Among the proteins identified were PhaC and the predicted PhaZ and PhaP proteins (Figure 2, shaded proteins). Carbonosome-associated proteins are generally predicted as soluble or in some cases have a hydrophobic region that can be predicted as a transmembrane helix. To reduce the number of candidate proteins, we predicted the subcellular location of the identified proteins (Figures 1e and 2). Most of the proteins identified were predicted as integral inner membrane proteins, indicating that the strongest contamination of the sample came from this membrane. This analysis allowed us to discard 55 candidates predicted to be periplasmic or outer membrane proteins. Conserved domain analysis of the identified proteins did not reveal any other protein with a function directly related to PHB metabolism (Figure 2).

PhaH is a new type of carbonosome-associated protein

As a first step to identify possible PGAPs, from the remaining list of 103 proteins, we discarded those for which a clear function not related to PHB could be assigned by their conserved domains composition. A putative function could be assigned for the majority of the proteins identified in the proteomic analysis of the purified carbonosomes so we decided to further investigate 9 candidates, 6 predicted cytoplasmic and 3 putative transmembrane proteins (Figure 2, red dots and Table 1). Of these only CC_3441 has a clear enzymatic domain, but was included as a candidate because, similarly to the PhaM and PhaF phasins of C. necator and P. putida (Galan et al., 2011; Pfeiffer et al., 2011), the C-terminal region of CC_3441 is mainly composed of alanine, lysine, and proline residues that probably arrange in a AKP helix. To evaluate if any of the candidate proteins associates with the PHB granules, we examined the localization of fluorescent fusions of these proteins (Figure 3a and Figure S1). For this, C-terminal fusions with the fluorescent YFP protein were generated. All the gene fusions were expressed from their native promoter and as single copy, with the exception of CC_3559 that was expressed as a second copy. Of all the proteins tested, only CC_1098 showed a strong localization in cytoplasmic foci similar to that observed in cells stained with Nile Red to visualize the PHB granules (see Figure 4b WT for an example). Proteins CC_0547 and CC_3559 showed a localization in foci that were not present in all the cells and the fusion of CC_3441, that has a probable AKP helix, showed a possible membrane localization. We tested the localization of all the protein fusions in high phosphate growth conditions but only CC_1098-YFP showed a localization similar to that observed in Nile Red stained cells. In a Western blot against YFP using α-GPF antibodies (Figure 3b), we detected the presence of all the proteins although some of them showed an apparent molecular weight different than expected (see Figure 3 and Figure S2).

A bioinformatic analysis showed that CC_1098 is a cytoplasmic protein that consists of a C-terminal helix-hairpin-helix domain (HhH) and an N-terminal sequence with no conserved domain (Figure 4a). HhH domains bind DNA in a no-site-specific manner and are mainly present in proteins involved in DNA metabolism (Doherty et al., 1996). This is the first phasin described with this domain and has no other detectable sequence similarity with other phasins. As previously mentioned, phasins PhaM and PhaF have a histone-like domain that binds DNA in the same way. A secondary structure prediction of the N-terminal region of CC_1098 with Ali2D (Gabler et al., 2020) showed that similar to other phasins (Maestro et al., 2013; Maestro & Sanz, 2017; Neumann et al., 2008), it mainly consists of α-helixes. Analysis of the predicted α-helixes with HeliQuest (Gautier et al., 2008) showed the presence of a hydrophobic helix (hH, amino acids 16–49 with hydrophobicity score above 0.5 calculated with a window of 18), that could be involved in PHB binding. A nonstructured region rich in acidic residues connects the hydrophobic helix with the HhH domain. From here on we refer to CC_1098 as PhaH.

To corroborate that PhaH associates with the carbonosomes, we tested if the fluorescence from stained PHB granules colocalizes with that of the fluorescent fusion protein. Since PHB granules stained with Nile Red were fluorescent in the YFP channel, but barely detectable in the CFP channel (Figure 4b, WT panels), we constructed a phaH-CFP gene fusion. As can be observed in Figure 4b, the fluorescence of PhaH-CFP colocalized with the carbonosomes (Pearson's R value of 0.84), supporting that PhaH is a carbonosome-associated protein. To test if the interaction of PhaH with the carbonosome was mediated by its N-terminal hydrophobic domain, we replaced the wild-type phaH allele with fluorescent fusion alleles deleted of the HhH domain (phaHΔHhH-CFP, coding for PhaH_1-159-CFP) or of the hH region (phaHΔhH-CFP, coding for PhaH_160-229-CFP). While the PhaH_1-159-CFP fluorescent fusion showed a localization similar to that of the Nile Red stained granules, PhaH_160-229-CFP was mainly dispersed in the cytoplasm although a weak colocalization could be observed in some granules (Pearson's R value of 0.71 and 0.32 respectively). In a Western blot using α-GFP antibody (Figure 4c) we were able to identify a band corresponding to the fluorescent fusion although a minority band of a lower molecular weight was observed for the PhaH-CFP fusion. This result indicates that all the CFP fusions are stable and that the delocalization of PhaH_160-229-CFP is not caused by proteolysis of protein fusion. We also observed that the PhaH_(160-229)-CFP was in a lower concentration than the other two fusions, since we did not detect any smaller molecular weight bands suggestive of degradation, this could be due to the modified start of the protein resulting from the deletion of the hH coding region. These results indicate that PhaH is a phasin that is recruited to the carbonosome mainly through its hH domain, either by directly binding the polymer or mediating the interaction with another PGAP. The weak localization of PhaH_160-229-CFP fusion suggests that this region of PhaH may interact with other PGAPs.

PhaH is necessary for full PHB accumulation and maintenance

To determine the role of PhaH in the carbonosome we obtained a phaH null mutant as well as phaH truncated mutants lacking the regions coding for the hH or the HhH domain. The truncated phaH alleles were expressed from the native promoter and locus. The phaH mutant was complemented by introducing the phaH gene under the control of its native promoter in the chromosomal xylX locus, the same strategy was used to introduce a second copy of phaH in the wild type strain to test the effect of the resulting overexpression of the gene. We then determined the growth characteristics of these strains (maximal optical density, duplication time, and colony forming units) in PHB accumulating growth conditions. These parameters were initially determined in the previously reported M5GG and M6Higg media but after 12 h of growth, the pH of these media started shifting downward and upward respectively. To solve this problem we substituted the buffering agent of M5GG (Tris 20 mM pH 7) with Imidazole 20 mM pH7. This was done after testing different concentrations of Imidazole for pH stability and cell morphology, we refer to this media as M5IGG.

All the strains showed the same duplication time during the exponential growth phase that extended for the first 24 h of incubation (Figure 5a and Figures S3 and S4). After this, the OD₆₆₀ of all the strains continued to increase at a slower rate and all the strains reached their maximal optical density at around 32 h (Figure 5b, first bar of each strain), only the maximal OD₆₆₀ of the ΔphaH and the phaHΔhH mutants were significantly different from that of the wild type. While the OD₆₆₀ of the rest of strains remained stable until the last time point at 48 h, the OD₆₆₀ of the ΔphaH and the phaHΔhH mutants started to decrease after reaching their maximal OD₆₆₀ (Figure 5b second bar of each strain). Interestingly the strain that expresses only the N-terminal region of PhaH (ΔHhH) shows a wild-type phenotype, indicating that this region is sufficient to reach and maintain the same OD₆₆₀ of the wild-type. The phenotype of the phaH mutant was complemented by the expression of phaH from the xylX locus (see ΔphaH/pXH) and the overexpression of PhaH did not have an effect on the maximal optical density (see WT/pXH). It is well known that besides the number of cells, also the accumulation of PHB can modify the optical density of a culture. To determine the cause in the difference in the OD_660, we first determined the number of viable cells as the number of colony forming (CFUs) at the early stationary phase (28 h) when all the strains had the same OD₆₆₀ and at 48 h when the OD₆₆₀ of the ΔphaH and the phaHΔhH cultures had dropped (Figure 5c). No significant changes in the number of CFUs between early and late stationary phase were observed for any of the strains, indicating that the difference in maximal OD₆₆₀ is not due to a lower number of cells in the ΔphaH and the phaHΔhH mutants and that the drop in OD₆₆₀ observed for these two strains during the late stationary phase is not caused by cell death. We then quantified the amount of PHB of cultures in early stationary phase (28 h), after they reached their maximal OD₆₆₀ (40 h) and in late stationary phase (48 h). All the strains had the same amount of PHB in early stationary phase (Figure 6, 28 h). At 40 h when the strains had reached their maximal OD₆₆₀, the amount of PHB was significantly lower in the ΔphaH and phaHΔhH mutants than in the rest of the strains and showed a similar amount to that detected at the 28 h time point, while it almost doubled for the rest of the strains. In the late stationary phase, the only strains that had a lower PHB content than in the previous time point were the ΔphaH and the phaHΔhH mutants. In accordance with our CFU results, the protein content was similar for all the strains. As it was the case for the maximal OD₆₆₀, the ΔphaH mutant was complemented by expression of phaH from the xylX locus and no effect was observed by overexpression of phaH. These results indicate that PhaH is required to achieve wild-type levels of PHB accumulation to maintain the accumulated PHB. The fact that no drop in the OD₆₆₀ or in the amount of PHB was observed in the phaHΔHhH strain indicates that the PhaH N-terminal hydrophobic helix is sufficient for these activities.

PhaH can bind DNA

Our initial bioinformatic analysis of PhaH revealed only the presence of a single HhH domain, however the majority of the proteins have at least two HhH domains and it has been proposed that two HhH domains (HhH2) is the minimal functional unit (Shao & Grishin, 2000). A closer examination of the primary structure and the alphafold predicted tertiary structure of the C-terminal domain of PhaH showed the presence of an HhH2 domain (Figure 7a), suggesting that this region of PhaH should be able to bind DNA. To test this, the PhaH_54-228 protein was expressed as a 6-His fusion and purified by metal affinity chromatography (Figure 6b). Increasing amounts of this protein were incubated with a purified 550bp PCR product containing the sequence of an unrelated gene and run in a native poly-acrylamide gel (Figure 7c). As expected from the structure analysis, 6His-PhaH_54-228 was able to bind DNA and retard the migration of the DNA fragment. To further test the ability of the HhH2 domain of PhaH to interact with the DNA, we decided to determine if it was capable of binding to the nucleoid of live cells. For this, we generated a plasmid that expresses in Escherichia coli cells an N-terminal fluorescent fusion of PhaH_54-228 with mCherry (mCh-PhaH_54-228). To label the cytoplasm of the cells, the same plasmid expresses the fluorescent YFP protein. Cells expressing these two proteins were then filamented with cephalexin and then treated with chloranphenicol, since this antibiotic has been shown to induce the condensation of the nucleoids, creating significant DNA-free regions inside the cell. Cells treated in this way were then stained with DAPI to visualize their nucleoids. In these cells we could observe an accumulation of mCh-PhaH_54-228 in the same regions in which the nucleoids were present (Figure 7d, white dots). However the mCh-PhaH_54-228 was not only present in the nucleoid regions, indicating an excess of this protein or that the affinity for the nucleoid is not sufficient to maintain all the protein in a bound state. In contrast, the fluorescence of the YFP protein was evenly distributed in the cytoplasm. This result further supports the idea that the HhH2 domain of PhaH is functional, allowing it to interact with the DNA.

The HhH domain of PhaH is required for carbonosome size and organization

Phasins frequently have a role in regulating carbonosome number, size and distribution inside the cell (Maestro & Sanz, 2017; Mezzina & Pettinari, 2016). To investigate the possible role of PhaH on these characteristics, we first stained with Nile Red wild type C. crescentus cells grown to stationary phase (28 h) in M5IGG low phosphate medium (Figure 8a upper panels). These cells showed a uniform distribution of carbonosomes along the cell in what appeared to be a series of granules, frequently giving a beaded string appearance similar to that described for P. putida (Galan et al., 2011). In contrast, the cells carrying the ΔphaH mutant allele showed a less clear definition of the PHB granules, and frequent PHB free spaces (Figure 8a, white dots) that were not caused by cell constriction. The same was observed for the phaHΔhH and phaHΔHhH strains. The PHB free spaces are not due to different amounts of PHB between the strains since at this growth phase all the strains have similar levels of accumulated PHB (Figure 6). Complementation of the mutants restored the wild-type distribution of the granules (Figure 8a, lower panels). Since in the phaHΔhH and phaHΔHhH complemented strains the truncated and full version of the proteins were present, this result shows that the mutant version of the proteins does not interfere with the function of the full protein. To better understand the reason behind these two PHB distribution patterns, we obtained electron microscopy images of cells grown in the same condition and for the same time as those used for the Nile Red staining (Figure 8b), we did not include the phaHΔhH mutant strain since it shows the same phenotype of the ΔphaH strain. Likewise the fluorescent images, the electron micrographs showed that the PHB granules were linearly distributed along the cell in the majority of the WT cells. In contrast, we could frequently observe PHB-free regions and a more frequent clustering of PHB granules of smaller size in the ΔphaH and the phaHΔHhH mutants. The white regions inside the granules are probably caused by an artifact of the staining used and were present in all the strains. To better describe the effect of the absence of the HhH domain of PhaH in the size of the PHB granules, we determined the average granule area and the number of granules per cell length in the wild type and phaHΔHhH strains (Figure 8c). To avoid artifacts, only cells that had been clearly sectioned by the middle longitudinal plane were used in this analysis, this plane can be easily identified in C. crescentus cells due to the polar morphology and the presence of the base of the stalk. This analysis showed that the phaHΔHhH strain had a larger proportion of smaller PHB granules and a wider distribution of granule size than the wild type, making the size distribution between the two strains significantly different (Figure 8c, left panel). The fact that the phaHΔHhH strain has the same amount of PHB as the wild type can likely be explained by the significant increase in the average number of granules per cell length in this strain (Figure 8c, right panel). These results indicate that although the presence of the hydrophobic helix (phaHΔHhH) is sufficient to reach and maintain maximal OD₆₆₀ and PHB concentration, to maintain the PHB granule size and distribution the HhH region is also required.

Role of PhaH in carbonosome segregation

Phasins with a domain that interacts in a non-site specifically with the DNA have been involved in carbonosome segregation between the daughter cells (Galan et al., 2011; Pfeiffer et al., 2011). The HhH domain present in PhaH could interact in the same way with the DNA, suggesting a possible role in this process. Although all the phaH mutant alleles showed a modified carbonosome distribution (Figure 8), no cells without PHB were found, indicating that PhaH is not required for carbonosome segregation during division. To further test this possibility, we compared the PHB granule distribution in wild-type and ΔphaH cells when grown in high phosphate medium. Under this condition the cells synthesize lower amounts of PHB which allows the visualization of single carbonosomes by fluorescent staining. If PhaH was relevant for carbonosome segregation we expected a higher proportion of cells with more and without granules in the ΔphaH mutant than in the wild type. We visualized cells from samples of cultures at different stages of the growth curve to obtain the distribution of foci number per cell when the cells had different number of foci (Figure 9a). Cells from exponential cultures (OD₆₆₀ 0.4–0.75) showed similar distributions although an increase in fluorescent intensity was observed, for this reason we only characterized the distribution of exponential cells (OD₆₆₀ 0.55) to minimize false negatives while still being in the exponential growth phase. Exponential cells have mainly one fluorescent foci, and in both strains, there is a relevant proportion of cells without a detectable focus (Figure 9b, exponential). Cells at the onset of the stationary phase (OD₆₆₀ 1.2) showed an increase in the number of cells with two granules while the percentage of cells without signal decreased (see Figure 9b stationary). In contrast to low phosphate conditions, no change in the granule distribution pattern was observed if the cells were further incubated (24 h total time, Figure 9b, mid stationary). The brightness of the fluorescent foci increased as the cells reached the stationary phase, indicating an increment in the amount of PHB. No significant differences were observed in the carbonosome number distribution between the wild type and ΔphaH mutant strains at any of the time points, supporting the idea that PhaH is not required for carbonosome segregation.

Bacterial strains and growth conditions

Strains of E. coli were grown in LB medium with the appropriate antibiotic at 37°C (Ausubel, 1987). Plasmids were maintained and purified from E. coli DH5α or TOP10 strains. Strains of C. crescentus were grown at 30°C in PYE rich medium, Tris-based M5GG medium (20 mM Tris-HCl pH 7, NaCl 1 mM, KCl 1 mM, NH₄Cl 0.05%, Sodium glutamate 1 mM, MgSO₄ 0.5 mM, Glucose 0.2%, FeSO₄ EDTA chelated 0.01 mM, CaCl₂ 0.5 mM) supplemented with 0.05 mM Na/K phosphate buffer pH 7 (Domian et al., 1997; Ely, 1991; Jacobs et al., 2001) or in M5IGG, with the same composition as M5IGG but with Imidazole 20 mM pH 7 as buffer instead of Tris-HCl. Antibiotics were used at the following concentrations (μg/mL) for E. coli: gentamycin, 20; kanamycin, 50; spectinomycin, 50; nalidixic acid, 20; chloramphenicol, 30. For C. crescentus, antibiotics were added at the following concentrations (μg/mL) for liquid and solid media, respectively: gentamycin, 2 and 5; kanamycin, 5 and 20; spectinomycin, 15 and 100; chloramphenicol 2 and 5. Strains and plasmids are listed in Table 2, plasmids and primers are listed in Table 3.

Molecular biology techniques and Western blot

Standard methods were used to purify and analyze chromosomal or plasmid DNA (Ausubel, 1987). Restriction enzymes and T4 DNA ligase were bought from New England Biolabs or Invitrogen. Polymerase chain reactions were done with TaKaRa PrimeSTAR-HS enzyme with high GC buffer. For Western blots, SDS-PAGE resolved proteins were transferred to nitrocellulose membranes that were then incubated with a mouse polyclonal α-RFP antibody raised against 6His tagged mRFP (Ginez et al., 2014) or a commercial polyclonal α-GFP antibody (Clontech). For detection, an alkaline phosphatase-conjugated α-mouse IgG antibody (Sigma) was used together with CDP-Star/Nitro-Block substrate. Samples were collected at OD₆₆₀ 0.6 or 1.2, the cells from 1.5 mL of the culture were harvested and resuspended in 100 μL of lysis buffer containing SDS and β-mercaptoethanol. The samples were then boiled and sonicated, and 10 or 5 μL were loaded depending on the original OD₆₆₀ of the culture.

Detailed information on plasmid construction and strain selection can be found in supplementary methods in the supplementary information. Briefly, fluorescent protein fusions were obtained by either amplifying with the appropriate primers (Table 3) the 3′ terminal region of the selected gene and the product cloned in the suicide vector pYFPC-4 or pCFPC-4 or by a specific strategy. Suicide plasmid carrying a fluorescent fusion were electroporated into CB15 cells and integration of the plasmid into the gene locus was selected by the appropriate antibiotic resistance. Deletion alleles were obtained by cloning the upstream and downstream regions of the region to be deleted in the pNPTS138 suicide vector, for the SP1287 (ΔphaH::Ω^Spc) the ΩSpc cassette was then inserted between the two fragments. The resulting plasmids were then electroporated into CB15N cells and allele replacement was selected by plating on PYE plates containing 3% sucrose and spectinomycin in the case of SP1287. The mutant alleles were confirmed by PCR. To complement the different phaH mutant alleles, plasmid pXphaHprom-5 was constructed by cloning a PCR product containing the phaH coding region and 600bp up-stream of it (primers del1098F1 HindIII and 1098R1 EcoRI) in pXTCYC-5. This plasmid is not replicative in C. crescentus and can integrate in the xylX locus.

Native PHB granule isolation

Isolation of PHB granules was carried by a modification of a previously reported method (Preusting et al., 1993). A 500 mL culture medium was inoculated with an overnight culture grown in M5GG medium to an OD₆₆₀ 0.03 and grown to a final OD₆₆₀ 0.6, the cells were concentrated to a final volume of 20 mL, and mixed with two volumes of sucrose 0.5 M in Tris 10 mM pH8. Lysozyme and EDTA pH8 were added to a final concentration of 170 μg/mL and 4 mM, respectively. After the cells converted to spheroplasts (approximately 5 min), 600 μL of Tris 1 M pH8 was added (10 mM final concentration), and the spheroplasts were lysed by sonication in an ice bath. A discontinuous sucrose gradient (Tris 10 mM, pH8) was prepared (5 layers of 5 mL – 1, 1.3, 1.5, 1.7, and 1.9 M sucrose) and 5 mL of the cell lysate was put on the top. The gradients were centrifuged in a Beckman SW28 rotor at 122,000 g for 15 h. The bands were recovered, washed twice with 30 mL of Tris 10 mM, pH8 and concentrated by centrifugation at 12,000 g in a Beckman JA-20 rotor. The pellets were resuspended in 1 mL of Tris 10 mM, pH8 and stored at −70°C.

Proteomic analysis

The purified native PHB granules were subjected to MudPIT analysis (ITSI Biosciences). Briefly, the sample was precipitated, resuspended, and then reduced and alkylated. 25 μg of protein was precipitated and resuspended in 100 mM triethylammonium bicarbonate buffer and digested with trypsin. For MudPIT, five fractions eluted at 75, 150, 250, 350, and 450 mM of ammonium acetate were collected and individually analyzed by nano-LC/MS/MS. A linear acetonitrile gradient was used in the LC phase to separate the tryptic peptides based on their hydrophobicity prior to MS analysis on a Thermo Scientific LTQ XL mass spectrometer. The total run time was 150 min per sample. For MS, a data-dependent Top 5 method was used where a full MS scan from m/z 350–1700 was followed by MS/MS scans of the five most abundant ions.

Each ion was subjected to CID (Collision Induced Dissociation) for fragmentation and peptide identification. Raw data files from all 5 fractions were searched against the fasta database for C. crescentus from Uniprot using Proteome Discoverer 1.4 (Thermo Scientific) and Sequest HT algorithm. Target Decoy PSM Validator was used for PSM (peptide-spectrum match) validation in the database searches. Trypsin was the selected enzyme allowing for up to two missed cleavages per peptide in protein digestion. High-confidence peptides were used in sorting these search results. Protein filters (minimum number of unique peptide: 1 per protein, only rank 1 peptides, peptides only in top-scored proteins) were also used to refine the search results. A total of 175 proteins were identified using stringent quality parameters, and peptides that were identified with high confidence. Out of these 90 proteins were identified with 2 or more unique peptides.

The 175 identified proteins were manually annotated, domain analysis was carried out using the NCBI CD-Search database (Lu et al., 2020), protein localization was predicted with the SignalP, LipoP, and TMhMM servers (Almagro Armenteros et al., 2019; Juncker et al., 2003).

Microscopy

For fluorescence microscopy, cells were grown and stained (see below), after which a 1.5 μL sample was spotted on a microscope slide covered with a 1.5% agar bed prepared with 10 mM Na/K phosphate buffer pH 7. Epifluorescence images were acquired using a Nikon Eclipse 600 microscope equipped with a Hamamatsu Orca-ER cooled charge-coupled-device (CCD) camera, and an X-Cite 120 as light source or with a Nikon Ti microscope equipped with a SCMOS pco.edge camera, controlled with MicroManager (Edelstein et al., 2014). Fluorescence cubes were acquired from Chroma (EYFP, 49003; ECFP, 49001; mCherry, 49008). Nile Red staining was carried out on concentrated cell samples to an OD₆₆₀ of 2. A 100 μL sample was incubated with a final concentration of 0.005% of Nile Red for 5 min. Images were processed with ImageJ and quantification was done manually in ImageJ or with MicrobeTracker (Sliusarenko et al., 2011). PHB granule distributions were determined using the spotFinderZ program. Filamented cells were discarded and cells in late divisional cells were split. Granule detection was optimized for each condition. Colocalization was determined using the Coloc2 plugin for ImageJ, and Pearson values without threshold are reported.

Electron microscopy

C. crescentus pellets were processed for standard transmission electron microscopy (Nepomuceno-Mejia et al., 2016). Briefly, samples were fixed in 2.5% glutaraldehyde and 4% paraformaldehyde in PBS for 4 h, postfixed in 1% osmium tetroxide overnight, dehydrated using a series of ascending concentrations of ethanol and embedded in epoxy resin (epon 812, Electron Microscopy Science). Sample integrity was evaluated in semithin sections of about 500 nm stained with toluidine blue and observed with bright-field microscopy (Axiostar, Carl Zeiss). Ultrathin sections (50–60 nm) were mounted on formvar-coated copper grids, contrasted with 4% uranyl acetate and 0.3% lead citrate, examined at 80 kV with a JEM-1010 (JEOL) transmission electron microscope. Images were captured with a CCD camera model Gatan Orius SC600 and a digital micrograph software.

PHB quantification

Quantification of PHB was carried out by the crotonic acid assay (Law & Slepecky, 1961). Briefly, cells from 5 mL of culture were collected in 1.5 mL Eppendorf tubes, washed with 1 mL of MgSO₄ 10 mM and then digested with 1 mL of 30% sodium hypochlorite at 30°C for 1 h. The samples were then successively washed with 1 mL of deionized water, ethanol and acetone and then air-dried at 50°C. To generate crotonic acid, 0.5 mL of concentrated H₂SO₄ was added and the samples were incubated at 95°C for 30 min, with occasional mixing. Crotonic acid was quantified by serial dilutions (1:100 to 1:1000) of the samples in 0.045 N H₂SO₄. The absorbance of these samples was measured at a wavelength of 208 nm in a spectrophotometer and compared to a reference curve of PHB (Karr et al., 1983). PHB concentrations were divided by the amount of protein in the original samples quantified in a Bradford assay (BioRad).

Protein structure analysis

The original PDB file (AF-Q9A996-F1) was downloaded from the AlphaFold database (Varadi et al., 2022) and then manipulated with RasMol.

Protein purification and Electrophoretic Mobility Assay (EMSA)

Plasmid pETphaHΔhH was transformed into Rosseta strain, an exponential (DO₆₀₀ 0.6)100 mL culture in LB media at 30°C was induced with 1 mM IPTG for 3 h. The culture was concentrated and resuspended in 1 mL of 100 mM phosphate buffer pH7 EDTA and lysozyme and sonicated. The lysate was then cleared by centrifugation at 4°C and passed through a 0.5 mL resin bed (cOmplete His-Tag purification resin, Roche), washed with 10 volumes of phosphate buffer and eluted in 4 fraction of 1 mL with 40 mM Imidazole in the same buffer.

For EMSA assays the DNA (550 bp fragment) and protein were incubated for 30 min at 4°C in a final volume of 10 μL in a buffer with final concentration of 10 mM sodium phosphate buffer pH 6, 1 mM MgCl₂, 10 μg/mL BSA and 30 mM NaCl. Glycerol was added to the reactions to a final concentration of 10%. The samples were then loaded in a native 5% polyacrylamide gel made with 0.5% TBE and 5% glycerol and run with 0.5% TBE buffer at 1 V/cm²

Nucleoid binding

A 5 mL culture was inoculated with an aliquot of an overnight culture of strain BL21 carrying plasmid pDCHYphaHΔhH/YFP to an OD₆₀₀ of 0.01 and incubated at 37°C with shaking until an OD₆₀₀ of 0.1, at this point IPTG and arabinose were added to a final concentration of 1 mM and 0.1%, and incubated for an additional hour. To 1 mL of this culture, cephalexine (3 μg/mL) was added and incubated for 30 min and then chloramphenicol (600 μg/mL) was added. It should be noted that pDCHYphaHΔhH/YFP confers resistance to chloranphenicol, so a higher concentration of this antibiotic was required to induce chromosome condensation. The cells were incubated for 45 min, concentrated to 200 μL and DAPI (0.5 μg/mL) was added. The cells were incubated at room temperature for 15 min and then mounted on 1% agarose pads made with PBS.