Control of proline utilization by the Lrp-like regulator PutR in Caulobacter crescentus

Cellular metabolism recently emerged as a central player modulating the bacterial cell cycle. The Alphaproteobacterium Caulobacter crescentus appears as one of the best models to study these connections, but its metabolism is still poorly characterized. Considering that it lives in oligotrophic environments, its capacity to use amino-acids is often critical for its growth. Here, we characterized the C. crescentus PutA bi-functional enzyme and showed that it is required for the utilization of proline as a carbon source. We also found that putA transcription and proline utilization by PutA are strictly dependent on the Lrp-like PutR activator. The activation of putA by PutR needs proline, which most likely acts as an effector molecule for PutR. Surprisingly, we also observed that an over-production of PutR leads to cell elongation in liquid medium containing proline, while it inhibits colony formation even in the absence of proline on solid medium. These cell division and growth defects were equally pronounced in a ΔputA mutant background, indicating that PutR can play other roles beyond the control of proline catabolism. Altogether, these findings suggest that PutR might connect central metabolism with cell cycle processes.


Results
Caulobacter crescentus needs PutA for proline utilization. Most bacteria use a bi-functional PutA enzyme to catabolize the conversion of proline into glutamate, the first step in proline utilization 3 . In order to study how C. crescentus uses proline as a source of nutrients, we searched for a putA homolog in its genome and found the CCNA_00846 protein that shares ~47% of identity with bi-functional PutA proteins from other Alphaproteobacteria 9,14 . It encodes a protein carrying two domains: one with similarities with the PRODH domain and the other one with similarities with the P5CDH domain, as expected for a bi-functional enzyme 3 . We constructed a complete ΔputA deletion strain and found that the growth rate ( Supplementary Fig. S1) and the morphology ( Supplementary Fig. S2) of mutant cells cultivated in complex peptone-yeast extract (PYE) medium (in which amino-acids serve as the primary carbon source) were unaffected, compared to isogenic wild-type (WT) cells. We then tested its capacity to grow in minimal media with proline (M2P) or glucose (M2G) as the sole carbon source. As expected, the ΔputA and its isogenic wild-type (WT) strain could both grow at a similar rate in liquid (Fig. 1A) or solid (Fig. 1B) M2G. We also observed that the WT strain could grow on M2P, but at a lower rate than on M2G, showing that glucose is a better carbon source than proline for C. crescentus. As for the ΔputA strain, we could not detect any growth in liquid (Fig. 1A) or on solid (Fig. 1B) M2P media, showing that  PutR is a proline-dependent activator of the putA but not of the relB 1 promoters. Promoter activities in Miller Units (MU) were measured by β-galactosidase assays from cells cultivated in exponential phase and carrying the placZ290-putAP (PputA-lacZ) or placZ290-relB 1 P (PrelB 1 -lacZ) plasmids. Error bars correspond to standard deviations from minimum six independent experiments. Statistical analysis were performed using student t-tests. Stars (*) highlight significant (p < 0.0005) but also relevant differences between two strains. (A) Activity of the putA promoter in WT (NA1000) or ΔputR (JC1040) strains cultivated in PYE rich medium. (B) Activity of the putA promoter in a WT strain carrying pBXMCS6 (empty vector) or pX-putR. Cells were cultivated over-night and re-diluted into PYE + 0.2% glucose (PYEG). Once cells reached exponential phase again, 0.3% xylose was added into half of the culture (PYEGX) to induce the expression of putR from pX-putR. β-galactosidase assays were done 90 minutes after. (C) Activity of the putA promoter in WT or ΔputR strains cultivated in M2 minimal medium containing 0.2% glucose (M2G) supplemented or not with 10 mM proline. (D) Activity of the putA promoter in a WT strain carrying pBXMCS6 (empty vector) or pX-putR. Cells were cultivated over-night PutA is required for proline utilization by C. crescentus. This result strongly suggests that C. crescentus PutA can convert proline into glutamate as previously shown for PutA homologs from other Alphaproteobacteria 9,11,14,15 . PutR is critical for putA transcription and proline utilization. C. crescentus must control PutA levels or activity to make sure that it does not catabolize proline when it grows in a medium that does not contain proline, in order to maintain intracellular levels of proline that are sufficient for efficient protein synthesis. Interestingly, a gene (CCNA_00847) encoding a protein sharing ~40% of identity with PutR from other Alphaproteobacteria 9,14 lies next to the putA gene but in an opposite orientation, suggesting that it may encode a regulator of putA transcription. To test if PutR regulates putA transcription, we cloned the putA promoter region upstream of the lacZ gene and introduced this PputA-lacZ reporter into a WT, a ΔputR deletion strain and a strain over-expressing putR. We then used these strains to perform β-galactosidase assays to evaluate the impact of PutR on putA transcription in complex PYE medium. We observed that the putA promoter was ~7-fold less active in ΔputR cells than in WT cells ( Fig. 2A), reaching very basal levels that are most likely not sufficient for proline utilization. Indeed, we also observed that ΔputR cells could not grow in M2P medium containing proline as the only carbon source, while they could grow as WT cells in M2G (Fig. 1) or PYE media ( Supplementary  Fig. S1). In addition, the activity of the putA promoter was induced ~2-fold upon putR-overexpression when cells were cultivated in rich medium for 90 minutes (Fig. 2B). All together, these results show that PutR is a strong activator of putA transcription and that PutR is thus required for proline utilization by C. crescentus.
PutR is a proline-responsive activator of putA. PutR belongs to the Lrp/AsnC family of transcription factors that often use effector molecules, such as amino-acids, to modulate their activity 16 . Considering that PutR regulates the proline utilization enzyme PutA, we tested if proline may affect the capacity of PutR to activate putA transcription. We grew the WT, ΔputR and putR-over-expression strains carrying the PputA-lacZ reporter in M2G media supplemented or not with proline and performed β-galactosidase assays. In the WT strain, the activity of the putA promoter was ~8-fold higher when proline was added into the medium, showing that proline induces putA transcription (Fig. 2C). Consistent with the proposal that PutR senses proline levels, we observed that proline could not induce the putA promoter in the absence of PutR. Interestingly, proline activated the putA promoter even more efficiently (~11-fold) when putR was over-expressed (Fig. 2D). Overall, these observations show that proline is required for the activation of putA transcription by PutR.
The overproduction of PutR in proline-containing medium leads to cell elongation in a PutA-independent manner. Surprisingly, we observed that the overexpression of putR under the control of the xylX promoter on the medium-copy number vector pBXMCS6 leads to cell morphology defects. Wild-type cells over-expressing putR appeared significantly elongated compared to control cells carrying the empty vector, but only when cultivated in the presence of proline in minimal medium (Fig. 3) or in complex medium ( Supplementary Fig. S3). This observation suggests that PutR inhibits cell division in liquid cultures. Since this phenotype is dependent on the presence of proline, we hypothesized that it might be connected with proline catabolism due to excessive transcription of putA when PutR is over-produced. Indeed, previous studies indicated that proline catabolism produces hydrogen peroxides, which can influence redox homeostasis in certain bacterial species 5,6 . Considering that oxidative stress may lead to DNA or protein damage and that such damage can lead to cell cycle arrests 32,33 , it could have been an explanation for the phenotype of cells over-expressing putR. To test this hypothesis, we looked at the effect of PutR overproduction in ΔputA cells that cannot catabolize proline. We observed that the cell division defect was similar in wild-type and ΔputA cells (Fig. 3), indicating that PutR can inhibit cell division independently of its function as a transcriptional activator of putA.
The overproduction of PutR inhibits colony formation in a PutA-independent manner. In addition to the observed cell division defect, we noticed that wild-type cells over-expressing putR while cultivated on plates could rarely form colonies (Fig. 4). When growth could nevertheless be detected, colonies appeared as much smaller than colonies of cells carrying the empty vector. Interestingly, this additional phenotype associated with the over-production of PutR was neither dependent on the presence of proline in the medium, nor on the capacity of cells to catabolize proline using PutA. Thus, it appeared as independent of the effect of PutR on putA regulation or on cell division.
PutR and proline levels do not influence relBE1 expression. Interestingly, we noticed that the relBE 1 genes, encoding a functional toxin-antitoxin (TA) module 34 , are located right after the putA gene on the C. crescentus genome (Fig. 5A). In addition, an RNA-sequencing experiment suggested that putA and relBE 1 may belong to the same polycistronic transcript 35,36 . Then, an excess of PutR might result in an excessive transcription of relBE 1 , potentially leading to RelE 1 -mediated toxicity on plates or to cell division defects. It is however important to mention that the relBE 1 genes are also transcribed from their own promoter located just upstream of the relB 1 gene (Fig. 5A) 34 . To test if this second promoter is dependent on PutR and/or proline like the putA promoter, we constructed a transcriptional fusion between the relB 1 promoter and the lacZ gene, and introduced this construct into wild-type and ΔputR strains. These strains were then cultivated in M2G medium supplemented or not with proline and promoter activity was evaluated by β-galactosidase assays. We found that the relB 1 promoter was in M2G and re-diluted into M2G supplemented or not with proline. Once cells reached exponential phase again, 0.3% xylose was added into half of the culture to induce the expression of putR from pX-putR. β-glactosidase assays were done 180 minutes after. (E) Activity of the relB 1 promoter in WT or ΔputR strains cultivated in M2 minimal medium containing 0.2% glucose (M2G) supplemented or not with 10 mM proline.
ScIeNtIFIc REPORTs | (2018) 8:14677 | DOI:10.1038/s41598-018-32660-3 neither dependent on PutR, nor on the presence of proline in the culture medium (Fig. 2E). If relBE 1 and putA do belong to the same operon 35 , it was however still possible that relBE 1 transcription might be affected by PutR and proline through their regulation of the putA promoter (Fig. 2C,D). To test this possibility, we performed quantitative reverse transcription PCR (qRT-PCR) with four different probes allowing us to distinguish transcripts carrying putA, relB 1 , relE 1 or the intergenic region between putA and relB 1 (IG) independently (Fig. 5A). First, these experiments confirmed that putA transcription is strongly dependent on PutR and proline (Fig. 5B), as previously shown using transcriptional reporters (Fig. 2C,D). Second, it showed that the IG region is much less transcribed than the putA gene in all tested conditions (Fig. 5B), indicating that transcription starting at the putA promoter mostly ends before the relB 1 gene (Fig. 5A). All together, these qRT-PCR results demonstrate that the relBE 1 TA  . Excess of PutR inhibits colony formation and growth in a proline-independent manner. NA1000 (WT) and isogenic ΔputA cells, or CB15 (WT') and isogenic ΔrelBE 1 cells, carrying the pBXMCS6 (empty vector) or the pX-putR plasmid were cultivated in M2G and diluted the next morning into M2G supplemented or not with 10 mM proline. Once cultures reached stationary phase, 10-fold serial dilutions in M2G medium were prepared and 5 μL of each dilution was spotted onto M2GA plates containing 0.3% xylose (M2GAX) to induce the expression of putR from pX-putR over-night. Proline was also added into plates when indicated. Representative images from three independent experiments are shown in this figure.
module is only transcribed from the PutR-independent relB 1 promoter under standard growth conditions. Then, it is unlikely that the overproduction of PutR will affect relBE 1 expression, leading to defects in colony formation (Fig. 4) or in cell division (Fig. 3). Supporting this proposal, we observed that the over-production of PutR in a relBE 1 deletion strain still inhibits colony formation on solid media (Fig. 4) and cell division in liquid media (Fig. 3). All together, these results suggest that PutR has other targets than the putA promoter and that PutR may not always need proline as an effector.

Discussion
This study aimed at characterizing proline catabolism and its control in C. crescentus. Such a control is critical to ensure that cells can use proline as carbon or nitrogen sources, but that they do not degrade proline when endogenous levels are too limiting. This is all the more important for environmental bacteria like C. crescentus that thrive in oligotrophic environments where amino-acids often serve as major carbon sources. Also, the catabolism of proline by PutA enzymes can generate hydrogen peroxide, which can be deleterious to cells [4][5][6][7] . We showed that the PutR regulator plays a key role in this process and unexpectedly also discovered that it may affect other PutA-independent cellular functions.
The putA gene of C. crescentus encodes a bi-functional proline dehydrogenase (Figs 1 and 6). Although indispensable for growth on minimal media containing proline as the only carbon source (Fig. 1), we observed that the absence of PutA does not affect growth or cell morphology in complex PYE media where amino-acids serve as major carbon sources ( Supplementary Figs S1 and S2). However, C. crescentus may frequently find conditions where proline becomes an important source of nutrients in its natural oligotrophic environment. Indeed, Alphaproteobacteria have been shown to be often exposed to environments where proline catabolism by PutA becomes indispensable, such as during nodulation or infection by Sinorhizobium meliloti and B. abortus, respectively 8,9 . As in many Alphaproteobacteria, the putA gene of C. crescentus is located immediately next to a gene that is divergently transcribed and which encodes a putative DNA binding protein of the Lrp/AsnC family. We found that this gene encodes a strong activator of putA transcription in the presence of proline (Figs 2 and 5) and thus named it PutR for proline utilization regulator. Without PutR, the levels of PutA are so low that C. crescentus can no more use proline as a carbon source (Fig. 1). The fact that PutR requires proline to activate putA transcription (Figs 2 and 5) suggests that proline acts as an effector molecule to stimulate PutR activity. Transcription factors from the AsnC/Lrp family often require such effector molecules to bind to some of their promoter targets, most likely through conformational changes 16 that may also take place with C. crescentus PutR.
Our observation that the putA promoter is efficiently activated by proline but still efficiently repressed without proline (Fig. 2C), suggests that this promoter could be used as a useful genetic tool for inducible protein expression in C. crescentus cells cultivated in minimal media. Compared to the two other promoters currently available for such applications (from the xylX and vanA genes) 37 , the putA promoter appears to have properties similar to the xylX promoter 38 in terms of efficiency of induction and repression. Moreover, the putA promoter does not respond to xylose levels (Fig. 2B), making the combined use of xylX and putA promoters, to induce the expression of two different proteins, potentially feasible.
Interestingly, several members of the Lrp/AsnC family of transcription factors are so-called global regulators controlling large regulons. The best characterized example is the leucine-responsive protein Lrp of E. coli that directly regulates ~130 genes encoding proteins involved in various processes such as metabolism, pili synthesis or adhesion to host cells [39][40][41] . Other members of this family are more specific regulators, like PutR homologs from a few other Alphaproteobacteria such as A. tumefaciens, S. meliloti and R. capsulatus 9,11,14,15 . In C. crescentus, we observed that an overproduction of PutR leads to cell division and colony formation defects, which are independent of the activation of putA by PutR (Figs 3 and 4). This observation indicates that PutR may have other activities in C. crescentus, despite no apparent defects in growth or cell morphology for ΔputR cells cultivated with or without proline ( Fig. 1 and Supplementary Figs S1 and S2). Future experiments should notably aim at determining under which conditions C. crescentus may use this other activity to block or slow down cell division in response to proline levels. Interestingly, recent observations made comparing the phenotypes of ΔputA and ΔputR strains of B. abortus also lead to conclude that PutR may have additional functions beyond simply regulating putA transcription, such as adaptation to oxidative stress 9 .
Interestingly, we observed that the inhibition of cell division by PutR was dependent on proline (Fig. 3), while the inhibition of colony formation on solid media was not (Fig. 4). Also, the proline and PutR-dependent inhibition of cell division (Fig. 3) in liquid media is not associated with a drop of cell viability as shown by live/dead staining assays and microscopy ( Supplementary Fig. S4). Then, PutR may carry minimum two other functions unrelated with putA activation, with only one of these that is dependent on proline (Fig. 6). Another possibility is that the putA-independent target(s) of PutR have more impact when C. crescentus is cultivated on solid rather than in liquid media. Importantly, we ruled out the possibility that PutR inhibits cell division through a polar effect on the transcription of the downstream relBE 1 genes (Figs 3 and 5) encoding a functional toxin-antitoxin system in C. crescentus 34 . Then, how PutR can inhibit cell division in liquid medium or colony formation on plates through direct or indirect effects remains to be clarified (Fig. 6). As a DNA binding protein, it may bind to other promoter regions to regulate the transcription of other genes or, alternatively, have more general effects on chromosome condensation as other members of the Lrp/AsnC family of transcription factors 16,18 . Consistent with this last possibility, results from flow cytometry experiments looking at the DNA content of elongated C. crescentus cells over-expressing PutR, suggest that the elongation of DNA replication may be slowed down or frequently arrested in these cells (Supplementary Fig. S5). This may then interfere with cell division through SOS-dependent or SOS-independent mechanisms that tend to block late steps of the cell division process in C. crescentus 32,33 . In agreement with this proposal, we found that the FtsK core divisome protein could still localize at mid-cell in elongated cells over-expressing PutR ( Supplementary Fig. S3), suggesting that cell constriction, rather than core divisome assembly is inhibited when PutR is over-expressed. Considering that the proteolysis of the CtrA response regulator, which regulates the initiation of DNA replication and the transcription of several cell division proteins in C. crescentus 42 , is induced by proline uptake in the intracellular Alphaproteobacterium Ehrlichia chaffeensis 43 , we also checked the intracellular levels of CtrA in elongated C. crescentus cells over-expressing PutR, but observed no significant difference compared to control cells (Supplementary Fig. S6). Then, although a common point between C. crescentus and E. chaffeensis is that proline can affect their cell cycles, it appears to do so using different mechanisms in different Alphaproteobacteria. Our findings on PutR are reminiscent of so-called "moonlighting" proteins that are multifunctional proteins that perform multiple autonomous functions without partitioning these functions into different protein domains 44 . Interestingly, moonlighting proteins are the most often involved in the glycolytic pathway or in the TCA cycle. An example found in Alphaproteobacteria is the GdhZ moonlighting enzyme that not only converts glutamate into α-ketoglutarate, but also inhibits FtsZ polymerization to modulate cell division in response to glutamate levels 21,45 . Although an excess of PutR in cells exposed to high proline levels may lead to an imbalance in glutamate intracellular levels through PutA, we ruled out the possibility that this is responsible for the observed cell division defect since PutR-overexpressing cells lacking putA were still elongated (Fig. 3).
Altogether, one of our surprising finding was that C. crescentus PutR may connect the catabolism of proline with other cellular functions such as cell division, but in a PutA-independent manner (Fig. 6). Recent discoveries also highlighted other unexpected connections between cell division and the central metabolism in diverse bacteria 25 . Beyond the example of the GdhZ enzyme 21 , it has, for example, been shown that intracellular α-ketoglutarate levels can influence peptidoglycan synthesis and cell division 22 or that the transcription of the ftsZ cell division gene is influenced by the glutamine-sensitive phosphotransfer system (PTS Ntr ) 23,24 in C. crescentus. Also, the pyruvate metabolite has been shown to modulate Z-ring assembly for the division of the Gram-positive Bacillus subtilis bacterium 46 . Such connections may play an important role in coordinating bacterial growth with cell division, potentially participating to cell size homeostasis 26 .

Materials and Methods
Bacterial strains and growth conditions. C. crescentus strains used in this study are described in Table 1.
The TOP10 E. coli strain (Invitrogen, USA) was used for plasmid constructions. Plasmid constructions. Plasmids used in this study are described in Table 1 and their construction is described below. All PCR amplifications were done using NA1000 genomic DNA and the KOD hot start DNA polymerase (Merck, Germany).
Construction of placZ290-putAP and placZ290-relB1P. The promoter regions of putA (445 base pairs upstream of the annotated translational start site) or relB 1 (568 base pairs upstream of the annotated translational start site) were amplified using primers described in Supplementary Table S1, carrying EcoRI, PstI or XbaI restriction sites. PCR products and the placZ290 vector were digested by EcoRI and XbaI or PstI. Promoters were then ligated into the vector, giving placZ290-putAP and placZ290-relB 1 P.
Construction of pX-putR and pRX-putR. The putR coding sequence (464 base pairs) was amplified using primers described in Supplementary Table S1, carrying NdeI, EcoRI or NheI restriction sites. The amplified putR sequence and the pBXMCS6 or the pRXMCS6 vectors were digested by NdeI and NheI or EcoRI. The putR sequences and the vectors were then ligated together, giving pX-putR and pRX-putR. With these plasmids, putR expression is controlled by the xylX promoter 31 , which is induced in the presence of xylose in the medium and repressed in the presence of glucose in the medium.
Construction of pNPTS138-putA, pNPTS138-putR and pNPTS138-putR:Ω. From 500 to 730 base pairs upstream (UP) and downstream (DOWN) of the putA and putR coding sequences were amplified using primers described in Supplementary Table S1. PCR products were digested by EcoRI, BamHI and NheI (putA) or PstI, BamHI and NheI (putR). Digested UP and DOWN sequences were then both ligated into a pNPTS138 vector digested with the same enzymes, giving pNPTS138-putA and pNPTS138-putR. An Ω cassette extracted from pBOR digested by BamHI and encoding Spec/Strep resistances was then introduced at the BamHI restriction site between the UP and the DOWN sequences of putR, giving pNPTS138-putR:Ω.
Strain constructions. Plasmids were introduced into NA1000 or CB15 C. crescentus strains as published in 47 .
To construct the ΔputA deletion strain, the pNPTS138-putA plasmid was integrated into the C. crescentus chromosome at the putA locus by single homologous recombination, selecting for Km-resistant colonies. The resulting strain was grown to stationary phase in PYE medium lacking Km. Cells were plated on PYEA + sucrose 3% and incubated at 28 °C or 30 °C. Single colonies were picked and transferred in parallel onto PYEA plates containing or not Km. Km-sensitive clones, which had lost the integrated plasmid due to a second recombination event were isolated. Colonies were further tested for their capacity to grow on M2GA medium but not on M2PA medium.The ΔputA deletion was then verified by colony-PCR using specific primers.
To construct the ΔputR deletion strain, the pNPTS138-putR:Ω plasmid was integrated at the putR locus of the chromosome of a C. crescentus NA1000 strain by single homologous recombination, selecting for Km-resistant colonies. The pRX-putR plasmid was then introduced into this strain. The resulting strain was grown to stationary phase in PYE medium supplemented or not with 0.3% xylose and Cam, but lacking Km. Cells were plated on PYEA + sucrose 3% + xylose 0.3% or 0.1% + Cm and incubated at 28 °C. Single colonies were picked and transferred in parallel onto PYEA + xylose 0.3% + Cam + Strep + Spec or onto PYEA + xylose 0.3% + Cam + Km. Km-sensitive and Spec/Strep-resistant clones, which had lost the integrated plasmid due to a second recombination event were isolated. These were supposedly ΔputR::Ω. The ΔputR::Ω deletion was then verified by ScIeNtIFIc REPORTs | (2018) 8:14677 | DOI:10.1038/s41598-018-32660-3 colony-PCR using specific primers and transduced into a NA1000 WT strain using ΦCR30 phages 48 selecting for Strep/Spec-resistances, giving the so-called ΔputR strain. β-galactosidase assays. β-galactosidase assays were done using standard procedures as previously described 49 . RNA extraction. Total RNA were extracted from cultures at an OD 660 of 0.6, using the RNeasy miniprep kit (Qiagen, GER) including the recommended DNase I treatment. A second DNase I treatment was done to remove the remaining gDNA using the Turbo DNase I (Ambion, USA) at 37 °C for 30 minutes and RNA were then purified using the RNeasy MinElute Kit (Qiagen, GER). RNA samples were tested for purity by PCR. RNA quantity was measured using a Nanodrop fluorospectrometer, while quality was checked by agarose gel electrophoresis.
Quantitative Real-Time PCR (qRT-PCR) analysis. cDNA were synthetized from 5 µg of total RNA using the Superscript III reverse transcriptase and 250 ng of random hexamers according to the manufacturer's instructions (Invitrogen, USA). The qRT-PCR was performed using the GoTaq qPCR master mix (Promega, USA), 2 μl of cDNA (diluted 1:100) and 0.5 µM of specific primers (Supplementary Table S1) for each gene. The enzyme was activated by heating at 95 °C for 2 minutes. Then, the cDNA was amplified by qPCR (40 cycles) with a denaturation step at 95 °C for 15 seconds, an annealing step at 60 °C for 60 seconds. A final step of dissociation was achieved by increasing the temperature by 0.3 °C every 5 seconds from 60 °C to 95 °C. qPCRs were performed in triplicates using a Rotor-GeneQ instrument (Qiagen, GER). Melting curves were analyzed and the presence of a single peak was checked. For each pair of primers, primer efficiency was evaluated using standard curves from gDNA from 5 ng/µL to 0.001 ng/µL. Absolute quantifications x were determined using the following formula: x = (y − b)/m; y corresponds to the Ct value, while m and b are automatically calculated using standard curves.
Microscopy. Cells were immobilized on slides using a thin layer of PYE + 1.5% agar and imaged immediately. Phase contrast microscopy images were taken with a Plan-Apochromat 100X/1.45 oil Ph3 objective on an AxioImager M1 microscope (Zeiss) with a Prime-95B back-illuminated CMOS camera (Photometrics) controlled by the VisiView software (Visitron Systems, Germany). Images were processed using Adobe Photoshop. The MicrobeJ software 50 was used to measure the length of cells on images using default parameters. Measurements from 100-200 cells were used to evaluate cell size distributions.