Ciprofloxacin sensitivity of the ΔlexA ΔhigBA mutant is due to excessive production of a transcription factor downstream of higBA

To investigate the source of the unexpected HigB-independent ciprofloxacin sensitivity in ΔlexA ΔhigBA, we focused on the CCNA_03131 (hereafter, higX) gene downstream of higBA (Fig 1A). We constructed in-frame deletions of higX in WT, ΔlexA or ΔhigBA strains and investigated the effect on ciprofloxacin sensitivity. Consistent with our previous observations [15], loss of higA in the WT background sensitized cells to ciprofloxacin (through HigB activation), while deletion of higB from the ΔlexA strain was protective against it. Deletion of higX in WT, with or without deletion of higBA, had no effect. However, in the ciprofloxacin-sensitive ΔlexA ΔhigBA background, deletion of higX improved ciprofloxacin resistance to a similar level to the ΔlexA ΔhigB strain, showing that higX is indeed responsible for the ciprofloxacin sensitivity in the ΔlexA ΔhigBA strain in the absence of the HigB toxin (Fig 1B).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. The higBA toxin-antitoxin system has a transcription factor closely associated with it.

(A) Operon structure of higBA and location of the downstream gene CCNA_03131 (higX). The promoter upstream of higBA is negatively regulated by LexA and HigA. The higB and higA genes are translationally coupled, with 7 bp overlap between the stop codon of higB and the start codon of higA, while the higX start codon is 42 bp downstream of the higA stop codon. (B) Efficiency of plating assay of WT Caulobacter (NA1000) and mutant strains lacking combinations of lexA, higB, higA and/or higX, on 0.5 μg/ml ciprofloxacin or vehicle control. Images are representative of three independent biological replicates. (C) Quantitative RT-PCR of mRNA levels of higB and higX, normalized to the control transcript rpoD. Data was obtained from three independent biological replicates.

https://doi.org/10.1371/journal.pgen.1011986.g001

The ΔlexA ΔhigBA strain had been constructed by in frame deletion of higBA in the ΔlexA single mutant background. However, we subsequently realized that this would exert polar regulatory effects on higX expression, because (1) deletion of the higBA coding sequence brings higX directly downstream of the strong promoter regulating higBA, and (2) both repressors of this promoter, LexA and HigA, are absent. Indeed, we observed that higX was transcribed at very low levels in WT, ΔhigA and ΔhigBA but much higher in ΔlexA ΔhigBA (Fig 1C). Therefore, the ciprofloxacin sensitivity phenotype of ΔlexA ΔhigBA is associated with elevated HigX RNA expression, suggesting why higX deletion is protective.

HigX regulates the expression of higBA alongside LexA and HigA

In other three-component TA systems with a transcription factor as their third component, the transcription factor regulates the promoter upstream of the TA system [18,24]. To investigate whether HigX also has this activity, we examined activity of a transcriptional P_higBA-lacZ reporter construct (Fig 2A and 2B) in WT or strains lacking combinations of lexA, higBA or higX. Promoter activity was unaffected by higX deletion in the WT or in ΔhigBA backgrounds, but decreased in ΔlexA ΔhigBAX compared to ΔlexA ΔhigBA, suggesting that HigX can activate the promoter upstream of higBA when LexA and HigA are absent (Fig 2B). We attempted to confirm this finding by complementation with an overexpression construct containing higX under the control of the vanillate-inducible promoter (pMT335-higX). However, we could not transform the ΔlexA ΔhigBA strain with this plasmid, suggesting that its increased level of higX expression is sufficiently toxic to make the cells intolerant of transformation with a multi-copy plasmid containing higX, even in the absence of inducer. Expression of higX from this plasmid in WT, ΔlexA or ΔhigBA strains containing the P_higBA-lacZ transcriptional reporter had a weak inhibitory effect on promoter activity, which was only statistically significant in the ΔhigBA background (Fig 2C). Therefore, higX can activate or repress the promoter upstream of higBA, dependent on which other regulators are present, but its regulatory activity is weaker at this locus than that of LexA or HigA.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Translational reporter constructs show that higBA and higX are independently regulated.

(A) Construction of the transcriptional (red) or translational (green) reporter fusions to lacZ from regions of the higBAX genomic locus. The transcription start sites in this region previously detected by global mapping [25] are shown as red arrows. (B-C) Transcriptional beta-galactosidase reporter assays in WT and mutant strains lacking combinations of lexA, higB, higA and/or higX (B) and WT, ΔlexA and ΔhigBA strains containing the empty vector pMT335 or the higX overexpression plasmid pMT335-higX. Overexpression of higX was induced by 50 μM vanillate (C). (D-E) Translational beta-galactosidase reporter assays in WT and mutant strains lacking combinations of lexA, higB, higA and/or higX for a lacZ in-frame fusion to the higB upstream region and first ten codons of HigB (D), and a lacZ in-frame fusion to the higX upstream region and first four codons of HigX (E). A promoterless negative control of the translational fusion construct was tested in WT and had undetectable beta-galactosidase activity (zero to two Miller units).

https://doi.org/10.1371/journal.pgen.1011986.g002

The different levels of mRNA expression for higB and higX observed by qRT-PCR (Fig 1C) were suggestive of regulation of the two transcripts by separate promoters, consistent with previous genome-wide transcriptional profiling which identified independent primary transcriptional start sites for higBA and higX [25], at genome positions 3281528 and 3280865 respectively. We therefore constructed translational beta-galactosidase fusions of the regions immediately upstream of higB and higX (including the first few codons of each of the genes) to lacZ, such that the higB and higX reporter fusions should employ the native transcription start sites. Surprisingly, the translational fusion P_higB-higB::lacZ did not exhibit the same pattern of activity as the transcriptional reporter (Fig 2D, compare to 2B). For the translational reporter, the basal activity in WT, ΔhigBA, ΔhigBAX or ΔhigX strains was almost undetectable and high promoter activity was only observed in the three strains lacking lexA, in contrast to the transcriptional reporter where basal expression appeared high and the fold increase in the ΔlexA mutant strains was much less pronounced. Conversely, the activity of the translational reporter for higX expression was much weaker (Fig 2E). It was slightly LexA-dependent, but much less so than the promoter of higB. We conclude that higX is primarily regulated by the strictly LexA-dependent promoter upstream of higB, despite having an independent transcription start site, and that the different mRNA levels observed for higB vs. higX (Fig 1C) are due to other factors, for example processing or different stability. Moreover, the strikingly low basal level of P_higB-higB::lacZ activity, relative to the transcriptional lacZ fusion of the same region, is suggestive of an unknown post-transcriptional negative regulator of higBA.

HigX is a membrane-associated transcription factor

It was previously noted in a bioinformatic study of LytTR-family transcription factors [23] that all members of this family in the Caulobacter crescentus genome, including higX, are likely to encode transmembrane proteins. Bioinformatic prediction suggested that HigX contains four transmembrane helices N-terminal to the LytTR DNA binding domain, with the transmembrane and DNA-binding domains separated by a short low-complexity region (LCR) (Figs 3A and S1A). The four helices are predicted to form a tightly packed bundle with no apparent periplasmic domain (S1B Fig). To investigate whether the transmembrane domain was required for DNA binding activity, we constructed another overexpression plasmid where the higX gene was truncated to convert amino acid 133 (Val) to a new start codon, creating the higX-noTM variant containing only the LCR and the LytTR domain. We expressed this alongside the full-length higX overexpression plasmid and compared HigX abundance by western blotting using a custom polyclonal antibody to HigX alongside cell extracts containing pMT335 empty vector and NA1000 without any plasmid as control. It was only possible to detect HigX protein in the pMT335-higX overexpression strain, showing that while the antibody can detect HigX, the native level of this protein is very low (Fig 3B, left). We could not detect the truncated HigX protein from the pMT335-higX-noTM overexpression strain, suggesting that it may be unstable. Comparing the Caulobacter western blot results to cell extracts of E. coli Rosetta cells containing pET28a-higX or pET28a-higX-noTM constructs intended for protein purification, we observed that both the full-length and truncated proteins could be detected in E. coli with our HigX polyclonal antibody (Fig 3B, right). Therefore, we conclude that our antibody is capable of detecting both the full-length and the truncated HigX proteins, but that the abundance of the truncated variant in Caulobacter is below the detection limit under these experimental conditions.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. The N-terminal region of HigX localizes it to the cell periphery and is required for higBA promoter regulatory activity.

(A) Predicted domain structure (from Uniprot) of HigX with transmembrane helices (TM1–4) indicated in blue, low-complexity region (LCR) in green and the LytTR DNA-binding domain in orange. The black pin indicates where in the higX-noTM construct, amino acid Val133 was converted to a new Met start codon to create a truncated version consisting of the DNA binding domain without any transmembrane helices. (B) Western blots of cell extracts of Caulobacter cells expressing higX or higX-noTM in pMT335 with 50 µM vanillate induction (left) or E. coli Rosetta expressing His6-HigX in pET28a with 0.1 mM IPTG induction (right), with the anti-HigX polyclonal antibody. Solid arrows indicate the size of full-length HigX (28.2 kDa) and the dotted arow indicates the size of HigX-noTM (14.6 kDa). (C) Phase-contrast and fluorescence microscopy of WT and ΔlexA ΔhigX expressing a HigX-GFP translational fusion from the vanillate-inducible promoter. Scale bars indicate 2 μm. No GFP fluorescence was observed for cells carrying the pMT335 empty vector. (D) Beta-galactosidase assay of the transcriptional P_higBAX-lacZ reporter in WT and ΔhigBA strains containing the empty vector pMT335, the full length higX overexpression plasmid pMT335-higX or the truncated higX overexpression plasmid pMT335-higX-noTM. No statistically significant difference was observed between the empty vector control and the higX-noTM overexpression plasmid. (E) Relative abundances of HigX peptides in ΔhigBA, ΔhigX, ΔlexA and ΔlexA ΔhigBA membrane-enriched cell extracts detected by mass spectrometry and normalized to WT. Values are the averages of three independent biological replicates. Peptide abundance values (averages), CV% values (% coefficient of variation) and p values are in Table A in S1 Text. Note that non-zero values for HigX peptides in the ΔhigX strain are likely derived from the vestiges of the start and end of the higX coding region that remain in place after the in-frame deletion, and/or low-frequency detection of peptides derived from other proteins that coincidentally have the same mass/charge ratio as HigX-derived peptides.

https://doi.org/10.1371/journal.pgen.1011986.g003

To investigate whether HigX is localized to the membrane in living cells, we constructed C-terminal GFP fusions to full length HigX and to the truncated HigX-noTM allele lacking the transmembrane helices. Since the native level of HigX expression was so low, these were created as plasmid-borne constructs under control of the vanillate-inducible promoter in order to have sufficient expression to detect GFP fluorescence. Fluorescence microscopy showed that full length HigX localizes uniformly around the cell periphery in both WT and filamentous cells (Fig 3C), indicating that normal cell division is not required for HigX subcellular localisation. No fluorescence was observed in cells carrying the truncated HigX-noTM-GFP fusion protein, and overexpression of the truncated HigX-noTM protein (without GFP fusion) had no effect on activity of the P_higBA-lacZ transcriptional reporter (Fig 3D), confirming that the transmembrane domain is essential for stability and/or activity of HigX in the cell.

In order to detect the HigX native expression levels in WT and mutant strains, we employed quantitative mass spectrometry. Protein extracts were prepared from WT, ΔhigBA, ΔhigX, ΔlexA, and ΔlexA ΔhigBA strains, with separation of the soluble (cytoplasmic) proteins from the insoluble (membrane-associated) proteins before trypsinization and isotope labeling. A total of 1332 membrane-associated proteins were detected with two or more peptides, including HigX. No HigX peptides were detected in the soluble fraction, indicating that the fluorescence observed at the cell periphery in Fig 3C is due to membrane localisation. The abundance of HigX peptides was not significantly changed between WT and ΔhigBA, strongly depleted in ΔhigX and enriched in both ΔlexA and ΔlexA ΔhigBA strains by 30–40% (Fig 3E and Table A in S1 Text), consistent with the trend observed in the translational reporter experiments (Fig 2).

HigX acts independently of the toxin-antitoxin system HigBA and is toxic in filamentous cells independently of the SOS response

To test our hypothesis that overproduction of HigX could be responsible for ciprofloxacin sensitivity in ΔlexA, we expressed it in trans from the vanillate-inducible promoter using pMT335-higX in WT, ΔlexA and ΔlexA ΔhigB. We found that higX overexpression had no effect on cell viability compared to the empty vector control on normal media. However, on ciprofloxacin, higX overexpression reduced viability of the ΔlexA strain compared to the empty vector. Strikingly, the higX overexpression plasmid also reversed the improvement in ciprofloxacin resistance seen in the ΔlexA ΔhigB strain relative to empty vector (Fig 4A). In addition to confirming that higX overexpression was at least partially responsible for the ciprofloxacin sensitivity phenotype, this result also indicated that HigX does not act solely through regulation of the TA system promoter, since the phenotype was clearly visible in the absence of the HigB toxin. In support of the independence of HigX from HigBA, we also observed an improvement of ciprofloxacin resistance in a ΔlexA ΔhigX double mutant relative to ΔlexA, where both of these strains have the higBA TA system genes intact (Fig 4B). This result demonstrates that the increased abundance of HigX seen in the ΔlexA single mutant strain (Fig 3D) contributes to its ciprofloxacin sensitivity in addition to (and independently of) the overexpression of HigB.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Overexpression of higX in trans causes ciprofloxacin sensitivity independently of the higBA toxin-antitoxin system.

(A) Efficiency of plating assay of WT, ΔlexA and ΔlexA ΔhigB containing empty vector pMT335 or the higX overexpression plasmid pMT335-higX on ciprofloxacin (0.5 μg/ml) or vehicle control. (B) Efficiency of plating assay of WT, ΔlexA and ΔlexA ΔhigX on ciprofloxacin (0.5 μg/ml) or vehicle control. (C) Efficiency of plating assay of WT, ΔtipN, Δlon and ΔpopZ containing empty vector pMT335 or the higX overexpression plasmid pMT335-higX on ciprofloxacin (0.5 μg/ml) or vehicle control. (D) Efficiency of plating assay of WT, ΔlexA, ΔlexA ΔhigX and ΔlexA ΔsidA containing empty vector pMT335 or the higX overexpression plasmid pMT335-higX on ciprofloxacin (0.5 μg/ml) or vehicle control. In (A), (C) and (D), all plates contained gentamicin and 50 μM vanillate in addition to ciprofloxacin (0.5 μg/ml) or vehicle control. Images are representative of three independent biological replicates.

https://doi.org/10.1371/journal.pgen.1011986.g004

To investigate whether the HigX-mediated ciprofloxacin sensitivity was unique to the ΔlexA strain, we performed the same experiment on filamentous mutant strains of Caulobacter lacking cell cycle regulators (TipN, Lon and PopZ) [26–30] but with intact SOS reponse. Here, we observed that the filamentous cell cycle mutants were also more sensitive to ciprofloxacin, and that the sensitivity was potentiated by HigX overexpression, approximately in proportion to the degree of filamentation (Fig 4C). Furthermore, when we overexpressed HigX in the ΔlexA ΔsidA strain, which has a constitutively active SOS response but which does not become filamentous due to the absence of the SidA cell division inhibitor [31], this strain was slightly more resistant to HigX overexpression than the ΔlexA mutant, although it was still sensitive to HigX overexpression combined with ciprofloxacin (Fig 4D). Therefore, the HigX overexpression-associated phenotype is not unique to the ΔlexA genetic background, but instead linked to cell elongation. Finally, we investigated whether the transmembrane domain alone had positive or negative effects by cloning the four-helix transmembrane domain alone into pMT335 and overexpressing it in WT, ΔlexA, ΔlexA ΔhigX and ΔlexA ΔsidA strains. In contrast to full-length HigX, this construct was unexpectedly protective, on control medium and even more so on ciprofloxacin and cephalexin (S1C Fig). Although we cannot quantify how strongly this construct is expressed since the N-terminal section of the protein does not contain the epitope recognized by our HigX polyclonal antibody, it is clear from the phenotype that (i) it is possible to express the N-terminal domain on its own, and (ii) both domains together are required for the toxic effect of HigX overexpression to be seen.

HigX preferentially binds cell envelope, transport, and metabolism-associated genes

To assess the DNA binding capability of HigX, we performed chromatin immunoprecipitation sequencing (ChIP-Seq) on wild type, ΔlexA and ΔlexA ΔhigX strains using our polyclonal antibody to HigX. Using the “MACS2 callpeak” program, we identified 596 significantly enriched peaks in the wild-type strain (S1 Dataset). In contrast, peak calling for the ΔlexA strain revealed genome-wide depletion of the enriched peaks, with 413 peaks identified with significant enrichment, but with enrichment signal values uniformly lower than in WT (Fig 5A). In the genetic negative control strain ΔlexA ΔhigX, we observed a similar profile to the ΔlexA strain except with a higher signal to noise ratio due to a much lower total number of reads (leading to overall amplified values when calculating reads per million) (S2 Fig). Therefore, the DNA-binding and thus regulatory function of HigX is strongly reduced upon deletion of lexA, despite the 40% upregulation of HigX peptide abundance (Fig 3E). Significant peaks in the WT strain were further filtered through a 95% threshold based on their signal value (≥ 3.22), resulting in 30 significant peaks that corresponded to the top 5% of the dataset. The highest of these peaks was located between CCNA_03217 (a TetR-family transcriptional regulator) and CCNA_03218 (periplasmic multidrug efflux lipoprotein precursor), with the summit being closest to CCNA_03218 (Fig 5B).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Genome-wide analysis of HigX DNA-binding and functional enrichment.

(A) ChIP-seq profiles of HigX in Caulobacter NA1000 WT (blue) and ∆lexA (yellow) strains, with the data plotted as reads per million (RPM) across the genome. (B) Zoom-in of the highest peak in the dataset located between CCNA_03217 and CCNA_03218, with a gene track displaying the transcriptional orientation. The peak region is highlighted in red, and the summit position is indicated by a red dashed line. (C) COG analysis of significant ChIP-seq peaks in the WT strain. Blue bars represent all significant peaks identified by MACS2, while red bars indicate peaks filtered by a signal value threshold (≥ 3.22). Two categories were omitted: “Function unknown (S)” with 66 (blue) and 3 (red) counts, and “Unassigned (no COG)” with 177 (blue) and 9 (red) counts. (D) Sequence logo of the HigX binding motif identified through MEME-ChIP from the peak summit regions of the filtered peaks from the WT (signal value ≥ 3.22). The motif consists of an imperfect inverted repeat of 8-bp with no intervening spacer. The left half of repeat is located at positions 4–11 (ATGTTASC) and the right half at positions 12–19 (GSTACCAT) with a single mismatch between the repeat halves. (E) Transcriptional beta-galactosidase reporter assay of the CCNA_03218 promoter in WT containing empty vector pMT335 or the overexpression plasmids pMT335-higX or pMT335-CCNA_03217. Overexpression was induced by 50 µM vanillate. (F) Efficiency of plating assay of WT, ΔlexA and ΔlexA ΔhigX on cephalexin (5 μg/ml) or vehicle control. (G) Efficiency of plating assay of WT, ΔlexA, ΔlexA ΔhigX and ΔlexA ΔsidA containing empty vector pMT335 or the higX overexpression plasmid pMT335-higX on cephalexin (5 μg/ml). The vehicle control for this experiment is shown in Fig 4D. Images in (F) and (G) are representative of three independent biological replicates.

https://doi.org/10.1371/journal.pgen.1011986.g005

To determine the overall function of HigX target genes and provide a holistic view of its potential regulatory role, we extracted the summit locations of all 596 peaks and mapped them to the nearest genes, followed by COG (Clusters of Orthologous Groups) annotation and functional classification. From this, we observed that most peaks were associated with the following COG categories, listed from most to least abundant: M, C, T, P, K, and J (Fig 5C). However, when we focused on the top 30 significant peaks (signal value ≥ 3.22), genes associated with carbohydrate transport and metabolism (G) became the most prominent, while most of the previously dominant categories (M, C, T, P, and K) remained to a lesser extent. In contrast, genes associated with translation, ribosomal structure, and biogenesis (J) were no longer represented. This COG distribution suggests that, for the 5% highest confidence ChIP peaks, HigX preferentially targets genes associated with membrane, transport, and metabolism-related functions.

To further characterize the DNA-binding specificity of HigX, we performed a motif discovery analysis by utilizing the summits of the 30 significant (top 5%) ChIP-seq peaks in conjunction with MEME-ChIP. This analysis identified a significantly enriched 19-bp motif (E-value = 1.4 × 10 ‒²³) present in 13 of the 30 submitted sequences, with a strong central enrichment, which indicates that this motif likely represents the direct binding site of HigX (Fig 5D). The motif is characterized by an 8-bp imperfect inverted repeat, with the left half spanning positions 4–11 (ATGTTASC) and the right half spanning positions 12–19 (GSTACCAT), with no intervening spacer and a single mismatch between the repeat halves. This motif was observed upstream of and on the same strand as the CCNA_03218 periplasmic multidrug efflux lipoprotein precursor, at genome positions 3378630–3378644 (TGTTATCAATCACTG), where we had observed the highest peak in our ChIP-Seq dataset. Since the initial HigX-dependent phenotype we had observed was ciprofloxacin sensitivity, we hypothesized that it could be due to HigX-mediated regulation of this putative efflux pump. To test in vivo whether HigX regulates this gene through binding this motif, we measured activity of a P_{CCNA_03218}-lacZ transcriptional reporter in WT cells overexpressing either HigX or the divergently encoded TetR-like protein CCNA_03217, since this class of protein often negatively regulates the efflux pump operons they are co-encoded with. Compared to the empty vector control, overexpression of HigX had no effect on this promoter, while overexpression of CCNA_03217 repressed it (Fig 5E). Therefore, the efflux pump is likely to be regulated primarily by CCNA_03217 in WT cells (although we do not exclude that HigX and CCNA_03217 could both bind this promoter) and is probably not dysregulated in HigX-overexpressing cells or responsible for the antibiotic sensitivity phenotype.

Thus far, we had only investigated ciprofloxacin sensitivity of the ΔlexA strain and derivatives. However, the ChIP-Seq analysis suggested that the dysregulation of HigX in the ΔlexA strain, either at the level of overproduction of the protein or failure to bind the target genes, may also lead to cell envelope irregularity that could lead to sensitivity to cell envelope-targeting antibiotics. We therefore tested the ability of the ΔlexA and ΔlexA ΔhigX strains to grow on cephalexin, which targets septal peptidoglycan synthesis. The ΔlexA strain was sensitized to cephalexin relative to WT, and cephalexin resistance was restored in the ΔlexA ΔhigX double mutant (Fig 5F). The cephalexin sensitivity of the ΔlexA mutant was likely due to its filamentation, since the ΔlexA ΔsidA strain was more resistant to this antibiotic, including when HigX was overexpressed (Fig 5G). Therefore, even though HigX production is under SOS control, its downstream effects in the cell are more likely to be involved with cell envelope maintenance than with the SOS response.

Suppressor mutations in cell envelope-related genes improve ΔlexA ΔhigBA ciprofloxacin resistance

To identify possible downstream effectors of HigX responsible for the overexpression phenotype, we designed parallel forward genetic screening strategies to take advantage of (1) the ΔlexA ΔhigBA strain’s sensitivity to ciprofloxacin and (2) its intolerance of transformation with the pMT335-higX plasmid (Fig 6A). We constructed a pooled library of random himar1 transposon (Tn) mutants in the ΔlexA ΔhigBA background in order to select for (1) ciprofloxacin-resistant Tn mutants and (2) Tn mutants which can be stably transformed with pMT335-higX, using the ΔlexA ΔhigBA strain in preference to the ΔlexA strain in order to avoid selection of phenotypes due to the HigB toxin. The first screen was performed by passage of the mutant pool through medium containing 1 μg/ml ciprofloxacin, followed by selection of ciprofloxacin-resistant single colonies. The second was performed by preparation of electrocompetent cells from a culture of the pooled library, electroporation with pMT335-higX and plating on medium containing gentamicin and vanillate (to select for the plasmid antibiotic resistance marker and to induce higX overexpression, respectively). Such mutants should have Tn insertions in loci which suppress the HigX-mediated phenotypes.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Forward genetic screening for suppressors of ΔlexA ΔhigBA ciprofloxacin sensitivity reveals cell envelope-related genes.

(A) Schematic of the forward genetic screening protocol in the ΔlexA ΔhigBA strain for ciprofloxacin resistance (left) and pMT335-higX transformation tolerance (right). (B) Location of the transposon insertions in the four verified suppressor mutants. The insertion sites at the CCNA_03248 locus are at 3415650 (upstream) and 3417411 (inside CDS), the insertion site in the CCNA_00669 gene is at 727111 and the insertion site upstream of the amiC gene is at 2097088. (C) Zoom-in of the ChIP-Seq profiles of WT (blue) and ΔlexA (yellow) at the CCNA_03248 locus. (D) Efficiency of plating assay of WT, ΔlexA ΔhigBA and ΔlexA ΔhigBA transduced with three of the suppressor transposon mutations, on ciprofloxacin (0.5 μg/ml) or vehicle control. (E) Transcriptional beta-galactosidase reporter assays of the CCNA_03248 (left) and amiC (right) promoters in WT containing empty vector pMT335 or the overexpression plasmid pMT335-higX. Overexpression was induced by 50 µM vanillate. (F) Efficiency of plating assay of WT, ΔlexA, ΔlexA ΔhigX and ΔlexA ΔhigBAX containing empty vector pMT335 or the amiC overexpression plasmid pMT335-amiC on ciprofloxacin (0.5 μg/ml) or vehicle control. Overexpression was induced by 50 µM vanillate. Images in (D and F) are representative of three independent biological replicates.

https://doi.org/10.1371/journal.pgen.1011986.g006

From our Tn mutant library of approximately 25’000 clones, we were able to obtain four stable Tn mutants where we could confirm the desired phenotypes by re-plating on selective media (ciprofloxacin, or gentamicin + vanillate) followed by backcrossing of the transposon mutations into ΔlexA ΔhigBA cells which had not been through the mutagenesis or selection procedure. All four suppressor mutations were in or near genes associated with cell envelope-related functions (Fig 6B). The two Tn mutants obtained from the ciprofloxacin resistance screen both had a Tn insertion in or near CCNA_03248, encoding an uncharacterized TonB-dependent receptor. One insertion was found within the coding sequence of CCNA_03248, while the other was in an intergenic region 218 bp upstream of the CCNA_03248 start codon. Notably, the Tn insertion site in the intergenic region, at 3415650, is only 188 bp away from the peak summit of one of the significant (although not top 5%) HigX binding sites, at 3415838, suggesting that this gene is a direct HigX target (Fig 6C). The two Tn mutants obtained from the pMT335-higX tolerance screen had Tn mutations in CCNA_00669, encoding a lipopolysaccharide biosynthesis protein, and approximately 100 bp upstream of the peptidoglycan amidase gene amiC. However, the two mutants obtained from the pMT335-higX tolerance screen were also ciprofloxacin-resistant (Fig 6D), and the mutants obtained from the ciprofloxacin resistance screen could be transformed with pMT335-higX, demonstrating that even if the genetic screens were not saturated, the selection strategies have succeeded in isolating suppressor mutants with convergent phenotypes.

Consistent with the ChiP-Seq profile at the CCNA_03248 locus, we saw that overexpression of HigX from pMT335-higX reduced promoter activity of a P_{CCNA_03248}-lacZ reporter construct, while it had no effect on the amiC promoter where no ChIP peak had been observed (Fig 6E). Examining the genomic region where the CCNA_00669 gene was located, we could see that this gene was located in the middle of a probable multi-gene operon, meaning that the insertion may well have caused polar effects on downstream genes in addition to disrupting CCNA_00669 itself. Moreover, no ChIP-Seq peak for HigX was found within 4 kbp either side of CCNA_00669. In contrast, extra HigX peaks were observed at the promoters of the flanking genes of CCNA_03248 (Fig 6C), making the significance of this gene harder to interpret. Therefore, we focused on the amiC suppressor. We anticipated that the mechanisms of the CCNA_03248, CCNA_00669 and amiC suppressors are probably different, since the CCNA_03248 and CCNA_00669 intragenic insertions will probably lead to a loss of function, while the amiC promoter insertion is more likely to be an overexpression suppressor since amiC is an essential gene in Caulobacter [32,33]. In order to mimic the effect of the P_amiC insertion, we overexpressed amiC from an inducible promoter on pMT335. On media with vehicle control, no difference in viability was seen between any of WT, ΔlexA, ΔlexA ΔhigX or ΔlexA ΔhigBAX with empty vector or with pMT335-amiC. As expected, the ΔlexA mutant was sensitized to ciprofloxacin compared to the other strains. AmiC overexpression could partially rescue ΔlexA viability on ciprofloxacin, but did not have any further positive effect on WT or the mutant strains lacking higX (Fig 6F). Hence, the suppressor mutation in P_amiC is most likely to be an indirect promoter-up mutation that increases AmiC levels, which then counteracts the negative effect of HigX overproduction.

Taken together, these data support a model where overproduction of HigX in a filamentous cell context, either from endogenous overproduction in ΔlexA strains or in trans from pMT335-higX in ΔlexA or other filamentous mutants, leads to a specific reduction of viability associated with cell elongation. Since higX deletion protected the ΔlexA strain against Cipro and cephalexin sensitivity even though HigX did not bind its target genes properly in the ΔlexA mutant, the phenotype cannot only be due to changes in HigX-dependent gene regulation but could be a combination of target gene dysregulation together with having too much HigX protein in the cell membrane, or having the protein present but nonfunctional. We infer that our suppressor Tn mutants counteract the destabilizing effect of HigX on the cell envelope, either directly or indirectly.

higX is widespread among alpha-proteobacteria but synteny with higBA is not conserved

To investigate whether this unusual combination of TA system with adjacent but independent transcription factor was conserved among other bacteria, we evaluated the genomic conservation of this locus using the FlaGs software [34] with the Caulobacter crescentus NA1000 HigX protein sequence as query and the limit of aligned genomes set to 99, of which 68 were successfully aligned. We observed that higX is not unique to Caulobacter species but found in many other alpha-proteobacteria as well. Within the Caulobacter genus, the flanking gene context of higX was well conserved, immediately upstream of the master cell cycle regulator ctrA. Moreover, it was co-conserved with a conserved uncharacterized gene, annotated as a putative hydrolase of the COG4188 family in all except five of the more distantly related genomes, including in the non-Caulobacter species which did not have their higX orthologue adjacent to ctrA (S3 Fig). However, in the 68 aligned higX-proximal regions, only four of the genomes contained probable higBA orthologues upstream of higX. Hence, higX is much more conserved in its genomic context in alpha-proteobacteria than higBA is (although we do not exclude that the other bacteria may encode higBA at a different site in the genome, since at least in the Caulobacter genus it is conserved between species [35]) and higX is almost universally found adjacent to the uncharacterized hydrolase gene.

While examining the alignments of S3 Fig we noted that the intergenic region between the putative hydrolase and the higB gene appeared consistently larger than the intergenic region between the putative hydrolase and the higX gene (in the majority of genomes which lacked higBA), suggesting that some of the intergenic sequence may be associated exclusively with the higBA genes. We therefore aligned the intergenic regions between the putative hydrolase gene and hig(BA)X from a subset of 15 Caulobacter genomes, including NA1000 and the four others which contained higBA upstream of higX, and observed a consistent breakpoint in the intergenic region where the intergenic sequence stopped in the genomes lacking higBA, but continued in the genomes which had higBA upstream of higX. (Fig 7A). This breakpoint was at base 3281622 of the Caulobacter crescentus NA1000 genome, relative to the start codons of the putative hydrolase at base 3281744 and higB at base 3281524. From this, we inferred that of this intergenic region, the 122 bp adjacent to the putative hydrolase gene is universal to all the genomes, while the additional 98 bp of intergenic sequence found only in the genomes of the strains possessing higBA is likely to be a higBA-specific regulatory element. Notably, this additional sequence contained the high-affinity LexA binding site that is responsible for the LexA-mediated repression of the higBA promoter. Therefore, the LexA-mediated repression of higX under normal conditions and the overproduction of HigX when this repression is released is likely unique to NA1000 and the other Caulobacter crescentus/vibrioides strains which contain higBA and its LexA-dependent promoter immediately upstream of higX. In all other higX-encoding alphaproteobacteria, we would instead anticipate higX to be expressed in a LexA-independent manner.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. HigX is broadly conserved in its genomic context while higBA is restricted to the Caulobacter crescentus/vibrioides lineage.

(A) Alignment of the intergenic region upstream of hig(BA)X containing the breakpoint between +higBA and -higBA strains. The DNA sequence shown corresponds to bases 3281574–3281632 of the NA1000 genome sequence. The LexA binding motif is shown in yellow. (B) Multivariate compositional analysis of relative synonymous codon usage (RSCU) across all coding sequences in Caulobacter NA1000 with individual CDSs represented as grey points and the 95% confidence intervals around the genomic centroid indicated as a dashed ellipse. The three genes of interest are indicated as higB (orange), higA (blue) and higX (green). Below the PCA plot, density plots display the genome-wide distributions of Mahalanobis distance (MD²) under both standard two-dimensional (PC1–PC2) and robust minimum covariance determinant (MCD) models (k = 35). Empirical FDR < 0.05 cutoffs (Benjamini–Hochberg), are shown as black dotted lines and the MD² values for higB, higA and higX are indicated in their respective colors. (C) Multivariate compositional analysis of tetranucleotide frequencies using the same approach as in (B). The tetranucleotide frequency composition of each gene was normalized and projected into principal component space, with the 95% confidence intervals around the genomic centroid indicated as a dashed ellipse. Again, higB, higA, and higX are shown in their respective colors, and the density plots display the corresponding MD² distributions for both two-dimensional and robust MCD models (k = 160) with FDR-based significance thresholds. (D) Efficiency of plating assay of the ΔlexA ΔhigX strain containing empty vector pMT335 or the overexpression plasmids pMT335-higX, pMT335-segnis-higX or pMT335-brev-higX on ciprofloxacin (0.5 μg/ml) or vehicle control. Overexpression was induced by 50 µM vanillate. Images are representative of three independent biological replicates.

https://doi.org/10.1371/journal.pgen.1011986.g007

In order to investigate whether higBA is a recent arrival in a minority of the Caulobacter genomes, or whether it is an ancestral factor that has been lost from the majority of them, we employed compositional analysis of relative synonymous codon usage (RSCU) and tetranucleotide frequency. Using multivariate compositional analysis of these two parameters (S2 Dataset), we detected a minor atypicality of higX in its codon usage (q = 0.0211), but no significant difference in tetranucleotide composition (q = 0.805) relative to other coding sequences of the genome. In contrast, higB and higA displayed significant compositional atypicality in their RSCU (higB: q = 3.83 x 10^-37, higA: q = 5.25 x 10^-7) and tetranucleotide composition (higB: q = 9.46 x 10^-187, higA: q = 4.39 x 10^-144), compared to all CDS within the genome. This degree of deviation from the rest of the genome is potentially indicative of horizontal gene transfer or, at the least, evidence for a recent arrival (Fig 7B - C). The small deviation in codon usage for higX is likely biologically negligible, with an effect size of 1.08 (i.e., 8% above the 95% cutoff; S3A Fig). By comparison, higB and higA exhibited larger effect sizes (RSCU: 2.34 and 1.39; tetranucleotide: 2.66 and 2.43; S3A Fig). Thus, higBA appear to be compositionally highly atypical, whereas higX is compositionally typical relative to the rest of the genome. We conclude from this that the presence of higBA at this locus is most likely due to an insertion event into a recent ancestor of the NA1000 lab strain of Caulobacter and a few closely related species, while the presence of higX alone at this locus is the “ancestral” state.

To evaluate whether HigX function is conserved across genera, we investigated whether two HigX orthologs from species that do not have HigBA located upstream had the same effect as native HigX overexpression in the NA1000 genetic background, by heterologous complementation EOP assay. ΔlexA ΔhigX cells were transformed with pMT335 overexpression plasmids containing higX from Caulobacter segnis (segnis-higX, 68% identity/76% similarity to NA1000 HigX protein sequence) or Brevundimonas lenta (brev-higX, 46% identity/57% similarity to NA1000 HigX protein sequence) and assessed for growth on ciprofloxacin or vehicle control (Fig 7D). Native higX overexpression in ΔlexA ΔhigX complemented the phenotype to that seen in the ΔlexA single mutant, with slightly reduced colony growth on control plates and strongly reduced viability on ciprofloxacin. Segnis-higX also slightly reduced colony growth on control medium, similarly to native higX, but did not sensitise the cells to ciprofloxacin. Brev-higX overexpression resembled the empty vector control and did not cause any viability changes on either medium. Since the higX orthologs did not fully complement, we conclude that changes in higX activity in the NA1000 strain have probably occurred in parallel with the acquisition of higBA, possibly to mitigate the disadvantage of having excessive HigX produced upon activation of the SOS response.

higX is ancestral but higBA is of likely foreign origin

To support the compositional bias analysis, we investigated if the compositional typicality of higX and atypicality of higBA was reflected in their encoded proteins’ phylogeny, as an indication of their most likely origin. The 16S rRNA phylogenetic tree showed that all genera cluster into separate clades, with the only exception being Asticcacaulis excentricus CB 48, whose placement is within the Brevundimonas clade (S3B Fig). Regarding the presence/absence heatmap, as expected, HigX was found to be the most prevalent protein across the order, with 13 organisms containing it in different combinations and 9 containing it as the sole protein. Interestingly, the prevalence of all 3 proteins was not limited to members of the Caulobacter genus. In fact, one member of Brevundimonas contained all three proteins. However, it is important to note that this heatmap does not account for the synteny of the genes encoding the proteins. Thus, it is possible that even if an organism contains all three proteins, they are not within the same gene cluster or operon. higBA could likely be positioned away from higX somewhere else on the genome.

The protein phylogenetic trees to be compared against the 16S rRNA phylogenetic tree were generated by utilizing BLASTx with higB, higA or higX as queries. Hits (n = 100) were filtered and used for creating a presence/absence heatmap of each gene per organism and in different combination classes (S4 Fig). Occurrence of all three proteins was limited to members of the Caulobacter genus (n = 5), closely matching results of the heatmap from the 16S rRNA phylogenetic tree (S3B Fig). HigB and HigA co-occurring in the absence of HigX were only found in one species out of the 100 used and the co-occurrence of HigA and HigX was limited to the Caulobacter genus (n = 7). Interestingly, in the remaining combination classes, a stark contrast was observed. The class for HigB as the sole protein (n = 41) contained no organisms within the Caulobacterales order (S4 and S5 Figs). In fact, twelve of these were members of the Enterobacter genus, which is part of Gammaproteobacteria, showing a significant discrepancy with the 16S rRNA phylogenetic tree. A similar pattern was observed for the HigA-only class (n = 59), whose members consisted of organisms outside Caulobacterales (S4 and S6 Figs). By comparison, the HigX-only class (n = 56) displayed a phylogeny similar to the 16S rRNA phylogenetic tree, with only two organisms coming from outside the Caulobacterales order (S4 and S7 Figs). Overall, the results from the phylogenetic analysis strongly indicate that higX is both ancestral to and phylogenetically consistent with the Caulobacterales lineage, whereas the phylogenies of higB and higA are indicative of a foreign genetic origin, possibly enterobacterial in the case of higB.

General growth conditions

Caulobacter crescentus strains were routinely grown in peptone-yeast extract (PYE) medium at 30°C and E. coli strains in LB at 37°C. Antibiotics were used at the following concentrations; tetracycline at 1 μg/ml for Caulobacter and 10 μg/ml for E. coli, gentamicin at 1 μg/ml for Caulobacter and 10 μg/ml for E. coli, nalidixic acid at 20 μg/ml for Caulobacter and kanamycin at 20 μg/ml (solid media) or 5 μg/ml (liquid media) for Caulobacter and 20 μg/ml for E. coli. Ciprofloxacin was prepared as 20 mg/ml stock solution in 0.1 M HCl and in all experiments involving ciprofloxacin, an equivalent volume of 0.1 M HCl was added to the control cultures or plates. Vanillate stock solution was prepared at 50 mM stock solution (adjusted to pH 8.0 with NaOH) and used at 50 or 100 μM as indicated in the figure legends.

Strain and plasmid construction

DNA fragments for cloning were PCR-amplified with Phusion DNA polymerase (Thermo Fisher Scientific) from stationary phase cultures of wild type (WT) Caulobacter crescentus, using PCR primers listed in Table B in S1 Text. Products were purified by agarose gel electrophoresis. Cloning of the correct region was confirmed by sequencing and plasmid stocks were maintained in E. coli EC100 or TOP10. Plasmids used in this work are listed in Table C in S1 Text. Replicating plasmids (pMT335 [51], pJC327 [52] and plac290 derivatives) were transferred into Caulobacter strains by electroporation while suicide plasmids for generation of deletion mutants (pNPTS138 derivatives) and transposon mutants (pHPV414 [53]) were transferred by conjugation from E. coli S17-1 λpir [54]. After integration of suicide vectors by recombination through one of the homologous flanking regions, secondary recombination events were induced by counterselection on PYE agar containing 3% sucrose and mutants carrying resulting in-frame deletions were screened for by PCR. Double mutants were made by introducing the ΔhigX or ΔhigBAX alleles into the ΔlexA mutant strain. Generalised transductions with bacteriophage ΦCr30 were performed as previously described [55]. C. crescentus strain numbers and genotypes are listed in Table D in S1 Text. pET28a derivatives containing WT or truncated HigX fused to an N-terminal His-tag were expressed in E. coli Rosetta(DE3) (Novagen).

Efficiency of plating (EOP) assays

Resistance to cephalexin and ciprofloxacin was assessed by dilution spot plating. Cultures of strains to be tested were grown overnight to stationary phase, then inoculated into new medium to grow to mid-exponential phase. Culture density was measured, normalised to the OD600 of the least dense culture (OD600 = 0.5 or less), serially diluted in PYE to 10^-6 and 5 µl spotted onto plates containing PYE medium with sub-inhibitory concentrations of cephalexin (5 μg/ml) or ciprofloxacin (1 μg/ml). For higX overexpression experiments, the plates additionally contained gentamicin in order to maintain selection on the plasmid, and all plates (including empty vector controls) contained vanillate (50 μM) to induce overexpression of higX. Plates were imaged after 3 days growth at 30°C. Images are representative of at least three independent biological replicates.

Beta-galactosidase assays

β-galactosidase assays were performed on strains carrying low copy plasmid-borne transcriptional or translational fusions of the promoters of interest to lacZ. Cultures were grown to early exponential phase (OD₆₀₀ = 0.1 – 0.4) with exposure to ciprofloxacin or vanillate as described in the main text or figure legends, followed by β-galactosidase assays using the method of Miller (35) on three independent biological replicates.

RNA extraction and quantitative RT-PCR

RNA was extracted from 4 ml mid-exponential phase cultures which were treated with 2500 U Ready-Lyse (Bionordika) and homogenized using QiaShredder columns, prior to RNA extraction with the RNEasy Mini Kit (Qiagen) according to the manufacturers’ instructions, including on-column DNase digestion. RNA quality was assessed by agarose gel electrophoresis and concentration was measured in a Nanodrop spectrophotometer. cDNA was prepared using the SuperScript IV Reverse Transcriptase Kit (Thermo Fisher) according to the manufacturers’ instructions, on 400 ng RNA template using random hexamer primers. Quantitative RT-PCR was performed in technical and biological triplicates on a 96-well LightCycler real-time PCR system (Roche) using SYBR Green (Roche). The higB transcript was amplified with primers higB_qrt_fwd and higB_qrt_rev, the higX transcript was amplified with primers CC_3036_qrt_fwd and CC_3036_qrt_rev, and the reference gene rpoD was amplified with primers rpoDfow and rpoDrev. Quantification was by the standard curve method and higB and higX transcript levels were normalized to rpoD.

Western blot and anti-HigX antibody

Custom polyclonal antibody to HigX for western blotting and ChIP-Seq was developed by Nordic Biosite. Rabbits were immunized with the synthetic peptide (C-)LRMEDHYVRIRTEHGSRLE corresponding to amino acids 178–196 of the HigX protein with cysteine conjugated at the N-terminus. Antiserum was extracted from the rabbits and affinity purified on a Sepharose column containing the immobilized synthetic peptide to obtain purified polyclonal antibody. Cell extracts for western blotting were prepared by growing Caulobacter cultures in PYE (containing 50 µM vanillate for strains containing pMT335 derivatives), or E. coli Rosetta(DE3) cultures in LB with 0.1 mM IPTG, to mid-exponential phase, normalizing to OD600 = 0.5, centrifuging and resuspending in 100 µl SDS-PAGE loading buffer. Extracts were run on a 16% SDS-PAGE gel alongside PageRuler prestained markers (Thermo Fisher). Proteins were transferred to PVDF membranes and blocked by incubation in TBS containing 5% dried skim milk. HigX protein was detected by incubation with primary antibody diluted 1/5000 in TBS with 2% skim milk at 4°C overnight, secondary antibody diluted 1/4000 in TBS with 2% skim milk at 4°C for one hour, and HRP substrate at room temperature for one minute. Between incubations, membranes were washed 2 x 5 minutes then 1 x 15 minutes in TBS.

Microscopy

Cultures for imaging were grown to mid-exponential phase with 50 μM vanillate to induce HigX-GFP expression and immobilized on 1% agarose (in water) microscope slides. Phase-contrast and GFP fluorescence images were obtained with an inverted Olympus IX83 microscope equipped with a 100x/1.46 oil objective and a Photometrics Prime scientific complementary metal-oxide-semiconductor (sCOMS) camera. GFP images were taken before phase-contrast images to minimize bleaching. Images are representative of three independent biological replicates.

Quantitative mass spectrometry

Cultures for mass spectrometry were grown to mid-exponential phase, harvested by centrifugation and cell pellets were immediately frozen in liquid nitrogen. Cells were sonicated in ice-cold 100 mM Na₂CO₃ containing cOmplete protease inhibitor and PhosSTOP phosphatase inhibitor (Sigma) for 2x20 sec at 60% amplitude on ice using a probe sonicator. After sonication the samples were incubated at 4 °C for 1h with rotation. After incubation, the samples were ultracentrifuged at 100.000g for 90 min using a Hitachi Micro ultracentrifuge (CS 150 NX). After ultracentrifugation, the supernatants (soluble protein fraction) were transferred to another tube and the membrane pellets were gently washed with 100µL 50 mM TEAB, pH 8.5. The membrane pellets were resolubilized in 5% sodium deoxycholate (SDC) in 50 mM HEPES buffer, pH 8.5, by probe sonication (1x20 sec at 20% amplitude). After sonication, the samples were centrifuged at 20.000g for 10 min, the supernatants were transferred to new tubes and the protein concentrations were measured by Nanodrop (Implen Nanophotometer N60). The membrane protein samples were reduced and alkylated using 5 mM DTT for 30 min and 10 mM iodoacetamide for 30 min, respectively. After alkylation, trypsin was added (5%) and the samples were incubated at 37 °C over-night. After incubation, additional trypsin (1%) was added and incubated at 37 °C for 1 hour.

A total of 25 µg of peptides from all conditions were labeled with Tandem Mass Tag (TMT) pro 18 plex (Thermo Fisher Scientific) according to the manufacturer’s protocol. The TMT labeling was performed as following: NA1000 (TMT 126, 127N and 127C); ∆higBA (TMT 128N, 128C and 129N); ∆higX (TMT 129C, 130N and 130C); ∆lexA (TMT 131N, 131C and 132N); ∆lexA ∆higBA (TMT 132C, 133N and 133C). After labeling, all samples were combined into one sample and the SDC was precipitated by adding formic acid to 2% followed by centrifugation at 20.000g for 20 min. The supernatant was transferred to a new Eppendorf tube and dried by vacuum centrifugation. The dried TMT labeled peptides were desalted using Poros Oligo R3 Reversed Phase (RP) material and subsequently fractionated using high pH RP fractionation into 12 concatenated fractions as described previously (PMID: 35240313). Each fraction was dried and resolubilized in 10 µL 0.1% formic acid prior to liquid chromatography tandem mass spectrometry (LC-MS/MS).

An aliquot of each peptide fraction was loaded onto a 20 cm analytical column (100 μm inner diameter) packed with ReproSil – Pur C18 AQ 1.9 μm RP material using an EASY nanoLC system coupled to an Orbitrap Eclipse Tribrid (Thermo Fisher Scientific). The peptides were eluted with an organic solvent gradient from 100% phase A (0.1% FA) to 22% phase B (95% ACN, 0.1% FA) for 100 min, then from 22% B to 40% B for 20 min before the column was washed with 95% B. The flow rate was set to 300 nL/min during elution. The automatic gain control target value was 1.5x10⁶ ions in MS and a maximum fill time of 50 ms was used. Each MS scan was acquired at high-resolution (120,000 full width half maximum (FWHM) at m/z 200) in the Orbitrap with a mass range of 350–1500 Da. The instrument was set to select as many precursor ions as possible in 3 sec between the MS scans. Peptide ions were selected from the MS for higher energy collision-induced dissociation (HCD) fragmentation (collision energy: 34%). Fragment ions were detected in the orbitrap at high resolution (50,000 FWHM) for a target value of 1,5x10⁵ ions and a maximum injection time of 150 ms using an isolation window of 0.7 Da and a dynamic exclusion of 20 sec. All raw data were viewed in Xcalibur v3.0 (Thermo Fisher Scientific).

All LC-MS/MS raw data files were searched in Proteome Discoverer (PD) version 2.5.0.400 (Thermo Fisher Scientific) using the SEQUEST HT search algorithm against the Caulobacter crescentus (strain NA1000/ CB15N) OX = 565050 protein FASTA database. The searches had the following criteria: enzyme, trypsin; maximum missed cleavages, 2; fixed modifications, TMTpro (N-terminal), TMTpro (K) and Carbamidomethyl (C).

The TMTpro reporter ion signals were quantified using S/N and they were normalized to the total peptide S/N in the PD program. The in-built ANOVA test in PD was used to generate p-values for all the proteins identified in the database searches with 1% FDR.

ChIP-Seq sample preparation and sequencing

Chromatin immunoprecipitation (ChIP) was performed essentially as described in Fumeaux et al [56] with the following modifications. Cultures were grown to mid-exponential phase in 80 ml PYE before formaldehyde crosslinking. Sonication was carried out in a Covaris ultrasonicator with the following settings: temperature 6°C, duration 3600 seconds, peak power 75.0, duty % factor 10.0, cycles per burst 200, average power 7.5, delay 60 seconds. All of the cleared sonicated lysate was used for ChIP without taking an input sample out first. HigX-bound DNA was recovered by immunoprecipitating with 20 µg purified anti-HigX antibody per sample, divided between two Eppendorf tubes. After immune complex recovery, reversal of crosslinks and protease treatment, DNA was recovered in 50 µl MilliQ water per Eppendorf and pooled into 100 µl, of which 40 µl was used for sequencing. DNA was prepared for Illumina sequencing using the NEBNext UltraExpress DNA library preparation kit (New England Biolabs) according to the manufacturer’s instructions and 50 bp paired-end sequencing was carried out on a NovaSeq 6000 (Illumina). Details of the sequence processing, alignment, quantification and data analysis are in the Supplementary methods, S1 Text.

Transposon mutagenesis, forward genetic screening and insertion mapping

A pooled library of transposon mutants was constructed in the ΔlexA ΔhigBA background by performing five independent conjugations of the transposon carrier plasmid pHPV414 from E. coli S17-1 λpir into the ΔlexA ΔhigBA recipient strain. Transconjugant colonies were selected on PYE plates with nalidixic acid and kanamycin after three days incubation at 30°C. Colony counting indicated that the library consisted of > 25’000 independent clones. Colonies were pooled in PYE and stored in aliquots at -80°C in 10% DMSO.

The ciprofloxacin resistance screen was performed by growing the pooled library to stationary phase and diluting it into medium containing 1 μg/ml ciprofloxacin and 20 ug/ml kanamycin, followed by plating on PYE plates containing 1ug/ml ciprofloxacin and 20 ug/ml kanamycin, and incubation at 30°C for five days. 40 colonies were re-streaked on PYE plates containing 1 μg/ml ciprofloxacin and 20 ug/ml kanamycin followed by incubation for 4 days to confirm growth.

The HigX overexpression tolerance screen was performed by growing the pooled library to stationary phase, preparing electrocompetent cells and electroporating with pMT335-higX. Plasmid-tolerating colonies were selected by plating electroporated cells on PYE plates containing gentamicin and 50 μM vanillate. Colonies were re-streaked to new gentamicin/vanillate plates to confirm stability of the phenotype.

Transposon insertion sites were mapped by plasmid rescue. Genomic DNA was extracted from the ciprofloxacin-resistant or HigX-tolerant transposon mutants, partially digested with SauIIIA and ligated. Ligations were used to transform E. coli EC100D, in which the transposon can be maintained as a replicating plasmid, and selected for kanamycin resistance. Plasmids were extracted and sequenced with the transposon-specific himup2 primer across the junction between the transposon insertion site and the flanking genomic DNA. To confirm specificity of the transposon insertions for the observed phenotypes, the transposon insertions were back-crossed into the ΔlexA ΔhigBA strain by bacteriophage ΦCr30-mediated generalized transduction. The resulting strains were tested for both ciprofloxacin resistance and tolerance of electroporation with pMT335-higX.

Sequence alignment and analysis

The genomic context conservation of HigX orthologues was investigated with FlaGs [34]. Transmembrane segment prediction was performed with the DAS TM prediction program [57] and the structure was modelled in Alphafold using default parameters [58]. Multiple sequence alignment of the region spanning CCNA_03134 to ctrA was performed with MAFFT [59] (default parameters) and visualized with Molecular Evolutionary Genetics Analysis version 11 [60] (MEGA version 11.0.13). Details of compositional bias analysis, comparative genomics and phylogenetic analysis of proteins and 16S RNA sequences are in the Supplementary Methods.

Statistical analysis

All numerical data are reported as mean of all biological replicates performed and error bars indicate the standard deviation. Statistical significance was analysed by non-paired equal variance 2-tailed Student’s T test for simple pairwise comparisons between strains or treatment conditions and by a Dunnett’s test in R (version 4.2.2) for multiple comparisons. Dunnett’s test in R was carried out by setting the wild-type strain (NA1000) as a reference and comparing all other strains against it. * signifies p < 0.05, ** signifies p < 0.01 and *** signifies p < 0.001 throughout.