NAL-induced adaptive resistance to β-lactams requires AcrAB-NodT

C. crescentus Δbla mutant cells lack the chromosomally encoded metallo-β-lactamase (MBL) and are, thus, unable to grow on plates supplemented with certain β-lactams, including the cephalosporin cephalothin (CEF, 10 μg/mL, CEF¹⁰) or the ureidopenicillin piperacillin (PIR, 40 μg/mL, PIR⁴⁰). To search for determinants conferring β-lactam resistance in cells lacking MBL, we conducted a transposon (Tn) mutagenesis of Δbla mutant cells, delivering the Tn by intergeneric conjugation from an E. coli donor. Initially, we counterselected the donor on plates containing NAL (20 μg/mL, NAL²⁰), but we unexpectedly observed that Δbla cells grew on PIR⁴⁰ or CEF¹⁰ plates supplemented with NAL²⁰. NAL is not known to affect β-lactam resistance directly, as it targets the A subunit of DNA gyrase (GyrA) in susceptible bacteria. However, since C. crescentus is naturally resistant to NAL owing to the polymorphism F96D in GyrA [9], we speculated that NAL activates a cryptic or an alternative β-lactam resistance mechanism in C. crescentus. To confirm that NAL confers β-lactam resistance in Δbla cells, we conducted Kirby–Bauer-based diffusion assays using β-lactam discs placed on plates with or without NAL²⁰. Indeed, NAL induces resistance of Δbla cells to various cephalosporins (Fig 1B: CEF, cefotaxime and cephalexin) and penicillins (PIR, ampicillin, and amoxicillin), yet it provided little protection against carbapenems (meropenem, imipenem, and doripenem).

To dissect the genetic basis for the NAL-induced resistance to these β-lactams, we mutagenized Δbla cells with a himar1 Tn-conferring kanamycin (KAN) resistance, counterselecting the Tn-delivering E. coli donor cells with colistin or aztreonam (instead of NAL) because C. crescentus is also naturally resistant to these antibiotics. We selected resistant mutants on plates containing KAN²⁰ and PIR⁴⁰ and mapped the Tn insertions of 2 different clones to the tipR gene (Fig 2A and 2C) [9,13]. Next, we backcrossed the tipR::Tn mutation into Δbla cells and compared the adaptive β-lactam resistance conferred by NAL in Δbla;tipR::Tn double-mutant cells to that of Δbla parental cells by Kirby–Bauer antibiotic disc diffusion assays (Fig 1B), as well as by growth on plates containing CEF¹⁰ or PIR⁴⁰ (Fig 2C). These tests revealed an identical β-lactam resistance spectrum between the Δbla strain grown in the presence of NAL and the Δbla;tipR::Tn strain without NAL, indicating that NAL responsiveness is lost without TipR (Figs 1B and 2C), the transcriptional repressor of P_acrA.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Deep mutational dissection of the tipR-acrAB-nodT locus.

(A) Positions of mutation in tipR and the acrAB-nodT operon conferring resistance to cephalothin (10 μg/mL) in Δbla cells. The list of mutations is a combination of sequences from the Mut-Seq experiment and single-clone sequencing. The line represents, from the tallest to the smallest, the loss of start codon, new stop codon, missense mutations, frameshift, and a synonymous variant. (B) Intergenic region between acrA and tipR. The inverted repeats region is a hot spot for mutations that confer resistance to cephalothin (10 μg/mL) to Δbla cells. The * represent SNPs; (α, σ, γ) are double mutations per strain and (i) insertion. Position of the putative −10 and −35 sequences was predicted using SoftBerry. (C) Resistance of Δbla cells and various mutant derivatives to cephalothin or piperacillin. Strains were straked plates containing 10 μg/mL or 40 μg/mL of cephalothin (CEF¹⁰ or CEF⁴⁰), or 40 μg/mL of piperacllin (PIR⁴⁰) with or without NAL (10 μg/mL, NAL¹⁰) and grown for 72 hours at 30°C. Note that ΔABT is used as abbreviation for the ΔacrAB-nodT mutation.

https://doi.org/10.1371/journal.pbio.3002040.g002

RNA-Seq analysis of NAL-induced cells revealed a 20-fold induction of acrAB-nodT transcripts compared to the (uninduced) control state (Figs 1CB and S2A and S1 Data) and a 7-fold induction of tipR transcripts, but also a 2-fold induction of transcripts encoding the DnaJ-like co-chaperone DjlA (CCNA_02218; see below). Moreover, several transcripts encoding components of respiratory chain/cytochrome biosynthesis pathway (CCNA_01767, CCNA_2850, CCNA_3065) were up-regulated between 1.9- and 1.6-fold. Thus, the transcriptional response to NAL predominantly affects the acrAB-nodT and tipR locus. No signature of a general transcriptional response to altered DNA supercoiling or DNA damage was apparent, a response that typically occurs in the presence of certain quinolone antibiotics. We hypothesized that the induction of the djlA transcript could reflect a direct interplay with TipR and AcrAB-NodT (see below), whereas the weak up-regulation of transcripts encoding respiratory proteins likely stems from increased respiratory demand owing to PMF consumption by elevated AcrAB-NodT-mediated efflux arising from TipR derepression.

To confirm that NAL interferes with TipR-mediated repression of P_acrA, we conducted β-galactosidase (LacZ)-based promoter probe assays (Fig 1D and S2 Data) in cells transformed with a P_acrA-lacZ transcriptional promoter probe plasmid (pP_acrA-lacZ). In wild-type (WT) cells, the addition of NAL lead to a 4-fold induction of P_acrA-lacZ activity (405%) relative to the reference state without NAL (set to 100%). By contrast, P_acrA-lacZ activity was 427% in tipR::Tn cells grown without NAL and remained unchanged upon the addition of NAL. Matching results were obtained by monitoring the accumulation of AcrA by immunoblotting using polyclonal antibodies to AcrA (Figs 1E and S3A). AcrA abundance was strongly induced upon the addition of NAL in WT cells. By contrast, AcrA steady-state levels are elevated in tipR::Tn cells and not further augmented by the addition of NAL. We conclude that NAL induction of P_acrA is likely governed through inactivation or removal of TipR. Moreover, and consistent with the induction of tipR mRNA by RNA-Seq, immunoblotting using polyclonal antibodies to TipR revealed a congruent accumulation of TipR and AcrA after 20 minutes during a time course analysis with NAL¹⁰ (S3B Fig), suggesting that acrA and tipR promoters are simultaneously derepressed upon the addition of NAL.

Next, we conducted ChIP-Seq experiments using polyclonal antibodies to TipR in pursuit of TipRs promoter occupancy in vivo. We discovered that TipR significantly binds 4 chromosomal targets in vivo. Importantly, TipR occupancy of these sites is reduced by ≈500% after exposing cells to NAL (for 30 minutes; Figs 1F and S2A and S3 Data), indicating that NAL antagonizes repression of P_acrA and P_tipR by TipR in vivo, allowing recruitment of RNAP (and TrcR [10]). The coordinated induction of AcrAB-NodT and TipR by NAL points towards a homeostatic control mechanism in which the adaptive and transient resistance to β-lactam antibiotics is conferred by induction of AcrAB-NodT, followed by TipR synthesis to reestablish the repressed ground state once the inducers have been expelled by AcrAB-NodT.

In support of this notion, we found that NAL no longer confers adaptive β-lactam resistance in Δbla ΔacrAB-nodT cells (Figs 2C and S1). While this result indicates that the adaptive resistance is mediated by NAL-based induction of AcrAB-NodT, the basal level expression of AcrAB-NodT in the uninduced state suffices to confer resistance to the fourth-generation cephalosporin cefepime. Such residual constitutive resistance by AcrAB-NodT is also detectable for low levels of cephalothin (first generation) and cefotaxime (third generation; S1B Fig). In the case of the former, we asked if AcrAB-NodT could even confer resistance to higher levels of cephalothin than the resistance arising from inactivation of tipR (CEF¹⁰). To this end, we conducted a second step selection for CEF⁴⁰ (cephalothin 40 μg/mL) resistance of Δbla;tipR::Tn cells and isolated several such mutants, each harbouring single missense mutations in acrB (Q177P, L241P, Q528H, P565A, E566K, N676Y, I762F; Figs 2C and S2B). We surmise that these mutations likely increase the specificity, capacity, or abundance of AcrAB-NodT to expel cephalothin (Fig 2C) and that they specifically point to AcrB as a key plasticity determinant enabling high-level MDR and the main element underlying NAL-dependent adaptive resistance (to cephalothin) in C. crescentus.

Ectopic long-term AcrAB-NodT induction perturbs envelope structure and function

Having dissected the role of AcrAB-NodT in adaptive β-lactam resistance, we probed for elevated resistance conferred by the tipR::Tn mutation compared to the parent by Kirby–Bauer-based antibiotic disc diffusion assays using a diverse panel of discs with different classes of antibiotics. We observed increased resistance toward the quinolones sparfloxacin and ciprofloxacin as well as to the macrolide antibiotics erythromycin and azithromycin, likely owing to increased efflux conferred by AcrAB-NodT overexpression (S3 Fig). Surprisingly, these experiments also revealed that tipR::Tn cells are more susceptible CW-targeting antibiotics such as vancomycin, teicoplanin, bacitracin, and fosfomycin (S4 Fig). As the first 3 molecules are large soluble antibiotics that are usually poorly active against gram-negative bacteria due to the OM, which prevents their entry into cells [21], we wondered whether envelope integrity might be compromised in tipR::Tn cells, facilitating the entry of these antibiotics. Indeed, phase contrast microscopy shows frequent envelope blebs and other cell deformations on tipR::Tn cells. While such blebs were seen less frequently when AcrAB-NodT induction occurred through NAL than through inactivation of TipR or through simple AcrAB-NodT overexpression form a plasmid (S4A Fig, red arrows), we suspect that NAL-dependent induction could be a transient phenomenon and/or the induction is attenuated because of the expulsion of NAL by AcrAB-NodT. In support of this idea, we found that the envelope deformations do not require the efflux activity of the pump, as cells exposed to the efflux pump inhibitor 1-(1-Naphthylmethyl)-piperazine (NMP) still harbour these deformations (S5A Fig). Thus, a strong induction and subsequent assembly of this trans-envelope structure is sufficient to disfigure cells.

To investigate the nature of those perturbations in detail, we imaged cells by cryo-electron tomography (cryoET). We imaged Δbla cells overexpressing AcrAB-NodT either via tipR::Tn mutation and via a multicopy plasmid carrying acrAB-nodT (pSRK-acrAB-nodT) coupled with native induction of AcrAB-NodT from the chromosomal locus by NAL¹⁰ for maximal overexpression. The cryo-electron tomograms of Δbla cells showed a canonical cell envelope architecture, with a continuous IM, OM, and a paracrystalline surface (S-layer) layer (S5B Fig). In contrast, Δbla;tipR::Tn and Δbla;pSRK-acrAB-nodT cells revealed abnormal cell shapes, manifesting in bulging of the entire cell envelope (S6A Fig). Furthermore, these mutants also showed significant perturbations of the structure of the cell envelope, including membrane blebs and diverse types of vesiculations of the membranes (S5B and S6 Figs). While the precise content and origin of those vesicles remains unknown, we suspect that they contain phospholipid- and/or LPS-derived OM material and/or soluble periplasmic content.

In summary, these structural defects are consistent with increased susceptibility toward antibiotics that are typically kept at bay by the OM, and they highlight the necessity for tight and transiently regulated control by TipR for an efflux system that should be induced only under life threatening conditions.

Genetic dissection of TipR and its target

To dissect the molecular mechanism of induction by TipR, we took advantage of the chemical-genetic relationship between AcrAB-NodT induction and CEF¹⁰ resistance. With the goal of discovering loss-of-function mutations in tipR, or potentially its (unknown) target sequence in P_acrA, we isolated spontaneous CEF¹⁰-resistant point mutants by plating Δbla cells on CEF¹⁰ plates. After collecting approximately 10,000 CEF¹⁰-resistant Δbla colonies, we PCR amplified tipR and acrAB-nodT from the mutant pool and deep sequenced the PCR products. Sequence analysis (Fig 2A and S4 Data) of this Mut-Seq dataset indeed revealed missense, nonsense, and frameshift mutations scattered throughout tipR, with a slight hotspot in the region predicted to encode the N-terminal DNA-binding domain. The fact that nonsense and frameshift mutations are also found in the C-terminal half of the protein indicates that these residues are also important for TipR function. Interestingly, we also found 14 (gain-of-function) mutations within the acrAB-nodT coding sequence that presumably increase the stability, structure, and/or affinity of AcrAB-NodT components towards CEF (matching the CEF⁴⁰ experiments described above).

Next, we isolated individual CEF¹⁰-resistant Δbla mutant clones from this pool and sequenced their genomes. Three TipR missense mutants were chosen for further study, each encoding a different mutation in TipR:S53R (in the DNA-binding domain), as well as E119V and L204Q, respectively in the central region and C-terminal domain (Fig 2A, corresponding to mutants Δbla;tipR-S53R, Δbla;tipR-E119V, and Δbla;tipR-S53R). We confirmed the strong increase in P_acrA activity in all these mutants using pP_acrA-lacZ-based assays (S7 Fig and S2 Data).

In addition to these loss-of-function mutations in TipR, we also obtained clones with a mutation in the operator site for the promoter of tipR (P_tipR, since tipR and acrAB-nodT are transcribed divergently from the same intergenic region; Fig 2B) that would prevent TipR from binding and repressing P_acrA. Since the TipR-binding site sequence is shorter than the 615-nucleotide tipR gene, mutations in the former would likely surface less frequently than nonsense mutations in tipR. Nonetheless, genome sequencing of CEF¹⁰-resistant mutants revealed 2 strains with mutations in P_acrA (strains Δbla P_acrA-RIR and Δbla P_acrA-IR^up), attesting to the strength and specificity of the selection. To favour the enrichment of additional loss-of-function mutations in the TipR-binding site (Fig 2B), we modified our starting genotype for a revised screen. To this end, we conducted the CEF¹⁰-resistant selection with Δbla cells carrying a multicopy plasmid expressing TipR (pSRK-Gm-tipR) to eliminate mutants with a dysfunctional tipR gene that would otherwise surface. With this strategy, we uncovered strains with additional P_acrA mutations, 80% of which reside in a region of dyad symmetry comprising an 11-base pair inverted repeated (IR, 5′-TGAGAATGAAC-3′; Fig 2B), 2 adjacent mutations are 11 nucleotides upstream of the IR (IR^up mutation), and another mutation lies 33 nucleotides downstream of the IR (mutation IR^dw).

Lastly, we chose the following IR mutant strains for follow-up confirmation studies: Δbla P_acrA-MIR (middle inverted repeat), Δbla P_acrA-RIR (right inverted repeat), Δbla P_acrA-2IR (2 mutations inverted repeat) and Δbla P_acrA-IIR (insertion inverted repeat; Fig 2B), as well as one with a mutation upstream of the IR (Δbla P_acrA-IR^up) and one downstream of it (Δbla P_acrA-IR^dw; Fig 2B). Immunoblotting using polyclonal antibodies to AcrA and TipR confirmed that AcrA is overexpressed in all selected mutants compared to the Δbla parent (S8A–S8D Fig), explaining why these mutants grew on CEF¹⁰ plates. TipR was also overproduced in all mutants to similar levels, except for strains Δbla P_acrA-IR^up and Δbla tipR-L204Q (S8A and S8B Fig) in which the TipR levels resemble the basal amount of the Δbla parental strain. The addition of NAL led to the induction of TipR in these 2 mutants, albeit to lower steady-state levels compared to Δbla cells, whereas AcrA levels were not substantially affected. This result suggests that the effect of the mutation differs for P_acrA versus P_tipR and/or that NAL affects the abundance of TipR through a post-translational mechanism. Below, we provide additional evidence for the latter model.

In summary, the strong selection for CEF¹⁰ resistance in Δbla cells unveiled mutational hotspots in the region predicted to be required for DNA binding by TipR, but also in the IR of in the suspected promoter target of TipR.

Defining the TipR target sequence and functional residues in TipR

We used electrophoretic mobility assays (EMSAs) to confirm that TipR binds the IR in vitro. In these experiments, a Cy5-labelled P_acrA probe centred on the IR was mixed with recombinant (untagged) TipR purified in 2 steps from an E. coli overexpression strain (see Experimental procedures). We observed a retardation of the labelled probe by TipR in a concentration-dependent fashion, even in the presence of nonspecific calf thymus competitor DNA (Fig 3A). EMSA experiments with E. coli extracts containing WT TipR also revealed retardation of Cy5-labelled P_acrA compared to E. coli extracts without TipR. Next, we probed E. coli extracts each containing a different TipR mutant described above (E119V, S53R, or L204Q) for binding of P_acrA by EMSA. While the L204Q variant retards P_acrA akin to WT TipR, the E119V or S53R mutations abolish P_acrA binding (Fig 3A). Immunoblots conducted with C. crescentus cell extracts revealed the E119V and S53R variants to accumulate to the same steady-state level as WT TipR upon NAL induction (S8A Fig), indicating that the E119V and S53R mutations promote DNA binding. By contrast, the abundance of TipR-L204Q is severely reduced, even in the presence of NAL (S8A Fig), suggesting that this mutation interferes with protein stability rather than with DNA-binding activity as shown by antibiotic chase experiments, revealing a reduced half-life of TipR-L204Q (Fig 3B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Determinants in TipR and P_acrA required for repression.

(A) EMSA with 200 ng of a Cy5-labelled acrA promoter (P_acrA) probe in the presences of increasing concentration of cell extract from E. coli overexpressing different mutated TipR protein from 0.031 μM to 16 μM. (B) Immunoblot showing the result of a 1-hour chase experiment using chloramphenicol (5 μg/mL) on exponential NA1000 (WT) and Δbla;tipR-L204Q cells in PYE. (C) EMSA with 200 ng of different Cy5-labelled P_acrA mutants analysed for retardation by gradual concentration of purified TipR protein from a concentration of 0.0156 μM to 16 μM. (D) EMSA with 100 ng of Cy5-labelled acrA with gradual concentration of heparin-purified TipR protein from a concentration of 0.25 μM to 8 μM. (E) EMSA with 100 ng of Cy5-labelled djlA, ccrM, and acrA promoters (P_djlA, P_ccrM, and P_acrA) with gradual concentration of purified TipR protein at a concentration of 1.125 μM to 18 μM. (F) Activity of β-galactosidase of the P_djlA-lacZ or P_ccrM-lacZ in WT C. crescentus. The « + » indicates induction by Nal (10 μg/mL) for 2 hours. All levels are indicated in percentage of expression regarding the basal level of NA1000 (WT) without induction. The data from the analysis are deposited in S2 Data. EMSA, electrophoretic mobility assay; PYE, peptone-yeast extract; WT, wild-type.

https://doi.org/10.1371/journal.pbio.3002040.g003

Having identfied determinants in TipR that promote DNA binding, we next determined the effects of the isolated promoter mutations. To this end, P_acrA promoter mutants with alterations in the IR were labelled with Cy3 and incubated with purified WT TipR in EMSAs (Fig 3C). While P_acrA-RIR is no longer retarded by TipR in EMSAs, we only observed a partial loss of binding by TipR to P_acrA-MIR. As expected, TipR still retarded the P_acrA-IR^up probe that carries a mutation outside the IR. Thus, these EMSAs validate the requirement of the IR as the likely TipR docking site in P_acrA. With these results, we investigated further how P_acrA-IR^up can cause overexpression of the pump without crippling DNA binding of TipR. EMSAs with a new Cy5-P_acrA probe combining the P_acrA-RIR mutant (that abolish binding of TipR) with the P_acrA-IR^up mutation, EMSA (Fig 3D) revealed that P_acrA-IR^up increased the affinity of TipR for the P_acrA-RIR probe, suggesting that P_acrA-IR^up mutation creates a secondary docking site upstream of the IR.

Indeed, the consensus target sequence derived for TipR (S9 Fig) resembles the motif created by the P_acrA-IR^up mutation. In search for additional binding sites of TipR, we conducted ChIP-Seq experiments (S3 Data) and found that in vivo TipR binds 4 sites on the chromosome: the promoter of acrAB-nodT and tipR, the promoter of the djlA gene predicted to encode a DnaJ-like co-chaperone, as well as a site upstream of the DNA methyltransferase gene ccrM and another site in between the qor and rho genes oriented towards each other. The RNA-Seq experiments described above confirmed the NAL induction of 3 genes neighbouring two of the TipR binding sites (acrAB-nodT, tipR, and djlA), and ChIP-Seq experiments conducted with NAL-treated cells revealed that the binding of TipR to all 4 chromosomal sites is impaired in vivo (Figs 1F and S10). However, the binding of TipR to P_acrA was approximately 250% more efficient than to the other sites, suggesting that the IR in P_acrA represents the preferred recognition sequence of TipR. Nonetheless, we scanned the other target sites for sequences comparable to the IR and detected one upstream of djlA and of ccrM, delivering the consensus sequence 5′-WTGaGWMtGAWC-3′ by MEME analysis (S9 Fig). Using this consensus, we analysed the intergenic sequence between the qor and rho genes and found a single sequence that could explain the presence of TipR at this location; however, the configuration of those genes facing each other makes a TipR-dependent transcriptional regulation of these 2 genes unlikely.

Next, we tested the binding ability of TipR to djlA and ccrM promoters in vitro by EMSA using purified recombinant TipR. As seen in Fig 3E, TipR was able to retard the djlA promoter, but not the promoter of ccrM. Moreover, LacZ-based promoter probe assays with P_djlA-lacZ and P_ccrM -lacZ reporters revealed an induction of P_djlA in WT cells in response to NAL, but no significant induction of P_ccrM-lacZ (Fig 3F and S2 Data). Since TipR binds P_ccrM weakly in vivo, likely because of the divergence of its IR from the preferred TipR target sequence, the weak interaction might result in a TipR-promoter complex too fragile to be maintained during electrophoresis. Moreover, in the absence of a significant induction of transcription by NAL in our RNA-Seq experiment (Fig 1C), we speculate that TipR is a poor repressor of P_ccrM in vivo, yet it clearly down-regulates P_djlA.

Screening for chemical inducers of P_acrA

Having identified the IR of P_acrA as the target of TipR’s DNA-binding activity, we engineered a reporter for rapid screening of chemical libraries for compounds that induce P_acrA (independent of requirement of AcrAB-NodT efflux function as in CEF¹⁰ resistance screen above). To this end, we fused to the promoterless nptII kanamycin resistance gene to P_acrA, creating a convenient readout for molecules like NAL that can activate P_acrA in vivo scored as kanamycin resistance. We transformed the resulting P_acrA-nptII promoter probe reporter plasmid (pP_acrA-nptII) into WT cells, embedded the resulting reporter cells in soft agar containing kanamycin (10 μg/mL, KAN¹⁰) overlaid on KAN¹⁰ plates, and then spotted 4 μL drops of each compound (at 10 μM) of the Maybridge chemical library for inducers of P_acrA on these seeded indicator plates. Small molecule inducers were identified by their ability to promote growth of cells around the disc owing to activation of P_acrA-nptII by the compound(s) (S11 Fig). This chemical screen along with tests of candidate compounds reported in the literature for other systems, unearthed 14 different inducers of P_acrA-nptII (Figs 4A and S11) that were subsequently grouped into 2 classes based upon pP_acrA-lacZ inducer strength as determined by LacZ activity measurements. One group of strong inducers (triggering P_acrA-lacZ activity exceeding 230% activity relative to WT) includes the quinolones NAL, flumequine (FLU), and sparfloxacin, quinolone precursors such as chloroquine and dyes such as rhodamine 6G (Rh6) and crystal violet (CV). The weak inducers include the dye malachite green (MG), DNA intercalants such as acridine orange (AO), ethidium bromide (EtBr), and SYBR safe, but also various other compounds such as glafenine and doxylamine. We also noticed that dopamine and aminolevulinic acid fail to induce P_acrA-lacZ after 3 hours; however, they allowed growth on kanamycin plates with P_acrA-nptII after overnight incubation (Figs 4A and S11 and S2 Data), suggesting that they act slowly, through another pathway, or that they counteract kanamycin in an alternative way.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Different chemicals relieve repression of P_acrA by TipR.

(A) Expression of β-galactosidase assayed from P_acrA-lacZ in NA1000 (WT) after 3 hours of induction by different compounds. All concentrations are in μg/mL. All levels are indicated as percentage of expression regarding the basal level of the control: NA1000 (WT) without induction. The data from the analysis are deposited in S2 Data. (B) Immunoblot revealing the abundance of TipR during a 2-hour chase experiment with chloramphenicol (5 μg/mL) added to exponential tipR::Tn;pSRK-tipR cells grown in PYE or and without selected inducers: Nal (10 μg/mL), Flu (10 μg/mL), Rh6 (2 μg/mL), CV (2 μg/mL), EtBr (2 μg/mL), and MG (2 μg/mL). (C) EMSA using 200 ng of Cy5-tagged DNA of the promoting region of acrA with 8 μM of heparin-purified TipR, supplemented with an increasing concentration of selected inducers. All quantities are in μg/mL. (D) β-galactosidase measurements in C. crescentus WT and different mutants containing P_acrA-lacZ before or after induction with Nal (10 μg/mL), Rh6 (Rho, 2 μg/mL Rho), CV (2 μg/mL), and MG (2 μg/mL) for 2 hours. All measurements are expressed as percentage of expression relative to basal activity of the control: WT or ΔsocB cells before induction. The data from the analysis are deposited in S2 Data. (E) Immunoblot analysis of cell extracts from different protease mutants performed with polyclonal antibodies to TipR and to AcrA. All inductions (+) were performed after 2 hours of induction with Nal (10 μg/mL). Shown as loading control are blots with polyclonal antibodies to CCNA_00163. (F) Immunoblot following a 2-hour chase experiment with chloramphenicol (5 μg/mL) on exponential cells of WT with a clpP* inserted at the P_xyl locus. Cells were grown in PYE, followed by the addition of glucose or xylose, and, subsequently, the inducer was added: Nal (10 μg/mL), and, finally, protein synthesis was blocked with chloramphenicol. CV, crystal violet; EMSA, electrophoretic mobility assay; EtBr, ethidium bromide; Flu, flumequine; MG, malachite green; Nal, nalidixic acid; PYE, peptone-yeast extract; Rh6, rhodamine 6G; WT, wild-type.

https://doi.org/10.1371/journal.pbio.3002040.g004

Some of the identified molecules are known substrates of the enterobacterial AcrAB-TolC RND pump and would thus likely be exported to reduce the active concentration in cells. Hence, we wondered if induction of P_acrA would be accentuated in cells lacking AcrAB-NodT (S12A Fig and S2 Data) that can no longer expel these molecules, resulting in their accumulation to higher intracellular levels. While the level of P_acrA induction with compounds like NAL or EtBr was similar in WT versus ΔacrAB-nodT cells, AO clearly showed a 4-fold elevated level of induction compared to the 2-fold induction in WT cells. Thus, the intracellular concentration modulates the level of acrAB-nodT expression. To eliminate the possibility that the compounds cause a nonspecific global increase of gene expression or LacZ enzymatic activity, we tested P_bla-lacZ reporter expressing LacZ expressed from the constitutive promoter of the C. crescentus MBL gene CCNA_02223 and found that these compounds did not alter LacZ activity from P_bla-lacZ (S12B Fig and S2 Data).

Different inducers control TipR at the post-translational level

To determine if the new inducers act in the same way on TipR as NAL, we first explored if they affect TipR stability in vivo. To this end, we uncoupled TipR synthesis from its inducible transcription by expressing it from the IPTG-controllable P_lac promoter (of E. coli) on plasmid pSRK-tipR in tipR::Tn cells. AcrAB-NodT is still inducible from P_acrA in these cells, but TipR synthesis is independent from the tester compounds such as NAL, Rh6, and MG (S13A and S13B Fig). We then compared the half-life of TipR in these cells after exposing them to the chemical inducers. To stop translation of TipR, we treated cells with high levels of the protein synthesis inhibitor chloramphenicol, a phenicol antibiotic that is (likely) not a substrate of the AcrAB-NodT efflux pump. In the presence of the quinolones NAL or FLU, TipR is rapidly turned over (Fig 4B), reminiscent of the instability observed for TipR-L204Q. Such an induced instability was not observed with the other P_acrA-activating compounds.

Next, we examined the effect of these compounds on the DNA binding ability of purified TipR by EMSA (Fig 4C). While at physiologically relevant concentrations, NAL or FLU did not perturb retardation of the Cy-labelled P_acrA, other inducers such as EtBr or MG interfered with binding of TipR to P_acrA. Rh6 and CV affect TipR binding in the same manner, but less efficiently in vitro compared to EtBr or MG. To assess the specificity of these compounds on TipR, rather than interfering with protein–DNA interactions in a general fashion, we conducted control EMSA using the histone-like IHF protein and its target DNA sequence attR (S14 Fig) as probe. Increasing concentration of CV did not prevent IHF binding even at a concentration sufficient to change the charge of the probe resulting in its upward migration in the gel during electrophoresis. Therefore, CV, Rh6, EtBr, and MG apparently interfere with TipR’s ability to bind the IR in P_acrA, explaining the derepression of P_acrA by these compounds in vivo.

To rule out the possibility that induction of AcrAB-NodT by NAL or FLU acts through the SOS (DNA damage) response by corrupting DNA gyrase (encoded by gyrAB) as in E. coli, we confirmed that inhibition of GyrA by the fluoroquinolone ciprofloxacin (CIP) or of GyrB by the aminocoumarin novobiocin (NOV) alone did not induce of P_acrA (S15 Fig and S2 Data). Moreover, NAL and CIP did not synergise to enhance P_acrA induction, yet NAL and NOV together resulted in elevated P_acrA-lacZ induction. Consistent with these findings, we next assayed P_acrA-lacZ activity in WT cells expressing an additional GyrA variant: GyrA from Brucella melitensis or a mutant variant of C. crescentus GyrA, GyrA*(F96D) that is NAL sensitive and causes DNA damage and the SOS response in the presence of NAL. C. crescentus WT cells expressing either of these GyrA variants from pMT335 still induce P_acrA upon the addition of NAL (S15 Fig and S2 Data), even more efficiently than WT cells. In fact, the level of NAL induction in cells expressing the NAL-sensitive forms of GyrA attains the level of induction when NAL and NOV are added jointly to WT cells, indicating that Gyrase inhibition can enhance AcrAB-NodT induction, but only after repression of P_acrA by TipR has been relieved by the addition of NAL.

Prompted by the observation that, at physiological concentrations, NAL and FLU reduce the half-life of TipR without interfering with TipR binding to P_acrA in vitro in EMSAs (Fig 4B), we speculated that a transient destabilization by a dedicated protease could account for the induction of AcrAB-NodT. Using a candidate approach, we then tested for protease(s) conferring instability of TipR by immunoblotting using polyclonal antibodies to TipR and AcrA (S16 Fig). While no significant changes in AcrA or TipR steady-state levels was observed in ftsH and lon mutant cells compared to WT cells, loss of ClpP or ClpX led to an accumulation of AcrA and a reduction in TipR levels in the absence of NAL. Indeed, previous work showed that AcrA and AcrB are substrates of the ClpP degradation machinery [22]. Yet, AcrA and TipR are still inducible upon exposure of clpP and clpX mutant cells to NAL (Figs 4E and S13), but promoter probe assays using the P_acrA-lacZ reporter revealed a strong reduction of LacZ induction when clpP and clpX mutant are exposed to NAL, Rh6, CV, or MG compared to WT cells (Fig 4D and S2 Data). Taken together, our findings suggest that the ClpXP protease promotes high level of induction of acrAB-nodT and tipR, whereas inactivation of the clpA chaperone gene in WT cells did not cause a major AcrA or TipR abundance before after induction with NAL (Fig 4E).

To test if ClpP controls the half-life of AcrA and TipR, we conducted antibiotic chase experiments in cells expressing a dominant negative version of ClpP (ClpP*) in which the catalytic serine is mutated to alanine (Figs 4F and S11A). This ClpP* variant is expressed from the xylose-inducible P_xyl promoter at the xylX locus (xylX::P_xyl-clpP*). Immunoblotting using antibodies to TipR and AcrA revealed that TipR is turned over normally upon the addition NAL in the presence of glucose (to repress expression of ClpP* from P_xyl). However, when cells are grown in PYE containing xylose to induce of the dominant negative ClpP* variant), TipR stability is increased. Lastly, to demonstrate the interaction of TipR and AcrA with ClpXP biochemically, we used a GFP Trap matrix to pull down a ClpX-YFP fusion protein from lysates prepared from WT xylX:.P_xyl-ClpX-YFP cells before or after induction with NAL. Immunoblotting revealed the presence of TipR and AcrA in the pulled down material (S17 Fig) regardless of the presence of NAL.

In summary, while some compounds dislodge TipR from the IR sequence, others like first-generation quinolones act to destabilize TipR. The ClpXP protease confers partial instability of TipR in the presence of NAL, and it also reduces TipR and AcrA steady-state levels in the absence of NAL. Lastly, we discovered that ClpXP is required for efficient transcriptional induction of acrAB-nodT.

Co-induction and interaction of DjlA and AcrAB-NodT

Since DjlA, a putative DnaJ-like co-chaperone, is co-induced with AcrAB-NodT and TipR, we investigated whether DjlA interacts with TipR or AcrAB-NodT or affects their function. We first examined whether DjlA overexpression alters induction of P_acrA or TipR and AcrA steady-state levels upon addition of NAL. Neither P_acrA-lacZ-activity nor AcrA or TipR abundance was affected as a function of DjlA, regardless of whether NAL was present or not (Figs 5A and S18 and S2 Data). However, when we probed for AcrAB-NodT-mediated efflux activity, we found that ectopic expression of DjlA enhances resistance to several β-lactam antibiotics belonging to the cephalosporin and penicillin substrates of AcrAB-NodT (S19 Fig). To assess if DjlA also promotes AcrAB-NodT-dependent efflux of other substrates, we conducted survival (efficiency of plating (EOP)) assays on plates containing EtBr⁶ (EtBr, 6 μg/mL) and found that cells ectopically expressing DjlA (from pMT335-djlA) exhibited an elevated EOP compared to WT cells with the empty vector (Fig 5B). The positive effect of DjlA was also observed by a reduction of ΔdjlA cells, survival on EtBr⁴ (EtBr, 4 μg/mL) plates. This effect was restored in the presence of a complementing plasmid (pMT335-djlA) but completely lost when AcrAB-NodT was inactivated (i.e., in ΔacrAB-nodT cells; Fig 5B and 5C). Together, these results support the view that the activity or assembly of the pump, and thus the ability to expel EtBr, is enhanced by DjlA. Interestingly, forcing overexpression of AcrAB-NodT in ΔdjlA cells by introducing the tipR::Tn mutation or by the addition of NAL can mitigate the reduced EOP on EtBr⁴ plates, likely because AcrAB-NodT levels are no longer limiting for expulsion of EtBr to enable growth.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Role of DjlA co-chaperone in the TipR-dependent control loop.

(A) Measurement of β-galactosidase expressed from P_acrA-lacZ in different C. crescentus ΔdjlA mutants before or after a 2-hour induction with Nal [10 μg/mL (+) and/or vanillate 100 μM (V)]. All measurements are indicated as percentage relative to the basal activity of the WT control before induction. The data from the analysis are deposited in S2 Data. (B) EOP assay determined by 10-fold serial dilutions of ΔdjlA::pNPTS and ΔacrAB-nodT mutants on plates containing EtBr (6 μg/mL) to probe for efflux pump activity. All strains contain pMT335 plasmid or a derivative (pMT335-djlA) expressing DjlA, grown on Van at 100 μM and Gent 1 μg/mL. (C) EOP assay by 10-fold serial dilutions of ΔacrAB-nodT and ΔdjlA::pNPTS mutants on plates with EtBr (4 μg/mL) to probe for pump efficiency. Nal induction was performed at 10 μg/mL, Van at 100 μM, and Gent in μg/mL. (D) ChIP-Seq analysis of DjlA-HA expressed from the xylX locus in WT cells before and after NAL induction. DjlA-HA was pulled down using an anti-HA affinity matrix. The acrA-tipR promoter region was the only region showing a >2.9-fold increase in abundance compared to the control ChIP-seq performed with cells that were treated with NAL for 30 minutes or not treated. The data from the analysis are deposited in S3 Data. (E) Immunoblot analysis of aggregated cellular proteins in various strains: A = ΔdjlA::pNPTS, B = WT, and C = WT pMTdjlA. Inductions (+) were performed for 2 hours with Nal (10 μg/mL). The total fraction contains the cell lysate. The insoluble fraction corresponds to the material that remained in the pellet after several extractions with 1% Triton X-100. Strain C was grown in PYE with Gent (1 μg/mL) and Van (100 μM). (F) AcrAB-NodT enables growth of C. crescentus in the vicinity of an unknown fungus. C. crescentus strains grown in contact with an unknown fungus on a PYE plate. The fungus was pregrown on the plate for 2 weeks, after which the C. crescentus strains were streaked from the edge of the plate towards the centre until in contact with the fungus. The plate was then incubated at 30°C for another 72 hours. The fungus had been isolated in the laboratory as a contaminant on a PYE plate. AcrAB-NodT encoded in the C. crescentus chromosome protects against (a) compound(s) produced by the fungus to enable growth of C. crescentus in the vicinity of the fungus. (G) Proposed model of the acrAB-nodT operon and djlA gene regulation, induction and function. EOP, efficiency of plating; EtBr, ethidium bromide; Gent, gentamicin; Nal, nalidixic acid; PYE, peptone-yeast extract; Van, vanillate; WT, wild-type.

https://doi.org/10.1371/journal.pbio.3002040.g005

Since co-chaperones like DjlA typically assist in protein folding and/or disaggregation using the conserved DnaK chaperone, we constructed 2 DjlA variants with mutations in the HPD domain (H187A and H187Q) that is necessary for proper DnaK ATPase activity. Those 2 constructs were put under the control of the inducible P_van promoter on the pMT335 plasmid and transformed in strains lacking the endogenous djlA gene. The complementation assay performed on EtBr⁶ agar plate demonstrates that the unmodified allele can complement the absence of the chromosomally encoded DjlA, while neither DjlA H187A or H187Q restore the phenotype (S20 Fig). Hence, the DnaK proper function is fully required for DjlA co-chaperone activity.

We reasoned that co-induction of DjlA and AcrAB-NodT by NAL may serve to manage a stress condition (for the envelope) imposed by the massive synthesis and assembly of the AcrAB-NodT structure, perhaps facilitating disaggregation of AcrA before secretion from the cytoplasm where DjlA is located. In support of this idea, pull-down experiments (S21A Fig) using a DjlA variant with a C-terminal HA-tag (DjlA-HA) revealed that DjlA directly or indirectly associates with AcrA and TipR. We conducted a similar experiment using the DjlA-H187A-HA variant in which a key residue affect the DnaK chaperone cycle is inactivated and found that the interaction of this DjlA-HA variant with AcrA and TipR still occurs, suggesting that the interaction of DjlA with AcrA and TipR does not depend on DnaK function (S21B Fig).

The interaction between DjlA-HA and TipR was further confirmed in ChIP-Seq (chromatin immunoprecipitation) experiments (Figs 5D and S22) revealing DjlA-HA can associate with TipR at the P_acrA chromosomal site in vivo following exposure to NAL. In these experiments, crosslinked chromatin was prepared from DjlA-HA expressing cells after NAL treatment, followed by immunoprecipitation of DjlA-HA with bound chromatin using an anti-HA affinity matrix. Deep sequence analysis of the precipitated chromatin revealed a near 3-fold increase in P_acrA abundance when DjlA-HA was immunoprecipitated from NAL-treated cells versus untreated cells. No association with P_acrA was observed in control ChIP-Seq experiments with other HA-tagged DnaJ-like proteins. Moreover, the specific occupancy of DjlA-HA at P_acrA required the presence of TipR (S22 Fig), as revealed by ChIP-seq experiments conducted in which tipR::Tn cells expressing DjlA-HA from the xylX locus had been treated with NAL. This result indicates that DjlA-HA specifically occupies P_acrA, that this association requires TipR, and that it is enhanced by NAL.

Finally, knowing that DjlA can interact with TipR and AcrA, we investigated whether DjlA also affects aggregation of TipR or AcrA. To test this, we first induced WT cells, ΔdjlA cells, and DjlA-overexpressing cells (WT+ pMT335-djlA) with NAL and then separated extracts into soluble fractions and those that remain insoluble in Triton X-100 by centrifugation (Figs 5E and S23). Immunoblotting revealed approximately 40% less aggregated AcrA in the insoluble fraction from WT cells overexpressing DjlA after NAL¹⁰ induction, while conversely, there was a 40% increase of AcrA aggregates induced by NAL¹⁰ in ΔdjlA cells. By contrast, no TipR was detectable in the insoluble fraction. To determine if DjlA also acts on other (envelope) proteins, we conducted LC-MS/MS (liquid chromatography followed by tandem mass spectrometry) analyses of the DjlA-HA pull-downs described above (S24 Fig). Pull-downs of DjlA-HA expressing cells grown in the absence of NAL revealed 8 of the 14 most expressed TonB-dependent receptors (TBDRs) and porins destined for the OM as clients of DjlA. In the presence of NAL¹⁰, there was a drastic reduction in the number of OM proteins (OMPs) with a concomitant increase in abundance of AcrA, AcrB, and NodT, suggesting that DjlA has higher affinity for efflux pump components than for OMPs, which is consistent with its role in enhancing efflux activity by preventing its aggregation.

Growth conditions

All C. crescentus strains were cultivated in peptone-yeast extract (PYE) and incubated at 30°C. ФCr30-mediated generalized transductions were done as described [30]. E. coli strains were grown in lysogeny broth at 37°C. All media were supplemented with the appropriate antibiotics or indicated inducer. The strain used for cloning is E. coli EC100D, while the strain used for protein expression is E. coli BL21(DE3). As both WT (NA1000) and Δbla C. crescentus cells are naturally resistant to colistin (COL, 4 μg/mL) or aztreonam (AZT, 3 μg/mL), COL or AZT was used to counterselect E. coli Tn delivery strains for Tn (himar1) mutagenesis encoded on plasmid pHPV414 [31]. The himar1 encodes resistance to kanamycin, and, therefore, the PYE plates were supplemented with kanamycin (20 μg/mL, KAN²⁰) and CEF¹⁰ or PIR⁴⁰. Strains, plasmids, primers, and synthetic genes are detailed in S6 Data.

β-galactosidase assays

All β-galactosidase assays were done using freshly electroporated strains harboring the pLac290-derived promoter probe plasmids to quantify the activity of the transcriptional fusions between the studied promoters and lacZ. The assays were done at room temperature. Between 50 and 100 μL of cells (OD₆₀₀nm of 0.3 to 0.7) were lysed in 30 μL of chloroform and vigorously mixed with Z buffer (60 mM Na₂HPO₄; 40 mM NaH₂PO₄; 10 mM KCl and 1 mM MgSO4 (pH 7)) to obtain a final volume of 800 μL. Followed by the addition of 200 μl of ONPG (Ortho-nitrophenyl-β-D-galactopyranoside, at 4 mg/mL in 0.1 M potassium phosphate (pH 7)) to begin the reaction. Assays were stopped using 500 μL of 1 M Na₂CO₃ when the solution turned light yellow. The OD₄₂₀nm of the supernatant was collected and used to calculate the Miller units as follows: U = (OD₄₂₀ * 1,000) / (OD₆₀₀ * t(min) * v(ml)). Error was calculated as standard deviation from at least 3 biological replicates (S2 Data).

Kirby–Bauer disk diffusion susceptibility test

In a 14-cm diameter Petri dish, 50 mL of 1.5% agar media supplemented with indicated antibiotic and/or inducer were cast. After the polymerization, 12 mL of 0.375% agar media maintained at 50°C were mixed with 400 μL of overnight bacterial culture prior to be spread evenly on top of the solid media. Antibiotics discs were from Bio-Rad or made in the lab using sterile disks. The pictures were taken after the plates were incubated overnight at appropriate temperate. Images were captured on a Bio-Rad illuminator and with the Image Lab 4.1 software.

Efficiency of plating (EOP) assays

According to the OD₆₀₀, the same quantity cells from an overnight culture were placed in the first line of a 96-well plates. All the wells were filled with 180 μL of appropriate media except the first line that was adjusted to 200 μL. The dilutions were done with 20 μL of the first line of wells transferred to the following line using a multipipette; this step was repeated until the 8 lanes were done. Finally, 5 μL of each dilution was dropped on a Petri dish with the solid media supplemented with the appropriate chemical(s).

Microscopy

Around 2 μL of an exponential phase culture of C. crescentus grown in PYE supplemented with indicated chemicals were spotted on a thin layer of 1% agarose pad for immobilization. Contrast microscopy pictures were taken using a phase 100× objective with an oil interface (Zeiss, alpha plan achromatic 100×/1.46 oil phase 3) using an Axio Imager M2 microscope (Zeiss), with appropriate filter (Visitron Sys-tems GmbH) and a cooled CCD camera (Photometrics, CoolSNAP HQ2) controlled via the Metamorph software (Molecular Devices).

Electrophoretic mobility shift assay (EMSA)

Cy5-labeled probes were generated by PCR with primers chemically modified with the fluorescent dye. The probes were purified by agarose gel electrophoresis. The reaction took place in a buffer containing: 40 mM Tris-HCl (pH 7.6), 60 mM KCl, and 0.1% glycerol, 240 μg of calf DNA, 800 μg of BSA, and 200 ng of labeled probe. After the addition of the indicated quantity of purified protein, the samples were placed in the dark for 30 minutes at room temperature. After a prerun of 30 minutes, the samples were loaded and migrated by electrophoresis in a nondenaturing 4% Tris-Bore-EDTA acrylamide (19:1) gel. After migration, the samples were observed in a Bio-Rad illuminator (Chemidoc MP) using the preinstalled parameters for Cy5 imaging; pictures were collected through Image Lab 4.1 software (Bio-Rad).

Chemical library screening

The Maybridge Chemical Library was delivered aliquoted into 96-well plates and dissolved in 50% DMSO. Petri dish of 22 cm containing PYE kanamycin 10 μg/mL were prepared using the Kirby–Bauer technique described above, with a NA1000 strain carrying the pLac290 plasmid harboring a fusion between the P_acrA and the nptII kanamycin resistance gene. Potential inducers were detected when a growth area occur at the position of a drop after 48 hours of incubation at 30°C.

Immunoblot analysis/chase

The appropriate strain was grown for 2 to 4 hours at 30°C under constant agitation up to OD_600nm of 0.4 to 0.6, then the inducer was added, and the cells were grown for 2 additional hours. For chase experiment, the strain was grown similarly up to OD_600nm of 0.3 to 0.5, then chloramphenicol 5 μg/mL was supplemented with vigorous mixing, immediately followed by the addition of mentioned inducer. Protein samples from exponentially growing cells were separated on a SDS–polyacrylamide (37.5:1) gel electrophoresis and blotted on 0.45 μm pore PolyVinyliDenFluoride (PVDF) membranes (Immobilon-P from Sigma Aldrich). Membranes were blocked for 2 hours with 1× Tris-buffered saline (TBS) (50 mM Tris-HCl, 150 mM NaCl [pH 8]) that contain 0.1% Tween-20% and 8% powdered milk, followed by overnight incubation with the primary antibodies diluted in the same milk solution. The polyclonal antisera to AcrA (1:15,000 dilution), TipR (1:5,000 dilution), and CCNA_00164 (1:20,000 dilution) were used. The detection of primary antibodies was done using HRP-conjugated donkey anti-rabbit antibody (Jackson ImmunoResearch) with Western Blotting Detection System (Immobilon from Milipore), and an imaging was performed in Bio-Rad illuminator (Chemidoc MP, Bio-Rad).

Protein purification

TipR protein was expressed from pET21 in E. coli BL21(DE3)/pLysS. Cells were cultivated in LB at 37°C and induced by 1 mM of IPTG for 4 hours, up to an OD_600nm of 0.5. The bacteria were harvested at 8,000 RPM for 30 minutes at 4°C. The pellet was resuspended in 25 mL of buffer (Tris-HCL 40 mM (pH 7.6), 50 mM KCl) before to be sonicated in a water–ice bath (15 cycles of 30 seconds ON, 30 seconds OFF). After centrifugation at 5,000g for 20 minutes at 4°C and filtration through 0.22 μm filters, the proteins contained in the supernatant were purified under native conditions successively using successively Qsepharose and Heparin columns (HiTrap systems from Cytia) according to manufacturer manual. Protein concentration was determined using the Bradford quantification method. Proteins were stored at −80°C in 40 mM Tris-HCl (pH 7.6), 50 mM KCl, and 10% glycerol.

Two-step resistance selection and Mut-seq

To isolate the Δbla;tipR::Tn derivatives that grow on CEF⁴⁰ (Δbla;tipR::Tn acrAB-nodT*), Δbla;tipR::Tn cells were plated on CEF⁴⁰, resistant mutant clones were pooled, and ФCr30-lysates were prepared from this pool. The tipR::Tn allele was then retransduced into Δbla cells, and transductants were selected on plates with KAN²⁰ and then tested for growth on plates with CEF⁴⁰. The acrAB-nodT locus was sequenced in several such CEF⁴⁰ -resistant clones.

To isolate CEF¹⁰-resistant clones for Mut-Seq, the Δbla strain was spread on PYE cephalothin 10 μg/mL and incubated at 30°C for 48 hours. Around 10,000 colonies were harvested and pooled prior to DNA extraction using Ready-Lyse Lysozyme (Epicentre Lucigen) and DNAzol (Thermo Fisher). PCR were performed using the Q5 High-Fidelity DNA Polymerase according to manufacturer instructions with a maximum number of 15 cycles. The amplicons were purified with GeneJET Gel Extraction Kit (Thermo Fisher) and sent to sequencing at the iGE3 genomic platform in the CMU of the University of Geneva. The mixed amplicons were sent to the Genomic platform iGE3 at the University of Geneva. Library preparation and sequencing were done using a HiSeq 2500 with 50-bp paired-end reads. Data analysis was done using Burrows–Wheeler Alignment Tool version 0.7.5a and Samtools version 1.2 prior to be blast against the C. crescentus NA1000 reference genome (NC_011916.1). SNPs with an abundance lower than 5% were considered background (PCR amplification errors) and excluded from the analysis. The analysis is presented in S4 Data.

RNA extraction, deep- sequencing, and bioinformatics analysis

Overnight cultures in PYE of C. crescentus NA1000 (WT) were freshly restarted in 10 mL of PYE (starting OD_660nm approximately 0.05) and incubated at 30°C under agitation to reach an OD_600nm of 0.5. NAL-treated cultures were supplemented with 20 μg/mL of antibiotic 30 minutes before being harvested. First, to provide immediate stabilization of RNA, each cell cultures (4 ml) are treated with 2 volumes of RNA Protect Bacteria reagent (Qiagen, Switzerland) according to the manufacturer’s protocol. Then, cells were lysed in a Ready-Lyse lysozyme solution (Epicentre Technologies) according to manufacturer’s instructions, and lysates were homogenized through QiaShredder columns (Qiagen, Switzerland). The RNeasy Mini Kit (Qiagen, Switzerland) was used for total RNA extraction according to the manufacturer’s protocol, which included a first DNase treatment using the On-column DNase I digestion kit (Qiagen, Switzerland). Extracted total RNA was subjected to a second DNase treatment using Promega RQ1 at 1 unit/μg of RNA. Another total RNA cleanup was performed after the DNase treatment. The RNA concentration was measured using a Nanodrop 1000 spectrophotometer (Thermo Scientific, USA), and RNA quality was assessed using a 2100 Bioanalyzer Instrument (Agilent Technologies, USA). Two independent biological replicates were analyzed per condition.

RNA-Seq library preparation and sequencing were performed at Fasteris SA (Geneva, Switzerland). Bacterial rRNAs were removed from each total RNA samples, and RNA-Seq libraries were prepared using the Ovation Complete Prokaryotic RNA-Seq library system kit according to the manufacturer’s instructions. Single-end runs were performed on an Illumina NextSeq 500 instrument (50 cycles), yielding several million reads (stored as fastq files). Using the web-based analysis platform Galaxy (https://usegalaxy.org), the single-end sequenced reads quality was checked (FastQC, Galaxy Version 0.72), and reads were mapped (Bowtie2, Galaxy Version 2.3.4.3) to the C. crescentus NA1000 genome (NC_011916.1). The counts with the number of reads mapping to each gene feature were prepared using the htseq-count software, Galaxy Version 0.9.1. For each experimental series, the counts normalization and the statistical differential expression analysis was performed using the DESeq2 software, Galaxy Version 2.11.40.6. Sequence data (S1, S3, and S4 Data) have been deposited to the Gene Expression Omnibus (GEO) database (GSE225489 accession).

Chromatin immunoprecipitation coupled to deep sequencing (ChIP-Seq) and data analysis

Overnight cultures in PYE of C. crescentus NA1000(WT) or Δbla; P_xylX::P_xyldjlA-HA were freshly restarted in 80 mL of PYE (starting OD_600nm of 0.05) and incubated at 30°C under agitation with 0.3% xylose when necessary. Cultures were treated with 20 μg/mL of NAL 30 minutes before fixation. Cultures of exponentially growing cells (OD_600nm of 0.5) were supplemented with 10 μM sodium phosphate buffer (pH 7.6) and then treated with formaldehyde (1% final concentration) at room temperature for 10 minutes to achieve crosslinking. Subsequently, the cultures were incubated for an additional 30 minutes on ice and washed 3 times in phosphate-buffered saline (PBS, pH 7.4). The resulting cell pellets were stored at −80°C. After resuspension of the cells in TES buffer (10 mM Tris-HCl (pH 7.5), 1 mM EDTA, 100 mM NaCl) containing 10 mM of DTT, the cell resuspensions were incubated in the presence of Ready-Lyse lysozyme solution (Epicentre, Madison, WI) for 10 minutes at 37°C, according to the manufacturer’s instructions. Lysates were sonicated (Bioruptor Pico) at 4°C using 15 bursts of 30 seconds to shear DNA fragments to an average length of 0.2 to 0.5 kbp and cleared by centrifugation at 14,000 rpm for 2 minutes at 4°C. The volume of the lysates was then adjusted (relative to the protein concentration) to 1 mL using ChIP buffer (0.01% SDS, 1.1% Triton X-84 100, 1.2 mM EDTA, 16.7 mM Tris-HCl [pH 8.1], 167 mM NaCl) containing protease inhibitors (Roche) and precleared with 80 μl of Protein-A agarose (Roche, www.roche.com) and 100 μg BSA. Approximately 5% of each precleared lysate were reserved as total input samples (negative control samples). The precleared lysates were then incubated overnight at 4°C with purified polyclonal rabbit anti-TipR antibodies (1:400 dilution) [9] or monoclonal rabbit anti-HA antibodies (1:250 dilution) (Clone 114-2C-7, Merck Millipore). The immunocomplexes were captured after incubation with Protein-A agarose beads (presaturated with BSA) during a 4-hour incubation at 4°C and then washed subsequently with low salt washing buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8.1), 150 mM NaCl), with high salt washing buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8.1), 500 mM NaCl), with LiCl washing buffer (0.25 M LiCl, 1% NP-40, 1% deoxycholate, 1 mM EDTA, 10 mM Tris-HCl (pH 8.1)) and, finally, twice with TE buffer (10 mM Tris-HCl (pH 8.1), 1 mM EDTA). The immunocomplexes were eluted from the Protein-A agarose beads with 2 times 250 μL elution buffer (SDS 1%, 0.1 M NaHCO3, freshly prepared) and then, just like total input samples, incubated overnight with 300 mM NaCl at 65°C to reverse the crosslinks. The samples were then treated with 2 μg of Proteinase K for 2 hours at 45°C in 40 mM EDTA and 40 mM Tris-HCl (pH 6.5). DNA was extracted using phenol:chloroform:isoamyl alcohol (25:24:1), ethanol-precipitated using 20 μg of glycogen as a carrier and resuspended in 50 μL of DNAse/RNAse free water.

Immunoprecipitated chromatins were used to prepare sample libraries used for deep sequencing at Fasteris SA (Geneva, Switzerland). ChIP-Seq libraries were prepared using the DNA Sample Prep Kit (Illumina) following manufacturer instructions. Single-end run was performed on an Illumina Next-Generation DNA sequencing instruments (NextSeq High); 50 cycles were performed and yielded several million reads per sequenced samples. The single-end sequence reads stored in FastQ files were mapped against the genome of C. crescentus NA1000 (NC_011916.1) using Bowtie2 version 2.4.2.+galaxy0 available on the web-based analysis platform Galaxy (https://usegalaxy.org) to generate the standard genomic position format files (BAM). ChIP-Seq reads sequencing and alignment statistics are detailed in S3 Data. Then, BAM files were imported into SeqMonk version 1.47.2 (http://www.bioinformatics.babraham.ac.uk/projects/seqmonk/) to build ChIP-Seq-normalized sequence read profiles. Briefly, the genome was subdivided into 50 bp, and for every probe, we calculated the number of reads per probe as a function of the total number of reads (per million, using the Read Count Quantitation option). Using the web-based analysis platform Galaxy (https://usegalaxy.org), TipR ChIP-Seq peaks were called using MACS2 Version 2.1.1.20160309.6 (no broad regions option) relative to the total input DNA samples. The q-value (false discovery rate (FDR)) cutoff for called peaks was 0.05. Peaks were rank ordered according to their fold-enrichment values (S3 Data). Peaks with a fold-enrichment values >4 for TipR were retained for further analysis. Consensus sequences common to the 4 enriched TipR-associated loci were identified by scanning peak sequences (+ or − 75 bp relative to the coordinates of the peak summit) for conserved motifs using MEME (http://meme-suite.org/) [32]. Sequence data (S3 Data) have been deposited to the Gene Expression Omnibus (GEO) database (GSE225489 series).

Purification of anti-TipR antibodies

We eluted high-affinity polyclonal antibodies to TipR from an immunoblot. To this end, 200 μg of purified TipR protein was blotted onto a PVDF membrane, and the blot was incubated overnight in 5 ml of anti-TipR serum at 4°C. The membrane was then washed 4 times with 5% fat-free milk and once with PBS. A band corresponding to the size of TipR was then cut out from the membrane and placed in a 15-ml falcon tube. The band was then incubated with 1 mL of buffer 1 [50 mM Glycine (adjusted to pH 2.8 with HCl), 500 mM NaCl] for 3 minutes prior to neutralization with 8 μL of Tris-HCl (pH 8.2). The eluate was harvested, and the membrane was incubated again in 1 mL of buffer 2 [50 mM Glycine (adjusted to pH 2.2 with HCl), 500 mM NaCl] for 3 minutes prior to neutralization with 16 μL of Tris-HCl (pH 8.2). The second eluate was combined with the first and used as purified antibodies at a 1:500 dilution for immunoblotting and 1:400 dilution for ChIP-Seq experiments.

Plunge freezing of Caulobacter crescentus

C. crescentus cells were mixed with 10 nm Protein A conjugated colloidal gold particles (1:10 v/v, Cytodiagnostics), and 4 μl of the mixture was applied to a glow-discharged copper EM grid (R2/1 or R2/2, Quantifoil). The grid was automatically backside blotted for 4 to 6 seconds in a Mark IV Vitrobot (Thermo Fisher Scientific) by using a Teflon sheet on the front pad and plunge frozen in a liquid ethane–propane mixture (37%/63%) cooled by a liquid nitrogen bath. Frozen grids were stored in liquid nitrogen until loaded into the microscope.

Cryo-electron tomography

C. crescentus cells were imaged by cryoET [33]. Images were recorded on Titan Krios 300 kV microscopes (Thermo Fisher Scientific) equipped with a Quantum LS imaging filter operated at a 20-eV slit width and K2 or K3 Summit direct electron detectors (Gatan). Tilt series were collected using a bidirectional tilt-scheme from −60 to +60° in 2° increments. Total dose was 130 to 150 e⁻/Å, and defocus was kept at −8 μm. Tilt series were acquired using SerialEM [34], drift corrected using alignframes, reconstructed, and segmented manually using the IMOD program suite [35]. To enhance contrast, tomograms were deconvolved with a Wiener-like filter [36].

In vivo protein aggregation assay

Exactly, not around 40 mL of exponentially growing cells (OD₆₀₀ between 0.4 and 0.6) were rapidly cooled in ice bath and pelleted by centrifugation (6,000g, 10 minutes). All steps were performed at 4°C, and buffers were all supplemented with protease inhibitor cocktail from Roche (cOmplete tablets, EASYpack). The pellets were washed once in buffer A (50 mM Tris/HCl (pH 8.0), 150 mM NaCl) and resuspended in 300 μL of buffer A supplemented with 100 U/mL Ready-Lyse Lysozyme (Epicentre Lucigen), 10 μg/mL of DNAse I (Roche), and 10 μg/mL of RNAse A (Invitrogen). Cells were lysed in a Bioruptor (Diagenode) (set to high, 15 cycles for 30 seconds at 4°C). Lysates were centrifuged (5,000g, 10 minutes) 2 times to remove nonlysed cells. The protein concentration of the lysate was quantified by Bradford assay, and an aliquot was harvested as the Total fraction. The remaining samples were centrifuged (14,000g, 30 minutes) to pellet the insoluble protein fraction. The supernatant was kept as the Soluble fraction. The pellets were then resuspended in 300 μL buffer A supplemented with 1% (v/v) Triton X-100 with and incubated for 1 hour on ice with regular vortexing prior to the sonication in a Bioruptor (set to high, 1 cycle for 30 seconds) followed by a centrifugation (20,000g, 20 minutes) for washing. This procedure was repeated 3 times. The protein pellet was resuspended in 100 μL 1xSDS loading buffer prior to heating to 95°C for 10 minutes (S5 Data).

Immunoprecipitation

A volume of 50 mL of exponentially growing cells were harvested by centrifugation (20 minutes, 4,500 rpm at 4°C). The pellets were washed in 50 mL 1xPBS and centrifuged for 15 minutes with 6,000 rpm at 4°C and then washed once in 1 mL 1xPBS before centrifugation (for 5 minutes, 14,000 rpm, at 4°C). The pellets were resuspended in 1 mL TES (10 mM Tris-HCl (pH 7.5); 1 mM EDTA; 100 mM NaCl) containing protease inhibitor (cOmplete tablets, EASYpack from Roche) at 2 tabs per 50 mL and 100 U of Ready-lyse/ml of culture (Ready-Lyse Lysozyme from Epicentre Lucigen) prior to 10 minutes of incubation at room temperature. Each sample were supplemented with 50 μL NP-40 10% (AppliChem); 1 μL EDTA 0.5 M (pH 8.0) (Sigma); 10 μL MgCl₂ 1 M; 10 μL DNAse I (Roche) 10 mg/mL; 10 μL RNAse A (Invitrogen) 10 mg/mL and incubated 20 minutes at room temperature with constant agitation. The nonlysed cells were removed by centrifugation (10 minutes, 8,000 rpm at room temperature). The supernatant was harvested and centrifuged for 20 minutes with 14,000 rpm at 4°C. The remaining supernatant was kept as IP input.

For each sample, 25 μL of GFP-trap (GFP-Trap_A from Chromotek) or anti-HA affinity matrix beads (Roche) was rinsed 4 times with 1 mL TES buffer and centrifugation 2 minutes at 3,000 rpm. Then, 25 μL of beads were added to each IP input and left overnight at 4°C with agitation. The beads were harvested by centrifugation (2 minutes at 3,000 rpm) and washed 4 times with wash buffer (10 mM Tris-HCl 7.5; 150 mM NaCl; 0.5 mM EDTA; 0.5% n-dodecyl-β-D-maltoside (DDM)). The beads were then resuspended in 40 μL of 2× SDS sample buffer [125 mM Tris-HCl (pH 6.8); 4% SDS; 20% glycerol; 10% β-mercaptoethanol; 0.004% bromophenol blue] and incubated 10 minutes at 95°C. Each sample was centrifuged for 2 minutes at 3,000 rpm, and the supernatant was collected (without pipetting the beads). Samples of 5 or 10 μL of each collected fraction was analyzed by SDS-PAGE and Coomassie or immunoblotting.