Cross regulation in a three-component cell envelope stress signaling system of Brucella

A multi-layered structure known as the cell envelope separates the controlled interior of bacterial cells from a fluctuating physical and chemical environment. The transcription of genes that determine cell envelope structure and function is commonly regulated by two-component signaling systems (TCS), comprising a sensor histidine kinase and a cognate response regulator. To identify TCS genes that contribute to cell envelope function in the intracellular mammalian pathogen, Brucella ovis, we subjected a collection of non-essential TCS deletion mutants to compounds that disrupt cell membranes and the peptidoglycan cell wall. Our screen led to the discovery of three TCS proteins that coordinately function to confer resistance to cell envelope stressors and to support B. ovis replication in the intracellular niche. This tripartite regulatory system includes the known cell envelope regulator, CenR, and a previously uncharacterized TCS, EssR-EssS, which is widely conserved in Alphaproteobacteria. The CenR and EssR response regulators bind a shared set of sites on the B. ovis chromosomes to control transcription of an overlapping set of genes with cell envelope functions. CenR directly interacts with EssR and functions to stimulate phosphoryl transfer from the EssS kinase to EssR, while CenR and EssR control the cellular levels of each other via a post-transcriptional mechanism. Our data provide evidence for a new mode of TCS cross-regulation in which a non-cognate response regulator affects both the activity and protein levels of a cognate TCS protein pair.

A common function of TCS pathways is to regulate the biochemical composition and transport capacity of the cell envelope (28,29), which serves as the barrier between the interior of the cell and the surrounding environment (30).As Gram-negative bacteria, the envelope of Brucella spp.contains two structurally distinct lipid bilayers with a peptidoglycan cell wall in the thin periplasmic space between.Only two of the six classical Brucella species, B. canis and B. ovis, are naturally "rough", meaning they do not synthesize repeating Opolysaccharide units that extend from the core LPS oligosaccharide of the outer membrane (31).These rough species are not as commonly studied as "smooth" Brucella species, which synthesize O-polysaccharide and are the major zoonotic pathogens.B. ovis therefore provides an interesting comparative model to investigate genes that regulate molecular features of the Brucella cell envelope.We sought to define the contribution of nonessential B. ovis TCS genes to cell survival under cell envelope stress conditions, with an additional goal of assessing functional relationships between TCS genes.To this end, we generated a collection of mutant strains harboring in-frame deletions of 32 non-essential TCS genes and measured growth and viability phenotypes of these strains on solid media containing sodium dodecyl sulfate (SDS) or carbenicillin, to target the cell membrane and cell wall, respectively.Five B. ovis TCS mutant strains were either more sensitive or more resistant than wild type (WT) to these specific cell envelope disruptors.Among these mutants, strains lacking the RR encoded by locus BOV_1929 (BOV_RS09460; cenR) or the RR and HK encoded by locus BOV_1472-73 (BOV_RS07250-55; essR-essS) had similar phenotypes under the tested conditions.Through a series of genetic, genomic, and biochemical experiments, we uncovered evidence for direct cross-regulation between these three TCS proteins, which control transcription of a common gene set to support B. ovis replication under envelope stress conditions and in the host intracellular niche.Specifically, the conserved alphaproteobacterial cell envelope regulator, CenR (32)(33)(34), stimulates phosphoryl transfer between the cognate RR-HK pair, EssR-EssS, while CenR and EssR regulate the steady-state levels of each other in B. ovis.Our results expand understanding of molecular mechanisms of two-component signal transduction and define an unusual mode of TCS crossregulation in Brucella that controls gene expression to support envelope stress resistance and intracellular replication.

Tn-seq identifies a set of candidate essential TCS genes in Brucella ovis
There are 47 TCS genes in the B. ovis genome (Genbank accessions NC_009504 and NC_009505); four of these genes contain frameshift mutations and are defined as pseudogenes (Table S2).Several histidine kinases and response regulators are reported to be essential in Brucella spp.based on previously published Tn-seq data in B. abortus (35) and we sought to comprehensively define the essential set of TCS genes in B. ovis using himar transposon sequencing.Analysis of Tn-himar insertion sites in a collection of approximately 200,000 B. ovis mutants revealed » 54,000 unique TA dinucleotide insertions in the B. ovis genome (36).We used Hidden Markov Model and Bayesian-based approaches (37) to identify candidate essential genes based on this Tn-seq dataset (Table S1).As expected, many of the known TCS regulators of Brucella cell cycle and cell development (38) are defined as essential using this approach including ctrA, cckA, chpT, pdhS, divL, and divK.Though insertion counts were lower in the developmental/cell cycle regulator cpdR relative to local background, this gene did not reach the essential threshold by our analysis.The bvrS-bvrR two-component system, which is homologous to the chvG-chvI system of other Alphaproteobacteria (39), is also essential.This result is consistent with a report by Martín-Martín and colleagues that the bvrSR genes cannot not be deleted in B. ovis (40).Notably, the bvrSR system is not essential in the smooth strains, B. melitensis (39) and B. abortus (41).Finally, the ntrX regulator is also essential.Considering a recent report of strong genetic interactions between ntrX, chvG, and chvI in Caulobacter (42), the essential phenotypes of bvrS-bvrR and ntrX may be related.As this study is focused on cell envelope regulation in a rough Brucella species, we note that several genes with cell envelope functions were identified as essential in B. ovis that are not essential in the smooth species, B. abortus (35), including a putative L,D-transpeptidase (BOV_0757) and a putative D-alanyl D-alanine carboxypeptidase (BOV_1129).Additionally, a DegQfamily serine endoprotease (BOV_0610), hydroxymethylpyrimidine phosphate kinase ThiD (BOV_0209), the putative magnesium transporter MgtE (BOV_A0822), the iron sulfur cluster insertion protein ErpA (BOV_0886), and the RNA chaperone Hfq (BOV_1070) are essential in B. ovis but not in B. abortus (35).
All five mutants with envelope stress defects had reduced viability on TSAB containing 2 mg/ml carbenicillin relative to WT (Figure 1 and Figure S1), though the carbenicillin sensitivity of ∆BOV_1602 was less severe than other strains (Figure 1A).The ∆BOV_1929 single mutant and ∆BOV_1472-1473 double mutant had growth defects on plain TSAB, as evidenced by smaller colony size (Figure 1B and Figure S1B).Growth of WT B. ovis was impaired on TSAB containing 0.0045% SDS, but ∆BOV_1929 and ∆BOV_1472-1473 were even more severely impacted under this condition (Figure 1B).Conversely, deletion of BOV_1602, a cytoplasmic HWEfamily histidine kinase (43) containing an amino-terminal PAS domain (44), strongly enhanced B. ovis viability on SDS plates: growth was apparent at titers approximately three log10 units more diluted than WT (Figure 1A).
BOV_1602 is adjacent to the nepR-ecfG-phyR general stress response (GSR) locus (8) in B. ovis.It was reported that orthologs of BOV_1602 do not regulate Brucella spp.GSR in vitro or in vivo (7,45), but control of GSR transcription is complex (2,46) and may involve BOV_1602 under select conditions.In all cases, the observed agar plate phenotypes were genetically complemented by integration of a single copy of the deleted gene(s) with their native promoter at the glmS locus (Figure 1 and Figure S1A).

A genetic connection between BOV_1472-1473 (essRS) and BOV_1929 (cenR)
There was notable congruence in the agar plate growth defects of the ∆BOV_1929 and ∆BOV_1472-1473 strains across the tested conditions, though ∆BOV_1929 was more sensitive to carbenicillin than ∆BOV_1472-1473 (Figure 1B).BOV_1929 has high sequence identity to CenR, which is a known regulator of cell envelope structure in the Alphaproteobacteria (33,34): BOV_1929 and Caulobacter crescentus, Rhodobacter sphaeroides, and Sinorhizobium meliloti CenR are reciprocal top-hit BLAST pairs (ranging from 65-80% identity over the full length of the protein).We have thus named BOV_1929, cenR.Consistent with our data, Brucella melitensis cenR (also known as otpR) has a reported role in beta-lactam tolerance (47) and acid stress response (48).Studies in C. crescentus and R. sphaeroides have identified the histidine kinase CenK as the cognate regulator of CenR (33,34), but the B. ovis genome does not encode a protein with high sequence identity to Caulobacter or Rhodobacter CenK.This is consistent with a previous report that B. melitensis does not encode an evident CenK ortholog (49).The B. ovis sensor HK most closely related to C. crescentus and R. sphaeroides CenK is encoded by gene locus BOV_0289 (33-37% amino acid identity).BOV_0289 encodes the most likely HK partner for CenR based on a Bayesian algorithm to predict TCS HK-RR pairs (50).However, the ∆BOV_0289 deletion mutant grew the same as WT in the tested stress conditions and a ∆BOV_0289 ∆cenR double mutant phenocopied the ∆cenR single mutant (Figure 1B).These data indicate that BOV_0289 is not an activator of CenR in these conditions.Given the envelope stress survival phenotypes of the ∆BOV_1472-1473 double deletion mutant, we hereafter refer to the DNA-binding response regulator gene BOV_1472 as essR and the sensor histidine kinase gene BOV_1473 as essS.To our knowledge, EssS and EssR have not been functionally characterized, though EssS is 68% identical to the RL3453 sensor kinase that has been functionally linked to Rhizobium leguminosarum plant root attachment (51).EssR is an OmpR-family response regulator, while the EssS sensor kinase has two predicted transmembrane helices and primary structure features that resemble the well-studied CpxA and EnvZ cell envelope regulators (52)(53)(54).To assess phenotypic relationships between the ∆essR-essS and ∆cenR mutants, we generated B. ovis strains harboring in-frame deletions of either essR or essS and a ∆cenR ∆essR-essS triple mutant and evaluated growth of these strains on TSAB containing SDS or carbenicillin as above.The ∆essR strain phenocopied the ∆essR-essS double mutant in all conditions.The ∆essS mutant was indistinguishable from WT in the presence of 0.0045% SDS, however, at a lower SDS concentration (0.004%) ∆essS showed an intermediate sensitivity phenotype (Figure S1B).∆essS also showed an intermediate carbenicillin sensitivity phenotype.The ∆cenR ∆essR-essS triple mutant phenocopied the ∆cenR and ∆essR single mutants (Figure 1B & S1B).We further evaluated ∆cenR, ∆essR, and ∆essS mutants in three additional stress conditions that perturb the cell envelope in different ways: ethylenediaminetetraacetic acid (EDTA) is a divalent cation chelator that can destabilize the outer membrane (55); polymyxin B is a cationic antimicrobial peptide that targets the outer membrane; and NaCl can function as an osmotic stressor that disrupts envelope integrity.∆cenR and ∆essR again phenocopied each other in all three treatment conditions.Both response regulator mutants were more sensitive to EDTA and polymyxin B, and both were more resistant to NaCl compared to wild type.These results provide further support that CenR and EssR function in the same pathway.Like the RR mutants, ∆essS was sensitive to EDTA and polymyxin B. However, ∆essS was not NaCl resistant (Figure S2).From these results, we hypothesize that the EssS sensor kinase has distinct regulatory roles in different stress conditions.The sequence relatedness between EssR and CenR is moderate when compared to pairings of all B. ovis response regulators with DNA binding domains, with 32% identity and 50% similarity.Specific regions of identity and similarity in their primary structures are presented in Figure S1C.

Contribution of the CenR and EssR aspartyl phosphorylation sites to stress survival
CenR and EssR are both DNA-binding response regulator proteins, which are typically regulated by phosphorylation of a conserved aspartic acid residue in the receiver domain.To assess the functional role of the CenR aspartyl phosphorylation site (D55), we tested whether expression of non-phosphorylatable (cenRD55A) or putative phosphomimetic (cenRD55E) alleles of cenR could genetically complement the defects of the ∆cenR strain in the presence of SDS or carbenicillin.Both mutant alleles of cenR restored ∆cenR growth/survival to WT levels, indicating that the CenR phosphorylation site does not impact CenR function under these conditions (Figure 2).We conducted the same experiment for EssR, testing whether essRD64E and essRD64A could genetically complement the ∆essR growth defects.essRD64E expression restored wild-type like growth to ∆essR on plain TSAB and TSAB-SDS plates but resulted in a strain that was even more sensitive than ∆essR to carbenicillin.Expression of essRD64A failed to complement ∆essR under all conditions (Figure 2).We conclude that EssR phosphorylation at D64 contributes to in vitro stress survival of B. ovis.

cenR and essR mutants have equivalent cell size defects
Previous studies of CenR in C. crescentus and R. sphaeroides have shown that depletion or overexpression of cenR leads to large defects in cell envelope structure and/or cell division (33,34).Considering these results and the sensitivity of B. ovis ∆cenR to SDS and carbenicillin, we inspected ∆cenR, ∆essR, and ∆essS cells by phase contrast light microscopy and cyro-electron microscopy for defects in cell envelope structure or cell morphology.Deletion of cenR, essR or essS did not result in apparent changes in cell morphology or cell division as assessed by phase contrast microscopy at 630x magnification (Figure S3A).However, an analysis of cell size revealed that both ∆cenR and ∆essR mutant cells were larger than WT; the average area of the mutant cells was approximately 12% greater than WT in 2D micrographs (p<0.0001)(Figure S3B).Again, the parallel phenotype of ∆cenR and ∆essR supports a model in which these two genes execute related functions.An intact phosphorylation site was not required for CenR or EssR to affect cell size, as expression of either phosphorylation site mutant allele of cenR(D55) or essR(D64) restored cell size to WT levels (Figure S3B).We did not observe major cell membrane defects in ∆cenR and ∆essR mutant cells by cryo-EM (Figure S3C).We conclude that the SDS and carbenicillin resistance defects we observe in the DcenR and DessR strains are associated with a defect in cell size control.

B. ovis ∆cenR, ∆essR, and ∆essS strains have equivalent fitness defects in a macrophage infection model
As B. ovis is an intracellular pathogen, we sought to examine the importance of cenR, essR and essS in the intracellular niche.The interior of mammalian phagocytes more closely models conditions encountered by the bacterium in its natural host, and many mutants with defects in cell envelope processes are attenuated in infection models (56).∆cenR, ∆essR, and ∆essS strains had no defect after entry (2h post-infection (p.i.)) or at early stages of infection (6h & 24h p.i.) of THP-1 macrophage-like cells relative to WT.By 48h p.i., recoverable CFU of WT B. ovis increased, consistent with adaptation and replication in the intracellular niche; CFU of ∆cenR, ∆essR, and ∆essS did not increase appreciably between 24 and 48h.The 1 log10 unit intracellular replication defect of the ∆cenR, ∆essR, and ∆essS strains at 48h was complemented by expression of the deleted gene(s) from an ectopic locus (Figure 3).Thus, we conclude that cenR, essR, or essS do not affect macrophage entry or early survival but all three genes similarly affect replication and/or survival after 24h.These results indicate that essS, essR, and cenR contribute to B. ovis fitness after the establishment of the replicative niche inside the Brucella-containing vacuole.
Additionally, we tested the infection phenotypes of strains harboring alleles of essR and cenR in which the conserved aspartyl phosphorylation site was mutated.Expression of cenRD55A or cenRD55E partially -and equivalently -complemented the 48-hour infection defect of ∆cenR (Figure 3A).Expression of either essRD64E or essRD64A failed to complement ∆essR in this assay (Figure 3B).We conclude that an intact aspartyl phosphorylation site in both the CenR and EssR receiver domains is required for WT levels of replication in a THP-1 macrophage infection model.

cenR, essS and essR mutants are not sensitive to low pH
The Brucella containing vacuole (BCV) is acidified at early time points after infection (1 to 10 hours) in J774 murine macrophages and HeLa cells (57,58).We observed no defect in recoverable CFU of our mutants at the 2h time point in THP-1 macrophage-like cells (Figure 3), which suggested that acid tolerance was not perturbed in ∆cenR, ∆essR, or ∆essS.Nonetheless, we sought to more rigorously test whether sustained exposure to acid impacted the survival of ∆cenR, ∆essR, and ∆essS strains.The acidified BCV has a pH in the 4.0-4.5 range (59), so we tested whether exposure to acidified Brucella broth (pH 4.2) for 2 hours differentially impacted mutant viability.We did not observe significant differences in viability between WT and mutant strains after in vitro acid exposure (Figure S4) and conclude that sensitivity to acid in the BCV cannot alone explain the intracellular defects of cenR, essS and essR mutants.The slower growth rates of cenR and essRS mutants, which are evident on solid media (Figures 1 & 2 and S1) and in broth (k =0.0028 min - 1 for WT, 0.0021 min -1 for ∆cenR, and 0.0022 min -1 for ∆essRS), could be the primary determinant of their replication defect after 24h.Nonetheless, the more severe defect of the ∆essS strain between 24 and 48 hours relative to the other TCS mutants provides evidence that intracellular defects of these TCS mutants are likely complex and multifactorial.

EssR and CenR regulate a common gene set
Considering that TCS proteins typically function to regulate transcription, we used RNA sequencing (RNAseq) to assess the relationship between EssRS-and CenR-regulated gene sets.The global transcriptional profiles of ∆cenR and ∆essRS mutant strains were highly correlated (Figure 4).Filtering genes based on a minimum fold change ³ |2| (FDR p-value < 10 -4 ) revealed 46 transcription units, containing 53 genes, that were regulated by essR-essS.Fifty-two transcription units, containing 61 genes, were regulated by cenR (Figure 4; Table S3).These gene sets largely overlap: thirty-eight (38) are differentially expressed in the same direction in both datasets.With few exceptions, genes that met the criteria for differential regulation in only one strain showed similar, but more modest, changes compared to WT in the other strain.These results indicate a high degree of functional overlap of EssRS and CenR, with respect to gene expression.One potential explanation for the observed regulatory overlap is that EssR and CenR transcription depend on each other.However, transcript levels of essR-essS in a ∆cenR background, and cenR in ∆essR-essS background do not differ significantly from WT (Table S3).We conclude that neither response regulator significantly affects the transcription of the other.Rather, CenR and EssR either independently or coordinately regulate transcription of an overlapping set of B. ovis genes.
To further investigate the connection between CenR-and EssR-dependent transcription, we performed chromatinimmunoprecipitation (ChIP)-seq with both EssR and CenR.To promote RR binding to chromosomal target sites, we expressed putative phosphomimetic alleles (DàE) of each RR from their native promoters integrated at the glmS locus in each mutant strain; genes were fused to a C-terminal 3xFLAG tag.ChIP of EssRD64E-3xFLAG yielded 65 significant peaks across two biological replicates, and ChIP of CenRD55E-3xFLAG yielded 47 significant peaks across three biological replicates.Thirtythree peaks were shared between CenRD55E and EssRD64E (Figure 4B and Table S4) providing additional support for a model in which CenR and EssR regulate transcription of a shared gene set, functioning either separately or as a heteromeric complex.We did not observe CenR binding to the essRS promoter or EssR binding to the cenR promoter in the ChIP-seq data.This is further indication that CenR and EssR do not regulate each other transcriptionally, consistent with the RNA-seq results.

Genes with cell envelope functions are abundant in the CenR-EssRS regulon
The CenR/EssR regulon prominently features genes encoding membrane transport proteins, several of which are contained in co-regulated clusters (Figure 4).The largest of these clusters encodes carbohydrate metabolism enzymes (BOV_A0354-A0355), transporter subunits (BOV_A0347-A0348; BOV_A0350-A0352), a DapAfamily dihydrodipicolinate synthase/N-acetylneuraminate lyase (BOV_A0349), Gfo/Idh/MocA family oxidoreductases (BOV_A0345-A0346), and a FadR-family transcription regulator (BOV_A0353).Transcription of this entire cluster is significantly reduced following cenR or essRS deletion (Figure 4; Table S3).Expression of B. melitensis orthologs of these genes is highly upregulated in supramammary lymph nodes of goats, specifically at late stages of infection (60) suggesting a role in long-term colonization of animal hosts.Homologs of these genes are also regulated by CenR in R. sphaeroides (33) (Table S3), providing evidence for conservation of CenR-dependent transcription across Alphaproteobacterial genera.
Other genes with reduced expression upon cenR or essRS deletion include a third Gfo/Idh/MocA family oxidoreductase (BOV_0234) and the secreted BA14K protein (BOV_A0688), which is required for normal spleen colonization in a mouse model of B. abortus infection (8).Transcripts of three genes immediately adjacent to the general stress response regulator, PhyR, also decreased upon cenR or essRS deletion including the previously discussed HWE-family sensor kinase, BOV_1602, which impacts B. ovis SDS resistance (Figure 1).Additionally, the predicted sulfate ABC transport operon, cysWTP, decreased significantly in both deletion mutant strains (Figure 4A; Table S3).Nine genes had significantly higher transcription in ∆cenR and ∆essRS (Figure 4A; Table S3) including a site-specific DNA integrase (BOV_0631), an ABC transporter (BOV_A0247), a C-type lysozyme inhibitor (BOV_0447) and LysM-domain protein (BOV_0448), envelope integrity protein B (eipB; BOV_1121), and BOV_1296, which is directly regulated by the virulence regulator VjbR under select conditions (61).BOV_1296 encodes acid shock protein 24 (Asp24), which contributes to virulence in later stages of infection in B. abortus and B. melitensis (62,63).BOV_1399, encoding a periplasmic soluble lytic murein transglycosylase (SLT) enzyme, exhibited the most significant expression difference in our RNA-seq datasets.The promoter of BOV_1339 is directly bound by CenR and EssR (Table S4) and transcript levels were approximately 48 times lower than WT in both ∆cenR or ∆essRS RNA-seq datasets (Figure 4; Table S3).To test whether loss of this predicted cell wall metabolism enzyme contributed to the stress survival phenotypes of ∆cenR and ∆essRS, we deleted BOV_1399 and subjected this strain to the same agar plate growth assays described above.However, the phenotypes of ∆BOV_1399 in these assays were indistinguishable from WT. Notably, the promoters of several genes with known cell envelope functions are bound by CenR, EssR, or both but do not exhibit differential transcription in ∆cenR or ∆essRS.For example, EssR binds to the promoter of BOV_0115 (Omp25d), which contributes to cell envelope integrity (64,65), and B. ovis-host interaction (66).The outer membrane autotransporters, bmaA and bmaC, which are important for protein translocation to the cell surface ( 67) and for host cell adherence (11) are also bound by CenR/EssR but do not change in expression upon cenR or essRS deletion.Additional studies are necessary to determine what CenR/EssR-bound or regulated genes determine B. ovis cell size and support B. ovis fitness under envelope stress and in the intracellular niche.

CenR and EssR physically interact via their REC domains in a heterologous system
Genetic, transcriptomic, and ChIP-seq data all indicate that the response regulators EssR and CenR directly control expression of a common gene set in B. ovis to enable growth in the presence of SDS and carbenicillin.While the molecular processes that determine signaling via a particular TCS pathway are typically insulated from other TCS proteins (20), we postulated that the congruent genetic and molecular phenotypes of ∆cenR and ∆essR strains could arise through direct molecular interactions of CenR with either EssS, EssR, or both.In fact, in a systematic analysis of B. abortus TCS protein interactions, Hallez and colleagues previously reported the surprising observation that EssR (then, generically annotated as TcbR) and CenR interact in a yeast twohybrid assay (68).EssR and CenR were the only two B. abortus RRs that showed interaction in their genomescale screen.To test EssR and CenR protein-protein interaction in a heterologous system, we used a bacterial two-hybrid approach based on the T18 and T25 domains of the split adenylate cyclase enzyme (69).CenR and EssR showed strong interactions when fused to either adenylate cyclase domain, while the homomeric CenR-CenR and EssR-EssR combinations showed no evidence of interaction (Figure 5A).
To test the contribution of the conserved aspartyl phosphorylation sites of CenR and EssR to the observed two-hybrid interaction, we fused putative phosphomimetic (DàE) or non-phosphorylatable (DàA) alleles of either CenR or EssR to the T18 and T25 fragments of adenylate cyclase.Both EssRD64A and EssRD64E interacted with CenR to the same extent as WT EssR (Figure 5B).CenRD55E had significantly reduced interaction with EssR, while CenRD55A interaction with EssR was not significantly different.These results provide evidence for a model in which CenR phosphorylation attenuates its interaction with EssR.Homomeric CenR-CenR or EssR-EssR interactions were again not observed in our two-hybrid assay for either the DàA or DàE mutants (Figure 5B).We used the neural network models of AlphaFold2 (70), as implemented in AF2Complex (71), to develop hypotheses about the structural basis of CenR-EssR interaction.The results of this computation predicted that these two response regulators interact primarily through their receiver (REC) domains rather than their DNA-binding (DBD) domains (Figure 5C).Predicted (1:1) CenR:EssR complex structures showed parallel REC domain heterodimers with significant buried surface area in a region of similar primary structure of CenR and EssR corresponding to the α4-β5-α5 structural face of each protein (Figure S1); this is a well-established REC domain interaction interface (72).To test if CenR and EssR interact via their REC domains, we again used a bacterial twohybrid assay.The measured interaction of the isolated CenR and EssR REC domains was comparable to that of the full-length proteins (Figure 5D).None of the other DBD-DBD or DBD-REC combinations had significant interactions.

CenR and EssR interact in B. ovis
Given that CenR and EssR strongly interact via their REC domains in a heterologous system, we sought to test whether these two proteins interact in B. ovis cells by co-immunoprecipitation.Briefly, we expressed a cenR-3xFLAG fusion from the native cenR promoter on a lowcopy plasmid in strains lacking either endogenous CenR (∆cenR) or lacking both response regulators (∆cenR ∆essR) and applied the crosslinked lysates to anti-FLAG magnetic beads.After multiple washing steps we eluted bound protein, reversed the crosslinks, and resolved the eluate by SDS-PAGE.Western blot using polyclonal EssR antiserum revealed clear EssR bands in the lysates of both ∆cenR / cenR-3xFLAG and ∆cenR vector control strains, but not in ∆cenR ∆essR / cenR-3xFLAG.Among the three eluate fractions, only the strain expressing both cenR-3xFLAG and endogenous essR yielded a strong EssR band on the Western blot (Figure 5E).These results provide evidence that the CenR-EssR interaction demonstrated by bacterial two-hybrid assay occurs in B. ovis cells.

CenR enhances the rate of phosphoryl transfer from EssS to EssR
The related phenotypes of cenR and essR mutants and the fact that these two response regulators physically interact in cells raised questions about the biochemical consequence(s) of EssR/CenR interaction on histidine kinase autophosphorylation and phosphoryl transfer.The data presented to this point do not clearly establish the identity of a sensor kinase in this system, though we considered EssS to be a strong candidate based on genomic proximity of essS and essR, and largely overlapping phenotypes between ∆essS and ∆essR strains.To test this hypothesis, the cytoplasmic kinase domain of EssS (aa 191-488) and full-length EssR and CenR response regulators were purified.The EssS kinase domain autophosphorylated in vitro; incubating EssS kinase domain with EssR resulted in rapid loss of phospho-EssS (EssS~P) signal and a concomitant increase in phospho-EssR (EssR~P) signal within 30s, indicating phosphoryl transfer (Figure 6A).Incubation of EssS with CenR resulted in no detectable phosphoryl transfer by 600s, even when CenR was added in 50 molar excess.We conclude that EssS is the cognate kinase for EssR, and that EssS does not directly phosphorylate CenR.Though EssS does not phosphorylate CenR in vitro, we postulated that CenR may influence activity of the EssS-EssR TCS.To test this idea, we first mixed equimolar EssS and EssR with increasing concentrations of CenR.Supplementing an EssS-EssR reaction mixture with CenR at a 1:1:1 molar ratio resulted in a 20% increase in EssS~P dephosphorylation after 20s relative to a 1:1 EssS-EssR reaction (Figure S5).Adding CenR to the reaction mixture at 5x molar excess (1:1:5), further enhanced EssS~P dephosphorylation (Figure S5).The reduction in EssS~P upon addition of CenR coincides with an increase in RR~P.The effect of CenR addition on EssS~P levels (at 20 s) saturated at a 1:1:5 ratio (Figure S5).Taken together, the data indicate that phosphoryl transfer from EssS to EssR is accelerated by addition of CenR.To further investigate the effect of CenR on the kinetics of phosphoryl transfer between EssS and EssR, we measured both EssS~P and EssR~P signal over a 1-minute time course.EssS to EssR phosphoryl transfer reactions containing 5-molar excess CenR had an enhanced rate of (apparent) EssR~P production and an enhanced rate of EssS~P loss compared to reactions with EssS and EssR only (Figure 6B-D); these experiments do not rule out the possibility that CenR is phosphorylated by EssS when EssR is present as EssR and CenR have similar molecular weights.In the absence of CenR, we observed maximal EssR~P signal after 45-60s.When CenR was present, levels of the band we attribute to EssR~P were maximal by 15s.These results provide evidence that CenR stimulates phosphoryl transfer from EssS to EssR.

EssR and CenR determine protein levels of each other via a post-transcriptional mechanism
Stimulation of phosphoryl transfer from EssS to EssR by the non-cognate CenR protein provides one explanation for the related phenotypes of the ∆cenR, ∆essS and ∆essR mutants.We sought to test whether the CenR protein may have other regulatory effects on the cognate EssS-EssR TCS pair in B. ovis.EssR and CenR do not regulate each other's transcription (Table S3) but do physically interact (Figure 5).We hypothesized that CenR interaction with EssR could impact steady-state EssR protein levels through a post-transcriptional mechanism, and vice versa.To test this hypothesis, we measured EssR protein by Western blot in WT and ∆cenR.Deletion of cenR resulted in a significant reduction (~70%) of EssR protein levels.The ∆essS strain also had significantly reduced EssR levels, though the effect was not as large as ∆cenR (Figure 7).The impact of EssS on EssR protein levels is not apparently a consequence of phosphorylation as steady-state levels of EssR, EssR(D64A) and EssR(D64E) did not differ significantly.The mechanism by which CenR affects EssR protein levels is not known, but reduced EssR levels in a ∆cenR background provide an additional explanation of the phenotypic congruence of the ∆cenR and ∆essR strains.The intermediate level of EssR protein in the ∆essS strain is consistent with the intermediate phenotype of this mutant in plate stress assays.We further tested whether CenR protein levels are impacted by essR by measuring CenR-3xFLAG in ∆cenR / cenR-3xFLAG and ∆cenR ∆essR / cenR-3xFLAG strains.CenR-3xFLAG was »60% lower in a strain lacking essR.We conclude that CenR and EssR regulate each other's protein levels through a post-transcriptional mechanism.

Discussion
Multiple TCS genes are typically present in bacterial genomes, and the proteins they encode most often function separately to regulate distinct transcriptional responses (20).Our systematic analysis of B. ovis TCS genes revealed two DNA-binding response regulators, cenR and essR, that had related morphological, stress resistance, and infection phenotypes when deleted.These results informed the hypothesis that CenR and EssR work together to execute their functions in the cell and led to the discovery of a new cell envelope regulatory system (EssS-EssR) and a new mechanism of TCS regulation in bacteria.

Discovery of a cell envelope regulatory system in
Brucella TCS proteins play a key role in the regulation of cell envelope biogenesis and homeostasis in the bacterial kingdom (29,73,74).The function of the EssR-EssS TCS protein pair had not been defined in any species prior to this study, though this system is conserved in many Alphaproteobacteria.Our data provide evidence that these two proteins play an important role in Brucella resistance to cell envelope disruptors in vitro, and in regulating processes important for intracellular replication in a macrophage infection model.Apparent orthologs of the sensor kinase, EssS, are present in select genera across the orders Hyphomicrobiales, Caulobacterales, Rhodobacterales, Rhodospirillales, and Rickettsiales; EssR has a similar phylogenetic distribution (Figure S6).The functional importance of EssR-EssS in Alphaproteobacteria is evidenced by the fact that it is 1 of only 5 TCS signaling pairs (NtrXY, PhoBR, RegBA, ChvGI/BvgRS, & EssRS) in the highly streamlined SAR11 genome (Pelagibacter ubique).The HK domain of EssS has low sequence identity to other well-studied Gramnegative envelope regulators (e.g.CpxA and EnvZ), but multiple sequence alignment models in the conserved domain database (CDD) (52) suggest that the EssS, CpxA, and EnvZ HKs have common ancestry (e-value < 10 -40 ).Likewise, EssR is most closely related to the OmpR sequence family in the CDD (e-value < 10 -70 ).OmpR functions as the cognate regulator of the EnvZ kinase in enteric bacteria (53).EssS and EssR clearly form a cognate signaling pair in vitro as evidenced by specific phosphoryl transfer from the EssS kinase domain to EssR on a fast time scale (Figure 6).However, the phenotypes of the ∆essS and ∆essR strains are not equivalent in an in vitro model of cell envelope stress.Under most challenges (SDS, carbenicillin, EDTA, and polymyxin B) the defect of ∆essR was more severe than ∆essS (Figure 1, S1 and S2), and these mutants have opposite phenotypes when exposed to high NaCl: the essR mutant is NaCl resistant while the essS mutant is sensitive compared to WT (Figure S2).The mechanism underlying the opposing NaCl phenotypes of these strains merits further investigation.The phenotypes of ∆essS and ∆essR are equivalent in a macrophage infection model (Figure 3).This result provides evidence that EssS-dependent phosphorylation (or dephosphorylation) of EssR is more important for system function in the complex environment of the intracellular niche than it is in a simple in vitro agar plate assay.

An unexpected functional role for the conserved cell envelope regulator, CenR
The result that B. ovis CenR confers resistance to SDS and carbenicillin (Figures 1 & 2) was not unexpected considering the phenotypes of cenR mutants in other Alphaproteobacteria.CenR was first described as an essential RR in C. crescentus, where it functions to regulate cell envelope structure (34), and is now known to be conserved in many alphaproteobacterial orders (33).Recent work has shown that cenR is essential in R. sphaeroides, where it controls transcription of the Tol-Pal outer membrane complex and other cell envelope genes (33), and in Sinorhizobium meliloti where it mediates osmotolerance and oxidative stress resistance (32).In all three of these species, CenR is regulated by a cognate histidine kinase, CenK.Our data show that cenR is not essential for B. ovis growth or division under standard culture conditions and indicate that it may be an orphan response regulator, which is consistent with a previous report in B. melitensis (49).B. ovis cenR (and essR-essS) do not function to mitigate acid stress in vitro.Though a B. melitensis cenR mutant was previously reported to be acid sensitive (48), the treatment protocol and measured pH range differ substantially between our B. ovis study and the B. melitensis study.The possibility that a Brucella HK phosphorylates (or dephosphorylates) CenR under certain conditions cannot be conclusively ruled out from our data.Indeed, there is some evidence that the conserved CenR aspartyl phosphorylation site can impact CenR-EssR interaction (Figure 5) and replication in the intracellular niche (Figure 3).Nonetheless, both cenRD55A and cenRD55E alleles fully complement the agar plate stress phenotypes of DcenR.And unlike R. sphaeroides and C. crescentus, where cenR depletion results in major defects in cell envelope structure, the impact of cenR deletion on B. ovis cell morphology is small: B. ovis DcenR mutants are slightly (but significantly) larger than WT (Figure S3), but the morphology of the mutant cells otherwise appears normal.These results indicate that CenR function in Brucella ovis differs somewhat from Caulobacter, Rhodobacter, and Sinorhizobium.

CenR is a post-transcriptional regulator of the EssS-EssR two-component system
Cross-regulation between TCSs is uncommon, though there is experimental support for direct interactions between otherwise distinct TCS HK-RR protein pairs for limited number of systems (20).For example, the NarXL and NarQP systems of Escherichia coli crossphosphorylate to tune nitrate and nitrite respiratory processes (75), and in C. crescentus, a consortium of sensor kinases that coordinately regulate cell adhesion in response to a range of environmental cues physically interact in cells (23).In this manuscript, we present evidence for a mode of TCS cross-regulation in which a non-cognate RR (CenR) directly stimulates the phosphoryl transfer activity of a cognate HK-RR protein pair (EssS-EssR) (Figure 6 and Figure 8).We further demonstrate that CenR and EssR reciprocally regulate their protein levels in B. ovis cells via a post-transcriptional mechanism (Figure 7).EssR and CenR physically interact via their receiver domains (Figure 5), and it seems most likely that CenR and EssR protect each other from proteolytic degradation in the Brucella cell though we cannot rule out other post-transcriptional models at this time.The positive effect of CenR on EssS-EssR phosphoryl transfer activity and the positive effect of EssR and CenR protein levels on each other are consistent with the congruent phenotypes of strains lacking cenR and essR.However, CenR does not simply control the activity and levels EssS-EssR.CenR is itself a DNA binding protein, and we have presented evidence that EssR and CenR both directly and indirectly regulate transcription of a highly correlated set of genes that includes multiple transporters and cell wall metabolism genes (Figure 4).These two transcriptional regulators directly bind shared and unique sets of sites on B. ovis chromosomes 1 and 2. It is possible that CenR and EssR bind DNA as heterodimers, which has been described for the BldM and WhiI response regulators of Streptomyces (76), and the RcsB regulator of E. coli with GadE (77) and BglJ (78).These heterodimeric regulators are competent to control different classes of promoters depending on oligomeric state; a similar mechanism may exist for EssR and CenR though we have not identified distinct promoter classes in our data.It is plausible that CenR and EssR bind as homodimers and heterodimers considering the pattern of unique and overlapping genes/promoters in the transcriptomic and ChIP-seq datasets.Future studies aimed at deciphering molecular features of environmental signal detection by the EssS sensor kinase, allosteric regulation of TCS activity by CenR, and transcriptional control by the CenR and EssR regulators will generally inform our understanding of the evolution of cell envelope regulatory systems in bacteria.More specifically, investigation of these proteins will illuminate mechanisms by which Brucella replicate in the intracellular niche and spread from cell to cell in face of harsh immune stresses encountered within the host.

Materials and Methods
Bacterial strains and growth conditions All Brucella ovis ATCC25840 and derivative strains were grown on Tryptic soy agar (BD Difco) + 5% sheep blood (Quad Five) (TSAB) or in Brucella broth at 37˚C with 5%

Essential gene calculations using B. ovis Tn-himar sequencing data
We constructed a library of B. ovis transposon mutants as described previously (15).Briefly, the E. coli strain APA752, harboring the pKMW3 mariner transposon library, was conjugated into WT B. ovis and transposon bearing strains were selected on TSAB supplemented with 50 µg/ml kanamycin.Following an outgrowth of pooled mutants in 250 ml Brucella broth to OD600 ≈ 0.6, cells were frozen in 25% glycerol in 1 ml aliquots.An aliquot was thawed for gDNA extraction and subsequent insertion site mapping, as described (79).The library contained insertions at over 50,000 unique sites in the genome.Insertion site sequencing data is available in the NCBI sequence read archive under accession SRR19632676.Using the insertion site mapping data from this Tn-seq dataset, we applied the HMM and Gumbel algorithms in the TRANSIT (37) software package to identify candidate essential genes (see Table S1).

Chromosomal deletion strain construction
The double-crossover recombination method was used to generate all B. ovis strains bearing in-frame, unmarked gene deletions.Approximately 500 base pair (bp) upstream or downstream of the target gene, including 9-120 bp of the 3' and 5' ends of the target gene, were amplified by PCR using KOD Xtreme polymerase (Novagen) using primers (Table S5) specific to these regions and the B. ovis genomic DNA as template.The DNA fragments were then inserted into the sacBcontaining suicide plasmid pNPTS138 either through Gibson assembly or restriction enzyme digestion and ligation.The ligated plasmids were first chemically transformed into competent E. coli Top10.After sequencing confirmation, plasmids were transformed into chemically competent E. coli WM3064 (strain originally produced by W. Metcalf), a DAP auxotroph conjugationcompetent donor strain.Plasmids were transferred to B. ovis through conjugation.Primary recombinants were selected on TSAB supplemented with kanamycin.After outgrowth in non-selective growth in Brucella broth for 8 hours, clones in which a second recombination removed the plasmid were identified through counter selection with 5% sucrose.Colony PCR was performed on kanamycinsensitive colonies to distinguish clones bearing the deletion allele from those bearing the WT allele.

Complementation strain construction
To engineer genetic complementation constructs, target genes were amplified by KOD polymerase, including ~300 bp upstream and ~50 bp downstream of each target gene.The PCR products were purified and inserted into plasmid pUC18-mTn7 by restriction enzyme digestion and ligation, followed by chemical transformation into E. coli TOP10 cells.After sequence confirmation, the mTn7 plasmids were transformed into chemically competent E. coli WM3064.These plasmids were co-conjugated into B. ovis strains with a Tn7 integrase expressing suicide helper plasmid, pTNS3, which is also carried by WM3064.B. ovis colonies carrying the integrated mTn7 constructs at the glmS locus were selected on TSAB containing kanamycin.
Polymyxin B stress assay After 2 days of growth on TSAB supplemented with kanamycin, B. ovis cells were collected and resuspended in 1 mL of Brucella broth at an OD600 =0.3.Cells were split and one portion was treated with 1 mg/mL polymyxin B and the other was untreated.Both treated and untreated groups were incubated at 37˚C with 5% CO2 supplementation for 80 minutes.Each culture was then 10-fold serially diluted in PBS and 5 µl of each dilution was spotted onto TSAB.After 3 days of incubation for untreated group and 4 days for treated group, growth was documented photographically.

Macrophage infection assay THP-1 cells were grown in Roswell Park Memorial
Institute Medium (RPMI) + 10% heat inactivated fetal bovine serum (FBS) at 37˚C with 5% CO2 supplementation.Three days prior to infection, THP-1 cells were seeded in a 96-well plate at a concentration of 10 6 cells/ml in 100 µl/well of fresh RPMI + 10% FBS with an addition of 50ng/ml phorbol 12-myristate-13-acetate (PMA) to induce differentiation.After three days at 37˚C in 5% CO2, B. ovis cells were resuspended at a concentration of 10 8 CFU/ml (OD600 = 0.15) in RPMI + 10% FBS and added to THP-1 cells at a multiplicity of infection (MOI) of 100.The 96-well plate containing B. ovis and THP-1 cells was centrifuged at 150 x g at room temperature for 5 minutes and incubated at 37˚C with 5% CO2 for 1h.Media was removed and fresh media containing 50 µg/ml gentamicin was added to kill extracellular B. ovis that were not internalized.The plate was incubated for another hour at 37˚C with 5% CO2.
Media was removed and fresh media containing 25 µg/ml gentamicin was added to each well except for 2h-p.i.wells.A 2h, 6h, 24h, and 48h post-infection, B. ovis were enumerated by removing the media and washing the cells with PBS once and incubating at 37˚C for 10 minutes.Then PBS was removed and 200 µl dH2O was added to each well and the plate was incubated at room temperature for 10 minutes to lyse the THP-1 cells.B. ovis cells were collected and serial diluted with a 10 -1 dilution factor in PBS and plated on TSAB.Plates were incubated for 3 days and CFU were enumerated.
Acid stress assay As previously described (7), Brucella cells were grown on TSAB for 2 days before being inoculated into 5 mL Brucella broth and shaken for 24h.OD600 was adjusted to 1.5 and 50 µl of cells were added to either 5 mL plain Brucella broth or 5 mL Brucella broth pH 4.2 (final OD600 = 0.015).After 2h of shaking incubation, cells were serially diluted and plated onto plain TSAB.After 3 days of incubation, CFU/ml were enumerated.
Phase contrast microscopy Samples of B. ovis cells were grown on TSAB supplemented with kanamycin and incubated for 2 days.Cells were resuspended in milliQ H2O and 1 µl of cells were spotted on an agarose pad on a cover slide and imaged on a Leica DMI 6000 microscope in phase contrast with an HC PL APO 63x/1.4numeric aperture (NA) oil Ph3 CS2 objective.Images were captured with an Orca-ER digital camera (Hamamatsu) controlled by Leica Application Suite X (Leica).Cell areas were measured from phase contrast images using BacStalk (80).
Cryo-electron microscopy B. ovis cells were grown on TSAB for 2 days and resuspended in PBS to a OD600 ≈ 1.Five µl of cells were placed on a glow-discharged R 3.51 grid (Quantfoil), blotted for 3.5 s, air dried for 20 s, and plunged into liquid ethane using the Vitrobot robotic plunge freezer (Thermo).Samples were stored in liquid nitrogen before imaging.Cells were imaged on a ThermoScientific Talos Arctica Cryo-EM (Thermo) with a -10.00 µm defocus and 17,500x magnification.Images were captured with a Ceta camera (Thermo) with 2.0 s exposure time.
RNA extraction and sequencing B. ovis strains were grown in Brucella broth at 37˚C in 5% CO2 overnight.The next day, cultures were back diluted to OD600 = 0.05 in 12 ml Brucella broth and incubated on a rotor at 37˚C with 5% CO2 for 7h. 9 ml of each culture was harvested by centrifugation, and the pellets were immediately resuspended in 1 ml TRIzol.Samples were stored at -80˚C until RNA extraction.To extract the RNA, samples were thawed and incubated at 65˚C for 10 minutes.200 µl of chloroform was added to each sample and mixed by vortexing for 15 seconds.Samples were then incubated at room temperature for 5 minutes.Phases were separated by centrifugation at 17,000 x g for 15 minutes at 4˚C.The aqueous layer was transferred into a fresh tube.Sample was mixed with 500 ml of isopropanol and stored at -20˚C for 2h.After thawing, samples were centrifuged at 13,000 x g for 30 minutes at 4˚C to pellet the RNA.Supernatants were discarded and the pellets were washed with 1 ml of ice cold 70% ethanol.Samples were centrifuged at 13,000 x g at 4˚C for 5 minutes.Supernatants were discarded and pellets were air dried.Pellets were resuspended in 100 µl RNAse-free H2O and incubated at 60˚C for 10 minutes.RNA samples were DNase treated using RNeasy Mini kit (Qiagen).RNA samples were sequenced at Microbial Genome Sequencing Center (Pittsburgh, PA) on an Illumina NextSeq 2000 (Illumina).RNA libraries for RNA-seq were prepared using Illumina Stranded RNA library preparation with RiboZero Plus rRNA depletion.

RNA-seq analysis
All RNA-seq analysis was conducted in CLC Genomics Workbench 21.0.4 (Qiagen).Reads were mapped to reference genome (Brucella ovis ATCC 25840, Genbank accessions NC_009504 and NC_009505).Differential expression for RNA-seq default settings were used to calculate the fold change of all annotated genes in mutant strains versus WT.Raw and processed RNA-seq data are publicly available in the NCBI GEO database at accession GSE229183.

ChIP DNA extraction and sequencing
To build the constructs for ChIP-seq analysis of strains expressing CenRD55E and EssRD64E, regions containing each gene (including ~300 bp upstream) were PCR amplified from the B. ovis cenRD55E and essRD64E mutant strains with KOD polymerase.Amplified fragments were purified and inserted into plasmid pQF through restriction enzyme digestion and ligation to yield a 3xFLAG fusion at the 3' end of each gene.This C-terminal FLAG fusion plasmid construct was conjugated into B. ovis mutants using WM3064 as described above.Cells were grown on TSAB plates supplemented with 30 mg/ml oxytetracycline for three days and resuspended in Brucella broth.Formaldehyde was added to a final concentration of 1% (w/v) to crosslink, and samples were incubated for 10 minutes on shaker.Crosslinker was quenched by adding 0.125 M glycine and was shaken for an additional 5 minutes.Cells were then centrifuged at 7,196 x g for 5 minutes at 4˚C and pellets were washed 4 times with cold PBS pH 7.5 and resuspended in 900 µl buffer [10 mM Tris pH 8, 1 mM EDTA, protease inhibitor cocktail tablet (Roche)].100 µl of 10 mg/ml lysozyme was added to each sample, which were then vortexed and incubated at 37˚C for 30 minutes before adding a final concentration of 0.1% SDS (w/v) to each sample.The samples were then sonicated on ice for 15 cycles (20% magnitude, 20 seconds on/off) and centrifuged at 15,000xg for 10 minutes at 4˚C.Supernatants were collected, Triton X-100 was added to a concentration of 1%, and lysates were added to 30 µl SureBead Protein A magnetic agarose beads (BioRad) equlibrated with binding buffer [10 mM Tris pH 8, 1 mM EDTA pH8, 0.1% SDS, 1% Triton X-100] and incubated for 30 minutes at room temperature; this step has empirically improved signal for our FLAG IP protocols.Beads were collected on a magnetic stand, the supernatant was transferred to a fresh tube, and 5% of each sample supernatant was removed as the input DNA samples.100 µl of α-FLAG magnetic agarose beads were then washed three times with binding buffer and incubated overnight at 4˚C shaking in 500 µl binding buffer plus 1% BSA to equilibrate the beads.The following day, α-FLAG beads were washed four times in binding buffer and cell lysates were added to the pre-washed beads and gently vortexed.Samples were incubated for 3h at room temperature with mixing and α-FLAG beads were collected on a magnetic stand and serially washed in 500 µl low-salt buffer [50 mM HEPES pH7.5, 1% Triton X-100, 150 mM NaCl], 500 µl high-salt buffer [50mM HEPES pH 7.5, 1% Triton X-100, 500 mM NaCl], and 500 µl LiCl buffer [10 mM Tris pH 8, 1 mM EDTA pH 8, 1% Triton X-100, 0.5% IGEPAL ® CA-630, 150 mM LiCl].To elute the protein-DNA complex bound to the beads, 100 µl elution buffer [10 mM Tris pH 8.0, 1 mM EDTA pH 8.0, 1% SDS, 100 ng/µl 3x FLAG peptide] was added to the beads and incubated at room temperature for 30 minutes with mixing.The eluate was collected, and the elution step was repeated.Input samples were brought to the same volume as output samples with elution buffer.For RNase A treatment, a final concentration of 300 mM NaCl and 100 µg/ml RNase A were added to each input and output sample and incubated at 37˚C for 30 minutes.Proteinase K was added to a final concentration of 200 µg/ml and the samples were incubated overnight at 65˚C to reverse crosslinks.ChIP DNA was purified with Zymo ChIP DNA Clean & Concentrator kit.The ChIP-seq library was prepared using Illumina DNA prep kit and IDT 10bp UDI indices; DNA samples were sequenced at SeqCenter (Pittsburgh, PA) on an Illumina Nextseq 2000.
ChIP-seq analysis Output and input DNA sequences were first mapped to the reference genome (Brucella ovis ATCC 25840, Genbank accessions NC_009504 and NC_009505) in Galaxy using bowtie2 (81).ChIP-seq enriched peak calls from the mapping output data were carried out in Genrich (https://github.com/jsh58/Genrich)with parameters maximum q-value = 0.05, and minimum AUC (area under the curve) = 20; PCR duplicates were removed.bamCoverage (82) was used to create bigwig file tracks for each replicate, and peaks were visualized in IGB (83).Raw and processed ChIP-seq data are publicly available in the NCBI GEO database at accession GSE229183.

Bacterial two hybrid β-galactosidase assay
The DNA fragments of the full-length, point mutant, and specific domains of cenR and essR were PCR amplified by KOD polymerase and cloned into split adenylyl cyclase plasmids (either pKT25 or pUT18c) (69) through restriction digestion and ligation.Each pair of pKT25 and pUT18c plasmids were co-transformed into chemically competent E. coli BTH101 cells.BTH101 strains carrying both pKT25 and pUT18c plasmids were grown in LB and incubated at 30˚C overnight while shaking.Fresh LB + 500 µM IPTG was inoculated with 100 µl of overnight culture and incubated at 30˚C until OD600 ≈ 0.3-0.4.To assess two-hybrid interaction via reconstitution of adenylyl cyclase activity, 100 µl of culture was mixed with 100 µl chloroform and vortexed vigorously.700 µl Z buffer (0.06 M Na2HPO4, 0.04 M NaH2PO4, 0.01 M KCl, 0.001 M MgSO4) and 200 µl ortho-nitrophenyl-β-galactoside (ONPG) was added to each sample.The reactions were stopped with 1 ml of Na2CO3 after 7.8 minutes and colorimetric conversion of ONPG was measured at 420 nm (A420) was measured.Miller Units were calculated as MU = A420 x 1000 / (OD600 x time x volume of cells).

Structure prediction
The heterodimeric structure of CenR and EssR was predicted using the protein complex prediction package AF2Complex (71), which utilizes the neural network models of AlphaFold2 (70).Computations were carried out on the Michigan State University high-performance computing cluster.
EssR-CenR co-immunoprecipitation assay Strain construction, lysate production, and α-FLAG magnetic agarose beads preparation were the same as described above for ChIP-seq sample preparation.After preparation and centrifugation of the lysates, the supernatants were collected and 50 µl of each sample was mixed with equal amount of SDS loading buffer and stored at -20˚C.For FLAG IP, remaining lysates were applied to pre-washed α-FLAG beads and incubated at room temperature for 3h with shaking.Beads were washed and eluted as described above, eluate from each sample was incubated at 65˚C overnight to reverse crosslinking, and samples were then mixed with equal volume of SDS loading buffer.All samples collected were heated to 95˚C for 5 minutes before resolving on a 12% mini-PROTEAN precast gel (BioRad).A Western blot using polyclonal EssR antiserum was conducted following the protocol outlined in the Western blot method section below.The membrane was imaged using BioRad Chemi-Doc Imaging System (BioRad).
Protein purification DNA fragments that encode full length CenR and EssR, and EssS residues 191-488 (EssS for short) were PCR amplified by KOD polymerase and inserted into a pET23b-His6-SUMO expression vector through restriction digestion and ligation.Vectors were transformed into chemically competent E. coli BL21 Rosetta (DE3) / pLysS.All strains were grown in LB medium at 37˚C and induced with 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at OD600 ≈ 0.5.Cell pellets were harvested after 2h of induction at 37˚C and stored at -80˚C until purification.For protein purification, cell pellets were resuspended in 25 ml of lysis buffer (25 mM Tris pH8, 125 mM NaCl, 10 mM imidazole) with the addition of 1mM phenylmethylsulfonyl fluoride (PMSF) and 5 µg/ml DNase I. Samples were lysed through sonication on ice (20% magnitude, 20 seconds on/off) until the lysates were clear.Lysates were centrifuged, and the supernatant was added to 3 ml of Ni-nitrilotriacetic acid (NTA) superflow resin (Qiagen) and the mixture was applied onto a gravity drip column.The resin was washed with 20 ml of lysis buffer, 50 ml of wash buffer (25 mM Tris pH8, 150 mM NaCl, 30 mM imidazole) and the proteins were collected in 20 ml of elution buffer (25 mM Tris pH8, 150 mM NaCl, 300 mM imidazole) at 4˚C.For CenR and EssR, the elution was dialyzed and the His6-SUMO tags were cleaved with ubiquitin-like-specific protease 1 (Ulp1) overnight at 4˚C in dialysis buffer (25 mM Tris pH8, 150 mM NaCl).The next day, the digested protein was mixed with NTA superflow resin.The mixture was applied onto a gravity column and the purified and cleaved protein was collected from the flow through.His6-SUMO-EssS was dialyzed in dialysis buffer overnight at 4˚C and collected.All proteins were stored at 4˚C.
Western Blotting of EssR protein B. ovis strains were grown on plain TSAB or TSAB supplemented with kanamycin for two days.Cells were collected and resuspended in sterile PBS to equivalent densities (measured optically at 600 nm).Cell resuspensions were mixed with an equal volume of SDS loading buffer and heated to 95˚C for 5 minutes.10 µl of each sample was loaded onto a 12.5% SDS-PAGE gel and resolved at 180V at room temperature.The proteins were transferred to a PVDF membrane (Millipore) using a semi-dry transfer apparatus at 10V for 30 minutes at room temperature.The membrane was blocked in 20 ml Blotto (Tris-buffered saline Tween-20 (TBST) + 5% milk) for 1h at room temperature.The membrane was blocked in 10 ml Blott + polyclonal rabbit α-EssR antiserum (1:1,000 dilution), and the membrane was incubated for 1h at room temperature.The membrane was washed three times with TBST.Goat α-rabbit IgG poly-horseradish peroxidase secondary antibody (Invitrogen; 1:10,000 dilution) was added to 10 ml Blotto and the membrane was incubated for 1h at room temperature.The membrane was washed three times with TBST, developed with ProSignal Pico ECL Spray (Prometheus Protein Biology Products) and imaged on a BioRad ChemiDoc Imaging System.Bands were quantified using ImageLab Software (BioRad).
Western Blotting of CenR-3xFLAG protein B. ovis strains carrying pQF plasmids were grown on TSAB supplemented with 30 µg/ml oxytetracycline for three days and samples for Western blot were collected and and prepared as described above.10 µl of each sample was loaded onto a 12.5% SDS-PAGE containing 0.5% 2,2,2-Trichloroethanol (TCE) and resolved at 180V at room temperature.The gel was imaged on a BioRad Chemi-Doc Imaging system using the stain-free protein gel setting to activate TCE and visualize total proteins in the gel.The proteins were then transferred to a PVDF membrane (Millipore) using a semi-dry transfer apparatus at 10V for 30 minutes at room temperature and imaged under the stain-free blot setting on Chemi-Doc to visualize the total transferred proteins.The membrane was blocked in 20 ml Blotto (Tris-buffered saline Tween-20 (TBST) + 5% milk) for 1h at room temperature, then incubated in 15 ml Blotto + monoclonal α-FLAG antibody (Thermo; 1:10,000 dilution) for 1h at room temperature.The membrane was subsequently washed three times with TBST and goat α-mouse IgG poly-horseradish peroxidase secondary antibody (Thermo; 1:10,000 dilution) was added and incubated for 1h at room temperature.Membrane incubated with secondary antibody was subsequently washed three times with TBST, developed with ProSignal Pico ECL Spray (Prometheus Protein Biology Products) and imaged on ChemiDoc Imaging System.Bands were quantified using ImageLab Software (BioRad).Conditions and dilution plating as in Figure 1; Plain TSA blood (TSAB) plates, TSAB + 2 µg/ml carbenicillin (+carb) and TSAB + 0.0045% SDS (+SDS).DcenR and DessR each carry an empty vector as control (EV).Experiments were repeated at least three times for all strains and one representative experiment is shown.ovis TCS gene loci BOV_1929 (cenR), BOV_0289, BOV_1472 (essR), BOV_1473 (essS), alone and in combination show a larger dynamic range at an SDS concentration of 0.004% compared to 0.0045% (as presented in Figure 1).Dilution plating experiments were repeated at least three times for all strains, and one representative experiment is shown.(C) Amino acid sequence alignment of CenR and EssR.Amino acids highlighted in yellow show the primary structure of the α4-β5-α5 proteinprotein interaction interface predicted by AFComplex2 (see Figure 5C).Protein secondary structure is presented above the alignment (helix in orange; strand in green).Alphaproteobacterial genomes were queried for reciprocal top BLAST hits of CenR, EssR, and EssS.A gene was determined as present (colored box) if a reciprocal top-hit pair was found, and as absent (white) if no such pair was found.The color of the boxes reflects the percent amino acid (AA) identity (over the full protein length) to the corresponding B. ovis sequences.B) A phylogenetic tree based on the amino acid sequence of EssS orthologs.Global alignment with free end gaps was used as the alignment type and Blosum62 as the cost matrix.Jukes-Cantor was used as the genetic distance model and neighbor-joining was the tree build method with no outgroup.

Figure 1 :
Figure 1: Analysis of B. ovis TCS deletion mutants identified five genes/operons that contribute to carbenicillin and SDS resistance in vitro.(A) Strains harboring in-frame unmarked deletions (D) of B. ovis TCS genes BOV_0577, BOV_0611-0612, and BOV_1602, carrying integrated empty vectors (EV) or genetic complementation vectors (::gene locus number) were plated in log10 dilution series on plain TSA blood agar (TSAB), TSAB containing 2 µg/ml carbenicillin (+carb) or TSAB containing 0.0045% SDS (+SDS).(B) Strains harboring in-frame unmarked deletions (D) of B. ovis TCS gene loci BOV_1929 (cenR), BOV_0289, BOV_1472 (essR), BOV_1473 (essS), alone and in combination plated as in panel A. Genetic complementation of panel B strains using a lower SDS concentration is shown in Figure S1.Dilution plating experiments were repeated at least three times for all strains, and one representative experiment is shown.(C) (top) Cartoon of the cenR and essR-essS genetic loci with bov gene locus numbers.(bottom) Protein domain models of CenR, EssR, and EssS with number of amino acid residues in each protein.Domains are labeled as follows: REC -receiver domain; HTH -helix-turn-helix DNA binding domain, sensor -extracellular sensor domain, HK -histidine kinase domain.Predicted transmembrane helices flanking the sensor domain are indicated by heavy black lines.

Figure 2 :
Figure 2: Functional analysis of the conserved aspartyl phosphorylation sites of the CenR and EssR response regulators.Test for genetic complementation of DcenR and DessR envelope stress phenotypes by expression of either non-phosphorylatable alleles of cenR (D55A) and essR (D64A) or putative phosphomimetic alleles (D55E / D64E).Conditions and dilution plating as in Figure1; Plain TSA blood (TSAB) plates, TSAB + 2 µg/ml carbenicillin (+carb) and TSAB + 0.0045% SDS (+SDS).DcenR and DessR each carry an empty vector as control (EV).Experiments were repeated at least three times for all strains and one representative experiment is shown.

Figure 4 :
Figure 4: CenR and EssRS regulate an overlapping set of genes and have a correlated ChIP-seq profile.(A) Heat map of log2 fold change in gene expression in ∆cenR and ∆essRS deletion strains relative to wildtype (WT).Genes presented have a fold change ³ |2| with an FDR p-value < 10 -4 .Genes highlighted in orange are adjacent on the chromosome.(B) EssR and CenR chromatin immunoprecipitation (ChIP)-seq peaks (q-value £ 0.05; minimum AUC = 20) along chromosomes 1 and 2. Red lines mark significant DNA peaks that are immunoprecipitated by CenR and EssR.Blue lines indicate peaks that are unique to either CenR or EssR.

Figure 5 :
Figure 5: CenR and EssR physically interact in a heterologous system and in B. ovis.(A,B) Measurement of homomeric and heteromeric interactions between EssR and CenR (and their aspartyl phosphorylation site mutants; D®A & D®E) using an E. coli bacterial two-hybrid assay.Proteins were fused to split adenylate cyclase fragments in vectors pUT18c and pKT25.Positive control (zip) and empty vector (EV) negative controls are shown.(C) CenR:EssR heterodimer structure showing interaction at the α4-β5-α5 structural face of each protein as predicted by AFComplex2 (71); CenR (green) and EssR (blue) (D) Test for interactions between DNA-binding domain (DBD) and receiver domain (Rec) fragments of EssR and CenR by bacterial two-hybrid.Error bars represent the standard deviation of 4 biological replicates.Statistical significance was calculated using one-way ANOVA, followed by Dunnett's multiple comparisons test (p < 0.0005, ***; p < 0.0001, ****) to EV control or WT CenR-EssR interaction.(E) Co-immunoprecipitation of EssR and CenR-3xFLAG from B. ovis lysate.(left) CenR-3xFLAG was captured using anti-FLAG beads, which were washed before elution.(right) EssR association with CenR-3xFLAG in the eluate was monitored by Western blot using polyclonal antiserum to EssR (a-EssR).Non-specific (n.s.) cross-reactive band is shown as an indicator of loading.Representative blot from three biological replicates.

Figure 8 :
Figure 8: Model of EssS-EssR-CenR-dependent gene regulation in B. ovis.Upon detection of host signals, EssS (pink) autophosphorylates and transfers a phosphoryl group (P) to EssR (blue).CenR (green) supports EssS-EssR signal transduction by directly stimulating phosphoryl transfer from EssS and EssR via a receiver domain interaction.Loss of CenR results in diminished EssR levels (light blue with dashed outlines) and loss of EssR results in diminished CenR levels (light green with dashed outlines), via a post-transcriptional mechanism.CenR and EssR directly interact with an overlapping set of promoters, and three possible modes of transcriptional regulation by EssR and CenR are shown: EssR-EssR homodimer, EssR-CenR heterodimer, and/or CenR-CenR homodimers binding to target promoters.Genes regulated by the EssS-EssR-CenR system impact B. ovis cell size, contribute to growth/survival during cell envelope stress in vitro, and support intracellular replication in a macrophage infection model.

Figure S2 :
Figure S2: cenR, essR, and essS contribute to cell survival in the presence of diverse envelope stressors.Strains harboring in-frame unmarked deletions (D) of cenR, essR, essS, carrying integrated empty vectors (EV) or genetic complementation vectors (::gene locus number) were plated in log10 dilution series on plain TSA blood agar (untreated), TSA-B containing 2.75mM EDTA (+EDTA), TSA-B containing 215 mM NaCl (+NaCl), or were cells treated with 1 mg/mL polymyxin B for 80 minutes before being plated on TSA-B (polymyxin B treated).Dilution plating experiments were repeated three times for all strains, and one representative experiment is shown.

Figure S4 :
Figure S4: ∆cenR, ∆essS, and ∆essR are not sensitive to acidic pH.Strains harboring in-frame deletions of cenR, essS, and essR were incubated in Brucella broth at pH 7.0 or Brucella broth at pH 4.2 for 2 h before being serially diluted and plated on TSAB.CFU of treated cultures (pH 4.2) were normalized to CFU of untreated cultures (pH 7).Error bars represent the mean ± standard deviation of the four replicates.Statistical significance was calculated using one-way ANOVA, followed by Dunnett's multiple comparisons test to WT (n.s.= non-significant, p > 0.05).

Figure S5 .
Figure S5.CenR enhances dephosphorylation of EssS~P in the presence of EssR.A) In vitro phosphoryl transfer assay with purified EssS kinase domain and EssR (1:1) and increasing ratiometric amounts of CenR.All reactions were stopped after 20s.B) Quantification of EssS~P levels; mean EssS~P band intensity is set to 1 in the EssS~P only reaction.Normalized EssS~P levels 20s after addition of EssR and CenR in varying ratios is plotted.Error bars represent the standard deviation of three biological replicates.

Figure S6 :
Figure S6: CenR, EssR, and EssS are widely distributed in the class Alphaproteobacteria. A) RepresentativeAlphaproteobacterial genomes were queried for reciprocal top BLAST hits of CenR, EssR, and EssS.A gene was determined as present (colored box) if a reciprocal top-hit pair was found, and as absent (white) if no such pair was found.The color of the boxes reflects the percent amino acid (AA) identity (over the full protein length) to the corresponding B. ovis sequences.B) A phylogenetic tree based on the amino acid sequence of EssS orthologs.Global alignment with free end gaps was used as the alignment type and Blosum62 as the cost matrix.Jukes-Cantor was used as the genetic distance model and neighbor-joining was the tree build method with no outgroup.