Protein with negative surface charge distribution, Bnr1, shows characteristics of a DNA‐mimic protein and may be involved in the adaptation of Burkholderia cenocepacia

Abstract Adaptation of opportunistic pathogens to their host environment requires reprogramming of a vast array of genes to facilitate survival in the host. Burkholderia cenocepacia, a Gram‐negative bacterium with a large genome of ∼8 Mb that colonizes environmental niches, is exquisitely adaptable to the hypoxic environment of the cystic fibrosis lung and survives in macrophages. We previously identified an immunoreactive acidic protein encoded on replicon 3, BCAS0292. Deletion of the BCAS0292 gene significantly altered the abundance of 979 proteins by 1.5‐fold or more; 19 proteins became undetectable while 545 proteins showed ≥1.5‐fold reduced abundance, suggesting the BCAS0292 protein is a global regulator. Moreover, the ∆BCAS0292 mutant showed a range of pleiotropic effects: virulence and host‐cell attachment were reduced, antibiotic susceptibility was altered, and biofilm formation enhanced. Its growth and survival were impaired in 6% oxygen. In silico prediction of its three‐dimensional structure revealed BCAS0292 presents a dimeric β‐structure with a negative surface charge. The ΔBCAS0292 mutant displayed altered DNA supercoiling, implicated in global regulation of gene expression. Three proteins were identified in pull‐downs with FLAG‐tagged BCAS0292, including the Histone H1‐like protein, HctB, which is recognized as a global transcriptional regulator. We propose that BCAS0292 protein, which we have named Burkholderia negatively surface‐charged regulatory protein 1 (Bnr1), acts as a DNA‐mimic and binds to DNA‐binding proteins, altering DNA topology and regulating the expression of multiple genes, thereby enabling the adaptation of B. cenocepacia to highly diverse environments.

soil and watercourses, and their success as human pathogens relies on their plasticity to adapt to varying conditions (Cullen & McClean, 2015;Hogardt & Heesemann, 2010;Howard et al., 2012;Jeong et al., 2010;Marvig et al., 2014;Trunk et al., 2010). Antimicrobial resistance (AMR) ranks among the most important threats to human health. Among the most problematic AMR bacteria, the so-called ESKAPE pathogens (acronym for E. faecium, Staphylococcus aureus, Klebsiella pneumoniae, A. baumannii, P. aeruginosa and Enterobacter spp), are opportunistic pathogens that are highly adaptable in their ability to colonize a diverse range of niches including the human host (De Oliveira et al., 2020). Insights into the adaptability of these successful opportunistic pathogens will enable the development of alternatives to antimicrobial therapies.
Other examples of exquisitely adaptable pathogens which colonize very diverse ecological niches ranging from contaminated soils, water sources, and pharmaceutical plants to the human lung include the Gram-negative genus Burkholderia. These bacteria possess large genomes (7.6-8 .5 Mb; Garcia-Romero & Valvano, 2020;Holden et al., 2009) and have enormous genetic potential for adaptation within the host. Within this genus, Burkholderia cepacia complex (Bcc) is a group of closely related bacteria that causes opportunistic chronic infections in people with cystic fibrosis (CF; Mahenthiralingam et al., 2008). The complex currently comprises at least 22 validly named species (Bach et al., 2017;Depoorter et al., 2020) that are highly resistant to different classes of antibiotics (Zhou et al., 2007). Some members of the complex, notably B. cenocepacia and B. multivorans have been shown to survive in macrophages and to attach to human epithelial cells (Caraher et al., 2007;Cullen et al., 2017;Saini et al., 1999;Valvano, 2015). Further, B. cenocepacia can persist in 6% oxygen through a coregulated 50-gene cluster of genes, designated the low-oxygen activated (lxa) locus (Sass et al., 2013) which we have shown to be upregulated during chronic infection in response to the hypoxic environment of the CF lung (Cullen et al., 2018).
Previously, we identified an immunogenic protein encoded by the BCAS0292 gene in B. cenocepacia using serum from Bcc-colonized people with CF (Shinoy et al., 2013). This gene encodes a hypothetical protein and is located within a virulence gene cluster on replicon 3 of the B. cenocepacia genome (Figure 1), which also includes aidA (BCAS0293). The aidA gene resides in the same operon as BCAS0292 and encodes a protein that is required for virulence in C. elegans (Huber et al., 2004). Both BCAS0292 and AidA proteins have been classified as PixA proteins, a poorly understood group of inclusion body proteins associated with cells in the stationary phase (Goetsch et al., 2006;Winsor et al., 2008). Both BCAS0292 and aidA are the most highly regulated genes in the quorum-sensing mutants, ΔcepR, and ΔcepRcciIR (O'Grady et al., 2009). Both genes are also involved in adaptation to the host in a rat chronic infection model (O'Grady & Sokol, 2011). Further, transcripts of both genes are upregulated under low oxygen conditions, and BCAS0292 is also the most upregulated gene in stationary phase growth, suggesting it may be involved in the stress response of B.
cenocepacia during anoxic and low nutrient conditions (Sass et al., 2013). Overall, these studies suggest a role for the BCAS0292encoded protein in the stress response of B. cenocepacia; however, its specific function remained unknown. In this study, we constructed a ΔBCAS0292 deletion mutant strain to investigate its role in B. cenocepacia pathogenesis. We demonstrate that loss of BCAS0292 alters the abundance of more than a thousand proteins and causes a range of pleiotropic effects, suggesting the protein is involved in the global regulation of gene expression. We also provide evidence indicating the protein has surface characteristics comparable to known DNA mimic proteins and propose that BCAS0292 functions as a DNA mimic by interacting with DNA-binding proteins and altering the state of DNA supercoiling and have dubbed the protein Burkholderia negative surface charged regulatory protein 1 (Bnr1).
2 | Experimental procedures 2.1 | Bacterial and epithelial cell culture maintenance The CFBE41o − cell line, which is homozygous for the ΔF508 mutation, was maintained in fibronectin/vitrogen coated flasks containing minimal essential medium with 10% fetal bovine serum, 1% penicillin/streptomycin, 1% L-glutamine, and 1% nonessential amino acids and incubated in 5% CO 2 environment at 37°C (Wright et al., 2011). 2.2 | Mutagenesis of B. cenocepacia K56-2 and complementation Unmarked, nonpolar gene deletions were constructed as previously described . The deletion of BCAS0292 was confirmed by DNA sequencing of the polymerase chain reaction (PCR) product spanning the deletion site. To complement B. cenocepacia K56-2_ΔBCAS0292, wild-type (WT) BCAS0292 was amplified from B. cenocepacia K56-2 (see primer pairs in Table A1), digested with the restriction enzymes NdeI and XbaI, and ligated into similarly digested pMH447. The complementation plasmid was introduced into the mutant by conjugation. Once transferred into the target mutant strain the complementation vector integrates into the genome at aminoglycoside efflux genes (BCAL1674-BCAL1675; Hamad et al., 2010). As before, pDAISce-I was then introduced, resulting in the replacement of BCAL1674-1675 by BCAS0292. The complementation of BCAS0292 was confirmed by PCR and then confirmed by phenotype analysis. The mutant and complement strains were cured by growth on LB agar enriched with 5% (wt/vol) sucrose. The growth of all the strains was measured by inoculating LB broth in a 96-well plate pre-equilibrated in a BioTek Synergy H1 Hybrid reader at 37°C with log growing cultures in replicates of six. Growth was monitored by measuring the OD 600 every 30 min for 24 h.

| Galleria mellonella infection model
Log-phase grown bacteria were suspended to a starting OD 600nm in 0.1 in sterile phosphate buffer saline (PBS) and serially diluted to 10 −7 . Each dilution (20 µl) was injected into the hindmost left proleg of G. mellonella larvae using a sterile Terumo 0.3-ml syringe (10 larvae per group; Costello et al., 2011). An equal volume of each dilution was also plated onto LB agar counted at 48 h to accurately quantify bioburden injected. Ten larvae injected with PBS alone served as controls. Larvae were incubated at 37°C and examined for survival at 24, 48, and 72 h following injection.

| Cloning and expression of recombinant BCAS0292
The forward primer was designed to contain the "CACC" sequence at the beginning of the primer to introduce a 5′ overhang allowing directional cloning of BCAS0292 into the pET100/D-Topo vector in frame with the N-terminal 6xHis tag. PCR amplification was carried out using HotStar Taq polymerase (Qiagen) and amplicons cloned into the pET100/D-Topo vector as previously described (Dennehy et al., 2017). Each transformation reaction was spread onto prewarmed selective plates containing 50 μg/ml of ampicillin and incubated overnight at 37°C and transformants identified by PCR. Escherichia coli BL21 Star™ (DE3) One Shot ® cells (Thermo Fisher Scientific) were used for protein expression (Dennehy et al., 2017). Following the confirmation of recombinant BCAS0292 (rBCAS0292) expression in a pilot study, 1 L cultures of BL21 cells transformed with the expression plasmid, were grown in LBampicillin (100 μg/ml), and induced with isopropyl ß-D-1-thiogalactopyranosideI (IPTG) at a final concentration of 1 mM overnight at 25°C. Bacteria were then pelleted using centrifugation at 2500 g for 10 min. Supernatants were discarded, pellets weighed and stored at −80°C.

| Recombinant BCAS0292 purification
Purification of rBCAS0292 was carried out as previously described (Dennehy et al., 2017). Ice-cold lysis buffer containing 6 M Guanidine-HCL, 100 mM sodium phosphate, 10 mM Tris-HCL, (pH 8), and 1X EDTA-free protease inhibitor was added to bacterial pellets at a volume of 30 ml/7 g of pellet. Cells were resuspended in lysis buffer and lysed on ice with a cell disruptor (maximum output, two 30-s pulses). Cell debris and any remaining intact cells were pelleted using centrifugation at 4800g for 35 min. Supernatants were filtered through a 0.45 μM filter and incubated with a rotating HisPur Ni-NTA column (Thermo Fisher Scientific), which was pre-equilibrated with Buffer A, for 1 h at 4°C. Unbound proteins were removed by a series of washes and bound protein was subsequently eluted in two steps with 250 mM and 300 mM imidazole.
Following purification, the protein concentration of each fraction was determined using the Bradford assay. Protein purity was determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).

| Attachment of bacteria to epithelial cells
To quantify the attachment of Bcc or BCAS0292-expressing E. coli strains to lung epithelial cells, CFBE41o − cells were seeded in coated chamber slides 24 h before the 30 min incubation of individual bacterial strains at an MOI of 50:1, as previously described (Dennehy et al., 2017). In the case of studies with recombinant E. coli strains, rBCAS0292 expression was induced in all strains (E. coli BL21, E. coli BL21_BCAS0292) with 1 mM IPTG for 3 h before incubation with epithelial cells. Unattached bacteria were removed by washing with sterile PBS four times for 5 min each. The CFBE41o − cells and attached bacteria were fixed in 3% paraformaldehyde (pH 7.2) for 10 min at room temperature (RT). Cells were gently washed once with PBS and then blocked in PBS containing 5% bovine serum albumin (BSA) for 1 h at RT. Cells were then incubated overnight at 4°C in PBS containing 1% BSA and either a rabbit anti-Bcc antibody (1:800 dilution) or an anti-E. coli fluorescein isothiocyanate (FITC)-conjugated antibody (1:100 dilution). The following day cells were washed three times with PBS before the addition of FITC-conjugated anti-rabbit antibody (1:500 dilution), in the case of Bcc strains, for 1 h in the dark. The secondary antibody was removed, and cells were washed DENNEHY ET AL. | 3 of 21 twice for 5 min each with PBS. Cells were then counterstained with 4′,6diamidino-2-phenylindole and phalloidin conjugated to Alexa fluor 568 (5 U/ml) for 15 min in the dark at a concentration of 1 μg/ml in PBS. A coverslip was applied to each chamber before bacteria and cells were visualized using confocal microscopy. The bacterial attachment was counted in 20 randomly selected fields for each strain and values were expressed as the number of bacteria/100 cells.

| Membrane permeability
Membrane permeability was determined by measuring the accumulation of the fluorescent dye H33342 as outlined previously (Dennehy et al., 2017). Briefly, strains were aliquoted in replicates of eight in a 96well plate and pre-equilibrated in a BioTek Synergy H1 Hybrid reader at 37°C. H33342 (10 μM) was added to each well and the plates were incubated for 90 min at 37°C with fluorescent readings being taken every minute (λ ex = 355 nm, λ em = 460 nm). The mean fluorescence of eight replicates relative to the control fluorescence at each time point and at zero time was used to determine the dye accumulation in each strain.

| Biofilm formation
Biofilm formation was determined as outlined previously (Caraher et al., 2007) by inoculation of 100/well of mid-log phase cultures (OD 600 = 0.6) into microtiter plates in replicates of eight. The plates were incubated at 37°C for 24, 48, or 72 h, rinsed five times with water to remove unadhered bacteria, and air-dried, airdried, and stained with crystal violet (0.125%) for 30 min at RT. The plates were then rinsed again five times and the dye dissolved with 95% ethanol.
The contents of the wells were then transferred to a new polystyrene microtitre plate and absorbance was determined at 590 nm (Tecan Plate Reader; Alpha technologies). Biofilm formation was defined as the wells showing A590 > 0.05.

| Whole proteome analysis
To extract whole cell lysates were extracted from 18 h stationary phase cultures of three biological replicates of K56-2 WT and the ΔBCAS0292 mutant strains. Bacterial cells were pelleted with centrifugation of 6000g for 10 min. Ice-cold lysis buffer containing 8 M urea, 25 mM Tris-HCL, 10 mM 1,4-dithiothreitol (DTT), and 1X EDTA-free protease inhibitor cocktail (pH 8.6) was added to pellets, using a volume of 8 ml/g bacterial pellet. Pellets were resuspended and cells lysed using a cell-disruptor probe set with an output power of 16, using four 30-s bursts on ice. Lysates were then centrifuged at 14,000g for 15 min, supernatants were transferred to fresh tubes, and a Bradford assay was used to determine the protein concentration of each sample. DTT (1 M) was added to each supernatant (10 μl/ml lysate) and incubated at 56°C for 30 min followed by 1 M iodoacetamide (55 μl/ml lysate) which was incubated in darkness at RT for 20 min. Lysates were then dialyzed in dialysis tubing with a cut-off of 3.5 kDa against 100 mM ammonium bicarbonate overnight with stirring at 4°C then the ammonium bicarbonate was replaced with fresh solution and incubated for a further 6 h. Trypsin (400 ng/μl) was added to dialyzed protein (5/100 μl protein) and tubes were incubated overnight at 37°C. Aliquots (20 μl) from each sample were placed in fresh tubes and samples were dried in a speedy vac using medium heat and resuspended in 20 μl resuspension buffer containing 0.5% TFA in MilliQ water. Tubes were sonicated for 2 min and centrifuged briefly. ZipTips (Millipore) were wetted with 10 µl of a wetting solution containing 0.1% TFA in 80% acetonitrile slowly five times. Equilibration solution (10 µl) containing 0.1% TFA in MilliQ water was aspirated and dispensed into the ZipTip, five times before 10 µl of the resuspended sample was then aspirated and dispensed slowly into the ZipTip, 15 times. Then 10 µl of the washing solution containing 0.1% trifluoroacetic acid (TFA) in MilliQ water was aspirated into the ZipTip and dispensed to waste.
This step was repeated five times before 10 µl of the elution solution containing 0.1% TFA in 60% acetonitrile was aspirated and dispensed from the ZipTip to a fresh 1.5 ml tube six times. Samples were then dried under medium heat and resuspended in 15 μl of loading buffer containing 0.05% TFA in 2% acetonitrile. Tubes were sonicated for 2 min, centrifuged briefly and samples were transferred into fresh Perseus to organize the data (version 1.4.1.3; Cox & Mann, 2008;Cox et al., 2011;Tyanova Temu, & Cox, 2016a;Tyanova, Temu, & Sinitcyn, 2016b). For protein identification, the following search parameters were used: precursor-ion mass tolerance of 1.5 Da, fragment ion mass tolerance of 6 ppm with cysteine carbamidomethylation as a fixed modification, and a maximum of two missed cleavage sites allowed. False discovery rates (FDR) were set to 0.01 for both peptides and proteins and only peptides with a minimum length of six amino acids were considered for identification. Proteins were considered identified when a minimum of two peptides for each parent protein was observed. LFQ intensities measured for individual runs were grouped based on their experimental treatment. The data were log2 transformed and an analysis of variance (ANOVA) (p < 0.05) was performed between the control and individual treatment samples to identify differentially abundant proteins.
A log2 fold change ≥ 1.5 and log2 fold change ≤−1.5 with adjusted p < 0.05 were considered. Proteins with statistically significant differential expression were extracted and these were used to generate maps of expression. To improve the visual representation of differentially abundant proteins, mean values were generated for each treatment and used to build the maps. A qualitative assessment was also conducted.
This involved the identification of proteins that were completely absent in a specific strain. Those proteins that were completely missing from all replicates of a particular group but present in other groups were determined manually from the data matrix. These proteins are not considered statistically significant as the values for absences are given as not a number which is not a valid value for an ANOVA analysis. However, the complete absence of a protein from a group may be biologically significant and these proteins are reported as qualitatively differentially expressed.

| Bacterial cell shape analysis
Cell shape was examined according to published methods (Shiomi et al., 2008). Cells were grown at 37°C in LB overnight before reinoculation and culture to achieve log growth (OD 600 of 0.5-0.7).
Optical sectioning was performed with an Olympus FV-10 ASW confocal microscope using a Plan Apo X60 1.40 oil immersion objective lens and Olympus FLUOVIEW Ver.3.1 software. Section images were captured along the z-axis at 0.2-μm intervals.

| Circular dichroism spectroscopy
Circular dichroism measurements were carried out using a Jasco J-810 spectropolarimeter equipped with a Peltier temperature control system (Model PTC-423-S). The molar ellipticity per mean residue, [θ](deg −1 was calculated from the equation obs is the ellipticity measured in degrees, mrw is the mean residue molecular mass (105.3 Da), C is the protein concentration in g × L −1 and l is the optical path length of the cell in centimeter. Far-UV spectra (190-260 nm) of recombinant BCAS0292 protein were recorded at 293 K using a 0.1 cm optical path-length cell, with a protein concentration of 0.2 mg ml −1 . For thermal denaturation experiments, ellipticity was monitored at 222 nm with a temperature slope of 1°C/min from 20 to 100°C. Melting temperature (T m ) was calculated from the maximum of the first derivative of the unfolding curves. Tris-HCl, 150 mM NaCl, 5% glycerol, pH 7.5 and. The online measurement of the intensity of the Rayleigh scattering as a function of the angle as well as the differential refractive index of the eluting peak in SEC was used to determine the weight-average molar mass (MW) of eluted protein, using the Astra 5.3.4.14 (Wyatt Technologies) software.

| Homology modeling
The homology model structure of BCAS0292 was built after consensus-based sequence comparison using HHPRED. The best model template was identified as the structure of the PixA inclusion body protein from Burkholderia cenocepacia (PDB code 4lzk, sequence identity 23.2%, E-value: 7.2e−38). Using this alignment, the homology model was built using the program MODELLER (Bitencourt-Ferreira & De Azevedo, 2019). As a validation tool, an independent model was computed using i-TASSER (J. Yang & Zhang, 2015) and the two models were compared. The meta-threading approach LOMETS was used to retrieve template proteins of similar folds from the PDB library. The program SPICKER was used to cluster the decoys based on the pair-wise structure similarity. The best confidence model presented a C-score −0.34 (J. Yang & Zhang, 2015). The two approaches provided consistent models, with root-mean-square deviation computed on Cα atoms of 1.5 Å. Based on the estimated local accuracy of the model computed by i-TASSER, the C-terminal 10 residues were removed from the model. The electrostatic potential surface was computed using the program Chimera (Z. Yang et al., 2012). 2.14 | Chloroquine gel electrophoresis B. cenocepacia strains were transformed with the pDA-12 reporter plasmid and cultured on plates containing tetracycline (150 μg/ml), ampicillin (100 μg/ml), and polymyxin B (50 μg/ml). The reporter plasmids were extracted from either log or stationary phase cultures using a Qiagen Miniprep Kit (Thermo Fisher Scientific). The plasmids were analyzed on chloroquine gels (1.5% agarose, 10 μg/ ml chloroquine) separated for 24 h at 100 V, 150 mA (Scanlan et al., 2017). The gels were washed 8-12 times over 8 h to remove chloroquine, before staining with ethidium bromide (10 μg/ml) for 2 h. The gel was washed for 1 h before imaging under UV light (Scanlan et al., 2017).

| FLAG-pulldown assay to identify BCAS0292 binding partners
The BCAS0292 gene was cloned into a pT7-FLAG-2 plasmid (Sigma-Aldrich) and transformed into E. coli BL21 STAR (DE3) One Shot cells (Invitrogen) for expression. Expression of the recombinant FLAG-BCAS0292 protein on induction with 1 mM IPTG for 24 h was DENNEHY ET AL. | 5 of 21 confirmed using SDS-PAGE. For the preparation of the bait lysates, E. coli BL21 cells transformed with FLAG-tagged BCAS0292 plasmid were incubated in the presence of 1 mM IPTG at 37°C for 2 h and overnight at RT without shaking. Prey lysates were prepared from K56-2 WT cells grown to OD 600 1.2 at 37°C at 200 rpm. Cells were centrifuged and pellets washed in PBS. The prey lysates pellets were also washed in Tris buffer saline (TBS) (5 ml) three times with centrifugation. Pellets were lysed in 50 mM Tris HCl, 150 mM NaCl, 1 mM EDTA, 1% Triton with protease inhibitor cocktail and DNAse (1 IU) and sonicated for 10 pulses on ice. Cell debris was pelleted by centrifugation at 4800 g (30 min) and lysates were transferred to fresh tubes and stored at −20°C until use. Anti-FLAG resin was equilibrated in TBS and washed five times before adding to all tubes.
The bait cell lysate was immobilized by adding to bait control and bait:prey tubes. TBS was added to the prey control tube. Tubes were incubated at 4°C with shaking for 4 h. The tubes were centrifuged and the resin was washed five times with TBS. The prey cell lysates were added to prey control and bait:prey tubes and TBS added to bait controls and incubated at 4°C overnight before centrifuging and washing with TBS five times. Pull-down proteins were eluted in 2X sample buffer (10% SDS, 20% glycerol, 125 mM Tris [pH 6.8], 0.004% bromophenol blue) and boiled for 3 min before loading on 12% SDS-PAGE gels. This was performed three times with independent preparations of prey and bait from independent cultures on each occasion. Protein bands of interest were manually excised from the gels and in-gel trypsin digestion was performed (Shevchenko et al., 2002) followed by Zip-tip clean up and identified in technical triplicates with a Thermo Scientific Q Exactive Mass Spectrometer connected to a Dionex Ultimate 3000 RSLCnano chromatography system. Proteins were separated on a C18 column (C18RP Reposil-Pur; 100 × 0.075 mm × 3 μm) over 60 min at a flow rate of 250 nl/min with a linear gradient of increasing ACN from 1% to 27%. The MS was operated in datadependent mode; a high resolution (70,000) MS scan (300-1600 m/z) was performed to select the 12 most intense ions and fragmented using high energy C-trap dissociation for MS/MS analysis.
Raw data from the Q-Exactive was processed using MaxQuant (Cox & Mann, 2008;Tyanova Temu, & Cox, 2016a;version 1.6.3.4) incorporating the Andromeda search engine (Cox et al., 2011). MS/ MS spectra were matched against Uniprot B. cenocepacia database (UP000001035) containing 6933 entries and E. coli BL-21 database (UP000002032) containing 4156 entries. All searches were performed using the default setting of MaxQuant, with trypsin as specified enzyme allowing two missed cleavages and a FDR of 1% on the peptide and protein level. The database searches were performed with carbamidomethyl (C) as fixed modification and acetylation (protein N terminus) and oxidation (M) as variable modifications. The label-free quantification (LFQ) algorithm was used to generate normalized spectral intensities and infer relative protein abundance.
Statistical analysis was performed using Perseus software (Tyanova, Temu, & Sinitcyn, 2016b; version 1.6.15.0). To examine changes in protein expression using volcano plots, data were log-transformed, and missing values were imputed with values from a normal distribution. Quantitative analyses were performed using a Student's t-test to compare bait:prey and prey samples. Proteins with significant changes in abundance (p < 0.05) were included in the quantitative results. Data were normalized using z-score and visualized using heat maps.

| Statistical analysis
Analysis of virulence data was performed using a log-rank (Mantel-Cox) test on the survival curves. Statistical analysis of host cell attachment, biofilm formation, and hypoxic growth and survival were carried out by two-way ANOVA using Prism software.

| RESULTS
BCAS0292/Bnr1 belongs to the PixA protein family. These are inclusion body proteins first identified in Xenorhabdus nematophila which are typically between 173 and 191 amino acids in length (Goetsch et al., 2006). They are elevated in cells transitioning to the stationary phase (Goetsch et al., 2006), but are poorly understood.
Bnr1 has a calculated molecular weight of 19.2 kDa, a calculated pI of 4.81, and has 34 homologs in several sequenced B. cenocepacia strains, and in B. anthina, B. cepacia, B. pyrrocinia, and other Burkholderia species (Winsor et al., 2008). A BlastP search showed that it is not found outside Burkholderia (Altschul et al., 1997). The it also shares sequence similarity with PixB of X. nematophila (Lucas et al., 2018). The aidA gene is more widely distributed across Burkholderia species with 417 orthologs identified to date (Winsor et al., 2008). AidA also shares a high level of identity with PixB proteins across Gammaproteobacteria, Alphaproteobacteria, Actinobacteria, and species (Lucas et al., 2018). Although Bnr1 is immunoreactive, indicating that it is expressed during human infection, we did not detect AidA in our immunoproteomic experiments (Shinoy et al., 2013). To investigate the potential function of Bnr1 in infection, we constructed a deletion mutant of BCAS0292 in K56-2, a B. cenocepacia strain that is amenable to genetic manipulation and a complemented strain (referred to as ΔBCAS0292 and Δ0292_0292, respectively) and compared their phenotypes to the WT strain. The growth of the mutant and complement strains was comparable to that of the WT K56-2 strain (Figure 1b

| Potential role in host-cell interactions
We have previously shown that other immunogenic proteins, including peptidoglycan-associated lipoprotein, are involved in attachment to host epithelial cells (Dennehy et al., 2017, McClean et al., 2016. The mutant strain ΔBCAS0292 displayed a 2.2-fold reduction in attachment to CFBE41o − cells relative to the WT strain 3.3 | Analysis of the susceptibility of K56-2 WT and ΔBCAS0292 to polymyxin B and meropenem Bnr1 expression was previously shown to be increased in response to antibiotics but deletion of the genes encoding Bnr1 and BCAS0293 did not alter antibiotic susceptibility (Sass et al., 2011). To further examine antimicrobial susceptibility, the sensitivity of the ΔBCAS0292 mutant to meropenem and polymyxin B (not previously examined) was measured to determine if the absence of this protein affected susceptibility to either of these antibiotics. The ΔBCAS0292 mutant was eightfold more sensitive to polymyxin B compared to the WT and complemented strains with a minimum inhibitory concentration (MIC) of 12 μg/ml for ΔBCAS0292 compared to 96 μg/ml for the WT and complemented strains (Figure 3a). Polymyxin B targets lipopolysaccharide components (Mares et al., 2009) and destabilizes the outer membrane; suggesting that Bnr1 may play a role in the maintenance of outer membrane integrity. This was supported by the enhanced permeability of the ΔBCAS0292 mutant to the fluorescent dye Hoechst H33342 relative to WT (Figure 3b). Membrane permeability was partially restored to WT levels in the complement strain. In contrast, ΔBCAS0292 was more resistant to the β-lactam meropenem in comparison to the WT and complemented strains with a MIC of 4 μg/ml for ΔBCAS0292 compared to 1.5 μg/ml for the WT and complemented strains (Figure 3c).  (Table 1).

| Whole proteome analysis of WT and ΔBCAS0292 strains
The lack of a direct effect on host cell attachment, combined with the apparent repression of biofilm formation and behavior in response to antibiotics suggested that the Bnr1 protein has a regulatory effect. To determine if Bnr1 is a regulatory protein we compared the proteome profiles of the ΔBCAS0292 mutant with WT under stationary phase conditions. Stationary-phase cultures were used for this experiment since the expression of Bnr1 is maximal in the stationary phase (Sass et al., 2013). Despite there being only a single gene deletion, we identified 2305 proteins showing different abundance between the strains, of which 9 proteins were unique to the ΔBCAS0292 mutant while 19 were unique to K56-2 WT cells (Table A3) We identified 406 proteins with significantly increased abundance by ≥1.5-fold in ΔBCAS0292 while 545 proteins (57%) showed significantly reduced abundance by 1.5-to 104-fold, suggesting Bnr1 is a global positive regulator of protein expression (Figure 4a).
Among the proteins showing a reduced abundance in the ΔBCAS0292 mutant were proteins involved in metabolism, virulence, stress responses, and uncharacterized proteins. The fold reduction in protein abundance was substantial: 81 proteins showed reduced abundance (ranging from 3-to 104-fold) in the mutant, and 232 proteins were reduced by over twofold (Table 2).
In contrast, the greatest fold increase in abundance was only 11.7, including proteins with roles in metabolic processes, translation, and the stress response (Table S4 at   The abundance of many proteins with roles in the stress response was reduced in the ΔBCAS0292 mutant, representing 6% of the total number of downregulated proteins in this group. Of particular interest was the reduced abundance of 19 proteins encoded on the low-oxygen activated (lxa) locus (BCAM0272 to 0323) in the mutant strain (Tables 2 and A3). This locus is upregulated under hypoxic conditions and in the stationary phase (Sass et al., 2013) and in sequential clinical B. cenocepacia isolates over time of chronic infection (Cullen et al., 2018). Eight lxa-encoded proteins that showed reduced abundance have stress response functions, including a heat shock/α-crystallin protein (BCAM0278) which was dramatically reduced (104.7-fold) relative to K56-2 WT.
Five of the six universal stress proteins (USPs) encoded on the lxa locus showed reduced abundance in the ΔBCAS0292 mutant strain by up to 4.5-fold. Other stress response proteins were also reduced in abundance (1.5-2.2-fold). Four other low-oxygen coregulated gene loci have been identified (Sass et al., 2013) and five proteins encoded by these lxa coregulated genes were also reduced in abundance: BCAM1480-BCAM1482, another USP (BCAM1495), and BCAM1496 (Table 2).
Proteins with roles in various metabolic processes represented over a third of the total number of proteins reduced in abundance, including a plasmid-encoded NUDIX hydrolase family protein  (Tables 2 and Table S4 at  A range of virulence proteins, including three T6SS secretion system proteins BCAL0337, tssD, and hcp1, the quorum sensing regulated metalloprotease, ZmpA, nematocidal protein AidA, and an Flp pilus protein (BCAL1529) also showed reduced abundance in the mutant. Bnr1 is identified as a hypothetical protein (Winsor et al., 2008) and 19% of the proteins showing a reduced abundance in the ΔBCAS0292 mutant were also classed as uncharacterized proteins/hypothetical proteins, including BCAL3228 and BCAS0636 (20-and 12.8-fold, respectively). All five PixA proteins in the sequenced J2315 genome (Winsor et al., 2008), showed reduced abundance (up to 17-fold) suggesting a coregulation or codependence of these uncharacterized proteins. Two of these PixA proteins (BCAM1412 and BCAM1414) are considered orthologs of AidA. Membrane proteins with roles in adhesion and possible roles in structure/transport represented 4% of the total number of proteins reduced in abundance in the mutant strain (Table 2). Among these were proteins associated with roles in adhesion that included putative lipoproteins (BCAM0843 and BCAM0944) and a putative membrane protein (BCAM1261). In addition, two membrane proteins were undetectable in the mutant strain, a lipoprotein (BCAL2166), and a putative membrane protein (BCAM0264) ( Table A3).

| ΔBCAS0292 mutant shows gross morphological changes
Given the vast number of proteins with altered abundance in the mutant strain, we compared the morphology of the mutant with the WT and complement strains. Unsurprisingly, the ΔBCAS0292 mutant cells had a grossly altered morphology, with a highly irregular cell surface and complete loss of their rod-like shape (Figure 4b). This was reversed in the complement suggesting that Bnr1 has a direct or indirect role in maintaining cell structure which may contribute to the reduced adherence of this strain to lung epithelial cells.

| ΔBCAS0292 mutant has limited growth at 6% oxygen
Given that several proteins encoded on the lxa locus showed reduced abundance in the ΔBCAS0292 mutant, we compared the growth of WT and ΔBCAS0292 mutant in 6% oxygen in a controlled hypoxia chamber. The growth (measured as CFU/ml) of the mutant strain was severely reduced under hypoxic conditions (p = 0.0138; Figure 4c). In contrast, growth recovered to WT levels in the complemented strain. Cells were cultured in a hypoxia chamber at 6% oxygen for up to 48 h. Growth is plotted as OD600 nm. Survival was determined by serially diluting aliquots of cultures that had been incubated at 6% oxygen in a controlled hypoxia chamber, plated at time intervals, and cultured for 48 h in a normoxic environment. WT, wild-type His-tagged Bnr1 was purified by SEC to remove any contaminating proteins or aggregates (Figure 5a). Circular dichroism analysis of rBnr1 showed a peak minimum at 215 nm indicating it has a β-sheet secondary structure (Figure 5b). The purified protein was highly stable with a melting temperature (T m ) of 49°C, (Figure 5c). LS indicated the oligomeric state of the Bnr1 protein while Rayleigh scattering showed it has a dimeric organization, under reducing and oxidizing conditions, with a molecular weight of 39.4 ± 0.5 kDa ( Figure 5d). Despite several attempts, we were unable to obtain crystals of Bnr1, therefore molecular modeling was used to predict the three-dimensional structure of the protein using consensus sequence alignment with PixA protein BCAM1413 from B. cenocepacia as a template (PDB code 4lzk, E-value: 3.5e −55 ; Nocek et al., 2013).
Consistent with CD analysis (Figure 5b), the Bnr1 homology model presents a structure with mostly β-folds, and only one small α helix in the N-terminal region of the protein (residues 13-21; Figure 6a).
Also, in agreement with LS studies (Figure 5d), the molecule has a dimeric organization, where two jelly-roll-like monomers are compactly held together (Figure 6a). Electrostatic potential calculations show that the entire surface of the protein has a strong negative charge distribution (Figure 6b), a finding which excludes that Bnr1 regulatory function occurs through interactions with DNA or RNA.
Similar electrostatic surface distributions are a distinctive feature of DNA mimicry (Wang et al., 2014).

| Bnr1 may function as a DNA mimic protein
Given that Bnr1 had an overall negative surface charge, we hypothesized that it bound to DNA-binding proteins to mediate its global regulatory activity. To corroborate this hypothesis we examined whether Bnr1 alters DNA supercoiling, an activity associated with global changes in bacterial gene regulation (Scanlan et al., 2017;Shortt et al., 2016)  Three proteins meeting our stringent selection criteria were reproducibly identified in the prey:bait lanes which were enriched or absent in the two control lanes, (Figure 7a,b; Table S7 at https://doi. org/10.5281/zenodo.5837496). There was a consistent band just below the FLAG-Bnr1 band, which was identified as HctB, a histone H1-like protein that is localized at the outer membrane (Winsor et al., 2008). Another putative histone-like protein was also identified (BCAM1012). Interestingly, the protein translocase subunit, SecA was also identified.

| DISCUSSION
The adaptation of opportunistic pathogens to enhance their fitness in the host and enable their successful colonization requires substantial rearrangement of gene expression. Understanding the processes of bacterial regulation in pathogens that adapt to chronic infection is crucial, particularly in those that are virtually impossible to eradicate.
Our initial studies on the B. cenocepacia K56-2 ΔBCAS0292 mutant  (Mhatre et al., 2020). Given that the ΔBCAS0292 mutant shares these phenotypes and uniquely expressed RpfR, it is likely that the Bnr1 protein mediates some of its pleiotropic effects via RpfR suppression.
Purified rBnr1 protein forms a dimer that is consistent with many functionally important regulatory proteins, such as transcription factors (Ispolatov et al., 2005;Zhanhua et al., 2005). The unexpected finding that Bnr1 has a highly negatively charged surface indicated that Bnr1 is likely to bind to the nucleic acid-binding protein(s) and conferring a more invasive phenotype, while SPI2 genes upregulated by DNA relaxation are essential for intramacrophage survival (Croinin et al., 2006;Dorman & Corcoran, 2009). Alterations in C. jejuni supercoiling were associated with differences in pathogenesis and an altered secretome (Scanlan et al., 2017). Differences in DNA supercoiling between the ΔBCAS0292 mutant and WT at stationary phase combined with enhanced ciprofloxacin susceptibility reinforced the view that Bnr1 can alter DNA topology, suggesting this is the mechanism by which it wields its regulatory effect.
DNA mimic proteins generally have a molecular weight <25 kDa, pI values <5.0 and are characterized by a negative surface charge distribution, allowing them to mimic the phosphate backbone of DNA (Wang et al., 2019), and Bnr1 matches each of these criteria. Very few (∼20) have been discovered to date because consistent with Bnr1, they have unique sequences and structures and are hard to predict using bioinformatic tools. Half of those identified are bacterial and are involved in stress responses (Wang et al., 2014). Two were identified in Neisseria spp., namely DMP12 and DMP19 (Wang et al., 2012(Wang et al., , 2013. Like Bnr1, DMP19 forms a homodimer, has a variety of potential binding partners, including a nitrogen response transcription factor (Huang et al., 2017;Wang et al., 2012). DMP12 maintains the stability of unbound HU-binding regions (Wang et al., 2014(Wang et al., , 2013. A DNA mimic in Mycobacterium tuberculosis, MfpA, protects the DNA binding site of gyrase from fluoroquinolones (Hegde et al., 2005). Pull-down analyses to identify possible binding partners of Bnr1 have identified HctB a histone H1-like protein and a putative histone-like protein, BCAM1012. A Blast search found no significant similarity between these two proteins (Altschul et al., 1997). HctB is particularly interesting as the hctB gene-encoded Hc2 protein is associated with the transformation of Chlamydia trachomatis into the metabolically inactive elementary body in the late stage of its developmental cycle (Barry et al., 1992;Hackstadt et al., 1991). Histone H1-like proteins, Hc1 and Hc2 are also considered to act as global transcriptional regulators in the transformation of vegetative reticulate bodies into infectious elementary bodies of Chlamydophila pneumoniae (Murata et al., 2007). The chlamydial hctB gene caused chromatin condensation and downregulation of transcription when expressed in E. coli (Brickman et al., 1993). A histone H1-like protein with the same H5 signature has also been associated with virulence in Leishmania (Papageorgiou & Soteriadou, 2002 (Shinoy et al., 2013). Future work will involve cloning these proteins to confirm their direct interaction and to elucidate this intriguing mechanism of regulation in more detail.
Overall, we propose that Bnr1, which is dramatically upregulated in response to hypoxic stress interacts with DNA-binding protein, HctB, altering DNA topology and influencing the expression of genes involved in virulence, host cell attachment, antibiotic susceptibility, and biofilm formation (Figure 7). Thirteen of the lxa-encoded proteins with reduced abundance in the ΔBCAS0292 mutant were also identified in chronic infection isolates suggesting a common adaptive stress response in the hypoxic CF lung ( cenocepacia in the hypoxic environment of the CF lung. We propose a mechanism whereby Bnr1 regulates the expression of multiple groups of genes which facilitate its adaptation to stress (Figure 7c). This would allow an environmental pathogen such as Burkholderia which is exquisitely successful at thriving in a wide variety of niches from the rhizosphere to the CF lung and disinfectants. The finding that over half of the proteins previously shown to be upregulated in chronic infection isolates (Cullen et al., 2018) are among those of reduced abundance in the mutant suggests that Bnr1 may also play a role in the adaptation to the hypoxic CF lung. Future work will focus on elucidating the mechanism by which Bnr1 and its binding partners mediate this effect. (c) Schematic of the proposed mechanism of regulation by Bnr1 in response to stress, low oxygen, and stationary phase. Previous work has shown that BCAS0292/Bnr1 is elevated in response to stress, low oxygen, and stationary phase and elevated in late isolates from chronically colonized patients. The present work shows that the absence of Bnr1 reduces the abundance of ≥500 proteins, particularly those including those involved in the stress response, metabolic processes, transcription cell signaling, and virulence. In addition, we have shown that the presence of Bnr1 alters DNA topology, providing a mechanism by which it globally regulates protein expression in response to stress

CONFLICT OF INTERESTS
None declared.

DATA AVAILABILITY STATEMENT
The mass spectrometry proteomics data have been deposited to the