The NarX-NarL two-component system is a global regulator of biofilm formation, natural product biosynthesis, and host-associated survival in Burkholderia pseudomallei

In the environment, Burkholderia pseudomallei exists as a saprophyte inhabiting soils and surface waters where denitrification is important for anaerobic respiration. As an opportunistic pathogen, B. pseudomallei transitions from the environment to infect human and animal hosts where respiratory nitrate reduction enables replication in anoxic conditions. We have previously shown that B. pseudomallei responds to nitrate and nitrite in part by inhibiting biofilm formation and altering cyclic di-GMP signaling. Here, we describe the global transcriptomic response to nitrate and nitrite to characterize the nitrosative stress response relative to biofilm inhibition. To better understand the roles of nitrate-sensing in the biofilm inhibitory phenotype of B. pseudomallei, we created in-frame deletions of narX (Bp1026b_I1014) and narL (Bp1026b_I1013), which are adjacent components of the conserved nitrate-sensing two-component system. Through differential expression analysis of RNA-seq data, we observed that key components of the biofilm matrix are downregulated in response to nitrate and nitrite. In addition, several gene loci associated with the stringent response, central metabolism dysregulation, antibiotic tolerance, and pathogenicity determinants were significantly altered in their expression. Some of the most differentially expressed genes were nonribosomal peptide synthases (NRPS) and/or polyketide synthases (PKS) encoding the proteins for the biosynthesis of bactobolin, malleilactone, and syrbactin, in addition to an uncharacterized cryptic NRPS biosynthetic cluster. We also observed reduced expression of ribosomal structural and biogenesis loci, and gene clusters associated with translation and DNA replication, indicating modulation of growth rate and metabolism under nitrosative stress conditions. The differences in expression observed under nitrosative stress were reversed in narX and narL mutants, suggesting that nitrate sensing is an important checkpoint for regulating the diverse metabolic changes occurring in the biofilm inhibitory phenotype. Moreover, in a macrophage model of infection, narX and narL mutants were attenuated in intracellular replication, suggesting that nitrate sensing is important for host survival. Author Summary Burkholderia pseudomallei is a saprophytic bacterium inhabiting soils and surface waters throughout the tropics causing severe disease in humans and animals. Environmental signals such as the accumulation of inorganic ions mediates the biofilm forming capabilities and survival of B. pseudomallei. In particular, nitrate metabolism inhibits B. pseudomallei biofilm formation through complex regulatory cascades that relay environmental cues to intracellular second messengers that modulate bacterial physiology. Nitrates are common environmental contaminants derived from artificial fertilizers and byproducts of animal wastes that can be readily reduced by bacteria capable of denitrification. In B. pseudomallei 1026b, biofilm dynamics are in part regulated by a gene pathway involved in nitrate sensing, metabolism, and transport. This study investigated the role of a two-component nitrate sensing system, NarX-NarL, in regulating gene expression, biofilm formation, and cellular invasion. Global gene expression analyses in the wild type, as compared to Δ narX and Δ narL mutant strains with nitrate or nitrite implicate the NarX-NarL system in the regulation of biofilm components as well as B. pseudomallei host-associated survival. This study characterizes a conserved nitrate sensing system that is important in environmental and host-associated contexts and aims to bridge a gap between these two important B. pseudomallei lifestyles.


206
Removal Kit for bacteria (Illumina) and purification using magnetic beads (AMPure). RNA-seq 207 libraries of cDNA were generated using ScriptSeq™ Complete v2 RNA-seq Library Preparation 208 Kit (Illumina) and purified using Monarch DNA cleanup kit (New England Biolabs). Unique 209 barcodes were added to each sample library using ScriptSeq™ Index PCR Primers (Illumina).

210
Libraries were analyzed on a Tapestation using HS D1000 tapes and reagents (Agilent) to TopHat was used for transcriptome assembly. HTseq-count (version 0.11.0) was used to count 225 accepted hits before the DEseq2 (version 1.20.0) (41) package was employed in R (version 3.6.1) 226 for comprehensive differential expression analysis. Raw read count coverage values were used 227 to compare the differential gene expression between temperature treatments, mutants, and 228 untreated controls. Using a negative binomial distribution to estimate variance and a Bayesian 229 approach for variance shrinkage, the DEseq2 package produced logarithmic fold-change values 230 between the conditions tested. Wald tests were used to calculate p-value and the Benjamini-

231
Hochberg multiple testing correction was used to correct for the false discovery rate.

233
Gene expression and quantitative real-time PCR

234
Genomic DNA-depleted RNA samples, isolated in quadruplicate, were pooled and cDNA was 235 synthesized using 1 µg total RNA, using the Transcriptor First Strand cDNA Synthesis kit (Roche).

263
The NarX-NarL two-component system (TCS) has been described extensively in -264 proteobacteria (9), in which it was discovered to regulate nitrate respiration in coordination with 265 an additional NarQ-NarP TCS depending on environmental nitrate availability (45-47). In -266 proteobacteria, Burkholderia and Ralstonia spp. encode only the NarX-NarL system, a feature 267 shared with Pseudomonadaceae in the -proteobacteria clade but not Neisseriaceae (also -268 proteobacteria) (9). The NarX-NarL TCS is often part of a larger regulon including the dissimilatory 269 nitrate reductase NarGHJI and the transporters/permeases NarK-1 and NarK-2, as is the case in 270 B. pseudomallei 1026b (Fig 1A). Analysis of this system via the SMART algorithm revealed a 271 predicted HAMP domain in NarX and a predicted REC domain in NarL (Fig 1B) 1C) and NarL receiver domain ( Fig 1D)  statically in LB supplemented with either 10 mM NaNO 3 or 10 mM NaNO 2 for 24h (Fig 2A). Biofilm 293 inhibition via nitrate and nitrite supplementation was observed for the wild type while both narX 294 and narL mutants were resistant to nitrate biofilm inhibition but not nitrite. Nitrite inhibited biofilm 295 formation in the nitrate sensing-deficient mutants at similar levels to wild type (Fig 2A). Nitrate-296 mediated biofilm inhibition of narX and narL mutants was restored by complementation of 297 Bp1026b_I1014 (narX) and Bp1026b_I1013 (narL)) with IPTG-induction and was comparable to 298 the wild type ( Fig 2B). Consistent with our previous transposon insertional mutants (5), both 299 components of the NarX-NarL system respond to nitrate by biofilm inhibition, however this system 300 is not similarly affected by nitrite at the concentration tested. These results suggest that the NarX-

301
NarL system has specificity for nitrate sensing in the regulation of biofilm dynamics in B.

304
Nitrite, but not nitrate, suppresses anaerobic biofilm growth in B. pseudomallei

305
To examine the effects of nitrate and nitrite on oxygen-deprived bacterial cells, as 306 commonly found in tissue-associated biofilm (51) or intracellular infections (52), we adapted an anaerobic biofilm model for B. pseudomallei (53). Static biofilm growth was initiated in an oxygen-308 deprived anaerobic jar with the addition of an added carbon source (0.75% glucose) and 309 enhanced via supplemented nitrate, which increased anaerobic biofilm formation starting at 10 310 mM ( Fig 3A). In contrast to the beneficial effect of nitrate, the addition of sodium nitrite had an 311 inhibitory effect on anaerobic biofilm growth ( Fig 3B). Significant biofilm growth defects were 312 observed starting with the addition of 10 mM NaNO 2 , while subsequent concentration increasingly 313 favored growth in NaNO 3 -supplemented media ( Fig 3B). The observed cessation of growth in the 314 nitrite treatment corresponds to a similar phenotype involving mycobacterial growth repression 315 due to endogenous nitrite accumulation (52). Interestingly, the addition of exogenous nitrate for 316 anaerobic swim motility assays showed a robust increase in flagellar motility in response to nitrate 317 sensing, considering that absence of both narX and narL inhibited swimming even with nitrate 318 present ( Fig 3C).These results suggest that B. pseudomallei uses nitrate but not nitrite as an 319 alternative terminal electron acceptor during anaerobic biofilm growth, and that nitrite may inhibit 320 anaerobic growth, suggesting hierarchical control during the shift to anaerobiosis.

321
To analyze relative function of the primary nitrate reductase in B. pseudomallei under both 322 aerobic and anaerobic conditions, we next examined the production of nitrite in culture media 323 using the Griess test. Using Griess reagent, which enables colorimetric quantification of nitrite ion 324 in solution, we measured 1.4 mM NO 2 and 3 mM NO 2 in anaerobic ( Fig 3D)  Because of the observed disparities between nitrite-mediated biofilm inhibition in narX 340 and narL strains (Fig 2A), as well as our previous observations involving decreased cyclic di-

341
GMP production in response to nitrate (5), we next aimed to characterize the global transcriptional  wild-type cultures grown in LB supplemented with or without 10 mM NaNO 2 , showing that the first principal component explains 74% of the variance ( Fig 4D). However, when comparing the 359 transcriptome data to account for nitrate and nitrite effects on wild type, most differentially 360 regulated transcripts are affected similarly by nitrate and nitrite ( Fig S3). Venn diagrams display 361 181 up-regulated genes ( Fig S3A) with fold changes greater than two and 203 down-regulated 362 genes ( Fig S3D) with fold changes less than two and false discovery thresholds below q < 0.01 363 that share expression profiles between nitrate-and nitrite-treated sample groups.   Table) and 21 clusters on 406 chromosome II (S3 Table) as differentially regulated in response to exogenous nitrate. The

413
and I1020 (narK-1) were collectively up-regulated at mean fold change of 9.7 in the nitrate 414 treatment group, and at mean fold change of 10.5 in the nitrite group. Among these loci, the -

415
and -subunits of the dissimilatory nitrate reductase narG-1, and narH-1, respectively, were the 416 most highly and significantly expressed in the nitrate and nitrite groups (S4 Table, S6 Table).

417
Noticeably, only loci in the narXL-narGHJI-narK 2 -narK 1 regulon were significantly regulated in 418 response to either nitrate or nitrate, omitting loci on chromosome II and the duplicate dissimilatory 419 nitrate reductase narGHJI-2. Genes encoding assimilatory nitrate and nitrite reductases, nitrite 420 reductases, nitric oxide, and nitrous oxide reductases were all absent from our expression data.

421
The absence of other nitrate metabolism genes in this analysis suggests that the normoxic   Table) and -5.8 in the nitrite 494 condition (S7 Table). bsaN has been shown to encode a positive transcriptional regulator of the  cluster has yet to be characterized in B. pseudomallei as a classical RND efflux relevant to clinical 523 antibiotics (77), thus this expression may not be associated to increased drug resistance.

524
Interestingly, this putative RND efflux cluster is encoded adjacent to an ABC-type export system

557
Polysaccharide biosynthetic genes. In conjunction with the biofilm inhibitory phenotype 558 resulting from nitrate and nitrite dose responses in wild type (Fig 2A, (5)), our analysis revealed 559 significant downregulation of several gene clusters associated with polysaccharide biosynthesis.

560
The   similarly differentially regulated in both the nitrate and nitrite stress comparisons (Fig 7).

594
Bactobolin is an antibiotic encoded by a 120-Kb DNA element in B. pseudomallei and B.   669 narX and narL mutants were significantly impaired at intracellular survival ( Fig 9A). Two hours 670 after infection, the wild type was recovered at lower titers than either of the two mutant strains 671 (Fig 9B). narX and narL were internalized at 166% and 168% compared to the wild type,

699
GMP levels (5). We further characterized the narX-narL system in conjunction with biofilm 700 inhibition and discovered a complex regulatory system encompassing key elements of the 701 stringent response, secondary metabolism, virulence, and antibiotic tolerance (Fig 10).

702
Our initial observation of a biofilm inhibition phenotype that is comparable in both nitrate 703 and nitrite treatment in oxic conditions is complicated by the fact that nitrite-dependent biofilm 704 inhibition does not require the NarX-NarL system (Fig 2A). This disparity lead us to hypothesize 705 that the NarX-NarL system in B. pseudomallei can discriminate between nitrate and nitrite ligands, suggesting that exogenous nitrite is not regulated by NarX-NarL (Fig S3C, S3F). Future studies 715 are required to characterize the complex regulatory systems that facilitate both nitrate and nitrite 716 signals in B. pseudomallei, although this is beyond the scope of the current study. Sensing of 717 similar N-oxides, nitrate and nitrite, are facilitated by twin two-component systems NarX-NarL and 718 NarQ-NarP in enteric bacteria (101), of which B. pseudomallei only encodes NarX-NarL (9). Given 719 that our previous in silico analyses did not discover a duplication of the NarX-NarL system on 720 chromosome II in B. pseudomallei (5), and given the data presented here, we propose that the 721 NarX-NarL system preferentially binds nitrate as a ligand.

722
The NarX-NarL system regulates biofilm formation in a nitrate-dependent manner, through

771
Nonetheless, the responses of in vitro biofilms to nitrate and nitrite were strikingly similar in our 772 study (Fig 5A, 5B). The transcriptional trends identified by the nitrate stress response are 773 dependent on both components of the NarX-NarL system, as the differential expression patterns 774 of both narX and narL in the nitrate condition were opposite to that of the wild type (Fig 7). nitrate and nitrite, yet are dependent on a functioning NarX-NarL system (Fig 7). In B.

958
We thank Grace Borlee and Ian McMillan for critical review and discussions of this manuscript.

959
We are also grateful to Justin Lee, Marylee Layton, Mark Stenglein, and the MIP NGS Illumina

960
Core for technical assistance with library prep and sequence generation.