DksA is a central regulatory switch for stress protection and virulence in Acinetobacter 1

27 Bacterial coordination of stress resistance mechanisms in harsh environments is key to long-term 28 survival and evolutionary success. In many Gram-negative pathogens, both general- and specific- 29 stress response are controlled by alternative sigma factors such as RpoS. The critically important 30 pathogen Acinetobacter baumannii is notoriously recalcitrant to external stressors, yet it lacks RpoS, 31 so the molecular control of its resilience remains unclear. Here, we used transposon insertion 32 sequencing to characterize the molecular responses of Acinetobacter baumannii to two biologically- 33 important metals stressors, zinc and copper, and discovered that the transcriptional regulator DksA 34 acts as a major regulatory stress-protection switch. We mapped the highly pleiotropic nature of DksA 35 using transcriptomics and phenomics and found that it controls ribosomal protein expression, 36 metabolism of gluconeogenic substrates and survival in stresses that cause oxidative damage. A. 37 baumannii strains lacking DksA were no longer virulent in both murine and Galleria mellonella in 38 vivo models. In vitro, DksA mutants exhibited increased sensitivity to human serum and antibiotics 39 yet promoted biofilm and capsule formation. Our study provides detailed insight into the unique role 40 that DksA plays in stress protection and virulence for A. baumannii and lays the groundwork for 41 understanding of RpoS-independent regulatory general stress response. 42 43 44 45

Only two potential global regulators were identified that showed opposite effects in copper and zinc 168 stresses: the two-component system gacS/A and transcriptional regulator dksA. While gacS/A has 169 been studied extensively in A. baumannii and is known to be a dynamic coordinator of tolerance to 170 stress, virulence, motility and antibiotic resistance 11, 55 , the role of DksA is largely uncharacterized 171 in A. baumannii. The TraDIS data suggested that DksA may act as a molecular switch in responses 172 to the two similar but distinct metal stress conditions (Fig. 2b). Phenotypic fitness assays of the dksA 173 mutant confirmed that DksA disruption is deleterious to the bacteria under zinc stress (Fig. 2d), 174 whereas it is beneficial under copper stress (Fig. 2e). We further noticed that the DdksA mutant had a 175 comparable growth rate to wild-type but reached stationary phase much earlier than wild-type with a 176 significantly lower growth yield (Fig. 2c) showing the overlap of genes involved in tolerance and sensitivity to copper (purple) or zinc (yellow) 184 stress. Genes represented by grey color are involved in tolerance to both copper and zinc. killed significantly fewer larvae compared to wild-type, which killed all larvae within 3 days post-202 infection (Fig. 3a). These promising results spurred us to investigate the role of DksA in a mammalian 203 host. For this, we intranasally challenged BALB/c mice with A. baumannii strain AB5075_UW or its 204 DdksA mutant derivative and after 24 h the mouse was sacrificed and organs were removed and 205 bacterial load counted. Strikingly, DdksA mutants could not be recovered from the blood of any mice 206 (<10 2 cells/mL), compared to 2.5x10 6 cells/mL for wild-type (Fig. 3b). For all tissues, the dksA 207 mutant could still colonize, but not as well as the wild-type Fig. 3c-g), except for liver (Fig. 3h). 208 Recovery of the DdksA mutant from the respiratory tract (nose, bronchoalveolar and lung tissue), was 209 at least 2 orders of magnitude lower than that seen for wild-type cells (Fig. 3c, intranasally challenged with 2 × 10 8 CFU and colonization was examined 24 h post-challenge. 218 Growth and respiration in presence of 50% human serum in LB (i), box and whiskers plots (min to 219 max with all data points) showing estimates of crystal violet based biofilm (j) and density gradient 220 qualitative estimation of capsule (k). See methods for detail. For each panel, the data represent the 221 mean of at least two biological triplicates (±SEM). Statistical analyses were performed using a one-222 way ANOVA; * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, and ns = not significant. 223 224 To understand the differences in the observed lack of ability of the dksA mutant to survive in the 225 blood compared to other issues, we performed in vitro virulence assays on both A. baumannii strains 226 (ATCC 17978 and AB5075_UW) and their dksA mutants. First, we tested the mutant's ability to 227 propagate in human serum, which we found was greatly reduced for both DdksA mutant strains (Fig.  228   3i). Next, we tested the mutant's ability to form biofilm and capsule, and found that it was increased 229 compared to wild-type ( Fig. 3j and 3k). Taken together, these data show that DksA is needed for 230 serum resistance and ultimately to infect the bloodstream but seems to repress other virulence 231 determinants, like biofilm and capsule formation. We speculate that the increase in biofilm density 232 resulting from dksA loss is what allows this mutant to still partially colonize tissue. 233 234

DksA acts as a global transcriptional regulator in A. baumannii 235
To identify the molecular mechanism underlying the divergent role of DksA in stress protection and 236 virulence in A. baumannii, we conducted RNA-sequencing (RNAseq) on the ATCC 17978 DdksA 237 mutant and wild-type with and without a shock treatment with copper or zinc. We used the same 238 concentration of copper and zinc in both TraDIS and transcriptomic assays ( Fig S1). For DdksA 239 compared to wild-type without treatment, differential expression of 12.1% (461) of the total genes in 240 the ATCC 17978 genome was observed (using a cut-off of log2FC >1.5 change and Padj<0.05, 241 Supplementary Table 2). Under copper and zinc stress, this increased so that the expression of ~1/5 242 of all genes (20.0% and 18.4% respectively) in DdksA were significantly altered, compared to treated 243 wild-type. 244

245
To obtain a functional overview of the genes with altered expression in the dksA mutant, we used the 246 "Pathway Omics Dashboards Tool" in the MetaCyc database, based on gene ontology 59 . The cellular 247 processes of translation, respiration, ATP synthesis, amino acid synthesis, aromatic compound 248 degradation, co-factor synthesis, nucleoside and nucleotide synthesis and oxidative stress protection 249 were amongst the most highly impacted, suggesting a crucial role of DksA regulation in both stress 250 protection and metabolism ( Supplementary Fig. 4). Individual genes and operons likely to be 251 switched on or off under stress included genes known to be responsible for A. baumannii metal efflux 252 and biofilm formation (e.g. csuA/BABCDE) (Supplementary Table 2). We found that the 253 csuA/BABCDE operon was significantly upregulated (3.8 to 7.1-fold) in the DdksA strain. The csu 254 operon encodes a pilus synthesis and assembly system required for initial bacterial attachment and 255 biofilm formation 60, 61 , consistent with our observation of higher biofilm formation in the DdksA 256 mutant in both ATCC 17968 and AB5075_UW backgrounds (Fig. 3j). The transcription of rRNA and r-proteins is highly correlated with the cellular concentration of 292 initiating nucleotide triphosphates, ATP and GTP 64-66 . Therefore the divergent regulation of r-protein 293 e genes under copper and zinc stresses by DksA could be due to differences in the cellular energy status 294 under these two stresses. Most microorganisms use a branched electron transport chain composed of 295 NADH-quinone oxidoreductases and quinol oxidases to efficiently couple electron exchange for ATP 296 production by the F1F0 ATPase during aerobic respiration 67, 68 . In A. baumannii, enzymes necessary 297 for ATP synthesis are encoded by the atpIBEFHAGDC operon whereas NADH:quinone 298 oxidoreductase and cytochrome bd-I ubiquinol oxidase subunits are encoded by 299 nuoABCEFGHIJKLMN and cydAB operons respectively. For the dksA mutant without treatment, the 300 expression of many of the genes (9 out of 24, log2FC>1.5 and Padj<0.05) in these operons were 301 increased (2.4 -6.1-fold), whereas expression remained largely (0/24 genes, log2FC>1.5 and 302 Padj<0.05) unaffected under both copper and zinc stresses ( Fig. 4b and 4c). In wild-type A. baumannii 303 under copper stress, expression of both the atp and nuo/cyd operons genes (22 out 24, log2FC>1.5 304 and Padj<0.05) were decreased up to 7-fold, but were largely unaffected under zinc stress ( Fig. 4b and  305 4c). 306

307
To test whether DksA impacts on respiration, we directly assayed respiration activities in wild-type 308 and DdksA with and without copper and zinc stress using a tetrazolium redox based assay (Fig. 4d). 309 Both wild-type and DdksA exhibited similar levels of respiratory activities under zinc stress, which 310 were also indistinguishable from the untreated controls (Fig. 4d). In contrast, copper stress resulted 311 in a drastic reduction in respiration for wild-type cells. A reduction of respiratory activity was also 312 noted in the DdksA strain under copper stress, but the effect was not as severe as in wild type, 313 suggesting that copper stress inhibits respiration in A. baumannii and DksA plays a role in 314 exacerbating this effect under copper stress. Collectively, these observations indicate that a 315 decoupling of electron exchange in the respiratory system and subsequent reduction of ATP 316 production may contribute to induction of the stringent response under copper stress. 317 318

DksA controls transcription of aromatic compound catabolism pathways 319
A. baumannii are metabolically versatile and can efficiently catabolize a large number of aromatic 320 and aliphatic compounds 69 . In particular, aromatic degradation pathways are known to be important 321 for A. baumannii virulence 11 . Most bacteria use the phenylacetate and β-ketoadipate pathways to 322 metabolize aromatic compounds (Fig. 5a). A variety of aromatic compounds such as catechol and 323 protocatechuate, can be degraded via these two pathways and are widely distributed among soil 324 microorganisms 70, 71 . In the phenylacetate pathway, aromatic compounds are broken down into 325 succinyl-CoA, whereas the β-ketoadipate pathway generates succinyl-CoA and acetyl-CoA before 326 entering into TCA-glyoxylate cycle (Fig. 5a) 71 . 327

328
We noted that the phenylacetate and β-ketoadipate pathways encoded by genes in paa 329 (paaNABCDEFGHK) and pca (pcaIJFBDKCHG) operons respectively were the two most enriched 330 pathways detected in our analysis of transcription in DdksA ( Fig. 5b and 5c) but displayed specific 331 induction conditions. The expression of genes in the paa operon decreased (between 12-330-fold) in 332 DdksA cells in both the presence and absence of copper stress (Fig. 5b) and was decreased up to 8-333 fold under zinc stress. By contrast, when wild-type cells are treated with copper, expression of these 334 genes was increased (5 to 14-fold; Fig. 5b). The effect of copper stress on expression of pca operon 335 in wild-type strain was found to be similar to the paa operon, as it also increased relative to untreated 336 cells (28 to 180-fold; Fig. 5c). When we gather all genes belonging to aromatic compounds together, 337 it is clear that DksA acts as a transcriptional switch for regulating secondary gluconeogenic pathways 338   Previously, it has been proposed that the GacS/GacA two-component system operates as a switch 361 between primary and gluconeogenic secondary metabolites in number of bacteria 73 , and also aliphatic 362 carboxylic acids such as acetate and propionate have been shown to be an environmental cue for the 363 GacS/A system 74, 75 . In A. baumannii gacS is essential for the expression of paa operon 11 . In line 364 with the reduced expression of the paa operon, expression of gacA was decreased 3.7-fold in the 365 DdksA mutant in both the presence and absence of copper. In ATCC 17978 wild-type, expression of 366 gacA remained unaffected in both copper and zinc stress conditions. Based on these observations, we 367 conclude that DksA is required for the GacS/GacA-dependent metabolic switch during stress. 368 369

DksA controls the glyoxylate shunt in A. baumannii 370
Growth on aromatic compounds, acetate, or fatty acids also requires the activation of the glyoxylate 371 shunt in the tricarboxylic acid cycle (TCA) and gluconeogenesis pathways 76 . More importantly, the 372 glyoxylate shunt that bypasses the NADH producing steps is required within the electron transport 373 chain for the production of ATP and plays important roles in oxidative stress, antibiotic resistance 374 and pathogenesis 77-79 . In DdksA cells, two important genes responsible for the glyoxylate shunt, aceA 375 encoding isocitrate lyase and glcB encoding malate synthase were reduced in expression by 18-and 376 5-fold respectively (Fig. 5a, Supplementary Table 2). When treated with zinc, expression of only 377 aceA (18-fold) was decreased in DdksA, whereas this pathway was not affected under copper stress 378 in both wild-type and DdksA strains (Fig. 5a) respiratory curve similar to the wild type (Fig. 5e). As expected, the DdksA mutant showed growth 385 defects in media requiring a functional glyoxylate shunt, such as acetic and ketoglutaric acid (Fig.  386 5e). Similarly, we found that the DdksA mutant had a significant growth defect on a number of 387 aromatic carbon sources requiring the paa and pca operons including phenylalanine and 4-hydroxy 388 benzoic acid (Fig. 5e). Taken together, our transcriptomic and phenotypic data indicate that DksA 389 controls pathways associated with aromatic and aliphatic compounds.    Table 3.  498   499 For routine overnight culturing of A. baumannii strains, a single colony from cation adjusted Mueller 500 Hinton (MH) agar (for AB5075 low switching opaque type was chosen to minimise phase variation) 501 was used to inoculate 5 mL of MH broth medium. 502 503

Construction of transposon mutant library 504
The ATCC 17978 A. baumannii dense transposon library used in this study was constructed using 505 the protocol as previously described 49 . Briefly, transposomes were prepared by using EZ-Tn5 506 transposase (Epicentre Biotechnology) and a custom Tn5 transposon carrying a kanamycin resistance 507 cassette amplified from the plasmid pUT_Km 96 using the primer set as described previously 97 . The 508 transposomes (0.25 µL) were electroporated into 60 µL of freshly prepared electrocompetent cells 509 using a Bio-Rad GenePulser II set to 1.