Proteomic analysis of the Pseudomonas aeruginosa iron starvation response reveals PrrF sRNA-dependent regulation of amino acid metabolism, iron-sulfur cluster biogenesis, motility, and zinc homeostasis

Iron is a critical nutrient for most microbial pathogens, and the innate immune system exploits this requirement by sequestering iron and other metals through a process termed nutritional immunity. The opportunistic pathogen Pseudomonas aeruginosa provides a model system for understanding the microbial response to host iron depletion, as this organism exhibits a high requirement for iron as well as an exquisite ability to overcome iron deprivation during infection. Hallmarks of P. aeruginosa’s iron starvation response include the induction of multiple high affinity iron acquisition systems and an “iron sparing response” that is post-transcriptionally mediated by the PrrF small regulatory RNAs (sRNAs). Here, we used liquid chromatography-tandem mass spectrometry to conduct label-free proteomics of the P. aeruginosa iron starvation response, revealing several iron-regulated processes that have not been previously described. Iron starvation induced multiple proteins involved in branched chain and aromatic amino acid catabolism, providing the capacity for iron-independent entry of carbons into the TCA cycle. Proteins involved in sulfur assimilation and cysteine biosynthesis were reduced upon iron starvation, while proteins involved in iron-sulfur cluster biogenesis were paradoxically increased, highlighting the central role of iron in P. aeruginosa metabolism. Iron starvation also resulted in changes in the expression of several zinc-responsive proteins, as well as the first experimental evidence for increased levels of twitching motility proteins upon iron starvation. Subsequent proteomics analyses demonstrated that the PrrF sRNAs were required for iron regulation of many of these newly-identified proteins, and we identified PrrF complementarity with mRNAs encoding key iron-regulated proteins involved in amino acid metabolism, iron-sulfur biogenesis, and zinc homeostasis. Combined, these results provide the most comprehensive view of the P. aeruginosa iron starvation response to date and outline novel roles for the PrrF sRNAs in the P. aeruginosa iron sparing response and pathogenesis. AUTHOR SUMMARY Iron is central for the metabolism of almost all microbial pathogens, and as such this element is sequestered by the host innate immune system to restrict microbial growth. Defining the response of microbial pathogens to iron starvation is therefore critical for understanding how pathogens colonize and propagate within the host. The opportunistic pathogen Pseudomonas aeruginosa, which causes significant morbidity and mortality in compromised individuals, provides an excellent model for studying this response due to its high requirement for iron yet well-documented ability to overcome iron starvation. Here we used label-free proteomics to investigate the P. aeruginosa iron starvation response, revealing a broad landscape of metabolic and metal homeostasis changes that have not previously been described. We further provide evidence that many of these processes are regulated through the iron responsive PrrF small regulatory RNAs, which are integral to P. aeruginosa iron homeostasis and virulence. These results demonstrate the power of proteomics for defining stress responses of microbial pathogens, and they provide the most comprehensive analysis to date of the P. aeruginosa iron starvation response.


23
Iron is a critical nutrient for most microbial pathogens, and the innate immune system 24 exploits this requirement by sequestering iron and other metals through a process termed 25 nutritional immunity. The opportunistic pathogen Pseudomonas aeruginosa provides a model 26 system for understanding the microbial response to host iron depletion, as this organism 27 exhibits a high requirement for iron as well as an exquisite ability to overcome iron deprivation 28 during infection. Hallmarks of P. aeruginosa's iron starvation response include the induction of 29 multiple high affinity iron acquisition systems and an "iron sparing response" that is post-30 transcriptionally mediated by the PrrF small regulatory RNAs (sRNAs). Here, we used liquid 31 chromatography-tandem mass spectrometry to conduct label-free proteomics of the P.

47
Iron is central for the metabolism of almost all microbial pathogens, and as such this 48 element is sequestered by the host innate immune system to restrict microbial growth. Defining 49 the response of microbial pathogens to iron starvation is therefore critical for understanding how 50 pathogens colonize and propagate within the host. The opportunistic pathogen Pseudomonas 51 aeruginosa, which causes significant morbidity and mortality in compromised individuals, 52 provides an excellent model for studying this response due to its high requirement for iron yet 53 well-documented ability to overcome iron starvation. Here we used label-free proteomics to 54 investigate the P. aeruginosa iron starvation response, revealing a broad landscape of 55 metabolic and metal homeostasis changes that have not previously been described. We further 56 provide evidence that many of these processes are regulated through the iron responsive PrrF 57 small regulatory RNAs, which are integral to P. aeruginosa iron homeostasis and virulence. Iron is an essential micronutrient for nearly all forms of life, and presents a central 64 paradigm for nutritional immunity, whereby the host sequesters iron from invading microbial 65 pathogens. (1, 2). In turn, pathogens express a variety of high affinity iron acquisition systems to 66 scavenge host iron (3, 4). In aerobic environments, iron poses the potential for toxicity through 67 the production of reactive oxygen species (5). To balance the essentiality of iron with its 68 potential for toxicity, bacteria must regulate the uptake, use, and storage of this nutrient in 69 response to iron availability. In iron-replete conditions, iron uptake systems are repressed, while 70 proteins involved in iron storage and oxidative stress protection are induced (6). Upon iron 71 starvation, iron uptake systems are upregulated, while non-essential iron containing proteins are 72 repressed in a strategy referred to as the iron-sparing response (7). Due to the central role of 73 iron in numerous metabolic pathways, the iron sparing response is likely to elicit a substantial 74 reorganization of bacterial metabolic networks. However, the full impact of iron starvation on 75 bacterial metabolism, and how these changes impact pathogenesis, remain unclear.

76
As a pathogen with a substantial metabolic requirement for iron, Pseudomonas 77 aeruginosa is an ideal model to elucidate the metabolic adaption to low iron starvation and the 78 subsequent impact of this response on pathogenesis. P. aeruginosa is a Gram-negative 79 opportunistic pathogen of significant concern for hospital-acquired infections, diabetic foot 80 wound infections, and cystic fibrosis lung infections (8-11). To overcome iron limitation in the 81 host, P. aeruginosa induces the expression of numerous exotoxins and proteases that cause 82 tissue damage and may release host cell iron stores (12-14), as well as multiple high affinity iron 83 acquisition systems to scavenge iron from host iron-sequestering proteins (15). In aerobic 84 environments, iron can be acquired via the synthesis and secretion of two distinct siderophores, 85 pyoverdine and pyochelin, which scavenge oxidized ferric iron (Fe 3+ ) from host proteins such as 86 transferrin and lactoferrin (16). In anaerobic environments, reduced ferrous iron (Fe 2+ ) is 87 acquired through the inner membrane associated Feo transport system (17). In addition to these 88 5 labile iron transport systems, P. aeruginosa utilizes two non-redundant heme uptake systems,    Table 1). While some of these pathways were previously 150 known to be iron-responsive, iron regulation of proteins in ketone body metabolism and 151 twitching motility has not been described. A closer examination of this dataset as described 152 below revealed several potential mechanisms for adapting to iron starvation, including changes 153 in proteins for amino acid metabolism, Fe-S cluster biogenesis, and zinc homeostasis.

219
We also observed modest but significant increases in proteins that synthesize L-220 ornithine from glutamate and acetyl-CoA (ArgA, ArgB, ArgC, and AruC) (28) (Fig 2A, and   221 Supplementary Materials, Fig S2A). The idea that the cell is metabolizing glutamate to form L-11 ornithine is attractive, as the DTSB medium used for this study is supplemented with an excess 223 of mono-sodium glutamate as a nitrogen source, and ornithine is needed for pyoverdine 224 production (22). Interestingly, the ArcD L-ornithine/arginine antiporter, and ArgF, which 225 incorporates ornithine into the arginine biosynthesis pathway, were both significantly 226 downregulated in iron-depleted conditions (Fig 2A, and Supplementary Materials, Fig S2A).

260
Interestingly, almost all of the known proteins involved with Fe-S cluster biosynthesis, 261 including IcsR, HscAB, IcsX, and the potential iron donor CyaY (40), were significantly 262 upregulated in iron-depleted conditions (Fig. 4). In contrast, IcsU, which is encoded on the 263 same operon as other Fe-S synthesis proteins, was significantly downregulated in iron-depleted 264 conditions. Differential impacts of iron on proteins encoded by the Isc-hsc-fdx operon may be 265 due to either dis-coordinate post-transcriptional regulatory activities or differences in protein 266 half-life. We also identified a partial Suf-like locus encoding a SufS-like desulfurase, PA3668, 267 and a SufE-like sulfur transport protein, PA3667, both of which were significantly upregulated in 268 low iron conditions (Fig. 4). The upregulation of the SufE-like protein (PA3667) may be able to 269 compensate for the reduced levels of IscU in low iron conditions (Fig. 4B). Overall, these results peptides was not sufficient for differential quantification. Therefore we differentiated the 294 expression of the two operons using PhzB expression: PhzB2 was significantly repressed under 295 low iron conditions, while PhzB1 was not affected by iron (Fig. 5C). Based on these data, we 296 hypothesize that iron specifically regulates proteins encoded by the phz2 operon, though further 297 studies will be necessary to thoroughly test this hypothesis. with previous studies, the twitching diameter for wild type PAO1 was significantly larger under 316 low iron conditions than high iron conditions ( Fig. 6B and Supplementary Materials, Fig. S3).

317
Combined, these data demonstrate that increased twitching in low iron conditions is likely due to 318 increased levels of many components of the Type IVa pili machinery.

320
Proteomics reveals regulatory crosstalk between iron and zinc. As discussed above, iron 321 starvation in P. aeruginosa results in the induction of iron-independent paralogs of certain 322 metabolic enzymes (e.g. FumC and Mqo-mediated reactions in Fig. 1A). In some cases, these 323 enzymes rely on other transition metals, such as zinc and manganese, to support their structure 324 or activity. Thus, one possible strategy to compensate for iron starvation may be to induce 325 15 pathways for the uptake of other transition metal ions. In support of this idea, our data 326 demonstrate that iron limitation results in increased levels of proteins encoded by the cnt operon, 327 which mediate the synthesis, secretion, and uptake of a novel opine metallophore involved in 328 zinc uptake (Fig 7A-B). Interestingly, CntL, the nicotianamine synthase (51) Table 2). Some inconsistencies in iron-regulated pathways between the 367 two experiments were noted, including a lack of iron regulation of the heme acquisition and 368 ketone body synthesis like pathways (compare Table 1 and Table 2). This is likely due to 369 changes in the DTSB media composition between the two experiments, as a result of the 370 variability of TSB batches and differences that can occur during dialysis of the media (see 371 Materials and Methods). However, the results of this later experiment were largely consistent 372 with the first experiment (see the heat maps in Fig. 1-7), demonstrating the overall 373 reproducibility of our reported results. 374 375

376
Notably, this second dataset showed that iron induction of several proteins within newly 377 identified iron-activated pathways was dependent upon the prrF locus. Specifically, iron 378 activation of these proteins was lost in the ∆prrF mutant, and we observed an increase in the 379 levels of these proteins in the ∆prrF mutant compared to wild type when grown under low iron 380 conditions proteins. As expected, PrrF-dependent iron regulation was observed for all previously 381 identified iron-regulated TCA cycle proteins, with the exception of SdhB (Fig. 1), as well as for 382 multiple other proteins encoded by PrrF-regulated mRNAs (Supplementary Materials, Dataset 383 S1). We additionally found that iron regulation of the branched chain amino acid biosynthesis 384 proteins IlvA1 and IlvD was dependent on the prrF locus (Fig. 2), as well as several proteins for 385 sulfur metabolism (SsuF, SsuA, CysH, CysN, and CysD; Fig. 3) and Fe-S cluster biosynthesis 386 proteins (IscU, and IscS; Fig. 4). We further identified novel PrrF regulation of the PhzA 387 phenazine biosynthetic proteins (Fig 5), although it was unclear whether PrrF regulation was 388 occurring through regulation of one or both phzABCDEFG transcripts due to the high identity of 389 the encoded proteins. Lastly, we observed PrrF-dependent iron regulation of the zinc-390 responsive PA5535 and PA5534 proteins (Fig. 7). Combined, these results show an even 391 broader impact of PrrF-dependent regulation on iron and metabolite sparing pathways than was 392 18 previously appreciated, and they highlight a novel role for these sRNAs in mediating metallo-393 regulatory cross-talk in in P. aeruginosa.

395
The PrrF sRNAs share complementarity with mRNAs encoding several iron-and PrrF-396 responsive proteins. To determine whether PrrF is likely to mediate direct regulation of the 397 iron-regulated pathways in this study, we employed CopraRNA (59) to search for 398 complementarity between the PrrF sRNAs and mRNAs encoding PrrF-regulated proteins. Our 399 analysis revealed significant regions of complementarity between the PrrF sRNAs and mRNAs 400 for the TCA cycle enzymes SdhC, PA4333, and PA0794 ( Fig. 1C-E), none of which were 401 identified as sharing complementarity with the PrrF sRNAs in previous studies. We also 402 identified PrrF complementarity with the mRNA encoding IlvD, which catalyzes two distinct 403 reactions in the branched chain amino acid biosynthesis pathway (Fig. 2D). We further 404 identified PrrF complementarity with the mRNA encoding CysD (Fig. 3C), encoding the first step 405 in sulfate assimilation into cysteine, and at the 5' end of iscS (Fig. 4D). The location of 406 complementarity at the 5' end of iscS is consistent with the discoordinate regulation of the Isc 407 proteins observed in our proteomics analyses (Fig. 4A,C-D) and is also consistent with 408 discoordinate regulation of the isc regulon in E. coli by the iron-responsive RyhB sRNA (7).

409
Lastly, we identified PrrF complementarity with the 5' UTR and into the coding region of the 410 PA5535 mRNA (Fig. 7D), encoding a putative metal chaperone that is induced upon zinc 411 starvation (54, 60) and repressed upon iron starvation (Fig. 7B-C). The location of 412 complementarity with the dksA-PA5535-PA5534 operon is consistent with the observed 413 discoordinate regulation of these proteins by iron and PrrF, as DksA2 levels were not affected 414 by iron or prrF deletion (Fig. 7B-C) was eliminated in the ∆prrF mutant, and the levels of these proteins were reduced in the ∆prrF 423 mutant as compared to wild type grown in low iron conditions (Fig 6C). The minor pilin proteins 424 play a key role in initiating twitching motility (61), suggesting the prrF locus is required for this 425 activity. In support of this hypothesis, the ∆prrF mutant exhibited decreased motility when 426 compared to wild type PAO1 ( Fig. 6B and Supplementary Materials, Fig. S3). Moreover, iron 427 regulation of twitching motility was abolished in the ∆prrF mutant. (Fig. 6B and Supplementary   428 Materials, Fig. S3). Thus, our results demonstrate that the prrF locus is critical for iron-429 regulated twitching motility in P. aeruginosa, highlighting a novel role for these sRNAs in P. to NAD + but also helps to generate a proton motive force that can be used to generate ATP (39).

471
Likewise, the cbb 3 -type cytochrome oxidases are the predominant terminal oxidases, as they 472 interact with the cytochrome bc 1 complex, which also contributes to the proton motive force (63).

473
Also for the first time, this study shows that P. aeruginosa upregulates enzymes in

514
The rewiring of metabolic networks that is suggested by our proteomics study also 515 appears to contribute to alterations in virulence factor production. We show that iron starvation transcriptional regulatory activities. This latter hypothesis is supported by our subsequent 539 studies of the ∆prrF mutant, which lacked the ability to mediate iron regulated twitching motility 540 ( Fig. 6), highlighting a novel mechanism for how iron may regulate the switch from planktonic to 541 biofilm growth. Future studies will be needed to determine the mechanism by which PrrF 542 promotes the expression of iron-regulated twitching motility proteins.

587
Spectra were acquired using this ion mobility linked parallel mass spectrometry (UDMS e ) and