The consequences of reciprocally exchanging the genomic sites of Integration Host Factor (IHF) subunit production for subunit stoichiometry and bacterial physiology in Salmonella enterica serovar Typhimurium

Integration host factor (IHF) is a heterodimeric nucleoid-associated protein that plays roles in bacterial nucleoid architecture and genome-wide gene regulation. The ihfA and ihfB genes encode the subunits and are located 350 kilobase pairs apart, in the Right replichore of the Salmonella chromosome. IHF is composed of one IhfA and one IhfB subunit. Despite this 1:1 stoichiometry, mass spectrometry revealed that IhfB is produced in 2-fold excess over IhfA. We re-engineered Salmonella to exchange reciprocally the protein-coding regions of ihfA and ihfB, such that each relocated protein-encoding region was driven by the expression signals of the other’s gene. Mass spectrometry showed that in this ‘rewired’ strain, IhfA is produced in excess over IhfB, correlating with enhanced stability of the hybrid ihfB-ihfA mRNA that was expressed from the ihfB promoter. Nevertheless, the rewired strain grew at a similar rate to the wild type, had identical cell morphology, and was similar in competitive fitness. However, compared to the wild type, it was less motile, had growth-phase-specific reductions in SPI-1 and SPI-2 gene expression and was engulfed at a higher rate by RAW macrophage. Our data show that while exchanging the physical locations of its ihf genes and the rewiring of their regulatory circuitry are well tolerated in Salmonella, genes involved in the production of type 3 secretion systems exhibit dysregulation accompanied by altered phenotypes. IMPACT STATEMENT Integration Host Factor (IHF) is an abundant nucleoid-associated protein that organises DNA architecturally, influencing gene expression globally in Salmonella and other bacteria. IHF is composed of two related, non-identical, subunits, produced by genes that are 350 kilobase pairs apart. Each ihf gene has unique expression controls and is embedded in a complex genetic network that supports mRNA translation. Given that the subunits are thought to be required in a 1:1 ratio to form functional IHF, we were surprised by this physical and regulatory separation. We rewired the Salmonella genome so that each subunit was produced using the other’s regulatory signals and gene location. This revealed a high degree of tolerance to the effects of this rewiring. However, we discovered that bacterial motility was disrupted, as was the expression of virulence genes that have been acquired by horizontal gene transfer. Proteomic analysis using mass spectroscopy (MS) showed the extent of the alterations to cell composition. Our MS data also demonstrated that the subunits of IHF are not produced in a 1:1 ratio in either the wild type or the rewired strain. We discuss this finding in terms of the ability of each subunit to stabilise its partner. DATA SUMMARY Whole genome sequence data for strain OrfSwapihfA-ihfB are available from the European Nucleotide Archive with accession number ERS4653309. Data from mass spectrometry analyses are available via ProteomeXchange with identifier PXD027465 (login: reviewer_pxd027465@ebi.ac.uk and password: GVeNIUB2). All supporting data have been provided in the article or through supplementary data files. Repositories Whole genome sequence data for strain OrfSwapihfA-ihfB are available from the European Nucleotide Archive with accession number ERS4653309. Data from mass spectrometry analyses are available via ProteomeXchange with identifier PXD027465 (login: reviewer_pxd027465@ebi.ac.uk and password: GVeNIUB2).


INTRODUCTION
complexes lie at the heart of many of the molecular processes listed above. tabolism: ihfA has its own promoter but is also transcribed with the pheST op-117 eron, encoding the phenylalanine tRNA synthetase (Fig. 1b) [15,57] while 118 ihfB has its own promoter and is co-transcribed with rpsA, the gene encoding 119 the S1 ribosomal protein (Fig. 1b) [15,58,59]. At first glance, this genetic ar-120 rangement seems counterintuitive. Why produce the two IHF subunits from 121 genes that are separated physically on the chromosome and are regulated 122 independently? Is the physical location of each gene significant for the life of 123 the bacterium, or might other arrangements work just as well? Several lines of 124 evidence suggest that gene location on bacterial chromosomes is important 125 [60-66]. For example, gene distance from the origin of chromosome replica-126 tion (oriC) influences gene copy number during rapid growth. This is because 127 several rounds of replication are initiated, producing more copies of oriC-128 proximal genes than there are copies of genes close to the terminus of repli-129 cation, in a phenomenon known as replication-associated gene dosage ef-130 were placed in a 37°C incubator without stacking to ensure equal oxygen ac-262 cess. After 5 h, the diameters of the resulting swarm zones were measured 263 and expressed as the ratio of the WT zone to that of the mutant. To obtain high-quality chromosomal DNA for whole genome sequencing, a 290 basic phenol-chloroform method was used [79]. Two ml of an overnight cul-291 ture were centrifuged at 16000 x g for 1 min to harvest cells and the cell pellet 292 was resuspended in 400 µl of TE buffer pH 8 (100 mM Tris-HCl pH 8.0, 10 293 mM EDTA pH 8.0 (BDH, Poole, England)). 1% SDS and 2 mg/ml proteinase K were added and incubated for 2 h at 37°C to complete lysis. DNA was isolat-295 ed by the addition of 1 volume phenol pH 8. 0 incubator grown in 75 cm 3 tissue-culture flasks. When approximately 80% 335 confluent growth was achieved, cells were split to a fresh flask. Cells within 336 the 9-16 passage number range were used for infections. All media and PBS 337 used for cell culture were pre-warmed to 37°C. To split cells, old DMEM was 338 removed and the monolayer was rinsed with 10 ml of sterile PBS. Fresh 339 DMEM (10 ml) was pipetted into the flask and the monolayer was scraped 340 gently with a cell scraper to dislodge the cells. Scraped cells were centrifuged 341 at 450 x g for 5 min in an Eppendorf 5810R centrifuge and the cell pellet was 342 resuspended in 5 ml DMEM+FBS. One ml of the cell suspension was added 343 to 14 ml of fresh DMEM+FBS in a 75 cm 3 flask, gently rocked to mix and in-344 cubated at 37°C, 5% CO 2 . To seed cells for infection, cells were treated as for 345 splitting. After resuspension in 5 ml DMEM+FBS, viable cells were counted on 346 a haemocytometer using trypan blue exclusion dye. A 24-well tissue culture 347 plate was filled with 500 µl DMEM+FBS. 1.5×10 5 cells were added to each 348 well, gently rocked to mix and incubated at 37°C, 5% CO 2 for 24 h. Label-free quantification (LFQ) intensity values were used to compare the pro-406 tein abundance between the samples (Table S2). These values were obtained 407 by the delayed normalisation algorithm that eliminates differences that arise 408 from separate sample processing and the extraction of maximum peptide ratio 409 information by using only common peptides for the pair-wise sample compari-  The proteins that were differentially expressed at a statistically significant level 421 were filtered out. Hierarchical clustering was performed on these proteins (the 422 log 2 -transformed LFQ values were normalized by Z-score) to distinguish be-sided Student's T-test was performed on these imputed data using the same 426 FDR and S 0 parameters. Manual review was performed on proteins that were 427 found to be differentially expressed after imputation to assess the effect of im-428 putation on statistical significance. This was necessary because among pro-429 teins that were detected in at least two samples in one strain and only in one The btuC gene, encoding the permease of the vitamin B12 transport system, 495 is located 3' to ihfA in the WT and 3' to the 5'-ihfA[UTR]-ihfB[ORF]-3' hybrid 496 SL_ycaI_qPCR_Pr, Table S1) for btuC and ycaI, respectively. Transcription of 502 both downstream genes had increased in the OrfSwap ihfA-ihfB , compared to 503 SL1344 (Fig. S3). With the exception of btuC at the 2-h time-point (when no 504 difference was found) small-but-statistically-significant increases in down-505 stream gene expression were detected. The fold changes in gene expression 506 were well below 1, relative to hemX expression, suggesting that they were un-507 likely to be physiologically important. 508 509 IhfA and IhfB production is unequal in both the SL1344 and the 510 The accurate absolute molar quantities of IHF subunits in SL1344 and the 512 OrfSwap ihfA-ihfB strain proteomes were compared by mass spectrometry (MS; 513 see METHODS) [86]. Data from MS analyses are summarised in Table 2. 514 Samples for protein preparation were taken from LB cultures at the 7-h time-515 point of the growth cycle, corresponding to the early stationary phase of the 516 growth cycle (the point where IHF production is at its maximum). It is also the 517 time point at which ihfA and ihfB gene expression had been measured by RT-518 qPCR (Fig. 2). IhfB to IhfA ratios were determined by dividing the correspond-519 ing iBAQ intensity values. This was done separately for each biological repli-520 cate, with the resulting ratios being averaged and plotted (Fig. 3). The results 521 showed that in SL1344, the IhfB subunit was present at approximately twice 522 the molar quantity of IhfA protein. However, in the OrfSwap ihfA-ihfB strain, the 523 IhfA protein was produced at a molar quantity that exceeded that of the un-524 tagged IhfB subunit, a reversal of the pattern seen in SL1344 (Fig. 3). 525 526

Comparing the proteomes of SL1344 and OrfSwap ihfA-ihfB strains 527
A total of 884 proteins were detected by MS in at least two samples of either 528 strain. The full dataset has been uploaded to ProteomeXchange (identifier 529 PXD027465). There were 213 proteins that were downregulated significantly 530 in the OrfSwap ihfA-ihfB strain, compared to SL1344, and 83 that were signifi-531 cantly upregulated (Table S2). In light of the reduced quantity of IhfB relative 532 to IhfA in the OrfSwap ihfA-ihfB strain, it was interesting to note that the B subunit 533 of IHF's paralogue HU was also present in reduced molar quantities.
Of the 126 proteins significantly downregulated in the OrfSwap ihfA-ihfB strain, 22 535 were involved in translation, 53 were involved in metabolic pathways, 16 pro-536 teins were associated with the cell envelope and transport, 12 proteins were 537 involved in genetic information processing, five in stress responses, seven in 538 motility and chemotaxis, four in pathogenesis, two in the bacterial cytoskele-539 ton, and five have no known function. In addition, 48 proteins were significant-540 ly upregulated in the OrfSwap ihfA-ihfB strain. Of these, 15 proteins were in-541 volved in metabolism, nine were associated with the outer membrane, two 542 were involved in genetic information processing, four in translation, one in mo-543 tility and chemotaxis, one in bacterial cytoskeleton, one in pathogenesis, five 544 in stress resistance and ten proteins with no known function. 545 546 Differential mRNA stability correlates with the relative IHF subunit levels 547 in the wild type and the OrfSwap ihfA-ihfB strain 548 Since swapping the protein-coding regions of ihfA and ihfB had created hybrid 549 mRNA in the OrfSwap ihfA-ihfB strain (Fig. 1) we investigated the possibility that 550 differences in wild type and hybrid ihfA and ihfB mRNA stabilities might ac-551 count for the differences in protein production. Rifampicin treatment [87] was 552 used to arrest transcription and then total RNA was harvested at fixed time 553 intervals. RT-qPCR, using primers specific for ihfA and ihfB mRNA (Table  554 S1), was used to amplify the transcripts of these genes [88]. Messenger RNA 555 half-life (T½) was measured as described by Chen et al. [89]. In SL1344 at 556 3.5 h, the T½ of ihfA mRNA was 2.78 min while that of ihfB mRNA was 3.54 557 min (Fig. 4a). Interestingly, at 7 h, T½ ihfB mRNA was 4.93 min, indicating 558 significantly greater stability than ihfA mRNA, where T½ was 2.19 min (Fig.  559   4b). This difference was consistent with the almost two-fold excess of IhfB 560 protein over IhfA at 7 h (Fig. 3). However, at this time point in the OrfSwap ihfA-561 ihfB strain the relative stabilities of ihfA and ihfB mRNA were found to be simi-562 lar, with ihfA mRNA becoming slightly more stable than ihfB mRNA. At 3.5 h, 563 the T½ of ihfA mRNA was 2.62 min, while the T½ of ihfB mRNA was 2.32 min 564 (Fig. 4c). At 7 h, the T½ of ihfA mRNA was 3.38 min, while the T½ of ihfB 565 mRNA was 2.64 min (Fig. 4d). These OrfSwap ihfA-ihfB strain data showed that 566 the hybrid mRNAs generated when the protein-coding segments of the ihfA and ihfB genes were exchanged, had half-lives that differed from those of the 568 native genes in SL1344. Therefore, relative mRNA stability correlated with the 569 levels of IHF proteins in SL1344 and in the OrfSwap ihfA-ihfB strain. Overall, our 570 data reveal that reciprocally exchanging ihfA and ihfB resulted in IhfA being 571 produced in excess of IhfB, a reversal of the SL1344 production pattern. What 572 were the physiological consequences of this reversal? 573 574 The growth, morphological, and competitive fitness characteristics of 575 strains with repositioned and rewired ihf genes 576 To assess the impact of the reciprocal exchanges of the coding regions of the 577 IHF-encoding genes on the growth characteristics of Salmonella Typhimuri-578 um, the growth profiles of the OrfSwap ihfA-ihfB strain was monitored in liquid 579 medium by optical density at 600 nm (OD 600 ) and compared to SL1344. The 580 growth pattern of the OrfSwap ihfA-ihfB strain was identical to that of the SL1344 581 (Fig. S4a). Growth was also measured by calculating the number of colony-582 forming units in a liquid culture by spreading aliquots onto agar plates at fixed 583 time intervals. Once again, no differences were seen between OrfSwap ihfA-ihfB 584 strain and the WT (Fig. S4b). Similarly, SL1344 and the OrfSwap ihfA-ihfB strain 585 had identical cell morphologies (Fig. S5) and were equal in competitive fitness 586 (Fig. S6). with the known positive role of IHF in motility, the OrfSwap ihfA-ihfB displayed a 596 statistically significant decrease in motility compared to SL1344 (Fig. 5). The 597 motility phenotype of the OrfSwap ihfA-ihfB strain resembled that previously de-system is sensitive to even modest adjustments to IHF subunit production pat-600 terns. 601 602

SPI-1 and SPI-2 gene expression in SL1344 and OrfSwap ihfA-ihfB strains 603
Salmonella pathogenicity islands (SPIs) are horizontally acquired genetic re-  The expression of the reporter fusions derivatives of SL1344 was in agree-621 ment with published data [101]: SPI-1 expression peaked during mid-622 exponential growth phase (Fig. 6a), while SPI-2 gene expression peaked in 623 early stationary phase and plateaued thereafter (Fig. 6b). In the OrfSwap ihfA-624 ihfB strain, expression of both pathogenicity island genes was almost identical 625 to that seen in SL1344 wild type control except for a small, but statistically 626 significant, decrease during late-stationary phase in SPI-1 gene expression 627 (Fig. 6a) and during mid-exponential phase in SPI-2 genes (Fig. 6b). OrfSwap ihfA-ihfB strain internalized by macrophages 1 h post infection were cal-662 culated relative to the infection mix and also relative SL1344. The data ob-than SL1344 (Fig. 6c). This enhanced phagocytosis suggested that the cell 665 membrane composition of the OrfSwap ihfA-ihfB is different from that of SL1344. 666 This may have allowed more efficient complement deposition on the microbial 667 cell surface and hence better phagocytosis. points tested (Fig. 6d). 678

679
The lack of any difference in survival rates between the strains may seem to 680 be inconsistent with the observation that the expression levels of the SPI 681 genes in the repositioned strains were slightly lower than in SL1344 (Fig. 6a,  682 6b). However, SPI expression depends on the mechanism of Salmonella up-683 take by the macrophage. Following phagocytosis of Salmonella, both SPI-1 684 and SPI-2 regulon genes are expressed to a much smaller magnitude than 685 when Salmonella invades macrophages using the SPI-1-encoded T3SS [102]. 686 Therefore, an infection model with SPI-1-induced bacteria was tested. No dif-687 ferences in the active SPI-1 mediated entry was found between SL1344 and 688 the OrfSwap ihfA-ihfB strain (Fig. 6e). This suggests that the magnitude of the 689 SPI expression changes observed in Fig. 6a and Fig. 6b was too low to cause 690 detectable alterations in the ability of the bacterium to infect mammalian mac-691 rophages. viable, although they exhibit dysregulation of many systems [44,46]. This 728 suggests that IHF may be redundant, or partially redundant in some molecular 729 processes to which it contributes, or that the contribution made by IHF is dis-730 pensable, at least under some conditions. The related nucleoid-associated 731 protein, HU, also a heterodimeric DNA binding protein, and with a structure 732 that is quite similar to that of IHF, may perform some IHF roles when IHF itself 733 is unavailable. Although it is a non-specific DNA binding protein, HU can sub- represent an attempt by the cell to address this issue. This situation is in 756 marked contrast to that of HU, also an alpha-beta heterodimeric protein, but each of which has distinct cellular functions [114]. Transcriptomic data from 759 hupA, hupB and hupA hupB knockout mutants indicate that each HU subunit 760 has a substantial regulon, but that these overlap incompletely with each other 761 [112]. 762 763 A need to produce IhfB and IhfA at different levels may also explain why the 764 ihfA and ihfB genes are not organised in a single operon. While it is known 765 that operon-encoded gene products can be tuned at the level of translation 766 initiation to achieve a required stoichiometry [115], it may also be advanta-767 geous to achieve the required balance by differential mRNA stability, and/or 768