Reciprocally rewiring and repositioning the Integration Host Factor (IHF) subunit genes in Salmonella enterica serovar Typhimurium: impacts on physiology and virulence

The 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 kbp apart, in the Right replichore of the Salmonella chromosome. IHF is composed of one IhfA and one IhfB subunit. Despite this 1 : 1 stoichiometry, MS 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. MS 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 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.

It binds to DNA with the consensus sequence WATCAANNNNTTR (where W is A or T, R is purine and N is any base) [18,19]. This protein was discovered originally as a host-encoded factor for integration of bacteriophage lambda into, and excision from, the Escherichia coli chromosome [10,11,16,[20][21][22].
IHF is involved in recombination processes [23][24][25][26][27], transposition [28,29], chromosome replication [30,31], integrative and conjugative element (ICE) transmission [32], plasmid and phage replication [33,34], plasmid transfer [35], CRISPR-Cas acquisition of novel DNA elements [36,37] and the regulation of transcription [38][39][40][41][42][43]. IHF influences the expression of large numbers of genes in several bacterial species [44][45][46][47][48][49][50]. Although IHF is not essential for bacterial life, knockout mutations in ihfA and ihfB result in widespread dysregulation of transcription [44,46,47]. IHF seems to be an architectural component of most systems that it influences, playing this role by bending DNA by up to 180° [2,3,[51][52][53]. These sharp bends promote contact between DNA sequences located upstream and downstream of the IHF-bound site and between any proteins that may be attached to those flanking sequences. Thus, IHF-dependent nucleoprotein complexes lie at the heart of many of the molecular processes listed above. For example, it can promote contact between DNA-bound transcription factors and RNA polymerase, with positive or negative effects on transcription initiation, depending on the distribution of the participating proteins along the bent DNA [54]. The genes ihfA and ihfB are located in the Right replichore of the Salmonella chromosome, 350 kbp apart (Fig. 1a) [55,56]. The ihfB gene is in the Right macrodomain and ihfA is in the Ter macrodomain, one of the major superstructural territories of the chromosome [57]. Each ihf gene is part of a complex operon that includes genes involved in various aspects of bacterial metabolism: ihfA has its own promoter but is also transcribed with the pheST operon, encoding the phenylalanine tRNA synthetase (Fig. 1b) [15,58], while ihfB has its own promoter and is co-transcribed with rpsA, the gene encoding the S1 ribosomal protein (Fig. 1b) [15,59,60]. At first glance, this genetic arrangement seems counterintuitive. Why produce the two IHF subunits from genes that are separated physically on the chromosome and are regulated independently? Is the physical location of each gene significant for the life of the bacterium, or might other arrangements work just as well? Several lines of evidence suggest that gene location on bacterial chromosomes is important [61][62][63][64][65][66][67]. For example, gene distance from the origin of chromosome replication (oriC) influences gene copy number during rapid growth. This is because several rounds of replication are initiated, producing more copies of oriC-proximal genes than there are copies of genes close to the terminus of replication, in a phenomenon known as replication-associated gene dosage effects [61,68,69]. In addition, the regulatory regimes that control the expression of genes, including genes that encode NAPs, are important for bacterial physiology [70]. Many NAPs exert a pervasive influence on the biology of the bacterium, and if their regulatory patterns are disturbed, significant impacts on cell physiology can result [61].
Homodimers of IhfA or IhfB bind in vitro to the same DNA sequences as heterodimeric IHF, but the protein-DNA complexes are much less stable than those formed by the heterodimer [71]. Data from experiments where ihfA or ihfB were overexpressed separately in vivo revealed a mutual dependency of the subunits for stability: ihfA overexpressed in the absence of IhfB results in unstable peptides and ihfB overexpressed in the absence of IhfA results in insoluble peptides [12]. Whether or not this results in the presence of IhfB aggregates is unknown. In contrast, overexpression of ihfA and ihfB in the same cell, either from separate, compatible plasmids or from an ihfBA operon on a single plasmid, results in the production of functional IHF [12]. These data suggest that coordinated expression of ihfA and ihfB is likely to be essential for the production of functional IHF protein, yet these genes are regulated independently and are separated by hundreds of kilobases on the chromosome.
We wished to investigate the significance of ihfA and ihfB gene position, and the consequences for bacterial physiology, of IHF production from relocated and rewired ihf genes. Therefore, we swapped the protein-encoding portion of the ihfA and ihfB genes, and therefore their physical location, but left the regulatory regions of the original genes intact. We monitored IHF protein production by MS. Our data revealed that the IhfB subunit of IHF is produced at a higher level than IhfA in wild-type Salmonella and that this pattern is reversed in the rewired strain. Differences in mRNA stability were found to

Impact Statement
The Integration Host Factor (IHF) is an abundant nucleoid-associated protein that organizes 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 kbp 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 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 stabilize its partner. The ihfA gene is located between the pheT and the btuC genes and is transcribed from the same DNA strand; ihfB is located between the rpsA and ycaI genes and is transcribed from the same DNA strand. Transcription promoters are illustrated using angled arrows and block arrows show directions of transcription, with stem-loop structures indicating the positions of transcription terminators, based on similarities to the E. coli orthologues listed at Ecocyc.org for ihfA (https://ecocyc.org/ECOLI/substring-search?type=NIL&object=ihfA) and ihfB (https://ecocyc.org/gene?orgid=ECOLI&id=EG1044 1#tab=TU). underlie these differences in IHF subunit production. Reciprocal exchange of the protein coding regions of ihfA and ihfB altered bacterial motility, the expression of SPI-1 and SPI-2 encoded virulence genes and Salmonella pathogenicity. We discuss our findings in the context of chromosome organization and IHF biology.

Bacterial strains and culture conditions
The bacterial strains used in this study were derivatives of S. Typhimurium strain SL1344 and their details are listed in Table 1; plasmids are described in Table 2. Bacteriophage P22 HT 105/1 int-201 was used for generalized transduction during strain construction [72]. Phage lysates were filter-sterilized and stored at 4 °C in the dark. Bacterial strains were stored as 35 % glycerol stocks at −80 °C and freshly streaked on agar plates for each biological replicate. A single colony was used to inoculate 4 ml LB broth and this was then incubated for 18 h. This overnight culture was subcultured into fresh 25 ml LB broth normalizing to an OD 600 of 0.003, unless otherwise stated, and grown to the required growth phase. The standard growth conditions for all bacterial strains were 37 °C and 200 r.p.m., unless otherwise stated.  pKD4 Amp R (Carb R ), Kan R [73] pKD46 Amp R (Carb R ), λ Red genes γ, β, exo under the control of an arabinose inducible promoter [73] pCP20 Amp R (Carb R ), Cm R , FLP recombinase-expressing, temperature-sensitive replicon [78] Amp R (Carb R ), ampicillin (carbenicillin) resistance; Cm R , chloramphenicol resistance; Kan R , kanamycin resistance.
To measure growth characteristics of a bacterial culture, an overnight culture was adjusted to an OD 600 of 0.003 in 25 ml of fresh LB broth and grown at the standard conditions for 24 h in the appropriate liquid medium. The optical density of the culture at OD 600 was measured at 1 h intervals for the first 3 h and then every 30 min until 8 h; the last reading was taken at 24 h. Measurements were taken using a Thermo Scientific BioMate 3S spectrophotometer with liquid cultures in plastic cuvettes.
The growth characteristics of bacterial cultures in LB broth were also measured by viable counts. The culture was grown in the same way as for spectrophotometry, and an aliquot was taken at 2, 4, 6, 8 and 24 h, serially diluted and spread on LB agar plates to give between 30 and 300 colonies after overnight incubation at 37 °C. The bacterial colony counts were expressed as colonyforming units per millilitre (cfu ml −1 ).

Lambda Red recombination
The Lambda Red-mediated recombination system was extensively used for strain construction [73], allowing us to circumvent bacterial defences while modifying bacterial genome with linear dsDNA [74][75][76]. Plasmid pKD46 (Table 2) carries the Lambda Red system under the control of an arabinose-inducible promoter. The replication of plasmid pKD46 is temperature-sensitive, allowing it to be eliminated by growth at the non-permissive temperature once the recombination process is completed [77].

Generation of the orfSwap ihfA-ihfB strain
To create the OrfSwap ihfA-ihfB strain, intermediate 2xihfA and 2xihfB strains were first constructed that contained, respectively, two copies of the ihfA gene or two copies of the ihfB gene (Table 1). Positioning ihfA in place of ihfB to construct 2xihfA is schematically shown in (Fig. S1) (available in the online version of this article) to represent the general strategy employed throughout this project. Briefly, a gene to be moved was tagged with an antibiotic resistance gene using plasmid pKD3 (carrying a chloramphenicol resistance cassette) or plasmid pKD4 (carrying a kanamycin resistance cassette) as a template ( Table 2). For 2xihfA construction, a pair of primers (infA.cmR.Pfwd and infA.cmR.Prev; Table S1) was used to amplify a chloramphenicol resistance cassette, producing an amplicon that had a first overhang homologous to a region immediately downstream of the gene of interest (ihfA). All insertions of the antibiotic resistance cassettes and amplicons in the subsequent steps were typically made 3 nt downstream of the stop codon to avoid disrupting the terminator sequence and excluding other regulatory regions. The second overhang was designed to delete 2-4 nt from the target region to make insertion more efficient. PCR was carried out using high-fidelity Phusion DNA Polymerase. The resulting amplicon was purified with the Qiagen PCR purification kit and transformed by electroporation into a strain harbouring pKD46. This strain was grown with arabinose at 30 °C to drive expression of the Lambda Red system encoded on pKD46 (Table 2) [73]. The integrity of successful transformant colonies was confirmed by PCR with 'confirmation' primers and sequenced by Sanger sequencing. Taking 2xihfA as an example, the ihfA::Cm R construct was similarly amplified with primers that had overhangs homologous to regions upstream and downstream of the ihfB ORF ( infB. int. Pfwd and infB. int. Prev; Table S1). The amplicon was purified and transformed into the wild-type (WT) strain harbouring pKD46, and successful transformations were confirmed as described above. Alongside the 2xihfA, a 2xihfB strain was constructed using a similar approach but using a kanamycin resistance cassette for selection. A P22 lysate of 2xihfB was used to transduce the ihfB::kan R into the 2xihfA background to yield the OrfSwap ihfA-ihfB . Antibiotic resistance cassettes were removed by the FLP-mediated site-specific recombination with aid of pCP20 (Table 2) [78]. The resulting strain had its ihfA ORF under the control of the regulatory region of ihfB and the ihfB ORF under the control of the regulatory region of ihfA (Table 1).
IHF plays a role in tyrosine-integrase-mediated site-specific recombination [13,20]. Therefore, attempts to remove the antibiotic resistance cassettes flanked by FRT sites by FLP-integrase-mediated site-specific recombination, in intermediate strains where the IHF heterodimer was not present (2xihfA and 2xihfB), failed. Hence, it should be noted that the final 2× versions of these strains contain antibiotic resistance cassettes, while the cassettes were removed from the final OrfSwap ihfA-ihfB strain to avoid potential fitness costs associated with carrying those cassettes.

Bacterial motility assays
These assays were carried out as follows, with particular attention being paid to the plate incubation times: 0.3 % LB agar was melted in a 100 ml bottle in a Tyndall steamer for 50 min and allowed to cool in a 55 °C water bath for 20 min, and six plates were poured and left to dry near a Bunsen flame for 25 min. One microlitre of bacterial overnight culture was pipetted under the agar surface, with two inocula per plate. Plates were placed in a 37 °C incubator without stacking to ensure equal oxygen access. After 5 h, the diameters of the resulting swarm zones were measured and expressed as the ratio of the WT zone to that of the mutant.

Competitive fitness assays
Flasks of broth were inoculated with the pair of competing bacterial strains in a 1 : 1 ratio. Derivatives of each competitor were constructed that carried a chloramphenicol acetyl transferase (cat) gene cassette within the transcriptionally silent pseudogene SL1483. This cat insertion is known to be neutral in its effects on bacterial fitness [79] and allows the marked strain to be distinguished from its unmarked competitor. Competitions were run, in which WT SL1344 was the marked strain or in which OrfSwap ihfA-ihfB was the marked strain. Strains to be competed were pre-conditioned in separate 25 ml cultures for 24 h without antibiotics. Then, 10 5 cells of each strain were mixed in 25 ml of fresh LB broth and grown as a mixed culture for another 24 h. The number of colony-forming units was determined by plating the mixture on chloramphenicol-containing plates and on plates with no antibiotic at T=0 h and T=24 h. WT SL1344 was competed against its SL1344 derivative OrfSwap ihfA-ihfB SL1483::cat and, as a control, SL1344_SL1483::cat was competed against the OrfSwap ihfA-ihfB strain. Competitive fitness was calculated according to the formula: f.i. = (ln (Nc(24)/Nc(0)))/(ln ((Nwt(24)/Nwt(0))), where Nc(0) and Nc (24) are the initial and final counts of a competitor and Nwt(0) and Nwt (24) are initial and final counts of the WT. Competitor is a strain other than the WT. An f.i.<1 means that the competitor is less fit than the WT, while f.i.>1 indicates the opposite.

DnA isolation for whole genome sequencing
To obtain high-quality chromosomal DNA for whole genome sequencing, a basic phenol-chloroform method was used [80]. Two millilitres of an overnight culture were centrifuged at 16 000 g for 1 min to harvest cells and the cell pellet was resuspended in 400 µl of TE buffer pH 8 (100 mM Tris-HCl pH 8.0, 10 mM EDTA pH 8.0; BDH). Then, 1 % SDS and 2 mg proteinase K ml −1 were added and incubated for 2 h at 37 °C to complete lysis. DNA was isolated by the addition of 1 volume of phenol pH 8.0/ chloroform/isoamyl alcohol (AppliChem) (25 : 24 : 1) to the reaction in a phase lock tube. After thorough mixing, the sample was centrifuged at 16 000 g for 15 min at 4 °C to separate the aqueous phase. The upper aqueous layer containing DNA was collected and the phenol/chloroform extraction was repeated twice more. When approximately 80 % confluent growth was achieved, cells were split to a fresh flask. Cells within the 9-16 passage number range were used for infections. All media and PBS used for cell culture were pre-warmed to 37 °C. To split cells, old DMEM was removed and the monolayer was rinsed with 10 ml of sterile PBS. Fresh DMEM (10 ml) was pipetted into the flask and the monolayer was scraped gently with a cell scraper to dislodge the cells. Scraped cells were centrifuged at 450 g for 5 min in an Eppendorf 5810R centrifuge and the cell pellet was resuspended in 5 ml DMEM +FBS. One millilitre of the cell suspension was added to 14 ml of fresh DMEM +FBS in a 75 cm 3 flask, gently rocked to mix and incubated at 37 °C, 5 % CO 2 . To seed cells for infection, cells were treated as for splitting. After resuspension in 5 ml DMEM +FBS, viable cells were counted on a haemocytometer using trypan blue exclusion dye. A 24-well tissue culture plate was filled with 500 µl DMEM +FBS. Then, 1.5×10 5 cells were added to each well, gently rocked to mix and incubated at 37 °C, 5 % CO 2 for 24 h.

Protein sample preparation for mS
An overnight culture was adjusted to an OD 600 of 0.003 in 25 ml of LB broth and grown to the required growth stage. A volume of cells, equivalent to a cell density of 1 ml of culture at an OD 600 of 2, was harvested by centrifugation at 3220 g for 10 min at 4 °C. The cell pellet was resuspended in 600 µl of 8 M urea and moved into a sonication tube. The sample was sonicated on ice with 10 rounds of sonication pulses at an amplitude of 10 µm for 20 s with 20 s resting periods between. The resulting lysate was transferred into a fresh Eppendorf tube and centrifuged at 1500 g for 15 min at 4 °C to remove cell debris. Then, 400 µl of supernatant was transferred into a fresh tube and centrifuged at 9300 g for 10 min at 4 °C; 200 µl of supernatant was transferred into a fresh tube and kept at −20 °C. The protein concentration was determined by the Bradford assay [82].
To reduce disulphide bridges, 100 µg of protein in 50 µl total volume was treated with 5 mM DTT with shaking at 350 r.p.m. for 10 min at 60 °C. To prevent cysteine bridges re-forming, exposed cysteine residues were alkylated by adding 10 mM of iodoacetamide (IAA) for 30 min in the dark. To quench the reaction, to dilute 8 M urea and to neutralize pH in order to enable trypsinization, 150 µl of 200 mM of ammonium bicarbonate was added. Twenty microlitres of the sample preparation was transferred into a fresh tube, and 2 µg of trypsin (Roche) was added and incubated at 350 r.p.m. for 18 h at 37 °C. Trypsin digestion was stopped by adding 0.1 % trifluoroacetic acid (TFA). Protein samples for MS were purified and concentrated using ZipTip C-18 (Merck).

mS and data analysis
Complex trypsin-digested peptide mix was first separated via a 1 h HPLC gradient using reversed-phase C18 columns on an Dionex Ultimate 3000 UPLC (ThermoFisher Scientific) followed by a run on a Q-Exactive (ThermoFisher Scientific) mass spectrometer located at the Mass Spectrometry Core, Conway Institute, UCD, Ireland.
MS output files in RAW format were loaded into MaxQuant version 1.6.10.43. The RAW files were searched against the Salmonella enterica serovar Typhimurium strain SL1344 complete proteome sequence obtained from UniProt (proteome ID UP000008962).
Carbamidomethylation of cysteine was chosen as a variable modification, while oxidation of methionine and N-terminal acetylation were chosen as fixed modifications. Trypsin digestion was selected with a maximum of two missed cleavages. Label Free Quantification (Fast LFQ) mode was selected in group-specific parameters and iBAQ quantification was added in global parameters.
Other settings were default. The search was performed using the Andromeda peptide search engine integrated into MaxQuant.

Protein ratio determination
Intensity-based absolute quantification values (iBAQ) were used to calculate the ratios of IhfA to IhfB. The iBAQ intensity values are generated by the MaxQuant algorithm that calculates the sum of all peptide peak intensities divided by the number of theoretically observable peptides during tryptic digestion [83]. The protein entries corresponding to IhfA and IhfB were identified in the proteinGroups. txt file. IhfA to IhfB ratios were determined by dividing the corresponding iBAQ intensity values.

Statistical analysis of mS
Label-free quantification (LFQ) intensity values were used to compare the protein abundance between the samples (Table S2). These values were obtained by the delayed normalization algorithm that eliminates differences that arise from separate sample processing and the extraction of maximum peptide ratio information by using only common peptides for the pair-wise sample comparison [84]. Statistical analysis was performed in Perseus (version 1.6.10.50). Potential contaminants, proteins only identified by site and reverse hits were filtered out. Annotations were downloaded from annotations. perseus-framework. org. The LFQ intensity values were log 2 -transformed, and the samples were grouped to account for the experimental design. A protein was considered to be present for the purpose of the analysis if it was detected in at least two samples in any of the sample groups. A two-sided Student's t-test was performed on the data prior to imputation, using a false discovery rate (FDR)=0.01 and S 0 =1. Principal component analysis (PCA) of the dataset was carried out to ensure that the MS samples cluster according to the strain. The proteins that were differentially expressed at a statistically significant level were filtered out. Hierarchical clustering was performed on these proteins (the log 2 -transformed LFQ values were normalized by Z-score) to distinguish between up-and downregulated proteins. For all other proteins, missing LFQ intensity values were replaced by imputation from the normal distribution. A two-sided Student's t-test was performed on these imputed data using the same FDR and S 0 parameters. Manual review was performed on proteins that were found to be differentially expressed after imputation to assess the effect of imputation on statistical significance. This was necessary because among proteins that were detected in at least two samples in one strain and only in one sample in the other strain and where detected LFQ values were similar, imputation frequently resulted in the creation of false conclusions about statistical significance. As a result, statistically significant up-and down-regulated proteins, as well as the unique proteins, were deemed to be differentially expressed.

RESuLTS
Constructing orfSwap ihfA-ihfB , a derivative of S. Typhimurium with the protein coding regions of its ihf genes exchanged reciprocally The ihfA and ihfB genes are located on the positive strand of the S. Typhimurium SL1344 chromosome and are separated by approximately 350 kbp [56,85]. The ihfA and ihfB genes in Salmonella and E. coli are at different locations on the chromosome and possess their own regulatory regions (Fig. 1b) [15,60]. A derivative of the S. Typhimurium WT strain SL1344 was constructed in which the complete protein-coding regions (CDS) of the genes were exchanged reciprocally. Here, the protein-coding region of ihfA and ihfB were brought under the transcriptional control of each other's 5′ and 3′ regulatory regions (Fig. 1a). These CDS exchanges resulted in the production of hybrid mRNA species consisting of the 5′ and 3′ untranslated regions (UTRs) of one gene and an mRNA corresponding to the CDS of the other. In addition, the IhfA protein was now produced from a transcript expressed at the genomic location previously occupied by the WT ihfB gene, and ihfB was transcribed from the erstwhile genomic location of ihfA (Fig. 1). This strain was designated OrfSwap ihfA-ihfB (Fig. 1a, Table 1). The process of strain construction, which relied on the Lambda Red recombination system [73], is described in detail in the Methods. By using whole genome sequencing, the OrfSwap ihfA-ihfB strain ihfA and ihfB genes were identical to the parental strain, but also carried harmless SNPs/rearrangements in the shufflon region. The SL1344 parental strain used in this study has two previously described SNPs, in manX (E95V) and menC (L148L), compared to the serotype reference strain genome [65].

Expression patterns of the rearranged ihfA and ihfB genes
The effects of these gene rearrangements on ihf transcription were tested by RT-PCR at different stages of growth using pairs of gene-specific primers: SL_ihfA_qPCR_Pf and SL_ihfA_qPCR_Pr for ihfA, and SL_2ihfB_qPCR_Pf and SL_2ihfB_qPCR_Prev for ihfB (Table S1). Transcript levels were calculated relative to hemX -a gene whose expression does not change in the WT under the growth conditions used here [85] and is not affected by the loss of IhfA or IhfB, or of both IhfA and IhfB [44]. The strain-to-strain comparisons of ihf gene expression made in the present study are time-dependent in nature.
In WT S. Typhimurium strain SL1344 (SL1344 , Table 1), the pattern of ihfA and ihfB mRNA expression resembled that of IHF protein production in E. coli [17,86,87], peaking at the 7 h time point, corresponding to the early stationary phase of growth (Fig. 2). In the OrfSwap ihfA-ihfB strain, ihfA gene expression dropped at 5 and 7 h compared to SL1344 but increased at 2 h, while ihfB gene expression matched that in the SL1344, except for a decrease at 7 h. Note that in the OrfSwap ihfA-ihfB strain,

Assessing the impacts of the ihfA and ihfB rearrangements on downstream gene expression
Experiments were conducted to investigate the effect of each ihf gene's relocation on the expression of downstream genes. RNA-sequencing (RNA-seq) data show no read-through transcription from ihfA or ihfB into downstream genes in SL1344 [85], indicating that the ihfA and ihfB gene transcription terminators are effective in SL1344.
We investigated gene expression of the genes immediately downstream from the repositioned ihfA and ihfB genes in the OrfSwap ihfA-ihfB strain by RT-qPCR. The btuC gene (Fig. 1b), encoding the permease of the vitamin B12 transport system, is located 3′ to ihfA in the WT and 3′ to the 5′-ihfA[UTR]-ihfB[ORF]−3′ hybrid locus in the OrfSwap ihfA-ihfB strain. The ycaI gene (Fig. 1b), encoding a putative competence-related protein, is located 3′ to ihfB in SL1344 and 3′ to the 5′-ihfB[UTR]-ihfA[ORF]−3′ hybrid locus in the OrfSwap ihfA-ihfB strain. Expression of btuC and ycaI was measured by RT-qPCR using the specific primer pairs (SL_btuC_qPCR_Pf and SL_btuC_qPCR_Pr; SL_ycaI_qPCR_Pf and SL_ycaI_qPCR_Pr, Table S1) for btuC and ycaI, respectively. Transcription of both downstream genes had increased in OrfSwap ihfA-ihfB compared to SL1344 (Fig. S3). With the exception of btuC at the 2 h time point (when no difference was found) small but statistically significant increases in downstream gene expression were detected. These data revealed impacts of ihf gene rewiring on downstream gene transcription, despite the presence of intergenic transcription terminators [85]. The reasons for this were not determined, but may have involved cis-acting influences such as perturbations to local DNA topology that stimulated downstream promoters and/or suppressed the effectiveness of the intergenic transcription terminators.

IhfA and IhfB production is unequal in both the SL1344 and the orfSwap ihfA-ihfB strains
The accurate absolute molar quantities of IHF subunits in SL1344 and the OrfSwap ihfA-ihfB strain proteomes were compared by MS (see Methods) [88]. Data from MS analyses are summarized in Table S2. Samples for protein preparation were taken from LB cultures at the 7 h time point of the growth cycle, corresponding to the early stationary phase of the growth cycle (the point where IHF production is at its maximum). It is also the time point at which ihfA and ihfB gene expression had been measured by RT-qPCR (Fig. 2). IhfB to IhfA ratios were determined by dividing the corresponding iBAQ intensity values. This was done separately for each biological replicate, with the resulting ratios being averaged and plotted (Fig. 3). The results showed that in SL1344, the IhfB subunit was present at approximately twice the molar quantity of IhfA protein. However, in the OrfSwap ihfA-ihfB strain, the IhfA protein was produced at a molar quantity that exceeded that of the untagged IhfB subunit, a reversal of the pattern seen in SL1344 (Fig. 3).

Comparing the proteomes of SL1344 and orfSwap ihfA-ihfB strains
A total of 884 proteins were detected by MS in at least two samples of either strain. The full dataset has been uploaded to ProteomeXchange (identifier PXD027465). There were 213 proteins that were downregulated significantly in the OrfSwap ihfA-ihfB strain, compared to SL1344, and 83 that were significantly upregulated (Table S2). In light of the reduced quantity of IhfB relative to IhfA in the OrfSwap ihfA-ihfB strain, it was interesting to note that HupB, the B subunit of IHF's paralogue HU, was also present in reduced molar quantities. HU is a heterodimeric, non-specific DNA binding protein, whose subunits show amino acid sequence similarities to IHF [1]. The significance of this change to HupB production was not determined.
Of the 126 proteins significantly downregulated in the OrfSwap ihfA-ihfB strain, 22 were involved in translation, 53 were involved in metabolic pathways, 16 were associated with the cell envelope and transport, 12 proteins were involved in genetic information processing, five in stress responses, seven in motility and chemotaxis, four in pathogenesis, two in the bacterial cytoskeleton and five had no known function. In addition, 48 proteins were significantly upregulated in the OrfSwap ihfA-ihfB strain. Of these, 15 proteins were involved in metabolism, nine were associated with the outer membrane, two were involved in genetic information  processing, four in translation, one in motility and chemotaxis, one in bacterial cytoskeleton, one in pathogenesis, five in stress resistance and ten had no known function.

Differential mRnA stability correlates with the relative IHF subunit levels in the WT and the orfSwap ihfA-ihfB strain
Since swapping the protein-coding regions of ihfA and ihfB had created hybrid mRNA in the OrfSwap ihfA-ihfB strain (Fig. 1) we investigated the possibility that differences in WT and hybrid ihfA and ihfB mRNA stabilities might account for the differences in protein production. These might reflect, for example, differences in folding of the native and hybrid mRNAs, leading to differences in their turnover by RNases. Rifampicin treatment [89] was used to arrest transcription and then total RNA was harvested at fixed time intervals. RT-qPCR, using primers specific for ihfA and ihfB mRNA (Table S1), was used to amplify the transcripts of these genes [90]. Messenger RNA half-life (T½) was measured as described by Chen et al. [91]. In SL1344 at 3.5 h, the T½ of ihfA mRNA was 2.78 min while that of ihfB mRNA was 3.54 min (Fig. 4a). Interestingly, at 7 h, T½ ihfB mRNA was 4.93 min, indicating significantly greater stability than ihfA mRNA, where T½ was 2.19 min (Fig. 4b). This difference was consistent with the almost two-fold excess of IhfB protein over IhfA at 7 h (Fig. 3). However, at this time point in the OrfSwap ihfA-ihfB strain the relative stabilities of ihfA and ihfB mRNA were found to be similar, with ihfA mRNA becoming slightly more stable than ihfB mRNA. At 3.5 h, the T½ of ihfA mRNA was 2.62 min, while the T½ of ihfB mRNA was 2.32 min (Fig. 4c). At 7 h, the T½ of ihfA mRNA was 3.38 min, while the T½ of ihfB mRNA was 2.64 min (Fig. 4d). These OrfSwap ihfA-ihfB strain data showed that 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 native genes in SL1344. Therefore, relative mRNA stability correlated with the levels of IHF proteins in SL1344 and in the OrfSwap ihfA-ihfB strain. Overall, our data reveal that reciprocally exchanging ihfA and ihfB resulted in IhfA being produced in excess of IhfB, a reversal of the SL1344 production pattern. What were the physiological consequences of this reversal?

Growth and competitive fitness characteristics of strains with repositioned and rewired ihf genes
To assess the impact of the reciprocal exchanges of the coding regions of the IHF-encoding genes on the growth characteristics of S. Typhimurium, the growth profiles of the OrfSwap ihfA-ihfB strain were monitored in liquid medium based on the optical density at 600 nm (OD 600 ) and compared to SL1344. The growth pattern of the OrfSwap ihfA-ihfB strain was identical to that of SL1344 (Fig.  S4a). Growth was also measured by calculating the number of colony-forming units in a liquid culture by spreading aliquots onto agar plates at fixed time intervals. Again, no differences were seen between OrfSwap ihfA-ihfB and the WT strain (Fig. S4b). Similarly, SL1344 and the OrfSwap ihfA-ihfB strain were equal in competitive fitness (Fig. S5).

motility in SL1344 and orfSwap ihfA-ihfB strain
As a master regulator, IHF influences the expression of multiple genes, including that of the entire motility regulon: IHF is a positive regulator of flagellar gene expression at all stages of flagellum production [44]. The motilities of the OrfSwap ihfA-ihfB strain and SL1344 were compared by measuring the diameters of motility zones after 5 h of incubation at 37 °C on 0.3 % LB agar. Motility differences were recorded as the fold change relative to SL1344. In keeping with the known positive role of IHF in motility, the OrfSwap ihfA-ihfB strain displayed a statistically significant decrease in motility compared to SL1344 (Fig. 5). However, the magnitude of the reduction (10 %) was less than that seen previously for the SL1344 ihfA ihfB double mutant (40 %) [44]. These data indicated that the motility system is sensitive to even modest adjustments to IHF subunit production patterns.
Fusions of the gfp reporter gene to the promoters of prgH and of ssaG, genes that encode needle components of SPI-1 and SPI-2, respectively, were used as proxies of SPI-1 and SPI-2 transcription [85,102], allowing these to be compared in SL1344 and OrfSwap ihfA-ihfB . Details of the strain constructions and genotypes (Table 1) are given in the Methods. GFP-mediated fluorescence at 528 nm was measured every 20 min over 24 h and adjusted to the OD 600 of the culture. The expression of the reporter fusion derivatives of SL1344 was in agreement with published data [103]: SPI-1 expression peaked during the mid-exponential growth phase (Fig. 6a), while SPI-2 gene expression peaked in the early stationary phase and plateaued thereafter (Fig. 6b). In the OrfSwap ihfA-ihfB strain, expression of both pathogenicity island genes was almost identical to that seen in the WT SL1344 control except for a small, but statistically significant, decrease during the late-stationary phase in SPI-1 gene expression (Fig. 6a) and during the mid-exponential phase in SPI-2 genes (Fig. 6b).

Infection and survival of SL1344 and orfSwap ihfA-ihfB strains in murine macrophages
RAW264.7 murine macrophages are a well-studied cell line, frequently used for studying Salmonella infection. Gentamycin protection assays were carried out to assess the performance of SL1344 and OrfSwap ihfA-ihfB strains at different stages of the infection process: efficiency of entry to the macrophage, intracellular survival and replication over a 20 h period post-infection. Survival and replication of Salmonella in macrophages is dependent upon the mechanism of entry [104]. The two main modes of entry utilized by Salmonella are: (1) SPI-1-mediated active invasion and (2) passive, host-mediated phagocytosis. Both modes are important as they occur in Salmonella's natural host environment. Macrophages may encounter highly invasive Salmonella, in which SPI-1 genes are induced, as the bacteria emerge from the intestinal epithelial cell barrier [105,106]. However, Salmonella can also be taken up by macrophages via phagocytosis. This occurs when the microbe escapes its original macrophage replication niche while expressing the SPI-2 T3SS needle on its surface or when Salmonella is disseminated to the circulatory system during systemic infection [107].
To achieve active entry to macrophages via SPI-1, bacteria were grown to the mid-exponential growth phase in LB broth to maximize SPI-1 expression [108]. To ensure passive entry via phagocytosis, bacteria were grown to the late stationary growth phase and complement-opsonized [109]. To promote phagocytosis by RAW264.7 macrophages, these cells were infected with SPI-2-induced and opsonized Salmonella. Both infection entry efficiency and long-term intracellular survival were tested. SL1344 was used as a reference strain in all experiments. Counts of bacterial cells of the OrfSwap ihfA-ihfB strain internalized by macrophages 1 h post-infection were calculated relative to the infection mix and also relative to SL1344. The data obtained showed that the OrfSwap ihfA-ihfB strain was phagocytosed more readily than SL1344 (Fig. 6c). This enhanced phagocytosis suggested that the cell membrane composition of the OrfSwap ihfA-ihfB is different from that of SL1344. This may have allowed more efficient complement deposition on the microbial cell surface and hence better phagocytosis. Salmonella survival and replication inside a macrophage were tested using the same infection model. Colony-forming units within macrophages were counted at further time points and calculated relative to those at 1 h. For both strains analysed, after slight initial growth at 3 h post-infection, a killing event was observed at 8 h post-infection (p.i). At 16 h p.i., Salmonella appeared to adapt and replicate significantly, before undergoing a further decline by 20 h p.i., possibly caused by macrophage cell death [110]. When the OrfSwap ihfA-ihfB strain was compared to the WT, no significant difference were found at any of the time points tested (Fig. 6d).
The lack of any difference in survival rates between the strains may seem to be inconsistent with the observation that the expression levels of the SPI genes in the repositioned strains were slightly lower than in SL1344 (Fig. 6a, b). However, SPI expression depends on the mechanism of Salmonella uptake by the macrophage. Following phagocytosis of Salmonella, both SPI-1 and SPI-2 regulon genes are expressed to a much smaller magnitude than when Salmonella invades macrophages using the SPI-1-encoded T3SS [104]. Therefore, an infection model with SPI-1-induced bacteria was tested. No differences in active SPI-1-mediated entry was found between SL1344 and the OrfSwap ihfA-ihfB strain (Fig. 6e). This suggests that the magnitude of the SPI expression changes observed in Fig. 6a, b was too low to cause detectable alterations in the ability of the bacterium to infect mammalian macrophages.

DISCuSSIon
A gene's location on the bacterial chromosome is known to be an important factor in establishing the appropriate spatiotemporal expression pattern of that gene and its influence on physiology [61-63, 65, 66, 70, 111]. The most influential parameter is the distance of the gene from the origin of chromosome replication, oriC [61,[68][69][70]. In rapidly growing bacteria with multiple rounds of replication underway, genes closest to oriC will be present in more copies than those closer to the terminus. The resulting differences in gene dosage contribute to variations in the output of the gene product at different stages of the growth cycle. As stationary phase approaches and replication ceases, oriC-proximal genes lose their copy number advantage over those closer to the terminus. Even so, moving fis, the gene encoding the FIS NAP, from its normal oriC-proximal position to a location near the terminus had a relatively mild impact on the physiology of the bacterium in S. Typhimurium [61] and in E. coli [66] compared with the strong phenotypic effects seen when fis has both the chromosomal location and the regulatory signals of the terminus-proximal dps gene [61].
In this study, we explored the significance of the chromosomal location of each gene that encodes one of the two subunits of IHF in S. Typhimurium, and the significance of each gene's individual regulatory pattern, for the physiology of the bacterium. To do this, we exchanged, reciprocally, the protein-coding regions of the ihfA and ihfB genes. The exchange placed each gene under the control of the expression signals of the other gene (Fig. 1). Whole genome sequencing confirmed that the only changes made to the chromosome in the OrfSwap ihfA-ihfB strain were those introduced in the CDS exchange. The ihf genes are in a quadrant of the chromosome that is close to the replication terminus, and IHF is normally expressed maximally at the transition from exponential growth to stationary phase [15,17]. In the OrfSwap ihfA-ihfB strain, each sequence has moved to a location that was approximately 350 kbp away from its native position within the Right replichore of the chromosome.
IHF contributes to many vital processes in the bacterial cell, yet it is not an essential protein. Bacteria with deletions of both the ihfA and the ihfB genes are viable, although they exhibit dysregulation of many systems [44,46]. This suggests that IHF may be redundant, or partially redundant in some molecular processes to which it contributes, or that the contribution made by IHF is dispensable, at least under some conditions. The related nucleoid-associated protein, HU, also a heterodimeric DNA binding protein, and with a structure that is quite similar to that of IHF, may perform some IHF roles when IHF itself is unavailable. Although it is a non-specific DNA binding protein, HU can substitute functionally for IHF in bacteriophage lambda integrase-mediated DNA excision, but not integration [112]. This is because HU provides the missing architectural function in the lambda intasome in the absence of IHF [113]. The regulons of IHF and HU in S. Typhimurium show overlapping, but non-identical, memberships [44,114] suggesting partial, rather than complete functional redundancy between the two NAPs. The suggestion that redundancy between IHF and HU is only partial is supported by work showing that E. coli mutants devoid of HU have phenotypes that cannot be complemented by overexpression of IHF [115]. Our proteomic data revealed that, like IhfB, the B subunit of IHF's paralogue HU was also present at a reduced level in the OrfSwap ihfA-ihfB strain. This finding was consistent with transcriptomic data showing that the gene encoding HU's B subunit, hupB, is downregulated in the absence of IhfB at the early stationary phase [44]. Perhaps this adjustment to HU subunit production ameliorated the impact of inverting the relative abundances of IhfA and IhfB in the rewired strain. Whether this involved interactions between HU and IHF subunits was not determined.
The mutual dependency of IhfA and IhfB on one another for the assembly of stable IHF protein indicates that it is important to maintain the relative levels of these proteins within certain limits. IhfA is especially unstable in the absence of IhfB and can form IhfA homodimers on DNA only when present at very high concentrations [71]. Our finding that IhfB is produced in excess over IhfA may represent an attempt by the cell to address this issue. This situation is in marked contrast to that of HU, also an alpha-beta heterodimeric protein, but one that also exists both as stable alpha-homodimers and beta-homodimers, each of which has distinct cellular functions [116]. Transcriptomic data from hupA, hupB and hupA hupB knockout mutants indicate that each HU subunit has a substantial regulon, but that these overlap incompletely with each other [114].
A need to produce IhfB and IhfA at different levels may also explain why the ihfA and ihfB genes are not organized in a single operon. While it is known that operon-encoded gene products can be tuned at the level of translation initiation to achieve a required stoichiometry [117], it may also be advantageous to achieve the required balance by differential mRNA stability, and/or by differential transcription, by expressing the messages from separate transcription units under distinct control regimes. The case of IHF production seems even more sophisticated when we take into account the observation that the ihfA and ihfB genes are embedded in operons that express components of the cell's translational machinery. It has been proposed previously that this might account for the presence of ihfA (himA) in an operon with pheST [118]. In this way, a rise or fall in demand for translation capacity (a proxy for metabolic flux) can feed back onto a regulator (IHF) that modulates many key cellular processes. A connection between IHF production and translation capacity is consistent with the observation that transcription of both ihfA and ihfB responds to (p)ppGpp, a signal that indicates an accumulation of uncharged tRNA molecules, and hence over-capacity in the translation apparatus [15]. The ihfA promoter lies within the pheT gene and is autoregulated by IHF, as is ihfB transcription [15]. The pheST-ihfA operon is controlled by transcription attenuation in response to phenylalanine and by the SOS response, indicating sensitivity to changes in translational capacity and to processes involved in DNA damage repair. Certainly, genes encoding components of the DNA metabolism and the translation machinery are well represented in the Salmonella IHF transcriptome [44] and their products in the proteome (Table S2). Most translation-associated proteins were present in reduced amounts in the OrfSwap ihfA-ihfB strain, mimicking the impact of loss of IhfB on the transcriptome at early stationary phase [44]. The majority of proteins with an altered production level in the OrfSwap ihfA-ihfB strain were present in lower amounts than in the WT SL1344 strain. The reduction in translation-associated proteins may reflect an attempt to keep overall protein production within acceptable limits, minimizing the number and severity of impacts on phenotype.
Despite reversal of the relative levels of IhfA and IhfB in the OrfSwap ihfA-ihfB strain, our data show that giving each ihf gene the transcription control region and the chromosome location of the other had, at best, subtle effects on Salmonella physiology. For example, the reduced expression of the rod-shape-determining cytoskeletal protein MreB [119] in the stationary phase correlated