Distinct ecological fitness factors coordinated by a conserved Escherichia coli regulator during systemic bloodstream infection

Significance Bacterial pathogens often acquire virulence genes that drive infection. However, preexisting genes related to metabolism can also be tailored to enhance pathogen fitness within the host by adapting their transcriptional control. Here, we describe a unique mechanism by which uropathogenic Escherichia coli (UPEC), the most common cause of urinary tract and bloodstream infections, regulates its pathogenic capability in this way. We found that the ancestral transcriptional regulator YhaJ coordinately controls the genetically unlinked processes of fimbrial adhesion and tryptophan biosynthesis to maximize UPEC survival during infection of the bloodstream. These findings reveal insights into the underlying basis of UPEC pathogenesis and shed light onto how repurposing preexisting genes and regulators can enhance the efficacy of pathogens during systemic infection.


Bacterial growth conditions
The strain CFT073 was referred to as UPEC throughout this study. All strains, derivatives and plasmids used are detailed in Tables S1 and S2. Overnight bacterial cultures of a single colony grown in 5 ml of LB (~16 hours) at 37 o C with 200 RPM shaking were washed in sterile PBS and diluted 1/100 into LB, M9 minimal media or MEM-HEPES for bacterial growth assays. Pooled human serum was purchased from Life Science Group Ltd. Pooled human urine was obtained from healthy volunteers.
Serum and urine were filtered, pre-warmed to 37 o C and typically added at a concentration of 10 or 50 % to the indicated media. Control cultures contained an equal proportion of sterile PBS to serum or urine. Where indicated, serum was inactivated by heat treatment at 56 o C for 30 minutes. All other chemicals and antibiotics were purchased from Sigma Aldrich.

Lambda Red mediated mutagenesis
Isogenic strains of E. coli containing single or multiple gene deletions were generated by Lambda Red recombineering (1). Briefly, resistance cassettes from pKD3 (chloramphenicol) or pKD4 (kanamycin) were amplified by PCR using primers (Table   S3) flanked with 50 bp overhangs complementary to the sequence directly adjacent to the 5' and 3' ends of the gene of interest. PCR products were purified and concentrated by phenol:chloroform extraction followed by ethanol precipitation. The parental strain was transformed with pKD46 and grown overnight at 30 o C in LB containing 100 µg/ml ampicillin. This was used to inoculate SOB containing 100 µg/ml ampicillin followed by culture at 30 o C for approximately 2 hours (OD600 of ~0.2) before the addition of 10 mM arabinose to the cultures. After a further hour of growth, cells were chilled on ice for 5 minutes and harvested by centrifugation at 3,500 RPM for 5 minutes. The pellet washed five times in 1 ml of ice-cold distilled water before a final resuspension in ice cold water at 100X concentration relative to the starting culture volume. 50 µl of cells was electroporated with ~1 µg of PCR product and cells were recovered in 1 ml of prewarmed SOC for 2 hours at 37 o C. Half the reaction was plated out onto the appropriate antibiotic containing LB agar plates and incubated overnight at 37 o C. The remaining mixture was left overnight at room temperature and plated the following day. Colonies were screened by PCR using check primers flanking the deleted region to identify successful recombinants. Positive mutants were subsequently transformed with pCP20 (ampicillin, 30 o ) and re-streaked non-selectively at 42 o C to remove the resistance cassettes. Clean deletions were confirmed by colony PCR.

Cloning procedures
Complementation and reporter plasmids were generated by standard cloning using restriction digestion/ligation. All primers are listed in Table S3. For complementation plasmids, genes were amplified by PCR from purified wild type genomic DNA using primers containing BamHI (5') and XbaI (3') overhangs. PCRs were then gel extracted, digested, phosphatase treated and ligated into pSU-PROM according to the manufacturer's specifications (2). This created constructs where the gene of interest was expressed constitutively from the E. coli Tat promoter. For reporter plasmids, promoter regions containing approximately 200 bp upstream of the gene of interest were amplified by PCR from genomic DNA using primers containing EcoRI (5') and BamHI (3') overhangs. PCRs were cloned into pMK1-LUX, as described above (3). This created transcriptional reporters for genes of interest where their respective promoters were fused to the luxCDABE cassette. For all cloning reactions, restriction enzymes and Q5 high fidelity polymerase were purchased from New England Biolabs, and T4 ligase was purchased from Invitrogen. All plasmid inserts were sequenced (Eurofins) to confirm accurate construct generation.

LUX-promoter fusion transcriptional reporter assays
Promoter activity was determined by measuring in parallel the cell density (OD600) and absolute luminescence of cultures carrying LUX reporter fusions at the same phase of growth using a FLUOstar Omega plate reader (BMG Labtech). Assays were performed in white walled/clear bottom microtiter plates and relative luminescence units (RLU) were determined by dividing the absolute luminescence values by OD600.
Experiments were performed in biological triplicate and statistical significance determined using the students t-test.

RNA-seq and ChIP-seq analysis
RNA-seq and ChIP-seq data from our previous study was retrieved from the European Nucloetide Archive (project accession number PRJEB12065) and reanalyzed for this study (4,5). Raw fastq files were imported into CLC Genomics Workbench version 7 (Qiagen) and the reads were aligned to the CFT073 reference genome (NCBI accession number: NC_004431). For RNA-seq, differential expression was performed using the EdgeR tool implemented in CLC (6). Relative expression levels were determined as absolute fold change and genes were considered as significantly differentially expressed by a false-discovery rate corrected p-value threshold of ≤ 0.05.
For ChIP-Seq, peaks were called using the shape analysis tool in CLC (p-value ≤ 0.05) and visually inspected as read tracks aligned to the reference, whereby a TF binding site is defined by a bimodal intersecting peak signifying forward and reverse strand reads aligning either side of the binding site (7). Network analysis of biological functional groups associated with differentially expressed genes and TF binding sites was carried out using the STRING tool (8).

Murine models of urinary tract and bloodstream infection
The UTI model was carried out essentially as described previously (9). The bacterial inoculum was prepared by resuspending twice sub-cultured overnight cultures of UPEC or mutant derivatives grown statically in LB media in sterile phosphate-buffered saline (PBS) to a final concentration of ~2 x 10 8 CFU/ml. Female 8-week-old C57BL/6 mice (Charles River) were inoculated trans-urethrally with 50 µl of the inoculum (~1 x 10 7 CFU) using a lubricated catheter. For co-infections of wild and mutant strains, inoculums for each were prepared identically and mixed 1:1 prior to administration to the animals. After 48 hours of infection, mice were euthanised humanely using increasing CO2 and cervical dislocation.
The BSI model was carried out essentially as previously described (10). The bacterial inoculum was prepared by resuspending overnight cultures of UPEC or mutant derivatives cultured in LB media in sterile PBS to a final concentration of ~1 x 10 8 CFU/ml. Female 8-week-old BALB/c mice (Charles River) were restrained using a Braintree restrainer (Fisher Scientific). The inoculum was administered by tail vein injection using 100 µl of the inoculum (~10 7 CFU). For co-infections of wild and mutant strains, inoculums for each were prepared identically and mixed 1:1 prior to administration to the animals. After 18 hours of infection, mice were euthanised humanely using increasing CO2 and cervical dislocation.

RNA extraction from bacterial cultures and infected tissue
For in vitro samples, 2 ml of bacterial culture was normalised to the same OD600, centrifuged, and resuspended in RNAlater (Ambion). Total RNA was extracted from the cell pellets using a Monarch RNA extraction kit (New England Biolabs) according to the manufacturer's specifications. Residual genomic DNA was removed using TURBO DNase (ThermoFisher Scientific) followed by PCR validation. The RNA was concentrated using phenol:chloroform extraction followed by ethanol precipitation. For in vivo samples, 5 mm sections of infected liver and spleen were stored in RNAlater immediately after dissection for 24 hours at 4 o C. The tissues were subsequently homogenised using a TissueLyser LT (Qiagen) and total RNA extracted using a mirVANA kit (Ambion) according to the manufacturer's specifications. Genomic DNA was removed using TURBO DNase. RNA samples were analyzed on a Qubit (ThermoFisher Scientific) and assessed for degradation using agarose gel electrophoresis.

Quantitative real time PCR (RT-qPCR)
RNA samples were normalised and cDNA synthesis performed using the LunaScript RT SuperMix kit (New England Biolabs) according to the manufacturer specification.
RT-qPCR was performed on the resulting cDNA using a LightCycler 96 Real-Time PCR system (Roche) and the Luna Universal qPCR Master Mix kit (New England Biolabs). The reactions were performed in technical triplicate and each gene that was analysed was performed in biological triplicate. All genes were normalised against the housekeeping gene, gapA. All primers used in RT-qPCR were checked for efficiency (90-110%) using standards made from template gDNA (100, 20, 4, 0.8 and 0.16 ng/µl).
The data was then analysed using the 2 -DDCT method (11).

Type I fimbriae phase variation assay
A PCR based approach to determine the proportion of a cell population expressing Type I fimbriae (fim) was carried out as previously described (4). Briefly, 2 µl of bacterial culture was added to 8 µl of nuclease free water and boiled. This solution was used as a PCR template for amplification of the fim phase-variable promoter region. The PCR product was purified using an NEB Monarch kit and digested with HinfI before being separated on a 2 % agarose gel by electrophoresis. The promoter region contains a unique HinfI restriction site and digestion of the PCR product (559 bp) results in a banding pattern of 74 bp and 485 bp (fim phase ON) or 202 bp and 357 bp (fim phase OFF). Relative band density was determined using ImageJ. The fim promoter region from E. coli TUV93-0 was used as a negative control as it is permanently in a locked OFF state.

Immunoblot analysis
Samples (1 ml) of bacterial were normalised for cell density and harvested by centrifugation. Pellets were resuspended in 100 µl 4x LDS sample buffer (Thermo Fisher), acidified with concentrated HCl to de-polymerise FimA subunits and boiled for 10 minutes before being neutralised with 10 M NaOH. 20 µl of lysate was separated by SDS-PAGE on a 4-12 % Bis-Tris NuPAGE gel (Thermo Fisher) at 180 volts for 45 minutes before being transferred to a 0.45 µm nitrocellulose membrane (GE Healthcare) at 30 volts for 90 minutes. Membranes were blocked with 5 % skim milk powder in PBST followed by incubation with primary (anti-FimA and anti-DnaK) and secondary antibodies (anti-Rabbit HRP-conjugated), separated by 3x 10 minute PBST washes. Immunoblots were incubated with SuperSignal West Pico chemiluminescent substrate for 5 minutes before imaging using the G:Box Chemi system (Syngene).

Recombinant YhaJ purification
6x Histidine tagged YhaJ cloned into pET28 was overexpressed in E.coli BL21-DE3 cells using 1 mM IPTG and purified as previously described (12). Cell pellets were resuspended in wash buffer (200 mM NaCl, 50 mM Tris, 40 mM Imidazole, 10% glycerol) and lysed by French press. YhaJ was immobilised on a HisTrap column (GE Healthcare) using an AKTA-prime and purified by size-exclusion using a Superdex S200 column (GE Healthcare).

Electrophoretic mobility shift assay (EMSA)
EMSA analysis was performed as previously described (4). Promoter regions of interest were amplified by PCR and ddUTP-11-DIG labelled using the DIG Gel Shift Kit (Roche). Binding reactions were set up in 20 µl using 0.2 ng/µl of labelled DNA with increasing concentrations of purified YhaJ (0, 0.3, 0.6 and 1 µM) for 45 minutes at room temperature. Reactions were separated on a 6 % DNA retardation gel (Invitrogen) and transferred to a positively charged nylon membrane (Roche) using the NOVEX system (Thermo Fisher). Transferred membranes were UV crosslinked, blocked, and then probed with AP conjugated anti-DIG antibody at 1/10000. EMSAs were imaged on a ChemiDoc imaging system (Bio-Rad). EMSAs were performed in triplicate.

Statistical analysis
ChIP-seq and RNA-seq analysis was performed using CLC Genomic Workbench