Salmonella enterica Serovar Typhimurium Uses PbgA/YejM To Regulate Lipopolysaccharide Assembly during Bacteremia

Salmonella enterica serovar Typhimurium (S. Typhimurium) relies upon the inner membrane protein PbgA to enhance outer membrane (OM) integrity and promote virulence in mice. The PbgA transmembrane domain (residues 1 to 190) is essential for viability, while the periplasmic domain (residues 191 to 586) is dispensable. Residues within the basic region (residues 191 to 245) bind acidic phosphates on polar phospholipids, like for cardiolipins, and are necessary for salmonella OM integrity. S.

amplified from the pBAV1k template, restriction enzyme digested, and ligated into pGRG37 between the left and right insertion elements of the Tn7 transposon (1). The pGRG37 vector allowed for transposition of the pbgA rescue allele at the neutral site 3' of glmS. We were initially concerned that the tetracycline resistance cassette and selection might have impacted pbgA mutant phenotypes and the evolution of suppressor genotypes. Therefore, we constructed an identical mutant allele using an alternative antibiotic resistance gene cassette. The pbgAΔ191-586::kan deletioninsertion genotype, which encoded for kanamycin resistance, was constructed using the phage lambda-red recombinase system (4). The deletion-insertion allele was then horizontally transferred into a fresh non-pKD46-bearing wild-type wza-lacZ salmonellae using bacteriophage P22 HT105/1 int-201. A similar procedure was used to delete phoPQ from the chromosome.
Growth curves (continued for the Main Text). The initial inoculum was prepared after resuspending a single colony in 170μl of LB broth and serial diluting 10 -1 , 10 -2 , and 10 -3 10 -4 . Bacteria were incubated at 37°C for 24h with continuous agitation and OD600 measurements were recorded at 30min time intervals. Four biological replicates were performed for each genotype and the results reflect the average. The 5ml growth curve was generated using a roller drum (Fig. S1A). A single colony was inoculated into 5ml of LB broth and incubated at 37°C at ~250 rpm. OD600 readings were taken hourly for 16h. Three biological replicates were performed per genotype and experiment. The results reflect the average for three experiments. The 1L growth curve was produced in a shaker incubator. A single colony was inoculated into 1L of LB and incubated at 37°C and ~250 rpm. OD600 readings were recorded hourly for 24h (Fig. S1B). The results reflect the average from two independent experiments.
Beta-galactosidase (β-Gal) assay (continued for the Main Text). The β-Gal assays were performed using standard procedures. Briefly, 5ml of log-and stationary-phase bacterial cultures were pelleted and re-suspended in Z-buffer. OD600 readings were taken to quantify culture density. Bacteria were permeabilized with chloroform and 0.1% (wt/vol) SDS. The time to develop a yellow color was recorded following addition of ortho-Nitrophenyl-b-galactoside (ONPG) (4mg/ml in PBS) followed by incubation at 30°C. Na2CO3 at 1M was added to stop the reaction and the mixture was centrifuged.
The supernatant was withdrawn and OD420 measurements were made. The levels of LacZ, or b-Gal, activity were calculated according to the formula (OD420)/(time x volume x OD600). Three biological replicates were measured for each genotype in each experiment. The mean values were calculated from three experiments.
Rifampin sensitivity assay (continued for the Main Text). Plating efficiency on rifampin (Rif) was determined by first adjusting the density of 5ml log-and stationary-phase cultures to OD600 1.0. Cells were serially diluted ten-fold in PBS and 2µl aliquots were spotted onto LB agar with Xgal, with or without rif at 2.5µg/ml. The plates were incubated at 37ºC OVN. The complementation genotype was grown in broth with and without 0.5mM theophylline. However, theophylline was not added to the plates. This allowed us to distinguish the effect of low-level PbgA expression and overexpression in broth culture.
Whole-genome sequencing (continued for the Main Text). To isolate genomic DNA for sequencing, strains were grown OVN in 5ml cultures of LB broth at 37°C. Genomic DNA was isolated using a QIAamp Ò DNA Mini Kit. A library of random fragments was generated for each genome using standard Illumina Nextera libraries. Libraries were sequenced according to manufacturer's standards on an Illumina MiSeq system, and 300-or 600-bp paired-end reads were generated at a minimum of 35-fold genome coverage. The genomes for the wild type and the pbgAΔ191-586::tetRA mutant, which carried the wza-lacZ reporter, were sequenced and compared to the 14028s reference genome to first identify pbgA-independent variations that might exist in this strain background. The sequenced suppressor genomes were then compared against the mutant to identify candidate non-synonymous single nucleotide polymorphisms (snps).
Additional details are provided in Table 1. After sequencing, the snps were individually confirmed by primer targeted sequencing of the LapB/YciM, FtsH, and LpxC coding region. Membrane fractionation (continued from the Main Text). Total membranes were collected and separated as follows. Log and stationary phase cultures were normalized to OD600 0.7 and harvested by centrifugation. The pellets were re-suspended in a sucrose solution to begin the lysis procedure, which allows for the efficient separation of the bilayers into two defined fractions, an IM fraction and an OM fraction (5). Briefly, the total membranes were resuspendend in 20% sucrose in 10mM Tris-HCl pH 7.8 and 0.5mM EDTA and 200μl was collected as a total membrane fraction for each genotype and stored at -20°C. The remaining 800μl of total membrane was applied to a discontinuous sucrose density gradient in order to isolate the IM and OM fractions. The gradient consisted 2ml of 73% sucrose, 4ml of 53%, 1ml of resuspended membranes in 20% sucrose. The gradient was filled to volume with the 20% sucrose solution. Each sucrose solution was prepared in 10mM Tris-HCl pH 7.8 at 0.5mM EDTA. Density gradients were centrifuged for 16h at 35k rpm using a Beckman Optima L-90K

Murine macrophage infections
Ultracentrifuge. Low-density IM fractions were carefully collected as a defined upper interface, brown in hue, using a pipette. High-density OMs were collected as a defined lower white interface. The membranes were washed and resuspended in 1ml of 1mM Tris, pH7.5, and stored at -20ºC

Protein quantification and phospholipid extraction (continued for the Main Text).
Protein concentrations were measured using Pierce Coomassie Plus Bradford assay reagent (Thermo Scientific) and a standard curve generated from bovine serum albumin standard (BSA). The equivalence of 1mg of protein for each membrane was extracted by Bligh-Dyer method to obtain the phospholipid extract that was ultimately analyzed by liquid chromatography tandem mass spectrometry (LC-MS/MS) (6).

Clearing of anti-PbgA rabbit sera (continued for the Main Text). Antibodies to
PbgA(191-586) were raised in rabbits as described previously (7). 5ml of rabbit antisera was eluted over a protein-A column (Thermo) to isolate Fc/Fab fragments. Non-specific S. Typhimurium cross-reacting antibodies were cleared by incubating 50μL of the Phase-contrast and epifluorescence microscopy Bacteria were labeled as described previously (7). Briefly, each strain was incubated with FM4-64 (0.5 µg/ml) for 1 h at 37 °C in Luria-Bertani Broth. Next, bacteria were diluted in PBS and immobilized on 0.2% agarose pads. After gentle drying the pads, the pads were sealed within a coverslip, using a hot-glue gun. Bacteria were imaged using the appropriated UV-filters. Mouse infections (continued from the Main Text). Homogenates were serially diluted and plated onto LB agar with X-Gal and chloramphenicol (10µg/ml) and incubated overnight at 37°C. The antibiotic resistance cassette is linked to the wza-lacZ allele, so that all genotypes in this study are resistant to chloramphenicol and possess the gene reporter of Rcs-signaling activity. Colony counts were normalized to individual organ weight. For murine lethality studies, 6-8-week-old male and female mice were intraperitoneally infected with roughly 5 × 10 3 cfu/mouse. The weight, appearance, and behavior of each infected animal were monitored daily using our standard mouse pain, distress and morbidity scoring system. Measurements were taken until at least 21days post infection (dpi) or until mice achieved a threshold morbidity score. The livers and spleens of deceased mice were dissected, weighed, homogenized, and plated to enumerate cfu. Randomly selected C57BL/6J mice infected with pbgAD191-586 and pbgAD191-586 lpxCY113C bacteria were euthanized at day 21, to assess cfu in the livers and spleens. During experiments were the complementation genotype was tested, 0.5mM theophylline was added to the drinking water of the cage housing the infected animals.
Lipopolysaccharide (LPS) extraction, electrophoresis, and detection (continued from the Main Text). Log and stationary-phase cultures were normalized to OD600 2.0, pelleted, and resuspended in 200µl of DEPC treated endotoxin free water. To corroborate that each sample contained approximately the same cfu level, 20μl was used for serial dilutions in PBS and 100µl aliquots were plated to enumerate bacteria concentrations in the solution. Next, 2µl of 2% SDS was added to the remaining 180µl of sample and the solution was boiled for 5min. 5ul of Proteinase K (NEB) was added to the reaction and the samples were incubated at 59ºC, overnight. Ice-cold Tris-saturated phenol was added (1:1) and the samples were vortexed and incubated at 70ºC for 15min. After cooling, diethyl ether was added (4:1) to generate a two-phase solution.
The solution was centrifuged and the bottom layer was harvested and re-extracted.