The Salmonella enterica serovar Typhimurium virulence factor STM3169 is a hexuronic acid binding protein component of a TRAP transporter

The intracellular pathogen S. Typhimurium is a leading cause of foodborne illness across the world and is known to rely on a range of virulence factors to colonize the human host and cause disease. The gene coding for one such factor, stm3169, was determined to be upregulated upon macrophage entry and its disruption reduces survival in the macrophage. In this study we characterize the STM3169 protein, which forms the substrate binding protein (SBP) of an uncharacterized tripartite ATP-independent periplasmic (TRAP) transporter. Genome context analysis of the genes encoding this system in related bacteria suggests a function in sugar acid transport. We demonstrate that purified STM3169 binds d-glucuronic acid with high affinity and specificity. S. Typhimurium LT2 can use this sugar acid as a sole carbon source and the genes for a probable catabolic pathway are present in the genome. As this gene was previously implicated in macrophage survival, it suggests a role for d-glucuronate as an important carbon source for S. Typhimurium in this environment.


INTRODUCTION
Non-typhoidal Salmonella is a leading cause of foodborne illness worldwide [1]. In particular, Salmonella enterica serovar Typhimurium (S. Typhimurium) is a common cause of gastroenteritis and can lead to inlammatory diarrhoea and bacteraemia [2,3]. his intracellular pathogen is able to survive and replicate within both phagocytic and non-phagocytic cells. he intracellular environment of a macrophage phagosome is harsh, containing acid, bactericidal compounds and reactive oxygen species [4,5]. However, this also provides a niche for adaptive enteropathogenic bacteria such as S. Typhimurium to colonize and replicate within, isolated from patrolling immune cells [6]. hese bacteria rely on a global change in gene expression to adapt to the intracellular environment, employing a range of virulence factors to mediate host cell entry, survival and growth.
Haneda et al. investigated the shit in protein levels in S. Typhimurium strain SH100 upon macrophage entry and identiied over 50 proteins whose increased accumulation was dependent on the stringent response-related signalling molecule ppGpp [7]. he gene encoding the most diferentially regulated protein, stm3169, was then disrupted on the S. Typhimurium chromosome, and the resulting strain had a signiicant growth defect inside macrophage-like RAW264.7 cells [7]; as such, STM3169 was suggested to be important for macrophage survival and virulence.
In this study we have characterized the function of this protein, which encodes a substrate binding protein of an uncharacterized tripartite ATP-independent periplasmic (TRAP) transporter. We provide a novel perspective into important cellular functions needed for macrophage survival.

Cloning, expression and purification of STM3169
he gene stm3169 was ampliied from S. Typhimurium LT2 (ATCC 700720) and cloned into pETYSBLIC3C [8] by ligation-independent cloning. his plasmid was transformed into the expression host Escherichia coli BL21 (DE3). Protein expression was carried out in 1 litre of Luria Bertani (LB) medium supplemented with kanamycin at 50 μg ml −1 , with 0.4 mM IPTG added once the optical density (OD 600 ) reached 0.6. Cells were incubated at 37 °C at 120 r.p.m. overnight, then harvested by centrifugation at 7800 g for 25 min. he cell pellet was resuspended in wash bufer (50 mM KPi pH 7.8, 200 mM NaCl, 20 % glycerol, 40 mM imidazole) containing 0.4 mM AEBSF protease inhibitor (Sigma Aldrich). he sample was sonicated at 70 W for 3 min, followed by centrifugation at 27 000 g at 4 °C for 30 min. Ni 2+ ainity chromatography was used to purify the His 6 -tagged protein. he cell lysate was loaded onto a HisTrap column (GE Healthcare) and washed with the aforementioned wash bufer, then eluted with the elution bufer (50 mM KPi pH 7.8, 200 mM NaCl, 20 % glycerol, 500 mM imidazole). he protein of interest was identiied by SDS-PAGE. he puriied protein was bufer exchanged into 50 mM KPi pH 7.8 and 200 mM NaCl using the HiTrap Desalting column (GE Healthcare). Protein concentration was measured by determining the absorbance at 280 nm and using the Beer-Lambert law.

Protein renaturation
A fraction of the desalted protein was then loaded onto a 5 ml HisTrap column (GE Healthcare). he column was then washed with 30 column volumes (CV) of denaturation bufer (50 mM KPi pH 7.8, 200 mM NaCl, 20 % glycerol, 40 mM imidazole, 2 M guanidine hydrochloride) to unfold the bound protein. A gradient wash across 12 CV was performed from the denaturation bufer to the previously mentioned guanidine hydrochloride-free wash bufer to refold the protein. A further 8 CV wash using the wash bufer was performed before elution using the elution bufer as before. Circular dichroism was used to check correct refolding of the renatured protein. CD spectra were determined using a J810 spectropolarimeter. Protein samples were dialysed into 20 mM KPi pH 7.8 bufer and diluted to a concentration of 0.2 mg ml −1 . he spectra were determined in a 1 mm pathlength quartz cuvette between 260 and 180 nm at 100 nm min -1 with 1 nm pitch. Prior to any downstream experiments, the protein was desalted into 50 mM KPi pH 7.8 and 200 mM NaCl. he concentration of the refolded protein was measured by determining the absorbance at 280 nm and using the Beer-Lambert law and predicted extinction coeicient from ProtParam (Expasy).

Thermal shift assay
he Protein hermal Shit Dye kit (Applied Biosystems) was used to carry out the thermal shit assay. A inal concentration of 5 µM protein in 50 mM KPi, 20 mM NaCl and each ligand at inal concentrations of 1 nM, 1 µM and 1 mM was added to the thermal shit dye in triplicate. he experiments were performed using a StepOne Real-Time PCR system (Applied Biosystems) with a temperature ramp of 1.5 °C every 2 min from 25 to 99 °C. he melting temperatures (T m ) were calculated by determining the inlection point of the plot of the normalized luorescence measurements against temperature.

Intrinsic fluorescence spectroscopy
All intrinsic luorescence measurements were performed using a FluoroMax 4 luorescence spectrometer (Horiba Jobin-Yvon). STM3169 in 50 mM KPi pH 7.8 plus 200 mM NaCl bufer was used at a concentration of 0.5 µM. he protein was excited at 297 nm with slit widths of 3.5 nm and emission was monitored at 330 nm. To test for potential ligands, various sugar acids were sequentially added at concentrations of 100 µM.
To quantitatively assess binding of speciic ligands, protein was titrated with increasing concentrations of ligand in order to elucidate the K D of ligand binding. Cumulative luorescence was then plotted in Prism 5 (GraphPad) and the K D was calculated from the hyperbolic it of the binding curve.

Microscale thermophoresis
he luorescent dye NT-647-NHS was used at 30 µM to label 15 µM of STM3169. Any unreacted dye was removed using a HiTrap Desalting column (GE Healthcare). Protein and dye concentrations were measured by absorbance at 280 and 650 nm, respectively. he concentration of ligand was varied from 200 µM to 5 nM in a 16-step dilution, then mixed with labelled protein in capillaries for data acquisition using the Monolith NT.115 (NanoTemper). Outliers were detected using the NanoTemper analysis sotware.

Mass spectrometry by FT-ICR-MS
he natively folded STM3169 containing an N-terminal tag ( MGSS HHHH HHSS GLEV LFEGPA) attached to the mature peptide was used. A sample of this protein was dialysed into 35 mM ammonium acetate pH 5 and diluted to a inal protein concentration of 100 µM. Fourier-transform ion cyclotron resonance MS (FT-ICR-MS) was performed using a solariX FT-MS device (Bruker Daltonics) with a 9.4-T superconducting magnet. Analysis was performed in denaturing conditions where the protein was diluted to 1 µM in 50 % (v/v) acetonitrile containing 0.1 % (v/v) formic acid before introduction via a TriVersa NanoMate (Advion BioSciences) source in positive-ion mode. he applied voltage was adjusted to 1.4-1.7 kV to achieve a stable ion current. A 120 °C nitrogen dry gas was supplied at 1.3 l min -1 to aid desolvation. Instrument control and data acquisition used Compass 1.4 (Bruker Daltonics). Spectra were generated by the accumulation of 20 scans with 0.1 s ion cooling time and 0.4 s scan time with 8 million data points recorded. To improve resolution and reduce space-charge efects, the 42+ protein charge states were isolated in the hexapole before scanning in the FT-ICR cell. Spectra from isolated charge states were deconvoluted to monoisotopic protein mass using

Homology modelling of STM3169
he predicted structure of the mature STM3169 was prepared by automated homology modelling on SWISS-MODEL ( swissmodel. expasy. org) [9][10][11][12]. he structure of Bamb_6123 from Burkholderia ambifaria AMMD (PDB ID: 4N15) was used as a template for homology modelling of the mature peptide of STM3169. he QMEAN Z-score was predicted to be −0.69, which suggests that the model of STM3169 is consistent with other experimentally determined structures of a similar size. he derived GMQE (global model quality estimation) for the model was determined to be 0.74, indicating high reliability.

S. Typhimurium LT2 growth experiments
A stationary phase culture of S. Typhimurium LT2 (ATCC 700720) grown in Luria Bertani (LB) media was centrifuged and washed with M9 salts [13]

RESULTS AND DISCUSSION
he stm3169 gene from S. Typhimurium LT2 is encoded within a typical operon for a TRAP transporter, in that it also includes genes encoding the two required membrane subunits (stm3170 and stm3171) [14] and so is probably part of a functioning transporter (Fig. 1). STM3169 is the substrate binding protein (SBP) component that initially binds the substrate in the periplasm; this component is usually expressed at high levels relative to the membrane domains, consistent with the high abundance of the STM3169 protein in the Haneda et al. study [7,14].
TRAP transporters form a large family of diverse SBPdependent secondary transporters in bacteria, the functions of which are, largely, to move organic acids into bacterial cells [15][16][17]. A hallmark sign of this substrate preference is a conserved arginine residue in the SBP subunit that coordinates the carboxylate group of the acid ligand [18,19]. his residue is conserved in STM3169 (Arg170), suggesting a typical organic acid-type ligand. Beyond predicting that the ligand contains a carboxylate group, the binding sites of diferent TRAP transporter SBPs are highly variable, allowing them to accommodate ligands of many diferent sizes and shapes, oten using complex water networks [17,20].
In S. Typhimurium LT2, the stm3169-stm3171 TRAP transporter gene cluster is not found in a region containing any other associated genes that give any clue to its substrate and hence cellular function (Fig. S1a, available in the online version of this article). However, genome analysis of stm3169 orthologues from related gamma-proteobacteria revealed a range of diferent cluster architectures that strongly suggest a role in sugar acid import and catabolism for these strains (Fig. 1). A common feature of these clusters is the presence of genes coding for proteins predicted to be involved in hexuronate catabolism. hese proteins, UxuA, UxuB and UxuC, have been characterized in Erwinia chrysanthemi where they are responsible for the utilization of d-glucuronate following uptake by the hexuronate MFS transporter ExuT [21]. he Fig. 1. Homologues of STM3169 may be encoded from operons containing hexuronate metabolic genes. In S. Typhimurium LT2, stm3169 is found on an operon with stm3170 and stm3171, the two putative membrane components of a TRAP transporter. This operon is not surrounded by other genes that might be involved in uptake or sugar metabolism. Some genes neighbouring stm3169 homologues were predicted to code for the hexuronate catabolism enzymes UxuAB and a glucuronate isomerase, UxuC/UxaC. Genes are represented by coloured bars, identifying its encoded protein's putative function. Homologues were identiied in other strains using MicrobesOnline.
presence of homologues of UxuABC (STM3135-STM3137) with ExuT (STM3134) in S. Typhimurium (Fig. S1b) suggests its ability to catabolize this sugar. Additionally, this pattern of co-occurrence is found in other genomes represented in MicrobesOnline [22], suggesting that this TRAP transporter might be involved in uptake of hexuronic acids.
To test this hypothesis, we expressed and puriied recombinant STM3169 from E. coli using nickel-ainity chromatography (Fig S2a). A portion of the puriied protein was denatured using guanidine hydrochloride and refolded, removing any endogenous ligand potentially carried over during the puriication, a common phenomenon with these SBPs [23,24]. Analysis of the protein by circular dichroism suggested minimal diferences in the folding when comparing the native and refolded protein, suggesting successful protein refolding (Fig. S2b). he protein was analysed by MS to have a mass of 35 741.96 Da, which matches the predicted monoisotopic mass of the protein (Fig. S2c). An additional signal at mass 35 840.4 was observed, which, being 98 Da greater than the predicted signal, is probably a single phosphate adduct on the protein, an artefact we have observed with other SBPs [25].
To screen potential substrates for this TRAP transporter we used an established luorophore-based protein thermal shit assay [17] using a range of monosaccharides and their related acid forms. Analysis of the natively puriied protein showed no signiicant changes in protein thermal stability in the presence of any of the compounds tested: d-glucuronate, d-galacturonate, d-gluconate, d-galactonate, d-galactose and the negative control, l-arabinose (data not shown). However, for the refolded and presumably ligand-free protein, increases in protein thermal stability were observed in the presence of two sugar derivatives, d-galactonate and d-glucuronate (Fig. 2a). hese data suggest that the puriied protein from E. coli came with pre-bound ligand and also suggested ligands for continuation of in vitro binding experiments.
To further assess binding we used luorescence spectroscopy, measuring the change in intrinsic protein luorescence upon ligand binding. Using protein excitation at 295 nm, the same set of sugar acids were added sequentially to refolded STM3169, detecting luorescence in time acquisition mode. Whilst addition of d-galactonate had little efect on luorescence, d-glucuronate caused an ~10 % quench in luorescence (Fig. 2b). his discrepancy may be attributed to the nature of protein thermal shit experiments as non-speciic interactions could give rise to a shit in thermal stability.
We proceeded to determine the binding ainity of STM3169 for d-glucuronate by titration and measuring the resulting changes in intrinsic luorescence (Fig. 2c). he experimental data suggested d-glucuronate was binding to STM3169 with a binding constant (K D ) of 0.17±0.077 µM. As a complementary biophysical methodology, we examined this interaction using microscale thermophoresis (MST), measuring a ligand-dependent change in thermophoresis of dyed protein molecules to indicate conformational changes that accompany ligand binding. his technique also suggested d-glucuronate was binding STM3169, and through repeated 16-step titrations a K D of 0.25±0.048 µM was estimated (Fig. 2d). No clear signal suggesting a conformational change upon d-galactonate binding was observed (data not shown). Together, these data are consistent with d-glucuronate being a ligand for STM3169, with ainities in the range determined for other high-ainity TRAP transporters such as the SBP of the sialic acid transporter of Haemophilus inluenzae, SiaP, with a reported K D of ~0.1 µM [26].
A recent study by Vetting et al. [17] screened 158 SBPs against a large number of potential ligands to classify these proteins into isofunctional clusters. A number of these SBPs were then characterized structurally, some of which were successfully co-crystallized with d-glucuronate. We sought to compare these crystal structures against a model of STM3169 prepared on SWISS-MODEL ( swissmodel. expasy. org) [9][10][11][12] to supplement our in vitro indings. hese proteins were superposed using the secondary-structure matching (SSM) algorithm [27] on CCP4MG [28] and the measured root mean square distances of the alignments were found to be between 1 Å and 1.56 Å, suggesting highly similar overall structures. Analysis of the aligned binding sites (Fig. 3) initially revealed two conserved arginines (Arg150, Arg170 in STM3169) and an asparagine (Asn210 in STM3169), consistent with known d-glucuronate binding proteins [18]. As predicted, Arg170 in the model of STM3169 aligns with the residues responsible for coordinating the carboxyl group of the ligand in the d-glucuronate bound structures of Apre_1383 and Bpro_3107. he two other conserved residues, Arg150 and Asn210, have been shown to further coordinate the ligand by forming multiple hydrogen interactions. STM3169 residues Asn35, Gln36, Glu74 and Asp237 are fully conserved in the d-glucuronate-speciic binding protein Apre_1383 and its corresponding residues form more hydrogen bonds with d-glucuronate. Bpro_3107, a d-glucuronate and d-galacturonate binding protein, possesses alternative residues also capable of coordinating the ligand. his intricate network of multiple predicted hydrogen bonds is likely to give rise to the high binding ainity for d-glucuronate shown though our in vitro assays.
hrough in vitro studies and homology modelling, we have characterized a high-ainity substrate binding component of the TRAP transporter with apparent speciicity for d-glucuronate. As S. Typhimurium also has homologous genes coding for proteins involved in d-glucuronate catabolism, it should be able to import d-glucuronate and utilize it as a carbon source. We conirmed this, an observation irst observed in 1969 from the Ames group [29] (Fig. 4a); furthermore, we demonstrated the utilization of a selection of other sugars used in this study. From these data we propose a model (Fig. 4b) whereby the STM3169 protein forms part of a functioning TRAP transporter that functions alongside the putative UhpC/ExuT hexuronate transporter (STM3134) that is divergently transcribed from the uxuABC genes (stm3135-stm3137). Ater transport, the bacterium would use the UxuABC proteins and the enzymes of the Entner-Doudorof pathway to convert d-glucuronate into central metabolites. he high-ainity nature of the TRAP transporter, compared to the lower ainity major facilitator superfamily protein ExuT, might suggest that d-glucuronate is present in the macrophage as a carbon source but at low micromolar concentrations, which would explain the signiicant increase in protein levels in the macrophage to ensure its eicient uptake [7]. his ability to metabolize sugar acids is key for the colonization of intracellular niches, where other substrates for glycolysis may not be present [16]. he landmark study of Eriksson et al. [30], which provided the irst comparison of S. Typhimurium transcriptional changes upon moving to the intravacuolar environment of a macrophage, revealed that the organism appeared to be using the related sugar acid d-gluconate and the Entner-Douderof pathway during growth [30]. In fact, the UhpT/ExuT-like transporter STM3134 was identiied as being speciically upregulated as well as dgoK and dgoA, which encode enzymes that interconvert the UxuABC reaction products into pyruvate and glyceraldehyde-3-phosphate, suggesting a broader role for sugar acid molecules as metabolites. While STM3134 may be responsible for the transport of multiple hexuronate carbon sources, we have demonstrated that STM3169-STM3171 is a selective high-ainity transport system for d-glucuronate.
Despite the lack of evidence to suggest that the STM3169-STM3171 TRAP transporter is able to transport d-galactonate, our deined growth media assay (Fig. 4a)  another transporter capable of eicient d-galactonate uptake. his and other sugars found in the macrophage could be taken up by other high-ainity transporters, including further members of the TRAP family. here is previous evidence suggesting the two other members of the TRAP family in S. Typhimurium LT2, STM3671-STM3673 and STM4052-STM4054, have transport functions which could include carbon sources in the macrophage. he SBP of STM3671-STM3673 is 88 % identical to the previously characterized E. coli 2,3-diketo-l-gulonate binding protein, YiaO [31][32][33], whilst STM4052-STM4054 is thought to be coregulated with proteins of the rhamnose utilization pathway by the activator RhaS (STM4048) through promoter analysis by RegPrecise [34].
Finally, we have shown evidence of function for the longidentiied virulence protein STM3169 and suggested a role for its associated transport system in cell metabolism. Our discovery adds to our understanding of how pathogenic bacteria such as S. Typhimurium are able to survive within hostile host intracellular environments. Given its key role in virulence as observed by Haneda et al. [7], this knowledge could potentially aid in the design of competitive inhibitors.

Funding information
The work was supported by the Department of Biology, University of York, through support of C.B.N., an Integrated Masters in Biochemistry student, supported by R.H. who is employed on a BBSRC grant (BB/ N01040X/1).