ABSTRACT
Salmonella species can gain access into nonphagocytic cells, where the bacterium proliferates in a unique membrane-bounded compartment. In order to reveal bacterial adaptations to their intracellular niche, here we conducted the first comprehensive proteomic survey of Salmonella isolated from infected epithelial cells. Among ∼3,300 identified bacterial proteins, we found that about 100 proteins were significantly altered at the onset of Salmonella intracellular replication. In addition to substantially increased iron-uptake capacities, bacterial high-affinity manganese and zinc transporters were also upregulated, suggesting an overall limitation of metal ions in host epithelial cells. We also found that Salmonella induced multiple phosphate utilization pathways. Furthermore, our data suggested upregulation of the two-component PhoPQ system as well as of many downstream virulence factors under its regulation. Our survey also revealed that intracellular Salmonella has increased needs for certain amino acids and biotin. In contrast, Salmonella downregulated glycerol and maltose utilization as well as chemotaxis pathways.
INTRODUCTION
While infectious diseases continue to be a major threat to human health, there is an urgent need in rational design and development of new treatment strategies. As a model for studying bacterial pathogenesis, Salmonella spp. cause millions of infections per year ranging from food poisoning to life-threatening systemic typhoid fever (1). Central to its pathogenesis is a syringe-like structure known as the type III secretion system (T3SS), which delivers bacterial proteins called effectors directly into eukaryotic host cells (2, 3). The injected proteins are able to modulate various host cellular functions, and over the years extensive studies have been carried out on these effector proteins and/or their interacting host targets. While these studies have contributed substantially to our initial understanding of infection biology, daunting challenges are encountered because conventional reductionism-based studies certainly cannot explain the complex multifactorial nature of host-pathogen interactions (4). Systems-level analyses can provide a panoramic view of the functional host-pathogen interplay and thus are promising in this regard.
During infection, the intracellular environment is likely to be very different from what bacteria encounter in rich culture media. It has long been known that various nutritional and environmental differences within host cells force bacterial pathogens to reprogram their gene expression. In fact, transcriptomic studies of intracellular Salmonella within infected macrophages and epithelial cells contributed to our initial understanding of bacterial adaptations in the host environment (5, 6). Because proteins are the final gene products and their changes may not correlate with those of mRNA, direct readout of the bacterial proteome is highly desired. Such measurements, however, have been technically challenging due to the presence of vast amounts of host proteins (7). As a matter of fact, until now >90% of Salmonella proteomics research has focused on bacteria grown in vitro either from different strains or under culture conditions with various perturbations (8, 9, 10, 11, 12, 13, 14, 15, 16, 17).
The first in vivo Salmonella proteome was reported by Becker et al. in 2006, and a flow cytometry-based strategy was used to sort out bacteria from infected mouse tissues (18). In total, they detected 370 and 835 proteins from the spleen and cecum, respectively. Further mutational analyses revealed that most of these observed proteins/metabolic enzymes are nonessential for Salmonella virulence, suggesting robust bacterial metabolism and availability of diverse nutrients in the host. In the same year, Shi et al. reported the first quantitative time course study of intracellular Salmonella proteome in infected macrophages (19). They found that 39 of 315 bacterial proteins were strongly induced after infection and, notably, that deletion of STM3117 led to markedly reduced replication, implicating STM3117 in a critical role in Salmonella colonization inside macrophages. It is also noteworthy that studies of closely related bacterial pathogens (i.e., Escherichia coli or Shigella spp.) have expanded our knowledge on bacterial adaptations upon interactions with their host cells (20, 21, 22, 23, 24, 25). For instance, Pieper et al. studied the Shigella flexneri proteome within infected epithelial cells (24). Notably, they found that mixed-acid fermentation was required for S. flexneri intracellular replication and cell-to-cell spread. Taking these data together, some common themes may start to emerge regarding adaptation mechanisms of Gram-negative bacterial pathogens.
Salmonella likely encounters distinct differences in its intracellular niches in epithelial cells compared to macrophages. While substantial work has illuminated bacterial strategies to evade host antibacterial activities in macrophages that are exerted by the phagocyte NADPH oxidase and inducible nitric oxide synthase (iNOS) (26, 27, 28, 29), much less is known about Salmonella adaptation mechanisms in host epithelial cells. Given that Salmonella encodes ∼4,500 proteins, adequate proteome sampling would be crucial to our understanding of Salmonella infection biology. Here we report the most comprehensive survey to date of the Salmonella enterica serovar Typhimurium proteome upon infection of nonphagocytic epithelial cells. Using a proteomic strategy that we described previously (30) to study intracellular Gram-negative bacteria, we quantitatively analyzed Salmonella protein expression at 1 h and 6 h postinfection along with extracellular bacteria. Among >3,000 detected proteins, we found ∼100 Salmonella proteins that were significantly altered during the transition of the bacteria from the extracellular to the intracellular niche. Our comprehensive proteomic survey revealed extensive Salmonella adaptations within host epithelial cells and provided exciting insight into the biology of Salmonella-host interactions.
MATERIALS AND METHODS
Bacterial strains, cell lines, and culture conditions.The S. enterica serovar Typhimurium SL1344 wild-type strain was kindly provided from Feng Shao's laboratory (National Institute of Biological Sciences, Beijing, China). All strains were maintained frozen at −80°C in the peptone solution (2% peptone, 5% glycerol). The frozen bacteria were routinely grown on LB plates with 15% agar and 30 μg/ml streptomycin at 37°C. A single colony picked from the plate was inoculated into 3 ml LB medium, and then the overnight culture was diluted 1:20 into 3 ml LB broth (with 0.3 M NaCl to increase bacterial invasion). It took ∼3.5 h for the bacterial culture to reach an optical density at 600 nm (OD600) of 0.9, and that OD was used for infection assays. To examine protein expression in vitro under iron-limiting conditions, Salmonella grown in LB medium (OD600 = 0.9) without and with 2,2′-dipyridine (DIP) at 70 and 280 μM was compared. To assess the impact of the medium shift, we compared the proteomes of Salmonella grown in LB medium (OD600 = 0.9) without and with further incubation in Hanks' balanced salt solution (HBSS) for 30 min. HBSS contains CaCl2, MgCl2, MgSO4, KCl, KH2PO4, NaHCO3, Na2HPO4, and glucose at millimolar concentrations in addition to NaCl (∼140 mM). HeLa cells were grown in Dulbecco's modified Eagle medium (DMEM; HyClone) supplemented with 10% fetal bovine serum (FBS; Gibco/Life Technologies, USA) under an atmosphere of 5% CO2 at 37°C.
Salmonella infection and isolation from infected human epithelial cells.Salmonella invasion of HeLa cells was performed when cell monolayers reached 70% to 85% confluence. For isolation of intracellular bacteria, infection was carried out for 30 min in HBSS with a multiplicity of infection (MOI) of 100. After infection, cell monolayers were washed with prewarmed HBSS and incubated further in prewarmed DMEM supplemented with 100 μg/ml gentamicin to kill extracellular bacteria. After 1 h, cell monolayers were washed again with prewarmed HBSS, and fresh DMEM supplemented with 10 μg/ml gentamicin was added. At 1 h and 6 h postinfection, cells were washed extensively with phosphate-buffered saline (PBS) and lysed in 20 mM Tris-HCl (pH 7.6) buffer containing 150 mM NaCl and 0.1% Triton X-100. To recover intracellular bacteria, first, collected cell lysates were centrifuged at 600 × g for 5 min to remove nuclei and cell debris, and then the supernatant was centrifuged at 4,000 × g for 20 min to pellet bacteria. The pellets were immediately washed with radioimmunoprecipitation assay (RIPA) buffer (25 mM Tris-HCl [pH 7.6], 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS) to remove residual host proteins. The final bacterial pellets were kept frozen at −20°C until further analyses.
Proteomic and metabolomic sample preparations.Bacterial pellets were resuspended in the SDS-PAGE sample buffer containing 60 mM Tris-HCl (pH 6.8), 1.7% (wt/vol) SDS, 6% (vol/vol) glycerol, 100 mM dithiothreitol (DTT), and 0.002% (wt/vol) bromophenol blue and then heated at 95°C for 5 min. Bacterial samples were then prefractionated by 10% SDS-PAGE. Once the band of bromophenol blue reached about 1 cm below the stacking gel, the electrophoresis was stopped. Subsequently, each sample was processed into eight gel bands and subjected to in-gel trypsin digestion as previously described (31). The resulting tryptic peptides were extracted from the gel twice by equilibrating the samples with 50% acetonitrile (ACN) and 5% formic acid (FA) for 20 min at 37°C. Finally, peptide samples were vacuum dried and reconstituted in high-performance liquid chromatography (HPLC)-grade water (Fisher Scientific, USA) prior to liquid chromatography-tandem mass spectrometry (LC-MS/MS) analyses. We analyzed four biological replicates of intracellular Salmonella, one set of in vitro-grown bacteria under iron-limiting conditions and three biological replicates of extracellular bacteria upon incubation in HBSS. With eight gel fractions per bacterial sample, in total we carried out 136 LC-MS/MS experiments (no technical runs).
For metabolite extraction, infected mammalian cells or bacterial pellets were extensively washed in PBS prior to addition of 50% ice-cold methanol. Resuspended samples were snap-frozen in liquid nitrogen and subjected to 15 freeze-thaw cycles. The supernatants were collected by centrifugation at 15,000 × g for 10 min and vacuum dried. Metabolite samples were reconstituted in 5% ACN and analyzed directly by LC-MS/MS without further fractionation.
LC-MS/MS analyses.A hybrid LTQ-Orbitrap Elite mass spectrometer (equipped with an EASY-nLC 1000 system; Thermo Scientific) was used for proteomic and metabolomic analyses. The capillary column (75 μm by 150 mm) with a laser-pulled electrospray tip (model P-2000; Sutter Instruments) was home-packed with 4-μm-diameter, 100-Å Magic C18AQ silica-based particles (Michrom BioResources Inc., Auburn, CA). The mobile phase was comprised of solvent A (97% H2O, 3% ACN, and 0.1% FA) and solvent B (100% ACN and 0.1% FA). The LC separation was carried out with the following gradient: solvent B was started at 7% for 3 min and then raised to 35% in 40 min; subsequently, solvent B was rapidly increased to 90% in 2 min and maintained for 10 min before 100% solvent A was used for column equilibration. Eluted fractions from the capillary column were electrosprayed directly onto the mass spectrometer for MS and MS/MS analysis. A data-dependent acquisition mode was enabled. One full MS scan (m/z 350 to 1,800 for proteomic analyses and m/z 100 to 1,100 for metabolomic analyses) was acquired by Orbitrap Elite, and then MS/MS analyses of the 10 most intense ions were performed in the ion trap. Dynamic exclusion was set with a repeat duration of 24 s and an exclusion duration of 12 s.
Proteomic data analyses.Raw MS files were processed by Proteome Discoverer (Version 1.4.0.288; Thermo Scientific) and searched with Mascot (Version 2.3.02; Matrix Science Inc.) against an S. enterica serovar Typhimurium LT2 protein database (downloaded from UniProt). Monoisotopic mass was selected with a precursor mass tolerance of 20 ppm and a fragment mass tolerance of 0.8 Da. Maximum missed cleavage was set to 2. Cysteine carbamidomethylation was set as a fixed modification, and methionine oxidation was set as a variable modification. The resulting assignments were filtered to achieve a false-discovery rate (FDR) of <1% at both the peptide and protein levels using the target-decoy method. The relative abundances of protein from different samples were assessed using a spectral counting method. Spectral counts represent the total number of repeated identification of peptides for a given protein during the entire analysis and provide a semiquantitative measurement of protein abundance. Raw spectral counts were normalized against the total counts of all identified proteins in a given sample. Only the proteins that were detected in >40% of all data sets and/or the relatively more abundant proteins (i.e., the top 50% of all identifications) were included for differential analyses. Filtered proteins with average fold change values of >2 or <0.5 and P values of <0.05 (Student's t test) were considered to represent up- or downregulation.
Genetic manipulation and characterization.Salmonella deletion mutants and 3×FLAG-tagged strains were generated by the standard homologous recombination method using a suicide plasmid, pSR47s. All primers used in this study are listed in Table S1 in the supplemental material. For constructing the ΔentC single mutant, a PCR fragment containing two flanking sequences of the target gene was cloned into pSR47s. The resulting targeting vector was transferred into the wild-type strain through E. coli DH5α (λpir)-mediated conjugation. The desirable transconjugants were selected on LB agar plates containing 50 μg/ml kanamycin and 30 μg/ml streptomycin. These colonies were further screened for markerless in-frame deletion by growth on LB agar plates containing 15% sucrose without NaCl, and deletion of the target gene was confirmed by sequencing and PCR analyses. The ΔentC ΔfeoAB double mutant was constructed in the same way. For chromosomal gene tagging, the sequences encoding 3×FLAG epitopes were inserted in-frame at the C termini of targeted bacterial genes right before the stop codons.
To assess mutant growth in iron-limiting media, an overnight culture of each of the wild-type, ΔentC, and ΔentC ΔfeoAB strains was diluted 1:3,000 into 3 ml LB broth supplemented with 280 μM DIP. For the intracellular replication assay, HeLa cells in 6-well plates were infected with an MOI of 10. At 1 h and 20 h postinfection, infected cells were lysed and viable intracellular bacteria were enumerated by CFU assays.
Validation of proteomic results by Western blotting.The 3×FLAG-tagged strains were used to infect HeLa cells in 6-well plates with an MOI of 100. At the appropriate times (1 h and 6 h) after infection, mammalian cells were lysed and very crude fractions of intracellular bacteria were obtained by differential centrifugation. Isolated bacterial pellets were resuspended in the SDS loading buffer and run by SDS-PAGE. Gel-separated bacterial proteins were further transferred to polyvinylidene difluoride (PVDF) membranes and blotted against primary antibodies specific for Salmonella DnaK (Enzo Life Sciences) (1:5,000) or FLAG (Cwbio, China) (1:2,000) followed by horseradish peroxidase (HRP)-conjugated secondary antibodies (Cwbio, China) (1:5,000).
RESULTS
The global proteomic profile of intracellular Salmonella upon infection of host epithelial cells and bacteria grown in vitro.We analyzed the proteome of four Salmonella populations: bacteria grown in the LB medium, bacteria upon incubation in HBSS (the infection medium), and intracellular bacteria at 1 h and 6 h postinfection. We managed to recover a sufficient number of Salmonella bacteria (on the order of 108) from infected HeLa cells. We identified ∼3,300 bacterial proteins in total, and, on average, we detected 1,675 and 2,080 proteins per sample for intracellular and extracellular bacteria, respectively. In general, extracellular bacteria were observed with ∼20% to ∼30% more proteins. In addition, we found that <15% of all protein identifications from intracellular samples were of human origin, thereby suggesting minimal host contamination.
As HBSS differs substantially from the LB medium, it may induce changes prior to bacterial internalization. Thus, we first examined the Salmonella proteome upon incubation in HBSS and found that 67 proteins were induced and 22 proteins were repressed. Notably, the vast majority of upregulated proteins were found to be associated with amino acid uptake and biosynthesis (see Fig. S1 in the supplemental material). For example, almost a complete set of proteins that are encoded by the arginine biosynthesis operon (ArgABCDEGHI) were observed at substantially higher expression levels. Next, we focused on differentially expressed proteins of intracellular Salmonella by comparing bacteria harvested at 1 h and 6 h postinfection. In total, we found that the abundance of 100 Salmonella proteins was significantly altered and that 61 of the proteins were upregulated and 39 proteins were downregulated. To get a quick view of altered proteins, we graphed protein fold differences (6 h over 1 h) as a scatter plot (Fig. 1). Notably, many upregulated proteins (highlighted in the gray circle) are involved in Salmonella iron-uptake pathways such as the Iro (IroNBCD) and Ent (EntACEF and FepA) systems. In addition, PhoPQ-regulated proteins (e.g., proteins encoded on Salmonella pathogenicity island 2 [SPI-2]), phosphate utilization proteins, and zinc and manganese transporters as well as biotin biosynthesis proteins were induced at 6 h postinfection. In comparison, downregulated proteins have more diverse functions. A complete list of all protein identifications as well as altered proteins is provided as Table S2 in the supplemental material.
Distribution of protein fold difference values for intracellular Salmonella at 6 h relative to 1 h postinfection. The gray circle highlights most of upregulated proteins that are associated with iron uptake and utilization.
Salmonella encounters severe metal ion starvation in host epithelial cells and upregulates iron, manganese, and zinc-uptake systems.As discussed above, a sizable fraction of upregulated proteins are involved in iron-acquisition systems, indicating iron limitation in host epithelial cells. Salmonella possesses two types of siderophores, enterobactin and salmochelin (32). We mapped our differential proteomics data onto these pathways, with the extent of induction being color coded (Fig. 2). We found that siderophore synthesis proteins (e.g., EntC) were induced the most for siderophore secretion pathways, whereas outer membrane receptors (e.g., IroN, CirA, and FepA) were elevated the most for uptake pathways. Interestingly, both classes of proteins participate in the very first steps of individual pathways that may be rate limiting. It is also interesting to note that IroC, which is involved in the early steps of salmochelin synthesis, was also highly induced.
Upregulation of Salmonella siderophore secretion and uptake pathways. The extent of protein upregulation (fold difference) is color coded. DHB, 2,5-dihydroxybenzoic acid; DGE, diglucosyl-enterobactin; DBS, 2,3-dihydroxybenzoylserine; OM, outer membrane; CM, cytoplasmic membrane; ND, not determined.
We next sought to determine if Salmonella siderophores were made and present in increasing amounts within infected host cells. We first performed metabolic measurements of Salmonella grown in vitro under iron-replete and iron-depleted (with addition of 280 μM DIP) conditions. Figure 3a shows the extracted-ion chromatograms of various forms of Salmonella siderophores. As expected, both enterobactin and salmochelin were observed with substantially increased levels in vitro when iron levels were limiting. When metabolites from Salmonella-infected epithelial cells were extracted and analyzed again by LC-MS/MS, we detected considerably higher levels of enterobactin and salmochelin as well at 6 h postinfection (Fig. 3b). In addition to the accurate mass afforded by high-resolution MS measurements, MS/MS analyses of their fragments unambiguously confirmed the identities of these siderophores (Fig. 3c).
Metabolomic analyses of Salmonella grown in vitro as well as infected host cells. (a) Extracted-ion chromatograms of four known Salmonella siderophores. Only specific ion signals (m/z values) corresponding to siderophores are shown in these graphs. Metabolites were extracted from Salmonella cultured under iron-sufficient and -deficient conditions and then analyzed by LC-MS/MS. (b) Extracted-ion chromatograms of enterobactin and salmochelin S1. Metabolites were extracted from infected HeLa cells at 1 h and 6 h postinfection and then analyzed by LC-MS/MS. (c) The MS/MS spectrum of enterobactin (m/z = 670.15).
Salmonella also induced transport systems for exogenous siderophores, as both FoxA and Fhu proteins were found at higher levels at 6 h postinfection. Furthermore, siderophore-independent pathways were induced as well. For example, the expression of a ferrous iron transport system (Feo) went up ∼8-fold at 6 h postinfection. To further support the notion that induction of iron-uptake systems is due to iron shortage in host epithelial cells, we carried out additional analyses of the bacterial proteome in vitro under iron-limiting conditions. Compared to iron-replete conditions, we found robust elevation of a nearly complete set of proteins that are associated with iron uptake and/or stress (see Fig. S2a in the supplemental material), including proteins encoded by the ent, iro, fhu, feo, and suf gene clusters as well as the tonB-exbB-exbD system. To facilitate the comparison of proteomic alterations in vitro when iron is limiting to those differences associated with intracellular bacteria, we graphed protein fold changes for these two data sets in the same scatter plot (see Fig. S2b in the supplemental material) and found that most upregulated proteins correlated very well (highlighted by the gray circle). Therefore, most proteomic alterations for Salmonella grown in vitro under iron-deficient conditions recapitulated those changes observed for intracellular bacteria during infection of epithelial cells.
Other than iron uptake, we also found that Salmonella substantially increased its capacity for transporting manganese and zinc ions. An ATP-binding cassette high-affinity manganese transporter, SitABCD, was found to be ∼8-to-9-fold more abundant at 6 h postinfection, although this transporter can also mediate low-affinity iron uptake (33). Furthermore, our proteomic data suggested zinc starvation in epithelial cells and an ABC-type high-affinity Zn transporter was strongly induced, including ZnuA (∼13-fold) and ZnuC (∼3-fold). Taking the results together, we have provided evidence that Salmonella encounters a general shortage of metal ions in host epithelial cells.
Mutational studies of S. Typhimurium iron-uptake systems.Next, we sought to test genetically the importance of iron-acquisition pathways for bacterial growth in vitro and in vivo during infection. As EntC is the very first enzyme responsible for siderophore synthesis (34), deletion of entC would disrupt siderophore-dependent systems (Ent and Iro). Thus, we constructed two deletion mutants, the ΔentC and ΔentC ΔfeoAB mutants. When cultured in iron-sufficient LB broth, the ΔentC ΔfeoAB mutant showed a severe growth defect whereas the ΔentC mutant exhibited only slight growth inhibition (Fig. 4a). When iron levels were limiting, growth of both mutants was strongly inhibited whereas the wild-type strain was affected only slightly (Fig. 4b). Next, we tested if the replication of these mutants is affected within infected epithelial cells. Interestingly, the ΔentC ΔfeoAB mutant was found to be incompetent with respect to invading HeLa cells (Fig. 4c), likely due to its impaired growth/fitness in vitro prior to bacterial invasion. It is also noteworthy that this mutant strain grew poorly on solid agar media, and, during its screening, we had to supplement with “siderophores” from the spent media of wild-type Salmonella that was cultured under iron-limiting conditions. For the invasion-competent ΔentC strain, we enumerated intracellular bacteria at 1 h and 20 h postinfection and found that their replication was indistinguishable from that of the wild-type strain (Fig. 4d), indicating that the Feo system likely provides sufficient iron supply during intracellular growth.
Characterization of Salmonella mutants deficient in major iron-uptake systems. (a and b) Growth curves of wild-type (WT), ΔentC, and ΔentC ΔfeoAB strains in the LB medium without (a) and with (b) 280 μM DIP. (c) Invasion rate of these strains determined by Gentamicin-protection assays (n = 3). The value of the wild-type strain was set to 100%. (d) Fold change in replication of bacteria (wild-type and ΔentC strains) within HeLa cells determined by Gentamicin-protection assays (n = 3).
Induction of multiple S. Typhimurium pathways associated with phosphate utilization during its intracellular growth.Figure 5a shows the heat map of all 61 upregulated proteins at 6 h postinfection, a number of which are notably indicative of phosphate limitation. For example, levels of both UgpQ and UgpB were markedly increased. UgpQ is a glyceryl phosphoryl phosphodiesterase which hydrolyzes diesters during their transport at the cytoplasmic side of the inner membrane (35). Therefore, sn-glycerol-3-phosphate (G3P) is generated and eventually serves as a phosphate source. The upg operon is regulated by a two-component system, PhoBR (35), and an outer membrane esterase, ApeE, has been shown to be PhoBR regulated as well (36). Consistently, we observed a 12-fold increase in ApeE abundance, likely in response to phosphate starvation as well (36). In addition, we found that PhnS, a component of an ABC transporter for aminoethylphosphonate, was ∼8-fold more abundant at 6 h postinfection. The Salmonella phosphonatase pathway is responsible for breaking down phosphonate and thus yielding phosphate if needed. Previously, this pathway was reported to be under PhoB regulon control and inducible by phosphate availability (37). It is also interesting to note that the Pst high-affinity phosphate transport system (PstSB) was moderately (∼1.75-fold to 1.95-fold) induced as well. Importantly, this system was heavily (∼5-fold to 8-fold) induced in response to HBSS, thus suggesting that Salmonella was to some extent primed for host phosphate starvation. Another protein whose induction is likely due to phosphate limitation is PhoN, and it is a nonspecific acid phosphatase regulated by the PhoPQ system (38). Lastly, a putative phosphatase (STM3595) was also heavily (∼13-fold) induced, and its role in phosphate stress has yet to be defined. To further verify the proteomic alterations, we constructed Salmonella strains chromosomally expressing 3×FLAG-tagged proteins and analyzed their expression levels by Western blotting analyses upon infection of HeLa cells. As seen in Fig. 6a and b, the immunoblotting data of both PhoN and STM3595 agreed well with our proteomic results.
Heat maps of Salmonella proteins that differed significantly in expression at 6 h relative to 1 h postinfection. (a) Upregulated proteins. (b) Downregulated proteins. Lanes I to IV represent 4 biological replicates at each sampling time. TCA, tricarboxylic acid cycle; Min, minimum; Max, maximum.
Western blot analyses of selected Salmonella proteins as well as their spectral counts derived from LC-MS/MS measurements. (Top panels) Comparison of protein spectral counts in 1-h and 6-h samples from four independent measurements. (Bottom panels) Representative immunoblots of DnaK and 3×FLAG-tagged PhoN (a), STM3595 (b), PipB (c), and SseL (d).
S. Typhimurium upregulates other classes of proteins important for its intracellular survival and proliferation.We found that PhoP was abundantly expressed at a slightly (∼1.4-fold) increased level at 6 h postinfection. The level of PhoQ was much lower, and we observed more-pronounced induction in three of four replicates. PhoPQ regulates a wide spectrum of proteins important for bacterial virulence, including those encoded on SPI-2 (39). We observed upregulation of several SPI-2 proteins such as SsaQN, consistent with the notion that SPI-2 expression coincides with the onset of bacterial replication (2). We also detected many SPI-2 effectors, though they were not included in our differential analyses due to their rather low spectral counts. Nevertheless, our proteomic data to some extent did suggest induction of these proteins (i.e., PipB and SseL). To corroborate these findings, we measured their abundance by immunoblotting analysis, and both effectors were indeed shown to be upregulated at 6 h postinfection (Fig. 6c and d). In addition, PhoPQ also regulates bacterial resistance to antimicrobial peptides (40). PgtE, an outer membrane protease involved in such resistance (41), was found to be induced substantially (∼8-fold). Another PhoPQ-regulated protein is PagN/STM0306 (∼4-fold-higher induction), which has been shown to function as an adhesion/invasion protein (42, 43). Collectively, our data showed that PhoPQ activated expression of many genes important for Salmonella survival and replication in epithelial cells.
During intracellular replication, Salmonella also exhibits an increased demand for certain amino acids. We found marked increase of STM1269/AroQ levels at 6 h postinfection. The AroF level was also higher, although it did not meet our significance criteria (P = 0.07). Consistently, it has been shown that aromatic amino acid synthesis is required for Salmonella virulence (44). In addition, a periplasmic d-alanyl–d-alanine dipeptidase (PcgL) and an ABC-type transporter for d-alanine (DalS/STM1633) were also upregulated. Finally, we also observed robust (∼8-fold) elevation of the level of biotin synthesis protein BioD. Considering the drastic (>14-fold) induction of BioBD for extracellular Salmonella in HBSS, our data strongly suggested that biotin is scarce in the Salmonella-containing vacuole (SCV).
Downregulated pathways during the early stages of S. Typhimurium intracellular growth.Figure 5b shows the heat map of all 39 downregulated proteins at 6 h postinfection. In contrast to induction of the Ugp system, another major G3P uptake system, Glp, including glycerol kinase GlpK, periplasmic glycerophosphodiester phosphodiesterase GlpQ and anaerobic G3P dehydrogenase GlpA, was markedly repressed. Interestingly, incubation of extracellular Salmonella in HBSS repressed GlpC expression (∼3-fold) as well. The aerobic dehydrogenase GlpD was not significantly altered, though it was abundantly expressed. The level of G3P-phosphate antiporter GlpT was fairly low for intracellular Salmonella. Altered expression was also observed for maltose utilization proteins. The levels of MalEM and LamB were significantly reduced (∼2 to 5-fold) at 6 h postinfection. We also observed moderate (∼2×-lower) downregulation of enzymes involved in the central metabolism such as aconitate hydratase AcnB. Other notable examples of downregulation include bacterial chemotaxis proteins such as Tsr, Aer, STM2314, and STM3216, suggesting that chemotaxis is no longer required for Salmonella residing in the vacuole.
DISCUSSION
S. Typhimurium has the capabilities to invade and replicate within nonphagocytic epithelial cells. To study the adaptation mechanisms of Salmonella during its transition from the extracellular to the intracellular environment, we performed large-scale profiling of the bacterial proteome from infected HeLa cells (a classical model for Salmonella infection of epithelial cells). At 1 h postinfection, internalized bacteria just begin to be located within the SCV (45), and at 3 to 4 h, Salmonella intracellular replication begins (46). We chose to sample bacteria at 6 h postinfection, when S. Typhimurium is thought to be well adapted into the host environment and actively engaged in replication (6). Thus, we compared the Salmonella protein expression levels at 1 h and 6 h postinfection. As controls, we also included bacteria that were cultured in LB broth and those that were further incubated in HBSS. Therefore, the proteomic changes we intended to capture are likely to reflect Salmonella's adaptations to its intracellular niche in host epithelial cells.
Overall, our work significantly advanced quantitative proteomics of intracellular Salmonella, as only 315 bacterial proteins were measured previously (19). With ∼3,300 proteins identified in total, our LC-MS-based approach covered ∼73% of the Salmonella proteome, which contains ∼4,500 proteins. We found that HBSS rapidly induced amino acid biosynthesis pathways. This is not surprising, after all, because HBSS contains minimal inorganic salts in addition to glucose (see details in Materials and Methods). Such observations underscore the importance of considering the impact of the infection medium when one attempts to interpret intracellular proteome data. When intracellular Salmonella was compared at 6 h versus 1 h postinfection, notably, we detected upregulation of nearly all known iron-responsive pathways. Furthermore, proteomic changes in vitro due to iron depletion closely resembled those observed for intracellular Salmonella, thus further supporting the notion of iron starvation in host epithelial cells. The importance of iron in host-pathogen interactions has been well documented (47, 48, 49, 50, 51). Several recent reports also highlighted the significance of iron acquisition during Salmonella infection (mostly through genetic studies) (52, 53, 54).
For bacterial pathogens, iron scavenging is mediated by siderophore-dependent pathways to a great extent. Intriguingly, our expression data suggested that Salmonella has evolved remarkable regulatory mechanisms to ensure the maximum induction of the very first steps of these pathways, which may be rate limiting. Consistent with our proteomic findings, we found substantially increased signals for major Salmonella siderophores as well within host cells. Therefore, we provided compelling evidence on both the protein and metabolite levels that intracellular Salmonella dramatically upregulates iron-scavenging pathways, and these findings significantly extend current understanding of the adaptations of Salmonella in response to host iron shortage.
In addition to iron limitation, we also found evidence of starvation of other metal ions such as Zn2+ and Mn2+. Consistent with our findings, several reports highlighted the importance of the ZnuABC zinc acquisition system during Salmonella infection of the mammalian host (55, 56). Salmonella encodes another zinc transporter, the ZIP family permease ZupT (57), and, interestingly, this protein was not detected in our study. A high-affinity manganese transporter, SitABCD, was also heavily induced, though we failed to detect another major transporter, MntH (58, 59). Given our extensive proteome coverage, MntH and ZupT may be present at extremely low levels or may not be expressed at all. It is well recognized that magnesium levels are low in the SCV (60), and, unfortunately, the MgtBC Mg2+ transporter was measured with rather low abundance that prevents us from getting reliable quantification. Consistent with an overall limitation of metal ions within host cells, intriguingly, we found increased levels of the PspA phage shock protein at 6 h postinfection. In an elegant study, Karlinsey and coworkers demonstrated that PspA can facilitate Mn2+, Zn2+, and Fe2+ transport by maintaining proton motive force (61).
Our comprehensive survey also suggested that intracellular Salmonella might experience severe phosphate shortage in host epithelial cells, an idea which is supported by elevated levels of a high-affinity phosphate transport system, Pst, and of many other proteins (e.g., Ugp) that are involved in phosphate utilization. As one of two major sn-glycerol-3-phosphate (G3P) uptake systems, the ugp operon is often induced when bacteria are starved for inorganic phosphate (62). Interestingly, while the Ugp system was induced in response to phosphate limitation, the other G3P transporter, the Glp system, was significantly repressed at 6 h postinfection. This can be ascribed to some extent to the fact that GlpT is a G3P importer at the price of simultaneously being a phosphate exporter (63). In agreement with this notion, previous studies suggested that Glp-transported G3P can serve as the sole source for both carbon and phosphate, whereas Ugp-transported G3P can serve as the sole source only for phosphate (62, 64). Thus, we believe that Ugp induction is mostly driven by phosphate starvation, whereas the Glp repression is likely due to unavailability of G3P as a carbon source as well as to phosphate limitation. Other repressed proteins include maltose utilization and chemotaxis proteins as well as a few enzymes in the central metabolism pathways.
It is noteworthy that the intracellular Salmonella transcriptome within both infected macrophages (5) and epithelial cells (6) has been reported previously. While our proteome survey shares some consistent findings (i.e., induction of iron uptake, the ugp system, and biotin synthesis), notable differences between protein and mRNA expression data were also found. For instance, we found induction of PhoPQ and many PhoPQ-regulated virulence proteins (e.g., SPI-2, PgtE, PagN, and PhoN) at 6 h postinfection, and yet decreased levels of their transcripts were reported (6). As one of the master regulators of Salmonella virulence gene expression, the PhoPQ two-component system has been shown to promote bacterial survival and replication in the host (65). To confirm and/or dispute our proteomics results, we conducted Western blot analysis of several PhoPQ-regulated proteins and the immunoblotting data in fact correlated well with the spectral counting data that were derived from our LC-MS/MS measurements. In addition, our proteomic data indicated substantial (∼8-fold) elevation of DalS, though a slight decrease of its transcriptional level was found (6). Interestingly, it has been shown in an animal model that DalS was required for Salmonella survival and fitness (66). Discrepancies between mRNA and protein expression data can be explained by minor differences associated with bacterium/cell culturing as well as with sample preparations. Nevertheless, it has become increasingly clear that protein translation can be subject to extensive posttranscriptional regulations that affect protein stability and function and thus that it is advantageous to measure the proteome.
In summary, our comprehensive proteomic analyses revealed a broad spectrum of adaptation mechanisms utilized by Salmonella residing within its unique intracellular niche of host epithelial cells. As mechanisms that govern bacterial virulence as well as intracellular growth may share common features, our findings might be applicable to other bacterial pathogens as well. Importantly, the proteomic strategy we described here would permit systems-level studies of any intracellular bacteria during infection. If combined with omics analyses of their host cells, these high-throughput molecular profiling approaches would allow us to move beyond the reductionist approach and achieve an integrative understanding of host-pathogen interactions.
ACKNOWLEDGMENTS
We thank members of the Liu laboratory for careful review of the manuscript and Feng Shao's laboratory for generously providing us the bacterial strains. We thank Ting Li from Chinese Center for Disease Control and Prevention for his help with bacterial mutant construction as well as for the DnaK antibody. We are also grateful for the help from Yang Cao, Zhirong Shen, and Xiaohui Liu with metabolomic measurements and from Changfa Yin with data analyses.
This work was financially supported by Peking University and by the Thousand Young Talents program from the Chinese government as well as by the National Natural Science Foundation of China (grants 21305006 and 21475005).
FOOTNOTES
- Received 6 November 2014.
- Returned for modification 16 December 2014.
- Accepted 27 April 2015.
- Accepted manuscript posted online 4 May 2015.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.02882-14.
- Copyright © 2015, American Society for Microbiology. All Rights Reserved.
