ABSTRACT
Salmonella enterica elicits intestinal inflammation to gain access to nutrients. One of these nutrients is fructose-asparagine (F-Asn). The availability of F-Asn to Salmonella during infection is dependent upon Salmonella pathogenicity islands 1 and 2, which in turn are required to provoke inflammation. Here, we determined that F-Asn is present in mouse chow at approximately 400 pmol/mg (dry weight). F-Asn is also present in the intestinal tract of germfree mice at 2,700 pmol/mg (dry weight) and in the intestinal tract of conventional mice at 9 to 28 pmol/mg. These findings suggest that the mouse intestinal microbiota consumes F-Asn. We utilized heavy-labeled precursors of F-Asn to monitor its formation in the intestine, in the presence or absence of inflammation, and none was observed. Finally, we determined that some members of the class Clostridia encode F-Asn utilization pathways and that they are eliminated from highly inflamed Salmonella-infected mice. Collectively, our studies identify the source of F-Asn as the diet and that Salmonella-mediated inflammation is required to eliminate competitors and allow the pathogen nearly exclusive access to this nutrient.
INTRODUCTION
Salmonella is among the most burdensome of foodborne pathogens in the United States and globally, causing 100 million illnesses and 50,000 deaths each year (1–7). Salmonella thrives in the host environment by inducing inflammation in order to disrupt the normal microbiota, allowing nutrients to accumulate and to generate electron acceptors for respiration (8–20). One of these nutrients is fructose-asparagine (F-Asn; Fig. 1) (21). F-Asn is an Amadori product generated from glucose and asparagine. Amadori products are formed by the nonenzymatic condensation of the carbonyl group of reducing sugars and the amino group of amino acids and proceed through the formation of a Schiff base that spontaneously rearranges to form a stable ketoamine (22–26). The Amadori product is the first stable compound in the sequence of reactions collectively referred to as the Maillard reaction (22, 27). The ability of Salmonella to utilize F-Asn is encoded by five horizontally acquired genes located between the conserved genes gor and treF (Fig. 1). Four of the genes appear to be in an operon, fraBDAE, while the fifth, fraR, is located upstream and is predicted to encode a transcription regulator of the GntR family. FraE is a periplasmic fructose-asparaginase that releases ammonia from F-Asn to form fructose-aspartate (F-Asp) (28). F-Asp is likely transported to the cytoplasm by FraA, a Dcu-type transporter (anaerobic C4-dicarboxylate), before its phosphorylation by the sugar kinase FraD. The resulting 6-phosphofuctose-aspartate (6-P-F-Asp) is then cleaved by the deglycase FraB, yielding glucose-6-phosphate (glucose-6-P) and aspartate (Fig. 1) (29, 30).
Structure of F-Asn (A) and schematic of the F-Asn utilization pathway as well as the fra locus (B).
A genetic screen revealed that the fra locus is required for Salmonella fitness in a variety of mouse colitis models (21). Interestingly, this locus was not required for fitness in conventional mice, unless those mice had been pretreated with streptomycin to disrupt their intestinal microbiota. Conventional mice are highly resistant to the inflammation induced by Salmonella, but disruption of the microbiota using streptomycin increases susceptibility to Salmonella-mediated inflammation (31–34). Thus, the fra locus was required for fitness only in mice with an inflamed intestinal tract (21), a finding that was further supported by the observation that the fra locus was not required for fitness when the fra mutation was present in a Salmonella strain that could not induce inflammation (a mutant lacking Salmonella pathogenicity island 1 [SPI1] and SPI2 [the ΔSPI1 ΔSPI2 mutant]) (21). Follow-up studies determined that F-Asn is not an essential nutrient during infection of the inflamed gut (there are redundant nutrient sources) but, rather, that F-Asn is toxic to a fraB mutant of Salmonella (30). Mutants lacking other genes of the fra locus or even deletion of the entire fra locus had no detectable effect on Salmonella fitness in mice. FraB is the last enzyme in the F-Asn utilization pathway, and the absence of this enzyme leads to the accumulation of a toxic metabolite, 6-P-F-Asp. Deletion of the entire locus (fraR fraBDAE) does not lead to accumulation of 6-P-F-Asp, because the transporter and kinase are required for formation of 6-P-F-Asp (Fig. 1) (30).
Other Salmonella nutrient sources include 1,2-propanediol, which is derived from the microbiota (35); ethanolamine, which is derived from damaged cells (14); glucarate and galactarate, which are derived from Nos2-mediated oxidation of glucose and galactose (9); and l-lactate, which is derived from altered host metabolism (36). Here, we determined that F-Asn is derived from the diet (mouse chow) and that the fitness defect of the fraB mutant is related to the availability of F-Asn in the intestinal tract. When F-Asn is available, the fraB mutant is inhibited. We found that F-Asn is not depleted in germfree mice but is consumed by the microbiota in conventional mice. We previously identified several members of the Clostridia class that utilize F-Asn (37). Here, we found that these organisms are eliminated from the intestinal tract during Salmonella-mediated inflammation. Thus, inflammation is required for Salmonella to eliminate competing consumers and gain nearly exclusive access to this nutrient.
RESULTS
F-Asn is derived from the diet rather than the microbiota.A fraB mutant of Salmonella is dramatically attenuated in germfree mice and in other mouse models that are susceptible to Salmonella-induced inflammation (21). The fraB mutant is attenuated due to the accumulation of a toxic metabolic intermediate, 6-P-F-Asp, during utilization of F-Asn as a carbon or nitrogen source (30). However, the fraB mutant is not attenuated in conventional mice. These results suggest that there is F-Asn within the intestinal tract of germfree mice but not in that of conventional mice, where other organisms presumably consume F-Asn and decrease F-Asn availability to Salmonella. To test this hypothesis, we used mass spectrometry (MS) to measure the F-Asn concentration in the cecal contents of germfree and conventional Swiss Webster mice (Fig. 2). F-Asn was detected in the intestines of germfree mice (2,700 pmol/mg), suggesting that mice cannot alter or metabolize the F-Asn that is within the gastrointestinal tract. This concentration is higher than what was observed in the mouse chow (400 pmol/mg), suggesting that either F-Asn may accumulate or F-Asn can form in the mouse gastrointestinal tract (see below). The concentration of F-Asn is much lower (8 pmol/mg) in the ceca of germfree mice infected with Salmonella, consistent with the ability of Salmonella to consume and utilize F-Asn. The concentration of F-Asn was also very low in the cecum of conventional mice (9 pmol/mg), consistent with the hypothesis that some member(s) of the normal microbiota can utilize F-Asn. These measurements were repeated with a slightly different MS protocol and different samples, and similar results were observed (see Fig. S1 in the supplemental material).
Presence of F-Asn in mouse chow, conventional mouse cecum, germfree mouse cecum, and germfree mouse cecum infected with Salmonella. Each data point represents a mouse chow pellet (LabDiet recipe 5066) (3 data points) or a Swiss Webster mouse cecum at 1 day postinfection (5 data points each), with the geometric mean indicated by a horizontal line and error bars representing 95% confidence intervals. The limit of detection (4.5 pmol/mg) is indicated by the dotted line.
F-Asn does not form in the mouse intestinal tract.The F-Asn concentration in the ceca of germfree mice is higher than that in mouse chow (Fig. 2 and S1). We hypothesized that either F-Asn originated in the mouse chow and accumulated in the cecum or was formed de novo in the mouse gastrointestinal tract. To test the latter hypothesis, we provided five germfree mice with drinking water containing labeled precursors of F-Asn (15N-labeled asparagine and 13C-labeled glucose) and then used MS to measure the labeled F-Asn in the cecal contents. No labeled F-Asn was detected in any of the samples. We then hypothesized that de novo formation may be dependent upon Salmonella-mediated inflammation. To test this hypothesis, we used three groups of five mice each, all of which were provided with labeled 15N-labeled asparagine and 13C-labeled glucose in their drinking water, as described above, and infected each of the groups with a different Salmonella strain. The first group received strain ASD215 (14028 Δfra80 ΔansB80::Kan). This strain was used to eliminate the possibility of Salmonella consuming any newly formed F-Asn, as it lacks the fra locus and also lacks ansB, a periplasmic asparaginase that can convert F-Asn to F-Asp (28). Another group was infected with strain ASD203 (14028 ΔSPI1 ΔSPI2 Δfra4 ΔansB80::Kan). This strain is similar to ASD215, except that it lacks SPI1 and SPI2 and thus cannot initiate inflammation (38). A third group was infected with Salmonella strain HMB175 (14028 ΔansB80::Kan), which can consume F-Asn and can initiate inflammation. No labeled F-Asn was detected in any of the samples, suggesting that F-Asn is not formed within the murine gastrointestinal tract, regardless of whether the gastrointestinal tract is inflamed or not by Salmonella (MS data from one mouse infected with ASD215 are shown in Fig. S2; data for the rest of the mice are not shown).
Competitors for F-Asn are eliminated by high levels of inflammation.The gastrointestinal tracts of many mouse strains fail to become inflamed upon infection with Salmonella. The resistance to Salmonella-mediated inflammation appears to be due to the normal microbiota, since germfree mice are susceptible to Salmonella-mediated inflammation. Disruption of the microbiota by treatment with streptomycin or other antibiotics (31–34) increases susceptibility to Salmonella-mediated inflammation. However, the use of antibiotic-treated or germfree mice is not ideal for studying Salmonella-mediated inflammation and its effects on the microbiota. Therefore, we sought an alternative.
The CBA/J strain of mouse is unusual in that it allows Salmonella to persistently colonize the intestinal tract for a long period of time, which leads to inflammation after 8 to 14 days of infection (13, 39). We have previously used this model to study the microbiome disturbances caused by Salmonella (40). Here, we leveraged these earlier collected samples to quantify the F-Asn concentration in the ceca after 16 days and tested the hypothesis that disruption of the microbial community (by either inflammation or antibiotic treatment) causes F-Asn to accumulate in the intestinal tract. Such an expectation is based on the idea that inflammation eliminates members of the microbiota that would otherwise consume F-Asn (Fig. 3). Previously, we reported a relationship between Salmonella relative abundance and inflammation levels and identified two distinct Salmonella responses (a high response, in which there is a >46% Salmonella relative abundance in the cecum, and a low response, in which there is a <7% Salmonella relative abundance in the cecum) in mice treated with the same Salmonella inoculum (40). Inflammation, measured using lipocalin-2 as a marker (9, 41–44), correlated with the Salmonella response; i.e., inflammation was high when the Salmonella response was high. The F-Asn concentration was relatively low in all samples compared to that in the germfree mice (Fig. 2), but it was the highest in the highly inflamed samples (2,700 pmol/mg F-Asn in germfree mice compared to 100 pmol/mg in the high-Salmonella-responder samples) (Fig. 3). The lipocalin-2 concentrations observed in streptomycin-treated mice were not different than those observed in untreated control mice, indicating that the mice did not become inflamed from this treatment or that they had recovered by day 15. In fact, recovery by day 4 has been noted previously (36). The lipocalin-2 concentrations observed in dextran sulfate sodium (DSS)-treated mice were statistically significantly higher than those observed in the controls, but the practical significance was low (1 ng/g versus 3 ng/g). Consistent with the lack of inflammation, the F-Asn concentrations in these groups were not statistically significantly different (28 pmol/mg in control mice, 24 pmol/mg in streptomycin-treated mice, and 37 pmol/mg in DSS-treated mice; Fig. 3).
F-Asn and lipocalin-2 concentrations in CBA/J mice. The F-Asn concentration in each individual mouse cecum at day 16 postinfection is indicated by a bar, with the color denoting the treatment. Mice infected with the same amount of Salmonella are denoted as high and low responders, with high responders having a >46% Salmonella relative abundance and low responders having a <7% Salmonella relative abundance (40). The lipocalin-2 concentration from feces collected on day 15 postinfection is indicated by a black triangle on each bar. The F-Asn concentrations in the high-Salmonella-responder group but no other groups were statistically significantly different than those in the control group (Mann-Whitney test, P = 0.0357). The lipocalin-2 concentrations in the DSS-treated, high-Salmonella-responder, and low-Salmonella-responder groups were all statistically significantly different than those in the control group (Mann-Whitney test, P = 0.0159, 0.0357, and 0.0025, respectively). The result for the high-Salmonella-responder group was statistically significantly different than that for all other groups (Mann-Whitney test, P < 0.03).
Metagenomics analysis of one highly inflamed Salmonella-infected mouse (1,130 ng/g feces of lipocalin-2) and three uninfected control mice (76 ± 17 ng/g feces of lipocalin-2) showed that the number of distinct fraBD homologs was greatly decreased in the inflamed mouse. The three control fecal samples contained between 13 and 18 fraB and fraD homolog pairs, with each pair colocalized on the same contig. The majority of these contigs belonged to the Firmicutes (93% ± 1%), with the exception of an actinobacterial Collinsella contig that was detected in each of the control metagenomes. In the Salmonella-infected metagenome, only the Salmonella fraBD was detected, indicating that all other F-Asn consumers were eliminated from the Salmonella-infected gut. However, 70% of the reads in this sample were Salmonella reads. To account for differences in the underlying community richness or sequencing depth, the recovered fraBD homologs were normalized to the S3 ribosomal gene number. The relative abundance of non-Salmonella fraBD-containing organisms was 23% of the community in the noninflamed controls, whereas it was 0% in the Salmonella-infected sample (Fig. 4). This observation demonstrates that the ratio of putative F-Asn consumers to nonconsumers decreased in the Salmonella-infected sample compared to the noninflamed control sample.
fraBD homologs are eliminated during Salmonella-mediated inflammation. The ratio of fraBD homologs to S3 ribosomal protein genes detected in CBA/J mouse fecal metagenomes at day 11 postinfection is plotted on the left y axis (green). Measurements of the lipocalin-2 concentration in the feces of the same mice are plotted on the right y axis (orange).
Our metagenomics analyses revealed that Salmonella infection led to decreased fraBD gene diversity, presumably due to elimination of other F-Asn-consuming competitors. Because the cost-intensive metagenomics studies were performed with a single highly inflamed Salmonella-infected mouse and three control mice, we sought an alternative method where we could examine multiple samples in parallel for the presence or absence of a target sequence and thereby cross-validate the findings from sequencing of the mouse fecal DNA. Specifically, we examined if any member of the Clostridia, whose select representatives we recently validated catabolize F-Asn (37), decreased upon Salmonella infection. To fulfill this objective, we chose quantitative PCR (qPCR) as the method of choice.
We first identified a 21-nucleotide (nt) sequence present in the fraB gene of Clostridium sp. strain MGS:81 but not in Salmonella fraB. This sequence was used to design our qPCR probe. Moreover, we generated a 144-bp amplicon that encompasses the probe sequence and used it as our standard. We generated a standard curve using a 10-fold dilution series ranging from 10 pg to 1 fg of the clostridial fraB 144-bp amplicon (Fig. 5A and B). The threshold cycle (CT) values ranged from 13 to 29 and led to a standard curve with a slope of −3.93 (∼80% efficiency) and an R2 value of 0.999. This standard curve was then used to determine the copy numbers in the genomic DNA (gDNA) samples from different mice.
qPCR analysis of gDNA samples isolated from the cecum of control, streptomycin-treated, DSS-treated, and low- and high-Salmonella-responder CBA/J mice at day 16 postinfection. (A) Amplification plots that indicate the change in normalized fluorescence (ΔRn) for the five indicated amounts of the 144-bp standard. (B) Standard curve of the threshold cycle (CT) versus the copy number obtained using nine technical replicates for each standard. Error bars represent standard deviations of the mean values calculated using the data from nine replicates. (C) Amplification plots that indicate the ΔRn for the indicated mouse gDNA samples together with those obtained for the 1-fg standard and negative control. (D) Clostridium sp. MGS:81 copy number calculated using the CT values determined in the assay whose results are shown in panel C. Error bars represent standard deviations of the respective mean values calculated from triplicate measurements for each mouse. Neg. cont., negative control; Strep., streptomycin; Sal., Salmonella.
We used qPCR assays to determine the Clostridium sp. MGS:81 copy number in five control, five DSS-treated, five streptomycin-treated, seven low-Salmonella-responder, and three high-Salmonella-responder mice (Fig. 5C). The copy numbers were calculated for all mice using the triplicate-averaged CT value from each and a nine-way averaged standard curve (Fig. 5B). After removal of outliers using Grubbs' tests (one DSS-treated mouse and one low-Salmonella-responder mouse), the average number of Clostridium sp. MGS:81 copies was calculated for each mouse type (Fig. 5D). Two-sample t tests (assuming unequal variances, with α = 0.05) showed that the control, streptomycin-treated, DSS-treated, and low-Salmonella-responder mice each contained Clostridium sp. MGS:81 copy numbers statistically significantly different than those in the high-Salmonella-responder mice, but all other relationships between groups were not statistically significant. There was a striking decrease, if not complete elimination, of the clostridial representative in the high Salmonella-responder mice (Fig. 5D).
DISCUSSION
Salmonella is one of the most significant foodborne pathogens affecting humans and livestock (1–7). Previously, we utilized a genetic screen to identify Salmonella genes required for optimal fitness in the gastrointestinal tract of germfree mice (21). The fra locus was identified in this screen and then further demonstrated to be extremely important for Salmonella fitness in several other mouse models (21). The fitness defect is observed only with fraB mutants and not with mutants lacking other genes of the fra locus (30). The fitness defect is not caused by a lack of F-Asn acquisition but, rather, is caused by the accumulation of a toxic metabolic intermediate, 6-P-F-Asp (30). The fraB fitness defect was observed only in mouse models that are susceptible to inflammation and was abrogated by additional mutations in the genes required for Salmonella to cause inflammation, the genes encoding SPI1 and SPI2 (21). Thus, the Salmonella fraB mutant has a fitness disadvantage only during gastrointestinal inflammation, which, we propose here, is simply due to the availability of F-Asn. Most mouse strains are resistant to Salmonella-mediated inflammation and become susceptible only after a perturbation of the microbiota, most commonly achieved with streptomycin (33, 34, 45, 46). Germfree mice are also susceptible (34). In this study, we measured the F-Asn concentration in a variety of mice and found that the concentrations are consistent with fraB phenotypes (a fraB mutant has a fitness defect only when F-Asn is present). In germfree Swiss Webster mice, F-Asn is abundant (2,700 ± 826 pmol/mg) (Fig. 2; see also Fig. S1 in the supplemental material), and a Salmonella fraB mutant is defective compared to the wild type in these mice (21). In conventional Swiss Webster mice, F-Asn is not abundant (9 ± 4 pmol/mg) (Fig. 2 and S1) and the fraB mutant is not defective compared to the wild type (21, 30).
The concentration of F-Asn is high in germfree mice and low in conventional mice, presumably because members of the microbiota consume F-Asn. We have determined that some members of the Clostridia class carry F-Asn utilization gene homologs (fraB, fraD, and fraE) and can consume F-Asn (37). If these members of the microbiota are drastically reduced in number during inflammation, then F-Asn would become available, and this availability would explain how the Salmonella fraB mutant becomes inhibited during inflammation. CBA/J is an unusual strain of mouse that can become inflamed by Salmonella without the need for a germfree state or antibiotic perturbation of the normal microbiota (13, 39). The CBA/J mouse allows a persistent colonization of the gastrointestinal tract that eventually leads to inflammation after 8 to 10 days. This is slow compared with the time to inflammation in antibiotic-treated mice (1 day), but the absence of antibiotics allows a more relevant study of the effects of Salmonella on the gut microbiota (13, 39, 40).
Using this CBA/J model, we determined that the inflammation achieved by Salmonella is quite variable but that the F-Asn concentration was the highest in the most highly inflamed mice (Fig. 3). The F-Asn concentration in these highly inflamed CBA/J mice was 113 ± 32 pmol/mg, which is much lower than the 2,700 pmol/mg present in germfree mice (presumably due to Salmonella consuming the F-Asn as it becomes available) but is higher than what is available in the uninfected conventional Swiss Webster (9 ± 4 pmol/mg) or CBA/J (28 ± 16 pmol/mg) mice. Since a fraB mutant is not inhibited in conventional mice but is inhibited in highly inflamed mice, the concentration of F-Asn required to inhibit the fraB mutant in vivo is therefore likely to be between 28 and 113 pmol/mg (or 28 and 113 μM, respectively). The 50% and 90% inhibitory concentrations (IC50 and IC90) of F-Asn for a Salmonella fraB mutant grown in vitro are 19 μM and 174 μM, respectively (30). Therefore, even with wild-type Salmonella consuming F-Asn, the F-Asn concentration (113 ± 32 pmol/mg [dry weight]) in the ceca of these highly inflamed CBA/J mice appears to be in a range that would be expected to inhibit a Salmonella fraB mutant.
While Salmonella-mediated inflammation led to increased F-Asn concentrations in CBA/J mice, we were curious if streptomycin or DSS treatment would give the same result. Streptomycin might negatively impact normal microbiota populations that consume F-Asn, leading to increasing concentrations of F-Asn. Similarly, DSS might inflame the gastrointestinal tract and decrease the populations of normal microbiota that consume F-Asn. However, streptomycin treatment did not cause lipocalin-2 or F-Asn concentrations to rise compared to those in untreated mice, suggesting that the F-Asn consumers are not susceptible to streptomycin, that the streptomycin did not kill a sufficient proportion of the consumers, or that the consumers had simply repopulated the intestine by the time that we took the samples (Fig. 3). Since the streptomycin treatment was a one-time administration and our measurements were 15 or 16 days later, repopulation is the most likely explanation (36). DSS treatment also failed to increase the lipocalin-2 or F-Asn concentrations. DSS was administered throughout the experiment, and the dose was high, so it is not clear why the CBA/J mice failed to become more inflamed. It has previously been noted that CBA mice predominantly become inflamed in the distal colon rather than the proximal colon, although the cecum was not tested (47). Also, while CBA mice were not tested, it has been determined that there is a wide range of responsiveness of mice to DSS (48).
Metagenomic analysis of the microbial community revealed that Clostridia carrying F-Asn utilization genes were abundant in the three uninfected control mice but absent from a highly inflamed Salmonella-infected mouse (Fig. 4). The metagenomic data from a single highly inflamed mouse showed that all species that carry fraBD homologs are eliminated. This change in the microbial landscape was confirmed for a single species of Clostridium using qPCR with all 25 of the mice, including the three highly inflamed Salmonella-infected mice (Fig. 3). This particular Clostridium species was essentially eliminated from all of the mice with a high Salmonella abundance, while its abundance in mice with a low Salmonella abundance or in the streptomycin- or DSS-treated mice was not different from that in the control mice (Fig. 5). This elimination of competitors would be expected to make F-Asn available to Salmonella. The availability of F-Asn can be determined most readily using a fraB mutant of Salmonella that is strongly inhibited by F-Asn. In all mouse models of inflammation that we have examined, the fraB mutant is inhibited, while in mice that are not inflamed, the fraB mutant is not inhibited (21, 30). Furthermore, the inhibition of the fraB mutant is absent when using a Salmonella genetic background lacking SPI1 and SPI2, which are required for Salmonella-mediated inflammation. These phenotypes of the Salmonella fraB mutant combined with the direct measurements of F-Asn performed here, both in mouse chow and in mouse intestines, provide strong evidence that the source of F-Asn is the diet (mouse chow) and that it becomes available to Salmonella upon inflammation-mediated elimination of competitors within the microbiota, predominantly of the Clostridia class.
In the original genetic screen in which we identified the fra locus, the fraB mutant was more attenuated in germfree mice monocolonized with Enterobacter cloacae than it was in completely germfree mice (21). Since Enterobacter cloacae does not utilize F-Asn, we hypothesize that E. cloacae removed many of the nutrient sources that could be utilized by Salmonella, leaving a relatively high proportion of F-Asn. This same phenomenon is seen using in vitro toxicity assays in which medium containing F-Asn but depleted of other nutrients by the growth of other organisms is more toxic to the Salmonella fraB mutant than medium that has not been previously depleted by other organisms (37).
We have recently confirmed that Clostridium acetobutylicum, C. bolteae, and C. clostridioforme can utilize F-Asn during growth in vitro (37). We also identified numerous human foods that contain F-Asn (49). There are some foods that have as much F-Asn as mouse chow (400 pmol/mg or higher), including fresh apricots, lettuce, asparagus, and canned peaches, but far higher concentrations (11,000 to 35,000 pmol/mg) are found in heat-dried apricots, heat-dried apples, and heat-dried asparagus (49). Since the fra locus is far older than the human invention of heat drying of foods, the moderate concentrations of F-Asn present in fresh apricots, asparagus, etc. (at levels mirroring those in mouse chow), are likely responsible for Salmonella's maintenance of the fra locus. It is not known if the presumably modern inventions of heat-dried fruits and vegetables can alter a Salmonella infection. Regular consumption of dried apricots may create a microbial community that is excellent at consuming F-Asn and actually increase resistance to Salmonella. On the other hand, a single meal of dried apricots by a person that does not normally consume them may temporarily increase susceptibility to Salmonella. These hypotheses remain to be tested.
MATERIALS AND METHODS
Bacterial strains and growth.Salmonella enterica serovar Typhimurium strains were used throughout. The wild-type strain in Fig. S1 in the supplemental material is strain MA43 (IR715 phoN::aadA), and the fraB mutant is strain MA59 (IR715 fraB1::Kan) (21). IR715 is an ATCC 14028 derivative resistant to nalidixic acid (50). In all other figures, the wild type is strain ATCC 14028. HMB175 is ATCC 14028 ΔansB80::Kan (28). ASD215 is ATCC 14028 Δfra80 ΔansB80::Kan. It was constructed by using phage P22HTint to transduce the ΔansB80::Kan allele from HMB107 (28) into HMB215 (30). ASD203 is ATCC 14028 Δ(avrA-invH)1 Δ(ssrB-ssaU)1 Δ(fraR-fraBDAE)4 ΔansB80::Kan. It was constructed by using phage P22HTint to transduce the ΔansB80::Kan allele from HMB107 (28) into ASD202. ASD202 was constructed by using the FLP recombinase encoded on pCP20 to remove the kanamycin resistance cassette (Kan) from the fra island of strain ASD201 (38). All strains were grown overnight with shaking at 37°C in LB broth (Fisher Bioreagents).
Animal experiments.All animal work was performed using protocols approved by The Ohio State University Institutional Animal Care and Use Committee (IACUC; OSU 2009A0035). Conventional Swiss Webster mice were obtained from Taconic. The germfree mice were Swiss Webster mice obtained from the breeding colony at The Ohio State University. They were of mixed sex and of variable age (range, 6 weeks to 6 months old). They were fed irradiated chow recipe 5066 (LabDiet). Germfree mice were inoculated with Salmonella overnight cultures that had been centrifuged at 10,000 × g and resuspended in water. The bacterial suspension was diluted to 105 CFU/ml in water and administered to mice by oral gavage with 100 μl (104 CFU total). At 1 day postinfection, the mice were euthanized for organ harvest.
The CBA/J mouse samples used in the assays whose results are shown in Fig. 3 and 5 were taken from a study published elsewhere (40). The lipocalin-2 results from these mice were published previously (40), but the F-Asn and qPCR data are unique to this report. New CBA/J mice were obtained for the metagenomic studies. All CBA/J mice were obtained from The Jackson Laboratory (Bar Harbor, ME), were females from 6 to 10 weeks old, and were fed chow recipe 7912 (Teklad). They were inoculated orally with Salmonella as described above for the germfree mice, except that they each received 109 CFU total. Metagenomic libraries were prepared from day 11 fecal samples (see below), and the lipocalin-2 concentration was measured in fecal pellets from day 15 postinfection (40). At 16 days postinfection, the mice were euthanized for organ harvest and the F-Asn concentration in the cecal contents was determined as described below.
Synthesis of fructose-asparagine and heavy fructose-asparagine.Fructose-asparagine was synthesized as described previously (23). 13C-labeled fructose-asparagine was made with the same protocol but using glucose uniformly labeled with 13C (Cambridge Isotope Laboratories).
LC-MS.To measure the F-Asn concentrations from mouse ceca, the ceca were cut open and the contents were gently scraped from the inner surface and rinsed with water. The cecal contents were then stored at −80°C until further use. Prior to analysis, the cecal contents were lyophilized, weighed, and ground on dry ice with a pestle that fit 1.5-ml microcentrifuge tubes. Mouse chow (recipe 5066; LabDiet) samples were also lyophilized and ground on dry ice with a pestle in a 1.5-ml tube. For all sample types, approximately 10 mg of dry material was then transferred to a new preweighed 1.5-ml tube and weighed, followed by the addition of 500 μl chilled methanol and 500 μl H2O spiked with 0.16 nmol [13C]F-Asn. After being vortexed and centrifuged at 14,800 × g for 1 h, the supernatant was transferred into a new 1.5-ml centrifuge tube, frozen, and lyophilized. Before MS analysis, these dried pellets were resuspended in 500 μl acetonitrile-water (80%:20%) with 0.1% (vol/vol) formic acid (liquid chromatography [LC]-MS grade; Thermo Scientific) and filtered through a 0.2-μm-pore-size polytetrafluoroethylene filter (Thermo Scientific). The flowthrough was analyzed by LC coupled to MS. A nano-Acquity ultraperformance LC (UPLC) system (Waters, Milford, MA, USA) with a UPLC M-class BEH 130 amide column (75 μm by 100 mm; particle size, 1.7 μm; Waters) was coupled to a triple-quadrupole mass spectrometer (Xevo TQ-S; Waters) for F-Asn quantification. Buffer A (0.1% formic acid [FA] in water with 10% acetonitrile) and buffer B (0.1% FA in acetonitrile) were used as mobile phases for gradient separation, which started with 80% buffer B for 6 min at a flow rate of 0.5 μl/min and was then followed by the following gradient: from 6 to 20 min, 80 to 50% buffer B; from 20 to 26 min, 50% buffer B; from 26 to 28 min, 50 to 80% buffer B; from 28 to 35 min, 80% buffer B. The mass spectrometer was operated in positive-ion nanoelectrospray ionization mode (nano-ESI+) with a capillary voltage of 3.5 kV, a source temperature of 70°C, a cone voltage of 2 V, and a source offset of 2 V. The gas flow rate for the collision cell was 0.15 ml/min. While transition m/z 295 → 211 of F-Asn with a collision energy of 13 eV was selected for quantitation, m/z 301 → 216 of [13C]F-Asn with a collision energy of 13 eV was used for normalization. While m/z 295 → 211 was used as the quantifier, transitions m/z 295 → 277 and m/z 295 → 259 were used as qualifiers to verify the presence of F-Asn. Skyline-daily (v3.5; M. J. MacCoss lab, Department of Genome Sciences, University of Washington, Seattle, WA, USA) was used for calculating the peak area of transitions.
A Q Exactive mass spectrometer (Thermo Fisher Scientific) with a Nanospray Flex ion source coupled with the same LC system described above was used for detecting the de novo formation of F-Asn. The LC settings were the same as those described above, and the mass spectrometer was operated in positive mode with a capillary voltage at 1.68 kV, a source temperature at 350°C, and the S-lens level at 50. A full mass scan (m/z 50 to 750) was performed at a resolution of 35,000 with an automatic gain control (AGC) target at 3e6 and a maximum injection time (IT) at 200 ms. Thirteen ions with m/z from 295.1 to 307.1, which differed by 1 Da in between, were selected for parallel reaction monitoring (PRM) scan with a resolution at 17,500, an automatic gain control (AGC) target at 1e6, a maximum injection time (IT) at 100 ms, an isolation window at 0.7 m/z, and a stepped collision energy of 20 eV.
Metagenomic profiling of FraBD from CBA/J mice.Total nucleic acids were extracted from mouse fecal samples (day 11 postinfection) using a PowerSoil DNA isolation kit (MoBio), eluted in 100 μl of the elution buffer provided, and stored at −20°C until sequencing. DNA was submitted for sequencing to the Genomics Shared Resource facility at The Ohio State University. Libraries were prepared with the Nextera XT library system in accordance with methods described previously (51, 52). All analysis methods and accompanying scripts from assembly to gene annotation are provided in Github (https://github.com/TheWrightonLab). Known FraB and FraD amino acid sequences for Salmonella enterica LT2 were used to recover homologs (E value < 1e−20) from each metagenome using the NCBI BLASTP program (53). The ratio of the recovered FraBD homologs to single-copy S3 ribosomal proteins (housekeeping gene) was used to estimate the relative abundance of F-Asn consumers in each sampled microbial community. This normalization was performed to account for significant differences in the richness of the control and Salmonella-infected microbial communities, as previously reported (40). FASTA files of the recovered FraBD amino acid sequences are listed in the Data File S1 in the supplemental material. We mined metagenomes from 4 CBA/J mouse metagenomes (NCBI BioProject PRJNA348350).
qPCR analysis.We used Jalview (54) to align the FASTA files of all fraBD sequences present in the metagenomic DNA libraries of fecal samples from control mice and absent in the highly inflamed Salmonella-infected mouse. This alignment, in conjunction with the Salmonella fraB sequence, helped identify a 21-nt sequence present in the fraB gene of Clostridium sp. MGS:81 and notably absent in Salmonella fraB. We exploited this distinctive feature and designed a qPCR probe to investigate the change in copy number of Clostridium sp. MGS:81 upon Salmonella infection, as well as DSS and streptomycin treatment.
Using 5′-AAGAAGGCTAAGGAGAAG-3′ and 5′-GCTGTTGTAATGGATCAG-3′ as the fraB forward and reverse primers (Integrated DNA Technologies [IDT], Coralville, IA), respectively, we first obtained a 144-bp PCR amplicon corresponding to the target DNA. This PCR was performed using PrimeSTAR GXL DNA polymerase (Clontech, Mountain View, CA) and control mouse cecum-derived genomic DNA as the template. The 144-bp amplicon was sequenced to ascertain its bona fides and validate the specificity of the primers used. Subsequently, a gel-purified preparation of this validated 144-bp amplicon was employed as the standard in our qPCR assays.
All qPCR assays were performed in triplicate using an Applied Biosystems StepOne 48-well instrument and the PerfeCTa qPCR ToughMix (QuantaBio, Beverly, MA) with carboxy-X-rhodamine (ROX) as a passive reference dye. The qPCR probe (IDT) used was 5′-GCGGATGAGTGGTATCAGG-3′ and had a fluor/quencher system consisting of 6-carboxyfluorescein (6-FAM) at the 5′ end, an internal ZEN dark quencher, and an Iowa Black FQ quencher at the 3′ end. Twenty-microliter qPCR mixtures contained 10 ng of cecal contents genomic DNA, 900 nM fraB forward and reverse primers, and 250 nM probe (final concentrations). All qPCRs were initiated with a 95°C hold (2 min), followed by 40 cycles of 95°C (3 s), 47°C (15 s), and 68°C (20 s). Three separate runs were conducted to evaluate cecal DNA samples from 25 different mice. Each run contained its own set of negative controls as well as three technical replicates of standard DNA solutions with amounts ranging from 10 pg to 1 fg. The StepOne instrument software determined the threshold normalized fluorescence (ΔRn) values and calculated the CT values. One threshold ΔRn value was computed for each triplicate set of gDNA samples from each mouse and, likewise, for each set of standards ranging from 1 fg to 10 pg. The threshold values ranged from 0.018 to 0.47 ΔRn units.
ACKNOWLEDGMENTS
We thank Edward J. Behrman and Alex Bogard for providing fructose-asparagine.
This work was funded by grant number 1R01AI116119 from the National Institutes of Health (to V.G., K.C.W., V.H.W., and B.M.M.A.) and a seed grant from The Ohio State University Foods for Health-Food Innovation Center (to K.C.W., V.H.W., and B.M.M.A.). Part of the research reported here was supported by the NIH National Institute of Allergy and Infectious Diseases (Y1-AI-8401) to J.N.A. and utilized capabilities developed through support from the NIH National Institute of General Medical Sciences (GM094623). Part of the work was performed in the Environmental Molecular Sciences Laboratory, a national scientific user facility located on the campus of Pacific Northwest National Laboratory (PNNL) in Richland, WA, and supported by the U.S. Department of Energy Office of Biological and Environmental Research. Battelle operates PNNL for the DOE under contract no. DE-AC05-76RLO01830.
FOOTNOTES
- Received 3 January 2018.
- Accepted 20 February 2018.
- Accepted manuscript posted online 26 February 2018.
Supplemental material for this article may be found at https://doi.org/10.1128/IAI.00945-17.
- Copyright © 2018 American Society for Microbiology.
