The Norepinephrine Metabolite 3,4-Dihydroxymandelic Acid Is Produced by the Commensal Microbiota and Promotes Chemotaxis and Virulence Gene Expression in Enterohemorrhagic Escherichia coli

ABSTRACT

Enterohemorrhagic Escherichia coli (EHEC) is a commonly occurring foodborne pathogen responsible for numerous multistate outbreaks in the United States. It is known to infect the host gastrointestinal tract, specifically, in locations associated with lymphoid tissue. These niches serve as sources of enteric neurotransmitters, such as epinephrine and norepinephrine, that are known to increase virulence in several pathogens, including enterohemorrhagic E. coli. The mechanisms that allow pathogens to target these niches are poorly understood. We previously reported that 3,4-dihydroxymandelic acid (DHMA), a metabolite of norepinephrine produced by E. coli, is a chemoattractant for the nonpathogenic E. coli RP437 strain. Here we report that DHMA is also a chemoattractant for EHEC. In addition, DHMA induces the expression of EHEC virulence genes and increases attachment to intestinal epithelial cells in vitro in a QseC-dependent manner. We also show that DHMA is present in murine gut fecal contents and that its production requires the presence of the commensal microbiota. On the basis of its ability to both attract and induce virulence gene expression in EHEC, we propose that DHMA acts as a molecular beacon to target pathogens to their preferred sites of infection in vivo.

INTRODUCTION

Foodborne pathogens have long been known to sense and exploit a multitude of host signals (1). Such interkingdom signaling outcomes range from enhanced bacterial survival to enhanced virulence. For example, pathogens utilize ethanolamine from host epithelium to gain a competitive advantage over resident microbiota (2). Similarly, pathogens induce virulence genes upon the sensing of host neurotransmitters. In fact, the role of catecholamine neurotransmitters in enhancing bacterial infections has been recognized for over 60 years (3). Norepinephrine (NE), the main neurotransmitter of the sympathetic nervous system (4), is known to increase the virulence of numerous pathogens, including Campylobacter, Salmonella, and enterohemorrhagic Escherichia coli (EHEC) (5–9). NE is abundant in gut-associated lymphoid tissues (GALTs) (10), which are the preferred targets for initial infection by most intestinal pathogens (11–16). Therefore, host signals indicative of GALTs, such as NE, are potential signals for pathogens to identify their favored sites for infection.

Bacteria detect NE and other catecholamine neurotransmitters through the QseBC two-component system (TCS) (17). In pathogens, QseBC are required for the catecholamine-induced expression of virulence genes (18). We hypothesized that, in addition to inducing virulence, NE may serve as a chemotaxis signal for pathogens to migrate to GALTs. Our previous data indicated that pathogenic and nonpathogenic E. coli strains are attracted to NE (19). However, we recently found that the actual attractant for a nonpathogenic E. coli K-12 strain is not NE but 3,4-dihydroxymandelic acid (DHMA), a metabolite of NE (20). NE is metabolized to DHMA through the action of the bacterial enzymes TynA and FeaB, which are induced in a QseC-dependent manner (20). DHMA is sensed by the Tsr chemoreceptor of strain RP437 at concentrations as low as 5 nM (20), which makes it an extraordinarily potent chemoattractant.

In this study, we investigated the role of DHMA as a signaling factor for pathogens, using EHEC O157:H7 (strain 86-24) as a model organism. EHEC O157:H7, also known as Shiga toxin-producing E. coli (STEC) or Verotoxin-producing E. coli (VTEC), is the causative agent in multiple widespread outbreaks every year, and the resulting infections are estimated to cost $1.9 billion annually in the United States alone (21). Here, we report that DHMA is a chemoattractant for EHEC. In addition, we show that DHMA induces expression of virulence genes and attachment of EHEC to epithelial cells in vitro in a QseC-dependent manner. We further demonstrate that DHMA is present in fecal contents from conventional specific-pathogen-free (SPF) but not germfree (GF) mice. Our results suggest that DHMA produced from NE by the gut microbiota serves as a molecular beacon for bacterial pathogens in the gut, targeting them to preferred sites of infection as well as inducing their virulence.

RESULTS

DHMA is a chemoattractant for EHEC.

Following our prior work showing that DHMA is a chemoattractant for nonpathogenic E. coli, we investigated whether DHMA is also a chemoattractant for EHEC. Figure 1 shows that EHEC 86-24 was attracted to an agarose plug containing 500 μM DHMA, as seen by the accumulation of green fluorescent protein (GFP)-labeled cells at the edge of the agarose plug after 30 min. Note that although the plug contains 500 μM DHMA, the cells must be responding to much lower concentrations in the gradient that forms as DHMA diffuses out of the plug into the pond. This response was similar to that seen with plugs containing 100 μM l-serine and markedly different from the negative response to plugs containing chemotaxis buffer (CB) alone, where no accumulation was observed over the same period of time.

FIG 1.

DHMA is a chemoattractant for EHEC 86-24. The chemotaxis response of EHEC was investigated using the plug-in-pond assay. EHEC cells expressing GFP were suspended in CB and incubated with an agarose plug containing either CB alone (control) or CB plus a signal molecule. The accumulation of cells at the plug edge was observed after 30 min with an inverted fluorescence microscope. The increase in the intensity of fluorescence at the plug boundary serves as a qualitative measure of the attractant response to the chemical in the plug. (A) Plug with CB plus 500 μM DHMA; (B) plug with CB only (negative control); (C) plug with CB plus 100 μM l-serine (positive control). The images shown are from one representative experiment of three independent experiments.

Figure 2A shows the chemotaxis response of EHEC 86-24 to 50 μM DHMA in a microfluidic device. The motility migration coefficient (MMC; see Materials and Methods) was measured with different concentrations of DHMA in stable concentration gradients (ranging from 0 to 50 μM to from 0 to 5,000 μM), and the dose-dependent response is shown in Fig. 2B. The dose-dependent MMC response to DHMA of EHEC 86-24 was similar to that of E. coli RP437 (see Fig. S1 in the supplemental material).

FIG 2.

EHEC 86-24 responds to DHMA in a dose-dependent manner. The dose-response of EHEC was measured using a microflow-based MMC assay. EHEC 86-24 cells expressing GFP were suspended in CB and introduced into the center of a microflow channel containing buffer or a fixed concentration of DHMA. One hundred images were collected over a period of 6 min and were overlaid to prepare representative snapshots. (A) Representative snapshot of cells in buffer. (B) Representative snapshot of cells in 50 μM DHMA. (C) The dose-dependent chemotaxis response of EHEC 86-24 to DHMA. The migration of cells from the center of the channel was determined by image analysis and used to calculate the MMC. The data represent the means and standard errors of the means from three independent experiments. *, **, and #, statistical significance for each response compared to that for the buffer control using Student's t test at significance levels of P < 0.01, 0.005, and 0.0005, respectively.

DHMA increases expression of virulence genes.

We investigated the effect of DHMA exposure on the expression of virulence genes in EHEC 86-24. Exposure to DHMA increased the expression of several genes from the locus of enterocyte effacement (LEE) pathogenicity island and of two non-LEE virulence genes (Fig. 3). The increase in expression with DHMA was comparable to that observed with NE. Transcription of all of the LEE genes tested was induced by both compounds at least 8-fold, and transcription of the genes for the adhesion intimin (eae) and the type 3 secretion system filament protein (espA) was induced over 30-fold. The prophage-associated virulence genes stx_2a and nleA were induced 2-fold and 9-fold, respectively. The EHEC 86-24 ΔqseC mutant did not show any induction of the LEE or non-LEE virulence genes tested in response to DHMA, as has been observed previously with NE and epinephrine (18), indicating that DHMA is sensed by QseC in the same manner as those compounds.

FIG 3.

DHMA induces expression of virulence genes in EHEC 86-24 in a QseC-dependent manner. The increase in the expression of LEE and non-LEE virulence genes upon exposure to DHMA (left) or NE (right) was determined by qRT-PCR. Untreated cultures were used as controls, and the housekeeping gene rpoA was used to normalize the data. The means and the standard errors of the means are shown for three biological replicates with two experimental replicates each. *, **, and #, statistical significance of the fold change in expression by the wild type (WT) relative to that by the mutant using Student's t test at significance levels of P < 0.1, 0.05, and 0.01, respectively.

DHMA increases attachment of EHEC to HeLa cells.

Because the protein products of the LEE genes are involved in attachment to host cells, we examined the effect of DHMA on the attachment to HeLa cells in vitro. About 1 × 10⁵ EHEC 86-24 or EHEC 86-24 ΔqseC bacteria attached to the HeLa cells in the absence of NE or DHMA, representing ∼10% of the 1 × 10⁶ cells that were inoculated. Attachment with DHMA-treated EHEC 86-24 cultures increased 2.4-fold, similar to the effect observed with NE (Fig. 4). No increase in attachment was seen with EHEC 86-24 ΔqseC bacteria treated with NE or DHMA. There was no change in the attachment of EHEC 86-24 to the wells as a result of either DHMA or NE treatment (Fig. S2).

FIG 4.

DHMA increases attachment of EHEC 86-24 to HeLa cells. The effect of DHMA and NE on attachment of EHEC 86-24 to HeLa cells was determined. The increase in attachment relative to that for the control (i.e., the ratio of the number of EHEC cells attached when the cells were grown with DHMA or NE relative to the number of EHEC cells attached when the cells were grown only with solvent) was calculated. The means and standard errors of the means are shown for six biological replicates with three experimental replicates each, using the EHEC 86-24 wild-type strain or the isogenic ΔqseC derivative. *, statistical significance of P < 0.001 by Student's t test.

DHMA is present in the gut of specific-pathogen-free but not completely germfree mice.

We used targeted liquid chromatography-mass spectrometry (LC-MS) metabolomics to determine whether DHMA is present in the murine gastrointestinal (GI) tract. In order to overcome the potential influences of the sex and age of the mice, we collected fecal samples from male and female mice of a wide age range (4 to 12 weeks). Metabolites were extracted from fecal samples from SPF and GF mice and tested for the presence of DHMA. Figure 5 shows that fecal samples from SPF mice contained approximately 2 μM DHMA. There was no significant variation in the DHMA concentrations in male and female mice of different ages. On the other hand, no DHMA was detected in fecal pellets from GF mice, which suggests that the production of DHMA by the host gut is not sufficient to be detected in this assay and therefore requires the normal microbiota. The absence of DHMA in the fecal extracts from GF mice was not due to poor extraction or a low sample load, as the total ion current (TIC) chromatograms of samples from GF mice (Fig. S3), which reflect the presence of all metabolites, were similar to those of samples from SPF mice.

FIG 5.

Presence of DHMA in fecal samples. Metabolites were extracted from fecal samples of SPF (n = 9; 4 to 12 weeks old) and GF (n = 6; 3 to 8 weeks old) mice. DHMA was detected by LC-MS and quantified using a standard curve. DHMA (exact mass, 184.037) was detected in the negative-ion mode as an ion of m/z 183.032 with a confirmatory fragment ion of m/z 137.025. (Top) Extracted ion current (XIC) chromatogram indicating the elution of pure DHMA using a Chromolith RP-18 column. (Inset) Standard curve of DHMA determined over a concentration range of 1 nM to 1 × 10³ nM. (Middle) Representative XIC chromatogram showing the elution of DHMA in a fecal extract from an SPF mouse and a GF mouse. The peak area of DHMA in the fecal extract from the SPF mouse corresponds to a DHMA concentration of 1.9 ± 0.3 μM. (Bottom) Quantitation of DHMA in SPF and GF mice. Note that no DHMA was detected in extracts from GF mice.

DISCUSSION

We previously reported that both pathogenic (19) and nonpathogenic (22) E. coli strains migrate toward NE. However, we recently showed that NE is not the actual molecule that is sensed as a chemoattractant by nonpathogenic E. coli strains (20). Rather, DHMA, a metabolite of NE formed by the activity of the bacterially encoded tyramine oxidase (TynA) and aromatic aldehyde dehydrogenase (FeaB), is the actual attractant. We have shown that the nonpathogenic E. coli strain RP437 responds to DHMA using the serine-binding site of the Tsr chemoreceptor (20). In this study, we show that DHMA is an attractant for EHEC 86-24 as well. We did not confirm experimentally that Tsr is the chemoreceptor involved in the response of EHEC 86-24 to DHMA. However, the EHEC genome contains the same complement of genes encoding methyl-accepting chemoreceptors (MCPs) that is present in E. coli RP437 (23), including Tsr, and EHEC responds robustly to serine (Fig. 1). Thus, it is likely that DHMA is also sensed in EHEC 86-24 by Tsr.

In addition to being a chemoattractant, we show that DHMA increases expression of EHEC 86-24 virulence genes and attachment to HeLa epithelial cells in vitro. The increase in both of these responses depends on the QseC sensor kinase. Previous reports have shown that the catecholamine neurotransmitters epinephrine (Epi) and NE also require the QseC sensor kinase for the induction of virulence in EHEC 86-24 (17). QseC also modulates virulence in pathogenic strains of Salmonella (24), Aggregatibacter (25), Haemophilus (26), and Actinobacillus (27), and QseC has been targeted for antimicrobial therapy (28, 29). DHMA shares the catechol moiety common to NE and Epi. Indeed, our data show that NE and DHMA are equally effective in inducing virulence in EHEC 86-24. We predict that DHMA also induces virulence in other enteric pathogens, as the QseBC TCS is found in many Gram-negative bacteria, including pathogens in the genera Shigella, Yersinia, and Vibrio (17, 30).

The ability of DHMA both to attract and to induce virulence in EHEC suggests that DHMA is an important signaling molecule in the gut microenvironment. Indeed, the presence of DHMA in fecal material from normal (SPF) mice, together with its chemoattractant and virulence-inducing properties, strongly suggests that DHMA plays an important role in EHEC infections. Although some DHMA may be produced directly by the host (31), our results with SPF and GF mice suggest that the gut microbiota produces most of the DHMA that is present in fecal matter. There is some evidence that the microbiota plays a role in the production of norepinephrine in the gut (32), and it is possible that this activity is similarly responsible for the production of DHMA. These results underscore the importance of microbial metabolites derived from host neurotransmitters as influential signals for infection and pathogenesis. Our findings suggest that the presence of the normal microbiota enhances the virulence of enteric pathogens, a conclusion that is consistent with the observations of Goswami et al. (33).

The production of DHMA by gut microbiota presumably begins with the sensing of NE via the QseBC TCS, which induces the monoamine regulon genes tynA and feaB (20). It has previously been shown that monoamines, including catecholamines like NE, induce the expression of the monoamine regulon via the FeaR regulator in Klebsiella and that a homologous operon is present in E. coli (34, 35). The activity of TynA, a primary amine oxidase, converts NE to 3,4-dihydroxyphenyl-glycol-aldehyde, which is then converted by FeaB, an aromatic aldehyde dehydrogenase, to DHMA. This mechanism leads us to propose that the NE released by the enteric nervous system into the gut, primarily at gut-associated lymphoid tissues (GALTs), is converted to DHMA by the local commensal microbiota. A search of the BLAST database indicates that feaR, tynA, and feaB are present in a number of members of the Enterobacteriaceae family. It is also possible that amine oxidases and aldehyde dehydrogenases in other bacterial taxa can also lead to the conversion of NE to DHMA. This DHMA then functions as a beacon to direct pathogenic bacteria to GALT-associated sites, such as Peyer's patches, where they establish infection.

A scheme showing how DHMA may function as a signal molecule in the gut is presented in Fig. 6. Presumably, any bacteria that possess Tsr or a similarly functioning chemoreceptor can hack into this system and migrate toward epithelial surfaces. Chemotaxis is negatively correlated with the infectivity of Vibrio cholerae (36, 37). On the other hand, chemotaxis function is an important factor in the virulence of many pathogens, including adherent invasive E. coli (38), Salmonella (39, 40), and Campylobacter jejuni (41, 42). Indeed, chemotaxis has been shown to be important for colonization by Helicobacter pylori (43, 44) and, more specifically, for its preference to colonize at sites of gastric injury (45).

FIG 6.

Proposed model for the effect of DHMA on EHEC chemotaxis and virulence. (Left) (A) Mesenteric nerves from lymphoid follicles release norepinephrine (NE; •) into the gut lumen. (Right) Enlarged image of the boxed area in the left panel. (B) NE is sensed by QseC in commensal bacteria (green), resulting in expression of tynA and feaB, whose products convert NE to DHMA (▲). (C) A gradient of DHMA is formed from NE. (D) DHMA attracts pathogens, such as EHEC (red), through Tsr-mediated chemotaxis and induces pathogenesis through QseC.

The novel function of DHMA as both a chemoattractant and a virulence inducer therefore places it in a new class of signaling molecules relevant to pathogenesis in the gut, where locally high concentrations of NE, such as those resulting under conditions of stress, would result in localized high concentrations of DHMA that recruit pathogens to sites of infection. It may be that, at preferred sites of bacterial colonization, like the GALTs that are prevalent in the ileum, the concentrations of DHMA are particularly high. This would especially be true if most of the NE released at GALTs is converted to DHMA by the resident commensal microbiota. It should be noted that the concentration of DHMA measured in fecal material (∼2 μM) likely underrepresents the actual concentration that pathogens encounter in the GI tract, as it represents a steady-state concentration and does not account for the loss of DHMA due to extraction, partitioning into different phases, and ion suppression and the selectivity of the analytical method used. It is likely that DHMA is produced locally to generate detectable gradients only in specific subregions of the gut and becomes highly diluted in the total fecal fraction.

Although DHMA is one metabolite that can be produced from NE, there may be others as well. For example, DHMA could be metabolized to 3,4-dihydroxybenzaldehye via tyrosinase (46). It remains to be seen if additional downstream products of NE are present in the GI tract and whether they play a role in virulence. Other catecholamine and analogous neurotransmitters in the gut may also undergo transformation by the intestinal microbiota to produce metabolites that function in a similar manner as DHMA. Therefore, inhibition of the production and/or sensing of DHMA may provide a target for the development of new therapies for treating infections by foodborne pathogens.

MATERIALS AND METHODS

Bacterial strains, growth conditions, and materials.

EHEC O157:H7 strain 86-24 and its isogenic ΔqseC derivatives (47) were kindly provided by V. Sperandio (University of Texas Southwestern, Dallas, TX). The bacterial strains were routinely grown and maintained in Luria-Bertani (LB) medium. For virulence studies, fresh cultures were grown to an optical density at 600 nm (OD₆₀₀) of ∼1.0 in Dulbecco's modified Eagle's medium (DMEM) with or without treatment with 50 μM DHMA or NE. For chemotaxis studies, fresh cultures were grown to OD₆₀₀ of ∼0.5 in tryptone broth (TB; 1% Bacto tryptone, 0.8% NaCl). The pCM18 plasmid (48) was used for the constitutive expression of green fluorescent protein (GFP) in EHEC for chemotaxis studies utilizing fluorescence microscopy. HeLa S3 cells (ATCC, Manassas, VA) were routinely cultured and propagated in DMEM with 10% fetal bovine serum according to standard protocols (ATCC). DHMA was purchased from Sigma-Aldrich (St. Louis, MO), and norepinephrine bitartrate was purchased from EMD Chemicals Inc. (San Diego, CA).

Chemotaxis assays.

The plug-in-pond assay, also known as the agarose-in-plug bridge assay (49), was used to investigate the response of EHEC to DHMA and serine. Briefly, 4% low-melting-point agarose in chemotaxis buffer (CB; 50 mM phosphate-buffered saline [PBS], pH 7.4, 10 mM d,l-lactic acid, 1 mM methionine, 0.1 mM EDTA) containing 0.25% bromophenol blue (to provide optical contrast) was melted at 75°C and then cooled to 52°C before addition of the chemoeffector. Approximately 5 μl of the molten agarose was placed on a glass slide, and a coverslip was placed on top of the agarose to form the plug. The coverslip was kept in place using double-sided adhesive tape on either side of the plug. GFP-expressing cells were introduced under the coverslip to form the pond around the plug. Images of the cells around the plug were captured with an inverted fluorescence microscope after incubation at 30°C for 30 min.

The dose-dependent chemotaxis response of EHEC 86-24 was studied using the motility migration coefficient (MMC) assay described previously (20). GFP-expressing cells were introduced into the center of a microfluidic channel in which a constant and stable concentration of DHMA was maintained at a flow rate of 1.5 μl/min. The flow was allowed to stabilize for 10 min before the cells were imaged. One hundred images were taken over a period of 6 min, and MMC values were calculated for different concentrations of DHMA (or buffer) by using image analysis to measure the migration of the cells away from the center of the channel.

qRT-PCR of virulence genes.

EHEC O157:H7 strain 86-24 and its isogenic ΔqseC derivative were grown and treated with 50 μM NE or DHMA as described above. RNA extraction and quantitative reverse transcription (RT)-PCR (qRT-PCR) protocols were performed as described by Vega et al. (50), with slight modifications. Cells were treated with the RNAprotect reagent (Qiagen Inc., CA) before collection by centrifugation and were stored at −80°C prior to RNA extraction. Total RNA was isolated from the cell pellets by using an RNeasy minikit (Qiagen, CA) and the protocol provided by the manufacturer. The RNA preparations were treated with DNase from an Ambion Turbo DNase kit (Thermo Scientific, MA) to remove contaminating genomic DNA. RNA quality was assessed spectrophotometrically using the A₂₆₀/A₂₈₀ ratio and gel electrophoresis of RT-PCR-amplified products. cDNA was synthesized using an Invitrogen SuperScript III first-strand synthesis system following the instructions provided by the manufacturer, and real-time quantitative PCR was performed with SYBR green master mix (Thermo Scientific, MA) and gene-specific primers (listed in Table S1 in the supplemental material). All qRT-PCR experiments were carried out with two technical replicates for each of three biological replicates; rpoA was used as the housekeeping gene for data normalization. Changes in expression were calculated using the ΔΔC_T threshold cycle (C_T) method (51).

In vitro attachment assays.

HeLa cell monolayer cultures are commonly used as a model to study the attachment of EHEC to host epithelial cells (52, 53). Attachment of EHEC O157:H7 strain 86-24 and its isogenic qseC derivative to HeLa S3 cells was performed in the presence and absence of DHMA (or NE) using a previously described in vitro protocol (19). Briefly, HeLa cells were cultured in 24-well tissue culture plates and grown at 37°C in a humidified environment with 5% CO₂ until ∼80% confluence. The HeLa cell monolayers were washed with sterile PBS thrice to remove unattached cells, and the growth medium was replaced with DMEM containing 10% heat-inactivated fetal bovine serum prior to the assay. Fresh cultures of EHEC (OD₆₀₀, ∼1.0), grown with or without NE or DHMA, were washed with sterile PBS, and approximately 1 × 10⁶ cells were added to each well containing a HeLa cell monolayer. The well plate was centrifuged at 1,000 × g for 90 s and incubated for 1 h at 37°C with 5% CO₂. Unattached bacterial cells were removed by washing each well thrice with sterile PBS, and the HeLa cells were lysed using 0.1% Triton X-100 in PBS. The cell suspension in each well was vigorously vortexed, and serial dilutions were plated on LB agar. Colonies were counted after overnight incubation at 37°C, and the fold increase in the counts of NE- or DHMA-treated cells relative to the counts of untreated cells was calculated. The means from six independent experiments are shown.

Detection of DHMA in murine fecal material.

Fresh fecal samples were collected from conventional (SPF) male and female C57BL/6 (B6) mice (n = 9; 4 to 12 weeks old) in accordance with Texas A&M University System Institutional Animal Care and Use Committee guidelines. Fecal samples collected from B6 germfree (GF) mice (n = 6; 3 to 8 weeks old) were obtained from Taconic (Albany, NY). The fecal samples were weighed, flash frozen, and stored at −80°C until further processing. Metabolites were extracted from the fecal material as described previously (54). Briefly, each sample was homogenized in 1.5 ml of chilled methanol-chloroform (2:1, vol/vol) and centrifuged (15,000 × g) at 4°C for 10 min. Six hundred microliters of ice-cold distilled water was added to the supernatant, and the samples were vortexed vigorously before centrifugation to achieve phase separation. The upper polar phase was collected and concentrated by freeze-drying. The lyophilized samples were resuspended in 100 μl methanol and stored at −80°C for LC-MS analysis.

DHMA was detected in the samples by LC-MS using an Exactive Orbitrap system (Thermo Scientific, Waltham, MA). Samples were maintained at 4°C on an autosampler prior to injection. Chromatographic separation was achieved on a Chromolith RP-18e column (100 by 2 mm; EMD Millipore, Billerica, MA) using a methanol gradient (95 to 5%, 15 min) in water. The column was cleaned after the testing of each sample with a gradient of acetonitrile (0 to 90%, 10 min), followed by equilibration with 95% methanol, and a solvent blank was run before the testing of each sample. Peak identification and integration were performed using TraceFinder software (v 3.1; Thermo Scientific, Waltham, MA).

The identity of DHMA was established using three criteria: (i) the exact mass (183.0320 m/z), (ii) the retention time on the C₁₈ column, and (iii) coelution of the 137.0253 m/z DHMA fragment ion. The total number of moles of DHMA in each sample was calculated from a standard curve (1 nM to 1 μM) of pure DHMA. The number of moles of DHMA was normalized to the mass of the fecal sample, and the concentrations of DHMA were calculated by assuming that the density of fecal material is 1.1 g/cm³. This value of density was calculated using the method of Lupton and Ferrell (55) on the basis of the fiber content (5% cellulose) of the AIN-76A diet (Harlan Teklad, Madison, WI) used to feed the mice in this study.