Cycasin, a glycoside from cycads, is strongly carcinogenic when administered orally to conventional rats [reviewed by Laqueur and Spatz (1968)]. However, germ-free rats failed to develop tumors even after large doses. Cycasin is hydrolytically activated by bacterial enzymes in conventional rats but is excreted unchanged in germ-free animals. Differences in biotransformation between germ-free and conventional animals or animals associated with specific microbial species have also been reported for many other xenobiotics, including propachlor (Bakke et al., 1980), 1-nitropyrene (El-Bayoumy et al., 1983), and quercetin (Schneider et al., 2000). A direct involvement of bacterial enzymes in the biotransformation has been demonstrated for some of these compounds. However, it is now evident that the intestinal microbiota has many important effects on the host. For instance, butyrate produced by bacteria is a major energy source for colonocytes (Roediger, 1980). The microbiota is also involved in the modulation of the mucosal immune system, the synthesis of vitamins and the conversion of bile acids (O'Hara and Shanahan, 2006; Albiger et al., 2007). Furthermore, the cecum is substantially enlarged in germ-free animals (Wostmann and Bruckner-Kardoss, 1959). The microbial status also affects hepatic and renal enzymes of the endogenous metabolism; for example, hepatic glucose-6-phosphate dehydrogenase activity is reduced and ATP-citrate lyase activity is elevated in germ-free rats (Reddy et al., 1973). In a number of studies, presented in detail under Discussion, effects of the microbiota on the expression of xenobiotic-metabolizing enzymes (XMEs) in the host had been reported. However, all of these studies dealt only with individual groups of XME or with activities that cannot be directly linked with a specific enzyme form. Overall, the information on the impact of microbiota on host XME is scarce. Therefore, we conducted a study on effects of microbiota on a large set of specific forms of XME. Bacterial enzymes can hydrolyze glycosylated phytochemicals and xenobiotic glucuronides released by the liver into bile. The resulting aglycons already contain functional groups and, therefore, may preferentially interact with phase 2 enzymes as substrates and possibly also as inducers. Therefore, we focused our study on this group of XMEs.

Materials and Methods

Animals. Germ-free Fischer 344 rats were obtained from the germ-free breeding colony of the Institute (Alpert et al., 2008). They were maintained in positive-pressure isolators (Metall and Plastik, Radolfzell, Germany) and housed in polycarbonate cages on irradiated wood chips at 22 ± 2°C and 55 ± 5% relative humidity on a 12-h light/dark cycle. They had free access to an irradiated diet (Altromin fortified type 1314; Altromin, Lage, Germany) and autoclaved distilled water. The germ-free status of the animals was monitored twice a month. A total of 18 male and 18 female animals were used in the present study. Six rats of each gender were associated at 4 weeks of age via intragastric application of 1 ml of a 50-fold diluted fecal sample collected either from a specific pathogen-free rat or from a healthy 30-year-old female human volunteer who consumed a normal Western diet and had not taken antibiotics for at least 1 year before the study. The remaining six rats of each gender were kept under germ-free conditions. Total bacterial cell counts in fecal samples were determined 1 week after the association with rat or human microbiota. They were similar in both groups to those found in conventional rats. One male animal reassociated with human microbiota had to be sacrificed the day after the association because of severe signs of intestinal inflammation and, therefore, could not be included in this study. The remaining animals were killed 4 weeks after the association. No macroscopic signs of inflammation were observed in any animal at this time. In other studies using similar association procedures, we observed an enhanced invasion of T cells into large intestinal tissues in the first 5 to 7 days, followed by normalization to the level observed in conventional animals. The animal experiment was approved by the Ministry of Rural Development, Environment and Consumer Protection, Brandenburg, Germany.

Subcellular Preparations. Small bowel, cecum, and colon were opened after a longitudinal cut, freed from intestinal contents, placed on an ice-cooled glass plate, and rinsed with ice-cold phosphate-buffered saline (pH 7.4). The mucosa was scraped off with a rubber scalpel. Intestinal mucosa and liver were homogenized using a Potter-Elvehjem tissue grinder. The homogenization medium (3–5 ml/g tissue) contained sodium phosphate buffer (10 mM, pH 7.4), KCl (150 mM), and Complete Protease Inhibitor Cocktail diluted according to the manufacturer's protocol (Roche Diagnostics, Mannheim, Germany). Cell debris was removed by centrifugation (15 min, 3000g, 4°C). The resulting supernatant was separated into cytosolic and microsomal fractions by ultracentrifugation (60 min, 105,000g, 4°C). The microsomal fraction was resuspended in 150 mM KCl containing 10 mM sodium phosphate buffer (pH 7.4). It was washed by repeating the ultracentrifugation steps twice. The fractions were stored at –80°C until used. The protein content was determined using the bicinchoninic acid protein assay (Pierce, Rockford, IL) according to the manufacturer's instructions.

Nomenclature of XMEs. The revised Guidelines for Nomenclature of Genes, Genetic Markers, Alleles, and Mutations in Mouse and Rat were used for gene, mRNA, and protein symbols (http://www.informatics.jax.org/mgihome/nomen/gene.shtml#gene_sym). Symbols for rat genes, cDNA, and mRNA are written in italic font, with the first letter in upper case and the following letters in lower case (all upper-case for human nucleic acids); symbols for proteins from any species, including rat, are written in roman font using all upper-case letters.

The nomenclature of many XMEs has changed much over the years. For GST we use the nomenclature proposed by Mannervik et al. (2005). Thus, GSTA1 refers to Ya₁ or 1-1, GSTA3 to Yc₁ or 2-2, GSTA4 to Yk or 8-8, GSTM1 to Yb₁ or 3-3, GSTP1 to Yf or 7-7, and GSTT2 to Yrs or 12-12 in former publications. SULTs are designated following the recommendations of Blanchard et al. (2004). GPX2, EPHX1, and EPHX2 have the trivial designations gastrointestinal glutathione peroxidase, microsomal epoxide hydrolase, and soluble epoxide hydrolase, respectively. Note that rat NAT1 is similar to human NAT2, and rat NAT2 is similar to human NAT1 with respect to structural stability and substrate selectivity (Walraven et al., 2006), whereas the degree of the nucleotide and amino acid sequence does not allow a clear association between individual human and rat NAT forms.

Antisera. Antisera raised in rabbits against rat GSTA1, GSTA3, GSTA4, GSTM1, and GSTP1 were purchased from Biotrin International (Dublin, Ireland). Anti-EPHX1 and anti-EPHX2 antisera, raised in rabbits against rat EPHX1 (Oesch and Bentley, 1976) and human EPHX2 (M. Arand, unpublished data), respectively, were a generous gift from M. Arand (University of Zürich, Zürich, Switzerland). The anti-GPX2 antiserum was generated in a rabbit using a 17-amino acid peptide of the C terminus of human GPX2 coupled to dextran (Böcher et al., 1997). SULT1A1, SULT1B1, and SULT1C2 were detected with antisera raised in sheep against human SULT1A3 (Richard et al., 2001), human SULT1B1 (Stanley et al., 2005), and human SULT1C2 (Stanley et al., 2005), respectively. Rat SULT1E1 was detected using an antiserum raised in rabbits directly against this protein (Forbes-Bamforth and Coughtrie, 1994). These antisera were generous gifts of M. W. H. Coughtrie (University of Dundee, Dundee, Scotland). SULT2A forms and SULT1C1 were detected with antisera raised in rabbits against human SULT2A1 (Comer et al., 1993) and human SULT1B1 (Wang et al., 1998), respectively. They were kindly provided by C. N. Falany (University of Alabama at Birmingham, Birmingham, AL). NAT1 and NAT2 were detected with an antiserum raised in rabbit against purified bacterial inclusion bodies of human NAT2 (Muckel et al., 2002). The same method was used for generating an antiserum against rat GSTT2 (W. Meinl, S. Hessel, and H. R. Glatt, unpublished data).

XME Protein Standards. Several rat XMEs were expressed in Salmonella typhimurium TA1538 using procedures described previously for human SULT (Meinl et al., 2006) and NAT enzymes (Muckel et al., 2002). The following rat cDNA sequences were used for expression: Sult1a1 (GenBank accession no. X52883), Sult1b1 (U38419), Sult1c1 (L22339), Sult1c2 (AJ238391), Sult1e1 (U50205), Sult2a1 (ST-20) (M31363), Sult2a3 (STa, ST-41) (X63410) and Sult2a4 (ST-60) (D14989), Sult2b1b (AJ827148), Nat1 (U17260), Nat2 (U17261), and Gstt2 (NM_ 012796).

Immunoblots. Cytosolic or microsomal protein fractions were resolved by SDS-polyacrylamide (11%, w/v) gel electrophoresis. After electrophoresis, proteins were electrotransferred to Hybond ECL membrane (GE Healthcare, Freiburg, Germany). To avoid unspecific binding of the antibody to the membranes, these were blocked with bovine albumin (1%) or dry milk powder (2%), dissolved in TBS-T [50 mM Tris-HCl (pH 7.6), 150 mM sodium chloride, and 0.1% Tween 20], at 25°C for 1 h. TBS-T was also used for diluting antisera [1:2000 to 1:10,000 for primary antisera or 1:2000 for secondary anti-rabbit or anti-sheep antibodies coupled with horseradish peroxidase (Sigma-Aldrich, Deisenhofen, Germany)]. Bands recognized by the antibodies were visualized using an enhanced chemiluminescence detection kit according to the manufacturer's instructions (GE Healthcare, Braunschweig, Germany) together with the Fuji LAS-1000 imaging system (Raytest, Straubenhardt, Germany).

To remove bound antibodies for reprobing with other antisera, membranes were incubated with 62.5 mM Tris-HCl (pH 6.7) containing 100 mM β-mercaptoethanol and 2% SDS at 50°C for 1 h. Afterward, they were washed twice with TBS-T for 15 min before new blocking reagent was applied and the membranes could be probed again with another antiserum as described above. Stripping and reprobing has advantages but involves some risks. This process saves material and workload. If one of the proteins studied shows low variation between individuals, tissues, and genders, as was the case for EPHX2, it can be used to verify that comparable amounts of protein have been loaded and that blotting efficiency was uniform for the different lanes. Risks of multiple usage of blots involve incomplete removal of an antibody and washing out of antigen before reprobing. Recombinant standards were available for several XMEs examined (all SULTs and NATs plus GSTT2). They were used in edge lanes (not included in the figures of this article) for ready detection of erroneous or insufficient signals. When such standards were not available, the antisera were used in an order that ruled out confusion between the proteins studied in adjacent analyses (e.g., due to strongly differing electrophoretic mobility). In the initial experiments, levels of XMEs were compared between tissues and genders. In several rounds of analyses, we adjusted the amount of total protein loaded for each tissue to the expression levels of the corresponding XME. Because the adjustment varied among the XMEs, reprobing was the exception in the final blots used for quantification. However, reprobing was widely used (up to 10 times in sequence), when XME levels in the same tissues of control and germ-free (or human microbiota-associated) animals were compared. In these analyses, all samples were contained on the same blot, minimizing possible errors due to washing out. Interindividual variation for most XMEs within a group was relatively low and did not appear to increase with the number of reblotting cycles. In particular, this result was verified with the control colon samples, which were analyzed at least twice with all antisera in a different sequence: on one blot in comparison with the germ-free animals and on another blot in comparison with the rats associated with human microbiota. Thus, antigen washing out was insignificant or homogeneous and, therefore, unproblematic under these conditions.

Identification of XMEs in the Immunoblots. EPHX1 and EPHX2 are unique proteins in microsomal and cytosolic fractions without close relatives. All tissues showed a clear band with the same electrophoretic mobility and negligible background signals. Therefore, there was no risk of confusion with other proteins. The same was true for the peptide antiserum used for detecting GPX2, whose immunoreactivity clearly differs from those of the other three GPX proteins of the rat (Wingler et al., 1999).

The anti-SULT and anti-NAT antisera showed some cross-reactivity, detecting several forms. However, standards of NAT1, NAT2, and most SULT forms, expressed in bacteria, were available to determine their electrophoretic mobility and immunoreactivity. The NAT standards were separated in the electrophoresis and showed the same electrophoretic motilities as the immunoreactive proteins in hepatic and intestinal samples. Therefore, an association was trivial. The situation was somewhat more complicated for SULTs, as more different forms exist. The electrophoretic mobility of different forms and their reactivity with the various antisera used are listed in Supplemental Table 1. All three SULT2A forms showed the same electrophoretic mobility and reacted well with the various anti-SULT2A antisera available in our laboratory. Therefore, they were analyzed as a group rather than as individual proteins.

A relatively large number of GSTs exist in the rat: at least 14 class Alpha, Mu, Pi, and Theta forms (Mannervik et al., 2005). Only one form, GSTT2, was available as a standard. Fortunately, all of anti-GST antisera used showed rather high specificity. Each of them only recognized a single band in hepatic cytosolic preparations, when reasonable protein levels were loaded. One or two additional bands, with a signal intensity of less than 1% compared with the main band, were seen in overloaded blots with some antisera. Our electrophoretic conditions were comparable to those used by Sherratt et al. (1998), who reported the following order of mobility (apparent molecular mass): GSTP1 (24.8 kDa) > GSTA4 (25.0 kDa) > GSTA1/2 (25.5 kDa) > GSTA5 (25.8 kDa) > GSTM3 (26.0 kDa) > GSTM1 (26.3 kDa) > GSTT2 (26.5 kDa) > GSTA3 (27.5 kDa) > GSTT1 (28.0 kDa). Other studies indicate that GSTM2 migrates close to GSTM1. Estimates on the resolution of our electrophoresis are based on extensive experience with numerous SULT forms. Equal bands are fully separated in one lane, when their apparent molecular mass differs by at least 0.6 kDa. Shifts in bands in neighboring lanes may be seen for differences of 0.2 kDa. The comigrating forms GSTA1 and GSTA2 only differ in eight amino acid residues. Therefore, it is likely that both forms were well recognized by the antiserum raised against GSTA1. We have no information on possible cross-reactivities of anti-GSTM1 antiserum with other GSTM forms (commercial antiserum, no longer available). Cross-reactivity with GSTM3 would not create any problem, because this form is absent in liver and gut of rats (Hayes and Mantle, 1986). However, these tissues express significant levels of GSTM2 in addition to GSTM1 (Hayes and Mantle, 1986). Therefore, we do not know whether and to which extent GSTM2 contributed to the signals labeled GSTM1 in the present study on the basis of the designation of the commercial antiserum used.

Results

Tissue Distribution of XMEs in Male and Female Control Animals (Reassociated with Rat Microbiota). For this analysis, subcellular preparations from a given tissue and gender were pooled, using equal protein amounts from each of the six animals of the group. Various protein levels were analyzed in initial immunoblots. Subsequently, protein levels were adjusted such that the signals for the various samples were in the same densitometric range. This adjustment is exemplified in Fig. 1 for GSTA1/2, the XME showing the largest (1000-fold) variation in expression between tissues (ignoring XMEs that were not detectable in some tissues). From this blot it can be seen that 3 μg of small intestinal protein produced a slightly stronger signal than 1 μg of hepatic cytosol. Likewise, 150 μgof colonic protein led to a somewhat weaker signal than did 0.3 μgof hepatic protein. The cecal sample elicited a signal of half the intensity of that for an equal amount of colonic sample. Males and females showed nearly equal GSTA1/2 levels in all tissues investigated. These evaluations by visual inspection were verified and specified using densitometry. The resulting values are listed in Table 1.

Anti-GSTA1 antiserum recognized a single band in liver but lost some specificity when very high amounts of total protein had to be used with the colon and cecum samples. Then some additional bands were detected (X, Y, and Z in Fig. 1). Band Z comigrated with GSTA3, detected with high sensitivity with another antiserum in these samples. Bands X and Y do not represent any known GSTA forms, as these migrate faster (see Materials and Methods for apparent molecular masses). Despite the reduced specificity of the antiserum at this low level of expression, it is probable that the immunoreactive band with the migration of GSTA1/2 indeed represented this protein, as others had demonstrated the presence of small levels of GSTA1 in colonic cytosolic fraction after enrichment by S-hexylglutathione affinity chromatography (Hayes and Mantle, 1986).

Analogous analyses were conducted for the other XMEs. Most positive results involved the exclusive staining of the appropriate band(s) without any background signals, except those described for GSTA1/2 in the preceding paragraph. The results are summarized in Table 1. Ten XMEs (all GSTs, GPX2, both EPHXs, and NAT1) were detected in all tissues investigated, although at strongly varying levels. The remaining seven XMEs were detected only in some tissues: four SULT forms (SULT1A1, SULT1C1, SULT1E1, and SULT2A) only in liver; the remaining two SULT forms (SULT1B1 and SULT1C2) in liver and, at a clearly lower level, in large intestine, but not in small intestine; and NAT2 was absent in liver and small intestine, but present in large intestine, with a higher level in colon than cecum. Three XMEs (NAT2, GSTP1, and GPX2) demonstrated their highest expression in colon; all of the other XMEs had their highest levels in liver.

Several XMEs were expressed in a gender-dependent manner in liver. Differences were greatest for several SULT forms. SULT1E1 was detected exclusively in males. SULT1C1 was also expressed with high selectivity in males. In contrast, SULT2A levels were approximately 100-fold higher in females than in males. Several GSTs showed higher expression by a factor of 2 to 3 in one gender: GSTM1 in males and GSTA3, GSTA4, and GSTT2 in females. The hepatic level of GPX2 was 2.5-fold higher in females than in males, but both levels were low compared with those in large bowel. In contrast to the liver, gender-dependent differences in expression were negligible for any XME studied in gut. They maximally reached a factor of 1.5, which may be within the experimental and interindividual variation.

Levels of XMEs in Germ-Free Animals Compared with Control Animals (Reassociated with Rat Microbiota). For this analysis, subcellular preparations from the individual animals were analyzed separately. The corresponding samples from germ-free and control animals were studied concurrently on a single blot, as exemplified in Fig. 2 for colonic samples. Densitometric analyses provided the data presented in Table 2. Levels of several XMEs were statistically elevated in colon mucosa of germ-free animals compared with those in control animals. In general, the effect was similar in both genders. The increase was strongest for GSTA1/2 (4.0- and 5.0-fold in males and females, respectively) and EPHX1 (3.5- and 2.4-fold). Weaker increases in both genders, statistically significant in at least one gender, were observed for GSTA4, GSTM1, EPHX2, SULT1C2, NAT1, and NAT2. In contrast, colonic levels of SULT1B1 were significantly reduced in germ-free animals in both genders (to 40 and 50% of the control levels in males and females, respectively). Several statistically significant changes in levels of XME also occurred in cecum. They paralleled the changes observed in colon but were somewhat weaker in general. However, expression of NAT2 was more strongly enhanced in the cecum than in the colon of germ-free animals compared with that of control animals. The GSTM1 level was enhanced in colon but was reduced in cecum of germ-free females. The enhancement of this protein in female colon was quantitatively moderate but nevertheless statistically highly significant; the decrease in cecum was not significant. Male rats showed virtually no changes in GSTM1 expression in dependence on the association with microbiota. Thus, the apparent effect in female colon mucosa may be accidental despite the statistical significance. The level of SULT1B1 was significantly decreased in colon but enhanced in cecum of female rats. The effect in colon, but not in cecum, was reproduced in males.

Comparison of Colonic Levels of XMEs in Rats Associated with Rat and Human Gut Microbiota. Subcellular preparations from the individual animals were analyzed separately. The samples from rat and human microbiota-associated animals were concurrently studied on a single blot. The blots are shown in Supplemental Figure 1. Densitometric analyses provided the data presented in Table 3. Levels of most XMEs were similar in both groups (associated either with rat or human microbiota). However, GSTT2 was significantly elevated in both genders in rats associated with human microbiota; four other enzymes (SULT1B1, SULT1C1, and both NAT forms) were significantly enhanced only in males of this group. All increases were quantitatively moderate (≤1.6-fold increase).

Discussion

Levels of XMEs in Liver and Intestine of Control Animals. The most homogeneous expression among the XMEs investigated was observed for EPHX2. Its level was nearly identical in liver, small intestine, and colon and was unaffected by gender; cecal levels were half of those in the other tissues. SULT1E1 was the XME expressed with the highest specificity. It was exclusively detected in male liver but not in female liver or intestinal tissues of either gender.

Several XMEs, including all GSTs as well as GPX2, were present in all tissues studied, although with large (up to 1000-fold) variations in levels. Our study confirms previous observations demonstrating the presence of GPX2 protein in rat liver (Brigelius-Flohé et al., 2002). In an initial study, GPX2 mRNA was readily detected in human liver, whereas its orthologs in rats and mice were only found in the intestinal tract (Chu et al., 1993). However, in more recent studies Gpx2 mRNA was also detected in rat liver, for example, by Nishimura et al. (2007).

Four SULTs were detected exclusively in liver; two other forms were also present at moderate levels in large bowel. No SULT was detected in small intestine. These findings strongly differ from observations in humans, whose gut is particularly rich in SULT, with the highest levels and the broadest diversity occurring in small intestine (Teubner et al., 2007). For example, the level of SULT1A3, a form not existing in nonprimates, is very high in all sections of the human intestine, reaching its peak in ileum. This form is absent in liver. Likewise, levels of human SULT1A1 and SULT1B1 were higher in ileum than in liver (1.5- and 17-fold, respectively). Both enzymes were also detected at high levels in other sections of human intestine. A further species difference in SULT involves the influence of gender on the expression in liver. Gender-dependent differences in hepatic expression are minute or absent in humans, depending on the SULT form (Pacifici, 2005). In contrast, we show here that SULT1C1, SULT1E1, and SULT2A proteins are expressed in rat liver with high gender selectivity, in accordance with previous observations on the mRNA level (Liu and Klaassen, 1996a,b).

The intestine of Fischer 344 rats (the same strain as used in the present study) has demonstrated substantial NAT activity with various substrates (Ware and Svensson, 1996). However, to the best of our knowledge, our study is the first one demonstrating the presence of NAT proteins in this tissue. NAT1 showed its highest level in liver but was also present in small and large intestine. In contrast, NAT2 was highest in colon but was not detected in liver and small intestine. These findings contrast with observations made on the mRNA level. The mRNAs of both Nat genes were detected in many tissues, in a rather constant ratio of 4:1 in favor of Nat1 (Walraven et al., 2007). Our divergent results for the corresponding proteins indicate that there must be further tissue-dependent post-transcriptional regulation mechanisms.

Influence of Intestinal Microbiota on Levels of XMEs in Rat Tissues. In the present study, we found that the microbial status affects the expression of diverse XMEs in host tissues. The strongest effects were observed in colon. Colonic levels of many XMEs were higher, but levels of one enzyme (SULT1B1) were lower in the germ-free state compared with the control (association with rat microbiota). With few exceptions, expression of the same enzymes was changed in cecum in the same direction, usually to a somewhat smaller extent. No changes were observed in small bowel, a tissue exposed to microbiota to a much lesser extent than large bowel. However, effects were seen in liver. In this tissue, several SULTs were up-regulated, including two forms (SULT1A1 and SULT1C1) that were not detected in intestinal tissues at any microbial status. Intestinal bacteria metabolize bile acids and thereby affect patterns and levels of reabsorbed bile acids. Lithocholic acid formed by intestinal bacteria is hepatotoxic but can be detoxified by SULT2A enzymes in the liver; sulfation also enhances the excretion of other bile acids (Alnouti, 2009). However, the levels of the hepatic SULT2A were not affected by the microbial status of the animals in our study. Moreover, the large gender-dependent difference of SULT2A expression suggests that factors and functions that are more gender-dependent than bile acid metabolism are important in the regulation of these enzymes.

Substituting human microbiota for rat microbiota led to a number of statistically significant but quantitatively minor changes in levels of XMEs. This result suggests that the kind of colonization of the gut could be important in the regulation of XMEs. However, complex microbiota populations may not be the ideal means to elucidate this aspect.

We have not elucidated mechanisms underlying altered XME expression. Taking into account the fact that in nearly all cases enzyme levels were highest in the germ-free status, one might speculate that bacteria eliminated feed-borne enzyme inducers. However, the complex nutritional and immunological microbiota-host interactions may allow for alternative mechanisms.

More recently, global gene expression at the mRNA level has been compared in conventional germ-free and variably reassociated mice (Hooper and Gordon, 2001). Studies specifically addressing XME levels in dependence on the microbial status are discussed in the subsequent paragraphs.

Esworthy et al. (2003) reported a strong elevation of Gpx2 mRNA levels in ileum and colon after 7 days of reassociation of germ-free Gpx1(+/–)/Gpx2(+/–) mice with murine bacteria. In contrast, we did not observe any effect of the microbial status on intestinal GPX2 at the protein level. It cannot be judged whether the difference in the result is due to differences in the read-out (mRNA versus protein), the species, the overall Gpx status, the different time of reassociation (4 weeks versus 7 days), or other experimental parameters. However, higher levels of Gpx2 mRNA and protein were found in the intestines of germ-free mice when they were fed a selenium-deficient diet (Hrdina et al., 2008). Thus, commensal bacteria might compete with the host for selenium only when it becomes limiting, which was not the case here.

Treptow-van Lishaut et al. (1999) analyzed the profiles of GST in rat liver colon chromatographically after affinity extraction on hexylglutathione-Sepharose. GSTP1 was the most predominant form in colon, followed by GSTM1 and GSTM2. GSTA1, close to the limit of detection, was only observed in individual animals. GST levels were then studied in animals maintained under various diets and also in some germ-free animals. Levels of GSTP1 were statistically significantly decreased in germ-free compared with conventional rats (by nearly 20% in four female and two male animals pooled per group), whereas GSTM1 and GSTM2 were unaffected. These data are not directly comparable to our study because they were obtained from animals fed a high-risk diet (containing high amounts of sucrose, saturated fat, and sodium caseinate) for 42 days. GSTP1 was unchanged in our study.

Edalat et al. (2004) analyzed GST expression in colon of germ-free and reassociated mice using immunohistochemistry. They used an antiserum that specifically recognized GSTA4 and other antisera that were cross-reactive with the various forms within a class. Levels of GSTA, and specifically of GSTA4, were increased, whereas those of GSTM and GSTP were decreased in the period of 4 to 18 days after conventional reassociation. However, all GST forms were elevated in animals monoassociated with Lactobacillus strain GG (only studied after 4 days). After 34 days, all differences in GST expression between germ-free and conventionally reassociated mice had disappeared. Therefore, the up- or down-regulation of the individual GST forms was interpreted to be a transient host reaction to the process of bacterial colonization. We analyzed the animals 28 days after reassociation, close to the last observation time in the study in mice. Nevertheless, we observed substantial changes in the colonic levels of various GST forms. Moreover, the effects on GSTA forms in our in rats were opposite (decreased expression in reassociated animals) to those observed in mice (increased expression). The situation is complicated as the kind of microbial colonization has a pronounced impact on expression of Gsta4 mRNA, as shown in a microarray-based comparison of ileal expression of 25,000 genes in age-matched germ-free and variably associated mice (Hooper and Gordon, 2001): Gsta4 expression was unchanged between germ-free mice and mice reassociated with conventional microbiota, down-regulated by a factor of 2 after monoassociation with Bacteroides thetaiotaomicron, and up-regulated by a factor of 4 after monoassociation with Escherichia coli K12.

Lhoste et al. (2003) compared the influence of catechins on several xenobiotic-metabolizing activities and cytochrome P450 levels in germ-free and human microbiota-associated Fischer 344 rats. Constitutive activities of microsomal glutathione transferase and 4-methylumbelliferone UDP-glucuronosyltransferase in liver and intestine were higher in germ-free than in associated rats. These activities and hepatic cytochrome P450 levels were differentially induced by catechins, depending on the microbial status. The authors suspect an involvement of bacterial metabolites from catechins.

In conclusion, we demonstrated that intestinal microbiota can affect levels of many XMEs in large intestine and liver. Few studies had been conducted on this microbe-host interaction before, and they only addressed individual XMEs. Nevertheless, the results indicate that influence may vary between different models, e.g., mice and rats, and probably also with different diets. However, the effects observed in the present study and also in the previous studies were moderate compared with tissue-dependent expression differences.