| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
ANIMAL GROWTH, PHYSIOLOGY, AND REPRODUCTION |
Department of Animal and Poultry Science, University of Saskatchewan, 51 Campus Drive, Saskatoon, Saskatchewan, Canada S7N 5A8
 |
Abstract |
|---|
 Top Abstract INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
(TNF), Fas ligand (FasL), CD3
, caspase 3 (casp3), and toll-like receptors (TLR)2, 4, and 9 expression were measured by quantitative PCR. Activated casp3 was measured by Western blot. Increased abundance of activated casp3 and transcripts encoding proliferating cell nuclear antigen, TNF, CD3, and FasL was observed in SF and EC treatment groups compared with GF and LF. Expression of TLR2 was increased (P < 0.05) in the SF treatment and tended to be greater (P < 0.08) in EC relative to LF and GF. Results indicate that conventional bacteria and E. coli but not L. fermentum increase overall cell turnover by stimulating increased apoptosis through the expression of FasL and TNF and by increasing cell proliferation. The differential regulation of TLR expression indicates that microbially induced changes may be mediated in part by these receptors. Induction of inflammatory responses and activation of apoptosis through death receptors appears to play a significant role in enterocyte turnover mediated by commensal bacteria.
Key Words: commensal bacteria • enterocyte turnover • gnotobiotic pig
|
INTRODUCTION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
), villous capillary networks (Stappenbeck et al., 2002), and development of Peyer’s patches (Rothkotter et al., 1991). To investigate bacterial contributions to intestinal development, we established a gnotobiotic pig model and reported marked differences in intestinal morphology in pigs monoassociated with Escherichia coli or Lactobacillus fermentum (Shirkey et al., 2006), suggesting increased cell turnover.
Because enterocyte turnover comes at a substantial metabolic cost (Nyachoti et al., 2000), the mechanisms by which bacteria affect cell turnover are of significant interest. One mechanism the host recognizes and responds to bacteria is through toll-like receptors (TLR) including TLR2, TLR4, and TLR9, which have been shown to respond to gram-positive bacteria, gram-negative bacteria, and unmethylated CpG DNA, respectively (Medzhitov, 2001). In vitro, Ruemmele et al. (2002) demonstrated that intestinal epithelial cells show a growth response to activation of TLR4 with lipopolysaccharide (LPS) that is mediated by tumor necrosis factor (TNF) . In vivo, Rakoff-Nahoum et al. (2004) showed TLR and downstream activation of IL-6 and TNF are essential in intestinal homeostasis.
We hypothesized that the composition of commensal intestinal bacteria affects enterocyte turnover affecting nutrient requirements and digestive function. We therefore examined intestinal morphology; indicators of cell proliferation and apoptosis and the expression of TLR2, 4, and 9; the death ligand, Fas ligand (FasL); and TNF in pigs monoassociated with model bacterial species known to colonize the neonatal digestive tract. Monoassociated bacteria also represented the major gram-negative and gram-positive cell wall divisions, namely E. coli and L. fermentum.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Experimental protocols were approved by the Animal Care Committee of the University of Saskatchewan and were performed in accordance with the recommendations of the Canadian Council on Animal Care (1993).
Derivation of germ-free piglets and preparation of isolators was performed as described in a previous gnotobiotic study (Shirkey et al., 2006). Briefly, 16 piglets (>800 g of BW, Large White x White Duroc) were derived via caesarian section from 2 sows and were transferred into a high-efficiency particulate air-filtered transfer unit through a betadine-filled dip tank. The pigs were then transported to the gnotobiotic facility, where they were fed, weighed, and allocated to 1 of 4 treatment groups balanced for litter of origin and BW. For the first 24 h, the piglets were fed a 1:1 mixture of sterile porcine serum (Gibco, Burlington, Ontario, Canada) and infant formula (Similac, Abbot Laboratories, Abbott Park, IL) to simulate immunoglobulins normally obtained from colostrum. Subsequently, they were ad libitum-fed a 2:1 (vol/vol) Similac:water mixture 3 times daily for the remainder of the experiment.
Experimental Design
In 2 experiments, 32 piglets (16 per experiment) were allocated to 4 treatment groups (4 piglets per treatment): pigs that remained germ-free (GF), those that were monoassociated with nonpathogenic E. coli (EC) or with L. fermentum (LF), and those that were conventionalized with sow feces along with doses of EC and LF (SF). The inoculating feces sample was freshly collected, separately for each experiment, from a clinically healthy sow that had recently farrowed in a research swine herd (Prairie Swine Center Inc., Saskatoon, Saskatchewan, Canada). Escherichia coli and L. fermentum inoculants were isolated from the cecum of a healthy pig. For isolation, cecal digesta were diluted in peptone and cultured on MacConkey (Difco Laboratories, Sparks, MD) and MRS (BBL, Sparks, MD) agar, respectively. Selected colonies were subcultured for biochemical (API50CHL and API20E, bioMrieux Vitek, St. Laurent, Quebec, Canada) and phylogenetic typing by sequencing of the chaperonin 60 universal target [cpn60 UT; http://cpndb.cbr.nrc.ca/ (Hill et al., 2002)]. Cultures (18 h) of E. coli in trypic soy broth (BBL) and L. fermentum in lactobacilli MRS broth (Difco Laboratories) were prepared, subsampled for enumeration, and used to orally inoculate piglets at 24 and 30 h after birth in their feed. By mixing into their feed and bottle-feeding, monoassociated piglets (LF and EC treatments) received 2 mL of their appropriate inoculants. Piglets in the SF group received 1 g of feces mixed with 1 mL of sterile peptone plus 2 mL of each of the E. coli and L. fermentum inoculants, for a total volume of 6 mL.
Tissue Collection
At 14 d of age, piglets were removed from the isolators, weighed, and killed by submersion in CO2 and exsanguination, at which time digesta from the jejunum, ileum, and cecum were taken to confirm treatment status and to enumerate culturable bacteria. The small intestine was carefully dissected from the mesentery, and its length was recorded. A 2-cm segment obtained at 75% (cranial to caudal) of the small intestinal length was placed in 10% buffered formalin for 24 h and subsequently transferred to 70% ethanol before embedding in paraffin and staining with hematoxylin and eosin for histological analysis. Two 10-cm segments for mRNA and protein analysis were obtained at 75% of the small intestinal length, snap-frozen, and stored at –80°C.
Enterocyte Isolation
Enterocytes were obtained using the distended intestinal sac method modified from Fan et al. (2001). Briefly, an 80-cm segment immediately cranial to 75% of the intestinal length was initially rinsed with a prein-cubation buffer (PBS with 0.2 mM phenylmethylsulfonyl fluoride (PMSF) and 0.5 mM dithiothreitol (pH 7.4). The caudal end was then clamped, and the segment was filled with isolation buffer (PBS with 1.5 mM EDTA, 0.2 mM PMSF, and 0.5 mM dithiothreitol, pH 7.4) until fully distended. The cranial end was then clamped, and the distended segment was placed in a saline bath at 37°C for 30 min. The contents of the segment, including isolated cells, were then emptied into a conical tube and centrifuged at 400 x g for 3 min at 4°C. The supernatant was discarded, and the cell pellet was resuspended in ice-cold PBS, repelleted, and snap-frozen in liquid N. The isolation procedure was repeated 6 times to obtain cells along the entire villous-crypt axis. Cell recovery was confirmed by evaluation of hematoxylin and eosin-stained cross-sectional tissues taken during the enterocyte isolation procedure. For pigs within each treatment, a pool of isolated enterocytes was prepared at each isolation step, generating 1 sample per treatment representing cell fractions along the villous tip to crypt axis. Enterocyte isolates from each isolation step were also pooled within pig (1 sample per pig representing combined cells along the entire villous tip to crypt axis).
Confirmation of Treatment Status and Microbial Enumeration
Swabs were taken perianally from GF pigs daily and placed in brain heart infusion broth (Difco Laboratories) with 0.5% Cys hydrochloride to confirm bacteria-free status through the course of the experiment. Before harvesting tissues, digesta from the ileum and cecum were collected in sterile peptone; diluted 10–1, 10–2, and 10–4; and plated using the Autoplate 4000 (Spiral Biotech Inc., Bethesda, MD) onto blood agar base (BBL) with 5% defibrinated sheep blood and cultured aerobically and anaerobically. Digesta of LF, EC, and SF pigs were also plated on MacConkey’s agar (Difco Laboratories) and cultured aerobically and on MRS agar (BBL) and cultured anaerobically. The plates were incubated for 24 to 48 h at 37°C before assessment of colony morphology and enumeration of ileal digesta. Selected colonies from EC and LF treatments were gram-stained to confirm cell wall type and morphology. Colonies or cells with morphology inconsistent with the inoculated organism were isolated and identified by sequencing of the chaperonin 60 UT sequence (Hill et al., 2002).
Intestinal Morphology
Formalin-fixed intestinal cross-sections were stained for hematoxylin and eosin by Prairie Diagnostic Services (Embury-Hyatt et al., 2005). Villous height and crypt depth were measured by an observer blinded to treatment assignment and reported as the mean length of 10 well-oriented and representative villi and crypts, respectively, from each pig using an Axiostar plus light microscope and AxioVision 3.1 software (Carl Zeiss Canada Ltd., Toronto, Ontario, Canada).
Caspase 3 Western Blot
A Western blot assay was performed on enterocyte homogenates. Tissue (200 mg) was homogenized in 5 mL of lysis buffer with a Brinkman Homogenizer (Westbury, NY) at low speed for 1 min. Lysis buffer contained 50 mM HEPES, 150 mM NaCl, 1.5 mM MgCl2, 1 mM EDTA, 1% Triton X-100 (Fischer Scientific, Pittsburgh, PA), 500 IU/mL of aprotinin, and 1 mM PMSF (pH 7.5). The homogenate was centrifuged for 5 min at 3,000 x g, and 1 mL of supernatant was transferred to a microfuge tube and centrifuged again for 3 min at 17,000 x g. A small aliquot of the supernatant was diluted 1:100 with double-distilled H20 to measure protein content using the BioRad protein microassay procedure (BioRad, Hercules, CA). The remainder of the supernatant was frozen at –80°C until the blot was performed. An aliquot containing15 µg of protein was diluted in loading buffer (4:1, loading buffer:sample, vol/vol), heated to 97°C for 6 min, loaded into an 18% acrylamide precast gel (Assorted Ready Gel, BioRad), and then run at 150 V until the dye front reached the bottom of the gel. A positive control of human caspase 3 (casp3; active subunit 17 kDa) recombinant protein (CC119, Chemicon International, Temecula, CA) and a standard of 15 µg of pooled protein from SF enterocytes was included in each gel. Protein was transferred to a 0.2-µm polyvinylidene difluoride membrane (Immuno-Blot, BioRad) at 100 V for 1 h using the Mini TransBlot Electrophoretic Transfer Cell (BioRad). The polyvinylidene difluoride membrane was then rinsed with double-distilled H20 (3 x 5 min) and then blocked with 5% nonfat dry milk (Blotting Grade NonFat Dry Milk, BioRad) in PBS with 0.1% Tween-20 (PBS-T) for 2 h. The membrane was then washed in PBS-T (3 x 5min) before applying the primary antibody, which was a rabbit antiactive human casp3 polyclonal antibody (Chemicon International; Schauser et al., 2005) diluted 1:2,000 in PBS-T (0.1% BSA), for 90 min at room temperature, washed again in PBS-T (6 x 5 min) before the secondary antibody, goat antirabbit conjugated with horseradish peroxidase (Opti-4CN Substrate Kit, BioRad) diluted at 1:8,000 in PBS-T (0.1% BSA), was applied for 90 min at room temperature. Colorimetric detection was performed using the Opti-4CN Substrate Kit (BioRad). Band intensity was measured using Scion Image software (Frederick, MD), and band intensity was standardized between gels to the pooled SF homogenate. Preliminary tests using a dilution series of SF homogenate showed a linear relationship between casp3 concentration and band densitometric quantification.
Ki-67 Immunohistochemistry
Formalin-fixed, paraffin-embedded cross-sections from the 75% location of small intestine (SI) length were subjected to immunohistochemical staining for the proliferation marker Ki-67 using a streptavidin-biotin complex technique as previously described (Haines and Chelack, 1991; Ellis et al., 1998). Immunohistochemical staining was performed by Prairie Diagnostic Services using a rabbit polyclonal antibody against human Ki-67 (PCNA); Abcam Inc., Cambridge, MA) diluted 1:100 and incubated overnight at 4°C and secondary antibody goat antirabbit (Vector Lab, Burlingame, CA) diluted 1:400 and incubated for 20 min at 42°C. Tissues were observed using an Axiostar plus light microscope (Carl Zeiss Canada Ltd.).
Gene Expression Analysis
Whole intestinal segments and enterocyte pellets were ground using a mortar and pestle, and total RNA was extracted from 20 to 30 mg of tissue using an RNeasy Mini Kit (Qiagen, Mississauga, Ontario, Canada). Genomic DNA was removed from RNA using RNase-free DNase (Qiagen). Amount of RNA was quantified by optical density at 260/280 nm using a spectrophotometer (Ultrospec 2000, Pharmacia Biotech, Baie d’Urfe, Quebec, Canada), and 1 µg of RNA was used to generate first-strand cDNA using i-Script cDNA Synthesis Kit (BioRad). Porcine PCNA sequence was not available; thus, a segment of the gene was sequenced by amplification with primers designed based on human sequence (GenBank accession BC062439). The resulting sequence (GenBank accession DQ473295) showed 93% identity to human PCNA transcript. Primers (Table 1) were designed for TNF, FasL, PCNA, casp3, CD3, and TLR2, 4, and 9 and where possible were designed to land on known or predicted intronexon splice regions. Primers for TLR were not intronspanning due to the lack of introns. Primers were designed using Oligo 6 (Molecular Biology Insights Inc., Cascade, CO) and Beacon Designer (PREMIER Biosoft International, Palo Alto, CA) software. Transcript abundance was measured by quantitative PCR using SYBR Green detection (iCycler iQ Real-Time PCR detection system, BioRad). Expression of CD3 and PCNA expression was measured in villous fractions along the villous axis to indicate whether intraepithelial lymphocytes (IEL) were the source of proliferative activity. Glyceraldehyde phosphate dehydrogenase was used as the internal control. Relative gene expression was corrected for PCR efficiency according to the methods of Pfaffl (2001).
|
Data were analyzed separately for each experiment as a 1-way ANOVA using the GLM procedure (SPSS program, Chicago, IL). Treatment means were separated using least significant difference, with significance of P < 0.05. Correlation coefficients between variables were determined by the Pearson correlation procedure (SPSS).
|
RESULTS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
All pigs, with the exception of 1 SF pig in Exp. 1, were observed as healthy throughout the experiment based on visual appraisal and appetite. The SF pig mortality was associated with alveolar flooding and edema in the lung, suggesting aspiration. For each variable, results represent the mean of 4 observations with exception of the SF group in Exp. 1, where n = 3. Swabs taken before the removal of pigs from isolators and culture of ileal and cecal digesta indicated that in both experiments, all treatment groups were maintained successfully as germ-free or monoassociated with the exception of the LF treatment group in Exp. 2. This group was contaminated with a Klebsiella pneumoniae, making the treatment group diassociated and subsequently denoted as LFKP. The distinct colony morphology associated with K. pneumoniae was not observed on EC or GF MacConkey plates, indicating that this was the only contaminated treatment. Culture of ileal digesta from EC pigs on MacConkey agar indicated colonization of 8.8 ± 0.11 and 7.6 ± 0.34 log cfu/g of digesta in Exp. 1 and 2, respectively. In LF pigs, ileal counts on MRS agar were 6.7 ± 0.28 log cfu/g of digesta. In LFKP, total anaerobes enumerated on blood agar were 8.3 ± 0.69 log cfu/g, consistent with growth of K. pneumoniae, whereas lactobacilli counts on MRS agar were 6.7 ± 0.12 log cfu/g. Culture of SF ileal digesta (Exp. 1, Exp. 2) resulted in diverse colony morphologies with aerobic blood agar counts of 7.2 ± 0.54 and 7.8 ± 0.39 log cfu/g and anaerobic blood agar counts of 7.6 ± 0.77 and 8.3 ± 0.54 log cfu/g for Exp. 1 and 2, respectively. Inoculants for the SF treatment were obtained fresh from different sows; thus, although colonization levels were similar, the composition of colonizing bacteria in SF treatment groups may have been different.
Villous Morphology
Conventionalization reduced villous height (P < 0.05) and increased (P < 0.05) crypt depth relative to GF (Figure 1). Intestinal morphology in LF pigs was similar to GF, whereas villous height and crypt depth in EC pigs was intermediate between SF and LF treatments. Klebsiella pneumoniae contamination of the LF group in Exp. 2 reduced (P < 0.05) villous height and increased (P < 0.05) crypt depth compared with GF in contrast to LF monoassociation in Exp. 1.
|
Compared with GF, activated casp3 abundance in enterocytes was increased (P < 0.05) by conventional microbiota in both experiments (Figure 2). In EC, activated casp3 abundance was intermediate to SF and GF, whereas LF elicited a similar response as GF. Similar to SF, LFKP induced greater levels (P < 0.05) of activated casp3 compared with GF. The casp3 gene expression in enterocytes was unaffected by treatment in both experiments (Table 2). Active casp3 was negatively correlated (P < 0.01) to villous height (r = –0.732) and positively correlated (P < 0.02) with FasL (Exp. 1: r = 0.880; Exp. 2: r = 0.587).
|
|
The majority of cells positive for Ki-67 staining in the intestinal epithelium were in the villous crypt for all treatment groups (Figure 3). Although Ki-67-positive cells were not enumerated, staining appeared concentrated in crypt epithelium and similar for SF and EC pigs, whereas LF pigs showed considerably less staining and GF only minimal staining. Few Ki-67-positive cells were observed within in the lamina propria; however, they appeared more abundant in the SF group. In Peyer’s patches, numerous Ki-67-positive cells were evident, again particularly in the SF group but also evident in the monoassociated and GF groups (data not shown).
|
Abundance of PCNA transcript was measured in isolated enterocytes as an indicator of proliferative activity. In Exp. 1, conventionalization significantly increased (P < 0.05) PCNA transcript abundance compared with GF (Table 2
), whereas LF pigs were similar to GF and EC pigs had intermediate (P < 0.05) PCNA expression falling between GF-LF and SF. In Exp. 2, the same trends were seen, although differences were not significant. Expression of PCNA was also positively correlated (P < 0.01) with crypt depth (r = 0.602) measured at 75% of SI length.
TLR Expression
Expression of TLR2, 4, and 9 was observed in both isolated enterocytes (Table 2) and whole intestinal tissue (Table 3). Generally, expression of TLR was increased in SF and EC pigs relative to GF, whether examined in enterocytes or whole tissue. Lactobacillus fermentum monoassociated pigs did not show increased TLR expression relative to GF in enterocytes or whole tissue. In the LFKP group, TLR expression was similar to EC and SF. With the exception of TLR2, increases in TLR expression associated with bacterial colonization were greater in enterocytes vs. whole tissue.
|
Expression
Expression of inflammatory cytokine TNF was induced over 3-fold (P < 0.05) in enterocyte isolates by both SF and EC compared with GF in both experiments (Table 2). Expression of TNF in LFKP was intermediate to GF and SF but not significantly different from either, whereas LF pigs had TNF expression equal to GF. The response observed in homogenates of whole tissue (Table 3) showed a similar, although reduced, response compared with enterocyte isolates. Abundance of TNF transcript abundance was positively correlated [Exp. 1: r = 0.820; Exp. 2: r = 0.666 (P < 0.01)] with PCNA transcript abundance.
FasL Expression
Transcript abundance of death ligand, FasL, in isolated enterocytes was increased (P < 0.05) over 5-fold in SF, intermediate in EC and LFKP (P < 0.05), and unaffected by LF compared with GF (Table 2). The response was similar in whole tissue, although less pronounced, with the only significant increase (P < 0.05) in expression being observed in SF pigs in both experiments (Table 3).
CD3 Expression
Transcript abundance of CD3 was measured as an indicator of T-cell infiltration into the intestinal epithelium. Expression of CD3 in isolated enterocytes was increased (P < 0.05) in SF, LFKP, and EC pigs and unaffected by LF compared with GF (Table 2). Increased CD3 was highly correlated [Exp. 1: r = 0.943; Exp. 2: r = 0.941 (P < 0.01)] with the expression of FasL.
CD3 and PCNA Expression Along the Villous Axis
The expression of T-cell marker CD3 and PCNA in isolated enterocytes from conventionalized pigs pooled according to the villous tip-to-crypt axis is shown in Figure 4. Expression of Ki-67 increased from villous tip to crypt consistent with Ki-67 immunohistochemistry. Expression of CD3 was consistent through all fractions and did not correlate with cell proliferation activity as measured by PCNA.
|
|
DISCUSSION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
), but high cell turnover comes at a significant metabolic cost (Le Floc’h and Seve, 2000). In this study, we demonstrated that the commensal intestinal microbiota play a role in the regulation of enterocyte turnover rate in the caudal intestine and that host response is dependent on the colonizing organism. Previously, we have observed that intestinal structure as well as proinflammatory cytokine gene expression is dependent on microbial colonization and varies with colonizing species (Shirkey et al., 2006). Using our established gnotobiotic neonatal pig model, we measured microbial-dependent modifications in villous morphology, including reduced villous length and increased crypt depth, that agree with our previous findings as well as observations of others in germ-free pigs (Thompson and Trexler, 1971), chickens (Furuse and Okumura, 1994), and rats (Coates, 1973). We further showed that these differences are attributed to both apoptotic and proliferative activity. Early comparisons of germ-free and ex-germ-free mice demonstrated that a full intestinal microbiota stimulates enterocyte turnover (Savage et al., 1981), yet the effort to unravel the complex relationship between specific commensal bacteria and activation of host gene expression pathways affecting cell turnover remains in infancy.
The role of apoptosis in the regulation of cell number in the intestinal epithelium of the mouse accounts for the majority of cell loss in adult tissues (Hall et al., 1994). This supports the strong negative correlation between activated casp3 abundance and villous height observed in this study. Although it is possible that apoptosis may occur without participation of casp3 and thus would not be identified in this study, casp3 is a primary executioner caspase, and its activation has correlated well with other apoptotic indicators in bacterial infection models (Schauser et al., 2005). In fact, the 2-fold increase in activated casp3 activity associated with SF pigs may be an underestimate of the rate of cell apoptosis, because apoptotic cells are quickly cleared within 1 to 2 h through phagocytosis by neighboring parenchymal cells (Coles et al., 1993).
The difference in apoptotic activity between EC and LF pigs is consistent with the differences observed in villous morphology. We suggest that E. coli either produces a toxic metabolite that L. fermentum does not or that the difference is associated with their ability to induce an inflammatory response. Apoptosis induced by bacteria in the intestinal epithelium can be caused by 2 means. The first is a direct effect associated with the production of toxic metabolites such as ammonia (Suzuki et al., 2002), hydrogen sulfide (Roediger and Babidge, 1997), and deconjugated bile acids (Leschelle et al., 2002). The second is indirect, through the induction of an inflammatory response, including increased infiltration of IEL and the expression of death ligands, TNF and FasL. Receptors for both of these death ligands are expressed in enterocytes and are important in the regulation of apoptosis (Strater and Moller, 2000). Increased infiltration of IEL and the expression of both TNF and FasL in SF and EC, but not LF and GF pigs, indicates that the induction of an inflammatory response likely plays a major role in the ability of E. coli and conventional bacteria to induce apoptosis. However, the apoptotic response may not be exclusive to inflammation, because the presence of luminal toxins was not measured in this experiment.
Increased expression of CD3 in enterocyte isolates from SF, EC, and LFKP pigs indicated increased infiltration of IEL into the intestinal epithelial layer. Fas ligand is expressed by lymphatic cells including IEL, and increased expression in SF and EC pigs in enterocyte isolates may have been a product of both increased infiltration of IEL into the mucosal surface as well as increased FasL expression within these cells. Inflammatory conditions markedly enhance enterocyte apoptosis induced by FasL (Ruemmele et al., 1999), thus infiltration of cells expressing FasL is not sufficient to induce apoptosis in enterocytes alone.
Enterocyte proliferation as indicated by Ki-67 immunohistochemistry and PCNA gene expression mirrored apoptotic activity in response to microbial treatments, with greatest proliferation in SF intermediate in EC compared with LF and GF. Increased intestinal epithelial proliferation has also been shown in conventional compared with germ-free mice (Abrams et al., 1963). In accordance with our findings, Savage et al. (1981) found that a strain of indigenous Lactobacillus sp. did not stimulate transit rate in monoassociated, ex-germ-free mice as indicated by the accumulation of [3H]TdR. A good correlation between PCNA transcript abundance and crypt depth indicated variation in proliferative response to different bacteria. Although PCNA protein abundance by immunohistochemistry is a more common method for assessing proliferative activity, PCNA transcript abundance as reported here is considerably more quantitative. The pattern of staining of the proliferating cell marker Ki-67 in ileal cross-sections and PCNA transcript abundance in isolated enterocytes was comparable across treatments and along the villous tip-to-crypt axis and is in agreement with PCNA transcript abundance reported in mouse enterocytes (Mariadason et al., 2005). Because CD3 transcript abundance tended to decline along the villous tip-to-crypt axis, whereas PCNA transcript abundance increased dramatically, it is very unlikely that IEL contributed to differences in PCNA expression observed between treatment groups.
The enterocyte proliferative response to E. coli and conventional microbiota may be a response to reduced number of mature enterocytes as well as enterotropic factors produced by these bacteria. It has been shown that short-chain fatty acids have a tropic effect on the intestinal epithelium (Sakata, 1987; Tappenden et al., 2003) and may account for increased expression of the enterotropic factor proglucagon, observed in monoassociated pigs (Siggers et al., 2003). Providing nondigestible fructooligosaccharides to the microbial population has been shown to increase n-butyrate concentrations and subsequently increase the number of mitotic enterocytes (Tsukahara et al., 2002). A 2.4-fold increase in the amount of ornithine decarboxylase antizyme mRNA in the enterocytes of mice colonized with Bacteroides thetaiotaomicron compared with GF (Hooper et al., 2001) indicates a potential role of polyamines produced by bacteria. Enterotoxic products of bacterial metabolism such as ammonia may also contribute to altered enterocyte replacement rate associated with different species of commensal bacteria (Suzuki et al., 2002), given that ammonia has been shown to increase epithelial cell turnover by affecting intermediary metabolism and DNA synthesis (Visek, 1978).
Gram-negative-induced host responses were greater than for gram-positive bacteria and may have been mediated by highly immunostimulatory LPS abundant in gram-negative cell wall. Indeed, an increased inflammatory response has been observed in conventional and E. coli-associated pigs as indicated by IL-1ßand IL-6 expression (Shirkey et al., 2006). Because proinflammatory cytokines have been shown to enhance enterocyte renewal (Corredor et al., 2003), this is one mechanism by which enterocyte replacement rate may have been affected. The family of TLR recognize many distinct bacterial products and likely mediate, at least in part, enterocyte responses to different bacterial species (Rakoff-Nahoum et al., 2004).
The growth modulatory effects of LPS in IEC-6 cell line are dependent on endogenously produced TNF and act in an autoparacrine-paracrine way (Ruemmele et al., 2002). The correlation between TNF and cell proliferation markers in this study is supported by the findings of Ruemmele et al. (2002), where blocking of TNF- signalling using an antagonistic anti-TNF receptor antibody abolished the effect of LPS on IEC proliferation. The importance of TLR in gut repair and intestinal homeostasis has been demonstrated (Rakoff-Nahoum et al., 2004). The induced expression of inflammatory cytokines and reparative factors, IL-6 and TNF, is dependent on the presence of commensal bacteria and an intact TLR pathway (Rakoff-Nahoum et al., 2004).
In this study we show that TLR2, 4, and 9 are expressed in enterocyte isolates as well as in whole SI tissue in the neonatal pig and that their expression is affected by microbial colonization. Increased expression of TLR2 in murine alveolar macrophages treated with LPS (Oshikawa and Sugiyama, 2003) agrees with the consistent increase in TLR2 expression associated with EC and SF pigs. Although human colonic epithelial cells seem to be fairly unresponsive to LPS (Abreu et al., 2002), small intestinal enterocytes are believed responsive to LPS by clathrin-dependent mechanisms in which TLR4 is present in the Golgi. However, we did not identify the location of TLR4 protein within the cell in this experiment. Gene expression of TLR2, 4, and 9 in mouse dendritic cells is stimulated by LPS and CpG (An et al., 2002), and this response is suppressed by inhibition of NF-
B activation. An et al. (2002) also found that TLR are differentially regulated, because blocking p38 reduced TLR2 and TLR4 expression, whereas it increased TLR9 expression. Microbial metabolites have also been shown to affect TLR expression. Saegusa et al. (2004) found that treating Caco-2 cells with butyric acid induced an increased expression of TLR1 and 6 but had no effect on TLR2 expression. The implications of differential TLR expression, based on microbial colonization, are not clear, but we suspect that the downstream effects of inflammatory response including cytokine production and eventual recruitment of IEL play an important role in host responses to luminal bacteria.
The immunostimulatory responses to gram-negative and gram-positive bacteria vary greatly and do not necessarily separate based on cell wall structure (Aderem and Ulevitch, 2000). For example, E. coli LPS induces apoptosis in a primary culture of guinea pig gastric mucous cells at 1/1,000 the concentration of Helicobacter pylori LPS (Durkin et al., 2006). Similarly, it should also be noted that lipoteichoic acid of all gram-positive organisms is not equal in its ability to activate TLR; thus, the inability of L. fermentum to induce cell death and proliferation should not be transferred to all gram-positive organisms. The probiotic Lactobacillus casei has, for example, been shown to increase infiltration of immune and IL-6-producing cells into the lamina propria of mice (Galdeano and Perdigon, 2006), although this could have been an indirect response to a change in microbial population. In addition, Lactobacillus helveticus and L. casei were shown to stimulate IL-6 production in large intestinal epithelial cells above levels observed in the absence of bacteria; however, E. coli induced double the response at the same bacterial cell concentration (Vinderola et al., 2005). Our results are supported by Erickson and Hubbard (2000), who state that PG may need to be 1,000-fold the concentration LPS to induce similar cytokine secretion and by the relative inability of Lactobacillus johnsonnii to induce inflammatory cytokine production compared with E. coli in cultured HT-29 cells (Delneste et al., 1998).
Klebsiella pneumoniae is a gram-negative bacterium in the Enterobacteriaceae family similar to E. coli. Interestingly, K. pneumoniae contamination of L. fermentum monoassociated pigs resulted in a response profile similar to E. coli. Because we have no data on the effect of K. pneumoniae in monoassociation, we cannot conclude the L. fermentum was unable to ameliorate or that it exacerbated the effect of K. pneumoniae; however, the result does seem to support a relatively benign effect of L. fermentum on host response.
Here, we present evidence that intestinal bacteria differentially affect enterocyte turnover by increasing both apoptotic and proliferative activity. Microbial induction of an inflammatory response and increased expression of death ligands appear to play a significant role in increased apoptosis and turnover. We suggest that, even in a low-pathogen environment, manipulation of the commensal microbial population colonizing the intestine could improve growth efficiency by improving digestive function and reducing the metabolic cost of replacing the intestinal epithelium. Although we report here that E. coli increases metabolic costs by increasing cell turnover, sufficient activation of the intestinal epithelium in forming tight junctions and maintaining adequate cell renewal is important. Further investigation of the role of early colonizing commensal bacteria on intestinal development is required.
1 Corresponding author: andrew.vankessel{at}usask.ca
Received for publication June 1, 2007. Accepted for publication August 22, 2007.
|
LITERATURE CITED |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
B signal transduction in regulation of TLR2, TLR4 and TLR9 gene expression induced by lipopolysaccharide in mouse dendritic cells. Immunology 106:38–45.[CrossRef][Medline] Gut 51:842–848.This article has been cited by other articles:
|
|
 |
|
  A. Castro-Alonso, N. Castro, J. Capote, A. Morales-delaNuez, I. Moreno-Indias, D. Sanchez-Macias, P. Herraez, and A. Arguello Short Communication: Apoptosis Regulates Passive Immune Transfer in Newborn Kids J Dairy Sci, May 1, 2008; 91(5): 2086 - 2088. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
