Final Progress Report
Proposal No. IBD-0098
Principal Investigator: James M. Anderson, M.D., Ph.D.
Applicant Organization: University of North Carolina at Chapel Hill (U.S.A.)
Project Title: The role of intestinal flora in transcriptional and functional control of the tight junction barrier
Period of Award: January 1, 2004 - February 28, 2005
A. Summary of project aims.
The aims of this project were to test the hypotheses that:
Gut flora influence intestinal tight junction barrier properties. We were to compare mice (129) raised germ-free (GF), normally-colonized specific pathogen free (SPF) and mono-associated (L. plantarum 299v) comparing mid-jejunum, ileum, cecum and colon for:
1. apparent permeability for non-charged solutes: [3H]-mannitol and graded sizes of fluorescent dextrans (reflecting the size of pro-inflammatory bacterial products).
2. histology and freeze-fracture electron microscopy.
3. mRNA transcript and protein profiles for the claudins and other proteins that control the barrier.
4. immunofluorescence microscopy for tight junction proteins.
We predicted that germ-free gut will show a leaky phenotype and that commensals and probiota will induce a tighter phenotype coinciding with a different claudin profile.
B. Accomplishments towards meeting those aims.
Preliminary description of claudin levels and locations. Before looking for bacterial influences on claudin levels, we first needed to define which claudins are expressed in the mouse intestine and their cellular localizations. qPCR profiles were done for claudins 1 to 19, ZO-1, occludin, E-cadherin and JAM in the duodenum, jejunum, ileum, cecum and proximal and distal colon. Some claudins are not expressed in the gut. Some are expressed at similar levels throughout and other show gradients up or down the length of the gut. Each has a complex subcellular location with some showing gradients up or down the crypt-villus axis. We intend to publish this large survey of normal tissue as a resource for those interested in transport and cancer. Claudin levels are now being used experimentally as prognostic factors in outcome of several GI cancers, yet their normal locations have not been described. This represents a large collection of data and will be submitted to the BMRP when the paper is accepted.
At steady state, the mRNA expression profile for TJ proteins in the colon is not significantly affected by bacterial colonization. We used qPCR to determine whether the expression profile of TJ mRNAs differed between GF (germ free) and SPF (specific pathogen free, i.e., no Helicobactor species) animals. This work encompassed several different kinetic paradigms, including comparing mice who were raised GF or SPF from birth (studied at 6-9 months) and two other studies in which GF animals were inoculated at 8 weeks, 11 weeks or 6 months of age and studied 5-8 weeks later. The pattern of mRNA expression within all groups was extremely consistent (for example, Figure 1).
Figure 1. qPCR profiles in proximal colon of 15 week old 129SvEv mice; GF compared with littermates who were colonized with SPF flora at 10-11 weeks of age. N=6 in each group. RNA levels reported relative to EF1α, a very abundant and constant reference.
We were initially surprised by these results. However, a review of the literature confirmed that most in vivo studies are performed either at short times after colonization or in states of frank inflammation. Also, the many recent studies showing bacterial and cytokine affects on cultured intestinal cells are performed over hours to days. Some of these recently reported effects include either increases or decreases in TJ resistance, mRNA and protein levels and MAPKinase phosphorylation (1,2). Thus, our findings are significant because evidence is lacking on the long term steady state influence of flora on the barrier at the level of the epithelial transcriptome. As presented below, similar results were observed after monoassociation with the non-pathogenic gram positive commensal E. faecalis (Figure 4).
Barrier physiology in the colon is not strikingly different between GF and SPF mice. We next asked whether the colonic barrier was different at 4 months of age in mice raised in germ-free conditions compared with littermates, which had been inoculated with SPF flora at two months. Consistent with the results cited above, the profile of TJ mRNAs was not different between the two groups. Segments of proximal colon were mounted in Ussing chambers and studied over 4 hours for electrical resistance, short circuit current, and flux of 70kD 14C-dextran, 3H-mannitol and 22Na (N=5-6 mice; three colon samples from each animal). As can be seen in Figures 2 & 3,
Figure 2. Proximal colonic epithelial resistance in GF and SPF mice described in Figure 1 and text. N=5 and 6 mice in each group and 3 tissues from each mouse. There is a trend for GF tissues to be leakier but this is not statistically significant when compared at individual time points.
Figure 3. Short circuit currents in proximal colonic epithelial samples from GF and SPF mice described in Figure 3.
the GF mice showed a trend toward lower electrical resistance and Isc, but there was no statistically significant difference in any electrical or flux parameter (not shown). These studies were conducted in collaboration with our colleagues at the Gnotobiotic Facility at NC State University College of Veterinary Medicine, Robin M. Hopwood-Courville DVM, and Jody L. Gookin DVM-PhD.
Wild–type and IL-10 -/- mice show very different acute and long term responses to colonization with E. faecalis. The gram positive bacterium E. faecalis induces colitis in IL-10 deficient mice after 12-14 weeks of colonization, but not in WT 129SvEv mice (3). Our colleagues recently documented transient signaling events after colonization, which differ between WT and IL-10-/- mice (D. Haller and B. Sartor). These include a transient activation of NF-κB and TGFβ/Smad signaling peaking at 7 days and a long term loss of toll-like receptor-2. Presumably this represents a bacterial-epithelial interaction that contributes to tolerance and prevents colitis. Downregulation of the TLR may provide adaptive tolerance to the continued presence of bacteria. In contrast, IL-10 deficient mice fail to activate TGFβ/Smad signaling and have persistent activation of NF-κB and expression of TLR2. Presumably, this represents a lack of adaptation to colonization that 12 weeks later correlates with overt colitis. Using the same experimental paradigm, we asked whether the profile of TJ mRNAs differed between WT and IL-10 -/- mice at short times and after establishment of colitis. These studies were performed in the UNC Gnotobiotic Facility in collaboration with Drs. Dirk Haller (Technical University of Munich) and Balfour Sartor (University of North Carolina). Mice were examined at t0, 7 days, 14 days and 14 weeks after monoassociation. We have not yet had time to analyze proteins by immunoblotting and immunofluorescence microscopy (n=3-6/group). H & E stained sections confirmed colitis at 12 week in the IL-10-/-, but not WT mice.
Under germ-free conditions, the profile of TJ mRNAs is the same in wild–type and IL10-/- mice; i.e., the absence of IL-10 does not itself result in a change in the TJ profile, Figure 4.
Figure 4. qPCR profiles in proximal colon of GF WT and IL-10 deficient mice. RNA levels are reported relative to EF1α, a very abundant and constant reference. Transcript levels were the same in GF WT and IL-10-/- mice and in both groups after colonization with E. faecalis for 14 weeks.
In contrast, WT and IL-10-/- mice differ significantly in their responses at 7 days, 14 days and during established colitis at 14 weeks (Table 1).
Table 1. Fold change in mRNA from GF levels at 7d, 14d and 98d in WT and IL-10-/- mice inoculated with E. faecalis. “-“= no change from pre-colonization. Numbers in bold are significant at the p=0.05 significance level.
WT mice show a striking transient increase at 7 days in mRNA for occludin (~5-fold) and ZO-1 (>8-fold) mRNA levels, which increase is coincident with a transient rise in TNFα (19-fold) and INFγ (~9-fold). Signaling pathways stimulated by bacterial colonization including induction of the NF-κB and TGF-β/Smad have been demonstrated to result in decreased expression and mislocalization of occludin and ZO-1 (>8-fold) mRNA levels, which increase is coincident with a transient rise in TNFα (19-fold) and INFγ (~9-fold). Signaling pathways stimulated by bacterial colonization including induction of the NF-γB and TGF-β/Smad have been demonstrated to result in decreased expression and mislocalization of occludin and ZO-1 (4,5). The observed increase in TJ mRNAs may indicate a compensatory response to transient disruptions of expression of these proteins, perhaps through a proliferative response in the crypts. We have previously reported that disruption of TJs in cultured Caco-2 cell causes mRNA levels for ZO-1 to increase dramatically (10- fold). In contrast, levels are very low in well-differentiated monolayers (6). We strongly suspect the increase in ZO-1 mRNA occurs in the proliferating crypt or stem cell compartments, based on our previous in situ hybridization studies showing that the only ZO-1 mRNA levels above background are in this cell population, Figure 6 (6). Claudin-7, which is unusual among the claudins in that it is expressed on the basolateral surface along the entire crypt/villus axis, also shows a small increase at 7d. These changes trend back to or below baseline over the following week. Along with ZO-1 and occludin, JAM-1 and JAM-4 mRNA levels decrease at 14d. JAMs are immunoglobulinsuperfamily members expressed by leukocytes and platelets as well as by epithelial cells. They concentrate at cell-cell contacts and are specifically enriched at tight junctions. Members of this family have been demonstrated to be important in mediating leukocyte/endothelial cell interactions and in epithelial cell polarity (7). The decrease in their mRNA levels may represent part of the recovery from the acute response to bacterial colonization.
We conclude that mRNAs encoding specific TJ markers show significant initial increases in response to bacterial colonization that return to baseline or below within two weeks. These increases likely represent a mechanism to repair barrier function disrupted by short-term inflammatory responses. The long term changes (decreased occludin and JAM-4 mRNA levels) may represent adaptive responses to colonization unrelated to inflammation. The long term decrease in mRNA for JAM4 and occludin following monoassociation with E. faecalis was not seen after colonization with the complex SPF mixture, suggesting there may be long-term bacteria-specific changes in TJ mRNAs.
The TJ mRNA profile in IL-10-/- mice colonized with E. faecalis is very different. These mice fail to mount the initial response of ZO-1 and occludin expression and during the later phase of inflammation, develop significant changes in the transcript levels for claudins 2, 3, 8, occludin and ZO-1, Table 1. This pattern is selective since the levels of other claudin mRNAs do not change. In addition, the direction of change is specific; claudin-3 transcript levels decreases ~3-fold, while that of claudin-2 and -8 go up ~2-fold and ~6-fold, respectively. A recent study performed in IL2-/- mice documented increases in several claudins during colonic inflammation that were positively associated with enhanced epithelial tightness (↑impedance and ↓mannitol flux) (8). Diarrhea was the result of decreased ENaC-dependent Na transport and the authors speculated that the enhanced TJ barrier was a physiologic adaptation serving to limit diarrhea. It is not our goal to study the TJ changes associated with chronic inflammation, but rather to focus on the novel observation of an initial transient response, which may represent the initial perception of bacteria by the epithelium and be required to prevent longer term inflammation or damage.
C. Future Applications.
We do not consider the work ready for requesting additional funding. We were initially surprised by the lack of a robust effect of bacteria on barrier properties. Feedback from investigators at the BMRP Annual Meeting reinforced our concerns about how difficult it will be to perform interpretable experiments. For example, we have been concerned that germ-free mice are exposed to bacterial products in their feed. Innate epithelial responses are likely already activated before the mice are colonized. The consensus of investigators at the meeting was that this line of investigation will not lead to a near term cure for IBD or insights into therapies to ameliorate symptoms. We are grateful for the initial funding and will continue to consider how our work can be applied to IBD.
REFERENCES:
1. Cario, E.; Gerken, G.; Podolsky, D. K. Gastroenterology 2004, 127(1), 224-238.
2. Otte, J. M.; Podolsky, D. K. American Journal of Physiology-Gastrointestinal and Liver Physiology 2004, 286(4), G613-G626.
3. Kim, S. C.; Tonkonogy, S. L.; Balish, E.; Warner, T.; Sartor, R. B. Gastroenterology 2001, 120(5), A82-A83.
4. Minagar, A.; Long, A.; Ma, T.; Jackson, T. H.; Kelley, R. E.; Ostanin, D. V.; Sasaki, M.; Warren, A. C.; Jawahar, A.; Cappell, B.; Alexander, J. S. Endothelium-Journal of Endothelial Cell Research 2003, 10(6), 299-307.
5. Wachtel, M.; Bolliger, M. F.; Ishihara, H.; Frei, K.; Bluethmann, H.; Gloor, S. M. Journal of Neurochemistry 2001, 78(1), 155-162.
6. Anderson, J. M.; Vanitallie, C. M.; Peterson, M. D.; Stevenson, B. R.; Carew, E. A.; Mooseker, M. S. Journal of Cell Biology 1989, 109(3), 1047-1056.
7. Ebnet, K.; Aurrand-Lions, M.; Kuhn, A.; Kiefer, F.; Butz, S.; Zander, K.; Meyer zu Brickwedde, M. K.; Suzuki, A.; Imhof, B. A.; Vestweber, D. J.Cell Sci. 2003, 116(Pt 19), 3879-3891.
8. Barmeyer, C.; Harren, M.; Schmitz, H.; Heinzel-Pleines, U.; Mankertz, J.; Seidler, U.; Horak, I.; Wiedenmann, B.; Fromm, M.; Schulzke, J. D. American Journal of Physiology-Gastrointestinal and Liver Physiology 2004, 286(2), G244-G252.
D. Lay Summary
The inner surface of the bowel creates a barrier to entry into the body of potentially harmful intestinal contents. The first line of defense is formed sheets of cells and a specialized intercellular barrier between them called the “tight junction.” While we are highly dependent on the normal resident bacteria for proper digestion, inflammatory bowel disease results from an abnormal immune response to these normal bacteria. One hypothetical reason for this inappropriate response is that the barrier lining of the bowel becomes leaky and allows proinflammatory bacterial products to enter the body. Work from another laboratory recently shown that bacteria can induce changes in gene expression of the bowel and this is a normal symbiotic response. In addition, so-called “probiotic” species of bacteria have proven used in treating some aspects of human IBD and can protect genetically colitis-prone mice from colitis that is induced by colonization with normal bacteria. Our studies addressed the question of whether bacteria can regulate the genes encoding tight junction proteins and barrier tightness. This might tell us how probiotic bacteria can protect from colitis.
We studied mice with defined intestinal bacteria available from the UNC Gnotobiotic Rodent Core. We predicted the germ-free gut will have a leaky barrier and that normal bacteria and probiota will induce a tighter barrier. Mice were germ-free from birth and colonized with a mixture of normal bacteria or a single species at defined times after birth from week to months. Barrier tightness was measured across pieces of colon using electrophysiologic tests and the ability of small tracer molecules to diffuse across the bowel. The activation of genes for tight junction proteins was measured. We did not observed a significant change in the tightness or response of tight junction genes following the introduction of bacteria. Transient changes at about one week were observed in some genes, but these returned to germ-free levels. Colonization studies were also performed in a genetically altered mouse strain that develops colitis when colonized with normal bacteria (Interleukin-10 knockout). These animals showed significant changes in tight junction genes when they develop overt colitis. In conclusion, our data do not support a strong effect of bacteria on barrier properties. This may not be the final word because of potential limitations in our studies.