UBC Faculty Research and Publications

Requirement of epithelial integrin-linked kinase for facilitation of Citrobacter rodentium-induced colitis Assi, Kiran; Bergstrom, Kirk; Vallance, Bruce; Owen, David; Salh, Baljinder Sep 11, 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12876_2013_Article_1000.pdf [ 1.11MB ]
JSON: 52383-1.0223749.json
JSON-LD: 52383-1.0223749-ld.json
RDF/XML (Pretty): 52383-1.0223749-rdf.xml
RDF/JSON: 52383-1.0223749-rdf.json
Turtle: 52383-1.0223749-turtle.txt
N-Triples: 52383-1.0223749-rdf-ntriples.txt
Original Record: 52383-1.0223749-source.json
Full Text

Full Text

RESEARCH ARTICLE Open AccessRequirement of epithelial integrin-linked kinase forfacilitation of Citrobacter rodentium-induced colitisKiran Assi1, Kirk Bergstrom2, Bruce Vallance2, David Owen3 and Baljinder Salh1*AbstractBackground: Integrin-linked kinase (ILK) is a serine-threonine kinase that transduces extracellular matrix-relatedcues into intracellular signals, with fundamental roles in cell motility, development and cancer. Recently ILK beenshown to have an important role in bacterial epithelial cell attachment, through ILK-bacterial OspE binding. Herewe report on the role of epithelial derived ILK in response to Citrobacter rodentium infection.Methods: C. rodentium was administered to both control and intestinal epithelial cell ILK knockout mice.Histological inflammatory scores were assessed, and cytokines measured by ELISA as well as RT-PCR, in mousecolons. Bacterial colonization was determined by plating homogenates onto MacConkey agar, andimmunofluorescence microscopy performed using anti-LPS and anti-Tir antibodies.Results: ILK-ko mice exhibited reduced weight loss at 15 days post-infection (p < 0.01) and demonstrated reducedhistological inflammatory scores (p < 0.01), reduced CCL2 and pro-inflammatory cytokines. This was not due toreduced colonization, but was associated with an altered pattern of C. rodentium bacterial migration. Attenuatedfibronectin expression was found in the ILK-ko mice. C. rodentium exposure was shown to increase ILK expression incell lines, and in murine epithelium in vivo. In ILK-ko mice reduced activation of ser473Akt and reduced cryptproliferation, together with reduced cyclin D1 expression were observed.Conclusions: ILK influences the host response to C. rodentium -induced infection, independently of reducedcolonization in the ILK knockout mice. The reduced inflammation and dramatically attenuated hyperplastic cryptalresponse to infection in this group, are at least in part the result of, the reduction in CCL2 and cyclin D1 expressionrespectively.Keywords: ILK, Citrobacter, Colitis, FibronectinBackgroundGastrointestinal infection is an important cause of mortalityin the developing world and morbidity in the developedworld [1,2]. Although a variety of bacteria and viruses areknown to cause gastroenteritis, the underlying mechanismsinvolved remain unknown. Several barriers are knownto mitigate against intestinal infection and these includephysical defenses such as the surface mucus layer, cell-cell junctions, rapid epithelial cell turnover, the presenceof commensal bacteria, as well as the innate immune sys-tem responsible for the production of immunoglobulin A,defensins and resident immune cells [3,4]. Infection occurswhen organisms are successfully able to breach thesebarriers. Some of the most important organisms causingbacterial infection world-wide are Enteropathogenic andenterohaemorrhagic Escherichia coli (EPEC and EHECrespectively). An organism that is useful to study mechan-istic aspects of this process is C. rodentium, colonizationby which results in epithelial injury through the develop-ment of development of F-actin-rich pedestals, otherwiseknown as attaching and effacing lesions in mice [5,6]. Thisprocess is known to rely on a type III secretion systemused to inject bacterial effectors into host epithelium. Dueto similarities with human idiopathic inflammatory boweldisease, such as a predominant Th1 response, attended bythe elaboration of cytokines such as interferon gamma,the C. rodentium model has also been used to investigatemechanisms involved in that group of disorders.* Correspondence: bill.salh@ubc.ca1Division of Gastroenterology, Department of Medicine, The University ofBritish Columbia, 5th Floor, 2775 Laurel Street, V5Z 1M9 Vancouver, BC,CanadaFull list of author information is available at the end of the article© 2013 Kiran et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.Assi et al. BMC Gastroenterology 2013, 13:137http://www.biomedcentral.com/1471-230X/13/137ILK was first discovered as a beta 1 (β1) integrin bind-ing protein via a yeast 2-hybrid assay. Since then it hasbeen shown to play an important role in focal adhesionformation, which it achieves by complexing with Pinchand the Parvin proteins [7]. A significant body of workhas shown that ILK plays a role in tumor biology. Otherwork indicates a role in cardiac development, blastocystimplantation, skin, connective tissue, hepatic and gut de-velopment [8].ILK has been shown to be involved in the uptake ofStreptococcus pyogenes and other bacteria into epithelialcells [9]. More recent work has shown that host intes-tinal ILK may be subverted by Shigella flexnerii in orderto stabilize focal adhesions [10]. This facilitates blockingof cell detachment and hence the bacteria are able to gaina foothold for infection to proceed. An effector proteinOspE, which is conserved in enteropathogenic E. coli,Salmonella and C. rodentium, was shown to bind to ILKand co-localize with it at focal adhesions.It is presently unclear exactly how ILK within thegastrointestinal epithelium may modulate the bindingof, and response to infection with organisms such asC. rodentium. In our previous work we have shown thatconditional knockout of ILK in epithelial cells bluntsthe response to inflammation-induced cancer develop-ment in the colon [11]. We have also demonstrated thatepithelial ILK deficiency leads to attenuation of DSS-induced colitis, an effect that was associated with areduction in fibronectin expression, as well as an al-teration in the ratios of lymphocyte populations [12].In this work we have investigated how ILK deficiencyaffects the host response to C. rodentium infection.MethodsAnti-Tir and anti-LPS antibodies were obtained fromDr B Vallance; anti-fibronectin antibody from Abcam(Cambridge, MA); ILK, Akt, Gadph, actin, cyclin D andKi-67 antibodies from Santa Cruz (Santa Cruz, CA),ser473 Akt from Cell signaling, ILK si-RNA from Qiagen,Akt and ILK antibodies from Santa Cruz; anti-CD3.ELISA kits for TNFα, IFNγ, IL-10, CCL2 were obtainedfrom BD Biosciences (Mississauga, ON). Horse-radishperoxidase conjugated secondary antibodies were obtainedthrough Calbiochem (San Diego, CA). EGTA, EDTA,MOPS, PMSF, sodium orthovanadate, leupeptin, aprotinin,benzamidine, dithiothreitol and β-glycerolphosphate, werepurchased from Sigma.ILK-ko mice, disease activitiesWe have previously described our ILK knockout mice[11]. Briefly, mice on an FVB (Friend virus B-type) back-ground were kept in conventional housing in the animalcare facility at Jack Bell Research Centre. They were fedchow ad libitum and had liberal access to drinking water.All experiments were approved by the UBC Animal EthicsCommittee (A05-1580). Inactivation of ILK in colonicepithelial cells was achieved by crossing the Fabp –Cremice with ILKflox/flox animals. The resulting offspringwere then backcrossed with the homozygote floxed miceto generate the genotype: ILKflox/flox,Cre. Genotyping forCre and ILK were carried out as previously described.Briefly, tail DNA was obtained and the following primersused to detect Cre expression: 5′–CCTGGAAAATGCTTCTGTCCG–3′ and 5′- CAGGGTGTTATAACAATCCC-3′.ILK genotype was monitored using: 5′- CCAGGTGGCAGAGGTAAGTA-3′ and 5′-CAAGAAATAAGGTGAGCTTCAGAA-3′.For infection experiments mice were used out at ap-proximately 6 to 8 weeks of age. C rodentium (strain DBS100) was given by gavage as an inoculum of 2.5 × 108 bac-teria per 100 ul of Luria broth and the mice terminatedeither on day 6 or 15.Their colons were examined for macroscopic and,using hematoxylin and eosin, for microscopic diseaseactivity as previously described with some modifications[13]. After removal, the colons were fixed in 10% bufferedformalin for immunohistochemistry, protein lysates wereprepared for western analysis and ELISA. With referenceto disease activity scoring: A. Macroscopic assessmentof disease activity was scored from 0–4 as follows: 0, nosigns of inflammation, normal pellet, and from 1–4depending on the degree (slight, moderate, severe) ofliquidity of stool, presence of hyperemia and thicken-ing of the distal bowel, presence of blood, as well asthe degree of weight loss. B. Microscopic scores werescored by a gastroenterological pathologist with experi-ence in murine mucosal pathology (D Owen). This scoregrades the severity of the lesion from 0–4, based on theseverity of inflammation, the extent of inflammation(depending on mucosal → transmural inflammation),ulceration, crypt damage, and percentage involvementobserved across 5 different microscopic fields per mouse.SDS-polyacrylamide gel electrophoresisWestern immunoblotting was performed as previouslydescribed [12]; colonic tissue or cultured cells were ho-mogenized in buffer containing 20 mM MOPS, 150 mMNaCl, 50 mM β-glycerophosphate, 5 mM EGTA, 50 mMNaF, 1 mM DTT, 1 mM sodium vanadate, 0.5% NP40and 1 mM PMSF. After sonicating for 15 s (×2) and cen-trifuging at 14,000 RPM for 15 min, the protein concen-tration in the supernatant was determined by the Bradfordassay (Bio-Rad, Mississauga, Ont). 25 ug of protein fromeach sample was resolved using 10% SDS-PAGE beforetransferring to nitrocellulose membranes (Bio-Rad). Theblots were blocked in 5% skim milk in TBST (20 mMTris–HCl pH 7.4, 250 mM NaCl, 0.05% Tween-20) for 1 hbefore probing for 2 h using the appropriate primaryAssi et al. BMC Gastroenterology 2013, 13:137 Page 2 of 13http://www.biomedcentral.com/1471-230X/13/137antibody. The blots were washed with TBST for 10 minthree times, before being incubated with the appropriatesecondary antibody for 1 h. Following 3 further washes inTBST, they were developed using the enhanced chemilu-minescence detection system (ECL, Amersham, Montreal,Quebec).ImmunohistochemistryParaffin-embedded colonic tissue samples were de-waxedin xylene twice for 5 min, rehydrated in a series of ethanol(100% - 70%) for 3 min each followed by rehydration inPBS for 30 min. After rehydration the endogenous perox-idase was blocked with 0.3% hydrogen peroxide followedby antigen retrieval by microwaving sections in citrate buf-fer pH 6.0 (10 mM Na citrate). Following antigen retrieval,the sections were washed three times with PBS, blockedin 1% BSA for 1 h, and then stained using the VectastainABC kit (Vector laboratories, Burlingame, CA) men-tioned below according to manufacturer’s recommenda-tions but with the following modifications. Sectionswere incubated with the following primary antibodiesat 4°C overnight: cyclin D1 (Santa Cruz, CA), fibronec-tin (1:200, Abcam, Cambridge, MA), Ki-67. Followingincubation, the sections were rinsed three consecutivetimes with PBS and then incubated with the appropriatebiotinylated secondary antibody for 1 h followed byincubation with peroxidase-labelled streptavidin. Nova –red and DAB were used as the chromagens, and thesections were counterstained with haematoxylin. Threeblinded observers independently examined all stainedsections [12].For detection of Tir and LPS by immunofluorescence,the slides were processed as for IHC and the followingantibodies were used: Tir and LPS (Vallance). Sectionswere stained with Vectastain ABC elite kit and biotinylatedant-rabbit for DAPI, or eFluor650 Nanocrystal conjugationkit, cat no. 88-7072-98 antibody, and Avidin D-FITC (orAvidin-Texas Red) used for immunofluorescence (Vectorlaboratories, CA, USA). A Zeiss LSM-780 microscope wasused to capture images. Each section had its own controlusing the secondary antibody only. Pre-immune serum wasinitially used to ensure specificity of the signal with eachof the antibodies.Cell cultureHCT 116 cells were a kind gift of Bert Vogelstein (JohnsHopkins, Baltimore, Maryland) and were cultured inMcCoys 5A Medium (Gibco, Burlington, Ontario) con-taining 10% heat inactivated fetal bovine serum (FBS)(Hyclone, Logan, Utah). Protein lysates were obtainedusing homogenization buffer as described above.CMT93 cells are derived from a murine rectal cancerand were obtained from B Vallance (Vancouver, BC).They were cultured in DMEM containing 10% FBS and2 mM glutamine; experiments were performed when cellswere approximately 90% confluent.Semiquantitative RT –PCR1 ug of RNA, obtained using Trizol from HCT 116 cells(or murine colon), was reverse transcribed using randomhexamers (Perkin-Elmer Applied Biosystems, Branchburg,NJ) and 20 units of Moloney murine leukemia virusreverse transcriptase M-MLV (Invitrogene) in 20 μl oftotal volume at 25°C for 10 min and at 37°C for 60 min.The resulting first-strand complementary DNA (cDNA)was used as template for the semi- quantitative-PCR.Amplification of the following cDNAs was performedusing the primers listed: CCL2 (F):ATGCAGGTCCCTGTCATGCTTCTG; (R):CTAGTTCACTGTCACACTGGTCACTCC; b-actin(F):AGAGGGAAATCGTGCGTGAC; (R):CAATAGTGATGACCTGGCGGT. Relative quantificationof gene expression was performed using densitometry andbeta-actin as a control.Si-RNA-mediated knockdown of ILKThis was performed as described previously using a 21-merto transfect HCT116 cells, grown to 60% confluency, usingSilentfect (Biorad). Two separate ILK si-RNA and control(scrambled) sequences were purchased from Qiagen Inc(Mississauga, ON), and from Santa Cruz BiotechnologyCorporation Inc (Santa Cruz, Ca). Gene knockdown wasconfirmed using western blotting and Q-PCR.Bacterial countsAfter homogenization of either cecal or colonic tissue,or stool pellets, they were serially diluted. They werethen plated onto MacConkey agar plates and bacterialcolonies were enumerated after 1 day [14].Statistical analysisAll macroscopic and histological disease scores, as wellas cytokine levels were expressed as mean + SD, withp < 0.05 being considered significant using the Student’st-test (unpaired, two-tailed). Where indicated ANOVAwas performed with Tukey post-hoc testing.ResultsC. rodentium induces ILK and activates Akt in epithelial cellsOur first objective was to investigate if bacterial exposureof epithelial cells led to any change in the levels of expres-sion of ILK. CMT 93 cells were exposed to C. rodentiumand cells were harvested at the time points shown(Figure 1A). As the data indicates this led to an increasein the expression of ILK without affecting the levels ofAkt, an important kinase involved in cellular survival.However there is a clear increase in the intensity of theser473 Akt signal, the site known to be critically involvedin its activation. In order to address whether ILK wasAssi et al. BMC Gastroenterology 2013, 13:137 Page 3 of 13http://www.biomedcentral.com/1471-230X/13/137responsible for the activation of Akt we used HCT 116cells. We have found that these cells respond reliably toLPS and have been able to knock down ILK using si-RNAin this system. The data (Figure 1B) shows that knock-down of ILK attenuates the response of these cells to LPSinduced ser473 Akt activation. Collectively these findingsindicate that epithelial cells have the capacity to activateAkt via ILK.C. rodentium –induced colitis is attenuated in ILK-ko mice,and ILK is induced in response to infectionThe C. rodentium induced murine colitis model is a verywell characterized system to investigate host-microbialinteractions, as well as the ensuing inflammatory response.A mild colitis usually results after 10 to 15 days post-infection, which is usually accompanied by a mild degreeof weight loss [15], however a fatal form of the disease,characterized by severe inflammation has been describedin the FVB strain [16]. We investigated this in our ILK-komice in comparison with their littermate control mice.The data (Figure 2A) obtained from six mice per group,indicates that the weight-loss response is attenuated in theILK-ko mice. In Figure 2B there is a clear reduction ofmacroscopic inflammation in the knockout mice. Similarchanges were seen at the level of histological inflammation(Figure 2C). In order to verify that there was an inductionof ILK protein in response to C. rodentium in vivo weperformed a time course experiment. Using mucosalscrapes from mice terminated at days 1–3, and westernimmunoblotting, we observe an increase in ILK at be-tween 2–3 days (Figure 2D). A comparison of the levelsof ILK present in the 2 sets of mice used in these stud-ies, shows a difference at the end of 15 days (Figure 2E).Whilst both panels show positive staining within theimmune cells coursing between the crypts, there is aclear reduction of the ILK signal from the epithelialcells, in the representative ILK-ko example shown.C. rodentium induces CCL2 and macrophage infiltration;both of which are blunted in ILK-ko miceWe next characterized the levels of expression of CCL2in these groups. We have shown that this chemokine isreduced in another model of colitis induced by dextransodium sulfate. As we have previously shown that ILKmay impact on the level of expression of CCL2 expres-sion by RT-PCR in an epithelial system, this was investi-gated in this model also. Similar to our findings in theDSS-induced colitis, we showed a reduction of CCL2 inthe ILK-deficient mice, both at the level of message andprotein (Figure 3A/B). To determine whether or not thiswas associated with any change in infiltration of cells ofthe monocytic series, we performed immunohistochemistrywith the F4/80 antibody, which recognizes monocytes andmacrophages. The data clearly shows an impressive numberof cells in the submucosa of the wild-type mice, which aremissing in the ILK-ko example shown. Typical histologicalsections (Figure 3D) show the crypt hyperplastic responsetogether with inflammation in the wild-type mouse, whichis blunted in the ILK-ko section shown.ILK-ko mice have reduced levels of pro-inflammatorycytokinesMeasurement of key cytokines, representative of pro- andanti-inflammatory effectors, tumor necrosis factor alpha(TNFα), interferon gamma (IFNγ) and interleukin 10(IL-10), revealed that there were significant reductionsin TNFα and IFNγ in the knockout mice, which was asso-ciated with an increase in the level of IL-10 (Figure 4A-C).Figure 1 ILK is induced in response to C. rodentium exposure of CMT 93 cells, together with activation of Akt, and ILK regulates Aktactivation in response to LPS. A. Colonic CMT 93 cells were exposed to C. rodentium for the times indicated. Cellular protein was obtained bylysing cells in homogenization buffer, and western blotting performed for the molecules indicated. B. After knocking down ILK using si-RNA(or scrambled control oligo) HCT116 cells were stimulated with LPS (1 ug/ml) for 30 min and western immunoblotting performed. GAPDH andβ-actin were used as the loading controls.Assi et al. BMC Gastroenterology 2013, 13:137 Page 4 of 13http://www.biomedcentral.com/1471-230X/13/137Overall this data indicates that ILK deficiency specificallyin the epithelial cell component, directly or indirectly, isassociated with a reduced inflammatory cytokine responsedue to C. rodentium.ILK-ko mice have diminished Akt activation in responseto C. rodentiumNext we wanted to assess the impact of inflammation onthe expression of key ILK regulated proteins in the intes-tine. In accordance with the data in the cell lines, weobserved an attenuation of the activation of Akt in theILK-ko mice, as well as a reduction in the level of expres-sion of the transcription factor Snail, but not E-cadherin,using western blotting (Figure 5A/B). Previous work hasshown that expression of Snail and E-cadherin, can bemodulated by ILK in epithelial cells [17]; ILK impactingon Snail expression thereby de-repressing E-cadherinexpression. Furthermore we have shown a correlationbetween Snail and ILK-ko in a colitis - associated cancermodel, where the ILK-ko mice had reduced expressionof Snail in the neoplasms [11]. These findings indicatethat genetic deletion of ILK in epithelial cells results inattenuation of the Akt activation response and Snailexpression, although regulation of E-cadherin may bemore complex.C. rodentium binding to apical epithelium is unimpairedin ILK-ko miceBased on these findings we hypothesized that perhapsthe blunted inflammatory and response to C. rodentiumwas due to impaired epithelial binding in the knockoutmice. By employing a bacterial plating assay we measuredFigure 2 ILK-ko mice are protected from C. rodentium induced wasting disease and inflammation, and exhibit reduced expression ofthe chemokine CCL2. A. Colitis was induced in 6 animals per group by administering C. rodentium by gavage. Animal weights were recordeddaily. There was an observed significant difference in their weights from days 13–15 (p < 0.05). B. Assessment of macroscopic damage in ILK-komice (day 15) indicated a significant reduction in comparison to control mice (**p < 0.01). C. Microscopic damage scores (day 15) were determinedusing criteria outlined in Materials and methods, based on severity and extent of inflammation. The data indicate there was reduced damage in theILK-ko mice. D. ILK expression was assessed in response to Citrobacter at the time points indicated using western immunoblotting. The representativedata indicated an increase in ILK expression after 48–72 hours (b-actin is shown as a loading control). E. ILK immunohistochemistry is shown 15 dayspost-infection in a wild-type and ILK-ko sample. There is a reduction in the ILK signal intensity in the epithelial cells as compared with the comparableintensity in the infiltrating immune cells.Assi et al. BMC Gastroenterology 2013, 13:137 Page 5 of 13http://www.biomedcentral.com/1471-230X/13/137the levels of bacteria in the cecum and colon, in both theluminal contents and in the mucosal lining. As the data inFigure 6 (A/B) indicate there appears to be no differencein the levels of bacteria in the two sets of mice, regard-less of whether the colonic (or cecal) contents or tissuewere examined. This clearly indicates that the responsesobserved must be occurring at or distal to the epithelialcell system in the mice, and importantly, that the responseswere not simply due to a failure of the C. rodentiumto bind.C. rodentium cryptal migration is impaired in ILK-ko miceIn order to verify that bacterial binding occurred in asimilar distribution in the two sets of mice we performedimmunofluorescence using previously described antibodies,one against C. rodentium LPS (green) and the other againstTIR (red); the latter being a widely recognized method fordemonstrating C. rodentium infection. The upper panels(Figure 6C) depict representative staining in the wild-typemice. As can be seen bacterial binding occurs at the apicalsurface and appears to migrate downwards along the lateralaspects of the crypts. In the lower panels for the ILK-komice (Figure 6D), it is evident that bacteria are able to bindto the apical region, however, there does not appear tobe a comparable signal migrating downwards. This maybe a factor in the reduced inflammatory response ob-served. In order to determine whether the response wasbeing delayed in the ILK-ko mice we repeated the ex-periment at day 14. The data again show no significantdifferences in bacterial binding, or any increased migra-tion at this later time-point (Figure 7A-C). This intri-guing observation at 2 different time-points, indicates thatABCDFigure 3 ILK-ko mice exhibit reduced expression of the chemokine CCL2, and reduced monocyte-macrophage influx. A. CCL2 induction(RT-PCR, message compared to that of β-actin) was significantly (**p < 0.01) reduced in the ILK-ko group. B. Distal colonic lysates (pooled) wereused to determine CCL2 levels by ELISA. The data are for 6 mice per group and show a reduction in levels of CCL2 in the ILK-ko mice (*p < 0.01).C. IHC for F4/80 which is a marker for cells of the monocyte/macrophage lineage shows reduced numbers in the ILK-ko mice. D. Histology of arepresentative section from each of the WT and ILK-ko mice shows less inflammation and hyperplasia in the ILK-ko example shown. Notably,there are elongated crypts seen in the ILK-ko mice as can be seen to the right side of the ILK-ko panel.Assi et al. BMC Gastroenterology 2013, 13:137 Page 6 of 13http://www.biomedcentral.com/1471-230X/13/137Figure 4 ILK-ko mice express reduced pro-inflammatory cytokines. A-C. Distal colonic lysates (pooled) were used to determine TNF alpha,IL-10 and IFN gamma levels by ELISA. The data are for 6 mice per group and show a reduction in levels of TNF alpha and IFN gamma, as well asan increase in IL-10 in the ILK-ko mice (*p < 0.05).Figure 5 ILK-ko mice have reduced Akt activation, and reduced expression of Snail, but not E-cadherin after C. rodentium inducedcolitis. A. Lysates prepared from distal colonic homogenates were resolved using Western analysis and the resulting membranes probed withthe antibodies indicated (4 mice from ILK-ko and littermate controls). B. Densitomtry was performed and the resulting data is depicted in thebarcharts (Akt, Snail and E-cadherin).Assi et al. BMC Gastroenterology 2013, 13:137 Page 7 of 13http://www.biomedcentral.com/1471-230X/13/137Figure 6 (See legend on next page.)Assi et al. BMC Gastroenterology 2013, 13:137 Page 8 of 13http://www.biomedcentral.com/1471-230X/13/137an ILK-dependent mechanism, following bacterial bind-ing, facilitates the intercryptal migration of bacteria.C. rodentium –associated hyperplasia is reduced inILK-ko miceAs one of the key mechanisms involved in prevention ofbacterial adherence by epithelia is related to increasedepithelial cell turnover, we explored cellular proliferationusing Ki-67 immunohistochemistry, in the ILK-ko andlittermate control wild-type mice. As the representativepictures show there is clearly more enhanced proliferationin the wild-type versus the knockout mice (Figure 8A/B).The data in the barchart (Figure 8C) show the positivestaining in the knockout mice is less than half of that seenin the wild-type mice. The crypt heights were measuredand there was a clear reduction at both the 6 and 15 daytime-points, in the ILK-ko mice. Interestingly, we notedan increase in the crypt height between the 2 time-points in the ILK-ko mice indicating that perhaps adelayed response to the bacteria was occurring. How-ever as the data in Figure 8 indicate, there was no differ-ence in the binding/distribution patterns at this latertimepoint.In order to determine the specific role of cyclin D1, atarget of ILK, in this response we utilized immunohisto-chemistry. The data indicates (Figure 9) that the level ofcyclin D is reduced in the ILK-ko mice, a finding inkeeping with changes observed and previously reportedby us in the ILK-ko mice in response to induction ofcancer-associated colitis.Reduced C rodentium –induced fibronectin expression inILK-ko miceThe extracellular matrix may be an important determin-ant of bacterial ability to infect epithelial systems. In thisregard a number of different bacteria utilize fibronectin toaid binding and/or invasion, whilst some are capable ofexpressing fibronectin-binding proteins [18]. As others,(See figure on previous page.)Figure 6 Effects observed in ILK-ko mice are not due to impaired binding of C rodentium. ILK-ko and control mice were gavaged withC. rodentium and their cecal and colonic tissues assessed for levels of C. rodentium by plating homogenates and counting the number ofcolonies, in both the colon (A) and cecum (B). The tissues were also processed for immuno-fluorescence and probed with antibodies to LPS(green) and Tir (red), and overlayed (yellow). As can be seen bacteria are capable of binding in the ILK-ko mice but may be impaired in theirability to migrate downwards in between the crypts (C/D).Figure 7 Effects observed in ILK-ko mice are preserved at day 14 post-infection. ILK-ko and control mice were gavaged with C. rodentiumand their cecal and colonic tissues assessed (day 14) for levels of C. rodentium by plating homogenates and counting the number of colonies asfor Figure 6 (A/B). The tissues were also processed for immuno-fluorescence in the same manner (C/D). It can be seen that although there arestill equivalent amounts of bacterial bainding, bacteria are still unable to migrate downwards in between the crypts, in the ILK-ko mice at thissecond time-point.Assi et al. BMC Gastroenterology 2013, 13:137 Page 9 of 13http://www.biomedcentral.com/1471-230X/13/137besides ourselves, have shown that ILK is involved in epi-thelial expression of fibronectin [19], we explored the pos-sibility that this was the case in this system also. We haveshown using immunohistochemistry that in the DSS-induced colitis model there is significantly less fibronectinexpression in the ILK-ko mice. When we assessed this inthe C. rodentium -induced colitis model we observe asimilar finding, specifically that fibronectin expression isdownregulated in the ILK-ko mice (Figure 10). This maybe another mechanism to explain the reduced migrationof the bacteria downwards between the crypts.DiscussionIn this report we have added to our understanding ofthe role of ILK in intestinal pathophysiology, specificallyin the setting of bacterial infection. Similar to the find-ings reported for DSS-induced colitis we show that thereis a reduced inflammatory response, associated with areduction in CCL2 expression, an important immunecell chemoattractant. Furthermore, our findings indicatethat the pattern but not the magnitude of epithelial C.rodentium binding is altered in the ILK-ko mice, withpreserved apical binding but reduced lateral epithelialcell binding/migration, in between adjacent crypts. Al-though the reduced expression of fibronectin may ac-count for this finding, we cannot exclude alterations inother extracellular matrix components, or alterations inthe levels of other cell surface integrins as being involvedin this response. An additional role for ILK is indicatedby the reduced crypt hyperplasia observed associatedwith decreased cyclin D1 on immunohistochemistry, inILK-ko mice.Bacteria utilize multiple mechanisms to gain a footholdfacilitating their colonization, and several of these involvecomponents of the extracellular matrix and integrin net-work [20]. The production of bacterial fibronectin-binding proteins (FnBPs) enables pathogens to bind hostcell integrins through a fibronectin bridge. Organismssuch as Yersinia pseudotuberculosis and Shigella flexneriundergo ingestion via integrin receptors. Streptococcuspyogenes uses M1 or PrtF1 surface proteins bound tofibronectin, to facilitate invasion via the α5β1 integrinreceptor. As ILK interacts with the cytoplasmic domainof β1 integrin, this places ILK in a unique position tomediate downstream signaling. In the case of C. rodentiumwe can demonstrate equivalent levels of apical epithelialFigure 8 Reduced proliferation in ILK-ko mice in response to C. rodentium. infection. A, B. Sections were processed for immunohistochemistryusing an antibody to Ki-67 at the time-points indicated. The photomicrographs are representative of data from 5 mice. C. The number of Ki-67 positivecells per crypt were counted in 5 HPFs at days 6 and 15. D. Crypt heights were measured using a micrometer in 5 different fields, in each of 5 miceper group.Assi et al. BMC Gastroenterology 2013, 13:137 Page 10 of 13http://www.biomedcentral.com/1471-230X/13/137binding but downwards basolateral migration of thebacterium is impaired. This leads us to suspect thatother receptors (possibly ILK-dependent integrins) arerequired for this. Future work will attempt to examinewhich integrins are specifically regulated by ILK in theintestinal epithelium.OspE is a virulence factor common to several pathogens,including C. rodentium, EPEC, EHEC and Salmonella spe-cies, and is injected into host cells by the type III secretionsystem [21]. This has recently been shown to bind to ILK,resulting in an increase in the number of focal adhesions;OspE and ILK were shown to co-localize at focal adhesionsthrough vinculin staining. It also caused an increase in thecell surface levels of β1 integrin, whilst at the same timereducing phosphorylation of focal adhesion kinase (FAK)and paxillin [10]. Together, this results in stabilizationof focal adhesions and thereby facilitates bacterial celladherence, through an attenuation of cell shedding. Ourfindings do not indicate a specific defect in bacterial ad-herence in ILK-ko mice, indicating that for C. rodentiumthis may not be a critical event. However there were cleardifferences in bacterial migration, indicating other (bacterial-mediated) ILK-dependent events that may be important.We cannot exclude the possibility that the effects weobserve are the sum of two opposing events, the firstdue to a primary ILK-dependent reduction in cellularturnover thereby facilitating bacterial binding, and thesecond due to a reduction in ILK and OspE mediatedreduction in bacterial binding.Recent work has characterized the role of anotherimportant matrix protein, osteopontin (OPN) in thedevelopment of murine intestinal inflammation [22]. Thishas been previously shown to be upregulated in inflamma-tory bowel disease, but the exact significance of thisobservation remains unknown. In that work, OPN wasinduced in response to infection with C. rodentium, and itwas noted that mice lacking OPN were colonized to aA CBFigure 10 Reduced fibronectin expression in response to C rodentium infection in ILK-ko mice. After performing immunohistochemistryas described in Methods, fibronectin scoring was performed by examination of 5 HPFs in 4 mice from each group, using the following criteria:0- no signal; 1- less than 10% of field, low intensity; 2-between 10-50% of field, moderate intensity; 3- over 50% of field, highest intensity. In theexample shown, the ILK-ko has been scored as 1, and for the WT sample, 2 (A/B). The overall data is shown in C.Figure 9 Reduced cyclin D1 expression in ILK-ko mice inresponse to C rodentium infection. Sections were processed forimmunohistochemistry using an antibody to cyclin D1. The numberof positively staining nuclei in 5 crypts per mouse in each of 5 micewere counted at days 6 and 15 (black: WT, grey:ILK-ko).Assi et al. BMC Gastroenterology 2013, 13:137 Page 11 of 13http://www.biomedcentral.com/1471-230X/13/137significantly reduced degree as compared with littermatecontrols. Consequently reduced pedestal formation andepithelial proliferation were observed and the former wasreversed by the administration of human OPN. Thisindicates that bacteria have varying degrees of depend-ence on extracellular matrix components in facilitatingtheir colonization, since apically at least we observedequivalent levels of binding (Figures 6 and 7), despitereduced fibronectin levels in ILK-ko mice.Interestingly, infection with C. rodentium has not beenassociated with profound changes in apoptosis. This maybe related to activation of the phosphatidylinositol 3-kinase (PI3K), a growth factor and TNFα -activated lipidkinase, which is associated with a cellular survival response.Using a pharmacological inhibitor Ly294002 it has beenshown that PI3K is required for the host response forbacterial clearance, as well as the epithelial proliferativeresponse [23]. This was reported to occur without anychanges in inflammation. As previous work indicatesthat ILK is downstream of PI3K, some of our observa-tions mirror these findings (reduced hyperplasia) whilstothers may be dissociated from PI3K, most notably, thereduced inflammation in conjunction with the effect onthe extracellular matrix.Presently it is not known what specific molecules areinvolved in the sensing of epithelial damage and theresulting effectors of epithelial proliferation (and/or re-pair). The reduced epithelial proliferation consequentupon C. rodentium infection observed in our study maybe due to 2 reasons. Firstly, because ILK is involved inthe regulation of cyclin D1 this may be a direct effect atthe level of the epithelial cell and independent of anybacterial-mediated mechanism [24]. As β-catenin is acti-vated in response to C rodentium infection [25], and itscasein kinase 1 (CK1) -mediated serine phosphorylationon residue 45 appears to coincide with hyperplasia [26],it is likely that cyclin D1 is activated directly in responseto this. However in the FVB strain mice used in our workwe were unable to demonstrate nuclear localization ofβ-catenin at either of the time-points investigated (days6 or 15). The second reason may be an indirect one, andit is hypothesized to occur via the inability of the bac-teria to migrate downwards in the ILK-ko in a similarmanner to that observed in the wild-type mice.Using Swiss Webster mice it has been shown thatNF-κB is activated in response to infection withC. rodentium, and importantly that this is not corre-lated with the observed hyperplastic response. This wasinitially reported with the NF-κB inhibitor Velcade, whichwas demonstrated to inhibit NF-κB activation only [27].More recently it has been shown that Mek, a componentof the TLR4-Mek-MAPK pathway may mediate activationof NF-κB in vivo, demonstrated using the Mek1/2 inhibi-tor PD98059 [28].Our study raises some important questions about therole of ILK in intestinal physiology and pathophysiology.We, and others have shown that ILK is upregulated atthe protein level in intestinal and other tumors, indicat-ing a role in tumorigenesis [29-31]. This is supported bya wealth of data regarding ILK’s role in various proper-ties fundamental for cancer development such as prolif-eration, avoidance of apoptosis, angiogenesis and EMT.Our previous work using a colitis-associated cancer modelshowed a trend towards smaller tumors in ILK-ko micethat, was accompanied by a reduction in both cyclin Dand Snail expression. This has been replicated in themodel described in this report, which is also character-ized by profound changes in cellular proliferation, indi-cating an important role for ILK in these two processesin the intestine. The reduction in Snail expression,which has been linked to EMT [32], in our ILK-ko miceis also of interest as the FVB strain of mice are knownto undergo more fibrosis [15], and this is attenuated inthe ILK-ko mice.ConclusionsOur findings indicate that C rodentium -induced colitisis impaired in mice lacking expression of ILK within thecolonic epithelium. This appears to be dependent upon,or at least associated with, a reduction in epithelial pro-liferation as well as a reduction in inflammation. How-ever, the observed effects do not appear to be related toimpaired bacterial binding to the apical epithelium.Competing interestsThe authors declare that they have no competing interests.Authors’ contributionsKA performed the cell culture and mouse studies, as well as the statisticalanalyses. KB carried out the bacterial counts. BS conceived and coordinatedproject and wrote the manuscript. DO carried out the histologicalassessments. BV helped with experimental design. All authors read andapproved the final manuscript.GrantsThis study was supported in part by funding from the Canadian Society ofIntestinal Research, the Geraldine Dow and Ferry Dow Endowment for IBDresearch, and BS received support from a Vancouver Coastal Health ResearchInstitute In-it-for life Clinician-Scientist award.Author details1Division of Gastroenterology, Department of Medicine, The University ofBritish Columbia, 5th Floor, 2775 Laurel Street, V5Z 1M9 Vancouver, BC,Canada. 2Child and Family Research Institute, Vancouver, Canada.3Anatomical Pathology, Vancouver V5Z 1M9, BC, Canada.Received: 8 January 2013 Accepted: 6 September 2013Published: 11 September 2013References1. Guarino A, Dupont C, Gorelov AV, Gottrand F, Lee JK, Lin Z, Lo Vecchio A,Nguyen TD, Salazar-Lindo E: The management of acute diarrhea inchildren in developed and developing areas: from evidence base toclinical practice. Expert Opin Pharmacother 2012, 13(1):17–26.2. Getto L, Zeserson E, Breyer M: Vomiting, diarrhea, constipation, andgastroenteritis. Emerg Med Clin North Am 2011, 29(2):211–237.Assi et al. BMC Gastroenterology 2013, 13:137 Page 12 of 13http://www.biomedcentral.com/1471-230X/13/1373. Sansonetti PJ: War and peace at mucosal surfaces. Nat Rev Immunol 2004,4(12):953–964.4. John LJ, Fromm M, Schulzke JD: Epithelial barriers in intestinalinflammation. Antioxid Redox Signal 2011, 15(5):1255–1270.5. Vallance BA, Deng W, Knodler LA, Finlay BB: Mice lacking T and Blymphocytes develop transient colitis and crypt hyperplasia yet sufferimpaired bacterial clearance during citrobacter rodentium infection.Infect Immun 2002, 70(4):2070–2081.6. MacDonald TT, Frankel G, Dougan G, Goncalves NS, Simmons C: Hostdefences to citrobacter rodentium. Int J Med Microbiol 2003, 293(1):87–93.7. Wickstrom SA, Lange A, Montanez E, Fassler R: The ILK/PINCH/parvincomplex: the kinase is dead, long live the pseudokinase! EMBO J 2010,29(2):281–291.8. McDonald PC, Fielding AB, Dedhar S: Integrin-linked kinase–essential rolesin physiology and cancer biology. J Cell Sci 2008, 121(Pt 19):3121–3132.9. Wang B, Yurecko RS, Dedhar S, Cleary PP: Integrin-linked kinase is anessential link between integrins and uptake of bacterial pathogens byepithelial cells. Cell Microbiol 2006, 8(2):257–266.10. Kim M, Ogawa M, Fujita Y, Yoshikawa Y, Nagai T, Koyama T, Nagai S, Lange A,Fassler R, Sasakawa C: Bacteria hijack integrin-linked kinase to stabilize focaladhesions and block cell detachment. Nature 2009, 459(7246):578–582.11. Assi K, Mills J, Owen D, Ong C, St Arnaud R, Dedhar S, Salh B: Integrin-linkedkinase regulates cell proliferation and tumour growth in murine colitis-associated carcinogenesis. Gut 2008, 57(7):931–940.12. Assi K, Patterson S, Dedhar S, Owen D, Levings M, Salh B: Role of epithelialintegrin-linked kinase in promoting intestinal inflammation: effects onCCL2, fibronectin and the T cell repertoire. BMC Immunol 2011, 12:42.13. Assi K, Pillai R, Gomez-Munoz A, Owen D, Salh B: The specific JNK inhibitorSP600125 targets tumour necrosis factor-alpha production and epithelialcell apoptosis in acute murine colitis. Immunology 2006, 118(1):112–121.14. Bergstrom KS, Kissoon-Singh V, Gibson DL, Ma C, Montero M, Sham HP, Ryz N,Huang T, Velcich A, Finlay BB, et al: Muc2 Protects against lethal infectiouscolitis by disassociating pathogenic and commensal bacteria from thecolonic mucosa. PLoS Pathog 2010, 6(5):e1000902.15. Luperchio SA, Schauer DB: Molecular pathogenesis of citrobacterrodentium and transmissible murine colonic hyperplasia. Microbes Infect2001, 3(4):333–340.16. Borenshtein D, Nambiar PR, Groff EB, Fox JG, Schauer DB: Development offatal colitis in FVB mice infected with citrobacter rodentium. Infect Immun2007, 75(7):3271–3281.17. Tan C, Costello P, Sanghera J, Dominguez D, Baulida J, de Herreros AG,Dedhar S: Inhibition of integrin linked kinase (ILK) suppresses beta-catenin-Lef/Tcf-dependent transcription and expression of the E-cadherinrepressor, snail, in APC−/− human colon carcinoma cells. Oncogene 2001,20(1):133–140.18. Joh D, Wann ER, Kreikemeyer B, Speziale P, Hook M: Role of fibronectin-binding MSCRAMMs in bacterial adherence and entry into mammaliancells. Matrix Biol 1999, 18(3):211–223.19. Gagne D, Groulx JF, Benoit YD, Basora N, Herring E, Vachon PH, Beaulieu JF:Integrin-linked kinase regulates migration and proliferation of humanintestinal cells under a fibronectin-dependent mechanism. J Cell Physiol2010, 222(2):387–400.20. Hoffmann C, Ohlsen K, Hauck CR: Integrin-mediated uptake of fibronectin-binding bacteria. Eur J Cell Biol 2011, 90(11):891–896.21. Krachler AM, Woolery AR, Orth K: Manipulation of kinase signaling bybacterial pathogens. J Cell Biol 2011, 195(7):1083–1092.22. Wine E, Shen-Tu G, Gareau MG, Goldberg HA, Licht C, Ngan BY, Sorensen ES,Greenaway J, Sodek J, Zohar R, et al: Osteopontin mediates citrobacterrodentium-induced colonic epithelial cell hyperplasia and attaching-effacing lesions. Am J Pathol 2010, 177(3):1320–1332.23. Brown JB, Cheresh P, Goretsky T, Managlia E, Grimm GR, Ryu H, Zadeh M,Dirisina R, Barrett TA: Epithelial phosphatidylinositol-3-kinase signaling isrequired for beta-catenin activation and host defense against citrobacterrodentium infection. Infect Immun 2011, 79(5):1863–1872.24. D’Amico M, Hulit J, Amanatullah DF, Zafonte BT, Albanese C, Bouzahzah B,Fu M, Augenlicht LH, Donehower LA, Takemaru K, et al: The integrin-linkedkinase regulates the cyclin D1 gene through glycogen synthase kinase3beta and cAMP-responsive element-binding protein-dependentpathways. J Biol Chem 2000, 275(42):32649–32657.25. Sellin JH, Umar S, Xiao J, Morris AP: Increased beta-catenin expression andnuclear translocation accompany cellular hyperproliferation in vivo.Cancer Res 2001, 61(7):2899–2906.26. Sellin JH, Wang Y, Singh P, Umar S: Beta-catenin stabilization impartscrypt progenitor phenotype to hyperproliferating colonic epithelia. ExpCell Res 2009, 315(1):97–109.27. Wang Y, Xiang GS, Kourouma F, Umar S: Citrobacter rodentium-inducedNF-kappaB activation in hyperproliferating colonic epithelia: role of p65(Ser536) phosphorylation. Br J Pharmacol 2006, 148(6):814–824.28. Chandrakesan P, Ahmed I, Anwar T, Wang Y, Sarkar S, Singh P, Peleg S,Umar S: Novel changes in NF-{kappa}B activity during progression andregression phases of hyperplasia: role of MEK, ERK, and p38. J Biol Chem2010, 285(43):33485–33498.29. Marotta A, Parhar K, Owen D, Dedhar S, Salh B: Characterisation ofintegrin-linked kinase signalling in sporadic human colon cancer. Br JCancer 2003, 88(11):1755–1762.30. Schaeffer DF, Assi K, Chan K, Buczkowski AK, Chung SW, Scudamore CH,Weiss A, Salh B, Owen DA: Tumor expression of integrin-linked kinase(ILK) correlates with the expression of the E-cadherin repressor snail: animmunohistochemical study in ductal pancreatic adenocarcinoma.Virchows Arch 2010, 456(3):261–268. Epub 2010 Jan 21.31. Hannigan G, Troussard AA, Dedhar S: Integrin-linked kinase: a cancertherapeutic target unique among its ILK. Nat Rev Cancer 2005, 5(1):51–63.32. Thiery JP, Acloque H, Huang RY, Nieto MA: Epithelial-mesenchymaltransitions in development and disease. Cell 2009, 139(5):871–890.doi:10.1186/1471-230X-13-137Cite this article as: Assi et al.: Requirement of epithelial integrin-linkedkinase for facilitation of Citrobacter rodentium-induced colitis. BMCGastroenterology 2013 13:137.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitAssi et al. BMC Gastroenterology 2013, 13:137 Page 13 of 13http://www.biomedcentral.com/1471-230X/13/137


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items