DIABETES-INSULIN-GLUCAGON-GASTROINTESTINAL

The Intestinal Epithelial Insulin-Like Growth Factor-1 Receptor Links Glucagon-Like Peptide-2 Action to Gut Barrier Function Charlotte X. Dong, Wen Zhao, Chloe Solomon, Katherine J. Rowland, Cameron Ackerley, Sylvie Robine, Martin Holzenberger, Tanja Gonska, and Patricia L. Brubaker Departments of Physiology (C.X.D., W.Z., C.S., K.J.R., P.L.B.), Pediatrics (T.G.), and Medicine (P.L.B.), University of Toronto, Toronto, Ontario M5S 1A8 Canada; Department of Paediatric Laboratory Medicine (C.A.), and Physiology and Experimental Medicine, Research Institute (T.G.), The Hospital for Sick Children, Toronto M5G 1X8, Canada; Centre National de la Recherche Scientifique (S.R.), Institut Curie, Paris, 75248 France; and Inserm (M.H.), Hôpital St-Antoine, Paris, 75005 France

Glucagon-like peptide-2 (GLP-2) is an intestinal growth-promoting hormone used to treat short bowel syndrome. GLP-2 promotes intestinal growth through a mechanism that involves both IGF-1 and the intestinal-epithelial IGF-1 receptor (IE-IGF-1R). GLP-2 also enhances intestinal barrier function, but through an unknown mechanism. We therefore hypothesized that GLP-2– enhanced barrier function requires the IE-IGF-1R and is mediated through alterations in expression and localization of tight junction proteins. Conditional IE-IGF-1R–null and control mice were treated with vehicle or degradation-resistant Gly2-GLP-2 for 10 days; some animals also received irinotecan to induce enteritis. Mice were then examined for gastrointestinal permeability to 4-kDa fluorescein isothiocyanate-dextran, jejunal resistance using Ussing chambers, tight junction structure by electron microscopy, and expression and localization of tight junction proteins by immunoblot and immunohistofluorescence, respectively. GLP-2 treatment decreased permeability to 4-kDa fluorescein isothiocyanate-dextran and increased jejunal resistance (P ⬍.05–.01), effects that were lost in IE-IGF-1R–null mice. Electron microscopy did not reveal major structural changes in the tight junctions in any group of animals. However, the tight junctional proteins claudin-3 and -7 were upregulated by GLP-2 in control (P ⬍.05–.01) but not null mice, whereas IE-IGF-1R deletion induced a shift in occludin localization from apical to intracellular domains; no changes were observed in expression or distribution of claudin-15 and zona occludins-1. Finally, in irinotecan-induced enteritis, GLP-2 normalized epithelial barrier function in control (P ⬍ .05) but not knockout animals. In conclusion, the effects of GLP-2 on intestinal barrier function are dependent on the IE-IGF-1R and involve modulation of key components of the tight junctional complex. (Endocrinology 155: 370 –379, 2014)

lucagon-like peptide-2 (GLP-2) is a nutrient-dependent intestinal L-cell hormone that promotes intestinal growth and function (1, 2). Teduglutide, a long-acting analog of GLP-2, has recently been approved for the treatment of adult short bowel syndrome (3) and is in clinical trials for Crohn’s disease (4). Because of the advances related to the clinical application of GLP-2, it is crucial to delineate the mechanisms of action of this intestinal hormone.

G

Nutrient-induced GLP-2 secretion follows a biphasic pattern, with a rapid increase followed by a more prolonged response (5). After release into the circulation, GLP-2 exerts its actions through the GLP-2 receptor (GLP-2R), which is expressed predominantly in the intestinal tract (6 – 8). The major biological roles ascribed to endogenous GLP-2 are the promotion of basal intestinal growth and of adaptive intestinal regrowth after a period

ISSN Print 0013-7227 ISSN Online 1945-7170 Printed in U.S.A. Copyright © 2014 by the Endocrine Society Received September 17, 2013. Accepted November 1, 2013. First Published Online November 21, 2013

Abbreviations: EM, electron microscopy; FD4, 4-kDa fluorescein isothiocyanate-dextran; GLP-2, glucagon-like peptide-2; GLP-2R, GLP-2 receptor; IE-IGF-1R, intestinal-epithelial IGF-1 receptor; KO, knockout; PI3K, phosphoinositide 3-kinase; ZO-1, zona occludins-1.

370

endo.endojournals.org

Endocrinology, February 2014, 155(2):370 –379

doi: 10.1210/en.2013-1871

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 July 2014. at 12:38 For personal use only. No other uses without permission. . All rights reserved.

doi: 10.1210/en.2013-1871

A

KO

-ve 100bp ladder

Animals Animals were housed in an animal facility with a 12-hour light, 12-hour dark cycle at the University of Toronto. All studies were approved by the University of Toronto Animal Care Committee. As described previously (20), IE-IGF-1R– knockout (KO) mice were generated by crossing villin-CreERT2⫹/0 and Igf1rflox/ flox mice, both on a C57BL/6 background (24, 25). The villinCreERT2⫹/0;Igf1rflox/⫹ offspring were then backcrossed to Igf1rflox/flox mice to generate the villin-CreERT2⫹/0;Igf1rflox/flox

B

** 600 Crypt-to-villus height (μm)

Control KO

390bp 312bp

C

**

500 400 300 200 100 0

Control

KO

D 6

120

* Jejunal resistance (Ω·cm2)

5

* 4 3 2 1

Materials and Methods

371

animals. Mice were genotyped by detection of the floxed allele in control animals (5⬘-ATCTTGGAGTGGTTGGGTCTGTTT-3⬘ and 5⬘-ATGAATGCTGGTGAGGGTTGTCTT-3⬘) and by the additional presence of the Cre allele in CreERT2⫹/0;Igf1rflox/flox animals (5⬘-CCTGGAAAATGCTTCTGTCCG-3⬘ and 5⬘-CAG GGTGTTATAAGCAATCCCC-3⬘; Figure 1A). Age-matched (8 –13 weeks) and sex-matched littermate Igf1rflox/flox control and IE-IGF-1R–KO mice were used in all experiments. IE-IGF-1R exon 3 cleavage was induced by ip injection of tamoxifen (100 ␮L of 10 mg/mL in ethanol/sunflower oil; MP Biomedicals) for 5 days (20); control animals were also injected with tamoxifen. After the fifth day of induction, 3 protocols were followed to study mice under healthy conditions or in models of altered barrier function: 1) enteritis induced by the topoisomerase I inhibitor irinotecan hydrochloride and 2) streptozotocininduced diabetes. For healthy animals, female and male mice were combined. Animals were injected sc with a pharmacological dose of a degradation-resistant GLP-2 analog (0.1 ␮g/g h(Gly2)GLP-2; American Peptide Company) or vehicle (PBS) daily for 10 days, with a final injection 3 hours before assessment, as previously reported (19, 20). For the enteritis model, only males were included because of higher sensitivity to irinotecan treatment in preliminary studies (data not shown). These animals received 10 days of GLP-2 or vehicle, as above, followed by 2 daily ip injections of 0.15 ␮g/g irinotecan (SigmaAldrich), based on the results of a preliminary study in which 8 mice were treated with a lower dose than that used in (17) to prevent animal mortality. In vivo and ex vivo permeability

Plasma FD4 (μg/ml)

of fasting or in response to enteritis, such as induced by the chemotherapeutic agent irinotecan (9 –11). However, exogenously administered GLP-2 also exerts profound effects to expand the crypt-villus epithelium through enhanced proliferation and survival as well as to increase nutrient digestion, absorption, and blood flow (2, 7, 12– 14). Importantly, GLP-2 also improves intestinal barrier function, an effect that has been observed under conditions of health as well as in models of disease (11, 15–17). Within the intestine, the GLP-2R is expressed in rare endocrine cells, enteric neurons, and subepithelial myofibroblasts but is not localized to target tissues such as the crypt epithelium and enterocytes (6 – 8). It has thus been established that GLP-2 exerts its tropic effects on the intestine indirectly, through pathways that involve ErbB, keratinocyte growth factor, and IGF-1 (8, 18, 19). In addition to IGF-1, we have further demonstrated that the IGF-1 receptor expressed in the intestinal epithelium (intestinal-epithelial IGF-1 receptor [IE-IGF-1R]) is required for the promotion of mucosal growth by GLP-2 (20). Interestingly, one study has shown that GLP-2R signaling is also required for prebiotic-induced increases in intestinal barrier function and that these effects are mediated, at least in part, via tight junctional proteins (16). The tight junctions form a semipermeable/selective paracellular barrier between the contents of the intestinal lumen and the basolateral side of the intestinal epithelium (21). Hence, understanding the mechanism underlying the barrier effects of GLP-2 has implications with respect to conditions associated with disruption of this protective layer. No mediator has been identified to date for the barrier effects of GLP-2. However, because IGF-1 is known to enhance intestinal barrier function (22, 23), we hypothesized that IGF-1 signaling is required for the intestinal barrier effects of GLP-2. We now provide evidence that GLP-2 requires the IE-IGF-1R to modulate intestinal barrier function and that these effects are induced, at least in part, through changes in the expression as well as the distribution of key proteins that form the intestinal tight junctions.

endo.endojournals.org

0

Control

KO

***

100 80 60 40 20 0

Control

KO

Figure 1. GLP-2 enhances intestinal barrier function in an IE-IGF-1R– dependent fashion. A, Genotype-specific primers were used to amplify the floxed and Cre alleles, generating bands of 390 and 312 bp, respectively, with –ve indicating PCR in the absence of template. Control and KO mice were treated with vehicle (open bars) or GLP-2 (closed bars), and crypt-villus height as well as barrier function were examined. B, Jejunal crypt-villus height was measured in control and KO mice (n ⫽ 4 – 8). C, Plasma levels of FD4 were determined after oral gavage (n ⫽ 11–14). D, Jejunal transmural resistance was quantified in Ussing chamber studies (n ⫽ 6 –9). *, P ⬍ .05; **, P ⬍ .01; ***, P ⬍ .001.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 July 2014. at 12:38 For personal use only. No other uses without permission. . All rights reserved.

372

Dong et al

GLP-2, Barrier Function, and the IE-IGF-1R

assays were performed 4 and 6 days later, respectively. Streptozotocin studies are described in the Supplemental Data (published on The Endocrine Society’s Journals Online website at http://endo.endojournals.org). Mice were killed, and a 2-cm segment from the midjejunum was fixed in 10% neutral-buffered formalin and cut into 3 to 4 pieces for paraffin embedding in a single block (University Health Network). For electron microscopy (EM), two 1-mm jejunal segments were fixed in 2.5% glutaraldehyde, postfixed in 2% phosphate-buffered OsO4, dehydrated in acetone, and embedded and polymerized in Epon-Araldite before cutting and staining (Hospital for Sick Children, Toronto, Ontario, Canada). One 2-cm jejunal segment was frozen for immunoblotting, and two 5-mm segments of fresh jejunum were collected for analysis of resistance. Segments of jejunum were also weighed before and after freeze-drying to determine percent dry mass. The jejunum was selected for all studies because it expresses the highest levels of the GLP-2R and, hence, serves as a functionally significant tissue (6).

Intestinal permeability Gastrointestinal permeability was measured in vivo by determining the ability of the relatively impermeant macromolecule 4-kDa fluorescein isothiocyanate-dextran (FD4) to cross from the lumen into the circulation. In brief, mice were fasted overnight and then orally gavaged with FD4 (0.5 mg/g body weight of a 50 mg/mL solution; Sigma-Aldrich). After 1.5 hours, 120 ␮L of tail vein blood was collected and plasma fluorescence was measured at excitation and emission wavelengths of 485 and 535 nm, respectively, as compared with a standard curve using FD4 in normal plasma (16). Transmural resistance was measured ex vivo in 2 contiguous segments of midjejunum, using electrophysiological measurements. Segments were opened along the mesenteric border and mounted in Ussing chambers (Physiologic Instruments). Tissues were incubated in modified Meyler solution at 37°C with continuous oxygenation (26). After equilibration, samples were clamped with intermittent current pulses of 0.001 mA, and the corresponding voltage changes across the intestinal wall were continuously recorded. Integrity of the tissue was confirmed using forskolin/3-isobutyl-1-methylxanthine (10␮M and 200␮M, respectively; Sigma-Aldrich) to induce cystic fibrosis transmembrane conductance regulator-mediated chloride flux; tissues with no response were excluded from analysis. Resistance was calculated according to Ohm’s law (voltage ⫽ current ⫻ resistance).

Microscopy Crypt-villus height was measured on hematoxylin/eosinstained slides, as previously reported (19, 20), using an average of 38 well-oriented crypt-villus units from 3 to 4 cross-sections per mouse. For jejunal tight junction immunostaining, 1 mouse from each group of animals was represented on each slide (eg, n ⫽ 1 for each group on each slide). Antigen retrieval for the claudins was heat-induced by microwaving in citrate buffer for 20 minutes. Antigen retrieval for occludin and zona occludins-1 (ZO-1) was induced by incubation with pronase E (1 mg/mL; SigmaAldrich) for 15 minutes at 37°C. Sections were incubated overnight at 4°C with primary antibodies as listed in Supplemental

Endocrinology, February 2014, 155(2):370 –379

Table 1. Secondary antibody incubation was performed at room temperature for 1 hour (Supplemental Table 1). An AxioPlan epifluorescence microscope (Carl Zeiss, Canada) was used to acquire all images using constant exposure levels for all 4 samples per slide.

Immunoblotting Jejunal mucosal scrapes (20) were homogenized in RIPA lysis buffer with a mini EDTA-free protease inhibitor tablet (Roche Diagnostics Corp). Protein concentration was assessed by Bradford assay (Bio-Rad, Hercules, CA), and samples were run on a 15% (for claudins) or 7% (for ZO-1 and occludin) PAGE gel. Proteins were transferred onto an Immun-Blot polyvinylidene difluoride membrane (Bio-Rad) and incubated overnight with primary antibodies, as listed in Supplemental Table 1. Membranes were probed with a secondary antiserum (Supplemental Table 1), and bands were detected by electrochemiluminescence (Amersham GE Healthcare).

Statistical analyses All data are expressed as mean ⫾ SE. Results were analyzed by two-way ANOVA, followed by Student’s t test when appropriate.

Results Conditional IE-specific IGF-1R–KO mice were identified by genotyping for IGF-1Rflox/flox (present in both control and KO mice) and villin-CreERT2⫹/0 (present only in KO animals; Figure 1A). Functional validation of gene deletion was performed by assessment of GLP-2–induced jejunal crypt-villus growth, which is known to require the IE-IGF-1R (20). Control mice treated with GLP-2 demonstrated an increase in crypt-villus height by 27.6% ⫾ 2.9% (P ⬍ .01), compared with vehicle-treated control animals. In contrast, crypt-villus height did not differ between vehicle- and GLP-2–treated KO mice, indicative of successful gene deletion (Figure 1B). To determine the in vivo effects of chronic GLP-2 treatment on gastrointestinal permeability from the gut lumen into the blood stream, fluorescence was measured in the plasma after oral gavage of FD4. In control mice, plasma FD4 levels were markedly reduced after GLP-2 treatment by 61.9% ⫾ 7.8% (P ⬍ .05; Figure 1C). Basal FD4 permeability was not altered in vehicle-treated KO mice; however, the ability of GLP-2 to reduce permeability to FD4 was abolished in the absence of the IE-IGF-1R. Transmural resistance was then quantified in jejunal tissues ex vivo using Ussing chamber studies. Consistent with the results of the FD4 study, GLP-2 treatment increased jejunal resistance in control mice by 44.1% ⫾ 6.0% (P ⬍ .001). Although basal resistance was not significantly altered in KO animals as compared with controls (P ⫽ .29 for a gene effect by two-way ANOVA), there appeared to

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 July 2014. at 12:38 For personal use only. No other uses without permission. . All rights reserved.

doi: 10.1210/en.2013-1871

endo.endojournals.org

500 nm

TJ

4.0 2.0 0.0

KO

D

Cl-15

D

1.0

0.5

0.0

Control

KO

KO GLP-2

C

D

E

2.0

*

**

1.5 1.0 0.5 0.0

Control

KO

Occludin β-acn 1.5

1.5

D

KO vehicle

Control

β-acn

AJ

AJ

*

6.0

B C

*

Cl-7 relave expression

8.0

Cl-7 β-acn

β-acn

Control GLP-2

TJ

B

Cl-3

Cl-15 relave expression

A

A

Occludin relave expression

Control vehicle

din-15, and the scaffolding protein ZO-1. GLP-2 treatment markedly increased expression of claudin-3 in control animals by 456% ⫾ 174% (P ⬍ .05; Figure 3A). Similarly, claudin-7 levels were increased in response to GLP-2 by 52% ⫾ 21% (P ⬍ .01) in control mice (Figure 3B). However, despite similar basal levels of both claudins in IE-IGF-1R–null animals, GLP-2–induced expression of claudin-3 and -7 was abrogated (P value not significant vs vehicle-treated KO mice; P ⬍ .05–.01 vs GLP-2–treated control mice. In contrast, neither GLP-2 treatment nor IE-IGF-1R knockout significantly affected expression of claudin-15, occludin, or ZO-1 (Figure 3, C–E).

Cl-3 relave expression

be a slight elevation in resistance in the vehicle-treated KO mice, suggestive of a compensatory response to IE-IGF1R– knockout mice (Figure 1D). Nonetheless, the effects of GLP-2 were profoundly impaired in the null mice (P value not significant) as compared with the response seen in control animals. These results confirm that GLP-2 enhances intestinal barrier function in mice (11, 15, 16) and demonstrate that this action requires the IE-IGF-1R. When examined at the ultrastructural level by EM, the tight junctions from vehicle- and GLP-2–treated control mice did not exhibit any gross differences in appearance (Figure 2, A and B). Similarly, tight junctions from vehicleand GLP-2–treated KO mice did not demonstrate any abnormalities when compared with control animals (Figure 2, C and D). Selected tight junctional proteins that regulate jejunal paracellular permeability were analyzed to determine possible mechanisms underlying the GLP-2–induced increase in barrier function, including several sealing proteins (claudin-3 and -7 and occludin), the pore-forming clau-

373

1.0

0.5

0.0

Control

KO

ZO-1 β-acn

TJ

AJ

AJ

D

D

1.2 ZO-1 relave expression

TJ

0.8

0.4

0.0

Figure 2. Gross jejunal tight junction morphology is independent of GLP-2 and the IE-IGF-1R. Control (A and B) and KO (C and D) mice were treated with vehicle (A and C) or GLP-2 (B and D), and jejunal sections were examined by EM (n ⫽ 4). Representative photomicrographs are shown. Abbreviations: AJ, adherens junction; D, desmosome; TJ, tight junction.

Control

KO

Figure 3. GLP-2 enhances jejunal claudin-3 and -7 protein expression in an IE-IGF-1R– dependent fashion. Control and KO mice were treated with vehicle (open bars) or GLP-2 (closed bars), and jejunal mucosal protein expression was determined by immunoblot. A, Claudin (Cl)-3 (n ⫽ 6 –9); B, Cl-7 (n ⫽ 8 –9); C, Cl-15 (n ⫽ 5–7); D, occludin (n ⫽ 10 – 11; note P ⫽ .13 overall by two-way ANOVA); E, ZO-1 (n ⫽ 6 – 8). Representative blots are shown. *, P ⬍ .05; **, P ⬍ .01.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 July 2014. at 12:38 For personal use only. No other uses without permission. . All rights reserved.

374

Dong et al

GLP-2, Barrier Function, and the IE-IGF-1R

Endocrinology, February 2014, 155(2):370 –379

which were distributed along the apical-basolateral axis of the epithelial cells (Figure 4, A and B) and around the circumference of the cell (Supplemental Figure 1, A and B). Both the expression and distribution of claudin-15 were unchanged in response to GLP-2 treatment and were not significantly altered after IE-IGF-1R knockout alone (P ⫽ .06 vs PBSControl Control KO KO Negave treated control) or in response to vehicle GLP-2 vehicle GLP-2 control treatment with GLP-2 (Figure 4C A 20μm and Supplemental Figure 1C). In contrast, occludin staining was localized to the apical border in vehicle-treated control mice, and although this distribution was not altered by GLP-2 treatment, occludin appeared to be predominantly B cytoplasmic in both vehicle- and GLP-2–treated KO animals (Figure 4D). Finally, no differences in expression or distribution of ZO-1 were noted between any of the animals (Figure 4E). To develop a model of altered barC rier function, control and KO animals were first subjected to multiple low-dose injections of streptozotocin, using a protocol known to induce diabetes in normal C57BL/6 mice (27). Although glucose tolerance was markedly impaired in the Negave streptozotocin-treated animals, FD4 control D permeability was unexpectedly de20μm creased as compared with the expected impairment in barrier function as reported for other models of diabetes (Supplemental Figure 2) (16). Mice were therefore next exposed to irinotecan to induce a GLPE 2–responsive enteritis that has been associated with enhanced bacterial translocation (11, 17, 28). Jejunal sections from irinotecan-treated control vehicle-treated animals were found to have a 26% ⫾ 5% increase (P ⬍ .01) in dry weight, suggestive of Figure 4. GLP-2 enhances jejunal claudin-3 and -7 immunofluorescence intensity in an IE-IGFenhanced tissue density. Furthermore, 1R– dependent fashion, whereas localization of occludin is GLP-2–independent but IE-IGF-1R– the morphological characteristics of dependent. A–C, Claudin-3 (A), -7 (B), and -15 (C) were immunostained in red, the brush border the jejunum in vehicle-treated control marker sucrase in green, and nuclei in blue. All claudin staining was performed with the same secondary antibody, and negative control staining was performed without primary antiserum. D mice after administration of irinoteand E, Occludin (D) and ZO-1 (E) were immunostained in red and nuclei in blue; negative staining can indicated extensive crypt-villus without primary antiserum is shown. All images subjected to comparison were captured from a damage with patches of immune insingle slide using the same exposure settings, and representative images of the longitudinal villus filtration. A similar degree of damaxis are shown (n ⫽ 4 each). ZO-1

Occludin

Claudin-15

Claudin-7

Claudin-3

Consistent with the immunoblot results, increased immunofluorescence intensity was observed in jejunal villi for both claudin-3 and –7 in GLP-2-treated control animals, and this effect was completely absent in IE-IGF1R–KO mice (Figure 4, A and B). No differences in localization of the proteins were apparent in any of the animals,

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 July 2014. at 12:38 For personal use only. No other uses without permission. . All rights reserved.

doi: 10.1210/en.2013-1871

endo.endojournals.org

age was observed in the vehicle-treated KO animals (Figure 5A). However, although the crypt-villus architecture was noticeably improved by GLP-2 treatment of control mice, these changes were not observed in KO animals. To quantitate these observations, crypt-villus height was determined for all groups of mice (Figure 5B). In comparison with healthy vehicle-treated control mice, the crypt-villus unit was shortened in control-vehicle irinotecan-treated animals. The ability of GLP-2 to increase the length of this axis was maintained in these animals (by 18% ⫾ 4%, P ⬍ .01) and, as also found in healthy mice (Figure 1B), was abrogated in the absence of the IE-IGF-1R. After the establishment of enteritis, the barrier function appeared to be improved in irinotecan-treated vehicle-control mice

compared with healthy vehicle-treated controls, as indicated by a decrease in permeability to FD4 and an increase in jejunal resistance (Figure 5, C and D). Furthermore, treatment of these animals with GLP-2 unexpectedly reversed these changes. Notwithstanding, all of these changes were absent in mice that lacked the IE-IGF-1R. Finally, to examine the effects of GLP-2 and IE-IGR-1R deletion on the tight junctional proteins, both Western blot and immunostaining of jejunal tissues from irinotecan-treated mice were conducted as performed for the healthy animals. Interestingly, no significant change in claudin-7 expression was found after GLP-2 treatment of control mice (P ⫽ .06), although the trend to a decrease was consistent with the effects of GLP-2 on gut permea-

**

B

500

Crypt-villus height (μm)

A

D

*

4

2

0

Control

KO

Jejunal resistance (Ω·cm2)

Plasma FD4 (μg/ml)

6

*

375

250

125

0

C

375

120

Control

KO

*

80

40

0

Control

KO

Figure 5. GLP-2 restores intestinal barrier function in an IE-IGF-1R– dependent fashion in mice with irinotecan-induced enteritis. Control and KO mice were treated with vehicle (open bars) or GLP-2 (closed bars) for 10 days, followed by administration of irinotecan for 2 days and then examination of jejunal integrity, crypt-villus height, and barrier function (n ⫽ 7–10). A, Representative photomicrographs of jejunum from irinotecan-treated control vehicle-treated (i), control GLP-2–treated (ii), KO vehicle-treated (iii), and KO GLP-2–treated (iv) mice. B–D, Crypt-villus height (B), plasma levels of FD4 after oral gavage (C), and jejunal transmural resistance as quantified in Ussing chambers (D). For comparison purposes, the dashed lines in B–D indicate the values obtained for healthy control vehicle-treated animals as shown in Figure 1. *, P ⬍ .05; **, P ⬍ .01.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 July 2014. at 12:38 For personal use only. No other uses without permission. . All rights reserved.

Dong et al

GLP-2, Barrier Function, and the IE-IGF-1R

Endocrinology, February 2014, 155(2):370 –379

bility and jejunal resistance in this model (Figure 6). None of the other tight junctional proteins appeared to be affected by GLP-2 treatment. In further contrast to the findings in the healthy animals, KO mice demonstrated elevated levels of claudin-3 but reduced claudin-7 expression, although, again, GLP-2 treatment had no effect on any of the proteins in the null mice. Finally, no marked differences were noted in the expression or localization of any of these proteins by immunofluorescence (Supplemental Figure 3).

Discussion A long-acting analog of GLP-2, teduglutide, is an important new therapeutic tool for the treatment of short bowel syndrome, increasing both the growth and the function of the residual bowel in these patients (3). However, the mechanism of action of GLP-2 is complex due to highly

A

B

Cl-3 β-acn

Cl-7 relave expression

Cl-3 relave expression

C

Cl-7

*

4.0 3.0 2.0 1.0

3.0

1.5

* 1.0 p=0.06 0.5

0.0 Control

Cl-15 β-acn

β-acn

5.0

0.0

restricted expression of the GLP-2R in cells that do not reside within the proliferative compartment of the intestine (6 – 8). GLP-2 thus requires multiple mediators to exert its intestinotropic actions including, most notably, IGF-1 and the IGF-1R expressed specifically on the intestinal epithelial cells (19, 20). In contrast, although GLP-2 is known to enhance the barrier function of the intestine (11, 15–17), the mediators of this biological action are unknown. The results of the present study demonstrate for the first time that treatment of healthy mice with GLP-2 enhances intestinal barrier function through a pathway that requires the IE-IGF-1R and that this effect is associated with IE-IGF-1R– dependent changes in the expression and localization of specific tight junctional proteins. Both gastrointestinal permeability to FD4 and jejunal conductance were reduced by 10 days treatment of healthy animals with GLP-2, consistent with previous reports (15, 16). These changes were associated with increased expres-

Cl-15 relave expression

376

KO

Control

KO

2.0

1.0

0.0

D

E

occludin β-acn

β-acn 1.5 ZO-1 relave expression

Occludin relave expression

KO

ZO-1

2.0 1.5 1.0 0.5 0.0

Control

1.0

0.5

0.0 Control

KO

Control

KO

Figure 6. IE-IGF-1R KO modulates jejunal claudin-3 and -7 protein expression. Control and KO mice were treated with vehicle (open bars) or GLP2 (closed bars) for 10 days, followed by administration of irinotecan for 2 days and then examination of jejunal mucosal protein expression by immunoblot. A, Claudin (Cl)-3 (n ⫽ 5– 6); B, Cl-7 (n ⫽ 4 – 6); C, Cl-15 (n ⫽ 4 – 6); D, occludin (n ⫽ 4 – 6); E, ZO-1 (n ⫽ 4 – 6). Representative blots are shown. *, P ⬍ .05.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 July 2014. at 12:38 For personal use only. No other uses without permission. . All rights reserved.

doi: 10.1210/en.2013-1871

sion but normal distribution of the sealing proteins claudin-3 and -7, members of the claudin superfamily that determine paracellular permeability based on size and charge selectivity (29). Although claudin-2, -3, -7, and -15 are expressed at particularly high levels in the rodent intestine (30, 31), claudin-7 appears to play a key role with respect to formation of the tight junctions on the apical side of the epithelial cells (32), in addition to the regulation of interactions between the epithelial cells and the underlying basolateral matrix (33). In contrast, claudin-15, which is a pore-forming protein, was not modulated by GLP-2 in the present study, whereas claudin-2 was not investigated because it is expressed predominantly in the crypts (30) and is therefore unlikely to contribute to overall intestinal barrier function. A previous study has shown that prebiotic induction of barrier function in ob/ob mice occurs consequent to GLP-2–associated increases in the sealing tight junction protein occludin and the tightening protein ZO-1, as assessed by immunofluorescent scoring (16). However, administration of GLP-2 to healthy animals in the present study did not alter the levels or distribution of either of these proteins. Hence, of the tight junctional proteins examined in healthy mice, only the expression of claudin-3 and -7 was associated with the actions of GLP-2 on barrier function. Importantly, and consistent with our findings of a requirement for the IE-IGF-1R in the intestinotropic actions of GLP-2 (20), the effects of GLP-2 on barrier function as well as on claudin-3 and -7 were abrogated in the absence of the IE-IGF-1R. Furthermore, the distribution of occludin appeared to be predominantly intracellular in the KO animals, suggesting an independent defect in the absence of IGF signaling, albeit one that did not appear to alter basal barrier function. Although IGF-1 is known to enhance intestinal barrier function in several models of disease (22, 23), little is known about the mechanisms underlying these actions. However, GLP-2 increases the levels of phosphorylated Akt in murine intestinal epithelial cells through a pathway that is at least partly IGF-1R– dependent (34), and Akt, which is a downstream mediator of the IGF-1R, has been shown to modulate tight junction protein distribution and barrier function in a number of different tissues (35). Furthermore, the IE-IGF-1R binds both IGF-1 and IGF-2, and studies on T84 human colonic epithelial cells have indicated that both of these growth factors increase transepithelial resistance through a mechanism that requires protein synthesis (36). Finally, the transcription factor Cdx2, which is a downstream effector of IGF-1R signaling (37), stimulates expression of claudin-2 in colonic epithelial cells in a phosphoinositide 3-kinase (PI3K)-dependent fashion (38). PI3K/Akt signaling has also been implicated in the translocation of occludin,

endo.endojournals.org

377

as well as of ZO-1, to the tight junctional complex (39). Collectively, these findings suggest that GLP-2 modulates intestinal barrier function through a pathway that leads from the GLP-2R to IGF-1 and the IE-IGF-1R and then to Akt and the tight junction proteins. Because the present study is the first to our knowledge that implicates IGF signaling as a direct regulator of intestinal tight junction protein expression (claudin-3 and -7) and localization (occludin), additional studies will be required to determine through which downstream signaling pathway these effects are mediated, particularly because the effects of PI3K/Akt on the tight junctional proteins appear to be both protein- and context-dependent (35). Although irinotecan treatment of mice has been associated with increased bacterial translocation across the gut wall (11, 17), intestinal permeability was actually found to decrease in the present study in mice with irinotecan-induced enteritis, in association with an increase in jejunal resistance. Precedent literature does exist for enhanced small intestinal resistance in response to irinotecan-induced mucositis in the rat, which was suggested to be due to an edema-associated limitation of ion movement across the gut wall (28). Similarly, a paradoxical increase in colonic transmural resistance was found in a murine model of ulcerative colitis consequent to submucosal edema as well as to inflammatory infiltration, rather than to enhanced function of the epithelial barrier (40). The difference between findings made on bacterial translocation vs those using small molecules such as FD4 and ions may therefore be due to physical limitations to small molecule flux rather than to alterations in tight junctional expression or distribution. Indeed, although there was no evidence of edema in irinotecan-treated mice in the present study, the increased tissue density and enhanced immune infiltration found in this model are both consistent with this hypothesis, at least at the time point selected for analysis. In further support of this notion, irinotecan-induced enteritis was associated with paradoxical effects of GLP-2 to enhance barrier function toward normal levels that occurred in association with restoration of the normal architecture of the gut. Similar findings have been reported for the tropic effects of GLP-2 in several models of experimental intestinal damage, with the abnormally high levels of crypt cell proliferation observed in these animals restored to normal by GLP-2 treatment, in association with a reduction in inflammation (41). Finally, the alterations in barrier function in these animals occurred in the absence of changes in expression or distribution of the tight junctional proteins. Hence, in disease models, the ability of GLP-2 to reduce damage and restore normal intestinal function appears to be more important than its effects on the tight junctions.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 July 2014. at 12:38 For personal use only. No other uses without permission. . All rights reserved.

378

Dong et al

GLP-2, Barrier Function, and the IE-IGF-1R

Regardless of the direction of change in barrier function found for GLP-2 in the disease model as compared with healthy animals, the IE-IGF-1R was found to play an important role in both the GLP-2–induced growth of the intestine as well as the regulation of the tight junctional proteins in irinotecan-treated mice. Hence, the ability of GLP-2 to increase crypt-villus length was abolished in these animals, as in the healthy mice. Furthermore, baseline expression of claudin-3 was increased in the absence of the IE-IGF-1R, whereas that of claudin-7 was decreased. Of some interest, the pattern in claudin-7 expression, but not that in claudin-3, appeared to match the changes found in barrier function. However, whether the altered barrier function was due to these differences in protein expression will clearly require further study. Finally, the distribution of occludin was again found to be predominantly cytoplasmic, at least in intact villi, although it is recognized that these may not be fully representative of the jejunum due to the extensive degree of structural damage observed in this model. Because tight junctions are believed to play a role in the pathogenesis of chemotherapy-induced mucositis (42), these findings may have implications with respect to the prevention or treatment of the side effects of this therapy. In conclusion, the results of this study demonstrate that the effects of GLP-2 on intestinal barrier function are mediated, at least in part, through the IE-IGF-1R and modulation of specific proteins that constitute the intestinal tight junction. Because the intestinal epithelium must restrict passage of harmful substances while permitting absorption of nutrients, elucidation of the pathways that regulate these processes are integral to our understanding of not only normal intestinal physiology but also the pathophysiology of intestinal disease. These findings may be of particular importance to the therapeutic application of long-acting GLP-2 analogs in patients with inflammatory bowel disease (4), in whom intestinal barrier function is known to be impaired (43).

Acknowledgments Address all correspondence and requests for reprints to: Dr Patricia L. Brubaker, Room 3366 Medical Sciences Building, 1 King’s College Circle, University of Toronto, Toronto, ON M5S 1A8, Canada. E-mail: [email protected]. This work was supported by operating grants from the Canadian Institutes of Health Research (MOP-9940 and MOP12344) and by an equipment grant to the 3D (Diet, Digestive Tract, and Disease) Centre from the Canadian Foundation for Innovation and Ontario Research Fund (Project 19442). C.X.D. was supported by graduate studentships from Ontario Graduate Scholarship and the Banting and Best Diabetes Centre, Univer-

Endocrinology, February 2014, 155(2):370 –379

sity of Toronto; W.Z. by a summer studentship from the Canadian Association of Gastroenterology; C.S. by a University of Toronto Research Opportunity Program Summer Studentship; and P.L.B. by the Canada Research Chairs Program. C.X.D., T.G., and P.L.B. designed the study; C.X.D., W.Z., C.S., C.A., and T.G. performed data acquisition, analysis, and interpretation; C.X.D. and P.L.B. drafted the manuscript; K.J.R., C.A., S.R., M.H., T.G., and P.L.B. revised the manuscript; S.R. and M.H. provided material support; and P.L.B. obtained funding and supervised the study. Disclosure Summary: P.L.B. has received consulting fees from NPS Pharmaceuticals; all other authors have nothing to disclose.

References 1. Estall JL, Drucker DJ. Glucagon-like peptide-2. Annu Rev Nutr. 2006;26:391– 411. 2. Rowland KJ, Brubaker PL. The “cryptic” mechanism of action of glucagon-like peptide-2. Am J Physiol Gastrointest Liver Physiol. 2011;301:G1–G8. 3. Jeppesen PB. Modern treatment of short bowel syndrome. Curr Opin Clin Nutr Metab Care. 2013;16:582–587. 4. Blonski W, Buchner AM, Aberra F, Lichtenstein G. Teduglutide in Crohn’s disease. Expert Opin Biol Ther. 2013;13:1207–1214. 5. Xiao Q, Boushey RP, Drucker DJ, Brubaker PL. Secretion of the intestinotropic hormone glucagon-like peptide 2 is differentially regulated by nutrients in humans. Gastroenterology. 1999;117:99 – 105. 6. Yusta B, Huang L, Munroe D, et al. Enteroendocrine localization of GLP-2 receptor expression in humans and rodents. Gastroenterology. 2000;119:744 –755. 7. Guan X, Karpen HE, Stephens J, et al. GLP-2 receptor localizes to enteric neurons and endocrine cells expressing vasoactive peptides and mediates increased blood flow. Gastroenterology. 2006;130: 150 –164. 8. Ørskov C, Hartmann B, Poulsen SS, Thulesen J, Hare KJ, Holst JJ. GLP-2 stimulates colonic growth via KGF, released by subepithelial myofibroblasts with GLP-2 receptors. Regul Pept. 2005;124:105– 112. 9. Shin ED, Estall JL, Izzo A, Drucker DJ, Brubaker PL. Mucosal adaptation to enteral nutrients is dependent on the physiologic actions of glucagon-like peptide-2 in mice. Gastroenterology. 2005;128: 1340 –1353. 10. Li J, Mizukami Y, Zhang X, Jo WS, Chung DC. Oncogenic K-ras stimulates Wnt signaling in colon cancer through inhibition of GSK3␤. Gastroenterology. 2005;128:1907–1918. 11. Lee SJ, Lee J, Li KK, et al. Disruption of the murine Glp2r impairs paneth cell function and increases susceptibility to small bowel enteritis. Endocrinology. 2012;153:1141–1151. 12. Drucker DJ, Erlich P, Asa SL, Brubaker PL. Induction of intestinal epithelial proliferation by glucagon-like peptide 2. Proc Natl Acad Sci U S A. 1996;93:7911–7916. 13. Brubaker PL, Izzo A, Hill M, Drucker DJ. Intestinal function in mice with small bowel growth induced by glucagon-like peptide-2. Am J Physiol Endocrinol Metab. 1997;272:E1050 –E1058. 14. Cheeseman CI. Upregulation of SGLT-1 transport activity in rat jejunum induced by GLP-2 infusion in vivo. Am J Physiol Regul Integr Comp Physiol. 1997;273:R1965–R1971. 15. Benjamin MA, McKay DM, Yang PC, Cameron H, Perdue MH. Glucagon-like peptide-2 enhances intestinal epithelial barrier function of both transcellular and paracellular pathways in the mouse. Gut. 2000;47:112–119. 16. Cani PD, Possemiers S, Van de Wiele T, et al. Changes in gut mi-

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 July 2014. at 12:38 For personal use only. No other uses without permission. . All rights reserved.

doi: 10.1210/en.2013-1871

17.

18.

19.

20.

21. 22.

23.

24.

25.

26.

27.

28.

29.

crobiota control inflammation in obese mice through a mechanism involving GLP-2-driven improvement of gut permeability. Gut. 2009;58:1091–1103. Boushey RP, Yusta B, Drucker DJ. Glucagon-like peptide (GLP)-2 reduces chemotherapy-associated mortality and enhances cell survival in cells expressing a transfected GLP-2 receptor. Cancer Res. 2001;61:687– 693. Bahrami J, Yusta B, Drucker DJ. ErbB activity links the glucagonlike peptide-2 receptor to refeeding-induced adaptation in the murine small bowel. Gastroenterology. 2010;138:2447–2456. Dubé PE, Forse CL, Bahrami J, Brubaker PL. The essential role of insulin-like growth factor-1 in the intestinal tropic effects of glucagon-like peptide-2 in mice. Gastroenterology. 2006;131:589 – 605. Rowland KJ, Trivedi S, Lee D, et al. Loss of glucagon-like peptide2-induced proliferation following intestinal epithelial insulin-like growth factor-1-receptor deletion. Gastroenterology. 2011;141: 2166 –2175.e7. Steed E, Balda MS, Matter K. Dynamics and functions of tight junctions. Trends Cell Biol. 2010;20:142–149. Huang KF, Chung DH, Herndon DN. Insulinlike growth factor 1 (IGF-1) reduces gut atrophy and bacterial translocation after severe burn injury. Arch Surg. 1993;128:47–53; discussion 53–54. Lorenzo-Zúñiga V, Rodríguez-Ortigosa CM, Bartolí R, et al. Insulin-like growth factor I improves intestinal barrier function in cirrhotic rats. Gut. 2006;55:1306 –1312. el Marjou F., Janssen KP, Chang BH, et al. Tissue-specific and inducible Cre-mediated recombination in the gut epithelium. Genesis. 2004;39:186 –193. Kappeler L, De Magalhaes FC, Dupont J, et al. Brain IGF-1 receptors control mammalian growth and lifespan through a neuroendocrine mechanism. PLoS Biol. 2008;6:e254. Bijvelds MJ, Bot AG, Escher JC, De Jonge HR. Activation of intestinal Cl⫺ secretion by lubiprostone requires the cystic fibrosis transmembrane conductance regulator. Gastroenterology. 2009;137: 976 –985. Li Y, Hansotia T, Yusta B, Ris F, Halban PA, Drucker DJ. Glucagonlike peptide-1 receptor signaling modulates ␤-cell apoptosis. J Biol Chem. 2003;278:471– 478. Nakao T, Kurita N, Komatsu M, et al. Irinotecan injures tight junction and causes bacterial translocation in rat. J Surg Res. 2012;173: 341–347. Lal-Nag M, Morin PJ. The claudins. Genome Biol. 2009;10:235.

endo.endojournals.org

379

30. Rahner C, Mitic LL, Anderson JM. Heterogeneity in expression and subcellular localization of claudins 2, 3, 4, and 5 in the rat liver, pancreas, and gut. Gastroenterology. 2001;120:411– 422. 31. Fujita H, Chiba H, Yokozaki H, et al. Differential expression and subcellular localization of claudin-7, -8, -12, -13, and -15 along the mouse intestine. J Histochem Cytochem. 2006;54:933–944. 32. Lei Z, Maeda T, Tamura A, et al. EpCAM contributes to formation of functional tight junction in the intestinal epithelium by recruiting claudin proteins. Dev Biol. 2012;371:136 –145. 33. Ding L, Lu Z, Foreman O, et al. Inflammation and disruption of the mucosal architecture in claudin-7-deficient mice. Gastroenterology. 2012;142:305–315. 34. Dubé PE, Rowland KJ, Brubaker PL. Glucagon-like peptide-2 activates beta-catenin signaling in the mouse intestinal crypt: role of insulin-like growth factor-I. Endocrinology. 2008;149:291–301. 35. González-Mariscal L, Tapia R, Chamorro D. Crosstalk of tight junction components with signaling pathways. Biochim Biophys Acta. 2008;1778:729 –756. 36. McRoberts JA, Riley NE. Regulation of T84 cell monolayer permeability by insulin-like growth factors. Am J Physiol. 1992;262: C207–C213. 37. Modica S, Morgano A, Salvatore L, et al. Expression and localisation of insulin receptor substrate 2 in normal intestine and colorectal tumours. Regulation by intestine-specific transcription factor CDX2. Gut. 2009;58:1250 –1259. 38. Suzuki T, Yoshinaga N, Tanabe S. Interleukin-6 (IL-6) regulates claudin-2 expression and tight junction permeability in intestinal epithelium. J Biol Chem. 2011;286:31263–31271. 39. Little D, Dean RA, Young KM, et al. PI3K signaling is required for prostaglandin-induced mucosal recovery in ischemia-injured porcine ileum. Am J Physiol Gastrointest Liver Physiol. 2003;284: G46 –G56. 40. Barmeyer C, Horak I, Zeitz M, Fromm M, Schulzke JD. The interleukin-2-deficient mouse model. Pathobiology. 2002;70:139 –142. 41. Sigalet DL, Wallace LE, Holst JJ, et al. Enteric neural pathways mediate the anti-inflammatory actions of glucagon-like peptide 2. Am J Physiol Gastrointest Liver Physiol. 2007;293:G211–G221. 42. Wardill HR, Bowen JM. Chemotherapy-induced mucosal barrier dysfunction: an updated review on the role of intestinal tight junctions. Curr Opin Support Palliat Care. 2013;7:155–161. 43. Jäger S, Stange EF, Wehkamp J. Inflammatory bowel disease: an impaired barrier disease. Langenbecks Arch Surg. 2013;398:1–12.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 July 2014. at 12:38 For personal use only. No other uses without permission. . All rights reserved.