The intestinal hormone, glucagon-like peptide-2 (GLP-2), stimulates growth, survival, and function of the intestinal epithelium through increased crypt cell proliferation, and a long-acting analog has recently been approved to enhance intestinal capacity in patients with short bowel syndrome. The goal of the present study was to determine whether GLP-2-induced crypt cell proliferation requires a full complement of B-cell lymphoma Moloney murine leukemia virus insertion region-1 homolog (Bmi-1), using the Bmi-1eGFP/+ mouse model in comparison with age- and sex-matched Bmi-1+/+ littermates. Bmi-1 is a member of the polycomb-repressive complex family that promotes stem cell proliferation and self-renewal and is expressed by both stem cells and transit-amplifying (TA) cells in the crypt. The acute (6 h) and chronic (11 d) proliferative responses to long-acting human (Gly2)GLP-2 in the crypt TA zone, but not in the active or reserve stem cell zones, were both impaired by Bmi-1 haploinsufficiency. Similarly, GLP-2-induced crypt regeneration after 10-Gy irradiation was reduced in the Bmi-1eGFP/+ animals. Despite these findings, chronic GLP-2 treatment enhanced overall intestinal growth in the Bmi-1eGFP/+ mice, as demonstrated by increases in small intestinal weight per body weight and in the length of the crypt-villus axis, in association with decreased apoptosis and an adaptive increase in crypt epithelial cell migration rate. The results of these studies therefore demonstrate that a full complement of Bmi-1 is required for the intestinal proliferative effects of GLP-2 in both the physiological and pathological setting, and mediates, at least in part, the proliferation kinetics of cells in the TA zone.
Glucagon-like peptide-2 (GLP-2) is an intestinal hormone that promotes gut growth and function in healthy rodents as well as in preclinical models of intestinal damage (1–7). Importantly, a degradation-resistant analog of GLP-2, teduglutide, also increases measures of intestinal growth as well as the capacity for enteral nutrition in humans with short bowel syndrome (8), leading to its recent approval for this indication (9). Acute treatment of mice (0.5–6 hours) with the long-acting analog, human (h)(Gly2)GLP-2, enhances crypt cell proliferation in association with increased canonical wingless plus chromosomal integration site of mouse mammary tumour virus on mouse chromosome 15 (Wnt) signaling, including crypt cell nuclear translocation of β-catenin and induction of cellular myelocytomatosis oncogene (c-Myc) and Sry-type high-mobility-group box (Sox9) expression (7, 10, 11). When administered chronically (ie, 10–14 d), GLP-2 also stimulates crypt cell proliferation and decreases epithelial cell apoptosis, resulting in an increase in the length of the functional crypt-villus axis (1–7, 12). Despite these findings, the exact cellular targets of GLP-2 action within the crypt proliferative unit have yet to be defined.
There are 3 main pools of proliferating cells within the intestinal crypt: the actively proliferating crypt-base columnar stem cells that reside in cell positions approximately 1–3; the relatively quiescent reserve stem cells localized at positions approximately 4–6; and the transit-amplifying (TA) cells in the remainder of the crypt, which rapidly divide thereby driving the newly differentiated cells into the villus epithelium (13–15). Proliferation of cells in the TA zone in response to chronic administration of GLP-2 has been observed in several studies (12, 16). However, stimulation of the intestinal stem cell population after acute administration of GLP-2 has also been reported (7, 17). Furthermore, the intestinal reserve stem cells are not only radiation resistant as compared with the active stem cell pool (15, 18) but have also been suggested to play an essential role in GLP-2-induced crypt regeneration after lethal irradiation in mice (5). The reserve intestinal stem cells were first found to be marked by expression of B-cell lymphoma Moloney murine leukemia virus insertion region-1 homolog (Bmi-1), a member of the polycomb-repressive complex family of chromatin-remodeling proteins that promote stem cell proliferation and self-renewal through transcriptional silencing (14, 15, 18). Notwithstanding, it now appears that most crypt cells, including not only the stem cells but also the TA cells, express Bmi-1 transcripts (14, 15, 19, 20). The importance of a full complement of Bmi-1 to the maintenance of normal gut morphology was demonstrated by the findings that diphtheria toxin-mediated ablation of Bmi-1+ cells causes crypt cell death, while whole-body Bmi-1−/− mice demonstrate thin and shortened intestines with reduced proliferative capacity (14, 21). The goal of the present study was, thus, to use the Bmi-1eGFP/+ haploinsufficient murine model to determine whether Bmi-1 plays a role in the intestinotropic actions of GLP-2, under physiologic conditions and after lethal irradiation to ablate the active intestinal stem cell population.
Materials and Methods
In vivo studies
Bmi-1eGFP/+ mice (The Jackson Laboratory) express a single copy of eGFP, which was knocked-into exon 2 of the Bmi-1 locus in C57Bl/6/Ka mice that also express the hematopoietic markers, Thy1.1 leukocyte alloantigen (Thy1a) and CD45.2 (Ptprcb). Age- and sex-matched littermate Bmi-1+/+ mice on a C57Bl/6/Ka-Thy1.1, CD45.2 background were used as controls. Both mouse models grow and survive equally well and have similar hematopoietic parameters (22, 23). Mice were genotyped by PCR using the following primers: common forward, 5′-GAG AAT CCA GCT GTC CAG TGT-3′; Bmi-1 reverse wild type, 5′-TAC CCT CCA CAC AGG ACA CA-3′; and Bmi-1 reverse mutant, 5′-GAA CTT CAG GGT CAG CTT GC-3′. All animal protocols were approved by the Animal Care Committee of the University of Toronto.
Mice were injected sc with vehicle (PBS) or degradation- resistant h(Gly2)GLP-2 (American Peptide Co, Inc) (24). 5-ethynyl-2′-deoxyuridine (EdU) and/or 5-bromo-2′-deoxyuridine (BrdU) (both at 100-mg/kg body weight; Thermo Fisher Scientific, Inc) was administered ip to label proliferating cells (11).
Animals were killed by deep anesthesia using isoflurane and the small intestine was harvested, flushed with PBS and weighed. Two-centimeter sections of jejunum were then collected on dry ice and stored at −80°C for molecular analysis, and 2-cm sections of jejunum were collected in 10% buffered formalin at room temperature (RT) for morphometric analyses. Two-centimeter sections of jejunum were also placed in 4% paraformaldehyde at 4°C overnight and then transferred into 30% sucrose for cryo-protection and stored at 4°C for immunometric analyses.
To study intestinal proliferation in the absence of intestinal growth, h(Gly2)GLP-2 (0.2 μg/g) or vehicle was administered acutely to 11- to 14.5-week Bmi-1+/+ and Bmi-1eGFP/+ mice, at t = −6 hours and t = −3 hours, followed by EdU administration at t = −1 hour and then killing of the animals (n = 6) (11).
To examine intestinal growth and proliferation, h(Gly2)GLP-2 (0.1 μg/g) or vehicle was administered to Bmi-1+/+ and Bmi-1eGFP/+ mice once per day for 10 days (6–16 wk of age at start; n = 6–8) (12). On day 11, h(Gly2)GLP-2 or vehicle, as appropriate, was administered at t = −3 hours, and BrdU was injected at t = −1 hour followed by animal killing. To assess intestinal cell migration rates (25), 6- to 9-week-old mice were similarly administered h(Gly2)GLP-2 or vehicle for 11 days but were injected with BrdU at t = −24 hours and EdU at t = −1 hour before killing (n = 6).
To determine the proliferative effects of GLP-2 in the setting of radiation-induced intestinal damage, h(Gly2)GLP-2 (0.2 μg/g) or vehicle was administered to Bmi-1+/+ and Bmi-1eGFP/+ mice bid (twice a day) for 18 days (5). On the 15th day, mice (9–14 wk of age at start; n = 6) were exposed to 10-Gy radiation (Gammacell 40 Exactor; MDS Nordion), followed by 3 days of continued treatment with h(Gly2)GLP-2 or vehicle, as appropriate. On the 19th day, h(Gly2)GLP-2 or vehicle, as appropriate, was administered at t = −3 hours, and EdU was injected at t = −1 hour followed by animal killing, as above. This postradiation time point was selected to allow crypt regeneration, wherein the killing of some clonogenic stem cells results in crypts of normal size, but the killing of all but a single stem cell results in crypt overgrowth (26). After radiation treatment, the health status of the mice was monitored bid by recording body weight and using a cage-side observation score calculated from the sum of 3 different categories, each assigned a value from 0 (normal) to 3 (profound change): posture, eye appearance, and activity level (27). This score was combined with a body weight score for which each 10% decrease was assigned an additional value of 1. No mice reached the termination end point, which was predefined as a combined health score of 7 and/or a decrease in body weight of 30%.
Analyses
For morphometric analyses, formalin-fixed jejunal tissues were divided into 4 pieces, paraffin embedded together along the vertical axis, and 4-μm slices were stained with hematoxylin and eosin (University Health Network Histological Services, Toronto, ON, Canada). In the studies using healthy animals, a minimum of 12 villus heights and 26 crypt depths were measured using an Axioplan 2 Imaging Microscope with AxioVision Rel. 4.8 software (Carl Zeiss). The minimum number of villi measured in the radiation study was 7, and the minimum number of crypts was 20. The total number of crypts and intestinal circumference was determined in more than 3 intestinal cross-sections per mouse. All measurements were conducted in a blinded fashion.
Tissue sections stored in 30% sucrose were cut into 4 pieces and oriented vertically in Tissue Tek Optimum Cutting Temperature Compound (Sakura Finetek USA Ltd) and were then flash frozen in isopentane at −80°C, mounted onto a Leica CM1510S CryoStat (Leica Biosystems Nussloch GmbH), cut into 4-μm sections placed on Superfrost Plus microslides (VWR International LLC), and stored at −80°C. Sections were stained for EdU using the EdU Click-iT reaction cocktail (Thermo Fisher Scientific, Inc) and visualized in the presence of 4′,6-diamidino-2-phenylindole (Vector Laboratories, Inc) using an Axioplan 2 Imaging Microscope. Positional analysis was conducted for a minimum of 21 right-half crypts for the physiological studies and a minimum of 8 right-half crypts for the radiation study. All analyses were conducted in a blinded manner. Proliferative index was determined as the proportion of labeled cells at each crypt cell position, and area under the curve for the proliferative index was calculated using the trapezoidal rule.
For immunohistochemical staining of BrdU-labeled cells, formalin-fixed, paraffin-embedded jejunal sections were deparaffinized and rehydrated, and endogenous peroxidase activity was quenched with 3% hydrogen peroxide at 37°C for 30 minutes. Sections were then treated with 2M HCl followed by 10mM HCl with 0.4% pepsin, both at 37°C for 30 minutes. After neutralization with 100mM sodium borate (pH 8.5) at RT for 10 minutes, slides were washed in PBS, blocked with 5% normal donkey serum, and incubated overnight at 4°C with mouse anti-BrdU antibody (1:100; Thermo Fisher Scientific, Inc) in PBS with 1% BSA and 0.05% Tween 20. After washing in Tris-buffered saline with 0.05% Tween 20, sections were incubated with biotinylated goat antimouse antibody (1:200; Vector Laboratories, Inc) in Tris-buffered saline at RT for 30 minutes, followed by incubation with UltraStreptAvidin-Horseradish Peroxidase (Cedarlane), for 30 minutes, Sigmafast DAB tablets (Sigma-Aldrich Co) for 20 minutes at RT and then a 20-second rinse in tap water. Mayer's hematoxylin solution (Sigma-Aldrich) was applied for 5 minutes to label nuclei, and slides were then rinsed for 1 minute in tap water, dehydrated in ethanol and xylene, and mounted with Vectashield Mounting Medium (Vector Laboratories). The number of BrdU-positive cells was counted in a blinded fashion in a minimum of 21 right-half crypts, and the proliferative index was determined as above.
For colabeling of EdU and BrdU, frozen-fixed jejunal sections were washed in PBS and then submerged in 10mM citrate buffer (pH 6) for 40 minutes at approximately 98°C for antigen retrieval. Slides were then treated with 2N HCl for 30 minutes at 37°C, neutralized with 100mM sodium borate (pH 8.5) for 10 minutes, blocked in PBS with 1% BSA for 1 hour, both at RT, and incubated with mouse anti-BrdU antibody (1:100; Thermo Fisher Scientific, Inc) overnight at 4°C. After washing in PBS, and incubation with goat antimouse Alexa Fluor 488 secondary antibody (1:1000; Thermo Fisher Scientific, Inc) at RT for 1 hour in the dark, slides were again washed in PBS, permeabilized with 0.5% Triton X-100 in PBS for 20 minutes, and then subjected to EdU Click-iT reaction, as above. The distance from the bottom of the lowest labeled cell to the top of the highest labeled cell was determined in a blinded manner for each of BrdU and EdU, in at least 8 continuous right-half crypt-left-half villus units per mouse. Cell migration rate was calculated as the distance between the length of the individual BrdU and EdU labels divided by the 23-hour administration interval.
For coimmunofluorescence of Bmi-1 and sucrase-isomaltase, paraffin-embedded jejunal sections were deparaffinized and rehydrated, followed by antigen retrieval in sodium citrate and washing in Tris-buffered saline (TBS)-Triton X-100. Frozen-fixed jejunal sections were washed in PBS. Sections were then blocked in normal donkey serum and incubated overnight at 4°C with rabbit anti-Bmi-1 antibody (1:100; Abcam) and goat antisucrase-isomaltase antibody (1:150; Santa Cruz Biotechnology, Inc) or vehicle alone (negative control). After washing, sections were incubated with cyanine-3 (Cy3)-donkey antirabbit antibody and fluorescein isothiocyanate (FITC) donkey antigoat antibody (1:500 and 1:200, respectively; Jackson ImmunoResearch), washed, and visualized in the presence of 4′,6-diamidino-2-phenylindole. As the staining patterns within and between the 2 methods of fixation were highly variable, no quantification of cell numbers or further identification of the labeled cells was conducted (Supplemental Figure 1).
RNA was extracted from jejunal sections using an RNeasy Plus Mini kit with QiaShredder (QIAGEN, Inc). RNA was then reverse transcribed with 5x All-In-One Reverse Transcriptase MasterMix (Applied Biological Materials, Inc), and relative mRNA transcript levels were determined using TaqMan Universal Master Mix and primers (Applied Biosystems) for Bcl2-associated X protein (Bax) (Mm00432051_m1), B-cell lymphoma-2 (Bcl2) (Mm00477631_m1), Bmi-1 (Mm00776122_gH), doublecortin and calmodulin kinase-like-1 (DCAMKL1) (Mm00444950_m1), eGFP (Mr04097229_mr), homeodomain-only protein homeobox (Hopx) (Mm00558630_m1), leucine-rich repeats and immunoglobulin-like domains-1 (Lrig1) (Mm00456116_m1), leucine-rich repeat-containing G protein-coupled receptor-5 (Lgr5) (Mm01251805_m1), musashi RNA binding protein-1 (Msi1) (Mm01203522_m1), Noggin (Mm01297833_s1), Notch1 (Mm00627185_m1), and 18S rRNA (Hs99999901_s1; housekeeping gene, as previously validated) (12). Mean C(t) values for the data used for normalization are shown in Supplemental Table 1. Relative changes in gene expression were calculated using the ΔΔC(t) method (28).
Data are presented as mean ± SE. Two-way ANOVA was performed using GraphPad Prism (GraphPad Software, Inc). Some data were log10 transformed to normalize variance. Post hoc analysis was conducted by 2-tailed Student's t test if significance was found for treatment, genotype, or an interaction between these variables.
Results
Acute effects of GLP-2 in healthy mice
Administration of h(Gly2)GLP-2 at t = −6 and −3 hours to Bmi-1+/+ and Bmi-1eGFP/+ mice had no effect on either body weight or small intestinal weight (data not shown). Crypt cell positional analysis revealed that basal rates of proliferation were not different between vehicle-treated Bmi-1+/+ and Bmi-1eGFP/+ animals (P = not significant). Bmi-1+/+ animals demonstrated a crypt cell proliferative response to acute GLP-2 treatment, with significant increases by 30%–76% detected at cell positions 10–13 (P < .05–.01) (Figure 1A and Supplemental Figure 2). In contrast, Bmi-1eGFP/+ mice showed enhanced proliferation in cell positions 6 and 8 only, by 29%–30% (P < .05) (Figure 1B and Supplemental Figure 2). When the different stem cell zones of the crypt were examined more specifically (ie, active stem cell zone: cell positions 1–3; and reserve stem cell zone: cell positions 4–6), no significant effects of GLP-2 were observed for either of the genotypes (Figure 1, C and D). In contrast, GLP-2 treatment increased proliferation in the TA zone (ie, cell positions 7–16) of Bmi-1+/+ mice, by 44% (P < .01) (Figure 1E). However, no effect of GLP-2 was observed in the TA zone of Bmi-1eGFP/+ mice, resulting in a proliferative index that was not different from that of vehicle-treated genotyped-matched animals but that was significantly lower than that found for GLP-2-treated Bmi-1+/+ mice (P < .05) (Figure 1E).
Bmi-1+/+ and Bmi-1eGFP/+ mice were acutely treated with vehicle (open symbols and bars) or h(Gly2)GLP-2 (closed symbols and bars) at t = −6 and −3 hours and with EdU at t = −1 hour, before killing. The proportion of EdU-positive cells was then determined in jejunal crypts. A and B, Proliferative index in Bmi-1+/+ (A) and Bmi-1eGFP/+ (B) mice (representative images are shown in Supplemental Figure 2). C–E, Area under the curve for the proliferative index in the active stem cell zone (C; cell positions 1–3), reserve stem cell zone (D; cell positions 4–6), and TA zone (E; cell positions 7–16). *, P < .05; **, P < .01 vs vehicle or as indicated.
Chronic effects of GLP-2 in healthy mice
Treatment of Bmi-1+/+ and Bmi-1eGFP/+ mice for 11 days with h(Gly2)GLP-2 had no effect on body weight (Figure 2A), but significantly increased small intestinal weight normalized to body weight regardless of genotype, by 27% and 28%, respectively (P < .001) (Figure 2B). Chronic GLP-2 administration also significantly increased villus height (by 30% and 28%, respectively) and crypt depth (by 15% and 11%, respectively) in both groups of animals (P < .05–.001) (Figure 2C).
Bmi-1+/+ and Bmi-1eGFP/+ mice were chronically treated with vehicle (open bars) or h(Gly2)GLP-2 (closed bars) for 10 days and then on day 11 received an additional treatment at t = −3 hours, as appropriate, as well as BrdU at t = −1 hour, before killing. Body weight (BW) (A), small intestinal weight normalized to BW (SI weight/BW) (B), and crypt depth and villus height (C). *, P < .05; ***, P < .001 vs vehicle.
To examine the mechanisms underlying the growth responses to chronic GLP-2 treatment, crypt cell proliferative and apoptotic indices were determined. As found in the acute study, the proliferative index did not differ between vehicle-treated Bmi-1+/+ and Bmi-1eGFP/+ animals (P = NS). In response to GLP-2 administration, Bmi-1+/+ mice demonstrated significant increases in crypt cell proliferation, by 264%–357% at cell positions 14 and 15 (P < .05) (Figure 3A and Supplemental Figure 3). This effect was not observed in the Bmi-1eGFP/+ animals (Figure 3B and Supplemental Figure 3). The genotype-dependent changes in proliferation also translated into differential responses in the proliferative index in the TA zone (P < .05), although again, no effects were observed in either of the stem cell zones (Figure 3, C–E). In contrast, the ratio of the proapoptotic marker Bax to the survival marker Bcl2 was significantly reduced in the Bmi-1eGFP/+ mice as compared with the control animals (P < .01), and the ratio was also decreased by chronic GLP-2 treatment in both groups of animals (P < .05) (Figure 3F).
Bmi-1+/+ and Bmi-1eGFP/+ mice were chronically treated with vehicle (open symbols and bars) or h(Gly2)GLP-2 (closed symbols and bars) for 10 days and then on day 11 received an additional treatment at t = −3 hours, as appropriate, as well as BrdU at t = −1 hour, before killing. The proportion of BrdU-positive crypt cells and mRNA transcript levels were then determined in jejunal samples. A and B, Proliferative index in Bmi-1+/+ (A) and Bmi-1eGFP/+ (B) mice (representative images are shown in Supplemental Figure 3). C–E, Area under the curve for the proliferative index in the active stem cell zone (C; cell positions 1–3), reserve stem cell zone (D; cell positions 4–6), and TA zone (E; cell positions 7–16). F, Jejunal Bax/Bcl2 mRNA transcript levels. G–J, Jejunal mRNA transcript levels. DCAMKL-1, doublecortin and calmodulin kinase-like-1; Hopx, homeodomain-only protein homeobox; Lrig1, leucine-rich repeats and immunoglobulin-like domains-1; Msi-1, musashi RNA binding protein-1. *, P < .05; **, P < .01 as indicated.
To examine the effects of chronic GLP-2 treatment on specific intestinal crypt cells, mRNA transcript levels for several cell markers were determined in jejunal segments. GLP-2 significantly increased expression of Bmi-1 in Bmi-1+/+ mice, by 51% (P < .05) (Figure 3G). Unexpectedly, given the haploinsufficiency of Bmi-1 in the Bmi-1eGFP/+ mice (23), Bmi-1 transcript levels were not reduced in the vehicle-treated animals (P = NS vs vehicle wild-type mice). However, Bmi-1 transcript levels in the GLP-2-treated Bmi-1eGFP/+ animals were significantly reduced in comparison with both vehicle-treated Bmi-1eGFP/+ and GLP-2-treated Bmi-1+/+ mice (P < .05 and P < .01, respectively). As expected, eGFP mRNA transcripts were detected only in the transgenic animals and, as found for Bmi-1, these were decreased, albeit not significantly, by GLP-2 treatment (Figure 3H). The specificity of the changes in Bmi-1 was demonstrated by the finding that transcript levels for Lgr5, a marker of active intestinal stem cells (13, 15) in the same animals were not affected by either GLP-2 treatment or genotype (P = .07 for treatment effect in Bmi-1eGFP/+ mice) (Figure 3I). Chronic GLP-2 treatment also did not alter the expression of any other markers of intestinal stem cells and progenitor cells in Bmi-1+/+ mice (Figure 3J).
Finally, changes in the rate of cell movement along the crypt-villus axis were examined as a possible mechanism underlying the intestinal growth response to GLP-2 in Bmi-1eGFP/+ mice (Figure 2C) in the absence of any change in crypt cell proliferation (Figure 3B). Interestingly, BrdU labeling distance over a 24-hour period was increased by GLP-2 treatment in the Bmi-1eGFP/+ mice, by 49% (P < .05) but not in the Bmi-1+/+ animals (Figure 4, A and D). In contrast, crypt cell labeling with EdU over a 1 hour period did not differ between treatment or genotype (Figure 4, B and E). Occasional colabeled cells were also detected (Figure 4C), suggesting that these cells underwent mitosis twice during the 24-hour labeling period. When the movement of the 2 S-phase labels over time was determined, an increase in cell migration rate in response to chronic GLP-2 administration was evident (by 60%; P < .01) in the Bmi-1eGFP/+ mice only (Figure 4F).
Bmi-1+/+ and Bmi-1eGFP/+ mice were chronically treated with vehicle (open bars) or h(Gly2)GLP-2 (closed bars) for 10 days with an additional treatment at t = −3 hours on day 11, as appropriate, as well as BrdU at t = −24 hours and EdU at t = −1 hour, before killing. A–C, Representative photomicrographs of a jejunal cross-section immunostained for BrdU (A), EdU (B) with the overlapping image shown in C (*, double-labeled cell); the red line indicates the measured labeling distance. D and E, Migration distances for BrdU (D) and EdU (E). F, Cell migration rate was determined as the distance between the length of the individual BrdU and EdU labels divided by the 23-hour administration interval. *, P < .05; **, P < .01 vs vehicle.
Chronic effects of GLP-2 in irradiated mice
The inability of GLP-2 to increase crypt cell proliferation in Bmi-1eGFP/+ mice was also examined in the well-established model of intestinal damage, 10-Gy irradiation (5, 26). Four days after radiation exposure, body weight was not different by genotype and/or treatment (Figure 5A). Although all animals demonstrated declining health over this period, as indicated by an increase in the combined score, there was again no difference between the groups (Figure 5B). Interestingly, small intestinal weight normalized to body weight was increased by 48% in irradiated vehicle-treated Bmi-1+/+ mice as compared with healthy controls (P < .001) (Figures 2B vs 5C), in association with a marked 117% increase in crypt depth indicative of crypt regeneration (P < .001) (Figures 2C vs 5D). However, no differences between vehicle-treated irradiated Bmi-1+/+ and Bmi-1eGFP/+ mice were found for small intestinal weight, crypt depth or villus height (Figure 5, C and D). Both groups of animals also demonstrated similar increases in small intestinal weight in response to GLP-2 treatment, by 25% and 32%, respectively (P < .05–.001), and villus height was increased in the Bmi-1+/+ mice (P < .01), with a similar trend observed in the Bmi-1eGFP/+ animals (Figure 5, C and D). Positional analysis demonstrated a profoundly enlarged region of crypt cell proliferation (from ∼20 cells in healthy mice to approximately 40 cells in irradiated mice) (Figures 1, A and B, and 3, A and B, vs 5, E and F, and Supplemental Figure 4), consistent with the observed increase in crypt depth (Figure 5D). However, no significant increases in proliferation were observed in either genotype in response to GLP-2 treatment and, indeed, an unexpected 45%–50% decrease in proliferation was observed for both groups of animals in the position occupied by the reserve stem cell (P < .05). In contrast, GLP-2 increased the total number of crypts in the Bmi-1+/+ mice after radiation (by 11.8%, P < .05) (Figure 5G), an effect that was not observed in the Bmi-1eGFP/+ animals. No changes in jejunal circumference were detected, regardless of treatment and genotype (Figure 5H).
Bmi-1+/+ and Bmi-1eGFP/+ mice were chronically treated with vehicle (open bars and symbols) or h(Gly2)GLP-2 (closed bars and symbols) for 18 days, with exposure to 10-Gy irradiation (IR) on day 15. On the 19th day, mice received an additional treatment, as appropriate, at t = −3 hours, as well as EdU at t = −1 hour, before killing. Body weight (BW) (A), combined health scores (B), small intestinal weight normalized to BW (SI weight/BW) (C), crypt depth and villus height (D), proliferative index in Bmi-1+/+ (E) and Bmi-1eGFP/+ (F) (representative images are shown in Supplemental Figure 4) mice, total number of crypts per jejunal circumference (G), and jejunal circumference (H). *, P < .05; **, P < .01; ***, P < .001 vs vehicle.
Discussion
The biological effectiveness of GLP-2 to enhance intestinal growth in preclinical models as well as in humans with short bowel syndrome has been established, leading to development of a long-acting analog of GLP-2 for therapeutic use in patients with short bowel syndrome (9). The results of the present study indicate that both acute and chronic administration of GLP-2 to Bmi-1+/+ mice increase intestinal proliferation, mainly in the TA zone of the crypt. However, this effect was abolished in healthy Bmi-1eGFP/+ mice, as was the ability of GLP-2 to increase crypt number in haploinsufficient animals subjected to irradiation-induced intestinal damage. Collectively, these findings indicate that a full complement of Bmi-1 is required for the proliferative actions of GLP-2 in the murine small intestine.
Although not well understood, a role for polycomb-repressive complex proteins in regulating TA zone proliferation through the suppression of cell differentiation has been suggested (29). Bmi-1 has been shown to effect gene silencing through histone modification in hematopoietic stem cells and mouse embryonic fibroblasts (30, 31), and histone modification of key differentiation markers has been shown to change along the length of the crypt-villus axis (32). Thus, although Bmi-1 may be required to directly support TA zone proliferation in response to GLP-2, Bmi-1 may also play an indirect role through suppression of cell differentiation and, thus, the movement of cells out of the TA zone. This theory is supported by the increased migration of BrdU-labeled cells over a 24-hour period in response to GLP-2 that was observed in the Bmi-1 haploinsufficient animals. Future studies using inducible lineage-tracing approaches will be required to examine this phenomenon further.
The results of the present study confirm those of a previous report, which showed that administration of h(Gly2)GLP-2 for 6 or 14 days before 11- to 16-Gy irradiation increases the number of crypts that survive 4 days after damage (5). Because crypt regeneration in this model of intestinal damage is dependent upon the proliferation of at least 1 clonogenic stem cell per crypt (26), and the reserve stem cells are relatively radiation resistant as compared with the rapidly proliferating active stem cells (15, 18), these findings still suggest a role for the reserve stem cells in the mechanism of action of GLP-2. Because no increase in the stem cell zones was noted in these studies and, indeed, a small decrease was observed in cell positions 4–5 after 10-Gy radiation, these findings are consistent with a proposed model of radiation-induced reserve stem cell activation, in which proliferation of the position 4 stem cells results in increased numbers of active stem cells which, in turn, serve to repopulate the crypt-villus epithelium (18).
Although it has proven difficult to directly assess the role of the quiescent, Bmi-1+ stem cell in the hierarchy of physiological crypt proliferation, ablation of Bmi-1 results in abnormal crypt architecture and/or complete loss of crypts (14, 21). Conversely, survival of the position 4 reserve stem cells after radiation enhances crypt regeneration (15). In the present study, the massive crypt regeneration that was induced by 10-Gy radiation appeared to override the proliferative effects of chronic GLP-2 administration, possibly as proliferation had already reached maximal capacity. Nonetheless, intestinal weight was increased in a genotype-independent manner, and villus height was increased in the wild-type animals in this setting (although this was attenuated in the haploinsufficient animals), suggesting that the tropic effects of GLP-2 were retained after 10-Gy radiation, likely mediated through an increase in the total number of crypts. Previous studies have suggested that GLP-2 can induce crypt fission, possibly through an IGF-2-dependent pathway (12, 33). Hence, due to lack of expression of the GLP-2 receptor in the intestinal epithelial cells, the proliferative effects of GLP-2 in the gut are mediated indirectly, through an IGF- and intestinal epithelial IGF-1 receptor-dependent mechanism (12, 16, 34, 35). Furthermore, IGF-1 signaling has been shown to promote the growth of the reserve intestinal stem cells in organoid culture (18), as well as to confer radiation resistance to mouse embryonic fibroblasts (36). These findings therefore provide further support for a role of the Bmi-1+ reserve intestinal stem cell in the proliferative responses to GLP-2 after radiation-induced injury.
The identity of the crypt cells that proliferate in response to GLP-2 has remained elusive, likely due to experimental variations in the proliferative marker used, the duration of treatment, and the genotype of the mouse. The results of the present study using 1-hour EdU or BrdU incorporation as the proliferative marker clearly implicate the midcrypt/TA zone as the main region of GLP-2-responsive proliferating cells after both acute and chronic administration to C57Bl/6 Bmi-1+/+ mice. Although similar findings on BrdU incorporation have been made by others in response to acute treatment of C57Bl/6 mice (37), cells lower in the crypt are labeled in CD1 mice (11). However, expression of the proliferation-associated Ki-67 antigen, is increased high in the crypt of C57Bl/6 (16) mice, and in the mid-TA region of both C57Bl/6 (34) and CD1 (12) mice, but is barely increased by the same treatment in 129Sv/J animals (12). Although total mitotic index is increased after treatment with GLP-2 analogs, nothing has been reported to date as to the specific crypt cells that proliferate in response to GLP-2 in humans (8). Collectively, these findings indicate the need for specific studies of biopsy samples collected from humans administered GLP-2 analogs to determine the identity of the responsive cells, rather than extrapolating the results of studies on GLP-2-induced proliferation in murine models.
Consistent with a role for Bmi-1 in the actions of GLP-2, chronic administration of this intestinotropic peptide enhanced expression of Bmi-1 in the jejunal mucosa. However, an unexpected finding of the present study was the presence of normal levels of Bmi-1 mRNA transcripts in the Bmi-1eGFP/+ mice. Notwithstanding, GLP-2 treatment failed to increase Bmi-1 transcript expression in the haploinsufficient animals. The specificity of these observations is indicated by the absence of similar changes in either Lgr5, a marker for the active stem cells (13), or other markers of intestinal stem cells and progenitor cells. These findings therefore suggest that healthy Bmi-1eGFP/+ mice undergo an adaptive response to the loss of 1 copy of Bmi-1, thereby maintaining normal intestinal structure, but that no further increase in Bmi-1 expression and, hence, in crypt cell proliferation, can be mounted in the face of the increased proliferative demand imposed by GLP-2 treatment.
An unexpected finding of the present study was the maintenance of normal intestinal weight and crypt-villus height responses to chronic GLP-2 treatment in the Bmi-1eGFP/+ mice, despite the absence of a proliferative response in these animals. Similar observations have been reported for the growth effects of GLP-2 in intestinal epithelial IGF-1 receptor null mice, which also do not demonstrate GLP-2-induced proliferation (16). However, our finding of decreased markers of apoptosis in the current study suggests that this mechanism may contribute to the maintenance of intestinal growth in the Bmi-1 haploinsufficient mice.
Finally, a long-acting analog of GLP-2, h(Gly2)GLP-2, has been found to increase dysplasia and promote the growth of intestinal tumors in rodent models of carcinogenesis in association with increased expression of β-catenin, a mediator of the proproliferative canonical Wnt pathway (38, 39). These findings have contributed to the recommendation for frequent colonoscopy screening in patients administered teduglutide. Of interest to the current study, Bmi-1 expression is increased by β-catenin signaling (21, 40), and Bmi-1 contributes to tumorigenesis through interaction with the proproliferative canonical Wnt, cellular-myelocytomatosis oncogene (c-Myc) pathway (41). Furthermore, although only low Bmi-1 expression has been found in human colonic tumors, Bmi-1 has been linked to both the recurrence and chemoresistance of cancer cells (42, 43). These findings therefore provide a plausible mechanism by which GLP-2 may promote the growth of intestinal tumors, at least in murine models.
In summary, the results of the present study indicate that a full complement of Bmi-1 is required for the intestinotropic actions of GLP-2. The consistent observation of a proliferative effect of GLP-2 in the TA zone of the crypt further suggests that the role for Bmi-1 in the actions of this hormone may be to prevent differentiation and/or support proliferation of the TA cells. As a long-acting GLP-2 analog has recently been implemented in the clinic for the treatment of patients with short bowel syndrome, these findings provide evidence for a complex mechanism of action, which may also include protective effects in the setting of radiation-induced mucositis.
Significance
The intestinal hormone, GLP-2 is established to enhance intestinal growth and function in both preclinical models and humans with intestinal insufficiency due to massive bowel resection. These beneficial actions of GLP-2 have led to the development of a long-acting analog for therapeutic use in patients with short bowel syndrome. However, the cellular targets of GLP-2 action, and the mechanisms by which GLP-2 enhances intestinal growth remain poorly defined. The results of the present study demonstrate that the intestinal proliferative actions of GLP-2 require a full complement of the polycomb-repressive complex family protein and intestinal crypt cell marker, Bmi-1, as effected, at least in part, through enhanced proliferation of the cells in the TA zone of the crypt.
Appendix
Antibody Table
Antibody Table
| Peptide/Protein Target | Antigen Sequence (if Known) | Name of Antibody | Manufacturer, Catalog Number, and/or Name of Individual Providing the Antibody | Species Raised in; Monoclonal or Polyclonal | Dilution Used |
|---|---|---|---|---|---|
| Bmi-1 | Anti-Bmi-1 | Abcam, ab38295 | Rabbit; polyclonal | 1:100 | |
| BrdU | Anti-BrdU | Thermo Fisher Scientific, Inc, B35128 | Mouse; monoclonal | 1:100 | |
| Sucrase-isomaltase | Antisucrase-isomaltase | Santa Cruz Biotechnology, Inc, sc-393470 | Goat; polyclonal | 1:150 |
| Peptide/Protein Target | Antigen Sequence (if Known) | Name of Antibody | Manufacturer, Catalog Number, and/or Name of Individual Providing the Antibody | Species Raised in; Monoclonal or Polyclonal | Dilution Used |
|---|---|---|---|---|---|
| Bmi-1 | Anti-Bmi-1 | Abcam, ab38295 | Rabbit; polyclonal | 1:100 | |
| BrdU | Anti-BrdU | Thermo Fisher Scientific, Inc, B35128 | Mouse; monoclonal | 1:100 | |
| Sucrase-isomaltase | Antisucrase-isomaltase | Santa Cruz Biotechnology, Inc, sc-393470 | Goat; polyclonal | 1:150 |
Antibody Table
| Peptide/Protein Target | Antigen Sequence (if Known) | Name of Antibody | Manufacturer, Catalog Number, and/or Name of Individual Providing the Antibody | Species Raised in; Monoclonal or Polyclonal | Dilution Used |
|---|---|---|---|---|---|
| Bmi-1 | Anti-Bmi-1 | Abcam, ab38295 | Rabbit; polyclonal | 1:100 | |
| BrdU | Anti-BrdU | Thermo Fisher Scientific, Inc, B35128 | Mouse; monoclonal | 1:100 | |
| Sucrase-isomaltase | Antisucrase-isomaltase | Santa Cruz Biotechnology, Inc, sc-393470 | Goat; polyclonal | 1:150 |
| Peptide/Protein Target | Antigen Sequence (if Known) | Name of Antibody | Manufacturer, Catalog Number, and/or Name of Individual Providing the Antibody | Species Raised in; Monoclonal or Polyclonal | Dilution Used |
|---|---|---|---|---|---|
| Bmi-1 | Anti-Bmi-1 | Abcam, ab38295 | Rabbit; polyclonal | 1:100 | |
| BrdU | Anti-BrdU | Thermo Fisher Scientific, Inc, B35128 | Mouse; monoclonal | 1:100 | |
| Sucrase-isomaltase | Antisucrase-isomaltase | Santa Cruz Biotechnology, Inc, sc-393470 | Goat; polyclonal | 1:150 |
Acknowledgments
We thank Dr C. Collymore, University of Toronto, for assistance with the radiation studies, and to Dr J. Gommerman, University of Toronto, for use of the irradiator.
This work was supported by a Department of Physiology, University of Toronto Graduate Stimulus Award (B.R.S.); a summer studentship from the Canadian Association of Gastroenterology/Crohn's and Colitis Foundation (H.Y.M.P.); and the Canada Research Chairs program (P.L.B.). This work was also supported by the Canadian Institutes of Health Research Grant MOP-12344.
Disclosure Summary: The authors have nothing to disclose.
Abbreviations
- Bax
Bcl2-associated X protein
- Bcl2
B-cell lymphoma-2
- Bmi-1
B-cell lymphoma Moloney murine leukemia virus insertion region-1 homolog
- BrdU
5-bromo-2′-deoxyuridine
- EdU
5-ethynyl-2′-deoxyuridine
- eGFP
enhanced green fluorescent protein
- GLP-2
glucagon-like peptide-2
- h
human
- Lgr5
leucine-rich repeat-containing G protein-coupled receptor-5
- RT
room temperature
- TA
transit-amplifying
- Wnt
wingless plus chromosomal integration site of mouse mammary tumour virus on mouse chromosome 15.
