Loss of enteroendocrine cells in mice alters lipid absorption and glucose homeostasis and impairs postnatal survival (original) (raw)
Generation of intestinal-specific Ngn3-knockout mice. To study the consequence of a complete and specific ablation of Ngn3 expression in the small and large intestine, we have generated mice carrying a floxed Ngn3 allele (Ngn3+/flox) (see Methods, Figure 1, and Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI40794DS1). Ngn3+/flox and Ngn3flox/flox mice developed normally, reached adulthood, were fertile, and showed normal glucose levels in the urine. In order to specifically ablate Ngn3 in the intestinal epithelium, we used transgenic mice expressing the Cre recombinase, under the control of a 9-kb regulatory region of the murine villin gene (vil-cre) (19). The 9-kb regulatory region of the villin gene has been shown to target stable and homogeneous expression of the Cre recombinase in small and large intestine along the crypt-villus axis, in the immature, undifferentiated cells of the crypt as well as in differentiated enterocytes (19). The following crosses were set up: Ngn3 heterozygous mice (Ngn3+/Neo) (11) or Ngn3+/flox mice carrying the vil-Cre transgene (Ngn3+/neo;vil-cre, Ngn3+/flox;vil-cre) were crossed with Ngn3+/flox or Ngn3flox/flox mice to obtain Ngn3neo/flox;vil-cre or Ngn3flox/flox;vil-cre mice, respectively. In the following, all experiments were done with mice coming from a pure CD1 background. As the efficiency of Ngn3 deletion in Ngn3neo/flox;vil-cre or Ngn3flox/flox;vil-cre mice was identical, judged by the complete absence of chromogranin A–positive cells at P3.5 or adult stages, they will in the following be referred to as Ngn3 knockout (Ngn3_Δ_int) or mutant mice. As controls, Ngn3+/+ or Ngn3+/+;vil-cre littermates were used.
Generation of animals with a conditional Ngn3 allele. (A) Schema depicting the Ngn3 locus and the targeting construct. (B) Targeted Ngn3 allele before and after the excision of the FRT flanked “PGK-_Neo_” selection cassette by the FLP recombinase. Stars in B indicate the position of the 5′- and 3′-external probes used for Southern blot analysis (see Supplemental Figure 1). (A and B) The black boxes indicate the Ngn3 coding sequence. The PGK-Neomycin selection cassette, the loxP, and FRT sites are indicated as well. X, XbaI; Sp, SpeI; E, EcoRI; Pm, PmeI; As, AscI.
Ngn3Δint mice exhibit severe growth retardation. Ngn3_Δ_int mice are born with the expected Mendelian frequency and are at P0.5 visually indistinguishable from control littermates. However, at this stage, Ngn3_Δ_int mice already show, in average, a slightly lower body weight (control mice, 1.5 ± 0.17 g; mutant mice, 1.4 ± 0.14 g; P = 0.00809) and are from P3.5 on clearly smaller than control littermates (Figure 2A and Figure 3). In a CD1 background, 50% of the mutant mice died within the first 8 days of life after minimal weight gain. In order to rule out that the early death of some mutant mice is due to a problem in nursing, we dissected out the whole intestinal tract from dead and living mutant mice. The macroscopic analyses of the dissected whole intestine clearly showed the presence of milk in the stomach of surviving mutant mice (Figure 2C) and mutant mice found dead (Figure 2D). Importantly, surviving mice had soft stool, which did not cease with age. In the following, surviving male and female Ngn3_Δ_int mice exhibited variant degrees of growth retardation, and even at adult stages, mutant mice stayed about 30% smaller than control littermates (Figure 3). However, although the intestinal tract of adult mutant mice seems visually to be shorter than the intestinal tract of control mice, its length is proportional to body weight (Supplemental Figure 2).
Fifty percent of mutant mice with an intestinal deletion of Ngn3 (Ngn3_Δ_int) die within the first 8 days of life. (A) Photography taken at P3.5 of a wild-type, mutant (MTa), and mutant mouse found dead (MTb) from the same litter. From P3.5 on, mutant animals start to be visibly smaller than control littermates. (B–D) Photography of the dissected intestinal tractus from the wild-type and mutant mice shown in A, taken with the same magnification (original magnification, ×0.8). The presence of milk in the stomach of mutant mice indicates their ability to suck milk.
Mutant mice gain less body weight than control littermates. During the first 2 weeks of life, the weight of wild-type (control) and mutant mice was taken every day (left graph) and thereafter once a week for a period of 6 more weeks (right graph). Mutant mice do gain less weight than control mice and keep, at adult stages, about 30% lower body weight than control littermates. Control, n = 66; mutant, n = 35, for all time points measured; P < 0.008 (left graph). Male control, n = 8; male mutant, n = 5; female control, n = 5; female mutant, n = 4; *P < 0.05, #P < 0.01, †P < 0.001 (right graph).
Specific and efficient ablation of Ngn3 and of enteroendocrine cells in the intestine of Ngn3Δint mice. To evaluate the ablation efficiency of Ngn3 in Ngn3_Δ_int mice, total RNA from the intestine of E19.5 and adult mutant and wild-type mice was prepared, and the expression of Ngn3 was evaluated by quantitative RT-PCR (RT-QPCR). These analyses showed a 90%–100% reduction of Ngn3 mRNA expression (Supplemental Figure 3) all along the intestinal tract at embryonic and adult stages. Likewise, RT-QPCR and immunohistochemistry for chromogranin A showed an 95%–100% reduction of its mRNA (Supplemental Figure 3A) and a complete loss of chromogranin A+ cells (Supplemental Figure 4) in the intestine of E19.5 mutant embryos. The low amount of chromogranin A (Chga) mRNA detected in the intestine of some E19.5 mutant animals is most likely due to the contamination with pancreatic tissue, as we detected in 2 out of 6 animals analyzed significant levels of insulin-1 (Ins1) mRNA (data not shown), whereas the mRNA for Gip, an intestinal-specific endocrine hormone, is completely gone (Supplemental Figure 3A). Furthermore, immunohistochemistry analyses for Ngn3 showed a complete loss of Ngn3+ enteroendocrine progenitors along the proximal-distal axis of adult mutant intestine (Figure 4B). Taken together, these results show that already at embryonic stages Ngn3_Δ_int mice showed a complete lack of all enteroendocrine cells, suggesting that the survival of some mutant mice is not due to a mosaic deletion of Ngn3 in these animals. We have previously shown that at embryonic stages all enteroendocrine cell development is Ngn3 dependent (12). However, since Ngn3 global knockout mice die shortly after birth, we could not analyze whether at adult stages enteroendocrine cell differentiation or the differentiation of the other intestinal cell types from intestinal stem cells is Ngn3 dependent or not. To evaluate the latter we performed immunohistochemistry analyses for chromogranin A, which marks all enteroendocrine cell types except CCK- and motilin-expressing cells, and for CCK/gastrin on tissue sections of Ngn3_Δ_int and control mice at different adult stages. This analysis revealed a complete lack of chromogranin A– and CCK/gastrin-positive enteroendocrine cells all along the intestinal tract, demonstrating that enteroendocrine cell development is Ngn3 dependent at adult stages also (Figure 4, C–F). Importantly and as expected, glucagon- and insulin-expressing α- and β-cells, respectively, are detected in pancreatic islets (see below), confirming the specific intestinal deletion of Ngn3.
Conditional inactivation of Ngn3 in only the intestine results in a complete loss of all enteroendocrine cells all along the proximal-distal axis of the intestine. Sections of adult duodenum, jejunum, ileum, and colon were examined for the presence of endocrine cells in control and mutant animals by immunofluorescence. Images presented are from the jejunum of control (A, C, E, G, and I) and mutant (B, D, F, H, and J) animals and are representative for the general loss of all enteroendocrine cells in mutant animals. Villin-Cre–mediated inactivation of Ngn3 results in a complete loss of all Ngn3+ enteroendocrine progenitors (B), which in wild-type animals are located in the intestinal crypt compartment (A, arrows). Likewise, mutant animals are also devoid of chromogranin A+ (D), Cck/gastrin+ (F), Glp1+ (H), and Gip+ (J) cells, normally located in the villi of wild-type mice (arrows in C, E, G, and I), respectively. The age of the animals analyzed is 10–12 weeks. Original magnification, ×10.
Intestinal ablation of Ngn3 leads to a perturbed intestinal morphology. Histological analyses of the small intestine of adult Ngn3_Δ_int and control mice revealed that in mutant Ngn3_Δ_int mice villi are frequently blunted or club shaped and often show dilatation of up to 400 microns in diameter (Figure 5B). Immunohistochemistry staining for total laminin further demonstrates the frequent dilatation of mutant villi and the strong detachment of the epithelium from the basement membrane (Figure 5D). Published data showing a slight but significant increase of goblet cells in Ngn3 global knockout mice (12), which die shortly after birth, and cell lineage studies showing that Ngn3 progenitors can also contribute to some goblet and Paneth cells (15) prompted us to look at the distribution of these 2 cell types. Ngn3_Δ_int mice showed no obvious change in the number and location of Paneth cells in the small intestine (Figure 5, E and F, and Supplemental Figure 5A). In contrast to the increased number of goblet cells found at early postnatal stages in Ngn3 global knockout mice, we did not observe in Ngn3_Δ_int mice at adult stages a similar increase (Figure 5, G and H, and Supplemental Figure 5B). The histological analyses of the large intestine revealed a reduction in the length of the glands of up to 1.5 times compared with the wild-type mice (Figure 6, A, B, and E). In addition, goblet cells seemed to be larger and mostly devoid of mucus in mutant compared with control large intestine (Figure 6, C and D).
Intestinal ablation of Ngn3 leads to an altered morphology of the small intestine but normal Paneth and goblet cell differentiation. Sections of adult wild-type and mutant duodenum, jejunum, and ileum were examined for their overall appearance (A–D) and the presence of Paneth (E and F) and goblet cells (G and H). Images presented are from the jejunum of control (A, C, E, and G) and mutant (B, D, F, and H) animals and are also representative for the phenotype observed in the duodenum and ileum of mutant animals. (A and B) H&E staining clearly shows the frequent blunt or club-shaped appearance of the villi and the disorganization of the crypt compartment of mutant animals compared with control small intestine. (C and D) Immunofluorescence analyses with an antibody recognizing all laminins, showing the frequent detachment of the intestinal epithelium from the lamina propria in mutant small intestine. (E and F) Immunohistochemistry with an anti-lysozyme antibody demonstrates normal appearance and location of Paneth cells (arrows in E and F) in mutant small intestine. (G and H) Likewise, periodic acid–Schiff staining shows that intestinal ablation of Ngn3 does not alter the location or number of goblet cells (arrows in G and H) in mutant animals. For Paneth and goblet cell counts, see Supplemental Figure 5. The age of the animals analyzed is 10–12 weeks.
The large intestine of mutant mice shows shorter glands. Sections of adult wild-type (A and C) and mutant (B and D) large intestine were examined for their overall appearance (A and B) and the presence of goblet cells (C and D). (A and B) H&E staining clearly shows the reduction in the length of the glands in the large intestine of mutant mice (B, measurement in E) compared with control tissue. (C and D) Periodic acid–Schiff staining of goblet cells. (E) The colonic glands in mutant animals are on average 26% shorter than the colonic glands of control animals (n = 4; 50–60 glands were analyzed per genotype). **P < 0.01. The age of the animals analyzed is 10–12 weeks.
Ngn3-deficient small intestine showed an enlarged multifocal proliferating crypt compartment and an accelerated cell turnover. The most striking feature of Ngn3_Δ_int intestine was the frequent disorganization of the crypt compartment. The transiently amplifying crypt compartment seemed to be larger and more abundant in Ngn3_Δ_int mice than in control mice of the same litter and age (Figure 5, A and B, and Figure 7, E and F). To evaluate whether this is due to an increase in cell proliferation, we analyzed by immunohistochemistry the proliferation marker Ki67. This analyses clearly showed that the enlargement of the crypt compartment seen in mutant mice is due to an increase in the number of proliferating cells of up to 44% (±11%) compared with wild-type mice (Figure 7, A and B). However, the enlarged proliferative crypt compartments in mutant mice did not result in longer or more villi. In contrary, measurement of their length clearly showed that mutant villi are up to 40% shorter than the villi in control littermates (Figure 7G). We came up with 3 possibilities for why mutant mice have shorter villi despite the enlarged transiently proliferating crypt compartment: increased apoptosis, accelerated cell turn over, or both. Immunohistochemistry staining for the apoptotic cell marker caspase 3 revealed no difference between mutant and wild-type mice (data not shown). However, by performing a 24-hour BrdU chase, it became clear that mutant mice have an up to 1.6-fold accelerated cell turnover (Figure 7, C and D, and measurement in H), which most likely is the reason for the shorter villi seen in mutant mice. Importantly, at embryonic stage E19.5, the intervillus region seemed to be slightly de-organized but otherwise did not show an increase in proliferating Ki67+ cells (Supplemental Figure 4).
Altered cell homeostasis in _Ngn3_-deficient small intestine. Sections of adult control (A, C, and E) and mutant (B, D, and F) intestine were examined for the status of the proliferative crypt compartment (A, B, E, and F), villus length (measurements in G), and cell turn over (C, D, and measurement in H). (A and B) Immunofluorescence staining for the proliferative cell marker Ki67 clearly demonstrates an up to 2-fold enlargement of the proliferative crypt compartment (dashed bars in A and B) in _Ngn3_-deficient intestine. Arrows point to chromogranin A+ cells. (G) Measurement of the villi length indicates an approximately 40% reduction in their length in mutant intestine. (C and D) Twenty-four hours before dissection, adult control and mutant mice were injected with BrdU, and BrdU-labeled cells were then visualized by immunofluorescence staining. Then the distance from the villus base to last labeled BrdU+ cell was measured (dashed bars in C and D), demonstrating a 1.6-fold accelerated cell turnover in _Ngn3_-deficient intestine. Arrows point to chromogranin A+ cells. (E and F) H&E staining showing the enlargement of the crypt compartment seen in Ngn3 mutant intestine. n = 3. The age of the animals analyzed is 10–12 weeks.
Strongly reduced intestinal absorptive surface area and impaired lipid absorption in Ngn3Δint mutant mice. The growth retardation of mutant mice, their frequent death during the weaning period, the appearance of soft yellowish liquid stool, which suggested that they might have steatorrhea, and the perturbed intestinal morphology prompted us to further characterize the ultrastructure of the absorptive cells and look as well at the absorption of lipids. The presence of ample milk in the stomach of mutant mice (Figure 2C) already suggested that the frequent death and growth retardation of the surviving mutant mice results from malabsorption rather than malnutrition. The apical microvilli of the absorptive cells greatly enhance the absorptive surface area of the intestine. Surprisingly, electron microscopy analyses showed the microvilli on the absorptive cells of mutant mice to be sparser, approximately 60% shorter, but twice as large than in control littermates (Figure 8, I and J), resulting in an approximately 44% reduction of the brush border of the absorptive cells in the small intestine of mutant mice. As mentioned above, the yellowish stool during the weaning period of mutant mice, which at postweaning stages still is of a lighter brownish color compared with that of control mice, prompted us to look at the absorption of lipids by the enterocytes. Oil red O staining of the mutant gut revealed a clear reduction of the presence of lipids in the enterocytes and the lamina propria compared with control samples (Figure 9, A and B, and Supplemental Figure 6). In addition, electron microscopy analyses revealed a strong reduction in the number of chylomicrons, which transport dietary lipids from the intestine to other locations in the body, in the absorptive cells of mutant mice (Figure 9, C and D). These findings are also in agreement with the reduced levels of total cholesterol, HDL cholesterol, and triglycerides found in mutant mice (Figure 10A). The enteroendocrine hormones CCK and secretin are known to regulate the secretion of digestive enzymes from the pancreatic exocrine cells. As reduced levels of digestive enzymes, notably that of lipase, could have an effect on lipid absorption, we analyzed the levels of lipase in the blood of mutant and control mice. Control and mutant mice showed similar levels of lipase in the blood (Figure 10B), despite a complete loss of CCK- and secretin-secreting CCK and S cells, respectively. In addition, control and mutant mouse pancreas showed no difference in quantity or distribution of zymogen granules in acinar cells (Figure 9, E and F). Taken together, these results suggest that the reduced numbers of Oil red O–positive lipid droplets and chylomicrons and the reduced serum levels of total cholesterol, HDL cholesterol, and triglycerides found in mutant mice is rather due to impaired lipid absorption than due to impaired lipid digestion.
Strong reduction of the intestinal absorptive surface area but normal expression of brush border enzymes and glucose transporters in Ngn3-deficient mice. Sections of control (A, C, E, G, and I) and mutant (B, D, F, H, and J) intestine were examined for the status of the absorptive cell population. Analyses of the lactase activity (A and B) and immunofluorescence staining for sucrase-isomaltase (C and D), the active glucose transporter Glut2 (E and F), and the passive glucose transporter SGLT1 (G and H) did not show any difference between control and mutant tissue, respectively. (I and J) Ultrastructural analysis of the brush border of the absorptive cells demonstrates a strong reduction of the microvilli length in mutant mice. The dashed lines in A and B indicate the bottom of the crypt compartment. The age of the mice analyzed in A and B is P1.5 and in C–J is 10–12 weeks. Original magnification, ×20 (A–H); ×40,000 (I and J).
Impaired lipid absorption in Ngn3_Δ_int mutant mice. Oil red O, which stains neutral fats, was used to visualize lipid droplets in control (A) and mutant (B) tissue. Mutant small intestine clearly shows a strong reduction in the amount of lipid droplets (B) compared with control tissue (A). Arrowheads in B point to some lipid droplets found in mutant tissue. Ultrastructural analysis of the absorptive cells furthermore demonstrates a clear reduction in the number of neutral lipid-containing chylomicrons (arrows in C and D) in Ngn3–deficient intestine (D). (E and F) Ultrastructural analysis of exocrine pancreas at adult stage shows no difference in the number or quality of zymogen granules (arrowheads) in acinar cells in control (E) and mutant animals (F). The age of the mice analyzed is 10–12 weeks. Original magnification, ×10,000 (C and D); ×4,000 (E and F).
Reduced levels of cholesterol and triglyceride in the blood of Ngn3_Δ_int mice. (A) Twenty-seven–week-old Ngn3_Δ_int mice have reduced cholesterol and triglyceride levels in the blood (n = 6; *P < 0.05, **P < 0.01). (B) Similar levels of lipase, a digestive enzyme mainly produced and secreted by the acinar cells in the pancreas, are found in the blood of 27-week-old control and mutant mice (n = 4–5).
Furthermore, we also analyzed by immunohistochemistry the expression of the brush border enzymes lactase and sucrase-isomaltase (Figure 8, A–D), respectively, and the active glucose transporter Glut2 (Figure 8, E and F) (20) and the passive glucose transporter SGLT1 (Figure 8, G and H) (20), which revealed no qualitative difference between mutant and wild-type mice. Likewise, RT-QPCR for Glut2 did not show any difference in the expression level between control and mutant intestinal tissue (data not shown), and the oral glucose tolerance tests (OGTTs) revealed that mutant animals take up glucose into the blood from the gastrointestinal tract with the same efficiency as wild-type littermates (Figure 11A, see time point 15 minutes).
Altered glucose homeostasis in Ngn3_Δ_int mice. Control (filled squares) and mutant mice (filled circles) were subjected to either an oral (A, OGTT) or intraperitoneal (B, IPGTT) glucose challenge or an ITT. Blood glucose levels were then measured at the indicated time points. (A) In the OGTT, mutant mice show a slightly improved glucose clearance from the blood. (B) In the IPGTT, at all time points measured, blood glucose levels of mutant mice do not rise to the same levels as in control mice. (C) The ITT clearly shows an improved insulin sensitivity of mutant compared with control mice. n = 6–7, for control and mutant mice. “0” indicates the blood glucose level before the glucose challenge. *P < 0.05, **P < 0.01, ***P < 0.001. The age of the mice analyzed is 7–10 weeks.
Altered glucose homeostasis in Ngn3Δint mice. GLP-1 and GIP, 2 incretin hormones produced and secreted after food ingestion by intestinal L-cells and K-cells respectively, have been shown to potentiate glucose stimulated insulin secretion of pancreatic β cells (1). Compound knockout mice for GLP1 and GIP receptor display a greater glucose intolerance following oral glucose challenge (21). The absence of chromogranin A–positive cells in Ngn3_Δ_int mice (Figure 4D) already suggest that mutant mice lack L and K cells secreting GLP1 and GIP, respectively. Indeed, immunohistochemistry and RT-QPCR analyses for GLP1 and GIP demonstrated their complete absence in mutant tissue (Figure 4, G–J, and Supplemental Figure 3). In addition, following a glucose challenge, we found in the blood serum of Ngn3_Δ_int mice a complete lack of GIP (Figure 12). Unfortunately, our attempts to measure GLP1 in the blood serum failed, which is most likely is due to its known rapid degradation (22, 23). However, mutant mice showed also a complete lack of PYY (Figure 12), further supporting the ablation of L-cells. In the following, to evaluate the impact of a complete loss of all incretin hormones on disposal of a glucose load, we performed an OGTT (Figure 11A) and compared the results to an intraperitoneal glucose tolerance test (IPGTT) (Figure 11B). In the OGTT, mutant and control mice showed, after an overnight fasting period, the same fasting blood glucose concentration. However, although initially the glucose concentration in the blood of mutant and control mice rose to the same levels, mutant mice showed an improved glucose clearance from the blood (Figure 11A). Surprisingly, the IPGTT revealed an even more pronounced phenotype in our Ngn3_Δ_int mice, where at all time points taken mutant mice had far lower glucose levels in the blood than control mice (Figure 11B). Already during all our dissections of adult Ngn3_Δ_int mice, we observed a strong reduction in the extent of abdominal fat compared with control littermates of the same sex. Measurement of the body composition of 27-week-old mice clearly demonstrated that Ngn3_Δ_int mice have strongly reduced body fat content, an improved lean mass and a better BMI than control littermates of the same sex (Figure 13, A–C). As reduced body fat results in improved insulin sensitivity and also an improved glucose uptake by the peripheral tissue, we subjected 14-week- and 27-week-old mutant and control mice to an insulin tolerance test (ITT). At both ages, analyzed Ngn3_Δ_int mice have a slightly improved insulin sensitivity (Figure 11C and Figure 13D) compared with control littermates, which is also seen by their difference in average under the curve (AUC) between 0–45 minutes after the injection of insulin (Figure 13D, left inset). In addition, the ITT also revealed that Ngn3_Δ_int mice had an approximately 30% lower fasting blood glucose level than control littermates (blood glucose levels at time 0 [T0], control mice, ~125 mg/dl, Ngn3_Δ_int mice, ~88 mg/dl; Figure 13D, right inset) and a strongly blunted hypoglycemic response to the injected insulin (Figure 11C and Figure 13D). As our mutant mice showed improved insulin sensitivity, we could not exclude that in the IPGTT we have missed an early blood glucose peak. We therefore repeated the IPGTT and measured blood glucose levels, also at 5 minutes after glucose injection. This revealed that even in the early phase the blood glucose levels are much lower than in control mice (Supplemental Figure 7).
Reduced levels of several intestinal and pancreatic hormones in the blood of intestinal Ngn3-deficient mice. After an oral glucose challenge, blood was taken from Ngn3_Δ_int and control mice and analyzed. Ngn3_Δ_int mice show a complete lack of the intestinal hormones GIP and PYY and reduced levels of ghrelin. Likewise, blood serum concentration levels of the pancreatic hormones insulin, amylin, and PP are also strongly reduced. The age of the mice analyzed is 9–10 weeks. n = 6, for mutant and control mice.
Improved BMI and insulin sensitivity in 27-week-old Ngn3_Δ_int mice. (A and B) The body composition of age- and sex-matched mutant and control mice was analyzed. Ngn3_Δ_int mice have approximately 30% less body fat (A), are leaner (B), and show an improved BMI (C) compared with control mice. (D) Control (filled squares) and mutant (filled circles) mice were fasted and subjected to an ITT. Mutant mice show improved insulin sensitivity, seen also by the reduction in the average under the curve between 0–45 minutes after insulin injection (AUC, left inset). In addition, mutant mice show a clear difference in the fasting blood glucose level (right inset). n = 6–7. *P < 0.05, **P < 0.01.
Altered islet morphology in Ngn3Δint mice. Results obtained by different groups suggest that GLP-1 not only potentiates glucose-stimulated insulin secretion but also stimulates islet neogenesis and β cell proliferation (4–6). In addition GLP–1 receptor–knockout (GLP-1R–knockout) mice exhibit an altered islet morphology with and shift from large islets to more medium or single islets (24). Moreover, these mice showed an increased proportion of islets with centrally located α-cells, which are normally located at their periphery. As our mutant mice showed no intestinal Glp1 expression (Figure 4 and Supplemental Figure 3B), we evaluated the distribution of islet sizes in Ngn3_Δ_int mice according to the published criteria (24), which classified the islets as single (<300 μm2), small (300–5,000 μm2), medium (5,000–20,000 μm2), or large (>20,000 μm2). In each experimental group we analyzed 1,500–2,000 islets. This showed that, like in the GLP-1R–knockout mice (24), our Ngn3_Δ_int mice, which lack all enteroendocrine cells, including the Glp1-secreting L-cells, show an shift from large to single islets (Figure 14, A and B). In addition, and like also seen in the GLP-1R–knockout mice (24), we also found an increasing number of medium and large islets with centrally located α-cells (Figure 14, C and D). These data confirm the published data (24) showing the importance of Glp1 for the organization of the adult endocrine islet cells. However, in contrast to Ngn3_Δ_int mice GLP-1R–knockout mice exhibit mild fasting hyperglycemia and glucose intolerance after an oral glucose challenge (25).
Altered islet architecture in Ngn3_Δ_int mice. (A) Ngn3_Δ_int mice show a decrease of large and an increase of single islets compared with control mice. (B) The contribution of islets of a particular size to the total islet volume is shifted from large to single and small islets in mutant mice compared with control mice. (C) Immunostaining for insulin (green) and glucagon (red) on a pancreatic section of control and Ngn3_Δ_int mutant mice. In mutant mice, glucagon-positive cells appear in the periphery and ectopically in the center of the islets, whereas in control mice, they are normally exclusively located at the periphery. (D) Quantitative analyses of C. (A, B, and D) Control, black columns; mutant, white columns. Islet sizes were classified as follows: single, <300 μm; small, 300–5,000 μm; medium, 5,000–20,000 μm; and large, >20,000 μm; n = 5, all male; 1,500–2,000 islets counted per genotype. The age of the mice analyzed is 8–9 weeks. *P < 0.05, **P < 0.01, ***P < 0.001.
The intestinal food transit is accelerated in Ngn3Δint mice. As mentioned in the beginning, Ngn3_Δ_int mice fed with a low-fat standard diet showed all their life the appearance of soft stool. Likewise, so far, in all dissected Ngn3_Δ_int mice, we hardly found “normal” dry excrement pellets in the colon. In severe cases, almost the whole intestinal tract was filled with liquid excrements, and these mice were in general in bad physical condition. Several hormones secreted by different enteroendocrine cells have been shown to regulate directly or indirectly the gastrointestinal motility (26). For example, ghrelin and motilin stimulate gastrointestinal motility, whereas GLP-1 and PYY inhibit it. To test whether in Ngn3_Δ_int mice intestinal food transit is affected, mutant and control mice were fasted for a prolonged period and then given simultaneously access to artificially colored alimentation (27). The time of the appearance of colored stool was then measured and showed in average a 2.3-times faster intestinal food transit in Ngn3_Δ_int mice compared with control littermates (Figure 15A). In addition, mutant mice excrete up to 2 times more stool than control littermates (Figure 15B). Importantly, the diarrhea seen in our mutant mice could also be the result of reduced expression of the water channels “Aquaporins” or deregulation of ionic transport. To address this point, we analyzed the expression of the Aquaporins 3 (Aqp3), 4 (Aqp4), and 8 (Aqp8), which are specifically expressed in the epithelium of the colon (28, 29), and performed and extensively analyzed the blood serum chemistry. These analyses showed no modulation of the expression of the Aqp3, Aqp4, and Aqp8 at embryonic stage E19.5 and only a very variable but weakly statistical significant increase of Aqp3 and Aqp4 in adult mutant mice (Supplemental Figure 8). In addition, blood chloride, sodium, calcium, potassium, magnesium, phosphorus, and iron levels were not altered and bicarbonate levels were only slightly increased in Ngn3_Δ_int mice (Supplemental Figure 9).
Accelerated food transit and increased feces production in Ngn3_Δ_int mice. (A) Age- and sex-matched mutant and control mice were fasted overnight and given then simultaneously access to colored food. The time of the appearance of the first colored stool was then taken and normalized to their body weight. Mutant mice have a 2.3-fold accelerated intestinal food transit. (B) The feces of control and mutant mice were collected after 1 or 4 days, and their weight was taken. Mutant mice show an up to 2-fold increase in feces production compared with control mice. (A and B) n = 4–5. **P < 0.01, ***P < 0.001.