We have generated mice with a conditional Ngn3
allele to specifically inactivate this gene in the intestine using the villin-Cre
) mice, in order to address its requirement for the genesis and differentiation of enteroendocrine cells in the adult.
In the mouse embryo, the development of enteroendocrine cells relies on the proendocrine transcription factor neurogenin 3, which promotes the endocrine fate in pluripotent intestinal stem cells (12
). Indeed, in Ngn3
-deficient mice, neither enteroendocrine nor pancreatic endocrine cells develop, and the mice die postnatally due to severe diabetes (11
). Enteroendocrine cell differentiation starts mostly at late embryonic and postnatal stages, and Ngn3 expression persists in the adult intestine, in which enteroendocrine cells are constantly renewed throughout life. However, the perinatal death of Ngn3
global knockout mice precluded the analysis of the role of Ngn3 in intestinal cell differentiation, and thus it was not known whether Ngn3 has a similar proendocrine function in the adult mouse or other yet unknown functions.
As our data show, the villin-Cre
transgene leads to a specific and complete inactivation of Ngn3
in the intestine and importantly, also to a complete lack of all differentiated enteroendocrine cell types. This result suggests that even at adult stages enteroendocrine cell differentiation is Ngn3 dependent. Surprisingly, histological analyses revealed a disorganized and enlarged crypt compartment in mutant Ngn3Δint
animals. This finding was unexpected, since several enteroendocrine hormones have been shown to positively influence cell proliferation in the intestine, especially GLP2 but also PYY and gastrin (30
), which suggested that the lack of one or several of these hormones might result in a reduction of the proliferative crypt. Our data showed that Ngn3Δint
mice lack all enteroendocrine cells and, in particular, a complete absence of PYY and GLP2. However, in contrast to a possible reduction, Ngn3Δint
mice clearly showed an enlargement of the proliferative crypt compartment, accompanied by an accelerated cell turnover and shorter villi. In addition, Ngn3Δint
mice showed shorter and sometimes misshaped microvilli. These results are in part similar to the results published by Fre and coworkers, who used the villin promoter to target the expression of a constitutively active form of the mouse Notch 1 receptor in all the intestinal epithelium (31
). These mice, which were referred to as Rosa-Notch/Cre+
mice, also show a strong reduction of Ngn3 expression and a marked reduction of enteroendocrine cells. However, the phenotype of Ngn3Δint
mice is clearly different, in that, in mutant animals we do not see an increase in apoptosis or a change in the goblet or Paneth cell numbers or location. This difference most likely is due the fact that in Rosa-Notch/Cre+
mice, not only is Ngn3 expression downregulated but the expression of Math1 is also, which is essential for the development of all the secretory cell lineages. In contrast to the Notch 1 receptor gain-of-function results, Notch 1-2 compound knockout mice or gut-specific inactivation of Rbpj results in an almost complete loss of proliferating crypt progenitors (32
). In Ngn3Δint
mice, in which we have conditionally inactivated Ngn3
in the intestine, the proliferative crypt compartment and cell turn over seemed to be deregulated, which is not due to increased inflammation, since we only find low numbers of lymphocytes and rare plasma cells. In addition, it is unlikely that this is due to the lack of endocrine progenitors at embryonic stages, since we do not see at E19.5 a difference in the proliferative status of the intervillus region in mutant animals. Taken together, our results suggest that at postnatal stages, enteroendocrine progenitors and/or differentiated enteroendocrine cells or their secreted hormones have a so far unknown role to our knowledge in maintaining the homeostasis of the intestinal crypt compartment.
As mentioned above, new born Ngn3
global knockout mice are diabetic showing high elevated urine glucose levels, which was suggested to be the most likely cause for their early death around day 3 (11
). However, Ngn3
global knockout mice also do not develop any enteroendocrine cells (12
), and although none of the so far published knockout mice for the diverse enteroendocrine hormones or their receptors, or even compound mutants (21
), show severe metabolic changes, one cannot exclude that first, the simultaneous loss of all enteroendocrine cells in the full Ngn3
knockout is contributing to their early lethality, and second, that the loss of only the enteroendocrine cells would lead to severe metabolic alterations and by itself be life threatening. As our data show, on a 100% CD1 background, approximately 50% of newborn mice with an intestinal Ngn3
ablation die within the first week of life. The capacity of mutant newborns to survive is not due to a mosaic deletion of Ngn3
, since in all surviving mutant animals analyzed so far, embryonic, post-natal, or adult, we never found any chromogranin A– or hormone-positive enteroendocrine cell. The presence of milk in the stomach of dead mutant mice indicated that they do take up nutrition. We also separated mutant newborns and their mother from the rest of the litter to rule out that competition for parental resources is the reason for the mortality of Ngn3Δint
mice. This step did not change the lethality ratio and suggests that surviving mutant mice reflect individual variability in sensitivity to the lack of enteroendocrine hormones. We can also rule out that Ngn3Δint
mice, like Ngn3
global knockout mice, die from hyperglycemia, as from over 15 tested P3.5 Ngn3Δint
mice, none showed elevated urine glucose levels. Overall, this high lethality ratio clearly shows the importance of enteroendocrine cells and their secreted hormones to sustain life.
The incretin hormones GLP-1 and GIP have been shown to potentiate glucose-stimulated insulin secretion (1
), and compound knockout mice for GLP-1 and GIP receptor display a greater glucose intolerance after oral glucose challenge (21
). Surprisingly, Ngn3Δint
mice, which completely lack GLP-1 and GIP, showed, in contrast to the GLP-1/GIP compound knockout mice, a slightly improved blood glucose clearance after an oral glucose challenge. In addition, in an IGTT, which does not stimulate intestinal GLP-1 and GIP secretion, mutant mice showed an even more blunted curve than control mice. Mutant mice have less body fat, a better lean mass, and showed slightly improved insulin sensitivity, which might contribute to the improved blood glucose clearance in the OGTT. As mutant mice showed very low accumulation of body fat, the improved insulin sensitivity might also enhance a direct uptake of the intraperitoneal injected glucose by the peripheral tissue. However, this might not fully explain the altered glucose homeostasis seen in our mutant mice, and we cannot exclude that the loss of several other enteroendocrine hormones contributes as well to the improved glucose clearance.
During the weaning period, mutant mice showed frequent yellowish stool, which suggested that they might have steatorrhea due to a problem with the absorption of lipids. In support of this hypothesis, surviving mutant mice did not gain weight at the same ratio as control littermates and even at adult stages still showed stool of a light brownish color, suggesting sustained lipid malabsorption. Accordingly, the strongly reduced Oil red O staining and the decreased number of chylomicrons in the intestine of mutant animals would suggest that lipids are not efficiently absorbed by the enterocytes. Lipid malabsorption is not due to an altered pancreatic exocrine function, which would result in an impaired enzymatic processing of lipids, as judged by the normal appearance and distribution of zymogen granules in acinar cells and the presence of similar blood lipase levels in mutant and control animals. Deficient lipid absorption is also not a consequence of an altered transport of lipids from the enterocytes to the lacteal vessels, which would result in an accumulation of chylomicrons and Oil red O staining in the enterocytes (35
). The impaired lipid absorption might in part be due to the lack of GLP-2. Hsieh and coworkers (36
) have recently shown that GLP-2 increases intestinal lipid absorption. However, the absorption of lipids, cholesterol and lipid soluble vitamins mostly depends on the function of bile acids (37
). In fact, bile acids owing to their amphipathic nature are essential for the solubilization of dietary lipids and their subsequent absorption in the digestive tract (37
). In addition, there is increasing evidence that bile acids also affect energy and glucose homeostasis (39
). Considering the reduced body fat, the impaired lipid absorption and glucose homeostasis seen in our Ngn3Δint
mice suggest that one or several enteroendocrine hormones may partially regulate bile acid homeostasis. A lack of enteroendocrine hormones could therefore lead to a deregulated bile acid homeostasis. In addition, the accelerated food transit and the reduced intestinal absorptive surface area of Ngn3Δint
mice might contribute to their impaired weight gain. Taken together, our data suggest that the early lethality and the impaired weight gain of surviving Ngn3Δint
mice are mostly due to malabsorption rather than malnutrition and clearly show that enteroendocrine cells and their secreted hormones are necessary to sustain life. Importantly, recently several patients with mutations in Ngn3
have been identified, suffering from birth onwards from unremitting diarrhea and the profound malabsorption of all nutrients, except water (17
), and the pathologic term given for the first case was enteroendocrine cell dysgenesis. Immunohistochemical analyses of the small intestine revealed few or a complete absence of enteroendocrine cells but normal numbers of Paneth, goblet, and absorptive cells, reinforcing the importance of enteroendocrine cells and/or hormones for nutrient absorption. Importantly and in contrast to our mice, which have only an intestinal deletion of Ngn3
, Ngn3 function in these patients is affected globally, and besides impaired enteroendocrine cell function, they also develop diabetes by the age of 8 years. However, it is not known whether these patients, by the age of 8 years, still have some remaining endocrine cells left in the pancreas, like in the intestines of patients 2 and 3 (18
), or whether they are completely lost. It is thus not clear whether the pathophysiology of patients with point mutations in Ngn3
results solely from enteroendocrine cell dysgenesis or also from the simultaneous altered Ngn3 function in the pancreas and/or hypothalamus. The malabsorption of all nutrients in the human patients versus specific lipid malabsorption in our mutant mice supports the latter hypothesis.
Taken together, from our data 2 important conclusions can be drawn. First, our results clearly show the importance of enteroendocrine cells/hormones in the regulation of energy homeostasis and that their loss is life threatening. They also point to the importance of enteroendocrine cells/hormones for lipid absorption by enterocytes. Second, enteroendocrine progenitor cells and/or one or several of the hormones secreted by differentiated enteroendocrine cells contribute, directly or indirectly, to the mechanisms regulating the homeostasis of the intestinal crypt compartment. The data presented here might also help to shed some further light on the pathophysiology of Ngn3 deficiency in humans and the contribution of enteroendocrine cell dysgenesis in malabsorption.