During embryogenesis, the mammalian gastrointestinal epithelium develops from definitive endoderm that starts forming villi and clusters of proliferating cells at embryonic day 15 (Crosnier et al., 2006). After birth, proliferating cells are located in the crypt compartment. Within the proliferating cells are stem cells that generate all the gastrointestinal epithelial lineages, including enteroendocrine cells (Lee and Kaestner, 2004; Noah et al., 2011). The enteroendocrine population is subdivided into at least 15 mature cell types that, in concert with the enteric nervous system, regulate many aspects of gastrointestinal activity, including glucose metabolism, the delivery of bile and pancreatic juice, and gut epithelial renewal (Skipper and Lewis, 2000; Furness, 2012).
Several gene expression and transgenic knockout studies have identified some of the signaling pathways and regulatory factors that are involved in the development and differentiation of the enteroendocrine population (Yang et al., 2001; Lee and Kaestner, 2004; Bjerknes and Cheng, 2006; Desai et al., 2008). Expression of the transcription factor (TF) Math1 by progenitor cells initiates the secretory cell differentiation program (Yang et al., 2001), while the expression of Ngn3 activates the endocrine differentiation (Gradwohl et al., 2000; Jenny et al., 2002; Lee et al., 2002; Bjerknes and Cheng, 2006). The differentiation of secretin and CCK lineages in Ngn3-positive precursors is activated by BETA2/NeuroD (Naya et al., 1997) while Nkx2.2 is required for the appearance of the CCK, GIP, somatostatin, GLP-1, and gastrin cells (Desai et al., 2008). The TF Pax6 activates the glucagon gene expression (St-Onge et al., 1997; Larsson et al., 1998; Hill et al., 1999; Katz et al., 2009), while TF's Arx and Pdx1 are required for L and K cell differentiation, respectively (Offield et al., 1996; Fujita et al., 2008; Beucher et al., 2012; Du et al., 2012). Although these factors are important for enteroendocrine lineage development, it is unclear when intestinal progenitor cells become committed to the different cell lineages. Also unknown is the identity of the signals regulating the differentiation of the progenitor cells into mature endocrine cells.
Hormones secreted by the entero-endocrine L cells of the small and large intestine play a crucial role in glucose homeostasis. L cells, together with pancreatic alpha cells and hypothalamic neurons, synthesize proglucagon, a product of the glucagon gene (Holst, 1997; Steiner, 1998). Proglucagon is cleaved into glucagon by proprotein convertase subtilisin/kexin-type 2 (PCSK2), termed here PC2, in alpha cells (Steiner, 1998). In brain and in L cells, proglucagon is processed by prohormone convertase 3/1 (PCSK3, termed here PC3/1), generating glucagon-like peptide-1 (GLP-1), glucagon-like peptide-2 (GLP-2), and other small peptides (Holst, 1997; Steiner, 1998). The small intestine also contains a mixed type of enteroendocrine cells, called LK cells, that produce both GLP-1 and GIP (glucose-dependent insulinotropic peptide) (Mortensen et al., 2003). GIP and GLP-1 are termed incretins because they stimulate insulin secretion from pancreatic beta-cells following nutrient ingestion (Baggio and Drucker, 2007). While the primary site of endogenous GIP release is the endocrine K cell of the duodenum and proximal jejunum, several reports indicate the presence of LK cells in adult mouse ileum (Mortensen et al., 2003; Fujita et al., 2008; Grigoryan et al., 2012).
To date, many questions remain unanswered regarding the identity of the signals regulating the number of L, K, and LK cells. Analysis of changes in L and LK cell populations along the small intestine in normal mice (Habib et al., 2012), in ileum of mutant mice lacking the glucagon receptor (Gcgr−/−) (Grigoryan et al., 2012), and in small intestine of rats following bariatric surgery (Patriti et al., 2007; Kohli et al., 2011; Speck et al., 2011; Woods et al., 2011) suggests an important effect of nutrients and other extrinsic cues on LK, L, and K cell number. Since in embryos, the intestine is not exposed to external cues, we speculated that the proportion of these three cell types could differ from that in adult small intestine. Therefore, the first goal of this study was to test this possibility by determining the proportion of L, K, and LK cells in ileum of mouse embryos. The second goal of this study was to ascertain whether the specificity of GLP-1 production by L cells of adult mice is established during development or if L cells of embryos are able to synthesize both glucagon and GLP-1.
L Cells First Appear During Midgestation
To determine the time of appearance of L cells during development, we examined embryos from embryonic day (e)-12 to e-18. Occasional cells containing GLP-1 (one GLP-1+ cell every few sections of intestine) were first seen at e-15 (Fig. 1A) and became more abundant at e-17 (Fig. 1B) (2.8 ± 0.4 GLP-1+ cells/100 epithelial cells. At least 2,000 epithelial cells/embryo were scored in the ileum; N=4). At e-17, L cells were detected both in the “crypt” (Fig. 1B) and villus (not shown) compartments. While villi begin to form at e-15, proliferative pockets of the small intestine reshape into mature crypts only in the first week after birth (Crosnier et al., 2006). Therefore, in this study, the term “crypt” is used to indicate the location of transit-amplifying (TA) region containing progenitor cells (Fig. 2E,F).
The Majority of L Cells of Embryos Coexpress GIP
At e-17, the mucosa of ileum contains GLP-1+ GIP + cells (Fig. 1C–E). The number of LK cells in ileum of embryos (Fig. 1F–G) is threefold higher than that in adults (Mortensen et al., 2003; Grigoryan et al., 2012). In embryos, the percentage of GLP-1+GIP+ cells is similar along the crypt-villus axis.
It has been proposed that Pdx-1 and Pax6 are required for GIP expression (Offield et al., 1996; Jepeal et al., 2005; Fujita et al., 2008) while Pax6 has been reported to activate the expression of the glucagon gene (Hill et al., 1999). At e-17, ileum of embryos contain GIP+Pdx-1+ (Fig. 2A), GIP+Pax6+ (Fig. 2B), GLP-1+ Pax6+ (Fig. 2C), and GLP-1+Pdx-1+ cells (Fig. 2D). Approximately 70% of GIP+ cells express either Pdx-1 or Pax6 (Table 1). In addition, although 80% GLP-1+ cells were Pax6+, only 44% expressed Pdx-1 (Table 1), raising the possibility that not all LK cells express both Pax6 and Pdx-1.
Table 1. Expression of Pdx-1 and Pax6 by L and K Cells of e-17 Ileuma
The data are expressed as a mean value ± SEM. At least 75 L and 75 K cells were scored/embryo/antibody combination. N = 3.
Most L Cells of Embryos Are Quiescent
To ascertain whether embryonic L cells, unlike the adult counterparts (Cheng and Leblond, 1974; Grigoryan et al., 2012), are able to proliferate, we examined sections of e-17 ileum immunostained for visualization of the proliferation marker Ki67 and GLP-1. While the large majority of L cells were Ki67 negative both within and outside the TA region (Fig. 2E,F), 4% of GLP-1+ cells present in the crypt were GLP-1+Ki67+ (Fig. 2G,H). This observation indicates that, as in normoglycemic adults, most embryonic L cells are quiescent and do not proliferate.
L Cells of Embryos Co-Express Glucagon and GLP-1
Since a subpopulation of alpha cells of pancreas synthesizes Glu and GLP-1 during development (Lee et al., 1999; Wilson et al., 2002), we sought to determine whether embryonic L cells are also able to express both peptides. The selection of antibody was important because polyclonal antibodies to glucagon also bind to proglucagon, as documented by the labeling of alpha cells on PC2 mutant mice, which express proglucagon but not glucagon (Furuta et al., 1997; Vincent et al., 2003). Therefore, we used a mouse monoclonal antibody (mAb) that immunostained pancreatic alpha cells of CD-1 mice (Fig. 3A) but not alpha cells of PC2 knockout mice (Fig. 3B) or intestinal L cells of adult CD-1 mice (Fig. 3C, J-L), indicating that this antibody was specific for glucagon.
Examination of the ileum of e-17 CD-1 embryos revealed the presence of L cells co-expressing both Glu and GLP-1 (Fig. 3D–F). Morphometric analysis indicated that 21% of GLP-1+ cells co-expressed Glu (Table 2). The ileum of e-17 embryos contained cells co-expressing Glu and PC3/1 (Fig. 3G–I, Table 2) and Glu+ PC2+ cells (Fig. 4A–C, Table 2). While all Glu+ cells have PC3/1, only 12% of the Glu+ cells expressed PC2 (Table 2). Most Glu+ cells have very low levels of PC2 staining (Fig. 4D–F) when compared to the few Glu+ cells with strong immunolabeling (Fig. 4A–C) or with the level of staining in adult pancreatic islets (Fig. 4G). In contrast to embryos, ileum of pups at postnatal day 4 did not contain Glu+ cells (not shown). Similarly, ileal L cells of adults did not express PC2 (Fig. 4H–J, Table 2) or, as described above, Glu (Fig. 3C, J–L; Table 2).
Table 2. L Cells of Embryonic, But Not Adult Ileum, Express Glucagona
The data are expressed as mean value ± SEM. At least 75 GLP-1+ and 20 Glu cells were scored/embryo and at least 75 GLP-1 cells were scored/adult/antibody combination. N=3.
While the enteroendocrine cells comprise the largest endocrine organ of the body, the identity of the signals regulating the differentiation and the number of each endocrine cell type in embryonic and adult intestinal tract is unknown. In the present study, we sough to determine whether the percentage of L and LK cells is similar to that previously reported for adult mice (Grigoryan et al., 2012) and to ascertain whether L cells of embryos express both GLP-1 and glucagon in embryos and/or adults.
In embryos, we found that the ratio of LK cells to the total number of L cells scored is threefold higher than in adult control mice (Mortensen et al., 2003; Grigoryan et al., 2012) and that this increase is not due to LK cell proliferation. Presumably, the TA region contains precursors for K, L, and LK cells and the environment in the fetal intestine specifically activates LK cell precursors to generate more LK cells. Alternatively, it is possible that LK cells of embryos are in an intermediate stage of the L and K cell differentiation pathway and, in embryos, these “mixed” cells fail to generate the appropriate number of differentiated L and K cells. Changes in the percentages of L, K, and LK cells may be determined by the apico-caudal differentiation of the lower intestine during postnatal growth. During this process, more anterior signals would promote K cell differentiation while more posterior signals would favor the appearance of L cells.
A high number of LK cells also differentiate in the small intestine of mice and of humans with metabolic disorders (Theodorakis et al., 2006; Grigoryan et al., 2012) and after anatomical re-arrangement of the small intestine (Speck et al., 2011) suggesting the presence, in these models, of signals that activate LK cell growth. Perhaps similar mechanisms lead to the development of LK cell hyperplasia in embryos. In adult Gcgr−/− mice, the increase in LK cell number was not correlated with an augmentation in glucagon or GIP mRNA, or with the circulating levels of GIP (Grigoryan et al., 2012) suggesting that LK cells of adults are not a functionally active cell type. The functional activity of LK cells in embryos, however, remains to be determined.
Our studies also revealed that the embryonic ileum contains cells that synthesize glucagon. However, only 12% of the Glu cells are PC2+. Since most Glu+ cells have very low levels of PC2 staining, it is possible that the convertase is expressed by all Glu cells but that its concentration is below the threshold of detection by the immunohistochemical technique. In contrast to embryos, the ileum of postnatal and adult mice does not contain glucagon cells, indicating that its expression is inhibited soon after birth.
Since the intestine of mice with ablation of the glucagon gene (Gcg−/−) (Hayashi et al., 2009), PC2 (Furuta et al., 2001; Grigoryan et al., 2008), Gcgr (Grigoryan et al., 2012), or the GLP-1 receptor (Hansotia and Drucker, 2005) appeared to be morphologically normal, the presence of glucagon and GLP-1 in the embryonic small intestine is intriguing. One possibility is that products of the glucagon gene are involved in a regulatory loop that controls the number of cells expressing the gene. In adults, mutant mice lacking glucagon display alpha but not L cell hyperplasia (Furuta et al., 1997; Grigoryan et al., 2008; Hayashi et al., 2009) and restoration of glucagon to PC2 KO mice normalizes alpha cell number (Webb et al., 2002). The number of L cells is positively correlated with plasma level of GLP-1, which was high in Gcgr−/− mice (Gelling et al., 2003) but was normal in PC2−/− mice (Furuta et al., 1997). These observations suggest an important role for Glu and GLP-1 in the regulation of alpha and L cell number, respectively, in adults. Conceivably, glucagon and/or the glucagon-like peptides may play a similar role during development.
In conclusion, we found that the embryonic mouse ileum contains more LK cells than in adults. In addition, we also determined that a subset of ileal L cells of embryos coexpress glucagon, a cell type that is not present in normal adult rodents. However, enteric glucagon cells may reappear in adults during alterations in metabolic homeostasis (Knop, 2009). Therefore, the understanding of the signals that regulate the differentiation and number of cells expressing GIP and/or the glucagon family of peptides will provide fundamental insight into the maintenance of glucose homeostasis.
Pregnant CD-1 mice were purchased from Charles River (Wilmington, MA). The appearance of the vaginal plug was considered day 1 of gestation (e-1). PC2 mice were kindly provided by Dr DF Steiner (University of Chicago) and were genotyped as described by Furuta et al., 1997. Animals were fed ad libitum with free access to water and maintained in a murine hepatitis virus-free barrier facility on a 12-hr light-12-hr dark cycle. All animal protocols were approved by the Institutional Animal Care and Use Committee.
Preparation of Tissue Sections
At e-17, the small intestine was dissected into three parts from which the ileum (the distal 1/3 before the caecum) was isolated. Tissues were fixed in 4% paraformaldehyde (PF) in Phosphate Buffer Saline (PBS), pH 7.4, cryopreserved in a 30% sucrose, embedded in Shandon M1 matrix (Thermo Scientific, Pittsburgh, PA) and 20-μm frozen sections obtained using a cryostat microtome (Leica Jung 3050S). Embryos were examined at e-12, 15, and 17.
Sections were processed for immunostaining as previously described (Vuguin et al., 2006). Primary antisera raised in different hosts were used to localize two antibodies in the same tissue section. The bound antibodies were visualized with corresponding anti-“primary” host “X” secondary antibody linked to molecules that fluoresce at different wavelengths, respectively. Source and dilutions of antibodies (Ab): Rabbit monoclonal Ab to Ki67,1:400, Novocastra (Vector Labs, Burlingame, CA); Rabbit Ab to GLP-1, 1:5,000, Calbiochem (San Diego, CA); Rabbit Ab to Pdx-1, 1:1,500, generously provided by Dr. C.W.E. Wright (Vanderbilt University); Antiguinea Pig antibody to Pdx-1, 1:1,000, Abcam (Cambridge MA); Rabbit Ab to GIP, 1:2,000, Peninsula Labs (San Carlos, CA); Mouse monoclonal Ab to Pax6 1:150, Developmental Studies Hybridoma Bank, U. of Iowa); and a rabbit polyclonal antibody to Pax6 from Chemicon (Temecula, CA) (1:500). Antibodies to PC2 and PC3/1 are a generous gift of Dr. D.Steiner (U.of Chicago) and were used at a 1:1,000 dilution. Mouse Monoclonal Ab to Glucagon, 1:6,000, Sigma-Aldrich (St. Louis, MO).
Alexa Fluor 488 anti-mouse, anti-rat, and anti-rabbit IgG, Alexa Fluor 594 anti-guinea pig, anti-rabbit, and anti-mouse IgG were purchased from Molecular Probes/Invitrogen (Carlsbad, CA). Controls included replacing the primary or secondary antibody with pre-immune serum. All secondary IgG's were used at 1:200 dilution.
Proliferating cells were identified by immunostaining with antiserum to Ki67, a marker of proliferating cells.
Confocal images were obtained using a Radiance 2000 confocal microscope (BioRad, Hercules, CA) attached to a Zeiss Axioskop microscope (Carl Zeiss Inc., Thornwood, NY). Images at 1,260×1,260 pixels were obtained and processed using Adobe Photoshop 6.0 (Adobe Systems, Mountain View, CA). At least 3 embryos were examined/stage/antibody combination.
At least 15 non-consecutive sections were chosen to determine numbers of immunostained cells per villus or crypt. Sample number is indicated in individual experiments. To measure the density of L cells, the number of stained cells was expressed as a percentage of the number of epithelial cells.
Values indicate the mean ± SEM. For comparison between two groups, unpaired two-tailed Student's t-test was used. A P value < 0.05 was considered significant.