Nkx2.2 regulates cell fate choice in the enteroendocrine cell lineages of the intestine

Mice

Mice were maintained on a Swiss black (Taconic) background and were housed in the University of Colorado Health Sciences Center animal resource center. The mice were fed standard rodent chow and all animal procedures were performed according to UCHSC Institutional Review Board approved protocols.

Immunohistochemistry and in situ hybridization

Tissue was fixed overnight in 4% paraformaldehyde, cryopreserved and cryosectioned into 10 μm sections. At e18.5, n > 6 mutant embryos for each antibody and RNA in situ probe assayed; at e15.5, n >5 for each RNA in situ probe assayed. In each case, corresponding numbers of wild type littermates were simultaneously analyzed. Immunofluorescence was performed on cryo-preserved tissue with the primary antibodies listed in Table 1. Secondary antibodies include: Alexa-fluor 488, 594, or 305 (Molecular Probes, Eugene, OR; 1:800).

RNA in situ hybridization was performed as previously described on e15.5 and e18.5 tissue (Prado et al., 2004) using antisense riboprobes transcribed from linearized plasmids. Riboprobes for all genes except Nkx2.2 were generated from full-length cDNAs, which were PCR-cloned from e17.5 intestinal RNA. Full-length Nkx2.2 was cloned from a brain cDNA library (Porteus et al., 1992). RNA in situs were performed on tissue from Nkx2.2 null embryos and a corresponding number of wild type littermates: villin (n=5), IFABP (n=4), Cryptdin1 (n=9), Cryptdin 6 (n=6), Cryptdin7 (n=10), Hes1 (n=3), Cdx1 (n=3), Cdx2 (n=3), Tcf4 (n=3), Math1 (n=4), Ngn3 (n=4), NeuroD1 (n=7), Pax6 (n=5) and ghrelin (n=8). Nkx2.2 RNA in situs were performed on e12.5, e13.5, e15.5 and e18.5 tissue. For combined Nkx2.2 RNA in situ and anti-glucagon immunohistochemical analyses, the RNA in situs were performed to completion followed by immunohistochemistry using rabbit anti-glucagon antibody and biotinylated anti-rabbit secondary antibodies (Jackson Immunoresearch). The biotinylated secondary antibodies were visualized using the Vectastain ABC kit (Vector labs). Images were obtained with a Leica DM5000 microscope and a Qimaging Evolution MP Color camera or a Nikon Eclipse 80i microscope and a Qimaging Retiga EXi-Fast 1394 camera, and were processed using ImagePro software from Media Cybernetics.

Cell counting

The total number of chromograninA, CCK, ghrelin, GIP, glucagon, PYY, secretin, serotonin, PH3 and alcian blue positive cells were individually counted and the total intestinal area was measured using standard morphometric analysis with ImagePro software on every 10th section (10 μm each, 10 sections per embryo) for wild type (ChromograninA, CCK, GIP, glucagon, PH3 and alcian blue: n=3; ghrelin, PYY, secretin, serotonin: n=7) and Nkx2.2-/- e18.5 embryos (ChromograninA, CCK, GIP, glucagon, PH3 and alcian blue: n=3; ghrelin, PYY, secretin, serotonin: n=8). The number of immunoreactive cells per cross section were counted and normalized to the total intestinal area used in the measurements. The results obtained from each embryo were averaged and statistically analyzed.

Quantitative real-time PCR

Total RNA was extracted from e18.5 wild type and Nkx2.2-/- intestine tissue (dissected between the pyloric sphincter on the anterior end to the caecum, posteriorly) and prepared using the Qiagen RNeasy Micro kit. cDNAs were prepared with random hexamer primers and Superscript III (Invitrogen). Real-time PCR was performed on the ABI 7000 using Taqman probes (ABI Assays on Demand) or individually designed Taqman probes and primers. Each gene was normalized to Gapdh. See Table 2 for the list of primers and Table 3 for the numbers of samples tested for each probe. Samples were quantified with ABI prism software.

Nkx2.2 is expressed in scattered cell populations throughout the intestine during embryonic development

To determine when and where Nkx2.2 is expressed in the developing mouse intestine, we performed RNA in situ analysis to assess Nkx2.2 mRNA expression patterns throughout embryogenesis. At e12.5, Nkx2.2 is expressed robustly in the dorsal pancreas and in the emerging ventral pancreas, but cannot be detected in the developing intestine (Figure 1a). By e13.5, a small number of Nkx2.2-producing cells can be detected in a few cells of the proximal forestomach (Figure 1b), but cannot be detected in the developing proximal intestine (Figure 1c and data not shown). Between e14 and e15, cytodifferentiation of the intestinal epithelium begins, and the four epithelial cell types (goblet, enteroendocrine, enterocyte, and paneth) develop shortly after. At this critical e15.5 time point, Nkx2.2 is first detected in scattered cells throughout the duodenum and small intestine, and expression is maintained through birth (Figure 1d and e). At perinatal stages, the position, spatial distribution and shape of Nkx2.2 positive cells in the intestinal epithelium is similar to that of the enteroendocrine population. We confirmed that Nkx2.2 was expressed in a subpopulation of enteroendocrine cells by performing RNA in situ hybridization for Nkx2.2 combined with anti-glucagon immunohistochemical staining. As shown in Figure 2, Nkx2.2 is expressed in glucagon-positive cells of the intestine. Because it can be difficult to visualize the co-stained cells, representative coexpression of Nkx2.2 mRNA and glucagon protein in the pancreas (where Nkx2.2 is known to be co-expressed with glucagon (Sussel et al., 1998)) is also shown in Figure 2 e and f. Consistent with Nkx2.2 being expressed in the enteroendocrine cells, analysis of a transgenic line of zebrafish, Tg(nkx2.2a:mEGFP) demonstrated co-expression of nkx2.2a with the endocrine hormones, glucagon and somatostatin in the intestinal epithelium (Ng et al., 2005).

Nkx2.2 is specifically required for appropriate cell type specification of the enteroendocrine lineages

With the discovery that Nkx2.2 is expressed in specific cells of the developing intestine, we wanted to determine whether Nkx2.2 plays a role in intestinal epithelial development and cell type differentiation, similar to its role in the pancreatic islet. For this study, we analyzed mice that had been deleted for Nkx2.2 (Sussel et al., 1998). Because the Nkx2.2 null mice die shortly after birth with pancreatic islet and neural defects (Briscoe et al., 1999; Sussel et al., 1998), our analysis of the intestine was restricted to the late embryonic stages, shortly after differentiation of the intestinal epithelium is initiated. The Nkx2.2 null mice appear to have no gross alterations in their intestinal morphology when assessed by hemotoxylin and eosin (H&E) staining at any stage during embryonic development (Figure 5a and b). Since Nkx2.2 has a known role in regulating endocrine cell fate decisions in the pancreatic islet, the apparent expression of Nkx2.2 in intestinal enteroendocrine cells led us to analyze the enteroendocrine lineages of the gut epithelium in the Nkx2.2 knockout mice. Immunofluorescence analysis of several enteroendocrine hormones that are normally present in the proximal small intestine demonstrated that many of these cell populations were missing or reduced in the Nkx2.2 null mice (Figure 3 and data not shown). Quantification of serotonin, CCK, GIP and gastrin confirmed that there was a significant reduction in these hormone-producing endocrine populations in the absence of Nkx2.2 (Table 4). We confirmed that these and additional hormones were also down-regulated at the mRNA level by performing real-time PCR analysis on mRNA collected from the small intestine of wild type and Nkx2.2 null animals. These studies showed that the expression of CCK, gastrin, glucagon, GIP, neurotensin, and somatostatin is significantly lower in the Nkx2.2 null mice than in their wild type littermates. Interestingly, substance P and secretin expression are only moderately reduced. The numbers of PYY-producing endocrine cells of the distal small intestine were also unchanged in the Nkx2.2 null mice (Table 4) and PYY and NPY were expressed at similar levels in the wild type and Nkx2.2 null mice, suggesting that these more distal cell populations are unaffected by the loss of Nkx2.2.

Although there is a reduction in many of the hormone-producing cells in the Nkx2.2 null mice, immunohistochemical analysis of the pan-endocrine markers chromograninA and synaptophysin suggested that the total numbers of enteroendocrine cells is unchanged (Figure 3 and data not shown). Quantification of the chromograninA immunoreactive cells confirmed that there were similar numbers of enteroendocrine cells in the wild type and Nkx2.2 null small intestine (Table 4). ChromograninA and synaptophysin mRNA expression was also unchanged in the Nkx2.2 null mice (Figure 4). In contrast, however, the number of cells expressing the ghrelin hormone is significantly increased throughout the proximal small intestine in the Nkx2.2 null embryos (Figure 3 and Table 4) and ghrelin mRNA expression is correspondingly elevated (Figure 4). The ghrelin-producing cells in the Nkx2.2 null mice appear to retain the characteristics of endocrine cells as they continue to express the endocrine marker synaptophysin (Figure 3 inset and Supplemental figure 1). This analysis suggests that, similar to its role in the pancreatic islet, Nkx2.2 functions to regulate cell fate specification of the enteroendocrine populations in the intestine.

The loss of several hormone-producing populations and the simultaneous increase in ghrelin-producing cells suggests that the ghrelin cell population may be specified at the expense of the other endocrine cell lineages. However, it remains possible that the loss of Nkx2.2 activity increases the proliferation of the existing ghrelin cell population, independently of losing the other hormone-producing cells. To test this possibility, we analyzed the expression of phospho-Histone H3 to assess the mitotic status of the ghrelin-positive cells. Similar to our results in the pancreatic islet, we could not detect any proliferating ghrelin-positive cells in the intestine of wild-type or Nkx2.2 null embryos (data not shown) and the total number of proliferating cells in the intestine is not significantly altered when we quantified the total number of phospo-histone H3 positive cells (wild type: 216.2±18.6 (n=3); Nkx2.2: 169±11.1 (n=3). Therefore, we hypothesize that in the absence of Nkx2.2 several of the endocrine populations appear to be replaced by ghrelin-producing endocrine cells.

Nkx2.2 does not affect the differentiation of the other intestinal cell populations

The loss of several hormone-producing enteroendocrine populations and corresponding upregulation of ghrelin cells in the absence of Nkx2.2 activity confirms the importance of Nkx2.2 in intestinal enteroendocrine cell specification. However, because we could not perform co-staining analysis of Nkx2.2 with the other intestinal epithelial cell populations, we could not rule out that Nkx2.2 was either directly or indirectly involved in the development and differentiation of other intestinal cell populations. To determine whether Nkx2.2 affected the development of these remaining intestinal cell types we assessed the expression of characteristic cell-type specific markers during intestinal development (Figure 4). Villin is one of the first genes to be expressed in the intestine and is located globally throughout the intestinal epithelium (Braunstein et al., 2002). We do not detect any change in villin expression in the Nkx2.2 knockout mice suggesting that the early intestinal patterning events are not perturbed (Figure 5c and d). We were also unable to detect any defects in the development of the mature enterocyte, goblet, or paneth cell populations by assessing the expression of markers characteristic of each cell type (Figure 5). For this analysis we assessed the expression of the enterocyte marker IFABP, an intracellular protein abundantly expressed in the differentiated enterocyte population, which represents the largest cell population in the intestine (Figure 5g and h,); Alcian blue labeling, which characteristically stains goblet cells (Figure 5e and f), and quantitative PCR analysis of two goblet cell specific genes, TFF3 and mucin2 (n=5; Figure 5k); and cryptdin 1, 6 and 7 as markers of paneth cell differentiation (Figure 5i and j; data not shown). According to this marker analysis, there were no discernable differences in the non-endocrine cell populations between the developing wild type and Nkx2.2 null intestine. Furthermore, quantification of the alcian blue stained cells revealed that total goblet cell numbers were unchanged in the Nkx2.2 null mice (wild type: 182.9±5.6 (n=3); Nkx2.2-/-: 184.1±11.1 (n=3) and expression of Klf4, a zinc finger transcription factor that is essential for the terminal differentiation of goblet cells (Katz et al., 2002) is also maintained. These studies indicate that Nkx2.2 is predominantly involved in the differentiation of the enteroendocrine intestinal cells and, similar to its role in the pancreatic islet, plays an important role in cell specification decisions within this population.

Positioning Nkx2.2 in the intestinal regulatory pathway

The increase in ghrelin cells at the expense of the several intestinal endocrine populations has not been reported in other mouse knockout models known to affect intestinal enteroendocrine cell differentiation. To better understand the molecular basis of the defects observed in the Nkx2.2 knockout and to position Nkx2.2 in the hierarchy of transcription factors regulating intestinal cell fates, we assessed expression of many of the regulatory factors known to be involved in intestinal cell differentiation. Several different transcription factors have been shown to regulate the cell type differentiation process of all intestinal cell populations or the enteroendocrine lineages, specifically. We used RNA in situ analysis and quantitative real-time PCR to determine which, if any, of the transcription factors had altered expression in the Nkx2.2 knockout mice (Figure 6 a-l, Supplemental figure 2). Consistent with the Nkx2.2 mutant phenotype, which appears to be restricted to the enteroendocrine lineages, expression of transcription factors such as Cdx1 and Cdx2, which are involved in early gut patterning are not affected by the loss of Nkx2.2 (Figure 6a,b, Supplemental figure 2). In addition, expression of the stem cell factor Tcf4, the progenitor cell regulator Hes1, and the early cell lineage regulator Math1 are also not affected in the mutant mice (Figure 6c-f, Supplemental figure 2). We also assessed the expression of the Gata4 and Gata6 transcription factors, which have recently been shown to be differentially expressed in the enterocyte and enteroendocrine populations of the gastrointestinal tract and the endocrine cells of the pancreas (Bosse et al., 2006; Decker et al., 2006; Dusing and Wiginton, 2005). Both by RNA in situ analysis and quantitative PCR, we could not detect an alteration in Gata4 or Gata6 expression in the Nkx2.2 null intestine (Supplemental figure 2 and data not shown).

Many additional transcription factors are thought to be responsible for the correct specification and/or development of several enteroendocrine lineages. Ngn3 and NeuroD1/Beta2 (hereafter referred to as NeuroD1) are basic helix-loop-helix transcription factors that are also involved in the endocrine cell differentiation process. Ngn3 is required for the differentiation of all endocrine cells in the gastrointestinal tract and pancreatic islet; Ngn3 null mice fail to develop these populations. Ngn3 expression is not altered in the Nkx2.2 knockout mice (Figure 6g-h, Supplemental figure 2), which is consistent with the observation that there are normal numbers of enteroendocrine cells in the Nkx2.2 null embryos. NeuroD1 has been shown to function downstream of Ngn3 and appears to regulate the formation of CCK and secretin cells (Huang et al., 2000; Jenny et al., 2002; Naya et al., 1997). NeuroD1 mRNA expression is modestly, but consistently downregulated in the intestine of Nkx2.2 null mice (Supplemental figure 2), however RNA in situ analysis does not reflect a two-fold reduction in the area of intestine staining for NeuroD1 (Figure 6i-j). The discrepancy between the PCR results and the RNA in situ analysis may be due to the increased sensitivity of real time PCR assay or may reflect a decrease in the amount of NeuroD1 expressed in each cell. The non-quantitative nature of RNA in situ assays makes this issue difficult to resolve, however a reduction in NeuroD1 expression the Nkx2.2 null mice is consistent with the observed reduction in the numbers of CCK and secretin cell populations.

Islet1 (Isl1) and Pdx1 play important regulatory roles in the differentiation of islet and intestinal endocrine populations. It has been suggested that the LIM homeodomain protein, Isl1 is involved in the early differentiation of gastrin and somatostatin cells, however the early lethality associated with the Isl1 knockout mice precludes confirmation of an Isl1 intestinal function (Ahlgren et al., 1997; Larsson, 2000). A deletion of the homeodomain protein, Pdx1, however, results in a loss or reduction of several endocrine cell populations in the duodenum (Larsson et al., 1996). Although the Nkx2.2 knockout affects many of the same endocrine cell populations, the expression of both these genes is unaffected by the loss of Nkx2.2 (Supplemental figure 2).

Pax4 and Pax6 are also critical endocrine regulatory factors that, similar to Nkx2.2, have been implicated in both pancreatic and intestinal cell fate determination (Larsson et al., 1998). In the intestine, Pax4 is essential for the formation of serotonin, secretin, CCK, GIP, and PYY cells and Pax6 is required for GIP cell differentiation. Interestingly, although it appears that Pax4 and Nkx2.2 regulate an overlapping set of intestinal cell types, Pax4 expression is not affected in the Nkx2.2 knockout mice (Supplemental figure 2). This would suggest that Pax4 functions upstream of Nkx2.2 or in a parallel pathway for the formation of several different enteroendocrine cell types. This result is consistent with what we observe with corresponding Nkx2.2 and Pax4 functions in the pancreatic islet. Pax6, on the other hand, is down-regulated in the absence of Nkx2.2 (Figure 6k-p), suggesting that it functions downstream of Nkx2.2 to specify GIP cells. Again, this parallels the corresponding functions of Nkx2.2 and Pax6 in the islet.