Xenopus Insm1 is essential for gastrointestinal and pancreatic endocrine cell development

Abstract

In mammals it has been well established that gastrointestinal and pancreatic endocrine cells are specified by a cascade of different transcription factors, but whether these same pathways (or linear relationships) operate in Xenopus is currently unknown. We recently identified the endocrine-specific zinc finger transcription factor insulinoma associated protein 1 (insm1) as a dorsal-enriched gene. We found that insm1 is expressed in the dorsal pancreas as early as NF28, making it one of the earliest markers to be localized to the dorsal pancreas. Through morpholino-mediated knockdown, we demonstrate that insm1 is essential for proper specification of both gastrointestinal and pancreatic endocrine cells. In addition, we place insm1 downstream of ngn3 and upstream of pax6 and neuroD in the endocrine cell transcription factor cascade. These are the first results showing that the endodermal endocrine cell development in Xenopus uses the same transcriptional cascade as in mammals.

Introduction

In Xenopus tadpoles, endocrine cells develop in the gastrointestinal tract and pancreas, as is seen in mammals. Somatostatin- and glucagon-expressing endocrine cells are found throughout the gastrointestinal tract and pancreas, while insulin-expressing cells are only found in the pancreas (Horb and Slack, 2002; Pearl et al., 2009). These three differentiation markers are expressed in a non-overlapping punctate fashion. The first pancreatic cell type to be specified is the β cell; expression of insulin is first detected at NF32 in the dorsal pancreas (Kelly and Melton, 2000). In contrast, pancreatic α and δ cells are not specified until NF44/45; development of these cells does however, occur earlier in the stomach and duodenum. Somatostatin and glucagon are first detected in scattered cells in the developing stomach/duodenum at NF40; subsequently their expression extends throughout the entire gastrointestinal tract.

Development of these endocrine cell types has been well studied in mammals and it has been found that specification occurs through regulated expression of various transcription factors, which are expressed in both the gastrointestinal tract and pancreas (Gittes, 2009). As the study of Xenopus pancreas and gastrointestinal tract development is relatively young, little is known about whether the same signaling cascades operate in the frog as well. Although many of these same transcription factors have been identified in Xenopus (Pearl et al., 2009), their function has only been examined in the context of neural development, where these genes are also expressed. In fact, to date there has not been any studies published in Xenopus examining whether these transcription factors function in the same manner in endocrine cell development.

One of the earliest acting endocrine specific transcription factors is Insm1 (insulinomas associated protein 1). Insm1 (also known as IA-1) is an intronless gene encoding a zinc-finger protein that functions as a transcriptional repressor that was originally isolated as being expressed exclusively in human insulinoma cells (Goto et al., 1992; Breslin et al., 2002). It is expressed in a variety of neuroendocrine tumors, including insulinoma, small cell lung carcinoma and medulloblastoma (Lan et al., 1993; Lan et al., 1994). Mouse insm1 is not detected in adult tissue and is only expressed from E10.5 until 2 weeks postnatal (Xie et al., 2002; Breslin et al., 2003). Knockout of insm1 in mice resulted in impaired differentiation of endocrine cells within the pancreas and gastrointestinal tract (Gierl et al., 2006). Although initial specification of β cells was normal in insm1−/− mutant mice, later development of ins+ cells at the secondary transition did not occur properly. In contrast, initial specification of α cells was impaired, whilethe number of glucagon+ cells was normal by E18.5. In the transcriptional cascade of endocrine cell development, insm1 was found to be a direct target of ngn3, acting upstream of other endocrine transcription factors (Mellitzer et al., 2006). Although insm1 has been identified in other model systems (Xenopus, medaka and zebrafish) (Lukowski et al., 2006; Candal et al., 2007; Parlier et al., 2008), its function in endocrine cell development in these other organisms has not been established.

In a search for dorsal and ventral pancreas specific genes, we identified insm1 as a dorsal-enriched gene product (Jarikji et al., 2009). Expression analysis revealed it to be expressed in a punctate fashion throughout the gastrointestinal tract and pancreas. In this paper, we describe the functional characterization of Xenopus insm1. We found that insm1 was required for development of both pancreatic and gastrointestinal endocrine cells, downstream of ngn3. This is the first study in Xenopus illustrating that the same signaling cascade that operates in mammalian endocrine cell development also functions in Xenopus.

Results and Discussion

Insm1 is essential for proper specification of pancreatic and gastrointestinal endocrine cells

To identify new dorsal and ventral pancreas specific genes we isolated individual dorsal and ventral pancreatic buds at NF38/39 and compared their gene expression profiles with the Affymetrix Genechip (Jarikji et al., 2009). One of the dorsal-enriched genes (2.58 fold) identified in this screen was the zinc finger transcription factor Insulinoma Associated Protein 1 (insm1). Since insm1 had not yet been identified in X. laevis we cloned the full-length cDNA by PCR based on partial X. laevis and X. tropicalis sequence information (see materials and methods); it is identical to the recently published Xenopus IA-1 sequence (Parlier et al., 2008).

The neural expression of Xenopus insm1 was recently examined in detail (Parlier et al., 2008). In that paper they also reported insm1 to be expressed in the dorsal endoderm at NF27, but its endodermal expression was not examined in detail. To characterize the developmental expression pattern of insm1 in the endoderm we examined its spatial and temporal expression profile from tail bud through tadpole stages (Fig. 1). In agreement with the previous report, insm1 expression was first detected in the endoderm at NF28 in a small domain of the dorsal endoderm (Fig. 1A). By NF34 expression of insm1 increased and was now found to extend laterally on either side of the dorsal endoderm (Fig. 1B). At NF38 the bilateral expression of insm1 had resolved into a central circular punctate pattern in the dorsal endoderm (Fig. 1C). Shortly thereafter at NF39 this punctate expression of insm1 expression is displaced to the left side of the tadpole (Fig. 1D). Immediately after fusion of the dorsal and ventral pancreatic buds at NF40, scattered expression of insm1 is found only in the dorsal pancreas (Fig. 1E). This localization of insm1 to the dorsal pancreas was in agreement with the results we obtained from the microarray. However, by NF44 insm1 expression was now found throughout the entire pancreas (Fig. 1F). In other parts of the endoderm, we also found scattered expression of insm1 throughout the entire length of the gastrointestinal tract at NF40 and NF42 (Fig. 1G,H). This early endodermal expression of insm1 begins before expression of endocrine differentiation markers and suggested that it may play an important role in the specification of early β cells as well as other intestinal endocrine cells.

Fig. 1.

Expression pattern of insm1 in endoderm. A–D: Expression of insm1 in the dorsal endoderm at NF28, NF34, NF38 and NF39, respectively. Notice expression begins as a small point in the dorsal endoderm (arrowhead) and then is found in bilateral domains at NF34 (two arrows); by NF38 expression is now present in the dorsal midline in scattered cells (outlined). Shortly thereafter expression is now found displaced to the left side of the tadpole (outlined). E,F: Isolated liver/pancreas tissue. Insm1 is detected at NF40 in a punctate fashion throughout the gastrointestinal tract and in the dorsal pancreas, similar to that seen in the early endoderm. (The dotted line demarcates dorsal and ventral portions of the fused pancreas.) At later stages expression found throughout the entire pancreas, but still in a punctate manner. G,H: Whole guts from NF40 and NF44 tadpole. The pancreas is outlined. Expression is found in a punctate fashion throughout the entire gastrointestinal tract. Dp, dorsal pancreas; vp, ventral pancreas.

To investigate the function of insm1 in early endocrine cell development, we designed two morpholinos, one to the translation start and one to the 5’UTR, to specifically inhibit insm1 translation and injected them into the vegetal hemisphere targeting the endoderm (see materials and methods). Both morpholinos were co-injected (20ng each) as they produced the most consistent phenotypes. A 5bp mismatch morpholino was used in control injections and specificity was determined by in vitro transcription and translation (Fig. 2O). The injected embryos were grown to tadpole stages and whole guts were isolated and the development of endocrine cells was examined using whole mount in situ hybridization. Knockdown of insm1 resulted in reduced expression of all endocrine differentiation markers. In Xenopus, β cells are the first endocrine cell type to be specified at NF32 in the dorsal endoderm (Pearl et al., 2009). In control tadpoles at NF35, insulin (ins) RNA expression is present in a small domain in the midline of the dorsal endoderm (Fig. 2A). However, in insm1 knockdown tadpoles this early ins expression was absent (Fig. 2B). Even at the later stages, NF40/42, there was no ins expression present in the pancreas (Fig. 2C,D). These results demonstrate that insm1 is required for proper specification of β cells at both early and late stages.

Fig. 2.

Insm1 is required for endocrine cell differentiation. A–D: Insulin expression at both early (NF35; 90%, n=15) and late stages (NF40/42; 79%, n=52) is almost completely abolished. E–H: glucagon expression. I–L: somatostatin expression. At early stages expression of both gcg and som is reduced in the stomach/duodenum of insm1 knockdown animals (100%; n=33 and n=12, respectively), but begins to reappear at later stages (100%; n=8 for both). The initial expression of these markers in the pancreas is also reduced at later stages (100% for gcg and 80% for som). M,N: Expression of the exocrine pancreas marker XPDIp is normal in insm1 morpholino injected tadpoles (100%, n=10). O: In vitro transcription and translation of insm1 either alone (insm1), in the presence of the 5bp mismatch morpholino (insm1+mMO) or in the presence of the morpholino (insm1+MO). Pictures presented in control are of uninjected embryos, but in all cases mismatch morpholino embryos (n≥8 for each marker) gave identical results as control, with normal expression of these markers. The pancreas is outlined.

We next examined whether insm1 was also required for the development of other endocrine cell types. Unlike insulin, glucagon (gcg) and somatostatin (som) are not expressed in the pancreas until NF44, while at earlier stages they are both expressed in gastrointestinal endocrine cells. To determine whether gastrointestinal and pancreatic expression of these markers was affected by loss of insm1, we examined the expression of these markers at both early and late stages. At early stages, we found reduced expression of gcg and som in the stomach/duodenum (Fig. 2E,F,I,J). However, at later stages we found both markers were re-expressed in the stomach/duodenum (Fig. 2G,H,K,L). Similar to the early reduction in the gastrointestinal tract, we found initial expression of som and gcg in the pancreas at NF45 to be reduced in insm1 knockdown tadpoles (Fig. 2G,H,K,L). In contrast, exocrine differentiation appeared normal, as judged by expression of XPDIp (Fig. 2M,N). These results are similar to that seen in the insm1 knockout mouse where there was reduced numbers of glucagon-expressing cells during early development, whereas at later stages the number of glucagon-expressing cells was normal (Gierl et al., 2006).

We next determined at which stage in the transcriptional cascade of endocrine cell differentiation insm1 acted by examining whether expression of other endocrine specific transcription factors was affected. Whereas neuroD and pax6 expression were severely reduced in insm1 knockdown endoderm (Fig. 3C–F), we found normal expression of ngn3 (Fig. 3A,B). These results demonstrate that insm1 acts upstream of neuroD and pax6, but downstream of ngn3, and is in agreement with the results in mammals (Gierl et al., 2006; Mellitzer et al., 2006).

Fig. 3.

Expression of endocrine specific transcription factors in insm1 knockdown whole guts. A,B: Ngn3 expression. Ngn3 is expressed in endocrine progenitors and its expression is normal in insm1 morpholino injected samples (70%, n=24). C,D: NeuroD expression. E,F: Pax6 expression. Expression of neuroD and pax6 is absent both in the pancreas and the gut (100%, n=16 for both). Pictures presented in control column are of uninjected embryos, but in all cases mismatch morpholino embryos (n≥8 for each marker) gave identical results as control, with normal expression of these markers. The pancreas is outlined.

To determine whether knockdown of insm1 had a global effect on anterior endodermal organs, we examined whether the development of other endodermal cell types was affected upon loss of insm1. We found normal development of the liver and stomach/duodenum as judged by expression of hex and frp5, respectively (Fig. 4A–D). We did however find reduced expression of hnf6 in the duodenum, but normal expression of this marker in the gall bladder (Fig. 4E,F). In conclusion, morpholino-mediated knockdown of insm1 had no effect on liver or gall bladder, but did appear to affect development of duodenum.

Fig. 4.

Development of anterior endodermal organs is unaffected. A,B: Hex gene expression is normal in insm1 morpholino injected samples (88%, n=8). C,D: Frp5 expression is slightly reduced, but overall the stomach appears normal (66%, n=25). E,F: Hnf6 expression. Hnf6 is expressed in the posterior duodenum and the gall bladder. In insm1 knockdown tadpoles, expression in the gall bladder is normal, while its expression in the gut is absent (72%, n=18). Pa, pancreas; Li, liver; GB, gall bladder. Pictures presented in control column are of uninjected embryos, but in all cases mismatch morpholino embryos (n≥8 for each marker) gave identical results as control, with normal expression of these markers. The pancreas is outlined.

The above results demonstrated that final differentiation of endocrine cells is affected in insm1 knockdown embryos, but it does not address whether this effect is seen earlier in development when the pancreatic domain is first specified. This is especially important, as insm1 is one of the earliest markers of the dorsal pancreas. To determine if initial specification of the anterior endoderm was affected by inhibition of insm1 we examined whether there were defects in the early expression of three anterior endoderm markers, pdx1, ptf1a and hnf6. At NF35, when specification of this region is occurring, we found normal expression of all three markers in the anterior endoderm (Fig. 5). These results confirm that knockdown of insm1 did not affect initial specification of the dorsal and ventral pancreas.

Fig. 5.

Initial specification of the anterior endoderm is normal. A,B: Expression of ptf1a in the developing dorsal and ventral pancreatic buds was normal in insm1 knockdown tadpoles at NF35 (100%, n=10). C,D: Expression of pdx1 in the pancreas and duodenum was unaffected (95%, n=20). E,F: Initial expression of hnf6 in the anterior endoderm was also unaffected by the knockdown on insm1 (100%, n=8). Pictures presented in control column are of uninjected embryos, but in all cases mismatch morpholino embryos (n=15 for each marker) gave identical results as control, with normal expression of these markers.

We attempted to rescue this knockdown phenotype by overexpressing insm1 mRNA (lacking the 5’UTR sequence to which the morpholino was designed, flag tag at 5’end) along with the insm1 5’UTR morpholino, but were unable to obtain any rescue even when 2ng of mRNA was injected (data not shown). We are unsure why the insm1 morpholino induced phenotype cannot be rescued, but it may be due to the fact that overexpression of insm1 mRNA alone in the endoderm has no effect. We injected up to 2ng of insm1 mRNA alone into the vegetal hemisphere, but did not observe any phenotype (data not shown). As ectopic induction of endocrine cells may not be observable by morphology, we examined insm1-injected embryos for ectopic expression of several different endocrine-specific markers (ins, neuroD, pax6), but found no increased expression of these markers (data not shown). Therefore, although we can show that the insm1 morpholino specifically inhibits translation of insm1 mRNA (Fig. 1O), we were unable to demonstrate that insm1 mRNA was sufficient to rescue the knockdown phenotype. This inability to promote ectopic endocrine cell fates in naïve endoderm by such an early acting transcription factor suggests that insm1 function in endocrine cell fate determination requires other factors, and stands in contrast to the results found with other lineage specific factors (Pearl and Horb, 2008). In contrast, overexpression of medaka insm1 does produce a phenotype, not affecting cell fate specification, but rather cell cycle control (Candal et al., 2007). It will be of interest to see whether these differences are due to specific amino acid changes and whether mammalian insm1 has any ability to promote endocrine cell fates.

In recent years, the use of Xenopus to study endodermal organogenesis has increased, including liver, pancreas and gastrointestinal tract (Afelik et al., 2006; Jarikji et al., 2007; McLin et al., 2007; Zorn and Wells, 2007; Li et al., 2008; Jarikji et al., 2009; Pearl et al., 2009). However, it has not been established whether the differentiation of specific endocrine cells within the gastrointestinal tract and pancreas occurs in the same stepwise manner as seen in mammals. Given that the endoderm develops in a completely different manner from mammals and is used as a nutrient source during much of early development, it is important that this issue be resolved. In this paper, we show that Xenopus insm1 functions in a manner similar to that seen in mammals (Gierl et al., 2006; Mellitzer et al., 2006). We demonstrate that insm1 is expressed in both gastrointestinal and pancreatic endocrine cells and is required for their differentiation. Therefore, although Xenopus endoderm develops in a unique morphological manner, the molecular mechanisms underlying development and differentiation of endocrine cells appear similar.

Materials and Methods

The full length X. laevis insm1 cDNA was cloned based on X. tropicalis genomic sequence and a X. laevis IMAGE clone 4959608. Our nucleotide sequence is identical to the recently deposited GenBank sequence (EU076904). Whole mount in situ hybridizations with single probes were performed as described using BM Purple (Horb et al., 2003; Jarikji et al., 2009). Complete information for each clone is available upon request. For whole mount in situ analysis of insm1 and insulin in the endoderm at tail bud and early tadpole stages, we first fixed the embryos for 15 minutes and then cut them open using a razor blade and continued fixing for an additional 45 minutes. This method allows for better probe penetration into the deep endoderm than traditional whole mounts.

Antisense morpholino oligonucleotides were designed by Gene Tools, LLC. The antisense morpholinos were designed either to the translation start or in the 5’UTR. Specificity was confirmed by in vitro transcription and translation for each gene and morpholino. Morpholinos were injected into the dorsal vegetal blastomeres at the eight-cell stage, and targeting to the stomach, liver and pancreas was confirmed by monitoring fluorescence from labeled oligonucleotides. The morpholinos used were insm1 utr- 5’-CCAGTGAGTCTGGGCTGAATGCTTT-3’, insm1 start- 5’-GCTTGACCAGGAAACCTTTAGGCAT-3’, insm1 mismatch-5’-GGTTCACGAGCAAAGCTTTACGGAT-3’. The specificity of the morpholinos was tested in in vitro transcription/translation reactions using a FLAG antibody to detect the INSM1 protein. The insm1-flag clone used in these reactions was cloned in frame by PCR amplifying the insm1 ORF into pCS107-FLAG vector. We also constructed a flag-insm1 clone for the rescue experiments in similar fashion. We ensured that there were no mutations introduced by the PCR by sequencing the clones.