Expression of Sox transcription factors in the developing mouse pancreas



Previous work has identified members of the homeodomain and basic helix-loop-helix families of transcription factors as critical determinants of mammalian pancreatic development. Here, we describe the identification of HMG-box transcription factors of the Sox gene family in the mouse pancreas. We detected transcripts for Sox11, Sox4, Sox13, Sox5, Sox9, Sox8, Sox10, Sox7, Sox17, Sox18, Sox15, and Sox30 in embryonic pancreas and found Sox4, Sox9, and Sox13 in adult pancreatic islets. Expression of seven of these Sox factors was studied in more detail by in situ hybridization from the stage of early pancreatic outgrowth to birth. Expression of Sox11 was found in the mesenchyme surrounding the pancreatic buds, whereas Sox4 and Sox9 were confined to the pancreatic epithelium and later to islets. Sox13 and L-Sox5 showed expression in most of the pancreatic epithelial cells between embryonic days 12.5 and 14.5. Sox8 and Sox10 were detected in a thin layer of cells surrounding the islets. The expression patterns of Sox genes in the embryonic pancreas suggest that they could have important and possibly redundant functions in pancreas development. Developmental Dynamics 227:402–408, 2003. © 2003 Wiley-Liss, Inc.


A better understanding of the molecular programs that underlie organogenesis and tissue maintenance is fundamental to the development of therapeutic strategies aimed to induce organ regeneration or the production of replacement cells. Given the pivotal role of pancreatic insulin-producing cells in the control of glucose homeostasis, a growing number of studies have focused on exploring the molecular signals that control morphogenesis and cell differentiation of the mammalian pancreas (Gu et al.,2003; Wilson et al.,2003).

In mice, the pancreas develops as separate dorsal and ventral evaginations from the gut endoderm at the foregut/midgut junction (Slack,1995), which later fuse to form the organ. The pancreatic epithelium gives rise to exocrine cells, pancreatic ductal cells, and four different types of endocrine cells, which produce the hormones insulin, glucagon, somatostatin, and pancreatic polypeptide (PP), respectively.

Proliferation of the pancreatic epithelium depends on signals from the surrounding mesenchyme (Golosow and Grobstein,1962; Wessells and Cohen,1967). In the absence of mesenchyme, growth and morphogenesis of the pancreatic buds are severely impaired and the epithelium undergoes a “default” endocrine differentiation at the expense of exocrine acinar formation (Gittes et al.,1996; Miralles et al.,1998). Recent studies suggest that fibroblast growth factors and the mitogen-activated protein kinase signaling pathway are involved in mediating this interaction between mesenchyme and epithelium (Bhushan et al.,2001; Elghazi et al.,2002). However, the factors that confer the property to promote exocrine differentiation to the pancreatic mesenchyme remain to be studied.

The first differentiated cells detected are glucagon-producing cells in the dorsal pancreatic bud at embryonic day (E) E9.5. Few insulin-producing cells are already present at E10.5, but the majority differentiates between E13 and birth. At the end of gestation, endocrine cells cluster into morphologically distinct structures, called islets of Langerhans, which are found scattered throughout the exocrine acinar tissue. Within each islet, a large core of insulin-producing cells is surrounded by the three other endocrine cell types.

Through gain- and loss-of-function studies conducted primarily in mice, several genes involved in pancreatic organogenesis and cell differentiation have been identified. Many of these genes encode transcription factors (Edlund,1998).

The Sox gene family represents a group of developmentally regulated transcription factors, which are characterized by a highly conserved DNA-binding domain, the HMG-domain (Wegner,1999; Schepers et al.,2002). Despite their similar DNA-binding characteristics, Sox proteins differ in their structure outside the HMG-domain and, therefore, may perform different functions in vivo. Based upon structural similarities, the more than 20 different vertebrate Sox genes can be assigned to one of the seven subgroups, A–G (Schepers et al.,2002). Recently, several Sox genes have been recognized as key players in the regulation of embryonic development and cell fate determination. The analysis of Sox gene mutations in humans, mice, and zebrafish have demonstrated a role for Sox genes in endoderm specification, as well as the development of gonads, lens, heart, lymphocytes, bone, and glial cells (Kent et al.,1996; Schilham et al.,1996; Britsch et al.,2001; Dickmeis et al.,2001; Kamachi et al.,2001; Lefebvre et al.,2001; Stolt et al.,2002). Despite their broad function in embryonic development, the expression and function of Sox genes during development of the pancreas remain to be explored. In this study, we used reverse transcriptase-polymerase chain reaction (RT-PCR) as well as in situ hybridization to evaluate the expression of Sox genes in the developing mouse pancreas.


To explore which mouse murine Sox genes are expressed in the developing and adult pancreas, we used RT-PCR on RNA from either E12.5 or E15.5 total pancreas or adult islets, using degenerate oligonucleotide primers to the highly conserved HMG-box. These degenerate primers have been widely used to amplify different members of the Sox gene family (Stock et al.,1996; Roose et al.,1998). Random sequencing of independent cDNA clones revealed expression of group C, D, and E Sox genes in the pancreas (summarized in Table 1). Within group C, cDNA clones coding for Sox4 were most frequently isolated, but clones for Sox11 and Sox12 were also found in the embryonic pancreas. Among group D members, we found expression of Sox5 and Sox13. Within group E, Sox9 was predominantly isolated from embryonic pancreas, but we also identified transcripts for Sox8 and Sox10 during pancreas development. From 50 different clones sequenced from adult islets, we identified transcripts for Sox4, Sox9, and Sox13. These data show expression of at least eight different members of the Sox gene family in the embryonic pancreas, and three members in adult islet cells. However, it should be noted that the RT-PCR approach may not cover all Sox genes in the mouse pancreas and that the relative number of sequenced clones does not necessarily reflect the abundance of a particular Sox transcript at a given developmental stage.

Table 1. Degenerate RT-PCR for Sox Genes in Mouse Pancreasa
GroupGeneE12.5 (%)E15.5 (%)Adult islets (%)
  • a

    Reverse transcriptase-polymerase chain reaction (RT-PCR) was performed using degenerate oligonucleotides described in the Experimental Procedures section with RNA purified from isolated embryonic mouse pancreata at the indicated ages (E, embryonic day) or mouse adult islets. The values show the frequencies of each gene as a percentage of a total of 50 sequenced subclones per age.


Given the potential preference of the degenerate RT-PCR primers for certain Sox family members, we next used gene-specific primers to directly test if members of group A, B, F, G, or H are found in the developing pancreas. Although Sry, Sox1, Sox2, and Sox3 were not detected, transcripts for Sox7, Sox15, Sox17, Sox18, and Sox30 were amplified from embryonic pancreas (Table 2), showing that numerous Sox genes from almost all subgroups are expressed in the pancreatic anlage.

Table 2. RT-PCR with Gene-Specific Primers for Group A, B, F, G, H Sox Genes in Mouse Pancreasa
  • a

    Reverse transcriptase-polymerase chain reaction (RT-PCR) was performed using gene-specific primers described in the Experimental Procedures section with RNA purified from isolated embryonic mouse pancreata at the indicated ages (E, embryonic day). (+) means that transcripts were detected, (−) means that no transcripts were detected, despite amplification from a control RNA.


We next examined the expression domains of Sox11, Sox4, Sox13, Sox5, Sox9, Sox8, and Sox10 in the developing pancreas by in situ hybridization or enzymatic X-gal staining on pancreatic sections throughout development of the organ. Between E9.5 and E10.5, Sox11 was detected in the mesenchyme surrounding both pancreatic buds and in a few cells of the pancreatic epithelium (Fig. 1A; data not shown). At E12.5, Sox11 was exclusively found in a subset of pancreatic epithelial cells (Fig. 1B). Sox4-positive cells were first observed at E12.5, showing wide epithelial expression in the pancreas (Fig. 1F). At E14.5, both Sox11 (Fig. 1C,D) and Sox4 (Fig. 1G,H) were broadly expressed in the epithelium. At the end of gestation, around E18.5, the islet cells have begun to cluster into islets of Langerhans and islets can be morphologically distinguished from exocrine acinar cells and pancreatic ducts. At E18.5, Sox11 and Sox4 were detected in the entire forming endocrine islets (Fig. 1E,I). Both factors were coexpressed with glucagon (Fig. 1E,I), and their wide expression in the center of the islets suggests expression in insulin-producing cells. Transcripts for Sox4 were also found in a subset of exocrine acinar cells (Fig. 1I).

Figure 1.

Expression of Sox11 and Sox4 in the developing mouse pancreas. In situ hybridization for Sox11 (A–E) and Sox4 (F–I) on cryosections through the pancreas at the indicated ages. A: The dorsal pancreatic epithelium is circled, and Sox11 expression in the pancreatic mesenchyme indicated by an arrow. In C and G, the boxes represent the regions that are shown in D and H, respectively, in higher magnification. E,I: To visualize the outer boundary of the islets, sections were immunostained with an anti-glucagon antibody (red) after in situ hybridization. The fluorescence images for glucagon and the light microscopy images were overlaid in Photoshop. The arrows indicate islet cells (i) and exocrine acinar cells (ac). E, embryonic day; s, stomach. Scale bars = 100 μm in A–C,E–G,I.

Within group D, Sox13 was first expressed in a subset of epithelial cells at E10.5, and broadly detected in the epithelium at E12.5 (Fig. 2A,B). The first L-Sox5–expressing epithelial cells were observed at E12.5 (Fig. 2D), and at E14.5, both L-Sox5 and Sox13 were widely expressed in the pancreatic epithelium (Fig. 2C,E). At E18.5, Sox13 and L-Sox5 mRNAs were no longer detected by in situ hybridization but could be amplified by RT-PCR from whole pancreas RNA with gene-specific primers (data not shown). These data could either suggest that their expression levels decrease in late embryogenesis, or alternatively that the increased levels of RNases in the perinatal pancreas impair the detection of the transcripts.

Figure 2.

Expression of Sox13 and L-Sox5 in the developing mouse pancreas. In situ hybridization for Sox13 (A–C) and L-Sox5 (D,E) on cryosections through the pancreas at the indicated ages. A: The dorsal pancreatic epithelium is circled. E, embryonic day; s, stomach. Scale bars = 100 μm in A–E.

Among group E members, Sox9 was expressed in the epithelium of both pancreatic buds at E9.5 (Fig. 3A,B). Between E10.5 and E12.5, Sox9 was strongly expressed in the entire pancreatic epithelium (Fig. 3C,D). By E14.5, the majority but not all pancreatic epithelial cells were still Sox9-positive (Fig. 3E). In late gestation, Sox9 expression became more restricted and was found in the islets of Langerhans, in a subset of ductal epithelial cells, and in few exocrine acinar cells (Fig. 3F). The expression of Sox9 in the embryonic pancreatic ducts is of particular interest, because ducts have been suggested to contain the endocrine progenitor cells (Schwitzgebel et al.,2000). Similar expression patterns, as observed for Sox9, have been described for Nkx2.2 and Nkx6.1, which are crucial determinants of endocrine cell differentiation (Sussel et al.,1998; Sander et al.,2000). Expression of Sox9 has also been reported recently in the human pancreas (Piper et al.,2002). The main difference between our results and the study in humans is that the expression of Sox9 decreases in human islets during the second trimester. Of interest, the analysis of humans with haploinsufficiency for Sox9 revealed a functional role for Sox9 in islet cell development, as evidenced by defects in islet cell morphology, as well as reduced expression of markers for mature islet cells (Piper et al.,2002).

Figure 3.

A–F: Expression of Sox9 in the developing mouse pancreas. In situ hybridization for Sox9 on cryosections through the pancreas at the indicated ages. Expression of Sox9 in the ventral (vb) (A) and dorsal (db) pancreatic buds (B). To visualize the pancreatic epithelium, the right half of each image shows an immunostaining for Pdx1 on an adjacent section. C,D: Sox9 expression marks the entire dorsal pancreatic epithelium. F: The arrows indicate Sox9 signal in islet cells (i), exocrine acinar cells (ac), and ductal cells (d). Note that the majority of acinar cells do not express Sox9. E, embryonic day; s, stomach. Scale bars = 100 μm in A–F.

Expression of Sox8 and Sox10 was studied by enzymatic X-gal staining, using mice heterozygous for a gene replacement of the coding sequence with β-galactosidase (Britsch et al.,2001; Sock et al.,2001). Sox8 and Sox10 were found in scattered cells at the epithelial/mesenchymal boundary at E10.5 (Fig. 4A,E). At E12.5, individual Sox8- and Sox10-positive cells were dispersed between mesenchyme and epithelium and did not express glucagon (Fig. 4B,F). A very similar pattern of Sox8 expression was observed in the developing chick pancreas (Bell et al.,2000). In late embryogenesis, Sox8- and Sox10-expressing cells were found at the periphery of forming islets. Immunohistochemistry with anti-glucagon, anti-insulin, anti-somatostatin, and anti-PP antibodies on X-gal stained tissue sections revealed that these Sox8- and Sox10-positive cells do not express any of the four hormones (Fig. 4C,D,G,H; data not shown). The peri-islet expression of Sox8 and Sox10 was maintained in the adult pancreas (data not shown).

Figure 4.

Expression of Sox8 and Sox10 in the developing mouse pancreas. β-Galactosidase staining for Sox8 (blue in A–D) or Sox10 (blue in E–H) and subsequent immunodetection of glucagon (brown in A–C and E–G) or insulin (brown in D, H) on paraffin sections through the pancreas at the indicated ages. A,E: The dorsal pancreatic epithelium is circled. B: The inset shows a confocal image of a double immunofluorescence with anti–β-galactosidase (red) and anti-glucagon (green) antibodies at embryonic day (E) 12.5. C,D,G,H: The arrows indicate hormone-positive pancreatic islets (i). s, stomach. Scale bars = 100 μm in A–H.

To investigate possible roles for Sox8 and Sox10 in pancreas development, we analyzed pancreatic sections from Sox8 and Sox10 homozygous mutant late gestation embryos, as well as from adult Sox8 mutant mice. These sections showed no histological abnormalities (Fig. 5A–D), normal expression of the endocrine hormones (Fig. 5E–H), and normal production of amylase (data not shown), suggesting that endocrine and exocrine differentiation does not depend on Sox8 or Sox10 function. In the peripheral nervous system, Sox10 and Sox8 are expressed in glial cells and Sox10 has been shown to be required for the development of these cells (Britsch et al.,2001; Sock et al.,2001). Because pancreatic islets are enveloped in a sheath of glial cells (Teitelman et al.,1998), it is possible that the Sox10-positive cells in the pancreas are indeed glial cells. Our finding that no β-galactosidase–positive cells could be detected in pancreas of Sox10 mutant embryos (data not shown) further suggests that these cells require Sox10 for their development. To this end, the functional significance of glial cells in the pancreas remains to be determined and the perinatal lethality of Sox10 mutants precludes studies of islet cell function in adult Sox10 homozygous mutant mice.

Figure 5.

Expression of islet hormones in Sox8- and Sox10-deficient mice. A–D: Hematoxylin and eosin staining of pancreatic sections from adult wild-type (+/+) and Sox8 mutant mice (-/-), as well as from wild-type (+/+) and Sox10 mutant (-/-) embryos at embryonic day (E) 18.5. E–H: Immunofluorescence staining with anti-insulin (ins; shown in red) and anti-glucagon (glc; shown in green) antibodies. Five independent pancreata were analyzed from both wild-type and mutant mice. Scale bar = 100 μm in A (applies to A–H).

In conclusion, our data show that numerous members of the Sox gene family are expressed during pancreas development. Some expression domains, such as expression of Sox11 in the mesenchyme around the pancreatic buds, appear to be unique to just one Sox gene, whereas other domains may partially overlap. For example, pancreatic islets at E18.5 expressed Sox11, Sox4, and Sox9. Although only coexpression analysis with suitable antibodies will prove expression of more than one Sox factor in a cell, our observation that several members of an individual Sox gene subgroup are expressed in embryonic pancreas suggests that redundancy may exist. This knowledge will help to interpret descriptions of pancreatic phenotypes in Sox gene mutant mice. Finally, given the maintenance of Sox gene expression in adult islets, it will be interesting to investigate their role in islet cell function.


Amplification and Cloning of Sox Genes From Pancreas RNA

Total RNA was isolated from dissected mouse pancreatic epithelia and adult mouse islets, using the RNeasy kit (Qiagen). By using random-hexamer primers, first-strand cDNA was synthesized from 1 μg of RNA. To amplify and clone cDNA products, we used a degenerate RT-PCR approach based on high sequence homology within the HMG-domain of the Sox transcription factor family (Yuan et al.,1995). The PCR products were cloned into pGEM-T easy vector (Promega), individual clones were randomly picked, tested for the presence of an insert by restriction digestion, and sequenced. Expression of Sox genes was examined by using the following primers: Sry, upstream primer (nucleotides [nt] −6–13, Genbank accession no. NM_011564) 5′-GAG AGC ATG GAG GGC CAT, downstream primer (nt 258–246) 5′-CCA CTC CTC TGT GAC ACT; Sox1, (nt 617–639, NM_009233) 5′-TGCAGGAGGCACAGCTGGCCTAC, (nt 897–876) 5′-TGCCGCCACCGCCGAGTTCTGG; Sox2, (nt 376–399, NM_011443) 5′-AAGTACACGCTTCCCGGAGGCTTG, (nt 787–766) 5′-AGTGGGAGGAAGAGGTAACCAC; Sox3, (nt 925–948, NM_009237) 5′-TCTCCGCCGCCCGCCATCCGTTCG, (nt 1108–1089) 5′-CCGTTCCATTGACCGCAGTC; Sox5,(nt 1696–1715, NM_011444) 5′-TGGAGATTCTGACGGAAGCG, (nt 2378–2396) 5′-CTTGTCCCGCAATGTGGTT; Sox13, (nt 787–806, NM_011439) 5′-CCCCACAACCACTGAACCTC, (nt 1487–1470) 5′-TGGACGCCGTGTCCTCAT; Sox7, (nt 152–173, NM_011446) 5′-ACCTTCAGGGGACAAGAGTTCG, (nt 497–476) 5′-GTTTTTCTCAGGCAGCGTGTTC; Sox17, (nt 383–404, NM_011441) 5′-AAGGCGAGGTGGTGGCGAGTAG, (nt 871–852) 5′-CCTGGCAGTCCCGATAGTGG; Sox18, (nt 265–287, NM_011441) 5′-CGAATCAGGGCGCTATGGCTTTG, (nt 676–657) 5′-AGTGGGTAGCTCGCGGAAGG; Sox15, (nt 108–127, AB014474) 5′-TGGAGCGTCTGGGGGACTTC, (nt 633–611) 5′-TGGGGATAGGTAAGGGGAGAAAG; Sox30, (nt 1922–1941, AV255326) 5′-CGGTTCTCCTTTCATCACCC, (nt 2255–2236) 5′-CCAAGGCTCCAATGTCCAGA.

Expression Analysis

In situ hybridizations on 10-μm frozen sections were performed as previously described (Gradwohl et al.,1996). The following probes were used: Sox4 (Schilham et al.,1996), Sox9 (Wright et al.,1995), probes for L-Sox5 (nt 12–726, GenBank accession no. NM_011444), Sox11 (nt 1103–1907, NM_009234), and Sox13 (nt 1725–2430, NM_011439) were generated by PCR amplification and subcloning into pBlueScript (KS+; Stratagene). Immunodetection with guinea pig anti-insulin (Linco), guinea pig anti-glucagon (Linco), rabbit anti-somatostatin (Dako), rabbit anti–pancreatic polypeptide (Dako), rabbit anti–β-galactosidase (ICN Pharmaceuticals), and rabbit anti-Pdx1 (kind gift from Helena Edlund) antibodies was performed as described (Sander et al.,2000). As previously detailed, β-galactosidase activity was enzymatically detected in heterozygous Sox8- (Sock et al.,2001) and Sox10-deficient mice (Britsch et al.,2001).


We thank H. Clevers for Sox4 probe, P. Serup for providing mouse islet cDNA, and Christoph Janiesch for technical assistance. M.S. was funded by a CDA from JDRF.