GATA4 and GATA6 control mouse pancreas organogenesis (original) (raw)
Single inactivation of Gata4 and Gata6 does not affect pancreas formation. To investigate the role of GATA4 and GATA6 in pancreas development, mice with a conditional (flox) allele of Gata4 (Gata4flox/flox) or Gata6 (Gata6flox/flox) were crossed to a transgenic mouse line that expresses Cre recombinase under the control of the pancreatic and duodenal homeobox gene 1 (Pdx1) promoter (Pdx1-Cre mice) (15). Pdx1 is expressed early in the multipotent pancreatic progenitors that give rise to all pancreatic cell types, thus allowing gene inactivation in the entire pancreas (15). Expression analysis of Cre activity in Pdx1-Cre;Rosa26R transgenic mice revealed robust lacZ staining in the pancreatic epithelium as early as E9.5 (Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI63240DS1). Gata4flox/flox and Gata6flox/flox mice have been used previously to successfully analyze the role of these transcription factors in other organs, such as heart and lung (16, 17). Gata4flox/flox;Pdx1-Cre and Gata6flox/flox;Pdx1-Cre mice were born at the expected Mendelian frequency and appeared overtly normal. Immunohistochemical staining demonstrated an efficient recombination of Gata4 and Gata6 floxed alleles by Cre recombinase in pancreatic tissue (Figure 1, A–D). Gross morphological examination and histological analysis by H&E staining of newborn pancreata did not reveal obvious defects in either Gata4flox/flox;Pdx1-Cre or Gata6flox/flox;Pdx1-Cre mice when compared with control littermates (Figure 1, E–J). Occasional, mild ductal dilation in acinar cells of Gata4flox/flox;Pdx1-Cre newborn pancreata was detected (Figure 1I). However, this phenotype was not observed in adult mice (data not shown), indicating a transient role for GATA4 in early exocrine formation. Immunohistochemical analysis for different epithelial cell lineages (acinar, ductal, and islet) within the pancreas revealed that pancreatic cell differentiation was unaffected in Gata4flox/flox;Pdx1-Cre and Gata6flox/flox;Pdx1-Cre mutant mice (Figure 1, K–P). Furthermore, Gata4flox/flox;Pdx1-Cre and Gata6flox/flox;Pdx1-Cre mice exhibited normal glucose tolerance to intraperitoneal glucose load (Supplemental Figure 2). Taken together, our results show that pancreas morphogenesis is not affected by inactivation of either GATA4 or GATA6 alone.
Single inactivation of Gata4 and Gata6 does not affect pancreas formation. (A–D) Immunohistochemical analyses show strong expression of GATA4 in acinar cells (arrowheads in A) and GATA6 in endocrine cells (arrows in C) in pancreatic sections of control mice at P1. Loss of GATA4 (B, arrowheads) and GATA6 (D, arrows) in newborn conditional mutant mice is confirmed by immunohistochemical analysis. (E–G) Gross morphology of neonatal WT and conditional mutant guts. H&E-stained sections of newborn control (H), Gata4 (I), and Gata6 (J) conditional knockout pancreata does not reveal major defects in pancreas architecture. Mature acinar (amylase), ductal (mucin) (K–M), and islet (insulin and glucagon) markers (N–P) are normally expressed in single Gata4 and Gata6 conditional knockout mice. Insets in H–M show higher magnification of acinar cells. Mild ductal dilation in acinar cells of Gata4flox/flox;Pdx1-Cre is observed (I and L, insets). Nuclei are counterstained with DAPI in K–P. Scale bars: 50 μm; 25 μm (insets).
Pancreatic agenesis in GATA4 and GATA6 double-mutant mice. Previous studies have shown that Gata4 and Gata6 have overlapping expression patterns at early stages of pancreas formation, when the pancreatic epithelium comprises mainly multipotent pancreatic progenitor cells (MPCs) (9) (see also Supplemental Figure 4). Since pancreas formation is unaffected in individual Gata4 and Gata6 conditional knockout mice, we reasoned that these transcription factors might have redundant roles during pancreas development. To test this hypothesis, we generated double Gata4flox/flox;Gata6flox/flox;Pdx1-Cre mutant mice (double-mutant mice hereafter). Double-mutant mice displayed growth retardation and hyperglycemia and died shortly after birth (data not shown). Dissection of newborn double-mutant mice revealed a near total absence of pancreatic tissue (Figure 2D). Close histological analysis of the pancreatic remnant revealed the presence of epithelial cysts and abundant stroma (Figure 2, H and Q), while exocrine and endocrine tissue were almost completely absent (Figure 2, L and P). Cysts comprised epithelial cells that express mature ductal markers, including cytokeratin 19 (Figure 2S), and react with lectin Dolichos biflorus agglutinin (DBA) (Figure 2T). Although the majority of double-mutant mice displayed pancreatic agenesis, the degree of pancreatic tissue loss was variable among individual mutant pups, with approximately 20% of double-mutant mice displaying severe pancreatic hypoplasia. Histological analysis of hypoplastic pancreata revealed well-differentiated tissue (Supplemental Figure 3). However, GATA4 and GATA6 expression was still observed in these cells, indicating inefficient excision of Gata4 and Gata6 floxed alleles, perhaps relating to the number of floxed alleles requiring recombination (Supplemental Figure 2 and data not shown). _Pdx1_-Cre activity in pancreas is reported to be mosaic (15), and thus it is conceivable that unrecombined Gata4+Gata6+ progenitor cells might have partially repopulated the pancreas in double-mutant mice. The phenotype of the double-mutant mice suggests that loss of both GATA4 and GATA6 confers a competitive disadvantage to developing pancreatic cells. A dramatic reduction in pancreatic mass was also observed in Gata4flox/flox;Gata6flox/+;Pdx1-Cre mutant mice (Figure 2, B and F), indicating that pancreas formation is sensitive to reduced Gata6 dosage. Gata4flox/flox;Gata6flox/+;Pdx1-Cre mutant mice displayed severe acinar cell loss, and the architecture of remaining acini appeared impaired (Figure 2, F and J). Interestingly, tissue architecture of Gata4flox/+;Gata6flox/flox;Pdx1-Cre mutant pancreata was completely normal (Figure 2, C, G, K, and O), indicating that GATA4 and GATA6 are not completely equivalent in regulating pancreas formation. Together, our results indicate that pancreatic morphogenesis requires GATA4 and GATA6 activity and that these transcription factors play redundant functions in this process.
Pancreatic agenesis in Gata4/Gata6 double mutant. Gross appearance of neonatal WT and conditional mutant guts (A–D) and pancreatic sections stained with H&E (E–H) reveal the abnormal morphology of double-mutant pancreata at P1. Gata4flox/flox;Gata6flox/+;Pdx1-Cre mice show pancreatic hypoplasia with scarcity of acinar cells (B and F). Gata4flox/+;Gata6flox/flox;Pdx1-Cre mice display normal pancreatic mass and architecture (C and G). Immunohistochemical analysis shows reduced expression of the acinar marker, amylase, in Gata4flox/flox;Gata6flox/+;Pdx1-Cre pancreatic sections (J) compared with Gata4flox/+;Gata6flox/flox;Pdx1-Cre (K) and control littermates (I). The double-mutant pancreatic remnant displays cystic structures surrounded by abundant stroma (H, L, P, and Q). The cystic structures express mucin (L) and cytokeratin 19 (S) and react with DBA lectin (T), which are markers of differentiated ductal cells. Immunostaining for E-cadherin confirms the epithelial nature of the cysts (R). Insulin and glucagon staining reveals normal differentiation of the endocrine lineage in Gata4flox/flox;Gata6flox/+;Pdx1-Cre (N) and Gata4flox/+;Gata6flox/flox;Pdx1-Cre (O) mutant mice in comparison with control mice (M). In contrast, Gata4/Gata6 double-mutant mice lack endocrine cells (P). Counterstaining with DAPI was performed to reveal nuclei. Scale bars: 50 μm.
Pancreatic epithelium of Gata4/Gata6 double-mutant embryos fails to expand. To delineate the role of GATA4 and GATA6 during pancreas development, we characterized the pancreatic morphology of double-mutant mice at different developmental stages. During the early stages of pancreatic development, when bud formation occurs (E11.5), Gata4 and Gata6 expression completely overlapped in the pancreatic epithelium (Supplemental Figure 4). However, no major morphological changes in pancreatic epithelia of the double-mutant mice were observed (Supplemental Figure 4). Efficient inactivation of the Gata4 and Gata6 genes by Cre recombinase at this developmental stage was confirmed by immunohistochemistry (Supplemental Figure 4). By E13.5, the pancreatic epithelium expands and undergoes extensive branching to form a network of tubules. Concomitantly, a massive wave of endocrine and acinar cell differentiation, known as the secondary transition, begins (18, 19). At this embryonic stage, Gata6 is homogeneously expressed throughout the pancreatic epithelium (Supplemental Figure 4). In contrast, Gata4 displays a very specific expression pattern, restricted to the tips of the branching ductal epithelium, which will further adopt an acini fate (Supplemental Figure 4). Histological analysis of double-mutant embryos revealed a smaller and disorganized epithelium compared with Gata4flox/flox;Gata6flox/+;Pdx1-Cre, Gata4flox/+;Gata6flox/flox;Pdx1-Cre, and control embryos, suggesting that epithelial expansion was disrupted (Figure 3, A–D). Furthermore, immunohistochemical staining for mucin on whole-mount pancreata revealed a significant reduction in the pancreatic epithelial area of the double mutant when compared with control littermates (Supplemental Figure 5). Consistent with these data, pancreatic epithelial morphogenesis was severely affected in double-mutant embryos at E15.5 (Supplemental Figure 4). These results indicate that pancreatic agenesis in newborn double-mutant mice is a consequence of defective pancreatic morphogenesis during early pancreatic development.
Pancreatic epithelial expansion is impaired in the absence of GATA4 and GATA6 activity. The pancreatic epithelia of control, _Gata4flox/flox;Gata6flox/+;Pdx1_-Cre, and _Gata4flox/+;Gata6flox/flox;Pdx1_-Cre embryos at E13.5 display normal morphology (A–C), whereas double-mutant pancreatic epithelium appears disorganized and reduced in epithelial area (D). Immunohistochemistry analysis of the mitotic marker phospho-histone H3 (PHH3) reveals a significant reduction in proliferating pancreatic epithelial cells in the double-mutant (H) compared with littermate embryos at E13.5 (E–G). Immunostaining with another proliferation marker, Ki67, confirms the reduction in proliferation of E13.5 pancreatic epithelial cells in the double mutant compared with littermates (I–L). Counterstaining with DAPI was performed to reveal nuclei. Quantification of proliferating cells, measured as the number of PHH3 and Ki67-positive cells (M and N, respectively) per E-cadherin–positive cells. *P < 0.05. Scale bars: 50 μm.
Defects in proliferation of pancreatic progenitors in Gata4/Gata6 double-mutant embryos. During pancreatic bud growth, extensive proliferation of the pancreatic epithelial cells occurs. To determine whether pancreatic growth arrest in double-mutant mice was due to defects in proliferation of pancreatic epithelial cells, we performed immunohistochemical analyses for proliferation markers at E13.5. Quantification for the mitotic marker phospho-histone H3 revealed a 50% reduction in proliferating cell number in the pancreatic epithelium of the double-mutant embryos when compared with control, Gata4flox/flox;Gata6flox/+;Pdx1-Cre, and Gata4flox/+;Gata6flox/flox;Pdx1-Cre littermates (Figure 3, E–H, and M). These results were confirmed using another proliferation marker, Ki67, a protein present during all active phases of the cell cycle. Based on Ki67 staining, a high percentage of proliferative pancreatic cells was found in control, Gata4flox/flox;Gata6flox/+;Pdx1-Cre, and Gata4flox/+;Gata6flox/flox;Pdx1-Cre littermate embryos (Figure 3, I–K, and N). In contrast, double-mutant mice showed a decreased number of Ki67-positive cells in the pancreatic epithelium compared with control mice (Figure 3, I, L, and N). Although not statistically significant, we also observed a decreasing trend in proliferation of pancreatic epithelial cells in Gata4flox/flox;Gata6flox/+;Pdx1-Cre embryos compared with control mice that might partially explain the hypoplastic phenotype of these mice at neonatal stages (Figure 3, F, J, M, and N). These results indicate that GATA4 and GATA6 are necessary for the normal proliferation rate of the pancreatic cells to sustain embryonic epithelial growth.
Endocrine and exocrine differentiation is arrested in the absence of GATA4 and GATA6 activity. Our previous results indicated that pancreatic epithelial growth and expansion were impaired in the absence of GATA4 and GATA6 function at E13.5. The epithelial remodeling that occurs during this period appears to be intimately connected to the formation of different progenitor domains in the progenitor ductal epithelium. Different pancreatic lineage progenitors display a characteristic distribution within the pancreatic epithelium at this embryonic stage (20). Multipotent progenitor cells are found at the tips of the branching pancreatic epithelium, while ductal-endocrine progenitor cells are located along the trunk (20–22). The tip and trunk domains can be easily identified by the expression of Carboxypeptidase A1 (_Cpa_1) and Neurogenin 3 (Ngn3), respectively (20, 23). Cpa1 and Ngn3 expression were analyzed by immunohistochemistry to determine whether the establishment of different pancreatic cell lineages within MPCs was affected in double-mutant mice. As expected, cells stained for Cpa1 were observed at the tips of the pancreatic epithelium, and Ngn3+ cells were located in the epithelial trunk in control, Gata4flox/flox;Gata6flox/+;Pdx1-Cre, and Gata4flox/+;Gata6flox/flox;Pdx1-Cre embryos (Figure 4, A–C, and E–G). In stark contrast, the pancreatic epithelium of the double-mutant embryos lacked clear tip-trunk structures, and Cpa1+ cells were totally absent in Gata4/Gata6 mutant pancreatic epithelium (Figure 4D). Indeed, no clear acinar morphology could be observed in double-mutant pancreas by H&E staining (Figure 3D). Endocrine commitment was affected as well in double-mutant embryos, as only a low number of _Ngn3_-expressing cells were found in the epithelial trunk (Figure 4H). In concordance with this observation, the number of cells expressing the transcription factor Nkx2.2, an endocrine progenitor marker regulated by Ngn3 (24), was also dramatically decreased in the double-mutant pancreata compared with littermate embryos (Figure 4, I and L). Thus, differentiation of pancreatic multipotent progenitor cells toward acinar and endocrine lineages requires GATA4 and GATA6 function.
Endocrine and acinar differentiation are compromised in Gata4/Gata6 double-mutant embryos. The enzyme Carboxipeptidase A1 (Cpa1) is expressed in the multipotent progenitor cell population located at the tip of the E13.5 branching epithelium in control, Gata4flox/flox;Gata6flox/+;Pdx1-Cre, and Gata4flox/+;Gata6flox/flox;Pdx1-Cre mice (A–C, arrowheads). In stark contrast, Cpa1+ cells were not detected in the double-mutant pancreatic epithelium (D). The proendocrine markers Ngn3 (E–G) and Nk2.2 (I–K) were mainly expressed in the epithelial trunk of control, Gata4flox/flox;Gata6flox/+;Pdx1-Cre, and Gata4flox/+;Gata6flox/flox;Pdx1-Cre embryos at E13.5. On the contrary, endocrine differentiation is disrupted in the double-mutant embryos as the number of cells expressing Ngn3 (H) and Nkx2.2 (L) are reduced compared with littermate embryos. The pancreatic epithelium is outlined in white in H and L. Counterstaining with DAPI was performed to reveal nuclei. Scale bars: 50 μm.
GATA4 and GATA6 are required to maintain the pancreatic progenitor cell pool. Exocrine and endocrine progenitors failed to form properly in Gata4/Gata6 mutant mice, raising the question of whether MPC formation and/or identity might be affected as well. To test this hypothesis, we decided to analyze the expression of several transcription factors that define MPC identity, including Pdx1, Ptf1a, Sox9, and Nkx6.1, in double-mutant embryos at E13.5 (22, 23, 25, 26). MPCs of control and Gata4flox/+;Gata6flox/flox;Pdx1-Cre mice displayed normal distribution and expression levels of progenitor markers (Figure 5, A, C, E, G, I, K, M, and O). In contrast, the numbers of _Pdx1_-, _Ptf1a_-, _Sox9_-, and _Nkx6.1_-expressing cells in double-mutant embryos were dramatically reduced compared with control and Gata4flox/+;Gata6flox/flox;Pdx1-Cre littermate embryos (Figure 5, D, H, L, and P). The decrease of _Pdx1_- and _Ptf1a_-expressing cells was particularly remarkable, as only a few positive cells could be observed in the epithelium of the double-mutant embryos (Figure 5, D and H). The reduced pool of cells expressing MPC markers in double-mutant embryos led to an overall decrease in the expression of progenitor markers (Figure 5Q). A decrease in the number of cells expressing Pdx1, Ptf1a, and Nkx6.1 was also found in Gata4flox/flox;Gata6flox/+;Pdx1-Cre, consistent with these mice displaying defects in pancreas formation at birth (Figure 5, B, F, J, and N). These results indicate that GATA4 and GATA6 activity are required to maintain normal numbers of MPCs during pancreatic development.
Reduced number of MPCs in Gata4/Gata6 double-mutant mice. Control embryos show strong expression of all multipotent pancreatic progenitor markers, Pdx1, Ptf1a, Sox9, and Nkx6.1 at E13.5 (A, E, I, and M, respectively). Similarly, Gata4flox/+;Gata6flox/flox;Pdx1-Cre embryos display normal distribution and expression levels (C, G, K, and O). However, Gata4flox/flox;Gata6flox/+;Pdx1-Cre mice show a significant decrease of cell numbers expressing Pdx1, Ptf1, and Nkx6.1 (B, F, and N) and a moderate reduction in the number of Sox9+ cells (J). The reduced number of cells expressing all pancreatic progenitor markers is even more dramatic in the double mutant (D, H, L, and P). (Q) qPCR analysis of multipotent pancreatic progenitor markers in E13.5 pancreata. *P < 0.02; **P < 0.001. Scale bars: 50 μm.
GATA4 and GATA6 bind to 2 conserved GATA sites in the Pdx1 promoter region. The impairment in pancreatic epithelial growth and cell differentiation observed in Gata4/Gata6 mutant mice is reminiscent of that in Pdx1- and _Ptf1a-_null mice (25, 27). This observation, in combination with the dramatic reduction in the number of cells expressing both Pdx1 and Ptf1a in the absence of GATA4 and GATA6 activity, prompted us to investigate whether GATA factors might regulate Ptf1a and Pdx1 expression in pancreatic progenitors. Previous studies have identified a 5′ distal enhancer region that controls the expression of Ptf1a in both dorsal and ventral pancreatic buds as early as E10.5 (28). Our bioinformatics analyses did not identify any conserved GATA site within these regulatory sequences, suggesting that the activity of this enhancer during pancreatic development might be independent of GATA factors. In spite of this, we cannot discard a role for GATA factors in the transcriptional regulation of Ptf1a through different regulatory sequences. Indeed, the relatively nonconserved proximal promoter contains several GATA sites that were able to bind recombinant GATA4 and GATA6 proteins (Supplemental Figure 6). However, this region has only so far been associated with the transcriptional control of Ptf1a at later stages of pancreas development (28).
The transcriptional regulation of Pdx1 has been extensively studied, and several regulatory sequences or enhancers in the 5′ conserved region of the Pdx1 promoter have been identified (29, 30). These enhancers, known as areas I, II, III, and IV, are bound by transcription factors, including HNF1a, Foxa2, Foxa1, HNF6, Pax6, MafA, and Ptf1, to direct Pdx1 expression in pancreatic progenitors and/or in adult β cells (29, 31–37). Areas I and II direct Pdx1 expression in pancreatic endocrine cells, but not exocrine cells, whereas the conserved area III controls the expression of Pdx1 in pancreatic progenitor cells (37). With this consideration, we analyzed the conserved enhancer areas in the Pdx1 locus for candidate GATA sites. Interestingly, 2 perfect and conserved candidate GATA-binding sites were found in area III of the previously identified Pdx1 enhancer (Figure 6A). To determine whether these 2 putative sites, called G1 and G2, represent bona fide sites for GATA factors in the Pdx1 enhancer, we assessed GATA4 and GATA6 protein binding (Figure 6B). First, we performed EMSA using recombinant proteins. Recombinant GATA4 protein bound efficiently to labeled oligos encompassing the Pdx1 G1 and Pdx1 G2 GATA sites (Figure 6B, lanes 2 and 8). The binding to G1 and G2 sites was specific. as it was competed off by excess, unlabeled self probe (Figure 6B, lanes 3 and 9) and by excess, unlabeled GATA control probe (Figure 6B, lanes 4 and 10). In contrast, binding of GATA4 to G1 and G2 sites was not competed off by unlabeled mutant versions of the G1 and G2 sites, G1m and G2m (Figure 6, A and B, lanes 5 and 11), or GATA mutant control site (Figure 6B, lanes 6 and 12). Similarly to GATA4, recombinant GATA6 protein strongly bound to both G1 and G2 sites of the Pdx1 enhancer (Figure 6B, lanes 14 and 20). This binding was specific, as it was competed off by an excess of unlabeled self probe (Figure 6B, lanes 15 and 21), but not by an excess of unlabeled mutant probes (Figure 6B, lanes 17 and 23). In addition, we found 2 conserved GATA sites in conserved area I that were able to bind GATA4 and GATA6 in EMSA experiments (Supplemental Figure 7), suggesting that other GATA sites may also participate in the transcriptional regulation of Pdx1 at later stages of pancreas development.
GATA4 and GATA6 bind to the Pdx1 conserved area III in vitro and in pancreatic cell line. (A) Highly conserved region in the _cis_-regulatory area III of Pdx1. Two conserved GATA sites, as revealed by bioinformatics analysis, are shown in blue boxes. Numbers indicate the position of the GATA sites relative to the Pdx1 translational start site. Point mutations introduced into GATA sites, G1m and G2m, are indicated in red lowercase. Asterisks denote nucleotides that have been perfectly conserved between mouse and human. (B) Recombinant GATA4 and GATA6 proteins are able to bind to G1 and G2 GATA sites of the Pdx1 enhancer as shown by EMSA. Competition experiments were performed by adding excess unlabeled probes of G1, G2, or control (denoted as c in competitor row) GATA sites, and the corresponding mutant versions (G1m, G2m, or cm) to the binding reaction. (C) ChIP experiments performed in mouse pancreatic ductal cells (mPAC cells) using specific GATA4 and GATA6 antibodies (lanes 2, 3, respectively) and nonspecific anti-IgG (lane 4) show that anti-GATA4 and anti-GATA6 antibodies are able to immunoprecipitate the GATA sites of the Pdx1 conserved area III, but not nonspecific genomic regions. Lane 1 contains PCR products from input DNA (Inp) amplified prior to immunoprecipitation. Sizes of the PCR products in bp are shown on the right. (D) A WT Pdx1 promoter-luciferase construct (pGL3-_Pdx1_-WT) is significantly activated by endogenous factors present in mPAC cells compared with the activity of the empty reporter pGL3 vector. Mutations in the GATA sites of Pdx1 (pGL3-_Pdx1_-mut) significantly attenuate the luciferase activity. *P = 0.002; **P = 0.005.
To investigate whether GATA4 and GATA6 could bind to conserved area III of the Pdx1 _cis_-regulatory sequences in pancreatic cells, we performed a ChIP assay in a murine pancreatic ductal cell line, mPAC, that endogenously expresses GATA4, GATA6, and PDX1 (data not shown). Anti-GATA4 antibody was able to specifically immunoprecipitate DNA fragments encompassing the G1 and G2 sites of the Pdx1 conserved area III, but not nonspecific genomic regions (Figure 6C, lane 2). Incubation of sheared DNA with anti-GATA6 also resulted in specific immunoprecipitation of DNA fragments containing the GATA sites of the Pdx1 enhancer (Figure 6C, lane 3). The addition of nonspecific IgG to the reaction did not result in the immunoprecipitation of the Pdx1 conserved area III or nonspecific genomic regions (Figure 6C, lane 4), confirming the specificity of the immunoprecipitation by the corresponding GATA antibodies. These results demonstrate that endogenous GATA4 and GATA6 bind to the GATA sites in the endogenous Pdx1 enhancer.
The observation that GATA4 and GATA6 bound to the Pdx1 conserved area III by EMSA and in mPAC pancreatic cells suggests that GATA4 and GATA6 transcription factors are required to transcriptionally regulate Pdx1 expression. Therefore, we examined the requirement of the G1 and G2 sites to activate the Pdx1 promoter in mPAC cells. A reporter construct containing the conserved area III of Pdx1 was fused to the luciferase gene to generate the pGL3-_Pdx1_-WT reporter plasmid. Likewise, the 2 conserved GATA sites were mutated to generate the pGL3-_Pdx1_-mut reporter plasmid. The luciferase activity of these plasmids was measured in transiently transfected mPAC cells. mPAC cells transfected with pGL3-_Pdx1_-WT display significantly higher luciferase activity compared with mPAC cells transfected with the parent reporter vector pGL3-basic, indicating that the Pdx1 enhancer region is activated by endogenous factors in mPAC cells (Figure 6D). Activation of the Pdx1 promoter significantly decreased when the 2 GATA sites in area III were mutated (Figure 6D). Taken together, these results demonstrate that GATA factors are required for the transcriptional regulation of Pdx1 in at least some pancreatic cell lines in vitro, suggesting that they may function through these sites in the MPCs.
Pdx1 transcriptional activity depends on GATA sites in vivo. Previous studies have reported that a region 4.6 kb upstream of the Pdx1 translational start site containing conserved areas I, II, and III is sufficient to faithfully recapitulate endogenous Pdx1 expression throughout development in transgenic mice (38, 39). We generated transgenic mice harboring these upstream sequences of the Pdx1 promoter fused to the _lac_Z reporter gene, which we refer to as _Pdx1-_WT-_lac_Z. To test the requirement of the GATA sites for the Pdx1 enhancer activity in vivo, we introduced mutations in the GATA sites of Pdx1 area III to generate the _Pdx1_-mut-_lac_Z transgene. The introduced mutations were identical to those used in EMSA analyses that abolished the binding of GATA4 and GATA6 recombinant proteins (Figure 6, A and B). We generated 3 stable transgenic lines for the _Pdx1_-WT-_lac_Z construct and 2 stable transgenic lines for the _Pdx1_-mut-_lac_Z construct. The analysis of the _lac_Z expression by X-gal staining was performed in at least 6 transgenic embryos from each founder at different developmental stages. We observed that the _lac_Z expression pattern of the transgenic embryos was temporally and spatially very consistent among the founders from each construct. As expected, the 4.6 kb _cis_-regulatory region was sufficient to recapitulate the endogenous expression of Pdx1 in transgenic embryos from early embryonic stages. At E9.5, β-gal activity was observed in both the dorsal and ventral pancreatic buds in _Pdx1_-WT-_lac_Z transgenic embryos (Figure 7, A and B). In contrast, no X-gal staining was observed in any _Pdx1_-mut-_lac_Z transgenic embryo (Figure 7, E and F). By E10.0, very robust X-gal staining was observed in the pancreatic foregut of _Pdx1_-WT-_lac_Z transgenic embryos (Figure 7, C and D), while only a few _lac_Z-positive cells were detected in the pancreatic epithelium of _Pdx1_-mut-_lac_Z embryos (Figure 7, G and H). At midgestation, around E13.5, the _Pdx1_-WT-_lac_Z transgene continued to be active in the branching epithelial tree of the pancreas (Figure 7I). The β-gal expression directed by the _Pdx1_-WT-_lac_Z transgene completely mirrored the endogenous expression of Pdx1 (Figure 7, J–L). In sharp contrast, mutations in Pdx1 area III GATA sites resulted in a dramatic decrease in the _Pdx1_-mut-_lac_Z transgene activity in the pancreatic epithelium (Figure 7N). Immunohistochemical analysis confirmed the reduction of β-gal expression in _Pdx1_-mut-_lac_Z embryos (Figure 7, O–Q). At later stages of development, the activity of the _Pdx1_-mut-_lac_Z transgene was even more diminished compared with the activity of the _Pdx1_-WT-_lac_Z transgene (Figure 7M). Furthermore, X-gal staining was observed only in scattered cells of _Pdx1_-mut-_lac_Z pancreatic embryos (Figure 7R). These results indicate that GATA sites are required for Pdx1 enhancer activity during pancreas development and place GATA factors upstream of Pdx1 in the regulatory network controlling pancreas formation.
Conserved GATA sites in area III are required for Pdx1 enhancer activation in vivo. Whole-mount (A, C, E, and G) and transversal sections (B, D, F, and H) of representative _Pdx1_-WT-_lac_Z and _Pdx1_-mut-_lac_Z transgenic embryos stained with X-gal. β-gal activity in both dorsal and ventral pancreatic buds is first observed in _Pdx1_-WT-_lac_Z embryos at E9.5 (A and B). In contrast, no X-gal staining is observed in the pancreatic buds of _Pdx1_-mut-_lac_Z embryos (E and F, asterisk). By E10.0, β-gal activity in the _Pdx1_-mut-_lac_Z embryos (square in G, arrowhead in H) is dramatically reduced compared with _Pdx1_-WT-_lac_Z embryos (C and D). At E13.5, all the pancreatic epithelial cells in _Pdx1_-WT-_lac_Z embryos show homogeneous β-gal activity (I). Similarly, strong X-gal staining is observed in most of the pancreatic cells of _Pdx1_-WT-_lac_Z embryos at E17.5 (M). In contrast, _lac_Z expression is markedly reduced in the pancreatic epithelium of _Pdx1_-mut-_lac_Z embryos both at E13.5 (arrows in N) and at E17.5 (R). Immunofluorescence staining at E13.5 reveals a complete overlapping expression pattern of Pdx1 and β-gal in _Pdx1_-WT-_lac_Z pancreas (J–L), while only a fraction of Pdx1+ cells express β-gal in _Pdx1_-mut-_lac_Z embryos (O–Q). dp, dorsal pancreas; vp, ventral pancreas. Scale bars: 50 μm.