Gata6 and Gata4 are expressed in the embryonic mouse pancreas
To begin to ascertain whether Gata factors are involved in the development of the embryonic mouse pancreas, we first determined which of the known mouse Gata family members are expressed during pancreas development. RNA in situ hybridization using RNA probes specific for each of the six Gata transcription factors was performed on frozen sections of e12.5 embryos. Our analysis shows that mRNA for Gata4 and Gata6 is expressed in the developing pancreas (). mRNA could not be detected for Gata1, Gata2, Gata3, or Gata5 in the pancreas, although we detected strong expression of these genes in other organs including the kidney, liver, and brain ( and data not shown). We performed RNA in situ hybridization at additional developmental stages between e9.5 and e16.5. This time period encompasses both initial pancreas specification (~ e9.5) and the secondary transition (~ e12.5–e15.5), a stage of extensive expansion and differentiation of the pancreatic epithelium. Gata4 and Gata6 expression is maintained throughout these stages of pancreas development (– and data not shown). To confirm the RNA in situ hybridization results and rule out the possibility that Gata1, Gata2, Gata3, and Gata5 are expressed at levels below detection by non-radioactive RNA in situ hybridization, we performed RT-PCR on an e12.5 pancreas cDNA library using gene-specific primers for each of the six Gata factor family members. In addition, we assessed expression data previously generated from Affymetrix microarray analyses of cDNA generated from e12.5 pancreas RNA (Prado et al., 2004
). Consistent with the RNA in situ hybridization expression data, Gata4 and Gata6 mRNA was strongly detected in the e12.5 pancreas, but Gata1, Gata2, Gata3 or Gata5 mRNA was not present (data not shown).
Fig. 1 In situ hybridization of e12.5 tissue using antisense RNA probes for each Gata transcription factor family member reveals that only Gata4 and Gata6 are expressed in the embryonic mouse pancreas (d, f). Gata1, Gata2, Gata3, and Gata5 are not expressed (more ...)
Fig. 3 Gata6 mRNA expression analysis. At e9.5, Gata6 mRNA (blue) is expressed globally throughout the pancreatic epithelium, along with Pdx1 (brown, b) and glucagon (green, a); a and b are adjacent sections. By e14.5, Gata6 mRNA is highly expressed in the pancreatic (more ...)
Gata4 and Gata6 expression becomes restricted to distinct pancreatic lineages at later developmental stages
To further characterize the spatial and temporal expression of Gata4 and Gata6 within the developing pancreas, we compared expression of Gata4 mRNA, Gata4 protein, and Gata6 mRNA with the expression of pancreatic cell type-specific markers at different embryonic stages. We were unable to analyze Gata6 protein expression because a reliable Gata6 antibody is not available. In contrast to previous studies (Ketola et al., 2004
), our analysis shows that both Gata4 and Gata6 are expressed throughout the pancreatic epithelium at e9.5, when the pancreas first begins to form. Gata4 mRNA is expressed from the onset of dorsal bud evagination and overlaps extensively with Pdx1 (). In addition, immunofluorescence studies demonstrate that Gata4 protein is detected throughout the pancreatic epithelium at e9.5 but is absent from glucagon-expressing cells (), suggesting that Gata4 is excluded from endocrine cells at the earliest stage of pancreas development. Gata4 mRNA is observed in most pancreatic epithelial cells by e11.5 (data not shown), but expression progressively moves toward the tips of the branching ducts (i.e., the presumptive acini) as development proceeds (, inset). By e16.5, Gata4 protein expression is restricted to the exocrine pancreas where it is observed in the nucleus of cells expressing amylase and carboxypeptidase A ( and data not shown). However, Gata4 is excluded from both the ducts and developing islets, and is not found in cells expressing glucagon ( and data not shown). At all stages of embryonic pancreas development, the pattern of Gata4 protein expression is consistent with the pattern of Gata4 mRNA expression and Gata4 protein is localized to the nucleus. By e18.5, a subset of acini continues to express Gata4 (data not shown). Our in vivo results are supported by immunofluorescence analysis in the AR42J exocrine pancreas cell line, confirming that Gata4 is present in cells also expressing the exocrine enzyme amylase (). Interestingly, in the adult pancreas, Gata4 is expressed in a subset of α and β cells of the adult islet, but is no longer detectable in the exocrine tissue ( and data not shown). Nuclear and non-nuclear expression of Gata4 can be detected in the adult islet.
Fig. 2 Gata4 mRNA and protein expression analysis. At e9.5, Gata4 RNA (blue, a) and protein (green, b) are expressed globally in the pancreatic epithelium along with Pdx1 (brown; a) and glucagon (red; b). By e16.5, Gata4 expression becomes restricted to the (more ...)
Similar to Gata4, Gata6 mRNA is expressed globally in the e9.5 dorsal pancreatic bud and is co-expressed with Pdx1 (). In contrast to Gata4, Gata6 mRNA is expressed in cells also expressing glucagon at this stage (). As development proceeds, Gata6 expression becomes restricted to cells of the endocrine pancreas and the ductal epithelium (). By e12.5, Gata6 is expressed throughout the differentiating pancreatic epithelium in regions similar to endocrine transcription factor Nkx2.2 () and a marker of endocrine progenitor cells, Ngn3 (). During the secondary transition, glucagon is coexpressed in some, but not all cells that express Gata6 (). Between e14.5 and e15.5, Gata6 mRNA is expressed strongly in the pancreatic epithelial ducts and becomes down-regulated in cells differentiating into acini ().
Nkx2.2 physically interacts with Gata6, but not Gata4
Gata4 and Gata6 are known to interact with tissue-specific transcription factors to regulate genes expressed during the differentiation of numerous embryonic tissues. We originally pursued our analysis of the Gata factors in the pancreas due to the close link between Gata factors and Nkx2 factors in the development of several different organs (Stennard et al., 2003
). Early in pancreas development, expression of Gata4 and Gata6 overlaps with Nkx2.2, a transcription factor known to be essential for normal pancreas development (Prado et al., 2004
; Sussel et al., 1998
). Later in development, only Gata6 is coexpressed with Nkx2.2 in endocrine cells of the islet. To determine whether Gata4 and/or Gata6 are able to interact with Nkx2.2, we performed a series of yeast two-hybrid assays in which a derivative of Nkx2.2 that lacks its two activation domains (see Materials and methods) was fused to the GAL4 DNA binding domain and used as bait (BD-Nkx2.2). Chimeric proteins containing full-length Gata4 and Gata6 fused to the GAL4 activation domain (AD) were tested for their ability to co-activate GAL4 UAS-lacZ with BD-Nkx2.2. As shown in Supplemental Fig. 1a
, BD-Nkx2.2 alone or in combination with AD-Gata4 is incapable of activating GAL4 UAS-lacZ, whereas coexpression of BD-Nkx2.2 with AD-Gata6 results in the activation of lacZ expression. Furthermore, coexpression of BD-Nkx2.2 with a C-terminal truncation of AD-Gata6 (pACT:Gata6-N-terminal), which deletes the zinc finger domain and the C-terminal extension required for interaction between Gata4 and Nkx2.5 (Durocher et al., 1997
), prevented the interaction and failed to activate lacZ expression. These results suggest that Gata6, but not Gata4, interacts with Nkx2.2.
We confirmed the physical association of Gata6 and Nkx2.2 using in vitro pull down assays and co-immunoprecipitation assays of mammalian cell extracts. Full-length Nkx2.2 was fused with maltose binding protein (MBP) to generate MBP-Nkx2.2. Immobilized MBP-Nkx2.2 retained 35
S-labeled Gata6, whereas immobilized MBP alone did not (Supplemental Fig. 1b
). Co-immunoprecipitation was performed on whole cell extracts from COS-7 cells transfected with Gata6 and myc-tagged Nkx2.2, individually or in combination. Immunoprecipitation of extracts using α-Gata6 antibody or α-myc antibody followed by Western blotting for Nkx2.2 confirmed that Nkx2.2 interacts with Gata6 in vivo (Supplementary Fig. 1c
). These results suggest that Nkx2.2 and Gata6 may cooperate in the regulation of early pancreas or islet specific gene expression, similar to Nkx2.5 and Gata4 cooperativity in heart-specific gene expression (Durocher et al., 1997
; Sepulveda et al., 1998
). We are currently attempting to identify direct targets of Nkx2.2 and Gata6 in the islet.
Production of Gata dominant repressor transient transgenic mice
The dynamic expression patterns of Gata4 and Gata6 in the pancreas, combined with the physical interaction between Nkx2.2 and Gata6, suggest that Gata4 and Gata6 may have overlapping functions during the initial specification of the embryonic mouse pancreas, but may be uniquely involved in the differentiation of endocrine cell types (Gata6) or exocrine cell types (Gata4). To investigate whether Gata factors play a functional role in the developing mouse pancreas, we utilized a dominant repressor transient transgenic mouse system modeled after experiments assessing the role of Gata6 in lung development (Koutsourakis et al., 2001
; Liu et al., 2002b
; Yang et al., 2002
). As illustrated in Supplemental Fig. 2
and described in Materials and methods, we designed two Gata6 dominant repressor transgenes (Pdx1:G6FLEnR and Pdx1:G6DBDEnR), one Gata4 dominant repressor transgene (Pdx1:G4FLEnR), and one control transgene (Pdx1:NLSEnR). Based on the previous studies, we predicted that fusion of the Engrailed repression domain to full-length Gata4 or Gata6 would repress genes normally activated by Gata4 and Gata6, respectively. Further, since the DNA binding domains of Gata4 and Gata6 are highly conserved at the DNA and protein levels, we hypothesized that fusion of the Engrailed repressor domain to the Gata6 DNA binding domain would result in repression of all Gata4 and Gata6 target genes. The control transgene, Pdx1:NLSEnR, was not expected to affect pancreas development, similar to the findings of Yang et al. (Yang et al., 2002
) who expressed an EnR-only control transgene in the lung epithelium without phenotypic consequences. Expression of each fusion protein is driven by the 4.5 kb Pdx1 promoter element, which is active starting at e8 in the developing endoderm and becomes restricted to the developing pancreas around e9 (Gannon et al., 2000
; Norgaard et al., 2003
). The Pdx1 promoter is well characterized, recapitulates endogenous Pdx1 expression which is largely pancreas-specific, and mirrors the early pancreas expression pattern of Gata4 and Gata6 (Apelqvist et al., 1997
; Li and Edlund, 2001
; Stoffers et al., 1999
). While later stage expression of Pdx1 is predominantly restricted to the endocrine pancreas (similar to Gata6), Pdx1 is also expressed at lower levels in exocrine cells and is required for proper differentiation of these cell types (Hale et al., 2005
Transient transgenic mice carrying each of the Gata derivatives or the control were generated after several attempts to establish stable mouse lines proved unsuccessful. We chose to analyze transient transgenic embryos due to the possibility that the Gata transgenes were significantly disrupting pancreas development, resulting in neonatal lethality. Pregnant mice were sacrificed and embryos were harvested between 15.5 and 17.5 days following pronuclear injection. For each of the four transgenes, phenotypic analysis was performed on transgenic and wildtype littermates. For this study, we generated twenty-five embryos which carried a transgene and assessed transgene expression by RNA in situ analysis on sectioned pancreatic tissue. Of these, two Pdx1:G6FLEnR, one Pdx1:G6DBDEnR, one Pdx1:NLSEnR (control), and eight Pdx1:G4FLEnR embryos expressed the transgene. Three additional Pdx1: G6FLEnR embryos exhibited a dramatic “absence of pancreas” phenotype (see below); assessment of transgene expression in the absent tissue was not possible. Injected embryos that did not contain the transgene were used as littermate controls.
Analysis and classification of Gata dominant repressor transient transgenics
To assess whether the Gata dominant repressor transgenes affect development of the embryonic mouse pancreas, we first performed histological analysis of the late stage embryos. Frozen sections from transgenic and wildtype embryos were stained with hematoxylin and eosin (H&E). As summarized in , we observed three distinct pancreatic phenotypes in the transgenic embryos: absence of pancreas (hereafter referred to as Class I); severely disrupted pancreas morphology (Class II); and little to no observable phenotype (Class III). All transgenic embryos exhibited overall normal morphology of non-pancreatic tissue. Three Pdx1:G6FLEnR embryos exhibited a Class I defect, completely lacking a pancreas or any recognizable pancreatic rudiment. Two Pdx1:G6FLEnR embryos and the Pdx1: G6DBDEnR embryo had Class II defects. A Class III phenotype was not observed for any Gata6 transgenic embryo. Interestingly, only one Pdx1:G4FLEnR transgenic embryo displayed disrupted islet morphology, while seven Pdx1:G4FLEnR transgenic embryos had no observable phenotype despite strong transgene expression (). The Pdx1:NLSEnR embryo had normal pancreas morphology and no apparent phenotype ( and data not shown).
Summary of transgenic phenotypes
Fig. 7 Analysis of Gata4 transgenic embryos. H&E staining (a, b, d, e). RNA in situ analysis of transgene expression (c, f). One Pdx1:Gata4FLEnR transgenic embryo exhibited a Class II phenotype (b) in comparison to its e16.5 wildtype littermate (a). (more ...)
Fig. 4 Histological analysis of Gata6 transgenic mice. H&E stained sections of an e17.5 Pdx1:G6FLEnR wildtype littermate (a) and two e17.5 Pdx1:G6FLEnR transgenic littermates with a Class I (no pancreas) phenotype (b, c). One embryo exhibited morphological (more ...)
The pancreas is absent in Class I Gata6 repressor transgenics
Following initial analysis and classification of the transgenic embryos, a more thorough investigation of the Pdx1:G6FLEnR and Pdx1:G6DBDEnR phenotypes was pursued. Strikingly, in three independent e17.5 Pdx1:G6FLEnR transgenic embryos, pancreas formation appeared to be completely disrupted (Class I phenotype; ). Although there was no morphological evidence of the pancreas, it was important to determine whether any pancreatic rudiments or cell types had formed, as seen in Pdx1 null mice (Offield et al., 1996
), or whether there were ectopic islet cells, as observed in Ptf1a null mice (Krapp et al., 1998
). Tissue sections throughout each of the three Class I Gata6 transgenics were analyzed with a panel of diagnostic pancreatic markers to ascertain whether pancreatic rudiments were present anywhere within these Pdx1:G6FLEnR embryos. Using immunohistochemical analysis, we were unable to detect expression of amylase, insulin, glucagon, somatostatin, PP, or ghrelin (data not shown). Furthermore, we were unable to detect pancreatic expression of the markers Pdx1 and Ptf1a (data not shown), although Pdx1 was still expressed in the duodenum and stomach (). In addition, we did not detect pancreatic expression of Ngn3, Nkx2.2, Nkx6.1, or Pax6, although these transcription factors were still expressed in non-pancreatic tissues (data not shown). Although we cannot rule out the possibility that the pancreas initially formed and subsequently degenerated, our analyses suggest that no pancreatic rudiments or cell types have formed in the Class I Pdx1:G6FLEnR embryos.
Pancreas development is disrupted in the second class of Gata6 repressor transgenics
In another subset of Pdx1:G6FLEnR and Pdx1:G6DBDEnR embryos we observed a second phenotypic class in which there was a severe loss of differentiated pancreas cell types although a small amount of pancreatic tissue remained (Class II phenotype; ). Class II transgenic embryos form both the dorsal and ventral pancreatic domains; however, there is less overall pancreatic tissue and islets are not properly formed (). In addition, fewer acini are present and some embryos display a disruption of acinar cell morphology that correlates with a reduction in enzyme expression (, arrow). Interestingly, a phenotypic difference was not observed between Class II Pdx1:G6FLEnR embryos and the Pdx1:G6DBDEnR embryo. In addition, the phenotype was similar in each of the Class II embryos and we did not observe a gradient of phenotypes between the Class I and Class II phenotypes, suggesting that Gata6 may function at two discrete stages in pancreatic development (see Discussion).
Fig. 5 Immunohistochemical staining for pancreatic hormones and enzymes. In comparison to the pancreas of the wildtype littermate (a, c, e, and g), Class II transgenic embryos have reduced numbers of cells expressing insulin (b, d, f, and h), glucagon (b), amylase (more ...)
To explore the possibility that the less severe phenotype associated with the Class II Gata6 transgenic mice is due to weaker or mosaic expression of the transgene, we performed in situ hybridization using an RNA probe for the Engrailed portion of the transgene in these embryos. Strong expression of the transgene was detected in differentiated regions of the pancreas in all Pdx1:G6FLEnR and Pdx1:G6DBDEnR embryos with a Class II phenotype (, and data not shown). Similar to the Class I embryos, transgene expression could no longer be detected in undifferentiated regions of the pancreas, regions that had differentiated into exocrine cells, or tissue that appeared more mesenchymal in structure (). It remains possible that timing or initiation of transgene expression may be affecting the phenotype; however, this cannot be explored in transient transgenic embryos.
Class II Gata6 repressor embryos have fewer differentiated cells
Histological analysis of the Class II Gata6 embryos suggested that few differentiated pancreatic structures had formed. To assess the degree to which cell differentiation was disrupted in these mice, we performed immunofluorescence to detect the presence and localization of exocrine enzymes and endocrine hormones. Exocrine tissue is clearly present; however, the relative numbers of acini clusters are reduced concomitant with a reduction in amylase expression (). Furthermore, there appear to be acini clusters that express reduced levels of amylase (, arrows). In addition, these embryos lack the well-formed islets normally seen at this stage of development, and have a significant reduction in the number of cells expressing insulin (), glucagon (), somatostatin (), and ghrelin (). The hormone-producing cells that are present tend to be scattered throughout the pancreas and around the ductal tissue. To quantify the loss of insulin and glucagon cells in the Class II embryos, we quantified insulin-positive and glucagon-positive cells from two independent Class II transgenic embryos as well as their wildtype littermates. This analysis revealed a 3- to 4-fold reduction in the number of insulin-expressing and glucagon-expressing cells in the Gata6 transgenic embryos (). Traditional morphometric analysis of the islet was not possible due to the severe disruption of general pancreas and islet morphology.
In comparison to wildtype littermates, Class II Gata6 transgenic embryos also display aberrant development of the ductal epithelium (; ). Ducts in a Class II embryonic pancreas are generally larger, greater in number, and morphologically abnormal compared to ducts in a typical e17.5 pancreas. To determine whether the ducts of Class II embryos were epithelial in nature, we performed immunofluorescence using a fluorescein-conjugated antibody against DBA-Lectin, a marker of pancreatic ductal epithelium. In the wildtype littermate, DBA-Lectin marks discrete areas of ductal epithelia that permeate throughout the pancreas (). By contrast, in the Class II embryos, DBA-Lectin delineates greatly enlarged ducts that are epithelial in nature (). Interestingly, DBA-Lectin is not expressed in all duct cells in these embryos (, arrow). In some cases, the epithelia of these oversized ducts are surrounded by a stratified cell type (, arrows) that express the mesenchymal marker, Vimentin (data not shown). In wildtype embryos, pancreatic mesenchyme is significantly reduced by e17.5.
Fig. 6 Analysis of aberrant duct development. H&E staining (a, b); immunofluorescence staining of insulin (red) and the ductal marker DBA-Lectin (green, c–e). In comparison to the wildtype littermate (a, c), Class II transgenic embryos display (more ...)
In Class II embryos, we also observe a significant reduction in the overall numbers of endocrine cells and a large proportion of the cells that remain are localized to the ducts (; ). Occasionally, cells that express the ductal marker DBA-Lectin also express hormones, such as insulin ( inset). Potentially, these cells represent an endocrine progenitor cell population. To determine whether Class II embryos maintain a larger pool of endocrine progenitors than wildtype littermates, we performed RNA in situ hybridization to analyze expression of Ngn3, a transient marker of all endocrine progenitor cells. By e17.5 in a wildtype embryo, the endocrine progenitor population has decreased to a low level and few cells express Ngn3 (, arrows). In comparison, Class II embryos show an apparent increase in the number of Ngn3-expressing endocrine progenitor cells, despite a clear reduction in pancreatic mass (). However, due to loss of overall pancreatic tissue and significant disruption to pancreas morphology in the Class II embryos, it is unclear whether this increase is statistically significant.
To further investigate the differentiation state of the pancreatic islet cell types in Class II Gata6 embryos, we analyzed expression of the transcription factors Pdx1, Nkx6.1, and Pax6. Pdx1 is normally expressed in endocrine progenitors along with mature β and δ cells (Guz et al., 1995
; Ohlsson et al., 1993
). Nkx6.1 is expressed in duct cells as well as terminally-differentiated β cells, while Pax6 is expressed in all mature endocrine cell types (Sander et al., 1997
; Sander et al., 2000
; St-Onge et al., 1997
). Consistent with the reduction in total pancreatic islet tissue, there is a reduction in the numbers of cells expressing these transcription factors in the transgenic embryos (Supplementary Figs. 3b, d, and f
). In addition, cells expressing these transcription factors in the Class II embryos are often localized to the periphery of the ducts (Supplementary Figs. 3d and f
). For example, the majority of Nkx6.1-expressing cells are within the ductal epithelium in Class II embryos (Supplementary Fig. 3d
, inset,). Since Gata factors are known to co-regulate one another, we also analyzed Gata4 expression in Class II Gata6 embryos by performing immunofluorescence using an α-Gata4 antibody. Although the total number of exocrine cells is reduced, Gata4 expression is maintained in the nucleus of these cells (data not shown).
The loss of pancreatic mass in the Class II Gata6 transgenics could be due to a block in cell differentiation, a reduction in cell proliferation, and/or an increase in cell death. To determine if there are any abnormalities in cell death or proliferation, we compared levels of apoptosis and cellular proliferation between transgenic and wildtype embryos. We observed no change in the level of cell death in Class II or Class I embryos, as determined by TUNEL assay at e17.5 (data not shown); however, these embryos have such an extreme loss of pancreatic tissue by e17.5 that it is possible there is substantial cell death at an earlier period, which is difficult to assess in the transient transgenic system. We also performed immunohistochemical analysis using an antibody against phospho-Histone H3, which marks cells in M phase, and determined that there is no significant difference in the proliferative state of pancreatic cell types between the wildtype and Class II transgenic littermates (data not shown). The absence of obvious proliferative defects or increased cell death at e17.5 supports our previous findings that suggest Pdx1:G6FLEnR and Pdx1:G6DBDEnR may cause a disruption of pancreatic cell differentiation, however, we cannot rule out that increased cell death or decreased proliferation occurred at time points prior to our analysis at e17.5.
The majority of Pdx1:G4FLEnR transgenic embryos do not display disrupted pancreas morphology or development
Both Gata4 and Gata6 are expressed globally throughout the pancreatic epithelium at the earliest stages of pancreas development, but their expression becomes restricted to mutually exclusive domains at later stages of development. Consequently, it was possible the Pdx1:G4FLEnR transgene would produce phenotypes that affect early pancreas development as well as differentiation of exocrine cell types. We analyzed eight Gata4 transgenic embryos along with their wildtype littermates. RNA in situ
hybridization to detect the Engrailed portion of the transgene demonstrated that all eight Gata4 transgenic embryos strongly express the Gata4 dominant repressor transgene (, and data not shown). Surprisingly, in contrast to the Gata6 transgenics, most of the Gata4 transgenic embryos appear to be phenotypically normal () with no obvious defects in endocrine hormone, exocrine enzyme, and pancreatic transcription factor expression (data not shown). To ensure that the G4FLEnR construct was functional, we assessed its ability to interfere with Gata4 transcriptional activation. Since Gata4 has been shown to be essential for the activation of the a myosin heavy chain gene (αMyHC) (Molkentin et al., 1994
; Kathiriya et al., 2004
) we determined that the G4FLEnR fusion protein is able to effectively block Gata4-dependent activation of an αMyHC-luciferase reporter in HeLa cells (Supplementary Fig. 4
). This suggests the G4FLEnR fusion protein is functional in an in vitro reporter assay, however we cannot rule out that the Gata4 dominant repressor transgene is less effective in the pancreas in an in vivo context. One Gata4 transgenic embryo exhibited a phenotype similar to the Class II Gata6 transgenic embryos: there is a reduction in the mass of exocrine and endocrine tissue and a corresponding loss of hormone and enzyme expression ( and data not shown). This phenotype may be due to off-target effects of the G4FLEnR transgene. Conditional knockout analysis of Gata4 in the pancreas will help resolve this issue.