Multiple, distant Gata2 enhancers specify temporally and tissue-specific patterning in the developing urogenital system

Abstract

Transcription factor GATA-2 is expressed in a complex temporally and tissue-specific pattern within the developing embryo. Loss-of-function studies in the mouse showed that GATA-2 activity is first required during very early hematopoiesis. We subsequently showed that a 271-kbp yeast artificial chromosome (YAC) transgene could fully complement the loss of Gata2 hematopoietic function but that these YAC-rescued Gata2 null mutant mice die perinatally due to defective urogenital development. The rescuing YAC did not display appropriate urogenital expression of Gata2, implying the existence of a urogenital-specific enhancer(s) lying outside the boundaries of this transgene. Here we outline a coupled general strategy for regulatory sequence discovery, linking bioinformatics to functional genomics based on the bacterial artificial chromosome (BAC) libraries used to generate the mouse genome sequence. Exploiting this strategy, we screened >1 Mbp of genomic DNA surrounding Gata2 for urogenital enhancer activity. We found that the spatially and tissue-specific functions for Gata2 in the developing urogenital system are conferred by at least three separate regionally and temporally specific urogenital enhancer elements, two of which reside far 3' to the Gata2 structural gene. Including the additional enhancers that were discovered using this strategy (called BAC trap) extends the functional realm of the Gata2 locus to greater than 1 Mbp.

FIG. 1.

Strategy to identify Gata2 urogenital enhancer. (A) Diagram of Gata2 in its genomic position on mouse chromosome 6. Constructs referred to in the study are diagrammed. A list of end points of the BACs used in the study is available in Table 1. Other genes in the locus are denoted by black boxes. (B) Schematic diagram of the targeting cassette used to generate modified BACs. (C) Confirmation of modification was achieved using restriction digest fingerprinting with EcoRI (left panel). Diagnostic changes are indicated: R, recombinant vector band; P, parental vector band; T, targeting cassette band; Neo, neomycin resistance gene. Lanes 1 and 2, parental BAC; lane 3, recombinant BAC with neomycin resistance; lanes 3 to 7, recombinant BAC with Neo removed. These changes were verified by Southern analysis (right panel) using a PstI digest and Neo probe (top) and HindIII digest and LacZ probe (bottom). Expected sizes are indicated. Clones which have had the neomycin resistance gene removed by Flp induction are marked. Lane 8, parental BAC; lanes 9 and 11, recombinant BAC with neomycin resistance; lanes 10 and 12, recombinant BAC with Neo removed; lane 13, targeting vector positive control.

FIG. 2.

Identification of BAC containing urogenital enhancer activity. (A) BACs were modified as described above and injected into fertilized oocytes. Founder embryos were collected on e12.5 and analyzed by X-Gal staining. Embryos bearing the 333I12-modified BAC show strong expression in the urogenital system (arrow). None of the other modified BACs show this activity. Other sites of expression are present in other BACs, indicating either the presence of other enhancers or ectopic sites. Pictures shown are representative founder embryos. The expression data are presented in tabular form in Table 2.

FIG. 3.

Normal expression of Gata2 in the development of the mouse urogenital system. Immunohistochemistry using an anti-GFP antibody on frozen sections of mice bearing a Gata2/GFP knock-in allele. (A) Saggital section of e10.5 embryo showing GFP expression in the mesenchyme surrounding the nephric duct (ND) and mesonephric mesenchyme. There is no staining in the epithelium of the nephric duct per se, nor is there staining in the mesonephric tubular (mt) epithelium. (B) Saggital section of e10.5 embryo showing staining in the cloaca (cl), which is partitioned into the urogenital sinus and hindgut. (C) Transverse section of e12.5 embryo showing staining in the mesenchyme surrounding the nephric and paramesonephric ducts, as well as in the ureteric bud (ub). The epithelium of the nephric duct does not appear to stain, but both the epithelium and the mesenchyme surrounding the ureteric bud show strong GFP expression. There is also no staining in the mesonephric mesenchyme (mm). (D) Expression of GFP is widespread in the region surrounding the urogenital sinus (us). There is staining in both the mesenchyme and epithelium of the walls of the urogenital sinus, and surrounding the areas where the ureter meets the prospective bladder. (E) Whole mount GFP fluorescence of P14 male mouse, showing intense GFP signal from the derivatives of the nephric duct, the vas deferens (vd), and epididymis. k, kidney; t, testis; bl, bladder; u, ureter. There is no signal from the testes (t). The ureter and bladder both show strong GFP expression. (F) Section of P14 mouse kidney, showing scattered positive tubules in the kidney. These appear to be the collecting ducts, which are derived from the ureteric bud.

FIG. 4.

333I12 partially recapitulates Gata2 urogenital expression. Transgenic lines bearing the 333I12Z BAC transgene were mated to Gata2/GFP knock-in mice. Immunohistochemistry using anti-GFP antibody (A, B, and C) or X-Gal-stained adjacent sections (D, E, and F) on frozen sections from e14.5 mouse embryos shows that the BAC transgene expresses LacZ in some, but not all, of the places where GFP is expressed. (A and D) Transverse sections of the head show expression of lacZ in olfactory mesenchyme completely overlaps with expression of GFP. (B and E) Transverse sections at the level of the metanephros and mesonephros show that expression in the mesonephric mesenchyme (mes) is identical between lacZ and GFP. However, the epithelium of the ureteric bud (ub) shows only GFP expression, not lacZ. The mesenchyme surrounding the ureteric bud shows both GFP and lacZ staining. (C and F) Transverse sections at the level of the urogenital sinus (us) show a good overlap of lacZ and GFP staining. The notable exception is the expression in the epithelium of the ureter (ue [t]), which shows only GFP staining (B). (G) Direct GFP fluorescence from a transverse section at the level of the metanephros shows signal from both the epithelium (ue) and the mesenchyme (um) of the ureteric bud. (H) Anti-LacZ immunofluorescence in the same section shows signal only from the mesenchyme surrounding the ureteric bud, not from the epithelium. (I) This lack of overlap can be seen clearly in the merged image, where there is colocalization of GFP and lacZ in the mesenchyme but only GFP signal in the epithelium.

FIG. 5.

Isolation of urogenital enhancers by comparative genomics. (A) PIPMAKER analysis of the critical 70-kbp region that contains urogenital enhancer activity. The circled regions represent conserved noncoding sequences that were identified as longer than 200 bp with >80% conservation between the human and mouse. In the case of UG3, two of these elements were located in close proximity and were grouped together into one region. Other highly conserved regions that are not circled represent the exons of the Eefsec gene, which is transcribed in the opposite direction of Gata2. Length and percent identity of each CNS are listed in Table 2. Approximately 2 kbp surrounding these conserved regions were amplified by PCR and cloned into a vector containing the Gata2 promoter driving lacZ. (B) Transgenic founders bearing each of these conserved regions were analyzed by X-Gal staining. Founders bearing both UG2 and UG4 show expression of lacZ in the urogenital system (arrows). However, each transgene has a unique pattern of activity, where UG2 activates expression in a rostral urogenital domain and UG4 activates expression in a caudal domain. Photos are of representative embryos; total transgenic data are shown in Table 2.

FIG. 6.

Deletion of UG enhancers from 333I12Z demonstrates that two independent enhancers control Gata2 in urogenital mesenchyme. By using homologous recombination in E. coli, UG2 and UG4 were individually deleted from the parental 333I12Z BAC (A and B). Two representative 333I12ΔUG2 founder embryos at e12.5 show a similar staining pattern as 333I12 mice, except the rostral domain of urogenital staining is absent. This absent region corresponds to the region activated by the UG2 enhancer. (C and D) Two representative 333I12ΔUG4 founder embryos at e12.5 show a reciprocal pattern to the 333I12ΔUG2; they retain lacZ expression in the rostral domain but lose expression in the caudal urogenital system. This region of activation corresponds to the UG4 transgene. (E) Transverse section of 333I12ΔUG2 at the level of the metanephros shows staining in the mesenchyme surrounding the ureter (u), indicating that UG2 is not responsible for this domain. (F) Sections at the level of the mesonephros (m) show no staining, this is the rostral domain which is lost when UG2 is deleted (g, gonad). (G) Staining in the urogenital sinus (us) is preserved, indicating that this domain is not controlled by UG2. (H) Sections of 333I12ΔUG4 show no staining in the mesenchyme surrounding the ureter, indicating that UG4 controls activation in this region. (I) Sections show lacZ expression in the mesonephros, which confirms that this expression is not regulated by UG4. (J) No staining is present in the urogenital sinus, suggesting this region is controlled by UG4.

FIG. 7.

Summary of different activation domains controlled by Gata2 urogenital enhancers. (A) This study demonstrates the existence of two separate enhancers, lying more than 70 kbp downstream from the Gata2 promoter, which control activation in the urogenital system. These enhancers, in addition to enhancers present near the gene, as well as a hematopoietic regulatory element more than 100 kbp upstream of the gene, collaborate to give rise to the appropriate expression of Gata2. This study postulates the existence of a third urogenital enhancer, which cannot be localized in this study. Thus, the minimal Gata2 locus is at least 313 kbp, although the existence of a third, epithelial urogenital enhancer suggests that this locus must be larger than 1 Mbp, as we have now determined. (B) The spatial distributions of the UG2 and UG4 enhancers are clearly separated. Although both enhancers control expression in mesenchymal tissues in the urogenital system, our data suggests that UG2 is responsible for expression in the mesonephros and mesenchyme surrounding the nephric duct, while UG4 is responsible for expression in the urogenital sinus and mesenchyme surrounding the ureter. Mes, mesonephros; Met, metanephros; UG sinus, urogenital sinus; u, ureter; ND, nephric duct.