If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Genomics Division, MS 84-171, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USAU.S. Department of Energy Joint Genome Institute, Walnut Creek, CA 94598, USASchool of Natural Sciences, University of California, Merced, Merced, CA 95343, USA
ChIP-seq identifies the CRMs bound by endogenous HAND2 in embryos and limb buds
HAND2 controls key transcriptional regulators acting upstream of SHH in limb buds
These transcriptional circuits define proximal, anterior, and posterior identities
HAND2 establishes anterior and posterior compartments by regulating Gli3 and Tbx3
The genetic networks that govern vertebrate development are well studied, but how the interactions of trans-acting factors with cis-regulatory modules (CRMs) are integrated into spatiotemporal regulation of gene expression is not clear. The transcriptional regulator HAND2 is required during limb, heart, and branchial arch development. Here, we identify the genomic regions enriched in HAND2 chromatin complexes from mouse embryos and limb buds. Then we analyze the HAND2 target CRMs in the genomic landscapes encoding transcriptional regulators required in early limb buds. HAND2 controls the expression of genes functioning in the proximal limb bud and orchestrates the establishment of anterior and posterior polarity of the nascent limb bud mesenchyme by impacting Gli3 and Tbx3 expression. TBX3 is required downstream of HAND2 to refine the posterior Gli3 expression boundary. Our analysis uncovers the transcriptional circuits that function in establishing distinct mesenchymal compartments downstream of HAND2 and upstream of SHH signaling.
The limb bud is an excellent model to study the gene networks that govern growth and patterning during vertebrate organogenesis (
). Localized Shh expression depends on the interaction of different transcriptional regulators with the ZRS. In particular, the interaction with HOX, PBX, ETS, and HAND2 transcriptional complexes has been implicated in activation of Shh in the limb bud, whereas TWIST1, ETV, and GATA factors prevent anterior ectopic expression (
). How the ZRS integrates these various inputs over time is unknown, but the resulting posterior restriction of SHH signaling is essential for proliferative expansion and anterior-posterior (AP) patterning of the distal limb bud mesenchyme that will form the skeletal elements of the zeugopod and autopod (
). It has been shown that proximal mesenchymal progenitors express several transcriptional regulators belonging to the Pbx, Meis, and Irx gene families, which participate in specification and/or morphogenesis of proximal skeletal elements (
Hand2 is genetically required for limb bud, branchial arch, and heart development and the lethality of Hand2-deficient mouse embryos around embryonic day E9.5 is caused by the severe heart malformations (
). In particular, the Hand2 and Gli3 genetic antagonism is required to establish AP asymmetry and pentadactyly, as limb buds deficient for both these transcriptional regulators lack discernible AP polarity, Shh expression, and are extremely polydactylous (
). However, the molecular nature of the underlying transcriptional and cis-regulatory networks remained largely unknown, as Shh is the only known direct transcriptional target of HAND2 in limb buds.
We have inserted a 3xFLAG epitope tag into the endogenous HAND2 protein to first determine the range of genomic regions enriched in endogenous HAND2 chromatin complexes. In a second step, we focused our in-depth analysis predominantly on HAND2 target genes that encode transcription factors expressed and/or required during early limb development. This analysis established that during the onset of limb bud outgrowth, HAND2 controls the expression of transcriptional regulators in the proximal mesenchyme that are involved in the formation of the most proximal forelimb skeletal elements. In addition, our study reveals the gene regulatory logic by which HAND2, in cooperation with GLI3 and TBX3, establishes AP axis polarity in the early limb bud mesenchyme. In summary, our analysis uncovers the HAND2-dependent molecular circuits that function in establishing proximal, anterior, and posterior compartments and activating Shh expression during the onset of limb bud development.
A Hand23xFLAG Allele Generated by dRMCE Identifies the Genomic Regions Enriched in Endogenous HAND2 Chromatin Complexes
A 3xFLAG epitope tag was inserted into the endogenous HAND2 protein using dual recombinase-mediated cassette exchange (dRMCE;
) to replace the conditional allele with a Hand23xFLAG cassette in mouse embryonic stem cells (ES) (Figure 1A; Supplemental Experimental Procedures available online). Insertion of the 3xFLAG epitope tag into the N terminus does not alter HAND2 functions, as homozygous Hand23xFLAG (Hand23xF/3xF) mice develop normally and do not show any embryonic lethality (
; data not shown). The 3xFLAG epitope tag allows sensitive detection of the endogenous HAND2 protein isoforms in embryonic tissues (limb buds, heart, and branchial arches; Figures 1B and 1C). During the onset of forelimb bud outgrowth, HAND2-positive nuclei are detected in the posterior and proximal-anterior limb bud mesenchyme (green fluorescence; Figures 1D, S1A, and S1D). As HAND2 is required to activate Shh expression (
). The Shh-expressing cells define a subdomain within the posterior HAND2-positive limb bud mesenchyme (Figures 1D and S1D). We also colocalized HAND2 with the GLI3 repressor isoform (GLI3R) to determine their spatial distribution during activation of Shh expression. GLI3R proteins were detected by immunofluorescence using monoclonal GLI3 N-terminal antibodies (
) as they recognize the nuclear GLI3R isoform rather than the cytoplasmic full-length GLI3 and/or nuclear GLI3A activator isoforms in limb bud sections (Figures S1B and S1C). This analysis reveals the complementary distribution of the nuclear HAND2 (green) and GLI3R proteins (purple) in mouse forelimb bud mesenchymal cells (Figures 1D, S1B, and S1D).
The only known direct target of HAND2 in limb buds is Shh, whose transcriptional activation is controlled by interaction of HAND2 with the ZRS cis-regulatory region located ∼800 kb upstream of the Shh locus (
). Hand23xF/3xF limb buds were used for chromatin immunoprecipitation (ChIP) using anti-FLAG antibodies in combination with quantitative real-time PCR (ChIP-qPCR). While our previous analysis did not unambiguously identify the HAND2-binding region(s) within the ZRS (
). This interaction is specific, as another Ebox sequence is not enriched (ZRS5; Figure 1E). The enrichment of the ZRS2-4 region in HAND2 chromatin complexes is highest in early limb buds (E10.5; Figure 1E) in agreement with the early but transient genetic requirement of Hand2 for Shh activation (≤E10.5;
To define the range of potential HAND2 targets during its early genetic requirement in mouse limb buds, we used chromatin immunoprecipitation sequencing (ChIP-seq). To this aim, the HAND2 binding profile in Hand2-expressing embryonic tissues at E10.5 (eT, limb buds and flank tissue, hearts, and branchial arches; Figure 1B) and dissected limb buds (Lb) was determined by HAND23xF ChIP-seq analysis (Figures 2A–2D and S2A). The primary data sets were analyzed using model-based analysis for ChIP-seq (MACS)-based peak calling, and regions enriched in wild-type eT controls were excluded as nonspecific (Tables S1 and S2 list all statistically validated regions enriched ≥2-fold). For both validated ChIP-seq data sets, the top 1,000 peaks were selected for the initial meta-analysis based on their fold enrichment over the input control and number of reads (Figures 2A–2D and S2A; Tables S3 and S4). The majority of the top 1,000 HAND2-binding peaks in both data sets map to conserved sequences located ≥10 kb away from the closest transcriptional start site (TSS; Figures 2A and 2B). De novo motif discovery showed that also the endogenous HAND2 chromatin complexes interact preferentially with the Ebox sequence defined in vitro (Figure 2C;
Intersection of the top 1,000 eT and Lb data sets shows that 259 of them are shared (Figure S2A; Table S5). GREAT analysis showed that the shared HAND2-binding regions are most often associated with genes known to function in limb bud and/or skeletal development (Figures 2D and S2B). Hence, this set of shared regions and associated genes likely corresponds to HAND2 target genes in limb buds. However, it is important to consider that the sequence coverage was overall significantly lower for the Lb than for the eT ChIP-seq data set. This is a likely consequence of the much lower amounts of tissue and HAND2 chromatin complexes recovered from the ∼1,000 fine-dissected limb buds used for ChIP-seq analysis (Supplemental Experimental Procedures). Furthermore, the ZRS, which is an established HAND2 target region (Figure 1E), was moderately enriched only in the eT ChIP-seq data set (Table S4). This could reflect the fact that the HAND2-ZRS interaction occurs only in a small fraction of posterior mesenchymal cells in early limb buds, and ChIP-qPCR analysis relying on specific oligos is more sensitive (Figures 1D and 1E). The limited sensitivity of the Lb ChIP-seq data set, in particular, indicates that not all relevant HAND2 binding peaks have been detected and represented in the top 1,000 peaks used for the comparative analysis. This is likely a consequence of a significant fraction of HAND2 target genes being restricted very proximally in early limb buds and extending into the flank mesenchyme (see below). However, the flank mesenchyme was only included in the eT, but not in the Lb ChIP-seq sample. Therefore, all potential HAND2 target regions selected for further analysis were first verified in early limb buds (E10.5), using the more sensitive HAND2 ChIP-qPCR analysis (n = 3; Experimental Procedures).
As we mainly wanted to get insight into the transcriptional circuitry controlled by HAND2 during prepatterning of early limb buds (
), the current study focused on the genomic regions associated with transcription factors functioning and/or expressed upstream of activating Shh expression. To assess the expression of the associated genes in mutant mouse limb buds, Hand2 was conditionally deleted during limb bud initiation using the Hoxb6-Cre transgene (
). This inactivates Hand2 in the posterior two-thirds of the forelimb bud mesenchyme (Figures S2C and S2D), resulting in almost complete penetrance of the Hand2 loss-of-function limb bud phenotype (n = 9/10). Our initial analysis of the curated list of genes expressed and/or functioning in early limb buds showed that most often their expression is altered in the posterior-proximal mesenchyme of Hand2Δ/Δc forelimb buds (≤E10.25; n = 17/29 genes analyzed; Table 1). Their expression is either significantly reduced or lost from the posterior mesenchyme (n = 10/17; including Ets1, Ets2, Gsc, Tbx2, and Tbx3; Table 1; Figure S2E) or expands posteriorly (n = 7/17; including Gli3, Tbx18, Irx3, and Irx5; Table 1; Figure S2F). This initial analysis pointed to the existence of a network of transcriptional regulators that are direct targets of HAND2 during the onset of limb bud development (Table 1; data not shown).
Table 1Summary of the RNA In Situ Hybridization Analysis of a Select Set of HAND2 Target Genes in Hand2-Deficient Limb Buds
Validated ChIP-seq peaks associated with genes encoding transcriptional regulators and/or essential roles during the onset of mouse limb bud development were selected. This table summarizes the alterations in their expression in Hand2-deficient mouse forelimb buds (≤E10.5). Genomic landscapes associated with genes functioning during progression of limb bud development and genes expressed either in the limb bud ectoderm or during advanced stages (e.g., chondrogenesis) were not considered for this study.
For example, two candidate CRMs located 23 kb upstream of the Ets1 and 146 kb downstream of the Ets2 TSS are significantly enriched in HAND23xF chromatin complexes as confirmed by ChIP-qPCR in early limb buds (Figures 2E and 2F). Indeed, the expression of both Ets1 and Ets2 is reduced in Hand2-deficient limb buds (Figure 2G). The Ets2 +146kb CRM likely encodes an enhancer as the orthologous human region is active in the posterior limb bud mesenchyme of transgenic mouse embryos (Figure 2H, left panel; Vista enhancer hs1516;
). The domain of the LacZ reporter not only recapitulates major aspects of Ets2 expression (Figure 2H, middle panel) but also overlaps the endogenous Hand2 domain (Figure 2H, right panel). These results establish the Ets1 and Ets2 genes as bona fide HAND2 target genes in early limb buds. As the ETS binding sites in the ZRS are essential for Shh expression in the posterior limb bud (
), HAND2-mediated upregulation of the Ets transcription factors likely helps to reinforce Shh expression.
HAND2 Target Genes Are Required in the Proximal Limb Bud for Morphogenesis of the Scapula and Humerus
As part of the initial analysis, HAND2 peaks were detected in the genomic landscapes of several transcription factors expressed in the proximal limb bud mesenchyme (Figure 3; Table 1) and that function in morphogenesis of proximal limb skeletal elements (scapula and humerus;
). In particular, two prominent HAND2 peaks, whose enrichment in early limb buds was confirmed by ChIP-qPCR, map to the Goosecoid (Gsc) genomic landscape (at positions +33 and −133 kb; Figure 3A). Gsc is expressed in the proximal and flank mesenchyme during early limb bud development and is required for scapular head development in the mouse (
). In agreement with Gsc being a direct transcriptional target of HAND2, its expression is lost from the posterior mesenchyme of Hand2Δ/Δc forelimb buds (Figure 3A). Genetic analysis underscores the functional importance of Gsc regulation by HAND2, as Hand2Δ/Δc and Gsc-deficient limbs display strikingly similar scapular defects affecting the glenoid cavity and coracoid process (n = 4/8; Figures 3B and S3A; compare to
). In limb buds, HAND2 chromatin complexes also interact with a specific region in the Tbx18 genomic landscape (at position −338 kb; middle panel, Figure 3C). Indeed, Tbx18 expression expands posteriorly in Hand2Δ/Δc forelimb buds (Figure 3C), and previous genetic analysis established that Tbx18 also functions during scapula and humerus development (
Recently, it has been shown that the Irx3 and Irx5 transcriptional regulators, which are part of the same gene cluster (Figure 3D), are required to specify the progenitors of proximal and anterior limb skeletal elements during limb bud initiation (
). Irx3 and Irx5 are likely part of the same cis-regulatory landscape as they are coexpressed in the anterior-proximal limb bud mesenchyme and their expression expands posteriorly in Hand2Δ/Δc forelimb buds (Figure 3E). As this region encompasses binding regions for both HAND2 and GLI3R chromatin complexes, its enhancer potential (blue bar in Figure 3D) was assessed in transgenic mouse embryos (Figure 3F). Indeed, this region functions as a CRM in limb buds, but the mesenchymal LacZ reporter activity is rather uniform in contrast to the proximal-anterior restriction of the endogenous Irx3 and Irx5 expression domains (n = 10/10; Figure 3F). Furthermore, as the GLI3R binding regions in limb buds (
) often map close to the ones for HAND2 (red bars in Figures 3A, 3C, and 3D; see also Figure S3B), the expression of these proximal genes was also analyzed in limb buds lacking both Hand2 and Gli3 (Figures S3C and S3D). While Tbx18 expression is not significantly altered, the expression of Gsc and Irx3 is much more reduced in double than in wild-type or in single mutant limb buds (Figures S3C and S3D; data not shown). These alterations in gene expression are paralleled by a more severely dysplastic scapula and humerus in Hand2Δ/ΔcGli3Δ/Δ limbs (Figure S3E). Together, these results indicate that HAND2 acts in concert with GLI3R to control a transcriptional network that functions in morphogenesis of proximal limb skeletal elements.
Direct Cross-Regulation between HAND2 and GLI3R Underlies Molecular Establishment of an Anterior and Posterior Limb Bud Compartment
Genetic analysis has shown that Hand2 and Gli3 are mutually antagonistic, and this has been proposed to prepattern the limb bud mesenchyme along its AP axis prior to activation of SHH signaling (
). To gain insight into the underlying molecular and cellular interactions, we colocalized the endogenous HAND23xF and GLI3R proteins during forelimb bud formation (E9.25; Figure 4A). HAND2 protein levels are highest in posterior and proximal-anterior nuclei (green fluorescence; Figure 4A), whereas GLI3R is most abundant in anterior nuclei (red fluorescence, Figure 4A). In spite of their opposing distribution, most mesenchymal cells coexpress nuclear HAND2 and GLI3R proteins during forelimb bud formation (yellow overlap; n = 4/4 at E9.25; Figures 4A and S5A). Subsequently, HAND2 becomes undetectable in anterior cells, which retain GLI3R. Conversely, GLI3R is lost from posterior cell nuclei, which retain HAND2 (n = 3/3 at E9.5-9.75; Figures 4B and S5B). Cells at the interface of the two domains continue to coexpress both proteins (Figures 4B and S5B). These spatial dynamics reveal how the populations of GLI3R-positive anterior and HAND2-positive posterior cells segregate, likely concurrently with molecular establishment of AP axis polarity in forelimb buds. The SHH-independent nature of these interactions is revealed by the fact that the HAND2 and GLI3R domains are initially normal in Shh-deficient limb buds (E9.5–9.75; Figure S5C). During progression of forelimb bud outgrowth, the HAND2 and GLI3R domains are increasingly separated by cells expressing neither of these proteins (n = 3/3 at E10.25; Figure 4B), which likely reflects the inhibition of GLI3R formation by SHH signaling. As the two proteins are initially coexpressed, direct cross-regulation could underlie establishment of the mutually exclusive HAND2 and GLI3R domains. Indeed, GLI3R interacts with two CRMs in the Hand2 genomic landscape that function in repressing Hand2 from the anterior limb bud mesenchyme (
). The CRM at position +13 kb is associated with one overlapping and a close-by region (at +11 kb) enriched in HAND2 chromatin complexes from limb buds (Figures S5D and S5E). This indicates that in addition to mediating repression by GLI3R, this CRM could also participate in autoregulation of Hand2 expression.
As Gli3 expression expands posteriorly in Hand2-deficient mouse limb buds (
), we screened also the Gli3 genomic landscape for HAND2-interacting regions (Figure 4C). Indeed, a region located 120 kb upstream of the Gli3 transcription unit is prominently enriched in early limb buds (right panel, Figure 4C). Transgenic analysis showed that this CRM is indeed active both in fore- and hindlimb buds (Figure 4D). In particular, this CRM is able to drive LacZ expression in a spatial pattern reminiscent of the endogenous Gli3 transcript distribution, with LacZ being excluded from the most posterior mesenchyme (n = 6/7; Figure 4D; data not shown).
Tbx3 Is a HAND2 Target Gene that Positions the Posterior Gli3 Expression Boundary
HAND2 chromatin complexes are also enriched in two specific regions of the Tbx3 genomic landscape that is required for expression in limb buds (Figures 5A and 5B ;
). While the eT and Lb HAND2 ChIP-seq data sets do not unequivocally identify the two regions for the reasons stated above, HAND2 ChIP-qPCR using early limb buds established that the region located 58 kb upstream of the Tbx3 transcription unit is ∼3-fold more enriched than the one located at −19 kb (left panel, Figure 5B). The enrichment of the Tbx3 −58 kb region is most prominent in early limb buds (right panel, Figure 5B), pointing to an involvement of HAND2 in regulating the early Tbx3 expression. This was corroborated by transgenic analysis, which established that the Tbx3 −58 kb CRM drives LacZ expression in the posterior-proximal forelimb bud and flank mesenchyme in a pattern strikingly similar to the endogenous Tbx3 expression (n = 8/10; Figures 5C and S6A). In contrast, the Tbx3 −19 kb region is not active in mouse limb buds (n = 0/7; Figure S6A). The Tbx3 −58 kb CRM regulates Tbx3 expression during early limb bud development, as the LacZ expression domain fails to expand distally as limb bud development progresses (Figure S6A). As neither of the two HAND2 binding regions is active in hindlimb buds (Figures 5C and S6A; data not shown), the expression of Tbx3 in hindlimb buds must be controlled by different cis-regulatory interactions.
Indeed, Tbx3 expression is lost from the posterior mesenchyme of Hand2-deficient forelimb buds, while it remains comparable to wild-types in mutant hindlimb buds (Figure 5D;
). Furthermore, the expression of Tbx5, which is located ∼163 kb downstream of Tbx3 as part of the same gene cluster, is not affected in Hand2-deficient forelimb buds (Figure S6B). Several HAND2 ChIP peaks were also detected in the Tbx2-Tbx4 genomic landscape (Figures S6C and S6D). Similar to the Tbx3-Tbx5 genomic landscape, Tbx2, but not Tbx4, expression is reduced in Hand2-deficient limb buds (Figures S6E and S6F). These apparent similarities in cis-regulation are likely related to the emergence of these gene clusters by duplication of an ancestral complex (
). Therefore, we focused our analysis on the HAND2-TBX3 interactions during the onset of forelimb bud development. HAND2 and TBX3 proteins are initially coexpressed (Figure S6G). However, TBX3 becomes rapidly more restricted than HAND2 such that the double-positive cells hallmark the proximal-posterior forelimb bud and posterior flank (Figure 5E). In Shh-deficient forelimb buds, both protein domains are more restricted but are otherwise maintained (right panels, Figure 5E). In contrast, TBX3 is absent from Hand2Δ/Δc and Hand2Δ/ΔcGli3Δ/Δ forelimb buds, with the exception of few positive cells at the proximal border (right panels, Figure 5F). Taken together, these results provide compelling evidence that Tbx3 is a direct transcriptional target of HAND2 in the posterior and flank mesenchyme during the onset of limb bud development.
To gain insight into the potential involvement of TBX3 in AP patterning of the early limb bud mesenchyme, the spatial distribution of the mesenchymal cells actively transcribing Gli3 was determined in relation to HAND2 and TBX3 proteins (Figures 6A and 6B ). To achieve cellular resolution, the Gli3 allele expressing EGFP under control of the endogenous locus (
) was used in combination with GFP antibodies (red fluorescence; Figures 6A and 6B). While the GLI3R and HAND2 protein domains are increasingly separated (Figure 4B), a significant fraction of boundary cells continue to coexpress HAND2 protein and Gli3 transcripts in wild-type limb buds (yellow fluorescence, left panels, Figure 6A). In contrast, far fewer cells coexpress TBX3 and Gli3 in the boundary region (right panels, Figure 6A). This difference is even more striking in Gli3Δ/Δ forelimb buds, which continue to express the mutant Gli3 transcripts (Figure 6B). There is a large population of cells coexpressing HAND2 and Gli3 due to the anterior expansion of the Hand2 expression domain in Gli3-deficient limb buds (left panels, Figure 6B;
). However, TBX3 remains posteriorly restricted in Gli3-deficient limb buds, and similar to wild-types, only few cells coexpress TBX3 proteins and Gli3 transcripts (right panels, Figure 6B). This boundary is also retained in Shh-deficient forelimb buds in spite of the posterior expansion of Gli3 (Figures S7A and S7B). These results show that HAND2 is not sufficient to establish a sharp boundary but rather that TBX3 defines the limit between the anterior Gli3-positive and posterior Gli3-negative mesenchymal cells. Indeed, the comparative analysis of wild-type, Shh-deficient, and Tbx3-deficient forelimb buds reveals that Tbx3 is required to establish this boundary (Figures 6C and 6D). In Shh-deficient limb buds, Gli3 transcription expands posteriorly with more restricted HAND2 and TBX3 domains but fails to reach the posterior margin (right panel, Figures 6C, 5E, and S7A). Rather strikingly, Gli3 expression expands into the posterior flank mesenchyme in Tbx3-deficient forelimb buds as in Hand2-deficient limb buds (Figure 6D). Hand2-deficient forelimb buds lack Tbx3 (Figures 5D and 5F) in their posterior mesenchyme, whereas Hand2 transcripts remain in Tbx3-deficient limb buds albeit at reduced levels (Figure S7C). Together, these results corroborate the proposal that the HAND2 transcriptional target Tbx3 is required to establish the posterior Gli3 expression boundary by inhibiting its expression in the posterior-most mesenchyme (Figures 6 and 7).
There is ample evidence that tissue-specific CRMs orchestrate the dynamics of gene expression during embryonic development and fine-tune morphogenesis as a consequence of their often highly dynamic regulation (
). Therefore, it is important to identify the trans-acting factors and complexes that regulate these CRMs in a tissue and stage-specific manner. To this aim, we have inserted a 3xFLAG-tag into the endogenous HAND2 protein, which permitted us to uncover genomic regions enriched in HAND2-containing chromatin complexes in mouse embryos. The current study focuses on an in-depth functional analysis of HAND2-CRMs in genomic landscapes of transcriptional regulators that are required during early mouse forelimb bud development. This analysis uncovers the architecture of a network of HAND2 target transcription factors that function in early determinative events that set up the proximal, anterior, and posterior domains during the onset of forelimb bud outgrowth (Figure 7). An essential role of HAND2 in proximal skeletal development was revealed by the fact that its early Hoxb6-Cre mediated conditional inactivation disrupts scapular head morphogenesis. Our study provides good evidence that this is a consequence of HAND2 directly regulating the expression of transcription factors that function in the development of the proximal limb skeletal elements, such as Gsc, Irx3, Irx5, and Tbx18 (Figure 7). In particular, the HAND2 ChIP analysis identifies a CRM located upstream of Irx3 that is active in the limb bud mesenchyme. In turn, IRX3 regulates Gli3 expression by directly interacting with a limb bud mesenchymal CRM located in the Gli3 genomic landscape (
). These direct interactions begin to reveal the underlying complexity of the cis-regulatory circuitry operating during the onset of limb bud development (Figure 7). In summary, our genetic analysis supports the proposal that the HAND2-regulated gene networks are required to pattern the early mesenchymal territory that gives rise to the proximal limb skeletal elements in a Shh-independent manner (
showed that the competence to activate Shh expression is widespread and without posterior bias in the forelimb field of early mouse embryos (E9.0). Together with other studies (see Introduction), this suggests that the limb bud mesenchyme might only be polarized along the AP axis during the onset of limb bud development in mouse embryos. The present study shows that the nuclear HAND2 and GLI3R proteins are initially coexpressed by the mesenchymal progenitors that give rise to the forelimb bud. However, during initiation of limb bud development, the spatial distributions of the two transcriptional regulators rapidly segregate into a distinct GLI3R-positive anterior and HAND2-positive posterior compartment. In addition to establishing this dynamic segregation with cellular resolution, we identify the HAND2-dependent cis-regulatory interactions and transcription factor networks that establish these anterior and posterior compartments (Figure 7). The loss of Hand2 transcripts and proteins from the anterior limb bud mesenchyme (
). The HAND2 ChIP analysis identifies one CRM in the Gli3 genomic landscape that is active in limb buds, with exception of the most posterior mesenchyme. The transcriptional regulation of Gli3 in early limb buds is complex and likely controlled by several CRMs with similar activities, one of which is bound by IRX3, a direct transcriptional target of Hand2 (
; this study). While the genetic analysis shows that Hand2 is required to repress the expression of Gli3 and other direct targets in the posterior mesenchyme, our transgenic studies of HAND2-interacting CRMs rather reveals their ability to activate LacZ expression in the limb bud mesenchyme. HAND2 was first described as a transcriptional activator (
), which could determine the repressive activity of HAND2 chromatin complexes interacting with CRMs in, e.g., the Gli3 genomic landscape. Alternatively, the repressive effect of HAND2 on Gli3 could be mediated via the transcriptional target Tbx3, which is required to repress Gli3 in the posterior mesenchyme. In fact, TBX3 has an essential role in positioning the Gli3 expression boundary (Figure 7), and its overexpression in forelimb buds of chicken embryos inhibits Gli3 expression (
). Our study now uncovers distinct transcriptional networks interlinked by HAND2 that are required to set-up the initial proximal, posterior, and anterior mesenchymal compartments prior to the onset of SHH signaling. These networks define proximal fates and the AP boundaries with cellular precision through selective activation and/or repression of downstream transcriptional regulators. The direct cross-regulation among these transcriptional regulators not only defines the limb bud mesenchymal compartments but also enables activation of SHH signaling, which then elaborates these compartments during distal progression of limb bud outgrowth.
Experiments involving mice were performed with strict adherence to Swiss law, the 3R principles, and the Basel Declaration. All animal studies were approved by the cantonal animal welfare and ethics committees. The procedures for generating transgenic mice at the Lawrence Berkeley National Laboratory (LBNL) were reviewed and approved by the LBNL Animal Welfare and Research Committee. The Hand23xF and Gli3C3xF alleles were maintained in an NMRI background. The Hand2f (floxed), Hand2Δ, and Gli3ΔGFP alleles were maintained in mixed backgrounds. The Gli3C3xF allele was constructed by inserting a 3XFLAG epitope tag in frame at the carboxy terminus of the endogenous GLI3 protein. Western blotting detects the 190 kD full-length GLI3 protein. Mice and embryos homozygous for this allele are phenotypically wild-type and will be described elsewhere (J.L.-R. and R.Z., unpublished data). The ShhGFPCre allele (
) were maintained in a C57BL/6J background. The Tbx3Venus allele used to generate Tbx3-deficient embryos recapitulates all known Tbx3 loss-of-function phenotypes and will be described elsewhere (R.M. and V.M.C., unpublished data).
Limb buds were fixed in 4% paraformaldehyde for 2–3 hr at 4°C, and proteins were detected on 10 μm cryosections. Primary antibodies against the FLAG epitope (M2; 1:500; Sigma), GLI3 (3.6 ng/ml; clone 6F5;
), TBX3 (1:300; E-20; Santa Cruz), and GFP (1:1,000; Life Technologies) were used. Goat anti-mouse (FLAG/GLI3), goat anti-rabbit (GFP), and rabbit/donkey anti-goat (TBX3) secondary antibodies conjugated to Alexa Fluor 488 or 594 (1:1,000; Life Technologies) were used for detection. For colocalization of HAND2 and GLI3R, anti-FLAG (M2; F3165, Sigma) antibodies were labeled with Alexa Fluor 488 using the APEX antibody labeling kit (Life Technologies). Sequential treatment with rabbit anti-Alexa Fluor 488 (Life Technologies) and goat anti-rabbit 488 Alexa Fluor enhanced the signal. Nuclei were counterstained with Hoechst-33258. The ShhGFPCre and Gli3ΔGFP null alleles were used in combination with immunodetection of GFP to visualize cells expressing Shh or Gli3, respectively. Autofluorescence, e.g., of blood cells, was detected equally in all channels. Therefore, it was captured utilizing the 633 nm laser of the confocal microscope and digitally removed using the subtraction mode in Photoshop CS5 (Adobe) as shown in Figure S1A. Images were acquired using a Leica SP5 confocal microscope.
ChIP-qPCR and ChIP-Seq Analysis
Fore- and hindlimb buds dissected from ∼50 Hand23xF/3xF embryos at E10.5 were processed for ChIP as described (
) using M2 anti-FLAG antibody (F1804; Sigma). For time course experiments, 20 fore- and hindlimbs of E11.5 or E12.5 Hand23xF/3xF embryos were processed in an identical manner. Chromatin was fragmented for 15 (E10.5) or 20 min (E11.5, E12.5) using a S220 Covaris Ultrasonicator, which yielded average fragment sizes in the range of 200–300 bp. All results (mean ± SD) represent three independent biological replicates. The details of the ChIP-seq analysis are given in the Supplemental Experimental Procedures.
We wish to thank the mouse transgenic core facility of the University of Basel for generation of chimeric mice and A. Offinger and her team for excellent animal care. Frédéric Laurent is thanked for performing the OPT analysis. We are grateful to Jennifer Akiyama for generation of constructs, Ina Nissen (D-BSSE) for the preparation of the ChIP-seq library, and Isabelle Ginez for preparing the tissue sections. This research was supported by Swiss National Science Foundation grants ( 31003A_130803/146248 ) and both cantons of Basel (to R.Z.), an EU Reintegration grant ( PERG-GA-2009-246576 ) to J.L.-R.), a Netherlands Heart Foundation grant ( 2010B205 ) (to V.M.C.), and NIH grants ( R01HG003988 and U01DE020060 ) (to A.V.). A.V.’s research was conducted at the E.O. Lawrence Berkeley National Laboratory and performed under a Department of Energy Contract ( DE-AC02-05CH11231 ), University of California . X.W. and S.J.S. are employees of Genentech, a member of the Roche group.
The GEO database accession number for all data sets reported in this paper is GSE55707.