BAC transgenic analysis reveals enhancers sufficient for Hoxa13 and neighborhood gene expression in mouse embryonic distal limbs and genital bud

Abstract

We previously demonstrated that a ~1 Mb domain of genes upstream of and including Hoxa13 is co-expressed in the developing mouse limbs and genitalia. A very highly conserved non-coding sequence, mmA13CNS, located ~350 kb upstream of the Hoxa13 gene, was shown to be insufficient in transgenic mice to direct precise Hoxa13-like expression in the limb buds or genital bud, although some LacZ expression from the transgene was found in those tissues. In this report, we used overlapping β-globin minimal promoter LacZ recombinant BAC mouse transgenes encompassing mmA13CNS to localize genital bud and distal limb enhancers. Hoxa13-like embryonic genital bud expression was observed with both BACs, suggesting that a genital bud enhancer lies within the region of BAC overlap. In contrast, at least two separate regions of sequence remote to the HoxA cluster are required to drive Hoxa13-like expression in developing distal limbs. Given that the paralogous posterior HoxD and neighboring genes have been shown to be under the influence of long-range distal limb and genital bud enhancers, we hypothesize that both HoxA and HoxD long-range enhancers have one ancestral origin, which diverged in both sequence and function after the HoxA/D cluster duplication.

Introduction

Hox genes encode transcription factors that are conserved among all vertebrates and are essential for appropriate axial and appendicular embryonic patterning (McGinnis and Krumlauf, 1992; Veraksa et al., 2000). Over the course of vertebrate evolution, a single ancestral Hox cluster underwent individual gene, and entire cluster, duplications leading to 39 Hox genes organized into four clusters (HoxA - HoxD) characteristic of all mammals (Kappen et al., 1989; Prince, 2002; Ruddle et al., 1994). During embryonic axial patterning the Hox genes are colinearly expressed, both spatially and temporally (Duboule and Morata, 1994). For appendicular patterning, the posterior HoxA and HoxD genes are re-expressed in the presumptive appendicular tissues. Specifically, the Abd-B derived genes in the posterior HoxA and HoxD clusters are necessary for limb and urogenital development (Fromental-Ramain et al., 1996; Mortlock and Innis, 1997; Warot et al., 1997; Zakany et al., 1997). Mutations in human HOXA13 are associated with dominantly inherited hand-foot-genital syndrome (HFGS; OMIM #140000) (Goodman et al., 2000; Mortlock and Innis, 1997) which is characterized by malformations of the distal limbs as well as urogenital structures. Similar malformations were also observed in the spontaneous mouse mutant, Hypodactyly (Mortlock et al., 1996), and in engineered Hoxa13 null mouse models (Fromental-Ramain et al., 1996; Stadler et al., 2001).

The function of posterior HoxA and HoxD genes in developing appendicular structures is believed to have been acquired secondarily to the primary, ancestral role of all four clusters in axial patterning (Herault et al., 1998; Spitz et al., 2001). Regulatory sequences responsible for the expression of the HoxD genes in the developing limb buds and genital bud have been mapped to locations remote from the cluster (Gonzalez et al., 2007; Spitz et al., 2003; Spitz et al., 2001), and we have hypothesized that the HoxA cluster is under a similar mode of regulation (Lehoczky et al., 2004). In previous work we showed that a set of genes upstream of Hoxa13, (Evx1, Hibadh, Tax1bp1, and Jazf1), are expressed in Hoxa13-like distal limb and genital bud domains between E10.5 and E13.5 (Lehoczky et al., 2004). We hypothesized that a broadly acting enhancer is responsible for the shared spatial and temporal gene expression patterns. In addition, we hypothesized that the enhancer exerts function over a finite domain because genes immediately proximal and distal to the neighborhood (Hoxa11 and Crebpa, respectively) are not expressed similarly.

A similar global enhancer has been described for the HoxD locus (Gonzalez et al., 2007; Spitz et al., 2003). At this locus, Evx2 and Lnp are expressed in the distal limb and genital bud in a spatial and temporal pattern in common with the posterior HoxD genes. Given that the four mammalian Hox clusters are derived from a single ancestral cluster, we hypothesized that a single regulatory innovation occurred in the HoxA/D lineage prior to cluster duplication. This regulatory mechanism was retained by both the HoxA and HoxD clusters, driving posterior HoxA/D gene expression in the developing appendicular and urogenital structures. Results of comparative sequence analysis supported our hypothesis, as five paralogously conserved (HoxA locus vs. HoxD locus) non-coding DNA sequences (putative regulatory elements) are shared between the HoxA/D upstream regions (paraCNS1-5) (Lehoczky et al., 2004). No similar sequence conservation was found between HoxA/D and the HoxB or HoxC genomic regions, consistent with absence of posterior HoxB/C expression in the distal limbs and genital bud. On the HoxA chromosome, paraCNS4 and 5 map to a conserved 2.25 kb element (mmA13CNS) and they also correspond to the ‘distal limb enhancer’ critical region on the HoxD chromosome (Spitz et al., 2003). We tested the ability of mmA13CNS to direct LacZ expression in the developing limbs and urogenital system and found that mmA13CNS was insufficient for appropriate posterior HoxD-like/Hoxa13-like distal limb or genital bud expression. Although inappropriate LacZ expression was reproducibly found in those tissues, in multiple lines, we hypothesized that mmA13CNS may function in Hoxa13 tissue-specific expression, but additional surrounding sequences were necessary for appropriate expression.

In this report we created transgenic mice with BACs encompassing mmA13CNS with β-globin LacZ reporters in cis to examine our hypothesis and to begin to identify broad enhancer domains within which to focus future attention. By deduction, expression from overlapping BAC clones allowed us to demonstrate the involvement of a single critical region for Hoxa13-like embryonic genital bud expression. By contrast, the same BACs illustrate the existence of at least two separate loci remote to the HoxA cluster involved in Hoxa13-like distal limb expression from E10.5-E13.5.

Materials and Methods

LacZ expression from mmA13CNS mouse transgenic lines generated from β-globin-LacZ (data not shown) corroborated the expression found from the Hsp68-LacZ lines (Lehoczky et al., 2004). Both constructs were insufficient to drive appropriate Hoxa13-like distal limb or genital bud expression at any embryonic stage. We concluded that the two minimal promoters were comparable for the purposes of our experiments and proceeded to use the β-globin minimal promoter LacZ construct for all future experiments.

BAC Modification

BAC clones from the HoxA locus RPCI-23-347D13 and RPCI-23-420L19 were identified by end sequences AZ032610, AZ032607, AZ068097, and AZ068091 (www.tigr.org). Original clones were both fingerprint (NotI and EcoRI) and PCR verified to assure identity and structure. The minimal promoter LacZ reporter cassette, pFNeoH1H2βglob-LacZ, was constructed to target the sacB gene in the pBACe3.6 backbone in either BAC. The neo cassette flanked by FRT sites cloned into the SmaI site of pBKS was obtained (gift of J.D. Engel) and modified to insert the β-globin minimal promoter LacZ reporter (kindly provided by R. Krumlauf) into the SalI site.

BAC homology arms H1 and H2 were individually PCR amplified with restriction site tailed primers (H1_PmeI TTGTTTAAACGTTACGACTGCACTTCTGG, H1_PacI.R TTTTAATTAACGAATTGAGGCACTTGGT, H2_XhoI.F AACTCGAGCGATATTTACATGCTTGGTT, and H2_KpnI.R AAGGTACCCATGTAGCTTGTGATAACCA) and directionally cloned into the targeting construct. The linear fragment for recombineering was generated from the resultant clone by insert excision with a PmeI/KpnI double digest followed by gel purification. Targeting of the linear fragment into the BAC vector was done as previously reported (Khandekar et al., 2004; Lee et al., 2001; Yu et al., 2000) with the following exceptions. EL250 (kindly provided by N.A. Jenkins and N.G. Copeland) heat induced, electrocompetent cells (100 mL culture) were prepared when the density of the culture reached an A600 of 0.2. The culture was centrifuged 15 minutes at 10,000 rpm at 4 degrees and the resultant pellet was resuspended in 10 mL of ice cold 10% glycerol. This was repeated 4 times and the final pellet was resuspended in 100-200 uL of 10% glycerol. 100 ng of the linear targeting construct was electroporated into 25 uL of freshly prepared, heat-induced cells. Clones that were chloramphenicol- and kanamycin-resistant were verified by PCR for the presence of the targeting event, and were used for subsequent arabinose induction of flp recombinase to excise the neo cassette. Resultant clones that were chloramphenicol-resistant and kanamycin-sensitive were both fingerprint and PCR verified to assure that the intended recombinant clones were obtained.

Transgenic Mice and Embryo LacZ Staining

BAC transgene preparation was performed by standard alkaline lysis protocols followed by sepharose-CL-4B column purification as previously described (Yang et al., 1997). Undigested BAC transgene DNA was microinjected into (C57BL/6 × SJL) F2 fertilized eggs at ~1 ng/ul (University of Michigan TAMC). All DNA samples extracted from yolk sacs or tail snips were genotyped for the BAC insert/BAC vector SP6 junction as well as the β-globin/LacZ junction. The preliminary 347D13-LacZ transgene microinjection was used for an E12.5 founder embryo harvest. This injection yielded 5 transgenic of 80 total embryos, 2 of which had LacZ expression. Additional 347D13-LacZ microinjections were performed to obtain permanent lines. 8 transgenic founders were obtained from a total of 331 pups. 3 founder lines were used for timed embryo analysis (3 founders did not pass the transgene through the germline and 2 founder lines did not express the transgene). A single 420L19-LacZ transgene microinjection was performed and yielded 10 transgenic animals out of 108 total offspring. 7 founder lines were used for timed embryo analysis (2 founders did not transmit the transgene and one founder line did not express the transgene). All timed matings and LacZ embryo staining were done as previously described (Lehoczky et al., 2004).

BAC Transgene Contribution Analysis

Because the BAC clones were derived from the C57Bl/6 mouse strain, the transgenic lines were crossed onto a polymorphic strain to assess, by SNP analysis, what portion(s) of BAC transgene had integrated into the genome. Four 347D13-LacZ and seven 420L19-LacZ transgenic lines were individually crossed onto the DBA2/J strain, which is polymorphic relative to C57Bl/6J throughout and upstream of the HoxA locus. In all cases, F1 litters were genotyped for the presence of the transgene and these positive animals were obligate DBA2/J heterozygotes. Transgene+/-; DBA+/- animals were subsequently backcrossed onto the DBA2/J strain and N1 litters were once again screened for the presence of the appropriate transgene. Transgenic DNA samples were then genotyped for the C57/SJL/DBA polymorphic SNPs, RS3023069 and RS13478756, positioned adjacent to the endogenous region from which the BAC clones were derived. Resultant DNA samples that were transgenic and DBA congenic for the upstream HoxA region were used to PCR amplify across SNPs throughout the region of the BAC transgene. SNPs RS29851258, RS29857058, RS30707421, RS13478755, RS29863749, and RS30170317 were used for the 347D13-LacZ lines and only the latter three were used for the 420L19-LacZ lines (Supplemental Figure 2). Lines 250 and 268 were excluded from this analysis due to repeated unsuccessful attempts to generate transgenic DBA +/+ animals. PCR products were directly sequenced (University of Michigan Sequencing Core), and resultant chromatograms were visually inspected for the DBA SNP only (no transgene contribution), or DBA/C57 SNP (transgene contribution).

Transgene copy number analysis was performed with phenol/chloroform prepared spleen DNA from N1 or N2 C57Bl/6J backcross animals. Southern blots were prepared by standard methods. In brief, 10 ug of transgenic or C57Bl/6J wild-type genomic DNA was digested with EcoRI to completion. Prior to electrophoresis, EcoRI digested pBACe3.6 vector was spiked in fixed amounts into the wild-type digests to serve as transgene copy number controls. An EcoRI/KpnI pBACe3.6 fragment was used for random hexamer primed labeling with ³²P-dATP. Blot hybridization and washing was performed by standard methods. Transgene copy number was determined by relative densitometry using the publicly available ImageJ software (NIH).

Results

In this report we compare transgenic expression to the endogenous expression patterns of Hoxa13 in the limbs and genital bud, yet we believe our transgenic expression data reflects the entire Hoxa13 neighborhood of genes (Evx1, Hibadh, Tax1bp1, and Jazf1) (Lehoczky et al., 2004). We use Hoxa13 expression as the representative gene for comparison because of its established necessity in limb and genital development.

BAC 347D13-LacZ is Sufficient for Hoxa13-like Expression

To address the hypothesis that additional sequences in the vicinity of mmA13CNS are needed for appropriate Hoxa13-like distal limb expression, we obtained a ~220 kb mouse BAC clone, RPCI-23-347D13, which overlaps mmA13CNS by a wide margin (~73 kb on one side and ~145 kb on the other) skewed toward the HoxA cluster (Figure 1). We used homologous recombination to place the β-globin-LacZ reporter construct into the BAC vector backbone (Khandekar et al., 2004; Lee et al., 2001; Yu et al., 2000) and used the resultant recombinant BAC, 347D13-LacZ, for transgenic microinjection. E12.5 founder embryos were harvested and three of five transgenic founders had no LacZ expression in any tissue and were thus excluded from the analysis. The two remaining embryos both had LacZ expression in the central nervous system, limbs, and genital bud. The expression pattern found in the hindbrain and neural tube of both embryonic founders was consistent with the original mmA13CNS transgene expression (Lehoczky et al., 2004). We used this central nervous system Evx1-like (not Hoxa13) pattern of expression as a positive control for expression from the injected transgene.

Figure 1.

(Top) To-scale representation of the 1Mb region of mouse chromosome 6 upstream of the HoxA cluster that is under the influence of a long-range distal limb and genital bud enhancer module. Base 1 is the first nucleotide adjacent to the Hoxa11 promoter and base 1041528 is the nucleotide prior to the start of the 5’ UTR of Creb5. Red arrowhead denotes the position of mmA13CNS, found within intron 4 of Hibadh. 347D13 and 420L19 are the two BAC clones used in this report. (Bottom) (A-C) X-gal stained E12.5-E13.0 347D13-LacZ founder embryo 69 and (D) E12.0 – E12.5 347D13-LacZ founder embryo 18. LacZ expression was detected in the distal autopod and genital bud in a Hoxa13-like pattern (E-H). Central nervous system LacZ expression was also found in both transgenic embryos in an Evx1/Evx2-like pattern, albeit at a weaker level in founder embryo 69 (A).

The expressing founder embryos showed Hoxa13-like distal limb and genital bud LacZ expression. Autopod expression in the fore- and hindlimbs of founder 69 was identical to the expected Hoxa13 expression domain (Figure 1B, C, F, G) (Haack and Gruss, 1993; Innis et al., 2004; Lehoczky et al., 2004; Mortlock et al., 1996; Williams et al., 2005). Expression in the genital bud was posteriorly restricted (discussed in later section; data not shown, similar to Figure 4E’). The limb expression in founder 18 was restricted to the posterior autopod (discussed in later section; shown in Supplemental Figure 1), yet genital bud staining was consistent when compared with endogenous expression of Hoxa13 (Figure 1D and H) (Lehoczky et al., 2004; Scott et al., 2005; Stadler et al., 2001). Collectively, these embryos demonstrate that 347D13-LacZ is sufficient for E12.5 Hoxa13-like expression in the distal limb and genital bud.

Figure 4.

Representative E10.5 – E13.5 genital bud LacZ expression common to both lines 347D13-LacZ and 420L19-LacZ. Embryos are pictured with anterior to the left. When any genital bud expression was present at E10.5-E11.5 in embryos from either transgene it was consistently expressed in the same domains (A and B). Between stages E11.5 – E13.5, both BAC transgenes were observed to have expression in the genital bud in two different ways: (C, D, and E) complete Hoxa13-like genital bud expression or (C’, D’, and E’) posteriorized Hoxa13-like genital bud expression.

To further interrogate the ability of 347D13-LacZ to direct Hoxa13-like expression at additional developmental stages, we repeated the transgene injection and generated 3 permanent lines (8, 710, and 733). Lines 8 and 733, while exhibiting the expected central nervous system LacZ staining pattern, had no additional LacZ activity at any stage (data not shown). In contrast, line 710 had strong distal limb and genital bud LacZ expression in addition to the expected central nervous system staining between stages E10.5 and E13.5 (Figure 2A, D, H, L, P, and T). A description of stage-specific LacZ expression in this line follows and the data is summarized in Table 1.

Figure 2.

Timed embryo analysis of LacZ expression in a 347D13-LacZ transgenic mouse line 710. This transgene insertion drives distal limb and genital bud LacZ expression similar, but not identical, to the expression pattern of Hoxa13. All embryos were stained with X-gal and photographed with distal positioned to the top and anterior positioned to the left. Embryos were analyzed at multiple stages: (A-C) E10.5, (D-G) E11.5, (H-K) E12.0, (L-0) E12.5, (P-S) E13.0, and (T-W) E13.5. B, E, I, M, Q, and U, are the forelimbs, photographed at a higher magnification, of embryos in A, D, H, L, R, and T, respectively. Black arrowhead in panel B shows distal autopod expression domain. (U inset) Eosin counter-stained cross-section of X-gal stained forelimb at level of black arrowhead with dorsal to the top and ventral to the bottom. C, F, J, N, R, and V are the stained hindlimbs and G, K, O, S, and W are the stained genital buds (circled).

Table 1.

Summary of LacZ expression from BAC transgenic lines

	CNS	Distal Limbs				Genital Bud
		E10.5	E11.5	E12.5	E13.5	E10.5	E11.5	E12.5	E13.5
347D13-LacZ
8	Y	N	N	N	N	N	N	N	N
18	Y	?	P	?	?	?	Y	?	?
69	Y	?	?	Y	?	?	?	P	?
710	Y	Y	Y	P	P	Y	Y	Y	Y
733	Y	N	N	N	N	N	N	N	N
420L19-LacZ
215	Y	N	P	P	P	Y	Y	P	P
222	N	N	N	N	P	N	N	N	N
237	Y	N	P	P	P	Y	P	P	P
250	Y	N	P	P	P	Y	P	P	P
267	Y	N	P	P	P	Y	P	P	P
268	Y	N	N	P	P	N	N	N	N
273	Y	N	N	N	N	N	N	N	N

Y= Appropriate expression

P= Partial expression

N= No expression

? = Not done

At E10.5, line 710 expresses the 347D13-LacZ transgene in the distal forelimb and the ventral, distal trunk mesenchyme surrounding the hindlimbs, urogenital tissues, and tail. Forelimb expression is largely similar to the endogenous E10.5 RNA expression domain of Hoxa13, with appropriate anterior/posterior boundaries and distal placement (Figure 2B arrowhead). Therefore, 347D13-LacZ is sufficient for E10.5 Hoxa13-like distal forelimb expression. The LacZ expression pattern found in the trunk is not consistent with the known E10.5 expression of Hoxa13 or any of the genes in the Hoxa13 expression neighborhood (Evx1, Hibadh, Tax1bp1, and Jazf1). However, expression does encompass the expected genital bud domain of Hoxa13, so in addition to its additional activity, 347D13-LacZ is sufficient for Hoxa13-like genital bud expression (Figure 2C).

The sufficiency of 347D13-LacZ to drive Hoxa13-like distal forelimb expression is maintained through E11.5 where the LacZ expression remains restricted to the distal boundary and extends to appropriate Hoxa13-like anterior/posterior boundaries of the autopod (Figure 2E). The broad trunk staining from the previous stage becomes strikingly restricted to the mesenchyme of the developing genital bud at E11.5 as is expected for the endogenous expression of Hoxa13. Thus, the transgene is sufficient for E11.5 Hoxa13-like forelimb and genital bud expression. The hindlimb expression in this line is not sufficient for complete Hoxa13-like limb expression (Figure 2F). While the proximal/distal boundaries are appropriate, the anterior/posterior expression boundaries are posteriorized.

For stages E12.0-E12.5 in line 710, 347D13-LacZ is sufficient for Hoxa13-like expression in the developing genital bud (Figure 2K and O). This corroborates the results obtained from founder 18 (Figure 1D). In contrast, the limb staining is not consistent between line 710 and either founder embryo (69 or 18) (Figure 2I, J, M, and N). Instead, this expression, while appropriate at the distal boundary, extends beyond the expected posterior and proximal boundaries. In addition, in comparison to endogenous Hoxa13, the transgenic LacZ expression domain does not completely extend into the anterior autopod. Variation in expression patterns among founders and lines is common in such experiments and can be due to complexity in the BAC transgene insert DNA configuration and/or genomic site effects.

At E13.0-E13.5 347D13-LacZ in line 710 remains sufficient for Hoxa13-like genital bud expression (Figure 2S and W). The expression remains concentrated in the genital bud mesenchyme and has appropriate anterior/posterior boundaries. However, at this stage line 710 does not exhibit complete distal fore- or hindlimb Hoxa13-like transgene expression (Figure 2Q, R, U and V). As was seen with the previous stage, limb staining extends to the distal tip of the autopod but does not reach the expected anterior boundary, and reporter expression extends beyond the expected posterior and proximal boundaries. In addition, at this stage there is minor regression of reporter activity in the anterior/central autopod. It is unclear whether this regression is similar to that of endogenous HOXA13 protein, but a similar regression of reporter expression was observed in the autopods of Hoxa13-GFP mice at similar stages (Stadler et al., 2001). Despite the skewed expression domains on the other axes, the expression on the dorsal/ventral axis is appropriate (Figure 2U inset), as was true with the earlier stages (data not shown). Thus, line 710 demonstrates appropriate early expression of the transgene reporter, however later expression domains are skewed in the autopod along the anterior/posterior and proximal/distal axes. Yet, complete Hoxa13-like expression of this transgene at E12.5-E13.0 in founder 69 suggests sufficiency at this stage for appropriate proximal/distal, anterior/posterior and dorsal/ventral expression.

BAC 420L19-LacZ is Partially Sufficient for Hoxa13-like Expression

In order to better define sequences responsible for directing Hoxa13-like expression, we selected another BAC clone, RPCI-23-420L19 that overlaps with mmA13CNS by a wide margin (~38 kb to one side and ~139 kb to the other), but is skewed further away from the HoxA cluster and overlapping RPCI-23-347D13 by 114 kb (Figure 1). We modified this BAC to contain the β-globin-LacZ minimal reporter in cis and used the resultant construct, 420L19-LacZ, to create transgenic mice. 7 expressing transgenic lines (215, 222, 237, 250, 267, 268, and 273) were used for timed embryo analysis. Embryos from all 7 lines were collected between stages E10.5 – E13.5 and analyzed for LacZ expression. Combined results from all lines revealed this transgene was capable of expressing LacZ in the distal limbs and the genital bud, as well as the expected domains of the central nervous system (Figure 3A, C, G, K, O, and S). As previously mentioned, central nervous system LacZ activity was used to gauge if our transgenic lines (either 347D13-LacZ or 420L19-LacZ) were expressing the incorporated transgene. While this served as a suitable positive control, we still observed inter-line differences in timing/onset, levels, and domains of LacZ expression in the developing genital bud and limbs (Figure 4 and Supplemental Figure 1). While there was inter-line variation, line 215 was representative of the expression timing and distribution of all 420L19-LacZ lines, and thus we present it here as an example of the enhancer capabilities of 420L19-LacZ. We conclude that 420L19-LacZ is sufficient for Hoxa13-like genital bud expression at E10.5 and E11.5, yet insufficient for this complete expression from E12.5-E13.5. We also conclude that this transgene is insufficient for complete Hoxa13-like distal limb expression at any stage. Detailed stage-specific LacZ expression follows and all results are summarized in Table 1.

Figure 3.

Timed embryo analysis of LacZ expression in a 420L19-LacZ transgenic mouse line 215. This transgene is capable of expressing LacZ in a peripheral subdomain of the expected Hoxa13 expression pattern in the autopod and genital bud. All embryos were stained with X-gal and photographed with distal positioned to the top and anterior positioned to the left. Embryos were analyzed at multiple stages: (A and B) E10.5, (C-F) E11.5, (G-J) E12.0, (K-N) E12.5, (O-R) E13.0, and (S-V) E13.5. A, D, H, L, P, and T, are the forelimbs, photographed at a higher magnification, of embryos in A, C, G, K, O, and S, respectively. (T inset) Eosin counter-stained cross-section of X-gal stained forelimb at level of black arrowhead with dorsal to the top and ventral to the bottom. B, E, I, M, Q, and U are the stained hindlimbs and F, J, N, R, and V are the stained genital buds (circled) at these stages. Note there is no expression in forelimbs at stage E10.5.

At E10.5, 420L19-LacZ embryos have no discernible forelimb LacZ expression (Figure 3A). As was seen with 347D13-LacZ, these embryos have a broad region of distal/ventral trunk LacZ expression, encompassing both the presumptive hindlimbs and the urogenital tissues but not the tail (Figure 3B). This expression, while not identical to the expected Hoxa13 genital bud expression, does encompass the expected Hoxa13 expression domain. We conclude that 420L19-LacZ is sufficient to direct expression in the genital mesenchyme and insufficient for forelimb expression at E10.5.

At E11.5, consistent with 347D13-LacZ, 420L19-LacZ reporter expression in the caudal/ventral mesenchyme becomes excluded from the surrounding tissues and is solely expressed in the genital bud (Figure 3F). This property was observed for both transgenes and may represent a true capability of the enhancer module that is used or modified in the endogenous locus in a promoter-specific manner. The expression in the genital bud appears identical to the expected endogenous pattern of Hoxa13 RNA, thus we conclude that 420L19-LacZ is sufficient for Hoxa13-like genital bud expression at E11.5. In contrast, this transgene is insufficient for appropriate Hoxa13-like distal limb expression. LacZ expression is found in both fore- and hindlimbs (Figure 3D and E), yet the domain is posteriorly restricted. In addition to the incorrect anterior/posterior boundaries, the proximal/distal boundaries are also misplaced. The 420L19-LacZ expression domain extends past the appropriate proximal boundary (arrowhead Figure 3D), and does not extend to the distal-most region of the autopod. This transgenic expression domain represents only a small posterior subdomain of the expected Hoxa13 expression pattern, thus we conclude that sequences within 420L19-LacZ are insufficient for Hoxa13-like E11.5 autopod expression.

Between stages E12.0 and E12.5, 420L19-LacZ no longer directs Hoxa13-like genital bud expression. In the transgenic embryos, LacZ expression is only seen in the posterior half of the tubercle (Figure 3J and N), whereas the endogenous expression of Hoxa13 is found throughout the entire genital bud at this stage. In addition, we conclude that this transgene is insufficient for Hoxa13-like distal limb expression between E12.0-E12.5. At E12.0, anterior and distal Hoxa13-like LacZ domains begin to be expressed in addition to a broadening of the posterior expression domain from the previous stage (Figure 3H, I, L, and M). This LacZ expression is in conflict with many aspects of the endogenous Hoxa13 expression pattern. While there is expression in both the anterior and posterior portions of the autopod, there is absence of LacZ expression throughout the central autopod. In addition, the proximal/distal boundaries are inappropriate. LacZ expression never reaches the distal tip of the autopod, leaving a thin strip of non-expressing mesenchyme, and LacZ expression extends too far proximally.

At stages E13.0 and E13.5, the conclusions drawn at E12.5 were corroborated. 420L19-LacZ is insufficient for complete Hoxa13-like genital bud expression (Figure 3R and V). The LacZ staining, while partially appropriate, is posteriorly restricted and only represents half of the endogenous Hoxa13 expression pattern. In addition, the distal limb LacZ expression only represents a portion of the stage-appropriate Hoxa13 expression (Figure 3P, Q, T, and U). While the expression does extend to the correct anterior/posterior and dorsal/ventral boundaries (as is true at all stages, Figure 3T inset), expression throughout the central autopod is absent. In addition, the proximal/distal boundaries are still skewed such that the LacZ expression domain does not extend to the distal-most tip of the autopod, and the proximal boundary extends past the expected boundary into the zeugopod. Because of the ability of this transgene to recapitulate some, but not all, endogenous Hoxa13 expression domains of the autopod, we conclude that 420L19-LacZ is insufficient for Hoxa13-like distal limb expression at any stage. We also conclude that 420L19-LacZ is sufficient for Hoxa13-like genital bud expression at E10.5 and E11.5, yet insufficient for this complete expression from E12.5-E13.5.

Transgene Integrity

Analyzing all of the 347D13-LacZ/420L19-LacZ lines collectively (Supplemental Figures 1 and 2) enabled us to tease out different aspects of the transgenic expression, namely: timing of expression onset, posterior and anterior expression domains, distal and proximal expression boundaries, and central autopod expression domain. To address whether the differences found between our BAC transgenes and between lines of the same transgene were due to technical issues arising from the transgene injection/insertion or to subtleties in the function of the enhancer(s), we pursued experiments to determine BAC copy number and to validate the integrity of the transgenes. By Southern hybridization, we quantified the number of integrated transgenes in fully constituted, N1 transgenic animals. We found that among our transgenic lines, there is no correlation between transgene copy number and level of LacZ expression (Supplemental Table 1).

Because BAC transgene copy number was not playing a significant role in the inter-line expression variation (Supplemental Figure 1), we turned to analyzing the integrity of the incorporated transgene by polymorphic SNP analysis (see Methods). Results from this analysis for 347D13-LacZ transgenic lines correlated well with the LacZ expression. Lines 8 and 733 had LacZ expression in the central nervous system and no additional tissues. Marker analysis showed that in both of these cases the incorporated BAC was fragmented into at least two (line 8) or three pieces (line 733) and large portions of the transgene missing (Supplemental Figure 2). These lines were excluded from the analysis.

Line 710 contained the entire BAC as assessed by markers A-D and F-I (Supplemental Figure 2). This analysis could not be done for the 2 founder embryos. Because circular BAC clones were used as the substrate for microinjection, random BAC breakage must have occurred prior to integration of the transgene into the genome. Our analysis did not determine the structure of the transgene integration and does not rule out a microdeletion between any two of the retained markers upon transgene breakage. Thus the anterior and central autopod expression loss observed late in this line cannot be readily explained by content of transgenic material.

Differences in BAC integrity were also found for the 420L19-LacZ lines. Lines 215, 237, and 267 all had the full complement of BAC markers E-H and J (Supplemental Figure 2). The finding that these three lines largely reflect the same transgenic sequences is consistent with the similar LacZ expression patterns in these embryos. We did not detect any major deletions and we conclude that the lack of central or distal autopod expression, or late onset of the expression are consistent with true insufficiency of 420L19-LacZ to perform these enhancer functions and that the regulatory elements responsible for these domains must lie outside of this clone. Line 273 also contains all of the tested markers, yet no limb or genital bud staining was found for this line. We hypothesize that potential limb and genital bud expression capabilities in this transgene are likely being silenced by the site of integration or the overall integrated structure is not conducive to expression. Line 222 was only positive for markers E and J. The absence of markers F and G indicate a probable deletion of mmA13CNS in this transgene, likely explaining the absence of central nervous system staining in this line.

Discussion

Genital Bud Enhancer(s)

Our data demonstrate that 347D13-LacZ is sufficient for Hoxa13-like genital bud expression at stages E10.5 – E13.5. 420L19-LacZ is sufficient for Hoxa13-like genital bud expression at stages E10.5 - E11.5 (Figure 3). This indicates that the region of intersection of the two overlapping transgenes is sufficient for E10.5 –E11.5 Hoxa13-like genital bud expression (Figure 4). While the genital bud expression from E12.5-E13.5 in the 420L19-LacZ lines was consistently posteriorized, variability was found for the E11.5 expression domain which shows the potential of the transgene to drive two types of expression. Line 215 was sufficient for entire E11.5 genital bud expression, whereas all of the remaining expressing lines had a posteriorized expression domain even in the context of the same transgenic sequence content (similar to Figure 4 C vs. C’). The ability of the same transgene to be expressed in either a complete or incomplete manner indicates the presence of an intact regulatory element whose function is perhaps context dependent. This is further demonstrated by the results from 347D13-LacZ. This transgene was shown to be sufficient for genital bud expression at all stages (line 710 and founder embryo 18), yet founder embryo 69 had a posteriorized genital bud expression domain (similar to Figure 4E’). We conclude that a regulatory element(s) sufficient for Hoxa13-like genital bud expression is contained within the ~114 kb fragment in common between BACs 347D13 and 420L19, yet our data does not exclude redundant enhancer contribution from outside our tested domain. We also conclude that the enhancer(s) within the BACs is sensitive to genomic context and is capable of being expressed in either a Hoxa13-like domain or in a restricted posterior domain between E11.5-E13.5.

Distal Limb Enhancer(s)

We propose a model to combine the expression data we have obtained from our 347D13-LacZ and 420L19-LacZ transgenic embryos that requires enhancers influencing the onset of expression, proximal/distal and anterior/posterior boundaries, and tissue expression subdomains. We considered the commonalities in Hoxa13-related limb expression between the two transgenes, because the enhancer(s) directing that expression should be localized to the overlapping BAC sequences. Generally speaking, both BACs are capable of directing distal autopod expression between E11.5-E13.5. This expression is Hoxa13-like in the anterior/posterior domains, the dorsal/ventral domains, as well as at the proximal boundary. We conclude that an enhancer(s) within this ~114 kb fragment is responsible for E11.5 expression onset and these related expression domains and boundaries.

Our data demonstrates that at least one additional enhancer is located within the ~107 kb fragment of 347D13 that does not overlap with 420L19, though we have not determined whether it functions autonomously or cooperatively with the enhancer(s) in the adjacent ~114 kb fragment. We showed that the 347D13-LacZ transgene was expressed in the autopod at stage E10.5 and it was expressed in the central and distal-most boundary of the autopod. This stage of expression onset and these tissue sub-domains were never observed in any of seven 420L19-LacZ lines, thus our model proposes the existence of an enhancer(s) in this ~107kb region responsible for this expression. Effectively, the remote enhancers for Hoxa13-like limb expression fall within two separate, yet still broad regions. A Hoxa13 regulatory network contingent upon multiple remote enhancers is supported by the recent finding of a similar, multi-component system upstream of the HoxD locus (Gonzalez et al., 2007).

Evolution of Embryonic Distal Limb and Genital Bud Expression Domains

We recognize the experiments in this report are testing the sufficiency of putative regulatory elements that are comparable to, but not proven to be biologically involved in, Hoxa13 gene expression. Our prior work showed that Hoxa13, Evx1, Hibadh, Tax1bp1, and Jazf1 all have distal autopod and genital bud expression from E10.5 through E13.5 (Lehoczky et al., 2004), and we have no direct evidence that the sequences we are mapping are not merely local regulatory elements specific to one of these genes. Yet, we interpret our LacZ reporter data as expression from regional enhancers rather than gene-specific enhancers primarily due to our original HoxA/D paralogous sequence comparison that led us to focus on this region (ParaCNS4 + 5).

The similarities in expression and function of the posterior HoxA/D genes, combined with their common genomic origin, are most parsimonious with the existence of a limb/genital, or more simply a distal appendage, regulatory module prior to the HoxA/D cluster duplication. Under this model, the remote regulation that exists for the extant posterior HoxA and posterior HoxD genes resulted from divergence over the course of vertebrate evolution. Presumably, introduction and/or degradation of genes upstream of the HoxA and HoxD clusters led to the extremely different upstream genomic neighborhoods (Lehoczky et al., 2004), yet the enhancers required for distal limb and genital bud expression were retained. The distinct upstream regions inevitably placed different evolutionary constraints on these pre-existing regulatory sequences. On the HoxD chromosome, the distal limb enhancers were recently mapped to two discrete, remote loci: GCR and Prox (Gonzalez et al., 2007). GCR exhibits sequence conservation to the HoxA upstream region corresponding to mmA13CNS (Lehoczky et al., 2004), but there is no paralogous Prox element in the HoxA locus detectable by standard sequence conservation. Absence of conservation between the Prox HoxD region and any sequence at the HoxA locus does not preclude the existence of a functionally comparable enhancer element at the HoxA locus. One explanation could be that Prox and a functionally paralogous enhancer at the HoxA locus diverged since the HoxA/D cluster duplication thus escaping the sequence conservation thresholds.

GCR and Prox enhancers are proposed to interact with two genes (Evx2 and Lnp) in addition to the posterior HoxD genes. Similarly, on the HoxA chromosome our data supports the existence of at least two discrete enhancers functioning in long-range regulation. Yet distinct from the HoxD regulatory organization, which has remote enhancers positioned within a gene desert, the HoxA limb/genital bud remote enhancers are within a gene neighborhood. The remote HoxA enhancers that we infer from our data appear to interact, bidirectionally, with four other genes (Evx1, Hibadh, Tax1bp1, and Jazf1) in addition to Hoxa13. Future work should identify the smallest functional sequence elements within these broad, remote upstream HoxA domains and explore their function(s) by comparison with the HoxD GCR and Prox regions in transgenic mice or via engineered deletion.