Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. 2002 Jan 22; 99(2): 780–785.
Published online 2002 Jan 15. doi:  10.1073/pnas.012584999
PMCID: PMC117382
Developmental Biology

Genomic structure and functional control of the Dlx3-7 bigene cluster


The Dlx genes are involved in early vertebrate morphogenesis, notably of the head. The six Dlx genes of mammals are arranged in three convergently transcribed bigene clusters. In this study, we examine the regulation of the Dlx3-7 cluster of the mouse. We obtained and sequenced human and mouse P1 clones covering the entire Dlx3-7 cluster. Comparative analysis of the human and mouse sequences revealed several highly conserved noncoding regions within 30 kb of the Dlx3-7-coding regions. These conserved elements were located both 5′ of the coding exons of each gene and in the intergenic region 3′ of the exons, suggesting that some enhancers might be shared between genes. We also found that the protein sequence of Dlx7 is evolving more rapidly than that of Dlx3. We conducted a functional study of the 79-kb mouse genomic clone to locate cis-element activity able to reproduce the endogenous expression pattern by using transgenic mice. We inserted a lacZ reporter gene into the first exon of the Dlx3 gene by using homologous recombination in yeast. Strong lacZ expression in embryonic (E) stage E9.5 and E10.5 mouse embryos was found in the limb buds and first and second visceral arches, consistent with the endogenous Dlx3 expression pattern. This result shows that the 79-kb region contains the major cis-elements required to direct the endogenous expression of Dlx3 at stage E10.5. To test for enhancer location, we divided the construct in the mid-intergenic region and injected the Dlx3 gene portion. This shortened fragment lacking Dlx7-flanking sequences is able to drive expression in the limb buds but not in the visceral arches. This observation is consistent with a cis-regulatory enhancer-sharing model within the Dlx bigene cluster.

The mammalian Dlx family consists of six genes with homeoboxes related to that of Drosophila Distal-less (Dll). The mammalian genes take the form of bigene clusters. The paired genes, termed Dlx2–1, Dlx5–6, and Dlx3–7, are organized in an inverted, convergently transcribed manner (13). It has been proposed that a single Dlx gene duplicated to form an ancestral inverted bigene cluster during the evolution of the early chordates and subsequent duplications gave rise to multiple bigene clusters (4). This hypothesis is based on phylogenetic clustering of Dlx2, 3, 5 to the exclusion of Dlx1, 6, 7. It is likely that the three Dlx clusters underwent duplication together with the Hox gene clusters, because Dlx2–1, 5–6, and 3–7 are closely linked to the Hox D, A, and B clusters, respectively (1, 3, 5). There is a certain degree of spatial and temporal expression overlap between the linked Dlx genes around embryonic stage (E) E10.5 (6). Models are proposed to explain this overlapping pattern, including cis-regulatory sharing between paired genes (7). In this report, we will focus on the Dlx3–7 bigene cluster with the ultimate goal of testing the shared enhancer hypothesis of bigene expression control.

The mouse Dlx3 gene is first expressed weakly in the rostral ectoderm, anterior to the neural plate during the head fold stage (6). Dlx3 is expressed by E8.5 in the ectoplacental cone and chorionic plate, developing into a high level of expression by day E10.5 in the labyrinthine layer of the embryonic placenta (8). Dlx3 is expressed in the first and second visceral arches and fronto-nasal ectoderm at E9.5 (9). Later in development, Dlx3 is also expressed in the external respiratory epithelium of the nares, in whisker follicles, taste bud primordia, dental and mammary gland epithelia, apical ectodermal ridge (AER) of limb buds, genital primordia, and in several additional sites of epithelial-mesenchymal interaction (8, 9). By the end of embryonic development, Dlx3 is down-regulated, except in the skin, where it is transcribed in stratified epidermis and in the matrix cells of the hair follicles (10). The Dlx3 pattern of expression is different in some respects in comparison with the bigene clusters Dlx2–1 and Dlx5–6 in that Dlx3 is not expressed in the central nervous system and is largely expressed in recently acquired structures common to the mammals. Dlx7 has an overlapping expression pattern with the Dlx3 gene in the visceral arches and limb buds before E10.5 (ref. 11 and unpublished data).

The Dlx genes provide an interesting model for the regulation of clustered genes because their arrangement in pairs is the simplest case of gene clustering. The fact that all of the mammalian clusters have the same genomic arrangement suggests that the clusters have not undergone extensive rearrangement since their formation and may help elucidate the fate of cis-regulatory elements after gene duplication. The Dlx3–7 cluster is of special interest to us because its expression pattern lacks the complication of CNS expression. In this report, we provide detailed information on the structural organization of the Dlx3–7 bigene cluster, identify conserved sequence elements in the coding and noncoding domains, and by means of functional transgenic analysis test the enhancer activity for Dlx3–7 expression control.

Materials and Methods

P1 Artificial Chromosome (PAC) Clones and Sequencing.

The clones P1–972 and P1–1490 were isolated from Incyte Genomics (Palo Alto, CA) mouse and human PAC libraries, respectively, by PCR screening of pooled libraries by using primers for the 3′ end of the Dlx3/DLX3 homeodomain (3, 12).

Nucleotide sequences were primarily determined by using the shotgun method, and unsequenced gaps were filled by primer walking (see ref. 13 for methods). The sequence was confirmed by comparing a restriction map deduced from genomic sequence with an experimentally constructed restriction map.

Cluster Alignments and Sequence Comparisons.

Alignments were computed with pipmaker (14) (available at http://bio.cse.psu.edu/pipmaker/). The blast 2 sequences program accessed through the National Center for Biotechnology Information website (15) was also used for identifying local sequence conservation.

Construction of the Mouse Dlx3-lacZ Reporter Gene.

The 80-kb insert of the P1–972 was captured into the pPAC-ResQ vector by using homologous recombination in yeast (16) to make construct Dlx37-A8. In brief, 50 ng of linearized pPAC-ResQ vector and 400 ng of circular PAC P1–972 were cotransformed into yeast strain Y724 (17). More than 60% of the 200 clones obtained were identified as recombinant by hybridization tests by using the Dlx3 5′ upstream region as probe. Yeast genomic DNA from recombinant clones was isolated by using the PureGene kit (Gentra Systems), transformed into Escherichia coli DH10B cells by electroporation (16), and checked for rearrangement by restriction analysis.

The lacZ reporter gene was inserted in-frame into the first exon of the mouse Dlx3 gene by using another round of homologous recombination in yeast. A targeting vector (pLZFSV-Dlx3AB) was generated in which a lacZ-Ura3 cassette (18) was flanked by PCR generated recombinogenic ends (Fig. (Fig.44a) from 5′ of the Dlx3-coding region [primers KL001 (5′-GGT AAC AAC AAA GAG GGT TGA GA-3′) and KL 002 (5′-GAA GGA GCC GCT CAT GCT-3′)], and portions of exon 1 and the downstream intron [primers KL003 (5′-CAC GCG GCC GCC AGC ATC CTC ACC GAC ATC T-3′) and KL 004 (5′-GCG GCG GCC GCA AGC TTC CTC TTC GGA GTG TCG TTC A-3′)]. The resulting pLZFSV-Dlx3AB was linearized and transformed into yeast harboring construct Dlx37-A8. One recombinant clone (construct Dlx37-lacZ-79kb) was transformed into E. coli DH10B and verified by restriction analysis which confirmed the lacZ insertion and lack of genomic rearrangement (Fig. (Fig.44b). The 5′ integration site was sequenced to confirm proper integration. Construct Dlx37-lacZ-19kb was subsequently made by HindIII digestion of construct Dlx37-lacZ-79kb.

Figure 4
Procedure diagram to make the Dlx37-lacZ-79kb construct. (a) Linearized pPAC-ResQ yeast-bacteria shuttle vector captures P1–972 insert by homologous recombination in yeast. Further homologous recombination makes lacZ-URA3 insertion after eighth ...

Production of Transgenic Mice.

Inserts were released from the 9.8-kb vector by digestion of 150 μg with I-Sce I (Boehringer Mannheim) and purified by ultra centrifugation on a 10–40% sucrose gradient. Concentration of dialyzed DNA was adjusted to 4 ng/μl and injected into pronuclei of one-cell mouse embryos as described previously (19). Injected embryos were transferred into the oviduct of pseudopregnant CD1 mice, dissected at appropriate stages, and stained for β-galactosidase activity.

In Situ Hybridization.

Sense and antisense RNA probes were transcribed from a Dlx3 partial cDNA (clone 62895–57). This 757-bp probe contains the most 3′ 27 bp of the homeobox, a further 300 bp of coding sequence, and 468 bp of 3′ untranslated region (positions 538-1295 of GenBank databank accession no. S81932). The protocol for in situ hybridization was modified from that of T. Sanders and C. Ragsdale (personal communication) and is available from the authors on request. Control embryos were hybridized with sense probe verify the specificity of the antisense probe signal.


Genomic Organization of the Dlx3–7 Bigene Cluster.

We obtained human and mouse P1/PAC clones covering the entire Dlx3–7 bigene clusters including flanking regions. The clones were completely sequenced: the human clone termed hP1–1490 measures 75,022 nt, whereas the mouse clone termed mP1–972 measures 78,651 nt.

The general genomic organization of the human and mouse Dlx3–7 bigene clusters was determined by comparisons of genomic and cDNA sequences. The Dlx7 and Dlx3 genes in both species are transcribed convergently and contain three exons in agreement with our earlier studies (3). Thus, we can divide the bigene clusters into three domains: an intergenic 3′ region shared by both genes, a flanking 5′ upstream region unique to the Dlx7 gene, and a flanking 5′ region unique to the Dlx3 gene (Figs. (Figs.11 and and2).2). The intergenic regions measure 17 kb in both species. At least 7 kb of sequence upstream of the Dlx3 translation start site and at least 37 kb upstream of the Dlx7 translation start site are captured in both the human and mouse clones. A close comparison of the human and mouse clones showed no other genes in the P1 clones, although several expressed sequence tag sequences show similarity in the first 20 kb upstream of Dlx7.

Figure 1
Dot plot analysis between mouse Dlx3–7 cluster (P1–972: horizontal line) and human DLX3–7 cluster (P1–1490: vertical line). Black boxes indicate exons. Exon and intron positions of mouse Dlx3 and Dlx7 are color coded as ...
Figure 2
Percentage identity plot of mouse Dlx3–7 cluster (horizontal line) to human DLX3–7 cluster. Vertical axis shows percentage of sequence identity from 50% (bottom) to 100% (top). Large black and gray boxes on the plot represent ...

We compared the mouse Dlx3-7 bigene cluster to its human paralogs, namely, DLX2–1 and DLX5–6 (data not shown). The general genomic organization is similar among the clusters. All show convergent translation, three exons, and 5′-flanking and 3′-intergenic domains. The DLX2–1 and DLX5–6 intergenic regions both measure 10 kb. Sequence similarities are found only in the homeodomains, whereas other coding and noncoding domains show low sequence conservation. The absence of sequence similarity in the noncoding domains is interesting, considering the paralogous nature of the bigene clusters and their overlapping patterns of expression during development.

Protein Sequence Comparisons Between Dlx7 and Dlx3.

The nucleotide sequences of coding regions of the mouse and human exons was subjected to comparative analysis to determine the extent of functional evolutionary constraints on the Dlx7 and Dlx3 proteins. We calculated synonymous and nonsynonymous substitution rates for coding regions of both Dlx7 and Dlx3 genes between human and mouse in each exon by using the method of Nei and Gojobori (20). The result shows that the Dlx3 gene coding sequence is highly conserved, whereas the Dlx7 gene is much less conserved in all three exons (Table (Table1).1). Note that a nonsynonymous to synonymous ratio (dN/dS) of one is predicted for neutrally evolving genes. Both genes in all three exons show ratio values significantly less than one, indicating that Dlx3 and Dlx7 are under functional constraint. However, all three exons of Dlx7 have ratio values much higher than Dlx3, indicating that the Dlx7 protein is evolving more rapidly than the Dlx3 protein. We will consider the implications of this situation in the Discussion.

Table 1
Genetic distances of synonymous and nonsynonymous substitutions between human and mouse Dlx3 and Dlx7 genes computed by using Nei and Gojobori's method

Noncoding Genomic Domain Comparisons Between the Mouse and Human Dlx3–7 Bigene Clusters.

We carried out several types of homology plot analyses between the human and mouse Dlx3–7 bigene clusters to identify putative cis-regulatory elements. We predict that cis-regulatory elements will be highly conserved unless the gene expression patterns have changed dramatically during species divergence. First, we performed dot plot analysis between mouse and human (Fig. (Fig.1)1) and observed similarities in most regions except for the mouse 12- to 21-kb region (Fig. (Fig.2)2) equivalent to the human map position 8–15 kb. Genomic rearrangement may have taken place in this region.

Second, we used percentage identity plot analysis to compare mouse (reference sequence) and human genomic sequences to find shared conserved motifs (Fig. (Fig.22).

An interesting aspect of the sequence comparison analysis is the striking conservation of a 30-kb region at mouse map position 44–74 kb that includes the Dlx7 and Dlx3 genes (Fig. (Fig.2).2). There are three strong matches of ≈85% similarity between human and mouse in the flanking domains (Fig. (Fig.2;2; red color code). There are also five conserved noncoding regions in the Dlx3–7 intergenic region (Fig. (Fig.2,2, orange color code, and Fig. Fig.3).3). Element I37–1 is the most highly conserved element with 90% identity between human and mouse. Element I37–5 appears to be a partial duplication of I37–1 with 80% identity >40 bp (Fig. (Fig.3).3). Element I37–1 also shows similarity with an element in the zebrafish dlx3–7 intergenic domain (M. Ekker, unpublished data). The three remaining conserved elements, I37–2, I37–3, and I37–4 have identities of 88%, 82%, and 87%, respectively, between human and mouse (Fig. (Fig.3).3). All of these noncoding genomic regions are more highly conserved than the exon 1 coding domains (Fig. (Fig.2).2).

Figure 3
Diagram showing locations and information of five highest sequence similarities in the human and mouse Dlx3–7 intergenic region (from I37–1 to 5). Numbers below are length of conservation and the percentage of sequence identity within ...

Functional Properties of Intergenic Domain Conserved Elements.

We carried out functional characterization of the intergenic putative control elements to explore their specific regulatory features. The P1–972/79-kb insert was transferred into a yeast-bacterial shuttle vector, pClasper, by homologous recombination in yeast cells (17) (Fig. (Fig.44a). Homologous recombination was also used to insert a lacZ reporter gene together with a ura3 yeast selection marker in-frame into the first exon of the Dlx3 gene. The transfer of the large insert into pClasper and its subsequent modification into a reporter construct was successfully accomplished without apparent rearrangement (Fig. (Fig.44b). The resulting Dlx37-lacZ-79-kb reporter was excised from pClasper and injected into the pronuclei of mouse zygotes to produce transgenic animals. Two types of transgenic mice were studied: “transient” transgenic animals sampled after injection at various developmental stages, and “stable” transgenic animals obtained from stably transformed lines derived from independent founder transformants. The stable transformants are advantageous in that they can be analyzed reproducibly. We analyzed four transient embryos and five independently derived stable transformed lines.

The expression pattern of the lacZ reporter in the transient and stable embryos was largely similar but differed in some instances in signal intensity (Fig. (Fig.5).5). Reporter expression is found in fore and hind limb buds and tail bud at stage E9.5 (Fig. (Fig.55bB). Expression is also seen in the first and second visceral arches. Expression is observed in the AER of the limb buds and some ventral ectodermal cells on the fore and hind limb buds at stage E10.5. Strong expression is found in the mesenchyme of the distal caudal portion of the mandibular process of the first visceral arch and the distal lateral portion of the second visceral arch (Fig. (Fig.55bE). Both E9.5 and E10.5 embryos show similar reporter expression patterns in limbs and visceral arches in comparison to the endogenous patterns of expression as determined by in situ hybridization (Fig. (Fig.55 bA and bD). We conclude from these results that a considerable proportion of the endogenous expression of the Dlx3 gene is reconstituted by the Dlx37-lacZ-79-kb reporter construct, implying that critical control elements necessary for the appropriate expression of Dlx3 in the visceral arches and limb buds are present in the construct.

Figure 5
(a) Diagram of transgenes and summary of reporter expression patterns in transgenic animals. Blue boxes and open circles indicate exons and conserved motifs, respectively. Dlx37-lacZ-79kb has the full-length of P1–972, whereas Dlx37-lacZ-19kb ...

Other regions show differences between reporter and endogenous patterns of expression. First, expression in frontonasal ectoderm and rostral midline ectoderm of the first arch is detected in in situ hybridization experiments (Fig. (Fig.55 bA and bD) but not in transgenics (Fig. (Fig.55 bB and bE). Second, some transgenic embryos show LacZ expression on the dorsal midline, which is not detected by in situ hybridization. One transient embryo and one stable transgenic line show pronounced expression on the dorsal midline and in the cranio-facial region. Interestingly, the stable transformant animals in this line show a neural tube closure defect with 50% penetrance. We have not determined the reason for this anomaly; however, it may result from the over- or misexpression of Dlx7, which is not disabled in the construct.

In an effort to discover specific control elements in the Dlx3-7 bigene cluster, we initiated experiments in which subregions of the Dlx37-lacZ-79kb reporter construct are tested for expression properties in transgenic mice. In an initial experiment, we have tested a subconstruct, Dlx37-lacZ-19kb that lacks Dlx7 and its flanking elements. This reporter construct contains the Dlx3-coding exons with lacZ insertion and intergenic conserved elements I37–1 and I37–3 (Fig. (Fig.55a). We obtained three transient and two stable transgene lines with this construct. Once again, the expression patterns between the independent founders are highly consistent (Fig. (Fig.55 bC and bF). Expression is seen in the AER of both fore and hind limb buds at stages E9.5 and E10.5. Moreover, the pattern is highly similar in this respect to that of the Dlx37-lacZ-79kb construct. However, there is a striking difference in that there is no detectable expression of the Dlx3 reporter in the first and second visceral arches. This result provides strong initial evidence that the conserved elements proximal to Dlx7 are necessary for visceral arch expression of Dlx3.


The Dlx3–7 bigene cluster is one of three Dlx clusters in mammals. These clusters share a convergent transcriptional orientation and are linked to the Hox clusters, suggesting that they are all derived from cluster duplications subsequent to an initial tandem duplication event (4).

Nonsynonymous to synonymous rate analysis, between human and mouse protein sequences, shows that Dlx3 is more evolutionarily conserved than Dlx7 (Table (Table1).1). There are two possible explanations. In the first, Dlx3 has retained more of its original functions since the two genes duplicated, whereas Dlx7 has been evolving more free of functional constraints. In the second, Dlx7 is evolving under directional selection. Some evidence points to important differences in function of the two genes despite partially overlapping patterns of expression. First, Dlx7 appears to have an important role in hematopoeisis (3), which has not been found for Dlx3. Both genes, on the other hand, are expressed in the placenta, but Dlx7 cannot compensate for a loss-of-function Dlx3 mutation, which causes embryonic death because of placental failure (8). The divergence in the Dlx7 protein sequence between human and mouse may relate to as yet unknown changes in protein function.

A comparison between human and mouse Dlx3–7 bigene clusters reveals striking conservation in noncoding regions, including the 5′ upstream promoter regions of both genes and the 3′ downstream intergenic region common to both. We have focused on the intergenic region because we hypothesize that this region might be the site of critical shared enhancers produced in the tandem duplication that created the bigene cluster (7). There are five putative control elements that show a high level of sequence similarity between human and mouse in the intergenic region (Fig. (Fig.3).3). Moreover, element IG73–1 shows a sequence match to an element in the corresponding zebrafish dlx3–7 bigene cluster. The zebrafish has additional Dlx7 gene, termed dlx8, in comparison to mammals. One of these, designated dlx8, is actually an ortholog of the mammalian Dlx7 gene. Therefore, it is possible that additional conserved elements between mammals and zebrafish might exist between the mammalian Dlx3–7 cluster and the zebrafish dlx8 gene as a consequence of cis-element subfunctionalization between duplicated gene clusters (21).

We have made sequence comparisons, inter se, among the three Dlx bigene clusters found within mammalian species. We find significant similarity only in the homeodomains but none in other coding or the noncoding regions. We find this anomalous because all of the clusters are derived from a common ancestor and the expression patterns of all genes partially overlap. How might expression patterns be conserved, in part, while control sequences undergo major modification after cluster duplication? We postulate two interrelated processes. In the first, termed “fine-tuning”, the transcription factor binding elements that themselves cluster in modules or enhanceosomes undergo loss, gain, or modification randomly, but some of these changes are advantageous, and are selected or drift to fixation, resulting in novel adaptive patterns of expression (22). In the second process, termed “compensation”, sequence changes occur randomly and are not immediately eliminated by purifying selection because of redundancy or buffering in the overall enhancer module. Constant expression patterns are maintained by balancing selection for subsequent compensatory changes, as proposed by Kreitman and coworkers (2326). We hypothesize that both these processes are occurring in the Dlx bigene clusters. Thus, sequence modifications are constantly introduced, erasing the similarity in the sequences as originally inherited from the common ancestral cluster. Fine tuning in control elements results in functionally important changes in individual gene or cluster expression patterns while compensation maintains the overall similarities in the overlapping patterns of expression between the genes.

We report here initial transgenic experiments to directly ascertain the function of noncoding conserved elements. A reporter construct consisting of a 79-kb genomic fragment with a LacZ insertion in exon 1 of the Dlx3 gene is capable of reconstituting expression in tail bud, limbs, and visceral arches in a manner closely resembling endogenous Dlx3 expression in stage E9.5 and E10.5 embryos (Fig. (Fig.5).5). This finding supports the view that most of the cis-elements required for proper Dlx3 expression are contained in the construct. In the visceral arches, the Dlx37-lacZ-79kb construct can initiate mesenchymal expression in the distal tip of the arches at E9.5 where postmigratory neural crest cells are present. At stage E10.5, reporter expression is restricted to the caudal portion of the mandibular process in conformity with endogenous expression (9). In the fore and hind limb buds, expression is detected at E9.5. At this developmental stage, the reporter is expressed in a broad area of lateral epithelium and weakly in the AER. Expression is restricted to the AER later at E10.5. Reporter expression is absent in the fronto-nasal process although Dlx3 transcripts are detected by in situ hybridization. This may be explained by the absence of certain control elements in our construct. In particular, the Dlx3 5′-flanking portion of Dlx37-lacZ-79kb is relatively short, being only 7 kb in length.

The Dlx bigene clusters are notable in that the two genes have overlapping patterns of expression ≈E10.5. Ellies et al. (6) proposed that the similarities in expression pattern within a cluster are due to enhancer sharing between the linked genes, and they put forward three hypothetical models of the original tandem gene duplication to explain similarity in expression patterns. In model A, a single control element upstream of one gene can regulate both genes similarly. In model B, a single control element located in the intergenic region regulates both genes similarly. In model C, separate elements located upstream of each gene regulate both similarly. Models A and B are consistent with enhancer sharing between Dlx7 and Dlx3. Our transgene results with the construct Dlx37-LacZ-19kb rule out model C because visceral arch expression of Dlx3 is lost when cis-elements proximal to Dlx7 are deleted. This result argues in favor of enhancer sharing. We will reexamine this question by transgenic experiments in which the putative enhancer elements are deleted singly and in combination.


We thank Marc Ekker for sharing zebrafish sequence data and Günter Wagner for insightful comments on the manuscript. This research was supported by National Institutes of Health Grant GM 09966 and National Science Foundation Grant IBN-9905403 (to F.H.R.); National Science Foundation/Sloan Postdoctoral Fellowship in Molecular Evolution (to S.Q.I.); and Japan Society for the Promotion of Science Postdoctoral Fellowship for Research Abroad (to K.S.).


Eembryonic stage
AERapical ectodermal ridge
PACP1 artificial chromosome


1. Simeone A, Acampora D, Pannese M, D'Esposito M, Stornaiuolo A, Gulisano M, Mallamaci A, Kastury K, Druck T, Huebner K, et al. Proc Natl Acad Sci USA. 1994;91:2250–2254. [PMC free article] [PubMed]
2. McGuinness T, Porteus M, Smiga S, Bulfone A, Kingsley C, Qiu M, Liu J, Long J, Xu D, Rubenstein J. Genomics. 1996;35:473–485. [PubMed]
3. Nakamura S, Stock D W, Wydner K L, Bollekens J A, Takeshita K, Nagai B M, Chiba S, Kitamura T, Freeland T M, Zhao Z, et al. Genomics. 1996;38:314–324. [PubMed]
4. Stock D W, Ellies D L, Zhao Z, Ekker M, Ruddle F H, Weiss K M. Proc Natl Acad Sci USA. 1996;93:10858–10863. [PMC free article] [PubMed]
5. Rossi E, Faiella A, Zeviani M, Labeit S, Floridia G, Brunelli S, Cammarata M, Boncinelli E, Zuffardi O. Genomics. 1994;24:34–40. [PubMed]
6. Quint E, Zerucha T, Ekker M. J Exp Zool. 2000;288:235–241. [PubMed]
7. Ellies D L, Stock D W, Hatch G, Giroux G, Weiss K M, Ekker M. Genomics. 1997;45:580–590. [PubMed]
8. Morasso M I, Grinberg A, Robinson G, Sargent T D, Mahon K A. Proc Natl Acad Sci USA. 1999;96:162–167. [PMC free article] [PubMed]
9. Robinson G W, Mahon K A. Mech Dev. 1994;48:199–215. [PubMed]
10. Beanan M, Sargent T. Dev Dyn. 2000;218:545–553. [PubMed]
11. Zhao Z, Stock D W, Buchanan A V, Weiss K M. Dev Genes Evol. 2000;210:270–275. [PubMed]
12. Weiss K M, Ruddle F H, Bollekens J. Connect Tissue Res. 1995;32:35–40. [PubMed]
13. Kawasaki K, Minoshima S, Nakato E, Shibuya K, Shintani A, Asakawa S, Sasaki T, Klobeck H-G, Combriato G, Zachau H G, et al. Eur J Immunol. 2001;31:1017–1028. [PubMed]
14. Schwartz S, Zhang Z, Frazer K A, Smit A, Riemer C, Bouck J, Gibbs R, Hardison R, Miller W. Genome Res. 2000;10:577–586. [PMC free article] [PubMed]
15. Tatusova T A, Madden T L. FEMS Microbiol Lett. 1999;174:247–250. [PubMed]
16. Bhargava J, Shashikant C, Carr J, Bentley K, Amemiya C, Ruddle F. Genomics. 1999;56:337–339. [PubMed]
17. Bradshaw M S, Bollekens J A, Ruddle F H. Nucleic Acids Res. 1995;23:4850–4856. [PMC free article] [PubMed]
18. Bradshaw M S, Shashikant C S, Belting H-G, Bollekens J A, Ruddle F H. Proc Natl Acad Sci USA. 1996;93:2426–2430. [PMC free article] [PubMed]
19. Shashikant C S, Bieberich C J, Belting H G, Wang J C, Borbely M A, Ruddle F H. Development (Cambridge, UK) 1995;121:4339–4347. [PubMed]
20. Nei M, Gojobori T. Mol Biol Evol. 1986;3:418–426. [PubMed]
21. Force A, Lynch M, Pickett F B, Amores A, Yan Y, Postlethwait J. Genetics. 1999;151:1531–1545. [PMC free article] [PubMed]
22. Tautz D. Curr Opin Genet Dev. 2000;10:575–579. [PubMed]
23. Kreitman M, Ludwig M. Semin Cell Dev Biol. 1996;7:583–592.
24. Ludwig M Z, Kreitman M. Mol Biol Evol. 1995;12:1002–1011. [PubMed]
25. Ludwig M Z, Patel N H, Kreitman M. Development (Cambridge, UK) 1998;125:949–958. [PubMed]
26. Ludwig M Z, Bergman C, Patel N H, Kreitman M. Nature (London) 2000;403:564–567. [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try


Save items

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...