cis-regulatory mutations are a genetic cause of human limb malformations

Authors

  • Julia E. VanderMeer,

    1. Department of Bioengineering and Therapeutic Sciences, and Institute for Human Genetics, University of California San Francisco, San Francisco, California
    Search for more papers by this author
  • Nadav Ahituv

    Corresponding author
    1. Department of Bioengineering and Therapeutic Sciences, and Institute for Human Genetics, University of California San Francisco, San Francisco, California
    • 513 Parnassus Avenue, HSE901H Box 0794, San Francisco, CA 94143
    Search for more papers by this author

Abstract

The underlying mutations that cause human limb malformations are often difficult to determine, particularly for limb malformations that occur as isolated traits. Evidence from a variety of studies shows that cis-regulatory mutations, specifically in enhancers, can lead to some of these isolated limb malformations. Here, we provide a review of human limb malformations that have been shown to be caused by enhancer mutations and propose that cis-regulatory mutations will continue to be identified as the cause of additional human malformations as our understanding of regulatory sequences improves. Developmental Dynamics 240:920–930, 2011. © 2011 Wiley-Liss, Inc.

HUMAN LIMB MALFORMATIONS

Congenital limb malformations are the second most common type of human malformation, occurring in as many as 1 in 500 births (Moore and Persaud,1998). These malformations can range in severity from limb truncations to changes in digit number or morphology. Limb malformations can be caused by genetic mutations or environmental factors such as exposure to teratogens or physical constraints including amniotic bands and vascular disruptions. The genetic etiology of limb malformations is important for genetic counseling and may be helpful in determining the best treatment when multiple options are available, as is the case with clubfoot where the standard of care is nonsurgical treatment consisting of casting and bracing, but where severe cases can require multiple surgeries (Alvarado et al.,2010). Uncovering the mutations that cause human limb malformations also provides insight into tetrapod limb development by highlighting important genes and pathways. Although human limb malformations have been studied since the early 19th century (Farabee,1903), identifying the causal mutations has been a challenge. Of the gene mutations that have been found, the majority lead to syndromes comprised of a range of other phenotypes in addition to limb malformations (Schwabe and Mundlos,2004). Compared with syndromic limb malformations, the genetic causes of isolated limb malformations have proven even more difficult to discover. In this article, we review the growing evidence that isolated malformations can be caused by changes to gene regulatory sequences that affect the expression of important developmental genes.

BACKGROUND ON cis-REGULATORY ELEMENTS

One of the revelations to come out of the publication of the complete human genome sequence was the fact that it contains only approximately 20,500 protein-coding genes (Clamp et al.,2007). This number reinforced the idea that many genes must serve multiple roles, which is especially evident in genes involved in early embryonic development and tissue patterning. In order for these roles to be executed at the proper time and place, the expression of these developmental genes must be tightly regulated. Some of this regulation occurs through the actions of cis-regulatory elements—sequences of noncoding DNA that control the expression of nearby genes. Many different types of cis-regulatory elements have been identified and while this review cannot adequately cover all of them, we will provide a brief introduction to the most established elements with references to more detailed reviews.

The best-characterized cis-regulatory element is the promoter (Fig. 1). The promoter is a region at the beginning of the gene that serves as the site where transcription machinery assembles and transcription of the gene begins. The promoter region can also include a nearby cis-regulatory element known as the proximal promoter that extends up to a few hundred base pairs upstream of the transcription start site and include many transcription factor binding sites (TFBS). For a more detailed view of promoters and gene regulation, see Smale and Kadonaga (2003).

Figure 1.

Diagram of cis-regulatory elements described in this review. Two different promoters are located next to the two different genes. By binding to Promoter 1, an enhancer actively regulates Gene 1 and leads to its transcription, as indicated by the black arrow above Gene 1. An insulator prevents this enhancer from activating Gene 2, which is not transcribed due to regulation by the silencer.

The cis-regulatory element that is the primary focus of this review is the enhancer (Fig. 1). Enhancers are DNA sequences that up-regulate gene expression in a specific temporal and spatial manner. They often have modular expression patterns and a gene that is active in many tissues is likely influenced by multiple enhancers (Visel et al.,2009a). Enhancers can be located as much as 1 megabase upstream or downstream of the gene they regulate, within the regulated gene's intron or within introns of other genes. Enhancer function is usually considered to be independent of location or orientation relative to the gene. Developmental enhancers are often conserved through evolution, but this is not always the case (Blow et al.,2010). In the tissue context where they are active, enhancers usually show a characteristic pattern of histone modifications including monomethylation at both H3K4 and H2BK5, low H3K4 tri-methylation, H3K27 acetylation and enrichment of the H2A.Z histone variant (Barski et al.,2007; Heintzman et al.,2009). Enhancers are thought to function through the recruitment of transcription factors (TFs) and subsequent physical interactions with the gene promoter. Detailed overviews of enhancers can be found elsewhere (Maston et al.,2006; Levine,2010).

A related cis-regulatory element is the silencer (Fig. 1), which acts to repress gene expression. Like enhancers, silencers are thought to be orientation-independent and can be located almost anywhere with regard to the genes that they regulate. Silencers are thought to act by recruiting repressor proteins, which can influence the local chromatin or by inhibiting the recruitment of activating factors (Maston et al.,2006; Cecchini et al.,2009; Perissi et al.,2010).

Insulators, also known as boundary elements, are cis-regulatory regions that prevent the transcriptional activity of one gene from affecting neighboring genes (Fig. 1). The mechanisms by which insulators delineate regions of active gene transcription is unknown, but involves limiting the range of permissive chromatin modifications. One protein that was shown to be important for insulator activity is the CCCTC-binding factor (CTCF), which is thought to insulate by modifying DNA methylation (West et al.,2002; Maston et al.,2006).

TECHNIQUES FOR STUDYING cis-REGULATORY ELEMENTS

One major challenge in the study of cis-regulatory elements is our relatively poor ability to detect and characterize them. Currently, other than the promoter, the most studied element is the enhancer. Enhancers can be identified using various techniques. One such method is the enhancer trap; the integration of a reporter gene along with a minimal promoter (a promoter that is not sufficient to drive reporter expression without the presence of a functional enhancer) randomly in the genome followed by a screen for the reporter gene expression (Korn et al.,1992; Bellen,1999; Parinov et al.,2004). Two drawbacks of this method are that it does not pinpoint the exact location of the enhancer or the identity of the gene that it regulates.

A more efficient way to identify enhancers is by using comparative genomics. By aligning genomes of different species, it is possible to identify noncoding regions with a high degree of conservation, suggesting a functional role for those regions. Depending on the degree of conservation and the species used for the alignments, ∼10–60% of these conserved noncoding elements are usually shown to function as enhancers (Woolfe et al.,2005; Pennacchio et al.,2006). Though this approach can identify some enhancers, there is strong evidence that not all enhancers can be detected by evolutionary conservation alone (Blow et al.,2010). Another approach used to detect enhancers focuses on the interactions with TFs either by computationally identifying regions where the TFBS sequences appear at some frequency in specific combinations, or by directly assaying for TF-DNA interactions through chromatin immunoprecipitation (ChIP; Vokes et al.,2008; Gotea et al.,2010). A final common approach is to use more general properties of cis-regulatory elements to detect them in a genome-wide manner, using ChIP in combination with microarrays (ChIP-chip) or next-generation sequencing technologies (ChIP-seq). ChIP-chip and ChIP-seq have been used to identify regions bound by enhancer-associated proteins like p300 (Visel et al.,2009b) or displaying chromatin modifications associated with enhancers like H3K4 me1 or H2BK5me1 (Barski et al.,2007; Heintzman et al.,2009). ChIP-based methods have the advantage of identifying tissue-specific enhancers and the ability to detect enhancers that do not show significant evolutionary conservation (Visel et al.,2009b; Blow et al.,2010).

A major reason that enhancers are well studied compared with other cis-regulatory elements is that the functional assays to characterize them are straightforward, highly reproducible and used by many researchers in the field. These assays are based on a common design: the putative enhancer is placed in a vector that contains a reporter gene controlled by a minimal promoter. This construct allows the enhancer to direct the expression of the reporter gene. This type of assay is commonly used in cell culture systems with a luciferase reporter gene (Heintzman et al.,2007), in mouse embryos with a LacZ reporter (Pennacchio et al.,2006) and in zebrafish with a green fluorescent protein reporter (Fisher et al.,2006). Each system has its advantages—the high throughput capacity of cell culture assays, the similarity to human development of the mouse model, and the relative ease and temporal tracking of reporter expression in zebrafish—but these systems share a common flaw. None of these assays is truly able to show quantitative enhancer activity in vivo where the enhancer normally functions.

Many attempts are being made to study other regulatory elements using similar principles. As is the case for enhancers, silencers and insulators are not readily identifiable from sequence data alone. Insulators have been identified through ChIP-chip and ChIP-seq studies looking for CTCF binding regions in specific cell types (Kim et al.,2007; Cuddapah et al.,2009). Functional assays for in vivo silencer or insulator activity have proven difficult to design and while some assays have been developed (Petrykowska et al.,2008; Bessa et al.,2009), there are currently no robust in vivo functional assays for these elements in widespread use.

THE MODULAR NATURE OF cis-REGULATORY ELEMENTS AND HUMAN DISEASE

Many developmental genes play key roles in different tissues and at different developmental stages. It is likely that the phenotype caused by a coding mutation in one of these genes would be different from the phenotype caused by a change to a cis-regulatory element that controls one aspect of that gene's expression. While the gene mutation could affect all tissues where the gene is active, the regulatory mutation would likely be less severe or involve only a subset of the phenotypes. This hypothesis is based on the proposed modular nature of cis-regulatory elements and suggests that a malformation that occurs in both syndromic and isolated forms would be a good phenotype to study; the isolated malformation cases could be caused by regulatory mutations. Human limb malformations occur in both these forms and are relatively common. The depth of knowledge we already have of how tetrapod limbs normally develop and of normal gene expression patterns makes it relatively simple to recognize when genes in early limb development are expressed abnormally. These factors make isolated human limb malformations a good phenotype in which to search for cis-regulatory mutations.

LIMB DEVELOPMENT AND EARLY TISSUE PATTERNING

Many human limb malformations can arise from defects in the early stages of limb tissue patterning and development. Normal limb development requires the coordinated determination of three axes: anteroposterior (AP), proximodistal (PD), and dorsoventral (DV). Much has been learned about the genes that control these axes and the networks through which they interact. The AP axis is the earliest axis to be defined in the developing limb bud. A critical signaling center known as the zone of polarizing activity (ZPA) in the posterior mesenchyme of the limb bud was shown in the 1960s to be responsible for defining the AP axis (Saunders and Gasseling,1968). The ZPA expresses Sonic Hedgehog (SHH), which is responsible for its ability to pattern the AP axis. The PD axis is controlled by the apical ectodermal ridge (AER), a small thickening of cells at the distal margin of the limb bud. The AER produces growth factors and signals that are required for the PD growth of the limb. A feedback loop is established between the ZPA and AER that maintains them both and allows for coordinated growth along both axes and maintenance of DV polarity.

Studies from classical embryology show that disruption of these signaling centers causes changes in limb morphology. Grafting a secondary ZPA, or an alternative source of SHH signal, to the anterior region of a chick limb bud can induce the development of supernumerary preaxial digits that develop as a mirror image to the normal digits (Riddle et al.,1993). Similarly, disrupting the AER causes problems with development along the PD axis, resulting in limb truncations (Summerbell,1974). Given our understanding of the basic roles of these signaling centers in normal limb development, it is clear that disruption of the primary patterning of limb axes can cause limb malformations.

HUMAN LIMB MALFORMATIONS AND THE ZRS

One of the most studied developmental enhancers is the enhancer known as the ZPA regulatory sequence (ZRS), which controls the expression of SHH in the developing limb bud. This enhancer was discovered through a combination of mouse models and the study of human patients with preaxial polydactyly. More complete reviews of this important regulatory sequence can be found (Lettice and Hill,2005; Hill,2007), but because of its prominence in studies of human limb malformations, we include an overview.

Because ectopic expression of Shh in the anterior part of the limb bud is able to induce ectopic digits, many scientists began to study its expression in mouse mutants that have preaxial polydactyly. In several of these mice, the extra digits grow as a mirror image to the normal digits, similar to what was seen from chicken ZPA-grafting experiments. Unsurprisingly, many of these mice were found to have Shh expression that extended far beyond the normal posterior ZPA and even formed a second anterior ZPA (Chan et al.,1995; Masuya et al.,1995; Blanc et al.,2002). Some of these mice were discovered to have defects in genes that function upstream of Shh to restrict its expression (xt mutant, Hui and Joyner,1998; lst mutant, Qu et al.,1998). In other mice, the phenotype was mapped to the Shh locus, but no Shh coding mutations were found (Sharpe et al.,1999). One such mouse is the Sasquatch mutant (ssq), which was the result of a transgenic insertion that created a 20 kilobase (kb) duplication within intron 5 of the gene limb region 1 (Lmbr1; Lettice et al.,2002). The ssq mutant showed not only ectopic Shh expression, but also expression of a similar pattern for the transgene, suggesting the presence of an endogenous regulatory element in the genomic region where it integrated.

The homologous human LMBR region was also recognized to be important in limb patterning through studies of human patients with preaxial polydactyly (PPD), an extra digit on the anterior side of the hand or foot. Through linkage analysis, PPD was mapped in several families to a region of approximately 500 kb on chromosome 7 (Fig. 2; Heus et al.,1999). The identification of a de novo translocation in a patient with PPD allowed Lettice et al. (2002) to map the translocation breakpoint to a region within intron 5 of LMBR1; the same region disrupted by the mouse ssq transgene insertion. A highly conserved region in this intron was found to have enhancer activity that directs Shh expression in the posterior limb bud (Fig. 3A) and was thus named the ZPA regulatory sequence (ZRS). The ZRS has since been found to be critical for limb development (Sagai et al.,2005) and to harbor mutations that cause polydactyly in additional mouse models (Lettice et al.,2003; Sagai et al.,2004; Masuya et al.,2007) as well as chickens (Maas and Fallon,2004), dogs (Park et al.,2008) and cats (Lettice et al.,2008).

Figure 2.

Genomic context of zone of polarizing activity regulatory sequence (ZRS) mutations and duplications. A: The ZRS Duplications track shows duplications that include the ZRS and that lead to complex polysyndactylies. Each bar in the track indicates the region that was duplicated in a family with these malformations. While the duplicated regions vary between families, they all include the ZRS. B: The ZRS Mutations track shows point mutations that are distributed throughout a ∼750 base pair conserved ZRS region that cause human limb malformations. Genomic coordinates and citations for duplications and SNPs can be found in Table 1. Both UCSC Genome browser tracks are available from http://bts.ucsf.edu/ahituv/limb.html. Images generated using the UCSC genome browser, http://genome.ucsc.edu/ (Kent et al.,2002).

Figure 3.

Phenotypic severity of human limb malformation is correlated with the degree of misexpression determined by the specific ZRS mutation. A: Wild-type (wt) human zone of polarizing activity regulatory sequence (ZRS) enhancer expression is restricted to the ZPA. B: In a mouse enhancer assay, the Cuban mutation shows strong anterior expression (arrowheads) and an expanded posterior expression domain (arrows). C: The Cuban mutation leads to a more severe limb malformation, including bilateral tibial aplasia in addition to polydactyly and triphalangeal thumb in this patient. D: The Belgian2 mutation leads to weak enhancer expression in the anterior portion of the developing limb (D′ arrow). E: The Belgian 2 mutation causes a milder human phenotype with triphalangeal thumbs on both hands (arrows) and an additional digit on the left hand (arrowhead). Images used with permission from (Zguricas et al.,1999; Lettice et al.,2003,2008).

Subsequent work on human patients with isolated limb malformations has revealed many mutations affecting the ZRS (Table 1; Fig. 2). The mutations identified in these patients consist of either single base pair substitutions or duplications that encompass the ZRS region. The number of mutations found in this region is certainly due in part to the focus it has received, but the diverse phenotypes associated with these mutations highlights the complexity of interpreting enhancer mutations. Phenotypes caused by changes to the ZRS include those limited to the autopod, including various forms of PPD and syndactyly, and those that include more proximal limb malformations such as tibial hypoplasia and Werner mesomelic syndrome (OMIM # 188770).

Table 1. Enhancer Defects Known to Cause Limb Malformations in Human Patients
Mutation NameMutationLocation (hg19)PhenotypeReference
BMP2 limb enhancer
 Family 1, Datheduplication∼chr20:6,860,129-6,866,024Brachydactyly type A2Dathe et al.,2009
 Family 2, Datheduplication∼chr20:6,860,477-6,866,024Brachydactyly type A2Dathe et al.,2009
DLX5/6 BS1 enhancer (∼chr7:96,357,368-96,357,92)
 Patient, Kouwenhovendeletion∼chr7:95,552,064-96,432,064Split hand/foot malformation1Kouwenhoven et al.,2010
SHH ZRS enhancer (∼chr7:156,583,562-156,584,711)
 739 A>G, Family A,CSNPchr7:156,583,831Preaxial polydactyly & triphalangeal thumbGurnett et al.,2007
 621 C>G, Family BSNPchr7:156,583,949Preaxial polydactyly & triphalangeal thumbGurnett et al.,2007
 463 T>GSNPchr7:156,584,107Preaxial polydactyly & triphalangeal thumbFarooq et al., 2010
 404 G>C, Family 2SNPchr7:156,584,166Werner mesomelic syndromeWieczorek et al.,2009
 404 G>A, Family 1SNPchr7:156,584,166Werner mesomelic syndromeWieczorek et al.,2009
 404 G>A, CubanSNPchr7:156,584,166Preaxial polydactylyLettice et al.,2003
 396 C>T, Turkish 1SNPchr7:156,584,174Preaxial polydactyly & triphalangeal thumbSemerci et al.,2009
 334 T>G, French 2SNPchr7:156,584,236Preaxial polydactylyAlbuisson et al., 2010
 323 T>C, Belgian 2SNPchr7:156,584,241Preaxial polydactylyLettice et al.,2003
 30 5A>T, Belgian 1SNPchr7:156,584,266Preaxial polydactylyLettice et al.,2003
 297 G>A, French 1SNPchr7:156,584,273Preaxial polydactylyAlbuisson et al., 2010
 295 T>CSNPchr7:156,584,275Triphalangeal thumbFurniss et al.,2008
 105 C>G, DutchSNPchr7:156,584,465Preaxial polydactylyLettice et al.,2003
 Case, Letticetranslocationt(5,7)(q11,q36)Preaxial polydactyly & triphalangeal thumbLettice et al.,2002
 Family, Klopockiduplication∼chr7:156,143,386-156,732,204Triphalangeal thumb-polysyndactylyKlopocki et al., 2008
 Family 6, Sunduplication∼chr7:156,241,020-156,699,998Triphalangeal thumb-polysyndactylySun et al.,2008
 Family 2, Sunduplication∼chr7:156,241,020-156,677,759Triphalangeal thumb-polysyndactylySun et al.,2008
 Family 5, Sunduplication∼chr7:156,241,020-156,619,399Syndactyly type IVSun et al.,2008
 Family 4, Sunduplication∼chr7:156,354,085-156,687,613Triphalangeal thumb-polysyndactylySun et al.,2008
 Family 3, Sunduplication∼chr7:156,354,085-156,619,399Triphalangeal thumb-polysyndactylySun et al.,2008
 Family 3, Wieczorekduplication∼chr7:156,368,541-156,661,877Triphalangeal thumb-polysyndactylyWieczorek et al.,2009
 Family 1, Sunduplication∼chr7:156,539,605-156,699,998Triphalangeal thumb-polysyndactylySun et al.,2008
 Family, Wuduplication∼chr7:156,547,469-156,644,074Syndactyly & tibial hypoplasiaWu et al., 2009
 Family 4, Wieczorekduplication∼chr7:156,572,751-156,661,877Triphalangeal thumb-polysyndactylyWieczorek et al.,2009
SOX9 limb enhancer
 Critical regionduplication∼chr17:65,642,665-66,847,686Brachydactyly-anonychiaKurth et al.,2009

ZRS POINT MUTATIONS

Thirteen single base mutations have been identified in the ZRS that cause human limb malformations. These mutations result in phenotypes that include PPD and triphalangeal thumb (TPT) with or without additional supernumerary digits. Because the mutations are distributed throughout the ∼2.2-kb region of the ZRS, it is difficult to predict which mutations will be more severe based on sequence alone. There are hints, however, that the degree of SHH misregulation corresponds to the severity of the phenotype. While not all ZRS mutations have been tested for in vivo enhancer activity in mice, those that were tested show a correlation between stronger expression of the reporter gene—in the ectopic anterior region and/or the normal posterior region—and more severe human phenotypes. Two of the first reported mutations, referred to as Cuban and Belgian2 (Table 1), demonstrate this correlation well (Fig. 3). In the Cuban mutation, which causes a severe form of polydactyly (Fig. 3C; including some tibial malformations in one patient) a very strong anterior and posterior expression pattern was observed in mouse embryos (Fig. 3B; Zguricas et al.,1999; Lettice et al.,2003,2008). In the Belgian2 mutation, which leads to a less severe phenotype of PPD II (Fig. 3E; an extra thumb anterior to a TPT), only weak anterior reporter expression was observed in mouse embryonic limbs (Fig. 3D; Lettice et al.,2003,2008).

While most of these single base pair substitutions show complete or near complete penetrance and are inherited in a dominant pattern, this is not the case for all mutations. The 295C>T mutation (Table 1) was originally reported as a neutral variant occurring in 10–30% of unaffected controls (Lettice et al.,2003). Later, this mutation was reported to be associated with TPT in multiple English families, but with incomplete penetrance (Furniss et al.,2008). When this mutation was tested in a mouse enhancer assay, the reporter gene was weakly, but significantly, detectable in the anterior portions of the limb buds, further supporting the link between the degree of ectopic expression and phenotypic severity (Furniss et al.,2008). This study suggests that regulatory mutations may have small effects that cause malformations in only a subset of carriers of the mutation.

Some patients with ZRS point mutations have malformations that are not limited to the autopod. In patients with Werner mesomelic syndrome, which includes hypoplastic tibia in addition to TPT polysyndactyly, mutations have been found repeatedly at a specific site, ZRS 404 (Wieczorek et al.,2009). This led the authors to suggest that mutations at this site cause Werner mesomelic syndrome while other single base mutations may cause less severe phenotypes such as TPT or PPD (Table 1; Wieczorek et al.,2009). However, another family with a mutation at the ZRS 404 site had a phenotype that was reported as PPD (Lettice et al.,2003). Of interest, this mutation comes from a family where the limb malformation phenotype includes bilateral tibial hypoplasia in one patient (Fig. 3C), causing Wieczorek et al., to suggest that the phenotype of this family is Werner mesomelic syndrome rather than PPD (Zguricas et al.,1999).

ZRS DUPLICATIONS

In addition to single base mutations, human limb malformations have been attributed to duplications that encompass the ZRS sequence (Table 1). Duplications including the ZRS region in humans have been shown thus far to only cause complex polysyndactyly phenotypes like triphalangeal thumb polysyndactyly (TPTPS) and syndactyly type IV. In one family, a single base pair mutation was initially reported to cause TPTPS (Wang et al.,2007), but that family was later shown to have a duplication as well, leading the authors to suggest that ZRS duplications, rather than mutations, are the cause of syndactyly malformations (Sun et al.,2008). It is worth noting that the human ZRS duplication phenotype is very different from the mouse ssq phenotype which also has a duplicated ZRS, but shows only polydactyly with no fusion of the digits (Sharpe et al.,1999).

The ZRS duplications found in humans do not have any recurring breakpoints nor any discernable relationship between the size of the duplication and the severity of the phenotype. The minimal region covered by all of the duplications appears to be approximately 47 kb (Table 1) covering intron 4 of LBMR1 and continuing 3′ of the ZRS conserved region. Interestingly, there is another limb phenotype associated with this region. Acheiropodia (OMIM #200500) is a severe limb malformation that has almost complete truncations of all limbs and aplasia of the hands and feet. A very rare disorder, acheiropodia is caused by a homozygous deletion of a ∼6-kb region (∼chr7: 156,616,000-156,622,000) that includes exon 4 of LMBR1, but does not include the ZRS (Ianakiev et al.,2001). This is similar to the phenotype seen in the ZRS knockout mouse, which reportedly does not have disruptions in this region (Sagai et al.,2004). Enhancers have not been identified in the acheiropodia deletion, although it is possible that the exon has a regulatory function in addition to its coding function as has been shown for a few other exons (Neznanov et al.,1997; Tumpel et al.,2008).

To date, there is no definitive way to predict the phenotype that will arise from a given ZRS mutation. Animal models with ZRS mutations appear to be of little use in these predictions because there are important differences between human and mouse ZRS mutations. Mice with ZRS mutations tend to have a stronger phenotype in the hindlimbs (Knudsen and Kochhar,1982; Sharpe et al.,1999), while human ZRS point mutations predominantly affect the hands and, unlike mice, human patients homozygous for ZRS mutations show phenotypes no more severe than heterozygotes (Semerci et al.,2009). Adding to this complexity is the observation that within a specific family there can be great phenotypic variability among affected individuals and incomplete penetrance in the family. There are also numerous families with strong linkage to the ZRS that appear to have no mutation or duplication in the ZRS or any portion of the acheiropodia deletion (Lettice et al.,2003; Gurnett et al.,2007; Li et al.,2009). Whether other mechanisms are behind these patients' malformations or there are other SHH limb cis-regulatory elements in this locus remains to be seen.

BMP2 LIMB ENHANCER DUPLICATIONS

In addition to the ZRS, mutations involving other cis-regulatory elements have been shown to cause human limb malformations (Table 1). A duplication including an enhancer of the Bone morphogenic protein 2 (BMP2) gene can cause brachydactyly (shortened digits). BMP2 is a signaling molecule that plays a role in limb development (reviewed in Robert,2007). BMPs are expressed in the condensing mesenchyme in the autopod that will form the digits and mutations in genes involved in BMP signaling can cause brachydactyly. A region ∼110 kb 3′ of BMP2 recapitulates a portion of BMP2 limb expression in a distinct region of the developing distal autopod and is thought to contain a BMP2 limb enhancer (Dathe et al.,2009). Two families with autosomal-dominant brachydactyly type A2 (OMIM# 112600) were identified with overlapping 5.5-kb duplications that include this enhancer (Table 1). These microduplications are proposed to increase BMP2 expression specifically in the limb (Dathe et al.,2009) and result in the brachydactyly phenotype.

EVIDENCE FOR ADDITIONAL cis-REGULATORY MUTATIONS

There are likely to be many more cis-regulatory mutations that cause limb malformations that have not yet been observed in human patients. For example, several cis-regulatory mutations have been shown to cause limb malformations in model organisms but have not yet been found in humans. One example is a duplication of a noncoding region 5′ of SRY (sex determining Y)-box 9 (SOX9), that is associated with brachydactyly–anonychia (Kurth et al.,2009). SOX9 is a gene that is involved in chondrocyte differentiation and skeletal development. Inactivation of Sox9 specifically in the mouse limbs leads to their complete absence, but the primary axes of the limb appear to be patterned correctly (Akiyama et al.,2002). A transgenic mouse designed to overexpress Sox9 specifically in the limb mesenchyme showed polydactyly as well as short, broad digits (Akiyama et al.,2007). Enhancers in the duplicated region have not been identified, but the phenotype of the human duplication is consistent with the mouse overexpression phenotype and suggests that changing the expression of SOX9 may be able to cause multiple limb malformation phenotypes.

It has been shown that removing genes from their normal chromosomal context by inversions or translocations can lead to limb malformations. This appears to be caused by the removal of the genes from their regulatory environment, perhaps separating a gene from its enhancers. One such example, is a chromosomal rearrangement near the homeobox A (HOXA) gene cluster that was found to lead to postaxial polysyndactyly (Lodder et al.,2009). In this patient, a balanced inversion on chromosome 7 that does not disrupt developmentally important genes removes the HOXA cluster from a region of several putative cis-regulatory elements more than 1 megabase away from the cluster. It is possible that other cases of postaxial polydactyly could be caused by mutations in HOXA-associated cis-regulatory elements. Chromosomal breakpoints near the homeobox D (HOXD) cluster also cause various limb malformations without disrupting genes (Dlugaszewska et al.,2006). Similarly, a translocation with a breakpoint near the parathyroid hormone-like hormone (PTHLH) gene, an important chondrogenic regulator, was shown to lead to brachydactyly type E by down-regulating gene expression (Maass et al.,2010).

Misexpression of genes due to genetic defects other than noncoding regulatory mutations suggests that regulatory mutations could affect those genes and lead to limb malformations. One such example is a duplication that includes the T-box 4 (TBX4) gene and causes isolated clubfoot (Alvarado et al.,2010). In other studies, isolated clubfoot was shown to be caused by mutations in paired-like homeodomain 1 (PITX1), which is thought to directly regulate the expression of TBX4 (Logan and Tabin,1999; Gurnett et al.,2008). Together, these provide a compelling argument that expression levels of TBX4 may be related to this limb malformation. Two hindlimb-specific TBX4 enhancers have been identified in the mouse (Menke et al.,2008) providing excellent candidates for cis-regulatory regions that may harbor mutations leading to human isolated clubfoot. Other cases of small duplications or deletions causing limb malformations, presumably by altering gene expression levels, have been reported (Schluth-Bolard et al.,2008; Tsai et al.,2009; van der Zwaag et al.,2010).

The depth of our current understanding of the genes and gene networks involved in tetrapod limb development provides a long list of candidate genes for which changes in cis-regulation could cause human limb malformations. Enhancers proposed to regulate some of these genes have already been identified (Sasaki et al.,2002; Cretekos et al.,2008; Feng et al.,2008; Durand et al.,2009; Abbasi et al.,2010). The study of human limb malformations has also led to the identification of specific genomic loci associated with various limb phenotypes. In cases where no coding mutations can be detected within the locus, the causal mutation may be in a cis-regulatory element. One such example is the split hand–foot malformation (SHFM), a limb malformation that is linked to six different genomic loci, only two of which have a coding mutation shown to cause SHFM. Recently, a study using ChIP-seq for the transcription factor tumor protein p63 (TP63) identified an enhancer within the SHFM1 locus that is thought to control the expression of distal-less homeobox 5 and 6 (DLX5/6) genes specifically in the limb AER (Kouwenhoven et al.,2010). Fine mapping of a patient with an 880-kb deletion (Table 1) was shown to encompass this enhancer, suggesting that its removal might lead to this phenotype (Kouwenhoven et al.,2010). In addition, studies in model organisms can also lead to the identification of cis-regulatory elements where mutations cause limb malformations in the model, but where mutations have not yet been identified in human patients (Feng et al.,2008; Liska et al.,2009).

INTERPRETING REGULATORY MUTATIONS

Deciphering the effect of cis-regulatory mutations is difficult in part because the mechanisms by which enhancers affect gene expression are poorly understood. Because transcription factors are known to bind to enhancers, a mutation that destroys a TFBS sequence could affect the recruitment of that transcription factor. However, many TFBS sequences are degenerate and a mutation might not have a strong impact on TF binding in vivo. Additionally, these binding sites often occur in clusters so mutations in a single site may not be critical (Gotea et al.,2010). It would also be possible for a mutation to create a new TFBS that could recruit the wrong TF to that region and conceivably cause ectopic expression or prevent normal expression. A better understanding of the flexibility of TFBS sequence and the importance of multiple sites would assist in the understanding of cis-regulatory mutation consequences.

Enhancers have been shown to physically “loop” to the promoters of the genes they regulate. This kind of long-range genomic interaction is correlated to gene expression, but remains poorly understood (reviewed in Sexton et al.,2009). This interaction is likely related to the mechanisms enhancers use to activate transcription. For example, it has been shown that the ZRS enhancer interacts with the SHH promoter only in regions of the limb where SHH is “poised” for transcription: the posterior ZPA and part of the anterior mesenchyme (Amano et al.,2009). In addition to this looping interaction, the same study showed that the looped SHH–ZRS locus moves out of its chromosome territory only in the ZPA region where it is actively transcribed. Other studies show a role for nuclear matrix proteins in the interaction of SHH and the ZRS (Zhao et al.,2009). While the exact mechanisms of looping remain unclear, it appears to be one way that cell-type specific DNA interactions are formed in a pattern that correlates with gene expression (Kagey et al.,2010).

PERSPECTIVES AND FUTURE DIRECTIONS

It is reasonable to speculate that cis-regulatory elements other than enhancers play a causative role in limb malformations. The best-studied enhancer, the ZRS, clearly fulfills the canonical role of an enhancer—up-regulating a specific gene in cis—and yet mutations within it can increase SHH expression beyond the normal level. Understanding the function of cis-regulatory elements like silencers and insulators is compounded by the lack of robust and reproducible in vivo assays for these elements. Even the commonly used enhancer assays are not quantitative because of copy number and integration site variability. The creation of functional assays to quantitatively validate cis-regulatory elements in an in vivo system will allow more rigorous study of regulatory mutations than is currently possible. Further advancements in tools to identify interactions between regions of DNA and nuclear organization will also allow better identification and understanding of how regulatory elements control gene expression (van Steensel and Dekker,2010).

The number of isolated limb malformations with no identified gene coding mutation suggests that many of them could be caused by cis-regulatory mutations. While there are only a few enhancers that have been shown to have the ability to cause human limb malformations thus far, our understanding of cis-regulatory elements continues to grow. The advancement of next-generation sequencing technologies will allow better detection, characterization and mutation analysis of regulatory elements involved in human limb malformations. Alongside these advancements, future high-throughput assays will be needed to rapidly assess the functional outcomes of the different nucleotide variants. Combined, these advances will lead to a growth in the depth of knowledge we have about how genes and gene networks interact in normal limb development and provide us with novel insight into how misexpression could cause human malformations in general.

Acknowledgements

We thank members of the Ahituv lab for helpful comments on the manuscript. J.E.V. and N.A. are supported by the National Institutes of Health. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NICHD, NHGRI or NIH.

Ancillary