Article Text

Original article
Deletions of 5′ HOXC genes are associated with lower extremity malformations, including clubfoot and vertical talus
  1. David M Alvarado1,
  2. Kevin McCall1,
  3. Jacqueline T Hecht2,
  4. Matthew B Dobbs1,3,
  5. Christina A Gurnett1,4,5
  1. 1Department of Orthopedic Surgery, Washington University, St. Louis, Missouri, USA
  2. 2Department of Pediatrics, University of Texas Medical School, Houston, Texas, USA
  3. 3Shriners Hospital for Children, St Louis, Missouri, USA
  4. 4Department of Neurology, Washington University, St. Louis, Missouri, USA
  5. 5Department of Pediatrics, Washington University, St. Louis, Missouri, USA
  1. Correspondence to Dr Christina A Gurnett, Department of Neurology, Washington University School of Medicine, 660 S Euclid Ave, St Louis, MO 63110, USA; gurnettc{at}


Background Deletions of the HOXC gene cluster result in variable phenotypes in mice, but have been rarely described in humans.

Objective To report chromosome 12q13.13 microdeletions ranging from 13 to 175 kb and involving the 5′ HOXC genes in four families, segregating congenital lower limb malformations, including clubfoot, vertical talus and hip dysplasia.

Methods Probands (N=253) with clubfoot or vertical talus were screened for point mutations and copy number variants using multiplexed direct genomic selection, a pooled BAC targeted capture approach. SNP genotyping included 1178 probands with clubfoot or vertical talus and 1775 controls.

Results The microdeletions share a minimal non-coding region overlap upstream of HOXC13, with variable phenotypes depending upon HOXC13, HOXC12 or the HOTAIR lncRNA inclusion. SNP analysis revealed HOXC11 p.Ser191Phe segregating with clubfoot in a small family and enrichment of HOXC12 p.Asn176Lys in patients with clubfoot or vertical talus (rs189468720, p=0.0057, OR=3.8). Defects in limb morphogenesis include shortened and overlapping toes, as well as peroneus muscle hypoplasia. Finally, HOXC and HOXD gene expression is reduced in fibroblasts from a patient with a 5′ HOXC deletion, consistent with previous studies demonstrating that dosage of lncRNAs alters expression of HOXD genes in trans.

Conclusions Because HOXD10 has been implicated in the aetiology of congenital vertical talus, variation in its expression may contribute to the lower limb phenotypes occurring with 5′ HOXC microdeletions. Identification of 5′ HOXC microdeletions highlights the importance of transcriptional regulators in the aetiology of severe lower limb malformations and will improve their diagnosis and management.

  • Copy-number
  • Clubfoot
  • Vertical Talus

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Congenital vertical talus is a fixed dorsal dislocation of the talonavicular joint combined with contracture of the hindfoot that causes a rigid flatfoot deformity.1 It is commonly referred to as ‘rocker-bottom foot’ because of its clinical appearance, and diagnosis of congenital vertical talus requires radiographic confirmation. Talipes equinovarus (clubfoot) is a closely related disorder that may occur in families with vertical talus, or even in the same individual (ie, clubfoot on the right foot and vertical talus on the left).2 ,3 Both clubfoot and vertical talus occur more frequently in male subjects, with a male: female ratio of approximately 2:1.4 Clubfoot has a birth prevalence of 1 in 1000 while 1 in 10 000 has been suggested for vertical talus,5 but may be an underestimate because of the difficulty in distinguishing from other more common and benign positional foot anomalies.1

Vertical talus is associated with additional congenital anomalies in slightly more than half of all cases.3 ,5 Syndromic causes of vertical talus include distal arthrogryposis and myelomeningocele, as well as trisomy 13 and trisomy 18.2 ,6 ,7 Many of these disorders are associated with significant morbidity and mortality and require specialised surgical or neonatal intensive care in the newborn period; therefore genetic analysis may be essential for prenatal diagnosis to distinguish them from other causes of isolated vertical talus.

Congenital vertical talus has previously been associated with a single missense mutation (M319K) in HOXD10 in two separate families.8 ,9 However, the genetic cause of congenital vertical talus in five other families was not found, and HOXD10 mutations were not found in sporadic cases of vertical talus.10 While the genetic basis of clubfoot is also poorly understood, abnormalities in genes involved in early limb development, including PITX1 and TBX4, are responsible for approximately 1–5% of familial clubfoot.11–15 A deletion of HOXC13 was also identified previously in one family with familial clubfoot.11 Here, we describe chromosome 12q13 microdeletions involving the 5′ HOXC genes in four families with vertical talus or clubfoot, and demonstrate trans-acting effects of this deletion on HOXD gene expression.


Patient samples

The study protocol was approved by the institutional review board of Washington University in St Louis and University of Texas at Houston and all subjects and/or parents gave informed consent. Blood and saliva samples were collected from probands and family members with lower limb malformations. The study included 1178 probands with clubfoot or vertical talus and 1775 controls (474 in-house and 1301 European Exome Aggregation Consortium (ExAC).16 All probands from Washington University in St Louis had either clubfoot or vertical talus. Most cases were isolated, but ∼10% of the cases recruited from Washington University had additional musculoskeletal abnormalities, including tibial hemimelia, fibular hemimelia or hip dysplasia. Affymetrix genome-wide human SNP array 6.0 was performed on 226 probands with clubfoot and 27 with congenital vertical talus and showed only the 175 kb deletion in family 1, as previously reported.11 All samples had previously been screened for PITX1 and TBX4 point mutations, duplications and deletions. All families with vertical talus had HOXD10 gene sequencing.9 ,10 Patients were excluded from the study if they had multiple congenital anomalies or known causes of clubfoot or vertical talus. DNA was extracted using either the DNA isolation kit for mammalian blood (Roche) or the Oragene purifier for saliva (DNA Genotek). MRI was performed as previously described.17

Multiplexed direct genomic selection targeted resequencing

Probands (N=253) with clubfoot or vertical talus were screened for copy number variants (CNVs) using multiplexed direct genomic selection (MDiGS), a pooled BAC capture approach for targeted CNV detection.18 Patient 1 was included as a positive control, as a 175 kb deletion had previously been detected with Affymetrix 6.0 microarray.11 Advantages of the MDiGS method are that the sequence data can be used to detect point mutations and for high-resolution copy number analysis. Samples were individually indexed and hybridised to a biotinylated BAC bait RP11-578A18 (BACPAC Resources) in pools of 96 samples, and post-capture samples were sequenced on a MiSeq personal sequencer (Illumina). The human chromosome 12 BAC RP11-578A18 encompasses a 184 kb region that includes HOXC13, HOXC12, HOXC11, HOXC10 and HOXC9 genes together with 108 kb non-coding region 5′ of the HOXC gene cluster. Five RefSeq lncRNAs, including HOTAIR, were also located in the captured region. Copy number was determined using normalised read counts for each test sample compared with the average normalised read counts for all other co-captured samples, and exact breakpoints were identified by gapped alignment of one-end anchored read pairs, as previously described.18 CNVs were validated by quantitative PCR using three PCR primers for each CNV.

Restriction digest genotyping

The rs189468720/HOXC12 SNP was genotyped in 625 clubfoot cases, 7 vertical talus cases and 474 controls using restriction digest (Washington University cohort) and in 336 clubfoot cases using TaqMan probes (University of Texas cohort). A 425 bp region containing rs189468720/HOXC12 was PCR amplified (F-primer: GTGGAGGACGGCAAGGG, R-primer: CTAGCTCAGTCCTGTTCTGCC) from genomic DNA. PCR products were restriction digested with EcoNI (37°C, 16 h), which specifically cuts the rs189468720 alternate allele, producing 163 and 262 bp products.

Human skin fibroblasts

Human skin fibroblasts were cultured using previously described methods.19 Skin was obtained during clubfoot surgery from patient 1 (HOXC deletion) and two reference samples without HOXC deletions.

Gene expression analysis

Total RNA was extracted from confluent human skin fibroblasts, 1 T75 flask per sample/replicate, using Trizol (Invitrogen) recommended protocol, DNase-digested with TURBO DNase and further purified using RNeasy (Qiagen). Total RNA was run on an agarose gel to inspect for degradation and quantified on a Nanodrop 2000. RNA (1 μg) was used for SuperScript II reverse transcription with random hexamers.

Gene expression was assayed on an Mx3005P (Stratagene) using iTaq Universal SYBR Green Supermix (BioRad), 40 cycles, followed by a melt curve analysis. All cDNA samples were run in triplicate with mean Ct values compared with a β-actin (ACTB) reference for ddCt calculation. Relative expression ratios represent mean ratios for three deletion biological replicates compared with two controls, three biological replicates each. Error bars represent SD across all case/control comparisons and p values were calculated using Student's t test for each gene compared with five housekeeping controls (18s rRNA, PUM1, GAPDH, HPRT1 and SDHA).


To investigate the role of HOXC gene cluster variation in human congenital lower limb malformations, we captured and sequenced18 a 184 kb region including 5′ HOXC genes (HOXC13, HOXC12, HOXC11, HOXC10 and HOXC9) and lncRNAs (HOTAIR, HOXC13-AS, HOXC-AS3, HOXC-AS2 and HOXC-AS1) in 226 probands with clubfoot and 27 with congenital vertical talus. Copy number analysis revealed three small deletions ranging in size from 13 to 52 kb, as well as our previously reported 175 kb chr12:54165001-54335668 deletion in patient 111 (figure 1). While these four microdeletions overlap within a 5.3 kb non-coding region (hg19, chr12:54311194-54316500) located 16 kb 5′ of the HOXC gene cluster, there are no genes within the overlapping region. The 175 kb deletion in family 1 contains only exon 2 of HOXC13 and HOXC13-AS lncRNA. The 13 kb microdeletion in family 2 contains no genes and is located entirely 5′ of the HOXC gene cluster but has been previously reported in 2/2504 healthy individuals in the Database of Genomic Variants (DGV).20 There are no reported microdeletions involving HOXC13-AS and HOXC13 in the DGV. The nearly identically sized 50 kb deletions in families 3 and 4 share the same distal breakpoint but have different proximal breakpoints, and include HOXC13, HOXC12 and part of HOTAIR lncRNA. The microdeletions segregate with lower limb malformations in each family and were fully penetrant with variable expressivity (figure 2).

Figure 1

5′ HOXC gene microdeletions identified in patients with clubfoot and vertical talus. Four partially overlapping deletions involving the 5′ HOXC genes and upstream regulatory region were identified in probands with clubfoot and vertical talus.

Figure 2

Segregation of 5′ HOXC microdeletions in patients with vertical talus and clubfoot. Black circles and squares indicate clubfoot in families 1 and 2 and vertical talus in families 3 and 4. Grey indicates syndactyly (2-003) and hammertoes (1-002). Del indicates deletion and WT indicates normal copy number.

Sequence analysis of the MDiGS data revealed two rare variants predicted to be damaging by SIFT.21 One was a rare HOXC11 p.Ser191Phe missense variant (EU-minor allele frequency (MAF)=0.0002) that completely segregates with clubfoot in six affected individuals of family 5 (figure 3). We also identified a rare HOXC12 p.Asn176Lys (rs189468720) missense variant in 5/210 (MAF=0.0119) Caucasian clubfoot and vertical talus cases compared with 0/321 in-house controls and 4/1301 European (MAF=0.0015) individuals from the ExAC (p=0.0016, OR=9.84).16 After genotyping the rs189468720 variant in a further 961 clubfoot and 7 vertical talus cases and 153 controls, we confirmed a weak association of rs189468720 with lower limb malformations (combined cases=15/1178, controls=6/1775, p=0.0057, OR=3.8).

Figure 3

Segregation of HOXC11 p.Ser191Phe with clubfoot. The HOXC11 p.Ser191Phe missense variant completely segregates with clubfoot in six affected individuals in family 5.

Clinical features of patients with HOXC deletions

The largest deletion identified was 175 kb, which was present in four affected members of family 1 and involved only HOXC13 and HOXC13-AS lncRNA. Three members of this family had severe treatment-resistant clubfoot, requiring multiple operations (figure 4A,B) (see online supplementary table S1), and one had hammertoes that required surgical treatment (figure 4C). One individual had nail hypoplasia involving all fingers but not toes (figure 4D).

Figure 4

Clinical features of patients with 5′ HOXC microdeletions. Treated clubfoot in (A) proband (1-001) and (B) sibling (1-005) with 175 kb 5′ HOXC gene deletion. Note overlapping toes. (C) Surgically treated hammertoes (1-002) and (D) fingernail hypoplasia (1-004). Congenital vertical talus with (E) rocker-bottom appearance (3-007) before treatment and (F and G) persistent flatfoot (3-007) and overlapping toes (3-008) after treatment.

The smallest 13 kb intergenic microdeletion, also reported in the DGV, occurred in the proband of family 2 who had right-sided clubfoot and fibular hemimelia and postaxial hypodactyly with absent fourth and fifth toes on the right foot. The microdeletion also occurred in his father who had mild intoeing and unilateral 2–3 toe syndactyly but no other limb abnormality.

Congenital vertical talus occurred with complete penetrance in all nine affected individuals from families 3 and 4 who harboured the similar, but not identical, 50 kb microdeletions containing HOXC13, HOXC12 and part of HOTAIR lncRNA (figure 4E–G) (see online supplementary table S1). Individuals from both families had recurrent deformity requiring additional casting and surgery. In addition, hip dysplasia and hip muscle weakness were present in childhood, and severe arthritis of the hip and knee were present in older individuals with the microdeletion. One individual was noted to have moderate bilateral adducted thumbs in infancy that resolved with time. Brittle toenails were noted in two individuals.

Muscle volume is concordantly reduced in patients with HOXC deletions

MRI of the lower limbs was performed to characterise the limb malformation, as some morphological abnormalities can only be detected with imaging.22 MR images were obtained at age 17 (1-001) and age 64 (1-004) from two affected members of family 1 who had bilateral clubfoot that was more severe on the right. MRI revealed smaller muscle compartments in the calf, with the peroneus muscles in the lateral compartment most severely affected and partially replaced with fat (figure 5A,B). Peroneus muscle hypoplasia was concordant in both individuals and was more prominent in the more severely affected right legs. Globally reduced muscle volume of the affected limb was also observed on MR images obtained from two siblings with unilateral clubfoot in family 5 (right limb, figure 5C; left limb figure 5D).

Figure 5

Magnetic resonance images show reduced muscle volume in the calf of patients with clubfoot and 5′ HOXC gene deletions. Lateral compartment peroneus muscles are most severely affected and replaced with fat (arrows) in the more severely affected limb (left) in (A) 1-001 and (B) 1-004. Globally reduced muscle volume in the unilaterally affected clubfoot limbs of siblings with HOXC11 p.Ser191Phe mutations (C) 5-007 (left leg) and (D) 5-008 (right leg).

Cis and trans effects of HOXC microdeletions

To determine if deletion of 5′ HOXC genes and upstream regulatory regions affects expression levels of downstream genes, we performed qPCR on human skin fibroblasts from the family 1 proband with the 175 kb microdeletion and two reference samples without HOXC deletions. The deletion removes the first exon of HOXC13 and its transcript was reduced by approximately 50% compared with controls (figure 6). Although HOXC12 is not located within the deletion, its transcript was also decreased. The expression of HOTAIR lncRNA, which is outside the deleted interval, was unaffected; however, lncRNAs HOXC-AS2 and HOXC-AS3, also outside the deletion interval, were decreased. Because HOXC lncRNAs regulate HOXD gene expression23 and HOXD10 mutations have previously been identified in patients with vertical talus,8 ,9 we also examined the expression levels of HOXD genes in human skin fibroblasts. The expression of four HOXD genes—HOXD13, HOXD12, HOXD11 and HOXD10—were significantly reduced compared with controls.

Figure 6

HOXC and HOXD gene expression is downregulated in the skin fibroblasts of a patient with 5′ HOXC gene deletion. HOXC and HOXD gene expression in human skin fibroblasts from a patient with clubfoot (1-001) with 175 kb HOXC deletion (HOXC13 and HOXC13-AS deleted) compared with control fibroblasts. *p<0.05, **p<0.005.


To interrogate the role of rare HOXC gene cluster variation in congenital lower limb malformations we used MDiGS, a new multiplexed targeted capture approach that is useful for both highly accurate CNV and SNP/INDEL (insertion/deletion) discovery in large genomic regions.18 Four unrelated families with congenital vertical talus and clubfoot were found to have small microdeletions of 5′ HOXC genes. While vertical talus and clubfoot associated with 5′HOXC microdeletions is almost completely penetrant and more difficult to treat clinically, the prognosis is much better than many other more common causes of vertical talus, such as trisomy 18,2 and these may therefore be clinically important to identify. Notably, the 5′ HOXC gene microdeletions we detected are small (<200 kb) and may be difficult to detect with routine chromosomal microarray. Only one of our four deletions was detected by chromosomal microarray (Affymetrix 6.0), and the other three required higher-resolution MDiGS. With our current methods, 5′ HOXC microdeletions were detected in two out of six previously described autosomal dominant isolated vertical talus families,9 and we expect that improvements in technology will allow for the identification of even smaller microdeletions that may be responsible for additional cases.

Earlier studies in mice demonstrated expression of 5′ HoxC genes in the developing hindlimb but not the forelimb, suggesting that they might play an important role in hindlimb morphogenesis.24 The phenotypes of our patients with 5′ HOXC deletions are consistent with hindlimb expression, as none of our patients had hand malformations, with the exception of one patient with fingernail hypoplasia. Despite their unique expression pattern, however, it was believed that 5′ HoxC genes were unnecessary for limb development because loss of the entire HoxC gene cluster or loss of individual 5′ HoxC genes (Hoxc9, Hoxc10, Hoxc13) did not result in lower limb skeletal abnormalities in mice.25 While loss of the entire HOXC gene cluster has been described in three individuals with large chromosome 12 deletions (>1 Mb) and cognitive impairment, the loss of additional genes precludes the assignment of HOXC gene effects. Skeletal abnormalities described previously in these patients include ulnar deviation of hands, flexion deformities of fourth and fifth fingers and contractures of Achilles tendons,26 severe kyphoscoliosis, ulnar deviation of the hands and finger flexion contractures,27 and arthrogryposis with valgus ankle position and pectus excavatum.28 Microdeletions of 5′ HOXC genes as described here have not been reported by other investigators and none are annotated in the DGV or DECIPHER database.29 Interestingly, the reciprocal microdeletion, in which all HOXC genes except HOXC13, HOXC12 and HOTAIR are deleted, was not associated with any skeletal defects or limb contractures,30 suggesting that the limb malformations in our families are specifically related to abnormalities in one of these genes. The smallest 13 kb intergenic microdeletion that we identified in one family had been reported in 2 out of 2504 controls in the DGV, suggesting that it may be either a benign variant or a modifier.

Because there are strong global regulators controlling the coordinated expression of Hox genes and positional effects of deletions can alter the colinear expression of HoxC genes,31 it is difficult to ascribe functional effects of single genes within the HOXC locus. Certainly, the incompletely penetrant nail hypoplasia seen in one of our cases is consistent with loss of HOXC13 within the deleted interval, although pure hair and nail ectodermal dysplasia due to HOXC13 mutations has only been described as an autosomal recessive condition.32 The association of single missense mutations in HOXC11 and HOXC12 with isolated clubfoot and vertical talus also suggests a possible role for single 5′ HOXC gene alterations in their pathogenesis, although additional studies are needed for confirmation.

Our data also support an intriguing possibility that modification of the HOXC locus on chromosome 12 trans-regulates expression of HOXD genes on chromosome 2. An indirect effect is plausible because mutations in HOXD10 cause congenital vertical talus.8 ,9 The 5′ HOXC region contains several lncRNAs including HOXC13-AS, HOXC-AS2, HOXC-AS3 transcripts, and HOTAIR; the last of these has been shown to be a repressor of HoxD gene expression in mice.23 ,33 Our study likewise demonstrates in human disease fibroblasts from a patient with a HOXC microdeletion, altered lncRNA expression and downregulation of HOXD gene expression. Overall, this suggests that HOXC locus differential trans-regulation of the HOXD genes may be an important factor in some human lower limb malformations.

The reduced limb muscle mass and peroneus muscle hypoplasia observed in two patients with 5′ HOXC gene microdeletions was similar to the morphological abnormalities identified in humans and mice with clubfoot secondary to PITX1 deletions.12 ,13 As a consequence of these morphological abnormalities,22 patients with 5′ HOXC deletions are more likely to have vertical talus and clubfoot that is difficult to treat and additional skeletal comorbidities, such as hip dysplasia. More severe clinical phenotypes have also been described in patients with clubfoot with TBX4 duplications15 ,34 and PITX1 mutations.12 ,13 Because Pitx1 acts upstream of Tbx4 and Hox genes during hindlimb specification35 and binds directly to regulatory elements near Hoxc10 and Hoxc11,36 our results support a critical role for hindlimb transcriptional regulators in clubfoot and vertical talus pathogenesis.

Most importantly, this study provides clinically relevant information about the genetic basis of clubfoot and vertical talus that will be immediately applicable to the diagnosis and genetic counselling of patients with these disorders. Because HOXC microdeletions are associated with an increased risk of severe treatment resistance and hip dysplasia, accurate molecular diagnosis will ultimately translate into personalised and improved management of patients with lower limb malformations.


We thank the patients and their families for their role in this work. We thank the Genome Technology Access Center in the Department of Genetics at Washington University School of Medicine for providing Illumina MiSeq Personal Sequencer support.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Contributors DMA: conceived the analysis, performed experiments, analysed the data and wrote the manuscript; KM: performed experiments. JTH: collected the samples and provided genotype data; MBD: collected the samples and carried out clinical analysis; CAG: designed the study and supervised the research.

  • Funding This work was supported by Shriners Hospital for Children, the Children's Discovery of St Louis Children's Hospital and Washington University, and the Eunice Kennedy Shriver National Institute of Child Health and Human Development (grant number R03HD068649).

  • Competing interests None declared.

  • Ethics approval Institutional review board of Washington University in St Louis and University of Texas at Houston.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The dataset of targeted capture sequencing is available upon request.