Introduction

Expression of some genes, in particular of developmental control genes, is influenced by regulatory elements at some distance from the transcription and promoter regions.1 Because many of these genes are transcription factors with tissue- and developmental stage-specific expression patterns, such position effects are difficult to study in humans. The best evidence for position effects causing human genetic disease comes from chromosomal rearrangements with breakpoints and/or microdeletions well outside the relevant genes. In a number of human developmental anomalies, including X-linked deafness,2 holoprosencephaly,3 campomelic dysplasia,4 and aniridia,5 it has been shown that the distance of the disease-causing chromosome breakpoint to the 3′ or 5′ end of the misregulated gene may be more than 1 Mb.

De novo balanced chromosomal rearrangements occur in approximately one in 2500 newborns. Because the rate of abnormal phenotypes in de novo translocation or inversion carriers is approximately twice as high as in random newborn populations,6 it is plausible to assume that in about half of these cases the observed phenotype is caused by inactivation of a specific gene(s) in the breakpoint region(s). The cytogenetic and molecular characterization of disease-associated balanced chromosome rearrangements has already led to the identification of numerous (>100) disease genes, showing the general usefulness of this approach.7, 8 In most published cases the pathological state was caused by disruption or deletion of a gene(s) in the breakpoint region(s). Relatively few cases exhibiting position effects have been described so far.

Here, we present a de novo t(12;17) translocation in a developmentally delayed boy with skeletal malformations. The breakpoint on chromosome 17q21.3 is of particular interest, because this region contains the HOXB gene cluster. The homeobox (HOX) genes belong to a family of transcription factors that play an established role in skeletal development. Specific mutations in different HOX genes (ie, HOXA11, HOXA13, HOXD10, and HOXD13) cause various limb malformations.9, 10 Limb and/or skeletal defects may also result from balanced translocations affecting regulatory elements around the HOXD gene cluster.11, 12

Materials and methods

Classical and molecular cytogenetic techniques

Metaphase chromosome spreads of the patient and his parents were prepared from peripheral blood lymphocytes and analyzed by classical GTG banding. An EBV-transformed lymphoblastoid cell line of the patient was established for chromosome, DNA, and RNA preparations. BAC and PAC clones (Table 1) were selected from the Wellcome Trust Sanger Institute Ensembl contigs and obtained from the Resource Center Primary Database of the German Human Genome Project and ResGen (Invitrogen). Amplification of larger (10–15 kb) BAC subfragments was carried out with a series of primer pairs (Table 2) chosen from the genomic sequence of BAC RP11-463M16, as described previously.13, 14 Genomic BAC DNAs and their long-range PCR products were labeled with biotin-16-dUTP or digoxigenin-11-dUTP (Roche) by standard nick translation and FISH mapped on metaphase chromosomes.15

Table 1 FISH mapping results of BAC/PACs from chromosome 17q21.3
Table 2 Long-range PCR fragments of the breakpoint-spanning BAC RP11-463M16

Expression analyses

Cytoplasmic RNAs of lymphoblastoid cell lines from the patient and 10 normal individuals were isolated using Trizol reagent (Invitrogen) and reverse transcribed with Superscript III reverse transcriptase (Invitrogen) and random primers. HOXB3, HOXB4, HOXB8, HOXB9, HOXB13, TTLL6, CALCOCO2, and SNF8 transcripts were amplified from 40 and 200 ng cDNA each, using PCR conditions and gene-specific primer pairs, as described.16 To test biallelic versus monoallelic expression SNP-containing RT-PCR products were sequenced, using a Beckman CEQ 8000 Genetic Analysis System. Real-time quantitative RT-PCR analysis of HOXB3, HOXB9, HOXB13, CALCOCO2, and SNF8 transcripts was performed with predesigned and optimized TaqMan Gene Expression Assays (Applied Biosystems) on an Applied Biosystems 7500 Real-Time PCR System according to the manufacturer's instructions. Relative quantification was carried out with the delta-delta-CT method, using GAPDH as endogenous control.

Results

Clinical presentation

The boy was born in the 37th gestational week with a weight of 2520 g, a length of 51 cm, and microcephaly. He had low-set and posteriorly rotated ears, macrostomy, coarse facial features, postaxial hexadactyly of the feet, clinodactyly of the fingers, and bilateral inguinal hernia. He also had a funnel chest and a bell-shaped thorax (Figure 1). Owing to an intracerebral hemorrhage he developed a posthemorrhagic hydrocephalus that required insertion of a vetriculoperitoneal shunt. At 5 years of age his gross and fine motor skills, his mental development, and his language skills were all markedly delayed.

Figure 1
figure 1

Patient with de novo t(12;17)(p13.3;q21.3) at the age of 5 years. Note the funnel chest and hexadactyly (inset) of both feet.

Breakpoint mapping

G-banding analysis (at a 500 band level) of the patient revealed an apparently balanced reciprocal translocation between the short-arm tip of chromosome 12 and the proximal long arm of chromosome 17 (Figure 2a), 46,XY,t(12;17)(p13.3;q21.3). The parental karyotypes were normal.

Figure 2
figure 2

Cytogenetic and molecular characterization of the t(12;17)(p13.3;q21.3) translocation. (a) Ideograms of the patient's normal and derivative chromosomes 12 and 17. The breakpoints predicted by G-band analysis are indicated by arrows. (b) Hybridization of a 12p subtelomeric probe (8M16/SP6, Vysis) to the normal 12 and the der(12) (arrow). (c) Hybridization with telomeric hexamer repeats revealed an interstial telomere (arrow) on the der(12) (inset). (df) High-resolution FISH mapping of BAC/PACs and BAC subfragments to patient's metaphase spreads. PACs CTC-221K18 (labeled by Cy3 in red) and 202I17 (labeled by FITC in green) from the 12q and 17p subtelomeres were used for chromosome identification. PAC CTD-2377D24 (d, red signals indicated by arrows) produced signals on the normal 17 and the der(17), proximal to the breakpoint. BAC RP11-463M16 (e, red signals indicated by arrows) hybridized to the normal 17 and both derivative chromosomes and, thus, contains the 17q21.3 breakpoint. PCR fragment 463M16G (f, red signals indicated by arrows) hybridized to the normal 17 and the der(12), distal to the breakpoint. (g) Chromosome 17q21.32 breakpoint contig of clones RP11-357H14 (green bar), CTD-2377D24 (blue bar), and RP11-463M16 (red bar). The distal end of the breakpoint-flanking (by FISH) PAC CTD-2377D24 corresponds to basepair 49 931 of the breakpoint-spanning BAC RP11-463M16 (vertical red dotted line). Smaller black bars below the BAC contig indicate the location of genes in the breakpoint region. Arrows below the corresponding gene names (above the BAC contig) indicate the direction of transcription. Green star indicates the TTLL6 stop codon outside the breakpoint region. Small orange bars below the BAC contig indicate the location of long-range PCR products used for FISH mapping. The breakpoint-containing interval (corresponding to the region from 40 to 61 kb of BAC RP11-463M16) is indicated by horizontal purple arrows. (h) VISTA conservation plot of the enlarged breakpoint interval (window size 50 bp, homology threshold 75%). Conserved non-genic sequences between humans and mouse are represented as pink peaks; conserved genic sequences are light blue. The distal blue peak (indicated by a blue arrow) corresponds to the 3′ UTR of TTLL6.

A 12p subtelomeric probe (8M16/SP6, Vysis) hybridized to the normal and derivative chromosome 12 of the patient (Figure 2b), whereas a 17q subtelomeric probe (D17S928, Vysis) produced signals on the der(12) and the normal 17 (data not shown). A telomeric hexamer repeat probe (Q-Biogene) hybridized to both ends and an interstitial site corresponding to the 12p13.3 breakpoint on the der(12) and to both ends of the der(17), as well as to other chromosome ends (Figure 2c). This implies that the 12p breakpoint occurred within the TTAGGG repetitive region and only telomeric repeats were translocated onto the der(17). On the other hand, it is possible that the distal long arm of chromosome 17 was added to an intact chromosome 12 and the chromosome 17 breakpoint was repaired by the addition of telomeric hexamer repeats. Telomeric sequences can be inserted at chromosomal breaks either directly by the telomerase enzyme or by non-homologous end-joining of a blunt-ended double-stranded telomeric DNA fragment.17, 18 However, regardless of whether the der(12) is the result of a true translocation exchange or capture of telomeric sequences, gene content and order of the 12p13.3 region remained undisturbed. Thus, we focussed our positional cloning efforts on the der(17), looking for loss, disruption, or otherwise inactivation of a dosage-sensitive gene(s) in the breakpoint region.

To narrow down the critical region, BAC and PAC clones from chromosome 17q21.3 were cohybridized with chromosome 12qter and 17pter identification probes to the patient's metaphase spreads and mapped relative to the breakpoint (Table 1). A breakpoint-containing BAC contig (Figure 2g) was assembled from the database. PAC CTD-2377D24 (Figure 2g, blue bar) hybridized to both the normal 17 and the der(17) (Figure 2d, red probes indicated by arrows) and, thus, appears to lie proximal to the breakpoint. BAC RP11-463M16 (Figure 2g, red bar), which partially (50 kb) overlaps CTD-2377D24, produced FISH signals on the normal 17 and on both derivative chromosomes (Figure 2e, red probes indicated by arrows), indicative of a breakpoint-spanning clone.

Four long-range PCR products of BAC RP11-463M16 were generated for higher-resolution mapping (Figure 2g, orange bars; Table 2). As expected, fragment 463M16A, which is also contained in PAC CTD-2377D24, hybridized to the normal 17 and the der(17), proximal to the breakpoint. Fragments 463M16G (Figure 2f, red probes indicated by arrows), C, and F all hybridized to the normal 17 and the der(12), which would be consistent with a location distal to the breakpoint. Because PAC CTD-2377D24 and BAC fragment 463M16G did not produce split hybridization signals, but hybridized to different derivative chromosomes, the 17q21.3 breakpoint must be contained in the very distal part (<10 kb) of PAC CTD-2377D24, the proximal part (<10 kb) of BAC fragment 463M16G, or the small interval (1090 bp) between these two DNA segments. In our experience, 10 kb of genomic DNA sequence is more than sufficient to generate a detectable FISH signal at the target region. Collectively, our FISH mapping results narrowed the breakpoint down to a 21-kb interval from 40 to 61 kb of the breakpoint-spanning BAC RP11-463M16 (Figure 2g, indicated by horizontal purple arrows).

Genes near the 17q21.3 breakpoint region

The breakpoint-spanning clone RP11-463M16 contains six validated genes (Figure 2g; Table 3). The coding sequence of tubulin tyrosine ligase-like family, member 6 (TTLL6), a gene of unknown function, lies distal to the breakpoint region. The stop codon of TTLL6 (Figure 2g, green star) corresponds to basepair 64 217 of RP11-463M16, implying that only the 3′ UTR lies within the distal breakpoint region (Figure 2h, indicated by vertical blue arrow). Interestingly, the transcription factor HOXB13 lies in close proximity (<30 kb) at the proximal side of the breakpoint on the der(17). The HOXB gene cluster on chromosome 17q21.3 contains HOXB1 to HOXB9 and HOXB13, which is approximately 100 kb upstream of HOXB9 (Figure 2g). All 10 HOXB genes are transcribed in the same direction with the 5′ end toward HOXB13 and the 3′ end toward HOXB1.

Table 3 Genes in the breakpoint region

Because the patient's phenotype, including hexadactyly and other skeletal abnormalities, is consistent with misregulation of a HOXB gene(s), we conclude that the 17q21.3 rearrangement separated the HOXB gene cluster from control elements and/or changed its local chromatin environment. Comparison of human and mouse genomic sequences with the GenomeVISTA program revealed three conserved non-genic segments of >50 bp with >75% sequence similarity in the 21-kb breakpoint interval (from 40 to 61 kb of BAC RP11-49931) at the 5′ region of the HOXB cluster (Figure 2h, indicated by vertical pink arrows). These evolutionarily highly conserved segments may be good candidates for cis-regulatory elements. Considering that the 17q21.3 breakpoint acquired telomeric hexamer repeats, the 5′ region of the HOXB gene cluster on the der(17) was fused to a repetitive DNA domain, which may have changed its higher-order chromatin structure and/or nuclear localization. Mammalian telomeres are located in specific cell-cycle-dependent areas of the nucleus.19 In yeast it was shown that the transcriptional states of telomeric genes depend on spatial positioning in transcription-compentent versus transcription-incompetent nuclear compartments.20

To provide additional evidence for a possible position effect, the expression of HOXB3, HOXB4, HOXB8, HOXB9, HOXB13, TTLL6, CALCOCO2, and SNF8 was analyzed by standard RT-PCR (using 40 ng cDNA) in lymphoblastoid cell lines of the patient and 10 normal control individuals. Unfortunately, HOXB8, which is likely to play a role in limb development,21 as well as the breakpoint-flanking genes HOXB13 and TTLL6, were not or only very lowly expressed in lymphoblasts. When using increased amounts (200 ng) of cDNA in the PCR, low HOXB8, HOXB13, and TTLL6 transcript levels were detected in some of the tested cell lines. However, repeated experiments using the same cDNA samples under identical conditions as well as experiments with different cDNA preparations from the same cell lines suggested that RT-PCR amplification occurred in a more or less stochastic manner. Therefore, these genes were excluded from further expression analyses. In contrast, HOXB3 and HOXB4, which are more 3′ in the HOXB cluster, and HOXB9, which is relatively 5′, as well as CALCOCO2 and SFN8 from the distal breakpoint region, were expressed at quantifiable levels in lymphoblastoid cell lines. Transcripts were easily amplified by standard RT-PCR from lymphoblasts of the patient and all 10 control individuals. However, quantitative real-time PCR demonstrated that expression of all five tested genes varied considerably (by a factor of 5–10) between cell lines (data not shown). We therefore conclude that EBV transformation and/or prolonged cell culture caused aberrant/ectopic expression of these genes and that RT-PCR data from lymphoblastoid cells do not allow any conclusion as to their quantitative expression in affected tissues. For HOXB9, CALCOCO2, and SNF8, we identified transcribed SNPs in our patient. Sequencing of RT-PCR products revealed biallelic expression of these genes in the patient's lymphoblasts.

Discussion

HOX genes are crucial for positioning organs along the anterior–posterior axis of the developing vertebrate embryo.22, 23 In the human genome, 39 HOX genes belonging to 13 paralogous groups are arranged in four clusters on chromosomes 7p15.3 (HOXA), 17q21.3 (HOXB), 12q13.3 (HOXC) and 2q31 (HOXD).24 Gene expression within each cluster follows the so-called ‘colinearity rule’. This means that HOX genes in the 3′ region of a cluster are activated first and in anterior areas of the embryo, whereas genes in the 5′ region are transcribed later and in more caudal areas.22 In certain tissues, HOX genes also display ‘quantitative colinearity’. The gene at the 5′ end shows the highest expression level and the more 3′ located genes are expressed at progressively lower levels.25 Real-time PCR analysis of the 39 human HOX genes in normal adult organs revealed that 5′ located HOX genes are expressed preferentially in organs of the caudal body parts. In general, the expression patterns of neighboring HOX genes in the same cluster are more similar to each other than to their paralogs in other clusters.16

Natural mutants of mammalian Hox genes are rare. Hypodactyly mice result from a deletion within the Hoxa13 locus.26 The different functions of individual members of the Hoxb gene cluster have been extensively studied in knockout mice. Hoxb1, Hoxb2, and Hoxb3 are involved in hindbrain specification.27, 28, 29 Hoxb2 and Hoxb4 are important for sternum development.28, 30 Hoxb3 and Hoxb4 play roles in hematopoietic stem cell regeneration and proliferative response.31 Hoxb5 is necessary for differentiation of angioblasts and mature endothelial cells from their mesoderm-derived precursors.32 Hoxb6 controls generation, proliferation, and/or survival of erythroid progenitor cells.33 Hoxb7 knockout mice have fused ribs in the upper thoracic region.34 Hoxb8 knockouts show abnormal grooming behavior and central nervous system abnormalities,35 whereas ectopic expression of Hoxb8 results in a mirror-image duplication in the forelimb.21 Hoxb9 together with its paralogs Hoxa9 and Hoxd9 controls development of the mammary gland in pregnancy.36 Hoxb13 knockouts display an overgrowth of all major structures derived from the tail bud, including the secondary neural tube, the caudal spinal ganglia, and vertebrae, implying that Hoxb13 may function as an inhibitor of neuronal cell proliferation, an activator of apoptotic pathways in the neural tube, and a general repressor of growth in the caudal vertebrae.37 Hoxb13 is also required for normal differentiation and secretory function of the ventral prostate.38 Targeted deletion within the mouse Hoxb cluster from Hoxb1 to Hoxb9 resulted in a series of single segment anterior homeotic transformations along the cervical and thoracic vertebral column and defects in sternum morphogenesis.39 Interestingly, Hoxb13 expression was not affected by Hox1–Hox9 deletion, suggesting that its regulatory element(s) is located 5′ to Hoxb9. However, so far there is no evidence from animal models implicating a HOXB gene(s) in postaxial hexadactyly, which is the most specific symptom of our patient.

In humans, several genetic disorders of the skeletal system are due to mutations in different HOX genes. Single basepair deletions in HOXA11 cause amegakaryocytic thrombocytopenia and radioulnar synostosis.40 Heterozygous mutations in HOXA13 are associated with hand–foot–genital syndrome and the closely related Guttmacher syndrome.41, 42 A missense mutation in HOXD10 segregates in a family with rocker-bottom feet and Charcot-Marie-Tooth disease.43 Polyalanine coding repeat expansions in exon 1 of HOXD13 cause synpolydactyly.44, 45 Intragenic frameshift deletions,46 missense mutations in exon 247, 48 and an acceptor splice site mutation49 have been associated with novel hand and/or foot malformations. It is important to note that breakpoints in chromosome 2q31 near the HOXD cluster were found in four independent patients with balanced chromosome rearrangements and various limb and skeletal malformations.11, 12 Similar to our case, these disease-associated chromosome rearrangements did not disrupt a gene(s) but most likely exerted position effects by disturbing normal HOXD gene expression. Interactions between regulatory elements at the 5′ border of the Hoxd cluster and the digit enhancer that lies at least 100 kb 5′ to the Hoxd gene cluster appear to be crucial for establishing quantitative colinearity in the mouse Hoxd cluster.50

To our knowledge, the patient described here represents the first example of a disease-associated translocation breakpoint near the HOXB cluster. Circumstantial evidence suggests that misregulation of a HOXB gene(s) by position effect is responsible for the patient's phenotype. Mouse Hoxb genes are known to play important functions in skeletal and central nervous system development. The HOXB cluster lies within 30 kb of the breakpoint on the patient's chromosome 17q21.3 and was moved from a euchromatic chromatin environment in close proximity of a repetitive chromatin domain (telomeric hexamer repeats). The breakpoint region at the 5′ end of the HOXB cluster contains evolutionarily conserved non-genic sequences that may act as cis-acting DNA elements. Unfortunately, expression analyses are limited with human patients. In our patient with mental retardation and limb and rib malformations, it was not possible to study gene expression in the relevant cell types/tissues. It is well known that HOX and other developmental control genes show tissue-specific expression patterns.51, 52 The observed differences between established lymphoblastoid cell lines of the patient and normal control individuals appear to be secondary to the genetic condition of the donors. Nevertheless, it is plausible to assume that the translocation breakpoint near the 5′ end of the HOXB cluster interferes with the colinear expression pattern of HOXB genes in organs that are affected in our patient. Separation of mouse Hox genes from regulatory sequences located at a distance outside the cluster is known to cause misexpression.53, 54 Because both HOXB1355, 56 and the PRAC genes56, 57 in the breakpoint region have been implicated in human prostate cancer, the patient will be followed for this. The described t(12;17) translocation patient provides a promising starting point for the identification and characterization of regulatory elements of the HOXB cluster. Indeed, alterations of HOX genes may be an underestimated cause for human genetic disease.