INTRODUCTION

Congenital heart disease (CHD) is one of the most common types of birth defects, occurring in 1% of live births and 10% of spontaneously aborted fetuses. Genetic aberrant expressions of certain genes may lead to defects of heart development.1 In the past decade, molecular genetic studies have exploited CHD families with multiple affected individuals and have provided further insight into the genetic bases of several forms of CHD, such as atrial septal defect or patent ductus arteriosus. Fifteen representative chromosomal disorders associated with CHDs and 28 genes associated with CHDs in the young were summarized in Pierpont et al (2007).2 The CHDs are present in 75% to 80% of patients with a Homo sapiens autosomal 22q11 (HSA22q11) deletion and most cases were regional hemizygosities or monosomics, and resulting in haploinsufficiencies,3,4 including syndromic cases (CHD associated with at least 1 extracardiac anomaly) and non-syndromic familial cases.2,5,6 Septation of heart chambers starts at early stages in embryogenesis and is completed only at birth with the closure of the foramen ovale. Defects in heart separation, including atrial septal defect, atrioventricular canal septal defects, conotruncal septal defects, and ventricular septal defect (VSD), represent a major cause of congenital heart disease.7

A substantial proportion of patients with double outlet right ventricle, interrupted aortic arch, posterior malalignment type VSD, pulmonary atresia with VSD, subpulmonary VSD, truncus arteriosus, transposition of great arteries, tetralogy of Fallot, and VSD associated with other conotruncal heart defects have an HSA22q11 deletion.3,8-11 Fluorescent in situ hybridization (FISH) with DNA probes from the DiGeorge Syndrome chromosome region is the clinical diagnostic strategy that has been used for molecular genetic testing of HSA22q11 deletion.9,11,12 However, microsatellite genotying provides a higher resolution (locus-based) than karyotyping (Mb-based resolution) to diagnose HSA22q11 microdeletions in karyotypically normal patients.5,8,13 Therefore we selected microsatellite genotyping technology to perform the current study. Also, in our previous genetic analysis of HSA22q11 markers in CHDs, a significant deviation from the expected level of heterozygosity at a microsatellite D22S1648 locus implied that the HSA22q11 region might be hemizygous among syndromic CHD patients.13 The HSA22q11 deletion in syndromic CHD patients was further confirmed by haplotype analysis, dual-labeling FISH and quantitative polymerase chain reactions (PCR).13,14 To improve our understanding of whether specific CHD types are relevant to HSA22q11 deletion and to further narrow down causative candidate genes, microsatellite genotyping and gene dosage analysis were performed in this study.

METHODS

Blood Samples and DNA Extraction

A total of 82 CHD families that had been referred to Kaohsiung Veterans General Hospital (Kaohsiung, Taiwan), including 261 individuals (85 patients and 176 siblings and parents) participated in this study. Informed consent was obtained from each participant. The age of the 85 patients ranged from 1 to 14 years of age (mean [SD], 3.6 [2.9]). They were randomly selected from our pediatric cardiology outpatient department. Genomic DNA was prepared from peripheral blood leukocytes using the salting-out DNA extraction method.

Microsatellite Genotyping

Microsatellites are polymorphic loci present in genomes that consist of repeating units of 1-6 bp in length, typically neutral and co-dominant. Because their copy numbers could be highly polymeric within and between individuals, these markers have been used to identify loss of heterozygosity (LOH), ie, loss of 1 allele or more alleles in an individual, compared to the parents.15 A total of 9 polymorphic microsatellite markers covering the HSA22q11 region and 1 distal marker, D22S315, located at HSA22q12.1, were used for genotyping. Nine highly polymorphic markers in Asia population spanning a total of 8.1 Mb were selected based on previous studies13,14 and the distal marker, D22S315, was selected from Map Viewer (http://www.ncbi.nlm. nih.gov/mapview/; NCBI, USA)16 (Figure 1). Primer sequences for microsatellite genotyping (not shown) were directly retrieved from Electronic PCR (http:// www.ncbi.nlm.nih.gov/sutils/e-pcr/; NCBI, USA).16 To multiplex PCR-based genotyping, one primer of each pair was either labeled with FAM (D22S264, D22S303, D22S420, D22S941, D22S1638, and D22S1648), HEX (D22S427, D22S944, D22S1623), or TET (D22S315). Multiplex PCRs were performed in primers with different fluorescent labels or with the same label but a different range (in size) of amplicons. The PCR conditions and microsatellite genotyping methods have been described in our earlier studies.5,13 For each sample, all markers were analyzed at least 3 times to confirm the genotypes.

Figure 1. Genomic organization of the Homo sapiens (HSA) autosomal 22q11 (HSA22q11) and markers used in this study. D22S- are polymorphic microsatellites. A total of 8 low copy repeats, termed LCR22-1 to -8, have been mapped to this region. The common ~3-Mb typically deleted region, which is present in more than 85% of velo-cardio-facial syndrome (VCFS) and DiGeorge syndrome (DGS) patients and the ~1.5-Mb deletion (~8%) are shown. Atypical deletions are not shown. Delection in VCFS/DGS: common ~3-Mb deletion (>85%); ~1.5-Mb deletion (~8%).

Bioinformatics Strategies to Identify Candidate Genes

Ensemble Genome Browser17 (http://www.ensembl. org/index.html), UCSC Genome Browser18 (http:// genome.ucsc.edu/) and NCBI Map Viewer16 were used to identify CHD-related candidate genes in the HSA22q11 region. Tissue specificities of candidate genes were inspected with the Swiss-Prot database19 (http://www.expasy.ch/sprot/). The FatiGOplus data mining tool20 was further employed to search for genes with similar Gene Ontology terms21 to one well-known CHD candidate gene, TBX1.

Gene Dosage Analysis by Quantitative Polymerase Chain Reaction

Dosage analyses of candidate genes were examined by quantitative PCR (iCycler thermal cycler, BioRad, CA, USA) along with comparative Ct method (DDCt method).22 Primer pairs were designed with the Beacon Designer 2.1 (Biomedal, Spain) (Table 1). In addition to 7 candidate genes, an X chromosome-linked gene, 6-phosphofructo-2-kinase (PFKFB1),23 was analyzed as a dosage reference of single-copy gene. Each quantitative PCR experiment was performed in a 15 µL volume (7.5 µL of SYBR® Green Mix and 0.75 µL of forward and reverse primers and 6 µL [12.5 ng/µL] genomic DNA). The concentration of each primer was 4 µM and the annealing temperature was optimized to 55-60°C. Amplifications were carried out in 96-well plates for 55-60 cycles. Equal amounts from an aliquot of 12 genomic DNA samples were pooled to be a common reference genomic DNA for further calculation of the relative expression fold in each candidate transcript. To assure equivalent amplification efficiencies of all primer sets used in this study, validation assays were performed according to Senchenko et al24 (2003).

Quantitative PCR data were expressed as mean (SD). The starting copy number of unknown genomic DNA samples was determined relative to the known copy number of the calibrator sample (common reference genomic DNA) using the following formula: DDCt = DCt(candidate gene, unknown sample-[ACTB, unknown sample])-DCt(candidate gene, common reference gDNA-[ACTB, common reference cDNA]). The relative gene copy number was calculated by the expression 2-DDCt. Prior to the dosage analysis on candidate genes, standard ranges of 1- and 2-copy gene dosages were established by quantifying the X chromosome-linked housekeeping gene, PFKFB1,23 in genomic DNA from 12 males and 12 females.

RESULTS

The Phenotypes of CHD patients

No CHD phenotypes were reported in parents of any recruited families. The VSD was identified in 29 patients (34.1%), which is the most prevalent CHD in this cohort. The tetralogy of Fallot and the atrial septal defect were identified in 10 patients (11.7%) and 7 patients (8.2%), respectively. The VSD cases were further subtyped as follows: (1) membranous (n=18); (2) subpulmonary (n=6); (3) muscular (n=4); and (4) atrioventricular canal (n=1). The presence more than 3 anatomical defects in 1 patient was defined as a complex CHD (n=4; 4.7%) (Table 2). Four families, ie, numbers 102, 106, 260, and 352, had two children affected. Two children in family 352 had a membraneous VSD. The VSD was also diagnosed in two children in family 102, but the subtypes were apical muscular and membranous. Two CHD patients in families 106 and 260 were diagnosed as membraneous VSD (106-1) and pulmonary stenosis (106-2), and membraneous VSD (260-1) and atrial septal defect (260-2), respectively (Table 3).

The Loss of Heterozygosity in HSA22q11 Was Identified in Several Types of CHDs

In addition to 85 patients and 164 unaffected parents, 12 unaffected siblings were genotyped for 10 microsatellites as well (partial data are shown in Table 3). While both parents share one allele (eg, D22S1623 in family 260) or both alleles (eg, D22S941 in family 297), the LOH statuses in their offspring might not be identified (uninformative, denoted as "UI" in Table 3 and Figure 2), indicating that no information was provided. Namely, UI situation does not exclude the possibility that the region is involved. Alternatively, failure to inherit any allele from the parents with homozygotes [eg, D22S1648 in family 100, child 1 (100-1) and 3 (100-3)] or heterozygotes [eg, D22S420 in family 106, child 2 (106-2), pulmonary stenosis case; Table 3) might result in LOH. Number and percentage of patients with HSA22q11 deletions by microsatellite genotyping and LOH analysis for different types of CHD are shown in Table 2. Forty-two (49.4%) of 85 CHD patients were identified to have at least 1 LOH in the HSA22q11 region.

Figure 2. Combination of loss of heterozygosity (by microsatellite genotyping) and gene dosage (by quantitative polymerase chain reaction) data in 16 ventricular septal defect patients. For microsatellite markers (D22S-). H indicates retention of heterozygosity; L, loss of heterozygosity; UI, uninformative; for candidate genes: 1 indicates 1 copy; 2, 2 copies; UI, uninformative.

Although the children of family 352 were both affected by membranous VSD, microsatellte genotyping was not able to identify any LOH within the HSA22q11 region. Distinct HSA22q11 LOHs were identified in family 102 with 2 different subtypic VSD patients and in family 106 with 2 different pathologic CHDs (Table 3). The LOH of D22S944 was identified in both VSD and atrial septal defect patients of family 260. LOH HSA22q11 was not found in 2 of 4 families with more than 1 child, none of whom was affected. However, in the other 2 families (157 and 201), HSA22q11 LOHs were also identified in non-affected individuals (Table 3). Of 3 children in family #157 only 1 was affected by VSD, both siblings showed HSA22q11 LOHs. Similarly, in family 201, HSA22q11 LOH was detected in the other non-affected child (Table 3).

Identification of Candidate Genes

A total of 75 genes were first identified to be intersected from 3 genome browsers in the HSA22q11 region. NCBI gene annotations indicated that only 20 of them were related to embryogenesis, mesoderm differentiation or heart development (data not shown). By using the information of tissue specificities from the Swiss-Prot database, 5 genes, ie, CLTCL1, DGCR2, DGCR14, GNB1L, and TUBA8, were verified to be expressed in heart specifically. Another gene, HIRA, was found to be identical to TBX1 at level 5 of gene ontology terms in molecular function, cellular component, and biological process. The TBX1 is a gene encoding the T-box transcription factor, which is hemizygously deleted on HSA22q11, and was found to be a strong candidate for DiGeorge syndrome.25 As a consequence, including TBX1, 7 candidate genes within HSA22q11 LOH region were identified.

Gene Dosage Analysis

The results of the PCR efficiency assay for every candidate gene was plotted with the log input concentration for each dilution of DNA templates on the x-axis, and the difference in DCt (candidate gene-ACTB) for each dilution was plotted on the y-axis (data not shown). When the absolute value of the slope of the line was less than 0.1, the comparative Ct method would be appropriate.24 As shown in Table 1, all primer sets in this study had slopes <0.1.

The genomic dosage profiles of PFKFB1 gene in males (n=12) and females (n=12) were subsequently established. Because PFKFB1 is an X chromosome-linked gene, average relative fold changes (2-DDCt) in the male and female genomes, represent the relative gene dosages of 1 copy and 2 copies. By quantitative PCR, average gene dosages (ie, the relative fold changes, 2-DDCt) of 1 copy (PFKFB1 in males) and 2 copies (PFKFB1 in females) were detected as 0.52 (0.15) (mean [SD]) and 1.0 (0.17), respectively. Thus, ranges of 95% coverage (mean [2SD] in 2-tail distribution) of gene dosages with 1 and 2 copies were 0.22 (0.82) and 0.66 (1.34). "Uninformative" (UI) was designated when gene dosages fell below 0.22, between 0.66 and 0.82, or above 1.34 (data not shown).

Genomic dosages of 7 candidate genes, ie, TUBA8, DGCR2, DGCR14, CLTCL1, HIRA, TBX1, and GNB1L (from centromere to telomere), in HSA22q11 region as identified by bioinformatic strategies were further analyzed by quantitative PCR. Due to inadequate DNA samples in patient 119-1, a total of 16 VSDs encompassing at least 1 LOH in HSA22q11 region were examined. Genomic dosages of candidate genes were compared with ranges of the standard fold change (2-DDCt) as established above. No gene deletion was detected in six VSD patients (102-1 apical muscular; 101-1, 105-1, 194-1, 202-1, and 260-1 membranous type; data not shown). Patients 138-1 (subpulmonary) and 106-1 (membranous) were identified to be heterozygous deletion of 6 and 5 candidate genes, respectively. Sixteen examined VSD patients preserved both copies of TBX1 gene. The HIRA or TUBA8 were examined as the only heterozygous deleted genes in 2 (115-1 subpulmonary; 102-2 membranous) and 1 (233-1 subpulmonary) patients, respectively (Figure 2). Among 7 candidate genes, heterozygous deletion of TUBA8, which was found in 8 probands, was predominant. Heterozygous deletions of HIRA and GNB1L genes were found to be the second most prevalent (6 cases in each gene) (Figure 2).

DISCUSSION

The HSA22q11 region is prone to rearrangements that lead to congenital anomaly disorders due to its genomic organization of low-copy repeats. The most common rearrangement is the 3.0- and 1.5-Mb hemizygous deletions detected in velo-cardio-facial syndrome/DiGeorge syndrome patients26 (Figure 1). Both intra- and interchromosomal homologous recombination mechanisms mediate these deletions.27 However, atypical deletions are more common than has been reported.28 Accordingly, we hypothesized that the HSA22q11 region encompasses genes that associate with cardiac and craniofacial anomalies. Earlier, HSA22q11 deletions were found in 12 (10%) of 125 VSD patients by FISH examination, using the Oncor cosmid probe N25(D22S75).10 With Vysis probe set targeting TUPLE1 (ie, HIRA), Beauchesne et al12 (2005) identified 6 patients (5.8%) to have HSA22q11 deletions (3 tetralogy of Fallot, 2 pulmonary atresia/VSD, and 1 truncus arteriosus). Combining several probes (100c10, 48F8, KI-506, and Sc11.1), 50 (48%) of 104 newborns with conotruncal defects and HSA22q11 deletions were detected.29 Most of these cases were syndromic CHD. Few studies have used polymorphic microsatellites that span the HSA22q11 region to detect microdeletion.5,8,13 Microsatellite genotyping have detected 10 (48%) of 21 patients with tetralogy of Fallot and 10 (100%) of 10 patients with either velo-cardio-facial syndrome or DiGeorge syndrome showing HSA22q11 deletions in our previous report.13 In the present study, we as well achieved higher identification efficiency than previous FISH results.2, 10 The possible explanations were (1) higher resolution of microsatellite genotyping compared to FISH analysis and (2) some of our patients were syndromic CHD cases. It was difficult to evaluate whether patients had syndromic CHD because subtle noncardiac features might not be apparent in infancy. Some of these features are age-dependent. In addition, the diagnosis of facial dysmorphia is more subjective.30 Syndromic CHDs might exist in our patients. However, we speculated that the probable case number was small because they had been evaluated initially for facial dysmorphia by at least one cardiologist. We therefore suggest that the microsatellite genotyping technology provides a higher resolution than traditional FISH analysis in such diagnoses.

After genotyping, no LOH in HSA22q11 was identified from 43 CHDs patients (including 1 family with 2 VSD patients), indicating other chromosomal regions might be responsible for their abnormalities. On the other hand, the HSA22q11 LOHs were identified in both patients in 3 of 4 families with 2 CHD-affected individuals, suggesting that genes within this region might be responsible for more than 1 CHD phenotype and that these CHDs might not be sporadic but fundamentally familial. To our surprise, LOHs in HSA22q11 were also identified in 3 non-affected children from 2 patients' families, prompting us to monitor their CHD incidence carefully. These results indicate the clinical significance of screening all siblings of each proband for early diagnoses and of the design of treatment strategies.

Bioinformatics strategies additionally identified 7 candidate genes located within the HSA22q11 LOH region. To narrow down specific gene(s) that cause VSDs, genomic dosages of 7 candidate genes in 16 probands were examined by quantitative PCR. Results exhibited recurrent heterozygous deletion of GNB1L, HIRA, and/or TUBA8 genesin membranous and subpulmonary VSDs. In HSA22q11 region, TBX1 was thought to be a major candidate gene to the cardiac phenotype or its severity in patients carrying the HSA22q11 microdeletion.31 Although Tbx1+/Tbx1- heterozygous mutant mice survived at normal Mendelian ratios and had only mild cardiovascular defects, Tbx1-/ Tbx1- homozygous mice died at birth with cleft palate, thymus gland aplasia, a single outflow tract and VSD.25 Unexpectedly, in this study, no hemizygous or homozygous deletion of TBX1 gene was identified in 16 VSD probands with HSA22q11 LOH. In contrast, recurrent heterozygous deletion of GNB1L, HIRA and/or TUBA8 genes was found. These are genes that are frequently deleted in the DiGeorge syndrome critical region, thus suggesting that heterozygous deletion might play a crucial role in the development of VSD. Alternatively, mutations of TBX1 gene in patients with HSA22q11 deletion syndrome but without 22q11 microdeletion or any other detectable rearrangements in this region were identified.32,33 In our dosage screening, heterozygous deletion of any candidate gene was not observed in 6 of 16 VSD probands with H22q11 LOH, implying that potential mutations of TBX1 or other genes in these patients might exist; however, further experiments will be required to validate these assumptions.

Of 3 recurrently deleted genes, HIRA is a homolog of yeast transcriptional corepressors Hir1p and Hir2p, possessing 7 WD dipepetide motifs and 1 LXXLL motif in its N-terminal and C-terminal regions, respectively.34 Whole mount in situ hybridization indicated HIRA mRNA is expressed in a temporal and spatial pattern during early embryogenesis, with high levels of expression in the neuroepithelium during neurulation.35 By exposing a portion of the premigratory chick neural crest to phosphorothioate end-protected antisense oligonucleotides, ex ovo, following by orthotopic backtransplantation to the untreated embryos, Farrell et al36 (1999) showed that the functional attenuation of chicken HIRA in the chick cardiac neural crest results in a significantly increased incidence of persistent truncus arteriosus, a phenotypic change characteristic of DiGeorge syndrome. But this does not affect the repatterning aortic arch arteries, the ventricular function, or the alignment of the outflow tract.36 High levels of mouse HIRA transcripts were detected in the cranial neural folds, frontonasal mass, the first 2 pharyngeal arches, circumpharyngeal neural crest, and limb buds.37 The TUBA8 is a tissue-specific isoform of alpha-tubulin that is highly conserved in human and mouse. Reverse transcription-PCR assay performed on a panel of 16 different human tissue cDNAs using primers specific for TUBA8 isoform found preferential expression in heart, skeletal muscle, and testis. Moreover, sequence comparison of TUBA8with other known alpha-tubulins shows that TUBA8is the most divergent member of the mammalian alpha-tubulin family. The sequence peculiarity of the TUBA8 in the human and mouse suggests that it might have functional significance and, according to the multi-tubulin hypothesis, that it might play specific functional roles in the cell cytoskeleton.38 The GNB1L encodes a G-protein b-subunit-like polypeptide. Northern blot analysis by Gong et al39 (2000) revealed its high expression level in the heart, but low expression levels in brain, placenta, lung and pancreas, suggesting it might be a strong candidate for heart development. A chicken model using vector-based RNA interference to knock down TUBA8 and GNBlL genes in early embryogesis underway in our laboratory will perhaps provide functional interpretations of these VSD candidate genes in the near future.

Limitations

Minor noncardiac features might not be diagnosed at an earlier stage in CHD patients. Hence karyotyping or detection for HSA22q11 deletion could be neglected, resulting in limitations to the power of genetic counseling. In patients without the HSA22q11 deletion, CHD might be a result of other chromosome/gene abnormality/ deletion/mutation.

CONCLUSIONS

By applying microsatellite genotyping within HSA22q11 region to CHD families along with the LOH analysis, we identified the HSA22q11 LOH in several CHD types. Recurrent heterozygous deletion of 3 candidate genes, HIRA, TUBA8, and GNB1L, is potentially responsible for some VSD cases. More than 1 affected individual within 2 families with HSA22q11 LOH strongly suggests that CHD is a familial disease. To screen HSA22q11 LOHs in siblings of the proband will be beneficial for early diagnosis and the design of effective treatment strategies.

ACKNOWLEDGMENTS

Special acknowledgements are extended for the financial support provided by Kaohsiung Veterans General Hospital (VGHKS) (grant number VGHKS94-78; to CL Lee) and supported in part by National Sun Yat-sen University-Kaohsiung Medical University Joint Research Center, Kaohsiung (to YL Shiue).

ABBREVIATIONS

CHD: congenital heart disease
FISH: fluorescent in situ hybridization
HSA22q11: Homo sapiens chromosome 22q11
LOH: loss of heterozygosity
PCR: polymerase chain reactions
VSD: ventricular septal defect

SEE ARTICLE ON PAGES 242-5

Correspondence: Dr Yow-Ling Shiue,
Institute of Biomedical Science, National Sun Yat-sen University,
70 Lain-Hai Road, Kaohsiung, Taiwan
E-mail: ylshiue@mail.nsysu.edu.tw

Received April 4, 2008.
Accepted for publication November 6, 2008.