NKX2–5 is a homeodomain containing transcription factor crucial to mammalian heart development.¹ Lack of NKX2–5 in animal models results in lethality due to impaired cardiac looping. Genetic analysis of lymphocytic DNA identified germline mutations in the human NKX2–5 gene which are associated with cardiac anomalies, but most families have single and unique mutations.^2–10 Germline NKX2–5 mutations are rare among sporadic cases of congenital heart disease (CHD), genotype–phenotype correlation is difficult, and mechanisms leading to complex CHD remain unclear.

Next to inherited mutations, the importance of somatic mutations during early embryogenesis is now being recognised, but rarely studied.^11,12 Disease associated or disease causing somatic mutations are undetected by genetic analysis of lymphocytic DNA alone, and mosaicism may reduce the likelihood of detection in the affected tissue. Therefore, mutations present in cardiac tissues may be absent or sporadic in blood samples of the same individual. Unlike previous studies, we searched for mutations in the diseased tissues of 68 explanted hearts of patients with definitive atrial (ASD), ventricular (VSD), and atrioventricular (AVSD) septal defects. These patients, however, had complex cardiac malformations and died mostly in early infancy (fig 1A and B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

 NKX2–5 mutation analysis in complex CHD. (A) Kaplan-Meier survival curves showing AVSD and VSD patients with lower life expectancy than ASD patients. (B) Examples of cardiac malformations in patients. (C) Three NKX2–5 fragments were amplified from genomic DNA (size of about 2 kb) isolated from archival cardiac tissues using three known primer pairs: 1F24/1AR (485 bp), 2F/2R24 (472 bp), and 3F/3R24 (573 bp). About 20–50 ng genomic DNA was used for PCR reaction. M (kb ladder): 50, 100, 150 ng (λ DNA standard). (D) Detection of mutations by direct sequencing, verification by PCR-RFLP, or cloning of fragments with heterozygous loci, separating the two variant alleles.

The human NKX2–5 gene maps to chromosome 5q34 and consists of two exons encoding a protein of 324 amino acids. The homeodomain (HD) of NKX2–5, which lies within exon 2, consists of three α helices. Helix 3 is important for DNA binding specificity.¹³ Other conserved regions of NKX2–5 are the TN domain and NK2 specific domain (NK2-SD). Using validated PCR primers,^3,5 we amplified fragments from the two exons (fig 1C). Mutations were detected by direct sequencing of both DNA strands and confirmed by PCR-RFLP or cloning of the sequenced fragments, allowing separation of the two variant alleles in a heterozygous locus (fig 1D).

We identified two NCBI dbSNPs (rs2277923 and rs703752) and 35 non-synonymous NKX2–5 mutations in diseased heart tissues of patients with complex cardiac malformations (table 1). Three of these mutations (Arg25Cys, Thr178Met, and Ala219Val) are known. We detected 29 mutations in VSD patients, 12 in ASD, and 14 in AVSD. We found one mutation in the TN domain (Asn19Ser) and five mutations (Lys183Glu, Gln187Ter, Lys192Thr, Lys192Arg, Lys194Arg) in the third helix of HD (table 1). We observed only heterozygous genotypes based on the sequence electropherograms. As determined by haplotype analysis, this heterozygosity may simply be a reflection of genotypes of mixed wild type and mutant cells. Therefore, the occurrence of homozygous genotypes in analysed diseased heart tissues cannot be excluded, although germline homozygous mutations in animal models are lethal. Further, somatically occurring mutations as seen in cancerous diseases result in loss of heterozygosity. Again, such loss of heterozygosity may result in cardiac anomalies as seen in our study. The 35 mutations were absent in lymphocytic DNA of 50 healthy individuals and in DNA from six frozen normal hearts.

The explanted hearts were formalin fixed more than 40 years ago, but analysis of NKX2–5 and five other cardiac transcription factor genes (46 709 nt) in 10 normal hearts from the same collection of formalin fixed hearts basically revealed no sequence alterations except dbSNPs, showing the reliability of using valuable archival material. We found a non-synonymous NKX2–5 substitution (T984A, Cys270Ser) in two individuals who died as babies, but this mutation was different from those in patients with CHD. This sequence alteration may just be a polymorphism.

The NKX2–5 mutations were detected in heart tissues sampled within the septal defects of the hearts. To determine whether these mutations were also present in normal heart tissues (such as in patients with VSD, where tissue samples were taken from unaffected atria), we compared 27 distinct mutations in normal and diseased heart tissues of the same patients (table 1). We analysed a total of 92 samples, 146 PCR fragments, and 77 598 nucleotides and found mutations mainly absent in normal heart tissues. The absence of mutations in matched normal heart tissue of the same patient indicates the somatic nature and mosaicism of NKX2–5 mutations.

AVSD is highly associated with Down’s syndrome (trisomy 21)¹⁴ and some of our patient cohorts had Down’s syndrome. Analysis of 18 AVSD patients with Down’s syndrome for NKX2–5 mutations found no difference in the mutation spectrum between syndromic and non-syndromic AVSD patients (tables 1 and 2). The occurrence of both trisomy 21 and NKX2–5 mutations suggests genomic chaos in the heart tissues of patients with complex CHD. We also detected common mutations among unrelated patients and some mutations were specific to ventricular or atrioventricular septal defects (table 2). We found common mutations such as A232G (Asn19Ser), A723G (Lys183Glu), C1012T (Ser279Phe), C1018T (Ala281Val), and A1072G (Asp299Gly) (table 2). We also found multiple mutations in 57 patients; 66% had five or more mutations. The range of multiple mutations per patient was 2–14, 2–7, and 2–10 in VSD, ASD, and AVSD patients, respectively (table 2).

The presence of multiple gene mutations in a patient is an unlikely occurrence in inherited mutations, but multiple mutations are frequently observed in high capacity cell division such as in cancer.¹⁵ For example, mutational analysis of the p53 tumour suppressor gene in human mammary carcinomas evidenced up to six mutations in this gene.¹⁶ It is, however, uncommon to find carriers of multiple mutations within one gene, as described here. We observed, on average, five mutations in VSD and AVSD subjects, and three mutations in ASD subjects. Moreover, nine of 29 VSD subjects, nine of 23 AVSD subjects, and three of 16 ASD subjects had more than five mutations. Nonetheless, recent genetic studies also demonstrate the occurrence of multiple mutations in human genes including the genes coding for butyrylcholinesterase¹⁷ and hALG6, which codes for a glucosyltransferase.¹⁸ In both genes, three mutations per patient were identified. In our study, it remains unclear whether multiple NKX2–5 mutations are a result of primary or secondary events.

We cloned PCR products containing heterozygous loci, as seen on sequence electropherograms. In most cases, a locus was scored heterozygous when the ratio of the second peak to the highest peak was at least 50%. After cloning, we sequenced random clones, and scored only those loci previously identified as heterozygous after direct sequencing of the PCR product. Surprisingly, we obtained more than the two expected haplotypes in one individual, and identified clones carrying all the reference alleles only, mutated alleles only, or a combination of both (table 3). In most analysed clones, both alleles were affected. A likely explanation for multiple haplotypes would be a mixed population of cardiomyocytes carrying different mutations or de novo chromosomal rearrangements and gene duplications in the heart tissues of patients affected by various mutations. Heart defects are observed following chromosomal abnormalities including chromosome 5.¹⁹ It will be the subject of future research to determine the origin of these multiple haplotypes.

Our patient cohort of rare and complex cardiac malformations had non-synonymous mutations in other cardiac transcription factor genes including GATA4, HAND1, and TBX5, although mutation frequencies varied considerably (data not shown). Simultaneous occurrence of multiple mutations in several genes is uncommon, but may be explained by the sheer complexity of the observed malformations, with patients carrying multiple defects in the heart. Somatically occurring mutations may affect various cell types at different stages of organogenesis, with different transcription factor programs taking place. These patients, however, did not carry mutations in the sequences affecting binding regions of MEF2C, an upstream determinant of cardiac development.²⁰ NKX2–5, HAND1, and MEF2C are all located on chromosome 5: 5q34, 5q32, and 5q14, respectively, but differ profoundly in the spectrum of mutation. Deleterious mutations in MEF2C would most likely lead to early termination of pregnancy, for example, intrauterine resorption or spontaneous abortion due to its pivotal function in the transdifferentiation process during organogenesis of the heart.

We also analysed lymphocytic DNA in seven families and four unrelated patients with CHD, but did not identify NKX2–5 mutations, except for one synonymous substitution in an unaffected family member. Further sequencing of nine cardiac transcription factor genes evidenced only dbSNPs. Thus, analysis of lymphocytic DNA may not reveal the genetic aberrations which are associated with CHD. In conclusion, our results demonstrate absence of mutations in matched normal heart tissues of the same patients’ hearts, multiple mutations as well as multiple haplotypes in patients. These observations strongly suggest that NKX2–5 mutations occur somatically and that such somatic mutations are associated with complex CHD. These mutations probably resulted from postzygotic errors in cell division, particularly in cardiac progenitors cells. Presumably, various cell types at different stages of organogenesis with different transcription factor programs in place are affected by somatic mutations. Determining the onset of somatic mutations, effects of multiple mutations, and breakdown of transcription factor crosstalks resulting from genomic instability in cardiac cells, will be an interesting avenue to explore in future stem cell research.