Accurate measurement of gene copy number for human alpha-defensin DEFA1A3

Abstract

Background: Multi-allelic copy number variants include examples of extensive variation between individuals in the copy number of important genes, most notably genes involved in immune function. The definition of this variation, and analysis of its impact on function, has been hampered by the technical difficulty of large-scale but accurate typing of genomic copy number. The copy-variable alpha-defensin locus DEFA1A3 on human chromosome 8 commonly varies between 4 and 10 copies per diploid genome, and presents considerable challenges for accurate high-throughput typing.

Results: In this study, we developed two paralogue ratio tests and three allelic ratio measurements that, in combination, provide an accurate and scalable method for measurement of DEFA1A3 gene number. We combined information from different measurements in a maximum-likelihood framework which suggests that most samples can be assigned to an integer copy number with high confidence, and applied it to typing 589 unrelated European DNA samples. Typing the members of three-generation pedigrees provided further reassurance that correct integer copy numbers had been assigned. Our results have allowed us to discover that the SNP rs4300027 is strongly associated with DEFA1A3 gene copy number in European samples.

Conclusions: We have developed an accurate and robust method for measurement of DEFA1A3 copy number. Interrogation of rs4300027 and associated SNPs in Genome-Wide Association Study SNP data provides no evidence that alpha-defensin copy number is a strong risk factor for phenotypes such as Crohn's disease, type I diabetes, HIV progression and multiple sclerosis.

Figure 1

Schematic map of the DEFA1A3 CNV region, based on the human genome assembly (March 2006, NCBI36/hg18), and showing the expected structure of a 2-copy allele. DEFA1A3 genes are found in the copy-variable 19 kb full repeat sequence and (with exactly one copy per haplotype) in the centromeric partial repeat. The positions of the three SNPs rs7825750, rs4300027 and rs4512398 in the centromeric flanking DNA are indicated as a, b and c respectively. The specific measurement points used in this study in the determination of copy number are shown as differently-shaped symbols below the relevant locations, with the nearby DEFA4 gene providing the reference locus for the DEFA4 PRT measurement (triangles). The MLT1A0 PRT (star) has a target site only in full, and not partial, repeat sequences, and for this reason measures one copy less per haplotype than the other assays; calibration of MLT1A0 PRT values is undertaken using the predicted total repeat unit count per sample, for ease of comparison with other measures.

Figure 2

Copy number estimates determined by MLT1A0 PRT1 plotted against the corresponding copy number measure from DEFA4 PRT, showing the correlation between the two measures (r ²  = 0.731) and some evidence of clustering around integer values of 4, 5 and 6.

Figure 3

Segregation of haplotypes in the three-generation CEPH family 1341. In addition to the copy number, segregating haplotypes (boxed) are distinguished by the composition of other variants for which ratios were determined in this study. In this figure, only the DEFA3 content of each haplotype is shown (as the figure at the bottom of the haplotype box), but other variants can also be used to distinguish and characterise the haplotype content. In this family, both parents possess a 3-copy haplotype (haplotypes A and C), but they are distinguished by the fact that only the haplotype (A) carried by the father contains a copy of DEFA3.

Figure 4

Frequencies of measured integer DEFA1A3 copy number states in Europeans. (a) Histogram of copy number frequencies in 589 European DNA samples, (b) subdivided according to rs4300027 genotypes.