The next step involved applying genetic methodologies to identify candidate genes causing this disease. In this case, the project benefited enormously from the fact that the parents of the five siblings were third cousins, unaffected, and readily available for study. The most appropriate genetic platform to employ was the million SNP array, which detects copy number variants (deletions and duplications) and also identifies regions of homozygosity. In a product of a consanguineous mating, most homozygous areas derive from a common ancestor, and a single pathogenic mutation in a shared ancestral gene could cause a recessive disease. Hence, we were looking for a region of the human genome in which all five affected siblings were homozygous, but the parents were heterozygous. Only one area satisfied these requirements – a 22-Mb region on chromosome 6. That area contained 92 genes; any one of them could have contained a homozygous mutation that caused the arterial calcifications. We needed an expert in vascular biology to identify the best candidates to pursue.

This was the third, and most crucial, step in discovering a new human disease, i.e. identifying and connecting with a basic scientist whose lab possessed expertise in the field. In this case, Manfred Boehm and Cynthia St. Hilaire of the National Heart, Lung and Blood Institute were authorities on vascular cell metabolism, and cell-based assays were extant in the Boehm lab. In concert with the UDP, whose clinicians provided a skin biopsy, St. Hilaire cultured an affected patient’s fibroblasts and performed expression analysis of candidate genes in the inherited homozygous region. Only one gene, NT5E, which encodes the CD73 protein, was differentially expressed. Direct sequencing of NT5E identified null mutations that resulted in complete inactivation of CD73 protein in all affected siblings. This newly identified vascular disease was referred to as ‘arterial calcification due to deficiency of CD73′ (ACDC). CD73 is a membrane-bound enzyme that converts extracellular AMP to adenosine and inorganic phosphate; this process was found to be severely impaired in fibroblasts from individuals with ACDC. The failure to generate extracellular adenosine leads to an increase in tissue non-specific alkaline phosphatase (TNAP), a key enzyme in tissue calcification in vitro and in vivo. Cultured fibroblasts of individuals with ACDC not only contained increased TNAP activity, but they also calcified. Finally, TNAP activity and in vitro calcification could be reversed by genetic rescue with NT5E cDNA or treatment with exogenous adenosine.

Again, consanguinity was involved, but this time in the form of a first cousin marriage. Whereas third cousins share 1/128 of their genes, first cousins share 1/8 of their genes, meaning that more than 10% of the human genome would be homozygous on SNP array analysis. As a consequence, SNP arrays would yield too many candidate genes to guess which ones would warrant further pursuit. Rather, WES was employed to identify genes containing homozygous variants. WES involves massively parallel sequencing of ~40 million bases that constitute the <2% of the human genome (3.2 billion bases) encoding genes. Using a whole exome platform, not every gene is perfectly sequenced (i.e. ‘covered’), but the breadth is so great that there is a good chance of detecting a causative variant. The problem is that non-pathogenic variants abound, so the exome sequences of other family members are required to reduce (‘filter’) the total number of variants. In the case of the two brothers, a total of 120,469 variants were identified among the two parents and two affected brothers; this was reduced to 59,482 variants not found in the 1000 Genomes database. Of these, 11 were homozygous and, of the 11, three were not present in dbSNP, a database of known polymorphisms. One homozygous missense mutation (c.1847G>A; p.Y616C) in the AFG3L2 gene was identified as the best candidate. AFG3L2 encodes a subunit of a mitochondrial AAA protease that removes damaged or misfolded proteins (Koppen et al., 2007).