Two steroid 21-hydroxylase genes are normally present within the human major histocompatibility complex near the genes encoding the fourth component of complement (C4A and C4B). Steroid 21-hydroxylase is encoded by the CYP21 gene, while the highly homologous CYP21P gene is a pseudogene. We studied steroid 21-hydroxylase and complement C4 haplotypes in 33 Dutch patients (29 families) suffering form classical congenital adrenal hyperplasia (CAH) and in their 80 family members, and also in 55 unrelated healthy controls, using 21-hydroxylase and complement C4 cDNA probes. Eleven different haplotypes, defined in terms of gene deletions, gene duplications, conversions of CYP21 to CYP21P, and "long" and "short" C4 genes, were found. In 23% of the patients' haplotypes, the CYP21 gene was deleted; in 12%, it was converted into a CYP21P pseudogene. In the remaining 65%, the defect was apparently caused by a mutation not detectable by this method. The most common haplotype (with one CYP21 and one CYP21P gene) was significantly more often observed in patients with simple virilizing CAH than in those with salt-losing CAH. Comparison of the 21-hydroxylase haplotypes found in CAH patients from several countries shows evidence for considerable genetic variation between the groups studied.

Steroid 21-hydroxylase deficiency, the primary cause of congenital adrenal hyperplasia, is caused by defects of the CYP21A2 gene. As a complement to hormonal measurements, mutation analysis of CYP21A2 is an important tool in the diagnosis of steroid 21-hydroxylase deficiency. Contemporary mutation-detection protocols based on the polymerase chain reaction often depend on the assumption that no more than one CYP21A2 gene is present on each chromosome 6. We describe three haplotypes with two CYP21A2 genes on the same chromosome, with defects typical of salt-losing steroid 21-hydroxylase deficiency in one of those genes, but not necessarily in the other. The frequency of these haplotypes in the general population is 6/365 (1.6%), so they are no less common than other haplotypes that indeed carry steroid 21-hydroxylase deficiency. Chromosomes that carry two CYP21A2 genes therefore represent a significant pitfall in the molecular diagnosis of steroid 21-hydroxylase deficiency. We recommend that, whenever CYP21A2 mutation analysis of an individual who is not a known carrier of steroid 21-hydroxylase deficiency is performed, the overall structure of the CYP21/ C4 region (the RCCX area) is determined by haplotyping to avoid erroneous assignment of carrier status.

Steroid 21-hydroxylase deficiency is caused by defectiveness of the CYP21 gene. Such defects have presumably originated from interactions with the nearby CYP21P pseudogene during evolution. We studied these mechanisms by comparing the genetic variability of CYP21, CYP21P, and CYP21P/CYP21 hybrids (resulting from large-scale rearrangements) at eight mutation sites in a group of Dutch steroid 21-hydroxylase deficiency patients, their family members, and controls. The most common CYP21 defect in patients with salt-losing steroid 21-hydroxylase deficiency was a splice junction mutation in intron 2. The most common defect in the simple virilising form of the disease was ile72 → asn. CYP21P showed considerable sequence variation in its central and 3' sections; the 5' section was constant. A single nucleotide (T) insert in exon 7 was found in all CYP21P genes. During the course of evolution, this was probably the third defect introduced into CYP21P after the splice junction mutation in intron 2 and the 8 bp deletion in exon 3. Gene conversions introducing CYP21-like sequences contribute to CYP21P variability. Such an event has occurred de novo in one family. A comparison of CYP21 and CYP21P mutations on the same chromosome shows that at least some of the small-scale gene conversions that supposedly transfer defects to CYP21 involve interaction between homologous chromosomes. The majority of the putative CYP21P-CYP21 transitions in hybrid genes appears to occur in a distinct zone that lies 5' of nucleotide 2108, which is further downstream than previously hypothesised. The other transitions lie upstream of nucleotide 999. Apparent 'large-scale' CYP21-CYP21P gene conversions lead to hybrid genes that are very similar to those found in CYP21 deletions, so these haplotypes have probably resulted from a meiotic double unequal crossover.

Steroid 21-hydroxylase deficiency is caused by a defect in the CYP21A2 gene. CYP21A2, the adjacent complement C4 gene and parts of the flanking genes RP1 and TNXB constitute a tandemly duplicated arrangement in the central (class III) region of the major histocompatibility complex. The typical number of repeats of the CYP21/C4 region is two, with one repeat carrying CYP21A2 and the other carrying the highly homologous pseudogene CYP21A1P. By comparison with this standard, three categories of CYP21A2 defects have traditionally been distinguished: CYP21A2 deletions, large-scale gene conversions of CYP21A2 into a structure similar to CYP21A1P, and smaller mutations in CYP21A2 (also derived from CYP21A1P, by means of small-scale gene conversions). The genetic mechanisms suggested by these designations have originally been inferred from the layout of the haplotypes involved and were later confirmed by observation of deletions and small mutations, but not large-scale conversions, as de novo events. Apparent large-scale conversions account for the defect in 9 out of 77 chromosomes in our patient group. We here demonstrate that 4 out of these 9 'conversions' extend into the flanking TNXB gene, which encodes tenascin-X. This implies that approximately 1 in every 10 steroid 21-hydroxylase deficiency patients is a carrier of tenascin-X deficiency, which is associated with a recessive form of the Ehlers-Danlos syndrome. Currently available data on the structure of 'deletion' and 'large-scale conversion' chromosomes strongly suggests that both are the result of the same mechanism, namely unequal meiotic crossover. Since it is unlikely that the term 'large-scale gene conversion' describes a mechanism that actually occurs between the CYP21A2 and CYP21A1P genes, we propose the discontinuation of that terminology.

No abstract