Recombination between palindromes P5 and P1 on the human Y chromosome causes massive deletions and spermatogenic failure

Abstract

It is widely believed that at least three nonoverlapping regions of the human Y chromosome-AZFa, AZFb, and AZFc ("azoospermia factors" a, b, and c)-are essential for normal spermatogenesis. These intervals are defined by interstitial Y-chromosome deletions that impair or extinguish spermatogenesis. Deletion breakpoints, mechanisms, and lengths, as well as inventories of affected genes, have been elucidated for deletions of AZFa and of AZFc but not for deletions of AZFb or of AZFb plus AZFc. We studied three deletions of AZFb and eight deletions of AZFb plus AZFc, as assayed by the STSs defining these intervals. Guided by Y-chromosome sequence, we localized breakpoints precisely and were able to sequence nine of the deletion junctions. Homologous recombination can explain seven of these deletions but not the remaining two. This fact and our discovery of breakpoint hotspots suggest that factors in addition to homology underlie these deletions. The deletions previously thought to define AZFb were found to extend from palindrome P5 to the proximal arm of palindrome P1, 1.5 Mb within AZFc. Thus, they do not define a genomic region separate from AZFc. We also found that the deletions of AZFb plus AZFc, as assayed by standard STSs heretofore available, in fact extend from P5 to the distal arm of P1 and spare distal AZFc. Both classes of deletions are massive: P5/proximal-P1 deletions encompass up to 6.2 Mb and remove 32 genes and transcripts; P5/distal-P1 deletions encompass up to 7.7 Mb and remove 42 genes and transcripts. To our knowledge, these are the largest of all human interstitial deletions for which deletion junctions and complete intervening sequence are available. The restriction of the associated phenotype to spermatogenic failure indicates the remarkable functional specialization of the affected regions of the Y chromosome.

Figure 1

Deletions between P5 and P1 and between P4 and P1 in relation to the sequence of the Y chromosome. A, Triangular dot plot, encompassing P5, P4, and AZFc (which contains P1, P2, and part of P3). The baseline represents 8 Mb of Y-chromosome long-arm sequence. This triangular display avoids the redundant, artifactual symmetries that would appear in a square self-versus-self plot. Each dot within the plot represents 100 bp of identity between two parts of the Y chromosome in a window of 100 bp. Direct repeats appear as horizontal lines of dots, and inverted repeats appear as vertical lines. Palindromes appear as vertical lines that almost intersect the baseline. Five major palindromes are labeled “P1” through “P5” from right to left, and arrows along the baseline indicate the extents of their inverted-repeat arms. P1.1 and P1.2 are minipalindromes within P1 (see also fig. 6). The arms of P1.1 and P1.2 are too short (10 kb) to be visible at this scale. The “b2” and “b4” direct repeats that bound AZFc are shaded blue on the baseline. Sequences that are homologous between P5 and P1 are shaded orange and green on the baseline. Diagonal gray guide lines connect the P5-P1 homologous sequences on the baseline to the two areas within the plot that contain the corresponding dots. These areas are shaded orange and boxed. The prominent triangle of dots near the baseline and labeled “DYZ19” is a satellite repeat array. B, STSs used for low-resolution breakpoint mapping. Tick marks show STS positions; asterisks indicate new STSs. STSs used in the original definition of AZFb are in boldface. Results for sY1227 and sY1228 are difficult to represent at this scale; see table A1. C, Low-resolution plus/minus STS results and deletion breakpoint positions for 12 men with spermatogenic failure. At the left are the identifiers of the men studied, and to the right of each man’s identifier is a representation of those parts of his Y chromosome that were determined to be present or absent. Horizontal black bars indicate confirmed presence of Y-chromosome DNA, and minuses indicate confirmed absence of Y-chromosome DNA, as assayed by low-resolution STSs. White boxes represent STS positives that were disregarded because of cross-amplifying loci elsewhere and because of negative results for flanking STSs. Horizontal gray bars represent the intervals to which breakpoints were localized by low-resolution breakpoint mapping. (Where STSs fall within gray bars, their results were positive). Short red vertical lines indicate the locations of amplified breakpoint junctions for nine of the patients and, for AMC0111, the 10-kb interval to which high-resolution STS mapping localized the distal breakpoint. AZFc-deleted patient WHT3060 is shown for comparison. D, Genes with significant confirmed or predicted ORFs (see the “Electronic-Database Information” section). E, Spliced but apparently noncoding transcripts (see the “Electronic-Database Information” section).

Figure 2

FISH, showing DAZ clusters in an unaffected man and in a man with a P5/proximal-P1 (AZFb) deletion. A, Representative interphase nucleus from a fertile, male control individual hybridized with cosmid 18E8 (yellow), showing two dots (corresponding to two DAZ clusters, each containing two DAZ genes). B, Representative interphase nucleus from P5/proximal-P1–deleted patient WHT4396, showing one dot (corresponding to one DAZ cluster). C, Histogram of number of nuclei observed grouped by number of dots per nucleus.

Figure 3

P5/P1 deletion junctions, indicating homologous recombination as the cause of the deletion. A, Junction sequence in AMC0110 aligned with proximal and distal Y-chromosome reference sequence. Dots indicate base pairs identical to the junction sequence. B, Model of production of the deletion-junction sequence in AMC0110 by homologous recombination. The bases shown (C/T, A/G, and C/T) differ between the proximal and distal copies of the sequence and therefore define the location of the deletion breakpoint within the junction sequence. C, Junction sequence in WHT4396 aligned with proximal and distal Y-chromosome reference sequence. D, Junction sequence in WHT2943 aligned with proximal and distal Y-chromosome reference sequence. E, Junction sequence in WHT3516, WHT3642, WHT4426, and WHT4486 aligned with proximal and distal Y-chromosome reference sequence.

Figure 4

P5/distal-P1 deletion junction in WHT3410. A, Junction sequence aligned with proximal and distal Y-chromosome reference sequences. Ovals mark 3 bp that differ between WHT3410 and the Y-chromosome reference sequence. The white box marks the 3 bp that contain the deletion junction. B, Breakpoint locations within palindrome P5 and minipalindrome P1.1, and the resulting sequence organization in WHT3410. Orange and green shading is as in figures 1 and 6.

Figure 5

P4/distal-P1 deletion junction in WHT2825. Junction sequence is aligned with proximal and distal Y-chromosome reference sequence.

Figure 6

“High-magnification” dot plots of the two regions of high similarity between P5 and P1 (see fig. 1A). Each dot represents 50 bp of identity in a window of 50 bp. Orientation and orange and green shading is as in figure 1A. A, Central P5 versus P1.2 and the neighboring region of the proximal arm of P1. Locations of the P5/proximal-P1 deletions in AMC0110 and WHT4396 are indicated by red arrows. (The coordinate of a deletion in the plot is the location of the proximal breakpoint paired with the location of the distal breakpoint. Deletions that extend between direct repeats map to horizontal lines of dots.) B, Central P5 versus P1.1 and the neighboring region of the distal arm of P1. Locations of the P5/distal-P1 deletions in WHT2943, WHT3410, WHT3642, WHT3516, WHT4426, and WHT4486 are indicated. WHT3410’s deletion does not extend between homologous sequences and therefore does not map to a line of dots.

Figure 7

Previous model of recurrent, interstitial Y-chromosome deletions that cause infertility in men (Vogt et al. 1996) contrasted with current model.