AZFc Deletions and Spermatogenic Failure: A Population-Based Survey of 20,000 Y Chromosomes

Abstract

Deletions involving the Y chromosome’s AZFc region are the most common known genetic cause of severe spermatogenic failure (SSF). Six recurrent interstitial deletions affecting the region have been reported, but their population genetics are largely unexplored. We assessed the deletions’ prevalence in 20,884 men in five populations and found four of the six deletions (presented here in descending order of prevalence): gr/gr, b2/b3, b1/b3, and b2/b4. One of every 27 men carried one of these four deletions. The 1.6 Mb gr/gr deletion, found in one of every 41 men, almost doubles the risk of SSF and accounts for ∼2% of SSF, although <2% of men with the deletion are affected. The 1.8 Mb b2/b3 deletion, found in one of every 90 men, does not appear to be a risk factor for SSF. The 1.6 Mb b1/b3 deletion, found in one of every 994 men, appears to increase the risk of SSF by a factor of 2.5, although <2% of men with the deletion are affected, and it accounts for only 0.15% of SSF. The 3.5 Mb b2/b4 deletion, found in one of every 2,320 men, increases the risk of SSF 145 times and accounts for ∼6% of SSF; the observed prevalence should approximate the rate at which the deletion arises anew in each generation. We conclude that a single rare variant of major effect (the b2/b4 deletion) and a single common variant of modest effect (the gr/gr deletion) are largely responsible for the AZFc region’s contribution to SSF in the population.

Main Text

The 4.5 Mb AZFc region (MIM 41500; Figure 1) of the Y chromosome is remarkable for its structural complexity (including the largest known palindrome in any genome sequenced to date), its propensity to suffer massive deletion, and the contribution of those deletions to spermatogenic failure while apparently sparing all other body functions.^1–7 This last characteristic suggests a regional functional specialization that is uncommon in eukaryotic genomes. In fact, deletions affecting the AZFc region are the most common known genetic cause of severe spermatogenic failure (SSF), defined by a sperm count of less than five million per milliliter of semen in the absence of physical obstruction.

Figure 1.

The AZFc Region of the Y Chromosome and Deletions Affecting It

(A) Overview of the Y chromosome, including the male-specific region (MSY). The AZFc region is located in the euchromatic portion of the long arm. STS sY14, located in the sex-determining gene SRY (MIM 480000), served as a positive control for the presence of detectable Y chromosome DNA. The four deletions observed in the present study are schematized here and in the panels below.

(B) Sequence organization of the AZFc region and STSs used for detecting and categorizing deletions involving AZFc. Colored arrows indicate large, nearly identical segmental duplications, termed “amplicons” in this context. Arrows of the same color are >99.82% identical to each other.¹Table S1 provides assay details for STSs. We used STSs sY1191 and sY1291, marked by asterisks, in stage 1 of the screen. STS sY254 detects multiple sites in the AZFc region. A green “bow” indicates amplicons involved in ectopic crossing over in the gr/gr deletion and regions and STSs affected; the gr/gr deletion results in loss of only sY1291 and sY1189.

(C) The b2/b3 deletion can arise on the two inverted variants of the AZFc region shown, but not on the reference organization;² the b2/b3 deletion arising on either of the two inverted variants results in the same organization of amplicons and results in the loss of only sY1192 and sY1191.

(D) Ectopic crossing over and the extents of the b1/b3 and b2/b4 deletions. The b1/b3 deletion (upper bow) results in loss of all copies of STSs sY1192, sY1191, sY1291, and sY1189 but spares two copies of sY254, as well as sY142 and sY1201. The b2/b4 deletion (lower bow) has a similar pattern but removes all copies of sY254.

(E) Patterns of PCR positives and negatives used for identifying four types of recurrent deletion. Black indicates the presence of STS PCR product, and “−” indicates absence.

The complex structure of the AZFc region, which is composed of massive, near-perfect amplicons, posed special challenges for sequencing the region and thereby characterizing deletions that affect it. To achieve the required level of accuracy and completeness, we previously developed a method (SHIMS, or single-haplotype iterative mapping and sequencing) to assemble its reference DNA sequence.^1,8 This AZFc-inspired strategy was subsequently employed for assembling the DNA sequences of the male-specific regions of the human, chimpanzee, and rhesus Y chromosomes and the chicken Z chromosome.^9–12 Sequencing of the AZFc region enabled our laboratory to identify and molecularly define six recurrent interstitial deletions that remove large portions of the region.^1–3,5,6 Most of these recurrent deletions are the consequence of nonallelic homologous recombination between near-identical amplicons found within and near the AZFc region (Figure 1).^1,2,5–7

Although numerous studies have examined the relationship between SSF and deletions affecting the AZFc region,^{3–6,13–17} the frequencies of the various deletions in any general population, and hence their precise population contributions to SSF, have remained unknown. In this regard, the chief limitation of previous studies is that they were focused on men who were studied because of clinically ascertained infertility. In addition, sample sizes have been small: almost always <1,000 men.

With the goal of understanding the population frequencies of the six recurrent, interstitial deletions affecting the AZFc region, we undertook a screen for these six deletions in 20,884 men from five populations: India (city of Pune), Poland (city of Katowice), Tunisia (city of Monastir), the United States (multiple sites), and Vietnam (cities of Hanoi and Hue). We used anonymized DNA samples collected by Genomics Collaborative for disease studies. The DNA donors included healthy controls and men with osteoarthritis, rheumatoid arthritis, asthma, hypertension, coronary artery disease, myocardial infarction, hyperlipidemia, stroke, type 2 diabetes, or osteoporosis. Table 1 shows the numbers of DNA samples screened and their geographic origins. This study was approved by the institutional review board at the Massachusetts Institute of Technology, and proper informed consent was obtained from participants.

Table 1.

Proportions of Deletions by Study Population

	Number of Samples
	India	Poland	Tunisia	United States	Vietnam	All Populations
Total samples	404	4,671	578	15,124	107	20,884
No deletion	373 (92%)	4,445 (95%)	533 (92%)	14,668 (97%)	90 (84%)	20,109 (96%)
Any deletion	31 (7.7%)	226 (4.8%)	45 (7.8%)	456 (3%)	17 (16%)	775 (3.7%)
gr/gr	27 (6.7%)	115 (2.5%)	41 (7.1%)	312 (2.1%)	16 (15%)	511 (2.4%)
b2/b3	2 (0.5%)	105 (2.2%)	3 (0.52%)	121 (0.8%)	1 (0.93%)	232 (1.1%)
b1/b3	2 (0.5%)	2 (0.043%)	1 (0.17%)	16 (0.11%)	0	21 (0.10%)
b2/b4a	0	2 (0.043%)	0	7 (0.046%)	0	9 (0.043%)
P5/P1	0	0	0	0	0	0
P4/P1	0	0	0	0	0	0
Unusualb	0	2 (0.043%)	0	0	0	2 (0.0096%)

^aEnumerated in Table S3.

^bDeletions not falling into one of the categories of recurrent deletions; see Table S2 for details.

We screened for interstitial deletions involving AZFc in two stages (Figure 2). In stage 1, we tested for the presence or absence of the sequence tagged sites (STSs) sY1191 and sY1291, one or both of which are deleted in all known recurrent deletions involving AZFc (Figure 1 and Table S1, available online). In stage 2, we further tested the Y chromosomes that appeared to lack sY1191 and/or sY1291. In this stage, we confirmed the results for sY1191 and sY1291 both by testing these STSs on new aliquots of DNA and by testing different STSs (sY1192 and sY1189) that detect the same sites (Figure 2). In this stage, we also classified deletions on the basis of the patterns of positive and negative results of STSs at additional sites: sY14, sY142, sY254, and sY1201 (Figures 1 and 2 and Table S1).

Figure 2.

Workflow for Detecting and Categorizing Deletions Involving AZFc

In stage 1, we tested 21,261 DNA samples for the presence of STSs sY1191 and sY1291; 19,168 samples were scored positive for both STSs. The 2,093 samples that were scored negative for one or both STSs were subject to further testing in stage 2 with the use of the STSs shown. Also, sY1191 and sY1291, marked by asterisks, were retested. This allowed us to categorize the deletions as shown. As expected, stage 2 retesting revealed that some STSs deemed absent in stage 1 were in fact present. For example, 492 samples that were scored negative for sY1191 in stage 1 were scored positive in stage 2. This was because we set a liberal threshold in stage 1 for scoring sY1191 absence to avoid missing deletions. We also determined in stage 2 that 377 DNA samples were not usable; we could not reliably amplify positive control STSs in these samples. Thus, the total number of samples assayed was 20,884.

When the patterns of positive and negative STS results did not correspond to one of the recurrent deletion classes (Figure 1E), we repeated the STS assays to confirm results. After this repeated testing, we determined that two samples bore deletions different from any of the six previously described recurrent interstitial deletions (Figure 2 and Table S2). We used additional STSs to further characterize these deletions (Tables S1 and S2), one of which most likely represents an isodicentric Y chromosome.¹⁸ To our knowledge, the other deletion has not been reported previously.

As summarized in Table 1, we detected four of the six previously described interstitial deletions in one or more of the five study populations. Among the total 20,884 men studied, 773 men (or one in every 27 men tested) displayed one of these four deletions. We found the gr/gr deletion to be the most common (2.4% or 1/41 men; 95% confidence interval [CI] = 2.2%–2.7%); it was followed by the b2/b3 deletion (1.1% or 1/90 men; 95% CI = 1.0%–1.3%), the b1/b3 deletion (0.1% or 1/994 men; 95% CI = 0.064%–0.16%), and the b2/b4 deletion (0.043% or 1/2,320 men; 95% CI = 0.021%–0.085%; Table S3). The estimate for the prevalence of the b2/b4 deletion is statistically consistent with, but higher than, a previous estimate of 0.025%. (The prior estimate was based on the prevalence of the b2/b4 deletion among men with nonobstructive azoospermia [no spermatozoa in semen], as well as estimates of the prevalence of nonobstructive azoospermia¹). Notably, our survey of these populations did not identify any Y chromosome with either of the two largest recurrent interstitial deletions affecting the AZFc region—that is, the previously described P5/P1 and P4/P1 deletions.⁵ In our laboratory’s published studies of men with SSF, the number of P5/P1 or P4/P1 deletions was about one fourth of the number of b2/b4 deletions.⁵ The present study would have detected P5/P1 or P4/P1 deletions had they been present, but we found none in the 20,884 men tested (95% CI = 0%–0.014%). This is fewer than expected but is still statistically consistent with our published work.⁵

There are two previously reported instances in which the prevalence of an AZFc-region deletion varies strongly by population: (1) the high prevalence of the b2/b3 deletion around the Baltic Sea is due to the prevalence of haplogroup N1 chromosomes, all of which contain the b2/b3 deletion,^2,7 and (2) the high prevalence of the gr/gr deletion in Japan is due to the prevalence of haplogroup D2a chromosomes,⁶ all of which contain the gr/gr deletion. Motivated by these examples, we examined variation in the prevalence of each of the AZFc-region deletions across the five populations studied. We found that prevalence varies significantly for two of the AZFc-region deletions: gr/gr and b2/b3 (Table 1). Prevalence of the gr/gr deletion ranges from 2% (in the United States) to 15% (in Vietnam) (p < 10⁻²⁰ for the proportions of gr/gr-deleted versus non-gr/gr-deleted chromosomes by Fisher’s exact test, two-sided). Prevalence of the b2/b3 deletion ranges from 0.5% (in India) to 2.2% (in Poland) (p < 10⁻¹¹ for the proportions of b2/b3-deleted versus non-b2/b3-deleted chromosomes). By contrast, prevalences of the b1/b3 and b2/b4 deletions do not vary significantly across the five populations (p > 0.8 and p = 1 for the b1/b3 and b2/b4 deletions, respectively). The difference in prevalence of the b2/b3 deletion across the five populations appears to be largely due to differences in the prevalence of haplogroup N1 chromosomes: considering only non-N1 chromosomes, the prevalence of the b2/b3 deletion does not vary significantly across the five populations (p > 0.15; see Table S4).

As suggested by the examples of the b2/b3 deletion in haplogroup N1 and the gr/gr deletion in haplogroup D2a, variation in deletion prevalence across populations might be due to a combination of two factors: (1) variation in deletion prevalence across Y chromosome haplogroups and (2) the enrichment for these haplogroups in particular populations. Therefore, we investigated whether deletion prevalence varies by haplogroup within the two largest population samples, those from Poland and the United States. Using seven Y chromosome SNPs, we assigned deleted chromosomes and a subsample of nondeleted chromosomes to one of nine Y chromosome haplogroups (Table S5).¹⁹ Figure 3 summarizes the findings and analysis. We excluded haplogroup N1, in which essentially all chromosomes have a b2/b3 deletion, to allow us to focus on other differences in deletion prevalence across haplogroups. We found, for example, that haplogroup R1a is significantly enriched among gr/gr-deleted chromosomes in the Polish population and among b1/b3-deleted chromosomes in the United States population. In both the Polish and United States populations, non-N1 b2/b3-deleted chromosomes, as compared to nondeleted chromosomes, show significant differences in haplogroup distribution (Figure 3 and Table S6 provide statistical details). Thus, these populations include haplogroups (in addition to N1) that are enriched with the b2/b3 deletion. Of particular note is that two haplogroups—BC^∗xDEF and E—are enriched among b2/b3-deleted chromosomes in both Poland and the United States. The simplest explanation is that there exists in each of these haplogroups some yet-unidentified sub-branch that features the b2/b3 deletion—just as it is a universal feature of haplogroup N1.

Figure 3.

Proportions of Deleted and Undeleted Chromosomes by Y Haplogroup

(A) Proportions of chromosomes in each Y haplogroup within four deletion classes: undeleted (purple), gr/gr (green), b2/b3 (orange), and b1/b3 (blue). Haplogroups are defined in (B). The height of each bar indicates the proportion of Y chromosomes in that haplogroup within a particular deletion class. For example, the prevalence of gr/gr-deleted chromosomes across haplogroups in the United States sample mirrors that of nondeleted chromosomes, as shown by the fact that the heights of green columns (gr/gr-deleted chromosomes) are similar to the heights of purple columns (nondeleted chromosomes). By contrast, in the Polish sample, the column representing gr/gr-deleted chromosomes in haplogroup F^∗xK is lower than the column representing nondeleted chromosomes, but for haplogroup R1a, the gr/gr-deletion column is higher than the column representing nondeleted chromosomes. This indicates that gr/gr-deleted chromosomes are underrepresented in F^∗xK and overrepresented in R1a. As discussed in the text, branch N1 was excluded from the analysis. Triangles mark haplogroups that make major contributions to variation in prevalence across haplogroups on the basis of their standardized residuals.²⁰Table S6 provides statistical details, including counts.

(B) Genealogical tree of human Y chromosomes used in determining Y haplogroups for this study. Branch tips are labeled with haplogroup designations with the use of terminology from Karafet et al.¹⁹ SNPs shown along branches are summarized in Karafet et al.¹⁹ Three haplogroups are defined by a single variant and are monophyletic (E, N1, and R1a); remaining haplogroups are paraphyletic (Table S5). The SRY10831 polymorphism reflects an A>G mutation early in the history of extant Y chromosomes; its reversion in an M207-derived chromosome defines branch R1a.

The current results, together with previous findings, allow us to estimate several population-genetic and epidemiological parameters for three AZFc-region deletions that cause or predispose to SSF: gr/gr, b1/b3, and b2/b4 (in decreasing order of prevalence in the general population). We will discuss them in order of their relative contributions to SSF in the population: b2/b4, gr/gr, and b1/b3.

The 3.5 Mb b2/b4 deletion almost always causes SSF, and it is therefore extremely rare for men with the b2/b4 deletion to father children without medical assistance.^15,21,22 In addition, the available evidence indicates that the b2/b4 deletion does not otherwise impair viability or health. Accordingly, the prevalence of the b2/b4 deletion in the current study (one out of every 2,320 men; Table 2) should approximate the rate at which the deletion arises anew, by mutation, in each father-to-son transmission of a Y chromosome. We estimate that the b2/b4 deletion increases a man’s risk of SSF by a factor of 145 and that it accounts for about 5.9% of SSF cases in the population (Table 2). Given the present estimate of the prevalence of the b2/b4 deletion in the general population (9 deletions in 20,884 men tested) and an estimate of its prevalence among men with SSF (42/713),¹⁵ we can also estimate the prevalence of SSF in the population as (9/20,884) / (42/713) = 0.0073 (95% CI = 0.0034–0.013 by bootstrap resampling).

Table 2.

Population Genetic and Epidemiological Parameters for AZFc-Related Deletions that Cause or Might Predispose to SSF

	Deletion Type
	b2/b4	gr/gr	b1/b3
Prevalence in Population
Total number of men tested	20,884	19,113a	20,884
Number of men with deletion	9	427a	21
Percentage of men with deletion (95% CI)	0.043 (0.021–0.085)	2.2 (2.0–2.5)	0.10 (0.064–0.16)
Prevalence among Men with SSF
Total number of men tested	713	4,685	3,956
Number of men with deletion	42	194	10
Percentage of men with deletion (95% CI)	5.9 (4.3–7.9)	4.1 (3.6–4.8)	0.25 (0.13–0.48)
Source of data	Oates et al.15	new data and literature (Table S7)	new data and literature (Table S7)
Calculated Parameters
Relative risk of SSF (95% CI)	145 (85–310)	1.9 (1.6–2.2)	2.5 (1.2–4.6)
Percentage of deletion-bearing men who have SSF (bootstrap 95% CI)	100 (assumed 100)	1.4 (0.63–2.3)	1.8 (0.63–4.1)
Attributable risk percentage of SSFb (bootstrap 95% CI)	99 (99–100)	47 (39–54)	60 (17–78)
Population-attributable risk percentage of SSFc (bootstrap 95% CI)	5.9 (4.4–7.4)	2.0 (1.4–2.5)	0.15 (0.022–0.29)
μ (95% CI)	4.3 × 10−4 (= prevalence) (2.1 × 10−4 to 8.5 × 10−4)

We estimated parameters from prevalences in unselected populations and prevalences among men with SSF as discussed in the text and the Supplementary Note. The following abbreviations are used: CI, confidence interval; SSF, severe spermatogenic failure; and μ, mutation rate per father-to-son transmission of a Y chromosome.

^aFor gr/gr deletions, we considered only the Polish and United States populations, which best matched the bulk of the data in the literature on gr/gr-deletion prevalence among men with SSF (Table S7).

^bIn men with a given deletion, the percentage of SSF that is due to that deletion. Supplemental Data provide details of calculations.

^cThe percentage of SSF due to the given deletion in the population.

We now compare and contrast these b2/b4 parameters with those for the 1.6 Mb gr/gr deletion. As with the b2/b4 deletion, accumulated evidence from multiple studies indicates that the gr/gr deletion increases risk of SFF.^13,14,16,17 But here, the contrasts between the gr/gr and b2/b4 deletions begin. Using the prevalence results from the current study, together with previously published data (Table S7), we estimate that the gr/gr deletion increases a man’s risk of SSF by only a factor of 1.9 (Table 2) and that only about 1.4% of men with the gr/gr deletion are affected by SSF. Nonetheless, the high prevalence of the gr/gr deletion in the general population results in its accounting for about 2.0% of SFF cases (Table 2). We calculate that new gr/gr deletions arise at a rate of roughly 1.4 × 10⁻⁴ per father-to-son transmission of the Y chromosome (Supplementary Note).

We also performed parallel analyses of the 1.6 Mb b1/b3 deletion. In this case, we analyzed data from 15 published studies, along with our own unpublished data as detailed in Tables S7 and S8. Among a total of 3,956 men with SSF, there were ten with the b1/b3 deletion. Table 2 shows estimates of population-genetic and epidemiological parameters based on these data. The confidence intervals are wider than for the gr/gr deletion because the b1/b3 deletion is much rarer, but the analysis nevertheless indicates that the b1/b3 deletion increases a man’s risk of SSF by a factor of about 2.5, (p = 0.023 by Fisher’s exact test, two sided). Even so, only 1.8% of men with the b1/b3 deletion are affected by SSF, and the b1/b3 deletion accounts for only 0.15% of SSF (Table 2). The low prevalence of the b1/b3 deletion (0.1%, Table 2) appears to stem primarily from the relatively low rate at which it arises: 1.1 × 10⁻⁵ per father-to-son transmission of the Y chromosome (Supplementary Note).

In contrast to the b2/b4, gr/gr, and b1/b3 deletions, the b2/b3 deletion has not been shown to increase the risk of SSF above the population average in either published literature on populations of European ancestry or our own data.

In conclusion, the six previously described deletions that affect the AZFc region vary dramatically in prevalence in the general population—this prevalence ranges from undetectability of the P5/P1 and P4/P1 deletions in our sample of 20,884 men to a prevalence of 15% in the case of the gr/gr deletion in the Vietnamese population. On the basis of present and previous findings, we conclude that five of these six previously described deletions—including the b1/b3 deletion—increase a man’s risk of SSF. With regard to SSF and its occurrence in the population, we conclude that one rare variant of major effect (the b2/b4 deletion) and one common variant of modest effect (the gr/gr deletion) together account for about 8% of cases; these two deletions are largely responsible for the AZFc region’s contribution to SSF in the population.

On a broader scale, our findings raise important questions about the mutability of structurally complex regions on the X chromosome and autosomes. The high rates at which the gr/gr and b2/b4 deletions arise anew on the Y chromosome suggest that, aggregated across the entirety of the genome, large-scale deletions or amplifications might contribute substantially to the load of deleterious new mutations. Indeed, investigators have already reported that large-scale copy-number mutations often underlie intellectual disability, schizophrenia, developmental delay, and congenital anomalies,^23–27 consistent with the fact that these types of mutations contribute substantially to mutational load. The picture remains far from complete, however, partly because of the difficulty of comprehensively identifying, on a genome-wide scale, regions that are prone to massive structural change. In fact, regions that are rich in segmental duplications or structural polymorphism might be missing from, or misassembled in, the human reference sequence.^28–30 It will be important to generate more accurate reference sequence for such regions, which is possible with the use of approaches such as the SHIMS technique that we used to sequence the AZFc region.^1,8

Web Resources

The URL for data presented herein is as follows:

• Online Mendelian Inheritance in Man (OMIM), http://www.omim.org