I got an email from Greg Cochran last night, with the full text of this article: There’s Something Curious About Paternal Age Effects
There seems to be a good connection between paternal age and short-limbed dwarfism. A simillar pattern can be seen in Apert’s syndrome. Both these disorders seem to be present in the population at higher levels than the mutation rate would indicate. The explanation? Selection for mutant spermatogonia.
It now seems likely that there are three main classes of gene mutations causing genetic disorders: (i) nucleotide substitutions scattered along the gene, usually with substantial sex and age effects; (ii) small insertions and deletions, mainly deletions, with no age effect and a slight maternal excess; and (iii) hot-spots occurring almost exclusively in males and rising steeply with age. Three genes–fibroblast growth factor receptor 3 (FGFR3, mutated in achondroplasia), FGFR2 (mutated in Apert’s syndrome), and RET (mutated in multiple endocrine neoplasia)–are examples of the hot-spot class. In this class, genes carry mutations that are clustered at just one or two nucleotide sites.
Update from Razib: I have addressed ziel’s #2 complaint (don’t know about the first ;). See text below….
As early as 1912, Wilhelm Weinberg (1) reported that children with dominant achondroplasia (short-limbed dwarfism) born to normal parents were usually among the last-born children in the family. With astonishing insight, he suggested that this finding argued for a genetic mutation as the cause of sporadic achondroplasia. A deeper understanding had to await the work of Penrose, who in 1955 showed that the effect observed by Weinberg was due to paternal age, not maternal age or birth order (2). This, of course, implied a much greater mutation rate among males than females. Haldane demonstrated just such a disparity in mutation rates between male and female gametes in the X-linked disorder hemophilia (3). Since then, the sex-difference and paternal-age effect have been confirmed for several X-linked recessive and autosomal-dominant diseases (4, 5). Missing to date has been an analysis of mutations in spermatozoa, but new techniques have finally made this possible. The results are shocking. First, a recently published analysis of sperm from men of different ages reported only a slight increase in mutant sperm with paternal age, much less than would be predicted by the clinical data (6). Now, on page 643 of this issue, Goriely, Wilkie, and colleagues (7) report their analysis in men of different ages of sperm carrying the mutation that causes Apert’s syndrome, another classic disease where the clinical data predict a large sex and paternal-age effect. They argue that the mutation rate for this disorder is low, and that the apparent high rate is because mutant spermatogonia are positively selected before the start of meiosis (the two cell divisions that give rise to sperm). Very unorthodox.
It now seems likely that there are three main classes of gene mutations causing genetic disorders: (i) nucleotide substitutions scattered along the gene, usually with substantial sex and age effects; (ii) small insertions and deletions, mainly deletions, with no age effect and a slight maternal excess; and (iii) hot-spots occurring almost exclusively in males and rising steeply with age. Three genes–fibroblast growth factor receptor 3 (FGFR3, mutated in achondroplasia), FGFR2 (mutated in Apert’s syndrome), and RET (mutated in multiple endocrine neoplasia)–are examples of the hot-spot class. In this class, genes carry mutations that are clustered at just one or two nucleotide sites.
So far, only hot-spot mutations have been amenable to direct analysis in sperm. In contrast to the analysis of FGFR3 mutations in achondroplasia (6), the data for FGFR2 mutations obtained by Goriely et al. (7) agree with clinical observations on the paternal-age effect. These authors used a sensitive technique called pyrosequencing to examine the FGFR2 mutation rate in sperm. They used a restriction enzyme that cuts near the site of the mutation in FGFR2 (nucleotide 755, normally a cytosine) and that recognizes only the normal, not the mutant, sequence. They also cleverly exploited a single-nucleotide polymorphism (SNP), dimorphic for bases A and G, about 100 nucleotides upstream of the mutation. Mutant sperm from men heterozygous for this SNP showed the expected 1:1 ratio of A and G, but the variance among men was enormous. This is consistent with premeiotic selection, because a selected mutant will carry along whichever SNP marker it happens to be linked to, thus broadening the distribution of SNP ratios. This finding, together with the enormous number of mutations at two hot-spots in FGFR2, and the fact that these are gain-of-function mutations, argues for premeiotic selection of mutant spermatogonia (7). As long ago as 1996, Wilkie suggested that mutations in sperm might be selected premeiotically (8).
Traditionalists, such as myself, are reluctant to postulate strong selection of mutant spermatogonia. Are there alternative explanations? In a 30-year-old man, some 90% of the spermatogonial divisions occur during the stem-cell phase of division, where the pattern is linear. This is followed by four exponential divisions before meiosis begins (9). By producing 16 exact copies of a mutation that occurred during the stem-cell period, this would introduce a correlation between the SNP markers, and thereby an enhancement of the variance in SNP ratio among sperm donors. With a number of exponential divisions, one would expect a Luria-Delbrück jackpot effect, that is, a tremendous difference in the mutant frequency in different men (10). A full analysis, taking mutant sperm number into account, would be complicated, yet four divisions hardly seem sufficient to explain the observations of Goriely and co-workers.
The authors offer additional evidence for selection of mutant spermatogonia. In the FGFR2 gene, the transversion, CrarrowG, is unexpectedly more common than the transition, CrarrowT. They argue that the CrarrowG mutations are rarer, but are more strongly selected. Another argument for selection is that the variance in SNP ratios is greater for the mutation with the higher frequency, an unexpected result in the absence of selection. Surprising hypotheses call for unusually strong evidence. The evidence that Goriely et al. present for the positive selection of a deleterious mutation in the testis, though indirect, is indeed strong.
Clearly, something is peculiar. If it isn’t spermatogonial selection that causes these curious effects, then what does? What causes the discrepancy in results between the FGFR2 and FGFR3 studies? Measuring the mutation frequency in sperm requires very sensitive techniques, so there may be technical reasons for the discrepancy. The third hot-spot locus, RET (5), contains genes that are important for spermatogonial function (11), which may offer insights into how selection operates. Is it possible that spermatogonial stem cells do not follow the strict linear kinetics that are usually assumed for all stem cells, permitting them more stochastic variation? Stem cells in some somatic tissues have the strange and wonderful skill of directing mutant genes into cells destined to die, while maintaining normal genes in the stem-cell lineage (12). Could spermatogonial stem cells indulge in such unorthodox behavior? Happily, hot-spots such as those in FGFR2, FGFR3, and RET are amenable to further research, including linked SNPs, so we should see some answers soon. Further data on Apert’s syndrome (13) are forthcoming and may help to resolve some of the conflicting data regarding the paternal age effect and premeiotic selection of mutations in achondroplasia and Apert’s syndrome.
In previous examples of premeiotic selection, germinal and somatic selection act in the same direction (14). That Goriely and colleague
s find favorable selection of FGFR2 mutations in the germ line, despite the fact that they cause a devastating disease, is indeed surprising. But that’s how the data look.
Posted by Thrasymachus at 09:36 AM