Expansions and contractions of the FMR1 CGG repeat in 5,508 transmissions of normal, intermediate, and premutation alleles

Abstract Instability of the FMR1 repeat, commonly observed in transmissions of premutation alleles (55–200 repeats), is influenced by the size of the repeat, its internal structure and the sex of the transmitting parent. We assessed these three factors in unstable transmissions of 14/3,335 normal (~5 to 44 repeats), 54/293 intermediate (45–54 repeats), and 1561/1,880 premutation alleles. While most unstable transmissions led to expansions, contractions to smaller repeats were observed in all size classes. For normal alleles, instability was more frequent in paternal transmissions and in alleles with long 3′ uninterrupted repeat lengths. For premutation alleles, contractions also occurred more often in paternal than maternal transmissions and the frequency of paternal contractions increased linearly with repeat size. All paternal premutation allele contractions were transmitted as premutation alleles, but maternal premutation allele contractions were transmitted as premutation, intermediate, or normal alleles. The eight losses of AGG interruptions in the FMR1 repeat occurred exclusively in contractions of maternal premutation alleles. We propose a refined model of FMR1 repeat progression from normal to premutation size and suggest that most normal alleles without AGG interruptions are derived from contractions of maternal premutation alleles.


| INTRODUCTION
FMR1 is notable for instability of the cytosine guanine guanine (CGG) repeat in its 5 0 untranslated region. This repeat can expand to more than 200 units (full mutation) which causes intellectual disabilities in males and some females by silencing the gene (Oberle et al., 1991;Verkerk et al., 1991;Yu et al., 1991). The repeat, which is highly polymorphic, is classified into four categories (Maddalena et al., 2001). Normal alleles (~5 to 44 repeats) are passed stably from parent to child with rare changes in repeat size. Intermediate alleles (45-54 repeats) occasionally undergo small changes in repeat size in some families during transmission whereas premutation alleles (55-200 repeats) are often highly unstable with large changes in repeat size from one generation to the next, which may result in full mutation expansions and the fragile X syndrome (FXS; OMIM 300624; Nolin et al., 2003). These categories correspond to the N (normal), S (high end normal, predisposed), Z (premutation), and L (full mutation) alleles in the population genetic model of the mutational process of the fragile X repeat (Morris et al., 1995;Morton & Macpherson, 1992).
Previous studies have identified three factors that contribute to repeat instability in FMR1 premutation alleles: the sex of the transmitting parent, repeat size, and the presence of adenine guanine guanine (AGG) interruptions within the repeat. The first of these, parental sex, was recognized long before the gene was identified in 1991 (Sherman et al., 1985;Sherman, Morton, Jacobs, & Turner, 1984). Specifically, expansion to a full mutation occurs in maternal transmissions; virtually all premutation alleles from males are passed to daughters as premutation alleles although two rare examples of full mutation transmissions from fathers have been reported (Alvarez-Mora et al., 2017;Zeesman et al., 2004). While males with full mutations generally do not have offspring, several reports and studies in sperm indicate the daughters will inherit premutations, not full mutations from their fathers (Reyniers et al., 1993;Willems et al., 1992). Females with full mutations have 50% risk of transmitting full mutations to their offspring who inherit an even greater repeat tract length (Nolin et al., 2003;Nolin, Ding, Houck, Brown, & Dobkin, 2008). The second factor, increased repeat size, increases the likelihood of expansion of a maternal premutation allele to a full mutation Nolin et al., 2003). For paternal transmissions, the effect of repeat size has not been as well studied since the alleles are virtually always transmitted to the daughters as premutation alleles (Nolin et al., 1996;Sherman et al., 1984;Sherman et al., 1985). The third factor is the presence or absence of AGG triplets within the FMR1 CGG repeat. Studies of primates and genetically distinct human populations indicate that the conserved repeat structure includes CGG repeats interspersed with regular AGG repeats (Eichler et al., 1995;. In humans, the most common repeat structure of normal alleles includes an AGG as either the 10th or the 11th triplet and a second AGG after a stretch of nine more CGGs (Eichler et al., 1994;Gunter et al., 1998;Hirst, Grewal, & Davies, 1994;Kunst et al., 1996;Kunst & Warren, 1994;Snow, Tester, Kruckeberg, Schaid, & Thibodeau, 1994). Premutation alleles, in contrast, are less likely to contain AGGs and have long stretches of uninterrupted CGGs at their 3 0 end. In 1994, Eichler et al. hypothe-sized that long tracts of uninterrupted CGGs in premutation alleles contribute to repeat instability and that tracts with more than~34 uninterrupted CGGs can contribute to instability in smaller alleles as well. Recent studies (Nolin et al., 2013;Nolin et al., 2015;Yrigollen et al., 2012) have demonstrated that premutation alleles with no AGGs are at risk for expansion to full mutations in the next generation while alleles that include AGG interruptions are associated with greater intergenerational stability of the repeat.
In this study we surveyed more than 5,000 transmissions of normal, intermediate, and premutation alleles to examine the relationship of the sex of the transmitting parent, repeat size and pattern of AGG interruptions with allele instability. While repeat expansions and contractions were observed in all three size classes, repeat instability differed between maternal and paternal transmissions in each class.
The distribution of different AGG configurations in the three size classes and in unstably transmitted alleles sheds light on the progression of FMR1 alleles from normal to premutation repeat size over generations.

| FMR1 repeat structure in the general population
We first examined the repeat structure of 5,623 normal alleles, 406 intermediate alleles and 1,574 premutation alleles. The six most common normal alleles comprising 67.6% of alleles in the general population are shown in Table 1. Alleles with 30 repeats and the structure (CGG) 10 AGG(CGG) 9 AGG(CGG) 9 (represented as 10A9A9) were the most frequent (33.9%) followed by 29 repeat alleles (17.4%) with the structure 9A9A9. Eighty-five percent of normal alleles had nine or ten uninterrupted repeats at the 3 0 end while 12.4% (699/5,623) had more Second, as the repeat size lengthened, the proportion of alleles with an AGG as the 10th triplet increased while alleles with an AGG as the 11th triplet decreased. Among normal alleles, 32.6% (1,833/5,623) had an AGG as the 10th triplet and 52.5% (2,951/5,623) as the 11th triplet.
In contrast, 75.6% (307/406) of intermediate alleles had an AGG as the 10th triplet and 11.3% (46/406) as the 11th triplet. For premutation alleles, 59.7% (939/1,574) had an AGG at the 10th triplet and 9.2% (145/1,574) at the 11th triplet (χ 2 w 2 df; p < .0001).   (Table 5). Two mothers carrying 79 repeat alleles had two transmissions with contractions, and a third mother with 79 repeat alleles had contractions in two of three children that inherited the expanded allele. A fourth mother with an 83 repeat allele had three transmissions with contractions and a fourth transmission that expanded. In each case, the contractions resulted in smaller premutations that were similar in size within each family and with no AGGs lost. We calculated the likelihood of observing multiple contractions from these four mothers, assuming that each contraction within a family was an independent event. We estimated the probability of a contraction from women in the sample with 65-94 repeat alleles who had only one transmission and that underwent a contraction (i.e., p = 18/354 = .051; 95% confidence interval = 0.028-0.074). The likelihood estimates of the four observations are outlined in Table 5 and are based on a binomial distribution for contractions. This analysis indicated that the number of contractions observed in these four mothers would not be expected to occur in this study population if contractions were independent events.

| AGG loss on transmission
The number of AGG interruptions in intermediate and premutation repeats was determined for transmissions of 1,450 maternal and 387 paternal alleles ( Table 4). The loss of AGG interruptions is summarized in Table 6 for the 32 maternal and 68 paternal premutation alleles with AGGs that contracted during transmission. Eight maternal alleles lost AGGs in contractions whereas no AGG loss was observed in paternal contractions, a highly significant difference (two-tailed Fisher exact test p = .0002). Seven maternal alleles lost one AGG and an eighth lost two AGGs (Table 7)

| DISCUSSION
In this study of more than 5,000 transmissions of normal, intermediate, and premutation alleles, we observed expansions and contractions in all three size classes. This instability was strongly influenced by the sex of the transmitting parent and by the repeat structurenumber of repeats and location of the AGG interruptions-in the parental allele. The results of this survey suggest that instability is intrinsic to all repeat alleles and the frequency of these events is a continuous function influenced by repeat structure and parental sex.
Expansions of maternal premutation alleles on transmission have been well documented in various studies since the FMR1 gene was identified in 1991 Nolin et al., 2003;Snow et al., 1994). While maternal contractions were infrequent, loss of an AGG interruption in these contractions was common and occurred in 25.0% (8/32) of maternal alleles with AGGs. This is an unexpectedly high frequency particularly considering that 68 contractions of paternal premutation alleles were not associated with any AGG loss in transmission. The longest pure repeat associated with AGG loss was 118 CGGs. The mechanism of maternal contractions appears to be complex as a maternal allele with 9A9A42 was reduced to 45 CGGs and another maternal allele with 9A9A75 was transmitted to the offspring as 9A24. We observed no AGG loss in expansions or contractions of paternal alleles although there is one report of a paternal 52 repeat allele with two AGGs that was transmitted to a daughter as an expanded 56 repeat allele with no AGGs (Fernandez-Carvajal et al., 2009). Our study indicates that changes in AGG interruptions are rare aside from infrequent loss of AGG interruptions in transmission from parent to child. Thus, it is likely that some full mutation expansions include AGG interruptions.

In contrast to the general stability of AGG interruptions in the FMR1
repeat, intergenerational variations in location of repeat interruptions in myotonic dystrophy type 1 are not uncommon (Musova et al., 2009).
The importance of AGG interruptions in maintaining FMR1 repeat stability has been well documented in studies of human transmissions (Nolin et al., 2013;Nolin et al., 2015;Yrigollen et al., 2012), but has not shed light on the mechanisms. Studies in yeast (Rolfsmeier & Lahue, 2000) have shown that the presence of AGG interruptions prevent the formation of hairpin intermediaries. Other studies (Jarem, Huckaby, & Delaney, 2010;Pearson et al., 1998)   Numbers represent the number of CGG triplets; A represents an AGG interruption. b In this family the grandfather carried a 29 repeat allele and the grandmother a 55 repeat allele. The mother was mosaic for 16, 29, and 55 repeat alleles and the grandson inherited a 16 repeat allele.
observations suggest that alleles with an AGG at position 10 may be more unstable than those with an AGG at position 11. Other investigators have also observed an excess of AGGs at position 10 associated with intermediate and premutation alleles (Eichler et al., 1995;Gunter et al., 1998;Kunst & Warren, 1994;Murray et al., 1997;Snow et al., 1994). with the higher mutation rates for sperm (Crow, 2000). Males and females with full mutations also have very different risks for transmitting full mutation alleles to their offspring. Females with full mutations have 50% risk with each pregnancy of passing full mutation alleles to the next generation that inherit larger full mutation alleles than their mothers (Nolin et al., 2008). Most males with full mutations do not have offspring, but rare reports of such events indicate that the daughters inherited premutation alleles from their fathers (Willems et al., 1992). Furthermore, analysis of sperm from full mutation males revealed only premutation alleles (Reyniers et al., 1993) confirming the instability of full mutation repeats in the male germline. The different outcomes of FMR1 full mutation allele transmissions from males and females result from differences in gametogenesis. In both sexes, the gametes segregate as primordial germ cells between days 5 and 12 after fertilization. With sexual maturity in males, the germ cells undergo numerous mitotic divisions with perhaps as many as 549 divisions (Vogel & Rathenberg, 1975) leading to contractions of the long repeat tracts as a consequence of repair associated with genome maintenance and DNA replication before entering into meiosis (Pearson, 2003 Our study supports and refines the model of fragile X population genetics originally proposed by Morton and Macpherson (1992), Morris et al. (1995) and modified by  who suggested there are two pathways for expansion of FMR1 alleles. Our analysis of 5,508 transmissions suggests that the generation of unstable premutation alleles and their expansion to a full mutation is a multi-step process that often includes contractions of maternal premutation alleles to smaller alleles without AGGs. We propose a refinement of the two-pathway model for repeat expansion (Figure 2). The first path is slow with a gradual accumulation of CGG repeats in normal alleles containing AGG interruptions. Small increases are more likely to occur in paternal than maternal transmissions of normal and intermediate alleles. Alleles with an AGG at the 10th position within the repeat are also more likely to increase in size than those with an AGG at position 11. Rare increases of a cassette of 10 repeats from 29 to 39 repeats with the final structure (CGG) 9 AGG(CGG) 9 AGG(CGG) 9 AGG(CGG) 9 in a single transmission   Small premutation alleles without AGGs then have a high likelihood to expand rapidly to full mutations within a generation or two. These findings suggest contractions of maternal premutation alleles are a major mechanism for generating new alleles without AGG interruptions that are at high risk for eventual expansion to full mutation alleles.
F I G U R E 2 Model with two pathways for fragile X repeat expansion. The black arrows represent transmissions from males and the white arrows transmissions from females. The curved white arrows represent loss of AGGs in contractions of maternal premutation alleles. The width of the arrows indicates greater or less instability on transmission 4 | MATERIALS AND METHODS

| Ethical policies and ethical considerations
The samples used in the study were discarded clinical specimens.

| Subjects
To examine the repeat structure distribution of normal alleles in the general population, alleles from one person in each pedigree from our database as well as partners marrying into a family were included for a random sampling of alleles. Forty-six percent of the ethnicities were known: 67% Caucasian, 13% Hispanic, 8% African-American, 7% Asian, 3% mixed race, and 2% other. DNA was isolated from leukocytes or for prenatal samples from chorionic villi or amniocytes.
The 2020

| PCR protocols
The FMR1 repeat sizing was performed by PCR analysis using

ACKNOWLEDGMENTS
We thank the families whose support made this work possible. We also thank Michael Flory for assistance in statistical analysis.