Am. J. Hum. Genet. 54:575-585, 1994

Gonosomal Mosaicism in Patients: Involvement of Mitotic Events in (CTG)n Repeat Variation and Selection against Extreme Expansion in Sperm Gert Jansen,* Patrick Willems,* Marga Coerwinkel,* Willy Nillesen,t Hubert Smeets,t Lieve Vits,* Chris Howeler,1 Han Brunnert and Be Wieringa* Department of Cell Biology and Histology and tDepartment of Human Genetics, Faculty of Medical Sciences, University of Nijmegen, Nijmegen; $Department of Medical Genetics, University of Antwerp, Antwerp; and IDepartment of Neurology, Academic Hospital Maastricht, University of Maastricht, Maastricht

Summary Myotonic dystrophy (DM) is caused by abnormal expansion of a polymorphic (CTG). repeat, located in the DM protein kinase gene. We determined the (CTG). repeat lengths in a broad range of tissue DNAs from patients with mild, classical, or congenital manifestation of DM. Differences in the repeat length were seen in somatic tissues from single DM individuals and twins. Repeats appeared to expand to a similar extent in tissues originating from the same embryonal origin. In most male patients carrying intermediate- or small-sized expansions in blood, the repeat lengths covered a markedly wider range in sperm. In contrast, male patients with large allele expansions in blood (>700 CTGs) had similar or smaller repeats in sperm, when detectable. Sperm alleles with >1,000 CTGs were not seen. We conclude that DM patients can be considered gonosomal mosaics, i.e., combined somatic and germ-line tissue mosaics. Most remarkably, we observed multiple cases where the length distributions of intermediate- or small-sized alleles in fathers' sperm were significantly different from that in their offspring's blood. Our combined findings indicate that intergenerational length changes in the unstable CTG repeat are most likely to occur during early embryonic mitotic divisions in both somatic and germ-line tissue formation. Both the initial CTG length, the overall number of cell divisions involved in tissue formation, and perhaps a specific selection process in spermatogenesis may influence the dynamics of this process. A model explaining mitotic instability and sex-dependent segregation phenomena in DM manifestation is discussed.

Introduction 1991), myotonic dystrophy (DM) (Brook et al. 1992; Fu in simple-sequence motifs (Richards et al. 1992; Mahadevan et al. 1992), Huntington disease and Sutherland 1992) is a new principle in human mo- (HD) (Huntington's Disease Collaborative Research lecular genetics. It has now been demonstrated that Group 1993), spinocerabellar ataxia type 1 (Orr et al. simple-sequence-motif instability is causally involved in 1993), and fraxE (Knight et al. 1993). Though quite spinal and bulbar muscular atrophy (or Kennedy dis- different disorders, , DM, and HD ease; La Spada et al. 1991), fragile X syndrome (Fu et al. may have several mutational/mechanistic aspects in 1991; Kremer et al. 1991; Verkerk et al. 1991; Yu et al. common. All three diseases show atypical segregation patterns, referred to as the "Sherman paradox" in frag- ile X syndrome (Sherman et al. 1985) and as "anticipa- Received August 9, 1993; accepted for publication December 3, tion" in DM (H6weler et al. 1989) and HD (Ridley et al. 1993. 1988). In each of these disorders, founder chromo- Address for correspondence and reprints: Be Wieringa, Depart- somes are implicated (Conneally et al. 1989; Richards et ment of Cell Biology and Histology Faculty of Medical Sciences, al. 1992; Imbert et al. 1993), repeat length and disease University of Nijmegen, P.O. Box 9101, 6500 HB, Nijmegen, The severity are correlated (Yu et al. 1992; Harley et al. Netherlands. © 1994 by The American Society of Human Genetics. All rights reserved. 1993; Huntington's Disease Collaborative Research 0002-9297/94/5404-0001$02.OO Group 1993), and there is an almost-identical threshold 575 576 Jansen et al.

length for the trinucleotide motifs (approximately 35- To assess the occurrence of mitotic repeat-length varia- 40 trinucleotides) above which instability occurs (Brun- tions within DM patients, we analyzed a broad range of ner et al. 1992; Macpherson et al. 1992; Huntington's tissues from both single patients and twins with mild, Disease Collaborative Research Group 1993). Although average, and severe disease manifestation. To address this suggests that similar molecular mechanisms under- further the mechanism and developmental timing of the lie the unstable behavior of the mutations, in DM fami- CTG repeat instability, we also studied repeat lengths in lies the enormous fluctuations in repeat length and germ-line tissues and compared CTG length in father's clinical severity are most evident and easy to study. The sperm with that in his child's blood. Our findings show high prevalence of the disorder, the parental influences that intergenerational CTG-repeat-length changes oc- on disease severity (Harper 1989), the availability of cur mainly during early embryonic mitotic divisions, large families, and tissue biopsy samples render DM a resulting in both somatic and germ-line tissue mosaicism. particularly interesting model in which to study dy- namic mutations. Material and Methods The causative mutation in DM, an expanding (CTG)n repeat, is located in the 3' UTR of a gene encoding a Patient Tissues putative protein kinase, known as "DM-PK" (Brook et Blood samples were collected from a large panel of al. 1992; Fu et al. 1992; Jansen et al. 1992b; Mahadevan DM patients, investigated by the diagnostic DNA ser- et al. 1992). The gene is expressed predominantly in vices at the University of Nijmegen; the Department of smooth, skeletal, and heart muscle and at low levels in Medical Genetics, University of Antwerp; and the De- brain and endocrine tissues (Jansen et al. 1992b). This partment of Neurology, University of Maastricht. pattern of expression parallels the tissues affected by Sperm samples were obtained from selected patients, the DM phenotype. In normal individuals, the (CTG). by the diagnostic DNA services at the University of repeat may vary in length between 3 and 37 repeat units Nijmegen and the Department of Medical Genetics of (Brunner et al. 1992; H. Smeets, H. Brunner, and W. the University of Antwerp. DNA was isolated accord- Nillesen, personal communication) with length distri- ing to standard procedures (Miller et al. 1988). butions clustered at 5, 11-14, and 18-22 CTGs in the Tissue samples obtained at autopsy from one female normal population (Brook et al. 1992; Brunner et al. DM patient (patient F, who died at age 66 years and 1992; Fu et al. 1992; Mahadevan et al. 1992; Imbert et who presented classical DM features including myo- al. 1993). When tested within individual pedigrees, tonia, cataract, apathetic behavior, and progressive di- CTG repeats in the 3-37-trinucleotide size range segre- gestive and respiratory problems, with onset at age 42 gate stably in a Mendelian fashion. This suggests that years) were taken within 4 h of death. Samples of skele- the mutation rate for "normal" alleles has an upper tal muscle (quadriceps), heart muscle, ovary, uterus, limit similar to that estimated for many other microsat- skin, liver, kidney, brain (frontal cortex and thalamus), ellites typed in human and mouse Ueffreys et al. 1988), lung, adrenal gland, thymus, pancreas, mesenteric typically between 10-2 and 10'-. Expanded repeats are lymph node, and spleen of this patient were snap frozen much more unstable, and there is a positive correlation in liquid nitrogen and stored at -80'C. Similarly, tis- between the length of the repeat and its instability sues from a congenital case (patient L), two congenital (Hunter et al. 1992; Shelbourne et al. 1992; Brunner et twins (28-wk-old prematures who died 2 wk after birth al. 1993a; Lavedan et al. 1993b). The length of the re- from respiratory insufficiency), and an affected embryo peat ranges from 42 to approximately 150 in nonmani- (P) aborted in week 12 were prepared and stored at festing or mildly affected patients to several thousand -80°C for DNA analysis. DNA was isolated using CTG units in more severely affected individuals or con- standard procedures (Maniatis et al. 1988). Clinical genital cases of DM (Hunter et al. 1992; Brunner et al. diagnosis of patients was as based on the criteria estab- 1993a; Harley et al. 1993). lished by the Myotonic Dystrophy Working Group While studying CTG repeat lengths in tissue speci- (Griggs et al. 1989). mens from various DM individuals, we observed that CTG repeats up to approximately 100 repeat units were Validation and Optimalization of the (CTG)n Length detected as defined, though often faint, bands. Larger Analyses: Southern Blotting versus PCR Analysis repeat alleles appeared as a diffuse smear because of For the examination of genomic DNAs from individ- mitotic instability (Brunner et al. 1992; Hunter et al. ual tissue samples Southern blot analysis and PCR typ- 1992; Mahadevan et al. 1992; Shelbourne et al. 1992). ing were used. Southern blot analysis is the only suit- Gonosomal Mosaicism in Myotonic Dystrophy Patients S77 able method for the analysis of long and heterogeneous EDTA, 50 mM NaCl) were equilibrated for 30 min in CTG expansions in tissues from severely affected classi- 30 mM NaOH, 1 mM EDTA; and PCR samples were cal DM patients and cases of congenital DM. For each adjusted to 10 mM NaOH, 0.2 mM EDTA, 0.6% Fi- sample, 10 ,ug of genomic DNA was digested with coll-400, 0.01% bromophenol blue dye, before loading HindIII (Bethesda Research Laboratories), electro- on the gel. Alkaline gels were electrophoresed for 24 h phoresed on 0.7% agarose gels, and transferred onto at 20 V in 30 mM NaOH, 1 mM EDTA, and blotted nylon membranes (Hybond N+; Amersham). Southern and probed as described for the standard assay above. blots were probed with the 1.8-kbp HindIII/BamHI By using the denaturing electrophoresis where heterolo- fragment of probe pGB2.6 (Aslanidis et al. 1992), which gous strand pairing cannot occur, the extent of smear- detects the repeat expansion Blots were washed at 0.2 ing was significantly reduced. X SSC, 0.2% SDS final stringency, at 650C, and were exposed to Kodak X-Omat S film for 1-14 d at -80'C Results using two intensifying screens. Tissue Heterogeneity within Individuals Although the Southern blot analysis of DM-PK al- By comparing CTG repeat lengths among different leles can be significantly improved by the parallel use of tissues in four unrelated patients, we identified clear different restriction enzymes (i.e., EcoRI, BglI, BamHI, intraindividual somatic heterogeneity. The 15 tissue HindIII) that generate suitably sized DNA fragments samples from a female DM patient (F) with onset of spanning the CTG repeat, it is not particularly suitable muscle disease at age 42 years, who died of respiratory to detect small CTG expansions. Therefore, for the de- insufficiency at the age of 66 years, were especially inter- tection of DM alleles of up to 1,200 trinucleotide units, esting. Conventional Southern blot analysis of her tis- we applied the PCR/blotting method adapted for the sue DNAs revealed diffuse bands of 1,000-1,300 CTGs visualization of the variable simple-sequence motif in most tissues. The somatic heterogeneity became (VSSM; Jansen et al. 1992a; Mahadevan et al. 1992). more apparent when the PCR-based analyses on native When this method is used, large repeats typically appear and denaturing gels were used (fig. 1). The distribution as a broad smear on agarose gels, while acrylamide gels of CTG repeat lengths varied among tissues, though are better for the detection of length variations of some overlap was evident. Most strikingly, a distinct shorter alleles (Brunner et al. 1992). In the standard band of 1 kbp was seen in both thalamus and frontal assay, equal amounts of PCR products ('0.5 jg) were cortex. The size distributions and intensities of the sig- loaded on 1% agarose gels (1% agarose, 40 mM Tris, 1 nals were reproducibly seen in different DNA prepara- mM EDTA, adjusted to pH 7.6 with glacial acetic acid) tions from the same tissues in this patient. and resolved by electrophoresis. DNA was transferred Similar analyses were performed on material ob- to nylon membranes (Hybond N+) and probed with a tained at autopsy from two infants (patients L and P) (CTG)1O oligonucleotide end-labeled with 32p. with congenital DM. Heterogeneity of repeat length Initially, to test whether false priming of incom- was shown by Southern blot analysis in tissues of pa- pletely elongated products, followed by aberrant elon- tient L. The largest CTG repeat expansions were found gation during the subsequent PCR reaction cycles, in heart DNA, followed by fibroblast, liver, and brain could be involved in signal smearing in the PCR assay, DNAs (fig. 2A). An additional smaller band of approxi- we varied the reaction conditions (cycling times, anneal- mately 5 kbp can be seen in brain DNA from this pa- ing temperature or dNTP, and ion concentrations). tient, although the reason for the appearance of this This did not significantly change the length distribution extra band, which we have observed also in some other of PCR products and thus was ruled out as an impor- patients, remains unclear (H. Smeets and W. Nillesen, tant source of apparent length heterogeneity. For all unpublished observations). In muscle, brain, and cho- subsequent amplification reactions, therefore, condi- rion villi DNA of an affected embryo (P), aborted in wk tions as described elsewhere (Mahadevan et al. 1992) 12, major repeat expansions of approximately 600- were used. Still, illegitimate heteroduplex formation be- 1,000 trinucleotides were detected (not shown). These tween (CTG)n and (CAG). strand products may occur repeats were significantly larger than those in either of during the renaturation step in the last reaction cycle. the parental blood DNAs, but length variation between In order to circumvent this problem, PCR products tissues was not evident in this embryo. were also analyzed by denaturing gel electrophoresis, In all cases where no obvious allelic differences were using an adapted standard protocol (Maniatis et al. found between tissues, patients carried extremely long 1988). In brief, 1% agarose gels (1% agarose, 1 mM repeats, and, as a consequence, minor variations may 578 Jansen et al.

DENATURING NON-DENATURING MUSCLE HEART OVARY UTERUS SKIN LIVER KIDNEY FR.CORTEX THALAMUS LUNG THYMUS ADRENAL PANCREAS MES.LYMPHN. SPLEEN

o -L 0 P 0 c -'06 *O :A 01 Figure I Mutation analysis in 15 different tissue samples from one DM patient (F) with classical symptoms. Repeat length was determined using the PCR/blotting method combined with denaturing (left pane!) and nondenaturing (right pane!) agarose gel electrophoresis as described in Material and Methods. Between most tissue samples, clear differences in intensity and the extent of smearing of the larger allele can be observed. Note the similarities in DNAs from frontal cortex and thalamus. Fr.cortex = frontal cortex; Adrenal = adrenal gland; and Mes.lymphn. = mesenteric lymph node.

have remained undetected. Yet, from our analysis we compared repeat lengths in matched blood samples of conclude that unstable mitotic segregation may occur two MZ twins with either approximately 180 or 400 throughout all tissues, including peripheral blood lym- CTGs. No obvious differences were detected. Taken phocytes, liver, brain, heart, and sperm (see below). Fur- together, these results suggest that (CTG). instability is thermore, the similarity in number of repeats in thala- most pronounced during early embryonic stages, i.e., mus and frontal cortex of the one patient (F) whose before Carnegie stage Sb (day 10.5) of development, the tissues were analyzed in greatest detail suggests that the last possible moment for MZ twinning (Phillips 1993). postmeiotic variation is most prominent in early devel- Our findings do not exclude, however, that subsequent opment. expansion may continue to occur at a reduced rate in tissues with a high proliferative capacity. Developmental Stage and CTG Expansion To see whether there is indeed a developmental time Does Repeat Variation Occur during Meiotic Cell window for (CTG)n length alterations to occur, we ana- Division? lyzed a selected subset of DNAs from 35 different tis- In DM pedigrees there is a trend toward gradual sue biopsies taken from a pair of MZ twins with con- CTG length increase; size compression is a much rarer genital onset DM. Both children were born at wk 28 of event (Shelbourne et al. 1992; Abeliovich et al. 1993; gestation, and died after 2 wk from respiratory insuffi- Brunner et al. 1993b; Harley et al. 1993; Hunter et al. ciency. Southern blot analysis (fig. 2B) revealed small 1993; Lavedan et al. 1993b; O'Hoy et al. 1993). PCR but significant fluctuations in the size of their expan- analysis of blood DNAs of parent-child pairs in our sions (approximately 1,500 CTGs), with differences be- family material revealed many cases where a conspicu- tween tissues. It is significant that expansions were simi- ously large intergenerational increase in repeat length lar in identical tissues in both individuals. We also occurred. In these cases, the size distributions of the Gonosomal Mosaicism in Myotonic Dystrophy Patients 579

A% A ca

4, - kb B

14.0 E EXPANDED 12.0 <' DM ALLELE MIA BIA BIA BIA BIA BlA B

23.7 KBP - 8.5 NORMAL -' DM ALLELE <3 EXPANDED DM ALLELE 9.5 KBP - <,NORMAL DM ALLELE 6.7 KBP -

4.3 KBP -

Figure 2 A, Southern blot comparison of mutation expansion in four different tissues from one congenital DM patient (L). B, Six different tissues of congenital DM twins. Genomic DNAs (10 gg) from brain, heart, and liver biopsies and fibroblast culture cells from patient L (A) and matched biopsies from heart (both atrium and ventricle), lung, liver, kidney, and cerebrum frontale from the congenital twins (B) were digested with HindlIl, electrophoresed on a 0.7% agarose gel, and blotted and hybridized using standard procedures. In panel A, variation in mobility of the expanded allele (size range 12-14 kbp) can be seen, whereas the normal DM allele (8.5 kbp) shows no length heterogeneity. Similarly in panel B, small fluctuations in the repeat length in different tissues (expansion up to S kbp) with a remarkable match between identical tissues are evident. Marker sizes (phage lambda DNA digested with HindIll) are shown to the left of each panel. amplification products rarely display overlap between (family 3), the repeat was smaller in sperm DNA; in blood DNAs of parent and child (illustrated in the pedi- three cases no evident differences were detected. When grees shown in fig. 3). Densitometrically determined comparing father-child pairs, six of eight showed an estimates for the percentage overlap between smeared enlargement of the repeat in the affected offspring rela- signals in such parent-child combinations range gener- tive to the repeat length in the father's sperm (fig. 4A, ally from 0%-5%, that is, the chance that an event gen- families 1, 2, and 5). In two of these father-child pairs erating a nonpaternal allele was involved is at least 20:1 (e.g., see family 2) the repeat length in sperm and blood in these cases. of the father was similar, but in the affected child's If the repeat expansion (or regression) occurred dur- blood a longer repeat was present. Finally, in two other ing meiotic cell divisions, the gametes of the affected cases the repeat was longer in the sperm sample than in parent would be expected to display a length distribu- blood of the father but was considerably smaller than in tion similar to that in the offspring. To test this hypoth- the child's blood. esis we analyzed CTG repeat lengths in blood and In one father-child pair the paternal alleles of approx- sperm DNAs from 14 male DM patients with expan- imately 400 CTGs underwent compression to 24 CTG sions in blood up to approximately 1.5 kbp (500 CTGs) trinucleotides in the offspring (fig. 4A, family 6; Brun- and compared them with repeat lengths present in the ner et al. 1993b). As nonpaternity was excluded, this blood of their offspring. One male did not produce event represents a marked intergenerational length sperm and was excluded. In figure 4A we present six change. The repeat length in blood and sperm of the representative families. Among the 13 families, 9 cases father are of similar length (±400 and 600 CTGs), and revealed wider-ranged repeat lengths in sperm than in the allele of 24 CTG triplets (Brunner et al. 1993b) was blood (fig. 4A, families 1, 2, 4, 5, and 6). In one case undetectable in sperm. 580 Jansen et al.

020DM 021DM ples from nine males with large repeats in blood. Only D... two sperm samples contained enough DNA for South- ern blot analysis, because of the occurrence of extreme oligospermia in seven males, but all could be tested by PCR. In four cases the CTG lengths in sperm were ap- parently less than the lengths in blood (fig. 4B). Patient V also showed this expansion limitation on Southern blot, underscoring the validity of the analysis. In only kb one case, the only case with an intermediate (CTG). length in blood (patient IX), the mean repeat length was 6.0 - slightly longer in sperm. Although the number of sperm samples analyzed is limited, this is the first direct evi- dence that restriction of expansion occurs in male trans- mitters, as proposed by Lavedan et al. (1993a) and Mul- 1.0 - ley et al. (1993).

Discussion

0.15 - Mechanistic and Diagnostic Implications Figure 3 Non-Mendelian behavior of CTG repeat smears in How does the behavior of the unstable DM kinase two DM families (020 and 021). Repeat lengths were determined CTG repeat compare with that of other simple repeti- using PCR under nondenaturing conditions, and many cases of appar- tive DNA loci in other organisms? Typically, the muta- ent nonoverlap in repeat smears from parent and offspring are shown. Size markers are indicated to the left. All family members in tion frequency of VSSMs does not exceed 1 in 103 in the pedigrees shown are carriers of the mutation. Patients with pheno- meiotic cell divisions (Jeffreys et al. 1988). In contrast, typic expression of DM and asymptomatic carriers are represented by there are some hypervariable loci, such as the Ms6-hm blackened or unblackened circles and squares, respectively. (GGGCA). (Jeffreys et al. 1987; Kelly et al. 1989) and Hm-2 (Gibbs et al. 1991) loci in mice, the loci recog- Assuming that the repeat length in sperm provides an nized by probes MS32 Ueffreys et al. 1991) and CEB1 accurate estimate for the CTG length distribution at (Vergnaud et al. 1991), (CAC)Q loci in man (Niirnberg et the time of conception, we thus consider it unlikely al. 1989), and GAA motifs in chicken (Epplen et al. that a rare sperm cell with a significantly expanded, or 1991). These loci exhibit mosaicism in somatic tissues shrunken, CTG repeat, similar to those in the child but and germ lines, and although mutation rates may be undetectable in the father's sperm, contributed to zy- locus dependent (being as high as 2.5%-15% per segre- gote formation. While we cannot rule out meiotic in- gation event), there is evidence that de novo mutations volvement in expansion, our findings suggest that re- occur during early embryonic mitoses (Gibbs et al. peats expand during mitotic, postzygotic events that 1993). Recent analysis of the behavior of the (CGG). occur during the early formation of somatic and germ repeat associated with the fragile X syndrome revealed tissues. Of course we have to bear in mind that our a striking similarity to the DM situation in that repeat analyses may not give an accurate reflection of the CTG instability seems greatest during early postzygotic cell repeat lengths in the single gametes that were actually divisions (W6hrle et al. 1992, 1993). However, the two involved in the fertilization events. Moreover, we have disorders differ in that expansion limitation during sper- to assume that the length distribution of CTG repeats matogenesis seems to be more profound in fragile X in germ tissues of male DM patients does not signifi- males where only premutations have been detected in cantly alter over the years. This may not necessarily be paternal spermatozoa (Reyniers et al. 1993). the case, as the number of cell divisions involved in Imbert et al. (1993) recently suggested that loci with gamete (sperm) production increases with age (see below). 19-30 trinucleotides constitute a predecessor pool for unstable (>42 X CTG) alleles. In a family described by (CTG)n Repeats with More Than 1,000 Triplets Are our group, a complete reverse mutation occurred, in Absent in Sperm which approximately 400 CTGs reverted to 19 CTGs To study the behavior of longer CTG repeats, we (Brunner et al. 1993b). We now know that the son that compared repeat expansions in,sperm-blood DNA sam- inherited the normalized allele transmitted it stably to Gonosomal Mosaicism in Myotonic Dystrophy Patients 581 A

kbBSBBBSBBBBSBFBSBBBBSB BSB

6.0

1.0 0.15 B IV V VI VIl Vill IX kb BSBSBS B S B S B S B S B S B Skb E - - 1 6.0

II H e~~~~~~ 3@ °~~~~~. MAXIMUM REPEAT 6.0 EXPANSION IN SPERM DNA APP. 1000x CTG 3.0

0.15 0.16

Figure 4 A, Comparison of CTG repeat lengths in blood and sperm of mildly manifesting male patients and the blood of their children. Repeat lengths were determined in DNAs isolated from blood (B), sperm (S), or fibroblasts (F; family 4 only) of six male patients and were compared with repeat lengths in blood of their children, by using PCR and denaturing electrophoresis as described in Material and Methods. In two cases (families 3 and 5) the blood sample of the father's parent was also tested. Section 6 shows the follow-up analysis of repeat lengths in blood and sperm of the father of a case with complete mutation reversion, described elsewhere (Brunner et al. 1993b). It is striking that the average repeat length of 400 CTGs in blood is even further expanded in sperm (±600 CTGs), but the normalized (CTG)24 allele present in the child is not present at appreciable levels in the father's sperm. Patients with phenotypic expression of DM and asymptomatic carriers are represented by blackened and unblackened circles and squares, respectively. B, CTG-repeat expansion in blood (B) and sperm (S) samples of nine male patients with classical DM symptoms. Repeat length was determined as described above. Typically, the repeat is consistently shorter in sperm than in blood and does not exceed a threshold length of approximately 3 kbp (approximately 1,000 CTGs). Moreover, the signal from the expanded allele is clearly weaker in sperm, whereas the signals from the normal alleles are fully comparable between blood and sperm. Note that only in one case (IX) the repeat length span is wider in sperm; however, in this case the CTG repeat length in blood is of intermediate size. his first child. This finding may argue against the pres- able simple-sequence motifs? The following facts must ence of neighbouring cis-acting elements which, in cer- be taken into account: (i) the length alteration is largely tain families, render CTG repeats particularly unstable, due to mitotic events, predominantly during early devel- regardless of their length. opment; (ii) there is a clear bias for expansion and a low What then contributes to the extreme instability of tendency to contraction; (iii) the mutant alleles in the the (>42 X CTG) repeats in DM, and is there a com- offspring may be more than 3-5 times larger than those mon mechanism for the dynamic behavior of hypervari- in the affected parent (Shelbourne et al. 1992; Harley et 582 Jansen et al.

3-40 (CTG) s: slow VSSM expansion Normal mutation frequency 1 in 1000 meiosis 4 >42 (CTG) s: sister chromatid misalignment Healthy loops, triple-stranded helices or cruciform structures 4

n

on Ai Endonucl-eas 4 (CTG)n: recombin;ation gap repair

4 U VAW DNA pol.4 qua~ Affected 4 gap repair follovyved by mitosis \ n~~e \ n~~~~%

0 0

multiple consecutive events lead to extreme repeat expansion and somatic heterogeneity Figure 5 Putative (CTG)n expansion mechanism. In normal individuals the trinucleotide repeat segregates mitotically stably and has a low mutation frequency, comparable to other VSSMs. On expansion to a length of about 40 CTG units, the repeat may form misaligned structures (here shown as double-stranded, staggered loops, but triple-stranded helices or cruciform structures may also be formed). Enzymes involved in recombination and repair during mitosis resolve the unequal pairing of sister chromatids by endonucleolytic cleavage and double- stranded gap repair, as originally proposed by Jessberger and Berg (1991) (also see Nelson et al. 1989; Murti et al. 1992). Multiple consecutive events could lead to new (CTG)n alleles that are more than the sum of the parental allele lengths. Disequilibrium markers are shown as open circles or boxes (O and I). It is conceivable that failure to resolve the looped structure would ultimately lead to preferential loss of the larger allele (including neighboring sequences), chromosome fragility (as in fragile X), or perhaps even cell-cycle arrest. al. 1993; Lavedan et al. 1993b); (iv) segregation analysis Vergnaud et al. 1991). In DM there is evidence that the of flanking markers excluded meiotic crossing-over largest fluctuation in CTG length and disease manifes- events Jansen et al. 1992a; Imbert et al. 1993); (v) the tation is associated with paternal transmission of minor instability of the motif seems to increase with length; expansions (Brunner et al. 1993a). This may in fact re- and (vi) there is a clear paternal bias in the generation of flect the larger number of cell divisions during sper- new alleles (Jefreys et al. 1987; NUimberg et al. 1989; matogenesis. In females the ovum-to-ovum sequence Gonosomal Mosaicism in Myotonic Dystrophy Patients 583 involves approximately 30 cell divisions, whereas the (1993) proposed that such effects could underlie the number of cell divisions for male gamete production restriction of congenital DM cases to maternal trans- ranges between 50 and several hundred during the ef- mittance. fective fertile life span (Edwards 1989). Other gender In summary, our studies have demonstrated (CTG)n influences on the expansion mechanism cannot be length mosaicism in germ-line and somatic tissues of ruled out. DM patients. The occurrence of gonosomal mosaicism (Edwards 1989) has implications for risk assessments in CTG Expansion Model DM pedigrees. The strong tendency of the CTG repeat We suggest that a mechanism similar to that pro- to expand is presumably the basis for the high pene- posed for the repair of deletions and double-strand trance, the , and potentiation phenomena gaps in DNA by homologous recombination (Jess- in DM, and it may explain the progressiveness of the berger and Berg 1991) may explain many of these find- disease. If repeat length is directly correlated with dis- ings. Further support comes from the observation that ease severity, tissues with shorter repeats could be less events resembling gene conversion can occur during affected than tissues with longer repeats. Whether this mitosis in tissue-culture cells (Nelson et al. 1989) and forms the basis for the extreme variation in clinical are seen frequently in mitotic divisions in the germ-cell manifestation in DM will be uncertain until repeat lineage of mice (Murti et al. 1992). We propose that lengths in different tissues from mild, moderate, and expansion is triggered by unequal pairing of sister chro- severely affected patients can be analyzed. matids, which leads to formation of a four-stranded synaptic structure during the S, G2, or M phase of the cell cycle. Once the CTG repeat has attained a certain Acknowledgments critical size ('40 CTGs), the formation of staggered Drs. de Kleine and van Beek of the Veldhoven St. Joseph cruciform structures, triple-stranded structures, or Hospital and H. Croes and Dr. Jap of the Department of Cell loops (referred to as "loops" in fig. 5) in the double- Biology and Histology of the KUNijmegen are gratefully ac- stranded DNA helices of either of the two sister chro- knowledged for help in obtaining obduction samples. We matids could be initiated. This might lead to unequal thank David Iles and other colleagues of the Departments of pairing in the region of the repeat homology. Cleavage Cell Biology and Histology and Human Genetics, Katholieke of the strands opposite the looped structure by endonu- Universiteit Nijmegen, for critical evaluation of this work. clease and Part of this study was presented at the 42d annual American subsequent double-stranded repair of the Society of Human Genetics meeting in San Francisco. This gap, using the donor strands as a template, could then work was supported by the Muscular Dystrophy Association, lead to expansion of the recipient strand. If this oc- the Dutch Beatrixfonds, NWO-GBMW grant 900-501-140, curred during consecutive mitotic divisions, new and the Association Franqaise contre les Myopathies (grants (CTG)n alleles that are more than the sum of the paren- to B.W.). tal allele lengths could appear. Alternatively, cleavage within the looped segment and removal of this struc- ture would lead to length regression or even to reversal References of the mutation. In this scheme, coupling to markers in Abeliovich D, Lerer I, Pashut-Lavon I, Shmueli E, Raas- linkage disequilibrium is retained and is independent Rothschild A, Frydman M (1993) Negative expansion of from changes in repeat length. the myotonic dystrophy unstable sequence. Am J Hum Simple-sequence motifs, depending on their length, Genet 52:1175-1181 may serve as pause or arrest sites for DNA replication Aslanidis C, Jansen G, Amemiya C, Shutler G, Mahadevan M, (Baran et al. 1991), as has been described for fragile X Tsilfidis C, Chen C, et al (1992) Cloning of the essential syndrome (Hansen et al. 1993). If elaborate mispaired myotonic dystrophy region and mapping of the putative structures have to be resolved before mitosis can pro- defect. Nature 355:548-550 ceed and if failure to do so may delay or inhibit mitosis, Baran N, Lapidot A, Manor H (1991) Formation of DNA then this scheme could explain the influences on re- triplexes accounts for arrests of DNA synthesis at d(CT). and d(GA). tracts. Proc Natd Acad Sci USA 88:507-511 peat-length selection during gametogenesis. As sper- Brook JD, McCurrach ME, Harley HG, Buckler AJ, Church matogenesis needs more and more rapid mitotic cell D, Aburatani H, Hunter K, et al (1992) Molecular basis of divisions than does oogenesis, this model may explain myotonic dystrophy: expansion of a trinucleotide (CTG) differential effects on CTG sizes in male and female repeat at the 3' end of a transcript encoding a protein ki- transmittance. Lavedan et al. (1993a) and Mulley et al. nase family member. Cell 68:799-808 584 Jansen et al.

Brunner HG, Briiggenwirth HT, Nillesen W, Jansen G, Tsilfidis C, Korneluk RG (1993) Decrease in the size of the Hamel BCJ, Hoppe RLE, de Die CEM, et al (1993a) Influ- myotonic dystrophy repeat during transmission from par- ence of sex of the transmitting parent as well as of parental ent to child: implications for genetic counseling and ge- allele size on the CTG expansion in myotonic dystrophy netic anticipation. Am J Med Genet 45:401-407 (DM). Am J Hum Genet 53:1016-1023 Hunter A, Tsilfidis C, Mettler G, Jacob P, Mahadevan M, Brunner HG, Jansen G, Nillesen W, Nelen M, de Die CEM, Korneluk RG (1992) The correlation of age at onset with Hbweler CJ, van Oost BA, et al (1993b) Reverse mutation CTG trinucleotide repeat amplification in myotonic dys- in myotonic dystrophy. N Engl J Med 328:476-480 trophy. J Med Genet 29:774-779 Brunner HG, Nillesen W, van Oost BA, Jansen G, Wieringa Huntington's Disease Collaborative Research Group (1993) B, Ropers H-H, Smeets HJM (1992) Presymptomatic diag- A novel gene containing a trinucleotide repeat that is ex- nosis of myotonic dystrophy. J Med Genet 29:780-784 panded and unstable on Huntington's disease chromo- Conneally PM, Haines JL, Tanzi RE, Wexler NS, Penchasza- somes. Cell 72:971-983 deh GK, Harper PS, Folstein SE, et al (1989) Huntington's Imbert G, Kretz C, Johnson K, Mandel J-L (1993) Origin of disease: no evidence for locus heterogeneity. Genomics the expansion mutation in myotonic dystrophy. Nature 5:304-308 Genet 4:72-76 Edwards JH (1989) Familiarity, recessivity and germline mo- Jansen G, de Jong PJ, Amemiya C, Aslanidis C, Shaw DJ, saicism. Ann Hum Genet 53:33-47 Harley HG, Brook JD, et al (1992a) Physical and genetic Epplen JT, Ammer H, Kammerbauer C, Schwaiger W, characterization of the distal segment of the myotonic dys- Schmid M, Nanda I (1991) On the meaning of hypervari- trophy area on 19q. Genomics 13:509-517 able simple repetitive DNA loci in eukaryote genomes: an Jansen G, Mahadevan M, Amemiya C, Wormskamp N, initial attempt for a basic theoretical assessment. Adv Mol Segers B, Hendriks W, O'Hoy K, et al (1992b) Characteriza- Genet 4:301-310 tion of the myotonic dystrophy region predicts multiple Fu Y-H, Kuhl DPA, Pizzuti A, Pieretti M, Sutcliffe JS, protein isoform-encoding mRNAs. Nature Genet 1:261- Richards S, Verkerk AJMH, et al (1991) Variation of the 266 CGG repeat at the fragile X site results in genetic instabil- Jeffreys AJ, Neumann R, Wilson V (1991) Repeat unit se- ity: resolution of the Sherman paradox. Cell 67:1047-1058 quence variation in minisatellites: a novel source of DNA Fu Y-H, Pizzuti A, Fenwick RG Jr, King J, Rajnarayan S, polymorphism for studying variation and mutation by sin- Dunne PW, Dubel J, et al (1992) An unstable triplet repeat gle molecule analysis. Cell 60:473-485 in a gene related to myotonic muscular dystrophy. Science Jeffreys AJ, Wilson V, Kelly R, Taylor BA, Bulfield G (1987) 255:1256-1258 "Mouse DNA fingerprints" analysis of chromosome local- Gibbs M, Collick A, Kelly RG, Jeffreys A (1993) A tetranu- ization and germ line stability of hypervariable loci in re- cleotide repeat mouse minisatellite displaying substantial combinant inbred strains. Nucleic Acids Res 15:2823- somatic instability during early preimplantation develop- 2836 ment. Genomics 17:121-128 Jeffreys AJ, Wilson V, Thein SL (1988) Spontaneous muta- Gibbs M, Kelly R, Jeffreys AJ (1991) Evidence for somatic tion rates to new length alleles at tandem-repetitive hyper- mutation during early development at highly unstable variable loci in human DNA. Nature 332:278-281 mouse minisatellite loci. Genet Res 58:78 Jessberger R, Berg P (1991) Repair of deletions and double- Griggs RC, Wood DS, the Working Group on the Molecular strand gaps by homologous recombination in a mammalian Defect in Myotonic Dystrophy (1989) Criteria establishing in vitro system. Mol Cell Biol 11:445-457 the validity of genetic recombination in myotonic dys- Kelly R, Bulfield G, Collick A, Gibbs M, Jeffreys AJ (1989) trophy. Neurology 39:420-421 Characterization of a highly unstable mouse minisatellite Hansen RS, Canfield TK, Lamb MM, Gartier SM, Laird CD locus: evidence for somatic mutation during early develop- (1993) Association of fragile X syndrome with delayed repli- ment. Genomics 5:844-856 cation of the FMR1 gene. Cell 73:1403-1409 Knight SJL, Flannery AV, Hirst MC, Campbell L, Christo- Harley HG, Rundle SA, MacMillan JC, Myring J, Brook JD, doulou Z, Phelps SR, Pointon J, et al (1993) Trinucleotide Crow S, Reardon W, et al (1993) Size of the unstable CTG repeat amplification and hypermethylation of a CpG island repeat sequence in relation to phenotype and parental in FRAXE mental retardation. Cell 74:127-134 transmission in myotonic dystrophy. Am J Hum Genet Kremer EJ, Pritchard M, Lynch M, Yu S, Holman K, Baker E, 52:1164-1174 Warren ST, et al (1991) Mapping of DNA instability at the Harper PS (1989) Myotonic dystrophy, 2d ed. Saunders, Lon- fragile X to a trinucleotide repeat sequence p(CCG)n. don Science 252:1711-1714 Howeler CJ, Busch HF, Geraedts JP, Niermeijer MF, Staal A La Spada AR, Wilson EM, Lubahn DB, Harding AE, Fisch- (1989) Anticipation in myotonic dystrophy: fact or fiction? beck KH (1991) Androgen receptor gene mutations in X- Brain 112:779-797 linked spinal and bulbar muscular atrophy. Nature 352:77- Hunter AGW, Jacob P, O'Hoy K, MacDonald I, Mettler G, 79 Gonosomal Mosaicism in Myotonic Dystrophy Patients 585

Lavedan C, Hofmann-Radvanyi H, Rabes JP, Roume J, Jun- they tell us whether diseases are genetically determined? ien C (1993a) Different sex-dependent constraints in CTG Lancet 341:1008-1009 length variation as explanation for congenital myotonic Reyniers E, Vits L, De Boulle K, Van Roy B, Van Velzen D, dystrophy. Lancet 341:237 de Graaff E, Verkerk AJMH, et al (1993) The full mutation Lavedan C, Hofmann-Radvanyi H, Shelbourne P, Rabes J-P, in the FMR-1 gene of male fragile X patients is absent in Duros C, Savoy D, Dehaupas I, et al (1993b) Myotonic their sperm. Nature Genet 4:143-146 dystrophy: size- and sex-dependent dynamics of CTG mei- Richards RI, Holman K, Friend K, Kremer E, Hillen D, Sta- otic instability, and somatic mosaicism. Am J Hum Genet ples A, Brown WT, et al (1992) Evidence of founder chro- 52:875-883 mosomes in fragile X syndrome. Nature Genet 1:257-260 Macpherson JN, Nelson DL, Jacobs PA (1992) Frequent Richards RI, Sutherland GR (1992) Heritable unstable DNA small amplifications in the FMR-1 gene in fra(X) families: sequences. Nature Genet 1:7-9 limits to the diagnosis of "premutations." J Med Genet Ridley RM, Frith CD, Crow TJ, Conneally PM (1988) Antici- 29:802-806 pation in Huntington's disease is inherited through the Mahadevan M, Tsilfidis C, Sabourin L, Shutler G, Amemiya male line but may originate in the female. J Med Genet C, Jansen G, Neville C, et al (1992) Myotonic dystrophy 25:589-595 mutation: an unstable CTG repeat in the 3' untranslated Shelbourne P. Winqvist R, Kunert E, Davies J. Leisti J, Thiele region of the gene. Science 255:1253-1255 H, Bachmann H, et al (1992) Unstable DNA may be respon- Maniatis T, Fritsch EF, Sambrook J (1988) Molecular clon- sible for the incomplete penetrance of the myotonic dys- ing: a laboratory manual, 2d ed. Cold Spring Harbor Labo- trophy phenotype. Hum Mol Genet 1:467-473 ratory, Cold Spring Harbor, NY Sherman SL, Jacobs PA, Morton NE, Froster-Iskenius U, Miller SA, Dykes DD, Polesky HF (1988) A simple salting out Howard-Peebles PN, Nielsen KB, Partington NW, et al procedure for extracting DNA from human nucleated (1985) Further segregation analysis of the fragile X syn- cells. Nucleic Acids Res 16:1215 drome with special reference to transmitting males. Hum Mulley JC, Staples A, Donnelly A, Gedeon AK, Hecht BK, Genet 69:3289-3299 Nicholson GA, Haan EA, et al (1993) Explanation for ex- Vergnaud G, Mariat D, Apiou F, Aurias A, Lathrop M, clusive maternal origin for congenital form of myotonic Lauthier V (1991) The use of synthetic tandem repeats to dystrophy. Lancet 341:236-237 isolate new VNTR loci: cloning of a human hypermutable Murti JR, Bumbulis M, Schimenti JC (1992) High-frequency sequence. Genomics 11:135-144 germ line gene conversion in transgenic mice. Mol Cell Biol Verkerk AJMH, Pieretti M, Sutcliffe JS, Fu Y-H, Kuhl DPA, 12:2545-2552 Pizzuti A, Reiner 0, et al (1991) Identification of a gene (FMR-1) containing a CGG repeat coincident with a break- Nelson FK, Frankel W, Rajan TV (1989) Mitotic recombina- point cluster region exhibiting length variation in fragile X tion is responsible for the loss of heterozygosity in cultured syndrome. Cell 65:905-914 murine cell lines. Mol Cell Biol 9:1284-1288 Wohrle D, Hennig I, Vogel W, Steinbach P (1993) Mitotic Nirnberg P, Roewer L, Neitzel H, Sperling K, P6pperl A, stability of fragile X mutations in differentiated cells indi- Hundrieser J, P6che H, et al (1989) DNA fingerprinting cates early post-conceptional trinucleotide repeat expan- with the oligonucleotide probe (CAC)5/(CTG)5: somatic sion. Nature Genet 4:140-142 stability and germline mutations. Hum Genet 84:75-78 Wohrle D, Hirst MC, Kennerknecht I, Davies KE, Steinbach O'Hoy KL, Tsilfidis C, Mahadevan MS, Neville CE, Barcelo P (1992) Genotype mosaicism in fragile X fetal tissues. J, Hunter AGW, Korneluk RG (1993) Reduction in size of Hum Genet 89:114-116 the myotonic dystrophy trinucleotide repeat mutation dur- Yu S. Mulley J, Loesch D, Turner G, Donnelly A, Gedeon A, ing transmission. Science 259:809-812 Hillen D, et al (1992) Fragile-X syndrome: unique genetics Orr HT, Chung M, Banfi S, Kwiatkowski TJ Jr, Servadio A, of the heritable unstable element. Am J Hum Genet Beaudet AL, McCall AE, et al (1993) Expansion ofan unsta- 50:968-980 ble trinucleotide CAG repeat in type Yu S, Pritchard M, Kremer E, Lynch M, Nancarrow J, Baker 1. Nature Genet 4:221-226 E, Holman K, et al (1991) Fragile X genotype characterized Phillips DIW (1993) Twin studies in medical research: can by an unstable region of DNA. Science 252:1179-1181