Inverted duplications deletions: underdiagnosed rearrangements??

Authors


Orsetta Zuffardi, Genetica Medica, Università di Pavia, Via Forlanini, 14 27100 Pavia, Italy.
Tel.: +39 382 987733;
fax: +39 382 525030;
e-mail: zuffardi@unipv.it

Abstract

Molecular techniques led to the discovery that several chromosome rearrangements interpreted as terminal duplications were in fact inverted duplications contiguous to terminal deletions. Inv dup del rearrangements originate through a symmetric dicentric chromosome that, after asymmetric breakage, generates an inv dup del and a deleted chromosome. In recurrent inverted duplications the dicentric chromosome is formed at meiosis through non-allelic homologous recombination. In non-recurrent inv dup del cases, dicentric intermediates are formed by non-homologous end joining or intrastrand annealing. Some authors hypothesized that in these cases the dicentric may have been formed directly in the zygote. Healing of the broken dicentric chromosomes can occur not only in a telomerase-dependent way but also through telomere capture and circularization thus creating translocated or ring inv dup del chromosomes. In all the cases reported up to now, the duplicated region was always longer than the deleted one, but we can safely assume that there is another group of rearrangements where the deleted region is longer than the duplicated portion. In general, in these cases, the cytogeneticist will suspect the presence of a deletion and confirm it by FISH with a subtelomeric probe, but he/she will almost certainly miss the duplication. It is likely that the conventional analysis techniques used until now have led to a substantial underestimate of the frequency of inv dup del rearrangements and that the widespread use of array-CGH in routine analysis will allow a more realistic estimate. Obviously, the concomitant presence of deletion and duplication has important consequences in genotype/phenotype correlations.

Recurrent inv dup del(8p)

History of a recurrent rearrangement

The paradigmatic example of a rearrangement with an inverted duplication contiguous to a distal deletion is the inv dup del(8p). This is a recurrent rearrangement (about 1:20,000 newborns) reported in the literature since 1976 (1) associated with severe mental retardation, corpus callosum agenesis, important scoliosis in adulthood. The presence of an inverted duplication (“mirror duplication”(2)) had been hypothesized based on the banding pattern, whereas the demonstration that a concurrent deletion was also present in all inv dup(8p) cases had to wait for the availability of molecular techniques (3,4). Further molecular studies (5,6) on several cases demonstrated that the rearrangement had always the same breakpoints, with a single copy region of about 4–5 Mb interposed between the deletion and the duplication region. The deleted region has a size of 8 Mb whereas the duplication region varies in size from about 12 Mb to the entire short arm including the centromere. The findings that some inv dup del(8p) ended with an apparently inactive centromere (a second alphoid signal at distal 8p without any chromatids’ constriction) and that three alleles were present in the duplication region (5) reinforced the hypothesis of Weleber (1) that the rearrangement originated through end-to-end fusion between the short arms of chromosome 8 homologues. This would result in the formation of a symmetric dicentric chromosome. Since a dicentric chromosome is by definition unstable, it would subsequently undergo asymmetric breakage generating a duplicated-deleted chromosome with an inverted duplication and a simply deleted chromosome. The size of the duplication would vary according to where the breakage occurred between the two centromeres of the dicentric chromosome. The largest duplications reported up to now are those in which the dicentric ends with a second alphoid signal. This recurrent breakpoint (6 out 16 cases in ref. 5) indicates that, at anaphase, the dicentric is particularly fragile at the level of the second centromere. All published inv dup del (8p) cases have been ascertained by karyotype analysis that revealed an abnormal chromosome 8 with additional material on its short arm. Successive FISH analysis with chromosome 8 painting and subtelomeric 8p probes led to the identification of the chromosome 8 origin of the extra material on the short arm and of a distal deletion, respectively. Further dual-color FISH experiments with probes located in the duplicated segment demonstrated the inverted orientation of the duplicated segment (5,6). The introduction of whole genome array-CGH analysis allowed a rapid definition of the size of the deleted and duplicated segments (7,8) and led to the discovery of cases with a more complex situation with the inv dup del(8p) capped with a portion of 8q24.13-qter (7).

The rearrangement is mediated by homologous segmental duplications

As hypothesized (3), the dicentric was found to be the result of unequal recombination between homologous low copy repeats (LCRs, REPD and REPP, Fig. 1A), constituted by complex repeats containing olfactory receptor gene clusters lying at 8 and 12 Mb on the short arm of chromosome 8 and flanking a single copy region (6,9). The rearrangement was the result of non-allelic homologous recombination (NAHR) leading to the formation of a dicentric (qter->p23.1::p23.1->qter) and a reciprocal acentric chromosome (pter->p23.1::p23.1->pter) (Fig. 1B). As expected from classical cytogenetics, the dicentric and acentric chromosomes are the two products of a crossing-over within a cryptic paracentric inversion at 8p23.1 between the deletion and duplication regions (Fig. 1A). We have demonstrated that all these rearrangements originate at the meiosis of a parent who is heterozygous for the cryptic paracentric inversion (Fig. 2). The inversion is a polymorphism present in heterozygous state in about a quarter of the population (6,9). Among the twenty-seven inv dup del(8p) cases we studied for parental origin, only one originated at the paternal meiosis from a father heterozygous for the inversion, while all the others originated from a heterozygous inverted mother (Table 1). The derivative acentric chromosome is either lost or it acquires a neocentromere and gets rescued (6,10). The dicentric chromosome, in theory, can undergo breakage during meiosis or it can be inherited as such in the zygote. A few cases of mosaicism with different derivatives of the dicentric indeed suggest that the dicentric is usually present in the zygote: both Soler et al (11) and Pramparo et al (12) found in chorionic villi samples two cell lines, one with a del(8)(p11), the other one with an inv dup del(8p). In the case analyzed by Pramparo et al, a third cell line was also present with an inv dup del(8p) ending with the satellites of a D or a G short arm, suggesting that here the inv dup del(8p) had been stabilized through telomere capture leading to a translocated inv dup del chromosome. Although the mosaicism was not confirmed in the product of conception because it was not available, these findings indeed suggest that in the zygote the dicentric chromosome undergoes differential breakage in different cells leading to different cell lines. The most viable one(s) will be detected at birth.

Figure 1.

Mechanism of recurrent inv dup del(8p) formation. (A) Abnormal meiotic pairing between chromosomes 8 in a subject heterozygous (+/-) for the REPD/REPP inversion and non-allelic homologous recombination between the LCRs leads to the formation (B) of an acentric der(8p) and a dicentric 8qter-8p23.1::8p23.1-8qter chromosome. Breakage of the dicentric chromosome leads to the creation (C) of 8p- and inv dup del(8p) chromosomes. The REPD and REPP LCRs are indicated by red boxes. Recombination and breakpoint locations are shown by arrows.

Figure 2.

Metaphase from a mother transmitting the inv dup del(8p). FISH with BACs RP11-399J23 (green) and RP11-589N15 (red) shows that the green signal is located at pter in one chromosome 8 (arrow) and proximal to the red signal in the other one (arrowhead) indicating the presence of a cryptic paracentric inversion in the latter.

Table 1.  Microsatellites analyses in 27 cases of inv dup del(8p) and their parents
CasesMicrosatelitesDeletion regionDuplication regionParental origin
MFPMFP
  1. For each case, two microsatellites (the first deleted and the second duplicated) have been reported. None of these cases has been investigated by array-CGH analysis.

  2. M: mother; F: father; P: proband. ND: not determined

1D8S518;D8S552242/248ND226173/175ND171/173/175mat
2D8S262;D8S552114112/114112175/181165/171165/175/181mat
3D8S262;D8S261114110/116110121/133123/129121/123/133mat
4D8S262;D8S258112/114108/116116144/146142/146142/144/146mat
5D8S518;D8S1809226240/246240166/170158/166158/166/170mat
6D8S262;D8S552115/117ND113166/174ND166/172/174mat
7D8S262;D8S261109/111ND113125/129ND123/125/129mat
8D8S1819;D8S261217203/211203123/129135/137123/129/137mat
9D8S1819;D8S261203/213203/219213123/135125/129125/127/135pat
10D8S1824;D8S258230/244228/238238142/146146/150142/146/150mat
11D8S1819;D8S549203/216216/218218162/164166162/164/166mat
12D8S518;D8S258226/248226/240240142/144146142/144/146mat
13D8S1824;D8S1809230/244230/238238152/158170/172152/158/170mat
14D8S1824;D8S511232/252240/246240130/132128/130128/132*mat
15D8S1819;D8S511217/219203/217203128/132128/130128/130/132mat
16D8S262;D8S258108112/114114142/144138/144138/142/144mat
17D8S518;D8S511226ND242130ND128/130*mat
18D8S518;D8S511226/242226/248248128/130130/132128/130/132mat
19D8S262;D8S552115ND123172/178ND172/176/178mat
20D8S1819;D8S258211203/217203144/146144/148144/146/148mat
21D8S1824;D8S511245229/245229129127/131129*/131mat
22D8S262;D8S511116116/118118127/129127/131127/129*mat
23D8S262;D8S511110/114116116125/127127/129125/127/129mat
24D8S518;D8S511226ND242129ND127/129*mat
25D8S518;D8S552226/244226/242242176/178168/172172/174/178mat
26D8S518;D8S261242/250226/242226123/127129123/127/129mat
27D8S262;D8S552108112/114114174/176166/176174/176*mat

Other recurrent inv dup del chromosomes

Apart from supernumerary marker chromosomes (SMCs), the inv dup del(8p) is by far the most frequent among the rearrangements of this type. Among the others, very few cases are formed through NAHR. We are aware of only one other case of inv dup del clearly mediated by NAHR. It is an inv dup del(Xp) with breakpoints at the VCX gene family at Xp22.31 (Iascone et al, in preparation). On hindsight, it is clear that many SMCs belong to this category of rearrangements: apart from the acentric inv dup del(8)(pter-> p23.1::p23.1->pter), the well-known inv dup(15) and inv dup(22) seem to follow the same scheme.

The inv dup(15)

The inv dup(15) is the most common SMC in humans, accounting for as much as 60% of all those observed (13). It is generally asymmetric, with a single copy region between the two duplication regions. Similarly to inv dup del(8p), they originate at the maternal meiosis through either interchromosomal or intrachromosomal abnormal recombination. The 15q11-q13 region contains several blocks of homologous LCRs which may give rise to a variety of rearrangements. Depending on the LCRs involved in their formations, these SMCs may have variable size and structure. They can be divided in two main categories: small SMC(15)s and large SMC(15)s. Small SMC(15)s, whose frequency is 0.14–0.72 per 1000 newborns (14), do not contain the Prader-Willi/Angelman syndrome (PWS/AS) region, and their formation occurs proximally to this region which contains two blocks of LCRs known as BP1 and BP2. These marker chromosomes can be de novo or inherited and are not usually associated with a pathological phenotype, although male carriers may be infertile. Uniparental disomy associated with small SMC(15)s has been reported, responsible for either PWS or AS (15). These small SMCs have also been reported in association with the PWS/AS 15q11-q13 deletion (16). Large SMC(15)s are usually de novo and maternally derived. The presence of this marker chromosome is mostly associated with mental retardation, seizures, autistic behavior and/or other psychiatric problems (17,18). Because of the presence of several blocks of LCRs, large SMC(15)s may have more than one recurrent breakpoint but in all cases they fall distally to the PWS/AS region. As a consequence, these markers contain two copies of the PWS/AS region which seems to be at least partially the cause of the pathological phenotype. Paternally derived large SMC(15)s have never been reported. One explanation is that such paternally derived marker chromosomes may be lethal; however, paternally derived interstitial triplications of the PWS/AS region have been described (19,20). Another explanation is that spermatozoa carrying large SMC(15)s are selected against, as suggested by some studies (14,21). Two possible mechanisms have been proposed to explain the formation of SMC(15)s: a mechanism is based on NAHR between LCRs leading to the formation of a small dicentric and a large acentric chromosome which is subsequently lost. This recombination may occur either between homologues or between sister chromatids (18). Alternatively, a normal chromosome may undergo a premeiotic breakage losing a large segment of its long arm. After replication, the sticky ends of the two broken chromatids join to form the dicentric.

The inv dup(22)

Another common SMC is the inv dup(22). Although the most common genomic disorder occurring at 22q11 is the interstitial deletion causative of DiGeorge/Velocardiofacial syndrome (DGS/ VCFS), the richness of segmental duplication of this region catalyzes, through NAHR, the formation of other types of rearrangements, including inv dup(22). Depending on which segmental duplication(s) trigger(s) the formation of the rearrangement, inv dup(22) can be symmetric or asymmetric and can have variable size: these markers can contain zero, one or two copies of the DGS/VCFS region (22). These supernumerary markers are usually associated with the cat-eye syndrome (CES), characterized by ocular coloboma, anal atresia, congenital heart defects, renal malformations, male genital anomalies, skeletal defects, and borderline-to-moderate mental retardation. However, the phenotypic spectrum of this genomic disorder is extremely variable also because most of them are in mosaicism with a normal cell line. In contrast with SMC(15), inv dup(22) can originate both at maternal and paternal meiosis and non-allelic recombination may occur either between homologues or between sister chromatids (22). Here, again, no reciprocal acentric product has ever been reported, seemingly reflecting its lethality.

Non-recurrent inv dup del chromosomes

Inverted duplications/deletions have been identified at least at the following chromosome ends: 1p (23), 1q (24,25), 2p (26), 2q (27), 3p (28, 29), 4p (30,31), 4q (32), 5p (33,34), 7q (35,36), 9p (37), 9q (38), 10p (35), 10q (35), 11p (39), 14q (40,41), 15q (42), 18q (43), 21q (44) and Xp (45,46). In the majority of the above-mentioned cases, identification of the rearrangement occurred after a chromosome with an abnormally long arm had been detected by conventional cytogenetic techniques, then FISH analysis with a subtelomeric probe had shown a terminal deletion. Additional FISH with more proximal probes would then reveal the associated duplication and two color FISH would demonstrate its inversion. It is important to note that FISH analysis with commercial subtelomeric probes, that can be located even 300–400 Kb proximal to the telomeric end, may not be able to detect the distal deletion concurrent to the contiguous inverted duplication. We recently published the case of an inv dup del(2p) (26) in which the subtelomeric FISH marker was preserved and the deletion could only be demonstrated by high resolution array-CGH. In all these cases the duplicated region was always longer than the deleted one, but we can safely assume that there is another group of rearrangements where the deleted region is longer than the duplicated portion. In general, in the latter inv dup del cases, the cytogeneticist will suspect the presence of a deletion and confirm it by FISH with a subtelomeric probe, but he/she will almost certainly miss the duplication. It is likely that until now conventional analysis techniques have led to a substantial underestimate of the frequency of inv dup del rearrangements and that the widespread use of array-CGH in routine analysis will allow to reach a more realistic estimate.

Unlike the recurrent inv dup del rearrangements, in most cases occurring by NAHR, the non-recurrent inv dup dels are generated by alternative mechanisms (Fig. 3). In fact, illegitimate repair of a double-strand break (DSB) by non-homologous end joining (NHEJ) (47) or intrastrand annealing (48) can also form dicentrics. Breakpoint junction analysis at the DNA sequence level revealed that none of the inv dup del rearrangements studied were mediated by segmental duplications or facilitated by the presence of short inverted repeats (IR). Homology between the two breakpoint regions was limited to a dinucleotide (26) or trinucleotide (23) (Case 40), suggesting microhomology-mediated intra-strand repair of ds-breaks, which favours intrachromatid misalignment and recombination. The presence of a single copy sequence between deleted and duplicated regions was demonstrated in this type of rearrangements (23,26) (Cases 30 and 40 in ref 23). The size of the single-copy region ranges between 1.7 Kb (Case 30) (23) and 4.5 Kb (26). In 2 out of 3 cases (Case 30 in ref. 23 and ref. 26), the breakpoint between normal and deleted sequence is in a region without any apparent structural features, while the breakpoint between normal and duplicated sequence lies within a G/C-rich region or a low complexity repeat, seemingly able to assume a non-B DNA structure (49). This secondary structure can probably protect the chromosome end from further degradation and facilitate repair. Repair could be effected through a breakage-fusion-bridge (BFB) cycle, as stated by Ballif and colleagues (23), or by another process involving non-homologous end joining (NHEJ) or microhomology-mediated intrastrand repair (50). It is likely that the same general structure, and consequently the same process, may underlie all non-recurrent inverted duplications associated to terminal deletions. In all these cases we have no clear demonstration that the original dicentric chromosome originated at the parental gametogenesis. In fact, some authors hypothesized that the dicentric had been formed directly in the zygote after subtelomeric breakage and sister chromatid reunion (51).

Figure 3.

Mechanism of non-recurrent inv dup del(2p) formation. (A) Double-strand break followed by U-type exchange or micro-homology-mediated intrastrand pairing leads to the formation (B) of a dicentric 2qter-2p25.3::2p25.3-2qter chromosome. Breakage of the dicentric chromosome leads to the creation (C) of 2p- and inv dup del(2p) chromosomes. The U-type exchange is shown in red. Recombination and break locations are shown by arrows.

Inverted duplications/ triplications and interstitial triplications

A new type of rearrangement consisting of the duplication of 8p23.1 and the triplication of 8p23.2 [dup trp(8p)], mediated by the combined effects of two unrelated low copy repeats (LCRs), was found in two patients affected by mental retardation and minor facial dysmorphisms (52). The first set of LCRs consisted of the two OR-REPs involved in the recurrent inv dup del (8p). The second type of LCRs consisted of short (15 kb) repeats lying in inverted orientation at 8p23.2 and enclosing a non-repeated sequence of approximately 130 kb. The molecular characterization of a third case with a dicentric chromosome 8, deleted in its distal 8p portion, demonstrated that the rearrangement had been generated by NAHR between the short IRs. Based on our findings, we proposed a model showing that a second recombination event at the level of the OR-REPs leads to the formation of the dup trp(8p) chromosome. This rearrangement can only arise during meiosis in heterozygous carriers of the polymorphic 8p23.1 inversion, whereas in subjects with non-inverted chromosomes 8 or homozygous for the inversion only the classic dicentric chromosome can be formed.

Inv dup del and ring chromosomes

Recently two papers (41,53) have been published describing ring chromosomes in which array-CGH analysis showed not only the expected deletion at one or both chromosome ends but also a contiguous duplication (Fig. 4). In three of these cases FISH analysis demonstrated that the duplication was inverted (Fig. 1 in ref. 41 and Fig. 2 in ref 53). Rossi et al (53) found that 7 out 33 ring chromosomes were in fact not only deleted but also duplicated indicating that a precise mechanism is operating. Here, again, the presence of an inv dup del chromosome, although in a circular form, led us to hypothesize that the starting point was a symmetric dicentric chromosome present in the zygote or in early embryogenesis. Its breakage would have led to an inv dup del that in the absence of telomerase would stabilize itself through circularization. If we consider that an asymmetric breakage of such a dicentric chromosome results in the formation of an inv dup del and a simply deleted chromosome and that both need to be healed, we might hypothesize that also some “regular” ring chromosome having only the expected deletion can originate through an original dicentric chromosome.

Figure 4.

A: array-CGH (244A, AGILENT) of a r(13) (Case 7 in Rossi et al, 2008) showing a 20 Mb duplication concomitant to a distal 10 Mb deletion. B: dual colour fluorescence in situ hybridisation (FISH) of the same r(13) with BAC RP11-164b1, green, and RP11-122a8, red showing that the duplication is inverted.

Inv dup del and de novo unbalanced translocations

On 2002, Kostiner et al, (54) described an inv dup del(8p) case having a cytogenetically detectable piece of chromosome 18q on the distal end of its short arm, suggesting that the chromosome was stabilized by telomere capture. We found two similar cases that, on the basis of subtelomeric FISH on a p+ chromosome, were originally interpreted as unbalanced translocations. Both were de novo and only array-CGH analysis to size the unbalanced regions led to the discovery that they were inv dup del stabilized through telomere capture. These findings indeed suggest that also some de novo unbalanced translocations may derive from the same mechanisms of classical inv dup del rearrangements, once again demonstrating that in the absence of telomerase different mechanisms operate to heal broken chromosomes.

Recurrence risk

There is no recurrence risk for any of the de novo imbalances we discussed. Thus, no invasive prenatal investigations are indicated after the birth of a child with a de novo inv dup del chromosome. Even the inv dup del(8p)s, that are the recombinant products of a parental 8p23.1 heterozygous inversion, are always single events reflecting either the rarity of the abnormal recombination or the early lethality of zygotes with an almost complete trisomy 8 due to the presence of the dicentric chromosome. In fact, dicentric 8p with an inactive centromere has been reported only in mosaic with an 8p deletion cell line (55) or in uniparental disomy (52,56). It has been estimated that the frequency of the inv dup del(8p) is of about 1:20.000 newborns (5). However, we do not know the frequency of NAHR leading to this rearrangement or to its reciprocal acentric chromosome. This information can only be gained through sperm-based assays to measure the de novo rate of these rearrangements (57).

Final conclusions

We showed that several rearrangements in humans are secondary to the formation of a symmetric dicentric chromosome. In some cases this dicentric is formed at meiosis, whereas in other cases it is not definitely clear whether it is formed during parental gametogenesis or in early embryogenesis. The instability of the dicentric and the low or absent telomerase activity in cleavage stage embryos (58) lead the chromosome derived from breakage of the dicentric to form different types of rearrangements including classical inv dup del, ring and unbalanced translocation chromosomes. In several of them a single copy region interposed between the deleted and the duplicated regions has been demonstrated whenever high resolution genome-wide arrays and/or breakpoint cloning has been performed (26).

These new technological approaches are only now allowing investigators to bring to light the complexity and frequency of these types of rearrangements.

Acknowledgements

This work was supported by CARIPLO 2007 (5197) to OZ, by Telethon Grant GGP06208 to MCB and RG and by Telethon Grant GGP08226 to RC.

Ancillary