Organisation of the pericentromeric region of chromosome 15: at least four partial gene copies are amplified in patients with a proximal duplication of 15q
- J A Fantes1,2,
- S K Mewborn1,
- C M Lese1,
- J Hedrick1,
- R L Brown1,
- V Dyomin3,
- R S K Chaganti3,
- S L Christian4,
- D H Ledbetter1
- 1Department of Human Genetics, University of Chicago, Chicago, IL 60637, USA
- 2Medical Genetics Section, Department of Medical Sciences, University of Edinburgh, Edinburgh EH4 2XU, UK
- 3Cell Biology Program and the Department of Medicine, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021, USA
- 4Department of Psychiatry, University of Chicago, Chicago, IL 60637, USA
- Correspondence to: Dr D H Ledbetter, Department of Human Genetics, The University of Chicago, 920 East 58th Street, Chicago, IL 60637, USA;
- Accepted 19 November 2001
- Revised 15 November 2001
Clinical cytogenetic laboratories frequently identify an apparent duplication of proximal 15q that does not involve probes within the PWS/AS critical region and is not associated with any consistent phenotype. Previous mapping data placed several pseudogenes, NF1, IgH D/V, and GABRA5 in the pericentromeric region of proximal 15q. Recent studies have shown that these pseudogene sequences have increased copy numbers in subjects with apparent duplications of proximal 15q. To determine the extent of variation in a control population, we analysed NF1 and IgH D pseudogene copy number in interphase nuclei from 20 cytogenetically normal subjects by FISH. Both loci are polymorphic in controls, ranging from 1-4 signals for NF1 and 1-3 signals for IgH D. Eight subjects with apparent duplications, examined by the same method, showed significantly increased NF1 copy number (5-10 signals). IgH D copy number was also increased in 6/8 of these patients (4-9 signals). We identified a fourth pseudogene, BCL8A, which maps to the pericentromeric region and is coamplified along with the NF1 sequences. Interphase FISH ordering experiments show that IgH D lies closest to the centromere, while BCL8A is the most distal locus in this pseudogene array; the total size of the amplicon is estimated at ∼1 Mb. The duplicated chromosome was inherited from either sex parent, indicating no parent of origin effect, and no consistent phenotype was present. FISH analysis with one or more of these probes is therefore useful in discriminating polymorphic amplification of proximal pseudogene sequences from clinically significant duplications of 15q.
Molecular cytogenetic analysis has shown that apparent duplications of proximal 15q identified by G banding are not usually associated with a duplication of DNA probes in the Prader-Willi/Angelman critical region. The majority of these cases are not duplicated for SNRPN and have a normal phenotype or inconsistent phenotypes.1–4 They are usually inherited from a phenotypically normal parent and have been interpreted as euchromatic variants.
Two recent reports5, 6 used molecular cytogenetic techniques to show that amplification of a pseudogene cassette, mapping close to the centromere of 15q, causes the visible appearance of a duplication in banded chromosomes. The pseudogene cassette contains truncated gene sequences from a minimum of three genes transposed from other sites in the genome. One such sequence was shown to be a partial copy of the neurofibromatosis-1 (NF1) gene from chromosome 17q11.2.5, 7–10
To determine whether increased dosage of this NF1 pseudogene correlated with observed cytogenetic duplications of 15q11.2, Barber et al5 used FISH to show that the signals from a PAC probe containing the NF1 pseudogene were increased in intensity in two unrelated cases of chromosome 15 proximal duplications and estimated the degree of amplification from measurements of fluorescence intensities on metaphase chromosomes.
Ritchie et al6 found that a truncated copy of the GABRA5 (γ-aminobutyric acid type A receptor α 5 subunit) gene, the full length copy of which maps within the Prader-Willi/Angelman critical region, mapped to the pericentromeric region of chromosome 15, and was present in multiple copies on proximal duplication chromosomes. An estimate of the degree of amplification was made from the number of FISH signals in interphase nuclei and from released chromatin preparations. By isolating P1 clones from the region, they found that both the GABRA5 duplication sequence and one of the NF1 non-processed pseudogenes were present in the same P1 clone in the contig. A third sequence, the immunoglobulin heavy chain diversity segment gene (IgH D), had previously been mapped to proximal 15q11–13 and was shown to be amplified in proximal duplications, but was not part of the same contig.
Ritchie et al6 reported variation in the number of signals from GABRA5 and flanking probes in cytogenetically normal controls. In this paper, we present a more detailed study of the variation in copy number of NF1 and IgH D sequences in 20 cytogenetically normal controls and in eight proximal duplication cases. We quantified the degree of amplification by interphase FISH and established the normal range in controls. In addition, we show that a fourth pseudogene for BCL8 (Dyomin and Chaganti, unpublished data) is part of the amplified cassette. BCL8 was originally identified at the breakpoint of a translocation (t(14;15)(q32;q11-13)) in a diffuse large cell lymphoma patient.14 Recent work has shown that the copy on chromosome 15 lacks several exons, now termed BCL8A; the full length copy, BCL8B, maps to 13q11-q12 (Dyomin and Chaganti, unpublished data). By interphase FISH mapping we have established the order of the pseudogenes within the cassette and estimated the size of the region amplified.
MATERIALS AND METHODS
The control population consists of 20 cytogenetically normal subjects from the University of Chicago Clinical Cytogenetics Laboratory. Details of the control population's ascertainment are given in table 1.
The patient population was divided into two groups: subjects in group I are phenotypically normal and were ascertained during routine cytogenetic analysis by the presence of a proximal duplication on chromosome 15, or as the parent of a fetus with a proximal duplication, while subjects in group II were ascertained on the basis of an abnormal phenotype and were found by cytogenetic analysis to have a proximal duplication on chromosome 15.
Case A was a 31 year old, phenotypically normal female, who was found to have a proximal duplication of one chromosome 15 after identification of a similar chromosome in a pregnancy loss. No evidence of a duplication was found by FISH using probes for D15S11, SNRPN, and GABRB3. Cytogenetic analysis of her parents determined that the chromosome with the duplication was inherited from her mother.
Case B was a 21 week fetus that showed extra material on chromosome 15q during routine prenatal diagnosis by amniocentesis. No evidence of duplication was found using FISH with SNRPN and the same duplication was subsequently found in the phenotypically normal mother.
Cases C and E are phenotypically normal mothers whose amniocentesis showed a proximal duplication of one chromosome 15 in the fetus. Case D is the father of a fetus found to have a proximal duplication of 15q11-13. Subsequent chromosome analysis showed that subjects C, D, and E had a similar proximal duplication. No duplication of SNRPN was seen by FISH in any of these cases.
Case F was a 5 year old child with autism. Cytogenetic analysis showed extra material on one chromosome 15q, but SNRPN and more distal probes were not duplicated.
Case G was a 9 year old child with moderate mental retardation, developmental delay, hyperactivity, and a proximal duplication of 15q11-12. Subject G's parents are first cousins. The SNRPN probe was not duplicated.
Case H was a 7 year old child with a diagnosis of autism and mild dysmorphic features. A maternal cousin is mentally retarded. Previous cytogenetic analysis had shown extra material on chromosome 15, but SNRPN was not duplicated.
A P1 clone (P1-4) for the NF1 pseudogene on chromosome 15 was described by Purandare et al.8 The chromosome 15 specific IgH D cosmid (c13C6) was described by Nagaoka et al.11 GS5022 is a P1 clone containing the STS D15S18, described by Christian et al.15 A plasmid clone (pMC15) for the more distal of the two chromosome 15 specific alphoid sequences (D15Z3) is described by Finelli et al.16 The SNRPN probe was obtained commercially (Vysis Inc, Downers Grove, IL). Two PACs (P398C20, P406K8) and a phage clone (SS283) containing BCL8A have been described by Dyomin et al.14 Primers derived from the 5 kb GABRA5 duplication sequence (GenBank AF061786) were used to screen a total human BAC library (Genome Systems, St Louis, MO) by PCR and clone GS35F21 was isolated. DNA from genomic clones was isolated using an automated DNA isolation system (AutoGen 740, Autogen Inc, Framingham, MA).
Chromosome preparations were made from peripheral blood cultures or lymphoblastoid cell lines using standard methods. In dual colour FISH experiments, probe labelling, DNA hybridisation, and antibody detection were carried out using methods described previously.17 In three colour ordering experiments, probes were labelled with biotin-16-dUTP or digoxigenin-11-dUTP (Roche, Indianapolis, IN) or directly labelled with Spectrum Orange-dUTP (Vysis Inc, Downers Grove, IL). Biotin labelled probes were detected with avidin-Cy5 (Amersham Pharmacia Biotech Inc, Piscataway, NJ) and digoxigenin labelled probes with FITC anti-digoxigenin (Roche, Indianapolis, IN). FISH slides were analysed using a Zeiss Axiophot microscope with the appropriate filters (No 83000 for DAPI, FITC, and rhodamine; No 84000 for DAPI, FITC, Spectrum Orange, and Cy5; Chroma Technology, Brattleboro, VT), a cooled CCD camera (Nu 200; Roper Scientific, Tucson, AZ), and Quips PathVysion software (Applied Imaging, Santa Clara, CA).
The copy number for each probe in both control and proximal duplication subjects was determined by counting the number of signals in each chromosome 15 domain in at least 30 interphase nuclei and calculating the mode for each domain.
Interphase ordering and distance measurements
The order of the probes was scored in at least 20 interphase nuclei. The distance between the two probe signals in an interphase nucleus was measured using IPLab software (Scanalytics, Fairfax, VA). Probes were hybridised to interphase nuclei from a subject previously shown to have two copies of the NF1 and IgH D probes in one chromosome 15 domain. At least 20 interphase distances were measured for each pair of probe signals. Genomic distance was estimated from a calibration curve.18, 19
NF1 pseudogene and IgH D gene segments map proximal to PWS/AS deletions
Using fluorescence in situ hybridisation (FISH), we mapped an NF1 pseudogene P1 clone (P1-4) and a cosmid (c13C6) containing an IgH D segment in relationship to the common proximal breakpoints (BP1 and BP2) found in PWS/AS deletion patients and subjects with small supernumerary inv dup(15) chromosomes20, 21 (fig 1). When these probes were hybridised to metaphase chromosomes from a class I deletion patient (deletion extending from BP1 to BP3), signals were present on both the normal chromosome 15 and the deleted chromosome 15 (data not shown), placing both clones between the centromere and BP1, the most proximal breakpoint in PWS/AS deletions. Both clones showed signals on class I as well as class II small inv dup(15) marker chromosomes, confirming their position proximal to BP1. This result is consistent with the data obtained by Ritchie et al.6
Polymorphic variation in controls
In our initial hybridisations with the NF1 and IgH D clones on metaphase chromosomes in normal subjects, we observed differences in signal size and intensity between the two chromosome 15 homologues. To investigate this variation, we hybridised both clones to metaphase chromosomes and interphase nuclei from 20 cytogenetically normal subjects. An example is shown in fig 2A-C. In metaphase cells (fig 2A, B), there is a slight difference in the size of the signal on the two chromosomes 15 with both probes. However, in the interphase nucleus, where the DNA is less condensed (fig 2C), several distinct signals for each probe are clearly present. In one chromosome 15 domain, there is one green signal and one red signal, while the second domain has one green or possibly two fused green signals and two red signals. We assume that each signal represents one copy of the sequence but note that any contiguous duplication of sequence would not be detected. The additional signals on two D group chromosomes in fig 2A represent cross hybridisation of the IgH D cosmid to related sequences on 14q32, the site of the functional IgH D gene locus.11 This cross hybridisation can also be seen in interphase nuclei (fig 2C). In the nucleus in fig 2C, the copy number for the IgH D probe on each chromosome 15 is (1, 2) with a similar result for the NF1 probe.
The number of signals per chromosome 15 domain for each probe was analysed in 30 G1 nuclei from each control. Nuclei were classified as G1 nuclei both by their size and by the absence of signal doublets from the additional IgH D hybridisation sites on 14q32, which indicates replication status. We used the modal number of signals as a measure of the copy number of each probe because of the known variation in signal number in interphase nuclei caused by overlapping signals, split signals, or variation in hybridisation efficiency.22
These results are summarised in table 1. We found that the copy number for NF1 varies from one to four signals; IgH D generally has fewer copies and varies from one to three signals. These results show that there is variation in copy number of both probes in our control population and provide base levels for further studies.
Copy number in subjects with a proximal duplication
To test the hypothesis that these sequences are amplified in proximal duplication patients, we hybridised both probes to chromosome preparations from the eight proximal duplication cases, A-H. An example is shown in fig 2D, E. On metaphase chromosomes (fig 2D), both the NF1 and IgH D probes give large or multiple FISH signals in the pericentromeric region of one chromosome 15, the region that appears duplicated by conventional G banding analysis. In the interphase nucleus (fig 2E), the difference between the two chromosome 15 domains is striking: in one domain there are 10 green signals from the NF1 probe and seven red signals from IgH D, in the other domain there are two NF1 signals and one IgH D signal. Although the signals sometimes colocalise to give yellow spots, there are also separate green and red signals with intervening unlabelled chromatin, suggesting that other sequences in this region may also be amplified.
Table 2 summarises the copy numbers per chromosome for the proximal duplication cases and any available family members. In each case, one chromosome 15 domain clearly has a higher number of copies of the NF1 probe than the second domain. The copy numbers for NF1 on the duplicated chromosomes (5-10) were higher than those in the control population (1-4) (Mann-Whitney test, p<0.0001). In six subjects, IgH D is also increased in copy number on the duplicated chromosome; the modal copy number for IgH D ranges from 4-8, again higher than the copy number of 1-3 seen in the control population (Mann-Whitney test, p<0.0001). In all six cases, there are fewer copies of IgH D than NF1, consistent with the controls.
Surprisingly, in two subjects (cases G and H) the copy number for NF1 was increased but the copy number for IgH D was within the range for normal subjects (fig 2F, G). Thus, the apparent duplication seen by routine cytogenetic analysis in these two cases is associated with polymorphic amplification of only part of the pseudogene cassette.
Inheritance of proximal duplication chromosomes
To determine the inheritance of the proximal duplication chromosome, we obtained samples from additional family members for five cases and analysed them as described previously. The results are summarised in table 2. In three subjects (cases B, G, and H), the phenotypically normal mothers also carried similar proximal duplication chromosomes. One subject (case C) inherited the proximal duplication chromosome from the phenotypically normal father. Although FISH studies were not done on the parents, previous cytogenetic analysis had clearly shown that case A inherited a proximal duplication chromosome from her phenotypically normal mother. Case F has inherited a chromosome 15 with one copy of each probe maternally, but the second chromosome appears to have a slightly higher modal copy number for both NF1 (5) and IgH D (4) probes than either paternal chromosome 15, although the ranges clearly overlap.
Three probands have an abnormal phenotype. Two of these inherited the duplication chromosome from their phenotypically normal father and one from the phenotypically normal mother. Thus, no evidence of parent of origin effects was observed in the population studied here.
BCL8A pseudogene copy number is increased in proximal duplication chromosomes
BCL8A sequences were previously mapped close to the centromere by STS mapping of the chromosome 15 YAC contig.23 Two PAC clones were available for mapping to PWS/AS deletions by FISH: P406K8 contains the entire BCL8A sequence, while P398C20 lacks the 3` end. Neither clone contained NF1 or GABRA5 pseudogene sequences by PCR. Both clones mapped by FISH to the 15q pericentromeric region and gave weak cross hybridisation to the pericentromeric regions of other chromosomes (P406K8 to 13q12, the location of the functional gene; P398C20 to 14q11.2, 18p11.2, and 21q11.2). In interphase nuclei from controls, each PAC gave a cluster of two to three signals close to the pMC15 (D15Z3) signal on chromosome 15. This pattern suggested that the BCL8A sequence was another member of the cassette of sequences amplified in proximal duplications. To confirm this, hybridisation to metaphases from a class I deletion patient showed that the BCL8A PAC maps proximal to the PWS/AS critical region (fig 3A). Hybridisation to metaphase chromosomes from a proximal duplication patient showed large FISH signals on the duplicated chromosome in metaphase (fig 3B) and multiple signals in interphase (fig 3C). The copy number for BCL8A probes was comparable to NF1 and higher than IgH D. Additionally, in case G, the proximal duplication where IgH D was not increased in copy number, BCL8A probes were coamplified with NF1.
Orientation of the amplification cassette
We performed three colour FISH on interphase nuclei to determine the order of three of the pseudogenes, NF1, IgH D, and BCL8A, on chromosome 15. As the truncated GABRA5 sequence is within 80 kb of NF1,6 it was not included because it would be unlikely to give a separate signal in interphase nuclei.24 To simplify the analysis, we used preparations from a subject who has a single copy of NF1 and IgH D on one chromosome 15 (control 16 in table 1). NF1 and IgH D were ordered with respect to a centromeric alphoid probe, pMC15 (D15Z3), and to a distal probe, GS5022 (D15S18), which maps between BP1 and BP2 (fig 1). BCL8A and NF1 were ordered with respect to GS5022 only. Fig 4A shows the results of hybridisation of GS5022, NF1, and IgH D in an interphase nucleus. In the single copy domain (arrowed), the order is IgH D-NF1-GS5022, which indicates that IgH D lies closer to the centromere than NF1. This was confirmed by a second hybridisation with pMC15 (D15Z3), where the order was pMC15-IgH D-NF1. Although signals from IgH D and NF1 overlapped on metaphase chromosomes, IgH D was consistently more centromeric whenever order could be scored. BCL8A was also present as a single copy in this domain and was found to map distal to NF1. Thus, the most likely order for these probes is cen-IgH D-(NF1/GABRA5)-BCL8A-BP1-GS5022-tel.
Size of the amplified region
We estimated the size of the amplified cassette from interphase distance measurements. If every probe within the cassette gave two signals in one domain, we assumed that this represented a duplication of the amplification cassette. In a tandem duplication, the distances between the two signals from each probe should be similar and reflect the size of the cassette. In contrast, in an inverted duplication, the distance should vary between probes, depending on the position of the probe within the cassette. We hybridised probes for IgH D, NF1, BCL8A, and GABRA5 pairwise to interphase nuclei from a subject with two copies of IgH D and NF1 in one domain (control 15 from table 1) and measured the interphase distance between the two copies of each probe. A representative result is shown in fig 4B. Table 3 shows the mean interphase distance and estimated genomic distance for each probe. Estimates of the size of the amplification cassette from each probe were similar (800 kb to 1.1 Mb) and suggest a tandem, rather than an inverted, duplication. Note that the NF1 and GABRA5 probes, which map within 80 kb of each other, give very similar estimates of genomic distance.
Our results indicate that the pericentromeric region of chromosome 15q, between the centromere and BP1, contains truncated or non-functional copies of genes transposed from other pericentromeric regions (NF1, 17q11.2 and BCL8A, 13q11), from a telomeric region (IgH D and IgH V, 14q32.3) and from a more distal region of the same chromosome (GABRA5, 15q12). Pericentromeric regions of human chromosomes have acquired duplicated gene segments from elsewhere in the genome by a process referred to as pericentromeric directed gene duplication10, 25–27 and the centromeric region of 15q is a prime example.
Detailed analyses of pericentromeric regions have previously been performed on chromosomes 10, 16, and 21.28–30 In all three cases a similar pattern of complex juxtaposition of arm specific sequences, stable duplications, and unstable sequences with homologies to telomeric and centromeric sequences have been observed. Additionally, amplification of a pseudogene cassette has been observed in the pericentromeric region for chromosome 16 without phenotypic effects, similar to the amplification observed on chromosome 15.31
Our interphase FISH results indicate the following order for sequences in proximal 15q: cen-IgH D-(NF1/GABRA5)-BCL8A-BP1-tel. In silico analysis of data from the UCSC Golden Path (http://genome.cse.ucsc.edu) for 15q11 is consistent with this order (data not shown). However our results were obtained from one subject and there may be polymorphism in the population. The four pseudogenes are part of a large cassette or amplicon that varies in copy number in subjects with a normal karyotype and is increased in copy number in those with a visible proximal duplication of chromosome 15q. From interphase distance measurements, we estimated the size of this cassette to be ∼1 Mb, a result comparable to an earlier estimate derived from the number of copies found in a subject with a cytogenetic band sized duplication.6 We have also shown that the pseudogene cassette is duplicated in a tandem, or direct, orientation.
Our data show that the copy number of the cassette has a polymorphic distribution in the normal population, with 3- to 4-fold differences in copy number for the NF1 and IgH D probes between chromosome 15 homologues. This observation was reflected in differences in signal size and intensity on metaphase chromosomes and may contribute to subtle differences in banding patterns in the 15q proximal region. At one end of this distribution are subjects with a cytogenetically visible proximal duplication where one 15 homologue has copy numbers of 6-10 for several probes.
We analysed a larger number of proximal duplication subjects than previous studies and found that all eight proximal duplications involved significant amplification of NF1 pseudogenes, while only some also involved coamplification of IgH D. In both controls and subjects with proximal duplications, we found that the copy number for IgH D was the same or less than that of NF1. These two observations suggest that the length of the amplification cassette is variable and sometimes does not include IgH D.
For six of the subjects with proximal duplications, we obtained parental material and were able to identify the parental origin of the duplication chromosome by cytogenetic and/or FISH analysis. We observed both maternal and paternal transmission and found that the copy number of each probe was similar from parent to offspring. In one case we found a higher modal copy number of both probes than in either of the parents, and while this may reflect differences in hybridisation or overlap of signals, it may also represent an increase in copy number.
Five of the subjects with proximal duplications had a normal phenotype and were ascertained by the presence of a proximal duplication during routine cytogenetic investigations. Three subjects were ascertained on the basis of an abnormal phenotype (autism or mental retardation) and subsequent cytogenetic analysis indicated the presence of a proximal duplication. Does a proximal duplication of chromosome 15 have any phenotypic effect? These three cases have some phenotypic overlap (autism and mental retardation) but additional phenotypes (talipes, hernia, short stature, epilepsy) as well as developmental delay have been reported in earlier publications.5, 6 Maternal and paternal transmission from phenotypically normal parents was also reported. We suggest that there is no evidence of parent of origin effects, unlike the maternally derived duplication of the PWS/AS critical region which is associated with autism.2, 32 The link between autism and chromosome 15 duplications may lead to an increased ascertainment rate of subjects with a proximal polymorphism owing to the close scrutiny of subtle differences between homologues. Consistent with this is the fact that three of our group II cases have less total amplification in proximal 15q than group I cases: two cases were not amplified for IgH D sequences and the third has fewer copies of the amplified sequences than other proximal duplications with a normal phenotype. We therefore conclude that the presence of a proximal duplication is unlikely to cause any phenotypic consequences.
The new BAC and PAC clones identified in this study will allow a more accurate definition of the proximal breakpoints of deletions and marker chromosomes, which may contribute to our understanding of the underlying mechanisms of the frequent chromosome 15 rearrangements. For clinical testing purposes, it is recommended that any person with a cytogenetically visible proximal 15q duplication, which does not include the probes within the PWS/AS critical region, be analysed by FISH for the presence of an amplification of the pseudogene cassette. As NF1 is amplified in all reported proximal duplications, P1-4 appears to be the single most effective FISH probe for detecting proximal duplications in metaphase or interphase nuclei. The use of P1-4 will verify the amplification of these sequences as the cause of the cytogenetic variation observed by G banding. Although no consistent phenotype has been observed in subjects with an increased copy number of these pseudogenes, further studies will help to define genotype/phenotype correlations.
We thank Drs Mark Pasquarette, Sau W Cheung, J T Mascarello, Arturo Anguiano, Ed Cook, Osama Kamal Zaki, and Bennett Leventhal for providing patient material and clinical information. We thank Drs David Viskochil and Tasuku Honjo for providing genomic clones. This work was supported by NIH grants HD36111 (DHL) and CA72669 (RSKC).