Editor—Partial loss of chromosome 18q (MIM 601808) results in characteristic clinical features including mental retardation, short stature, developmental delay, CNS defects, dysmorphic facies, and hearing loss. By phenotype mapping in 26 patients, Strathdee et al 1showed that a region critical for many of the 18q− features lies in 18q22-23. Later, they were able to refine this region to an approximate 6 Mb segment within 18q23.2 However, the clinical picture of 18q− patients is extremely variable, rendering a precise prediction of the clinical outcome impossible, even when the extent of the deletion is determined.1 3 4 Among the factors contributing to this phenotypic variability, the genetic background of affected patients, environmental factors, and possibly genomic imprinting of genes in 18q may play a role. The selective expression of the paternal or maternal allele of a gene responsible for a phenotypic feature5-7 might influence the phenotype of 18q− patients, depending on whether the mutation arose in the maternal or paternal germline. Most of the 18q− patients have a paternal deletion.8 Assuming deletions originate with the same frequency in the maternal and paternal germline, imprinting of maternal genes could explain a more severe phenotype in patients with a paternal deletion, leading to a higher detection rate.

Evidence for imprinting in 18q came from linkage studies of bipolar affective disorder (BPAD). While several reports have shown linkage to 18q,9-15 in some studies most of the linkage evidence derived from families with affected phenotypes in only the paternal lineage and from marker alleles in 18q11 and 18q21 transmitted on the paternal chromosome.9 10 13 16 Genomic imprinting might explain the uniparental linkage, if critical maternal genes are imprinted (inactivated) and thus inheritance of a BPAD gene predisposes to the illness only if it is inherited from the father. Hence, depending on the exact location, imprinted genes in 18q would be strong candidates for the 18q− syndrome and for BPAD.

Encouraged by these observations, we started a PCR based screen in distal 18q to test for imprinting by analysing 22 expressed sequence tagged sites (ESTs) within 18q22-23. The ESTs consisted of unidentified as well as identified transcripts, including the myelin basic protein (MBP, MIM 159430), the galanin receptor (GALNR, MIM 600377), cerebellin 2 (CBLN2, MIM 600433), cytochrome b5, and the nuclear factor of activated T cells (NF-ATc, MIM 600489). The ESTs were chosen mainly because of their location within the 18q− critical region and their expression in brain which is severely affected in the 18q− syndrome.17-19 The strategy is similar to the one described by Wevrick and Francke,20 who showed that transcription of the small nuclear ribonucleoprotein associated polypeptide N (SNRPN) in peripheral blood lymphocytes and Epstein-Barr virus (EBV) transformed lymphoblastoid cell lines of Prader-Willi syndrome patients is abolished. We tested for expression by PCR analysis of cDNA derived from lymphoblastoid cell lines of 18q− patients using specific EST primers that are only present in one copy in these patients.

EBV transformed cell lines derived from three patients with 18q deletions (patients 11, 17, and 18) have been previously described. The breakpoint of sample 11 is located proximal to D18S214 and D18S269 at approximately 80 cM, the breakpoint of sample 18 is located proximal to D18S165 and D18S252 at approximately 100 cM, and the breakpoint of sample 17 is located proximal to D18S299 between the two other breakpoints.1 An EBV transformed cell line from a subject with a normal karyotype was used as a control (C). A lymphoblastoid cell line derived from a Prader-Willi syndrome patient (GM09189) who has a deletion involving the SNRPN gene, del(15)(pter→ q11.2::q13→qter), was obtained from the Coriell Cell Repository.

Cytoplasmic RNA was prepared using a kit (RNeasy, Qiagen, Hilden) according to the manufacturer's instructions. Subsequently, RNA was treated with DNAse I (AGS, Heidelberg) to remove any residual genomic DNA. cDNA synthesis was performed by using the MMLV reverse transcriptase (Promega, Madison) according to the manufacturer's protocol.

ESTs that mapped in 18q22.2-qter were identified from the gene map/Unigene map database of the National Center for Biotechnology Information (NCBI), National Institutes of Health (http://www.ncbi.nlm.nih.gov/SCIENCE96/ andhttp://www.ncbi.nlm.nih.gov/UniGene/Hs.Home.html). EST primer sequences were obtained from the websites of the Radiation Hybrid Database, European Bioinformatics Institute, Hinxton, UK (www.EBI.ac.uk/RHdb/index.html), and of the Whitehead Institute Center for Genome Research, Boston, MA, USA (www.genome.wi.mit.edu/). Three additional sets of oligonucleotides were used for controls. MBP-PCR1/2 (5′-GGA CCT CGT GAA TTA CAA TC-3′ and 5′-ATT TAC CTA CCT GTT CAT CC-3′) amplifies a polymorphic region 5′ of the MBPgene that is not transcribed. SNRPN-A/B (5′-AGA TGG CCG AAT CTT CAT TG-3′ and 5′-AGC AAC ACC AGA CCC AAA AC-3′) amplifies a 150 bp segment of the SNRPN gene that is known to be maternally imprinted.20 21 Finally β-actin-A/B (5′-TCG TGC GTG ACA TTA AGG AG-3′ and 5′-AGC ACT GTG TTG GCG TAC AG-3′) are primers derived from exon 4 and exon 5 of the β-actin gene, respectively, and will amplify a 274 bp product only in RNA samples.

PCR analyses were performed in 1 × Q buffer, 1 × Q solution (Qiagen, Hilden), 0.1 mmol/l dNTPs (Pharmacia, Uppsala), 0.02 U/μl Platinum Taq DNA polymerase (Life Technologies, Rockville) in a total volume of 25 μl in a Perkin Elmer DNA Thermal Cycler (Perkin Elmer, Norwalk). Optimal primer concentrations and annealing temperatures were established before testing. Cycling was performed for 30 seconds at 95°C, 30 seconds at the appropriate annealing temperature, and 30 seconds at 72°C for 34 cycles. Amplification products were analysed on a 3% agarose gel.

Three cell lines containing 18q deletions were used for imprinting analysis (fig 1). The origin of the deletion had been previously determined using polymorphic markers.1 Samples 11 and 18 contain paternally derived rearrangements while sample 17 contains a maternally derived rearrangement. The hemizygous deletion in all three cell lines includes the region 18q22.1-qter.1 cDNA was prepared from RNA for each of the three cell lines as well as a cell line containing a normal karyotype. To ensure that the RNA samples were free of genomic DNA contamination, PCR analysis was first performed using primers for a region that is located 5′ of the MBP gene. Amplification was not observed in any of the samples but was observed in genomic DNA controls (data not shown), indicating the lack of DNA contamination in the RNA samples. To ensure that cDNA samples were not degraded, PCR was performed using primers derived from two exons of the β-actin gene. Amplification of the appropriate size product was seen with all cDNA samples (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Schematic drawing of distal human chromosome 18q. The truncated chromosomes of the patients in this study are shown at the top; the approximate map positions of the ESTs tested are indicated by dashed bars on the right.

Twenty two ESTs that mapped to 18q22.2-qter were analysed for imprinting. Table 1 lists the ESTs that were tested as well as the composite results. Nine ESTs did not produce amplification products in the sample derived from a subject with a normal karyotype. Fig 2A shows a representative result from EST A007H41. This showed that the genes from which these ESTs were derived were not expressed in lymphoblastoid cells. Appropriate amplification was observed when genomic DNA was used, showing that the lack of amplification was not the result of inappropriate PCR conditions.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Imprinting analysis of 18q genes. PCR analysis of cDNA derived from patients with maternal 18q deletions (patient 17), paternal 18q deletions (patient 11 and 18), or normal control (C) is shown. Amplification using genomic DNA from two normal subjects (A and AM) as well as a sample lacking any DNA input (neg) are shown. Size markers are included in each gel. (A) EST A007H41. (B) EST WI-9340. (C) Amplification using a cDNA derived from a patient with Prader-Willi syndrome (GM09189) and a paternal 15q deletion is shown. Amplification with two 18q ESTs (WI-6843 and H81050) are shown as well as amplification using SNRPN primers, a known maternally imprinted gene.

The remaining 13 ESTs all produced an amplification product in the sample derived from a subject with a normal karyotype. For all 13 ESTs, amplification was observed in the samples that contained either maternally or paternally derived deletions. A representative result from EST WI-9340 is shown in fig 2B. This showed that imprinting did not occur in lymphoblastoid cells for the 13 genes that were tested.

To ensure that imprinting could be detected, PCR analysis was performed using primers derived from the SNRPN gene which is known to be maternally imprinted. As shown in fig 2C, no amplification was observed in GM09189, a cell line derived from a Prader-Willi syndrome patient, which contains a paternally deleted chromosome 15. Since the remaining homologue is maternally imprinted,SNRPN is not expressed and therefore no amplification was observed. ESTs that mapped to 18q (WI-6843 andH81050) did amplify, showing that amplification could be obtained from the RNA sample.

In this study, 13 genes that mapped to the distal region of chromosome 18 were investigated to determine whether they might show an imprinting effect. All 13 genes were expressed from the single remaining allele, regardless of whether it was maternally or paternally derived. The basis for the experiments was that there is significant clinical variability in patients with 18q deletions. It is possible that imprinting may play a role in this variability.

A clear phenotypic distinction between subjects bearing a paternal and a maternal deletion was not previously possible.1 In addition, translocation mouse models did not show any evidence for imprinting in distal MMU18,22 23 which represents the syntenic region of 18q23. However, in the study by Strathdeeet al,1 only very limited numbers of patients were available and the extent of the deletion differed in almost all of them. Also, imprinting is not always conserved between mouse and man24 and there are at least three human genes that have been shown to be imprinted despite their location within a region excluded from imprinting by translocation studies in mice.23 25-27 This is probably because of difficulties in detecting more subtle phenotypic alterations in animal models.

The lack of detection of an imprinting effect can be interpreted in several ways. First, imprinting of certain genes in 18q does exist, but we did not test these genes. We do not know whether any of the genes tested are involved in the phenotype of the 18q− syndrome, although the galanin receptor, the myelin basic protein, cytochrome 5b, and NF-ATc have been proposed to be candidates.19 28 29In our study, 13 of the 22 ESTs distal to 100 cM were informative compared to about 36 genes or so found in this region, according to the NCBI Unigene map. Thus, we cannot exclude other imprinted genes in 18q22-23, though in many cases clusters of genes are imprinted and might therefore be easier to detect when testing a set of genes that are distributed over a chromosomal region. Second, imprinting of genes in 18q may be restricted to certain tissues or developmental stages. Since only RNA from lymphoblastoid cell lines was tested, we may have missed imprinting that occurred in other tissues. While local and temporal restrictions of imprinting are well known,30-36EBV transformation of lymphocytes does not seem to modify imprinting mechanisms, as has been shown forSNRPN,20 37 38 PAR-1 andPAR-5,38 andIPW.39 Thus, the method described here should at least be able to detect ubiquitous imprinting of genes and all imprinting phenomena that affect lymphocytes. Finally, it is possible that there is simply no imprinting in the region investigated. If this holds true, we have to conclude that mechanisms such as modifying genes outside 18q and environmental factors influence the phenotypic picture of 18q− patients. In this case, we may have to consider that the parent of origin effects in bipolar disorder linked to 18q represent statistical artefacts, an assumption consistent with observations by Durner and Abreu40 and McMahonet al,41 who did not observe a consistently paternal parent of origin effect.

Clinical heterogeneity is well known in chromosomal syndromes. The parental origin of the rearrangement has been reported to influence the phenotype in a number of cases. Among them are the paternal duplication of 11pter-p15.4 which results in Beckwith-Wiedemann syndrome (MIM 130650),42 and monosomy 15q11-q13 which results in Prader-Willi syndrome (MIM 176270) if the paternal chromosome is affected and in the clinically different Angelman syndrome (MIM 105830) if the maternal chromosome is affected.43 More subtle influences of imprinted genes on the phenotype might be involved in other chromosomal rearrangements44 like the deletion associated retinoblastoma, which displays slower tumour progression if the maternal allele is deleted,45 suggesting that imprinted genes close to RB might influence tumour growth. Kato et al 25showed that the serotonin receptor 2 (HTR2) gene, which is located within the rearranged chromosomal region in 13q14, is paternally imprinted. The authors speculate thatHTR2 is a gene promoting the growth of retinoblastoma and that tumour progression depends on whether its active or inactive copy is retained. Other examples are Williams syndrome patients with a deletion of 7q11.23 who display significantly more severe growth retardation and microcephaly if the arrangement is maternally derived,46 and Turner syndrome patients (45,X) who show significantly poorer verbal and higher order executive function skills when the retained X chromosome is of maternal origin.47 48

This study presents the initial analysis for identifying imprinted genes in 18q. Although this study did not identify any imprinted genes, the approach can easily be expanded to investigate additional genes. The ability to detect imprinted genes located on other chromosomes in lymphoblastoid cells shows that the approach used here is a viable one and should be continued as a method for investigating whether imprinting effects might be involved in the clinical variability of the 18q− syndrome.