Hereditary paraganglioma (PGL) is characterised by slow growing, vascular tumours that can develop in any component of the paraganglia, a neuro-ectodermal system that is distributed from the skull base to the pelvic floor.¹ Common tumour sites include the carotid body in the head and neck and adrenal and extra-adrenal paraganglia in the abdomen. Heterozygous germline inactivating mutations in SDHD, SDHC, and SDHB, which encode three of the four subunits of mitochondrial complex II (succinate dehydrogenase), cause hereditary paraganglioma types 1, 3, and 4 (PGL1, PGL3, and PGL4), respectively. Mutations in the fourth subunit of mitochondrial complex II, SDHA, have yet to be demonstrated in hereditary paraganglioma. Germline loss of function mutations followed by somatic loss of non-mutant alleles in the tumours^2–4 suggests a tumour suppressor role for mitochondrial complex II in the paraganglia.

Over 25 mutations in SDHD and 25 mutations in SDHB have been detected in hereditary paraganglioma, including those reviewed by Baysal¹ and the more recent additions of multiple mutations in SDHB^4–6 and SDHD.^7–9 All reported mutations are single nucleotide alterations leading to splice site, missense, nonsense, or frameshift mutations, or intra-exonic deletions and insertions of up to four nucleotides, which have been detected through exonic PCR amplifications and sequencing. In contrast to the abundance mutations in SDHB and SDHD, only a single multiply affected family and an isolated case, containing a single nucleotide initiation codon and a splice site mutations in SDHC, respectively, have been described by Niemann et al.^3,10 However, analyses of SDHC in four series of patients with paraganglioma or pheochromocytoma ^6,8,11,12 yielded no definitive SDHC mutations. These findings indicate that the relative contribution of complex II subunit mutations to hereditary paraganglioma is not similar and may reflect currently unrecognised aspects of complex II biology. Hence, it is of utmost importance that role of SDHC in familial and sporadic paragangliomas be confirmed independently.

Penetrance of complex II mutations shows peculiar characteristics. Mutations in SDHD cause PGL1 only if the transmission occurs paternally, whereas maternal transmission does not cause disease,¹³ suggesting operation of genomic imprinting on SDHD. In contrast, SDHB mutations are transmitted both paternally¹¹ and maternally.¹⁴ Thus far, transmissions of SDHC mutations causing disease occurred through mothers in the one multiplex family¹⁴ and in one isolated case¹⁰. Because the molecular basis of the parent of origin effects in PGL1 is unknown, it is unclear whether transmissions of mutations in SDHC, the protein product of which couples with that of SDHD and forms the membrane spanning domain of mitochondrial complex II, also shows any parent of origin effects.

METHODS

The family and the isolated cases were ascertained from two US otolaryngology clinics (University of Pittsburgh School of Medicine, Pittsburgh, PA and House Ear Institute, Los Angeles, CA) under research protocols approved by the University Institutional Review Board committee. DNA isolation, genotyping of simple tandem repeat polymorphisms, PCR amplification and sequencing were performed using standard techniques² and all simple tandem repeat polymorphisms were amplified in the presence of 10% glycerol and 5% DMSO after labelling one oligonucleotide primer with ³²P and analysed on a 6% polyacrylamide gel. DNA from the multiplex family was isolated either directly from white blood cells or from transformed lymphoblastoid cell lines. DNA of the sporadic cases was isolated from cheek swabs as described earlier.¹¹

The separation of chromosome 1 for SDHC mutation analysis in somatic cell hybrids was performed commercially (GMP Genetics, Waltham, MA), using a conversion approach that employs fusion between human and rodent cells to create stable hybrids that contain only a subset of the human chromosomes. This approach significantly increases the sensitivity to detect unconventional mutations that could be missed by techniques based on PCR because the parental copies of a given chromosomal pair can be separated and tested individually.¹⁵ An EBV transformed lymphoblastoid cell line derived from an affected individual (4-2, fig 1) was used for chromosomal separation by the conversion approach. This subject was chosen for chromosome separation because he is affected and an obligate carrier of the familial mutation (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

 A simplified pedigree of the multiplex family with SDHC deletion. Differently patterned long vertical bars symbolise individual haplotypes spanning the SDHC gene. Tested simple tandem repeat polymorphisms around the SDHC gene are shown in a box and the alleles (in bp) segregating in the family are depicted for subjects 1, 2, and 3. A haplotype of a married and deceased subject (5) was inferred and is shown in parenthesis. A diagram depicting the SDHC genomic structure, its relative position to two flanking simple tandem repeat polymorphisms (STRP) (D1S484 and D1S2844) and the location of the deletion identified in this study is shown on the right. D1S484 is located ≈500 kb centromeric to SDHC. Pathology records for subject 9 (shown by an asterisk) and for an unrelated sporadic subject with an identical SDHC deletion were available. R, a recombination event in subject 15 between D1S2844 and D1S196. NC, non-carrier of the deletion spanning SDHC exon 6.

Three independent somatic cell hybrids for the disease chromosome 1 and three independent somatic cell hybrids for the normal chromosome 1 were obtained from subject 4-2 and tested by sequence tagged site analysis. In addition to the SDHC exon 6,¹¹ the following PCR primer pairs located near SDHC exon 6 amplified a product at the expected size (given in parenthesis) from the hybrids containing the normal chromosome 1 but did not amplify from the hybrids containing the disease chromosome 1: (1) 120F: 5′-TTGGATCGCCCCTGGGCT-3′ and 120R: 5′-AACAAGCATAGCTCTCAGGGT-3′ (512 bp): the amplicon is located ≈5 kb downstream of exon 6; (2) 123F: 5′-TTGTCTTCAGTTGGTATGCCT-3′ and 123R: 5′-GAAGATTTCTGGAAGGAGACAC-3′ (243 bp): the amplicon is located ≈7 kb downstream of exon 6. In addition to the SDHC exon 5,¹¹ the following PCR primer pairs located near SDHC exon 6 amplified a product of the expected size (in parenthesis) from the hybrids containing both the disease and the normal chromosomes 1: (1) 124F: 5′-GAACTAATTGATTGAACTAGTAG-3′ and 124R: 5′-CCATGTTTAACCTACAGCTTAAC-3′ (236 bp): the amplicon is located ≈8.5 kb downstream of exon 6; (2) 110FB: 5′-GGAGAAAATATATGTTTTTTAATGAAG-3′ and 110R: 5′-CAGTCAATCTCAGAATCTTT-3′ (283 bp): the amplicon is located ≈1.3 kb upstream of exon 6. On the basis of these findings, we attempted to PCR amplify the deleted allele using oligonucleotide primer pairs that are too far apart to amplify the normal genomic DNA by standard PCR.

The PCR primers, 112F; 5′-CCTTTAGAATACTAGTCCTCTGA-3′ and 124R: 5′-CCATGTTTAACCTACAGCTTAAC-3′, which are located 9405 bp apart (including primer binding sequences) in the normal genomic sequence (GenBank accession number AL592295) and span SDHC exon 6, captured the deletion junction in a 1037 bp fragment (including a four base pair non-templated junctional insertion). The PCR amplification was performed for 38 cycles after an initial 10 min denaturation at 94°C. Each cycle was composed of 45 s incubation at 94°C, 45 s incubation at an annealing temperature of 54°C and 2.5 min incubation at 72°C for extension. The reaction was terminated with a final 7 min of extension period. AmpliTaq Gold Taq polymerase enzyme was used in all PCR amplifications.

The degree of allelic loss was assessed by comparison of the intensities of parental alleles between the tumour sample and peripheral blood as described.¹⁶ Allele intensities were quantified after densitometric scanning of the x ray autoradiograms using VIDEK Harmony Bioscan Software (v 4.03, Aldus). We first calculated the ratio of alleles, RA = (normal allele/disease allele) in blood (RA_b) and in tumour (RA_t). We then calculated the percentage loss of normal alleles (PLNA) in the tumour by PLNA = 100×[1−(RA_t/RA_b)]. For example, if there is no relative loss of normal allele in the tumour (that is, RA_t = RA_b), then the percentage loss of normal alleles equals 0%. If there is a complete loss of normal allele in the tumour, then the percentage loss of normal alleles equals 100%.

The following web based resources were used in this study:

• RepeatMasker program at http://searchlauncher.bcm.tmc.edu/seq-util/seq-util.html;

• Blast 2 program at www.ncbi.nlm.nih.gov/BLAST/;

• UCSC genome database at www.genome.ucsc.edu/, which contains the full genomic sequence of the SDHC gene.

RESULTS AND DISCUSSION

Previously, we reported the presence of germline mutations in SDHB and SDHD in 70% (7/10) of familial and ≈8% (3/37) of non-familial clinic patients with head and neck paragangliomas but no mutations could be identified in the SDHC gene.¹¹ Here, we report discovery of a SDHC mutation in one of the remaining three families, family 4.^11,16 We have evaluated family 4 by recruiting additional members (fig 1). The detailed phenotype of the affected subjects from this family is listed in table 1. The information on phenotype of most affected subjects was obtained through their medical histories, although pathology reports, confirming diagnoses of paragangliomas, were also available for two subjects. Both paternal and maternal disease transmissions were observed in the pedigree suggesting that SDHD was not the underlying locus. No subject reported having been diagnosed with abdominal or metastatic paragangliomas, which may be associated with SDHB mutations.^4,14,17 We reasoned that either a mutation in a new hereditary paraganglioma gene or an unconventional mutation in the known hereditary paraganglioma genes was responsible for the disease.

To test for cosegregation of alleles among the affected subjects we genotyped simple tandem repeat polymorphisms near SDHB, SDHC, and SDHD genes. We found that haplotypes defined by simple tandem repeat polymorphisms near SDHB and SDHD genes did not cosegregate among the affected subjects, further excluding the role of these genes in this family (data not shown). However, multimarker haplotypes spanning SDHC were consistently shared among the five affected subjects (fig 1). Haplotype analysis further suggested the presence of two more mutation carriers, including an obligate carrier mother and a subject who died of metastatic breast cancer (fig 1). Neither carrier was clinically diagnosed with hereditary paraganglioma (table 1). The obligate carrier mother reported undergoing carotid artery surgery following an episode of stroke but denied ever been diagnosed with a head and neck tumour. The phenotypic information for both carriers was obtained through interviews; their medical records were not available for detailed investigation.

Because simple tandem repeat polymorphisms flanking SDHC were consistent with linkage, we hypothesised that a genomic rearrangement that escapes detection by PCR amplification of the exons might be responsible for PGL3 in this family. An amplification of SDHC cDNA prepared from lymphoblastoid total RNA from an affected subject by RT-PCR, using expressed oligonucleotide primers from exon 1 and exon 6,¹¹ revealed only the expected normal transcript size without any evidence of an aberrant band. To obtain direct evidence for gross gene alterations, we pursued separation of the disease and non-disease chromosomes to perform sequence tagged site content analysis. The SDHC gene is localised at the long arm of chromosome 1 at band q23.3 at the UCSC genome database, which is far more distal than was reported earlier.¹⁸ Following the separation of the two parental chromosomes 1 in somatic cell hybrids, we conducted sequence tagged site content analysis using three stable hybrids for each parental chromosomes 1 derived from subject 4-2 (fig 1). We found that oligonucleotide PCR primer pairs that span SDHC exon 6 did not amplify from any of the three hybrids containing the disease chromosome, suggesting a deletion spanning the 3′-end of the gene. Further sequence tagged site content mapping around exon 6 confirmed the deletion and enabled us to capture a junctional PCR fragment which amplified from the five affected and the two carrier subjects (fig 2), as predicted by haplotype analysis, but not from the other unaffected family members and 103 additional control subjects. The deletion was heterozygous in all six carriers since the normal copy of SDHC exon 6 could be amplified from constitutional genomic DNA.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

 Genomic amplification of the deletion junction in the members of the multiplex family and in the isolated case as a 1037 bp fragment. The numbers above the lanes correspond to the subjects in the pedigree (fig 1).

Sequence analysis of the junctional fragment indicated that an 8372 bp genomic fragment spanning exon 6 was deleted (fig 3). Both breakpoints mapped within AluY elements, which are normally located 8.25 kb apart on the genomic sequence in identical orientations. An alignment by the “Blast 2” program indicated an 84% sequence identity between the two AluY elements. We also detected a non-templated insertion of four nucleotides at the deletion junction (that is, filler DNA). Filler DNAs are random insertions at the breakpoint junctions of constitutional and somatic chromosomal rearrangements and are incorporated through a variety of mechanisms including non-homologous end repair.¹⁹ The adjoining of two similar AluY elements by the deletion and the identification of a 4 bp insertion at the junction suggests that a homologous Alu-Alu recombination between two highly similar elements, that occurred intrachromosomally or interchromosomally, and an ensuing non-homologous end repair was responsible for this genomic rearrangement. The deletion of exon 6 is predicted to remove the third transmembrane spanning domain of the SDHC protein product, cybL, as well as the RNA polyadenylation and termination signals.¹⁸

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

 The upper panel shows genomic structure and high repeat element content of the SDHC gene (data from USCS database). Vertical bars in the SDHC genomic sequence indicate six exons of the gene. The deletion occurred between two AluY elements (circled) that span exon 6. The two AluY elements have the same genomic orientation (arrows). Locations of two (CA/GT)n repeat elements, SDHC-CA-2 and SDHC-CA-3, used in haplotype and allelic imbalance analyses are indicated by arrows (table 2). The lower panel shows nucleotide composition around the deletion breakpoints and the sequence chromatogram of the deletion junction. The sequence of the telomeric AluY element is underlined. Four nucleotides shown in capitals at the deletion junction were non-templated insertions (filler DNA). SINE, short interspersed repeat elements, including Alus; LINE, long interspersed repeat elements; LTR, long terminal repeats.

Availability of paraffinated tumours from individual 9 allowed us to test for allelic imbalance near the SDHC gene. We tested four simple tandem repeat polymorphisms, SDHC-tetra, SDHC-CA-2, D1S484, and SDHC-CA-3, and found losses of normal alleles in the tumour of 66%, 30%, 38%, and 73%, respectively (fig 4). The lost alleles were located on the non-disease chromosome for each marker, suggesting that the normal copy of SDHC is somatically lost during tumour genesis. These results are in accord with those of Niemann and Muller³ and suggest that SDHC gene is subject to two hit inactivation consistent with a tumour suppressor role in human paraganglionic tissue.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

 Partial somatic losses of normal alleles in hereditary paraganglioma tumour from subject 4-9 for four simple tandem repeat polymorphisms near the SDHC gene. Comparison of relative intensities of the normal alleles between the constitutional (C) and tumour (T) DNAs from subject 4-9 (in lanes denoted by an asterisk) demonstrates partial losses of the normal alleles in the tumour for all markers (denoted by arrowheads). Genotypes of certain pedigree members (fig 1) are included to demonstrate the normal and disease alleles segregating in the pedigree. Note that affected subject 4-15 did not share SDHC-CA-2 disease allele (141 bp) with the other affected individuals in the pedigree. This was most probably caused by a spontaneous mutation of the STRP allele because subject 4-15 shared the disease alleles of other simple tandem repeat polymorphisms (fig 1) and was a carrier of the SDHC deletion (fig 2). Sizes (in base pairs) of the segregating alleles (denoted by filled arrowheads) are indicated for each marker.

To test whether this deletion might be present among isolated cases with solitary head and neck paragangliomas who previously failed to reveal any mutations in the SDHB, SDHC, and SDHD genes,¹¹ we conducted PCR analyses in 31 cases. We observed one subject, whose constitutional DNA from cheek swab and blood amplified a fragment at a similar size to that of the familial cases. Direct sequencing of the junctional PCR fragment from the isolated case revealed the same sequence found in the family, including the four base pair filler DNA insertion at the junction. To further assess the possibility of a common origin, we genotyped three previously described and four novel STRP markers that we characterised near the SDHC gene (fig 1 and table 2) and found that the sporadic case carried all the alleles located on the disease haplotype in the multiplex family. The tested simple tandem repeat polymorphisms, the shared disease alleles and their frequencies were as follows: D1S2635: 151 bp (0.160), D1S2707: 149 bp (0.130), D1S484: 142 bp (0.107), SDHC-CA-2: 141 bp (0.053), SDHC-CA-3: 211 bp (0.50), SDHC-3′-STRP: 151 bp (0.44), and SDHC-tetra: 207 bp (0.143).

The identity of the deletion junctions and the sharing of multiple uncommon alleles of simple tandem repeat polymorphisms distributed over a 4 cM distance strongly suggest that the family and the isolated case inherited the SDHC deletion from a common ancestor, who was not apparent by investigation of both extended pedigrees. Geographical proximity of residential locations of the familial and sporadic cases (both from the northeastern United States) further supports this conclusion. Multiple founder mutations have been detected for SDHD, but not for SDHB.^11,20,21 Whether the SDHC deletion is a founder mutation and might be responsible for additional HNP cases in the United States or elsewhere will be appreciated better by further testing of additional cases. Nevertheless, our findings clearly indicate that SDHC mutations contribute to the aetiology of familial and sporadic head and neck paragangliomas. Cosegregation of the SDHC deletion with head and neck paragangliomas in five affected subjects in the multiplex family and in one unrelated sporadic case provides the strongest association between a SDHC mutation and head and neck paragangliomas so far. For comparison, the initiation codon mutation described in the original PGL3 family was detected in only five subjects with head and neck paragangliomas.³

SUMMARY

Current data further expand the role of complex II mutations in the aetiology of paragangliomas and together with our previous findings¹¹ suggest that SDHD, SDHB, and SDHC germline mutations contribute to 50%, 20%, and 10% of the familial and ≈5%, ≈3%, and ≈3% of the sporadic cases with head and neck paragangliomas, respectively. To our knowledge, the SDHC deletion described in here is the first large gene deletion identified in a hereditary paraganglioma gene. The discovery of a large deletion that was missed by conventional mutation screening methods may warrant inclusion of screening methods for identifying large deletions. A repeat content analysis of the genomic sequence of SDHC, which spans 48 809 bp from the transcription start site to the end of exon 6 in the USCS database, by the RepeatMasker program blocks 42.31% of the sequence as repeat elements and 37% as Alu elements (fig 3). This density of Alu elements is comparable to that of BRCA1,²² which has 41.5% Alu content and which undergoes recurrent genomic deletions.²³ It is also well known that other genes with a high Alu repeat content, such as those encoding the LDL receptor and α-globin, are prone to recurrent genomic deletions through Alu-Alu recombination events.²³ Thus, on the basis of its high Alu content, our discovery of an Alu mediated deletion and the failure to find conventional mutations in several paraganglioma series, the SDHC gene might be considered further for genomic deletions in familial and sporadic paragangliomas. Because the SDHC deletion caused tumours after both paternal and maternal transmissions, it is unlikely that an absolute parent of origin effect operates at the SDHC locus. This observation, together with the earlier results on the transmission of SDHB mutations, strongly suggests that the parent of origin effect on the SDHD gene is not a functional consequence of complex II mutations but a locus specific epigenetic phenomenon operating exclusively on the SDHD gene at chromosomal region 11q23.