Background
Succinate dehydrogenase (SDH) plays an integral role in the tricarboxylic acid (TCA) cycle, where it couples the oxidation of succinate to fumarate to ubiquinone reduction. SDH is comprised of four nuclear encoded subunits (SDHA, B, C and D), and germline mutation in any of these SDH subunits is associated with a variable risk of developing neoplasia [
1]. Nevertheless, discordant phenotypes are observed both within and between families carrying the same
SDH mutation, suggesting that other environmental, genetic or epigenetic factors influence the clinical phenotype.
SDH genes (
SDHA,
SDHB,
SDHC,
SDHD) act as classical tumor suppressors, such that germline heterozygous inactivating mutations coupled with somatic loss of the remaining wild-type allele leads to complete loss of enzyme function and development of associated tumors. SDHx mutations have been linked to tumorigenesis as a result of a number of downstream consequences. SDH deficiency results in succinate accumulation, due to inability of SDH to catalyze the oxidation of succinate to fumarate. In turn, the elevated succinate can inhibit α-ketoglutarate-dependent dioxygenases, resulting in pseudo-hypoxia and hypermethylation of histones and DNA. Inhibition of the α-ketoglutarate-dependent dioxygenase, prolyl hydroxylase (PHD), leads to HIF stabilization, increased expression of HIF targets, and ultimately induction of a hypoxic response under normoxic conditions (pseudo-hypoxia). In line with this proposed mechanism of action, both increased stability of HIF and increased expression of HIF targets have been identified in
SDHx-mutated paragangliomas and pheochromocytomas [
2‐
4]. Additionally, accumulated succinate inhibits other α-ketoglutarate-dependent dioxygenases, such as histone demethylases of the Jumonji demethylase family and TET hydroxylases, resulting in hypermethylation of histones and DNA. This mechanism has been observed in
SDHx-mutated paragangliomas, pheochromocytomas, and gastrointestinal stromal tumors [
5‐
8].
Recently, SDH assembly factors (SDHAF1, SDHAF2, SDHAF3 and SDHAF4) have been identified as being crucial for maturation and effective functioning of SDH within mitochondria [
9‐
11]. To date, mutations affecting
SDHAF1 (involved in maturation of SDHB) and
SDHAF2 (required for covalent attachment of FAD to SDHA) have been associated with human diseases.
SDHAF1 mutations have been identified in individuals with leukoencephalopathy [
12,
13], but not yet in subjects with paragangliomas or pheochromocytomas. A
SDHAF2 loss-of-function mutation (p.Gly78Arg) has been reported in two unrelated families with head and neck paragangliomas [
9,
14,
15]. In the tumors of affected individuals, this mutation was shown to impair flavinylation of SDHA. Additionally, in vitro experiments showed that the p.Gly78Arg mutant leads to complete loss of SDH activity, through impaired covalent flavinylation of SDHA and destabilization of the SDHAF2 protein. Subsequent studies in large cohorts of apparently sporadic paragangliomas and pheochromocytomas have failed to identify germline or somatic
SDHAF2 mutations, suggesting that mutations within
SDHAF2 may be rare [
14,
15].
Recently, yeast studies showed that two LYR motif proteins Sdh6 (SDHAF1, human ortholog) and Sdh7 (SDHAF3, human ortholog) act in concert to promote the maturation of Sdh2 (SDHB, human ortholog) by shielding one or more of the three Fe-S clusters in Sdh2 from the deleterious effects of oxidants during assembly [
10]. Mutations in the LYR motif of human
SDHAF1 were shown to attenuate interaction with iron-sulfur biogenesis components supporting a role for SDHAF1 in maturation of the holo-SDHB complex [
16,
17]. L(I)YR motifs were also observed in SDHB itself in residues 44–46 (IYR) and 240–242 (LYR). The second motif is close to the binding site for SDHAF1 [
17]. We therefore hypothesized that mutations within the newly identified SDH assembly factor, SDHAF3, may be associated with the pathogenesis of pheochromocytoma and/or paraganglioma syndromes. Furthermore, given SDHAF3 is involved in the maturation of SDHB, we hypothesized that mutations within either of these genes may impair this process.
Discussion
In this study, we have identified a variant in the SDH assembly factor 3 (SDHAF3, c.157 T > C [p.Phe53Leu]) that may be associated with increased prevalence of pheochromocytoma and/or paraganglioma (PC/PGL). Our studies in yeast have confirmed this to be a hypomorphic variant, leading to reduced SQR activity. Furthermore, our in vitro studies in human cells show that SDHAF3 interacts with SDHB (residues 46 and 242), and that interaction between SDHAF3 p.Phe53Leu and SDHB is impaired.
SDH plays an integral role in both the tricarboxylic acid cycle and electron transport chain. Germline mutations within any of its four subunits (SDHA, B, C and D) have been associated with development of a number of tumors, including pheochromocytoma and/or paraganglioma, gastrointestinal stromal tumors, renal cancer, and pituitary adenomas [
1]. Recently, SDH assembly factors (SDHAF1–4) have been identified as playing a role in maturation of individual SDH subunits and assembly of the functioning SDH complex as a whole [
27]. To date, loss-of-function mutations in
SDHAF1 (biallelic) and
SDHAF2 have been associated with infantile leukoencephalopathy [
12,
13] and head and neck paragangliomas [
9,
14,
15], respectively. More recently, the yeast orthologs for SDHAF1 (Sdh6) and SDHAF3 (Sdh7) were shown to shield the Fe-S clusters of the yeast ortholog for SDHB (Sdh2), thereby promoting maturation of SDHB [
10]. In human cells, SDHAF1 was shown to associate with iron-sulfur cluster biogenesis components suggesting a role for SDHAF1 in mediating Fe-S cluster insertion [
16,
17]. Taken together, we hypothesized that mutations within the newly identified SDH assembly factor,
SDHAF3, may be associated with the pathogenesis of pheochromocytoma and/or paraganglioma syndromes.
In this study, we identified a
SDHAF3 c.157 T > C (p.Phe53Leu) variant in familial and sporadic cases of PC/PGL, observing a minor allele frequency (MAF) of 0.0658. Although this variant (rs62624461) is reported with a MAF of 0.0209 in the Exome Aggregation Consortium (ExAC) database (
http://exac.broadinstitute.org/), reflecting exome variant data from 60,422 individuals, the prevalence is significantly higher in PC/PGL (6.6% versus 2.1%,
p = 0.022). This prompted us to perform additional studies, to clarify the role that the SDHAF3 p.Phe53Leu variant may play in the pathogenesis of PC/PGL. Through yeast studies we were able to show that introduction of the SDHAF3 p.Phe53Leu variant, into
Sdh7 null yeast (ortholog of
SDHAF3 in humans) resulted in impaired function, observed by its failure to fully restore SDH activity when expressed in
Sdh7 null yeast relative to wild-type (WT) SDHAF3. Taken together, these findings indicate that although SDHAF3 p.Phe53Leu is at best a very low penetrance allele for PC/PGL per se, it may play a modifying role as observed by its hypomorphic activity. Hypomorphic alleles of SDHAF3 may contribute to the pathology in SDH-deficient tumors with residual SDH subunits, or alternatively through a secondary unidentified function of SDHAF3. In this study, two PC/PGL tumors from patients harboring germline
SDHB (IVS3 splice-site) mutation and
SDHAF3 (c.157 T > C) variant showed loss of SDHB staining by immunohistochemistry. This raises the question of how
SDHAF3 c.157 T > C can play a role in PC/PGL tumorigenesis, in SDH-deficient tumors. Clearly, by the time that inactivation of both
SDHB alleles has occurred in the tumor,
SDHAF3 c.157 T > C presumably has no additional role, as SDHAF3 appears to interact specifically with SDHB. Nevertheless, we conjecture that the germline presence of this hypomorphic
SDHAF3 c.157 T > C allele may over time lead to instability of SDH. Further, as
SDHB is a known tumor suppressor and hence requires inactivation of both alleles for tumorigenesis, the timeframe between
SDHB germline (first hit) and somatic loss of the normal
SDHB allele (second hit) provides a means by which the
SDHAF3 c.157 T > C allele could act. Further research is needed to resolve this issue.
To further understand the role of SDHAF3, and the impact of p.Phe53Leu in greater detail, we assessed its ability to interact with SDHB. Our in vitro studies in human cells, confirmed previous findings in yeast, [
10] with wild-type SDHAF3 shown to interact with SDHB. Interestingly, this interaction was attenuated on introduction of the p.Phe53Leu variant. We wanted to assess SDHAF3-SDHB interaction further by introducing clinically relevant SDHB mutations. Interaction between wild-type SDHAF3 and SDHB p.Arg242His mutant was not observed, implicating this region of SDHB as a direct binding site for SDHAF3. Maio et al. (2015) recently demonstrated that SDHAF1 interacts with SDHB with contacts between SDHB residues 146–153, 183–185 and 198–202 [
17]. Thus, two LYR-motif proteins, SDHAF1 and SDHAF3, may associate with different contact sites on SDHB during maturation of SDHB with Fe-S clusters.
This finding is supported by observations in yeast, whereby Sdh6 and Sdh7 (orthologs of human SDHAF1 and SDHAF3, respectively) have been shown to interact with Sdh2 (ortholog of human SDHB) [
10], although the specific binding site(s) in yeast Sdh2 were not identified. Our study shows that SDHAF3, in fact, is a direct binding partner for the LYR motif of SDHB (p.240–242). Reduced interaction was observed in all other SDHB mutants, with the exception of p.Ile127Ser in which interaction between SDHAF3 and SDHB appeared to be unaffected, indicating that this residue has no bearing on SDHB interaction with SDHAF3.
On introduction of the SDHAF3 p.Phe53Leu variant, SDHAF3-SDHB interaction was completely lost for SDHB p.Arg46Gln and p.Arg46Gly mutants, implicating residue 46 (contained within an IYR binding site [p.44–46]) as another region of SDHB that may interact with SDHAF3. Although these two regions are spatially separated in two distinct sub-domains, SDHAF3 may in fact bind the L(I)YR motifs residues in each of these domains. Alternatively, the impaired binding may arise from secondary consequences of the p.Arg46 mutation. Interestingly, our previous structural modeling of these
SDHB mutations had not identified the functional impact on SDHB, as both glycine and glutamine are capable of fitting within the space left by arginine, and the electron path is not nearby [
25]. The findings of our current study suggest that mutations affecting residue 46 of SDHB are pathogenic via preventing maturation of SDHB. A similar effect was noted by Maio et al. (2014), whereby the SDHB p.Arg46Gln mutation did not impair SDHB interaction with HSC20, although reduced binding to the HSC20 complex and SDHA were noted, suggestive of an effect on formation of a mature SDH complex [
16].
Interestingly, introduction of the SDHAF3 p.Phe53Leu variant resulted in a stronger SDHAF3-SDHB interaction in the presence of the SDHB p.Cys101Tyr mutant. Since Cys101 is a ligand to the 2Fe-2S center in the N-terminal domain of SDHB, the enhanced interaction is suggestive that SDHAF3 interacts with apo-SDHB. Consistent with this prediction is the observation that overexpression of SDH2 in yeast lacking Sdh1 results in a profound stabilization of Sdh7 (SDHAF3) (U.N., unpublished data). SDHAF3 may, therefore, contribute to the Fe-S insertion process in SDHB.
Acknowledgements
The authors are extremely grateful to all of the patients, families and clinicians associated with the Australian SDH Consortium; the Kolling Neuroendocrine Tumour Bank; the Kolling Institute Healthy Volunteers Bank; David Espinoza (NHMRC Clinical Trials Centre, The University of Sydney) for statistical support; and Adam Dwight for assistance with bioinformatics.