EIF2AK4 mutation in autosomal dominant HPAH
EIF2AK4 was first described in the autosomal recessively inherited pulmonary veno-occlusive disease (PVOD) [
11] and pulmonary capillary haemangiomatosis [
12]. Recently a single recessively inherited
EIF2AK4 mutation (c.3344C>T, p.(P115L)) was repeatedly identified in six consanguineous HPAH families with autosomal recessive mode of inheritance [
14,
15]. Only homozygous mutation carriers developed the disease plus a single heterozygous carrier of a distinct
EIF2AK4 mutation, in whom the authors suspected a second non-identified mutation in the same gene [
14]. Therefore, up to now
EIF2AK4 mutations have been believed to be a very rare in HPAH. Apart from the mentioned family we were able to identify a nonsense mutation in exon 8 in the
EIF2AK4 gene in another PAH patient with sporadic IPAH who had no other mutation in known candidate genes (data not shown). Thus, this gene might be more often affected than initially thought and contributes to the disease in an autosomal dominant manner. In contrast,
BMPR2 mutations occur in up to 85 % of familial cases and are autosomal dominantly inherited [
7,
8]. However, many
BMPR2 gene carriers have no clinical symptoms and do not develop manifest PH even during a more than 10 year follow-up period [
22]. Hence, the family described here provides an explanation for the decreased penetrance and suggests an autosomal dominantly contribution of
EIF2AK4 to disease manifestation.
EIF2AK4 mutation as “second hit” may explain variable penetrance
Up to date only four families with second hits have been described [
19,
20]. This might be due to the fact that usually only 3 genes (
BMPR2, ACVRL1, ENG) are analysed routinely in PAH patients in a sequential processes, i.e. if one mutation is discovered the other genes are not assessed. Thus, second hit mutations might be overlooked in general in the current routine diagnostic setting and particularly in genes such as
EIF2AK4, which are usually not included in the diagnostic analysis.
While second hits are still rarely described in PAH this model is often found in other diseases such as the nephrotic syndrome [
28] or the long QT-syndrome [
29]. At the same time the decreased penetrance in PAH is an acknowledged pattern. Thus, second hits or modifier genes might be more common in PAH than known to date. We therefore contrast two genetic models: Firstly, we propose the “second hit model” to explain the low disease penetrance in PAH. In this model a single mutation in each gene on its own has a very low penetrance. Two mutations however, lead to a synergistic effect resulting in a high disease penetrance. The second model is the “single gene model”, representing the classical view for PAH suggesting
BMPR2 mutations alone are responsible for disease manifestation. Under this assumption, the
EIF2AK4 mutation would randomly occur in this family and not impact PAH manifestation. In the latter model the penetrance for the
BMPR2 variant must be moderate, since 4 (obligate) carriers of the mutation did not develop PAH up to ages 31, 37, 49 and 59.
Considering both models, it is significantly unlikely that both mutations in two known PAH genes occurred by chance in this family. Moreover, the EIF2AK4 mutation clearly co-segregates with the disease in BMPR2 positive family members suggesting a second hit model.
We have not observed an
EIF2AK4 mutation alone within this family, albeit in a different IPAH patient (data not shown) indicating at least a low penetrance to be present. Disease severity might moreover be influenced by the location of the respective mutations within the protein and thus their variable impact on protein function [
8]. Therefore, we hypothesise according to the second hit model the penetrance to be intermediate if only
BMPR2 was positive, very low if only
EIF2AK4 was positive, and very high if
EIF2AK4 and
BMPR2 each harboured a mutation. However, a greater cohort study would be required assessing IPAH/HPAH patients for all known PAH genes to investigate the frequency of
EIF2AK4 mutations, second hits in PAH and their contribution to disease manifestation within affected families. Furthermore, animal studies would be required to investigate the proposed synergistic effect of two mutations in the same individual.
Not only the number of mutations in different genes but already the state (homo-/heterozygous) of the allele can affect disease severity [
30] and even define which disease is developed. For example
BRCA2 may cause Fanconi aenemia in the homozygous state and familial breast and ovarian cancer in the heterozygous state [
31]. In other diseases variants within the same gene might act dominantly or recessively depending on their localisation within the gene, e.g.
MAB21L2 can lead to eye malformations as a dominant or recessive trait [
32]. Moreover, the 1000 genome project revealed around half a million variants in regulatory sites which most likely act as modifiers on gene expression and are not routinely considered in the diagnostic setting [
33]. Thus, most Mendelian diseases are more complex than initially thought. The same most likely applies to PAH which is characterised by a reduced penetrance. Non-diseased mutation carriers may therefore only be provided with probabilities regarding disease manifestation by genetic counsellors. Any elucidation of further mechanisms refining the predictions of disease manifestation would reduce the uncertainty for patients and genetic counsellors. While we propose a synergistic effect of the two mutations, currently no different therapeutic approach is indicated. A close monitoring of these patients will be required to allow therapy escalation if necessary. A recent publication supports increased disease severity in patients with several mutations, younger age of onset and less effective treatment response in comparison to patients with a single mutation [
34]. Thus, a comprehensive overview will be required with a large cohort of patients analysing current treatment options and genetic mutation status to re-evaluated current therapeutic strategies.
Loss of protein function by EIF2AK4 mutation
The gene
EIF2AK4 encodes a kinase termed general control nonderepressable 2 (GCN2) which phosphorylates the eukaryotic translation initiation factor 2α leading to a global down regulation of protein synthesis in response to amino acid starvation, hypoxia and viral infection but the up-regulation of specific stress response proteins [
35]. Gene expression is increased in smooth muscle cells in the vessel wall and interstitial tissue [
11]. While the gene function has been studied a clear link to PVOD or PAH still remains to be detected. However, an interaction between
EIF2AK4 and the BMPR2 pathway genes
SMAD1, SMAD4, ACVRL1 and
ENG has been observed [
11,
36]. Thus, an impaired functioning of both, BMPR2 and GCN2 (
EIF2AK4), might potentiate its effect on the transcription of target genes of the BMPR2 pathway.
The
EIF2AK4 mutation of this family leads to the loss of a splice site and subsequently the loss of exon 38, presumably a frame shift and premature stop codon. In the last exons of the functional gene (31–39) lies the ribosomal binding domain and the dimerisation domain between amino acids 1396–1643 [
37]. The last 51 amino acids of this domain were missing or partly exchanged by wrong amino acids in the affected members of this HPAH family. The domain is essential to recruit ribosomes for protein synthesis [
38], thus a partial deletion will at least moderately affect protein-ribosome binding if not fully impair it. Moreover, in the same region the dimerisation domain is located. This facilitates the formation of a homodimer (2 EIF2AK4 proteins bind to each other) and thus a functional protein [
38]. The formation of homodimers has been shown to be conserved in mice and yeast [
39]. Single amino acid substitutions in the C-terminal domain in yeast already led to an inability of the protein to dimerise and to be functionally active [
38]. The gene sequence of the C-terminal domain is highly conserved from mice to mammals suggesting corresponding functional impairments in humans [
40]. A total deletion of this region in our HPAH family thus likely leads to a loss of function in the mutated gene.