©20

13 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

Nature GeNetics  ADVANCE ONLINE PUBLICATION �

l e t t e r s

Pulmonary veno-occlusive disease (PVOD) is a rare and devastating cause of pulmonary hypertension that is characterized histologically by widespread fibrous intimal proliferation of septal veins and preseptal venules and is frequently associated with pulmonary capillary dilatation and proliferation�,2. PVOD is categorized into a separate pulmonary arterial hypertension–related group in the current classification of pulmonary hypertension3. PVOD presents either sporadically or as familial cases with a seemingly recessive mode of transmission4. Using whole-exome sequencing, we detected recessive mutations in EIF2AK4 (also called GCN2) that cosegregated with PVOD in all �3 families studied. We also found biallelic EIF2AK4 mutations in 5 of 20 histologically confirmed sporadic cases of PVOD. All mutations, either in a homozygous or compound-heterozygous state, disrupted the function of the gene. These findings point to EIF2AK4 as the major gene that is linked to PVOD development and contribute toward an understanding of the complex genetic architecture of pulmonary hypertension.

PVOD was first recognized as a specific entity of pulmonary hyperten-sion in the 1960s5. PVOD is characterized by a low diffusing capacity for carbon monoxide, occult alveolar hemorrhage and high-resolution computed tomography of the chest that shows patchy centrilobular ground-glass opacities, septal lines and lymph node enlargement6. The true incidence of PVOD remains unknown because many cases are probably misclassified as idiopathic pulmonary arterial hypertension (PAH). The proportion of idiopathic cases of PAH that in reality fulfill the criteria for PVOD is likely around 10%1.

Mutations in BMPR2 are found in approximately 75% of familial cases of PAH and in almost 20% of apparently sporadic cases of PAH. Mutations in ACVRL1, which can complicate hereditary hemorrhagic

telangiectasia, have also been described in PAH7,8. PAH that is due to BMPR2 mutations segregates as an autosomal-dominant trait with incomplete penetrance9.

Familial cases of PVOD have been described in three different studies, and the disease typically occurs in the young siblings of one generation4,10,11. In the French referral center, we identified 13 PVOD families. In eight families (PVOD1, PVOD2, PVOD3, PVOD5, PVOD6, PVOD7, PVOD8 and PVOD12), we confirmed the PVOD diagnosis histologically after lung transplantation or lung biopsy in at least one family member (Fig. 1 and Supplementary Tables 1 and 2). In the five remaining PVOD families (PVOD4, PVOD9, PVOD10, PVOD11 and PVOD13), we considered the diagnosis to be highly likely on the basis of clinical and paraclinical data (Supplementary Tables 1 and 2). All PVOD families were characterized by the presence of at least two affected siblings and unaffected parents, suggesting that the disease segregates as a recessive trait.

To identify the genetic basis of familial forms of PVOD, we first adopted a genetic linkage mapping strategy in three families (PVOD1, PVOD2 and PVOD3). We observed suggestive linkage sig-nals at six regions, with a maximum log10 odds (LOD)-score peak above 1.5 at each locus, but we detected no genome wide–signifi-cant linkage (LOD > 3) (Supplementary Fig. 1). We then performed whole-exome sequencing on subjects from five families (PVOD1, PVOD2, PVOD3, PVOD4 and PVOD5). We selected homozygous or compound-heterozygous variants that were rare (minor allele fre-quency (MAF) <0.1%) or unknown in either the National Heart, Lung, and Blood Institute (NHLBI) Exome Sequencing Project Exome Variant Server (EVS) or the 1000 Genomes Project and that were shared by both affected subjects and were present in a heterozygous state in the unaffected parents. We found that variants of a single gene, EIF2AK4, met these criteria in two families, PVOD1 and PVOD4. In PVOD1, the two affected siblings carried heterozygous frameshift and

EIF2AK4 mutations cause pulmonary veno-occlusive disease, a recessive form of pulmonary hypertensionMélanie Eyries1–3, David Montani4–6, Barbara Girerd4–6, Claire Perret3,7, Anne Leroy2, Christine Lonjou8, Nadjim Chelghoum8, Florence Coulet2,3, Damien Bonnet9,10, Peter Dorfmüller6,11, Elie Fadel6,12, Olivier Sitbon4–6, Gérald Simonneau4–6, David-Alexandre Tregouët3,7, Marc Humbert4–6 & Florent Soubrier1–3

1Unité Mixte de Recherche en Santé (UMR_S 956), Université Pierre and Marie Curie Université Paris 06 (UPMC) and Institut National de la Santé et de la Recherche Médicale (INSERM), Paris, France. 2Genetics Department, Hôpital Pitié-Salpêtrière, Assistance Publique–Hôpitaux de Paris (AP-HP), Paris, France. 3Institute for Cardiometabolism and Nutrition (ICAN), Paris, France. 4Université Paris-Sud, Faculté de Médecine, Le Kremlin Bicêtre, France. 5Département Hospitalo-Universitaire (DHU) Thorax Innovation (TORINO), Service de Pneumologie, Hôpital Bicêtre, AP-HP, Le Kremlin Bicêtre, France. 6INSERM UMR_S 999, Laboratoire d’Excellence en Recherche sur le Médicament et l’Innovation Thérapeutique (LERMIT), Centre Chirurgical Marie Lannelongue, Le Plessis Robinson, France. 7UMR_S 937, UPMC, INSERM, Paris, France. 8Post-Genomic Platform (P3S), UPMC, INSERM, Paris, France. 9Cardiac Surgery Department, Hôpital Necker-Enfants Malades, AP-HP, Paris, France. 10UMR_S 765, INSERM and Université Paris Descartes, Paris, France. 11Department of Pathology, Centre Chirurgical Marie Lannelongue, Le Plessis-Robinson, France. 12Thoracic Surgery Department, Centre Chirurgical Marie Lannelongue, Le Plessis-Robinson, France. Correspondence should be addressed to F.S. (florent.soubrier@upmc.fr).

Received 6 August; accepted 6 November; published online 1 December 2013; doi:10.1038/ng.2844

©20

13 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

2  ADVANCE ONLINE PUBLICATION Nature GeNetics

l e t t e r s

splicing mutations, and in PVOD4, the two affected siblings harbored heterozygous nonsense and frameshift mutations. The genotypes in PVOD2 and PVOD3 were not initially consistent with recessive

transmission in the affected subjects, but further Sanger sequenc-ing of the corresponding EIF2AK4 coding sequences in these fami-lies showed that all affected subjects carried deleterious mutations

a b c

d e f

Figure 1 Pathology of heritable PVOD. (a) Septal vein displaying intimal fibrosis and overall thickening of the vessel wall. Scale bar, 500 µm. (b) Small vein with occlusive intimal fibrosis. Scale bar, 100 µm. (c) Intimal remodeling of a pulmonary artery; the adjacent bronchiole is also visible (bottom). Scale bar, 100 µm. (d) Numerous patch-like foci of capillary hemangiomatosis; small veins with intimal remodeling are visible (left and bottom). Scale bar, 1,000 µm. (e) PCH-like remodeling with a central muscularized arteriole or small artery. Scale bar, 100 µm. (f) Small artery displaying important concentric fibrosis of the intima; the slender media is visible, which is delimited by the internal and external lamina elastica (highlighted in black). Scale bar, 100 µm. Hematoxylin and eosin stains are shown in a–e, and orcein stain is shown in f.

table 1 Age at diagnosis of PVOD, gender and genotypes of subjects with EIF2AK4 mutationsFamily number Individual Gender Age at diagnosis (years) Mutationa

PVOD1 II-1 F 31 c.354_355del; c.1554-4C>A (p.Cys118Trpfs*7; p.Cys519Aspfs*17)

II-2 F 50

II-3 F 49

PVOD2 II-2 M 23 c.2319+1G>A; c.2319+1G>A (p.[?]; p.[?])

II-3 F 16

PVOD3 II-2 F 23 c.745C>T; c.2136_2139dup (p.Arg249*; p.Ser714Hisfs*21)

II-3 M 23

PVOD4 II-1 M 20 c.1392del; c.3802C>T (p.Arg465Valfs*38; p.Gln1268*)

II-2 F 20

PVOD5 IV-2 F 20 c.567dup; c.567dup (p.Leu190Glufs*8; p.Leu190Glufs*8)

IV-3 F 27

PVOD6 III-2 M 26 c.3159G>A; c.3159G>A (p.Lys975_Lys1053del; p.Lys975_Lys1053del)

PVOD7 II-1 M 32 c.3406C>T; c.3406C>T (p.Arg1136*; p.Arg1136*)

PVOD8 III-3 M 26 c.1754G>A; c.1754G>A (p.Arg585Gln; p.Arg585Gln)

PVOD9 II-5 M 36 c.4065+1G>C; c.4065+1G>C (p.[?]; p.[?])

PVOD10 II-1 M 19 c.1387C>T; p.1387C>T (p.Arg463*; p.Arg463*)

PVOD11 II-1 F 37 c.3448C>T; c.4728+1_4728+13delinsTTCT (p.Arg1150*; p.[?])

PVOD12 II-1 M 44 c.1387C>T; c.3244C>T (p.Arg463*; p.Gln1082*)

PVOD13 IV-3 M 11 c.1928T>G; c.1928T>G (p.Leu643Arg; p.Leu643Arg)

112160 F 32 c.560_564del; c.560_564del (p.Lys187Argfs*9; p.Lys187Argfs*9)

091769 F 15 c.3159G>A; c.3159G>A (p.Lys975_Lys1053del; pLys975_Lys1053del)

05220 M 20 c.2857C>T; c.3576+1G>T (p.Gln953*; p.[?])

05498 M 20 c.4205dup; c.4205dup (p.Ser1403Lysfs*45; p.Ser1403Lysfs*45)

06734 F 28 c. 2458C>T; c.2458C>T (p.Arg820*; p.Arg820*)aShown as the nucleotide changes with the respective protein effects in parentheses.

©20

13 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

Nature GeNetics  ADVANCE ONLINE PUBLICATION 3

l e t t e r s

on both alleles and that each unaffected parent was heterozygous for one of the two mutations (Table 1 and Supplementary Fig. 2a). In the consanguineous family (PVOD5), Sanger sequencing in both affected siblings identified a deleterious homozygous mutation located in exon 5 of EIF2AK4, which had not been captured by whole-exome sequencing. Of note, EIF2AK4 maps to the LOD-score peak on chromosome 15 that had been detected by nonparametric linkage (Supplementary Fig. 1).

We screened the index cases of eight additional PVOD families with typical recessive transmission (PVOD6, PVOD7, PVOD8, PVOD9, PVOD10, PVOD11, PVOD12 and PVOD13) and found biallelic EIF2AK4 mutations in all of these families (Table 1 and Supplementary Fig. 2b). In family PVOD13, we found a homozygous missense muta-tion encoding p.Leu643Arg, which is considered deleterious by in silico analysis (Online Methods), in the 11-year-old index case, whose brother and sister both died of pulmonary hypertension at the ages of 15 and 10 years, respectively. Histological aspects of the diagnostic lung biopsy of the index case, a borderline value of mean pulmonary arterial pressure, high-resolution computed tomography of the chest, substan-tial oxygen desaturation at exercise and familial context together argue in favor of an incipient form of PVOD in this homozygous mutation carrier (Supplementary Table 2). The rather high number of PVOD families collected in our referral center could be explained by both a long-term and careful search for the disease among patients studied and referral to the center of patients originating from North Africa, where recessive disease occurrence is favored by endogamy.

We also investigated the entire EIF2AK4 coding sequence in 20 apparently sporadic, histologically proven cases of PVOD, in which

pulmonary tissue either after lung transplantation or lung biopsy or at post-mortem was examined (Supplementary Table 3). We found biallelic EIF2AK4 mutations in 5/20 cases (25%), suggesting that a quarter of these sporadic cases could correspond to heritable PVOD (Table 1). We also investigated an additional series of 26 patients with pulmonary hypertension who had clinical and radiological signs that were strongly suggestive of PVOD but in whom PVOD was not histo-logically proven. No EIF2AK4 mutations were found in this series of patients. We also analyzed nine index cases from PAH families with-out BMPR2 mutations, but we detected no EIF2AK4 mutations.

Collectively we identified a total of 22 distinct EIF2AK4 mutations in this study. Most of these mutations are stop codons (n = 8) or insertions/deletions (indels) (n = 6) that disrupt the function of the gene. We identified two distinct missense mutations that resulted in p.Arg585Gln in PVOD8 and p.Leu653Arg in PVOD13. Both of these mutations are located in the kinase domain of EIF2AK4 (Fig. 2), alter conserved residues and are predicted to be deleterious by in silico tools (Online Methods). We also found six splice mutations, four of which affect a consensus donor site and two of which are pre-dicted to affect splicing by in silico analysis. In both cases, RT-PCR experiments using RNA extracted from circulating blood cells from the mutation carriers confirmed that mRNA splicing was defective (Supplementary Fig. 3). Notably, we found two mutations (encod-ing p.Lys975_Lys1053del and p.Arg463*) in two patients from two apparently unrelated families. In particular, we found the p.Lys975_Lys1053del alteration in two native Algerian patients. To estimate the frequency of this variant in the Algerian population, we performed further genotyping in 278 Algerian control subjects. With a sample

p.C118Wfs*7

p.K190Efs*8p.R465Vfs*38

p.C519Dfs*17

p.R249*

NH2 RWD Pseudokinase Protein kinase His-tRNA synthase_like

1649 aa1493

p.R463*p.R463* p.R585Q

p.L643R

p.R820* c.3576+1G>Tc.4065+1G>C

p.S1403Kfs*45c.4728+1_4728+13delinsTTCTp.Q1082*Q958*

p.R1150*

p.R1136* p.Q1268*

p.c.2319+1G>Ap.K975_K1053delp.K975_K1053del

p.S714Hfs*21p.K187Rfs*9

COOH

1 25 137 296 539 590 1001 1022

Figure 2 Location of the EIF2AK4 mutations detected in patients with PVOD. The locations and consequences of the recessive mutations in subjects with familial PVOD (black) or sporadic PVOD (blue) are shown on a schematic of the EIF2AK4 protein25. Mutations found in a homozygous state are shown in bold; the other mutations were found in a heterozygous state. For nonsynonymous and frameshift mutations, the protein effects are shown, and for splicing variants, the nucleotide changes are indicated. In the figure, amino acids are indicated by their one-letter abbreviations. aa, amino acid.

Figure 3 Immunohistochemical staining with an EIF2AK4 antibody in lungs from a control patient, a patient with PVOD not carrying mutations and a patient with PVOD who is homozygous for a EIF2AK4 mutation encoding p.Leu190Glufs*8. (a) Control lung. The arrows indicate the slender venule showing staining of the vessel wall (smooth muscle cells) and negative endothelial cells. Diffuse interstitial staining is indicated by arrowheads; macrophages and some mononuclear cells are stained (triangles). (b) Nonmutated PVOD lung. Smooth muscle cells within the vessel wall of the remodeled microvessel (arteriole or venule) are highlighted by the antibody (arrow). Again, diffuse interstitial staining is present (arrowhead). (c) EIF2AK4-mutated PVOD lung. The remodeled microvessel (arteriole or venule; center) and the interstitium are not labeled. All scale bars, 100 µm.

a Control b Nonmutated PVOD EIF2AK4-mutated PVOD(p.L190Efs*8)

c

©20

13 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

4  ADVANCE ONLINE PUBLICATION Nature GeNetics

l e t t e r s

of this size, we had 94% power to detect any variant with MAF as low as 0.005. However, we did not detect any carriers in this group, confirming that the p.Lys975_Lys1053del alteration is rare, even in the Algerian population.

Using immunohistochemistry, the EIF2AK4 protein is detected in smooth muscle cells of the vessel wall and interstitial tissue and in macrophages in control lungs. We observed a similar pattern of expression in a lung from a patient affected with PVOD that did not have EIF2AK4 mutations, whereas we detected no expression of EIF2AK4 in the lung from patient IV-3 of family PVOD5, who carries the homozygous mutation encoding p.Lys190Glufs*8 (Fig. 3).

The arguments that support EIF2AK4 as a major causal gene for PVOD are as follows: (i) the complete consistency between genotype and phenotype of the affected patients, together with the segregation of single-allele mutations from the parents to their progeny; (ii) the unambiguous loss of gene function caused by the mutations identi-fied; (iii) the presence of biallelic mutations in the same gene in all families tested that share the same specific PVOD phenotype; (iv) the presence of homozygous EIF2AK4 mutations in affected siblings of consanguineous families; and (v) the low genetic burden of EIF2AK4, as inspection of EVS showed only six loss-of-function mutations among nearly 12,000 sequenced alleles.

We describe a genetic cause for PVOD, which is a particular form of pulmonary hypertension whose recessive mode of transmission differs from that of dominant-heritable forms of BMPR2-linked PAH that have lower penetrance rates (14% and 42% in males and females, respectively12). The penetrance of the EIF2AK4 mutations is difficult to evaluate from our data because we genotyped only the affected subjects from sibships, but the single unaffected sibling tested from one family carries the mutation in the heterozygous state (subject II-1 from PVOD2; Supplementary Fig. 2a). In the nonfamilial but histologically proven cases of PVOD, we found a mutation rate of 25%, which indicates that these heritable-recessive forms are frequently unrecognized, as is the case for sporadic PAH caused by BMPR2 muta-tions. The development of a genetic test will allow these heritable forms to be recognized and genetic counseling to be performed.

The age of onset of PVOD in mutation carriers varies, with early death seen in some families (at 10 years of age or earlier) and onset delayed until the age of 50 years in others. The early occurrence sug-gests that additional genetic or environmental factors may accelerate the disease process, as there is no detectable genotype-phenotype rela-tionship in the families studied that could explain the variable age of onset. Similar to previous observations in patients with PAH who carry mutations in BMPR2 or ACVRL1 (ref. 8), we noted that patients with PVOD carrying EIF2AK4 mutations (n = 24) were significantly younger at the time of PVOD diagnosis compared to patients with PVOD who do not carry such mutations (n = 15) (26.7 ± 10.4 years (mean ± s.d.) compared to 44.3 ± 11.8 years, respectively; P < 0.0001).

Notably, patients II-1 and II-2 from family PVOD1 were initially diagnosed as having pulmonary capillary hemangiomatosis (PCH) on the basis of extensive precapillary proliferation with moderated venous involvement that was seen in explanted lung samples from the two siblings. In the current classification of pulmonary hypertension, PCH is grouped with PVOD because PCH shares many similarities with PVOD13,14. Researchers from a previous study demonstrated that capillary proliferation could be identified in 73% of cases diagnosed as having PVOD and that 80% of cases of PCH presented venous and arterial changes that are typical of PVOD15. Our results reinforce the hypothesis that PVOD and PCH may be two different histological patterns of the same disease with a common genetic risk factor.

EIF2AK4 encodes a serine-threonine kinase present in all eukaryo-tes that can induce changes in gene expression in response to amino acid deprivation. EIF2AK4 belongs to a family of four kinases that phosphorylate the α-subunit of the eukaryotic initiation factor 2 (eIF2α). eIF2 functions by directing the binding of initia-tor methionyl-tRNA to 40S ribosomal subunits in the early stages of protein synthesis from a small number of specific mRNAs. EIF2AK4- mediated phosphorylation of eIF2α inactivates the factor and leads to preferential synthesis of stress proteins16.

The pathophysiological link between biallelic EIF2AK4 loss-of-function mutations and vascular cell proliferation and remodeling of lung vessels remains elusive. Eif2ak4−/− mice on a C57BL/6J background are less viable than wild-type mice, as a subset of the mice lacking Eif2ak4 die shortly after birth; however, the lungs of these mice have not yet been studied17. EIF2AK4 has been shown to prevent oxidative damage induced by an amino acid–imbalanced diet, as Eif2ak4−/− mice show an increase in protein carbonylation, which is a marker of oxidative stress that is important for pulmo-nary hypertension development18,19. The involvement of EIF2AK4 in PVOD could be related either directly to its amino acid starva-tion sensor function and subsequent translational changes second-ary to its activation or to EIF2AK4 kinase activity, which might have substrates other than eIF2α. Two series of experimental data potentially connect EIF2AK4 to the bone morphogenetic protein (BMP) pathway, which has been implicated in PAH pathogenesis through mutations in BMPR2 and ACVRL1 (refs. 7,20). Notably, as a result of an interaction screen performed with the luminescence-based mammalian interactome mapping technique, EIF2AK4 was found to interact with SMAD4 and SMAD1 (ref. 21) and also with ALK-1, endoglin (ENG) and transforming growth factor-β receptor-2 (TGFBR2) (M. Letarte, personal communication). These proteins are receptors, co-receptors or signaling intermediates of the TGF-β–BMP superfamily. Second, TRIB3, the Tribbles homolog 3 gene, is transcriptionally regulated by ATF4, a stress protein whose translation is activated by EIF2AK4, as has been demonstrated in the mouse liver22. Downregulation of TRIB3 has been shown to inhibit BMP-mediated cellular responses23. One can speculate that EIF2AK4 inactivation would lead to TRIB3 downregulation and decreased BMP signaling, this latter condition leading to PAH in the case of BMPR2 haploinsufficiency24.

The identification of EIF2AK4 mutations as the major cause of heritable PVOD confirms the hereditary origin of this particular type of pulmonary hypertension and allows identification of heritable but apparently sporadic cases. Mutation identification will allow genetic counseling to be offered to families affected by this extremely severe disease. These results also open new research avenues into the role of EIF2AK4 in pulmonary vascular remodeling and pave the way for innovative therapeutic strategies.

URLs. The NHLBI Exome Sequencing Project EVS, http://evs.gs.washington.edu/EVS/; 1000 Genome Projects, http://www.1000genomes.org/; the Consensus Assessment of Sequence and Variation (CASAVA) software, http://www.illumina.com/software/genome_analyzer_software.ilmn; the Human Genome Variation Society, www.hgvs.org/mutnomen; SIFT, http://sift.jcvi.org/; PolyPhen-2, http://genetics.bwh.harvard.edu/pph2/; Align-GVGD, http://agvgd.iarc.fr; Mutation Taster, http://www.mutationtaster.org/; MaxEntScan, http://genes.mit.edu/burgelab/; NNsplice, http://www.fruitfly.org/seq_tools/splice.html; GeneSplicer, http://ccb.jhu.edu/software/genesplicer/; Human Splicing Finder, http://www.umd.be/HSF/.

©20

13 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

Nature GeNetics  ADVANCE ONLINE PUBLICATION 5

l e t t e r s

MeThODsMethods and any associated references are available in the online version of the paper.

Note: Any Supplementary Information and Source Data files are available in the online version of the paper.

ACkNOwLEDGMENTSWe thank F. Pires, A. Dion-Minière, S. Bakas, G. Legrand and N. Raymond for technical assistance. We thank W. Carpentier for supervising SNP array experiments. We thank R. Peat for kindly editing the manuscript. D.M. and P.D. are supported by a grant from the Association Hypertension Artérielle Pulmonaire (HTAP) France. This work was supported by Programme Hospitalier de Recherche Clinique (PHRC) AOM07-041, INSERM and UPMC. The tissue bank was supported in part by the Legs Poix, Chancellerie des Universités de Paris. Bioinformatics analyses benefit from the C2BIG computing centre funded by the Région Ile de France and UPMC.

AUTHOR CONTRIBUTIONSF.S. initiated and supervised the study. M.E., D.-A.T., M.H. and F.S. conceived and designed the experiments. D.M., B.G., D.B., O.S., G.S., E.F. and M.H. performed clinical phenotyping. D.M., B.G. and M.H. analyzed clinical data of collected patients. C.P. performed the whole-exome sequencing experiments. N.C. performed bioinformatic analyses. D.-A.T. supervised bioinformatic and biostatistical data. M.E. and F.S. analyzed whole-exome sequencing data. C.L. performed linkage analysis. A.L. performed Sanger sequencing. M.E. and F.C. analyzed Sanger sequencing data. E.F. collected lung sample specimens. P.D. performed tissue imaging. M.E., D.M., B.G., D.-A.T., M.H. and F.S. wrote the manuscript. All authors reviewed the manuscript.

COMPETING FINANCIAL INTERESTSThe authors declare no competing financial interests.

Reprints and permissions information is available online at http://www.nature.com/reprints/index.html.

1. Mandel, J., Mark, E.J. & Hales, C.A. Pulmonary veno-occlusive disease. Am. J. Respir. Crit. Care Med. 162, 1964–1973 (2000).

2. Montani, D. et al. Pulmonary veno-occlusive disease. Eur. Respir. J. 33, 189–200 (2009).

3. Simonneau, G. et al. Updated clinical classification of pulmonary hypertension. J. Am. Coll. Cardiol. 54, S43–S54 (2009).

4. Davies, P. & Reid, L. Pulmonary veno-occlusive disease in siblings: case reports and morphometric study. Hum. Pathol. 13, 911–915 (1982).

5. Heath, D., Segel, N. & Bishop, J. Pulmonary veno-occlusive disease. Circulation 34, 242–248 (1966).

6. Montani, D. et al. Pulmonary veno-occlusive disease: clinical, functional, radiologic, and hemodynamic characteristics and outcome of 24 cases confirmed by histology. Medicine 87, 220–233 (2008).

7. Trembath, R.C. et al. Clinical and molecular genetic features of pulmonary hypertension in patients with hereditary hemorrhagic telangiectasia. N. Engl. J. Med. 345, 325–334 (2001).

8. Girerd, B. et al. Clinical outcomes of pulmonary arterial hypertension in patients carrying an ACVRL1 (ALK1) mutation. Am. J. Respir. Crit. Care Med. 181, 851–861 (2010).

9. Machado, R.D. et al. Genetics and genomics of pulmonary arterial hypertension. J. Am. Coll. Cardiol. 54, S32–S42 (2009).

10. Rosenthal, A., Vawter, G. & Wagenvoort, C.A. Intrapulmonary veno-occlusive disease. Am. J. Cardiol. 31, 78–83 (1973).

11. Voordes, C.G., Kuipers, J.R. & Elema, J.D. Familial pulmonary veno-occlusive disease: a case report. Thorax 32, 763–766 (1977).

12. Larkin, E.K. et al. Longitudinal analysis casts doubt on the presence of genetic anticipation in heritable pulmonary arterial hypertension. Am. J. Respir. Crit. Care Med. 186, 892–896 (2012).

13. Humbert, M. et al. Pulmonary edema complicating continuous intravenous prostacyclin in pulmonary capillary hemangiomatosis. Am. J. Respir. Crit. Care Med. 157, 1681–1685 (1998).

14. Langleben, D. et al. Familial pulmonary capillary hemangiomatosis resulting in primary pulmonary hypertension. Ann. Intern. Med. 109, 106–109 (1988).

15. Lantuéjoul, S., Sheppard, M.N., Corrin, B., Burke, M.M. & Nicholson, A.G. Pulmonary veno-occlusive disease and pulmonary capillary hemangiomatosis: a clinicopathologic study of 35 cases. Am. J. Surg. Pathol. 30, 850–857 (2006).

16. Donnelly, N., Gorman, A.M., Gupta, S. & Samali, A. The eIF2α kinases: their structures and functions. Cell. Mol. Life Sci. 70, 3493–3511 (2013).

17. Anthony, T.G. et al. Preservation of liver protein synthesis during dietary leucine deprivation occurs at the expense of skeletal muscle mass in mice deleted for eIF2 kinase GCN2. J. Biol. Chem. 279, 36553–36561 (2004).

18. Chaveroux, C. et al. Identification of GCN2 as new redox regulator for oxidative stress prevention in vivo. Biochem. Biophys. Res. Commun. 415, 120–124 (2011).

19. Fessel, J.P. et al. Hyperoxia synergizes with mutant BMPR2 to cause metabolic stress, oxidant injury, and pulmonary hypertension. Am. J. Respir. Cell Mol. Biol. 49, 778–787 (2013).

20. Machado, R.D. et al. BMPR2 haploinsufficiency as the inherited molecular mechanism for primary pulmonary hypertension. Am. J. Hum. Genet. 68, 92–102 (2001).

21. Barrios-Rodiles, M. et al. High-throughput mapping of a dynamic signaling network in mammalian cells. Science 307, 1621–1625 (2005).

22. Carraro, V. et al. Amino acid availability controls TRB3 transcription in liver through the GCN2/eIF2α/ATF4 pathway. PLoS ONE 5, e15716 (2010).

23. Chan, M.C. et al. A novel regulatory mechanism of the bone morphogenetic protein (BMP) signaling pathway involving the carboxyl-terminal tail domain of BMP type II receptor. Mol. Cell. Biol. 27, 5776–5789 (2007).

24. Davies, R.J. et al. BMP type II receptor deficiency confers resistance to growth inhibition by TGF-β in pulmonary artery smooth muscle cells: role of proinflammatory cytokines. Am. J. Physiol. Lung Cell. Mol. Physiol. 302, L604–L615 (2012).

25. Sattlegger, E. & Hinnebusch, A.G. Separate domains in GCN1 for binding protein kinase GCN2 and ribosomes are required for GCN2 activation in amino acid–starved cells. EMBO J. 19, 6622–6633 (2000).

©20

13 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

Nature GeNetics doi:10.1038/ng.2844

ONLINe MeThODsSubjects. Patients were studied at the French referral center for severe pulmonary hypertension (Université Paris-Sud, AP-HP, Le Kremlin-Bicêtre), and patient IV-3 from family PVOD 13 was studied at the Cardiopediatrics Department of the Necker-Enfants Malades hospital (AP-HP). All clinical characteristics at PVOD diagnosis and follow-up were stored in the Registry of the French PAH Network. This registry was set up in agreement with French bioethics laws (French Commission Nationale de l’Informatique et des Libertés), and patients gave their consent to be included26. All patients and relatives gave their informed consent for genetic research, which was approved by the Comité de Protection des Personnes Ile de France-VI, decision ID RCB2007-AO1347-46. The diagnosis of precapillary pulmonary hypertension was defined by hemodynamic measurement during right-heart catheterization of all patients included in the study. Precapillary pulmonary hypertension was defined as a mean pulmonary arterial pressure ≥25 mm Hg (where normal values are 14 ± 3 mm Hg (mean ± s.d.)) associated with normal pulmonary capillary wedge pressure (≤15 mm Hg). In an early pediatric familial case (PVOD13), an abnormal borderline value was observed (mean pulmonary artery pressure of 23 mm Hg). Hemodynamic evaluation by right-heart catheterization was performed at baseline in all subjects according to previously described protocols27.

The diagnosis of PVOD was considered as confirmed when histological proof of veno-occlusive disease was available. Histological proof of veno- occlusive disease was based on hematoxylin-eosin-safran staining of lung biopsies, post-mortem lung samples or lungs obtained after lung transplan-tation. The pathologic hallmark of PVOD is defined as an extensive and diffuse obstruction of pulmonary veins and venules by intimal fibrosis, cellular proliferation and muscularization.

The diagnosis of PVOD was considered as highly probable if patients fulfilled the following characteristics: precapillary pulmonary hyperten-sion, presence of two or more radiological abnormalities on high-resolution computed tomography of the chest (including lymph node enlargement, centrilobular ground-glass opacities and septal lines), low diffusing capacity for carbon monoxide or occult alveolar hemorrhage. The diagnosis was even more strongly supported when patients with signs of highly probable PVOD developed pulmonary edema after the initiation of specific PAH therapy.

Genetic analysis. All patients were screened for BMPR2 mutations, and those with a family history of precapillary pulmonary hypertension were further screened for ACVRL1 mutations. Screening for point mutations and large rearrangements was performed as previously reported28,29.

Linkage analysis. Three pedigrees (PVOD1, PVOD2 and PVOD3) were ini-tially used for linkage analysis. In each of these pedigrees, two affected siblings and their unaffected parents were genotyped for genome-wide SNPs by the Illumina HumanOmniExpress_12v1 DNA beadchip. Genotyping quality controls, including Mendelian errors, were conducted using Pedcheck30 and Merlin31 software. Nonparametric linkage analysis was performed using the exponential model proposed by Kong and Cox32, which is particularly adapted to studies with few families where a strong genetic effect is anticipated, as in our project. Multipoint analyses were also performed. Because the presence of linkage disequilibrium (LD) between SNPs can introduce bias in multipoint linkage analysis, before the analysis, we eliminated strong LD between SNPs and selected SNPs regularly spaced every ~250 kb. The absence of LD between selected SNPs was further confirmed using the Haploview program33.

Exome sequencing and sequence analysis. All sequencing experiments were conducted at the Post-Genomic Platform of the Pitié-Salpêtrière (P3S) campus. Genomic DNA from selected individuals was prepared and subjected to exome capture using the Truseq Exome Enrichment kit (Illumina) followed by next-generation sequencing on the HiSeq2000 platform (Illumina). Sequence reads were aligned to the human genome reference sequence (hg19 build) using the Consensus Assessment of Sequence and Variation (CASAVA) software. PCR or optical duplicates were discarded, as were nonpaired reads and reads with low-quality mapping (Q score < 20). Variant calling was performed using the

SAMtools program34, and Annovar35 software was used to annotate the identi-fied variants. Variants reported with MAF > 0.1% in the 1000 Genomes Project or in the EVS, as well as variants in intronic (except variants considered to be splicing variants and located at exon-intron junctions ranging from −5 to +5) and intergenic regions, were filtered out. The analysis prioritized homozygous or compound-heterozygous protein-altering variants. Variants of interest were verified by Sanger sequencing.

To search for EIF2AK4 mutations in additional subjects, the entire cod-ing sequence and intronic junctions of EIF2AK4 were PCR amplified using specific oligonucleotide primer pairs and subjected to bidirectional Sanger sequencing. The EIF2AK4 primer sequences used are listed in Supplementary Table 4. The resulting sequence data were analyzed with the SeqScape soft-ware, version 2.5 (Applied Biosystems), in comparison with the EIF2AK4 reference sequence (NM_001013703). Nucleotide numbering reflects cDNA numbering, with +1 corresponding to the A of the ATG translation initiation codon in the reference sequence according to the Human Genome Variation Society (HGVS) recommendations.

In silico tools for variant interpretation. The missense mutations identified were analyzed by the prediction programs SIFT, Polyphen-2, Align-GVGD and Mutation Taster to predict the putative functional role of the amino acid sequence alteration. Potential splice variants identified were analyzed by five splice-site predictions programs, Splice Site Finder, MaxEntScan, NNsplice, GeneSplicer and Human Splicing Finder, to predict the potential effect on mRNA splicing. All these tools are available through Alamut v2.2 software (Interactive Biosoftware).

cDNA analysis. Total RNA from whole-blood samples was extracted with the PAXGene Blood RNA system (Qiagen), and 500 ng of RNA was used for RT-PCR according to the manufacturer’s recommendations (Superscript II, random primers, Invitrogen). PCR products were analyzed by bidirectional Sanger sequencing.

Immunohistochemical studies. Lung specimens were obtained at the time of lung transplantation from patients with PVOD. Control lung specimens were obtained from patients without any evidence of pulmonary vascular disease who underwent lobectomy or pneumonectomy for localized lung cancer, with normal tissue collected at a distance from the tumors. Immunohistochemical analysis was performed on formalin-fixed samples. Paraffin-embedded sections were stained with polyclonal rabbit anti-human primary antibody against EIF2AK4 (Abcam, reference ab137543; dilution, 1:100). Biotinylated horse anti-rabbit IgG (Vector, reference BA-1100) was used for primary anti-body detection, streptavidin-bound alkaline phosphatase (Vector, reference SA-5100) was used for amplification and Vector Red alkaline phosphatase substrate (Vector, reference SK-5100) was the chromogen.

26. Humbert, M. et al. Pulmonary arterial hypertension in France: results from a national registry. Am. J. Respir. Crit. Care Med. 173, 1023–1030 (2006).

27. Sitbon, O. et al. Long-term response to calcium channel blockers in idiopathic pulmonary arterial hypertension. Circulation 111, 3105–3111 (2005).

28. Sztrymf, B. et al. Clinical outcomes of pulmonary arterial hypertension in carriers of BMPR2 mutation. Am. J. Respir. Crit. Care Med. 177, 1377–1383 (2008).

29. Eyries, M. et al. ACVRL1 germinal mosaic with two mutant alleles in hereditary hemorrhagic telangiectasia associated with pulmonary arterial hypertension. Clin. Genet. 82, 173–179 (2012).

30. O’Connell, J.R. & Weeks, D.E. PedCheck: a program for identification of genotype incompatibilities in linkage analysis. Am. J. Hum. Genet. 63, 259–266 (1998).

31. Abecasis, G.R., Cherny, S.S., Cookson, W.O. & Cardon, L.R. Merlin—rapid analysis of dense genetic maps using sparse gene flow trees. Nat. Genet. 30, 97–101 (2002).

32. Kong, A. & Cox, N.J. Allele-sharing models: LOD scores and accurate linkage tests. Am. J. Hum. Genet. 61, 1179–1188 (1997).

33. Barrett, J.C., Fry, B., Maller, J. & Daly, M.J. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21, 263–265 (2005).

34. Li, H. et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078–2079 (2009).

35. Wang, K., Li, M. & Hakonarson, H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 38, e164 (2010).