Genetic Predisposition to High-Altitude Pulmonary Edema

Study subjects

In a family of Northern European descent with low-altitude heritage and re-entry HAPE, we collected family history and performed clinical examinations, including medical history, physical examination, laboratory parameters, electrocardiogram, lung function test, chest X-ray, echocardiography at rest and during exercise, as described previously (Bärtsch et al., 1991). Diagnosis of PH/PAH was performed following ERS/ESC guidelines, including right heart catheterization (Galiè et al., 2016).

Blood samples were also analyzed from 64 unrelated healthy Caucasian mountaineers from western/central Europe without acute or chronic pulmonary or cardiovascular diseases who had at least one prior HAPE episode, assessed in various studies at high altitude in the Capanna Margherita (4559 m altitude) (Dehnert et al., 2002, 2005, 2010; Maggiorini et al., 2006; Betz et al., 2015). Briefly, mountaineers ascended by cable car from Alagna (1100 m) to 3200 m and climbed to the Capanna Gnifetti (3611 m) to spend the night. On the following morning they climbed to the Capanna Margherita. Oxygen saturation was measured the following morning by pulse oximetry and sPAP was measured with Doppler echocardiography (Grünig et al., 2000b). HAPE was diagnosed by chest radiographs. Echocardiographic videos and X-rays were assessed in a double blinded fashion by two independent clinicians. This study was conducted in accordance with the amended Declaration of Helsinki. The study was approved by the Ethics Committee at Heidelberg University (065/2001, S-378/2017 and S-427/2017); written informed consent was obtained from all participants. For children, informed consent was issued by a parent and/or legal guardian for study participation.

DNA and complementary DNA analysis

DNA was extracted from peripheral blood from 8 family members of the HAPE family (Autogene Qiagen) and the 64 HAPE-S mountaineers (PaxGene DNA and Qiagen DNA extraction assays) using standardized procedures. Sequence variants in HAPE-S and the familial HAPE index patient were sought in the following PAH genes (activin A receptor (like) type 1; [ACVRL1], BMPR1B, BMPR2, caveolin 1, eukaryotic translation initiation factor 2 alpha kinase 4 [EIF2AK4], endoglin, growth differentiation factor 2 (bone morphogenetic protein 9) [GDF2], potassium voltage-gated channel subfamily A member 5, KCNK3, Krüppel-like factor 2 [KLF2], Mothers Against Decapentaplegic Homolog 4/5/6/7/9 [SMAD]1, SMAD4, SMAD9, T-Box transcription factor [TBX] 4, and DNA Topoisomerase II binding protein 1) and further candidate genes (ACVR1, BMP2, BMPR1A, Butyrophilin Like 2, CAMP responsive element binding protein 1, Cytochrome P450 family 1 subfamily B member 1 [CYP1B1], endothelial PAS domain protein 1 [hypoxia-inducible factor 2 alpha] [EPAS1], forkhead box O1 [FOXO1], HGR, inhibitor of DNA binding 1 [ID1], ID2, ID3, ID4, interleukin 6, Janus Kinase 2 [JAK2], KLF4, KLF5, notch 3 [NOTCH3], SMAD5, SMAD6, SMAD7, superoxide dismutase 2, TBX2, transmembrane protein 70, versican [VCAN], Von Hippel-Lindau tumor suppressor, and zinc finger FYVE-type containing 16) using a next-generation sequencing (NGS)-based PAH-specific gene panel. Procedures were followed as reported previously (Song et al., 2016) apart from the addition of further genes to the panel and the use of the SureSelect QXT (Agilent) system for library preparation.

Respective transcript IDs are listed in Supplementary Table S1. Variants in exonic regions and exon/intron boundaries were characterized following recommendations of the Human Genome Variation Society and the genetic variant interpretation tool (Richards et al., 2015). Nonsynonymous missense variants with a population frequency of <1% were assessed using four in silico prediction programs (MutationTaster, Sorting Intolerant From Tolerant [SIFT], Align Grantham Variation Grantham Deviation [GVGD], PolyPhen2); variants impacting splice sites were evaluated using the prediction programs SpliceSiteFinder-like, MaxEntScan, NNSPLICE, GeneSplicer, and Human Splicing Finder (Alamut Visual 2.9, interactive biosoftware). The Combined Annotation-Dependent Depletion Score was calculated to consider further algorithms (Kircher et al., 2014) and to exclude variants with a score <20. Variants of unknown significance (VUS, class III) and likely pathogenic variants (class IV) were confirmed by Sanger sequencing. Benign variants and likely benign variants (class I and II) were considered polymorphisms and not followed up.

Two variants, which were predicted by at least two programs to lead to a loss or gain of splice sites, were also assessed on RNA level. RNA was extracted from EDTA blood using standard protocols (RNeasy Mini Kit; Qiagen). Complementary DNA was transcribed with superscript II reverse transcriptase (Life Technologies) with random hexamer primers (Roche). Polymerase chain reactions (PCRs) were designed with primers located in adjacent exons to the predicted newly generated or lost splice site. Sizes of amplification products were measured by agarose gel electrophoresis and Sanger sequenced.

Statistics

Variants of uncertain significance and likely pathogenic variants in PAH genes in HAPE-S were compared with their frequency in a presumably healthy control population with unknown HAPE incidence listed in the genome Aggregation Database (gnomAD) (Lek et al., 2016). Variant classification criteria and allele frequency cutoffs were adapted from PAH diagnostic setting, considering any variant with an allele frequency in gnomAD <1% for further analysis, depending on its classification by in silico prediction programs and further functional evidence. Clinical parameters of HAPE-S were given in % for frequency distributions or as mean ± standard deviation. Frequency distributions of genetic variants in mountaineers were provided with respective 95% confidence intervals (95% CIs).

Clinical and genetic findings in the HAPE family

We assessed a family with Northern European descent of which the mother and her relatives originated from low altitude (120 m). The father, although also son of Northern European parents, and the two youngest children were born and grew up in La Paz, Bolivia (3640 m). The two oldest children were born in Germany at low altitude and moved to La Paz when they were between 6 and 8 weeks old (Fig. 1). At the age of 8 years, the two oldest siblings (III:3 and III:4) developed repeated episodes of severe re-entry HAPE during the first night at high altitude (diagnosed by local HAPE-specialists) when they returned to La Paz from holidays spent with the grandparents in Germany at low altitude. The sister of the father, who was also born in Germany at low altitude developed re-entry HAPE when the family settled to La Paz. She subsequently moved to the lowlands. No DNA could be obtained from her but from eight other family members.

OPEN IN VIEWER

FIG. 1.

Caucasian family with inherited re-entry HAPE, progression to PAH in the index patient. The siblings III:3 to III:6 and their parents II:3 and II:4 lived permanently at an altitude of 3640 m. Two of them, who were both born in the lowlands and shortly afterward moved to high altitude developed frequent re-entry HAPE. Their aunt II:2 also developed re-entry HAPE during childhood. II:3, III:5, and III:6 were born at high altitude and showed no signs of HAPE. A likely pathogenic variant in the gene JAK2 was identified in three of the siblings. The variant might have a reduced penetrance as II:3 and III:5 are healthy carriers or its disease manifestation is associated with place of birth at low altitude. One sibling (III:6), their mother II:4, and maternal grandparents carried the wild-type allele. No DNA was available from further family members. HAPE, high-altitude pulmonary edema; JAK2, Janus Kinase 2; PAH, pulmonary arterial hypertension.

Within the HAPE family one missense mutation in the gene JAK2 [c.1198T>G p.(Cys400Gly)] was identified in four family members (II:3, III:3, III:4, III:5, Fig. 1), which was only present in a single individual in the database gnomAD (>120,000 exomes). The exchange of the highly conserved amino acid cytosine by glycine within the Src Homology 2 domain of this tyrosine protein kinase, could impact ligand binding and was therefore judged maximally pathogenic by 4/4 prediction programs (Align GVGD: C65, PolyPhen2: 1, MutationTaster: damaging, SIFT: 0). Hence, due to its rareness in >120,000 exomes, the location of the JAK2 variant within a functional domain and the judgment of the prediction programs the variant was classified as a likely pathogenic variant (class IV) by us. Since pathogenic variants in JAK2 can be involved in erythrocytosis (Kralovics et al., 2005), hematocrit levels of all family members were measured, which were within the normal range.

The missense mutation of JAK2 was found in the father and 3 of the 4 siblings (Fig. 1). Interestingly, re-entry HAPE occurred not in all variant carriers but only in the two siblings (III:3, III:4). The index patient (III:3) of the family developed additionally to re-entry HAPE PAH at an age of ∼14 years. An elevated sPAP of 55 mmHg was measured during a routine checkup without an acute HAPE episode at ∼3500 m, which fell to 36 mmHg with supplemental oxygen. At an age of 16 years she moved permanently to low altitude to avoid further HAPE episodes. Nevertheless, at an age of 23 years she developed dyspnea (WHO functional class II) and PAH was diagnosed according to the ERS/ESC-PH guidelines 2009 (Galiè et al., 2009). A mildly elevated mean PAP (26 mmHg) was diagnosed using right heart catheterization with normal cardiac index [3.8 L/(min·m²)], normal pulmonary artery wedge pressure (13 mmHg), and only slightly elevated pulmonary vascular resistance (PVR; 2.4 WU). Hence, treatment with PAH-targeted medication (bosentan) was started. Since then she is doing well. Further clinical characteristics of family members are summarized in Table 1.

Clinical and genetic findings in HAPE-S mountaineers

HAPE was confirmed clinically and radiographically in all 64 Caucasian mountaineers, and anthropometric data and clinical measures are shown in Table 2. Arterial oxygen saturation dropped to 73% ± 10%, and sPAP increased to 60 ± 15 mmHg the next morning after arrival at the Capanna Margherita (4559 m). At this time point some of the HAPE-S had already developed HAPE, whereas others were still subclinical. Normal levels at this altitude range between 31.0–43.0 mmHg (Allemann et al., 2000).

Heterozygous likely pathogenic variants were found in 2 out of 64 HAPE-S mountaineers (3%; 95 CI: 1%–13%). One HAPE-S carried a missense mutation [c.1430A>T p.(Lys477Met)] in exon 3 in the CYP1B1 gene. The exchange of the lysine at amino acid position 477 by methionine was characterized as pathogenic by all four prediction programs (Align GVGD: C65, PolyPhen2: 0.52, MutationTaster: damaging, SIFT: 0).

The second HAPE-S mountaineer carried a deletion in the histidine-rich glycoprotein (HRG) gene. The in-frame deletion consisted of 42 base pairs or 14 amino acids leading to a partial loss of the histidine-rich region in exon 7: c.1110_1154del p.(His373_His387del). Due to the unanimous in silico prediction program characterization, the absence in the control database of >120,000 genomes and the location of the mutations within important protein domains we graded these variants following classification guidelines as likely pathogenic (class IV).

In 26 further HAPE-S (41%, 95 CI: 29%–54%), 31 genetic VUS were detected. These were characterized as pathogenic by at least one of the four in silico prediction programs and benign by remaining programs (see Supplementary Table S2). We classified these variants as VUS since they were either not judged pathogenic by all four programs or they were frequent in the control population. No alteration of the canonical splicing pattern on messenger RNA level was identified by PCR analysis for two different VUS. In 36 HAPE-S (56%, 95 CI: 43%–69%) only benign variants (polymorphisms) within PAH or candidate genes were identified.

An average coverage of 636 × was reached across all samples. The samples were sequenced in 10 different NGS runs. On average across the runs, 5.8 Gb were sequenced with base call error rate ≤0.1% of 92.9%, that is, 92.9% of the sequenced bases had a base call accuracy of 99.9%.

Family with re-entry HAPE

In the family described in this study two siblings suffered from re-entry HAPE, which occurs in permanent high-altitude residents returning after some time spent in the lowlands. In general, as also found in this study, children appear to be more often affected from this form of HAPE than adults (Scoggin et al., 1977; Houston et al., 2005) and it is more frequently described in residents from the Andes than the Himalayas (Roach and Schoene, 2002; Baniya et al., 2017). It was speculated that physical activity might be a trigger for re-entry HAPE (Hultgren et al., 1961). The degree of exercise also appears to contribute to ascent HAPE in mountaineers, which seems to be more often reported in men than in women (Hultgren, 1996; Bärtsch et al., 2005). However, both, re-entry and ascent HAPE, have also been documented in women (Hultgren et al., 1961; Hultgren, 1996). Moreover, the pathophysiology has been suggested to be the same as in native lowlanders developing HAPE during ascent (Bärtsch et al., 2005), but direct comparisons remain to be performed. The absence of overlapping genetic contributions could also point to the possibility of different underlying pathobiological mechanisms.

Four members of the family in this study carried a missense mutation in the JAK2 gene. Only the two lowland-born children with the missense mutation, who moved to the highlands between 6 and 8 weeks of age suffered from re-entry HAPE (III:3, III:4). In contrast, the younger sibling (III:5) but also the father (II:3) carrying the variant, who were born and raised at high altitude, never displayed HAPE symptoms, although they also spend several summer vacations in the lowlands. This either implies a reduced penetrance of the missense mutation, or possible epigenetic reprogramming in the two HAPE-S children during a crucial phase of pulmonary artery remodeling. At birth children display muscular pulmonary arteries to divert blood flow as a fetus redirects blood to the placenta away from the nonoxygen perfused lung (Vali and Lakshminrusimha, 2017). Directly after birth, a remodeling process sets in to lower the PVR and to optimize lung perfusion (Vali and Lakshminrusimha, 2017). Thus, the different oxygen levels and saturation of oxyhemoglobin at birth and in the subsequent weeks in the two older siblings could have altered epigenetic regulations predisposing pathogenic variant carriers to excessive HPV and HAPE.

JAK2 encodes a tyrosine kinase and activates signal transducers and activators of transcription (STATs) by phosphorylation. STATs, together with the canonical BMPR2 signaling pathway can activate the expression of target genes (Darvin et al., 2013), see Figure 2. Any disturbance of this pathway by pathogenic variants may foster PAH (Machado et al., 2015). STAT signaling promotes an adequate hypoxia response due to activation of hypoxia-inducible factor 1 alpha (HIF-1α) and the production of nitric oxide (Paulin et al., 2012). Decreased nitric oxide has also been related to HAPE (Busch et al., 2001). In mice, the induced loss of JAK2 led to a repression of endothelial nitric oxide synthase expression (Yang et al., 2013). Hence, the JAK2 gene represented a good candidate gene for PAH and possibly HAPE for the following reasons: first, JAK2 interacts with the BMPR2 pathway (Darvin et al., 2013); second, JAK2 is activated by interleukin-6 (Ni et al., 2004), which is elevated in PH patients and HAPE patients (Humbert et al., 1995; Altamura et al., 2015); third, JAK2 is expressed in human endothelial and smooth muscle cells (Janmaat et al., 2010); and finally JAK2 is involved in the regulation of nitric oxide production (Yang et al., 2013).

OPEN IN VIEWER

FIG. 2.

PAH gene signaling and the JAK2 gene. The BMPR2 pathway is the main signaling cascade disturbed in PAH patients. In healthy subjects, it leads to an increase of cell growth and a decrease of apoptosis. BMPs bind to the BMPR2-Alk1 receptor, which forms a complex with ENG. Phosphorylation passes on signals to SMAD proteins, which form a complex and translocate into the nucleus resulting in altered gene transcription. JAK2 can be activated by various receptors, such as platelet-derived growth factor receptor or growth hormone receptor and subsequently phosphorylates STAT proteins. These can bind the SMAD complex in the nucleus leading to gene transcription. BMP, bone morphogenetic protein; BMPR2, bone morphogenetic protein receptor 2; ENG, endoglin; STAT, signal transducer and activator of transcription.

The index patient of the family progressed to a mild form of PAH, although with a PVR below the current diagnostic cutoff of 3 WU (Galiè et al., 2016; Simonneau et al., 2019). However, her PVR level of 2.4 WU indicates cardiovascular disease. At the time of diagnosis, this cutoff was not yet established and therefore targeted PAH therapy was initiated in accordance with valid PH guidelines (Galiè et al., 2009). A progression of HAPE-S to mild forms of PH has been reported previously in several cases (Naeije et al., 1996; Corvinus et al., 2010; Brill et al., 2012). In a large family with hereditary PAH and a pathogenic variant in the gene BMPR2 the index patient developed HAPE a few years before the diagnoses of PAH was established (Grünig et al., 2000a; Hinderhofer et al., 2014). Due to a co-occurrence of high-altitude PH and HAPE, it has been speculated that PH may predispose to re-entry or ascent HAPE (Das et al., 2004). Thus, HAPE episodes can occur in patients who have a predisposition for PAH. This study shows for the first time that an autosomal dominantly inherited missense mutation could have led to re-entry HAPE in two out of four variant carriers and might also increase the risk for developing PAH.

A familial aggregation of HAPE-S has been previously described in a Han Chinese family with HAPE susceptibility across three generations (Lorenzo et al., 2009; Yang et al., 2017). However, in both studies, the exact genetic cause could not be identified. In the first study, Lorenzo et al. (2009) performed a genome-wide analysis of single nucleotide polymorphisms, which pointed toward the gene of HIF-2α, although no exact causative variant but only linkage with the HAPE phenotype could be demonstrated. Subsequent whole exome sequencing of the same family could neither identify a causative variant in the region coding of HIF-2α (Yang et al., 2017). Instead, 27 different variants were highlighted, which were shared by the Han Chinese HAPE-S family members and not the healthy mother. The variants were assessed by similar in silico prediction programs as used in this study. Of these variants only two missense variants and one deletion represented interesting follow-up candidates. The missense variants were located in the genes OXER1 and CFHR4 and were classified as pathogenic by all three programs used. These variants were, however, present in four and three control subjects with unknown HAPE history in the online data base gnomAD, respectively, reducing their likelihood to be uniquely responsible for HAPE development in this family. Moreover, one in-frame deletion of one amino acid in the gene KCNJ12 was highlighted as pathogenic and absent in the control data base. Thus, the latter variant might represent a better candidate for HAPE development in the Han Chinese family. The described variants by Lorenzo, Yang, and colleagues were distinct from the familial variant identified in this study, which could be due to the different techniques used, the ethnicity of the study participants, or the assumption of various genes contributing to HAPE development.

Pathogenic variant in HAPE-S mountaineers

Two out of 64 HAPE-S mountaineers presented with different likely pathogenic variants. Interestingly, these two variants were not located in the JAK2 gene. These could be risk variants, which might act together with so far unidentified other genetic and nongenetic factors leading to HAPE. One variant was located in the gene CYP1B1. It encodes a monooxygenase, which breaks down steroids, such as estrogens, and is highly expressed in the lung endothelium (Kwapiszewska et al., 2019). In addition, estrogen was also suggested to be a disease modifier for HAPE-S (Luks, 2014).

Additional pathogenic variants in CYP1B1 could act as modifiers for PAH disease penetrance in patients with a pathogenic variant in the BMPR2 gene (West et al., 2008). The messenger RNA of CYP1B1 was highly decreased in female PAH patients with a pathogenic variant in BMPR2 in comparison to carriers without PAH (West et al., 2008). Another potential involvement in PAH and HAPE pathology might be the upregulation of this enzyme by shear stress (Dekker et al., 2002) and hypoxia (White et al., 2012). The enzyme CYP1B1 has been suggested to contribute to vasodilatory response once activated (De Caterina and Madonna, 2009). This response may be hampered by the presence of a likely pathogenic variant reducing its functionality and thus leading to higher pressures.

The second likely pathogenic variant was identified in the HRG gene. HRG protein is found in plasma and binds various small molecules such as plasminogen and thrombospondin, thus modulating processes concerning apoptosis and angiogenesis (Poon et al., 2011). Moreover, HRG inhibits the angiogenic effects of the vascular endothelial growth factor (VEGF) (Olsson et al., 2004). VEGF in turn is known to increase vascular permeability of the endothelial layer and promote HAPE if overexpressed (Kaner et al., 2000). The likely pathogenic variant found in one of the HAPE-S mountaineers was characterized by a 14 amino acid deletion within the histidine-rich region responsible for the antiangiogenic effects of HRG (Poon et al., 2011). VEGF messenger RNA was elevated in HAPE-S during acute HAPE and 3 days after HAPE treatment in comparison to HAPE-free controls (Yuhong et al., 2018). Hence, the lack of this antiangiogenic domain in the HRG protein of the HAPE-S mountaineer may lead to a reduced inhibition of VEGF, resulting in a greater VEGF-induced increase in endothelial permeability and thus contribute to HAPE development. In PAH patients HRG has been identified as a candidate gene by whole exome sequencing (de Jesus Perez et al., 2014). The alteration in the protein may disrupt its effects on plasminogen regulation and thrombosis formation, which play a role in PAH manifestation (Galiè et al., 2016). However, neither of the two carriers of the likely pathogenic variants has developed PAH manifestation. Hence, neither the incidence of HAPE nor the presence of PAH-related pathogenic variants does imply a subsequent PAH manifestation.

The identified pathogenic variants, which could account for an exaggerated HPV and/or increased endothelial permeability, were found in 3% of HAPE-S mountaineers in our sample. This suggests that only a small fraction of HAPE cases shares a genetic basis with PAH. However, the genetic predisposition of HAPE is an interesting model for the development of PH and should be assessed further on a functional, genetic, and epigenetic basis to clarify mechanisms that might explain the high recurrence rate of HAPE in susceptible individuals. Moreover, in this study no overlapping pathogenic variants could be identified between the re-entry HAPE family and the HAPE-S mountaineers. Similarly, in the Han Chinese family with re-entry HAPE described by Lorenzo, Yang, and colleagues, none of our variants was identified, but variants in 18 other genes were highlighted (Lorenzo et al., 2009; Yang et al., 2017). This indicates that various genes may contribute to HAPE susceptibility, which could also differ between ethnic groups and types of HAPE (re-entry vs. ascent HAPE).