Transcriptomic profile of cationic channels in human pulmonary arterial hypertension

The dysregulation of K+ channels is a hallmark of pulmonary arterial hypertension (PAH). Herein, the channelome was analyzed in lungs of patients with PAH in a public transcriptomic database. Sixty six (46%) mRNA encoding cationic channels were dysregulated in PAH with most of them downregulated (83%). The principal component analysis indicated that dysregulated cationic channel expression is a signature of the disease. Changes were very similar in idiopathic, connective tissue disease and congenital heart disease associated PAH. This analysis 1) is in agreement with the widely recognized pathophysiological role of TASK1 and KV1.5, 2) supports previous preliminary reports pointing to the dysregulation of several K+ channels including the downregulation of KV1.1, KV1.4, KV1.6, KV7.1, KV7.4, KV9.3 and TWIK2 and the upregulation of KCa1.1 and 3) points to other cationic channels dysregulated such as Kv7.3, TALK2, CaV1 and TRPV4 which might play a pathophysiological role in PAH. The significance of other changes found in Na+ and TRP channels remains to be investigated.


KCNK1
TWIK www.nature.com/scientificreports/ forms of impaired regulation of channel activity. Deleterious genetic variants in the KCNK3, KCNJ8, ABCC8 and ABCC9 genes (which encode the TASK1 and K IR 6.1 channels and the auxiliary subunits SUR1 and SUR2, respectively) and possibly in KCNA5 (encodes for K V 1.5 channels) have been associated to PAH, representing paradigmatic examples of channelopathies 8 . Downregulation of K + channel expression is also a typical feature of the disease. Several K + channels have been reported to be downregulated in animal models of PAH including K V 1.1, K V 1.5, K V 1.6, K V 2.1, K V 3.1b, K V 7.1, K V 7.4 and TASK1 while others were upregulated such as K V 11.1, K Ca 1.1, K Ca 3.1 and K IR 6.1 9,10 . Some studies have also been carried out to analyze the alteration of specific cationic channels in patients with PAH. Notably, Yuan et al. demonstrated attenuated expression of Kv1.5 mRNA in lungs from PAH patients 9,10 and reduced Kv1.1 and Kv4.3 channel mRNA have also been shown 4 . Downregulation of TASK1 expression and activity has also recently been demonstrated in PASMC from PAH patients 11 .
In the present study, the expression of the cationic channel encoding genes in lungs from patients with PAH was analyzed using a transcriptomic approach.

Methods
Data was downloaded from the publically available Gene Expression Omnibus (GEO) dataset GSE113439 (https:// www. ncbi. nlm. nih. gov/ gds/). This dataset contained the gene expression profile of samples from explanted lungs of fifteen patients with pulmonary hypertension who underwent lung transplantation. Specimens were obtained from the peripheral area at the base of each organ. Lung samples obtained from the region of normal healthy tissue flanking lung cancer resections from eleven patients without pulmonary hypertension or left-side heart disease were used as controls. The PAH group (class 1 pulmonary hypertension) included six patients with idiopathic PAH (IPAH), four patients with PAH secondary to connective tissue disease (CTDPAH) and four patients with PAH secondary to congenital heart disease (CHDPAH). There was only one patient with chronic thromboembolic pulmonary hypertension (class 4 pulmonary hypertension) in the dataset which was not included for the present study. The sample preparation, RNA extraction, cDNA synthesis, microarray analysis, gene annotation, preprocessing and data curation and normalization followed standard procedures and are described in detail in the original paper and its supplement 12 . Briefly, microarrays were processed with Genechip® Human Gene Array and scanned on Affymetrix Genechip Scanners. The expression in the control and the PAH samples was compared for the whole transcriptome by the GEO2R analysis tool (https:// www. ncbi. nlm. nih. gov/ gds/) using the Benjamini & Hochberg approach for false discovery rate to calculate the adjusted p value. The principal component analysis was performed using the Past software (ver3.21, Oslo, Norway). The changes in channel expression between the different PAH subgroups (log[fold change] versus controls) were compared by regression analysis using Prism (Ver 7.04). P < 0.05 was considered statistically significant.

Results and discussion
A total of 22,195 mRNA products were identified and 12,408 (56%) of them were significantly dysregulated in PAH compared to control lungs (adjusted p < 0.05). This large transcriptomic change is consistent with the advanced stage of the disease in these PAH patients undergoing transplant, which show complex lesions and accumulation of different PASMC and endothelial cells (PAECs), fibroblasts, myofibroblasts and inflammatory cells 2 . Therefore, the hypertensive lung transcriptome must reflect not only changes in the regulation of gene transcription in the different specific cell types but also changes in cell phenotypes and a different proportion of cell types within the lung. In fact, PAH cells show a hyperproliferative phenotype and share several characteristics with cancer cells 13 . The Vulcano plot shows an apparently symmetrical distribution of the changes in expression (blue dots in Fig. 1A) with 40.2% of these mRNA products significantly upregulated and 59.8% downregulated.
The different families and subfamilies of mRNAs encoding the pore forming alpha subunits of cationic channels identified are shown in Fig. 1B. K + channels comprise the family of cationic channels with the highest relative abundance, followed by TRP channels; and to a lesser extent Ca 2+ , Na + and CNG/HCN channels (Figs. 1B, 2). As expected, there were large differences in the relative expression of the different cationic channel genes in control lungs (Fig. 2), which gives a rough estimate of the importance of each channel. Among the 143 mRNA products identified as cationic channels, 66 of them (46.2%) were dysregulated in PAH with a strongly asymmetrical distribution in the Vulcano plot (red dots in Fig. 1A), with most of them downregulated (83.1%) and only 16.9% of them upregulated. This broad decrease in channel expression might be partly attributed to a relative loss or a phenotypic change in the electrically excitable cells in the lung in which voltage-gated cation channels are abundant, including smooth muscle, neurons and endocrine cells. The principal component analysis (PCA) for the whole transcriptome previously published 12 indicated a complete separation between PAH subjects and normal controls. Herein, the unsupervised PCA applied to the 143 channel transcripts also clearly separated controls from PAH patients (Fig. 1C). This indicates that dysregulated cationic channel expression is a signature of the disease. However, the three different forms of PAH analyzed, i.e. IPAH, CTDPAH and CHDPAH, were indistinguishable in the PCA analysis. Two samples from the CTDPAH group were clearly separated from the rest of the PAH patients and further away from the controls. Moreover, the plots of the changes in channel gene expression in IPAH vs CTDPAH, IPAH vs CHDPAH and CTDPAH vs CHDPAH, shows an excellent correlation ( Fig. 1D, P < 0.001 for the three pairs of comparisons). That confirms that the changes in channel gene expression were very similar in the three forms of PAH. The data was very internally consistent. The exceptions to this rule were only two genes, Ca V 1.3 and Na V 2.1, which were downregulated specifically in CTDPAH (supplementary Fig. 1). If confirmed, these might represent biomarkers of this specific subtype of PAH. The heat map for the genes encoding cationic channel that displayed significant difference (adjusted p < 0.05) in expression is shown in Fig. 3.
The group of K + channels is the largest family of cationic channels ( Fig. 2A). The differential expression of K + channels in PAH with 33 (44%) mRNAs encoding K + channels significantly downregulated and 5 (7%) mRNAs upregulated is shown in Fig. 4A. Because K + conductance is essential to maintain a normally polarized cell www.nature.com/scientificreports/ membrane, this reduced expression of K + channels explains the depolarized Em observed in PASMC in PAH leading to increased contractility and cell proliferation 9 . The analysis of the expression of each channel family ( Fig. 4A-E) shows that the K + , Ca 2+ , TRP and CNG/HCN superfamilies of cationic channels present an asymmetrical distribution of their expression in PAH in the Vulcano plots with a majority of their members downregulated. The Na + channel family with a smaller number of members shows a nearly symmetrical plot. The data with significant changes are shown in supplementary Figs. 2-5.
The K V channel family contributed the largest number of dysregulated K + channel genes (3 up-and 20 downregulated, Fig. 4A). Some of them have been previously reported to be downregulated in PAH including K V 1.5 channels (KCNA5) which are key regulators of Em and are inhibited by vasoactive factors involved in PAH such as 5-HT, endothelin-1, thromboxane A 2 and hypoxia [14][15][16] . In fact, K V 1.5 downregulation is widely recognized as a hallmark of the disease in both humans and animal models 10,17,18 . Consistent with previous reports, other  were previously shown to be reduced in vitro by hypoxia in animal models of PAH 9,18 were unchanged. However, the electrically-silent K V 9.3 channels, which form a functional heteromer with K V 2.1, were downregulated in the present and in the vitro study 18 . Several other K V channels such as K V 2.2, K V 3.2, K V 3.3, K V 3.4, K V 5.1, K V 6.1, K V 6.2, K V 6.3, K V 8.1 and K V 12.2, that have been previously reported to be expressed in the pulmonary arteries 19 but whose functional importance is unclear, were also modestly downregulated. In contrast, K V 4.3 previously reported to be downregulated in vitro by hypoxia, as well as K V 4.2, both of which generate transient currents, were found to be upregulated in PAH patients. On the other hand, the expression of K V 11.1 channels, also known as hERG channels and involved in heart repolarization, was increased as measured by immunohistochemistry  www.nature.com/scientificreports/ in the pulmonary arteries of humans affected by chronic obstructive pulmonary disease-associated pulmonary hypertension (class 3) 20 but were significantly downregulated in the three forms of PAH in the present study.
In recent years, a key role of K V 7 channels (KCNQ1-5) in the control of pulmonary vascular tone has been demonstrated [21][22][23][24] . K V 7.1, K V 7.3, K V 7.4 and K V 7.5 mRNAs are known to be expressed in rat pulmonary arteries 24,25 but K V 7.3 was not detected in vascular smooth muscle cells. Based on the effects of K V 7 activators flupirtine and retigabine, these channels have been proposed as therapeutic targets for PAH 23,24 . In control human lungs, the five K V 7 were expressed, with K V 7.5 showing the lowest levels ( Fig. 2A). Downregulation of K V 7.1, K V 7.4 and K V 7.5 have been shown previously in animal models of PAH 21,23,24,[26][27][28] . Herein K V 7.1, K V 7.2 and K V 7.4 were downregulated in human PAH (Fig. 4A). Interestingly, K V 7.3 channels were upregulated in the three forms of human PAH seemingly compensating for the loss of the other K V 7 channels. This may also help to support the preserved Kv7 function in PAH 24 . Because K V 7.3 are sensitive to flupirtine and retigabine, it would be worth analyzing the role of these channels in the context of PAH.
K IR are expressed in vascular smooth muscle and endothelial cells 9,10 . The only change reported in PAH so far is an increase in K IR 6.1, which forms the vascular K ATP channel in the monocrotaline rat model of PAH 26 . The present results indicate that K IR 6.1 was strongly expressed in the lungs ( Fig. 2A) but unchanged by PAH and that most other K IR are unchanged as well, except K IR 2.2-K IR 2.4 and K IR 3.1 that were modestly downregulated (Fig. 4A).
K Ca channels are activated by cytosolic Ca 2+ . The K Ca 1.1 channel which is responsible of the large conductance K Ca current (BKCa), was strongly upregulated in lungs from PAH patients (Fig. 4A). Similarly, these channels were upregulated in laser microdissected pulmonary arteries from IPAH patients 29 , suggesting the utility of BKCa openers for pharmacological interventions in PAH. On the contrary, K Ca 2.1 which carries a small conductance Ca 2+ -activated K + channel (SKCa) was modestly downregulated.
K 2P channels, also referred to as leak or background conductance channels, comprise a large family of voltageindependent K + channels that control resting Em in smooth muscle cells. These K 2P channels are classified into six subfamilies according to their sequence and functional properties: the weak rectifying group (TWIK1, TWIK2, and K2P7.1), the mechano-gated group (TREK1, TREK2, and TRAAK), the acid-sensing group (TASK1, TASK3, and TASK5), the alkaline-sensitive group (TALK1, TALK2, and TASK2), the halothane-sensitive group (THIK1 and THIK2), and the spinal cord-expressed TRESK channel. Among this family, the mRNA of TASK1, TASK2, THIK1, TREK2 and TWIK2 are reported to be expressed in pulmonary arteries, playing a role in pulmonary vascular tone 30 . The present data also indicate a high expression of the mRNA of TALK2 (also named as TASK4), TASK5 and THIK2 in the whole lung ( Fig. 2A). Remarkably, TASK1 has gained a reputation as an important player in PAH. In fact, loss of function mutations in KCNK3 (TASK1) were the first channelopathy described in PAH 8,31 . Herein, the present results confirm the downregulation of TASK1 reported in both human and animal models of PAH 9-11 . In fact, TASK1 was the second most highly expressed K + channel and the strongest downregulated channel in the channelome. Interestingly, other K 2P channels were also markedly downregulated, including TALK2, TWIK2 and TASK5. TALK2 heterodimerizes with TASK1 32 and, thus, it might play a role together with TASK1 in controlling Em and tone in pulmonary arteries. Based on experiments carried in TWIK2 knockout mice, Pandit et al. 33 suggested that these channels are involved in PAH. This concept was challenged by Lambert et al. 4 , which showed that mRNA expression of TWIK2 was unchanged in pulmonary arteries from IPAH patients. The present data, however, would argue in favor of the former possibility. The physiological role of TASK5 is unclear as it produces a weak current in transfected mammalian cell lines 34 . Other members of the family that were modestly downregulated include THIK1, THIK2, TRAAK1 and TASK3 and none of the channels of this family were upregulated.
Ca V channels are activated by depolarization and the subsequent elevation in cytosolic Ca 2+ constitutes the main mechanism for the activation of electrically excitable cells, including PASMC, and some non-excitable cells 35 . Pulmonary artery contraction depends on Ca 2+ -dependent pathways (mediated mainly by L-type Ca 2+ channels) and Ca 2+ -independent pathways (mediated mainly by Rho kinase activation), and the relative role of each one may vary with age, the vasoconstrictor agent involved and the hypertensive state 36,37 . The downregulation of K + channels in PAH as described above and the subsequent depolarization increases the activation of these voltage-gated Ca V channels promoting contraction and proliferation. The Ca V 1 subfamily (Ca V 1.1-Ca V 1.4) corresponds with the voltage-dependent channels known as L-type channels which are sensitive to classic Ca 2+ channel blockers (verapamil, diltiazem and dihydropyridines such as nifedipine or amlodipine) 7 . These drugs constitute a first line therapy for PAH, but unfortunately only 10% of the patients respond to these drugs at clinically tolerable doses. Ca V 1.2 and Ca V 1.3 showed the highest expression in the lungs (Fig. 2B). Downregulation of Ca V 1.2 has been previously reported in arteries from a model of persistent pulmonary hypertension of the newborn 38 . However, hypoxia upregulated these Ca V 1.2 channels in mouse lungs 39 . The Vulcano plot shows that in PAH most Ca V channels are modestly downregulated including the L-type Ca 2+ channels Ca V 1.1, Ca V 1.2 and Ca V 1.4 (Fig. 4B). The downregulation of Ca V 1.x and the upregulation of Rho kinase in these patients 12 might account for a higher role of Ca 2+ -independent pathways for pulmonary artery contraction 37 . It may also account for the low proportion of responders to L-type Ca 2+ channel blockers. We also analyzed the relationship of the expression of the different Ca V 1.x with the expression of K V channels; the significant associations are shown in supplementary Fig. 6. It is interesting that in control patients there is an inverse association for certain K V and the different Ca V 1.x channels. Curiously, each K V channel is only associated with a single Ca V 1.x channel: K V 1.3, K V 4.2 and K V 6.3 with Ca V 1.1; K V 12.1 with Ca V 1.2; K V 1.2, K V 3.1, K V 4.1, K V 5.1 and K V 6.2 with Ca V 1.3 and finally K V 4.3 with Ca V 1.4 (black symbols in supplementary Fig. 6). On the contrary, in the lung from PAH patients these inverse associations were lost and all significant associations were direct (supplementary Fig. 6, blue symbols). The cause-effect relationships and their meaning are unclear. It is possible that K V activity negatively regulates Ca V 1.x expression and that this relationship is lost with the missing excitable phenotype in PAH cells. Ca V  www.nature.com/scientificreports/ correspond to the T-type channels. Hypoxia upregulated Ca V 3.2 channels in mouse lungs 39 while these channels were decreased in patients with PAH (Fig. 4B). ORAI channels are store-operated Ca 2+ entry channels which are activated by depletion of Ca 2+ in the endoplasmic reticulum and do not share a common structure with voltage-gated like channels. It has been suggested that ORAI1 channels play a role in PAH 40 but evidences are weak so far. The expression of ORAI2 was found increased in PASMC in culture from patients with iPAH 40 . ORAI2 was also upregulated in endothelial cells from PAH patients in a single cell transcriptomic analysis 41 . ORAI1 is involved in the transition of PASMC from a contractile to a proliferative phenotype 42 . However, the three ORAI genes were downregulated in the whole lungs from PAH patients in the present study (Fig. 4B) suggesting that its inhibition is not an appealing therapeutic strategy.
Na V channels play a major role in regulating neuron excitability. Nonetheless, human pulmonary arteries expresses multiple Na V. channels 4,43 although their functional role is unclear. The present study shows that the mRNA expression of Na V 2.1 predominates in the lung (Fig. 2C). In lungs from PAH patients there was a strong upregulation of Na V 1.1, Na V 1.3 and Na V 1.7 (Fig. 4C). In contrast, the sodium leak channel, Na Vi 2.1, a voltageinsensitive member of the Na V family, was downregulated in PAH lungs. The significance of these findings deserves further investigation.
The TRP channel family comprises seven subfamilies of cationic channels with diverse biophysical properties and physiological functions: the TRPC ("canonical"), the TRPV ("vanilloid"), the TRPM ("melastatin"), the TRPP ("polycystin"), the TRPML ("mucolipin"), the TRPA ("ankyrin") and the TRPN ("nompC", no mechanoreceptor potential C) families. TRP channels contribute to the modulation of cytosolic Ca 2+ directly by carrying Ca 2+ entry or indirectly by modulating the resting Em 4,44 . TRPC1, TRPC4 and TRPC6 are the major TRPC channels expressed in rat, mouse, and human pulmonary arteries 3,45 and in the human lung (present results). TRPV channels, except TRPV2, are involved in thermic sensitivity. However, TRPV1, TRPV2, TRPV3 and TRPV4 mRNA were also detected in pulmonary arteries 44 and in whole lung (present results). In contrast to other TRPs, TRPM4 and TRPM5 are impermeable to Ca 2+ and its role in the lung is unknown. TRPM6 and TRPM7 are closely related; together they form a heterotetrameric channel permeable to divalent (Ca 2+ , Mg 2+ , Zn 2+ ) and monovalent cations and play a role in Mg 2+ homeostasis 44 . The mRNA of TRPM2, TRPM3, TRPM4, TRPM7 and TRPM8 were detected in pulmonary arteries 44 . Similarly, in the whole lungs analyzed in the present study, the highest expression was found in TRPM7, TRPM4, TRPM2 and TRMP5 channels. Among other TRP channels belonging to the TRPML, TRPP, TRPA and TPCN families, little is known about their expression and function in the lungs. The present study, however, shows a strong expression of the TRPP channel PKD2, the TRPML channels MCOLN1, MCOLN2, and MCOLN3 and the TPCN channels TPCN1 and TPCN2.
A number of changes were found in the literature about the expression of TRPs in pulmonary arteries of PAH patients 44 . TRPC1, TRPC4 and TRPC6 isoforms are upregulated in rat BMP2 knockdown PASMC, leading to increased Ca 2+ entry and increased human PASMC proliferation and migration 46 . TRPC6 and TRPC3 mRNA and protein were also upregulated in lungs from IPAH patients 47 . Moreover, TRPC6 gene deletion suppresses chronic hypoxia-induced PH in mice 48 . All these findings strongly suggest a role for TRPC6 in PAH. However, the present results do not reproduce the upregulation of any TRPC mRNAs. The only significant change in this subfamily was a modest decrease in TRPC2 expression, probably irrelevant because TRPC2 is a pseudogene in humans. Both TRPV1 and TRPV4 were found to be upregulated in cultured PASMC from PAH patients 49 . In contrast, in the present PAH cohort there was a downregulation of lung TRPV1, TRPV2 and TRPV4. Interestingly, TRPV4 channels mediate a rise in endothelial intracellular Ca 2+ concentration, leading to the activation of KCa3.1 (IKCa) and KCa2.3 (SKCa) currents and activating endothelial-dependent hyperpolarization (EDH) and relaxation in pulmonary arteries. The downregulation of TRPV4 is consistent with reduced EDH in pulmonary arteries from hypoxic mice 50 . In the TRPM family, TRPM7 was downregulated in cultured PASMCs from PAH patients and in rats with hypoxia-induced PAH and its inhibition exacerbates PAH 40 . Similarly, mice with a deletion of the kinase domain in TRPM7 channels in heterozygosis, show a severe systemic hypertensive phenotype induced by angiotensin II 51 . However, herein, TRPM7 and, to a minor extent, TRPM6 were upregulated in PAH. Conversely, TRPM5 was modestly reduced and TRPM4 was strongly downregulated. Other TRP channels downregulated include MCOLN1, TPCN1 and TPC2. Taken together, there was little correlation of the present findings regarding changes of TRP channel expression in PAH between the present data compared to isolated PASMC in animal models. This may reflect that TRP expression is quantitatively important in lung cells different from PASMC.
The CNG channels are permeable to Ca 2+ and/or Na + and are gated by cyclic nucleotides. They are involved in fluid absorption and CNGA2 has been reported to regulate Ca 2+ in bovine PAECs 52 . CNGA3, CNGA4, CNGB2 and to a minor extent CNGA2 were expressed in the control lungs (Fig. 2E) and the CNGA3 and CNGA4 were downregulated in PAH lungs (Fig. 4E). HCN channels are permeable to Na + and/or K + and are activated by membrane hyperpolarization and by cyclic nucleotides 53 . They regulate neuronal excitability, pain and sinoatrial and atrioventricular node activities but their physiological role in the lung is unknown. Downregulation of the HCN4 channel in the atrioventricular node was observed in the monocrotaline rat model of PAH 54 . All isoforms of HCN (Fig. 2E) were present in the lung and HCN3 and HCN4 were moderately downregulated in PAH lungs (Fig. 4E).
A limitation of this study is that control samples were from neoplasia-free areas of lungs from patients with lung cancer, which might obviously affect the results. K + channels, including Kv1.5, have been implicated in the abnormal growth of lung cancer cells 55 . K V 1.5 expression negatively correlated with tumor grade; the higher the histologic tumor grade the lower the K V 1.5 expression. It is unknown whether healthy cells from cancer patients also show dysregulated K + channels. However, if anything these issues should hide the differences observed herein. Moreover, PAH samples were obtained from advanced-stage, severe ill patients undergoing transplantation. Then, some changes found may be the consequence of a critical health condition rather than a pathophysiological causative factor of the disease. An additional important limitation of the present study is www.nature.com/scientificreports/ that mRNA expression is not always translated into protein and even when protein is present it might not traffic to the membrane or may not form a functional channel. However, the study has the strength of using clinically relevant samples in contrast with much of the previous work done on cell culture from PAH patients or controls with or without exposure to hypoxia, which can be problematic because channel expression can be altered by culture passage. Some studies have analyzed specific channels genes in PASMC or PAECS or in pulmonary arteries. These include cells from manually dissected medium and small pulmonary arteries, illustrative of vascular changes but which may not be very representative of the complex lesions present in the microcirculation in PAH lungs, such as onion-skin lesions, plexiform core lesions and dilation lesions 2 .
In conclusion, the dysregulation of cationic channel expression is a signature of PAH with multiple channels being downregulated and a few of them upregulated. Changes were very similar in IPAH, CTDPAH and CHDPAH. When compared to changes reported in the literature, this analysis of the cationic channel transcriptome in PAH: 1) is in agreement with previously published changes in the expression of channels which play an important and widely recognized pathophysiological role in the disease such as the downregulation of TASK1 and K V 1.5 in humans and animal models; 2) supports previous preliminary reports pointing to the dysregulation of several K + channels including the downregulation of K V 1.1, K V 1.4, K V 1.6, K V 7.1, K V 7.4, K V 9.3 and TWIK2 and the upregulation of K Ca 1.1; 3) argues against the proposed role of other K + channels including K V 1.2, K V 2.1, K V 4.2, K V 4.3, K V 11.1, K IR 6.1 and ORAIs and 4) points to other novel cationic channels dysregulated such as Kv7.3, TALK2, Ca V 1 and TRPV4 which might play a pathophysiological role in PAH. The significance of other changes found in Na + and other TRPs remains to be investigated.

Data availability
The data that support the findings of this study are available in Gene Expression Omnibus (GEO) at https:// www. ncbi. nlm. nih. gov/ gds/, reference number GSE113439.