ATP13A3 variants promote pulmonary arterial hypertension by disrupting polyamine transport

Abstract Aims Potential loss-of-function variants of ATP13A3, the gene encoding a P5B-type transport ATPase of undefined function, were recently identified in patients with pulmonary arterial hypertension (PAH). ATP13A3 is implicated in polyamine transport but its function has not been fully elucidated. In this study, we sought to determine the biological function of ATP13A3 in vascular endothelial cells (ECs) and how PAH-associated variants may contribute to disease pathogenesis. Methods and results We studied the impact of ATP13A3 deficiency and overexpression in EC models [human pulmonary ECs, blood outgrowth ECs (BOECs), and human microvascular EC 1], including a PAH patient–derived BOEC line harbouring an ATP13A3 variant (LK726X). We also generated mice harbouring an Atp13a3 variant analogous to a human disease–associated variant to establish whether these mice develop PAH. ATP13A3 localized to the recycling endosomes of human ECs. Knockdown of ATP13A3 in ECs generally reduced the basal polyamine content and altered the expression of enzymes involved in polyamine metabolism. Conversely, overexpression of wild-type ATP13A3 increased polyamine uptake. Functionally, loss of ATP13A3 was associated with reduced EC proliferation, increased apoptosis in serum starvation, and increased monolayer permeability to thrombin. The assessment of five PAH-associated missense ATP13A3 variants (L675V, M850I, V855M, R858H, and L956P) confirmed loss-of-function phenotypes represented by impaired polyamine transport and dysregulated EC function. Furthermore, mice carrying a heterozygous germline Atp13a3 frameshift variant representing a human variant spontaneously developed a PAH phenotype, with increased pulmonary pressures, right ventricular remodelling, and muscularization of pulmonary vessels. Conclusion We identify ATP13A3 as a polyamine transporter controlling polyamine homeostasis in ECs, a deficiency of which leads to EC dysfunction and predisposes to PAH. This suggests a need for targeted therapies to alleviate the imbalances in polyamine homeostasis and EC dysfunction in PAH.


Introduction
Pulmonary arterial hypertension (PAH) is a progressive vascular disorder characterized by the narrowing and obliteration of small pre-capillary lung arterioles. 1 Endothelial cell (EC) dysfunction, proliferation of mesenchymal cells in the vascular wall, and aberrant inflammation 1,2 contribute to this pathological process.Despite the availability of licensed therapies, the survival of a patient with PAH remains poor, necessitating new targeted treatments.
The identification of heterozygous germline mutations in the bone morphogenetic protein Type II receptor (BMPR2) gene 3,4 and the more recent identification of loss-of-function mutations in other BMP pathway components 5 have underpinned potential PAH therapies to enhance BMP signalling. 6However, some rare PAH-related genes appear distinct from the BMP pathway, 5,7 suggesting additional mechanisms underlying the pathobiology of PAH that may be informative for alternative therapies.
In a European-wide PAH cohort study, we identified 11 rare heterozygous ATP13A3 variants with protein-truncating variants overrepresented (6 of 11), suggesting a loss of function in PAH. 7[10] Although ATP13A3 is expressed in various cell types, including pulmonary vascular cells, pulmonary macrophages, and dendritic cells, 7,11,12 its function remains unclear.ATP13A3 is a member of the P5B-type ATPase family (ATP13A2-5), and ATP13A2 has recently been identified as a polyamine transporter. 13ATP13A3 has close homology to ATP13A2.Furthermore, recent studies have shown that ATP13A3 mutations account for the polyamine uptake deficiency in CHO-MG cells, 14 and ATP13A3 facilitates polyamine transport in human pancreatic cancer cells, 15 strongly implicating ATP13A3 as a polyamine transporter.
Cellular polyamine levels are tightly regulated through the integration of their biosynthesis/catabolism and their transport, and disruption of these pathways can lead to diseases. 16,17In this study, we established that ATP13A3 mediates cellular polyamine uptake in human vascular ECs, whereas PAH-associated ATP13A3 variants reduce polyamine transport.Loss of ATP13A3 leads to PAH-associated phenotypes in pulmonary arterial ECs and in mice, harbouring a PAH-associated Atp13a3 frameshift variant (P452Lfs).Collectively, our data explain the impact of ATP13A3 variants in PAH and suggest that dysregulated polyamine homeostasis may contribute to its pathobiology.

Methods
Key protocols are described here, and additional detailed protocols are described in the Supplementary Material.

Animals
All animal procedures were performed in accordance with the Home Office Animals (Scientific Procedures) Act (1986) and were approved under Home Office Project Licence 70/8850.All studies were approved locally and conformed to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes.The Atp13a3 genetically modified mouse, C57BL/6Ntac-Atp13a3 em2H /H (MGI: 6450011), designated Atp13a3 P452Lfs , was generated at MRC Harwell using CRISPR-Cas9 editing to introduce a 1-nt deletion, resulting in a frameshift and termination codon after seven more amino acids (P452LfsTer7), equivalent to the human PAH variant P456Lfs. 7

Haemodynamic assessments of mice
Cardiac catheterization was performed by the closed-chest technique, as described previously. 18Measurements of right ventricular systolic pressure (RVSP) were performed under isoflurane anaesthesia (2.0-2.5% isoflurane, 100% oxygen 2 L/min) in spontaneously breathing animals.In the same animals, systolic blood pressure in the aorta was measured.Mice were then killed by exsanguination while still under anaesthesia.
Hearts were excised, and the right ventricle (RV) was dissected, weighed, and then fixed in 10% neutral-buffered formalin.The lungs were inflated with 10% neutral-buffered formalin and harvested for histological analyses as described in the Supplementary Material.
Echocardiographic ultrasound measurements of heart rate, RV dimensions, and PA pressure surrogates were conducted in spontaneously breathing 6-month-old male mice under isoflurane anaesthesia using an ultrasound machine (Vevo 3100 System; FUJIFILM VisualSonics, Amsterdam, The Netherlands) equipped with a 40 MHz linear array transducer.
Human blood outgrowth ECs (BOECs) were isolated from 40 to 80 mL of blood, as previously described. 19BOEC lines were grown in EGM2 with the addition of 10% FBS, antibiotics/antimycotics, and omission of heparin.For experiments involving BOEC generation, all donors provided informed written consent in accordance with the Declaration of Helsinki under human study 07/H0306/134 (Cambridgeshire 3 Research Ethics Committee) or REC-17/LO/0563 (ATP13A3-LK726X variant carrier).Demographic and variant information for the BOECs used in this study is specified in Supplementary material online, Table S1, and genomic information for the LK726X variant is detailed in Supplementary material online, Table S2.Both hPAECs and BOECs were used for experiments between Passages 4 and 7.

Cellular transfection and transduction
Cells were transfected with ATP13A3 siRNA and ATP13A3 expression plasmids or transduced with lentiviral expression particles, 20 as described in the Supplementary Material.All variants studied are detailed in Supplementary material online, Table S2.

Measurement of cellular polyamines
Aqueous metabolites were extracted from cell lysates and analysed for polyamine content by liquid chromatography-mass spectrometry (LC-MS), as described in the Supplementary Material.

BODIPY-labelled polyamine uptake assay
BODIPY-tagged spermine (SPM-BDP), spermidine (SPD-BDP), and putrescine (PUT-BDP) were synthesized, as previously described, 21 and dissolved in 0.1 M 3-morpholinopropane-1-sulfonic acid (MOPS), pH = 7.0 (AppliChem, A1076, Darmstadt, Germany).The uptake of polyamine-BDP in HMEC-1 was determined by flow cytometry.HMEC-1 cells were seeded in 12-well plates at 300 000 cells/well and left to attach overnight.After determining the kinetics of uptake to ensure that the cells were in the linear phase, they were incubated with the respective polyamine-BDP concentration (5 µM, if it was a single concentration) for 30 min after which they were trypsinized, and centrifuged at 300×g, and the pellet was washed with cold Dulbecco's phosphate-buffered saline (PBS) solution without calcium or magnesium (Sigma, D8537).The pellets were then re-suspended in 1% bovine serum albumin/PBS.Polyamine-BDP uptake was determined by flow cytometry on a BD FACSCanto™ II instrument (BD Life Sciences, Franklin Lakes, NJ), with 10 000 events recorded per treatment.

Statistical analysis
The data are presented as mean ± standard error of the mean (SEM) and are analysed using GraphPad Prism 7 (GraphPad Software, Boston, MA).All presented data are n = 3 (unless mentioned otherwise), where n represents the number of independent repeats.The Kolmogorov-Smirnov test was performed for assessing normality, and the data were analysed by one-way analysis of variance (ANOVA) with post hoc Tukey's honestly significant difference (HSD) analysis, one-/two-way ANOVA followed by multiple comparisons using Dunnett's/Tukey's post hoc tests, or an unpaired two-tailed Student's t-test as indicated.A value of P < 0.05 was considered statistically significant.

ATP13A3 is a polyamine transporter localized to recycling endosomes
hPAECs localized primarily to a perinuclear region with lower nuclear staining (Figure 1A).Both Green fluorescent protein (GFP)-tagged (Figure 1B and C ) and endogenous (see Supplementary material online, Figure S1) ATP13A3 in HMEC-1 cells colocalized mainly with Rab11, a Ras-like GTPase critically important in vesicle trafficking, although some colocalizations were also observed with EAA, LAMP1, and Rab7 (Figure 1B and C and Supplementary material online, Figure S1).These data suggest ATP13A3 shows a general expression pattern in the endolysosomal system, with the highest colocalization observed in recycling endosomes (Rab11).We hypothesized that ATP13A3 mediates polyamine transport in primary human ECs.ATP13A3 siRNA (siATP13A3) in hPAECs reduced ATP13A3 expression without affecting the expression of ATP13A1-2 (see Supplementary material online, Figure S2A-C).We could not detect ATP13A4-5 expression in these cells.siATP13A3 significantly reduced the basal cellular PUT, SPM, and SPD contents in hPAECs, while only the increase in PUT content was significantly attenuated upon exogenous polyamine supplementation (Figure 2A).For validation, we stably silenced ATP13A3 in HMEC-1 cells (HMEC-1 miATP13A3 ) using three lentiviral micro-RNAs (miRNA), targeting different regions of the ATP13A3 mRNA (Figure 2B and Supplementary material online, Figure S2D and E).Like in hPAECs, basal PUT, SPD, and SPM contents were reduced in HMEC-1 miATP13A3 cells (see Supplementary material online, Figure S2F) and uptake of PUT-BDP and SPD-BDP, but not SPM-BDP was impaired (Figure 2C).Furthermore, confocal imaging confirmed lower PUT-BDP uptake in HMEC-1 miATP13A3 , with the internalized polyamine predominantly confined to punctae (Figure 2D).In conclusion, ATP13A3 determines polyamine uptake, redistribution, and homeostasis in ECs.

ATP13A3 levels alter the expression of polyamine biosynthesis pathways
Polyamine homeostasis is maintained through the integrated functions of polyamine transporters and polyamine metabolism enzymes (see Supplementary material online, Figure S3).Ornithine decarboxylase (ODC) converts ornithine into PUT to initiate polyamine biosynthesis and is tightly regulated by cellular polyamine levels. 17ODC protein levels increased without altering mRNA expression in siATP13A3-transfected hPAECs or BOECs (Figure 3A and B and Supplementary material online, Figures S4 and S5A-C).

ATP13A3 deficiency leads to pulmonary artery endothelial dysfunction
Dysregulated proliferation and increased apoptosis and permeability of ECs contribute to the pathobiology of PAH. 1,2We previously reported that ATP13A3 knockdown in BOECs impaired their proliferation and increased apoptosis in reduced serum. 7Here, we established by cell counting that siATP13A3 reduced hPAEC proliferation (Figure 4A), which was associated with reduced mRNA expression of cyclins E (CCNE1), A (CCNA1), and B (CCNB1), suggesting impaired cell cycle progression (Figure 4B).Supplementation with 10 µM PUT, SPD, or SPM promoted hPAEC proliferation (see Supplementary material online, Figure S6A-C), which was attenuated in siATP13A3-transfected hPAECs (see Supplementary material online, Figure S6D).The reduced proliferation in Endothelial Basal Medium-2 (EBM)2/2% FBS was not due to increased apoptosis, as caspase 3/7 activity was not altered by ATP13A3 knockdown (Figure 4C).However, caspase 3/7 activity was increased by siATP13A3 in hPAECs incubated in low serum conditions, suggesting a greater susceptibility to intrinsic stress (Figure 4C).Moreover, siATP13A3 did not alter basal hPAEC monolayer permeability, but we observed 40% higher permeability when monolayers were exposed to 1 U/mL thrombin (Figure 4D).

PAH-associated variants impair ATP13A3-mediated polyamine uptake in HMEC1 cells and hPAECs
To establish whether PAH-associated ATP13A3 variants are pathogenic, we assessed the functional impact of five PAH-associated missense protein variants (L675V, M850I, V855M, R858H, and L956P; Supplementary material online, Figure S7A) in different EC models, comparing these with ATP13A3 wild-type (ATP13A3-WT) protein and an artificial transport dead mutant protein with a D498N substitution in the catalytic autophosphorylation domain.
Stable lentiviral overexpression in HMEC1 cells of the untagged WT, but not the artificial D498N mutant protein, increased the uptake of PUT-BDP and SPD-BDP, but not SPM-BDP (Figure 5A and B and Supplementary material online, Figure S7B), consistent with our knockdown results (Figure 2C).Benzyl Viologen, a polyamine uptake inhibitor, 14 blocked PUT-BDP uptake (see Supplementary material online, Figure S7C).
Interestingly, despite the mRNA expression of the V855M variant being comparable with the WT and the R858H being higher than the WT (see Supplementary material online, Figure S7E), both variants showed reduced protein expression (Figure 5C and Supplementary material online, Figure S7F).As the antibody epitope (488-631) lies outside the mutated region, this implies reduced protein stability.Consequently, these variants fail to increase PUT-BDP and SPD-BDP uptake Figure 5A and B).Though the other variant proteins expressed well, the uptake of both PUT-BDP and SPD-BDP was impaired only for the L956P variant protein.Intriguingly, the L675V variant protein only demonstrated reduced SPD-BDP uptake, suggesting altered substrate specificity.Polyamines are essential for cell survival but at high concentrations, can be cytotoxic. 13Stable ATP13A3-WT protein overexpression sensitized HMEC-1 cells to escalating PUT concentrations, whereas the artificial D498N mutant did not (Figure 5D).Mirroring their impact on PUT-BDP uptake, the three untagged PAH-associated protein variants (V855M, R858H, and L956P) also exhibited reduced sensitivity to PUT toxicity.
So far, the M850I protein variant did not differ from the WT, although the endogenous PUT, SPD, and SPM levels were reduced in HMEC-1 cells stably overexpressing the M850I, L956P protein variants, and artificial D498N mutant, albeit non-significantly (see Supplementary material online, Figure S8A-C).However, stable overexpression of the L675V and M850I missense variant proteins did significantly decrease basal SPD and SPM levels in neuroblastoma SH-SY5Y cells (see Supplementary material online, Figure S8D-F), suggesting a cell-type-specific phenotype for these variants.To analyse the intracellular localization of the protein variants, we transiently overexpressed GFP-tagged WT, D498N, and the five PAH-associated ATP13A3 protein variants in HMEC-1 cells.Surprisingly, all GFP-tagged variants and WT-GFP consistently colocalized with Rab11, suggesting that when transiently expressed, V855M-GFP and R858H-GFP express well (see Supplementary material online, Figure S9A).In hPAECs transiently overexpressing these constructs, basal PUT levels were similar to all variants (see Supplementary material online, Figure S9B).Supplementation with 1 mM PUT increased endogenous PUT levels mainly in WT-GFP expressing cells, since the ATP13A3 protein variants attenuated (R858H-GFP and L956P-GFP) or abolished (D498N-GFP, L675V-GFP, M850I-GFP, and V855M-GFP) this response (see Supplementary material online, Figure S10C).We confirmed similar ATP13A1-3 expression of all the constructs (see Supplementary material online, Figure S10A-C).
We further assessed the impact of high polyamine concentrations on apoptosis (caspase-3/7 activity) of hPAECs.Although ATP13A3-WT-GFP overexpression sensitized hPAECs to 10 mM PUT (see Supplementary material online, Figure S11A), this was reduced for the artificial D498N-GFP mutant and disease variants L675V-GFP, M850I-GFP, and V855M-GFP, consistent with the attenuated response to PUT supplementation in these cells.In contrast, the R858H-GFP and L956P-GFP variants behaved more akin to the ATP13A3-WT (see Supplementary material online, Figure S11A).No differences were seen among ATP13A3-WT, the D498N, and PAH-related protein variants to SPD and SPM toxicity (see Supplementary material online, Figure S11B and C).
Together, our analysis of complementary expression systems reveals that the ATP13A3 missense variants present different forms of loss-of-function phenotypes, affecting polyamine uptake and/or homeostasis.

The ATP13A3 LK726 frameshift variant predisposes BOECs to apoptosis by affecting ATP13A3-mediated polyamine transport
To cross-validate our findings in a disease-relevant endogenous system, we derived BOECs from a patient with PAH bearing a heterozygous ATP13A3 frameshift variant (ATP13A3 LK726X , c.2176_2180delTTAAA), confirmed by Sanger sequencing (see Supplementary material online, Figure S12).This variant creates a premature stop codon (TGA 733X), predicted to reduce ATP13A3 mRNA expression through nonsense-mediated decay.In support of this, both ATP13A3 mRNA and protein levels (Figure 6A and Supplementary material online, Figure S13A) were reduced in ATP13A3 LK726X BOECs compared with control BOECs, without changes in the expression of ATP13A1 and ATP13A2 (see Supplementary material online, Figure S13B and C).Compared to control cells, ATP13A3 LK726X BOECs exhibited lower PUT content under both basal and PUT-supplemented conditions, though the fold increase in PUT content with supplementation was similar for all the BOEC lines (Figure 6B).Although basal and supplemented SPM and SPM contents were unchanged (Figure 6B), the uptake of PUT-BDP and SPM-BDP was significantly lower than in control BOECs (Figure 6C).Functionally, caspase-3/7 activity in low serum was significantly elevated in ATP13A3 LK726X BOECs (Figure 6D), and this was partially rescued by overexpression of the ATP13A3-WT protein, but not the artificial D498N mutant (see Supplementary material online, Figure S14).
Interestingly, ODC mRNA and protein levels were increased in ATP13A3 LK726X BOECs (Supplementary material online, Figure S15A and B), which was not explained by changes in OAZ1 or AZIN1 expression (see Supplementary material online, Figure S15C and D).Intriguingly, the expression of the synthetic enzymes, ARG1, AMD1, SRM, and SMS, were also elevated (see Supplementary material online, Figure S15E), while the catabolic enzymes remained comparable between ATP13A3 LK726X and control BOECs (see Supplementary material online, Figure S16).Collectively, our data suggest the ATP13A3 LK726X variant disrupts polyamine homeostasis in BOECs, predisposing cells to apoptosis.

Mice harbouring an Atp13a3 P452Lfs variant spontaneously develop PAH
To explore the potential role of potentially pathogenic ATP13A3 variants in PAH, we generated a new mouse line harbouring an Atp13a3 variant (P452LfsTer7), homologous to a human disease-associated ATP13A3 variant (P456Lfs). 7The mice were viable and fertile, although homozygous males and females were born at a lower frequency than the expected Mendelian ratio [males (n = 110): 26.4% Wt: 57.3%, Het: 16.3% Hom; females (n = 107): 28.0%Wt: 56.1%, Het: 15.9% Hom].Otherwise, mice harbouring heterozygous or homozygous Atp13a3 P452LfsTer7 alleles did not exhibit any overt behavioural abnormalities, nor did they differ in appearance or weight from their WT littermates.
Male mice heterozygous or homozygous for the Atp13a3 P452LfsTer7 allele exhibited reduced lung Atp13a3 expression reflecting their genotypes (Figure 7A).When RVSPs were assessed in male and female mice aged to 3 months, there were no genotype-related differences (Figure 7B).When aged to 6 months of age, female mice carrying either heterozygous or homozygous Atp13a3 P452LfsTer7 exhibited identical pressures to WT littermate controls.However, heterozygous male Atp13a3 P452Lfs mice spontaneously and consistently developed pulmonary hypertension (PH) compared with their littermate controls at 6 months of age, with a significant elevation of RVSP (Figure 7B), while heart rate and systemic blood pressure did not differ (see Supplementary material online, Figure S17A  and B).Transthoracic echocardiography demonstrated a significant shortening of pulmonary artery acceleration time (a surrogate of pulmonary artery pressure and pulmonary vascular resistance) in heterozygous Atp13a3 P452Lfs mice (see Supplementary material online, Figure S17C).Heterozygous Atp13a3 P452Lfs mice also exhibited increased right heart dimensions, namely RV inner diameter (see Supplementary material online, Figure S17D) and RV end-diastolic anterior wall thickness (see Supplementary material online, Figure S17E) compared with the WT littermates.Again, heart rate did not differ between the two groups (see Supplementary material online, Figure S17F).
Consistent with the haemodynamic data suggesting a baseline PAH phenotype, we also observed histological changes in the pulmonary circulation and hearts of Atp13a3 P452Lfs mice.Lung vascular morphometric analysis revealed that heterozygous Atp13a3 P452Lfs mice had a significantly higher percentage of fully muscularized small pulmonary arteries (Figure 7C).Consistent with this phenotype, the RVs were also hypertrophic (Figure 7D).This was associated with a significant increase in cardiomyocyte cross-sectional diameter (Figure 7E) and in interstitial fibrosis (Figure 7F).Although milder than the PAH observed in induced models of disease, these data confirmed that Atp13a3 P452Lfs heterozygous mice spontaneously and consistently demonstrated a PAH phenotype without the need for an applied disease-promoting stimulus.

Discussion
In this study, we establish ATP13A3 in ECs as an endolysosomal polyamine transporter enriched in recycling endosomes.ATP13A3 enables cellular polyamine uptake and ATP13A3 deficiency causes EC dysfunction.Human disease-associated ATP13A3 variants present a loss of polyamine uptake phenotype, whereas male mice harbouring a human diseaserelevant Atp13a3 variant develop PAH.This suggests that disrupted ATP13A3 most likely functions as a regulator of cellular polyamine content from recycling endosomes. 14,15,24This function mirrors the closely related ATP13A2, a late endo/lysosomal polyamine exporter that delivers polyamines into the cytosol, preventing lysosomal accumulation while regulating cellular content. 13The expression patterns of ATP13A2 and ATP13A3 partly explain their differing disease associations.In mice, Atp13a2 is widely expressed albeit with higher expression in the brain 25 and pathogenic ATP13A2 variants in humans cause an early onset parkinsonism, Kufor-Rakeb syndrome. 26Atp13a3 is also widely expressed in mice, with particularly high expression in liver, 25 although in humans potentially pathogenic variants may cause PAH. 7The widespread expression of Atp13a2 and Atp13a3 suggests possible redundancy, but the distinct diseases associated with potentially pathogenic variants in humans suggest some non-redundant roles.
We studied several EC models of ATP13A3 deficiency, namely transient siRNA, stable miRNA, and patient-derived ATP13A3 LK726X BOECs in parallel with overexpression experiments.Of note, the mRNA expression of ATP13A1 and ATP13A2 was not altered by ATP13A3 deficiency, suggesting a lack of compensation by these related proteins.We found that ATP13A3 deficiency in these EC models suppressed the uptake of BDP-labelled polyamines.The reduction in uptake of each polyamine differed between the cell models, possibly reflecting model-specific variations in the receptormediated endocytosis pathway proximal to ATP13A3-mediated transport to the cytosol. 17We also observed reductions in endogenous polyamine levels in ATP13A3-deficient EC cell models, and an attenuated response to PUT supplementation.Conversely, ATP13A3-WT protein overexpression led to increased uptake of mainly PUT-BDP and SPD-BDP, and a higher intracellular PUT content.The differences observed between labelled polyamine uptake vs. endogenous levels are not unusual.Fluorescently labelled polyamines behave similarly to radiolabelled polyamines, thus directly representing uptake. 24In contrast, quantifying endogenous polyamine levels by MS reflects an integration of the polyamine metabolic pathways, so the uptake of one polyamine species may impact all the polyamine species.
In this study, we have also identified that the loss of ATP13A3 not only reduces intracellular polyamine levels but also leads to a reprogramming of the polyamine metabolism pathway.Despite subtle differences between our EC models, we generally observed increased polyamine biosynthesis (e.g.up-regulated ODC1/AZIN1 or reduced OAZ1) as well as a reduced polyamine catabolism (e.g.lower SAT1 and/or PAO), both of which may compensate for polyamine loss.This probably explains why multiple endogenous polyamine species are simultaneously affected.One limitation of cells isolated from patients with PAH is the potential impact of changes acquired due to the disease state 27 and the heterozygous expression of the variant compared with the greater reduction achieved with siRNA transfection.These may explain why transcript alterations of polyamine biosynthesis enzymes with short-term ATP13A3 loss of a large magnitude (siRNA transfection) differed from chronic deficiency (genetic defects).With specific reference to OAZ1, a question remains regarding the mechanism by which its expression is reduced in hPAECs and BOECs following ATP13A3 knockdown.This may be due to a transcriptional response mediated by changes in polyamines either directly, or by activation of other signalling pathways.The regulation of OAZ1 would be of interest in future studies.
In this context of altered polyamine metabolism, it remains difficult to deduce the precise polyamine transport specificity of ATP13A3 from cellular data.Our observations suggest that ATP13A3 displays a broad polyamine selectivity that exerts the strongest impact on PUT uptake and content.Similar findings were also reported for Atp13a3 in CHO-MG cells, although unlabelled SPD and SPM competed with ATP13A3-dependent PUT-BDP uptake. 14However, ATP13A3 loss in pancreatic cancer cells prevented the uptake of radiolabelled SPD and SPM, with little effect on PUT uptake. 15Whether the polyamine specificity of ATP13A3 differs in a cell-dependent manner or if the methods of ATP13A3 manipulation exert different effects have yet to be established.
The dual effect of ATP13A3 deficiency on polyamine uptake and polyamine homeostasis in ECs may explain the strong impact of ATP13A3 reduction on endothelial health and functionality.Polyamines are essential for cell growth, 16,17 with polyamine depletion leading to cell cycle arrest. 17ATP13A3 deficiency suppressed serum-dependent proliferation and cyclin A, E, and B mRNA expression in BOECs implying repression of G1-S transition and DNA synthesis in the cell cycle.Polyamine depletion may promote cell apoptosis in response to pathogenic insults. 28,29We show that ATP13A3 deficiency increases apoptosis in hPAECs, BOECs, and ATP13A3 LK726X BOECs when they are stressed by serum starvation, an effect that is rescued with ATP13A3-WT overexpression.Moreover, polyamines are essential for epithelial cell-cell junctions, 30 and we observed that ATP13A3 deficiency exacerbated thrombin-dependent hPAEC monolayer permeability, indicating an important role of ATP13A3 in maintaining endothelial integrity.
Using our EC models, we examined the functional impact of PAH-associated missense variants using lentiviral overexpression.A homozygous V855M variant was identified in a child with early onset PAH leading to early death. 31The other PAH missense variants (L675V, M850I, R858H, and L956P) were heterozygous and found in older patients with PAH. 7 Collectively, our data confirm pathogenicity due to a loss of function, although possibly by differing impacts on ATP13A3 protein function, negatively impacting on (i) protein expression, (ii) transport activity, (iii) polyamine homeostasis, and/or (iv) substrate specificity.These differences may relate to cell type (e.g. for M850I), transient vs. the lentiviral stable variant expression (the latter being more potent potentially increasing the window), and/or the possible stabilizing effect of the GFP tag (for V855M and R858H).The latter may explain the low protein stability of untagged V855M and R858H variants compared with GFP fusion products.Furthermore, the variable impacts of these ATP13A3 variants may suggest other modifiers are required to promote disease progression.Interestingly, interferon-β (IFN-β) therapy induced PAH in a multiple sclerosis patient with a nonsense ATP13A3 variant (Glu514*) and IFN-β withdrawal improved their PAH symptoms. 32ombined, our results convincingly show that the disease-associated ATP13A3 variants present a loss of function, which based on our knockdown studies, would have a major impact on polyamine uptake and homeostasis in ECs.Although many genes disrupted in PAH tend to be enriched in the endothelium, ATP13A3 mRNA is expressed at similar levels in hPAECs, BOECs, and human pulmonary artery smooth muscle cells. 7iven that pulmonary artery smooth muscle cell proliferation is a hallmark of medial thickening in PAH, future investigations into the impact of ATP13A3 disruption on smooth muscle cell function would be of interest.
Importantly, mice harbouring a heterozygous Atp13a3 P452Lfs variant associated with human PAH developed increased RVSP and RV hypertrophy without altered systemic blood pressure.Intriguingly, this was only observed in male mice aged to 6 months, while female mice were unaffected.This does not reflect the female predominance reported previously in patients with PAH, 7 though it is not known if human males with pathogenic ATP13A3 variants die at a young age.Alternatively, there may be sex differences in the susceptibility of humans and mice to pathogenic ATP13A3 variants.In addition to elevated RVSP in 6-month-old male mice, analysis of the lung vasculature revealed a significantly higher percentage of fully muscularized small pulmonary arteries, suggesting that genetic deficiency of Atp13a3 leads to the development of PAH by affecting small pulmonary vessels.Although the increase in pressure in heterozygous Atp13a3 P452Lfs variant mice was modest compared with the large pressure increases seen in hypoxia-or Sugen/hypoxia-induced mouse PAH models, the PAH phenotype in Atp13a3 P452Lfs variant mice arose spontaneously and reproducibly without requiring a stimulus.4][35] In future, it would be interesting to explore the impact of disease-promoting stimuli on PAH and the polyamine pathway in Atp13a3 P452Lfs variant mice.
In a broader disease context, polyamine dysregulation has been implicated in non-genetic rodent models of PH and human PAH.9][40] However, the different rodent models appear to be associated with different mechanisms of polyamine accumulation.In MCT rats, increased activity of the synthetic enzymes, ODC and AMD, suggests higher polyamine biosynthesis rates. 38,39dministration of the irreversible ODC inhibitor, DFMO, attenuated the increase in mean pulmonary arterial pressure (mPAP). 40Conversely, in hypoxic PH, 14 C-SPD accumulation was elevated in lung tissues, 41 suggesting that the rates of uptake were increased.Altered polyamine metabolism has also been documented in patients with PAH.Metabolomic analysis has reported increased ornithine and PUT in lung tissues from patients with PAH 42 and elevated plasma 4-acetamidobutanoate and N-acetyl-PUT have been reported in patients with idiopathic/heritable PAH. 43Recently, ODC mRNA expression was shown to negatively correlate with mPAP in patients with PAH. 44Although abnormal polyamine levels are observed in PAH, it is unclear whether this represents changes in the intracellular or extracellular environment and whether various cell types may exhibit different responses.At this juncture, the mechanisms linking polyamine dysregulation and the pathobiology of PAH have not been clarified.
In conclusion, we have demonstrated that ATP13A3 functions as a polyamine transporter and has a functional role in endothelial homeostasis.PAH-associated variants exhibited impaired ATP13A3-mediated polyamine transport, contributing to disease-associated cellular phenotypes.These findings shed light on the pathogenic mechanism of ATP13A3 genetic defects leading to a loss of function in PAH and provide new insight into a potential role for polyamine dysregulation in the pathobiology of PAH.

Translational perspective
Rare missense ATP13A3 disease-associated variants have been identified in patients with pulmonary arterial hypertension (PAH), although their pathogenicity has not been confirmed.In this study, we show that disease-associated variants hamper the ATP13A3 transport function.ATP13A3 deficiency impairs polyamine homeostasis and uptakes and drives endothelial dysfunction.Conversely, overexpression increases polyamine uptake and rescues the proapoptotic phenotype of cells harbouring a disease-associated variant.Mice heterozygous for a disease-associated Atp13a3 variant spontaneously develop PAH.These findings support the rationale for exploring dysregulated polyamine homeostasis in PAH and suggest a potential for therapeutic targeting of polyamine pathways in PAH.

Figure 4 ATP13A3
Figure 4 ATP13A3 deficiency leads to endothelial dysfunction.(A) Proliferation, determined by cell counting, of transfected hPAECs over 6 days in EBM2 with 2% FBS, with media replenished every 2 days.(B) Transcription of CCND, CCNE, CCNA, and CCNB mRNAs with ATP13A3 deficiency assessed by quantitative polymerase chain reaction.The data are fold-change relative to the DH1 control for each transcript.(C ) Apoptosis assessed by Caspase-Glo®3/7 assay (Promega, Madison, NI) of transfected hPAECs cultured in EBM2 with 0.1% FBS or 2% FBS.(D) Permeability of transfected hPAEC monolayers to horseradish peroxidase in the absence or presence of 1 U/mL thrombin assessed by colorimetric assay.The data are the raw absorbance values for the different groups at the 2 h time point.The data (n = 4 experiments) in A-D are mean ± SEM and were analysed using a one-way ANOVA with Tukey's post hoc test for multiple comparisons.*P < 0.05, **P < 0.01 compared with siCP.