|Home | About | Journals | Submit | Contact Us | Français|
Pulmonary arterial hypertension (PAH) is a lethal syndrome characterized by pulmonary vascular obstruction due in part to pulmonary artery smooth muscle cell (PASMC) hyperproliferation. Mitochondrial fragmentation and normoxic activation of hypoxia-inducible factor-1α (HIF-1α) have been observed in PAH PASMCs, however their relationship and relevance to the development of PAH is unknown. Dynamin-related protein-1 (DRP1) is a GTPase that, when activated by kinases that phosphorylate Serine-616, causes mitochondrial fission. It is however unknown whether mitochondrial fission is a prerequisite for proliferation.
We hypothesize that DRP1 activation is responsible for increased mitochondrial fission in PAH PASMCs and that DRP1 inhibition may slow proliferation and have therapeutic potential.
Experiments were conducted using human control and PAH lungs (n=5) and PASMCs in culture. Parallel experiments were performed in rat lung sections and PASMCs and in rodent PAH models induced by the HIF-1α activator, cobalt, chronic hypoxia, and monocrotaline. HIF-1α activation in human PAH leads to mitochondrial fission by cyclin B1/CDK1-dependent phosphorylation of DRP1 at Serine-616. In normal PASMC, HIF-1α activation by CoCl2 or desferrioxamine causes DRP1-mediated fission. HIF-1α inhibition reduces DRP1 activation, prevents fission and reduces PASMC proliferation. Both the DRP1 inhibitor Mdivi-1 and siDRP1 prevent mitotic fission and arrest PAH PASMCs at the G2/M interphase. Mdivi-1 is antiproliferative in human PAH PASMC and in rodent models. Mdivi-1 improves exercise capacity, right ventricular function and hemodynamics in experimental PAH.
DRP-1-mediated mitotic fission is a cell cycle checkpoint that can be therapeutically targeted in hyperproliferative disorders such as PAH.
Pulmonary arterial hypertension (PAH) is a syndrome in which obstructed, constricted and inflamed small pulmonary arteries (PA) increase pulmonary vascular resistance (PVR), leading to right ventricular hypertrophy and ultimately failure. The cause of PAH remains unclear, despite important advances in understanding the genetic and epigenetic basis of this syndrome. The discovery of mutations in bone morphogenetic protein receptor-2 (BMPR-2) in familial PAH1, 2, the recognition of somatic chromosomal abnormalities in sporadic PAH3 and the discovery of epigenetic silencing of mitochondrial superoxide dismutase 2 (SOD2) and the resulting activation of hypoxiainducible factor-1α (HIF-1α4 have yet to alter treatment. Moreover, despite effective vasodilator therapies, mortality rates remain high (15% at 1-year)5 suggesting a new treatment paradigm is needed. Excessive pulmonary artery smooth muscle cell (PASMC) proliferation plays an important role in PAH (reviewed in 6), suggesting that antiproliferative therapies are needed. We recently discovered that the hyperproliferative phenotype in PAH PASMCs (human and rodent) is associated with normoxic activation of HIF-1α and fragmentation of the mitochondrial network7. Normoxic activation of HIF-1α in human PAH occurs in PASMC4, 7 and endothelial cells8. HIF-1α activation can result from the redox changes initiated by epigenetic silencing of SOD2 in PAH4, consistent with prior descriptions of redox-regulation of HIF-1α9, 10. Activated HIF-1α suppresses mitochondrial oxidative metabolism (by increasing the expression of pyruvate dehydrogenase kinases, PDKs, thereby blocking pyruvate uptake into the Krebs cycle)11 while simultaneously upregulating enzymes and transporters that favor glycolysis (i.e. hexokinase-2 and the glucose transporter-1, GLUT1) Targeting these mitochondrial-metabolic abnormalities in PAH, using the PDK inhibitor dichloroacetate, reduces glycolysis, restores oxidative metabolism and regresses. PAH in several experimental models12-14. This demonstrates the susceptibility of PAH to mitochondrial-metabolic therapies. HIF-1α is activated early in experimental models of PAH and leads to a pulmonary glycolytic pattern which can be detected with a 18F-fluoro-deoxy-glucose positron emission tomography (FDG-PET) scan. This glycolytic switch is reversible with therapies including PDK inhibitors and Gleevec15. However, the link between HIF-1α, mitochondrial fragmentation, and cell proliferation is unknown.
Mitochondria exist in a dynamic network comprised of individual organelles that continuously join (fusion) and fragment (fission). The GTPases Mitofusin-1 (MFN1), Mitofusin-2 (MFN2), and Optic atrophy-1 (OPA1) regulate fusion. Conversely, Fission-1 (FIS1) and Dynamin-related protein-1 (DRP1) mediate mitochondrial fission16. DRP1’s activity results from phosphorylation by regulatory kinases, including the mitosis regulator, cyclin B1/Cyclin-dependent kinase 1 (CDK1), which causes mitotic phosphorylation of Serine 616 and activation of DRP1. Upon activation DRP1 moves from the cytosol to the mitochondria where it assembles in multimers that constrict and divide the mitochondria. Conversely, phosphorylation of DRP1 at Serine 637, by protein kinase A, inhibits fission17, 18.
Here we determine the molecular basis for mitochondrial fragmentation in PAH, its relationship to HIF-1α and assess whether mitochondrial fission offers a novel antiproliferative therapeutic target in PAH. Studies are conducted in human PASMCs and lung sections from PAH and control patients and in 3 rodent models of PAH, including a model induced by chemical activation of HIF-1α. We demonstrate that HIF-1α activation is a proliferative stimulus that causes DRP1-mediated mitochondrial fission. In vivo HIF-1α activation (by cobalt) is sufficient to produce PAH in rats and Mdivi-1, a small-molecule DRP1 inhibitor19, restores mitochondrial fusion and reduces the hemodynamic and histological severity of PAH in this model. Increased fission in PAH PASMC reflects DRP1 activation by cyclin B1/CDK1 activity and accompanies cell cycle progression from G2 to mitosis. We show that fission is required for normal cell division and that Mdivi-1 slows proliferation by locking mitochondria in fusion and inhibiting cell-cycle progression, causing G2/M arrest, as we recently also observed in lung cancer cells20. Mdivi-1 also has antiproliferative therapeutic benefit in monocrotaline- and hypoxia-induced pulmonary hypertension. We conclude that normoxic activation of HIF-1α creates a fissogenic, proliferative milieu in PAH. Cyclin B1/CDK1, which is upregulated in PAH, coordinates mitosis and mitochondrial fission, permitting rapid proliferation; conversely inhibition of mitotic fission causes cell cycle arrest, and regresses PAH. Thus, mitotic fission is a checkpoint that can be therapeutically targeted by inhibiting DRP1 in hyperproliferative syndromes.
Paraffin-embedded tissue sections and fresh tissue for PASMCs isolation were obtained from autopsied patients with idiopathic PAH or control patients (without PAH) under an IACUC-approved protocol (Online Table I).
The University of Chicago Animal Care Committee approved all protocols. Male Sprague-Dawley rats were purchased from Charles Rivers Laboratories (Wilmington, MA). To induce pulmonary hypertension in rats, we injected 2 mg CoCl2 i.p./day for 4-weeks (Sigma Aldrich, St. Louis, MO). This dose was well tolerated and did not induce mortality. Higher doses caused death or severe lethargy. Control rats received saline injections. Other animal models for PAH were chronic hypoxia (10% oxygen) and monocrotaline (60mg/kg i.p.). The DRP1 inhibitor Mdivi-1 (Enzo Life Sciences, Plymouth Meeting, PA) was dissolved in DMSO and injected at a dose based on the literature (50mg/kg)21 for prevention studies (weekly injections for CoCl2 and bi-weekly injections for chronic hypoxia), while 5 daily injections were given 3 weeks after monocrotaline injection for regression studies. Control animals received intraperitoneal DMSO injections.
Immunoblotting was performed on 25μg of protein from PASMCs. The primary antibodies used were anti-cyclin B1 (Santa Cruz Biotechnology, Santa Cruz, CA), anti-cyclin E (Cell Signaling Technology, Boston, MA), anti-DRP1 (Novus Biologicals, Littleton, CO), anti-phospho Ser616 DRP1 (Cell Signaling Technology, Boston, MA), anti-phospho Ser637 DRP1 (Cell Signaling Technology, Boston, MA), anti-HIF-1α (Novus Biologicals, Littleton, CO), anti-smooth muscle cell actin (MP Biomedicals, Solon, OH), and anti-vWF (Abcam, Cambridge, MA). To confirm equal protein loading, blots were stripped in Restore Western Blot stripping buffer (Thermo Scientific, Rockford, IL) for 15’ and then reprobed with anti-Actin (Millipore, Billerica, MA). Antigen retrieval, secondary antibodies, signal amplification and imaging parameters are described in detail in the Online Data Supplement.
For siRNA treatment, PASMCs were grown to ~80% confluence and then transfected using oligofectamine (Invitrogen, Carlsbad, CA) and 25nM validated Silencer Select siRNAs (Applied Biosystems, Carlsbad, CA). After 4 hours, normal culture medium was added and gene knockdown was assessed after 48 hours using qRT-PCR and immunocytochemistry. For each experiment there was a scrambled siRNA control (Applied Biosystems, Carlsbad, CA).
Mitochondria of live PASMCs were imaged with a Zeiss 510META confocal laser scanning microscope equipped with an environmental chamber to maintain humidity, 37°C, and 5% CO2. We used co-expression of mitochondrial matrix-targeted photoactivatable GFP (mito-PA-GFP, a kind gift of Dr. Richard Youle, NIH, Bethesda, MD)22 and a mitochondrial matrix-targeted DsRed Protein (mito-DsRed, a kind gift of Dr. Michael Frohman, Stony Brook University, Stony Brook, NY)23 to assess the degree of mitochondrial connectivity in live cells. PASMCs were transfected using Fugene HD transfection reagent (Roche, Indianapolis, IN). After acquiring 3 control images mito-PA-GFP was photoactivated with 10 passes of the 2-photon laser (set to 700nm) in a confined 200×60 pixel area. Photoactivated GFP was measured at 488nm excitation, 500-550nm emission. For assessing mitochondrial morphology in the whole cell, we used a mitochondrial matrix-targeted GFP protein (mito-GFP, a kind gift of Dr. Michael Frohman, Stony Brook University, Stony Brook, NY)23.
Two recently validated techniques were used for assessing mitochondrial morphology in the whole cell: measurement of the mitochondrial fragmentation count (MFC) and the mitochondrial networking factor (MNF)20. Briefly, images of mito-GFP transfected PASMCs were obtained, thresholded and binarized to identify mitochondrial segments using ImageJ (NIH, Bethesda, MD). The mitochondrial fragmentation count (MFC) was calculated by normalizing the number of individual mitochondria to the entire mitochondrial pixel count20. To quantify functional networking of the mitochondria, PASMCs were transfected with mitochondrial-targeted, photoactivatable green fluorescent protein (mito-PA-GFP) and mitochondrial-targeted DsRed protein. The extent of mitochondrial networking is reflected in the diffusion of the photoactivated mitochondrial GFP away from the site of photoactivation, as measured after 17.6 seconds using ImageJ (NIH, Bethesda, MD). A highly interconnected network allows the photoactivated GFP to distribute in a larger area, resulting in a higher MNF20.
Proliferation was quantified using the Click-It EdU kit according the manufacturer’s instructions (Invitrogen, Carlsbad, CA).
Freshly isolated pulmonary arteries (3rd-4th order) were fixed and processed for electron microscopy (see Online Data Supplement). Images were collected using a scanning transmission electron microscope at 300KV (Tecnai F30; FEI) with a high-performance Gatan CCD camera.
A biosensor for cyclin B1/CDK1 activity was recently developed. It consists of a plasmid encoding 2 fluorescent proteins (mCerulean and YPet) linked by a cyclin B1/CDK1 phosphorylation site and a phosphobinding domain25. Phosphorylation induces a conformational change which facilitates energy transfer from mCerulean to YPet. PASMCs were transfected with this FRET biosensor plasmid and mCerulean was exited with a 458 nm Argon laser. Emission was recorded at 465-510nm for mCerulean and 520-555nm for YPet. Increased YPet/mCerulean signal indicates cyclin B1/CDK1 activation25.
Values are stated as mean±SEM. Inter-group differences were assessed by an unpaired Student’s t-test, simple or repeated-measures ANOVA with post hoc analysis using Tukey’s test, as appropriate. When applicable, normality was confirmed with a Kolmogorov-Smirnov test. For comparing the number of cyclin E positive cells after Mdivi-1 treatment, a Fisher’s exact test was used. A p<0.05 was considered statistically significant.
Mitochondria are more fragmented in PASMCs from idiopathic PAH patients versus control PASMCs (Figure 1A). Quantitative RT-PCR for the genes that regulate mitochondrial fission and fusion16 shows a profile that favors fission (increased DRP1 and FIS1 and decreased MFN2) (Figure 1B). OPA1 and MFN1 expression are unchanged (Online Figure IA). We focused on the role of DRP1 because preliminary data indicated that Mdivi-1, a small molecule inhibitor of DRP1, substantially rescues the mitochondrial network and slows proliferation.
Expression of total and activated DRP1 (phosphorylated at Ser616) are upregulated in PAH PASMC (Figure 1C-E). Direct inhibition of CDK1, using 10μM of the specific inhibitor RO-330627, significantly reduces DRP1 Ser616 phosphorylation, implicating CDK1 as the key kinase driving DRP1 activation in PAH. In contrast, inhibition of other putative mediators of DRP1 phosphorylation, such as calmodulin, with 5μM W7, or calcium/calmodulin-dependent protein kinase II, with 20μM KN93, has no effect on DRP1 Ser616 phosphorylation (Figure 1E). Serum withdrawal (like CDK1 inhibition) reduces both the percentage of PASMCs with DRP1 Ser616 phosphorylation and the expression of the G2/mitotic-specific cyclin B1 in PAH PASMCs (Figure 1E-F), indicating that external growth factors/mitogens promote DRP1 activation. Using a FRET biosensor for cyclin B1/CDK1, we confirmed that increased levels of cyclin B1 in PAH PASMCs translate into higher cyclin B1/CDK1 activity (Figure 1G).
Mitochondrial networking, measured using the MNF, is decreased in human PAH PASMC and this is reversed by 25μM Mdivi-1 (Figure 1H). Online Movies I-III show representative 8-minute time-lapse movies of control, PAH, and PAH+Mdivi-1-treated PASMCs immediately prior to and following photoactivation, documenting the more rapid spread of the signal in cells treated with Mdivi-1.
Compared to control human PASMCs, PAH PASMCs have increased proliferation rates at baseline, confirming the persistence of their hyperproliferative phenotype in vitro. DRP1 inhibition by Mdivi-1 causes a dose-dependent reduction in proliferation rates (Figure 2A). Mdivi-1’s antiproliferative effects are due to induction of cell-cycle arrest in the G2/M phase (Figure 2B and Online Figure IIA), suggesting that cell-cycle progression is dependent on fission of mitochondria. To exclude non-specific effects of Mdivi-1, we used siRNA directed against DRP1 and found a similar reduction in proliferation (Figure 2C) due to accumulation of PASMCs at the G2/M phase of the cell cycle (Figure 2D). To more exactly establish at which point the cell cycle is interrupted, we used immunoblotting to assess whether CDK1 activation is altered by Mdivi-1. CDK1 activation during mitosis depends on dephosphorylation of Thr14 and Tyr1528. This dephosphorylation is not impaired by Mdivi-1 (Figure 2E), suggesting that cell cycle arrest occurs specifically in prometaphase (between late G2 and metaphase) since this is the last phase in which cells have dephosphorylated CDK1. Subsequently, CDK1 becomes phosphorylated/inactivated due to proteosomal degradation of cyclin B129 and this was not observed with Mdivi-1. Mdivi-1-induced G2/M arrest is reminiscent of the state observed with cell synchronization by the microtubule assembly inhibitor, nocodazole. Therefore, we synchronized PASMCs in prometaphase with 50ng/ml nocodazole for 16-hours. Synchronization was confirmed by an increase in cyclin B1-positive PASMCs (Figure 2F). After removal of nocodazole, PAH cells rapidly move through mitosis, evident in parallel decreases in cyclin B1 levels and DNA content; however, Mdivi-1 prevents decreases in cyclin B1 levels (Figure 2F), confirming that Mdivi-1 specifically blocks cell cycle progression beyond prometaphase in PAH PASMCs.
Mdivi-1 binds to an allosteric site on DRP1 and thereby blocks a conformational change required for self-assembly19. Since self-assembly is downstream of DRP1 Ser616 phosphorylation, Mdivi-1 should not decrease DRP1 Ser616 phosphorylation. Indeed, Mdivi-1 actually increases DRP1 Ser616 phosphorylation (Online Figure IB-C), consistent with the observed increased CDK1 activation in response to Mdivi-1 (Figure 2E). Thus Mdivi-1 is not interfering with CDK1-mediated phosphorylation of DRP1 Ser616. Recently, Mitra et al. described increased levels of the S-phase promoting cyclin E in response to Mdivi-1 (which we confirmed, Online Figure IIB)30. They also noted increased entry into S-phase, which supports our finding that Mdivi-1 inhibits proliferation mainly by inhibiting a G2/M checkpoint in the cell cycle.
We obtained human lung tissue from 5 autopsied idiopathic PAH patients and 5 age-matched controls (patient characteristics in Online Table I) and confirmed the presence of plexiform lesions and increased pulmonary artery muscularization (Figure 3A). Plexiform lesions and muscularized pulmonary arteries from idiopathic PAH patients have 10±7% and 12±3% increases in DRP1 expression, respectively compared to arteries from normal subjects (Online Figure IIIA). DRP1 Ser616 is clearly enriched in pulmonary arteries (versus lung parenchyma) in PAH and is greater in PAH versus control lungs (Figure 3B), consistent with the PASMC data (Figure 1D-E). Moreover, the G2/mitotic-specific cyclin B1 is not only upregulated in PAH PASMCs, but also activated, evident from its nuclear translocation (Figure 3C), which is in line with the increased cyclin B1/CDK1 activity measured in PAH PASMCs using a FRET biosensor (Figure 1G). Increased cyclin B1/CDK1 directly leads to DRP1 Ser616 phosphorylation31. While endothelial cells stain strongly for the inhibitory Ser637 form of DRP1, we did not detect this form of DRP1 in the media of control or PAH lungs (Online Figure IIIB) and therefore focused on DRP1 Ser616 in subsequent experiments.
We confirmed that HIF-1α activation occurs in PASMCs in lungs from PAH but not control patients (Figure 4A). Even when PAH PASMCs are cultured in vitro at oxygen concentrations of ~140mmHg, most have HIF-1α activation, evident from its nuclear translocation, greatly exceeding the rate in control PASMCs (Figure 4B). Metabolically, HIF-1α activation increased glycolysis, as measured with a Seahorse Bioscience XF24 analyzer (Online Data Supplement). Compared to control PASMCs, we observed a more than 3-fold higher lactate production rate and a 2-fold lower ratio of oxygen consumption/lactate production in PAH versus control human PASMCs (Online Figure IVA). The upregulated glycolysis is also evident in increased gene expression of GLUT1, PDK2 and PDK4 (Online Figure IVB).
We next examined the consequences of HIF-1α activation on proliferation and metabolic function in human PASMCs. The importance of HIF-1α activation for proliferation was studied using chetomin, a small molecule that inhibits the interaction of HIF-1α with its p300 coactivator, preventing hypoxia-inducible gene transcription32. Chetomin caused minimal cell death (Figure 4C) consistent with the low doses (5-50nM) we used, relative to those reported in the literature33, 34. Chetomin dose-dependently inhibited proliferation (Figure 4D) and DRP1 Ser616 phosphorylation (Figure 4E).
We used two chemically discrete HIF-1α activators (CoCl and desferrioxamine)35 2 to establish a direct relationship between normoxic HIF-1α activation and mitochondrial fragmentation in normal rat PASMCs. This approach avoids confounders, such as genetic abnormalities and upregulation of growth factor pathways that might exist in human PAH cells. CoCl2 activated HIF-1α in most PASMCs (evident from nuclear accumulation of HIF-1α, Online Figure VA). Whole-cell patch clamp traces show a reduction in voltage-gated K+ currents (Kv) with CoCl2, which is rescued with a HIF-1α dominant-negative virus (HIF-1α DN, Online Figure VIA-B). The reduced Kv currents lead to depolarization of the PASMC (lower EM, Online Figure VIC) and increased cytosolic calcium levels (Online Figure VID). Both abnormalities can be prevented by overexpression of a HIF-1α DN construct, confirming the role of HIF-1α in the current inhibition caused by ex vivo CoCl2 treatment. These HIF-1α-induced electrophysiologic changes recapitulate those seen in PAH PASMCs7, 36-39.
CoCl2 (500μM) rapidly induces mitochondrial fission, again mimicking PAH (Figure 4F). To exclude nonspecific effects of the chemical HIF-1α activators, we first showed that 50μg/ml cycloheximide, an inhibitor of protein translation, prevents cobalt or desferrioxamine-induced mitochondrial fragmentation. This indicates that protein synthesis is required for HIF-1α-induced fission (Figure 5A-B). Next we showed that HIF-1α siRNA largely prevents cobalt-induced fission, proving the fragmentation is indeed due to HIF-1α activation (Figure 5C). Efficacy of HIF-1α siRNA is demonstrated in Online Figure VB-C. Finally, we showed that, cobalt increases the phosphorylation of DRP1 Ser616 (Figure 5D). Thus cobalt-induced HIF-1α activation mimics the PAH phenotype described in Figure 1D.
We used Mdivi-1 to confirm that the mitochondrial fragmentation observed after CoCl2-induced HIF-1α activation is dependent on DRP1. Mdivi-1 treatment (25μM) not only normalizes mitochondrial morphology after CoCl2 treatment (Figure 5E-F) but also enhances networking, as demonstrated by the more rapid and broader spread of photoactivated, mitochondrial matrix-targeted GFP (Figure 5E-G). This is concordant with the effects of Mdivi-1 in human PAH PASMCs (Figure 1H).
Since cobalt recapitulates many of the features of PAH PASMCs in culture, we determined whether activation of HIF-1α using cobalt would cause PAH in vivo. As expected, chronic cobalt treatment increased hematocrit, confirming that the selected dose of CoCl2 was biologically active (Figure 6A). Moreover, cobalt decreased maximal walking distance (Figure 6B). Echocardiography confirmed that the pulmonary artery acceleration time (PAAT), a measure of vascular compliance that is inversely related to pulmonary artery pressures26, decreases significantly after cobalt treatment (Figure 6C, Online Figure VIIA), while right and left heart catheterization confirmed that cobalt increased pulmonary vascular resistance (Figure 6D-F, Online Figure VIIB-C). Cobalt-induced PAH is not caused by left ventricular dysfunction as the left ventricular end diastolic pressures did not change between groups (Figure 6F). Post mortem analysis revealed increasedright ventricular hypertrophy and percentage medial thickness of small pulmonary arteries (Figure 6G-I).
To determine whether inhibition of mitochondrial fission in vivo would attenuate cobalt-induced PAH, we co-injected 50mg/kg Mdivi-1 (or the DMSO vehicle) on a weekly basis in rats receiving chronic CoCl2 injections. Exercise capacity, which is reduced in cobalt-induced PAH, is normalized by Mdivi-1 treatment (Figure 6B). The PAAT increases significantly after Mdivi-1 treatment (Figure 6C, Online Figure VIIA). Similarly, Mdivi-1 significantly decreases PVR (Figure 6D). Mdivi-1 treatment decreases right ventricular hypertrophy (Figure 6G) and the percentage medial thickness of small pulmonary arteries, consistent with an antiproliferative effect (Figure 6H-I).
To prove that Mdivi-1’s antiproliferative effects in vivo reflected inhibition of fission, we performed transmission electron microscopy studies on pulmonary arteries isolated from control rats and rats receiving CoCl2 or CoCl2+Mdivi-1. These images demonstrate that mitochondrial fragmentation is present in vivo in the cobalt PAH model and that these morphologic changes are reversed with Mdivi-1 (Figure 6J).
We also studied whether similar benefits of Mdivi-1 therapy could be observed in additional experimental models of PAH. As shown in Figure 7A, Mdivi-1 prevents the decrease in exercise capacity caused by chronic hypoxia-induced pulmonary hypertension. Echocardiography confirms that Mdivi-1 improved PAAT (consistent with a reduction in pulmonary artery pressure) and TAPSE (indicating improved RV function), while reducing right ventricular hypertrophy (Figure 7A). Mdivi-1 also has anti-proliferative effects in the chronic hypoxia model. In line with our previous findings, we observed decreased muscularization and fewer proliferating PASMCs in rats treated with Mdivi-1 (Figure 7B-D).
Having shown that Mdivi-1 can attenuate the development of CoCl2 and chronic hypoxia-induced PAH, we tested Mdivi-1 as regression therapy in the monocrotaline model. More specifically, we wanted to test whether inhibiting fission reduces PASMC proliferation. Daily Mdivi-1 treatment was started 3 weeks after monocrotaline injection, when PAH is already established (confirmed by echocardiography), and continued for 5 days. This relative short treatment significantly improved TAPSE, and there were trends towards improved PAAT and reduced right ventricular hypertrophy, although neither achieved statistical significance (Online Figure VIIIA). In addition, muscularization of small pulmonary arteries was significantly decreased by Mdivi-1 and there was a pronounced decrease in PASMC proliferation (Online Figure VIIIB). HIF-1α activation was also present in the MCT model (Online Figure VIIIC).
We demonstrate that the hyperproliferative diathesis of human PAH PASMCs reflects increased DRP1 expression and activation (Figures 1--2).2). The molecular mechanism of DRP1 activation in human PAH patients is increased DRP1 Ser616 phosphorylation (Figures 1D-E and and3B),3B), reflecting increased activation of DRP1’s regulatory kinase (cyclin B1/CDK1). DRP1 Ser616 phosphorylation and fission can be induced in normal PASMCs by chemically activating HIF-1α. The relevance of HIF-1α-induced DRP-1 activation to PAH is evident in the discovery that cobalt-induced HIF-1α activation elicits PAH in normal rats and that PAH is attenuated by Mdivi-1. Notably, the hemodynamic benefits were associated with restoration of mitochondrial fusion in vivo (Figure 6J). The antiproliferative effects of blocking mitotic fission are also evident in two well-established models of pulmonary hypertension. By locking the mitochondria in fusion, Mdivi-1 and siDRP1 prevent mitotic fission and cause cell-cycle arrest in the G2/M phase (Figure 2A-D), which is consistent with the recent demonstration that inhibition of DRP1 regresses lung cancer cell xenografts20. Thus excessive fission, which occurs in human PAH PASMCs in vitro and in vivo, contributes to the disease by driving proliferation and can be therapeutically targeted.
Mitochondrial morphology reflects the balance of fission and fusion. Fusion merges the contents of adjoined mitochondria, creating a homogenous network of elongated mitochondria allowing the redistribution of mitochondrial DNA and proteins40. Conversely, fission is necessary during cell division to evenly distribute mitochondria between daughter cells31. Fission occurs when DRP1 is recruited from the cytoplasm to the mitochondrial surface where it forms oligomeric complexes that hydrolyse GTP and mechanically constrict the mitochondria causing fragmentation41
Taguchi et al. showed that the interconnected mitochondrial. network structures in interphase HeLa cells become fragmented in the early mitotic phase via cyclin B1/CDK1-mediated DRP1 phosphorylation31. In the current study we show that in human PAH there is both an upregulation of total and Ser616 phosphorylated DRP1. The pro-fission phenotype is seen in lungs from patients with PAH and persists in human PAH PASMCs in culture (Figure 1D-E, Figure 3B). The elevation of cyclin B1 and CDK1, which coordinate mitosis and mitotic fission, is a major stimulus for Ser616 phosphorylation in PAH (Figure 1F-G and Figure 3C). Cell proliferation is driven by many factors in PAH, including elevated levels of mitogens and growth factors (reviewed in 6) and mutations of the bone morphogenetic protein receptor (in familial PAH)42.
In carefully controlled experiments, using molecular tools including siHIF-1α and a HIF-1α dominant-negative virus, we showed that HIF-1α activation is sufficient to confer a PAH phenotype on previously normal PASMCs (Online Figure VI) and induce mitochondrial fission (Figure 5). The increased HIF-1α in human PAH lungs (Figure 4A) persists in culture, despite PASMCs being cultured at ambient oxygen concentrations (~140mmHg). This is consistent with the recently described epigenetic suppression of mitochondrial superoxide dismutase, which is an upstream, heritable cause of the HIF-1α activation and proliferation in human and fawn-hooded rat PAH that persist in cells in culture4. In addition, normoxic HIF-1α activation can be induced in aortic SMC by growth factors such as serotonin and PDGF-BB43.
HIF-1α activation causes a glycolytic shift that confers a proliferative advantage in PASMCs. Similar HIF-1α-dependent glycolytic metabolic changes also occur in the endothelial cells of PAH patients8, 44. Proof that HIF-1α underlies the proliferation advantage comes from experiments in which HIF-1α inhibition by chetomin32 reduces PASMC proliferation (Figure 4D). Activation of HIF-1α results in glycolytic metabolic changes which can be detected using lung FDG-PET scans. These glycolytic changes occur simultaneously with the development of pulmonary hypertension after monocrotaline administration and appear to be related to increased proliferation15. Reversing the glycolytic phenotype with dichloroacetate also regresses experimental PAH15. Our findings are consistent with the observations that HIF-1α haploinsufficient mice are relatively protected from chronic hypoxic pulmonary hypertension45 whilst HIF-1α activation increases proliferation of human PASMC in response to platelet-derived growth factor46. These data provide compelling evidence that HIF-1α activation contributes to disease progression in human PAH.
While CoCl2 clearly induces HIF-1α in PASMCs (Online Figure VA), it is possible that non-specific effects of cobalt contribute to our observations. Indeed gene expression arrays in fibroblasts and carcinoma cells have shown that CoCl2 and HIF-1α induce a common set of genes, but some genes are specifically induced by cobalt47, 48. Yet, desferrioxamine, another HIF-1α activator35, similarly leads to mitochondrial fragmentation (Figure 5A) and conversely, siRNA-mediated knockdown of HIF-1α prevents CoCl2-induced mitochondrial fragmentation (Figure 5C), indicating that mitochondrial fragmentation depends on HIF-1α activation. Moreover, changes in K+ currents, membrane potential, and cytoplasmic calcium concentrations induced by CoCl2 are almost completely negated by adenoviral overexpression of a HIF-1α dominant negative construct (Online Figure VI), again suggesting that in PASMCs most effects of CoCl2 are HIF-1α mediated. In addition, in vivo, CoCl2 will induce HIF-1α in other tissues, as exemplified by increased hematocrit which is dependent on renal erythropoietin production. Despite possible non-specific effects, the net result of chronic CoCl2 administration is pulmonary vascular remodeling (Figure 6).
We acknowledge that while Mdivi-1 clearly improves mitochondrial networking both in human (Figure 1H) and rat PASMCs (Figure 5E) and in vivo (Figure 6J), it could potentially have off-target effects. However, knockdown of DRP1 expression by using a specific siRNA causes a similar arrest of PAH PASMC in the G2/M phase and has the same antiproliferative effects as Mdivi-1 on PAH PASMC (Figure 2C-D). Since Mdivi-1 is currently the only commercially available small molecule inhibitor of mitochondrial fragmentation, the development of other inhibitors will be necessary to confirm the specificity of our findings. Even though DRP1 knockout mice have been generated, they die around embryonic day 12.5, precluding their use to study the role of DRP1 in the development of PAH49.
In patient-derived PAH PASMCs as well as cobalt treated PASMCs, Mdivi-1 prevents mitochondrial fragmentation and enhances fusion (Figure 1H and 5E-G). Mdivi-1 or knockdown of DRP1 prevent proliferation in human PAH PASMCs (Figure 2A,C). The notion that mitochondrial fragmentation is essential for cell cycle progression past the G2/M checkpoint is an important and possibly controversial discovery. In support of our findings, gene therapy with Mitofusin-2, which causes mitochondrial fusion, also slows SMC proliferation in systemic arterial injury50. Moreover, DRP1 activation has recently been shown to be crucial to the hyperproliferative phenotype in lung cancer and Mdivi-1 is therapeutic in regressing tumors in a xenotransplantation model20. More study is required to determine whether mitotic division of organelles (not just mitochondria) is relevant to proliferative disorders. Supporting this notion, blocking Golgi fragmentation prevents progression through mitosis51. Interestingly, CDK1 is involved in controlling Golgi fragmentation as well52.
Electron microscopy confirms that CoCl2 induces mitochondrial fragmentation in vivo (Figure 6J), and chronic cobalt administration leads to pulmonary hypertension in line with previous publications that describe its ability to increase hematocrit, pulmonary hypertension and right ventricular hypertrophy in rats53 and right ventricular hypertrophy in broiler chickens54. Importantly, Mdivi-1 has already been shown to protect against ischemia-induced kidney damage and cardiac ischemia-reperfusion injury21, 55. For the first time, we used Mdivi-1 for long-term treatment (during the 4 weeks of CoCl2 administration) and observed that it is well tolerated and significantly improves exercise capacity and reduces PVR and right ventricular hypertrophy (Figure 6B-G). Moreover, muscularization of the pulmonary small blood vessels is attenuated by Mdivi-1 treatment, consistent with its antiproliferative effects on PAH PASMCs (Figure 6H-I). We only looked at the effect of Mdivi-1 on PASMC proliferation and it is possible that Mdivi-1 could influence other cell types as well. However, we did observe increased hematocrit in CoCl2 rats treated with Mdivi-1, indicating that hematopoietic stem cell proliferation is not or only minimally impaired. Moreover, impaired intestinal stem cell proliferation would result in impaired nutrient absorption and weight loss, which we did not observe. After intraperitoneal injection, Mdivi-1 is taken up in the mesenteric vein and transported to the right ventricle and subsequently to the pulmonary circulation. Therefore, Mdivi-1 concentrations are expected to be higher in the pulmonary vascular bed compared to other vascular beds. A similar mechanism is responsible for the relative pulmonary specificity of monocrotaline-induced pulmonary hypertension. Future pharmacological studies are necessary to evaluate tissue distribution and retention times of Mdivi-1. Mdivi-1 reduces PASMC proliferation in both monocrotaline and chronic hypoxic pulmonary hypertension (Figure 7 and Online Figure VIII). Therefore, it is reasonable to anticipate that Mdivi-1 is more effective in preventing than in reversing PAH. Thus inhibition of mitochondrial fission may have broad therapeutic benefit in preventing disease progression in various forms of PAH. Future research is necessary to elucidate the role of FIS1 and MFN-2 in the hyperproliferative phenotype of PAH.
In conclusion, HIF-1α stimulated, DRP1-mediated mitochondrial fission leads to increased proliferation in human PAH PASMCs. This is the first direct evidence that an acquired abnormality of mitochondrial fission contributes to the maintenance of human PAH and demonstrates the feasibility of targeting mitochondrial dynamics as a therapeutic option for PAH. Taken together, our data indicate that mitotic fission is required for rapid proliferation and conversely, that locking the mitochondria in a networked state, prevents mitotic fission causing cell cycle arrest. A schematic representation of our conclusions Is provided in Figure 8. Our findings have great therapeutic potential in PAH, evidenced by the effective and well-tolerated response to Mdivi-1 in the human PAH PASMC and several experimental models of pulmonary hypertension (Figures 6--77 and Online Figure VIII). This work also demonstrates an important interaction between the mitochondrial cycle and the cell cycle, which can perhaps be exploited in other proliferative disorders, such as cancer.
Online Movie I: Control PASMCs were cotransfected with mitochondrial matrix-targeted DsRed and mitochondrial matrix-targeted photoactivatable GFP. Upon photoactivation, GFP spreads throughout fused mitochondria. Images were obtained during 8 minutes following photoactivation.
Online Movie II: Photoactivation experiment in PAH PASMC.
Online Movie III: Photoactivation experiment in PAH PASMC treated with 25μM Mdivi-1.
Online Figure I. Expression levels of additional mitochondrial fusion mediators and Mdivi-1 does not inhibit phosphorylation of DRP1 Ser 616.
A) Mitofusin-1 (MFN1) and Optic atrophy-1 (OPA1) are proteins involved in mitochondrial fusion and are present in the outer and inner mitochondrial membrane, respectively. Their expression levels were not changed between control and PAH human PASMCs.
B-C) Mdivi-1 does not prevent cyclin B1/CDK1 mediated phosphorylation of DRP1 at Ser616. Immunocytochemistry (B) and immunoblotting (C) demonstrated increased Ser616 phosphorylation, in agreement with the increased CDK1 activation in response to Mdivi-1.
Online Figure II. Cell cycle analysis after Mdivi-1 treatment of PAH PASMCs and increased cyclin E expression after Mdivi-1 treatment.
A) Human PASMCs were incubated for 1 hour with the thymidine analog 5-ethynyl-2’-deoxyuridine (EdU) to assess proliferation and 7-Amino-Actinomycin D (7-AAD) was used to determine DNA content. The percentage of cells in the G2/M phase of the cell cycle (no EdU incorporation, more than diploid DNA content) increased in response to 25μM Mdivi-1.
B) Increased percentage of PASMCs are positive for nuclear cyclin E staining after Mdivi-1 treatment, asterisks point at PASMCs with nuclear cyclin E staining. The nuclear intensity in approximately 60 nuclei was analyzed in each group. Based on the no antibody control, nuclei with a mean relative fluorescence intensity >1200 were considered positive. Scale bar = 100μm.
Online Figure III. Immunostaining for DRP1 and DRP1 phosphorylated at Ser637 of human lungs.
A) Immunofluorescent staining for DRP1 in human lung sections. Double staining for DRP1 (green) and smooth muscle cell actin (red) allows identification of the smooth muscle cell layer and permitted quantification of DRP1 staining in the PASMCs. PASMCs in PAH lungs have a 12±3% increased DRP1 staining compared to PASMCs in control lungs. Approximately 50 blood vessels were analyzed per group. Scale bar = 100μm.
B) Immunohistochemistry for DRP1 phosphorylation at Ser637. When phosphorylated at this Serine amino acid, DRP1 is inactivated. While there is strong staining in endothelial cells and inflammatory cells, we could not detect Ser637 phosphorylation in small precapillary resistance PASMCs of control or PAH lungs. Scale bar = 50μm.
Online Figure IV. Glycolytic shift in PAH PASMCs.
A) PAH PASMCs have increased lactate production and a decreased oxygen consumption/lactate production ratio as measured using the Seahorse analyzer. This suggests that PAH PASMCs generate a larger proportion of their ATP from glycolysis compared to control PASMCs.
B) This glycolytic switch is confirmed by the upregulation of glucose transporter 1 (GLUT1), and pyruvate dehydrogenase kinases (PDK) 2 and 4.
Online Figure V. Confirmation of HIF-1α induction by CoCl2 and of the effectiveness of siRNA mediated HIF-1α knockdown.
A) CoCl2 leads to a strong induction and nuclear accumulation of HIF-1α in the majority of PASMCs.
B) siRNA mediated HIF-1α downregulation prevents HIF-1α accumulation in response to CoCl2. Arrowheads point to examples of nuclei without HIF-1α, while asterisks point at nuclei with HIF-1α accumulation. There is a clear reduction in HIF-1α signal intensity when cells are pretreated with siHIF-1α. Scale bar = 100μm.
C) The glucose transporter 1 (GLUT1) is upregulated in response to HIF-1α and siRNA against HIF-1α prevent this increase, confirming the efficacy of this HIF-1α inhibition strategy.
Online Figure VI. HIF-1α recapitulates the features described for PAH PASMCs.
A-B) Representative patch clamp traces show a reduction in voltage-gated potassium (Kv) currents after chronic (24 hours) cobalt treatment of rat PASMCs. This is rescued with a HIF-1α dominant-negative virus (HIF-1α DN, n=4). The 4-aminopyridine (4-AP)-sensitive current is diminished in cobalt-treated cells, demonstrating decreased Kv currents. * P < 0.05 versus control.
C-D) Chronic (24 hours) cobalt treatment depolarizes cells (lower membrane potential, n=4) and increases cytosolic calcium concentrations ([Ca2+]cyt). Both effects are prevented by the HIF-1α DN virus (n=11).
Online Figure VII. Echocardiography and catheterization measurements in CoCl2/Mdivi-1 treated animals.
A) Representative pulse wave Doppler tracings of the pulmonary artery. These traces were used to measure the pulmonary artery acceleration time (PAAT).
B-C) Representative catheterization traces of the pulmonary artery (B) and the right ventricle (C). The mean pulmonary artery pressure was used to calculate pulmonary vascular resistance. These traces were obtained using a 22 gauge fluid-filled catheter in anesthetized, open-chest rats.
Online Figure VIII. Therapeutic benefit of Mdivi-1 in the chronic hypoxia model.
A) Rats were injected with monocrotaline and 3 weeks later, when pulmonary hypertension is present, we started daily treatment for 5 days with Mdivi-1. We found that Mdivi-1 improves functional capacity measured on a treadmill. There is a trend for improved pulmonary artery acceleration time (PAAT) and decreased right ventricular fractional weight, while there is a significant improvement in tricuspid annular plane systolic excursion (TAPSE).
B) Rats injected with monocrotaline develop pulmonary hypertension characterized by excessive PASMC proliferation. Mdivi-1 both reduces the degree of muscularization of the pulmonary arteries and the number proliferating cell nuclear antigen (PCNA) positive PASMCs. Scale bar = 50μm
C) We observed nuclear accumulation of HIF-1α in PASMCs of monocrotaline-treated rats. Scale bar = 50μm.
Online Table I. Characteristics of patients of which lung tissue was studied by immunohistochemistry
Dr. Archer is supported by NIH-RO1-HL071115, 1RC1HL099462-01, and the American Heart Association. Dr. Rehman is supported by NIH-R01-GM094220.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.