Following overnight fasting, animals were premedicated with ketamine (15 mg·kg⁻¹), midazolam (0.5 mg·kg⁻¹), and azaperone (4 mg·kg⁻¹). General anesthesia was induced with a propofol bolus (2 mg·kg⁻¹) and maintained via continuous infusion (10–20 mg·kg⁻¹·h⁻¹). Analgesia was provided using fentanyl (3 µg·kg⁻¹ bolus, followed by 2 µg·kg⁻¹·h⁻¹ infusion). Intravenous cefazolin (25 mg·kg⁻¹) was administered as antibiotic prophylaxis.

Animals were orotracheally intubated and ventilated with room air at a tidal volume of 8–10 mL.kg⁻¹, a respiratory rate of 15–20 breaths.min⁻¹, adjusted to maintain normocapnia. Positive end-expiratory pressure was set at 5 cmH₂O. An arterial catheter (Arterial Leadercath 20G, Vygon) was placed in a branch of the left or right superficial femoral artery for systemic blood pressure monitoring and recording (LabChart 8, ADInstruments). Continuous monitoring included peripheral oxygen saturation, capnography, body temperature, systemic blood pressure, and three-lead electrocardiography. Normothermia was maintained using a heating pad.

An 8 Fr introducer sheath (Radifocus® Introducer, Terumo) was placed in the right internal jugular vein under ultrasound guidance. A Swan-Ganz thermodilution catheter (Edwards) was advanced into the PA, and baseline PA pressure (PAP) and cardiac output (CO) were recorded. A 0.035-inch, 260 cm guidewire was then introduced through the catheter's distal lumen and advanced into a distal PA segment.

The Swan-Ganz catheter was exchanged over the guidewire for a 6 Fr long sheath (Flexor® Shuttle® Guiding Sheath, Cook Medical), positioned in either the left or right PA. The sheath's cross-cut valve was replaced with a Y-connector, enabling simultaneous introduction of a 3.5 Fr tipped Mikro-Cath (Millar), equipped with a radiopaque marker band for continuous pressure monitoring, and injection of either contrast medium or silk sutures. The experimental setup is illustrated in Figure 1B.

Following baseline pulmonary angiography, 15–30 sterile silk sutures (size 0, 15 cm length) were injected through the PA sheath in animals assigned to the CTEPH group (final n = 7). Each suture was manually loaded into a 10 mL syringe and purged with high pressure for deployment into the pulmonary vasculature (as illustrated in Supplementary Video 1). Throughout the procedure PAP, CO, and systemic blood pressure were continuously monitored.

Suture injection was halted upon meeting any of the following criteria: mPAP exceeding 40, 45, or 50 mmHg during the first, second, or third intervention, respectively; a drop in CO; systemic hypotension (mean blood pressure—mBP -<60 mmHg); or angiographic evidence of complete vessel occlusion. The procedure lasted approximately 1 h per session.

The embolization protocol was repeated weekly for 3 consecutive weeks, beginning with the left PA and alternating sides thereafter. A sham group (n = 6) underwent right heart catheterization (RHC) with saline infusion at matched time points; the average volume used to deliver sutures in one session was estimated at 150 mL.

All procedures were conducted under normothermic and normocapnic conditions, with room air ventilation to minimize external influences on vascular tone. The first intervention was systematically performed in the left PA, selected due to its relatively smaller vascular territory. The right cranial lobe artery, originating craniolaterally at the proximal right PA (19), was systematically preserved (as detailed in Figure 1B).

Representative angiograms obtained before and after left and right PA embolization, as well as following the final intervention, are shown in Figures 1C,D illustrates the evolution of mean PAP and CO in CTEPH animals. Data from Sham is not presented for simplicity's sake.

Four weeks after the last intervention, animals were anesthetized and ventilated as described in the animal preparation section. Transthoracic echocardiography was performed using a cardiac probe (4v1c, Acuson Sequoia C512, Siemens). Measurements included: RV free wall thickness (RVWT), PA acceleration/ejection time (PAAT/PAET), left ventricle (LV) eccentricity index at early diastole in short axis (EI) and tricuspid annular plane systolic excursion (TAPSE). At least three stable cardiac cycles were averaged for all measurements.

Following echocardiography, 10 Fr introducer sheaths were placed in a femoral artery and vein for LV catheterization with a pressure-volume (PV) catheter (Ventri-Cath 507, Millar) and inferior vena cava (IVC) occlusion balloon deployment, respectively. Two additional 8 Fr introducer sheaths were placed in the right internal jugular vein for RHC, one for CO monitoring with thermodilution and another for PV catheter (Ventri-Cath 507, Millar). To allow simultaneous recording of left and right PV loops, each catheter was connected to a separate control unit (MPVS Ultra, Millar).

In a closed-chest percutaneous setup, thermodilution CO measurements were used for calibration of field inhomogeneity (α), parallel conductance was derived by central injection of 15% NaCl bolus (5–7 mL), and IVC occlusions were achieved using the occlusion balloon. Acquisitions were performed upon a 15 min stabilization period with ventilation suspended at end-expiration. Data was analyzed using custom Python scripts. End-systolic (ES) PV relationships (ESPVR) were modelled with simple linear fitting upon checking that quadratic terms were dismal, end-diastolic (ED) PV relationships (EDPVR) were modelled with exponential fitting including an asymptote. The time constant of isovolumic relaxation, τ, was derived by logistic fitting.

After ensuring adequate anesthetic depth with propofol (10 mg·kg⁻¹) and fentanyl (3 µg·kg⁻¹) bolus administration, sternotomy was performed, the thoracic aorta and inferior vena cava were lacerated for exsanguination, and the heart and lungs were harvested en bloc and washed. Transmural RV samples were collected from the trabecular free wall at the mid-ventricular level and were either snap-frozen in liquid nitrogen and stored at −80 °C or collected for histopathology (formalin-fixed paraffin embedded tissue). The non-obstructed cranial right lung lobe was dissected and perfused with 10% (v/v) buffered formalin for adequate fixation to assess flow-induced distal vasculopathy.

The right and left ventricles were dissected and weighed separately. For histological assessment, 3 µm-thick sections of the right ventricle (RV) were stained with hematoxylin and eosin (H&E) to quantify cardiomyocyte cross-sectional area (CSA), and with picrosirius red (PSR) to evaluate myocardial fibrosis. Cranial right lung lobe as well as lower lobe samples were formalin-fixed, and paraffin embedded; 3 µm-thick sections were obtained and stained with hematoxylin-eosin-saffron (HES) to assess distal arteriole remodeling and lung parenchyma modifications.

Digital images were acquired and analyzed using ImageJ software. For CSA quantification, a minimum of 60 cardiomyocytes per animal were measured in transverse cross-section at the level of the nucleus. Myocardial fibrosis was quantified as the percentage of PSR-positive (red) area, excluding background, across at least 10 sections per animal. Distal arteriole remodeling was assessed by calculating the percent wall thickness from a minimum of 10 arterioles per animal. Lung HES were also blindly analyzed by an experienced pathologist (DL) to assess other lung structural modifications on a qualitative basis.

RV skinned cardiomyocytes were isolated as previously described (20). In brief, samples were defrosted in cold relax solution [5.95 mM Na₂ATP, 6.04 mM MgCl₂.6H₂O, 2.00 mM ethylene glycol tetra acetic acid (EGTA), 139.6 mM KCl, 10 mM imidazole, pH 7.0], mechanically disrupted and permeabilized with 0.1% Triton X-100. Using microscopic view (1X51, Olympus) and imaging software (VSL 900B, Aurora Scientific), single cardiomyocytes were attached to a force transducer and a length controller (403 A and 315 C-I, Aurora Scientific, respectively). Cell length was adjusted digitally (series 600 A digital controller, Aurora Scientific).

Cardiomyocyte stiffness was assessed through sarcomere length (SL)–passive tension (PT) relationships between 1.8 and 2.3 µm. Calcium-dependent active tension (AT) relationships were obtained at 2.2 µm SL in activating solutions with different Ca²⁺ concentrations [pCa: 5.0, 5.2, 5.4, 5.6, 5.8 and 6.0; diluted from a stock of 5.97 mM Na₂ATP, 6.28 Mm MgCl₂, 40.64 mM propionic acid, 100 mM N,N-Bis(2-hydroxyethyl)taurine, 7 mM Ca-EGTA and 14.5 mM phosphocreatine disodium salt hydrate, Na₂PCr]. Passive force was measured through a “slack” test (cells were shortened to 80% of the initial length) in relax solution. AT was derived by subtracting passive force from total force. Experiments were carried out at 15 °C. Force was normalized for myocyte CSA, which was derived assuming an elliptical shape. A minimum of 5 cardiomyocytes were analyzed per animal.

Total RNA was extracted from 30 mg of frozen RV tissue using NZYol (Nzytech) following the manufacturer's instructions. RNA concentration and integrity were assessed by Nanodrop (Thermo Fisher Scientific®) and electrophoresis. cDNA (100 ng.µL⁻¹) was synthesized using the SensiFAST™ cDNA Synthesis kit (Meridian Bioscience Inc®) and reactions were conducted in a T100 thermal cycler (Bio-Rad®). Real-time quantitative polymerase chain reaction (RT-qPCR) reactions were run in duplicate using SensiFAST™ SYBR® Hi-ROX Kit (Meridian Bioscience Inc®) in PikoReal™ Real-Time PCR System (Thermo Fisher Scientific®), according to manufacturer's instructions. Before gene expression quantification using the 2−ΔΔCT method, PCR efficiency of each gene including the internal control genes (glyceraldehyde-3-phosphate dehydrogenase, GAPDH, and 60S ribosomal protein L4, RPL4) was determined and it was ensured that they were identical. Target gene expression was normalized using the geometric mean of both housekeeping genes (21). No differences were observed in the expression of control genes. Data are presented relative to Sham expression. Primer sequences (Table 1) were designed in-house with Primer-BLAST (NCBI). Melting-curve analysis and primer-BLAST checks ensured amplicon specificity.

Myocardial phospholamban (PLB), its phosphorylated form (p-PLB), and sarcoendoplasmic reticulum Ca²⁺-ATPase (SERCA) expression were assessed by western blot. RV samples were homogenized in modified RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate and 0.1%–0.5% SDS) supplemented with protease (Sigma P8340) and phosphatase (Sigma P0044, Sigma P5726) inhibitors (1:100) using 1.4 mm zirconium oxide beads (Precelleys, Bertin Technologies) and tissue homogenizer (Minilys, Bertin Technologies). Total protein concentration was quantified using bovine serum albumin (BSA) as a standard (Bradford assay, Bio-Rad) and the sample concentration was adjusted accordingly. A volume of Laemmli Buffer 4× (8% SDS, 20% glycerol, 0.004% bromophenol blue and 100 mM Tris-HCl, pH 7.4) and DTT 10× (1.25 M) was added to the protein lysate and the solution was heated for 5 min at 95°C. 20 μg of RV protein were separated by electrophoresis using a 4%–20% gradient polyacrylamide gel (Criterion™ TGX™ Precast Gels, Bio-Rad) and then electroblotted onto nitrocellulose membranes (Bio-Rad). Blots were blocked for 1 h with 5% bovine serum albumin in Tris-buffered saline with Tween 20 (TBS-T; 50 mM Tris-HCl, 150 mM NaCl, pH 7.4, 0.1% Tween 20) for PLB and p-PLB detection and with 5% non-fat dry milk in TBS-T for SERCA analysis. PLB and p-PLB blots were transferred to 0.1% sodium azide solution (1:250 dilution) and primary antibodies to PLB (MA3-922, Invitrogen) and p-PLB (#8496, Cell Signaling) were added whereas blots for SERCA analysis were transferred to 5% non-fat dry milk in TBS-T (1:250 dilution) with primary antibodies to SERCA (ab3625, Abcam) and GAPDH (ab8245, Abcam). After overnight incubation at 4°C, blots were subsequently incubated in 0.5% non-fat dry milk in TBS-T with 700 nm and 800 nm infra-red dye-conjugated antibodies (IRDye® 680LT and 800W, LI-COR) for 60 min at room temperature and imaged by scanning at 800 and 700 nm with the Odyssey Infrared Imaging System (LICOR Biosciences). Experiments were carried out in duplicate. No differences were observed in GAPDH levels between groups. Data are presented relative to Sham levels.

Previously aliquoted plasma samples were thawed on ice and assayed for N-terminal pro–B-type natriuretic peptide (NT-proBNP; CSB-EQ027465PI, Cusabio) and endothelin-1 (ET-1; ADI-900-020A, Enzo Life Sciences Inc) in duplicate according to the manufacturer's instructions.

Tabular data and graphs show the mean ± standard error of the mean for normally distributed data or the median and range and box plots otherwise, respectively. For normally distributed data, groups were compared by Student's or Welch t-test in case of homogeneous or non-homogeneous variance, respectively. The Mann–Whitney U-test was employed whenever deviation from normality was observed according to the Shapiro–Wilk test. Analysis of covariance was employed for data derived from linear or exponential fitting in PV relationships, where data were previously log-transformed whenever necessary to meet assumptions. Mixed Linear models with fixed effect for group and random intercepts by animal plus a variance component for each myocyte were performed for skinned RV cardiomyocytes studies, according to Sikkel et al. (22). Significance was set at 2-tailed P < 0.05. Statistical analysis and figures were performed with Python (key packages: numpy, pandas, statsmodels, scipy, and matplotlib).

Repeated weekly embolization of the PA using long silk sutures resulted in a progressive and sustained elevation of mPAP, which was generally well tolerated by most animals (Figure 1D). Despite stringent criteria for discontinuing embolization, two CTEPH animals died during model induction. One animal succumbed to acute RV failure refractory to inotropic support during the third embolization (mPAP > 60 mmHg with mBP < 35 mmHg). The second animal was found dead the day after the second intervention, with acute RV failure considered the most likely cause.

At each embolization, major branches of the pulmonary arteries were occluded, and the obstruction persisted in follow-up. At necropsy, suture plugs accompanied by fibrin deposition were observed within large pulmonary arteries of CTEPH, but the right cranial lobe artery was unobstructed (Figures 1C,E). A cephalad shift in pulmonary perfusion was also observed and can be clearly appraised in the final angiography in Figure 1C. Supplementary Video 2 shows full PA angiograms.

On terminal evaluation, mPAP was substantially higher in CTEPH compared with Sham (51 ± 3 vs. 20 ± 1 mmHg, respectively; P < 0.001), there was also a clear reduction of CO (4.2 ± 0.2 vs. 6.6 ± 0.9 L.min⁻¹; P = 0.042) yielding remarkably elevated pulmonary vascular resistance (PVR; 10.0 ± 1.1 vs. 1.5 ± 0.5 WU; P < 0.001). Of note, since wedged pressure could not be obtained reliably, we employed LV end-diastolic pressure (EDP) from PV loops in the estimation of PVR. CTEPH was induced consistently across animals (Figure 1D). CTEPH also showed a trend towards decreased weight gain on follow-up (Table 2).

On echocardiography, CTEPH presented lower TAPSE, increased RVWT as well as both short PA acceleration times and lengthened ejection times with late systolic notching in PA flow on pulsed-wave Doppler. Septal displacement towards the LV was particularly prominent in early diastole, leading to increased EI (Table 2 and representative tracings in Figure 2A). Supplementary Video 3 further illustrates septal shift and RV hypertrophy.

In PV hemodynamic analysis, the RV of CTEPH showed decreased CO, ejection fraction (35 ± 2 vs. 51 ± 3%; P < 0.001) and a trend towards dilation. Still, maximum pressure derivative (dP/dt_max), preload recruitable stroke work (PRSW) and both pulmonary arterial elastance (E_a) and RV end-systolic elastance (E_es) were elevated, with preserved ventricle-vascular coupling (VVC), denoting overall stiffening of the right circulation in compensatory adaptation to PH. There was also delayed relaxation as assessed by time constant τ (20 ± 2 vs. 13 ± 3 ms; P = 0.041), EDP elevation (12 ± 1 vs. 8 ± 1 mmHg; P = 0.036) and an upward shift in the EDPVR, as denoted by the rise in chamber stiffness constant β [0.038 [0.021−0.176] vs. 0.015 [0.001−0.042] mmHg.mL⁻¹; P = 0.031]. No changes were observed in heart rate.

As for the LV, although systolic performance was unaltered there was a clear trend for unloading and impaired relaxation and filling, most likely due to ventricular interdependence. Averaged group PV loops, ESPVR and EDPVR from RV and LV IVC occlusions are depicted in Figure 2B and data is summarized in Table 2.

RV hypertrophy was evident at necropsy (Figure 3A), with increased RV weight in CTEPH (Table 2), as well as on histological sections from the RV myocardium, according to cardiomyocyte CSA (Figure 3B, top), nevertheless no significant differences in fibrosis percentage was noted (Figure 3B, bottom). These changes were paralleled by a shift towards cardiomyocyte fetal phenotype in gene expression, RV myocytes from CTEPH showed increased expression of MYH7, which codes for slow β-myosin heavy chain, and markedly reduced expression of fast α-myosin heavy chain encoded by MYH6, as well as raised expression of NPPB and TNF, which code B-type natriuretic peptide (BNP) and tumor necrosis factor-α, respectively. Collagen chain expression and RV myocardium gene expression of ET-1, however, were unchanged (Figure 4A). Moreover, despite RV overexpression of BNP no changes were observed in circulating plasma levels regarding N-terminal-proBNP whereas ET-1 plasma levels showed a trend for increase (Figure 4C; P = 0.059).

Skinned RV cardiomyocytes from CTEPH showed elevated Ca²⁺-active tension (AT) responses, with raised maximum AT but unchanged Ca²⁺ myofilament sensitivity as appraised by Ca²⁺ EC₅₀ compared with Sham (Figure 5A), as well as increased passive stiffness as evidenced by the upward shift in SL-PT (Figure 5B). Calcium-handling proteins were also disturbed in CTEPH, SERCA2a protein levels (Figure 4B; full representative Western blot membranes are presented as Supplementary Figure 1 in Supplementary Material) and its ATP2A2 gene expression (Figure 4A) were both reduced along with hypophosphorylation of phospholamban (Figure 4B), both suggesting impaired Ca²⁺ reuptake. No changes were observed in the expression of the cardiac ryanodine receptor.

Due to the pulmonary arteries anatomy of swine, the right cranial lobe artery was preserved. Indeed, a cephalad redirection of flow was clearly visible on angiograms. This overflow induced remodeling of distal pulmonary arterioles (<100 µm) in CTEPH lungs which showed increased media thickness compared with sham (43 ± 3 vs. 22 ± 2%, respectively; P < 0.001), as represented in histological sections from the right cranial lobe (Figure 6).

CTEPH lungs displayed abundant peribronchovascular lymphoid inflammation, mainly characterized by follicular organization with germinative centers (Figure 7A), often associated with multiple inflammatory cell types, including neutrophils, in both central and distal parts of the lung. The airways showed thickening of the mucosa, characterized by a pseudopolypoid appearance, sometimes sub-occlusive, raised by hyperplastic or aberrant bronchial vascularization (Figure 7B). Distant respiratory bronchioles were sometimes devoid of overlying epithelium, and extensive thickening of the alveolar wall was particularly noticeable in CTEPH lungs, while not very noticeable in Sham samples. These changes are morphologically like human Non-Specific Interstitial Pneumopathy (NSIP, Figure 7C), with a more cellular, or a mixed cellular and fibrous components in distal or proximal samples, respectively, suggesting a temporal and spatial gradient of CTEPH-related lesions (earlier lesions being more distant).

Pulmonary arteries showed vasculitis lesions without vascularitis or leukocytoclastic features, as well as intimal and medial hyperplasia with an increase in small distal muscularized vessels (Figure 7D), without significant intimal fibrosis or thickening. Aberrant or occlusive muscularization of the septal vessels was also observed. Bronchial vessels showed adventitial myxedema or centripetal remodeling of the media in CTEPH (Figure 7E). Although predominant in CTEPH, microthrombi were also found in Sham.

The obstructed lobes of CTEPH animals (basal portions) showed congestive and hemorrhagic changes in the bronchial tree and bronchial arteries similar to those described above. Follicular inflammation and expansion of the bronchial artery bed was evident (Supplementary Figure 2A,B in Supplementary Material), as well as pulmonary arterial medial hyperplasia (Supplementary Figure 2C in Supplementary Material). In addition, they were associated with an increase in the number of alveolar capillaries, which were also dilated, sometimes taking on a capillary hemangiomatosis-like appearance (Supplementary Figure 2D in Supplementary Material).

Inflammatory changes related to the intervention, such as intravascular foreign bodies, isolated or embedded in a giant cell resorptive inflammatory reaction, were observed in CTEPH (Figure 7F). No old nor organized thrombi or colander-like lesion were identified in the tissue samples taken from either group. No inflammatory changes of suspected septic origin, or tumoral features were identified.

On some of the sites of occlusion a fibrin and collagen cast is clearly identifiable around suture threads, with eccentric recanalization channels, a representative section is presented as Supplementary Figure 3 in Supplementary Material.