Our prospective observational cohort of CTEPH patients has been previously reported (22). Forty-one CTEPH patients were diagnosed according to the ERS/ECSC guidelines ( Table 1 ) (30). Similar to prior work from our group, exclusion criteria were incomplete diagnostic work-up and therefore no confirmed PH diagnosis, not treatment-naive, age <18 years, or not capable of understanding or signing informed consent (31). Additionally,17 HCs (41% female, mean age 55.3 ± 12.5), were included with the following exclusion criteria: autoimmune disease, active infectious disease, use of immunomodulatory drugs, history of cardiopulmonary disease. At 1-year follow-up patients visited the outpatient clinic and did not have any active infection. The study protocol was approved by the Erasmus MC medical ethical committee (MEC-2011-392). Written informed consent was provided by all patients and controls. The study was performed conform the principles outlined in the declaration of Helsinki.

Hemodynamic and clinical data at diagnosis were collected during the inpatient cardiopulmonary screening visit for analysis of PH (22, 31). Data were collected and stored in PAHTool (version 4.3.5947.29411, Inovoltus), an online electronic case report form. Patients were treated according to the ERS/ESC guidelines (30) and prospectively followed-up by half-yearly scheduled visits to the outpatient clinic. All patients were assessed for eligibility for either a PEA or BPA ( Table 1 ). The mean follow-up duration was 51.0 months. Survival was defined as all-cause mortality. In our cohort, 7 out of 41 patients died within the follow-up period.

Peripheral blood mononuclear cells were stained for intra- and extracellular markers, using flow cytometry procedures essentially as described previously (32), using the antibodies given in Supplementary Table 1 . Cell fractions were directly stained for chemokine receptors and extracellular markers for 60 minutes at 4°C. After fixation and permeabilization steps, cells were intracellularly stained for FoxP3 and CTLA4. For the measurement of cytokines, PBMCs were incubated for 4 hours at 37°C in RPMI Medium 1640 + GlutaMAX-I (Gibco) supplemented with 5% fetal bovine serum (Gibco), 10 ng/ml phorbol 12-mysristate 13-acetate (Sigma-Aldrich), 250 ng/ml ionomycin (Sigma-Adrich) and Golgistop (BD Bioscience), after which cells were stained as previously described (33). Non-specific labelling was prevented by blocking Fc receptors using human TruStain FcX (Biolegend) and dead cells were excluded with Fixable Viability Dye Live/Dead eF506 (eBioscience). Data was acquired using a FACSymphony A5 flow cytometer (Beckton Dickinson) and analyzed using FlowJo version 10 (Tree Star Inc software).

Hematoxylin and eosin (H&E) staining and multiplex immunofluorescence were performed as described previously (34). For immunohistochemical analysis of human lung tissue, we performed a 5-plex immunofluorescent multiplex by automated IHC using the Ventana Benchmark Discovery ULTRA (Ventana Medical Systems Inc.). Formalin fixed paraffin embedded (FFPE)-tissue of lung vascular lesions of CTEPH patients were handled and used according to the guidelines of the declaration of Helsinki. This study was conducted in accordance with the guidelines of the Biomedical Scientific Societies (Dutch Federa code of conduct 2011) for the use of anonymized residual tissue obtained during regular treatment.

Slides of 4 µm thick FFPE sections were stained for CD3, CD4, CD8, FOXP3 and CCR6 ( Supplementary Table 2 ). In brief, following deparaffinization and heat-induced antigen retrieval with CC1 (#950-224, Ventana) for 32 min, anti-CD3 was incubated for 32 min at 37°C followed by omnimap anti-rabbit HRP (#760-4311, Ventana) and detection with R6G (#760-244, Ventana). An antibody denaturation step was performed with CC2 (#950-123, Ventana) at 100°C for 20 min. Secondly, incubation with anti-FOXP3 was performed for 60 min at 37°C, followed by universal HQ kit (#760-275, Ventana) and detection with DCC (#760-240, Ventana) for 8 min. An antibody denaturation step was performed with CC2 at 100°C for 20 min. Thirdly, anti-CD4 was incubated for 32 min at 37°C, followed by omnimap anti-rabbit HRP (#760-4311, Ventana) and detection with Red610 (#760-245, Ventana) for 8 minutes. An antibody denaturation step was performed with CC2 at 100°C for 20 min. Fourthly, sections were incubated with anti-CCR6 for 120 min at 37°C followed by universal HQ kit (#760-275, Ventana) and detection with Cy5 (#760-238, Ventana) for 8 min. An antibody denaturation step was performed with CC2 at 100°C for 20 min. Lastly, incubation with anti-CD8 was performed for 60 min at 37°C followed by omnimap anti-rabbit HRP (#760-4311, Ventana) and detection with FAM (#760-243, Ventana). Finally, slides were washed in phosphate-buffered saline and mounted with Vectashield containing 4’,6-diamidino-2-phenylindole (Vector laboratories, Peterborough, UK). Analysis of the multiplex staining was performed using Qupath the open software platform for bioimage analysis (version 0.3.0). In short, sections were annotated by simple tissue detection after which cells were identified by the default cell detection command using dapi to identify nuclei. Using simple thresholding for all channels all cells were classified and annotated. These measurements were used for further analysis.

Principal component analysis (PCA) was performed using R and RStudio, and the packages FactoMineR and Factoextra (22, 35).

Statistical evaluations of T cell subsets, cytokine production, activation markers and PCA dimension coordinates were by Mann-Whitney U tests. Paired diagnosis and 1-year follow-up data were analyzed using a Wilcoxon matched-pairs signed rank test. Correlation coefficients were calculated using the nonparametric Spearman correlation. All statistical tests were two-sided; p-values <0.05 were considered statistically significant. Statistical analyses were performed using GraphPad Prism v8 (Graph Pad Software).

We characterized peripheral blood T cells in a cohort of 41 treatment-naive CTEPH patients at diagnosis and 17 HCs ( Table 1 ). Five subpopulations were separately analyzed: CD45RA⁺ CD4⁺ T cells (FoxP3^-; mainly naive T cells, but also CD45RA⁺ TEMRA cells, which are thought to be antigen-experienced T cells re-expressing CD45RA), CD45RA^- CD4⁺ T cells (FoxP3^-CXCR5^-; memory T cells), FoxP3⁺ CD4⁺ T cells (Tregs and activated T cells), CD45RA^- CD8⁺ T cells and CD45RA⁺ CD8⁺ T cells (see Figure 1 and Supplementary Figure 1 for gating strategy). The proportions of these five T cell subpopulations did not differ between CTEPH patients and HCs (data not shown).

We used the expression of surface chemokine receptors to distinguish distinct CD45RA^- CD4⁺ memory T cell subsets, including Th1 (CCR6^-CXCR3⁺CCR4^-), Th2 (CCR6^-CXCR3^-CCR4⁺), Th17 (CCR6⁺) and follicular T helper cells (Tfh; CXCR5⁺). Whereas the frequencies of Th1, Th2 and Tfh cells did not differ between CTEPH patients and HCs (data not shown), we observed an increase in Th17 cells within the memory CD4⁺ T cells ( Figure 1 ). A sub-analysis of CCR6⁺ CD4⁺ T cells, identifying CXCR3^-CCR4⁺ Th17, CXCR3⁺CCR4^- Th17.1, CXCR3⁺CCR4⁺ double positive (DP) and CXCR3^-CCR4^- double negative (DN) cells, showed a particularly strong increase in the proportions of CXCR3^-CCR4^- DN T cells in CTEPH patients compared to HCs ( Supplementary Figures 2A, B ). Although DN Th17 cells are less studied than the classic Th17 and Th17.1 cells, evidence was provided that these cells have pathogenic features in rheumatoid arthritis (36) and display a stable Th17-lineage commitment, even under Th1 polarization conditions (37). They have the capacity to produce IL-17, but little IFNγ and co-express the transcription factors TBX21 and RORγt, which are associated with the Th1 and Th17 subset, respectively. In this context, we confirmed intracellular expression of the key transcription factor RORγt in the total CCR6⁺ CD4⁺ T cell population ( Supplementary Figure 2C ).

Significantly more CCR6⁺ T cells were also present in the fractions of naive CD4⁺ and CD8⁺ T cells, as well as Tregs ( Figure 1 ). This was not the case for CCR6⁺ memory CD8⁺ T cells, which would reflect IL-17-producing Tc17 cells (28, 38).

In conclusion, circulating T cells in CTEPH patients show an aberrant phenotype, characterized by increased proportions of CCR6⁺ cells in naive and memory CD4⁺ T cells, naive CD8⁺ T cells and Tregs.

The activation status of a T cell is reflected by the expression of activation and inhibitory markers, such as T cell co-stimulator (ICOS), programmed cell death 1 (PD1) and cytotoxic T lymphocyte antigen 4 (CTLA4) (39). Their expression was examined in the five T cell subpopulations, as gated in Figure 1 . CTLA4 was significantly increased in all T cell fractions from CTEPH patients compared to HCs, except for Tregs in which only a slight trend was observed ( Figure 2A ). PD1 and ICOS expression were unchanged ( Figures 2B, C and Supplementary Figure 3A, B ). No significant correlations between the three different markers were found (shown for memory CD4⁺ T cell fractions in Supplementary Figure 3C ).

In summary, these data indicate that both naive and memory CD4⁺ and CD8⁺ T cells of CTEPH patients have increased expression of CTLA4.

Subsequently, we used PCA to investigate if a specific T cell profile could separate treatment-naive CTEPH patients from HCs. We observed a non-random distribution that was not due to gender or age ( Figure 3A and data not shown). The T cell profile was able to significantly separate patients from HCs in dimension 3 (Dim3) ( Figure 3B ), to which CCR6 and PD1 contributed most ( Figure 3C ). Although CTLA4 expression was significantly different between patients and HCs in most T cell populations, these differences did not significantly separate the two groups in Dim1 ( Figures 3B, C ).

Taken together, quantification of CCR6 and PD1 expression creates a T cell profile that separates CTEPH patients from HCs in a multivariate analysis.

Next, we determined the cytokine-producing capacity of T cells in a subgroup of 38 CTEPH patients (gating strategy as in Figure 1 , except that Tregs were now identified by CD127^-CD25⁺ expression; Figure 4A ). We observed that the CD45RA^- memory CD4⁺ T cell fractions of CTEPH patients were less capable of producing TNFα, IFNγ, granulocyte-macrophage colony-stimulating factor (GM-CSF) and IL-4 compared with HCs, but intracellular IL-6, IL-10 and IL-17 did not differ between patients and HCs ( Figure 4B ). A correlation matrix of the proportions of cytokine-expressing CD45RA^-CD4⁺ T cells in HCs indicated a strong coordination between all cytokines analyzed except IL-6, but this was partly lost in CTEPH patients ( Figure 4C , examples shown in Supplementary Figure 4 ). The finding of mainly positive correlations across individual cytokines indicated that both CTEPH patients and HCs did not show skewing toward distinct Th subsets (because this would then have been visible in the matrices as negative correlations). Nevertheless, the significant positive correlation between the proportions of IL-17⁺, IFNγ⁺ and IL-4⁺ T cells in HCs was weaker in CTEPH patients. This observation suggested that in CTEPH patients the decrease of IFNγ⁺ or IL-4⁺ CD4⁺ T cells was not associated with a concomitant decrease or increase in IL-17-producing CD4⁺ T cells.

CD45RA^- memory CD8⁺ T cells from CTEPH patients showed reduced cytokine production, reaching significance for TNFα and IFNγ, while their IL-6 production was increased ( Figure 4D ; Supplementary Figure 5A, B ). No differences were found in cytokine-producing capacity of CD45RA⁺ naive CD8⁺ T cells between HCs and CTEPH patients, nor in the cytokine correlation matrix (data not shown).

In summary, CD45RA^-CD4⁺ and CD45RA^-CD8⁺ T cells of CTEPH patients are less capable of producing cytokines, whereby correlations between the different cytokines produced by CD4⁺ T cells are weaker than in HCs.

Furthermore, we used PCA to investigate whether besides the T cell subset and activation marker characteristics, cytokine profiles could separate CTEPH patients from HCs ( Supplementary Figure 6A ). Indeed, cytokine T cell profiles separated patients from HCs in Dim1 to which all cytokines produced by CD4⁺ and CD8⁺ T cells contributed except IL-6 ( Supplementary Figure 6A ). A comprehensive PCA analysis that included expression of chemokine receptors, activation markers and cytokine production, showed that HCs were mainly separated from CTEPH patients by cytokine profiles ( Supplementary Figure 6B ). Given the differences in prognosis and treatment of CTEPH patients with central and peripheral lesions (5), we performed a sub-analysis comparing these two patient groups. We excluded patients with mid lesions eligible for BPA but not for PEA. We found that T cell subset and activation marker profiles did not separate patients with central (n=15) and peripheral (n=9) lesions at diagnosis ( Supplementary Figure 7A ). However, T cell cytokine profiles [determined in a smaller group of patients; central lesion (n=11) and peripheral lesion (n=6)] did separate the two patient groups in Dim2, largely based on IL-6 production by CD4⁺ and CD8⁺ T cells and GM-CSF expression by CD8⁺ T cells ( Supplementary Figures 7B, C ). We did not find significant correlations between cytokine producing abilities of CD4⁺ or CD8⁺ T cells and clinical parameters of disease course, such as patient 1-year survival (data not shown).

In conclusion, all T cell cytokines analyzed, except IL-6, constitute a T cell profile that separates CTEPH patients from HCs. In addition, CTEPH patients with central and peripheral lesions could be separated based on IL-6 and GM-CSF production by T cells.

Next, we evaluated the T cell profiles of CTEPH patients over time. For this sub-study, we focused on a group of 22 patients of whom we had paired samples at diagnosis and at 1-year follow-up. The increase in CCR6⁺ cells in the three CD4⁺ T cell fractions analyzed ( Figure 1 ) was significantly more pronounced at 1-year follow-up, especially in the CD45RA⁺CD4⁺ (naive) T cells ( Figure 5A ). An increase was also observed in the CD8⁺ T cell fractions.

Cytokine expression remained stable over time for both CD4⁺ and CD8⁺ T cell populations, except for IFNγ and IL-17 in CD4⁺ T cells, which were increased at 1-year follow-up ( Figures 5B, C ). CTLA4 expression significantly decreased over time in CD45RA^-CD4⁺ (memory) T cells and in CD8⁺ T cells. In contrast, PD1 and ICOS expression were similar for all T cell fractions at the two time points ( Figure 5D ). Within the follow-up group of 22 patients, three patients underwent a PEA before 1-year follow-up ( Figure 5 ; depicted in blue). The T cell profile of these patients at 1-year follow-up were within the range of the 19 patients who did not undergo a PEA.

These results indicate that at 1-year follow-up, CTEPH patients present with substantial changes in the expression of CCR6, cytokines and activation markers in CD4⁺ and CD8⁺ T cells, compared to baseline.

Since the T cell profile changed substantially over time, we determined whether PCA would separate CTEPH patients at diagnosis from patients at 1-year follow-up. Indeed, significant differences were observed in Dim1 and Dim3, to which CTLA4 expression and the proportions of CCR6⁺ T cells contributed most, respectively ( Figure 6A ). Likewise, patients at diagnosis and at 1-year follow-up were separated in a PCA based on T cell cytokine profiles, particularly by IL-17 in CD4⁺ and CD8⁺ T cells and TNFα in CD8⁺ T cells in Dim2 ( Figure 6B ). In these two PCAs the average positions of patient clusters over time did not move towards the position of HCs ( Figures 6A, B ).

Individual PCA analyses of patients with central and with peripheral lesions did not separate patients at diagnosis and at 1-year follow-up, using T cell subsets and activation marker profiles or cytokine expression, likely due to low sample sizes ( Supplementary Figure 8A, B ; data not shown)

Overall, integration of T cell profiles by PCA clearly separated patients at diagnosis from patients at 1-year follow-up.

Next, we examined the relationship between the phenotype of circulating T cells and the concentrations of various inflammatory mediators in plasma at diagnosis determined previously (22). Various significant correlations were observed, including a positive correlation between Th17 cells (gated as CCR6⁺CXCR5^-CD45RA^-CD4⁺ T cells) and CXCL9 (a ligand for CXCR3) ( Supplementary Figure 9 ) of which the plasma levels correlated with patient survival (22). This positive correlation would be consistent with the recent finding that CXCL9 promoted T17 differentiation in a murine model of liver disease (40). In addition, plasma levels of the anti-inflammatory cytokine IL-10 showed a negative correlation with proportions of intracellular GM-CSF⁺, IL-4⁺ and IL-17⁺ CD4⁺ T cells ( Supplementary Figure 9 ), in line with the capacity of IL-10 to suppress T cell proliferative and cytokine responses (41).

Therefore, the aberrant phenotype of circulating T cells of CTEPH patients at diagnosis displayed correlations with plasma levels of cytokines and chemokines.

Additionally, we determined the presence of CCR6⁺ T cells in paraffin-embedded vascular lesions of 25 CTEPH patients using multiplex immunofluorescence. Vascular lesions of 19 CTEPH patients contained one or more CD3⁺ T cell clusters, the majority of which contained CCR6⁺ T cells ( Figure 7A and Supplementary Figure 10A ). T cell clusters were mostly located in organized thrombi and surrounding neo-vasculature, as determined by H&E staining. In the vascular lesions of the remaining 6 CTEPH patients CCR6⁺ T cells were present but were more randomly spread in the lesion ( Figure 7A ). The staining further included CD4, CD8 and FoxP3, enabling the determination of different T cell subpopulations ( Figure 7B ). Using Qupath quantification, we found that the majority of the CD3⁺ T cells were double negative (CD4^-CD8^-) T cells, followed by CD4⁺ T cells ( Figure 7C and Supplementary Figure 10B ). CCR6 expression was found on up to 60% of all T cells present, a substantial fraction of which were CD4⁺ ( Figure 7D ).

These findings demonstrate the presence of CCR6⁺CD4⁺ T cells in vascular lesions of CTEPH patients, both in T cell clusters and randomly distributed, supporting a role for CCR6⁺ T cells in pathology.