Differential responses of pulmonary vascular cells from PAH patients and controls to TNFα and the effect of the BET inhibitor JQ1

Pulmonary arterial hypertension (PAH) is a rare and incurable progressive disease of the lung vasculature which leads to right heart failure and death. It encompasses a group of diseases characterized by raised pulmonary vascular resistance, resulting from vascular remodelling in the pre-capillary resistance arterioles [1, 2]. Many factors contribute to the vasoconstriction and vascular remodelling, but despite therapeutic advances for PAH the 3-year survival is only ~ 70% [2‐4]. Treatment with pulmonary vasodilators, such as endothelin receptor antagonists, prostacyclin analogues and phosphodiesterase type V inhibitors have improved both morbidity and mortality but are not a cure [2, 5]. It is important, therefore, to understand the mechanisms of vascular remodelling in PAH and to determine novel therapies targeting these abnormalities.

Vascular remodelling is common to all types of PAH and in early stages of the disease there is a significant increase in inflammation and oxidative stress causing cell damage and programmed cell death [6, 7]. In contrast, increased proliferation is a hallmark of later disease [8]. Markers of increased inflammation, such as nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) activation, is seen in pulmonary vascular endothelial cells in patients with idiopathic PAH (IPAH) [9]. Endothelial cells, in vitro, produce cytokines that are also found circulating in the plasma of patients with IPAH and correlate with worse survival [10].

Epigenetic changes such as histone acetylation regulate inflammatory gene expression and cell proliferation, migration and survival [11, 12]. The histone acetylation/deacetylation balance is maintained by histone acetyltransferases (HATs) and histone deacetylases (HDACs) which control chromatin structure and the accessibility of transcription factors for their target genes [13]. Histone acetylation, is linked to heightened inflammatory gene expression [14, 15]. Bromodomains (BRDs) are conserved structures that read acetylated lysine’s to interpret the histone acetylation topography. BRD-containing bromodomain and extra-terminal (BET) proteins co-ordinate the regulation of genes involved in cell proliferation, apoptosis, and inflammation [16, 17].

The expression of BET proteins, specifically BRD4, is increased in PAH lung tissue, distal pulmonary arteries (PAs) and right ventricle (RV) compared to control tissues [18, 19]. Overexpression was also demonstrated in isolated PAH pulmonary artery smooth muscle cells (PASMCs) compared to control cells [18]. However, both BRD2 and 4 regulate inflammatory gene expression in several murine and human cell types [20‐22] and molecular mimics such as JQ1+, which block binding to acetylated histones, attenuate cell proliferation and differentiation in vivo and in cell lines [23‐25]. I-BET151, a structurally similar BET mimic, reduced inflammation in vivo and in vitro following LPS challenge [23]. In the Sugen/hypoxia rat model, BRD4 inhibition using JQ1 or siBRD4 in vivo reversed established PAH and decreased proliferation, increased apoptosis and restored mitochondrial membrane potential in PAH-PASMCs, suggesting that BRD4 upregulation may be pathologically associated with PAH [18].

The clinically available BET inhibitor, RVX208 (Apabetalone) showed promise in preclinical trials [26], where it reversed PA remodelling in various PAH rat models, potentially via modulation of proinflammatory, proproliferative and prosurvival pathways. Apabetalone has undergone phase I-III cardiovascular trials where it failed to show a significant effect on cardiovascular death but did show promise on secondary outcomes and was well tolerated [27]. A recent pilot study, clinical trial (NCT 03655704) was carried out to test the feasibility of Apabetalone for a future early-stage trial to evaluate inhibition of BRD in PAH [28]. This single-arm open-label study reported that apabetalone was feasible and that further studies are required to confirm the efficacy signal but it may be associated with beneficial effects when added to current PAH therapies.

We hypothesised that BET mimics will effectively suppress inflammation-driven functions of human pulmonary artery smooth muscle (HPASMC) and pulmonary microvascular endothelial cells (HPMEC). We investigated the effect of the BET mimic JQ1+ on the regulation of TNFα-driven inflammatory responses in HPASMCs and HPMECs from PAH patients and non-PAH subjects and determined the expression of BRD2 and 4 in vascular cells from PAH patients and control lung tissue.

The authors would like to highlight that the HPMEC data from non-PAH controls presented in this manuscript has previously been published [31] but is shown here in this paper for comparison to the data in PAH.

We confirm the previously reported upregulation of BRD4 expression in endothelial and smooth muscle cells in PAH lung tissue compared to control lung tissue from patients with normal lung function [18, 26, 32] and extended those studies to also show increased nuclear localisation of BRD2 in PAH cells. TNFα-driven IL-6 protein release from both HPMECs and HPASMCs was greater in PAH cells than from control cells despite greater levels of IL-6 mRNA being induced in cells from control subjects. Similarly, TNFα-induced CXCL8/IL-8 mRNA was much greater in healthy control HPMECs and HPASMCs than in PAH cells although elevated levels of CXCL8/IL-8 protein release was only seen in HPASMCs with similar levels observed in HPMECs. These differences in mRNA expression between healthy control and PAH cells were not due to changes in TNFα-induced NF-κB activation or recruitment of activated NF-κB p65 to the IL-6 and CXCL8/IL-8 promoters as these were similar in both cell types and between subject groups. The BET mimic JQ1 suppressed TNFα-induced IL-6 and CXCL8/IL-8 release and mRNA expression to a similar extent in control and PAH HPMECs and HPASMCs reflecting a similar degree of suppression of TNFα-induced NF-κB p65 and BRD4 recruitment to the IL-6 and CXCL8/IL-8 promoters. However, the IC₅₀ for JQ1 suppression of CXCL8/IL-8 release was 3-times higher (~ 300 nM) than that for IL-6 release (~ 100 nM) in HPMECs whereas the IC₅₀ for IL-6 suppression in HPASMCs was double that for CXCL8/IL-8 suppression (~ 200 vs ~ 100 nM) in HPASMCs.

These results could be linked to dysregulated acetylation in PAH. Previously we have reported that the HAT/HDAC ratio is altered in IPAH patients [31] and here we now demonstrate higher levels of nuclear BRD4 expression in PAH and that PAH smooth muscle cells appear to express more BRD4 than diseased endothelial cells.

Our results show that PAH cells secrete greater levels of IL-6 and CXCL8/IL-8 than cells from healthy control subjects in response to TNFα despite control subjects generating greater levels of mRNA. This suggests that there is dysregulation of post-transcriptional control of the synthesis and release of these proteins in PAH cells. Translational regulation of mRNA transcripts involves 3 major stages: initiation, elongation and termination [33]. RNA binding proteins bind to the 3'-untranslated regions (3'-UTRs) of mRNA to modulate mRNA stability in response to cellular stressors including hypoxia but may also act as translational repressors. Interestingly, the expression of the RNA binding proteins quaking (QKI) and cold-inducible RNA-binding protein (CIRP) are altered in PAH and affect cellular function [34, 35]. Overall, JQ1 has marked anti-inflammatory effects in both cell types studied irrespective of any mechanistic differences regulating IL-6 and CXCL8/IL8 transcription and secretion.

TNFα-stimulated inflammatory responses are associated with activation of the NF-κB pathway [36]. Upon activation NF-κB p65 moves to the nucleus where it forms a transcriptional activator complex with the promotor regions of inflammatory genes [37]. For maximal activity post-translational modifications such as phosphorylation or acetylation can occur which enhances DNA binding and transcriptional activity. Acetylation of p65 at Lysine-310 (K310) is required for optimal NF-κB transcriptional activity [38]. Acetylated p65, as well as acetylated histones, provides an additional binding site for BRD4 and this interaction is thought to be an important target for BET inhibitors to suppress inflammation [39]. In this study we demonstrated that NF-κB was activated in vascular cells from PAH patients and controls, translocating into the nucleus as early as 30 min after TNFα stimulation. We also showed that TNFα increases the recruitment of p65 and BRD4 to the IL-6 and IL-8 promotors. The bromodomain inhibitor JQ1 decreased recruitment of both p65 and BRD4 to the IL-6 and IL-8 promoters and reduced IL-6 and CXCL8 protein release from vascular cells but had no effect on TNFα-stimulated nuclear translocation of NF-κB p65. These results are in accordance with other studies which indicate BET inhibitors can modulate the transcription of NF-κB target genes in macrophages [23], renal tubular epithelial cells [40], tumour cells [41, 42] and rheumatoid fibroblast‐like synoviocytes [43].

In our studies in human primary pulmonary vascular cells, JQ1 had no effect on TNFα-stimulated NF-κB nuclear translocation, however some groups have shown effects of BET inhibition on NF-κB activation as detected by TransAm assays. In TNFα-treated HUVEC, BET inhibition using JQ1 or siRNA attenuated IKK-mediated activation of NF-κB pathway and decreased activation of p38 and JNK MAPKs [44]. The authors also noted that p38 and JNK inhibitors also blocked TNFα-stimulated NF-κB activation. In rheumatoid fibroblast-like-synoviocytes (RA FLSs) BET inhibition with JQ1 or Brd shRNA decreased TNFα-induced NF-κB–dependent transcription of a luciferase reporter gene and NF-κB target genes [43] following attenuation of TNFα-induced phosphorylation of IKKβ and IκBα, and translocation of nuclear NF-κB. The authors suggested that nuclear BET proteins could regulate the NF-κB pathway through changes in early cytoplasmic IKK signaling events. BET inhibition also prevented the cytoplasmic phosphorylation of the p38 MAPK pathway in TNFα-stimulated RA FLSs. Whilst in diffuse large B-cell lymphomas BET inhibition prevented oncogenic IKK activity [45]. These findings indicate that in some cell types BET proteins may affect cytoplasmic signaling through an unknown mechanism(s). It would be interesting to study the cytoplasmic BRD2 protein interactome in PAH pulmonary vascular cells compared to cells from non-PAH control subjects.

We were unable to show any effects of JQ1 on NF-κB nuclear translocation in the time frame studied and we did not explore the effect of JQ1 on the MAPK pathways. In future studies it would be interesting to explore in more detail the mechanisms by which BET proteins could modulate cytoplasmic IKK activity and MAPKs. This may also provide a role for the enhanced cytoplasmic levels of BRD2 seen in PAH pulmonary vessels reported here.

Earlier studies have shown that BRD2 co-operates with BRD4 to enable RNA polymerase 2 recruitment and run through coding region [46]. The combination of genome-wide transcriptomics and epigenetics of BRD2 and BRD4 function in fibroblast-like synovial cells revealed a close co-operativity between BRD2 and BRD4 actions particularly at a super enhancer that controlled the expression of IL-6 and IL-8 [47]. In addition, BRD2 and BRD4 bind to distinct regions of target genes [47] and have distinct protein–protein interactomes that contribute to the JQ1-induced rewiring of the cellular transcriptome [46].

JQ1 inhibited IL-1β and TNFα-induced expression of inflammatory genes by attenuating the chromatin access for key inflammatory transcription factors AP-1 and NF-κB and by reducing inflammatory pathways controlled by these proteins and by the p38 MAPK pathway [47]. BRD2 binding was more affected than BRD4 binding by JQ1 with a predominant effect on genes controlled by joint BRD2/BRD4 complexes which results in enhanced RNAP2 pausing at promoter sites leading to suppressed mRNA expression [48].

In support of the concept that BRD2 and BRD4 may have distinct roles in specific cell types, the BRD4-selective BET mimic AZD5153, unlike the pan-selective JQ1, has differential effects on natural killer cell inflammatory and cytolytic responses [49]. However, RVX208, which preferentially binds to the second bromodomain in BRD2 and BRD4, reversed the inflammatory and proliferative phenotypes of PAH HPMECs and HPASMCs and reversed monocrotaline- and Sugen5416 + hypoxia-induced in vivo models of PAH [26]. In vivo, both nebulised JQ1 and siBRD4 reversed Sugen/hypoxia-induced PAH with improvements in right heart function and pulmonary pressures reported [18]. Additionally, JQ1 was able to increase mitochondrial spare respiratory capacity and decrease membrane potential to restore a healthy phenotype in PAH-PASMC. Elevated BRD4 expression in PAH has been linked to down-regulation of miR-204. BET inhibition by JQ1 or siBRD4 resulted in upregulation of miR-204. BRD4 inhibition decreased NFATc2 and Bcl-2 mRNA levels and Survivin protein expression with an increase in the cell cycle inhibitor p21 resulting in decreased proliferation and increased apoptosis [18]. In these experiments the authors report a greater effect with JQ1 compared to the specific siBRD4 which is not surprising as JQ1 is a pan-BET selective bromodomain inhibitor and may be having synergistic inhibitory effects on the other BRD proteins known to be involved in PAH. As previously discussed BRD4 and BRD2 are involved in inflammation via activation of NF-κB resulting in the release of proinflammatory cytokines which are capable of activating the STAT3 signaling pathway [50, 51] and also causing DNA damage and subsequent down-regulation of miR-204 [18]. Thus BRDs may play a role in initiation of PAH and also sustaining the inflammatory, proproliferative and antiapoptotic phenotype. In an in vivo model of chronic hypoxia together with pulmonary inflammation, I-BET151 restored hemodynamic parameters to levels of control animals and partially corrected right ventricle hypertrophy [21]. These results could be due to IBET151 changing the acetylation profile of HIF-2α and its target genes. These results suggest that BRD inhibition may also be relevant for COPD related PH and other group 3 PH patients [21].

The increased expression of BRD2 and BRD4 in PAH pulmonary vascular cells may reflect an impact of inflammation since IL-6 can enhance control human coronary artery smooth muscle cell expression of BRD4 to levels seen in cells from PAH subjects [32]. Together with results from JQ1 presented here and in similar ex vivo and in vivo models of PAH it highlights the importance of both BRD2 and BRD4 in controlling the function of PAH.

There are limitations to this study, the main one being the small number (n = 4 per group) of subjects studied. Experiments were performed twice per subject where possible, one at a low passage number and again at a higher passage to rule out any effects of passage number on results. The diseased cells were isolated from tissue obtained at late-stage disease so perhaps do not reflect early stages of disease progression. We only investigated IL-6 and IL-8 as NF-κB activated target genes but many more are known and should be investigated. Future studies should use RNA-sequencing (seq) analysis, possibly together with p65 and/or BRD2/4 ChIP-seq or ATAC-seq analysis, to determine widespread differential gene effects resulting from the altered PAH acetylome in both cell types. Only pharmacological inhibition of BET proteins was performed. JQ1 is not selective for BRD4 and effects other BRDs, therefore experiments using specific siRNA or more selective BRD-specific BET mimics need to be performed. In mitigation, we show that both BRD2 and BRD4 had enhanced nuclear expression in pulmonary vascular endothelial and smooth muscle cells in PAH lung tissue and so both may be playing a role in the molecular phenotype of PAH.

In summary, our data shows that JQ1 can suppress TNFα-stimulated inflammation by disrupting the recruitment of p65 and BRD4 to IL-6 and IL-8 promoters. JQ1 had a greater effect on IL-6 in HPMEC and CXCL8/IL-8 in HPASMC which suggests distinct cell-specific regulatory control of these genes. This was further suggested by the elevated mRNA expression in healthy control samples but enhanced protein secretion in PAH-derived samples. The data support further development of BET mimics as novel anti-inflammatory agents for the treatment of PAH.