Abstract
Cystic fibrosis (CF) is a life-threatening disorder characterised by decreased pulmonary mucociliary and pathogen clearance, and an exaggerated inflammatory response leading to progressive lung damage. CF is caused by bi-allelic pathogenic variants of the cystic fibrosis transmembrane conductance regulator (CFTR) gene, which encodes a chloride channel. CFTR is expressed in endothelial cells (ECs) and EC dysfunction has been reported in CF patients, but a role for this ion channel in ECs regarding CF disease progression is poorly described.
We used an unbiased RNA sequencing approach in complementary models of CFTR silencing and blockade (by the CFTR inhibitor CFTRinh-172) in human ECs to characterise the changes upon CFTR impairment. Key findings were further validated in vitro and in vivo in CFTR-knockout mice and ex vivo in CF patient-derived ECs.
Both models of CFTR impairment revealed that EC proliferation, migration and autophagy were downregulated. Remarkably though, defective CFTR function led to EC activation and a persisting pro-inflammatory state of the endothelium with increased leukocyte adhesion. Further validation in CFTR-knockout mice revealed enhanced leukocyte extravasation in lung and liver parenchyma associated with increased levels of EC activation markers. In addition, CF patient-derived ECs displayed increased EC activation markers and leukocyte adhesion, which was partially rescued by the CFTR modulators VX-770 and VX-809.
Our integrated analysis thus suggests that ECs are no innocent bystanders in CF pathology, but rather may contribute to the exaggerated inflammatory phenotype, raising the question of whether normalisation of vascular inflammation might be a novel therapeutic strategy to ameliorate the disease severity of CF.
Abstract
CFTR-impaired endothelial cells have a pro-inflammatory phenotype that can attract and reinforce leukocyte extravasation. Endothelial cells possibly contribute to the excessive inflammatory phenotype observed in cystic fibrosis. https://bit.ly/2GRijq8
Introduction
Cystic fibrosis (CF) is an autosomal recessive disorder caused by impaired function of the cystic fibrosis transmembrane conductance regulator (CFTR) protein, which primarily acts as a chloride channel, mainly in epithelial cells [1]. CF is a multi-systemic disorder characterised by progressive lung disease and pancreatic insufficiency, while also featuring conditions of gastrointestinal involvement, liver disease and diabetes [2]. Lung disease determines much of the morbidity and mortality in CF patients [3]. While the CF lung phenotype is considered to be primarily caused by infections [3], the subsequent inflammation is simultaneously exaggerated and ineffective at eradicating pathogens, initiating a vicious cycle of infection and inflammation [4]. Persistent high-intensity inflammation inflicts permanent structural damage on the airways and impairs lung function, which ultimately results in respiratory failure and death [3, 4].
Endothelial cells (ECs) line the inner wall of blood vessels [5]. These cells regulate blood clotting, inflammation, angiogenesis and vascular tone [6]. Although CFTR is expressed and functional in normal ECs [7, 8], the vascular role of CFTR is less well studied in comparison to its well-established role in epithelial cells. Yet, micro- and macrovascular dysfunction have been documented in CF patients [9, 10]. Endothelial dysfunction is involved in CF-associated manifestations including diabetes and cardiovascular-, lung- and liver-associated complications [11, 12]. Nevertheless, the underlying mechanism of EC dysfunction and its role in disease progression remain elusive [11, 12]. Endothelial dysfunction is a predictor of cardiovascular risk, and genome-wide association studies detected a link between single-nucleotide polymorphisms in CFTR and coronary artery disease and flow-mediated arterial dilation [13, 14]. Elucidating the role of CFTR in ECs is therefore relevant for understanding the pathophysiology of CF and cardiovascular diseases [13, 14]. We thus employed an unbiased transcriptomics approach complemented by functional characterisation of EC properties upon CFTR impairment, and studied the possible relevance of our findings for CF patients.
Materials and methods
Details regarding materials and methods are provided in the supplementary material.
Cell culture
Freshly isolated human umbilical vein endothelial cells (HUVECs) were obtained from different donors and used as single-donor cultures between passage 1 and 4 as previously described [15, 16]. Blood outgrowth endothelial cells (BOECs) were freshly isolated from peripheral blood obtained from different healthy donors and CF patients from the outpatient clinic in their usual state of health without intercurrent infections (clinical information is shown in supplementary table S1) following previously described protocols with minor modifications [17, 18].
In vivo mouse assays
Leukocyte infiltration into lung and liver was analysed as previously described with minor adaptations [19]. Briefly, Cftrtm1Unc-Tg(FABPCFTR)1Jaw/J (CFTR knockout (CFTRKO)) mice [20, 21] were perfused with PBS and organs were collected for immunostaining (CD45 and CD105) or digested for flow cytometry analysis (CD45, CD31, intercellular adhesion molecule 1 (ICAM1) and vascular cell adhesion molecule 1 (VCAM1)).
Bulk RNA sequencing and analysis
RNA extracted with TRIzol was subjected to sequencing library preparation with the Lexogen QuantSeq 3′ mRNA-Seq library preparation kit (Lexogen, Vienna, Austria). Samples were indexed to allow for multiplexing. Library quality and size range was assessed using a Bioanalyzer (Agilent Technologies, Leuven, Belgium) with the DNA 1000 kit (Agilent Technologies). Libraries (2 nM) were sequenced on an Illumina HiSeq 4000 platform (Illumina, Eindhoven, the Netherlands). Single-end reads of 50 bp length were produced with a minimum of 1 M reads per sample. Quality control of raw reads was performed with FastQC v0.11.7 [22].
Immunoblotting
Protein extraction and immunoblot analysis were performed using a modified Laemmli sample buffer (125 mM Tris-HCl, pH 6.8 buffer containing 2% SDS and 10% glycerol) in the presence of protease and phosphatase inhibitors (Roche, Anderlecht, Belgium). Mitochondrial isolation was performed using the mitochondria isolation kit for cultured cells according to the manufacturer's instructions (Thermo Fisher Scientific, Geel, Belgium). Lysates were separated by SDS-PAGE under reducing conditions, transferred to a nitrocellulose membrane, and analysed by immunoblotting.
In vitro functional assays
Endothelial proliferation, lactate dehydrogenase (LDH) viability assay, scratch wound, trans-endothelial electrical resistance (TEER) and leukocyte adhesion assays were performed as described previously [19, 23–25].
Intracellular and mitochondrial reactive oxygen species analysis
Intracellular and mitochondrial reactive oxygen species (ROS) levels were measured using CM-H2DCFDA (Thermo Fisher Scientific) and MitoSOX (Thermo Fisher Scientific) according to the manufacturer's instructions, and quantified as previously described [26].
Oxygen consumption rate
Bioenergetics of ECs were determined on a Seahorse XF24 instrument (Agilent, Diegem, Belgium) as previously described [27].
Detection of glutathione species and NADPH
Targeted metabolites were measured as previously described [27].
Quantification and statistical analysis
Data represent mean±sem of at least three independent experiments. Statistical significance was calculated by standard two-tailed t-test (with Welch's correction when variances were significantly different between groups), ANOVA (for multiple comparisons within one dataset) and one-sample t-test (for comparisons to point normalised data) using Prism v8.2 (GraphPad Software, La Jolla, CA, USA). Bioinformatic analysis was carried out using in-house-developed BIOMEX software [28]. A p-value <0.05 was considered statistically significant.
Results
CFTR impairment leads to transcriptomic alterations in ECs
We first confirmed earlier findings [7, 8] of detectable CFTR expression in cultured primary HUVECs (hereafter referred to as ECs) at the protein level via immunoblotting and immunocytochemistry (figure 1a, b). In many cases, CFTR mutations result in lower levels of functional CFTR protein in the plasma membrane [29]. To study CFTR's function, we therefore used two complementary models to mimic this situation: 1) we blocked CFTR activity using the allosteric CFTR inhibitor CFTRinh-172, which is commonly used in CF research [30]; and 2) we silenced CFTR expression using lentiviral-mediated expression of two non-overlapping CFTR-specific short hairpin RNAs (shRNAs) to obtain a CFTR knockdown (CFTRKD). We refer to CFTR-blocked or CFTR-silenced ECs as “CFTR-impaired” ECs.
CFTRKD lowered CFTR protein levels by >55% (figure 1a–c, supplementary figure S1a) without altering protein expression of the typical endothelial markers von Willebrand factor and vascular endothelial cadherin (VE-cadherin) (supplementary figure S1b). While CFTR blockade or silencing only minimally affected EC survival (supplementary figure S1c, d), increasing CFTRKD efficiency (>95% at the protein level) increased cell death (>50%, supplementary figure S1d, e). We therefore performed further experiments at an intermediate silencing depth.
We carried out an unbiased transcriptomics analysis to screen for global changes induced upon CFTR impairment. Principal component analysis of all genes and hierarchical clustering analysis of the highly variable genes revealed that control and CFTR-impaired ECs grouped into distinct clusters, suggesting broad transcriptomic changes (figure 1d, e, supplementary figure S2a, b). Gene set enrichment analysis (GSEA) comparing control and CFTR-impaired ECs was used to associate groups of significantly differentially expressed genes with biological processes (supplementary table S2) [31]. Upon CFTR impairment, GSEA revealed increased expression of pro-inflammatory, ion transmembrane transport and hypoxia-related pathways, while gene sets involved in proliferation (cell cycle), cell migration and cell adhesion were downregulated (figure 1f, supplementary figure S2c, supplementary table S2).
We then set out to obtain congruent markers that are conserved across CFTR blockade and silencing. We hypothesised that congruent pathways are related to the chloride channel function of CFTR, whose malfunction is shared between all CF patients. Congruent upregulated gene sets suggested a role in EC activation (pathways related to tumour necrosis factor-α (TNF-α), signal transducer and activator of transcription 3 (STAT3), hypoxia-inducible factor 1-α (HIF1-α), interleukin (IL)-2 and IL-6), antigen binding and processing (human leukocyte antigen (HLA) class I, transporters associated with antigen processing and cathepsins), matrisome remodelling and ion transmembrane transport, while congruent downregulated gene sets implied a possible role in the cell cycle, migration and adhesion (supplementary figure S2d, supplementary table S3). We confirmed these findings by performing a meta-analysis of control versus CFTR-impaired ECs to rank differentially regulated genes and gene sets (figure 1g, supplementary table S3). Among the top 20 ranking genes were CD9, CYR61, CXCL1 and CCL2, which have been linked to leukocyte recruitment to the endothelial membrane and trans-endothelial leukocyte migration (supplementary table S3) [32–35]. The top three upregulated gene sets were involved in EC activation, STAT3 signalling and H+ transmembrane transport, while gene sets involved in the cell cycle and cell migration were the top three downregulated gene sets (figure 1g, supplementary table S3). Taken together, CFTR impairment induced a prominent pro-inflammatory and EC activation gene signature, accompanied by a downregulated gene signature of EC migration and proliferation.
Several key EC functions malfunction upon CFTR impairment
To investigate in more detail the compromised EC proliferation transcriptome signature upon CFTR silencing, we compared the expression of genes involved in various phases of the cell cycle. Genes coding for proliferation markers, transcriptional regulators, cyclins and cyclin-dependent kinases were downregulated upon CFTR impairment (figure 2a). These transcriptomic findings were confirmed by a decrease in EC proliferation as measured by tritium-labelled thymidine incorporation into DNA upon CFTR impairment (figure 2b, c).
The transcriptomics analysis further revealed that CFTR silencing downregulated the expression of genes involved in integrin and vascular endothelial growth factor (VEGF) signalling, and actin remodelling, suggesting a novel role of CFTR in EC migration (figure 2d). In agreement, the scratch wound assay confirmed reduced migration of CFTR-impaired ECs (figure 2e, f). Overexpression of an exogenous haemaglutinine (HA)-tagged CFTR in ECs (to facilitate visualisation) followed by immunostaining for the HA-tag revealed that CFTR was primarily expressed and enriched in membrane ruffling structures (e.g. lamellipodia) of migrating ECs (supplementary figure S3a), further supporting a role in EC migration. Confluent monolayers of CFTRKD ECs displayed compromised junctional integrity as measured by a higher proportion of discontinuous and reticulated VE-cadherin and CD31 junctions (figure 2g, h, supplementary figure S3b, c) [18, 36, 37]. Functionally, this corresponded to a decrease in the trans-endothelial electrical resistance (TEER) of CFTR-impaired ECs (figure 2i, j), which is in line with previously published findings on ECs in CF [38, 39]. Our data thus indicate an underappreciated role for CFTR in EC proliferation and migration, and in maintaining endothelial barrier integrity.
CFTR impairment induces oxidative stress and mitochondrial dysfunction
Increased oxidative stress is a well-known and noxious hallmark of CF pathophysiology [40]. CFTR silencing led to the induction of several gene classes related to an anti-oxidative stress response, including transcription factors capable of binding antioxidant response elements (ARE), glutathione-consuming enzymes, superoxide dismutase, peroxiredoxin, thioredoxin, and DNA repair and chaperone proteins (figure 3a). We confirmed these transcriptomic changes by flow cytometry analysis of total cellular oxidative stress, which indeed revealed increased oxidative stress upon CFTR impairment (figure 3b, c). When key cellular antioxidant scavengers were measured by liquid chromatography-mass spectrometry, both glutathione (GSH) and its reducing co-factor NADPH were affected, as indicated by the increased ratios of oxidised/total GSH and NADP+/NADPH (figure 3d–g). The mRNA transcript levels of glutamate-cysteine ligase, the rate-controlling enzyme in GSH biosynthesis, were unchanged in CFTR-silenced ECs (supplementary figure S4a), suggesting that increased GSH consumption rather than reduced GSH synthesis contributed to the elevated oxidative stress.
Measurements of mitochondrial ROS using the MitoSOX dye by flow cytometry revealed that CFTR impairment led to significantly elevated mitochondrial O2− levels (figure 3h, i). This was associated with decreased mitochondrial respiration as reflected by a reduced basal oxygen consumption rate, ATP production and maximal oxygen consumption rate (figure 3j–m), suggesting mitochondriopathy. The clearance of damaged mitochondria through autophagy, a process called mitophagy, is vital for cellular functioning and survival [41]. Immunoblotting revealed increased mitochondrial levels of dynamin related protein 1 (DRP1) and Parkin (mediating mitochondrial fission), but reduced levels of OPA1 mitochondrial dynamin like GTPase (OPA1) (mediating mitochondrial fusion), indicative of increased mitochondrial fission but reduced fusion (figure 3n–q). Taken together, these data show that CFTR deficiency induces an oxidative stress response, with more damaged mitochondria with defective mitochondrial function.
Autophagy deficiency in CFTR-impaired ECs
Differential gene expression analysis of our transcriptomic data revealed that autophagy markers p62 (SQSTM1) and microtubule-associated protein 1 light chain 3B (LC3B) (MAP1LC3B) were elevated upon CFTR silencing (figure 4a). Real-time qPCR analysis of these marker genes confirmed these findings (figure 4b). Corroboratively, the protein levels of the autophagic substrate p62 and autophagosome markers LC3BI and LC3BII were also increased (figure 4c–j), suggesting an accumulation of autophagosomes and a defect in autophagy flux rather than increased functional autophagy. To validate our findings, we performed transmission electron microscopy (TEM), which highlighted a two-fold increase in autophagosome abundance upon CFTR impairment, confirming defective autophagy (figure 4k–n, supplementary figure S4b). CFTRKD ECs displayed upregulated mTORC1 activity (a known inhibitor of autophagosome formation and lysosome biogenesis [42]) as measured by the increased phosphorylation of its downstream canonical target P70S6K and its substrate S6 (figure 4o–r). Our results suggest that CFTR impairment may hinder endothelial autophagy through mTORC1 activation and autophagosome accumulation.
CFTR impairment induces EC activation
Differential expression analysis in control versus CFTRKD ECs revealed a pro-inflammatory phenotype with CXCL1, CCL2 and ICAM1 ranking as the top three upregulated genes in CFTRKD cells (supplementary table S4). CFTR-silenced ECs upregulated transcripts of various pro-inflammatory genes, adhesion molecules and TNF-related signalling (figure 5a). In baseline conditions and upon pro-inflammatory stimulus (lipopolysaccharide (LPS) challenge), CFTRKD ECs had a more pronounced pro-inflammatory phenotype than control cells, with elevated expression of VCAM1, ICAM1, SELE (E-selectin) and IL8. While LPS stimulation effectively increased the expression of activation markers in control ECs, it had no further effect on their already increased expression in CFTRKD ECs (figure 5b–e). Emerging evidence highlights that a loss in autophagy could contribute to EC activation [43]. However, rescuing the pro-inflammatory phenotype with rapamycin (an mTORC1 inhibitor inducing autophagy) did not alter the expression levels in either baseline or LPS challenge conditions (supplementary figure S4c). We hypothesised that increased expression of adhesion molecules and chemokines by ECs could result in more attraction and binding of leukocytes onto the EC surface. Therefore, we performed an in vitro leukocyte adhesion assay and confirmed that, compared to control ECs, more leukocytes adhered to the surface of CFTR-impaired ECs at both baseline and after LPS stimulation (figure 5f–i).
We then investigated the role of CFTR in a bi-transgenic mouse model of CFTR harbouring a targeted knockout (CFTRKO) mutation of CFTR and expressing the human CFTR transgene under the expression of the FABP1 promoter (rescuing the lethal intestinal occlusion phenotype occurring at birth [21]). Immunostaining of CD45, a marker of leukocytes, showed increased leukocyte infiltration into the lung and liver parenchyma of adult CFTRKO mice compared to wild-type littermates (figure 6a–d). Flow cytometry measurements of single-cell lung suspensions confirmed a heightened CD45+ cell fraction and increased expression of VCAM1 and ICAM1 in viable CD31+ CD45− cells isolated from CFTRKO mice (figure 6e–g, supplementary figure S4d). Our data thus suggest that across different model organisms, CFTR impairment leads to a pro-inflammatory phenotype in ECs, which can attract leukocytes and reinforce their extravasation across the endothelium.
CFTR mutations cause a pro-inflammatory phenotype in CF patient-derived ECs
In the clinical context that patients with CF can experience sustained periods of inflammation with an exaggerated immune response [44], we validated our findings in an ex vivo model using blood outgrowth endothelial cells (BOECs), cells derived from circulating EC progenitors that express typical EC markers [45]. BOECs were cultured from peripheral blood taken from CF patients with severe mutations (Class I–III; patients’ clinical information is shown in supplementary table S1), and retained endothelial features when cultured in vitro (figure 7a, supplementary figure S4e). Confirming our previous results in HUVECs, wound closure and barrier integrity were also impaired in CF patient-derived BOECs (figure 7b, c). Similar to CFTRKD ECs, CF patient-derived BOECs displayed increased discontinuous and reticulated VE-cadherin junctions compared to BOECs from healthy donors (figure 7d–f). It has been suggested that such reticulated junctions are important in controlling leukocyte transmigration [37, 46, 47], corroborating the role of CFTR in barrier integrity and inflammation. Furthermore, mRNA levels of the pro-inflammatory ICAM1, SELE and IL-8 were increased in BOECs from CF patients compared to healthy donors, which was mirrored in the increased adherence of leukocytes to CF patient-derived ECs (figure 7g, h). These results correlate with our earlier findings, in which CFTR loss of function in our models resulted in a pro-inflammatory phenotype in ECs. More importantly, we found that rescuing CFTR function with the combination of CFTR modulators VX-770 and VX-809 (Orkambi®) was also able to partially rescue the pro-inflammatory gene expression signature, and significantly diminished leukocyte adhesion onto CF patient-derived ECs (figure 7h, i).
Discussion
Inflammation plays a major role in CF pathophysiology, in which it affects mainly the lungs but also organs such as the liver, intestines and pancreas [2]. The origin of the inflammation in CF-affected organs is currently debated because it is unclear whether it is of primary or secondary origin [4, 48]. A pro-inflammatory phenotype has been observed in different cell types independent of pro-inflammatory challenges [49–51], e.g. in HUVECs treated with CFTR inhibitor, which produced more IL-8 [39, 52]. Our bulk transcriptomic data revealed upregulated expression of multiple pro-inflammatory genes in CFTR-impaired ECs. Upon CFTR impairment in vitro and in CF patient-derived ECs ex vivo, the expression of several EC pro-inflammatory genes (ICAM1, IL-8 and SELE), which are essential molecules for leukocyte adhesion and diapedesis, was upregulated. This was mirrored by increased leukocyte adhesion to CFTR-impaired ECs. These findings confirm that CFTR dysfunction by itself leads to a pro-inflammatory phenotype even in the absence of a concomitant infection. CFTR-modulator treatment has already been successfully used to rescue CFTR function in other in vitro systems, even showing clinical correlation to patient response [53, 54]. Accordingly, the combination treatment of VX-770 and VX-809 (Orkambi®) was able to partially rescue the pro-inflammatory phenotype of CF patient-derived ECs, as measured by inflammatory marker gene expression and leukocyte adhesion. Our results are in accordance with earlier findings in CF patients in which increased soluble EC activation molecules were detected in the plasma and serum, correlating, in some cases, with impaired lung function [55–60]. Additionally, in the gut-corrected CFTRKO mouse model, microscopic quantifications revealed more leukocyte infiltration in lung and liver, with increased expression of leukocyte adhesion molecules in lung ECs. Although these CFTRKO mice do not display the inflammatory lung phenotype [20, 61], they do show signs of subsymptomatic inflammation, for which the vasculature could thus at least be partly responsible by promoting leukocyte adhesion and infiltration. Hence, CFTR impairment leads to an inflammatory EC phenotype with increased activation and leukocyte extravasation, highlighting the potential, but overlooked role of the endothelium as a mediator in the inflammatory process in CF. In support of a role of ECs in immune regulation, we recently observed that freshly isolated tumour ECs from human lung cancers expressed a transcriptome signature associated with diverse immune functions, suggesting a more important function of disease ECs in immune surveillance than previously anticipated [62].
We propose that three potential interlinked underlying mechanisms give rise to this increase in inflammation: 1) defective autophagy, 2) increased oxidative stress and 3) mitochondrial dysfunction (figure 8). Our study provides several lines of evidence that autophagy might be defective in ECs with impaired CFTR function. We observed increased protein levels of the autophagy marker LC3BI, thus confirming autophagy induction. However, analysis of autophagic flux using LC3BII and TEM images revealed autophagosome accumulation, indicating decreased autophagic flux due to defective autophagosome maturation and/or autolysosome function. Although an autophagy impairment has been reported in CF-deficient epithelial cells and macrophages, in which accumulation of p62 is documented together with a decrease in LC3BII and autophagosomes [63, 64], our mechanism seems to be distinct because it is characterised by an accumulation of autophagosomes. Autophagy is a highly dynamic process with several components and mediators involved, including mTORC1, a key regulator of autophagosome formation and lysosome biogenesis [65]. We observed increased mTORC1 activity in CFTRKD ECs, suggesting inhibition of autophagy. Indeed, a prominent function of mTORC1 is to restrict autophagy [42]; however, reactivation of mTORC1 is required for the termination of autophagy [66]. Defective autophagy has already been associated with a pro-inflammatory phenotype in different cell types, including ECs [67, 68], and could contribute to the pro-inflammatory phenotype in CF. In addition, it is well established that impaired cellular proteostasis and autophagy are involved in the pathogenesis of CF [69]. In fact, correcting the underlying proteostasis and autophagy defect has emerged as a novel intervention strategy in CF [69].
Increased oxidative stress is another hallmark of CF pathophysiology [40]. We observed increased oxidative stress in CFTR-impaired ECs, together with a striking transcriptomic signature in which multiple antioxidant-related genes were upregulated. Although our data revealed that the rate-controlling enzyme for GSH synthesis (a key intracellular antioxidant) was unchanged upon CFTR silencing, glutathione disulfide (GSSG) and NADP+ levels were increased, suggesting increased oxidative stress. Our findings confirm earlier studies in CF patients that documented heightened vascular oxidative stress, probably scavenging the vasodilator nitric oxide (NO), leading to reduced NO bioavailability, EC activation and dysfunction [70–72]. The role of excessive ROS levels in the respiratory system has mostly been characterised in connection with chronic pulmonary infections and persistent inflammation [40]. Additionally, low GSH levels have been found in plasma and blood neutrophils from CF patients [73, 74] and systemic GSH dyshomeostasis has already been suggested in CF [40]. Because of the heightened oxidative stress in CF patients, CF research has also explored the potential of antioxidant treatment [40, 75]. Recently, an antioxidant cocktail treatment in CF patients improved both vascular and lung function [71]. Although these results are promising, the beneficial effect of antioxidant therapy remains difficult to evaluate in CF patients with chronic infections on intensive antibiotic treatments and warrants a large-scale, long-term study [75].
Mitochondria can be ROS-producing organelles and CFTR function is necessary for optimal mitochondrial function [76, 77]. We consistently found increased mitochondrial ROS in CFTR-impaired ECs. A previous study in human bronchial epithelial cells reported that oxygen consumption and both mitochondrial complex I and IV activities were impaired in CF epithelial cells, concomitantly associated with increased mitochondrial ROS production and membrane lipid peroxidation [78]. We confirmed these findings in ECs, finding that both basal and maximal respiration as well as mitochondrial ATP production were decreased in CFTR-impaired ECs. Currently, the molecular mechanisms explaining how CFTR impairment affects so many parameters of mitochondrial function remain unknown [78]. In the absence of efficient removal of damaged mitochondria (mitophagy), ROS levels would be expected to rise. Our observations of increased mitochondrial fission and impaired mitophagy suggest that increased mitochondrial damage is at least partly responsible for the observed mitochondrial phenotype. However, we cannot conclude which mechanism is initiating the mitochondriopathy: (mitochondrial) ROS inducing autophagy/mitophagy or vice versa (figure 8). The process of autophagy/mitophagy in ECs might be particularly important in CF because impairment of this pathway contributes to inflammasome activation [67, 79].
From our transcriptomics data, we observed downregulation of essential marker genes for proliferation, as well as cyclins and cyclin-dependent kinases [80], consistent with our functional validation in vitro. The mechanisms by which chloride fluxes are affecting interactions between cyclins and cyclin-dependent kinases, thereby regulating cell cycle, are poorly characterised, especially in ECs. Although interesting, how and if an EC proliferation defect might contribute to CF disease progression is currently unknown.
EC migration is a regulated multistep process involved in tissue formation, wound healing and regeneration [81]. Upon CFTR impairment, ECs exhibited decreased migration, which is in accordance with their transcriptomic signature. We also obtained evidence that CFTR is expressed at the lamellipodia of migrating ECs and therefore is possibly involved in EC polarisation and migration. Our findings are consistent with reports that CFTR regulates lamellipodia protrusion and cell migration in non-EC types [82, 83]. It is becoming increasingly obvious that CFTR regulates cell migration across a diverse range of cell types; however, the underlying mechanisms remain to be elucidated, given that delayed wound repair may augment inflammation in CF [82].
The endothelium also functions as a semipermeable barrier where EC junctions dynamically open to allow the passage of ions, nutrients and inflammatory cells for tissue homeostasis and immune surveillance. In accordance with published literature [38, 39], we observed increased membrane permeability in CFTRKD cells and CF patient-derived ECs. This was probably caused by the larger number of discontinuous junctions present when CFTR was impaired. More interestingly, CFTRKD cells and CF patient-derived ECs displayed higher fractions of reticulated adherens junctions. Such reticulated junctions are important in controlling leukocyte transmigration [37, 46, 47]. This reinforces our hypothesis that CFTR impairment is associated with a pro-inflammatory and leaky barrier, which could possibly contribute to the excessive leukocyte extravasation observed in the lung of CF patients [84].
Our study has limitations and outstanding questions. For instance, how does CFTR, a protein in the plasma membrane that regulates ion conductivity, regulate diverse processes such as endothelial proliferation, migration, autophagy and inflammation? It is well known that CFTR is a promiscuous protein that also functions as a protein hub for scaffolding proteins at the plasma membrane and in the Golgi apparatus [85–87]. Protein–protein interactome analysis revealed that CFTR is involved in a wide range of processes (including proteostasis, cytoskeleton remodelling, immune response, ROS signalling, metabolism and proliferation), highlighting the complex underlying regulatory network [88]. Limitations to our study are the low endogenous CFTR levels in ECs, making its detection challenging, and the fact that we had to use a limited CFTRKD efficiency to avoid cellular toxicity at higher CFTRKD efficiency. However, this incomplete CFTRKD was sufficient to induce the observed vascular inflammation and other phenotypes, indicating its physiological relevance. It was thought that CF carriers (50% CFTR expression) could remain healthy [89, 90]; however, it was recently observed that CF carriers have increased risk for a wide range of CF-related conditions [91]. Thus, studying and comparing carrier- versus CF patient-derived ECs would give insightful information. It is well established that gut-corrected CFTRKO mice are not an ideal model organism to study CF because this model does not accurately reflect the human phenotype [20, 61]. Given that CFTR is absent in almost all cell types, other cell types could also influence the increased leukocyte extravasation found in murine CF lungs. However, endothelial leukocyte adhesion markers were increased in murine CFTRKO lung ECs, suggesting that the endothelium could partially mediate the excessive extravasation. Obtaining primary ECs from CF patients is challenging. We therefore used BOECs as an alternative, though the extent to which these cells reflect in situ organ-derived ECs from CF patients needs studying. Nevertheless, we observed a similar pro-inflammatory signature in BOECs from CF patients as in ECs with impaired CFTR function.
In conclusion, our findings provide new evidence that EC function is altered upon CFTR impairment, with defective cell proliferation, migration and autophagy. Remarkably, altered CFTR expression led to EC activation and a pro-inflammatory state, increasing leukocyte adhesion and extravasation in vitro, in vivo and ex vivo. Our data raise the question of whether ECs should still be considered as passive bystanders in CF pathology, or whether they actively co-determine the exaggerated inflammatory response. Hence, our study may provide an incentive to consider EC normalisation as a new therapeutic approach to reduce the pro-inflammatory status of ECs and thereby improve the quality of life of CF patients.
Supplementary material
Supplementary Material
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary methods ERJ-00261-2020.Supplement
Supplementary table S1. Patient genotypes used for BOEC generation. The table shows CFTR genotype, gender, age and forced expiratory volume in one second (FEV1) from each CF patient enrolled for the generation of blood outgrowth endothelial cells (BOECs). ERJ-00261-2020.Table_S1
Supplementary table S2. GSEA and meta-analysis of CFTR-impaired ECs. Bulk RNA sequencing was performed in primary HUVECs after inducing CFTR impairment with: (i) the highly specific CFTR inhibitor CFTRinh-(172) (CFTRinh), or (ii) CFTR-specific short hairpin RNAs leading to CFTR knockdown (CFTRKD). The table presents the results of gene set enrichment analysis (GSEA) for (i) Ctrl (DMSO) versus CFTRinh-(172) and (ii) Ctrl (PLKO) versus CFTRKD. The table also contains the results of congruent significantly up/down-regulated gene sets between CFTRKD and CFTRinh. A curated list of human vascular gene sets was used. Only gene sets with a minimal set size of 10 and with a significant adjusted p-value were considered for GSEA. For each top up- or down-regulated gene sets, enrichment score, normalised enrichment score (NES, used to generate figure 1f and supplemental figure S2c), p-values and adjusted p-values are indicated. Each gene set is presented with its official identifier and the corresponding name was used. This annotation was made according to the known gene functions within each gene set. ERJ-00261-2020.Table_S2
Supplementary table S3. Meta-analysis upon CFTR impairment. The table presents the results of meta-analysis of differential analysis between Ctrl versus CFTRKD, and Ctrl versus CFTRinh-(172) at the gene and gene set levels. Low rank numbers mean genes/gene sets upregulated upon CFTR impairment, while high rank numbers translate into genes/gene sets that are downregulated upon CFTR impairment. The product rank method was applied to generate meta-analysis scores. ERJ-00261-2020.Table_S3
Supplementary table S4. Differential analysis of CFTR-impaired ECs. The table presents the results of differential analysis expressed as Log fold change for: (i) Ctrl (DMSO) versus CFTRinh-(172) and (ii) Ctrl (PLKO) versus CFTRKD. For each detected gene, the average expression, p-values and adjusted p-values are indicated. ERJ-00261-2020.Table_S4
Shareable PDF
Supplementary Material
This one-page PDF can be shared freely online.
Shareable PDF ERJ-00261-2020.Shareable
Acknowledgements
The authors thank P. Agostinis and A. Luttun for discussion and advice. They also thank M. Parys, S. Wyns, C. De Legher, J. Souffreau, I. Betz, P. Vanwesemael, B. Verherstraeten, A. Manderveld, A. Carton, K. Peeters, N. Dai, L. De Keersmaecker and W. Vermaelen for technical assistance. We thank the Genomics Core Leuven for the RNA sequencing. The authors would like to thank the CF patients and hospital staff of the UZ Leuven CF centre who participated in this study.
Footnotes
This article has supplementary material available from erj.ersjournals.com
Author contributions: M. Declercq, L. Treps, P. Carmeliet, D. Cassiman and P. Witters designed the research. M. Declercq, L. Treps, P. de Zeeuw, M. García-Caballero, K. Brepoels, S. Vinckier and B. Ghesquière performed the research. P. de Zeeuw, N. Vasconcelos Conchinha, V. Geldhof, M. Ensinck, M.S. Carlon, A.S. Ramalho, M.J. Bird, M. Proesmans, F. Vermeulen, L. Dupont, B. Ghesquière, G. Eelen and M. Dewerchin contributed vital new reagents or analytic tools. M. Declercq, L. Treps, P. Carmeliet and P. Witters analysed data. M. Declercq, L. Treps, P. Carmeliet and P. Witters wrote the paper. All authors discussed the results and commented on the manuscript.
Conflict of interest: M. Declercq has nothing to disclose.
Conflict of interest: P. de Zeeuw has nothing to disclose.
Conflict of interest: N.V. Conchinha has nothing to disclose.
Conflict of interest: V. Geldhof has nothing to disclose.
Conflict of interest: A.S. Ramalho has nothing to disclose.
Conflict of interest: M. García-Caballero has nothing to disclose.
Conflict of interest: K. Brepoels has nothing to disclose.
Conflict of interest: M. Ensinck has nothing to disclose.
Conflict of interest: M.S. Carlon has nothing to disclose.
Conflict of interest: M.J. Bird has nothing to disclose.
Conflict of interest: S. Vinckier has nothing to disclose.
Conflict of interest: M. Proesmans has nothing to disclose.
Conflict of interest: F. Vermeulen has nothing to disclose.
Conflict of interest: L. Dupont has nothing to disclose.
Conflict of interest: B. Ghesquière has nothing to disclose.
Conflict of interest: M. Dewerchin has nothing to disclose.
Conflict of interest: P. Carmeliet has nothing to disclose.
Conflict of interest: D. Cassiman has nothing to disclose.
Conflict of interest: L. Treps has nothing to disclose.
Conflict of interest: G. Eelen has nothing to disclose.
Conflict of interest: P. Witters has nothing to disclose.
Support statement: M. Declercq and P. Witters are funded by the King Baudouin Foundation, Mucovereniging and UZ Leuven. L. Treps and M.S. Carlon are funded by an FWO Fellowship. P. Carmeliet is supported by the long-term structural Methusalem funding by the Flemish Government, grants from the Research Foundation Flanders (FWO-Vlaanderen), an ERC Proof of Concept Grant and an ERC Advanced Grant. Funding information for this article has been deposited with the Crossref Funder Registry.
- Received February 7, 2020.
- Accepted October 4, 2020.
- Copyright ©ERS 2021