Endothelial cells present an innate resistance to glucocorticoid treatment: implications for therapy of primary vasculitis
- 1Institute of Immunology, University of Muenster, Muenster, Germany
- 2Interdisciplinary Centre of Clinical Research, University of Muenster, Muenster, Germany
- 3Department of Paediatric Pulmonology, Allergology and Neonatology, Hannover Medical School, Hannover, Germany
- Correspondence to Dorothee Viemann, Department of Paediatric Pulmonology, Allergology and Neonatology, Hannover Medical School, Carl-Neuberg-Strasse 1, D-30625 Hannover, Germany;
- Received 3 August 2011
- Accepted 15 November 2011
- Published Online First 27 December 2011
Background In contrast to other chronic inflammatory diseases glucocorticoids alone do not maintain sufficient remission in primary vasculitis. The reasons for this therapeutic failure remain unclear.
Objectives To investigate the molecular effects glucocorticoids exert on endothelial cells (EC) and to elucidate the molecular pathways responsible.
Methods A comparative approach was used to treat human micro and macrovascular EC as well as monocytes long and short term with glucocorticoids or glucocorticoids and tumour necrosis factor alpha (TNFα). Gene expression changes were analysed applying microarray technology, sophisticated bioinformatic work-up and quantitative reverse transcription PCR. Glucocorticoid receptor translocation processes were traced by cell fractionation assays and immunofluorescence microscopy.
Results In EC glucocorticoids completely failed to inhibit the expression of immune response genes both after sole glucocorticoid exposure and glucocorticoid treatment of a TNFα-induced proinflammatory response. In contrast, an impressive downregulation of proinflammatory genes was seen in monocytes. The study demonstrated that the glucocorticoid receptor is comparably expressed in EC and monocytes, and demonstrated good translocation of ligand-bound glucocorticoid receptor allowing genomic glucocorticoid actions. Refined gene expression analysis showed that in EC transactivation takes place and causes glucocorticoid side effects on growth and metabolism whereas transrepression-mediated anti-inflammatory effects as in monocytes are missing. Insufficient induction of SAP30, an important constituent of the Sin3A-histone deacetylase complex, in EC suggests impairment of transrepression due to co-repressor absence.
Conclusions The impressive unresponsiveness of EC to anti-inflammatory glucocorticoid effects is associated with deficiencies downstream of glucocorticoid receptor translocation not affecting transactivation but transrepression. The findings provide the first molecular clues to the poor benefit of glucocorticoid treatment in patients with primary vasculitis.
Glucocorticoids play a decisive role in the pharmacotherapy of inflammatory diseases and represent the most frequently used anti-inflammatory drugs.1 However, in vasculitis they fail to achieve long-lasting control of inflammation. Primary vasculitis is characterised by chronic inflammation of the vessel wall,2 and endothelial cells (EC) crucially contribute to the complex pathogenesis of vasculitides.3 4 The treatment strategy of vasculitis comprises an induction and maintenance therapy. Remission is induced with high-dose glucocorticoids followed by immunosuppressants plus low-dose glucocorticoids because glucocorticoid monotherapy does not sustain remission.5 6 The molecular reactions glucocorticoids induce in EC are largely unclear. Elucidating these mechanisms might help to understand the particular glucocorticoid resistance in vasculitis.
Glucocorticoid effects are mediated by non-genomic and genomic mechanisms, the latter including transactivation and transrepression. Whereas glucocorticoid side effects are mainly induced by transactivation, the majority of anti-inflammatory effects are supposed to be exerted by transrepression.1 In monocytes it has been shown that glucocorticoids suppress inflammation by repression of inflammatory and induction of anti-inflammatory genes.7,–,9 After glucocorticoid binding to the cytosolic glucocorticoid receptor the complex translocates into the nucleus where it binds to specific DNA sequences (glucocorticoid-responsive elements) mediating the transactivation or inhibition of gene expression.10 Other inflammatory genes are indirectly repressed through glucocorticoid receptor interference with activating transcription factors or their coactivators such as nuclear factor kappa B (NF-κB), activator protein 1 (AP-1), STAT5 and IRF3, called transrepression.10,–,13
To understand the clinically poor anti-inflammatory response of vasculitis patients to glucocorticoids and speculating that EC play an important role in the pathogenesis of primary vasculitis we investigated the molecular effects glucocorticoids exert on EC to identify differences to the well established anti-inflammatory effects in monocytes. In striking contrast to human monocytes, analysis of genomic actions in resting and inflammatory stimulated EC could not reveal any immunosuppressive glucocorticoid effects. Tracking the path glucocorticoids take to exert genomic actions we demonstrated that the reason for deficient anti-inflammatory glucocorticoid effects in human umbilical vein endothelial cells (HUVEC) is an impairment of transrepression due to co-repressor absence located downstream of glucocorticoid receptor translocation. To our knowledge this study for the first time provides evidence for an innate resistance of EC regarding glucocorticoid-induced anti-inflammatory actions.
Material and methods
Pooled HUVEC were purchased from Cambrex (USA) for microarray analysis. For all other experiments primary EC were harvested from human umbilical veins of nine different donors as per protocol.14 HUVEC were cultured in endothelial basal medium with 2% fetal calf serum (FCS), 10 ng/ml human epidermal growth factor, 12 µg/ml bovine brain extract, 50 µg/ml gentamicin and 50 ng/ml amphotericin B, without supplementing steroids. Before stimulation they were cultured in Dulbecco's modified eagle medium (DMEM) with 2% FCS overnight.
Human microvascular endothelial cells (HMEC) type 1 were provided by Dr Candal (Atlanta, Georgia, USA) and cultured in MCDB131 medium with 10 nM L-glutamine, 50 µg/ml gentamicin, 10 ng/ml epidermal growth factor and hydrocortisone and FCS as indicated. Human monocytes were isolated from ABO-identical buffy coats of three healthy donors using Pancoll (Pan-Biotech, Aidenbach, Germany) and Percoll (Pharmacia, Uppsala, Sweden) density gradient centrifugation. Culture conditions were used as described.15 Twenty-four hours before stimulation, monocytes were cultured in teflon bags.
Treatment settings for all cell types were stimulation for 4 h and 16 h with 2 ng/ml tumour necrosis factor alpha (TNFα), dexamethasone (both Sigma-Aldrich, Deisenhofen, Germany) as indicated or TNFα plus dexamethasone.
Quantitative real-time RT–PCR
Total cellular RNA was isolated from HUVEC, HMEC and monocytes using the RNeasy Mini Kit (Qiagen, Hilden, Germany); 1 µg RNA was used for complementary DNA synthesis with the RevertAid H Minus First Strand cDNA synthesis kit (Fermentas, St Leon-Rot, Germany). Quantitative reverse transcription (RT)–PCR was performed using the QuantiTect SybrGreen RT–PCR Kit (Qiagen). Data were acquired with the ABI Prism 7900HT sequence detection system (Applied Biosystems, Darmstadt, Germany). Primers were obtained from MWG-Biotech (Ebersberg, Germany) (see supplementary table S1, available online only). Expression was normalised to the housekeeping gene GAPDH. RPL served as the second confirmatory housekeeping gene. The comparative threshold cycle method was applied to determine relative expression differences.16 For statistical analysis the results of all quantitative RT–PCR were assessed by Student's t test and are expressed as means±SD.
DNA microarray hybridisation and statistical data analysis
After 4 h and 16 h exposure of HUVEC and monocytes to medium or 10−5 M dexamethasone total RNA was isolated from independent experiments and processed for hybridisation to human genome U133 Plus 2.0 Arrays (Affymetrix, Santa Clara, California, USA) according to the manufacturer's instructions. Fluorescent signals were detected by the GeneChip scanner 3000 and analysed by GeneChip operating software 1.4 (Affymetrix). For sophisticated data analyses we used the GeneData expressionist suite software (Basel, Switzerland) as described.17 Only genes with an expression fold change of 2.0 or greater and a p value less than 0.05 (paired t test) were retained. ‘On/off’ regulation of genes was analysed as described.17 To cluster for overrepresented functional gene categories we compared the distribution of GO annotations on the gene chip with the group of glucocorticoid-regulated genes applying Fisher's exact test. In the case of genes that are represented by two or more probe sets only one transcript was taken into account to avoid potential bias.
Western blot analysis
HUVEC and monocytes were exposed to medium or 10−7 to 10−5 M dexamethasone for 1 h and 2 h (glucocorticoid receptor analysis) or 24 h (SAP30 analysis). Using high-salt lysis buffer whole cell lysates were prepared. The Bradford assay was performed for protein quantification. Lysates were adjusted to equal protein concentrations, subjected to sodium dodecylsulphate polyacrylamide gel electrophoresis and blotted onto nitrocellulose. Staining was performed using anti-glucocorticoid receptor (E-20) polyclonal rabbit antibody (Santa Cruz, Heidelberg, Germany), anti-SAP30 polyclonal rabbit antibody (Upstate, Lake Placid, USA) and anti-β-actin monoclonal mouse antibody (Sigma-Aldrich). Horseradish peroxidase conjugated anti-rabbit (Cell Signaling, Frankfurt, Germany) and anti-mouse (Dao, Hamburg, Germany) secondary antibodies were used for enhanced chemiluminescence detection. HeLa cell nuclear extract (Upstate) served as a positive control of SAP30 staining. For evaluation of cellular glucocorticoid receptor distribution HUVEC lysates underwent differential centrifugation to separate nuclear and cytoplasmatic protein fractions and the aforementioned glucocorticoid receptor staining was performed. The nuclear marker Lamin B (Santa Cruz) served as a control for the successful separation of protein fractions.
In three independent experiments HUVEC and monocytes were treated with medium or varying concentrations of dexamethasone (10−9 to 10−5 M) for 1 h and 2 h. They were fixed with acetone and stained with anti-glucocorticoid receptor (E-20) polyclonal rabbit antibody (2–4 µg; Santa Cruz) or rabbit control antibody (Pierce, Thermo Scientific, Schwerte, Germany) and a FITC-coupled anti-rabbit secondary antibody (Dianova, Hamburg, Germany). Nuclear staining was performed using 4,6-diamidino-2-phenylindole (Sigma-Aldrich). Images were captured with an AxioObserver.Z1 inverted microscope (Zeiss, Oberkochen, Germany) using an EC Plan-Neofluar 40× and Plan-Apochromat 63× objective.
Inducibility of glucocorticoid-sensitive genes in HUVEC
To get access to glucocorticoid actions in EC we first checked the inducibility of genes such as IL1R2, DUSP1, FPR1 and FKBP5 in HUVEC, which are well known to be glucocorticoid regulated in other cell types.18,–,20 In a comparative approach HUVEC and monocytes were exposed to increasing doses of dexamethasone (10−9 to 10−5 M). We chose this dosage range because dexamethasone concentrations of 10−9 to 10−6 M are common and effective for monocyte treatment. To exclude insufficient glucocorticoid dosage as an issue in EC we scaled the dosage range up to 10−5 M.7,–,9 21 Performing quantitative RT–PCR revealed that short-term and long-term glucocorticoid treatment of monocytes resulted in an impressive upregulation of FKBP5, DUSP1, IL1R2 and SAP30, whereas in HUVEC glucocorticoids did not significantly change the expression of these genes after 4 h and 16 h (figure 1A,B). Furthermore, the expression of immune response-associated genes such as TLR7, CD83, CXCL9, CXCL10 and CXCL11 could be significantly repressed by glucocorticoids in monocytes in contrast with a lacking response in HUVEC (figure 1C,D). Varying the dexamethasone concentration did not alter our findings. To avoid culture conditions in HUVEC as an error source, we adapted them in many ways. We used different HUVEC batches, varied the confluence between 50–60% and 90–100%, cultured HUVEC without supplementing steroids, used phenol-free DMEM,22 and scaled the FCS content down. None of these measures achieved any effects regarding the expression of glucocorticoid-inducible genes or the suppression of proinflammatory genes in HUVEC (data not shown).
Glucocorticoid effects on inflammatory-activated HUVEC
To analyse whether glucocorticoids are able to inhibit the expression of inflammatory induced genes HUVEC and monocytes were exposed to TNFα (2 ng/ml) and TNFα plus dexamethasone (10−9 to 10−5 M) for 4 h and 16 h. By quantitative RT–PCR we verified a TNFα-mediated upregulation of several proinflammatory genes in HUVEC and monocytes, as described earlier.7 17 In monocytes glucocorticoids significantly inhibited the TNFα-induced expression of TNFSF10, IL1B and CCL5 after 4 h (figure 1E). Upon 16 h of glucocorticoid treatment the TNFα-induced expression of VCAM1, TNFSF10, IL1B, CCL5 and TNFAIP6 was significantly downregulated (figure 1G). In contrast, in TNFα-activated HUVEC neither after 4 h nor after 16 h could a significant inhibiting effect be observed (figure 1F,H). Varying the dexamethasone concentration between 10−9 and 10−5 M did not alter our findings.
To exclude EC type-specific differences23 24 all experiments were also performed with HMEC. Results were comparable to the effect that the glucocorticoid treatment of HMEC neither significantly altered basal or inflammatory-induced gene expression with respect to genes involved in inflammatory responses (data not shown).
Our data show that in EC glucocorticoids neither induce genes, which are known to be glucocorticoid-dependent in other cell types, nor do they suppress the expression of TNFα-induced proinflammatory genes.
Gene profiling of glucocorticoid-treated HUVEC
As our gene selection for quantitative RT–PCR was based on glucocorticoid-regulated genes known from other cell types we might have missed glucocorticoid-sensitive genes in HUVEC. To screen the whole genome for glucocorticoid-regulated genes HUVEC cultured in phenol-free DMEM and monocytes were stimulated with 10−5 M dexamethasone for 4 h and 16 h or left untreated and processed for microarray analysis. After the assignment of gene ontology annotations we looked in detail for expression changes of immune response-associated genes annotated by GO:0006968 cellular defence response, GO:0006955 immune response, GO:0006954 inflammatory response, GO:0006952 defence response, GO:0030236 anti-inflammatory response, GO:0050727 regulation of inflammatory response, GO:0019882 antigen presentation or GO:0030333 antigen processing.
Whereas in monocytes 22% of immune response-associated genes were downregulated after 4 h of glucocorticoid treatment and 28% of immune response-associated genes were downregulated after 16 h of glucocorticoid treatment,7 respectively, a negligible proportion of 2% was found in HUVEC (table 1; see supplementary tables S2 and S4, available online only). Apparently, in HUVEC glucocorticoids exert virtually no negative regulating effects on inflammation and immune response-associated genes.
Analysing the overrepresentation of functional gene categories confirmed statistically that in monocytes glucocorticoids primarily downregulate immune response-associated genes (figure 2A,B). In contrast, in HUVEC we could not identify overrepresented downregulated groups indicating an anti-inflammatory glucocorticoid effect. After 4 h glucocorticoid treatment genes mainly involved in cell cycle regulation, metabolism and cell adhesion were downregulated (figure 2C), whereas cell cycle regulatory genes were overrepresented after 16 h (figure 2D).
Glucocorticoid receptor expression in HUVEC and monocytes
As anti-inflammatory glucocorticoid effects are supposed to be mediated by genomic actions after glucocorticoid receptor binding25 we considered differences in glucocorticoid receptor expression in monocytes and HUVEC. Performing western blot analysis we investigated glucocorticoid receptor expression in whole cell lysates. The glucocorticoid receptor was comparably detectable in untreated and glucocorticoid-treated monocytes and HUVEC (figure 3A). In both cell systems glucocorticoid receptor expression was slightly decreased after glucocorticoid exposure. So differences in glucocorticoid receptor expression could be excluded to explain missing anti-inflammatory glucocorticoid responses in HUVEC.
Nuclear glucocorticoid receptor translocation in HUVEC and monocytes
Subsequently, we investigated whether glucocorticoid treatment of HUVEC induces nuclear glucocorticoid receptor translocation. The cellular distribution of glucocorticoid receptor was analysed by the western blot and immunofluorescence technique. For the western blot approach nuclear and cytoplasmatic proteins were separated after 1 h and 2 h glucocorticoid treatment (figure 3B). In HUVEC exposed to medium the amount of glucocorticoid receptor in the cytoplasm remained significantly higher than in the nuclear fraction. Upon glucocorticoid stimulation the amount of cytosolic glucocorticoid receptor decreased over time while it increased in the nuclear fraction suggesting a ligand-dependent glucocorticoid receptor shift in HUVEC after glucocorticoid treatment.
Immunofluorescence microscopy clearly confirmed this result. In untreated monocytes and HUVEC the glucocorticoid receptor was mainly localised in the cytoplasm whereas the nucleus remained nearly free. Upon glucocorticoid exposure we found a nuclear staining pattern of glucocorticoid receptor in both cell types, indicating glucocorticoid receptor translocation after steroid binding (figure 4). Varying dexamethasone concentrations from 10−7 to 10−5 M did not alter these findings.
Impaired induction of SAP30 in HUVEC
In contrast to monocytes quantitative RT–PCR showed an impaired glucocorticoid-mediated transcriptional induction of SAP30 in HUVEC (figure 1A,B). SAP30 is an important subunit of the Sin3A–histone deacetylase (HDAC) co-repressor complex. Co-repressor complexes with HDAC activity are crucially involved in the mediation of genomic anti-inflammatory glucocorticoid effects.26 Therefore, we investigated SAP30 protein expression in whole cell lysates. Whereas in monocytes the amount of SAP30 significantly increased after glucocorticoid treatment, SAP30 expression in HUVEC remained constantly low and was not altered by glucocorticoid treatment (figure 5). The results could be confirmed by flow cytometry (data not shown).
Glucocorticoids are mainstays of anti-inflammatory regimens in the treatment of inflammatory diseases. However, they achieve only a poor long-lasting benefit in most forms of vasculitis.27 In systemic inflammatory response syndromes activated EC tag the place of inflammation and coordinate the attraction of leucocytes. In primary vasculitis EC are important contributors to the complex pathogenesis of vasculitis,3 4 and might represent valuable primary targets of anti-inflammatory regimens. It remains unexplained why many forms of vasculitis are insensitive to glucocorticoids, at least regarding prolonged remission.
To exert genomic effects glucocorticoids have to bind the cytosolic glucocorticoid receptor20 afterwards translocating as complex into the nucleus.10 Classic gene induction by binding to glucocorticoid responsive elements in glucocorticoid-sensitive genes is mainly called to account for side effects.28 29 The major anti-inflammatory effects are accomplished by the inhibition of proinflammatory genes via transrepression.1 10 13 28 30 Thereby, the glucocorticoid receptor interacts with transcription factors such as NF-κB or AP-1 or their coactivators and represses their transactivating effects on proinflammatory genes.31
Until now the molecular effects glucocorticoids exert in human EC have been poorly examined. Few groups relate to the perturbing influence of glucocorticoids on mitogen-activated protein kinase pathways32 or glucocorticoid receptor interference with NF-κB and AP-1. However, either these studies do not demonstrate the attenuation of inflammatory surrogates32 or their results are conflicting concerning anti-inflammatory glucocorticoid effects.33 34 Our study aimed to elucidate the molecular glucocorticoid effects on EC in detail. We chose a comparative approach and analysed whether the well established anti-inflammatory glucocorticoid response in human monocytes7 9 35 36 could also be elicited in human EC.
Surprisingly, neither in macrovascular HUVEC nor in microvascular HMEC did we find an induction of anti-inflammatory genes such as IL1R2, DUSP1, FPR1 and FKBP5 or a repression of proinflammatory and immune response-associated genes such as CXCL9, CXCL10 and CXCL11 well known to be glucocorticoid sensitive in other cell types.7 18,–,21 30 37 Fürst et al38 described an induction of DUSP1 in HUVEC by 10−9 M dexamethasone but the detected fold changes were at most three on protein and less than two on the messenger RNA level. We were not able to reproduce this minor glucocorticoid-mediated induction of DUSP1.
Subsequently, we expanded the analysis onto the whole genome using microarray technology. We could not identify any anti-inflammatory gene expression pattern in EC but found primarily the repression of cell cycle regulation and metabolism-associated genes. Moreover, contrary to monocytes glucocorticoids were neither able to inhibit TNFα-induced proinflammatory responses in HUVEC or HMEC. Effects on growth and metabolism are generally counted among side effects and are primarily caused by transactivation.28 29 Our data suggest that in strict contrast to glucocorticoid-sensitive monocytes glucocorticoids seem not to be able to induce an anti-inflammatory phenotype in HUVEC. Anti-inflammatory effects are mainly accomplished by transrepression requiring glucocorticoid binding to glucocorticoid receptors and nuclear translocation.1 10 13 28 30 As glucocorticoid receptor expression is reported to vary in distinct cell types30 and to correlate with steroid resistance,39 we considered a lower glucocorticoid receptor expression in HUVEC compared with monocytes, which we were, however, able to exclude.
The exact mechanisms governing glucocorticoid receptor translocation are not yet identified. An impaired glucocorticoid receptor shift has been found in patients with steroid-resistant asthma.40 We therefore considered reduced nuclear glucocorticoid receptor import in HUVEC compared with monocytes. Analysing expression levels of transport proteins known to be important for translocation such as importin-α and importin-1310 our microarray data provided no evidence for expression differences in monocytes and EC. Finally, in both cell types we found a good nuclear shift upon glucocorticoid exposure using immunofluorescence as well as glucocorticoid receptor detection in cytosolic and nuclear cell fractions via western blot.
Consequently, the reason for steroid resistance of EC must be found downstream of glucocorticoid receptor translocation. Lack of anti-inflammatory effects despite regular glucocorticoid receptor translocation and the presence of glucocorticoid side effects has also been observed in patients with steroid-resistant asthma.31 Low ratios of the isoforms glucocorticoid receptor-α/glucocorticoid receptor-β might cause glucocorticoid receptor unresponsiveness as glucocorticoid receptor-β is suggested to exert dominant negative effects on glucocorticoid receptor-α.31 41 42 However, our microarray data show a comparable glucocorticoid receptor-α/glucocorticoid receptor-β ratio in EC and monocytes not suggesting such kind of disproportion. Finally, impaired co-repressor functions downstream of glucocorticoid receptor translocation, which are essential for the successful transrepression of proinflammatory transcription factors, have to be considered. In this respect Sin3A–HDAC is a crucial multiprotein complex consisting of Sin3A, HDAC1, HDAC2, retinoblastoma suppressor-associated protein 46 (RbAp46), RbAp48 and polypeptides such as SAP30 and SAP18, which modifies gene expression by histone deacetylation.43 44 Reduced activity of HDAC2, a pivotal component of Sin3A–HDAC, has been shown to correlate with experimental and clinical steroid resistance.45 46 Here we show that in HUVEC glucocorticoids completely fail to induce SAP30. Insufficient SAP30 expression resulting in impaired Sin3A–HDAC co-repressor activity is a reasonable explanation for the insufficient transrepression specifically affecting EC.
In summary, to our knowledge this is the first report demonstrating an impressive failure of glucocorticoids to regulate immune response-associated genes in human EC. We located the molecular deficiencies downstream of glucocorticoid receptor translocation and found an EC-specific impairment of glucocorticoid-induced SAP30 expression that is a crucial member of the Sin3A–HDAC co-repressor complex. Our findings might provide a first molecular insight explaining the poor benefit of glucocorticoid treatment in primary vasculitis.
Funding This work was supported by the Deutsche Forschungsgemeinschaft (DFG) grants VI 538/1-1 to DV and RO 1190/9-1 to JR, the Interdisciplinary Centrum of Clinical Research (IZKF) grant Ro2/004/10 to JR.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.