The murine lung microvascular endothelial cell line (L2 MVEC) was cultured in RPMI medium (Gibco, Belgium) supplemented with 2 mM l-glutamine (Gibco), 0.1 mg/mL streptomycin (Sigma, Belgium), 200 U/mL penicillin (Kela, Belgium), and 10% FCS (Gibco). Cells were grown in 5% CO₂ at 37°C.

All animal experiments were performed in accordance to the regulations as declared in Directive 2010/63/EU from the European Union and the Belgian Royal Decree of May 29, 2013, and were approved by the Animal Ethics Committee from the KU Leuven (project number P163-2014, License LA1210186, Belgium). All efforts were made to minimize suffering of animals. Unless otherwise indicated, male Balb/c mice were obtained from Janvier (7–8 weeks old, Le Genest-Saint-Isle, France). Mice were injected intraperitoneally with 10⁶ PbNK65-infected red blood cells (a kind gift of the late Prof. D. Walliker, University of Edinburgh). Mice were kept in a conventional animal house and drinking water was supplemented with 4-amino benzoic acid (0.375 mg/mL, PABA, Sigma-Aldrich, Bornem, Belgium). Parasitemia was determined by microscopic analysis of tail blood smears after Giemsa staining (1/10 dilution, VWR, Heverlee, Belgium). Mice were sacrificed 8 days after infection (when parasitemia was approximately 4%) by euthanasia with Dolethal (Vtoquinol, Aartselaar, Belgium; 200 mg/mL, intraperitoneal injection of 50 µL) and cardiac punctures were performed.

After cardiac punctures, blood was filtered with Plasmodipur filters (Europroxima) to remove leukocytes and washed with RPMI medium (Gibco) supplemented with 25 mM HEPES, 0.425 g NaHCO₃, 2 mM l-glutamine (Gibco), 5 mM glucose (Sigma), and 20% FCS. The pellet was resuspended and seeded into culture flasks. Cells were gassed with a mixture of 92.5% N₂, 5.5% CO₂, 2% O₂ and cultured overnight at 37°C. The next day the cultures were centrifuged, resuspended in RPMI medium and loaded on MACS columns (Miltenyi Biotec, NL) to purify the schizonts. After elution the schizonts were diluted to the concentration of 10⁸/mL in RPMI medium, aliquoted, and frozen at −20°C. Next they were thawed and frozen at −20°C again to produce the extract.

L2 MVECs were seeded in 6-, 24-, 48-, or 96-well plates at the concentration of 5 × 10⁴ cells/mL. After expanding the cells for 24 h, they were washed with medium and stimulated for 24 h with combinations of murine IFN-γ (20 ng/mL, PeproTech, USA), parasite extract (10⁷ infected red blood cells—iRBCs/mL), murine red blood cells extract (10⁷ RBCs/mL), and dexamethasone (100 nM dissolved in DMSO, Sigma). JNK, p38, and ERK inhibitors (SP600125, SB203580, and FR180204, dissolved in DMSO, R&D, UK) were used at 20, 5, and 10 µM, respectively. After stimulation, plates were centrifuged (5 min, 1,200 rpm, RT) and supernatants were collected for ELISA and stored at −20°C. Cytokines were analyzed by ELISA (R&D). Cells were washed with PBS, lysed with RLT buffer with β-mercaptoethanol from RNeasy Mini kit (Qiagen, Belgium), and stored at −80°C for RNA extraction.

RNA was extracted with RNeasy Mini Kit (Qiagen) according to the manufacturer’s protocol. RNA concentration and purity were evaluated with Nanodrop 1000 (Thermo Scientific, Belgium). RNA (0.25 µg) was converted to cDNA using high-capacity cDNA reverse transcription kit (Applied Biosystems). Quantitative RT-PCR was performed with 6.25 or 0.125 ng cDNA using predesigned primers (IDT) and TaqMan Universal PCR Master Mix (Applied Biosystems), respectively. RNA-Seq expression profiling was performed by the Genomics Core UZ Leuven. Per independent experiment RNA from three technical replicates per experimental condition was pooled and 3 µg RNA/experimental condition were sequenced from a total of three independent experiments. Illumina TruSeq stranded mRNA kit was used and the single-end sequencing was performed. 33 M 50 bp reads per sample were sequenced. Reads were aligned to mm10 murine genome using TopHat. A heatmap with highly variable genes across the samples was plotted using pheatmap package (pheatmap). The rlog transformed counts of each gene were centered across the samples. Ribosomal RNA genes and predicted genes that were increased in one sample (GC027190) were excluded from the rlog data used for heatmap since this increase was a result of an imperfect poly-A selection during the library preparation of this sample. Differential expression analysis was performed with DESeq2 (36). Differences in gene expression with a FDR adjusted p value below 0.1 were considered significant. Gene ontology analysis of differentially expressed genes was performed with clusterProfiler package (37). Motif analysis of the proximal promoter region (400 bp upstream of the transcription start site till 100 bp downstream) was performed using Homer software (Homer motif analysis). RNA-seq data were submitted to ArrayExpress (accession number E-MTAB-5921 (ArrayExpress URL)).

For Western blot analysis, protein extracts from L2 MVECs stimulated with IFN-γ, PbNK65 extract, IFN-γ and PbNK65 extract in the presence or absence of dexamethasone were separated on SDS PAGE gels and blotted onto a PVDF membrane. Blocking was performed with BSA (Carl Roth Gmbh, Belgium) or non-fat dry milk (Bio Rad, USA). The following specific primary Abs were used: JNK, pJNK, p38, p-p38 (1:2,000, Cell Signaling Technology, The Netherlands), p-GR S211 (1:1,000, Cell Signaling Technology), p-GR S226 (1:1,000, Abcam), GR H300 (1:1,000, Santa Cruz, Germany), p-STAT1 Tyr701 (1:1,500, Cell Signaling Technology), and STAT1 (1:5,000, Cell Signaling Technology). Fusion solo S system (Vilber, France) was used to take chemiluminescence Western blot images. Quantification of Western blot images was performed by densitometry (ImageJ software was used).

Cells were seeded on coverslips and incubated in phenol-red-free and serum-free medium for 4 h. Cell fixation, methanol permeabilization, and staining were performed according to Cell Signaling guidelines. GR was visualized with the GR polyclonal (H300) antibody (Santa Cruz, Germany), used at 1:200, followed by probing with Alexa Fluor 488 (Invitrogen, Belgium). Nuclei were visualized using 4′,6-diamidino-2-phenylindole (DAPI) staining. A motorized inverted IX81 FluoView FV1000 laser scanning confocal microscope (Olympus) was used to record high-resolution images. Assessment of intracellular localization of protein signal was done double blind.

Statistical analysis was performed with R (The R Project for Statistical Computing). Mann–Whitney or ANOVA was used to analyze ELISA and qPCR data. Statistical significance was set at a p value below 0.05.

IFN-γ plays a crucial role in the induction of chemokines in mouse models of complicated malaria (38, 39). To evaluate whether IFN-γ might cause GC resistance in lung endothelial cells, we stimulated L2 MVECs with IFN-γ in the presence or absence of dexamethasone for 24 h. IFN-γ induced CC chemokine ligand 2 (CCL2, also known as monocyte chemoattractant protein 1 or MCP-1), CXC chemokine ligand 10 (CXCL10, also known as IFN-γ-induced protein 10 or IP-10), and CCL5 (also known as regulated on activation normal T cell expressed and secreted or RANTES) on both protein and RNA levels (Figure 1). CXCL10 (IP-10) showed the strongest induction. Dexamethasone inhibited IFN-γ-induced chemokine secretion, suppressing CCL5 (RANTES) by 90% and both CXCL10 (IP-10) and CCL2 (MCP-1) approximately by 50% (Figure 1A). Real-time qPCR experiments revealed 90% inhibition of CCL5 (RANTES) and 71% inhibition of both CXCL10 (IP-10) and CCL2 (MCP-1). Moreover, treatment with dexamethasone (with or without IFN-γ) induced DUSP-1 (MKP1), GILZ (TSC22d3), and FKBP51 (Figure S1 in Supplementary Material), indicating that IFN-γ did not alter GC-mediated transactivation of these genes. These results indicate that IFN-γ-stimulated lung endothelial cells remain GC sensitive.

To further characterize the transcriptional targets indicating GC sensitivity in lung endothelial cells, we analyzed by RNA-Seq the transcriptome of L2 MVECs stimulated with IFN-γ in the presence or absence of dexamethasone. IFN-γ induced expression of various guanylate binding proteins (GBPs, Figure 2A), which trigger antimicrobial effector mechanisms via inhibition of replication of intracellular pathogens (40). Similarly, we found several interferon-induced proteins with tetratricopeptide repeats (IFITs, Supplementary File 1). These proteins inhibit viral replication (41). IFN-γ also increased expression of H-2 class II histocompatibility genes and adhesion molecules such as vascular cell adhesion protein 1 (VCAM-1) and intercellular adhesion molecule 1 (ICAM-1, Supplementary Files 1 and 2). Moreover, IFN-γ potently induced transcription of various proinflammatory CXC and CCL chemokines including CXCL9 (also known as monokine induced by gamma interferon or MIG), CXCL11 (also known as Interferon-inducible T cell alpha chemoattractant or I-TAC), CXCL10 (IP-10), CCL5 (RANTES), and CCL2 (MCP-1, Figure 2A; Table 1; Supplementary Files 1). These data on chemokine induction further confirm the ELISA and qRT-PCR experiments shown in Figure 1.

Dexamethasone significantly inhibited large clusters of IFN-γ-induced genes, which included several CXC and CCL chemokines (Table 1; Supplementary Files 1 and 3). In particular, dexamethasone markedly diminished transcription of CCL5 (RANTES, more than 85% reduction), CXCL11 (I-TAC, more than 80% reduction), CCL4, and CCL3 (more than 75% reduction, Figure 2B). Also, IFN-γ-induced interleukin 1β (IL-1β), IL-27, and IFITs were downregulated by dexamethasone. In contrast, the expression of GBPs, signal transducer and activator of transcription (STAT) family members, and the cytokine IL-15 was resistant to GC-mediated transcriptional inhibition (Supplementary File 1).

In addition, we found a large cluster of genes upregulated by GCs (Supplementary File 1). Dexamethasone induced several anti-inflammatory genes in IFN-γ-stimulated endothelial cells including orosomucoid-1 (Orm1, Fold Change 11.3), DEP domain-containing mTOR-interacting protein (DEPTOR, Fold Change 5.3), and TSC22 domain family, member 3 (TSC22d3 also known as GILZ, Fold Change 4.6, Figure 2C)—a known GR target gene. These genes were also upregulated upon treatment with dexamethasone alone (Supplementary File 4). Interestingly, another set of genes was repressed by IFN-γ alone, but the expression was restored when dexamethasone was added. These genes include CXCL12, Angptl7, TGF-β2, TLR7, and TNFRSf21 (Supplementary File 1). Altogether, these data indicate that lung endothelial cells are GC sensitive, when stimulated with IFN-γ. Furthermore, these cells provide a novel model to study GR-dependent responses.

As for the affected pathways, gene ontology (GO) analysis of genes induced by IFN-γ at least 2-fold and inhibited by dexamethasone at least by 50% (Figure 2D; Supplementary File 5) revealed a high prevalence of biological process GO terms related to regulation of immune response and proinflammatory signaling pathways such as MAPK (Figure 2E). Several binding partners have been involved in GR-mediated repression of proinflammatory genes including AP-1, NF-κB, and IRF3 (42–44). Interestingly, a computational analysis of the 400-bp region upstream of genes induced by IFN-γ at least 2-fold and inhibited by dexamethasone at least by 50% showed a significant enrichment of the binding sites for IRF and NF-κB families but not for GR (Table S1 in Supplementary Material). These results remain in line with previous findings since in LPS-stimulated macrophages, less than 6% of GR binding sites occurred at proximal promoter regions (44). Also, GR can tether to DNA-bound TFs such as NF-κB without requiring a GRE motif.

Although GC resistance has been observed in our murine model of MA-ARDS (9), the above results indicate that IFN-γ-stimulated lung endothelial cells remain GC sensitive. As sequestering parasites release a variety of products which can further activate endothelial cells (6, 7), we investigated whether addition of parasite extract might alter the GC sensitivity of L2 MVECs. PbNK65 extract in combination with IFN-γ increased the mRNA levels of CCL2 (MCP-1) and CCL5 (RANTES), respectively, 2-fold and 3-fold in comparison to IFN-γ alone (Figure 3A). PbNK65 alone increased only CCL2 (MCP-1) protein levels and interestingly, this induction was sensitive to dexamethasone (Figure 3B). Importantly, dexamethasone failed to inhibit mRNA and protein induction of CCL2 (MCP-1), CCL5 (RANTES), and CXCL10 (IP-10), when cells were challenged with the combination of PbNK65 extract and IFN-γ (Figures 3A,B). These results indicate that PbNK65 extract impairs GC-mediated transrepression of these inflammatory chemokines observed after stimulation with IFN-γ alone. In contrast, PbNK65 extract did not decrease GC-induced transactivation of DUSP-1 (MKP1), GILZ (TSC22d3), and FKBP51 (Figure S1 in Supplementary Material).

To evaluate the time course of the development of GC resistance, we stimulated lung endothelial cells with PbNK65 extract and IFN-γ for 6, 24, and 48 h in the presence or absence of dexamethasone and analyzed CCL2 (MCP-1) and CXCL10 (IP-10) secretion. After 6 h, chemokine levels remained low both in the resistant condition with PbNK65 extract and IFN-γ and in the sensitive one with IFN-γ alone (Figure S2 in Supplementary Material). GC resistance for CCL2 (MCP-1) occurred following 24-h stimulation with PbNK65 extract and IFN-γ and was still present after 48 h (Figure S2A in Supplementary Material). PbNK65 extract and IFN-γ induced GC resistance for CXCL10 (IP-10) already after 6 h (Figure S2B in Supplementary Material). Notwithstanding these gene-specific differences in kinetics, all studied genes underwent GC resistance upon combining IFN-γ with PbNK56 extract. In contrast, upon stimulation with IFN-γ alone endothelial cells remained fully GC sensitive even after 48 h (Figures S2C,D in Supplementary Material).

As a control for the PbNK65 extract, we stimulated lung endothelial cells with extract from non-infected red blood cells (RBC) and evaluated GC sensitivity. RBC extract did not induce CCL2 (MCP-1) or CXCL10 (IP-10) in endothelial cells (Figure 4). When combined with IFN-γ, RBC extract failed to enhance secretion of proinflammatory chemokines. Furthermore, dexamethasone inhibited CCL2 (MCP-1) and CXCL10 (IP-10) production induced by IFN-γ in the presence of RBC extract, showing that RBC extract is not able to induce GC resistance. These results confirm that parasite components but not RBC components mediate GC resistance.

GR undergoes homologous downregulation when incubated with its ligand for longer periods of time. Perturbations of this process may be lead to GC resistance (45, 46). To evaluate the levels of GR, we stimulated L2 MVECs for 6 or 24 h with vehicle (DMSO), red blood cell extract (RBC), IFN-γ, PbNK65 extract, IFN-γ and PbNK65 extract in the presence or absence of dexamethasone. The levels of GR remained unchanged in the resistant condition with IFN-γ and PbNK65 extract when compared to the sensitive condition with IFN-γ alone. Additionally, the capacity and extent of GC-induced GR homologous downregulation remained unaffected in either the sensitive or resistant condition. We also investigated the serine-211 (S211) phosphorylation of GR, since this modification is associated with transcriptionally active GR and provides a means for cross-talk with other signaling pathways (47). As expected, dexamethasone alone induced S211 phosphorylation. Moreover, phosphorylation of GR at S211 remained present when cells were challenged with IFN-γ or IFN-γ with PbNK65 extract in the presence of dexamethasone (Figure 5A; Figure S3 in Supplementary Material). We also evaluated the GR S226 phosphorylation, which inhibits GR function (48, 49). Dexamethasone inhibited GR S226 phosphorylation upon stimulation with IFN-γ or IFN-γ with PbNK65 extract (Figure 5B). These data indicate that GC resistance following IFN-γ/PbNK65 cotreatment is most likely not caused by defective GR phosphorylation or GR homologous downregulation mechanisms.

Since impaired GR nuclear translocation leads to GC resistance (31, 50), we assessed the ability of dexamethasone to induce GR translocation in lung endothelial cells. L2 MVECs were challenged for 24 h with PbNK65 extract and IFN-γ and dexamethasone was added during the last hour of stimulation (Figure 6). Immunofluorescence microscopy revealed that in unstimulated cells GR localized mainly in the cytoplasm and translocated to the nucleus upon dexamethasone exposure. GR also translocated to the nucleus upon treatment with IFN-γ and dexamethasone. Furthermore, the translocation occurred in the GC resistant condition with PbNK65 extract and IFN-γ (Figure 6). We also obtained similar data for a shorter time point of 2 or with 24 h cotreatment with IFN-γ, PbNK65 extract and dexamethasone (Figures S4A,B in Supplementary Material). These data indicate that GC resistance is not caused by any defect in GR translocation.

Since IFN-γ signals via STAT1 to induce gene transcription, we evaluated the effect of dexamethasone on STAT1 activation in the GC sensitive condition with IFN-γ and in the resistant one with IFN-γ and PbNK65 extract. RNA-Seq data showed that dexamethasone failed to affect STAT1 expression induced by IFN-γ and this result was validated by qRT-PCR (Figure 7A). Since STAT1 phosphorylation at Tyr 701 controls STAT1 signaling, we subsequently addressed the impact of dexamethasone on phosphorylated STAT1. We showed that STAT1 phosphorylation induced by IFN-γ remained unaffected by dexamethasone both in the GC sensitive (IFN-γ) and the GC resistant condition (IFN-γ and PbNK65 extract, Figure 7B).

MAPK family members play an important role in the generation and fine-tuning of inflammatory responses and are known to be activated by IFN-γ receptor signaling (51, 52). To address the role of specific members of the MAPK family in the induction of proinflammatory chemokines in our model, we used the JNK-, p38-, and ERK-specific inhibitors: SP600125, SB203580, and FR180204, respectively. Treatment with JNK and p38 inhibitors significantly reduced proinflammatory response induced by IFN-γ in lung endothelial cells. Inhibition of p38 reduced CCL2 (MCP-1) levels by 28%, whereas inhibition of JNK blocked CXCL10 (IP-10), CCL2 (MCP-1), and CCL5 (RANTES) by 55, 59, and 83%, respectively (Figure 8A). Accordingly, JNK and p38 inhibition blocked chemokine induction in the GC resistant condition upon combining IFN-γ with PbNK65 extract. SB203580 inhibited CXCL10 (IP-10), CCL2 (MCP-1), and CCL5 (RANTES) levels by 44, 49, and 54%, whereas SP600125 reduced CXCL10 (IP-10), CCL2 (MCP-1), and CCL5 (RANTES) by 69, 79, and 94%, respectively (Figure 8B). Inhibition of ERK failed to reduce CXCL10 (IP-10) CCL2 (MCP-1) or CCL5 (RANTES) levels upon challenge with IFN-γ and PbNK65 extract (Figure S6 in Supplementary Material).

Since previous studies indicated that GCs inhibit MAPK family members phosphorylation (53, 54), we investigated the effects of dexamethasone on p38 and JNK phosphorylation in endothelial cells stimulated with IFN-γ with and without PbNK65 extract in the presence or absence of dexamethasone. We found that dexamethasone blocked p38 and JNK phosphorylation upon challenge with IFN-γ (Figures 9A,B). Quantification of the Western blot data revealed that, upon challenge with IFN-γ, dexamethasone reduced p38 and JNK phosphorylation on average by 69 and 61%. In unstimulated cells, dexamethasone also blocked p38 and JNK phosphorylation by 58 and 63%, respectively (Figures 9C,D). In contrast, when IFN-γ was combined with PbNK65 extract dexamethasone failed to inhibit p38 and JNK phosphorylation by more than 14 and 4% (Figures 9A–D). These data suggest that dexamethasone inhibits IFN-γ-mediated induction of CXCL10 (IP-10), CCL2 (MCP-1), and CCL5 (RANTES) at least in part by blocking the activation of p38 and JNK. Furthermore, this inhibitory mechanism is impaired in the GC resistant condition.