An anti-inflammatory role for C/EBPδ in human brain pericytes

Neuroinflammation contributes to the pathogenesis of several neurological disorders and pericytes are implicated in brain inflammatory processes. Cellular inflammatory responses are orchestrated by transcription factors but information on transcriptional control in pericytes is lacking. Because the transcription factor CCAAT/enhancer binding protein delta (C/EBPδ) is induced in a number of inflammatory brain disorders, we sought to investigate its role in regulating pericyte immune responses. Our results reveal that C/EBPδ is induced in a concentration- and time-dependent fashion in human brain pericytes by interleukin-1β (IL-1β). To investigate the function of the induced C/EBPδ in pericytes we used siRNA to knockdown IL-1β-induced C/EBPδ expression. C/EBPδ knockdown enhanced IL-1β-induced production of intracellular adhesion molecule-1 (ICAM-1), interleukin-8, monocyte chemoattractant protein-1 (MCP-1) and IL-1β, whilst attenuating cyclooxygenase-2 and superoxide dismutase-2 gene expression. Altered ICAM-1 and MCP-1 protein expression were confirmed by cytometric bead array and immunocytochemistry. Our results show that knock-down of C/EBPδ expression in pericytes following immune stimulation increased chemokine and adhesion molecule expression, thus modifying the human brain pericyte inflammatory response. The induction of C/EBPδ following immune stimulation may act to limit infiltration of peripheral immune cells, thereby preventing further inflammatory responses in the brain.


Results
Characterisation of adult human brain pericyte cultures. Immunocytochemical analysis of early passage cell cultures obtained from human middle temporal gyrus tissue reveals a mixed population of astrocytes, microglia and pericytes as described previously 10 . Under our in vitro conditions microglia and astrocytes do not proliferate and are diluted out in subsequent passages. To ensure no contamination from brain glia in pericyte cultures these were grown until passage five before use. Late passage cultures showed positive immunocytochemical staining for the pericyte markers alpha smooth muscle actin (α SMA; Fig. 1a), platelet derived growth factor receptor beta (PDGFRβ ; Fig. 1b) and neural/ glial antigen 2(NG2; Fig. 1c) as well as the fibroblast markers prolyl-4-hydroxylase (P4H; Fig. 1d) and fibronectin (Fig. 1e). There were no cells positive for the microglia marker CD45 or the astrocyte marker glial fibrillary acidic protein (GFAP; Fig. 1f) at passage five onwards. Positive controls of GFAP (Fig. 1g) and CD45 (Fig. 1h) staining at passage two are shown. C/EBPδ is induced in human brain pericytes by IL-1β/IFNγ. Microarray analyis of human brain pericytes treated with IL-1β /IFNγ has previously revealed an induction of numerous inflammatory genes 10 . Due to the ability of the C/EBP family of transcription factors to modify cellular inflammatory responses, we chose to investigate the induction of three members C/EBPα , C/EBPβ and C/EBPδ in this dataset. By microarray analysis, C/EBPδ was found to be significantly increased by an IL-1β /IFNγ treatment (p < 0.001) whilst C/EBPα (p > 0.05) and C/EBPβ (p < 0.05) were not affected (Fig. 2a). Real time quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) was performed on independent RNA samples, which confirmed the induction of C/EBPδ (5.33 ± 0.69 fold; p < 0.001), whereas no significant change in C/EBPα (0.74 ± 0.18 fold; p > 0.05) or C/EBPβ (1.18 ± 0.15 fold; p > 0.05) was observed (Fig. 2b), consistent with the microarray data.
Time course and concentration-dependant induction of C/EBPδ expression. As IL-1β treatment resulted in the greatest induction of C/EBPδ all subsequent experiments were performed using this inflammatory cytokine. To further understand the profile of C/EBPδ expression in human brain pericytes, a concentration-response curve of IL-1β induced expression was performed. Using immunocytochemistry, C/EBPδ was found to be induced by IL-1β in a concentration-dependant manner (Fig. 4a,b). A significant induction was observed by concentrations as low as 0.1 ng/mL IL-1β (vehicle 13.07 ± 1.20%, IL-1β 26.18 ± 1.16%; p < 0.001) with maximal induction by 10 ng/mL (53.73 ± 2.49%; p < 0.001). Western blotting revealed a significant induction of C/EBPδ using 1 ng/mL (10.12 ± 3.06 fold; p < 0.05) and 10 ng/mL (8.57 ± 0.99 fold; p < 0.05) IL-1β (Fig. 4c,d). In order to investigate the temporal profile of C/EBPδ induction a time-course was performed with 10 ng/mL IL-1β . By immunocytochemistry C/EBPδ was induced as early as two hours after IL-1β stimulation (vehicle 13.95 ± 0.85%, IL-1β 23.03 ± 1.71%; p < 0.05), maximally induced four hours after treatment (72.83 ± 3.11%; p < 0.001) and remained elevated 48 hours later (34.43% ± 1.45%; p < 0.001; Fig. 4e). Western blotting analysis revealed a similar trend with significant induction of C/EBPδ at two (23.42 ± 5.67 fold; p < 0.05), four  . C/EBPδ is induced in human brain pericytes by IL-1β/IFNγ. Human brain pericytes were treated with vehicle or 10 ng/mL IL-1β + IFNγ for 24 hours and RNA was extracted. Expression of C/ EBPα , C/EBPβ and C/EBPδ was determined by microarray 10 (a) and qRT-PCR (b). Data is displayed as average fold change of five independent cases (a) or mean ± SEM of a separate three independent cases (b). *** = p < 0.001.
(27.32 ± 10.55 fold; p < 0.05) and 24 hour treatments (12.45 ± 3.24 fold; p < 0.05) with IL-1β . However, it was not significantly elevated at 48 hours (9.43 + 1.85 fold; p > 0.05; Fig. 4f,g). . C/EBPδ is differentially induced by IFNγ, IL-1β and LPS. Human brain pericytes were treated with vehicle or 10 ng/mL IL-1β , IFNγ , or LPS for 24 hours. Representative immunocytochemistry images of C/EBPδ with treatments are shown (a). The percentage of cells positive for nuclear C/EBPδ was determined by immunocytochemistry (b) and the intensity of C/EBPδ expression was analysed by western blotting (c,d). Blots are cropped to improve clarity. Full-length blots are presented in Supplementary Figure S2. Data is displayed as mean ± SEM of three independent experiments. * = p < 0.05 compared to vehicle control, ** = p < 0.01 compared to vehicle control *** = p < 0.001 compared to vehicle control. Scale bar = 100 μ m.  Figure S2. Human brain pericytes were treated with 10 ng/mL IL-1β for 0-48 hours. The percentage of C/EBPδ positive cells was determined by immunocytochemistry (e) and C/EBPδ intensity was analysed by western blotting (f,g). Data is displayed as mean ± SEM from three independent experiments. * = p < 0.05 compared to vehicle control, ** = p < 0.01 compared to vehicle control, *** = p < 0.001 compared to vehicle control. Scale bar = 100 μ m.

Discussion
C/EBPδ was found to be induced in human brain pericyte cultures by the pro-inflammatory cytokine IL-1β . Several studies have reported induction of C/EBPδ expression in microglia and astrocytes following inflammation 25,26,29 , however, C/EBPδ expression in pericytes has not been previously shown. Using C/EBPδ specific siRNA we were able to attenuate the induction of pericyte C/EBPδ following immune challenge in order to identify its contribution to the inflammatory response. Consistent with literature from rodent microglia and astrocytes, C/EBPδ knock-down attenuated COX-2 expression; however it did not significantly alter IL-6 expression 25,26 . COX-2 is an inducible enzyme that catalyses prostaglandin formation, many of which have roles in inflammation. Whilst typically thought to have a pro-inflammatory response, anti-inflammatory roles of prostaglandins are now being acknowledged 36 . Prostaglandin I2 (PGI2) is the most prominent prostaglandin synthesised by cells of the vasculature including endothelial cells and vascular smooth muscle cells and is predominantly synthesised via COX-2 37,38 . Furthermore, PGI2 is the major prostaglandin product from retinal pericyte cells 39 . PGI2 inhibits leukocyte adhesion to vascular endothelium and therefore may attenuate immune infiltration 40 . In order to understand the specific inflammatory role of COX-2 induction in pericytes the composition of prostaglandins needs to be studied in context of the system. Similarly, a dual role in both anti-inflammatory and pro-inflammatory responses has been observed for IL-6 with evidence supporting a neuroprotective role in the brain 41,42 . Care should be taken when classifying these as either pro or anti-inflammatory due to their context dependant nature.
Attenuation of IL-1β induced SOD-2 expression was also observed with C/EBPδ knock-down. Induction of SOD-2 by a range of inflammatory factors has previously been shown, whilst its promoter region contains a C/EBP binding site 43,44 . SOD-2 regulates oxidative stress in cells by converting superoxide into hydrogen peroxide. Superoxide ions enhance recruitment of immune cells, precipitate DNA damage and were found to augment behavioural deficits in a mouse model of Alzheimer's disease [45][46][47] . Furthermore, oxidative stress is believed to be a major contributor to neuronal damage in inflammatory conditions and elimination of superoxide formation by SOD-2 is often considered to be anti-inflammatory 48 . The finding that C/EBPδ enhances SOD-2 induction may indicate an anti-oxidant/ anti-inflammatory role for this transcription factor following inflammation. Interestingly, C/EBPδ knock-down was also found to enhance the expression of several IL-1β induced inflammatory genes including IL-8, ICAM-1, MCP-1 and IL-1β . Two of which, MCP-1 and ICAM-1, were confirmed to be upregulated by immunocytochemical staining, as well as secretions into culture media. MCP-1 is a major chemokine involved in the recruitment of monocytes. In the context of CNS immune responses, secretions of MCP-1 from brain pericytes could be involved in the trafficking of monocytes from blood into the brain. 49,50 . Several parenchymal brain cells produce MCP-1 under inflammatory conditions, including microglia, astrocytes and pericytes 10 . Peripheral immune cell recruitment to the brain may further enhance central inflammatory responses through potentiated inflammatory cytokine production. Furthermore, MCP-1 increases both local microglial proliferation, as well as migration to injured sites 19 . As the brain's predominant immune cell, increased microglial presence and activation may further worsen inflammatory responses, contributing to neuronal death. Indeed, enhanced MCP-1 expression has been observed in several neurological disorders and may worsen disease progression [51][52][53] . Following inflammatory responses, the finding that C/EBPδ induction acts to dampen MCP-1 expression in brain pericytes suggests potential for reducing peripheral immune cell infiltration. Interestingly, the only study currently proposing an anti-inflammatory role of C/EBPδ utilised pancreatic beta cells to show that C/EBPδ knock-down enhanced IL-β /IFNγ induced chemokine expression, corroborating the changes we see in MCP-1 expression 33 . Investigations into the effect of C/EBPδ on brain glia chemokine expression are currently lacking.
ICAM-1 is a transmembrane protein involved in stabilising cell-cell interactions 15 . It is a ligand for the CD11 family of leukocyte adhesion molecules, which are widely expressed and inducible on leukocytes and this interaction aids their attachment to brain vasculature and subsequent migration into tissue 54 . Under basal conditions ICAM-1 expression in unstimulated pericytes is low; however it was found to be significantly elevated by the pro-inflammatory cytokine IL-1β . Expression of ICAM-1 on pericytes has previously been shown to aid peripheral immune cell infiltration across tissue vasculature, potentially worsening CNS inflammatory responses 8,16,54 . The finding that C/EBPδ knockdown enhances ICAM-1 expression on brain pericytes suggests that C/EBPδ induction during immune challenge may be beneficial in preventing ICAM-1mediated immune cell infiltration, highlighting another anti-inflammatory role for this transcription factor. Interestingly, C/EBPβ knockdown in human astrocytes was also found to enhance IL-1β induced ICAM-1 expression suggesting that this protein is largely regulated by C/EBP family members 24 .
C/EBPδ knockdown also enhanced IL-1β induced IL-8 and IL-1β gene expression. IL-8 is a potent neutrophil chemokine that has previously been observed in brain pericytes and involved in enhanced CNS immune cell infiltration 8,55 . Like MCP-1 and ICAM-1, it represents another mechanism whereby pericytes can enhance central infiltration of peripheral immune cells. Interestingly, the change in IL-8 mRNA expression with C/EBPδ knockdown did not translate to a significant change in protein secretion. The reasons for this are currently unclear. IL-8 was the most concentrated of all secreted proteins measured in culture media and perhaps saturation of secretory pathways was observed at the measured time-point. C/EBPδ regulation of IL-8 therefore warrants further investigation. Due to the use of IL-1β to stimulate pericytes, its expression in culture media was unable to be measured. IL-1β is a potent inflammatory stimulus, and therefore its induction following C/EBPδ knockdown would appear to contradict several of the anti-inflammatory roles in pericytes. However, pericyte stimulation with other inflammatory stimuli (IFNγ , LPS, TNFα and IL-6) revealed an increase in IL-1β mRNA, which did not translate to IL-1β secretion (Fig. S1). It is possible then that the observed IL-1β gene change with C/EBPδ knockdown would not necessarily cause increased IL-1β secretion. Indeed no evidence for pericyte secretion of IL-1β exists from studies examining their secretome 7,8 .
In conclusion, we propose that C/EBPδ has a role in modifying inflammatory responses of human brain pericytes. Contrary to previous data involving brain glia, we have demonstrated an anti-inflammatory role for C/EBPδ induction in the CNS. This is achieved through limiting pericyte MCP-1 and ICAM-1 expression, which in turn may attenuate peripheral immune cell infiltration into the CNS. Of course whether these changes induced by C/EBPδ knock-down would actually lead to altered inflammatory events in vivo is not clear as currently we don't know whether the concentrations of cytokines/chemokines measured in vitro are biologically relevant. Despite this, the involvement of pericytes in neuroinflammation is becoming increasingly recognised and further studies investigating opportunities to control this response are warranted.

Methods
Tissue source. Human middle temporal gyrus brain tissue was obtained, with informed consent, from surgeries of patients with drug-resistant temporal lobe epilepsy. All specimens were collected with written patient consent. All protocols used in this study were approved by the Northern Regional Ethics Committee (New Zealand) and all methods were carried out in accordance with the approved guidelines.
Mixed glial cultures isolated from human brain tissue. Mixed glial cultures containing microglia, astrocytes and pericytes were isolated from adult human brain tissue as per 56 with minor modifications. Approximately 2 g of tissue from the middle temporal gyrus was washed in Hanks balanced salt solution (HBSS; Ca 2+ and Mg 2+ free, Gibco BRL, CA, USA), meninges and visible blood vessels removed and tissue diced into pieces approximately 1 mm 3 . Tissue was added to 10 mL g −1 enzyme dissociation mix (10 U mL −1 DNase (Invitrogen, CA, USA) and 2.5 U mL −1 papain (Worthington, NJ, USA) in Hibernate-A medium (Gibco, CA, USA) for 15 minutes at 37 °C with gentle rotation, triturated to aid digestion and incubated for a further 15 minutes. The tissue was further triturated, allowed to settle and the supernatant transferred to a new tube containing equal volume Dulbecco's modified Eagle medium: Nutrient mixture F-12 (DMEM/F12) (Gibco, CA, USA) supplemented with 10% foetal bovine serum (FBS; Gibco, CA, USA) and 1% Penicillin-Streptomycin-Glutamine (PSG; Gibco, CA, USA) henceforth referred to as DMEM/F12. The cell suspension was passed through a 100 μ m nylon cell strainer (Becton Dickenson, NJ, USA), centrifuged at 1000 RPM for 10 minutes and the pellet resuspended in DMEM/F12. 20 mL of the single cell suspension was added to T75 tissue culture flasks (Nunc, Roskilde, Denmark) and incubated overnight at 37 °C with 95% air / 5%CO 2 . The following day, media from flask containing unattached cells and debris was removed and centrifuged at 1000 RPM for 10 minutes. The resulting pellet was resuspended in DMEM/F12 and added back to the flask. On day three, media containing unattached cells or debris was removed and discarded. Cells were grown until confluent (1-2 weeks) at which point they were harvested with 0.25% Trypsin-1mM Ethylenediaminetetraacetic acid (EDTA) (Gibco, CA, USA) and gentle scraping. Cells were passaged up to passage eight with early passages (two-three) containing microglia, astrocytes and pericytes, whilst latter passages (four-eight) contained only pericytes. Detailed characterisation of both early and late passage cultures has been performed previously 10 . Briefly, late passage cultures were positive for pericyte markers α -smooth muscle actin, platelet derived growth factor receptor-β and neural/glial antigen-2 and showed no expression of the astroglial marker glial fibrillary acidic protein (GFAP) or microglial markers PU.1 and CD45. However, although we have used a number of markers, none of these are specific to pericytes so it is possible that our cultures contain other fibroblast-like cells. For simplicity, however, we will refer to late passage cultures as pericytes. We have also quantified the relative proportions of microglia, astrocytes, and pericytes in our cultures at different passages. In general (there is some variation between different donors), we find ~10% CD45-positive microglia, ~5% GFAP-positive astrocytes, and 85% pericytes in passage two cultures. Passage five and above cultures contain 100% pericytes and thus were used for all experiments. Collection of conditioned media and cytokine measurement by cytometric bead array. Conditioned media was collected from cells grown in a 96 well plate. Media was spun at 16 x g for five minutes to collect possible cells and debris. Supernatant was obtained and stored -20°C. The concentration of cytokines was measured using cytometric bead array (CBA)(BD Biosciences, CA, USA) as described previously 57 . CBA samples were run on an Accuri C6 flow-cytometer (BD Biosciences, CA, USA). Data was analysed using FCAP-array software (version 3.1) (BD Biosciences, CA, USA) to convert fluorescent intensity values to concentrations.
Immunocytochemistry. Cells were fixed in 4% paraformaldehyde for 15 minutes and washed in phosphate buffered saline with 0.1% triton X-100 (PBS-T). Cells were incubated with primary antibodies (Table S1) overnight at 4 °C in immunobuffer containing 1% goat serum, 0.2% triton X-100 and 0.04% thimerosal in PBS. Cells were washed in PBS-T and incubated with appropriate anti-species fluorescently conjugated secondary antibodies overnight at 4 °C. Cells were washed again and incubated with Hoechst stain for 20 minutes. Images were acquired using the Discovery-1 automated fluorescence microscope (Molecular Devices) or ImageXpress ® Micro XLS (Version 5. Western blotting and densitometry analysis. Following treatments, cells were harvested by rinsing in ice-cold PBS, scraping directly into sample harvesting buffer (Laemmli buffer 58 without bromophenol blue or reducing agents) and boiling for 10 minutes. Protein concentrations were quantified using the BioRad DC Protein Assay (BioRad, Hercules, CA, USA). Samples were diluted to give 20 or 30 μ g protein per lane in sample harvesting buffer with final 100 mM DTT and 0.02% bromophenol blue and boiled for a further 10 minutes. Samples were then resolved by electrophoresis on 4-12% NuPAGE Bis-Tris SDS-PAGE gels (Life Technologies). Gels were transferred onto Immobilon-FL PVDF membrane (Millipore) using the X-Cell II blot module (Life Technologies). Blots were blocked in 5% non-fat dried milk in Tris-buffered saline with 0.1% tween (TBS-T) for 30 min. Blots were probed overnight at 4 °C with primary antibodies, then for two hours at room temperature with secondary antibodies (all in 50% Odyssey blocking buffer in TBS-T, see Table S1 for antibody details). Blots were scanned on the Li-Cor Odyssey Fc Imager (Li-Cor Biotechnology, Lincoln, NE, USA). Blot images in TIF format were quantified using the gel analyser tool in ImageJ (version 1.49d, NIH, Bethesda, USA, http://rsb.info.nih.gov/ij/). Integrated band intensities were normalised to band intensities of GAPDH loading controls.
Real time quantitative reverse transcriptase PCR. Cells were washed in PBS and RNA extraction and purification was performed using the RNeasy ® mini kit (Qiagen, Limberg Netherlands) as per manufacturer's instructions. RNA was treated with DNase (1 μ g DNase/1 μ g RNA) using the RQ1 RNase-free DNAse kit (Promega, WI, USA) and cDNA made using the Superscript ® First-Strand Synthesis kit (Life Technologies, CA, USA). Quantitative real-time reverse transcriptase polymerase chain reaction was performed using Platinum ® SYBR ® Green qPCR SuperMix-UDG with Rox (Life Technologies, CA, USA).
Standard curves were run for all primers used and efficiencies were all 100 ± 10% (Table S2). Relative gene expression analysis was performed using the Δ Δ C t method to the housekeeping gene GAPDH.

Microarray.
Microarray experiment was performed as described in 10 . Briefly, cells destined for microarray experiment where treated for 24 hours with vehicle (0.1% BSA in PBS) or 10 ng/mL IL-1β / IFNγ . Cells were washed in PBS and RNA extracted as above for qRT-qPCR experiments. RNA quality was analysed using the Experion ™ system (Bio-Rad, CA, USA) and the Experion ™ RNA StdSens Analysis kit. RNA was labelled and hybridised to Affymetirx Genechip ® Primeview ™ Human Gene Expression Arrays (Santa Clara, CA, USA) according to manufacturer's instructions.
Statistical analysis. Unless otherwise stated, all experiments were performed at least three independent times from three different tissue donors. Statistical analysis was carried out using one-way analysis of variance (ANOVA) followed by Dunnett's multiple comparison test to compare treatments versus vehicle control (Graphpad Prism 5.02). For siRNA experiments a Two-way ANOVA with Bonferroni post-test was used to compare between treatments. For statistical analysis of qPCR data Δ Ct values were used. Statistical analysis of microarray experiment was performed as previously described 10 .