The cell-specific induction of CXC chemokine ligand 9 mediated by IFN-gamma in microglia of the central nervous system is determined by the myeloid transcription factor PU.1

Abstract

The IFN-gamma-inducible chemokines CXCL9 and CXCL10 are implicated in the pathogenesis of T cell-mediated immunity in the CNS. However, in various CNS immune pathologies the cellular localization of these chemokines differs, with CXCL9 produced by macrophage/microglia whereas CXCL10 is produced by both macrophage/microglia and astrocytes. In this study, we determined the mechanism for the microglial cell-restricted expression of the Cxcl9 gene induced by IFN-gamma. In cultured glial cells, the induction of the CXCL9 (in microglia) and CXCL10 (in microglia and astrocytes) mRNAs by IFN-gamma was not inhibited by cycloheximide. Of various transcription factors involved with IFN-gamma-mediated gene regulation, PU.1 was identified as a constitutively expressed NF in microglia but not in astrocytes. STAT1 and PU.1 bound constitutively to the Cxcl9 gene promoter in microglia, and this increased significantly following IFN-gamma treatment with IFN regulatory factor-8 identified as an additional late binding factor. However, in astrocytes, STAT1 alone bound to the Cxcl9 gene promoter. STAT1 was critical for IFN-gamma induction of both the Cxcl9 and Cxcl10 genes in microglia and in microglia and astrocytes, respectively. The small interfering RNA-mediated knockdown of PU.1 in microglia markedly impaired IFN-gamma-induced CXCL9 but not STAT1 or IFN regulatory factor-8. Cells of the D1A astrocyte line showed partial reprogramming to a myeloid-like phenotype posttransduction with PU.1 and, in addition to the expression of CD11b, acquired the ability to produce CXCL9 in response to IFN-gamma. Thus, PU.1 not only is crucial for the induction of CXCL9 by IFN-gamma in microglia but also is a key determinant factor for the cell-specific expression of this chemokine by these myeloid cells.

Figure 1. Regulation of CXCL9 and CXCL10 gene expression by IFN-γ in CNS glial cells

CXCL9 (A) and CXCL10 (B) levels in supernatants from mixed glial cells, astrocytes, primary microglia and EOC-13 microglial cells. Supernatants were collected from IFN-γ-treated (24 hours) or untreated (0 hours) (n=3 per timepoint) cells were analyzed by ELISA. Autoradiographic images showing CXCL9 and CXCL10 mRNA levels in primary mixed glial cells (C). Cells were cultured as described in the Materials and Methods and treated either with medium alone (0 hour) or with medium containing recombinant IFN-γ (1000U/ml) for 4 hours or 24 hours with or without cycloheximide (CHX) 20μg/ml). In CHX treated groups, samples were pretreated for 30 min before treatment with IFN-γ. Chemokine mRNAs were detected by RPA analysis of total RNA (3 μg per sample) using a multiprobe set as described in the Materials and Methods. Quantification of CXCL9 (D) and CXCL10 (E) mRNA levels was performed by densitometry as described in the Materials and Methods. For statistical significance: *p<0.05, **p<0.01, ***p<0.001.

Figure 2. Expression of selected transcription factors in astrocytes and microglia

All cells were cultured as described in the Materials and Methods and treated either with medium alone (0 hour) or medium containing recombinant IFN-γ (1000U/ml) at the time points shown. Total RNA was isolated from purified primary astrocytes, primary microglia and EOC-13 microglial cells and 3 μg was used for analysis by RPA (A) as described in the Materials and Methods section. The relative mRNA levels were quantified by densitometry and normalized to the L-32 loading control (B, C, D). PU.1 gene expression was detected only in microglial cells and not in astrocytes. Selected transcription factors were also examined at the protein level by immunoblot analysis (E) as described in Materials and Methods. Transcription factor levels in cultured glial cells were quantified by densitometry and normalized to the β-tubulin loading control (F, G, H). For statistical significance: *p<0.05, **p<0.01, ***p<0.001.

Figure 3

Localization of PU.1 and STAT1 in glial cells. The PU.1 and STAT1 proteins were analyzed by immunofluorescence staining of mixed glial cells (A–H) treated either with medium alone (0 hour) or medium containing recombinant IFN-γ (1000U/ml) for 1 hour as described in Materials and Methods. PU.1 (green; A,B,E,F, arrows) colocalized with F4/80+ (red; E,F) cells and not GFAP+ (red; A,B) cells. PU.1 was localized in the nucleus independent of IFN-γ treatment (E, F). In contrast, STAT1 (green; C,D,G,H) was found in the cytoplasm of GFAP+ (red; C) and F4/80+ (red; G) cells cultured in medium alone but was found in the nucleus of GFAP+ (D; arrow) and F4/80+ (H; arrow) cells following IFN-γ treatment. Dual-label immunohistochemical staining on brain sections from control mice (I,J) or mice at peak EAE (K,L) was performed as described in the Materials and Methods. PU.1 (purple) co-localised with lectin+ (red) microglia/macrophages (I,K; arrows). However, PU.1 (purple; arrowheads) did not co-localize with GFAP+ astrocytes (red; arrows) (J, L). Asterisks denote blood vessels (K,L). Original magnification of panels I and J: 1000x and K and L: 400x.

Figure 4

Interaction of STAT1, PU.1 and IRF-8 with the Cxcl9 gene promoter in microglia and astrocytes. A schematic illustration of the murine Cxcl9 gene promoter (A) showing the location of the primer sites for ChIP analysis (black bars) and the sites corresponding to the oligonucleotide probe used for EMSA (open bars). ChIP analysis was performed on EOC microglia (B) or astrocytes (C). Cells were incubated in the absence or presence of IFN-γ (1000U/ml) for 0, 4 or 24 hours, and cross-linked with formaldehyde and soluble chromatin was subjected to immunoprecipitation with antibodies against STAT1, PU.1, IRF-8 or normal IgG as described in the Materials and Methods. Representative PCR images taken from 1 of 3 independent experiments is shown for EOC-13 cells (A) and astrocytes (B). Fold-change as compared with IgG was quantified for each transcription factor binding for EOC13 cells (D) and astrocytes (E). For statistical significance: *p<0.05, **p<0.01, ***p<0.001. Binding activity to γ-RE (F) or EIRE1 (G) radiolabeled oligonucleotides with nuclear extracts (5 μg) from EOC-13 cells treated with IFN-γ for the times shown and incubated with the indicated antibodies was performed as described in Materials and Methods.

Figure 5

Essential role for STAT1 in the induction of CXCL9 and CXCL10 by microglia and astrocytes in response to IFN-γ. Autoradiographic images showing CXCL9 and CXCL10 RNA levels in primary microglia (A) or primary astrocytes (B) from Wt or Stat1^−/− mice. Cells were cultured as described in the Materials and Methods and treated with or without IFN-γ (1000 U/ml) for the times shown and total RNA (3 μg per sample) analyzed by RPA using a multiprobe set as described in the Materials and Methods. The relative mRNA levels were quantified by densitometry and normalized to the L32 loading control for microglia (C) and astrocytes (D). Cultured WT or Stat1^−/− mixed glial cells were treated with or without IFN-γ (1000 U/ml) and the supernatants collected and analyzed by ELISA for CXCL9 (E) and CXCL10 (F) as described in the Material and Methods. For statistical significance: *p < 0.05, ***p<0.001.

Figure 6

siRNA mediated knockdown of PU.1 in EOC-13 cells and its impact on IFN-induced-CXCL9 production. siRNA knockdown was performed as described in the Materials and Methods. Non-transfected, transfection medium (Dharmafect®) alone, control (CTRL) siRNA transfected and PU.1 siRNA transfected EOC-13 cells were treated with medium alone or IFN-γ (1000 U/ml) for 24 hours. Total RNA was extracted and 3 ug RNA was subjected to RPA analysis as described in the Materials and Methods (A). Quantification of STAT1, IRF-8 and PU.1 mRNA levels was performed by densitometry and normalized to L32 loading control (B, C, D). Protein lysates were prepared and subjected to immunoblotting with anti-PU.1, anti-STAT1, anti-IRF-8 and anti-GAPDH antibodies (E). Quantification of STAT1, IRF-8 and PU.1 protein levels was performed by densitometry and normalized to GAPDH (F, G, H). Quantification of CXCL9 mRNA levels was performed by densitometry and normalized to the L32 loading control (I). Supernatants from PU.1 siRNA transfected, CTRL siRNA transfected or Dharmafect® transfected controls treated with or without IFN-γ treated were analyzed by ELISA for CXCL9 protein levels (J). For statistical significance: *p < 0.05, **p<0.01, ***p<0.001.

Figure 7

Properties of PU.1 transduced C8-D1A astrocytic cells. Cells of the C8-D1A astrocyte cell line were transduced with the pHAGE lentiviral vector containing GFP alone (empty vector control) or GFP plus PU.1 as described in the Materials and Methods. Flow cytometry was used to collect and enrich for GFP-positive cells which were then analyzed for CD11b expression compared with EOC-13 microglial cells as a positive control (A). Following treatment with or without IFN-γ, cells were either lysed and the PU.1 protein was analyzed by immunoblotting (B) or total RNA was prepared and analyzed by RPA (C) as described in the Materials and Methods. The relative level of PU.1 (D), STAT1 (E), CXCL9 (F) or CXCL10 (G) mRNA was quantified by densitometry and normalized to the L32 loading control. The production of CXCL9 (H) and CXCL10 (I) by the different cell types in response to IFN-γ was determined by ELISA. For statistical significance: *p < 0.05, **p<0.01, ***p<0.001.