Background
Persistent demyelination often follows recurrent inflammation in multiple sclerosis (MS), even though oligodendroglial progenitor cells (OPCs) are present in the adult CNS as a potential source of oligodendrocytes for remyelination after loss of myelin [
1‐
3]. As a pathological mechanism underlying this remyelination failure, accumulating evidence indicates that interferon-γ (IFNγ), the only type II IFN secreted into the lesions by infiltrating T helper 1 (T
H1) cells and natural killer cells, induces cytotoxic effects on OPCs, and inhibits their differentiation, leading to failure in
de novo myelination by OPCs [
4‐
9]. We also demonstrated in our previous study that actively proliferating OPCs are far more susceptible to cytotoxic effects of IFNγ than are post-mitotic mature myelinating oligodendrocytes [
10]. In contrast to IFNγ, interferon-β (IFNβ), a type I IFN, is used successfully to reduce relapse rates in relapsing remitting MS [
11]. However, though IFNβ has minimal adverse effects on proliferation, migration and differentiation of oligodendrocytes
in vitro [
12,
13], it does inhibit remyelination after cuprizone-induced demyelination
in vivo [
14]. Given the extensive overlap in type I and type II IFN signaling pathways, our goal in the present study was to determine what molecular mechanisms are responsible for the much greater OPC toxicity of IFNγ than IFNβ.
The janus kinase (JAK)/signal transducer and activator of transcription (STAT) pathway has been well-studied as a principal intracellular signaling pathway activated by IFNs (Reviewed in [
15]). Binding of IFNγ to the type II IFN receptor results in the rapid autophosphorylation and activation of the receptor-associated JAK1 and JAK2, which in turn activates cytoplasmic STAT1 by phosphorylation at Tyr 701. Activated STAT1 proteins form homodimers, which translocate into the nucleus, and initiate transcription by binding to a specific motif, known as the IFNγ-activated site (GAS), in the promoters of various IFN-stimulated genes (ISGs). IFNβ utilizes another receptor, the type I IFN receptor, associated with JAK1 and tyrosine kinase 2 (TYK2), and regulates formation of the heterotrimeric transcription complex, interferon-stimulated gene factor 3 (ISGF3), composed of the activated forms of STAT1 and STAT2, and IRF9/ISGF3γ. ISGF3 recognizes the IFN-stimulated response element (ISRE) which is distinct from the GAS, and activates transcription of another set of ISGs. Although there is a substantial overlap between IFNβ-inducible and IFNγ-inducible ISGs stemming from their common dependence on activation of STAT1 [
16], we hypothesized that IFNγ-mediated cytotoxic effects on OPCs are attributable to ISGs which are differently induced in OPCs by IFNγ than by IFNβ.
As an example of the ISGs which are differently induced between IFNγ and IFNβ, we previously examined IFN-mediated transcriptional induction of major histocompatibility complex class II (MHC-II) molecules in the oligodendroglial lineage [
17]. Surface expression of MHC-II becomes detectable in OPCs after treatment with IFNγ, whereas IFNβ fails to induce expression of MHC-II. Our results indicated that the distinct difference in transcriptional activation of interferon regulatory factor 1 (IRF1) between IFNγ and IFNβ is attributed to the difference in subsequent MHC-II expression. Thus, IRF1 is also a promising example of the ISGs responsible for IFNγ-mediated cytotoxic effects in OPCs. In agreement with this idea, involvement of IRF1 in IFNγ-induced OPC apoptosis has recently been reported [
18].
IRF1 was originally isolated as a transcriptional activator of the IFNβ gene in response to viral infection [
19,
20]. IRF1 and eight other subsequently identified factors share a highly-conserved amino-terminal DNA binding domain (DBD) with five conserved tryptophan repeats, and thereby constitute a family of transcription factors, termed the IRF family (Reviewed in [
21‐
24]). The DBD forms a helix-turn-helix domain and recognizes similar, if not identical, DNA motifs containing the consensus IRF recognition sequence, 5'-AANNGAAA-3' [
25]. IRF2, which shares the highly homologous DBD with IRF1, is considered a transcriptional repressor for IRF1-mediated transcription by competing for the same
cis elements [
19,
26,
27]. In addition, most of the members, except IRF1 and IRF2, have an IRF association domain (IAD) at the C-terminal region, through which they interact with other members or other transcription factors. Despite the possible functional overlap and interplay among members of the IRF protein family, however, there have been only a few studies on the IRF family members in the oligodendroglial lineage [
28], particularly with respect to their roles in IFNγ-mediated and IFNβ-mediated signaling [
17]. In this study, using primary cultures of highly pure OPCs from rats, we performed a comprehensive analysis of all members of the IRF family in OPCs in response to IFNγ and IFNβ, and examined the synergistic roles for IRF1 and IRF8 (also known as interferon consensus sequence binding protein (ICSBP)), in IFNγ-induced OPC apoptosis.
Methods
Reagents and chemicals
All reagents and culture media used in this study were purchased from SIGMA (St. Louis, MO, USA) and Invitrogen (Carlsbad, CA, USA), respectively, except for the following products: Human recombinant fibroblast growth factor 2 and platelet-derived growth factor A homodimer were from R&D systems (Minneapolis, MN, USA); rabbit anti-IRF1 and anti-IRF2 antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA, USA), and mouse anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibodies were from Chemicon (Temecula, CA, USA). Rabbit anti-IRF8 antibody was produced by Ozato's laboratory. Small interfering RNA (siRNA) for IRF2 (siRNA ID: s220597), IRF8 (siRNA ID: s146232), and Negative control siRNA were from Ambion (Austin, TX, USA).
Mixed glial culture
Primary mixed glial cultures from rats were prepared as reported previously [
29]. Briefly, whole brains were dissected from 0 to 2-day-old Lewis rats, and submerged in ice-cold Leibovitz's L-15 medium. Under a dissecting microscope, olfactory bulbs, cerebral cortices and hindbrains were removed. After cleaning off meninges and vessels including choroidal plexus, the remaining brain tissues were cut into small chunks with a 21-gauge needle, and digested by 0.0625% (w/v) trypsin in Ca
2+ and Mg
2+-free Hank's Balanced Salt Solution (HBSS) for 20 min. Dissociated cells were obtained by passing the softened chunks through a 1 ml pipette tip several times, and collected by centrifugation at 365 xg for 5 min. The cells were resuspended in minimum essential medium alpha containing 5% (v/v) fetal bovine serum and 5% (v/v) calf serum, and plated onto a 10 cm culture dish. One day after plating, attached cells (designated as passage 0) were washed with HBSS to remove serum, and thereafter maintained in the medium (GM), a 3:7 mixture (v:v) of B104 neuroblastoma-conditioned medium and the N1 medium (high glucose Dulbecco's modified Eagle's medium supplemented with 6 mM
L-glutamine, 10 ng/ml biotin, 5 μg/ml insulin, 50 μg/ml apo-transferrin, 30 nM sodium selenite, 20 nM progesterone and 100 μM putrescine as final concentrations). Cultures were fed with fresh GM medium every other day for approximately 5 days, at which time the proliferating glial cells were almost confluent.
Immunopanning for purification of A2B5+rat OPCs
The mixed glial cultures were washed with Ca
2+ and Mg
2+-free HBSS, suspended in the N1 medium containing 0.1% (w/v) BSA, and plated and incubated on negative immunopanning plates coated with RAN-2 antibody for 30 min at 37°C to exclude RAN-2-positive cells [
29]. Following two rounds of this negative selection, nonadherent cells were transferred to the A2B5 positive panning plates. After the serial immunopanning, purified cultures contained more than 95% of OPCs which were A2B5-positive, O4-negative, and glial fibrillary acidic protein-negative.
Immunocytochemistry
Cells cultured on poly-D-lysine-coated coverslips were incubated with A2B5 hybridoma supernatants (undiluted) at room temperature for 30 min. After washing with phosphate-buffered saline (PBS), cells were fixed with 4% paraformaldehyde at room temperature for 15 min, and then permeabilized with 100% methanol at -20°C for 15 min. For IRF8 staining, cells were incubated with anti-IRF8 antibody diluted at 1:50 in PBS containing 5% normal goat serum and 0.03% Triton-X 100 at room temperature for 2 h, after permeabilization by 100% methanol. After incubation with fluorophore-conjugated secondary antibodies (1:50, v:v) in PBS at room temperature for 30 min, nuclei were counterstained with 4,6-diamidio-2-phenylindole (0.5 μg/ml) for 10 min, and then the coverslips were mounted on slide glasses with VectorShield (Vector laboratory, Burlingame, CA, USA).
Immunoblots
Protein lysates were prepared in the lysis buffer as described previously [
10]. Twenty μg of protein from each sample were size-fractioned by SDS-polyacrylamide gel electrophoresis, transferred onto a nitrocellulose membrane (Schleicher & Schnell, Keene, NH, USA) and probed with primary antibodies for IRF1 (1:400, v:v) and IRF8 (1:5000, v:v) for 1 h. Full range recombinant Rainbow Molecular Weight Markers (Amersham Biosciences, Piscataway, NJ, USA) were used as a reference for molecular sizes. Immunoreactive signals were detected by enhanced chemiluminescence according to the manufacture's protocol (Amersham Biosciences). Equal protein loading was confirmed by subsequent probing with the mouse monoclonal antibody against GAPDH in each experiment.
Caspase activity assay
Cells were homogenized in lysis buffer (100 mM HEPES; 10% (w/v) sucrose; 0.1% (w/v) 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate; 10 mM dithiothreitol; 1 mM EDTA; 1 mM phenylmethylsulfonyl fluoride, 2 μg/ml aprotinin; 1 μg/ml pepstatin; 5 μg/ml leupeptin) [
30]. The protein lysates were stored at -80°C until use as a 1:1 (v:v) mixture with glycerol. Caspase activity was measured by a fluorometric method; protein samples (10 μg) were incubated with the fluorogenic substrate, acetyl-Asp-Glu-Val-Asp-α-(4-methylcoumaryl-7-amide) (12.5 μM) (Ac-DEVD-AMC, Peptides international, Luoisville, KY, USA) in 250 μl of the lysis buffer, and cleavage of Ac-DEVD-AMC was monitored by a multiplate spectrofluoromater, Gemini EM (Molecular devices, Sunnyvale, CA, USA) for 60 min at 25°C. The DEVD-cleavage activity was expressed as delta RFU (relative fluorescence unit)/μg protein/h.
5-bromo-2'-deoxyuridine (BrdU)-incorporation assay
OPCs cultured in 60 mm dishes were exposed to a 4 h BrdU pulse (10 μM) just prior to harvesting. The trypsinized cells were collected in GM and resuspended in 1.5 ml PBS. After fixation by 70% (v/v) ethanol at -20°C for overnight, 5 × 104 cells were washed with 1 ml of the washing buffer (0.1% (w/v) BSA in PBS), and denatured by resuspension in 2N HCl at room temperature for 20 min. After resuspending once more in washing buffer, the cells were incubated in 0.1 M sodium borate at room temperature for 2 min to neutralize any residual acid. Cells that had incorporated BrdU following incubation were identified by incubation with a fluorescein isothiocyanate (FITC)-conjugated mouse anti-BrdU monoclonal antibody at room temperature for 20 min in dilution buffer (0.1% (w/v) BSA, 0.5% (v/v) Tween-20 in PBS) followed by another resuspension in washing buffer. The labeled cells were detected in the green (FL1) channel of a flow cytometer, CyAn-ADP (Dako cytomation, Carpinteria, CA). FITC-conjugated mouse monoclonal IgG1 was used as isotype control.
MTT assay
Cell viability was estimated by the enzymatic conversion of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) to formazan crystals in live cells. Formazan was dissolved in dimethyl sulfoxide at 90 min after addition of MTT (0.5 mg/ml) to the culture medium, and quantified by a spectrophotometer or a microplate reader at 560 nm.
Annexin-V and propidium iodide binding assay
The OPC cultures were maintained in 60 mm dishes and subjected to various experimental treatments. At 0, 24, and 48 h after these treatments, the culture medium containing detached dead cells was collected, and the attached cells were washed once with 2 ml of Ca
2+ and Mg
2+-free HBSS. The attached cells were removed from the plate by exposure to 0.5 ml of 0.05% trypsin at 37°C for 2 min, suspended in 2 ml of GM with 625 μg/ml trypsin inhibitor, and collected into a 15 ml tube together with the saved medium and the Ca
2+ and Mg
2+-free HBSS used for wash. After centrifuge at 520 xg for 10 min, the pellet was resuspended into 0.4 ml of binding buffer (0.1 M HEPES, pH 7.4; 140 mM NaCl; 2.5 mM CaCl
2; 0.45% (w/v)
D-glucose). Five μl of FITC-conjugated annexin-V solution and propidium iodide (PI; 8 μg/ml at final concentration) were added into 0.1 ml of the cell suspension. After incubation at room temperature for 15 min in the dark, 0.3 ml of the binding buffer was added to the cell suspension. To determine the absolute number of cells in each preparation, Flow-Count™ fluorospheres were added at a concentration of 19 beads/μl just before flow cytometry by CyAn-ADP (DakoCytomation, Carpinteria, CA, USA). Fluorescence of annexin-V-FITC and PI were detected in FL-1 and FL-4 channels, respectively. Gatings and data acquisition and analysis were carried out using Summit software (DakoCytomation) as described previously [
10].
Cell death and loss of mitochondrial membrane potential assay
Rat OPCs cultured in 24-well plates were treated with the GM or the GM supplemented with IFNγ for 12, 18, and 24 h. Prior to collection, cells were incubated with tetramethylrhodamine ethyl ester (TMRE, 0.1 μM) at 37°C for 30 min. Then, culture medium containing dead cells was collected, and cells were washed once with 0.5 ml of the Ca
2+ and Mg
2+-free HBSS. Attached cells were removed with 150 μl of 0.05% trypsin for 1 min, suspended in 1 ml of the GM, and collected into a 15-ml tube together with the saved medium and the Ca
2+ and Mg
2+-free HBSS used for washing. After centrifugation with 1500 rpm for 5 min, the supernatant was aspirated and the pellet was kept on ice. Pellets were resuspended with 0.5 ml PBS containing 5 μM DAPI and 0.1% BSA immediately prior to analysis by flow cytometry employing a Cyan-ADP Flow Cytometer (DakoCytomation). Live and dead cell populations were gated as described previously [
10], and TMRE and DAPI were detected in the FL-2 and the FL-6, respectively.
Real-time PCR
Real-time PCR analyses were performed by MX3005P (Stratagene, La Jolla, CA, USA) using TaqMan
® Assay-on-Demand™ assay kits (assay nos.: Rn00561424_m1, Mm00515204_m1, Rn01764369_m1, Rn01435145_m1, Rn01500522_m1, Rn01762216_g1 and Rn01751474_m1 for detection of IRF1, IRF2, IRF3, IRF4, IRF5, IRF8 and interferon gamma induced GTPase (IGTP) cDNA, respectively) [
17]. For detection of IRF6, IRF7, and IRF9 cDNA, each set of primers and a probe was obtained from Applied Biosystems as a Custom TaqMan
® Gene Expression Assay, because the kits for these cDNA were not available at the time of these experiments. For standardization, GAPDH cDNA levels were quantified with TaqMan Rodent GAPDH Control Reagents according to the manufacturer's instructions, and the absolute cDNA amounts were expressed as ratios to GAPDH cDNA. We present representative data from at least two independent analyses for each mRNA.
Plasmid construction
For the forced expression vectors, the open reading frame of rat IRF1 or IRF8 was inserted in the pcDNA3.1 mammalian expression vector followed by the internal ribosome entry site (IRES) and humanized Renilla reniformis green fluorescent protein (hrGFP, Stratagene) in order to facilitate identification of transfected cells by flow cytometry or by fluorescence microscopy. The expression vector for the dominant-negative form of IRF1 (IRF1DN-hrGFP) and was constructed by inserting the coding sequence of truncated rat IRF1 (amino acids 1 to 144) into the pcDNA3.1 mammalian expression vector. The truncated IRF1 coding sequence was fused to hrGFP in frame with a spacer sequence, Pro-Gly-Gly-Gly-Gly-Pro (P4GP) hinge, in order to facilitate identification of transfected cells and to evaluate the intracellular localization and stability of the dominant negative protein. For forced double expression of IRF1 and IRF8, the expression construct for IRF8 lacking hrGFP reporter (PCMV-IE-IRF8-pA) was prepared by inserting the open reading frame of rat IRF8 into pcDNA3.1.
Transfection by electroporation
Trypsinized OPCs (2 × 106) were resuspended in 100 μl of N1 medium with 10 μg plasmid DNA or 2 μM siRNA, and put into a 2 mm cuvette. A square pulse with 110 mV for 25 msec was applied to the mixture of cells and plasmid DNA with BioRad GenePulser Xcell (BioRad, Hercules, CA, USA). Cells were resuspended into GM, plated on 24 well plates, and subjected to further experimental procedures.
Statistical analysis
Data are presented as mean ± SD unless otherwise noted. Statistical significance was determined by two-tailed ANOVA followed by Student-Newman-Keuls post hoc test.
Discussion
Proapoptotic effects of IFNγ and at most minimal cytotoxic effects of IFNβ on OPCs have been reported previously [
4‐
9,
12,
13]. In the present study, however, we have directly compared effects of IFNγ and IFNβ on OPCs in the same
in vitro condition, and confirmed a substantial difference in proapoptotic effects between the two IFNs. Furthermore, IFNβ was not protective against IFNγ-induced OPC apoptosis, despite several prior reports that IFNβ antagonizes IFNγ signaling [
39‐
43]. As far as we could determine by transcriptional induction of IRF1, simultaneous application of IFNβ failed to reduce IFNγ-mediated robust induction of IRF1 [
17]. Although the mechanisms underlying the beneficial therapeutic effects of IFNβ on relapsing-remitting MS are still largely unknown, recent studies have indicated that IFNβ and type I IFN receptor-mediated signaling limit CNS autoimmunity by regulating innate immune responses in peripheral tissues [
44,
45] and the production and properties of T
H17 cells, a pathogenic T helper subset largely responsible for CNS autoimmunity [
46]. Despite the beneficial effects of IFNβ which is further ensured by far less cytotoxicity of IFNβ to OPCs, we observed that IFNβ did inhibit the cell cycle in OPCs, though to a lesser extent than IFNγ. It is thus conceivable that, as demonstrated by Trebst et al.[
14], IFNβ attenuates the endogenous capability for remyelination, which is presumably masked by its profound beneficial effects on the immune system.
Based on the marked difference in proapoptotic effects between IFNγ and IFNβ on OPCs, our next aim in this study was to identify those ISGs responsible for IFNγ-mediated OPC apoptosis. IFNγ induces robust and sustained elevation of IRF1, whereas IFNβ elicits only a transient elevation of IRF1, which ends up being undetectable at the protein level at 24 h after treatment, indicating that IRF1 is a candidate for such an ISG [
17]. In support of this, Balabanov's group has recently reported that, using a lentiviral expression system, down-regulation of IRF1 by IRF1 shRNA partially protected against IFNγ-induced OPC apoptosis, and that forced expression of IRF1 reduced the viability of OPCs [
18]. We employed a different forced expression system and a dominant negative approach in this study, and confirmed significant involvement of IRF1 in IFNγ-mediated OPC apoptosis. We further provided direct evidence for activation of the mitochondrial apoptotic pathway by overexpression of IRF1 alone. Notably, however, both approaches used in our study and the study by Balabanov's group to down-regulate IRF1-mediated transcription failed to completely inhibit IFNγ-mediated apoptotic events, suggesting possible functional redundancy of the ISGs involved in IFNγ-mediated transcriptional activation leading to apoptosis of OPCs. Given the structural and functional similarity among members of the IRF family and their known interactions, transcriptional activity of IRF1 is likely to be modified or compensated by the other members of the IRF protein family.
In an effort to obtain a comprehensive expression profile of the IRF family in OPCs stimulated by either IFNγ or IFNβ, we found that IRF8 was also up-regulated by IFNγ but not by IFNβ. IRF8 was originally identified as a protein that binds to the ISRE in the promoter region of the MHC class I gene H-2LD [
47], and was believed to be expressed exclusively in the hematopoietic lineage (Reviewed in [
48]). Our result indicates that OPCs are also capable of expressing IRF8 in response to IFNγ. In contrast to overexpression of IRF1, however, overexpression of IRF8 alone resulted in only transient depolarization of the mitochondrial membrane in OPCs, but failed to reduce their viability. More importantly, despite this weak proapoptotic effect of overexpressed IRF8 itself, it significantly enhanced the IFNγ-induced apoptosis and proapoptotic effect of overexpressed IRF1 in OPCs even in the absence of IFNγ. Unlike other IRF members, IRF8 is capable of binding to the target DNA motif only following association with IRF1, IRF2 or non-IRF transcription factors such as PU.1 [
36,
49]. As an example, IRF8 and IRF1 synergistically induce several genes, such as IL-12 and iNOS [
50,
51], in activated macrophages. A study from Ozato's group also demonstrated that IRF8 induced by activated STAT1 forms a multiprotein transcriptional complex with other nuclear proteins, which binds to GAS, and, in turn, potentiates transcriptional activation of the ISGs in a GAS-dependent manner [
52]. Therefore, it is conceivable that, although IRF8 alone is not sufficient to activate the apoptogenic cascade in OPCs, IRF8 enhances IFNγ-induced OPC apoptosis by interacting with other transcription factors activated by IFNγ. Indeed, IRF8 is known to function as a proapoptotic transcription factor like IRF1. IRF8-deficient mice are characterized by a myeloproliferative phenotype resulting in a syndrome similar to human chronic myelogenous leukemia [
53]. This oncogenic phenotype is attributable to cytokine hypersensitivity and apoptosis resistance of IRF8 deficient myeloid progenitor cells [
54]. During differentiation of the myeloid lineage, IRF8 down regulates anti-apoptotic genes such as Bcl-X
L, one of anti-apoptotic member of the Bcl-2 family, and PTPN13, which encodes an inhibitor of Fas-mediated apoptosis [
55,
56]. The anti-oncogenic roles of IRF8 are associated with its proapoptotic function in the other types of tumors as well. In colon carcinoma cells, IRF8 induced by IFNγ sensitizes them to Fas-mediated apoptosis, but the silencing of the IRF8 gene by methylation of its promoter region renders them resistance to IFNγ-mediated apoptosis [
57].
Reduction of IRF8 by siRNA failed to enhance viability of OPCs after treatment with IFNγ, although it partially but significantly decreased the number of preapoptotic cells. However, we still could not rule out a contribution of the endogenous IRF8 in the IFNγ-induced OPC apoptosis, because the transfection of the IRF8 siRNA resulted in only a partial suppression against the robust IRF8 induction by IFNγ. Together, these results support the notion that endogenous IRF8 positively regulates the IFNγ-induced OPC apoptosis depending on its induced dosage.
We previously demonstrated that, unlike OPCs, mature myelin-producing oligodendrocytes were totally resistant to IFNγ-induced apoptosis [
10]. Nevertheless, IRF1 was similarly induced by IFNγ in mature oligodendrocytes compared with OPCs [
10,
17]. We also confirmed that IFNγ induced IRF8 mRNA at similar levels in both OPCs and mature oligodendrocytes (Data not shown.). These results indicate that IRF1-mediated transcriptional activations may be necessary to activate the apoptotic cascade in OPCs, but are not sufficient. We speculate that differences in cellular context between OPCs and mature oligodendrocytes such as activities of ERK signaling are the other necessary components for IFNγ-induced OPC apoptosis as well [
10,
58,
59].
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
MH prepared the cultures and carried out most of the experiments and data analysis, and wrote the manuscript. AI carried out RNA isolation and qPCR experiments, and helped development of the expression constructs. DP participated in data interpretation and critical reading of the manuscript. KO provided anti-IRF8 antibody and participated in data interpretation and critical reading of the manuscript. TI conceived the study, contributed to the experimental design, and helped to write the manuscript. All authors read and approved the final manuscript.