Interleukin-6 (IL-6) Production by Astrocytes: Autocrine Regulation by IL-6 and the Soluble IL-6 Receptor

MATERIALS AND METHODS

Cell lines and primary human astrocyte cultures. The U373-MG astroglioma cell line was grown in DMEM supplemented with 10 mm HEPES, pH 7.2, 2 mml-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, 0.1 mm nonessential amino acid mixture, 1 mmsodium pyruvate, and 10% fetal bovine serum (FBS), as described previously (Oh et al., 1998). CRT human astroglioma cells were grown in RPMI 1640 medium supplemented with 10 mm HEPES, pH 7.2, 2 mml-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and 10% FBS as described previously (Rosenman et al., 1995). Biopsy material from patients undergoing surgery to treat intractable epilepsy was used to prepare human adult astrocyte cultures as described previously (Barnum et al., 1992). Primary human adult astrocytes were obtained after 30 d in culture and were grown in DMEM, high glucose formula supplemented with 2 mml-glutamine, 0.1 mm nonessential amino acid mixture, 0.1% gentamicin, 2.5 μg/ml amphotericin B, and 10% FBS. Astrocytes were 87–90% GFAP positive (Barnum et al., 1992). Human astrocytes obtained from this source have many of the same functional responses as do primary cultures of rat astrocytes (Benveniste et al., 1994; Oh et al., 1998).

Reagents. Human recombinant TNF (rTNF)-α (9.0 × 10⁷ U/mg) and human rIL-1β (2.0 × 10⁸ U/mg) were purchased from Genzyme (Cambridge, MA), and human rIL-6 (7.0 × 10³ U/μg), sIL-6Rα [100 active units (AU)/μg], rTGF-β₁(2.5 × 10⁴ U/μg), rOSM (4.4 × 10³ AU/μg), rLIF (8.3 × 10⁴ U/μg), rCNTF (10 AU/μg), soluble rCNTFR (3.3 AU/μg), and rIL-11 (6.6 × 10³ AU/μg) were purchased from R & D Systems (Minneapolis, MN). Hyper-IL-6 (H-IL-6) was the kind gift of Professor Stefan Rose-John (Universität Mainz) and was prepared as described previously (Fischer et al., 1997). Anti-human IL-6, anti-human IL-6R, and anti-human gp130 neutralizing antibodies were obtained from R & D Systems. Purified mouse IgG₁ (anti-TNP) (used as an isotype control) was purchased from PharMingen (San Diego, CA).

IL-6 production in astrocytes. U373-MG cells, CRT cells, and primary human astrocytes were resuspended in their respective media containing 10% FBS and plated at 0.5 × 10⁶cells/well into six-well (35 mm) plates (Costar, Cambridge, MA). The cells were allowed to reach ∼90% confluency, media was aspirated, and fresh serum-free media was added to cells for 12 hr. After this time, cells were washed once with sterile PBS, and fresh serum-free media was added to each well. Astrocytes were treated with various reagents (TNF-α, IL-1β, IL-6, sIL-6R, TGF-β, CNTF, sCNTFR, LIF, and IL-11) alone or in combination for 24 hr, then supernatants were collected, centrifuged to remove contaminating cells, and stored at −70°C until use.

Measurement of IL-6 activity. IL-6 activity in astrocyte culture supernatants was determined in a biological assay using the IL-6-dependent B cell hybridoma B9, as described previously (Norris and Benveniste, 1993). Briefly, B9 cells (6 × 10³cells/well) were plated in 96-well microtiter plates, then serial dilutions of astrocyte-conditioned medium and recombinant human IL-6 (used as a standard) were added and incubated at 37°C for 72 hr. Triplicate cultures were set up for each condition. After 72 hr, B9 cell growth was assessed using the MTT assay as described previously (Norris and Benveniste, 1993). In this assay, the amount of IL-6 in astrocyte-conditioned supernatants was determined by comparison to a recombinant human IL-6 standard in which colorimetric change versus recombinant human IL-6 concentration (picograms per milliliter) is known. The reagents used in this study to stimulate IL-6 production (TNF-α, IL-1β, sIL-6R, TGF-β, OSM, LIF, CNTF, sCNTFR, and IL-11) do not support B9 growth when added directly to B9 cells or when added to astrocyte cultures immediately before collecting supernatants (data not shown).

RNA isolation, riboprobes, and RNase protection assay (RPA).U373-MG cells were plated at 4 × 10⁶ cells per 100 mm² dish (Costar, Cambridge, MA). Upon confluency, cells were serum-starved for 12 hr. Total cellular RNA was isolated from resulting confluent monolayers of astrocytes that had been incubated for various times with cytokines. RNA isolation was performed as described previously (Lee et al., 1997). A pGem4Z plasmid containing a 349 bp fragment corresponding to 194–542 bp of the human IL-6 cDNA was linearized with EcoRI. In vitrotranscription of this linearized plasmid with T7 RNA polymerase generated an antisense probe 408 nucleotides in length (note that this probe contains a portion of the pGem4Z plasmid). A pAMP-1 vector containing a fragment of the human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA (corresponding to 43–531 bp) was linearized with NcoI, which digests within the GAPDH cDNA insert.In vitro transcription of this linearized plasmid with T7 RNA polymerase generates a 290 bp antisense RNA probe.

In vitro transcription of riboprobes and RPA were conducted as described previously (Lee and Benveniste, 1996). Total RNA (15 μg) was hybridized with IL-6 (30 × 10³ cpm) and GAPDH (25 × 10³ cpm) riboprobes at 42°C overnight. The hybridized mixture was then treated with RNase A/T1 (1:200 dilution) at room temperature for 1 hr and analyzed by 5% denaturing (8 m urea) PAGE, and the gels were exposed to x-ray film. The protected fragments of the IL-6 and GAPDH riboprobes are 349 and 230 nucleotides in length, respectively. Quantification of protected RNA fragments was performed by PhosphorImager (Molecular Dynamics, Sunnyvale, CA). Values for IL-6 mRNA expression were normalized to GAPDH mRNA levels for each experimental condition. GAPDH mRNA was used as a control gene because its levels are not affected by cytokine treatment. The RPA shown in this article is overexposed for GAPDH because the signal for IL-6 is weaker. Quantification of the original gel was performed on a PhosphorImager to arrive at accurate values.

Nuclear run-on analysis. Nuclear run-on analysis was performed as described previously (Bethea et al., 1992; Chung et al., 1992; Shrikant et al., 1994). Nuclei were isolated from confluent monolayers of astrocytes that were incubated for 30–45 min in the presence or absence of stimulus (IL-6/sIL-6R, TNF-α, IL-1β, TNF-α + IL-6/sIL-6R, and IL-1β + IL-6/sIL-6R). The cells (30 × 10⁶) were collected, washed once with cold PBS, and pelleted. Nuclei were isolated by lysing the cells with 10 mm Tris, pH 7.4, 2 mm MgCl₂, 10 mm NaCl, and 0.5% Nonidet P-40, followed by centrifugation at 1000 × g. The nuclei were stored at −80°C in buffer containing 50 mm Tris, pH 8.3, 40% glycerol, 5 mm MgCl₂, and 0.1 mm EDTA. To perform run-on transcriptional analysis, the nuclei were thawed on ice and incubated for 30 min at 30°C in reaction buffer containing 10 mm DTT, 1 mm each of ATP, CTP and GTP, and 0.25 mCi [³²P]UTP (3000 Ci/mm) (Amersham, Arlington Heights, IL). After the reaction, DNA digestion was performed in the presence of 0.5m NaCl, 50 mm MgCl₂, 2 mm CaCl₂, and 10 mm Tris, pH 7.4, and DNase (10 mg/ml) (Promega, Madison, WI) for 5 min at 30°C. Protein digestion was then performed in the presence of 5% SDS, 0.5m Tris, pH 7.4, 0.125 m EDTA, and proteinase K (20 μg/sample) for 30 min at 42°C. Nuclei were lysed and RNA was harvested as described in RNA isolation. Denatured circular plasmid DNA was immobilized on nitrocellulose paper using a Millipore (Bedford, MA) Milliblot S system. After UV cross-linking the DNA to the nitrocellulose, prehybridization was performed at 65°C for at least 3 hr in a solution of 50% formamide, 0.1% SDS, 5× SSC, 2.5× Denhardt’s, 250 μg/ml tRNA, and 50 mmNa₂PO₄, pH 6.5. For hybridization, 9 × 10⁶ cpm of labeled and denatured RNA was used in 1 ml of hybridization solution and incubated at 65°C for 48 hr. The filters were washed twice for 30 min at 65°C in 2× SSC, 30 min at 37°C in 2× SSC with 10 μg/ml ribonuclease A/T1, and finally for 30 min at 37°C in 2× SSC. The filters were exposed to the PhosphorImager (Molecular Dynamics) for quantification. The increase of transcriptional activation was determined by comparing the ratios of IL-6/GAPDH values obtained for each stimulus.

Statistical analysis. Significance between experimental values was determined using two-way ANOVA. *p < 0.05, **p < 0.01, and ***p < 0.001.

RESULTS

DISCUSSION

IL-6 has been shown to have beneficial potential in the CNS because of its neurotrophic and neuroprotective effects (Satoh et al., 1988; Hama et al., 1989; Gadient and Otten, 1997; Loddick et al., 1998;März et al., 1998b), and anti-inflammatory actions through the inhibition of VCAM-1, intercellular adhesion molecule-1, and TNF-α expression by CNS cells (Benveniste et al., 1995; Shrikant et al., 1995; Oh et al., 1998). However, other reports suggest that IL-6 is detrimental and adds to the pathophysiology associated with CNS disorders (Campbell et al., 1993; Klein et al., 1997). Collectively, these studies illustrate the pleiotropic nature of IL-6 in the CNS and stress the need for an understanding of IL-6 regulation in the CNS. Previously, we have shown that human astrocytes express low levels of the IL-6R and require the addition of the sIL-6R for IL-6-mediated responses (Oh et al., 1998). Herein, we have shown that treatment of astrocytes with IL-6 and the sIL-6R leads to modest increases in IL-6 mRNA expression. However, co-treatment with either TNF-α or IL-1β plus IL-6/sIL-6R leads to synergistic increases in IL-6 gene expression.

We speculated that TNF-α stimulation of astrocytes would provide a source of endogenous IL-6 that would then interact with the sIL-6R and possibly influence IL-6 expression. The addition of TNF-α and the sIL-6R led to synergistic increases in IL-6 expression by U373-MG cells, primary human astrocytes, and CRT cells (Fig. 1, Table 1). These results suggest that the sIL-6R plays a pivotal role in determining the levels of IL-6 expressed by astrocytes in the CNS and furthermore influences IL-6 function in the CNS. This is supported by several reports that CNS cells that are normally slightly responsive or unresponsive to IL-6 become responsive on addition of the sIL-6R. In the absence of nerve growth factor, IL-6 weakly enhances sympathetic neuron survival, and the addition of the sIL-6R enhances neuronal survival (März et al., 1998a). In human astrocytes, IL-6 alone has no effect on α₁-antichymotrypsin expression, whereas addition of the sIL-6R leads to expression (Kordula et al., 1998). In the presence of the sIL-6R, IL-6 inhibits TNF-α-induced VCAM-1 expression in astrocytes (Oh et al., 1998). These findings suggest that the sIL-6R is a functionally relevant CNS molecule. In this regard, sIL-6R is detectable in the CSF of normal individuals (0.8–1.67 ng/ml) (Frieling et al., 1994; Rose-John and Heinrich, 1994), although the source of sIL-6R in the CNS is unknown. A likely source of CNS sIL-6R is Purkinje neurons, which exhibit intense immunostaining for the IL-6R (Nelson et al., 1999). As well, IL-6R immunoreactivity has been detected in hypoglossal nerve cell bodies in the brainstem and in dorsal root ganglion neurons (Hirota et al., 1996; Thier et al., 1999). Other potential CNS sources of sIL-6R, particularly under inflammatory conditions, are activated macrophages and T-cells (Müllberg et al., 1993; Banning et al., 1998; Jones et al., 1998). Preliminary results from our laboratory indicate that sIL-6R, in the range of 300–900 pg/ml, can be detected from human brain extracts (our unpublished observation). Circulating levels of sIL-6R are considerably higher (∼70 ng/ml) and may increase CNS levels on disruption of the BBB, thereby providing an endogenous CNS source of sIL-6R sufficient to elicit functional responses.

Addition of TNF-α plus IL-6/sIL-6R led to an increase in the overall level of IL-6 mRNA as well as the duration of expression (Fig. 2). To determine the basis of this increase, mRNA stability assays were performed. As shown in Figure 5A, thet_½ of TNF-α-induced IL-6 message was ∼90 min, and the inclusion of IL-6/sIL-6R did not affect thet_½ of the IL-6 message. Therefore, the synergistic effect of TNF-α and IL-6/sIL-6R on IL-6 gene expression is likely caused by transcriptional regulation. Nuclear run-on analysis indicated that both TNF-α and IL-6/sIL-6R induce IL-6 transcription, which is enhanced in the presence of both stimuli (Fig. 6).

IL-1β alone is a more potent inducer of IL-6 mRNA and protein than TNF-α (Fig. 4, Table 2). Our results provide the explanation for this: enhanced IL-1β induced IL-6 transcription and stabilization of the IL-6 message. Nuclear run-on analysis revealed that IL-1β induced IL-6 gene transcription to a greater extent than TNF-α (6.7-fold for TNF-α vs 13.3-fold for IL-1β) (Fig. 6). As well, IL-1β induces a more stable IL-6 message than TNF-α (Fig. 5A,D). Interestingly, IL-1β stimulation of cells for 1 hr induced an IL-6 message with a t_½ of ∼60 min, whereas a 6 hr stimulation with IL-1β resulted in an IL-6 message that did not undergo any appreciable decay for up to 18 hr (Fig.5C,D). Thus, it appears that there is a time dependence for IL-1β-mediated stabilization of the IL-6 message. In this regard, there are several reports of IL-1β stabilization of IL-6 message. In murine mast cells, IL-1β alone has no effect on IL-6 expression, but it enhances expression induced by c-kit ligand and IL-10 (Lu-Kuo et al., 1996). In fibroblasts, TNF-α and IL-1β synergistically stimulate IL-6 expression that is caused in part by stabilization of the IL-6 message (Elias and Lentz, 1990). Interestingly, in those two studies, IL-1β alone did not stabilize the IL-6 message, which is what we have observed in the U373-MG cells. Thus, the ability of IL-1β to stabilize the IL-6 message may occur in cell type-specific manner.

IL-1β induction of IL-6 protein expression was significantly enhanced in the presence of sIL-6R (Table 2), and elevated levels of IL-6 mRNA were detected in cells treated with IL-1β plus IL-6/sIL-6R over treatment with either stimulus alone (Fig. 5B). The synergistic effect of IL-1β plus IL-6/sIL-6R appears to be caused by increased transcription of the IL-6 gene in the presence of both stimuli (Fig. 6) because IL-6/sIL-6R does not affect thet_½ of IL-6 message induced by IL-1β at either 1 or 6 hr (Fig. 5C,D). IL-6/sIL-6R has been shown to induce its own synthesis in osteoblasts by transcriptional mechanisms (Franchimont et al., 1997), and our study also documents an effect of IL-6/sIL-6R on IL-6 transcription.

Co-treatment with TGF-β plus the sIL-6R did not lead to synergistic increases in IL-6 mRNA and protein expression (Fig. 4, Table 2). The difference in the ability of TNF-α and IL-1β, versus TGF-β, to synergize with IL-6/sIL-6R may be dependent on transcriptional regulation. TNF-α and IL-1β regulate IL-6 transcription by similar mechanisms that involve use of NF-κB and CCAAT enhancer binding protein (C/EBP) binding sites (Akira et al., 1990; Sparacio et al., 1992; Matsusaka et al., 1993; Miyazawa et al., 1998). Treatment of U373-MG cells with TNF-α and IL-1β, but not TGF-β, leads to NF-κB interactions with the NF-κB response element in the IL-6 promoter, as determined using electrophoretic mobility shift assays (data not shown). We hypothesize that synergy between TNF-α or IL-1β and IL-6/sIL-6R requires NF-κB activation as well as the activation of an undetermined IL-6 inducible transcription factor, possibly C/EBPβ. C/EBPβ has been shown to function in the positive regulation of the IL-6 promoter as well as other IL-6 inducible promoters (Miyazawa et al., 1998). Maximal IL-6 transcription as seen by TNF-α or IL-1β plus IL-6/sIL-6R may involve the interactions of NF-κB and C/EBPβ, as has been shown for IL-8 and serum amyloid A gene expression (Stein and Baldwin, 1993;Kunsch et al., 1994; Ray et al., 1995).

Most members of the IL-6 cytokine family act through ligand-specific receptors that interact with gp130 on binding ligand. OSM differs from this model in that OSM interacts first with gp130 (Gearing et al., 1992; Sporeno et al., 1994), then recruits either the LIFRβ (Thoma et al., 1994), a receptor used by various members of this family, or the OSMR, a receptor specific to OSM (Mosley et al., 1996). Other members of the IL-6 cytokine family were tested for effects on IL-6 expression; OSM induced IL-6 protein expression and strongly synergized with TNF-α for enhanced IL-6 expression. LIF alone induced a modest level of IL-6 protein but did not synergize with TNF-α (Fig. 7). On the basis of the ability of OSM to significantly influence IL-6 expression, we propose that OSM along with IL-6 may affect the function of astrocytes. To our knowledge, no studies have been conducted to determine OSM expression in the brain. However, preliminary data indicate that human astrocytes, on stimulation, can produce OSM protein (our unpublished observation). This suggests a potential CNS source of this cytokine.

In summary, data presented herein support a positive autoregulatory role of IL-6 and the sIL-6R in astrocytes for IL-6 regulation. These findings may be relevant to inflammatory conditions in the brain, where cytokines such as TNF-α and IL-1 play an essential role in this process through the induction of chemokines and adhesion molecules, recruitment of immune cells into the CNS parenchyma, and ultimately, activation of immune cells and endogenous glial cells (for review, seeBenveniste, 1997). In this context of elevated CNS expression of TNF-α, IL-1β, and IL-6, and the presence of the sIL-6R, a cytokine circuitry is established by which high levels of expression of IL-6 can occur, which may lead to some of the neuropathology associated with CNS inflammation. The ultimate biological effect of IL-6 in the CNS will depend on the availability of IL-6 receptors, both membrane bound and soluble forms, as well as the levels of soluble gp130, which can neutralize IL-6/sIL-6R complexes, thereby acting as an antagonist (for review, see Heinrich et al., 1998). Whether soluble gp130 exists in the CNS is not currently known; however, the expression of this component of the IL-6 receptor signaling complex will impact on IL-6 biological effects in the CNS.