Address correspondence and reprint requests to Iain L. Campbell, School of Molecular and Microbial Biosciences, University of Sydney, NSW 2006, Australia. E-mail: firstname.lastname@example.org
An imbalance in the matrix metalloproteinase (MMP) : tissue inhibitor of MMP (TIMP) ratio may be associated with tissue injury. Here, we studied the regulation of TIMP and MMP gene expression in primary glial cultures to ascertain the factors involved in the regulation of these genes in conditions of inflammatory neuropathology. Astrocytes were found to basally express TIMP-1 and TIMP-3 mRNA while microglia expressed only TIMP-2 mRNA. TIMP-4 mRNA was not detectable in either cell type. Treatment with interferon-α (IFN-α), IFN-γ, interleukin-3 (IL-3), IL-6 or tumor necrosis factor-α (TNF-α) did not alter expression of the TIMP genes. However, in astrocytes, but not in microglia, serum, IL-1β or lipopolysaccharide (LPS) evoked a dose- and time-dependent increase in TIMP-1 mRNA and a coincident down-regulation of the TIMP-3 gene. Astrocytes were found to express mRNA constitutively for MMPs -3, -11 and -14. In contrast, microglia expressed only MMP-12 mRNA under basal conditions. IL-1β enhanced MMP-3 mRNA levels while LPS increased the MMP-3, -9, -12, -13 and -14 mRNAs. Our findings reveal that regulatory control of TIMP and MMP gene expression by glial cells is agonist- and cell-type specific, and suggest that innate immune signals govern the temporal and spatial expression patterns of TIMP and MMP genes in neuroinflammatory conditions of the CNS.
Glia have important roles as non-neuronal mediators of CNS homeostasis. As sentinel cells associated with the blood–brain barrier (BBB), astrocytes and microglia are both active and reactive participants in the CNS response to neuronal injury and neuroinflammation (Becher et al. 2000; Hoek et al. 2000; Broderick et al. 2002; Campbell 2002). Accordingly, glia can produce factors, such as cytokines, that have the potential to quell immune cell migration and promote CNS repair (Trajkovic et al. 2004; Aldskogius 2005). However, disparate negative functions have also been attributed to glia as mediators of the recruitment and co-activation of T-cells trafficking into the CNS during immune-mediated injury (Ford et al. 1996). Accumulating studies have demonstrated that astrocyte-driven expression of pro-inflammatory cytokines is sufficient to elicit CNS neuropathology, but can also provide protection from CNS injury (Campbell et al. 1993; Carr et al. 1998; Pagenstecher et al. 2000b; Minagar et al. 2002; Penkowa et al. 2003). Hence, defining how glia respond to pro-inflammatory stimuli may delineate key processes underlying their roles in shaping the responsiveness of the CNS to inflammation in a variety of pathological states.
Matrix metalloproteinases (MMPs) are zinc-dependent endopeptidases responsible for the dynamic remodeling of the extracellular matrix. MMP-mediated proteolysis is critical for cellular homeostasis. Dysregulation of MMPs has been associated with diverse human disease conditions, including cancer metastasis, cardiac injury and brain injury. Elevated expression and activity of MMPs are associated with an increasing number of neurodegenerative and neuroinflammatory conditions (reviewed in Rosenberg 2002; Crocker et al. 2004). Current data suggest that increased proteolytic activity of MMPs during disease contributes to many common CNS diseases, including multiple sclerosis (MS), Parkinson's disease, cerebral ischemia and spinal cord injuries (Pagenstecher et al. 1998; Noble et al. 2002; Nygardas and Hinkkanen 2002; Rivera et al. 2002; Lorenzl et al. 2003a). The MMPs are regulated, in part, through the expression of endogenous MMP inhibitor proteins, the tissue inhibitors of metalloproteinases (TIMPs). In most neurodegenerative diseases studied, reports have indicated that elevated expression and/or activities of MMPs are also accompanied by altered expression of TIMPs (Pagenstecher et al. 1998; Lorenzl et al. 2003a,b). In animal models of autoimmune-mediated demyelination, abrogation of MMPs by either targeted-gene knockout or through administration of chemical inhibitors can reduce pathology and improve clinical outcome (Gijbels et al. 1994; Clements et al. 1997; Dubois et al. 1999). Comparatively less is known of the precise functions of TIMPs in neurological disease. However, during the symptomatic phase of experimental autoimmune encephalomyelitis (EAE) in mice, expression of TIMP-1 is dramatically up-regulated (Pagenstecher et al. 1998; Teesalu et al. 2001), where it is expressed by astrocytes surrounding inflammatory lesions (Pagenstecher et al. 1998).
The current and prevailing hypothesis regarding metalloproteinase involvement in CNS inflammatory disease is that an overall increase in MMP activity overwhelms the potential endogenous protective response of TIMPs, and the net outcome of this process is represented by the pathology of disease (Yong et al. 1998). This notion of an imbalance in the MMP : TIMP axis underlying disease etiology is supported not only by models of disease but also, by findings that increased serum MMP : TIMP ratios in MS can be reduced by IFN-β therapy (Waubant et al. 2001, 2003). Thus, inflammatory mediators produced during disease likely represent an important process underlying CNS injury by evoking dysregulation of MMP : TIMP homeostasis (Waubant et al. 1999; Mandel et al. 2004). As central players in the responsiveness of the CNS to inflammation, analysis of inflammatory regulation of TIMP and MMP gene expression in glia is therefore central to our understanding of the physiological and disease-associated functions of the TIMP/MMP axis. Thus, the aim of this study was twofold: first, to identify the expression patterns of the MMP and TIMP genes in astrocytes and microglia; and second, to determine what factors can regulate their mRNA levels in these primary glial cells.
Materials and methods
Primary glial cultures
Glial cultures were prepared from post-natal (day 0–2) C57BL/6 mouse pups, as previously described (Milner and Campbell 2002, 2003). Briefly, cerebral cortices were extracted into Dulbecco's modified Eagle's medium (DMEM; Sigma-Aldrich, St. Louis, MO, USA) containing papain, DNAse 1-type IV and l-cysteine, and then incubated for 60 min at 37°C. Fresh medium was then exchanged and tissues were mechanically dissociated. The cells were plated out onto poly d-lysine (Sigma-Aldrich)-coated T75 flasks (Falcon, Franklin Lakes, NJ, USA). Glia were maintained in DMEM medium containing 10% fetal calf serum (Sigma-Aldrich) for 7–10 days and then either plated out onto 6-well plates (Nunc, Naperville, IL, USA) as mixed astrocyte/microglial cultures, or microglia were dissociated and plated separately, as previously described (Milner and Campbell 2002). To characterize cell purity, parallel cultures of mixed glial cells were established on poly d-lysine-coated glass coverslips and immunocytochemistry was performed with glial fibrillary acidic protein (GFAP), a marker for astrocytes, and the microglial marker, Mac-1. The mixed glial cultures prepared in this way were shown to contain greater than 95% astrocytes and between 4 and 5% microglia (Milner and Campbell 2002). Microglial cultures prepared by shaking off the loosely-attached microglial cells yielded cultures which were greater than 99% Mac-1 cells (data not shown). The use of animals in this project was approved by the Animal Ethics Committee, Department of Animal Resources, The Scripps Research Institute.
Cytokines and chemokines
Murine cytokines were purchased from commercial sources and used at concentrations listed in Table 1. The cytokine concentrations assayed in this study were based upon a previous study where we had determined the lowest effective dose mediating the activation of microglia (Milner and Campbell 2003). Reagents were prepared in accordance with the manufacturer's specifications and then diluted to working concentrations in DMEM containing antibiotics, either with or without 10% bovine serum where indicated. For experiments performed under serum-free conditions, cytokines were added to serum-free medium and/or administered to glial cultures following overnight serum withdrawal. Cycloheximide (1 µg/mL; Sigma-Aldrich) was administered in serum-free DMEM for 30 min prior to treatment with interleukin-1β (IL-1β) or lipopolysaccharide (LPS).
Table 1. Summary of cytokines and chemokines used to assess for effects on the expression and/or induction of TIMP and MMP gene expression
Effective dose (ng/ml unless otherwise indicated) of half maximal response in murine cells based on current literature and/or product information provided by the commercial supplier.
n/a, Not applicable.
R & D Systems
R & D Systems
R & D Systems
Tumor necrosis factor-α
RNase protection assay
Multi-probe RNase protection assays (RPA) were used to quantitatively assess changes in the expression of MMP and TIMP genes. RPA probe sets were developed and used as previously described (Pagenstecher et al. 1997), with the additional inclusion of an antisense probe directed against nucleotides 229–498 (GenBank Accession NM_080639) of TIMP-4 mRNA. RPA analyses were performed on total RNA samples (5–10 µg) prepared using TriReagent (Sigma-Aldrich). For analyses, each lane represented an RNA sample derived from individually treated culture wells at the times and/or treatments indicated. RPA gels were visualized using film autoradiography, and densitometry was performed on scanned images of the developed autoradiographs using NIH ImageJ software (version 1.33). Expression was reported as ‘arbitrary units’ that represent the ratio of the intensity of the signal intensity for each gene relative to the respective internal loading control, the gene for the ribosomal protein L32. It was determined that longer exposure times (2–7 days) were required to resolve the expression of MMP and TIMP genes that resulted in saturation of L32 signals. To correct for this, autorad signal intensities for the MMP and TIMP genes were corrected to the L32 band intensities from shorter exposure times (8–12 h) of the same gels.
Measurement of MMP and TIMP mRNA levels by RPA was carried out in triplicate. Statistical analysis of differences between results was performed using a two-way anova followed by Bonferroni tests for pairwise comparison. The null hypothesis was rejected when p < 0.05.
Regulation of TIMP family genes in primary glial cell cultures
RPA analysis of TIMP family gene expression in primary glial cultures revealed that the TIMP-2 mRNA was present at the highest level in the basal state. TIMP-3 mRNA was also found to be abundant, while the level of TIMP-1 mRNA was detectable at very low levels and TIMP-4 mRNA was the lowest in overall abundance (Fig. 1a). Next, we determined whether expression of the TIMP genes could be directly modulated and/or induced by pro-inflammatory factors in primary murine glial cultures. We evaluated a limited panel of agents at doses that we had previously found could induce microglia activation (Milner and Campbell 2003), as indicated in Table 1. The factors we tested included: interleukin (IL)-1β, IL-3, IL-6, interferon (IFN)-α, IFN-γ, tumor necrosis factor (TNF)-α or the bacterial endotoxin, LPS. Primary glial cultures were treated for either 2 h (Figs 1a and c) or 24 h (Figs 1b and c), and RNA was isolated and analyzed by RPA. Under these conditions, we determined that treatment with either IL-1β or LPS, but none of the other cytokines, significantly increased TIMP-1 mRNA at both time points tested. The increase in TIMP-1 mRNA in these glial cultures by IL-1β was also accompanied by a significant reduction in the level of TIMP-3 mRNA (Figs 1b and c). The level of either TIMP-2 (Fig. 1c) or TIMP-4 (data not shown) mRNA was not modified by any of the pro-inflammatory factors at either time point tested.
Regulation of MMP genes in primary glial cell cultures
In view of the notable role of TIMPs as endogenous regulators of MMP activities we also sought to determine, in parallel with the TIMP genes, the pattern and regulation of MMP genes expressed by the cultured glial cells. Analysis of MMP gene expression in primary glial cell cultures revealed constitutive expression of the MMP-11, -12 and -14 mRNA transcripts (Fig. 2). Exposure to IL-1β had markedly increased the level of MMP-3 mRNA by 2 h (Figs 2a and b) that was sustained for 24 h (Figs 2b and d). In contrast, although administration of IFN-α or, to a lesser degree, IFN-γ had increased the level of MMP-13 mRNA by 2 h, this effect was short-lived and not sustained through the 24 h time point (Fig. 2d). Of the agents we had selected to treat the primary glial cultures, administration of LPS produced the most dramatic differences in the level of the different MMP mRNAs, both in terms of the number altered and their extent of change. A 2 h exposure to LPS significantly increased the level of the mRNA for MMPs -3, -9, -13 and -14 (Figs 2a and b). Longer (24 h) exposure to LPS resulted in profoundly increased levels of the MMP-3, -9, -12, -13, and -14 mRNA transcripts (Figs 2c and d). Treatment with LPS did not effect basal expression of the MMP-11 gene, neither did treatment of primary glial cultures with IL-3, IL-6 or TNF-α influence the levels of any MMP mRNAs at the doses tested or the time points examined.
Differential expression of TIMP genes in astrocytes and microglia
Primary glial cultures as prepared herein are a heterogeneous population of astrocytes and microglia comprising approximately 95% astrocytes and 5% microglia (Milner and Campbell 2002). While astrocytes represented the most abundant cell type in this culture system, the potential involvement of microglia cannot be overlooked. Therefore, to identify which cell type(s) were expressing TIMP mRNAs, we next analyzed RNA derived from primary mixed cultures and purified microglial cultures devoid of astrocytes, treated with either IL-1β or LPS. As shown in Fig. 3(a), cultures enriched for astrocytes expressed TIMP-1, -2 and -3 mRNAs and exhibited increased levels of TIMP-1 mRNA in response to IL-1β or LPS stimulation (Figs 3a and b). By contrast, purified cultured microglia expressed TIMP-2 mRNA, but not TIMP-1 or TIMP-3 mRNAs, and the level of TIMP-2 mRNA was unaffected by administration of IL-1β (Figs 3a and c).
MMP genes were also found to be differentially expressed by astrocyte-enriched versus purified microglial cells (Fig. 3d–f). Astrocyte-enriched cultures were found to basally express MMP-11, -12, -3 and -14 mRNAs (Figs 3d and e), whereas MMP-12 mRNA was the most abundantly expressed MMP gene in the microglia cultures (Figs 3d and f). Next, we determined whether the astrocyte or microglia responded differently to treatment with IL-1β or LPS. Astrocytes treated with IL-1β exhibited a modest increase in MMP-3 mRNA and a significant reduction in MMP-12 mRNA, while LPS induced expression of the MMP-12, -3 and -13 mRNAs (Fig. 3e). Whereas expression of MMP-3 mRNA was induced in astrocytes by LPS or IL-1β, similar treatment of microglial cultures did not increase the levels of this MMP mRNA (Fig. 3f). However, microglia were found to have robust expression of MMP-13 mRNA in response to LPS, but not IL-1β, treatment at this time point. Expression of MMP-12 mRNA in microglia was not modulated by either LPS or IL-1β (Fig. 3f).
Characterization of TIMP-1 gene expression by astrocytes
Analysis of the time course of LPS-stimulated TIMP-1 gene expression in mixed glial cultures revealed that increased TIMP-1 mRNA was initiated within 2 h but was also sustained through to the 24 h experimental endpoint (Fig. 4a). As LPS is known to induce cytokines, we next determined whether administration of LPS to the mixed glial culture also induced cytokines and, specifically, IL-1β. RPA analysis indicated that LPS treatment of primary mixed glial cultures resulted in the increased levels of four cytokine mRNA transcripts: IL-1β, IL-6, IL-1α and TNF-α (Fig. 4b). While the increased levels of the TNF-α, IL-6 and IL-1α mRNAs were transient and returned to basal levels within 18 h of LPS treatment, IL-1β mRNA levels remained elevated throughout.
To characterize further the selective nature of increased TIMP-1 gene expression mediated by IL-1β, we examined the dose–response relationship and time course of changes of TIMP mRNAs in the mixed glial cultures treated with IL-1β. Cultures treated with a range of IL-1β concentrations revealed a significant dose–response effect on TIMP-1 mRNA levels, with little induction at lower doses (0.01–1 ng/mL) but progressively increased levels of TIMP-1 mRNA at higher doses (1–100 ng/mL) (Fig. 5a). In response to a single dose of IL-1β (10 ng/mL), TIMP-1 mRNA levels increased within 2 h and were sustained up to 18 h, but they had returned to basal levels by 24 h (Fig. 5b).
Interestingly, and consistent with our results from the screen of potential inducers of TIMP genes above, IL-1β stimulation of TIMP-1 mRNA in primary glial cultures was coincident with a reduction in expression of TIMP-3 mRNA (Fig. 5b). Of further note, and similar to the observed inverse temporal regulation of the TIMP-1 and -3 genes with direct administration of IL-1β (10 ng/mL), exposure to LPS (20 ng/mL) also resulted in a marked reduction in TIMP-3 mRNA levels that was temporally and proportionally opposite to the levels of TIMP-1 mRNA (Fig. 4a).
Modulation of MMP and TIMP genes by serum
In terms of the responsiveness to neuroinflammation in vivo, we next evaluated the impact of serum in the culture medium as a potential factor that might influence the expression of either MMP or TIMP genes. In vivo, astrocytes would only be exposed to serum components under conditions resulting from a breach or breakdown in the integrity of the blood–brain barrier. Accordingly, we investigated the influence of serum on MMP and TIMP gene expression by determining whether the re-introduction of serum following overnight withdrawal could alter the expression of these genes. Exposure of primary glial cultures to serum (10%) following an overnight serum withdrawal resulted in a diminution of MMP-11 and -12 mRNA and did not increase the levels of any MMP mRNAs (Figs 6a and b). In contrast, analysis of TIMP gene expression by following re-introduction of serum (10%) revealed a marked increase in TIMP-1 mRNA levels without measurable changes in the level of the other TIMP mRNAs (Fig. 6c). When compared with basal levels of TIMP-1 mRNA in the absence of serum, re-introduction of serum resulted in a significant increase in mRNA levels (p < 0.01; Fig. 6d). In contrast to the counter-regulation of the TIMPs -1 and -3 mRNAs by IL-1β or LPS, re-introduction of serum did not reduce the TIMP-3 mRNA.
The finding that serum was sufficient to induce TIMP-1 gene expression suggested that there might be an important influence of serum on the regulation of TIMP-1 mRNA by IL-1β. To test whether the presence of serum modified the levels of TIMP-1 mRNA in response to IL-1β, primary mixed glial cultures were treated with either 1 or 10 ng/mL IL-1β concomitant with the re-introduction of serum following overnight serum withdrawal (Fig. 6c). In contrast to treatment with IL-1β alone, the level of the TIMP-1 mRNA was potentiated by exposure to serum (10%), albeit that this effect was pronounced only at the higher concentration tested (10 ng/mL, IL-1β; p < 0.001; Fig. 6c).
Induction of TIMP-1 gene expression by LPS or IL-1β requires de novo protein synthesis
The temporal profile of astrocyte TIMP-1 mRNA changes in response to either IL-1β or LPS indicated that there was a delay in the time to induction. As an initial step towards better understanding the mechanisms of transcriptional regulation of the TIMP-1 gene in astrocytes, we next determined whether the increase of TIMP-1 mRNA levels in response to IL-1β or LPS required de novo protein synthesis by administering the general protein synthesis inhibitor, cycloheximide (CHX), at the time of treatment of the glial cultures with either IL-1β (10 ng/mL) or LPS (20 ng/mL) under serum-free conditions (Fig. 7). Analysis of TIMP gene family mRNAs revealed that the increased TIMP-1 mRNA induced by either IL-1β or LPS, was completely blocked by CHX treatment (Fig. 7).
Regulation and potential physiological roles of glial-derived TIMPs
To our knowledge, our report here is the first to examine the co-ordinated response of multiple members of the TIMP and MMP gene families, with the demonstration that there is differential expression of these genes by astrocytes versus microglia. Previous studies have reported that the expression of TIMPs and/or MMPs can be induced by a variety of inflammatory factors in a wide range of cell types (Gottschall and Yu 1995; Bugno et al. 1999; Muir et al. 2002; Suryadevara et al. 2003). We found that the TIMP-1 and -3 genes were expressed solely by astrocytes, while only the TIMP-2 gene was expressed by microglia. We also identified similar distinctions among the expression of TIMP family genes in response to cytokines. On the basis of our findings presented here, we conclude that the regulation of TIMP genes in primary glial cultures by cytokines is very specific both in terms of the stimulating signal and the responding cell type.
In response to either IL-1β or LPS treatment, we determined that astrocytes significantly increased TIMP-1 mRNA levels. The absence of any influence of IL-3, IL-6, IFN-α, IFN-γ or TNF-α on TIMP gene expression indicated that the specificity of this response is unique to IL-1β and LPS. The basis for the increased and sustained expression of the TIMP-1 mRNA in the presence of IL-1 (and LPS) is currently unknown but may reflect either continued transcription of the TIMP-1 gene in the presence of the stimulus and/or decreased TIMP-1 mRNA degradation. Further experiments will be required to clarify this mechanism.
Our findings also indicate that there is an inverse regulation of TIMP-1 and TIMP-3 mRNAs in response to IL-1β in astrocytes, but not microglia. This observation is consistent with work by others (Bugno et al. 1999; Suryadevara et al. 2003), although an inverse relationship between TIMP-1 and -3 gene expression had only been reported previously in response to TNF-α in combination with IL-1β (Bugno et al. 1999). It is of particular interest, with regard to a potential significance of this counter-regulation of the TIMP-1 and -3 mRNAs in astrocytes, that there are several instances where the net physiological effects of TIMP-1 and TIMP-3 are known to be antagonistic, even though TIMP proteins have seemingly redundant roles in the regulation of metalloproteinase (MMP) activities (Crocker et al. 2004). For instance, expression of TIMP-1 has been shown to prevent glutamate excitotoxicity of cultured neuronal cells while in contrast, enhanced expression of TIMP-3 can sensitize cells to toxic stimuli (Bond et al. 2002; Ahonen et al. 2003; Tan et al. 2003). TIMP-3 has also been reported to antagonize the protective effects of virus-expressed TIMP-2 in a mouse model of demyelination (Nygardas et al. 2004).
We found that serum increased TIMP-1 mRNA levels and modulated the effect of IL-1β on TIMP-1 mRNA in astrocytes. This suggested that in vivo, serum may represent an early danger signal that induces TIMP-1 gene expression by astrocytes as part of an attempt to alleviate further destruction of the BBB, or the CNS parenchyma, by MMPs (La Fleur et al. 1996; Mun-Bryce and Rosenberg 1998; Rosenberg 2002). As such, it is proposed that serum-induction of the TIMP-1 gene expression may play a fundamental role in preventing excessive BBB breakdown during neuroinflammatory conditions and thus, may contain the spread of tissue injury. Previous work has reported evidence for increased permeability of the BBB in MS, and for the presence of serum proteins within the brain parenchyma of post-mortem MS tissues in all clinical stages of MS (Vos et al. 2005). As astrocytes comprise an ultrastructural basement membrane that surrounds blood vessels in the brain, our findings suggest that serum proteins represent an important signal in the regulation of TIMP gene expression in neuroinflammatory conditions.
The lack of TIMP gene expression in response to IFN-α or IFN-γ in our primary glial cultures was unexpected as increased expression of TIMP-1 is a consistent feature of CNS viral infections (Crocker and Campbell 2005). However, our results are in agreement with earlier work by others demonstrating that IFN-γ is not sufficient to evoke TIMP-1 mRNA or protein expression in astrocytes (Bugno et al. 1999; Suryadevara et al. 2003; Leveque et al. 2004). Therefore, we would propose that components of the innate immune response, such as IL-1β, as well as serum-derived products from the breakdown of the BBB that accompanies many viral infections of the CNS (Soilu-Hanninen et al. 1994; Stephens et al. 2003), are the primary factors responsible for the localized and early regulation of TIMP-1 gene expression by glia following viral infections. Our in vitro findings are consistent with the results of previous work where a tight spatial and temporal expression of the TIMP-1 gene by astrocytes surrounding inflammatory lesions was observed in the spinal cord of EAE mice (Pagenstecher et al. 1998). This correlation led us to determine which inflammatory factors, such as pro-inflammatory cytokines released from infiltrating immune cells or serum entering the brain following breakdown of the BBB, are directly related to the regulation of the glial TIMP-1 gene during CNS inflammation. Together, our results support an emerging hypothesis that a primary role for astrocyte-derived TIMP-1 during CNS inflammatory attack is to minimize the spread of immune cells into the CNS parenchyma (Pagenstecher et al. 1998). In addition, these data provide strong evidence to support the use of TIMP-1 as a marker of tissue injury or immune stress affecting the CNS.
It is important to consider that while the best characterized role of TIMP-1 is its capacity to function as an endogenous MMP inhibiting protein, TIMP-1 was first identified as a trophic factor for erythrocytes (Gasson et al. 1985; Hayakawa et al. 1992; Chesler et al. 1995). Indeed, studies have shown that TIMP-1 can act independently of blocking MMPs, for example, by stimulating cellular proliferation (Chesler et al. 1995) or preventing excitotoxic neuronal injury (Tan et al. 2003), thereby suggesting that the biological actions of TIMP-1 may extend beyond its action as an MMP inhibitor (Crocker et al. 2004).
Regulation and potential physiological roles of glial-derived MMPs
Analogous to our examination of TIMP gene regulation, we also found that the MMP genes are differentially expressed by astrocytes or microglia in response to IL-1β or LPS. Whereas astrocytes expressed the genes for MMPs -3 and -11, microglia expressed the genes for MMPs -12 and -13. This differential expression of MMP genes in astrocytes or microglia was also found to reflect the significant segregation in the expression of MMP mRNAs in response to the cytokines tested in this study. Our findings are consistent with previously reported work describing robust microglial responsiveness to LPS, but limited direct stimulation by IL-1β (Pinteaux et al. 2002). These data would also indicate that microglia play a more central role in the expression of MMP genes during inflammation that may also reflect their physiological role as endogenous CNS macrophages (Raivich and Banati 2004). An important difference between our results and those previously reported was the lack of MMP gene expression changes in response to TNF (Muir et al. 2002; Leveque et al. 2004). While we did observe a modest decrease in MMP-2 gene expression in response to TNF treatment (data not shown), as also reported by others (Qin et al. 1998), the reasons underlying these differences are presently unclear.
MMPs have been implicated in a variety of neuroinflammatory conditions, including EAE (Pagenstecher et al. 1998; Nygardas and Hinkkanen 2002), multiple sclerosis (Gijbels et al. 1994), Parkinson's disease (Lorenzl et al. 2003b), cerebral ischemia (Cunningham et al. 2005) and nerve regeneration (Larsen et al. 2003). In many instances, the activation of MMPs is associated with tissue injury and cell death. For example, MMP-3 has recently been implicated in the neuroinflammatory cascade as a secreted factor released from apoptotic cells (Kim et al. 2005). In their study, Kim and co-workers showed that active MMP-3 increased expression of cytokines that resulted in the activation of microglia, accounting for the localized activation of microglia around apoptotic neurons in the CNS (Kim et al. 2005). Elevated MMP-3 levels have been reported in several autoimmune and inflammatory CNS disease states (Maeda and Sobel 1996; Keyszer et al. 1999; Leppert et al. 1999; D'Souza et al. 2002). As an extension of these findings, our results would implicate IL-1β as an important stimulus for astrocytic production of MMP-3, which may then serve as a potential activator of microglia. In addition to MMP-3, in our primary glial cultures, we observed increased MMP-13 mRNA levels in response to IFN-α and a potent induction in microglia following LPS exposure. At this time, the function and exact contribution of MMP-13 gene expression by glia during neuroinflammatory and neurodegenerative disease is not clear and warrants further study.
Another finding of our study was the expression and regulation of MMP-9 mRNA by astrocytes. Increased levels of MMP-9 protein have been reported in the cerebrospinal fluid of a variety of human neurodegenerative and neuroinflammatory conditions (Lorenzl et al. 2003b). In an animal model of demyelination, expression of MMP-9 protein is correlated with increased disease severity (Nygardas and Hinkkanen 2002; Esparza et al. 2004). In our study, we determined that MMP-9 mRNA expression in astrocytes was increased by the same stimuli that induced expression of TIMP-1 gene expression. Given that TIMP-1 mRNA is also expressed during the peak clinical illness in EAE (Pagenstecher et al. 1998), and protein levels of both MMP-9 and TIMP-1 have been associated with clinical relapse in human MS (Waubant et al. 1999), the expression of MMP-9 and TIMP-1 mRNAs by LPS in primary glia likely reflects a mode for dual regulation of both enzyme and inhibitor. The physiological importance of this phenomenon is presently unclear. Nevertheless, it provides an example of both enzyme and inhibitor being induced by the same stimulus in the same cell types.
Glial regulation of TIMP and MMP genes by IL-1β and LPS
We hypothesize that glial expression of the genes for TIMPs -1 and -2 with MMPs -3, -9, -12 and -13 are regulated by both direct and indirect signals (Fig. 8). In this scheme, we propose that the reported low abundance of LPS receptor on astrocytes (Jack et al. 2005) indicates that LPS directly stimulates the expression of TIMP-2, MMP-12 and MMP-13 genes by microglia. By extension, the stimulation of TIMP and MMP genes in astrocytes by LPS is indirect, and mediated by the secretion of LPS-induced IL-1β (see Fig. 4 and Chauvet et al. 2001). Accordingly, this model would predict that IL-1β derived from infiltrating cell types and/or microglia acts directly on astrocytes (via IL-1 receptors; Pinteaux et al. 2002) to stimulate the expression of the TIMP and MMP genes such as TIMP-1, MMP-3 and MMP-9. Similarly, in the course of neuroinflammation in vivo, circulating factors derived from serum leaching through a destabilized BBB could also stimulate and/or enhance IL-1β-mediated induction of TIMP-1 mRNA in astrocytes. It is important to note that the MMP and TIMP genes expressed by glia in response to LPS in vitro were similar to those found altered by LPS in endotoxemia-sensitive mice (Pagenstecher et al. 2000a). This suggests that our proposed model of MMP and TIMP gene regulation in primary glial cell cultures recapitulates important aspects of their regulation during inflammation of the CNS.
In conclusion, our findings reveal agonist- and cell-specific regulatory control of TIMP and MMP gene expression by glial cells and suggest that specific ‘danger’ signals govern the highly selective and differential expression of TIMP-1 and MMP genes found in neuroinflammatory conditions of the CNS. Future studies will be required to identify the intracellular signals mediating IL-1β- and LPS-induced TIMP and MMP gene expression in glia, and to determine whether the increases in mRNA for these genes are reflected by changes in corresponding protein levels. Knowledge of the expression and regulation of the MMP and TIMP genes by inflammatory stimuli in vitro may lead to a better understanding of the factors governing their expression in vivo and thus, may potentially identify novel strategies for modulating TIMP or MMP gene expression in neuroinflammation and neurodegenerative diseases.
We thank Ricardo Frausto for expert technical assistance. This work was supported by grants from the National Institutes of Health (MH62231, NS36979 and MH62261 to ILC). SJC is the recipient of an Advanced Postdoctoral Fellowship from the National Multiple Sclerosis Society FA 1552-A-1 (USA). RM was the recipient of a Wellcome Trust International Traveling Fellowship. This is manuscript #NP-17564 from The Scripps Research Institute.