The Expression of Matrix Metalloproteinase-12 by Oligodendrocytes Regulates Their Maturation and Morphological Differentiation

Introduction

The levels of several matrix metalloproteinases (MMPs) are increased in pathology, including tumorigenesis, stroke, multiple sclerosis (MS), and inflammation in which they are thought to have detrimental effects. Hardly appreciated is that such MMP expression may serve beneficial functions. We demonstrated that the elevation of MMP-9 after a demyelinating injury to the adult spinal cord is required for subsequent remyelination by virtue of removing the NG2 proteoglycan that inhibited the maturation of oligodendrocyte precursor cells (OPCs) (Larsen et al., 2003). MMP activity is required for neurite extension (Zuo et al., 1998; Webber et al., 2002) in which the mechanism may similarly involve the MMP-mediated degradation of inhibitory proteoglycans (Zuo et al., 1998). Further supporting the concept of beneficial properties of MMPs is the knowledge that many MMP members are elevated during CNS development (Canete Soler et al., 1995a,b; Vaillant et al., 1999; Sekine-Aizawa et al., 2001). Understanding the role of these proteases during development may shed light on the beneficial components of MMPs required to achieve recovery from pathological conditions.

Previous studies from our laboratory have implicated MMPs in the biology of oligodendrocytes (OLs), which are the cells that extend processes in the CNS to contact axons to form myelin. MMP-9 was found to mediate OL process extension in vitro, and MMP-9 expression correlated with myelination in vivo (Uhm et al., 1998; Oh et al., 1999). In correspondence, MMP-9 removed injury-induced deposition of inhibitory NG2 proteoglycan to allow OPCs to mature into myelinating OLs (Larsen et al., 2003). Because the MMPs are a large group of proteinases capable of processing similar substrates, we asked the question of whether other MMP members might be playing a role in OL biology in addition to MMP-9.

The current study focuses on MMP-12 and its role in OL biology. MMP-12, also known as macrophage metalloelastase, has most often been studied in vascular diseases and lung inflammation in which an increase in MMP-12 has been associated with disease progression (Matsumoto et al., 1998; Annabi et al., 2002). MMP-12 has also been shown to play a role in macrophage infiltration after emphysema (Shipley et al., 1996) and leukocyte and eosinophil recruitment to the lung (Lanone et al., 2002). With respect to the CNS, the literature on MMP-12 is sparse. A study by Vos et al. (2003) reported that MMP-12 is present in active MS lesions in which its expression could be localized to phagocytic macrophages. Although these studies indicate that cells of monocytoid lineage (i.e., macrophages and microglia) are responsible for MMP-12 expression, we demonstrate here that OLs express this protein in large amounts. Furthermore, we show that MMP-12 has functional roles for both the maturation and morphological differentiation (process extension) of cells of the OL lineage.

Materials and Methods

Establishment of cultures of OLs, astrocytes, and microglia. For most experiments, mouse OLs obtained from 3-week-old 129/SvEv wild-type or MMP-12 null mice were used. In the series involving various monoclonal antibodies to MMP-12 (see Figs. 3, 4), human OLs were used because these MMP-12 antibodies were generated to human determinants and their cross-reactivity to murine cells was unclear. Human OLs were derived from surgical resection of adult intractable epilepsies; tissue samples contained areas adjacent to, but not containing, the epileptic focus.

For murine and human samples, OLs were purified as described previously (Oh et al., 1999). Briefly, cells were dissociated by trypsin digestion and isolated by Percoll density centrifugation. The initial cell isolates were plated in T-25 cm² flasks overnight. The floating cells (OLs) were then isolated and plated onto poly-l-ornithine-coated 16-well Lab-Tek chamber slides (Nunc, Naperville, IL) or in six-well chambers. OLs were maintained in Minimal Essential Medium (MEM) containing 5% fetal calf serum, 0.1% dextrose, 10 μm glutamine, and 1 mm sodium pyruvate. Using this protocol, the purity of adult human OLs was routinely in excess of 90% as determined using an antibody (O1 antibody) (Sommer and Schachner, 1981) and immunohistochemistry to an OL-specific glycolipid, galactocerebroside. The purity of mouse OLs was between 80 and 90%.

We reported previously that human OLs initiated processes very slowly in culture in comparison to the robust growth properties of murine OLs (Yong et al., 1991; Oh et al., 1999). Thus, to determine whether process extension from the human OLs could be blocked by MMP-12 antibodies, cells were treated with phorbol-12-myristate-13-acetate (PMA) (10 ng/ml) to initiate OL process extension. In parallel, mouse OLs were similarly treated with PMA for the RNase protection assay (RPA) experiments (see Figs. 1, 2).

Neonatal mouse astrocytes were obtained as described previously (Corley et al., 2001). In brief, brain tissues were dissociated using trypsin digestion and then seeded into poly-l-ornithine (10 μg/ml)-coated 100 mm culture dishes at an approximate density of 5 × 10⁶ per dish. After 7-10 d, which provided for a rapid expansion of the astrocyte population, the purity of astrocytes was assessed to be over 95% as determined using an antibody to glial fibrillary acidic protein.

Neonatal mouse microglia were isolated similarly to the OLs described above. However, microglia attached to the flasks unlike the floating OLs, and they were purified by the removal of other (floating) cells. The purity of mouse microglia as determined by immunohistochemistry to ionized calcium-binding adaptor molecule 1 (IBA1) (Chemicon, Temecula, CA) is routinely in excess of 90%.

OPC purification. Mouse OPCs were purified by immunomagnetic sorting using the A2B5 antibody and a secondary IgM antibody linked to magnetic beads (MACS; Miltenyi Biotech, Auburn, CA). Briefly, brains were removed from 3-d-old 129/SvEv mice, and the cells were dissociated by 0.25% trypsin treatment (20 min). The cells were then separated from debris using Percoll density centrifugation (Oh et al., 1999) and incubated with the A2B5 monoclonal antibody (Eisenbarth et al., 1979) for 1 hr at 4°C, followed by incubation with bead-coated anti-mouse IgM for 30 min at 4°C. Finally, the cells were purified by magnetic sorting according to the instructions of the manufacturer.

OPCs were also cultured from MMP-12 null mice (Shipley et al., 1996) obtained from Dr. Steve Shapiro (Harvard Medical School, Boston, MA). These were similarly isolated on the 129/SvEv background.

OPC maturation and differentiation assay. Purified OPCs from wild-type or MMP-12 null mice were plated onto 16-well Lab-Tek chamber slides coated with poly-l-ornithine at a density of 50,000 cells per well. After cell plating, the A2B5 precursors were allowed to mature for 48 hr at 37°C in a 5% CO₂ environment. Feeding medium consisted of an OPC differentiation medium (DMEM-F-12) containing N2 supplement (Invitrogen, San Diego, CA), 100 μg/ml bovine serum albumin, 30 ng/ml T₃ (triiodothyronine), 30 ng/ml T₄ (thyroxine), 10 nm d-biotin, 10 ng/ml platelet-derived growth factor, and 0.5% fetal calf serum (FCS). The number of galactocerebroside-positive OLs (Oh et al., 1999) present after 48 hr in each condition was obtained by counting the total number of labeled (O1⁺) cells within each well (n = 4 wells per condition in each experiment; experiments were repeated three times). To assess whether there still were similar numbers of cells after 48 hr of culture for both genotypes, the total number of cells was counted from eight random fields using Hoechst dye as a nuclear marker.

We also assessed whether OPCs and OLs from wild-type and MMP-12 null mice were morphologically different. Cells were labeled with Ranscht mouse monoclonal antibody (RmAb) (Ranscht et al., 1982; Bansal et al., 1989), which detects both immature and mature cells of the OL lineage. This allowed for characterization of OPCs and OLs at different stages of process extension. Cells were scored using four different morphological classes (see Fig. 6), as follows: low, in which an RmAb-positive cell has few or short processes radiating from soma; low-med (medium), whereby a cell has three to six processes up to a length of three cell soma with minimal interconnection; med-high, indicating a cell with over six processes longer than three cell soma with clear interconnections; and high, whereby there are many processes with extensive interconnections all extending far from the cell soma in all directions. These studies were conducted by counting the entire well of three wells per condition in each experiment; experiments were repeated three times. All morphological examinations were performed under blinded conditions.

Process elongation from human OLs. Purified adult human OLs were used to investigate process extension in the presence of monoclonal antibodies raised to different human MMP-12 domains (all from Chemicon). OLs were treated with antibodies (10 μg/ml) to the N-terminal catalytic (N-term), propeptide, C-terminal (C-term), or hinge regions (Chemicon). MMP-12 proenzyme (54 kDa) is released extracellularly from cells and consists of four domains (propeptide, catalytic, hinge, and C-terminal). The activation process involves the removal of the propeptide, rendering an intermediate form (Inter) (45 kDa). The C-terminal domain is then cleaved by autocatalysis leaving fully active MMP-12 (∼22 kDa) consisting only of the catalytic N-terminal domain and part of the hinge domain. A function-blocking antibody to MMP-9 (Cambio, Cambridge, UK) was used in one set of experiments.

Human OLs were first treated with individual antibodies, and process outgrowth was stimulated with PMA (10 ng/ml) 3 hr later. As noted previously, rodent OLs can extend processes readily, whereas human OLs require PMA treatment to prompt their morphological differentiation. Process extension was analyzed after 48 hr in culture by counting the percentage of OLs with processes extended more than three cell soma in length. Five views using a 40× microscope objective were analyzed for each well in three separate wells for each condition. Experiments were repeated three times.

RPA. RNA from purified murine OLs was obtained by applying 1 ml of Trizol reagent (Invitrogen) to each well. Total RNA was extracted according to the instructions of the manufacturer, and 10 μg were used for each lane. To analyze the expression of various MMP transcripts, we used a mouse MMP multiprobe set courtesy of Dr. Iain Campbell (Scripps Research Institute, La Jolla, CA) (Pagenstecher et al., 1998). RPA was performed as described by the manufacturer (PharMingen, San Diego, CA). In brief, RNA samples were hybridized to radioactive (α-³³P-UTP) labeled antisense probe overnight at 56°C. After 1.5 hr of digestion of unhybridized probe with an RNase A and T1 mix (PharMingen), the protected fragments were treated with proteinase K (10 mg/ml) and extracted with phenolchloroform. Samples were separated on a 6% polyacrylamide gel and analyzed by phosphor imaging (Molecular Dynamics, Sunnyvale, CA).

Casein zymography of conditioned medium from peritoneal macrophages. Peritoneal macrophages were obtained from mice that had been injected with 1 ml of 3% thioglycollate medium (Sigma, St. Louis, MO) into the peritoneum. After 3 d, mice were killed and immediately injected with 7 ml of MEM containing 5% FCS into the peritoneum. The medium now containing macrophages was recovered from the peritoneum and plated in a 100 mm dish. Macrophages from five animals were pooled for this experiment. Once the cells attached onto the culture dish, the medium was changed to MEM containing 0.5% FCS and N2 supplement (Invitrogen). The cells were stimulated with lipopolysaccharide (500 ng/ml) and incubated for 48 hr.

Conditioned medium was collected from cultured peritoneal macrophages from either wild-type or MMP-12 null mice. The medium was concentrated 100 times (Centricon; Millipore, Bedford, MA), and samples were separated on a 12% SDS-polyacrylamide gel containing 1 mg/ml casein (Worthington, Lakewood, NJ). After separation, the gel was incubated for 6 hr on a shaker at room temperature with rinse buffer (2.5% Triton X-100, 50 mm Tris-HCl, pH 7.5, and 5 mm CaCl₂) to extract the SDS, allowing proteins to renature within the gel. The gel was then incubated in incubation buffer (50 mm Tris-HCl, pH 7.5, 5 mm CaCl₂, and 1.25 μm ZnCl₂) for 36-48 hr at room temperature to allow in-gel proteolytic degradation of casein. Subsequently, the gel was stained with Coomassie blue for 4 hr and destained (10% acetic acid and 30% methanol) until areas of proteolysis could be visualized.

Western blot analysis. Conditioned medium from human OLs was collected after 24 hr in culture and incubated with MMP-12 antibody bound to microbeads. Purification was performed according to the instructions of the manufacturer (MMP-12 purification kit; R & D Systems, Minneapolis, MN). Purified protein samples were electrophoresed on 12% SDS-PAGE under reducing conditions and transferred to a polyvinylidene difluoride membrane (Millipore), which was probed for 1 hr with a rabbit anti-MMP-12 antibody [MMP-12 (hinge); 1:3000; Chemicon]. This antibody recognizes the pro (Pro) form, the Inter form, and the active (Act) form of MMP-12. A secondary antibody (anti-rabbit HRP) was added for 1 hr, and blots were detected by enhanced chemiluminescence detection kit (Amersham Biosciences, Piscataway, NJ).

Statistical analysis. Statistical evaluation of the effect of MMP-12 function-blocking antibodies on OL process extension was performed by ANOVA with Dunnett multiple comparisons (see Fig. 3). An unpaired Student's t test was applied to compare the means between mature OLs from wild-type and MMP-12 null mice (see Fig. 5). Similarly, Student's t test was used to evaluate the morphology of OL lineage cells between wild-type and MMP-12 null mice at each classification of morphological differentiation (see Fig. 6). Error bars are all shown as SEM.

Results

Discussion

This work has addressed the functional importance of MMP-12 activity for OL development. In recent years, metalloproteinase activity has been correlated with many aspects of development. Examples include the observations that angiogenesis (Canete Soler et al., 1995a; Vu et al., 1998; Werb et al., 1999) and the migration of neuronal precursor cells and OPCs (Amberger et al., 1997; Vaillant et al., 2003) seem to require metalloproteinase activity. A growing list has ascribed novel functions of metalloproteinases in normal development, such as in cytokine activation-inactivation and controlling apoptotic pathways (for review, see Sternlicht and Werb, 2001). These findings have left us with a more complex picture of the functional roles metalloproteinases have in different cell types during developmental processes.

In this study, we investigated the role of MMP-12 in OL biology, prompted by the finding that MMP-12 transcripts were produced in high amounts by OLs in culture (Fig. 1). We also found MMP-12 in microglia, which is not surprising because MMP-12 is reported in cells of monocytoid lineage (Shapiro et al., 1993). Indeed, this is the first description of MMP-12 in OLs and one of the very few to place MMP-12 expression outside of cells of the monocytoid lineage.

We showed previously that MMP-9 is important for OL process extension and myelin development, and the finding that MMP-12 is expressed by OLs prompted us to investigate similar functions for MMP-12. We first tested the effects of function-blocking antibodies to MMP-12 on OL process outgrowth. Three antibodies targeted against different domains of MMP-12 demonstrated function-blocking effects, and PMA-induced OL process extension was significantly reduced (Fig. 3). A fourth antibody raised against the hinge region did not show any blocking effects. One possible explanation for the various inhibitory effects between the different MMP-12 antibodies may be the different domains to which they bind. It is likely that the antibody binding to the hinge region of MMP-12 (which is not a true part of the catalytic domain) did not interfere with the activity of this enzyme, whereas antibodies interacting at the catalytic site and propeptide regions hindered correct activation of this enzyme, resulting in the observed functional blockage of OL process extension. An antibody to the C terminus of MMP-12 also reduced process extension, which could be the result of interfering with the binding of MMP-12 to its substrate because C termini of MMPs are involved in substrate binding (Overall, 2002).

Additional evidence for MMP-12 in process extension comes from a blinded analysis of OL lineage cells with different morphological complexities. We found significantly fewer OL lineage cells with highly branched processes in the MMP-12 null cultures compared with wild-type cells (Fig. 6B). In correspondence, blocking MMP-12 of wild-type cells reduced their differentiation, whereas MMP-12 added exogenously to MMP-12 null cells enabled morphological differentiation (Fig. 6C,D). Future experiments will determine whether manipulating MMP-9 also have similar effects.

We also investigated a role for MMP-12 in OL maturation by examining the number of mature O1⁺ OLs present after 48 hr in A2B5-sorted cultures (Fig. 5). There was an ∼40% reduction in the number of O1⁺ cells in the MMP-12 null cultures compared with wild-type ones, suggesting that MMP-12 plays a role in the timely maturation of OLs.

Although the substrate that is processed by MMP-12 to promote differentiation and maturation of OLs is not known, possible candidates include neuregulin, insulin-like growth factor-1 (IGF-1), or IGF binding proteins and thyroid hormone, which are all involved in OPC development (D'Ercole et al., 1996; Rodriguez-Pena, 1999; Fernandez et al., 2000; Kuhl et al., 2003). Many other molecular pathways have been described to be involved in OPC maturation and for myelin formation. For example, the Notch pathway has been found to both inhibit and promote OPC maturation depending on the activating ligand for Notch (Wang et al., 1998; Hu et al., 2003). Two metalloproteinases, ADAM (a disintegrin and metalloprotease) -10 and -17, have been proposed to cleave Notch (Pan and Rubin, 1997; Brou et al., 2000). Thus, it is possible that MMP-12, in concert with other metalloproteinases, could be responsible for regulating the activity of the Notch pathway in OL biology.

MMP-12 is able to cleave a variety of extracellular molecules and growth factors in vitro, and this may also account for the MMP-12 activity in OPC differentiation and maturation. Additional investigation of the role and substrates of MMP-12 in developmental myelination is required.

Previous reports have mostly focused on the role of MMP-12 in macrophages in which it is highly expressed. This study shows for the first time that MMP-12 is expressed in CNS glia cells in which it plays a role in OL maturation and process extension. The lack of MMP-12 could have profound effects on developmental myelination if the maturation of OLs is delayed. However, it is possible that other metalloproteinases have redundant effects and can substitute for the lack of MMP-12 activity as a result of their broad substrate specificity. In this regard, MMP-9 has been implicated previously in OL process extension and increases early during developmental myelination (Uhm et al., 1998; Oh et al., 1999). The finding that MMP-12 null OLs still mature, although in lower numbers, suggests that MMP-12 is not the only MMP member capable of driving this event.

In conclusion, we have identified an MMP member that is being highly expressed in OLs. With respect to a functional role, this study implicates MMP-12 activity in the control of morphological differentiation (process extension) and maturation of cells of the OL lineage.