The up-regulation of stromelysin-1 (MMP-3) in a spontaneously demyelinating transgenic mouse precedes onset of disease.

The matrix metalloproteinases (MMPs) are a family of endoproteinases that degrade various components of the extracellular matrix and have been implicated in the pathogenesis of multiple sclerosis. To determine whether up-regulation of MMP-3, or stromelysin-1, was a causative factor during the development of demyelination, we have examined the expression of MMP-3 mRNA and protein in brain tissue of a spontaneously demyelinating mouse model overexpressing DM20 (ND4 line) prior to and during the progression of disease. Stromelysin-1, but not other MMP mRNA was elevated approximately 10-fold in transgenic mice between 5 days and 1 month of age, more than 2 months before the onset of disease, and was coordinately expressed with the DM20 transgene. Stromelysin-1 protein levels were also up-regulated as was tissue inhibitor of metalloproteinase-1 (TIMP-1), an in vivo regulator of stromelysin-1 mRNA. When we crossed our ND4 mice with a line of transgenic mice overexpressing TIMP-1 in brain, clinical signs in these mice were attenuated, and the level of stromelysin-1 protein was reduced. Thus, in this transgenic model of demyelinating disease up-regulation of DM20, MMP-3, and TIMP-1 represent important changes in the chemical pathogenesis in brain, which precede the onset of disease.

The matrix metalloproteinases (MMPs) are a family of endoproteinases that degrade various components of the extracellular matrix and have been implicated in the pathogenesis of multiple sclerosis. To determine whether up-regulation of MMP-3, or stromelysin-1, was a causative factor during the development of demyelination, we have examined the expression of MMP-3 mRNA and protein in brain tissue of a spontaneously demyelinating mouse model overexpressing DM20 (ND4 line) prior to and during the progression of disease. Stromelysin-1, but not other MMP mRNA was elevated ϳ10-fold in transgenic mice between 5 days and 1 month of age, more than 2 months before the onset of disease, and was coordinately expressed with the DM20 transgene. Stromelysin-1 protein levels were also up-regulated as was tissue inhibitor of metalloproteinase-1 (TIMP-1), an in vivo regulator of stromelysin-1 mRNA. When we crossed our ND4 mice with a line of transgenic mice overexpressing TIMP-1 in brain, clinical signs in these mice were attenuated, and the level of stromelysin-1 protein was reduced. Thus, in this transgenic model of demyelinating disease up-regulation of DM20, MMP-3, and TIMP-1 represent important changes in the chemical pathogenesis in brain, which precede the onset of disease.
In recent years, several members of the matrix metalloproteinase (MMP) 1 family have received considerable attention as important enzymatic components involved in the pathogenesis of multiple sclerosis (MS) (1)(2)(3). Multiple sclerosis, an inflammatory demyelinating disease of the human central nervous system is characterized by multiple lesions within the white matter (4,5). Although the etiology is unknown, genetic, environmental, and immune factors in various combinations are involved possibly accounting for the heterogeneity of the disease. An important role has been assigned to the extracellular matrix, the dissolution of which permits a variety of cells including astrocytes, macrophages, and sensitized lymphocytes access to myelin. The migration of cells through the brain parenchyma and the transmigration of T cells across the blood brain barrier are mediated by degradation of extracellular matrix components by powerful proteases, the matrix metalloproteinases (6).
The MMPs are a family of more than 20 proteases that are activated extracellularly in most cases by specific proteolytic hydrolysis to generate an active enzyme. They are Zn 2ϩ and Ca 2ϩ requiring neutral endopeptidases, which include collagenases, stromelysins, gelatinases, and membrane-type metalloproteinases (7)(8)(9). These enzymes are tightly regulated at the transcriptional level, and with the exception of the membrane type-MMPs, they are secreted as inactive zymogens that require activation. Their activities are controlled by other proteins called tissue inhibitors of metalloproteinases (TIMPs), which bind the enzymes in a 1:1 stoichiometry (10).
Gelatinase B (MMP-9) has been implicated in the transmigration of lymphocytes through the blood brain barrier in in vitro studies (11,12). Interferon ␤-1b, which is used in the treatment of MS, decreased the ability of T-lymphocytes to cross an artificial basement membrane in vitro, suggesting that the beneficial effects of interferon ␤-1b in MS may be ascribed to its inhibition of MMP-9 activity (13). MMP-9 has been reported to be selectively elevated in the cerebrospinal fluid during relapses and stable phases of MS (14). Ö zenci et al. (15) investigated the expression of MMP-9, stromelysin-1 (MMP-3), TIMP-1, and TIMP-2 in blood mononuclear cells obtained from normal individuals and patients with MS and other neurological diseases including inflammatory diseases. Numbers of MMP-9 mRNA-expressing cells were higher in MS than other neurological diseases, and MS patients had higher levels of MMP-3 and TIMP-1 mRNA than normal patients as well as other neurological diseases, suggesting that several MMPs may be elevated in MS (15). In a study of cells in the human central nervous system, which expressed MMPs, endothelial cells in MS brain expressed MMP-3 and MMP-9. Macrophages in active and necrotic lesions expressed MMP-1, MMP-2, MMP-3, and MMP-9, whereas astrocytes expressed MMP-2, MMP-3, and MMP-9 (16). Therefore, MS tissue expresses a number of MMPs in addition to MMP-9.
To determine whether MMPs have a causative role in the pathogenesis of MS, it is essential to carry out studies prior to the development of demyelinating disease. These studies require a relevant animal model, because studies on human disease can only be done after the disease is established and diagnosis is definite. The animal model, which we have used, is a transgenic model obtained by the incorporation of various copies of the cDNA for DM20, the major myelin proteolipid protein in early development, into the genome. Mice carrying 70 copies of the transgene (ND4 line) undergo a normal birth and development up to 3 months of age. At this time, they begin to have tremors, shake, show unsteady gait, and lose weight. These early signs become worse between 3 and 8 months, and by 10 months the animals become moribund (17,18). Mice carrying 17 copies of the transgene (ND3a line) are also born and develop normally. They begin to show signs consistent with demyelinating disease at 6 -8 months and become moribund by 16 months (19), suggesting a gene-dosage effect.
The MMP which we have focused on is stromelysin-1 (MMP-3), because it has a central position among MMPs for several reasons. Firstly, MMP-3 has a broad substrate specificity permitting it to degrade fibronectin, laminin, elastin, collagen IV, and proteoglycans (20,21). Secondly, MMP-3 activates other MMPs such as MMP-9, for example, overexpressing cell cultures needed to be supplemented with active MMP-3 to initiate the activation of pro-MMP-9 (22). Stromelysins have also been shown to activate pro-collagenase (23) and gelatinase A/TIMP-2 complex (24). Thirdly, MMP-3-dependent generation of a macrophage chemoattractant in a model of herniated disc resorption has been reported previously (25).
In this report, we demonstrate that MMP-3 was up-regulated both at RNA and protein levels prior to the appearance of disease, whereas other MMPs were not affected. The overexpression of TIMP-1 in vivo, which binds the enzyme, ameliorated the disease in a double transgenic model.

EXPERIMENTAL PROCEDURES
Animals-Normal, ND3a, and ND4 transgenic mice were of CD-1 background (18). The transgenic mice were produced by the incorporation of cDNA for human DM20 under the control of the human proteolipid protein (PLP) promoter into the genome of a normal CD-1 mouse. The presence of an SV40 site at the 3Ј end permitted specific detection of the transgene. On Northern blots, the transgene RNA resolved into species of 1.7 and 1.25 kb, whereas the transcripts from the endogenous PLP gene were detected at 3.2 and 2.4 kb. The ND3a line carried 17 copies of the cDNA for DM20, whereas the ND4 line carried 70 copies of the cDNA. The disease course was milder in the ND3a line with first signs at 6 months of age, whereas first signs were observed at 3 months in the ND4 line. All ND3a and ND4 transgenic mice were heterozygous for the DM20 transgene. Littermates not carrying the transgene were used as normal mice. TIMP-1 transgenic founder mice were C57BL/6 (line 277-34) (26) and were gifts from Dr. Rama Khokha. The TIMP-1 transgene was under the control of the murine metallothionein-1 promoter. All animals were genotyped by Southern blot analysis of DNA extracted from mouse tail clips.
Clinical Scores-Mice were scored for clinical signs three times a week starting at 2 months of age and continued until they were 8 months old. Clinical scores were calculated based on several criteria such as the extent of body shaking, tremors of the hind limb and head area, a jerky head, and an unsteady gait. Mice were scored on a four-point scale with a "1" representing very mild signs and a "4" representing severe signs. The sum of scores/week was calculated, averaged, and plotted against animal age.
RNA Isolation-RNA was isolated from brain tissue by the method of Chirgwin et al. (27). One half of a mouse brain was homogenized in 5 ml of guanidine thiocyanate solution (4.2 M guanidine thiocyanate, 60 mM sodium acetate, pH 6.5, 0.1% sarkosyl, 25 mM EDTA, pH 8.0, 100 mM 2-mercaptoethanol). The homogenate was layered over a 2-ml cesium chloride cushion (5.7 M CsCl, 50 M sodium acetate, pH 6.5, 1 mM EDTA) and centrifuged at 32,000 rpm at 15°C for 18 h. After centrifuging, the supernatant was removed, and the pellet was solubilized with two 200-l washes of diethylpyrocarbonate-treated water. The washes were pooled, and the RNA was precipitated with one-tenth volume of 3 M sodium acetate, pH 6.5, and two volumes of ethanol. The RNA was recovered by centrifuging, and the pellet was washed in 70% ethanol, dried, and redissolved in 200-l diethylpyrocarbonate-treated water. The concentration of RNA was determined from the relationship E 1 cm 1% ϭ 200 at 260 nm. Northern Blotting-Total RNA (10 g) was run on a 1.2% agarose formaldehyde gel and transferred to nylon membrane in 20ϫ SSC (3 M NaCl, 0.3 M sodium citrate). The membrane was prehybridized in prehybridization solution (1% bovine serum albumin, 0.35 M sodium phosphate (0.26 M Na 2 HPO 4 , 0.34% orthophosphoric acid), 7% SDS, 30% formamide) for 3 h at 55°C. Blots were then hybridized with 32 P-labeled cDNA probes at 55°C overnight. The membrane was washed twice in wash buffer (0.5% 20ϫ SSC, 0.5% SDS) at 55°C for 30 min. Blots were exposed to a phosphorimaging screen (Molecular Dynamics), and band intensities were quantitated on a phosphorimaging device (Image Quant).
Protein Assay-A one-half brain from normal and transgenic mice was homogenized in 4 ml of protein extraction buffer (1% Triton X-100, 0.5 M Tris-HCl, pH 7.5, 0.2 M NaCl, 10 mM CaCl 2 ), and the resulting homogenates were fractionated into supernatant and pellet fractions by centrifuging at 12,000 ϫ g for 30 min. Brain supernatant and pellet fractions were collected, and the protein concentration of each fraction was determined by the method of Peterson (35). 100 l of 0.15% sodium deoxycholate was added to each sample and allowed to stand for 10 min. 100 l of 72% trichloroacetic acid was then added, and the samples were centrifuged at 10,000 rpm for 10 min. The pellet was resuspended in 400 l of H 2 O. 400 l of reagent A containing 1 volume each of coppertartrate-carbonate, 10% sodium dodecyl sulfate, 0.8% NaOH, and H 2 O was added to each sample and reacted for 10 min. 200 l of reagent B containing 1 part Folin reagent (ICN Biomedicals Inc.) to 5 parts of H 2 O was then added, and the mixture allowed to react for 30 min before the absorbance was read at 750 nm. Protein concentrations were obtained by interpolation from a standard curve generated with bovine serum albumin.
SDS-PAGE and Western Blotting-Brain supernatant fractions (100 g) from normal and transgenic mice at various ages were run on 10% SDS-polyacrylamide gels (36) and transferred electrophoretically onto nitrocellulose membranes by the method of Towbin et al. (37). Western blot analysis was carried out using a mouse monoclonal anti-MMP-3 antibody (Oncogene) against both active and latent forms of the enzyme.

The Development of Clinical Disease in Transgenic Mice-
Transgenic mice carrying 70 copies of the cDNA for DM20 were observed for the development of clinical disease over an 8-month period. To assess the extent of the disease, each mouse was scored for clinical signs three times/week. At the end of each week, the scores were summed and plotted (see "Experimental Procedures" for scoring criteria). The data are plotted in Fig. 1, which represents the mean and standard deviation for six normal and six transgenic mice. The curve is typical of a total of more than 70 mice. Signs of the disease began at 3 months of age, increased slowly until 4 months and then rapidly between 4 and 6 months, and remained unchanged there-  (Table I). Of the six MMPs studied, the expression of gelatinase A and collagenase-3 was low at all ages, whereas the expression of gelatinase B, stromelysin-3, and matrilysin expression was undetectable. Stromelysin-1 was the only enzyme examined that showed a dramatic increase during development in transgenic animals.
Stromelysin-1 mRNA Expression Correlates with the Expression of the DM20 Transgene-In a more extensive developmental study of stromelysin-1 mRNA in ND4 transgenic mice ( Fig.  2A, curve A), the expression was low at 5 days but increased rapidly between 5 and 18 days, reached maximal values between 1 and 3 months, and remained at this high level until 8 months. The expression of stromelysin-1 mRNA in normal littermates shown in Fig. 2A, curve B, was barely detectable at all ages. Although these data are from a single set of Northern blots, they have been repeated three times by two different observers with identical results. The expression of stromelysin-1 was increased approximately 10-fold in transgenic animals between 5 days and 1 month of age. However, a 40-fold increase in stromelysin-1 was observed when compared with normal animals (Table I). Because the expression of stromelysin-1 was increased in transgenic animals between 5 days and 1 month of age although clinical signs were not evident until 3 months of age, a causative role for stromelysin-1 was suggested.
To determine when the DM20 transgene was expressed, Northern blots were run with RNA isolated from transgenic mice. To distinguish transgene mRNA from the transcripts from the normal PLP gene, the cDNA used for DM20 transgene contained an SV40 site at the 3Ј end. The probe used to detect DM20 detects both the endogenous PLP gene transcripts at 3.2 and 2.4 kb as well as the DM20 transgene mRNA at 1.7 and 1.25 kb. The two transcripts resulting from the transgene are thought to be attributed to an alternative polyadenylation site within the transgene (18). The expression of the DM20 transgene during development is shown in Fig. 2B. The expression of the DM20 transgene was low at 5 and 9 days after birth but increased rapidly between 9 and 18 days and remained at this high level for up to 6 months. The expression pattern of strome-lysin-1 and DM20 transgene was remarkably similar, suggesting that the expression of the two genes was coordinated.
Another series of Northern blots were run in which the expression of stromelysin-1 was correlated with DM20 transgene dosage. To perform this experiment, RNA was isolated from normal mouse brain and transgenic mouse brain carrying 17 (Fig. 3, ND3a) and 70 (ND4) copies of the cDNA for DM20 at 1 and 8 months of age. Northern blots were probed with 32 Plabeled cDNA probes for stromelysin-1 and DM20. Quantitation of these blots is shown in Fig. 3. The expression of stromelysin-1 and DM20 mRNA in normal, ND3a, and ND4 lines at 1 and 8 months of age is shown. The expression of stromelysin-1 was highest in the ND4 line carrying 70 copies of the transgene, which correlated well with the highest expression of the DM20 transgene. The amount of stromelysin-1 mRNA in the ND3a line was approximately 25% of that in the ND4 line, and again this correlated well with the DM20 transgene expression. Neither stromelysin-1 nor the DM20 transgene was expressed in kidney or liver at 8 months of age. These experiments demonstrate that the expression of stromelysin-1 occurred only in the brain, and the level of expression was proportional to the number of copies of the DM20 transgene.
MMP Protein Levels in Normal and Transgenic Mice-To determine whether the increased expression of stromelysin-1 mRNA resulted in a corresponding increase in stromelysin-1 protein, the levels of stromelysin-1 protein were examined in normal and transgenic animals by Western blot analysis (Fig.  4). At all ages, a 45-and a 28-kDa band were detected, both corresponding to the active forms of the enzyme. At 9 days of age, both 45-and 28-kDa stromelysin-1 expression was easily The mRNA expression of stromelysin-1 (MMP-3), stromelysin-3 (MMP-11), gelatinase A (MMP-2), gelatinase B (MMP-9), matrilysin (MMP-7), and collagenase-3 (MMP-13) was studied during development in the mouse from 5 days to 1 month. Total RNA was extracted by the method of Chirgwin et al. (27) from half of a mouse brain. Northern blots were run and probed with the appropriate 32 P-labeled cDNA (see "Experimental Procedures"). Blots were exposed to a phosphorimaging screen, and band intensities were quantitated on a phosphorimaging device. Band intensities were then normalized to the level of GAPDH mRNA. MMP levels are expressed as a ratio of transgenic/normal. ND, not detected.  2. Expression of stromelysin-1 and DM20 transgene mRNA during development of clinical disease. A, Northern blots were run with RNA (10 g) isolated from brain tissue from normal and ND4 transgenic mice at different ages and were probed for stromelysin-1 and GAPDH. Blots were exposed to a phosphorimaging screen, and band intensities were quantitated on a phosphorimaging device. Levels of stromelysin-1 mRNA were normalized to the level of GAPDH mRNA. Curve A, ND4 transgenic mice; curve B, normal mice. B, Northern blots were run with RNA (10 g) isolated from ND4 transgenic mouse brain of various ages from 5 days to 8 months and probed with a 32 P-labeled cDNA probe for DM20. Specific detection of the transgene mRNA was possible because of the SV40 site at the 3Ј end of the transgene, and it fractionated as two bands at 1.7 and 1.25 kb. Blots were exposed to a phosphorimaging screen, and band intensities were quantitated on a phosphorimaging device. Levels of DM20 transgene mRNA were normalized to the level of GAPDH mRNA. detected in the brain supernatants from normal mice. In the transgenic mice, the 28-kDa protein was detected in similar amount to that in normal mice, whereas the 45-kDa band was greatly reduced. At 1 month of age, protein expression was low in both normal and transgenic mice, although levels were higher in transgenic animals (Fig. 4B). At 6 months of age, both 45-and 28-kDa proteins were detected again in both normal and transgenic mice. However, both proteins were increased in transgenic mice compared with normal mice. Therefore, both mRNA ( Fig. 2A) and stromelysin-1 protein were increased in ND4 mice at 1 month of age and 2 months prior to onset of the disease.
Expression of TIMPs in Normal and Transgenic Mice-The tissue inhibitor of metalloproteinases (TIMPs) are capable of combining with the active form of MMPs in a 1:1 stoichiometry, thereby inhibiting their activity. The interaction between TIMP-1 and stromelysin-1 has been studied extensively (38 -40). To determine whether the expression of TIMP-1, TIMP-2, TIMP-3, and TIMP-4 in normal and transgenic mice carrying 70 copies of the DM20 transgene was elevated, RNA was isolated from brains in a developmental study from 5 days to 6 months for Northern blot analysis. The data are shown in Table  II. TIMP-3 expression was undetectable. Although the expression of TIMP-2 and TIMP-4 mRNA showed little change during development, TIMP-1 mRNA increased gradually from 5 days to 1 month and then to a transgenic/normal ratio of 2.24 at 6 months. A more detailed study of TIMP-1 expression (data not shown) showed that at 8 months TIMP-1 was approximately 5-fold increased in ND4 mice over normal mice.
The clinical course at different ages is shown in Fig. 5. In the crosses, which are C57BL/6 Ϫ CD-1, the TϪDϩ (DM20 overexpressors) had a clinical course that rose gradually from 3 to 5.5 months of age (Fig. 5, curve B). The double transgenic mice fell into two populations. One group of the double transgenics showed similar scores as the TϪDϩ mice (Fig. 5, curve A). The other group accounting for approximately 50% of the total (Fig.  5, curve C) showed an attenuated clinical course. At 5 months, the clinical scores were approximately 50% of the TϪDϩ, suggesting that in this group, the inhibition of stromelysin-1 by TIMP-1 was beneficial. The reason for the difference in clinical scores for the two groups of TϩDϩ mice is not understood at this time. In some way, it may be related to the mixed C57BL/ 6 Ϫ CD-1 background.
Correlation of Stromelysin-1 and DM20 in Double Transgenic Mice-To determine whether the double transgenics (Fig.  6, TϩDϩ) that showed the attenuated clinical course expressed stromelysin-1 at the same levels as the DM20 overexpressors (TϪDϩ), Northern blots examining the expression of stromelysin-1 mRNA, DM20 mRNA, and TIMP-1 mRNA in double transgenic mice were done (Fig. 6). Both the normal mice (N on a CD-1 background, TϪDϪ on a mixed C57BL/6-CD-1 background) and the DM20 transgenics (ND4, TϪDϩ) were 6-month-old mice. The normal mice did not express stromely-

FIG. 3. Quantitation of stromelysin-1 and DM20 transgene dosage.
Northern blots were run with RNA (10 g) isolated from brain, kidney, and liver from normal (N), ND3a (17 copies of DM20 transgene), and ND4 (70 copies of DM20 transgene) mouse lines at 1 and 8 months of age. Blots were probed with 32 P-labeled cDNA probes for stromelysin-1, DM20, and GAPDH. Blots were exposed to a phosphorimaging screen, and band intensities were quantitated on a PhosphorImager.

FIG. 4. Expression of MMP-3 protein.
A, supernatants (100 g) from normal (N) and ND4 transgenic mouse brain homogenates were run on SDS-PAGE, transferred to nitrocellulose membranes, and probed with a mouse monoclonal anti-MMP-3 antibody (Oncogene) against both active and latent forms of the enzyme. The 45-and 28-kDa active forms of stromelysin-1 were detected in 9-day-old, 1-month-old, and 6-month-old mice. B, the 45-and 28-kDa bands were quantitated on a Macintosh computer using the public domain NIH Image program (developed at NIH), and results were expressed as a ratio of transgenic/normal.

TABLE II Expression of TIMP mRNA in transgenic and normal mice
The mRNA expression of TIMP-1, TIMP-2, TIMP-3, and TIMP-4 was studied during development in the mouse from 5 days to 6 months. Total RNA was extracted by the method of Chirgwin et al. (27) from half of a mouse brain. Northern blots were run and probed with the appropriate 32 P-labeled cDNA (see "Experimental Procedures"). Blots were exposed to a phosphorimaging screen, and band intensities were quantitated on a phosphorimaging device. Band intensities were then normalized to the level of GAPDH mRNA. TIMP levels are expressed as a ratio of transgenic/normal. ND, not detected.  1 and 3).
The DM20 transgenics showed a high expression of stromelysin-1, the DM20 transgene, and increased expression of endogenous TIMP-1 expression at 6 months of age (lanes 2 and 4). In the developmental study at 2, 4, and 6 months with the TϩDϪ, i.e. the TIMP-1 overexpressors, the expression of stromelysin-1 and DM20 was not observed. As expected, TIMP-1 expression was very high (lanes 5-7). The double transgenics TϩDϩ expressed stromelysin-1, DM20, and TIMP-1 at all ages (lanes 8 -10). These data demonstrated that (i) the normal mice expressed the transcripts produced from the normal PLP gene but none of the transgene (Fig. 6B, lane 1); (ii) the DM20 transgenics with 70 copies of the transgene (ND4) expressed stromelysin-1, the transcripts from the endogenous PLP gene, the DM20 transgene, and endogenous TIMP-1 (Fig. 6B, lane 2); (iii) the TϪDϪ mice, generated from the cross, expressed small amounts of the transcripts from the normal PLP gene (lane 3); (iv) the TϪDϩ mice expressed stromelysin-1, the DM20 transgene, and endogenous TIMP-1 similar to the ND4 (lane 4); (v) the expression of stromelysin-1 was dependent on the presence of the DM20 transgene (compare lane 3 with 4); (vi) in the absence of the DM20 transgene, the TIMP-1 overexpressors (TϩDϪ) did not express stromelysin-1, but they expressed the endogenous PLP transcripts and high amounts of TIMP-1 as expected (lanes 5-7); (vii) in the presence of the DM20 transgene, the TIMP-1 overexpressors (TϩDϩ) expressed stromelysin-1, the DM20 transgene, and TIMP-1. These data demonstrate that expression of stromelysin-1 and TIMP-1 correlated with the presence of the DM20 transgene.

Expression of MMP-3 Protein in TϩDϩ Mice-
To determine the level of stromelysin-1 protein in the double transgenics (Fig. 7, TϩDϩ), which showed high levels of stromelysin-1 mRNA but the attenuated clinical course, stromelysin-1 protein was studied. Western blots were done (Fig. 7) on brain supernatants from 6-month-old TϪDϩ and 6-month-old TϩDϩ mice. Fig. 7A shows MMP-3 protein levels in brain from 6-month-old normal, 6-month-old ND4, 6-month-old TϪDϩ, and 6-month-old TϩDϩ mice. Fig. 7B shows quantitation of the 45-and 28-kDa bands. ND4 mice showed a dramatic increase in MMP-3 protein levels, especially the 28-kDa form, compared with normal mice. The TϪDϩ mice showed a similar level of both the 45-and 28-kDa forms of MMP-3 as the ND4. Interestingly, the TϩDϩ mice showed a dramatic reduction in levels of the 28-kDa form, whereas the expression of the 45-kDa form was barely detectable. It is probable that TIMP-1 binds more efficiently to the 45-kDa protein than to the 28-kDa form, because this smaller protein is missing the C-terminal hemopexin-like domain. Thus, increased TIMP-1 levels abolished the 45-kDa stromelysin-1 protein and greatly reduced the 28-kDa protein, suggesting that the beneficial effects of TIMP-1 overexpression on clinical course was the result of a reduction of stromelysin-1 protein. DISCUSSION MMPs have been implicated in the pathogenesis of demyelinating disease (i.e. MS) in humans and in experimental models of MS, particularly the experimental allergic encephalomyelitis model (1,3,(41)(42)(43). Although several MMPs may be involved depending on the stage of the disease, gelatinase B (MMP-9) has received the most attention in MS (11,12,44,45) and in experimental allergic encephalomyelitis (46). However, other MMPs such as MMP-3 and MMP-7 have also been reported to be elevated in MS (16,43).
Of the various members of the MMP family, stromelysins have special importance. They have a broad substrate specific-  7. MMP-3 protein in T؉D؉ mice. A, supernatants (100 g) from 6-month-old normal, ND4, TϪDϩ, and TϩDϩ mouse brain homogenates were run on SDS-PAGE, transferred to nitrocellulose membranes, and probed with a mouse monoclonal anti-MMP-3 antibody (Oncogene) against both active and latent forms of the enzyme. The 45and 28-kDa active forms of stromelysin-1 were detected. B, the 45-and 28-kDa bands were quantitated using the public domain NIH Image program on a Macintosh computer. ity, degrading fibronectin, laminin, elastin, collagen IV, and proteoglycans (20,21), suggesting they are capable of carrying out widespread damage. They activate MMP-9 (gelatinase B) produced by MMP-9-overexpressing cells. In these studies, active MMP-9 was only generated if active stromelysin-1 was added to the cultures (22). These authors postulated that MMP-3 represents the in vivo mechanism for the conversion of pro-MMP-9 to active MMP-9. Stromelysins also activate procollagenase (23), pro-gelatinase B (47), and gelatinase A/TIMP-2 complex (24).
In a model of herniated disc, MMP-3 was found in high amounts. The recruitment of macrophages to the herniated disc was dependent on chondrocyte MMP-3, which generated a macrophage chemoattractant with subsequent infiltration of the disc by proteolytically active macrophages (25). In active MS lesions, macrophages can be seen ingesting myelin debris and have been shown to be the principal cellular component in fulminating disease of the Marburg's type (48). Although not known, macrophage activation in MS may also be MMP-3-dependent. In rheumatoid arthritis, serum levels of MMP-3 were reported to be a predictor of joint destruction and correlated with disease activity (49 -51). Vacuolation in murine prion disease occurred because of the destruction of the extracellular matrix by stromelysin-1, which was up-regulated 25-fold demonstrating an important role for this MMP (52).
The studies mentioned above suggest that stromelysin-1 (MMP-3) has a special role in several autoimmune diseases, specifically in MS and rheumatoid arthritis. Most of the reported studies of gelatinase B in MS and stromelysin-1 in rheumatoid arthritis were done on samples obtained after the disease became evident clinically. MMP-9 (gelatinase B) was found to be elevated in cerebrospinal fluid during relapses and stable phases of MS (14). In adoptive transfer experimental allergic encephalomyelitis, MMP-9 mRNA peaked with maximum disease severity (46). In all cases an inflammatory response was either induced in vitro with interleukin-2 as in the case of the transmigration of lymphocytes in vitro (12), or MMP was measured in samples obtained from diseased tissue. From such studies, it is difficult to determine whether the up-regulation of MMP-9 was the result of the inflammatory process or was related to disease induction. To resolve this issue, we carried out our studies in a relevant animal model with a 3-month period of normal development so that changes in MMPs prior to the onset of disease could be correlated with disease induction, i.e. prior to the appearance of clinical signs.
In the studies presented in this report, we have used a transgenic mouse model of demyelination that carries 70 copies of the cDNA for DM20 with many features that are relevant to MS in humans (17). All mice carrying the transgene are clinically normal up to 3 months of age when the earliest signs of demyelinating disease appear. The signs worsen until the animals become moribund at approximately 10 months. Neuropathological examination was normal at 2 months of age and showed an occasional demyelinated axon at 2.5 months. Therefore, these mice with a predictable disease course provide us with a 2-3 month period in which changes that precede demyelination may be observed.
The choice of the DM20 transgene to generate the various transgenic lines was not fortuitous. DM20 represents the major proteolipid in embryonic and neonatal development but is a minor proteolipid in the adult (53,54). Evolutionary studies have suggested that DM20 is the ancestral gene product that acquired the insert of amino acids 116 -150 to generate the adult PLP protein. Despite this finding, DM20 is usually considered to arise by alternative splicing of the PLP mRNA (55). Therefore, by maintaining a high level of DM20 and a low level of PLP (56), the ND4 transgenic mice maintain several features associated with an early stage of development. Developmental immaturity has been suggested as a mechanism that predisposes humans to MS (57). With this tool, we were able to investigate changes in MMPs during the development of demyelinating disease. As shown in Table I, only stromelysin-1 (MMP-3) showed a dramatic increase in mRNA (18-fold over normal at 1 month). Our subsequent studies focused on this MMP.
A developmental time course study from 5 days to 8 months of age revealed a large (10-fold) increase in the mRNA for stromelysin-1 within the first month in transgenics, which remained at a high level of up to 8 months. A similar time course was observed for the DM20 transgene, suggesting coordinate expression of the two genes. Furthermore, when the ND3a line (17 copies of transgene) was examined, the increase in stromelysin-1 was found to be proportional to the transgene copy number. This transgene dosage effect was evident at both 1 and 8 months of age (Fig. 3). Although we do not have an explanation for this apparent coordinated expression, a possible explanation may be found in the promoter region of stromelysin-1, which contains a stromelysin-1 platelet-derived growth factor-responsive element. A novel transcription factor, stromelysin-1 platelet-derived growth factor-responsive element-binding protein, binds to this site (58,59). This element is not found in the promoter region of other MMP genes. A direct effect of DM20 transgene on either stromelysin-1 platelet-derived growth factor-responsive element or stromelysin-1 platelet-derived growth factor-responsive element-binding protein may explain the coordinated expression. These possibilities represent future studies.
The high protein levels of stromelysin-1 seen in 9-day-old normal and transgenic mice coincide with the onset of myelination in the mouse (60). It is known that MMPs are required for tissue remodeling during early developmental processes such as oligodendrocyte process extension (61). High levels of stromelysin-1 have also been observed in the developing rat cerebellum at postnatal day 10 (62). The expression of this protein at this stage of development was suggested to be related to the migration of granular cell precursors. At 1 month of age, ND4 animals showed very low levels of MMP-3 protein, although high levels of MMP-3 mRNA were detected. Because myelin synthesis in the mouse is complete at 1 month of age, the low stromelysin-1 protein levels suggest that in normal development, low stromelysin-1 levels correlated with stable compact myelin. The increase in MMP-3 protein at 6 months in the normal mice probably reflects myelin turnover and correlates with aging, whereas the larger amounts of MMP-3 in the transgenic mice correlate with demyelination.
The level of MMP-3 mRNA at 1 month was high, however, MMP-3 protein levels were low at this time. This observation may result from decreased translational efficiency, increased protein degradation, or increased mRNA stability. The 5Ј-untranslated region of mRNA has been implicated in modulating translational efficiency (63,64). Specific mRNA-binding proteins such as cap-binding proteins, which bind to this region, may be differentially regulated during postnatal development in normal and ND4 mice, resulting in the differential regulation of MMP-3 mRNA and protein.
The exact mechanism by which high MMP-3 levels induce or aggravate demyelination is not known. However, MMP-3 has been shown to have the second highest activity after MMP-2 on the degradation on myelin basic protein in vitro (65). A role for MMP-3 in activating a protease cascade is also probable. High levels of MMP-3 protein detected at 6 months in ND4 mice have the potential to activate both the gelatinases and collagenases, which would result in further tissue damage. The improvement in disease course observed in animals that overexpressed TIMP-1 (TϩDϩ), suggests that the inhibition of this MMP may be beneficial and MMP-3 may be a useful therapeutic target.
Coordinated up-regulation of MMPs and TIMPs has been reported in several studies (15,66). We have found an upregulation of TIMP-1 mRNA in mice that overexpress stromelysin-1 (Fig. 6, A and C, lanes 2 and 4). This coordinated regulation may occur through common promoter elements such as activator protein-1 and polyomavirus enhancer-A-binding protein-3 (67,68). Although up-regulation of stromelysin-1 correlates with the up-regulation of TIMP-1 mRNA (Fig. 6, lanes  2 and 4), the reverse does not hold, i.e. increased TIMP-1 mRNA via transgene expression does not lead to up-regulation of stromelysin-1 (Fig. 6, lane 7). Coordinate up-regulation of MMPs and TIMPs is expected, because an increase in MMP activity would necessitate an increase in TIMP. A developmental time course analysis of TIMP-1 expression in normal and transgenic mice revealed a large increase in TIMP-1 mRNA expression in transgenic mice from 6 to 8 months. This upregulation of the endogenous TIMP-1 may be a response to the appearance of high MMP-3 protein levels at these later stages of disease (Fig. 4).
Increased TIMP-1 expression has been shown to have protective effects in vivo, e.g. resistance to experimental brain metastases in fibrosarcoma (26). Invasiveness of human brain tumors has been inversely correlated with TIMP-1 expression (69). An invasive human astrocytoma cell line, which overexpressed gelatinases A and B, showed decreased in vitro invasive potential when transfected with TIMP-1 (70). This study demonstrated that up-regulation of TIMP-1 via transgene expression in a spontaneously demyelinating model attenuated disease in 50% of the double transgenics. Because we found no differences in the DM20 or TIMP-1 transgene copy numbers between the two populations of the double transgenics (data not shown), the C57BL/6 Ϫ CD-1 cross must be responsible.
The level of MMP-3 protein was dramatically reduced in DM20 overexpressing mice that expressed high TIMP-1 (Fig. 7, TϩDϩ) compared with the level in the DM20 overexpressors that did not overexpress TIMP-1 (TϪDϩ). The attenuation of the disease seen in TϩDϩ mice may reflect MMP-3 inhibition by increased levels of TIMP-1. Although the source of MMP-3 in our model is not known, it is probable that this enzyme is secreted by inflammatory cells and/or glial cells. The attenuation of the disease seen in TϩDϩ mice may also reflect decreased levels of inflammatory and activated glial cells within the brain, which would result in the diminished levels of the MMP-3 observed.
In summary, our data suggest that stromelysin-1 up-regulation is probably a causative factor in the onset of demyelinating disease in our transgenic model, because both mRNA and protein up-regulation occurred prior to clinical or pathological evidence of demyelination. Increased TIMP-1 transgene levels were protective in 50% of the double transgenic animals (TϩDϩ), implicating MMP activity in disease progression. Although coordinate up-regulation of MMP and TIMP was expected, the coordinate regulation of stromelysin-1 and DM20 transgene was not. The correlation of stromelysin mRNA levels with DM20 transgene copy number suggests tight regulation of these two processes.