A novel intracellular role of matrix metalloproteinase-3 during apoptosis of dopaminergic cells


  • Dong Hee Choi,

    1. Department of Biochemistry and Molecular Biology, University of Ulsan College of Medicine, Seoul, Korea
    2. Institute for Neurochemistry, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Eun-Mee Kim,

    1. Department of Biochemistry and Molecular Biology, University of Ulsan College of Medicine, Seoul, Korea
    2. Institute for Neurochemistry, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Hyo Jin Son,

    1. Department of Biochemistry and Molecular Biology, University of Ulsan College of Medicine, Seoul, Korea
    2. Institute for Neurochemistry, University of Ulsan College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Tong H. Joh,

    1. Department of Neurology and Neuroscience, Weill Medical College of Cornell University, New York, USA
    Search for more papers by this author
  • Yoon Seong Kim,

    1. Department of Neurology and Neuroscience, Weill Medical College of Cornell University, New York, USA
    Search for more papers by this author
  • Donghou Kim,

    1. Department of Anatomy and Cell Biology, University of Ulsan College of Medicine, Seoul, Korea
    2. Brain Research Center, Asan Institute for Life Science, Seoul, Korea
    Search for more papers by this author
  • M. Flint Beal,

    1. Department of Neurology and Neuroscience, Weill Medical College of Cornell University, New York, USA
    Search for more papers by this author
  • Onyou Hwang

    1. Department of Biochemistry and Molecular Biology, University of Ulsan College of Medicine, Seoul, Korea
    2. Institute for Neurochemistry, University of Ulsan College of Medicine, Seoul, Korea
    3. Brain Research Center, Asan Institute for Life Science, Seoul, Korea
    Search for more papers by this author

Address correspondence and reprint requests to Onyou Hwang, PhD, Department of Biochemistry and Molecular Biology, University of Ulsan College of Medicine, 388-1 Pungnap-dong, Seoul 138-736, Korea. E-mail: oyhwang@amc.seoul.kr


We have previously demonstrated that the active form of matrix metalloproteinase-3 (actMMP-3) is released from dopamine(DA)rgic neurons undergoing apoptosis. Herein, whether actMMP-3 might be generated intracellularly, and if so, whether it is involved in apoptosis of DArgic neurons itself was investigated in primary cultured DArgic neurons of wild-type, MMP-3 knockout animals, and CATH.a cells. During apoptosis, gene expression of MMP-3 is induced, specifically among the various classes of MMPs, generating the proform (55 kDa) which is subsequently cleaved to the catalytically active actMMP-3 (48 kDa) involving a serine protease. Intracellular actMMP-3 activity is directly linked to apoptotic signaling in DArgic cells: (i) Pharmacologic inhibition of enzymatic activity, repression of gene expression by siRNA, and gene deficiency all lead to protection; (ii) pharmacologic inhibition causes attenuation of DNA fragmentation and caspase 3 activation, the indices of apoptosis; and (iii) inhibition of the pro-apoptotic enzyme c-Jun N-terminal protein kinase leads to repression of MMP-3 induction. Under the cell stress condition, MMP-3 is released as actMMP-3 rather than the proform (proMMP-3), and catalytically active MMP-3 added to the medium does not cause cell death. Thus, actMMP-3 seems to have a novel intracellular role in apoptotic DArgic cells and this finding provides an insight into the pathogenesis of Parkinson’s disease.

Abbreviations used

active MMP-3






fetal bovine serum


c-Jun N-terminal protein kinase




lactose dehydrogenase


matrix metalloproteinase-3


N-isobutyl-N-(4-methoxyphenylsulfonyl) glycyl hydroxamic acid


phosphate-buffered saline


phenylmethylsulfonyl fluoride


the proform of MMP-3


sodium dodecyl sulfate


tyrosine hydroxylase


transferase biotin-dUTP nick-end labeling



The role of intracellular proteases in neuronal death signaling and neurodegeneration has been demonstrated in recent years. In addition to the well-known cysteine aspartate-specific caspases (reviewed by Troy and Salvesen 2002), calpains, cathepsins, and serine proteases have been shown to contribute to cell death (reviewed by O’Connell and Stenson-Cox 2007; Leist and Jaattela 2001; Nakanishi 2003). Matrix metalloproteinases (MMPs), a class of zinc-dependent proteases known to degrade the extracellular matrix have received much attention in recent years for their apparent role in the pathophysiology of neurodegenerative diseases (reviewed by Cauwe et al. 2007; Crocker et al. 2004; Rosenberg 2002; Yong 2005). Specifically, MMPs have been linked to neurodegeneration associated with ischemia, brain trauma, neuroinflammation, multiple sclerosis, Alzheimer’s disease, Parkinson’s disease, stroke, human immunodeficiency virus-associated dementia, and glioma.

We have recently observed that MMP-3 plays a critical role as an intercellular signaling molecule that modulates neuroinflammatory responses in both cellular and animal models of Parkinson’s disease (Kim et al. 2005b, 2007). That is, dopamine(DA)rgic neurons that are under cellular stress and undergoing apoptosis release MMP-3 which in turn triggers microglial activation and production of proinflammatory biomolecules. Interestingly, this MMP-3 released from the apoptotic DArgic neurons was mostly the cleaved form rather than the proform (Kim et al. 2005b). Whereas MMPs in general are thought to be secreted as the proform and are processed to the active form outside the cell (Lijnen 2002), our finding that the cleaved MMP-3 is released suggest that in addition to the previously known extracellular cleavage of the proform, MMP-3 activation might occur inside cells that are under stress. In fact, so far intracellular generation of active form has been observed in the case of two other MMPs (Pei and Weiss 1995, 1996; Luo et al. 2002). Furthermore, if the catalytically active MMP-3 (actMMP-3) is indeed generated inside the cell, it was possible that MMP-3 might participate in intracellular events in DArgic neurons.

Lines of evidence have demonstrated both in vivo and in vitro that DArgic neurons selectively undergo apoptotic death when exposed to tetrahydrobiopterin (BH4) or when the level of endogenously present intracellular BH4 is increased (Choi et al. 2000, 2003a,b, 2005; Anastasiadis et al. 2001; Kim et al. 2003, 2004, 2005a; Lee et al. 2007). Therefore, the BH4-induced apoptosis of DArgic cells can be used as an experimental model to study the DArgic cell vulnerability. CATH.a is a clonal cell line of the CNS producing primarily DA and to a lesser extent, norepinephrine (Suri et al. 1993), and has been extensively used as an in vitro system model to study the DArgic cell death (Masserano et al. 1996; Choi et al. 2000, 2003a,b, 2004, 2005, 2006; Miyazaki et al. 2006; Han et al. 2007). We have demonstrated that the BH4-induced apoptotic phenomena observed in CATH.a cells are reciprocated in primary cultured mesencephalic DArgic neurons (Lee et al. 2007) making CATH.a, a valid substitute for the DArgic neurons in experiments that require a large number and/or homogenous population of cells.

This study, using both primary cultured neurons and CATH.a cells, sought to determine whether there is an intracellular mechanism for actMMP-3 generation and whether actMMP-3 has an intracellular role during apoptosis of DArgic neurons. We demonstrate that intracellular MMP-3 activity is altered in response to cellular stress and that this is because of changes in gene expression and activation by cleavage of proMMP-3. In addition, we demonstrate that the elevation of MMP-3 activity is an early event in the apoptotic death signaling occurring upstream of caspase 3 and downstream of c-Jun N-terminal protein kinase (JNK) and that suppression of MMP-3 activity by various methods leads to protection of DArgic cells.



Fetal bovine serum (FBS), horse serum, RPMI 1640, l-glutamine, trypsin/EDTA, and penicillin–streptomycin, neurobasal medium, and B-27 were from GibcoBRL (Gaithersburg, MD, USA). BH4, poly-l-lysine, phenylmethylsulfonyl fluoride (PMSF), Nonidet P-40, SP600125, Brij35, and bupropion were purchased from Sigma Chemicals (St. Louis, MO, USA). N-isobutyl-N-(4-methoxyphenylsulfonyl) glycyl hydroxamic acid (NNGH) and MMP-3 fluorescent assay kit were purchased from Biomol (Plymouth Meeting, PA, USA). In situ cell death detection kit for transferase biotin-dUTP nick-end labelin (TUNEL) staining was from Roche Diagnostics GmbH (Penzberg, Germany). Mouse monoclonal anti-tyrosine hydroxylase (TH) and goat polyclonal anti-MMP-3 were obtained from R&D systems (Minneapolis, MN, USA), and rabbit polyclonal anticleaved caspase 3 antibody was from Cell Signaling (Beverly, MA, USA). Anti-goat IgG and anti-rabbit IgG were from Sigma Chemicals and Alexa Fluor 488 donkey anti-goat IgG, Alexa Fluor 488 goat anti-rabbit IgG, and Alexa Fluor 546 donkey anti-mouse IgG were from Molecular Probes (Eugene, OR, USA). Recombinant MMP-3 containing only the catalytically active domain (22 kDa) was purchased from Calbiochem (San Diego, CA, USA). Enhanced chemiluminescence kit was obtained from Pierce (Rockford, IL, USA), [3H]DA from Amersham Biosciences (Piscataway, NJ, USA), and Taq polymerase from Roche Applied Science (Indianapolis, IN, USA). Laminin, Trizol reagent, superscript II reverse transcriptase, LipofectaminTM 2000, and zymogram casein gel were purchased from Invitrogen (Carlsbad, CA, USA). All other chemicals were of reagent grade from Sigma Chemicals or Merck (Rahway, NJ, USA).

CATH.a cell culture

Cells were grown in RPMI 1640 containing 8% horse serum, 4% FBS, 100 IU/L penicillin, and 10 μg/mL streptomycin at 37°C in 95% air and 5% CO2 in humidified atmosphere. For experiments, the cells were plated on polystyrene tissue culture dishes at a density of approximately 1.5–3 × 105 cells/well in 24-well culture plates or 5.5 × 105–1.5 × 106 cells/well in 6-well culture plates. After 24 h, the cells were fed with fresh medium and used for experiments.


All procedures were approved by the Animal Experiment Review Board of the University of Ulsan Asan Institute for Life Science and performed in compliance with the guidelines set forth by the International Society for Neurochemistry. Pregnant Sprague–Dawley rats were obtained from Orient Charles River Technology (Seoul, Korea). MMP-3 knockout (KO) mice (C57BL/6x129SvEv), originally developed by Mudgett et al. (1998), and their wild-type (WT) animals were obtained from Taconic Farms (Germantown, NY, USA) and bred at the specific pathogen-free animal facility of Asan Institute for Life Science, University of Ulsan of College Medicine (Seoul, Korea). To obtain mouse embryos for primary culture experiments, adult female mice (5–7 week old) were superovulated by injecting 5 IU of pregnant mare serum gonadotropin i.p., followed by administration of 5 IU of human chorionic gonadotropin 48 h later. The animals were mated with fertile male mice. The presence of vaginal plugs were determined on the following day, which was considered embryonic day zero.

Primary culture of mesencephalic neurons

The ventral mesencephalon was removed from 14-day gestation rat embryo or 13-day gestation mice embryo and incubated with 0.01% trypsin in Hank's balanced salt solution (HBSS) for 15 min at 37°C. After trituration, 3 × 105 cells were plated on each polystyrene cover slide that had been pre-coated with 100 μg/mL poly-l-lysine and 4 μg/mL laminin and placed in a 24-well culture plate. The cells were maintained at 37°C in a humidified atmosphere with 5% CO2 in neurobasal medium supplemented with B-27, 2 mM glutamine, 100 IU/L penicillin, and 10 μg/mL streptomycin. On day 5 or 6 in vitro, the neurons were fed with fresh medium and treated.

Cell staining

Cells grown and treated on a cover slide were fixed in cold 4% paraformaldehyde in 0.1 M phosphate-buffered saline (PBS), pH 7.4 for 30 min at 24°C. After washing twice in PBS, the cells were incubated for 1 h in blocking solution (5% FBS and 0.3% Triton X-100 in 0.1 M PBS). The cells were then incubated overnight with appropriate primary antibody (TH, 1 : 500; MMP-3, 1 : 500; cleaved caspase 3, 1 : 100) diluted in incubation solution (5% FBS and 0.2% Triton X-100 in 0.1 M PBS) at 4°C and washed twice in PBS. The samples were incubated at 24°C for 1 h with appropriate fluorescence-labeled secondary antibody (TH, Alexa Fluor 546 goat anti-mouse IgG; MMP-3, Alexa Fluor 488 donkey anti-goat IgG; cleaved caspase 3, Alexa Fluor 488 goat anti-rabbit IgG) diluted 1 : 200 in the incubation solution. As a negative control, the samples were incubated with the respective secondary antibodies only. The cells were washed in PBS, mounted on a glass slide, and viewed by confocal microscopy (TCS-ST2; Leica, Wetzlar, Germany). TUNEL staining was performed according to the manufacturer’s protocol, labeled with fluorescein and analyzed by confocal microscopy.

Western blot analysis

Cells were washed with ice-cold PBS and lysed on ice in lysis buffer (50 mM Tris–HCl, pH 8.0, 137 mM NaCl, 1% Nonidet P-40, 10% glycerol, 50 mM sodium fluoride, 10 mM sodium pyrophosphate, 100 μM molybdic acid, 1 mM sodium orthovanadate, 10 μg/mL leupeptin, 10 μg/mL aprotinin, and 1 mM PMSF). The soluble fraction was obtained and equal amounts (30 μg) of cell lysate protein were loaded in each lane of sodium dodecyl sulfate (SDS) polyacrylamide gel. After electrophoresis and transfer onto polyvinylidene difluoride membrane, specific protein bands were detected using appropriate antibodies (MMP-3, 1 : 1000; cleaved caspase 3, 1 : 1000) and secondary antibodies followed by enhanced chemiluminescence.

Lactate dehydrogenase assay

Aliquots (50 μL) of cell culture medium were incubated at 24°C in the presence of 0.26 mM NADH, 2.87 mM sodium pyruvate, and 100 mM potassium phosphate buffer (pH 7.4) in a total volume of 200 μL. The rate of NAD+ formation was monitored for 5 min at 2-s intervals at 340 nm using a microplate spectrophotometer (SPECTRA MAX 340 pc; Molecular Devices, Sunnyvale, CA, USA).

MMP-3 activity assay

Matrix metalloproteinase-3 activity was measured using MMP-3 fluorescence assay kit following the manufacturer’s instructions. To measure the activity of released MMP-3, 100 μL of the medium was collected and transferred to 96 wells, to which 99 μL of assay buffer (50 mM 2-(N-morpholino)ethanesulfonic acid, 10 mM CaCl2, and 0.05% Brij-35, pH 6.0) was added. To measure the activity of intracellular MMP-3, cells were washed in PBS and lysed by brief sonication in the assay buffer. About 50 μg of cell lysate protein was transferred to 96 wells, to which the assay buffer was added to a total volume of 199 μL. After incubation for 1 h at 37°C, reaction was started by adding 1 μL (4 μM final concentration) of substrate (Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2). The plates were read continuously in a fluorescence microplate reader (Molecular Devices) over 30 min at Ex/Em = 328/393, and the rate of product formation was determined from the linear range.

Zymographic analysis

Cell lysates containing equal amounts of protein (50 μg) were mixed with 2× Tris-glycine SDS sample buffer and then loaded without reduction or heating onto 12% Tris-glycine gel with β-casein incorporated as a substrate for MMP-3. Samples were subjected to electrophoresis using the Tris-glycine SDS running buffer. Following electrophoresis, the gels were incubated in the renaturing buffer [2.7% (w/v) Triton X-100] for 30 min with gentle agitation to remove SDS. The gels were equilibrated in the developing buffer (50 mM Tris base, 40 mM HCl, 200 mM NaCl, 5 mM CaCl2, and 0.02% (w/v) Brij 35) for 30 min at 24°C, followed by incubation in the fresh developing buffer overnight at 37°C with gentle agitation. The gels were stained with 0.5% (w/v) Coomassie Brilliant Blue R250 [25% (v/v) methanol, 10% (v/v) acetic acid in water] and then destained [25% (v/v) methanol, 10% (v/v) acetic acid in water without Coomassie blue] to visualize the area of protease activity as clear bands.

DA uptake assay

The primary cultured neurons were rinsed with Krebs–Ringer solution and incubated in the same solution containing 1 mM ascorbate, 2 mM β-alanine, and 100 μM pargyline at 37°C for 5 min in a humidified 5% CO2 atmosphere. The neurons were further incubated in the presence of 25 nM [3H]DA (45 Ci/mmol) for 15 min. Control neurons were treated in the same solution but containing the DA uptake blocker, bupropion (10 μM). The cells were then washed with ice-cold Krebs–Ringer solution and lysed with 0.5 M NaOH. Radioactivity was measured using a liquid scintillation counter (Beckman, Fullerton, CA, USA).

Total RNA extraction and RT-PCR analysis

Total RNA was extracted from CATH.a cells using Trizol reagent. RT was performed for 40 min at 42°C with 2 μg of total RNA using 1 unit/μL of superscript II reverse transcriptase. Oligo (dT) 18 was used as a primer for this reaction. The samples were then heated at 94°C for 5 min to terminate the reaction. The cDNA obtained from 0.5 μg total RNA was used as a template for PCR amplification. Oligonucleotide primers were designed based on Genebank entries for mouse MMP-1 (sense, 5′-CTGAGAGCTATGAAGAAGCCCAG-3′; antisense, 5′-TCTGTTAACTGGATGGGATTTGG-3′), MMP-2 (sense, 5′-AGCGTGAAGTTTGGAAGCATC-3′; antisense, 5′-GCTGGTTAACTACAGAGGAGGACAG-3′), MMP-3 (sense, 5′-GATCTCTTCATTTTGGCCATCTCTTC-3′; antisense, 5′-CTCCAGTATTTGTCCTCTACAAAGAA-3′), MMP-7 (sense, 5′-GTGAGGACGCAGGAGTGAAC-3′; antisense, 5′-ACAGGTGCAGCTCAGGAAGG-3′), MMP-9 (sense, 5′-CCTTACCAGCGCCAGCCGAC-3′; antisense, 5′-GAAAGGCGTGTGCCAGAAGG-3′), and β-2M (sense, 5′-GGGAAGCCGAACATACTGAA-3′; antisense, 5′-CGGCCATACTGTCATGCTTA-3′). PCR mixes contained 10 μL of 2× PCR buffer, 1.25 mM of each dNTP, 100 pmol each of forward and reverse primers, and 2.5 units of Taq polymerase in the final volume of 20 μL. Amplification was performed in 38 cycles for 1 min at 95°C, 1 min at 58.5°C, and 1 min at 72°C. After the last cycle, all samples were incubated for an additional 10 min at 72°C. PCR fragments were analyzed on 2% agarose 1× Tris acetate-EDTA gel containing ethidium bromide and their amounts were normalized against β-2M amplified, in parallel. Each primer set specifically recognized only the gene of interest as indicated by amplification of a single band of the expected size.

Preparation and transfection of siRNA

Sense and antisense oligonucleotides corresponding to the following cDNA sequences of mouse MMP-3 were used: AAUUCCAACUGCGAAGAUCCACUGA (#1), UCAAGGAGGUUUCUACGCGUUGAUU (scrambled sequence of #1), AUACCAUCUACAUCAUCUUGAGAGA (#2), and UCUGGCAACUGUAGUGUAUAGAUAU (scrambled sequence of #2). The sense and antisense oligonucleotides were annealed following the manufacturer’s protocol (Invitrogen) to generate double-stranded siRNAs at the final concentration of 20 μM. CATH.a cells grown to 80% confluency in 6-well culture plates were subjected to transfection by adding 5 μL of LipofectaminTM 2000 and 4 μL of 20 μM siRNAs (final concentration 40 nM). After 6 h of incubation, the culture medium was changed and cells were maintained for additional 18 h before analysis.

Data analyses

Comparisons were made using anova and Newman–Keuls multiple comparisons test. < 0.05 was considered statistically significant for all analyses.


MMP-3 is increased in dying DArgic cells

We first determined whether the level of MMP-3 might be altered in DArgic cells that have been insulted. As shown in Fig. 1a, primary cultured rat mesencephalic neurons were exposed to 10 μM BH4 for 24 h, the condition previously determined to exert preferential death of DArgic neurons (Lee et al. 2007). Immunocytochemistry against TH, the marker enzyme for DA cells, revealed that though the untreated control DArgic neurons appeared healthy with long fibers, the BH4-treated neurons lost their fibers considerably (Fig. 1a). Immunocytochemistry against MMP-3 performed on the same preparation revealed a dramatic increase in MMP-3 in response to exposure to BH4 in the TH-positive neurons. Under this condition, non-DArgic neurons have been observed not to be vulnerable (Lee et al. 2007); interestingly, in non-DArgic neurons, though there was a very low but detectable level of MMP-3 immunoreactivity in the untreated culture, the level was not significantly affected by BH4 (Fig. 1a). No non-neuronal cells were present in the culture prepared by our method, as previously demonstrated (Lee et al. 2007).

Figure 1.

 Matrix metalloproteinase-3 (MMP-3) protein is increased in dying dopaminergic cells exposed to tetrahydrobiopterin (BH4). (a) A representative double immunofluorescence staining for tyrosine hydroxylase (TH, red) and MMP-3 (green) in primary cultured rat mesencephalic neurons exposed to 10 μM BH4 for 24 h (scale bar = 10 μm). (b) MMP-3 immunocytochemistry of CATH.a cells exposed to 100 μM BH4 for 24 h (scale bar = 50 μm). (c) MMP-3 western blot of CATH.a cell lysates exposed to 100 μM BH4 for various periods. (d) Densitometric analysis of western blots. **< 0.01; ***< 0.001 vs. time zero. actMMP-3, active MMP-3; proMMP-3, the proform of MMP-3.

To ensure that CATH.a cells behave similarly and to assess the degree of induction, we tested the phenomenon in CATH.a cells. Indeed, the cells exposed to BH4 for 24 h also showed a dramatic increase in MMP-3 immunoreactivity (Fig. 1b). Western blot analysis using an antibody that recognizes both proMMP and actMMP-3 (Fig. 1c) and its densitometric analysis (Fig. 1d) showed that the 55 kDa proMMP-3 band increased by 2.00 ± 0.02- and 3.22 ± 0.03-folds in 3 and 6 h, respectively, and then diminished in 24 h (0.43 ± 0.03-fold). On the other hand, the 48-kDa band corresponding to actMMP-3 was increased by 2.12 ± 0.06-fold in 6 h, and more dramatically in 24 h (5.23 ± 0.34-fold). Therefore, it appeared that proMMP-3 protein was elevated first in response to the cellular stress, which was subsequently cleaved to the active form.

Intracellular activity of MMP-3 is increased in dying DArgic cells

To determine whether the increase in actMMP-3 was accompanied by enhanced catalytic activity, we performed MMP-3 activity assay in CATH.a cells. The results (Fig. 2a) showed that the enzyme activity indeed incremented in a similar time course as actMMP-3. At 6 and 24 h, the enzyme activity was increased by 1.65 ± 0.13- and 1.98 ± 0.06-folds, respectively. Casein gel zymography also confirmed the rise in MMP-3 activity (Fig. 2b). The amount of the 48-kDa band corresponding to actMMP-3 was elevated in time after the BH4 exposure. The 55-kDa band, corresponding to proMMP-3, was also observed to be increased; although the proform would not normally have enzymatic activity, and in situ activation of casein-degrading activity during zymographic analysis was often observed. Densitometric analysis of the zymogram revealed a 2.40 ± 0.01-fold elevation of the 48-kDa band at 24 h (Fig. 2c).

Figure 2.

 Matrix metalloproteinase-3 (MMP-3) activity is increased in CATH.a cells exposed to tetrahydrobiopterin (BH4). (a) MMP-3 enzyme activity assay and (b) zymographic analysis on lysates of CATH.a cells exposed to 100 μM BH4 for various periods. (c) Densitometric analysis of the zymograms; *< 0.05, **< 0.01; ***<0.001 vs. time zero. actMMP-3, active MMP-3; proMMP-3, the proform of MMP-3.

MMP-3 expression is selectively induced among MMPs in dying DArgic cells

Whether the rise in the level of MMP-3 protein might involve induction of MMP-3 gene expression was tested by RT-PCR. Untreated CATH.a control cells revealed barely detectable level of proMMP-3 mRNA (Fig. 3). BH4 treatment for 6 h, on the other hand, resulted in a dramatic increase by 10.65 ± 1.24-fold. To determine whether this induction was specific for MMP-3 among MMPs, we also assessed the mRNA levels of other MMPs. RT-PCR performed on the same RNA preparations revealed that the mRNA levels of MMP-1 and MMP-7, belonging to the collagenase and matrilysin subtypes, respectively, were already present in detectable levels in the untreated cells and were not altered by the BH4 treatment. MMP-9, which belongs to the gelatinase subtype, was barely detectable in both untreated and BH4-treated cells. The finding that MMP-9, whose induction usually depends on certain stimuli (St-Pierre et al. 2003), was not increased in response to the cellular stress exerted by BH4, supported the argument that MMP-3 expression is indeed selectively increased in DArgic cells. We also performed RT-PCR for MMP-2 on the same samples and observed that MMP-2 expression was also unaltered by the BH4 treatment. That MMP-2, which is expressed constitutively in almost all cells and body fluids (Liao et al. 2003), is indeed expressed to a considerable degree in untreated CATH.a cells confirmed the validity of our experimental system.

Figure 3.

 Gene expression of matrix metalloproteinase-3 (MMP-3) is specifically induced among MMP members in response to cell stress. (a) A representative RT-PCR performed on various MMP members from RNA preparation of CATH.a cells treated with 100 μM tetrahydrobiopterin (BH4) for 6 h. (b) Densitometric analysis of the RT-PCR results normalized against β2M and expressed as fold induction compared with respective untreated control. ***< 0.001.

Mechanism for generation of actMMP-3 from proMMP-3 exists inside the cell

The intracellular generation of actMMP-3 from proMMP-3 would require the presence of protease(s) inside the cell, similar to plasmin that cleaves proMMP-3 outside the cell (Yong et al. 1998). We therefore tested for the presence of an intracellular protease responsible for the cleavage. Whether inhibition of serine protease, like plasmin, might block the production of actMMP-3 was determined. The serine protease inhibitors aprotinin, PMSF, and leupeptin were all effective in attenuating the formation of the 48 kDa actMMP-3 (Fig. 4a); the level of actMMP-3 protein in BH4-treated CATH.a cells was increased only by 1.41 ±0.07-, 1.56 ± 0.07-, and 1.97 ± 0.16-folds, respectively, compared to 5.23 ± 0.34-fold in the absence of the inhibitors (Fig. 4b). These inhibitors did not influence the catalytic activity of MMP-3 per se and had no cytotoxic effect at the concentrations used (not shown). Therefore, there appeared to be a serine protease involved in the intracellular cleavage of proMMP-3 to actMMP-3. In the presence of the protease inhibitors, the level of proMMP-3 was not significantly different from the BH4-alone control and the reason for this is currently unknown.

Figure 4.

 Formation of active form of matrix metalloproteinase-3 (actMMP-3) is attenuated in the presence of serine protease inhibitors. (a) A representative western blot against MMP-3 on lysate of CATH.a cells pre-treated with various serine protease inhibitors for 1 h followed by exposure to 100 μM tetrahydrobiopterin (BH4) for 24 h. (b) Densitometric analysis of the western blots expressed as fold increase compared with untreated control; ***< 0.001 vs. untreated control; +++< 0.001 vs. BH4 alone. AP, aprotinin (100 nM); PMSF, phenylmethylsulfonyl fluoride; LP, leupeptin (1 μM); actMMP-3, active MMP-3; proMMP-3, the proform of MMP-3.

Intracellular MMP-3 activity participates in the cell death

Whether the intracellular actMMP-3 might participate in the cell death process itself was then tested. First, we determined whether inhibition of MMP-3 activity might have a protective effect using the MMP-3 inhibitor NNGH (MacPherson et al. 1997). As shown in Fig. 5a, primary cultured rat mesencephalic neurons exposed to BH4 showed a decrease in DA uptake ability, used as an index of membrane integrity of DA neurons, to 60.30 ± 3.10% of untreated control. In the presence of NNGH, on the other hand, the decrease in DA uptake ability was recovered to 86.90 ± 15.80%.

Figure 5.

 Matrix metalloproteinase-3 (MMP-3) activity participates in death of dopaminergic (DA) cells. (a) Primary cultured rat mesencephalic neurons were pre-treated with 1 μM N-isobutyl-N-(4-methoxyphenylsulfonyl) glycyl hydroxamic acid (NNGH) for 30 min followed by 10 μM tetrahydrobiopterin (BH4) for 24 h. Degree of cell damage was assessed by DA uptake assay, expressed as percent of untreated control; ***< 0.001 vs. untreated control, +++< 0.001 vs. BH4 alone. (b) Primary cultured mesencephalic neurons isolated from MMP-3 knockout (KO) mice and their wild-type (WT) animals were treated with 10 μM BH4 for 24 h. Degree of cell damage was assessed by DA uptake assay, expressed as percent of untreated control; **< 0.01 vs. untreated control, ++< 0.01 vs. BH4-treated WT. CATH.a cells were transfected with two different siRNAs against MMP-3 and their scrambled sequences and subsequently treated with BH4. Down-regulation of MMP-3 mRNA and protein in the siRNA-transfected cells was verified by RT-PCR (c) and immunofluoroscence staining (d, scale bar = 40 μm), respectively. (e) CATH.a cells transfected with MMP-3 siRNAs and their scrambled sequences were exposed to BH4 and the extent of cell death was monitored by lactose dehydrogenase activity in the culture medium. Data are expressed as percent of respective untreated control; **< 0.01 vs. respective untreated control, +< 0.05; ++< 0.01 vs. mock-transfected and BH4-treated.

Whether the deficiency in MMP-3 might render the cells more resistant was also tested using primary cultured mesencephalic neurons isolated from MMP-3 KO mice. While BH4 caused a decrease in DA uptake ability to 39.58 ± 8.80% of untreated control in the WT culture, the cells deficient of MMP-3 showed significant resistance (68.51 ± 7.72% of untreated control) (Fig. 5b).

For further verification, we tested the effect of inhibition of MMP-3 expression by utilizing the siRNA technique. Two different siRNA sequences (#1 and #2) and their scrambled sequences (#1ss and #2ss) were designed based on the known mouse MMP-3 sequence (Hammani et al. 1992). As shown in Fig. 5c and d, both siRNA’s were shown to be effective in repressing the induction of MMP-3 expression that normally occured in response to BH4 exposure. In contrast, mock transfection and transfection with scrambled sequences were unable to repress the MMP-3 expression, demonstrating the specificity of our siRNA sequences. Under this condition, we tested whether the sensitivity of these cells to BH4 might be altered. As shown in Fig. 5e, the mock-transfected cells were sensitive to BH4, accompanied by an increase in the released lactose dehydrogenase (LDH) to 202.86 ± 9.79%. On the other hand, the cells transfected with #1 and #2 MMP-3 siRNA’s showed resistance to BH4 exposure; the LDH activity stayed at 116.46 ± 11.78% and 127.44 ± 12.58%, respectively, of untreated control. The cells transfected with the scrambled sequences showed no significant difference from the mock-transfected (178.33 ± 5.63% and 200.54 ± 5.49% of untreated control, respectively).

MMP-3 leads to apoptotic cell death

On the basis of findings that BH4 causes apoptotic death of DArgic cells (Choi et al. 2003a; Kim et al. 2003; Lee et al. 2007) and that lack of MMP-3 activity leads to protection (Fig. 5), it was possible that MMP-3 might play a role in the apoptotic signaling. As shown in Fig. 6a, exposure to BH4 caused DNA fragmentation, one of the final events in apoptosis of primary cultured DArgic neurons, as evidenced by an increase in TUNEL staining. In comparison, the TUNEL reactivity was not detected in these cells pre-treated with NNGH. BH4 treatment was also found to cause a time-dependent increase in the active, cleaved form of caspase 3, another marker of apoptosis, with the maximal elevation of 4.26 ± 0.50-fold observed at 24 h in CATH.a cells (Fig. 6b). This was also evident in primary cultured DArgic neurons, where the immunoreactivity against cleaved caspase 3 was dramatically increased upon BH4 treatment (Fig. 6c). Under this condition, pre-treatment with NNGH caused attenuation of the immunoreactivity against cleaved caspase 3, resulting in a shift in color of the merged image from greenish yellow (BH4-treated) to orange (BH4 and NNGH-treated) (Fig. 6c). The results taken together suggested involvement of MMP-3 in the apoptotic signaling pathway, acting upstream of caspase 3 activation.

Figure 6.

 Matrix metalloproteinase-3 (MMP-3) mediates apoptotic signaling. (a) A representative double immunofluorescence staining for tyrosine hydroxylase (TH) (red) and transferase biotin-dUTP nick-end labeling (green) in primary cultured rat mesencephalic neurons exposed to 10 μM tetrahydrobiopterin (BH4) in the presence or absence of 1 μM N-isobutyl-N-(4-methoxyphenylsulfonyl) glycyl hydroxamic acid (NNGH) for 24 h (scale bar = 8 μm). (b) A representative western blot of cleaved caspase 3 in CATH.a cells exposed to 100 μM for various periods and densitometric analysis data expressed as fold induction of time zero; ***< 0.001 vs. untreated control. (c) A representative double immunofluorescence staining for TH (red) and cleaved caspase 3 (green) in primary cultured rat mesencephalic neurons exposed to 10 μM BH4 in the presence or absence of 1 μM NNGH for 24 h (scale bar = 20 μm). (d) A representative western blot and densitometric analysis of MMP-3 in CATH.a cells treated with SP600125 (1 μM, SP) and/or 100 μM BH4 for 24 h. Densitometric data are expressed as fold induction of untreated control. **< 0.01; ***< 0.001 vs. respective untreated control; +++< 0.001 vs. BH4 alone.

As JNK activation is known to be involved in DArgic neuronal apoptosis (Choi et al. 2004; Wilhelm et al. 2007), we also attempted to assess where in the apoptotic signaling pathway MMP-3 exerts its action in relation to JNK. For this, CATH.a cells were treated with BH4 alone or with BH4 and the JNK inhibitor SP600125 and western blot was performed against MMP-3. The results showed that the JNK inhibitor repressed the BH4-induced increase in actMMP-3 by 67.29 ± 1.06% (Fig. 6d). SP600125 also lowered the proMMP-3 level by 79.50 ± 5.70%. Taken together, the results suggested that the increase in MMP-3 gene expression and consequent elevation of MMP-3 activity occur downstream of JNK activation and in turn leads to activation of caspase 3 in DArgic cells.

actMMP-3 form is released but does not cause cell death from outside

We determined whether the actMMP-3 is ultimately released. For this, CATH.a cells were treated with BH4, and the activity and protein levels of MMP-3 were measured in the medium at different time-points. MMP-3 activity increased with time and reached a 10.43 ± 0.45-fold increase in 24 h (Fig. 7a). Western blot analysis of the medium also showed a dramatic increase in the 48-kDa band corresponding to actMMP-3 that reached a 7.60 ± 0.13-fold increase in 24 h (Figs. 7b and c). The band corresponding to the 55 kDa proMMP-3 was barely detected and did not change under this condition. To eliminate the possibility that this released actMMP-3 might cause cell death, we determined the degree of cell death after adding to the medium the catalytically active recombinant MMP-3 (cMMP-3). As shown in Fig. 7d, no apparent cell death was observed in the range of cMMP-3 concentrations that was at least 10-fold of the concentration of actMMP-3 released into the medium during BH4-induced cell death.

Figure 7.

 Active form of matrix metalloproteinase-3 (actMMP-3) is released from dying dopaminergic cells. CATH.a cells were treated with 100 μM tetrahydrobiopterin (BH4) for various periods and the medium was subjected to (a) MMP-3 enzyme activity assay (expressed as fold increase of time zero) and (b) western blot analysis against MMP-3. (c) Densitometric analysis of the 48 kDa actMMP-3 band on western blots, expressed as fold induction compared with time zero. ***< 0.001. (d) CATH.a cells were exposed to various concentrations of the catalytically active recombinant MMP-3 (cMMP-3) and the extent of cell death was monitored by lactose dehydrogenase (LDH) activity in the medium. > 0.05 vs. untreated control for all concentrations tested.


It has been generally thought so far that the only mechanism for MMP-3 activation is via extracellular cleavage subsequent to secretion as the inactive zymogen. In this study, we demonstrate that the active, cleaved form (actMMP-3) can be produced inside DArgic cells that are under cellular stress. We also show that the increase in intracellular MMP-3 activity can occur via induction of gene expression of proMMP-3 as well as the cleavage to actMMP-3. MMP-3 activity thus raised seems to play an important role in the apoptotic signaling.

Our evidence that actMMP-3 is produced intracellularly includes the following observations. First, the MMP-3 immunopositive 48-kDa band, corresponding to actMMP-3, is dramatically increased in the lysate of DArgic cells exposed to BH4 on western blot. Secondly, MMP-3 enzyme activity is increased in the cell lysate, as determined by MMP-3 activity assay and casein zymography. Thirdly, because the cell culture medium would not contain plasmin, proMMP-3 would not be cleaved to actMMP-3 and a majority of the protein should remain as proMMP-3 if released in this proform. In this study, however, the 48-kDa MMP-3 is primarily detected in the medium after the BH4 treatment. This suggests that MMP-3 is released from the cells as the cleaved 48 kDa form in a cell stress condition. Similar observation has been made with DArgic cells under various apoptosis-inducing conditions, such as in serum-deprived PC12 (Kim et al. 2005b), etoposide-treated SY5Y cells (Y. K. Kim, unpublished data), and MPP+-treated mesencephalic neurons (Y. K. Kim, unpublished data). Fourthly, exposure to protease inhibitors and subsequent analysis of the cell lysate reveals that the generation of actMMP has been inhibited, suggesting that the cleavage occurs intracellularly.

The possibility that actMMP-3 is generated inside the cell is corroborated by cases of some other MMPs. ProMMP-11 (stromelysin-3) (Pei and Weiss 1995) and membrane type 1-MMP/MMP-14 (Pei and Weiss 1996) have been reported to be activated intracellularly by furin, a serine protease that recognizes the basic motif RXXR in the propeptide. Several other MMPs, including MMP-23 (Velasco et al. 1999), MMP-28 (Lohi et al. 2001; Marchenko and Strongin 2001), and other MT-MMPs (Woessner and Nagase 2000; Sternlicht and Werb 2001) have a similar basic motif in their proforms, although direct evidence for their intracellular activation is not available. Although proMMP-3 also possesses the furin recognition sequence (Cao et al. 2005), the cleavage would still leave nine amino acids belonging to the prodomain. We have observed that the cleavage of proMMP-3 by furin does not generate a catalytically active MMP-3 (data not shown). Interestingly, a furin cleavage in the prodomain of proMMP-2 can cause generation of a non-functioning protease (Cao et al. 2005). Therefore, it is possible that a serine protease other than furin is involved in the functional processing of proMMP-3 inside the cell.

Brain cells have been suggested to express both constitutive and inducible MMPs in response to cellular stress (Herron et al. 1986). We observed in this study that MMP-3 is induced by treatment with BH4, an endogenous molecule present in DArgic cells and a source of oxidative stress (Choi et al. 2000, 2003a;Kirsch et al. 2003). On the basis of this observation, it is possible to speculate that the constitutive and inducible forms of proMMP-3 may be differentially processed. That is, while the constitutive proMMP-3 is released and cleaved outside the cell by plasmin (Chandler et al. 1997), the stress-inducible form may be processed to actMMP-3 inside the cell by the action of another protease.

We show in this study a critical role of MMP-3 activity in cell death. This is evidenced by the following findings. First, pharmacologic inhibition of MMP-3 activity can suppress the cellular phenomena that occur in dying DArgic neurons, such as a decrease in DA uptake ability and an increase in the release of soluble protein LDH. Second, deficiency of the MMP-3 gene also attenuated the decrease in DA uptake ability in DArgic neurons. Third, down-regulation of MMP-3 expression by siRNA led to protection. Furthermore, actMMP-3 was found to participate in the apoptotic signaling itself. Pharmacologic inhibition of MMP-3 activity was able to attenuate activation of caspase 3, the executioner enzyme in apoptosis and the increase in DNA fragmentation that occurs during apoptosis. Incubation of cMMP-3 and procaspase 3 revealed that procaspase 3 is not a direct substrate of MMP-3 (not shown). Therefore, MMP-3 seems to cause the cleavage of procaspase 3 indirectly, possibly through activating other proteases.

Dopaminergic neurons are particularly prone to accumulation of reactive oxygen species because of the presence of molecules such as DA, BH4, tyrosinase, monoamine oxidase, iron, TH, as well as the low anti-oxidants and anti-oxidant enzymes in the brain. This elevated reactive oxygen species can lead to activation of JNK, which can in turn induce apoptosis in a number of cells including DArgic cells (Chun et al. 2001). This study shows that induction of MMP-3 gene expression occurs downstream of JNK. As MMP-3 gene is known to have an activator protein-1 (AP-1) element on its promoter (Westermarck and Kahari 1999), it is possible that c-Jun, phosphorylated by JNK, acts as a transcription factor that binds to the AP-1 element of the MMP-3 gene.

The actMMP-3 thus generated intracellularly is also accumulated outside the cells (Fig. 7). This catalytically active form can trigger microglial activation and production of proinflammatory agents including tumor necrosis factor α, interleukin-1β, and superoxide (Kim et al. 2005b, 2007). These cytotoxic molecules can in turn exert damaging effects and lead to additional production of actMMP-3 in neighboring neurons. Indeed, MMP-3 mRNA and protein levels have been shown to be up-regulated by interleukin-1β and tumor necrosis factor α in human corneal epithelial cells (Li et al. 2003). Therefore, MMP-3 activity seems to play roles in DArgic degeneration both intracellularly by mediating apoptosis of DArgic neurons and extracellularly by triggering neuroinflammation.

Overall, our results suggest that MMP-3 plays an important role in the apoptotic signaling, after intracellular processing of proMMP-3 to actMMP-3 under cell stress conditions. In combination with its previously observed role in microglial activation, MMP-3 seems to be a critical molecule that plays a role in the perpetual degeneration of neurons as in Parkinson’s disease. Therefore, MMP-3 can potentially serve as a cellular target against which neuroprotective therapy might be designed.


DHC and EMK made equal contributions. This work was supported by the Brain Research Center of the 21st Century Frontier Research Program of the Ministry of Science and Technology (M103KV010011-07K2201-01110), Korea Research Foundation (KRF-2006-J00800), and Asan Institute for Life Science (07-053; 08-053). It was also supported in part by Parkinson’s Disease Foundation (YSK, MFB) and Michael J. Fox Community Fast Track 2006 (THJ, YSK).