Matrix metalloproteinases limit functional recovery after spinal cord injury by modulation of early vascular events

Abstract

Inflammation in general and proteinases generated as a result are likely mediators of early secondary pathogenesis after spinal cord injury. We report that matrix metalloproteinase-9 (MMP-9) plays an important role in blood-spinal cord barrier dysfunction, inflammation, and locomotor recovery. MMP-9 was present in the meninges and neurons of the uninjured cord. MMP-9 increased rapidly after a moderate contusion spinal cord injury, reaching a maximum at 24 hr, becoming markedly reduced by 72 hr, and not detectable at 7 d after injury. It was seen in glia, macrophages, neutrophils, and vascular elements in the injured spinal cord at 24 hr after injury. The natural tissue inhibitors of MMPs were unchanged over this time course. MMP-9-null mice exhibited significantly less disruption of the blood-spinal cord barrier, attenuation of neutrophil infiltration, and significant locomotor recovery compared with wild-type mice. Similar findings were observed in mice treated with a hydroxamic acid MMP inhibitor from 3 hr to 3 d after injury, compared with the vehicle controls. Moreover, the area of residual white matter at the lesion epicenter was significantly greater in the inhibitor-treated group. This study provides evidence that MMP-9 plays a key role in abnormal vascular permeability and inflammation within the first 3 d after spinal cord injury, and that blockade of MMPs during this critical period attenuates these vascular events and leads to improved locomotor recovery. Our findings suggest that early inhibition of MMPs may be an efficacious strategy for the spinal cord-injured patient.

Fig. 1.

Time course of MMP-9 activity increases after spinal cord injury. A 3 mm length of spinal cord, centered over the impact site, was flash-frozen and homogenized. Soluble fractions were analyzed by gelatin zymography (A) or by reverse gelatin zymography (B). A, MMP-9 activity increases acutely after spinal cord injury and decreases by 1 week after injury. Note that the absence of MMP-9 activity in the null mouse does not result in a compensatory increase in MMP-2 activity in the injured spinal cord. 1,10-phenanthroline, a general inhibitor of metalloproteinases, completely blocks the inactive and active forms, thus confirming the specificity of these molecules. The positions of migration of active and zymogen forms of MMP-9 and MMP-2 and the MMP-9/TIMP-1 complexes determined from standards are marked.B, TIMP activity, seen by reverse zymography, is unchanged after spinal cord injury. The positions of migration of TIMP-1 and TMP-2 determined from standards are marked, as well as the migration of proMMP-9 and active MMP-9. WT, Wild type.

Fig. 2.

Immunolocalization of MMP-9 in the uninjured spinal cord and at 24 hr after injury. A, B, Uninjured spinal cord. C–F, Injured spinal cord.A–D, Anti-MMP-9 in the wild-type mouse. E, F, Anti-MMP-9 in the MMP-9-null mouse. MMP-9, visualized by HRP immunohistochemistry, is localized in meninges (A,arrowhead) and ventral horn motoneurons (B, arrowheads). After spinal cord injury there is prominent expression of MMP-9 at the lesioned epicenter (C). At higher magnification, MMP-9 is localized within vascular structures (D, arrows), as well as in round cells bearing no processes (D,arrowheads). There is no staining within the epicenter (E) or motoneurons in the adjacent penumbral zone (F, arrowhead) in the MMP-9-null animal. Scale bars: A, C, E, 500 μm; B, D, 50 μm; F, 100 μm.

Fig. 3.

Immunolocalization of MMP-9 at 24–72 hr after injury. Based on double immunofluorescence, MMP-9 (A, B, E) is localized in blood vessels (C, PECAM immunolocalization), macrophages (D, F4/80 immunolocalization), and astrocytes (F, glial fibrillary acidic protein immunolocalization). Controls had no immunofluorescence (data not shown). Scale bars: A–D, 50 μm; E, F, 100 μm.

Fig. 4.

Localization of gelatinolytic activity in situ after spinal cord injury. Unfixed spinal cords from mice (uninjured or at 24 hr after injury) were frozen, and cryosections were prepared for in situ gelatin zymography as described in Materials and Methods. Fluorescence is indicative of gelatinolytic activity. In the uninjured wild-type spinal cord, small amounts of gelatinase activity are identified in the meninges (A, arrowheads). After spinal cord injury, gelatinase activity is prominent in the meninges (B, arrowhead) as well as within the epicenter (C). The gelatinase activity within the epicenter is localized at least in part to blood vessels (D). In the MMP-9-null, injured mouse (E, F), gelatinase activity is not as robust in the epicenter (E). Activity still appears in the meninges (E, F, arrowheads) and blood vessels (F, arrow). Scale bars: A, B, 500 μm; C, 100 μm; D, 50 μm.

Fig. 5.

Blood–spinal cord barrier disruption to HRP at 24 hr after injury in the wild-type (A, C) and the MMP-9-null (B, D) mice. The lateral white matter is characterized by radial spokes of intraparenchymal hemorrhage (A, B, arrows). HRP, appearing as adark brown diffuse reaction product, is more pronounced in the wild-type (C) compared with the MMP-9-null (D) spinal cord. Scale bars: A, B, 500 μm; C, D, 100 μm.

Fig. 6.

Pattern of blood–spinal cord barrier disruption to HRP at 24 hr after injury in MMP-9 wild-type (WT) and null mice and in mice treated with GM6001. The relative intensity of staining for HRP, scaled from 1 to 3, was determined in 18 serial sections centered over the impact site. Within each cross section, 11 regions of the spinal cord, indicated in the schematic, were evaluated. The maximal score, indicative of intense HRP reactivity for any given section, was scored 33. The most pronounced staining for HRP occurred at the epicenter in all animals. There was a marked increase in permeability to HRP in the wild-type mice compared with either the MMP nulls or the wild-type mice treated with GM6001. DC, Dorsal column; DH, dorsal horn; L, peripheral lateral white matter;M, pericentral lateral white matter; V, ventral white matter; VH, ventral horn.

Fig. 7.

Effect of blocking MMPs on permeability to luciferase after spinal cord injury. Abnormal permeability to luciferase was quantified in the epicenter at 24 hr after injury in MMP-9-null (n = 6) and wild-type (WT) (n = 5) littermates and in mice treated with vehicle (n = 4) or GM6001 (begun at 3 hr after injury; n = 6). There is a significant reduction in barrier permeability in the MMP-9-null compared with the wild-type littermates (*p = 0.02) and in drug-treated compared with vehicle controls (*p = 0.04). Values are means ± SD.

Fig. 8.

Effect of blocking MMPs on recruitment of inflammatory neutrophils into lesions after spinal cord injury. According to chloroacetate esterase staining, there appear to be greater numbers of neutrophils in the spinal cord-injured wild-type (A) compared with the MMP-9-null (B) animals at 24 hr after injury. The numbers of neutrophils were quantified within the epicenter of spinal cord-injured mice treated with either GM6001 (n = 4) or vehicle (n = 4) at 24 hr after injury (C). There is a significant reduction in the numbers of neutrophils in mice treated with GM6001 compared with vehicle controls (*p = 0.01). Values represent means ± SD. Scale bars: A, B, 100 μm.

Fig. 9.

Effect of blocking MMPs on locomotor activity after spinal cord injury. Locomotor recovery was evaluated over a 6 week period, using a 21 point scale, in wild-type (WT) (n = 7) and MMP-9-null (n = 7) animals (A) and in GM6001-treated (n = 9), and vehicle-treated (n = 4) animals (B). Both the nulls and GM6001-treated animals exhibited greater locomotor recovery compared with their respective controls. Values represent means ± SD; ∗p = 0.05, ∗∗p ≤ 0.01.

Fig. 10.

Effect of blocking MMPs on preservation of spinal cord white matter. A typical appearance of residual white matter at 42 d after injury, as identified with a Luxol fast blue stain, is shown for representative sections of GM6001-treated (A) and vehicle-treated (B) animals. There is significantly greater preservation of white matter in the GM6001-treated compared with the vehicle-treated animals (C). Values represent means ± SD; *p = 0.01.