Increased Matrix Metalloproteinase-9 Activity in Mild Cognitive Impairment

Abstract

Nerve growth factor (NGF)-dependent cholinergic basal forebrain neurons degenerate during the progression of Alzheimer disease (AD). Elevated proNGF and reduced levels of the TrkA high-affinity NGF receptor occur in prodromal and advanced stages of AD. We recently described a protease cascade responsible for the conversion of proNGF to mature NGF (mNGF) in which matrix metalloproteinase 9 (MMP-9) degrades mNGF in the extracellular space. To determine whether this proteolytic cascade is altered during the progression of AD, we examined human frontal and parietal cortex tissue from aged subjects with a clinical diagnosis of AD, mild cognitive impairment (MCI) or no cognitive impairment (NCI). The analysis demonstrated greater MMP-9 activity in both AD and MCI compared to NCI brain samples (p < 0.01), which supports the notion that a metabolic failure in the NGF-maturation/degradation pathway may be associated with an exacerbated degradation of mNGF in the cerebral cortex in early AD. Moreover, there were inverse correlations between Global Cognitive Score and Mini-Mental State Examination score and MMP-9 activity. These findings suggest that a reduction in mNGF as a consequence of MMP-9-mediated degradation may in part underlie the pathogenesis of cognitive deficits in MCI and AD.

INTRODUCTION

Alzheimer disease (AD) is characterized by a progressive chemical phenotypic down-regulation of specific cellular markers within the cholinergic basal forebrain (CBF) nucleus basalis neurons that provide the major cholinergic innervation to the entire cortical mantle (1). Degeneration of these cholinergic neurons and reduction in cortical choline acetyltransferase activity are associated with cognitive decline in AD (2). The explanation for the selective vulnerability of these CBF neurons remains to be determined. Over the past 25 years one of the most prevailing hypotheses attributed to CBF system degeneration is the loss of trophic support in AD (3). In particular, nerve growth factor (NGF) and its high (TrkA) and low (p75^NTR) affinity receptors, which support the survival and maintenance of these neurons from development to adulthood have been implicated. In this regard, cholinotrophic markers such as the NGF receptors TrkA and p75^NTR decline early as the disease progresses (4, 5) and correlate with cognitive dysfunction, as measured by the Global Cognitive Score (GCS) (4) and the Mini-Mental State Examination (MMSE). In addition to changes in receptor levels, the precursor molecule for the formation of NGF (proNGF), is increased by 40% to 50% in individuals with mild cognitive impairment (MCI) (6) and AD (7).

For over 2 decades it has been assumed that mature NGF (mNGF) accounts for the biological activities (i.e. promotion of cell survival, neurite outgrowth and neuronal differentiation) of this neurotrophin (8, 9). The realization that proNGF plays a key biological role in neuronal activity raises questions about the molecular mechanisms underlying its release, modulation of the proNGF to NGF ratio and the degradation of NGF. Recently, our group described the involvement of a protease cascade responsible for both the maturation of NGF from proNGF to mNGF via the plasminogen/plasmin system and the degradation/inactivation of NGF in the extracellular space by matrix metalloproteinase 9 (MMP-9) (10). We also found an increase in MMP-9 activity in the cortex of patients with AD compared to non-demented aged-matched controls (11). Thus, it is plausible that alterations of the MMP-9 component of this protease cascade might contribute to the vulnerability of the cholinotrophic basal forebrain neurons during the progression of AD.

MMP-9 is a member of the family of Zn²⁺ containing and Ca²⁺ requiring endoproteases capable of degrading compounds of the extracellular matrix (12). MMP-9 is localized and released from neurons, astrocytes and microglia where its expression is regulated by growth factors, cytokines and free radicals (13). MMP-9 induces amyloid-β peptides (Aβ) (14, 15), degrades amyloid β (16, 17) and compact plaques (18), and is released, along with proNGF, in an activity-dependent manner (10). In postmortem AD brain, MMP-9 expression is elevated in neurons, senile plaques, and neurofibrillary tangles and within the vascular wall (19). Additionally, MMP-9 levels are higher in the plasma (20, 21), hippocampus (19, 22) and cerebral cortex (11) of AD patients. Whether the alterations in MMP-9 seen in end stage AD cases occur during the earlier stages of this disease is unknown. The purpose of the present study was to investigate whether the NGF protease degradation process is altered in the cortex of people who died with a clinical diagnosis of MCI, a precursor syndrome to AD. Specifically, MMP-9 activity was analyzed in frontal and parietal cortex harvested from a cohort of brains in which upregulation of cortical proNGF was found to be associated with loss of mild cognitive function (6).

MATERIALS AND METHODS

Subjects

Brain tissues from 29 frontal cortex and 24 inferior parietal cortex samples were harvested from subjects who died with a premortem clinical diagnosis of no cognitive impairment (NCI), MCI, and mild AD (Tables 1, 2). Subjects were participants in the Religious Orders Study (ROS), a longitudinal clinical-pathologic study of aging and AD in older Catholic nuns, priests and brothers (23–25). Sample overlap occurred between 8 frontal and parietal cortex cases. The Human Investigations Committee of Rush University Medical Center approved the study.

Table 1.

Clinical, Demographic, and Neuropathological Characteristics by Diagnosis Category in Frontal Cortex Samples

		Clinical Diagnosis				Comparison by diagnosis group
		NCI (n=8)	MCI (n=11)	AD (n=10)	Total (n=29)
Age at death (years):	Mean ± SD (Range)	81.8 ± 5.6 (72-90)	84.9 ± 5.8 (78-97)	85.7 ± 7.7 (70-95)	84.3 ± 6.4 (70-97)	p = 0.4a
Number (%) of males:		5 (62.5%)	6 (55%)	7 (70%)	18 (62%)	p = 0.9b
Years of education:	Mean ± SD (Range)	19.1 ± 3.0 (16-23)	18.5 ± 3.9 (8-22)	15.5 ± 4.7 (6-21)	17.6 ± 4.2 (6-23)	p = 0.1a
Number (%) with ApoE ε4 allele:		1 (12%)	2 (18%)	4 (40%)	7 (24%)	p = 0.5b
MMSE:	Mean ± SD (Range)	28.4 ± 1.2 (27-30)	26.5 ± 1.9 (24-30)	16.9 ± 5.9 (5-25)	23.7 ± 6.2 (5-30)	p <0.0001a*
Global Cognitive Score (GCS):	Mean ± SD (Range)	0.52 ± 0.32 (0.16-1.15)	0.24 ± 0.24 (−0.25, 0.66)	−1.06 ± 0.57 (−2.22, −0.29)	−0.14 ± 0.81 (−2.22, 1.15)	p < 0.0001a*
Post-mortem interval (hours):	Mean ± SD (Range)	7.8 ± 5.3 (2.8-16)	5.7 ± 4.0 (3-13.9)	6.5 ± 3.3 (3-12)	6.5 ± 4.1 (2.8-16)	p = 0.6a
Distribution of Braak scores:	0	1	0	0	1	p = 0.2a
	I/II	2	1	1	4
	III/IV	4	9	5	18
	V/VI	1	1	4	6
Distribution of NIA Reagan diagnosis (likelihood of AD):	No AD	1	0	0	1	p = 0.8a
	Low	3	4	4	11
	Intermediate	3	7	4	14
	High	1	0	2	3

^aKruskal-Wallis test.

^bFisher's exact test.

^*Pair-wise comparisons with Bonferroni correction showed that there was no significant difference between NCI and MCI, but both were significantly higher than AD.

AD = Alzheimer disease; MCI = mild cognitive impairment; MMSE = Mini Mental State Examination; NCI = no cognitive impairment.

Table 2.

Clinical, Demographic, and Neuropathological Characteristics by Diagnosis Category in Parietal Cortex Samples

		Clinical Diagnosis				Comparison by diagnosis group
		NCI (n=10)	MCI (n=8)	AD (n=6)	Total (n=24)
Age at death (years):	Mean ± SD (Range)	80.9 ± 8.0 (67-92)	84.3 ± 6.8 (75-97)	83.0 ± 6.6 (73-90)	82.5 ± 7.1 (67-97)	p = 0.8a
Number (%) of males:		5 (50%)	3 (37.5%)	3 (50%)	11 (46%)	p = 0.9b
Years of education:	Mean ± SD (Range)	18.1 ± 3.5 (12-23)	17.4 ± 2.3 (14-21)	18.0 ± 3.7 (11-21)	17.8 ± 3.1 (11-23)	p = 0.7a
Number (%) with ApoE ε4 allele:		1 (10%)	3 (38%)	4 (67%)	8 (33%)	p = 0.072b
MMSE:	Mean ± SD (Range)	28.0 ± 1.6 (25-30)	26.1 ± 3.0 (20-29)	11.0 ± 7.7 (0-20)	23.7 ± 7.9 (0-30)	p = 0.0019a*
GCS:	Mean ± SD (Range)	0.54 ± 0.32 (−0.08, 1.15)	0.08 ± 0.20 (−0.25, 0.35)	−1.11 ± 0.76 (−2.10, −0.26)	−0.03 ± 0.81 (−2.10, 1.15)	p = 0.0002a*
Postmortem interval (hours):	Mean ± SD (Range)	8.3 ± 6.0 (2.3-24)	7.2 ± 4.3 (3-13.9)	3.9 ± 1.0 (2.7-5)	6.8 ± 4.8 (2.3-24)	p = 0.066a
Distribution of Braak scores:	0	0	0	0	0	p = 0.2a
	I/II	3	0	1	4
	III/IV	7	8	2	17
	V/VI	0	0	3	3
Distribution of NIA Reagan diagnosis (likelihood of AD):	No AD	0	0	0	0	p = 0.07a
	Low	5	2	1	8
	Intermediate	5	6	2	13
	High	0	0	3	3

^aKruskal-Wallis test.

^bFisher's exact test.

^*Pair-wise comparisons with Bonferroni correction showed a significant difference between AD vs. (MCI and NCI) for the Mini-Mental State examination (MMSE); all 3 groups were significantly different from each other for the Global Cognitive Score (GCS).

AD = Alzheimer disease; MCI = mild cognitive impairment; NCI = no cognitive impairment.

Clinical Evaluation of the Religious Orders Cohort

Details of the clinical evaluation in the ROS cohort have been published elsewhere (25–27). Briefly, investigators performed annual complete clinical evaluations that included an assessment for stroke (28, 29) and parkinsonian (30, 31). Trained neuropsychology technicians performed a battery of cognitive tests, including the MMSE (32), Boston Naming, Word List Memory, Word List Recall and Word List Recognition (33) and Logical Memory (Story A). Many of these tests were slightly modified to facilitate recording and storage by computer (34). Cut-off points for rating impairment on each test were adjusted for education because neuropsychological tests do not measure cognition uniformly across different levels of education. A computer algorithm applied these cut-off points uniformly converting the score of each individual into ratings of impairment in 5 domains of cognition (orientation, attention, memory, language and perception) (35). A score of impairment required impairment on multiple tests within each domain. A board-certified neuropsychologist employed these findings to summarize impairment in each of the 5 cognitive domains as not present, possible, or probable. The neuropsychologist was blind to age, gender, race, clinical data other than education, occupation and information about sensory or motor deficits and effort. Following review of clinical data from that year and examination of all subjects, a board-certified neurologist with expertise in geriatrics made a clinical diagnosis. The diagnosis of dementia and AD followed the recommendations of the joint working group of the National Institute of Neurological and Communicative Disorders and the Stroke and the Alzheimer's Disease and Related Disorders Association (36). Although there are no clear consensus criteria for the clinical classification of MCI (37), the criteria used were similar or compatible with those used by others to describe persons who were not cognitively normal but do not meet accepted criteria for dementia (38, 39). The MCI subjects were defined as those rated as impaired on neuropsychological testing by the neuropsychologist but who were not found to have dementia by the examining neurologist (4, 24, 40) Postmortem interviews were performed to determine medical conditions that occurred during the interval between the last clinical evaluation and death. Subsequently, a consensus conference of neurologists and neuropsychologists reviewed all available clinical findings, medical records and neuroimaging studies, and assigned a summary clinical diagnosis. Each member of the ROS was cognitively assessed using the MMSE and by the GCS measurement, a neuropsychological battery of tests summarized as a composite z-score indicative of cognitive status (27).

Pathological Evaluation the Religious Orders Cohort

At autopsy, brains were processed as previously published (4, 23, 24, 40). The neuropathologist examined the brains and identified atrophy, number and locations of any hemorrhages or infarcts, degree of atherosclerosis and the presence and location of other abnormalities. Cases that exhibited significant non-AD types of pathologic conditions (e.g. tumor, encephalitis, large or multiple lacunar infarcts, Lewy body pathology) were excluded. Brains were cut into 1-cm-thick slabs using a plexiglass jig. One hemisphere was immersion-fixed in 4% paraformaldehyde, as previously described (24). From the opposite hemisphere, slabs alternating around an anchor slab at the level of the crossing of the anterior commissure were either snap-frozen in liquid N₂ or placed in 4% paraformaldehyde. Cortical samples of grey matter were dissected based on standard anatomic landmarks. Frozen dissections were performed on dry ice to prevent tissue thawing. Samples were stored at –80°C prior to biochemical testing.

From the remaining immersion-fixed brain slabs, select brain regions were dissected, paraffin embedded, cut at 8 μm, and stained with hematoxylin and eosin, modified Bielschowsky, thioflavin-S, and with an anti-ubiquitin antibody (41) for complete neuropathological analysis. The neuropathologist making the diagnoses was blinded to the clinical diagnosis. Designations of “normal” (with respect to AD or other dementing processes), “possible” or “probable AD,” and “definite AD” were based on Consortium to Establish a Registry for Alzheimer's Disease (CERAD) Criteria (42). Each case also received Braak score based upon neurofibrillary tangle pathology (43) and a NIA-Reagan Diagnostic criteria (44). The investigators were blinded to the selection process.

Gelatin Zymography

Zymography was used for detection of MMP-2 (gelatinase A, 72 kD) and MMP-9 (gelatinase B, 92 kD) proteolytic activities (45, 46). Differences in molecular weight of proenzymes and activated enzymes allow estimation of relative amounts of proenzymes (pro-MMP-9 running above the 92 kD band at 105 kD) as well as active enzymes in samples. The identity of these bands has been well established (11, 20, 21) and we recently confirmed this using correlative zymograms and Western blots (11).

Frozen grey matter samples from the frontal and parietal cortices were sonicated in ice-cold homogenization buffer (20 mM Tris, 1 mM EGTA, 1 mM EDTA, 10% sucrose, pH 7.4) containing protease inhibitors and S1 fractions were obtained by centrifugation at 1,000 rpm for 10 minutes at 4°C; 60 μg of each sample was loaded on 10% SDS-polyacrylamide gels containing 0.1% gelatin and run with 1x Tris-Glycine SDS running buffer (0.25 M Tris pH 8.6, 1.92 M glycine, 1% SDS, 10x stock solution) to maintain non-reducing conditions. After running, the gels were incubated in zymogram renaturing buffer (Triton X-100, 25% (v/v) in water) with gentle agitation for 30 minutes at room temperature. After decanting the zymogram renaturing buffer, gels were incubated overnight at 37°C with gentle agitation in developing buffer (50 mM Tris-HCl, 0.2 M NaCl, 5 mM CaCl2, 0.02% Brij 35, pH 7.6). The gels were then stained with Coomassie Blue R-250 (0.5% in ethanol 40%) for 30 minutes. Gels were de-stained with destaining solution (Methanol: Acetic acid: water [40: 10: 50]). Areas of protease activity appeared as clear bands against a dark blue background at molecular weights established to correspond to MMP-2 and MMP-9 (11, 20, 21). The degree of gelatin digestion was quantified by densitometry of the scanned gels using MCID image analysis system. Gels were scanned in a grey scale mode and the images were digitally inverted, so that the integration of bands was reported as positive values. The pixel density was determined after background subtraction and the value used to calculate the integrated density of a selected band. Values of integrated density were reported as volume units of pixel intensity per mm². The integrated density of each band was reported as the mean of 3 different measurements of the same gel for each sample run in duplicate.

Statistical Analysis

Demographic and neuropathological variables were compared across clinical diagnostic groups by Kruskal-Wallis test or Fisher exact test; pair-wise comparisons were performed with Bonferroni correction for multiple comparison. Associations between proMMP-9, MMP-9, and MMP-2 activity on the one hand, and demographic and neuropathological variables on the other, were assessed by Spearman rank correlation or Wilcoxon rank-sum test. The level of statistical significance was set at 0.05 (two-sided).

RESULTS

Case Demographics

The 3 clinical groups were comparable in age, sex, educational level, ApoE ε4 allele prevalence, postmortem interval and neuropathologic diagnosis (Braak and NIA/Reagan scores) (Tables 1, 2). MMSE and GCS scores were significantly lower in the AD group than in the MCI or NCI groups (pair-wise comparison, p < 0.01; Tables 1, 2); this was also the case for GCS scores in the frontal cortex sample cases but they were significantly different from each other in the parietal cortex sample cases in which the scores were lowest in AD and intermediate in MCI. AD and MCI subjects did not have co-existing clinical conditions contributing to their cognitive impairments as judged by the examining neurologist.

Frontal Cortex proMMP-9, MMP-9 and MMP-2 Activity

Gelatin zymography of frontal cortex homogenates revealed upregulation in proMMP-9 and MMP-9 activity in MCI and AD groups vs. the NCI group (Fig. 1A). MMP-9 activity of the NCI group (mean ± SD = 76.7 ± 22.6) was significantly lower than those of the MCI (291.4 ± 76.0) and AD (304.7 ± 33.1) groups (p = 0.0002). Similarly, proMMP-9 gelatinolytic activity in the MCI (127.9 ± 38.9) and AD (138.2 ± 40.5) cases was significantly greater than in the NCI (37.7±11) group (p = 0.0002). By contrast, MMP-2 activity was stable across all 3 clinical groups (p = 0.4) (Fig. 1B; Table 3).

Figure 1.

(A) Gelatin zymograph showing an increase in matrix metalloproteinase (MMP)-9 and proMMP-9, but not in MMP-2 activity in the frontal cortex of individuals with mild cognitive impairment (MCI) and Alzheimer disease (AD). (B) Box plots show the significant increased in proMMP-9 and MMP-9 but not MMP-2 activity in frontal cortex in MCI and AD. MMP-9 activity of the NCI group was significantly lower than those of the MCI and AD groups (p = 0.0002). Relative units in the Y-axis are described in Material and Methods

Table 3.

Summary of Matrix Metalloproteinase Activity Levels in Frontal And Parietal Cortex Samples

		Clinical Diagnosis				P-valuea
		NCI	MCI	AD	Total
Frontal cortex:		n=8	n=11	n=10	n=29
proMMP-9	Mean ± SD (Range)	37.7 ± 11.0 (23.3-53.3)	127.9 ± 38.9 (88.6-187.8)	138.2 ± 40.5 (87.0-209.5)	106.6 ± 54.7 (23.3-209.5)	0.0002*
MMP-9	Mean ± SD (Range)	76.7 ± 22.6 (44.3-122.7)	291.4 ± 76.0 (198.9-397.0)	304.7 ± 33.1 (259.0-360.4)	236.7 ± 112.6 (44.3-397.0)	0.0002*
MMP-2	Mean ± SD (Range)	37.5 ± 6.8 (26.6-48.7)	40.2 ± 5.2 (30.5-45.5)	41.7 ± 8.6 (299.9-51.3)	40.0 ± 6.9 (26.6-51.3)	0.4
Parietal cortex:		N=10	N=8	N=6	N=24
proMMP-9	Mean ± SD (Range)	92.2 ± 49.0 (36.8-167.5)	141.3 ± 41.2 (87.5-200.3)	113.0 ± 31.7 (78.8-149.3)	113.8 ± 46.3 (36.8-200.3)	0.15
MMP-9	Mean ± SD (Range)	232.9 ± 81.3 (115.8-326.4)	352.8 ± 47.9 (298.0-410.5)	312.5 ± 41.7 (256.6-354.0)	292.7 ± 81.1 (115.8-410.5)	0.0094*
MMP-2	Mean ± SD (Range)	39.2 ± 5.4 (28.5-45.5)	38.9 ± 7.4 (29.8-50.8)	40.1 ± 8.9 (28.1-50.6)	39.3 ± 6.8 (28.1-50.8)	0.9

^aKruskal-Wallis test.

^*Pair-wise comparisons with Bonferroni correction showed no significant difference between mild cognitive impairment (MCI) and Alzheimer disease (AD) groups; both were significantly higher than the no cognitive impairment (NCI) group. MMP = matrix metalloproteinase; proMMMP-9 = pro-matrix metalloproteinase-9.

In accordance with the above results, frontal cortex MMP-9 activity showed a strong inverse correlation with GCS (Spearman rank correlation, r = −0.63, p= 0.0003; Fig. 2A) and MMSE (r = −0.45, p = 0.013; Fig. 2B), as did proMMP-9 activity (r = −0.47, p = 0.012 for GCS; r = −0.40, p = 0.033 for MMSE; Fig. 2C). MMP-2 levels did not correlate with GCS (Fig. 2D).

Figure 2.

(A–C) Scattergrams show inverse correlations between frontal cortex matrix metalloproteinase (MMP)-9 activity and Global Cognitive Score (GCS) (Spearman rank correlation, r = −0.63, p = 0.0003) (A), frontal cortex MMP-9 activity and Mini-Mental State Examination (MMSE) (r = −0.45, p = 0.013) (B); frontal cortex proMMP-9 and GCS (r = −0.47, p = 0.012) (C). (D) frontal cortex MMP-2 and Global GCS are not correlated (r = −0.27, p = 0.16). Relative units in the Y-axis are described in Material and Methods

Frontal Cortex proMMP-9, MMP-9 and MMP-2 Activity and Neuropathologic Data

Both MMP-9 and proMMP-9 activities also correlated with Braak scores (r = 0.45, p = 0.013 for MMP-9; r = 0.39, p = 0.038 for proMMP-9), but not NIA-Reagan diagnosis. On the other hand, there was an association between frontal cortex MMP-2 activity and cognitive function or neuropathological diagnosis. No other demographic variables were associated with proMMP-9, MMP-9, or MMP-2 activities.

Parietal Cortex proMMP-9, MMP-9 and MMP-2 activity and Clinical Data

Gelatin zymography of parietal cortex homogenates from different case groups shown in Figure 3A revealed that MMP-9 activities in AD (292.7 ± 81.1) and in MCI (353.8 ± 47.9) were significantly greater than that in the NCI (232.9 ± 81.3) group (p = 0.0094, Fig. 3B; Table 3). Although there was an elevation of proMMP-9 activity among the MCI cases, the difference across the 3 groups did not reach statistical significance (p = 0.15), suggesting a higher degree of conversion of proMMP-9 to activated MMP-9. Similar to that reported for the frontal cortex samples, there was little variability of MMP-2 activity in parietal cortex across clinical groups (Fig. 3B; Table 3).

Figure 3.

(A) Gelatin zymography shows greater matrix metalloproteinase (MMP)-9 levels compared to proMMP-9 and MMP-2 levels in the parietal cortex. (B) Box plots show significantly increased MMP-9 but not proMMP-9 or MMP-2 activity in parietal cortex in mild cognitive impairment (MCI) and Alzheimer disease (AD) compared to the no cognitive impairment (NCI) group (p = 0.0094). Relative units in the Y-axis are described in Materials and Methods

Parietal Cortex proMMP-9, MMP-9 and MMP-2 Activity and Neuropathologic Data

In the parietal cortex, the associations between MMP-9 activity, cognition and Braak scores confirmed those in the frontal cortex (r = −0.48, p = 0.020 for GCS (Fig. 4A); r = −0.42, p = 0.046 for MMSE (Fig. 4B); r = 0.45, p = 0.027 for Braak scores). Neither proMMP-9 nor MMP-2 showed similar correlations.

Figure 4.

(A, B) Scattergrams showing inverse correlations between parietal cortex matrix metalloproteinase (MMP)-9 activity and Global Cognitive Score (Spearman rank correlation, r = −0.48, p = 0.020) (A) and parietal cortex MMP-9 activity and Mini-Mental State Examinaiton (MMSE) (r = −0.42, p = 0.0046) (B). Relative units in the Y-axis are described in Material and Methods.

DISCUSSION

We demonstrate an increase in MMP-9 activity during the progression of AD that correlates with cognitive impairment measured by GCS and the MMSE tests. Since MMP-9 plays a key role in degrading mNGF in the CNS (10), the findings suggest that the NGF protease degradation cascade plays a prominent role in the dysregulation of the NGF system during the prodromal stage of AD.

Unlike previous studies using the ROS cohort (27, 47), case selection for this study was based upon premortem clinical classification and not postmortem neuropathological evaluation. For those studies, the biochemical data were evaluated across clinical groups, cognitive test data and subject demographics. A secondary statistical analysis was then performed to correlate clinical and demographic data with the neuropathologic data derived for each case including Braak scores, NIA-Reagan and CERAD evaluations. In the present study, postmortem neuropathological evaluation revealed that half or more cases in the NCI group had Braak scores of III-IV and or NIA-Reagan criteria of intermediate likelihood of AD. This discrepancy between clinical and neuropathological diagnosis in people without dementia has been reported by our group using ROS cases and by others (5, 40, 48). In fact, a recent multi-center study of non demented aged people revealed that of the 97 cases examined, 41% fell into the CERAD criteria of possible, probable or definite AD and 19% were classified with a NIA-Reagan criteria of intermediate to high AD and 23% had a Braak score of III-IV (limbic stage) (49). These findings support the fact that several of the frontal and parietal cases we examined were classified as Braak stage IV and fell within the range of intermediate to high likelihood of AD according to NIA/Reagan criteria (Tables 1, 2). In the present study, slightly less than 50% of our NCI cases were Braak stage 111-1V but most were rate in the low to intermediate stage of AD according to the NIA-Reagan criteria. Interestingly, MMP-9 and proMMP-9 activities were also correlated with Braak neurofibrillary tangle scores but not with NIA/Reagan diagnoses; this stresses a strong correlation between NFT topography determined by Braak scores and neuritic plaque density (50). NFT pathology is also found in the cholinergic neurons of the nucleus basalis, which innervate the entire cortical mantle during the early stages of AD (51, 52); neuronal loss is in seen in this nucleus at this stage of the disease (23), despite an increase in MMP-9 activity. Since NCI subjects have NFT pathology similar to the MCI and AD (49), it is possible that neuronal tau accumulation is not a necessary precondition for cell dysfunction early in the disease state but that the staging of cytoskeletal and amyloid lesions (53) play a crucial interactive role in the cellular deficits leading to the reduction of cortical MMP-9. Alternatively, molecular and protein pathologic studies indicate gene expression dysfunction of the high affinity TrkA signal transduction NGF receptor within nucleus basalis cholinergic neurons (2, 54), as well as within the cortical projection sites of these neurons in cases of MCI (55). Indeed, growth factors appear in the corona of plaques prior to the appearance of phosphorylated tau protein (56). These observations support the possibility that there is an axonal transport defect which interacts with the formation of early tau protein aggregation (53) and/or an alteration in the ratio of 3 repeat to 4 repeat tau expression (57) within cholinergic neurons of the nucleus basalis during the onset of AD, Perhaps these combinational pathologic events trigger the up regulation of MMP-9 early in the disease process. Moreover, whether NCI cases with a Braak stage III-IV may constitute pre-symptomatic AD (49) remains to be determined.

The consequences of the upregulation of MMP-9 during the progression of AD remain to be elucidated. One possibility is that the increased levels of MMP-9 and of its proteolytic activity diminish the levels of available mNGF. Alterations in the levels of MMP-9 might affect CBF neuron phenotype or even survival by changing the balance of the NGF receptors TrkA and p75^NTR. Clinical and pathological investigations indicate that the dysregulation of NGF and its cognate receptors underlies CBF neuron system dysfunction during the onset of AD (55). mNGF binds to the TrkA receptor, which stimulates signal transduction pathways mediating the majority of the survival and growth effects of NGF (58). NGF also binds to the p75^NTR receptor, which is a positive modulator of NGF/TrkA binding (58). p75^NTR also has apoptotic or cell death actions that depend upon its interaction with various receptor chaperones, however (59–61). The physiological consequences of TrkA and p75^NTR signalling may depend upon their interactions with proNGF (62). We previously demonstrated that cortical levels of TrkA decrease (55), whereas proNGF levels increase in subjects diagnosed with MCI or mild AD compared to age-matched cognitively intact controls (6). These observations together with the present findings support the hypothesis that alterations occur in the metabolic pathways regulating the maturation and degradation of NGF early in the development of AD (11). The observations are also in line with the notion that a protease cascade converts proNGF to mNGF and that MMP-9 degrades extracellular mNGF by a process that is coordinated by activation of plasminogen to plasmin by tissue plasminogen activator and regulated by neuroserpin (10).

The increase in MMP-9 seen in MCI may be another example of the shifting balance of select components of the NGF cascade, which determines whether CBF neurons degenerate during the course of AD. In this regard, the increase in MMP-9 may alter the ratio of proNGF to mNGF by enhancing the degradation of mNGF in a context where pro-NGF conversion to mNGF is already diminished (11). The molecular consequence of this shift in the balance from mNGF to proNGF may result in pro-apoptotic signalling via binding to the p75^NTR receptor with proNGF, which could promote neuronal apoptosis (60) perhaps involving a p75^NTR/sortilin mediated mechanism (62). Hence, increased MMP-9 and proNGF concomitant with reduced TrkA early in the progression of AD may result in a shift away from cell survival to proNGF apoptotic signalling. It is also important to note that TrkA reduces and p75^NTR activates ß-secretase cleavage of the amyloid precursor protein (APP), which requires NGF binding and activation of the second messenger ceramide (63). Aging itself activates Aß generation in the brain by ‘switching’ from TrkA to the p75^NTR (63), suggesting that NGF receptor balance is a molecular link between normal brain aging and AD in relation to amyloid processing. The role MMP-9 plays in the activation of Aß peptide generation is an intriguing question and suggests a role for proNGF in this process.

The source of increased cortical MMP-9 activity in MCI and AD is unknown. It is possible that it could arise from neurons, astrocytes and/or activated microglia and its expression is regulated by a variety of growth factors, cytokines and free radicals (13). Following exposure to Aβ peptides, hippocampal and astrocytic cell cultures show upregulated MMP-9 (14, 15), suggesting that MMP-9 contributes to extracellular Aβ clearance by promoting Aβ catabolism (18, 64). On the other hand, in response to Aβ exposure, activated microglia release proinflammatory cytokines (e.g. tumor necrosis factor, interleukin-1ß and interleukin-6 [65, 66]) that also regulate MMP-9 expression (13). Moreover, Aβ-induced iNOS expression results in an overproduction of nitric oxide, which reacts with superoxide to yield peroxynitrite and that may increase the overall free radical burden in Aβ-loaded regions of the AD brain (67-69). In support of this, aberrant expression of MMP-9, iNOS and nitrated and oxidized proteins (70, 71), including proNGF (11), have been colocalized in Aβ burdened brain regions in AD. The chronic neuroinflammatory cascade produced by a local neuronal insult caused by the presence of Aß may initiate a vicious cycle (i.e. an autotoxic loop) that maintains and amplifies the inflammatory cascade and MMP-9 activity, thereby contributing to increased mNGF degradation. This cycle may ultimately impact CBF neuron dysfunction characteristic of the prodromal (i.e. MCI) to later AD stages.

Interestingly, increased cortical MMP-9 activity correlated with mental decline. Previous studies report that reduced cortical TrkA levels are positively associated with lower cognitive performance (55) and increased cortical proNGF levels negatively correlate with cognitive impairment (6). Thus, the concomitant accumulation of cortical MMP-9 and proNGF, along with the reduction of TrkA, may represent a group of early pathobiological markers of the onset of AD. In this regard, increased NGF cerebrospinal fluid levels are detectable in AD (72) and plasma MMP-9 was increased in MCI and AD (21).

In conclusion, this report provides evidence for an upregulation of activated MMP-9 signalling with the potential consequent exacerbated degradation of mNGF early in the MCI stage of AD. This upregulation of MMP-9 in cerebral cortex clearly differentiates MCI from NCI populations and their values are correlated to the degree of cognitive impairment.