Protein kinase Mζ (PKMζ), an atypical protein kinase C (PKC) isoform, plays a key role in the maintenance of long-term potentiation (LTP), a persistent enhancement of AMPA receptor-mediated synaptic transmission, as well as in the persistence of memory in Drosophila. Because memory impairment in Alzheimer disease (AD) has been attributed to disruption of synaptic plasticity, we investigated the expression and distribution of PKMζ in this disorder. We found that PKMζ accumulated in neurofibrillary tangles (NFTs), whereas conventional and novel PKC isoforms did not. Unlike tau, which is present in all NFTs regardless of location, PKMζ was found in a subset of NFTs restricted to limbic or medial temporal lobe structures (i.e. hippocampal formation, entorhinal cortex, and amygdala), areas implicated in memory loss in AD. Interestingly, PKMζ was not identified in any NFTs in control brains derived from 6 elderly individuals without known cognitive impairment. In medial temporal lobe structures in AD, PKMζ also occurred within abnormal neurites expressing MAP2, GluR1 and GluR2 as well as in perisomatic granules expressing GluR1 and GluR2, suggesting that aggregation of PKMζ disrupts glutamatergic synaptic transmission. Together, these findings suggest a link between PKMζ-mediated synaptic plasticity and memory impairment in AD.
The limbic system, a group of associated structures involved in memory, is a site of early changes in Alzheimer disease (AD) (1-3). Although the mechanism underlying memory impairment in AD is not known, a failure of synaptic plasticity has been proposed (4). Specifically, long-term potentiation (LTP), a widely studied animal model of synaptic plasticity and memory, may be disrupted in AD. LTP can be mechanistically divided into 2 phases: a brief induction phase triggering synaptic enhancement, followed by a sustained maintenance phase, which may represent the physiological substrate of memory storage. Whereas many signaling events have been implicated in induction, including activation of postsynaptic NMDA receptors, increases in intracellular calcium, and stimulation of several protein kinases, the molecular mechanism of maintenance is not as well characterized (5). Recently, however, a specific protein kinase C (PKC) isoform, PKMζ, has been shown to be both necessary and sufficient for maintaining LTP (6).
Clarifying the role of PKC in memory function and dysfunction is complicated, in part, because the PKC family consists of 9 similar genes (7), each with distinct functions. The kinases encoded by these genes are classified by their cofactor activation profiles, which vary according to sequence differences in the kinase regulatory domains. Conventional PKC (α, β, γ) is activated by diacylglycerol (DAG) and calcium, novel PKC (δ, ε, η, θ) by DAG, and atypical PKC (aPKC; ζ and ι/λ) is activated by other lipid second messengers and binding proteins, but not DAG and calcium (8). In general, cofactor binding to the autoinhibitory regulatory domain transiently activates PKC by inducing a conformational change in the enzyme, thus exposing the catalytic domain. In vitro, PKC also can be activated by proteolytic cleavage between the regulatory and catalytic domains, liberating an independent catalytic domain termed PKM. Free of regulatory domain inhibition, PKM phosphorylates substrates in the absence of cofactors. PKMζ, the only stable PKM form in the brain, is not, generated by partial proteolysis, but from a brain-specific PKMζ mRNA transcribed from an internal promoter within the PKCζ gene (9, 10). The PKMζ mRNA is actively transported to dendrites (11) and translated during LTP maintenance (10). Because the PKMζ message does not translate the autoinhibitory regulatory domain of PKCζ, PKMζ is autonomously active, persistently phosphorylating substrates even in the absence of second messengers.
This persistent increase in PKMζ is a critical core molecular mechanism for maintaining LTP in rodents. The increase in PKMζ results in increased phosphotransferase activity that causes and sustains an enhancement of alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor responses for hours (12). Recent work indicates that PKMζ enhances AMPA receptor responses exclusively through regulation of receptor trafficking into the synapse (13). The role of PKMζ in maintaining LTP appears specific because the activity of conventional PKC and novel PKC, or other kinases implicated in LTP induction, such as Ca2+/calmodulin-dependent protein kinase II (CaMKII) and cAMP-dependent protein kinase, is not required during maintenance (6). Furthermore, experiments in Drosophila melanogaster show that overexpression of atypical PKM prolongs a simple form of classic conditioning, suggesting evolutionary conservation in memory storage (14).
The critical role of PKMζ in synaptic plasticity prompted our study of the atypical PKC isoforms in AD brain. Although previous reports have documented the distribution of conventional PKC in AD (15, 16), the relationship of atypical PKC to AD neuropathology has not, to our knowledge, been studied. This article focuses on the expression and distribution of PKMζ in AD, whereas an accompanying report highlights a second atypical PKC isoform, PKCι/λ, and its expression in a spectrum of neurodegenerative disorders (17).
Materials and Methods
Antibodies used in this study are listed in Table 1. The polyclonal antibodies against PKC isoforms (α, βII, γ, PKCζ catalytic and aPKC C-terminal) were generated as previously reported (10, 18, 19). Briefly, peptides corresponding to the C-terminal regions of the PKC isoforms were synthesized (Quality Control Biochemicals, Hopkinton, MA). Additionally, a peptide within the catalytic domain of PKCζ, without homology to PKCι/λ, was generated (10). The peptides were coupled to bovine serum albumin and injected intramuscularly into female New Zealand rabbits. After 1 to 3 boosts at 4-week intervals, rabbit antisera were affinity-purified on peptide-conjugated Sulpholink gel columns (Pierce, Rockford, IL). The mouse anti-PKCι (human) and mouse anti-PKCλ (rodent) both recognize human PKCι/λ protein and showed identical results in this study (Transduction Laboratories, Lexington, KY). Thus, these antisera are referred to collectively as "anti-PKCι/λ" in this article.
Autopsy brain tissue was derived from 8 patients with neuropathologically confirmed AD (20, 21) and 6 individuals (controls) with no known history of dementia (Table 2). The mean age of death of the AD group was 78.4 years (range, 60-91 years) and that of the controls was 75.7 years (range, 60-98 years). As expected in an elderly cohort, neurofibrillary changes were observed in a restricted distribution in the 6 control cases; none of the controls, however, met neuropathologic criteria for the diagnosis of AD (20, 21). Paraffin sections (8 μm) of hippocampus, amygdala, frontal cortex, temporal cortex, parietal cortex, occipital cortex, nucleus basalis, midbrain including dorsal raphe, and pons including locus ceruleus were used for immunohistochemistry. The postmortem intervals (PMI) on 7 AD and 4 control cases ranged from 1.5 to 14 hours (the PMI was not available on one AD and 2 control cases).
For Western blot, fresh-frozen autopsy brain tissue was derived from 4 individuals with a mean age of 83 years (range, 72-88 years) without neurologic disease or neuropathology (provided by the Rush Alzheimer's Disease Center, Chicago, IL). The PMI of these 4 cases ranged from 3 to 7 hours.
Brain tissue from hippocampus, caudate nucleus, and gray and white matter from superior temporal cortex and cerebellum were homogenized in buffer (50 mM 4-[2-hydroxyethyl]-1-piperazineethanesulfonic acid [HEPES] [pH 7.5], 5 mM ethylene diamine tetra-acetic acid [EDTA], 5 mM ethylenebis[oxyethylenenitrilo]tetraacetic acid [EGTA], 5 mM 2-mercaptoethanol) containing protease inhibitors (0.1 mM phenylmethylsulfonyl fluoride, 17 kallikrein U/mL aprotinin, 5 mM benzamidine, 0.1 mM leupeptin) and phosphatase inhibitors (50 mM sodium fluoride, 40 mM β-glycerol phosphate, 10 mM pyrophosphate). Protein concentration was measured by bicinchoninic acid assay (Pierce). Fifteen micrograms of total protein was subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to nitrocellulose, probed with isoform-specific antisera for 1 hour, and developed with BCIP/NBT as described previously (18).
To assess the postmortem stability of PKMζ, levels within hippocampus were measured in 1-month-old Sprague-Dawley rats (n = 16) using quantitative immunoblots at PMIs ranging from 1 to 24 hours. The brains were kept at 4°C. Two additional rat brains were examined at 48 hours after storage at room temperature. No diminution in the level of PKMζ was observed over time (data not shown), indicating stability of the protein.
Immunohistochemistry and Confocal Laser Microscopy
Paraffin-embedded sections (8 μm) were submerged in 10 mM citrate buffer (pH 6.0) and microwaved for 5 minutes. Slides were rinsed in phosphate-buffered saline (PBS) for 5 minutes, treated with 99% formic acid for 2 minutes, and rinsed in PBS. Sections were treated with 3% hydrogen peroxide for 10 minutes, rinsed in PBS for 5 minutes, blocked in 4% normal horse serum for 20 minutes, and incubated in primary antibody in a humidity chamber overnight. Sections then were rinsed in 2 changes of PBS for 10 minutes each and developed using the ABC system (Vector Laboratories, Burlingame, CA), which yields a dark brown reaction product, as described previously (19). Sections were rinsed in water, counterstained with Gill's hematoxylin for 5 minutes, rinsed twice, dipped in bluing solution 10 times, rinsed again, dehydrated, and coverslipped. For controls, sections were treated as previously mentioned with omission of primary antisera or with antisera preincubated for 1 hour with the immunizing peptide at a concentration of 10 μg/mL. As an additional control, sections were incubated in serum (1:200) from the rabbit from which the antisera was derived before immunization.
For quantification of the proportion of aPKC-positive NFTs, some sections were incubated with PHF1 after labeling with specific aPKC antisera, rinsed, incubated in biotinylated secondary (1:200), rinsed, incubated in ABC reagent, and developed with the VECTOR VIP substrate kit (Vector Laboratories), which yields a purple reaction product. The proportion of aPKC-positive NFTs was determined by dividing the number of aPKC-positive NFTs (dark brown reaction product) by the total number of NFTs as determined by PHF1-positivity in 3 20× fields of maximum tangle density in CA1-subiculum, entorhinal cortex, amygdala, midfrontal gyrus, and superior temporal gyrus.
For confocal microscopy, sections were incubated in fluorescein isothiocyanate (FITC)-conjugated donkey antirabbit (1:200) and tetramethyl rhodamine isothiocyanate (TRITC)-conjugated donkey antimouse (1:500; Jackson ImmunoResearch Laboratories, West Grove, PA) for 1 hour after incubation in primary antisera. To minimize nonspecific label, these secondary antibodies had been solid phase-adsorbed by the vendor against a panel of nontarget antibodies derived from other species. To reduce background immunofluorescence, sections were then treated with Sudan black as described (22). Sections were then imaged using a MRC 1024 ES confocal microscope with a krypton /argon laser and analyzed using LaserSharp 2000 (Bio-Rad Laboratories, Hercules, CA). To confirm fluorescent channel specificity, parallel sections treated under identical conditions, except for omission of each primary antibody, were evaluated.
We first confirmed that normal human brain derived from elderly individuals without neurologic disease expresses PKMζ. Western blots with ζ-specific catalytic domain antisera showed a prominent band at 55 kD corresponding to PKMζ in the hippocampus, caudate nucleus, superior temporal cortex, and cerebellum (Fig. 1A). The cerebellum displayed an additional species at 72 kD corresponding to full-length PKCζ. This predominance of PKMζ in forebrain of adult humans is identical to its distribution in rat brain (10). A third minor band was inconsistently observed at 63 kD. The other closely related aPKC, PKCι/λ (the amino acid sequences of the human PKCι/λ catalytic domain and human PKMζ are 83.8% identical), was also expressed in all brain regions tested (Fig. 1B). Consistent with work in rats (10), no truncated PKM form of this isozyme was observed. An antiserum targeting the C-terminus of PKMζ that also recognizes PKCι/λ confirmed the presence of these isoforms in human brain (Fig. 1C). Hence, the distribution of the atypical PKC isoforms in human brain parallels that seen in rodents with PKMζ being the predominant product of the ζ gene in forebrain (10).
PKMζ immunopositivity was found in a subset of NFTs within medial temporal lobe structures (Fig. 2A, B; Table 3), i.e. hippocampus, entorhinal cortex, subiculum, and amygdala. These NFTs were recognized based on their typical fibrillary nature and flame-shaped appearance. Moreover, the PKMζ-immunopositive structures were present in cells exhibiting NFTs on adjacent sections of the same blocks. Interestingly, no PKMζ-containing NFTs were observed in tangle-containing regions outside the limbic system, i.e. frontal, parietal, and occipital neocortex, striatum, nucleus basalis, dorsal raphe nucleus, and locus ceruleus. In contrast, PKCι/λ was found in NFTs in all these regions, whether limbic or extralimbic (Fig. 2C, D). Antibodies against both aPKC isoforms revealed modest, finely granular cytoplasmic immunopositivity within neurons throughout the brain in a manner similar to that described in rodents (19). Senile plaques, on the other hand, failed to label with either antibody.
Within limbic structures, PKMζ immunolabeling identified a subset of tangle-bearing neurons (Table 4). Ghost or extracellular tangles were not immunoreactive for PKMζ (data not shown). Double-labeling experiments with PHF1 confirmed the presence of PKMζ in NFT-bearing neurons (Fig. 3A-C). The accumulation of PKC in tangles was specific to the aPKC subclass. NFTs failed to label with antibodies to isoforms of conventional PKC, confirming previous reports (15, 16). NFTs also were immunonegative using antibodies to novel PKC (data not shown).
Prominent PKMζ label of nonplaque-associated distended neurites was variably observed in the limbic regions containing PKMζ-positive tangles (Fig. 4); many of these neurites were negative for PHF1 (Fig. 5A-C). PKMζ also colocalized with the dendritic marker MAP2 (Fig. 5D-F), supporting the dendritic nature of these neurites.
Given PKMζ's role in modulating AMPA receptors at synapses during maintenance of LTP, we examined the relationship of PKMζ to the AMPA receptor subunits GluR1 and GluR2 in AD. These antisera labeled some of the PKMζ-positive dendrites (Fig. 5G-I) and perisomatic granules around CA1 neurons in the hippocampus (Fig. 6A), consistent with previous reports (25, 26). Interestingly, PKMζ also labeled some of these granules; Hirano bodies also were immunoreactive (Fig. 6B-E).
We then examined autopsy brains derived from elderly individuals with no known history of dementia (n = 5; case no. 13 was excluded because of limited tissue availability). Although these cases failed to meet the neuropathologic diagnosis of AD, they all exhibited sparse to moderate numbers of NFTs in the entorhinal cortex, with 4 of the 5 cases also exhibiting sparse to moderate NFTs in the hippocampus, as is commonly encountered in normal elderly subjects without dementia (2). In contrast to the AD cases in which some NFTs exhibited PKMζ immunoreactivity, no PKMζ aggregation was observed within the NFTs in the normal elderly subjects without dementia, i.e. control cases (Fig. 7A). PKCι/λ immunolabeling, however, was observed in entorhinal tangles in 3 of the 5 control brains (Fig. 7B).
We have shown that, within the family of PKC isoforms, only PKCι/λ and PKMζ are present in NFTs in AD. Furthermore, whereas PKCι/λ antisera label NFTs in multiple sites in AD, PKMζ-positive NFTs are restricted to medial temporal lobe structures associated with memory function. This regional specificity is intriguing in view of the role of PKMζ in the maintenance of synaptic plasticity in animal models of memory (6) and contrasts with the usual widespread distribution of tangle-associated proteins (e.g. tau and other tau-associated kinases) in AD. This unusual topographic pattern of PKMζ-positive NFTs cannot be explained on the basis of selective expression of this molecule within these limbic structures, because PKMζ is widely expressed in the forebrain of rodents (10) and humans (Fig. 1). Although there is no clear explanation for this observation, it is possible that increased PKMζ synthesis occurring during LTP (10) renders those neurons actively involved in memory such as hippocampal pyramidal cells especially vulnerable to PKMζ aggregation. This aggregation may contribute to the pathophysiological basis of memory impairment in AD.
Another unusual finding is the absence of PKMζ-positive NFTs in limbic or medial temporal lobe regions in the brains of 5 elderly individuals without AD (Fig. 7). Although modest numbers of NFTs, some immunopositive for PKCι/λ, were present in hippocampus and entorhinal cortex in these control cases, PKMζ was absent in these tangle-bearing neurons. This observation is unexpected given that the properties of NFTs in aged, nondemented individuals generally recapitulate those found in NFTs in AD. This intriguing finding further strengthens the link between PKMζ aggregation and memory loss and warrants additional study and confirmation.
Additionally, we found PKMζ within perisomatic granules in association with AMPA receptors, key molecules in the expression of LTP (25). AMPA receptors have been demonstrated within perisomatic granules, ubiquitin-containing structures variably associated with Hirano body filaments described in AD and other conditions (23, 24). The coexistence of PKMζ and AMPA receptors within perisomatic granules and abnormally distended neurites within medial temporal structures in AD links both molecules to the dysfunction of synaptic plasticity believed to be important in the disease.
Unlike other kinases that have been associated with NFTs, PKMζ is autonomously active, an essential property enabling this enzyme to maintain the long-term enhancement of synaptic transmission during LTP (6). The sequestration of PKMζ within medial temporal tangles may inhibit the normal activity of this kinase in modulating the trafficking of AMPA receptors at synapses. Together, our data suggest that disruption of the normal function of PKMζ in maintaining LTP may contribute to memory loss in AD. Investigation into the mechanism of PKMζ aggregation may provide a new target for treatment of AD.
The authors thank Dr. Marla Gearing of the Emory Alzheimer's Disease Center, Atlanta, Georgia, and Dr. David Bennett of the Rush Alzheimer's Disease Center, Chicago, Illinois, for providing autopsy brain tissue from AD and control cases. The authors also thank the patients and staff of both centers. Dr. Peter Davies generously shared the PHF1 antisera. Lorraine Braithwaite-Harte, William Oxberry, and Andrew Tcherepanov provided excellent technical assistance.
- protein kinase c
- alzheimer's disease
- alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid
- entorhinal cortex
- long-term potentiation
- neurodegenerative disorders
- neurofibrillary tangles
- neuronal plasticity
- protein isoforms
- protein kinase
- ampa receptors
- synaptic transmission
- temporal lobe
- cognitive impairment
- memory impairment
- hippocampal formation
- mapt gene