Amyloidogenesis in Alzheimer's Disease and Animal Models

Sangram S. Sisodia and Donald L. Price

Alzheimer's disease (AD) is the most common type of dementia in adults (13, 41). The major risk factors for AD are age (13) and, in families with early-onset disease, genetic loci on chromosomes 21 and 14 (16, 23, 42, 50, 57). The disease involves a variety of neuronal circuits in several regions of the brain, including nerve cell populations within the brainstem, basal forebrain, thalamus, amygdala, hippocampus, and neocortex. The consequence of these lesions is deafferentation of targets, particularly in the hippocampus and neocortex (5, 24, 30, 62, 70). Affected neurons develop neurofibrillary tangles (NFTs), neuropil threads, and neurites, all of which reflect abnormalities of the neuronal cytoskeleton (6, 17, 37). A characteristic feature of AD is the presence of deposits of the b-amyloid protein (Ab) in the hippocampus and neocortex. Derived from the amyloid precursor proteins (APPs), Ab is a 4-kD peptide comprised of 11–15 amino acids of the transmembrane domain and 28 amino acids of the extracellular domain of APP (15, 28, 39).

This review, which complements other chapters dealing with aspects of AD, describes the current status of our understanding of (a) the biology of APP, (b) mechanisms that lead to the formation of deposits of Ab, (c) the pathogenesis of amyloidogenesis in the brains of aged nonhuman primates, and (d) the results of attempts to develop transgenic animal models that show ADtype abnormalities (see also New Developments in Dopamine and Schizophrenia and The Effects of Neuroleptics on Plasma Homovanillic Acid, this volume).

Located on the midportion of the long arm of human chromosome 21 (28), the APP gene, encompassing ~ 400 kb of DNA (B. Lamb and J. Gearhart, personal communication), gives rise by alternative splicing of APP pre-mRNA to at least four transcripts that encode Ab containing proteins of 695, 714, 751, and 770 amino acids (18, 28, 31, 46). APP-751 and -770 contain a domain that shares homology with the Kunitz class of serine protease inhibitors (31, 46). In cultured mammalian cells, full-length APP isoforms are modified by the addition of both N- and O-linked carbohydrates and terminal sulfation events (68). Dependent on cell type, levels of newly synthesized APP molecules appear at the cell surface (21), and some of these proteins are cleaved by an enzyme, designated APP secretase, within the Ab sequence (2, 12, 54, 65) to release the ectodomain of APP, including residues 1–16 of Ab, into the medium. Thus, APP cleavage within the Ab domain precludes the formation of Ab. The presence of secreted APP isoforms that contain Ab epitopes in cerebrospinal fluid suggests that similar processing events occur in vivo (44, 68). A fraction of cell-surface APP is also internalized and degraded via endosomal–lysosomal pathways (19, 21). It appears that the sequence NPXY in the cytoplasmic domain of APP is required for internalization by a clathrin-coated-pit-mediated pathway (S. Sisodia, personal observations). Processing via the endosomal–lysosomal pathway results in the production of potentially amyloidogenic C-terminal-containing fragments (19, 21). Moreover, recent reports indicate that peptides similar to Ab (4 kD) and truncated forms of Ab (~3 kD) are secreted to the media of primary cell cultures and various tissue culture cell lines. Ab-related peptides have also been detected in cerebrospinal fluid (52, 53). Although these studies suggest that Ab may be produced and released in vitro and in vivo, it is not clear whether these Ab-related fragments of variable lengths are the source of cerebral Ab (which is principally 42–43 amino acids) found in cases of AD. In any event, the observation that APP mutations in early-onset familial AD invariably flank the Ab sequence indirectly suggests that altered processing of APP is central to the formation of amyloid in these individuals (see below).

Despite advances in our understanding of APP metabolism in cultured cells and the description of APP mutations in early-onset familial AD, little is known regarding the biological function of APP in the brain and its metabolism in in vivo settings. In the central nervous system, APP transcripts and proteins are expressed in most neurons. In many regions of the nervous system, including the neocortex, APPs are present in cell bodies, proximal dendrites, and axons (10, 38). In the rat peripheral nervous system, neuronal APPs are delivered to axons and terminals via the fast anterograde axonal transport system (34), and holo-APP-695 is the predominant transported isoform (55). It has been suggested that APPs may play roles in synaptic adhesion, interactions, and plasticity. Recent studies indicate that a sequence in the APP cytoplasmic domain catalyzes GTP exchange with the trimeric G protein, Go. These results suggest a role of APP as a Go-coupled receptor (43). Additional studies (56, 66) have documented the expression of APP homologues in vertebrates. Unfortunately, relatively little is known about the subcellular distributions, targeting, processing, or functions of APP or the amyloid-precursor-like proteins (APLPs) at pre- or postsynaptic sites. The distribution and fates of transported APP isoforms in the central nervous system can be examined in vivo by labeling APP synthesized by specific populations of neurons followed by examination of APP processing in defined target fields. In transgenic mice, this approach will be invaluable in assessing the processing of overexpressed or mutated APP isoforms in transgenic animals.

Senile plaques, comprised of amyloid deposits in proximity to neurites, are common in amygdala, hippocampus, and neocortex (30) and are a histological hallmark of AD (41). Amyloid, visualized by staining with thioflavin S or Congo red and consisting of 8-nm straight extracellular filaments, typically appears as diffuse parenchymal deposits in the cores of plaque and around blood vessels. The presence of amyloid has also been documented in the brains of aged primates (38, 51, 59, 72) and in the brains of older patients with Down's syndrome (48), as well as in cases of hereditary cerebral hemorrhage with amyloidosis, Dutch type (HCHWA-D).

Several mutations have been described in the APP gene that are linked to early-onset AD (22). Mutations at codon 717 (of APP-770) and 693 (of APP-770) are linked genetically to early-onset familial AD and HCHWA-D, respectively (8, 16, 23, 42). In nine families of early-onset autosomal dominant AD, the valine residue at position 717 is replaced by either isoleucine, phenylalanine, or glycine; mutations occur within the transmembrane domain of APP, two amino acids downstream from the C-terminus of Ab (16). Although the conservative substitution of isoleucine for valine at position 717 does not appear to alter APP processing in transfected cells (7), the effects of isoleucine and other 717 substitutions on APP metabolism in vivo remain to be established. In addition, a mutation in APP that leads to a Glu-Gln substitution at position 693, corresponding to amino acid 22 of Ab, is associated with HCHWA-D, a disease in which Ab is deposited around blood vessels. As with the 717 mutation, the maturation of APP harboring this substitution is indistinguishable from wild-type APP (S. Sisodia, unpublished observations). More recently, a double mutation at codons 670 and 671 has been demonstrated in two large, related, early-onset AD families from Sweden (42). This double mutation results in substitution of the normal Lys-Met dipeptide for Asn-Leu. As yet, the pathology has not been fully documented. Remarkably, Ab-related peptides are secreted at elevated levels from cells transfected with cDNA encoding APPs that harbor this double amino acid distribution (7, 9). Finally, recent studies have shown that the majority (~80%) of early-onset non-Volga German kindreds exhibit linkage to markers localized to 14q24.3 on the distal part of chromosome 14 (50, 57). The gene involved in these early-onset cases is presently unknown, although candidate loci mapping to 14q24 includes genes encoding c-fos and HSP-70. Future research should clarify the relationship of the product encoded by this locus to amyloidogenesis.

The cellular origins of APP that give rise to Ab are beginning to be clarified. Neurons are one likely source (10, 38), and morphological evidence suggests that APP-immunoreactive neurites, often decorated or capped by Ab deposits (10, 38), are one source of parenchymal amyloid. However, others cells, including astroglia, microglia, and vascular cells, may contribute to the formation of Ab and may produce other constituents that colocalize with Ab (1, 35, 58, 67, 73).

With a lifespan of 25–30 years (61), Macaca mulatta develop age-associated impairments in performance on cognitive and memory tasks early in the third decade of life (3). It is likely that these impairments in performance of specific behavioral tasks are associated, in individual animals, with the formation of neurites (32, 38), the deposition of amyloid (51,59), and alterations in neurotransmitter markers (4, 20, 69). In many old animals, Ab appears as diffuse deposits, as the cores of senile plaques, and as congophilic angiopathy (i.e., Ab within the walls of blood vessels) (1, 51, 59, 63, 64). In neural parenchyma, Ab is readily demonstrable in proximity to swollen APP-enriched neurites filled with lysosomes and abnormal membranes (38). It has been speculated that, normally, APP may play an important role in synaptic interactions, and alterations in the biology of APP at synapses could lead to changes in synaptic functions, including synaptic disjunction (a potentially reversible process) followed by irreversible synaptic disconnection. Loss of synapses has been well documented in AD (11, 49, 60). Disconnected axons may then "die back," and retrograde abnormalities, similar to those that occur following axotomy, appear in cell bodies. Subsequently, the organization of cytoskeletal elements in perikarya becomes perturbed, leading to the formation of NFT, a pathology eventually associated with death of neurons. Research during the next few years will identify the influences of other cells (i.e., astroglia, astrocytes, and vascular cells), colocalized proteins [i.e., apolipoprotein E (58)], and constituents of the complement cascade (40) on Ab amyloidogenesis.

Transgenic strategies have been used to test some of the hypothetical mechanisms of amyloidogenesis. Although initial studies designed to examine age- or disease-associated alterations in ratios of different APP transcripts and isoforms did not disclose consistent patterns in these measures (25, 33, 45), one in situ hybridization study suggested that levels of APP-751 mRNA (relative to APP-695 mRNA) were increased in subsets of affected neurons in AD (25). The hypothesis that the overexpression of APP-751 could facilitate the formation of Ab was tested in transgenic mice that expressed a human APP-751 cDNA under the control of a neuron-specific enolase promoter (47). Immunocytochemistry of the brains of transgenic animals was interpreted to suggest a relative increase in APP in neurons and the presence of some extracellular Ab deposits in hippocampus and cortex. Although the present report suggests that increased neuronal expression of APP-751 may promote amyloidogenesis, absolute levels of transgene-derived products or the prevalence of deposits in aged animals are presently unknown.

Transgenic animals were also generated by introduction of a transgene that encodes the 4-kD Ab peptide under the transcriptional control of ~1.8 kb of the human APP promoter (71). Although initial immunocytochemical analyses were interpreted to show small clusters of Ab immunoreactivity in the hippocampus of older mice, it soon became apparent that C57BL/6J mice, a contributor to the Ab transgenic mouse line, develop nonspecifically stained age-related clusters of intracytoplasmic inclusions within astrocytic processes (26).

On the basis of several lines of evidence suggesting that Ab-containing C-terminal APP fragments may exhibit potential neurotoxicity, Gordon and colleagues (29) produced transgenic mice that utilized the human Thy-1 promoter to drive the expression of the C-terminal 100 amino acids of APP. Animals were reported to show neuritic plaques, NFT, and neuronal degeneration in silver-stained preparations similar to those seen in cases of AD (29). However, discrepancies were noted in the original report, and subsequent studies of additional littermates failed to replicate the initial observations.

Neve and colleagues (27) generated transgenic animals that express the C-terminal 100 amino acids of APP under the control of a brain dystrophin promoter. The authors documented (a) accumulations of Ab-immunoreactive deposits in cell bodies and neuropil of the brains of 4- and 6-month-old transgenic animals and (b) the presence of abnormal aggregates of C-terminal epitopes in vesicular structures in the cytoplasm and in abnormal-appearing neurites in hippocampal CA2/3 regions. However, because polypeptide expression has not been documented by these investigators, the relationship between transgene expression and the observed phenotypes is unclear.

In related experiments, Fukuchi and colleagues (14) created transgenes that encode the APP signal peptide fused to APP C-terminal 100 amino acids driven by a cytomegalovirus (CMV) promoter. Cells transfected with this gene showed evidence of cytotoxicity (14). In transgenic mice, the CMV-driven transgene expressed substantial levels of the transgene-derived APP C-terminal 100 amino acids. However, these animals have not shown clinical signs or histopathological abnormalities in any tissues (G. M. Martin, personal communication).

Thus, at the present time, none of the transgenic models reproduce the features of AD-type pathology. It is uncertain whether the relative lack of success is caused by differences between species in their ability to form Ab, the inappropriate selection of transgenes, or insufficient levels of transgenic product expression in cells that have the capacity to generate Ab.

Because it has been difficult to overexpress APP by using conventional transgenic technologies, Lamb et al. (36) have recently used yeast artificial chromosome(s)/embryonic stem cell (YAC-ES) technologies to create a dosage imbalance and overexpression of APP in mice. A 650-kb YAC that contained the entire unrearranged 400-kb APP gene was transferred into ES cells by lipidmediated transfection; ES cells that expressed human APP were introduced into mouse blastocytes to generate a number of chimeric mice. Subsequent breeding efforts resulted in mice that harbor human sequences in the germline. Levels of expression of the human transgeneencoded mRNA and protein in brain and peripheral tissues are approximately equivalent to endogenous levels, and, remarkably, the splicing pattern of human APP transcripts mirrors the stoichiometry spliced products derived from the endogenous mouse APP gene. This approach, which allows the introduction and expression of large DNA fragments, represents a technology that should prove useful in generating animal models of disorders that involve large genes, such as APP, and that has proven intractable for manipulation and expression by standard transgenic technologies.

Over the past 5 years, cellular and molecular biological approaches have significantly advanced our understanding of Ab amyloidogenesis. Research in the area of trafficking and processing of APP in specific cell types, along with in vivo paradigms, will provide additional insights into APP biology. Moreover, the descriptions of mutations in the APP gene in some early-onset familial AD provide compelling reasons to pursue transgenic approaches in which the effects of these mutations on Ab amyloidogenesis can be evaluated.

The authors thank Drs. Lary Walker, Lee Martin, Linda Cork, Cheryl Kitt, Bruce Lamb, David Borchelt, and John Gearhart for helpful discussions.

This work was supported by grants from the U.S. Public Health Service (NS AG 05146, NS 20471) as well as from the American Health Assistance Foundation, the Metropolitan Life Foundation, and the Claster Family Fund. Dr. Price is the recipient of a Leadership and Excellence in Alzheimer's Disease (LEAD) award (AG 07914) and a Javits Neuroscience Investigator Award (NS 10580).

published 2000