Human Disease and the Genetically Engineered Animal Models

Alzheimer's disease (AD) represents a great challenge for science and medicine because of its prevalence, cost, lack of reliable treatments, and often devastating impact on individuals and caregivers. This age-associated chronic illness involves genetic risk factors, a well-defined clinical syndrome with a progressive course, evidence of dysfunction and/or death of populations of neurons, pathological and biochemical abnormalities, and intraor extra-cellular protein aggregates (Price, Tanzi, Borchelt, and Sisodia, 1998). Patients become severely disabled and often die of intercurrent illnesses. There are treatments for symptoms but no cure. However, recent research, particularly in animal models, has begun to provide new insights into the mechanisms of Alzheimer's disease and has identified new targets for therapy.

Clinical-Pathological Features of Alzheimer's Disease

In most cases of AD, the initial impairments of memory and cognition appear gradually during the seventh decade. The accuracy of clinical diagnoses improved from 1980 to 2000, and early diagnosis will become increasingly important as mechanism-based treatments become available.

AD involves the brain (and not other organs) and certain neuronal populations are selectively vulnerable. AD results from the selective degeneration of nerve cells in the brain's regions and neural circuits that are critical for memory, cognitive performance, and personality (Albert, 1996). The dysfunction and/or death of these neurons reduces the numbers of generic and transmitter specific-synaptic markers in their target fields; the disruption of synaptic communication in affected regions/circuits can lead to mental impairments and, finally, severe dementia.

AD typically involves intracellular or extracellular protein aggregates in brain. Neurofibrillary tangles (NFTs), inclusions located within cell bodies and proximal dendrites, are composed of poorly soluble paired helical filaments (PHF), which are, in turn, composed principally of hyperphosphorylated isoforms of tau. PHF are also present in dystrophic neurites, the filamentous swellings of distal axons/terminals (usually seen in proximity to A β deposits). Perturbations related to hyperphosphorylated tau seem to play a role in disturbances in intracellular transport.

The extracellular aggregates in brain are abnormal accumulations of A β, a 4kD β pleated sheet amyloid peptide, derived by β -and γ -secretase cleavages of the amyloid precursor protein (APP). Levels of A β are elevated in brain, and A β monomers form oligomers and multimers that assemble into protofilaments and then fibrils. Eventually, A β fibrils are deposited as the amyloid cores of neuritic or senile plaques, which are complex structures also containing dystrophic neurites, astrocytes, and microglia. Plaques are preferentially localized to the cortex, hippocampus, and amygdala.

The levels and distributions of APP and its cleavage enzymes in neurons lead to the selective appearance of A β in brain. It seems that toxic A β peptides, particularly oligomers, accumulate near synapses and may impair transsynaptic communication, eventuating in the disruption of synaptic connections between neurons and their targets (other nerve cells). Nerve cells are functionally damaged, changes occur in tau phosphorylation; microtubule stability is compromised; intracellular transport processes are impaired; intracellular PHF appear; and cell geometry is altered, with synapses, axon terminals, and dendrites appearing to be most vulnerable. The neuron is incapable of performing its normal functions for a significant interval before the ensuing demise of the cell.

Mutant Genes/Proteins Implicated in Familial Alzheimer's Disease (FAD)

In some individuals with early onset AD, the illness may be inherited as an autosomal dominant with mutations in three different genes: the APP; PS1; and PS2 (Price, Tanzi, Borchelt, and Sisodia, 1998).

APP

Encoded by a gene on chromosome 21, APP is expressed in many cells and tissues but is particularly abundant in neurons. This type-1 transmembrane protein is cleaved endoproteolytically by an enzyme, β -site APP-cleaving enzyme 1 (BACE1), and by an activity termed " γ -secretase," which, in concert, generate the N-and C-termini of the A β peptide, respectively. The levels and distributions of APP and the activities of proamyloidogenic cleavage enzymes, particularly BACE1, in neurons seem to be lead to the formation of A β in brain. The formation of A β 1-40,42 is precluded by the endoproteolytic cleavage of APP within the A β sequence by α -secretase, now thought to be either tumor necrosis factor (TNF) α converting enzyme (TACE) or a disintegrin and metalloproteinase 10 (ADAM 10), which cut between residues 16 and 17 of A β, and by BACE2, a protease sharing features with BACE1, but cleaving APP after residues 19 and 20 of A β (i.e., within the A β domain). These different endoproteolytic cleavages generate various C-terminal peptides, including the APP intracellular domain (C60), which may play a role in the activation of transcription.

A variety of APP mutations, including APPswe (a double mutation at the N-terminus of A β) and APP-717 (near the C-terminus of A β), have been reported in cases of FAD. These mutations, strikingly situated near several secretase cleavage sites, are proamyloidogenic, and cells that express mutant APP show aberrant APP processing: the APPswe mutation, which enhances BACE1 cleavage, is associated with elevated levels of Aß; the APP 717 mutations, which affect γ -secretase activity, lead to a higher secreted fraction of longer, more toxic A β peptides (A β 42) relative to cells that express wild-type APP (Price, Tanzi, Borchelt, and Sisodia, 1998; Citron et al., 1992).

PS1 and PS2

Localized to chromosomes 14 (PS1) and 1 (PS2), respectively, these genes encode highly homologous 43-to 50-kD proteins with multiple transmembrane (TM) domains. Oriented toward cytoplasm are a hydrophilic acidic "loop" region, an N-terminal region, and a C-terminal domain. PS1 is synthesized as an ≈ 42-to 43-kD polypeptide, but the preponderant PS1-related species that accumulate in vitro and in vivo are ≈ 27-to 28-kD N-terminal and ≈ 16-to 17-kD C-terminal derivatives, which accumulate and/or associate in a 1:1 stochiometry and are stable, tightly regulated, and saturable. PS genes are widely expressed at low abundance in the CNS.

PS1 influences APP processing, but it is not clear whether PS1 itself is an aspartyl protease (i.e., γ -secretase), is a cofactor critical for the activity of γ -secretase, or plays a role in trafficking of APP to the proper compartment for γ -secretase cleavage (De Strooper et al., 1998).

The PS1 gene harbors more than fifty different FAD mutations in more than eighty families, whereas only a small number of mutations have been found in PS2-linked families. The vast majority of the abnormalities in PS genes are missense mutations that result in single amino-acid substitutions. However, researchers have found a mutation that deletes exon 9 from PS1 in several different FAD families. The various PS mutations appear to influence γ -secretase activity and increase the generation of the A β 42 peptide.

Transgenic Models of Aβ Amyloidogenesis

Some of the lines of mutant APP mice, although they do not reproduce the full phenotype of AD, represent excellent models of A β amyloidosis and are of great value for testing causal effects of mutant genes, analyses of pathogenic pathways, determination of the molecules participating in A β amyloidogenesis, and identification of therapeutic targets. What follows is a review of selected examples of lines of mice expressing autosomal dominant FAD-linked mutant transgenes.

Mutant APP Mice

Several promoters have been used to drive the expression of an APP minigenes that encode the FAD-linked APP mutants (swe and 717) in strains of mice. The pathology is influenced by the level of transgene product and the specific mutation. The hippocampi and cortices of these mice show elevated levels of A β, diffuse A β deposits, and plaques consisting of dystrophic neurites displayed around an A β core. Astrocytes and microglia are clustered in and around plaques. NFT are not apparent. In some lines of mice there may be mild loss of neurons. Some mice show abnormalities of synapses in hippocampal circuits that precede the deposition of A β. In some lines, mice may exhibit learning deficits, problems in object-recognition memory (related to the number of amyloid deposits in specific regions) and impairments of alternation/spatial-reference and working memory (perhaps related to reductions in synaptic densities in the hippocampus).

APPswe/PS1 Mutant Transgenic Mice

Transgenic mice that coexpress A246E HuPS1 and Mouse/Human-APPswe have elevated levels of A β in the brain and develop numerous amyloid deposits in the hippocampus and cortex (Borchelt et al., 1997; Borchelt et al., 1996). The A β deposits are associated with dystrophic neurites that contain APP, PS1, and BACE1 immunoreactivities; thus, the key participants in amyloidosis appear locally at some of the possible sites of formation of A β.

Gene-Targeted Mice Particularly Relevant to AD

APP and APLP2 Null Mice

Homozygous APP-/- mice are viable and fertile, but they appear to have subtle decreases in locomotor activity and forelimb grip strength. The absence of substantial phenotypes in APP-/- mice may be related to the functional redundancy provided by homologous amyloid precursorlike proteins (APLP1 and APlP2), molecules expressed at high levels with developmental and cellular distributions similar to APP. Consistent with this idea are observations that APLP2-/- mice appear normal but that mice with either both APP and APLP2 targeted alleles or both APLP1 and APLP2 null alleles show significant post- natal lethality.

BACE1-/- Mice

These null mice are viable and healthy, have no obvious phenotype or pathology, and can mate successfully (Cai et al., 2001). In cortical neurons from BACE1 null embryos, there is no cleavage at the +1 and + 11 sites of A β (Cai et al., 2001), and the secretion of peptides is abolished even in the presence transfected mutant APP transgenes. Moreover, A β peptides are not produced in the brains of BACE1 null mice. These results establish that BACE1 is the neuronal β -secretase required to cleave APP to generate the N-termini of A β they further establish that BACE1 is an excellent therapeutic target for drug development for AD.

PS1 and PS2 Null Mice

To examine the roles of PS1 in development, several groups have produced PS1-/- mice (4). Homozygous mutant mice fail to survive beyond the early postnatal period and show severe perturbations in the development of the axial skeleton and ribs (defects in somitogenesis) that resemble a particular Notch1 null phenotype. Because PS1 homologues interact with Notch1, a receptor protein involved in critical cell- fate decisions during development, it is not surprising that cells lacking PS1 show reductions in proteolytic release of the Notch1 intracellular domain (NICD), a cleavage that is thought to be critical for Notch1 signaling. Both wild type and mutant human PS1 trans- genes rescue the spectrum of developmental defects in PS1 null mice. These results indicate that the FAD- linked PS1 variants retain sufficient normal function to allow normal mammalian embryonic development. With regard to the role of PS proteins in A β biology, mutations in PS genes increase the formation of A β 42, and ablation of PS1 reduces the secretion of A β. Significantly, cells from PS1-/- mice show reductions in the levels of γ -secretase cleavage products and levels of A β

Vulnerability of Neurons in Alzheimer's Disease

Among the most challenging mysteries of neurodegenerative diseases is the identification of factors that render neurons susceptible in specific diseases (the principle of selective vulnerability). For example, APP and SOD1 are abundant in many cells. Why, then, do mutations of genes encoding these proteins cause neurological diseases? And why are mutations in APP associated with the development of a dementia syndrome and mutations in SOD1 with an MND phenotype? Recent research has begun to provide exciting new clues concerning the biological basis for vulnerabilities of neurons. AD serves as an illustration. We suggested that the basis for the vulnerabilities of the brain to AD are related to the levels and distributions of APP and its cleavage enzymes in neurons as opposed to other cells (Cai et al., 2001). APP is one of the most abundant proteins in neurons, and available evidence indicates that neurons are the principal source of A β. In nerve cell, APP is transported within axons by the fast anterograde system. APP processing can occur at nerve terminals. In the terminal fields of the perforant pathway, BACE1 cleavage generates soluble C-terminally truncated APPs and amyloidogenic C-terminal fragments. Moreover, in mutant APP transgenic mice, APP, BACE1, and PS1, the key proteins in the formation of amyloid, have been seen in swollen neurites in immediate proximity to A β deposits. These observations conform to the idea that neurons and their processes, including axon terminals, are one source of the APP that gives rise to A β peptide species.

However, the presence of APP in neurons, although necessary, is not sufficient to explain why the brain is particularly vulnerable to A β amyloidogenesis whereas other organs, such as the pancreas, are not β. The patterns of APP-cleavage enzymes in different cell populations are of equal importance. The cellular distributions, relative levels, and sites of APP cleavage of BACE1, BACE2, and α -secretase are principal determinants of such vulnerability. Although both BACE1 and BACE2 are expressed ubiquitously, BACE1 mRNA levels are particularly high in the brain and pancreas, whereas the levels of BACE2 mRNA are relatively low in all tissues except the brain, where it is nearly undetectable. As indicated above, A β is generated by biochemical pathways involving the endoproteolytic cleavages carried out by BACE1 and by an activity termed " γ -secretase," which generate the N-and C-termini of the A β peptide, respectively. Most importantly, BACE1 is the principal β -secretase necessary to cleave APP at the +1 and +11 sites of A β in neurons (Cai et al., 2001). In contrast, BACE2 cleaves APP more efficiently at residues +19 and +20 of APP compared to the +1 site of A β. Significantly, levels of A β 1-19 and A β 1-20 are undetectable in brain. APP can also be cleaved endoproteolytically before residue +17 within the A β sequence by " α -secretases," either TACE or ADAM10. These three cleavages within the A β domain of APP preclude the formation of A β 1-40,42. Because BACE1 is the principal β -secretase in neurons (Cai et al., 2002) and BACE2 may serve to limit the secretion of A β peptides, we suggest that BACE1 is a proamyloidogenic enzyme, wherease BACE2 is an antiamyloidogenic protease, and that the relative levels of BACE1 and BACE2 are major determinants of A β amyloidosis. Significantly, γ -secretase, which may be PS1 (or has its activity influenced by PS), is present in a relatively low level in brain and does not form A β without BACE1 cleavages.

In this model (Cai et al., 2001), the secretion of A β peptides would be highest in neurons/brain as compared to other cell types/organs because neurons express high levels of BACE1 coupled with low expression of BACE2. If the ratio of the level of BACE1 to BACE2 is a critical factor that selectively predisposes the brain to A β amyloidosis, AD would likely involve the brain selectively as opposed to other organs. A seemingly contradictory study shows a high level of BACE1 mRNA expression in the pancreas. Since APP is expressed in the pancreas, why do AD and diabetes mellitus not occur together? It now appears that some of the pancreatic BACE1 mRNAs are alternatively spliced to generate a BACE1 isoform that is incapable of cleaving APP. Taken together with the observations that the pancreas possesses low levels of BACE1 and low amounts of BACE1 activity, these results are consistent with the view that a high ratio of BACE1 to BACE2 activity leads to selective vulnerability of neurons and not pancreatic cells to A β amyloidosis. To test this hypothesis at the level of specific cell populations, it will be important to define the levels and distributions of BACE1 and BACE2 in specific brain regions, circuits, and neurons using specific BACE1 and 2 antisera and to attempt to correlate these measures with the regional vulnerabilities to A β amyloidosis seen in AD.

Treatment in Models of Aβ Amyloidogenesis

Research on model systems relevant to AD illustrates the value of studies of transgenic and genetargeted mice for experimental treatments. Although they do not model the full phenotype of AD, these mutant mice represent excellent models of A β amyloidogenesis and are highly suitable for analyses of pathogenic pathways, determination of the molecular participants in amyloidogenesis, and identification of therapeutic targets. Moreover, they are invaluable for examining the effects of the introduction and/or ablation of specific genes, administration of pharmacological agents (secretase inhibitors), and A β vaccination or passive transfer of A β antibody. Researchers need to assess the efficacy of anti-A β therapies in transgenic mice exhibiting tau pathology.

Bibliography

Albert, M. S. Cognitive and neurobiologic markers of early Alzheimer disease (1996). Proceedings of the National Academy of Sciences of the United States of America 93, 13,547-13,551.

Borchelt, D. R., Ratovitski, T., Van Lare, J., Lee, M. K., Gonzales, V. B., Jenkins, N. A., Copeland, N. G., Price, D. L., and Si sodia, S. S. (1997). Accelerated amyloid deposition in the brains of transgenic mice co-expressing mutant presenilin 1 and amyloid precursor proteins. Neuron 19, 939-945.

Borchelt, D. R., Thinakaran, G, Eckman, C. B., Lee, M. K., Davenport, F. Ratovitsky, F. Prada, C.-M., Kim, G. Seekins, S., Yager, D., Slunt, H. H., Wang, R., Seeger, M. Levey, A. I., Gandy, S. E., Copeland, N. G., Jenkins, N. A., Price, D. L., Younkin, S. G., and Sisodia, S. S. (1996). Familial Alzheimer's disease-linked presenilin 1 variants elevate A β 1-42/1-40 ratio in vitro and in vivo. Neuron 17, 1,005-1,1013.

Cai, H., Wang, Y., McCarthy, D., Wen, H., Borchelt, D. R., Price, D. L., and Wong, P. C. (2001). BACE1 is the major β -secretase for generation of A β peptides by neurons. Nature Neuroscience 4, 233-234.

Citron, M., Oltersdorf, T., Haass, C., McConlogue, L., Hung, A. H., Seubert, P., Vigo-Pelfrey, C., Lieberburg, I., and Selkoe, D. J. (1992). Mutation of the β -amyloid precursor protein in familial Alzheimer's disease increases β -protein production. Nature 360, 672-674.

De Strooper, B., Saftig, P., Craessaerts, K., Vanderstichele, H., Guhde, G., Annaert, W., Von Figura, K., and Van Leuven, F. (1998). Deficiency of presenilin-1 inhibits the normal cleavage of amyloid precursor protein. Nature 391, 387-390.

Price, D. l., Tanzi, R. E., Borchelt, D. R., and Sisodia, S. S. (1998). Alzheimer's disease, genetic studies and transgenic models. Annual Review of Genetics 32, 461-493.

Donald L.Price

Sangram S.Sisodia

Revised byPhilip C.Wong

andDonald L.Price