CA2040077A1 - Recombinant app minigenes for expression in transgenic mice as models for alzheimer's disease - Google Patents

Abstract

ABSTRACT OF THE DISCLOSURE
Functional recombinant APP minigene constructs and their introduction into the germline of transgenic mice. Such transgenic mice are useful to generate models of Alzheimer's disease and pathogenesis and are also useful to identify molecular mechanisms of the pathogenesis of Alzheimer's disease.

Description

2 ~ 7 BAC~GROUND OF THE INVENTION
This invention relates to recombinant gene constructs, minigene constructs, and transgenic mice for phenotypic expression of Alzheimer-like pathology. The invention further relates to transgenic animal models ior Alzheimer's disease. In particular, the present in~ention provides a variety of minigene constructs which include all or portions of the coding sequences of the amyloid pre~ursor proteins, and which can be expressed in a cell and tissue in a specific manner in transgenic mice carrying the minigene constructs.
Alzheimer's disease (AD) is the most common single cause of dementia in late life. Individuals with AD are characterized by progressive memory impairments, loss of language and visuospatial skills and behavior deficits (McKhann et al., 1986, Neurology 34:
939-944). The cognitive impairment of individuals with AD is the result of degeneration of neuronal cells located in the cerebral cortex, hippocampus, basal forebrain and other brain regions (for reviews, see Kemper, in Clin. Neurol. Aging, M.L. Albertj ed., pp.
9-52, Oxford University Press, New York, 1984; Price, 1986, Annu.
Rev. Neurosei. 2: 489-512). Histologic analyses of AD brains obtsined at autopsy demonstrated the presenee of neurofibrillary tangles (NFT) in perikarya and axons of degenerating neurons, extracellular neuritic (senile) plaques, and amyloid plaques inside and around some blood vessels of affected brain regions (Alzheimer, 1907, Allg. Z. Psychiat. u. Psych. Gerichtl. Med. 64: 146-148).
Neurofibrillary tangles are abnormal filamentous structures containing fibers (about lO nm in diameter) that are paired in a helical fashion, therefore also called paired helical filaments (Kidd, 1963, Nature 197: 192-193; Wisniewski et al., 1976, J. Neurol.
Sci. 27: 173-181; Selkoe et al., 1982, Scienee 215: 1243-124S; Brion et al., 1985, J. Submicrosc. Cytol. 17: 89-96; Grundke-Iqbal et al., 1986, J. Biol. Chem. 261: 6084-6089; Wood et al., 1986, Proe. Natl.
Acad. Sci. USA 83: 4040-4043; Kosik et al., 1986, Proc. Natl. Acad.
Sei. USA 83: 4044-4048; Goedert et al., 1988, Proe. Natl. Acad. Sci.
USA 85: 4051-4055; Wischik et al., 1988a, Proc. Natl. Aead. Sci. USA

, `` 2 ~ 7 ~

85: 4884-4888; Wischik et al., 1988b, Proc. Natl. Acad. Sci. USA 85:
4506-4510). Neuritic plaques axe located at degenerating nerve terminals (both axonal and dendritie), and contain a eore eomposed of amyloid protein fibers (Masters et al., 1985a, EMBO J. _: 2757-2763; Masters et al., 1985b, Proe. Natl. Aead. ~ei. USA 82: 4245-4249). Cerebrovaseular amyloid protein material is found in blood vessels in the meninges and the eerebral eortex (Glenner and Wong, 1984a, Biochem. Biophys. Res. Commun. 120: 885-890; Glenner and Wong, 1984b, Biochem. Biophys. Res. Commun. 122: 1131-1135; Wong et al., 1985, Proe. Natl. Acad. Sci. USA 82: 8729-8732).
During the past se~eral years, primary pathological markers associated with AD have been characterized. The biochemieal analyses of three forms of Alzheimer brain lesions (for reviews, see Kemper, supra; Wurtman, 1985, Sci. Amer. 252: 62-74; Katzman, 1986, N. Engl.
J. Med. 314: 964-973; Priee, 1986, supra; Selkoe, 1989, Ann. Rev.
Neurosei. 12: 463-490; Muller-Hill and Beyreuther, 1989, Ann. Rev.
Bioehem. 58: 287-307), tangles, neuritie plaques, and eerebrovaseular plaques, has revealed protein sequence information, and has facilitated subsequent cDNA eloning and chromosomal mapping of some of the corresponding genes. Immunological studies have identlfied several eandidates for protein eonstituents of the paired helieal filaments (PHF), including mierotubule-associated protein 2 (MAP-2), tau, ubiquitin and the amyloid protein (A4). Degenerating nerve eells express speeifie antigens sueh as A68, a 68 kDa protein. This abnormal antigen is deteetable with the monoclonal antibody ALZ-50 (Wolozin et al., 1986, Seienee 232: 648-650; Wolozin et al., 1987, Ann. Neurol. 22: 521-526; Wolozin et al., 1988, Proe. Natl. Acad.
Sei. USA 85: 6202-6206).
A eentral feature of the pathology of AD is the deposition of amyloid protein within plaques. The 4 kDa amyloid protein (also referred to as A4 (APC, ~-amyloid or BAP) ~s a truneated form of the larger amyloid preeursor protein (APP) which is encoded by a gene loealized on chromosome 21 (Goldgaber et al., 1987, Seienee 235:
877~880; Rang et al., 1987, Nature 325: 733-736; Jenkins et al., 1988, Bioehem. Biophys. Res. Commun. 151: 1-8; Tanzi et al., 1987, Science 235: 880-885). Genetie linkage analysis, using DNA probes .

'~ :
' ' ' ' ~:

, that detect restriction fragment-length polymorphisms (RFLPs, Botstein et al., 1980, Am. J. Hum. Genet. 32: 314-331), has resulted in the localization of a candidate gene (FAD, familial AD) on human chromosome 21 in families with high frequencies of AD (St. George-5 Hyslop et al., 1987a, Science 235: 885-890). However, the FAD locus has not been localized precisely, ant very little is known about its function. Initial studies of individuals with Down syndrome (DS), caused by trisomy of part or all of chromosome 21, indicate that these individuals develop Alzheimer-like pathology beyond the second decade of life. However, analysis of multiple Alzheimer pedigrees revealed that the APP gene does not segregate with familial AD (Van Broeckhoven et al., 1987, Nature ~: 153-155; Tanzi et al., 1987, Nature 329: 156-157). Furthermore, two recent studies with new families demonstrated the absence of a linkage of chromosome 21 15 markers to familial AD (Schellenberg et al., 1988, Science 241: 1507-1510; Roses et al., 1988, Neurology 38: 173).
Age, genetic elements, and possibly environDental factors appear to contribute to cellular pathology of AD. A fundamental but unanswered question in the pathogenesis of AD is the relationship between abnor~alities of neurons and the deposition of amyloid.
Specifically, the cellular origin of pathological events leading to the deposition of amyloid fibrils ad~acent to some areas of the blood-brain barrier (cerebrovascular amyloid) and in the proximity of nerve terminals (neuritic plaques) in specific brain regions as well as extracellular amyloid in plaque cores is not known. Glenner and Wong have described the purification and characterization of meningeal amyloid from both brains of individuals with AD (Glenner and Wong, 1984a, supra) or DS (Glenner and Wong, 1984b, su~ra) and determined the N-terminal peptide sequences. Among 24 residues analyzed, the two amyloid peptides showed only one difference, namely, at amino acid position 11 (glutamine in AD amyloid versus glutamic acid in DS amyloid) among 24 residues analyzed. Subsequent studies of amyloid from Alzheimer brain plaque cores revealed amino acid sequences identical to the reported DS cerebrovascular amyloid 35 data (Masters et al. 1985b, Proc. Natl. Acad. Sci. USA 82: 4245-4249). Copy-DNA analysis of APP tran~cripts from both normal tis8ue ~' ' ' .

-s-and Alzheimer brain material demonstrated the presence of the codon for glutamic acid at this position (Kang et al., 1987, supra;
Goldgaber et al., 1987, supra; Robakis et al., 1987, Lancet: 384-385; Tanzi et al., 1987, Science ~ 880-884; Zain et al., 1988, 5 Proc. Natl. Acad. Sci. USA 85: 929-933; Vitek et al., 1988, Mol.
Brain Res. _: 121-131).
The availability of protein sequence information from the amyloid protein in Alzheimer brains enabled the design of synthetic oligonucleotides complementary to the putative messenger RNA
transcripts. Four groups independently reported successful cloning of cDNAs including the region of the amyloid protein sequence (Goldgaber et al., 1987, su ra; Kang et al., 1987, su~ra; Robakis et al., supra; Tanzi et al., 1987, supra). One gro~p (Kang et al.) cloned the apparent full-length transcript (approximately 3.4 kb) for APP from a human fetal brain cDNA library. The 695-residue amyloid precursor protein (APP-695) shows typical features of a glycosylated cell-surface transmembrane protein. The C-terminal 12 to 14 residues of the A4 protein reside in the putative transmembrane domain of the precursor and 28 N-terminal residues are in the "extracellular domain" (Dyrks et al., 1988, EMBO J. 7: 949-957).
Genomic mapping localized the APP gene on human chromosome 21 using human/rodent somatic cell hybrids (Goldgaber et al., 1987, suDra;
Kang et al., 1987, suora; Tanzi et al., 1987, su~ra). Applying in situ hybridization techniques, this gene was sublocalized to chromosome 21q21 (Robakis et al., supra) and more recently at the border of 21q21-22 (Blanquet et al., 1987, Ann. Genet. 30: 68-69;
Patterson et al., 1988, Proc. Natl. Acad. Sci. USA 85: 8266-8270~.
Chromosome 21 has been the subject of intensive studies because of its involvement in DS (trisomy 21). While 95% of individuals with DS are trisomic for the entire chromosome 21, 2-3~ are mosaics, i.e., trisomic in only some cells, and 3-4~ are caused by triplication (translocation) of the distal part of the long arm (21q22) of chromosome 21 (Crome and Stern, 1972, Patholoey of Mental Retardation, Churchill Livingstone, Edinburgh). The occurrence of such translocations has led to the conclusion that DS can be attributed to trisomy of the distal part (the "pathological region~) , 2 ~

of chromosome 21 (Summitt, 1981, in TrisomY 21 (Down Syndrome~:
Research Prospectives, de la Cruz and Gerald, eds., pp. 225-235, University Park Press, Baltimore). To date, it is not known precisely where the breakpoint on the q arm of chromosome 21 is located, and it is not known whether individuals with DS, who have partial trisomy, develop Alzheimer pathology. In this context, it will be of particular interest to determine if the APP gene maps within the ~pathological region" of chromosome 21. The localization of the APP gene on the long arm of chromosome 21, together with the apparent development of AD pathology in individuals with DS, provides a potential mechanism for the formation of amyloid on the basis of over-expression of a number of genes on chromosome 21, including the APP gene and the FAD gene locus. Initial studies of genomic DNA from sporadic (non-familial) AD cases and "karyotypically normal"
individuals with DS have implicated the presence of microduplication of a segment of chromosome 21 including the APP gene (Delabar et al., 1987, Science 235: 1390-1392; Schweber et al., 1987, Neurology 37:
222). However, subsequent analyses of large numbers of individuals with AD by several laboratories has not confirmed these findings 20 (Tanzi et al., 1987c, Science ~: 666-669; St. George-Hyslop et al., 1987b, Science 2~: 664-666; Podlisny et al., 1987, Science 238: 669-671; Warren et al., 1987, Genomics 1: 307-312).
Chromosomal mapping experiments, using human APP probes in human/rodent cell hybrids, have shown cross-hybridization with mouse and hamster genomic DNA (Kang et al., 1987, su~ra; Tanzi et al., 1987a, supra; Goldgaber et al., 1987, su~ra). Southern-blot analysis of DNA from various species has indicated that the APP gene is highly conserved during evolution. Comparison of the mouse APP sequence (Yamada et al., 1987, Biochem. Biophys. Res. Commun. 158: 906-912) 30 with the sequence from rat (Shivers et al., 1988, EMB0 J. 7: 1365-1370) shows 99% homology on the protein level; furthermore, the human sequence is 96.8% homologous to the mouse sequence and 97.3%
homologous to the rat sequence. Based on the striking conservation of APP proteins, Yamada et al., ~yp~_, have calculated the evolutionary rate of changes at the amino acid level to be 0.1 x 10-9/site/year, which is comparable to that of cytochrome C, and 2 ~ 4 ~

suggests an essential bLological function for APP proteins.
Recently, K. White and colleagues have cloned a Drosophila gene (vnd locus) which is highly homologous to large regions of the APP
sequence. Northern-blot experiments have confirmed these data at the level of mRNA and have demonstrated for various m = alian species the ubiquitous expression of APP transcripts in a number of different tissues (Manning et al., 1988, Brain Res. 427: 293-297).
Kang et al., supra, reported the presence of two distinct bands (-3.2 kb and -3.4 kb) by Northern-blot analysis of human fetal brain mRNA using APP cDNA as a probe. This finding suggests either differential splicing of mRNA or alternative usage of polyadenylation sites. Both post-transcriptional events were found to be operative following detailed investigation by several groups. First, Kang et al., supra, indicated a potential polyadenylation signal (AATAAA
tandem repeat) 259 bp upstream of the 3'-end of the reported APP
full-length cDNA. The analysis of eight other full-length APP cDNA
clones obtained from a human fetal brain cDNA library (Unterbeck, 1986, Dissertation, University Cologne, FRG) demonstrated in a 1:1 ratio between shorter cDNAs (-3.2 kb) using the first polyadenylation signal versus the original cDNA forms (~3.4 kb) using the second polyadenylation signal. Interestingly, all eight clones encoded for 695 residues of APP. The alternative use of different polyadenylation signals in APP transcripts was confirmed by other laboratories (Goldgaber, 1988, in The Molecular Biologv of Alzheimer's Disease, Finch and Davis, eds., pp. 66-70, Cold Spring Harbor Laboratory, Cold Spring Harbor, New York; Jahnson et al., 1988, Exp. Neurol. lQ~: 264-268). A number of groups have screened several tumour cell-line derived cDNA libraries for the presence of APP transcripts and identified clones encoding new APP molecules containing an additional domain. This domain possesses striking homology to the Kunitz family of serine protease inhibitors (Tanzi et al., 1988, Nature 351: 528-530; Ponte et al., 1988, Nature 331:
525-527; Kitaguchi et al., 1988, Nature 331: 530-532). In particular these cDNA sequences contain an additional 167 bp insert at residue 35 289 of the APP-695 precursor (Figure 1) which encodes a 56 amino acid sequence of high sequence of homolo~y to aprotinin (Laskowski and ' 2 ~

Kato, 1980, Ann. Rev. Biochem. 49: 593-626), a well-characterized inhibitor of "trypsin-like" serine proteases. The peptide sequences flanking this region of insert are identical to the original APP-695 clone, resulting in an open readin8 fræme of 751 residues (APP-751). Kitaguchi et al., supra, isolated a third APP form withanother addition of a 19 amino acid domain at the C-terminal end of the 56 amino acid "aprotinin-like" region of APP-751, thus resulting in a larger protein of 770 residues (APP-770). Transient expression of APP-770 in COS-l cells conferred a marked inhibition of trypsin activity in cell lysates (Kitaguchi et al., supra). Both additional domains have been found to be encoded by discrete exons (Kitaguchi et al., su~ra) and all three transcripts (APP-695, APP-751, APP-770) are generated by differential splicing of a single gene on chromosome 21. These protease inhibitor domains have also recently been found to be present in mouse (Yamada et al., 1989, Biochem.
Biophys. Res. Co~un. 158: 906-912) and rat (Kang and Muller-Hill, 1989, Nucleic Acids Res. 17: 2130) species.
The relationship between the three different amyloid precursor forms and the formation of amyloid in AD is not known. In particular, it is not known whether a specific form of APP
contributes to A4 deposition. It is possible that either an imbalance in the relative expression levels of the three APP forms or their over-express~on might be involved in AD pathology. Initial i~ situ hybridization analyses using APP cDNA probes in human CNS
sections indicated that many neuronal cell types express these mRNAs (Bahmanyar et al., 1987, Science 237: 77-79; Goedert, 1987, EMBO J.
6: 3627-3632; Cohen et al., 1988, Proc. Natl. Acad. Sci. USA 85:
1227-1231; Higgins et al., 1988, Proc. Natl. Acad. Sci. USA 85: 1297-1301; Lewis et al., 1988, Proc. Natl. Acad. Sci. USA 85: 1691-1695;
30 Schmechel et al., 1988, Alzheimer Dis. Assoc. Disord. (US) 2: 96-111), but because of the nature of the probes used, these studies did not allow a differential analysis of the various APP transcripts.
Furthermore, there is little documented correlation between APP mRNA
levels, amyloid deposition and neuronal degeneration in AD. However, it appears that high levels of APP mRNAs alone do not form a sufficient prerequisite for cellular pathology in either the aging ~, . .. .. . .

, .

or AD brain (Higgins et al., supra). Spscific probes which discriminate between thP APP transcripts have been used for Northern analysis and the results suggest a developmental and tissue-specific pattern of expression of these mRNAs (Tanzi et al., 1988, ~E~;
Kitaguchi et al., 1988, supra; Neve et al., 1988, Neuron 1: 669-677).
Recently, S'-end cDNA probes from full-length APP cDNA clones (Kang et al., 1987, suDra), have been used to isolate genomic clones containing the 5'-end of the APP gene, also referred to as precursor of Alzheimer's Disease A4 amyloid protein (PAD) gene (Salbaum et al., 1988, ENBO J. 7: 2807-2813; La Fauci et al., 1989, Biochem.
Biophys. Res. Commun. 159: 297-304). Approximately 3.7 kb of sequences upstream of the strongest RNA start site have been analyzed by Salbaum et al., 1988, supra. By a combination of primer extension and Sl protection analyses, five putative transcription initiation sites have been determined within a 10 bp region. This ~3.7 kb region lacks a typical TATA box and displays a 72~ GC-rich content in a region (-1 to -400) that confers promoter activity to a reporter gene in an ~B vivo assay system (Salbaum et al., 1988, supra). The absence of a typical TATA and CMT box and the presence of multiple RNA start sites Is suggestive of its function as a housekeeping gene but does not imply constitutive gene expression (Salbaum et al., 1988, supra). The regulatory region contained within 400 bp upstream of the strongest RNA start site shows a variety of typical promoter-binding elements, including: two AP-l consensus sites (Lee et al., 1987, Nature 325: 368-372), a single heat shock recognition consensus element (Wu et al., 1987, Science 238: 1247-1253), and several copies of a 9 bp-long GC-rich consensus sequence where sequence-specific binding has been shown to occur by gel-retardation studies (Salbaum et al., 1988, suDra). In addition, the CpG:GpC ratio in this promoter region has been found to be 1:1 in contrast to a 1:5 ratio found in many eucaryotic DNAs (Razin and Riggs, 1980, Science 210:
604-610); CpG dinucleotides are known to control gene expression via DNA methylation (Doerfler, 1983, Annu. Rev. Biochem. 52: 93-124).
In addition, palindromic sequences capable of forming hairpin-like ~ ~3 ~ r~

structures are found around the RNA start sites ~La Fauci et al., 1989, supra).
Recently, several groups of investigators have determined the consensus binding sequence (AT rich decamer) for a number of 5 different homeobox proteins (Desplan et al., 1988, Cell 54: 1081-1090; Hoey and Levine, 1988, Nature ~: 858-861; Ko et al., 1988, Cell 55: 135-144; Odenwald et al., 1989, Genes Dev. 1: 482-496), which act most likely as transcription factors in specific regions during embryogenesis (for review, see Gehring, 1987, Science 236:
1245-1252; Holland and Hogan, 1988, Genes Dev. 2: 773-782). As yet, target genes, which might be developmentally regulated by the homeobox proteins have not been identified. Such genes, however, will have an important role during embryogenesis and potentially throughout the entire life span. The APP gene promoter contains at least five homeobox binding sites upstream of the RNA start sites.
Preliminary experiments have shown that the homeobox protein Hox-1.3 (Odenwald et al., 1987, Genes Dev. l: 482-496; Odenwald et al., 1989, Genes Dev. 3: 158-172) can bind at two of these sites. Thus, the APP gene, whose expression is developmentally regulated, appears to be a candidate gene for homeobox protein regulation. It is not known whether any of these putative recognition consensus elements modulate the expression of the APP gene promoter.
Despite all that is known about the APP gene, the primary defect leading to AD is not yet known, and specific mutations in the APP gene or other genes which cause AD in humans have not been defined. Uith the exception of aged primates (Price et al., 1989, BioEssays 10: 69-74), no laboratory animal model for AD exists. The introduction of genes into the germline of animals is an extremely powerful technique for the generation of disease models which will lead to a better understanding of disease mechanisms (Cuthbertson and Klintworth, 1988, Laboratory Investigation 58: 484-501; Jaenisch, 1988, Science 240: 1468-1474; Rosenfeld et al., 1988, Ann. Rev.
Neurosci 11: 353-372), including the mechanisms of AD. Cell culture and in vitro systems cannot duplicate the complex physiological interactions inherent in animal systems. Transgenic animals have been successfully generated from a number of species including mice, 2 ~ 7 ~

sheep, and pigs (Church, 1987, Trends in Biotech. 5: 13-19; Clark et al., 1987, Trends in Biotech. 5: 20-24). The gene or genes of interest are microinjected directly into the pronuclei of a one-cell embryo. A high percentage of reimplanted embryos develop S normally and, in a significant proportion of progeny, the transgene becomes integrated into the chromosonal DNA. Usually, multiple copies of the transgene i~ltegrate as a head-to-tail array. Although mosaic animals can be generated, germline transmission of the transgene usually occurs (Hogan et al., 1986, in Manipulating the Mouse Embryo: A Laboratorv Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, New York; DePamphilis et al., 1988, BioTechniques 6: 662-680). The generation of a transgenic mouse would be useful in defining the role of APP in the pathology of AD. For example, mice carrying APP transgenes which have been altered in either their protein-coding sequences or in their expression levels, might display dominant mutant phenotypes resembling those displayed in AD
pathology. The construction of recombinant genes and minigenes for expression in transgenic mice is a critical step in the development of transgenic mouse models. Particularly critical is the choice of an appropriate gene promoter for the minigene and other regulatory elements for the cell and tissue speciiic expression of the minigene.
A gene promoter must be utilized which will facilitate the expression of recombinant genes with a cell and tissue specificity consistent with the formation of amyloid plaque and perhaps with the expression pattern of the endogenous mouse APP gene.
To date, the identification of essential regulatory elements for many genes has not been straightforward and is, at best, unpredictable. A number of critical factors contribute to the complexity of this problem. Firstly, gene promoters that exext cell specific regulation in DNA transfection experiments do not necessarily confer cell and tissue specificity in transgenic animals.
For example, transfection experiments have revealed that important cell specific regulatory elements reside within 400 bp upstream of the cap site of the rat albumin gene (Ott et al., 1984, EMB0 J. 3:
2505-2510; Friedman et al., 1986, Mol. Cell Biol. 6: 3791-3797).
However, an additional enhancer, located 10 kb upstream from the -12- ~ 7~
albumin promoter, was found to be necessary to obtain liver-specific expression in transgenic mice (Pinkert et al., 1987, Genes Dev. 1:
268-276). While promoter sequences of the ~-fetoprotein gene confer cell specificity in cell culture (Godbout et al., 1986, Mol. Cell Biol. 6: 477-487; Muglia and Rothman-Denes, 1986, Proc. Natl. Acad.
Sci. USA 83: 7653-7657; Widen and Papaconstantinou, 1986, Proc. Natl.
Acad. Sci. USA 83: 8196-8200), additional enhancer elements located between -1 kb and -7 kb, were found to be necessary for liver specific expression in transgenic animals (Hammer et al., 1987, Science 235: 53-58~. Sscondly, the organization of various genes differs considerably and essential regulatory elements have been found in numerous positions. In some cases, the necessary regulatory elements are located within a compact region proximal to the cap site. For example, sequences residing within nucleotide -205 to nucleotide +8 of the rat elastase I gene are sufficient to confer an appropriate expression pattern in transgenic mice (MacDonald et al., 1987, Progress in Brain Research 71: 3-12). A tightly defined regulatory region has also been identified in the human ~-crystallin gene (Goring et al., 1987, Science 235: 456-458). The human ~-globin gene, however, has at least four separate regulatory elements:
a positive globin specific promoter element, a negative regulatory element, and two gene enhancers, one located within the second intron and the other located 3' of the structural gene (Behringer et al., 1987, Proc. Natl. Acad. Sci. USA 84: 7056-7060; Grosveld et al., 1987, Cell 51: 975-985). Thirdly, in many cases, the site of integration exerts a strong influence on the level and pattern of expression of transgenes. Regions of several genes have been identified which overcome, at least in part, these position effects.
DNase I hypersensitive sites located approximately 50 kb 5' to and 20 kb 3' of the ~-globin gene facilitate position-independent, high-level expression of a ~-globin minigene in transgenic mice (Grosveld et al., 1987, ~D~)- Furthermore, introns of the rat growth hormone and mouse metallothionein genes increase transcriptional efficiency of transgenes on average 10- to 100-fold (Brinster et al., 1988, Proc. Natl. Acad. Sci. USA 85: 836-840). Rat growth hormone lntronic sequences exerted a positive effect even on heterologous gene , `

-13- ~ 7 ~
constructions utilizing either the mstallothionein or elastase promoters. The effect of these introns is not related to an increased efficiency of RNA processing but is due to an actual increase in the rate of transcription (Brinster et al., 1988, supra).
It is also possible that introns and other genomic regions contain sequence elements which are recognized at particular stages of development or may contain elements which influence chromatin structure. In many cases, the inclusion of genomic elements which diminish position effect~ ~ay be essential for a transgene to maintain an expression level sufficient to generate a phenotype.
The identification of these elements may in some cases be a formidable task; for example, the APP gene locus encompasses at least 50 kb (Lemaire et al., 1989, Nucleic Acids Res. 17: 517-522). The identification of such elements would be extremely useful in the construction of recombinant APP minigenes. These minigenes can then be introduced into the germline of transgenic mice, thus providing animal models for AD.

. . ~ ., ~ . . .

-14- 2 ~
SUMNARY OF T~E INVENTION
The present invention provides recombinant minigenes for the expression of alternative forms of APP, including APP-695, APP-751, APP-770, and a variety of mutant forms of APP. The present invention also provides for the introduction of such functional APP minigene constructs into the germline of mice thereby generating transgenic animal models of AD useful $n the identification of the molecular mechanisms of AD pathogenesis. The recombinant minigenes according to the present invention contain essentially five different elements:
(1) gene promoter (DNA elements responsible for gene regulation), (2) and (3) APP protein coding region (cDNA or mutated cDNA), (4) mRNA
polyadenylation signals, (5) RNA splicing signals and genomic elements required for developmentally appropriate and cell~tissue-specific expression of the APP-encoding DNA. The identification of such genomic elements is highly unpredictable. The ].ocation of these sequence elements varies from gene to gene and may be found in the 5' regions, within introns, in 3' regions, or in other locations.
It has now been unexpectedly found that an -4.6 kb EcoRI human genomic fragment (or portions thereof), comprising ~2.8 kb 5' to the APP in RNA start site, the first exon of the APP gene and -1.6 kb of the first intron, is sufficient to direct cell and tissue-specific expression of a reporter gane in transgenic mice, and in a manner consistent with the expression pattern of the endogenous mouse APP
gene. This genomic fragment contains a promoter and perhaps other regulatory elements that facilitate the expression of recombinant APP minigenes with a cell and tissue specificity consistent with the formation of amyloid plaque and the expression patterns of the APP
gene. Since the primary defect leading to AD has not yet been determined, and specific mutations which cause AD in humans have not been identified, transgenic mice with recombinant APP minigenes according to the present invention provide animal models for the disease. For example, the generation of transgenic animal models for A4 amyloidosis is essential for defining the role of A4 in the pathogenesis of AD. Transgenic mice, according to the present invention, carrying APP genes altered in their protein-coding sequences or in their expression levels, provide models for ', ' ~

-15- 2 ~
exhibition of dominant mutant phenotypes resembling some aspects of AD pathology. As Alzheimer's pathology is restricted to specific regions of the brain, only those minigene constructs with the appropriate cell and tissue-specific genomic regulatory elements, such as those provided by the present invention, will enable for the development of transgenic mouse models of AD.

8RIFF DE~CRI~TION 0~ 5H~ ~RAWINGS
Figure 1 is the cDNA sequence of the amyloid precursor protein (APP) cloned in pFC4.
Figure 2 is a circular map of pFC4.
Figure 3 is an illustration of the 5'-end of the APP gene.
Figure 4a is an illustration of gene products of nonmutated forms of APP encoded in APP minigenes.
Figure 4b is an illustration of gene products of mutated forms of APP encoded in APP minigenes.
Figure 5 is an illustration of the construction intermediates and products: pMTI-2302, pMTI-2303, pMTI-2305 and pMTI-2304.
Figure 6a is an illustration of polylinkers in cloning vectors:
pWB16, pMTI-2110, pMTI-2300 and pMTI-2301.
Figure 6b is a circular map of pMTI-2301.
Figure 7a i9 an illustration of construction intermediates pNTI-2306, pMTI-2307, pMTI-2311 and pMTI-2312 and minigene pMTI-2314.
Figure 7b is a circular map of pMTI-2307.
Figure 7c is a circular map of pMTI-2312.
Figure 8a is an illustration of minigene constructs pMTI-2310, pMTI-2314, pMTI-2319, pMTI-21320, pMTI-2321, pMTI-2322, and pMTI-2325, encoding alternate forms of APP.
Figure 8b is a circular map of pMTI-2314.
Figure 9 is an illustration of construction intermediates pMTI-2307, pMTI-2316, and pMTI-2317 and minigene pMTI-2318.
Figure lOa i9 an illustration of construction intermediates pMTI-2312 and mouse metallothionein-I genomic sequences and minigenes pMTI-2323, pMTI-2331, pMTI-2332, pMTI-2324, and pMTI-2326.
Figure lOb is a circular map of pMTI-2323.

, . , 2 ~ 7 7 Figure lla is an illustration of construction intermediate pMTI-2323 and minigenes pMTI-2327 and pMTI-2337.
Figure llb is a circular map of pMTI-2337.
Figure 12 shows the DNA/amino acid sequence of 5p - spacer A4 and MC-100.
Figure 13 shows the DNA/amino acid sequence of sp-spacer A4 and SP-A4.
Figure 14 is an illustration of construction intermediate pMTI-2328, a pFC4 fragment, and minigenes pMTI-2329, pNTI-2333, pMTI-10 2334, pMTI-2335 and pMTI-2336.
Figure 15a is an illustration of APP 3'-end genomic clone pSVl and minigene pMTI-2339.
Figure 15b is a circular map of pMTI-2339.
Figure 16 is the DNA sequence of the 3'-end of the APP gene.
Figure 17 is a circular map of pNotSV2neo.
Figure 18a is an illustration of pNotSV2neo subclones for minigenes pMTI-2360, pMTI-2361, pMTI-2362, pMTI-2363, pMTI-2364, pMTI-2365, pMTI-2366, pMTI-2367, pMTI-2368, and pNTI-2369.
Figure 18b is a circular map of pMTI-2360.
Figure 19 is an illustration of pNotSV2neo and minigenes pMTI-2339, pMTI-2369, pMTI-2342, pMTI-2343 and pMTI-2344.
Figure 20 is an illustration of the APP-lacZ reporter gene pMTI-2402.
Figure 21(a-d) illustrates the cellular distribution of APP
mRNA in normal mouse detected by in situ hybridization with labeled single-stranded human APP DNA probe. (a) Section of mouse cerebral cortex. (b) Section of mouse cerebellar cortex. (c) Section of mouse trigeminal ganglia. (d) Section of mouse liver.
Figure 22 illustrates the histochemical staining pattern of E.
coli ~-galactosidase activity in brain section of a BE803 transgenic mouse.
Figures 23(a-d) illustrates the histochemical staining pattern of E. coli ~-galactosidase activity in serial brain sections of a BE803 transgenic mouse (a, b and c) and in a section of a normal mouse brain (d).

.
.~ ' .

.

' 2 ~

Figure 24(a-d) illustrates the histochemical staining pattern of E. coli ~-galactosidase activity in sections of cerebellar cortex (a and b), trigeminal ganglion (c), and liver (d) from transgenic BE803 mouse.
5Figures 25(a-d) illustrates the cellular and subcellular distribution of E. coli ~-galactosidase in transgenic BE803 mouse brain. Light microscopic image of histochemical staining pattern of E. coli ~-galactosidase activity in cerebral co 'a). Normarksi optic image of histochemical staining pattern of E. coli ~-galactosidase activity in cerebral cortex (b and c). Immunogold localization of E. coli ~-galactosidase in cerebral cortex section from a BE803 transgenic mouse.
Figure 26 shows an Sl analysis of human APP RNA expression in the brain of a series of transgenic mice.
15Figure 27 shows an Sl analysis of human APP RNA expression in the brain of a second series of transgenic mice.
Figure 28a is a Western-blot of APP-695, APP-751 and APP-770 protein expression in the brain of a normal mouse and transgenic mice carrying human APP minigenes using monoclonal antibody (mAb) 22C-II.
Figure 28b is a Western-blot of human APP-751 protein expression in the brain of a normal mouse and transgenic mice carrying human APP minigenes using mAb 56-1.
Figure 29 is a ~estern-blot (using mAb 22C-ll) of APP-695, 25APP-751 and APP-770 protein expression in COS cells transfected with human APP minigenes.
Figure 30a illustrates circular maps of pMTI-4 and pMTI-38.
Figure 30b illuqtrates circular maps of KS Bluescript, pMTI-41, pMTI-43 & 44 and pMTI-42.
30Figure 31 is an illustration of construction intermediates and products pMTI-52 to pMTI-53, pMTI-57 and pMTI-58.
Figure 32 illustrates reactivities of Kunitz monoclonal antibodies ~56-1, 56-2, 56-3) with APP proteins.
Figure 33 illustrates the primate specificity of monoclonal antibody 56-1.

.19~
Fl~sr~ 34 ~how~ the i~rDunbcytoch~mlcal ~Snin6 o~ a ~r-in tl~3ub 8~tlon ~ro~ transgenic ~ou~o AE301~207 ~ Airlg r~bblt polyclonAl An~ y pAb 90-29.
~ leuro 3S ~hows the ~munocytoch~m~c~ a~nln3 of ~ br~-n t~ u~ ~ctlon from tran6go"~4 30uso A~;301+207~5'1) u~in~ rabbSt polyolona~ ~nt~body pAb 90-29 (hl~hsr ~gnL~lcatlon o~ ~mil~ ld t~ r~ed in F~e~r~ 34) 6 ~bo~r- thc i~ mocy~cochemic~l ~ta~ng o~ a brc~n tlJ4t~ ~ect~on ~rom trl~nsg~c ~ou~o AE301+207~Fl) s~tng ~bblt po~yclonnl anelbod.y pAb 90-28 ~aE5nlf~catlon ~imllct to th~
d~crlbct ~n ~Ieure 35).
Figure 37 ~how~ thR lm~unocyto~h~mlcal staln~ng o~ ~ braln ti~uc ~actlon ~ro~ tranYg~nIc D10U5- F~8~31~l05~1) u~ g s~
pDlyclonal ~ntlbody pAb 93-29 ~m~gni~lcatlon ~mll~r to th~t 1~ descrlb~t ln ~gur~ 35~.
Flgurc 3~a 1~ ~n olRctron ~l~rogr~ph of ~ ~h~n oactio~ o~ ~r~ln t~eou- ~ro~ tho h~ppocampal regton of trsn~gflnl~ ~ou4a A~301~201(F2~.
F~tur- 3Bb i~ ~n clcc~ro~ ~lc~ogrJph of ~ ~h~n D~otlon o~ braln tl~su~ ~rom th~ hlppocampal regto~ o~ t~an~g~nlc mous- ~E301~201(F2) (t~ nt ~eld eb~n doscribed ~n Flgure 38a).
Figurc 39 ~ ~n alectron ml¢ro~r~ph of l~muno6old ~talnlnt o~
ultr~thln cryosectlon~ of braln tIuu~ ~roD~ h~o¢amp~Ll r~g1on of ~ran~g~ntc ~OUBO ~E3al+201(F2~ w lng rabblt poly41Ona1 ant~body ~pA~) 90-29.
F~yure 40 ~ ~ c~rcular ~sp of p~TI-70.
~Ur~ B a c~rcul~r ~ap of p~T-2371.
FiB~Y~ 42 I~ ~ ~e~tonn-blot, u~n~ pAb SG369, of MC-laO p~4~n oxpro~u~on ~n ~S-70 tr~n~focto~ oell lin~ ch~70-31, cMT170~a2, ant cMTI70-~3, nd cQll llne~ cMT152-A~, cHT~66-B6, c~SS66-CS, 30 c~TI69~C6, ~MTI69-A4, ~nt oM~69-A5 wh~ch ~r~ Includ~d c~ n-~Ativc ¢on~role .

. .

2 ~
Figure 43 is a Western-blot, using pAb SG369, of MC-100 protein expression induced with cadmium in transfected cell lines cMTI70-A2, cMTI70-A3, cMTI70-A6, cMTI70-Bl, cMTI70-B2, and cMTI70-B3 (transfected with pMTI-70). Cell lines cMTI63-Bl, cMTI63-C2, and cMTI53-Al are included as negative controls.
Figure 44a shows the immunofluorescence, using pAb SG369, of MC-100 protein expression in transfected cell line cMTI70-A6.
Figure 44b shows the immunofluorescence, using pAb SG369, of MC-100 protein expression in transfected cell line cMTI70-A6 (higher magnification, different field than described in Figure 44a).
Figure 44c shows the immunofluorescence, using pAb SG369, of control transfected cell line pMTI53-Al (same magnification as described in Figure 44a).
Figure 45 is a Western-blot of immunoprecipitated (using pAb SG369) MC-100 protein from cadmium-induced, pMTI-70 transfected cell line cMTI70-A6.
Figure 46 demonstrates human APP RNA expression, using riboprobe analysis, in the brain of a series of transgenic mice.

' :

:

-20- ~ 3~ a~ ~
DESCRIPTION OF THE PREFERRED EMBODIMENT
AD (Alzheimer, 1907, su~ra) is characterized by a widespread functional disturbance of the human brain. Fibrillar amyloid proteins are deposited inside neurons as neurofibrillary tangles (Katzman, 1983, supra) and extracellularly both as amyloid plaque cores (Katzman, 1983, su~ra) and as cerebrovascular amyloid, (Katzman, 1983, supra). The major protein subunit (A4) of the amyloid fibril of plaques, blood vessel deposits, and potentially of tangles is an insoluble, highly aggregating 4~-42 residue peptide of relative molecular mass 4,500 (Masters et al., 1985, suDra and 1985, su~ra; and Glenner and Wong, 1984, supra). The A4 peptide which derives from a larger amyloid precursor protein is encoded by a gene on chromosome 21 (Kang et al., 1987, su~ra; Goldgaber et al., 1987, su~ra; Tanzi et al., 1987, su~ra). APP mRNAs are detected in neurons and in other tissues both within and outside the brain (Goedert, 1987, su~ra; Cohen et al., 1988, supra; Higgins et al., 1988, ~a~3).
Age, genetic elements, and, potentially, environmental factors appear to contribute to cellular pathology in AD, but mechanisms that lead to these brain lesions are not yet understood. A
fundamental question in the pathogenesis of AD is the relationship between the observed neuronal abnormalities and the deposition of amyloid.
Because the primary defect leading to AD is not yet known, and specific mutations which cause AD in humans have not been defined, animal models for the study of AD wou$d be especially useful. With the exception of aged primates, no laboratory animal model for AD
exists. Due to these limitations, the generation of transgenic mouse models for AD may bs the best approach in deiining the role APP plays in the etiology of AD. Transgenic mice carrying APP genes which have been altered either in their protein-coding sequences or in their expression levels may lead to dominant mutant phenotypes resembling those displayed by the AD pathology. The introduction of functional minigene constructs described herein into the germline of mice has been used to generate models of AD and to identify the molecular mechanisms of pathogenesis.

/ - --21- 2~ 77 A critical step for the development of a transgenic mouse model for AD was the design of a minigene that allows high-level expression of a foreign gene in a predictable tissue-specific fashion.
Recombinant minigenes according to the present invention contain essentially five different elements: gene promoter (DNA elements responsible for gene regulation), protein coding region (cDNA), mRNA
polyadenylation signals, RNA splicing signals, and genomic elements required for correct developmental expression of DNA that has participated in a developmental program (the location of these sequence elements can vary from one gene to another and can be found within introns, 3' regions, and in other locations).
The following paragraphs and examples describe essential steps leading to the design and construction of such minigenes for the generation of animal models for AD. A gene promoter has been isolated and characterized which in transgenic animals confers an expression pattern of foreign genes that is comparable with the pattern of expression of the endogenous mouse APP gene. A series of minigenes comprising the APP gene promoter and a variety of different APP gene products including mutant forms have been generated.
Transgenic animals, expressing these minigenes, are useful in the investigation of the i~ vivo function of various APP gene products, the regulation and expression patterns of the APP gene, and the relationships of these processes to the formation of amyloid. The use of various RNA splicing and polyadenylation signals in the minigenes allows for the optimization of post-transcription processing and stability of human APP transcripts in transgenic animals.
Appropriate recombinan~ minigenes were generated and tested.
The minigenes were microin;ected directly into the male pronucleus of mouse l-cell embryos. The manipulated embryos were subsequently transferred to the oviducts of pseudopregnant females. Litters from recipients were screened for the presence of the transferred minigene (transgene) in their genome by polymerase chain reaction (PCR) analysis and Southern-blot analysis of DNA derived from tail biopsies.

2G4~Q~7 Because Alzheimer's pathology is restricted to specific regions of the brain (Price, 1986, su~ra), the choice of an appropriate gene promoter for minigene constructions was critical for the development of the transgenic mouse model. A regulatory element comprising a gene promoter and perhaps other regulatory sequences must be utilized which will facilitate the expression of recombinant genes with a cell and tissue specificity consistent with the formation of amyloid plaque and the expression patterns of the APP gene. The present invention provides such a regulatory element.
An ~4.5 kb genomic fragment described herein encompassing the 5'-end of the human APP gene (Figure 3) had sufficient sequence information to direct cell and tissue-specific expression of thæ
protein product of a reporter gene, E. coli lacZ, in transgenic mice (Figures 20 to 25). The expression pattern of the reporter gene product ~-galactosidase in the central nervous system (CNS) was strikingly consistent with the expression pattern of the endogenous mouse APP gene and is consistent with the pattern of senile plaque deposition characteristic of AD pat$en~s. In situ hybridizations, using human APP cDNA as probe, revealed APP mRNA expression in specific brain regions, including: hippocampus, dentate gyrus, cerebral cortex, cerebellar cortex, pons, and spinal cord. ~-galactosidase staining, in transgenic brain tissue, was restricted to areas containing neuronal perikarya. In most cases, the ~-galactosidase staining in the CNS of BE803 transgenic mice was consistent with i~ situ hybridization patterns of mouse APP mRNA.
One exception was the CA3 region of the hippocampus where the ~-galactosidase staining was not as intense as would be expected from the observed levels of mouse APP mRNA. This difference may have been due to a lowered expression level of the reporter gene in this region or due to altered stability of the ~-galactosidase fusion protein. The ma~ority of ~-galactosidase fusion protein was looalized in secondary lysosomes within neuronal perikarya, therefore, E. coli ~-galactosidase fusion protein may be relatively unstable in neurons.
A variety of cDNAs encoding various forms of APP and mutants of APP were constructed. Three alternate forms of APP exist, : ' ` ' .

, .

designated APP-695, APP-770 and APP-751, all of which are encoded by a single common gene on human chromosome 21. The mRNA of the APP
gene is differentially spliced to yield thres gene products of 695 amino acids (aa), 751 aa, and 770 aa in length. The 751 aa and 770 aa forms contain an additional domain which has striking homology to a Kunitz type serine protease inhibitor (Kitaguchi et al., 1988, Nature ~ 530-532; Ponte et al., 1988, Nature 331: 525-527; Tanzi et al., 1988, Nature ~1: 528-530). Expression of one or more forms of the human APP gene products in transgenic mice provides a model with which to test the hypothesis that over-expression or anomalous expression of one or more forms of the gene results in Alzheimer's pathology. This hypothesis is not inconsistent with the observation that Alzheimer's pathology (i.e., A4 plaques in brain tissue) has been found in individuals with DS past the age of 30-40 (Glenner and Wong, 1984, Biochem. Biophys. Res. Commun. 122: 1131-1135). Such individuals are trisomic for chromosome 21 which contains the APP
gene, and the levels of APP mRNA in these individuals appears to be elevated (Tanzi et al., 1988, Science 235: 880-885).
Human APP cDNA clones corresponding to the three alternative APP forms were isolated. Plasmid pFC4 contains the full length cDNA
for tbe 695 aa form of APP (Kang et al., supra). Using pFC4 as a probe, a human neuroblastoma cDNA library was screened for the presence of additional transcripts corresponding to additional forms of human APP. An -1.8 kb cDNA was identified which contained both additional exons found in APP-770 (167 bp plus 58 bp), and represented a partial cDNA of the mRNA. Unique restriction sites (AccI and BglII) were used to subclone this 1775 bp fragment into the original pFC4 full-length clone, thus generating a full-length cDNA clone (pFC4-770) for the 770 aa form of APP. The 751 residue APP encoding cDNA clone (pFC4-751) was engineered by }~ vitro mutagenesis of pFC4-770. The deletion of 58 bp (M13-looping-out) was confirmed by DNA sequence analysis.
Using the various APP cDNAs, minigenes expressing each APP
form were constructed: pMTI-2314 for APP-695 expression, pMTI-2319 for APP-770 expression and pMTI-2320 for APP-751 expression. As an initial step in the construction of an APP-695 minigene, the EcoRI

.:

.

2 ~ 7 1 promoter fragment of APP was inserted into the HindIII site of pMTI-2301 by blunt-end ligatlon to produce pMTI-2307. The construction of the APP-695 minigene was completed in a stepwise fashion. pMTI-2311 was generated by ligating the ~3~HI fragment from pFC4 into the ~HI site of pMTI-2307. Next, the ~h~I fragment from pFC4 was inserted into the ~h~I site of pMTI-2311 to generate pMTI-2312.
Finally, a ~hI fragment of pMTI-2304 containing the SV40 RNA
splicing and polyadenylation signals was inserted at the ~hI site of pMTI-2312 to yield pMTI-2314. The APP-751 and APP-770 minigenes were constructed by subcloning the AccI-~glII fragments from pFC4-751 and pFC4-770 into the AccI/~glII sites of pMTI-2314 to generate pMTI-2320 and pMTI-2319, respectively.
A second minigene series for expression of APP-695, APP-770 and APP-751 can be constructed using a truncated APP promoter. For minigene, pMTI-2310 (APP-695), the -2.6 kb HindIII fragment of the 5'-end of the APP gene [Figure 3] was inserted into the HindIII site of the pMTI-2301 to generate pMTI-2306 (Figure 7a). The minigene expressing the 695 form of APP, pMTI-2310, was constructed in the same manner as pMTI-2314 described above (Figure 7a). The corresponding 751 and 770 minigenes can be generated as described above using the -1.8 kb AccI-~glII restriction fragment.
The accumulation of A4 pep~ide in amyloid plaques may be the result of anomalous proteolytic degradation of one or more APP forms (695, 751, 770). Minigenes have been constructed which can directly express either the A4 peptide or other fragments of APP that may exist as proteolytic intermediates during n vivo generation of A4.
Such APP fragments may, if they contain the A4 region, self-aggregate, and be further processed by the cell to alternately generate A4. The types of minigene which were constructed and which express such mutants are summarized in Figure 4b. Gene product IV
is devoid of a portion of the transmembrane domain and the entire cytoplasmic domain, leaving the A4 domain intact. This mutant gene product is expected to be secreted from the cell and perhaps further degraded to produce the A4 peptide. The secreted protein may also have other biological effects because at least some portion of APP
has been shown to be shed from cell surfaces. Gene product V

.

7 ~

(designated as MC-100) is translated into the membrane and, therefore, a pre~ursor protein was constructed which contains the 17-residue signal peptide of APP at the N-terminus. If the signal peptide is omitted, the C-100 protein (gene product VI) would be translated into the cytoplasm and perhaps have significantly different properties than if inserted into a membrane. Gene product VIII in which the signal peptide is also omitted should produce intracellular A4 directly, and which will not be inserted into a membrane. Another construct also expressing the A4 peptide including the APP signal peptide (gene product VII) was prepared. After the signal peptide cleavage point, gene product VII includes 40 amino acids encompassing the A4 peptide as well as 12 additional amino acids N-terminal of the A4 peptide region. This protein is expected to translocate through the cellular membrane and aggregate following proteolytic cleavage by the cell to generate A4.
To construct mutant APP minigenes for expression of truncated APP product, gene product IV and VII mutants, C-terminal frameshift mutations were generated. Frameshift mutations (-1, +2) of the cDNA
sequences immediately following the A4 coding region brought a translation 9top codon into the reading frame following the A4 peptide coding region. The resulting sequence encodes a truncated APP species (Figure 4b, gene product IV). A frameshift mutation (deleting nucleotide C) at the nucleotide position 2045 generated a stop codon after 40 amino acids from the N^terminus of the A4 sequence (amino acids 38, 39 and 40 are different than the native A4 sequence), and a +2 mutation (TG) after nucleotide position 2050 generated a stop codon after 41 amino acids from N-terminus of the A4 sequence (the last amino acid is different than the native A4 sequence). The +2 mutation was utilized in construct pMTI-2321 (Figure 8a). The generation of these frameshift mutations is described in co-assigned patent application U.S. Serial No. 194,053.
A third frameshift mutation, "mutant 40-1, n deleted an adenosine nucleotide at nucleotide 2055 (APP-695 cDNA sequence; Figure 15) and brought a translation stop codon into the reading frame directly following the 40th codon of the A4 peptide coding region (used in plasmids pMTI-2322, pMTI-2326, pMTI-2341, pMTI-2343, pMTI-2361).

rl r~

The frameshift mutations were inserted into pMTI-2314 (APP-695), pMTI-2319 (APP-770), or pMTI-2320 (APP-751) by swapping sequence domains between the unique ~glII and ClaI restriction sites (Figure 8a). The deletion mutation was also generated by site-directed mutagenesis which placed the stop codon directly past the A4 sequellces (pMTI-26). The mutation in pMTI-26 was inserted into the minigenes in a similar manner as described above.
To construct mutant APP minigenes for expression of gene product VIII mu~ant, the following steps were taken. Minigene pMTI-2318 (gene product VIII; Figure 4b) was generated in stepwise fashion (Figure 9). A ~glII-BamHI fragment from pMTI-26 containing the 42 aa A4 peptide sequence was inserted into the BsmHI site of pMTI-2307. Next, the BamHI to BamHI fragment of pFC4 was inserted into the BamHI site of pMTI-2316. Finally, the SDhI fragment containing the SV40 RNA processing signal was inserted into the ~hI site of pMTI-2317.
Because of substantial sequence homology between mouse and human APP gene products, it has been difficult to generate adequate antibodies that will allow unequivocal identification of APP using immunohistochemical analysis of tissue sections. To circumvent this problem, a highly antigenic epitope of Chlamvdia wss inserted into the APP-695 sequences at either the site of the Kunitz inhibitor domain insertion or the extreme C-terminus of the protein. The sequences were transferred into the minigenes using either the AccI
and ~glII restriction sites to generate pMTI-2325 (Figure 8a) or pMTI-2324 (Figure lOa).
In another series of minigene constructs, alternate RNA
processing signals were used. Because minigenes utilizing RNA
processing signals derived from the human APP gene or from an exogenous mouse gene might be expressed more efficiently in transgenic mice than those derived from SV40, constructs were generated which utilize RNA splicing and polyadenylation sequences of the mouse metallothionein gene. Alternatively, a genomic fragment from the 3'-end of the human APP gene which encompassed the APP
polyadenylation ~ignals was utilized. Minigenes expressing all of ~.

-27- 2~4~
the gene products described above and additional forms were generated using the alternate RNA processing signals as follows.
Using the metallothionein gene body (Figure lOa) as a source of RNA processing signals, minigenes expressing the three alternate forms of APP (695, 770, 751) and mutant APP forms described above were constructed. To generate minigene pMTI-2323 for expression APP-695, the -2.2 kb BelII to EcoRI fragment from the EcoRI genomic clone of the mouse metallothionein-I gene, pJYMMT(L), was inserted into the ClaI ~ite of pHTI-2312 by blunt-end ligation to generate pMTI-2323. Minigenes expressing alternates AP~ forms, APP-770 and APP-751, were generated by switching sequence domains (AccI to ~glII
fragment) fron minigenes (expressing APP-751 or APP-770) utilizing the SV40 polyadenylation sequence (Figure 8a) to pMTI-2323 (Figure 10) to generated pMTI-2331 and pMTI-2332. To generate a minigene pMTI-2326 expressing an APP C-terminal frameshift mutant, the BglII
to ClaI fragment from pMTI-2322 (Figure 8a), containing mutation 40-1, was switched with sequences between ~glII to ~l~I site of pMTI-2312 (Figure 7a) to generated pMTI-2326a. Then, the -2.2 kb metallothionein fragment (~glII to EcoRI fragment, see Figure lOa) was inserted into the ClaI site of pMTI-2326a to generate pMTI-2326.
Alternatively, the ~glII to SpeI fragment of pMTI-2322 was swapped directly into pMTI-2323 to produce pMTI-2326 (Figure lOa).
To generate minigene pMTI-2327 coding for C-100 (gene product VI, Figure 4b), the sequences between the NruI and ~glII restriction sites of pMTI-2323 were deleted (Figure lla). A translation initiation codon, ATG, directly precedes the first codon of the A4 sequences. To generate minigene pMTI-2337 coding for MC-100, the sequences between the KpnI and ~glII sites of pMTI-2323 (Figure lla) were deleted and the clone was ligated using synthetic oligonucleotide adaptor, sp-spacer-A4 (Figure 12). MC-100 (gene product V, Figure 4b) required the 17 residue signal peptide of APP
to direct insertion of the translated mutant protein into the membrane. The signal peptide should be cleaved and eliminated during the process. The nucleotide and a~ino acid sequence of MC-100 is presented in Figure 12.

... . .

, ~

To generate minigene pMTI-2341, a process analogous to that used to generate pMTI-2337 (Figure lla) was used. This involved deleting the sequences between the ~2~I and ~glII sites of pMTI-2326 (Figure lOa) and ligating the clone uslng synthetic oligonucleotide adaptor sp-spacer-A4 (Figure 13). Minigene pMTI-2341 (gene product VII or sp-A4; Figure 4b) thereby generated should express an A4 peptide linked to the APP signal peptide. The nucleotide and amino acid sequence of sp-A4 is shown in Figure 13.
For the alternate series of the minigenes incorporating the human APP gene-derived RNA processing signals, the 3'-end of the APP
gene was isolated in clone pVS-l. Clone pVS-l contained an -1.5 kb EcoRI fragment of human genomic DNA which encompasses the 3'-end of the terminal exon of human APP and the APP polyadenylation signal inserted into the EcoRI site of pUCl9, Figure 15a. The -1.5 kb ~PP
genomic fragment was isolated from a charon 21A lambda library of human chromosome 21 DNA (A.T.C.C. No. LA21NS01). The SmaI-SphI
(nucleotides 3102 to 3269) fragment from the APP cDNA clone, pFC4, was labeled as probe and used to screen the lambda library for the APP 3'-end genomic. The nucleotide sequence of the -1.5 kb APP
genomic fragment is shown in Figure 16.
The minigene construct, pMTI-2339, designed to express APP-695 was generated by inserting the -1.3 kb SphI fragment from pVS-1 into the ~hI site of pNTI-2312 (Figure 7a) yielding pMTI-2339 (Figure 15a). Minigenes expressing APP-751 and APP-770 and the mutants indicated below were generated by switching sequence domains of pNotSV2neo subclones of the APP constructs (Figure 18a). The subclones were utilized for switching sequence domains because of the presence of convenient restriction enzyme sites. NotI fragments of many APP minigenes were subcloned into pNotSV2neo (see Figure 18a) so that APP expression could be determined in transient transfections of COS cells. The construct encoding APP-770 (pMTI-2342, Figure 19) was generated by swapping the PvuI to SpeI fragment from pMTI-2363 (Figure 18a) to pMTI-2369 (Figure 19). A construct which encodes APP-751, pNTI-2345, was generated in an analogous fashion.

' ; "' ' .

7 ~

To generate minigene pMTI-2343 expressing mutant 40-1, the sequence domain encompassing the framashift mutant, 40-1, was inserted into pMTI-2369 by swapying sequences between the PvuI and Spel restriction sites from pMTI-2361 (Figure 18a) to p.~TI-2369 (Figure 19).
To generate minigene pMTI-2340 encoding mutant, MC-100, the sequences between the KpnI and BglII sites of pMTI-2339 (Figure 15a) were deleted and the clone ligated using synthetic oligonucleotide linkers (sp-spacer-A4, Figure 12).
To generate minigene pMTI-2344 encoding mutant sp-A4, the sp-A4 mutant was inserted into pMTI-2369 by swapping sequences between the PvuI and SpeI restriction sites from pMTI-2365 (Figure 18a) to pMTI-2369 (Figure 19).
A llumber of the APP minigenes prepared as described in the Examples which follow were used to prepare transgenic mice expressing an APP transgene. Such transgenic mice are useful as models for the study of AD.

E~AMPLE 1 Elements of APP ~in~genes:
Gene Promoter and Regulator Elements for Cell- ant Tissue-Specific Expression A critical step in the development of a transgenic mouse model for the pathology of AD is the isolation of an appropriate gene promoter for minigenes to be used as transgenes. A gene promoter and perhaps other regulatory elements must be identified that facilitate expression of recombinant APP minigenes with a cell and tissue specificity consistent with the formation of amyloid plaque.
As a first step, fragments containing various portions of the 5'-end of the human APP gene from human genomic libraries were isolated.
The 5'-end of the APP gene has been shown to contain DNA sequence elements which function as gene promoters in cell culture (Salbaum et al., supra). The starting material for the isolation of an -2.5 kb HindIII fragment was a hu~an genomic library available from the A.T.C.C. under accession number LL21NS02. This library was prepared by uslng a fluore9cence-activated cell sorter to obtain a fraction enriched for human chromosome 21. This frAction was digested with .

HindIII and cloned into the lambda veetor Charon 21A. This HindIII
human chromosome 21 library was plated on 6 plates at an approximate density of 5 X 104 pfu/plate. Screening of duplicate filters of the library representative of 3 X 105 total pfu was done by conventional S methods (Benton and Davis, 1977, Scienee ~ 180) using an -1.0 kb cDNA probe derived from plasmid pFC4. Plasmid pFC4 (Figure 2) is described ln Example 3. It contains an -3.3 kb cDNA insert having the sequence shown in Figure 1. The -1.0 kb probe was obtained from the AoaI site at nucleotide position 52 to the XhoI site at nucleotide position 1056. A 91 bp probe was also used to confirm the initial screen with the -1.0 kb probe. This small confirming probe was derived from pFC4 as an ApaI (nucleotide 52) to NruI (nucleotide 144) fragment. Clones which hybridized were plaque-purified through three subsequent cycles of screening and purification until a 100 positive hybridization response was obtained. One such plaque-purified clone was designated ~MTI 3509 (~12A). ~MTI 3509 contained a genomic insert of -2495 bp. This HindIII insert was subcloned into the HindIII site of plasmid pUC18 (Yanisch-Peron et al., 1985, Gene 33: 103) and designated pMTI-3501 (pUC18/pAL12A-12). Plasmid pMTI-3501 was found to contain -487 bp of sequence 5' to the first nueleotide of the cDNA insert of plasmid pFC4 (shown in Figure 1).
Using an -181 bp genomic probe derived from pMTI-3501 (from the ApaI site at -128 to ~eaI at 52), an EcoRI human genomic chromosome 21 cell sorter-enriched library available from the A.T.C.C. under accession number LA21NSOl was screened in a manner similar to that described above for the ~iadIII library. One plaque-purified clone, designated ~MTI 3522 (~pE-l) contained an insert of -4638 bp. Thi~ -4.6 kb insert was subeloned into the EeoRI site of plasmid pUCl9 (Yaniseh-Peron et al., supra~ and designated pMTI-3515 (pUCl9/pE-12). Plasmid pMTI-3515 was found to cantain -2831 bp 5' to the first nueleotide of the eDNA insert of plasmid pFC4 (Figure 1) .
The genomie inserts of both pMTI-3501 and pMTI-3515 eontaining sequenees 5' to the cDNA sequence of pFC4 interrupt the ~I site of the eDNA insert of the pFC4 at position 207. This ~p~I site was not present in the genomie DNA at the junetion of the boundary between ' . , ~

r~

exon 1 and intron 1, but was created at the splice site of the mRNA.
Plasmid pMTI-3501 and plasmid pMTI-3515 encode -1.6 kb and -1.4 kb of intron 1, respectively. Plasmid pMTI-3515 was shown to contain -2.8 kb of sequences 5' to the APP start site of transcription, along with exon 1 (encoding amino acids 1-19 of APP) and part of the first intron (-1.6 kb) as shown in Figure 3.

Elements of APP Minlgenes:
SV40-derived Polyadenylation and RNA Splicing Signals 10SV40 virion DNA (Hay and DePamphilis, 1982, Cell 28: 767-779;
also commercially available from International Biotechnologies, Inc.
(IBI) as catalog no. 33200) was digested with BamHI and BclI. The small -0.2 kb BamHI-BclI fragment (2533 bp to 2770 bp) containing two RNA polyadenylation signals (one in each strand) (see, 15DePamphilis and Bradley, 1986, in The Papovaviridae, Volume 1, Salzman (ed.), Plenum Publishing Corp., NY) was "shotgun" cloned into plasmid pUC19 as follows. The ~mHI- and ~lI-digested SV40 DNA was mixed with ~3~HI-digested pUCl9 DNA. The restriction enzymes were removed via phenol-chloroform extractions and the DNA was ligated overnight at 12C using T4 ligase (commercially available from New England Biolabs (NEB) as catalog no. 202). Impurities, including any residual phenol or chloroform were removed from the ligated DNA by GENECLEAN (available from BI0 101 Inc., PØ Box 2284, La Jolla, CA. 92038). This DNA was used to transform competent DH5~
E. coli cells (commercially available from Bethesda Research Laboratories (BRL), Gaithersburg, MD 20877). The transformants were screened by miniprep analysis using BamHI digestion and HDaI/HindIII
digestions to determine the orientation of the inserted DNA. The desired -2.9 kb plasmid was designated pMTI-2302 (Figure 5).
30SV40-derived RNA splicing signals from plasmid pFC4 were inserted into pMTI-2302 as follows. First, the ~I restriction endonuclease site at nucleotide position 144 of the APP cDNA sequence (shown in Figure l) was converted to a BglII restriction endonuclease site using blunt-end linkers to yield plasmid pMTI-2303 (Figure 5).
3S For this conversion, pFC4 was digested with ~_I and the linear -6.4 kb DNA fragment was purified with GENECLEAN and then ligated with 0.5 . .,~, .

, -OD260 units of ~glII linkers (~EB catalog no. 1036) using T4 ligase [incubation for 5 hours at 16C]. Linkers were removed by gel-filtration using "Quick Spin Columns" (Boehringer Mannheim, catalog no. 100408). The linear DNA was recovered using GENECLEAN and was circularized using T4 ligase to generate pMTI-2303 [diagnostic minipreps with ~glII and XhoI digestion revealed fragments of -0.35 kb, -1.4 kb, and -2.95 kb]. This procedure deleted APP encoding sequences from nucleotide position 145 to the BglII site downstream at nucleotide position 1915. An -0.3S kb fragment containing the SV40 splicing signals could be excised from plasmid pMTI-2303 DNA by digestion with XhoI and BglII. This XhoI-BglII fragment was gel-purified on a 5~ polyacrylamide gel, eluted from the gel, then used for ligation with SalI-~HI digested pNTI-2303 DNA to generate plasmid p~TI-2305 (Figure 5). In the next step, an SphI site was inserted into the EcoRI site of pMTI-2305 to generate plasmid pMTI-2304 (Figure 5). Plasmid pMTI-2305 was digested with EcoRI to yield an -3.2 kb fragment and then dephosphorylated using CIAP (calf intestine alkaline phosphatase, reaction conditions suggested by manufacturer, Boehringer Mannheim, catalog no. 1097075). The DNA was extracted with phenol/chloroform/isoamylalcohol (24/24/1) and precipitated in 2.5 M ammonium acetate and 70~ ethanol. The DNA
fragment was ligated, using T4 ligase, to an EcoRI-~I adaptor. The adaptor is a self-annealing oligonucleotide (Sequence 5'-AATTCCCGCATGCGGG-3'; synthesized using an Applied Biosystems instrument and manufacturer's recommended methods, model no. 380A DNA
Synthesizer) and was annealed by heating in solution (1 mM EDTA, 10 mM Tris pH 7.6) to 65C and allowing sample to slowly cool to room temperature. Diagnostic minipreps of pMTI-2304 with ~hI revealed fragments of -0.6 and -2.6.
An -0.6 kb S~hI cassette containing SV40-derived splicing signals and polyadenylation signals could be excised from pMTI-2304 DNA by digestion with SphI. This cassette was useful in the construction of APP minigenes described in the Examples below.

:

~, 2 ~

EXA~PLE 3 Elements of APP ~inigenas:
APP-695, -770, -751 Protein Coding Regions Plasmid pFC4 is a cDNA clone similar to clone 9-110 described by Kang et al., 1987, Nature 325: 733-736 and contains a full-length cDNA encoding APP-695 (Figure 2). The cDNA library described by Kang et al., suDra, was constructed by the method of Okayama and Berg, 1983, Mol. Cell. Biol. 3: 280-289, using polyAt RNA isolated from brain cortex of a 5-month-old aborted human fetus. This cDNA library (~105 E. coli HB101 transformants) was originally screened using a mixture of 64 20-mers as probes. The 20-mers had the sequence S'-T T 2 T G A T G A T G C A C T T C A T A-3' C G G C G G
This sequence was deduced from the amino acid sequences of residues 10-16 of the A4 peptide. Nine positive clones were obtained, and one (clone 9-110) was selected for sequence analysis. The complete sequence of the clone 9-110 insert encoding a full-length APP-695 sequence is shown in Figure 1 of Kang et al., su~ra. This cDNA
library was replated and screened with two different synthetic 20oligonucleotide probe mixtures of 17 and 20 nucleotides. The 17 mers had the sequence:
5'-A C G T C T T C N G C G A A G A A-3' A C A A
where N is A, G, C or T. The 20-mers had the sequence:
255'-T T T T G G T G G T G N A C T T C G T A-3' C A A C A
where N is A, G, C or T. Two positive clones, designated pFC4 and pFC7, were obtained which contained identical inserts as determined by restriction endonuclease mapping and partial DNA sequence analysis. Clone pFC4 was selected for further analysis and contained an -3.4 kb insert encoding the full-length APP-695 sequence shown in Figure 1. The nucleotide sequence is identical to the sequence of 9-110 shown as Figure 1 Kang et al., supra.
A human neuroblastoma cDNA library was purchased from Clontech, Palo Alto, CA, catalog no. HL1007a, and screened for the presence of APP transcripts of the 751 aa and 770 aa forms of human APP. This library was probed with an -1.4 kb BamHI fragment (nucleotide 99-1475) of pFC4. Two positive clones (~El-bl-l and ~El-bl-3) with ,..,, ..~. ,.

.

. . .: . .
.

~. ,,, . . ~ ~

7~

identical inserts were obtained. These two clones each contained an -2.0 kb cDNA insert comprising both of the additional exons (167 bp and 58 bp) found in APP-770 but represented only a partial cDNA of the full-length mRNA for APP-770. The -2.0 kb insert was subcloned into the EcoRI site of plasmid pUCl9 to generate plasmid pMTI-3525.
A full-length cDNA for APP-770 was obtained by replacing the -1.7 kb I-~glII fragment of pFC4 (nucleotide 207 - nucleotide 1915 of pFC4 sequence shown in Figure 1) with the -.96 kb ~eaI-~glII fragment of pMTI-3525, generating plasmid pMTI-3521 (pFC4-770). Specifically, pMTI-3525 was digested with KpnI and ~glII, and the -.96 kb KpnI-~glII fragment was gel-purified. Plasmid pFC4 was similarly digested with ~BI and B~lII, and the -4.7 kb K nI-~glII fragment was gel-purified. The two gel-purified fragments were mixed, ligated and used to transform E. coli DH5 cells. The resulting -6.6 kb plasmid pMTI-3521 was the source of the APP gene fragment for the construction of minigenes for the expression of the APP-770.
In order to obtain a full-length cDNA encoding the 751 aa form of APP, in vitro mutagenesis of plasmid pMTI-3521 was performed to delete the 58 bp sequence encoding the C-terminal 19 amino acids of the 75 aa protease inhibitor domain of APP-770. This was achieved by the M13-looping out procedure as described by Geisselsoder et al., 1987, Biotechniques 5: 786-791. DNA sequence analysis confirmed the successful deletion of the 58 bp seg~ent of pMTI-3521 to generate plasmid pMTI-3524. Plasmid pMTI-3524 was the source of the APP gene fragment for the construction of minigenes for the expression of APP-751.
Plasmid pMTI-3524 was prepared according to the following series of steps. First, plasmid pMTI-3522 was constructed as follows. Plasmid pMTI-4 was partially digested with AccI, then filled-in with the Klenow fragment of DNA polymerase (Klenow) to remove one of the two AccI sites, ligated and used to transform ~.
coli DH5 cells to yield plasmid pMTI-3526. Plasmid pMTI-3526 was digested with AccI and ~glII; the -3.8 kb large fragment was gel-purified and ligated with the -1.7 kb AccI-~glII gel-purified fragment from pMTI-3525, then used to transform E. coli DH5~ cells.
The desired transformant plasmid was designated pNTI-3522. Plasmid , .

7Y~

pMTI-3522 was then used to transform competent E. coli CJ236 cells which are uracil N-glycosylase-deficient. Several transformants were propagated and single-stranded uracil containing pMTI-3522 (phage) DNA was generated with the use of R408 helper phage (available from Stratagene as catalog no. 200252). MUTA-l, a synthetic 60-mer which spans the ~unction of APP-751 and APP-695 was used to "loop out" the 57 nucleotides of pMTI-3522 to generate pMTI-3523. MUTA-l has the sequence:
5'agtactgcatggccgtgtgtggcagcgccattcctacaacagcagccagtacccctgatg 3' and was 5' phosphorylated and annealed to the single-stranded pMTI-3522 DNA in the presence of Gene 32 protein which assists the enzyme T4 DNA polymerase in copying the complementary strand. This mixture was used to transform competent E. coli XL-l Blue Cells (available from Stratagene as catalog no. 200268) which are uracil N-glycosylase positive. Colorless plaques were picked for miniprepand sequence analysis. This procedure (Geisselsoder et al., 1987, Biotechniques 5: 786-791) greatly reduced the propagation of parental phage, thus enriching for the mutant strand. One of thesa positive clones was designated pMTI-3523.

Construction of ~inigenes pMTI-2314 ant pMTI-2310 for E~pression of APP-695 A. Minigene pMTI-2314 For the first step of the construction of minigene pMTI-2314 for the expression of APP-695, an -4.6 kb EcoRI fragment derived from plasmid pMTI-3515 (Example l; Figure 3) was inserted into the HindIII site of plasmid pMTI-2301 (Figures 6a and 6b) by blunt-end ligation to yield plasmid pMTI-2307 (Figures 7a and 7b). Plasmid pMTI-2301 contains a unique HindIII cloning site, flanked by NotI
restriction endonuclease sites, and was prepared by first replacing the EcoRI-HindIII polylinker of pUCl9 (obtained from Bethesda Research Laboratories, Life Technologies, Inc., Gaithersburg, MD, catalog no. 95357) with the polylinker of pWE16 (available from Stratagene as catalog no. 251202), and then converting the ~l~dIII
~ite to sn EcoRI site u9ing adaptors. For the first step in the construction of pMTI-2301, two oligonucleotides purchased from NEB

7 ~

(catalog nos. 1107 and 1105) were annealed to yield the following double-stranded adaptor:
5'- MTTCGAACCCCTTCG-3' (#1105) 3'-GCTTGGGGM GCTCGA-5' (#1107) S Then, plasmid pUC19 DNA was digested with EcoRI and ~dIII and the -2.7 kb fragment was gel-purified (using low melt agarose), ligated with the adaptor, and used to transform E. coli DH5~ cells. The desired transformant was designated pMTI-2110 (Figure 6a). For the second step in the construction of pMTI-2301, plasmid pMTI-2110 DNA
was digested with EcoRI, then treated with calf intestine alkaline phosphatase (CIAP), then gel-purified. CIAP was obtained from Boehringer Mannheim Biochemicals, Biochemicals Division, P.O. Box 50816, Indianapolis, IN 46250 (catalog no. 713 023). Plasmid pWE16 DNA was also digested with EcoRI. The EcoRI-digested pWE16 and gel-purified EcoRI-digested pMTI-2110 DNAs were mixed, ligated, treated with GENECLEAN and used to transform E. coli DH5~ cells. The desired transformant plasmid was designated pNTI-2300. Niniprep analysis showed that NotI linearizes the -2.7 kb plasmid. For the third step in the construction of pMTI-2301, plasmid pMTI-2300 DNA was digested with ~mHI and ligated to self-annealing synthetic oligonucleotide adaptor (sequence 5'-GATCGGGAAGCTTCCC-3'; synthesized using an Applied Biosystems instrument and manufacturers recommended methods, model no. 380A DNA Synthesizer) in order to convert the BamHI site to HindIII. The oligonucleotide was annealed to yield the following double-stranded adaptor:
5'-GATCGGGMGCTTCCC-3' 5'-CCCTTCGAAGGGCTAG-5' The ligated DNA was treated with GENECLEAN and used tD transform ~.
coli DH5~ cells. Miniprep analysis of transformant DNA was performed using BamHI (plasmid remains uncut) and HindIII (linearizes plasmid).
The desired transformant was designated pNTI-2301.
Plasmid pMTI-2301 DNA, thus obtained, was digested with HindIII, gel-purified, then treated with Klenow and CIAP. Then, plasmid pNTI-3515 DNA was digested with EcoRI and an ~4.6 kb fragment was gel-purified, treated with Klenow, and blunt-end ligated with the pMTI-2301 DNA prepared as described above. The ligated DNA was treated with GENECLEAN and used to transform E. coli DH5~ cells.

'' .: .

, ~ .
:. ' - .. -- ~ :

S~ 7 Transformants were screened by miniprep analysis using EcoRI. The desired transformant plasmid was designated pMTI-2301 and contain~d EcoRI fragments of ~4.7 kb and -2.6 kb by miniprep analysis.
For the second step of the con-qtruction of pMTI-2314, an -1.4 kb BamHI fragment of pFC4 comprising nucleotides 100-1475 (Example 3, Figure 1) was ligated into the ~3~HI site at nucleotide position 99 of the APC cDNA sequence in plasmid pMTI-2307 to yield plasmid pMTI-2311 (Figure 7a). Diagnostic miniprep analysis of pMTI-2311 DNA digested with Xhol and NotI revealed fragments of -3.9 kb, -2.7 kb and -2.2 kb.
For the third step of the construction of pMTI-2314, an -2.4 kb XhoI fragment of pFC4 comprising nucleotides 1056-3353 and including 3' sequences, poly A track and SV40 sequences found in the Okayama and Berg vector (Okayama and Berg, 1983, supra) was ligated into the XhoI site at nucleotide position 1056 to yield plasmid pMTI-2312 (Figures 7a and 7c).
For the iinal step of the construction of pMTI-2314, an -0.6 kb SDh I fragment of pMTI-2304 containing SV40-derived RNA splicing and polyadenylation signals was ligated into the ~8hI site of pMTI-20 2312 to yield plasmid pMTI-2314 (Figures 7a and 8b). Plasmid pMTI-2314 DNA waq used as a minigene for the construction of APP-695 expressing transgenic mice as described in Example 11 below.
B. Minigene pMTI-2310 A second minigene for the expression of APP-695, pNTI-2310 (Figure 8a), was constructed according to the same four steps as outlined above for the construction of pMTI-2314, except that in the first step, an -2.4 kb HindIII fragment derived from plasmid pMTI-3501 (Example 1) was inserted into the HindIII site of pMTI-2301 (part A above) to yield plasmid pMTI-2306. Diagnostic miniprep analysis of pMTI-2306 DNA digested with NotI and ~a~HI revealed fragments of -2.7, -0.9 and -0.6 kb. The subsequent three steps yielded plasmids pMTI-2308 (diagnostic minipreps with NotI and XhoI
revealed fragments of -2.6, -2.3 and -1.6 kb), pMTI-2309 (diagnostic minipreps with HindIII to determine orientation revealed fragments 35 of -3.4, -2.9 and -2.6 kb) and pMTI-2310 (diagnostic minipreps with EcoRI revealed fragments of -2.7, -2.4, -2.3, -l.I and -0.9 kb), .~
.

7 ~

respectively. Plasmid pMTI-2310, containing the same APP-695 gene as pMTI-2314 but with a truncated regulatory region, was also used as a minigene for the construction of APP-695 expressing transgenic mice as described in Example 11 below.

EXA~PLE 5 Construction of Minigenes pMTI-2319 and pMTI-2320 for Expression of APP-770 ant APP-751 Minigenes pMTI-2319 and pMTI-2320 (Figure 8a) for the expression of APP-770 and APP-751, respectively, were each constructed in a single step digestion and ligation procedure via a simple interchange of AccI-BglII fragments. Plasmid pMTI-3521 DNA
(Example 3) encoding a full-length cDNA for APP-770 was digested with AccI and BglII. An -1.8 kb AccI-BglII fragment of pFC4-770 was ligated into the AccI and BelII site~ of pMTI-2314 to yield pMTI-2319. Diagnostic miniprep analysis of pNTI-2319 DNA digested with ScaI revealed fragments of -7.3 and -4.8 kb. A ScaI site exists in the inhibitor encoding domains of APP-770 and APP-751. Similarly, plasmid pMTI-3524 DNA (Example 3) encoding a full-length cDNA for APP-751 was digested with _ç~I and ~glII. An -1.75 kb AccI -~g~II
fragment of pMTI-3524 was ligated into the AccI and ~glII sites of pMTI-2314 to yield pMTI-2320. Diagnostic miniprep analysis of pMTI-2320 DNA digested with ScaI revealed fragments of -7.2 and -4.8 kb.
Plasmids pMTI-2319 and pMTI-2320, containing a full-length cDNA for APP-770 and APP-751, respectively, were used as minigenes for the construction of APP-770 and APP-751 expressing transgenic mice as described in Example 12 below. Thus, the inhibitor encoding domains found in APP-770 and APP-751 may be inserted in the APP-695 sequence of pMTI-2314 by swapping sequence domains between the unique AccI
and ~glII sites.

Construction of p~TI-2321 and pMTI-2322 Minigenes for Expression of APP C-Terminal Frameshift ~utants Minigenes pMTI-2321 and pMTI-2322 (Figure 8a) for the expression of a truncated APP protein were constructed by making frameshift mutations in the C-terminal region of the APP coding , 2 ~

sequence. These frameshift mutations were made in the APP cDNA
sequences immediately following the A4 coding region so as to bring a translation stop codon into the reading frame (i.e., in-frame termination) following the A4 peptide ccding region. The resulting mutated sequence codes for a truncated APP species as exemplified by gene product IV shown in Figure 4b.
1~ vitro mutagenesis procedures described by Kunkel et al., 1987, Methods In Enzymol. 154: 367-382, were used to generate the frameshift mutants briefly summarized as follows. The starting material for the mutagenesis was plasmid pMTI4 DNA. Plasmid pMTI4 is the mutagenesis vector KS Bluescript (+) available from Stratagene into which the -2.3 kb NruI-SpeI fragment of pFC4 containing a segment of the APP-695 cDNA has been inserted. For this construction, pFC4 DNA was digested with NruI and SpeI, treated with Klenow, then blunt-end ligated into the SmaI site of SmaI-digested KS Bluescript(+) DNA to yield pMTI-4. Single-stranded pMTI4 DNA was prepared from E. S~l~ W236 host cells, in which cells uracil replaces thymine in DNA. The DNA was then made double-stranded by ~ vitro DNA synthesis using one of three mutagenizing synthetic oligonucleotides described below as primer for a particular preparation. The heteroduplex DNA (one uracil-containing normal pMTI4 strand and one newly synthesized thymine-containing mutated strand) was used to transform E. coli MVll90 cells, in which cells the sequence of the thymine-containing mutated strand is selectively propagatedbecause the uracil-containing wildtype strand is degraded.
Miniprep plasmid preparations from such transformed E. coli colonies were screened for incorporation of the mutation by direct DNA
sequence analysis. For sequence analysis, the primer was an oligonucleotide having the sequence from nucleotide position 1881-30 1897 of the APP cDNA. The sequence -200 nucleotides downstream of the primer was analyzed to confirm the mutated sequence. Those clones having the desired mutation of the APP coding sequence were used for preparative purification of mutant plasmid DNA by conventional methods, for use in the construction of truncated APP
minigenes.

' ~ ' ' ~ ," ' .

Three types of mutants (a, b, c) were generated which introduced premature translation terminat~on signals in APP mRNA to yield truncated APP proteins. The wildtype (wt) and mutant sequences beginning at nucleotide position 2040 are shown with the termination codons underlined as follows:
a b c wt GTG GGC GGT GTT GTC A~ GCG ACA GTG _TC
Val Gly Gly Val Val Ile Ala a GTG GGG GTG TTG TCA TAG CGA CAG TGA TCG
Val Gly Val Leu Ser *
b GTG GGC GGT GTT GTG TCA TAG CGA CAG TGA
TGC
Val Gly Gly Val Val Ser *
c GTG GGC GGT GTT GTC TAG CGA CAG TGA TCG
Val Gly Gly Val Val *
The synthetic oligonucleotides used for mutagenesis were:
a 5'-CATGGTGGGGGTGTTGTCATAGC-3' [23-mer, 2036-2059]
b 5'-GGGCGGTGTTGTGTCATAGCGACAG-3' [25-mer, 2042-2064]
c 5'-GGCGGTGTTGTCTAGCGACAGTGA-3' [24-mer, 2043-2067]
For mutant-a, one nucleotide (C) that is marked above the wildtype sequence with the letter "a", i9 deleted, generating two in-frame termination codons. The first in-frame termination codon in mutant-a is the codon for aa 40 of the A4 sequence. In mutant-a, amino acids 38, 39 and 40 are different than those of the wildtype sequence. For mutant-b, two nucleotides (TG) were inserted at the position marked above the wildtype sequence with the letter "bn, generating two in-frame termination codons. The first in-frame termination codon in mutant-b is after the codon for aa 41 of the A4 sequencs. In mutant-b, aa 41 i9 different than that of the wildtype sequence. Mutant-b (also known as the +2 mutant) was utilized in the construction of plasmid pMTI-2321 described below. For mutant-c, one nucleotide (A) that is marked in the wildtype sequence shown above with the letter "c" is deleted, generating an in-frame termination codon in the reading frame directly following the codon for aa 40 of the A4 sequence. Mutant-c has been designated mutant 40-1, and was utilized in the construction of plasmid pMTI-2322 described below.

2~ r~

The portion of APP-695 cDNA that contains the A4 coding sequence lies within an -0.7 kb B lII-ClaI fragment (corresponding to nucleotide position 1915-2620) that can conveniently be moved from one APP gene construct to another since BglII and ClaI each cleave APP-695 cDNA only once. The following steps were used to insert the mutated part of pMTI4 into another APP construct to generate minigenes for expression of truncated APP proteins. In the first step, mutated pMTI4 DNA was digested with ~glII and ClaI. The ~0.7kb ~glII-,ClaI fragment wa~ then isolated from a preparative agarose gel. In the second step, DNA from the other construct was digested with BelII and ClaI and then treated with CIAP. For the preparation of plasmids pMTI-2321 and pMTI-2322, this other construct was pMTI-2314 (Example 4) encoding a full-length (wildtype) cDNA for APP-695. The small -0.7 kb BglII-ClaI fragment of pMTI-2314 was removed from the digest by preparative agarose gel electrophoresis.
In the third step, the large -11.1 kb BglII-ClaI fragment of pMTI-2314 remaining after removal of the -0.7 kb fragment to be replaced (step 2) was mixed with the -0.7 kb mutated ~glII-ClaI fragment (step 1), then ligated and,used to transform E. coli DH5~ cells.
Transformant plasmids were initially screened for appropriate inserts by miniprep analysis. Diagnostic miniprep analysis of the plasmids using EcoRI revealed fragments of -4.7, -2.7, -2.6, -1.1 and -0.9 kb.
Then, the integrity of each selected plasmid preparation was confirmed by DNA sequence analysis of the mutated sequence and sequences surrounding the mutation. The resulting selected plasmids designated pMTI-2321 (mutant-b) and pMTI-2322 (mutant 40-1), were used as minigenes for the construction of transgenic mice expressing truncated APP proteins. A plasmid for the expression of mutant-a may be constructed and selected by fragment swapping as described above for mutant-b (pMTI-2321) and mutant 40-1 (pMTI-2322).

E~IPLE 7 Construction of Min$gene pMTI-2318 for Espression of A4 Amyloid Peptide In order to construct plasmid pMTI-2318 (Figure 9) containing a minigene for the expression of A4 peptide, first a portable gene encoding the 42 aa A4 peptide sequence was prepared. This gene was ' , ' -42- 2~ 7 7 obtained by in vitro mutagenesis of a fragment of the cDNA for APP-695 as described in Example 6 above. The starting material was the same as that described in Example 6, plasmid pMTI-4. A 38-base oligonucleotide mutagenesis primer corresponding to the A4 sequence, S but with desired changes for the creation of an in-frame termination (TAG) and a convenient ~mHI restriction site immediately following the in-frame termination codon, was synthesized chemically. The sequence of this synthetic primer was:
BamHI
5'-GGTGTTGTCATAGCGTAGGATCCGTCATCACCTTGGTG-3' Ter This primer was used to mutate the APP-695 sequence in pMTI4 in substantial accordance with the procedure in Example 6 above to create plasmid pMTI-26. The wildtype (native) and mutated sequences are shown as follows:
BglII
wt AGATCTCTG MGTGAAGA~ GAT---GGT GTT GTC ATA GCG ACA GTG ATC-MET asp gly val val ile ala thr val ile BglII BamHI
mutant AGATCTCTG MGTGMG~ GAT---GGT GTT GTC ATA GCG TAGGATCCGT
MET asp gly val val ile ala ter The newly created ~HI site ant the ~glII site preceding the ATG
codon provide convenient restriction sites for cloning the A4 gene into other APP constructs to generate minigenes for expression of 42 aa A4 peptide. One such minigene construct was pMTI-2318, prepared according to the following steps (Figure 9). In the first step pMTI-26 DNA was digested with BglII and BamHI. The -150 bp fragment was then isolated from a 5~ polyacrylamide gel by electroelution. In the second step, DNA from another construct, for example, pMTI-2307 (Example 4) was digested with ~3~HI, gel- purified and treated with CIAP. In the third step, the -150 bp mutated ~glII-BamHI fragment (step 1) was mixed with the ~HI-cut pMTI-2307 DNA
(step 2), then ligated and used to transform E. coli DH5~ cells.
Transformant plasmids were screened for appropriate inserts by miniprep and DNA sequence analysis. For each miniprep analysis restriction endonucleases were chosen that would allow starting and ending materials to be distinguished and also allow determination o~

", j ~, ... . . .

-43- ~ 7 ~
desired orientation. Then, other re-qtriction endonucleases were chosen to confirm the integrity of the construction (e.g., no anomalous rearrangements). The resulting -7.6 kb plasmid was designated pMTI-2316. Miniprep analysis with ~EHI and EcoRI
S revealed fragments of -6.8 and ~0.8 kb. In the next steps, pMTI-2316 DNA was digested with ~3~HI, then mixed with and ligated to the -2.0 kb BamHI fragment of pFC4 to yield pMTI-2317 by transformation and selection as described above. Miniprep analysis of the -9.6 kb pMTI-2317 DNA with HindIII revealed fragments of -7.3 and -2.2 kb.
Insertion of this ~HI fragment provided a convenient ~E_I site along with a portion of 3' untranslated sequences so as to be a template for a messenger RNA transcript of stable size. In the final steps, an -0.6 kb ~E~I fragment of pMTI-2304 containing SV40 splicing and polyadenylation signals was inserted at the ~hI site of pMTI-2317 by appropriate restriction endonuclease digestion, ligation,transformation and selection as described above, to yield desired plasmid pMTI-2318. Miniprep analysis of pMTI-2318 DNA with EcoRI
revealed fragments of -2.9, -2.7, -2.0, -1.1, -0.9 and -0.6 kb.
Plasmid pMTI-2318 was utilized as a minigene for the construction of transgenic mice expressing a 42 aa A4 peptide.

Epitope Tagging of Recombinantly Expressed APP
Immunochemical studies of the subcellular localization and processing of alternative forms of APP, including APP-695, APP-770 and APP-751, and mutated forms of APP, using antibodies elicited to synthetic peptides and/or recombinant precursor proteins are difficult for the following reasons. Firstly, the APPs are highly conserved among species (mouse and rat 99%, human and rat 97.3~) and are ubiquitously expressed. In order to easily obtain adequate antibody production against a variety of APP peptides and recombinant proteins, a highly antigenic epitope of Chlamvdia (Huguenel et al., 1989, Intl. Soc. Sex. Trans. Dis. Res., 8th Meeting, Copenhagen, Denmark, Abstract no. 22) was inserted into the APP sequence at either the Kunitz inhibitor domain or the extreme C-terminus of the protein to produce "tagged" APP cDNAs. Minigenes containing such ~ ~3 ~

tagged APP cDNA can be used to prepare transgenic mice, and the APP
translation products in such mice can be detected using antibodies against this epitope.
The peptide sequence of the Chlamvdia epitope is TVFDVTTLNPTI.
This epitope has been shown to be very antigenic as an isolated peptide and as part of a larger protein (Huguenel et al., supra;
Baehr et al., 1988, Proc. Natl. Acad. Sci. USA 85: 4000-4004;
Stephens et al., 1988, J. Exp. Med. ~ 817-831). Synthetic oligonucleotides were generated for the site-directed mutagenesis in the APP coding region to insert the sequences for the ChlamYdia epitope by M13 "looping-in" experiments. The synthetic oligonucleotide 5~ Aat Il GGCTGCTGTG GCGGGEGTCTA AAT AGT TGG GTT CAG AGT GGT GAC GTC AAA
* I T P N L T T r V D F
GAC AGT GTT CTG CAT CTG CTC AAA GA 77-mer (CC-TAG) V T N Q M Q B F F
was used to engineer pMTI-63 which carries a C-terminal addition of sequences encoding the Chlamvdia epitope to APP695. The synthetic oligonucleotide 5~ Aat II
G Ç~ ACT GGC TG~C TGT TGT AGG AAT AGT TGG GTT CAG AGT GGT GAC GTC
T S A A T T P I T P N L T T V D
AAA GAC AGT Q~ AAC CAC CTC TTC C 77-mer (IC-TAG) F V T V R V V E E
was used to engineer pMTI-35 which carries an addition of the amino acid sequences encoding the Chlamvdia epitope into the APP sequence of amino acid residue 289, where the (Kunitz) protease inhibitor-like domain is spliced into the APP-695 molecule.
Antibodies prepared against this Chlamydia epitope are useful to investigate the tissue, cellular and subcellular localization of tagged APP proteins ~ vivo, to study the biochemical properties and processing of APP in transfected animal cells including cell-sorting of transfectants, to study APP ~ vitro translation products and APP
transformed E. coli products on Western-blots, and to follow processing of APP and its subcellular localization in transgenic 2 ~ 3 7 ~

animals. Such studies should permit the identification of the functional role of APP in normal individuals and in individuals with AD or DS. Minigenes pMTI-2324 and p.~TI-2325 (Figure 8a) for the expression of APP-695 (IC-TAG) were each constructed in a single step digestion and ligation procedure via a simple interchange of AccI and ~glII fragments. Plasmid pMTI-35 DNA was digested with AccI and BglII. An -1.6 kb AccI a-~elII fragment of pNTI-35 was gel-purified and ligated èither into the AccI and BglII sites of the gel-purified large DNA fragment of pMTI-2314 to generate pMTI-2325 (Figure 8a) or into the AccI and BelII sites of the gel-purified large fragment of pMTI-2323 to generate pMTI-2324 (Example 9, Figure lOa). Diagnostic miniprep analysis of pNTI-2324 DNA digested with HindIII revealed fragments of ~1.3, -4.4 and -7.7 kb. Diagnostic miniprep analysis of pNTI-2325 DNA digested with AatII revealed fragments of -4.2, -3.3, -3.1, -0.6 and -.055 kb. In an analogous manner, the -1.6 kb AccI-~glII fragment of pMTI-35 can be ligated into the AccI and BglII sites of ~he gel-purified large DNA fragment of pMTI-2329 (Example 9, Figure 14) to generate pMTI-2335.

APP Minigenes with Metallothionein-Derived Regulator or Processing Sequences A. APP Minigenes with Metallothionein-Derived RNA Processing Si~nals Minigenes utilizing RNA processing signals derived from an exogenous mouse gene might be more efficiently expressed in transgenic mice as compared with minigenes utilizing SV40-derived ~NA processing signals as described in Examples 4-8 above.
Therefore, alternate minigene constructs were generated in which RNA
splicing and polyadenylation sequences of the mouse netallothionein gene were utilized. One source of the mouse metallothionein gene is plasmid pJYMMT(L) (alternatively designated pCL-28 or T25). Plasmid pJYMMT(L) is an -12.4 kb genomic clone of the metallothionein gene described by Hamer and Walling, 1982, J. Mol. App. Gen. 1: 274-28c.
This alternate series of minigenes utilizing metallothionein RNA
signals was constructed using pJYNMT(L). Many of these alternate minigenes were generated via fragment swaps using the minigenes containing SV40 RNA signals described in Examples 4-8.
1. Alternate Minigenes Exp~essing APP-695. APP-770 and APP-751 A minigene utilizing mouse metallothionein-I gene RNA
processlng signals (splicing and polyadenylation) and expressing APP-695 was constructed in a single step as follows. A Klenow- treated ~2.2 kb ~g~ EcoRI fragment of pJYMMT(L) containing all of the mouse metallothionein-I genomic gene sequence except the promoter was inserted by blunt-end ligation into the ClaI site of plasmid pMTI-2312 DNA (Example 4) that had been digested with ClaI, and treated with Klenow and CIAP to generate plasmid pMTI-2323 (Figure lOa and lOb). Plasmid pMTI-2323 was selected in a two-step screening procedure. First, transformant plasmids from the blunt-end ligation were screened by colony hybridization using the insert fragment (-2.2 kb BelII-EcoRI) of pJYMMT(L) labelled with 3ZP as probe.
Colony hybridization was used as a first step in screening a variety of constructs disclosed herein when the background of transformant plasmids that were vector alone (i.e., no insert) was high. In the second screening step, the desired plasmid was selected from those positively hybridizing colonies by miniprep analysis of restriction endonuclease digested DNA. For pMTI-2323 (~13.3 kb), miniprep analysis using HindIII revealed fragments of -7.7, -4.4 and -1.3 kb.
Minigenes utilizing metallothionein RNA signals and expressing APP-770 or APP-751 can be constructed via AccI-BglII fragment exchanges (Figure lOa) with either pMTI-3521 or pMTI-3524 (Example 3) respectively. Specifically, the -1.8 kb AccI-~glII fragment of pMTI-3521 was inserted into the AccI-~glII sites of pMTI-2323, replacing the -1.5 kb AccI-~glII fragment of pNTI-2323, to generate plasmid pMTI-2331 (Figure lOa). For example, for pMTI-2331, pMTI-2323 DNA was digested with AccI and aglII and the -11.8 kb fragment was gel-purified, treated with CIAP and ligated with the -1.8 kb AccI-~glII fragment that was gel-purified from pMTI-3521 DNA. The ligation mixture was used to transform E. coli DH5~ cells. The desired plasmid, pMTI-2331 (-13.3 kb), was identified by miniprep analysis. Using ScaI, miniprep analysis of pMTI-2331 revealed fragments of -9.0 and -5.0 kb.

., .

2 ~ 7 ~

2. Alternate Minigenes Expressing APP C-Terminal Frameshift Mutants A minigene utilizing metallothionein RNA signals and expressing a truncated APP protein was constructed via a fragment swap using minigene pMTI-2322 (Example 6, Figure 8a) containing SV40 RNA
signals. Specifically, the -0.7 kb ~lII-ClaI fragment of pNTI-2322 containing mutation 40-1 (Example 6), was inserted into pNTI-2312 (Example 4, Figure 7a), via ligation of the -0.7 kb fragment of pMTI-2322 with the -10.5 kb ~glII-ClaI fragment of pMTI-2312 (that had been gel-purified and treated with CIAP prior to ligation), to gsnerate pMTI-2326a. Miniprep analysis of the -11.2 kb pMTI-2326a DNA using HindIII revealed fragments of -7.8 and -3.4 kb. Then, the -2.2 kb BglII-EcoRI metallothionein fragment of pJYMNT(L) was inserted into the ClaI site of pMTI-2326a by blunt-end ligation to 15 generate pMTI-2326. Miniprep analysis of the -13.4 kb pMTI-2326 DNA
using ~lgdIII revealed fragments of -7.7, -4.4 and -1.3 kb.
Alternatively, plasmid pMTI-2326 can be constructed in a one-step fragment swap. Specifically, the ~0.6 kb BglII-SpeI fragment of pMTI-2322 (Example 6, Figure 8a) can be inserted directly into pMTI-20 2323, replacing the -0.6 W ~ p~I fragment of pMTI-2323 to 8enerate pMTI-2326 (Figure lOa). Alternative minigenes can be generated by analogous ~lII-S~eI fragment exchanges between pMTI-2323 and constructs encoding alternate truncated forms of APP-695 (Examples 6 and 7).
25 3. Alternate Minigenes Ex~ressing C-100 (Plasmid pMTI-2327) and MC-100 (Plasmid pMTI-2337) APP Mutants A minigene utilizing metallothionein RNA signals and coding for the ~utation designated C-100 (gene product VI, Figure 4b) was 30 generated by deleting the -1.8 kb NruI-~glII fragment of pMTI-2323 ~this example, part A) with blunt-end ligation to generate plasmid pMTI-2327 (Figure lla). Plasmid pMTI-2323 DNA was digested with NruI and BglII; the -11.5 kb fragment was gel-purified, treated with Klenow, ligated, and used to transform ~. coli DH5~ cells.
Transformant plasmids were screened by miniprep analysis and the desired plasmid pMTI-2327 was selected. Miniprep analysis of pMTI-2327 DNA using ~i~dIII revealed fragments of -7.7, -2.6 and -1.3 kb.

.
', ' `~ ' ' - . . .

.

2 ~

A translation initiation codon, ATG, directly precedes the first codon of the A4 sequences in C-100.
A minigene for expressing the mutation designated MC-100 (gene product V, Figure 4b) was also prepared from plasmid pMTI-2323 to generate pMTI 2337 (Figures lla and llb). Specifically, pMTI-2337 was generated by deleting the ~1.7 kb ~ 21II fragment of pMTI-2323 via gel purification of the ~11.6 kb fragment and ligating using a synthetic oligonucleotide linker, sp-spacer-A4. The linker sp-spacer-A4 was inserted between the ~e~I site at position 207 and the BglII site at position 1915 in APP-695, and had the following sequence:
5'-GGACGGAGGA-3' 10-mer 3'-CATGCCTGCCTCCTCTAG-5' 18-mer The two oligonucleotide sequences that comprise sp-spacer-A4 were synthesized (using an Applied Biosystems instrument and manufacturer's recommended methods, model no. 380A DNA Synthesizer), kinased and annealed according to conventional methods before ligation with the gel-purified -11.6 kb KpnI-BglII fragment of pMTI-2323 to generate pMTI-2337. Miniprep analysis of pMTI-2337 DNA with HindIII revealed fragments of -7.7, -2.6 and -1.3 kb.
MC-100 requires the 17 residue signal peptide of APP to direct translation and insertion of the mutant protein into the membrane.
The signal peptide will be clea~ed and eliminated during the membrane insertion. The nucleotide and amino acid sequence of MC-100 is shown in Figure 12.
4. Alternate Minigene Ex~ressing A4 Pe~tide (Plasmid pMTI-2341) A minigene utilizing metallothionein RNA signals for expressing the A4 peptide (gene product VII or sp-A4, Figure 4b) was prepared by deleting the -11.6 kb KpnI-BglII fragment of plasmid pMTI-2326 (this example, part A,2) and ligating using the sp-spacer-A4 linker (described in part A,3 above) to generate plasmid pMTI-2341.
Minigene pMTI-2341 generates sp-A4, which an A4 peptide linked to the APP signal sequence. The nucleotide and amino acid sequence of sp-A4 is shown in Figure 13.

.

, . . . ;

~6~7 ~

B. APP Minigenes Uith A Metallothlo~ç~n Derived Promoter The generation of transgenic mice which express APP (or derivatives of APP) in cells and tissues not normally expressing the gene may lead to dominant phenotypes. The new pheno~ypes may facilitate a better understanding of the function of APP. To this end, a series of minigenes was constructed which minigenes are under the regulation of the mouse metallothionein gene promoter (Figure 14).
1. Alternate minigenes expressing APP-695~ APP-770 and APP-751 A minigene utilizing a metallothionein promoter and expressing APP-695 was constructed as follows. Plasmid pMTI-2301 DNA (Example 4) was digested with HindlII, treated with Klenow and CIAP. Plasmid pJYMNT(L) DNA (part A above) was digested with EcoRI. An -4.0 kb EcoRI fragment was gel-purified, treated with Klenow and blunt-end ligated to the pMTI-2301 DNA treated as described above. The desired transformant plasmid was designated pMTI-2328 (-6.7 kb). In the next step, the pMTI-2328 DNA thus obtained was digested with ~glII, treated with Rlenow and CIAP, gel-purified and then blunt-end ligated to an -2.8 kb gel-purified SmaI-HindlII fragment of pMTI-2314 (-11.8 kb, Example 4). Transformant plasmids were screened by miniprep analysis and the desired plasmid pMTI-2329 (Figure 14) was selected.
Miniprep analysis of pMTI-2329 DNA using EcoRI revealed fragments of -3.7, 3.1 and -2.7 kb.
Minigenes utilizing a metallothionein promoter from pMTI-2329 and expressing APP-770 or APP-751 are constructed via fragment swaps with pMTI-2319 (alternatively, pMTI-2331 or pMTI-2342) or pMTI-2320 (alternatively, pMTI-2345), respectively. Specifically, an -7.4 kb AccI-SpeI fragment is gel-purified and ligated with an -2.4 kb AccI-SpeI fragment of pMTI-2319 or pMTI-2330 to yield pMTI-2333 for APP-770 expression and pMTI-2334 for APP-751 expression, respectively.
2. Alternate Minigenes Expressing APP C-Terminal Frameshift Mutants By a similar fragment swap, a minigene utili~ing a metallothionein promoter and expressing a truncated APP protein is constructed. Specifically, an -2.1 kb AccI-SpeI gel-purified fragment of pMTI-2322 (alternatively, pMTI-2343 or pNTI-2326) !

'` ~ ' . .

~ ' . .

7 ~
so containing mutation 40-1 (Example 6) was ligated with the -7.4 kb AccI-SpeI gel-purified fragment of pMTI-2329 described above.
3. Alternate Minigene Expressing MC-100 APP Mutants An alternate minigene for the expression of the ~C-100 mutation (part A above) using a metallothionein promoter may also be prepared.
For example, the -1.7 kb ~ glII fragment of pMTI-2329 may be deleted via digestion with ~e~I and ~elII, then gel purification of the -7.8 kb ~nI-~glII fragment and finally ligation with the sp-spacer-A4 (part A above). The desired plasmid i9 confirmed by sequence analysis.

APP ~inigenes with Genomic APP-Derived RNA Processing Signals APP minigenes utilizing RNA processing signals derived from the human APP gene might be more efficiently expressed in transgenic mice as compared with minigenes described in Examples 4-8 above utilizing SV40 derived RNA processing signals or minigenes described in Example 9 above utilizing metallothionein gene-derived signals.
Therefore, minigene constructs were generated in which RNA
polyadenylation signals of the human APP gene were utilized. The source of the human APP genomic sequences for these constructs was plasmid pVS-l. Plasmid pVS-l is an -4.3 kb genomic clone of the human APP gene which comprises an -1.5 kb EcoRI genomic fragment inserted into the EcoRI site of pUCl9 in the orientation shown in Figure 15a, so that the APP polyadenylation signal can be recovered as an -1.3 kb ~ihI fragment. The -1.5 kb EcoRI fragment encompasses the 3'-end of the terminal exon of human APP and the APP
polyadenylation signal and was isolated as follows. A Charon 21A
lambda library of human chromosome 21 DNA (available from the A.T.C.C. as accession no. LA21NS01) was screened for clones containing 3'-end genomic sequences with a small SmaI-~e~I fragment (nucleotides 3102-3269) from plasmid pFC4 labelled as a probe. The nucleotide sequence of the -1.5 kb APP genomic fragment is shown in Figure 16. An alternate saries of minigenes utilizing APP RNA
signals derived from pVS-l were constructed. Many of these alternate minigene9 were generated ~ia fragment swaps using pNotSV2neo -51- ~ 7 subclones of the APP constructs. These pNotSV2neo s~bclones were utilized for switching sequence domains via fragment swaps because of the presence of convenient PvuI and ~I restrlction enzyme sites.
NotI fragments of many of the APP minigenes described in Examples 4-8 were subcloned into pNotSV2neo tsee Figures 17 and 18a) so that APP
expression could be determined in transient transfections of COS
cells (Gluzman, 1981, Cell ~: 175-182). Plasmid pNotSV2neo tFigure 17) was prepared by converting the unique ~HI site of pSV2-neo (available from the A.T.C.C. as accession no. 37149) to a NotI site 10 using linkers (NEB catalog no. 1045). Plasmid pSV2-neo was digested with ~HI, treated with Klenow and CIAP, and ligated to NotI linkers as recommended by the supplier. The pNotSV2neo subclones of the APP
constructs used in the preparation of alternate Dinigenes were prepared as summarized in Table I below and Figure 18a. The construction of each of the APP minigenes utilizing APP genomic RNA
processing signals from pVS-l listed in Table I is described below.

- ~

~ r~ ~
.
TABLE I
pNotSV2neo APP Subclones Subclone APP Sequence Insert in p~otSV2neoa pMTI-2360 APP-695 -10.6 kb ~I fragment of pMTI-2323 pMTI-2362 APP-695 -6.8 kb NotI fragment of pMTI-2329 pMTI-2369 APP-695 -9.8 kb MotI fragment of pMTI-2339 pMTI-2363 APP-770 -10.8 kb NotI fragment of pMTI-2331 pMTI-2368 APP-751 -9.2 kb NotI fragment of pMTI-2320 pMTI-2361 Mutation 40-1 -10.8 kb NotI fragment of pMTI-2326 pMTI-2364 MC-100 ~8.0 kb NotI fragment of pMTI-2340 pMTI-2366 MC-100 -8.8 kb NotI fragment of pMTI-2337 pMTI-2365 Sp-A4 -8.8 kb NotI fragment of pMTI-2341 pMTI-2367 ~-gal -8.6 kb NotI fragment of pMTI-2402 ~ For preparation of the subclones, each insert was gel-purified and ligated into pNotSV2neo vector DNA that had been digested with NotI, gel-purified and treated with CIAP.

A. Alternate Minigenes Expressing APP-695. APP-770 and APP-751 ~Plasmids pMTI-2339. pMTI-2342 and pMTI-2345~
The minigene construct designed to express APP-695 was generated by inserting an -1.3 kb Sphl fragment from pVS-l into the SehI site of pMTI-2312 (Exa~ple 4) to generate pMTI-2339 (Figures 15a and 15b). Minigene pMTI-2342 expressing the APP-770 alternate form of APP was generated by inserting the -6.9 kb PvuI-SpeI fragment of pMTI-2363 (Table I, Figure 18a) into the PvuI-SpeI fragment of pMTI-2369 (Figure 19). Plasmid pMTI-2369 was itself generated by 2 ~

inserting the ~9.8 kb ~I fragment of pMTI-2339 into pNotSV2neo (Table I and Figure 19). Minigene pMTI-2345 expressing the APP-751 alternate form of APP was generated analogously by inserting the -6.9 kb PvuI-S~eI fragment of pMTI-2368 (Table I) into the -8.8 kb PwI-SpeI fragment of pMTI-2369 (Table I).
B. Alternate Minigene~ fQ~y_~xpressing C-Terminal Frameshift Mutants 1. Mutant 40-1 (Plasmid pMTI-2343) Minigene pMTI-2343 expressing the 40-1 frameshift mutant was generated by a fragment swap. The -6.7 kb PvuI-S~eI fragment of pMTI-2361 (Table I, Figure 18a) was inserted into pMTI-2369 (Figure 19) ~
C. Alternate Minigene For MC-100 Mutant Minigene pMTI-2340 expressing the MC-100 deletion mutant was generated by deleting an -1.7 kb ~a~ glII fragment of pMTI-2339 (Figures 15a and 15b) and ligating using the sp-spacer-A4 synthetic linker described in Example 9 above.
D. Alternate Mini~ene for A4 Peptide Minigene pMTI-2344 expressing the A4 peptide was generated by a fragment swap (Figure 19). The -5.0 kb PvuI-SpeI fragment of pMTI-2365 (Table I, Figure 18a) was inserted into the -8.8 kb ~y_I-SpeI
fragment of pMTI-2369 (Figure 19).

E~LWPLE 11 Preparation and Analysis of Transgenic Mice Espressing APP Minigene3 Transgenic mice are mice that contain exogenous DNA integrated into their genomes (Gordon and Ruddle, 1981, Science 214: 1244-1246). The DNA thereby integrated is called a transgene. APP
minigenes prepared as described in Examples 4-10 may be used to prepare corresponding transgenic mice expressing these transgenes.
The technical aspects of the procedure for preparing transgenic mice have been the subject of extensive review by Gordon and Ruddle, 1983, Methods Enzymol. lOlC: 411-433, and Hogan et al., 1986, Manipulation of the Mouse Embryo: A Laboratorv Manual, Cold Spring Harbor Lab., Cold Spring Harbor, NY, and are hereby incorporated by reference.

., ~ . .

-2 ~ 'd 7 The general procedure involves gene transfer by microinjection.
Fertilized l-cell mouse embryos are dissected from superovulated female mice [strain: Hsd:(ICR)BR] mated with male mice (strains: Hsd:
(ICR)BR or B6D2Fl/HsdBR). Transgenic mice generated from a homozygous, or inbred, strain of mice are created using embryos from C57BL/6NHsdBR mating partners. Embryos are cultured in vitro as described in Hogan et al. (suDra). Microin~ections were performed as described in DePamphilis et al., (supra). Approximately 100 to 500 copies of a linear NotI fragment (-6-11 kb in size) of an APP
minigene (listed in Table II) are loaded into a microinjection pipet and expelled into one of the pronuclei of a l-cell mouse embryo.
Approximately 1 to 3 pl of DNA in;ection fragment solution (approximately 5-lO ~g/ml linear DNA, 0.3 mM EDTA, and lOmM Tris pH
7.5) is injected into a pronucleus of each l-cell embryo. During injection, mouse embryos are held in-place by a microscopic cell holder. Surviving embryos were then surgically reimplanted into pseudo-pregnant foster mice (strain: Hsd:(ICR)BR) as described in DePamphilis et al. (supra) and Hogan et al. (su~ra). Progeny mice are born approximately 19 dayq post-implantation and approximately 10-30% of the progeny are transgenic (i.e., their chromosomes carry one or more copies of the in~ected APP minigene) and are designated as transgenic founders. Positive transgenic mice are designated by either Southern-blot or PCR analysis of tail-biopsy DNA (See below).
Transgenic founder mice are bred with appropriate partners, strain:
Hsd:(ICR)BR for outbred strain background, or C57BL/6NHsdBR) for inbred strain background, to generate heterozygote Fl progeny.
Transgenic siblings (Fl) are then inbred to generate a homozygous (for transgene) line of mice. Glass cell-holders are constructed using borosilicate glass capillaries (l~m od. and 0.58m~
30 id.; from Sutter Instruments Co., San Rafael, CA, part #B100-58-15) on a microfuge (de Fonbrune-type; Technical Products International Inc., St. Louis, M0). The tips of the cell-holders are fire-polished and have a diameter of approximately 50 microns.
Microinjection pipets are beveled and have a diameter of approximately 2 microns at their terminus. To make microin;ection pipets, glass capillaries (lmm od. and 0.78mm id.; from Sutter .

2 ~ ~ ~ O I

-ss-Instruments Co., part #B100-78-15) were pulled on a Sutter Instruments Co. micropipet puller (model #P-80/PC) and then the tips were beveled on a Sutter micropipet beveler (model #BV-10; bevel angle approximately 25 to 30). Pulled pipets are siliconized by incubation in a glass chamber saturated with hexamethyldisilazane (HMDS; Pierce #84769) for approximately 8 hour~ at room temperature.
Microin~ections are performed on a Zeiss IM-35 inverted microscope using Nomarski optics. Microinjection pipets and cell holders are controlled using Narishigi (Japan) micromanipulators (model #MO-102M and #MN-2). The flow of the injection solution in microinjection pipets is controlled using an Eppindorf Microinjector (model #5242). Surgical reimplantations are performed using a Zeiss SV8 dissection microscope.
DNA in;ection fragments were isolated from vector sequences by NotI digestion and agarose gel electrophoresis. Linear DNA
fragments were recovered from the agarose gels by electrophoresis onto a NA45 membrane (Schleicher and Schuell, catalog no. 23410).
The NotI linear DNA fragment was recovered from the membranes according to the manufacturer's instructions. Ethidium bromide was extracted from the DNA solution using isopropanol (buffer saturatet, 1 mM EDTA and 10 mM Tris pH 7.6). DNA was precipitated by addition of a half volume of 7.5 M ammonium acetate and then by 2.4 volumes of absolute ethanol. The DNA pellet was resuspended in TE buffer (1 mM EDTA and 10 mM Tris pH 7.6) and then reprecipitated in ammonium acetate and ethanol as described above. DNA was reprecipitated a total of three to four times. DNA in;ection fragments were finally resuspended in injection buffer (0.3 m~ EDTA, and 10 mM Tris pH 7.5).
DNA concentration of the fragment was estimated by ethidium bromide staining on diagnostic agarose gels against known concentrations of DNA as standards. Fragments obtained in this manner were diluted to a concentration of 5 ~g/ml.
Positive transgenic mice are identified by either Southern-blot or PCR analysis of tail-biopsy DNA. Southern-blot analysis is performed as described in Uirak et al~, 1985, Mol. Cell Biol. 5:
35 2924-2935 and Maniatis et al. (supra). PCR analysis of tail-biopsy DNA is described below.

Tail biopsies are performed by dissecting approximately 1 cm of mouse tail from each mouse. Tail segments are cut into small fragments and incubated in 1.0 ml of tail extraction buffer (0.5 mg/ml proteinase K, 0.5~ SDS, 100 mM EDTA, and 50 mM Tris pH 8.0) at 55C for 12 to 16 hours. Samples are then extracted with 1.0 ml phenol (equilibrated with 1 mM EDTA and 10 mM Tris pH 7.6). The samples were further extracted with addition of 1.0 ml of CIA
(chloroform: isoamylalcohol; 24:1). Samples are centrifuged at 10,000 x g for 10 minutas at room temperature and 0.7 ml of the aqueous phase is transferred to an Eppendorf tube. DNA is precipitated at room temperature by addition of 0.07 ml sodium acetate, pH 6.0, and 0.7 ml 100~ ethanol. DNA is pelleted by centrifugation at 12,000 x g for 2 minutes at room temperature.
Ethanol is decanted and DNA pellets are washed with 1.0 ml 70%
ethanol and samples are centrifuged at 12,000 x g for 1 minute at room temperature. DNA pellets are dried in vacuum and resuspended in 0.05 ml TE (1 mM EDTA and 10 mM Tris pH 7.6). Samples are incubated at 55C for 5 minutes and then refrigerated overnight to rehydrate DNA. DNA concentrations were determined by reading absorbance at 260 nm in a spectrophotometer.
PCR analysis of tail-biopsy DNA was performed using two sets of oligonucleotides; one set (either oligonucleotides #11 and #12 or #40 and #41) which generates a 322 bp or 320 bp DNA fragment, respectively. These oligonucleotides amplify DNA sequences specifically from human APP minigenes. A second set of oligonucleotides (oligonucleotides #6 and #7) is included with each reaction which serves as an internal control for the PCR reaction and which amplifies a 154 bp DNA fragment from the mouse ribosomal protein L32 gene (Dudov and Perry, 1984, Cell 37: 457-468). The sequences of the oligonucleotides are as follows:
oligonucleotide #6:
5'-CCTCGGCCTTTGGTGTGTGTTTTATATGACATGACCCCCTTGA-3' oligonucleotide #7:
5'-CACCCCTGTTGTCAATGCCTCTGGGTTTCCGCCAGTTTCG-3' oligonucleotide #11:
5'-ATGMCTTCATATCCTGAGTCCATGTCGGMTTCT-3' , `

- 2 ~ 7 ~J

oligonucleotide #12:
5'-GGCAACATGATTAGTGAACCAAGG-3' oligonucleotide #40:
5'-GGAGGGTGCTCTGCTGGTCTTCAATTACC-3' oligonucleotide #41:
5'-AAGGGTTTGTCCAGGCATGCCTTCCTCATCC-3' The PCR reaction conditions are: 50 ~g/ml DNA, 5.0 ~g/ml of each oligonucleotide, 25 units/ml Taq polymerase (Cetus), 0.2 mM dATP, 0.2 mM dGTP, 0.2 mM dCTP, 0.2 mM TTP, 50 mM KCl, 1.5 mN MgCl2, 0.01~
gelatin, and 10 mM Tris pH 8.3. In many cases the oligonucleotides are end-labelled with 32p using polynucleotide kinase as described in Example 13. The specific activity of each ollgonucleotide is approximately 2 x 105 cpm/~g. The PCR reactions are performed in a Perkin Elmer DNA thermal cycler using the following reaction cycles (files): twenty-one cycles of 1 minute at 94C, then 2 minutes at 55C, then 3 minutes at 72C with an auto extension for sequence 3 of 10 seconds/cycle, followed by a cycle of 1 minute at 94C, then 2 minutes at 55C, then 12 minutes at 72C with an auto extension for sequence 3 of 10 seconds/cycle. The samples are then maintained at 18C until removal from thermal cycler. DNA fragments are separated by electrophoresis on a 5~ polyacrylamide gel and visualized by either staining with ethidium bromide or by radioautography.
Table II shows a number of APP minigene constructs useful for the preparation of transgenic mice. Table III shows a listing of representative APP transgenic founder mice generated according to the above-described methods. The transgenic founder mice are bred to establish permanent strains as described above. Table III also summarizes RNA and protein expression of APP minigenes in various transgenic mice as described in Examples 12, 13, 14 and 15.

; ' .

-58- 2~4~7 TABLE II
APP MINIGENE CONSTR~CTS
Promoter and Splicing and/
Genomic or Poly-Regulatory APP cDNA Adenylation Construct Elements Sequences Signals pMTI-2310 ~2.4kb HindIII (APP) APP-695 (pFC4) SV40 pMTI-2314 -4.6kb EcoRI (APP) APP-695 (pFC4) pNTI-2319 APP-770 (pFC4-770) pMTI-2320 APP-751 (pFC4-751) pMTI-2321 APP-695 +2 frame shift pMTI-2322 APP-695 - mutant pMTI-2325 APP-695 + Chlamvdia antigen pNTI-2318 A4 pMTI-2323 ~4.6kb ~çQRI (APP) APP-695 (pFC4) Mouse metallo-thionein pMTI-2331 APP-770 pMTI-2332 APP-751 pMTI-2324 APP-695 + Chlamvdia antigen pMTI-2326 APP-695 - mutant pMTI-2327 C-100 pMTI-2337 MC-100 pMTI-2341 A4 pMTI-2329 -2.2 kb EcoRI/BglII APP-695 ~pFC4) Mouse mouse metallothionein metallo-thionein pMTI-2333 APP-770 pMTI-2334 APP-751 pMTI-2335 APP-695 + Chlamvdia antigen pMTI-2336 APP-695 - mutant pMTI-2330 APP mutant 40 -alternative con-struct . ~ .. . . . .. .
.

204~7 Promoter and Splicin~ and Geno~ic Poly-Regulatory APP cDNA Adenylation Construct Elements Se~uences Si @als pMTI-2339 -4.6kb EcoRI (APP) APP-695 APP 3'-end pNTI-2342 APP-770 pMTI-2345 APP-751 pMTI-2343 Mutant 40-1 pMTI-2340 MC-100 10 pMTI-2344 sp-A4 - ~ , .
.:

:, .
:

, .

-60- 2~ 77 TABLE III
Transgenic ~ouse Strains with ~uman APP ~inigenes Strain Transgenic Gene Exp~ession (brain) Constructs Designation Founders RNA Protein s pMTI-2401 HB HB805 pMTI-2402 BE BE803 BE1805 . +
BE3002 . +
pMTI-2310 DH DH106 DHllO

pMTI-2314 ED ED106 ED1001 +
pMTI-2318 AE AE101 pMTI-2319 JE JE711 pMTI-2320 IE IE205 pMTI-2321 FE FE403 FE1001 +
pMTI-2322 GE GE106 - 2~00~7 Strain Transg~nic Gene ~xpression (brain) Constructs Designation Founders RNA Protein pMTI-2323 DM DM101 ++

DM406 +
DM606 +

15 pMTI-2329 DL DL110 pMTI-2331 JM JM201 pMTI-2344 SA SA110 pMTI-2343 FA FA105 FA201 +

pMTI-2340 CA CA507 CA507 +
CA1102 - .

pMTI-2342 JA JA407 ++ +
JA1301 ++ +

+ EXPRESSION
- EKPRESSION NOT WITH LIMITS OF DETECTION
. NOT DETERMINED

. ~.. ~- - . :

.
, , .
.
.. ,. .. ~ : .
, ' ', ., ~, .
' ., ~ :

.

~ r~

Tissue-Speclfic E~pression of APP Ninigene in Transgenic Nice Uslng the lacZ Reporter Gene The design of recombinant minigenes is a critical step in the generation of a transgenic mouse model for A4-amyloidosis. An essential element is a gene-regulatory region required for tissue-specific gene expression. Minigene constructs should exhibit expression patterns in transgenic animals which are consistent with the occurrence of amyloid in AD brains and preferentially resemble expression patterns of the endogenous mouse APP gene. For this purpose, the human APP gene regulato~y region was isolated as described in Example 1, and utilized for the construction of APP
minigenes. To monitor the tissue specificity of this human regulatory element in transgenic mice, a reporter gene, E. coli lacZ, encoding ~-galactosidase was utilized. This reporter gene allows for the convenient histochemical localization of protein expression regulated by the human APP gene regulatory region. Using this reporter gene, the 5'-end sequences of the human APP gene were demonstrated to contain sufficient information to target expression in the CNS of transgenic animals with patterns consistent with endogenous ~ouse APP gene expression as iollows.
The minigene pMTI-2402 (Figure 20) was constructed by fusing the -4.6 kb EçQRI fragment tescribed in Example 1 comprising the human APP gene regulatory region including the APP promoter, with the lacZ reporter gene in the following steps. First, the cloning vector pMTI-2301 was prepared. Plasmid pMTI-2301 contains a unique HindIII cloning site, flanked by ~I restriction sites, and was constructed as described in Example 4. Second, the ~4.6 Xb EcoRI
fragment isolated from the chromosome 21 library as described in Example 1 encompassing the 5'-end of the human APP gene was inserted by blunt-end ligation into the HindIII site of pMTI-2301 to generate pMTI-2307. Finally, minigene pMTI-2402 containing the lacZ gene was constructed by inserting an -3.9 kb HindIII-BamHI fragment from plasmid pCH126, containing the lacZ fusion protein and SV40 polyadenylation signal, into the ~I site of pMTI-2307 by blunt-end ligation. Plasmid pCH126 i9 identical to plasmid pCHllO
described by Hall et al., 1983, J. Mol. Appl. Gen. 2: 101-109, except `
.
- -that the SV40 promoter tthe uII-_i~dIII fragment. See Figure 1 in Hall et al., 1983, J. Mol. Appl. Gen. 2: 101-109.) has been deleted but the HindIII site remains. (Goring et al., 1987, Science 235:
456-458).
Plasmid pNTI-2402 DNA was double-purified in CsCl/ethidium bromide equilibrium density gradients. The -8.6 kb linear DNA
fragment, encompassing the APP/lacZ reporter gene, was excised from vector sequences using NotI and isolated from an agarose gel using NA45 paper (Schleicher and Schuell, ~eene, NH). The DNA was precipitated in ethanol-ammonium acetate three times and resuspended, at a concentration of 6 ~g/ml, in filtered (0.2 ~ membrane) in~ection buffer (10 mM Tris, pH 7.5, and 0.3 mM EDTA; Brinster et al., 1985, Proc. Natl. Acad. Sci. USA 82: 4438-4442). Purified DNA fragments were microinjected into l-cell embryos of Hsd:(ICR)BR female mice and B6D2Fl/Hsd BR male mice and reimplanted into Hsd:(ICR)BR female mice as described (DePamphilis et al, 1988, BioTechniques 6: 662-680).
Transgenic founder mice were identified by PCR analysis of tail biopsy DNA using 30 bp oligonucleotides #15 and #16, complementary to the E. coli lacZ gene and internal control oligonucleotides #6 and 20 #7 (described in Example 11). The sequence of oligonucleotides #15 and #16 are as follows:
#15 S'-CCTGGCGTTACCCM CTTM TCGCCTTGCAGCACAT-3' #16 5'-AATAAATGTGAGCGAGTAACAACCCGTCGGATTCT-3' DNA from transgenic mice was further analyzed by restriction enzyme 2~ digestion and Southern-blot analysis (Wirak et al., 1985, Mol. Cell.
Bio. 5: 2924-2935).
A. In situ bvbridization In ~ hybridization techniques were used to establish the cellular distribution of APP mRNA in normal mice. The distribution of APP mRNA within the central nervous system (CNS) of other species (humans, primates, rats) has been previously determined with the ma~ority of the CNS APP mRNA localized to neuronal cytoplasm. For these experiments, four-to-five-week-old mice were anesthetized and perfused with 4~ paraformaldehyde in 0.08 N phosphate buffer, pH
7.6. The following tissues were removed and processed to paraffin by standard procedures: cerebral hemispheres plus diencephalon, ... ., .. - , . . , :
: ,, ' ';

- .

.
' ' ' ~ ;.

' pons, medulla, cervical and lumbar spinal cord, trigeminal nerve, and liver. All tissues from individual mice were embedded in a single block. Sections were cut at a thickness of 6 ~m and hybridized according to published procedures tTrapp et al., 1987, Proc. Natl. Acad. Sci. USA 84:7773-7777; Trapp et al., 1988, J. Neuro Sci. 8: 3515-3521). Briefly, pre-hybridization treatment consisted of 0.2 N HCl for 20 minutes and 25 ~g/ml protease K for 15 minutes at 37C. Slides were hybridized at room temperature for 16 hours in a standard buffer containing 0.2 ng/~l of single-stranded, full-length human APP cDNA, labeled with 35S by the Klenow procedure (specific activity, 2.3 x 109 cpm/~g). Stringency washes included 50~ formamide containing 0.3 M NaCl, 1 mM EDTA, and S mM Tris (pH
8.0) for 30 minutes at room temperature and 2 X SSC (1 X SSC - 0.3 M NaCl, 0.03 ~ sodium citrate, pH 7.4) in 1 mM EDTA for one hour at 55C. Slides were then dehydrated, air dried, dipped in emulsion (Kodak, NTB-3), exposed for 7 days, developed for autoradiography and counter-stained with hematoxylin. Sections were photographed with a Zeiss Axiophot using dark-field and bright-field optics.
Specific brain regions and neuronal subpopulations were identified according to published criteria (Sidman et al., 1971, ~tlas of the Mou~e ~rain and Spinal Cord, Harvard University Presg, Cambridge, MA). Silver grains, representing APP mRNA, occurred in clusters that reflected the general distribution of neurons in all brain regions studied (Figure 21). For example, neuronal perikarya present in the pyramidal layer of the hippocampus and granular layer of the dentate gyrus were labeled intensely (Figure 21a). Significantly less hybridization signal was detected in other layers of the hippocampal formation in subcortical white matter. The cerebral cortex contained significant levels of APP mRNA (Figure 21a); and the labeling pattern in ~arious cortical areas (i.e., occipital, temporal, and frontal) reflected the distribution of neuronal perikarya in the various layers. Layer I, which contains few neurons, had the lowest hybridization signal in all cortical regions. In sections of cerebellar cortex, Purkin~e and granular cells were labeled by APP
cDNA (Figure 21b). Sections of trigeminal ganglia (from peripheral nervous system) were hybridized with APP cDNA and the neuronal 2~877 perikarya, which occur in clusters, were labeled intensely but little hybridization signal was found in myelinated fiber tracts in the PNS
(Figure 21c). APP mRNA was not detected in sections of liver (Figure 21d), a finding consistent with Northern-blot analysis (Yamada et al., 1989, Biochem. Biophys. Res. Commun. 158: 906-912). The distribution of silver grains was concentrated within perinuclear cytoplasm of neurcns (Figure 21e), Few silver grains were present over neuronal nuclei or scattered throughout the neurophil.
B. Histochemical ~etection of ~-Galactosidase For the light microscopic histochemical detection of ~-galactosidase in the transgenic mice carrying the APP/lacZ reporter gene, transgenic mice and normal mice as controls, four-to-five-weeks of age, were anesthetized and perfused with 4~ paraformaldehyde and 0.08 N phosphate buffer, pH 7.6. The CNS, trigeminal nerve, and liver were removed and placed in the fixati~e overnight at 4C.
These tissues were cut into 0.5 cm thick slices that were either stained histochemically for ~-galactosidase or sectioned at a thickness of 20 ~m on a vibrating microtome prior to staining.
~-galactosidase activity was detected histochemically by incubating the tissue in a reaction buffer [2.7 mM KH2PO4, 8.0 m~
Na2HPO47H20, pH 7.6, 2.7 mM KCl, 140 mM NaCl, 2 mM MgCl2, 22.5 mM
K~Fe(CN)5, 25.0 mM K3Fe(CN)5, 0.27 mg/ml sodium spermidine, 0.5 mg/ml X-gal (20 mg/ml stock in diethylformamide), 0.02% NP-40, and 0.01 sodium deoxycholate] that was maintained at 30C for 18 to 24 hours.
Vibratome sections were infiltrated with 100~ glycerol, mounted on glass slides, and then photographed with a Zeiss Axiophot microscope using bright-field or Nomarski optics.
The tissue and cellular expression pattern of an APP promoter-lacZ reporter gene in adult transgenic mice as determined by the above-described histochemical method was strikingly similar to the distribution of endogenous mouse and endogenous human APP mRNA.
Minigene pMTI-2402, for the expression of the reporter gene in transgenic mice, was constructed as described above by inserting sequences encoding a lacZ fusion protein and SV40 polyadenylation signals into an -4.5 kb genomic fragment encompassing the 5'-end of the human APP gene. The genomic fragment contains 2831 bp of ~ . . . ` .
.

~ .

2 ~

sequences 5' to the primary transcriptional start site, exon I, and approximately 1.6 kb of the first intron. Three lines of transgenic mice were identified which carried multiple head-to-tail integrations of the intact reporter gene (Table III). Tissue distribution analysis of the three lines showed that one line, BE803, exhibited intense ~-galactosidase expression throughout the CNS, while two lines (BE1805 and 8E3002) exhibited lower levels of expression.
In adult ~E803 mouse brain, staining was concentrated in regions having high concentrations of neuronal perikarya (Figures 22 and 23a-c). Thus, cerebral cortex, dentate gyrus, basal ganglion, thalamus, and regions of the hippocampus were stained intensely.
Prominent white matter tracts such as the corpus callosum and internal and external capsule were not stained. Staining of brain stem and spinal cord tissue was observed in a pattern similar to endogenous mouse APP mRNA. ~-galactosidase was not detected in slices of normal mouse brain, used as a control (Figure 23d).
Regions of cerebellar cortex that contain high concentrations of neuronal perikarya were positive for ~-galactosidase (Figures 24a and 24b) as were neuronal perikarya in the trigeminal ganglion.
White matter tracts in the cerebellum and trigeminal nerve (Figure 24c), and slices of liver (Figure 24d) from BE803 mice were negative for ~-galactosidase. Identical ~-galactosidase staining patterns were observed in tissue slices from several BE803 transgenic mice.
The cellular and subcellular distribution of ~-galactosidase was determined in several brain regions by light microscopic procedures.
~-galactosidase was localized histochemically in 20-~m-thick vibratome sections. In these sections, ~-galactosidase reaction product occurred as small dots that were restricted to regions of the CNS that contained neuronal perikarya. Reaction product was detected in all layers of the cerebral cortex (Figure 25a), including occasional deposits in Layer I. ~hen examined at higher magnification with Nomarski optics, ~-galactosidase reaction product was restricted to perinuclear regions of neurons (Figures 25b and 25c). ~-galactosidase was not detectable in endothelial cells and cellular perikarya within white matter tracts. In the cerebellar cortex, ~-galactosidase was localized in perinuclear regions of .
; ~ , . , . . . .: ~ , .

~4~

Purkin~e cells (Figure 24b). Analysis of vibratome sections also detected the presence of ~-galactosidase in regions of the CNS that were not labeled intensely in the brain ~lices. For example, consistent but weak staining of ~ome but not all neurons in CA-3 region of the hippocampus was found.
C. Immunocvtochemical Detection of ~-Galactosidase For the electron microqcope (~H) i,,unocytochemical detection of ~-galactosidase, transgenic mics, between four-to~five-weeks of age, were perfused with 2.5% glutaraldehyde and 4~ paraformaldehyde in 0.08 M phosphate buffer. The brains were removed and placed in the fixative overnight at 4C. Segments of the cerebral cortex (2 mm2) were infiltrated with 2.3 N sucrose and 30% polyvinyl pyrrolidon, placed on specimen stubs, and frozen in liquid nitrogen.
Ultrathin frozen sections (-120 nm-thick) were cut in a Reichert Ultracut-FC4 ultracryomicrotome maintained at approximately -110C.
The sections were transferred to formvar and carbon-coated hexagonal mesh grids and stained by immunogold procedures using a modification of standard procedures. Following immunostaining, the grids were placed in PBS containing 2.5% glutaraldehyde for 15 minutes and rinsed. The sections were stained with neutral uranyl acetate followed by embedding in 1.3~ methylcellulose containing 0.3% uranyl acetate. Grids were examined in a Hitachi H600 electron microscope.
The cellular and subcellular distribution of ~-galactosidase in the cerebral cortex and other brain regions as determined by the immunogold procedure revealed that the majority of gold particles in electron micrographs was localized to the perinuclear cytoplasm of neurons (Figure 25d). Glial cells and endothelial cells were not labeled.
The striking conclusion from the ~n situ hybridization, light microscopic and electron microscopic detection of mouse APP and ~-galactosidase was that the -4.5 kb genomic fragment encompassing the 5'-end of the human APP gene isolated as described in Example 1 had sufficient sequence information to direct cell- and tissue-specific expression of a reporter gene, E. coli lacZ, in transgenic mice. The expression pattern of the reporter gene in the C~S was strikingly consistent with the expression pattern of the endogenous mouse APP

.........

2 ~

gene. This -4.5 kb genomic fragment which includes the APP promoter and perhaps other regulatory elements was incorporated in nearly all APP minigene constructs described in Examples 4-10 above. Such constructs are particularly useful in the preparation of transgenic mice as described in Example 11. The identification of such an appropriate gene promoter and other regulatory elements for minigene constructs is a critical step for the development of transgenic mouse models for AD, since AD pathology is restricted to specific regions of the brain [Price, 1986, Annu. Rev. Neurosci. 9: 489-512). The ~4.5 kb genomic fragment described and characterized herein is the type of regulatory element that must be utilized to facilitate the expression of recombinant APP genes with a cell and tissue specificity that is consistent with the formation of amyloid plaque and the expression patterns of the APP gene.

Espression of Human APP ~RNA in Transgenic Mice Several transgenic mouse lines express human APP mRNA in brain tissue (Figures 26, 27, and 46). Expression of human APP mRNA in transgenic animals was determined by nuclease Sl protection analysis (Figures 26 and 27) and by riboprobe analysis (Figure 46).
A. Nuclease Sl Protection Analysis Sl nuclease digests single-stranded DNA and RNA but not double-stranded species. Therefore, specific 32P-labeled oligonucleotides that hybridize with complementary mRNA sequences, are protected from digestions by Sl nuclease and can be identified by denaturing polyacrylamide gel electrophoresis (PAGE). A human specific oligonucleotide (designated oligonucleotide #29 and described below) will produce an approximately 70 bp-protected fragment in an Sl digestion when annealed to human APP mRNA. A mouse-specific oligonucleotide (designated oligonucleotide #30 and described below) will produce an approximately 50 bp-protected fragment when annealed to mouse APP mRNA in an Sl assay. RNA from the human cell line, Hela (A.T.C.C. No. CCL2) was used for a positive control for human APP RNA
(Hela cells express APP; Weidemann et al., 1989, Cell 57: 115-126) , 0 7 ~

and RNA from a control non-transgenic mouse was used for a negative control in the assay.
Oligonucleotides complementary to eitherhuman (oligonucleotide #29) or mouse (oligonucleotide #30) APP mRNA sequences were synthesized using an automated Applied Biosystems oligonucleotide synthesizer (model 380A). Oligonucleotides were generated using reagents and protocols provided by tbe manufacturer. The sequence of oligonucleotide #29 is:
5'-GAGATAGMTACATTACTGATGTGTGGATTMTTCAAGTTCAGGCATCTACTTGTGTTACA
GCACAGCTGGGCGTCCATA-3' This 80 bp oligonucleotide contains a 10 bp non-homologous sequence domain at the 3'-end so that, after Sl digestion, the protected oligonucleotide fragment (approximately 70 bp) can be distinguished from non-hybridized oligonucleotide probe. The actual size of the protected fragment(s) can only be determined by experimentation because specific single- and double-stranded nucleotide sequences exhibit variability in their sensitivity to Sl. The sequence of oligonucleotide #30 is:
S'-CGCGGGTGGGGCTTAGTTCTGCATTTGCTCAAAGM CTTGTAAGTTGGATAGGTTCCMG-3' This 60 bp oligonucleotide contains a 10 bp non~homologous sequence domain at the 3'-end so that, after Sl digestion, the protected oligonucleotide fragment (approximately 50 bp) can be distinguished from non-hybridized oligonucleotide probe.
The 5'-end of each oligonucleotide was labeled with 32p using T4 polynucleotide kinase and ~gamma-32P]dATP. The reaction conditions were as follows: 200 ng oligonucleotide,-l ~1 (10,000 units/ml) polynucleotide kinase (NEB), 1.0 mCi[gamma-3ZP]dATP (3000 Ci/nmole; Amersham PB15068), LX kinase buffer (Maniatis et al., supra), and incubation at 37C for 45 minutes. Unincorporated nucleotide was removed by gel-filtration (Sephadex G-50). The specific activity of each probe was: oligonucleotide #29, 6.04 x 108 cpm/~g; oligonucleotide #30, 5.72 x lOa cpm/~g.
RNA was extracted from mouse brain and Hela cell pellets using a procedure described in Basic Methods in Moleçula~ Biolog~ (Davis et al., 1986, Elsevier, New York, Amsterdam, and London; pp. 130-135).

2~ r~ r~

Total RNA, 50 ~g/sample, was mixed with 1 x 106 cpm of each 32p labelled oligonucleotide (oligonucleotide #29 and oligonucleotide #30) and then dried in vacuum. The RNA/oligonucleotide pellet was resuspended in 20 ~l of Hybridization buffer (l mM EDTA, 0.4M NaCl, 50~ formamide, and 40 mM Pipes pH 6.4). Hybridization was performed in a Perkin Elmer Cetus DNA Temperature Cycler (model #PCR-10000).
Samples were incubated at 90C for 10 minutes and then at 70C for 20 minutes. The temperature was then lowered 1C every 18 minutes until the temperature reached 30C. The reaction was terminated by placing samples on ice. Sl nuclease digestion was initiated by addition of 300 ~1 of Sl reaction buffer (0.2 M NaCl, 5 mM ZnCl2, 30 mM sodium acetate pH 4.5, and 400 units Sl) and samples were incubated at 20C for 2 hours. Sl reaction was terminated by adding EDTA to a final concentration of 25 mM. Samples were extracted with equal volumes of phenol and then phenol/chloroform/isoamylalcohol (24/24/1). The oligonucleotides in each sample were precipitated at -70C for 1 hour by addition of 10 ~g tRNA, 175 ~l of 7.5M NH4-acetate, and 875 ~l of absolute ethanol. The oligonucleotides were resuspended in lO ~l of 10 mM Tris and l mM EDTA, pH 7.6. Samples were denatured by addition of 10 ~1 of 2X Sequencing loading 8uffer (from USB) and incubation at 90C for 3 minutes. Samples were then transferred to ice and then loaded onto a 10% denaturing polyacrylamide gel (lX TBE and 7M urea) that had been prerun for 20 minutes at 1600V, constant voltage. The samples were electropXoresed at 1600 V for approximately one hour. The gel was dried and the migration of the oligonucleotides was detected by autoradiography using Kodak X-ray film.
Fig~re 26 demonstrates that transgenic lines AE301, AE101, FE801, FE403, ED1001, ED801, and DHl06 express human APP RNA in brain (i.e., have an approximately 70 bp-protected fragment after Sl digestion). The intensity of the approximately 70 bp band of the protected fragment in these samples was greater than the background observed in control mouse brain RNA (lane 3). The level of human-specific expression, however, is low compared to the endogenous mouse APP expression level. For size markers: gel lane 1 contains oligonucleotides 29 and 30 (9.7 x 102 and 6.1 x 102 cpm respectively) . , . . :

' 2 ~

and lane 2 contains a 1 bp DNA sequencing ladder. Both oligonucleotides 29 and 30 were annealed to 50 ~g of brain RNAs and samples were digested with Sl nuclease as described below. The gel contains the following ~NA samples: lane 3, mouse normal brain;
lane 4, Hela cell; lane 5, AE301 brain; lane 6, AE302 brain; lane 7, AE601 brain; lane 8, AE101 brain; lane 9, FE801 brain; lane 10, FE403 brain; lane 11, ED1001 brain; lane 12, ED106 brain; lane 13, ED801 brain; lane 14, JE711 brain; lane 15, JE1005 brain; lane 16, DH106 brain; lane 17, GE107 brain.
Figure 27 demonstrates that transgenic lines IE504, IE801, IE301, IE606, IE206, DM101, DM405, DM406, and DM606 express human APP RNA in brain (i.e., have an approximately 70 bp-protected fragment after Sl digestion). The intensity of the approximately 70 bp band of the protected fragment in these samples was significantly greater than the background observed in control mouse brain RNA (lane 3). The level of human-specific expression, however, is low compared to the endogenous mouse APP expression level. For size markers: gel lane 1 contains oligonucleotides 29 and 30 (9.7 x 102 ar.d 6.1 x 102 cpm respectively) and lane 2 contains a 1 bp DNA sequencing ladder.
Both oligonucleotides 29 and 30 were annealed to 50 ~g of brain RNAs and samples were digested with Sl nuclease as described below. The gel contains the following RNA samples: lane 3, normal mouse brain;
lane 4, Hela cell; lane 5, IE602 brain; lane 6, IE504 brain; lane 7, IE801 brain; lane 8, IE301 brain; lane 9, IE205 brain; lane 10, IE606 brain; lane 11, IE206 brain; lane 12, IE505 brain; lane 13, IE803 brain; lane 14, DM101 brain; lane 15, DM309 brain; lane 16, DM405 brain; lane 17, DM406 brain; lane 18, DM606 brain.
B. Ribo~robe Analvsis RNase A and RNase Tl digest single-stranded RNA but not double-stranded RNA species. Therefore, specific riboprobes (32P-labelled anti-sense RNA) that hybridize with complementary mRNA sequences, are protected from digestion by a cocktail of RNase A and RNase Tl and can be identified by denaturing polyacrylamide gel electrophoresis (PAGE). The Bluescript M13 phagemid (Stratagene, San Diego, CA) contains a multiple restriction enzyme polylinker flanked by promoters for T7 and T3 RNA polymerase. The promoters are positioned .,, ... ~ .

' ' 2 ~ 7 ~

in opposite orientations and can be utilized to transcribe 32p_ labelled anti-sense RNA probes specific to any sequence inserted into the polyl$nker region. Clone pMTI-2371 (see Example 16, part B, and Figure 41) contains the human APP sequences encoding the MC-100 gene product (gene product V, Figure 4b; see also Figure 12 and Example 9, part A, section 3) inserted into Bluescript KS+. A riboprobe which specifically hybridizes to human APP mRNA was generated using T7 RNA
polymerase and linearized pMTI-2371 (phagemid digested with HincII) as template. The riboprobe was -408 bp in length and the portion complementary to human APP was -373 bp. Therefore, RNase A/RNase Tl digestion of the riboprobe, which has been hybridized with human APP
mRNA, would generate an -373 bp-protected fragment. RNase A/RNase Tl digestion of riboprobe, which has been hybridized with mouse APP
mRNA, would result in numerous fragments which are considerably 15 smaller than 373 bp. The template was prepared and the 32p labelled riboprobe was generated (using 60 ~Ci of 32P-rUTP [sp. act.: 800 mCi/mmol] obtained from Amersham (Arlington Heights, IL). RNA was prepared from the Hela cell line, the brain of a normal mouse, and the brains of individuals from the following lines of transgenic 20 mice: AE101, AE301, CA507, FA201, FE1001, FE403, IE801, JA407, JA1301, SAllO, SA602, and SA706 using methods de~cribed in Example 13, part A. RNA samples (20 ~g) were precipitated with 1/10 volume, 3 M sodium acetate pH 5.2, and 2.5 volumes ethanol. Each RNA sample was resuspended in 20 ~1 of lX hybridization buffer (80% formamide, 25 40 mM PIPES pH 6.4, 0.4 M NaCl, and 1 mM EDTA) and 10 ~1 of riboprobe (2 x 105 cpm in lX hybridization buffer). Samples were incubated at 85C for 10 minutes and then incubated at 45C overnight. The RNA
samples were digested by addition of 350 ~1 of ribonuclease buffer ~10 mM Tris pH 7.5, 300 mM NaCl, and 5 mM EDTA) with 40 ~g/ml RNase 30 A and 2 ~g/ml RNase Tl and incubation at 30C for 60 minutes. To each sample was added 20 ~1 of 10~ SDS and 2.5 ~1 of 20 mg/ml proteinase K. Samples were incubated at 37C for 15 minutes and then extracted with phenol/isoamylalcohol/chloroform. The samples were precipitated by addition of lO ~g of tRNA and 1 ml of ethanol.
Samples were resuspended and electrophoresed on a denaturing polyacrylamide/urea gel as described in Example 13, part A. The gel 2 ~

represented in Figure 46 contains the following RNA samples: lane 1, Hela cell RNA; lane 2, normal mouse; lane 3, AE301; lane 4, AE301;
lane 5, AElOl; lane 6, CA507; lane 7, FA201; lane 8, FE1001; lane 9, FE403; lane 10, IE801; lane 11, JA407; lane 12, JA1301; lane 13, SAllO; lane 14, SA602; lane 15, SA706; lane 16, blank; lane 17, riboprobe (undigested); and lane 18, riboprobe (undigested). The protected riboprobe fragments were detected by autoradiography as shown in Figure 46. The experiment demonstrated that the following transgenic mouse lines express human APP RNA: AE101, AE301, CA507, FE1001, IE801, JA407, JA1301, SA602, and SA706 (see Table III).

Exprossion o~ Human APP and APP Derivatives in ~ransgenic Mice A. Ex~ression of APP-751 Transgenic mouse line IE801 (see Table III) expresses human APP-751 protein in the brain (Figure 28a and 28b). Human APP-751 expression (Figure 28b) was detected in protein extracts of transgenic mouse brain by Western-blot analysis using the human-specific monoclonal antibody (mAb), mAb 56-1 (see Example 17).
Western-blots of protein extracts from transgenic mouse brains were also stainet using mAb 22C-ll which reacts with APP-695, APP-751 and APP-770 from both human and mouse (Figure 28a). The monoclonal antibody, mAb 22C-ll, was a gift from Dr. Beyruether (Weidemann et al., 1989, Cell 57: 115-126).
Figure 28a contains the following samples: lane 1, low molecular weight protein markers; lane 2, DH106 brain lysate; lane 3, DM606 brain lysate; lane 4, JE711 brain lysate; lane 5, IE508 brain lysate; lane 6, IE801 brain lysate; lane 7, IE301 brain lysate;
lane 8, normal mouse (ICR strain) brain lysate; lane 9, media from cell line, cMTI-53; and lane 10, high molecular weight protein markers.
Figure 28b contains the following samples: lane 1, high molecular weight protein markers; lane, 2, cell line cMTI-53; lane 3, normal mouse (ICR) brain lysate; lane 4, IE301 brain lysate; lane 5, IE801 brain lysate; lane 6, IE508 brain lysate; lane 7, JE711 S~ P~r~

brain lysate; lane 8, DM606 brain lysate; lane 9, DH106 brain lysate;
and lane 10, low molecular weight protein markers.
Figure 28a demonstrates that each brain extract contains approximately equal amounts of APP protein and that APP-695 is the predominant form of APP in ~ouse brain extracts. The extracellular forms of APP-695 and APP-751 (or 770) have apparent molecular weights of ~93-105 kDa and -112-125 kDa respecti~ely (Weidemann et al., 1989, suDra and Palmert et al., 1989, Proc. Natl. Acad. Sci. USA 86: 6338-6342) Protein from the culture ~edia of a mouse cell line (line 10 cMTI-53; see Example 16) which secretes human APP-751 was included as control (Figure 28b, lane 2). We could not determine whether transgenic mouse lines DM101 or DH106 expressed human APP-695 because of the cross-reactivity of mAb 22C-ll for mouse and human APP-695.
Figure 28b demonstrates that transgenic mouse line IE801 (lane 5) expresses a protein which reacts with mAb 56-1 and has a gel migration mobility equal to that of APP-751 secreted by the cell line cMTI-53 (lane 2). A non-transgenic mouse (lane 3) or transgenic mice carrying minigenes encoding human APP-695 (DMlOl and DH106) do not exhibit immunostaining of this protein. Transgenic mouse line 20 IE508 al90 expres~ed cross-reactive proteins species. However, the migration of the proteins doeq not correspond to human APP-751. It is possible that human APP-751 is anomalously expressed or metabolized in the IE301 line and no APP-770 expression was observed in the JE711 line.
Protein was extracted from the brain of a non-transgenic control mouse (IC~ strain) and the brains of transgenic animals from the following lines: DH106, DM606, JE711, IE508, and IE301. Whole brains were dissected from the animals and weighed to estimate tissue volume. Two volumes of lysis buffer (0.2N ~aCl, 1% Triton X-100, 2 30 m~ PMSF (Sigma #P-7626), 1 mM DFP, LX protease inhibitor solution, 10 mM Tris pH 8.0) was added to each brain. Protease inhibitor solution, lOOX, consisted of: 1 mg/ml leupeptin (Sigman #L-2884), 1 mg/ml pepstatin-A (Sigman #P-4265), 10 TIU/ml aprotinin (Sigma 3A-6012), 0.1 mN EDTA, and 0.2N Tris pH 8Ø Brain tissue was then 35 homogenized for -1 minute with a Polytron homogenizer (model CH6010).
Each sample was centrifuged at 10,000 x g at 4C for 30 minutes and ~ 7 the supernatant (lipid layer removed), or "brain lysate," was stored at -70C. Protein in culture media for cell line cMTI-53 was concentrated by acid precipitation. Approximately 1.5 ml of culture media, ice cold, was harvested and a 1.5 ml aliquot of 25~
trichloroacetic acid (TCA), ice cold, was added. Samples were centrifuged at 15,000 x g for 10 minutes at room temperature. The protein pellets were washed three times with 100% acetone and then centrifuged after each wash at 15,000 x g for 10 minutes at room temperature. The pellets were dried in a vacuum for -20 seconds, 10 resuspended in 100 ~1 of NRS8 buffer (2~ SDS, 5~ betamercaptoethanol, 5% loading dye, 10% glycerol, 0.125 M Tris pH 6.8) and boiled for 5 minutes.
The "brain lysate" proteins and cMTI-53 cell supernatants were fractionated by polyacrylamide gel electrophoresis (10% running gel and 4~ stacking gel) and transferred to Immobilon-P membrane by the technique of electroblotting using a Biorad Mini-Protean II apparatus and using procedures recommended by the manufacturer. Prior to electrophoresis, 10 ~1 human APP-751 control cell supernatant (cell line cMTI-53), 1 ~1 of control mouse brain lysate, and 2 ~1 aliquots o transgenic mouse brain lysates were denatured by addition of LX
NRSB and boiling for 5 minutes. Each gel also included pre-stained high and low molecular weight standards (BRL catalog #6041LA and #6040SA, respectively).
Human and mouse APP proteins, transferred from the polyacrylamide gels onto Immobilon-P membrane, were detected by Western-blot staining. Mouse and human APP-695, APP-751 and APP-770 proteins were detected using mAb 22C-ll. Human APP-751 was detected using mAb 56-1; this antibody does not recognize mouse APP-751. After protein transfer, the Immobilon membranes (12 x 12 cm) 30 were incubated with 50 ml lX blocking buffer (0.15M NaCl, 5~ non-fat dry milk, and 10 mM Tris pH 8.0) for one hour at room temperature. Membranes were then stained with 10 ml of "first"
antibody solution (22C~ 1:10,000 dilution of mAb stock into blocking; or mAb 56-1: 1:100 dilution of mAb stock into blocking buffer) for 2 hours at room temperature. Membranes are next washed with blockin~ buffer and then stained with 15 ml of "second" antibody .

solution [goat anti-mouse IgG conjugated with alkaline phosphatase (Promega): 1:7500 dilution of antibody into blocking buffer] for 30 minutes at room temperature. Membranes are then washed with blocking buffer and then with AP buffer (O.lM NaCl, 5 mM MgCl2, O.lM Tris pH

9.5). Membranes are next stained with "AP substrate~ solution (15 ml AP buffer, 99 ~1 NBT stock solution, and 49 ~1 BCIP stock solution) for one hour at room temperature. NBT stock solution consists of 50 mg/ml nitro blue tetrazolium (Sigma #N-6876) in 70 dimethylformamide and BCIP stock solution consists of 50 mg/ml 5-bromo-4-chloro-3-indolyl phosphate in 100% dimethylformamide. The AP staining reaction was determined by washing membranes in deionized water.
B. Expression of A4 APP Peptide Transgenic mouse line AE301 (see Table III) carries minigene pMTI-2318 (gene product VIII, Figure 4b), which encodes the 42 amino acid A4 peptide of APP (see Example 7 above). This line of transgenic mice has been shown to express APP RNA in brain (see Example 13 above). In further studies, it was shown that AE301 transgenic mice exhibit A4 aggregates in the hippoca~pus region of the brain. This transgenic line can be used to examine the neurotoxicity of the A4 peptide in brain tissue. In addition, the A4 aggregates present in the transgenic mice may represent an early stage of senile plaque formation. These transgenic mice can serve, therefore, as a model for early pathological events occurring in patients affected with AD. Aggregation of A4 peptide was demonstrated by several methods, including immunocytochemical analysis and electron microscopic (EM) analysis.
1. Immunocvtochemical Analvsis of A4 Aggregates Rabbit polyclonal antibodies (pAb) 90-25, 90-28 and 90-29, used for the immunocytochemical analysis, were generated by standard methods. Subcutaneous injections of the A4 peptide (amino acid residues 1 to 28 for pAb 90-25, and amino acids 1 to 42 for pAbs 90-28 and 90-29) were administered to rabbits using Freund's ad~uvant. Rabbit sera were screened for immunoreactivity to the A4 peptide and several, including pAb 90-25, 90-28 and 90-29, tested positive. These positive antibodies were further characterized by .

2 ~ 7 ~

reaction with pathological human brain tissue from a patient with AD.
The pAb 90-25, 90-28 and 90-29 immunostain A4 amyloid plaques (senile plaques) found in the pathological tissues.
Once the specificity of pAb 90-25, 90-28 and 90-29 with A4 peptide had been established, these antibodies were used to immunostain cross-sections of brain tissue from uice. Light microscopic immunochemistry was performed using paraffin tissue sections, according to the method of Trapp et al., 1983, J.
Neurochem. 40: 47-54. The results showed immunostaining of specific areas of the hippocampus region of the brain from an AE301 transgenic mouse as shown in Figures 34, 35, and 36. The transgenic mouse, designated AE301+207 (Fl), used for this immunocytochemical analysis was a transgenic progeny of a mating between AE301(FO), the founder mouse, and a non-transgenic female (IC2200). Transgenic progeny of this mating were identified by PCR analysis as described in Example 11 above.
Figure 34 illustrates a cross-section of brain from mouse AE301+207(Fl) immunostained with pAb 90-29. A4 immunoreactive regions can be observed as dark-brown areas, are punctate in nature, and appear in clusters in the hippocampus (Figure 34, representative immunostained clusters are highlighted with arrows). Figure 35 is a higher magnification of the hippocampal region of mouse AE301+207(Fl) brain tissue stained with pAb 90-29 (representative immunoreactive regions are highlighted with arrows). Similar 25 immunostaining in the hippocampus of AE301+207(Fl) brain tissue can be observed with a second A4 immunoreactive antibody, pAb 90-28 (Figure 36, representative immunoreactive regions are highlighted with arrows). A third A4 immunoreactive antibody, pAb 90-25, also showed similar immunostaining.
This immunostaining was specific to the AE301 transgenic line because an age-matched mouse, designated FE803+105(Fl), from transgenic line FE803 which carries pNTI-2321 (see Example 6 and Table II) does not exhibit immunostaining with pAb 90-29 in the hippocampus or in other regions of the cross-section of brain (Figure 37)-2. ~l~ctron Sr~n~lssion elact~on mlcro~copic ~n~ly8~s of ~h~n oec~lon~ o~
~lX~d ~nd ot~ln~d braln t~U4 ~a~ par~onse~ ~cco~dln~ to thQ ~ethod ~ app ~t ~1., 1982, J. Nouroacl. ~: 986^993. The tr-n~gon~c ~a4 5 u~ed ~r th~ B alactron ~croocopic cnalyoi3 W~ ~esigna~a~
A~301+201~P2) ~nd ~s the prog~ny of a mat~n~ betwesn ~301l210~F1) and ~301~207(~ 301+210~ nd AE301l207~Fl) are ~rageny a~
~at~ng ~two~n ~E301~(F0) ~nd B non-tr-n~enic fe~A1a (IC~200).
Th~e tra~e6n~c p~ogony wara ~tentifled by PC~ n~ly~ de~cribed lo ~n ~x~plo 11 ~bove.

~h~ r~ult~ chowet ~ctron~d~nJc a~r4~At~ ~n ~p~c~c ~re~e of ~hfl hl~ocamp~l reg~on o~ the br~n from thle tran~nIc mouse.

T~o ol~ctron.d4noe aggrqgate~ w~re found l~ tho oa~e br~n re~iono whic~ exh~blt~d Sm~unochem~cAl ~talnlng with pA~ 90-2~, 90-28 and 15 90-29. Sh~ ~6sr4g~te~ ~ppe~r to bo loc~ed ~Ithin tho IntrAcellular ~pcce of ne~ron dendr~tos. ~U~9 38a ~nd 38b Lllu~tr~te ~l~ctron-donoo aggr~g~tc~ ~n thln ~ect~o~ o~ hlppocu~p~l br~n tlcoue l~ol~t-t from tr~n~on~c mow e AR30l+20l(F2). Sho bord-~ of the ~l~ceron-den~4 ~ggreg~to3 aro h~thl~ht-d wich ~rrow~.

~0 Th~t ~loceYon-d-n~Q r~glonJ r~ ~ggrogat~ o~ the A4 yoptlde w~ domon~tr~ted ~lnco im~unor~qtivlty wlth pAb 90-29 co-loc~ ed with tho elactron-d-n~e ~3gr~g~tQ4 (Fl~ure 3~). Sh~ co-loccllzation v-g ~hown u4Ing EM ~mmunocytoche~try o~ ultr~thln cryo~-ction4 a~ po~or~od cccord~ to eho ~tho~ of ~rapp t ~
25 19~9, J, C~ll BSol. 10~: 2417-2426, The i~unorQact~Vtty of pAb 90-29 ~a~ tetoct~d i~ th- lec~ron ~lcrographs u-lng S~uno~old p~re~clo~ unotol~ pdreicle~ a~yo-r ~ d1sor~t~ dot~ of unl~ar~
ol~e 1D tho olectron micsosr-ph~. a~pr4~0n~-t~v- r~ion~. ~b~-gold ~rticles Co~loc~ltz~ ~Itb the olactron-~anJo cggrag~to~, Ara ~0 lndl~ d by arro~.

' .

;
' ' EXA~PLE 15 Expression of Human APP Gene Product3 in COS Cell Transfections DNA transfections of COS cells (Gluzman, 1981, Cell 23: 175-182) demonstrate that pMTI-2360, pMTI-2362, pMTI-2369, and pMTI-46 express and secrete huoan APP-695 as described below and shown in Figure 29.
For DNA transfections, 60 mo culture dishes were seeded with approximately 2-5 x 105 COS cells/dish t-50% confluency) in 3 ml DMEM
and 10% fetal calf serum. Cells were cultured overnight at 37C in a 6% CO2 atmosphere. Cells were washed with PBS (no Ca~ or Ng~) and then 2 ml of DMEM plus 10% NuSeruo (catalog #50000) was added to each plate. Then 2 ~l of lOOOX chloroquine stock solution (0.1 M
chloroquine, Sig~a no. C-6628), 32 ~1 of DEAE dextran sulfate stock solution (25 mg/ml DEAE dextran sulfate, Sigma no. D-9885), and 4 ~g of DNA was added to each plate. Cells were incubated for 3.5 hours at 37C in a 6% CO2 atmosphere. Cells washed with PBS and then "shocked~ with 2 ml of 10% DMSO in PBS and incubated at 37C in a 6%
C02 atmosphere with DMEN plus 10% fetal calf serum for 48 hours then washed 3 times with PBS and cells were further incubated at 37C in 20 a 6% C02 atmosphere with "Cutter" media for an additional 24 hours.
Protein in culture media from COS cells, transfected COS cells, and huoan neuroglioma cell line H4 (A.T.T.C. no. HTB148) was concentrated by acid precipitation. Approximately 3 ml of each culture media, ice cold, was harvested and a 3 ml aliquot of 25%
trichloroacetic acid (TCA), ice cold, was added. Saoples were centrifuged at 20,000 x g for 30 minutes at 4C. The protein pellets were washed three tioes with 100% acetone and then centrifuged after each wash at 10,000 x g for 15 minutes at 4C. The pellets were dried in a vacuuo for -20 seconds, resuspended in 10 ~1 of NRBS
30 buffer (2~ SDS, 10% betaoercaptoethanol, 5% loading dye, 10%
glycerol, 0.125 M T-is pN 6.8), and boiled for 5 oinutes.
Cell supernatant protein was fractionated by polyacrylamide gel electrophoresis (8% running gel and 4% stacking gel) and transferred to Immobilon-P membrane by the technique of electroblotting using a Biorad Mini-Protean II apparatus and using procedures recomoended by the manufacturer. Huoan and oouse APP

i - ' ' ~' ~

-80- 2~ 7~
proteins, transferred from the polyacrylamide gels into Immobilon-P membranes, were detected by Western-blot staining as described in Example 14. Mouse and human APP-695, APP-751 and APP-770 proteins were detected using monoclonal antibody (mAb) 22C-ll.
APP protsin secreted into media from various transfected cell cultures was detected in Western-blots using mAb 22C-ll. COS cells express predominately APP-751 (ant/or 770) and a smaller amount of APP-695 (Figure 29, lane 8). The secreted forms of APP-695 and APP-751 (or 770) have apparent molecular weights of -93-105 kDa and -112-125 kDa, respectively (Weidemann et al., 1989, supra, and Palmert et al., 1989, su~ra). Human cerebral spinal fluid (CSF) contains predominately APP-695 (Palmert et al., 1989, su~ra.) and is included on the Western-blot as a control for APP-695 expression (lane 9, Figure 29). Several DNA transfections (pMTI-2360, lane 7; pMTI-2362, lane 6; pMTI-2369, lane 5; and pMTI-46, lane 4) exhibit significant increases in APP-695 immunostaining relative to APP-751(770) immunostaining (Figure 29). Therefore, these constructs express human APP-695 in COS cells. These APP-695 encoding minigenes are used as a "template" for construction of minigenes encoding alternate or mutant forms of APP. Because the parent APP-695 constructs express protein, it is highly likely that the other constructs also will express their proteins.

E~pression of APPs in Mammalian Cell Lines Stable cell lines expressing the 695, 751 and 770 forms of APP, as well as a mutated form of APP MC-100, were constructed as follows using bovine papilloma virus- (BPV) based vectors.
A. Cell Lines for APP-695. APP-751 and APP-770 Plasmid pMTI-4 described in Example 6 was mutagenized at the 5'-end of the APP-695 cDNA to create a new SalI restriction site.
In addit~on, during the mutagenesis procedure, the bases flanking the initiation codon, AUG, were altered to conform to the optimum sequence for translation initiation described by Kozak (Rozak, 1989, Cell Biol. lQ~: 229-241). The oligonucleotide primer used in the ' ~ ' .
, : .
. : ~

r mutagenesis and map of the resulting vector, pMTI-38, are shown in Figure 30a.
Plasmid pMTI-41 was constructed by deleting the unique KpnI
site in Bluescript KS (Stratagene; parent vector for pNTI-4).
S Bluescript KS was digested with ~paI and the overhangs were digested with mung bean nuclease by standard methods. The digested DNA was treated with ligase to circularize the vector and pMTI-41, lacking the ~E~I site was isolated.
The XbaI - HindIII fragment from pMTI-38, containing the APP-10 695 cDNA, was introduced into XbaI - HindIII digested pMTI-41 as shown in Figure 30 to obtain pMTI-42 which has only one Kpn site within the APP cDNA. pMTI-43 and pMTI-44 containing respectively the 751 and 770 forms of APP were constructed by replacing the KpnI
- ~glII fragment in pMTI-42 with the corresponding fragments from 15 pFC4-751 and pFC4-770 described in Example 3.
SalI fragments containing the APP regions in pMTI-42, pMTI-43 and pMTI-44 were introduced into the XhoI site of the BPV vector pMTI-52, placing them under the control of the mouse metallothionine promoter illustrated in Figure 31. As shown in Figurc 31, pMTI-52 contains the colEl replicon, the ampicillin resistance gene, the mouse metallothionine promoter a unique cloning site for cDNAs followed directly by the polyadenylation signal of SV40.
Specifically, pMTI-52 contains ~HI and XhoI cloning sites for introduction of cDNAs of interest. In addition, the pMTI-52 vector contains the entire 8 kb genome of BPV. The presence of BPV
sequences allows the vector to replicate as a multicopy episome in mouse C127 and NIH3T3 cells resulting in stably transformed cell lines. The plasmid pMTI-52 was constructed by ligating the -237 bp ~amHI-BclI fragment (containing the viral polyadenylation signals) from SV40 viral DNA into the unique ~HI site of pMTI-32.
Diagnostic restriction digestion of pMTI-52 with BamHI and PvuII gave the following DNA restriction fragments: -11.5 kb, -0.55 kb, and -0.25 kb. pMTI-32 was generated by ligating an -1.8 kb BamHI-~glII
restriction fragment from pMTI-29 (this DNA fragment contains the mouse metallothionein gene promoter, which can be obtained from alternative sources, for example, the -l.9 kb EcoRI-~lII fragment from plasmid pJYMMT(L) described in Example 9 also contains an analogous promoter fragment) into the unique BamHI restriction site of plasmid BPV-240.7. Plasmid BPV-240.7 was used as a source of the entire BPV genome and is a variant of the BPV vectors described and prepared by Howley et al., 1983, in Methods of Enzvmologv, Volume 101, ~u et al., eds., Academic Press, NY, pp. 387-402. Alternative sources of the -8 kb BPV genome may be used, in particular, any number of the BPV vectors described by Howley et al., suvra, with minor changes in restriction enzyme cleavage sites, can serve as a source of the BPV genome in place of BPV 240.7 in the construction of pMTI-52. Diagnostic restriction digestion of pMTI-32 with BamHI
and ~BdIII gave the following DNA restriction fragments: 8.0 kb and 4.1 kb. pMTI-29 was generated by inserting BglII, ~_I, and SalI
restriction sites (using a synthetic DNA linker) into the unique EcoRI restriction site of plasmid pM~Bneo. Plasmid pMVBneo has been described by Pavlakis et al., lg87, in Gene Transfer Vectors, Miller and ~alos, eds., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp. 29-58, and was used as a source of the mouse metallothionine gene promoter. Alternative sources of this promoter may be used, for example, plasmid pJYMMT(L) (see Example 9).
Diagnostic restriction digestions of pMTI-29 with either SalI or XbaI
yielded a single 6.7 kb DNA restriction fragment. The BPV vectors pMTI-53, pMTI-57 and pMTI-58 contain the 695, 751 and 770 forms of APP, respectively.
Each BPV vector, pMTI-53, pMTI-57 or pMTI-58 was transfected into mouse C127I cells ~a variant of C127 obtained from Dr. D.
DiMaio, Yale University) which are permissive for the high-copy-number, episomal replication of BPV vectors (Howley et al., 1983, Methods in Enzymol. 101 387-403). Vectors were introduced into cells by calcium phosphate precipitation-and the trans~ormed foci were isolated as described (Howley et al., 1983, ~y~). Alternatively, in some cases, the vectors were co-transfected (Howley et al., 1983, su~ra) with pSV2neo (Southern and Berg, 1982, J. Mol. Appt. Gen. 1:
327-341) which is capable of conferring resistance to the antibiotic G418. BPV vector DNAs were mixed with pSV2neo DNA at 5- to 10-fold molar excess (BPV vectors in excess) and transfected into C127I cells by calcium phosphate precipitation. Colonies resistant to G418 were isolated. The molar excess of BPV DNA over pSV2neo D~A ensured that almost every G418 resistant colony contained the cotransfected BPV
vector.
Transfection of APP cDNAs into various cell types has shown that the amino-terminal region of APP, including the Kunitz domain, is released into the medium (Weidemann et al., 1989, Cell 57: 115-126 and Palmert et al., 1989, supra. Therefore, serum-free, 24-hour supernatants from transformed foci and the G418 resistant colonies were screened for the appropriate form of APP by Western-blot analysis. Proteins in 1.5 ml of supernatants from semi-confluent to confluent 25 square cm flasks were concentrated approximately 15-fold by precipitation with trichloroacetic acid (TCA) prior to loading onto polyacrylamide gels. APP bands were visualized using the mouse monoclonal antibody 22Cll (see Example }4). Clones producing high levels of the appropriate APP form were expanded and propagated in culture. Supernatants from these cell lines, cMTI-53, cMTI-57 and cMTI-58 provided standards for the three forms of human APP.
B. Cell Lines for MC-100 Further transfections of mouse C127I cells were performed using the plasmid vector pMTI-70 (-12.9 kb, Figure 40). This plasmid was constructed by cloning an -615 bp XhoI-PvuII fragment from the vector pMTI-2371 (Figure 41) into the BPV vector pMTI-52. pMTI-52 was first digested with ~mHI and then a blunt-end was generated using Klenow.
The vector was then digested with XhoI and the large restriction fragment was gel-purified and ligated with the -615 bp ~h~I-PvuII
fragment from the vector pMTI-2371 to generate pMTI-70. Diagnostic digestion of pMTI-70 with HindIII revealed an -3.6 kb and an -9.3 kb restriction fragment. The pMTI-2371 plasmid was derived by cloning the -707 bp BamHI-SpeI fragment from pHTI-2337 between the BamHI and the XbaI sites of the Bluescript KS+ vector (see Example 6).
Construction of plasmid pMTI-2337 is described in Example 9 (part A, section 3).
Plasmid pMTI-70 contains the sequences derived from pMTI-2337 which encode the mutation designated MC-100 (gene product V, Figure .

4b, see also Figure 12). The fragment obtained from pMTI-2337 (with pMTI-2371 as an intermediate) used for the construction of pMTI-70 encodes the C-terminal segment common to the three forms of APP (695, 751, 770), including the A4 region, preceded by the secretion signal (i.e., signal peptide) of the APPs. Thus, translation of this APP
minigene i5 expected to result in the incorporation of the APP C-terminus into the membrane of the cell transfected with this minigene.
Plasmid pMTI-70 was transfected into mouse C127 cells as described above and colonies resistant to G418 were isolated to generate stable transfectant cell lines which included lines: cMTI70-A2, cMTI70-A3, cMTI70-A6, cNTI70-Bl, cMTI70-B2, and cMTI70-B3. Cell lysates of such resistant clones were analyzed by Western-blotting using a rabbit polyclonal antibody (pAb) SG369. The pAb SG369 15 (described in Buxbaum et al., 1990, Proc. Natl. Acad. Sci. USA 87:
6003-6006 was raised by immunization of a rabbit with a synthetic peptide corresponding to the C-terminus of human APP-695 using standard immunization procedures and techniques (as described in Buxbaum et al., 1990, su~ra). The synthetic peptide consisted of APP
20 amino acid residues 645-694, wherein the numbering of amino acids corresponds to those of human APP-695 as described in Kang et al., 1987, su~ra) and was prepared by the Yale Protein and Nucleic Acid Chemistry Facility (New Haven, CT). Rabbit polyclonal antibodies with similar characteristics as those of pAb SG369 have also been generated by other laboratories using various human APP-695 C-terminal peptides (Ishii et al., 1989, Neuropath. Appl. Neurobiol.
15: 135-147; Palmert et al., 1989, supra; and Bush et al., 1990, J.
Biol. Chem. 265: 15977-15983). Figure 42 shows the results of the Western-blot analysis of cell lysates of pMTI-70 transfected BPV
30 cell lines cMTI70-Bl, cMTI70-B2, and cMTI70-B3. The Western-blot shown in Figure 42 also shows the results of using cell extracts from control BPV cell transfectants which do not express MC-100.
Cell cultures were grown to 100% confluency, washed with 1 mM EDTA
in PBS, and extracted by boiling for 10 minutes in LX SSB (2~ SDS, 35 63 mM Tris pH 6.8, and 10~ glycerol), 5~ ~-mercaptoethanol, and 5~
bromphenol blue. The Western-blot analysis was performed as described above. The Western-blot illustrated in Figure 42 contains the following samples: lane 1, molecular weight markers; lane 2, cMTI52-A4 cell extract; lane 3, cMTI66-B6 cell extract; lane 4, cMTI66-C5 cell extract; lane 5, cMTI69-C6 cell extract; lane 6, cMTI69-A4 cell extract; lane 7, cMTI69-A5 cell extract; lane 8, cMTI70-Bl cell extract; lane 9, cMTI70-B2 cell extract; and lane 10, cMTI70-B3 cell extract. Polyclonal antibody SG369 was used in this Western-blot analysis. BPV cell line cMTI52-A4 was transfected with pMTI-52 ~BPV cloning vector); BPV cell transfectant lines cMTI66-10 B6 and cMTI66-C5 carries BPV vector pMTI-66 which encodes the A4 peptide of human APP (gene product VIII, Figure 4b; see Example 7 above); and BPV cell transfectant lines cMTI69-C6, cMTI69-A4, and cMTI69-A5 carry the BPV vector pMTI-69 which encodes the Sp-A4 peptide of human APP (gene product VII, Figure 4b; see Example 9A, 15 section 4). A major immunoreactive band between 14 kD and 21 kD
representing the product of the APP minigene is seen. Also present are immunoreactive bands of higher molecular weights consistent with their being aggregation products of the primary translation product (indicated by arrows).
The transcription of the APP minigene in pMTI-70 is under the control of the mouse metallothionine promoter which is inducible by heavy metals such as cadmium (Hamer, D.H. and Welling, M.J., 1982, J. Mol. Appl. Genet. 1: 273-288). Induction of cell lines cMTI70-A2, cMTI70-A3, cMTI70-A6, cMTI70-Bl, cMTI70-B2, and cMTI70-B3 with cadmium would be expected to result in increases in mRNA levels and resultant increases in MC-100 protein levels as shown in the Western-blot illustrated in Figure 43. Cell cultures were grown to 100~
confluency, washed with PBS, incubated with DMEM with 5 ~g/ml cadmium chloride at 37C in 5% CO2 for 16 hours, and then extracted by boiling for 10 minutes in LX SSB (2~ SDS, 63 mM Tris pH 6.8, and 10~ glycerol), 5% ~-mercaptoethanol, and 5~ bromophenol blue. The Western-blot analysis was performed as described above. The Western-blot illustrated in Figure 43 contains the following samples: lane 1, cMTI63-Bl cell extract; lane 2, cMTI63-C2 cell extract; lane 3, molecular weight ~arkers; lane 4, cMTI53-Al cell extract; lane 5, cMTI70-A2 cell extract; lane 6, cMTI70-A3 cell extract; lane 7, 2~1~a~7 cMTI70-A6 cell extract; lane 8, cMTI70-Bl cell extract; lane 9, cMTI70-B2 cell extract; and lane 10, cMTI70-B3 cell extract. BPV
cell transfectant lines cMTI63-Bl and cNTI63-C2 carry BPV vector pMTI-63 which encodes the human APP-695 with a C-terminal addition of the Chlamvdia epitope (see Example 8) and BPV cell transfectant line cMTI53-Al which carries BPV vector pMTI-53 which encodes human APP-695. The higher molecular weight bands corresponding to the aggregated molecules increase in intensity upon cadmium induction (as indicated by arrows). This observation is consistent with the expectation that aggregation is a concentration dependent phenomenon.
The pMTI-70 transfected and G418 selected cells were also analyzed by immunofluorescence of stained cells and immunoprecipitation of cell lysates using the SG369 antibody. The results demonstrated the accumulation of the MC-100 fragment in the transfected cells. Figures 44a and 44b show immunofluorescence results of two representative fields where a limited number of cells in the population of cMTI70-A6 cells show intense fluorescence.
Transfected cell line cMTI-53 (which expresses human APP 695) does not exhibit these immunofluorescent cells (Figure 44c). Cultures of cell line~ cMTI70-A6 (transfected with pMTI-70, see above) and cMTI53-Al (express human APP 695) were grown to 70% confluency using standard culture conditions, the cells were washed with PBS, and incubated for 16 hours with DMEM supplemented with 5 ~g/ml cadmium chloride. The induced cMTI53-Al and cMTI70-A6 cells were then washed twice with PBS, and fixed using 4% paraformaldehyde in PBS at room temperature for 10 minutes. The cells were permeabilized with 0.2~
Triton X-100, 10 mM Tris pH 8.0, 0.2 mM EDTA at room temperature for 5 minutes. The fixed and permeabilized cells were then incubated with affinity purified pAb SG369 (1:200 dilution of stock) in PBS and 3% bovine serum albumin (BSA) at room temperature for 60 minutes.
The cells were washed 5 timeq with 3~ BSA in PBS and then incubated with goat anti-rabbit IgG con~ugated with rhodamine (obtained from Boehringer Mannheim) in PBS and 3% BSA at room temperature for 30 minutes. The cells were then washed 5 times with 3~ BSA in PBS. The fluorescence of the cells was observed on mounted slides using a Zeiss IM fluorescent microscope. As shown in Figure 44, the staining .
. .
'' ' ~ .

- . : ; -- :.

;. :, 7'~

is punctate in nature and localized at the cell periphery, away from the site of synthesis in the endoplasmic reticulum (ER) and Golgi.
This staining pattern suggests highly localized concentrations of MC-100 protein. Upon continued passage of the C127I/pMTI-70 transfected S cells, it has been observed that fluorescent cells are lost from the population with continued passage. This suggests that production of MC-100 may confer a selective disadvantage to these cells.
Figure 45 shows a Western-blot of immunoprecipitated NC-100 from extracts from the cell line cMTI70-A6 (transfected with pMTI-70, see above). The results indicate that the MC-100 aggregates, observed in Figures 42 and 43, can be immunoprecipitated from cell lysates (Figure 45, lane 6). Cultures of cell line cMTI70-A6 (transfected with pMTI-70, see above) were grown to 100% confluency using standard culture conditions, the cells were washed with PBS, and incubated for 16 hours with DNEM supplemented with 5 ~g/ml cadmium chloride. The induced cMTI70-A6 cells were resuspended twice, washed with 1 mM EDTA in PBS, and the cells were pelleted by centrifugation (1000 x g) and resuspended in IP (lysis) buffer (100 mM Tris pH 7.4, 150 mM NaCl, 2 mM NaN3, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, and 40 units/ml aprotinin). An aliquot of this cell lysate appears in Figure 45, lane 1. The cell lysate was incubated with a 1:50 dilution of affinity purified pAb SG369 for 2 hours at 4C. The extract was then incubated with a 1:10 dilution of protein-G Sepharose (stock, 2 mg/ml in PBS; obtained from Sigma) at 4C overnight with gentle agitation. The protein-G Sepharose bead~ were then collected by centrifugation (12,000 x g for 15 seconds at 4C) and an aliquot of the supernatant appears in Figure 45, lane 2. The pellet was washed 3 times with IP (lysis) buffer.
An aliquot of each wash appears in Figure 45, lanes 3, 4, and 5. The proteins were solubilized, boiling for 10 minutes in lX SSB (2% SDS, 63 mM Tris pH 6.8, and 10% glycerol), 5% ~-mercaptoethanol, and 5%
bromophenol blue. An aliquot of the solubilized immunoprecipitant appears in Figure 45, lane 6. The Western-blot analysis was performed as described above using the SG369 antibody.
The C127I/pMTI-70 clones thus provide a mammalian cell host/vector system in which the aggregation of a segment of APP

.

2 ~ r~

containing the A4 region was observed. Since this reaction is a critical step in amyloid formation, this host/vector system is valuable for studying: (i) the steps in amyloid formation by studying the aggregation process i~ vitro and (ii) methods for intervention into this process by characterizing chemical and physical agents that accelerate or interfere with amyloid aggregation. In addition, the NC-100 minigene in pMTI-70 may be expressed in other cell lines including neurons to study amyloid formation in different cell lines.

EgAMPLE 17 Generation of Human APP-specific house ~onoclonal Antibodiss, 56-1 and 56-2 Monoclonal antibodies reactive to the 56 and 75 amino acid Kunitz domain inserts of APP were generated as follows:
Immunogen: The immunogen used in the immunization of mice was the enriched pellet fraction of bacterium E. coli expressing the 75 amino acids of the Kunitz domain as a fusion to the first 36 amino acids of E. coli recA protein (Sancar et al., 1980, Proc. Natl. Acad.
Sci. USA 77: 2611-2615). The fusion protein, which segregated into the pellet fraction of the expressing strain, was enriched by detergent and water washes. The resulting insoluble pellet was solubilized in 8 M urea and the urea was removed by dialysis against phosphate buffered saline (PBS). Dialysis caused the precipitation of a part of the solubilized material. The emulsion resulting from the dialysis was used to immunize mice. The recA-75 fusion represented over 30% of the protein in the emulsion.
Immunizations. Three 8-week-old Balb/c mice were immunized intraperitoneally with 100 ~g of immunogen emulsified with an equal volume of complete Freunds ad~uvant. Then, 21 and 28 days later, mice were given additional intraperitoneal in~ections of 100 ~g of immunogen emulsified in an equal volume of incomplete Freunds ad~uvant. After an additional seven days, the mice were boosted intravenously with 20 ~g of immunogen. Three days later the spleens were removed and somatic cell hybrids were prepared by the method of Herzenberg (Nerzenberg et al., 1978, Handbook of Ex~erimental Immunologv (D. M. Weir, ed.) Blackwell Scientific Publications, ', ' .

7 ~

Oxford, pp. 25.1-25.7) with some modifications (Lerner et al., 1980, J. Exp. Med. 152: 1085-1101).
ELISA assav: The enriched pellet fraction containing the recA-75 fusion was dissolved in PBS and used in an ELISA assay. As a negative control, similar pellet fraction prepared from an E. coli strain expressing a fusion of the same 36 amino acids of recA (as in recA-75) with a segment of APP-695 (which does not have the Kunitz insert) was used.
The ELISA assay was conducted as follows. Culture fluids from growing hybridomas were tested for the presence of specific antibody using ELISA. 1 ~g of extracts containing recA fusion proteins was allowed to adsorb to each well of Immunolon II EIA plates (Dynatech, Chantilly, VA) by overnight incubation at 4C in 50 ~1 O.OlM sodium carbonate pH 9.5. Non-specific protein binding sites in each well were blocked by incubation with 200 ~1 PBS containing O.05~ Tween-20 and 1~ BSA followed by washing with PBS/0.05% Tween-20. ~ells were then sequentially incubated with 100 ~1 of hybridoma tissue culture supernatant, washed, and 100 ~1 of a 1:1,000 dilution (in PBS/Tween-20) of peroxidase labelled affinity purifled goat anti-mouse IgG (Kirkegaard and Perry, Gaithersburg, MD). All incubation steps, lasting one hour each, were done at room temperature. Bound peroxidase labelled "second antibody" was detected using the peroxidase substrate tetramethylabenzidine (TMB) according to the manufacturer's instructions (Kirkegaard and Perry, Gaithersburg, MD); optical density at 450 nanometers was then determined for each well. Isotypes of positive hybrid culture fluids was determined using an ELISA assay in which 1 ~g of anti-mouse Fab was adsorbed to each well of Immunolon II EIA plates followed by sequential incubations with culture supernatants and peroxidase labelled antiserum specific for mouse IgGl, IgG24, Ig2b, IgG3~ and IgM-Characterization of the monoclonal antibodies: The antibodieswere characterized on Western-blots by comparing their reactivities against the whole recA protein and with fusions of the first 36 amino acids of recA protein with the 56 and 75 amino acids of Kunitz domain. These comparisons were used to eliminate antibodies directed against the recA portion of the immunogen and to localize the : .

:

-9o-reacting epitopes to the 56 or the 19 amino acid regions comprising the 75 amino acids of the Kunitz antigen. Three antibodies, 56-1, 56-2, 56-3 reacting with the 56 amino acid Kunitz domain were isolated by this procedure. They were then tested similarly for 5 reactivity against the 695, 751 and 770 APP forms secreted from mammalian cells described in Example 16. All three were found to react with human APP-751 and APP-770 from transfectants but not with the APP-695 form.
It has been observed (Weidemann et al., 1989, ~ and Palmert et al., 1989, supra) that many cell lines in culture secrete all three forms of APP to various extents, with the APP-751 and APP-770 forms predominating in most cases. The 56-1 and 56-2 mAbs showed no cross-reactivity with the endogenous mouse versions of APP (Figure 32). All forms of mouse and human APPs were found to react with the 15 22Cll antibody raised against human 695 precursor. The 56-1 mAb was further tested against supernatants of 751 and 770 transfectants described in Example 16 and also against supernatants of mouse L-cells and COS monkey cells. As shown in Figure 33, the 56-1 mAb reacted strongly with supernatants of the 751 transfectant and with the monkey APP but not against mouse APPs either endogenous in the C127 mouse cell host or in mouse L-cells. The 22Cll mAb detected all forms of APP from all animal species tested here. Thus, the results in Figure~ 32 and 33 establish that the 56-1 and 56-2 mAbs are being specific for primate (human and monkey) APPs.

, . , ' ', ' ~ .

.
.

.
.:

Claims (10)

1. A minigene for expression of an amyloid precursor protein (APP) or derivatives thereof comprising (a) a regulatory region, said regulatory region capable of directing tissue and cell specific expression, (b) a gene construct encoding said APP or derivative thereof, and (c) genetic sequences containing a RNA polyadenylation signal.

2. A minigene according to Claim 1 wherein said regulatory region further includes genetic elements conferring a developmental expression pattern of said gene construct of (b) similar to the developmental expression pattern observed in the endogenous APP gene.

3. A minigene according to Claim 1 or 2 wherein said minigene further comprises (d) an intronic sequence containing acceptor and donor sites for splicing:

4. A minigene according to Claim 1 wherein said gene construct encodes APP-695, APP-751, APP-770, a mutated APP, a truncated APP or an A4 peptide.

5. A minigene according to Claim 1 wherein said gene construct of (b) is replaced by a reporter gene, said reporter gene capable of being monitored to assess the function of said regulatory region, or by a fusion protein containing a reporter gene and a gene encoding said APP or derivative thereof.

6. A minigene according to Claim 3 wherein said minigene further comprises (e) an antigenic tag for the expression of a tagged APP or APP derivative, said tagged APP or APP
derivative capable of being detected to assess said expression.

7. A minigene cassette for transfer and expression in transgenic mice of APP or derivatives thereof comprising a NotI fragment containing (a) a regulatory region, said regulatory region capable of directing tissue and cell specific expression.
(b) a gene construct encoding said APP or derivative thereof, (c) genetic sequences containing a RNA polyadenylation signal, and (d) an intronic sequence containing acceptor and donor sites for splicing.

8. A transgenic mouse, including progeny, embryo or cell derived from said transgenic mouse, capable of expressing an amyloid precursor protein (APP) or derivatives thereof in a tissue and cell specific manner.

9. A transgenic mouse, including progeny, embryo or cell derived from said transgenic mouse, comprising a transgene for the expression of an amyloid precursor protein (APP) or derivative thereof.

10. A transgenic mouse according to Claim 9 wherein said transgene comprises a NotI minigene cassette containing (a) a regulatory region, said regulatory region capable of directing tissue and cell specific expression, (b) a gene construct encoding said APP or derivative thereof, (c) genetic sequences containing a RNA polyadenylation signal, and (d) an intronic sequence containing acceptor and donor sites for splicing.