From the amyloid protein A4 to isolation of the first Alzheimers disease gene amyloid A4 precursor protein APP

Super Memory Formula

New Alzheimer Cure

Get Instant Access

Rudolph E. Tanzi1

Introduction

As life expectancy continues to increase, so will the prevalence and incidence of Alzheimer's disease (AD) in our elderly population; by 2050, as many as 14 million AD cases are expected in the USA, alone. AD is characterized by global cognitive decline in association with specific brain pathological lesions, neuronal loss, and synaptic pruning. The disease takes its name from Dr. Alois Alzheimer, a German psychiatrist who in the fall of 1906 suggested that specific physical aberrations in the brain were driving dementia in his female patient, Auguste D (Alzheimer 1907a). Alzheimer had been treating Auguste D since she was first admitted at age 51 to the Hospital for the Mentally Ill and Epileptics in Frankfurt for "frenzied delirium." Shortly after the patient's death at age 56, Alzheimer presented the results of his post-mortem examination of her brain at a meeting in Tubingen. He wisely took advantage of Camillo Golgi's new silver staining technique to examine the neurons in his patient's brain tissue. Alzheimer was not the first to describe the appearance of senile plaques (clusters he called "miliary bodies"); neither did he know that the core was made of amyloid, despite Virchow's description of "amyloid" decades earlier. However, with the help of Golgi's silver stain, Alzheimer does appear to have been the first to suggest that the plaques were associated with "dense bundles of fibrils" choking the inside of cortical neurons, i.e., neurofibrillary tangles, and that these lesions were the cause of dementia in Auguste D. Thus, the pathogenic mechanism presented by Alzheimer in 1906 can in some ways be considered the earliest form of the "amyloid hypothesis."

It was not until the 1960s that Robert Terry, Michael Kidd, Henry Wisniewski, and others would employ both light and electron microscopy to reveal the ultrastructural details of plaques and tangles (reviewed in Tanzi and Parson 2000). However, the question of primacy remained. Did plaques or tangles come first, and which lesion, if either, was killing off neurons? While these questions could not be immediately addressed, by the late-60s, Bernard Tomlinson, Gary Blessed, and Martin Roth provided the next boost for the emerging amyloid hypothesis when they suggested for the first time that dementia was correlated with senile plaque counts in the cerebral gray matter (Roth et al. 1966). Later, in 1968, these same investigators showed that over 60% of the demented elderly (the "senile") harbored the same lesions observed by Alzheimer in his

1 Genetics and Aging Research Unit, MassGeneral Institute for of Neurology, Massachusetts General Hospital, Harvard Medical School

Genetics and Aging Research Unit, Department of Neurology, Massachusetts General Hospital, 114 16th Street, Charlestown, MA, 02129, USA

pre-senile patient, Auguste D (Blessed et al. 1968). By 1970, as average life expectancy was hitting 70 years old, it became clear that AD was a prevalent cause of most cases of "senility." And, a growing amount of attention was being paid to the origins of senile plaques.

Discovery of the first AD gene: amyloid p (A4) Precursor Protein

The steadily emerging amyloid hypothesis received perhaps its greatest infusion of support in the early 1980s when Dr. George Glenner, an amyloidologist, entered the scene and argued that amyloid was the central player in AD pathology (Glenner 1981). By the summer of 1983, Glenner and his assistant, Cai'ne Wong, had started obtaining their first amino acid sequences from cerebral blood vessel amyloid isolated from a patient with Down syndrome (Tanzi and Parson 2000). In May, 1984, they published the first sequence of the 4 kDa peptide (called the amyloid p protein), which was found to be the major component of p-amyloid (Glenner and Wong 1984a). In a follow-up paper in August 1984, Glenner and Wong (1984b) showed the same amino acid sequence for amyloid p-protein deposits in both Down syndrome and AD and, since Down syndrome is caused by trisomy 21, made the prophetic statement that a genetic defect causing AD might be localized on chromosome 21. A year later, Colin Masters, who had teamed up with Konrad Beyreuther, reported that the amino acid sequences of senile plaque core proteins from AD and Down syndrome brains were virtually identical to the sequence published by Glenner (Masters et al. 1985a). The stage was nowset for "reverse genetics" to potentially furnish the first AD gene and first molecular target for drug discovery.

Between 1980 and 1985, I had been working with Dr. Jim Gusella on a project that would become the first to identify and employ human DNA variants to localize a disease gene where no biochemical clues were available regarding etiology or patho-genesis of the disease. While considered quite routine today, this approach had never been used before, and the project had more than its share of "doubting Thomases." Yet, Jim Gusella believed (and convinced me) that if we could track the inheritance of a sufficient number of common DNA variants [restriction fragment length polymorphisms (RFLPs)] through families with the devastating movement disorder, Huntington's disease (HD), we might be able to find the location of that disease gene in the human genome. This finding would then pave the way for later pinpointing the exact genetic cause of the disease, all solely through genetics. When we started the project in 1980, only one human gene polymorphism was known, AW101 (amidst a handful of protein polymorphisms).

In the fall of 1980, we set out to identify as many human RFLPs as possible with the hopes of finding one that co-segregated with the onset of HD. The problem, Jim said, was that we might have to test hundreds, if not thousands, of RFLPs before we found one that revealed the location of the HD gene. As it turned out, not just luck, but miraculous luck was on our side. Among the first 12 RFLPs we pulled from the genome, one, which we had named "G8," was tightly linked to HD and mapped to the short arm of chromosome 4 (Gusella et al. 1983). Ten years later, we would learn that G8 sits less than 200 kilobases from the HD gene mutation. The odds against pulling out a RFLP so close to the HD gene mutation in a genome of three billion basepairs were greater than

15,000:1! And, if that were not lucky enough, even if we did not find G8, the very next RFLP that was randomly chosen, G9, was only 20 million basepairs away from the HD gene, which meant that, eventually, with more families, we would have found linkage to HD with G9, even in the absence of G8!

While working on the HD screen, I had also initiated a side project in Jim's lab in collaboration with Paul Watkins - an attempt to build the first complete genetic linkage map of a human chromosome. We chose the smallest one, chromosome 21, partly because of its role in Down syndrome (trisomy 21). We began isolating RFLPs from chromosome 21 (using chromosome 21-specific somatic cell hybrids) in an attempt to build a complete genetic linkage map that could be used to map features of Down syndrome. While pursuing the chromosome 21 genetic linkage map, I had read Glenner's prediction of an AD gene on chromosome 21 in his 1984 paper. In the meantime, Jim was able to obtain cell lines from a Canadian family with early-onset (< 65) familial AD (FAD) from Ron Polinsky and Linda Nee at the NIH. In the summer of 1984, after testing that family for linkage to markers on our chromosome 21 map, Jim had me bring the FAD-chromosome 21 genotype data to Michael Conneally's lab at the University of Indiana. There I would test for genetic linkage of chromosome 21 to FAD in the Canadian family, with the help of his graduate student, Peggy Wallace. The results were dismally negative. Meanwhile, Jim's lab had collected another FAD family of Italian origin with the help of Robert Feldman (Boston University) and Jean-Francois Foncin (La Salpetriere Hospital, Paris). I started testing our chromosome 21 markers as soon as I got back to Boston. But once again, by early 1985, we had found no signs of genetic linkage of chromosome 21 with FAD in the second family.

Around this time, Peter St. George-Hyslop had joined Gusella's lab as a postdoctoral fellow and assumed responsibility for the FAD-chromosome 21 linkage study, which was now extended to two additional kindreds, one from Germany and one from Russia (from Dan Pollen, University of Massachusetts). Meanwhile, by the fall of 1985, I had become a graduate student in the Neuroscience Program at Harvard Medical School. For my first rotation project, I had decided to employ a "reverse genetics" approach to isolating the gene responsible for the amyloid p-protein gene. For this purpose, I had joined up with renowned Down syndrome geneticist, Dr. David Kurnit, at Boston Children's Hospital. He set me up to work with his post-doctoral fellow, Rachael Neve, who had been constructing human cDNA libraries. My chromosome 21 map collaborator, Paul Watkins, and I then designed "best-guess" oligonucleotides to the amyloid p protein amino acid sequences published by Glenner and Masters.

To isolate the amyloid p protein gene, we employed what we called the "genomic window" strategy. This novel approach was based on comparative Southern blot analysis using various restriction fragment patterns. Briefly, we designed two non-overlapping oligonucleotides to the amyloid p protein amino acid sequence: the first, a 21-mer corresponding to amino acids 1-7, and the second, a 48-mer corresponding to amino acids 9-24. We then screened Rachael Neve's human fetal brain and fetal liver cDNA libraries and only pursued cDNA clones that hybridized independently to both oligonucleotides. We next picked cDNAs that hybridized to the same or overlapping sets of bands (on Southern blots containing human DNA cut with various restriction enzymes) as those detected by the two oligonucleotides used to screen the libraries. Next, just in case Glenner's prediction was correct, we asked whether any of those cDNAs mapped to chromosome 21, using human-rodent somatic cell hybrid cell lines

(from David Patterson, University of Colorado, and Margaret Van Keuren, University of Michigan). Two clones, FB63 and FB68, each detected a set of bands that matched subsets detected by the two screening oligonucleotides. Surprisingly, both clones mapped to chromosome 21, in agreement with Glenner's prediction. The two clones contained the same 1.1 kb EcoRI fragment and sequencing of FB68 by Susan Pagan revealed that this was a partial cDNA clone encoding a protein containing residues 3-29 of the amyloid p protein (Tanzi et al. 1987a). The final proof that the cDNA corresponded to a single copy gene on chromosome 21 came from hybridization to a whole genome somatic cell hybrid panel (from Gail Bruns, Harvard).

We next examined the expression profile for the gene, which we called the "p protein" gene, and found it be a 3.2 kb message ubiquitously expressed in all human tissues tested, with its highest expression in brain, heart, kidney, spleen, and pancreas. The gene was also expressed throughout adult brain, with highest expression in brain regions, A40, A44, A20/21, A10 and cerebellar cortex (Fig. 1). Next, we showed that an extra dose of the gene led to excessive amounts of message in Down syndrome patients, most likely explaining how these patients accumulate excessive amounts of amyloid p protein in their brains by middle age (Fig. 2). Finally, we used our chromosome 21 markers and RFLPs detected by FB68 to genetically map the p protein gene near marker, D21S1, from our chromosome 21 linkage map. In parallel studies, Peter St. George-Hyslop was continuing to test markers from our chromosome 21 map in the four FAD families, and together with genetic analyst, Jonathan Haines, had found evidence for genetic linkage of FAD to the region of chromosome 21 around the same marker, D21S1 (St. George-Hyslop et al. 1987a). Statistical significance for the linkage result derived primarily from the Italian FAD family. However, my own earlier analyses of the chromosome 21 markers in the Italian and Canadian FAD families had yielded only negative results. Two possible explanations for the discrepancy were 1) Peter had simply tested additional chromosome 21 markers in all four FAD families, and 2) Jonathan had employed the relatively new method of "multi-point linkage analysis," which tests several markers for linkage simultaneously and can, thus, yield different linkage results from the single locus analyses that I had carried out earlier.

Soon after our p protein gene [later renamed amyloid p (A4) precursor protein {APP)] cloning paper (Tanzi et al. 1987a) and the chromosome 21-FAD linkage paper (St. George-Hyslop et al. 1987a) were published in Science, I completed the definitive experiment aimed at asking whether the APP gene was linked to FAD in the four Massachusetts General Hospital (MGH) FAD pedigees by analyzing the segregation of APP gene RFLPs in all four families. The genetic linkage results were all negative. APP was clearly not the genetic culprit in these four FAD kindreds (Tanzi et al. 1987b). This finding meant that, even if there were a gene on chromosome 21 responsible for FAD in these four pedigrees, as purported in the St. George-Hyslop et al. (1987a) study, it was not APP. Later, these same four FAD pedigrees would be shown to actually be linked to chromosome 14 and to contain mutations in the presenilin 1 gene (Sherrington et al. 1995). Ironically, however, the most likely spurious multi-point linkage of these four FAD kindreds to chromosome 21 in 1987 (St. George-Hyslop et al. 1987a) had motivated other groups to analyze their own independent FAD families, some of which would be genuinely linked to chromosome 21, and which would later reveal pathogenic FAD mutations in the APP gene.

Fig. 1. Expression of APP (FB63) in (a) 20- to 22-week-old human fetal tissues and (b) adult brain regions (from Tanzi et al., 1987a)
Fig. 2. Expression of APP (FB68L) andTauin 19-week-old normal (lane a) and trisomy 21 (laneb) brain; adult normal (lane c) and AD (lane d) cerebellum; adult normal (lane e) and AD (lane f) frontal cortex (from Tanzi et al., 1987a)

Ultimately, the amyloid p protein (A4; Ap) sequence was successfully employed by four different groups to isolate the APP gene (Goldgaber et al. 1987a; Kang et al. 1987; Robakis et al. 1987a; Tanzi et al. 1987a). However, only the Kang et al. study isolated the entire APP cDNA. This full-length APP clone revealed APP to be a type I integral transmembrane protein. Later in 1988, we discovered a novel transcript of

APP (APP751) containing an alternatively spliced exon encoding a Kunitz protease inhibitor domain. We found this to be the main form of APP in the periphery (Tanzi et al. 1988), and interestingly, the secreted portion of this form of APP had previously been identified as protease nexin II, which plays a key role in the coagulation pathway (Van Nostrand et al. 1989). To date, this is the clearest physiological role known for APP. As an interesting side bar to the APP cloning study, one of the cDNA clones that we had pulled out with our original oligonucleotide screen for APP turned out tobe the gene that causes Wilson's disease, a copper toxicity disorder mainly affecting the brain and liver (Tanzi et al. 1993). Interestingly, the 48-mer oligonucleotide corresponding to amino acids 9-24 of the Ap region was homologous to a sequence in the Wilson's gene encoding a copper-binding motif, thus explaining how this gene was fished out in the same experiment that landed APP. We would later show that the 9-24 amino acid region of Ap also binds copper, which, along with zinc, drives oligomerization and aggregation of the peptide (Bush et al. 1994). Oddly enough, however, while the homologous DNA regions between the APP and the Wilson's disease gene both encoded copper-binding motifs, they were of two different types: the motif in APP was histidine-based whereas the motif in the Wilson's disease protein was cysteine-based. Yet, both types of copper-binding motifs in APP and the Wilson's disease protein were encoded in a single homologous stretch of DNA, in two different reading frames! To this day, is it is unclear whether this is due to simple coincidence or an example of evolutionary economy of function in the genome. Interestingly, APP contains another copper-binding site in its N-terminus, and this motif is uniquely conserved in the two other homologs of the human APP family that Wilma Wasco and I later identified, APLP1 (Wasco et al. 1992) and APLP2 (Wasco et al. 1993).

The first disease mutation in APP was reported in 1990 by Frangione, Van Broeck-hoven, and colleagues, who after sequencing exons 16 and 17 of the APP gene (encoding the Ap domain), discovered a mutation that caused hereditary cerebral hemorrhage with amyloidosis in a Dutch family linked to chromosome 21 (Levy et al. 1990; Van Broeckhoven et al. 1990). Sequencing of these same two APP exons in FAD families (that were genuinely linked to chromosome 21) then led to the discovery of the first FAD mutation in 1991 (Goate et al. 1991). Later, in the summer of 1995, St. George-Hyslop's group, in collaboration with our and other laboratories, reported that the original four MGH FAD pedigrees actually harbored mutations in the gene called S182 (now prese-nilin 1; PSEN1) on chromosome 14 (Sherrington et al. 1995). A month and a half later, in collaboration with Jerry Schellenberg, we first reported FAD mutations in the S182 homolog, STM2 (now called presenilin 2; PSEN2) on chromosome 1 (Levy-Lahad et al. 1995a). Collectively, the > 160 known mutations in APP and the presenilins account for roughly half of all cases of early-onset FAD and < 1% of all AD cases. However, studies of these three genes have arguably provided the most valuable clues we currently have regarding the etiology and pathogenesis of AD.

Was this article helpful?

0 0
Anti Aging Made Easy

Anti Aging Made Easy

When it comes to reducing wrinkles, you really have to take your needs seriously. There are a number of factors that play a role in the health and well being of your skin. It is often hard to understand how products work and why they may not work even if they promise to do so. If you are considering wrinkle creams and a regimen of facelifts, you may go broke in the process. Before you do all of that, consider going through a process of determine the very best solution for your needs.

Get My Free Ebook


Post a comment