Allen D. Roses1
The past 14 years from 1992 to 2006 provide an interesting perspective for the application of genetic methods to Alzheimer's disease (AD) research. In the science of genetics, these years saw the beginning of linkage analyses for the identification of inherited disease mutations as well as susceptibility genes for complex diseases. This was followed in 1999-2001 with maps of single nucleotide polymorphisms (SNPs) across the now-sequenced genome, ushering in the latest era of whole genome association studies of sporadic patients with complex diseases. For our AD studies, clinical collection of patients from families in which there were two or more AD patients started in 1981. The hypothesis was that AD was a complex disease with multiple contributing genetic influences that could possibly be identified by a new method of genetic research: linkage analysis using a growing number of genetic variants distributed across the genome. At that time, few authorities considered AD to be a genetic disease - susceptibility genes were a concept of the 1990s. Therefore, the initial work was carried out alongside genetic studies of inherited muscular dystrophies, hereditary neuropathies, and other genetic neuromuscular diseases. By 1990, the Duke Joseph M Bryan ADRC at Duke University had linked late-onset "sporadic" AD to a region on chromosome 19 spanning millions of base-pairs, or hundreds of genes. The linkage method was then, and is now, hypothesis generating; the initial assumption was that there are genetically determined factors contributing to the disease.
As more information about genes located on chromosome 19 began to accumulate, testingeachnew gene in thebroad linkageregionbecameaone gene:oneresearchfellow investigation. When Warren Strittmatter was recruited to Duke, he continued his earlier studies of amyloid. Examination of CSF amyloid on various gel separations always was accompanied by another prominent unidentified protein band. We hypothesized that the band might be an important factor, so Dr. Strittmatter cut out the bands from the separation gels and investigated proteolytic fragments of the extracted proteins for sequence analyses. In late 1992, after a frustrating five months of experiments, Dr. Strittmatter found two peptide fragments from the extracted band contained amino acid sequences that were identical to sequences of fragments from apolipoprotein E. Dr. Strittmatter came into my office quite upset about wasting his time on sequencing a protein that had no known relationship to AD. However, as soon as he said "APOE," a synaptic connection fired in my head.
From my linkage studies in myotonic muscular dystrophy (DM), which is located on chromosome 19, I was aware that APOE was the first gene localized to chromosome 19. APOE was instrumental in localizing a long-known linkage of DM to the Lutheran blood
1 GlaxoSmithKline Research and Development Research Triangle Park, NC 27709
group and a salivary secretor locus to chromosome 19. The entire laboratory, including several post-doctoral fellows, was then focused on proteins within the chromosome 19 linkage region for AD. APOE was located in the middle of this linkage region but was not ever considered to be a candidate gene for AD. We immediately searched the textbooks on cardiology and found references to a new PCR method to measure nucleotide polymorphisms of APOE for the three common alleles, APOE2, APOE3, and APOE4. My problem was that not a single fellow or technician would take time off their own candidate genes to organize this APOE allele assay. I had never run a PCR myself, being from another era of biochemistry.
Several days before this revelation 14 years ago, my daughter Stephanie Margaux Roses was born. Her mother, Ann Saunders, was on maternity leave from her fellowship in a mouse chromosome mapping laboratory - and she knew how to perform PCR. A deal was made: if she came into my laboratory for a couple of weeks and set up the APOE assay, I would take care of Stephie at home. Within two weeks, the assay was running and validated with samples of known APOE genotypes obtained from the reporting laboratory in Texas. Approximately 50 DNA samples from AD patients were run along with 50 controls. The differences were obvious to the eye: there were many more APOE4 alleles in the AD patients. The next weeks allowed multiple confirmations using DNA from several hundred patients and several hundred controls. It was clear that there was an extraordinarily statistical significance for the association between APOE4 and AD (Strittmatter et al. 1993a; Saunders et al. 1993).
Since we had reasonably good historical age-of-onset information on the patients collected over the past decade, an age-of-onset analysis based on each APOE genotype clearly established that APOE4 was associated with a significantly lower age-of-onset distribution than APOE3 or APOE2. The epidemiologic data were important as a disease-specific independent confirmation of the genetic association, rather than simply additional series of AD patients. These data were also published in 1993, the same year as our publications reporting the higher allele frequency of APOE4 in both familial and sporadic cases (Strittmatter et al. 1993a; Saunders et al. 1993; Corder et al. 1993). The laboratory also initiated a series of biologic experiments to accompany the genetic and epidemiologic data, including the binding of apoE protein to amyloid and tau and the association of apoE4 > apoE3 with increased amyloid plaques in AD patients.
Since the AD field has had a greater interest in the amyloid causation hypotheses (see most of the chapters in this book), it was not until after several clinical papers and multiple letters in Lancet confirmed the association of APOE4 and AD that a few AD research laboratories accepted the relationship, but always as a factor secondary to the central dogma of amyloid cascade hypotheses.
In 1995, we began a series of experiments examining expressed brain proteins for pathway analysis using 2-dimensional gel electrophoresis of APOE knock-out and human allele-specific genomic APOE transgenic mice (reviewed in Saunders et al. 2000). We quickly identified and confirmed that multiple glucose metabolic enzymes were coordinately affected (increased or decreased expression]) in APOE KO and transgenic mice. A seminal PET study of glucose utilization was published by Eric Reiman's laboratory in 1996 (shortly after the birth of Maija Diane Roses). This study also noted differences in glucose utilization in AD but showed that APOE4/4 normal subjects (averaging 50 years or about 20 years before the mean age of onset of AD for
Apolipoprotein E and Alzheimer's disease: A brief retrospective 285
that genotype) metabolized glucose less than APOE3/3 subjects of the same age range (or about 30-40 years younger than the mean age of onset for the APOE3/3 genotype; Reiman etal. 1996).
Our protein pathway experiments had been supported by a research collaboration with GlaxoWellcome (GW). I left Duke University in 1997 to set up "genetics" of common diseases within the company, but also with the opportunity to pursue the APOE and AD data. The APOE Team very quickly demonstrated an APOE-specific effect on brain glucose utilization from transgenic and KO mice with several PPARy agonists that were being developed for treatment of type 2 diabetes mellitus, or had been developed by other companies. By late 1999, however, the focus of GW was on merger with SmithKlineBeecham (SB), and almost a year was lost from these experiments during the process of merger until GSK was formed. SB had marketed a PPARy agonist and this molecule was chosen over an ex-GW compound that was still in clinical trials at the time of the merger. We were then free to use the ex-GW compound in these experiments, but we had already previously tested the ex-SB compound. (In those ancient days, a company tested competitor compounds but avoided testing their own -no longer the practice!) The data were strong that several PPARy agonists increased glucose utilization. Rosiglitazone (the ex-SB compound) was on the market and has been used in more than 2 million diabetes mellitus patients. A small Phase IIA clinical proof of concept trial was supported using rosiglitazone in the laboratory of Susan Craft. Using the results of this study, which suggested some clinical improvement in patients who were APOE3/3 and had not inherited an APOE4 allele, a formal Phase IIB trial was begun. This monotherapy trial prospectively designated APOE genotype as a biomarker for rosiglitazone efficacy in the clinical protocol. It was initiated in late 2003 and completed in mid-2005. The results of the study were remarkable from several points of view (Risner et al. 2006).
AD patients who had not been exposed to other AD therapies were recruited into a 24-week monotherapy trial using three doses of rosiglitazone, including two doses that were lower than the therapeutic dose for T2DM. A total of 511 patients were in the intention-to-treat group and, when the clinical parameters (ADAS-cog and multiple other measures) were analyzed, there was no statistically significant effect of the drug at any dose. Following this analysis, the APOE genotype data were assigned. All three doses of rosiglitazone decreased the ADAS-cog (increased function), as well as multiple secondary measures compared to placebo in patients who did not carry an APOE4 allele. The placebo group for the APOE4-positive patients made that arm of the study difficult to interpret but, if only the rosiglitazone-treated patients were analyzed, there appeared to be a dose effect with some improvement at higher doses in the APOE4-positive patients. This finding suggests that APOE genotyping might be used to select dosage for AD patients. These pharmacogenetic analyses were sufficient to design and execute a Phase III registration program.
With the consultation process of the FDA called (Voluntary Genomic Data Submission (VGDS), plans were put forward for APOE genotype-specific trial designs that would not only test patients who carried no APOE4 allele but would also look for the need for larger doses in patients with one or two APOE4 alleles. This registration program, for both adjuvant and monotherapy treatments, was initiated in the summer of2006.
A great deal of data supporting a novel mechanism of action data has been accumulated during the past three years. The Gladstone Institute is the premier institution for APOE research, dating back to the 1980s with the work of Robert Mahley and Karl Weisgraber. Recently, Yadong Huang's laboratory has concentrated on abnormal mitochondrial function in the presence of proteolytic fragments of apoE4 protein. Two recent papers have reviewed these data (Roses and Saunders 2006). In fact, if the data supporting mitochondrial proliferation had been generated before the existing peroxisome studies, PPARy agonists would probably be known as "MPARy agonists!" A strong rationale exists for relating the apoE4 > apoE3 effects on neuronal sprouting and maintenance of connections on the difference in the protein structure of apoE4 compared to apoE3, leading to increased cleavage to produce the apoE1-272 fragment from apoE4. This fragment acts as a slow toxin to mitochondria, diminishing their dynamic movements, speed and distance traveled within the confining architecture of neurons. Mitochondria movement to the base of neuronal spouts is diminished so markedly that, over time, there is a decreased rate of energy-dependent dendritic plasticity, leading to simplified dendritic trees, decreased connectivity, and increased scarring with the accumulation of amyloid and other secondary proteins in the areas vacated by neurites. It is remarkable that the pathological data were referred to by Ramón y Cajal but were demonstrated using rapidly autopsied brain in 1994 (Ramón y Cajal 1906; Einstein et al. 1994). Rosiglitazone rapidly increases mi-tochondrial proliferation so mitochondria that have not been exposed to the apoE fragment increase effective energy metabolism function. Time will tell whether the functional improvement of symptomatic patients can be maintained by rosiglitazone. Once registered as a medicine, epidemiologic studies can be designed to study disease modification and prevention in at-risk subjects whose expected ages of onset are based on differential APOE genotype-specific risks.
That's where we are, 100 years after Alzheimer's first paper. Planning for the success of the current clinical trials, new discovery projects are in progress using functional assays developed around this novel mechanism of therapy. At this time, the research is mostly limited to GSK and the Gladstone Institut, and our collaborators. Should rosiglitazone achieve registration as a new chemical entity for the treatment of AD, future directions for neurodegenerative disease research may be established.
Was this article helpful?