The use of gene expression profiling to characterize Alzheimer disease pathology has, unsurprisingly, focused on changes in brain tissue (see ARF related news story), but the search for biomarkers that might allow early diagnosis is moving outside of that anatomical box. Reaching for a more easily attainable tissue, researchers from Sumitomo Pharmaceuticals in Osaka, Japan, and the Karolinska Institute in Sweden chose fibroblasts as the starting point for comparing global gene expression patterns between people with FAD mutations and their wild-type siblings. Their results show a clear gene expression signature that can distinguish FAD gene carriers from non-carriers long before signs of dementia appear. Interestingly, the changes in gene expression caused by three different FAD genes (the Swedish and Arctic APP mutations and PSEN1 H163Y) were all similar, suggesting that mutations in either APP or PS1 can cause common changes in the physiology of cells outside the brain long before clinical disease sets in. The research appears online in the September 29 PNAS Early Edition.
For the study, first author Yosuke Nagasaka and colleagues probed the genomewide expression of cultured fibroblasts from skin biopsies. The investigators found 56 genes that were most highly differentially expressed and 200 that showed smaller but significant differences. The gene expression profile using 200 genes predicted FAD status with 97 percent accuracy, regardless of whether the subjects displayed signs of dementia or not. Other factors, such as age, ApoE status, or gender did not seem to contribute to the difference in gene expression observed.
The authors did not list the identities of the differentially expressed genes in the paper but note they will make them available on request. The Alzheimer Research Forum has made such a request. The authors characterize the significance of their finding as “a unique gene expression signature for FAD caused by three different mutations in two different genes…that…can be detected in fibroblasts, which may seem to be an organ completely unrelated to the tissue affected by the disease.” Within the 56 or 200 genes may lay smaller sets or even individual genes that represent potential biomarkers for early AD.
The elephant on the chip is the open question of whether the gene signature exposed in FAD fibroblasts will translate in any way to sporadic AD. The similarities of the profiles with three mutations suggest that underlying changes in gene expression might be central to all types of AD. But fibroblasts express APP and PS1 proteins, putting them directly in line for FAD-related effects. The question of whether similar alterations, or any at all, occur in fibroblasts in sporadic AD will be of considerable interest for the development of surrogate markers for brain pathology.—Pat McCaffrey
Updated 5 October 2005:
Q&A with Toru Kimura and Caroline Graff. Questions by Pat McCaffrey.
Q: Of possible tissues, why did you choose fibroblasts for your analysis?
A: We agree that the much more common and easy procedure of sampling peripheral blood would make lymphocyte or lymphoblast studies preferable. Actually, in parallel with the fibroblast biopsies, peripheral lymphocytes (blood) were sampled from the same family members, followed by microarray hybridizations. Unfortunately, the inter-individual variation was very high and in some cases (more often than in the fibroblasts), the RNA quality was too poor and we were thus unable to interpret the hybridization signals. Actually, we experience from this study that skin biopsies appear to be less sensitive to differences in external conditions directly after sampling as compared to lymphocytes, which require direct handling after sampling.
Besides the bad RNA quality of our lymphocyte samples, there are several other reasons why fibroblasts are more attractive than lymphoblasts. First, we would like to challenge the appropriateness of using lymphoblasts since these cell lines are established after immortalization, typically with Epstein-Barr virus transformation. This in itself changes the genetic make-up of the cells, leading to uncontrolled changes in the genome which theoretically may lead to spurious changes in gene expression. Second, the use of RNA from peripheral blood lymphocytes may be an attractive alternative, but these cells are very reactive to acute stimuli such as nutritional status, fever, infections, and drug treatment; this makes them less attractive. Moreover, if it were possible to make a presymptomatic diagnosis using fibroblasts and there were drugs which could delay the onset of symptoms, we believe that a skin biopsy would be tolerated and requested by most patients. For example, many muscle dystrophies and myopathies can only be diagnosed based on results from muscle biopsies which are routinely sampled on patients for this purpose.
Q: Your gene expression profiles clearly distinguish FAD mutation carriers from wild-type siblings, but what about sporadic AD? Do you expect to see the same differences in the absence of a causative mutation?
A: Naturally we are interested in investigating the gene expression profile in patients with the common, sporadic forms of AD. It can be anticipated that there will be shared changes on the gene expression level between sporadic and familial AD since the diseases are clinically indistinguishable except for the age at onset and the family history. However, it is also plausible that the gene signature we have identified is related to the biochemical pathways perturbed by the specific FAD mutations included in this study. If this is true, we may find a similar profile in other AD-causing APP, PSEN1, and PSEN2 mutations. Therefore our next step will be to characterize family members with other FAD-causing mutations. If we can validate the gene signature in additional FAD mutation carriers, it may be possible to use the signature in order to identify sporadic AD patients who share the same gene expression profile. That is, the heterogeneous nature of sporadic AD suggests that the etiology is also heterogeneous. This makes it unlikely that all sporadic AD patients will share the same gene expression profile. However, such gene expression classification may serve as a tool to subcategorize the disease etiologies in the common forms of the disease.
Q: Your paper doesn't talk about which genes were affected by the mutations. Did these fall into any interesting classes, for example, cell cycle genes, or other groups?
A: Actually, the 200 differentially expressed genes were subjected to functional classification based on their known functions using the FatiGO program. FatiGO is a Web interface which carries out simple data mining using Gene Ontology for DNA microarray data. The FatiGO results showed that effects were seen in virtually all biological processes. The three largest functionally categorized groups are those of metabolism, cellular growth, and/or maintenance, as well as cell communication.
Q: Do your results suggest any good candidate genes for stand-alone analysis as biomarkers?
A: We intend to follow up the study in order to validate our findings and, if possible, reduce the number of informative genes. At this point it is unlikely to expect a single gene or a handful of genes to be sufficient. However, it may be true that some of the gene products could be used as biomarkers to categorize the heterogeneous sporadic AD patients into more homogeneous subgroups.
Q: What happens next to validate or further develop the gene expression profiles as diagnostic tool? Will you be pursuing that, and if so, how?
A: Yes, we are willing to and going to pursue this approach of gene expression analysis of RNA from fibroblasts. First, we are planning to analyze more FAD samples, which are independent from the 30 samples we analyzed in this study, in order to see if the same expression differences can be observed. Then we will assess if the expression differences can be observed in sporadic AD. We consider this study as exploratory, and our ambition is to make a large and carefully performed validation study with additional samples. We are continuously collecting samples from these rare FAD mutation families; however, it is a very slow and time-consuming process, and we hope that the data from this paper will encourage further collaboration and perhaps thereby also speed up the follow-up validation. Beside the collection of FAD mutation families, we are planning to validate the gene signature in sporadic AD patients, as well as validation with respect to other neuropathological conditions.
- de Leon MJ, Desanti S, Zinkowski R, Mehta PD, Pratico D, Segal S, Rusinek H, Li J, Tsui W, Saint Louis LA, Clark CM, Tarshish C, Li Y, Lair L, Javier E, Rich K, Lesbre P, Mosconi L, Reisberg B, Sadowski M, DeBernadis JF, Kerkman DJ, Hampel H, Wahlund LO, Davies P. Longitudinal CSF and MRI biomarkers improve the diagnosis of mild cognitive impairment. Neurobiol Aging. 2006 Mar;27(3):394-401. PubMed.
- Blalock EM, Geddes JW, Chen KC, Porter NM, Markesbery WR, Landfield PW. Incipient Alzheimer's disease: microarray correlation analyses reveal major transcriptional and tumor suppressor responses. Proc Natl Acad Sci U S A. 2004 Feb 17;101(7):2173-8. PubMed.
- Nagasaka Y, Dillner K, Ebise H, Teramoto R, Nakagawa H, Lilius L, Axelman K, Forsell C, Ito A, Winblad B, Kimura T, Graff C. A unique gene expression signature discriminates familial Alzheimer's disease mutation carriers from their wild-type siblings. Proc Natl Acad Sci U S A. 2005 Oct 11;102(41):14854-9. PubMed.