AT A GLANCE
Daniel Geschwind, assistant professor of neurology, UCLA
MD, PhD in neurobiology, Yale University: Identified proteins involved in early neuronal development using 2D-gel electrophoresis
Neurology residency, UCLA School of Medicine
Co-edited book entitled Microarrays for the Neurosciences: An Essential Guide, published in 2002 by The MIT Press
What are your research interests, and what role do microarrays play in your research?
We are interested in determining how genes lead to brain structures which underlie human cognitive phenotypes. For example, we are interested in the development of language, and that relies on brain asymmetry formed during embryonic development. So we have been using microarrays and other genetic technologies to look at gene expression in the two hemispheres of the brain. We have also used this [technology] to study other issues in developmental neurobiology, namely looking for markers for neuronal stem cells, and that work has been rather successful. So far we have identified a larger number of genes involved in neural stem cell proliferation and differentiation and compared neural stem cell populations to each other and to other stem cell populations, like hematopoietic stem cells, in collaboration with other groups.
What types of microarrays have you been using for these experiments?
For the laterality studies in humans we have used Affymetrix arrays, for other studies we have made custom arrays, where we have done a partial subtraction or enrichment for clones of interest and made arrays from these subtracted libraries. We have used the Incyte 9K mouse array: We print all the arrays ourselves in the UCLA core facility, but the clone set is from Incyte and clones that we have picked. We have also made a list of 6,500 human genes of great importance to brain development and function and are printing a 4,500-clone set based on that logically and empirically derived list.
Do you also use oligo arrays?
We are printing oligo arrays currently and trying to assess how they perform vs. cDNA arrays. Currently nobody has really done a published study that shows that custom-printed oligo arrays work as well as cDNA arrays. Nobody has shown to the contrary either. It’s clear that the Agilent oligo arrays work quite well, but there is a difference in performance between those and the printed arrays. Until we have proof that they work as well, we will stick to the cDNA arrays, even though they are more difficult to make.
Your book is entitled Microarrays for the Neurosciences. So what are the specific issues of applying microarrays in that field?
We focus on some issues that plague the neuroscientist specifically, and that is tissue heterogeneity and very low abundance RNAs. And this, coupled with a huge number of cell classes in a given neural tissue can create a number of issues when doing microarray experiments. One of these is detection. For example, a gene may go down tenfold in one cell that represents five percent of the tissue and go up two- or threefold in the remaining tissue. We have a discussion of tissue heterogeneity and methods for getting around that using subtractions coupled to microarrays to enrich for species of interest, [as well as how to do] microdissections, and amplification of RNA from single cells or cell populations. Also, a lot of neuroscience is focused on neurobiology of disease. If you take diseases like schizophrenia and autism, there aren’t good animal models, so one needs to study post-mortem human tissue.
What are the main issues with using post-mortem tissues?
One has to be very careful that your n is large enough to control for variability, and that you have done your best to ethnically match populations. Number two would be RNA quality. Even if the post-mortem interval was very short, the brain tissue may have suffered quite a bit of anoxic damage, and RNA may be quite degraded. The cause of death [and] the post-mortem interval will cause changes in gene expression that aren’t related to the underlying condition of interest. Also, in many cases, high-quality frozen tissue is often not available in the quantities necessary. One of the issues is, can one use fixed tissue in any ways for these studies?
Where do you see future developments?
[An] issue that would be really interesting is the next generation kind of array, arrays that are focused at picking up all the different splice variants and isoforms of genes. And I think in the brain, this is going to be particularly important. We have been studying a few genes that have many isoforms and some very minor isoforms that seem to be very enriched in the brain relative to other tissues, and of course our current arrays would never pick this up. This is something that the oligonucleotide arrays are especially well suited for. One of the issues is, some of these exons that are spliced in and out can be quite small, 30 base pairs, 90 base pairs.
Currently, if you are going to design two oligos per gene, you have a lot of choice, and therefore you can optimize an array, so that each oligo has essentially almost identical hybridization conditions. This is going to be a lot more difficult when looking at splicing because people will be limited to the small sequence in a particular exon. In that case it may be that the smaller oligonucleotide arrays with smaller overlapping oligos actually work better.
Where do you see the greatest potential for the use of microarrays in the neurosciences? Also, do you think they will be used outside of basic research anytime soon?
In basic research, microarrays have the possibility for us to collect and curate a molecular anatomy of the nervous system on a cell-by-cell basis. I think that [it] would be pretty extraordinary to understand how gene expression leads to particular specific neuronal phenotypes. That’s a fundamental question in developmental neurobiology, and of course will have importance for disease. Plus, we would also like to understand individual variability in gene expression [and] how the environment leads to differences in gene expression.
In terms of where this may go in the clinic, there were several very intriguing papers recently published — one from Frank Sharp’s group, where they looked at rats that had four different acute injuries; seizures, stroke, hypoglycemia, and hypoxic injury. [Ann Neurol, 50(6), p. 699-707]. What was fascinating is that 24 hours after the insult, the peripheral lymphocyte gene expression was remarkably different for each of those conditions. If you take that one step further, it may allow us to say that neurodegenerative diseases like Alzheimer’s disease also have peripheral blood fingerprints. It’s possible that we will be able to pick that up in lymphocyte gene expression.
Probably the advance that’s closest around the corner is using arrays in the clinic for tumor prognosis and treatment protocols. But beyond that in neurology, it certainly might help you understand who is going to be a responder and non-responder to a particular treatment. It may help you decide who is going have side effects to a particular medicine or not. Some medications have been taken off the market that are very good anti-seizure drugs. If we could just identify those people who we shouldn’t give [a drug] to, and those it would save, that would be quite helpful, because the medicines are quite effective. It could be that this is unwarranted optimism, but that study in the rat leads me to believe that many CNS diseases, especially acute diseases, will have particular fingerprints in peripheral blood, that will help in diagnosis, maybe prognosis, and then in choice of therapy.
One other thing with the nervous system that’s really important to realize is that gene expression is obviously just one part of the equation, but in the nervous system, posttranslational modifications play an enormous role in development and maintenance of structures, especially glycosylation, myristylation, et cetera. Tour de force methods to study that will be really great to couple with microarray studies in the future.