Yu J, Othman MI, Farjo R et al 2002 Evaluation and optimization of procedures for target labeling and hybridization of cDNA microarrays. Mol Vis 8:130^137

DISCUSSION

Swaroop: Initially, when you and I began studying the regulation of rhodopsin we identi¢ed NRL, then CRX. Since then, many transcription factors and their interacting proteins (e.g. RX, QRX, BAF and FIZ) have been implicated. I’m sure there are many more factors and interacting proteins that will soon be found. How do we go about looking for the in vivo relevance of all of these interactions?

Zack: My dream is to be able to take a non-photoreceptor cell such as a 293 cell, and transfect a subset of factors so that we could turn on the endogenous rhodopsin gene that is not normally expressed. It is not clear whether this is theoretically possible. This is one approach: to ¢nd the subset of proteins that will allow that. Another avenue that might prove fruitful is chip^chip assays, which involves bringing down the chromosomes with an antibody against a speci¢c transcription factor, and then you ¢gure out where they are binding the chromosome.

Swaroop: Unless there is a master gene that turns all the genes on, it will be hard to take 15 di¡erent genes and put them in a cell!

Zack: We could express perhaps half a dozen genes in 293 cells and get expression.

Thompson: Some of them may already be there. Zack: The more interesting ones won’t be.

Daiger: One of the interests in genetics is applying computational methods to detect binding sites for expression factors. This runs into the problem that most of the sites are 6 or 8 nucleotides, hence they are found throughout the genome. But if you look at these as a cluster of required sites for this whole set of complementary expression factors, can you have more success at a computational level in ¢nding the regulatory regions around rhodopsin or the other opsins?

Zack: I hope so. When we initially compared rhodopsins from mouse, human and bovine, we were fortunate that the RER locus control region came out of a sequence comparison. We have tried doing this a lot since then, but it hasn’t worked at all. Microarrays are likely to be an e¡ective way to address this because we can use them to identify over 200 expressed genes in the retina. After we get these putative sites the question then remains about how we test them and show that they are important in vivo.

Farber: One of ways to address the issue Anand Swaroop raised is ex vivo transfection of Xenopus embryos to see whether you get the same results as transfecting cell lines. We have been doing this in collaboration with Barry Knox. Once we have ex vivo transfections we can do transgenics and see what happens there.

Zack: Any of these systems deals with basically one gene at a time, but we want to be able to address these issues at a multifactorial, complex level. We want to integrate the modi¢er genes so we can understand the complex matrix of interactions.

Swaroop: The issue is one of evaluating the critical trans-factors for regulation in vivo, out of the many that we know of. One way to do this might be in vivo footprinting.

Zack: That is a di¡erent method, but it is basically looking at the same thing as chromatin immunoprecipitation.

Daiger: What are the sizes of the families in which you found the pathogenic variants?

Zack: Unfortunately they are very small . they are singletons. Of the three, one person had a brother in India who could not be found. It has been very hard.

Thompson: I would like to ask about your microarray work. You have 10 000 genes on your array. What is the redundancy? Are there 10 000 unique sequences on your slide? How di¡erent are your slides from the cDNAs that other people have analysed, and does this relate to why di¡erent groups have seen di¡erent things with array methods?

Zack: There is about 5% redundancy. One approach we have that is di¡erent from what Anand Swaroop and you have done is that all the spots on your slides have been sequenced in house whereas we got sequences from many di¡erent places. We didn’t have the resources to sequence all 10 000 again. Our approach is to identify the interesting di¡erentially expressed sequences, and then we go back to that spot and sequence it. We think this is a good way of correcting errors in the original sequences. In terms of why di¡erent groups see di¡erent things, we have mentioned the di¡erences between the genes on the arrays. I am sure there is signi¢cant overlap as well as signi¢cant non-overlap.