Graham Watson's Paper Review

This web page was produced as an assignment for an undergraduate course at Davidson College

The paper to be reviewed:

Analysis of the eye development pathway in Drosophila using DNA microarrays

Lydia Michaut, Susanne Flister, Martin Neeb, Kevin P. White, Ulrich Certa; and Walter J. Gehring

Biozentrum, University of Basel, CH-4056 Basel, Switzerland; Roche Bioinformatics and Roche Center for Medical Genomics, F. Hoffmann-La Roche, CH-4070 Basel, Switzerland; and Yale University School of Medicine, New Haven, CT 06520

PNAS, Volume 100, No. 7, April 1, 2003, p. 4024-4029

This goal of this paper is to investigate and explore the cascade of genes that are involved in eye development in Drosophila. The researchers realized that Pax-6 genes, which encode two DNA-binding domains, are able to induce all of the genes required for eye development. In fact, the Pax-6 genes are so influential that they can be applied to almost any part of a developing fly ectopically, and an eye will develop in that region. Eye development was demonstrated on leg, antennae, and wing in Drosophila as a result of the ectopic placement of a Pax-6 gene (Halder et al., 1995). In this study, Michaut et al. determined a set of genes that are involved in eye development by comparing the genes that are up-regulated when ey, a Pax-6 gene, is ectopically placed on a leg disc, and genes that are normally expressed in eye discs. To validate these results, an analysis of endogenous expression of selected genes in the eye disc was performed. They used two distinct Drosophila full-genome high-density oligonucleotide arrays, roDROMEGa and DrosGenome1, that were developed independently to insure the validity of their data.

The introduction starts with an appropriate explanation of Pax-6 genes and their important role in eye development. The researchers smartly pointed out that this pathway is highly conserved and is seen in both vertebrates and invertebrates, citing that Pax-6 genes trigger vertebrate camera-type eye development as well as insect eye development. Since the model organism here is Drosophila, an invertebrate, the conserved nature of this process in more complex organisms increases the paper’s significance.

The researchers contextualized the research with regard to stages of development. The process of retinal differentiation is described in some detail and the role and timing of eyeless (ey) in this part of development is explained. The developmental stage in which the cells start differentiating into photoreceptors is also analyzed. Ey encodes one of the two Pax-6 genes which can induce the expression of all the genes required for eye development. The key to this procedure is that this induction can occur when ey is ectopically expressed in other imaginal discs. The set of 371 genes that are involved in eye development was determined by comparing the genes expressed in a normal eye discs to those up-regulated when ey was ectopically expressed in the leg. Because the oligonucleotide sequences chosen as probes in the microarray analysis can effect the results, two different arrays were done. Some expected genes were identified, such as those that are downstream of the ey gene, but many genes that were previously uncharacterized seem to be involved in eye development.

The introduction was written succinctly and described all of the pertinent information. Although knowing the stage and details of eye development is not necessary to understanding the rest of the paper, it is important to understand exactly when and where the experiments take place on the fly. At this point, I am convinced that this is a valid way to approach this research objective. The use of two independently developed microarrays does, as they claim, “significantly increase the validity of our results” because different oligonucleotide probes were used and found these 55 genes in common. It would be more convincing, however, if the two data sets had a larger amount of genes in common. There will likely be some overlap by chance and so a higher correlation would have further proved that the genes were in fact involved in retinal development.

Materials and Methods
In the Materials and Methods section, four different processes are described. In the DNA microarray section the authors point out that two different microarrays were performed and that different oligonucleotide probes were used for each. Further, there are different amount of splice variants represented on the two different microarray analysis. This may not be the best method for detailing a comprehensive list of genes involved in eye development although it seems sufficient for finding new genes, never before shown to be involved.

Results and Discussion
Screen Design and Strategy of Identification of Genes Expressed Downstream of ey During Drosophila Eye Development.
This section begins by describing how flies engineered to produce eyes larger than normal were used in this experiment to magnify the effects of gene expression during eye development. Next, the process by which eye-specific genes were determined was outlined. Ey is a transcription factor that is active in several tissues during Drosophila development. In order to recognize which genes are up-regulated due to involvement in eye development and which genes are not, the researchers compared the set of genes ectopically induced by ey and normal gene expression in eye imaginal disc using microarrays. While this is a logical way to determine which genes are truly involved in eye development, the case could be made stronger. If they could have proved some sort of dysfunction in eye development with an absence of one or a group of these genes, it would be more convincing. Granted this would be time consuming and difficult but it would prove that the genes named were not coincidentally expressed in ey induced and normal eye development.

Discrepancies Between the Two Microarrays.
This section describes the differences between the two microarrays and also goes into some discussion regarding the differences between the ey induced genes and those that are expressed in normal eye discs. Only 40% of the genes were both ectopically induced by ey in the leg discs and transcribed in the eye disc. The researchers attempt to claim that this fact supports the idea that a comparison was necessary and that when coupled with biological replicated is a “strong validation of the data.” However, this low percentage weakens their case that the genes they selected are in fact involved in retinal development. A higher percentage would have proven that there are not a lot of other genes being detected and that there is less reason to assume coincidence of a mistaken overlap.

The researcher also note that the 61% of genes that were detected by roDROMEGa were also detected by DrosGenome1 and conversely, 65.5% of genes detected by DrosGenome1 were detected by roDROMEGa. They state that this low percentage is likely a result of the varying oligonucleotide probes. While this is on plausible explanation it is certainly not something that can be claimed with any sort of certainty. Trying to cover up for the low percentage of overlap by claiming that they were “probably” a result of the probes only draws attention to the inconsistency.

This section then moves into describing some of the results seen in specific genes with regard to the differences between the two microarrays. While these examples are a good way to describe some of the effects that were seen, there is not a lot of detailed analysis given. If the goal of the paper is to gain an overview of the cascade a more specific look at some interactions may be called for. Granted, they are dealing with a large number of genes and so looking more generally is not completely detrimental to their efforts. They also have a great deal of tables that are not shown that may support some of their claims. I understand that these were probably omitted to fit the article into a certain space, and it is good that they refer the reader to the tables online.

Figure 1A is an effective visual aid to accurately describe the amount of overlap that was seen between the two microarray analyses. It does however highlight the seemingly week data. Maybe it would be worth highlighting had the overlap been greater. Figure 1B is also effective at describing the roles that some of these genes play. However, if they are only looking at overlaps from the two microarrays because it is more valid, why do they spend time listing the other genes on this graph? These genes are admittedly not necessarily involved in eye development.

Table 1 is also a succinct display of only the 55 genes that have more convincingly shown to be involved eye development. The color coding is effective at showing which genes are more induced that others although there is a discrepancy between the key and the table. The yellow shading is supposed to show IF < 1.5 but several of the values are above 1.5. They likely meant < 2.0. They did a good job of trying to group the genes in row based on a combination of function and level of induction. This type of data is what can be used to draw larger conclusion although the researchers to not offer any at this juncture.

The ey-induced Genes Function High up in the Retinal Differentiation Pathway.
This section brings up the important observation that although some genes are not up-regulated in ectopic ey. The fact that they may also be involved in other processes in the leg would cause their expression pattern to remain the same while really they are involved in both. This is a result of the relative nature of DNA microarrays, which compare two expression patterns and do not analyze one on its own. The researchers do a good job here of offering a legitimate explanation as to why some genes that may in fact be at the top of the retinal differentiation pathway may not have the same types of percentages, shown in Table 2, as it would seem they should. They site dac as a pertinent example.

Known Genes Not Yet Associated with Eye Formation.
This section essentially makes some speculation with regard to the genes that are not yet known to be involved in eye development. As much as is possible, the data seems to support the authors’ suggestions that certain genes such as, lola and sequoia, may play a role in eye development because of the transcription in the eye disc in conjunction with up-regulation in ectopically induced eye development. As is often the case with microarray data, however, the researchers can only suggest function and propose that further research be done to investigate the role of certain genes in this pathway. No conclusive data is provided. I am convinced, however, that a closer look is necessary for many of these genes.

Figure 2, which displays the results from an in situ hybridization experiment, shows endogenous expression of the ey induced genes in wild type eye discs. This figure serves to strengthen the case that some of these genes really are involved in eye development. The microarrays have shown that they are transcribed in a normal eye disc as well as when eye development is induced by ey. This in situ figure further shows that these genes are expressed in the right place as well as the right time during development to be involved in the way that the researchers hypothesized. Again, this does not prove anything, but suggests more strongly the role of these genes (Sur-8, sprint, and APPL-interacting protein).

The researchers suggest that some of the proteins such as QUAIL, which are shown in Figure 2 to be located in the eye discs posterior to the morphogenetic furrow, are involved in cell shape and actin bundle assembly (QUAIL). Although these proteins have not been proven to serve these functions in eye development specifically the implication remains. It is fair to hypothesize that these proteins serve the same function here as they do in other tissues although further testing is needed.

Previously Uncharacterized Genes Expressed During Eye Development.
It is clear that several previously uncharacterized genes have been identified by this study as having possible roles in eye development. While no conclusions can be drawn by the researchers the likelihood of certain scenarios is made more believable by analyzing the roles that similar genes play in other areas of the organism. The authors do not overstep their bounds, but do a nice job of incorporating a good deal of research and comprehensive knowledge of the genes in question to make valid claims with regard to specific genes.

In the conclusion, the authors do a nice job of making general statements about their project and, in the last sentence, relating their work back to the larger picture of analyzing camera-type eyes in mammals.

General Thoughts
Although some unfounded claims were made early on in this paper, I think the authors did a fine job of using microarrays to get a clearer picture of the complex process of eye development. The use of two independent microarray’s with different oligonucleotide probes did not increase the validity of the results as much as the authors would like to claim that it did, but there are still general conclusions that can be drawn from the data. More importantly, future researchers will be able to formulate focused hypotheses to solidify the roles of some otherwise uncharacterized genes.

Thoughts for Future Research
The most important thing that this paper accomplished is opening the door for further investigation into the specific interactions and processes that occur every day in the development of the Drosophila eye. The researchers used microarrays effectively and we can now design focused experiments that will uncover the true interactions that occur. Scientists are no longer tied to the point and shoot methods, known as reverse genetics. The possible follow-up experiments are as abundant as the number of genes that have been implicated by this paper. One possible approach would be to test knockouts for loss of function to determine that certain genes are not just involved but are necessary to eye development.

One issue with this experiment was the amount of overlap that was seen between genes that are expressed in the developing eye and genes that are expressed in the developing leg, the organ where ey was ectopically expressed. If there was some way to compare the expression of normal eye disc with another organ that does not have as much similarity in its transcription factors it would be worth exploring.

A natural next step for research in this area to take would be to defining specific pathways in the development of the eye in this and other organisms. Specific and detailed analysis of protein shape may be one way to determine which proteins are interacting and on what level. There are several tools including chime imagery that can help to determine information regarding protein-protein interactions.

Sticking with the theme of microarrays, there are some other interesting microarray experiments that could be done that would provide further insight into the pathway in question. Here, the researchers compared normal leg development to leg development that was ectopically induced with ey. It would be beneficial to compare a microarray from a normal eye with a microarray from eyes that have been exposed to drugs that inhibit different types of processes such as ion channels or enzyme interactions. This may provide insight which gene products are involved in which types of cell functions.

Also, similar experiments could be performed on other organisms so that the claims made about the conserved nature of Pax-6 genes could be confirmed. Mice might be a good place to start, although a different approach may have to be taken because the same ectopic expression may not be possible.



Halder, G. et al. (1995). Induction of Ectopic Eyes by Targeted Expression of the eyeless Gene in Drosophila. Science. 267.


This page was designed by Graham Watson, a senior Biology major at Davidson College.

Click here to go back to my Home Page

Molecular Biology Home Page


© Copyright 2003 Department of Biology, Davidson College, Davidson, NC 28035

Send question to