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

Ashley's Paper Review:
Analysis of the eye developmental pathway in Drosophila using DNA microarrays
by: Lydia Michaut, Susanne Flister, Martin Neeb, Kevin P. White, Ulrich Certa, and Walter Gehring.


In this paper, the authors intend to identify the genes involved in the Drosophila eye morphogenisis, specifically those under the control of the eyeless (ey) gene. To do this, they used two DNA full-genome Drosophila microarrays to compare expression in an imaginal leg disk ectopically expressing eye growth and a wild-type leg disk. This comparison revealed a group of genes associated with eye development under the influence of the ey gene. To ensure that the expressed genes were involved in eye development rather than leg development (as ey has been shown to initiate a broad range of developmental functions), the authors compared the gene set from the leg experiments to genes that are expressed in normal Drosophila eye development.
The authors identified 371 genes that are both expressed in normal Drosophila eye development and are up-regulated in tissues when eyes are ectopically expressed. They identified a number of genes already known to act in Drosophila eye development, but also a number of genes previously uncharacterized. Most of the genes they found encoded transcription factors, signal transducers, actin binding proteins, cell adhesion molecules, and proteins involved in cell division.

Methods and Experimental Design

One of the strengths of this paper was the experimental design. The authors tried to ensure that the results they got were not ambiguous by using two different DNA assays to compare gene expression in Drosophila legs with ectopic eyes and wild type Drosophila legs, then checked those results against gene expression in normal Drosophila eye growth to ensure that the genes they identified were truly involved with eye expression.

Method Summary: RNA was extracted from the target tissues (imaginal disks in wild type Drosophila leg and Drosophilaleg induced with ey to express ectopic eyes, or primordal tissue of a normal Drosophila eye), and used to produce cDNA that was hybridized to the assays. The cDNA was hybridized first to one assay and then to the other, but an excess of cDNA was used and the signal was strong even after three such hybridizations.
Two high-density oligonucleotide assays of the full Drosophila genome were used: roDROMEGa and DrosGenome1. The assays are based on different genome sequences (DrosGenome1 is based on a later release) and contain different nucleotide probes. Thus some genes were detected by only one assay, and some genes were detected multiple times by each. Expression had to be at least 1.5 times higher (confidence level 95%) in ectopic eye tissues than in wild type tissues in order for the gene to be categorized as "expressed."

The difference in detection between the two assays indicates that neither assay is complete in itself, and perhaps there are more genes that neither assay was able to detect. The authors are very forward with this shortcoming with assay methods and tried to correct for it by using two different assays.


Results Summary: The authors discussed three kinds of genes that they identified: a number of genes that are involved in eye development but act parallel or upstream of ey and thus are not significantly up-regulated by ey (such as toy or optix), 18 genes known to be downstream of ey that were significantly up-regulated by ey (such as atonal or rough), and 20 genes that had previously not been identified with eye development but were identified as such in this paper. Eight of these previously unidentified genes had been known to be involved in development.

Figures and Tables:

Fig. 1A is a schematic diagram showing the number of genes identified by each assay and the genes that were identified by both. Out of 371 genes identified by one or the other assay, only 55 were detected by both.

Fig. 1B shows the functional classes of each of the identified genes for which a function could be determined. Transcription factors and other enzymes are the most prevalent.

Table 1 displayed the 55 genes identified by both assays. It lists gene name, function if known, and the signal strength (fold induction). The cut-off fold induction determining expression was 1.5, and the values of fold inductions displayed range from 1.6 to 71.5. A number of the values were very close to the cut-off value, which raises the issue of determining cut-off values. It is often difficult to know where they should fall, and as the cut-off value determines contents of the data set, it is important to assign it appropriately. If the authors chose an inappropriate cut-off value, it is possible that they are claiming certain genes are expressed when they are not, or vice-versa. However, the authors said they were following a standard statistical procedure, and as I have little statistical analysis background, I choose to accept their cut-off value.

Table 2 reports the array values for some of the genes discussed in the text. Again, it is important to understand that cut-off values, confidence levels, etc, affect the data set that is chosen, but the authors claim to be adhering to standard statistical procedures.

Fig. 2 shows panels of ey-induced genes in wild-type disks. Each disk has been probed with an RNA probe specific for a certain gene identified in the paper. The genes probed are ken, ske, Sur-8, aplip 1, sprint, CG13532, gh11973, CG9134, SP1173, gh11415, CG12605, CG11849, and CG13651. Each disk displays some sort of signal, though the signal for CG12605 is very weak while the signal for gh11415 is quite strong.

Future Research

Future research could fall under two categories: repeating this experiment for more complete or more accurate data, and using this experiment as a jump-off point to explore more about eye development and eyeless.

Though the authors try to account for the potential error in the microassay method by using two different microassays, it is still possible that there were genes neither assay identified. Also, out of 371 identified genes, only 55 were detected by both assays. In addition, a few of the genes the authors said were “expressed” (i.e., were above the 1.5 fold induction cut-off) were only identified at levels high enough to be called expressed by one assay- the other assay picked up a weaker signal, below the cut-off value. Repeating the experiment with a third assay might increase the accuracy of the data by allowing more comparison.

Future research could include performing the experiment again in other species to see if there is an overlap in the genes used for eye expression for each species. Would the genes identified by this type of experiment in Drosophila have homologs in other species? How many of the genes are conserved? This would be especially interesting if the comparison species had a different type of eye (for example, simple eyes like mice) because it could give another clue to how closely related different types of eyes are in terms of evolution.


Michaut, L., S. Flister, M. Neeb, K. P. White, U. Certa, and W. J. Gehring. (2003). Analysis of the eye developmental pathway in Drosophila using DNA microarrays. Proc. Natl. Acad. Sci. USA, Vol. 100, Issue 7, 4024-4029.

Ashley's Homepage

Molecular Bio Home

Davidson College Home