Discussion
The purpose of this experiment was to analyze the effects of knocking out hexose transporter genes on gene expression in S. cerevisiae. The two yeast strains examined were the Δhexose20 mutant, EBY, and the wild type strain, CEN. The intent was to survey the genome to observe genes with significant changes in expression. It was expected that the genes that have the most change in expression would be directly linked to the sugar metabolic pathway.
In order to evaluate gene expression, RNA had to be isolated from the two different yeast strains. To make sure that the RNA was comparable between the two strains, the optical density (OD) reading was taken at 600 nm. This was to determine which stage of growth that the yeast were in. Due to inconsistencies in the original data, the data presented in Table 1, is based on the data from Daniel 2. An OD of 6.6 shown in Table 1, shows that the yeast were in a stationary phase with an estimated concentration of 2x10E8 yeast cells per milliliter. The RNA isolation procedure was followed. The RNA was then quantified to determine not only its concentration, but also it's purity. Based on the A260/A280 ratio's of 1.56 for each strain suggested that the RNA extract was not entirely pure, a pure RNA sample would have yielded a ratio between 1.9 and 2.2. The RNA was then checked for degradation using agarose gel electrophoresis. Figure 1 shows the results of running the RNA. As expected there were three bands present for the wild type strain at 28S, 18S, and 5S. However, there was some RNA contamination present as signified by the smearing located at the 28S band. The absence of any bands with the mutant strain signifies poor RNA extraction procedure and/or RNA degradation.
The RNA isolated by Daniel 2 was loaded on to a microarray slide that contained two copies of the S. cerevisiae genome. The original slide number was 136 and the mutant strain was labeled with the red dye while the wild type was labeled with the green dye. Upon return of the microarray, the results were unable to be analyzed due to extreme background noise and limited RNA presence as seen by the lack of dye fluoresced on the array. This could be to several experimental errors such as failure during the labeling procedure, ozone degradation of the dyes, as well as the degradation of the RNA itself by RNases. Due to the inability to analyze chip #136, a slide from previous years, #3304 (Fall '04) was used for analysis.
Based on the type of experiment that was being conducted a list of genes was compiled that were hypothesized to show changes in expression levels when comparing the mutant to the wild type yeast strains. Table 2 shows this list of genes and their levels of expression in a log2 form. The data was converted to log2 form in order to better show the repression and induction of gene expression. Table 2 shows that nine out of the ten genes are repressed, while HOM6 is induced. Thus the cell does not appear to over express any sugar transport proteins to compensate for the loss of the hexose transporters.
Figure three is a representation of genes with expression changes over 25 in both samples. These values are not fold increases as the data was not taken to the log2. The relation of genes when compared using this method should generate straight lines due to the fact that both array chips are identical. Obviously, straight lines or identical expression is not observed in figure three. Microarray analysis is composed of many steps, but has few which would make it prone to error. Most likely, the reason for this idiosyncrasy is that the gridding steps done by Daniel group one and Daniel group two do not coincide. Why the data does not relate is still unclear, but it indicates that the computer is assigning florescence values or gene expression ratios to unrelated genes.
Figure three indicates a problem that becomes a common theme in the rest of the data obtained from analysis. The question remains: if the gridding is not synchronized and the wrong expression values are being computed for genes, then how can correct conclusions be drawn? Figure four simply indicates the genes with the most similar change in expression or the strongest cluster. When these genes are explored more in depth, table three, there is no clear relation between them. Ideally there would have been a correlation with the metabolic pathway if the microarray had been analyzed correctly.
If sugar transporters were knocked of the S. cerevisiae genome it would be expected that other genes involved in sugar metabolic pathways would show the largest changes in expression. However, the most similar cluster shows genes that are involved in DNA repair, GTPase activity, macromolecular recycling, and cell wall maintenance. Of these four FUN12 and ATG7 are the only ones that appear to have any possible correlation to hexose transporter genes while the others are superfluous. The possible relation of the FUN12 gene to hexose transporters is that the GTPase enzyme encoded by this gene is involved in the metabolism of more complex sugars that need to be broken down because the more simple sugars are not available to the cell. ATG7 is thought to play a role in conserving energy by recycling macromolecules. The cell must do this because the energy normally obtained by the sugars brought into the cell by the hexose transporters is not available. It is not known what the correlation between the other two genes are to the sugar metabolic/transport pathway. Again, this clustering anomaly appears to be due to miss-gridding and incorrect analysis of the data. Because the level of expression for genes has been incorrectly assigned through gridding the computer has determined the wrong expression and in turn the wrong dissimilarity values of genes that would otherwise be unrelated, show a correlation.
The final relation of gene expression determined in this experiment is most certainly flawed. These results lead to two courses of action. More time must be taken with the data obtained to try and determine why the gridding of groups one and two did not relate properly. Furthermore, this experiment should be conducted again using a separate Δhexose20 knockout. Comparing identical yeast samples did not lead to strong results and promoted possible errors that were inflated into misinterpreted readings.