| Discussion |
It is hypothesized that ZMSD1 and ZMSD2 are transcription factors involved in the regulation of ZWF1 which acts in the pentose phosphate pathway. This pathway is key to reducing oxidative stress because it produces NAPH, a powerful anti-oxidant. In order to study the ZMSD1 and ZMSD2 genes, microarray technology was employed. Expression of ZMSD1/2 knockout strains were compared to wild-type via the creation of dye labeled cDNA from cellular mRNA. The arrays prepared by our team were unable to be analyzed (Figure 2), and this poor data can be attributed to several factors. Improper preparation of the dyes, incomplete hybridization, and procedural errors in preparation of the slides are all possible sources of error in this experiment. In preparation of the red and green dyes, it is possible that these dyes were exposed to light, which would significantly reduce their ability to fluoresce. Additionally, since a gel of the cDNA was not run, it is unclear whether or not the reverse transcriptase performed its function. Any error in the preparation of the microarray slides could also lead to incomplete hybridization. Due to time restraints, microarray slides were kept incubated with a small amount of water at 37oC. Although this is procedurally sound, the optimal situation would be to hybridize the microarray slides immediately to minimize both degradation of the sample and contamination to the slide.
Due to the poor quality of the microarray data, data from previous years were used to compare ZMSD1 vs ZMSD2 mutants. Our group obtained data from Fall 2008 with a red dye labeled ZMSD2 mutant and a green dye labeled wild type. This data was combined with a ZMSD1 vs wild type data set so a comparison of ZMSD1 and ZMSD2 could be performed. The data was imputed into scanalyze software where the amount of pixels of light from each spot were measured. This data was then imputed excel where the background intensities were subtracted out and the ratio of mutant vs wild type obtained. The data was imputed into magic tool software for a log base 2 transformation and subsequent normalization and standardization. The purpose of the log base 2 transformation was to compact the data and make it easier to work with as far as seeing fold increases in gene expression. The normalization and standardization helped to minimize dye bias and fit the data to a standard bell curve for statistical analysis. Due to the low sample number (1 slide) no significant statistical tests could be done. In order to feel confident about what was up regulated and down regulated, the very extremes in gene expression were taken. Other methods of determining up and down regulation are to analyze the expression of housekeeping genes or focusing on two to three standard deviations away from the mean to determine significant variable regulation.
In our analysis, we found a cluster of genes that varied significantly between ZMSD1 and ZMSD2 (Figure 5). All these genes were upregulated in ZMSD2 and downregulated in ZMSD1. These genes were involved in a wide variety of cell processes, however, there were a few interesting genes involved in production of molecules used or synthesized the pentose phosphate pathway (Table 2). These genes are YJR009C and YJL052W which are both involved in glycolsis and glycogenesis. Both genes work to catalyzes the reaction of glyceraldehyde 3-phosphate to 1,3 bis-phosphoglycerate. There was also an unknown gene in this set called YLR243W which might be interesting for further studies. The expression differences between these genes may suggest that ZMSD1 and ZMSD2 function in different ways to reduce oxidative stress. There were also two genes that were significantly down regulated across both ZMSD1 and ZMSD2. YPR185C is a retrotransposon and YOR178C is a regulatory subunit for Glc7p type-1 protein phosphatase (PP1) thought to function in a heat shock pathway. It is hard to see a significant connection between these genes and what function they would serve in a cell when down regulated to help deal with oxidative stress.
Further studies may involve using RT-PCR to confirm up or down regulation of certain genes of interest. The strength of the microarray information could be improved in several ways. Using multiple microarray slides (increasing the same size n) would allow for more sound statistical tests to be used. Commonly used tests to analyze microarray data are T-tests and ANOVA. Additionally, in our study, the dye bias was not completely removed. Perhaps a Loess Regression could be employed along with a dye swap in order to make this data more sound. In magic tool, a mean normalization and standardization was used to make the data fit a standard bell curve. Instead of relying on this software so heavily several types of normalization could be used besides mean normalization. Median or quantile normalization could be tested to obtain a better fit for the data. Although the results of this experiment were not statistically founded, the microarray data gave some possible ideas on avenues of study and some interesting information concerning the relationship of ZMSD1 to ZMSD2.