Primary Filtering

Once the preliminary steps, described here, are complete, the microarray experiment will result in a list of genes with associated ratio changes and p-values. In a diagnostic, two-color slide experiment like the current example of forskolin vs. vehicle, the ratio represents the change in gene expression levels between the two conditions and the p-value represents whether the measured expression levels were significantly different from each other.

In the current experiment, two different comparisons, forskolin vs. vehicle and ICI + forskolin vs. ICI + vehicle, were performed at 3 different time points, t = 3hrs, 6hrs, and 24hrs. To begin with I will focus on the forskolin vs. vehicle without ICI condition.

Microarray experiments generally result in very large volumes of data; for instance, this experiment resulted in a spreadsheet depicting 1,591 probe sequences, some of them corresponding to the same genes. A first step in dealing with such a large set of data is to filter it. Searching for genes that exhibited a 2-fold change, either up or down, we should apply two different filters. Those genes that exhibit a ratio > 2 and a p-value < 0.05 are considered to be up-regulated in a statistically significant manner; in this experiment 36 genes met this criteria at some time point and this data was saved to a spreadsheet that can be found here. Similarly those genes with a ratio < 0.05 and a p-value < 0.05 can be considered down-regulated; in this experiment 10 genes met this criteria and they can be found here. In this case, these filters were applied to find genes that had a significant affect at at least one time point, not necessarily every time point.

Based off these filters we have a relatively small (46) assortment of genes that appeared to be significantly affected by the presence of forskolin. A good next step is to review the function of these genes by looking them up in MGI. Based off knowledge of the genes functions and interactions, as well as our foreknowledge of what forskolin does, we can start identifying how these genes fit in to the larger picture of the cell.

The PKA Pathway

Forskolin acts to raise cAMP levels and initiate the protein kinase A signaling pathway, among other effects. Consequently, a number of the genes identified as having their expression levels changed are part of the PKA pathway, including perhaps these four:

Standing out right away with a very large positive fold change is Crem, cAMP responsive element modulator, an important factor in the PKA pathway. Crem is also known as ICER, induced cAMP early repressor, which describes its function rather aptly. The PKA pathway is initiated by raised cAMP levels which in this case are caused by forskolin. Crem is a transcription factor that helps transcribe genes that work in reaction to this signal pushing the cell back to homeostasis. It regulates various genes including suppressing its own expression.

When we look at Crem's expression level over time we see that very quickly, by t=3hrs, it raises to a very high level which is maintained through t=6hrs, but that its levels drop at t=24hrs. This is most likely a result of its auto-regulation. Although we see its expression levels dropping at t=24hrs, it is important to remember that the microarray only gives us information about mRNA levels not protein levels. The previous high levels of transcription resulted in lots of Crem mRNA that in turn produced large amounts of Crem protein, and it is the protein rather than the mRNA that is affecting transcription of other genes such as itself. I would expect that the decrease in expression is not associated with equally low levels of protein at t=24hrs, but rather especially high levels of protein at that point.

Another gene that was up-regulated and might act as a part of the PKA pathway is Enpp2, ectonucleotide pyrophosphatase/phosphodiesterase 2. Enpp2 is a phosphodiesterase which breaks down phosphodiester bonds such as those found in cAMP. The increased Enpp2 gene expression levels, possibly being regulated by the transcriptional factor Crem, will result in increased levels of phosphodiesterase activity and less cAMP. While both Crem and Enpp2 are activated in response to the effects of forskolin raising cAMP levels, they work through very different methods with Crem acting as a transcription factor to modify the expression levels of other genes and Enpp2 acting directly as an enzyme on cAMP.

The slow, steady rise in Enpp2 levels over time, as compared to the quick rise and eventual drop of Crem levels, fits with the theoretical model of the PKA pathway that I am describing here. Whereas Crem transcription is most likely increased as a very direct response to the effect of forskolin, Enpp2 is likely to have its expression levels modified by a transcription factor that was itself affected by forskolin, such as Crem. The initial effect of forskolin would serve to increase transcription of this transcription factor, and as the protein levels of that transcription factor rise over time it in turns increases transcription of Enpp2.

Thirdly two protein phosphatases, Ptpn21 and Ppp1r3a, were seen to be down-regulated. Ptpn21 is protein tyrosine phosphatase, non-receptor type 21 and Ppp1r3a is protein phosphatase 1, regulatory (inhibitor) subunit 3A. Protein phosphatases act by removing phosphate groups that were attached to proteins by protein kinases, such as protein kinase A in the PKA pathway. Phosphatase activity would serve to inhibit the signal through the PKA pathway, so by down-regulating the production of these two phosphatases the signal should have an increased effect. The down-regulation of the phosphatases was relatively constant over time.

Coming soon

The genes discussed above are only a small subset of those genes identified as having their expression levels changed as a reaction to forskolin and I plan to discuss more about the others soon.