Since I started my PhD, I have been interested in miRNAs, their regulation and how miRNAs regulate the functions of their targets. I also proposed miRNA and mRNA networking as part of my PhD thesis. Exactly two years ago, August 2009, we started a collaborative project on miR-11 with Dr. Maxim Frolov and his post-doc researcher Dr. Mary Truscott at the department of Biochemistry and Molecular Genetics, University of Illinois at Chicago, USA in the Drosophila melanogaster model system. Our interest was in studying two microRNAs, miR-11 and miR-998, which are encoded within an intron of the Drosophila E2F1 gene, de2f1, and are likely to be co-expressed with de2f1. In flies, E2F1 is a potent inducer of proliferation and apoptosis, and one consequence of E2F1 overexpression is apoptosis. Conversely, in the absence of endogenous dE2F, cells do not die following irradiation-induced DNA damage.
The figure illustrates the role of miR-11 in repressing members of the reaper/hid (wrinkled)/grim/sickle family of proapoptotic molecules, thereby protecting against cell death in Drosophila. The image was originally designed by Frolov lab members, and drawn by Iveel Mashbat, as a label for beer brewed and bottled for the celebration of the first author’s wedding July 1, 2011. The inset shows increased cell death following DNA damage (red, antibody recognizing active caspases) in mutant tissue (lack of green fluorescence protein) compared to wild-type tissue (presence of GFP).
We found that co-expression of miR-11 with dE2F1 suppressed dE2F1-induced apoptosis, but did not modulate dE2F1-induced proliferation in transgenic flies. We hypothesized that a normal function of miR-11 is to protect cells from E2F1-induced apoptosis during S phase entry. Since miR-11 is co-expressed with dE2F, an increase in dE2F at the G1/S transition could be accompanied by an increase in miR-11, which we predicted would protect cells from apoptosis in S phase. Microarray experiments were done using RNA from the wild-type eye disc tissue, and from tissue in which miR-11, dE2F1, or miR-11 and dE2F1 were overexpressed. The idea was that we will learn which pro-apoptotic genes are induced by dE2F and how miR-11 affects their expression. This would tell us about potential targets of miR-11. I found that GOBP enrichment analysis confirmed what was found in genetic interactions: expression of dE2F1 alone induced both proliferation- and cell death-associated genes. However, following co-expression of miR-11 with dE2F1, cell death genes were no longer overrepresented.
Mary generated a mutant that specifically deleted mir-11 but did not affect the expression of dE2F1, in which she did not observe an increase in cell death in tissues in which cells were proliferating. She did, however observe that cells were more sensitive to dE2F1-dependent irradiation-induced cell death. Proapoptotic molecules reaper and hid were expressed at higher levels in mir-11 mutant cells, and their expression increased further following exposure to irradiation. Significantly, dE2F1 was found bound to the reaper and hid gene promoters in irradiated cells. mir-11 repressed 3’UTR sensor constructs that contained the predicted mir-11 target sequences of reaper and hid. This led us to hypothesize that dE2F1 and mir-11 target the same genes. I compared predicted miR-11 targets that had been identified using the targetscan and, using an orthology mapping approach, I identified predicted dE2F1 targets. Surprisingly ~25% of miR-11 targets were also dE2F1 targets, and cell death genes were overrepresented. This unbiased bioinformatic approach complemented the findings from genetic interaction experiments. Using these findings, we tested cell death genes predicted to be regulated by dE2F1 and miR-11 in chromatin immunoprecipitation, and 3’UTR sensor assays, respectively. We identified a common set of genes implicated in cell death, which were regulated by both dE2F1 and miR-11.
Also, together with a graduate student in Dr. Maxim Frolov’s lab (Brandon Nicolay) we previously analyzed microarray data where two proteins, Rbf and Warts, in the Hippo pathway were mutated (Nicolay et al, Genes & Development 2011). We thought that microarray of the eye discs which overexpress dE2F might complement nicely to the RBF mutant array which was analyzed within the Hippo/RBF project. So, together with already known information, we hypothesized that cell cycle and proapoptotic genes will be upregulated in the de2f samples, but that the proapoptotic genes will not be upregulated in the de2f+miR11 samples. Also, importantly, cell cycle genes that are upregulated in Rbf1 mutant tissue might be upregulated upon overexpression of de2f.
Main challenges in this data analysis were finding a suitable algorithm for raw microarray data normalization and differential expression. This is because RNA was extracted from overexpressed cells which contained large proportions of wild type cells. Also, to find putative miR-11 targets that would be true positive. We tried almost all available miRNA target prediction algorithms and did several combinations (union set), intersections and tested our hypothesis. Well, in most cases our hypothesis suits well, so we later combined targets from five well known databases. Another approach that helped us to prove our hypothesis computationally was enrichment analysis of Gene Ontology and pathways. For this we used our laboratory developed tool Gitools.
After two years for bioinformatic and wet lab experiments and validation in various ways, finally we could prove this hypothesis with enough evidence and we are pleased to announce that it has been just published in Genes and Development journal: