Meta-analysis of SUMO1

An abundantly growing body of literature implicates conjugation of SUMO in the regulation of many proteins and processes, yet the regulation of SUMO pathways is poorly understood. To gain insight into the players in the SUMO1 pathway I have performed an in-silico co-expression meta-analysis of SUMO1, comparing many different multi-microarray studies of various normal and human tumour tissues, from the Oncomine database. This serves as a data-driven predictor of pathway partners of SUMO1. While the data obtained need to be confirmed by future independent experiments and can currently only be considered a hypothesis, results implicate defender against cell death (DAD1) and the anti-apoptotic DEK oncogene as new pathway partners of SUMO1.


Discussion
Oncomine [1] meta-analysis was performed as previously described [2,3]. Briefly, 15 multi-array studies were analyzed for common overlapping co-expressed genes of SUMO1, using muti-array studies within the Oncomine integrated cancer database. This technique gives insight into which pathways the searched gene (in this case SUMO1) are involved in, although it is impossible to tell if co-expressed gene products are complexed to SUMO1, act upstream of SUMO1 or downstream of SUMO1. Therefore, while limited, this technique is important for generating leads to assess both the pathways SUMO1 is important for, and regulation of SUMO1 itself.
After meta-analysis there were over 400 consistently coexpressed genes at the cutoff of 3 studies (Additional File 1). Table 1 shows the genes with the higher cutoff of 4 studies. This high number may be expected as SUMO1 is a general factor and involved in many processes. I note that the archetype SUMO1-modified promyelocytic leukemia (PML) was co-expressed with SUMO1, acting as validation of the results [4]. While the Ubc9 conjugation enzyme was not found to be co-expressed many other ubiquitin-conjugating enzymes were (UBE2N, UBE4A,  UBE2G1, UBE2V2, UBE2E1, UBE2D2, UBE2A, UBE1C,  CUL4A), as was the SUMO1 activating enzyme subunit 2 (UBA2). Transcription factors shown to be modified by SUMO were also co-expressed, such as HIF1α, Rb, YY1, and SMAD4 [5][6][7][8][9]. Interestingly RARα is also co-expressed and while it has never been shown to be a target of SUMO1 the PML-RARα fusion has been shown to be a target of SUMO1 mediated degradation [10]. It would be interesting to investigate if RARα itself is a SUMO1 target. Also co-expressed is the NF-κB subunit RelA. While RelA also is not a proven target of SUMO1 NF-κB is regulated indirectly by SUMO1 modification of Iκ Kgamma/NEMO or IκB [11,12].
A similar meta-analysis was attempted for SUMO2 and SUMO3. However, SUMO2 was not expressed to levels that allowed for meta-analysis, and the results of SUMO3 meta-analysis gave fewer co-expressed genes than for SUMO1 (Additional File 2). There was a small overlap (37 genes) of co-expressed genes of SUMO1:SUMO3, but this does not necessarily imply that both are involved in completely distinct pathways. Rather, the meta-analysis tech- 27% dual-specificity tyrosine-(Y)-phosphorylation regulated kinase 1A Oncomine meta-analysis of SUMO1 co-expressed genes at a cutoff of 27% overlap (4 studies). nique has a high false-negative rate meaning that while the co-expressed genes we see are significant we will never get full coverage of every co-expressed gene as the stringency level of analysis is high.
SUMO1 was also seen to be involved in cell death pathways. In 67% (10 out of 15) of the studies analyzed SUMO1 was co-expressed with the defender against cell death (DAD1) gene. This was the highest co-expression with SUMO1 in the meta-analysis. As the name suggests DAD1 is anti-apoptotic and can be upregulated in cancer [13,14]. Other SUMO1 co-expressed genes involved in cell death pathways include RELA, FADD, BCL2A1, BAK1, TNFRSF1A. The high co-expression with DAD1 is a novel finding and may prove important to SUMO1 pathways.
DEK oncogene was the next highest co-expressed gene (53%) with SUMO1. The DEK protein is important for chromatin structure, and may also play a role in cell death pathways by inhibiting apoptosis [15][16][17].
While co-expression meta-analysis data has previously been shown to have a high correlation with known pathways in other studies [2,3], prudence should still be used when interpreting novel findings until they can be proven in a separate experimental system. For this reason the meta-analysis list is presented here only as a predictive data-driven hypothesis. The next step is experimental analysis of DEK and DAD1 proteins to assess whether they are targets of SUMO1 conjugation, protein-complex partners of SUMO1, or act upstream or downstream of SUMO1.
In summary, it is interesting that both of the highest coexpressed genes of SUMO1 are anti-apoptotic, and it is tempting to speculate that this may be an important pathway of SUMO1 regulation.

Conclusion
Using co-expression meta-analysis from the Oncomine database SUMO1 co-expressed with many gene products, some which are already known to be in SUMO1 pathways. Novel predicted pathway partners include the DEK oncogene and DAD1, both of which co-expressed in over half of all studies analyzed. However, in what regard they take part in SUMO1 pathways remains to be further investigated.