Vignettes highlighting breadth of novel insights into DUB biology

Rapid growth in publicly available and “functional genomic” datasets affords an opportunity for extensive analysis of large gene families such as human DUBs. A total of nine different public resources measuring the following were mined for data, in most cases starting with a CRISPR-Cas9 knockout or small molecule signature we collected in our laboratories (using the high-throughput DGE RNA-seq):

• transcript signatures following gene perturbation (CMap)
• gene essentiality (DepMap, IMPC ad MGI)
• protein co-expression (CCLE proteomics)
• protein-protein interaction (BioGRID, IntAct, Pathway Commons PPID, and NURSA PPID)

Click here to view a table summarizing the data and insights gained from each resource.

Comparison of enriched gene sets or GO terms made it possible to bridge different types of data. Overall, we observed substantial and encouraging consistency among datasets. For example, genes identified as co-dependent with DUBs in DepMap data frequently exhibited similar transcript signatures, were co-expressed across cell lines, and were physically associated.

Our analysis yielded three types of information:

1. potentially new or more precisely specified functions for several well-characterized DUBs, including a UCHL5, USP7, USP8, and USP14
2. potential pathways or functional roles for understudied DUBs, including a role for USPL1 in the Little Elongation Complex and UCHL3, USP38, VCPIP1, and USP42 in the regulation of TP53 signaling;
3. insight into the DUB family as a whole, including evidence that 23 DUBs play a role in stabilizing 33 E3 ubiquitin ligases, most likely by antagonizing their auto-ubiquitination activities. 52 DUBs were found to be essential for proliferation in at least five cancer cell lines, and 34 of these DUBs affected the proliferation of cell lines from one tumor type more than cell lines from all other tumor types, potentially providing insight into disease context. These data are summarized in a simple graphical form, as a series of tables suitable for computational analysis in supplementary materials, and online on this DUB Portal.

Conclusion

Our studies provide a diverse set of data on the DUB family as a whole as well as new insight into many individual DUBs, including several that have been studied intensively. One theme that emerges is that for genes with multiple proposed functions (USP7 and UCHL5 for example), a combination of profiling CRISPR-Cas9 knockouts or drug-induced perturbations with systematic mining of functional genomic databases makes it possible to distinguish among essential and no-essential phenotypes. A second is that more DUBs than anticipated have non-redundant roles in the tumor suppressor and oncogenic pathways, most notably TP53 regulation, suggesting new approaches to undruggable targets. The approaches described in this work are directly applicable to other gene families and therapeutic targets.