China
Africa
Explore chapters and articles related to this topic
Precision medicine in myelodysplastic syndromes
Published in Debmalya Barh, Precision Medicine in Cancers and Non-Communicable Diseases, 2018
While CSNK1A1 is the CRL4CRBN target in del(5q) MDS, CRL4CRBN targets in lower risk non-del(5q) remain to be determined. The mechanism of action of lenalidomide is still unclear in non-del(5q) MDS cells. Recent evidence shows that lenalidomide directly improves erythropoietin receptor (EPOR) signaling by EPOR upregulation mediated by a posttranscriptional mechanism (Basiorka et al., 2016a). Lenalidomide stabilizes the EPOR protein by inhibition of the E3 ubiquitin ligase RNF41 (ring finger protein 41, also known as neuregulin receptor degradation protein-1 [Nrdp1] and fetal liver ring finger [FLRF]) (Basiorka et al., 2016a) and induces lipid raft assembly to enhance EPOR signaling in MDS erythroid progenitors (McGraw et al., 2012, 2014).
Targeting TANK-binding kinase 1 (TBK1) in cancer
Published in Expert Opinion on Therapeutic Targets, 2020
Or-Yam Revach, Shuming Liu, Russell W. Jenkins
K63-linked polyubiquitination of TBK1 on residues Lys30 and Lys401 is a posttranslational modification that is essential for Ser172 phosphorylation, activation, and downstream signaling [19]. Several TRAF family E3 ubiquitin ligases, including MIB1, MIB2 RNF128, RNF144B, and RNF41/Nrdp1, have been shown to promote K63-polyubiquitination of TBK1 [29–32]. Thus, CYLD (cylindromatosis deubiquitinase) and ubiquitin-specific protease 2b (USP2b) have been shown to limit TBK1 activation by removing K63-linked ubiquitin chains [33,34]. In contrast, K48-linked polyubiquitination promotes TBK1 degradation serving as a mechanism to limit TBK1 signaling, providing a negative feedback loop on TBK1 activation [35–37]. Additional regulatory factors influencing TBK1 ubiquitination have been described [38], as well as SUMO-ylation on Lys694 which appears to contribute to anti-viral signaling [39]. A recent study suggested that TBK1 itself can serve as an E3-ligase and induce its own ubiquitination in vitro [40], although the significance of this finding is unknown and requires further exploration.
Replication of single-cell proteomics data reveals important computational challenges
Published in Expert Review of Proteomics, 2021
Christophe Vanderaa, Laurent Gatto
Furthermore, the imputation is performed at the protein level. As pointed out by [30], imputation at the protein level means that a first implicit imputation is performed at the peptide level and the authors instead suggest to use well-justified imputation methods directly at the peptide level. However, a good understanding of the missingness mechanism is required to justify the use of a suited imputation method. Further research is required to extend current work on bulk proteomics to the context of SCP data [30,31]. Finally, just like batch correction, imputation is an estimation process that generates estimates with some degree of uncertainty. Replacing missing data by imputed values ignores the variance associated to the estimates and this variance can become large when available data are scarce. Multiple imputation, i.e. the application of a range of imputation parameters or methods to estimate a range of plausible values rather than point estimates, would be a promising strategy here. This is best illustrated by an issue we noticed in the data. For instance, the E3 ubiquitin-protein ligase (RNF41) is quantified in only three MS runs and KNN imputation predicted the missing values for the remaining runs (Figure 4(a)). When comparing the resulting data distribution to the distribution for vimentin (VIM), a protein that is not missing, we can clearly observe that the imputation introduces two suspicious trends. First, the variability observed for imputed values is much lower than for acquired values, and second, the imputation does not exhibit batch effects. While reduced variability and absence of batch effects are desirable properties, in this case, we are faced with erroneous data that does not hold biologically meaningful information. The imputed data for RNF41 is unreliable and should be flagged accordingly. Furthermore, small quantification errors may get amplified during the imputation step, as observed during the replication of the SCoPE2 data (Figure 4(b)). While we observed minimal differences between the two workflows before imputation, those small differences where magnified after imputation, considerably increasing the proportion of values deviating from zero.