An integrated bioinformatics platform for investigating the human E3 ubiquitin ligase-substrate interaction network - PubMed (original) (raw)

doi: 10.1038/s41467-017-00299-9.

Ping Xie 1 2, Liang Lu 1, Jian Wang 1, Lihong Diao 1 3, Zhongyang Liu 1, Feifei Guo 1, Yangzhige He 1, Yuan Liu 1, Qin Huang 3, Han Liang 4, Dong Li 5, Fuchu He 6

Affiliations

An integrated bioinformatics platform for investigating the human E3 ubiquitin ligase-substrate interaction network

Yang Li et al. Nat Commun. 2017.

Abstract

The ubiquitination mediated by ubiquitin activating enzyme (E1), ubiquitin conjugating enzyme (E2), and ubiquitin ligase (E3) cascade is crucial to protein degradation, transcription regulation, and cell signaling in eukaryotic cells. The high specificity of ubiquitination is regulated by the interaction between E3 ubiquitin ligases and their target substrates. Unfortunately, the landscape of human E3-substrate network has not been systematically uncovered. Therefore, there is an urgent need to develop a high-throughput and efficient strategy to identify the E3-substrate interaction. To address this challenge, we develop a computational model based on multiple types of heterogeneous biological evidence to investigate the human E3-substrate interactions. Furthermore, we provide UbiBrowser as an integrated bioinformatics platform to predict and present the proteome-wide human E3-substrate interaction network ( http://ubibrowser.ncpsb.org ).Protein stability modulation by E3 ubiquitin ligases is an important layer of functional regulation, but screening for E3 ligase-substrate interactions is time-consuming and costly. Here, the authors take an in silico naïve Bayesian classifier approach to integrate multiple lines of evidence for E3-substrate prediction, enabling prediction of the proteome-wide human E3 ligase interaction network.

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing financial interests.

Figures

Fig. 1

Fig. 1

Data flow to predict the human E3-substrate interaction based on naïve Bayesian classification. Five types of evidence were labeled on the top of each data source. First, likelihood ratios were calculated as the weight for the source(s) of each type of biological evidence towards the prediction of ESI. Second, if there was more than one source for each type of biological evidence, the maximum LR from this data source was identified. Finally, the likelihood ratio from individual evidence was integrated into a composite likelihood ratio by the naïve Bayesian classifier

Fig. 2

Fig. 2

Diverse types of biological evidences contributing to the reliable evaluation. a Domain pairs enriched among E3-substrate interaction. DER was used to measure domain pair enrichment, which was calculated as the probability (Pr) of observing a pair of domain in a set of known E3-substrate interactions divided by the product of probabilities of observing each domain independently. b GO term pairs enriched among E3-substrate interaction. GER was used to measure GO term pair enrichment, which was calculated as the probability of observing a pair of GO term in a set of known E3-substrate interactions divided by the product of probabilities of observing each GO term independently. c Network Topology. The number of the three- and four-interaction loops was calculated based on the integrated network with the query interaction and the HPRD protein interaction data. d E3 recognition consensus motif. Recognition consensus motif for each E3 was identified based on two parallel sequence data sets: one was the sequence data of this E3’s substrates in GSP for motif building, and the other was that of all the proteins interacting with this E3 in HPRD database for background probability calculations. Please refer to Supplementary Methods for details of the calculation of E3 recognition consensus motif

Fig. 3

Fig. 3

TP/FP ratio as a function of likelihood ratio cutoff for ESI prediction. The ratio of the true to false positive (TP/FP) was plotted as the function of the cutoff of likelihood ratio. The number of true positives and false positives were from the fivefold cross-validation (see text for details)

Fig. 4

Fig. 4

ROC curves for various assessment models using fivefold cross validations against the golden standard data sets. Each point on the ROC curves of various assessment models corresponds to sensitivity and specificity against a particular likelihood ratio cutoff. Different assessment models corresponding to these curves are labeled in legends. The numbers in parentheses refer to the AUC under ROC curves for each model. Sensitivity and specificity are computed during the fivefold cross-validations. SPSS software is used to smooth the curves. (see text for details)

Fig. 5

Fig. 5

Network and sequence view for predicted E3-substrate interactions in UbiBrowser web services. In network view the central node is the queried E3 ligase or substrate, and the surrounding nodes are the predicted substrates or corresponding E3s. In the confidence mode of network view, the edge width and surrounding node size are positively correlated with the confidence of prediction, and in the evidence mode of network view, each E3-substrate interaction will be presented by multiple lines with different colors for different type of evidence. Clicking each edge will lead to a popup illustrating the supporting evidence, and clicking the surrounded node a popup for the sequence view of the involved substrate. In the sequence view for each E3-substrate interaction, the substrate’s sequence is shown in PRIDE format with multiple signs: black lines under the sequence denote the potential domain interacting with related E3, gray lines under the sequence mark the inferred E3 recognition consensus motif and the yellow background of character K means known ubiquitination site

Fig. 6

Fig. 6

Experimental validation of predicted E3-substrate interaction. a, b Smurf1 destabilizes Smad3 in MDA-MB-231 cells. a MDA-MB-231 cells were transfected with Myc-Smad3 and Smurf1, after 36 h, cells lysates were analyzed by western blot. b MDA-MB-231 cells were transfected Myc-Smad3, with Flag-Smurf1 (WT or C699A: 0.5 μg of Flag-Smurf1-WT was transfected into the second band and 1 μg into the third one, and 1 μg of Flag-Smurf1-C699A into the fourth one). Myc-Smad3 level was analyzed by immunoblotting. c Smurf1 promotes the ubiquitination of Smad3. MDA-MB-231 cells were transfected with HA-Ub, Myc-Smad3, control vector, or Flag-Smurf1, Smad1 was used as a positive control, and treated with MG132 as indicated. Ubiquitinated Smad3 was immunoprecipitated (IP) with anti-Myc antibody and detected by immuneblotting with anti-HA antibody. d, e Smurf1 interacts with Smad3. Co-immunoprecipitation of Smurf1 and Smad3 in MDA-MB-231 cells. In Fig. 6d, Flag-Smurf1 was used as immunoprecipitated, and in Fig. 6e, Myc-Smad3 was used as immunoprecipitated. To avoid the degradation of Smad3, MG132 (20 µM) was added for 8 h before harvested. Cell lysates were immunoprecipitated with anti-Myc antibody and analyzed by immunoblotting

References

    1. Giasson BI, Lee VM-Y. Are ubiquitination pathways central to Parkinson’s disease? Cell. 2003;114:1–8. doi: 10.1016/S0092-8674(03)00509-9. - DOI - PubMed
    1. Popovic D, Vucic D, Dikic I. Ubiquitination in disease pathogenesis and treatment. Nat. Med. 2014;20:1242–1253. doi: 10.1038/nm.3739. - DOI - PubMed
    1. Perry G, Friedman R, Shaw G, Chau V. Ubiquitin is detected in neurofibrillary tangles and senile plaque neurites of Alzheimer disease brains. Proc. Natl Acad. Sci. 1987;84:3033–3036. doi: 10.1073/pnas.84.9.3033. - DOI - PMC - PubMed
    1. Song S, Jung Y-K. Alzheimer’s disease meets the ubiquitin–proteasome system. Trends Mol. Med. 2004;10:565–570. doi: 10.1016/j.molmed.2004.09.005. - DOI - PubMed
    1. Leroy E, et al. The ubiquitin pathway in Parkinson’s disease. Nature. 1998;395:451–452. doi: 10.1038/26652. - DOI - PubMed

Publication types

MeSH terms

Substances

LinkOut - more resources