iLIR: A web resource for prediction of Atg8-family interacting proteins

Abstract

Macroautophagy was initially considered to be a nonselective process for bulk breakdown of cytosolic material. However, recent evidence points toward a selective mode of autophagy mediated by the so-called selective autophagy receptors (SARs). SARs act by recognizing and sorting diverse cargo substrates (e.g., proteins, organelles, pathogens) to the autophagic machinery. Known SARs are characterized by a short linear sequence motif (LIR-, LRS-, or AIM-motif) responsible for the interaction between SARs and proteins of the Atg8 family. Interestingly, many LIR-containing proteins (LIRCPs) are also involved in autophagosome formation and maturation and a few of them in regulating signaling pathways. Despite recent research efforts to experimentally identify LIRCPs, only a few dozen of this class of-often unrelated-proteins have been characterized so far using tedious cell biological, biochemical, and crystallographic approaches. The availability of an ever-increasing number of complete eukaryotic genomes provides a grand challenge for characterizing novel LIRCPs throughout the eukaryotes. Along these lines, we developed iLIR, a freely available web resource, which provides in silico tools for assisting the identification of novel LIRCPs. Given an amino acid sequence as input, iLIR searches for instances of short sequences compliant with a refined sensitive regular expression pattern of the extended LIR motif (xLIR-motif) and retrieves characterized protein domains from the SMART database for the query. Additionally, iLIR scores xLIRs against a custom position-specific scoring matrix (PSSM) and identifies potentially disordered subsequences with protein interaction potential overlapping with detected xLIR-motifs. Here we demonstrate that proteins satisfying these criteria make good LIRCP candidates for further experimental verification. Domain architecture is displayed in an informative graphic, and detailed results are also available in tabular form. We anticipate that iLIR will assist with elucidating the full complement of LIRCPs in eukaryotes.

Keywords: Atg8-family interacting proteins; LC3 interacting region-motif; macroautophagy; prediction; selective autophagy; web server.

Figure 1. PSSM score distributions for different classes of hexapeptides. Box-plot representation of PSSM score distributions for xLIR-motifs in the 26 sequences of LIRCPs (verified and unverified; left and middle, respectively) and the remaining hexapeptides (“background”; right), obtained by scoring a sliding-PSSM along the sequences in the set of 26 sequences reported by Alemu and colleagues. The differences indicated here suggest that PSSMs may be able to reliably discriminate between functional and nonfunctional xLIRs. In particular, a Wilcoxon rank sum test with continuity correction, demonstrates significant differences between both verified and unverified xLIRs compared with background (P < 2.2 × 10⁻¹⁶ and 1.2 × 10⁻¹⁴, respectively) and verified vs. unverified xLIRs (P: 6.0 × 10⁻⁶).

Figure 2. iLIR server user interface. A simple user interface enables sequence data entry in FASTA format either by copying-pasting data in the respective text area or by uploading a FASTA formatted text file. Currently, a single sequence can be processed at a given execution and no user-defined parameters are necessary/supported. Pre-run examples of known LIRCP sequences help users get accustomed to the iLIR output format.

Figure 3. iLIR results page. The output page for human SQSTM1 (UniProt Accession: Q13501) is displayed. A graphical depiction of identified domains (top) is accompanied with detailed tables (bottom). Some domains/features are kept hidden for maintaining an uncluttered graphic but are present in the tables. By moving the mouse over any domain/feature on the graphic, a pop-up tip (blue panel, top right) displays further information (in this case, details regarding the name, borders and origin database of the ubiquitin associated [UBA] domain). Notice that the tables containing anchors and SMART-derived domains may be shown/hidden according to the user’s preference.