Figures and data
Human peptidome library screening enriches known and novel LC3B-binding peptides.
(A) Schematic of FACS-based bacterial display enrichment screening. Peptides, 36 amino acids in length, that tile the human proteome were expressed as FLAG-tagged fusions to eCPX on the surface of E. coli. Expression was detected using the fluorescence signal from an allophycocyanin (APC)-conjugated anti-FLAG antibody. Binding to LC3B (grey), which was tetramerized through binding to streptavidin conjugated to phycoerythrin (SA-PE, pink), was detected using PE fluorescence. Peptide-expressing cells were sorted based on APC and PE fluorescence and sequenced. Panel created using BioRender.com. (B) Diverging bar chart plotting the number of unique sequences detected in each round of sorting that lacked (grey, left axis) or contained (right, blue) a LIR motif. Sort 2, a negative sort used to eliminate peptides nonspecifically bound to SA-PE, was not sequenced. (C) Peptide enrichment profiles across enrichment sorts. Black lines show trajectories for a random subset (0.1%) of all peptides that persisted to sort 6. Overlaid are the enrichment profiles for the best-enriching peptide (BLM552-570*), positive controls (FYCO11277-1312 and ATG4A363-398), and the worst-enriching peptide that reached sort 6 (EXTL3871-906) colored green, pink, yellow, and blue, respectively. Peptides sequences detailed in Supplementary Table 1. (D) Cumulative density function plot of the mean z-scores across sorts 4, 5, and 6. Mean z-scores for the 427 peptides of 12,158 that surpassed the threshold of 1.70 that were used to define the HC-set are colored in black, and select peptides are colored as in panel C. (E) Peptide enrichment profiles for the 427 peptides in the HC-set (black), with select peptides colored as in panels C-D.
Highly enriched peptides share annotations with LC3B and bind LC3B tightly.
(A) Peptides in the HC-set, annotated with various lines of evidence of binding to LC3B. For each protein, circles indicate: LIR - experimentally validated LIR motifs annotated in LIRcentral (Chatzichristofi et al., 2023); IP - proteins co-purified with LC3B, from BioGrid (Stark et al., 2006); GO – peptides with GO-annotations shared with LC3B, from Uniprot (The UniProt Consortium, 2023), with heatmap of enrichment z-score plotted at bottom. (B) Binding to monomeric LC3B assayed by BLI for peptides ATG4A363-398 (VPPAKPEVTTTGAEFIDSTEQLEEFDLEEDFEILSV), SCYL1640-675 (TADRWDDEDWGSLEQEAESVLAQQDDWSTGGQVSRA), HEAT3377-412 (EDPSDDEWEELSSSDESDAFMENSFSECGGQLFSPL), and BLM552-587 (DIDNFDIDDFDDDDDWEDIMHNLAASKSSTAAYQPI). Assay performed in triplicate, with error bars indicating standard deviation. Data fit to a standard binding isotherm: 
Highly enriched peptides feature W-type core LIR motif flanked by acidic residues.
(A) Sequence logo derived from analysis of the HC-set peptides (see Methods), and plotted using logomaker (Tareen and Kinney, 2019). Acidic residues are colored maroon and those colored red are significantly enriched compared to the input library. Other residues are colored gray with [FWY]0 and [LVI]3 in black. The core LIR is highlighted in yellow. (B) BLI measurements of binding affinity between LC3B and peptides pCONSLIR (EEEVEEKEEEDDDEEWEILDIEEGSDSEQKLISE), ANK21578-1613 (VQSSRSERGLVEEEWVIVSDEEIEEARQKAPLEITE), and FYCO11277-1312 (DAVFDIITDEELCQIQESGSSLPETPTETDSLDPNA). Error bars report the standard deviation of two or more technical replicates. Data fit to a standard binding isotherm, as in Figure 1. (C) Bar chart plotting binding affinities (KD) of sequential truncations of pCONSw measured via BLI, with error bars reporting standard error of the mean. Brackets report statistical significance of pairwise t-tests (****p ≤ 0.0001; ns not significant). (D) Structure of BLM552-571* bound to LC3B, resolved to 2.2 Å resolution (see Supplementary Table 4). Here, and in panels E-G, the BLM peptide is displayed in stick representation (green backbone) and the LC3B surface is colored by hydrophobicity, as computed by ChimeraX (Meng et al., 2023). Hydrophobic pockets HP1 and HP2, and the N-terminal flanking residues, are indicated. (E) Inset highlighting contact between BLM E568 and LC3B R70, and BLM D569 and LC3B K30, near HP2 (black dashed lines). LC3B residues that form HP2 are annotated and shown in stick representation. (F) Inset highlighting docking of BLM W567 in LC3B HP1, with residues forming that pocket annotated and shown in stick representation. (G) Inset highlighting interactions between BLM N-terminal acidic residues D564 and D566 with R11, K49, and K51 of LC3B. Putative contacts within 4.5 Å marked with dashed black lines.
LIR+ motifs support binding to LC3B.
(A) Affinities of LC3B for peptides from PAR118-37, CTSL2233-262*, and DYH12422-445*, as measured by BLI, plotted as a bar chart (mean ± s.e.m, n≥3). Candidate binding sites that were mutated to Ala are red in the mutated sequence. No detectable binding, up to 40 µM LC3B, is indicated with N.B. (B) AlphaFold3 (Abramson et al., 2024) structural prediction of CTSL2233-262* (colored by pLDDT) in complex with LC3B (colored by hydrophobicity). Insets: canonical LIR YYFI258-261 is not predicted to engage the LDS (left), in contrast to WEVF252-255 (right).
Mutations at the LC3B LIR docking site alter LIR-peptide binding specificity.
(A) Structure of LC3B bound to FYCO11276-1288 LIR peptide (PDB 5d94) (Olsvik et al. 2015) shown as a surface representation colored by hydrophobicity. Insets depict hydrophobic pockets formed by residues F52, L53 (stick representation) in wild-type LC3B (left), or a predicted surface of the LDS* mutant (right), with each alanine substitution colored red. (B) Enrichment profiles across three rounds of the LC3B LDS* enrichment sort, with nine peptides spanning the range of observed profiles colored and sequences provided. Four peptides that were enriched and bound with measurable affinity to LC3B LDS* are in blue. Known, canonical LC3B LDS binders that were depleted in the LDS* sort, are colored purple and red. Black lines indicate a random sample of 10% of peptides that persisted through LC3B LDS* sort 3. The histogram shows the frequencies of peptides with a given enrichment ratio in LDS* sort 3, with the dashed red line marking an enrichment ratio of 0. (C) Affinities of LC3B (blue) and LC3B LDS* (light red) for peptides FYCO11277-1312 and ATG4A363-398 and for four peptides found to enrich in across the LDS* sorts: PPM1H436-471, OSBL71-36, EFC13262-278*, TRIM5288-120. Error bars indicate standard error of the mean across three or more replicates. Hashed bars mark peptides without detectable binding at the highest measured LC3B concentration (40 µM).
