Decoding isozyme-specific substrate recognition in protein arginine deiminases by in vitro lysate-based citrullination mapping

Figures

• 

  Fig. 1. Lysate-based citrullination mapping delineates PAD isozyme-specific substrate landscape. A Schematic overview of the workflow. Created in BioRender. Laposchan, S. (2026) BioRender.com/8qb0alk. B Total number of peptides and proteins identified per experiment. Light bars represent the mean number of peptides across replicates (n = 3 for all experiments except HeLa PAD2, where n = 2), ± standard deviation shown as error bars, and with markers for individual data points (X). Dark bars indicate number of peptides shared across all replicates, and dots represent proteins shared across replicates. C Number of citrullination sites and citrullinated proteins identified per experiment. Light bars represent the average number of sites across replicates (n = 3 for all experiments except HeLa PAD2, where n = 2), ±standard deviation shown as error bars, and with markers for individual data points (X). Dark bars indicate number of sites shared across all replicates, and dots represent proteins shared across replicates. D Principal component analysis (PCA) of all citrullination sites shared between all replicates, isozymes and lysates (n = 2221). E Overlap of citrullination sites catalyzed by…

• 

  Fig. 2. Motif and structural analyses reveal distinct preferences of PAD1/2 versus PAD3/4. A Motif enrichment analysis of citrullination sites generated by each PAD isozyme in H4 lysate compared to the background arginines of all identified proteins. Significance = log-odds of the binomial probability. B, C Co-occurrence of residues flanking the modified arginine for PAD2 and PAD4. Only sites containing a favored residue at either position, as defined by motif analysis in (A), are included. Connections indicate the frequency of residue pairings between positions −1 and +1. D Intensity distributions of all citrullination sites compared to sites containing favored residues at −2 or +2 (−2| +2), −1 or +1 (−1| +1), or both −1 and +1 (−1& +1) positions. E, F Frequency of amino acid combinations at −1 and +1 positions in PAD2- and PAD4-modified sites. Rows and columns are hierarchically clustered by Euclidean distance. G Predicted relative solvent accessibility (RSA) of modified arginine residues compared to all arginines within identified proteins (background, BG). Box plots indicate the median, with box bounds representing the 25th and 75th percentiles (interquartile range, IQR). Whiskers…

• 

  Fig. 3. Time-resolved profiling demonstrates consistent substrate specificity of PAD2 and PAD4. A Number of identified citrullination sites (bars) and their summed intensity (lines) across incubation times with PAD2 and PAD4 (n = 1 per timepoint). Light bars denote total sites per timepoint; dark bars indicate sites shared with previous timepoints. Control sample was incubated without enzyme. B, C Proportion of citrullination sites containing favored residues (as defined in Fig. 2A) at −2/ +2, −1/ +1, or both −1 and +1 positions for PAD2 and PAD4. Background (BG) represents the proportion of arginine residues bearing favored neighboring residues (regardless of modification) within all proteins identified in H4 lysate. AA, amino acid. D, E Hierarchical clustering (Euclidean distance) of citrullination sites reveals early, intermediate, and late conversion clusters for PAD2 and PAD4, respectively. Right panels show motif enrichment for sites within each cluster compared to background proteome. Significance = log-odds of the binomial probability. F, G Predicted relative solvent accessibility (RSA) of PAD2- and PAD4-modified arginines across all timepoints (All) and within Early, Mid, and…

• 

  Fig. 4. Single point mutations in PAD4 expand substrate repertoire. A Sequence alignment excerpt of PAD1–4 near known substrate-binding regions. Residues selected for mutagenesis are highlighted in red. B Crystal structure of PAD4 (gray) in complex with histone H4 substrate (yellow) [PDB ID: 2DEY]. Catalytic residues are shown in blue, and mutated residues in red. C Number of citrullination sites shared across replicates (n = 3) for each enzyme or PAD4 mutant compared to control (no enzyme). *WT PAD2 data shown here for comparison is from endpoint WT experiments and was not generated along with the mutant data from this batch. Shading indicates sites shared with wild-type PAD2, PAD4, or both. D Motif enrichment analysis of sites newly generated by each mutant (i.e., not shared with wild-type PAD4) compared to background proteome. E Motif enrichment analysis of sites lost in each mutant (i.e., present in wild-type PAD4 but absent in the corresponding mutant) compared to background proteome. Significance = log-odds of the binomial probability. Source data are provided as a Source Data file.

• 

  Fig. 5. Multiple PAD4 mutations drive broad substrate recognition similar to PAD2. A Number of citrullination sites shared across replicates (n = 3) for each enzyme or PAD4 mutant compared to control (no enzyme). *WT PAD2 data shown here for comparison is from endpoint WT experiments and was not generated along with the mutant data from this batch. Shading indicates sites shared with wild-type PAD2, PAD4, or both. B PCA of all citrullination sites shared between all replicates and enzymes (n = 1956). PAD4 set is the WT PAD4 run with the endpoint wild-type PAD1-4 experiment, *PAD4 set is the WT PAD4 control run with the single-point mutant experiment, **PAD4 set is the WT PAD4 control run with the multiple-point mutant experiment. C Heatmap of citrullination sites identified in each replicate of wild-type PAD1–4 and PAD4 mutants. Rows and columns are hierarchically clustered by Euclidean distance. D Motif enrichment analysis of PAD4 quadruple mutant (ACDE) (top) and wild-type PAD2 (mirrored, bottom) sites, each relative to background of all identified proteins. Significance = log-odds of the binomial probability. Source data are provided as a Source Data file.