Confetti: a multiprotease map of the HeLa proteome for comprehensive proteomics

Abstract

Bottom-up proteomics largely relies on tryptic peptides for protein identification and quantification. Tryptic digestion often provides limited coverage of protein sequence because of issues such as peptide length, ionization efficiency, and post-translational modification colocalization. Unfortunately, a region of interest in a protein, for example, because of proximity to an active site or the presence of important post-translational modifications, may not be covered by tryptic peptides. Detection limits, quantification accuracy, and isoform differentiation can also be improved with greater sequence coverage. Selected reaction monitoring (SRM) would also greatly benefit from being able to identify additional targetable sequences. In an attempt to improve protein sequence coverage and to target regions of proteins that do not generate useful tryptic peptides, we deployed a multiprotease strategy on the HeLa proteome. First, we used seven commercially available enzymes in single, double, and triple enzyme combinations. A total of 48 digests were performed. 5223 proteins were detected by analyzing the unfractionated cell lysate digest directly; with 42% mean sequence coverage. Additional strong-anion exchange fractionation of the most complementary digests permitted identification of over 3000 more proteins, with improved mean sequence coverage. We then constructed a web application (https://proteomics.swmed.edu/confetti) that allows the community to examine a target protein or protein isoform in order to discover the enzyme or combination of enzymes that would yield peptides spanning a certain region of interest in the sequence. Finally, we examined the use of nontryptic digests for SRM. From our strong-anion exchange fractionation data, we were able to identify three or more proteotypic SRM candidates within a single digest for 6056 genes. Surprisingly, in 25% of these cases the digest producing the most observable proteotypic peptides was neither trypsin nor Lys-C. SRM analysis of Asp-N versus tryptic peptides for eight proteins determined that Asp-N yielded higher signal in five of eight cases.

Fig. 1.

A multi-protease approach for bottom-up proteomics. Forty-eight single, double, and triple-enzyme sequential digests were analyzed by LC-MS/MS using Orbitrap Elite CID and QExactive HCD. The five best digests by complimentary proteome amino acid coverage for CID and HCD were then identified, and SAX fractionated to achieve deeper proteome coverage. All resulting data was used to build Confetti, our web-accessible coverage map of the HeLa proteome.

Fig. 2.

A, As up to 48 single, double, and triple enzyme HeLa lysate digests are combined, total proteome amino acid coverage (PAAC) increases, reaching ∼3x of trypsin alone (HCD analysis). The percentage improvement with each addition drops rapidly as the total number of digests grows. B, Mean sequence coverage and number of protein groups identified for CID and HCD analyses of SAX-fractionated digests, using the best five enzyme combinations for each method.

Fig. 3.

Resolution of SAX fractionation for five digests analyzed with Q Exactive HCD LC-MS/MS. A, Graphs show the mean number of peptide identifications for each fraction, across three replicate injections. Colored stacked bars indicate the number of peptides in each fraction that are unique to that fraction (green, labeled) or present in 2–8 fractions. Total peptide IDs per fraction are given above bars. B, Proportions of fraction-unique and duplicated peptide identifications across the trypsin SAX data set. Across three replicate injections a mean of 57.5% of peptide sequences were identified in a single SAX fraction.

Fig. 4.

Multiple digests substantially improve protein identification and sequence coverage. A, Total number protein groups identified and mean percentage sequence coverage for combinations of tryptic digests, 48 unfractionated single/double/triple enzyme digests, and SAX fractionated “best 5” digests. B, Distribution of sequence coverage among all protein groups identified at a 1% FDR using various digests. Using multiple digests sequence coverage decreases more slowly as protein depth increase than for trypsin alone.

Fig. 5.

A, The Top5 SAX data set was used to identify the single digest that produced the three most abundant SRM candidate peptides per protein group. Chart shows the percentage of proteins in which each enzyme produces the most abundant three candidate peptides by spectral counting. B, The SRM workflow used to validate digestion reproducibility and determine peak areas of Asp-N and tryptic peptides for eight proteins.

Fig. 6.

Testing the suitability of alternative enzymes (Asp-N) for SRM assay development. A comparison of summed peak areas for eight proteins - three peptides per protein, with three transitions measured per peptide. AspN peak areas exceed tryptic signal for five proteins, indicating improved ionization and/or transition response can be achieved for some assays using alternative digests. CV of peak area measurements is comparable.