658 JOURNAL OF COSMETIC SCIENCE tandem mass spectra (2–32 Hz, depending on intensity) for 3 s. A preference of selection of multiple-charged precursors (2+ to 5+) was made, and an isolation width of 0.03 Th was set. Dynamic exclusion duration was set to 20 s. PROTEIN IDENTIFICATION Potential protein–protein cross-links were searched with Kojak v1.6 (7). The following parameters were used: a precursor mass tolerance of 10 ppm was allowed with fragment bin offset = 0, fragment bin size 0.03. A custom in-house database was created containing all human keratin protein sequences, keratin-associated proteins, and their reverse complements as decoy proteins. Trypsin was specified as a digestive enzyme, and up to three missed cleavages were allowed. A minimum peptide mass of 245 Da was set. Carbamidomethyl (C), oxidation (M), and deamidation (NQ) were chosen as variable modifications, and a maximum of three modifications per peptide was allowed. A lanthionine bond (Cys–Cys or Ser–Cys) or lysinoalanine bond (Lys–Cys) was specified as a cross-link. For each spectrum, the top 20 scoring peptides were used for cross- link combination. To minimize false positives, peptide spectrum matches exported from Kojak were further validated by Percolator V2.08 (8). The threshold for peptide spectrum matches after validation was set at q value smaller than 0.01. RESULTS CONFIRMATION OF CROSS-LINK PRESENCE IN HAIR FIBERS AND THEIR PROTEIN EXTRACTS Amino acid analysis was used to validate the presence of lysinoalanine or lanthionine cross-links in samples that were used for proteomics cross-link mapping. The amount of lanthionine and lysinoalanine present both in hair fiber snippets (one biological replicate run in duplicate) and hair snippet protein extracts (three separate protein extracts run in duplicate) is shown in Figure 1. Cross-links were detected in both the fiber and in the protein extract from the fiber, though some variability in the samples was noticed. Figure 1. The quantities of lanthionine and lysinoalanine cross-links (mg/100 mg) present in hair snippets (black) and triplicate hair extracts (blue) as determined by amino acid analysis represented with 95% confidence intervals.

659 MAPPING PROTEIN CROSS-LINKS IN HUMAN HAIR LOCATION OF LANTHIONINE CROSS-LINKS WITHIN HAIR PROTEIN EXTRACTS VIA MASS SPECTROMETRY As cross-linked peptides are very challenging to characterize by mass spectrometry, a data acquisition method that allows for better fragmentation of these low-abundant cross-linked peptide precursors was developed. By allowing a more dynamic sampling of precursors in the mass spectrometer, a higher chance of detecting low-abundant, multiple-charged precursors was created. To maximize the chance of detecting lanthionine and lysinoalanine cross-links between keratin-related proteins and to minimize the amount of false positive hits within this process, we designed a custom database that consisted of all human keratins and keratin-associated proteins, as it is believed that cross-links formed within these protein families are largely responsible for the hair fiber rigid structure. Using this FASTA file (2019.02), we analyzed for cross-links between 298 proteins (including decoy proteins), resulting in 22,204 in silico tryptic peptide combination possibilities, of which 8,759 peptides are linkable. Table I lists lanthionine cross-links that could be detected in the hair digests. The table lists the two potential peptide sequences that are linked by a lanthionine cross-link as well as their corresponding first hit proteins. It also provides the spectrum number so that further manual validation of the fragmentation spectrum is possible. The spectrum-match validation settings were chosen as very stringent, and only q values smaller than 0.05 were considered confident. Because the algorithm only reports q values to three decimal places, the q values that made it through the filter are reported as 0. A further look at the fragmentation spectrum of the detected cross-link is depicted in Figure 2. A few observations can be made. First, as lanthionine derives from dehydroalanine intermediates (6), the bond can exist between cysteine–cysteine or serine–cysteine in the cross-linked peptides. All the cross-linked peptides detected consisted of a cysteine– cysteine bond. Second, none of the annotated fragment ions are ions that are directly linked to the Cys–Cys bond, though several other ions are detected. Third, the detected fragment ions are still all very low abundance, yet through our dynamic acquisition method, just detectable. Fourth, the detected b-ions and y-ions originating from the two peptides, the dominant alpha peptide, as well as the beta peptide are indicated. Table I List of Lanthionine Cross-Links Found in Human Virgin Hair Shaft Digest Scan number q value Peptide Peptide Protein 1 Protein 2 19,599 0 .LLEGEEHRLC[57.02]​ EGVGSVN​ VCVSSSRGGVSC​ [57.02]GGLSYSTTPGR NVSVSPIDIGC[57.02] QPGAEAN​ IAPM[15.99] CLLANVAHAN​ [0.98]R HUMAN_K85 HUMAN_K38 6,471 0 .QVVSSSEQLQ[0.98] SCQAEIIELR ISSGCGVTRNFSS​ CSAVAPK HUMAN_K34 HUMAN_K85 4,490 0 .SQQQEPLVCASYQSYFK CEIGNVK HUMAN_K33A HUMAN_K74 4,479 0 .SQQQEPLVCASYQSYFK CEIGNVK HUMAN_K33A HUMAN_K74 3,540 0 .SQQQEPLLCPSYQSYFK MSCRSYR HUMAN_K34 HUMAN_K84