Global Analysis of Citrullination
Prior to efficient enrichment [103, 104], global analysis of citrullination was challenging. Generally, analysis of individual proteins was performed after confirmation of citrullination by Western blot as demonstrated by Bhattacharya et al. . They analyzed healthy and glaucomatous optic nerve tissue; it was observed by Western blot analysis that both PAD2 expression and citrullina- tion levels increased in the glaucomatous optical nerve tissue compared to healthy tissue. They quantified the levels of PAD2 in both the healthy and glaucomatous tissues and observed approximately an eightfold increase in PAD2 in the glaucomatous tissue. The tissue samples were homogenized and the proteins separated by SDS-PAGE, followed by trypsin digestion and LC-MS/MS analysis. Two hundred and fifty proteins were identified, 68 of which were only identified in the glaucomatous samples. Following the proteomics analysis, immunoprecipitation using an anticitrulline antibody was performed on a glaucomatous sample. The immunoprecipitate was also analyzed by LC-MS/ MS, and 36 proteins were identified. Several of these proteins had previously been identified as citrullinated including annexin A2 , MBP , and histone H4 . As discussed earlier, citrullination is known to demethylate monomethylarginine. Bhattacharya et al. performed Western blots using an antibody specific to protein methylarginine. Proteins containing methylargi- nine were identified in the control groups but not in the glaucomatous samples. This result adds to the evidence that citrullination is involved in the regulation of monomethylation and therefore the regulation of gene expression.
Hermansson et al.  analyzed synovial tissue from RA patients. The presence of citrullination was detected by antibody staining of four samples. Though the analysis was of a tissue digest and therefore very complex, the target of the experiment was to identify citrullinated sites on fibrinogen. To achieve this, purified fibrinogen was separately citrullinated using a rabbit PAD enzyme (PAD type not specified) and digested using Lys-C. LC-MS/MS analysis followed by a Mascot search of this digest was then used to identify sites of citrullination. They identified 33% of the arginine residues from the fibrinogen chains (both unmodified and citrullinated). The analysis resulted in a list of citrullinated peptides together with accurate mass and time measurements. In order to use the identified peptides as markers in the in vitro study, several criteria had to be met: (i) no asparagine or glutamine residues, (ii) the noncitrullinated peptide had to be identified in the Mascot search with high confidence, and (iii) the citrullinated peptide had to be identified with a Mascot ion score >25 and greater than 95% confidence. Two citrullinated peptides met these criteria: one from the a-fibrinogen (559-575) and one from p-fibrinogen (52-77). For two of the tissue samples where the citrullinated a-fibrinogen peptide was identified, the citrullination occupancy was calculated as 1.4% and 2.5%, respectively (occupancies calculated by dividing the intensity of the citrullinated peptide by the intensity of the unmodified counterpeptide + citrullinated peptide). The citrullinated peptide identified in p-fibrinogen was identified in one of the samples with occupancy of 1.2%.
Bennike et al.  analyzed the synovial fluid from a patient with RA. This study was one of the first large-scale analyses of citrullinated peptides from a complex sample. The proteins were digested with trypsin and separated over a 120 min LC gradient. No chemical modifications were employed, and a standard CID fragmentation method was used. As discussed earlier, citrullinated arginine hinders trypsin digestion. Using this feature, they interrogated the results from an LC-MS/MS analysis and database search of a synovial fluid trypsin digest. In total, 58 citrullinated peptides were identified following a protein database search employing the Mascot algorithm. On manual inspection, 14 of the peptides were found to be false positives (24%), 64% of the sites were unambiguously identified (fragment ions for N-CO bonds either side of the modified arginine), and the remaining 12% of sites were identified by flanking fragment ions (the N-CO bonds N- and C-terminal to citrulline are not cleaved, but sufficient fragment ions are observed to localize the site of modification). Over 360 proteins were identified from synovial fluid, highlighting the low stoichiometry and specificity of citrullination.
Christophorou et al.  measured the differential levels of citrullination in mouse embryonic stem (ES) cells that overexpressed human PAD4 compared to normal mouse ES cells. The quantification was performed by the use of stable isotope labeling by/with amino acids in cell culture (SILAC) [56, 110]. Briefly, in SILAC quantification, one cell population is grown in “light” media (arginine and lysine have 12C or 14N), and another cell population is grown in “medium”/“heavy” media (arginine + 13C6 and 15N4 and lysine 13C6). The isolated proteins are mixed, proteolytically digested, and analyzed by LC-MS/MS. The light and heavy peptides coelute, and extracted ion chromatograms for each pair of peptides can be used to quantify the relative abundance of the peptides. SILAC allows the analysis of three samples at once (light, medium, and heavy). In this experiment, mouse ES cells expressing an empty vector were grown in “light” lysine, and mouse ES cells that overexpressed human PAD4 were grown in “heavy” lysine (13C6). The sample was digested with Lys-C and analyzed by LC-HCD-MS/MS over a 2 h LC gradient. In the four citrullinated HCD spectra presented in this work, the peak corresponding to neutral loss of HNCO, which is observed in CID, was not present. This observation suggests that the modification remains on the arginine, increasing the probability of successfully identifying the site of citrullination. In total, 162 cit- rullinated peptides were quantified in this first example of large-scale citrullination quantitation by mass spectrometry. Thirteen of the identified citrullinated peptides have localization probabilities of less than 75%. The localization probability used here calculates the chance that a given peak that allows localization was a random match. A localization probability of 75% means that there is a 25% chance that the citrullination is not located to where the search software has assigned it. Given the challenges associated with citrul- lination site assignment, citrullinated peptide assignments may need to be manually validated to ensure no false sites are reported.
Van Beer et al.  were able to confidently identify 149 citrullinated peptides from the synovial fluid of two patients with RA. All citrullinated peptides were manually validated. Synovial fluid from 80 patients was separated into soluble (supernatant) and insoluble (pellet) fractions. The two fractions from each patient were analyzed by Western blotting to identify the two fractions (one soluble and one insoluble) with the greatest levels of citrullinated protein. The soluble and insoluble fractions with the greatest levels of citrullination (from different patients) were depleted for albumin, resulting in two pellet fractions and a supernatant (predominantly albumin) for each sample . The proteins were solubilized and separated by SDS-PAGE. The resulting lanes were dissected into 18 even slices. Each gel slice was analyzed by LC-MS/MS after trypsin digestion using CID fragmentation followed by a Mascot database search. The combined database search of all samples identified 192 proteins with Mascot scores of 30 or greater. Of these, 40 and 45 were identified solely in the soluble and insoluble fractions, respectively. Of the 192 proteins, 53 were identified as citrullinated. Thirty five of these were identified from the soluble fraction. Only six citrulli- nated proteins were identified in both the patient fractions. Of the modified proteins identified in both fractions, five were identified with multiple citrullination sites in common between the fractions. In p-actin, 11 citrullination sites were identified solely in the supernatant sample. In addition, some peptides identified as citrullinated in the supernatant were only observed as unmodified in the pellet.
Fert-Bober et al.  performed a large-scale data-independent (sequence window acquisition of all theoretical fragment ion spectra (SWATH) ) quantitative proteomics experiment on human heart tissue from three different groups: ischemic heart disease (n = 10), idiopathic cardiomyopathy (n = 10), and nonfailing hearts (n = 10). No modification of the samples was performed and the citrullinome was compared. SWATH is a two-step process: initially, the sample is analyzed by a traditional data-dependent acquisition (DDA) method. This analysis provides a peptide ion library; this library of accurate mass, time, and fragment measurements is used in step 2. The mass spectrometer rapidly cycles through a set of mass ranges, isolating and fragmenting everything in a specified m/z window (m/z width 25). The mass spectrometer moves to the next mass window and repeats over the whole mass range. The cycle usually takes 2-4 s
(instrument, m/z range, and m/z window size dependent). Peptides are therefore fragmented multiple times as they elute from the LC column, allowing for quantitation and identification. Myofilaments were isolated using IN sequencing . Proteins are separated into three fractions based on their solubility at various pH levels. IN sequencing was developed specifically for proteomics analysis of the heart tissue. The heart tissues were fractionated into myofilament- and cytosolic-enriched fractions. To maximize the citrullinome coverage, the samples were treated with a PAD cocktail containing all five PADs. Each sample (both fractions) was analyzed with DDA and data-independent acquisition (DIA) methods using a 2 h 5-35% acetonitrile gradient. To create the SWATH ion library of verified citrullinated peptides, each citrullinated peptide was required to have a partner unmodified peptide with a retention time shift of at least 5 min. For citrullinated peptides, which had unmodified partner peptides with less than 5 min retention time shift or no partner peptide, transitions were manually selected to ensure unambiguous assignment of citrullination, not deamidation. SWATH analysis resulted in the identification and relative quantitation of 304 citrullination sites from 145 proteins. Of the 304 citrullination sites identified, 53 were altered in the heart failure samples compared to the nonfailing controls. A set of citrullinated peptides were identified, which were upregulated in ischemic heart disease and downregulated in idiopathic cardiomyopathy.