Technical comment on “Synovial fibroblast-neutrophil interactions promote pathogenic adaptive immunity in rheumatoid arthritis”

See allHide authors and affiliations

Science Immunology  31 Jan 2020:
Vol. 5, Issue 43, eaax5672
DOI: 10.1126/sciimmunol.aax5672

This article has a correction. Please see:


Reassessment of citrullinome cargo in neutrophil extracellular traps confirms the presence of citrullinated peptides.

In an article previously published in Science Immunology (1), we reported a previously unidentified mechanism of induction of pathogenic adaptive immunity against citrullinated antigens in rheumatoid arthritis (RA). This mechanism was mediated by an interaction between neutrophil extracellular traps (NETs) and synovial fibroblasts (FLS) that promote the presentation of citrullinated peptides identified in NETs by the FLS to the adaptive immune system. The identification of citrullinated peptides in NETs was performed by a variety of methods described in this and previous manuscripts and included the use of citrulline-specific probes, antibodies, and mass spectrometry (MS) (1, 2).

Following this publication, we were contacted by an investigator with concerns about our MS analysis. Given the magnitude of the mass shift expected between a citrulline and an arginine residue (+0.984 Da), misannotation of citrulline-containing peptides can occur because of fragmentation of misassigned monoisotopic mass peaks. In addition, the +0.984-Da mass increase upon citrullination is identical to the mass increase observed upon protein deamination of asparagine and glutamine residues within a peptide, thereby confounding the correct assignment of citrullinated species. Last, most of the previously identified peptides contained a C-terminal citrulline. Whereas several endogenous proteases, including cathepsin B, cleave after citrulline, trypsin, the protease used in sample preparation, does not (3). These discrepancies prompted us to reevaluate the initial proteomic data more stringently. Upon reevaluation, the spectra were either not consistent with citrullination or not definitive enough to be consistent with a citrullinated residue. Therefore, we generated new samples for analysis. Herein, we report the identification of 58 sites of citrullination on 32 proteins that are present in NETs using a highly stringent data analysis workflow.

In the analysis, healthy neutrophils were purified as described (1), resuspended in RPMI, seeded in 12-well plates, and incubated with 2.5 μM calcium ionophore (A23187, Sigma-Aldrich) or 100 μg of rheumatoid factor for 4 hours at 37°C. Supernatants were carefully removed, and NETs were harvested with micrococcal nuclease (MNase) (10 U/ml) in RPMI for 15 min at 37°C. Supernatant-containing NETs were isolated via centrifugation and stored at −80°C until analysis. Note that the proteomic analysis was performed on NETs and not from cells treated with ionophore or rheumatoid factor; hence, it is not possible to generate control proteomes.

NETs (50 μg) were precipitated with 20% trichloroacetic acid. The pellets were washed with acetone and then resuspended in urea (8 M). Ammonium bicarbonate (100 mM) was then added to the solution. Samples were reduced with dithiothreitol (15 mM) and alkylated with iodoacetamide (12.5 mM). Next, samples were diluted to 1.8 M urea with phosphate buffer (pH 7.4). Samples were then digested overnight at 37°C with Trypsin Gold (2 μg, Promega) and 1 mM CaCl2. Tryptic peptides were separated on a ZORBAX extended C18 column (Agilent) over a 1-hour, biphasic gradient from 0% buffer A (10 mM ammonium bicarbonate plus 5% acetonitrile) to 100% buffer B (10 mM ammonium bicarbonate and 90% acetonitrile). The resulting 96 fractions were pooled to yield 6 fractions for liquid chromatography–tandem MS (MS/MS) analysis.

Fractions were dried and resuspended in 25 μl of 5% acetonitrile/0.1% trifluoroacetic acid. Peptides were eluted from a gravity-pulled analytical column packed with 3-μm (100 Å) Magic C18AQ particles using a biphasic linear gradient from 5 to 60% of B (acetonitrile and 0.1% formic acid) in mobile phase A (water and 0.1% formic acid) over 120 min. Ions were introduced by positive electrospray ionization at 1.4 kV into a Thermo Scientific Q Exactive hybrid mass spectrometer. Raw data files were peak processed by MaxQuant, and identifications were searched by Andromeda against the Human SwissProt database. First-search mass tolerance was 20 parts per million (ppm) for precursors and 20 ppm for fragment ions, and main search tolerance after precursor mass recalibration was 4.5 ppm for precursors and 20 ppm for fragments. All positive citrullination sites were verified manually.

To unambiguously detect and verify the presence of citrullinated arginines, we followed a recently proposed workflow (4). First, we filtered the data to remove C-terminal citrullines. Next, we compared the observed MS isotopic envelopes to predicted envelopes for arginine- and citrulline-containing (+0.984 Da) peptides. If the MS spectra demonstrated a mass shift correlating with citrulline, then we searched for a neutral loss of isocyanic acid (─CNOH, −43.0058 Da) in the MS/MS fragmentation spectra. Neutral loss of isocyanic acid, which is unique to citrulline-containing ion fragments, is frequently observed during high-energy collision dissociation (4). It is important to note that, although deamination of glutamine and asparagine results in a similar mass shift, it does not yield the neutral loss. Using this algorithm, we identified 58 sites of citrullination on 32 proteins (Table 1; spectra data available upon request).

Table 1 Identified sites of citrullination.

r, citrulline; c, alkylated cysteine; m, oxidized methionine; n or q, deamination; m/z, mass/charge ratio; IO, ionophore; RF, rheumatoid factor.

View this table:

Notably, three of the proteins we found as being citrullinated, actin, heat shock protein 90 (HSP90), and plastin-2 (PLSL), were also identified as being citrullinated in RA patient samples (5). In addition, over half of the proteins found in the NETs were also previously found in the proteome of RA synovial fluid and neutrophils treated with ionophores (6). Together, our new data uncovered 15 previously unreported citrullinated proteins present in NETs.

The identified proteins include vimentin and myeloperoxidase (MPO), the latter also identified in the original report (1). Notably, citrullinated vimentin, α/β-tubulin, β-actin, and NCF1 are known RA antigens (1, 4, 7, 8). In addition, we identified S100-A8, plastin, coronin, moesin, and erzin, which are all linked to RA (9). Several inflammation-associated proteins were also observed (S100A8, MPO, LKHA, and HMGB1). Furthermore, our list of citrullinated proteins includes numerous cytoskeletal and cell motility proteins such as PLSL, myosin, vimentin, tubulin, and actin. S100A8 is interesting because several targets of the iNOS-S100A8 transnitrosylase complex (i.e., EZR1, MOES, and VIME) are also present as citrullinated proteins in these NETs. Last, linking our analysis to NETosis is our identification of several citrullinated chromatin binding proteins (HMGB1, HP1B3, and HNRPU) and NCF1, which is a component of the nicotinamide adenine dinucleotide phosphate oxidase complex.

In summary, these findings continue to support the conclusions of our previous publications (1, 2), mainly, that NETs are a source of citrullinated autoantigens. We have identified 58 sites of citrullination in 32 proteins from purified NET samples induced via rheumatoid factor or calcium ionophore. Because citrullination is a modification in such low abundance, generating a comprehensive database of citrulline-containing peptides found in any complex proteome is challenging (10). In addition, the need to verify each identified peptide via manual inspection of MS and MS/MS spectra has hampered the growing field exploring this specific modification. Methods to enrich for citrulline-containing peptides and/or computationally automate the process of picking out true citrulline peptides by neutral loss MS/MS analysis would further facilitate the process of decoding the citrullinome.


Funding: This work was supported by the Intramural Research Program at the National Institute of Arthritis and Musculoskeletal and Skin Disease/NIH (ZIAAR041199). This work was also supported in part by NIH grants R35GM118112 (P.R.T.), F32GM128231 (A.J.S.), 1R01GM117004 (E.W.), and 1R01GM118431-01A1 (E.W.).
View Abstract

Stay Connected to Science Immunology

Navigate This Article