Summary

Title

Identification of In Vivo HLA-DR-Presented Self-Peptides (T Cell Epitopes) in Synovial Tissue, Synovial Fluid and Peripheral Blood in Patients with Rheumatoid Arthritis or Lyme Arthritis

Description

HLA-DR molecules are highly expressed in synovial tissue (ST), the target of the immune response in chronic inflammatory arthritides, including in rheumatoid arthritis (RA) and Lyme arthritis (LA). In the current study, we identified HLA-DR-presented self-peptides (T cell epitopes) in ST, synovial fluid mononuclear cells (SF), and peripheral blood mononuclear cells (PB) from five patients with RA and eight with LA. Altogether, from 22 samples, 1,532 non-redundant HLA-DR-presented peptides were identified that were derived from 778 source proteins. Moreover, as the study progressed, application of newer, high sensitivity LC-MS instruments resulted in the identification of larger numbers of HLA-DR-presented peptides. Surprisingly, the source proteins for HLA-DR-presented peptides were as likely to have intracellular as extracellular locations. Although the number of peptides identified was greater in ST than in SF or PB, 68% of the peptides identified in SF were found in ST, and 55% of the peptides in PB were found in ST. Two RA patients had the same HLA-DR genotype (DRB1*0401 and 0101), and 44% of the peptides identified in these two patients were the same. In contrast, among the patients with RA or LA who had no shared HLA-DR alleles, only 6% of the peptides were the same. In the RA patients, HLA-DR-presented peptides were identified from source proteins that are thought to serve as potential autoantigens: collagen, enolase, fibrinogen, fibronectin, immunoglobulin and vimentin. Interestingly, peptides from these same source proteins were also commonly found in LA patients. However, citrullinated T cell epitopes were not identified in any patient. Thus, LC-MS/MS techniques are now sensitive enough to identify large numbers of T cell epitopes from tissue or fluids in individual patients. Importantly, analysis of SF or PB allowed us to identify a broader range of HLA-DR-presented self-peptides from individual patients and from patients seen earlier in the disease.

Sample Processing Protocol

ST, SF, and PBMC Preparation — ST samples (8-10 g) were prepared using the protocol described previously (19). Heparinized SF and PBMC (~3 to 8 x 107 cells) were obtained by Ficoll-Hypaque (MP Biomedicals) separation and stored in liquid nitrogen prior to analysis. On the day of purification, SF and PB samples were quickly thawed by placing in a 37 °C water bath for two minutes. Cells were washed twice in phosphate buffered saline (PBS) (Life Technologies) by centrifuging at 800 × g for 10 min. The pellet was resuspended with 10 mL lysis buffer: 150 mM sodium chloride, 20 mM tris(hydroxymethyl)aminomethane -HCl (pH 8.0), 5 mM ethylenediaminetetraacetic acid disodium solution, 0.04% sodium azide, 1 mM 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride, 10 µg/mL leupeptin, 10 µg/mL pepstatin A, 5 µg/mL aprotinin, and 1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate. Immunoaffinity purifications of HLA-DR complexes from SF and PBMC were performed using the protocol previously described for synovial tissue samples. LTQ-Orbitrap XL mass spectrometer: A nanoAcquity UPLC system (Waters Corp., Milford, MA) was coupled to an LTQ-Orbitrap XL mass spectrometer (ThermoFisher Scientific, San Jose, CA) through a TriVersa NanoMate ion source (Advion, Ithaca, NY). Peptide enrichment was performed with a trapping column (180 µm × 20 mm, 5 µm 100 Å Symmetry C18, Waters Corp) at a flow rate of 15 µL/min for 1 min, and separation was achieved with a capillary column (150 µm × 10 cm, 1.7 µm 130 Å BEH C18, Waters Corp). Buffer A contained 1% acetonitrile (ACN) and 0.1% formic acid in water, and buffer B contained 1% water and 0.1% formic acid in ACN. A linear gradient of buffer B from 2% to 40% over 52 min was used at a flow rate of 0.5 µL/min. The capillary voltage was set to 1.7 kV using the NanoMate, and the capillary temperature was set to 120 °C. The mass spectra were recorded over the range of m/z 300-2000 at a resolution of 60,000 (the width of the peak at half its maximum height at m/z 300) at a scan rate of approximately 1.2 s/spectrum. Tandem MS was performed for the five most abundant, multiply-charged species in the mass spectra that had a signal intensity threshold of 8000 NL. The normalized collision energy for collision-induced dissociation (CID) was set to 35%, and helium was used as the collision gas. MS/MS spectra were recorded with LTQ XL linear ion trap. Q Exactive plus mass spectrometer: A nanoAcquity UPLC system was coupled to a Q Exactive plus mass spectrometer (ThermoFisher Scientific, San Jose, CA) through a TriVersa NanoMate ion source. The same peptide enrichment and separation columns, HPLC solvents and gradients, and NanoMate capillary voltage and temperature used on the LTQ-Orbitrap XL mass spectrometer were also applied on the Q Exactive plus mass spectrometer. Survey MS scans were performed over the range m/z 400-2000 with a resolution of 70,000. The 15 most abundant, multiply-charged ions were selected for higher-energy collisional induced dissociation (HCD) MS/MS with a resolution of 17,500, an isolation width of 1.5 m/z, and a normalized collision energy of 27%. All spectra were recorded in profile mode. 6550 iFunnel QTOF LC/MS: Two 1260 HPLC systems were coupled to a 6550 iFunnel QTOF mass spectrometer through a chip cube interface (Agilent Technologies, Santa Clara, CA). A Polaris-HR-3C18 chip was used for peptide enrichment and separation. The flow rate on the enrichment column was set at 2 µL/min with sample flush volume set as 4 µL. With the same solvents used on the Waters UPLC systems, a gradient of buffer B from 2% to 5% over 0.1 min, 5% to 20% over 74.9 min, 20% to 40% over 10 min, 40% to 90% over 10 min, stayed at 90% buffer B for 5 min, returned to 2% buffer B and stayed for 10min was used at a flow rate of 0.3 µL/min for peptide separation. The capillary voltage was set to 1.7 to 1.95 kV, and the gas temperature was set to 225 °C. The survey mass spectra were recorded over the range of m/z 295-1700 with acquisition rate of 8 spectra/second. Tandem MS was recorded over the range of m/z 50-1700 with acquisition rate of 3 spectra/second. Precursor ion isolation was performed with narrow isolation window of 1.3 m/z. Tandem MS was performed for the 20 most abundant, multiply-charged species in the mass spectra that had a signal intensity threshold of 5000 counts or a relative threshold of 0.001%. The collision energy for CID was set as using the formula [3.1×(m/z)/100+1] for doubly charged precursor ions and [3.6×(m/z)/100-4.8] for triply and higher charged precursor ions. Active exclusion was enabled after one tandem spectrum and released after 0.15 min. All spectra were recorded in centroid mode.

Data Processing Protocol

Protein Database Searching — In our original study, the International Protein Index (IPI) database was used to identify HLA-DR-presented peptides from ST samples from two patients each with RA or LA (identified here as patients RA1, RA2, LA1 and LA2) (19). Because the IPI database is now closed and no longer maintained, the spectra of these four patients were reanalyzed using the UniProt protein database and the findings were compared to our original study. Using the UniProt database, ~96-98% of the peptides identified were the same as those identified using the IPI database. Therefore, in the current study, HLA-DR-presented self peptides were identified by searching the Mascot generic file from each sample against UniProt human database concatenated with a decoy database. The decoy databases were generated by randomizing each protein sequence in the database using the Perl script decoy.pl (Matrix Science). Mascot 2.4.0 (Matrix Science), OMSSA Browser 2.1.1 (NCBI), and X!Tandem tandem-win-12-10-01-1 (The Global Proteome Machine Organization www.thegpm.org) were used for database searches. Cysteinylation of cysteine, deamidation of glutamine and asparagine, pyroglutamic acid from amino-terminal glutamine and glutamic acid, and oxidized methionine were specified as the variable modifications in all searches. In addition, evaluation for the post-translational modification of arginine to citrulline was also included as a variable modification. For this purpose, peaks exhibiting the difference in molecular weight between arginine and citrulline (0.98402 Da) were searched for each arginine residue in the database. For data acquired using LTQ-Orbitrap XL mass spectrometer, precursor ion error tolerance was set as 0.01 Da and fragment ion error tolerance was set as 0.5 Da; for data acquired using Q Exactive plus mass spectrometer, precursor ion error tolerance was set as 10 ppm and fragment ion error tolerance was set as 20 mmu; for data acquired using 6550 iFunnel QTOF mass spectrometer, precursor ion error tolerance was set as 20 ppm and fragment ion error tolerance was set as 50 mmu. “No enzyme” was specified for all the protein database searches. Mascot score cutoff was set at ≥20; OMSSA e-value cutoff was set to ≤0.01; and X!Tandem e-value cutoff was set at ≤10. In addition, consensus identifications of peptides by at least two search programs were acquired using a Microsoft Access query. Removal of redundant peptide sequences - Unique consensus peptides were submitted to a software generated in-house that collates and sets aside redundant peptides that have amino acid sequences overlapping in a core sequence. Peptide sequences with at least six continuous overlapping amino acid residues were considered redundant, and the peptide with the longest amino acid sequence was output from the software program. The same software was also used to compare the consensus peptides among different samples. Peptide sequences with at least six continuous overlapping amino acid residues were considered consensus. Identification of source proteins – All of the unique consensus peptides identified by at least two protein database search engines were submitted to another software program generated in-house that screens the peptide sequences with the UniProt human protein database. This database excludes decoy sequences. All the accession numbers of the source proteins containing the screened peptide sequence were shown. Although a number of peptides could have been derived from more than one source protein, the proteins were usually closely related. Therefore, a given peptide was counted as having only one source protein. The consensus source proteins among different samples were acquired by matching the first 252 characters (the maximum allowed) of each source protein accession numbers using a Microsoft Access query. Go annotation for protein location — To determine protein location, the first protein accession number from each source protein was submitted to the Software Tool for Rapid Annotation of Proteins (STRAP) 1.5 (Boston University School of Medicine, Boston, MA) (24, 25). The output was condensed manually to one of four groups: extracellular, plasma membrane, intracellular, or multiple locations. Statistics – All p values are two-tailed. A p value of ≤0.05 was considered statistically significant.

Contact

Qi Wang, Boston University
Catherine E. Costello, Center for Biomedical Mass Spectrometry, Department of Biochemistry, Boston University School of Medicine ( lab head )

Submission Date

14/10/2015

Publication Date

13/10/2016

Publication

    Wang Q, Drouin EE, Yao C, Zhang J, Huang Y, Leon DR, Steere AC, Costello CE. Immunogenic HLA-DR-Presented Self-Peptides Identified Directly from Clinical Samples of Synovial Tissue, Synovial Fluid, or Peripheral Blood in Patients with Rheumatoid Arthritis or Lyme Arthritis. J Proteome Res. 2016 Oct 11 PubMed: 27726376