Study subjects
Subjects were recruited from the Studies of the Etiology of RA Lung Study that is described in detail elsewhere [13, 14]. Briefly, this lung study was designed to study the biomarkers of autoimmunity in the lung during different phases of RA development. For this cross-sectional study, stored samples were included from RA subjects (N = 37) and healthy controls (N = 25). RA subjects were all serum ACPA-positive based on anti-cyclic citrullinated peptide (CCP) ELISA testing (QUANTA Lite CCP3.1 IgG/IgA or CCP3 IgG, Inova Diagnostics, San Diego, CA, USA) and had a medical chart review to confirm RA diagnosis by 1987 RA classification criteria and 2010 RA classification criteria or previously diagnosed with anti-CCP+ RA by a board-certified rheumatologist. Healthy controls had no personal or family history of RA and were serum anti-CCP-negative. We also included 46 subjects at-risk for RA defined as having a first-degree relative with RA (N = 20), having serum anti-CCP positivity identified through community or clinical screening (N = 15), or having both (N = 11). At-risk subjects had no clinical or historical evidence of inflammatory arthritis at the time of sample collection.
Study visit
All subjects had a paired collection of blood and sputum. The majority (79%) also had saliva collected prior to sputum collection. Standardized questionnaires were used to obtain demographic information and self-reported histories of smoking and chronic lung disease.
HLA-DRB1 testing
Immunogenetic analysis was performed on DNA isolated from the whole blood of each patient to determine the presence of HLA-DRB1 alleles containing the shared epitope using previously described methodologies [17].
Sputum and saliva collection and processing
Induced sputum was collected using inhaled hypertonic saline and established protocols that have been previously described [13, 14]. A portion of subjects (33 RA, 21 controls, and 31 at-risk) provided an unstimulated saliva sample prior to sputum induction. Saliva samples with volume remaining after anti-PAD4 antibody testing were also tested for ACPA, and this included 28 RA, 19 controls, and 28 at-risk subjects (see below for a description of antibody testing methodologies). All samples were stored at −80°C.
Serum, sputum, and saliva ACPA testing
Paired serum, sputum, and saliva were tested for ACPA using anti-CCP3.1 (IgG/IgA, Inova Diagnostics, San Diego, CA, USA) ELISA. In serum, the cutoff level for anti-CCP3.1 positivity established by the manufacturer was used. For sputum, a cutoff level for anti-CCP3.1 positivity was set at the 95th percentile of sputum anti-CCP3.1 levels in a separate healthy control group (N = 100, median age 37 years, 71% female, 22% ever smokers). For saliva, a cutoff level for anti-CCP3.1 positivity was set at the 95th percentile of salivary anti-CCP3.1 levels in a separate healthy control group (N = 80, median age 48 years, 60% female, 24% ever smokers).
Serum, sputum, and saliva anti-PAD testing
Anti-PAD4 and anti-PAD3/4 antibodies were tested using an established two-tiered quantitative immunoprecipitation method [2]. Briefly, subject sera (1 μl), sputum (10 μl), or saliva (20 μl) were incubated with 1 μl of 35S-methionine-labeled PAD4 or PAD3 generated via in vitro transcription and translation (Promega) for 1 h at 4°C. Radiolabeled immune complexes were immunoprecipitated with 40 μl Protein A beads and washed, and bound antigen was eluted using 2× sodium dodecyl sulfate buffer. The immunoprecipitated proteins were separated by gel electrophoresis, visualized by radiography, and quantified using densitometry. Densitometry values were background-corrected and normalized to a known positive reference serum analyzed in parallel. A normalized value of >0.140 anti-PAD arbitrary units was considered positive for either anti-PAD4 or anti-PAD3 antibodies based on the analysis of known negative samples. Samples that were negative for reactivity to both PADs were considered “anti-PAD negative”; those that were positive for PAD4 reactivity but negative for PAD3 were considered “anti-PAD4 mono-reactive,” and those that were positive for both PAD4 and PAD3 reactivity were defined as “anti-PAD3/4 cross-reactive,” based on our previous work [2]. Isotype-specific immunoprecipitation was also performed, using a similar protocol as described above, for all samples using anti-IgA, IgM, or IgG-coupled agarose beads (Sigma; cat#A3316, A2691, and A9935, respectively) to define the proportion of anti-PAD4 or anti-PAD3/4 antibodies of each isotype present.
PAD4 activity testing
To determine the effect of autoantibodies on PAD4 enzymatic activity, IgG and IgA were co-purified from anti-PAD-negative, anti-PAD4 mono-reactive, and anti-PAD3/4 cross-reactive serum and sputum using an equal mixture of Protein A/G agarose (Pierce; cat#20423) and Peptide M agarose (InvivoGen; cat# gel-pdm-2) beads. This mixture purifies predominantly all four IgG subclasses and IgA isotypes, with minimal purification of IgM. The concentration of total Ig was determined by NanoDrop (Thermo), >95% purity confirmed by Coomassie stain, and composition confirmed by immunoblot with goat anti-human IgG antibody and rabbit anti-human IgA antibody (Jackson Laboratories). The effect of purified Ig on PAD4 activity was evaluated using recombinant human PAD4, purified in-house as previously described [18]. PAD4 (10 nM) was pre-incubated with 1 μM purified Ig for 45 min at 4°C, followed by incubation with 700 μM histone H3 substrate for 3 h at 37°C at increasing calcium chloride concentrations (i.e., 0.2 and 2 mM), as previously described [2]. Citrullination of histone H3 was evaluated by anti-citrullinated histone H3 immunoblotting (ab5103, Abcam).
Statistical analysis
Subject characteristics were compared between the groups using Kruskal-Wallis testing for age and chi-square/Fisher’s exact test for dichotomous variables, including the prevalence of anti-PAD antibody positivity. All analyses were performed using the SPSS software, version 25, and figures were generated using GraphPad Prism version 8.