Journal of Immunological Methods 405 (2014) 15–22
Contents lists available at ScienceDirect
Journal of Immunological Methods journal homepage: www.elsevier.com/locate/jim
Research paper
Generation of monoclonal antibodies against peptidylarginine deiminase 2 (PAD2) and development of a PAD2-specific enzyme-linked immunosorbent assay Dres Damgaard a, Yaseelan Palarasah b, Karsten Skjødt b, Anca I. Catrina c, Sanne M.M. Hensen d, Ger J.M. Pruijn d, Claus H. Nielsen a,⁎ a
Institute for Inflammation Research, Department of Infectious Diseases and Rheumatology, Copenhagen University Hospital, Rigshospitalet, Denmark Department of Cancer and Inflammation Research, Institute of Molecular Medicine, University of Southern Denmark, Denmark c Rheumatology Unit, Department of Medicine, Karolinska University Hospital and Karolinska Institutet, Sweden d Department of Biomolecular Chemistry, Institute for Molecules and Materials, Nijmegen Center for Molecular Life Sciences and Netherlands Proteomics Centre, Radboud University Nijmegen, Nijmegen, The Netherlands b
a r t i c l e
i n f o
Article history: Received 22 November 2013 Received in revised form 18 December 2013 Accepted 20 December 2013 Available online 31 December 2013 Keywords: Peptidylarginine deiminase PAD2 Citrullination Monoclonal antibody ELISA
a b s t r a c t The enzyme peptidylarginine deiminase 2 (PAD2) has been associated with inflammatory diseases, such as rheumatoid arthritis and neurodegenerative diseases including multiple sclerosis. To investigate the association of various diseases with extracellular PAD2, we raised monoclonal antibodies (mAbs) against rabbit PAD2 and evaluated their cross-reactivity with human PAD2 by indirect enzyme-linked immunosorbent assay (ELISA), western blotting and immunohistological staining of inflamed synovial tissue. Moreover, we established a sandwich ELISA detecting human PAD2, based on two different monoclonal antibodies, mAbs DN2 and DN6. The assay had a lower detection limit of 200 pg/mL in serum and plasma samples, and showed dilution linearity and recovery ranging from 95 to 106%. The mAbs and the ELISA showed isotype specificity for PAD2. Circulating PAD2 was found in 8/28 (29%) serum samples from healthy donors. In conclusion, several of our mAbs proved useful in western blotting and immunohistochemistry, and the ELISA described here reliably measures PAD2 levels in blood. This allows investigation of PAD2 as a possible biomarker and further investigation of PAD2's involvement in various inflammatory diseases. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Citrullination (deimination) designates a posttranslational modification in proteins resulting in a conversion of the amino acid residue arginine into the non-standard residue citrulline. This reaction is catalyzed by enzymes of the peptidylarginine deiminase (PAD) family (Vossenaar et al., 2003) and is important for homeostatic processes such as keratinocyte differentiation (Senshu et al., 1996), maintenance of myelin sheath insulation
Abbreviations: ELISA, enzyme-linked immunosorbent assay; LOD, limit of detection; mAb, monoclonal antibody; PAD, peptidylarginine deiminase. ⁎ Corresponding author. E-mail address: [email protected] (C.H. Nielsen). 0022-1759/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jim.2013.12.008
(Harauz and Musse, 2007) and formation of neutrophil extracellular traps (NETs) (Li et al., 2010). PADs and citrullinated proteins are involved in a number of autoimmune- and inflammatory diseases, and have especially been drawn into focus after discovery of their pathogenic role in rheumatoid arthritis (RA) (Schellekens et al., 1998). During inflammation, including that associated with RA, proinflammatory stimuli and increased cell death allow protein citrullination to occur; either as a result of calcium-influx, followed by intracellular citrullination (Vossenaar et al., 2004; Asaga et al., 1998; Orrenius et al., 2003), or by leakage of PADs from dying cells into synovial fluid, where the calcium content is sufficient for PAD to have enzymatic activity (NakayamaHamada et al., 2005), followed by citrullination of extracellular
16
D. Damgaard et al. / Journal of Immunological Methods 405 (2014) 15–22
proteins (Kinloch et al., 2008). Extracellular PAD2 may originate from dying macrophages or mast cells (Vossenaar et al., 2004; Arandjelovic et al., 2012). Approximately 80% of RA patients carry MHC class II molecules containing the ‘shared epitope’ (SE), capable of binding citrullinated self-antigens (Gregersen et al., 1987). A similar proportion of patients produce anti-citrullinated protein antibodies (ACPAs) (Snir et al., 2009; Huizinga et al., 2005; van Venrooij et al., 2011), which have become a major diagnostic and prognostic marker of RA (van Venrooij et al., 2011). A significant positive correlation exists between the serum titer and clinical, biologic, and radiologic data related to ACPA-positive RA activity and severity (Forslind et al., 2004). Increased levels of PADs and citrullinated proteins have also been found in neurodegenerative diseases, such as multiple sclerosis (Wood et al., 2008). Hypercitrullination of myelin basic protein (MBP) disrupts the formation of normal myelin sheath structure and makes MBP more susceptible to be degraded (Pritzker et al., 2000). Five members of the human PAD enzyme family exist; PAD1, PAD2, PAD3, PAD4 and PAD6 (Vossenaar et al., 2003). PAD2 and PAD4 are the isotypes associated with RA, as their expression/ presence has been demonstrated by semi-quantitative techniques, such as immunoblotting and immunohistochemistry, in RA synovium, synovial fluid cells and extracellularly in synovial fluid (Foulquier et al., 2007; Chang et al., 2005; Kinloch et al., 2008). PAD2 and PAD4 differ with respect to citrullination efficiency of different substrates and optimal enzymatic conditions (Nakayama-Hamada et al., 2005; Darrah et al., 2012). Their differential roles in the pathogenesis of RA have not been elucidated. It is also unclear which role intracellular and extracellular citrullination, respectively, play in the pathogenesis, and whether both events are required for the initiation or maintenance of the disease. Specific monoclonal antibodies are needed to investigate the pathogenic role of PAD2 (Cherrington et al., 2012). We generated a panel of monoclonal anti-PAD2 antibodies, and developed a sandwich ELISA specific for PAD2 allowing investigation of disease associations with PAD2 in blood and other body fluids. 2. Material and methods
(Sigma–Aldrich, Brøndby, Denmark). The mice received an intravenous injection of 25 μg rPAD2 administered with adrenalin four days prior the fusion. The fusion and selection were based on the principles described by Kohler and Milstein (Kohler and Milstein, 1975). The SP2/0-AG14 myeloma cell line (Palarasah et al., 2010) was selected as fusion partner. Positive clones were identified by ELISA using microtiter plates (Maxisorp; Nunc, Roskilde, Denmark) coated with 0.25 μg/mL rPAD2 in coating buffer (35 mM NaHCO3, 15 mM Na2CO3, pH 9.6). Cells from positive wells were cloned at least four times by limiting dilution. Single clones were grown in culture flasks in RPMI-1640 medium (RPMI) (Sigma–Aldrich) containing 10% foetal calf serum (FCS), and mAbs were purified from culture supernatant by Protein G affinity chromatography using the Äkta fast-performance liquid chromatography (FPLC) system according to the manufacturer's instructions. The eluate was dialyzed against phosphate-buffered saline (PBS), pH 7.4. Subsequently, the purified mAbs were dialyzed against 0.1 M carbonate buffer, pH 8.5, and biotin N-hydroxysuccinimide ester (BNHS) (Sigma–Aldrich, Glostrup, Denmark) was added 1:6 (w:w) of the total protein amount. The mixture was incubated with gentle mixing for 3 h at room temperature (RT). Unbound biotin was removed by dialysis against PBS. 2.3. Western blotting Gel-electrophoresis was performed on 4–12% Novex Bis-Tris gels (Invitrogen, CA, USA) using the NuPAGE® system (Invitrogen) according to the manufacturer's recommendations. Proteins were electro-blotted onto polyvinylidene difluoride membranes (PVDF-HyBond; Amersham Bioscience, Uppsala, Sweden). The membrane was blocked in washing buffer (PBS, 0.05% Tween-20, pH 7.4) and sliced into strips. Antibody culture supernatants (2 mL) were added 1:1 in wash buffer and incubated O/N at 4 °C. Horseradish peroxidase (HRP)-conjugated rabbit anti-mouse IgG antibodies (Dako Denmark A/S, Glostrup) were added to a dilution of 1:3000 in washing buffer, followed by incubation 1 h at RT. Finally, the strips were washed, incubated 10 min with acetate buffer (50 mM CH3COOH, 33.75 mM NaOH, pH 5.0) and stained in carbazole staining solution (0.04% 3-amino-9-ethylcarbazole and 0.015% H2O2) in acetate buffer.
2.1. PADs 2.4. Immunohistochemistry PAD2 from rabbit skeletal muscle (rPAD2) and recombinant murine PAD2 (rmPAD2) were purchased from Sigma, St Louis, MO. Recombinant human PAD (rhPAD) of various isotypes, i.e. rhPAD2, rhPAD3, rhPAD4 and rhPAD6, was produced and purified as described previously (Raijmakers et al., 2007). Different deletion mutants of rhPAD2 were prepared likewise, i.e. C254 (amino acids 1–254 of rhPAD2), I385–463 (rhPAD2 with deletions in the catalytic site), N165 (amino acid sequence from 165 to the C-terminus), N343 (amino acid sequence from 343 to the C-terminus). 2.2. Generation and purification of PAD2 mAbs Three BALB/cxNMRI mice were immunized subcutaneously three times with 25 μg of rPAD2 adsorbed to Al(OH)3 (Alhydrogel 2.0%, Brenntag Biosector A/S, Frederikssund, Denmark) and mixed 1:1 with Freund's incomplete adjuvant
Synovial biopsy specimens from inflamed synovium were snap frozen within 30 s in liquid isopentane, stored at −70 °C, and then embedded in OCT compound (TissueTek; Sakura Finetek, Zoeterwoude, The Netherlands) when sectioning was performed. Cryostat sections from the biopsy samples were placed on SuperFrost Plus slides (Menzel-Glaser, Braunschweig, Germany) and air dried for 30 min. Sections were initially fixed for 20 min with freshly prepared 2% (volume/ volume) formaldehyde (Sigma, St. Louis, MO), dissolved in phosphate buffered saline (PBS; pH 7.4) at 4 °C, washed in PBS, and then left to air-dry before storage at − 70 °C. The staining procedures have been described in detail previously (Makrygiannakis et al., 2012). Either 0.1% saponin or 0.3% Triton-X100 was used for cell permeabilization. All antiPAD2 antibodies were used as primary antibodies in a dilution of 1:3 (estimated to correspond to 1.25–5 μg/mL).
D. Damgaard et al. / Journal of Immunological Methods 405 (2014) 15–22
As positive control, polyclonal anti-PAD2 antibodies (ROI002) (CosmoBio; Tokyo, Japan) were used, and as negative control an irrelevant, in house-raised mAb against chicken complement component 3 (C3) was used. After three washes, the slides were blocked with 1% normal horse serum, followed by incubation for 30 min at RT with biotinylated horseanti-mouse IgG antibodies (Vector Laboratories; Burlingame, CA), diluted 1:640 in wash buffer containing 1% horse serum. The polyclonal anti-PAD2 positive control was diluted in PBS containing 0.3% Tween-20, 3% BSA (0.02 μg/mL). The slides were then incubated with ABC Elite streptavidin–biotin complex (Vector Laboratories) for 40 min at RT, and developed using diaminobenzidine-containing peroxidase Substrate Kit SK-4100 (Vector laboratories) for 7 min at RT in dark. Counterstaining was performed using hematoxylin (Vector Laboratories). Photography was carried out with identical exposures and concurrent processing. The staining was scored as: 1) not reacting (binding pattern similar to that of the irrelevant control and of slides with no primary antibody added), 2) weakly reacting (a weak staining pattern) and 3) strongly reacting (staining pattern similar to that of the positive anti-PAD2 control).
17
2.7. Evaluation of accuracy To assess the variation of the ELISA, plasma- and serum samples were spiked with rhPAD2 and applied in 2-fold serial dilutions in the assay. To determine the intra-assay variation, plasma and serum samples were measured ten times. To determine inter-assay variation, three samples were tested in the assay on 5 different days. Recovery was tested by spiking samples with rhPAD2 at different concentrations covering the whole working range of the assay. The mean values, standard deviation (SD), and the coefficient of variation (CV = SD/mean × 100%), were calculated for each of these experiments. 2.8. Calculations and statistical analysis All standards and samples were measured in duplicates. Concentrations were calculated by regression analysis for the standard curve using four-parameter logistic curve-fitting on MARS software. GraphPad Prism 5.0 (GraphPad Software, CA) was used for statistical analyses. 2.9. Stability of PAD2 in samples
2.5. ELISA for PAD2 Maxisorp plates were coated overnight at 4 °C with 100 μL mAb DN2 (1 μg/mL). Wells were washed three times and blocked with washing buffer (PBS, 0.05% Tween-20, pH 7.4) for 20 min at room temperature. The calibrator, quality control, and samples (see below) were diluted in dilution buffer (PBS, 0.5% Tween-20, 2% bovine serum [Sigma], 20 μg/mL Mouse IgG isotype control [Novus Biologicals; Cambridge], UK, pH 7.4) and incubated for 90 min in the wells, at room temperature. Following three washes, the wells were incubated for 1 h at room temperature with 100 μL biotinylated mAb DN6 (1 μg/mL) in washing buffer. After another three washes, the wells were incubated at room temperature with 100 μL of streptavidin-conjugated HRP (Invitrogen; Invitrogen, CA, USA) diluted 1:3000 in washing buffer for 45 min. Finally, the plates were washed three times and incubated with 0.4 mg of o-phenylene-diamine (Kem-En-Tec; Taastrup, Denmark) per mL developing buffer (35 mM citric acid, 65 mM Na2PO4, pH 5). After 10 min the color development was stopped with 1 M H2SO4, and absorbance was measured at 490–650 nm using SPECTROstar nano Microplate Reader (BMG Labtech; Ortenberg, Germany) and the data were processed using MARS software (BMG Labtech). An irrelevant, in house-raised mAb, anti-chicken C3, was used as negative control, either as capture or as detecting mAb. 2.6. Calibrator and controls The calibrator consisted of rhPAD2, spiked in dilution buffer. The quality control consisted of a pool of normal serum spiked with rhPAD2, and stored as aliquots at −80 °C. As isotype controls, i.e. rhPAD3, rhPAD4 and rhPAD6, were used. Moreover, we also tested rmPAD2. A two-fold serial dilution of these reagents was used to generate 10-point calibrator curves with a range from 25 to 0.05 ng/mL. A four-parameter fit model was applied to estimate the concentration of unknown samples. The calibrator was stored as aliquots at −80 °C.
Serum samples were stored at 4 °C or 20 °C for 24 or 48 h to test stability of PAD2 in non-frozen samples. Furthermore, serum samples were subjected to nine freeze–thaw cycles (− 80 °C and room temperature, respectively) and compared with respect to PAD2 levels. 2.10. Serum and plasma samples Serum and plasma were isolated from peripheral venous blood, drawn in dry tubes or EDTA-tubes (BD; Plymouth, UK). Serum samples were collected from 28 self-reported healthy, anonymous donors (14 men, 14 women, age: 45 ± 13 years), attending the Blood Bank at Copenhagen University Hospital Rigshospitalet. 3. Results 3.1. Human PAD2 specific monoclonal antibodies After immunization of mice with rPAD2 and generation of hybridoma cells, culture supernatants were screened for reactivity with rPAD2 in an indirect ELISA. Hybridoma cells in positive wells were sub-cloned until positive culture supernatants were formed by single clones. From the screening and sub-cloning procedure, 35 hybridomas were successfully isolated (mAbs DN1–35). These were screened for reactivity against rhPAD2 by indirect ELISA (Fig. 1A) and western blotting (Fig. 1B). Thirty three of these hybridoma supernatants and 23 out of 31 tested samples were reactive with rhPAD2 in ELISA and western blotting, respectively. The mAbs DN2 and DN6 were found to bind to an epitope in the N-terminal region between amino acids 1–165, demonstrated by western blotting using different deletion mutants of rhPAD2 (Fig. 1C). Immunocytochemistry performed on synovial tissue from an ACPA-positive RA patient showed that a large majority of the tested clones had staining patterns similar
18
D. Damgaard et al. / Journal of Immunological Methods 405 (2014) 15–22
Fig. 1. Reactivity of mAbs with human PAD2. Culture supernatants from 35 hybridomas producing mAbs against rabbit PAD2 were tested for reactivity with rhPAD2: (A) Absorbance at 490 nm values, after allowance for background (absorbance at 650 nm), is shown for six representative mAbs (DN1–6) tested in indirect ELISA on plates coated with serial dilutions of rhPAD2 or rhPAD4 (compared at 50 ng/mL). The supernatants were diluted 1:1 in dilution buffer. Column C1 represents mouse IgG isotype control, and column C2 represents buffer alone. (B) Reactivity of DN1–6, as determined by western blotting. The supernatants were tested in a 1:1 dilution in dilution buffer against reduced rhPAD2 (~3 ng/lane). Lane C1 represents mouse IgG isotype control and lane C2 represents buffer alone. (C) Epitope mapping of the anti-PAD2 mAbs PAD2 was carried out using western blotting of different deletion mutants of rhPAD2. Reactivity of selected mAbs is shown: WT (full length wild type rhPAD2), C254 (amino acids 1–254 of rhPAD2), I385–463 (rhPAD2 with deletions in the catalytic site), N165 (amino acid sequence from 165 to the C-terminus), N343 (amino acid sequence from 343 to the C-terminus). All mAbs bound to an epitope in the N-terminal region between amino acids 1 and 165. (✓) reacting mAbs (✗) non-reacting mAbs.
to a commercially available polyclonal anti-PAD2 antibody (Fig. 2A) (Makrygiannakis et al., 2012), as exemplified for mAb DN6 (Fig. 2B) and mAb DN34 (Fig. 2C). No signal was detected when an irrelevant mAb, anti-chicken complement component C3, was used for staining (Fig. 2D). A summary of the reactivity pattern in all assays is given in Fig. 3.
3.2. Specificity of ELISA for human, rabbit and mouse PAD2 Aiming to develop a human PAD2 (hPAD2) specific, noncompetitive sandwich ELISA, we tested combinations of the most strongly reacting mAbs against rhPAD2 in the indirect ELISA (data not shown). A combination of mAb DN2 as capture
Fig. 2. Immunohistochemistry showing binding of anti-PAD2 mAbs to inflamed synovial tissue from an ACPA-positive patient with rheumatoid arthritis. Frozen synovial biopsy sections show brown diaminobenzidine (DAB) staining for PAD2 and blue hematoxylin counterstaining. Shown are micrographs obtained at 25× magnification. (A) Staining with rabbit polyclonal anti-PAD2 as positive control. (B) Staining with mAb DN6. (C) Staining with mAb DN34. (D) Staining with anti-chicken complement component 3, as negative control.
D. Damgaard et al. / Journal of Immunological Methods 405 (2014) 15–22
19
Fig. 3. Overview of mAb reactivity with human PAD2. Culture supernatants from 35 clones (DN1–35) were tested in indirect ELISA against rhPAD2, Western blot (WB) against rhPAD2, and against synovium from RA patients by means of immunohistochemistry (IHC).
antibody and DN6 as detection antibody was chosen. It should be noted that even when the same mAb was used as captureand detector Abs, signals could also be obtained (data not shown). The assay was specific for PAD2, since strong signals were obtained when rhPAD2 was applied, while no signals were registered after application of rhPAD3, rhPAD4, or rhPAD6 (Fig. 4A). No signals were obtained in absence of capture antibody or detecting antibody, nor when an irrelevant mAb (anti-chicken C3) was used as capture or detection antibody (data not shown). The maximal background signal allowed was an absorbance of 0.03.
In accordance with rPAD2 being the immunizing antigen, rPAD2 was also readily detected in the assay while also rmPAD2 was detected, albeit less efficiently (Fig. 4B).
3.3. Calcium-dependency of the ELISA Since calcium is known to alter the tertiary structure of PAD2, the presence of calcium may affect the ELISA. To assess this, rhPAD2 was added to TBS containing various concentrations of calcium. As shown in Fig. 5, the ELISA readings were largely independent of calcium, with slightly higher measurements in
Fig. 4. Specificity of ELISA for PAD2. (A) Dilutions of rhPAD2, rhPAD3, rhPAD4 and rhPAD6 in washing buffer were applied to the ELISA using DN2 as capture antibody and DN6 as detection antibody. Among the human isoforms, reactivity was only observed with rhPAD2. (B) Absorbance at 490 nm was compared at 50 ng/mL for rhPAD2, rhPAD3, rhPAD4, rhPAD6, murine PAD2 (mPAD2) and rabbit PAD2 (rPAD2) after allowance for background (absorbance at 650 nm). Duplicate measurements are shown as mean and range.
20
D. Damgaard et al. / Journal of Immunological Methods 405 (2014) 15–22
Hence, the LOD of the assay was 200 pg/mL for serum and plasma samples. The intra-assay- and inter-assay coefficients of variation (CV) ranged from 5.1% to 7.6% and from 11.7% to 12.1%, respectively (Table 1). Experiments addressing recovery of PAD2 were performed by spiking serum and plasma samples with rhPAD2. The mean recoveries for the spiked serum and plasma samples were 98 % and 102 %, respectively (Table 2). Following addition of 10 mM EDTA to serum samples, the signal tended to be slightly lower than in the absence of EDTA (p = 0.07; data not shown). 3.5. PAD2 levels in healthy donors
Fig. 5. Calcium-dependency of the ELISA for PAD2. rhPAD2 added to TBS buffer was measured in the presence of different concentrations of calcium, i.e. 5, 2.5, 1.25 and 0.63 mM, or without calcium. Duplicate measurements of absorbance are shown as mean and range.
The PAD2 concentrations were determined in serum samples from 28 healthy blood donors (Fig. 7). Eight out of 28 donors (29%) had PAD2 levels above the lower limit of detection of 0.2 ng/mL with a median concentration of 0.54 ng/mL (range 0.36 – 1.28 ng/mL). 3.6. Effect of storage and freezing–thawing on PAD2 measurements
the presence of calcium, however (p b 0.05 for 5 mM or 2.5 mM calcium versus no calcium). 3.4. Assay range, limit of detection, and accuracy of the ELISA A representative standard curve consisting of serial dilutions of rhPAD2 is shown in Fig. 6A. The limit of detection (LOD) of the sandwich ELISA was defined as the mean of the background absorbance plus three times the standard deviation, i.e. 50 pg/mL. Within a range of 50 pg/mL to 5 ng/mL, the standard curve was linear. To test for parallelism between the calibrator and PAD2 in different matrixes, serum and plasma samples were spiked with rhPAD2, followed by serial dilutions in dilution buffer containing 25% normal serum or plasma, respectively (Fig. 6B). Below 5 ng/mL of rhPAD2, parallelism was observed between calibrator and analyte levels. Serum and plasma samples had to be diluted at least 1:4 to obtain parallelism with the calibrator.
Storage of a serum and plasma sample at 4 °C or 20 °C, for both 24 and 48 h, resulted in significantly decreased measurements of PAD2 (losses of 24% to 83%), compared to storage at −80 °C (p b 0.0003–0.06, data not shown). The effect of repeated freezing (−80 °C) and thawing on measurement of serum PAD2 levels was tested. A modest, albeit statistically significant, decrease in PAD2 levels by 11% was observed, when the samples spiked with rhPAD2 had been frozen and thawed more than five times (p b 0.05; data not shown). Thus, PAD2 seems to be relatively robust upon freezing and thawing. 4. Discussion Using reactivity with rhPAD2 indirect ELISA as read-out, we identified 35 clones producing mAbs against PAD2, the specificity of which was examined in both western blotting and immunohistochemistry. Most of the produced mAbs
Fig. 6. (A) Calibrator curve of the ELISA for human PAD2. rhPAD2 was diluted serially from 25 ng/mL in dilution buffer and tested in the ELISA using DN2 as capture antibody and DN6 as detection antibody. Duplicate measurements of absorbance (absorbance 490 nm) values, after allowance for background (absorbance at 650 nm), are shown as mean and range (too narrow to be visible). (B) Serum or plasma were spiked with rhPAD2 and serially diluted in dilution buffer with a constant content of 25% normal serum or plasma, respectively, to test parallelism with the calibrator. Means of 2 measurements are shown.
D. Damgaard et al. / Journal of Immunological Methods 405 (2014) 15–22
21
Table 1 Intra-assay and inter-assay variation of ELISA for human PAD2. Experiment#
n
Sample material
rhPAD2 (ng/mL)
SD
CV (%)
Mean Intra-assay variationa 1 2 3 4
10 10 10 10
Serum Serum Plasma Serum
1.93 2.69 4.42 5.98
0.15 0.17 0.22 0.44
7.6 6.4 5.1 7.3
1 2 3
5 5 5
Serum Plasma Serum
1.41 3.97 12.8
0.17 0.47 1.56
12.1 11.7 12.1
Inter-assay variationb
a The intra-assay variation of the PAD2 ELISA using DN2 as capture antibody and DN6 as detection antibody was determined by four measurements of 10 samples, spiked with rhPAD2. b The inter-assay variation was determined by measuring three samples on five different days. The coefficient of variation (CV) was determined for each experiment.
showed similar patterns of reactivity in all three assays tested (WB, ELISA and immunohistochemistry), with some variations as expected. Only a minority of the tested mAbs (10/35) were found not to react with inflamed synovial tissue from an ACPA-postive RA patient. Among these, five were also non-reactive in western blotting. In contrast, a majority of the mAbs showed reactivity comparable with a commercially available anti-PAD2 antibody, suggesting that these antibodies might be useful immunohistochemistry tools with the known advantages of using monoclonal rather than polyclonal antibodies for such studies. Four mAbs (DN6, DN18, DN31 and DN34), were among highest responders in all assays. It is possible that these mAbs will also prove useful in other techniques, such as in chromatin Immunoprecipitation, for which no monoclonal anti-PAD2 antibody has proved useful so far (Cherrington et al., 2012). Using two mAbs, DN2 and DN6, we developed an ELISA, which to our knowledge is the first assay to measure soluble PAD2 quantitatively. The observation that positive signals could be obtained when the same mAb was used as captureand detection Ab indicates that PAD2 exists as a dimer. This is in accordance with the findings of Liu et al. (2011), who showed that PAD4 exists as homo-dimers and suggests that Tyr435 in PAD4 and Phe435 in PAD2 have similar roles in the interaction between the monomers.
Table 2 Recovery of recombinant human PAD2 in serum/plasma. Spiked rhPAD2 (ng/mL) 10 3,33 1,11 0,37
Seruma (N = 2)
Plasmaa (N = 2)
Mean
SD
Recovery
Mean
SD
Recovery
10.0 3.17 1.12 0.36
0.46 0.05 0.02 0.05
100% 95% 101% 96%
10.6 3.27 1.14 0.37
0.30 0.09 0.16 0.01
106% 98% 102% 101%
a Plasma and serum samples were spiked with rhPAD2 and the PAD2-concentration was measured by the ELISA using DN2 as capture antibody and DN6 as detection antibody.
Fig. 7. Variability of PAD2 levels in normal blood. Serum concentration of PAD2 in 28 healthy donors, as assessed by the ELISA for human PAD2.
Numerous investigators have reported associations between PAD2 and PAD4 with inflammatory diseases, in particular with RA (Foulquier et al., 2007; Kinloch et al., 2008; Basu et al., 2011) but also other diseases, such as multiple sclerosis (Wood et al., 2008; Moscarello et al., 2013) and cancer (Cherrington et al., 2012). Enzymatic differences between the two isoforms have been shown in vitro (Nakayama-Hamada et al., 2005), but it is still not known if one or both of them contribute to the pathogenesis of RA. Our PAD2 ELISA may prove useful in future studies of PAD2's role as mediator of extracellular citrullination. Interestingly, PAD2 was measureable in 29% of 28 normal serum samples tested. These donors all fulfilled the criteria for Danish blood donors, i.e. they had no history of autoimmune disease or cancer, and had no recognized ongoing infection. In several inflammatory diseases, for example RA, circulating PAD2 levels can be expected to be elevated, and our assay facilitates studies aimed at the question whether circulating PAD2 levels or PAD2 levels in synovial fluid are correlated to clinical features like disease activity. Acknowledgments This study was supported by proof-of-concept means from the Danish Research Council and by Novo Nordic exploratory pre-seed grant. Part of this work was financially supported by The Netherlands Proteomics Centre, a program embedded in the Netherlands Genomics Initiative. The authors would like to thank Marianne Engström (Rheumatology Unit, Karolinska Institutet, Stockholm), and Judith Stammen-Vogelzangs (Department of Biomolecular Chemistry, Radboud University Nijmegen) for expert technical assistance with the immunohistochemistry procedures and expression of recombinant proteins, respectively. References Arandjelovic, S., McKenney, K.R., Leming, S.S., Mowen, K.A., 2012. ATP induces protein arginine deiminase 2-dependent citrullination in mast cells through the P2X7 purinergic receptor. J. Immunol. 189, 4112. Asaga, H., Yamada, M., Senshu, T., 1998. Selective deimination of vimentin in calcium ionophore-induced apoptosis of mouse peritoneal macrophages. Biochem. Biophys. Res. Commun. 243, 641. Basu, P.S., Majhi, R., Ghosal, S., Batabyal, S.K., 2011. Peptidyl-arginine deiminase: an additional marker of rheumatoid arthritis. Clin. Lab. 57, 1021.
22
D. Damgaard et al. / Journal of Immunological Methods 405 (2014) 15–22
Chang, X., Yamada, R., Suzuki, A., Sawada, T., Yoshino, S., Tokuhiro, S., Yamamoto, K., 2005. Localization of peptidylarginine deiminase 4 (PADI4) and citrullinated protein in synovial tissue of rheumatoid arthritis. Rheumatology (Oxford) 44, 40. Cherrington, B.D., Zhang, X., McElwee, J.L., Morency, E., Anguish, L.J., Coonrod, S.A., 2012. Potential role for PAD2 in gene regulation in breast cancer cells. PLoS One 7, e41242. Darrah, E., Rosen, A., Giles, J.T., Andrade, F., 2012. Peptidylarginine deiminase 2, 3 and 4 have distinct specificities against cellular substrates: novel insights into autoantigen selection in rheumatoid arthritis. Ann. Rheum. Dis. 71, 92. Forslind, K., Ahlmen, M., Eberhardt, K., Hafstrom, I., Svensson, B., 2004. Prediction of radiological outcome in early rheumatoid arthritis in clinical practice: role of antibodies to citrullinated peptides (anti-CCP). Ann. Rheum. Dis. 63, 1090. Foulquier, C., Sebbag, M., Clavel, C., Chapuy-Regaud, S., Al, B.R., Mechin, M.C., Vincent, C., Nachat, R., Yamada, M., Takahara, H., Simon, M., Guerrin, M., Serre, G., 2007. Peptidyl arginine deiminase type 2 (PAD-2) and PAD-4 but not PAD-1, PAD-3, and PAD-6 are expressed in rheumatoid arthritis synovium in close association with tissue inflammation. Arthritis Rheum. 56, 3541. Gregersen, P.K., Silver, J., Winchester, R.J., 1987. The shared epitope hypothesis. An approach to understanding the molecular genetics of susceptibility to rheumatoid arthritis. Arthritis Rheum. 30, 1205. Harauz, G., Musse, A.A., 2007. A tale of two citrullines—structural and functional aspects of myelin basic protein deimination in health and disease. Neurochem. Res. 32, 137. Huizinga, T.W., Amos, C.I., van der Helm-van Mil, A.H., Chen, W., van Gaalen, F.A., Jawaheer, D., Schreuder, G.M., Wener, M., Breedveld, F.C., Ahmad, N., Lum, R.F., de Vries, R.R., Gregersen, P.K., Toes, R.E., Criswell, L.A., 2005. Refining the complex rheumatoid arthritis phenotype based on specificity of the HLADRB1 shared epitope for antibodies to citrullinated proteins. Arthritis Rheum. 52, 3433. Kinloch, A., Lundberg, K., Wait, R., Wegner, N., Lim, N.H., Zendman, A.J., Saxne, T., Malmstrom, V., Venables, P.J., 2008. Synovial fluid is a site of citrullination of autoantigens in inflammatory arthritis. Arthritis Rheum. 58, 2287. Kohler, G., Milstein, C., 1975. Continuous cultures of fused cells secreting antibody of predefined specificity. J. Immunol. 174, 2453. Li, P., Li, M., Lindberg, M.R., Kennett, M.J., Xiong, N., Wang, Y., 2010. PAD4 is essential for antibacterial innate immunity mediated by neutrophil extracellular traps. J. Exp. Med. 207, 1853. Liu, Y.L., Chiang, Y.H., Liu, G.Y., Hung, H.C., 2011. Functional role of dimerization of human peptidylarginine deiminase 4 (PAD4). PLoS One 6, e21314. Makrygiannakis, D., Revu, S., Engstrom, M., af, K.E., Nicholas, A.P., Pruijn, G.J., Catrina, A.I., 2012. Local administration of glucocorticoids decreases synovial citrullination in rheumatoid arthritis. Arthritis Res. Ther. 14, R20.
Moscarello, M.A., Lei, H., Mastronardi, F.G., Winer, S., Tsui, H., Li, Z., Ackerley, C., Zhang, L., Raijmakers, R., Wood, D.D., 2013. Inhibition of peptidylarginine deiminases reverses protein-hypercitrullination and disease in mouse models of multiple sclerosis. Dis. Model. Mech. 6, 467. Nakayama-Hamada, M., Suzuki, A., Kubota, K., Takazawa, T., Ohsaka, M., Kawaida, R., Ono, M., Kasuya, A., Furukawa, H., Yamada, R., Yamamoto, K., 2005. Comparison of enzymatic properties between hPADI2 and hPADI4. Biochem. Biophys. Res. Commun. 327, 192. Orrenius, S., Zhivotovsky, B., Nicotera, P., 2003. Regulation of cell death: the calcium-apoptosis link. Nat. Rev. Mol. Cell Biol. 4, 552. Palarasah, Y., Skjodt, K., Brandt, J., Teisner, B., Koch, C., Vitved, L., Skjoedt, M.O., 2010. Generation of a C3c specific monoclonal antibody and assessment of C3c as a putative inflammatory marker derived from complement factor C3. J. Immunol. Methods 362, 142. Pritzker, L.B., Joshi, S., Gowan, J.J., Harauz, G., Moscarello, M.A., 2000. Deimination of myelin basic protein. 1. Effect of deimination of arginyl residues of myelin basic protein on its structure and susceptibility to digestion by cathepsin D. Biochemistry 39, 5374. Raijmakers, R., Zendman, A.J., Egberts, W.V., Vossenaar, E.R., Raats, J., SoedeHuijbregts, C., Rutjes, F.P., van Veelen, P.A., Drijfhout, J.W., Pruijn, G.J., 2007. Methylation of arginine residues interferes with citrullination by peptidylarginine deiminases in vitro. J. Mol. Biol. 367, 1118. Schellekens, G.A., de Jong, B.A., van den Hoogen, F.H., van de Putte, L.B., van Venrooij, W.J., 1998. Citrulline is an essential constituent of antigenic determinants recognized by rheumatoid arthritis-specific autoantibodies. J. Clin. Invest. 101, 273. Senshu, T., Kan, S., Ogawa, H., Manabe, M., Asaga, H., 1996. Preferential deimination of keratin K1 and filaggrin during the terminal differentiation of human epidermis. Biochem. Biophys. Res. Commun. 225, 712. Snir, O., Widhe, M., von, S.C., Lindberg, J., Padyukov, L., Lundberg, K., Engstrom, A., Venables, P.J., Lundeberg, J., Holmdahl, R., Klareskog, L., Malmstrom, V., 2009. Multiple antibody reactivities to citrullinated antigens in sera from patients with rheumatoid arthritis: association with HLA-DRB1 alleles. Ann. Rheum. Dis. 68, 736. van Venrooij, W.J., van Beers, J.J., Pruijn, G.J., 2011. Anti-CCP antibodies: the past, the present and the future. Nat. Rev. Rheumatol. 7, 391. Vossenaar, E.R., Zendman, A.J., van Venrooij, W.J., Pruijn, G.J., 2003. PAD, a growing family of citrullinating enzymes: genes, features and involvement in disease. Bioessays 25, 1106. Vossenaar, E.R., Radstake, T.R., van der Heijden, A., van Mansum, M.A., Dieteren, C., de Rooij, D.J., Barrera, P., Zendman, A.J., van Venrooij, W.J., 2004. Expression and activity of citrullinating peptidylarginine deiminase enzymes in monocytes and macrophages. Ann. Rheum. Dis. 63, 373. Wood, D.D., Ackerley, C.A., Brand, B., Zhang, L., Raijmakers, R., Mastronardi, F.G., Moscarello, M.A., 2008. Myelin localization of peptidylarginine deiminases 2 and 4: comparison of PAD2 and PAD4 activities. Lab. Invest. 88, 354.
Contents lists available at ScienceDirect
Journal of Immunological Methods journal homepage: www.elsevier.com/locate/jim
Research paper
Generation of monoclonal antibodies against peptidylarginine deiminase 2 (PAD2) and development of a PAD2-specific enzyme-linked immunosorbent assay Dres Damgaard a, Yaseelan Palarasah b, Karsten Skjødt b, Anca I. Catrina c, Sanne M.M. Hensen d, Ger J.M. Pruijn d, Claus H. Nielsen a,⁎ a
Institute for Inflammation Research, Department of Infectious Diseases and Rheumatology, Copenhagen University Hospital, Rigshospitalet, Denmark Department of Cancer and Inflammation Research, Institute of Molecular Medicine, University of Southern Denmark, Denmark c Rheumatology Unit, Department of Medicine, Karolinska University Hospital and Karolinska Institutet, Sweden d Department of Biomolecular Chemistry, Institute for Molecules and Materials, Nijmegen Center for Molecular Life Sciences and Netherlands Proteomics Centre, Radboud University Nijmegen, Nijmegen, The Netherlands b
a r t i c l e
i n f o
Article history: Received 22 November 2013 Received in revised form 18 December 2013 Accepted 20 December 2013 Available online 31 December 2013 Keywords: Peptidylarginine deiminase PAD2 Citrullination Monoclonal antibody ELISA
a b s t r a c t The enzyme peptidylarginine deiminase 2 (PAD2) has been associated with inflammatory diseases, such as rheumatoid arthritis and neurodegenerative diseases including multiple sclerosis. To investigate the association of various diseases with extracellular PAD2, we raised monoclonal antibodies (mAbs) against rabbit PAD2 and evaluated their cross-reactivity with human PAD2 by indirect enzyme-linked immunosorbent assay (ELISA), western blotting and immunohistological staining of inflamed synovial tissue. Moreover, we established a sandwich ELISA detecting human PAD2, based on two different monoclonal antibodies, mAbs DN2 and DN6. The assay had a lower detection limit of 200 pg/mL in serum and plasma samples, and showed dilution linearity and recovery ranging from 95 to 106%. The mAbs and the ELISA showed isotype specificity for PAD2. Circulating PAD2 was found in 8/28 (29%) serum samples from healthy donors. In conclusion, several of our mAbs proved useful in western blotting and immunohistochemistry, and the ELISA described here reliably measures PAD2 levels in blood. This allows investigation of PAD2 as a possible biomarker and further investigation of PAD2's involvement in various inflammatory diseases. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Citrullination (deimination) designates a posttranslational modification in proteins resulting in a conversion of the amino acid residue arginine into the non-standard residue citrulline. This reaction is catalyzed by enzymes of the peptidylarginine deiminase (PAD) family (Vossenaar et al., 2003) and is important for homeostatic processes such as keratinocyte differentiation (Senshu et al., 1996), maintenance of myelin sheath insulation
Abbreviations: ELISA, enzyme-linked immunosorbent assay; LOD, limit of detection; mAb, monoclonal antibody; PAD, peptidylarginine deiminase. ⁎ Corresponding author. E-mail address: [email protected] (C.H. Nielsen). 0022-1759/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jim.2013.12.008
(Harauz and Musse, 2007) and formation of neutrophil extracellular traps (NETs) (Li et al., 2010). PADs and citrullinated proteins are involved in a number of autoimmune- and inflammatory diseases, and have especially been drawn into focus after discovery of their pathogenic role in rheumatoid arthritis (RA) (Schellekens et al., 1998). During inflammation, including that associated with RA, proinflammatory stimuli and increased cell death allow protein citrullination to occur; either as a result of calcium-influx, followed by intracellular citrullination (Vossenaar et al., 2004; Asaga et al., 1998; Orrenius et al., 2003), or by leakage of PADs from dying cells into synovial fluid, where the calcium content is sufficient for PAD to have enzymatic activity (NakayamaHamada et al., 2005), followed by citrullination of extracellular
16
D. Damgaard et al. / Journal of Immunological Methods 405 (2014) 15–22
proteins (Kinloch et al., 2008). Extracellular PAD2 may originate from dying macrophages or mast cells (Vossenaar et al., 2004; Arandjelovic et al., 2012). Approximately 80% of RA patients carry MHC class II molecules containing the ‘shared epitope’ (SE), capable of binding citrullinated self-antigens (Gregersen et al., 1987). A similar proportion of patients produce anti-citrullinated protein antibodies (ACPAs) (Snir et al., 2009; Huizinga et al., 2005; van Venrooij et al., 2011), which have become a major diagnostic and prognostic marker of RA (van Venrooij et al., 2011). A significant positive correlation exists between the serum titer and clinical, biologic, and radiologic data related to ACPA-positive RA activity and severity (Forslind et al., 2004). Increased levels of PADs and citrullinated proteins have also been found in neurodegenerative diseases, such as multiple sclerosis (Wood et al., 2008). Hypercitrullination of myelin basic protein (MBP) disrupts the formation of normal myelin sheath structure and makes MBP more susceptible to be degraded (Pritzker et al., 2000). Five members of the human PAD enzyme family exist; PAD1, PAD2, PAD3, PAD4 and PAD6 (Vossenaar et al., 2003). PAD2 and PAD4 are the isotypes associated with RA, as their expression/ presence has been demonstrated by semi-quantitative techniques, such as immunoblotting and immunohistochemistry, in RA synovium, synovial fluid cells and extracellularly in synovial fluid (Foulquier et al., 2007; Chang et al., 2005; Kinloch et al., 2008). PAD2 and PAD4 differ with respect to citrullination efficiency of different substrates and optimal enzymatic conditions (Nakayama-Hamada et al., 2005; Darrah et al., 2012). Their differential roles in the pathogenesis of RA have not been elucidated. It is also unclear which role intracellular and extracellular citrullination, respectively, play in the pathogenesis, and whether both events are required for the initiation or maintenance of the disease. Specific monoclonal antibodies are needed to investigate the pathogenic role of PAD2 (Cherrington et al., 2012). We generated a panel of monoclonal anti-PAD2 antibodies, and developed a sandwich ELISA specific for PAD2 allowing investigation of disease associations with PAD2 in blood and other body fluids. 2. Material and methods
(Sigma–Aldrich, Brøndby, Denmark). The mice received an intravenous injection of 25 μg rPAD2 administered with adrenalin four days prior the fusion. The fusion and selection were based on the principles described by Kohler and Milstein (Kohler and Milstein, 1975). The SP2/0-AG14 myeloma cell line (Palarasah et al., 2010) was selected as fusion partner. Positive clones were identified by ELISA using microtiter plates (Maxisorp; Nunc, Roskilde, Denmark) coated with 0.25 μg/mL rPAD2 in coating buffer (35 mM NaHCO3, 15 mM Na2CO3, pH 9.6). Cells from positive wells were cloned at least four times by limiting dilution. Single clones were grown in culture flasks in RPMI-1640 medium (RPMI) (Sigma–Aldrich) containing 10% foetal calf serum (FCS), and mAbs were purified from culture supernatant by Protein G affinity chromatography using the Äkta fast-performance liquid chromatography (FPLC) system according to the manufacturer's instructions. The eluate was dialyzed against phosphate-buffered saline (PBS), pH 7.4. Subsequently, the purified mAbs were dialyzed against 0.1 M carbonate buffer, pH 8.5, and biotin N-hydroxysuccinimide ester (BNHS) (Sigma–Aldrich, Glostrup, Denmark) was added 1:6 (w:w) of the total protein amount. The mixture was incubated with gentle mixing for 3 h at room temperature (RT). Unbound biotin was removed by dialysis against PBS. 2.3. Western blotting Gel-electrophoresis was performed on 4–12% Novex Bis-Tris gels (Invitrogen, CA, USA) using the NuPAGE® system (Invitrogen) according to the manufacturer's recommendations. Proteins were electro-blotted onto polyvinylidene difluoride membranes (PVDF-HyBond; Amersham Bioscience, Uppsala, Sweden). The membrane was blocked in washing buffer (PBS, 0.05% Tween-20, pH 7.4) and sliced into strips. Antibody culture supernatants (2 mL) were added 1:1 in wash buffer and incubated O/N at 4 °C. Horseradish peroxidase (HRP)-conjugated rabbit anti-mouse IgG antibodies (Dako Denmark A/S, Glostrup) were added to a dilution of 1:3000 in washing buffer, followed by incubation 1 h at RT. Finally, the strips were washed, incubated 10 min with acetate buffer (50 mM CH3COOH, 33.75 mM NaOH, pH 5.0) and stained in carbazole staining solution (0.04% 3-amino-9-ethylcarbazole and 0.015% H2O2) in acetate buffer.
2.1. PADs 2.4. Immunohistochemistry PAD2 from rabbit skeletal muscle (rPAD2) and recombinant murine PAD2 (rmPAD2) were purchased from Sigma, St Louis, MO. Recombinant human PAD (rhPAD) of various isotypes, i.e. rhPAD2, rhPAD3, rhPAD4 and rhPAD6, was produced and purified as described previously (Raijmakers et al., 2007). Different deletion mutants of rhPAD2 were prepared likewise, i.e. C254 (amino acids 1–254 of rhPAD2), I385–463 (rhPAD2 with deletions in the catalytic site), N165 (amino acid sequence from 165 to the C-terminus), N343 (amino acid sequence from 343 to the C-terminus). 2.2. Generation and purification of PAD2 mAbs Three BALB/cxNMRI mice were immunized subcutaneously three times with 25 μg of rPAD2 adsorbed to Al(OH)3 (Alhydrogel 2.0%, Brenntag Biosector A/S, Frederikssund, Denmark) and mixed 1:1 with Freund's incomplete adjuvant
Synovial biopsy specimens from inflamed synovium were snap frozen within 30 s in liquid isopentane, stored at −70 °C, and then embedded in OCT compound (TissueTek; Sakura Finetek, Zoeterwoude, The Netherlands) when sectioning was performed. Cryostat sections from the biopsy samples were placed on SuperFrost Plus slides (Menzel-Glaser, Braunschweig, Germany) and air dried for 30 min. Sections were initially fixed for 20 min with freshly prepared 2% (volume/ volume) formaldehyde (Sigma, St. Louis, MO), dissolved in phosphate buffered saline (PBS; pH 7.4) at 4 °C, washed in PBS, and then left to air-dry before storage at − 70 °C. The staining procedures have been described in detail previously (Makrygiannakis et al., 2012). Either 0.1% saponin or 0.3% Triton-X100 was used for cell permeabilization. All antiPAD2 antibodies were used as primary antibodies in a dilution of 1:3 (estimated to correspond to 1.25–5 μg/mL).
D. Damgaard et al. / Journal of Immunological Methods 405 (2014) 15–22
As positive control, polyclonal anti-PAD2 antibodies (ROI002) (CosmoBio; Tokyo, Japan) were used, and as negative control an irrelevant, in house-raised mAb against chicken complement component 3 (C3) was used. After three washes, the slides were blocked with 1% normal horse serum, followed by incubation for 30 min at RT with biotinylated horseanti-mouse IgG antibodies (Vector Laboratories; Burlingame, CA), diluted 1:640 in wash buffer containing 1% horse serum. The polyclonal anti-PAD2 positive control was diluted in PBS containing 0.3% Tween-20, 3% BSA (0.02 μg/mL). The slides were then incubated with ABC Elite streptavidin–biotin complex (Vector Laboratories) for 40 min at RT, and developed using diaminobenzidine-containing peroxidase Substrate Kit SK-4100 (Vector laboratories) for 7 min at RT in dark. Counterstaining was performed using hematoxylin (Vector Laboratories). Photography was carried out with identical exposures and concurrent processing. The staining was scored as: 1) not reacting (binding pattern similar to that of the irrelevant control and of slides with no primary antibody added), 2) weakly reacting (a weak staining pattern) and 3) strongly reacting (staining pattern similar to that of the positive anti-PAD2 control).
17
2.7. Evaluation of accuracy To assess the variation of the ELISA, plasma- and serum samples were spiked with rhPAD2 and applied in 2-fold serial dilutions in the assay. To determine the intra-assay variation, plasma and serum samples were measured ten times. To determine inter-assay variation, three samples were tested in the assay on 5 different days. Recovery was tested by spiking samples with rhPAD2 at different concentrations covering the whole working range of the assay. The mean values, standard deviation (SD), and the coefficient of variation (CV = SD/mean × 100%), were calculated for each of these experiments. 2.8. Calculations and statistical analysis All standards and samples were measured in duplicates. Concentrations were calculated by regression analysis for the standard curve using four-parameter logistic curve-fitting on MARS software. GraphPad Prism 5.0 (GraphPad Software, CA) was used for statistical analyses. 2.9. Stability of PAD2 in samples
2.5. ELISA for PAD2 Maxisorp plates were coated overnight at 4 °C with 100 μL mAb DN2 (1 μg/mL). Wells were washed three times and blocked with washing buffer (PBS, 0.05% Tween-20, pH 7.4) for 20 min at room temperature. The calibrator, quality control, and samples (see below) were diluted in dilution buffer (PBS, 0.5% Tween-20, 2% bovine serum [Sigma], 20 μg/mL Mouse IgG isotype control [Novus Biologicals; Cambridge], UK, pH 7.4) and incubated for 90 min in the wells, at room temperature. Following three washes, the wells were incubated for 1 h at room temperature with 100 μL biotinylated mAb DN6 (1 μg/mL) in washing buffer. After another three washes, the wells were incubated at room temperature with 100 μL of streptavidin-conjugated HRP (Invitrogen; Invitrogen, CA, USA) diluted 1:3000 in washing buffer for 45 min. Finally, the plates were washed three times and incubated with 0.4 mg of o-phenylene-diamine (Kem-En-Tec; Taastrup, Denmark) per mL developing buffer (35 mM citric acid, 65 mM Na2PO4, pH 5). After 10 min the color development was stopped with 1 M H2SO4, and absorbance was measured at 490–650 nm using SPECTROstar nano Microplate Reader (BMG Labtech; Ortenberg, Germany) and the data were processed using MARS software (BMG Labtech). An irrelevant, in house-raised mAb, anti-chicken C3, was used as negative control, either as capture or as detecting mAb. 2.6. Calibrator and controls The calibrator consisted of rhPAD2, spiked in dilution buffer. The quality control consisted of a pool of normal serum spiked with rhPAD2, and stored as aliquots at −80 °C. As isotype controls, i.e. rhPAD3, rhPAD4 and rhPAD6, were used. Moreover, we also tested rmPAD2. A two-fold serial dilution of these reagents was used to generate 10-point calibrator curves with a range from 25 to 0.05 ng/mL. A four-parameter fit model was applied to estimate the concentration of unknown samples. The calibrator was stored as aliquots at −80 °C.
Serum samples were stored at 4 °C or 20 °C for 24 or 48 h to test stability of PAD2 in non-frozen samples. Furthermore, serum samples were subjected to nine freeze–thaw cycles (− 80 °C and room temperature, respectively) and compared with respect to PAD2 levels. 2.10. Serum and plasma samples Serum and plasma were isolated from peripheral venous blood, drawn in dry tubes or EDTA-tubes (BD; Plymouth, UK). Serum samples were collected from 28 self-reported healthy, anonymous donors (14 men, 14 women, age: 45 ± 13 years), attending the Blood Bank at Copenhagen University Hospital Rigshospitalet. 3. Results 3.1. Human PAD2 specific monoclonal antibodies After immunization of mice with rPAD2 and generation of hybridoma cells, culture supernatants were screened for reactivity with rPAD2 in an indirect ELISA. Hybridoma cells in positive wells were sub-cloned until positive culture supernatants were formed by single clones. From the screening and sub-cloning procedure, 35 hybridomas were successfully isolated (mAbs DN1–35). These were screened for reactivity against rhPAD2 by indirect ELISA (Fig. 1A) and western blotting (Fig. 1B). Thirty three of these hybridoma supernatants and 23 out of 31 tested samples were reactive with rhPAD2 in ELISA and western blotting, respectively. The mAbs DN2 and DN6 were found to bind to an epitope in the N-terminal region between amino acids 1–165, demonstrated by western blotting using different deletion mutants of rhPAD2 (Fig. 1C). Immunocytochemistry performed on synovial tissue from an ACPA-positive RA patient showed that a large majority of the tested clones had staining patterns similar
18
D. Damgaard et al. / Journal of Immunological Methods 405 (2014) 15–22
Fig. 1. Reactivity of mAbs with human PAD2. Culture supernatants from 35 hybridomas producing mAbs against rabbit PAD2 were tested for reactivity with rhPAD2: (A) Absorbance at 490 nm values, after allowance for background (absorbance at 650 nm), is shown for six representative mAbs (DN1–6) tested in indirect ELISA on plates coated with serial dilutions of rhPAD2 or rhPAD4 (compared at 50 ng/mL). The supernatants were diluted 1:1 in dilution buffer. Column C1 represents mouse IgG isotype control, and column C2 represents buffer alone. (B) Reactivity of DN1–6, as determined by western blotting. The supernatants were tested in a 1:1 dilution in dilution buffer against reduced rhPAD2 (~3 ng/lane). Lane C1 represents mouse IgG isotype control and lane C2 represents buffer alone. (C) Epitope mapping of the anti-PAD2 mAbs PAD2 was carried out using western blotting of different deletion mutants of rhPAD2. Reactivity of selected mAbs is shown: WT (full length wild type rhPAD2), C254 (amino acids 1–254 of rhPAD2), I385–463 (rhPAD2 with deletions in the catalytic site), N165 (amino acid sequence from 165 to the C-terminus), N343 (amino acid sequence from 343 to the C-terminus). All mAbs bound to an epitope in the N-terminal region between amino acids 1 and 165. (✓) reacting mAbs (✗) non-reacting mAbs.
to a commercially available polyclonal anti-PAD2 antibody (Fig. 2A) (Makrygiannakis et al., 2012), as exemplified for mAb DN6 (Fig. 2B) and mAb DN34 (Fig. 2C). No signal was detected when an irrelevant mAb, anti-chicken complement component C3, was used for staining (Fig. 2D). A summary of the reactivity pattern in all assays is given in Fig. 3.
3.2. Specificity of ELISA for human, rabbit and mouse PAD2 Aiming to develop a human PAD2 (hPAD2) specific, noncompetitive sandwich ELISA, we tested combinations of the most strongly reacting mAbs against rhPAD2 in the indirect ELISA (data not shown). A combination of mAb DN2 as capture
Fig. 2. Immunohistochemistry showing binding of anti-PAD2 mAbs to inflamed synovial tissue from an ACPA-positive patient with rheumatoid arthritis. Frozen synovial biopsy sections show brown diaminobenzidine (DAB) staining for PAD2 and blue hematoxylin counterstaining. Shown are micrographs obtained at 25× magnification. (A) Staining with rabbit polyclonal anti-PAD2 as positive control. (B) Staining with mAb DN6. (C) Staining with mAb DN34. (D) Staining with anti-chicken complement component 3, as negative control.
D. Damgaard et al. / Journal of Immunological Methods 405 (2014) 15–22
19
Fig. 3. Overview of mAb reactivity with human PAD2. Culture supernatants from 35 clones (DN1–35) were tested in indirect ELISA against rhPAD2, Western blot (WB) against rhPAD2, and against synovium from RA patients by means of immunohistochemistry (IHC).
antibody and DN6 as detection antibody was chosen. It should be noted that even when the same mAb was used as captureand detector Abs, signals could also be obtained (data not shown). The assay was specific for PAD2, since strong signals were obtained when rhPAD2 was applied, while no signals were registered after application of rhPAD3, rhPAD4, or rhPAD6 (Fig. 4A). No signals were obtained in absence of capture antibody or detecting antibody, nor when an irrelevant mAb (anti-chicken C3) was used as capture or detection antibody (data not shown). The maximal background signal allowed was an absorbance of 0.03.
In accordance with rPAD2 being the immunizing antigen, rPAD2 was also readily detected in the assay while also rmPAD2 was detected, albeit less efficiently (Fig. 4B).
3.3. Calcium-dependency of the ELISA Since calcium is known to alter the tertiary structure of PAD2, the presence of calcium may affect the ELISA. To assess this, rhPAD2 was added to TBS containing various concentrations of calcium. As shown in Fig. 5, the ELISA readings were largely independent of calcium, with slightly higher measurements in
Fig. 4. Specificity of ELISA for PAD2. (A) Dilutions of rhPAD2, rhPAD3, rhPAD4 and rhPAD6 in washing buffer were applied to the ELISA using DN2 as capture antibody and DN6 as detection antibody. Among the human isoforms, reactivity was only observed with rhPAD2. (B) Absorbance at 490 nm was compared at 50 ng/mL for rhPAD2, rhPAD3, rhPAD4, rhPAD6, murine PAD2 (mPAD2) and rabbit PAD2 (rPAD2) after allowance for background (absorbance at 650 nm). Duplicate measurements are shown as mean and range.
20
D. Damgaard et al. / Journal of Immunological Methods 405 (2014) 15–22
Hence, the LOD of the assay was 200 pg/mL for serum and plasma samples. The intra-assay- and inter-assay coefficients of variation (CV) ranged from 5.1% to 7.6% and from 11.7% to 12.1%, respectively (Table 1). Experiments addressing recovery of PAD2 were performed by spiking serum and plasma samples with rhPAD2. The mean recoveries for the spiked serum and plasma samples were 98 % and 102 %, respectively (Table 2). Following addition of 10 mM EDTA to serum samples, the signal tended to be slightly lower than in the absence of EDTA (p = 0.07; data not shown). 3.5. PAD2 levels in healthy donors
Fig. 5. Calcium-dependency of the ELISA for PAD2. rhPAD2 added to TBS buffer was measured in the presence of different concentrations of calcium, i.e. 5, 2.5, 1.25 and 0.63 mM, or without calcium. Duplicate measurements of absorbance are shown as mean and range.
The PAD2 concentrations were determined in serum samples from 28 healthy blood donors (Fig. 7). Eight out of 28 donors (29%) had PAD2 levels above the lower limit of detection of 0.2 ng/mL with a median concentration of 0.54 ng/mL (range 0.36 – 1.28 ng/mL). 3.6. Effect of storage and freezing–thawing on PAD2 measurements
the presence of calcium, however (p b 0.05 for 5 mM or 2.5 mM calcium versus no calcium). 3.4. Assay range, limit of detection, and accuracy of the ELISA A representative standard curve consisting of serial dilutions of rhPAD2 is shown in Fig. 6A. The limit of detection (LOD) of the sandwich ELISA was defined as the mean of the background absorbance plus three times the standard deviation, i.e. 50 pg/mL. Within a range of 50 pg/mL to 5 ng/mL, the standard curve was linear. To test for parallelism between the calibrator and PAD2 in different matrixes, serum and plasma samples were spiked with rhPAD2, followed by serial dilutions in dilution buffer containing 25% normal serum or plasma, respectively (Fig. 6B). Below 5 ng/mL of rhPAD2, parallelism was observed between calibrator and analyte levels. Serum and plasma samples had to be diluted at least 1:4 to obtain parallelism with the calibrator.
Storage of a serum and plasma sample at 4 °C or 20 °C, for both 24 and 48 h, resulted in significantly decreased measurements of PAD2 (losses of 24% to 83%), compared to storage at −80 °C (p b 0.0003–0.06, data not shown). The effect of repeated freezing (−80 °C) and thawing on measurement of serum PAD2 levels was tested. A modest, albeit statistically significant, decrease in PAD2 levels by 11% was observed, when the samples spiked with rhPAD2 had been frozen and thawed more than five times (p b 0.05; data not shown). Thus, PAD2 seems to be relatively robust upon freezing and thawing. 4. Discussion Using reactivity with rhPAD2 indirect ELISA as read-out, we identified 35 clones producing mAbs against PAD2, the specificity of which was examined in both western blotting and immunohistochemistry. Most of the produced mAbs
Fig. 6. (A) Calibrator curve of the ELISA for human PAD2. rhPAD2 was diluted serially from 25 ng/mL in dilution buffer and tested in the ELISA using DN2 as capture antibody and DN6 as detection antibody. Duplicate measurements of absorbance (absorbance 490 nm) values, after allowance for background (absorbance at 650 nm), are shown as mean and range (too narrow to be visible). (B) Serum or plasma were spiked with rhPAD2 and serially diluted in dilution buffer with a constant content of 25% normal serum or plasma, respectively, to test parallelism with the calibrator. Means of 2 measurements are shown.
D. Damgaard et al. / Journal of Immunological Methods 405 (2014) 15–22
21
Table 1 Intra-assay and inter-assay variation of ELISA for human PAD2. Experiment#
n
Sample material
rhPAD2 (ng/mL)
SD
CV (%)
Mean Intra-assay variationa 1 2 3 4
10 10 10 10
Serum Serum Plasma Serum
1.93 2.69 4.42 5.98
0.15 0.17 0.22 0.44
7.6 6.4 5.1 7.3
1 2 3
5 5 5
Serum Plasma Serum
1.41 3.97 12.8
0.17 0.47 1.56
12.1 11.7 12.1
Inter-assay variationb
a The intra-assay variation of the PAD2 ELISA using DN2 as capture antibody and DN6 as detection antibody was determined by four measurements of 10 samples, spiked with rhPAD2. b The inter-assay variation was determined by measuring three samples on five different days. The coefficient of variation (CV) was determined for each experiment.
showed similar patterns of reactivity in all three assays tested (WB, ELISA and immunohistochemistry), with some variations as expected. Only a minority of the tested mAbs (10/35) were found not to react with inflamed synovial tissue from an ACPA-postive RA patient. Among these, five were also non-reactive in western blotting. In contrast, a majority of the mAbs showed reactivity comparable with a commercially available anti-PAD2 antibody, suggesting that these antibodies might be useful immunohistochemistry tools with the known advantages of using monoclonal rather than polyclonal antibodies for such studies. Four mAbs (DN6, DN18, DN31 and DN34), were among highest responders in all assays. It is possible that these mAbs will also prove useful in other techniques, such as in chromatin Immunoprecipitation, for which no monoclonal anti-PAD2 antibody has proved useful so far (Cherrington et al., 2012). Using two mAbs, DN2 and DN6, we developed an ELISA, which to our knowledge is the first assay to measure soluble PAD2 quantitatively. The observation that positive signals could be obtained when the same mAb was used as captureand detection Ab indicates that PAD2 exists as a dimer. This is in accordance with the findings of Liu et al. (2011), who showed that PAD4 exists as homo-dimers and suggests that Tyr435 in PAD4 and Phe435 in PAD2 have similar roles in the interaction between the monomers.
Table 2 Recovery of recombinant human PAD2 in serum/plasma. Spiked rhPAD2 (ng/mL) 10 3,33 1,11 0,37
Seruma (N = 2)
Plasmaa (N = 2)
Mean
SD
Recovery
Mean
SD
Recovery
10.0 3.17 1.12 0.36
0.46 0.05 0.02 0.05
100% 95% 101% 96%
10.6 3.27 1.14 0.37
0.30 0.09 0.16 0.01
106% 98% 102% 101%
a Plasma and serum samples were spiked with rhPAD2 and the PAD2-concentration was measured by the ELISA using DN2 as capture antibody and DN6 as detection antibody.
Fig. 7. Variability of PAD2 levels in normal blood. Serum concentration of PAD2 in 28 healthy donors, as assessed by the ELISA for human PAD2.
Numerous investigators have reported associations between PAD2 and PAD4 with inflammatory diseases, in particular with RA (Foulquier et al., 2007; Kinloch et al., 2008; Basu et al., 2011) but also other diseases, such as multiple sclerosis (Wood et al., 2008; Moscarello et al., 2013) and cancer (Cherrington et al., 2012). Enzymatic differences between the two isoforms have been shown in vitro (Nakayama-Hamada et al., 2005), but it is still not known if one or both of them contribute to the pathogenesis of RA. Our PAD2 ELISA may prove useful in future studies of PAD2's role as mediator of extracellular citrullination. Interestingly, PAD2 was measureable in 29% of 28 normal serum samples tested. These donors all fulfilled the criteria for Danish blood donors, i.e. they had no history of autoimmune disease or cancer, and had no recognized ongoing infection. In several inflammatory diseases, for example RA, circulating PAD2 levels can be expected to be elevated, and our assay facilitates studies aimed at the question whether circulating PAD2 levels or PAD2 levels in synovial fluid are correlated to clinical features like disease activity. Acknowledgments This study was supported by proof-of-concept means from the Danish Research Council and by Novo Nordic exploratory pre-seed grant. Part of this work was financially supported by The Netherlands Proteomics Centre, a program embedded in the Netherlands Genomics Initiative. The authors would like to thank Marianne Engström (Rheumatology Unit, Karolinska Institutet, Stockholm), and Judith Stammen-Vogelzangs (Department of Biomolecular Chemistry, Radboud University Nijmegen) for expert technical assistance with the immunohistochemistry procedures and expression of recombinant proteins, respectively. References Arandjelovic, S., McKenney, K.R., Leming, S.S., Mowen, K.A., 2012. ATP induces protein arginine deiminase 2-dependent citrullination in mast cells through the P2X7 purinergic receptor. J. Immunol. 189, 4112. Asaga, H., Yamada, M., Senshu, T., 1998. Selective deimination of vimentin in calcium ionophore-induced apoptosis of mouse peritoneal macrophages. Biochem. Biophys. Res. Commun. 243, 641. Basu, P.S., Majhi, R., Ghosal, S., Batabyal, S.K., 2011. Peptidyl-arginine deiminase: an additional marker of rheumatoid arthritis. Clin. Lab. 57, 1021.
22
D. Damgaard et al. / Journal of Immunological Methods 405 (2014) 15–22
Chang, X., Yamada, R., Suzuki, A., Sawada, T., Yoshino, S., Tokuhiro, S., Yamamoto, K., 2005. Localization of peptidylarginine deiminase 4 (PADI4) and citrullinated protein in synovial tissue of rheumatoid arthritis. Rheumatology (Oxford) 44, 40. Cherrington, B.D., Zhang, X., McElwee, J.L., Morency, E., Anguish, L.J., Coonrod, S.A., 2012. Potential role for PAD2 in gene regulation in breast cancer cells. PLoS One 7, e41242. Darrah, E., Rosen, A., Giles, J.T., Andrade, F., 2012. Peptidylarginine deiminase 2, 3 and 4 have distinct specificities against cellular substrates: novel insights into autoantigen selection in rheumatoid arthritis. Ann. Rheum. Dis. 71, 92. Forslind, K., Ahlmen, M., Eberhardt, K., Hafstrom, I., Svensson, B., 2004. Prediction of radiological outcome in early rheumatoid arthritis in clinical practice: role of antibodies to citrullinated peptides (anti-CCP). Ann. Rheum. Dis. 63, 1090. Foulquier, C., Sebbag, M., Clavel, C., Chapuy-Regaud, S., Al, B.R., Mechin, M.C., Vincent, C., Nachat, R., Yamada, M., Takahara, H., Simon, M., Guerrin, M., Serre, G., 2007. Peptidyl arginine deiminase type 2 (PAD-2) and PAD-4 but not PAD-1, PAD-3, and PAD-6 are expressed in rheumatoid arthritis synovium in close association with tissue inflammation. Arthritis Rheum. 56, 3541. Gregersen, P.K., Silver, J., Winchester, R.J., 1987. The shared epitope hypothesis. An approach to understanding the molecular genetics of susceptibility to rheumatoid arthritis. Arthritis Rheum. 30, 1205. Harauz, G., Musse, A.A., 2007. A tale of two citrullines—structural and functional aspects of myelin basic protein deimination in health and disease. Neurochem. Res. 32, 137. Huizinga, T.W., Amos, C.I., van der Helm-van Mil, A.H., Chen, W., van Gaalen, F.A., Jawaheer, D., Schreuder, G.M., Wener, M., Breedveld, F.C., Ahmad, N., Lum, R.F., de Vries, R.R., Gregersen, P.K., Toes, R.E., Criswell, L.A., 2005. Refining the complex rheumatoid arthritis phenotype based on specificity of the HLADRB1 shared epitope for antibodies to citrullinated proteins. Arthritis Rheum. 52, 3433. Kinloch, A., Lundberg, K., Wait, R., Wegner, N., Lim, N.H., Zendman, A.J., Saxne, T., Malmstrom, V., Venables, P.J., 2008. Synovial fluid is a site of citrullination of autoantigens in inflammatory arthritis. Arthritis Rheum. 58, 2287. Kohler, G., Milstein, C., 1975. Continuous cultures of fused cells secreting antibody of predefined specificity. J. Immunol. 174, 2453. Li, P., Li, M., Lindberg, M.R., Kennett, M.J., Xiong, N., Wang, Y., 2010. PAD4 is essential for antibacterial innate immunity mediated by neutrophil extracellular traps. J. Exp. Med. 207, 1853. Liu, Y.L., Chiang, Y.H., Liu, G.Y., Hung, H.C., 2011. Functional role of dimerization of human peptidylarginine deiminase 4 (PAD4). PLoS One 6, e21314. Makrygiannakis, D., Revu, S., Engstrom, M., af, K.E., Nicholas, A.P., Pruijn, G.J., Catrina, A.I., 2012. Local administration of glucocorticoids decreases synovial citrullination in rheumatoid arthritis. Arthritis Res. Ther. 14, R20.
Moscarello, M.A., Lei, H., Mastronardi, F.G., Winer, S., Tsui, H., Li, Z., Ackerley, C., Zhang, L., Raijmakers, R., Wood, D.D., 2013. Inhibition of peptidylarginine deiminases reverses protein-hypercitrullination and disease in mouse models of multiple sclerosis. Dis. Model. Mech. 6, 467. Nakayama-Hamada, M., Suzuki, A., Kubota, K., Takazawa, T., Ohsaka, M., Kawaida, R., Ono, M., Kasuya, A., Furukawa, H., Yamada, R., Yamamoto, K., 2005. Comparison of enzymatic properties between hPADI2 and hPADI4. Biochem. Biophys. Res. Commun. 327, 192. Orrenius, S., Zhivotovsky, B., Nicotera, P., 2003. Regulation of cell death: the calcium-apoptosis link. Nat. Rev. Mol. Cell Biol. 4, 552. Palarasah, Y., Skjodt, K., Brandt, J., Teisner, B., Koch, C., Vitved, L., Skjoedt, M.O., 2010. Generation of a C3c specific monoclonal antibody and assessment of C3c as a putative inflammatory marker derived from complement factor C3. J. Immunol. Methods 362, 142. Pritzker, L.B., Joshi, S., Gowan, J.J., Harauz, G., Moscarello, M.A., 2000. Deimination of myelin basic protein. 1. Effect of deimination of arginyl residues of myelin basic protein on its structure and susceptibility to digestion by cathepsin D. Biochemistry 39, 5374. Raijmakers, R., Zendman, A.J., Egberts, W.V., Vossenaar, E.R., Raats, J., SoedeHuijbregts, C., Rutjes, F.P., van Veelen, P.A., Drijfhout, J.W., Pruijn, G.J., 2007. Methylation of arginine residues interferes with citrullination by peptidylarginine deiminases in vitro. J. Mol. Biol. 367, 1118. Schellekens, G.A., de Jong, B.A., van den Hoogen, F.H., van de Putte, L.B., van Venrooij, W.J., 1998. Citrulline is an essential constituent of antigenic determinants recognized by rheumatoid arthritis-specific autoantibodies. J. Clin. Invest. 101, 273. Senshu, T., Kan, S., Ogawa, H., Manabe, M., Asaga, H., 1996. Preferential deimination of keratin K1 and filaggrin during the terminal differentiation of human epidermis. Biochem. Biophys. Res. Commun. 225, 712. Snir, O., Widhe, M., von, S.C., Lindberg, J., Padyukov, L., Lundberg, K., Engstrom, A., Venables, P.J., Lundeberg, J., Holmdahl, R., Klareskog, L., Malmstrom, V., 2009. Multiple antibody reactivities to citrullinated antigens in sera from patients with rheumatoid arthritis: association with HLA-DRB1 alleles. Ann. Rheum. Dis. 68, 736. van Venrooij, W.J., van Beers, J.J., Pruijn, G.J., 2011. Anti-CCP antibodies: the past, the present and the future. Nat. Rev. Rheumatol. 7, 391. Vossenaar, E.R., Zendman, A.J., van Venrooij, W.J., Pruijn, G.J., 2003. PAD, a growing family of citrullinating enzymes: genes, features and involvement in disease. Bioessays 25, 1106. Vossenaar, E.R., Radstake, T.R., van der Heijden, A., van Mansum, M.A., Dieteren, C., de Rooij, D.J., Barrera, P., Zendman, A.J., van Venrooij, W.J., 2004. Expression and activity of citrullinating peptidylarginine deiminase enzymes in monocytes and macrophages. Ann. Rheum. Dis. 63, 373. Wood, D.D., Ackerley, C.A., Brand, B., Zhang, L., Raijmakers, R., Mastronardi, F.G., Moscarello, M.A., 2008. Myelin localization of peptidylarginine deiminases 2 and 4: comparison of PAD2 and PAD4 activities. Lab. Invest. 88, 354.
