peptidylarginine deiminase 2 is required for tumor ... · 2.7. immunoﬂuorescence staining and net...

lable at ScienceDirect

Journal of Autoimmunity xxx (2017) 1e9

Contents lists avai

Journal of Autoimmunity

journal homepage: www.elsevier .com/locate/ jaut imm

Peptidylarginine deiminase 2 is required for tumor necrosis factoralpha-induced citrullination and arthritis, but not neutrophilextracellular trap formation

Mandar Bawadekar a, Daeun Shim a, Chad J. Johnson a, Thomas F. Warner b,Ryan Rebernick a, Dres Damgaard c, Claus H. Nielsen c, Ger J.M. Pruijn d, Jeniel E. Nett a, e,Miriam A. Shelef a, f, *

a Department of Medicine, University of Wisconsin, Madison, WI, USAb Department of Pathology and Laboratory Medicine, University of Wisconsin, Madison, WI, USAc Institute for Inflammation Research, Center for Rheumatology and Spine Diseases, Copenhagen University Hospital, Rigshospitalet, Copenhagen, Denmarkd Institute for Molecules and Materials and Radboud Institute for Molecular Life Sciences, Radboud University, Nijmegen, The Netherlandse Department of Medical Microbiology and Immunology, University of Wisconsin-Madison, Madison, WI, USAf William S. Middleton Memorial Veterans Hospital, Madison, WI, USA

a r t i c l e i n f o

Article history:Received 21 November 2016Received in revised form23 January 2017Accepted 31 January 2017Available online xxx

Keywords:CitrullinationPeptidylarginine deiminaseTNFaNeutrophil extracellular trapRheumatoid arthritisMurine model

* Corresponding author. UW Medical Foundation4130), 1685 Highland Avenue, Madison, WI 53705, US

E-mail address: [email protected] (M.A.

http://dx.doi.org/10.1016/j.jaut.2017.01.0060896-8411/Published by Elsevier Ltd.

Please cite this article in press as: M. Bawcitrullination and arthritis, but not neutrj.jaut.2017.01.006

a b s t r a c t

Citrullination, the post-translational conversion of arginines to citrullines, may contribute to rheumatoidarthritis development given the generation of anti-citrullinated protein antibodies (ACPAs). However, it isnot known which peptidylarginine deiminase (PAD) catalyzes the citrullination seen in inflammation.PAD4 exacerbates inflammatory arthritis and is critical for neutrophil extracellular traps (NETs). NETsdisplay citrullinated antigens targeted by ACPAs and thus may be a source of citrullinated protein.However, PAD4 is not required for citrullination in inflamed lungs. PAD2 is important for citrullination inhealthy tissues and is present in NETs, but its role in citrullination in the inflamed joint, NETosis andinflammatory arthritis is unknown. Here we use mice with TNFa-induced inflammatory arthritis, amodel of rheumatoid arthritis, to identify the roles of PAD2 and PAD4 in citrullination, NETosis, andarthritis. In mice with TNFa-induced arthritis, citrullination in the inflamed ankle was increased asdetermined by western blot. This increase was unchanged in the ankles of mice that lack PAD4. Incontrast, citrullination was nearly absent in the ankles of PAD2-deficient mice. Interestingly, PAD2 wasnot required for NET formation as assessed by immunofluorescence or for killing of Candida albicans asdetermined by viability assay. Finally, plasma cell numbers as assessed by flow cytometry, IgG levelsquantified by ELISA, and inflammatory arthritis as determined by clinical and pathological scoring wereall reduced in the absence of PAD2. Thus, PAD2 contributes to TNFa-induced citrullination and arthritis,but is not required for NETosis. In contrast, PAD4, which is critical for NETosis, is dispensable forgeneralized citrullination supporting the possibility that NETs may not be a major source of citrullinatedprotein in arthritis.

Published by Elsevier Ltd.

1. Introduction

Citrullination, the post-translational conversion of arginines in aprotein to citrullines, has gained attention related to the observa-tion that anti-citrullinated protein antibodies (ACPAs) can be

Centennial Building (RoomA.Shelef).

adekar, et al., Peptidylargininophil extracellular trap form

detected in about 75% of people with rheumatoid arthritis [1]. Thus,protein citrullination (and similar post-translational modifications)may be a trigger in genetically susceptible individuals for the initialloss of tolerance seen in rheumatoid arthritis. Once tolerance is lost,inflammation increases, inflammatory cytokines like TNFa rise, andclinical rheumatoid arthritis develops.

Although ACPAs are a hallmark of rheumatoid arthritis, cit-rullination increases in a variety of inflamed tissues [2] includingthe rheumatoid joint [3]. Peptidylarginine deiminases (PADs)

e deiminase 2 is required for tumor necrosis factor alpha-inducedation, Journal of Autoimmunity (2017), http://dx.doi.org/10.1016/

M. Bawadekar et al. / Journal of Autoimmunity xxx (2017) 1e92

catalyze citrullination, but which PAD catalyzes inflammation-induced citrullination in rheumatoid arthritis and other diseasesis unclear. Single nucleotide polymorphisms (SNPs) in PADI2 andPADI4, the genes that encode PAD2 and PAD4, have been associatedwith rheumatoid arthritis [4e12], suggesting that theymay both beinvolved in disease pathogenesis. Further, both PAD2 and PAD4 arefound in immune cells [13e15] as well as in the rheumatoid joint[16,17] and a pan-PAD inhibitor reduces collagen induced arthritis[18]. Also, both PAD2 and PAD4 can hypercitrullinate proteinsduring immune-mediated membranolysis, which has been impli-cated in ACPA formation [19]. Adding complexity, both PAD4 andPAD2 can autocitrullinate [20e22], which may regulate their ac-tivity in disease. Thus, PAD2, PAD4, or both could citrullinate pro-teins in inflamed areas. Understanding the individual contributionsof the PADs is important since inhibiting specific PAD enzymesmight prove useful in treating rheumatoid arthritis. Indeed, this isan active area of investigation by both academic and commercialresearch labs with several isoform-specific inhibitors generated todate [23,24].

There is additional evidence that PAD4 may be important forcitrullination in rheumatoid arthritis. PAD4 contributes to inflam-mation and arthritis in three different murine models of rheuma-toid arthritis [25e27]. Also, PAD4 is required for LPS-inducedhistone citrullination and neutrophil extracellular trap (NET) for-mation [28,29] and NETs display some of the same citrullinatedantigens that are targeted by ACPAs [30]. Together, these findingssuggest that PAD4 may contribute to the increased citrullinationseen in arthritic joints. However, no reduction in serum [25] or lung[31] protein citrullination is seen in TNFa-induced arthritis whenPAD4 is absent.

Thus, PAD2 may be the major contributor to joint citrullinationin inflammatory arthritis. In support of this idea, synovial fluidPAD2 levels correlate with overall PAD activity as well as diseaseactivity in rheumatoid arthritis [32]. Moreover, PAD2-deficientmice have a reduction in PAD activity in healthy tissues as well asreduced citrullination in the central nervous system in experi-mental autoimmune encephalomyelitis [33,34]. Also, PAD2 displaysless restrictive substrate specificity than PAD4 [35], which couldlead to citrullination of more arginines. Finally, PAD2 is present onNETs [36] and is required to citrullinate histones in breast cancer[37], but it is not known if PAD2 is required for citrullination ininflamed joints, NETosis or arthritis.

Here we use mice with TNFa-induced arthritis to determine ifPAD2 or PAD4 is required for citrullination in inflamed joints and ifPAD2 is required for NETosis and inflammatory arthritis.

2. Material and methods

2.1. Animals

Tg3647 mice, which overexpress systemically one copy of theTNFa transgene (TNFþ mice) [38], were crossed with PAD4-deficient mice [28] to generate TNFþPAD4þ/þ and TNFþPAD4�/�

mice. TNFþ mice were also crossed with PAD2-deficient mice [34]to generate TNFþPAD2þ/þ and TNFþPAD2�/� mice. Littermateswere used for all experiments and sex matched whenever possible.Approximately equal numbers of male and female mice were usedfor experiments. PAD2 and PAD4 were confirmed to be deleted.Mice were maintained in pathogen-free conditions and all murineexperiments were approved by the University of Wisconsin AnimalCare and Use Committee.

2.2. Protein lysates

One ankle joint per mouse was harvested, flash frozen and

Please cite this article in press as: M. Bawadekar, et al., Peptidylarginincitrullination and arthritis, but not neutrophil extracellular trap formj.jaut.2017.01.006

homogenized using a Bullet Blender (Next Advance, Averill Park,USA) in RIPA buffer (Sigma Aldrich, St. Louis, USA) supplementedwith protease inhibitor cocktail (Sigma Aldrich). Lysates weresonicated briefly and total protein in the supernatants wasquantified.

2.3. Western blots to detect citrullination

Equal amounts of protein were subjected to gel electrophoresisusing two polyacrylamide gels. One gel was stained with brilliantblue G colloidal solution (Sigma Aldrich) and the other was trans-ferred to a polyvinylidene difluoride (PVDF) membrane (Bio-Rad,Hercules, USA) using a semi-dry blotting system (Bio-Rad). Themembrane was then subjected to the Anti-Citrulline (Modified)Detection Kit (AMC kit) (Millipore Sigma) to visualize citrullinatedproteins. Imaging was performed using ImageQuant LAS 4000 (GEHealthcare, Madison, USA), followed by densitometry usingQuantity One 1-D Analysis Software (Bio-Rad). Brilliant blue Gstained gels were imaged using an Odyssey infrared scanner (LI-COR Biosciences, Lincoln, USA), followed by densitometry. AMCsignal was normalized to total protein as detected by brilliant blueand blots were normalized to each other for overall signal intensity.Blots treated with the AMC kit without primary antibody showedno signal (data not shown).

2.4. Western blots to detect PAD2

Equal amounts of protein were subjected to polyacrylamide gelelectrophoresis followed by transfer to PVDF membrane as above.The membrane was blocked with 5% milk, incubated at 4 �C over-night with rabbit anti-actin antibody (Sigma Aldrich) and anti-PAD2 mouse monoclonal IgG (clone DN2) [39] diluted 1:100 and1:200 respectively in 2.5% milk, washed, incubated with anti-mouse IgG conjugated to IRDye 680LT and anti-rabbit IgG conju-gated to IRDye 800LT (LI-COR Biosciences, Lincoln, NE, USA), andwashed. Blots were imaged using the Odyssey infrared scannerfollowed by densitometry.

2.5. Neutrophil purification

Femurs and tibias were flushed using RPMI1640 (Thermo FisherScientific, Waltham, USA) with 2% fetal bovine serum (FBS)(Thermo Fisher Scientific) and 2 mM EDTA (Sigma Aldrich). Neu-trophils were purified using EasySepMouse Neutrophil EnrichmentKit (Stemcell Technologies, Vancouver, Canada) according to themanufacturer's instructions. Purity of neutrophils was confirmedby flow cytometry to be 92%.

2.6. Quantitative polymerase chain reaction (qPCR)

RNA was isolated using Trizol reagent according to the manufac-turer's instructions (Thermo Fisher Scientific) followed by treatmentwith DNase I (Sigma Aldrich). cDNA was synthesized using iScriptcDNA synthesis kit according to themanufacturer's instructions (Bio-Rad). qPCR was performed on equal amounts of cDNA using iTaqUniversal SYBR Green Supermix (Bio-Rad) on a StepOnePlus Real-Time PCR system (Thermo Fisher Scientific). Primers for PAD2 (for-ward: 50-AGCAGCGGAGGGCTTAC-30, reverse: 50- CACGCGGCTCCCA-TACT-30) and 18s (forward: 50-GAATAATGGAATAGGACCGCGG-30,reverse: 50-GGAACTACGACGGTATCTGATC-30) were designed usingIDT PrimerQuest Tool. All genes were normalized to 18S rRNA andrelative gene expression (fold change) was calculated by 2�DDCt

method.


M. Bawadekar et al. / Journal of Autoimmunity xxx (2017) 1e9 3

2.7. Immunofluorescence staining and NET quantification

All chemicals, unless otherwise indicated, are from SigmaAldrich. Purified neutrophils were treated with 100 ng/ml TNFa inRPMI for 30 min on poly-L-lysine coated glass coverslips, followedby the addition of 10 mg/ml lipopolysaccharide (LPS) and then in-cubation for 4 h at 37 �C with 5% CO2. Coverslips were fixed andpermeabilized with a solution of 4% paraformaldehyde, 1% NP-40,and 0.5% Triton X-100 in phosphate buffered saline (PBS) for30 min, washed, and blocked for 1.5 h at room temperature with2.5% bovine serum albumin (BSA), 5% goat serum, and 0.5% Tween-20 in PBS. The coverslips were then incubated for 1 h with rabbitanti-histone H4 (citrulline 3) antibody (MilliporeSigma) diluted1:200 in block solution, washed, incubated with a 1:200 dilution ofdonkey anti-rabbit TRITC and a 1:1000 dilution of 40,6-diamidino-2-phenylindole (DAPI) in blocking solution for 30 min at roomtemperature, washed, and mounted using Aquamount (ThermoFisher Scientific). Five predetermined locations on the coverslipwere imaged using a Leica Fluorescence Microscope at 400�. Thetotal number of neutrophils and NETs (defined as large, cloud-likemasses with protrusions staining positive for DAPI and citrulli-nated histone H4) were counted manually in a blinded manner foreach of the 5 images.

2.8. Candida killing assay

All chemicals, unless otherwise indicated, are fromSigmaAldrich.A modified XTT (2,3-bis-(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide) assay was performed [40,41]. Candidaalbicans strain SC5314 [42]was stored in 15% glycerol stock at�80 �Cand maintained on yeast extract-peptone-dextrose (YPD) mediumsupplemented with uridine (1% yeast extract, 2% peptone, 2%dextrose medium, 0.08% uridine) prior to experiments. A singlecolony was used to propagate a culture in YPD with uridine at 30 �Cwith orbital shaking at 220 RPM overnight. To obtain exponentiallygrowing cells, cultures were diluted 20-fold and incubated for anadditional 2 h. Candida were collected by centrifugation, washedtwice with PBS and adjusted to a concentration of 4� 107 cells/ml inRPMI 1640 (without phenol red) supplemented with 2% FBS. Neu-trophils were suspended at 4 � 106 cells/mL in identical culturemedia and pre-incubated with TNFa (100 ng/ml) for 30 min on ice.Next, 40 mL of neutrophils or culture media and 4 ml of Candidawereadded to each well of a 384 well clear plate (Corning, USA) with100 nM phorbol myristate acetate (PMA). To somewells, 10 mg/ml ofcytochalasin D was added [43]. Plates were incubated at 37 �C with5% CO2 for 4 h. Neutrophils were then lysed for 15 min at 37 �C in afinal concentration of 0.03% TritonX-100. Background absorption at492 nm was pre-read in a Synergy H1 microplate reader (BioTek,Winooski, USA) followed by the addition of 30 mL of 9:1 XTTworkingsolution (0.75 mg/ml XTT in PBS with 2% glucose: phenazinemethosulfate 0.32 mg/ml in ddH2O). Following a 25 min incubationat 37 �C in 5% CO2, absorption was read at 492 nm and the baselineabsorbance was subtracted. Experiments were performed in tripli-cate. To determine percent killing values, neutrophils withoutCandidawere subtracted fromneutrophilswith Candida andallwerenormalized to Candida alone.

2.9. Enzyme-linked immunosorbent assay (ELISA)

Sera were collected by cardiac puncture and subjected to amouse IgG ELISA quantification kit according to manufacturer'sprotocol (Bethyl Laboratories Inc., Montgomery, USA).


2.10. Flow cytometry

Red blood cells were lysed using a hypotonic NaCl solution.Remaining cells were suspended in MACS buffer (1% BSA, 2% FBS,0.03% sodium azide and 2 mM EDTA) and stained using anti-mouseB220 conjugated to allophycocyanin (clone RA3-6B2, eBioscience,San Diego, USA) and anti-mouse CD138 conjugated to phycoery-thrin (clone 281-2, BD Biosciences, San Jose, CA, USA). Sampleswere acquired using a BD FACSCalibur machine and data wasanalyzed using FlowJo software.

2.11. Clinical arthritis scoring

Scoring was performed in a blinded manner on a scale of 0e3 aspreviously described [25]. Scale: 0, no arthritis; 0.5, mild jointdeformity and swelling; 1.0, moderate joint deformity andswelling; 1.5, moderate/severe joint deformity, moderate swelling,and reduced grip strength; 2.0, severe joint deformity, moderateswelling, and no grip strength; 2.5, severe joint deformity, mod-erate/severe swelling, and no grip strength; 3.0, severe jointdeformity and swelling and no grip strength.

2.12. Pathology

Rear hind legs were fixed in 10% neutral buffered formalin,decalcified for 30 h with Surgipath Decalcifier 1 (Leica Biosystems,Lincolnshire, USA), processed with a VIP processor, embedded inparaffin (Sakura Tissue Tek, Torrance, CA, USA), sectioned at 5 mmon a Leica microtome, and stained with hematoxylin and eosin(H&E). A board certified pathologist scored the severity of synovitisat the tibio-talar joint on a scale of 0e4 (0, none; 1, mild; 2, mild tomoderate; 3, moderate; 4, severe) and determined the percentageof the surface of the tibio-talar joint that was eroded in a blindedmanner.

2.13. Statistics

All data was analyzed using a paired t-test with the p valuereported as significant if less than or equal to 0.05.

3. Results

3.1. Joint citrullination is increased in TNFa-induced arthritiswithout a significant requirement for PAD4

TNFþ mice develop an inflammatory, destructive arthritis verysimilar to rheumatoid arthritis [38]. Protein lysates from the anklesof wild type and TNFþ mice were assessed for total protein cit-rullination by western blot using an anti-modified citrullinedetection kit. As shown in Fig. 1A and B, generalized protein cit-rullination was increased in TNFþ compared to wild type joints. Todetermine if PAD4 is required for joint protein citrullination inTNFa-induced arthritis, we prepared ankle lysates fromTNFþPAD4þ/þ and TNFþPAD4�/� mice for western blot as above.There was no loss of citrullination in TNFþPAD4�/� compared toTNFþPAD4þ/þ ankles (Fig. 1C and D). Therefore, PAD4 is notrequired for gross protein citrullination in the arthritic joints ofTNFa-overexpressing mice as detected by the anti-modifiedcitrulline detection kit.

3.2. PAD2 is required for joint citrullination in TNFa-inducedarthritis

To determine the requirement for PAD2 in joint citrullination,we assessed total protein citrullination by western blot as above in


Fig. 1. Joint citrullination is increased in TNFa overexpressing mice without a requirement for PAD4. Ankle protein lysates from 4 to 5 month old TNFþ and littermate control(WT) mice were subjected to western blot to detect citrullinated proteins and gel electrophoresis to detect total protein. A. Representative blot (left) and gel (right). B. The density ofthe entire lanes was determined and citrullinated protein signal was normalized to total protein signal. Average and SEM are graphed (n ¼ 9, *p < 0.05). Ankle lysates from 4 to 5month old TNFþPAD4þ/þ and TNFþPAD4�/� littermates were similarly assessed. C. Representative blot (left) and gel (right). D. Graph with average and SEM (n ¼ 8).


arthritic ankles from TNFþPAD2þ/þ and TNFþPAD2�/� mice. Asshown in Fig. 2A and B, there was a dramatic loss of protein cit-rullination in TNFþPAD2�/� compared to TNFþPAD2þ/þ arthriticankles. Since PAD activity in multiple tissues is lost in PAD2-deficient mice [33], we hypothesized that PAD2 also would beresponsible for citrullination in the non-inflamed joint. Thus, weassessed protein citrullination by western blot as above in the an-kles of PAD2þ/þ and PAD2�/�mice that do not overexpress TNFa. Asshown in Fig. 2C and D, citrullination was significantly reduced innon-inflamed PAD2�/� ankles. Taken together, these data suggestthat PAD2 is required for baseline and TNFa-induced citrullinationin the joint.

3.3. PAD2 is not increased in inflamed joints in the absence of PAD4

PAD2 expression has been shown to be increased in the absenceof PAD4 in neutrophils [26]. Therefore, we hypothesized that theunaltered citrullination in PAD4-deficient arthritic joints might bedue to upregulation of PAD2. To test this hypothesis, first we pu-rified neutrophils from PAD4þ/þ and PAD4�/� mice and assessedPAD2 expression by qPCR. As shown in Fig. 3A, PAD2 was onlymildly upregulated in PAD4-deficient neutrophils. Althoughmodest, these levels are significantly different. We then used amonoclonal antibody against PAD2 [39] to determine if PAD2 is

Fig. 2. PAD2 is required for joint citrullination. Protein lysates from the ankles of 4e5 mocitrullinated proteins and gel electrophoresis to detect total protein. A. Representative blot (protein signal was normalized to total protein signal. Average and SEM are graphed for thePAD2þ/þ and PAD2�/� littermates were similarly assessed. C. Representative blot (left) and g


increased in arthritic joints in the absence of PAD4. As a control, wesubjected ankle lysates from TNFþPAD2þ/þ and TNFþPAD2�/� miceto western blot to detect PAD2 (Fig. 3B). We then assessed PAD2protein levels in TNFþPAD4þ/þ and TNFþPAD4�/� arthritic anklesand found no significant increase in PAD2 in TNFþPAD4�/�

compared to TNFþPAD4þ/þ ankles (Fig. 3C). Thus, upregulation ofPAD2 does not appear to explain the unaltered levels of citrulli-nation in TNFþPAD4�/� ankles.

3.4. PAD2 is not required for NETosis

Given the loss of citrullination in the inflamed joints of PAD2-deficient mice, we hypothesized that PAD2 might be required forNETosis. To test this hypothesis, we purified neutrophils fromPAD2þ/þ and PAD2�/� mice, stimulated with TNFa and LPS, per-formed immunofluorescent staining, and quantified NETosis. As acontrol, PAD4þ/þ and PAD4�/� neutrophils were treated identically.As shown in Fig. 4A and B, PAD4�/� neutrophils were unable tomake NETs as expected [28]. In contrast, PAD2�/� neutrophils wereable to make NETs (Fig. 4C and D). NETs can kill a variety of path-ogens including Candida albicans [41,44]. To test if PAD2�/� neu-trophils could kill Candida, we subjected PAD2þ/þ and PAD2�/�

neutrophils to a modified XTT assay. As shown in Fig. 4E, PAD2�/�

neutrophils could effectively kill Candida regardless of whether

nth old TNFþPAD2þ/þ and TNFþPAD2�/� mice were subjected to western blot to detectleft) and gel (right). B. The density of the entire lanes was determined and citrullinatedentire lane (TNFþPAD2þ/þ n ¼ 9, TNFþPAD2�/� n ¼ 8, *p < 0.05). Ankle lysates fromel (right). D. Graph with average and SEM (PAD2þ/þ n ¼ 4, PAD2�/� n ¼ 5, **p < 0.01).


Fig. 3. PAD2 is not increased in inflamed joints in the absence of PAD4, but is upregulated in PAD4-deficient neutrophils. A. RNA from PAD4þ/þ and PAD4�/� neutrophils wassubjected to qPCR to assess PAD2 gene expression levels with average and SEM graphed for the fold change (n ¼ 6). B. Protein lysates from the ankles of TNFþPAD2þ/þ andTNFþPAD2�/� mice were assessed for PAD2 protein by western blot. Representative blots (upper panel) and average with SEM for quantification of PAD2 normalized to actin in thelower panel (TNFþPAD2þ/þ n ¼ 9, TNFþPAD2�/� n ¼ 8). C. Protein lysates from the ankles of TNFþPAD4þ/þ and TNFþPAD4�/� mice were assessed for PAD2 protein by western blot.Representative blots (upper panel) and average with SEM for quantification of PAD2 normalized to actin in the lower panel (n ¼ 7). For all panels, *p < 0.05.


phagocytosis was blocked with cytochalasin D or not. Takentogether, these data suggest that PAD2 is not required for neutro-phils to form NETs or kill Candida.

3.5. PAD2 is required for plasma cells and IgG in TNFa-inducedarthritis

NETosis occurs in the absence of PAD2, but other immunefunctions could be impaired in PAD2-deficient mice. TNFþPAD4�/�

mice have reduced IgG [25]. Therefore, we tested if PAD2 was alsorequired for IgG levels in TNFa-induced arthritis. IgGwas quantifiedfor TNFþPAD2þ/þ and TNFþPAD2�/� sera by ELISA. As shown inFig. 5A, TNFþPAD2�/� mice had reduced serum IgG compared toTNFþPAD2þ/þ mice. Of note, female TNFa-overexpressing micetend to have higher levels of IgG, but the reduction in IgG in theabsence of PAD2 was independent of the sex of the mice.

We have previously shown that PAD4-deficient mice that do notoverexpress TNFa have normal levels of IgG [25]. Thus, we deter-mined if serum IgG levels in PAD2-deficient mice in the absence ofTNFa overexpression were normal. As shown in Fig. 5B, despite anoverall lower level of IgG in mice that do not overexpress TNFa,PAD2�/� mice have reduced total IgG compared to PAD2þ/þ mice.Therefore, PAD2 appears to be required to achieve the IgG levelsseen in arthritic and non-arthritic mice.

Since IgG is primarily made by B220LOCD138HI plasma cells inthe bone marrow [45], we determined if bone marrow plasma cellswere reduced in TNFþPAD2�/�mice by flow cytometry. As shown inFig. 5C and D, bone marrow plasma cells were, indeed, reduced inTNFþPAD2�/� compared to TNFþPAD2þ/þ mice. These data suggestthat PAD2 is required for normal plasma cell numbers and IgGlevels in TNFa-induced arthritis.

3.6. PAD2 is required for maximal severity of TNFa-inducedarthritis

We have previously shown that PAD4 exacerbates TNFa-


induced arthritis [25]. To determine if PAD2 contributes to TNFa-induced arthritis, we clinically scored arthritis severity inTNFþPAD2þ/þ and TNFþPAD2�/� mice based on joint swelling, jointdeformity and grip strength monthly [25]. We saw no difference inarthritis severity at 2 and 3 months of age (Fig. 6A, left panel).However, at 4 months of age, there was a significant, althoughmodest, reduction in arthritis in TNFþPAD2�/� compared toTNFþPAD2þ/þ mice (Fig. 6A, left panel). Since PAD2 is expressed inmuscle [46], we were concerned that TNFþPAD2�/� mice mighthave reduced grip strength due to a role for PAD2 in muscle makingarthritis scores higher for PAD2-deficient mice unrelated to in-flammatory arthritis severity. Therefore, we also calculated clinicalarthritis scores using only joint swelling and deformity. As shownin Fig. 6A (right panel), arthritis scores were reduced at both 3 and 4months of age in TNFþPAD2�/� compared to TNFþPAD2þ/þ micewhen grip strength was not included in the score. Finally, we scoredthe joints histologically for the percent erosion of the tibio-talarjoint surface as well as the extent of synovitis. As shown inFig. 6BeD, there was reduced erosion and synovitis of the tibio-talar joint in TNFþPAD2�/� compared to TNFþPAD2þ/þ mice.Together, our data suggest that PAD2 is required for maximalseverity of TNFa-induced arthritis.

4. Discussion

As expected, we found that TNFa overexpression leads toincreased citrullination in inflamed joints consistent with thefinding that TNFa drives citrullination in the serum [25] and lung[31]. TNFa may also drive citrullination in the human rheumatoidjoint providing fodder for continued inflammation particularly inACPAþ individuals. If true, this would provide further support for acircular model of rheumatoid arthritis pathogenesis in which cit-rullination contributes to the formation of ACPAs, which deposit inthe joint and activate immune cells to produce TNFa and othercytokines, which then induce citrullination to fuel the fire.

However, despite the known requirement for PAD4 in histone


Fig. 4. PAD2 is not required for NETosis. Neutrophils from PAD4þ/þ and PAD4�/� mice were untreated or treated with TNFa and LPS followed by staining for citrullinated histoneH4 (red) and DNA (blue). A. Representative images at 400�. B. Graph with average and SEM for the percent of cells that formed NETs (n ¼ 3, *p ¼ 0.05). C. Neutrophils from PAD2þ/þ

and PAD2�/� mice were treated identically to the PAD4þ/þ and PAD4�/� neutrophils. C. Representative images at 400�. D. Graph with average and SEM for the percent of cells thatformed NETs (n ¼ 5). E. PAD2þ/þ and PAD2�/� neutrophils were incubated with Candida in the presence or absence of Cytochalasin D. The percent of Candida killed was detected byXTT. Average and SEM are graphed (n ¼ 3). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)


citrullination [28,29], PAD4 does not appear to be required forgeneralized TNFa-induced joint citrullination as detected by theanti-modified citrulline antibody. This method likely does notdetect every citrullinated protein, but use of the citrulline proberhodamine-phenylglyoxal (Rh-PG) [47] is limited by extensivebackground and use of the mouse monoclonal anti-peptidylcitrulline antibody F95 [48] is limited by binding of theanti-mouse IgM secondary antibody inmurine joint tissue (data notshown). Nonetheless, our finding is consistent with the observationthat PAD4 is not required for TNFa-induced lung [31] or serum [25]protein citrullination, as determined by other methods. Thus, weconclude that PAD4 is not required for at least a significant portionof the increased citrullination seen in TNFa-induced arthritis.

A role for PAD4 in TNFa-induced arthritis [25] without a clearrequirement for generalized protein citrullination suggests thatPAD4 may contribute to rheumatoid arthritis apart from antigencitrullination. This idea is consistent with the finding that geneticvariants in PADI4 associatewith both ACPAþ and ACPA� rheumatoidarthritis [49] as well as ulcerative colitis [50], an inflammatorydisorder that does not involve ACPAs. Further, PAD4 contributes toIgG levels and T cell activation [25,27]. Thus, even though PAD4 is acitrullinating enzyme and protein citrullination is increased ininflammation, PAD4 may contribute to rheumatoid arthritis apartfrom the generation of citrullinated antigens.


In this manuscript, we have shown a role for PAD2 in TNFa-induced arthritis, providing the first evidence that PAD2 contrib-utes to inflammatory arthritis. The reduction in arthritis that wedetect in the absence of PAD2 is modest. Previously, we demon-strated that PAD4 contributes to TNFa-induced arthritis at 5months of age, with little to no effect at younger ages [25]. Unfor-tunately, TNFþ mice, either with or without PAD2, die earlier forunclear reasons in our current animal facility than in our previousfacility. Thus, we were unable to score arthritis consistently at 5months of age. It is possible that if we had been able to scorearthritis regularly at 5 months, we would have detected an evenlarger reduction in arthritis in PAD2-deficient mice. Also, mice withTNFa-induced arthritis do not produce classic ACPAs [25,51]. Giventhe requirement for PAD2 in generating citrullinated protein, onemight expect an even larger role for PAD2 in ACPA-dependentarthritis, which could be the case in human rheumatoid arthritis.

The reduced arthritis in PAD2-deficient mice may be related to areduction in plasma cells and IgG levels, especially since PAD2 is notrequired for NETosis. Interestingly, TNFþPAD4�/� mice have areduction in serum IgG levels, but not plasma cells [25]. It ispossible that plasma cells or serum IgG are reduced at least in partsecondary to overall reduced inflammation in the PAD-deficientmice. Alternatively, the PADs may regulate plasma cells and anti-bodies. PAD2 and PAD4 may be important within B lineage cells for


Fig. 5. PAD2 contributes to plasma cells and IgG levels in TNFa-induced arthritis. A.Total serum IgG was quantified for 4e5 month old TNFþPAD2þ/þ and TNFþPAD2�/�

mice by ELISA with average and SEM graphed (n ¼ 8, **p < 0.01). B. Total serum IgGwas quantified for 2 month old PAD2þ/þ and PAD2�/� mice by ELISA with average andSEM graphed (n ¼ 7, *p < 0.05). Bone marrow from TNFþPAD2þ/þ and TNFþPAD2�/�

mice was stained for B220 and CD138 and subjected to flow cytometry. C. Represen-tative zebra plots of live cells. D. Average and SEM for the percentage of B220LOCD138HI

plasma cells is graphed (n ¼ 9, *p < 0.05).

Fig. 6. PAD2 is required for TNFa-induced inflammatory arthritis. A. Severity ofclinical arthritis was scored on a scale of 0e3 in 2, 3, and 4 month old TNFþPAD2þ/þ

and TNFþPAD2�/� mice including (left) or not including (right) grip strength withaverage and SEM graphed (2 and 3 months: n ¼ 25 TNFþPAD2þ/þ and n ¼ 24TNFþPAD2�/� mice; 4 months: n ¼ 19 pairs; *p < 0.05). The ankles of 4e5 month oldTNFþPAD2þ/þ and TNFþPAD2�/� mice were fixed, sectioned, and stained with H&E. B.Representative images with arrows indicating inflamed synovium/pannus invading thetibio-talar joint (Magnification 100�, bar 500 mm). The percentage of surface erosion(C) and extent of synovitis on a scale of 0e4 (D) at the tibio-talar joint was determinedin a blinded manner with average and SEM graphed (TNFþPAD2þ/þ n ¼ 11,TNFþPAD2�/� n ¼ 10, *p < 0.05).


regulating IgG production or other pathways, although expressiondata for the PADs in B cells is conflicted [9,14,15,52], or important incells that regulate B cells. For example, PAD2 is upregulated in thebone marrow mesenchymal stem cells of patients with multiplemyeloma, a plasma cell malignancy, driving production of IL-6, amolecule that supports plasma cells [53,54]. Perhaps in the absenceof PAD2, IL-6 is reduced in the bone marrow microenvironmentleading to reduced plasma cells and IgG levels. Further work isneeded to clarify the role of PAD2 and PAD4 in the regulation ofplasma cells and antibodies.

Finally, we show that PAD2 is required for joint citrullination inTNFaeinduced arthritis. If the relative contributions of PAD2 andPAD4 to citrullination that we observed in mice apply to humans,then PAD2 may be responsible for the bulk of citrullinated proteinstargeted by ACPAs in rheumatoid arthritis. Interestingly, despite therequirement for PAD2 in TNFa-induced joint citrullination and theobservation that NETs are source of citrullinated antigens [30],PAD2 is not required for NETosis in our experiments. In contrast,PAD4, which is critical for NETosis [28,29], does not play a majorrole in gross citrullination in TNFa-induced arthritis. Together,these findings suggest that NETs may not be the main source ofcitrullinated protein in arthritic mice. Perhaps the main source of


citrullinated proteins in TNFa-driven arthritis, and potentiallyrheumatoid arthritis given the high levels of TNFa, is immune-mediated membranolysis-induced hypercitrullination, which canbe catalyzed by PAD2 [19]. It is impossible to directly extrapolateour findings to human disease. Nonetheless, even if NETs do notprovide the majority of citrullinated proteins in rheumatoidarthritis, they may still be important for generating and/or dis-playing key citrullinated antigens targeted by ACPAs.

5. Conclusion

Our data support an important role for PAD2 in citrullinationand arthritis severity in TNFa-induced inflammatory arthritiswithout a major role in NETosis suggesting that NETs may not bethe main source of citrullinated protein in rheumatoid arthritis.

Acknowledgements

This work was supported by the Rheumatology Research



Foundation via a Scientist Development Award and a NationalInstitute of Health (NIH) K08 AR065500 to MAS as well as a NIHK08 AI108727 and Burroughs Wellcome Fund 1012299 to JEN.

References

[1] W.J. van Venrooij, J.J. van Beers, G.J. Pruijn, Anti-CCP antibodies: the past, thepresent and the future, Nat. Rev. Rheumatol. 7 (2011) 391e398.

[2] D. Makrygiannakis, E. af Klint, I.E. Lundberg, R. Lofberg, A.K. Ulfgren,L. Klareskog, et al., Citrullination is an inflammation-dependent process, Ann.Rheum. Dis. 65 (2006) 1219e1222.

[3] A. Kinloch, K. Lundberg, R. Wait, N. Wegner, N.H. Lim, A.J. Zendman, et al.,Synovial fluid is a site of citrullination of autoantigens in inflammatoryarthritis, Arthritis Rheum. 58 (2008) 2287e2295.

[4] S. Hou, G.P. Gao, X.J. Zhang, L. Sun, W.J. Peng, H.F. Wang, et al., PADI4 poly-morphisms and susceptibility to rheumatoid arthritis: a meta-analysis, Mod.Rheumatol. 23 (2013) 50e60.

[5] C.P. Kang, H.S. Lee, H. Ju, H. Cho, C. Kang, S.C. Bae, A functional haplotype of thePADI4 gene associated with increased rheumatoid arthritis susceptibility inKoreans, Arthritis Rheum. 54 (2006) 90e96.

[6] F. Gandjbakhch, I. Fajardy, B. Ferre, S. Dubucquoi, R.M. Flipo, N. Roger, et al.,A functional haplotype of PADI4 gene in rheumatoid arthritis: positive cor-relation in a French population, J. Rheumatol. 36 (2009) 881e886.

[7] K. Panati, S. Pal, K.V. Rao, V.D. Reddy, Association of single nucleotide poly-morphisms (SNPs) of PADI4 gene with rheumatoid arthritis (RA) in Indianpopulation, Genes Genet. Syst. 87 (2012) 191e196.

[8] R.M. Plenge, L. Padyukov, E.F. Remmers, S. Purcell, A.T. Lee, E.W. Karlson, et al.,Replication of putative candidate-gene associations with rheumatoid arthritisin >4,000 samples from North America and Sweden: association of suscep-tibility with PTPN22, CTLA4, and PADI4, Am. J. Hum. Genet. 77 (2005)1044e1060.

[9] A. Suzuki, R. Yamada, X. Chang, S. Tokuhiro, T. Sawada, M. Suzuki, et al.,Functional haplotypes of PADI4, encoding citrullinating enzyme peptidy-larginine deiminase 4, are associated with rheumatoid arthritis, Nat. Genet. 34(2003) 395e402.

[10] Y. Takata, H. Inoue, A. Sato, K. Tsugawa, K. Miyatake, D. Hamada, et al.,Replication of reported genetic associations of PADI4, FCRL3, SLC22A4 andRUNX1 genes with rheumatoid arthritis: results of an independent Japanesepopulation and evidence from meta-analysis of East Asian studies, J. Hum.Genet. 53 (2008) 163e173.

[11] C.L. Too, S. Murad, J.S. Dhaliwal, P. Larsson, X. Jiang, B. Ding, et al., Poly-morphisms in peptidylarginine deiminase associate with rheumatoid arthritisin diverse Asian populations: evidence from MyEIRA study and meta-analysis,Arthritis Res. Ther. 14 (2012) R250.

[12] X. Chang, Y. Xia, J. Pan, Q. Meng, Y. Zhao, X. Yan, PADI2 is significantly asso-ciated with rheumatoid arthritis, PLoS One 8 (2013) e81259.

[13] E. Darrah, A. Rosen, J.T. Giles, F. Andrade, Peptidylarginine deiminase 2, 3 and4 have distinct specificities against cellular substrates: novel insights intoautoantigen selection in rheumatoid arthritis, Ann. Rheum. Dis. 71 (2012)92e98.

[14] X. Chang, R. Yamada, A. Suzuki, T. Sawada, S. Yoshino, S. Tokuhiro, et al.,Localization of peptidylarginine deiminase 4 (PADI4) and citrullinated proteinin synovial tissue of rheumatoid arthritis, Rheumatol. Oxf. 44 (2005) 40e50.

[15] E.R. Vossenaar, T.R. Radstake, A. van der Heijden, M.A. van Mansum,C. Dieteren, D.J. de Rooij, et al., Expression and activity of citrullinating pep-tidylarginine deiminase enzymes in monocytes and macrophages, Ann.Rheum. Dis. 63 (2004) 373e381.

[16] C. Foulquier, M. Sebbag, C. Clavel, S. Chapuy-Regaud, R. Al Badine,M.C. Mechin, et al., Peptidyl arginine deiminase type 2 (PAD-2) and PAD-4 butnot PAD-1, PAD-3, and PAD-6 are expressed in rheumatoid arthritis synoviumin close association with tissue inflammation, Arthritis Rheum. 56 (2007)3541e3553.

[17] D. Makrygiannakis, S. Revu, M. Engstrom, E. Af Klint, A.P. Nicholas, G.J. Pruijn,et al., Local administration of glucocorticoids decrease synovial citrullinationin rheumatoid arthritis, Arthritis Res. Ther. 14 (2012) R20.

[18] V.C. Willis, A.M. Gizinski, N.K. Banda, C.P. Causey, B. Knuckley, K.N. Cordova, etal., N-alpha-benzoyl-N5-(2-chloro-1-iminoethyl)-L-ornithine amide, a proteinarginine deiminase inhibitor, reduces the severity of murine collagen-inducedarthritis, J. Immunol. 186 (2011) 4396e4404.

[19] V. Romero, J. Fert-Bober, P.A. Nigrovic, E. Darrah, U.J. Haque, D.M. Lee, et al.,Immune-mediated pore-forming pathways induce cellular hypercitrullinationand generate citrullinated autoantigens in rheumatoid arthritis, Sci. Transl.Med. 5 (2013), 209ra150.

[20] F. Andrade, E. Darrah, M. Gucek, R.N. Cole, A. Rosen, X. Zhu, Autocitrullinationof human peptidyl arginine deiminase type 4 regulates protein citrullinationduring cell activation, Arthritis Rheum. 62 (2010) 1630e1640.

[21] D.M. Lewallen, K.L. Bicker, V. Subramanian, K.W. Clancy, D.J. Slade, J. Martell,et al., Chemical proteomic platform to identify citrullinated proteins, ACSChem. Biol. 10 (2015) 2520e2528.

[22] N.E. Blachere, S. Parveen, M.O. Frank, B.D. Dill, H. Molina, D.E. Orange, Hightiter rheumatoid arthritis antibodies preferentially bind fibrinogen citrulli-nated by peptidyl arginine deiminase 4, Arthritis Rheumatol. (2016).

[23] V. Subramanian, J.S. Knight, S. Parelkar, L. Anguish, S.A. Coonrod, M.J. Kaplan,


et al., Design, synthesis, and biological evaluation of tetrazole analogs of Cl-amidine as protein arginine deiminase inhibitors, J. Med. Chem. 58 (2015)1337e1344.

[24] H.D. Lewis, J. Liddle, J.E. Coote, S.J. Atkinson, M.D. Barker, B.D. Bax, et al., In-hibition of PAD4 activity is sufficient to disrupt mouse and human NET for-mation, Nat. Chem. Biol. 11 (2015) 189e191.

[25] M.A. Shelef, J. Sokolove, L.J. Lahey, C.A. Wagner, E.K. Sackmann, T.F. Warner, etal., Peptidylarginine deiminase 4 contributes to tumor necrosis factor alpha-induced inflammatory arthritis, Arthritis Rheumatol. 66 (2014) 1482e1491.

[26] A. Suzuki, Y. Kochi, H. Shoda, Y. Seri, K. Fujio, T. Sawada, et al., Decreasedseverity of experimental autoimmune arthritis in peptidylarginine deiminasetype 4 knockout mice, BMC Musculoskelet. Disord. 17 (2016) 205.

[27] Y. Seri, H. Shoda, A. Suzuki, I. Matsumoto, T. Sumida, K. Fujio, et al., Pepti-dylarginine deiminase type 4 deficiency reduced arthritis severity in aglucose-6-phosphate isomerase-induced arthritis model, Sci. Rep. 5 (2015)13041.

[28] P. Li, M. Li, M.R. Lindberg, M.J. Kennett, N. Xiong, Y. Wang, PAD4 is essential forantibacterial innate immunity mediated by neutrophil extracellular traps,J. Exp. Med. 207 (2010) 1853e1862.

[29] S. Hemmers, J.R. Teijaro, S. Arandjelovic, K.A. Mowen, PAD4-mediatedneutrophil extracellular trap formation is not required for immunity againstinfluenza infection, PLoS One 6 (2011) e22043.

[30] R. Khandpur, C. Carmona-Rivera, A. Vivekanandan-Giri, A. Gizinski,S. Yalavarthi, J.S. Knight, et al., NETs are a source of citrullinated autoantigensand stimulate inflammatory responses in rheumatoid arthritis, Sci. Transl.Med. 5 (2013), 178ra40.

[31] M. Bawadekar, A. Gendron-Fitzpatrick, R. Rebernick, D. Shim, T.F. Warner,A.P. Nicholas, et al., Tumor necrosis factor alpha, citrullination, and peptidy-larginine deiminase 4 in lung and joint inflammation, Arthritis Res. Ther. 18(2016) 173.

[32] D. Damgaard, L. Senolt, C.H. Nielsen, Increased levels of peptidylargininedeiminase 2 in synovial fluid from anti-CCP-positive rheumatoid arthritispatients: association with disease activity and inflammatory markers, Rheu-matol. Oxf. 55 (2016) 918e927.

[33] J.J. van Beers, A.J. Zendman, R. Raijmakers, J. Stammen-Vogelzangs, G.J. Pruijn,Peptidylarginine deiminase expression and activity in PAD2 knock-out andPAD4-low mice, Biochimie 95 (2013) 299e308.

[34] R. Raijmakers, J. Vogelzangs, J. Raats, M. Panzenbeck, M. Corby, H. Jiang, et al.,Experimental autoimmune encephalomyelitis induction in peptidylargininedeiminase 2 knockout mice, J. Comp. Neurol. 498 (2006) 217e226.

[35] C. Assohou-Luty, R. Raijmakers, W.E. Benckhuijsen, J. Stammen-Vogelzangs,A. de Ru, P.A. van Veelen, et al., The human peptidylarginine deiminases type2 and type 4 have distinct substrate specificities, Biochim. Biophys. Acta 1844(2014) 829e836.

[36] J. Spengler, B. Lugonja, A. Jimmy Ytterberg, R.A. Zubarev, A.J. Creese,M.J. Pearson, et al., Release of active peptidyl arginine deiminases by neu-trophils can explain production of extracellular citrullinated autoantigens inrheumatoid arthritis synovial fluid, Arthritis Rheumatol. 67 (2015)3135e3145.

[37] X. Zhang, M. Bolt, M.J. Guertin, W. Chen, S. Zhang, B.D. Cherrington, et al.,Peptidylarginine deiminase 2-catalyzed histone H3 arginine 26 citrullinationfacilitates estrogen receptor alpha target gene activation, Proc. Natl. Acad. Sci.U. S. A. 109 (2012) 13331e13336.

[38] E. Douni, K. Akassoglou, L. Alexopoulou, S. Georgopoulos, S. Haralambous,S. Hill, et al., Transgenic and knockout analyses of the role of TNF in immuneregulation and disease pathogenesis, J. Inflamm. 47 (1995) 27e38.

[39] D. Damgaard, Y. Palarasah, K. Skjodt, A.I. Catrina, S.M. Hensen, G.J. Pruijn, et al.,Generation of monoclonal antibodies against peptidylarginine deiminase 2(PAD2) and development of a PAD2-specific enzyme-linked immunosorbentassay, J. Immunol. Methods 405 (2014) 15e22.

[40] J.E. Nett, M.T. Cain, K. Crawford, D.R. Andes, Optimizing a Candida biofilmmicrotiter plate model for measurement of antifungal susceptibility bytetrazolium salt assay, J. Clin. Microbiol. 49 (2011) 1426e1433.

[41] C.J. Johnson, J. Cabezas-Olcoz, J.F. Kernien, S.X. Wang, D.J. Beebe,A. Huttenlocher, et al., The extracellular matrix of Candida albicans biofilmsimpairs formation of neutrophil extracellular traps, PLoS Pathog. 12 (2016)e1005884.

[42] A.M. Gillum, E.Y. Tsay, D.R. Kirsch, Isolation of the Candida albicans gene fororotidine-50-phosphate decarboxylase by complementation of S. cerevisiaeura3 and E. coli pyrF mutations, Mol. Gen. Genet. 198 (1984) 179e182.

[43] C.F. Urban, U. Reichard, V. Brinkmann, A. Zychlinsky, Neutrophil extracellulartraps capture and kill Candida albicans yeast and hyphal forms, Cell. Microbiol.8 (2006) 668e676.

[44] C.F. Urban, U. Reichard, V. Brinkmann, A. Zychlinsky, Neutrophil extracellulartraps capture and kill Candida albicans yeast and hyphal forms, Cell Microbiol.8 (2006) 668e676.

[45] M. Shapiro-Shelef, K. Calame, Regulation of plasma-cell development, Nat.Rev. Immunol. 5 (2005) 230e242.

[46] H. Takahara, Y. Oikawa, K. Sugawara, Purification and characterization ofpeptidylarginine deiminase from rabbit skeletal muscle, J. Biochem. 94 (1983)1945e1953.

[47] K.L. Bicker, V. Subramanian, A.A. Chumanevich, L.J. Hofseth, P.R. Thompson,Seeing citrulline: development of a phenylglyoxal-based probe to visualizeprotein citrullination, J. Am. Chem. Soc. 134 (2012) 17015e17018.

[48] A.P. Nicholas, J.N. Whitaker, Preparation of a monoclonal antibody to



citrullinated epitopes: its characterization and some applications to immu-nohistochemistry in human brain, Glia 37 (2002) 328e336.

[49] S.Y. Bang, T.U. Han, C.B. Choi, Y.K. Sung, S.C. Bae, C. Kang, Peptidyl argininedeiminase type IV (PADI4) haplotypes interact with shared epitope regardlessof anti-cyclic citrullinated peptide antibody or erosive joint status in rheu-matoid arthritis: a case control study, Arthritis Res. Ther. 12 (2010) R115.

[50] C.C. Chen, H. Isomoto, Y. Narumi, K. Sato, Y. Oishi, T. Kobayashi, et al., Hap-lotypes of PADI4 susceptible to rheumatoid arthritis are also associated withulcerative colitis in the Japanese population, Clin. Immunol. 126 (2008)165e171.

[51] G.A. Schellekens, B.A. de Jong, F.H. van den Hoogen, L.B. van de Putte, W.J. vanVenrooij, Citrulline is an essential constituent of antigenic determinantsrecognized by rheumatoid arthritis-specific autoantibodies, J. Clin. Invest. 101


(1998) 273e281.[52] K. Nakashima, S. Arai, A. Suzuki, Y. Nariai, T. Urano, M. Nakayama, et al., PAD4

regulates proliferation of multipotent haematopoietic cells by controlling c-myc expression, Nat. Commun. 4 (2013) 1836.

[53] M. Jourdan, M. Cren, N. Robert, K. Bollore, T. Fest, C. Duperray, et al., IL-6supports the generation of human long-lived plasma cells in combinationwith either APRIL or stromal cell-soluble factors, Leukemia 28 (2014)1647e1656.

[54] G. McNee, K.L. Eales, W. Wei, D.S. Williams, A. Barkhuizen, D.B. Bartlett, et al.,Citrullination of histone H3 drives IL-6 production by bone marrow mesen-chymal stem cells in MGUS and multiple myeloma, Leukemia 31 (2) (2017Feb) 373e381.


peptidylarginine deiminase 2 is required for tumor ... · 2.7. immunoﬂuorescence staining and net...

Documents