Lupus (2013) 22, 778–792

http://lup.sagepub.com

PAPER

Pristane-induced lupus as a model of human lupus arthritis:

evolvement of autoantibodies, internal organ and joint

inflammation

H Leiss1, B Niederreiter1, T Bandur1, B Schwarzecker1, S Bluml1, G Steiner1,2, W Ulrich3, JS Smolen1,2 andGH Stummvoll1

1Department of Rheumatology, Medical University of Vienna, Austria; 2Department of Rheumatology, Medical University of Vienna, Austria;and 3Department of Pathology, Hietzing Hospital, Austria

Objective: Arthritis is frequently seen in human lupus, but rarely in lupus models. Pristane-induced lupus (PIL) can be induced in various mouse strains such as BALB/c and C57BL/6.We herein characterize clinical and histological features of arthritis in the context of systemiclupus and provide a prudent comparison with models of rheumatoid arthritis (RA). Methods:

A total of 57 BALB/c mice received pristane (PIL group) and were analyzed for serumautoantibodies (anti-chromatin-, -histone, -Sm, -dsDNA), as well as for clinical featuresand histopathology of joints, lungs and kidneys. Joint pathology was quantified by imageanalysis and tissue cytometry. Ten C57BL/6 mice (Bl/6-PIL) and historical groups of twodifferent RA models were analyzed accordingly. Results: In BALB/c PIL, clinical arthritisstarted at three months, occurred finally in 79% of PIL (but not in controls, p< 0.001) andcorrelated with areas of inflammation, erosion, cartilage damage, osteoclast numbers and totalseverity score (for all: r> 0.7, p< 0.001). After eight months, 58% of PIL (but no controls,p< 0.001) had mild-erosive arthritis. In contrast to RA, the most frequent inflammatory celltype of the pannus was granulocytes (17.7%), PIL had lower numbers of osteoclasts, erosionsrarely affected both layers of the cortical bone and there was no progression to complete jointdestruction (even after one year of observation). Serum autoantibodies (auto-abs) precededarthritis and became significantly elevated in all PIL; affected joints showed increased depositsof IgG (and IgM) within the inflammatory tissue, indicative of an ab-mediated process. PILmice with arthritis also showed signs of pulmonary (100%) and renal (46%) lupus. In contrastto BALB/c, Bl/6-PIL mice did not develop any signs of arthritis. Conclusion: PIL in BALB/cmice is characterized by severe organ involvement, typical autoabs and by a mild-erosivearthritis with similarities to, but also with distinct differences from, RA. PIL may help tostudy arthritis along with other key features of systemic lupus erythematosus after therapeuticinterventions or in knock-out models based on a BALB/c but not on a C57BL/6background. Lupus (2013) 22, 778–792.

Key words: Lupus arthritis; pristane-induced lupus (PIL); systemic lupus erythematosus (SLE)

Introduction

Systemic lupus erythematosus (SLE) is a multi-organ disease characterized by widespread involve-ment of the skin, the inner organs, the hematopoieticand the musculoskeletal systems. Arthralgia isoften reported as a first symptom of SLE and ispresent in up to 95% of patients; the frequency of

joint swelling and thus frank arthritis varies between35% and 88%.1,2 Although musculoskeletalinvolvement in general and ‘‘joint pain’’ in particu-lar constitute a major concern in SLE-patient-basedsurveys,3,4 arthralgia and arthritis are not inthe focus of lupus research and there is little in-depth information on lupus arthropathy in theliterature.

By virtue of the involvement of the small jointsof the hand and the wrist, lupus arthritis clinicallyoften resembles early rheumatoid arthritis (RA),but typical radiographic erosions are seen in only2%–6% of patients; in those who have more appar-ent erosions, the disease is often referred to as

Correspondence to: Georg Stummvoll, Dept. of Rheumatology,

Medical University of Vienna, Waehringer Guertel 18-20, 1090

Vienna, Austria.

Email: georg.stummvoll@meduniwien.ac.at

Received 20 September 2012; accepted 30 April 2013

! The Author(s), 2013. Reprints and permissions: http://www.sagepub.co.uk/journalsPermissions.nav 10.1177/0961203313492869

‘‘rhupus.’’5 Interestingly, magnetic resonance ima-ging (MRI) in patients with lupus arthritis revealsbony erosions in 93% of the examined wrists and in64% of metacarpophalangeal joints (MCPs), yield-ing a similar score as observed in early RApatients.6 In contrast, patients with typical deform-ing but nonerosive Jaccoud’s arthropathy do notshow MRI erosions despite long-standing historyof lupus arthritis.7 In light of these MRI-basedobservations, minimally erosive arthritis may bemuch more frequent in human SLE than hithertosupposed, but in contrast to RA, it does not pro-ceed to frank destruction or ankylosis.

Histologically, SLE is characterized by a mild tomoderate inflammatory synovitis that is similar incharacter but less ‘‘angry’’-appearing than RA.8,9

In older reports with rather low numbers ofpatients, synovial histopathology was described asexhibiting lining layer hyperplasia, microvascularchanges and a perivascular infiltrate with mono-nuclear cells, but at fewer numbers than found inRA.10–12 While lupus arthritis could not clearly bedifferentiated from RA by means of histology anddespite the extensive connective tissue change, littlecartilage and bone destruction seems to occur.13

Although arthritis is frequently observed inhuman SLE, it is rarely observed in murine modelsof SLE: It has been reported in MRL-lpr mice andMRL-gld mice,14–19 but not in other widely usedmodels such as (New Zealand Black (NZB)�NewZealandWhite (NZW)) F1. To our knowledge, pris-tane-induced lupus (PIL) is the only induced andnon-spontanous lupus model with arthritis.

Pristane, a mineral oil (2,6,10,14-tetramethylpen-tadecane, or TMPD), can induce lupus-like diseasein humans20,21 and can also induce systemic lupuswith characteristic organ involvement and autoan-tibodies (auto-abs) in various mouse strains, butwith differences in clinical presentation suggestingthat genetic factors play an important additionalrole.15,22–27 In early reports on pristane-inducedarthritis the systemic, lupus-like nature of pris-tane-induced autoimmune disease has not beencharacterized and PIL arthritis has been consideredtypical for osteoarthritis or, more often, forRA:26–29 BALB/c, DBA-1 and CBA strains havebeen considered susceptible, while (based on clin-ical observations) C57BL/6 have not.15,26,30

Since BALB/c and especially C57BL/6 mice oftenserve as background for genetically altered mousemodels (whose use might allow the development ofvaluable insights into lupus pathogenesis), we hereininvestigated BALB/c and C57BL/6 mice for theoccurrence and detailed clinical and histopatho-logical presentation of arthritis. We applied

procedures originating in experimental RA researchto also observe minor changes in cartilage and bonehistology and tried to highlight some histologicalsimilarities and differences in comparison with twomurine models of RA analyzed in the same way. Wealso embed PIL arthritis in the context of the sys-temic nature of the pristane induced immuneresponse and report on the course of typical auto-abs and the occurrence of major organ involvement.

Methods

Mice and induction of PIL

Female BALB/c mice (Charles River Laboratories)at an average age of 8–10 weeks (six independentexperiments) and female C57BL/6 mice (JacksonLaboratories) were injected intraperitoneally (i.p.)with either 0.5ml of pristane (PIL-group and Bl/6-PIL group, respectively) or saline (controls) andsacrificed after eight months; a limited number ofmice were sacrificed at six or 12 months to allow ahistopathological estimation of the time course.Animal work was performed under the animalprotocol 1919/115-97/98, approved by the animalcare committee (Austrian Ministry of Science andResearch).

Clinical scoring of arthritis

Clinically affected and nonaffected mice could easilybe distinguished.30 Arthritis was scored by a blindedobserver as established for murine RA models.31–34

The assessment was performed every two weeksstarting at two weeks after injection of pristane (orsaline, respectively). Paw swelling as well as gripstrength were assessed by using semiquantitativescores (for paw swelling: 0¼ no swelling, 1¼mildswelling of the toes and ankle, 2¼moderate swellingof the toes and ankle, and 3¼ severe swelling of thetoes and ankle; for grip strength: 3¼ normal gripstrength, 2¼mildly reduced grip strength, 1¼severely reduced grip strength, and 0¼ no grip atall). After acquiring a total score per animal, themean per paw was calculated and used for furtheranalysis.31,32

Detection of auto-antibodies

Anti-histone and anti-chromatin abs were measuredby enzyme-linked immunosorbent assay (ELISA)(Inova Diagnostics, San Diego, CA, USA). Resultsare presented as units/ml (U/ml) employing thereference sera provided by the manufacturer.Autoabs to citrullinated epitopes were determined

Pristane-induced lupus as a model of human lupus arthritisH Leiss et al.

779

Lupus

by a commercial anti-cyclic citrullinated peptide(anti-CCP)3 ELISA (Inova Diagnostics, SanDiego, CA, USA) employing anti-CCP-positivehuman sera as positive controls. Immunoglobulin(Ig)M- and IgG-rheumatoid factor (RF) wasdetected by ELISA (Hycor Biomedical Inc,Garden Grove, CA, USA). Horseradish peroxi-dase-conjugated goat-anti-rat IgG and IgM ab(Southern Biotech; 1:2.000) served as secondaryabs in all ELISAs. A positive result was defined asa level exceeding the mean plus 2 standarddeviations of control BALB/c mice. Antinuclearantibodies (ANA) were determined by immuno-fluorescence (IF) on Hep-2 cells as described,35 andanti-Sm antibodies by immunoblotting.36 Anti-double-stranded DNA (anti-dsDNA) abs weredetected by IF on Crithidia luciliae as described.37,38

Histology of joints, lungs and kidneys

In each mouse, the hind paw showing more pro-nounced paw swelling was prepared and analyzedby histopathological techniques as previouslydescribed.39 Staining with hematoxylin and eosin(HE) allowed a general assessment of inflamma-tion and structural damage; toluidine blue (TB)destaining was used to determine cartilage matrixloss; and tartrate-resistant acid phosphatase(TRAP)-staining identified osteoclasts and osteo-clast precursors.40

Additional immunohistochemistry using absagainst macrophages (clone F4/80, Serotec), gran-ulocytes (G1, Frima), T- (anti-CD3, Novo CastraLaboratories, Newcastle, UK) and B-Cells (anti-CD45 receptor, BD Biosciences PharMingen, SanDiego, CA, USA) was performed in a limitednumber of slides;40–42 quantitative analysis of theinflammatory cellular infiltrate was performed bytissue cytometry using HistoQuestTM

(TissueGnostics, Vienna, Austria).43,44

Staining for IgG and IgM was performed withbiotinylated goat anti-mouse abs (Vector,Burlingame, CA, USA).

Lungs, kidneys, brains and spleens wereobtained from PIL and healthy controls (HC), pro-cessed according to standard laboratory proceduresand stained with HE and periodic acid-Schiff (PAS;for kidneys), respectively. Sections were analyzedby a blinded pathologist (WU).

Histomorphometry

The quantification of histomorphometrical param-eters was performed with a bone-specific imageanalysis system (OsteoMeasureTM, OsteoMetrics,Decatur, GA, USA) to determine the

histomorphometry of hind paw specimens to calcu-late areas of (i) inflammation, (ii) erosion, (iii) car-tilage destruction (indicated by loss of proteoglycangiven as a ratio of destained cartilage/total cartil-age) and (iv) the number of osteoclasts as an add-itional marker for the extent of the destructiveprocess.40–42

Arthritis severity score (ASS)

To express the severity of joint involvement by asingle value (which would also allow for a prudentcomparison with other murine arthritis models),we scored the histomorphometric features of arth-ritis as obtained by histomorphometry and calcu-lated the ASS based on its results: We assignedone to three points relating to the size of theinflammatory area (�0.1, �0.5, �1mm2, respect-ively), one to three points relating to the size oferosions (�0.01, �0.05, �0.1mm2, respectively)and one to three points according to the numberof osteoclasts (1, 2–5, �5, respectively). Sinceminor cartilage damage was also observed inHC, only one point was assigned for damageratios greater than the mean of HC� 2 standarddeviations (SD) (i.e. >0.16).

To validate this scoring system, HE stainedhistological specimens were evaluated for the pres-ence of histologic arthritis by a blinded expert(BN), who found all mice with a severity score �2to have arthritis, but all mice with an ASS of 1 asnegative (not shown). Mice with a score of 1 hadeither isolated cartilage degradation withoutinflammatory synovitis or showed a single osteo-clast, but no inflammatory infiltrate or erosion.Consequently, we defined an ASS score �2 as indi-cative of arthritis. The maximum ASS of 9 indicatesmassive erosive arthritis.

We also compared our scoring system in PILwith two historical cohorts from establishedmodels of RA: In the spontaneous model ofhuman tumor necrosis factor (TNF)-transgenicmice (hTNFtg) with 100% penetrance of arthritis,all affected mice scored positive (ASS �2), whilewild-type (wt) controls did not.42 Similarly, allaffected mice with collagen-induced arthritis(CIA)32,45 scored �2, while the unaffected animalsor the controls were negative (not shown).

Statistical analysis

All group results are expressed as mean� SD, if notstated otherwise. Student’s t test or Fisher’s exacttest (two tailed) were used for the comparison ofgroup values and discriminatory parameters, whereappropriate. For comparing group values that did

Pristane-induced lupus as a model of human lupus arthritisH Leiss et al.

780

Lupus

not follow Gaussian distribution, Mann-Whitney’stest (two tailed) was used. Welch’s correction wasapplied if variances were significantly different.Pearson and Spearman correlation coefficientswere calculated for variables following or not-following Gaussian distributions, respectively.P values less than 0.05 were considered significant.All statistical analyses were performed using PASWStatistics 18.0 (version 18.0.3, 09.09.2010) or Prism5 for Mac OS X (version 5.0a, December 24, 2007)software.

Results

Paw swelling and loss of grip strength in BALB/c

Clinical evidence of arthritis, paw swelling and lossof grip strength occurred after pristane injection in

BALB/c but not in C57BL/6 mice. The first signs ofarthritis were seen at three months after pristaneinjection (17%), with a predilection for the hindpaws (Figure 1(a), (b)). The frequency of arthritisincreased over time: After six months, more than50% were clinically affected, after eight months,79% of PIL mice (45 out of 57) had at least oneclinically detectable episode of arthritis, whilesaline-injected controls stayed free from arthritis(Figure 1(a), (b)).

Time course of autoabs in BALB/c PIL

Serum levels of anti-histone and anti-chromatin abswere measured for detecting early autoab produc-tion, since anti-dsDNA abs are known to becomepositive as late as seven months after pristane injec-tion.46,47 Within the first month, both anti-histoneand anti-chromatin abs were found in >20% of

Figure 1 Time course of clinical characteristics and levels of auto-antibodies in BALB/c mice.Paw swelling and loss of grip strengths ((a), (b), respectively) became detectable in PIL mice (black diamonds) three months afterinjection of pristane at the earliest. After six months, however, more than half of PIL mice showed clinical signs of arthritis, whileHC were not affected (white diamonds).Auto-antibodies (auto-abs) production preceded the clinical onset of arthritis. Anti-chromatin (c) as well as anti-histone abs (d)were detectable as soon as one month after pristane injection and became significantly elevated a month later, leading to positiveresults in more than 50% of PIL mice.*: Denotes a p> 0.05. Black diamonds: PIL; white diamonds: HC. One positive ab levels were defined as >meanþ 2SD ofrespective controls. PIL: pristane-induced lupus; HC: healthy controls; SD: standard deviation.

Pristane-induced lupus as a model of human lupus arthritisH Leiss et al.

781

Lupus

PIL mice and after two months more than 50% of

PIL mice tested positive, leading to significant ele-

vations in mean serum levels compared to HC

(Figure 1(c), (d)). Thus, the increase in these

serum auto-abs preceded the first clinical signs of

arthritis by one to two months (see above). After

eight months, 100% of PIL mice were positive for

anti-chromatin, 93% were positive for anti-histone

abs (Figure 1(c), (d), Appendix 1(a)).In comparison to anti-chromatin and anti-his-

tone abs, anti-Sm abs occurred later in the courseof PIL, yielding first positive results after threemonths; after eight months, however, 80% of PILmice tested positive (Appendix 1(a)). Interestingly,86% of PIL mice with histologically proven arth-ritis were positive for all three types of abs, whilewe found this combination in only 50% of nonaf-fected PIL mice (not shown).

In line with the literature, we also observed aslow increase in anti-histone and anti-chromatinabs in the PBS-injected control group, but theirlevels never reached those of PIL mice (Figure1(c), (d));46 none of the controls ever developedanti-Sm abs.

In a limited number of BALB/c mice, we alsotested for ANA and anti-dsDNA abs by IF eightmonths after induction of PIL: All PIL mice werepositive for ANA, but at low titers (1:40); only 24%had positive results at either 1:80 or 1:160 dilutions.ANA showed a homogenous (60%) or speckledpattern (40%); all mice positive for ANA by IFalso had increased levels of anti-chromatin absand anti-histone abs compared to controls (notshown). In line with the literature, we found anti-dsDNA abs in 47% of PIL sera, with titers rangingfrom 1:20 to 1:160, while HC had either no (80%)or low (1:20, 20%) titers of dsDNA abs (Appendix1(a), (b)).15

BALB/c PIL mice first showed increased levels ofrheumatoid factor (RF) (compared to age-matchedcontrols) after two months (67� 21 vs. 41� 7U/ml,p¼ 0.04), but, in the course of eight months, RFserum levels significantly increased in PIL as well asHC compared to baseline levels, leading to a loss ofsignificance when analyzed after eight months(145� 37 vs. 141� 34, p¼ n.s.). Thus, in line withthe literature, RF levels appear to be triggered bypristane injection and are elevated in early PIL, butalso constitute a feature of senescing BALB/c mice(Appendix 1(c)). In addition, anti-CCP abs devel-oped in 100% of PIL (593� 380U/ml) after eightmonths, but in only 25% of HC (84� 51U/ml,p¼ 0.002) (Appendix 1(a)).

Erosive arthritis in BALB/c PIL

In the histological analysis, inflammatory changeswere detected in the BALB/c PIL group, but not inHC or in pristane-injected C57BL/6 mice (Bl/6PIL). PIL arthritis was characterized by cartilagedamage (loss of proteoglycans) and cell-rich inflam-matory infiltrates. Cortical erosions with TRAP-positive osteoclasts were found in all affected ani-mals, but in most cases (88%), the inner corticallayer stayed intact (Figure 2(a), (c), (e)).

Quantitative analysis of immunohistochemistrystaining by tissue cytometry showed that theinflammatory pannus consisted mainly offibroblasts, but also of inflammatory cells with apredominance of granulocytes (17.7� 8.7%) andonly scarcely scattered macrophages (1.9� 0.8%),T- (1.7� 0.7%), and B-cells (0.3� 0.1%)(Figure 3).

Histomorphometric analysis and correlation withclinical variables

Histomorphometric analysis of the paws showedevidence of inflammatory synovitis in 58% ofBALB/c PIL mice (n¼ 57), but not in controls(n¼ 35); cartilage destaining was seen in 65% ofthe PIL mice, but only one healthy mouse hadsigns of minor cartilage degradation (p< 0.0001).A subgroup of PIL developed clinical signs of arth-ritis, but showed only minor (or no) histologicalinflammatory infiltrates or erosions (Table 1). Incontrast to previous reports, we did not identifyanimals with synovial hyperplasia without clinicalsigns of arthritis.29,48

In PIL, the two assessed clinical features of arth-ritis (paw swelling and loss of grip strength, respect-ively) correlated well with each other (r¼ 0.949,p< 0.0001) and also correlated with histologicalresults obtained by histomorphometry (inflamma-tory area, cartilage destaining, erosive area, numberof osteoclasts) and the severity score ASS (for all:r> 0.7, p< 0.0001; Figure 4).

Frequency and time course of arthritis in BALB/cPIL

When analyzed after six months, 50% of PIL micehad histologically proven arthritis; after eightmonths the percentage increased to 58% and fur-ther to 75% after 12 months, while no controlmouse was affected. Although the frequency ofarthritic mice increased the ASS did not deteriorateover time (8.3� 0.9 after six months, vs. 7.0� 1.8after eight months, vs. 8.0� 0.8 after 12 months, allp¼ n.s.). Also, the mean eroded area did not

Pristane-induced lupus as a model of human lupus arthritisH Leiss et al.

782

Lupus

Figure 2 PIL leads to erosive arthritis.HE staining illustrates the erosive nature of PIL arthritis in a mouse with typical disease severity as found after eight months(arthritis severity score (ASS) of 7) and allows for an overview of arthritis in PIL mice (a) and controls (b). Cartilage damagedefined as loss of proteoglycans frequently occurred in PIL. Typical areas of destaining are shown in (c) as opposed to a healthycontrol mouse in (d). Although the frequency of PIL arthritis increased over time, its severity remained stable even after 12 monthsof observation and did not progress to highly destructive, erosive stages (12 months old PIL in (e), RA models of hTNF and CIAmice after 4 months in (g) and (f); pictures of RA courtesy of Dr. S. Hayer).PIL: pristane-induced lupus; HE: hematoxylin and eosin; RA: rheumatoid arthritis; hTNFtg: human tumor necrosis factor (TNF)-transgenic; CIA: collagen-induced arthritis.

Pristane-induced lupus as a model of human lupus arthritisH Leiss et al.

783

Lupus

change and yielded similar results when analyzedafter six, eight and 12 months, respectively(0.12� 0.07 (six months), 0.11� 0.13 (eightmonths) and 0.09� 0.02mm2 (12 months),p¼ n.s.).

Disease severity in PIL arthritis as opposed to RAmodels

In our laboratory, different murine models of arth-ritis are analyzed by the same standardized meth-odology, which allows for a prudent comparison ofPIL arthritis with other forms of experimental arth-ritis. Compared to the spontaneous disease inhTNFtg mice or the induced model of CIA withclinical onsets at five and three weeks, respectively,PIL arthritis develops more slowly and shows firstclinical signs of arthritis as late as three monthsafter induction.31,45

For comparison of histomorphometry, wematched two historical groups of hTNFtg and

CIA mice with PIL groups based on disease sever-ity as indicated by the ASS. Comparing PIL tosimilarly affected hTNFtg mice (ASS 6.6� 1.4and 6.6� 0.5, respectively), PIL had larger areasof inflammation (0.77� 0.24 vs. 0.47� 0.39mm2,p¼ 0.043), slightly smaller areas of erosion(0.07� 0.11 vs. 0.08� 0.14mm2, p¼ n.s.) and sig-nificantly lower numbers of osteoclasts (7.9� 2.6vs. 19.7� 18.5, p< 0.04).

Comparing PIL to CIA (ASS 8.0� 0 and8.0� 1.2, respectively), PIL had, again, slightlylarger inflammatory areas (1.19� 0.28 vs.0.87� 0.50mm2, p¼ n.s.), smaller eroded areas(0.08� 0.02 vs. 0.48� 0.32mm2, p¼ 0.003) andlower numbers of osteoclasts (10.4� 4.5 vs.39.8� 3.2, p¼ 0.02).

In line with these observations, erosions in bothRA models frequently reached the bone marrowcavity by destroying both layers of the corticalbone. Analyzing ASS-matched groups, we found67% complete erosions in hTNFtg vs. 8% in PIL

Figure 3 Composition of the inflammatory infiltrate.Granulocytes were the dominant cell type of the inflammatory infiltrate (Gr-1 staining in brown, (a)). Quantitative analysis wasperformed by tissue cytometry, where positive cells could be identified (shown for granulocytes in (b), red circles), separated fromother cells ((b), green circles) and further analyzed (c). Besides granulocytes, we found low numbers of T and B-lymphocytes andmacrophages (d).

Pristane-induced lupus as a model of human lupus arthritisH Leiss et al.

784

Lupus

(p< 0.0001) and 80% complete erosions in CIA vs.13% in matched PIL (p< 0.0001). In contrast toerosive RA in humans and mice, in which distinctsubcortical bone marrow aggregates (with a pre-dominance of B cells) were described as a reactionto synovial invasion, such a subcortical accumula-tion of immune cells was not observed in PIL (anexample is given in Appendix 2).32,34,45,49,50

As mentioned, the frequency of arthritic mice inPIL increased over the course of several months,but disease severity remained rather stable, evenin mice analyzed after 12 months. In contrast,peak frequencies of arthritis in hTNFtg arthritisand CIA occur after a few weeks and disease ultim-ately progresses to complete joint destructionwithin six months or less.

Synovial accumulation of IgG and IgM

IgG (and—to a lesser extent—IgM, not shown)were found within the inflammatory pannus tissuein PIL arthritis. IgG accumulation appeared toincrease with disease severity and a blinded obser-ver (GHS) could distinguish a group of moderately(ASS 6) from highly affected mice (ASS 9), whilesevere arthritis in hTNFtg or CIA mice (for all ASS9) was not featured by IgG (or IgM) within theinflammatory pannus (Figure 5(a)–(d)).

Pneumonitis and nephritis as indicators of systemiclupus

BALB/c mice developed pneumonitis characterizedby perivascular inflammatory infiltrates with hemo-siderin-laden macrophages, similar to the observa-tions in C57BL/10 mice.51,52 All mice suffering

from PIL arthritis had pulmonary involvementand 45% of them also had glomerulonephritis.Interestingly, PIL mice without arthritis showed atendency to less frequent and less severe pulmonaryand renal involvement, while controls were free ofany organ damage (Figure 5 (e)–(h), Table 2). PILspleens showed marked germinal center hyperplasia(not shown), while there were no signs of inflam-mation in PIL brains (not shown).

No arthritis in C57BL/6 PIL

As did BALB/c, C57BL/6 mice showed inflamma-tory pulmonary involvement upon pristane injec-tion (not shown), but—in contrast to BALB/c—did not develop clinical signs of arthritis, nordid we observe increased cartilage degradation orsynovial reaction as possible minor signs of arthritisin the histological analysis.

Discussion

PIL is a model for systemic lupus characterized bytypical auto-abs, organ involvement and also byarthritis.

We found elevated anti-nucleosome-abs in allpristane-injected mice and evidence of pulmonalinvolvement in 94%, and we herein show that arth-ritis in PIL is frequent, with 81% of mice experien-cing paw swelling and 58% showing clearhistological arthritis.

Interestingly, despite its high frequency inhuman SLE, only the MRL/lpr (and MRL/gld)mice lupus models are characterized by arthritis,

Table 1 In-depth analysis of arthritis in PIL in BALB/c mice

PIL AllPIL Provenarthritis

PIL Only clinicalarthritis HC pb PIL vs. HC

Number (n¼) 57 33 13 35 —

% of PIL 100% 58% 23% n.a. —

Clinics (mean� SD)

Paw swelling 0.87� 0.85 1.36� 0.78 0.36� 0.31 0� 0 <0.001

Grip strength 2.44� 0.61 2.13� 0.6 2.83� 0.15 3� 0 <0.001

Histology (mean�SD)

Severity (ASS) 4.23� 3.63 7.06� 1.78 0.38� 0.51 0� 0 <0.001

Cartilage degradationa (ratio) 0.35� 0.28 0.51� 0.25 0.12� 0.14 0.06� 0.05 <0.001

Inflammatory areaa (mm2) 0.55� 0.62 0.96� 0.52 0� 0 0� 0 <0.001

Area of erosiona (mm2) 0.05� 0.06 0.08� 0.07 0� 0 0� 0 <0.001

Osteoclastsa (number) 4.81� 4.97 8.21� 3.82 0.23� 0.6 0� 0 <0.001

Most animals in the PIL group developed arthritis leading to significant differences in clinical features and OsteoMeasure results when compared to

controls. A subgroup of PIL mice developed clinical signs of arthritis, but showed only minor (or no) changes in histology without evidence of

inflammatory infiltrates or erosions. a: Values obtained by OsteoMeasure analysis. b: p denotes comparison of all PIL mice vs. controls. ASS:

arthritis severity score (see Methods section); PIL: pristane-induced lupus; HC: healthy controls; SD: standard deviation.

Pristane-induced lupus as a model of human lupus arthritisH Leiss et al.

785

Lupus

Figure 4 Clinical and histological features correlate well in PIL arthritis.In PIL, clinical features of arthritis (paw swelling and loss of grip strength) correlated well with histological results gained byOsteoMeasure: inflammatory area (a), erosive area (b), cartilage destaining (c) and arthritis severity score (d).PIL: pristane-induced lupus; ASS: arthritis severity score.

Pristane-induced lupus as a model of human lupus arthritisH Leiss et al.

786

Lupus

Figure 5 IgG, IgM in joints, organ involvement.Indicative of an antibody-mediated disease, there was accumulation of IgG (and to a lesser degree, IgM) within the inflammatorytissue of affected PIL joints (a), but not in similarly affected RA models (hTNFtg, (c) and CIA, (d), respectively or healthy controls(b). All mice suffering from PIL arthritis also had pulmonary involvement and 45% of them had nephritis. Pneumonitis wascharacterized by perivascular inflammatory infiltrates (black arrow) and hemosiderin-laden macrophages (white arrow) (e). PILnephritis presented as either mesangial or proliferative glomerulonephritis, comparable with WHO class II and WHO class III/IVlupus nephritis in humans. An example of focal-proliferative nephritis is given in (b). Interestingly, mice without arthritis also had atendency to less frequent and less severe pulmonary and renal involvement, while healthy controls were free of any organ damage((f), (h)).Ig: immunoglobulin; PIL: pristane-induced lupus; RA: rheumatoid arthritis; hTNFtg: human tumor necrosis factor (TNF)-trans-genic; CIA: collagen-induced arthritis; WHO: World Health Organization.

Pristane-induced lupus as a model of human lupus arthritisH Leiss et al.

787

Lupus

but even in MRL/lpr, the incidence of arthritis israther low at 15%–25%.53

Although a prominent and often first sign ofhuman SLE, lupus arthritis is not in the focus ofresearch and is still far from being well defined. Inconventional X-rays, erosions are rather rare, butminor erosions are frequently observed in MRI.6

As a consequence of the milder course of arthritisin lupus, joint replacement has been less often per-formed (compared to RA) and, therefore, histo-logical studies had to be performed with a ratherlimited number of patients. Nevertheless, all evi-dence describes arthritis in lupus as similar to butless aggressive than RA.8,9

As in MRL/lpr, arthritis in PIL is erosive, whichalso may constitute an inherent feature of murinedisease. Compared to murine RA, however, PILarthritis appears less aggressive as indicated bylower numbers of osteoclasts and a lower frequencyof complete cortical destruction, by the lack of sub-cortical inflammatory aggregates in the paracorticalbone marrow cavity and by the lack of progressionto complete joint destruction or ankylosis, evenafter one year of observation.32,49,50 As a furtherdifference from RA, tissue cytometry showed thatthe inflammatory synovial infiltrate was dominatedby granulocytes, while there were only a few macro-phages, T and B cells; this contrasts with arthritis inhTNFtg and CIA mice as well as in human RA, inwhich macrophages prevail.32,49,54,55 Despite allinherent limitations of such comparisons betweendifferent mouse models, these findings indicate thatPIL arthritis has—despite all similarities—at leastsome distinct histological differences comparedwith typical RA models.

The pathogenesis of PIL is not completely clear,but B-cell hyperreactivity and the occurrence oftypical (pathogenic) auto-abs appear to playmajor roles.15,56 PIL is associated with hypergam-maglobulinemia57 and PIL spleens show germinalcenter hyperplasia. Abs in PIL have pathogenicpotential and promote nephritis,58,59 and there isa pathogenic role of B cells in development of dif-fuse alveolar hemorrhage induced by pristane.52 Asin human lupus, there is a need for CD4 T cell help,since pristane-induced arthritis was reduced anddelayed by anti-CD4, but not by anti-CD8 abs.60

Locally, TNFa, interleukin (IL)1b and IL-6 havebeen reported as elevated in affected joints (DBA/1mice)28 and there is IL-6 dependency for the pro-duction of anti-dsDNA abs, but not for e.g. anti-Sm abs (BALB/c),61 while interferon (IFN)g isrequired for lupus nephritis (BALB/c).62 Sincethere was no pristane found directly in affectedPIL joints, an indirect pathogenesis is likely.48

Fitting with the hypothesis of an antibody/immune complex-mediated disease, we found IgG(and IgM) accumulation in synovitic inflammatoryinfiltrates in PIL, but not in similarly affectedhTNFtg or CIA mice.

PIL mice developed auto-abs typical for SLE,such as anti-chromatin and anti-histone abs,which became detectable in all and clearly positivein more than 90% of PIL mice; 80% of PIL (but nocontrol animals) were also positive for the highlySLE-specific anti-Sm abs and 47% of PIL devel-oped anti-dsDNA abs (Appendix 1(a)).

While anti-chromatin as well as anti-histone absoccurred as early as one month after pristane injec-tion and were significantly elevated in 50% of PILmice after two months, clinical signs of arthritisoccurred later and could be detected after threemonths at the earliest. Thus, our findings fit thetheory that anti-nuclear antibodies precede the clin-ical onset of lupus as shown in human SLE.63

Interestingly, all PIL, but also all controlsbecame positive for RF, but with an earlier onsetfor PIL. This increase in RF might thus have beentriggered by pristane injection, but also reflects acommon feature observed in senescing but other-wise healthy BALB/c mice (Appendix 1(c)). Boththe occurrence of—mostly low-titer—RF (in SLEpatients up to 61%) as well as an increase in vari-ous autoabs in the elderly is also seen inhumans.64–66 In line with the literature, healthymice also developed low-titer anti-histone andanti-chromatin abs over time, but at significantlylower titers than PIL.61 In contrast, anti-CCP absand anti-Sm abs predominantly occurred in PIL,but not in controls.

Table 2 Major organ involvement in PIL

PneumonitisFrequency

NephritisFrequency Histology

PIL Class II 24%

Proven arthritisa 100% 55% Class III 18%Class IV 12%

PIL Class II 17%

Only clinical arthritisb 100% 50% Class III 33%Class IV 0%

PIL

No arthritis 70% 20% class II 10%class III 10%class IV 0%

HC 0% 0%

a: Positive histology and clinical evidence of arthritis, b: Only clinical

arthritis without inflammation in histology. Renal histology was clas-

sified by mesangial (class II) and proliferative (class III and IV)

involvement according to the World Health Organization (WHO) cri-

teria of human lupus nephritis. PIL: pristane-induced lupus; HC:

healthy controls.

Pristane-induced lupus as a model of human lupus arthritisH Leiss et al.

788

Lupus

Few SLE models are characterized by arthritis,and PIL represents—to our knowledge—the onlyexample of inducible SLE-associated arthritis inpreviously healthy mice. PIL can be regarded as asuitable tool to follow the course of arthritis in thecontext of systemic lupus and the possible influenceof various therapeutic approaches in vivo.28 In con-trast to e.g. the Fas-defective arthritis-prone MRL/lpr or MRL/gld strains, PIL is an induced model,which may also allow for studying lupus-associatedarthritis in certain knock-out or transgenic mousemodels based on a BALB/c but not on a C57BL/6background. Because of its rather high (possiblystill underestimated) frequency in human SLE,arthritis may deserve more attention—and PILoffers an in vivo model to study it.

Acknowledgments

We thank Dr Silvia Hayer for fruitful discussionsand forthcoming support. We also thank her andDr Esther Jimenez-Boj for the permission to showthe histology specimens depicted in Figure 2(e)and (f) and Appendix 2.

Funding

The research leading to these results has receivedfinancial support from St. Anna Research Institute,Vienna (Grant for Research in Paediatric andInternal Rheumatology 2010) and the InnovativeMedicines Initiative Joint Undertaking undergrant agreement number 115142, resources ofwhich are composed of financial contributionfrom the European Union’s Seventh FrameworkProgramme and European Federation ofPharmaceutical Industries and Associations(EFPIA) companies in-kind contribution.

Conflict of interest statement

The authors have no conflicts of interest to declare.

References

1 Dubois EL, Tuffanelli DL. Clinical manifestations of systemic lupuserythematosus. Computer analysis of 520 cases. JAMA 1964; 190:104–111.

2 Grigor R, Edmonds J, Lewkonia R, Bresnihan B, Hughes GR.Systemic lupus erythematosus. A prospective analysis. Ann RheumDis 1978; 37: 121–128.

3 Stamm TA, Bauernfeind B, Coenen M, et al. Concepts importantto persons with systemic lupus erythematosus and their coverageby standard measures of disease activity and health status. ArthritisRheum 2007; 57: 1287–1295.

4 Bauernfeind B, Aringer M, Prodinger B, et al. Identification ofrelevant concepts of functioning in daily life in people with sys-temic lupus erythematosus: A patient Delphi exercise. ArthritisRheum 2009; 61: 21–28.

5 van Vugt RM, Derksen RH, Kater L, Bijlsma JW. Deformingarthropathy or lupus and rhupus hands in systemic lupus erythe-matosus. Ann Rheum Dis 1998; 57: 540–544.

6 Boutry N, Hachulla E, Flipo RM, Cortet B, Cotten A. MR ima-ging findings in hands in early rheumatoid arthritis: comparisonwith those in systemic lupus erythematosus and primary Sjogrensyndrome. Radiology 2005; 236: 593–600.

7 Ostendorf B, Scherer A, Specker C, Modder U, Schneider M.Jaccoud’s arthropathy in systemic lupus erythematosus:Differentiation of deforming and erosive patterns by magnetic res-onance imaging. Arthritis Rheum 2003; 48: 157–165.

8 Armas-Cruz R, Harnecker J, Ducach G, Jalil J, Gonzalez F.Clinical diagnosis of systemic lupus erythematosus. Am J Med1958; 25: 409–419.

9 Cruickshank B. Lesions of joint and tendon sheaths in systemiclupus erythematosus. Ann Rheum Dis 1959; 18: 111–119.

10 Goldenberg DL, Cohen AS. Synovial membrane histopathology inthe differential diagnosis of rheumatoid arthritis, gout, pseudog-out, systemic lupus erythematosus, infectious arthritis and degen-erative joint disease. Medicine (Baltimore) 1978; 57: 239–252.

11 Labowitz R, Schumacher Jr HR. Articular manifestations of sys-temic lupus erythematosus. Ann Intern Med 1971; 74: 911–921.

12 Natour J, Montezzo LC, Moura LA, Atra E. A study of synovialmembrane of patients with systemic lupus erythematosus (SLE).Clin Exp Rheumatol 1991; 9: 221–225.

13 Wallace D. The musculoskeletal system. In: Wallace D (ed.),Dubois’ Lupus Erythematosus, 5th ed. Baltimore, MD: Williams& Wilkins, 1997. p. 635–651.

14 Ito MR, Terasaki S, Itoh J, Katoh H, Yonehara S, Nose M.Rheumatic diseases in an MRL strain of mice with a deficit inthe functional Fas ligand. Arthritis Rheum 1997; 40: 1054–1063.

15 Reeves WH, Lee PY, Weinstein JS, Satoh M, Lu L. Induction ofautoimmunity by pristane and other naturally occurring hydrocar-bons. Trends Immunol 2009; 30: 455–464.

16 Furukawa F, Tanaka H, Sekita K, Nakamura T, Horiguchi Y,Hamashima Y. Dermatopathological studies on skin lesions ofMRL mice. Arch Dermatol Res 1984; 276: 186–194.

17 Furukawa F, Yoshimasu T. Animal models of spontaneous anddrug-induced cutaneous lupus erythematosus. Autoimmun Rev2005; 4: 345–350.

18 Hang L, Theofilopoulos AN, Dixon FJ. A spontaneous rheuma-toid arthritis-like disease in MRL/l mice. J Exp Med 1982; 155:1690–1701.

19 Koopman WJ, Gay S. The MRL-lpr/lpr mouse. A model for thestudy of rheumatoid arthritis. Scand J Rheumatol Suppl 1988; 75:284–289.

20 Dahlgren J, Takhar H, Anderson-Mahoney P, Kotlerman J, TarrJ, Warshaw R. Cluster of systemic lupus erythematosus (SLE)associated with an oil field waste site: A cross sectional study.Environ Health 2007; 6: 8.

21 Mayes MD. Epidemiologic studies of environmental agents andsystemic autoimmune diseases. Environ Health Perspect 1999;107(Suppl 5): 743–748.

22 Satoh M, Kuroda Y, Yoshida H, et al. Induction of lupus auto-antibodies by adjuvants. J Autoimmun 2003; 21: 1–9.

23 Kuroda Y, Nacionales DC, Akaogi J, Reeves WH, Satoh M.Autoimmunity induced by adjuvant hydrocarbon oil componentsof vaccine. Biomed Pharmacother 2004; 58: 325–337.

24 Wooley PH, Sud S, Whalen JD, Nasser S. Pristane-induced arth-ritis in mice. V. Susceptibility to pristane-induced arthritis is deter-mined by the genetic regulation of the T cell repertoire. ArthritisRheum 1998; 41: 2022–2031.

25 Chapdelaine JM, Whalen JD, Wooley PH. Pristane induced arth-ritis. II. Genetic regulation in F1 hybrid mice and cellular immune

Pristane-induced lupus as a model of human lupus arthritisH Leiss et al.

789

Lupus

abnormalities following pristane injection. Autoimmunity 1991; 8:215–220.

26 Wooley PH, Seibold JR, Whalen JD, Chapdelaine JM. Pristane-induced arthritis. The immunologic and genetic features of anexperimental murine model of autoimmune disease. ArthritisRheum 1989; 32: 1022–1030.

27 Potter M, Wax JS. Genetics of susceptibility to pristane-inducedplasmacytomas in BALB/cAn: Reduced susceptibility in BALB/cJwith a brief description of pristane-induced arthritis. J Immunol1981; 127: 1591–1595.

28 Patten C, Bush K, Rioja I, et al. Characterization of pristane-induced arthritis, a murine model of chronic disease: Response toantirheumatic agents, expression of joint cytokines, and immuno-pathology. Arthritis Rheum 2004; 50: 3334–3345.

29 Hopkins SJ, Freemont AJ, Jayson MI. Pristane-induced arthritis inBalb/c mice. I. Clinical and histological features of the arthropa-thy. Rheumatol Int 1984; 5: 21–28.

30 Bedwell AE, Elson CJ, Hinton CE. Immunological involvement inthe pathogenesis of pristane-induced arthritis. Scand J Immunol1987; 25: 393–398.

31 Hayer S, Redlich K, Korb A, Hermann S, Smolen J, Schett G.Tenosynovitis and osteoclast formation as the initial preclinicalchanges in a murine model of inflammatory arthritis. ArthritisRheum 2007; 56: 79–88.

32 Bluml S, Binder NB, Niederreiter B, et al. Antiinflammatoryeffects of tumor necrosis factor on hematopoietic cells in amurine model of erosive arthritis. Arthritis Rheum 2010; 62:1608–1619.

33 Zwerina K, Koenders M, Hueber A, et al. Anti IL-17A therapyinhibits bone loss in TNF-alpha-mediated murine arthritisby modulation of the T-cell balance. Eur J Immunol 2012; 42:413–423.

34 Redlich K, Hayer S, Ricci R, et al. Osteoclasts are essential forTNF-alpha-mediated joint destruction. J Clin Invest 2002; 110:1419–1427.

35 Paisansinsup T, Deshmukh US, Chowdhary VR, Luthra HS,Fu SM, David CS. HLA class II influences the immuneresponse and antibody diversification to Ro60/Sjogren’s syn-drome-A: Heightened antibody responses and epitope spreadingin mice expressing HLA-DR molecules. J Immunol 2002; 168:5876–5884.

36 Dumortier H, Monneaux F, Jahn-Schmid B, et al. B and T cellresponses to the spliceosomal heterogeneous nuclear ribonucleo-proteins A2 and B1 in normal and lupus mice. J Immunol 2000;165: 2297–2305.

37 Yang JY, Oh EJ, Kim Y, Park YJ. Evaluation of anti-dsDNAantibody tests: Crithidia luciliae immunofluorescence test, immu-noblot, enzyme-linked immunosorbent assay, chemiluminescenceimmunoassay [in Korean]. Korean J Lab Med 2010; 30:675–684.

38 Melegari A, Bonaguri C, Russo A, Luisita B, Trenti T, Lippi G. Acomparative study on the reliability of an automated system for theevaluation of cell-based indirect immunofluorescence. AutoimmunRev 2012; 11: 713–716.

39 Hayer S, Redlich K, Korb A, Hermann S, Smolen J, Schett G.Tenosynovitis and osteoclast formation as the initial preclinicalchanges in a murine model of inflammatory arthritis. ArthritisRheum 2007; 56: 79–88.

40 Schett G, Tuerk B. Bone histomorphometry in arthritis models.Methods Mol Med 2007; 135: 269–283.

41 Parfitt AM, Drezner MK, Glorieux FH, et al. Bone histomorpho-metry: Standardization of nomenclature, symbols, and units.Report of the ASBMR Histomorphometry NomenclatureCommittee. J Bone Miner Res 1987; 2: 595–610.

42 Redlich K, Gortz B, Hayer S, et al. Repair of local bone erosionsand reversal of systemic bone loss upon therapy with anti-tumornecrosis factor in combination with osteoprotegerin or parathyroidhormone in tumor necrosis factor-mediated arthritis. Am J Pathol2004; 164: 543–555.

43 Eferl R, Hasselblatt P, Rath M, et al. Development of pulmonaryfibrosis through a pathway involving the transcription factor Fra-2/AP-1. Proc Natl Acad Sci USA 2008; 105: 10525–10530.

44 Leonard JP, Lacasce AS, Smith MR, et al. Selective CDK4/6inhibition with tumor responses by PD0332991 in patients withmantle cell lymphoma. Blood 2012; 119: 4597–4607.

45 Bluml S, Bonelli M, Niederreiter B, et al. Essential role ofmicroRNA-155 in the pathogenesis of autoimmune arthritis inmice. Arthritis Rheum 2011; 63: 1281–1288.

46 Richards HB, Satoh M, Shaw M, Libert C, Poli V, Reeves WH.Interleukin 6 dependence of anti-DNA antibody production:Evidence for two pathways of autoantibody formation in pris-tane-induced lupus. J Exp Med 1998; 188: 985–990.

47 Savarese E, Steinberg C, Pawar RD, et al. Requirement of Toll-likereceptor 7 for pristane-induced production of autoantibodies anddevelopment of murine lupus nephritis. Arthritis Rheum 2008; 58:1107–1115.

48 Hopkins SJ, Freemont AJ, Jayson MI. Pristane-induced arthritis inBALB/c mice. Curr Top Microbiol Immunol 1985; 122: 213–220.

49 Jimenez-Boj E, Redlich K, Turk B, et al. Interaction between syn-ovial inflammatory tissue and bone marrow in rheumatoid arth-ritis. J Immunol 2005; 175: 2579–2588.

50 Hayer S, Polzer K, Brandl A, et al. B-cell infiltrates induce endo-steal bone formation in inflammatory arthritis. J Bone Miner Res2008; 23: 1650–1660.

51 Chowdhary VR, Grande JP, Luthra HS, David CS.Characterization of haemorrhagic pulmonary capillaritis:Another manifestation of pristane-induced lupus. Rheumatology(Oxford) 2007; 46: 1405–1410.

52 Barker TT, Lee PY, Kelly-Scumpia KM, et al. Pathogenic role of Bcells in the development of diffuse alveolar hemorrhage induced bypristane. Lab Invest 2011; 91: 1540–1550.

53 Hahn B. Animal models of systemic lupus erythematosus.In: Wallace D (ed.), Dubois’ Lupus Erthematosus, 5th ed.Baltimore, MD: Williams & Wilkins, 1997.

54 Holmdahl R, Mo J, Nordling C, et al. Collagen induced arthritis:An experimental model for rheumatoid arthritis with involvementof both DTH and immune complex mediated mechanisms. ClinExp Rheumatol 1989; 7 Suppl 3): S51–S55.

55 Holmdahl R, Andersson ME, Goldschmidt TJ, et al. Collageninduced arthritis as an experimental model for rheumatoid arth-ritis. Immunogenetics, pathogenesis and autoimmunity. APMIS1989; 97: 575–584.

56 Boumpas DT, Eleftheriades EG, Tsokos GC. Cellular immunity insystemic lupus erythematosus. In Vivo 1988; 2: 41–45.

57 Hamilton KJ, Satoh M, Swartz J, Richards HB, Reeves WH.Influence of microbial stimulation on hypergammaglobulinemiaand autoantibody production in pristane-induced lupus. ClinImmunol Immunopathol 1998; 86: 271–279.

58 Satoh M, Kumar A, Kanwar YS, Reeves WH. Anti-nuclear anti-body production and immune-complex glomerulonephritis inBALB/c mice treated with pristane. Proc Natl Acad Sci USA1995; 92: 10934–10938.

59 Calvani N, Satoh M, Croker BP, Reeves WH, Richards HB.Nephritogenic autoantibodies but absence of nephritis in Il-12p35-deficient mice with pristane-induced lupus. Kidney Int2003; 64: 897–905.

60 Levitt NG, Fernandez-Madrid F, Wooley PH. Pristane inducedarthritis in mice. IV. Immunotherapy with monoclonal antibodiesdirected against lymphocyte subsets. J Rheumatol 1992; 19:1342–1347.

61 Richards HB, Satoh M, Shaw M, Libert C, Poli V, Reeves WH.Interleukin 6 dependence of anti-DNA antibody production:Evidence for two pathways of autoantibody formation in pris-tane-induced lupus. J Exp Med 1998; 188: 985–990.

62 Richards HB, Satoh M, Jennette JC, Croker BP, Yoshida H,Reeves WH. Interferon-gamma is required for lupus nephritis inmice treated with the hydrocarbon oil pristane. Kidney Int 2001; 60:2173–2180.

63 Arbuckle MR, McClain MT, Rubertone MV, et al. Developmentof autoantibodies before the clinical onset of systemic lupus ery-thematosus. N Engl J Med 2003; 349: 1526–1533.

64 Feinglass EJ, Arnett FC, Dorsch CA, Zizic TM, Stevens MB.Neuropsychiatric manifestations of systemic lupus erythematosus:Diagnosis, clinical spectrum, and relationship to other features ofthe disease. Medicine (Baltimore) 1976; 55: 323–339.

Pristane-induced lupus as a model of human lupus arthritisH Leiss et al.

790

Lupus

65 Hayahara C, Ikeda K, Sakanishi Y, Ota T. Incidence of serumrheumatoid factors in elder non-rheumatic individuals [inJapanese]. Rinsho Byori Jpn J Clin Pathol 2010; 58: 211–215.

66 Mariz HA, Sato EI, Barbosa SH, Rodrigues SH, Dellavance A,Andrade LE. Pattern on the antinuclear antibody-HEp-2 test is a

critical parameter for discriminating antinuclear antibody-positivehealthy individuals and patients with autoimmune rheumatic dis-eases. Arthritis Rheum 2011; 63: 191–200.

Appendix 1 Auto-antibodies in BALB/c PIL.Auto-antibodies of PIL mice and PBS-injected controls were analyzed eight months after induction of PIL for the occurrence ofauto-abs. Anti-dsDNA abs were analyzed by IF (Crithidia luciliae), anti-Sm abs by immunoblotting; anti-histone abs, anti-chromatin abs, RF and anti-CCP abs, respectively, by ELISA (see Methods section). aELISA results were regarded positive ifvalues exceeded the meanþ 2� SD of controls at baseline (a).An example of anti-dsDNA abs staining by IF (Crithidia luciliae) is given in (b).RF became positive in PIL after two months; interestingly, controls also developed RF over time (c).PIL: pristane-induced lupus; PBS: phosphate-buffered saline; anti-dsDNA: anti-double-stranded DNA; IF: immunofluorescence;abs: antibodies; RF: rheumatoid factor; CCP: anti-cyclic citrullinated peptide; ELISA; enzyme-linked immunosorbent assay; SD:standard deviation.

Pristane-induced lupus as a model of human lupus arthritisH Leiss et al.

791

Lupus

Appendix 2 No inflammatory bone marrow aggregates at sites of complete erosions in PIL.In most erosions in PIL, the inner cortical layer stayed intact and a penetration of the inflammatory pannus into the bone marrowoccurred significantly less frequently than in ASS-matched RA mice (8% in moderately affected PIL mice as opposed to 67% inhTNFtg and 13% in highly affected PIL vs. 80% in CIA (see Results section). Sites of complete cortical destruction are markedwith black arrows in (a)–(d).As a reaction to the cortical penetration of the pannus tissue, subcortical inflammatory aggregates in the bone marrow cavity (witha predominance of B cells) were described in murine and human RA (c), (d), but were absent in PIL-associated arthritis (a), (b).49,50

(Pictures (c) and (d) courtesy of Dr S. Hayer and Dr E. Jimenez-Boj, respectively.)PIL: pristane-induced lupus; ASS: arthritis severity score; RA: rheumatoid arthritis; hTNFtg: human tumor necrosis factor (TNF)-transgenic; CIA: collagen-induced arthritis; HC: healthy controls.

Pristane-induced lupus as a model of human lupus arthritisH Leiss et al.

792

Lupus