ARTHRITIS & RHEUMATISMVol. 54, No. 8, August 2006, pp 2523–2532DOI 10.1002/art.22015© 2006, American College of Rheumatology

Association of �-Actinin–Binding Anti–Double-Stranded DNAAntibodies With Lupus Nephritis

Yves Renaudineau,1 Sabine Croquefer,1 Sandrine Jousse,1 Eric Renaudineau,2

Valerie Devauchelle,1 Paul Gueguen,1 Catherine Hanrotel,1 Boris Gilburd,3 Alain Saraux,1

Yehuda Shoenfeld,3 Chaim Putterman,4 and Pierre Youinou1

Objective. Anti–double-stranded DNA (anti-dsDNA) antibodies may contribute to the pathogenesisof glomerulonephritis (GN) by cross-reacting with�-actinin in murine models and in some patients withsystemic lupus erythematosus (SLE). We thereforesought to determine possible disease associations withserologic and clinical features and to characterize thisnew autoantibody specificity.

Methods. One hundred patients with SLE wererecruited into this multicenter study, as well as 100rheumatic disease controls and 2,100 healthy blooddonors. Clinical disease was evaluated by the SLEDisease Activity Index (SLEDAI; excluding the anti-DNA component). Anti-dsDNA antibodies were detectedby conventional enzyme-linked immunosorbent assay(ELISA) and by a commercial enzyme immunoassay(EIA). Anti–�-actinin antibodies were detected byELISA, and their specificity was confirmed by Westernblotting and by indirect immunofluorescence using ratkidney sections and mesangial cells as substrates.Highly positive sera were selected for absorption exper-

iments and were affinity-purified for cross-reactivitystudies and measurement of antibody avidity.

Results. Sera from 62 of the SLE patients hadanti-dsDNA antibodies; 21 of these sera also had anti–�-actinin antibodies, as compared with 1 of the 38 serawithout anti-dsDNA antibodies. Of the 22 patients withanti–�-actinin antibodies, 10 had GN, as compared with14 of the 78 without anti–�-actinin antibodies (P <0.01). In patients with GN, anti–�-actinin, but notanti-dsDNA, antibodies correlated with the SLEDAIscore (minus the anti-DNA component) and with treat-ment. The fraction of serum anti-dsDNA antibodies thatcross-reacted with �-actinin exhibited high avidity fordsDNA, as determined using a commercial EIA forhigh-avidity anti-dsDNA antibodies and an in-houseconventional ELISA.

Conclusion. The �-actinin–binding antibodies aresignificantly associated with GN in SLE. Whether suchautoantibodies may anticipate the development of thiscomplication of SLE remains to be verified.

Nephritis is a common feature of systemic lupuserythematosus (SLE), making it critical to recognizevariables that would be predictive of this feared compli-cation (1). Early reports have implicated anti–double-stranded DNA (anti-dsDNA) antibodies as pathogenicmediators (2), with serum autoantibody levels correlat-ing with the severity of kidney damage (3). However,later studies cast doubt on the tacit inference from thisobservation that glomerulonephritis (GN) in SLE resultsfrom the deposition of dsDNA–anti-dsDNA antibodyimmune complexes (ICs) (4).

Since some anti-dsDNA antibodies bind to theglomerulus, whereas others do not (5), this reactivitymust encompass an assortment of autoantibodies. Whataccounts for these differences? Support for the view thatthe pathogenic potential of certain dsDNA autoantibod-

Dr. Putterman’s work was supported by grants from the NIH(R01-AR-48692 and P01-AI-51392). Dr. Youinou’s work was sup-ported by a grant from Brest Metropole Oceane.

1Yves Renaudineau, PharmD, PhD, Sabine Croquefer,PharmD, BSc, Sandrine Jousse, MD, Valerie Devauchelle, MD, PhD,Paul Gueguen, BSc, Catherine Hanrotel, MD, Alain Saraux, MD,PhD, Pierre Youinou, MD, DSc, MACR: Brest University MedicalSchool, Brest, France; 2Eric Renaudineau, MD: Caen UniversityMedical School, Caen, France; 3Boris Gilburd, BSc, Yehuda Shoen-feld, MD, FRCP (Hon): Chaim Sheba Medical Center, Tel-Hashomer,Israel, and Sackler Faculty of Medicine, Tel-Aviv University, Tel-Aviv,Israel; 4Chaim Putterman, MD: Albert Einstein College of Medicine,Bronx, New York.

Address correspondence and reprint requests to Pierre Youi-nou, MD, DSc, MACR, Laboratory of Immunology, Brest UniversityMedical School Hospital, BP824, F29609 Brest, France. E-mail:youinou@univ-brest.fr.

Submitted for publication December 19, 2005; accepted inrevised form April 28, 2006.

2523

ies lies in their avidity for dsDNA came from thedemonstration of a higher avidity of dsDNA antibodiesin kidney eluates than in sera from the same patients (6).Yet, amino acid substitutions in monoclonal antibodies(mAb), while increasing avidity for dsDNA, may pre-clude their glomerular sequestration (7). Similarly, sev-eral studies have associated specific anti-dsDNA IgGsubclasses with GN (8), whereas other studies could notestablish such a connection (9). Hence, the determinantsof the pathogenicity of dsDNA antibodies remain elu-sive.

Such inconsistencies in previous results favor thealternative view that autoantibodies bind directly toglomerular antigens. Consistent with this idea is thefinding that cross-reactivity characterizes anti-dsDNAantibodies that are nephritogenic (10). Specificity fordsDNA is not sufficient for GN to occur, because not allSLE patients with elevated anti-dsDNA levels developthis complication (11), and all SLE patients with GN donot have elevated anti-dsDNA levels (12). Further fuel-ing the idea that specificity for dsDNA is dispensablewith regard to renal complications is the finding thatantibodies reacting with dsDNA and related nuclearantigens account for as little as 10% of the IgG elutedfrom the kidneys of SLE patients (13). This signifies thatin the glomeruli, a large proportion of antibodies recog-nize antigens other than dsDNA.

Given the possible progression of GN to end-stage renal disease, the need for serologic tools withwhich to evaluate its severity has come to be pressing,and conclusive identification of target antigens for ne-phritogenic antibodies has become essential. A fewcandidates have been acknowledged, such as nucleo-somes (14), the collagen-like regions of C1q (15), or theribosomal P proteins (16). Although the occurrence ofanti-dsDNA antibody–negative SLE has proved to be astumbling block for the idea that dsDNA antibodies arepivotal in the pathogenesis of GN, a subset of theseantibodies may cross-react with endothelial cell–associated antigens, such as �-enolase (17), histones(18), or heat-shock protein 60 (19), or with glomerularcomponents, such as laminin (20), heparan sulfate (21),or fibronectin (22). Recently, �-actinin has been en-dorsed as an additional reasonable candidate antigen.This is a family of 4 closely related gene products, allexisting as head-to-tail dimers that are ideally shaped forcrosslinking actin filaments (23). Whereas �-actinin 1and �-actinin 4 are widely expressed, only �-actinin 4 ispresent at significant levels in the human kidney. Anti-dsDNA antibodies cross-reacting with this constituent ofglomerular podocytes and mesangial cells were discov-

ered in mice (24,25), and their presence was confirmedin small numbers of SLE patients (26,27).

Following previous reports of the cross-reactivityof anti-DNA antibodies with �-actinin (24–27), we ex-amined SLE sera and identified the following 3 groups:anti-dsDNA positive/anti–�-actinin positive, anti-dsDNA positive/anti–�-actinin negative, and anti-dsDNA negative/anti–�-actinin positive. Our analysesfailed, however, to discern an association of anti-dsDNAor anti–�-actinin antibodies with musculoskeletal, hema-tologic, cardiac, or neurologic involvement in thesepatients. In contrast, anti–�-actinin, rather than anti-dsDNA, antibodies were significantly associated withGN. Interestingly, we also found that there was cross-reactivity with �-actinin in high-avidity anti-dsDNAantibodies, and that the anti–�-actinin response wasrelated to the actin-binding site of �-actinin. Our find-ings further support a role of anti–�-actinin antibodiesin the pathogenesis of GN. Additional longitudinalstudies are warranted to determine whether the pres-ence of anti–�-actinin antibodies anticipates the devel-opment of this complication.

PATIENTS AND METHODS

Patients and controls. Blood samples were collectedfrom 100 patients who fulfilled the American College ofRheumatology revised criteria for SLE (28). There were 76patients from rheumatology units (58 at Brest UniversityHospital, and 18 at Albert Einstein College of Medicine) and24 patients from nephrology units (14 at Brest UniversityHospital, and 10 at the Caen University Hospitals). These 14men and 86 women ranged in age from 18 years to 77 years.

The SLE Disease Activity Index (SLEDAI) was calcu-lated (29), but because increased DNA binding is part of theinstrument, this item was omitted from the total SLEDAIscore (SLEDAI minus the anti-DNA component). Fifty-fivepatients were designated as having inactive disease based ontheir scores of �3, and the remaining 45 patients weredesignated as having active disease based on their scores of �3(30). An SLE flare was defined as an increase of �1 on thephysician’s global assessment (31). Twenty-four of the 45patients with active disease presented with GN that wasclassified as class III or class IV according to the World HealthOrganization (WHO) revised criteria (32). At the time ofblood sampling, the remaining 76 SLE patients did not haveevidence of renal disease, such as active urinary sediment orproteinuria. That does not imply they had not had suchsymptoms in the past.

The clinical characteristics and treatments of the 100SLE patients are given in Table 1. Seventy-one of the patientswere receiving therapy for SLE, such as prednisone and/orimmunosuppressive agents, while another 12 were not receiv-ing treatment. None of the patients was receiving long-termhemodialysis.

Rheumatic disease control subjects and normal control

2524 RENAUDINEAU ET AL

subjects were selected from 2 large cohorts of patients and 2large groups of blood donors, respectively, in order to matchthem as precisely as possible to the SLE patients (Table 2).Thus, 70 of 345 sera from patients with rheumatoid arthritis(RA) and 30 of 157 sera from patients with primary Sjogren’ssyndrome (SS) were evaluated. These patients fulfilled theproposed diagnostic criteria for their respective conditions(33,34). The disease duration in the SLE, RA, and primary SSpatients was similar, and none of the rheumatic diseasecontrols had evidence of GN or interstitial nephritis (35).Rheumatic disease controls were similar to the SLE patients interms of age, sex, and race. In addition, 100 of 293 sera fromblood donors at the Tel-Hashomer Hospital and 100 of 378sera from blood donors at the blood bank in Brest werestudied. These healthy volunteer donors were similar to theSLE patients in terms of age, sex, and race.

Before the series of experiments conducted in thepresent study, another 80 aliquots of sera from healthy volun-teer donors were used to determine the conditions for theenzyme-linked immunosorbent assays (ELISAs) performed inthe Laboratory of Immunology at Brest University. Thesedonors were members of the laboratory and clinic staff, as wellas residents of a nursing home.

All SLE patients, rheumatic disease controls, andnormal controls gave their informed consent for study. Thestudy was approved by the Ethics Committees of the Brest andCaen Hospitals, the Albert Einstein College of Medicine, andthe University Hospital at Tel-Hashomer.

Cell culture. BALB/c mouse–derived, SV40-transformed mesangial cells were derived as described previ-

ously (25). Cells were maintained in Dulbecco’s modifiedEagle’s medium supplemented with 10% fetal calf serum.

Autoantibody tests. An ELISA was set up to determinedsDNA reactivity. DNA (Sigma, St. Louis, MO) was diluted to5 �g/ml in phosphate buffered saline (PBS) and immediatelyspread onto poly-L-lysine–coated plates (Nunc, Roskilde, Den-mark). After an 18-hour incubation at 4°C, bound single-stranded DNA was digested with S1 nuclease (Sigma). Thisconventional ELISA was performed as described elsewhere(36). The optical density (OD) cutoff was fixed at 0.150, whichis 3 SD above the mean value in sera from 80 healthy volunteerdonors.

Conventional ELISAs measure anti-dsDNA antibod-ies, irrespective of their avidity. In contrast, due to the high saltconcentration required to precipitate ICs, the Farr radioim-munoassay selects high-avidity autoantibodies (37). This fea-ture was exploited to develop ELISAs that exclusively detectedhigh-avidity anti-dsDNA antibodies (38). We chose the Farr-zyme enzyme immunoassay (EIA; The Binding Site, Birming-ham, UK) as an alternative to the traditional Farr immunoas-say in order to detect high-avidity anti-dsDNA of the IgGsubclass. Following the manufacturer’s instructions, the testwas calibrated (in IU/ml) against a WHO standard and usingthe threshold established by the manufacturer (30 IU/ml,which is 3 SD above the mean value in sera from 150 Britishblood donors).

As described previously (39), our IgG-specific ELISAfor anti–�-actinin antibodies was adapted from a methoddescribed previously (26,27). Linbro microplates (Flow, Irvine,UK) were coated with 10 �g/ml of chicken �-actinin (Sigma) inbicarbonate buffer, pH 9.6, left to evaporate overnight at 37°C,washed 3 times in PBS, blocked for 1 hour at 37°C with 2%bovine serum albumin (BSA) in PBS (PBS–BSA), and washedwith PBS containing 0.05% Tween 20 (PBST). Sera werediluted 1:200 in PBS–BSA and incubated for 90 minutes at37°C. Following another 3 washes, bound antibody was visual-ized using a horseradish peroxidase (HRP)–conjugatedF(ab�)2 fragment of goat anti-human IgG (Jackson Immu-noResearch, West Grove, PA). The OD of each sample boundto �-actinin–free wells was automatically subtracted from theOD of the test sample. Standard curves were constructed withstrongly positive SLE sera, which thereafter served as internalreference sera. Sera were scored positive if the OD was�0.150, which is 3 SD above the mean value in sera from 80healthy volunteer donors.

A specific ELISA measured antibodies against actin.Microtiter plates were coated with 100 �l of a 10-�g/mlpreparation of actin (Sigma). SLE test sera, positive controlsera from patients with autoimmune chronic hepatitis, andnegative control sera from healthy blood donors were diluted1:100 in PBS–BSA and incubated for 90 minutes at 37°C. AnHRP-conjugated goat anti-human IgG was used to reveal theantibody binding. The cutoff level for positivity was set at anOD of 0.170, which is 3 SD above the mean value in sera from80 healthy volunteer donors.

Confirmation of anti–�-actinin activity. The presenceof anti–�-actinin antibodies was confirmed by Western blottingand indirect immunofluorescence. Two anti-dsDNA–positive/anti–�-actinin–positive sera (sera 1 and 2) and 2 anti-dsDNA–positive/anti–�-actinin–negative sera (sera 3 and 4) were se-lected on the basis of their high reactivity in the dsDNA and

Table 1. Clinical characteristics and treatment of 100 patients withsystemic lupus erythematosus at the time their sera were examined

SLEDAI score, mean (range)* 3.6 (0–18)No. with renal disease 24No. with musculoskeletal involvement 74No. with hematologic disease 35No. with cardiac involvement 21No. with neurologic disease 1Treatment

No. taking prednisone 71Prednisone dosage, mean (range) mg/day 7.5 (0–50)No. taking immunosuppressive agents 12

* SLEDAI � Systemic Lupus Erythematosus Disease Activity Index(minus the anti-DNA component).

Table 2. Demographic features of the systemic lupus erythematosuspatients, rheumatic disease control patients, and normal control subjects*

No. of males/females

Age, mean(range) years

Systemic lupus erythematosus(n � 100)

14/86 43.8 (18–77)

Rheumatoid arthritis controls(n � 70)

10/60 52.3 (27–77)

Primary Sjogren’s syndromecontrols (n � 30)

4/26 41.6 (18–69)

Israeli blood donors (n � 100) 14/86 43.1 (20–65)French blood donors (n � 100) 20/80 44.4 (20–65)

ASSOCIATION OF ANTI–�-ACTININ ANTIBODIES WITH LUPUS NEPHRITIS 2525

the �-actinin antibody tests and for their high reactivity only inthe dsDNA antibody test, respectively. These 4 sera, plus aserum sample from a healthy volunteer donor, were loaded ona protein G–Sepharose column (Pharmacia, Uppsala, Swe-den), and IgG was eluted with 0.1M HCl-glycine, pH 2.8.In-house ELISAs and Western blotting showed that only IgGwas present.

The �-actinin was subjected to 10% sodium dodecylsulfate–polyacrylamide gel electrophoresis and then electro-eluted for 3 hours to polyvinylidene difluoride sheets (Bio-Rad, Hercules, CA). The membranes were blocked overnightusing 5% nonfat dry milk in PBS, probed with IgG at aconcentration of 10 �g/ml in PBST containing 1% milk, andrevealed by biotinylated F(ab�)2 fragment of goat anti-humanIgG (Zymed, South San Francisco, CA). IgG from a healthyblood donor served as the negative control and anti–�-actininmAb EA-53 (Sigma) as the positive control. Samples were thenincubated for 1 hour with HRP-labeled streptavidin or HRP-labeled goat F(ab�)2 fragment of anti-mouse IgG (Zymed).

Similar analyses were performed by indirect immuno-fluorescence using mesangial cell smears or rat kidney sections.The cells were grown to confluence on coverslips, treated with4% paraformaldehyde for 30 minutes, washed 3 times withPBS–BSA, and incubated for 45 minutes with 20 �g/ml ofpatient IgG, negative control, or positive control. The prepa-rations were washed 3 times with PBS, and bound antibodieswere stained with tetramethylrhodamine isothiocyanate–labeled donkey anti-mouse IgG (Dakopatts, Glostrup, Den-mark) or fluorescein isothiocyanate–labeled rabbit F(ab�)2fragment of anti-human IgG (Jackson ImmunoResearch).Confocal images were acquired using a Leica TCS-NT fluores-cence microscope (Leica Microsystems, Bensheim, Germany).

To exclude artifactual DNA in the �-actinin prepara-tion, plates were incubated with 100 �l of a 100-�g/ml solutionof DNase I (Roche, Indianapolis, IN) in DNase buffer (0.01MTris, 0.01M MgCl2, and 0.01M CaCl2, pH 7.5), as describedpreviously (26). The ODs obtained with 5 anti-dsDNA–positive/anti–�-actinin–positive SLE sera were compared withthose obtained without DNase I treatment in this anti–�-actinin ELISA.

Competition experiments. One 100-�l aliquot of sera1–4 at dilutions giving 50% of their maximum OD in theanti-dsDNA ELISA was mixed with an equal volume of PBScontaining dsDNA diluted from 0.1 �g/ml to 500 �g/ml. Asecond 100-�l aliquot of sera 1 and 2 at dilutions giving 50% oftheir maximum OD in the anti–�-actinin ELISA and of sera 3and 4 at dilutions set arbitrarily halfway between that of serum1 and that of serum 2 was mixed with an equal volume of PBScontaining �-actinin diluted from 0.1 �g/ml to 500 �g/ml. Inaddition to these specific competitions, 2 anti-dsDNA–positive/anti–�-actinin–positive/antiactin-negative sera, 2 anti-dsDNA–positive/anti–�-actinin–negative/antiactin-positivesera, and 2 anti-dsDNA–positive/anti–�-actinin–positive/antiactin-positive sera were diluted and mixed with an equalvolume of PBS containing actin diluted from 0.1 �g/ml to 500�g/ml.

Following 1 hour of incubation, the samples weredepleted of ICs by precipitation with 2% polyethylene glycol6000, and after a 30-minute centrifugation at 15,000g, thesupernatants were removed and assayed in the anti-dsDNA,

the anti–�-actinin, and the antiactin ELISAs. The respectivepercentages of inhibition were calculated as follows: [(ODbefore inhibition – OD after inhibition)/(OD before inhibi-tion)] � 100.

Affinity-purification of autoantibodies. IgG from sera1–4 and from a normal control serum were purified. One halfof each IgG preparation was applied to a dsDNA column(Amersham, Aylesbury, UK), and the other half was runthrough a column containing 1 mg of �-actinin coupled toHiTrap NHS-activated Sepharose (Amersham).

The IgG that bound to the dsDNA column and the IgGthat bound to the �-actinin column were recovered with 15 mlof 0.15M NaCl, 0.4M NaCl, and 6M urea–2M NaCl. Thecollected fractions were dialyzed against PBS. The concentra-tions of specific IgG were then determined using the Bradfordmicro–bicinchoninic acid assay (Bio-Rad) and were adjustedto 20 �g/ml in PBS–BSA before being assayed for anti-dsDNAand anti–�-actinin.

Determination of autoantibody avidity. This techniquehas been described elsewhere (40). Avidity was measured by anELISA, based on the use of diethylamine, to dissociate low-avidity antibodies from dsDNA. The sera were serially dilutedon a dsDNA-coated plate, in the presence or absence of 10mM diethylamine (Sigma). Following incubation, the plateswere washed 3 times with PBST, and antibody binding wasrevealed using an HRP-conjugated F(ab�)2 fragment of goatanti-human IgG. After a 60-minute incubation and 3 washes,dose-response curves were plotted, and the diethylamine-induced decrease in the log10 titer was measured at half-maximal OD. The extent of this inhibition index was taken asan estimate of the avidity, to which it correlates inversely;therefore, the inhibition index is low when the avidity is highand vice versa.

Statistical analysis. All results from triplicate wellswere averaged. In accordance with the approximately normaldistribution of the anti-dsDNA and the anti–�-actinin antibod-ies, the illustrations display the arithmetic mean and SD. Datawere compared using chi-square or Fisher’s exact tests, as wellas the Mann-Whitney U test for unpaired data. The correla-tions were established by Spearman’s correlation test. TheBonferroni correction was applied for multiple comparisons,and corrected P (Pcorr) values are shown.

RESULTS

Autoantibodycombination.Theconventionalanti-dsDNA ELISA was positive for 62 of the 100 SLE sera.Twenty-one of these 62 SLE sera were also positive inthe anti–�-actinin ELISA, as compared with only 1 ofthe 38 sera that were negative in the anti-dsDNAELISA. SLE sera were thus categorized on the basis oftheir reactivity in the 2 ELISAs: 21 group A sera weredsDNA positive/�-actinin positive, 41 group B sera weredsDNA positive/�-actinin negative, 1 group C serum wasdsDNA negative/�-actinin positive, and 37 group D serawere dsDNA negative/�-actinin negative. Among the

2526 RENAUDINEAU ET AL

control groups, 9 and 4 of the 100 rheumatic diseasecontrol sera, 3 and 2 of the 100 Israeli blood donor sera,and 4 and 1 of the 100 French blood donor seradisplayed dsDNA and anti–�-actinin reactivity, respec-tively. Double positivity was found in none of these 3sets of control sera.

Anti–�-actinin reactivity was verified using 2additional methods. Western blotting confirmed thatIgG from group A sera recognized �-actinin (Figure1A), and indirect immunofluorescence analysis of thesame IgG preparations showed that they bound to

�-actinin on rat kidney sections (Figure 1B, top panels)and mesangial cell smears (Figure 1B, bottom panels).These anti–�-actinin IgG colocalized with the anti–�-actinin mAb, as demonstrated by the yellow fluorescenceseen on the merged images of the green-stained humanIgG with the red-stained murine IgG (Figure 1C).

To exclude the possibility that positivity in theanti–�-actinin test was induced by trace amounts ofdsDNA in the �-actinin preparation, the plates wereincubated with DNase I prior to the anti–�-actinin assay.This treatment reduced the OD values of 5 of the group

Figure 1. Confirmation of the presence of anti–�-actinin (Act) antibodies in the serum of patients withsystemic lupus erythematosus (SLE). A, Two sera from SLE patients, one anti–double-stranded DNA(anti-dsDNA) positive/anti–�-actinin negative (group B) and the other anti-dsDNA positive/anti–�-actinin positive (group A), were used to probe �-actinin. Anti–�-actinin monoclonal antibody (mAb) wasused as the positive control and a healthy blood donor serum as the negative control. B, Sera used in Awere examined by indirect immunofluorescence on rat kidney sections (top) and mesangial cell smears(bottom) (original magnification � 640 top; � 1,000 bottom). C, These IgG colocalized with anti–�-actinin mAb, as demonstrated in the overlay of green-stained human IgG and red-stained murine IgG(yellow) (original magnification � 640). D, Influence of actin on the binding of antibodies to �-actinin.Increasing amounts of actin were added to 2 antiactin antibody–negative group A sera (solid lines), 2antiactin antibody–negative group B sera (dotted lines), and 2 antiactin antibody–positive group A sera(broken lines). These 6 sera were then examined for their anti–�-actinin activity.

ASSOCIATION OF ANTI–�-ACTININ ANTIBODIES WITH LUPUS NEPHRITIS 2527

A sera by �5% as compared with the OD valuesobtained without DNase I treatment.

Association between serologic features and clin-ical features. There were no associations of anti-dsDNAor anti–�-actinin antibodies with musculoskeletal, hema-tologic, cardiac, or neurologic involvement. Anti–�-actinin antibodies, rather than anti-dsDNA antibodies,were associated with GN irrespective of the assay used.Indeed, 16 of the 62 anti-dsDNA–positive SLE patientshad GN (Table 3), compared with 8 of the 38 anti-dsDNA–negative SLE patients (P not significant),whereas 10 of the 22 anti–�-actinin–positive SLE pa-tients had GN, compared with 14 of the 78 anti–�-actinin–negative SLE patients (Pcorr � 0.05). The levelsof conventional anti-dsDNA and anti–�-actinin antibod-ies were similar in SLE patients with and those withoutGN (mean � SD 0.467 � 0.469 versus 0.260 � 0.345 ODunits, and 0.154 � 0.169 versus 0.106 � 0.106 OD units).

Searching analyses disclosed that in the 24 SLEpatients with GN, only a SLEDAI score (minus theanti-DNA component) of �3 (Pcorr � 0.005) and treat-ment (Pcorr � 10–6) were associated with anti–�-actinin,and that treatment was associated with high-avidityanti-dsDNA antibodies (Pcorr � 2 � 10–4) but not withconventional anti-dsDNA antibodies (Table 3).

This group of 24 SLE patients with GN wasfurther categorized into 2 subgroups according to theoccurrence (n � 12) or no occurrence (n � 12) of adisease flare at the time the blood was collected, andaccording to the prescribing (n � 14) or no prescribing

(n � 10) of a treatment (Table 4). Flares and treatmentwere associated with high-avidity anti-dsDNA antibod-ies, as determined by the Farrzyme EIA (Pcorr � 0.003 andPcorr � 0.03, respectively) and with anti–�-actinin antibod-ies (Pcorr � 0.03 and Pcorr � 5 � 10–6, respectively).

Relationships between anti-dsDNA and anti–�-actinin autoantibodies. There seem to be 2 likely expla-nations for these findings. One is that 2 independentgroups of autoantibodies coexist in a given serum, one ofwhich directed toward dsDNA and the other toward�-actinin. Alternatively, there is a subset of anti-dsDNAantibodies that cross-react with �-actinin. To distinguishbetween these 2 interpretations, 2 group A SLE sera(sera 1 and 2) and 2 group B SLE sera (sera 3 and 4)were preincubated with increasing amounts of dsDNAprior to retesting in the anti–�-actinin ELISA or withincreasing amounts of �-actinin prior to retesting in theanti-dsDNA ELISA.

While 100% of anti-dsDNA was absorbed bydsDNA in SLE sera 1–4 (Figure 2A), the binding ofgroup A sera, but not group B sera, to �-actinin wasreduced by 50% (Figure 2B). Conversely, 50% of anti-dsDNA antibodies were removed from group A sera byincubation with �-actinin, and again, there were nochanges in group B sera (Figure 2A�). As expected, thebinding of group A sera to �-actinin was absorbed byincubation with �-actinin (Figure 2B�). Thus, 50% ofanti–�-actinin antibodies in group A sera belonged tothe anti-dsDNA antibody population in group A sera,implying that 50% of their anti-dsDNA antibodies cross-

Table 3. Relationship between nephritis and serologic variables in 100 patients with systemic lupuserythematosus*

Anti-dsDNA

Conventional ELISA Farrzyme EIA Anti–�-actinin

Present(n � 62)

Absent(n � 38)

Present(n � 30)

Absent(n � 70)

Present(n � 22)

Absent(n � 78)

NephritisYes (n � 24) 16 8 12 12 10 14No (n � 76) 46 30 18 58 12 64Corrected P NS NS �0.05

SLEDAI score�3 (n � 45) 33 12 18 27 17 28�3 (n � 55) 29 26 12 43 5 50Corrected P NS NS �0.005

TreatmentYes (n � 73) 42 31 14 59 4 69No (n � 27) 20 7 16 11 18 9Corrected P NS �2 � 10�4 �10�6

* Anti-dsDNA � anti–double-stranded DNA; ELISA � enzyme-linked immunosorbent assay; EIA �enzyme immunoassay; NS � not significant; SLEDAI � Systemic Lupus Erythematosus Disease ActivityIndex (minus the anti-DNA component).

2528 RENAUDINEAU ET AL

reacted with �-actinin, as compared with 0% of thegroup B SLE sera.

An aliquot of sera 1–4 was then affinity-purifiedover a dsDNA column and tested in the dsDNA and the�-actinin ELISAs. A second aliquot of group A sera, butnot group B sera, was affinity-purified over an �-actinincolumn and tested in the �-actinin and the dsDNAELISAs. Two populations of anti-dsDNA antibodieswere recovered from the dsDNA column (Figure 2C).The one eluted by 0.4M NaCl did not bind to �-actinin,whereas the one eluted by 6M urea–2M NaCl bound to�-actinin (Figure 2C�). Not surprisingly, the flow-through reacted with �-actinin, but not with dsDNA. Incontrast, the �-actinin column retained one populationof anti-dsDNA antibodies from group A sera and nonefrom group B sera (Figure 2D). The anti–�-actininantibodies eluted from the �-actinin column bound to�-actinin (Figure 2D�). These affinity purificationsstrengthen our interpretation in that high-avidity anti-dsDNA antibodies, but not low-avidity anti-dsDNA an-tibodies, cross-react with �-actinin.

Characterization of anti-dsDNA antibodies. Thesimplest approach to assessing avidity was to comparedsDNA antibodies found using our conventional ELISAwith dsDNA antibodies found using the Farrzyme EIA.Unlike those of the conventional ELISA, the Farrzymeanti-dsDNA antibodies were significantly associated(Pcorr � 0.03) with GN. Assuming that low-avidityanti-dsDNA antibodies go undetected in the latter test,these differences indicate that anti-dsDNA autoantibod-

ies that contribute to the development of GN have thehighest avidity. Such observations were confirmed byshowing that the avidity for dsDNA was higher in 5patients with GN than in 10 patients without GN(mean � SD 0.55 � 0.16 versus 1.03 � 0.53 OD units;P � 0.05).

Finally, given that �-actinin is an actin-crosslinking protein and that antiactin IgG are commonin SLE, this additional specificity warranted determina-tion in our 100 SLE sera. Low levels of antiactinantibody existed in 31 of the 100 sera, with ODs rangingfrom 0.180 to 0.307. Six sera were preincubated withincreasing amounts of actin and reassayed in the anti–�-actinin ELISA (Figure 1D). The anti–�-actinin reac-tivities of the 2 antiactin antibody–negative group A serawere reduced in the presence of actin in a dose-dependent manner, those of the 2 antiactin antibody–positive group B sera were unchanged, and those of the2 antiactin antibody–positive group A sera were in-creased (Figure 1D).

These results indicate that the reduced anti–�-actinin OD of some sera was due to the masking of�-actinin by actin. In contrast, the modest elevation ofthe anti–�-actinin reactivity of other sera suggests thatliquid-phase actin sticks on solid-phase �-actinin andretains antiactin antibodies. Taken together, our find-ings support the view that the anti–�-actinin response iselicited by the actin-binding site of the �-actinin mole-cule.

Table 4. Relationship between the occurrence of SLE flares or the prescribing of a treatment and thepresence of anti–�-actinin or anti-dsDNA antibodies in 24 patients with lupus nephritis*

Antibodies

SLE flare Treatment

Yes(n � 12)

No(n � 12)

Yes(n � 14)

No(n � 10)

Anti-dsDNA by conventional ELISAYes 10 6 7 9No 2 6 7 1Corrected P NS NS

Anti-dsDNA by Farrzyme EIAYes 10 2 3 9No 2 10 11 1Corrected P �0.003 �0.03

Anti–�-actininYes 9 1 0 10No 3 11 14 0Corrected P �0.03 �5 � 10�6

* Anti–double-stranded DNA (anti-dsDNA) antibodies were determined by conventional enzyme-linkedimmunosorbent assay (ELISA) and by Farrzyme enzyme immunoassay (EIA) methods (see Patients andMethods for details). SLE � systemic lupus erythematosus; NS � not significant.

ASSOCIATION OF ANTI–�-ACTININ ANTIBODIES WITH LUPUS NEPHRITIS 2529

DISCUSSION

Approximately one-third of SLE patients developovert renal disease (41), but specific autoantibody may

accumulate before the emergence of this complication(42). The paradigm of systemic autoimmunity based onanti-DNA antibodies stems from their discovery ininflamed kidneys (7). However, one is struck by theacceptance of the theory that the mechanism of tissueinjury is by DNA-containing ICs, without any physicalisolation of DNA from glomeruli (43). Nephritogenicitymay, in fact, be restricted to subsets of the spectrum ofanti-dsDNA antibodies. A first subpopulation would beassociated with dsDNA, a second would be trapped bynegatively charged molecules (the latter cross-reactingwith glomerular antigens), and a third would be seques-tered by virtue of their genuine specificity for localantigens. The report that non–DNA-binding antibodiesrecognize the same renal antigens as DNA-bindingantibodies (44) is compatible with our view that cross-reactivity of anti-dsDNA antibodies with kidney antigensdetermines their renal pathogenicity.

Three combinations of autoantibodies were iden-tified in our analysis of 100 SLE sera. Group A (anti-DNA positive/anti–�-actinin positive) consists of high-avidity anti-dsDNA antibodies that were eluted from theDNA column with difficulty, retained in the �-actinincolumn, and absorbable with dsDNA as well as�-actinin. Group B (anti-DNA positive/anti–�-actininnegative) consists of low-avidity anti-dsDNA antibodiesthat were easily eluted from the DNA column, flowedthrough the �-actinin column, and bound to dsDNA, butnot to �-actinin. Group C (anti-DNA negative/anti–�-actinin positive) consists of antibodies that flowedthrough the DNA column, bound to the �-actinin col-umn, and were absorbed with �-actinin. A few SLE seraharbored only group A antibodies and high-avidityanti-dsDNA antibodies that did not cross-react with�-actinin.

Among the anti-dsDNA antibodies, those cross-reacting with a protein expressed by a given tissue wouldbe pathogenic for that particular organ. Interesting inthis respect is the model in which dsDNA antibodies thatrecognize laminin have been implicated in recurrentfetal loss (45), because laminin is crucial to the implan-tation of the ovum in humans. In our own study, themechanisms of the renal insult remain unclear. How-ever, point mutations in the ACTN4 gene have beenrecognized (46) in individuals with familial focal seg-mental glomerulosclerosis (FSGS), and they alter theactin-binding activity of �-actinin 4 (47). Althoughknockin mice with a disease-associated ACTN4 mutationdevelop clinical manifestations similar to those in hu-mans with FSGS, the pathogenicity of the anti–�-actininantibodies remains unexplained. Inasmuch as damage to

Figure 2. Competition and affinity-purification experiments. Twoanti–double-stranded DNA (anti-dsDNA)–positive/anti–�-actinin(Act)–positive sera from group A (solid lines) and 2 anti-dsDNA–positive/anti–�-actinin–negative sera from group B (dotted lines) wereselected in the competition and affinity-purification experiments. Aand B, Sera were mixed with dsDNA diluted 0.1–500 �g/ml and thenreassayed for anti-dsDNA (A) and anti–�-actinin (B) antibodies. A�

and B�, The same sera were mixed with �-actinin diluted 0.1–500mg/ml and than reassayed for anti-dsDNA (A�) and anti–�-actinin (B�)antibodies. C and D, Purified IgG from the 4 sera were applied to adsDNA column (C) or to an �-actinin column (D). C� and D�, Theeffluents and eluates from the dsDNA column were tested for anti-dsDNA (C) and anti–�-actinin (C�), and those from the �-actinin columnwere tested for anti-dsDNA (D) and anti–�-actinin (D�) antibodies.

2530 RENAUDINEAU ET AL

the kidney requires target antigen display, the expressionof �-actinin contributes to lupus GN.

Anti–�-actinin antibodies specifically target theactin-binding site of the molecule. In order to refine thisobservation, we are developing an ELISA with syntheticpeptides that mimic the actin-binding site of �-actinin asthe substrate. Based on the demonstration that a patho-genic anti-dsDNA/�-actinin mAb binds more strongly toimmortalized mesangial cells derived from a lupus-prone MRL-lpr/lpr mouse than to those from normalmice, susceptibility to autoantibody-mediated GN maybe envisioned as being dependent upon genetics in termsof antigen display (25). The actin-binding site might beselectively uncovered in the context of genetic predispo-sition. The cytokine milieu can also be invoked as afeature of divergences in autoantigen expression. In-deed, it has been shown that basic fibroblast growthfactor and transforming growth factor � favor the ex-pression of �-actinin (48); thus, cytokines may up-regulate the local expression of the target antigen forpathogenic antibodies.

Titers of antibodies to �-actinin were significantlymore likely than those to dsDNA to be increased inpatients with GN as compared with those without GN.The anti–�-actinin reactivity was associated with high-avidity, but not low-avidity, anti-dsDNA antibodies. Wespeculate that such correlations result from a progres-sion of the avidity from low-avidity to high-aviditydsDNA binding (49). Nonpathogenic autoantibodiesmay thus converge toward dsDNA with increasing avid-ity through clonal expansion and somatic mutation (50).Stimulation is sustained by dsDNA along with �-actininor by epitopes shared by dsDNA and �-actinin. Al-though unproven, it is possible that a new population ofautoantibodies emerges that is directed toward�-actinin, but no longer toward dsDNA (51).

Clarification of how these anti–�-actinin antibod-ies cross-react with dsDNA may help identify whichpatients will experience an aggressive course of theirGN. Serial determinations of anti–�-actinin antibodytiters, before and after the onset of kidney involvement,remain to be completed. Prompted by the promise ofnovel treatment approaches (52), we have launchedthese studies in our laboratories.

ACKNOWLEDGMENTS

Thanks are due to Cindy Sene and Simone Forest forsecretarial assistance.

REFERENCES

1. Cameron JJ. Clinical manifestations of lupus nephritis. In: LewisEJ, Schwartz MM, Korbet SM, editors. Lupus nephritis. Oxford:Oxford University Press; 1999. p. 159–84.

2. Koffer D, Shur PG, Kunkel HG. Immunological studies concern-ing nephritis of systemic lupus erythematosus. J Exp Med 1967;126:607–24.

3. Koffer D, Carr R, Agnello V, Thoburn R, Kunkel HG. Antibodiesto polynucleotides in human sera: antigenic specificity and relationto disease. J Exp Med 1971;134:294–312.

4. Swaak AJ, Aarden LA, Statius Van Eps LW, Feltkamp TE.Anti-dsDNA and complement profiles as prognostic guides insystemic lupus erythematosus. Arthritis Rheum 1979;22:226–35.

5. Budhai L, Oh K, Davidson A. An in vitro assay for detection ofglomerular binding IgG autoantibodies in patients with systemiclupus erythematosus. J Clin Invest 1996;98:1585–93.

6. Winfield JB, Faiferman I, Koffler D. Avidity of anti-DNA anti-bodies in serum and IgG glomerular eluates from patients withsystemic lupus erythematosus: association of high-avidity antina-tive DNA antibody with glomerulonephritis. J Clin Invest 1997;59:90–6.

7. Katz JB, Limpanasithikul W, Diamond B. Mutational analysis ofan auto-antibody: differential binding and pathogenicity. J ExpMed 1994;180:925–32.

8. Devey ME, Lee SR, Le Page S, Feldman R, Isenberg DA. Serialstudies of the IgG subclass and functional affinity of DNAantibodies in systemic lupus erythematosus. J Autoimmun 1988;1:483–94.

9. Gershwin ME, Steinberg AD. Qualitative characteristics of anti-DNA antibodies in lupus nephritis. Arthritis Rheum 1974;17:947–54.

10. Pankewycz OG, Migliorini P, Madaio MP. Polyreactive autoanti-bodies are nephritogenic in murine lupus nephritis. J Immunol1987;139:3287–94.

11. Foster MH, Cizman B, Madaio MP. Nephritogenic autoantibodiesin systemic lupus erythematosus: immunochemical properties,mechanisms of immune deposition, and genetic origins. Lab Invest1993;69:494–507.

12. Lloyd W, Schur PH. Immune complexes, complement and anti-DNA in exacerbations of systemic lupus erythematosus. Medicine(Baltimore) 1981;60:208–17.

13. Mannik M, Merrill CE, Stamps LD, Wener MH. Multiple auto-antibodies from the glomerular immune deposits in patients withsystemic lupus erythematosus. J Rheumatol 2003;30:495–504.

14. Burlingame RW, Boey ML, Starkebaum G, Rubin RL. The centralrole of chromatin in autoimmune responses to histones and DNAin systemic lupus erythematosus. J Clin Invest 1994;94:184–92.

15. Uwatoko S, Mannik M. Low molecular weight C1q-binding immu-noglobulin G in patients with systemic lupus erythematosus con-sists of autoantibodies to the collagen-like region of C1q. J ClinInvest 1988;82:816–24.

16. Hulsey M, Golstein R, Scully L, Swbeck W, Reichlin M. Antiri-bosomal P antibodies in system lupus erythematosus: a case-control study correlating hepatic and renal disease. Clin ImmunolImmunopathol 1995;74:252–6.

17. Pratesi F, Moscato S, Sabbatini A, Chimenti D, Bombardieri S,Migliorini P. Autoantibodies specific for �-enolase in autoimmunedisorders. J Rheumatol 2000;27:109–15.

18. Chan TM, Frampton G, Staines NA, Hobby P, Perry GJ, CameronJS. Different mechanisms by which anti-DNA MoAbs bind tohuman endothelial cells and glomerular mesangial cells. Clin ExpImmunol 1992;88:68–74.

19. Jamin C, Dugue C, Alard JE, Jousse S, Saraux A, Guillevin L, etal. Induction of endothelial cell apoptosis by the binding ofanti-endothelial cell antibodies to hsp60 in vasculitis-associatedsystemic autoimmune diseases. Arthritis Rheum 2005;52:4028–38.

ASSOCIATION OF ANTI–�-ACTININ ANTIBODIES WITH LUPUS NEPHRITIS 2531

20. Sabbaga J, Line SR, Potocnjak P, Madaio MP. A murine nephri-togenic monoclonal anti-DNA autoantibody binds directly tomouse laminin, the major non-collagenous protein component ofthe glomerular basement membrane. Eur J Immunol 1989;19:137–43.

21. Faaber P, Rijke TP, van de Putte LB, Capel PL, Berden JH.Cross-reactivity of human and murine anti-DNA antibodies withheparan sulfate: the major glycosaminoglycan in glomerular base-ment membranes. J Clin Invest 1986;77:1824–30.

22. Atta MS, Powell RJ, Hopkinson ND, Todd I. Human anti-fibronectin antibodies in systemic lupus erythematosus: occur-rence and antigenic specificity. Clin Exp Immunol 1994;96:20–5.

23. Otey CA, Carpen O. �-actinin revisited: a fresh look at an oldplayer. Cell Motil Cytoskeleton 2004;58:104–11.

24. Mostoslavsky G, Fischel R, Yachimovich N, Yarknoni Y, Rosen-mann E, Monestier M, et al. Lupus anti-DNA autoantibodiescross-react with a glomerular structural protein: a case for tissueinjury by molecular mimicry. Eur J Immunol 2001;31:1221–7.

25. Deocharan B, Qing X, Lichauco J, Putterman C. �-actinin is across-reactive renal target for pathogenic anti-DNA antibodies.J Immunol 2002;168:3072–8.

26. Zhao Z, Weinstein E, Tuzova M, Davidson A, Mundel P, Maram-bio P, et al. Cross-reactivity of human lupus anti-DNA antibodieswith �-actinin and nephritogenic potential. Arthritis Rheum 2005;52:522–30.

27. Mason LJ, Ravirajan CT, Rahman A, Putterman C, Isenberg DA.Is �-actinin a target for anti-DNA antibodies in lupus nephritis?Arthritis Rheum 2004;50:866–70.

28. Tan EM, Cohen AS, Fries JF, Masi AT, McShane DJ, RothfieldNF, et al. The 1982 revised criteria for the classification of systemiclupus erythematosus. Arthritis Rheum 1982;25:1271–7.

29. Bombardier C, Gladman DD, Urowitz MB, Caron D, Chang CH,and the Committee on Prognosis Studies in SLE. Derivation of theSLEDAI: a disease activity index for lupus patients. ArthritisRheum 1992;35:630–40.

30. Gladman DD, Urowitz MB, Kogal A, Hallett D. Accuratelydescribing changes in disease activity in systemic lupus erythema-tosus. J Rheumatol 2000;27:377–9.

31. Petri M, Genovese M, Engle E, Hochberg M. Definition, inci-dence and clinical description of flare in systemic lupus erythem-atosus: a prospective cohort study. Arthritis Rheum 1991;34:937–44.

32. Weening JJ, D’Agati VD, Schwartz MM, Seshan SV, Alpers CE,Appel GB, et al, on behalf of the International Society ofNephrology and Renal Pathology Society Working Group on theclassification of lupus nephritis. The classification of glomerulo-nephritis in systemic lupus erythematosus revisited. J Am SocNephrol 2004;15:241–50.

33. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF,Cooper NS, et al. The American Rheumatism Association 1987revised criteria for the classification of rheumatoid arthritis.Arthritis Rheum 1988;31:315–24.

34. Vitali C, Bombardieri S, Jonsson R, Moutsopoulos HM, Alex-ander EL, Carsons SE, et al, and the European Study Group onClassification Criteria for Sjogren’s Syndrome. Classification cri-teria for Sjogren’s syndrome: a revised version of the Europeancriteria proposed by the American-European Consensus Group.Ann Rheum Dis 2002;61:554–8.

35. Goules A, Masouridi S, Tzioufas AG, Ioannidis JP, Skopouli FN,

Moutsopoulos HM. Clinically significant and biopsy-documentedrenal involvement in primary Sjogren’s syndrome. Medicine (Bal-timore) 2000;79:241–9.

36. Dueymes M, Barrier J, Besancenot JF, Cledes J, Conri C, Dien G,et al. Relationship of interleukin-4 to isotypic distribution ofanti-double-stranded DNA antibodies in systemic lupus erythem-atosus. Int Arch Allergy Immunol 1993;101:408–15.

37. Smeenk R, van der Lelij G, Aarden L. Avidity of antibodies todsDNA: comparison of IFT on Crithidia luciliae, Farr assay andPEG assay. J Immunol 1982;128:73–8.

38. Budd R, Hughes RG, Taylor D, Isenberg DA. New anti-dsDNAenzyme immunoassay [abstract]. Autoimm Rev 2004;3 Suppl 2:82.

39. Croquefer S, Renaudineau Y, Jousse S, Gueguen P, Ansart S,Saraux A, et al. The anti-alpha-actinin test completes anti-DNAdetermination in systemic lupus erythematosus. Ann N Y Acad Sci2005;1050:170–5.

40. Youinou P, Casburn-Budd R, Paolozzi L, Lamour A, Keusch J, LeGoff P, et al. Tiopronine-induced reduction of rheumatoid factorfunctional affinity and asialylated IgG in patients with rheumatoidarthritis. Clin Exp Rheumatol 1992;10:333–8.

41. Foster MH, Kelley VR. Lupus nephritis: update on pathogenesisand disease mechanisms. Semin Nephrol 1999;19:173–81.

42. Arbuckle MR, McClain MT, Rubertone MV, Scofield RH, DennisGJ, James JA, et al. Development of autoantibodies before theclinical onset of systemic lupus erythematosus. N Engl J Med2003;349:1526–33.

43. Eilat D. Cross-reactions of anti-DNA antibodies and the centraldogma of lupus nephritis. Immunol Today 1985;6:123–7.

44. Chowdhry IA, Kowal C, Hardin J, Zhou Z, Diamond B. Autoan-tibodies that bind glomeruli: cross-reactivity with bacterial antigen.Arthritis Rheum 2005;52:2403–10.

45. Qureshi F, Yang Y, Jacques SM, Johnson MP, Naparstek Y,Ulmansky R, et al. Anti-DNA antibodies cross-reacting withlaminin inhibit trophoblast attachment and migration: implicationfor recurrent pregnancy loss in SLE patients. Am J ReprodImmunol 2000;44:136–42.

46. Kaplan JM, Kim SH, North KN, Rennke H, Correia LA, TongHQ, et al. Mutations in ACTN4, encoding �-actinin-4, causefamilial focal segmental glomerulosclerosis. Nat Genet 2000;24:251–6.

47. Michaud JL, Lemieux LI, Dube M, Vanderhyden BC, RobertsonSJ, Kennedy CR. Focal and segmental glomerulosclerosis in micewith podocyte-specific expression of mutant �-actinin-4. J Am SocNephrol 2003;14:1200–11.

48. Hsu DK, Guo Y, Alberts GF, Peifley KA, Winkles JA. Fibroblastgrowth factor-1 inducible gene FR-17 encodes a non muscle�-actinin isoform. J Cell Physiol 1996;167:261–8.

49. Shlomchick M, Mascelli M, Shan H, Radic MZ, Pisetsky D,Marshak-Rothstein A, et al. Anti-DNA antibodies from auto-immune mice arise by clonal expansion and somatic mutation. JExp Med 1991;171:265–97.

50. Conger JD, Pike BL, Nossal GJ. Clonal analysis of anti-DNArepertoire of murine B lymphocytes. Proc Natl Acad Sci U S A1987;84:2931–5.

51. Shoenfeld Y. The idiotypic network in autoimmunity: antibodiesthat bind antibodies that bind antibodies. Nat Med 2004;10:17–8.

52. Shoenfeld Y. Anti-DNA idiotypes: from induction of disease tonovel therapeutical approaches. Immunol Lett 2005;100:73–7.

2532 RENAUDINEAU ET AL