Endogenous circulating DNA in systemic lupus erythematosus Occurrence as multimeric complexes bound to histone

(1)

Endogenous circulating DNA in systemic lupus

erythematosus. Occurrence as multimeric

complexes bound to histone.

P M Rumore, C R Steinman

J Clin Invest.

1990;

86(1)

:69-74.

https://doi.org/10.1172/JCI114716

.

Little is known about endogenous systemic lupus erythematosus (SLE) plasma DNA even

though it is the presumed precursor of DNA-containing immune complexes, thought to play

a central role in lupus glomerulonephritis. DNA purified from SLE plasma formed discrete

bands, corresponding to sizes of about 150-200, 400, 600, and 800 bp, closely resembling

the characteristic 200 bp "ladder" found with oligonucleosomal (ON) DNA. By radiolabeling

DNA while in whole plasma, the very small amounts present could be further characterized.

All of 24 such specimens formed two or more discrete bands on 6% PAGE. Detergent

treatment of plasma resulted in a DNA migration pattern similar to that of purified DNA,

suggesting disruption of DNA-protein complexes. DNA purified from authentic ON and

detergent-treated ON behaved similarly. A significant portion of DNA, labeled in SLE

plasma could be specifically immunoprecipitated with monoclonal antihistone antibody as

was the case with ON. These immunoprecipitates, when redissolved, exhibited the

expected size distribution upon PAGE. It is concluded that DNA in SLE plasma occurs as a

series of multimeric complexes, at least a portion of which is noncovalently bound to

histone. These results are consistent with an ON-like structure for SLE plasma DNA as had

been suggested by theoretical considerations and may have important implications for its

immunologic behavior in SLE and perhaps other disorders.

Research Article

(2)

Endogenous Circulating

DNA in

Systemic

Lupus

Erythematosus

Occurrence

as

Muftimeric Complexes Bound to Histone

Peter M. Rumore and Charles R. Steinman

The Department ofMedicine, Division ofRheumatology, TheState UniversityofNew York,

Health Science CenteratBrooklyn, Brooklyn,New York11203

Abstract

Little is known about endogenous systemic lupus erythema-tosus(SLE) plasma DNA even though it is the presumed pre-cursorof DNA-containing immunecomplexes, thought to play a central role in lupusglomerulonephritis.DNApurified from SLE plasma formed discrete bands,correspondingtosizes of about150-200, 400, 600, and 800 bp, closely resembling the characteristic 200 bp "ladder" found with oligonucleosomal

(ON)DNA.Byradiolabeling DNA while in whole plasma, the verysmall amounts present could be further characterized.All

of 24 such specimens formed two or more discrete bands on 6%

PAGE.Detergent treatment of plasma resulted in a DNA mi-gration pattern similar to that of purified DNA, suggesting disruption of DNA-protein complexes. DNA purified from

au-thentic ONand detergent-treated ON behaved similarly. A significant portion of DNA, labeled in SLE plasma could be

specifically immunoprecipitated with monoclonal antihistone antibodyas was the case with ON. Theseimmunoprecipitates,

when redissolved,exhibitedthe expected sizedistributionupon

PAGE. It is concluded thatDNA inSLEplasma occurs as a

seriesofmultimericcomplexes,at leastaportionof whichis noncovalently boundto histone. These results are consistent withanON-likestructurefor SLE plasma DNAashad been

suggested bytheoretical considerations and may have

impor-tant implications for its immunologic behavior in SLE and

perhapsotherdisorders.(J.Clin.Invest.1990.86:69-74.)Key words: nucleosome* immunecomplexes*autoimmunity-

glo-merulonephritis*apoptosis

Introduction

DNA-containing immune complexes are believedto have a

central causal role in lupus glomerulonephritis (1). The

proxi-matesourceof theDNA componentof thesecomplexes,

pre-sumedto be

plasma

DNA,has been difficulttostudy,inpart

because of the low concentrationspresent in most SLE

speci-mens(2-4).

Previous studieshave shownSLEplasmaDNAtobeoflow

molecular weight, largely double-stranded, but with

single-Apreliminary report of this work has appeared in abstract form in 1988.(Clin.Res.36:537a.)

Addressreprint requeststo Dr.Steinman, Division of Rheumatol-ogy, TheState University of NewYork, Health Science Centerat

Brooklyn, Box 42, 450 Clarkson Avenue, Brooklyn,NY11203.

Receivedfor publication15September1988 and in revisedform18

September1989.

stranded regions, and to contain nucleotide base sequences

most or allofwhich are presentin thehuman genome (3-6). Thepresent resultsindicate thatDNAin SLEplasma

ex-hibits structural characteristics similartothoseof

oligonucleo-somes as would be consistent with its having been derived fromthesemacromolecular complexes with retentionof some

oftheir structural features. Suchanoligonucleosomal

(ON)'-like structure might have important implications for its appar-ent role in SLE,

particularly

with regardtoits immunologic behavior andtoits cellular

origins.

Methods

Patients andplasma specimens. Patients with SLE were selected for havingpersistentlycirculatingimmunoprecipitabledsDNAassociated withactive systemic vasculitis and/or CNS involvement as described (7). All fulfilled American Rheumatism Associationdiagnostic criteria

for SLE (8). Plasmawashandledaspreviouslydescribed (7) and stored below -20'C. Storage of these EDTA-containingspecimenshadno

discernable effectonthe results obtained. Plasmaspecimenswere in-activatedat

560C

for 30 min anddialyzed overnightin 10 mM Tris-HC1, 1mMEDTA,pH 8.0 (TE).Omissionof heat inactivation didnot

alter the results. Plasma DNAconcentrationswerebetween 20 and 300 ng/ml (9). None exhibitedincreasedanti-dsDNA(10),ashasgenerally

been foundtobe thecasein suchpatients (7).

DNAandoligonucleosomes. Plasma DNA waspurifiedby a modi-fication ofa previousmethod that was notsizeselective and gave 30-80% recovery (4). Plasma,100Ml wasmade0.1MinTris-HC1,pH 8.0, 20 mM in EDTA, 1 Minsodiumperchlorate,and 1% in SDS. ProteinaseK(SigmaChemical Co., St.Louis, MO), 100

Mg/ml,

was

thenadded and the mixture incubatedat

550C

for 2 h. Itwasthen extracted sequentially, as described elsewhere (11), with phenol, phenol-chloroformandchloroform, precipitatedwith ethanol and dis-solved in 50ju of TE. HumanplacentalDNA(SigmaChemicalCo.) wasdenaturedat0.5 mg/ml inTEbyboilingfor 10 min followed by

rapidcooling. DNAasedigestionof radiolabeledplasmaorONwas

with DNAase1,50ug/ml,at

370C

for 30 min.Incontrols, the DNAase

I washeat inactivatedat 100ICfor 10 min. ONwerepreparedfrom chickerythrocytes (Whittaker Bioproducts,Walkersville,MD), by

di-gestionofnuclei with micrococcal nuclease and storedat-20'C in the presence ofphenylmethylsulfonylfluorideasaprotease inhibitor(12).

Anti-DNA reactionswerecarried outwith antisera obtained from threepatientswithSLE,andwereallcapableofprecipitatingdsDNA (9). 10 glof ON or DNA in plasma wereincubated with 20 Ml of antiserumat

370C

for 30 min and thenat

40C

for 1 h,followedby centrifugationat13,600g for 15 min. 20Mlof supernatant, treatedas

beforewithSDS,werethen examinedbyPAGE. Controlsemployed

antiserum that had been absorbed with 25

Mg/ml

each of native and heat-denatured humanplacentalDNAat371Cfor 30 min and

40C

overnight.

RadiolabelingandPAGE.PurifiedDNAwasradiolabeledby nick-translationas

described

elsewhere(I 1), using

[a-32PldCTP

from New

1. Abbreviations usedinthis paper:ON, oligonucleosome; TE,

Tris-EDTAbuffer.

CirculatingDNA inSystemicLupusErythematosus 69 J.Clin. Invest.

0021-9738/90/07/0069/06 $2.00

(3)

England Nuclear(Boston, MA), 2-3,000 Ci/mM.DNAin dialyzed plasma and inON was labeled in exactly the same manner except that DNAasedigestion could be omitted for plasma, apparently because it issufficiently nicked in vivo. PAGE was on 6% polyacrylamide in 89 mMTris, 89mMboricacid,2 mMEDTA, pH8.0 (TBE) (I 1).Ina typicalexperiment, each sample lane contained theDNAderivedfrom 10 ploriginal plasma.DNAsize standards consisted of a Hae III digest of phi-XDNA(NewEnglandBiolabs, Beverly, MA). Radiolabeling of standards waseither by nick translation orby 5'-end-labeling (11). Whereindicated, labeledspecimens were treated with 1% SDS for> 10 minat20'C. Each sample lane contained 10,lof a 100-Mul labeling reaction mixturecontaining80

_Al

of plasmaorON.Autoradiography was at4VC on Kodak X-Omat AR film (Eastman Kodak Co., Roches-ter,NY). Exposure timeswerevariedtoallowoptimum resolution of bandssothat bandintensitiesare notnecessarily comparable among different experiments.

Experiments toexamine themobilities ofplasmaDNAbands upon re-electrophoresisweredone after electroelutionof individual excised bandsasdescribed elsewhere (1 1). The eluatewasthenreconcentrated by vacuumdialysisusingcollodion bags (Schleicher and Schuell, Inc., Keene,NH),following thedirections of the supplier.

Antihistone experiments.Amonoclonal antihistoneantibody

reac-tive with atrypsin-sensitive epitope accessible in oligonucleosomes and,ascontrol,amonoclonalantiacetylcholinesteraseantibody of the

same isotype(catalogue No. MABO52 and MAB304, respectively)

were purchased from ChemiconInternational, Inc. (El Segundo, CA). Thereactivity of the antihistoneantibodywithatrypsin-sensitive epi-topeonONwasconfirmed inpreliminary experimentswhere labeled ONwereincubatedat37°C for 30 min with 10or 100Mg/mltrypsin (Sigma) in the presence orabsence ofatwofold excessof soybean

trypsininhibitor(Sigma).

Antihistoneprecipitation experiments wereperformedon

dupli-catespecimens of five SLE plasmas andanONpreparation.Amixture

containing 10 MlofplasmacontaininglabeledDNA(generally

con-tainingbetween50,000and150,000cpm)ornucleosomal DNA,5Ml of either the monoclonal antihistoneortheanticholinesterase control antibody, 5 ul of 0.15Msodium chloride, 0.01 Msodium phosphate, pH 7.2(PBS) and Ig1ofsalmon sperm DNA(Sigma),10 mg/ml in TE,

wasincubated at 37°C for 45 min. Then 5 Ml of normalmouse serum

(Sigma),diluted 1:5 in PBSwasaddedand, afterincubatingat37°Cfor

15min, 15,ulof rabbit anti-mouse IgG(Sigma), 2 mg/ml in PBS,was addedfollowed by incubationat37°C for 60 min andat4°C

over-night.Themixturewasthencentrifugedat13,600 g for 15 min, the supernatantwasremoved, and theprecipitatewashedtwice with 20 Ml PBS. 50 MlofTEcontaining 1% SDSwasthen addedtothewashed

precipitateandthemixture incubatedat100°Cfor 1 hduringwhichit dissolved. Theduplicatesetsofspecimenswerethenexamined both by

electrophoresison 6% polyacrylamide in 1% SDS and TBE and by TCAprecipitation andcounting in ascintillation counter(Packard

Tri-Carb; model 300C) with a background of 29 cpm, which was subtractedfrom theresultingcounts. Anadditionalsetof control

ex-perimentswasperformed usingthesameprocedure andstarting with thesamefive SLE plasmas andpreparationof ON but with exogenous labeledDNAin place oflabeled endogenous plasma DNA. This

exoge-nousDNAconsisted ofelectrophoretically purified 190-310 bp Hae IIIfiragmentsofphi-XDNAthat hadbeen5'-end-labeled to an

ap-proximatespecificactivityof 1.8X107cpm/Mg.

Results

DNA

_purifiedfrom

SLE

plasma

exhibits

multiple discrete

size

distributions.

DNA, first

purified

from whole SLE plasma and then

radiolabeled,

generally demonstrated four bands on

PAGE withsizes correspondingto - 150-200, 400, 600, and

800

bp

(Fig.

1). Withsome

specimens

the

separation

between

the 600- and 800-bp bandswas indistinct so thatthe two

blended

together.

Mostorall ofthe

radioactivity appeared

to

603 _l

31 0 - _{_/_}

194-Figure1. DNA extractedfrom arepresentative group of12SLE plasmaspecimens,labeled by nick-translation and separated by PAGE.At the left are locations of DNA size markers in base pairs.

Inmostcases the bandsmigrating at - 150-200 bpis the strongest.

Bands at- 400, 600,and 800 bp are alsovisible although the sepa-ration between the 600- and 800-bp bands is frequently indistinct.

be associated withoneof these bands. Small amountsoflow molecularweightradioactivity were variably seen atthe bot-tomsof some lanes andrepresentedmolecules - 50bp longor

less. Mostoftheradioactivitywasusuallyin the 150-200-bp

bandwithprogressivelyless in each of the slower bands. All of 20 DNAspecimens that were examined were similar.

DNA radiolabeled in

plasma forms

discrete bands on PAGE. PAGE of 24 DNA specimens labeled in whole plasma

consistently revealedtwo prominent bands on

autoradiogra-phy(Fig. 2A). Aslower, faint bandwasalso

frequently

seen. Bands ofother mobilities were also seen, but infrequently. Autoradiographic intensity was usually higher for the faster bands.

Avariableamount ofradioactivity failedto enterthegel.

Treatmentwith SDS largelyeliminated thisbut alsochanged

the mobilities of the bands, as willbe discussed. To confirm

thatthe labeledmaterialwas in fact DNA, itwasshownthat, inasubgroupof specimens, digestion withDNAaseI, butnot

inactivated DNAaseI,eliminatedtheradioactive bands. Also, incubation with each of three different anti-dsDNA-contain-ing antisera removed>90% oftheradioactivity from the gel. Priorabsorption ofthe antisera with DNAabolished this

ac-tivity.

ThefastestDNA band inplasma,whichalso wasthe most

intense, migrated moreslowlythandidthefastest band inthe DNA extracted from plasma (Fig. 2 B), consistent with the

interpretation that, in plasma, the migration of thisDNA

spe-cies had been retarded, presumably by

binding

to

protein.

Purified DNA that had first been radiolabeled while in

whole plasma and then extracted gave a bandingpattern on

PAGE similar to that ofDNA labeled after extraction from plasma. Thus, each of the molecular species seen in DNA

labeled after purification from plasma was accessible to the labelingreagentsbefore extraction. Also,nomolecular

species

was labeled in plasma that was not also labeledafter

extrac-tion.Toexplore the effect ofplasma

protein

binding

onthe

gel

mobility ofDNA, an attempt was made to add purified,

la-beled plasma DNA back to unlabeled plasma. This resulted

(4)

electro-A.

C,,

OZ

U)a z

z cln a

B.

SDS

P

S

F

S

F

Figure 2. Autoradiographicbandingpatternof SLE plasma DNAon

PAGE.Specimens fromthree patientsareshown in each of the three

panels, A,Band C. DNA marker locationsareindicatedatthe leftas

inFig. 1. (A) Whole plasma containing labeled DNA, 10g perwell. Bandsmigratingnearthe603- and 194-bp markerswereuniformly

observed.Also,abandjust ahead of the 872-bp markerwas

occa-sionallyseen asinthespecimenatthe far left.(B) DNA extracted fromplasma and thenradiolabeled, similartoFig. 1and shown here

forcomparison. (C) DNA in plasma, labeledasbefore but with 1% SDSadded. Thebandingpatternisgenerally similartothatseen

with DNAextracted fromplasma. Autoradiographic signals extend-ing irregularly toward thebottoms of the lanesareartifactual.

static interaction with plasma components. Incubation at

370C for 30min did notalter this. A similar result was

ob-tained when purified plasmaDNAwasreplaced byalabeled Hae IIIdigest of phi-X DNA.

SDStreatment.Treatment ofplasmawith SDS resultedin

aDNA bandingpatternthatwassimilar tothat of DNA ex-tracted fromplasma (Fig.2C).Thissuggestedthat SDS treat-ment largely released plasma DNA from association with

re-tarding molecules and allowed ittomigratein plasma essen-tially as free DNA. The mobility of uncomplexed control DNA, added toSDS-treated plasma, migrated no differently from DNA inbuffer alone. SDStreatmentgave more homoge-neous DNA bands in plasma than were seen with purified DNA, probably because of damage during purification.

Comparison with ON. Radiolabeled ON behaved on PAGEasshown inFig. 3 A. Attemptstoexamine themobility of labeled ON addedtoplasmaresulted in mostorall of the

addedradiolabel failingto enter thegelandremainingas

ag-gregatesattheorigin. Insomeexperimentsafaint band could

be observedto enterthegelbut thiswasnotreproducible.This

observationwassimilartothat made after addition ofpurified labeled DNAtoplasma, just described, and is assumed alsoto

be dueto aggregateformationwithplasma proteins. Replacing plasma with serum, incubation ofplasma-ON mixtures at

370C foraslongas 16 h andpredigestion of ON withtrypsin allfailedtoalter this result. Inbuffer,ONgaveaseries of bands migratingmoreslowlythan did DNA inplasma. Partial

pro-teolysis with trypsin increased their mobility, but they still

Figure 3. Autoradiographicpatternsof DNAonPAGE. (A) Behavior

of chick ON containing labeledDNAonPAGE. As indicatedatthe top,thelanescontain, from lefttoright, ON inbuffer, ON treated with SDS and DNA extracted from ON and then labeled. The multi-plebandingpatternof ON in buffer is similartothat of DNA in plasma, but slower,atleast for the fast band. Treatment of ON with SDS resultsinapatternsimilartothat ofDNAextracted from ON which shows the expected 200-bp ladder and resembles thatseenin

Fig. 1B forpurified plasma DNA. (B) Behavior inSDS of individual

plasma DNA bands. Lane P contains labeled DNA in plasma (10IL), nottreated with SDS. Thenext twolanes, S andF,contain the

ex-cised, electroeluted slow and fast bands, respectively, again without SDS. Thenexttwolanes, labeled SDS, contain thesesameslow and fast bands butafterSDStreatment.The slowbandyieldeda

pre-dominant bandat400bp andafainter bandat200bp, identical in mobilitytothe single bandseenafter SDStreatmentof the fast band, showninthe last laneattheright. Each S and F lane contains elec-troeluted and reconcentratedDNAobtained from 27gloforiginal plasma.

remained slower thanDNA inplasma whosemobilitywasnot

consistently altered by trypsin. Treatment of ON with SDS

resulted in DNA bands that migrated almost identicallywith

bands of labeled, purified ONDNA and whichexhibited the

200-bp ladder typical of ON (13). Thus, both theSDS-treated

ON and their extracted DNA gave banding patterns that

closely resembled those of plasmaDNA.

Studiesofthe isolatedplasmaDNA bands. DNApurified

from material recovered after electroelution ( 11) of the faster of the two majorDNA bands in plasma showed it to be

150-200 bp in size. Similar experiments with the slower, fainterDNAbandrepeatedly failedtoyield enoughDNAfor

visualization. However,as seen in Fig. 3 B, treatment of the separately isolatedfast and slowbands withSDS,followedby

repeat electrophoresis, resulted in the fast band migratingat

150-200 bp, similar to its purified DNA. The slow plasma band yielded a major band at 400 bp and a faint band at

150-200 bp.

Detection of histone bound toDNA inplasma. In an

at-temptatdirect identification of histones in the fast DNA band

in plasma, the latter was examined, after electroelution, by

A

1 B

C

-I(F.

872

603 -

310-194

i

872 -603

-310

1

(5)

194-SDS-PAGE (12). Protein staining failed to reveal a typical

histonebandingpattern, although these would probably have beenobscured by the largeamountof plasma proteinpresent. In an alternate approach, five radiolabeled

DNA-contain-ingplasmaspecimenswereimmunoprecipitatedwith either a monoclonal antihistone or with a control monoclonal

anti-body.Theantihistone-produced precipitates containedtwo- to

eightfoldmoreTCA-precipitableradioactivity (range 1.7-16%

of the total TCA radioactivityineachspecimen)thandidthe controls (Table I), a statistically significant difference (P

< 0.05). As already noted, attempts to add labeled ON to

plasma consistently yielded mainly large aggregates,

prevent-ing itsuse as apositivecontrol. However, ONinbuffergave

immunoprecipitation results similar to those seen with the plasmaspecimens.

Examination ofthe redissolved immunoprecipitates by

electrophoresis (Fig. 4) demonstrated,inthosethat gave a visi-ble

signal,

DNAbandscorrespondinginsize to about 200and 400bpasexpectedwhereaslittleor no autoradiographicsignal

was seen in thecorrespondingareasforcontrolprecipitates. In

addition a faint DNA band, correspondingto - 50-100 bp

was seen bothinthe plasmaspecimensas well as in the

prepa-rationof ON. Finally,the absence of a visible 400-bp band in

this

preparation

of ON suggested thatit may have been

par-tially degraded.

Toexaminethepossibilitythatfree histones, which might

have been present in plasma, could have nonspecifically boundtolabeledDNAthereby accounting for its precipitation by antihistone, another set of simultaneous control

experi-ments was performed. Purified, radiolabeled DNA,

- 200-300 bp long,wasaddedtoeach

of

thesameplasma and

ON specimens, this time without radiolabeling the endoge-nous plasma-derived DNA so that all the radioactivity was

containedin theadded "naked" DNA. Thespecimens were then

subjected

tothe same

immunoprecipitation

procedure. Therewas no significant difference(P > 0.5) betweenthe

ra-dioactivityin theantihistone (mean=909cpm or 0.1 0%ofthe

866,180 cpm added) and control

precipitates

(mean = 794

cpm)sothat

nonspecific

binding ofDNAby free plasma

his-toneisanunlikely

explanation

for plasmaDNA

precipitation

by antihistone.

Discussion

Thepresent dataindicate thatDNApurified fromSLEplasma

does not exist as a continuous distribution of size species as

TableI.

TCAprecipitablecpm

Plasma Antihistone Control

specimen Total precipitate precipitate

Plasma1 103,302 1,818 883 Plasma 2 169,664 9,068 1,032 Plasma3 43,803 1,729 743 Plasma4 57,276 9,199 1,134 Plasma5 77,167 3,536 1,504

Mean(±SD) 5,070(±3,779) 1,059(±289)

ON 1.7 X106 67,597 434

1

2

3

4

5 NS

H C

HC

603 -194

-Figure 4. Radiolabeled plasmaDNAafter precipitation with

antihis-tone(H)oranticholinesterase control (C) antibodies, after resolubili-zationat 100°C in SDS and separatedonSDS-PAGE.Specimens 1-5arederivedfrom SLE plasmas and NS is from chick ON. The size markersareheat-denaturedHae IIIphi-Xfragments.Lanes at

theleft edge of the gelareslightlydistorted dueto adrying artifact.

Ineach antihistone precipitate a radioactive signalwasvisible on the

originalautoradiographand exceeded inintensitythe controlsignal

which was generally not visible. These results were consistent with thoseobtained fromdeterminingtheTCA-precipitable radioactivity

by scintillation counting, shown in TableI.

would be expected ifit consisted ofrandomly fragmented

DNA molecules. Instead, it occurs ina series of distinct size distributions whose gel mobilitiesare consistent with their being derived from multimeric molecules witha unit size of

- 200 bp. The largestspecies clearlyseen was - 800bp,but

small amountsof larger forms

might

nothavebeendetected.

Theseapparent multimersappear to account formost ofthe

plasmaDNA. Theeffects of SDStreatmentand of

immuno-precipitation

suggest further that plasmaDNA is

complexed

withprotein that consists, atleastinpart,of histone.

Failureof antihistoneto reactwithahigher proportion of plasmaDNAcouldhaveresultedfrom eitherloss or

inaccessi-bility

ofthe

trypsin-sensitive

ONepitopesknown to bereactive with the monoclonal antihistone antibody

(14).

This could have been due, for example, either to prior proteolysis or,

alternatively,

to blockage ofthe monoclonal antihistone by previously bound endogenousantihistones. Neither ofthese

explanations

wasexaminedhere.

These molecular characteristics of plasmaDNA aresimilar

to those of ON, the structural units of

chromatin,

where, in

eachmonomer,bound histone

partially

protects about 200bp of associated DNAfrom nuclease

digestion

in vitro

(13, 15).

The consequence is that DNA extracted from partially

nu-clease-digested

chromatin exhibits a characteristic

200-bp

"ladder" on electrophoresis. However, mononucleosome

DNA can befurther

digested

sothat, after

purification,

addi-tionalmolecules

ranging

insizefrom 150-200

bp

are seen. Also, several size

species

inthe rangeof 50to 1 10bphave been

(6)

purified from plasma isstrikinglysimilar to thatfound forON, bothhere andinthe literature.

It is therefore concluded that the ON-like structure ob-served here for plasma DNAisconsistentwithits having been derived from cellular ON with atleast partialretention oftheir

structuralfeatures.

TheelectrophoreticbehaviorofDNAin the electroeluted

fastandslow plasma bandsfollowingtreatmentwithSDS sug-gests DNA contents consistent with ON-like monomers and

dimers, respectively. However, this could be directly

con-firmedwith purified DNA onlyforthefastband.

Whether theslow band infact contains DNA molecules of 400

bp depends

upon the

assumption

thatSDStreatmentof thecomplexesin theslow band resultsin DNA thatmigrates

asfree molecules, a

likely

but unprovedassumption. Itthus

remainspossiblethat theON-like mobilityofDNAin whole

plasma isdetermined by bindingtoplasmaproteins,although

this seems unlikely since, asnoted, that would leave

unex-plainedthemobilities ofDNA observed afterSDS treatment

of theisolated slowandfast bands.Itis emphasized, however, that identification of specific DNA bands seen in whole plasma with specific nucleosome-related structures, a large numberof which have been described(14-16), would be pre-mature.

Even though it appears likely that the present data are

explained byan ON-like structure ofplasma DNA, itis

ap-propriate to consider alternate explanations. Although

non-histone, DNA-binding proteinsare presentinthe nucleus(17) and couldperhapsgenerateplasma DNA-protein complexes, theyarepresent inmuch smalleramountsthanhistones and,

further,

cannot as

readily explain

the observedsize distribution of thepurifiedDNA. However,our currentunderstanding of these

proteins

is stilltoo

rudimentary

toeliminate this

possibil-ity with certainty. Similarly, neither plasmaDNA

binding

proteins (18) norbound anti-DNA

antibodies,

ifpresent,

could

easily

accountfortheobserved plasmaDNAsize distri-bution. Although anti-dsDNA antibodies

partially

protect DNAfrom

nucleases,

thelengthprotected is

only

about40 bp (19). Moreover, these

explanations

wouldnotaccountfor the reactivity of plasma DNAwith antihistone

antibody.

How-ever, theseconsiderations do notexcludethe

possibility

that DNA present in plasma, even in the form of ON-like

com-plexesaspostulated,

might

form additional

bonds,

either

spe-cific or

nonspecific,

with other

plasma

molecules. Inthis

re-gard, it is

emphasized

that most SLE

plasma

specimens

ex-hibiting

free DNA,asfound

here,

andashasbeenour

general

experience,

donotalso contain free anti-dsDNA detectable

by

binding

assays.

However,

thepresenceofbound

antibody,

un-detectable in this way

by

virtue ofexcess dsDNA

antigen,

cannotbe excluded bythese data

but,

as

noted,

wouldseem

unlikely

toaffect thepresent results.

Because the study patientswere selected forhaving large

amountsof plasmaDNAinassociation with vasculitisor CNS

involvement,

theextent towhich these resultscanbe general-izedis uncertain.However, severalpreviousreports(3, 5, 20, 21) ofDNA isolated from plasma of SLE patientswho had

apparently

not been selected in this manner found it to be between100 and 200

bp

insizeasis thecaseforthe

predomi-nant

species

described here and thussupportsthe

generality

of these

findings.

In

addition,

one of these reports describes a

300-bp

species

ofDNA fromserum ofoneSLE

patient (21).

Fromthe datapresented itseems

possible

thatthisDNA

spe-cies was the same as that considered to be - 400 bp in the

presentstudy since the 8% polyacrylamide gel used there pro-videdless precise size discrimination in this region than does the 6% gel used here. Failure in those studies to describe larger molecular species consistent with multimeric forms as de-scribed here could have resulted from other differences in tech-nique. For example, the use here of nick-translation, which incorporates the radiolabel throughout the DNA molecule,

would beexpected to enhancesensitivityof detectionof longer

molecules relative toshorterones, in contrast tomethods that incorporate the radiolabel only at chain termini, as employed in atleastonesuchstudy. Also, failureinthe presentstudyto recognize a 17.5-kb DNA molecule, as described in another report(22), cannot beattributedtoomission ofproteolysisas wassuggested toexplainthat result. It wasobservedhere, how-ever,that, without SDS, much of the DNA was in aggregates toolargeto enterthegel, as was also reportedinthatstudy. However,this could have been due to the low ionicstrengthof thegel buffer.Nonetheless,thepossibilitythat a small fraction of DNA of large size eluded detection in the present study cannotbe excluded.

The significance of recognizing an ON-like structure for SLE plasma DNA is severalfold. First, it is consistent with otherevidencesuggestingthat this DNA islargelyor entirely derivedfrom DNA in thehumannucleus (4, 23), rather than from a non-genomically integrated viral agent, for example, as has been described in at least one other clinical state (24).

Second, it hasbeen proposedthat

autoimmunity

in SLE re-sultsfrom animmunogen-driven mechanism, incontrast to a

lessspecific mechanismsuch aspolyclonalimmuneactivation

(25). One piece ofevidence for this isthe reported

concor-dance,in individual patients, ofantibodies directedat

mole-cules that occur in nature asmacromolecularcomplexes but

thatotherwise appear

quite

different structurally (25). Anti-bodiesto histonesand DNA havebeen reported to occurin

such "linkedsets,"thussuggestingthat ahistone-DNA

macro-molecular complex, presumably ON derived, mightact as

theircommon

autoimmunogen.

The present

finding

thatsuch

anextracellular histone-DNA complex does in fact occur in vivo, in SLE plasma, and, in addition, that it possesses the spectrum of DNA sizes expected of an ON-like complex is

consistent with suchan

immunogen-driven

mechanism.

In adifferentcontext,theoccurrenceof SLE plasmaDNA

in ON-like complexes

implies

potential

autoantibody-recogni-tionofnon-DNA components of thecomplex(13, 26), which could thereforecontributetoDNA-containing immune

com-plex formation

independently

of anti-DNAandwhich might meritfurther study.

Similarly, positively

charged sitesknown to bepresent on such non-DNA componentsofON(15)could

allowbindingto

negatively

charged moietiesin the glomerular basement membrane and mediate formation of

DNA-an-tiDNA immune complexesin situas hasbeenpostulatedto

occur

(1).

IfSLEplasmaDNA occursintheform ofadouble

super-coil ofDNA

tightly

wound about ahistonecore,asis thecase

for ON

(13, 15),

consequent structural constraints

might

be

expectedtohavean

important influence,

difficult to

predict,

onits

reactivity

with anti-DNA antibodies andon thebehavior ofimmune

complexes

so

formed,

asfor example,in

comple-mentactivation ortissue deposition, thatwouldrequire

fur-ther

investigation

ofthese

important

phenomena. Also, the

possibility shouldbeconsidered that antibodies might

(7)

entially react with both of the two adjacent dsDNA strands wound about the same core, either as two separate but contigu-ousbinding sites that might permit a form of monogamous

bivalentbinding (27)or, morespeculatively, that the separate

chains might form asingle, new epitopeif they were spaced

closelyenough.

ExtracellularON-likecomplexes generatedin vivo do not appear to have beenstudied previously. Therefore, assuming it is correct that SLE plasma DNA has such astructure,it might differ in unexpected ways fromONpreparedin vitro, chiefly from avian erythrocytes, which provide muchofour current

knowledgeaboutthese complexes (13).Suchdifferencescould beimmunologically important and,inaddition, might provide insight into mechanisms oftheir release into thecirculation

which is not nowunderstood.

In this last regard, a particularly significant question is whether the ON-like structureof SLE plasma DNA is gener-ated within an intact nucleus, a phenomenon that is a hall-mark of apoptotic (in contrast to necrotic) cell death, a

well-definedprocess that characterizesavariety of important

physiologic events, including some with immunologic rele-vance (28).

Itis uncertain whether ON generated during apoptosiscan escapephagocytosis, with which it is closely associated, and thusgain access to theextracellular fluid, at least on a large

enoughscalefor detectioninplasma. The present data would

be consistent withthat

possibility,

however, atleast in SLE.

Clearly, however,much moreinformation isneeded to

clarify

thesequestions.

Acknowledgments

WewishtothankMs. M.Langefor her technical assistanceand Dr. D.

Kaplanforacriticalreading of the manuscript.

This workwassupported inpartbygrantsfrom the National

Insti-tutesof Health, The SLE Foundation, Inc. andfromthe NewYork Chapter oftheArthritis Foundation.

References

1.Zvaifler, N.Z., andV. L. Wood, Jr. 1985.Etiologyand

patho-genesis of systemic lupus erythematosus.InTextbookof

Rheumatol-ogy. W. N.Kelly, E.D.Harris, Jr., S. Ruddy,andC.B.Sledge, editors.

W. B.SaundersCo.,Philadelphia,PA. 1042-1070.

2.Davis, G. L., Jr., and J. S. DavisIV. 1973. Detection of circulat-ingDNAbycounterimmunoelectrophoresis (CIE). ArthritisRheum. 16:52-58.

3. McCoubrey-Hoyer,A., T. B.Okarma, andH. R.Holman. 1984. Partialpurificationandcharacterization of plasmaDNAandits

rela-tiontodisease activityin SLE.Am.J.Med. 77:23-34.

4.Steinman,C.R. 1984.CirculatingDNAin systemic lupus

ery-thematosus. Isolation and Characterization. J. Clin. Invest. 73:832-841.

5.Sano,H., and C. Morimoto. 1982. DNAisolated from DNA/

anti-DNAantibodyimmune complexes in systemic lupus

erythema-tosusisrich inguanine-cytosinecontent.J.Immunol. 128:1341-1345. 6.Raptis,L., andH. A.Menard. 1980.Quantitationand

character-ization of plasmaDNAin normals andpatientswith systemic lupus erythematosus. J. Clin.Invest.66:1391-1399.

7. Steinman,C. R. 1979.CirculatingDNA insystemic lupus ery-thematosus. Associationwith centralnervoussysteminvolvement and

systemicvasculitis.Am.J.Med.67:429-435.

8.Tan E.M.,A. S.Cohen,J. F.Fries,A.T.Masi,D. J.McShane,

N. F.Rothfield,J.G.Schaller,N.Talal,andR. J. Winchester. 1982. The 1982 revised criteria for theclassification ofsystemic lupus

ery-thematosus. Arthritis Rheum. 25:1271-1277.

9.Steinman,C.R. 1982. Detection andsemiquantitationof DNA

by counterimmunoelectrophoresis (CIE).MethodsEnzymol. 84:187-193.

10.Steinman,C.R.,U.Deesomchok,andH.Spiera. 1976. Detec-tion of anti-DNAantibody using synthetic antigens.Characterization

andclinicalsignificanceofbindingof

poly(deoxyadenylate-deoxythy-midylate) byserum.J. Clin.Invest. 57:1330-1341.

11.Maniatis, T.,E. F. Frisch,andJ. Sambrook. 1982.Molecular Cloning.Alaboratorymanual. ColdSpring Harbor Laboratory,Cold

Spring Harbor,NY.545pp.

12. Goodman, M.,and A.Dahlberg. 1982.Electrophoresisof

nu-cleoproteins.In GelElectrophoresisofNucleicAcids,aPractical

Ap-proach.D.Rickwood and B. D.Hames,editors. IRL PressLtd., Ox-ford. 199-225.

13. Kornberg, R. D. 1977. Structure ofchromatin. Annu. Rev.

Biochem. 46:931-954.

14.Bakayev,V. V.,T.G.Bakayeva,andA. J.Varshavsky. 1977.

Nucleosomes andsubnucleosomes: Heterogeneity andcomposition.

Cell. 11:619-629.

15.McGhee,J.D.,and G.Felsenfeld. 1980. Nucleosomestructure. Annu.Rev. Biochem. 49:1115-1156.

16. Sollner-Webb, B., R. D. Camerini-Otero, and G. Felsenfeld.

1976.Chromatinstructure asprobed by nucleases andproteases: evi-dence for the central role of histones H3 and H4.Cell.9:179-193.

17.Lewin, B. 1983. Repressor isadimer that binds cooperativelyat eachoperator. In Genes. 2nd Edition. B. Lewin, editor. John Wiley

andSons,New York. 266-271.

18. Hoch,S., andE.McVey. 1977.Purification and

characteriza-tion of twomajor DNA-binding proteins inhumanserum. J. Biol.

Chem.252:1881-1887.

19. Emlen,W.R.,Ansari, and G. Burdick. 1984. DNA-anti-DNA

immunecomplexes: Antibody protection ofadiscreteDNAfragment

fromDNAase digestion in vitro. J. Clin. Invest.74:185-190. 20. Fournie, G. J. 1988. Circulating DNA andlupus nephritis. KidneyInt. 33:487-497.

21.Morimoto, C.,H.Sano,T.Abe,M.Homma, andA.D. Stein-berg. 1982. Correlation between clinical activity of systemic lupus erythematosus and theamountsof DNA inDNA/anti-DNA antibody immune complexes. J. Immunol. 139:1960-1965.

22.Rieber, M., C. E.Contreras,M.S.Rieber,and N. E.Bianco.

1986.NovelDNA-protein complex andalarge DNA in SLE cryopre-cipitates.Clin. Exp. Immunol. 66:61-67.

23.Li, J. Z.,andC. R.Steinman. 1989. PlasmaDNAin systemic lupus erythematosus.Arthritis Rheum. 32:726-733.

24. Hung,P.P.,J.C. Mao, C.M.Ling,and L. R.Overby. 1975. Hybridization ofDaneparticleDNA with the free plasmaDNA of hepatitis carriers. Nature (Lond.). 253:571-572.

25. Craft, J., and J. A. Hardin. 1987. Linkedsetsof antinuclear antibodies: What do they mean? J. Rheum. 14(Suppl.): 106-109.

26.Bustin, M. 1979. Immunological approachestochromatin and chromosome structure and function. Curr. TopicsMicrobiol.

Im-munol. 88:105-142.

27. Day,E. D.1972.Antibody reactions with multivalent antigens. AdvancedImmunochemistry.WilliamsandWilkins,Baltimore,MD.

311-356.

28.Wylie,A.H.1987.Apoptosis: Cell death in tissue regulation.J.