Naturally Occurring Agglutinins MaybeEncoded byDifferent ......Naturally OccurringAnti-i/l ColdAgglutinins MaybeEncoded byDifferent VH3GenesasWellasthe VH4.21 GeneSegment Leigh C

Naturally Occurring Anti-i/l Cold Agglutinins May be Encodedby Different VH3 Genes as Well as the VH4.21 Gene SegmentLeigh C. Jefferies, Catherine M. Carchidi, and Leslie E. SilbersteinHospital of the University of Pennsylvania, Department of Pathology and Laboratory Medicine, Philadelphia, Pennsylvania 19104

Abstract

In the current study, we wished to determine if the V regionsencoding the naturally occurring anti-i/I Cold Agglutinins(anti-i /I CA) differ from pathogenic anti-i /I CAthat are exclu-sively encoded by the VH4.21 gene. After EBVtransformationof B lymphocytes, we generated one anti-I secreting clone fromeach of two individuals; clone 4G (individual CM, PBL) andclone Spl (individual SC, spleen). Clone 4G expresses a VH3gene sequence that is 92% homologous to the germline geneWHG26. Clone Spl also expresses a VH3 gene that is 98%homologous to the fetally rearranged M85/20P1 gene. An-other clone, Sp2 (anti-i specificity), from individual SCis 98%homologous to the germline gene VH4.21. For correlation, westudied anti-i /I CA fractions purified from 15 normal sera andfound no or relatively small amounts of 9G4 (VH4.21 relatedidiotype) reactive IgM. Five cold agglutinin fractions containedlarge amounts of VH3-encoded IgM (compared to pooled nor-mal IgM) by virtue of their binding to modified protein Staph A(SPA), and absorption of three CA fractions with modifiedSPA specifically removed anti-i/I binding specificity entirely.Collectively, the data indicate that naturally occurring anti-i/ICA may be encoded to a large extent by non-VH4.21-relatedgenes, and that the VH4.21 gene is not uniquely required foranti-i/I specificity. (J. Clin. Invest. 1993.92:2821-2833.) Keywords: immunoglobulin - autoantibody * red blood cell * coldagglutinin * autoimmunity

Introduction

Autoantibodies with various specificities are frequently detect-able in the sera of healthy individuals and are referred to asbenign, natural, or naturally occurring. However, curiously,antibodies of the same specificity may also be associated withautoimmune disease (1, 2). For example, virtually all serafrom healthy individuals contain low concentrations of natu-rally occurring anti-i /I IgM anti-red blood cell (RBC) antibod-ies or cold agglutinins (CA)' that cause no apparent immune

Address correspondence and reprint requests to Leigh C. Jefferies,M.D., Transfusion Medicine, 6 Floor Founders Pavilion, Hospital ofthe University of Pennsylvania, 3400 Spruce St., Philadelphia, PA19104.

Receivedfor publication 5 February 1993 and in revisedform 12July 1993.

1. Abbreviations used in this paper: CA, cold agglutinins; HA, hemag-glutination; RBC, red blood cells; SPA, Staphylococcal protein A.

hemolysis. The pathogenic counterparts of these autoantibod-ies may cause severe immune hemolytic anemia and are gener-ally derived from monoclonal B cell expansions that may pro-gress to frank lymphoma (3). The anti-i and anti-I specificitiesof CA are broadly defined by their differential reactivities tocord (predominantly i-expressing) vs adult (predominantly I-expressing) RBCs. Normal adult sera contain a heterogeneouspopulation of naturally occurring CAthat almost always haveanti-I specificity, although rarely small amounts of anti-i CAmay also be detected. In contrast, pathogenic CA (e.g., derivedfrom B cell lymphoma) mayhave either anti-I or anti-i specific-ity (4).

In the current study, we have begun to investigate the rela-tionship between naturally occurring, nonpathogenic anti-i/ICAand pathogenic anti-i/I CAderived from clonal B cell ex-pansions. In recent years, several laboratories have demon-strated a highly restricted VH4 gene use among pathogenic CAfrom different individuals. The heavy chain variable geneVH4.2 1 (or a closely related gene) has been found in all patho-genic monoclonal CAstudied to date, suggesting that this genemay play an important role in defining the i/I specificity (5-9). Our study addressed the question whether the same re-stricted pattern of V gene usage by pathogenic CA might befound in naturally occurring CA. The simplest hypothesis, thatthe VH4.2 1 gene product is the principle determinant of anti-i /I CAactivity, would predict that this same gene would encodenaturally occurring CA with similar specificity. Alternatively,naturally occurring anti-i/I CAcould be encoded by VHgenesother than VH4.2 1. A schematic diagram illustrating these pos-sibilities is shown in Fig. 1.

To evaluate the genetic diversity of naturally occurring CA,we have established EBV-transformed B cell clones secretingantibodies with anti-i/I binding specificity from two individ-uals without evidence of immune hemolysis. VHand VL regionnucleotide sequence analyses of these B cell clones indicate thatVH genes other than VH4.21 may encode anti-i/I specificity.These results were subsequently correlated with additional sero-logical analyses of anti-i/I enriched CA fractions from normalsera to further understand the relationship between natural andpathogenic, monoclonal anti-i/I CA.

Methods

Generation of monoclonal anti-i/I secreting EBV-transformed B celllines. PBL were isolated by Ficoll-Hypaque gradient centrifugationfrom a healthy nonanemic individual (CM) without symptoms of coldagglutinin disease. Serum from this individual agglutinated untreatedhomologous red blood cells at 40C up to a titer of 1:16. PBL were

enriched 10% before EBV transformation by panning with murinemonoclonal anti-idiotypic antibodies previously described (10). Fourprevious attempts to establish B cell clones secreting natural CA fromnormal individuals had been unsuccessful by EBV transformationmethodology without panning. The anti-idiotypic antibodies used inthe panning step are light chain restricted and bind both benign and

Naturally Occurring Anti-i/I Cold Agglutinins 2821

J. Clin. Invest.© The American Society for Clinical Investigation, Inc.0021-9738/93/12/2821/13 $2.00Volume 92, December 1993, 2821-2833

I Idiopathic(Preneoplastic) Lymphoma I

4~~~~~~.2

H~~~~.

lonal Monoclonalanti-i/I

Monoclonalanti-i/I

Figure 1. The relationship between naturally occurring and patho-genic anti-i/I CA. The origin of pathogenic (e.g., monoclonal) CAderived from clonal B cell expansions has been previously investigated(6, 12, 18, 25).

pathologic CA. By Western blot analysis, the anti-idiotypic antibodiesbind isolated kappa light chains, and, therefore, did not bias the selec-tion of B lymphocytes expressing selected VH genes in these experi-ments. Polystyrene petri dishes were coated with a mixture of antibod-ies (2.0 gg/ml IV.8, 1.0 gg/ml IV.6, 1 ,ug/ml IV.2, and 1.0 Ag/mlIV.9) suspended in PBS (0.01 MNa P04 and 0.15 MNaCl, pH 7.4) at220C for 1 h. Unbound antibody was removed by gentle washing fourtimes using PBSwith 1%FBS. Lymphocytes were washed twice in PBS,resuspended in 10 ml 5% FBS in PBS (1051_06 cells/ml), and incu-bated in the petri dish for 1 h at 40C. Unbound cells were removed bygentle washing four times with 1% FBS/PBS. Adherent cells were re-

moved with flushing and scraping using 30 ml 1%FBS/PBS, pelleted,and resuspended in 5 ml ex vivo cloning media (Whittaker Bioprod-ucts, Walkersville, MD), 1%glutamine, 10% Pen Strep, 1 Ag/ml Mitoserum extender, 1 pg/ml PHA, and 10% FBS).

PBL from donor CMand splenic lymphocytes from a second non-

anemic individual (SC) were similarly enriched by panning with the ratmonoclonal anti-idiotypic antibody 9G4 (1 1). This reagent recognizesa conformation-dependent epitope that is highly associated withVH4.2 1 expressing autoantibodies to the i and I antigens.

EBV transformation of PBL and splenic lymphocytes was accom-

plished as previously published ( 12 ). After 3-4 wk, supernatant fromwells with confluent growth was added to 2-4% suspension of ficintreated adult RBC. Hemagglutination was assessed after a 30 min incu-bation at 4°C. Wells positive for hemagglutination were subcloned bylimiting dilution ( 13). After subcloning, supernatants from monoclo-nal cell lines grown in RPMI 1640 medium were retested for hemagglu-tination with both ficin-treated and untreated adult (I) and cord (i)group 0 RBCusing previously established methods ( 14). Serial dilu-tions of supernatant or CA purified from supernatant in PBS, pH 7.4,were performed to compare relative degrees of agglutination with adultand cord RBC.

Isolation of anti-i/I CA from normal sera. Serum was obtainedfrom healthy individuals whose serum agglutinated untreated adult (I)RBCat 40C at a dilution of 2 1:8. Cold reactive autoantibodies wereisolated by absorption at 40C to allogeneic RBCs, followed by heatelution. This procedure was performed by a cold/warm/cold incuba-tion method ( 15) that limits the amount of nonspecific protein in thefinal eluate. Split serum samples were absorbed with ficin-treatedgroup 0 adult (I) and cord (i) RBCsto obtain enriched anti-I and anti-ifractions, respectively. The CA fractions are termed "enriched" be-cause they have been isolated from sera using RBCsthat are enrichedfor either the I (adult RBC) or i (cord RBC) antigen. Anti-i vs anti-Irelative binding specificities of purified fractions were confirmed bystandard methods using both ficin-treated and untreated adult (I) andcord (i) cells ( 14).

Monoclonal antibodies. The IgG2a rat antibody 9G4 recognizes a

heavy chain associated idiotype highly associated with cold agglutinins

encoded by VH4.21 or a closely related VH4 heavy chain genes (6, 8,1 1). Cold agglutinins VOG(IgM, K) and CAP (IgM, K) have beenpreviously described (6). Antibody VOGwith anti-I specificity, is en-coded by a VH4 gene with 97% homology to the published VH4.21germline gene (16), and expresses the 9G4 idiotype. Antibody CAPwith anti-i specificity is encoded by a VH4 gene identical to the pub-lished VH4.2 1 germline sequence and expresses the 9G4 idiotype.

ELISA. Except where indicated, the following general proceduresand buffers were used: Microtiter plates for ELISA (Dynatech Labora-tory, Chantilly, VA) were coated (150 ,dl/well) with antibodies sus-pended in PBS, pH 7.4, and incubated overnight at 40C. All subse-quent incubations were for 1.5 h at 370C. All washing steps used threewashes with PBS containing 0.1% Tween 20. To decrease nonspecificbinding, wells were blocked with blocking/diluent buffer (0.5% nonfatdry milk and 0.0 1%Thimerosol in PBS, pH 7.4), 300 ul/well. Antibod-ies were diluted in blocking/diluent buffer ( 150 AL/well). All sampleswere tested in duplicate wells, and alkaline phosphatase-conjugatedantibodies were used at 1:1,000 dilution. Alkaline phosphatase activitywas measured with 5 mgp-nitrophenylphosphate (Sigma Immuno-chemicals, St. Louis, MO) per 40 ml of developing buffer (1 MTrisHCI, 0.5 mMZnCI2, 2 mMMgCl2, 0.02% NaN3, pH 8.2), 150,l/well,at 22°C for 30 min. Activity was quenched with 3 MNaOH, 50 ,ul perwell and optical density readings were measured at 405 nM(MR 700ELISA Reader; Dynatech Laboratories).

To detect IgM in cell culture supernatants, microtiter plates werecoated with rabbit anti-human IgM (Jackson Immunoresearch, WestGrove, PA) blocked and washed. After incubation with supernatants orstandard IgM, wells were incubated with mouse anti-human IgM(Zymed Laboratories, Inc., South San Francisco, CA), 1:1,000 dilu-tion, and bound human IgM was detected using alkaline phosphataseconjugated goat anti-mouse IgG and IgM (Jackson Immunoresearch).

To detect light chain isotype of cell culture supernatants or purifiedCA, microtiter plates were coated with rabbit anti-human IgM,blocked, and washed. Duplicate wells were incubated with cell culturesupernatants, or IgM kappa and IgM lambda (Tago, Burlington, CA)controls. After washing, paired wells were incubated with either alka-line phosphatase-conjugated rabbit anti-human kappa or anti-humanlambda (Sigma Immunochemicals) to detect bound human immuno-globulin.

Cell culture supernatants and purified CAwere tested for bindingby anti-idiotypic antibody 9G4 as follows: Microtiter plates werecoated with 2 ,g/ml rabbit anti-human IgM, blocked, washed andincubated with cell culture supernatant. After incubation and washing,rat monoclonal antibody 9G4 (1.25 gg/ml) was added. Either VOGor

CAP VH4.21 proteins (6) were used as a positive control. Alkalinephosphatase-conjugated goat anti-mouse IgG and IgM (H and L) was

used to detect bound rat immunoglobulin. To quantitate the relativeamounts of 9G4-positive IgM in purified CA, a standard curve was

established correlating the IgM concentration of a known VH4.2 1 ex-

pressing monoclonal IgM CAwith 9G4 reactivity. Subsequently, serialdilutions of CA fractions (known IgM concentration) were tested forbinding by 9G4, and the amount of 9G4 reactive IgM in CAfractionswas extrapolated from the standard curve.

To detect VH3 family-encoded IgM, plates were coated with rabbitanti-human IgM (2 Ag/ ml), subsequently blocked and coated in tripli-cate wells with serial dilutions of purified CA(known IgM concentra-tion), with equivalent concentrations of pooled polyclonal IgM (Cal-biochem Corp., San Diego, CA) and with serial dilutions of a positivecontrol VH3 IgM. After a 4-h incubation and washing, modified (Staph-ylococcal protein A (SPA) modified by iodination of tyrosines, biotin)(5.0 pg/ml) was added for a 2-h incubation at 22°C. Modified SPAhasbeen shown to bind 90% of VH3-encoded IgM via Fab-binding sitesthat are functionally and structurally distinct from the Fcy-bindingsites of SPA ( 17). Bound IgM was detected with alkaline phosphatase-labeled streptavidin. Modified SPAreactivity of CAfractions was com-

pared to that of pooled normal IgM (both tested at 0.25 ug/ml) toassess for enrichment of VH3-encoded IgM; we did not attempt to iso-

2822 L. C. Jefferies, C. M. Carchidi, and L. E. Silberstein

'ring Pathogenicanti-i/I

Naturally Occuranti-i/II

Polyclonal/Oligoclanti-i/I

late the total IgM fraction of each individual from whomthe naturallyoccurring CA fractions were purified.

Absorption of CA fractions with modified SPA. Absorption of CAfractions with modified SPAwas used to determine if depletion of VH3IgM in CA fractions would diminish or abolish anti-I binding specific-ity. To this end, wells were coated overnight (40C) with modified SPA(5 ,ug/ml) and blocked with 1%BSA. Each anti-i/I CAfraction (200 ,1aliquot) was then added sequentially to a series of six wells with a30-min 220C incubation in each well for maximal absorption of VH3IgM from the CA fraction. After removal from the last well, the CAfraction was then tested for agglutination of either untreated or ficin-treated adult RBCs (I) and compared to agglutination of an unab-sorbed CAaliquot. CAfractions were also sequentially incubated withwells not coated with SPA but blocked with 1% BSA to evaluate fornonspecific binding of CA to microtiter plates. Goat anti-human IgMalkaline phosphatase was used to detect binding of VH3 IgM to plates.VH4.2 1 CA CAPand a VH3-encoded monoclonal IgM without RBCspecificity were used as controls.

Hemagglutination inhibition. Serial dilutions of anti-I (VOG) andanti-i (CAP) (6) in PBSwere tested against ficin treated RBCs(adult Ior cord i, respectively) to establish a range of CAconcentration yielding4+ hemagglutination (HA). Hemagglutination inhibition curves wereestablished using 50-,al aliquots of each CA ( 1.25 and 5.0 gg/ ml) incu-bated with an equal volume of anti-idiotypic antibody, 9G4 (serialtwofold dilutions, starting from 10.0 ,ug/ml). After a 30-min incuba-tion at 370C, ficin-treated RBCswere added and HAwas assessed aftera 30-min incubation at 4VC. Purified CA (0.2-6.0 ,gg/ml) were evalu-ated using high concentrations of 9G4 (> 5.0 ag/liter). HAwas also

Table I. Nucleotide Sequence of Primers Used in First-strandcDNA and PCRReactions

Application Sequence (5'-3')

cDNA synthesisC,, CGAGGGGGAAAAGGGTTGGGGCC, AGATGGCGGGAAGATGAAGACHJL* ACCTAGGACGGTCAGCTTGGT

AmplificationVH1LEADER ATGGACTGGACCTGGAGGGTCVH2LEADER ATGGACATACTTTGTTCCACVH3LEADER ATGGAGTTTGGGCTGAGCTGGVH4LEADER ATGAAACACCTGTGGTTCTTVH5LEADER ATGGGGTCAACCGCCATCCTVH6LEADER ATGTCTGTCTCCTTCCTCATVKILEADER ATGGACATGAG(GA)GTCCVJIILEADER ATGAGGCTCCCTGCTCAGCTCCTGVIIILEADER I ATGGAAACCCCAGCTCAGCTTCTCVKIIILEADER 2 CCCCAGCTCAGCTTCTCVJIVLEADER ATGGTGTTGCAGACCCAGGTCTTCHLFW1* CAGTCTGTGCTGACTCAG4G DJ* GGCCCCAGTAATCCCTGGGTA4G VHCDR2 GGGTATGGTAGTAGCCACACTGTCVH26CDR2 GTATGTGCTACCACCACTACCACT4G VLCDRI CAAGTAGGTGCCAACACTvgCDRI TAAGTAGCTGCTAACACTVKIIIaCDR3/INTRoN TGTGGGAGGCCAGTTGCTVH3 NTRON CTGGGCGCACAATGACT

Sequences of oligonucleotide primers used for first strand cDNAsyn-thesis and PCRamplification of V genes. * HJL, human A joiningregion; HLFW1, human A framework 1; DJ, the junctional sequencegenerated by the heavy chain diversity and joining region genes.

Table II. Red Blood Cell Binding, Specificity of SerumAntibodies, and Clonal Supernatants

Adult I Cord iAdult I ficin untreated Cord i ficin untreated

RBCs RBCs RBCs RBCs Specificity

CMserum 1:128 1:16 1:64 1:4 I4G SN 1:16 1:8 Neg Neg I

SC serum 1:8 1:4 1:8 1:2 ISpi SN 1:8 1:4 1:2 1:2 ISp2 SN 1:8 1:2 1:32 1:8 i

Donor CMserum, donor SC serum and supernatants from EBVtransformed B cell clones were tested for hemagglutination of RBCs.Samples were serially diluted in PBS and incubated with a 2-4%suspension of RBCsat 4VC for 30 min. Samples were designated ashaving either anti-i or anti-I binding specificities based on the pre-dominance of cold reactive agglutination with umbilical cord 0 RBCsor adult 0 RBCs, respectively. Test RBCswere also treated with ficinto show that binding was not decreased by protease treatment. Thedilution indicated is the highest dilution at which hemagglutinationwas observed. CMserum, SCserum, and 4G supernatant (4.88 1Ag/mlIgM) were tested "neat"; i.e., not concentrated or diluted. For Sp Iand Sp2, the clone supernatants were concentrated to 15.6 and 12.3Ag/ml IgM, respectively, for agglutination studies. Media controls(similarly concentrated) and PBSdid not agglutinate RBCs.

compared to a dilutional control (antibody incubated with 50 Al PBSinplace of the anti-idiotypic antibody). A decrease in the score of hemag-glutination in the presence of inhibitor is expressed as percent inhibi-tion of hemagglutination.

Sequencing and cDNAcloning of V region genes. The experimentalapproach for the cloning and sequencing of the expressed IgV genes hasbeen previously described ( 18). Total RNAwas isolated from EBV-transformed cell lines using thiocyanate-phenol chloroform extractiontechniques. First-strand cDNAsynthesis from 1 gg total cellular RNAwas performed with reverse transcriptase (Life Sciences, Inc., St. Pe-tersburg, FL) and an oligonucleotide primer specific for the 4 chainconstant region (C,,) (Table I). The cDNAwas amplified by PCRwithAmplitaq recombinant stable Taq polymerase (Perkin-Elmer CetusCorp., Norwalk, CT), the same C,, primer and 5' primers complemen-tary to the conserved portion of published leader sequences for each VHgene family ( 1-6) (Table I). The same approach was used to isolate theVL region gene sequences. The primers corresponding to K constantregion (C), the joining region of the X light chain (JA), the leader se-quences of each of the V, families ( 1-4) and the framework I region ofV. genes (consensus primer HLFW1A) are listed in Table I. For eachclone, amplification occurred with only one of the 5' primers; e.g.,corresponding to leader or framework 1 sequences. PCRamplifica-tions were carried out using a programmable thermocycler (Coy Labo-ratory Products, Inc., Grasslake, MI). 30 cycles were performed withsoak temperatures of 94°C, 55°C, and 72°C for 1 min each followed by9 min at 72°C. The amplified products were blunt ended using Klenowfragment DNApolymerase I (Sigma Immunochemicals), gel purified(2% agarose), electroeluted, and phosphorylated with T4 polynucleo-tide kinase (BRL Life Technologies, Gathersburg, MD). AmplifiedDNAwas then ligated with T4 ligase (International Biotechnologies,Inc., NewHaven, CT) into the phosphatased SmaI site of pBS M13 -

(Stratagene Inc., La Jolla, CA) and transfected by electroporation (BioRad Laboratories, Rockville Center, NY) into XL- I blue strain of Esch-erichia coli. Clones were screened by blue/white selection and restric-tion analysis of plasmid DNA. Clones containing the insert of expectedsize (400-600 bp) were sequenced by dideoxynucleotide terminationwith Sequenase (United States Biochemical Corp., Cleveland, OH)


U I 01

HD I 01

< I 01

U u

EH <01 <I

0 AI

< u uHI H I

01 H

0 0II E--'0U 4i

O

l

H0 E--00v <

AH

HvI HvIHI

< u

<04 U

(D E-- I u

l<C HI

0) H

U

4-I <A HI

I Co 0110 O o40'

4 <E I

4 01

O4-)

I 4 E I4

I 0NE1 I

14 HI

I 1

14 01

I la

14i H

14-i 01

ON O

<

I E-4 I

I14Co

10 01l1 0 01I

14 01OCo

10 01

1 4 01I1 0 01II Co H

I Co 010' 01I

I Co zI Co HvI

Co 0Co O

1 0' 01Co H

I Co

I Co )

0v1 <I

H:I 01IHDI O I

I O0IA () EF I

I tO4 E

A: l HI

I E- E

I E-4

E-4 I

I <E I

1 0 I H-I

u I E- I

HI <

H I O:oH 0U101 <Il

E I u I

u Iu I

<1 <IU

HE I u0

H I 01I

0 < IHI 01CrU

HOI 01

104 <I

HvI 01U

<1 <IU

HO< < I

H4I 00H HI

c' I 01

VHI H I

H I 04

H I < IH I 01I

H I < I

<:I <:I

0DI <:I

H I 01

< I < I

< I 01I

O<I 01

C:0

-H0'0)

V:

V-

A

!U

C;U

H4

H4

04

UU

E0-

u

10

H HvU

u

uE-

<

u

OI

HI

HI

OI

<II

AI

<0

I E1

HI

01

<IIOo

IU

IE-4

I ocv u

<I01<0:01E

01

H

Kx

Kc

0

0v0v

EH

0EH0E00H500

H

106

IO<H

I0EI <

I0UI0O

l0

I O

H

IOU

IOEI10

HHE

I1010

HH

I <

10t

I OI10I <10

HEI10I1010I1O

H

CoO0)H

-K-K-K

-K-K-K

-K-K-K

-K-K-K

-K-K-K

-K-K-K

-K-K-K

%D %D %D %D(N4 (N4 (N

: u = = u3. -W 3: qw 3 'o 3 q

E0 H <

0 E- I 01

<I 051 01I01

I <OI < I

u0 A U I I

H I I < I <vIu

I Ev I E- IE I HI I

u < I I

HI 10IE I HIl 01l1<O1 l 01l)

HE4 I 1I1<S:1 101 01D

1 0 1 I I

I H C) I

H I I II

I I tI C)

101 I E1< I I10 I I EH I 0

O I 0IC I 01V <:l 1<1 <I

<: I 100II

uY

<Ev I 0I I 01

Ot I I O- I O I

0 HI I 1 I

(F) I u LD I :

H I I ) I <Hq I I101 <I

EH I HI 0 1

E-4 I I O I E

H I H I 1 I

0 10 1 H-4I

H I I<1E I <I

EeI I IE-4 I I O I <

u

<v I I101 I 1 IH I 1 <1 H:I01 1<:1 <4IH I 1<HSIUI

01) I HI HvI

H I I<I 01

E0 I HI 0

H I 00 01<

< I 0101I0 0I1 I <vI<zI X <I0 <-)I

H I 04<HIoI

H I HI < I

H I H I HI

01 I <qI < I

0 HoI <SI

H I 01 I <vI01 01: < I071 01 <OI< I <I 01< I <I H I

O<I 01 01H)I 01 H<I

01I 01 <I

071 <I 01I

H I 01 01)HvI 0O I <)I


C0

4C

V

4

0'4-

4-

U

l 4

0'4

HE-

E-4

E-4E-4

uS I

Co

E-1I

E--IErI

Co

D H I

F4

t I *

0Co ICoICoICo e0'I0' I4 sCoI04

4

< I<OIO<Io<OIHOIHbIH:I

I-

IW'S

AUIH

101U

< I

H

H

I<I<

a1000u

-H <

wu0

VHUV E--f

A

I uI <I EHI <I E-4I EHI <I E-4I E-4I uI uI HI UI <I

HIHI E-4

I E-4C EH0 <

.-I E-4-H H0)0a

u<:

M E-4

a <a;

E-VO

0

HE-0

0L)0

000H

000

00

H-I

(Nj

04CI

00._C

0'0C

e0j

rSC-,

Lf4LL0 LP4CO X P4

m: uU):co PCI) 0C/)

using both "-40" and "reverse" universal primers ( 19). Oligonucleo-tide primers were synthesized by an oligonucleotide synthesizer (Ap-plied Biosystems, Inc., Foster City, CA) using phosphotriester chemis-try. For each clone, we sequenced three to four bacterial isolates in bothdirections. The obtained sequences were identical and thus confirmedhomogeneity of the cell populations.

Results

Establishment of CA-secreting B cell clonesPBL from normal donor CMwere exposed to EBVand seededat 1 X 103 cells/well of 96-well microtiter plates (Corning).After 4 wk of culture, supernatants were assayed for CAactivity(20). CAactivity was detected in only 3 of 60 wells with growth(identified on different plates), designated 4G, 8B, and 3C.These wells were subcloned by limiting dilution ( 13). CA-se-creting cell lines that were considered clonal based on statisticalanalysis were subsequently expanded. These three clones (4G,8B, and 3C) were derived from PBL selected with the V.III-spe-cific anti-idiotype. In contrast, two trials of PBL (donor CM)selection using the anti-idiotypic antibody 9G4 did not yieldany CA-secreting cell lines.

EBV-transformed splenic lymphocytes from donor SCwere exposed to EBV and seeded at 100, 10, and 1 cell/well.Two wells were considered to contain clonal cell lines based onstatistical analysis and were designated Sp I and Sp2. These celllines were subcloned again at 1 cell/well, tested for CAactivity,and expanded.

Southern blot analyses were performed on DNAfrom es-tablished clones that confirmed clonality of the cell lines (datanot shown). Furthermore, clones 4G, 3C, and 8B from donorCMshowed the same JH restriction fragment pattern usingthree different restriction enzymes. Subsequent sequence analy-ses indicated that these clones were in fact identical. Conse-quently, we present data from only one of the clones (4G).

Serology of cold agglutinins secreted by B cell clonesThe binding specificity of the RBCautoantibodies was deter-mined by their differential reactivity against adult and cord testgroup 0 RBCs(Table II) (4, 21 ). The CApresent in the serumand in clonal supernatant of normal donor CMreacted morestrongly with adult (I) RBCsthan umbilical cord (i) RBCsandwas therefore classified as anti-I. Similarly, CA present in theserum of donor SC and in supernatant of splenic clone Sp 1preferentially reacted with adult (I) RBCsand was also classi-fied as anti-I. In contrast, splenic clone Sp2 reacted relativelystronger with cord (i) RBCs and was, therefore, classified asanti-i.

The VHgenes of anti-I specific B cell clones of donor CMand SCbelong to the VH3f amilyThe VHsequences of clones 4G (3C and 8B) are most homolo-gous (92%) to the published germline gene WHG26(22) (Fig.2 A). However, based on the level of sequence complementar-

ity, the VHgene WHG26cannot reliably ascribed as the germ-line precursor.

The VHsequence of clone Sp 1 was most homologous (98%)to the fetally rearranged M85/20P1 VH3 gene (23) (Fig. 2 B).

The VL genes of the anti-I specific clones of donor CMandSCare VJ and VJII genes, respectivelyThe VL genes from clones 4G (3C and 8B) were shown to bemost closely related (96%) to the VKIIIa germline gene Vg (24)(Fig. 3 A). Other antibody specificities encoded by rearrangedgenes most closely related to the Vg germline gene include apathologic CA with anti-Pr2 specificity (25) and rheumatoidfactor protein (26).

The VL gene expressed by clone Sp 1 was only 94%homolo-gous to the VKI germline gene Ll 1 (27) (Fig. 3 B) and its germ-line precursor, therefore, remains speculative, since VKI repre-sents a complex VL gene family.

The anti-i specific B cell clone of donor SCexpressesVH4.21 and VHII genesThe expressed VHgene of clone Sp2 showed closest homology(98%) with the germline gene VH4.2 1 (8), while the expressedVL gene was 98% homologous to a rearranged VII gene (VIIfamily) (28) (Fig. 4).

Evidence that the expressed VH and VL genes encodinganti-I of Clone 4G are somatically diversifiedThe expressed VHand VL genes appeared somatically mutatedwhen compared to known germline gene sequences. To inves-tigate whether these genes were in fact mutated rather thanderived from yet unreported germline genes, we used a PCR-based approach previously described by van Es et al. (29).Specific oligonucleotide primer pairs were made correspondingto putative mutated regions of the CDRsand to the potentialgermline precursor sequences (Table I).

In addition, we isolated the putative germline precursor VLgene Vg, from autologous granulocyte DNAof individual CM.A similar attempt to isolate the precursor VH3 gene was notperformed because of the complexity of the VH3 family, whichcontains many homologous gene segments.

VHgene analysis. In this experiment, autologous granulo-cyte DNAcould be amplified with primers corresponding togermline sequences but not with primers corresponding to ex-pressed CDR2 sequences. Amplification of CMgranulocyteDNAwas achieved using primer pairs VH3LEADERVH3INTRONand VH3LEADERVH26cDR2, but not using primer pairs,VH3LEADER-4GVHCDRl (Fig. 5 A). Based on sequence homol-ogy to published VH3 germline genes, the VH3LEADERprimermay be expected to amplify all VH3 genes. In contrast, theVH3 INTRON primer was designed to amplify only VH26-likegenes. These amplifications were included as positive controlsand to show the presence of commonly identified VH3 genesegments. The lack of amplification using 4GVHCDI suggeststhat the expressed VH gene of clone 4G represents a mutatedrearranged VH3 gene. Additional analysis of the expressed VH3

Figure 2. Expressed VH sequences encoding anti-I. (A) Sequence analysis of expressed VH3 genes encoding anti-I of clone 4G. The the relevantreference nucleotide sequences of germline gene WHG26(22), and JH4 (33) are shown on the top lines. Homology of 4G nucleotides is shownas a dash. *Oligonucleotide sequences used as flanking primers (VH3L"DER and C,,) for amplification of the VH gene of 4G. (B) Sequence analysisof expressed VH3 encoding anti-I of clone Spl. The nucleotide sequences of the relevant reference sequences, fetally rearranged M85/20P1(23), DXPI (30), and JH4 (33) are shown on the top line. Homology of Spl nucleotides with reference sequences is shown as a dash. *The oli-gonucleotide sequence used as a flanking primer (5' VH3 family specific) for amplification of VH of clone Sp 1.


'I

UP

UP

E-4

UP

UP

E-4

UP

0-1

II

I4

U

E4:

E4:

U

E4:

E4:

UE4:

(5j

(5E-

H-

E4:

U

I

I

II

I

It

III

I

I

I

I

III

III

I

II

I

III

III

III

Iu

I

III

III

III

III

011

111

111

111

4:u

EP4:

14:4

E4:

14:4

-E1(5

I

m

II

II

III

I

tpI

I

UI

III

II

I

I

III

III

III

III

III

III

III

III

III

III

III

0

~4-Vi

(00)

m0'( 0 0'(5u

(0

4-)

4-)

4(0

4-)

4-i

4-)

U

4-)

U

U

4-4

U

U

E4:U)U)

H-U

E4:

U

E4:

A E4:

4:

4:

U-

14:

14:

P4-

IPU

P4i

I1 co

IPU

P4-)

P4-i

ji

4-

V-Cl2826 L. C. Jefferies, C. M. Carchidi, and L. E. Silberstein

U-H

(5H

UU

U

E4:U

E4:U4:

U4:

(5H

H

E4:

E4:

E4:

A-4'('(

'-4

-4:

-4:PU

PUPUPH

PUcou

'4 <

(0

0

-H

ON01)

A

0

UV

III

0)

I

I

I

III

III

III

III

III

I

III

III

III

III

III

ONION

III

III

III

III

III

III

III

IONI

II

III

III

E4:

U

-4:

-4:

E4:

E4:

E-:

E4:

PUE(5

4:4

4:4

U)

4:4

U-

4:E-

H

E4:

4:4

U

U

I

H- I

<:

UPP4

<:

UP

UP

UP

-J

H-4U

I

E4:O

E4:

UP<:O

PO

H-4 I

I

E4:

E4:

E4:

H-4 I

H-4 I

UP

E4:

E4:

III

I

II

I

II

I

I

I

I

I

III

III

I

II

III

III

III

II

III

III

III

III

UII

III

II

9cIII

III

r-4 r-I r-4 r-4 V-1 1-4 r-4 r-4r-q 04 W-4 a4 r-4 04 r-q 04$.-I cn 1-.3 Cf) 4 U) 64 (n

I101 01 A H-I 01 <0 A 0101) II C0 IE-l 0 01 0 fUIu C)I U U~~~~~~~~~O

leH <I AUp <i HI<I10 HI 10 10 <I0 < H-I <

<<I E-II u

<0 E-I10 1 10 01

0101101~~~~~uE-

10 AH-HII 0 <OH~~~~~~~~~~~~~~~~~~~~~~~~~A-E-4 < H-I H 0U

HI <I 1<1~~~~~~~~E- HI 01 IE 4 HI4 <I 01 C <

IIE-f HC I <--, <

HuI 10110 10 HI Iu0u0 I I 1101 IHI-l~~~~~l HII<EI i. KH 1 <

I I~~~~~~~A IIIHI <0 .0

HI H IOuI 101 0I< L E-I0 <I 0 1-4E

<10 E

0I10I< H-I (0 ~ ~~~ ~~01 10

lIE-l HIE I < 01 < 0 HI4 E-I H

<' <10<

IHIK 10 01 1~~~ ~~~~~~~~~~~~~<1<I IE-HI I <1U 01~ ~~~~~~~~~~~~~~~~A <I II

I PI 01 10 IuI uI01K () H 10 HI 0E- K HI- HI HO

1 P <I 10 01 10 00 <0111 uP0 10

I U EC)01 1< 1014 HIK 101 01 HO

I P II III101 E14001 0 HHEIH-0KIu 01f I<

4() 01<1H 10 1 I 1 11(a

I01uC 01IH H)EI HI0') I

I< H IH~~~~~~I II u10 HI<<

C) <I 1<1 01 001101IHIIH<I< I mIII<101 0'< IHIEH

0') I101 01-qPt 0-IE-4

1 01u4-) 01

w'.0 O 1 0

O H-I V l <I Eo1 141 IC) V H4I

P) HI 01 I IA H E01 A0 <I

0' 0< I E- I

P 01 HI 0 0 10 10 H-I10 .0 H-I 01I < 10 10 <I

0' <I <I HI- EH 1< 0 HI0' 01 01 01 < 10 E01 IP- <I 01 HI EH 100 <I

to uU u0

< EI0' 01 <I <I 10 C01 1<I 010

Ff <HI <

0' HI 01 OP~~~~~~~~~~~~4 IE-HO IHI HI 01P 01 HI <I I<H- 00 101 <I1001 01 <I H~~~~~~~~~~~~~~~~~~E-< V < 1<1 0

pN H- I E I CI

0' E-I 0I EI HI 01 lO <Ito E-I < FfI I H-I Cl g 0

PHI <I 01 ~~~~~~~~~~~~~~~~IEHI01 0C10 HI <I HI 10 HI VEl 0

0' H 1 01 1 1H<I 01C0140<I HI< <I HI 0

O ) I E4C HI Ep4 <I< 1 IH 0< HI 0

C)4 01I 0 <I IEH-0 H0' ~~~~~~~~~~<u< I C-

~~~PC/) >C/) >C) >0 >0U) 0 1 m >0n >0) >C/) >C0)


1 2 3 4 5 Figure 5. (A) Evalua-A tion of CMgranulocyte

DNAfor the presenceof target sequenceVH D2 of clone 4G.Lane 1, pGEMstan-dards; lanes 2-4 repre-

470 sent CMgranulocyte350 DNAamplified with the350- following primer pairs:

lane 2,VH3LEADERVH3INTRON(corresponds to 17 nu-cleotides located be-tween the heptamer and

1 2 3 4 5 6 nonamer sequences, 3'B of the VH26 gene seg-

ment); lane 3,VH3 -4G VH ;lane 4, VH3LEADER_

375 - VH26CDI . Lane 5 rep-resents 4G plasmid

180 - DNAamplified withprimer pair VH3LEADER4GVHCDR2. Bars indi-cate 470- and 350-bp

size bands. (B) Evaluation of CMgranulocyte DNAfor presence oftarget sequence VL CDR1of clone 4G. The 5' leader primer, VIII-LEADER2(shorter than VJIII LEADERI) was constructed to approximatethe length of the 3' primer 4G VLCDRI to optimize amplificationconditions. Lane 1, pGEMStandards; lane 2, 4G plasmid DNAam-plified with primer pair VKIII LEADER24G VL RI; lane 3, Vg clonedfrom CMamplified with VKIIILEADER 2_VgcDRI; lane 4, CMgranulo-cyte DNAamplified with primers VjIIILEADER 2-4G VLCDRI; lane 5,CMgranulocyte DNAamplified with VKIIILEADER 2_Vg Rl; lane 6,CMgranulocyte DNAamplified with VH3L9DER4GDJ. Bars indi-cate 375- and 180-bp size bands.

gene sequence is not possible because a putative germline pre-cursor was not determined; the latter experiment was per-formed in the VL gene analysis (see below).

VL gene analysis. First, oligonucleotide primers VKIII LEADERand V IIIaCDR3/INTRoN were used to isolate and sequence germ-line VHIII genes from autologous CMgranulocyte DNA. Of four colo-nies sequenced, two were identical to germline gene Vg while two weremore closely related to a highly homologous VKIII germline gene Vh(24) (Fig. 6), thus confirming the presence of Vg in the germ-line repertoire of CM. Second, oligonucleotide primers weremade corresponding to the putative mutated regions of CDR1and to the potential germline precursor Vg. As expected,primer pair VKIIILEADERv gCDR1 amplified granulocyte DNA,while the primer pair VKIIILEADER-4GVLCDR' did not (Fig. 5 B).Taken together, these data suggest that the expressed VL gene ofclone 4G is a somatically mutated Vg gene.

Weapplied a binomial probability model to the pattern ofmutations in this VL gene as a test for positive selection (30).The expected fraction of replacement mutations in CDRfromthis VL sequence is 0.22, calculated by multiplying the relativesize of the CDR(0.28) by the fraction of all mutations in CDRthat could result in amino acid replacements (0.79). Using thebinomial probability model, we calculated that the distributionof six replacement mutations in CDRof a (corrected) total of

13 mutations had a low probability of occurring at random (P= 0.046). Positive selection of replacement mutations in CDRhas been observed in murine immune responses and presum-ably reflects selection for enhanced antigen binding (31 ). Thestatistical model provides evidence that positive selection led tothe accumulation of amino acid substitutions in the light chainof this anti-I antibody. It is tempting to speculate that the ob-served somatic mutations are the result of selection by the Iantigen or another cross-reactive antigen.

Diversity region sequence analysisThe D region segments of the three clones (4G, Sp 1, and Sp2)with specificities for the related i and I antigens are heteroge-neous. Comparison of nucleotide sequences to published Dregion germline genes shows homology of the Sp 1 D segment(1 1 nucleotides) with Dxp1 (32) and homology of the Sp2 Dsegment ( 17 nucleotides) with Dxp4 (33) (Figs. 2 and 4). TheD region sequence of clone 4G could not be reliably alignedwith any of the published D gene segments. The D segmentsfrom the natural anti-i/I secreting clones 4G, Spl, and Sp2were compared to published pathogenic anti-i/I RBCautoanti-bodies (6-8). However, a commonD region sequence, possi-bly contributing to the similar serological specificities of theseautoantibodies, was not apparent.

Evaluation of clonal supernatants for expression of 9G4idiotypeEach of the clonal supernatants was evaluated for expression ofthe 9G4 idiotype (Table III). As expected, only one of theseclones, Sp2, which expresses the VH4.21 gene, was bound bythe monoclonal anti-idiotypic antibody 9G4 (OD > 1.0).These findings indicate that Sp2 expresses the 9G4 idiotype,which has previously been highly associated with pathologicanti-I/i autoantibodies of (pre)neoplastic B cell origin (6, 8).

Evaluation of naturally occurring CApurified from serumfor expression of 9G4 idiotypePurified CA, isolated from the serum of 15 individuals withoutimmune hemolysis (including CMand SC), were evaluated byELISA for expression of the 9G4 idiotype. For each individual,both anti-i- and anti-I-enriched fractions were evaluated. Thequantity of 9G4-reactive IgM content of CA fractions was as-sessed by establishing a standard curve correlating the IgM con-centration of a known VH4.2 1 -expressing monoclonal IgM CAwith 9G4 reactivity. Wefound that of 30 fractions tested, 7 hadno 9G4 reactivity, while the remainder had relatively little 9G4reactivity (range = 1-30%). These results are summarized inFig. 7. These data suggest that the majority of naturally occur-ring circulating CAare encoded by non-VH4.2 1 immunoglobu-lins.

9G4 fails to inhibit hemagglutination by naturallyoccurring anti-i/I CApurifiedfrom serumThe 9G4 antibody is known to efficiently inhibit hemagglutina-tion by pathogenic, monoclonal anti-i/I CA encoded by theVH4.2 1 gene. Consequently, hemagglutination inhibitioncurves were first established for VH4.2 1-encoded proteins VOGand CAP(6) using anti-idiotypic antibody 9G4 as the inhibitor(Fig. 8). As little as 0.15 tl g/liter and 0.62 ,ug/ml of 9G4 com-pletely inhibited hemagglutination by 1.25 ,ug/ml of VOGandCAP, respectively. Complete inhibition of 5 ugg/ml of CAPwas


14-I l w0 II > <lI1 '-4 U II IW< l* Q-ir-4

I I I0jg (001< 1 C(<lI 1I 4001< IIuI~04 I1114- 91(a301ruI (E-II -0ouII uC'

lll I~ 0<11I 1-011u AO0-I 14 n(1 :u I: -0) 1 A 0E-4 I14 -W

I 0)0- III mII(011-f<w CC)>i11 01 E 1i0CI >01OIlow 4' <~-

E1E I '0-11 '-1<II 14(91 04491( '4' P-

I~<Ill '011 (9011 <-H 9(11 <<14I0)0-4111011 H FqU) H E4I

I0)04III '011 ~ ~ ~ 4 <

1..u 1 41 1 w<9lI <IACO 040E-II C)-'I ~~~I04C' )0 I -I0) I-l 0E-) l 1 (_9

I 0) E-' I I I a I I IIII '11 '1(11 I1401< a~(9 Im zl

1011 '00) -II1 >ElI1 14(9II C'I'-40)0 I1I1I411 >(1I 11-4 E-I a4E<II I(90S1I0~0-4III'011 0)0-411 401 '01 IO(9

)E-4 IIII4E'0<<11 CI.-04 I1I14-11 1< 14 1 0 1 d1-'I 1411I~ '--4-(I-lI0(94'1114 1 1(9(911<l (9 4-4<1 E-'0II

'-1<4 uIIIII'011IE-4I 00uI'011II IM 0)9(A:9l s~

1(00-41(11 0~l 04 <II CI-I<l4 (9(11 -0II'-404'III '011><I, FII

I Ul ll ) 0 --4<1 l 00 (0-411II0<4'III 0)I '0-'<C) 1(011 1-4<11 >(9< 'oa

141m ~ 4< 0)(9I '--401 F-4011'-404'III0~~~~IA 0~~l OH

<(4' '(041 '1 I'01 (9 0)-11 u 10 1 011E4I I-41 1-( .04

( IU11'-4<9l << 0- EC

0)0)0'4'11011 01 I4l -l 1 (a)0I 11401E-<l II '

1411 411 >~~~~(9l 0 01E--4<1

1411 411 14011 '0<11 00-4ErIC.)I '011 411 0)0U-I c'<( c)n0 '

I '011 411 ~~ 0l II ~04U~ 3<I

1411 014- IE H I uI 0~I 411 14 11 14 I < 01410 411 -411 0-01< 0E-4o II

1(01 11<9114- -4<1< 0<11loll 0~~~~~~~I '0:3

1411 411 <<II~ ~U-)En< (~(9 IE-l91 1401(011 411 (9(911 <<II 0-<I

I0~~~~~lI 4 1 1 E4 C

1411 014 0<11~ ~~~~~00 1<0 0,1(011 411 1-4011 14011 04)0-1

AJ' 41 0<0-11 < 911 -4<1

4(011 1 7 0)-11 w-1<11 .E0-' WII

VO~ (1140-11n01 Q.II 01 0)011 (a'-4 -4(911 11<1

4'011 CI)E-4I <((91 < 91

4JCOI 4(011N 0) E4 E-111 0 1 1III (011 ~ 4E0I (Iu l 0-<l4I r

0. 0 0 004-) 0% tovr.~~~~~~~~~~~~I0

'-4 '-4 '-4 '-- 0 0

u 0E-4II <~~~~~~ 0

.~~.(()~~~~Cfl 4 ~ ~ E- 0 C') ') C

00 0 EU E'a EU E-ON '-4 C" ON7' C-(Nm0(0)(NV a~)0u 00) N ON 0) CN to 0> 00 >00 >(90 > U >(90 U >(90 z ~>.

Natural/v Occurring Anti-i/I Cold Agglutinins 2829

Table III. Ig V Region Use and Idiotype of Autoantibody-secreting Clones

Ig variable region gene Idiotype

Clone VL VH 9G4

4G VKIII VH3 NegSpl1 KI VH3 NegSp2 VA VH4 Pos

VHand VL was determined by nucleotide sequence analysis for clones4G, Spl, and Sp2. Expression of idiotype 9G4 was determined byELISA. Sp2 was bound by 9G4 (OD > 1.0). None of the other clonalsupernatants were bound by 9G4 (OD < 0.02), but were bound bymurine anti-IgM as a control to indicate adequate concentrations ofclonal IgM present. (OD > 1.0) Cold agglutinin VOG(a previouslydescribed VH4.2 1 anti-I) (6) was used as a positive control for 9G4reactivity (OD > 1.0). Background values are <0.02.

achieved using 2.5 ,ug/ml of 9G4. Naturally occurring CAweresubsequently tested at concentrations of total IgM rangingfrom 0.1 1 to 6.4 ,ug/ml. For 14 of 14 CAevaluated, hemaggluti-nation was not inhibited, even using high concentrations of9G4 (5.0-15.0 ,g/ml) (Fig. 8). This failure to inhibit hemag-glutination using anti-idiotypic antibody 9G4 supports the find-ing that only a small fraction of circulating naturally occurringCAexpresses the 9G4 idiotype.

Evaluation of naturally occurring CAfor VH3-encoded IgMcontentA serologic marker that would recognize all VH3 gene productsis to our knowledge, currently not available. Nevertheless, wehave used modified SPA, which binds to - 90% of VH3-en-coded IgM through Fab-binding sites ( 17), to evaluate the pres-ence of VH3 IgM in the CA purified from normal serum andalso to show that the absorption with modified SPA results in

the abrogation of hemagglutination activity of the CA frac-tions.

Five CA fractions contained quantities of VH3 encodedIgM (OD range = 0.41-0.49) similar to the VH3 content ofpooled normal human IgM (OD = 0.45) as defined by bindingto modified SPA. In two of these CAfractions (including anti-Ifrom donor CM), the modified SPAbinding exceeded that ofpooled IgM. The remaining CA fractions (19/24) containedmodified SPA reactivity that was less than that of pooled nor-mal IgM (OD < 0.20). The varying amount of modified SPAreactivity suggests that the IgM antibodies in these CAfractionsare encoded by (a) VHgenes other than VH3 (or VH4.2 I ); (b)certain germline VH3 genes that are not bound by modifiedSPA, or alternatively (c) VH3 genes that have lost modifiedSPA reactivity caused by somatic mutations ( 17). A positivecontrol antibody (representing 100% VH3-encoded IgM, at0.25 ,ug/ml) gave an average ODvalue of 0.55. These experi-ments suggested that some purified CA fractions containedvery high amounts of VH3-encoded IgM, which was unlikelycaused by nonspecific Ig entrapment during the RBCabsorp-tion and elution procedure.

Next, we used modified SPAbound to microtiter plates tospecifically absorb VH3 IgM from CA fractions and subse-quently test the absorbed anti-I CA fractions for CA activity.After six sequential absorptions with SPA, three separate anti-ICA fractions (individuals CM, MS, and MG)failed to aggluti-nate adult (I) RBCs, whereas the unabsorbed fractions titeredto 1:16, 1:16, and 1:32, respectively. This finding suggests thatthe anti-I binding specificity virtually completely results fromVH3 IgM in these CA fractions. As expected, SPA-bound IgM(absorbed from the CA fractions) was detected in each of sixwells on the microtiter plate (Table IV). There was no absorp-tion of IgM from the naturally occurring CAonto wells withoutSPA coat. There was also no absorption of a VH4.2 1-encodedIgM CACAPonto wells with or without SPA coat. Together,these studies suggest that naturally occurring CA may be en-coded by VH3 genes.

Relative 9G4 reactivity of a-i versusa-l CA fractions in normal serum

%of totalIgM with 9G4

reactivity

100

Ba

60

40

20

a/i Cxajl aial citl aial aialreference 1 2 3 4 5

CA

11 ffl liill-aial aial aial aial aial ai al6 7 8 9 10 11 12

l13 14 15Iai a] ai aI CA al

1 3 1 4 15

Figure 7. Relative 9G4 reac-tivity in enriched anti-i vsanti-I CA fractions purifiedfrom normal serum of 15healthy individuals. The per-centage 9G4 reactivity wascalculated by dividing 9G4-reactive IgM content by totalIgM content. The 9G4-reac-tive IgM content ranged from1 to 30% of total IgM. Forall individuals studied, anti-ifraction (grey bars) con-tained greater 9G4 contentthan the corresponding anti-Ifraction (black bars). Indi-vidual 14 represents CM, thesource of clone 4G (VH3-en-coded anti-I CA) and indi-vidual 15 represents SC, thesource of clone SpI (VH3-en-coded anti-I CA) and cloneSp2 (VH4.2 1-encoded anti-iCA).


C0

4- 4-'0 mC.

. 4Jo a3

.C ._

._ ma

X

I

0%

25%

50%

75%

100%

A e 1.25 jug/ml CAP anti-i A C0IB 0 5 pg/ml CAP anti-i 0 CMC * 1.25 g/mI VOG anti-I O SCiD * 5 ig/rrg VOG anti-I * CA fractiorns

Al& *0

0.15 0.30 0.62 1.15 2.5 " 5.0 15.0

Concentration of Inhibitor pg/mL

Figure 8. Hemagglutination inhibition studies. Serial dilutions of pu-rified VOGand CAP(monoclonal VH4.2 1-encoded anti-I and anti-iCA, respectively) were first tested to establish concentrations yielding4+ hemagglutination. Aliquots (50 ,l) of these antibodies (both at1.25 and 5.0 Atg/ml) were incubated with serial twofold dilutions ofanti-idiotypic antibody 9G4 (beginning at 10 qg/ml). After a 30-minincubation at 370C, ficin treated RBCswere added and hemaggluti-nation assessed after a 30-min incubation at 4VC. The 9G4 antibody(> 5.0 qg/ml), was also incubated with purified naturally occurringCA (0.20-0.60 ,g/ml) for inhibition of hemagglutination. CA wereincubated with adult (I) or cord (i) RBCsdepending on the prede-termined relative specificty. There was no inhibition of HA for anynaturally occurring CAtested. CMI and CMi represent anti-I-andanti-i-enriched CA fractions, respectively purified from serum of in-dividual CM. SC i represents anti-i-enriched CA fraction purifiedfrom serum of individual SC. *Represents 14 anti-i-and anti-I-enriched CA fractions from seven additional healthy individuals.

Discussion

The Ig V region gene use of pathogenic, monoclonal CA hasbeen extensively studied by serology, protein chemistry, andrecently by nucleotide sequencing, whereas the VH and VL re-gion structure of naturally occurring CA has not been previ-ously determined. Because these proteins are present in sera inmuch smaller quantities than the monoclonal CA, they havenot been amenable to purification and subsequent analyses asperformed on monoclonal CA. Thus, there is little or no infor-mation reported regarding their structure or origin. While thenaturally occurring CA are probably polyclonal in origin,based on analogy with other naturally occurring autoantibod-ies, such as anti-DNA antibodies, the polyclonality of these CAhas not been demonstrated adequately. Furthermore, becausethe V gene usage of naturally occurring CAis not known, the Bcell origin of naturally occurring vs pathogenic CA has re-mained speculative.

In the current study, naturally occurring anti-I autoantibod-ies from two individuals were found to use VH3 family genes,the largest of the VH families (34). This finding indicates thatnaturally occurring CA may derive from B cells that are dis-tinct from the VH4.2 1 expressing cells, which can give rise to Bcell neoplasms. The initial isolation of VH3encoded anti-I auto-antibodies from donor CMusing K chain-restricted anti-idio-types was somewhat surprising in view of the published re-stricted use of VH4.2 1 or a closely related VH4 gene by patho-genic anti-i/I RBC autoantibodies (5-9). Therefore, we

subsequently used the 9G4 anti-idiotypic antibody (associatedwith the VH4.2 1 gene) to attempt to isolate 9G4-positive autore-active B cell clones. Although no CA-secreting clones were ob-tained from PBL of donor CM, we did isolate from spleniclymphocytes of donor SC one clone SpI (anti-I), which, curi-ously, is encoded by a VH3 family gene and another clone Sp2(anti-i), which indeed is encoded by the VH4.21 gene. Theisolation of a clone expressing VH3 using the 9G4 reagent likelyoccurred because the panning procedure for B cell enrichmentdoes not exclusively select idiotype-positive B cells, and be-cause subsequent screening of the wells containing EBVtrans-formed splenic B cells was based on CAactivity and not on theexpression of 9G4.

Interestingly, one of the VH3 genes encoding anti-I specific-ity is 98% homologous to the 20p 1 gene (23). This gene be-longs to a set of VH genes associated with fetal ontogeny (e.g.,VH26 related), which are considered to be preferentially ex-pressed during fetal development and which, in germline con-figuration, are associated with several natural autoantibodies(35). In certain instances, antibodies encoded by these fetallyderived VH genes with few amino acid substitutions, maycrossreact with both self antigens (e.g., anti-DNA) and withbacterial polysaccharides (36). By analogy, it is possible thatthe few somatic changes observed in the VH3 gene expressed byclone Sp 1, may play a role in conferring the anti-I specificity.However, the comparison of VH region genes that encode bothnatural and pathogenic anti-i and anti-I CAhas not revealed ashared protein sequence that would potentially contribute toanti-i/I antigen binding.

Table IV. Absorption of VH3 Igfrom CA Fractions UsingModified SPA

Hemagglutination titer of adult (I) RBCs

Before adsorption After adsorption IgM boundCA with SPA with SPA to SPA

CManti-Ipurified fromserum 1:16 0 OD0.25

MSanti-Ipurified fromserum 1:16 0 OD0.17

MGanti-Ipurified fromserum 1:32 0 OD0.38

Anti-I fromVOG(VH4.2 1)cell line 1:32 1:32 OD0.0

CA fractions were sequentially incubated in a series of wells precoatedwith modified SPAand blocked with 1%BSA. Absorbed CA fractionswere then tested for agglutination of untreated adult (I) RBCsandcompared to titration of the unabsorbed CA fraction from the sameindividual. Goat anti-human IgM alkaline phosphatase was addedto wells after removal of the CA fractions to confirm binding of IgMto modified SPA as shown by ODvalues. There was no absorptionof CA onto wells without modified SPA coat. There was no absorp-tion of VH4.2 1 CA CAPonto wells with or without modified SPAcoat. The signal for a positive control monoclonal VH3 IgM (1 gg/ml)OD= 0.50.


The great majority of naturally occurring CAhave the anti-I specificity, and when the anti-i specificity is present, it isusually found in smaller amounts (titers) relative to anti-I.Therefore, we anticipated that clones screened for CAactivitywould most likely have anti-I specificity, as did clone Spl . Theisolation of clone Sp2, with anti-i specificity, from the samedonor SC raised the question of the relative frequency of thesetwo specificities. Since small numbers of independent cloneswere isolated and since enrichment by panning was used, it isdifficult to estimate the relative frequencies of anti-I vs anti-iexpressing B cells, or of VH4.2 1-encoded vs non-VH4.2 1-en-coded natural occurring anti-i/I CA.

It is possible, however, to assess circulating naturally occur-ring CA for the relative expression of VH4.2 1- vs non-VH4.2 1-encoded IgM, as defined by 9G4 reactivity. Wefound that of30 fractions tested, 7 had no 9G4 reactive IgM, while the re-mainder had relatively little 9G4 reactive IgM (Fig. 7). Fordonor CM, from whomclone 4G (B cell clone, secreting VH3encoded anti-I CA) was isolated, circulating anti-i and anti-Ifractions contained only 4%and 0%9G4 reactive IgM, respec-tively. In addition, the anti-i/I CA fractions of donor CM, aswell as the remainder of anti-i/I fractions, were not inhibitedby the 9G4 antibody (Fig. 8). VH3 IgM was detected in severalCAfractions and was found to be enriched in donor CManti-Ifraction compared to pooled IgM. Moreover, absorption of CAfractions with modified SPA to remove VH3-encoded Ig re-sulted in complete abrogation of anti-I binding specificity foreach of three CA fractions evaluated (Table IV). Together, thedata suggest that to a large extent, naturally occurring anti-i/ICA are encoded by non-VH4.21 genes.

In summary, the current work represents an extension ofour previously established model involving pathogenic CAde-rived from clonal B cell expansions (6, 12, 18, 25) to permit acomparison with nonpathogenic anti-i/I CA. The observeddifferences in V region gene use by naturally occurring vs patho-genic CA would suggest that natural/nonpathogenic anti-i/Iautoantibodies may not necessarily represent the precursors oftheir monoclonal counterpart, and that perhaps the mecha-nisms leading to the development of the two different anti-i/Iautoantibody populations may also differ. For example, thepresence of the I antigen on L. monocytogenes raises the possi-bility that naturally occurring anti-I CAmay arise from micro-bial stimulation, similar to naturally occurring anti-A, B, andH isoagglutinins, while pathogenic, monoclonal anti-i/I CAmay primarily arise by alternative mechanisms (37, 38).

This work was supported by a Transfusion Medicine Academic Awardfrom the National Heart, Lung, and Blood Institute and the NationalInstitutes of Health (l-KOH-HL-02673-01, a National Institutes ofHealth First Award (R29 DK39065), as well as by a North AtlanticTreaty Organization Collaborative research grant.

References

1. Carson, D. A., P. P. Chen, R. I. Fox, T. J. Kipps, F. Jirik, R. D. Goldfein, G.Silverman, V. Radoux, and S. Fong. 1987. Rheumatoid factors and immunenetworks. Annu. Rev. Immunol. 5:109-126.

2. Madaio, M. P., A. Schattner, M. Shattner, and R. S. Schwartz. 1986. Lupusserum and normal human serum contain anti-DNA antibodies with the sameidiotypic marker. J. Immunol. 137:2535-2540.

3. Pruzanski, W., and K. Shumak. 1977. Biologic activity of cold-reactingantibodies. Parts I and II. N. Engl. J. Med. 297:538-542 and 583-589.

4. Roelcke, D. 1989. Cold agglutination. Transfus. Med. Rev. III. 2:140-166.5. Silverman, G., and D. Carson. 1990. Structural characterization of human

monoclonal cold agglutinins: evidence for a distinct primary sequence-definedVH4 idiotype. Eur. J. Immunol. 20:351-356.

6. Silberstein, L. E., L. C. Jefferies, J. Goldman, D. F. Friedman, J. S. Moore,P. C. Nowell, D. Roelcke, W. Pruzanski, J. Roudier, and G. J. Silverman. 1991.Variable region gene analysis of pathologic human autoantibodies to the related iand I red blood cell antigens. Blood. 78:2372-2386.

7. Leoni, J., J. Ghiso, F. Goni, and B. Frangione. 1991. The primary structureof the Fab fragment of protein KAU, a monoclonal immunoglobulin Mcoldagglutinin. J. Biol. Chem. 266:2836-2482.

8. Pascual, V., K. Victor, D. Lelsz, M. B. Spellerberg, T. J. Hamblin, K. M.Thompson, I. Randen, J. Natvig, J. D. Capra, and F. K. Stevenson. 1991. Nucleo-tide sequence analysis of the V regions of two IgM cold agglutinins: evidence thatthe VH4-2 1 gene segment is responsible for the major cross-reactive idiotype. J.Immunology. 146:4385-4391.

9. Pascual, V., K. Victor, M. Spellerberg, T. J. Hamblin, F. K., Stevenson, andJ. D. Capra. 1992. VH Restriction among human cold agglutinins. J. Immunol.149:2337-2344.

10. Jefferies, L., J. Gregg, G. Silverman, C. Carchidi, and L. Silberstein. 1992.Idiotypic heterogeneity of VKIII autoantibodies to red blood cell antigens. Clin.Immunol. Immunopathol. 65:119-128.

11. Stevenson, F., M. Wrightham, M. Glennie, D. Jones, A. Cattan, T. Feizi,T. Hablin, and G. Stevenson. 1986. Antibodies to shared idiotypes as agents foranalysis and therapy for human B cell tumors. Blood. 68:430-436.

12. Silberstein, L. E., J. Goldman, J. A. Kant, and S. L. Spitalnik. 1988.Comparative biochemical and genetic characterization of clonally related humanB-cell lines secreting pathogenic anti-Pr2 cold agglutinins. Arch. Biochem.Biophys. 264:244-252.

13. McKearn, T. 1980. Cloning of hybridoma cells by limiting dilution influid phase. In Monoclonal Antibodies: Hybridomas. A NewDimension in Bio-logic Analysis. Plenum Press, NewYork. pp. 374-382.

14. Petz, L., and D. Branch. 1988. Serological Tests for the Diagnosis ofImmune Hemolytic Anemias. Churchill Livingstone, NewYork.

15. Goldberg, L. S., and E. V. Barnett. 1967. Mixed gammaG-gammaMcoldagglutinins. J. Immunol. 99:803-809.

16. Sanz, I., P. Kelly, C. Williams, S. Scholl, P. Tucker, and J. D. Capra. 1989.The smaller human VH gene families display remarkably little polymorphism.EMBO(Eur. Mol. Biol. Organ.) J. 8:3741-3748.

17. Silverman, G. J., M. Sasano, and S. B. Wormlsey. 1993. The variable-re-gion specificity of bacterial Fab-binding protein: the search for B-cell superanti-gens. Immunomethods. 2:17-32.

18. Friedman, D. F., E. A. Cho, J. Goldman, C. E. Carmack, E. C. Besa, R. R.Hardy, and L. E. Silberstein. 1991. The Role of clonal selection in the pathogene-sis of an autoreactive human B cell lymphoma. J. Exp. Med. 174:525-537.

19. Sanger, F., S. Nicklen, and A. Coulson. 1977. DNAsequencing with chainterminating inhibitors. Proc. Natl. Acad. Sci. USA. 74:5463-5467.

20. Silberstein, L. E., L. C. Jefferies, J. Goldman, and S. L. Spitalnik. 1989.Production of carbohydrate-specific human monoclonal antibodies in vitro.Methods Enzymol. 179:299-304.

21. Petz, L., and G. Garratty. 1980. Acquired Hemolytic Anemia. ChurchillLivingstone, NewYork.

22. Kuppers R., U. Fischer, K. Rajewsky, and A. Gause. 1992. Immunoglobu-lin heavy and light chain gene sequences of a human CD5 positive immunocy-toma and sequences of four novel VHIII germline genes. Immunol. Lett. 34:57-62.

23. Schroeder, H. W., and J. Y. Wang. 1990. Preferential utilization of con-served immunoglobulin heavy chain variable gene segments during human fetallife. Proc. Nail. Acad. Sci. USA. 87:6146-6150.

24. Pech, M., and G. Zachau. 1984. Immunoglobulin genes of different sub-groups are interdigitated with the V. locus. Nucleic Acids Res. 12:24-29.

25. Silberstein, L. E., S. Litwin, and C. E. Carmack. 1989. Relationship ofvariable region genes expressed by a human B cell lymphoma secreting pathologicanti-Pr2 erythrocyte autoantibodies. J. Exp. Med. 169:1631-1643.

26. Olee, T., E. Lu, D. Huang, R. Soto-Gil, M. Deftos, F. Kozin, D. Carson,and P. Chen. 1992. Genetic analysis of self-associating immunoglobulin Grheu-matoid factors from two rheumatoid synovia implicates an antigen-driven re-sponse. J. Exp. Med. 175:831-842.

27. Scott, M., D. Crimmins, W. McCourt, G. Chung, K. Schable, R. Thiebe,E. Quenzel, H. Zachau, and M. Nahm. 1991. Clonal characterization of thehuman IgG antibody repertoire to haemophilus influenzae type b polysaccharide.J. Immunol. 147:4007-4028.

28. Paul, E., A. Livneh, L. Manheimer, and B. Diamond. 1991. Characteriza-tion of the human IG VJI gene family and analysis of VJI and C, polymorphismin systemic lupus erythematosus. J. Immunol. 147:27771-2776.

29. van Es, J. H., F. H. J. G. Meyling, W. R. M. van de Akker, H. Aanstoot,


R. H. W. M. Derksen, and T. Logtenberg. 1991. Somatic mutations in the vari-able regions of a human IgG anti-double-stranded DNAautoantibody suggest arole for antigen in the induction of systemic lupus erythematosus. J. Exp. Med.173:461-470.

30. Shlomchik, M. J., A. H. Aucoin, D. S. Pisetsky, and M. G. Weigert. 1987.Structure and function of anti-DNA autoantibodies derived from a single au-toimmune mouse. Proc. Natl. Acad. Sci. USA. 84:9150-9154.

31. Sharon, J., M. Gefter, L. Wysocki, and M. Margolies. 1989. Recurrentsomatic mutations in mouse antibodies to p-azophenylarsonate increase affinityfor hapten. J. Immunopharmacol. 142:596-601.

32. Ichihara, Y., H. Matsuoka, and Y. Kurosawa. 1988. Organization of hu-man immunoglobulin heavy chain diversity gene loci. EMBO(Eur. Mol. Biol.Organ.) J. 7:4141-4150.

33. Ravetch, J. V., U. Siebenlist, S. Korsmeyer, T. Waldmann, and P. Leder.1981. Structure of the human immunoglobulin A locus: characterization of em-

variable region genes: organization, polymorphism, and expression. Adv. Im-munol. 49:1-74.

35. Pascual, V., J. Andris, and J. D. Capra. 1990. Heavy chain variable regiongene utilization in human antibodies. Int. Rev. Immunol. 5:231-238.

36. Adderson, E. E., P. G. Shackelford, A. Quinn, P. M. Wilson, M. W.Cunningham, R. A. Insel, and W. L. Carroll. 1993. Restricted immunoglobulinVHusage and VDJ combinations in the human response to Haemophilus influen-zae type b capsular polysaccharide. Nucleotide sequences of monospecific anti-Haemophilus antibodies and polyspecific antibodies cross-reacting with self anti-gens. J. Clin. Invest. 91:2734-2743.

37. Costea, N., V. Yakulis, and P. Heller. 1965. Experimental production ofcold agglutinins in rabbits. Blood. 26:323-340.

38. Springer, G. F., and R. E. Horton. 1969. Blood group isoantibody stimula-tion in man by feeding blood group-active bacteria. J. Clin. Invest. 48:1280-1291.

Documents

Naturally Occurring Agglutinins MaybeEncoded byDifferent ......Naturally OccurringAnti-i/l ColdAgglutinins MaybeEncoded byDifferent VH3GenesasWellasthe VH4.21 GeneSegment Leigh C