Proc. Nat. Acad. Sci. USAVol. 71, No. 12, pp. 5109-5114, December 1974Symposium

The Organzation and Diversity of Immunoglobulin Genes*

PHILIP LEDER, TASUKU HONJO, SEYMOUR PACKMAN, DAVID SWAN, MARION NAU,AND BARBARA NORMAN

Laboratory of Molecular Genetics, National Institute of Child Health and Human Development, National Institutes ofHealth, Bethesda, Maryland 20014

ABSTRACT We have used purified mouse ih auno-globulin light chain mRNA and synthetic DNA dich iscomplementary to it to assess the reiteration frequency ofgene sequences corresponding to the K constant region ofthe mouse immunoglobulin light chain. These studiesindicate that the constant region sequence is representedonly two to three times per haploid mouse genome, afinding that rules out a simple stringent germ linemechanism which would require the constant regionsequence to be represented hundreds if not thousands oftimes. Hybridization studies involving laI-labeled my-eloma light chain mRNA yield interesting results whichmay eventually permit us to distinguish between the re-maining somatic mutation and recombinational germline hypotheses. These results reveal a major componentof relatively unique frequency and a minor componentwith a reiteration frequency of approximately 30 to 50copies per haploid genome. As discussed, these results donot permit us to distinguish unambiguously between agerm line model and a type of somatic mutation modelthat permits germ line genes corresponding to each Ksubgroup. The results do, however, clearly rule out the ex-istence of thousands of variable region sequences soclosely related to the MOPC-41 V-region as to permitextensive stable cross-hybridization.

Introduction

The functional responsibilities of the immune system includethe potential to respond-with great specificity-to a virtuallylimitless array of antigens. Almost as a minimum, one mustpostulate that the immune system has the potential to producea million different antibody molecules-each fairly specificand each subject to some form of regulation which ensuresthat the correct antibodies are produced in very large amountsin response to specific antigenic stimuli. We have been in-terested in determining how this vast amount of information isorganized in the cellular genome and how we can account, ingenetic terms, for the tremendous diversity encoded in theimmune system.Many structural and serologic studies carried out over the

last decade have provided an important insight into the wayspecificity and, thereby, diversity are built into the antibodymolecule. A given class of antibody molecules can haveabsolutely identical amino-acid sequences in their C-terminalhalves, but all differ in their N-terminal regions. It is this

Abbreviations: cDNA, complementary DNA; Cot (or Crt) is de-fined as the concentration of DNA (or RNA) in annealing reac-

tion X the time of incubation and is expressed as moles X sec/liter; C-region, constant region of immunoglobulin; V-region,variable region of immunoglobulin.* This paper was presented at the Annual Meeting of the NationalAcademy of Sciences in Washington, D.C. on April 22-24, 1974.It was presented in A Symposium on Organization and Tranmcrip-tion of the Eukaryotic Genome.

5109

variable portion of heavy and light chains that forms theantibody combining site and, hence, gives the antibody mole-cule its unique specificity. The major questions, then, con-fronting modern immunology are how this diversity arises,how immunoglobulin genes are organized, and how theirexpression is regulated.A number of distinguished geneticists and immunologists

have provided a variety of very useful genetic models to ac-count for the diversity of antibody molecules (see refs. in 1).These models fall into three general categories which aredepicted diagrammatically in Table 1. Our representation ofthese models is based on the reasonable assumption thatabout 1000 light chains and 1000 heavy chains could com-bine to give a million different antibody molecules whichwould represent a reasonable antibody repertoire (2). Thesimplest, and in many ways most appealing model, is astringent germ line hypothesis which holds that each immuno-globulin subunit, light and heavy chain (we shall restrict ourarguments to the light chain, inasmuch as analogous argu-ments hold for the heavy chain) is represented, intact in thegenome of every cell of the organism. Thus, 1000 light chainsequences would require 1000 variable region genetic se-quences immediately adjacent to the 1000 constant regiongenetic sequences. In contrast to this straightforward model,the remaining two major hypotheses, the recombinational germline hypothesis and the somatic mutation hypothesis, have in-corporated a critical feature originally suggested by Dreyerand Bennett (3), namely, that variable and constant regionsare encoded by two separate genes which are (or whose prod-ucts are) ultimately joined to one another to form an intactimmunoglobulin sequence. These models differ from thestringent germ line hypothesis in that they require only one orvery few copies of the constant region gene sequence. They dif-fer from one another in that the recombinational germ linehypothesis holds that immunoglobulin diversity has arisenthrough evolutionary processes so that the variable regionsare represented individually in the germ line and thus in everycell of the organism. The somatic mutation hypothesis, incontrast, holds that diversity has arisen-at least in part-during somatic differentiation; and, therefore, only very fewcopies of the variable region sequences are required in thegerm line since these will differentiate during somatic develop-ment.The three models differ clearly from one another in the

number of constant and variable gene sequences which theyrequire. The stringent germ line hypothesis requires manycopies of the constant region sequences, the remaining hy-potheses require very few. The recombinational germ line hy-pothesis and the somatic mutation hypothesis differ from oneanother in that the former requires many more copies cor-

Proc. Nat. Acad. Sci. USA 71 (1974)

TABLE 1. Predicted frequencies of light chain constant and variable gene sequences in cloned immunocyte DNA

Expected frequency as determinedby hybridization with MOPC-41

Theoretical gene frequency light chain mRNA

Variable Constant Variable Constant

Stringent germ line hypothesisDNA sourceMOPC41 VU Ck VS Ck V3 Ch ... '1000 '1000 <50 -1000Other VU Ck V2 Ck VS Ck-.. '1000 --'1000 <50 '1000

Recombinational germ line hypothesisDNA sourceMOPC-41 Va V2 V V. ... V10--o Ck -1000 <10 <50 <10Other V"1 V2 VS V...* V102no- Ck -1000 <10 <50 <10

Somatic mutation hypothesisDNA sourceMOPC41 V41 V2 VS V4 **. V10-1o2 Ck <100 <10 <<50 <10Other V1 V2 Vs V4... V10-1*2 Ck <100 <10 <<50* <10

* According to the somatic mutation hypothesis, the gene sequence corresponding to the variable region of the MOPC41 subgroup in aMOPC41 immunocyte should differ from the comparable sequence in other clones. Therefore, hybrids of the MOPC-41 variable regionmRNA and the comparable DNA regions derived from other clones should be mismatched and incomplete were this hypothesis correct.

responding to variable region sequences than the latter. Onthe basis of these expectations, it is quite clear that we couldbegin to distinguish between these models were we able toquantitate the immunoglobulin constant and variable regiongene sequences. Our studies and those of a number of otherlaboratories have, in fact, been directed towards this goal(1, 4-23). This paper reviews the progress we have made.

Experimental strategyOur approach to this problem has relied heavily on our pre-vious studies that examined the organization and regulatedexpression of mammalian globin genes (12). In previousstudies, we and others have successfully translated the im-munoglobulin light chain mRNAs derived from mousemyeloma tumors (4-9, 13-16). This work, interestingly,revealed that the immunoglobulin light chain was synthesizedin cell-free systems as a molecule longer than the mature,secreted form of the light chain, suggesting that light chainmight be synthesized as a precursor (6, 9). Since a convenientcell-free system was available for the purpose of assayingmRNA during its purification, the light chain message wasthen purified by the very simple stratagem of repeated oligo-(dT)-cellulose chromatography and sucrose gradient centrifu-gation (1, 9). The most highly purified mRNA thus obtainedfrom the MOPC41 myeloma tumor ran, in polyacrylamide gelelectrophoresis in formamide, as a single band, corresponding

TABLE 2. Diagrammatic representation of MOPC-41 mRNAand its complementary ['HJcDNA

: Variable: Constant: Light chain-4 1300 Bases i

300-: 325e Co325 b U*150A32005': U* : Pret : Variable : Constant: U* : AAAAA 3' mRNA

3' : Constant TTT 5' cDNA

630 Bases

t Precursor sequence.* Putative untranslated sequence.

to a molecular weight of approximately 450,000 or approxi-mately 1300 nucleotides (1, 19, 24).

Synthesis of DNA complementary to purified light chainmRNA

The purified mRNA thus obtained served as an efficienttemplate for the synthesis of complementary MOPC-41 DNA(cDNA) (10). As we have pointed out previously, by annealingan oligo(dT) primer to the 3' poly(A) end of the message, wewere able to phase the RNA-dependent DNA polymerase andassure that copying occurs from the 3' to the 5' end. In thecase of the light chain message, the cDNA thus synthesizedwas approximately half the length of the immunoglobulinmessage. Careful chain length determinations carried outusing authentic restriction enzyme fragments of phage Xindicated that the MOPC41 cDNA had a chain length ofapproximately 630 nucleotides. Since reverse transcription isfrom the 3' end, the cDNA probe would correspond to the C-terminal portion or constant region-coding portion of theimmunoglobulin light chain mRNA (see diagram, Table 2).We will discuss the evidence supporting this assertion below.

An additional test of the homogeneity of the immuno-globulin light chain mRNA

The availability of the ['H]cDNA probe makes it possibleto analyze the purity of the immunoglobulin mRNA prepara-tion using the technique of hybridization kinetic analysis (25,26). Under conditions in which the mRNA concentration is inexcess, the Crt value at which one-half the cDNA probe ishybridized (the Crti/i) is proportional to the relative concen-

tration of mRNA complementary to it in the reaction mix-ture. Thus, we can use purified reticulocyte globin mRNA andits complementary probe as a standard for an mRNA se-

quence of 1200 bases in length (a plus P chain) and view thehybridization of a globin cDNA probe approximately 570nucleotides in length (27). The Crtl/2 value thus obtained is3.7 X 10-4. Thus, the Crti/, value for the MOPC41 mRNAwhich is slightly longer, 1300 nucleotides, would be 4 X 10'as shown in Fig. 1. The experimentally obtained value is 4.2X 10-4, which is in very close agreement to the expected theo-

5110 Symposium: Leder et al.

Symposium: Immunoglobulin Genes 5111

retical value. Also shown is the thermal elution profile of thehybrid formed between MOPC-41 mRNA and its ['H]cDNA.The sharp melting profile with a Tm of 87° indicates that thehybrid formed is congruently base-paired with little or no

mismatching.

Cross-hybridization between MOPC-41 cDNA and other Klight chain mRNAs

We have argued above from the known mechanism of theRNA-dependent DNA polymerase and the structure of themRNA that the [3H]cDNA probe corresponds, as a mini-mum, to a major portion of the constant region of the lightchain mRNA (see Table 2). Inasmuch as the mRNA is 1300bases long, contains a 200 base poly(A) sequence, and re-

quires about 650 bases to encode the mature form of the lightchain, we are left to account for a 450 nucleotide sequence.We must assign an additional 250 bases to encode the putativelight chain precursor (9). At least 150 of these bases are locatedat the 5'-terminal portion of the translated sequence (D.McKean, T. Honjo, M. Nau, D. Swan, and P. Leder, un-

published results). There still remain 200 nucleotides to beaccounted for and this sequence, presumably untranslated,can occupy either the 3' or the 5' end of the molecule. Ineither case, the cDNA probe which is transcribed from themiddle of the poly(A) region at the 3' end of the molecule willencode, as a minimum, about 70% of the constant regionsequence and, as a maximum, will be complementary to allof it as well as a small portion of the variable region sequence.Consistent with this expectation, the MOPC-41 [8H]cDNAprobe cross-hybridizes readily to mRNA derived from a

myeloma tumor which produces a K chain of a different sub-group (Table 3). As shown, the hybridization results in pro-

tection of approximately 70% of the cDNA probe. The re-

maining, unprotected portion of the probe might correspond toa region overlapping the variable or an untranslated sequence.

Genetic representation of the constant region sequence.Two genes encoding one polypeptide chain?

We have already indicated that the most stringent germ linehypothesis would require hundreds if not thousands of copiesof the C-region sequence. Nevertheless, a number of geneticanalyses (28-35) which have examined the linkage, allotype,and crossover of heavy and light chain C-region genes havebuttressed the original arguments of Dreyer and Bennett (3)suggesting that only a few copies of the C-region gene are re-

quired in the genome. Implicit in this argument is the notionthat (at least) two genes encode one polypeptide chain.A direct test of the stringent germ line hypothesis, there-

fore, can be performed using the MOPC-41 cDNA as a

hybridization probe to assess the reiteration frequency of theconstant region gene sequence. MOPC41 [3H]cDNA was

hybridized to a vast excess of unlabeled total DNA derivedfrom MOPC-41 tumor (Fig. 2). A similar analysis using mouseglobin cDNA is shown for comparison. The result lends littleroom for ambiguity. The Cot11, value for MOPC41 cDNA is1130 as compared to a Cotl/, value of 3000 for a "uniquecopy" mouse DNA. This corresponds to a reiteration fre-quency of 2.7 copies per haploid genome. The sharp thermalelution profile (Fig. 2, inset) indicates the closely matchednature of the hybrid formed. Similar analyses were carried outusing MOPC-41 [3H]cDNA and total DNA from a X chain-producing tumor, RPC-20, and from normal mouse liver.These results gave reiteration frequencies of 3.0 and 3.2 per

10k20

30 k

n

z0)

40

60

70 F

80

90

100

-6 -5 -4 -3LOG Crt (mole sec/liter)

G)

0

z

0Dz

-o

-2 -1

FIG. 1. Hybridization of purified mouse globin and MOPC41light chain mRNAs and their respective [3H] cDNAs. Thehybridization reaction was carried out in 0.6 M NaCl-0.2 mMEDTA-15 mM Tris-HCl, pH 7.2 at 75°. mRNA is present at 10-to 100-fold excess over cDNA. Aliquots were taken at timeintervals for the assay of the hybrid formed by Si nucleasedigestion (27). Thermal elution profiles (inset) were determinedby hydroxyapatite chromatography (12).

haploid genome, respectively, and, taken together, indicatethat only two or three copies of the C-region gene are presentper haploid genome. These are represented to the same extentregardless of whether the cells are actively producing K chains.

Naturally this result is not compatible with the stringentgerm line hypothesis which requires many copies of the C-region gene. It is, however, entirely consistent with one of thecardinal notions of the Dreyer-Bennett model which requiresonly a few copies of the constant region gene. Given the con-siderable data on the differences among mouse K chain V-region amino-acid sequences, it seems clear that a minimum of25 V-region sequences (and possibly many more) must beencoded in the germ line genome. These two conclusions forceus to ask how these many V-region sequences can be joined tothe few C-region sequences which we observed. The pos-sibility that the polypeptide products of these genes are joinedis ruled out by the .fact that V- and C-regions are encoded ina single mRNA molecule (1). The remaining stages at whichthis joining could occur would be at the level of mRNA or at

TABLE 3. Cross-hybridization between MOPC-41 [3H]cDNAand light chain mRNA derived from a different subgroup

Source of Maximum extentlight chain of hybridmRNA formation, % Tm, 0C

MOPC41 93 85MPC-11 71 84.5

Hybridization reactions and assays were carried out as indi-cated in the legend to Fig. 2. Thermal denaturation was assayedby hydroxyapatite elution as described previously (12).

Proc. Nat. Acad. Sci. USA 71 (1974)

50 F

Proc. Nat. Acad. Si. USA 71 (1974)

IDm

zie

0

<Eo

0

10

20

30

40

50

60

70

80

90

10C)

20 F030 Z

40%z

50>60 x

70 Xa

SoR90 3-

K)..K0 Ko 0 13v4 105Cot (mole x sec/liter)

FIG. 2. Kinetics of annealing of MOPC-41 [3H]cDNA to a vast excess of total cellular MOPC-41 DNA. Hybridization kinetics of

MOPC-41 [EH]cDNA were carried out in the presence of up to a 1.2 X 107-fold excess of unlabeled, sheared MOPC-41 cellular DNA.(For comparison, the kinetics of hybridization, under similar conditions, of mouse globin [3H]cDNA with MOPC-41 cellular DNA areshown.) The cDNA and cellular DNA were prepared as described (10, 12), with the former having a specific activity of 4.6 X 107 dpm/,ug,and a base length of 630 as determined by alkaline sucrose gradient centrifugation. The hybridization procedure and SI nuclease assayswere as described (38), and were performed at DNA concentrations and salt concentrations appropriate for each portion of the kineticcurve (see below). The Cot values on the abscissa are those that would obtain at 0.18 M Na+ (26). Coti/2 for MOPC-41 cDNA X MOPC-41DNA equals 1130. Coti/, for globin cDNA X MOPC-41 DNA equals 2100. The filled symbols represent MOPC-41 cDNA X MOPC-41DNA: (U) 0.99 mg/ml of DNA, 0.18 M Na+; (0) 9.9 mg/ml of DNA, 0.54 M Na+; (A) 9.9 mg/ml of DNA, 1.06 M Na+. The open

symbols (0) represent globin cDNA X MOPC-41 DNA. Thermal elution profiles (inset) were determined by diluting the hybrid to 0.24M Na+, heating at each temperature for 8-10 min, and determining resistance to SI nuclease (1). In the inset, the filled symbols (0)represent MOPC-41 cDNA X MOPC-41 DNA; the open symbols (0) represent MOPC-41 cDNA X mouse spleen DNA.

the level of the gene itself. While there is no evidence favoringeither of these possibilities, efficiency argues that a singlejoining event occurring at the level of the gene would be morelikely than thousands of joining events occurring at the levelof the message. Several interesting models for genetic recom-

bination or re-arrangement of these multiple genes have beenproposed (3, 36, 37).

Somatic mutation or evolution as a source of diversity

The studies noted above focus on the genetic representation ofthe C-region sequence and, therefore, do not address the ques-

tion of origin of diversity in the V-region sequences. Never-theless, a number of studies addressing this question havebeen carried out using [(25I]mRNA derived from myelomatumors (11, 20, 22, 23). The interpretation of data obtainedfrom such studies is necessarily limited by uncertainties re-

garding the purity of the hybridization probe and/or the por-

tion of the hybridization curve representing the V- and C-regions. Inasmuch as the MOPC-41 mRNA preparation we

have used appears pure by several criteria and the uniquenature of the constant region sequence has been determined,we have carried out preliminary studies using intact 125I1labeled mRNA (19).As we have indicated previously, the remaining hypotheses,

the recombinational germ line and the somatic mutation, dif-fer from one another in that the former requires many, pos-

sibly thousands, of variable region gene sequences whereas thelatter requires few, perhaps a hundred or less. Whereas, theserequirements are quite clear, there are consequences in termsof a hybridization experiment which require qualification.Referring to Table 1 we may ask what results might be ex-

pected from hybridization analyses were either model correct?The differences in amino-acid sequences between variable

regions among the K subgroups are so great that we might notexpect stable hybrid formation between sequences derivedfrom different subgroups. Variable sequences derived fromthe same subgroup in which over 80% of the amino-acidsequences are identical, in contrast, would be expected to formstable cross-hybrids (12). Were each subgroup to include ap-proximately 50 different variable sequences (a not unreason-able expectation) and were the germ line hypothesis correct,we would expect to observe a reiteration frequency of about50 or fewer copies for the variable region. Were the somaticmutation hypothesis correct and the subgroup represented bya single germ line gene, we would expect Cot'!2 values con-sistent with a unique representation of this gene. Further-more, the somatic mutation hypothesis stipulates that thegenetic makeup of the cloned immunocytes would differ fromone another due to their having diverged by a somatic muta-tion. We might then expect that hybrids formed betweenmRNA from one clone and DNA from another would be lessstable and hybridization less extensive were the somatic muta-tion hypothesis correct. Our initial hybridization kinetic an-

alyses of ['26I]mRNA obtained from MOPC41 in the presenceof vast excesses of total cellular homologous MOPC41 DNAare shown in Fig. 3. A similar analysis using '26I-labeled mouseglobin mRNA is shown for comparison. A major portion ofthe [125I]mRNA hybridizes with a relatively unique Cot1/2(3200). In contrast to the observation of others who used lesswell characterized RNA preparations, there is no very highlyreiterated material (11, 20). The Cot curve is not obviouslybiphasic. Nevertheless, approximately 25% of the (125I-mRNA does hybridize over a Cot range of between 2 and 600-800. This material can be assigned a reiteration frequency of30 to 50 copies per haploid genome, though it is difficult to feelsecure about this number. An important difference between

5112 Sympoeium: Wer et al.

Symposium: Immunoglobulin Genes 5113

0

10

20F

.s 30D

40

I-

r-z

E 50

Ur

60_

70

80-1 0 1 2

Log COT3 4 5

FIG. 3. Kinetics of annealing of MOPC-41 [125IJmRNA to a

vast excess of total, cellular MOPC-41 DNA. Hybridizationkinetics of MOPC-41 [125I]mRNA in the presence of up to a

2.94 X 106-fold excess of unlabeled, sheared MOPC-41 cellularDNA. (For comparison are shown the results of the hybridizationof mouse globin [126I] mRNA with mouse spleen DNA.) CellularDNA was prepared as described (38). MOPC-41 mRNA was pre-

pared as described (1). Iodination was performed according to theprocedure of Prensky et al. (39); specific activity of the MOPC-41[126IJmRNA is 3.2 X 106 cpm//Ag. Hybridization was performedat 65o-67' in 1.04 M Na+- 0.02 M Tris-HCl, pH 7.0, with0.18 M Na+ used at the lower Cot values. DNA concentration was10 mg/ml, with 0.94 mg/ml used at the lower Cot values. Cotvalues on the abscissa are those that would be obtained at 0.18 MNa+ (26) in a DNA-DNA reannealing reaction. At each timepoint, aliquots were diluted 100-fold into 0.3 M NaCl-30 mMNa-citrate, and assayed (at 370 after 40 min) for resistance to60 jg/ml of RNase A, according to the procedure of Melli et al.(40). Reaction mixtures were acid-precipitated and collected on

glass fiber (GF/C) filters. Filters.were eluted with NCS (Amer-sham-Searle), and counted in solution in a standard toluene-fluormixture in a Beckman LS3250 6-scintillation spectrometer. Thesymbols represent: (0) MOPC41 [l25I]mRNA X MOPC41DNA, Coti/, equals 3200, (0) mouse globin [l21IJ-mRNA Xmouse spleen DNA, Coti/2 equals 2100.

this mRNA and that derived from globin is that the latterdoes not contain these moderately reiterated sequences. Theextent of hybridization, the amount of hybrid formed whenthe reaction is complete, is approximately 70%.We have similarly analyzed DNA derived from a myeloma

of a different subgroup, MOPC-265 (Fig. 4). The results are

quite similar: a portion of the material is moderately re-

iterated, the major component is relatively unique and theextent of hybridization is about 66%. Indeed, the thermalelution profiles of hybrids formed between MOPC-41 [125I-mRNA and MOPC-41 and MOPC-265 DNAs are indistin-guishable (Fig. 5).Such results are entirely-but not exclusively-consistent

with the recombinational germ line hypothesis. A portion ofthe mRNA is moderately reiterated, consistent with a numberof genes that would be expected to constitute a single sub-

0

10 H

20 F

.N

.0C

z(rE6-

30MOPC-41COT,/2=2400

40

50

60 -

70

80 _

-1 0 1 2 3 4 5

Log COT

FIG. 4. Kinetics of annealing of MOPC-41 [vuIlmRNA to avast excess of total, cellular MOPC-265 DNA. Hybridizationkinetics of MOPC-41 ['2II]mRNA in the presence of up to a 2.9 X106-fold excess of unlabeled, sheared MOPC-265 cellular DNA.Procedures are as described in the legend to Fig. 3. The symbolsrepresent: (0) MOPC41 ['11I]mRNA X MOPC-265 DNA,Cotl/, equals 2400, (0) globin (mouse) [lutI]mRNA X mousespleen DNA, Cotl/, equals 2100.

group. The extent of hybridization does not differ significantlybetween two cloned myeloma tumors, suggesting that se-quences are represented equivalently in each cloned line.Still there are very strong reservations to such an uncriticalinterpretation of the data. It is clear that the extent of hy-bridization can only be estimated within a wide margin oferror (10-15%) and a significant difference might easily goundetected. Furthermore, there is no direct evidence to sug-gest that the moderately reiterated RNA which we observein fact corresponds to the variable region. Preparation of V-region specific probe and careful competition experimentsusing mRNAs derived from various subgroup tumors arenecessary to establish this point.

Conclusions

Studies, in which an apparently homogeneous mRNA andits complementary synthetic DNA are used, have allowed usto quantitate the number of gene sequences corresponding tothe constant region of the immunoglobulin light chain. Theseresults indicate that -the constant region sequence is repre-sented approximately three times per haploid mouse genomeand is represented equally in a variety of tumors and organs.Such a result is inconsistent with a stringent germ line hy-pothesis which requires many, possibly thousands of, copies ofthe constant region gene sequence. Taken together with otherserologic and immunochemical data, this result lends con-siderable support to the notion that immunoglobulin subunitchains are encoded by at least two gene sequences. This con-

clusion raises the further requirement of postulating an un-

precedented biochemical mechanism for joining constant andvariable sequences either at the level of the mRNA or at thelevel of the gene itself.

MOPC-41COT,,2=3200

Proc. Nat. Acad. Sci. USA 71 (1974)

Proc. Nat. Acad. Sci. USA 71 (1974)

100o

c

cFa01)

0

-e

0

E0

80

60

40L

201

60 -70 80 90 100TEMPERATURE (OC)

110

FIG. 5. Thermal denaturation of hybrids formed betweenMOPC41 [la5I]mRNA andDNA from MOPC41 and MOPC-265tumors. Hybrids were formed in 1.04 M Na+ as described in thelegend to Fig. 3 and diluted 4-fold at the initial temperature ofthe melt curve. The reaction mixture was kept in a glass testtube in a Haake circulating water bath, and kept at each tem-

perature for 8-10 min prior to sampling. At each temperature,aliquots were withdrawn, diluted 50-fold into 0.3 M NaCl-30mMNa-citrate and assayed for resistance to 60 ;sg/ml of RNase A,as described in the legend to Fig. 4. The ordinate represents the% of the total cpm made single-stranded during the melting.

Thus far, hybridization studies in which the entire lightchain mRNA is used have not permitted us to distinguishbetween the recombinational germ line and somatic mutationhypotheses. They do indicate, however, that we should beable to make this important distinction when our technologypermits us to focus on the variable region sequence.

We are very grateful to Dr. Michael Potter for supplying theoriginal myeloma tumors used in these studies. We are alsograteful for the expert editorial assistance of Ms. CatherineKunkle in the preparation of this manuscript.

1. Honjo, T., Packman, S., Swan, D., Nau, M. Leder, P.(1974) Proc. Nat. Acad. Sci. USA 71, 3659-3663.

2. Inman, J. K. (1974) The Immune System: Genes, Receptors,Signals, eds. Sercarz, E. E., Williamson, A. R. & Fox, C. F.(Academic Press, Inc., New York), pp. 37-52.

3. Dreyer, W. J. & Bennett, J. C. (1965) Proc. Nat. Acad. Sci.USA 54, 864-869.

4. Stavnezer, J. & Huang, R. C. (1971) Nature New Biot. 230,172-176.

5. Brownlee, G. G., Harrison, T. M., Mathews, M. B. &Milstein, C. (1972) FEBS Lett. 23, 242-248.

6. Milstein, C., Brownlee, G. G., Harrison, T. M. & Mathews,M. B. (1972) Nature New Biol. 239, 117-120.

7. Stevens, R. H. & Williamson, A. R. (1972) Nature 239, 143-146.

8. Storb, U. (1972) J. Immunol. 108, 755-764.9. Swan, D., Aviv, H. & Leder, P. (1972) Proc. Nat. Acad. Sci.

USA 69, 1967-1971.10. Aviv, H., Packman, S., Swan, D., Ross, J. & Leder, P.

(1973) Nature New Biol. 241, 174-176.11. Delovitch, T. L. & Baglioni, C. (1973) Proc. Nat. Acad. Sci.

USA 70, 173-178.12. Leder, P., Ross, J., Gielen, J., Packman, S., Ikawa, Y.,

Aviv, H. & Swan, D. (1973) Cold Spring Harbor Symp.Quant. Biol. 38, 753-761.

13. Mach, B., Faust, C. & Vassalli, P. (1973) Proc. Nat. Acad.Sci. USA 70, 451-455.

14. Schechter, I. (1973) Proc. Nat. Acad. Sci. USA 70, 2256-2260.

15. Smith, M., Stavnezer, J., Huang, R. C., Burdon, J. B. &Lane, C. D. (1973) J. Mol. Biol. 80, 553-557.

16. Tonegawa, S. & Baldi, I. (1973) Biochem. Biophys. Res.Commun. 51, 81-87.

17. Bernardini, A. & Tonegawa, S. (1974) FEBS Lett. 41, 73-77.18. Faust, C. H., Diggelman, H. & Mach, B. (1974) Proc. Nat.

Acad. Sci. USA 71, 2491-2495.19. Leder, P., Honjo, T., Packman, S., Swan, D., Nau, M. &

Norman, B. (1974) The Immune System: Genes, Receptors,Signals, eds. Sercarz, E. E., Williamson, A. R. and Fox,C. F. (Academic Press, Inc., New York), pp. 299-317.

20. Premkumar, E., Shoyab, M. & Williamson, A. R. (1974)Proc. Nat. Acad. Sci. USA 71, 99-103.

21. Rabbitts, T. H., Bishop, J. P., Milstein, C. & Brownlee,G. G. (1974) FEBS Lett. 40, 157-160.

22. Storb, U. (1974) Biochem. Biophys. Res. Commun. 57, 31-38.23. TQnegawa, S., Bernardini, A.' Weimann, B. J. & Steinberg,

C. (1974) FERS Lett. 40, 92-96.24. Staynov, D. Z., Pinder, J. C. & Gratzer, W. B. (1972)

Nature New Iiol. 235, 108-110.25. Britten, R. J. & Kohne, D. E. (1968) Science 161, 529-540.26. Britten, R. J. & Smith, J. (1970) Carnegie Inst. Washington

Yearb. 68, 378-386.27. Gielen, J., Aviv, H. & Leder, P. (1974) Arch. Biochem.

Biophys. 163, 146-154.28. Liberman, R. & Potter, M. (1966) J. Mol. Biol. 18, 516-528.29. Hamers, R. & Hamers-Casterman, C. (1967) Cold Spring

Harbor Symp. Quant. Biol. 32, 129-132.3Q. Koshland, M. (1967) Cold Spring Harbor Symp. Quant. Biol.

32, 119-127.31. Natvig, J. B., Kunkel, H. G. & Litwin, S. D. (1967) Cold

Spring Harbor Symp. Quant. Biol. 32, 173-180.32. Herzenberg, L. A. & McDevitt, H. 0. (1968) Annu. Rev.

Genet.2, 209-244.33. Kunkel, H. G., Natvig, J. B. & Joslin, F. G. (1969) Proc.

Nat. Acad. Sci. USA 62, 145-149.34. Tosi, S. L., Mage, R. G. & Dubiski, S. (1970) J. Immunol.

104, 641-647.35. Mage, R. G., Young-Cooper, G. 0. & Alexander, C. (1971)

Nature New Biol. 230, 63-6.36. Gally, J. A. & Edelman, G. M. (1970) Nature 227, 341-348.37. Smithies, 0. (1970) Science 169, 882-883.38. Packman, S., Aviv, H., Ross, J. & Leder, P. (1972) Biochem.

Biophys. Res. Commun. 49, 813-819.39. Prensky, W., Steffensen, D. M. & Hughes, W. L. (1973)

Proc. Nat. Acad. Sci. USA 70, 1860-1864.40. Melli, M., Whitfield, C., Rao, K. V., Richardson, M. &

Bishop, J. 0. (1971) Nature New Biol. 231, 8-12.

I ...

['21]mRNA DNA41-41 -41-265 o

0

~~90@_T,,,=87'

00Go

0~C%

R14 Symposium: Leder et cd.