Proc. Nati. Acad. Sci. USAVol. 73, No. 7, pp. 2476-2480, July 1976Immunology

Phylogenetic origins of immune recognition: Lymphocyte surfaceimmunoglobulins in the goldfish, Carassius auratus

(thymus/spleen/lactoperoxidase/polyacrylamide gel electrophoresis/immunofluorescence)

GREGORY W. WARR, DOMINICK DELUCA, AND JOHN J. MARCHALONISWalter and Eliza Hall Institute of Medical Research, P.O. Royal Melbourne Hospital, Melbourne 3050, Australia

Communicated by F. M. Burnet, April 21, 1976

ABSTRACT Membrane immunoglobulin (Ig) of splenocytesand thymocytes of the goldfish, Carassius auratus, was dem-onstrated by indirect fluorescent-antibody techniques. Obser-vations on shedding and resynthesis indicated that the thymo-cyte Ig was endogenously produced. The lymphocyte surfaceproteins were radioiodinated using the lactoperoxidase-cata-lyzed reaction, and the labeled Ig molecules were isolated byspecific precipitation and analyzed by sodium dodecyl sul-fate/polyacrylamide gel electrophoresis. The IgM-like mem-brane Igs of splenocytes and thymocytes were shown to differin their ease of solubilization with nonionic detergent, and inthe sodium dodecyl sulfate/electrophoretic mobility of theirheavy chains. The significance of these observations for theevolution of T-cell recognition is discussed.

Although it is likely that surface immunoglobulin (Ig) functionsin antigen-recognition by mammalian B (bone marrow-de-rived) and T (thymus-derived) cells, it is generally difficult todetect Ig on T cells by direct binding assays (1, 2). In contrast,thymus lymphocytes of larval amphibians (3), adult teleost fish(4), skates (5), and urodeles (6) possess surface immunoglobulindemonstrable by immunofluorescence, although present evi-dence (7, 8) makes it reasonable to presume that the thymusfunctions in a similar manner in these diverse vertebrate species.Investigations of teleost lymphocytes cast certain problems ofimmunoglobulin display into sharp relief because these speciespossess lymphocytes capable of performing both T-cell func-tions,.such as allograft rejection (9), and the B-cell function ofantibody production. Moreover, teleosts exhibit a "hapten-carrier" effect on secondary stimulation (10) which is suggestiveof collaboration between antigen-specific T and B cells.The present study was designed to ascertain the presence and

nature of surface Ig molecules on thymus and spleen lympho-cytes of the goldfish Carassius auratus, a teleost fish whichcontains only one type of serum Ig. This molecule is composedof polypeptide chains similar to those of IgM from mammals(11), but exists as a tetramer in the serum of Cyprinidae (11, 12).We provide evidence that both splenic and thymic lymphocytesexhibit surface Ig detectable by immunofluorescence. Fur-thermore, we describe partial characterization of these mole-cules isolated from cells which were surface radioiodinated ina lactoperoxidase-catalyzed reaction (13).

MATERIALS AND METHODSAnimals. Goldfish, less than 12 months old, were obtained

from commercial suppliers.Media. The media used, phosphate-buffered saline pH 7.3,

Eisen's balanced salt solution (EBSS) and RPMI 1640, were each

Abbreviations: EBSS, Eisen's balanced salt solution; NP 40, NonidetP 40; GARG, goat anti-rabbit globulin serum; SARG, sheep anti-rabbitgamma globulin; FITC, fluorescein isothyocyanate; RAGF IgM, rabbitanti-goldfish immunoglobulin M; NaDodSO4/electrophoresis, poly-acrylamide gel electrophoresis in the presence of sodium dodecylsulfate-containing buffers.

adjusted to the osmolality of goldfish plasma, 242 milliosmoleper kg. RPMI 1640 was buffered to pH 7.3 with 13 mM 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid.

Preparation of Cells. The thymus lobes were teased intoEBSS with a pair of needles. After clumps were allowed to settle,thymocytes were washed twice in EBSS. A viability of 95% (asjudged by eosin dye exclusion) was routinely obtained. Spleenswere pressed gently through an 80 gauge stainless steel sieveinto EBSS. The erythrocytes and the majority of dead cells wereremoved by centrifugation over a solution of Ficoll and sodiummetrizoate, density 1.088 g/ml (14) for 15 min at 2000 X g.Cells retained at the interface contained <5% erythrocytes andwere >90% viable. Recovery of total viable cells was approxi-mately 95%. Splenocytes were washed twice in EBSS beforeuse.

Antigens and Antisera. Goldfish IgM was prepared by starchblock electrophoresis and gel filtration on Sephadex G-200 (11),and was free of contamination as judged by immunoelectro-phoresis and sodium dodecyl sulfate/polyacrylamide gelelectrophoresis. Antiserum was raised in rabbits (11) and thegamma globulin fraction was prepared by two precipitationsin 40% saturated ammonium sulfate followed by gel filtrationon Sephadex G-200. The rabbit anti-goldfish IgM globulin(RAGF 1gM) used in this study was specific for light chains and,t chains. Goat anti-rabbit gamma globulin serum (GARG) waspurchased from Commonwealth Serum Laboratories (Mel-bourne, Australia). Fluorescein conjugated goat anti-rabbitgamma globulin (FITC-GARG) was purchased from CappelLaboratories, Downington, Pa. This reagent had a proteinconcentration of 3.1 mg/ml, and a fluorescein/protein ratio of1.6. Fluorescein conjugated sheep anti-rabbit gamma globulin(FITC-SARG), obtained from the Institute Pasteur, Paris, witha protein concentration of 45 mg/ml, and fluorescein/proteinratio of 7, was used interchangeably with the FITC-GARG.Immunofluorescence Studies. All cell suspensions were

prepared and maintained in EBSS supplemented with 5% fetalcalf serum at 00, unless otherwise stated. Cells (1 to 10 x 106)were incubated in 0.2 ml of RAGF IgM globulin (20,ug/ml) for30 min at 00, washed four times with 1 ml of medium eachtime, and incubated for 30 min with FITC-GARG (1 in 20dilution) or FITC-SARG (1 in 50 dilution) in a final volume of0.2 ml of medium. After five washes in medium, the cells werescored either immediately, or after post-fixation in 1% para-formaldehyde in cacodylate buffer, at pH 7.0, and storage forup to 1 week. Cells were scored using a Leitz Orthoplan mi-croscope equipped with the following: a Ploem vertical illu-minator, a mercury arc lamp (Osram HBO 200/4); and BG 38,G 475, and two KP 490 filters, plus a TK 510 dichroic mir-ror.

Surface Labeling, Extraction, and Characterization ofImmunoglobulin. Membrane proteins of Iymphocytes wereradioiodinated by the method of Marchalonis et al. (13),

2476

Dow

nloa

ded

by g

uest

on

Aug

ust 2

5, 2

021

Proc. Natl. Acad. Sci. USA 73 (1976) 2477

YL

a

b IaL*-w,

C T

d IFIG. 1. Analysis by NaDodSO4electrophoresis of high-molecular-weight Ig from goldfish serum. (a) Human myeloma IgM (Frymac, gl, lambda

chains) mixed with pooled human IgG. Note two light chain bands, probably representing kappa and lambda chains (22). (b) Mouse myelomaIgM from MOPC 104E cells. (c) and (d) Goldfish high-molecular-weight Ig. Samples a, b, and c were reduced with 2-mercaptoethanol, and allsamples were analyzed on gels of 10% acrylamide and 0.25% bisacrylamide concentration.

modified for use with small cell numbers (5 to 10 X 106 perreaction mixture). About 30-50% of the added counts were

associated with the cells. Lymphocytes were then either lysedin 1% Nonidet P 40 in phosphate-buffered saline (NP 40, BDH,Melbourne, 1.0 ml per aliquot of cells), or allowed to shed thelabeled membrane proteins during a period of incubation inmedium. For this process of metabolic release cells were sus-

pended in 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acidbuffered RPMI 1640 plus 5% fetal calf serum (1.0 ml of com-plete medium per aliquot of cells) and incubated at roomtemperature for up to 5 hr. The viability of cells was not ob-served to drop significantly during labeling, but dropped byup to 15% during the period of metabolic release. After removalof cells and cell debris by centrifugation, up to 96% of thecell-associated radioactivity was obtained in the supernatantboth by detergent lysis and metabolic release. The extracts weredialyzed overnight at 40 against two changes of 2 liters ofphosphate-buffered saline and the labeled immunoglobulinmolecules were specifically precipitated with a sandwich sys-tem. The dialyzed extracts of surface-labeled cells were cen-

trifuged (12,000 X g for 20 min) and to 100 to 200-,gl aliquotswas added 100 Ail of RAGF 1gM globulin or normal rabbitgamma globulin (0.2 mg/ml of phosphate-buffered saline), andthe samples were mixed and incubated at 370 for 2 hr. Onehundred microliters of GARG (1:4 in phosphate-buffered sa-

line) was then added, mixed, incubated at 370 for 2 hr, and at40 overnight. The precipitates were washed exactly as describedin detail elsewhere (15).

Polyacrylamide Gel Electrophoresis in the Presence ofSodium Dodecyl Sulfate (NaDodSO4/Electrophoresis). Thiswas carried out by the procedure of Laemmli (16), exactly as

described previously (15).

RESULTSCharacterization of the polypeptide chains of serum IgThe polypeptide chains of goldfish high molecular weightserum Ig, prepared as described in Materials and Methods,were analyzed by NaDodSO4/electrophoresis (Fig. 1). Themolecule was dissociable, with reducing conditions, into oneband with a mobility slightly less than that of mammalian ,uchains, and another band with a mobility similar to that ofmammalian light chains. There was some apparent hetero-geneity in the mobility of the goldfish light chains (Fig. Ic)which resembled that shown by human light chains (Fig.la).

Immunofluorescence studies of lymphocyte membraneIgA sandwich fluorescence technique showed that up to 84% ofsplenocytes and 99% of thymocytes were strongly positive formembrane Ig (Table 1). A dilution of RAGF IgM globulin out

to 4 Atg/ml still gave positive fluorescence in 99% of thymocytes.The appropriate controls, including omission of RAGF Ig4,using an indifferent rabbit antiserum (anti-mouse Ig), and ab-sorbing the RAGF IgM globulin with goldfish IgM (Table 1)demonstrated the specificity of the staining. To exclude thepossibility that the thymocyte Ig may have been passively ab-sorbed from serum, experiments were carried out to determinewhether or not surface Ig on thymocytes would reappear aftershedding of fluorescent-labeled Ig complexes during culture.

After the sandwich labeling reaction, thymocytes were cul-tured for 24 hr at approximately 20° (incubation at 370 resultedin poor shedding) in RPMI 1640 plus 10% fetal calf serum. Thecells were then divided into two groups. One group (Shed) wasfixed and scored for the presence of Ig positive cells. The secondgroup (Re-add), consisting of living cells, was treated once morewith RAGF IgM and FITC-SARG. The Re-add group testedthe capacity of the cells to regenerate membrane Ig after lossor "shedding" induced by the anti-Ig reagents. The results ofthese experiments (Table 2) implied that thymocytes werecapable of resynthesizing membrane Ig, and that this processwas sensitive to inhibition of protein synthesis with cyclohexi-mide.NaDodSO4/electrophoresis analysis of lymphocytemembrane IgThe characterization of the lymphocyte surface moleculesrecognized by the rabbit anti-goldfish IgM globulin was un-

Table 1. Goldfish lymphocyte membraneimmunoglobulin detected by immunofluorescence

% PositiveCell type First antibody* cells

Splenocytes RAGF IgMt 68; 84Splenocytes None <1Thymnocytes RAGF IgM 96; 99; 91;

98; 99; 82Thymocytes FITC-RAM IgM4 <1Thymocytes RAGF IgM

absorbed withgoldfish IgM§ < 1

* FITC-GARG (fluorescein isothyocyanate conjugated goat anti-rabbit gamma globulin), as a 1:20 dilution, was the secondantibody in all groups.

t RAGF IgM (rabbit anti-goldfish IgM globulin), at 0.02 mg/ml.I FITC-RAM IgM (fluorescein isothyocyanate conjugated rabbitanti-mouse IgM), as a 1:20 dilution.

§ Absorbed for 30 min at 370 and 13 jtg/ml with goldfish IgM at 1mg/ml before chilling and addition of cells to give a final con-centration of RAGF IgM of 6 Fg/ml. Control cells were incubatedwith RAGF IgM treated exactly the same way except that goldfishIgM was replaced by EBSS, and gave a value of 98% fluorescencepositive cells.

Immunology: Warr-et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

5, 2

021

Proc. Nati. Acad. Scd. USA 73 (1976)

'ii

0*5

RELATIVE MOBILITY

CY

x

E

B

is I-

101-

sF

0 0-5

RELATIVE MOBILITY

1-0

FIG. 2. Analysis of 125I-labeled splenocyte surface Ig by NaDodSO4electrophoresis. Material precipitated by RAGF IgM plus GARG (-*)or with NRG plus G-ARG (O--- -0) was analyzed with nonreducing conditions on gels of 5% acrylamide and 0.125% bisacrylamide concentration,or with reducing conditions on gels of 10% acrylamide and 0.25% bisacrylamide concentration. (a) Splenocyte surface Ig, extracted by NP 40,and analyzed by nonreducing conditions. (b) Splenocyte surface Ig, extracted by NP 40, and analyzed with reducing conditions. Immunoglobulin

standards were: unreduced, pooled human IgG; reduced, pooled human IgG and mouse myeloma IgM (from MOPC 104E cells) mixed together.NRG refers to normal rabbit globulin.

dertaken by NaDodSO4/electrophoresis analysis of the solu-bilized, 125I-labeled, membrane molecules which were boundby these antibodies. Quantitation of the labeled splenocyte andthymocyte membrane immunoglobulin precipitated in a

sandwich system is shown in Table 3. These data suggest thatimmunoglobulins constituted up to 3% of the labeled mem-brane proteins extracted from splenocytes by NP 40 and fromthymocytes by metabolic release. It is also apparent from thesedata (Table 3) that NP 40 lysis was far more efficient at solu-bilizing immunoglobulin from the membrane of splenocytesthan that of thymocytes. The labeled material present in theprecipitates was subjected to NaDodSO4/electrophoresisanalysis. Splenocyte surface immunoglobulin, analyzed undernonreducing conditions, was resolved into two peaks (Fig. 2A),

one with a mobility similar to that of intact human IgG, and onewith a faster mobility. On reduction, the splenocyte immuno-globulin resolved into two peaks with mobilities similar to thoseof mammalian ,u and light chains (Fig. 2B). In addition, otherpeaks were seen on these gels (Fig. 2B). A peak of radioactivitywith a mobility faster than mammalian y chain was routinelyobserved in the specific, but not the nonspecific precipitates.There was also observed, in the specific precipitate, a peak ofradioactivity with a relative mobility of around 0.1, which,because of its inconsistent nature, was not further investi-gated.The precipitated thymocyte Ig, when analyzed with re-

ducing conditions, was (Fig. 3) resolved into three major peaks.One possessed a mobility significantly faster than that of

Ic

cm

xsE

-j1.0

4

L.

2478 Immunology: Waff et al.

r

i

Dow

nloa

ded

by g

uest

on

Aug

ust 2

5, 2

021

Proc. Natl. Acad. Sci. USA 73 (1976) 2479

Table 2. Shedding and reappearance of membraneIg on cultured thymocytes

Percentage fluorescence positive cells

Preincubation Incubated for 24 hrcontrol Shed Re-add

Medium alone 96; 96 8; 30 96; 94Medium pluscycloheximide(100 Ig/ml) 8;16 40; 69

The results of two separate experiments are shown. Pre-incuba-tion control: cells treated with RAGF IgM plus F1TC-SARG,washed, and scored for Ig positive cells without culturing. Shed:cells treated with RAGF IgM plus FITC-SARG, washed, and cul-tured for 24 hr before scoring for Ig positive cells. Re-add: cellstreated with RAGF IgM plus FITC-SARG, washed, cultured for 24hr before the re-addition of RAGF IgM plus FITC-SARG, and thenscored for Ig positive cells. Duplicate cultures of the Shed and Re-add groups were incubated in the presence of cycloheximide (100,gg/ml) for the duration of the culture period.

mammalian ,u chains and, therefore, faster than splenocyteheavy chain, and another showed a mobility similar to that ofmammalian light chains. In addition, a third peak of a mobilityslightly faster than mammalian y chain (relative mobility 0.5)was also observed in all patterns. A molecule similar in mobilityto this component (which appeared to be present on both thy-mocytes and splenocytes, see Figs. 2B and 3) has been isolatedfrom the membrane of mammalian lymphocytes, where it hasbeen observed to associate with antigen/antibody complexes(15).

DISCUSSIONThe present results show that goldfish lymphocytes of both thespleen and thymus possess surface Ig which is readily demon-strable by fluorescent antibody techniques, and that thismembrane Ig can be extracted and characterized. Three salientpoints of this study merit comment.

First, rabbit antibodies directed against goldfish tetramericserUm IgM bound directly to both thymus and spleen lym-

phocytes. The ease of detection of goldfish thymus Ig contrasts

X L

7-

6

03

2

0 0.5 10RELATIVE MOBILITY

FIG. 3. Analysis of W25I-labeled thymocyte surface Ig by Na-DodSO4/electrophoresis. Surface labeled material extracted fromgoldfish thymocytes by metabolic release and precipitated byRAGFIgM plus GARG (o--) or NRG plus GARG (o--- -o) was analyzedwith reducing conditions on gels of 10% acrylamide and 0.25% bisac-rylamide concentration. Ig standards were the same as in the legendto Fig. 2. An essentially similar pattern was obtained when thymocyteIg extracted by NP 40 was analyzed under the same conditions.

Table 3. Precipitation of immunoglobulin from extractsof surface "'5I-labeled lymphocytes

Percentage ofnondialyzable radio-

activity precipitated by*

Extraction RAGF IgM NRG plusCell conditions plus GARG GARG

Splenocytes NP 40 5.3 ± 0.2 2.2 ± 0.1Thymocytes NP 40 0.8 ± 0.2 0.3 ± 0.1Thymocytes Metabolic 4.5 ± 0.4 1.5 ± 0.1

release

* Values expressed as mean 4 standard error for between three andsix replicates. The total nondialyzable radioactivity in the ex-tracted eell surface material was: splenocytes, NP 40 extract, 8 x106 cpm; thymocytes, NP 40 extract, 12 x 106 cpm; thymocytes,metabolic release, 4 X 106 cpm. NRG, normal rabbit globulin.

with the situation encountered in mammals (2) and chickens(17) but is consistent with those reported for larval Xenopus (3),adult carp (4), skates (5), and urodeles (6). A number of studiesshow that the lymphocytes of cyprinoids possess the capacityto redisplay surface ig after its loss following reaction withantiglobulin reagents. (ref. 4, and present results).

Second, isolated surface Igs of thymus and spleen lympho-cytes possessed light (L) chains identical to those of serum 1gM,and a heavy (H) chain comparable in mobility on NaDodSO4/gels to mammalian ,u chains. The heavy chain of thymus surfaceIg expressed a mobility faster than that of spleen ,u chain andthis mobility was comparable to that of the T-cell heavy chainof mice (15, 18). The intact unit of spleen Ig consisted of H2L2monomers and possibly HL half molecules linked covalentlyvia disulfide bonds. It is more difficult to obtain a clear pictureof the intact state of thymus surface Ig, but preliminary resultsindicate that light chains and A-like heavy-chains can be re-solved on NaDodSO4/electrophoresis in the absence of reducingagents. If the divergence among the heavy-chains results fromamino acid sequence differences, rather than, for example,carbohydrate differences, then these results suggest that thegene specifying the T-cell heavy chain (AT) diverged from thecistron for the B cell (AB) and serum A chains very early invertebrate evolution, and preceded the divergence of the an-cestors of teleosts and mammals. The third point of note is thatuse of the nonionic detergent Nonidet P 40, under conditionsroutinely used for murine B cells (19, 20), was satisfactory forrelease of goldfish splenocyte surface Ig, but was insufficientfor quantitative isolation of thymocyte surface Ig. This moleculewas isolated more effectively from thymocytes by metabolicrelease, a result which parallels recent observations on murinethymocytes (18) and T lymphoma cells (15). The plasmamembranes of goldfish thymus and spleen lymphocytes, thus,exhibit differences comparable to those reported betweenmurine T and B cells (19). This observation suggests that thedifferentiation of lymphocytes into T and B cells was an ancientevent in vertebrate evolution.Much of the tendency to discount the existence of surface Ig

on mammalian T-cell stems from the inability to detect thesemolecules with relatively insensitive techniques such as im-munofluorescence. The present results establish that virtuallyall lymphocytes of a teleost fish express readily detectablesurface Ig even though such species possess cells capable ofperforming functions attributed to T and B cells of mammals(7). Either T and B cells both carry exposed surface Ig as a usualmembrane component, or a primitive cell combining properties

Immunology: Warr et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

5, 2

021

Proc. Nati. Acad. Sci. USA 73 (1976)

of T and B cells exists throughout life in these fish. The latteralternative might be discounted because the heavy chains ofthymus and spleen lymphocyte surface Ig differ. Furthermore,results obtained here are relevant to the controversy whetherthe antigen recognition unit on T cells is immunoglobulin oranother molecule specified by genes lying within the majorhistocompatibility locus. It has been proposed that products ofthis locus represent a more primitive recognition system thanthat mediated by Ig variable regions (21). Teleosts are thoughtto possess strong histocompatibility antigens (9) but must alsopossess surface Ig on T- and B-type cells. The minimal argu-ment is that Ig functions in the recognition of foreign antigenby both cell types. Recent evidence indicates that a similarconclusion also pertains for urodeles and cartilagenous fishes,which lack strong histocompatibility antigens (9) but expresssurface Ig on the preponderance of lymphocytes (5, 6).Therefore, it is likely that Ig molecules are involved in therecognition of antigen by lymphocytes of all vertebratespecies.

This work was supported by grants from the American Heart As-sociation (AHA 75-877) and the USPHS (CA-20085; AI-12565).G.W.W. is in receipt of a Postdoctoral Fellowship from the ScienceResearch Council (U.K.). D.D. is in receipt of a Postdoctoral Fellowshipfrom the Leukemia Society of America. We are grateful to Ms. G.Marton for her excellent technical assistance, to Drs. J. Moseley, J.Decker, J. Dixon, and T. Mandel for their advice and to the Bio-chemistry Department, Royal Melbourne Hospital for the use of anosmometer. This is Publication no. 2194 from the Walter and Eliza HallInstitute of Medical Research.

1. Marchalonis, J. J. (1975) Science 190, 20-29.2. Warner, N. L. (1975) Adv. Immunol. 19,67-216.

3. DuPasquier, L., Weiss, N. & Loor, F. (1972) Eur. J. Immunol.2,366-370.

4. Emmrich, F., Richter, R. F. & Ambrosius, H. (1975) Eur. J. Im-munol. 5,76-78.

5. Ellis, A. E. & Parkhouse, R. M. E. (1975) Eur. J. Immunol. 5,726-728.

6. Charlemagne, J. & Tournefier, A. (1975) Adv. Exp. Biol. Med.64,251-255.

7. Marchalonis, J. J. (1976) Immunity in Evolution (HarvardUniversity Press, Cambridge), in press.

8. Brown, B. A. & Cooper, E. L. (1976) Immunology 30, 299-305.

9. Hildemann, W. H. (1972) in Transplantation Antigens: Markersof Biological Individuality, eds. Kahan, B. D. & Reisfeld, R. A.(Academic Press, New York), pp. 3-73.

10. Yocum, D., Cuchens, M. & Clem, L. W. (1975) J. Immunol. 114,925-927.

11. Marchalonis, J. J. (1971) Immunology 20,161-173.12. Shelton, E. & Smith, N. (1970) J. Mol. Biol. 54,615-617.13. Marchalonis, J. J., Cone, R. E. & Santer, V. (1971) Biochem. J.

124,921-927.14. Boyum, A. (1968) Scand. J. Clin. Lab. Invest. Suppl. 97, 21,

77-89.15. Haustein, D., Marchalonis, J. J. & Harris, A. W. (1975) Bio-

chemistry 14, 1826-1834.16. Laemmli, U. K. (1970) Nature 227,680-85.17. Hudson, L. (1975) Eur. J. Immunol. 5,694-698.18. Haustein, D. & Coding, J. W. (1975) Biochem. Biophys. Res.

Commun. 65, 483-489.19. Cone, R. E. & Marchalonis, J. J. (1974) Biochem. J. 140,345-

354.20. Vitetta, E. S., Bianco, C., Nussenzweig, V. & Uhr, J. W. (1972)

J. Exp. Med. 136,81-93.21. McDevitt, H. 0. & Landy, M., eds. (1972) Genetic Control of

Immune Responsiveness (Academic Press, New York).22. Virella, G. & Coelho, I. M. (1974) Immunochemistry 11, 157-

160.

2480 Immunology: Waff et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

5, 2

021