Proc. NatL Acad. Sci. USA Vol. 78, No. 2, pp. 1138-1142, February 1981

Switch from hapten-specific to secretion in a hybrid mouse cell line (immunoglobulin genes/fluorescence-activated cell sorting) MICHAEL S. NEUBERGER AND KLAUS RAJEWSKY Institute for Genetics, University ofCologne, D-5000 Cologne 41, Federal Republic ofGermany Communicated by Herman N. Eisen, October 13, 1980 ABSTRACT From a hybrid mouse cell' line (B1-8) that se- all monoclonal anti-Bl-8.IgM idiotopic except for creted an IgM,Al anti-(4-hydroxy-3-nitrophenyl)acetyl A6-24 (unpublished data) were gifts from M. Reth (Cologne); but that had no detectable surface IgM, selection for avariant with rabbit anti-Bl-8.IgM idiotypic antiserum was a gift from T. Im- Al chains on the surface resulted in the isolation ofa line that had anishi-Kari (Cologne). Immunosorbents were made by coupling switched from la to 8 expression. The surface and secreted Igs of this line were typed as IgD with two monoclonal antibodies, and purified antibodies to CNBr-activated Sepharose (7). the parental IgM and variant IgD molecules carried the same vari- Serology, Immunofluorescence, and Cell Sorting. Radioim- able regions as judged by hapten-binding and idiotypic analysis. munoassays were performed as described (8). Purified myeloma The surface and secreted. 6 chains of the IgD variant have appar- and purified - or rhodamine-coupled goat ent molecular weights of 64,000 and 61,000, respectively. How- ever, the unglycosylated secreted 8polypeptide chain has a molec- anti-mouse class antisera were gifts from H. Dorff, B. Liese- ular weight ofonly 44,000. The secreted IgD exists-predominantly gang, and A. Radbruch (Cologne). Preparation of samples for in the 2A2 form, does not contain J , is relatively stable in the examination ofsurface or cytoplasmic Ig in the fluorescence serum, and does not fix complement. microscope and cell sorting and analysis by using a fluores- cence-activated cell sorter (FACS-1; Becton Dickinson) have IgD was first identified as a new class ofhuman myeloma pro- been described (5). tein (1); subsequently, IgD myelomas have been isolated from Purification ofCell Surface Ig. Cells were surface labeled by the rat (2) and the mouse.* The level of IgD in normal serum lactoperoxidase catalyzed radioiodination (9) and the l25I-la- from all species studied is extremely lowalthough IgD is present beled Ig was purified from a Nonidet P-40 lysate ofthe cells on as a surface Ig on a large proportion ofB cells (3). anti-8 sorbents, anti-Al sorbent, or NIP-Acap Sepharose. The function of IgD remains unclear and in part this can be Biosynthetic Labeling of 1g. Cells were preincubated for 30 ascribed to the lack ofaserum containing or acell line producing min in RPMI-1640 medium lacking methionine and then la- an IgD ofknown binding specificity. Here we describe beled by the addition of 120 ,uai ofL-[LS]methionine (specific the isolation ofa variant cell line that has switched from ,ua to 8 activity, 1100 Ci/mmol; 1 Ci = 3.7 X 10's becquerels). If re- expression and produces both surface and secreted IgD with quired, tunicamycin (gift ofR. L. Davie, Eli Lilly) was included specificity for the 4-hydroxy-3-nitrophenylacetyl (NP) hapten. in the medium (8 uag/ml) throughout. Cytoplasmic Ig (from a Nonidet P-40 lysate of-the cells after a 20-min labeling) and se- MATERIALSAND METHODS creted Ig (from. the culture supernatant after a 6-hr labeling) were purified by adsorption on NIP-Acap Sepharose. Treat- Animals. Mice were obtained from the Zentralinstitut fur ment of [3SS]IgD with endo-(3-N-acetylglucosaminidase H (Sei- Versuchstierzucht (Hannover, Fed. Rep. Germany). kagaku Kogyo, Tokyo, Japan) was carried out by incubation for Cell Lines. B1-8.64 (IgM,Al, anti-NP) was kindly donated 6 hr at 37°C with 1.5 units of enzyme per ml in 100 mM Na ci- by C. Muller (Cologne) and 11-6.3.1 cells (4) were obtained from trate at pH 5.5. the Salk Institute (San Diego, CA). B1-8.64. 1 (IgM,A1, anti-NP) Gel Electrophoresis. Samples were subjected to polyacryla- is a subelone of B1-8.64. BALB/c-derived hybrid cell line mide gel electrophoresis as described (10), and gels containing 267.7, which secretes an IgM,Al anti-NP antibody was a gift [3S]methionine were soaked in EN3HANCE (New England from M. White-Scharf (Cologne). Cells were either cultured as Fed. and dried to described (5), except that supplemented RPMI 1640 medium Nuclear, Dreieich, Rep. Germany) prior (Seromed, Munich, Fed. Rep. Germany) was used; or were autoradiography. maintained intraperitoneally in Pristane-primed mice. Cells were cloned by limiting dilution. RESULTS Antisera and Monoclonal Antibodies. Ascites fluid from H6/ 31 anti-8" tumor-bearing mice (6) was obtained from Sera-lab Isolation ofB1-8.61. B1-8 is a hybrid cell line resulting from (Sussex, England) and the IgM anti-8" antibodies were purified fusion between a spleen cell from an NP-immunized C57BL/6 by gel filtration. 11-6.3.1 anti-6b was purified from ascites fluid mouse and a cell of the myeloma line X63.Ag8 (11). B1-8 cells by (NH4)2504 fractionation. and chromatography on DE-52. secrete both an IgM, Al anti-NP antibody and the yl and K Anti-NP IgM from Bl-8.64. 1 ascites fluid and anti-NP IgD from chains ofthe X63 fusion partner. However, a B1-8 variant, Bl- B1-8.81 ascites fluid (Bl-8.IgD) were purified by binding on to 8.64, has been isolated that has lost expression ofthe X63 yl and (4-hydroxy-5-iodo-3 nitrophenyl)acetyl-aminocaproic acid (NIP- Acap)-Sepharose and elution with 1 mM NIP-Acap in phos- Abbreviations: B1-8. IgD, secreted Ig ofB1-8.81; B1-8.IgM, secreted Ig ofB1-8.64. 1; Acap, 6-aminocaproic acid; FACS, fluorescence-activated phate-buffered saline. Monoclonal anti-Al antibody (Ls136) and cell sorter; NIP, (4-hydroxy-5-iodo-3-nitrophenyl)acetyl; NP, (4-hy- droxy-3-nitrophenyl)acetyl.V region, variable region. The publication costs ofthis article were defrayed in part by page charge * Finkleman, F. D., Kessler, S. W., Mushinski, J. F. & Potter, M. payment. This article must therefore be hereby marked "advertise- (1980) Abstracts of the 4th International Congress of Immunology, ment" in accordance with 18 U. S. C. §1734 solely to indicate this fact. Paris, 6.5.02. 1138 Downloaded by guest on September 27, 2021 Immunology: Neuberger and Rajewsky Proc. NatL Acad. Sci. USA 78 (1981) 1139

23 10 a~~~~

24-0 X LtA2

C.) .~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.....

.1o3~~ ~ ~ ~ ~ ~ ~ ~ a

Relativefluorescence~~~~S,chne o

FIG. 1. Fluorescence profiles of B1-8.64.1 (curve 1), B1-8.81 (curve 2), and 267.7 (curve 3) cells stained with fluorescent goat anti-/. antiserum (a), 11-6.3.1 anti-8" antiserum (b),H6/31 anti-c antiserum (c), and Ls136 anti-A antiserum (d). The fluorescence profiles of B1-8.81 cells stained with fluorescent 11-6.3.1 anti-8b' or Ls136 anti-A in the presence of B1-8.IgM protein (3 mg/mI) are also shown (curve 4).

K chains (C. Muller, personal communication). been enriched six times with anti-Al revealed no cells that had Although both B1-8 and B1-8.64 cells secrete IgM, neither switched to y3, yl, y2a, y2b, or a expression. Among 106 cells line possesses surface IgM as detected by FACS analysis ofcells screened, we also found no revertants of B1-8.61 that stained stained with goat anti-s antisera. That this is not a general defect for cytoplasmic p chains. in the synthesis or membrane incorporation ofsurface Ig is dem- Surface Ig ofBl-8.6i1. FACS analysis demonstrated that B1- onstrated by the fact that B1-8 cells do stain for surface IgGi. In 8.61 stained brightly with fluorescent anti-Al (Ls136) and anti- the hope of isolating B1-8.64 variants containing surface IgM, 6b (H6/31 and 11-6.3.1) antisera but not with goat anti-p. anti- we used the FACS to enrich for B1-8.64 variant cells staining serum (Fig. 1). Furthermore, whereas the binding ofanti-Al to with a fluorescein-conjugated monoclonal anti-Al antiserum the surface Ig could be competitively inhibited by soluble Bl- (Ls136). After four successive sortings, each time taking the 8.IgM protein, the binding ofH6/31 anti-6 could not. To deter- brightest 1% ofthe population, analysis in the fluorescence mi- mine the Mr ofthe surface 6 chains, Ig was purified on anti-6 or croscope revealed the presence of about 1 in 100 cells which anti-Al immunosorbents or on NIP-Acap-Sepharose from cells could be stained for A on the surface but not A; the rest of the cells remained surface-Ig negative. Analysis with fluorescein- conjugated monoclonal, antisera directed against the b- 1 2 3 4 5 6 chain (H6/31 or 11-6.3.1 anti-a) suggested that these variants -92 had switched from IgM to IgD expression. The cells were there- fore further sorted twice by using either theLsl36 anti-Al or the _68 FIG. 2. Analysis of radioiodin- H6/31 anti-81 antiserum. Both procedures yielded populations _1Il ated and reduced Ig samples in a in which more than 50% ofthe cells stained for surface 6 and in 10%NaDodSO4/polyacrylamide gel. Y Lanes: 1 and 5, secreted Ig of Bl- which a similar proportion of the cells typed in the cytoplasm 8.8; 2, spleen cell surface Ig puri- as &68+. The population from the second anti-6 sorting was -43 fied on 11-6.3.1 anti-OS sorbent; 3, cloned and, of39 clones, 35 secreted 6 but not p. and 4 secreted B1-8.81 surface Ig purified on NIP- p. but not 6 as judged by radioimmunoassay. One of the -S' Acap-Sepharose; 4, spleen cell sur- clones (B1-8.81) was chosen forfurther study. face Ig purified on H6/31 anti-dS Because linefor variants- Al sorbent. Positions of Mr markers enriching the'BI-8.64 expressing and light chains (L) are indicated. chains on the surface resulted in the isolation ofa line that had ab - L [Phosphorylase a (Mr 92,000), bo- switched from p. to 6 expression, it is interesting to note that cy- vine serum (Mr 68,000), toplasmic staining of =105 cells from the population that had and (Mr 43,000).] Downloaded by guest on September 27, 2021 1140 Immunology: Neuberger and. Rajewsky Proc. Natl. Acad. Sci. USA 78 (1981)

that had been surface radioiodinated. In all cases, NaDodSO4/ Table 1. Comparison ofB1-8.IgM and B1-8.IgD polyacrylamide gel electrophoresis revealed a band comigrat- Property B1-8.IgM Bl-8.IgD ing with the A1 light chain of B1-8.IgM and a diffuse band of Affinity for NP 2 x 10-6 1.6 x 10-6 the same mobility as the chains of Ig purified on immuno- sorbents from surface radioiodinated cells of hapten, M* spleen C57BL/6 Affinity for NIP 1 X 10-7 1.2 x 10-7 mice (Fig. 2). The surface 8 chain has an apparent M, of about hapten, M* 64,000. Structure (secreted (2,2A2)5 82A2 (major species) Secreted Ig of B1-8.61. B1-8.81 cells were propagated in Ig) mice as ascites and the Bl-8.IgD was purified on NIP-Acap- Heavy chain Mr 75,000 (apparent) 61,000 (apparent; Sepharose with a yield of approximately 0.6 mg/ml of acites major species); fluid. Serological binding inhibition assays (Fig. 3) demon- 44,000 strated that the Bl-8. IgD carried 8 but not ,u determinants. (unglycosylated) Comparison of the B1-8.IgM and B1-8.IgD Variable (V) Re- Presence ofJ + gions. The B1-8.IgM and B1-8.IgD antibodies had similar affin- proteint ities for the hapten NP as well as for the crossreactive hapten Binding to protein NIP (Table 1). It had been shown (11) that, on binding of NIP- At Acap to several monoclonal Al-bearing anti-NP antibodies, the Complement + NIP-Acap absorbance maximum shifts from 430 nm to 480 nm. fixation§ This absorbance shift, a characteristic ofthe binding site, is also * Affinities given are the concentrations of NP-Acap or NIP-Acap re- observed on NIP-Acap binding to B1-8. IgD. quired to inhibit the binding of 10-s M N[215I]P-Acap to the antibod- The V regions ofB1-8. IgM and B1-8.IgD were compared ser- ies by 50%. The values for B1-8.IgM are from ref. 11. ologically by using both a polyspecific rabbit antiserum directed t J protein was identified after electrophoresis ofalkylated Ig samples against the B1-8. IgM V region (anti-idiotypic serum) and several (1 mg) in alkaline gels (12); IgAs from myelomas MOPC 460 and TEPC 15 (which contain J protein) and IgGs from J 606 and HOPC 1 monoclonal anti-idiotopic antibodies. Bl-8.IgD could inhibit (which do not) were included for comparison. more than 98% of the binding of radioiodinated Bl-8.IgM to t Binding to protein A at pH 7.2 in the presence of 1% bovine serum al- the rabbit anti-idiotypic serum and, in fact, competed slightly bumin was tested by using 125I-labeled antibodies and protein A as better than did unlabeled B1-8.IgM itself (Fig. 3d). well as 125I-labeled protein A and antibodies bound to Costar plates. Similar inhibition curves were obtained with the monoclonal Binding was determined after washing with phosphate-buffered sa- anti-idiotopic antibodies (I50 values given in Table 2). line at pH 7.2. § Complement fixation was tested by antibody-mediated hemolysis of Molecular Weight and Glycosylation. After electrophoresis NP-coupled sheep erythrocytes with guinea pig complement (13). An- in a 10% polyacrylamide gel containing NaDodSO4, reduced tibodies were tested at concentrations up to 32 ,ug/ml (at least 10 Bl-8. IgD gave a light chain comigrating with the A1 light chain times the hemagglutination titer). The 50% hemolysis values were: of Bl-8.IgM and a diffuse heavy chain band that could be re- Bl-8.IgM, 0.02 ,ug/ml; S43 [IgG3, Al anti-NP antibody (11)], 6 ug/ solved into a major component (more than 80%) with an appar- ml; B1-8.IgD, no hemolysis at 32 pg/ml. ent Mr of61,000 and a minor component ofapparent Mr 57,000 (Fig. 4a). Both bands were present in samples purified from as- cites.fluid and culture supernatants whether or not a.cocktail of 100~~~~~~~~~* protease inhibitors was included in all, buffers; furthermore, polyacrylamide gel electrophoresis ofunreduced samples from several Bl-8.IgD purifications did not reveal the presence of Fab or F(ab')2 fragments. 50 Treatment of B.1-8.81 cells with tunicamycin, an inhibitor of glycosylation (14), decreased the apparent Mrs ofthe secreted 8 chains from 61,000 and 57,000 to below 47,000 with a limit prod- uct at 44,000; the electrophoretic mobility of the light chain, a b I which was not glycosylated, was unaffected. The cytoplasmic 6 , 100 1 chains, which have lower apparent Mrs than secreted chains (57,000 and 56,000), also were decreased to a limit product OfMr

Table 2. Serological comparison ofB1-8.IgM and B1-8.IgD a~~~~~~~~~~~50 Antiserum I50 ratio* Rabbit anti-Bl-8.IgM serum 0.3 Monoclonal group anti-B1-8.IgM idiotope antibodiest 0.5-0.9 Monoclonal group 2 anti-Bl-8.IgM idiotope 0 0.1 1 10 100 0 0.1 1 10 100 antibodiest 0.4-0.8 Inhibitor, ,ug/ml Monoclonal Ls136 anti-Al antibody 0.4 FIG. 3. Serological analysis of B1-8.IgM and B1-8.IgD in binding * I0 values are the amounts (jg) antibodies required to inhibit 50% of inhibition assays. 100% relative binding was binding without inhibi- the binding ofradioiodinated B1-8.IgM tothe appropriate antiserum. tor. Unlabeled B1-8.IgM (-) and Bl-8.IgD (o) were used to inhibit the This ratio is Iso ofBI-8.IgD to I50 ofB1-8.IgM. binding of radiolabeled B1-8.IgM to goat anti-, (a), radiolabeled Bi- t Binding to B1-8.IgM of group 1 anti-idiotopic antibodies (Ac17, Ac22, 8.IgD to H6/31 and-db (b), radiolabeled B1-8.IgD to 11-6.3.1 anti-86 (c), Ac146, and As79) is inhibited by free hapten whereas the binding of and radiolabeled B1-8.IgM to rabbit anti-B1-8.IgM idiotype antiserum group 2 antibodies (Ac38, Ac106, and A6-24) is not (ref. 8; unpub- (d). lished data). Downloaded by guest on September 27, 2021 Immunology: Neuberger and Rajewsky Proc. Natd Acad. Sci. USA 78 (1981) 1141 ducing only a reduction ofabout 1000 in their apparent Mr, the cytoplasmic 8 chains were decreased from apparent Mr values of 57,000 and 56,000 to values of 47,000 and 45,000 (Fig. 5). The differential sensitivity of the cytoplasmic and secreted 8 chains to endo-/3-N-acetylglucosaminidase H is analogous to observa- tions made previously on mouse u chains (15). a Unreduced BL-8.IgD contains a major component of appar- ent Mr 135,000 in NaDodSO4polyacrylamide gel electropho- resis (Fig. 4b) and about 140,000 by gel filtration (not shown). We interpret this as indicating the predominance of the 82A2 a1IgM IgG 92 A 68 form with the discrepancy between the reduced and unreduced Mr values being explained by the anomalous behavior ofglyco- protein in NaDodSOjpolyacrylamide gel electrophoresis (16). Electrophoretic analysis of unreduced B1-8.IgD (Fig. 4b) also demonstrated the presence of a minor species (3.5% of the protein) with an apparent Mr of 85,000; this could well be due b to the 8A form of the protein. General Properties. BL-8.IgM, but not B1-8.IgD, contained 0.2 0.4 0.6 J protein (Table 1). Another change resulting from the switch of 0.8 constant regions was that BL-8.IgM was potent at fixing guinea RF pig complement but BL-8.IgD was ineffective. B1-8.IgM and B1-8.IgD were similar in that neither bound to protein A of FIG. 4. Densitometry traces after Coomassie blue-staining of re- duced Bl-8.IgD electrophoresed in a 9% NaDodSO4/polyacrylamide Staphylococcus aureus at pH 7.2. gel (a) and unreduced B1-8.IgD electrophoresed in a 7% NaDodSO4/ Although no significant degradation of B1-8.IgD occurred polyacrylamide gel (b). Mobility (RF) is relative to the bromphenol blue during purification, on incubation of B1-8.IgM and B1-8.IgD dye front. The positions of unreduced IgM (Bl-8.IgM) and IgG1 (from with trypsin (trypsin/Ig ratio, 1:100) all 8 chains were cleaved X63) are as are the of markers (x indicated, positions Mr 10-3) (cf. within 1 1/2 hr whereas intact A chains were still apparent after Fig. 2). 12 hr. Samples ofradioiodinated B1-8.IgM and B1-8.IgD were incubated at 37°C in serum from C57BL/6 mice and in both 44,000 when synthesized in the presence of tunicamycin (Fig. cases the amount of radioactivity that could bind to NIP-Acap- 5). Sepharose after 50 hr ofincubation was 65-75% ofthe zero time To confirm that the Mr 44,000 chain band represents the value; NaDodSOJpolyacrylamide gel electrophoresis of the unglycosylated, but otherwise intact, chain, we treated bio- samples after the incubation demonstrated that no detectable synthetically labeled B1-8.IgD with endo-f3-N-acetylglucosa- (<5%) degradation to Fab or F(ab')2 had occurred in either Ig minidase H, an enzyme that degrades "core" sugar moieties that sample. have a high mannose content and are linked by an N-glycosidic bond to asparagine residues ofpolypeptide chains. Whereas the enzyme had little effect on the normal secreted chains, pro- DISCUSSION The Ig secreted by the B1-8. 81 line has 8 allotypic determinants 1 2 3 4 5 6 7 recognized by two monoclonal anti-8 antibodies which are known to be specific for sites located on opposite sides of the hinge region (17), and the protein does not react with polyspe- cific antisera directed against the ,u, 'y, or a mouse heavy chain classes. The B1-8. 81 surface heavy chain have the same electro- phoretic mobility as surface 8. Thus, the constant 68 region of the 8 chain secreted by B1-8. 81 appears to be repre- - sentative ofnormal mouse 8. Our estimates for the apparent Mrs of the B1-8.81 surface and secreted 8 chains (major species at 64,000 and 61,000, respectively) are slightly lower than those .10Is0 obtained by other groups for the apparent Mrs ofspleen cell sur- 43 face 8 and serum 8 [major species ofboth forms between 65,000 and 70,000 (18-21)]. However, because we find that the surface 8 chains of spleen cells and of B1-8.81 comigrate in our poly- acrylamide gel system, this discrepancydoes not indicate unusu- al features ofthe 8chains from B1-8.81 but is likely to reflect the _ a 0 L fact that, owing to the heavily glycosylated nature ofthis heavy _00 chain class, Mr values deduced from electrophoretic mobilities in NaDodSOJpolyacrylamide gels are probably inaccurate and FIG. 5. Analysis of biosynthetically labeled and reduced B1-8.61 the values determined will depend critically on the polyacryla- samples in a 10% NaDodSO/polyacrylamide gel. Lanies: 1, secreted Ig mide gel electrophoresis procedure used (16, 22). treated with endo-(3-N-acetylglucosaminidase H; 2 and 7, untreated se- B1-8.IgD is heavily glycosylated as indicated by the fact that creted Ig; 3, cytoplasmic Ig of cells labeled in the presence of tunica- mycin; 4, cytoplasmic Ig that had been treated with endo-3-N-acetyl- the unglycosylated 8 chain has an Mr of44,000, as compared to glucosaminidase H; 5, cytoplasmic Ig of untreated cells; 6, Ig secreted the apparent Mr value of61,000 for the major species of glyco- by tunicamycin-treated cells. Positions of Mr markers (x 10-3) and sylated 8 chain. Human 8 chains have also been shown to be light chains are indicated. heavily glycosylated (21). Downloaded by guest on September 27, 2021 1142 Immunology: Neuberger and Rajewsky Proc. Nad Acad. Sci. USA 78 (1981)

Several models ofIgD function are based on the susceptibil- 1. Rowe, D. S. & Fahey, J. L. (1965)J. Exp. Med. 121, 171-184. ity of this immunoglobulin to proteolytic fragmentation (21). 2. Bazin, H., Beckers, A., Urbain-Vansanten, C., Pauwels, R., However, we find that, although BL-8.IgD is much more sus- Bruyns, C., Tilkin, A. F., Platteau, B. & Urbain, J. (1978)J. Im- munol 121, 2077-2082. ceptible to tryptic fragmentation than is B1-8.IgM, both the 3. Abney, E. R., Hunter, I. R. & Parkhouse, R. M. E. (1976) Nature IgM and IgD are relatively stable in normal mouse serum. We (London) 259, 404-406. think that the low level ofIgD in serum may be explained by the 4. Oi, V. T., Jones, P. P., Goding, J. W., Herzenberg, L. A. & Her- small number ofIgD-secreting plasma cells (18). zenberg, L. A. (1978) Curr. Top. Microbiol Immunol. 81, 115-129. The class switch observed in B1-8.64 cells has a number of 5. Radbruch, A., Liesegang, B. & Rajewsky, K. (1978) in Immuno- features in common with the spontaneous class switches previ- fluorescence and Related Staining Techniques, eds. Knapp, W., ously observed in MPC 11 and MOPC 21 myelomacells (23, 24). Holubar, K. & Wick, G. (Elsevier/North-Holland, Amsterdam), First, the pp. 123-136. frequency of the pu-to-6 switch must be of the same 6. Pearson, T., Galfre, G., Ziegler, A. & Milstein, C. (1977) Eur. J. order of magnitude as the "forward" switches observed in the Immunol. 7, 684-690. myeloma lines ( 10-6 per cell per generation) because we did 7. March, S. C., Parikh, I. & Cuatrecasas, P. (1974) AnaL Biochem. not detect any S+ variants in 10 B1-8 cells screened and used 3 60, 149-152. X 107 cells in the initial FACS enrichment. Second, the present 8. Reth, M., Imanishi-Kari, T. & Rajewsky, K. (1979) Eur. J. Immu- study, as well as the previous ones, suggests that spontaneous nol. 9, 1004-1013. 9. Haustein, D. (1975) J. Immunol Methods 7, 25-38. class switches occur stepwise and in the order in which the 10. Neuberger, M. S., Patterson, R. A. & Hartley, B. S. (1979) heavy chain constant region genes are arranged on the chromo- Biochem. J. 183, 31-42. some (ref. 25; K. Moore and T. Hunkapillar, cited in ref. 26). 11. Reth, M., Hammerling, G. J. & Rajewsky, K. (1978) Eur. J. Im- Third, Bl-8.IgD appears to carry a V region identical or very munot 8, 393-400. similar to that of BL-8.IgM as indicated by hapten-binding and 12. Mosmann, T. R., Gavel, Y., Williamson, A. R. & Baumal, R. (1978) idiotypic analysis; at the same time it has lost all heavy chain Eur.J. Immunol 8, 94-101. constant region 13. Ey, P. L., Prowse, S. J. & Jenkin, C. R. (1979) Nature (London) determinants of Bl-8.IgM and possesses a 8 281,492-493. chain indistinguishable from that of normal in 14. Tkacz, J. S. & Lampen, J. 0. (1975) Biochem. Biophys. Res. Com- terms ofsize and serological properties. However, in contrast to mun. 65, 248-257. the situation encountered with the MOPC 21-derived switch 15. Tartakoff, A. & Vassali, P. (1979)J. Cell Biol 83, 284-299. variants (24), we have not observed any phenotypic revertants 16. Segrest, J. P., Jackson, R. L., Andrews, E. P. & Marchesi, V. T. in 106 B1-8.81 cells. Ifforward switches were due to deletion of (1971) Biochem. Biophys. Res. Commun. 44, 390-395. the previously expressed CH gene (25) 17. Kessler, S. W., Woods, V. L., Finkelman, F. D. & Scher, I. (1979) and the reversion as ob- J. Immunol 123, 2772-2778. served in MOPC 21 switch variants were a result of homologous 18. Bargellesi, A., Corte, G., Cosulich, E. & Ferrarini, M. (1979) Eur. mitotic recombination, then the absence of revertants in Bi- J. Immunol 9,490-492. 8.81 cells might simply indicate that these cells do not possess a 19. Mescher, M. F. & Pollock, R. R. (1979) J. Immunol 123, CIA-bearing chromosome. Indeed, a CAu gene could not be de- 1155-1161. tected in the DNA of BL-8.81 cells by the Southern blotting 20. Eidels, L. (1979)J. Immunol 123, 896-902. technique (F. Sablitzky, personal communication). 21. Moller, G., ed. (1977) Immunol Rev. 37. 22. Melcher, U. & Uhr, J. W. (1976)J. Immunol. 116,409-415. 23. Liesegang, B., Radbruch, A. & Rajewsky, K. (1978) Proc. Nati We thank C. Muller for the B1-8.64 line, W. Santfins for helpwith the Acad. Sci. USA 75, 3901-3905. FACS, T. Imanishi-Kari, B. Liesegang, H. Dorff, and M. Reth for 24. Radbruch, A., Liesegang, B. & Rajewsky, K. (1980) Proc. Natl gifts ofreagents, and M. Cramer and A. Radbruch for many helpful dis- Acad. Sci. USA 77, 2909-2913. cussions. This work was supported by the Deutsche Forschungsge- 25. Honjo, T. & Kataoka, T. (1978) Proc. Natl Acad. Sci. USA 75, meinschaft through Sonderforschungsbereich 74. M.S.N. is a recip- 2140-2144. ient of a long-term fellowship from the European Molecular Biology 26. Early, P., Rogers, J., Davis, M., Calame, K., Bond, M., Wall, R. Organization. & Hood, L. (1980) Cell 20, 313-319. Downloaded by guest on September 27, 2021