Programmed DNA Rearrangements During Differentiation: Immunoglobulin Class Switching Citation Davis, Mark Morris (1981) Programmed DNA Rearrangements During Differentiation: Immunoglobulin Class Switching. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/KFJS-G857. https://resolver.caltech.edu/CaltechETD:etd-04292005-082735 Abstract The events of B-lymphocyte differentiation can be reconstructed in part through an analysis of the organization of heavy-chain genes isolated from B-cell tumors (myelomas). A mouse immunoglobulin alpha heavy-chain gene is shown to be composed of at least three non-contiguous segments of germline DNA -- a V H gene segment, a J H gene segment adjacent to the C μ coding region, and a C α gene segment. These gene segments are joined together by two distinct types of DNA rearrangements: variable region formation and immunoglobulin class switching. Three examples of IgM → IgA elass switching were examined and in each case a different site adjacent to C μ and a different site adjacent to C α were joined together in the process of switching. Two of the three C μ sites shared significant homology to each other (15/25 nucleotides) and all three of C α sites were highly homologous (22/30 nucleotides). We believe these sequences serve as recognition sites for class switching. Furthermore, the lack of homology between the C α consensus sequence and sequences reported for C γ1 and c γ2b recombination sites suggests that this process is mediated by class-specific recognition sequences and, presumably, class-specific regulatory mechanisms. A number of predictions and possible explanations of immune phenomena result from this observation. Apparently nonproductive DNA rearrangements, occurring in the same tumor lines, seem also to utilize some of the _same regulatory apparati. In addition, it appears that in one example, MClOl, class switching has progressed from C μ → C α → C γ1 . This switching pathway presents difficulties for the simple deletional model of C H switching. Item Type: Thesis (Dissertation (Ph.D.)) Subject Keywords: (Molecular Biology) ; antibody gene rearrangement, heavy chain switching ; immunoglobin ; DNA rearrangements Degree Grantor: California Institute of Technology Division: Biology Major Option: Molecular Biology Awards: Caltech Distinguished Alumni Award, 2005. Milton and Francis Clauser Doctoral Prize, 1981. Thesis Availability: Public (worldwide access) Research Advisor(s): Hood, Leroy E. (advisor) Lewis, Edward B. (advisor) Group: Caltech Distinguished Alumni Award Thesis Committee: Lewis, Edward B. Davidson, Eric H. Delbruck, Max Davidson, Norman R. Maniatis, Thomas P. Defense Date: 28 August 1980 Record Number: CaltechETD:etd-04292005-082735 Persistent URL: https://resolver.caltech.edu/CaltechETD:etd-04292005-082735 DOI: 10.7907/KFJS-G857 ORCID: Author ORCID Davis, Mark Morris 0000-0001-6868-657X Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 1544 Collection: CaltechTHESIS Deposited By: Imported from ETD-db Deposited On: 29 Apr 2005 Last Modified: 19 Aug 2025 18:25 Thesis Files Preview PDF - Final Version See Usage Policy. 4MB Repository Staff Only: item control page

PROGRAMMED DNA REARRANGEMENTS DURING DIFFERENTIATION: IMMUNOGLOBULIN CLASS SWITCHING Thesis by Mark Morris Davis In Partial Fulfillment of the Requirements For the Degree of Doctor of Philosophy California Institute of Technology Pasadena, California 1981 (Submitted August 28, 1980) ii This thesis is dedicated to my family and friends, especially Chien, without whose love nothing would be possible. iii "Glad to be so young, Talkin' with my tongue. Glad to be so careless in my way. Glad to take a chance, and play against the odds. Glad to be so crazy in my day." StevE: Forbert 11 ••• But more than that, science is a quest for poetic, romantic, almost unattainable goals like growing a beard or having a car that runs." Mitch Kronenberg iv ACKNOWLEDGEMENTS I'd like to thank my professors, who, both by formal instruction and by their . example have taught me much of what I know about Science: Michael Tomko, P. Y. Johnson, Michael Beer, Eric Davidson, Tom Maniatis, Ed Lewis, Max DelbrUck, Ron Konopka, Norman Davidson, and especially Lee Hood, who knows he's important. I'd also like to thank those on the foothills (in some cases, the gullies) of Olympus who . taught me all the rest: Bill Klein, Glenn Galau, Barbara Hough-Evans, Amy Lee, Tom Sargent, Maggie Chamberlin, Jimbo Mullins, Marty and David Goldberg, Brian Seed, Randy Smith, Max (again) and Manny DelbrUck for their great camping trips and terrific family, all the Ingersolls, the Pechpeople, Stuart "Iron Butt" Kim, Mitchell Kronenberg, "Full-length Phil" Early (a general thorn-in-the-side but a first-rate collaborator), Steve and Aggie Crews, John Burdakin, Mark Warren, the honorable and highly esteemed C. Lai, Jerry Garcia, Jackson Browne, Richard Brown, "Sunny Phil" Patten, fencing buddies like "Fast Eddy" Rhodes, Jeff Hicks and Stuart Goodnick ("the euthanasia poster boy"), and others who have touched my life. Further thanks to Jane Chacon (lifeguard of the typing pool) and to Pat Lee and her gang at Graphic Arts, both of whom never tired of hearing me cry "Wolf!". Thanks also to the Jean Weigle Memorial Fund for support in the preparation of this thesis and to NIH for pocket money. Thanks especially to those Samurai scientists Susumu Tonegawa, 'Toshi ~akano and Tasuku Honjo for their high standards of competition. to Connie Katz for typing this. I am also grateful v ABSTRACT Th~ events of B-lymphocyte differentiation can be reconstructed in part through an analysis of the organization of heavy-chain genes isolated from B-cell tumors (myelomas). A mouse immunoglobulin alpha heavy-chain gene is shown to be composed of at least three non-contiguous segments of germline DNA--a VH gene segment, a .JH gene segment adjacent to the Cµ coding region, and a Ca gene segment. These gene segments are joined together by two distinct types of DNA rearrangements: variable region formation and immunoglobulin class switching. Three examples of IgM + IgA elass switching were examined and in each case a different site adjacent to Cµ and a different site adjacent to Ca were joined together in the process of switching. Two of the three Cµ sites shared significant homology to each other (15/25 nucleotides) and all three of Ca sites were highly homologous (22/30 nucleotides). We believe these sequences serve as recognition sites for class switching. Furthermore, the lack of homology between the Ca consensus sequence and sequences reported for Cyl and cy 2b recombination sites suggests that this process is mediated by class-specific recognition sequences and, presumably, classspecific regulatory mechanisms. A number of predictions and possible explanations of immune phenomena result from this observation. Apparently nonproductive DNA rearrangements, occurring in the same tumor lines, seem also to utilize some of the _same regulatory apparati. In addition, it appears that in one example, MClOl, class switching has progressed from Cµ+ Ca+ Cyl. This switching pathway presents difficulties for the simple deletional model of CH switching. vi Table of Contents IN'TRODUCTION . . . . . . . . . • . . • . . . . • • . . . . . CHAPTER f Immunoglobulin heavy chain gene organization in mice: Analysis of a myeloma genomic clone containing variable and a constant regions • • • • • • • • • 1 • • • • 11 CHAPTER 2 The organization and rearrangement of heavy chain immunoglobulin genes in mice • • • • • • • • • • • • • • • 16 CHAPTER 3 An immunoglobulin heavy chain gene is formed by at least two recombinational events • • • • • • • • • • . • • 31 CHAPTER4 DNA sequences mediating class switching in • • • • • • • • • 38 alpha immunoglobulin genes • • • • CHAPTER 5 Two types of DNA rearrangements in immunoglobulin genes • • • • • • • • • • • • • • • • • • • 62 1 INTRODUCTION All life processes are abstracted into the genome of each organism. Any fundamental understanding of evolution or development, of what we are and what we can be, has to take into account the information encoded in our DNA and the mechanisms which act upon it. A useful construct for subdividing genomes into manageable segments has been the idea of a "gene", typically a specific heritable trait which can be localized to a discrete region of the genome and associated with a specific protein product. In this light, it is perhaps understandable that antibody genes have been an anathema to many geneticists since they disrupt this concept of a gene and in general blur the distinction between gene and genome. 'Each antibody "gene" is made up of many parts, sets of which are interchangeable, spread over perhaps millions of nucleotides of DNA. But it is through the study of "exceptions" such as antibody genes that we seem to significantly expand our knowledge about the informational capabilities of genomes. In particular, this thesis concerns itself with immunoglobulin class switching, one of the two distinct forms of carefully regulated DNA rearrangement which act to expand the ·information inherent in antibody genes several orders of magnitude over that of "conventional" genes. Immunoglobul.ins and Their Genes . Antibodies are one of the major components of the body's defense against foreign organisms. It serves both to 'recognize' and bind to chemical moieties not native to the organism (antigens) and to trigger a variety of secondary immune mechanisms (effector functions). The antibody molecule itself is typically a dimer composed of two polypeptide chains-heavy (H) chains and light (L) chains. Early protein sequencing studies further subdivided each type of polypeptide into variable (V) and constant (C) regions (1, 2). The variable regions of heavy and light chains fold 2 together to form the antigen-binding site or V region domain. It was this striking distinction between variable and constant regions within the same polypeptide that first led Dreyer and Bennett (3) to propose that DNA rearrangement during B-cell dif(erentiation was responsible for the production of such disparate molecules. Molecular cloning and other forms of genomic analysis have confirmed and extended this original hypothesis and defined two distinct forms of DNA rearrangement which amplify the ·information inherent in the DNA many orders of magnitude more than if these genes were static. These types of DNA rearrangement are termed variable region formation and class switching. Since this thesis deals principally with class switching, variable formation or V(D)J joining will be dealt with only briefly, and for the purposes of comparison. Variable Region Formation In the mouse L chain polypeptides are encoded by two unlinked gene families, K and A, as defined by their constant regions CK and CA, respectively (4). The variable regions are encoded by two complementary gene segments, VL and J 1 (5-7). The JL regions are located adjacent to CK or CA. in undifferentiated DNA and it is the fusion of a particular VL with a particular J 1 which gives rise to a functional light chain gene (6-9). A large intervening sequence 1.5-3.0 kb between VLJ and c 1 1 is excised during RNA processing, resulting in a continuous reading frame progressing - from VLJL to CL (10, 11). In the kappa family a large number of V regions (>200) and at least four functional J regions allow a large number of possible antigen-binding sites to be formed. Similarly, in the third unlinked gene family, the heavy chains (4), it has now been shown that the VH region is composed of three gene segments, VH' D, and J (12, 13), which can also independently assort (14) to produce a large number of 8 different variable regions. Of particular importance to later arguments involving heavy chain class switching, an apparent recognition sequence CAC~GTG(NNNNNNN­ NNNN) .rJ._2GACATAAACC has been identified 3' to every V region (both light and 3 ,, ~ heavy) (8, 9, 12, 13), and 5' to every J region. This sequence of approximately 17 conserved residues apparently reflects the information necessary to generate sequence-specific intrachromosomal rearrangements, and in fact, would occur by chance only once or twice per haploid mammalian genome. The exact site of recombination can vary by at least six nucleotides outside of this recognition sequence (8, 12, 13). This further contributes to the diversity possible in antibody variable genes. Class Switching Antibody molecules are divided into five distinct classes: IgM, IgD, IgG, IgA, and IgE-which are determined by one of eight distinct heavy chain genes [i.e., Cµ, C 0 (Cyl' Cy 2a' Cy 2b' C'Y 3) Ca and CE]. These CH regions are tightly linked to each other (15) and to VH' D, and JH regions on chromosome 12 of mice (16, 17). Each class (and subclass) seems to be associated with unique functions such as complement fixation or the release of histamine from most cells (15). One of the first observations that the same VH region could be expressed with different constant (CH) regions was made by Wang et al. (18) who partially sequenced the N-terminal (V H) regions of human patients with myelomas which synthesize two or more different classes of immunoglobulin. This result was reproduced in a number of other cases as well (19-22). Concurrently, Gearhart et al. (23) showed that a single cultured lymphocyte could give rise to two or more different classes of antibody with serologically identical variable region domains. In a number of studies (24-26), and particularly in the elegant work of Coffman and Cohn (27), it was clear that the first class of antibody expressed was IgM, no matter what the subsequent product. In summary, these experiments indicated that IgM (hence Cµ) was always the first antibody expressed with the newly formed VH region (and light chain) and a second or third CH gene could then be expressed with this same VH region. This phenomenon is known as CH or class switching, and serves to combine a particular 4 antigen-binding capability ·with the different CH fWlctions and tissue localizations (15). All v6•s may be required to 'pass through' an IgM stage because of the crucial role that this molecule has as a cell-surface receptor for antigen, in particular, the clonal expansion of a particular cell line in response to antigen may require C µ specific functions. Recent History The first concrete evidence that class switching might arise through DNA rearrangement was that of Honjo and Kataoka in 1978 (28) who measured the gene number for the four different Cy subclasses in various myeloma tumors. Their findings indicated that specific CH producers tended to have lost copies of other CH genes in a particular order and suggested that this order was VH' Cµ, Cya' Cyl' cy 2b' cy 28, Ca. However, their results were suspect in that they were at the very limits of resolution of hybridization kinetics (1 copy vs. 1/2 copy) and used uncloned probes. Based on these data, they proposed a simple, linear deletional model for CH switching such that an IgM producer (Cµ) would have all of the CH genes and that an IgG3 (y3) producer would delete Cµ , etc. An alternative model utilizing differential RNA processing of a large· nuclear transcript was proposed by Rabbitts (29). With the widespread adoption of recombinant DNA techniques, a number of groups (30-35) repeated Honjo and Katauka's experiments using cloned probes and the more sensitive -techniques of Southern blotting (36). In general, they also showed that chromosomal deletions are associated with class switching and support the gene order mentioned above, although a number of anomalies remain-usually attributed to random myeloma rearrangements. The C'Y part of the gene order has recently been unequivocally confirmed by the linking-up of genomic clones (Honjo, personal commW1ication; Tonegawa, personal communication). The first direct evidence for class switching as the result of DNA rearrangement involved a heavy chain gene isolated by Phil Early and myself from a library (37) of myeloma DNA inserted into 5 the phage Charon 4A (38). Chapters 1 and 2 present the initial characterization of this gene and in the paper presented in Chapter 3, Kathryn Calame and I demonstrated that this gene was the fused product of at least three widely separated segments of DNA, a VH region, a JH-cµ-associated region, and a Ca region. This finding strongly implies that at least two different recombinational events had to have occurred to form this gene, namely, variable region formation [or VH(D)JH joining, shown by Phil Early (12)] and a CH switch (replacement of C µ with Ca ). Subsequently, other groups have found similar tripartite structures (V H-Cµ-CH) for all four CY subclasses (39-41; Honjo, personal communication; Sakano, personal communication). In order to examine the precise mechanisms involved in this process, Stuart Kim and I studied three different examples of lgM-IgA switching at the nucleotide sequence level (Chapter 4). We found that the recombination could involve C µ and Ca flanking sequences directly (simple switching), a discontinuous portion of C µ flanking sequence joined to Ca (complex switching) or that a VH region which has switched to one CH (from Cµ) could then switch to another (successive switching). The instance of successiv~ switching seems to run contrary to the simple deletional model of Honjo and Kataoka (28), however, since the sequence organization is VHeµ -Ca-Cyl" Possible escapes are suggested in Chapter 4 for this paradox but the issue is unresolved. We also found that multiple sites adjacent to both Cµ and Ca coding regions could be used for switching and that two of the three Cµ sites and all three of the C sites were highly homologous among themselves (C -C ) but not to a a a each other (C -C ). Thus, homologous recombination does not appear to be a µ important in CH switching as suggested by a number of groups (39-42). The Cµ consensus sequence is GGTNNTTANNNNNNGGTANNCAAAG and the Ca consensus sequence is PGCTPPGCTGGAATPPGYTGGGNTGPGCTG (where P = Purine, Y = Pyrimidine and N = any nucleotide). Hence there is no apparent similarity with 6 variable regioh formation ,recognition sequences, although the number of conserved nucleotides is similar (15 for Sµ, 22 for Sa, .r17 for V(D)J homology regions). The exact points' of recombination vary up to 16 nucleotides over the Cµ homology region and up to 9 nucleotides in the Ca homology region, roughly comparable to the variability seen in variable region formation (at least 6 nucleotides}. A prokaryotic model might be the restriction endonuclease Mnl I which recognizes a 4-nucleotide sequence, then cuts 5-10 nucleotides away from it (43). Comparison of the C 0 consensus sequence with those around analogous recombination sites for cy 2b and Cyl genes gives very little homology-leading us to propose a model in which each class has its own recognition sequence and hence can be specifically regulated (Chapter 4). Other predictions of this model are that a whole regulatory network might exist to induce switching to each c 8 region, e.g., there might be Ca-specific cell-surface receptors on B cells as well as a specific DNA recognition/splicing enzyme. Thus, the tissue specificity (15), frequency (15) and antigen preference (45) of the immunoglobulin classes and subclasses might be explained by class-specific regulation at the level of DNA rearrangement. Especially interesting in this regard is the finding of Stein et al. (45) that a non-CH linked locus (Sr-1) influences the preponderance of the different CH genes expressed in response to a polysaccharide antigen, thus Sr-1 might be a regulatory gene for CH switching. Other candidates for genes which may influence CH switching are the multitude of different B-cell specific cell-surface markers (seven identified to date) (46). Related Rearrangements As mentioned in Chapter 2, two other forms of the alpha constant region gene are seen in the tumor M603, neither of which seem contiguous to a VM603 , hence they are termed 'abortive' rearrangements. Characterization of one of these 7 abortive rearrangements, a30, by Kathryn Calame and myself, (Chapter 5) indicates that some sequence of unknown origin has been brought 5' to the Ca' precisely within an a homology region associated with c 8 switching in the work mentioned above. This provides a fourth example of rearrangement within an alpha homology region and indicates the possible continued presence of class-specific switching enzymes in myeloma cell lines. In short, immunoglobulin class switching promises to be a fruitful area for studying the precise regulation of DNA rearrangements during differentiation and the informational requirements of DNA sequence recognition in complex genomes. 8 REFERENC~ 1. Porter, M., Dreyer, W. J .. , Kuff, E. L., and Mcintire, K. R. J. Mol. Biol. 8, 814 (1964). 2. Bennett, J. c., Hood, L., Dreyer, W. J., and Potter, M. J. Mol. Biol. 12, 81 (1965). 3. Dreyer, W. J. and Bennett, J.C. Proc. Natl. Acad. Sci. USA 54, 864 (1965). 4. Mage, R., Lieberman, R., Potter, M., and Terry, W. Im The Antigens (Ed. Sela, M.) Academic, New York, (1973) p. 299. 5. Tonegawa, S., Maxam, A. M., Tizard, R., Bernard, 0., and Gilbert, W. Proc. Natl. Acad. Sci. USA '15, 1488 (1978). 6. Brack, C., Hirowa, A., Lenhard-Schueller, R., and Tonegawa, S. Cell 15, 1 (1978). 7. Seidman, J. G., Max, E. E., and Leder, P. NatW'e 280, 370 (1979). 8. Max, E., Seidman, J. G., and Leder, P. Proc. Natl. Acad. Sci. USA 76, 3450 (1979). 9. Satano, H., Huppi, K., Heinrich, G., and Tonegawa, S. Nature 280, 288 (1979). 10. Gilmore-Herbert, M~ and Wall, R. Proc. Natl. Acad. Sci. USA 75, 342 (1978). 11. Schibler, U., Marcu, K. B., and Perry, R. P. Cell 15, 1495 (1978). 12. Early, P., Huang, H., Davis, M., Calame, K., and Hood, L. Cell 19, 881 (1980). - 13. Sakano, H., Maki, R., Furosawa, Y., Roeder, W., and Tonegawa, S. Nature, submitted (1980). 14. Schilling, J., Cleavinger, B., Davie, J., and Hood, L. Nature 283, 35 (1980). 15. Lieberman, R. Springer Seminars in Immunopathology 1, 7 (1978). 16. Meo, T., Johnson, J., Beechey, C. V., Andrews, S. J., Peters, J., and Searle, A. G. Proc. Natl. Acad. Sci. USA '17, 550 (1980). 17. D'Eustachio, P., Prautcheva, D., Marcu, K., and Ruddle, R. J. Exp. Med. 151, 1545 (1980). 9 18~ Wang, A. C., Wilson, S. K., Hopper, J. E., Fudenberg, H. H., and Nisonoff, A. Proc. Natl. Acad. Sci. USA 66, 337 (1970). 19. Sledge, C., Fain, D. S., Black, B., Krieger, R. G., and Hood, L. Proc. Natl. Acad. Sci. USA 73, 923 (1976). 20. Fudenberg, H. H., Wang, A. C., Pink, J. R. L., and Levin, A. S. Ann. N.Y. Acad. Sci. 190, 501 (1971). 21. Wang, A. C., Geisely, J., and Fudenberg, H. H. Biochemistry 12, 528 (1973). 22. Levin, A. S., Fudenberg, H. H., Hopper, -J. E., Wilson, S., and Nisonoff, A. Proc. Natl. Acad. Sci. USA 68, 169 (1971). 23. Gearhart, P. J., Sigal, N. H., and Klinman, N. R. Proc. Natl. Acad. Sci. USA 72, 1707 (1975). 24. Raff, M. C. Cold Spring Harbor Symp. Quant. Biol. 41, 159 (1976). 25. Anderson, J., Coutinho, A., and Melchers, F. J. Exp. Med. 146, 1744 (1978). 26. Wabl, M. R., Forni, L., and Loor, F. Science 199, 1078 (1978). 27. Coffman, R. L. and Cohn, M. J. Immunol. 118, 1806 (1977). 28. Honjo, T. and Kataoka, T. Proc. Natl. Acad. Sci. USA 75, 2140 (1978). 29. Rabbitts, T. H. Nature 275, 291 (1978). 30. Cory, S., Jackson, J., and Adams, J. M. Nature 285, 450 (1980). 31. Cory, S. and Adams, J.M. Cell 19, 37 (1980). 32. Coleclough, C., Cooper, C., and Perry, R. P. Proc. Natl. Acad. Sci. USA '17, 1422 (1980). 33. Rabbitts, T. H., Forster, A., Dunnick, W., and Bentley, D. L. Nature 283, 351 (1980). 34. Yaoita, Y. and Honjo, T. Biomed. Res., in press (1980). 35. Yaoita, Y. and Honjo, T. Nature, in press (1980). 36. Southern, E. M. J. Mol. Biol. 98, 503 (1977). 37. Maniatis, T., Hardison, R. C., Lacy, E., Lauer, J., O'Connell, C., Quon, D., Sim, G. E., and Efstratiadis, A. Cell 15, 687 (1978). 10 38. Blattner, F. R., Williams, B. G., Blechl, A. E., Denniston-Thompson, K., Faber, H. E., Furlong, L.-A., Grunwald, D.J., Kiefer, D. O., Moore, D. D., Schumm, J. W., Sheldon, E. L., and Smithies, O. Science 196, 161 (1977). 39. Kataoka, T., Kawakami, T., Takahashi, N., and Honjo, T. Proc. Natl. Acad. Sci. USA 77, 919 (1980) 40. Takahashi, N., Kataoka, T., and Honjo, T. Gene, in press (1980). 41. Sakan<1, H., Maki, R., Kurosawa, Y., Roeder, W., and Tonegawa, S. Nature, submitted (1980). 42. Dunnick, W., Rabbitts, T. H., and Milstein, C. Nature, in press (1980). 43. Zabeau, M., Greene, R., Myers, P. A., and Roberts, R. J. Unpublished obser- vations cited by New England Biolabs. 44. Perlmutter, R. M., Hansburg, D., Briles, R. A., Vicolotti, R. A., and Davie, J. M. J. Immunol. 121, 566 (1978). 45. Stein, K. E., Bona, C., Lieberman, R., Chien, C., and Paul, W. E. J. Exp. Med. 151, 1088 (1980). 46. B Lymphocytes in the Immune Response. (M. Cooper, D. E. Mosier, I. Scher, and E. S. Vitetta, e~.) Elsevier North-Holland, Inc., New York (1979). " p oc Natl. Acad. Set. USA ~I. 76, No. 2, pp. 857-861, February 1979 GenE!tiCS ' 11 Jmmunoglobulin heavy chain gene organization in mice: Analysis of a myeloma genomic clone containing variable and a constant regions (heavy chain genomic clone/intervening DNA sequences/reslriclion maps/R loop mapping/V and C gene segments) PHILIP w. EARLY*, MARK M. DAVIS*, DAVID B. KABACKt, NORMAN DAVIDSoNt, AND LEROY Hoon** •Division of Biology and fDepartment of Chemistry, California Institute of Technology, Pasadena, California 91125 Contributed by Norman Davtdson, November 7, 1978 MATERIALS AND M~ODS ABSTRACT We have isolated a myeloma genomic DNA clone containing the variable and con~tant regions of a mouse a chain. Restriction enzyme analyses and electron microscopic R loop mapping have demonstrated that the variable region is separated from the constant region by 6.8 kilobases of inter· vening DNA. In addition, two intervening DNA sequences of 100-200 bases separate the constant region into three approxi· mately equal units. These intervening sequences may separate each of the segments coding for the three constant region domains of the a heavy chain. Southern blot analysis of embryo and myeloma DNA suggests that DNA rearrangement of heavy chain variable and constant regions occurs during the differ· entiation of antibody-producing cells. Biological and Physical Containment. Work described in this report was conducted in a P3 physical containment facility using EK2 host-vector systems, in compliance with National Institutes of Health guidelines for recombinant DNA research as published in the Federal Register ((1976) 4.1, 2790227943]. . Bacterial and Phage Strains. Escherichia coli K-12 strain xl776 (7) was provided by R. Curtiss. E. coli K-12 strain DP50SupF (8) and Charon 4A phage (9) were provided by F. Blattner. E.coli strains NS428 and NS433 (10) used for in vitro packaging were obtained from T. Maniatis. mRNA Preparation. BALB/c mouse myeloma tumors originally obtained from M. Potter or the Salk Institute were propagated subcutaneously. A postnuclear supernatant pre· pared from homogenized tissue was used to prepare poly(A)+ RNA from membrane-bound polysomes (11). Heavy chain mRNA, identified by in vitro translation (12), was isolated by sucrose gradient fractionation. cDNA Synthesis and Cloning. Double-stranded cDNA was synthesized by sequential reactions with avian myeloblastosis virus reverse transcriptase and E. colt polymerase I (13). After exclusion on Sephadex G-100, the major component of this cDNA migrated as a band of 1550 base pairs on a nonde· naturing agarose gel. Double-stranded cDNA was joined to Eco RI-cut pMB9 either by poly(dA), poly(dT) tailing (14), or by ligation to synthetic EcoRl linkers (15). Annealing or ligation mixtures were used directly to transform E.coli x1776 (16). Positive transformants were identified by the Grunstein-Hogness technique (17) using 32P-labeled M603 heavy chain mRNA. Construction of M603 Library. High molecular weight genomic DNA (18) prepared from M603 subcutaneous tumors was partially digested with EcoRI, and fragments in the range of 12 to 20 kilobases (kb) were isolated on a sucrose gradient. Ten micrograms of M603 DNA fragments was ligated to Charon 4A arms and packaged in vitro to obtain a library of 3 X 106 recombinant phage (19). The library was amplified on DP50SupF as a plate lysate prior to screening. Isolation of Clones from M603 Library. The constant region plasmid p603al labeled by nick translation with deoxynucleotide [32P]triphosphates (20) was used to screen 400,000 clones from the M603 library plated on DP50SupF (19, 21). Duplicate nitrocellulose filters from each plate were prehybridized in 1 M NaCl/0.045 M trisodium citrate/0.2% bovine serum albumin/0.2% Ficoll/0.2% polyvinylpyrrolidone /0.1 % sodium The antibody system affords a fascinating model for the study of gene organization and expression in eukaryotes. Antibodies or immunoglobulins are composed of two subunits, light and heavy chains, each of which contains an NHz-terminal variable (V) region and a COOH-terminal constant (C) region (I). The antibody polypeptides are divided into discrete domains or homology units, each encompassing approximately 110 resi· dues. Accordin_gly, the light chain has two domains (V and C) and the a heavy chain has four (V, CHI. Cff2, and Cn3) (2). The variable and constant regions of light chains are encoded by three distinct gene segments, V (approximately residues 1-99 ), J or joining (approximately residues 100-112), and C (approximately residues 113-219) (3, 4). The V and J segments encode the classical V region. Each of these DNA segments is separated by intervening nucleotide sequences in embryo or undifferentiated DNA {4). Studies of myeloma DNA suggest that these gene segments are rearranged during the differentiation of antibody-producing cells. However, in myeloma DNA, intervening sequences still separate the V and C segments (4). These interv~ning sequences must be removed from nuclear RNA transcripts of light chain_genes by RNA splicing (5). Thus, DNA rearrangements and RNA splicing appear to be important events in the differentiation of antibody-producing cells (6). We are interested in analyzing the genes coding for heavy chains in embryo and myeloma DNA to determine whether sequence rearrangements occur during differentiation that are comparable to those seen for light chain gene segments (4). Our initial approach has been to isolate genomic clones from a Ji. brary of recombinant Charon 4A bacteriophage containing long fragments of M603 myeloma DNA. In this paper we report the characterization of one clone containing both Vu and Ca regions. The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "adoertuement" in accordance with 18 U.S. C. §1734 solely to indicate Abbreviations: V, variable; C, constant; NaDodS04, sodium dodecyl sulfate; kb, kilobase(s); IVS, intervening sequence(s). *To whom reprint requests should be addressed. this fact. 857 12 858 Genetics: Early et al. dodecyJ sulfate (NaDodS04 ) at 68°C in a rotary water bath (D. Engel and J. Dodgson, personal communication). Denatured plasmid DNA was added to 10 rlg/mland hybridization was continued for 48 hr. Filters were washed extensively in 0.15 M NaCl/O.OI5 M trisodium citrate/0.1% NaDodS04 /10 mM Na4P20, at 68°C. Duplicated spots of hybridization were identified by autoradiography, and plaques corresponding to these locations on the filters were picked and rescreened at a )ow plating density to obtain pure clones. Electron Microscopic R Loop Mapping. Duplex Ch603a6 DNA was photochemicalJy crosslinked (one crosslink per 4 kb) in the presence of 4,5',8-trimethylpsoralen (trioxsalen). Crosslinking prevents DNA strand separation, thus permitting the R loop hybridization (22) to be carried out at a temperature close to or above the DNA strand separation temperature (unpublished data). Crosslinked DNA (5 µg/ml) was hybridized to mRNA (2 µg/ml) in 70% recrystallized formamide/0.4 M NaCl/Q.l M I,4-piperazinediethanesulfonic acid (Pipes), pH 7.2/10 mM EDTA at 56°C for 24 hr. The R-looped DNA was either spread from 70% onto 15% formamide or treated with 1 M glyoxal at I2°C to stabilize the R loops against branch migration (unpublished observations) and then spread from 50% formamide (23). RESULTS AND DISCUSSION Sequence Homologies of a Heavy Chain mRNAs. We isolated heavy chain mRNA from three mouse myeloma tumors secreting IgA imrnunoglobulins. Both the M603 and Sl07 immunoglobulins bind phosphorylcholine and contain nearly identical V region protein sequences (24). The M315 V region differs from M603 at 60% of its amino acid residues. To assay the extent of nucleotide homology between these mRNA species, 32P-labeled single stranded M603 cDNA was hybridized to each mRNA. The hybrid was digested with nuclease SI (25), and the resulting cDNA cleavage products were fractionated on an alkaline agarose gel (Fig. I). M3I5 mRNA protects an 1100-nucleotide piece of M603 cDNA from SI digestion; Sl07 and M603 mRNA both protect the fu]) length of 1550 nucleotides. We conclude that Sl07 and M603 mRNAs are closely homologous over the entire sequence and that M315 mRNA · shares this homology only for the C region, as expected from the protein sequences. In subsequent experiments with M603-like genomic sequences, we used Sl07 mRNA as a probe for the V and C regions and M315 mRNA as a probe for the C region only. FIG. 1. Hybridimtion of mRNAs to M603 cDNA Five 1W1ograms of first-strand M603 cDNA (specific activity, 106 cpm/µ.g) washybridized to 50 ng of the indicated heavy chain mRNA followed by digestion with nuclease SI {25) and separation on a 2% alkaline agaroae gel (26), a portion of which is shown by autoradiography. The first lane contains cDNA alone, hybridized and digested as above, and the last lane contains undigested input cDNA. The cDNA is somewhat longer than 1550 nucleotides, presumably due to an unfolded "hairpin" structure at the 5' end (27). Proc. Natl. Acad. Sci. USA 76 (1979) Characterization of cDNA Clones.· Double-stranded cDNA prepared from M603 mRNA was Used to construct the cDNA restriction map in Fig. 2a. We used poly(dA), poly(dT) tailing to obtain one recombinant plasmid, p603al, which was shown by electron microscopy to contain an insert of approximately 600 nucleotides. Comparison of the cDNA restriction map with that of the plasmid indicates that p603aI contains part of the Ca sequence (Fig. 2b). Subsequently, synthetic EcoRI linkers were used to "clone 5107 double-stranded cDNA. Regions of the cDNA included in two of these plasmids, p107aR5 and pl07aR6, are indicated in Fig. 2b. In order to verify that the cloned sequences are derived from 5107 heavy chain mRNA, a Hinfl restriction fragment of pl07aR6 was isolated, annealed to 5107 mRNA, and used to prime cDNA synthesis in the dideoxynucleotftle sequencing procedure (29). The partial sequence so determined matched the known protein sequence of the 5107 heavy chain between amino acids 92 and 125 (24) (data not shown). Characterization of Genomic Clones. We screened 400,000 plaques from the M603 library with the C region plasmid p600al, labeled with 32p by nick translation. Twenty-five clones hybridized to the probe. Only three of these clones also hybridized to a nick-translated V region probe (the portion of pl07aR6 shown to the left of the Hha I site in Fig. 2b ). These three clones showed identical Eco RI restriction patterns; one, Ch603a6, was selected for further characterization. Ch603a6 contained 16.4 kb of mouse DNA with two internal EcoRI cleavage sites yielding fragments of 7.2, 4.8, and 4.4 kb, which are ordered in the restriction map shown in Fig. 2c. Filter hybridizations by the Southern blot procedure (30} localized the variable region in Ch603a6 to the 7.2-kb EcoRI fragment (Fig. 3). C region probes (p603a I and the portion of pl 07aR6 to the right of the Hha I site in Fig. 2b) hybridized only to a .2.4-lcb section bounded by Xho I and Sma I sites in the 4.8- and 4.4-kb Ec.oRI fragments. These results, displayed schematically in Fig. 2c, demonstrate that the VH and Ca regions in Ch603a6 are separated by at least 4 kb of intervening DNA. Identification of Genomic DNA Fragments Hybridizing to cDNA Clones. Hybridization of the nearly full-length cDNA probe pI07aR5 to Southern blots of EcoRI-digested chromosomal DNA showed that the 7.2-, 4.8-, and 4.4-kb fragments of Ch603a6 all correspond to bands present in the M603 genome (Fig. 4). In embryonic DNA, the 4.8-kb middle piece of Ch603a6 was absent, although bands at 7.2 and 4.4 kb were present. These observations are consistent with a reduction in the distance between the Ch603a6 V and C regions having occurred in the derivation of the M603 genome from the embryo genome. Genetic evidence suggests that a closely related VH region, 5107, may be several hundred thousand base pairs from the Ca region in embryonic DNA (31, 32; but see ref. 33). The other changes evident from the Southern blots of embryonic and M603 DNAs cannot yet be fully explained. · Electron Microscopy Indicates That There Are Three Intervening Sequences in the Genomic Clone Ch603a6. Sl07 or M315 mRNA was hybridized to trioxsalen-cr~linked Ch603a6 DNA under conditions favoring R-loop formation. At least 80% of the DNA molecules examined in the electron microscope were full length and >90% of these contained R lodps. Typical micrographs are shown and interpreted in Fig. 5. Data from 52 molecules were combined to generate the map shown in Fig. 2d. These studies indicate that Ch603a6 contains three intervening sequences (IVS). IVS I has a length of 6.8 kb. The R loop to the left of IVS 1 in Fig. 2d is due to the VH region because it is seen in hybridizations of Ch603a6 DNA with Sl07 mRNA but not with M315 mRNA. The C region structures discussed below are seen in hybridizations with either SI07 or 13 Genetics: Early et al. Proc. Natl. Acad. Sci. USA 76 (1979) a) 859 ~~ ·~ ~ ii~ 5'.-,----)-++-\-ilt--.lt----+-1-l....+l-------.13' (poly A) I ii i i ii j ,' pl07aR6 r~............. ' 'r--------- "" ...... ....... ' 7 .2kb ' [Jzwzzmm??>mmzmzmwmwmam I lefl arm ii ~ Ch4A ~ "" D C} p603a, . pl07aR5 pi07aR5 b} 1550 \ ·.. '-. -- · 4 Bkb:-..... ~~.... \ \ ·... \ --...fl'· "t6kb \\ .•.=""_I ,,. .,. ,_ 4.4kb- i • .. ·....- ~ «§· •• ••••·• x w. DCJ · 1i riohl arm 8 Ch4A w IA!(IOO 360 140: 1·11 H 1 t I I - b c::I 051?g f!f ~ lO:- ..... IVS2 1210±10) 293±58 ~ 343±85 386±70_ Qm.C)-M)Q d) Dleft 0arm Ch4A 24.62:t.73kb - - s.8o:t.46kb...... _ --:14,39·:f.Sokbo O>----------....... ·- cx:x::> --··· 360±80 V IVS\ C . D rivnt arm Ch4A FIG. 2. Comparison of genomic and cDNA segments. (a) Restriction map of M603 double-stranded cDNA. The presence of a "hairpin" structure (determined by alkaline agarose gels) served as a marker for the 5' end (27). Only the HinfI site identified by mRNA sequencing is shown. The structural domains depicted above the cDNA are based on a partial protein sequence of M315 and M511 heavy chains and analogy with other IgA proteins (28). The protein sequence has been aligned so that the Hha I site sequenced in 8107 mRNA coincides with the Hha I site mapped in M603 cDNA. Thi~ alignment gives good correspondence between the Hpa II and Hin ell sites mapped in M603 cDNA and possible cleavage sites of these enzymes at codons for amino acids 158 and 191. (b) Sequences included in cDNA plasmid clones. The ends of p603a 1 are uncertain, as indicated by dotted lines. The larger of the two Eco RI fragments in p 107aR5 is present in inverted orientation relative to its position in the cDNA. This is presumably due to the independent ligation of these two fragments during the cDNA cloning procedure. The restriction fragments of Ch603a6 within which the various parts of the plasmid clones hybridize are indicated by arrows connecting the plasmid and genomic clones. The dotted line connects the int.ernal Eco RI site of the cDNA plasmids with the corresponding site in Ch603a6. (c) Restriction map and sequence organization of Ch603a6. Restriction fragments to which V and C region probes (see arrows from plasmids) hybridize are shaded in correspondence t.o the protein domains. The enlarged portion of the figure shows one of the short intervening sequences (IVS 2). Compare distances between restriction sites here and those shown in a. The Xho I site does not lie within the coding sequence. (d) R-Loop map of Ch603a6. The right and left arms of Ch4A were taken to be 10.7 and 20.1 kb long, respectively. The enlargement of the C region R loops is drawn with the total length of the RNA·DNA duplex equal to the length of the cDNA from the 3' end to the junction with the V region shown in a. IVS 2 and IVS 3 are both assumed to be the same length. The Eco RI site shown in the enlargement of the Ch603a6 restriction map above is aligned with a point 590 nucleotides (measured on the RNA·DNA duplex regions) from the 3' end of the C region R-loop complex. This corresponds to the distance of the Eco RI site from the 3' end of the cDNA. The wavy line in the R loops represents RNA. The dashed line for IVS 3 indicates that its length has not been accurately measured. Errors given are SD. M315 mRNAs. These results and the measured lengths of the several R loops indicate that IVS 1 occurs approximately at the junction of the VH and Ca regions (Fig. 2d). Two kinds of Rloop structures were seen for IVS 1, depending on whether a single mRNA molecule (Fig. 5 a and b) hybridized to both the C and V regions of the DNA or whether these regions were hybridized to two separate mRNA molecules (Fig. Sc). In addition to the 6.8-kb IVS, 58% of the molecules contained two structures that we interpret as due to two short IVS within the Ca region (IVS 2 and IVS 3 in Fig. 2d). In one type of short IVS structure, the single- and double-stranded arms of a C region R loop are joined at a reproducible point inside the R loop. We interpret this as due to a base-paired IVS (bpIVS in Fig. 5). Alternatively, the two arms of an R loop are not joined, but there ·is a small knob at a reproducible point on the double-stranded arm. We interpret this as an IVS that is not base paired to its complement on the opposite strand (ssIVS in Fig. 5). The positions on the C region R loop when~ bpIVS and ~IVS structures are seen are coincident. Because IVS 2 and IVS 3 are observable as knobs but are too short to be measured by electron micros- copy, we believe their lengths to be in the range of l 00 to 200 nucleotides. The reproducible junction points that we attribute to basepaired IVS could be due to site-specific trioxsalen crosslinking. However, the ssIVS structures cannot be explained this way. Furthermore, these structures occurred with approximately the same frequency in R loops with uncrosslinked Ch603a6 DNA. Both electron microscopy and restriction mapping give the same orientation of the VH and Ca regions relative to the right and left arms of Charon 4A. The position of the 3' poly(A) end of the mRNA also has been independently determined in some R loops by an electron microscopic labeling technique (34) (Fig. Sc). IVS 2 Is 210 Nucleotides in Length by 'Restriction Mapping. By comparing the sizes of restriction fragments produred from DNA of the genomic clone Ch603a6 with those from the cDNA clone p107aR6, we have confirmed the presence and determined the length of one of the short IVS in the Ca region (Fig. 6). Gel electrophoresis showed that the Hincll and Pst I 860 14 Genetics: Early et al. 2 3 Proc. Natl. Acad. Sci. USA 76 (1979) 4 7.2- 4.8-. 4.4- FIG. 3. Localization of the V and C regions in Ch603a6 by hybridization to Southern blots. Restriction enzyme digests of Ch603a6 DNA were electrophoresed on l'lir agarose gels, transferred to nitrocellulose filters (30), and hybridized to V or C region probes prepared from gel-purified fragments of Hha I-digested pl07aR6 (site of cleavage indicated in Fig. 2b). V region probe: ethidium bromide staining (lane 1) and blot (lane 2) of Eco RI-digested Ch603a6 DNA. Only the 7.2-kb fragment displayed strong hybridization to the V region. C region probe: ethidium bromide staining (lane 3) and blot (lane 4) of Sma l/Xho I-digested Ch603a6 DNA. Only the 2.4-kb fragment hybridized to this C region probe. The same result was obtained with probes from other parts of the C region. Unmarked bands seen by ethidium bromide fluorescence contained Ch4A vector DNA. The origins of the gels are not shown. sites of the cDNA plasmid pl07 aR6 are separated by 150 nucleotides, whereas in the genomic DNA of Ch603a6 they are 360 nucleotides apart. This demonstrates the presence of an IVS of 210 nucleotides in the genomic clone which corresponds to the position assigned to IVS 2 by electron microscopy. emb 603 : . . : . .,. =~ 19- ....... FIG. 4. Southern blots of EcoRI-cut genomic DNAs hybridized to pl07aR5. High molecular weight genomic DNA was prepared from 12-13 day BALB/c embryos (emb) or M603 subcutaneous tumors (603) (18). Approximately 15 µg of DNA completely digested with Eco RI. was electrophoresed on each lane of a 0. 7% agarose gel. Nitrocellulose filter replicas of the gels were hybridized to 3 2P-labeled p107 aR5 (specific activity, 108 cpm/µg). Upper portions of the blots are not shown. In the two lanes at the left, different plasmid DNAs were added as internal standards (arrows). This obscures the 7.2- and 5.3-kb bands in the embryo DNA, which are better aeen in the embryo lane at the right. The 19-kb band is not visible in the right embryo Jane, probably because of poor transfer to the filter. Some clones from the M603 library contained a 4.4-kb Eco RI fragment hybridizing to P603al, plus either a 12.5- or an 8.8-kb EcoRI fragment hybridizing to the 51 C., sequences in pl07aR6. Thus, there may be multiple copies of the C" gene segment, which would account for the relatively intense hybridization to the 4.4-kb band. All copies of the C,. gene in M603 DNA appeared to have undergone rearrangement or mutation from the embryo; the significance of this observation for the differentiation of antibody-producing cells is unknown. FIG. 5. Electron micrographs of R-loop structures observed with 8107 mRNA hybridized to crosslinked Ch603a6 DNA. (a) R loop showing the large intervening sequence (IVS 1) dividing the constant (C) and variable (V) regions, and two short base-paired intervening sequences (bplVS) in the constant region. (b) Similar too but the C region contains a single-stranded intervening sequence (ssIVS) and a base-paired intervening sequence. (c) Two mRNA molecules hybridized to one Ch603a6 DNA molecule. The poly(A)+ RNA at the 3' end of the C region R loop is labeled with a poly(BrdU)-tailed circular microcolicin El molecule(Col). Both the V and C region R loops contain unhybridized RNA tails adjacent to IVS 1. The C region contains one bplVS and one sslVS. Broken lines represent RNA. The Two Short C Region IVS May Separate the Three Ca Domains. The short IVS in the C., gene separate the C region into three roughly equal segments. At the protein level the C.,. region is divided into three roughly equal homologous structural domains: CHI, Cu2, and CH3 (28). Accordingly, the IVS in the DNA may separate the individual C,. domains, although this supposition will have to be verified by direct nucleic acid sequence analysis. There are several features of immunoglobulin evolution and structure that might involve IVS separating CH domains. First, all C regions, including C.,, contain homologous protein d<r mains presumably derived from a common ancestral gene (2, 28, 37, 38). Second, C regions do not all contain the same number of domains, indicating that new C regions may arise by the addition or deletion of domains. Third, certain aberrant immunoglobulins (heavy chain disease proteins) often appear to involve deletions with breakpoints occurring between dcr mains (39). Fourth, a variant myeloma cell line produced in culture shows deletion of the Cul domain (40). If IVS between domains facilitate unequal recombination, then the creation of new C regions with additional or deleted domains can be explained. Accordingly, the short IVS observed in the C,. region may permit the immunoglobulin domains to operate as fundamental units of evolution (41). 15 Genetics: Early et al. Proc. Natl. Acad. Sci. USA 76 (1979) ""c p107aR6 ... _,i.ii =.. 3 4 .. ---· Ch603a6 iii Q. 1 1so-• 620· 500· 2 :c" ::cc. iii ::c.E ::cc. • ' 5 6. 7 ,, "' . ' '' '' ' •' 250· 140· _ Q. • -· ' ' •·400 • ··300 FIG. 6. Identification and measurement of IVS 2 by restriction mapping. Ch603a6 DNA digested with EcoRI and Xho I was 32P end-labeled (35), and the 750-bp C region fragment (Fig. 2c) was eluted from a 1% agarose gel (36). An autoradiograph of part of a 5% acrylamide gel is shown comparing restriction digests of this endlabeled genomic DNA fragment and p107aR6end-labeled a&r EcoRI digestion. Lanes: 1, Ch603a6 750-bp Xho 1/EcoRI fragment; 2-4, products of digestion of this fragment with Pst I, Hincll, and Hpa II, respectively (the HincII digest is incomplete); 5--7, Pst I, Hin ell, and Hpa II digests of p107aR6 end-labeled after EcoRI digestion. The EcoRl/Pst I bands of the two DNAs are the same length (140 bp), whereas the EcoRI/HincII and EcoRI/Hpa II bands are both 200-220 bp longer in the genomic DNA than in the cDNA (Fig. 2c ). Comparing these results with the cDNA restriction map (Fig. 2a) shows that Ch603a6 contains a short intervening sequence (210 ± 10 bp} between the HincII and Pst I sites of the cDNA. This region includes the boundary between the CHl and CH2 domains. (Other bands in lane 7 derive from Hpa 11 cleavage of the end-labeled plasmid DNA. The 790-bp fragment produced from Hincll-digested p107aR6 is not shown on this part of the autoradiograph.) We thank H. Manor for electron microscopy of p603al, N. Johnson for providing M603 DNA, and Y.-H. Chien for electron microscopk characterization of mRNAs. E.coli polymerase I, DNA ligase T4, and EcoRI linkers were generous gifts from M. Goldberg, R. Scheller, and K. Itakura, respectively. Most of the in ottro packaging extracts used were the gift of T. Sargent. We thank K. Marcu, 0. Valbuena, and R. Perry for advice on mRNA preparation and M. Wickens for communicating cDNA synthesis procedures prior to publication. We benefited from helpful discussions with T. Maniatis, T. Sargent, D. Goldberg, D. Engel, J. Dodgson, R. Joho, B. Klein, and D. Anderson. The work reported here was supported by National Institutes of Health Grants GM 10991 and GM 20927 and Biomedical Research Grant· RR07003A and National Science Foundation Grant PCM 71-00770. P.W.E. and M.M.D. are supported by National Institutes of Health Training Grant GM 07616; D.B.K. is a National Institutes of Health .Postdoctoral Fellow. l. 2. 3. 4. 5. 6. Curtiss, R., Ill, Pereira, D. A., Hsu, J. C., Hull, S. C., Clarke, J. E., Maturin, L. J .• Sr., Goldschmidt, R., Moody, R., Inoue, M. & Alexander, L. (1977) in Proceedings of the 10th Miles International Symposium, eds. Beers, R. F., Jr. & ~tt. E. G. (Raven, New York), pp. 45-56. 8. Leder, P., Tiemeier, D. & Enquist, L. (1977) Science 196, 175-177. 9. Blattner, F. R., Williams, B. G., Blechl, A. E., DennistonThompson, K., Faber, H. E., Furlong, L.-A., Grunwald, D. J., Kiefer, D. 0., Moore, D. D., Schumm, J. W., Sheldon, E. L. & Smithies, 0. (1977) Science 196, 161-169. 10. Sternberg, N., Tiemeier, D. & Enquist, L. (1977) Gene 1, 2.55-280. 11. Marcu, K. B., Valbuena, 0. & Perry, R. P. (1978) Biochemutry 17, 1723-1733. 12. Pelham, H. R. B. & Jackson, R. J. (1976) Eur.]. Biochem. 67, 7. 247-256. • •140 Gally, J. (1973) in The Antigens, ed. Sela, M. (Academic, New York), Vol. l, pp. 162-299. Edelman, G. M., Cunningham, B. A., Gall, W., Gottlieb, P., Rutishauser, U. & Waxdal, M. (1969) Proc. Natl. Acad. Sci. USA 83, 78-85. Weigert, M., Gatmaitan, L., Loh, E., Schilling, J. & Hood. L. (1978) Nature (London) 276, 785-790. Brack, C., Hirawa, M., Lenhard-Schuller, R. & Tonegawa, S. (1978) Cell 15, 1-14. Gilmore-Hebert, M. & Wall, R. (1978) Proc. Natl. Acad. Sci. USA 75, 342-345. Hood, L., Huang, H. V. & Dreyer, W. J. (1977) ]. Supramol. Struct. 7, 531-559. 861 • . Wickens, M. P., Buell, G. N. &'Schimke, R. T. (1978) ). Biol. Chem. !53, 2483-2495. 14. Wensink, P. C., Finnegan, D. J., Donelson, J.E. & Hogness, D. s. (1974) Cell 3, 315-32.5. 15. Seeburg, P. H., Shine, J.. Martial, J. A., Baxter, J. D. & Goodman, H. M. (1977) Nature (London) 270, 486-494. 16. Villa-Komaroff, L., Efstratiadis, A., Broome, S., Lomedico, P., Tizard, R., Naber, S. P., Chick, W. L. & Gilbert, W. (1978) Proc. Natl. Acad. Sci. USA 75, 3727-3731. 17. Grunstein, M. & H~. D.S. (1975) Proc. Natl. Acad. Sci. USA 72, 3961-3965. 18. Blin, N. & Stafford, D. W. (1976) Nucletc Acids Res. 3, 23032308. 19. Maniatis, T., Hardison, R. C., Lacy, E., Lauer, J.. O'Connell, C., Quon, D., Sim, G. K. & Efstratiadis, A. (1978) Cell 15, 687701. 20. Maniatis, T., Jeffrey, A. & Kleid, D. G. (1975) Proc. Natl. Acad. Sci. USA 72, 1184-1188. 21. Benton, W. D. & Davis, R. W. (1977) Science 196, 180-182. 22. Thomas, M., White, R. C. & Davis, R. W. (1976) Proc. Natl. Acad. Sci. USA 73, 2294-2298. 23. Davis, R., Simon, M. & Davidson, N. (1971) Methmh Enzymol. !lD, 41$-428. 24. Hood, L., Loh, E., Hubert, J., Barstad, P., Eaton, B., Early, P., Fuhrman, J., Johnson, N., Kronenberg, M. & Schilling, J. (1977) Cold Spring Harbor Symp. Quant. Biol. 41, 817-836. 25. Berk, A. J. & Sharp, P.A. {1977} Cell 12, 721-732. 26. McDonell, M. W.,Simon, M. N. & Studier, F. W. (1977)).Mol. Biol. 110, 119-146. 27. Maniatis, T., Sim, G. K., Efstratiadi.s, A. & Kafatos, F. C. (1976) Cell 8, 163-182. 28. Robinson, E. A. & Appella, E. (1977} Proc. Natl. Acad. Sci. USA 74, 2465-2469. 29. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl. Acad. Sci. USA 74, 5463-5467. 30. Southern, E. M. (1975) ]. Mal. Biol. 98, 500-517. 31. Riblet, R. J. (1977) ICN-UCLA Symposia on Molecular and Cellular Biology (Academic, New York), Vol. 6, pp. 83-89. 32. Robinson, E. A. & Appella, E. (1977) Proc. Natl. Acad. Sci. USA 74, 2465-2469. 33. Gearhart, P. J. & Cebra, J. J. (1978) Nature (London) !72, 264-265. 34. Bender, W., Davidson, N., Kindle, K., Taylor, W., Silverman, M. & Firtel, R. (1978) Cel/ 15, 779-788. 3.5. Berkner, K. L. 6 Folk, W.R. (1977}). Btol. Chem. !52,31763184. 36. Maxam, A. M. & Gilbert, W. (1977) Proc. Natl. Acad. Sci. USA 74, 560-564. 37. Beale, D. & Feinstein, A. (1976) Q. Rev. Biophys. 9, 135-180. 38. Davies, D.R., Padlan, E. A. & Segal, D. M. (1975) Annu. Rev. Btochem. 44, 639-667. 39. Frangione, B., Lee, L., Haber, E. & Bloch, K. (1977) Proc. Natl. Acad. Sci. USA 70, 1073-1077. 40. Adetugbo, K., Milstein, C. & Seeber, D. (1977) Nature (London) 165, 299-304. 41. Gilbert, W. (1978) Nature (London) 271, 501. 13. 16 THE ORGANIZATION AND REARRANGEMENT OF HEAVY CHAIN IMMUNOGLOBULIN GENES IN MICE1 M. Davis, P. Early, K. Calame, D. Livant and L. Hood Division of Biology, California Institute of Technology Pasadena, California In: Eukaryotic Gene Regulation, edited by R. Axel, T. Maniatis and c. F. Fox. 1 91125 ICN-UCLA Symposium. Academic Press, in press This work was supported by NSF grant PCM 76-81546. 17 ABSTRACT 'A preliminary analysis of several heavy chain variable (V) and constant region (C) gene segments from sperm (undifferentiated) and myeloma (differentiated) DNA has revealed the following: 1) the v8 and Ca genes are separate in the germ line; 2) the VH and C genes are rearranged during the differentiation or tge antibody-producing cell; 3) multiple rearranged Ca genes are present in the DNA or a single myeloma tumor; 4) small intervening sequences may separate the domains or the and constant region genes; and 5) at least 8-9 germ line VH genes exist for antibodies binding phosphorylcholine. INTRODUCTION The antibody gene families have several interesting organizational features. There are three distinct gene families - two code for light (L) chains, ~ and K, and the third codes for heavy (H) chains. They are composed of three distinct coding segments which are separated from one another by intervening DNA sequences - V (variable), J (joining) and C (constant). The V and J segments together comprise the V region of the antibody polypeptide which encodes the immunoglobulin domain concerned with antigen recognition. Moreover, each antibody gene family appears to contain multiple V and J segments. The antibody gene families present two fascinating biological problems. FirstA it has been estimated that mammals can synthesize io5 to 10° different antibody .molecules. What genetic mechanisms are responsible for this diversity of antibody molecules? We hope to assess the relative contributions of three genetic mechanisms: multiple germ line V genes (1), somatic mutation (2), and the joining in a combinatorial fashion of multiple V and J segments (3). Second, 18 how are antibo4y gene segments rearranged during the differeptiation of antibody-producing cells? These DNA rearrangements presumably are fundamental components of the molecular events that commit the antibody-producing cell to the synthesis of a single type of antibody molecule as well as contributing to antibody diversity in the combinatorial joining of V and J segments (3,4). We have focused on the analysis of the heavy chain gene family because, in addition to being an excellent system for ~tudying the p~enomena mentioned above, it has intricacies not exhibited in light chains. The heavy chain gene family of the mouse is comprised of an unknown number of variable (VH) gene segments and at least eight different constant (Ca) gene segments (5) (Figure 1). Heavy Family FIGURE 1. Heavy chain antibody gene family in mice. The order of CH gene segments is uncertain, although indirect evidence supports the following alignment: Cv3Cy1Cv 2 bCy2aCa (20). The number of VH gene segments is stili a malter of controversy. The heavy chain gene family also has multiple J segments that are not depicted in this figure (see text). The various classes and subclasses of immunoglobulins are determined by the CH gene segments (e.g., Cµ-IgM, Cy-IgG, C -IgA, etc.). Moreover, during the differentiation of the antibody-producing cell, distinct classes of immunoglobulins are expressed in a reproducible order (Figure 2). First IgM is expressed; later IgD and IgM are expressed; and eventually the other classes of immunoglobulins are expressed (6). In the lineage of a particular antibody-producing cell, it appears that these developmental shifts in immunoglobulin class expression occur by associating a particular VH gene segment with different CH gene segments while maintaining the expression of the same light chain gene segments. Therefore, a question of particular interest is the nature of the DNA rearrangements which lead to sequential and at times, simultaneous, expression of different heavy chain classes. Fortunately, tumors of antibody-producing cells exist which "freeze" this developmental pathway at many different points. Thus in time we will understand how the antibody gene organization for sperm cells (undifferentiated DNA) differs from that of tumor cell lines producing IgM, IgM + IgD and IgA (i.e., various stages of differentiation). Accordingly, our our initial efforts are focused on understanding the gene 19 organization in DNA at the beginning (speMll or embryo) and the end (lgA-producing myeloma) of a heavy chain differentiation pathway. committed surface lg secreting IQ FIGURE 2. The differentiation of B cells. A B cell first becomes committed to the expression of a particular V domain {one VL region and one Vff region) which is associated with cytoplasmic IgM molecules. Subsequently the IgM molecule is expressed on the cell surface. Later, cell-surface IgD molecules appear. Subsequent differentiation events lead to a terminally differentiated cell which specializes in the synthesis of soluble antibodies of one of a variety of immunoglobulin classes. For an individual B cell, the same V domain is associated with the various classes of immunoglobulins throughout the differentiation pathway. THE PHOSPHORYLCHOLINE ANTIBODY SYSTEM We have chosen to examine some of the questions posed above for a series of antibody-producing cells which synthesize immunoglobulin binding phosphorylcholine because this system allows us to analyze directly the biology of the immune response to phosphorylcholine (PC). Let us summarize the salient features of this system. First, several thousand myeloma tumors have been screened and twelve appear to 20 synthesize immunoglobulins binding phosphorylcholine (7). Our laborato~y ,has determined the amino acid sequences or th~ Va regions for seven of these tumors (8,9) and other laboratories have analyzed several additional sequences (10) (Figure 3). The Va sequences· from myeloma proteins binding phosphorylcholine illustrate several features of V diversity. 1) Four Va sequences are identical. Since these identical VH sequences were expressed independently in different mice, it appears that they are encoded by a germ line Va gene segment designated T15. This reasoning argues that it is un)ikely that four somatic variants would be identical in amino acid sequence. 2) The variant sequences differ by one to eleven amino acid substitutions and also exhibit sequence gaps. Accordingly, one can hope to determine the nature and extent of diversity generated from somatic genetic mechanisms by sequencing germ line PC VH gene segments and comparing them with the protein diversity patterns reflected in their myeloma counterparts. Second, antisera have been raised which are specific for the V domains of several myeloma proteins binding phosphorylcholine. These antisera are termed anti-idiotypic antisera. Anti-idiotypic antisera to T15 can be used to map genetic elements which control the expression of this VH domain. The T15 idiotype maps about 0.4 cent1Morgans (cM) from the CH gene cluster (11) and simplistic genetic calculations suggest the PC Ve and .CH gene segments are separated by hundreds of thousands or even a million nucleotides. For example, mouse chromosomes have about 25 chiasmata per meiosis (12). With a genome of 3 x 109 nucleotide pairs, 0.4 cM of DNA in the mouse would span about 106 nucleotide pairs, if meiotic recombination were random. Third, the T15 idiotype appears to be present on at least one type of T cells ("helper T cells") (13), implying that T-cell receptors and B-cell immunoglobulins may share the same Ve repertoire of genes. Thus an analysis of the phosphorylcholine system may provide opportµnities to analyze T-cell receptors. Finally, the hybridoma system of Milstein and KBhler (14) has been employed to generate homogeneous antibodies to phosphorylcholine. In collaboration with Dr. Patricia Gearhart, we are analyzing 20 hybridomas to phosphorylcholine in order to broaden our knowledge about the phenotypic diversity patterns of the phosphorylcholine system. The importance of detailed protein sequence studies on the products of complex multigenic systems such as the antibody gene families cannot be overemphasized, for these phenotypic diversity patterns are one or the end results of heavy chain gene organization and rearrangements and any meaningful understanding of this system at the DNA level must account for the resultant diversity of its gene products. Thus we hope the phosphorylcholine system will provide 21 insights into antibody gene diversity and organization and permit us, in time, to begin analyzing the more complex r,egulatory events or this sophisticated system. 80 TIS _Hv._mL.,, 120 5107 $63 ·Y5236 H8 M603 • I ------------G-l. . . . ------~-TO.,-I I ' t W3207 _____,.___...,_ _...._,- - - - - - F- - - l·lYDl:V,...-- 1 MSll M167 I ft I ----------0---------'"" : :•o F ; I -v ---+---+--T--f S-H-R V-T-TiA'D"C~ FIGURE 3. The amino acid sequences or VH regions from immunoglobulins binding phosphorylcholine. Identities of these sequences to the VH region of T15 are indicated by a straight line. The one letter code of Dayhoff is used to indicate amino acid substitutions (28). Deletions are indicated by brackets. Insertions are denoted by a vertical bar. The three hypervariable regions which fold in three dimensions to constitute the walls of the antigen-binding site of the V domain are designated by HV1, HV11 , HVrrI and dotted lines. OUR APPROACH We have constructed libraries in Charon 4A bacteriophage from partial restriction digests of sperm, embryo, and myeloma DNA (15, 16). The sperm and embryo libraries are &·source of undifferentiated DNA. The myeloma library, derived from the tumor MOPC 603 which synthesizes IgA molecules binding phosphorylcholine, represents a terminal stage in the differentiation of an antibody-producing cell. We also have purified mRNA from a variety of myeloma tumors, and used these as templates for the synthesis of double-stranded DNA copies which were then inserted into plasmids (16). Our initial approach has been to compare the genomic organizations of undifferentiated (sperm or embryo) and differentiated (IgA myeloma tumor) DNAs. To this end we have isolated a number of genomic clones from both the M603 library and from a sperm library, using cDNA probes for the complete VffCa coding region of myeloma protein S107. The VH regions of the 22 S107 and the M603 immunoglobulins are very closely related (Figure 3) and the corresponding mRNAs completely protect one another in S1 nuclease digestion experiments (16). Certain of .these initial experiments have recently been published in a paper which describes for the first time a heavy chain genomic clone containing the VH and Ca gene segments and the presence of intervening sequences within the Ca coding region, probably separating the coding regions for immunoglobulin a domains (16). These results as well as more recent observations are summarized below. EXPERIMENTAL OBSERVATIONS The Variable and Constant Regions of a Heavy Chains Appear to be Encoded by Distinct VH and Ca Gene Segments which are Rearranged During Differentiation. We have analyzed a series of overlapping genomic clones from the M603 library which have the general structures illustrated in Figure 4. The V and the C gene segments are separated by 6.8 kilobases. Furthermore, idiotypic mapping, discussed above, suggests that these regions were separated by hundreds of thousands of nucleotides prior to differentiation of this antibody-producing cell with the concomitant DNA rearrangements. A heteroduplex comparison of a sperm VH clone with the myeloma M603 clone, which will be discussed subsequently, also provides evidence for the rearrangement of the VH gene segment in the myeloma DNA. Accordingly, the v8 and C gene segments are originally widely separated from one anotger. As the antibody-producing cell differentiates, DNA rearrangements of antibody V and C gene segments occur over extensive stretches of DNA. The Ca Gene Segments from the M603 Myeloma Library are Present in Multiple Rearranged Forms. A comparison of Southern blots on sperm M603 DNA using the Ca probe demonstrates that the myeloma DNA has three forms of the Ca gene, none of which are identical to their germ line counterpart (Figure 5). These three forms have been isolated from the M603 library as Charon 4A clones (Figure 6). Restriction enzyme analyses and heteroduplex comparisons demonstrate that, although they share 2.7 or more kilobases of homology just 5' to the Ca gene, each of these three clones is distinct from the others in their more 5' regions. These observations raise several interesting possibilities. The absence of a germ line-like Ca gene segment in the H603 DNA suggests that the·Ca gene segments in both the maternal ai:i.!1 paternal chromosomes coding for heavy chain genes have been rearranged. Immunoglobulin-producing cells exhibit allelic exclusion; that is, a particular antibody- 23 ALPHA GENOMIC CLONES FROM MYELOMA M603 ONA ~I 6.5kb RI 7.3kb • RI 5.lkb RI 4.4kb RI -------------------------CH603al25 -------------------------CH603a6 FIGURE 4. The organization of VH and Ca gene segments from DNA derived from myeloma tumor M603. Kb denotes kilobases. R1 denotes Eco R1 cleavage sites. The distances between Eco R1 sites are indicated. CH603a125 and CH603a6 are two clones derived from the phage library of M603 DNA. The VH gene segment is separated from the Ca gene segment by 6.8 kilobases of intervening DNA. R-loop mapping and restriction enzyme analyses demonstrate that the Ca segment is divided into three approximately equal segments, presumably coding regions for the three Ca domains, by two small intervening DNA sequences (16). producing cell may express the maternal or paternal allele for a particular immunoglobulin family, but not both alleles. In the past the phenomenon of allelic exclusion has been explained by suggesting that either the maternal or paternal chromosome does not rearrange at the DNA level and, accordingly, cannot express an immunoglobulin polypeptide. This suggestion has come from S0uthern blot analyses of myeloma DNAs in which the germ line pattern of constant gene segments for light chains appears to be preserved (17). Our data on the alpha constant region genes of the M603 myeloma DNA suggests that both the maternal and paternal chromosomes undergo rearrangements, but that one of these rearrangements is_abortive in the sense no gene product is expressed. It will be interesting to determine whether these abortive DNA rearrangements include V gene segments; or whether only the C gene segment is involved in the rearrangement. Moreover, it will be interesting to analyze carefully the myeloma examples that appear to have germ line C fragments to determine whether the DNA rearrangements have been missed due to technical limitations of the Southern blotting technique, orcontamination with somatic DNA. It may be that all myeloma DNAs in fact rearrange both the paternal and maternal chromosomes--one in a productive and the second in an abortive fashion. 24 •mltr_yo/R1 p107oR5 M603/Rt p5'Ca . C»-7.4 v •: ca..-.,c3· 'ea I 0,--- 12.5 llb e:..,_ l.Okb c~ 5 •• "a, FIGURE 5. Southern blots of embryo (undifferentiated) and myeloma M603 (differentiated) DNAs. The picture on the left is a Southern blot of 13-day embryo DNA after digestion with the Eco R1 enzyme, separation of the DNA fragments on agarose, and hybridization with a cDNA probe derived from mRNA of myeloma tumor 5107. This probe contains both the VH and C coding regions. Assignments of the Ca fragments are base~ on Southern blots with separated VH and Ca probes (d~ta not shown). The remaining fragments must be VH gene fragments. _ Thus there are at least 8-9 germ line VH genes which cross-hybridize with the VH probe from myeloma tumor S107. The exposure on the right is a Southern blot of tumor M603 DNA after Eco R1 digestion and hybridization to a plasmid containing the 5' half of the Ca coding region (an R1 site separates the 5' from the 3' half of the Ca gene segment; see Figure 4). The 5' Ca probe gives just one 9.5 kilobase band in the embryo DNA (data not shown) and 5.1, 9.0 and 12.5 kilobase bands in the M603 DNA. Hybridization to the 3' half of the Ca coding region gives a 4.4 kilobase band in both embryo and myeloma DNA (not shown). 25 GERMLINE AND MYELOMA 5'Ca EcoRI FRAGMENTS 9.5kb 1 CU Germline 5 Ca 5.lkb lt111111111111111J11111111111 WM a6 PPW as 9.0kb 11111111111111111111111111111 12.5kb 11111111111111111m Myelomo 5'Ca a30 FIGURE 6. Eco R1 genomic fragments including the 5' portion of the Ca gene from myeloma M603 DNA and sperm DNA. The genomic clones a6, a9, and a30 have been derived from the M603 phage library. The structure of the germ line Ca clone comes from a Southern blot analysis of sperm or embryo DNA (Figure 5). The boxes represent the 5' portion of the Ca coding sequence (see Figure 4), whereas the hashmarks represent DNA homologies revealed by heteroduplex analyses. One surprising observation that is difficult to explain is the presence of three distinct Ca clones in the M603 DNA. Several explanations may be offered, none really satisfactory. First, the germ line may contain two Ca genes, both the same size by Eco R1 restriction analysis. Both of these Ca genes may undergo rearrangements of several different types. Second, perhaps the abortive rearrangement is unstable and may be subject to additional DNA rearrangements. Third, perhaps there are several different M603 cell types in the uncloned tumor from which the DNA was derived. The possibility that the M603 Ca pattern is some aberration of this particular tumor line seems unlikely because at least one other phosphorylcholine binding tumor (HS) has an identical pattern on Southern blots (M. Davis and P. Early, unpublished). Thus in the case of the Ca gene segments, it appears that both the maternal and paternal chromosomes undergo DNA rearrangements, some of which are abortive (nonproductive) while others lead to the expression of one VH-CH pair of gene segments. The V and C Rearrangements in Heavy Chains Resemble Those of Light Chains in Some Respects but Not Others. The VL and CL gene segments are rearranged by a fusion at the DNA level of VL and JL gene segments with the removal (or rearrangement) of the intervening DNA (Figure 7) (4, 17). Accordingly, the DNA 5' to the VL gene segment is identical to that of the unrearranged VL gene and the intervening DNA between the V and C gene segments is derived from the region 26 Light Chain v Sperm J c Heavy Chain v --c:::c---1f-il-c::J- Mtl\ \\ \ I \ ·c /?• I \ Myeloma c J I I \ V\J C ~ ? FIGURE 7. A model of the joining of light and heavy chain gene segments. An analysis of A {~) and K (18} light chain gene segments indicate that the 3' side of a V segment is fused to the 5' side of a J segment. The intervening DNA sequence between the J segment and the C segment remains unchanged in the DNA rearrangement process. The heavy chain gene segments appear to rearrange in a similar fashion, although the organization of the intervening DNA sequence between the J and C gene segments is altered, presumably because of multiple DNA rearrangements between one VH gene segment and two (or more) CH gene segments (see text). The existence of Ja segments for heavy chains is strong~y implied from protein sequence data (18) and has recently been demonstrated by the DNA sequence analysis of a sperm clone containing a Vtt segment (P. Early and M. Davis, unpublished observation). Comparison of a sperm VH clone and the joined Va and Ca myeloma clone (a6) by DNA heteroduplex analysis demonstrates that those regions 5' to the V segment are homologous and those regions 3' ·to the V segment are nonhomologous (Figure 8). In this respect the· heavy chain variable region gene segment appears to rearrange in a manner similar to its light chain counterparts (Figure 7). The rearrangement of VH and CH gene segments differs from those of the light chains in one important regard. Certain of the intervening sequences between the Vtt and Ca gene segments of the a6 clone (Figure 4) are not derived from germ line.DNA 5' to the Ca gene. For example, a Southern blot analysis of germ line DNA with a Ca probe shows that the closest Eco R1 site is 9.5 kilobases from the 5' side of the Ca gene segment (Figure 5). However, the a6 clone 5' to the unrearranged CL gene. 27 . HOMOLOGY OF GERMLINE VH WITH MYELOMA VH + Ca Germ line &.5 7.4 1.4 ~------------~--------.IL----~-~~ Myeloma ~-------&.5 Ca ------1-i------- --·----+------·IW¥·iW·------~ 7.3 5.1 4.4 FIGURE 8: Homologies determined by heteroduplex analysis between the flanking sequences of germ line and myeloma v8 clones. The dotted lines indicate germ line sequences. Accordingly, the intervening DNA sequence between the v8 and Ca gene segments is not derived from the sperm Ve clone. The sperm library was constructed by M. Davis and R. Joho. from the M603 DNA has an Eco R1 site 5.1 kilobases trom the 5' end of the Ca gene segment. In addition, as discussed above, this a6 DNA is not homologous to DNA of the sperm v8 clone (Figure 8). Moreover, the Eco R1 site of the a6 clone in the DNA between the V and C gene segments does not seem to have been created by a spurious mutation, since Southern blots of DNA from an independently derived tumor line (88) show the same Gi Eco R1 fragment. One explanation for the origin of the DNA sequence between V9 and «;i gene segments in the a6 clone containing this Eco R1 site is that it arises from the DNA rearrangement events or an earlier stage in differentiation, in which this Va gene segment was formerly joined to a diff~rent c8 (or J) gene segment. Indeed, during the differentiation of antibody-producing cells, the V9 gene segment appears initially to be Joined to a Si gene (Figure 2), so we would predict that some of the intervening DNA in the a6 clone between the V9 and Ca gene segments may be derived from the 5' side of a germ line Cµ gene segment. The subsequent joining of this v8 segment to a Ca gene segment later 1n development might displace or delete (19) the cµ gene, but not all of its flanking sequences. Intervening Sequences Appear to Separate the Domains of the CH Genes. The Ca polypeptide is divided into three discrete molecular domains, each of which encompasses about 110 amino acid residues (20). We initially used R-loop mapping to demonstrate the existence or two small intervening sequences (IVS2, IVS3) which separate the Ca coding region into three roughly equal segments (Figure qJ (16). Subsequent restriction enzyme analyses of the M603 genomic clone (a6) places IVS2 within 30 amino acids of the domain boundary 28 between the Ca1 and Ca2 homology units (16; M. Davis, unpublished) •. Thus it appears likely that the two intervening ,equences will separate the Ca gene into three distinct coding segments, one for each Ca domain (Figure 4). In addition, we have analyzed a µ genomic clone from the M603 library by R-loop mapping. The Cµ region has four domains (21) and, as expected, R-loop analysis demonstrates that the Cµ coding region is divided by three small intervening sequences into four roughly equivalent segments {K. Calame, P. Early, M. Davis, D. Livant, unpublished observations). The analysis . of a genomic y1 clone has established that intervening sequences separate the three Cy 1 domains and the hinge region from one another precisely at the interdomain boundaries (22). Therefore it appears reasonable to conclude that intervening sequences will divide all or the immunoglobulin C genes coding into segments for structural domains (see Figure 1). The function of intervening sequences has generated spirited controversy and discussion. Individual domains of the immunoglobulin molecule carry out discrete and independent functions (20). Accordingly, the immunoglobulin intervening sequences appear to perform the important task of breaking the coding regions into discrete units which may then rearrange independently of one another through recombination at either the DNA level or the nuclear RNA level as proposed by Gilbert (23). Several lines or evidence suggest that the domains of immunoglobulins may be discrete evolutionary units. First, Ca regions with two, three, and four .domains are present in vertebrate antibodies. Second, heavy chain disease deletions (24) and spontaneous deletions in tissue culture lines (2.5) suggest that frequent nonhomologous crossing-over occurs at or between domain boundaries. Perhaps intervening sequences not only separate domains but facilitate recombination as well. It will certainly be interesting to determine the homology relationships, if any, of the various immunoglobulin intervening sequences to one another. The Germ Line V Gene Segments or Mouse Heavy Chains Appear to be as Diverse as Their VK Counterparts. The Vtt regions derived from myeloma proteins binding phosphorylcholine show a limited range of heterogeneity (Figure 3). We are interested in determining whether these different Va sequences are germ line or in part derived by somatic mutatio~. Southern blot analysis of embryo DNA employing the S107 cDNA probe reveals at least 8-9 restriction fragments which hybridize to the 5107 V region probe (Figure 5). The PC Va regions represent a single group of heavy chain variable regions (26). Approximately 20 other groups of 29 VH regions have been defined (26). Therefore, if each group is on the average encoded by ~10 germ line genes, the heavy chain gene fami'ly may be comprised of approximately 200 VH genes. Since the amino acid sequence analyses of mouse VH regions are relatively limited, it appears likely that in time many additional VH groups will be defined. By similar analyses, the VK family of mouse appears to be encoded by 200 or more germ line V genes {3, 27). We have isolated several different PC VH genes and are now in the process or sequencing them to determine the relative contributions of germ line diversity, somatic mutation, and combinatorial joining of VH and Ja segments to antibody variability. The Generality of Nucleic Acid Rearrangements. The intriguing general question posed by the studies on immunoglobulin genes is whether DNA rearrangements are a fundamental aspect of differentiation in other eukaryotic systems. An answer to this question will await more detailed analyses of other gene families, both simple and complex. ACKNOWLEDGMENTS The work here is supported by National Science Foundation Grant PCM 76-81546. MD, PE, and DL are supported by National Institutes of Health Training Grant GM 07616. KC is supported by National Institutes of Health Fellowship GM 05442. REFERENCES 1. 2. 3. 4. 5. 6. 7. B. Hood, L., Campbell, J. H., and Elgin, S. C. R. (1975). Ann. Rev. Genet. 9, 305. Cohn,M., Blomberg, B·., Geckeler, w., Raschke, W., Riblet, R., and Weigert, M. (1974). "The Immune System," ICN-UCLA Symp., p. 89. Academic Press. Weigert, M., Gatmaitan, L., Loh, E., Schilling, J., and Hood, L. (1978). Nature 276, 785. Brack, C., Hirawa, M., Lenhard-Schueller, R., and Tonegawa, S. (1978). Cell 15, 1. Mage, .R., Lieberman, R., Potter, M., and Terry, W. (1973). In "The Antigens" (M. Sela, ed. ) , Vol. I, p. 300. Academic Press. Goding, J. W., Scott, D. W., and Layton, J.E. (1977). Immunol. Rev. 37, 152. Potter, M-:--f1970). Physiol. Rev. 52, 631. Hood, L., Loh, E., Hubert, J., Barstad, P., Eaton, B., Early, P., Fuhrman, J., Johnson, N., Kronenbergs M., and Schilling, J. (1976). Cold Spring Harbor §l!!!E· Quant. Biol. 41, 817. 30 9. 10. 11. 12. 13. 14. Hubert, J .• , Johnson, N., Barstad, P., Rudikorr, s., and Hood, L. In preparation. Rao, D~ N., Rudikoff, S., and Potter, M. (1978). !!£chemistry 17, 5555. Riblet, R. J. (1977). "Molecular and Cellular Biology," ICN-UCLA Symp., Vol. 6, p. 83. Academic Press. Klein, J~ (1975). "The Biology or the House Histocompatibility Complex." Springer-Verlag. Cosenza, H., Augustin, A., and Julius, M. (1977). Cold S~rin~ Harbor~· Quant. !!.21· ~1, 709. ---Khler, G., and Milstein, c. (1976). Eur. J. Immunol. 6, 511. ~ - 15. Maniatis, T., Hardison, R., Lacy, E., Lauer, J., 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. O'Connell, c., Quon, D., Sim, G., and Efstratiadis, ~. (1978). Cell 15, 687. Early, P-:-:-Oavis, H., Kaback, D., Davidson, N., and Hood, L. (1979). Proc. Nat. Aoad. Sci. USA 76, 857. Seidman, J. G., and'Leder, P:-rf97~ NatUre 276, 790. Schilling, J., Clevinger, B., Davie, J., and Hood, L. In preparation. Honjo, T., and Kataoka, T. (1978). f!:.2£.· Nat.~­ Sci. USA 75, 2140. Edelman, G. M., Cunningham, B. A., Gall, W., Gottlieb, P., Rutishauser, U., and Waxdal, M. (1969). 12:.2£. ~­ Acad. Sci. USA 63, 78. Beale,-0:-, and Feinstein, A. (1976). Quart. Rev. Biophzs. 9' 135. Sakano, H., Rogers, J. H., Huppi, K., Brack, C., Traunecker, A., Maki, R., Wall, R., and Tonegawa, s. (1979). Nature 277, 627. Gilbert, W. (1978). Nature c'71, 501. Frangione, B., Lee, L~, Haber, E., and Bloch, K. (1917). Proc. Nat. Aoad. Sci. USA 70, 1073. ide'tugbo, K~ilstein~., and Secher, D. (1977). Nature 265, 299. Barstad, P., Rudik~ff, s., Potter, M., Cohn, M., Konigsberg, W., and Hood, L. (1974). Science 183, 962. Seidman, J., Leder., A., Nau, M., Norman, B., and Leder, P. (1978). Science 202, 11. Dayhoff, M. O. (1972). In "Atlas of Protein Sequence and Structure," Vol. 5, Bio1riedical Research Foundation, Washington, D.C. fr/aturt Vol. 283 21 February 1980 31 An imlnuIJ.oglobulin heavy-chain gene is formed by at least two recombinational events Mark M. Da~*, Kathryn Calame*, Philip W. Early*, Donna L. Livant*, Rolf Johot, Irving L. Weissmant & Leroy Hood* •Division of Biology, California Institute of Technology, Puadena, California 91125 t Laboratory of Eiq>erimental Oncology, Department of Pathology, Stanford Medical School, Stanford, California 94305 The events of B-cell differentiation can be reconstructed in pan through an analysis of the organisation of heavy-chain gene segments in differentiated B cells. A mouse immunoglobulin a heavy-chain gene is composed ofat least three noncontiguous germ-line DNA segments-a VH gene segment, a JH gene segment associated with the C,. gene segment, and the C• .gene segme1{t. These gene segments are joined together by two distinct types of DNA rea"angements-a V-J joining and a CH switch. THE antibody genes provide a unique opportunity for studying the molecular basis of one pathway of eukaryotic differentiation because the rearrangement of gene segments is correlated with the expression of antibody molecules. Antibody molecules are encoded by three unlinked gene families-A, IC and heavy chain (H) 1• The A and IC families encode light .(L) chains and the heavy-chain family encodes heavy chains. The light chains are encoded by three gene segments, VL (variable), JL (joining) and CL (constant), which are separated in the genomes of cells undifferentiated with regard to antibody gene expression 2•3• During differentiation of the antibody-producing or B cell, the VL and h gene segments are rearranged and joined together while the intervening DNA between the JL and CL gene segments remains unmodified 2°"'. This process of DNA re1rrangement is termed V-J joining. During the expression of the rearranged gene in the differentiated B cell, the coding regions as well as the intervening DNA between the JL and CL gene segments are transcribed as part of a high molecular weight nuclear transcript. The intervening region is then removed by RNA splicing to produce a light-chain mRNA with contiguous vL. JL and CL coding segments5 '6 • Recently, we demonstrated that the heavy chain contains three analogous gene segments, VH, JH and CH, which undergo a similar type of V-J joining during B-cell differentiation'-09. Each antibody-producing cell synthesises only one VLh polypeptide sequence and one V.JH sequence, which together form the antigen-binding (V) domain of the antibody molecule. The heavy-chain genes seem to have a. special role in the differentiation of the antibody-producing cell as reflected in a second phenomenon known as the ·switching of heavy-chain constant regions or CH switching. Antibody molecules can be divided into five different immunoglobulin classes-lgM, lgD, lgG, lgA and IgE-which are determined by one of eight distinct heavy chain genes [that is, C,., C,, (C,.i. C,.2a, C,.2.,. C,. 3 ), C. and c.]. Each class of immunoglobulin seems to be associated with unique functions such as complement fiution or the release of histamine from mast cells. The process of B-cell differentiation is complex and a variety of conflicting views exists on the transitional stages. It is generally agreed, however, that lgM is the earliest immunoglobulin class that is expressed in the differentiation of a B cell (see ref. 10). The immature B lymphocyte apparently has the capacity to differentiate along a variety of discrete pathways and produce progeny which may IWitch from IgM expression to the expression of any one of the other classes of immunoglobulins. The terminal stage of B-cell differentiation is the plasma cell, which is committed to synthesise and secrete large quantities of a single molecular species of antibody. Several studies suggest that the specificity or V domain docs not change as different immunoglobulin classes 002l-08_36/IO/OI0733-G'710l .OO are expressed throughout this differentiation process•1-1 5, During CH switching, light-chain expression remains unaltered. We are interested in studying the molecular mechanisms which permit a B cell (or its progeny) to express successive classes of antibody molecules. Here we demonstrate that an a heavy-chain gene derived from a terminally differentiated plasma cell is composed of three noncontiguous germ-line DNA segments. These gene segments are joined together by at least two distinct DNA rearrangements-a V-J joining and a CH switch. Organisation of the a heavy-chain gene is altered in difterentiated as opposed to undifterentiate~ DNAs We have constructed 'libraries' of recombinant phage 16 containing large (12-20 kilobases) inserts of mouse DNA in the vector Charon 4A (ref. 17). These libraries contain sufficient numben of recombinants (-106 ) for there to be a high probability that most single-copy sequences of a given genome are included. We have previously reported the construction of such a library from the DNA of the mouse IgA-producing myeloma tumour M603, and the subsequent isolation and characterisation of clones containing rearranged or differentiated genomic DNA. These clones, CH603a6 (a6) and CH603a125 (a125), gene segments on a single fragment of DNA'·'. have VH and The EcoRI restriction maps of these clones are shown in Fig. 1 a. The a6 clone has three EcoRI fragments-one 7.3 kilobases long containing the VH gene segment, a second of 5.1 kilobases containing the 5' portion of the gene segment, and a third of 4.4 kilobases containing the 3' portion of the c. gene segment. Approximately 6.8 kilobases of intervening DNA separate the VH and gene segments. The a 125 clone also has three EcoRl fragments-one of 6.5 kilobases which is located on the 5' side of the 7.3- and S.1-kilobase EcoRI fragments also described -for a6. The genomic clones were isolated using a cloned cDNA plasmid representing the entire heavy-chain mRNA of the lgA-producing tumour S107 (denoted S107 cDNA)7 • The Sl07 V" region is about 98% homologous to the M603 VH region at the nucleotide Ievel9. Analysis by the Southern blot procedure11.19 of EcoRI-digested M603 DNA hybridised to the S 107 cDNA probe demonstrates the presence of three EcoRI fra~­ ments corresponding to those described above in the a 6 clone . Thus, the rearranged a6 clone is not an artefact of the cloning or isolation procedures. When the Sl07 cDNA probe is med to examine a Southern blot .of EcoRl-digested sperm or embryo DNA, a somewhat Jlifterent pattern is observed'. In particular, the S.1-tilobase c. c. c. Nature Vol. 283 21 February 1980 32 fl: li 6.5111> 'ltJ 7.31ib Rt ~lkb Rt 4,41ib RI --------4"--"'4 ---------CH6()3a6. CH603at25 b Ullb - RI . TAiib Rll,41ibRI ---------tCH5p"'3 I IU 1.$11b I Rt 9..5kb .-. CHSpa22 ~ Rlt.5kbR I 6.2kb RI IQ.5kb T.Tkb Rt c. c.. RI 3.211b CHSpa~ c~ In view of the possibility that lymphocytes earlier in the M603 lineage might first have produced IgM molecules and later lgA molecules, we decided to investigate the possible contribution of germ-line sequences as well as VH and sequences to the myeloma a gene (Fig. la). We constructed several libraries of germ-line DNA (sperm) and proceeded to isolate clones containing VH, and gene segments using cloned cDNA probes (denoted Sl07 VH, C. and C,.) for the corresponding codin§ regions and the screening procedure of Benton and 0 Davis • EcoRI restrietion maps of several germ-line clones are shown in Fig. lb. We chose sperm (germ-line) DNA for our undifferentiated genomic libraries to eliminate any possibility of DNA rearrangements which ~ay occur in somatic tissues during embryogenesis. The and V H clones seem to be representative of germ-line sequence organisation because the EcoRI fragments containing the corresponding coding regions (in the clones) are identical in size to those found in the Southern blot analysis of undifferentiated DNA with and S107 VH cloned cDNA probes-9.5 and 7 .4 kilobases, respectively (Fig. 2a-d). The V H and C. probes were derived from restriction fragments of the S107 cloned cDNA' extending from approximately the 5'-untranslated region to codon 108 (VH) and from codons 108 to 274 (C. ). Figure 2a shows that a Southern blot of a germ-line c.. Ca Ca c. The a gene is composed of at least three difterent germ-line segments of DNA v ltl EcoRI fragment containing most of the large intervening sequence and the 5' portion of the gene segment is not present. This observation indicates that the a6 clone is the product of one or more DNA rearrangements which presumably occurred during B-cell differentiation'. CHSflfL27 c. c. ~T c. lkb SCALE COOING REGIONS • TRANSCRIPTION &'-3' Fi&· 1 o, a Heavy-chain clones isolated from a genomic library of myeloma M603 DNA. CH603 a6 and CH603 a125 are two overlapping clones derived from a partial EcoRI library of M603 DNA, constructed using the phage Charon 4A (ref. 7). The position of the respective gene segments, as well as their direction of transcri~tion, was determined by R-loop mapping and restriction analysis ·'. 11, Germ-line VH, c.. and c. clones isolated from a aenomic library of sperm DNA from inbred BALB/c mice. Sperm DNA33 •34 was partiall.y digested with restriction enzymes in two ways: (1) with a mixture of HoeIII plus Alul and (2) with EcoRI alone. After digestion in conditions designed to maximise the yield of 12-20-ltilobase fragments, these fragments were selected on sucrose gradients. The Haelll/ Alul fragments were methylated with EcoRI methylase and blunt end ligated with synthetic EcoRI deavage sites and then cleaved with EcoRI essentially as described previously 16 except that EcoRI linker ligations were done at 18 •c to lessen the endonuclease degradation. Both types of fragments were then ligated to the isolated arms of the bacteriophage Charon 4A (ref. 16) at 4 •c. The enzymatically recombined DNAs were packaged in vitro using the strains of Stemberg35 and the protocol of Hohn36 • The efficiency of packaging was 400,000 ·plaque forming units (PFU) per ,,..g insened DNA in the case of EcoRI partially digested DNA and 200,000 PFU per ..,g for Haelll/ Alul digestions. The background of non-recombinant Charon 4A was 25% an~ . < 5%, respectively. Approximately 500,000 EcoRI and 1,200,000 HaeIII/ Alul clones were constructed and amplified. A library of - 500,000 clones provides a 90% chance of finding a &iven si~e copy sequence and a library of 1,200,000 clones a 99% chance3 • The use of enzymes that recognise three different aequences reduces the possibility that a particular region of interest is lost from the library be~usc it had too many or too few restriction sites to fall within the sucrose gradient size cut. Ubraries •ere screened'· 16 with the cDNA clone (see text) S107 or Ml04E C,. cloned cDNA probes identified by DNA sequence analysis9 •38• The c.. clone used extends from codon 300 to the 3'-untranslated region (-1,000 base pairs). 31• Independent overlapping clones •ere obtained from the c. and C,. aene sepnents. Location of the eoding regions and the direction of transcription were determined by DNA sequence analyses for the CHSp PC-3 done, by beteroduplex analyses with the a6 clone for the CHSp a29 clone, and bl.R-loop mapping and restriction analyses of the CHSp,,.27 clone . The aerm-line aenomic clones will be denoted ,,.21 (CHSp,,.27), VH3 (CHSpVH3), and 029 (CHSpa2!il). Ca VH b c d 0 r CIL e I . I'.· ' -;-7.4kb ...... -9.5kb J! ;:.: f.f v fls. 2 ! l -12.2kb ·' ~~l Southern blot analyses of C., C,. and 8107 VH coding regions in germ-line DNA. Approximately 3-20 !Ao& of sperm DNA or 13-day embryo was digested with EcoRI and electrophoresed on a 0.7% neutral agarose gel for 10-12 hat 30-40 V. Gels were then blotted according to the procedures of Southern and Flavell 1 a. 19 and hybridised with 32 P nick -translated cDNA39 • Washing was for 1.5 bin 1 M NaCl, 1 M Tris pH 8, 0.1% SDS and 0.1% IOdium pyrophosphate at 65 •c and for 1 b in 1 x SSC, 0.1 % SDS and 0.1 % IOdium pyrophosphate. Lanes a and b are 029 and embryo DNAs, respectively, hybridised to the c. cDNA probe. Lane c is VH3 DNA hybridised to the Sl07 cDNA probe. Lane d i11perm DNA hybridised to the S107 VH cDNA probe. This probe crossreacts with eight or nine closely related VHgene sepnents1 . Lanese and fare ,,.21 and sperm DNAs, respectively, hybridised to the c.. cDNA probe. Identical results in all cues were obtained with BALB/c 13-day embao DNA or sperm DNA as has been found for a V. aene . Sizes (given in ltilobases) were determined by comparison with restriction fragments of PBR322 (ref. 40) or by the use of PBR322 multimers aenerated by BamHI cleavaae of PBR322 and limited li1ation of th.e resultifta monomer. Nature Vol. 283 21 February 1980 33 c· counterpart seen on a Southern blot analysis of sperm DNA witn a clonedµ cDNA probe (12.2 kiJobases; see Fig. 2{). We believe this discrepancy arises from one, (or more) dclction(s) in the DNA flanking the C,. coding region during the propagation of the recombinant phage. In isolating µ clones from the M603 library, we obtained several 9.5-10-kilobase EcoRI fragments containing C,. coding regions, whereas Southern blot analysis of M603 DNA with a cloned C,. cDNA probe demonstrated a genomic fragment of 12.2 kilobases, as in the sperm DNA (data not shown). Attempts to isolate C,. clones from a library of mouse liver DNA have led to similar results (N. Newell and F. Blattner, personal communication). We will present restriction enzyme data below which demonstrate that this apparent deletion in the µ clone does not affect our general conclusions. The germ-line V"' and C,. clones were compared with the myeloma a6 and a12S clones by hetcroduplex analysis. Representative hcteroduplexes from each of these comparisons show extensive homologies. The germ-line V H clone shares approximately 11.6 kilobases of homology with the myeloma a12S clone (Fig. 3a). This homology extends from the S' end of the al2S clone up to and including the VH coding region. The germ-line C,. clone shows 5.0 kilobascs of homology with the large intervening sequence of the myeloma a6 clone (Fig. 3b). Starting at its 3' end, the germ-line c. clone has about 6.4 kilobases of homology with the myeloma a6 clone (Fig. 3c ). The heteroduplex measurements for these analyses arc given in Table 1. .......... i clone (a29) digested With EcoRI and hybridised With the C. cDNA ptobc yields a 9.S-kilobase fragment. Asimilar analysis of cmbrY9 ))NA produces a band of identical size (Fig. 2b ). Southern blriis of a germ-line V H&lone (VH3) and embryo DNA digested with EcoRI and hybridised to a S107 VKcDNA probe gave. respectively. a 7 .4-kilobase fragment (Fig. 2c) and several fragments including a 7 .4-kilobase fragment (Fig. 2d). An iJ11portant question is the relationship between the germ-line VH3 and myeloma M603 V11 coding regions because there are at least eig~t.VH gene segments that hybridise to the 5107 probe'. The germ-line VH3 DNA sequence codes for the 5107 VH protein sequence• and, accordingly, differs from the M603 VH region by a minimum of four base changes leading to three amino acid subsitutions21 • Thus, the germ-line VH3 and the mycloma M603 V H gene segments may be encoded by two distinct germ-line VH gene segments or the VH3 gene segment may give rise to the M603 VHgene segment by somatic mutation and selection. As we shall show subsequently, the VH3 clone ICCIDS to be indistinguishable.from the M603 clone in the VH coding· region and in more than 11-kilobascs of S' -flanking sequence by hcteroduplcx and restriction enzyme analyses. Therefore, our analyses of the localisation of VH sequences in the myeloma a6 clone arc valid because the germ-line V Hclone serves as a probe for the M603 VH gene and its attendant S'-flanking sequence. The EcoRI fragment of the germ-line C,. clone containing the C,. coding region is smaller (10.S kilobascs; sec Fig. 2e) than its c. B 01603a125 ::<r::lf==C==::::Cvw:.-.,l:::< I) Ch5p~3 ~ A 0l603a6 ~27 0 ~ B ~a22 M~a6 Germ-line c,. E A Cll603a6 Myeloma a 125 Germ-line vH C D ~ Myeloma a6 Germ-line Ca B ....·--.... ,... ~,......,,.... Jlla. 3 Heteroduplex analyses of germ-line and somatic clones. The electron micrographs arc shown on the left, tracings of these heteroduplexcs in the middle and diagrammatic representations on the far right. a, Mycloma a125/germ-line VH PC3. b, Myeloma a6/germ-line p.27. c, Myeloma a6/germ-linc a22. Letters A-E indicate single-strand and double-strand regions for which measurements arc given in Table 1. In corresponding line drawinp, (r) and (I) refer to the ri&ht and left arms of the A vector. Blocks indicate coding regions in these figures, CsCI purified pbaac particles were treated with 0.1 M NaOH for 10 min at 20 •c to lyse the phage and denature the DNA simultaneously. After aeutralisation of the mixture, the DNA was allowed to reanneal in 50°1" (v/v~ three times recrystallised formamidc for 45 min at 20 •c before spreadina for electron microscopy••. . Table 1 M~utements of heteroduplex molecules fleteroduplex II . ft{yeloma cr12S/ ,crm-line VH3 b ft{yeloma cr6 / . ,crm-line p.27 t Myeloma cr6/ 1erm-line cr22 No.of Molecules A Distance in kilobases D B" c· 3S 4.S:t: 0.4 7.2:t: . 11.6:t: 0.4 0.7 30 4.8:t: S.1 :t: 26 S.O:t: 6.7:t 7.S:t o.s o.s 0.4 9.3:t: 10.4:t 0.9 6.4:t 1.2:t: 0.2 o.s E 0.4 o.s o.s Measurements were standardised relative to two circular DNA molecules on the same grid [single-strand ¢1Xl 74 DNA (S,375 bases) 111d double-strand pBR322 DNA (4,365 base pairs)]. Letters refer to ngions indicated in Fig. 3. The c:Omplete clone designations are given in fig. 3. In heteroduplex a, C refers to the region of duplex between the 1erm·line vH·clone and the myeloma clone. A and Bare the non· homologous single strands of these clones. Similarly, C for betero· duplexes b and c refers to the duplexes formed between myeloma and senn·line or clones. c.. c. Comparative restriction analyses confirm and extend the heteroduplex data discussed above. A detailed restriction map . for the M603 myeloma clones was o~tained by double digestion with pairs of restriction enzymes and is shown in Fig. 4a. To compare the placement of these cleavage sites with those of the germ-line clones, the 5.1-kilobase restriction fragment of the a6 clone (Fig. 1a) that spans the region joining the germ-line c. and C,. sequences was subcloned. Using this fragment as a probe, detailed restriction comparisons of the myeloma a6 clone and the germ-line C,. and C0 clones were made. Representative data are shown in Fig. 5 for these comparisons. Figure Sa, b and c represents HincII plus EcoRI digestions of the myeloma a6 5.1-kilobase RI fragment, the germ-line a29 clone and the germ-line C,. clone, respectively. These digests were electrophoresed on an agarose gel, blotted onto a nitrocellulose filter and hybridised with the labelled 5.1-kilobase RI fragment. This enables homologous restriction fragments to be identified rapidly and precisely (arrows indicate identical fragments). Figure Sd, e and f shows a similar analysis using Hi~III plus EcoRI digestions, respectively, of the myeloma 5.1-kilobase subclone, the germ-line a29 clone and the germ-line µ27 clone. These data, as well as additional restriction analyses of the germ-line V H3 clone, demonstrate that 4 out of 4 restriction sites in the germ-line VH3 clone, l 0 out of 10 sites in germ-line a29 clone and 9 out of l 0 sites in the germ-line µ27 clone corresponded exactly to those found in the myeloma a6 clone (Fig. 4). Not only do these restriction analyses independently confirm the heteroduplex results, but they also suggest that the component germ-line sequences of the a6 clone are very similar to their germ-line counterparts. Thus, the heteroduplex and restriction analyses demonstrate that the germ-line VH gene segment and its 5'-ftanking sequence, although not identical to its M603 counterpart, are very similar and may be used to analyse VH gene segment organisation in the myeloma M603 clones. DNA sequence analyses of the myeloma a6 clone and the germ-line µ.27 clone have demonstrated that the heavy~hain gene fami~ does have distinct VH and JH gene segments in the term line . Moreover' the germ-line JH gene segment corresponding to that expressed in the myeloma a6 clone is associated with the germ-line C,. aene9 • This JH gene segment contains the Hhal site that marks the end of the homology between the germ-line µ27 clone and the myeloma clones (Fig. 4 ). Accordingly, the distinct germ-line VHand germ-line JH gene segments are rearranged in the myeloma clones in a manner analogous to the Vi. and Ji. aene segments of myeloma light chain genes2 ·3• Although it seems unlikely. the M603 JH gene segment and flanking sequences could have beer:i fused to a germ-line C,. clone as the result of some cloning ~fact~ To eliminate this possibility, we decided to demonstrate the germ-line association of the JH flanking sequence and C,. sequences by Southern blot analyses. On different slots of the same agarose gel, we electrophoresed mouse sperm DNA cleaved with either Hincll or EcoRI, transferred these DNAs to a nitrocellulose filter and hybridised one lane of each digest with a C,. probe and a second lane with the 5.1-kilobase EcoRI fragment from the intervening sequence of the myeloma a6 clone (Fig. 6). Figure 6a and b shows Southern blots of EcoRI-digested sperm DNA hybridised with the C,. cDNA probe and the 5.1-kilobase EcoRI subclone &om the a6 clone, respectively. In both lanes, a 12.2-kilobase band corresponding to the germ-line C,. fragment can be identified. In a second digest (HinclI) of mouse sperm DNA, a similar result is obtained with both the C,. probe (Fig. 6c) and the 5.1-kilobase EcoRI subclone (Fig. 6d) hybridising to a 9.3-kilobase HinclI fragment. The 9.5-kilobase EcoRI band in Fig. 6b and the 5.0-kilobase band in Fig. 6d correspond to restriction fragments. In each case, one of the two bands from the S.1-kilobase EcoRI probe co-migrated with the single c.. DNA fragment. This analysis shows that at least part of the intervening sequence from the myeloma a6 clone is adjacent to the C,. gene in the germ line. Thus, the apparent deletion in the aerm-line µ27 clone is not a significant factor in our discussion. A summary of these analyses is presented in Fig. 7. The myeloma a6 clone is composed of three noncontiguous germline gene segments: (1) a VH gene segment and its 5'-ftanking c. i i I RI " I UM> i I RI h 1.i- - . ; ; y~ ~r n~k' 1t I T..11111 RI l.IM> RI 4.4 • RI i b ! y ...._....,.~ VH ....__,_______ I RI RI I • I I ~ p i I iI.Ii}JIJ c~ c,. RI RI Fla· C a. Restriction map of myeloma clones cr6 and a125. Specific cleavage sites were determined using double enzyme di&estion and sizing by gel electrophoresis. Size standuds were restriction digests of PBR322 and PBR322 multimers (see FiJ. 2 legend). HindlII, HincII, Pstl and Sad cleavage sites are only shown for the S.1-kilobase EcoRI fragment. b, Restriction sites corresponding to those of the myeloma clones detected in the VH. and aerm-line clones (Fig. 1). Restriction sites in re&ions corresponding to the 6.S~ and 7 .3-kilobue RI frapients of a6 and a12S were mapped by double digestion as above. All sites within the S.1-kilobase RI frasment were compared by co-migration and blotting as described and illustrated in Fig. S. The Hhol site marked by an.8$terisk in the C,. done was the one site found not to conform with those or the mycloma cr6 clone. c.. c. 737 Nature Vol. 283 21 February 1980 a b c e d Hine II Hind ill Eco RI Eco RI + f + Fig. 5 Comparative restriction digests of the a6 myeloma, germline C µ and C"' clones. Lanes a-f show parallel Hincll +EcoRI and HindIII + EcoRI digests of the 5 .1-kilobase EcoRI subclone of the a6 clone (Fig. la) and DNA from the germ-line a29 and µ27 clones (Fig. lb). Samples were electrophoresed in 1% agarose, transferred to nitrocellulose and hybridised with nick-translated 5.1-kilobase a6 subclone DNA. The co-electrophoresis of restriction fragments allows a large number of different restriction sites to be compared rapidly. In this way, all 16 mapped sites in the 5.1-kilobase a6 subclone were compared with equivalent sites on the a29 and µ27 clones. Arrows show coincident bands. The CH switch may be explained by any one of several genetic models Several mechanisms of CH switching have been proposed 11 •22 - 26 , many of which are similar to those proposed for V-J joining27 • These models can be categorised as involving either DNA rearrangements to replace one constant region with another (successive deletions, excision-insertions or inversions) or the differential processing of a large nuclear RNA transcript containing multiple heavy-chain constant-region genes 26 • The RNA processing model seems unlikely as a general mechanism for CH switching at the level of antibody-secreting plasma cells. High molecular weight nuclear RN As from three myeloma tumours hybridise only with a cDNA probe complementary to the class of immunoglobulin expressed in that tumour and not with probes from non-expressed immunoglobulin classes 28 • Moreover, cell fusion experiments hybridising two different myeloma cells demonstrate that the hybrid cells synthesise only parental VHCH combinations 29 , a result in conflict with the simple RNA processing model. Furthermore, neither Cy 1 nor C,, DNA probes hybridise on Southern blots with the M603 a clones (data not shown), contrary to one prediction of the RNA processing model. On the other hand, the evidence presented here strongly supports DNA rearrangement as a fundamental element in the mechanism for heavy-chain switching. As summarised in Fig. 7, a very large segment of Cµ flanking sequence has been brought adjacent to C"' and VH gene segments in the active a gene of myeloma tumour M603. The creation of the M603 a gene apparently requires two DNA rearrangements (Fig. 8). A CH switching a c b d ....... sequence, (2) flanking sequences located 5' to a germ-line Cµ gene which includes the JH gene segment, and (3) the germ-line C"' gene segment with its flanking 3' and 5' sequences. Note that within the limits of the methods used here these three germ-line gene segments and their attendant flanking sequences apparently cover the entire myeloma a6 clone. 12.2kb 9.5kb The a heavy-chain gene is formed by at least two recombinational events Two distinct DNA rearrangements have occurred to form the M603 a gene-V-J joining and CH switching (Fig. 8) 9 • The simplest interpretation of these observations is that the VHgene segment is first joined to a JH gene segment linked to the Cµ gene segment. This V-J joining, analogous to that which occurs in light chains, generates a rearranged µ gene that presumably leads to the expression of IgM molecules. V-J joining also commits an individual lymphocyte to the expression of a single V domain that remains invariant throughout subsequent steps of B-cell differentiation. Later, a CH switch joins the V-J gene segment to the C" gene segment to create a functional a gene and thus enables the differentiated lymphocyte to express lgA molecules. Thus, the myeloma a6 heavy-chain gene is assembled by two distinct and presumably independent DNA rearrangement events. Our heteroduplex, Southern blot and restriction mapping data show unequivocally that these two rearrangements occur at two distinct sites in the genome. These two sites of rearrangement are termed the V-J joining site and the CH switch site, respectively. These data are consistent with a differentiation pathway in which a B cell may switch from lgM to lgA synthesis while expressing the same V domain. We cannot establish that these two DNA rearrangement events occurred at different times, although this supposition is reasonable. We also cannot rule out the possibility of intermediate differentiation states in this B-cell lineage where IgG molecules were produced, although there is no direct evidence for such a stage. Cµ. 5.lkb ~I Eco RI 9.3kb Cµ. 5.lkb Eco RI '---v----' Hine II Fig. 6 Southern blots of the 5.1-kilobase EcoRI and C,,fragments. Mouse sperm DNA was digested with EcoRI or Hincll and 3 µg was loaded onto a 4 mm x 20 x 20 cm 0. 7% agarose gel, electrophoresed at 40 V for 10 hand blotted as described. Probes used were either the S.1-kilobase EcoRI fragment or a C,. cDNA clone nick-translated to 4 x 10 8 c.p.m. per µ.g. Washing was as described in Fig. 3 legend except that lanes hybridised with the 5.1-kilobase EcoRI fragment were washed further in 10 mM NaCl, 10 mMTris, 0.1 % SDS and 0.1% NaPPi for 2 hat 68 °C to reduce the signal strength of weakly homologous repeats. Filters were exposed for 12 h with an intensifying screen at - 80 °C. The faint band above the 12.2-kilobase band in b is a partial digestion product. All lanes shown were run on the same gel and blotted simultaneously, alignment being assisted by inclusion of pBR322 multimers as internal standards (see Fig. 2). 36 Nature Vol. 283 21 February 19~0 138 pNA rearrangement is not postulated in th~ ~N~ processing aiodels 26• The data. presented here d't not dtStmgutSh betwee'1 the various types of DNA rearrangements proposed,. but the h bridisation kinetics· experiments of Honjo ·and Kataoka are ~nsisteitt with a deletional mechanism for the CH switch 25• In addition, recent experiments suggest that V-J joining in mouse A genes is accomplished by a del~tio1:1al mechanism'. Thus, both types of DNA rearrangement, V-J joining and CH switching, aiay arise thtough deletional mechanisms. H the deletional 1Dodel is correct for either type of DNA rearrangement, the differentiation of B ceUs is irrevenible because chromosomal information is lost with the excision of each deletional loop of chromosome. It will be interesting to determine· whether all joined a genes have the same switch site. Southern blot analyses of the DNA from a second IgA-producing myeloma tumour, HS, with the 5.1-kilobase EcoRI probe yield a restriction fragment pattern identical to that of M603 DNA (data not shown). In particular, the 5.1-kilobase EcoRI fragment (Fig. 1) which contains the switch site seems the same. These data suggest that the expressed a genes in both M603 and H8 myeloma tumours have the same CH switch site. However, Southern blot analyses of several . other closely related IgA-producing tumoun do not show a S.1-kilobase EcoRI fragment (M.M.D., unpublished observation). Thus, there may be multiple CH switch sites for the C. gene segment. As it seems that B cells producing lgM or lgG may switch to the production of lgA, perhaps distinct CK switch points reflect distinct pathways of B-cell differentiation. It will also be interesting to determine the location and number of switch sites for other immunoglobulin classes. If each CH gene segment has a unique site or set of sites for CK switching, one may be able to trace the distinct pathways of B-cell differentiation by studying the sequence organisation of each functional heavy-chain gene. DNA rearrangements of antibody gene segments lead to combinatorial amplification of immunoglobulin information V-J joining and CH switching are mediated by DNA rearrangements which display combinatorial properties that amplify the germ-line information encoding the antibody gene families. (1) The V and J gene segments of one antibody gene family may be joined in a combinatorial manner to generate divenity in the third hypervariable regions of both " and heavy chains'·1·u0 ·31 • For example, mice may have at least 200 VK and S JH gene segments that may be joined combinatorially to generate 1,000 different V HJ Kcoding regions. (2) One V domain may be combinatorially switched among eight or more different CH regions to carry out a variety of different effector functions tillii@ I ii! RI RI Vt1 v" fii!iilil!ilirlilil!!iJil!Ji!li!Jiil!iiit' !i!!iJ!!ilii!!ll!i!!!!i!!!ID RI Co !i!i!iii!!il!!ill!iilil@hf.li!i@!li!!i'V;~ RI »?+Pfn RI t I I RI RI I RI I c,. man RI -5.0kb-~ I RI ' V•.J joi11i119 ' '"IWitclliftt "" .=. .; ltM Eoprau'°" ¢ •I-zm111=..-._c:---•·-•tp1 cA.,_:::>.._IZZl!Z'=m!i..,~-¢: . V.J fl&. 8 • Two types of DNA rearrangements leading to the creation of the myeloma a heavy-chain gene V-J joining and CH switching. V-J joining indicates a DNA rearrangement that joins the VK and JH gene segments. Because the JH segments seem to be associated aene9 , V-J joining permits a ~ chain and with the germ-line laM molecules to be expressed by the differentiating B cell. CH switching denotes a DNA rearrangement that replaces the C,. gene aene segment. This second rearregement segment with a presumably permits an a chain and lgA molecules to be expressed by the now fully differentiated lymphocyte. c. that arc directed at eliminating antigen or triggering defensive mechanisms such as complement fixation. Thus, each recognition (V) domain may be switched to many different effector (C) domains. The combinatorial properties of antibody gene segments and polypeptides therefore contribute to several fundamental aspects of the vertebrate immune response-V region divenity and the combinatorial switching of antigcnrecognition (V) domains with a variety of different effector (C) domains during B-cell differentiation. Other complex eukaryotic gene families may use similar DNA combinatorial mechanisms for information amplification 32 • This work was supponed by NSF grant PCM 76-81546, ACS grant IM56 and USPHS grant AI09072. M.M.D., P.W.E. and D.L.L. are supported by NIH training grant GM 07616. K.C. is supported by NIH Fellowship GM 05442. RJ. is a fellow of the Swiss National Foundation. I.L.W. is a faculty Research Awardee of the ACS. All experiments involving recombinant organisms were conducted in accordance with the revised NIH Guidelines on recombinant DNA, using P2, EK-2 or P3, EK-2 containment. We thank David Goldberg for bis gift of PBR322 multimcrs, Tom Sargent for packaging extracts, Keichi ltakura for synthetic RI linken, and David Anderson, Norman Davidson Max Delbriick, Richard Flavell, Henry Huang, Tom Ma~iatis and Tom Sargent for helpful discussions. a-Mid 1 Qc:lober 1979; .-pied 22 Ju_, 1980. nw l. Maae. Jl., Liebermu, Jl., Poller, M. It TenJ, W. in ~led. Sela, M.) 299-376 (Mademic. Net\' York, 1973). l5, 1-14 (1978). 2. Brack, c .. Hirowa. A., l..enbard·Scbueller, R. "Tonepwa, s. 3. Seidman, J. G., Mas, E. E. It Leder, P. N - 218, 37G-37S (1979). 4. Sakano, H., Huppi, IC, Heinrich. G. It Tonepwa, S. "'-" - · 218-294 (1979). 5. Gilmore·Herben, M. It Woll, R.. Proc. 11&111. At<MI. Sci. U.S.A. 75, 342-345 (1978). 6. Scllibler. u.. Marcil, IL •. " Perry, R. P. 15, 1495-1509 (1978). 7. Euly. P. w.. Dava. M. M.• Kabact. D. a.. N." Hood. L l'roc. AcM. Sci. U.S.A. 76, 157-861 (1979). I. Davil, M.• Early, P., Calame. K~ l.Mftt, D. It Hood. L. in E~ a.- Rt,..U.lion !eds Axel, R. Manialii. T. It F01t, C. f.) 393"406 llCN UCLA Symp., Aadeaiic. N-York.. 1979). t. Euty. P. W•• Huns. H. V., Da..a. M. M., c.llme. K. It Haod. L QU llllbmiftedl. 10. Raft, M. C. a.Id Spills Har6. .....,_Biol. 41, 15~162 0976). 11. 51edse, C., Fain, D.S., Blaclt. B., 9'rie1er, R. G. ii: Hood, L PNc.-. MM. Sci. 11.s.A '3, 92M27 (1976). 12. Fllllenbllr&. H. H.• Wua, A. C., l'ink. J. R. L It LeWi. A. S. AM. N. Y. A<U. S<i. 1"SOl-506 (1971). U. Wua. A. C., Geilely, J. It Fudenbcrs. H. H. BJocllnWlry U. 521-534 (1973). 14. Levin. A. S., F"*nber&. H. H.. Hopper, J.E., Willon. S. It Niaonolf. A. l'rrtc. MMI. ScL U.S.A. 61. 169-171 (1971) . . 15. Wana, A. C., Willon, S. IL, Hopper, J.E., FlllleDberJ, H. ff. It Nilonoll. A. l'rot. ,..,.._ Aall. ScL U.S.A. 66, 337-343 (1970). t6. ~ T. •.t. C.CU 15,687-701 (1978). 17. Blattner, F. R . .i aL titoa 19', 161-169 (1977). 11. Soutbern. E. M. I.-'«. Biol ti. 503-51? 0977). 19. Wreys, A. J. It Flavell, R. A. C.U U. 429-439 (1977). 20. llenlOll. W. D. It 0.vis, R. W. Scinoa 19', llG-11211977). 21. Hood. L•.t. Cold Spift1Har6. ....,.i Biol 41.,117-136 (1976). 22. Smilbilll. 0. Sdmct "'· 182-~13 (1970). 23. Hood. L. Fe'1t l'rec. 31, 177-178 U972l. . 24. 9evan. M. J., ~. R.. M. E .• Ww.-a, A. I.. It ~ B. A. ,.,..._ Bioplty•. • . . . . Biol. u. 131-162 (1972). c.eu c.ea o......... s.,...,.. RI I I RI RI ---ll.6kb---. I t c:jJlir:a:za-=m:a:m:!!l;12t.-111= c.. Gene organisation studies may delineate distind pathways of B-cell differentiation M6()~ ,1, I Rt RI -&.6kb- na. 7 Orisins of the three aerm-line components of the myeloma er heavy-chain acne. Various types of lbadina indic:ate bomolo&Y by heteroduplex ealyses ed by restriction enzyme analyses. s.,,..,,. 37 N,1turt Vol. 283 21 February 1980 :5. ffonjo. T. 4 Ka~.ka. T. Proc. iia1>1. A&&d. sCi. U.S.A. 11, 2140-2144 (1971). . 26. Jtabbiru, T. ~· Nilrv'< 2'75, 291-296 (1978). %7. T-pwa, $., HOflmli, V .. Manhys1Cn, G." Sf'ullcr, IL Cold $pillc '"'6. S,..,......t Biol. 41, 177~89 0976). . is. Mare11, K. 8 .• Schibler.. V. 4 Perry, IL P. SO.- IN. 1087-lDll (1979). '9· Schlliman. M. J. A Kohler. G. N•w'< 2'4, 917-919 (1971). JO. MU. E .• Seidman, J. G. 4 lAder, P. r.oc. Miii. A&Mt. Sci. U.S.A. 'N. J4~J454 (1979). JI. SchiUina. J., Oeavinaer. 8., Davie, J. 4 HOOd, L Ne""' JU, J~ (l980J. . J2, Hood. L. Huans. ff ... Dreyer. w. J. /. lllpnllllO/cc. 531-559 (1977). J3. Wilkin. S., Xantaold. G. A lleadich, A..r.oc. kU. Sd. U.S.A. '2, 3295-3299 $...a.'· . ~A . ,._,, J4. Joho,ll., Wcilulwl.l.L,Early,l'~Cole,J.Affood.Lflk. -MU.Sci. V.S.A.(intlle JS. Slcrnber&. N., T.-ier, D. A EeqUt. L 0... l, 25S-2IO Um). J6. Hohn. 8. A Mutm. It. r.oc....... Aud. Sd. U.S.A. 'N, 3259-3263 (Im). J7. Clarke. L. Cart.on. J. t, 91-99 (1976). JI. C'.alame, It., J~ Dnil. M.. Earlr, P~ Una!. D~ Wall, ll. A Haod. L ~ a a.n. c:.a ""bmined). J9. Maniatis, T ~Jd!ref. A.A Kleid. D. G . .PNc. Aud. ScL U.s.A. '2, 1114-llll 11975J. 40. Sondillc, J. G. Nw:lftc Nib Rn. J, 2721-nll (1971). 41. Dnil. ll.. Simon, M. A Dmdoan, N. ,.,._ 51z,_ UD, 41M21 Ctt71). The primary structure of 16S rDNA from Zea mays chloroplast is ho1ttologous to E. coli 16S rRNA Zs. Schwarz• & H. Kosselt lnstitut fir Biolo&ie Ill der lJniversitit Freibura. Schinzlestnsse 1, D·78 Freiblq i. Br.. FRG The nucleotide sequence of ribosomal DNA coding for 16S rRNA from Zea mays claloroplast has Men determined. A comparison with the 16S rRNA sequence from Escherichia coli reveals strong homology and thereby demonstrates the prokaryotic nature of chloroplast ribosomes from a higher plant. CHLOROPLASTS of green plants contain 70S ribosomes which differ from the eukaryotic SOS type ribosomes of the cytoplasm 1 • The organelle-specific ribosomes mediate translation of organelle-specific mRNAs, which in tum are encoded in chloroplast DNA 2•3• Organelle-specific rRNA• and tRNA 5 species are also encoded in the chloroplast genome, which thus provides the basis for the semiautonomous nature of the organelle. This, along with the various prokaryotic characteristics of the chloroplast-specific translation system, has lent strong support to the endosymbiont hypothesis~. This hypothesis postulates that modem chloroplasts are descendants of endosymbiotic prokaryotes-particularly cyanobac:teria ' which have entered eulcaryotic cells during evolution. Due to a supposedly heavy functional constraint of a highly complex entity such as the ribosome, it seemed likely that its components would have changed comparatively slowly during evolution. Therefore, a comparison of rRNA primary structures from chloroplasts and E. coli should provide a quantitative measure of their evolutionary relationship, even if they are separated by a large phylogenetic distance. As an extension of previous work based on T 1-oligonucleotide catalogues'... of 16S rRNAs, complete sequencing should not only allow the euct positioning of conserved T 1 oligonucleotides, but also a comparison of regions in which larger T 1 oligonucleotides are absent. Furthermore, a differentiation between strongly conserved regions and regions which have 1one through changes during the course of evolution will be possible only on the basis of complete sequences: We have, therefore, analysed rI>NA from ua mays chloroplasts at the nucleotide level, which should allow deduction of the primary structures of the corresponding rRNA species and comparison with the corresponding rRNA species from E. coli. Furthermore, it was anticipated that analysis of sequence homology between orpnelle and E. coli rRNA specie& would support the prediction of secondary structures if they are common or would help to exclude them if they are not preserved. In this article we report the primary •Present addresl: Bioloaical Laboratories, Harvard University, Cambridae, Musachusetts 02138. t To whom corrnpondenc::e should be lddreued. 002a-GIJ6/IO/OI0'7~IOl .OO structure of 16S rDNA from ua mays chloroplast and compare it with E. coli 16S rRNA recently analysed by others 10•11 • Sequencing of maize chloroplast 16S rDNA DNA from ua mays chloroplasts is a circular molecule of about 135 kilobase pairs on which two rRNA coding regions are positioned in opposite orientation within two 22 kilobase-pair inverted repeats~. Each inverted repeat includes within a 12 tilobase-pair EcoRI fragment one copy of a 16S rRNA, 23S rRNA, 4.SS and SS rRNA aene and a 2 kilobase-pair spacer region between the 16S and 23S rRNA p:nes. One such 12 tilobase-pair fragment bu been linked to the plasmid vector pMB9 within the E.coli clone pZmc134 (ref. 4). This clone was therefore used for isolation, mapping and sequencing of the 16S rRNA coding DNA fragments, which are ~tioned within the left half of fragment BamHl·C (Fig. l)'·1:i;n. In Fig. 2 the RNA-like strand of the entire 16S rRNA aene from maize chloroplast is depicted (lower row of each line). Together with maize chloroplast 4.5S and wheat SS rDNA aequences Dyer et al., unpublished) it represents-to our knowledge-the first sequence analysis of a complete acne originating from a bi&her plant. The exact termini of the struc:tural part of the p:ne cannot be located by DNA sequencing and must await sequencing of the RNA ends. Taking into account the inhomology of the first four positions, it teems not .-nllkely that the position of the 5'terminal nucleotide of maize 16S rRNA is different from posi· don l, although very dose to it. At the opposite end positions 1,540 (last homologous position) or 1.541 (non-homologous, but in accordance with the chain end of E. coli 16S rRNA) lppCU as the most likely candidates coding for the 3'-terminal nucleotide of the chloroplast 16S rRNA. Nepecting this uncertainty a chain length of 1,491 results, which, due to several deletions, is 50 residues shorter than the E.coli 16S rRNA. rr. Sequence homolou between maize chloroplast and E. coll 16S rRNA When the maize chloroplast 16S rDNA aequence is compared •with the 16S rDNA10 or rRNA 11 from E.coli (Fig. 2) eKtensive 38 DNA Sequences Mediating Class Switching in Alpha Immunoglobulin Genes Science (1980) In Press 39 , ·Abstract•.. Imipunoglobulin class switching involves specific DNA rearrangements of the gene segments coding for heavy chain constant regions during B-lymphocyte dif(erentiation. In two different cases of Cµ to ~ switching examined here (Tl5, M603) and one taken from the literature (MClOl), three different ·sites 5' to Cµ and three different sites 5' to Ca are joined together in the process of CH switching. The sequences surrounding the three germline Ca sites of recombination are highly conserved blocks of 30 nucleotides which may serve as recognition sequences for CH switching to the Ca gene. This putative recognition sequence is repeated 17 times in approximately 1400 nucleotides of germline Ca 5' flanking sequence. The lack of homology between this Ca sequence and sequences reported for the Cyt and C-y2b switch sites suggests that heavy chain switching is mediated by class-specific recognition sequences and, presumably, class-specific regulatory mechanisms. In addition, it appears that in one example, MC101, CH switching has progressed from C µ -+ Ca -+ Cy1. This switching pathway presents difficulties for the simple deletional model of CH switching. 40 · The antibady qiolecule is comprised of discrete molecular domains that . carry out two general. types of functions. The variable or V domain binds antigen and the c9nstant or C domains trigger effector functions such as complement fixation. The V and C domains arise from the interactions of two different polypeptides, light (L) and heavy (H), which in turn are encoded by a series of discrete gene segments-VL, JL (joining), and CL and Vff, D (diversity), JH and Cff, .respectively (1-4). During the differentiation of antibody-producing or B cells, two distinct types of DNA rearrangements of these gene segments occur (4, 5). One type generates the VL and Vff genes by joining directly the VL and JL and the Vff, D, and JH gene segments, respectively. These DNA rearrangements are termed V-J or V-D-J joining and they are, in part, responsible for the generation of antigen-binding diversity in V domains. A second type of DNA rearrangement, termed Cff switching, allows an important flexibility in the use of a given antigen-binding site. At an early stage of B-cell differentiation, an individual B cell initially expresses IgM molecules with a single V domain (VL-VH combination) (6, 7). Later, this B cell or its clonal progeny may express another immunoglobulin class while continuing to express the same V domain (8). Since the class of immunoglobulin is determined by the Cff region (e.g., Cµ-IgM, Cy-IgG, Ca-IgA), the B cell must shift from the expression of a Cµ gene to the expression of another Cff gene during differentiation. Thus, Cff switching associates a particular antigen-binding specificity, the V domain, with a series of different effector functions encoded by the various Cff regions. Two types of experiments have provided insights into the mechanism of Cff switching. First, Honjo and Kataoka (9) have employed hybridization kinetics to determine the numbers of CH genes in a series of mouse myeloma 41 tumors producing -diff~rent immunoglobulin classes. Their results suggest that the Va gene is rearranged from one Ca gene to a second by a deletion of the intervening DN~ between the Va .gene and the second CH gene. From these data a heavy chain gene order of 5'-Cµ-Cy3-Cyi-Cy2b-Cy2a-Ca-3' was suggested. ·Recent experiments using cloned probes and Southern blot analyses generally support the gene order and deletional mechanism proposed by aonjo (10). Second, we examined the rearranged (expressed) Cl gene in an IgAproducing myeloma tumor (M603) and obtained direct evidence that DNA rearrangement mediates Ca switching (4). The rearranged M603 a gene is composed of three distinct germline gene segments-Va, Ja with 5' Cµ flanking sequences and Ca with its flanking sequences. This tripartite structure of the rearranged a gene suggests that the Va gene was initially associated with the Cµ gene by DNA rearrangement through V-D-J joining and expressed as aµ chain in the IgM molecule. A subsequent DNA rearrangement could then replace the Cµ gene with the Ca gene by linking together 5' Cµ and Ca flanking sequences. The point at which the flanking Cµ and ~ sequences join in the rearranged gene and the corresponding breakpoints on the germline DN As are termed the switch (S) sites. Subsequently, other laboratories have obtained evidence for similar Cµ to Ca switches in rearranged yl (11) and y 2b (12, 13) genes. Three Examples of JgM + IgA Switching In order to investigate the molecular mechanisms underlying Ca switching, we sequenced the switch sites of two rearranged a genes and compared these rearranged switch sequences with their germline counterparts in the 5' flanking sequences of the Cµ and Ca genes. To this end, we have constructed genomic libraries (14) from the DNAs of M603 (15, 16) and an additional IgA-producing myeloma tumor, T15, and isolated the rearranged a genes. Homologies between 42 clones corresponding t:o the rearranged a genes of T15 and M603 are depicted in Fig. 1 and compared· with their germline Cµ and Ca. counterparts obtained from a genomic libFary of mouse sperm DNA (4). These homologies were established by detailed restriction enzyme analysis (data not shown). Both rearranged a genes exhibit the tripartite structure of Vu. Ju with Cµ flanking sequences and Ca gene segments. The size of the intervening sequence between the Vff an~ Ca coding regions is substantially different in these two cases-5.4 vs. 6.8 kilobases (Fig. 1). Since each of the two VH gene segments is joined to the same JH gene segment (17), the variation in size in the intervening sequences between the V and Ca coding regions may be the result of using different sites for CH switching in each of these rearranged Ca genes. In addition, we found that a rearranged yl gene, MClOl, whose sequence was reported by Honjo and colleagues (11) contains a 500 nucleotide region of Ca flanking sequence between Cµ and Cyl derived sequences (Fig. 1). The evidence for this supposition is that this 500 nucleotide sequence has been localized solely to a region 5' to the Ca gene by Southern blotting analyses {18) and restriction mapping using fragments containing all or part of this region as probes (11; M. Davis, data not presented). Therefore, this fragment is apparently represented just once in the genome and, accordingly, must have been derived from flanking sequences 5' to the Ca gene. Furthermore, DNA sequence analyses of the corresponding germline Ca sequence show virtual identity (approximately 95 percent) with the sequence found between the Cµ and Cy1 flanking sequences in the MClOl yl gene (Fig. 4a) (19). Thus we feel the MClOl yl gene is composed of several distinct germline sequences: a Vff gene (T. Honjo, personal communication), flanking sequence for the Cµ gene, flanking sequence for the Ca gene and the yl gene with its flanking sequences. 43 The arrows in Fig. 1 indicate the regions analyzed by DNA sequence analysis in our laboratory. The· DNA sequences of the MC101 S region (11) and the rearranged M603 and T15 a genes and their germline Cµ and Ca counterparts are shown in Fig. 2. Examination of the rearranged switch sequences indicates that all three examples juxtapose Cµ and Ca derived sequences. However, each switch sequence seems different from the others. The arrangement of sequences in these genes suggest that at least three distinct types of switching may occur. We denote these categories "simple," "complex," and "successive." In the simple category, T15, the Cµ flanking sequence joins directly to the· Ca flanking sequence. Similarly, a y2b gene (M141) has been found to join Cµ flanking sequence to that of Cy2b (12, 13). In the complex category, M603, a short sequence of 287 base pairs, is interposed between the Cµ and Ca flanking sequences. This sequence appears to derive from a region 3' to SM603 on the Cµ gene (Fig. 1). Probes containing this sequence hybridize strongly to restriction fragments containing or adjacent to the Cµ gene in Southern blotting analyses (data not shown). The region of hybridization corresponds to the 1.5-2.5 kb .region 5' to the Cµ gene (5) which deletes spontaneously upon cloning and hence is not present in our Cµ containing clones. Thus the complex category is explained by two distinct deletions-Ca swit~h and a deletion within the Cµ flanking sequence. This deletion of Cµ flanking sequence (at 8Mso3) does not appear to be a random event in that another a producing tumor line, M167, switches at exactly the same point adjacent to the Cµ gene. However, the complex category reflects DNA deletions seen to date only in myeloma cells and, accordingly, may or may not be biologically significant. In the third category, "successive" switching, the MC101 yl gene contains Cµ, Ca. and Cy1 flanking sequences between Va and C'Yl gene segments. Therefore, it appears to have switched twice, once from Cµ to Ca and subsequently from Ca to .Cy1 • • .. 44 The. DNA sequ,ence data in Fig. 2 show that the three rearranged genes · employ three germline Cµ switch sites up to 300 base pairs apart and three germline Ca switch sites up to 1350 base pairs apart. The locations of these germline switch sites, Sr15• SM603• and SMc101 are depicted in Fig. 1. Thus multiple switch sites exist for Ca. switching in sequences 5' to both the germline Cµ and Ca genes. ~NA Sequences Mediating Heavy Chain Switching The DNA sequences involved in V-D-J joining are quite distinct from those implicated in CH switching. The inverted repeat CAC~GTG occurs at the 3' end of antibody V gene segments and at the 5' end of the J gene segments (2, 4, 20). This inverted repeat is believed to be a recognition sequence that mediates the juxtaposition of the V, D, and J gene segments to allow subsequent joining by site-specific recombination (for a proposed mechanism, see ref. 4). This inverted repeat is not found in the fianking regions surrounding any of the Cu switch sites. This sequence also is missing from the switch sites for a yl (11) and a y2b gene (12, 13). Therefore, CH switching and V-J joining emp~oy distinct mechanisms for DNA rearrangement. In an effort to determine what sequences are important in Cff switching, the _sequences of the three Cµ sites and the three Ca sites were compared (Fig. 3). The germline Cµ MC101 and T15 switch sites share significant homology (15/25 nucleotides), although 16 nucleotides separate the actual switch points. We believe that these homologies ·are significant and may represent general ~equence requirements for Cff switching adjacent to the Cµ gene. Neither the MC101 nor the T15 Cµ sites share any homology with that of M603. The switch site of a y2b producer M141 (12, 13), also depicted in Fig. 3, is nine nucleotides away from that of Tl5 and may indicate that both y2b and a switching can utilize the same recognition sequence adjacent to eµ. 45 The sequences,, around ·each Ca switch site are even more highly conserved. Each germline Sa site occurs within a block of 30 conserved nucleotides (Fig. 3). Twenty:two bases are identical and seven of the remaining eight nucleotides are conserved with regard to type of base (i.e., purine-purine or pyrimidine-pyrimidine substitutions). The three points of recombination differ within each of these conserved sequences. Since the first three rearranged a genes examined switched at distinct Ca sites, we reasoned that there must be additional Sa sites. We determined the DNA sequence of some 1400 nucleotides in the region of these germline Sa sites (Fig. 4a) and found seventeen 30 nucleotide repeat sequences (Fig. 4b). These sequences are extremely homologous to each other (boxed regions) (Fig. 4c). Since these repeated sequences represent 510/1400 nucleotides in the region analyzed, the CH switching into three of these sites does not appear to represent random DNA rearrangement. We suggest that most of these repeats are potential germline S0 sites. Evidence for Class-Specific Regulation of Ca Switching The data presented in t~is paper, taken together with sequence data on y2b (12, 13) and yl (11) switch sites from the literature, lead to several important inferences about the mechanism of Cff switching. The germ line sequences for Cff switching appear to be class specific with regard to Cy1. C0 and Cy2b switch sites. The germline Cy1. C0 , and Cy2b "switch" sequences are compared in Fig. 5. The yl sequence is identical to the prototype Sa sequence in 10 out of 22 bases and the y2b sequence is identical for 7 of 22 nucleotides (only 3 of which are contiguous). Thus, various germline Sa sequences are far more similar to one another than to the germline Sy1 or Sy2b sequences (Fig. 3). One explanation for these sequence differences is that CH switching is mediated by class-specific recognition sequences. 46 Since the gerll!line Sa and Sµ sequences are not homologous (Fig. 3), ·homologous recombination cannot account for their joining. We believe the joining may be .mediated by a number of distinct. types of "switching" proteins (Fig.· 6). For example, one switching protein (Pa) may bind the germline Sa sequence and a second CPµ) may bind one of the germline Sµ sequences. These proteins may then interact to form a heterodimer which juxtaposes the V-D-J gene with ~he Ca gene segment (Fig. 6). . The multiple germline S0 sequences would increase the probability that Ca switching could occur once the appropriate joining protein is expressed. Because the germline Sa, Sy1. and 8y2b sequences seem distinct, different joining proteins could bind these sequences. Accordingly. the developmental regulation of the expression of these proteins would lead to class-specific regulation of CH switching. The Implications of Successive Cff Switching in MClOl The evidence for "successive" switching in the MClOl clone indicates that two or more CH switches can occur in a particular B-cell line. In the simple deletional model for class switching proposed by Honjo (9), Cn switching progresses in a linear fashion,. deleting intervening CH genes at each stage. As mentioned, the experiments supporting this model (9, 10) indicate a gene order of Cµ-Cy3-Cy1-C)'2b-Cy2a-Ca. Paradoxically, however, the MC101 clone appears to have switched from Cµ to Ca and then from Ca to Cyl• contrary to the linear deletional model for class switching proposed by Honjo (9). Several explanations seem plausible. 1) The CH gene order is Cµ-Ca-Cy· This seems difficult to support because of the large number of myeloma tumors which express the Cy1 gene, and still contain Ca genes (9, 10). In contrast, those myeloma tumors which express the Ca gene generally appear to have deleted their Cyl genes (10). However, until these CH genes are ordered in the germline DNA, this remains a formal possibility. 47 •- 2) lnterchromqsomal recombination. In this scheme, Cµ ~ Ca rearrange.ment on one chromosome could be followed by recombination with a Cy1 gene on another cflromosome to produce the MClOl yl mosaic gene CVH-G,aGx..:Cyt>· Moreover, a prediction of this model is that the reciprocally rearranged chromosome should have a rearranged <;i gene. 3) Episomal deletion. If the deleted DNA between the Vff gene and the Ca gene for med a circular intermediate (an episome) that was at least transiently stable, the episome could then re-integrate into the chromosome, replacing the Ca gene with the Cyt gene. This model is easily testable since it predicts the presence of a ~ gene, a Cy3 gene, etc. in the MClOl genome and, in particular, the Cµ and Ca _genes should be rearranged. The rearranged MClOl yl gene. accordingly, raises two general possibilities with regard to CH switching and normal B-cell differentiation. i) The successive (and complex) types of CH switching observed here may arise from one or more aberrant chromosomal rearrangements that are unique to myeloma cell lines and will not generally be seen in normal B cells. The numerous cell divisions that occur between myeloma tumor production and our analyses of the ~orresponding DN As, as well as the aneuploid nature of myeloma cells, make this a serious possibility. ii) Normal B cells may employ multiple CH switching mechanisms-some perhaps different from any cited above. The developmental regulation of CH switching may operate at several different levels. i) The nature of the sequences mediating V-D-J joining and CH switching implies that these phenomena are regulated independently. ii) The existence of distinct switch sequences for a, yl, and y2b genes implies that the expression of these classes may be developmentally regulated at the 48 level ·of DNA rearrQ.ngement, depending, for example, on which specific switching . protein is expressed. Mark M. Davis Stuart K. Kim Leroy E. Hood Division of Biology California Institute of TechnolOgy Pasadena, California 91125 49 References and Notes 1. C. Brack, A. Hirawa, R. Lenhard-Schueller, S. Tonegawa, Cell 15, 1 (19'18). 2. · H. Sakano, K. Huppi, G. Heinrich, S. Tonegawa, Nature 280, 288 (19'19). 3. J. G. Seidman, E. E. Max, P. Leder, Nature 280, 370 (1979). 4. P. Early, H. Huang, M. Davis, K. Calame, L. Hood, Cell 19, 981 (1980). 5. M. M. Davis, K. Calame, P. W. Early, D. L. Livant, R. Joho, L L. Weissman, L. Hood, Nature 283, 733 (1980). 6. M. C. Raff, Cold Spring Harbor ~- Quant. Biol. 41, 159 (19'16). 7. M. D. Cooper, J. F. Kearney, P. M. Lydyard, C. E. Grossi, A. R. Lawton, Cold Spring Harbor Symp. Quant. Biol. 41, 139 (1976). 8. B. Pernis, L. Forni, A. L. Luzzati, Cold Spring Harbor Symp. Quant. Biol. 41, 1'15 (19'16). 9. T. Honjo and T. Kataoka, Proc. Natl. Acad. Sci. U.S.A. 75, 2140 (19'18). 10. S. Cory and J. M. Adams, Cell 19, 37 (1980); C. Coleclough, C. Cooper, R. P. Perry, Proc. Natl. Acad. Sci. U.S.A. 77, 1422 (1980); T. H. Rabbits, A. Forster, W. Dunnick, D. L. Bentley, Nature 283, 351 (1980); Y. Yaoita and T. Honjo, Bfomed. Res., in press; Y. Yaoita and T. Honjo, Nature, in press. 11. T. Kataoka, T. Kawakami, N. Takahashi, T. Honjo, Proc. Natl. Acad. - Sci. U.S.A. 7'1, 919 (1980). 12. N. Takahashi, T. Kataoka, T. Honjo, Gene, in press. 13. H. Sakano, R. Maki, .Y. Kurosawa, W. Roeder, S. Tonegawa, Nature, submitted. 14. T. Maniatis, R. C. Hardison, E. Lacy, J. Lauer, C. O'Connell, D. Quon, G. K. Sim, A. Efstratiadis, Cell 15, 687 (1978). 15. P. W. Early, M. M. Davis, D. B. Kaback, N. Davidson, L. Hood, Proc. Natl. Acad. Sci. U.S.A. 76, 857 (1979). 50 16. M~ M. Davis, P. W. Early, K. Calame, D. L. Livant, L. Hood, in Eukaryotic ,} Gene Regulation, R. Axel; T. Maniatis, and C. F. Fox, Eds. (Academic Press, New York, 1980), p. 393. 17. N. Johnson, S. Rudikoff, P. Barstad, L. Hood, in preparation. 18. E. M. Southern, ~- Mol. Biol. 98, 503 (1977). 19. The minor sequence differences noted might arise from one or more of the following sources: polymorphism in these flanking sequences arising from mice in different localities; mutations arising in the DNA sequences of myeloma tumors during repeated passages; repair or correction mechanisms during the CH switching process; and/or DNA sequencing errors. 20. E. Max, J. G. Seidman, P. Leder, Proc. Natl. Acad. Sci. U.S.A. 76, 3450 (1979). 21. W. D. Benton and R. W. Davis, Science 196, 180 (1977). 22. A. Maxam and W. Gilbert, in Methods in Enzymology, Nucleic Acids Part I, L. Grossman and K. Moldave, Eds. (Academic Press, New York, 1980), vol. 65, p. 499. 23. This research was supported by NSF grant PCM 76-81546, and USPHS grant AI 09072. M.M.D .. and S.K.K. are supported by NIH training grant GM 07616. All experiments involving recombinant organisms were conducted in accordance with the NIH Guidelines on recombinant DNA. We wish to thank Nate Kupperman for assistance with restriction mapping and Stephen Crews, David Goldberg, and Mitch Kronenberg for helpful ·discussions. : 51 Figure Legends Fig. 1. Clones containing rearranged alpha heavy chain genes and germline Cµ and Cci genes~ The rearranged. Tl5 and M603 genes were isolated from genomic libraries of their respective myeloma tumors, and MClOl was isolated as described (11). Clones containing the germline Gi and Ca genes were isolated from libraries of mouse sperm DNA (5). regions. Raised boxes denote coding The shaded areas indicate homologies with germline genes and their IIlJ , Vff and flanking region; I] , Cµ and flanking region; E] , Ca and flanking regions; and 0 , uncertain origin, probably flanking sequences: Cµ derived (see text);~ , Cy1 and flanking regions. by detailed restriction mapping. follows: Homologies were determined Restriction enzyme sites are denoted as Hf = Hinfl, Hh = Hha, H = Hindlll, M = Mspl, R = EcoRI, Rs = Rsal, S = Sau3a, and Mb = Mboll. The position of the JH regions was determined from experiments described in refs. 4 and 13. The M603 clone was isolated and characterized as previously described (5, 15, 16). The T15 library was made from EcoRI partially digested DNA as described (5, 14). Recombinant phage were screened by the procedure of Benton and Davis (21) using 32p nick translated Vff + Cci cDNA clones as probes as described (5, 14). Southern blot analysis (18) of EcoRI digested myeloma DNA indicated identity of central EcoRI fragments in the Tl5 genomes with the clones shown here (data not shown), indicating that the clones isolated are representative of their genome of origin. (16). This analysis has been previously reported for M603 The DNA sequence analysis employed the procedures of Maxam and Gilbert (22). Arrows originating from a restriction site indicate 5' end labeling with y32P-ATP, while arrows ending in a restricting site indicate 3' end labeling with ci 32P-dNTPs. Fig. 2. Switch sequences for the rearranged T15, M603, and MC101 genes and their germline Cµ and Ca counterparts. Cy1 corresponding sequences for MClO~ 52 have·. been reported by aonjo and colleagues (11). Ca switch sites art! shown in ~' + 3' orientation. sequence identity of unrearranged Cµ ( D Underlining indicates ) and Ca ( l:J) flanking sequences with their counterparts in the individual rearranged 10 nucleotide spacings. Sequences surrounding genes. Dots indicate DNA sequence analysis was performed as described (22). Fig. 3. A comparison of the DNA sequences of the germline Cµ and Gi switch sites for the rearranged T15, M603, and MC101 genes. Sequences obtained from Cµ and Ca flanking regions were aligned for maximum homology around their Ca switch sites. Boxes indicate nucleotide identities and - indicates gap introduced for homology alignment. switch sites of each rearranged gene. for a y2b producer, Ml41 (12, 13}. Arrows indicate breakpoints for the Also indicated is the Cµ switch site A consensus sequence for at least some cases of Cµ switching might therefore be GGTNATTANNNNNNGGTANNCAAAG which does not occur elsewhere in any of the some 1900 nt of Cµ flanking region sequenced (13}. Elements of this Cµ homology region have been suggested previously to play a role in Ca switching (12, 13). A consensus sequence for Ca switching derived from the examples here would be PGTCPPGCTGGAATPPGYTGGGNTGPGCTG. Fig. 4. a) DNA sequence in the region of the germline Ca switch sites. Restriction sites are indicated. Recognition sequences are boxed. Also shown is a comparison of. the MClOl Gi derived sequence (11) with that of the germline Ca sequence. The MClOl sequence is bracketed by 8MC101 and SMClOl and differences from the corresponding germline ·sequence· are indicated above, ei~l)~r ~--~ase changes or gaps (-) Q.1'.J.:l~-~tc.). Precise localization of gaps in the tandem repeats GAGGA and GAGCT is not possible since the deletions are symmetrical. It is interesting that virtually all of the sequence differences are localized on the 3' half of the MClOl sequence, with the tandem repeats acting as a 53 ''b0rd¢r'' between highly conserved and mutated regions. } b) Schematic diagram of the location of seventeen 30 nucleotide repeated sequences adjacent to the germline Ca gene. Arrows indicate which repeats are used in the three rearranged genes described in this paper. c) DNA sequences of 1'1 repeats found interspersed in germ line 5' flanking Ca sequence. Boxes indicate nucleotide identities and - indicates a gap. Fig. 5. A comparison of the switch sequences from germline 5' flanking sequences for the Ca (see Fig. 3), Cy1 (11), and y2b (12, 13) genes. Boxes in the germline a sequence denote the conserved bases of the 30 nucleotide recognition sequence. Boxes in the germline yl and y2b sequences denote nucleotide identities to the germline Ca sequence. - indicates a gap. Fig. 6. Model for class-specific regulation of CH switching (see text). In this scheme the small boxes represent recognition (S) sequences which bind to switch proteins to mediate CH switching; the circle represents switching proteins and the large boxes represent coding regions. This model depicts a Ca gene order of Cµ, Cy. Ca (see text) but this is not a requirement. 54 lkb scale 1---i Tl5 R M R VDJ Ca ll!l! l!!lllll!l!!i!Ul~E-.:-:.:-:-~,B kb -:=:::r-:::::::::1-::::::::::-:::::::::-::::::::::~Jlfl---· M603 iJ-· R HSSM R CYI r: ..I ':&!l/111111!/lf/!tl/(#d'lk!!lll!!IJn. R MCIOI H scale 1 I kb 1 ·germ line Ca Fig. 1 55 . . cµ • • • • • • TGAGCTAGGCTGGGCTGAGCTGGAATGAGCTGGGTTGAGCTGAACTAGAATAAACITG~ 1::p·; a:.;:- -:;::r;;;::tt; - ·: c ·!: 44 TIS • • • • • • ACTTCCTGGTTGTTAAAGAATGGTATCAAATGGGTTGAGCTGAACTAGAATAAACTTGGC c:.. I -: ;-; ' : I ; ; ; :;:; : ; .; ~ • Complex Switching c1J AAGGGAACAAGGTTGAGAGCcCTAGTAAGCGAGGCTCTAAAAAGCATGGC • • • • • • GGACTAGGCTGGAATAGGTTGGGCTGGGCTGGTGCGAGCTGGGTTAGGCT . w j#rJ·*GZGS ft; i > : vs.; . - :+ ,ttF .. 4f. • • • • • • • AAGGGAACAAGGTTGAGAGCCCTAGCGTGAGTCTGAGCTGGGGTGAGCTGAGTGGGCTGAGTTGGGGTGA M603 . . . . . . . GCTGGGCTGAGTCTGGGGTGAGCTGAGCTGAGCTGGGGTGAGCTGAGCTGGGGTGAGCTGAGCTGAGCTG • • • • • • • AGCTGGGCTGAGCTGAGATGAGCTGGGGTGAGCTGAGCTGAGTTGAGCTGGGGTGAGCTGGAGCTGGGCT . . . . . . . AGCTGAGCTGGGGTGAGCTGAGCTGAGCTGGGGCTGAGCTGAGCTGAGCTGAGCTGGGCTGAGCTGGGCT • • • • • • GAGCTGGGCTGAGCTGGGCTGAGCTGGGCTGAGCTGGGCTGGTGCGAGCTGGGTTAGGCT • Successive Switching - . . . . . C AAAATGCGCTAAACTGAGGTGATTACTCTGAGGTAAGCAAAGCTGGGCTT ca • • • • • GGCTGAGAGCTGAGCTGAGCTGGAATGAGCTGGGATGAGCTGAGCTAGGC t.CIOI • • • • • AAAATGCGCTAAACTGAGGTGATTATGAGCTGGGATGAGCTGAGCTAGGC I •:- ;- -, ,;:;ea1 aa t-14 e 1J Fig. 2 Germ! tne cll Switch Sequences 5Ml41 ... 5T15 ... Tc c TfG G TIT GIT r AIA AG A A TIGG r Al- rlc AAAGIG Ac AG 5 MCIOt .... c T G Al G G i JG ALL T Al c T c T G AIG.Q. T~ A G 1£. A A A GI c T G G c 5 M603 ... AG AG cc c TA GIT AIA G c GA GIGlclTlc r A AIA A A Glc AT G G Germ! tne CCI Swttch Sequences en 5rt5 ... Cl) T G GfG c Tf G AjG c T G GA A TIG AIGfCIT G G GITIT GIAIG c T GIA A 5 M603 ... G G AjG c TIA GjG c T G GA A TIA GIGITIT G G Glclr GIGIG c T GIG T 5 t.el01 "' rlG A[§Jclr G G GIAl.!...2.JAIG c r GIA G r G AIG c T!G AIG c r G GA A Fig. 3 --~--~~~ ~ 0 c AAGGCTGAGCTG1'GCTGAGC TGGAATGAGCTGGGATGAGC"TGAGCTAGGC"TGGAATAGGC'TGGGCTGGGC"rGGTGTGT~GCTAGGTTA~GC1'~GGC1'G~· 1QQ GCTGGAATGAGCTGGGTTGAACTG'AGCAA'GGCTGGATGGAATAGGCTGGGCTGGGCTGtiGTGAGGCTGGGTTAGGCTGA°GCT~~GCTGA°GCTGAGCTGA · 200 GCTGAGCTGAGCTGAGCTGAGCTGAGCTGAGCTGAGCTGAGCTGGAAGGAGAGGAGAAGAGGA6R8GAGAGGAGAGGAGAGGAGAGGAGAGGAGAGGAGA 300 GGAGAGGAGAGGAGAGGAGAGGAGAGGAGAGGAGAGGAGAGGAGAGGAGAGGAGAGGAGAGGAGAGGAGAGGAGAGGAGAGGKGCf~GGCT°'G~AATAGGT 400 1 . 3 2 4 '--5 Mboll . s53 ' 6 r.1 o CG G G G G C T iGd'GCTGGGCTGGTGCGAGCTGGGTTAGGCT~AGCTGAGCTGGAATGAGCTAGGATGAGCTG',l\GCTAGGtGGAACAGGCTGGGTTTGGCiGGTGTGA~T ~ 7 A :-... - G :-. . G G T G • • ·~ SMC101 • C s+ •· • GAGTTAGGATGATCCGAGC1GAAATGAGC1GAGATAAGATlAGCTAGGCTGGAATAGGCTGGGCTGGGCTGbTGTGTGCTAGGTTGAGTCTAGCGGAAG Sau3A .,500 600 8 CTGGAATGAGCTGGCATGGGCTGN\GCTAGGCTGGAATAGGTTGGGCTGGGCfG}.;TGTTGAACTGGGTTAAGGCTGAACTGAACTGAATGAGCTAGGATG 9 10 . . 700 GTAAGACT~~CTAGGCTG~TAGGCTGGGTTTGGCTGttlGTTGAGCCAGGTTGATCCGAGCTGAATGAGCTGAGATAGATfAGCTAGGCTGGAATGGC 11 SauJA 12 800 TGGGCTGGGCf§GTGTGTGCTAGGTTGGTCTGAAGCTCGAAGCTGGAATGAGCTGGGATG~GGCTGGAATAGGTTGGGCTGG~GTGTG 13 14 900 ACAGCTGGGtrAGGCT~CTGAGCTGGAA~GAGTTGGAATAGGCTgGGCTGGCTGGATGTGAGCTGGGCTAGGCTGAGCTGAGCTAAAeTAGGCTGAAJ( 15 1000 TGGGCTGAGCAGAGCTGGACAAA~CTAGGCTACTGAGCACTGTCTGGCTA);GCTGTACTGGAATGAGCTGAGCTGAGCTGpGCTAAGCTGGGATGGACTA Rsa 16 1100 GGATAAACTA°AGCTGGGATGAGACAGGCTGACTGCAGGAGGAAGACTGGJO\GGGCTGGC1'GAGCTAGACTAGGCTGGGC'iGAGCTGGAAiGAG~· 1200 GAGCTljAACfAGATATAAAC"rTGGCTAGGCTACAATGGATTGAGCTGAGCTAGACTAGGGTGGAATGGGCTGAACAAGGCTGAGCTTACCTAGACGCGCC 1300 GGTGGCAGACtTAGATAGAG'rTGCACTGAGGTAGGTTAGAtAGGGTTGTCTGTAGCTTGAGCTTGACCTiAGGNGCTGTGCTTGTCTGTAGCTG Msp 1400 5115 17 Fig. 4a l C1I -:i 58 0 ts . .... bO ~ .c .:.:. v• - - 0 u~ cJf 59 Ca. consensus sequence (Fig. 3) P GCT 5MC101 5T15 5M603 . } + + + P P G C T G GA AT P P GYT G G GN T GP GC T G MC101 Sl A G C T - - G A G C T G G A A T G A G C G G GA - T G A GC T G 52 A G C T - - A G G C T G G A A T A G G C Gr S C - T G G GC T G 53 G G C T - - G A G C T G G A A T G A G C G G G T - T G A A C T G S4 G G C T - - - G G~ T G G A A T A G G C G G G C - T G. G G C T G 0 55 A G C T - - G A G C T G G A A G G A G A G G G A A G A G G~ G M603 56 . A G C T - - A·s G C T G G A A T A G G T T G G G C - T G G G C T G 57 A ~ C T - - G A G C T G G A A T G A G C T 0G G A - T G A G C T G 58 A G C T - - A G G C T G G A A T A G G C T G G G C - T G G G C T G 59 A G C G G - A A G C T G G A A T G A G C T G G C A - T G G G C T G 510 A G C T - - A G G C T G G A AT A G G T T G G G C - T G G G C T G 511 AG CT - - A G G CT G GA AT AG G CT G G GT TT G - G CT G 512 A G C T - - A G G C T G G A A T - G G C T G G G C - T G G G C T G 513 A G C T C G A A G C T G G A AT GA G C T G G G A - T G G G C T G 514 A G C T - - A G G C T G G A A T A G G T T G G G C - T G G G C T G 515 A G C T - - G A G C T G G A AT G A G T T G GA A - T A G G C T G 516 G G C T - - G T A C T G G A A T G A G C T G A G C - T G A G C T G T15 517 G G C T - - G A G C T G G A A T G A G C T G G G T - T G A G C T G P = purine Y = pyrimidine N = any nucleotide Fig. 4c 5K.:IOI 5T15 ... ~ Germ It ne a Germ I I ne ·YI 5M603 ... fFGGG1 N[[]I P IG c r GI 5K.:,OI I ... c A G G G A IG c ,.!., G G A IG c T G A -1 Tg GIT A T A A A A A PIG c ti P P ~IG""""'c~r'"""'G-G~A-Xfl,........r P P ml Y 5 Ml41 ... Germl tne Y2b l!J C C~ A(!] T G A G C A T G G G G G C le I A IG G A IG A G T T G 0) 0 Fig. 5 VDJ ----.o @ [}{)-{] ~~~ Cp. I I // ® Cy 111 ~~~ I I // D-lHl ~~~ Ca I I....- • heterodimer formation Cy C') ...... -Ca VDJ l VDJ _ _,.CJ 0-H I Fig. 6 site-specific recombination I ~ Jt--- 62 Two Types 9f DNA Rearrangements in Irnrnunoglobulin Genes L. HOOD, M. DAVIS, P. EARLY, K. CALAME, S. KIM, S. CREWS, AND H. HUANG Division of Biology, California Institute of Technology Pasadena, California 91125 Cold Spring Harbor Syrnp. Quant. Biol. (1980) In Press 63 Running Head: Send proofs to: L. Hood Division of Biology California Institute of Technology Pasadena, California 91125 Phone: (213) 795-6811, Ext. 1951 64 · Antibody g~nes~ along with most other eukaryotic genes, are examples of "split genes.'' One .unique feature of the antibody gene system is that it can rearrange indi~idual gene segments by two distinct mechanisms during the 'diffe{'.entiation of antibody-producing or B cells. These DNA rearrangements amplify the information encoded in the antibody gene families. The antibody or immunoglobulin molecule is composed of two distinct polypeptides, light (L) and heavy (H) chains, both of which are divided into variable (V) regions responsible for antigen binding and constant (C) regions responsible for effector functions such as complement fixation. There are three unlinked families of antibody genes-two code for lambda (),) and kappa (K) light chains, and the third codes for heavy chains. Light chains are encoded by three distinct gene segments-VL (variable), JL (joining), and CL (constant), whereas heavy chains may be encoded by four-VH, D (diversity), JH and CH (Fig. 1) (Bernard et al. 1978; Early et al. 1980). During the differentiation of antibody-producing cells, two types of DNA rearrangements occur. First, the VL and JL or the Vff, D and JH gene segments are joined to generate a contiguous coding sequence for the entire light or heavy chain variable region. V(D)J joining. We term this type of DNA rearrangement A second type of DNA rearrangement, CH or class switching, occurs in the heavy chain gene family (Davis et al. 1980; Kataoka et al. 1980). In the mouse this gene family contains at least eight closely linked heavy chain constant region genes (Cµ, C\S, Cy1. Cy3• Cy2b. Cy2a• Ca, and Ce;). During the development of a B cell, IgM (Cµ) is initially expressed, while later the B cell or its progeny may switch to expression of any of the other classes or subclasses of immunoglobulins (Fig. 2) {Cooper et al. 1976; Pernis et al. 1976). In molecular terms these distinct stages of B-cell differentiation are mediated by the two types of DNA rearrangements discussed above. The 65 V(D)J joining generat~s the functional VH gene 5' to the Cµ gene, leading to the synthesis of µ chains and IgM molecules. Later, the B cell or its clonal progeny can undergo a class switch ON A rearrangement in which another heavy chain constant region gene replaces the Cµ gene adjacent to the same functional VH gene. For example, a transcript can then be made which contains the VH and Ca genes, leading to the synthesis of a chains and IgA molecules. In the .functional gene, each domain of the immunoglobulin chain (the variable region and one or more constant region domains) is encoded by a separate exon, so that RNA splicing is required to generate the final V+C immunoglobulin mRNA from transcripts of the rearranged gene (Bernard et al. 1978; Early et al. 1979; Sakano et al. 1979b). Variable Region DNA Rearrangement Studies of immunogloublin gene organization have provided insights into the process of V(D)J joining and the mechanisms by which the enormous repertoire of specificities in the vertebrate immune system is generated. In the light chain gene families, A and K, the variable regions are er coded by two separate germline gene segments. The VL gene segments generally encode residues 1-96 from the N terminus, and the JL gene segments, residues 97-108, the remainder of the variable region (Bernard et al. 1978; Max et al. 1979; Sakano e-t al. 1979b). The heavy chain gene organization is somewhat more complex in that a third gene segment, D, appears to encode from one to fifteen residues between those encoded by the VH and JH gene segments (Early et al. 1980). For example, one heavy chain gene, M603, has tt•e germline VH and JH gene segments encoding variable region amino a(:ids 1-101 and 107-122, respectively. Thus, amino acids 102-106 must be encoded by a third gene segment, termed diversity (D), because this is the most 66 diverse segrnent 'Of heavy chain variable regions (Early et al. 1980; Schilling ·et al. 1980). The DNA sequence coding for the D segment of the M603 a gene is not found near the 3' end of the germline VH gene segment or the 5' end of the germline JH gene segment. Although a germline D gene segment has not yet been identified, it clearly must be encoded by DNA sequences sE-parate from VH and JH. In light chai~s, the JK and J A gene segments are approximately 2.4 kb and 1.2 kb 5' to the CK and Cx genes, respectively (Bernard et al. 1978; Max et al. 1979; Sakano et al. 1979b). Accordingly, VJ joining brings a VL segment into close association with its corresponding CL gene. In heavy chains, the JH gene segments are approximately 8 kb 5' to the Cµ gene, the first in the series of CH genes to be expressed by a B cell (Early et al. 1980). V(D)J joining creates a functional VH gene associated with the Cµ gene. Germline V (and D) gene segments are probably 5' to the J gene segments in each gene family, although the distances separating these gene segments are unknown. DNA Recognition Sequences for Joining Proteins May Mediate V(D)J Joining Immunoglobulin genes can be examined in the differentiated and undifferentiated state by studying their organization in myeloma (differentiated) and sperm or embryo (undifferentiated or germline) DNAs. Analyses of DNA sequences near potential sites of DNA joining flanking germline V and J gene segments has revealed blocks of highly conserved nucleotides in each of the antibody gene families (Bernard et al. 1978; Max et al. 1979; Sakano et al. 1979b;Early et al. 1980). Moreover, these blocks of nucleotides are the same in all the immunoglobulin gene families of the mouse, and a recent report indicates that the same conserved nucleotides are adjacent to human V 67 gene segments (T. Rabbitts, personal communication). Thus these nucleotides have been conserved for more than 500 million years, since the immunoglobulin light and heavy chain gene families diverged from one another. This strikir:g conservation argues compellingly that these blocks of nucleotides play a fundamental role in V(D)J joining. The structures of these blocks of nucleotides for each of the three immunoglobulin fal]lilies are presented in Table 1. These nucleotides are found at the 3' side of the V gene segments and at the 5' side of the J gene segments-precisely at the locations expected for sequences which mediate V(D)J joining. We have noted that these conserved nucleotides always occur as two blocks, of 7 and 10 nucleotides, separated by a nonconserved spacer sequence of either 11 or 22 nucleotides (Table 1) (Early et al. 1980). The spacer lengths correspond closely to separations of these blocks of recognition sequences by 1 or 2 turns of the DNA helix (1-turn or 2-turn spacers). The conserved nucleotides adjacent to V gene segments are nearly inverted complements of the nucleotides adjacent to J gene segments. The observation of these conserved nucleotides in the light chain gene families initially led to the suggestion that V-J joining occurs by virtue of a stem and loop structure that could be formed by the self complementarity of the inverted repeat sequences at the 5' end of the J gene segment and the 3' end of the V gene segment (Max _et al. 1979; Sakano et al. 1979b). Presumably a site-specific recombination event across the base of the stem structure could join the V and J gene segments. However, the existence of separated blocks of conserved nucleotides and the "1- and 2-turn" spacer relationships led us to consider a more attractive alternative mechanism of V(D)J joining (Early et al. 1980). In the K gene family the V gene segments have blocks of conserved nucleotides that are always separated by a 1-turn spacer (Table 1). In contrast, 68 the Jt< gene segments have their. blocks of conserved nucleotides separated by 2-turn spacers. The v>.:1 gene segment has a 2-turn spacer, and the Ju gene segment has a 1;,.ttirn spacer. '. have 2-turn spacers. Heavy chain VH and JH gene segments both This pattern of sequence organization has led us to postulate that the conserved nucleotides are recognition sites for two types of DNA joining proteins. sit~, One joining protein interacts with the 1-turn recognition whereas the second joining protein recognizes the 2-turn site. We suggest that after DNA binding a 1-turn joining protein may dimerize with a 2-turn joining protein, but not with another 1-turn protein. This dimerization apposes V and J gene segments so that a site-specific recombination event can splice together these gene segments. features. This model has several attractive First, it will prevent nonproductive V-V or J-J joining. this model is directly applicable to all antibody gene families. Second, When the model is applied to the heavy chain gene family, it makes an interesting prediction about the structure of germline D gene segments. Because the Va and JH gene segments both have 2-turn spacers, we predict that the D gene segments will have recognition sequences with 1-turn spacers on both their 5' and 3' sides. Two types of recognition sites have evolved in immunoglobulin gene segments, each of which contains the same two blocks of conserved nucleotides apparently involved in protein-DNA interactions. The only difference in the sites is the distance (1 or 2 helical turns) between the blocks of conserved nucleotides. This pattern suggests that the evoluti.on of the proteins involved in immunoglobulin variable region gene rearrangement occurred as follows. Separate protein subunits or domains probably once existed which could bind the individual blocks of nucleotides, CACTGTG and GGTTTTTGTA. These proteins may have functioned in DNA rearrangement, or could have served 69 other purposes requi~ing protein recognition of the specific DNA sequence• .In order to increase the specificity of DNA binding (an important consideration in a large vertebrate genome), compound proteins could have evolved which combined the specificities for the two original DNA binding sites. The spatial orientation of the protein subunits determined the length of any spacer DNA required between the blocks of CACTGTG and GGTTTTTGTA. The different spatial orientations in proteins requiring 11 and 22 nucleotide spacers may simply be the result of different polymers of the same subunits. In this fashion, two (or more) highly specific DNA-binding proteins could have evolved by combinations or existing proteins which specifically bound shorter sequences of DNA. This is likely to have been a much more rapid process than the evolution of entirely new binding sites in proteins specific for longer sequences of DNA. Support for this view of the evolution (and antiquity) of the DNAbinding proteins involved in immunoglobulin variable region gene rearrangement comes from a gene for the major flagellar protein in Salmonella. Silverman et al. (1980) have identified a 14 nucleotide sequence present in opposite orientations on the ends of the invertible DNA segment controlling which of two types of flagella protein is synthesized by Salmonella. The blocks of 10 conserved nucleotides in the immunoglobulin recognition sequences are almost completely identical to part of the Salmonella-inverted sequence. GGTTTTTGTA is the consensus sequence in immunoglobulin genes, whereas GGTTTTTGAT is the sequence of Salmonella. The difference between these two sequences is no greater than between some of the immunoglobulin sequences shown in Table 1. Accordingly, this sequence may be a very ancient recognition sequence for a protein involved in DNA rearrangements. The system of DNA inversion in Salmonella also is apparently closely related or identical to that of the 70 G-loqp }n mu phage (Katsukake and Ino 1980). Eukaryotes may have retained k. the recognition sequence and something of the function of this protein, which seems then to have combined with another DNA-binding protein in evolving the ,mechanism for variable region gene rearrangements in the immune system. Several Distinct Mechanisms Generate Antibody Diversity The controversy between germline and somatic theories of antibody diversity (see CSHSQB 1976) has largely disappeared because both kinds of mechanisms contribute significantly to antibody diversity. Germline genes. The DNA rearrangements necessary to generate complete variable region genes are a major source of diversity in antibody molecules. First, this process allows many alternative variable region germline gene segments to exist, any of which can be expressed with the constant region gene(s} in its family. There are perhaps 200 to 500 light chain variable region gene segments (Seidman et al. 1978), and presumably a comparable number for heavy chains. Both the kappa light chain and heavy chain gene families of mouse have four J .gene segments (Max et al. 1979; Sakano et al. 1979b; Sakano et al. 1980; D. Kemp, personal communication). The number of heavy chain D gene segments is unknown, but judging from the diversity of heavy_ chain protein sequences (Schilling et al. 1980) this number may be on the order of 10. Combinatorial joining. Protein sequence studies suggest that for any particular antibody gene family, any V gene segment may be joined to any J gene segment (Weigert et al. 1980). Presumably the same will be true of D gene segments with regard to their joining to VH and JH gene segments. Accordingly, the combinatorial joining of, for example, 200 VH, 10 D, and 4 71 Ja ge11e segments can generate 8000 distinct Va regions. In contrast, 200 VK and 4 ·JK gene segments will generate 800 VK regions. Thus the existence of a third gene segment, D, in the heavy chains considerably increases the potential for Va gene diversification. The joining of VL and JL as well as Vff. D, ano Ja gene segments occurs in the third hypervariable region-an important part of antigen-binding sites. Thus diversification in the third hyper.variable region by combinatorial joining of gene segments can lead to a diversity of antigen-binding sites. Somatic variations. Although the blocks of recognition sequences adjacent to variable region gene segments appear to guide the enzymes responsible for DNA rearrangement, the exact points at which the DNA joining occurs are quite variable. This gives rise to variable region junctional diversity, as first noted at the V-J junction of light chains (Max et al. 1979; Sakano et al. 1979b). Some examples of heavy chain junctional diversity at both V-D and D-J junctions are shown in Figure 4. The length of the D and JH gene segments in the rearranged variable region can vary over a few codons in the third hyper.variable region, and new hybrid codons can be created at the junction (Weigert et al. 1980).. An additional opportunity for variability unique to heavy chains arises from the presence of DNA junctions on both sides of the D gene segment. In or~er for an immunoglobulin chain to be synthesized, the Va and JH (and consequently Ca) gene segments must be translated in the same reading frame. There is no necessary constraint on the reading frame of the D gene segment, so long as it does not contain a termination codon. As long as any changes in the reading frame at the Va-D junction are compensated by changes at the D-JH junction, it should be possible for a given germline D gene segment to occur in any of three reading frames in rearranged variable ·~ 72 region "genes. A; probable example of the use of alternative reading frames for a D segment is shown in Figure 5. Comparing the protein sequence of a levan-binding (A4) VH region with the DNA sequences of the germline Ja3 gene ·segment, and an A4-like germline Vff gene segment indicates that the A4 protein probably contains a D segment (S. Crews, unpublished results), contrary to previous conclusions based only on the protein sequence (Schilling et al. 1980). The .nucleotide sequence of the A4 D segment, as deduced from the protein sequence, closely resembles or may be identical to the 8107 D gene segment shown in Figure 4. This nucleotide sequence is translated in different reading frames in the A4 and 8107 Vff regions. Generation of antibody diversity. As mentioned above, by the combinatorial joining of V, D, and J gene segments, mice may generate approximately 104 Vtt regions and 103 VL regions. Independent association of these light and heavy chains in antibody molecules produced by different B-cell clones will generate approximately 107 different antibody molecules. Accordingly, combinatorial joining and association will generate a vast repertoire of antibody molecules before employing any of the mechanisms of somatic mutation. Clearly higher vertebrates can easily generate 108 or more antibody molecules by combining all of the mechanisms discussed-multiple germline gene segments, combinatorial joining, somatic mutation, and independent association of light and heavy chains.Heavy Chain Switching Involves a Second Type of DNA Rearrangement We first observed the results of heavy chain switching at the DNA level in a functional a heavy chain gene from the M603 myeloma tumor (Figure 6) (Davis et al. 1980a). The M603 Vff exon is separated from the 73 Co. exon by a 6.8 kb,intervening sequence. Comparison of the M603 a gene with germline VH, Ca, and JH plus Cµ genes (Figure 6) shows that the M603 gene is composed qf at least three d.istinct segments of germline DNA. These are: ·a VH gene segment and its 5' flanking sequences, a JH gene segment and associated sequences originally adjacent to the Cµ gene, and the Ca gene with its 5' and 3' flanking sequences. DNA sequence analysis also shows that the M603 a gene contains a D gene segment between the VH and JH gene segments (Early et al. 1980). Thus the structure of the M603 a gene demonstrates that at least two separate DNA rearrangements must have occurred. The first type of DNA rearrangement, V(D)J joining, generated a functional Vtt gene associated with the Cµ gene. The second type of DNA rearrangement, Ctt switching, took place in intervening DNA sequences between the Jtt gene segment and the Cµ gene, and resulted in the replacement of the Cµ gene with the Ca gene. The points at which the flanking Cµ and Ca sequences join in the rearranged gene and the corresponding breakpoints on the germline DNAs are termed the switch (S) sites after Kataoka et al. (1980). Subsequently, other laboratories have obtained evidence for similar Cµ to Ctt switches in rearranged yl (Kataoka et al. 1980) and y2b (Takahashi et al. 1980; Sakano et al. 1980) genes. Three Examples of IgM + IgA Switching In order to investigate the molecular mechanisms underlying CH switching, we sequenced the switch site of M603 and that of an additional IgA-producing myeloma tumor, T15. Homologies between clones containing the rearranged a genes of T15 and M603 are depicted in Figure 6 and compared with their germline Cµ and Ca counterparts obtained from a genomic library of mouse sperm DNA (Davis et al. 1980a). These homologies were established by detailed restriction 74 enzyme analysis . (dat,11 not shown). Both rearranged a genes exhibit the tripartite structure of Vff, JH with Cµ flanking sequences and Ca gene segments. The size of the interv~ning sequence between the Va and Ca coding regions is substantially different in these two cases-5.4 vs. 6.8 kilobases (Fig. 6). Since each of the two Vff gene segments are joined to the same JH gene segment (Johnson et al. 1980, in preparation), the variation in size in the intervening sequences between the V and Ca coding regions may be the result of using different sites for Cff switching in each of these rearranged Ca genes. In addition, we found that a rearranged yl gene, MClOl, whose isolation and switch site sequence was reported by Honjo and colleagues (Kataoka et al. 1980) contains a 500 nucleotide region of Ca flanking sequence between Cµ and Cyl derived sequences (Fig. 6). The evidence for this statement is that this 500 nucleotide sequence has been localized solely to a region 5' to the Ca gene by Southern blotting analyses (Southern 19_75} and restriction mapping using fragments containing all or part of this region as probes (Kataoka et al. 1980; M. Davis, data not presented). Therefore, this fragment is apparently represented just once in the genome and, accordingly, must have been derived from flanking sequences 5' to the Ca gene. Furthermore, DNA sequence analyses of the corresponding germline Ca sequence show virtual identity (>95%) with the sequence found between the Cµ and Cy1 flanking sequences in the MC101 yl gene (Davis et al. 1980b}. Thus we feel the MC101 yl gene is composed of several distinct germline sequences: a Vff gene, flanking sequence for the Cµ gene, flanking sequence for the Ca gene and the yl gene with its flanking sequences. The arrows in Figure 6 indicate the regions analyzed by DNA sequence analysis in our laboratory. The DNA sequences of the MC101 S region (Kataoka et al. 1980) and the rearranged M603 and Tl5 a genes and their germline C 11 75 and Ccf counterparts are shown in Figure '1. k Examination of the rearranged switch sequences indicates that all three examples juxtapose, Cµ and Ca derived sequences. sequence seems distinct from the others. However, each switch The arrangement of sequences in these genes suggests that at least three distinct types of switching may occur. We denote these categories "simple," "complex," and "successive." In the simple category, T15, the Cµ flanking sequence joins directly to the Ca flanking sequence. Similarly, a y2b gene (Ml41) has been found to join Cµ flanking sequence to that of Cy2b (Takahashi et al. 1980; Sakano et al. 1980). In the complex category, M603, a short sequence of 287 base pairs, is interposed between the Cµ and Ca flanking sequences. This sequence appears to derive from a region 3' to SM603 on the Cµ gene (Fig. 6). Probes containing this sequence hybridize strongly to restriction fragments containing or adjacent to the Cµ gene in Southern blotting analyses (M. Davis, data not shown). The region of hybridization corresponds to the 1.5-2.5 kb region 5' to the Cµ gene (Davis et al. 1980a) which deletes spontaneously upon cloning and hence is not present in our Cµ containing clones. Thus the complex category is explained by two distinct deletions-CH switch and a deletion within the Cµ flanking sequence. This deletion of Cµ flanking sequence (at SMso3) does not appear to be a random event in that another a producing tumor_ line, Y5236, switches at exactly the same point adjacent to the Cµ gene. However, the complex category reflects DNA deletions seen to date only in myeloma cells and, accordingly, may or may not be biologically significant. In the third category, "successive" switching, the MC101 Yl gene contains Cµ, Ca, and Cy1 flanking sequences between Vtt and Cy1 gene segments. Therefore, it appears to have switched twice, once from Cµ to Ca and subsequently from Ca to Cy1. 76 ·The DNA sequ~ce data in Figure 7 show that the three rearranged genes employ three germline Si switch sites up to 300 base pairs apart and three germline Ca ,switch sites up to 1350 base pairs apart. The locations of these· germline switch sites exist for Ca switching in sequences 5' to both the germlihe Cµ and Ca genes. DNA Sequences Mediating Heavy Chain Switching The DNA sequences involved in V(D)J joining are quite distinct from those implicated in Ctt switching. The inverted repeat CAC~GTG occurs at the 3' end of antibody V gene segments and at the 5' end of the J gene segments (Sakano et al. 1979b;Early et al. 1980; Max et al. 1979). This inverted repeat is believed to be a recognition sequence that mediates the juxtaposition of the V, D, and J gene segments to allow subsequent joining by site-specific recombination (for a proposed mechanism, see Early et al. 1980). This inverted repeat is not found in the flanking regions surrounding any of the Ca switch sites. This sequence also is missing from the switch sites for a yl (Kataoka et al. 1980) and a y2b gene (Takahashi et al. 1980; Sakano et al. 1980). Therefore, Ctt swit~hing and V-J joining employ distinct mechanisms for DNA rearrangement. In an effort to determine what sequences are important in Ctt switching, the se_quences of the three Cµ sites and the three Ca sites were compared (Fig. 8). The germline Cµ MClOl and T15 switch sites share significant homology (15/25 nucleotides), although 16 nucleotides separate the actual switch points. We believe that these homologies are significant and may represent general sequence requirements for Ctt switching adjacent to the Cµ gene. Neither the MC101 nor the T15 Cµ sites share any homology with that of M603. The switch site of a y2b producer M141 (Takahashi et al. 1980; Sakano et al. 1980), also depicted in Figure 8, is nine nucleotides away 77 from that of T15 and may indicate that both y2b and a switching can utilize } the same. recognition sequence adjacent to eµ. The sequences around each Ca switch site are even more highly conserved. Each germline Sa site occurs within a block of 30 conserved nucleotides (Fig. 8). Twenty-two bases are identical and seven of the remaining eight nucleotides are conserved with regard to type of base (i.e., purine-purine or pyrimidine-pyrimidine substitutions). wi~hin The three points of recombination differ each of these conserved sequences. Since the first three rearranged genes examined switched at distinct Ca sites, we reasoned that there must be additional Sa sites. We determined the DNA sequence of some 1400 nucleotides in the region of these germline Sa sites (Davis et al. 1980b) and found seventeen 30 nucleotide repeat sequences (Fig. 9a). These sequences are extremely homologous to each other (boxed regions) (Fig. 9b). Since these repeated sequences represent 510/1400 nucleotides in the region analyzed, the Ctt switching into three of these sites does not appear to represent random DNA rearrangement. We suggest that most of these repeats are potential germline Sa sites. Multiple sites for Ctt switching may be advantageous since they would improve the ~fficiency of switching enzymes, allowing for a relatively low level of expression of what could be potentially deleterious gene products (switching enzymes). This strategy would also apply to the enzymes mediating V-J joining where multiple V and J gene segments provide large numbers of recombination sites. Evidence for Class-specific Regulation of Ce Switching The data presented here, taken together with sequence data on y2b (Takahashi et al. 1980; Sakano et al. 1980) and yl (Kataoka et al. 1980) switch sites from the literature which complement these experiments, suggest a mechanism by which class-specific regulation might be accomplished. 78 The germline sequenc;es for CH switching appear to be class specific with regard to Cyl• Ca and. Cy2b switch sites. The germline Cy1. Ca, and Cy2b "switch" sequences, are compared in :Figure 10. The yl sequence is identical to the prototype Sa sequence in 10 out of 22 bases and the y2b sequence is identical for 7 of 22 nucleotides (only 3 of which are contiguous). Thus, various germ line Sa sequences are far more similar to one another than to the germline Sy1 °.r Sy2b sequences (Figs. 8 and 9b). One explanation for these sequence differences is that CH switching is mediated by class-specific recognition sequences. A Mechanism for Cu Switching Since the germline Sa and Sµ sequences are not homologous (Fig. 8), homologous recombination cannot account for their joining. We believe the joining may be mediated by a number of distinct types of "switching" proteins (Fig. 11). For example, one switching protein (Pa) may bind the germline &x sequence and a second (Pµ) may bind one of the germline Sµ sequences. These proteins may then interact to form a heterodimer which juxtaposes the V(D)J gene with the Ca gene s~gment (Fig. 11). The multiple germline Sa sequences would increase the probability that Ca switching could occur once the appropriate joining protein was expressed. Because the germline Sa, Sy1, and Sy2b sequences seem distinct, different joining proteins could bind these sequences. Accordingly, the developmental regulation of the expression of these proteins would lead to class-specific regulation of CH switching. The Implications of successive Cu Switching in MClOl The evidence for more "successive switching in MC101 indicates that two or more CH switches can occur in a particular B-cell line. In the simple deletional model for class switching proposed by Honjo and Kataoka (1978), CH switching progresses in a linear fashion, deleting intervening CH genes at 79 each stage. As .menponed, the experiments supporting this model (Honjo and Kataoka 1°978; Cory anq Adams 1980; Coleclough et al. 1980; Rabbitts et al. 1980; Yaoita and H;onjo 1980; Cory et al. 1980) indicate a gene order of c;.iCy3.;.Cyi-Cy2b-Cy2a-Ca. Paradoxically, however, MC101 appears to have switched from Cµ to Ca and then from Ca to Cyl• contrary to the simple deletional model for class switching proposed by Honjo and Kataoka (1978). Several explanations seem plausible: · 1) The CH gene order is Cµ-Ca-Cy. This seems difficult to support because of the large number of myeloma tumors which express the Cyl gene, and still contain Ca genes (Honjo and Kataoka 1978; Cory and Adams 1980; Coleclough et al. 1980; Rabbitts et al 1980; Yaoita and Honjo 1980; Cory et al. 1980). In contrast, those myeloma tumors which express the Ca gene generally appear to have deleted their Cy1 genes (Cory and Adams 1980; Coleclough et al. 1980; Rabbitts et al. 1980; Yaoita and Honjo 1980; Cory et al. 1980). However, until these genes are ordered in the germline DNA, this explanation remains a formal possibility. 2) Interchromosomal recombination. In this scheme, Cµ ..... Ca rearrange- ment on one chromosome could' be followed by recombination with a Cy1 gene on another chromosome to produce the MC101 Yl mosaic gene (VH-CµCa-Cy1). Moreover, the reciprocally rearranged chromosome should have a rearranged Ca gene. 3) Episomal deletion. If the deleted DNA between the VH gene and the Ca gene formed a circular intermediate (an episome) that was at least transiently stable, this episome could then re-integrate into the chromosome, replacing the Ca gene with the Cyt gene. This model is easily testable since it predicts the presence of a Cµ gene, a Cy3 gene, etc. in the MC101 genome and, in addition, the Cµ and Ca genes should be rearranged. 80 k The .rearranged MClOl yl gene, accordingly, raises two general possibilities with regard to CH switching and normal B-cell differentiation. i) The successive (and co~plex) types of Cff switching observed here may arise from one or more aberrant chromosomal rearrangements that are unique to myeloma cell lines and that will not generally be seen in normal B cells. The numerous cell divisions that occur between myeloma tumor production and our analyses of the corresponding DNAs, as well as the aneuploid nature of myeloma cells, make this a serious possibility. ii) Normal B cells may employ multiple Cff switching mechanisms-some perhaps different from any cited above. The developmental regulation of Cff switching may operate at several different levels. i) The nature of the sequences mediating V(D)J joining and Ctt switching implies that these phenomena are regulated independently. ii) The existence of distinct switch sequences for a, yl, and y2b genes implies that the expression of these classes may be developmentally regulated at the level of DNA rearrangement, depending, for example, on which specific switching protein is expressed. iii) The ~vidence for a Cµ • Ca• Cyi "successive" switch (MClOl) presents difficulties for simple deletional models of Ce switching. Abortive Rearrangement Suggests that CH Switching Enzymes May Generally Operate on all Heavy Chain Chromosomes Myeloma cells are generally subtetraploid and may have between one and four or more heavy chain chromosomes. Often a particular expressed heavy chain gene from an individual myeloma may be rearranged on each of the heavy chain chromosomes. For example, the lgA-producing M603 myeloma tumor has three distinct a genes and each has been rearranged from its germline configuration (Davis et al. 1979). We have discussed the 81 ttctively expresserl M~03 a gene above. The other two a genes contain no M603 VH genes and are presumably not functional. said to be "abortiv~ly rearranged." These M603 a genes are Similar observations of abortive rearrange- ment .of light chain genes have been made by others (Steinmetz and Zachau 1980). One puzzling. aspect of Southern blotting data on various myeloma tumors is that in many cases all of the CH genes 5' to the expressed CH gene have been deleted (Cory et al. 1980). These DNA deletions are not random rearrangements in that one would not expect them to terminate at or near the expressed CH gene. We believe that these data independently support the existence of class-specific switching enzymes which, once induced, delete the CH genes 5' to the expressed CH gene on one or both heavyI chain chromosomes. One prediction of this interpretation is that abortive rearrangement of CH genes will involve the same DNA switching sequences as CH switching. In order to test this prediction, we sequenced one abortively rearranged M603 Ca clone. This clone, ChM603a30, and its homology to a germline Ca gene are illustrated in Figure 12a. There is no VH M603 gene segment present in the a3o clone, hence it appears to be an "abortive rearrangement." DNA sequence analysis· indicates that the a30 clone diverges from the germline a flanking sequence at one of the specific Sa switch sequences (Fig. 12b). Indeed, the a30 clone diverges two nucleotides 3' to the switch site used by the T15 gene (Fig. 8) and well within the 30 nucleotide homology region characteristic of Ca switching. The fact that abortive rearrangement of the Ca gene in M603 occurs at an Sa switch site strongly suggests that abortive rearrangements do not occur randomly during B-cell development, but are in fact mediated by a mechanism similar or identical to that responsible for CH switching. Furthermore, this 82 result adds weight to;. the suggestion that CH switching is mediated by class.. specific proteins which. recognize sequences 5' to germline CH gene segments. Acknowledgments This work was supported by grants from the NIH, NSF, and USPHS. M.D. and S.K. are supported by an NIH training grant. All experiments involving recombinant organisms were conducted in accordance with the NIH Guidelines on recombinant DNA. Portions of this review were adapted from an article in Science (Davis et al. 1980b). 83 REFERENCES Benton, W. D. and R. W. Davis. 1977. Screening >..gt recombinant clones by hybridization to single plaques in situ. Bernard, 0. and N. Gough. 1980. Science 196: 180. Nucleotide sequence of immunoglobulin heavy chain joining segments between translocated VH and µ constant region genes. Be~nard, Proc. Nat. Acad. Sci. 77: 3630. 0., N. Hozumi and S. Tonegawa. 1978. Sequences of mouse immuno- globulin light chain genes before and after somatic changes. Cell 15: 1133. Coleclough, C., C. Cooper and R. P. Perry. 1980. Rearrangement of immuno- globulin heavy chain genes during B-lymphocyte development as revealed by studies of mouse plasma cells. Proc. Nat. Acad. Sci. 77: 1422. Cooper, M. D., J. F. Kearney, P. M. Lydyard, C. E. Grossi and A. R. Lawton. 1976. Studies of generation of B-cell diversity in mouse, man, and chicken. Cold Spring Harbor Symp. Quant. Biol. 41: 139. Cory, S. and J. M. Adams. 1980. Deletions are associated with somatic rearrangement of immunoglobulin heavy chain genes. Cory, S., J. Jackson and J. M. Adams. 1980. Cell 19: 37. Deletions within the constant region locus can account for switches in immunoglobulin heavy chain expression. Nature (in press). Davis, M., P. Early, K. Calame and L. Hood. 1979. The organization and rearrangement of heavy chain immunoglobulin genes. Gene Regulation (ed. R. Axel et al.). In Eukaryotic ICN-UCLA Symp., 393, Academic Press, New York. Davis, M. M., K. Calame, P. W. Early, D. L. Livant, R. Joho, I. L. Weissman and L. Hood. 1980a. An immunoglobulin heavy chain g-ene is formed bv at least two recombinational events. Nature 283: 733. 84 '} Davis, M •. N1., S. K. Kim and L• Hood. 1980b. Sequences mediating heavy chain switching in alpha immunoglobulin genes. Science (in press). Early, P., H. Huang, M. Davis, K. Calame and L. Hood. 1980. An immunoglobulin heavy chain variable region gene is generated from three segments of DNA: VH' D and Jff Cell 19: 981. Early, P. W., M. M. Davis, D. B. Kaback, N. Davidson and L. Hood. 1979. Immunoglobulin heavy chain gene organization in mice: analysis of a myeloma genomic clone containing variable and ex constant regions. Proc. Nat. Acad. Sci. 76: 857. Honjo, T. and T. Kataoka. 1978. Organization of immunoglobulin heavy chain genes and allelic deletion model. £Toe. Nat. Acad. Sci. 75: 2140. Johnson, N., S. Rudikoff, P. Barstad and L. Hood. 1980. Sequences of phosphorylcholine binding variable regions (in preparation). Kabat, E. A., T. T. Wu and H. Bilofsky. 1979. Sequences of immunoglobulin chains. NIH publication no. 80-2008. Kataoka, T., T. Kawakami, N. Takahashi and T. Honjo. 1980. Rearrangement of immunoglobulin yl-chai~ gene and mechanism for heavy-chain class switch. Proc. Nat. Acad. Sci. 77: 919. Katsukake, K. and T. Ino. 1980. A trans-acting factor mediates inversion of a specific DNA segment in flagellar phase variation of Salmonella. Nature 284: 479. Maniatis, T., R. C. Hardison, E. Lacy, J. Lauer, c. O'Connell, D. Quon, G. K. Sim and A. Efstratiadis. 1978. The isolation of structural genes from libraries of eucaryotic DNA. Cell 15: 687. 85 Max, E. E., J. G. Sediman and P. Leder. 1979. Sequences of five potential recombination sites encoded close to an immunoglobulin K constant region gene. Proc. Nat. Acad. Sci. 76: 3450. Maxam, A. and W. Gilbert. 1980. Sequencing end-labeled DNA with base-specific chemical cleavages. In Methods in Enzymology, Nucleic Acids, Part I (Ed. L. Grossman and K. Moldave), vol. 65, p. 499. Academic Press, New York. Pernis, B., L. Forni and A. L. Luzzati. 1976. Synthesis ofmultiple immunoglobulin classes by single lymphocytes. Cold Spring Harbor~· Quant. Biol. 41: 175. Rabbitts, T. H., A. Forster, w. Dunnick and D. L. Bentley. 1980. The role of gene deletion in the immunoglobulin heavy chain switch. Nature 283: 351. Sakano, H., J. H. Rogers, K. Hilppi, C. Brack, A. Traunecker, R. Maki and S. Tonegawa. 1979a. Domains and the hinge region of an immunoglobulin heavy chain are encoded in separate DNA segments. Nature 277: 627. Sakano, H., K. Hilppi, G. Heinrich and S. Tonegawa. 1979b. Sequences at the somatic recombination sites of immunoglobulin light-chain genes. Nature 280: 288. Sakano, H., A. Maki, Y. Kurosawa, w. Roeder ands. Tonegawa. 1980. The two types of somatic recombination necessary for generation of complete immunoglobulin heavy chain genes. Nature (in press). Schilling, J., B. Clevinger, J.M. Davie and L. Hood. 1980. Amino acid sequence of homogeneous antibodies to dextran and DNA rearrangements in heavy chain V-region gene segments. Nature 283: 35. Seidman, J. G., A. Leder, M. Nau, B. Norman and P. Leder. 1978. Antibody diversity. Science 202: 11. 86 Silverman; M., J. Zieg, G. Mandel and M. Simon. 1980. Analysis of the functional components of the phase variation system. Cold Spring Harbor ~· Quant~ Biol. (in press). Southern, E. M. · 1°975. Detection or specific sequences among DNA fragments separated by gel electrophoresis. Steinmetz, M. and H. G. Zachau. ~· 1980. Mol. Biol. 98: 503. Two rea~ranged immunoglobulin kappa light chain genes in one mouse myeloma. Nucleic Acids Res. 8: 1693. Takahashi, N., T. Kataoka and T. Honjo. 1980. Nucleotide sequences around class switch recombination site of the immunoglobulin y2b chain gene of mouse. Gene (in press). Tonegawa, S., A. M. Maxam, R. Tizard, O. Bernard and W. Gilbert. 1978. Sequence of a mouse germ-line gene for a variable region of an immuncr globulin light chain. Proc. Nat. Acad. Sci. 75: 1485. Weigert, M., R. Perry, D. Kelley, T. Hunkapiller, J. Schilling and L. Hood. 1980. The joining of V and J gene segments creates antibody diversity. Nature 283: 497. Yaoita, Y. and T. Honjo. 1980. Deletion of immunoglobulin heavy chain genes accompanies the class switch rearrangement. press). Biomed. Res. (in 87 FIGURE LEGENDS Figure 1. Organization of immunoglobulin genes in BALB/c mice. gene family is loc~ted on a separate chromosome. Each The lambda gene family contains at least two units (A.I and A. 11> organized as shown. The probable order of the heavy chain constant region genes is shown beneath the diagram of variable region gene segments. Each CH gene contains multiple exons, as depicted for Cµ (Early et al. 1979; Sakano et al. 1979a; Early et al. 19SOb). Slash marks indicate linkage relationships which have not been firmly established. Figure 2. Distances are not to scale. B-cell differentiation. Cells initially committed to immunoglobulin synthesis first express variable regions on membrane-bound IgM (and IgD in many cases). After stimulation by antigen, these cells or their progeny can secrete immunoglobulins with the same variable region, but any one of the eight classes or subclasses of heavy chain. Figure 3. The upper diagram depicts possible germline D gene segments. Relative distances of the various gene segments from one another are undetermined. The short arrows indicate conserved noncoding sequences (Table 1) which may be involved in DNA rearrangement. ment joins VH-D-JH gene segments. In this model, DNA rearrange- Intervening DNA may be deleted, or could -undergo other types of rearrangement. The lower diagram shows paired VH and D (alternatively D and JH, or VL and JL) gene segments with 11 and 22 nucleotide spacers. Putative DNA-joining proteins might bind to the areas enclosed by dashed lines. The gene segments are represented as colinear to emphasize the symmetry of the conserved noncoding nucleotides. The actual structure may bring the ends of the two gene segments into close proximity. 88 Figure A. Heavy chain junctiona.l diversity. The genes whose partial sequences k are shown· all are joined to JHl (JH107 in Early et al. 1980). of these genes are homologous, although not identical. The D segments The three myeloma genes. encode phosphorylcholine-binding VH regions of subgroup Ill, while the two rearranged genes from normal IgM-producing B cells isolated by fluorescence activated cell sorting are subgroup I (VH49) and subgroup II (VH2) (Kabat et al. 1979; P. Early and C. Nottenberg, unpublished observations). The upper panel shows variations in D segment sequences and lengths, while the lower panel shows variations in the points of D-JH1 joining. Figure 5. Alternative possible reading frames for a D gene segment. The protein sequence of the 8107 D segment is shown in italics (Early et al. 1980). V(D)J joining at the sites indicated by dotted lines can generate the A4 protein sequence from a germline A4-like VH gene (S. Crews, unpublished observation), the D107 gene segment, and JH3 gene segment. Figure 6. Clones containing rearranged alpha heavy chain genes and germline Cµ and Ca genes. The rearranged T15 and M603 genes were isolated from genomic libraries of their respe·ctive myeloma tumors, and MC101 was isolated as described (Kataoka et al. 1980). Clones containing the germline Cµ and Ca genes were isolated from libraries of mouse sperm DNA (Davis et al. 1980a). Raised boxes denote coding regions. The shaded areas indicate homologies with germline genes and their flanking sequences: vertical lines, VH and flanking region; solid, Cµ and flanking region; dots, Ca and flanking regions; and white, uncertain origin, probably Cµ derived (see text); diagonal lines, Cy1 and flanking regions. restriction mapping. Homologies were determined by detailed Restriction enzyme sites are denoted as follows: Hf = Hinfl, Hh = Hha, H = HindIII, M = Mspl, R = EcoRI, Rs = Rsal, S = Sau3a, and Mb = Mboll. The position of the JH regions was determined from experiments 89 described in Early et,~ al. (1980) and Sakano et al. (1980). The M603 clone was isolated and characterized as previously described (Davis et al. 1980a; Early et al. 1979; J?avis et al. 1979).. The T15 library was made from EcoRI partially digested DNA by procedures described in Davis et al. (1980a) and Maniatis et al. (1978). Recombinant phage were screened by the procedure of Benton and Davis (1977) using 32p nick translated Vtt + Ca cDNA clones as probes. Southern blot analysis (Southern 1975 of EcoRI digested myeloma DNA indicated identity of central EcoRI fragments in the T15 genomes with the clones shown here (data not shown), indicating that the clones isolated are representative of their genome of origin. reported for M603 (Davis et al. ~979). the procedures of Maxam and This analysis has been previously The DNA sequence analysis employed Gilbert 1980. Arrows originating from a restriction site indicate 5' end labeling with y32P-ATP, while arrows ending in a restricting site indicate 3' end labeling with a32P-dNTPs. Figure ? . Switch sequences for the rearranged T15, M603, and MC101 genes and their germline Cµ and Ca counterparts. Cy1 corresponding sequences for MC101 have been reported by Honjo and colleagues. Ctt switch sites are shown in 5' -+ 3' orientation. Sequences surrounding Underlining indicates sequence identity of unrearranged Cµ (open blocks) and Ca (solid blocks) flanking sequences with their counterparts in the individual rearranged genes. Dots indicate 10 nucleotide spacings. Figure 8. A comparison of the DNA sequences of the germline Cµ and Ca switch sites for the rearranged Tl5, M603, and MC101 genes. Sequences obtained from Cµ and Ca flanking regions were aligned for maximum homology around their Ctt switch sites. Boxes indicate nucleotide identities and - indicate a gap introduced for homology alignment. Arrows indicate breakpoints for the switch sites of each rearranged gene. Also indicated is the Cµ switch 90 site for a Y2b producer, M141 (Takahashi et al. 1980; Sakano cL cu. • .a.::180). A Jc consensus sequence for .at least some cases of Cµ switching might therefore be GGTNATTANNNNNNGGTANNCAAAG which does not occur elsewhere in any of the some 1900 nucleotides of Cµ flanking region sequenced (Sakano et al. 1980). .Elements of this Cµ homology region have been suggested previously to play a role in CH switching (Takahashi et al. 1980; Sakano et al. 1980). A consensus sequence for Ca switching derived from the examples here would be PGTCPPGCTGGAATPPGYTGGGNTGPGCTG. (P = purine, Y = pyrimidine, N = any nucleotide). Figure 9a. Schematic diagram of the location of seventeen 30-nucleotide repeated sequences adjacent to the germline Ca gene. Arrows indicate which repeats are used in the three rearranged genes described in this paper. 9b. DNA sequences of 17 repeats found interspersed in germline 5' flanking Ca sequence. Boxes indicate nucleotide identities and - indicates a gap. Figure 10. A comparison of the switch sequences from germline 5' flanking sequences for the Ca {see Fig. 8), Cyi (Kataoka et al. 1980), and y2b (Takahashi et al. 1980; Sakano et al. 1980) genes. Boxes in the germline a sequence denote the conserved bases of the 30-nucleotide recognition sequences. Boxes jn the germline yl and y2b sequences denote nucleotide identities to the germline Ca sequence. Figure 11. - indicates a gap. Model for class-specific regulation of CH switching {see text). In this scheme the small boxes represent recognition (S) sequences which bind to switch proteins to mediate CH switching; the circle represents switching proteins and the large boxes represent coding regions. This model depicts a CH gene order of Cµ, Cy, Ca (see text) but this is not a requirement. 91 Figute·.12 . . An exan:iple of abortive rearrangement in a Ca gene isolated .from the myeloma tumor M603. The clone CH603a30 is identical to DNA flanking the germ line Ca gene up to approximately 2 kb from the 5' end of the· Ca coding region (A). DNA sequence analysis (B) shows that the breakpoint of the recombination occurs within one of the 30-nucleotide repeat sequences (#17 in Fig. 9) associated with heavy chain switching. 92 Table t. Comparisorr of recognition nucleotides adjacent to V and J gene segments. K sequences are from Max et al. (1979) and Sakano et al. (1979b; A from Bernard et· al. (1978) and Tonegawa et al. (1978); heavy chains from Early. et al. (1980) and Bernard and Gough (1980). The mRNA-sense strand is shown 3' to V gene segments and 5' to J gene segments. The asterisk indicates a possibly nonfunctional JK gene segment; it has not been found in a rearranged gene. The column to the right lists the number of nucleotides between the underlined hepta- and decanucleotides. .. • 93 . k :I. 0 0 " u ;II( aj (.) ~ >-.. (.)~ cJ- .., ! ;II( ..," Cr ' . '° 0: (.) N >"" • ~, ~ U> (\I C» - CD 0 en 0 0 E 0 e0 E 0 ~ s::. u e en .... &. 0 Getaline v8 •••• TAT'l'A~AAC TyrTyrCyaI I I I Dl07 7"rTyrGlySer TACTACGG,TAGTA ThrThrG. I .I Gerallne JR3 GC~AC •••• lyPheAlaTyr A4 protein (2,1 levan) •••• TyrTyrCyaThrThrGlyPheAlaTyr •••• 97 lkb scaler---; S region . ~ . VOJ · 5.4kb m< Ca _..~·. ·. ·· ......., TIS 1111' ' ' ' ' ' ""' 'hl; :=·= :, : t=== ~ ,. ··.: . . ,.,.,,>sr+·········>········ 1 • R M R VDJ Ca "f* ·· · · M603 6.8 kb-----,,...rJQJ'J 1111111111111111111ITt~:~:::-.-.- - -t_;,:=-=-:" " · ~"T'·.7(7 ·• ~:·7·. ••••0· . •0··<0 R 1 HSSM R ' . fJ!//j/J//JJJIJf)!!I(~ I R MCIOI H I kb scale ,. . . .-----, J region c l: ! ~'1') i' R p ( Hf germ line :· · ·:.: ·•· ,·: ·r~··~(·;t=:· ·1~·.. 1·· ······· 1 I Mb SS RsM Hf 5 R Ca 98 Sim:> le Switching Cµ • • • • • • ACTTCCTGGTTGTTAAAGAATGGTATCAAAGGACAGTGCTTAGATCCAAGGTGAGTGTGA TIS . . . . . . ACTTCCTGGTTGTTAAAGAATGGTATCAAATGGGTTGAGCTGAACTAGAATAAACTTGGC 41 Ki •-m :: ;•s <F : ·mA1 · • == Com;> lox Switching Cµ AAGGGAACAAGGTTGAGAGCCCTAGTAAGCGAGGCTCTAAAAAGCATGGC .GGACTAGGCTGGAATAGGTTGGGCTGGGCTGGTGCGAGCTGGGTTAGGCT . . . . . C ;; * a M603 •t a ? ¥* •ets t AAGGGAACAAGGTTGAGAGCCCTAGCGTGAGTCTGAGCTGGGGTGAGCTGAGTGGGCTGAGTTGGGGTGA . . . . . . . . . . . . . . AGCTGGGCTGAGCTGAGATGAGCTGGGGTGAGCTGAGCTGAGTTGAGCTGGGGTGAGCTGGAGCTGGGCT . . . . . . . AGCTGAGCTGGGGTGAGCTGAGCTGAGCTGGGGCTGAGCTGAGCTGAGCTGAGCTGGGCTGAGCTGGGCT .GAGCTGGGCTGAGCTGGGCTGAGCTGGGCTGAGCTC3GCTGGTGCGAGCTGGGTTAGGCT . . . . . . GCTGGGCTGAGTCTGGGGTGAGCTGAGCTGAGCTGGGGTGAGCTGAGCTGGGGTGAGCTGAGCTGAGCTG f# M Succosstvo Switching - . . . . %'f ggs i . Cµ AAAATGCGCTAAACTGAGGTGATTACTCTGAGGTAAGCAAAGCTGGGCTT c~ GGCTGAGAGCTGAGCTGAGCTGGAATGAGCTGGGATGAGCTGAGCTAGGC ~.CIOI AAAATGCGCTAAACTGAGGTGAT:~TGAGCTGGGATGAGCTGAGCTAGGC • I • • c =: • Q ; : : ; • ; =·A ,. ; 5s : 5 & Genni rno clJ Switch Soquonces SMl41 5 rt5 ~ ~ r cc TIGG TIT GIT r AIA AG A A TIGG r Al- rte A A A GIG Ac AG 5 MCIOI ... c T G AIG G TIG AIT T Ate Tc T G AIG GT AIA Glg_A A A Glc T G G c 5 M603 ~ . AG AG cc c TA GIT AIA G c GA GIGlclTIC TA AIA A A GIG AT G G Gorml t ne co co ca Switch Soquencos 5rt5 ~ T G GIG c Tf G AjG c T G GA A TfG A1G1c1r G G GITIT GIAlG c T GIA A 5 M603 ~ G G AjG c TjA GIG c T G GA A TIA GIGITIT G G GlclT GIGIG c r GIG T 5 tel01 ... T G AIG C TIG AIG CT G GA A TIG AlQJCIT G G GIAIT GIAIG CT GIA G 100 a (.) 101 Ca Consensus sequence (F;g. 3) P G C T MC101 5MC101 5T15 SM603 P P GC T GGA A "' T P P G Y"' T G G"' G N T G P G C T G Sl A G CT - - G A GCTGGAAT GA G c T G G G A - T G A G C T G 52 A G CT - - A G G C T G G A A T AG G c G ti G c - T Cl G G C T G 53 G G CT - - G A G C T G G A A T Gp, G c G GGT - T GA A C T G Ejr G G A A T A G G C G G G C - T G G G C T G S4 G G C T - - - G G SS A G CT - - GA G C T G G A A G G A G AG G0G A A G A G GBG M603 S6 A G CT - - A G G C T G G A A T AG G T T G G G c - T G G G C T G 57 A G C T - - GA GCTGGAAT GA G c T0G G A - T G A G C T G SB A G C T - - A G G C T G G A A T AG G c T G G G c - T G G G C T G s; A G C G G - A f\ G C T G G A A T GA G c T G G C A - T G G G CT G SlO A G C T - - A G G C T G G A A T AG G T T G G G c - T G G G CT G Sll A G C T - - A G G C T G G A A T A G G c T G G G T T T G - G CT G . 512 A G CT - - AG GCTGGAAT - G G c T G G G c - T G G G C T G 513 A G CT C G A A G C T G G A A T GA G c T G G G A - T G G G C T G S14 A G CT - - A G G C T G G A A T A G G T T G G G c - T G G G C T G 515 A G CT - - G A G C T G G A A T GA G T T G G A A - TA G G C T G 516 G G CT - - G T A C T G G A A T GA G c T GA G C - T G A G C T G T15 517 G G C T - - G A G C T G G A A T G A .G c T G G G T - T G A G C T G P = purjne Y = pyrimidine N = any nucleotide 5 MCIOI Gorml t ne a 5Tl5 5 M603 ~ p IG. c TI p p -,G_C_T_G_G_A_A---.TI p p 1§1 y"'~ N [[§) p IG c T GI "' 5MCIOI Germl tne YI "' C A G G G A IG C T G G AIG C T GA -1T G G GIT A T A A A A A I 5 M1·41 Germl Ina Y2b "' T G G G G G CIC IA IG G A IG A G T T GL!J C C~ A(!]T G A G C A ..... Q ~ VDJ ----o @ []-{}{] ~~~· Cµ. I I // Iii ® Cy t ~~~ I II I HI ~~~ heterodlmer formation Ca I ,......- c,, .... Q c:.:i -Ca ! site-specif le recombination VDJ ---10 CHll I ~ 1-I- - ·1 I ·1 I 0 rt> " J rt> 0 U> I :e c: -• 0 0 3 ·.i1 a:: ~ ~ . ~I ..!? u ".,, 0 I 105 .vec41 CACAG'l'GATACAAA'l'CATAACATAAACC Cll) VIC2 CACAGTGATTCAAGCCATGACATAAACC (11) VIC3 CACAGTGAT'l'CAAGCCATGACATAAACC (11) VIC2l CACAGTGC'l'CAGGGC'l'GAACAAAAACC (10) VH107 CACAGTGAGAGGACG'l'CATTG'l'GAGCCCAGACACAAACC (22) v>.1 CACAATGACATGTGTAGATGGGGAAGTAGATCAAGAACA (22) v>.11 CACAATGACATGTGTAGATGGGGAAGTAGAACAAGAACA (22) JICl GGT'l'Tl'TGTAGAGAGGGGCATGTCATAGTCC'l'CACTGTG (22) JIC2 *JIC3 GGT'l'Tl'TGTAAAGGGGGGCGCAGTGATATGAATCACTGTG (23) GGGTTTTGTGGAGGTAAAG'l'TAAAATAAATCACTGTA (20) JIC4 AGT'l"l'TTGTATGGGGGTTGAGTGAAGGGACACCAGTGTG (22) JICS GGT'l'TTTGTACAGCCAGACAGTGGAGTACTACCACTGTG (22) '1Bl AGT'l'TTAGTATAGGAACAGAGGCAGAACAGAGACTGTG (21) JB2 GGT'l"l"T'l'GTACACCCACTAAAGGGGTCTATGATAGTGTG (22) JB3 A'l'TTATTGTCAGGGGTCTAATCA'r?GTTGTCACAATGTG (22) J).1 GG'l'T'l'TTGCATGAG'l'CTATATCACAGTG (11) i{ cH__ ~ ,... 2 D-f Q-1 ~{ V.-.H,,_1_0_7__,ITACTACGGTAGTAl J"1 VH603 JH1 Vwl67 ITACTACGGTAGTAd I LACTACGGTAATAGCTACTTTG Jw1 ~GAAGTAI JN1 I ~ []-i~-.-.~-H4~9---1TATTACTACGGTAGTAGTa--a-H-,__, JH1 I{... 01rn v"gc3 VHl0'7' 1ii :a D-1 v1:11g7 JDIGcTACTGGTAC•••• g ICTGGTAC •••• D lGGTAC • • • • 9H2 biGCTACTGGTAC•••• •{ rn VH'I§ ! 01 I 12 lACTGGTAC••••