CELLULARIMMUNOLOGY17, 552--559 (1975)
COMMENTARY Theories of Antibody Diversity: The Great Debate FRED C. OSHER
Biological Laboratories, Harvard University, 16 Divinity Avenue, Cambridge, Massachusetts 02138 Received August 10, 1974 This is in essence the oldest of biological problems--the classic struggle between the Lamarckians and the Darwinians on the nature of adaptation in evolution. It is a theme that has been played on in an infinity of variations over the last 100' years, and I can only hope that in the present treatment there are scattered elements of novelty, either in regard to example or approach, which will make up for the inevitable inclusion of much that is trite and familiar. Sir 3/Iacfarlane Burnet (1) The Scene :--Professor George Line, on the way to deliver his Bio 14 lecture, runs into Professor Samuel Soma
Prof. G. Line: Say Sam, one of my students told me about the lecture you gave yesterday. It's ludicrous to propose an alternate mechanism for antibody diversity when other sets of related proteins have been shown to evolve along germ-line mechanisms. W e see this, for example, with hemoglobin and cytochrome c. Braunitzer (2) has shown that by comparing the variable regions of amino acid sequences in hemoglobin from different species, clear-cut evolutionary trends are found. Therefore, it's virtually unnecessary to look for further ad hoc mechanisms to generate antibody diversity. Prof. S. Soma: Slightly different matter, I ' m afraid--no one attempts to claim that there are a million different genes for cytochrome c in every cell. On the other hand, Burnet (3) points out that clonal phenomena in bacterial and viral populations are basic biological mechanisms. There's no reason why the same mechanisms which produce mutations and selection for different bacterial and viral lines can't operate on a somatic level with lymphocytes in the thymus or bursal equivalent. Burnet's approach led to the successful clonal selection theory of acquired immunity, so why can't the same approach work to explain the somatic generation of diversity ? G. Somatic theories force you to accept tremendous amounts of cell wastage in such processes. Isn't that a bit unreasonable, when the germ line can pass on a ready-made set of responses to each individual ? S. Well, the fact is that that sort of cell growth and death does seem to happen in the thymus ( 4 ) - - s o why not put it to use? And as Milstein and Munro (5) 552 Copyright © 1975 by Academic Press, Inc. All rights o~ reproduction in any form reserved.
ANTIBODY DIVERSITY DEBATE
point out, a somatic mutation and selection process may waste cells, but a germ line system wastes DNA, and selection is at the expense of entire individuals during evolution. The germ line theory needs 10" genes' worth of precious DNA for antibody coding--doesn't that seem like an awful lot ? G. Obviously you don't realize how small a percentage of the genome that represents. With random recombination of constant and variable regions you can get a large amount of diversity from relatively few subgroups of antibody genes. Wang's work (6) has shown that heavy chain variable regions associate with numerous constant regions. And suppose you have 10,000 VL and 10,000 VII regions. This accounts for 10s different antibody molecules. Hood (7) suggests that even if only 10% of the reciprocal sharings are possible, one could generate responses to 107 antigens. With 107 amino acids of a variable region, and the 321 base pairs needed to code for these amino acids, my proposed 20,000 genes take up only 0.1% of the 3.7 × 109 base pairs of genetic information for the human cell. Now really--is this an unreasonable proportion for such an important system ? S. Yes, I most certainly do think so, when somatic theories can yield the same diversity with a mere fraction of that number of genes. G. Listen, Sam--you're beating around the bush. Let's get down to brass tacks. Just the fact that antibodies are composed of a constant and a variable region necessitates your postulating a mechanism to explain this selective mutation. Any hypermutation theories you present are ad hoc explanations of this phenomenon. And on top of that, you're trying to tell me that from a few prototypical antibody genes you can generate enough diversity to explain all antibody specificities--106 is a lot of mutations ~o occur during ontogeny. S. Look, there are a number of well-documented, reasonable ways to get the rlecessary amount of mutation-G. Name one. S. First of all, Eisen (8) shows that special mechanisms for increasing mutation rates aren't really even necessary. With even ordinary mutation rates--say 10-7 per base pair per cell division--the mutation rate per V gene could be about one for every 10~ cell division, since there are about 330 base pairs per variable region gene. Even if only 10% of these mutations could eventually yield a functional immunoglobulin chain, this means that in a population of rapidly dividing lymphocytes--the 108 cells we have in the thymus--1000 immunologically competent cells with variable region gene mutations could arise each day. In fact, Coffino and Scharff (9) examined the rate of somatic mutation in immunoglobulin production in mouse myeloma cells themselves. In a simple, unadulterated soft agar culture, they observed that heavy and light chain producing lymphocytes spontaneously mutated to light chain-only producing cells at a rate of 1.1 × 10-2 per cell per generation. That's higher than other measured rates for somatic cell markers, which for mammalian cells have been between 10-~ and 10-8 per cell per generation. These findings even suggest something that Burnet had proposed (10)--that the production of antibody diversity goes on even after ontogeny.
OSttER AND NEAI.,
And then there's-G. Hold on just a minute. You can get all the mutants you want, but they're all random mutants. When you compare amino acid sequence variations, there are patterns--some regions never vary, while other positions d o . . . S. Not so fast there yourself, George. Efficient selection mechanisms could certainly give those patterns. And of course, there are other legitimate--call them "ad hoc" if you will--mechanisms which could both increase the number of mutants and at the same time do so in a nonrandom manner. Smithies (11) has suggested somatic recombination, which could perhaps be encouraged by certain coding strips along the DNA of the variable region genes. Brenner and Milstein (12) give the hypermutation idea a mechanism that both yields a high mutation rate and limits the region for mutation at the same time. They suggest that part of the variable region gene acts as a recognition site for an enzyme which cuts into the DNA. During the repair of this region, errors are introduced at random at a high rate--a documented feature of replication and repair processes--yet these mutations affect only the area which has been excised in the first place. The "hot spots" seen in sequence analyses could certainly be such areas, with accompanying enzyme recognition sites. Notice too that the process can be repeated in the same region again and again without accumulation. G. If it were a used car, I'd buy it, Sam. Are you finished yet ? S. As a matter of fact, I'm not. Speyer's research (13) with bacteriophage T4 shows that DNA polymerase mutants can cause mutations in the DNA. In addition, different DNA polymerase mutants have different effects on the DNA---so that a DNA polymerase mutant has the properties of being a specific kind of "mutator gene." Perhaps such mutants are responsible for the hypervariable regions of the V gene. G. Even if I did accept these ridiculous mechanisms--which I don't--according to your proposals the differences in antibodies accumulate separately in each individual. Fitch and Margoliash (14) have developed an analytical technique that refutes your somatic mechanisms! Genealogical analysis allows the determination of immediate and distant ancestral sequences of antibody molecules. By observing homologous patterns in amino acid sequences, the minimum number of mutational events required to construct a set of proteins from a single ancester gene can be determined. Statistical methods can then be applied to find the minimal number of prototype genes needed to explain antibody diversity. Dreyer (15) first noticed these patterns in amino acid sequences over nine years ago. Since that time, much more evidence has pointed out the evolutionary trend of antibody diversity. For example, you must accept that there are a bare minimum of 10 light regions coded for by separate genes-- VKI, n, m, CK, VxI, ii, m, iv, and Cxi, ii (7). Smith, Hood, and Fitch (16) have now positively identified four separate subgroups of the V~ region through genealogical analysis. Now, you must again raise your core group of germ line genes. When will you stop ? The trend is evident. Statistical calculations (17) of 52 randomly chosen human K chains suggest that there are at least 425 V~ genes. How about that ?
ANTIBODY DIVERSITY DEBATE
The branching pattern to the outermost terminal branches implies that the genes are subjected to intense selectional pressures even at this level. Hood and Talmage (18) have also used genealogical trees to show that the V and C regions of different species are diverging at the same rate. I see some s o r t of evolutionary trend . . . don't you ? The close agreement between individuals as evidenced by the genealogical analysis would necessitate excessive parallel mutation if you somato-boys are correct. I don't think you can get around this one, Sammy. S. Let's not get so self-confident quite yet, George. Who says that these "genealogic" relationships have to occur over millions of years ? Strictly speaking, this tree is independent of time altogether. Now, I'm certainly amenable to thinking that there's one gene, or maybe even a few, for each subgroup-G. Do I detect a change in your position ? oc. I'll ignore that rude interruption. As I was saying, this pattern could easily be the result of either a nonrandom mutation mechanism-G. A d hoc! A d hoc! A d hoc! !
S. - - o r an efficient, standardized selection mechanism operating throughout a species. Up until now, that's been the biggest obstacle for somatic theories : a selection mechanism powerfffl enough to explain the pattern of variability--the limited alternatives at certain positions, the "hot spots," and the rigid invariability at most positions. But Jerne (4) has devised just such a mechanism, one that not only presents a simple selection mechanism, but also explains the phenomena of self-tolerance and allo-aggression. In Jerne's theory, the germ line variable region genes of an animal species are the structural genes for antibodies specifically directed against the set of histocompatibility antigens of the species. Since an individual inherits only a certain number of these species histocompatibility antigens, the full set of antibody genes which the individual inherits can be divided into two subsets--those directed against the histocompatibility antigens which he happens to possess (the S, or self set) and those directed against all other of the species histocompatibility antigens (the A, or allo set). During ontogeny, stem cells of both sets enter the thymus where they proliferate, perhaps stimulated by some hormone and by contact with the individual's own histocompatibility antigens. It's been shown that only small numbers of lymphocytes ever leave the thymus, so it's quite reasonable to assume that negative selection against nonmutants may be taking place there. Nonmutants of the S set are identified by their affinity to the histocompatibility antigens present in the thymus, and they are prevented from leaving the thymus on the basis of this affinity. Thus only the nonmutant lymphocytes expressing the A variable region genes, and mutants of the S set which no longer have affinity for the individual's histocompatibility antigens are allowed to leave. Double and triple mutants are possible as the original mutants proliferate inside the thymus, unaffected by negative selection. The known physiology of the thymus is compatible with this model, and the presence of the A subset lymphocytes accounts for the strong alloaggression seen between animals of a species. And of course, self-tolerance is achieved in the same simple selection process. G. Hmmm. That's pretty interesting, but it still doesn't answer some important questions.
S. Oh yeah ? Like what ? G. It doesn't explain the developmental qualities of the immune system• Silverstein and Prendergast (19) have experimented with fetal lambs and have shown that there is a stepwise maturation of immunologic competence to different antigens at different stages• The timing of these responses is quite regulated and exact among fetuses• For example, antibody formation is first seen in response to bacteriophage • X, then later against ferritin and hemocyanin, and still later the ability to show homograft rejection comes about• Thus it appears that antibody diversity doesn't arise in random fashion, but is a carefully controlled and reproducible sequence. As the authors concluded, "the data from these developmental studies strongly suggests that the generation of immunologic diversity cannot be based on a series of somatic gene alterations in ontogeny." I concur• 5". Well, those first antibodies could be single mutants, the second could be double mutants, which would take more time to arise out of the colony of single mutants, and the third class of antibodies could be triple mutants . . . G. Aw, come off it. 5`. Okay, hotshot--if the generation of antibody diversity is so "carefully controlled" and "reproducible," then how about idiotypes?! Kelus and Gell (20): injected rabbits with one antigen, and they couldn't find two rabbits that responded with the same antibody--not even related rabbits! Not even parents and offspring! So how can you suggest that a rabbit inherits thousands of antibody specificities when experimenters can't even find one specificity handed down from a rabbit to its offspring ? ! G. Looks like you're subscribing to the wrong journal. Capra and Kunkel (21) found variable regions from two different human myeloma antibodies with amino acid sequences that were identical for the 40 terminal positions. 5`. Well, maybe those were Jerne's A set antibodies• And who says that you can generalize from myeloma cells--by definition abberant--to a normal immune system ? The last time that the big boys of antibody diversity got together for a talk (22), Smithies warned Milstein, Talmage, Edelman, Hood, and Jerne against generalizing with confidence from malignancy-derived myeloma proteins. G. Perhaps--but what else do we have to go on ? You're evading the issue. Let's get back to idiotypes, since yo~ brought them up in the first place. Weigart et al. (cited by Smith, Hood, and Fitch, 16) examined ten amino acid sequences from the mouse V X system. Six of these were found to be identical while the other four had very few mutational differences. I'd say this indicated a common ancestor • . . wouldn't you ? oc. I'd say that those six w e r e the "common ancestor," and that the other four were somatic mutants. And wait a minute, George--I'm asking the questions just right now. How about a problem that doesn't depend on what journal I've read: if useful antibody specificitles have been selected for and maintained by selection pressures over millions of years, and handed down from generation to generation without any random element, how's it possible for an individual to respond to an
ANTIBODY DIVERSITY DEBATE
antigen which no species has ever encountered before--newly synthesized chemicals like benzenearsonate, one that Eisen (8) points out ? G. That's funny--I think that Dr. Eisen himself suggested a pretty good explanation. In an earlier paper, Eisen et al. (23) showed that myeloma proteins with antibody-like activity had the ability to cross-react with a variety of antigens with certain stereochemical features. Yount et al. (24) verifies this possibility by finding that an antiserum may possess several sets of antibodies against a single antigen. These mechanisms also reduce the number of antibody genes necessary and make the germ line hypothesis that much more practical. The beauty of this reasoning is that the individual is fully prepared to fight off the existing environmental hazards. Over the course of time, selection via the germ line has given the individual protection against a wide variety of antigens. Do you understand ? S. Enough is enough! ! Perhaps, Professor Line, we should discuss allotypes--
G. (aside) My God! I knew that was coming . . . (aloud) I'm afraid I have to go now; it's time for my Bio 14 l e c t u r e S. Not so fast! You're staying right here! G. Okay, okay! but let go of my tie! S. The somatic theory can explain rabbit heavy variable region allotypes with perfect ease. These three sequences, each with their own marker al, a2, or a3, seem clearly to be alleles at one locus (16). So it's likely that each one is simply one of three basic rabbit heavy chain variable genes, two of which, in each rabbit, undergo somatic mutation to create the minor differences--minor compared to the tenresidue differences between the 34 amino terminal positions of the al and a3 chains. An apparent case of a limited number--only 3, mind you--of V genes undergoing somatic mutation. On the other hand, how can you possibly find some way--or reason--for your thousands of ancestral V~ genes to have separated into three allelic groups somewhere along the line ? Face it, George, you CAN'T! G. Let me try. Smithies (25) proposes that antibody differentiation is controlled by forks in DNA replication. He assumes that DNA exists in a twodimensional network where at branchpoints there are protein switches that can be either "left" or "right." With this model, you can have multiple V genes which are linked in tandem and for which DNA polymerases travel down only one path. Thus the allotypes in rabbits need not be explained by extreme selection pressures, but rather by an evolution and replication of a permanent D N A fork. S. E1 grande ad hoc! ! George, I never thought that you would stoop so low as to tamper with the sacred double helix! And you've got some more tales to weave-like an explanation for species and subgroup specificities. You insist that hundreds of variable region genes have been selected for and maintained within each subgroup. Well, that's hard to believe, since the selection pressures on one V gene out of hundreds would be negligible--especially if, as you say, multiple speeificities are the rule. But suppose we do assume that there's powerful selection pressure for diversity even within subgroups. N o w you've got to explain why certain segments have remained invariant all this time. So you say that there may not be any pressure on those positions after all. But now how do you explain the fact that the
OSHER AND N E A L
invariant residues are different, or specific, for each subgroup or species? You have to invoke that strong selection pressure again, which you've just denied for those segments. Only now that selection pressure has to occur at the exact moment of speciation, and act with unbelievable speed according to your own timetable. Obviously you've gotten yourself into a trap here, perverting the evolutionary process like you have to for the germ line theory. G. This poses no real problem. There are completely logical mechanisms to explainS. Like what ? G. With unequal crossing over between homologous genes (16) rapid expansion and contraction can take place. The ensuing duplication and deletion of these genes could easily result in different evolutionary lines. Other multigene systems do this, so why not the immune system ? S. That's ridiculous! Species specificity has got you beat! l G. What?! That's just flexible response to the environment, which a good multigene system provides ! S. Some service! Killing off individuals while somatic selection kills only cells[ G. You somato-boys would try to sell your own grandmother an ad hoc mechanism !
S. Me ? How about you and species specificities ? You and idiotypes ?! G. Nonrandom mutations! S. Allotypes ! G. Genealogical tree! S. Go sit in your genealogical tree! ! ! ! G. (Expletive deleted.) 1 ACKNOWLEDGMENTS The authors extend thanks to Tom Wegmann, whose suggestion prompted this cooperative venture, to Alwin Pappenheimer, Jr., for his encouragement, and to Leroy Hood, whose kind offer opened up a new world for the authors. 1 Originally submitted by the authors as a joint term paper for Biology 14, the Introductory Course in Genetics at Harvard College. REFERENCES 1. Burner, Sir Macfarlane, "The Clonal Selection Theory of Acquired Immunity," Cambridge University Press, 1959. 2. Braunitzer, G., 7". Cell Physiol. 67, 1, 1965. 3. Burnet, Sir Macfarlane, "The Clonal Selection Theory of Acquired Immunity," Cambridge University Press, 1959. 4. Jerne, N. K., Eur. J. Immunol. 1, 1, 1971. 5. Milstein, C., and Munro, A. J., Ann. Rev. Microbiol. 24, 335-358, 1970.
ANTIBODY DIVERSITY DEBATE
6. Wang, A. C., Pink, J. R. L., Fudenberg, H. H., and Ohms, J., Proc. Nat. Acad. Sci. 657, July, 1970. 7. Hood, L., and Talmage, D. W., In "Developmental Aspects of Antibody Formation and Structure" (J. Sterzl and I. Rhia, Eds.), Vol. II, Academic Press, New York. 1970. 8. Eisen, H. N., "Immunology," Harper and Row, 1974. 9. Co ffino, P., and Scharff, M. D., Proc. Nat. Acad. Sci. 68, No. 1, 219, January 1971. 10. Burnet, F. M., "Cellular Immunobiology," Cambridge University Press, London, 1969. 11. Smithies, 0., Science lS7, 267, 1967. 12. Brenner, S., and Milstein, C., Nature (London) 211, 242, 1966. 13. Speyer, J. F., J. Biochem. Biophys. Res. Commun. 21, (1), 6, 1965. 14. Fitch, W. M., and Margolish, E., Science 155, 279, 1967. 15. Dreyer, W. J., and Bennett, J. C., Proc. Nat. Acad. Sci. 54, 864, 1965. 16. Smith, G. P., Hood, L., and Fitch, W. M., Ann. Rev. Biochem. 40, 969, 1971. 17. Hood, L., and Prahl, J., Advan. Immun. 14, 291, i971. 18. Hood, L., and Talmage, D. W., Science 168, 325, 1970. 19. Silverstein, A. M., and Prendergast, R. A., In "Developmental Aspects of Antibody Formation and Structure (J. Sterzl and I. Rhia, Eds.,) Vol, I, pp. 69-77. Academic Press, New York, 1970. 20. Kelus, A. S., and Gell, P. G., J. Exp. Med. 127, 215, 1968. 21. Cap,ra, J. D., and Kunkel, H. G., Proc. Nat. Acad. Sci. 67, 87, 1970. 22. Discussion chaired by Crick, F. H. C., Theories of Antibody Diversity Cold Spring Syrup. O.uan. Biol. 32, 171, 1967. 23. Eisen, H. N., Michaelides, 3/I. C., Underdown, B. J., Schulenberg, E. P., and Simms, E. S., Fed. Proc. 29, 78, 1970. 24. Yount, W. J., Dorner, M. M., Kunkel, It. G., and Kabat, E. A., J. Exp. Med. 127, 633, 1968. 25. Smithies, O., Science 169, 882, 1970.