The efficiency of antibody affinity maturation: can the rate of B-cell division be limiting?

Immunology Today, Vol. 11, No. 9 1990

The

efficiency of antibody affinity maturation: can the rate of B-call division be limiting?

It has been known for manyyEarsthat the affinity of antibodies for antigen increaseswith time during an immJne response. It is now clear that two processes play fundamental roles in this affinity 'maturation' in the mouse - V gene somatic mutation and antigen affinity-based selection. Exactly how these two processes work in concert is not fully understood. In th:,sarticle Tim Manserargues that models of affinity maturation based on the assumption that somatic mutation, antigen se'ection and B-cell division are in~rdependent may not explain the high efficiency of the process, and he suggest an alternative model.

During most humoral immune responses the average affinity of serum antibodies for antigen increases with time. A large body of evidence now supports the idea that the concerted action of antibody V region gene somatic mutation and antigen affinity-based selection is largely responsible for this affinity 'maturation' in the mous@~. Whether the selection process is directly perceptive of affinity or specificity differences for the eliciting antigen is unclear 3t present. However, increased affinity has been shown to correlate strongly with antibody predominance during responses to several haptens. Affinity maturation has often been explained by Darwinian modifications of the clonal selection hypothesis, in which the structure of the surface immunoglobulin (slg) expressed by the responding B-cell population is continually changing due to V gene mutation, and the number of offspring a given cell r n ~r ,rn. .r f. i.n. .n ~ i~ . . ~ . t~nv t~l ~. . ~ ,: .' F f c el,-, :~r . . . . . . . .c;an . . . . . . . .pmdiJc~, i~ n r.., . .f,;,n, ,;.h. .~j ,~'f ,.t| iwL..J Jl~.l ~~,,J| antigen ~.2,4.5.The most straightforward versions of such models assume that mutation, selection and cell division are interdependent processes.

Antibody Vgene somatic mutation Mechanistically, both somatic mutation and affinitybased antigen selection are poorly understood. Nevertheless, analysis of the V genes expressed by hybridom3s isolated at various stages of different murine immune respo~,ses has provided insights into the properties and capabilities of these processes. Somatic mutation occurs predomlnantly, if not exclusively, during the immune response6; it can take place at a very high rate7 (in comparison with spontaneous 'background' mutation), and this rate can be regulated8-~°. The mutation process is site-specific, altering sequence only throughout the length, and in the immediate 5' and 3' flanking regions, of rearranged VH, V, and V~ genes11,12.The end product of this B-cell specific mutation process is predominantly point mutations, with small insertions and deletions also occasionally obser/ed ~-5.~~.~2.The idea that 'mutation' in the mouse may act,Jally be due to gene conversion events, which are similar to those that have been shown to be responsible for the primary diversification of the chicken VH and Vx repertoires ~3, has been discounted by several investigations~4.~s.

Departmentof Bblovy, ?rincetonUniversity,Princeton,NJ08544,USA. ~ ) 1990, Elsewer Science PublishersLid, U K. 0167--4919/90/$02.00

Tim The efficiency of affinity maturafinn The exquisite efficiency ot the affinity-based antigen selectior, process, working in conjunction with somatic mutation, has recent!y become apparent. Many mutations can be generated and fixed in a single V region during the response, as a V gene expressed by a hybridoma isolated during the late primary or early secondary response contains six mutations on average~-4. Some mutations that confer even small measurable increases in affinity for antigen are recurrently observed among the V regions expressed by hybridomas~.2.~6. When the structure of the antigen is altered, changes in the pattern of such mutations are observed, in a manner that directly correlates with the cognate affinity increases conferred by these mutations (M. Fleming, S. Fish, J. Sharon and T. Manser, submitted). Multiple mutations that individually result in increased affinity for antigen are observed either alone, or in various combinations in elicited Vregions, and so haveapparently been fixed in those V regions by a process oT stepwise mutation and affinity-based selection~6-~8. Such mutations can be sustained and selected many times in a single B-cell clone irl a period of oniy 16 clays after initial immunization ~9. Indeed, the efficiency of the mutationselection process is so high as to call into question the idea that this efficiency is ultimately limited by the rate of B-cell division. In the following discussion the data that led to this conciusion will be further reviewed, and it will be proposed that an alternative model for the affinity maturation process should be considered. Mutation does not take placethroughout the immune response As mentioned above, the simplest Darwinian models of affinity maturation are predicated on the assumption that mutation, selection and clonal expansion are interdependent processes that take place for extended periods during the immune response. However, most data currently available support the idea that mutation does not occur throughou~ the immune response. No V gene mutation can be detected by hybridoma technology Jntil approximately one week a~er immunization 2.4, and two weeks after immunization the frequency of mutation assayed by hybridomas (particularly translationally silent mutations not subject to selection) does not differ greatly from that seen in the early secondary response (Ref. 20 and T. Manser, unpublished observations). More direct sampling of expressed V genes during the primary response by cDNA cloning suggeststhat the onset of mutation is probably several days earlier than day 7, and that the frequency of mutation in expressed Vgenes does not increasesignificantly after day 7; this implies that the mutation period is only a few days2~. Moreover, 305

Immunology Today, Vol. 1 I, No. 9 1990

-rostrum 5 °

3 °

V

(i)--~V1

---~

~t(ii~

(iii) ~

~

Vl

V?

"R "k

c 3" 5"

~

(iv)

]

f.~ l. M~de~k~r~e s`°quen~a~generati~n~f distir`ctmutant f~rms~f a V regi~ngenein a sing~eB ~e~and ~xati~n~f [he f~rrnthat ~nfe~ increaseda~inityf~r antigen.

hybridomas that were derived from the same B-cell progenitor and that express V regions that share all somatic mutations8-~° are often isolated from single mice during early secondary immune responses. Such an observation can most easily be explained if ~tis assumed that mutation is turned 'off' during the primary response and is then followed by an extended period of clonal expansien.

Stepwise af..'umulationof mutationsmusttake placeover a shorttime The picture ~hat is suggested by the above considerations is one of saltatory evolution, in which periods of clonal expansion are punctuated by a burst (or bursts) of mutation followed by selection 2°.2~.Such a process could explain the dramatic mutational 'founder effect', mentioned above, that is often observed among secondary hybridomas".9.1°.23.However, the number of cell divisions that could take place during the short period of time in which mutation is active would be limited. It is, therefore, questionable whether several mutations that independently confer increased affinity, can be sustained and fixed by selection in any one V region. In a situation where mutation, selection and cell division are mutually interdependent, the number of mutational affinity maturation steps that can be taken is limited by the number of divisions that occur. Assuming that the mutation process is somewhat random, a particular mutation that results in increased affini~ will take place with a frequency of approximately 10-3 per mutational event in a VHor VLcoding sequence. Given current estimates for the rate of mutation (10-3-10 -4 per base pair per cell division) ~,7.2°,only 103-I 0 ~divisions would De required to make it probable that this mutation is generated. However, in the case of three independent mutations in one V gene. 108-I0 ~ divisions (or 27-37 generations starting from a single cell) would be required. Given even the shortest current estimates of B-cell doubling time (6-7 hours for B cells in germinal centers24), producing this many divisions starting from a single cell would require at least one week. Another way to describe this concept is that the waiting time 2s necessary for a clone to combine a new mutation with an existing one would be several days. If it is assumed that the three mutations discussed above took place at mutational hotspots20.26 the number of divisions necessary to generate them in combination would be reduced. However, it is unlikely that mutations occurring at

306

'hotspots' in a given V gene would, in general, encode amino acid substitutions that resulted in affinity increases for the eliciting antigen.

A new model for the affinity maturation process It, therefore, seems appropriate to question the generally held notion that somatic mutation, antigen selection and B-cell division are mutually interdependent. The fundamental question is: if this notion is correct, can enough celY divisions and large enough clone sizes be attained to account for the efficiency of V region mutant selection observed during an immune response? Figure I presents an alternative model for the mutation-selection process that can be constructed with this reservation in mind. In this model, V gene mutation is assumed to occur independently of chromosomal DNA replication, and in only the lagging strand of a unidirectional, site-specific DNA replication event that duplicates the original V, but i, vr ', ,, n "~j~., /~-+~,.., / i ~ II,-, ~ _l l ,gJ tVJg n nqe ~ . . l l . g ~ ~111. I I +i., U I .~. . .L. .O ~ I ~ ' O l.~ V H O . .n. .t. . . t h , ~~ (~- r,-g , ~. loci, the intronic enhancer might be involved in initiation of such local DNA replication, as previously suggested 27. Only dupFcation of the lagging strand is assumed to be error prone, as a result of the distinct nature and number of enzymatic events necessaryfor replication of this strand relative to the leading strand. Leading strand replication proceeds 5' to 3' in an uninterrupted process..~ile lagging strand replication also takes place 5' to 3' ~ :t in a discontinuous manner, involving the generation ,if multiple RNA primers (that must be removed and replaced) and Okazaki DNA fragments of 100-300 bases in length (thaL must b~. ligated) 28. An alteration or absence of any of the enz]/matic components necessary for efficient lagging ~trand replication (e.g. DNA polymerase 13),perhaps simply due to the fact that the cell is not in S phase, could lead to a high error rate during replication of this strand. It should be noted that when mutations are incorporated into the newly synthesized lagging strand, they must be fixed in tne complementary RNA template strand by repair mechanisms if they are to be expressed at the level of mRNA. After replication, he mutant form of the V gene (V2) is expressed with the unmutated Cgene. As shown in Fig. I, the unmutated Vgene (VI) can not be expressed since its RNA template strand ~s not contiguous with the C gene template strand. The antibody partially encoded by V2 is assumed not to have sufficient affinity to induce chromosomal DNA replication and mitosis. In step (ii),

Immunology Today, Vol. 11, No. 9 1990

therefore, the V2 duplex is excised and degraded, and asymmetric error-prone replication and repair generates a qew mutated, coding Vgene (step (iii)). The differential recognition of the V2 strand by degradative nucleases could result from residual unrepaired base mismatches that were incor,t,;ated during step (i). Such replication-expression trials could take place many times (step (iv)) without loss of the germline Vl gene information. If, at some point, the mutated V gene partially encoded an antibody with the requisite affinity for antigen (the lower the antigen concentration the higher the requisite affinity), conventional chromosomal DNA replication would be induced in the cell, followed by mitosis, and the two DNA duplexes would be segregated to different daughter cells (step (v)). The entire process could then be repeated in these cells, in one case using the new mutant form of the Vgene. Presumably, this process would occur simultaneously at both VH and VL loci, resulting in the combination of new V, and VL mutations being sampled by selection during each trial. Alternatively VH and VL mutant selection may take place sequentially. In principle, similar results could be obtained in a situation in which mutation was confined to the RNA template strand of a normal DNA duplex, as has been previously suggested 2~. The wider context

The model presented in Fig. 1 can easily account for the high efficiency of the affinity maturation process. For example, it provides the means by which a single B cell could survive mutational events that destroy V region function, and would allow a large number of independent mutation-selectiort trials to take place in a single cell in a short time. The length of each trial would be limited mainly by the time necessary to convert Vregion genotype to cell surface phenotype - perhaps 30 minutes 29. This model is distinguished from prewous models that have assumed that cell division is not a requirement for mutation (for example, see Refs 3, 30 and 31) in proposing that mutations in a single Vgene wi!l not accumulate in the absence of antigenic selection. Thus, mutations that increase affinity, and neutral mutations, will selectively accumulate during the growth of a B-cell clone. Interestingly, a process of adaptive mutation that does not depend on chromosomal DNA replication has also recently been proposed to occur in bacteria 32. Clearly, more and better data concerning the mechaT~isms of V gene mutation and affinity-based antigen selection are required before the validity of the model presented here can be adequately assessed. Some support for the model can, nevertheless, be garnered from the observation that ~/gene mutation appears to be sustained with a bias for one of the DNA strands 2°.33- implying intrinsic asymmetry in this process. Finally, it should be mentioned that histological exami.~,ations of lymphoid organs during Lne immune response have implicated the germinal centers of B-cell follicles as a possible site in which the mutation and selection of antibody V regions take place 3~. The rapid divis;on of germinal center B cells, and the indirect evidence for B-cell death in this microenvironment have been cited in support of this idea 3s. The hypothesis presented here is by no means mutually exclusive with this view. Extensive cional proliferation and differentiation to secretory phenotype are clearly required for the large-scale expression of any V

region mutant, and a certain amount of cell death would be expected to result from 'mistakes' in the process shown in Fig. 1 (e.g. those leading to double strand chromosome breaks). If, however, the ,'ate of B-cell division limited the adaptability of the responding B-ce!i population to natural pathogens, this would constItute a distinct strategic disadvantage. Owing to their short generation time, often high mutation rate, or specialized mechanisms for genome alteration, many pathogens are capable of rapid antigenic variation during the course ef infection. I would like to thank Alar, Perelson, Marty Weigert, Larry Wysocki and the members of my !aboratory for insightful discussions. This work was supported by grants from the NIH (AI-23739) and the ACS (IM-557). T. Manser i~ a Pew Scholar in the Biomedical Scierlces. Reference:.

1 Allen, D., Cumano, A., Dildrop, R. eta/. (1987)Immunol. Rev. 96, 5-22 2 Berek, C. and Milstein, C. (1987) Immunol. Rev. 96, 23-42 3 Malipiero, U.V., Levy, N.S. and Gearhart, P.J. (1987) ImmunoL Rev. 96, 59-74 4 Manser, T., Wysocki, L.J., Margolies, M.N. and Gefter, M.L. (1987) ImmunoL Rev. 96, 141 .-162 Manser, T., Wysocki, L.J., Gridley, T., Near, R.I. and Gefter, M.L. (1985)Immunol. Today 6, 94-101 6 Manser, T.and Gefter, M.L. (1986) Eur. J. ImmunoL 16, 1439-1444 7 McKean, D., Huppi, K., Bell, M. etal. (1984) Proc. Natl Acad. Sci. USA 81,3 ! 80-3184 8 Siekevitz, M., Kocks, C., Rajewsky, K. and Dildrop, R. (1987) Cell 48, 757-770 9 Claflin, Ji., Berry, J., Ftaherty, D. and Dunnick, W. (1987) J. ImmunoL 138, 3060-3068 10 Fish, S., Zenowich, E., Fleming, M. and Manser, T. (1989) J. Exp. Med. 170. 1191-1209 11 Kim, S., Davis, M., Sinn, E., Patten, P. and Hood, L. (1981) Cell 27,573-581 12. Gearhaff, P.J. and Bogenhagen, D.F. (1983) Proc. Natl Acad. Sci. USA 80, 3439-3443 13 Reynaud, C-A., Dahan, A., Anquez, V. and Weill, J.C. (1989) Ce1159, 171--183 14 Chien, N.C., Pollock, R.R., Desaymard, C. and Scharff, M.D. (19~8) J. Exp. Med. 167, 954-973 15 Wysocki, L. and Gefter, M.L. (1989)Annu. Rev. Biochem. 58, 509-529 16 Sharon, J., Gefter, M.L., Wysocki, L.J. and Margolies, M.N. (1989) J. ImmunoL 142, 596-601 17 Kochs, C. and Rajewsky, K. (198S) Proc. NatiAcad. Sci. USA 85, 8206-8210 18 Wysocki, L.J., Gefter, M.L. and Mar§olies, M.N.J. Exp. Med. (in press) 19 Manser, T. (1989)J. Exp. Med. 170, 1211-1230 20 Berek, C. and Milstein, C. (1588) ImmunoL Rev. 105, 5-26 21 Levy, N.S., Malipiero, U.V., Lebecque, S.G. and Gearhart, P.J. (1989) J. Exp. Med 169, 2007-2019 22 Dell, C.L., Lu, Y. and Claflin, J.L. (1989)J. Immunol. 143, 3364-3370 23 Blier, P.R. and Bothwell, A. (1987)J. Irnmunol. 139, 3996-4006 24 Zhang, J., MacLennan, I.C.M., Liu, Y-J. and Lane, P.J.L. (1988) Immunol. Letts 18, 297-300 25 Kauffman, S.A., Weinberger, E.D. and Perelson, A.S. (1988) in Theoretical Immunology, Part I (SFI Studies in the Science of Complexity), (Perelson, A.S., ed.), pp. 349-382, Addison-Wesley Publishing Company 26 Levy, S., Mendel, E., Kon, E. et al. (1988)J. EXP. Med. t68, 475-489 3O7

Immunology Today, VoL 11, No. 9 1990

-rostrum, 2? bothwell, A L M (19~4) in The Biology of Idictypes (M Greene and A Msoqoff, eds), pp 19-34, Plenum Publishing Co 28 Kornberg, A. (1.°.82)DNA ReplicationW H. Freemanand Company 2~ Watts, C. and Davidson, H.W. (1988) EMBOJ. 7, 1937-1945 Brenner, S. and Milstein, C. (1966)Nature 211,242-243 . . ~ . . . 31 Steele,EJ. and Pollard,J.W. (1987)Mol. ImmunoL 24,

Haman .

.

.

.

667-673 32 Cairns,J., Overbaugh, J. and Miller, S. (1988) Nature 335, 142-!45 33 Manser,T. in Somatic Hypermutation in V-Regions(E.J. Steele, ed.)oCRC Press(in press) 34 Heinen. E., Cormann, N. and Kinet-Denoel,C. (1988) ImmunoL Today 9, 240-243 35 MacLennan,I.C.M. and Gray, D. (1986) ImmunoL Rev. 91, 61-86

Fc /receptor I1: a standby receptor activated by proteolys's?

Three cl~es of Fc receptor for IgG have been defined: Fq,RI, Fc~IRiland FolRIII. Fe,/RIIis the most widely distributed, but it has a low affinity for human IgG and its function is not well understood. In this and the following paper, two possible functions of FoiRIf are presented. Here, Wil Taxand Jan van de Winkel argue that F~R!I ,'nay contribute significant.~ to FeyRmediated processes,such as the removal of immune complexes and antibody-coated particles, especiallyin view of the recent finding that the affinity of Fe~.R!Ican be upr~ulated by proteolys& The following paper considers the role of F~RII in the context of immune complex regulation of the immune system.

Receptors for the Fc moiety of IgG, Fc~/R, play a central role in a variety of immune defense processes. These include pinocytosis of immune complexes, phagocytosis of antibody-coated panicles and antibody-dependent cell-mediated cytotoxicity (ADCC). Furthermore, Fc~/R control the release of several mediators of inflammation. In recent years, the availability of monoclonal antibodies against the different Fc~/R has contributed greatly to progress in the identification and characterization of these receptors. Three classes of human Fc~/Rhave been defined, both at protein and complementary DNA (cDNA) level: Fc~/RI(CD64), Fc~R~!(C'732)ar,,~l m:c~RIII(CD16).

Expressionand affinity of huma~ Fc~/R The three classes of Fc~ receptors are differentially expressed on different cell ~jpes (for general reviews see Refs 1--4). Fc-yRI is constitutively expressed on human monocytes and macrophages. Gamma-interferon (IFN-~/) induces a 5-10-fold increase in the number of Fc~/RIon these cells and also induces de novo expression of Fc~/RI on neutrophils4. This receptor has a high affinity for monomeric human IgG (Ka = 108-109 M-1) Fc~/RIIand Fc~/RIII,in contrast, are low-affinity receptors that can only interact well with IgG complexes or IgGcoated particles. The difference in affinity for human IgG between Fc-yRIand Fc~/RIIappears to be large. Whereas Fc~/RI-positive cells readily bind erythrocytes sensitized with human IgG against the rhesus D antigen, no such

'Depa~mentof M~icine, Divisionof Nephrology,UniversityHospital Nijmegen,6525GANijmegen,TheNetherlands,2Departmentof Experimental Immunology, Universityof Utrecht, 3584 CX Utrecht, The Netherlands. Currentaddress: Ohio State UniversityCollegeof Medicine, 480 Westgt,5Avenue,Rm$2092, Columbus,0H43210, USA. 308

'

Wil J.M. Tax and Jan G.J. van de Winkelz binding is observed using the cell line K562, which expresses only Fc~/Ril (Ref. 5). Platelets, which also express only Fc-yRII, have a lO0-fold lower avidity for IgG dimers than monocytes6. Fc~/RIIis expressed on a large variety of cell types including monocytes, macrophages, neutrophils, basophils, eosinophils, B lymphocytes, platelets, endothelial cells and placental trophoblasts 7. Fc~RIII can occur in two different forms. As the predominant Fc~/R of neutrophils, Fc~/RIII is attached to the membrane by a phosphatidylinositol glycan linkage (Fc~/RIII-PI)8.9, while on macrophages and natural killer (NK) cells it is expressed as a transmembrane molecule (Fc~/RIII-TM).

Polymorphismof Fc~RII A genetically determined polymorphism of Fc~/RIIwas discovered in studies on the interaction between human monoq/tes and mouse IgG! antibodiesS,l°-12: monocytes from 70% of normal Caucasian individuals (high responders) exhibit strong interaction with mouse IgG1 whereas 30% (low responders) do not. The structural polymorphism of Fc-~RIIis reflected in isoelectric focussing patterns of purified receptor molecules 13 and in differential binding of the monoclonal antibody 41.1416 (Ref. 14). Differences between the high and low responder forms of this receptor have also been found at the cDNA levePs. Functional capacitiesof Fc~/R Although much is known about the interaction of human Fc~/RIIwith mouse IgG 1, the physiological role of the receptor is poorly understood. Since Fc~/RIhas a high affinity for human IgG, and its expression is strongly increased by IFN-~/, it has always appeared logical to postulate that this receptor plays a central role in phagocytosis and ADCC. Fc~/Riiand Fc~/Riii,however, can, under appropriate conditions, perform mo~t or all of the functions of Fc~/RI (Ref. 7). In ADCC studies using hybridoma cell lines bearing surface antibody against one of the Fc~,Ras targets, it could be demonstrated that all three Fc~/R classes could mediate cytotoxicity (although with respect to Fc~/RIII, only Fc~/RIII-TM was active in this © 1990, Elsev;erScience Publishers Ltd, UK. 0167--4919/90/$02.00