Portal de Periódicos da CAPES

Life and Death in Germinal Centers (Redux)

1996; Cell Press; Volume: 4; Issue: 2 Linguagem: Inglês

10.1016/s1074-7613(00)80675-5

1097-4180

Garnett Kelsoe,

Immunotherapy and Immune Responses

Curiously enough, B cell responses to thymus-dependent antigens begin in the T cell zones of secondary lymphoid tissues, where T and B cells initiate antigen- and costimulus-dependent proliferation. These initial cognate interactions are essential for humoral immunity, but alone result only transient low affinity antibody responses (31Klaus G.B.B Humphrey J.H Kunkel A Dongworth D.W Immunol. Rev. 1980; 53: 3-28Crossref PubMed Scopus (351) Google Scholar, 11Coico R.F Boghal B.S Thorbecke G.J J. Immunol. 1983; 131: 2254-2257PubMed Google Scholar, 18Han S Hathcock K Zheng B Kepler T Hodes R Kelsoe G J. Immunol. 1995; 155: 556-567PubMed Google Scholar). It is a subsequent set of cellular encounters, collectively known as the germinal center (GC) reaction, that drives affinity maturation by V(D)J hypermutation, B cell memory, and the continued self-tolerance of lymphocytes bearing mutated antigen receptor molecules. The GC is a complex cellular microenvironment that supports and directs post-V(D)J diversification and selection. In the GC, T and B lymphocytes balance precariously between receptor-driven activation and apoptotic death, processes common in primary lymphoid tissues. Increasingly, evidence suggests that the developmentally regulated proliferation and selection of newly generated lymphocytes in bone marrow and thymus may be mirrored in GCs. Here, I shall briefly describe the GC reaction in mice, focusing on the possibility and significance of T cell hypermutation, cellular selection, and the maintenance of self-tolerance in the periphery. I also note characteristics of the GC reaction that appear analogous to events in thymus, bone marrow, and the gut-associated follicles that support ontogenic diversification of rearranged immunoglobulin genes. Histologists prosper by finding and naming novel microanatomies: the relationship is not unlike that between biochemists and tyrosine kinases. Unfortunately, this process has resulted in an unwieldy jargon that can confuse the uninitiated. In this outline of the genesis and loss of splenic GCs (Figure 1), I provide only the names of prominent and important histologic structures. Those wishing for a fuller naming of parts are referred to recent fulsome reviews on the subject (35MacLennan I.C.M Annu. Rev. Immunol. 1994; 12: 117-139Crossref PubMed Scopus (1642) Google Scholar, 27Kelsoe G Adv. Immunol. 1995; 60: 267-288Crossref PubMed Scopus (131) Google Scholar). After their activation in T cell areas (in the splenic white pulp this is the periarteriolar lymphoid sheath [PALS]), selected T and B lymphocytes migrate into the lymphoid, or primary, follicles. There, they accumulate within the extensive processes of follicular dendritic cells (FDCs) that form a scattered network within the B cell–rich follicle. FDCs retain antigen–antibody or antigen–complement complexes on their surface for long periods and act as depots for the antigen that sustains the GC reaction (43Nossal G.J.V Ada G.L Austin C.M Aust. J. Exp. Biol. Med. Sci. 1964; 42: 311-330Crossref PubMed Scopus (70) Google Scholar, 61Tew J.G Phipps R.P Mandel T.E Immunol. Rev. 1990; 117: 185-211Crossref PubMed Scopus (284) Google Scholar, 56Schriever F Nadler L.M Adv. Immunol. 1992; 51: 243-284Crossref PubMed Scopus (70) Google Scholar). The immigrant lymphocytes rapidly proliferate, filling the FDC reticulum and acquiring novel phenotypic characteristics. For example, GC B cells avidly bind peanut agglutinin (PNA) and become positive for the activation marker recognized by the GL-7 monoclonal antibody (41Miller C Kelsoe G Han S Aging Immunol.Infect. Dis. 1994; 5: 556-559Google Scholar), and most GC T cells down-regulate expression of Thy-1 (20Harriman G.R Lycke N.Y Elwood N.J Strober W J. Immunol. 1990; 145: 2406-2414PubMed Google Scholar). This early phase of the GC reaction compresses the surrounding uninvolved follicular cells to form a mantle zone about the new GC, or secondary follicle. Shortly after this initial period of expansion, the GC polarizes to form a dark zone (DZ) proximal to the T cell area that contains rapidly dividing immunoglobulin-negative B cells called centroblasts and a distal light zone (LZ) that contains nondividing immunoglobulin-positive centrocytes, the bulk of the FDC network, and most of the helper T cells present in GCs. Cell labeling studies suggest that centroblasts proliferate rapidly, dividing every 6–7 hr and demonstrate that the centrocyte population is continuously derived from cells in the DZ (34Liu Y.-J Lane P.J.L Chan E.Y.T MacLennan I.C.M Eur. J. Immunol. 1991; 21: 2951-2963Crossref PubMed Scopus (591) Google Scholar). Recent evidence (19Han S Zheng B Dal Porto J Kelsoe G J. Exp. Med. 1995; 182: 1635-1644Crossref PubMed Scopus (195) Google Scholar) suggests that, in turn, LZ centrocytes reenter the DZ, join the centroblast population, and reinitiate proliferation. Interestingly, these cyclic migrations may help explain the rapid tempo of selection for high affinity B cells into the memory compartment (30Kepler T.B Perelson A Immunol. Today. 1994; 14: 412-415Abstract Full Text PDF Scopus (206) Google Scholar). Importantly, newly formed GCs represent oligoclonal B cell populations (23Jacob J Kassir R Kelsoe G J. Exp. Med. 1991; 173: 165-1175Crossref Scopus (583) Google Scholar, 24Jacob J Kelsoe G Rajewsky K Weiss U Nature. 1991; 354: 389-392Crossref PubMed Scopus (870) Google Scholar, 34Liu Y.-J Lane P.J.L Chan E.Y.T MacLennan I.C.M Eur. J. Immunol. 1991; 21: 2951-2963Crossref PubMed Scopus (591) Google Scholar). On average, each mature GC is derived from only 1–3 B cell clones that survive a dramatic reduction in clonal diversity that precedes the onset of significant V(D)J hypermutation (25Jacob J Przylepa J Miller C Kelsoe G J. Exp. Med. 1993; 178: 1293-1307Crossref PubMed Scopus (326) Google Scholar, 39McHeyzer-Williams M.G McLean M.J Lalor P Nossal G.J.V J. Exp. Med. 1993; 178: 295-307Crossref PubMed Scopus (265) Google Scholar). The GC reaction reaches its maximum by day 10–12 of primary responses, accounting for as much as 1%–3% of the total splenic volume. Without further antigenic stimulation, GCs wane by 21 days postimmunization, losing volume and avidity for PNA (34Liu Y.-J Lane P.J.L Chan E.Y.T MacLennan I.C.M Eur. J. Immunol. 1991; 21: 2951-2963Crossref PubMed Scopus (591) Google Scholar). By 32 days after immunization, GC residua occupy <5% of their peak volume and appear as infrequent collections of a few antigen-binding blast cells in association with FDCs. Outside its coterie of enthusiasts, the GC is best known as the site of antigen-driven V(D)J hypermutation and selection (24Jacob J Kelsoe G Rajewsky K Weiss U Nature. 1991; 354: 389-392Crossref PubMed Scopus (870) Google Scholar, 6Berek C Berger A Apel M Cell. 1991; 67: 1121-1129Abstract Full Text PDF PubMed Scopus (730) Google Scholar). Within GCs, antigen-specific B cells acquire (mostly) point mutations in the V regions of transcriptionally active rearranged immunoglobulin genes. Mutated immunoglobulin genes are first observed on day 7–10 of primary responses, coincident with the polarization of the GC and the expression of CD86 on centrocytes (18Han S Hathcock K Zheng B Kepler T Hodes R Kelsoe G J. Immunol. 1995; 155: 556-567PubMed Google Scholar). Mutations accumulate steadily at least until day 18 of the response by the step-wise introduction of 1–3 nucleotide substitutions, resulting in clonal genealogies that recapitulate the repeated rounds of intraclonal mutation, selection, and proliferation that take place in GCs (10Clarke S.H Huppi K Ruezinsky D Staudt L Gerhard W Weigert M J. Exp. Med. 1985; 161: 687-704Crossref PubMed Scopus (230) Google Scholar, 24Jacob J Kelsoe G Rajewsky K Weiss U Nature. 1991; 354: 389-392Crossref PubMed Scopus (870) Google Scholar, 25Jacob J Przylepa J Miller C Kelsoe G J. Exp. Med. 1993; 178: 1293-1307Crossref PubMed Scopus (326) Google Scholar, 39McHeyzer-Williams M.G McLean M.J Lalor P Nossal G.J.V J. Exp. Med. 1993; 178: 295-307Crossref PubMed Scopus (265) Google Scholar). Clonal evolution proceeds independently in each GC, as there is little or no B cell trafficking between GCs (22Jacob J Kelsoe G J. Exp. Med. 1992; 176: 675-687Crossref Scopus (392) Google Scholar), and in the absence of significant convergent selection driven by circulating antibody (63Vora K.A Mauser T J. Exp. Med. 1995; 181: 271-281Crossref PubMed Scopus (30) Google Scholar). Thus, each GC represents a local fitness optimum, one island in an archipelago of selected clones. Where and when interclonal competition among memory B lymphocytes takes place remains an important unanswered question. The mechanism of V(D)J hypermutation is unknown but it introduces a distinctive pattern of nucleotide misincorporations. Characteristically, hypermutation favors transition mutations and exhibits biased nucleotide exchange and strand polarity (15Golding G.B Gerhart P.J Glickman B.W Genetics. 1987; 115: 169-176Crossref PubMed Google Scholar, 9Both G.W Taylor L Pollard J.W Steele E.J Mol. Cell. Biol. 1990; 10: 5187-5196Crossref PubMed Scopus (120) Google Scholar, 65Weber J.S Berry J Manser T Claflin J.F.L J. Immunol. 1991; 146: 3218-3226PubMed Google Scholar; Gearhart and Levy, 1991; 25Jacob J Przylepa J Miller C Kelsoe G J. Exp. Med. 1993; 178: 1293-1307Crossref PubMed Scopus (326) Google Scholar, 7Betz A.G Neuberger M.S Milstein C Immunol. Today. 1993; 14 (_411): 405Abstract Full Text PDF PubMed Scopus (12) Google Scholar, 8Betz A.G Rada C Pannell R Milstein C Neuberger M.S Proc. Natl. Acad. Sci. USA. 1993; 90: 2385-2388Crossref PubMed Scopus (251) Google Scholar, 45Pascual V Liu Y.-J Magalski A de Bouteiller O Banchereau J Capra J.D J. Exp. Med. 1994; 180: 329-339Crossref PubMed Scopus (573) Google Scholar). Also, short sequence motifs have been identified that are intrinsic mutational hotspots within both human and murine VH and VL exons (54Rogozin I.B Kolchanov N.A Biochem. Biophys. Acta. 1992; 1171: 11-18PubMed Google Scholar, 7Betz A.G Neuberger M.S Milstein C Immunol. Today. 1993; 14 (_411): 405Abstract Full Text PDF PubMed Scopus (12) Google Scholar). Surprisingly, recent transgenic "switch-and-bait" experiments (2Azuma T Motoyama N Fields L Loh D Int. Immunol. 1993; 5: 121-130Crossref PubMed Scopus (72) Google Scholar, 8Betz A.G Rada C Pannell R Milstein C Neuberger M.S Proc. Natl. Acad. Sci. USA. 1993; 90: 2385-2388Crossref PubMed Scopus (251) Google Scholar, 71Yélamos J Klix N Goyenechea B Lozano F Chui Y.L Gonzalez-Fernandez A Pannell R Neuberger M.S Milstein C Nature. 1995; 376: 225-229Crossref PubMed Scopus (206) Google Scholar) have revealed that hypermutation can act on a variety of DNA substrates, including prokaryotic genes and Vκ genes driven by nonimmunoglobulin promoters. In fact, the only cis element required for V(D)J mutation in the well-studied Lκ transgene was the intronic enhancer/matrix attachment region (κi/MAR). Deletion of κi/MAR abrogated mutation in Peyer's patch B cells even though the expression of the transgene remained high in hybridomas, owing to the presence of the κ 3′ enhancer (8Betz A.G Rada C Pannell R Milstein C Neuberger M.S Proc. Natl. Acad. Sci. USA. 1993; 90: 2385-2388Crossref PubMed Scopus (251) Google Scholar). Mounting evidence indicates an important, or even necessary, role for transcription and/or transcription-linked DNA repair in V(D)J hypermutation (60Storb U Curr. Opin. Immunol., in press. 1996; Google Scholar). For example, in contrast with VH and Vκ, unrearranged Vλ gene segments can undergo hypermutation (67Weiss S Wu G EMBO. J. 1987; 6: 927-932PubMed Google Scholar, 42Motoyama N Okada H Azuma T Proc. Natl. Acad. Sci. USA. 1991; 88: 7933-7937Crossref PubMed Scopus (57) Google Scholar, 57Selsing E Storb U Cell. 1981; 25: 47-58Abstract Full Text PDF PubMed Scopus (113) Google Scholar, 16Gorski J Science. 1983; 220: 1179-1181Crossref PubMed Scopus (36) Google Scholar); unrearranged Vλ genes are transcribed in B cells, while Vκ genes are not (46Picard D Schaffner W EMBO J. 1984; 3: 3031-3035Crossref PubMed Scopus (22) Google Scholar, 37Mather E Perry R Nucl. Acids Res. 1981; 9: 6855-6867Crossref PubMed Scopus (44) Google Scholar). Likewise, the characteristics of V(D)J hypermutation, including strand bias, absence of mutation upstream of the transcriptional start site (9Both G.W Taylor L Pollard J.W Steele E.J Mol. Cell. Biol. 1990; 10: 5187-5196Crossref PubMed Scopus (120) Google Scholar; Gearhart and Levy, 1991; 50Rada C Gonzalez-Fernandez A Jarvis J.M Milstein C Eur. J. Immunol. 1994; 24: 1453-1457Crossref PubMed Scopus (86) Google Scholar, 53Rogerson B Mol. Immunol. 1994; 31: 83-98Crossref PubMed Scopus (64) Google Scholar), and dependence upon transcriptional enhancers (8Betz A.G Rada C Pannell R Milstein C Neuberger M.S Proc. Natl. Acad. Sci. USA. 1993; 90: 2385-2388Crossref PubMed Scopus (251) Google Scholar), suggest an association between hypermutation and transcription. Perhaps the strongest evidence for transcription-linked V(D)J hypermutation comes from a recent study (47Peters A Storb U Immunity. 1996; 4: 57-65Abstract Full Text Full Text PDF PubMed Scopus (360) Google Scholar) demonstrating that introduction of a redundant transcriptional promoter into the J–C intron of a κ transgene promoted a sharp increase in the frequency of mutations in Cκ. Thus, the modified transgene supported two distinct tracts of mutation, one focused over VJ and the other over the C region; the origins of both tracts were precisely coincident with transcription initiation. As the immune response progresses, evidence for phenotypic selection in GC B cells becomes increasingly obvious. Early GCs contain as many as 10 unique VDJ joint sequences, indicating colonization by at least 10 B cell clones; by day 8 of the response, CDR3 diversity in GCs falls to only 1.5 distinct VDJ sequences in each GC. This reduction in diversity is largely due to the loss of B cells expressing rearrangements that encode antigen-specific but low affinity antibody. Selection in GCs can also be inferred from the V(D)J mutations within GC B cell populations. Initially, mutations are uniformly distributed within VH/L exons but with time become focused within the complementarity-determining regions (CDRs). The proportion of VDJ rearrangements containing crippling mutations, e.g., misincorporations leading to termination codons or replacements at invariant amino acid residues, can be as high as 33% of total mutations on day 8 of a primary response but falls to less than 3% by day 14 (25Jacob J Przylepa J Miller C Kelsoe G J. Exp. Med. 1993; 178: 1293-1307Crossref PubMed Scopus (326) Google Scholar). Ratios of replacement:silent (R:S) mutations initially approximate random values but become biased towards S mutations in framework regions and R mutations in CDRs. The frequency of mutations that confer high affinity becomes enriched in GCs, presumably reflecting their selection (reviewed by5Berek C Ziegner M Immunol. Today. 1993; 14: 400-404Abstract Full Text PDF PubMed Scopus (181) Google Scholar). Although the GC is widely accepted as the histologic site for immunoglobulin hypermutation, a sequence analysis of Vα rearrangements indicates that the T cell receptor (TCR) may also mutate there (72Zheng B Xue W Kelsoe G Nature. 1994; 372: 556-559Crossref PubMed Scopus (121) Google Scholar). It must be noted (and has been, sometimes pointedly) that this finding is controversial (3Bachl J Wabl M Nature. 1995; 375: 285-286Crossref PubMed Scopus (4) Google Scholar, 38McHeyzer-Williams M.G Davis M.M Science. 1995; 268: 106-110Crossref PubMed Scopus (399) Google Scholar). T cell responses to pigeon cytochrome c (PCC) are dominated by clones bearing Vα11+ Vβ3+ TCR in mice expressing I-Ek (70Winoto A Urban J.L Lan N.C Goverman J Hood L Hansberg D Nature. 1986; 324: 679-682Crossref PubMed Scopus (184) Google Scholar). After intraperitoneal immunization with haptenated PCC, the number of CD4+ Vα11+ Vβ3+ cells increases dramatically in the spleen, first in the PALS and later in GCs. GC T cells probably represent immigrants from the PALS, as both populations often share complex junctional sequences. It is considerably surprising that Vα11, but not Vβ3, rearrangements recovered from CD4+ GC cells contained mutations far in excess of that expected from PCR errors (72Zheng B Xue W Kelsoe G Nature. 1994; 372: 556-559Crossref PubMed Scopus (121) Google Scholar, 29Kelsoe G Zheng B Kepler T Nature. 1995; 375: 286Crossref Scopus (2) Google Scholar). Mutations were not observed in the nearby clonally related PALS T cells; TCR mutations in GC T cells were confined to the V region of the α chain, exhibited the biased nucleotide substitutions and DNA strand polarity characteristic of immunoglobulin hypermutation, and while the observed mutations differed significantly from that expected for unselected meiotic point mutations, they were indistinguishable from mutations flanking hypermutated immunoglobulin genes (72Zheng B Xue W Kelsoe G Nature. 1994; 372: 556-559Crossref PubMed Scopus (121) Google Scholar). In contrast with immunoglobulin hypermutation, the majority of TCR mutations were recovered in out-of-frame rearrangements. This finding was interpreted as evidence for strong negative selection against GC T cells expressing mutant TCRs. Indeed, CD4+ cells containing the fragmented DNA characteristic of apoptosis (TUNEL+) (14Gavrieli Y Sherman Y Ben-Sasson S.A J. Cell Biol. 1992; 119: 493-501Crossref PubMed Scopus (8981) Google Scholar) are frequent in GCs. Further, recovery of TCR α rearrangements from single GC T cells is consistent with the notion of selection against TCR mutants; more than one-half of 16 VαJα rearrangements from TUNEL+ CD4+ cells dissected from GCs contained point mutations; most (≈ 90%), associated with in-frame VJ joints. In contrast, mutations observed in single TUNEL− Vα11+ cells from the same sites (n = 25) were less common (≈ 10% of sequences) and equally distributed among productive and out-of-frame rearrangements. This selection and confinement of TCR mutations to the small pool of GC T cells may explain why a recent flow cytometric study (38McHeyzer-Williams M.G Davis M.M Science. 1995; 268: 106-110Crossref PubMed Scopus (399) Google Scholar) found no mutated Vα11 rearrangements in PCC-specific T cells recovered 6 days after primary (n = 9) and secondary (n = 9) immunizations. Nonetheless, these observations are likely to prove unconvincing if TCR mutations can not be recovered from antigen-specific T cell lines or hybridomas. Mutational diversification of V(D)J genes within GCs must occasionally produce antigen-receptors that acquire new specificities; a fraction of these may be autoreactive and potentially detrimental (12Diamond B Scharff M.D Proc. Natl. Acad. Sci. USA. 1994; 81: 5841-5844Crossref Scopus (269) Google Scholar, 58Shlomchik M.J Marshak-Rothstein A Wolfowicz C.B Rothstein T.L Weigert M.G Nature. 1987; 328: 805-811Crossref PubMed Scopus (579) Google Scholar). Diversification of TCRs in the absence of thymic censoring would be especially problematic. However, the GC may maintain tolerance to self by the elimination of autoreactive lymphocytes. Several groups have developed experimental models to study the fate of GC B cells that acquire specificity for self-antigens present in the GC (49Pulendran B Kannourakis G Nouri S Smith K.G.C Nossal G.J.V Nature. 1995; 375: 331-334Crossref PubMed Scopus (276) Google Scholar, 59Shokat K.M Goodnow C.C Nature. 1995; 375: 334-338Crossref PubMed Scopus (334) Google Scholar, 19Han S Zheng B Dal Porto J Kelsoe G J. Exp. Med. 1995; 182: 1635-1644Crossref PubMed Scopus (195) Google Scholar). Injection of soluble antigen into recently immunized mice induces massive and rapid apoptosis in GCs but not elsewhere in the spleen, including extrafollicular sites of antigen-specific T and B cell proliferation. Cell death peaks by 5–8 hr postinjection and apoptosis is most obvious in the GC LZ, suggesting that the immunoglobulin-negative centroblasts are unaffected. Apoptosis is antigen specific, dose dependent, and sensitive to hapten density, suggesting that cell death may be mediated directly by surface immunoglobulin engagement. This conclusion is supported by VH sequence analysis of GC B cells that resist prolonged administration of soluble antigen. These B cells contain typically mutated VH rearrangements but with a distinctive spectrum of R mutations that indicates selection for decreased affinity for antigen (19Han S Zheng B Dal Porto J Kelsoe G J. Exp. Med. 1995; 182: 1635-1644Crossref PubMed Scopus (195) Google Scholar). The effects of soluble antigen do not stem from interference in cognate T–B collaboration nor are they the product of costimulatory disruption. Immune mice injected with anti-CD40L antibody instead of soluble antigen exhibited GC apoptosis only slightly above background levels (19Han S Zheng B Dal Porto J Kelsoe G J. Exp. Med. 1995; 182: 1635-1644Crossref PubMed Scopus (195) Google Scholar). Analogous experiments performed in C57BL/6 mice congenic for the lpr locus demonstrated that antigen-induced GC apoptosis was independent of the Fas pathway of programmed cell death (64Watanabe-Futuraga R Brannan C.I Copeland N.G Jenkins N.A Nagata S Nature. 1992; 356: 314-317Crossref PubMed Scopus (2661) Google Scholar, 1Adachi M.R Proc. Natl. Acad. Sci. USA. 1993; 90: 1756-1760Crossref PubMed Scopus (489) Google Scholar). Taken together, these observations support the notion that antigen causes GC B cell death directly by cross-linking surface immunoglobulin or by preventing centrocytes from interacting with the FDC. Other recent work on GCs (44Nossal G.J.V Karvelas M Pulendran B Proc. Natl. Acad. Sci. USA. 1993; 90: 3088-3092Crossref PubMed Scopus (40) Google Scholar, 48Pulendran B Karvelas M Nossal G.J.V Proc. Natl. Acad. Sci. USA. 1994; 91: 2639-2643Crossref PubMed Scopus (28) Google Scholar) has demonstrated another form of immunological tolerance mediated through the absence of T cell help. Interestingly, although this tolerance mechanism is clearly T dependent and distinct from that described above, it also affects only follicular responses, i.e., GCs but not PALS-associated plasmacytes. Thus, it is possible that GC cells are uniquely sensitive to tolerance induction, as if recapitulating that susceptibility observed in primary lymphopoiesis. Indeed, GC centrocytes express a collection of surface markers typically absent or weakly expressed on mature peripheral B lymphocytes but abundant on the immature/transitional B cells in bone marrow. For example, HSA, Fas, and GL-7 are commonly expressed on bone marrow B cell populations, but in the periphery are abundant only in GCs. Reciprocally, Bcl-2 is present in follicular lymphocytes but not in GC cells or immature B cells. In certain species, ontogenic diversification of immunoglobulin V region genes continues after V(D)J recombination. In chicken (51Reynaud C Anquez V Dahan A Weill J.C Cell. 1985; 40: 283-291Abstract Full Text PDF PubMed Scopus (233) Google Scholar), rabbits (4Becker R Knight K Cell. 1990; 63: 987-997Abstract Full Text PDF PubMed Scopus (222) Google Scholar, 66Weinstein P Anderson A.O Mage R.G Immunity. 1994; 1: 647-659Abstract Full Text PDF PubMed Scopus (130) Google Scholar), and sheep (52Reynaud C Garcia C Hein W.R Weill J.C Cell. 1995; 80: 115-125Abstract Full Text PDF PubMed Scopus (266) Google Scholar), B cells migrate from regions of primary lymphopoiesis, colonize epithelial crypts along gut mucosae, and proliferate to form prominent lymphoid follicles that share many histologic features of GCs. Here, B cells undergo postrearrangement V(D)J diversification by gene conversion (chickens), hypermutation (sheep), or both (rabbits). Even though birds, rabbits, and sheep represent widely divergent taxa, the many conserved features of this process imply a single underlying mechanism. It is unlikely that such a complex developmental pathway for postrearrangement diversification of the immunoglobulin repertoire arose independently three times during evolution (27Kelsoe G Adv. Immunol. 1995; 60: 267-288Crossref PubMed Scopus (131) Google Scholar, 36Maizels N Cell. 1995; 83: 9-12Abstract Full Text PDF PubMed Scopus (105) Google Scholar). Reynaud (1995) and colleagues have demonstrated that the spectrum of mutations introduced into the V regions of Peyer's patch B cells in fetal lambs is similar to the mutations observed in passenger κ transgenes in mice (7Betz A.G Neuberger M.S Milstein C Immunol. Today. 1993; 14 (_411): 405Abstract Full Text PDF PubMed Scopus (12) Google Scholar, 8Betz A.G Rada C Pannell R Milstein C Neuberger M.S Proc. Natl. Acad. Sci. USA. 1993; 90: 2385-2388Crossref PubMed Scopus (251) Google Scholar). This finding links antigen-driven and developmental V(D)J diversification, raising the possibility that the GC is homologous to the gut-associated follicular microenvironments. In contrast with V(D)J hypermutation (69Wilson M Hsu E Marcuz A Courtet M DuPasquier L Steinberg C EMBO J. 1992; 11: 4337-4347PubMed Google Scholar, 21Hinds-Frey K Nishikato H Litman R.T Litman G.W J. Exp. Med. 1993; 178: 815-824Crossref PubMed Scopus (115) Google Scholar, 17Greenberg A.S Arila D Hughes M Hughes A McKinney E.C Flajnik M.F Nature. 1995; 374: 168-175Crossref PubMed Scopus (464) Google Scholar), the GC reaction is not thought to be present in cold-blooded vertebrates (32Kroese F.G.M Leceta J Döpp E.A Herraez M.P Nieuwenhuis P Zapata A Dev. Comp. Immunol. 1985; 9: 641-652Crossref PubMed Scopus (16) Google Scholar). 69Wilson M Hsu E Marcuz A Courtet M DuPasquier L Steinberg C EMBO J. 1992; 11: 4337-4347PubMed Google Scholar note that affinity maturation is also absent in lower vertebrates and have suggested that GCs may represent an evolutionary adaptation to link V(D)J hypermutation to antigen-specific immune responses. If so, hypermutation might not require the GC microenvironment but be catalyzed by it. Could the mammalian GC reaction have evolved from the earlier gut-associated follicles? It may be time for renewed interest in immunology's zoo. I am grateful to K. Hathcock, R. Hodes, A. Peters, and V. Storb for sharing unpublished results. This work was supported in part by United States Public Health Service grants AI24335 and AG10207.