Abstract
In the first week of the primary immune response to the (4-hydroxy-3-nitrophenyl)acetyl (NP) hapten, plasmacytic foci and germinal centers (GCs) in C57BL/6 mice are comprised of polyclonal populations of B lymphocytes bearing the λ1 L-chain (λ1+). The Ig H-chains of these early populations of B cells are encoded by a variety of VH and D exons undiversified by hypermutation while later, oligoclonal populations are dominated by mutated rearrangements of the VH186.2 and DFL16.1 gene segments. To assess directly Ab affinities within these defined splenic microenvironments, representative VDJ rearrangements were recovered from B cells participating in the early immune response to NP, inserted into Ig H-chain expression cassettes, and transfected into J558L (H−; λ1+) myeloma cells. These transfectoma Abs expressed a remarkably wide range of measured affinities (Ka = 5 × 104-1.3 × 106 M−1) for NP. VDJs recovered from both foci and early GCs generated comparable affinities, suggesting that initial differentiation into these compartments occurs stochastically. We conclude that Ag normally activates B cells bearing an unexpectedly wide spectrum of Ab affinities and that this initial, promiscuous clonal activation is followed by affinity-driven competition to determine survival and clonal expansion within GCs and entry into the memory and bone marrow plasmacyte compartments.
Mature splenic B cells can differentiate along two distinct pathways in response to thymus-dependent Ags. Activated B cells initially proliferate in the periarteriolar lymphoid sheath (PALS3) and either develop locally into foci of Ab-forming cells (AFC) or return to the lymphoid follicle to initiate the germinal center (GC) reaction (1, 2, 3, 4). Although adjacent populations of AFC and GC B cells often share clonal origins (1), it is unclear whether entry into nascent GCs represents chance or whether the GC population is established by a deterministic mechanism.
Models for selective entry into the GC compartment have generally focused on lineage commitment (5) or the affinity of the BCR (6, 7). Lineage theories predict intrinsic commitment to the early AFC or GC pathways (3, 8, 9). In contrast, initial BCR affinity may determine the fate of B cells. For example, the intrinsic affinity of the BCR may translate into different degrees of activation, resulting in alternative cellular fates (10, 11). This model, in analogy to the development of thymocytes responding to agonist and antagonist peptides (12, 13, 14), suggests that very low affinity B cells might be capable of forming AFCs but unable to enter the GC/memory cell compartment (15, 16, 17, 18, 19, 20).
Early humoral immune responses are usually characterized by Abs bearing a wide range of affinities (15, 16, 17, 18, 21, 22, 23, 24). Indeed, a significant fraction of these early Abs have very low or even undetectable affinity for the immunizing Ag (17, 18, 25, 26, 27, 28, 29, 30, 31). While clonal selection theories allow for receptor heterogeneity (32, 33), production of “unspecific” Ab is unexpected. Very low affinity Abs may result from nonspecific or bystander effects; several studies of B cell repertoires and early Ab production have discounted the physiologic relevance of Abs with association constants below the micromolar range (Ka < 106 M−1) (10, 19, 34, 35, 36). On the other hand, the presence of early, unspecific Abs could indicate that the initial affinity requirement for B cell activation is extremely permissive. Indirect support for this comes from in vitro work demonstrating the ability of very low affinity (Ka ≈105 M−1) BCRs to transduce activation signals (29, 37, 38). Additionally, the participation of low affinity B cells in immune responses is suggested by the recovery of VDJ rearrangements from GCs that likely encoded Ab Ka values less than 106 M−1 (39, 40); however, a direct physical assessment of these affinities has not been performed.
Here we investigate the Ka of BCRs expressed by AFC and early GC B cells specific for the (4-hydroxy-3-nitrophenyl)acetyl (NP) hapten. Authentic VDJ rearrangements, recovered from AFC and GC B cells, were reexpressed as IgG1 transfectoma Abs, and their affinities were measured by fluorescence quenching. These Abs had Ka values for NP ranging from 5.0 × 104 M−1 to 1.3 × 106 M−1; a subset encoded Abs of such low affinity that specific binding could be discerned only when they were expressed as IgM pentamers. Early AFC and GC B cell populations possess equivalent and, often, very low affinities for Ag. The presence of very low affinity B cells in nascent GCs reveals the effective absence of affinity thresholds for cell entry into the differentiation pathway leading to humoral memory. Very low affinity B cells and Ab are physiologically relevant and specifically elicited by Ag. Nevertheless, affinity-driven selection becomes evident later in the GC reaction (41, 42, 43, 44, 45, 46, 47). Together, these observations suggest that the great specificity of humoral immune responses is not the consequence of highly selective clonal activation but of competitive survival and proliferation of higher affinity B cells.
Materials and Methods
Mice
C57BL/6 mice (6–8 wk of age) were purchased from The Jackson Laboratory (Bar Harbor, ME) or Taconic Farms (Germantown, NY). All mice were maintained in sterile microisolator cages on a 12-h day/night cycle and provided with sterile food and water ad libitum.
Ag and immunizations
The succinic anhydride ester of NP (Cambridge Research Biochemicals, Cambridge, U.K.) was reacted with chicken γ-globulin (CG; Sigma, St. Louis, MO). Mice were immunized with a single i.p. injection of 50 μg NP-CG precipitated in alum. Cohorts of mice were killed by cervical dislocation at various times after primary immunization; their spleens were removed, frozen, and processed for histology and DNA amplification, as described (1, 48). Six immunized mice provided the VDJ rearrangements expressed as transfectoma Abs (Table I⇓).
Affinities of transfectoma Abs for NP and NIPa
Staining and ELISA reagents
Goat anti-mouse Ig conjugated with horseradish peroxidase (anti-IgG1-HRP), streptavidin-HRP (SA-HRP), and streptavidin-alkaline phosphatase (SA-AP) were purchased from Southern Biotechnology Associates (Birmingham, AL). HRP-labeled peanut agglutinin (PNA-HRP) was purchased from E-Y Laboratories (San Mateo, CA). The anti-allotypic Ab RS-3.1 (anti-IgMa) (49) was a gift from Dr. T. Imanishi-Kari, Tufts University School of Medicine). The anti-mouse λ1 L-chain Ab Ls136 (50) was used to detect NP-specific serum Ab and B cells in situ (1). The Ls136 and RS-3.1 hybridomas were grown in culture, and their Ab products were purified over protein G-Sepharose (Pierce, Rockford, IL). BSA was purchased from United States Biochemical (Cleveland, OH). The succinic anhydride esters of NP and its derivative (4-hydroxy-5-iodo-3-nitrophenyl)acetyl (NIP) (Cambridge Research) were coupled to BSA as described (48, 51). Substitution ratios for NP-BSA or NIP-BSA were determined by absorbance at 430 nm. NIP-BSA, Ls136, and RS-1.3 were biotinylated with biotin-N-hydroxysuccinamide (Vector Laboratories, Burlingame, CA) using the manufacturer’s protocol.
Histologic staining and recovery of VDJ rearrangements from individual GC and foci
Serial, 6-μm-thick frozen sections of spleen were cut in a cryostat microtome (International Equipment, Needham Heights, MA), thaw mounted onto poly-l-lysine-coated slides, fixed in ice-cold acetone for 10 min, and stored at −20°C (1, 48). Rehydrated sections were stained in tandem with PNA-HRP and Ls136-biotin followed by SA-AP. The identity of NP-reactive AFC and GC cells was confirmed by staining adjacent sections with NIP-BSA-biotin (data not shown and 48 . Endogenous peroxidase activity was blocked by a 10-min incubation in 3% H2O2 before staining. HRP and AP activities were visualized using 3-aminoethylcarbazole and naphthol AS-MX phosphate/Fast Blue BB (Sigma), respectively (48). NP-specific GCs (surface λ+, PNA+) and AFC foci (cytoplasmic λ+, PNA−) were identified by bright field microscopy. Focus and GC B cells were recovered by micromanipulation as previously described (1, 48).
Cellular material from single GCs or foci were processed, and the recovered genomic VDJ rearrangements were amplified by PCR (40, 51). Briefly, an initial round of 40 amplification cycles used primers complementary to DNA 5′ of the transcription start site of V186.2 and to a region in the JH2-JH3 intron. Two-microliter aliquots of this reaction mixture were reamplified for an another 40 cycles using nested primers complementary to the first 20 nucleotides of the V186.2 gene segment and to the terminal portion of JH2. To facilitate the recovery of B cells carrying analogue VH gene rearrangements, both 5′-primers share significant homology (≥90%) with other (V186.2 and V3 subgroups) J558 VH gene family members, and the internal 3′-primer binds to both JH1 and JH2 segments. Amplified DNA was extracted, cloned into plasmid, and sequenced (40, 48).
Creation of transfectomas and site-directed mutants
Recovered VDJ segments were excised from the sequencing plasmid pBSK+ (pBluescript; Stratagene, La Jolla, CA) and inserted into a shuttle vector (pBSK.JJ) to obtain the 5′ regulatory region necessary for Ig H- chain expression (52). The modified VDJ segments were then excised, purified, and placed into an Igγ1 expression vector carrying the gpt selection marker, pECγ1-gpt (46), as described (52). The resulting constructs were linearized and transfected into J558L myeloma cells (H−;λ1+) by electroporation (53). Growth under mycophenolic acid selection and subcloning by limiting dilution was followed by screening for the production of λ1+, γ1+ Abs by ELISA; positive clones were further scored for their ability to bind NP and NIP. Some λ1+, γ1+ transfectoma proteins did not bind NP or NIP in ELISA, even when highly concentrated and purified.
The B1-8 hybridoma line (50) secretes a canonical, NP-specific λ1+/IgM Ab encoded by a V186.2/DFL16.1/JH2 rearrangement (41, 46) The B1-8 VDJ sequence was recovered and cloned into pECγ1; transfection and subcloning into J558L was performed to make an IgG1 Ab, B1-8γ1. Site-directed mutagenesis (United States Biochemical) of the B1-8γ1 sequence generated specific base pair mutations that significantly alter affinity for the NP hapten 1) in CDR1 at codon 33 (TGG→TTG) that causes a Trp-to-Leu exchange (H33Lγ1 and 2) in CDR2 at codon 50 (AGG→GGG) that yields an Arg-to-Gly replacement (H50Gγ1). The replacement at position 33 produces a 10-fold increase in Ab affinity (40, 41, 54), whereas the position 50 exchange dramatically reduces the strength of NP binding (45, 55, 56). These single substitution VDJ sequences were transfected into J558L and subcloned.
To create low affinity NP-specific IgM molecules, the position 50 mutant, H50G, and the VDJ rearrangement present in transfectoma T1 (V23/DFL16.1; Table I⇑) were subcloned into a vector that contains a 11.6-kb Cμ fragment including the intronic enhancer and membrane and secretory exons (57). The VDJ rearrangements were inserted into this vector as 4-kb EcoRI fragments that included their 5′-regulatory regions. Construct integrity was confirmed by restriction digest mapping and sequencing of the VH region. IgM expression constructs were transfected into J558L by electroporation. Transfectants secreting λ1+ IgM Abs were identified by ELISA and cloned.
Transfectoma affinity measurements
Abs were purified by concentration of transfectoma supernatants grown in miniPerm bioreactors (Heraeus, South Plainfield, NJ), followed by affinity chromatography over protein G-Sepharose. Ab concentrations and purity were determined by spectrophotometry and anti-IgG1/anti-λ1 sandwich ELISA, using B1-8γ1/λ1 as a reference standard. All purified proteins were found to consist of >90% IgG1 activity. Their Ka values for NP and NIP haptens were measured by fluorescence quenching in a Shimadzu RF-5301 fluorospectrophotometer (Shimadzu Scientific Instruments, Columbia, MD) (42, 58, 59). Excitation and emission wavelengths were 280 and 340 nm, respectively; temperature (25°C) and pH (7.4) were held constant. Titration was conducted by adding NP- or NIP-caproate (Cambridge Research) over a three-log range (10−8-10−5 M) to a known concentration of Abs (33 μM in 2.5 ml PBS) in quartz cuvettes. To correct for nonspecific quenching, MOPC 21, an anti-dextran monoclonal mouse IgG1/κ, and MOPC 104e, an anti-trinitrophenol monoclonal mouse IgM/λ1 (Sigma), were used. Emission signal loss, or quench, vs Ag concentration was plotted according to the Scatchard equation to derive association constants (Ka values) for half-maximal binding (42, 58). All samples were measured in triplicate, and affinities were expressed as mean Ka values (±SD).
Measurement of functional avidity of IgG1 and IgM Abs
B1-8, H50Gμa/λ1, and T1(V23)μa/λ1 were purified by affinity chromatography over goat anti-mouse IgM-Sepharose 4B (Sigma) and compared with IgG1 Abs (B1-8γ1/λ1, H50Gγ1/λ1, H33Lγ1/λ1, and T1(V23)γ1/λ1) to analyze avidity-dependent NP binding by IgG1 and IgM. In brief, 96-well ELISA plates (Falcon, Becton Dickinson, Oxnard, CA) were coated with 50 μg/ml NP22-BSA, NIP21-BSA, NP5-BSA, or CG in 0.1 M carbonate buffer (pH 8.6) at 4°C overnight. Plates were washed in PBS containing 0.05% Tween 20 (Sigma) and blocked with 5% FCS in carbonate buffer at 37°C for 3 h. Abs were serially titrated in ELISA plates from ≈1 ng/ml to 200–400 μg/ml and incubated for 2 h at 22°C. Plates were then washed, and Ls136-biotin was applied for 2 h; a subsequent wash was followed by SA-HRP. The TMB peroxidase substrate kit (Bio-Rad, Hercules, CA) was used to detect bound enzyme activity. Optical densities at 430 nm were determined on an ELISA reader (SoftMAX, Molecular Devices, Sunnyvale, CA) with values reported as the mean OD of three independent experiments.
The T1 and H50G IgG1 and IgM Abs showed enhanced binding if sample incubations were performed at 4°C (data not shown); this enhancement is characteristic of natural, low affinity cryoglobulin Abs (60).
Avidity/affinity determinations by 2-ME treatment
Abs were assessed for NP reactivity by ELISA after treatment with 2-ME. Mild reduction with 2-ME treatment causes 19S pentameric IgM to dissociate into 7S H2L2 subunits that remain capable of binding Ag (61). Briefly, equivalent (H2L2) concentrations of IgM and IgG1 Abs (0.02 mg/ml) were diluted in 0.1 M sodium phosphate/4 mM EDTA (PB/EDTA) and 1.0 M 2-ME was added to a final concentration of 0.01 M (61). Reduced and mock-treated samples (receiving an equal volume of PB/EDTA) were incubated at 37°C for 2 h and serially diluted in NP22-BSA-coated ELISA plates, followed by Ls136 to measure Ab binding (55). Relative NP reactivity was measured as a percentage of the B1-8γ1 end point. NP binding was identical for reduced, mock-treated, and untreated IgG1 Abs (data not shown).
Results
Affinity analysis of Abs isolated from PALS-associated foci and GCs
The Ab response of C57BL/6 mice to NP-CG is dominated by B cells bearing the λ1 L-chain, whereas immunization with carrier protein alone elicits virtually no λ1+ Ab or cells (20, 24, 45, 48, 62). By day 10 after immunization, the majority of these λ1+ B cells also express VH V186.2-to-DFL16.1 rearrangements that encode a tyrosine-rich CDR3 region with a consensus motif, YYGS (45, 48, 56, 63), a combination referred to as the canonical anti-NP BCR (43, 50). However, earlier in the response (days 4–8 postimmunization), a large proportion of activated, λ1+ B cells express noncanonical combinations of VH and D gene segments (Fig. 1⇓) (1, 48, 64). To investigate the affinity of these NP-dependent, noncanonical B cells, VDJ rearrangements were recovered from λ1+ focus and GC B cell populations by microdissection, amplified by a specific PCR, and cloned into plasmids for sequence analysis (40, 48). At the times examined (2–10 days postimmunization), H-chain genes recovered from GCs have few mutations (Table I⇑) and the λ1 L-chain genes from GCs were generally unmutated (data not shown, and Refs. 1, 40). This latter finding is consistent with previous studies demonstrating a ≥10-fold reduction in the frequency of mutations in λ1 compared with H-chain genes (62, 65, 66).
Frequency of V186.2 and V3 family VH genes recovered from plasmacytic foci and GCs during the primary anti-NP response. The frequency of J558 (subfamily V3) VH genes recovered from λ1+ PALS-associated foci or λ1+ GCs in spleens of C57BL/6 mice responding to NP-CG. Line plots (○) represent relative kinetics of focus and GC reactions during the primary response. Data shown combine previous studies (1, 40, 51) and the current work. Noncanonical VH genes include C1H4, CH10, V23, 24.8, V102, and V583.5 (41).
Thirteen VDJ fragments representing both canonical and noncanonical rearrangements from six plasmacytic foci and five early GCs were inserted into pECγ1 and transfected into J558L myeloma cells (Fig. 2⇓A). Transfectomas producing λ1+ IgG1 Abs were cloned, and their monoclonal Ig products were purified for determination of specificity and affinity. Table I⇑ shows the affinities of transfectoma Abs for NP- and NIP-caproate as determined by fluorescence quenching (42, 58, 59, 67). Three control IgG1 Abs, B1-8γ1, H33Lγ1, and H50Gγ1 (Fig. 2⇓B), were generated for affinity standards (see Materials and Methods). Representative quenching curves for the control Abs (Fig. 3⇓) illustrate that a wide range of affinities (Ka) for NP, 1.2 × 105 M−1 (H50Gγ1) to 2.0 × 107 M−1 (H33Lγ1), may be accurately determined. This range includes affinities reported for early Ab (41, 54, 55, 56, 68). To exclude the possibility that λ1+ B cells recovered from the plasmacytic foci and GCs of immunized mice were specific for the carrier protein, transfectoma Abs were assessed for reactivity to CG. No CG binding was detected by fluorescence quenching or ELISA for any transfectoma Ab (data not shown).
Ig H-chain constructs used for transfectoma affinity analysis and avidity studies. A, Elements of transfectoma IgG1 expression constructs, where X represents VDJ gene sequences recovered from spleens of immunized mice by microdissection. Recovered genes were PCR amplified and cloned into PstI-XbaI sites. Expression construct domains are indicated as follows: IgH promoter (P), leader (L), intronic enhancer (Ei), γ1 secretory domain (γS), and complimentarity determining regions (CDR1 to -3). B, Structure of control transfectoma IgG1 antibodies: B1-8γ1, containing the canonical anti-NP VDJ rearrangement recovered from the B1-8 hybridoma line; high affinity mutant, H33Lγ1 (codon 33: Trp→Leu); and low affinity mutant, H50Gγ1 (codon 50: Arg→Gly). Mutant constructs contain single site substitutions in the canonical VDJ engineered by site-directed mutagenesis. C, Transfectoma IgM expression constructs were made using a noncanonical gene segment (V23) recovered from an early GC (T1, Table I⇑) and the low affinity H50G mutant VDJ. The μ expression vector contains an 11.6-kb Cμa fragment and includes the intronic enhancer and membrane and secretory exons. All H-chain constructs were expressed in conjunction with identical λ1 L-chain proteins.
Fluorescence quenching curves for transfectoma Abs vs NP-caproate. Quenching curves are shown for the canonical anti-NP Ab B1-8γ1 (○) and the mutant Abs H33Lγ1 (□), and H50Gγ1 (▵). Fluorescence quenching was performed by measuring the change in 280 nm excitation/340 nm emission signal of Ab samples through a serial titration of NP-caproate (2.5 × 10−8–5.0 × 10−5 M). Curves represent the average (±SD) of three independent measurements for each sample. Values represent percent of maximum quench (max Q), which varied for each Ab sampled (max QB1-8γ1 = 50%, max QH33Lγ1 = 55%, and max QH50Gγ1 = 15%).
Kas for NP of Abs encoded by VH186.2/DFL16.1/JH2 rearrangements from λ1+ foci and GCs ranged from 9.0 × 104–1.3 × 106 M−1. This 10-fold range illustrates that significant CDR3 diversity can be generated even by joining identical V,D, and J gene segments. The average affinity of these canonical Abs was 6.0 (± 3.3) × 105 M−1, a value virtually identical to primary anti-NP serum Ab (42, 46, 50, 59). The average affinity encoded by noncanonical VDJ fragments (1.4 (± 0.8) × 105 M−1) was modestly, but significantly (p ≤ 0.05), lower than that of the VH186.2 set. To facilitate comparison, unmeasurably low affinities from the noncanonical set were arbitrarily assigned a value equal to the lowest Ka detectable by fluorescence quenching (≈5 × 104 M−1). This is a conservative estimate of their true affinities (see below). Of the 13 transfectoma Abs studied, only 2, T1 and T2, did not exhibit detectable binding for NP-caproate (Table I⇑). The absence of NP binding by the T2 Ab is likely due to replacement (R) mutations at codons 16 and 80 of the C1H4 gene segment, since its unaltered clonal relative T6 binds NP. The T1 Ab, however, is encoded by an unmutated V23 VH gene segment combined with an atypical DFL16.1 sequence and represents a true germline affinity.
Avidity-dependent binding of low affinity IgM Abs
To ensure that the VDJ rearrangement present in the T1 transfectoma (Table I⇑) was obtained from an NP-specific B cell, we constructed IgM-secreting transfectant lines. The high valency of IgM enhances weak binding to Ag by permitting multiple, concurrent interactions (i.e., increased avidity) (37, 38, 69, 70). Fig. 2⇑C shows the IgM constructs for H50Gμa and T1(V23)μa. Transfection of these constructs into the J558L myeloma line yielded μa+/λ1+ Abs. Canonical IgG (B1-8γ1), IgM (B1-8), and the high affinity mutant H33Lγ1 Abs were used as controls. The relative avidities of transfectoma IgG1 and IgM Abs were assessed in ELISA by measuring binding to highly (NP22-BSA) or sparsely (NP5-BSA) substituted substrates over a wide range of Ab concentrations. Avidity-dependent heteroclitic binding was determined on NIP21-BSA.
The λ1+ IgG1 Abs encoded by the H50G and T1(V23) VDJ rearrangements did not bind to solid phase NP- or NIP-BSA (Fig. 4⇓). However, H50G bound NP in ELISAs as an IgM, emphasizing the limitation of solid phase assays for detection of low affinity, bivalent Abs (71, 72, 73). Indeed, fluorescence quenching revealed specific NP binding by the H50Gγ1 Ab (Fig. 3⇑; Table I⇑), which could be detected in ELISA only with IgM proteins. Similarly, IgG1 Abs bearing the T1(V23) V region had no detectable NP binding by fluorescence quenching and demonstrated only slight heteroclitic binding to NIP (Table I⇑). However, Fig. 4⇓ shows that the T1(V23)μa Ab bound appreciably to NIP and marginally to NP in ELISA. Neither H50Gμa nor T1(V23)μa exhibited detectable binding to NP5-BSA; at low hapten densities, multivalency could not compensate for the very low affinity of these Abs. Interestingly, H50Gμa Abs remained unable to bind NIP, suggesting that Arg at residue 50 is crucial for heteroclicity (Fig. 4⇓). The H50Gμa and T1(V23)μa Abs do not bind CG in ELISA (data not shown), confirming the specificity of avidity-dependent hapten binding.
Relative binding efficiency of IgG1 and IgM transfectoma Abs. Purified transfectoma IgG1 and IgM Abs were compared by ELISA for NP binding (NP22), high affinity NP binding (NP5), and heteroclitic NIP binding (NIP21). Avidity contributions to binding were assessed by comparing the IgG1 and IgM forms of identical VDJ rearrangements, including the canonical anti-NP rearrangements B1-8 μ (□) and B1-8γ1 (▪) and lower affinity variants H50Gμ (⋄), H50Gγ1 (♦), T1(V23)μ (○), and T1(V23)γ1 (•). The high affinity transfectoma Ab H33Lγ1 (┌) was also tested. Curves represent means from triplicate experiments.
That the binding of H50Gμa and T1(V23)μa to NP depended upon the high valency of IgM was confirmed by limited reduction with 2-ME (Fig. 5⇓). This reduction dissociates 19S IgM into 7S subunits and resulted in the complete loss of hapten binding in ELISA by H50Gμa and T1(V23)μa (see Materials and Methods). In contrast, the ELISA titer of the higher affinity control IgM B1-8 was only slightly diminished by 2-ME treatment, and reduction did not affect the binding of IgG1 Abs.
Avidity-dependent binding of low affinity transfectoma Abs. Purified transfectoma Abs were compared for their ability to bind NP-BSA in ELISA after 2-ME reduction (dark bars; +) or mock treatment (light bars; −) 2-ME fragments IgM into 7S subunits without injuring the H2L2 structure. The relative reactivity of treated samples is shown as a percentage of the mock-treated B1-8γ1 reactivity. No difference was seen between binding of untreated and mock-treated samples. Results are presented as mean (±SD) from four independent assays.
Characterization of Abs from AFC foci and GCs
While differences in affinity existed between canonical and noncanonical Abs, our sampling did not reveal differences in the initial affinities of Ig produced by those B cells committed to early Ab production (AFC foci) or to the GC/memory pathway. Transfectoma Abs encoded by H-chain genes recovered from plasmacytic foci had an average affinity of 5.7 (± 3.9) × 105 M−1, a value not significantly different from the Ig receptors present in early λ1+ GCs, 3.9 (± 2.2) × 105 M−1 (Table I⇑).
Three pairs of VDJ rearrangements (T23 and T21; T30 and T25; and T6 and T2 (Table I⇑)), represent clonally related sets that were recovered from the same GCs and share identical CDR3 junctions. These related rearrangements differ from their partners by 1- to 5-point mutations representing their independent somatic evolution (40, 51, 74, 75). Although this sample of paired mutants is small, it is striking that mutations in each pair resulted in decreased affinity for NP. For example, the additional position 81 and 82 R mutations in the T21 rearrangement diminish its Ka for NP to a log below that of its progenitor, T23 (Table I⇑). Likewise, two R mutations present in T2 completely abolish the relatively robust NP binding (2.12 × 105 M−1) present in its precursor, T6. Even the effect of a single R mutation (position 41) in T25 lowers the affinity for NP compared with its progenitor, clone T30. Thus, mutations that reduce affinity, even within a single clonal genealogy, are frequent (54, 76, 77, 78, 79, 80). It may be that most V(D)J mutations, even those present in mature GCs, do not contribute to affinity maturation. Alternatively, the small size of the NP determinant may limit the mutations that lead to improved affinity (11, 81). Nevertheless, these lower affinity BCRs seem competent to survive within GCs, at least initially, and may even enter the memory compartment (44, 56).
The λ1 Abs characteristic of the anti-NP response generally demonstrate heteroclitic, or higher affinity, binding to NIP (24, 43, 50). With the exception of the mutated clone T21 and the position 50 mutant (H50Gγ1), canonical transfectoma Abs exhibited heteroclitic binding to NIP (Table I⇑). In contrast, Abs encoded by VH186.2-analogue gene segments were generally not heteroclitic (Table I⇑). Only the noncanonical T1 transfectoma, which had undetectable NP binding, demonstrated enhanced reactivity to NIP, albeit at the lowest threshold of detection (Ka ≤ 5.0 × 104 M−1).
The H50G mutation in GC B cells
Our conclusion that mutation commonly reduces affinity and that lower affinity mutants can survive in GCs is supported by studies on the population genetics of NP-specific B cells in AFC foci and GCs (1, 40, 48, 51). We have previously recovered three independent examples of the Arg→Gly exchange present in the H50Gγ1 transfectoma (Table I⇑). This high frequency suggests that position 50 may represent a mutational hotspot, especially since the Gly replacement severely reduces affinity (55). Nonetheless, this exchange does not appear to preclude clonal survival in GCs. Fig. 6⇓ illustrates the genealogy of a canonical B cell clone recovered from a single GC 14 days after immunization with NP-CG. In this GC, an Arg→Gly mutation at position 50 has occurred but allowed subsequent mutations in daughter cells (Fig. 6⇓). These findings and the recovery of an NP-specific hybridoma containing Gly at position 50 (54, 82) demonstrate that B cells bearing this deleterious mutation and its consequent low affinity (Table I⇑) are capable of activation, proliferation, and differentiation in vivo.
Genealogy of VDJ sequences, recovered from a single GC, that share a debilitating mutation. VDJ rearrangements (a-f) with identical CDR3 motifs (YDYYGS) were amplified from a single λ1+ GC 14 days after immunization with NP-CG. All recovered VDJ fragments contained the Arg→Gly R mutation present in the H50Gγ1 transfectoma. R mutations are indicated by codon number. Numbers in the dendogram represent the number of substitutions to next node. P in the first circle represents the putative founder, which is assumed to carry the unmutated, canonical anti-NP sequence.
Discussion
Forty years ago, Macfarlane Burnet postulated that B cells bearing surface receptors complementary to a single Ag would be selectively activated from a vast pool of lymphocytes (32). This clonal selection hypothesis is the central paradigm of immunology, and yet the affinity threshold for selection by Ag remains unknown. To define this threshold and its consequences for Ag-driven B cell differentiation, we have conducted the first direct assessment of Ig affinities in defined splenic microenvironments. Recovery of VDJ rearrangements that encode extremely low affinity Abs from plasmacytic foci and early GCs suggests that even those BCRs with marginal affinity for Ag allow cellular participation in normal immune responses. This in turn suggests that the frequency of biologically relevant B cells specific for any Ag is much higher than previously believed.
The exquisite specificity of serum Ab responses has led to the notion that only B cell receptors with Ka values well above 105 M−1 are specifically activated by Ag (10, 19, 34, 81). Low affinity Abs that bind Ag in an avidity-dependent manner are often dismissed as physiologically irrelevant (19, 83). Yet, almost 50% (6/13) of the recovered VDJ sequences we studied encoded λ1+ Abs with affinities that fell below this hypothetical threshold (Ka ≤ 3 × 105 M−1). A few exhibited only avidity-dependent binding and by the usual convention would have been considered nonspecific or insignificant. Nonetheless, each VDJ rearrangement was recovered by precise microdissection from collections of λ1+ AFCs and GCs that depend upon immunization with NP (48). Indeed, while only about 40% (5 of 13) of these rearrangements contained the V186.2 VH gene segment and the YYGS CDR3 motif that dominate the primary response to NP (45, 48, 56, 63), all but the mutated T2 VDJ sequence generated λ1 Abs that bound NP or NIP specifically.
We propose that very low affinity B cells normally participate in responses to thymus-dependent Ags. Indeed, many investigators have documented the production of apparently nonspecific AFCs and Ab early in humoral responses (26, 27, 29, 30, 31). These Igs had no detectable affinity for Ag but often shared phenotypic characteristics (e.g., Id) with the Ag-binding Ab set(s). Genetic analysis of NP-responsive B cells has shown that the AFC focus and early GC populations routinely contain many noncanonical VH rearrangements (1); initially, more than 80% of the VDJ rearrangements recovered from λ1+ GCs contain noncanonical VH segments. However, these genes are quickly replaced by the VH186.2 gene, such that, after day 10 of the response, only about 15% of VDJ rearrangements recovered from λ1+ GCs contained VH gene segments other than V186.2 (40). Our measurements of transfectoma Ab affinity suggest that this change in the GC repertoire is affinity driven, since most noncanonical VDJ rearrangements generated transfectoma Abs with two- to sixfold lower affinity for NP than canonical VDJ sequences. In a few cases, noncanonical Abs had no detectable NP binding by fluorescence quenching (Ka < 5 × 104 M−1).
Promiscuous activation by Ag would increase clonal participation and diversity early in the primary immune response, perhaps providing useful humoral protection during the interval required to generate high affinity Abs. Even very low affinity Ab can augment the activation of innate (84, 85) and specific immunity (86, 87), facilitating complement fixation, opsonization, and neutralization. Indeed, the bacterium Proteus morganii elicits specific Ab with virtually unmeasurable affinity (88). In GCs, a large and diverse progenitor population would ensure multiple trajectories of clonal evolution, maximizing the chance of achieving a specific, high affinity memory B cell pool (40, 67, 74, 89). The progressive loss of low affinity B cells during the immune response, i.e., in late GCs, and from the memory B cell population (45, 90, 91), is due either to an intrinsic inability of low affinity B cells to enter the memory compartment or clonal competition for survival within these immunologic niches (91).
We found no significant difference between the mean Ka for BCRs present in AFCs or GCs (Table I⇑). This supports our earlier observation of clonally related B cells in adjacent focus and GC populations (1) and suggests that, if distinct cell lineages give rise to these compartments (5, 8, 10), each lineage is very similar with respect to BCR genetics and phenotype. Later in the immune response, VH usage appears to be determined by strong clonal selection, most notably in GCs. Although there may be some selection for VH use in the PALS-associated foci (Fig. 1⇑ and 48 , these AFC remain terminally differentiated and short-lived.
Despite evidence for the overall improvement of B cell affinity in the immune response (11, 42, 45, 46, 62, 75, 90, 92, 93, 94, 95, 96), affinity maturation during the primary GC reaction appears to be local; i.e., competition takes place within but not between GCs. Previously, Vora and Manser showed that B cells expressing a high affinity Ig transgene and endogenous, lower affinity B cells matured concurrently within the same spleen (67). Canonical V186.2 rearrangements are common in mature GCs, but they are not recovered from nearby GCs that contain low affinity, noncanonical VDJ rearrangements (Refs. 40, 51, 89; and unpublished observations). Since our study focused on the early primary response, changes due to somatic mutation were limited. However, analysis of the few mutated VDJs expressed as λ1+ transfectoma Abs showed that most mutations led to neutral or deleterious effects on affinity. Additionally, the clonal genealogy shown in Fig. 6⇑ suggests that very low affinity B cells can proliferate and mutate in GCs. Thus, there does not appear to be an intrinsic affinity requirement for B cell entry or survival in GCs; instead, each GC defines a local fitness (affinity) optimum.
Alternatively, survival of low affinity GC B cells may depend upon signals or factors that are independent of the BCR. Recent work has shown that signaling via the CD21/CD19 complex is required for the survival of even very high affinity (Ka > 1010 M−1) (97). However, a substantial body of evidence suggests that GC B cells undergo apoptosis in the complete absence of BCR stimulation (98, 99, 100). We cannot rule out the possibility that, early in the GC reaction, Ag is sufficiently abundant not to constrain AFC development and GC B cell survival (101, 102). We are currently investigating this possibility by extending our transfectoma analysis to VDJs from later GCs when clonal competition is evident (M. Shimoda and G. Kelsoe, manuscript in preparation).
A potential consequence of the presence and persistence of low affinity GC B cells would be the generation of low affinity memory cells. Low affinity memory B cells are not usual in secondary immune responses (42, 45, 90, 91, 103). This could be the result of 1) an intrinsic inability of low affinity B cells to enter the long-lived memory pool (8) or 2) the failure of low affinity memory cells to persist due to their ineffective competition for stimulation by residual Ag (92, 104). In fact, long-lived bone marrow plasmacytes undergo affinity-driven selection outside of the GC microenvironment (92). Thus, even if low affinity B cells did enter the memory compartment, continued selection/competition could reduce their survival numbers over time.
Is clonal competition the sole mechanism for generating specific and high affinity Ab responses and B cell memory? The removal of competing, high affinity B cells should allow us to determine whether competition is the dominant selective mechanism during the immune response. In the absence of competition, very low affinity B cells will be either unable to mount and/or sustain humoral immunity or fully competent to initiate all facets of a T-dependent immune response. We have created two lines of transgenic mice that express BCR with very low affinities for NP in an attempt to test this hypothesis (J. M. Dal Porto et al., manuscript in preparation).
Acknowledgments
We thank J. Pryzlepa, K. Bobinet, and C. Himes for expert technical help, Dr. J. Jacob for providing VH gene usage data, and Drs. T. Manser and K. Vora for their assistance with fluorescence quenching measurements. We are grateful for the help of our laboratory colleagues, especially Drs. Jan Cerny and D. Cerasoli, who reviewed the manuscript.
Footnotes
-
↵1 This work was supported in part by U.S. Public Health Service Grants AI-24335, AG-10207, and AG-13789. A.M.H. was supported by a fellowship from the Donaghue Foundation.
-
↵2 Address correspondence and reprint requests to Dr. Garnett Kelsoe, Department of Immunology, Duke University Medical Center, Box 3010, Durham, NC 22710. E-mail address: ghkelsoe{at}duke.edu
-
↵3 Abbreviations used in this paper: PALS, periarteriolar lymphoid sheath; PNA, peanut agglutinin; AFC, Ab-forming cell; CDR, complementarity-determining region; CG, chicken γ-globulin; GC, germinal center; NP, (4-hydroxy-3-nitrophenyl)acetyl; NIP, (4-hydroxy-5-iodo-3-nitrophenyl)acetyl; R, replacement; SA, streptavidin; HRP, horseradish peroxidase; BCR, B cell receptor; H, heavy; L, light; AP, alkaline phosphatase.
- Received April 2, 1998.
- Accepted July 16, 1998.
- Copyright © 1998 by The American Association of Immunologists
References
In this issue
