A hallmark of T cell–dependent immune responses is the progressive increase in the ability of serum antibodies to bind antigen and provide immune protection. Affinity maturation of the antibody response is thought to be connected with the preferential survival of germinal centre (GC) B cells that have acquired increased affinity for antigen via somatic hypermutation of their immunoglobulin genes. However, the mechanisms that drive affinity maturation remain obscure because of the difficulty in tracking the affinity-based selection of GC B cells and their differentiation into plasma cells. We describe a powerful new model that allows these processes to be followed as they occur in vivo. In contrast to evidence from in vitro systems, responding GC B cells do not undergo plasma cell differentiation stochastically. Rather, only GC B cells that have acquired high affinity for the immunizing antigen form plasma cells. Affinity maturation is therefore driven by a tightly controlled mechanism that ensures only antibodies with the greatest possibility of neutralizing foreign antigen are produced. Because the body can sustain only limited numbers of plasma cells, this “quality control” over plasma cell differentiation is likely critical for establishing effective humoral immunity.
A major weapon used by the immune system to combat infection is the secretion of antibody molecules into bodily fluids. Antibodies, which bind to and eliminate foreign antigens, represent soluble versions of the cell surface Ig proteins that act as the B cell receptor for antigen (BCR). More than 70 yr ago, antibodies were found to alter their antigen-binding properties over the course of an immune response (1). The term “maturation of the immune response” was subsequently coined to describe the increase in antibody affinity that is now recognized to be a defining characteristic of T cell–dependent (TD) humoral immune responses (2).
To secrete antibody, antigen-activated B cells must first differentiate into plasma cells. During TD immune responses, plasma cells are initially produced in transient extrafollicular proliferative foci (3) but are subsequently derived from B cells participating in the follicular germinal center (GC) reaction (4, 5). Evidence that GCs might be connected with maturation of the serum antibody response was provided by the discovery that somatic hypermutation (SHM) of Ig genes occurs in GCs (6) and that rare mutant clones expressing BCRs with increased affinity for the immunizing antigen preferentially survive there (4, 7). Nevertheless, the precise factors that cause GC B cells to differentiate into plasma cells and, thus, drive affinity maturation of the antibody response remain unclear (8). In vitro experiments have suggested that stochastic or nonselective mechanisms are of primary importance in the regulation of plasma cell differentiation (9). On the other hand, indirect evidence suggests that plasma cell differentiation of GC B cells may be more selective, with only those cells that exceed a threshold antigen affinity contributing to the antibody response (10, 11). Distinguishing between these two possibilities has proven difficult because current experimental models do not allow affinity-based selection and plasma cell differentiation of GC B cells to be comprehensively tracked in vivo.
The SWHEL Ig knock-in mouse model was developed to analyze TD B cell responses to the protein antigen hen egg lysozyme (HEL) conjugated to the sheep RBC (SRBC) carrier (12). SWHEL B cells express the anti-HEL BCR encoded by the high affinity mAb HyHEL10 and can undergo both class switch recombination and SHM (13). Adoptive transfer of small numbers of SWHEL B cells into CD45.1 congenic recipients and challenge with HEL-SRBC results in a typical TD immune response dominated by secretion of IgG1 antibodies derived from donor SWHEL B cells (12). Responding SWHEL B cells can be tracked with precision by virtue of their expression of the anti-HEL BCR and the CD45.2 allotypic marker. The recombinant mutant HEL protein (HEL3X) binds HyHEL10 with >10,000-fold lower affinity than wild-type HEL (HELWT) (14). HEL3X-SRBC triggers migration of responding SWHEL B cells into GCs but is ineffective at eliciting an extrafollicular plasma cell response (14). In this report we exploit the low affinity of HEL3X to develop a system in which the affinity-based selection of GC B cells and their differentiation into plasma cells can be followed. This approach revealed that affinity maturation of TD antibody responses is driven by a mechanism that permits only GC B cells that have acquired high affinity for antigen to differentiate into plasma cells.
Results And Discussion
When SWHEL B cells are challenged with either high affinity (HELWT-SRBC) or low affinity (HEL3X-SRBC) antigen in CD45.1 congenic recipient mice, similar frequencies of donor-derived (CD45.2+) GC B cells are produced at over the first 15 d of the response (14), and these cells undergo equivalent rates of class switch recombination to IgG1 (Fig. 1 A). The extent of SHM measured during the early stages of the GC response (day 5) also does not differ (Fig. 1 B). However, as the responses progress, GC B cells responding to the lower affinity HEL3X-SRBC accumulate somatic mutations faster and by day 15 contain significantly more mutations per Ig heavy chain variable region gene than GC B cells responding to HELWT-SRBC (Fig. 1 B). These observations confirm previous analyses of TD antihapten responses showing similar rates of SHM when initial antigen affinity is high or low but enhanced selection for mutated variable regions in B cells with low initial antigen affinity (15).
SWHEL B cells challenged with HEL3X-SRBC do not produce the burst of extrafollicular plasma cells that typically peaks around day 5 of responses to higher affinity antigens such as HELWT-SRBC (14). As a result, the levels of both total anti-HEL antibody (14) and anti-HEL IgG1 (Fig. 1 C) present at day 5 are ∼100-fold lower when SWHEL B cells are challenged with HEL3X-SRBC compared with HELWT-SRBC. Nevertheless, the concentration of anti-HEL IgG1 in recipient serum increases progressively from days 5 to 20 of the response to HEL3X-SRBC (Fig. 1 C). This antibody is derived from SWHEL donor B cells, because it is not detected in recipients receiving HEL3X-SRBC alone (unpublished data). To examine whether the antibodies elicited in response to HEL3X undergo affinity maturation, serum anti-HEL IgG1 present at day 15 of the two responses was tested by ELISA for binding to HELWT and HEL3X. As expected, the IgG1 produced in response to HELWT-SRBC bound efficiently to HELWT but showed negligible binding to HEL3X (Fig. 1 D). In contrast, the serum anti-HEL IgG1 from recipients challenged with HEL3X-SRBC showed almost equivalent binding to HELWT and HEL3X (Fig. 1 D), indicating that affinity maturation to HEL3X had indeed occurred.
To track the appearance and ultimate fate of GC B cells acquiring increased affinity for HEL3X, donor-derived B cells within the spleens of mice immunized with HEL3X-SRBC were stained directly with fluorescently labeled HEL3X at days 5, 10, and 15 of the response (Fig. 2). Analysis of CD45.2+ donor cells revealed a progressive increase in HEL3X binding in mice challenged with HEL3X-SRBC but not HELWT-SRBC (Fig. 2). When HEL3X binding was counterstained for surface IgG1 so that HEL3X binding by IgG1+ responders could be visualized independently of BCR density, a distinct population of IgG1+ B cells with uniformly increased affinity for HEL3X was evident by day 10 of the HEL3X-SRBC response and subsequently dominated the IgG1+ donor population by day 15 (Fig. 2). This population was not apparent at any stage of the response to HELWT-SRBC (Fig. 2), indicating that the cells were specifically selected because of their increased affinity for HEL3X.
To determine whether B cells with high affinity for HEL3X carry specific somatic mutations that increase their affinity for HEL3X, single donor-derived GC B cells were sorted from immunized mice, and their Ig heavy chain variable region genes were sequenced. By day 15 of the response to HEL3X-SRBC, 82% (23 out of 28) of the clones analyzed carried a specific point mutation in the tyrosine 53 codon encoding its substitution with aspartate (Y53DHyHEL10; Fig. 3 A). No selection of any heavy chain mutation was evident in day 15 GC B cells produced in response to HELWT-SRBC (Fig. 3 A), which was consistent with the proposition that the affinity of the HELWT–HyHEL10 interaction is too high to permit further affinity maturation (16). None of the 24 clones analyzed from the day 15 HELWT-SRBC GC response contained a mutation in the Y53HyHEL10 codon, indicating that the Y53DHyHEL10 mutation is selected specifically in response to HEL3X-SRBC and is therefore likely to increase the affinity of HyHEL10 for HEL3X. This was confirmed by sorting high affinity anti-HEL3X IgG1+ donor B cells (see gate in Fig. 2), as subsequent sequence analysis revealed that 96% (23 out of 24) of these clones carried the Y53DHyHEL10 mutation (unpublished data).
Analysis of the binding of HEL3X to recombinant wild-type and Y53D-mutated HyHEL10 IgG1 antibodies showed that the Y53DHyHEL10 mutation increases the affinity of HyHEL10 for HEL3X by ∼85-fold (Fig. 3 B). This affinity increase was also evident from the ability of the mutated form of HyHEL10 to recognize plate-bound HEL3X efficiently in ELISA under conditions in which binding of wild-type HyHEL10 to HEL3X was virtually undetectable (Fig. 3 C). Computer modeling revealed that the arginine side chain introduced at position 101 of HEL (D101RHEL) to produce HEL3X (14) is likely to cause a major steric conflict with the phenol group of Y53HyHEL10 and that this conflict is resolved by the Y53DHyHEL10 substitution (Fig. S1, available at http://www.jem.org/cgi/content/full/jem.20061254/DC1).
To examine the selection of the Y53DHyHEL10 mutation over the course of the response to HEL3X-SRBC, Ig heavy chain gene sequence analysis of single GC B cells and plasma cells was performed on days 5, 10, and 15. On day 5, the Y53DHyHEL10 mutation was not detectable in either the GC or the small plasma cell compartment (Fig. 4 A), which was consistent with the absence of high affinity anti-HEL3X B cells at this time point (Fig. 2). By day 10, however, this mutation was detectable in some GC B cells (22% of sequences) but was already present in the great majority of splenic plasma cells (86% of sequences). Similar domination of the GC B cell population by this mutation was not evident until day 15 (Fig. 4 A). Because hypermutated plasma cells must have derived from GC B cell precursors, this result shows that GC B cells generated in response to HEL3X-SRBC do not undergo stochastic differentiation into plasma cells but instead differentiate upon acquisition of the high-affinity Y53DHyHEL10 mutation. Accordingly, Y53D-mutated clones that had left the GC and differentiated into plasma cells by day 10 of the response had a significantly lower overall rate of SHM than the Y53D-mutated clones that remained within the GC compartment (2.2 vs. 4.6 mutations/clone, respectively; P = 0.02).
To monitor affinity-based regulation of post-GC plasma cell differentiation more directly, we next challenged donor B cells from SWHEL × Blimpgfp/+ mice with HEL3X-SRBC. Plasma cells generated in Blimpgfp/+ mice express GFP under the control of the Blimp-1 promoter and, thus, can be detected via intrinsic green fluorescence (17). By using SWHEL × Blimpgfp/+ donor B cells in conjunction with our method for identifying somatically mutated B cells with high affinity for HEL3X (Fig. 2), we directly assessed the antigen affinity of post-GC IgG1+ plasma cells in the spleen. Analysis of GFP expression by low and high affinity donor-derived IgG1+ B cells on day 10 of the response to HEL3X-SRBC clearly demonstrated that GFP-expressing plasma cells originated almost exclusively from GC precursors that had acquired a high affinity anti-HEL3X BCR (2.54% of high affinity compared with 0.27% of low affinity cells; Fig. 4 B). These GFP+ cells also had low but detectable levels of surface CD45.2 and IgG1 (Fig. 4 B and not depicted), as is typical of plasma cells (14). Because >95% of high affinity clones detected by HEL3X binding have the Y53DHyHEL10 mutation, most of these GFP+ plasma cells are likely to secrete antibodies carrying this amino acid change. Indeed, the serum IgG1 found in mice challenged with HEL3X-SRBC shows almost identical antigen-binding characteristics to Y53D-mutated HyHEL10 (Fig. 1 D and Fig. 3 C).
The model system described in this report provides a unique window into the GC reaction by allowing the appearance, selection, and differentiation of high affinity somatically mutated B cells to be followed throughout the response. We have used this system to demonstrate that high affinity B cell specificities generated within the GC are harnessed to drive affinity maturation of the antibody response by a mechanism that ensures their rapid and selective differentiation into plasma cells.
Because low affinity B cells survive within GCs without undergoing plasma cell differentiation (Fig. 4 B), it is apparent that the affinity-dependent mechanism that regulates plasma cell differentiation from GC B cell precursors operates independently of the processes that govern GC B cell survival. The existence of this mechanism was not predicted from in vitro experiments, because these show that B cells can undergo stochastic plasma cell differentiation without requiring a BCR signal (9). This would suggest that specific controls exist within the GC microenvironment that suppress plasma cell differentiation in the absence of signals from high affinity antigen.
The requirement for antigen-dependent signals to drive plasma cell differentiation from GC B cells presents an interesting parallel with the regulation of the extrafollicular plasma cell response. These early plasma cells arise independently of the GC reaction (3) but are similarly biased toward high affinity specificities or epitopes present at high density (14). Therefore, the BCR–antigen interaction appears to play a key role in regulating TD plasma cell differentiation both before and after GC formation. It is possible that BCR signaling could facilitate plasma cell differentiation via the induction of Bcl-6 degradation (18) and subsequent lifting of Blimp-1 repression (19). Alternatively, responding B cells may stochastically commence plasma cell differentiation but require BCR signals to survive beyond the very earliest stages of this process. Whatever the precise mechanism, it is apparent that the immune system places a high priority on ensuring that it devotes resources primarily to the production of relatively high affinity antibodies that are most likely to be biologically effective. The importance of this stringent regulation of plasma cell differentiation is perhaps underscored by the relatively permissive affinity requirements for GC B cells to enter the memory B cell compartment (10, 20). The emphasis on quality control of in vivo plasma cell differentiation may have been a critical development during the evolution of the immune system. Because the body is known to have only a relatively limited capacity within the specialized microenvironments that sustain plasma cells (21, 22), it can be seen that tight control over the specificities that enter the plasma cell compartment would be essential for ensuring that the antibodies that are produced provide effective immune protection.
Materials And Methods
Mice and procedures.
SWHEL mice (13), Blimpgfp/+ mice (17), mutant HEL proteins (14), conjugation of HEL to SRBC, and the adoptive transfer system (12) have been previously described. SWHEL spleen cells were not purified before transfer. Mice were housed in a specific pathogen-free environment at Centenary Institute, and experiments were approved by the University of Sydney Animal Ethics Committee.
Serum anti-HEL antibody levels were analyzed by ELISA as previously described (13). The specificity of serum IgG1 antibodies for HELWT and HEL3X was determined by coating the ELISA plates with the respective antigens. The relative affinities of HyHEL10WT and HyHEL10Y53D for HEL3X were determined by capture ELISA. 5 μg/ml each of soluble HyHEL10WT and HyHEL10Y53D IgG1 antibodies was captured by plate-bound anti–mouse IgG1. Subsequent binding of HEL3X was detected with biotinylated HyHEL9, which recognizes an epitope on HEL distinct from that bound by HyHEL10. Nonlinear regression based on a sigmoidal binding curve was performed using software (Prism; GraphPad) to find the curve-fit and calculate the half-maximal binding concentration and relative affinity of HyHEL10WT and HyHEL10Y53D for HEL3X.
Flow cytometry and single cell sorting.
Splenocytes were prepared, stained for surface molecules with monoclonal antibodies, and analyzed on a FACSCalibur (BD Biosciences) as previously described (14). To track affinity maturation, cells were stained with recombinant HEL3X conjugated to Alexa Fluor 647 (Invitrogen). For analysis of Blimp-GFP expression, CD45.2 was conjugated to Pacific blue (Invitrogen) to allow identification of donor-derived cells and data acquired on a flow cytometer (LSR II; BD Biosciences). Single cells were sorted on a FACSAria (BD Biosciences) as previously described (14). The gates used for single cell sorting of GC B cells (CD45.2hi, syndecan-1−) and plasma cells (CD45.2int, syndecan-1+) were the same as those previously described (14). These gates have been verified through localization of antibody secreting activity (12) and high levels of intracellular Ig staining (14) to cells in the plasma cell gate and demonstration of high levels of GL7, PNA, and Fas on cells in the GC gate (12, 14).
The HyHEL10 Ig heavy chain variable region gene was amplified from single-responding SWHEL donor B cells and sequenced as previously described (14). Translated sequences were aligned with the original HyHEL10 protein sequence to determine the position and significance of the mutations, as previously described (12). An unpaired t test was used to calculate the probability (p-value) when comparing SHM frequency per clone in different responding populations.
Expression of wild-type and Y53D-mutated HyHEL10 IgG1 antibodies.
The canonical T to G mutation encoding the Y53DHyHEL10 substitution was introduced by PCR mutagenesis into a pcDNA3 vector (Invitrogen) encoding the HyHEL10 γ1-secreted Ig heavy chain. Wild-type and mutant heavy chain constructs were transiently expressed in Chinese hamster ovary cells along with wild-type HyHEL10 κ light chain construct, and culture supernatants were collected and concentrated.
Online supplemental material.
Fig. S1 shows Rasmol wire frame representations based on the HEL–HyHEL10 complex that model the effects of the D101RHEL and Y53DHyHEL10 substitutions on the interaction.
We thank Jenny Kingham and the Centenary Institute Animal Facility staff for animal husbandry; Chris Brownlee and Michelle Amesbury for mouse screening; and Adrian Smith and Vivienne Moore for FACS sorting. We also thank Stuart Tangye for his comments on the manuscript.
T.G. Phan was supported by a Dora Lush Scholarship from the National Health and Medical Research Council of Australia and an Early Career Development Award from the University of Sydney. This work was funded by program grant 183700 from the National Health and Medical Research Council of Australia.
The authors have no competing financial interests.
T.G. Phan and D. Paus contributed equally to this work.
T.G. Phan's present address is Dept. of Microbiology and Immunology, University of California, San Francisco, San Francisco, CA 94143.