Summary
Germinal centers (GCs) are the primary sites of antibody affinity maturation, sites where B-cell antigen-receptor (BCR) genes rapidly acquire mutations and are selected for increasing affinity for antigen. This process of hypermutation and affinity-driven selection results in the clonal expansion of B cells expressing mutated BCRs and acts to hone the antibody repertoire for greater avidity and specificity. Remarkably, whereas the process of affinity maturation has been confirmed in a number of laboratories, models for how affinity maturation in GCs operates are largely from studies of genetically restricted B cell populations competing for a single hapten epitope. Much less is known about GC responses to complex antigens, which involve both inter- and intraclonal competition for many epitopes. In this review, we (i) compare current methods for analysis of the GC B cell repertoire, (ii) describe recent studies of GC population dynamics in response to complex antigens, discussing how the observed repertoire changes support or depart from the standard model of clonal selection, and (iii) speculate on the nature and potential importance of the large fraction of GC B cells that do not appear to interact with native antigen.
(n=188)
Keywords: Germinal center, affinity maturation, dark antigen
Introduction
A fundamental feature of the B-cell response to T-cell-dependent antigens is the formation of germinal centers (GCs), transient microanatomical structures that comprise antigen-activated B cells, follicular dendritic cells (FDCs), and follicular T-helper (TFH) cells (1). GCs are the canonical sites of humoral affinity maturation, wherein B cells undergo cycles of clonal expansion, somatic hypermutation (SHM) of B-cell antigen receptors (BCRs), and enrichment of clones with higher affinity for antigen (2–5). After numerous iterations of this process, GCs are generally believed to be filled with high-affinity B cells, which differentiate into quiescent memory B cells or antibody (Ab)-secreting plasma cells to provide durable immunity (6, 7).
The specifics of the formation and cellular dynamics of GCs have been discussed extensively in recent reviews (6, 7), so we recount here only a sufficient background to understand the current models for antigen-driven clonal selection and affinity maturation. Briefly, humoral responses begin with the migration of antigen-activated, follicular B cells to the border of T-cell zones in secondary lymphoid tissues, where they receive cognate help from antigen-specific T-helper (TH) cells (8, 9). This TH:B interaction triggers local B-cell proliferation and a subsequent migration of some fraction of the expanded B- and TH-cell populations into the follicle, where nascent GCs form within the reticulum of FDCs (8, 10). With time, GCs become organized into light- and dark zones (LZ and DZ, respectively) (11, 12). In the DZ, GC B cells proliferate and acquire V(D)J mutations as a result of DNA lesions introduced by activation-induced cytidine deaminase (AID) – the process of Ig SHM (6, 13). These mutations become fixed by cell proliferation, and non-dividing mutant B cells enter the GC LZ, where they recover unprocessed antigen held by FDCs and present it to a subset of the resident TFH (14). Ongoing interaction between GC B cells and TFH are necessary to sustain the GC reaction, as blockade of CD40:CD154 signaling soon disrupts GCs (5, 15), and recent experiments link the quality of B:TFH interaction to the return of GC B cells to the DZ for additional rounds of SHM and proliferation (14, 16, 17).
Cyclic models for natural selection in GCs were first proposed by MacLennan (11, 12); shortly thereafter, mathematical models by Kepler and Perelson (18, 19) showed that rounds of mutation, proliferation, and selection could optimize B-cell selection for increased rates of affinity maturation. Remarkably, it took almost another 20 years before Victora et al. confirmed periodic B-cell migration between the GC DZ and LZ directly by a novel method of intravital microscopy (14).
The dynamics of B cells in GCs immediately suggested plausible models for affinity maturation, the most obvious being that mutant B cells entering the LZ would compete for survival signals by their capacity to recognize antigen held by FDCs. In this way, B cells expressing mutated BCR with lower or no affinity for antigen would die, either by an intrinsic signal (20, 21) or by their inability to interact with TFH (15). Interest in identifying an intrinsic signal that measured BCR affinity waned when transgenic B cells bearing very-low-affinity BCR were shown to be capable of supporting robust GC responses in the absence of higher affinity competitors (22, 23). In consequence, the standard model became focused on low affinity mutants being unable (or unlikely) to receive survival and/or proliferation signals from TFH as a consequence of their inability to gather and process antigen. Regardless of whether affinity selection was determined by an intrinsic signal or from TFH, these models were based on negative selection for less-fit GC B cells.
Whether BCR signaling per se directs selection of higher affinity clones remains a question; however, a number of recent studies support modified selection models in which the BCR acts primarily as a receptor for acquiring antigen to process and present to TFH as peptide associated with major histocompatibility complex class II molecules (pMHCII) (16, 17, 24, 25). In consequence, GC B cells with high affinity BCRs are more successful than their lower affinity competitors in acquiring and presenting processed antigen as pMHCII. GC B cells with high affinity BCR express high densities of pMHCII, which in turn, increase the quality of TFH proliferation signals that promote GC B-cell division (16, 17, 24, 25). It is unclear whether GC B cells with higher BCR affinity also gain fitness through increased survival; contrary to the conclusions of early studies (20, 21, 26), recent experiments suggest that GC apoptosis is linked weakly, if at all, to BCR affinity (27). Regardless, the current T-cell help model for affinity-driven selection in GCs posits that higher-affinity GC B cell populations outstrip lower-affinity competitors by positive – not negative – selection. The result of this selection, whether negative or positive, should be a progressive increase in average BCR affinity (i.e., fitness) coupled with a concomitant decrease in genetic (clonal) diversity as increasingly fewer high affinity clones outcompete less-fit, lower affinity B cells (28). Indeed, these outcomes are often observed in GC populations elicited by haptenated immunogens (4).
A major caveat of the standard model for antigen-driven affinity maturation is that it rests largely on studies of Ab responses to haptens that provoke genetically restricted humoral responses; for example, immunization of Ighb mice with (4-hydroxy-3-nitrophenyl)acetyl (NP) conjugates elicits B-cell responses wherein >90% of the responding clones express a BCR formed by heavy (H)- and light (L)-chains bearing the VH1–72 and Vλ1 gene segments (3, 29). Moreover, B cells isolated from late GCs frequently bear characteristic VH point mutations (e.g., the exchange, W33L) that significantly increase BCR affinity (30–32). Genetically restricted humoral responses are experimentally tractable; however, they are not representative of responses to the complex protein antigens found in vaccines or pathogens. Complex protein antigens elicit genetically diverse, polyclonal humoral responses driven by the many and various epitopes arrayed on the antigen (33, 34). Unlike responses to haptens, in which a single epitope drives affinity maturation largely by intraclonal competition (4), GC responses to complex antigens involve both intraclonal competition for a common epitope, as well as interclonal competition for single or distinct epitopes. Because of these fundamental differences, it has been unclear whether the hapten-focused standard model accurately describes humoral responses elicited by infection or vaccination.
In this brief review, we describe recent developments in repertoire analysis of humoral responses to complex protein antigens, with particular focus on studies that prioritize analysis of Ab specificity and binding strength. We begin by describing and comparing current technologies for B-cell repertoire analysis, and then discuss studies that address the gap in knowledge regarding complex antigen-driven competition and affinity maturation in GCs. Finally, we offer several thoughts on the surprisingly large portion of the GC response that is apparently not driven by native antigen, the agency we have described as “dark antigen.”
Current methods for B-cell repertoire analysis
Due to methodological advances in the past fifteen years, current studies of B-cell repertoires can sample much larger fractions (i.e., several hundreds or thousands of cells) of the populations of interest than was previously possible (tens). These advances have made it experimentally tractable to study the genetically diverse humoral responses to complex, polyepitopic antigens. Two different strategies account for the majority of contemporary repertoire analysis studies (Figure 1). The first (35, 36) is to sort single B cells into microwell plates and then use RT-PCR to amplify paired H- and L-chain V(D)J rearrangements. These amplicands are sequenced to determine the V(D)J rearrangements present in each cell, and paired H- + L-chain sequences are expressed as recombinant Abs for the determination of BCR specificity and binding strength. Variations on this method have gained widespread use for B-cell repertoire analysis. However, the amplification of paired H+L V(D)J DNA rearrangements from single cells is non-trivial and results in an aggregate cloning efficiency of ~30% (i.e., 30 H+L sequence pairs per 96 sorted cells) (37). Additionally, substantial effort is required to clone, express, and purify significant numbers of recombinant Abs, limiting the utility of this method for smaller laboratories (and budgets!). Finally, analysis of the Abs’ specificity, avidity, and functional properties occurs at the final step, potentially resulting in a substantial cost in time and resources if only a subset of Abs possess a relevant characteristic (e.g., virus-neutralization).
Figure 1. Dominant strategies for B-cell repertoire analysis.
A, Single-cell PCR begins with cloning paired VDJ and VJ rearrangements from single B cell mRNA. The amplified heavy- and light-chain variable regions are subjected to DNA sequencing to determine gene segment usage, and also cloned into expression vectors containing the heavy- or light-chain constant regions. Paired heavy- and light-chain vectors are co-transfected into an expression host for recombinant Ab (rAb) production, followed by rAb purification and characterization. B, After single-cell sorting, Nojima cultures use CD154-expressing feeder cells and cytokines to drive B-cell proliferation, class-switch recombination, and differentiation into IgG-secreting plasma cells. Culture supernatants are separated from the cells and screened for the presence of IgG. Clonal IgG+ supernatants are used for Ab characterization. After identifying Abs of interest, abundant B-cell mRNA is readily used to clone the paired V(D)J rearrangements for DNA sequencing. If necessary, rAb can be produced for further study.
We have recently developed an alternative approach: the cloning of individual B-cells in cultures supported by specialized feeder layers (Nojima cultures)(38–40). Single mouse or human B cells are sorted into 96-well plates pre-seeded with feeder cells (41, 42) that express low densities of CD154, the CD40 ligand. Additional cytokines, including BAFF, IL-2, IL-4, and IL-21, are either produced by the feeder cells or are added exogenously. By this method, single mouse or human B cells robustly proliferate (≥4 × 104 progeny cells), undergo class-switch recombination (CSR), and differentiate into plasmablasts and plasmacytes, such that culture supernatants harvested from mouse B cells contain 1–5 μg/mL monoclonal IgG, while those from human B cells contain ~50 μg/mL IgG (38, 40). The quantity of clonal IgG in each culture supernatant is more than adequate to screen for specificity and avidity [relative to known IgG standards; the avidity index (AvIn) (38)] by, for example, standard multiplex or microneutralization methods. Following this determination of BCR phenotypes, the amplification and sequence analysis of paired H- and L-chain V(D)J rearrangements in cell pellets from single culture wells is technically simple and efficient (>95% recovery) due to the large numbers of genetically identical sister B cells in each culture. Recombinant IgG Abs from paired rearrangements can then be used to confirm the initial screening results and to allow interesting Abs to be studied more intensively, e,g., high resolution structural studies (40) or in vivo protection studies. The concentration and yield of secreted IgG in Nojima cultures are typically sufficient to make recombinant IgG Abs an adjunct to repertoire analysis rather than a requirement.
A caveat to the Nojima culture is its reliance on the ability of B cells to proliferate in culture. In consequence, plasmacytes, which do not divide, cannot be studied by Nojima cultures. GC B cells, which may be fragile (1) or predisposed to differentiate into plasmacytes, have a lower cloning efficiencies (~25%) than naïve or memory B cells (70–80%) (38, 43). Despite these limitations, we have found Nojima cultures to be a useful tool for BCR repertoire analysis and an effective pathway for the identification and study of “interesting” Abs (38, 40). Additionally, we are currently developing and validating analogous culture methods for single B cells from Rhesus macaque, a non-human primate that is often used in pre-clinical studies and trials.
Following the trend in many disciplines, high-throughput (HT) platforms that interrogate >5 ×104 B cells in a single experiment are likely to have a prominent role in the future of BCR repertoire analysis (44–46). Indeed, numerous HT techniques – including commercially available ones – now exist for analysis of mouse and human Ab repertoires. For in-depth discussion of both the merits and pitfalls of these methods, we recommend two excellent reviews (47, 48). We note here two caveats of current HT approaches: (1) many do not retain H- and L-chain pairing, and consequently lose information regarding native Ab specificity and/or avidity; and (2) despite their capacity for large-scale analysis of BCR genetics, most HT methods rely on expression of limited numbers of recombinant Abs and determine only small samplings of the BCR binding repertoire. An exception to both points is a recently described technique in which paired H- and L-chain V(D)J rearrangements from >106 single B cells are physically linked and displayed as antigen-binding fragments (Fabs) on yeast cells (49). Fluorescently labeled antigens can be used to progressively enrich and isolate yeast expressing desirable Fabs. Determination of paired H- and L-chain V(D)J arrangements from single yeast colonies or from enriched populations enable synthesis of recombinant IgGs for further binding studies (49).
Analysis of GC B cell repertoires after immunization with complex antigens
Affinity maturation in response to complex antigens
Standard models for affinity maturation in GCs rest on antigen-driven clonal competition that stringently removes lower-affinity competitors and is largely the construct of studies of GC- and Ab-responses to haptens. To determine whether these models adequately describe the intra- and interclonal competition of GC responses to complex, polyepitopic antigens, like proteins, we recently characterized how immunizations with recombinant Bacillus anthracis protective antigen [rPA] or recombinant influenza hemagglutinin [rHA] drove changes in clonally diverse B-cell populations (38). We used Nojima cultures to obtain clonal IgGs from antigen-binding, mature follicular B cells from unimmunized mice and GC B cells from immunized animals (38). In the pre-immune population of C57BL6/J mice, mature B cells that avidly bound rPA or rHA were present at frequencies on the order of 10−4 (1/8,000 or 1/18,000, respectively). Consistent with standard models, antigen-driven selection increased the representation of rPA- or rHA-specific B cells ≥4,000-fold, from ≤0.02% among mature naïve B cells to 30% in early (day 8) GCs, and ~50% among late (day 16) GC B cells.
To assess affinity maturation, we then determined the AvIn of each clonal IgG relative to rPA- and rHA IgG Ab standards (Figure 2)(38). The median AvIn for rPA-binding IgG Abs increased ~12-fold from rPA-selected pre-immune mature follicular B cells to day 8 GC B cells, then increased an additional ~4-fold in day 16 GC B cells. Avidity for rHA also progressively increased: approximately 4-fold from mature follicular B cells to day 8 GCs, followed by a 20-fold increase in day 16 GCs. These changes in the median affinity distributions of each antigen-specific B-cell population were consistent with the standard models for affinity maturation, namely that antigen drives biased recruitment of antigen-specific B cells into GCs (25), then continues to select for higher BCR affinities after additional rounds of inter- and intra-clonal competition (4, 5).
Figure 2. Affinity maturation in GCs responding to complex antigens.
Clonal IgGs were generated from antigen-binding pre-immune mature B cells (gray) or day 8 (green) or day 16 (blue) GC B cells from rPA-immunized mice. The distributions of binned avidity indices from clonal IgGs binding to rPA are shown. Clonal IgGs that did not bind antigen were excluded from analysis. The cutoff for antigen binding was set six standard deviations above the mean value for culture supernatants from B-cell-negative samples. Reprinted from (38) with permission.
Nonetheless, our AvIn measurements for hundreds of GC B cell receptors specific for rHA or rPA revealed that a substantial fraction (20%–30%) of these GC populations bound the eliciting antigen with much lower avidity (38). Whereas some 70%–80% of antigen-binding GC B cells expressed BCRs that clustered in a single high-avidity peak (AvIn ranging from four-fold higher to six-fold lower than the standard Ab), the remaining fraction of GC B cells was distributed as a long, lower-avidity tail spanning a >100-fold range of avidities (AvIn that were six- to 5,000-fold lower than the standard Ab)(Figure 2)(38). Despite the efficient affinity maturation exhibited by the majority of GC B cells, GC responses to two complex protein antigens were more permissive for the survival of low-avidity B cells than standard models for selection in GCs predict.
The abundance of low avidity GC B cells was surprising, and we initially assumed that these cells represented members of high-avidity clones that had acquired deleterious mutations. However, this assumption proved incorrect, as many of the low-avidity clones proved to be unrelated to their higher-affinity competitors (38). We termed this inclusion of low- and high-avidity B cells in GCs “permissive selection” and developed a simple mathematical model to account for it (Figure 3)(38). Our model simply assumes that the probability of B-cell division increases linearly with the cumulative distribution function on BCR avidity; above some threshold, B cells are guaranteed to divide, providing positive selection for high-affinity cohorts, but allowing for the survival of less-fit B cells as well. Our model makes no claim about the mechanism for determining increased avidity/fitness, i.e. whether it is conferred by an intrinsic signal such as BCR signaling, or by qualitative changes in TFH interaction. However, this simple model accurately describes the complex avidity distributions that we observed in early and late rPA and rHA GCs, and we propose that this or some other form of permissive selection operates on GC B cells.
Figure 3. Mathematical model for permissive selection in GCs.
The probability of a LZ GC B cell being selected for cell division is a function of its BCR avidity relative to its GC competitors. GC B cells whose BCRs surpass some avidity threshold (T) are guaranteed to divide, while those with the lowest-avidity BCRs have a small-but-non-zero probability of receiving selection signals. Adapted from (38).
Whether these lower affinity B cells compete in the same GCs occupied by high affinity clones is not known and remains to be demonstrated by cloning individual B cells from single GCs recovered by microdissection. However, we have demonstrated that intraclonal affinity differences, likely representing clonal progeny within the same GC, can span a 40-fold range of AvIn values (38). While it remains possible that segregation of lower- and higher-affinity clones, either during the initiation or maturation of the GC response, could result in the generation of GCs containing low- or high-affinity B-cell populations, this sort of segregation would require the following conditions: first, that GC B cells do not migrate from one GC to another, and second, that no global agency of selection drives affinity maturation. These conditions are not trivial, as the first refutes the notion of “open” GCs (50), and the second is incompatible with a role for circulating Ab in driving affinity maturation (51). Indeed, B cells that exit GCs become refractory to GC re-entry for weeks (52). Our preferred explanation is that the driving forces of affinity maturation are sufficiently permissive to support individual GCs populated with clones having substantially different (10- to 100-fold) BCR affinities for antigen, even late in the immune response. Preliminary studies in our lab of BCR AvIn distributions in single GCs versus whole lymph node support this conclusion (Yeh et al., unpublished data).
A recent study from our laboratory supports this “permissive selection” model for GCs. Evidence has accumulated that affinity maturation is driven by increased quality (or probability) of TFH help for GC B cells displaying the highest density of pMHCII, with the accumulation of processed antigen acting as a surrogate measure for BCR affinity (14, 16, 17, 24, 25, 53). We quantitatively tested this hypothesis by generating chimeric mice in which B cells differing only by their capacity to express MHCII could compete for entry into and success within GCs. Control of MHCII expression was by haploinsufficiency, that is, B cells either were homozygous or were heterozygous for functional MHCII loci (MHCII+/+ or MHCII+/−, respectively)(43). In some instances, we fixed the initial BCR affinity in both populations by a shared VDJ knockin; in others, we allowed both populations to express the full BCR repertoires available to them (43). Absent competition, both MHCII+/+ and MHCII+/− B cells produced identical humoral responses, demonstrating that MHCII+/− and MHCII+/+ B cells are equally competent to mount Ab and GC responses. When MHCII+/+ and MHCII+/− B cells were placed in direct competition in chimeric mice, MHCII+/+ cells held a significant advantage over MHCII+/− cells in recruitment to nascent GCs, confirming an earlier report by Schwickert et al. (25). This advantage, however, did not persist in established GCs. Instead, MHCII+/+ and MHCII+/− GC B cell ratios remained stable over the entire course (24 days) of the GC response. Histological analysis revealed that every GC contained mixtures of MHC+/+ and MHCII+/− cells, and Nojima culture-based repertoire analysis demonstrated that MHCII+/+ and MHCII+/− GC B cells acquired equivalent frequencies of V(D)J mutations and achieved indistinguishable affinity maturation. The equivalent fitness of MHCII+/+ and MHCII+/− B cells during affinity maturation in GCs demonstrates that whatever the driver of affinity maturation in GCs may be, it is insensitive to two-fold differences in pMHCII expression. Given that the GC reaction is the paradigm for affinity-driven selection, it is remarkable that entry into the GC response exhibits a greater dependence on pMHCII density than affinity maturation during GC the GC response (43). How affinity-driven selection acts in GCs, and it surely does, remains something of a mystery.
Genetic diversity of GC B cells responding to complex antigens
Generally, standard models for affinity maturation in GCs predict a gradual reduction in genetic (clonal) diversity as increasingly fewer high-affinity clones prevail in competition with less-fit, lower-affinity, competitor B cells. As noted, this restriction has been observed in the genetically restricted responses to haptens (4), and we have tested whether similar reductions in clonal diversity occur in GCs elicited by immunization with rPA or rHA by determining paired H- and L-chain V(D)J rearrangements recovered from Nojima cultures of unselected and antigen-binding mature follicular B cells from naïve mice and GC B cells recovered from mice on days 8 and 16 after immunization (38).
In unselected mature follicular B cells, the distribution of VH gene segment usage was distributed widely over the IghV locus, with no more than 7% of population using any particular gene segment. In contrast, rPA- and rHA-binding mature follicular B cells were enriched for specific but distinct VH gene segments, with each constituting some 10%–30% of the antigen-specific BCR repertoire. Distributions of VH gene usage were largely shared between antigen-binding mature follicular B cells and day 8 (early) GC B-cell populations. However, these early “winners” in rPA- and rHA-specific GCs did not continue to dominate as the GC response progressed. Rather, in day 16 (late) GCs, the early dominant clones were supplanted by B cells that had been rare in early GC populations (38). This trend for increasing clonal diversity implies that neither the initial prevalence nor avidity of a BCR are sufficient to predict clonal fitness over the duration of GC responses to complex protein antigens.
Our observation of increasing clonal diversity in late GCs is consistent with a report by Victora and colleagues (54), who used Confetti mice and V(D)J analysis from single GCs to estimate clonal diversity in response to immunization with a protein antigen. In this study, B cells were stochastically labeled with up to ten different fluorescent protein color combinations, and color distributions were tracked in individual GCs as a function of time. While the frequency of GCs dominated by a single color (and presumably a single clone) increased with time after immunization, a majority of GCs were still not dominated by a single color by day 15, and a significant fraction of GCs retained a low-dominance phenotype (i.e., highly multi-colored) as late as day 23 after immunization. Moreover, by sampling large numbers of B cells from several individual GCs (>50 cells per GC) and determining V(D)J usage, the authors confirmed that many GCs retained substantial clonal diversity (a median of 9 clones per GC) late into the response. Rather than stringently restricting genetic diversity to a few clones carrying characteristic, affinity-enhancing mutations, late GCs support substantial clonal diversity even as selection for increased affinity proceeds (Figure 4).
Figure 4. Models for clonal competition and affinity maturation in GCs.
A, During hapten-specific responses, antigen activates a genetically-restricted population of B cells, which are then recruited to nascent GCs in a manner strictly proportional to their BCR affinities. Subsequent, mostly intraclonal, competition among GC B cells results in increased representation of high-affinity clones at the expense of BCR diversity. B, During humoral responses to complex antigens, antigen activates and drives a genetically diverse pool of B cells to nascent GCs, favoring GC recruitment of higher-affinity B cells. Inside GCs, intra- and interclonal competition select for expansion of higher-affinity clones, but tolerate a wide range of BCR affinities. Late GCs exhibit high BCR diversity and a significant frequency of B cells with undetectable avidity for native antigen. Reprinted from (38) with permission.
Dark Antigen: potential origins and importance
A surprising feature of GC responses to complex antigens is the high abundance of GC B cells with no measurable avidity for native antigen. In our studies of GCs elicited by rHA or rPA, approximately 70% of the BCRs expressed by early GC B cells and 50% of BCRs from late GCs did not detectably bind immunogen in standard in vitro immunoassays (38). This unanticipated result persisted even when we denatured the native rHA and rPA proteins in vitro to mimic the cryptic- or neo-epitopes potentially generated by formulating these antigens (i.e., alum adsorption) for immunization (38). Other laboratories have also observed GC B clones with no measurable affinity for native antigen, and note that such “unspecific B cells” can co-exist with high-affinity clones in the same GC (54). We find that unspecific GC B cells shared many of the features of antigen-specific GC B cells, namely they are elicited by immunization with antigen and not by adjuvant alone, they possess similar frequencies of V(D)J mutations that have replacement:silent (R:S) ratios indicative of selection, and they form extended clonal lineages indicating clonal proliferation (38). These characteristics rule out the possibilities that these cells are unspecifically activated by the inflammatory properties of the adjuvant, that they are not accidental recruits into the active GC response, and that they are free of phenotypic selection. Significantly, the unspecific GC B cells elicited by immunization with rHA or rPA exhibited patterns of VH gene usage that were distinct from their antigen-binding counterparts and dissimilar between the two pools of unspecific GC B cells. This observation lead to our conclusion that unspecific GC B cells are not elicited by the (adjuvant-mediated) release of self-antigens. Indeed, except for their lack of specificity, these unspecific GC B cells share the characteristics of antigen-selected GC B cells (38).
Two non-mutually exclusive hypotheses could explain the existence of late GC B cells whose BCRs have no detectable avidity for native immunogen: 1) the BCRs (and the IgG Abs secreted in Nojima cultures) are specific for native immunogen but bind with avidities too low to be detected in our ELISA and Luminex assays, or; 2) the BCRs are specific for epitopes not present in native immunogen. The first hypothesis is plausible in that low-affinity interactions between antigen and surface-bound BCR provoke intracellular signaling, and recruit B cells to early GCs, even when high concentrations of the same BCR do not detectably bind antigen as soluble IgG (23, 32, 55). Additionally, in the absence of higher-affinity competition, B cells with very low affinity for NP can initiate GCs, undergo SHM, and differentiate into memory B cells (22, 23, 56). Thus, very-low-affinity B cells are intrinsically competent for antigen-driven activation and GC participation; traditionally, the absence of such clones late in humoral responses to haptens has been interpreted as lack of fitness in the presence of high-affinity competitors. However, in light of recent observations (38, 54) indicating that complex antigens elicit late GCs that are both clonally diverse and permissive for BCR avidities spanning a ≥1000-fold range – or even with no measurable affinity – it is plausible that the GC response may recruit and tolerate B cells expressing an extraordinary range of BCR affinities.
Another precedent for antigen-driven activation of unspecific B cells is the non-classical humoral response to Salmonella typhimurium (Stm), in which large numbers of plasmablasts accumulate at extrafollicular sites, while GC formation is greatly delayed (57). A recent study of the anti-Stm extrafollicular response reported that the vast majority (>95%) of early plasmablasts appeared to be unspecific for Stm (58). At later time points, Stm-specific Abs were more frequently recovered, and a correlation between SHM and Stm-specificity was determined. When the Stm-specific Abs were stepwise-reverted to the germline sequence, their affinity for antigen progressively decreased until binding was very weak or undetectable in in vitro assays. Therefore, high frequencies of B cells with unmeasurably low avidity for antigen can participate in humoral responses and later acquire reasonable numbers of mutations to achieve highly-avid antigen binding.
The alternative hypothesis, that GC B cells with no detectable binding to native antigen are, in fact, responding to some modified form of antigen, is equally interesting. We have called the agency that recruits and selects unspecific GC B cells “dark antigen,” because, like dark matter, it has not been directly observed but its presence can be inferred by its effects. That unspecific GC B cells are responders to dark antigen is supported by our observations that these cells 1) are not induced by alum alone, 2) do not react with common autoantigens, 3) do not react with the immunizing antigen that has been denatured by heat or chaotropic agents, and 4) use biased VH gene distributions that are immunogen-specific, but distinct from the VH gene sets used by GC B cells that bind immunogen (38). Although we cannot rule out the possibility that dark antigen responses represent B-cell populations that recognize the native immunogen with unmeasurably low affinity, our working hypothesis is that dark antigen represents neo-epitopes derived from native antigen by in vivo processes (38). Once generated, dark antigen would function and behave as native antigen by activating and recruiting populations of specific B cells into GCs and there driving clonal competition and affinity maturation. As dark antigen-specific B cells are theoretically no different from any other GC B cell, we propose that they can probably differentiate into fully functional memory B cells and plasma cells.
Potential sources for dark antigen
The question of how dark antigen might be generated is interesting to consider. It is generally assumed that immunization or infection presents the immune system with antigen in primarily native conformation – this assumption is the foundation of immunological specificity. However, surprisingly little is known about the nature of antigen in vivo. In the field of autoimmunity, post-translational modification (PTM) of self-proteins – by phosphorylation, citrullination, proteolytic cleavage, etc. – is widely reported to generate neo-epitopes to which there is no immunological tolerance. These neo-autoantigens promote the induction and progression of many autoimmune diseases [reviewed in (59, 60)]. Inflammation increases the frequency of many PTMs in the extracellular environment, presumably by the release of enzymes from dying cells (59, 60). We are considering, therefore, whether the inflammation and local cell death that accompanies immunization with adjuvants or infection might release antigen-modifying enzymes into the extracellular milieu, and thereby create dark antigen by the generation of neo-epitopes. Alternatively, it is possible that these conditions for the generation of dark antigen exist constitutively, perhaps in association with the antigen-presenting cells that traffic antigen into secondary lymphoid organs. A third possibility is that dark antigen represents neo-epitopes created by complement fixation. Regardless of mechanism, dark antigen is targeted by a substantial fraction of GC B cells and this unanticipated and cryptic process merits close investigation. We have begun biochemical and genetic experiments to probe the nature of dark antigen.
Possible roles for immunity to dark antigen
If dark antigen is in fact neo-epitopes displayed by modified antigens, an important question arises: could Abs to dark antigen provide benefit? Could these Abs offer protection to the host?
For dark antigen to be useful in generating protective humoral immunity, generation of dark antigen must occur not only in the context of antigen/alum vaccination, but also during infection; otherwise, dark antigen has little or no biological consequence. To the best of our knowledge, a dark humoral response has not been observed during studies of pathogen infection. Nevertheless, we see no compelling reason that production of dark antigen would occur solely during artificial vaccination. Second, if dark antigen is generated by the modification of native antigen, this modification must act on pathogens themselves, otherwise Ab to dark antigen could have no effect on pathogen survival or proliferation. However, if substantial modification of microbial antigens occurs at the site of active infection, e.g. due to complement fixation or enzymatic activity around dying cells; Abs to dark antigen neo-epitopes will likely offer a protective benefit. This prediction suggests a simple experiment: determining whether passively transferred Abs to dark antigen offer any protection against the pathogen (e.g., rHA from influenza virus) that elicited the dark antigen response.
Dark antigen might impact the immune response in other ways. An interesting prospect is that Abs to dark antigen might possess advantageous cross-reactivities if directed to common PTMs. For example, dark antigen memory B cells specific for a citrulline-containing epitope in a viral envelope protein could also be activated by weak cross-reactivity with a citrulline-containing epitope in a bacterial protein. This might provide accelerated protection against new threats. We emphasize here that these thoughts about dark antigen are purely speculative. Much additional study will be necessary to demonstrate dark antigen directly and to clarify its immunologic functions.
Concluding remarks
Recent advances in single-cell-based assays have freed studies of B-cell repertoire analysis from a reliance on the use of BCR transgenic mice or hapten conjugates that activate B cells bearing genetically restricted BCRs directed to limited antigenic epitopes. Like hapten responses, GC responses to complex antigens drive substantial affinity maturation; however, unlike hapten responses, these GC responses retain clonal diversity of B cells, and are permissive for the presence of low affinity clones – including B cells with undetectable affinity to native conformations of immunogens (i.e. dark antigen-responsive B cells). Biologically, this permissive nature of clonal selection in GCs might be beneficial in preserving not only those memory B cells that are specific to eliciting antigens but also those that cross-react with variants of the antigen (e.g., different subtypes of viruses). These memory B cells might rapidly respond to subsequent challenge by heterologous antigens. Fully understanding the clonal selection and affinity maturation operating during GC reactions will be important for developing effective vaccination strategies to evolving infectious agents, such as influenza and HIV-1.
Acknowledgments
This work was supported in part by NIH awards AI100645, AI117892, and AI089618 (to G.K.).
Footnotes
Conflict of interests
The Authors declare there are no potential conflicts of interest.
References
- 1.MacLennan IC. Germinal centers. Annual Review of Immunology. 1994;12:117–39. doi: 10.1146/annurev.iy.12.040194.001001. [DOI] [PubMed] [Google Scholar]
- 2.Berek C, Berger A, Apel M. Maturation of the immune response in germinal centers. Cell. 1991;67(6):1121–9. doi: 10.1016/0092-8674(91)90289-b. [DOI] [PubMed] [Google Scholar]
- 3.Jacob J, Kelsoe G, Rajewsky K, Weiss U. Intraclonal generation of antibody mutants in germinal centres. Nature. 1991;354(6352):389–92. doi: 10.1038/354389a0. [DOI] [PubMed] [Google Scholar]
- 4.Jacob J, Przylepa J, Miller C, Kelsoe G. In situ studies of the primary immune response to (4-hydroxy-3-nitrophenyl)acetyl. III. The kinetics of V region mutation and selection in germinal center B cells. J Exp Med. 1993;178(4):1293–307. doi: 10.1084/jem.178.4.1293. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Takahashi Y, Dutta PR, Cerasoli DM, Kelsoe G. In situ studies of the primary immune response to (4-hydroxy-3-nitrophenyl)acetyl. V. Affinity maturation develops in two stages of clonal selection. Journal of Experimental Medicine. 1998;187(6):885–95. doi: 10.1084/jem.187.6.885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Mesin L, Ersching J, Victora GD. Germinal Center B Cell Dynamics. Immunity. 2016;45(3):471–82. doi: 10.1016/j.immuni.2016.09.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.De Silva NS, Klein U. Dynamics of B cells in germinal centres. Nat Rev Immunol. 2015;15(3):137–48. doi: 10.1038/nri3804. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Garside P, Ingulli E, Merica RR, Johnson JG, Noelle RJ, Jenkins MK. Visualization of specific B and T lymphocyte interactions in the lymph node. Science. 1998;281(5373):96–9. doi: 10.1126/science.281.5373.96. [DOI] [PubMed] [Google Scholar]
- 9.Okada T, Miller MJ, Parker I, et al. Antigen-engaged B cells undergo chemotaxis toward the T zone and form motile conjugates with helper T cells. PLoS Biol. 2005;3(6):e150. doi: 10.1371/journal.pbio.0030150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Coffey F, Alabyev B, Manser T. Initial clonal expansion of germinal center B cells takes place at the perimeter of follicles. Immunity. 2009;30(4):599–609. doi: 10.1016/j.immuni.2009.01.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Liu YJ, Zhang J, Lane PJ, Chan EY, MacLennan IC. Sites of specific B cell activation in primary and secondary responses to T cell-dependent and T cell-independent antigens [published erratum appears in Eur J Immunol 1992 Feb;22(2):615] Eur J Immunol. 1991;21(12):2951–62. doi: 10.1002/eji.1830211209. [DOI] [PubMed] [Google Scholar]
- 12.MacLennan IC, Liu YJ, Johnson GD. Maturation and dispersal of B-cell clones during T cell-dependent antibody responses. Immunol Rev. 1992;126:143–61. doi: 10.1111/j.1600-065x.1992.tb00635.x. [DOI] [PubMed] [Google Scholar]
- 13.Muramatsu M, Kinoshita K, Fagarasan S, Yamada S, Shinkai Y, Honjo T. Class switch recombination and hypermutation require activation-induced cytidine deaminase (AID), a potential RNA editing enzyme. Cell. 2000;102(5):553–63. doi: 10.1016/s0092-8674(00)00078-7. [DOI] [PubMed] [Google Scholar]
- 14.Victora GD, Schwickert TA, Fooksman DR, et al. Germinal center dynamics revealed by multiphoton microscopy with a photoactivatable fluorescent reporter. Cell. 2010;143(4):592–605. doi: 10.1016/j.cell.2010.10.032. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Han S, Hathcock K, Zheng B, Kepler TB, Hodes R, Kelsoe G. Cellular interaction in germinal centers. Roles of CD40 ligand and B7-2 in established germinal centers. J Immunol. 1995;155(2):556–67. [PubMed] [Google Scholar]
- 16.Shulman Z, Gitlin AD, Weinstein JS, et al. Dynamic signaling by T follicular helper cells during germinal center B cell selection. Science. 2014;345(6200):1058–62. doi: 10.1126/science.1257861. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Gitlin AD, Mayer CT, Oliveira TY, et al. HUMORAL IMMUNITY. T cell help controls the speed of the cell cycle in germinal center B cells. Science. 2015;349(6248):643–6. doi: 10.1126/science.aac4919. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kepler TB, Perelson AS. Cyclic re-entry of germinal center B cells and the efficiency of affinity maturation. Immunology today. 1993;14(8):412–5. doi: 10.1016/0167-5699(93)90145-B. [DOI] [PubMed] [Google Scholar]
- 19.Kepler TB, Perelson AS. Somatic hypermutation in B cells: an optimal control treatment. Journal of Theoretical Biology. 1993;164(1):37–64. doi: 10.1006/jtbi.1993.1139. [DOI] [PubMed] [Google Scholar]
- 20.Liu YJ, Joshua DE, Williams GT, Smith CA, Gordon J, MacLennan ICM. Mechanism of antigen-driven selection in germinal centres. Nature. 1989;342(6252):929–31. doi: 10.1038/342929a0. [DOI] [PubMed] [Google Scholar]
- 21.Liu YJ, Mason DY, Johnson GD, et al. Germinal center cells express bcl-2 protein after activation by signals which prevent their entry into apoptosis. Eur J Immunol. 1991;21(8):1905–10. doi: 10.1002/eji.1830210819. [DOI] [PubMed] [Google Scholar]
- 22.Dal Porto JM, Haberman AM, Kelsoe G, Shlomchik MJ. Very low affinity B cells form germinal centers, become memory B cells, and participate in secondary immune responses when higher affinity competition is reduced. J Exp Med. 2002;195(9):1215–21. doi: 10.1084/jem.20011550. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Shih TA, Meffre E, Roederer M, Nussenzweig MC. Role of BCR affinity in T cell dependent antibody responses in vivo. Nat Immunol. 2002;3(6):570–5. doi: 10.1038/ni803. [DOI] [PubMed] [Google Scholar]
- 24.Gitlin AD, Shulman Z, Nussenzweig MC. Clonal selection in the germinal centre by regulated proliferation and hypermutation. Nature. 2014;509(7502):637–40. doi: 10.1038/nature13300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Schwickert TA, Victora GD, Fooksman DR, et al. A dynamic T cell-limited checkpoint regulates affinity-dependent B cell entry into the germinal center. J Exp Med. 2011;208(6):1243–52. doi: 10.1084/jem.20102477. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Takahashi Y, Cerasoli DM, Dal Porto JM, et al. Relaxed negative selection in germinal centers and impaired affinity maturation in bcl-xL transgenic mice. J Exp Med. 1999;190(3):399–410. doi: 10.1084/jem.190.3.399. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Mayer CT, Gazumyan A, Kara EE, et al. The microanatomic segregation of selection by apoptosis in the germinal center. Science. 2017;358(6360) doi: 10.1126/science.aao2602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Hardin G. The competitive exclusion principle. Science. 1960;131(3409):1292–7. doi: 10.1126/science.131.3409.1292. [DOI] [PubMed] [Google Scholar]
- 29.Bothwell AL, Paskind M, Reth M, Imanishi-Kari T, Rajewsky K, Baltimore D. Heavy chain variable region contribution to the NPb family of antibodies: somatic mutation evident in a gamma 2a variable region. Cell. 1981;24(3):625–37. doi: 10.1016/0092-8674(81)90089-1. [DOI] [PubMed] [Google Scholar]
- 30.Allen D, Simon T, Sablitzky F, Rajewsky K, Cumano A. Antibody engineering for the analysis of affinity maturation of an anti-hapten response. Embo J. 1988;7(7):1995–2001. doi: 10.1002/j.1460-2075.1988.tb03038.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Weiss U, Rajewsky K. The repertoire of somatic antibody mutants accumulating in the memory compartment after primary immunization is restricted through affinity maturation and mirrors that expressed in the secondary response. J Exp Med. 1990;172(6):1681–9. doi: 10.1084/jem.172.6.1681. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Dal Porto JM, Haberman AM, Shlomchik MJ, Kelsoe G. Antigen drives very low affinity B cells to become plasmacytes and enter germinal centers. J Immunol. 1998;161(10):5373–81. [PubMed] [Google Scholar]
- 33.Benjamin DC, Berzofsky JA, East IJ, et al. The antigenic structure of proteins: a reappraisal. Annu Rev Immunol. 1984;2:67–101. doi: 10.1146/annurev.iy.02.040184.000435. [DOI] [PubMed] [Google Scholar]
- 34.Laver WG, Air GM, Webster RG, Smith-Gill SJ. Epitopes on protein antigens: misconceptions and realities. Cell. 1990;61(4):553–6. doi: 10.1016/0092-8674(90)90464-p. [DOI] [PubMed] [Google Scholar]
- 35.Wardemann H, Yurasov S, Schaefer A, Young JW, Meffre E, Nussenzweig MC. Predominant autoantibody production by early human B cell precursors. Science. 2003;301(5638):1374–7. doi: 10.1126/science.1086907. [DOI] [PubMed] [Google Scholar]
- 36.Wrammert J, Smith K, Miller J, et al. Rapid cloning of high-affinity human monoclonal antibodies against influenza virus. Nature. 2008;453(7195):667–71. doi: 10.1038/nature06890. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.von Boehmer L, Liu C, Ackerman S, et al. Sequencing and cloning of antigen-specific antibodies from mouse memory B cells. Nat Protoc. 2016;11(10):1908–23. doi: 10.1038/nprot.2016.102. [DOI] [PubMed] [Google Scholar]
- 38.Kuraoka M, Schmidt AG, Nojima T, et al. Complex Antigens Drive Permissive Clonal Selection in Germinal Centers. Immunity. 2016;44(3):542–52. doi: 10.1016/j.immuni.2016.02.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Su KY, Watanabe A, Yeh CH, Kelsoe G, Kuraoka M. Efficient Culture of Human Naive and Memory B Cells for Use as APCs. J Immunol. 2016;197(10):4163–76. doi: 10.4049/jimmunol.1502193. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.McCarthy KR, Watanabe A, Kuraoka M, et al. Memory B Cells that Cross-React with Group 1 and Group 2 Influenza A Viruses Are Abundant in Adult Human Repertoires. Immunity. 2018;48(1):174–84. e9. doi: 10.1016/j.immuni.2017.12.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Nojima T, Haniuda K, Moutai T, et al. In-vitro derived germinal centre B cells differentially generate memory B or plasma cells in vivo. Nat Commun. 2011;2:465. doi: 10.1038/ncomms1475. [DOI] [PubMed] [Google Scholar]
- 42.Luo XM, Maarschalk E, O’Connell RM, Wang P, Yang L, Baltimore D. Engineering human hematopoietic stem/progenitor cells to produce a broadly neutralizing anti-HIV antibody after in vitro maturation to human B lymphocytes. Blood. 2009;113(7):1422–31. doi: 10.1182/blood-2008-09-177139. [DOI] [PubMed] [Google Scholar]
- 43.Yeh CH, Nojima T, Kuraoka M, Kelsoe G. Germinal center entry not selection of B cells is controlled by peptide-MHCII complex density. Nat Commun. doi: 10.1038/s41467-018-03382-x. in press. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.DeKosky BJ, Kojima T, Rodin A, et al. In-depth determination and analysis of the human paired heavy- and light-chain antibody repertoire. Nat Med. 2015;21(1):86–91. doi: 10.1038/nm.3743. [DOI] [PubMed] [Google Scholar]
- 45.DeKosky BJ, Ippolito GC, Deschner RP, et al. High-throughput sequencing of the paired human immunoglobulin heavy and light chain repertoire. Nat Biotechnol. 2013;31(2):166–9. doi: 10.1038/nbt.2492. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Weinstein JA, Jiang N, White RA, 3rd, Fisher DS, Quake SR. High-throughput sequencing of the zebrafish antibody repertoire. Science. 2009;324(5928):807–10. doi: 10.1126/science.1170020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Friedensohn S, Khan TA, Reddy ST. Advanced Methodologies in High-Throughput Sequencing of Immune Repertoires. Trends Biotechnol. 2017;35(3):203–14. doi: 10.1016/j.tibtech.2016.09.010. [DOI] [PubMed] [Google Scholar]
- 48.Georgiou G, Ippolito GC, Beausang J, Busse CE, Wardemann H, Quake SR. The promise and challenge of high-throughput sequencing of the antibody repertoire. Nat Biotechnol. 2014;32(2):158–68. doi: 10.1038/nbt.2782. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Wang B, DeKosky BJ, Timm MR, et al. Functional interrogation and mining of natively paired human VH:VL antibody repertoires. Nat Biotechnol. 2018;36(2):152–5. doi: 10.1038/nbt.4052. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Schwickert TA, Lindquist RL, Shakhar G, et al. In vivo imaging of germinal centres reveals a dynamic open structure. Nature. 2007;446(7131):83–7. doi: 10.1038/nature05573. [DOI] [PubMed] [Google Scholar]
- 51.Zhang Y, Meyer-Hermann M, George LA, et al. Germinal center B cells govern their own fate via antibody feedback. J Exp Med. 2013 doi: 10.1084/jem.20120150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Lau D, Lan LY, Andrews SF, et al. Low CD21 expression defines a population of recent germinal center graduates primed for plasma cell differentiation. Sci Immunol. 2017;2(7) doi: 10.1126/sciimmunol.aai8153. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Victora GD, Nussenzweig MC. Germinal centers. Annu Rev Immunol. 2012;30:429–57. doi: 10.1146/annurev-immunol-020711-075032. [DOI] [PubMed] [Google Scholar]
- 54.Tas JM, Mesin L, Pasqual G, et al. Visualizing antibody affinity maturation in germinal centers. Science. 2016;351(6277):1048–54. doi: 10.1126/science.aad3439. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55.Lingwood D, McTamney PM, Yassine HM, et al. Structural and genetic basis for development of broadly neutralizing influenza antibodies. Nature. 2012;489(7417):566–70. doi: 10.1038/nature11371. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56.Shih TA, Roederer M, Nussenzweig MC. Role of antigen receptor affinity in T cell-independent antibody responses in vivo. Nat Immunol. 2002;3(4):399–406. doi: 10.1038/ni776. [DOI] [PubMed] [Google Scholar]
- 57.Cunningham AF, Gaspal F, Serre K, et al. Salmonella induces a switched antibody response without germinal centers that impedes the extracellular spread of infection. J Immunol. 2007;178(10):6200–7. doi: 10.4049/jimmunol.178.10.6200. [DOI] [PubMed] [Google Scholar]
- 58.Di Niro R, Lee SJ, Vander Heiden JA, et al. Salmonella Infection Drives Promiscuous B Cell Activation Followed by Extrafollicular Affinity Maturation. Immunity. 2015;43(1):120–31. doi: 10.1016/j.immuni.2015.06.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Doyle HA, Mamula MJ. Autoantigenesis: the evolution of protein modifications in autoimmune disease. Current opinion in immunology. 2012;24(1):112–8. doi: 10.1016/j.coi.2011.12.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 60.van Venrooij WJ, van Beers JJ, Pruijn GJ. Anti-CCP antibodies: the past, the present and the future. Nat Rev Rheumatol. 2011;7(7):391–8. doi: 10.1038/nrrheum.2011.76. [DOI] [PubMed] [Google Scholar]