Adaptive B Cell Responses to Influenza Virus Infection in the Lung

Introduction

A hallmark of adaptive immune response is the long-term recall of previous encounters with viruses. A phenomenon called “immunological memory,” through which a primed host can promptly and vigorously eliminate reinvading viruses, is the basis of the protective immunity provided by vaccinations or previous infections (51). Immunological memory comprises antibody-mediated humoral responses and T cell-mediated cellular responses; however, this review focuses primarily on the humoral response mediated by memory B cell populations. The contribution of T cell-mediated cellular responses is reviewed elsewhere in this journal.

The precursors for memory B cells and plasma cells develop within germinal centers (GCs) in which somatic hypermutations and affinity-based selection take place (32,57). Although GC-derived B cell populations express somatically mutated antibodies with high affinity and specificity, a GC-independent pathway also produces B cell populations bearing germline-encoded antibodies with low affinity and specificity (26,58,64). The heterogeneity of these pathways is advantageous in creating B cell repertoires that are diverse in terms of affinity, specificity, and clonality because B cell responses must cover a variety of viruses that have different antigenic compositions, structures, and conformations. In addition, some viruses generate antigenic variations rapidly through mutations.

B cell responses are typically orchestrated in secondary lymphoid organs equipped with preorganized lymphoid structures; however, after pulmonary virus infection, the lungs begin to support T cell and B cell responses, which results in the ectopic formation of GCs and protective humoral responses at this site (40). Eliciting B cell responses at the sites of viral replication may be advantageous for two reasons. First, if virus-specific B cells are present at the site of infection, this anatomical advantage allows them to respond to infecting viruses faster than B cells in secondary lymphoid organs (42,49). Second, B cells at the site of infection have a greater chance of recognizing replicating viruses that maintain virus signatures (14). Virus signatures include highly dense B cell epitopes on viral membranes and viral DNA/RNA, which are recognized by B cells through B cell antigen receptors (BCRs) and Toll-like receptors (TLRs), respectively (2,22).

The direct sensing of virus signatures may accelerate the kinetics of B cell responses and enhances antibody affinity, as discussed later. After viral replication subsides, viral antigens persist more abundantly at the site of infection (1,29) and become selecting antigens to fine-tune B cell repertoires. Exposure to higher amounts of persisting antigens likely modifies the threshold of clonal selection, thereby influencing the affinity and specificity of B cell repertoires formed at the site (1,30). Supporting this hypothesis, respiratory infection by influenza viruses in both humans and mice has repeatedly been observed to increase the breadth of antibody response significantly relative to vaccination through other systemic routes (36,38). This outcome is partly due to the induction of polymeric IgA antibodies that have greater avidity to virus variants (55); however, compared with IgG antibody repertoires generated by systemic vaccination, those formed after respiratory infection are more reactive to virus mutants, which suggests that the clonal selection processes after infection and vaccination are different (36,38).

In this study, we compare several phases of B cell responses that take place in the lungs (the sites of viral replication) with those in secondary lymphoid organs after influenza virus infection and highlight unique aspects of lung B cell responses associated with protective immunity against acutely infectious and rapidly mutating influenza viruses.

Two Types of Long-Lived B Cell Populations

Humoral memory response is achieved by two types of long-lived B cell populations: long-lived plasma cells and memory B cells (51). Plasma cells serve as the first line of defense by continuously secreting virus-neutralizing antibodies that confer immediate protection at the time of reinfection. Memory B cells, on the contrary, cannot contribute to immediate protection, but promptly divide and differentiate into plasma cells that secrete large amounts of neutralizing antibodies after restimulation by viral antigens. Therefore, memory B cells have been considered backups for long-lived plasma cells when viral titers exceed the protective capacity of the plasma cells.

Although plasma cells secrete antibodies with high affinity and specificity and provide effective protection against homologous viruses, they easily lose their binding capacity for antigenically divergent viruses (67). On the contrary, several lines of evidence illustrate that antigen recognition by memory B cells is more adjustable to antigenic variations (17,45). Functional compartmentalization is considered to exist in both long-lived B cell populations: long-lived plasma cells provide effective protection against previously encountered homologous viruses by providing highly specific antibodies, and memory B cells are more specialized to respond to future infection by mutant viruses. Understanding how memory B cells acquire broad specificity is an important topic for both basic and translational research because it involves the rational basis for the development of universal vaccines that can provide broad protection against rapidly mutating viruses, such as influenza viruses and human immunodeficiency viruses (7,15,25,59).

Early Phase of Adaptive B Cell Responses to Influenza Virus Infection

It is difficult to depict a generalized view of B cell responses to all virus types because they differ in antigen structure, entry routes, cellular tropisms, and pathogenesis. In this article, we focus on mouse B cell responses to influenza virus infection for which we have substantial information about antigen structure, virus entry routes, cellular targets, and innate and acquired immune responses. Under nonlethal infectious conditions, virus replication is largely restricted to the RT, including the lung, and does not usually cause viremia because the enzyme required for hemagglutinin (HA) cleavage is absent at sites other than the RT (28).

Viral titers in the RT typically peak within 1 week of infection and decline rapidly to levels below the detection limit of standard plaque-forming assays (24,29,69). However, viral antigens persist in the lungs and draining lymph nodes (LNs) for extended periods in several forms after influenza virus infection. First, viral antigens are present as Ag/MHC complex that can be recognized by transferred CD4 and CD8 T cells (24,29,69). Second, viral antigens are deposited for the maintenance of GCs that are preferentially long lived in the lungs and draining LNs (1). Third, as reflected in the detection of viral RNA at 1 month after infection in the lungs (1,29), continual viral replication appears to last for an extended period in this organ, where several unique aspects of B cell responses may be triggered by the persistent antigens.

In this study, we refer to the phase that lasts <1 week after infection as the acute phase and the phase that persists for >2 weeks after infection as the chronic phase. During the acute phase, viral antigens are actively transported from infected lungs to draining LNs by dendritic cells (33), which leads to the initial priming of B cells as late as 48 h after infection in the LNs (10). After priming by viral antigens, activated B and T cells migrate toward the borders of B cell follicles and T cell zones, respectively, enabling B cells to receive helper signals from cognate CD4⁺ T cells (3,52). Activated B cells migrate to the outer follicles, actively proliferate, and participate in either GC-independent or GC-dependent pathways for differentiation into long-lived B cell populations comprising plasma cells and memory B cells during the acute phase (26,37,58,64).

Compared with B cells participating in the GC-dependent pathway, those participating in the GC-independent pathway are characterized by a lack of somatic hypermutations/affinity maturation and earlier development (<1 week), as they skip the time-consuming GC reaction (26,37,58,64). Thus, the early GC-independent pathway is committed to supplying B cells with low affinity and specificity in the acute phase, whereas the GC-dependent pathway produces B cells with high affinity and specificity in the chronic phase and these cells gradually join memory compartment (56).

The early GC-independent pathway is proposed to recruit low-affinity, broadly reactive (polyreactive) B cells into memory compartments and increase the breadth of humoral protection (26,27,64). GC-independent memory B cells are clearly detected in secondary lymphoid organs after haptenated protein immunization (26,64). In addition, influenza vaccines stimulate the GC-independent pathway that induces long-lived plasma cells in secondary lymphoid organs and provides protective antibodies against homologous virus infection (37). Likewise, influenza virus infection appears to activate this pathway because unmutated memory B cells are detectable in secondary lymphoid organs (1). These data suggest that influenza viruses and vaccines promptly induce memory B cell populations through the early GC-independent pathway, at least in secondary lymphoid organs; however, whether this pathway contributes to the induction and breadth of memory B cell populations at the site of infection remains to be clarified.

Presumably, owing to the absence of preprogrammed lymphoid structures, B cell responses in the lungs are not initiated as fast as those in draining LNs (1,6,42,50). In mice maintained under conditions free of specific pathogens, B and T cells constitute ∼30% of lung cells, but those lymphocytes in naive lungs are within vasculature rather than in parenchyma, as most of them are stained by intravascularly injected antibodies (18,43). Due to the absence of lung-resident lymphocytes in naive lungs, antigen-reactive B and T cells are presumably recruited into the lungs through circulation and then accumulate into CD11c⁺ cell-rich areas (20). Thereafter, B cells begin to form aggregates and, by day 10 after infection, develop into tertiary lymphoid structures containing B cell- and T cell-rich areas around high endothelial venules (40). This tertiary lymphoid structure, known as induced bronchus-associated lymphoid tissue (iBALT), is formed not only during infections but also as a result of chronic inflammation and autoimmune diseases (46). Mediators of iBALT formation include lymphotoxin (35), several homeostatic chemokines (16,48), and IL-17 (16,47). IL-1α released from dying alveolar macrophages has been shown to initiate iBALT formation by promoting dendritic cell migration and follicular helper T (Tfh) cell development in response to inhaled fine particles (31). Notably, several studies suggest that influenza viruses inoculated through the RT infect alveolar macrophages and induce cell death (12). Therefore, IL-1α from alveolar macrophages may be a key initiator of iBALT formation after influenza virus infection as well.

Although iBALT structures vary depending on the virus strain and infection conditions, typical structures include B cell follicles and T cell-rich zones that support the initiation of B cell responses within the site of infection (20,40). After infection, the lungs harbor plasma cells and memory B cells for an extended period (42,65); however, the early wave of memory B cell development, as observed in the secondary lymphoid organs during the acute phase, does not occur in the lungs and, instead, memory B cells begin appearing in the chronic phase only after GCs emerge (42). Consistent with this feature of delayed appearance, unmutated memory B cells are rare in the lungs (1), further supporting the conclusion that the GC-independent pathway is attenuated in B cell responses in the lung. However, it is crucial to mention that a previous study focused on isotype-switched B cells (1), and therefore the contribution of the GC-independent pathway in the generation of IgM⁺ memory B cells remains to be determined.

GC Phase of Adaptive B Cell Responses to Influenza Virus Infection

GCs offer a unique microenvironment for the selection of B cells with the appropriate affinity and specificity to viral antigens and recruitment into either memory B cells or plasma cells (61). Within GCs, the clonal evolution of B cells takes place in a Darwinian process comprising somatic diversification and affinity-based selection. Many experimental systems mirror GCs that are transiently formed in the secondary organs after systemic immunization with nonreplicating protein antigens; therefore, these systems are insufficient for the dissection of local GCs that form ectopically in the lungs after influenza virus infections. We initially summarize canonical GC responses in secondary lymphoid organs and then point out the unique characteristics of lung GC responses that have been recently illustrated.

Imaging studies using genetically restricted and monoepitope-specific responses established the basis of GC dynamics in secondary lymphoid organs (3,62). GC response is governed by iterative cycles of somatic diversification in the dark zone and clonal selection in the light zone. The rapid clonal expansion of antigen-specific B cells in the dark zone is accompanied by BCR diversification through somatic hypermutation. The hyperproliferation of GC B cells was traditionally monitored by the uptake of thymidine analogs (e.g., 5-bromodeoxyuridine and ethynyl deoxyuridine) incorporated into DNA during active DNA synthesis with a short incubation period (19). Cells that exit the cell cycle relocate to the light zone, in which affinity-based selection takes place through competitive interactions with antigens on follicular dendritic cells and Tfh cells that express B cell activating cell surface molecules (e.g., CD40L) and cytokines (e.g., IL-21). Some of the B cell clones that receive abundant Tfh-derived helper signals return to the dark zone for repeat division and mutation regulated by c-myc and AP1 (8,9,13). As a result of these iterative cycles, the average number of somatic hypermutations and the affinity for priming antigens increase time dependently in proportion to the duration of the GC responses (30).

Concomitant with iBALT formation, GC structures, including follicular dendritic cells and Tfh cells, begin developing in the lungs (40). Although details about the cellular dynamics and regulatory mechanisms of lung GCs is limited, in part, by the technical difficulty of imaging lung GCs, the steady increase in somatic hypermutations in lung GCs over time demonstrates that, similar to GCs in secondary lymphoid organs, lung GCs can support the iterative cycles of GC reactions (1). IgV sequence analysis of iBALT in patients with chronic obstructive pulmonary disease provides further evidence of ongoing somatic hypermutations at these sites (60). Notably, we recently found that compared with splenic GCs, lung GCs more frequently incorporate ethynyl deoxyuridine in association with persistent antigen deposits (1), which suggests that cellular dynamics may be accelerated in lung GCs. Although the exact mechanisms underlying these accelerated cellular dynamics remain unknown, it is interesting to speculate that greater amounts of viral antigens or infection-induced cytokine milieu, or both, near the site of viral replication impact B cell dynamics in the lungs.

The most striking feature of lung GC responses is the enhanced selection for B cell repertoires that bind antigenically subdominant, but conserved, epitopes in influenza HA proteins. HA proteins have two distinct antigenic domains, globular head domain and stem domain (54). The globular head domain easily accumulates mutations that enable viral escape from protective antibodies, whereas the stem domain includes relative conserved regions necessary for transmission and replication (11). Once B cells target the conserved functional regions in stem domains, they acquire broad reactivity to a wide spectrum of mutant viruses; however, the viruses have a number of strategies for obscuring the conserved domains from antibody surveillance, with the functional result being that the conserved domains are antigenically subdominant.

By monitoring the ratios of broadly reactive GC B cells, we recently demonstrated that half of the HA-binding GC B cells in the lungs became reactive to antigenically divergent HA by targeting stem domains, whereas only 5% of splenic GC B cells bound the same HA variant (1). Broadly reactive lung GC B cells were then recruited into lung memory compartments with help of Bach2 transcriptional factor that instructs GC B cells into the memory pools (53). Weak signals from Tfh cells are the key determinants for increasing Bach2 expression and directing cells into the memory compartment (53). It is conceivable that broadly reactive B cells are guided into the memory compartment by weak Tfh signals, as the viral conserved domains are antigenically subdominant and poorly accessible to BCRs, which would reduce the amount of antigen uptake and presentation by B cells, and eventually attenuate helper signals from Tfh cells. Somatic genetics analysis of the IgV genes of broadly reactive memory B cells in the lungs showed a high number of somatic hypermutations (1), which are consistent with the results of mutational analysis of broadly reactive monoclonal antibodies from human memory B cells or memory-derived plasmablasts (44,66). Thus, the mouse B cell strategy of acquiring broad reactivity at local sites and targeting viral conserved domains through the GC-dependent pathway is likely generalized to humans. Although the contribution of the GC-independent pathway to the acquisition of broad reactivity through other strategies cannot be excluded (27), the available data suggest that the GC-dependent pathways are more critical in providing broad protection against influenza virus infection at local sites and the distinct GC selection in the lungs is one of the key events providing infection-induced humoral protection against a wide spectrum of viral mutants.

Current strategy for developing universal influenza vaccines focuses on a new design of HA-conserved domains with better antigenicity, which elicit high amounts of broadly reactive antibodies (23,68). The studies using animal models demonstrated the promising results; however, it is important to understand immunological aspects of broadly reactive memory responses for future application into vaccine strategies. In this context, dissecting the lung GC components required for the selection of broad reactivity may provide important clues for the immunological aspects. Especially, follicular dendritic cells and Tfh cells, crucial components for GC selection, need to be analyzed under the lung GC microenvironment. The information, combined with newly designed HA antigens, would contribute to the elicitation of broadly reactive memory responses and maximize the efficacy of universal influenza vaccines.

Post-GC Phase of Adaptive B Cell Responses to Influenza Virus Infection

The location of memory B cells and the B cell extrinsic factors arising from that location significantly influence how promptly and robustly memory B cells respond to virus reinfection. These properties of memory B cell responses are crucial to disease outcome, particularly when rapidly replicating and cytopathic influenza viruses invade. Broadly reactive memory B cells from lung GCs are preferentially localized in the lungs as tissue-resident cells (1). Although marginal zones and B cell follicles have been identified as the lodging sites of splenic memory B cells (4,26), only limited experimental data are available for the intraorgan localization of lung memory B cells. Histological data from chronic obstructive pulmonary disease patients suggest that CD27⁺ memory phenotype B cells are located in B cell follicles in iBALT (60). Impaired recall antibody responses to reinfecting viruses through the disruption of the preexisting iBALT structure also suggest that iBALT contributes to lung memory B cell responses (18). However, the possibility that memory B cells reside in other areas of inflamed lungs cannot be excluded because lung memory B cells are detectable long after iBALT formation subsides (39,42).

Due to their proximity to virus entry and replication sites, lung B cells have greater access to live replicating viruses with highly organized virus particle structures, whereas access by cells in other organs is limited. Indeed, a study using somatic cell transfer technology to develop a mouse strain expressing HA-binding BCRs demonstrated that virus binding B cells in the lungs, but not in the secondary lymphoid organs, has direct access to infecting viruses through BCRs (14). In this mouse model, virus-binding naive B cells in the lungs were targeted by viral infection, which caused cell death and subsequently delayed primary antibody responses at local sites. On the contrary, a study using nontransgenic mice repeatedly infected with influenza viruses demonstrated that compared with B cells in secondary lymphoid organs, lung memory B cells were reactivated promptly and differentiated more quickly into plasma cells after secondary infection (42). These results suggest that unknown mechanisms may protect lung memory B cells from infection-induced cell death, as observed in lung-resident memory T cells (63). Furthermore, these data indicate that the lung localization of memory B cells facilitates direct access to infecting viruses with intact virus particles, which allows lung memory B cells to receive stimulatory signals through virus signatures such as high-density HA epitopes and TLR agonists.

One significant outcome of direct stimulation by intact virus particles is B cell reactivation in the absence of CD4⁺ T cells. T-independent (TI) B cell activation is evident from in vitro B cell stimulation by virus particles in the absence of CD4 T cells (34). Likewise, the injection of virus particles has been shown to stimulate donor-derived, virus-specific memory B cells in recipient mice lacking T cells (5,21,41). Thus, virus-specific memory B cells are potentially reactivated, without T cell interaction, when they capture intact virus particles in the lung. Mechanistically, TI reactivation of memory B cells depends on high-density B cell epitopes in vesicular stomatitis virus models, and it is largely lost in response to poorly organized vesicular stomatitis virus antigens displayed on infected cells (5). Moreover, we recently found that TI reactivation of memory B cells for influenza vaccines is completely dependent on TLR7 signaling in memory B cells (41). Compared with T-dependent reactivation, which requires time for T cell interaction, TI reactivation of memory B cells has the clear advantage of supplying virus-neutralizing antibodies more quickly. In addition, we found that the TI pathway selectively reactivates high-affinity memory B cells (41), which supplies antibodies with higher affinity than those from the T-dependent pathway. Together, the direct contact between infecting viruses and lung memory B cells could facilitate the faster production of high-affinity antibodies through the TI pathway at the site of infection, which would significantly improve outcomes in illness caused by acutely infectious and cytopathic influenza viruses.

Conclusions

Influenza viruses have threatened human populations for generations. The most effective protection against such infections is the induction of protective antibodies through memory B cell responses. B cells are the only cell types that abundantly express antigen receptors and TLR for viral DNA/RNA, a property that may have been acquired through coevolution with several types of viruses. After an infection with viruses that replicate in nonlymphoid organs, B cells lose the opportunity to interact directly with the virus structure during the initial encounter; however, the ectopic development of cellular niches promotes GC responses and generates memory B cell populations through the GC pathway at sites of virus replication. Infection-induced immune modulation affects the clonal selection within lung GCs, which highly selects broadly reactive B cells and recruits them into memory B cells. During reinfection, these memory B cells potentially sense the structural signatures of replicated viruses, supply high-affinity plasma cells promptly and without assistance from T cells, and contribute to broad protection (Fig. 1). Key details remain unknown about the mechanism of selection for broadly reactive B cells within lung GCs and the types of antigens, follicular dendritic cells, and Tfh cells involved in the selection process. This information will shed new light on B cell biology and inform rational strategies for the development of universal vaccines that provide broad protection against influenza viruses.

OPEN IN VIEWER

FIG. 1.

Adaptive B cell responses to influenza virus infection. After primary influenza virus infection, adaptive B cell responses occur not only in secondary lymphoid organs but also in iBALT in the lungs, the site close to virus replication. Memory B cell populations develop through GC-dependent and GC-independent pathways in secondary lymphoid organs, whereas they appear to develop through the GC-dependent pathway with the help of Bach2 in the lungs. By possibly using a greater amount or different conformation of antigens in the presence of follicular dendritic cells and Tfh cells, lung GCs highly select and recruit broadly reactive repertoires to become memory B cells, which respond promptly to reinfecting viruses and provide protective antibodies at the site of infection, likely in a TI cell manner. GC, germinal center; iBALT, induced bronchus-associated lymphoid tissue; Tfh, follicular helper T; TI, T independent.