Privileged Antigen Presentation in Splenic B Cell Follicles Maximizes T Cell Responses in Prime-Boost Vaccination

This article is featured in In This Issue, p.4421

The CD8⁺ T cell subset plays an essential role in host defense against viruses, intracellular bacteria, and malignancies. Generation of protective CD8⁺ T cell immunity has been a central focus for the development of Ag-specific vaccines. Although numerous strategies have been designed to elicit high frequencies of circulating Ag-specific CD8⁺ T cells, sequential immunizations with viral vectors (conventionally known as prime-boost immunization) is one of the most effective (1). The acute phase of the primary response is dominated by highly differentiated CD8⁺ effector T cells (T_EFF) that display robust cytotoxicity and production of inflammatory cytokines, necessary for initial control of fatal infections and other severe diseases. However, T_EFF have a limited capacity to proliferate and booster immunizations need to optimally stimulate memory T cells, especially central memory T cells (T_CM), which proliferate rapidly and vigorously upon encounter with Ags (2). Although a small number of T_CM can be identified during the acute phase of a typical primary CD8⁺ T cell response, subsequent boosting is limited by pre-existing T_EFF that rapidly clear Ag-loaded cells, preventing adequate Ag presentation to T_CM (3–5). Thus, prime-boost immunizations in a prophylactic setting are typically spread across a period of weeks to months to allow time for the T_CM to dominate the available T cell pool and enable maximal secondary expansion. In the case of therapeutic vaccination for patients with lethal (e.g., Ebola) or chronic infection (e.g., HIV, hepatitis C virus) or cancer, where immediate and high levels of the T_EFF are needed for protection, boosting strategies that can bypass T_EFF-mediated negative feedback regulation would be desirable.

It is established that activation of naive T cells requires a coordinated process between migratory dendritic cells (DCs) from peripheral tissues and lymphoid-resident APCs (6–8), but little is known about the role of different APCs in Ag presentation to memory T cells. It was originally thought that memory T cells had a reduced threshold for activation and, thus, could be triggered to proliferate by both professional and nonprofessional APCs (9, 10). Recent work, however, has changed that perspective, as hematopoietic APCs have been shown to be required for secondary expansion (11); one study has suggested that DCs, in particular, may be required for maximal secondary responses (12). However, Ag-carrying DCs are sensitive to T_EFF-mediated killing and they can be rapidly eliminated prior to their engagement with memory cells in the secondary lymphoid organs (13–15).

Interestingly, we have recently demonstrated that vaccination with recombinant rhabdoviruses, such as vesicular stomatitis virus (VSV) and Maraba virus, could provoke massive secondary expansion of T cells as early as 4 d after priming or at the height of the primary CD8⁺ T cell response (7–14 d after priming) (16–18), suggesting that rhabdoviral boosting can overcome the T_EFF barrier, allowing Ag-loaded APCs to escape T_EFF-mediated killing. Understanding the mechanisms of Ag presentation that facilitate rapid boosting by rhabdoviruses will provide important information that can be used to design optimal vaccination strategies for rapid generation of protective responses.

In the present study, we demonstrate that a robust secondary CD8⁺ T cell expansion following booster immunization with VSV occurs primarily in the spleen. Inducing maximal secondary CD8⁺ T cell responses at the peak of the primary response requires i.v. delivery of VSV, which results in direct infection of follicular B cells and subsequent Ag presentation by neighboring DCs in the follicular region. Escape of Ag-loaded APCs from T_EFF-mediated killing is consistent with evidence that T_CM are located within B cell follicles in the spleen whereas T_EFF cannot traffic through follicular regions (19–21). These data reveal a distinct role of splenic APCs with regard to secondary CD8⁺ T cell activation, which may not only offer insights into engineering prime-boost vaccines for rapid expansion of CD8⁺ T cells but may also provide new fundamental mechanistic insight into the immune response to viremia.

C57BL/6 and BALB/c mice were purchased from Charles River Laboratories (Wilmington, MA). B6.PL-Thy1^a/CyJ (Thy1.1 congenic), B6.PL-Thy1^a/CyJ, B6.129S2-Igh-6^tm1Cgn/J (referred to as B^−/− mice), and B6/JiK^btm1D^btm1N12 (referred to as K^bD^b−/− mice) were obtained from The Jackson Laboratory (Bar Harbor, ME). CD11c–diphtheria toxin receptor (DTR) mice were bred in the Central Animal Facility at McMaster University. All animal experimentation was approved by McMaster University’s Animal Research Ethics Board and complied with the Canadian Council on Animal Care guidelines.

The recombinant human adenovirus (Ad)-BHG was a replication-deficient recombinant human serotype 5 Ad with no transgene inserted and served as a control vector. The Ad–human dopachrome tautomerase (hDCT) expressed the full-length hDCT gene. The Ad-SIINFEKL-luciferase (Luc) (Ad-SIIN) vector expressed the immunodominant epitope of chicken OVA (OVA_257–264) coupled to firefly luciferase. The Ad-gp33 expressed the immunodominant eptiope from the gp33 of lymphocytic choriomeningitis virus (LCMV). Recombinant VSV-hDCT and VSV-GFP have been described (22). VSV-SIIN (expressing OVA_257–264 and luciferase) and VSV-gp33 (expressing gp_33–41 from LCMV) were engineered by an identical construction strategy. The VSV-MT was a control vector lacking a transgene. Recombinant vaccinia virus (VV, Western Reserve strain) expressing SIINFEKL-Luc (VV-SIINFEKL-Luc) has been described (23). Recombinant Maraba virus expressing hDCT (Maraba-hDCT) was constructed based on the attenuated strain MG1 of Maraba virus (24).

K^b-restricted DCT (DCT_180–188, SVYDFFVWL) and OVA (OVA_257–264, SIINFEKL) peptides were synthesized by Pepscan Systems (Lelystad, the Netherlands). Using a peptide library of 15-mers spanning the full length of the DCT protein (Pepscan Systems) we identified several CD8⁺ T cell epitope–containing peptides, which were pooled for evaluating immune responses in BALB/c mice.

Abs and kits used for surface or intracellular staining were purchased from BD Biosciences (Mississauga, ON, Canada). The tetramers K^b-DCT_180–188-allophycocyanin, K^b-OVA_257–264-allophycocyanin, and D^b-gp_33–41-allophycocyanin were obtained from Baylor College of Medicine.

Immunized mice received i.p. injections of 1 mg BrdU and BrdU-containing drinking water (0.8 mg/ml) 24 h prior to harvesting tissues. To block T cell trafficking from lymphoid organs in some experiments, mice also received an i.p. injection of FTY720 (Cayman Chemical, Ann Arbor, Michigan) at 4 mg/kg body weight 24 h before boosting with VSV. After blocking Fc receptors, lymphocytes in single-cell suspensions from various tissues were stained with a tetramer and then Abs against surface CD8 and intranuclear BrdU (BD Biosciences).

Negatively selected splenic CD8⁺ T cells (Stemcell Technologies, Vancouver, BC, Canada) from Ad-SIIN–immunized congenic (Thy1.1⁺) mice were sorted into T_EFF (CD127⁻CD62L⁻), effector memory T cell (T_EM, CD127⁺CD62L⁻), and T_CM (CD127⁺CD62L⁺) populations. Sorted subsets (>90% purity) were adoptively transferred into naive wild-type (WT) C57BL/6 recipients (Thy1.2⁺) 1 d prior to vaccination with VSV-SIIN. Five days after administration of VSV, blood, bone marrow, and spleens were harvested to determine the frequency of congenic tetramer⁺ cells.

C57BL/6 mice received 1 × 10⁹ PFU VSV-GFP i.v. Spleens were harvested 1.5 h later to minimize the potential for the confounding effect of cell trafficking between tissues, while providing enough time for VSV to infect cells. Splenocytes were cultured for an additional 4.5 h to allow GFP to accumulate within infected cells for detection by flow cytometry.

For depletion of B cells, mice were given 250 μg anti-CD20 mAb (clone 18B12, Biogen Idec, San Diego, CA) by the i.v. route 7 d prior to vaccination and 2 wk later to maintain depletion. By flow cytometry, >98% of B cells were removed from the blood and spleen. Sham-treated mice received an isotype control.

C57BL/6 recipient mice were irradiated (2 × 550 rads; 48-h interval) and then received i.v. injections of T cell–depleted bone marrow–derived cells. Four sets of chimeric mice were made: group 1 received 5 × 10⁶ WT cells; group 2 received 4 × 10⁶ cells from B^−/− mice and 2 × 10⁶ cells from WT mice; group 3 received 4 × 10⁶ cells from B^−/− mice and 2 × 10⁶ cells from K^bD^b−/− mice; and group 4 received 5 × 10⁶ cells from B^−/− donors. CD11c-DTR chimeric mice were similarly prepared by transferring CD11c-DTR transgenic mouse-derived bone marrow into lethally irradiated C57BL/6 recipients. Mice were given 3 mo to reconstitute their hematopoietic systems, which was confirmed by flow cytometry prior to experimentation.

C57BL/6 or K^bD^b−/− mice were injected i.v. with 1 × 10⁹ VSV-GFP or VSV-gp33. Seventeen hours later, splenic B cells were purified using negative selection kits according to the manufacturer’s instructions (Miltenyi Biotec, Teterow, Germany). Naive P14 cells were enriched by positive selection from the spleens of P14 mice (EasySep mouse CD8a positive selection kit, Stemcell Technologies) and labeled with 1 mM CFSE (Sigma-Aldrich, Oakville, ON, Canada). Bone marrow–derived DCs were generated and matured as described previously (25).

Purified WT B cells (2 × 10⁷) were cocultured with 2 × 10⁶ bone marrow–derived DCs for 18 h at 37°C. DCs were separated using CD11c microbeads (Miltenyi Biotec) and cultured with CFSE-labeled P14 cells at a 1:1 ratio. Cultures were analyzed for proliferation after 72 h. Additionally, 5 × 10⁶ K^bD^b−/− B cells were directly mixed with 5 × 10⁵ CFSE-labeled P14 cells in a 24-well plate or separated by Transwell inserts (Corning, Tewksbury, MA) in the presence or absence of 5 × 10⁵ DCs for 72 h prior to flow cytometric analysis.

Mice primed with Ad-hDCT received splenectomies or sham surgeries 12 d later. At 48 h after surgery, mice were boosted with 1 × 10⁹ PFU i.v. VSV-hDCT. Five days after VSV, DCT_180–188-specific CD8⁺ T cell responses were quantified by flow cytometric analysis of intracellular cytokine staining following in vitro peptide restimulation.

GraphPad Prism for Windows 6.0 (GraphPad Software, San Diego, CA) was used for graphing and statistical analyses. When required, data were normalized by log transformation. A Student two-tailed t test, Mann–Whitney U test, or one- or two-way ANOVA was used to query immune response data. Differences between means were considered significant at p ≤ 0.05. Means with SE bars are shown. Survival data were analyzed using the Kaplan–Meier method and the log-rank test.

The choice of i.v. administration of rhabdoviruses (i.e., VSV and Maraba) in our previous studies was driven by our desire to exploit their oncolytic potential to target both primary and metastatic tumors (16, 18). Our observation that i.v. injection of rhabdoviruses also resulted in a rapid and robust secondary expansion of Ag-specific CD8⁺ T cells prompted us to determine whether i.v. delivery was a particularly efficient route for booster vaccinations. Mice were immunized with an Ad vector expressing an OVA-derived CD8⁺ T cell epitope (OVA_{257– 264}, SIINFEKL) coupled to luciferase (Ad-SIIN) (26). Fourteen days following immunization, OVA-specific CD8⁺ T cells had expanded to 12% of total circulating CD8⁺ T cells (Fig. 1A, lower panel) and primarily displayed a T_EFF phenotype (CD62L^loCD127^loKLRG1^higranzyme B^hi; data not shown). Despite the high frequencies of T_EFF, i.v. administration of a VSV vector encoding the same Ag (VSV-SIIN) provoked a massive expansion of OVA-specific cells, reaching a mean of 88% of circulating CD8⁺ T cells (Fig. 1A, lower panel) or a 58-fold increase by absolute numbers (Fig. 1A, upper panel). However, other routes, including i.m., i.p., intranasal, and s.c., were not as efficient as i.v. (Fig. 1A, upper panel).

We next examined the route dependence of the boosting immunization using another Ag. In this case, mice were primed with an Ad expressing the weakly immunogenic melanocyte-differentiation Ag DCT (Ad-hDCT) and boosted with a VSV expressing the same Ag (VSV-hDCT) (22). Interestingly, not only was i.v. injection the best route for the booster vaccination, it was the only route that generated a significant boost against this self-epitope (Fig. 1B). Similar boosting responses were achieved with Maraba virus, suggesting that rhabdoviruses share similar biological properties (Fig. 1C). Intravenous administration of VSV-hDCT also induced a dramatic expansion of CD8⁺ T cells in BALB/c mice, indicating that VSV-mediated boosting was not strain specific (Fig. 2A). We also determined that the VSV boost was effective in the same time frame (i.e., 7–14 d after primary immunization) in mice primed with DCs or VV expressing the same Ag (Fig. 2B, 2C). Thus, the remarkable potency of the VSV boost was not limited to a single strain, Ag, or priming method but did require i.v. delivery of the boosting virus.

The therapeutic relevance of this prime-boost approach was demonstrated by an aggressive murine melanoma model (B16) where a single vector alone only slightly slowed down the growth of pre-established B16 tumors (5 d old) and mice reached the endpoint rapidly in 12 d after treatment (Fig. 2D). However, priming animals bearing 5-d-old tumors with either Ad or DCs was sufficient to allow a boost by VSV 9 d after priming, leading to complete tumor regression (Fig. 2D).

To learn more about the mechanism of the VSV-mediated boost, we first determined the locations of secondary expansion of T cells following i.v. inoculation of VSV. Mice were primed with Ad-hDCT and boosted 14 d later by i.v. VSV-hDCT. As a control, a subset of the primed mice received only VSV-MT, which does not express any transgene. The frequency of DCT-specific CD8⁺ T cells was assessed in various lymphoid tissues on days 1, 2, 3, 4, and 7 following administration of VSV. To monitor local proliferation, BrdU was administered to the mice for 24 h prior to lymphocyte harvest. As shown in Fig. 3A, the proliferative response peaked around days 3–4 following boosting and returned to baseline by day 7. Notably, the earliest evidence of a proliferative response was observed in the spleen at day 2 after boosting, suggesting that the secondary expansion began in the spleen somewhere between 24 and 48 h after boost. Ultimately, proliferation was observed in all tissues, likely as a consequence of migratory activated CD8⁺ T cells. To further confirm that the spleen was the primary site where the proliferation of secondary T cells started, we carried out two more experiments. First, we treated mice with FTY720 to inhibit lymphocyte egress from lymphoid organs during boosting, and T cell proliferation was monitored by BrdU incorporation. Compared to the VSV-MT control, the earliest proliferation of hDCT-specific CD8⁺ T cells was evident only in the spleen 48 h after boosting with VSV-hDCT (Fig. 3A, 3B, Supplemental Fig. 1). Second, spleens in Ad-hDCT–primed mice were surgically removed 2 d before VSV-hDCT boosting (i.e., 12 d after priming) and secondary expansion was determined on day 5 after boosting. Data in Fig. 3C show that CD8⁺ T cell expansion was abrogated in the absence of a spleen, reinforcing a critical role of this organ in initiating the secondary T cell responses.

The population of Ad-induced CD8⁺ T cells within the spleen on the day of the boost was composed primarily of T_EFF, although a small population of T_CM was also detectable (Fig. 4A). To determine which CD8⁺ T cell subset was contributing to the secondary expansion, OVA_257–264-specific splenic CD8⁺ T cells (Thy1.1) were sorted into T_EFF, T_EM, and T_CM subsets based on CD127 and CD62L expression (Fig. 4B) (27). These subsets were adoptively transferred into naive congenic recipients (Thy1.2) that were subsequently immunized with VSV-SIIN. Owing to the low frequencies of Ag-specific CD8⁺ T cells in the T_CM population, we did not attempt to normalize the number of Ag-specific T cells that were transferred. Rather, we normalized the total number of adoptively transferred cells using naive splenocytes to ensure that equal numbers of total T cells were delivered to each recipient (10⁶ cells per mouse). Tetramer staining indicated that each bolus of 10⁶ cells contained 1.9 × 10⁵ OVA_257–264-specific T_EFF, 6.6 × 10⁴ T_EM, or 4 × 10³ T_CM. Frequencies of adoptively transferred cells (Thy1.1) were determined 5 d after boosting (Fig. 4C). The transgene-specific T_CM were the primary responders, as they underwent dramatic expansion 5 d after exposure to VSV-SIIN (i.e., 135-, 425-, and 206-fold increases in frequency relative to day −1 in the spleen, bone marrow, and blood, respectively). In contrast, T_EM contributed little to the expansion (∼50% increase over input). As expected, T_EFF did not contribute to the secondary expansion and actually contracted during the 5-d period. Our observation is consistent with previous reports that the T_CM possess the highest proliferative potential. The results also demonstrated that the T_CM subset was preferentially and effectively expanded by the VSV vaccine.

Previous reports have indicated that Ag presentation is rapidly shut off by CD8⁺ T_EFF through a feedback mechanism that involves killing Ag-loaded APCs (14, 28). To determine which cell types were infected by VSV, a vector expressing GFP (VSV-GFP) was injected i.v. into mice and spleens were harvested 90 min later to minimize the migration of infected cells between tissues. Single-cell suspensions were incubated for an additional 4.5 h to provide sufficient time for the VSV to express GFP and then infected cells were analyzed by flow cytometry (Fig. 5A). Strikingly, most infected cells within the spleen were B220⁺CD19⁺ B cells (∼82%; Fig. 5B). Further characterization of the VSV-infected B cells revealed most were CD21/35^intCD23^hiCD27⁻, which is indicative of follicular B cells (Fig. 5B) (29). Ag expression by B cells within the follicular region may represent a unique mechanism to avoid T_EFF-mediated clearance because T_EFF cannot circulate through this area (30). The GFP expression pattern in the spleen, together with the fact that T_CM were primarily reactivated in the spleen, appeared to suggest that follicular B cells might be serving as APCs for the secondary expansion.

To characterize the requirement of B cells in vivo for the secondary expansion of T_CM, we tested our prime-boost regimen in B cell–deficient (B^−/−) mice (31). WT and B^−/− mice were immunized with Ad-hDCT and boosted 14 d later with VSV-hDCT. The absence of B cells had no effect on the primary response to the DCT transgene carried by the Ad vector (Fig. 6A, 1°). In contrast, the secondary response to hDCT was severely attenuated in the absence of B cells (Fig. 6A, 2°), suggesting that B cells were required for secondary CD8⁺ T cell expansion. Because B cells are necessary for the organogenesis of lymphoid tissues (32), we repeated the same experiment in WT mice where B cells were depleted with an anti-CD20 mAb. Similar to the results in B^−/− mice, the ablation of B cells before boosting with VSV had a dramatic negative effect on the expansion of the memory DCT-specific T cells (Fig. 6B).

Having established a critical role for B cells in secondary expansion of T_CM, we next investigated whether B cells were required for Ag presentation. We generated chimeric mice using bone marrow cells from B^−/− and K^bD^b−/− mice to reconstitute lethally irradiated WT mice. In these chimeras, all B cells could only be derived from the K^bD^b−/− bone marrow whereas a large proportion of other APCs (from the B^−/− bone marrow) had intact MHC class I molecules (Supplemental Fig. 2). Similarly prepared recipient mice reconstituted with WT and/or B^−/− bone marrow were included as controls (Fig. 6C). Surprisingly, in contrast to the results in B^−/− and B cell–depleted mice, the secondary expansion was not attenuated when the B cells were unable to present Ag on MHC class I (Fig. 6C), suggesting that B cells do not have a direct role as an APC.

The above observations prompted us to determine whether splenic DCs were responsible for Ag presentation. To address this possibility, we generated chimeric mice with bone marrow from CD11c-DTR transgenic mice as described by Zammit et al. (12). These mice tolerated multiple injections of DT to remove CD11c⁺ DCs (Supplemental Fig. 3). To ensure that CD8⁺ T cells were not affected by DT treatment, we transferred Ag-primed CD8⁺ T cells from WT congenic mice to CD11c-DTR chimeric mice prior to boosting with VSV. Fig. 6D shows that injections of DT diminished expansion of endogenous T cells (Thy1.1⁻) in CD11c-DTR chimeric mice but not WT controls, suggesting that host DCs may mediate Ag presentation to trigger memory T cell proliferation. DT injections also reduced the secondary response of transferred WT CD8⁺ T cells (Thy1.1⁺), confirming the requirement of CD11c⁺ DCs instead of removal of certain CD8⁺ T cells that may also express CD11c (Fig. 6E) (33, 34).

Taken together, these results pointed to a possibility that infected B cells produced and transferred Ags to neighboring DCs, leading to Ag presentation to T_CM that were also localized in the follicular region (20, 35). To confirm this possibility, we isolated B cells from K^bD^b−/− mice 17 h after i.v. inoculation with VSV-gp33 and coincubated them with CFSE-labeled P14 T cells in the presence or absence of bone marrow–derived DCs. Three days after coculture, proliferation of P14 T cells was detected only in the presence of DCs, suggesting that B cell–produced Ags were taken by DCs and presented to T cells (Fig. 7A, upper panel). To confirm this, we used Transwell inserts to separate infected K^bD^b−/− B cells (top of well) from P14 T cells and DCs (bottom of well). P14 cell proliferation was observed only when DCs were present in the lower compartment, confirming that Ag was transferred from B cells to DCs in a soluble form (Fig. 7A, lower panel). To determine whether infected B cells produce viral progeny that reinfect DCs, we coincubated B cells from VSV-gp33–exposed K^bD^b−/− mice with Vero cells that are highly sensitive to VSV replication. After 48 h, Vero cells coincubated with B cells (Fig. 7B, middle panel) remained intact (compared with Vero cells alone; Fig. 7B, left panel), whereas cytopathic effect was visibly evident by direct infection with VSV (Fig. 7B, right panel), suggesting that B cells neither permit productive viral replication nor carry viruses on their surface.

T_EFF-mediated elimination of Ag-laden DCs has been widely accepted as a homeostatic mechanism to prevent overexuberant activation of CD8⁺ T cells (5, 14, 28). We report in the present study, however, that i.v. delivery of rhabdoviral vectors could overcome this negative regulation and maximize expansion of CD8⁺ T cells, even at the peak of the primary response. A unique cooperation between splenic B cells and DCs for Ag presentation and their anatomic location (within the follicle), which is separated from T_EFF but proximate to T_CM, appears to be a mechanism that enables engagement of T_CM during an ongoing effector phase in the presence of circulating viruses (i.e., viremia). Our data revealed a distinct role of splenic APCs during secondary T cell responses, which should be considered for the rational design of booster vaccines to rapidly and effectively expand Ag-specific CD8⁺ T cells. Our finding may also provide new fundamental mechanistic insight into the immune response to viremia where the primary CD8⁺ T cell response can be amplified to contend with an uncontrolled virus infection without compromising the regulatory feedback to DCs that limits further naive T cell recruitment and, presumably, minimizes inadvertent presentation of tissue Ags by migrating DCs.

Previous research has demonstrated that the secondary lymphoid organs are the primary sites of naive T cell priming following immunization or pathogen infection. Upon activation, T_EFF and T_EM enter the circulation and peripheral tissues, whereas T_CM home to the secondary lymphoid organs (36). This differential distribution poses a potential barrier to interaction between Ag-laden DCs and T_CM because cytolytic T_EFF kill migratory DCs in peripheral tissues (13, 15). Moreover, a recent study has shown that T_EFF can re-enter reactive lymph nodes and attenuate Ag presentation by killing newly arrived DCs and Ag-loaded residential DCs (37). These data clearly demonstrate that the efficacy of booster immunizations that rely on migratory DCs for Ag presentation will be limited by this T_EFF-mediated negative feedback mechanism, especially during primary and chronic immune responses.

In the spleen, CD8⁺ T_CM are located in the T cell zones of the white pulp surrounded by follicular B cells. Using TCR-transgenic CD8⁺ T cells, several groups have shown that after adoptive transfer to naive recipients, effector CD8⁺ T cells were only found in the red pulp, whereas memory CD8⁺ T cells homed to the B cell follicles (19, 30, 38). Subsequent studies confirmed that this localization pattern held true for endogenously activated CD8⁺ T cells. Lefrançois and colleagues (20) and Jacob and colleagues (21) showed that upon secondary exposure to LCMV or Listeria monocytogenes, memory CD8⁺ T cells rapidly expanded and mobilized from the B cell follicles to the red pulp via bridging channels. After the immune response subsided, memory CD8⁺ T cells homed back to the B cell follicles whereas effector cells remained in the splenic red pulp (20, 21). These results point to a possibility that follicular APCs were responsible for memory CD8⁺ T cell activation. Colocalization with T_CM in the follicles confers upon follicular APCs anatomical advantages to interact with T_CM while avoiding T_EFF-mediated killing. In contrast, migratory DCs, including the CD8α⁺ subset, primarily reside in the red pulp, making them highly susceptible to killing by T_EFF during secondary responses (39). Our results provide unequivocal evidence that this negative regulation can be bypassed when the Ag can be delivered into the follicular region. This discovery has important implications for optimizing therapeutic vaccination strategies where high numbers of T_EFF are needed for protection prior to boosting as well as in circumstances where T_EFF will inevitably be present for prolonged periods such as during chronic infections.

It was previously demonstrated that delivery of VSV into the flank leads to infection of the lymph nodes where most of the Ag is taken up by macrophages and DCs within the subcapsular sinus, so little of the virus gets into the B cell follicles (40). This may explain why immunization through discrete routes (e.g., s.c., i.m.), which result in Ag uptake through the lymphatics, are not effective routes for boosting. Additionally, the ability of VSV and Maraba virus to infect splenic B cells, especially follicular B cells, not only bypasses circulating T_EFF but also increases Ag production due to their enormous number and subsequent Ag presentation by neighboring DCs to T_CM, which are also located in the follicular region. Our in vitro analysis confirmed that infected B cells could transfer Ag to DCs without involving direct contact between the two cells although the exact form of the Ag remains to be determined. It was previously reported that B cells could carry viruses including VSV on their surface and transfer them to other cells (41, 42), but our results argue that the Ag is likely in a different soluble form.

Our data suggest a novel strategy for boosting immune responses in the presence of T_EFF by delivering Ags to a location that is anatomically separated from the T_EFF and in close proximity to T_CM. These findings are supported by recent publications revealing differential distributions of T_CM and T_EFF and offer important insights into how to rapidly boost immunity without the need to compromise the T_EFF population induced by primary immunization.

We thank Drs. R. Dunn and M. Kehry at Biogen Idec (San Diego, CA) for the gift of mAb against mouse CD20 (clone 18B12).

B.D.L. is a board member of Turnstone Biologics, which is developing Maraba virus as an oncolytic vaccine. The other authors have no financial conflicts of interest.