Main

Memory CD8+ T cells enhance resistance to infection by intra- cellular pathogens, and generation of these cells is an important goal of vaccination. Experimental models, including infection of mice with Listeria monocytogenes, show that increasing numbers of memory CD8+ T cells provide better immunity to infection1,2,3. These data suggest that effective vaccines should rapidly generate high numbers of memory CD8+ T cells. But multiple booster immunizations are often needed to reach protective numbers of memory cells4. In most vaccine regimens, including those applied to humans, a substantial interval between the initial and booster immunizations leads to the greatest amplification of immunological memory.

Because of phenotypic and functional heterogeneity, no single property can be used to identify memory CD8+ T cells. Operationally, memory T cells persist, are capable of vigorous secondary proliferation and mediate protective immunity against pathogen rechallenge. Much effort is currently directed at understanding the pathway(s) leading to the development of CD8+ T cells with memory characteristics. After infection or vaccination, antigen-specific CD8+ T cells undergo a program of extensive proliferation, lasting for 7–10 d, which is followed by a precisely regulated contraction phase in which the number of antigen-specific CD8+ T cells decreases by 90%2,5,6,7,8. The remaining antigen-specific CD8+ T cells form the initial pool of memory cells, which may remain stable in number for the life of the host. But a period of more than 40 d after infection is required to stabilize gene expression in memory T cells9, suggesting that full differentiation of memory T cells is a relatively slow process.

Data from experimental models indicate that memory CD8+ T cells present at the end of the contraction phase show so-called 'effector memory' phenotype and function (also known as Tem cells), associated with rapid induction of effector functions and a circulatory pattern that excludes lymph nodes10. In contrast, at later time points after initial stimulation, the population of memory CD8+ T cells in the circulation and secondary lymphoid organs show a predominantly 'central memory' phenotype (also known as Tcm cells) that includes expression of surface CD62L and CCR7, molecules that permit trafficking into lymph nodes. Notably, the ability to undergo vigorous secondary expansion after antigen challenge in vivo, a characteristic of memory T cells, roughly parallels the appearance of central memory T cells. Consistent with this notion, central memory T cells proliferate more vigorously than effector memory T cells in response to antigenic stimulation and provide better protection against some11, but perhaps not all12 microbial challenges. These data suggest that booster immunizations administered long after initial immunization may generate the highest number of secondary memory cells11,13.

Accelerating the rate at which memory CD8+ T cells are generated after immunization would enhance vaccine efficacy by decreasing the interval required to elevate the number of memory T cells by booster vaccinations. But the factors that govern the rate at which CD8+ T cells acquire memory characteristics are undefined. Here, we evaluated the rate of development of memory CD8+ T cells after vaccination with mature, peptide-coated DCs. We show that DC immunization, in the absence of overt inflammation, accelerates the generation of antigen-specific CD8+ T cells with the function and phenotype of memory cells. Notably, these early memory cells were able to respond vigorously to booster immunizations administered within days of the initial vaccination.

Results

Early booster response after DC immunization

Robust CD8+ T-cell responses and long-lasting protective immunity are generated after sublethal infection of mice with L. monocytogenes or immunization with lipopolysaccharide-matured, bone marrow–derived DCs coated with L. monocytogenes peptides14,15,16. Infection with 0.1 lethal dose in 50% of animals (LD50) of virulent L. monocytogenes or immunization with 2 × 105 LLO91–99 peptide–coated mature DCs (DC-LLO) generated similar numbers of LLO91–99–specific CD8+ T cells in BALB/c mice during the expansion, contraction and memory phases of the immune response (Fig. 1a). Booster infection with 1.0 LD50 of L. monocytogenes at day 40 after initial L. monocytogenes infection (LM+LM) resulted in increased secondary memory cell numbers compared to mice that received only the first or second infection (Fig. 1b). But, as previously observed17, the same booster L. monocytogenes infection at day 4, 6 or 11 after initial L. monocytogenes infection did not increase number of memory CD8+ T cells (Fig. 1b). In marked contrast, the same booster infection at day 4, 6 or 11 after initial DC immunization (DC+LM) resulted in six- to tenfold more memory cells than achieved with the booster infection alone (Fig. 1b). Thus, compared to L. monocytogenes infection, DC immunization substantially shortened the interval required for amplification of the number of antigen-specific memory CD8+ T cells by booster infection.

Figure 1: Accelerated response to booster infection after DC-immunization.
figure 1

(a,b) BALB/c mice were immunized with virulent L. monocytogenes (1 × 103) or DC-LLO and boosted with virulent L. monocytogenes (1 × 104) at various days. (a) Total number of LLO91–99–specific T cells/spleen after initial L. monocytogenes infection or DC-LLO immunization. (b) Total number of LLO91–99–specific T cells/spleen 40 d after booster-infection. ns, not significant; **P < 0.01. (c–e) Mice were immunized with virulent L. monocytogenes (1 × 103), uncoated DC (DC-none), or with DC-LLO on day 0 and boosted with virulent L. monocytogenes (1 × 104) at day 6. (c) Bacterial numbers on day 3 and day 5 after booster infection. One of three mice in LM+LM group had bacteria at day 3. LOD, limit of detection. (d) The percent of IFN-γ+ CD8+ T cells from representative mice in the presence (upper number) or absence (lower number) of LLO91–99 peptide stimulation. (e) Total number/spleen of LLO91–99–specific T cells. Data are mean + s.d. for three mice per group.

DC vaccination accelerates secondary response potential

The increased number of secondary memory cells in DC+LM mice was not a result of delayed clearance of the booster infection compared to LM+LM mice (Fig. 1c). To determine the pathway resulting in elevated memory cell numbers, LLO91–99–specific CD8+ T cells were enumerated at various days in DC-LLO+LM, LM+LM and DC-none+LM mice (Fig. 1d,e). Antigen-specific CD8+ T cells in LM+LM mice increased about two- to threefold between days 6 and 9 and then underwent a pronounced contraction to stable memory cell numbers that were not elevated compared to mice receiving a single infection with L. monocytogenes (Fig. 1e). In contrast, antigen-specific CD8+ T cells in DC-LLO+LM mice underwent substantial secondary expansion, reaching peak numbers at day 11 that were 25-fold higher than those found at day 6. Antigen-specific CD8+ T cells in DC-LLO+LM mice were then reduced in number; however, the resulting memory cell number was 12-fold higher than in LM+LM mice, and remained stable for at least 100 d (data not shown). Amplification of the number of effector and memory T cells did not occur in mice receiving DC without antigen (Fig. 1d,e) and did not result from trapping of antigen-specific cells in the spleen, as they were observed in all tissues analyzed (Supplementary Fig. 1 online). Finally, boosting DC-immunized mice with as few as 100 individual L. monocytogenes (0.01 LD50) increased the number of memory T cells, whereas boosting with 100,000 individual L. monocytogenes elevated memory T-cell numbers by >20-fold (Supplementary Fig. 2 online). Thus, antigen-specific CD8+ T cells at day 6 after DC immunization are capable of substantial secondary expansion and trafficking to tissues in response to even very weak booster infection.

Enhanced resistance in DC-LLO+LM mice

Antigen-specific CD8+ T cells at day 68 after booster infection of DC-LLO+LM and LM+LM mice showed a memory phenotype (CD44hi, interleukin (IL)-7 receptor–positive18,19,20, CD43 (1B11)lo21) and the same fraction of cells produced both IFN-γ and tumor necrosis factor (TNF) or IL-2 after antigen stimulation11 (Fig. 2a–c). DC-LLO+LM mice contained approximately fivefold more memory T cells in this experiment (Fig. 2b), which allowed these mice to control and survive high-dose (100 LD50) L. monocytogenes infection that was lethal in naive and LM+LM mice (Fig. 2d,e). Therefore, the antigen-specific T cells at day 68 in DC-LLO+LM mice show a memory phenotype and their increased numbers conferred a substantial survival advantage in response to high-dose infection.

Figure 2: Increased numbers of memory phenotype CD8+ T cells in DC+LM mice enhance protective immunity.
figure 2

(a) BALB/c mice were immunized with virulent L. monocytogenes (1 × 103) or DC-LLO and boosted with virulent L. monocytogenes (1 × 104) on day 6. At day 6 + 68 these mice and naive (control) mice were challenged with 1 × 106 virulent L. monocytogenes. (b) Frequency of LLO91-99–specific CD8+ T cells from representative mice at day 6+68. (c) Phenotypic and functional status of IFN-γ+ CD8+ T cells at day 6 + 68. (d) Bacterial numbers (mean + s.d., three mice per group) in the spleen and liver on day 2 after challenge. Double asterisk indicates mice that succumbed to infection. Percentages indicate percent decrease in bacteria from DC-LLO+LM compared to LM+LM mice. (e) Percent survival after challenge.

Multiple regimens boost DC-immunized mice

It is unlikely that booster immunizations in humans would be given with virulent L. monocytogenes. We therefore subjected DC-immunized mice to a range of booster immunizations including: vaccinia virus expressing the LLO91–99 epitope22 (Fig. 3a,b), attenuated actA-deficient L. monocytogenes23 (Fig. 3c,d), and LLO91–99–coated syngeneic spleen cells (Fig. 3e,f), all of which generated substantially elevated numbers of memory cells 40 d later compared to infected mice given the same booster immunization. The ability of LLO91–99–coated spleen cells to boost at day 6 after DC immunization but not L. monocytogenes infection (Fig. 3e,f) indicates that antibody24,25 or other concurrent T-cell responses do not prevent the secondary response in mice infected with L. monocytogenes. Together, the results suggest a fundamental difference in the developmental stage of antigen-specific T cells at day 6 after DC immunization compared to infection with L. monocytogenes. In addition, these data show that the accelerated secondary response of CD8+ T cells in DC-immunized mice can be stimulated by infectious and noninfectious booster immunizations.

Figure 3: Amplified secondary memory in DC-immunized mice in response to multiple boosting regimens and against weak antigens.
figure 3

BALB/c mice were immunized with (a) vaccinia virus expressing LLO (VV-LLO) (c,e,g) virulent (Vir) L. monocytogenes (1 × 103) (a,c,e) DC-LLO, or (g) p60449–457–coated DC. On day 6 mice were boosted with (a) VV-LLO (c) actA-deficient L. monocytogenes (1 × 107) (e) LLO91–99–coated splenocytes (2 × 107/mouse), or (g) Vir L. monocytogenes (1 × 104). (b,d,f) Total number per spleen of LLO91–99–specific, or (h) p60449–457–specific CD8+ T cells. Data are mean + s.d. from three mice per group. Fold increase in total numbers of antigen-specific CD8+ T cells in DC+boost at days 40–45 is indicated.

DC immunization permits boosting of weak responses

Many vaccines stimulate weak CD8+ T-cell memory, and prime-boost regimens are required to generate sufficient memory cells to achieve protective immunity4. To determine whether DC immunization would permit early boosting against weak antigens, we infected mice with L. monocytogenes or vaccinated with DC coated with p60449–457, a subdominant L. monocytogenes antigen16, and boosted 6 d later with L. monocytogenes (Fig. 3g,h). p60449–457–specific memory T cells were at or below the level of detection at day 40 after the boost in mice that initially received L. monocytogenes. In contrast, p60449–457–specific CD8+ T cells in DC-immunized mice increased by 300-fold by day 5 after booster infection and these mice generated >40-fold higher numbers of memory cells than those achieved in LM+LM mice. Similar amplification of weak memory responses also occurred after vaccination with DC coated with the H2-M3 (class Ib)-restricted f-MIGWII epitope from L. monocytogenes15,26 (Supplementary Fig. 3 online). Thus, DC priming and early boosting can be used to rapidly amplify number of effector and memory T cells, even in response to weak antigens.

DCs accelerate memory CD8+ T-cell generation

The ability to undergo vigorous secondary expansion, as observed early after DC immunization but not L. monocytogenes infection, is a characteristic of memory T cells. Consistent with this notion, antigen-specific CD8+ T cells at day 6 after DC immunization show a memory phenotype9 (CD44hi, CD127hi, CD43(1B11)lo) and 40% produced IL-2 after antigen stimulation (Supplementary Fig. 4 online), whereas antigen-specific CD8+ T cells at day 6 after L. monocytogenes infection show an effector phenotype1,27,28 (CD44hi, CD127lo, CD43(1B11)hi) and no IL-2 production after antigen stimulation (Supplementary Fig. 4 online). To confirm that early DC-stimulated T cells possess a memory phenotype, we vaccinated L. monocytogenes–immune mice (containing memory L. monocytogenes–specific T cells) with lymphocytic choriomeningitis virus NP118–126–coated DCs and determined the phenotype of day 75 L. monocytogenes–stimulated memory cells and day 5 DC-stimulated T cells (DC-NP) in the same immune mice. Both populations show similar memory phenotype (CD44hi, CD127hi, CD43 (1B11)lo) and >30% produced IL-2 after antigen stimulation, including high levels of CD27 expression, another marker of functional memory cells29 (Fig. 4a,b). Again, the phenotype of DC-stimulated cells differed substantially from L. monocytogenes–stimulated effector cells (Fig. 4c). Although not further explored here, we noted that DC-stimulated CD8+ T cells underwent normal contraction beginning at day 7 (Fig. 1a), even though the majority of the cells express the IL-7 receptor at day 6 (Fig. 4b). These data were not expected in light of recent results suggesting that T cells expressing high levels of the IL-7 receptor survive contraction18,30,31.

Figure 4: Rapid generation of memory CD8+ T cells and vigorous secondary expansion after booster infection of DC-immunized L. monocytogenes–immune hosts.
figure 4

(a) Naive or L. monocytogenes–immune (day 75 after infection with 1 × 106 individual L. monocytogenes deficient in the gene encoding actA) BALB/c mice were immunized with DC-NP and boosted at day 5 with virulent L. monocytogenes expressing NP118–126 (LM-NP; 1 × 104). (b) Phenotypic and functional status of NP118–126– or LLO91–99–specific CD8+ T cells at day 5 after DC-NP immunization. (c) Phenotypic and functional status of LLO91–99–specific CD8+ T cells at day 6 after virulent L. monocytogenes infection (1 × 103) of naive mice. (d) Frequency of NP118–126– or LLO91–99–specific CD8+ T cells from representative mice at the indicated days after initial (day 5) and booster (day 5+5) immunizations. (e) Total number/spleen (mean + s.d., three mice per group) of NP118–126–, LLO91–99– or p60217–225–specific CD8+ T cells. Fold increase in total numbers of NP118–126–specific CD8+ T cells at day 5 after booster immunization is indicated.

Finally, the day 5 DC-NP–stimulated CD8+ T cells in both naive and L. monocytogenes–immune mice were able to undergo vigorous secondary expansion after booster infection with L. monocytogenes expressing the NP118–126 epitope32. In fact, secondary expansion was similar in DC-NP–immunized naive (29-fold) and L. monocytogenes–immune (34-fold) mice (Fig. 4d,e) despite the presence of other L. monocytogenes–specific CD8+ T cells in immune mice that also underwent secondary expansion (Fig. 4d,e). Thus, DC-stimulated early memory cells can be boosted even in the presence of preexisting immunity. Together, these data show that DC immunization generates antigen-specific CD8+ T cells with both the phenotype and function of true memory cells within 5–6 d after immunization.

DC immunization could either directly generate memory cells or accelerate the transition from effector to memory cells. To resolve these possibilities, we transferred sufficient naive Thy1.1 OT-1 T-cell receptor–transgenic CD8+ T cells33 (specific for ovalbumin (OVA) 257–264) into naive Thy1.2 hosts to allow analysis at various days after infection with L. monocytogenes expressing OVA (LM-OVA34) or immunization with OVA257–264–coated DC (DC-OVA). Notably, the majority of OT-1 cells at day 6 after DC immunization showed a memory phenotype (CD44hi, CD127hi, CD43 (1B11)lo) and produced IL-2, whereas the OT-1 cells at day 6 after infection with L. monocytogenes showed an effector phenotype (CD44hi, CD127lo, CD43(1B11)hi) and did not produce IL-2 (Fig. 5 and Supplementary Fig. 5 online), thus recapitulating the phenotype and function showed by endogenous populations of T cells after DC immunization or infection (Fig. 4b,c). In contrast, the majority of OT-1 cells at day 3 after infection with L. monocytogenes or DC immunization showed an effector phenotype (CD44hi, CD127lo, CD43 (1B11)hi) and did not produce IL-2 (Fig. 5). Therefore, DC immunization did not directly generate memory phenotype CD8+ T cells, but instead substantially accelerated the transition from effector cells into cells with memory phenotype and function.

Figure 5: DC immunization accelerates the transition of CD8+ T cells from an effector to memory phenotype.
figure 5

Purified naive OT-I (Thy1.1) cells were transferred into C57BL/6 (Thy1.2) mice and 1 d later mice were immunized with actA-deficient LM-OVA (1 × 106) or DC-OVA. CD8+ Thy1.1+ cells in the spleens were analyzed at the indicated days. Nonimmunized recipient mice were used as controls (day 0). Percent of CD8+ Thy1.1+ cells that expressed the indicated molecules or produced IL-2 after antigen stimulation is indicated. Data are mean ± s.d. for three to five mice per group and are representative of three experiments.

Inflammation controls the rate of memory cell generation

DC subsets are thought to have specific roles in the immune response35,36,37, and accelerated generation of memory CD8+ T cells in DC-immunized mice could be the result of a specific property of the in vitro–matured DCs. Notably, preliminary analysis showed rapid elevation of IL-1β, granulocyte-macrophage colony-stimulating factor (GM-CSF), IFN-γ, TNF, IL-6, IL-12, granulocyte colony-stimulating factor (G-CSF), MIP-1α and RANTES protein in serum by 20 h after L. monocytogenes infection; however, these inflammatory mediators were not elevated in the serum of DC-immunized mice (data not shown), suggesting that accelerated generation of memory CD8+ T cells may also result from a lack of infection-induced inflammatory signals during T-cell priming.

To address these possibilities, we injected mice with LLO91–99–coated DCs and/or virulent L. monocytogenes carrying an epitope-destroying mutation at residue 92 of LLO38 and at day 6, boosted all groups with wild-type L. monocytogenes. Thus, LLO91–99–specific T cells were primed by the injected DCs, in the presence or absence of L. monocytogenes infection. L. monocytogenes infection prevented the generation of LLO91–99–specific CD8+ T cells able to undergo secondary expansion after booster infection (Supplementary Fig. 6 online), suggesting that the rate of memory cell generation was determined by whether priming occurred in the presence or absence of infection. Truncating L. monocytogenes infection by antibiotic treatment at day 2, which causes clearance of L. monocytogenes by day 3 versus day 6–7 in control mice5, did not change the effector phenotype of antigen-specific T cells at day 7 or permit these cells to undergo secondary expansion after booster immunization (Supplementary Fig. 6 online). Thus, the duration of infection does not determine whether CD8+ T cells can respond to early booster immunization. Again, these data suggest a fundamental difference in the antigen-specific CD8+ T cells stimulated by DC vaccination versus L. monocytogenes infection.

L. monocytogenes infection results in substantial activation of the innate immune system, including the production of proinflammatory cytokines such as IL-12 and type I and type II IFN16. To induce these same inflammatory mediators in the absence of infection, we immunized mice with LLO91–99–coated DCs with or without CpG oligodeoxynucleotide 1826 (refs. 39,40). The CpG treatment did not alter the magnitude of the LLO91–99–specific CD8+ T-cell response at day 6 after DC immunization (Fig. 6a) or the ability of mice to clear the booster L. monocytogenes infection (Fig. 6b). But CpG treatment substantially decreased the fraction of antigen-specific CD8+ T cells with memory phenotype (CD127hi, CD43lo; Fig. 6c). Although only a modest decrease in the percentage of IL-2–producing T cells occurred in CpG-treated mice (Fig. 6c), these cells were unable to respond to booster immunization (Fig. 6a). Thus, CpG treatment prevents accelerated generation of memory CD8+ T cells and early prime-boost response.

Figure 6: Inflammation prevents accelerated generation of memory CD8+ T cells and early prime-boost.
figure 6

(a c) BALB/c mice received DC-LLO alone (w/o CpG) or with CpG (w/CpG) and were boosted with L. monocytogenes (1 × 104) on day 6. One of four experiments is shown. (a) Total number and fold increase of LLO91–99–specific T cells in spleen at the indicated days. (b) Bacterial numbers in organs on day 2 after boost. (c) Percent of LLO91–99–specific T cells expressing CD127, CD43 and IL-2. (d) C57BL/6 (Thy1.2) mice received no immunization, DC-OVA alone (w/o CpG) or with CpG (w/CpG) and were injected with naive OT-I cells (Thy1.1; 5 × 105) on the indicated days. Results are the frequency of OT-I Thy1.1 cells in spleens at day 3 after injection. (e,f) BALB/c mice received DC-LLO alone (w/o CpG) or with CpG on day 0 (w/CpG d0) or day 3 (w/CpG d3) and all mice were boosted with virulent L. monocytogenes (1 × 104) on day 7. One of two experiments is shown. (e) Phenotypic and functional status of LLO91–99–specific T cells at day 7 after DC-LLO immunization. (f) Total number and fold increase of LLO91–99–specific T cells in spleen. (g) C57BL/6 (WT) and Ifngr2−/− mice were immunized with DC-OVA alone (w/o CpG) or with CpG (w/CpG) and were boosted with actA-deficient LM-OVA (5 × 106) on day 6. Total number and fold increase of OVA257–264–specific T cells in spleen. Data in a–d,f and g are mean + s.d., three mice per group.

DCs matured in vitro with lipopolysaccharide, CpG or lipopolysaccharide+CpG all stimulated memory phenotype CD8+ T cells at day 6 after immunization, and these T cells underwent vigorous secondary responses to L. monocytogenes booster immunization (Supplementary Fig. 7 online). Thus, it is unlikely that CpG treatment directly altered DC function to prevent early formation of the memory T-cell pool. CpG treatment could alter the lifespan of the injected DCs, and thus mimic infection by increasing the duration of antigen presentation. But there was no difference in DC recovery between control and CpG-treated mice at various days after immunization (DCs became undetectable at 48 h after injection; data not shown) and CpG treatment did not alter the duration of DC antigen display, detected by proliferation of naive OT-1 T-cell receptor–transgenic cells injected at various days after DC-OVA immunization (Fig. 6d). Thus, CpG treatment does not alter the lifespan of the injected DC or increase the duration of DC-mediated antigen presentation in vivo.

It remained possible that the CpG treatment was acting indirectly through the injected DCs to prevent early generation of antigen-specific CD8+ memory T cells. To address this, DC-immunized mice received no CpG, CpG at day 0 or CpG at day 3, a time point at which the DCs are no longer detectable and DC-mediated antigen presentation is substantially reduced (Fig. 6d). Notably, CpG treatment as late as day 3 after DC immunization prevented the generation of CD8+ T cells with early memory phenotype (Fig. 6e) and inhibited the response to booster immunization as well as CpG treatment on day 0 (Fig. 6f). These data show that CpG does not act through the injected DCs to prevent generation of early memory CD8+ T cells.

Because T cells do not express Toll-like receptor 9 (ref. 41), the results suggest that CpG treatment prevents early memory generation by inducing inflammatory cytokines that act to alter the program of the responding T cells. To determine whether CpG-stimulated IFN-γ39,40 has a role in preventing early generation of memory T-cell pool after DC immunization, we vaccinated wild-type or IFN-γ receptor II (IFN-γRII)–deficient mice with DC-OVA, with or without CpG treatment, and boosted the mice with LM-OVA at day 6. DC immunization stimulated a similar expansion of antigen-specific T cells in wild-type and IFN-γRII–deficient mice and these cells underwent vigorous secondary expansion in response to day 6 booster immunization (Fig. 6g). CpG treatment of wild-type mice prevented substantial secondary expansion in response to day 6 booster immunization. In contrast, CpG treatment did not prevent the booster response in DC-vaccinated IFN-γRII–deficient mice. These data provide strong support for the idea that early inflammation, specifically involving IFN-γ, acts through the IFN-γ receptor on T cells to prevent early generation of memory characteristics. From these data, we propose that inflammatory signals received by the T cells during or shortly after priming determine the rate at which T cells acquire memory characteristics.

Discussion

A single acute infection can generate substantial T-cell memory and lifelong protection from reinfection1,16,28, whereas multiple immunizations and a substantial amount of time may be required to achieve the same protective immunity after vaccination. Here, we report that DC immunization stimulates a pathway of accelerated CD8+ T-cell memory that substantially shortens the time required for booster amplification of even weak primary CD8+ T-cell responses.

What features of DC immunization permit the accelerated generation of memory CD8+ T cells compared to infection? After infection, DCs acquire antigens from pathogens and receive inflammatory signals (Toll-like receptor agonists, cytokines, etc.) that upregulate costimulatory molecules, induce cytokine production and direct DCs to encounter naive T cells in secondary lymphoid organs35,36. In this circumstance, inflammation is crucially important for maturation of the DCs. But T cells and DCs share many cytokine receptors, and thus, naive T cells will also receive inflammatory signals as they interact with mature DCs. For example, both DCs and T cells express the receptors for IL-12 and type I and type II IFN, and all of these cytokines are rapidly produced in response to L. monocytogenes infection16 (data not shown). Consistent with this notion, recent experiments from our laboratory show that CD8+ T cells receive an IFN-γ signal within 12 h of L. monocytogenes infection42 and that IFN-γ signals can regulate the contraction phase of the CD8+ T-cell response31,43. After infection or vaccination with adjuvants, when both the DCs and T cells receive inflammatory signals, the acquisition of memory characteristics by antigen-specific CD8+ T cells is relatively slow.

In contrast, DC immunization creates a situation in which the naive T cells receive T-cell receptor and costimulatory signals in the absence of infection-induced inflammatory signals. In this scenario, we propose that the lack of inflammatory signals to the T cells during priming substantially accelerates acquisition of memory characteristics, permitting a shorter interval between priming and booster immunizations. Consistent with this notion, CpG treatment to induce inflammation at the time of or shortly after DC immunization prevented the accelerated generation of memory cells and the rapid prime-boost response, despite the fact that the magnitude of the CD8+ T-cell response was unaltered and the CpG treatment did not alter the lifespan of the injected DCs or the duration of DC-mediated antigen presentation. In addition, the failure of CpG-treatment to inhibit early memory formation after wild-type DC immunization of IFN-γRII–deficient mice, an experimental design in which the T cells but not DCs lack the IFN-γRII, strongly suggests that specific inflammatory mediators acting directly on the T cells determine the rate of memory CD8+ T-cell generation.

DCs are under evaluation as vaccine vehicles to evoke CD8+ T-cell responses in humans. Our results show that DC immunization allows the rapid boosting of the number of antigen-specific CD8+ T cells, a finding that could improve vaccine efficacy. For example, rapid generation of large numbers of effector and memory CD8+ T cells by vaccination may be crucial to control the spread of pathogens introduced by bioterrorism. Similarly, a major constraint on the success of DC-based cancer immunotherapy is the requirement for rapid generation of high numbers of effector and memory CD8+ T cells specific for tumor antigens. These antigens are often tissue-specific self-peptides, where tolerance-induced limitations in the available T-cell repertoire may restrict the magnitude of the primary response. Our results show that DC immunization allows the rapid amplification of effector CD8+ T cells specific for weak, subdominant antigens. Thus, this strategy may address major limitations of DC vaccination for cancer immunotherapy. Finally, the amplification of numbers of effector and memory CD8+ T cells in DC-immunized mice occurs in response to a variety of booster immunizations, including those that could be readily applied to humans. It should be possible to rapidly determine whether similar strategies can be used to enhance vaccine efficacy against tumors and pathogens causing serious human disease44,45.

Methods

Mice, Listeria monocytogenes, Vaccinia virus, CpG and peptide-coated splenocytes.

We obtained BALB/c (Thy1.2+, H-2d major histocompatibility complex) and C57Bl/6 (Thy1.2+, H-2b) mice from the US National Cancer Institute. OT-I transgenic Thy1.1+ mice were previously described33. IFN-γRII–deficient mice were provided by P. Rothman (University of Iowa)46. We housed pathogen-infected mice in the appropriate biosafety conditions. We used all mice at 8–16 weeks of age. Virulent (10403s, XFL303 (L. monocytogenes expressing lymphocytic choriomeningitis virus–derived NP118–126 epitope))32, and DP-L2528 (LM LLO92F, nonfunctional LLO91–99 epitope)38 and attenuated (DP-L1942, which is actA-deficient)23 and actA-deficient LM-OVA34L. monocytogenes strains were grown, injected intravenously and colony-forming units (CFU) per spleen and gram of liver were determined on various days after infection as described47. The LD50 for virulent L. monocytogenes strains is 1 × 104, whereas the LD50 for actA-deficient L. monocytogenes is 1 × 107 for BALB/c mice. Vaccinia-virus expressing LLO was provided by J. Lindsay Whitton (The Scripps Research Institute) and was propagated and injected intraperitoneally as described22. We injected mice intraperitoneally with 60 μg CpG oligodeoxynucleotide 1826 (ref. 31) each at the time of DC immunization (unless otherwise stated). Naive BALB/c splenocytes were coated with LLO91–99 peptide (1 μM) at 37 °C for 1 h, washed three times and injected intravenously at 2 × 107 cells/mouse.

Antibodies and peptides.

We used antibodies with the indicated specificity and with appropriate combination of fluorochromes: IFN-γ (clone XMG1.2, eBioscience), CD8 (53-6.7, Pharmingen), Thy1.2 (53-2.1, Pharmingen), TNF (MP6-XT22, eBioscience), CD127 (A7R34, eBioscience), CD43 (1B11, Pharmingen), CD27 (LG.7F9, eBioscience), CD44 (Pgp-1, Pharmingen), IL-2 (JES6-5H4, Pharmingen), CD62L (MEL-14, eBioscience), CD25 (PC61, eBioscience), and isotype controls IgG2a, IgG2b and IgG1 (clones eBR2a, KLH/G2b-1-2, eBRG1, respectively, eBioscience). Synthetic peptides, which represented defined L. monocytogenes epitopes: LLO91–99, p60217–225, p60449–457, f-MIGWII, lymphocytic choriomeningitis virus-derived NP118–126 epitope, as well as OVA257–264 were previously described26,32,33,48.

Adoptive transfer of OT-I.

We transferred purified naive OT-I Thy1.1 cells (4 × 105/mouse) into naive C57Bl/6 Thy1.2 mice and 1 d later immunized the recipient mice either with actA-deficient LM-OVA (4 × 106) or DCs coated with OVA257–264 peptide (4 × 105 CD11c+ cells). In the experiments in which we addressed duration of antigen display after peptide DC immunization, we transferred purified naive OT-I Thy1.1 cells (5 × 105/mouse) into DC-OVA immunized C57Bl/6 Thy1.2 mice at various days after immunization.

Bone marrow–derived dendritic cells.

We generated bone marrow–derived CD11c+ DCs by 5–7 d of culture in GM-CSF and IL-4 as described15. We then added lipopolysaccharide (100 ng/ml; Sigma) and/or CpG oligodeoxynucleotide 1826 (10 μg/ml; unless otherwise stated) for 1 d to induce maturation, and then added peptide (1 μM) to cultures 2 h before we collected cells and washed them extensively before injection. The resulting cell populations consisted of 50–80% CD11c+ cells. These cells were also positive for H-2Ld, B7.1, B7.2, I-Ad and CD11b and negative for CD8α. Based on percentage of CD11c+ cells (determined before injection), 2.5 × 105 mature peptide-coated DCs were injected intravenously.

Quantification of antigen-specific CD8+ T-cell response.

The magnitude of the epitope-specific CD8+ T-cell response was determined by peptide-stimulated intracellular staining for IFN-γ, IFN-γ and TNF or IFN-γ and IL-2 as described49. We subtracted the percentage of IFN-γ+ CD8+ T cells in unstimulated samples from each mouse from the peptide-stimulated value to determine the percentage of antigen-specific CD8+ T cells. We calculated the total number of epitope-specific CD8+ T cells per spleen from the percentage of IFN-γ+ CD8+ T cells, the percentage of CD8+ T cells in each sample and total number of cells per spleen. The same procedure was used for detection of antigen-specific CD8+ T cells obtained from various organs as previously described50. LLO91–99–specific CD8+ T cells were also detected by phycoerythrin-conjugated tetramer complexes as described50.

Statistical analysis and ethical considerations.

Results are expressed as mean ± s.d.. Differences between groups were examined for statistical significance using Student t-test. Experimental protocols were approved by the University of Iowa Animal Care and Use Committee.

Note: Supplementary information is available on the Nature Medicine website.