Priming of Memory But Not Effector CD8 T Cells by a Killed Bacterial Vaccine

See allHide authors and affiliations

Science  23 Nov 2001:
Vol. 294, Issue 5547, pp. 1735-1739
DOI: 10.1126/science.1064571


Killed or inactivated vaccines targeting intracellular bacterial and protozoal pathogens are notoriously ineffective at generating protective immunity. For example, vaccination with heat-killed Listeria monocytogenes (HKLM) is not protective, although infection with live L. monocytogenes induces long-lived, CD8 T cell–mediated immunity. We demonstrate that HKLM immunization primes memory CD8 T lymphocyte populations that, although substantial in size, are ineffective at providing protection from subsequent L. monocytogenes infection. In contrast to live infection, which elicits large numbers of effector CD8 T cells, HKLM immunization primes T lymphocytes that do not acquire effector functions. Our studies show that it is possible to dissociate T cell–dependent protective immunity from memory T cell expansion, and that generation of effector T cells may be necessary for long-term protective immunity.

CD8 T lymphocytes mediate immunity to a broad range of viral, bacterial, and protozoal pathogens (1), and increasing evidence suggests that effector T cells primed during infection evolve into long-lived memory cells (2–4). Memory T cells can be subdivided into two categories on the basis of activation markers, homing receptor expression, and effector function (5). Central memory T cells, which express high levels of the chemokine receptor CCR7 and the adhesion molecule CD62L and do not express effector functions, may differentiate into effector memory T cells, which express low levels of CCR7 and CD62L but produce cytokines (6). Whether these memory T cell subsets differ in their capacity to mediate protective immunity is unknown. Furthermore, the stimuli that generate central versus effector memory T cells remain undefined.

Listeria monocytogenes is a Gram-positive, facultative intracellular bacterium that causes severe disease in immunocompromised patients (7). Studies in a mouse model of listeriosis have demonstrated that CD8 T cells mediate protective immunity after immunization with live bacteria (8–10). HKLM immunization, on the other hand, does not induce protective immunity (11). Infection of mice with live L. monocytogenes induces antigen-specific CD8 T cell responses, with peak frequencies 7 to 9 days after infection of naı̈ve mice and 5 days after reinfection of immune mice (12). To determine whether HKLM immunization primes L. monocytogenes–specific CD8 T cells, we immunized CB6 (C57BL/6 × BALB/c F1) mice with live L. monocytogenes or HKLM (13) and, 21 days later, intravenously infected these mice with live L. monocytogenes. The magnitude of the L. monocytogenes–specific CD8 T cell response was measured by H2-Kd tetramer staining of splenocytes (Fig. 1A). CD8 T cells specific for the immunodominant listeriolysin epitope LLO91-99 were detectable 7 days after infection of naı̈ve mice (Fig. 1A, upper left panel), whereas responses to the subdominant p60217-225 and p60449-457 epitopes were smaller. As expected, reinfection of mice immunized with live L. monocytogenes induced markedly larger CD8 T cell memory responses (Fig. 1A, middle row), and these mice were resistant to reinfection (Fig. 1B). In contrast, HKLM immunization was not protective (Fig. 1B). Surprisingly, HKLM-immunized mice mounted memory CD8 T cell responses that were indistinguishable in size from those detected in immune mice (Fig. 1A, bottom row). Mice immunized with HKLM derived from an avirulent strain of L. monocytogenes lacking LLO also primed CD8 T cell responses to p60 (14), indicating that residual LLO associated with HKLM is not allowing access to the cytosol of APCs.

Figure 1

HKLM immunization primes antigen-specific memory CD8+ T cell responses. (A) CB6 mice (6 to 8 weeks old) were immunized intravenously with PBS, 104live L. monocytogenes, or two consecutive daily doses of 109 HKLM (13). Mice were infected 21 days later with 104 (naı̈ve and HKLM-immunized mice) or 105 (immune mice) live L. monocytogenes. Splenocytes, 7 days after infection in PBS-treated mice and 5 days after infection in HKLM and live L. monocytogenes–immunized mice, were stained for CD8α and CD62L and with H2-Kd tetramers complexed with threeL. monocytogenes–derived epitopes (LLO91-99, p60217-225, and p60449-457). Dot plots are gated on live CD8 T lymphocytes and show CD62L and H2-Kd tetramer staining. The percentage of activated tetramer-positive CD8 T cells is shown in the upper left quadrant of each panel. (B) CB6 mice were immunized with two intravenous doses of 109 HKLM, 104 live L. monocytogenes, or PBS and were challenged 21 days later with 104 live L. monocytogenes. Protective immunity was measured by dissociating spleens 72 hours after infection and quantifying viable bacteria. Mean numbers of colony-forming units (CFU) from six mice per group are shown (error bars, SD). (C) BALB/c mice were immunized with 2000 live L. monocytogenes and, 4 weeks later, left untreated or depleted of CD8 T cells by intravenous administration of three consecutive daily doses of 100 μg of CD8α-specific mAb (anti Lyt 2, 53-6.72, American Type Culture Collection) before rechallenge with 105 live L. monocytogenes. Mice received an additional dose of CD8 mAb 3 days after infection. CD8 T cell depletion was greater than 90%. Mean numbers of bacteria in spleens (three mice per group) 72 hours after infection are shown (error bars, SD).

Although CD8 T cells can provide protective immunity toL. monocytogenes infection, CD4 T cells and antibodies have also been implicated in immunity (10, 15). To measure their contribution to protective immunity, we depleted CD8 T cells from immune mice and challenged them with L. monocytogenes. Depleted animals were markedly more susceptible than control immune mice to L. monocytogenes infection (Fig. 1C), demonstrating the essential role that CD8 T cells play in protective immunity. Depletion of CD8 T cells with a monoclonal antibody (mAb) specific for CD8β yielded similar results (16).

The robust CD8 T cell memory responses in HKLM-immunized mice could have resulted from direct priming of antigen-specific CD8 T cells or, indirectly, through priming of L. monocytogenes–specific CD4 T cells that would act to accelerate naı̈ve CD8 T cell responses upon challenge with live bacteria. To distinguish between these two mechanisms, we immunized mice with HKLM lacking the immunodominant LLO91-99 CD8 T cell epitope (HKLMLLO–). This strain of L. monocytogenes was generated by point mutation of an essential anchor residue of LLO91-99 and expresses functional LLO (17). If CD4 T cell responses accelerate CD8 T cell responses upon live challenge of HKLM-immunized mice, LLO91-99–specific responses would be expected to still be enhanced in HKLMLLO–-immunized mice. Infection of mice previously immunized with HKLMLLO– resulted in a primary-like CD8 T cell response to LLO91-99, similar to that obtained upon primary infection of naı̈ve mice with L. monocytogenes (Fig. 2A, top row), but a memory-like CD8 T cell response to p60217-225 (Fig. 2A, bottom row). This experiment demonstrates that HKLM immunization primed CD8 T cell responses independently of CD4 T cell priming. Indeed, CD4 T cells from HKLM-immunized and naı̈ve mice, in contrast to mice infected with live bacteria, do not respond to bacterial antigen (Fig. 2B), which suggests that HKLM immunization does not efficiently prime CD4 T cells. Furthermore, infection of HKLM-immunized mice did not result in memory CD4 T cell responses (14).

Figure 2

HKLM directly primes antigen-specific CD8+ T cells, but not L. monocytogenes–specific CD4+ T cells. (A) CB6 mice (three per group) were injected intravenously with PBS or HKLM derived from a mutant L. monocytogenesstrain lacking the LLO91-99 epitope (see text) and were rechallenged 21 days later with live L. monocytogenes. The frequencies of CD8 T cells specific for LLO91-99 and p60217-225 were determined by tetramer staining 5 days after infection. (B) Mice were immunized with live L. monocytogenes, HKLM, or PBS; 6 days later, splenocytes were restimulated in vitro with HKLM (top row) or PBS (bottom row) in the presence of brefeldin A (BFA) for 6 hours. IFN-γ production was measured by intracellular cytokine staining using manufacturer's protocols (Cytofix/Cytoperm, Pharmingen). Dot plots show CD4 and IFN-γ staining; percentages of CD4 T cells producing IFN-γ are shown. (C) BALB/c or CIITA–/– mice (8 to 10 weeks old) were infected intravenously with live L. monocytogenes and analyzed 7 days later for the frequency of LLO91-99–specific T cells by tetramer staining (top row). Mice that had received live L. monocytogenes 21 days earlier were rechallenged with 105 live bacteria, and the LLO91-99–specific CD8 T cell response was measured 5 days later. (D) The number of live bacteria in the spleens of mice 72 hours after primary or secondary infection was determined as described in Fig. 1. Means of two mice per group are shown (error bars, SD). In (A) and (C), the percentage of activated, tetramer-positive CD8 cells is shown in the upper left quadrant of each panel.

One explanation for a lack of protective immunity in HKLM-immunized mice may be that deficient CD4 T cell responses limitL. monocytogenes–specific CD8 T cell function. To determine whether CD4 T cells influence CD8 T cell responses toL. monocytogenes infection, we measured the LLO91-99–specific CD8 T cell response in mice lacking the class II transactivator gene (CIITA–/–). These mice have a 95% reduction in peripheral CD4 T cells and do not mount peripheral CD4 T cell responses because B cells, dendritic cells, and macrophages lack major histocompatibility complex class II molecules (18). The primary CD8 T cell response toL. monocytogenes infection was identical in CIITA-deficient and control mice (Fig. 2C). Similarly, reinfection of previously immunized CIITA–/– and control mice elicited indistinguishable memory LLO91-99–specific CD8 T cell responses. CIITA-deficient mice cleared bacterial infection and developed similar protective immunity (Fig. 2D). Thus, CD4 T cell responses do not account for the difference in protective immunity after live and HKLM immunization.

To further characterize CD8 T cell priming by in vivo HKLM immunization, we generated a transgenic mouse line expressing T cells with specificity for p60217-225 (19). Naı̈ve, antigen-specific T cells from these mice were labeled with carboxyfluorescein diacetate succinimidyl ester (CFSE) and transferred into naı̈ve recipient mice, allowing us to trace their fate after live infection or HKLM immunization. Donor CD8 T cells can be distinguished from recipient cells by a congenic Thy1 disparity. In the absence of infection, transferred p60217-225–specific CD8 T cells remained CFSE positive (Fig. 3A, top row), indicating that they had not undergone in vivo divisions after transfer. In contrast, infection with live L. monocytogenes resulted in a large population of CFSE-low donor CD8 T cells (Fig. 3A, middle row). Immunization with HKLM also induced the expansion of p60217-225–specific T cells, with the majority of transferred T cells entering division (Fig. 3A, bottom row). However, in contrast to priming by live infection, p60217-225–specific T cells primed with HKLM underwent fewer divisions (Fig. 3B). The proportions of naı̈ve p60217-225–specific T cells that proliferated after infection with live bacteria or immunization with HKLM were not significantly different (Fig. 3C), which suggests that transferred T cells encounter similar levels of processed antigen. It is unlikely that decreased T cell proliferation reflects insufficient antigen presentation, because mice immunized with 108, 109, and 1010 HKLM yielded similar results. Furthermore, we and others have demonstrated that naı̈ve CD8 T cells require only brief encounter with antigen to undergo prolonged division (20–22). Thus, it is unlikely that a shorter duration of in vivo antigen presentation after HKLM immunization accounts for decreased proliferation.

Figure 3

HKLM-primed CD8+ T cells undergo limited primary expansion and do not acquire effector functions. (A) TCR-transgenic mice specific for p60217-225were generated with the TCRα and β chain genes from CTL clone L9.6 and backcrossed to the BALB/c (Thy1.2) background (19). L9.6 TCR-transgenic splenocytes (2.2 × 107) were labeled with 5 μM CFSE and transferred into BALB/c Thy1.1 mice. One day later, mice were intravenously injected with PBS, live L. monocytogenes, or HKLM. Splenocytes were recovered 3, 5, and 7 days after immunization and stained with antibodies specific for CD8α and Thy1.2. CFSE staining intensity and staining for Thy1.2, which identifies transferred T cells, are plotted for live CD8 T cells. Percentages of Thy1.2+ cells that are CFSE low, intermediate, and high are shown in each plot. (B) Percentages of cells that have divided from 0 to >7 times (as determined by the intensity of CFSE fluorescence) are plotted for uninfected, live L. monocytogenes–infected, and HKLM-immunized mice on the fifth day after inoculation. (C) Absolute numbers of transferred Thy1.2+ T cells that remained undivided or that proliferated in uninfected, liveL. monocytogenes–infected, and HKLM-immunized mice were determined on the fifth day after inoculation. (D) For determination of the cytolytic activity of transferred L9.6 TCR-transgenic T cells, CD8 T cells were enriched with MACS anti-CD8α beads (Ly-2, Miltenyi Biotech) from mice immunized 7 days previously with PBS, live L. monocytogenes, or HKLM. Direct cytolytic activity of these CD8 T cells was measured in a 6-hour 51Cr release assay using p60217S-225–pulsed P815 target cells. Percent specific lysis is plotted for three different effector-to-target ratios for an uninfected control and two live and HKLM-immunized mice. (E) p60217-225TCR-transgenic spleen cells (Thy1.2; 2.2 × 106) were labeled with CFSE and transferred into Thy1.1 BALB/c recipient mice. One day later, mice were injected intravenously with PBS (upper panel), live L. monocytogenes (middle panel), or HKLM (lower panel). Splenocytes were taken 5 days later and stimulated in vitro for 6 hours with 10 μM p60217S-225 in the presence of BFA. IFN-γ synthesis by the specific T cells was determined by intracellular cytokine staining of Thy1.2+CD8 T cells. The percentage of IFN-γ–producing, CFSEloCD8 T cells is indicated in the upper left quadrant of each plot.

To determine whether L. monocytogenes–specific CD8 T cells activated by HKLM immunization express cytolytic activity, we transferred p60217-225–specific T cells and infected recipients with live L. monocytogenes or immunized with HKLM. Seven days later, CD8 T cells were assayed for cytolytic activity. CD8 T cells from infected recipients were cytolytic, whereas those from HKLM-immunized recipients displayed no lytic activity (Fig. 3D). Because the ratios of antigen-specific T cells to target cells were similar, these results demonstrate that CD8 T cells induced to proliferate by HKLM do not acquire cytolytic activity. Similarly, the proportion of p60217-225–specific T cells producing interferon-γ (IFN-γ) after HKLM immunization is reduced relative to live infection (Fig. 3E). These experiments show that HKLM immunization supports CD8 T cell expansion, but not differentiation into effector CD8 T cells.

Consistent with their disparate effector functions, p60217-225–specific CD8 T cells after live immunization became CD62Llo, whereas HKLM-primed T cells remained CD62Lhi (Fig. 4, B and C). To determine the possible impact of CD4 T cell help and/or inflammation on CD8 T cell priming, we coimmunized mice with HKLM and live bacteria. Concurrent immunization of mice with HKLM and live wild-type bacteria resulted in two distinct populations, one CD62Lhi and the other CD62Llo, suggesting that some T cells are primed by live infection while others follow the HKLM route (Fig. 4D). Concurrent immunization with wild-type HKLM and live bacteria lacking the p60217-225 epitope primed T cells that underwent fewer divisions and remained CD62Lhi (Fig. 4E), demonstrating that CD4 T cell responses and inflammation do not promote differentiation of HKLM-primed CD8 T cells. These results suggest that different antigen-presenting cells, perhaps in different sites, prime naı̈ve CD8 T cells after HKLM and live infection.

Figure 4

Concurrent live infection with L. monocytogenes does not alter the phenotype of HKLM-primed CD8 T cells. L9.6 TCR-transgenic splenocytes (Thy1.1; 106) were labeled with CFSE and transferred into Thy1.2 mice. One day later, mice were either left uninfected (A), infected with live wild-type L. monocytogenes (B), or immunized with HKLM alone (C) or together with live wild-type L. monocytogenes (D). One group of mice was immunized with HKLM and concurrently infected with live L. monocytogenes 218Ser (E), which contains a point mutation in p60, eliminating the p60217-225 epitope. After 4.5 days, splenocytes were stained with mAbs specific for CD8, Thy1.1, and CD62L. Dot plots show CFSE and CD62L staining of gated Thy1.1 CD8 T cells. Each plot shows results for a single animal and is representative of two mice per experimental group. The percentage of transferred CD8 T cells in the CD62Lhi and CD62Llo quadrants is shown.

Innate inflammatory responses to lipopolysaccharide enhance proliferation and survival of antigen-stimulated CD4 T cells (23), an effect that may result from the induction of antiapoptotic cellular factors (24). Although inflammatory responses to live infection and HKLM immunization differ, HKLM consists of a remarkable constellation of adjuvants: lipoteichoic acid, peptidoglycan, flagellin, lipoproteins, and bacterial DNA. Nevertheless, stimulation of CD40 during HKLM immunization augments the magnitude of L. monocytogenes–specific immune responses (25). It is possible that HKLM-derived molecules do not access relevant innate immune receptors, perhaps because killed and live bacteria are cleared by different cells. For example, HKLM uptake may be restricted to macrophages, whereas live infection may provide greater access of antigen to dendritic cells.

Recent studies have demonstrated that memory T cells reside predominantly in peripheral tissues (26, 27) and that central and effector memory cells differ with respect to trafficking (6). Previous studies have also shown that adoptively transferred CD8 T cells only limit bacterial growth if administered during the first day of infection (28). It is possible, therefore, that HKLM-primed memory CD8 T cells traffic and/or acquire effector functions too sluggishly to mediate protective immunity. Our findings suggest that differentiation of CD8 T cells into effec- tor cells during primary immune responses has important consequences for the development of protective immunity. We believe this has important implications for pathogen- and cancer-specific vaccine development.

  • * These authors contributed equally to this work.

  • Present address: Henry M. Jackson Foundation for the Advancement of Military Medicine, 1600 East Gude Drive, Rockville, MD 20850, USA.

  • Present address: LMU München, Institut für Immunologie, Goethestrasse 31, 803360 München, Germany.

  • § To whom correspondence should be addressed. E-mail: pamere{at}


Stay Connected to Science

Navigate This Article