Increased Dendritic Cell Numbers Impair Protective Immunity to Intracellular Bacteria Despite Augmenting Antigen-Specific CD8⁺ T Lymphocyte Responses¹

Protection against microbes is achieved through the coordinated actions of the innate and adaptive immune systems. Upon pathogen encounter, the innate immune system must sense a pathogen and control its early spread and replication. If innate defense mechanisms are not successful or if the numbers of the infecting organism are overwhelming, the adaptive arm of immunity is then required for controlling infection. APCs bridge innate and adaptive immunity by processing and presenting Ag to T lymphocytes, thereby influencing the quality and magnitude of the adaptive immune response and the outcome of the infection (1).

Dendritic cells (DCs)³form a network of motile sentinels that play a pivotal role in initiation and modulation of the adaptive immune response (2). Immature DCs present in peripheral tissues have high phagocytic activity, but are unable to stimulate T cells efficiently. As they capture microbes or microbial Ags, DCs mature in response to microbial pathogen-associated molecular patterns signaling through Toll-like receptors or in response to proinflammatory stimuli produced by other cells responding to pathogen-associated molecular patterns (3, 4, 5, 6). Mature DCs up-regulate MHC and costimulatory molecules and migrate from tissues to regional lymph nodes, where they efficiently present captured Ags to naive T cells. However, DCs represent a small population of cells in vivo (7), and thus may be a limiting factor in the rate of development and magnitude of Ag-specific T cell responses to infection.

T cells play a critical role in cellular immunity against intracellular bacterial pathogens, such as Listeria monocytogenes and Mycobacterium tuberculosis (8). Effector CD4⁺ T cells act as Th cells during Listeria infection by the production of Th1-type cytokines, such as IFN-γ, which activate macrophage microbicidal activity (9). CD8⁺ T cells, compared with CD4⁺ T cells, appear to play an even greater role in protection against Listeria infection. Mice genetically lacking or depleted of CD8⁺ T cells show increased susceptibility to Listeria, and adoptively transferred Listeria-specific CD8⁺ T cells provide greater protection against Listeria infection than CD4⁺ T cells (10, 11). During primary infection, Listeria-specific CD8⁺ T cells peak in numbers at day 7 (12) and lead to sterilizing immunity, which is mediated by the production of IFN-γ (13) and by lysis of Listeria-infected cells (14, 15). Cytotoxic CD8⁺ T cell-mediated lysis of infected cells releases the bacteria into the interstitial space in which activated macrophages can engulf and kill them. Furthermore, protective CTL immunity against Listeria requires priming by DCs, as demonstrated by Jung et al. (16), in which conditional depletion of DCs in vivo completely abrogates the development of CTL during primary infection.

Accordingly, there is great interest in manipulating DCs to augment the development of robust T cell responses to nominal as well as tumor and microbial Ags (17, 18, 19, 20, 21, 22). One approach is the administration of Flt3 ligand (Flt3-L). When given to mice, Flt3-L causes the in vivo proliferation and mobilization of early hemopoietic progenitor cells in the bone marrow that result in a massive expansion of both CD11c⁺CD8α⁺ and CD11c⁺CD8α⁻ DCs (23). In tumor-bearing mice, Flt3-L treatment augments tumor-specific T cell responses, and causes significant tumor regression (24, 25).

Similar to findings in tumor models, there is considerable experimental evidence that DCs are efficient vaccine vehicles for the induction of protective T cell and T cell-dependent immune responses to infectious diseases (26). These results, and results in the tumor models noted above, suggest that DC numbers may be a limiting factor in the initiation of Ag-specific immunity, and that treatment with agents that enhance DC numbers should augment T cell immunity and facilitate the resolution of active infection with intracellular pathogens. The study presented in this work addresses this question. Contrary to this prediction, we demonstrate that increasing DC numbers through the administration of Flt3-L or polyethylene glycol-conjugated GM-CSF (pegGM-CSF) undermined protective immunity to L. monocytogenes and M. tuberculosis despite an augmented Ag-specific T cell response. This suggests that the immune system evolved to maintain DC numbers within limits that are balanced to provide optimal host defense to infection with intracellular pathogens.

C57BL/6 (H-2^b), recombination-activating gene (RAG)-II-deficient (RAG null, 129/svj (H-2^b)), and BALB/c (H-2^d) mice used in the Flt3-L studies were females between the ages of 2 and 6 mo, and were purchased from The Jackson Laboratory (Bar Harbor, ME) or Taconic Farms (RAG null mice; Germantown, NY). Mice were housed under specific pathogen-free conditions and in accordance with the Institutional Animal Care and Use Committee. Mice were age matched in each experiment, and BALB/c mice were used, unless otherwise indicated.

Frozen (−80°C) 1-ml aliquots of a log-phase culture of L. monocytogenes (ATCC 43251), which had been passaged in C57BL/6 mice to promote virulence, were thawed and grown in trypticase soy broth for 3 h at 37°C. Serial dilutions were then made in sterile 0.9% saline or antibiotic-free cell culture medium to achieve the desired concentration (27).

M. tuberculosis (H37Rv strain), obtained from J. Belisle (Colorado State University, Fort Collins, CO, through National Institutes of Health National Institute of Allergy and Infectious Diseases Contract N01 AI-75320), was grown at 37°C to mid-log phase in Proskauer-Beckett medium. The bacteria were passaged once in mice via aerosol to promote virulence. After 4 wk, lung homogenates (in 0.05% Nonidet P-40 in PBS) were spread onto 7H10 agar, and a single colony was inoculated in modified Proskaur-Beckett medium and grown to mid-log phase at 37°C (28). Individual aliquots were frozen at −80°C until use.

For Listeria infections, mice were injected with bacteria (1 × 10⁴ CFU, unless otherwise indicated) in 100 μl of 0.9% saline by the i.p. route, as previously described (27). To determine bacterial burden, livers and spleens were aseptically removed en bloc from mice at the indicated days and weighed. Organs were placed in sterile Nonidet P-40 (0.1% in water) and ground in 15-ml round-bottom polypropylene tubes with a motorized tissue homogenizer (Biospec Products, Bartlesville, OK). Rotors were washed and flame sterilized between samples to avoid cross-contamination. Serial 10-fold dilutions of the organ homogenates were made in 0.1% Nonidet P-40 and spread in duplicates on trypticase-soy agar plates. Plates were then incubated for 36–48 h at 37°C, and the CFU were enumerated. The bacterial burden is expressed as the mean of the log₁₀ CFU/g tissue ± SE.

For M. tuberculosis infections, mice were infected with M. tuberculosis (H37Rv strain) by aerosol, which deposited ∼100 CFU to the lungs, as previously described (27). Lungs, livers, and spleens were removed on the indicated days, and organs were homogenized in 0.05% Nonidet P-40 in PBS, as above. Serial 10-fold dilutions of the homogenates were spread on 7H10 and then incubated at 37°C for 2–3 wk before CFU were enumerated.

Immunex (Seattle, WA), now Amgen, provided human rFlt3-L and recombinant pegGM-CSF. Flt3-L, produced in Chinese hamster ovary cells, was supplied as a sterile, lyophilized preparation with 40 mg of mannitol, 10 mg of sucrose, and 25 mM tromethamine (Tris) per 1.5 mg of FL. PegGM-CSF was supplied in PBS. Endotoxin levels were <10 pg/g protein. Sterile, nonpyrogenic water was used to reconstitute Flt3-L and pegGM-CSF to desired concentrations, and aliquots were frozen at −80°C. Working concentrations of Flt3-L and pegGM-CSF were diluted in 10 μg/ml mouse serum albumin in water and kept at 2–8°C before injection. For in vivo DC expansion, mice were injected i.p. with 10 μg of Flt3-L (9 days) or 2 μg of pegGM-CSF (5 days). Control mice received mouse serum albumin alone. All groups of mice received the same number of injections for each experiment.

Cell culture was performed using RPMI 1640 supplemented with 10% heat-inactivated FCS (Life Technologies, Grand Island, NY), 2 mM l-glutamine, 50 U/ml penicillin, 50 μg/ml streptomycin, 40 μg/ml gentamicin, 0.1 mM nonessential amino acids, 1.0 mM sodium pyruvate, and 50 μM 2-ME (RPMI-C⁺; antibiotic-free culture medium is referred to as RPMI-C). Where indicated, RBC were lysed with RBC lysis buffer (0.15 M NH₄Cl, 1.0 mM NaHCO₃, 0.1 mM EDTA, pH 7.2) for 5 min at room temperature. Cells were then washed twice in RPMI-C⁺ or RPMI-C. PBS plus BSA (PBSA) was used for flow cytometric assays. Where indicated, cells were exposed to the following: LPS derived from Escherichia coli strain 0111:B4 at a final concentration of 100 ng/ml (Sigma-Aldrich, St. Louis, MO); murine rIFN-γ (Genzyme, MA) at a final concentration of 200 U/ml; and brefeldin A (Sigma-Aldrich) for intracellular cytokine staining at a final concentration of 10 μg/ml.

Listeria peptides for the in vitro stimulation of CD8⁺ T cells were purchased from United Biochemical Research (Seattle, WA). The peptides were for the dominant listeriolysin O_91–99 (LLO_91–99) (G-Y-K-D-G-N-E-Y-I) and the subdominant p60_217–225 (K-Y-G-V-S-V-Q-D-I) H-2^d MHC class I-restricted peptide Ags.

Spleens from cell donors were harvested and placed in 10 ml of RPMI-C⁺. The spleens were ground between sterile frosted glass slides, and aspirated through a 20-gauge needle until a single cell suspension was achieved. RBC were lysed with filter-sterilized RBC lysis buffer for 5 min at room temperature and then washed. Spleen cells were enriched for T cells by depleting B cells (B220⁺) and MHC class II⁺ cells using the MACS system (Miltenyi Biotec, Bergisch Gladbach, Germany). An aliquot of this cell preparation was stained with mAbs for CD3, CD4, CD8, NK1.1, B220, and Mac-1 to monitor cells transferred to mice. The T cell-enriched fraction was resuspended in HBSS (Life Technologies) to the desired concentration in 100 μl and then injected in the lateral tail vein (i.v.), after which mice were challenged with Listeria.

Analysis of lymphoid populations was accomplished by staining cell suspensions in PBSA (PBS + 0.5% BSA) with fluorochrome- or biotin-conjugated mAbs specific for surface proteins found on T and B cells, and on macrophages, and immature and mature DCs (BD PharMingen, San Diego, CA), as indicated. Biotin-conjugated Abs were detected with streptavidin-fluorochrome conjugates. Stained suspensions were analyzed on a FACScan or LSR flow cytometer and analyzed using CellQuest software (BD Biosciences, Sunnyvale, CA).

A method modified from Badovinac and Harty (29) was used to assess CD8⁺ T cells that produce IFN-γ in response to LLO_91–99 and p60_217–224 class I-restricted peptides. Briefly, spleens were dissociated and, after RBC lysis, resuspended at 2 × 10⁷/ml in RPMI-C, and 200 μl was pipetted per well in a 96-well U-bottom tissue culture plate. For positive control wells, PMA (0.5 μg/ml) and ionomycin (7.5 μM) were added. For peptide stimulation, 20 μl of synthetic peptide was added to give a 5 μM final concentration. Brefeldin A (10 μg/ml final; Sigma-Aldrich) was added to all wells. After incubation for 6 h (37°C/5% CO₂), cultures were washed twice in cold PBSA. Splenocyte Fc receptors were blocked, and cells were stained with PE-labeled Abs to CD8α at a 1/100 dilution (BD PharMingen) or isotype control Abs. Cells were washed twice in PBSA and fixed on ice with 2% paraformaldehyde for 10 min. Next, cells were washed twice and allowed to permeabilize with PBS plus 0.1% saponin for 15 min on ice. Abs to IFN-γ conjugated to FITC (1/100 dilution) were added for 30 min on ice. After washing twice in PBS plus 0.1% saponin, cells were resuspended in 300 μl of PBSA and saved at 4°C in dark until analyzed by flow cytometry.

CTL activity was determined using a standard ⁵¹Cr release assay. Splenocytes were harvested from individual Listeria-infected control or Flt3-L-treated BALB/c (H-2^d) mice at day 7 postinfection. RBCs were lysed, and cells were resuspended at 1 × 10⁷ per ml in RPMI-C⁺. Target cells (P815 mastocytoma cell line) were prepared by incubating with 100 μCi of ⁵¹Cr and 100 μM peptide for 1 h at 37°C/5% CO₂. Effectors were plated in 100 μl in 96-well U-bottom tissue culture plate to give appropriate E:T ratios. Targets were seeded in the plate at 1 × 10⁴ per well in triplicate. Targets were plated in the absence of effectors for mininum (no effectors) and maximum (0.1% Nonidet P-40) lysis controls. After 6 h at 37°C/5% CO₂, the supernatants were harvested, and the lysis was determined as cpm on a gamma counter. CTL activity was reported as percentage of specific lysis = (sample lysis − minimum lysis)/(maximum lysis − minimum lysis) × 100 ± SE for each E:T ratio.

For evaluation of DC resistance to CTL lysis, the B9 CTL clone specific for LLO_91–99 presented on K^d was used as the effector cell (30). DCs and macrophages pulsed with 100 μM LLO_91–99 peptide and ⁵¹Cr were used as targets. DCs were isolated from two sources: 1) magnetic bead-enriched CD11c⁺ cells from Flt3-L-treated mice, or 2) bone marrow cells cultured for 9 days in the presence of 200 μg/ml Flt3-L (31). Macrophages were isolated from two sources: 1) adherent resident peritoneal exudate cells, or 2) bone marrow cells cultured for 9 days with 30% L cell conditioned medium in RPMI-C⁺. Isolation protocols for DCs and macrophages gave purities of >80%, and each was incubated with B9 effector cells for 6 h before supernatant harvest.

Previously, studies using Flt3-L to expand and mobilize DCs in vivo used either naive mice or mice challenged with tumor cells (23, 32). To verify that Flt3-L treatment expands DC numbers in the context of a Listeria infection, we infected mice and treated a cohort with Flt3-L on days −4 to +4 of infection (standard treatment protocol). We then analyzed the splenic DC populations in mice at various time points during infection. Listeria-infected mice treated with Flt3-L showed increased CD11c⁺CD8α⁺ and CD11c⁺CD8α⁻ DC numbers, compared with control-infected mice, throughout treatment: DC numbers peaked on day 5 of infection, 1 day after the final Flt3-L administration (Fig. 1 A). Thus, Flt3-L increased DC numbers in Listeria-infected mice, as previously shown in uninfected mice (23).

Because Flt3-L treatment augmented the development of Ag-specific T cell-mediated protection in other models, we predicted that Flt3-L-treated mice would exhibit enhanced bacterial clearance during primary Listeria challenge. Surprisingly, Flt3-L-treated BALB/c mice had increased numbers of Listeria in the liver at days 3–7 (Fig. 1,B). A lower inoculum of 10³ CFU Listeria revealed similar results (data not shown). Furthermore, in one experiment in which BALB/c mice were infected with the normally well-tolerated dose of 1 × 10⁵ CFU Listeria, 58% (7 of 12) of Flt3-L-treated mice, but only 8% (1 of 12) of controls died unexpectedly. Flt-L-treated C57BL/6 mice (which are more resistant to Listeria than BALB/c mice) also had increased numbers of Listeria in the liver and spleen at days 5 and 7, but not at day 3 (Fig. 1, C and D) or day 1 (data not shown). All subsequent experiments with Listeria were performed with the BALB/c mice.

In studies from other groups, mice were pretreated with Flt3-L before tumor challenge or infection (33, 34, 35). The unexpected deleterious effect of Flt3-L treatment on the control of primary Listeria infection led us to question the timing of our treatment regimen. Therefore, we compared our standard Flt3-L regimen (days −4 to +4) with a pretreatment regimen (days −9 to −1). Both Flt3-L treatments inhibited the control of Listeria infection (Fig. 1 E).

To determine whether the effects of Flt3-L treatment on resistance to Listeria were transient as are the effects on DC numbers, we performed an additional experiment. In this experiment, mice were treated with Flt3-L in the usual manner (days −4 to +4) or were treated from days −12 to −4 and then allowed to recover for 4 days before challenge, by which time numbers of total cells and DCs in the spleen had nearly returned to baseline (data not shown). Mice allowed to rest after Flt3-L treatment (days −12 to −4) cleared infection as well as control mice (Fig. 1 F), while mice receiving the standard Flt3-L treatment (days −4 to +4) had a significant increase in Listeria CFU in both the liver and spleen.

The deleterious effect of Flt3-L treatment on protection against Listeria infection was surprising. To determine whether the effect was restricted to Flt3-L treatment or was more general, we asked whether treatment with GM-CSF, which selectively expands CD11c⁺CD8α⁻ DCs (23), would have effects similar to Flt3-L. To determine this, we used GM-CSF coupled to polyethylene glycol (pegGM-CSF), which increases its t_1/2. As predicted, pegGM-CSF treatment increased numbers of CD11c⁺CD8α⁻ DCs in mice infected with Listeria (Fig. 2,A). Both pegGM-CSF-treated mice and Flt3-L-treated mice were less able to control primary Listeria infection than control mice (Fig. 2 B). Thus, high numbers of DCs impaired host defense to Listeria infection regardless of the agent used to increase their numbers in vivo.

To determine whether the harmful effect of increasing DCs in vivo was particular to Listeria or a more general characteristic of infections with intracellular bacteria, we treated mice with Flt3-L and pegGM-CSF and assessed their ability to control infection with M. tuberculosis. Protective immunity to M. tuberculosis requires the host to mount a T cell-mediated immune response in which CD4⁺ and CD8⁺ T cells, as well as IFN-γ, are important for bacterial clearance (36). The tempo of M. tuberculosis infection and the development of T cell immunity are much slower than for Listeria. Innate mechanisms slow bacterial replication during the first 2 wk, after which adaptive immunity develops and acts to control infection (36). Therefore, we assessed M. tuberculosis CFUs at 2 and 4 wk postinfection. Increasing DCs with either Flt3-L (given days −2 to +6) or pegGM-CSF (given days +2 to +6) inhibited M. tuberculosis clearance in all organs assessed at 2 wk (Fig. 2 C). At 4 wk, the results were even more dramatic. Surviving Flt3-L-treated and pegGM-CSF-treated mice had >10-fold more CFU in all organs than control mice. Furthermore, 5 of 10 Flt3-L-treated mice and 7 of 10 pegGM-CSF-treated mice died or had to be sacrificed by 4 wk compared with 0 of 10 controls. These results suggest that increasing DCs in vivo provides an environment in which intracellular bacteria thrive, and that such effects are not restricted to Listeria infection.

To determine the basis for the increased susceptibility of Flt3-L-treated mice, we first asked whether the development of Listeria-specific T cell responses was impaired by assessing the frequency and number of Listeria-specific CD8⁺ T cells in Flt3-L-treated and control mice. Similar to its effects in tumor models, Flt3-L treatment enhanced T cell responses at days 5 and 7, as indicated by the increased percentage (Fig. 3,A) and total numbers (Fig. 3,B) of CD8⁺ T cells able to produce the protective cytokine IFN-γ after in vitro stimulation with LLO_91–99 peptide (the dominant H2^d MHC class I-restricted Listeria epitope (37)). The percentage of CD8⁺ T cells producing IFN-γ to the subdominant p60_217–225 epitope (38) was similar to controls (Fig. 3,A). However, the absolute number of p60_217–225-specific CD8⁺ T cells in Flt3-L-treated mice was increased ∼4.3-fold at day 7 (Fig. 3 B) due to the overall increased splenic cellularity in these mice.

In addition to IFN-γ production, a crucial mechanism by which CD8⁺ T cells protect against primary Listeria infection is the lysis of infected cells (14, 15). To examine the cytolytic potential of T cells from Flt3-L-treated mice, target cells were pulsed with LLO_91–99 or p60_217–225 peptides in a standard ⁵¹Cr release assay. Consistent with the increased numbers of IFN-γ-producing Ag-specific CD8⁺ T cells in Flt3-L-treated mice (Fig. 3, A and B), splenocytes from Flt3-L-treated mice consistently lysed Listeria peptide-pulsed targets more efficiently than cells from control mice after restimulation in vitro (Fig. 3 C). In one of these two experiments, increased CTL activity was also observed directly ex vivo, although, as expected, lytic activity was less than that observed with cells that had been restimulated in vitro (data not shown).

We next sought to determine whether the Listeria-specific T cells in Flt3-L-treated mice were able to mediate protection. A classic method for determining the protective potential of T cells is by adoptive transfer (11). We used this assay, in which T cells from Listeria-infected mice are isolated and injected into naive mice that are subsequently rechallenged with a high dose of Listeria. T cell-enriched splenocytes from control and Flt3-L-treated mice at day 7 postinfection were pooled, divided in numbers equal to those obtained from the spleen of a single donor mouse, and injected into naive recipient mice that were then challenged with a high dose (1 × 10⁶ CFU) of Listeria (Fig. 4,A). In another experiment, recipient mice in all groups received an equivalent number of cells (3 × 10⁷) before high dose challenge (Fig. 4 B). In both experiments, mice receiving cells from Listeria-immune, Flt3-L-treated mice were clearly protected compared with those receiving cells from naive mice, and protection appeared to be superior to that achieved with Listeria-immune cells from control mice.

These data indicate that the expansion of DCs by treatment with Flt3-L augmented the Ag-specific CD8⁺ T cell responses in mice infected with Listeria, and that the increased susceptibility of Flt3-L-treated mice was not due to impaired T cell responses.

Despite the presence of increased numbers of Ag-specific CD8⁺ T cells that produced IFN-γ, were cytolytic, and could transfer protection, Flt3-L-treated mice were unable to control primary Listeria infection as well as control mice. We reasoned that Flt3-L treatment might be impeding an innate protective mechanism independent of the development or presence of T cell immunity.

First, we tested whether increasing DCs by Flt3-L treatment would undermine established T cell immunity in Listeria-immune mice. To assess this, mice were immunized with Listeria and allowed to resolve the infection at least 3–4 wk. Immunized mice were then treated with Flt3-L on days −6 to +2 and challenged with a high dose (1 × 10⁶ CFU) of Listeria on day 0. Flt3-L treatment abolished the early protection at day 3 of secondary Listeria infection, but Flt3-L-treated mice ultimately reduced the numbers of bacteria by day 5 to values similar to controls (Fig. 5 A).

Next, we tested whether increasing DCs with Flt3-L could inhibit innate immune control of Listeria. RAG-deficient mice lack T and B lymphocytes, but have increased NK cell activity. These mice controlled early growth of Listeria better than controls at day 3, as previously reported (39), but enhanced early control of Listeria was abrogated in Flt3-L-treated RAG-deficient mice (Fig. 5 B).

Together, the results in immunized and RAG-deficient mice indicate that Flt3-L treatment undermined both established T cell-mediated and innate control of Listeria infection. We next sought to determine whether DCs provided a favorable niche for Listeria in vivo.

DCs are known to ingest Listeria (40) and other intracellular bacteria in vivo, and bacteria primarily associate with DCs in vivo after infection with Salmonella or bacillus Calmette-Guerin (41, 42). Given that DCs are the cell type most increased in Flt3-L-treated mice during primary and secondary Listeria challenge (data not shown), we hypothesized that DCs were providing a reservoir in which Listeria could replicate and evade T cell-mediated immunity. To determine whether DCs preferentially harbor Listeria, CD11c⁺ DCs were purified from pooled spleens of control and Flt3-L-treated mice before (day 3) and at the peak of organ CFU (day 5) in primary. As indicated in Fig. 6,A, DCs from Flt3-L-treated mice contained more Listeria than the same number (1 × 10⁷) of DCs from control mice. To determine whether DCs, compared with other cell types, were preferentially infected with Listeria, we purified DCs, macrophages, and, for comparison, B cells at day 5 postinfection. Both in Flt3-L-treated and control mice, DCs contained more Listeria per cell than other cell types, and DCs from Flt3-L-treated mice contained more Listeria than DCs from controls (Fig. 6 B). Thus, DCs were preferentially infected with and harbored viable Listeria in vivo.

Listeria are killed in vivo inside activated macrophages, but can grow and replicate in resting macrophages and certain other cell types (43, 44). CD8⁺ T cells contribute to Listeria clearance, at least in part, by lysing infected cells that are unable to kill this organism, thereby releasing Listeria into the extracellular environment, where activated microbicidal phagocytes can engulf and efficiently kill the bacteria. We therefore hypothesized that Listeria persist and replicate in DCs in vivo in Flt3-L-treated mice due to decreased microbicidal action by DCs or by resistance of Listeria-infected DCs to CD8⁺ T cell lysis.

To address the first possibility, resting and activated (IFN-γ plus LPS) macrophages and Flt3-derived DCs were infected (multiplicity of infection = 10), and their ability to kill Listeria in vitro was compared. Both DCs and macrophages ingested bacteria, as indicated by the numbers of Listeria present after 1 h of phagocytosis (t = 0) (Fig. 7,A). After an additional 6 h of incubation, activated macrophages killed ∼83% of the Listeria, while unstimulated macrophages did not. By contrast, Listeria replicated in unstimulated DCs and, notably, DCs were unable to kill Listeria after stimulation with IFN-γ and LPS (Fig. 7 A).

To determine whether DCs were more resistant to CTL lysis compared with macrophages, the B9 CD8⁺ CTL line, specific for the dominant LLO_91–99 peptide (37), was used as the effector cell against peptide-pulsed DC targets labeled with ⁵¹Cr. When purified splenic CD11c⁺ DCs from Flt3-L-treated mice were used as targets compared with resident peritoneal macrophages, DCs were more resistant to CTL lysis than macrophages (Fig. 7,B). Similarly, splenocytes from Flt3-L-treated mice were more resistant to lysis than were splenocytes from control mice (Fig. 7,C), and DCs prepared by culturing bone marrow with Flt3-L were more resistant than bone marrow-derived macrophages (Fig. 7 D).

As predicted, we found that Flt3-L treatment dramatically increased DC numbers and augmented the Ag-specific T cell response during primary Listeria infection. Listeria-specific CD8⁺ T cells were present in greater numbers in Flt3-L-treated mice compared with controls, and produced IFN-γ, were cytolytic, and could transfer protection. Despite the enhanced CD8⁺ T cell responses, Flt3-L-treated mice did not control bacterial replication as well as control mice during primary or secondary infection with Listeria. The increased Listeria replication in Flt3-L-treated RAG null mice suggested that Flt3-L treatment impaired innate immune mechanisms. Consistent with this, Listeria were preferentially associated with DCs in vivo, and DCs were unable to kill Listeria and were relatively resistant to lysis by Listeria-specific CTL in vitro. This suggests that the higher numbers of DCs in Flt3-L-treated mice created a favorable environment for Listeria in vivo that could not be overcome by an augmented Listeria-specific CD8⁺ T cell response. We cannot exclude the possibility that treatment with Flt3-L impaired protection in part through other mechanisms. However, treatment with pegGM-CSF also impaired protection from Listeria. The common feature of Flt3-L and pegGM-CSF treatment is that they increase DC numbers, supporting the notion that this was the principal mechanism by which they increased susceptibility to Listeria. Similar to the findings with Listeria, both Flt3-L and pegGM-CSF inhibited control of M. tuberculosis infection. Together, these data show that the impaired immunity is not particular to one therapeutic cytokine, treatment regimen, or pathogen, but is a more general phenomenon, in which intracellular bacteria can evade an otherwise protective Ag-specific immune response when DCs are present in exaggerated numbers in vivo.

The finding that Flt3-L treatment enhanced the CD8⁺ T cell response to Listeria infection is consistent with other reports using DCs to augment T cell immunity. DCs, pulsed in vitro with nominal, microbial, or tumor Ags, are potent inducers of Ag-specific CD4⁺ and CD8⁺ T cell responses in vivo (21, 45, 46). Furthermore, DCs infected in vitro with M. tuberculosis and used to immunize mice can induce a high level of protection to challenge with virulent M. tuberculosis (47). Increasing DCs in vivo with Flt3-L treatment enhanced CD4⁺ and CD8⁺ T cell responses to normally nonimmunogenic tumor Ags. In fact, Flt3-L treatment of tumor-bearing mice caused complete tumor regression that was both CD4⁺ and CD8⁺ T cell dependent (24, 25, 32, 35).

The unexpected observation in our study is that despite augmented T cell responses, the presence of DCs in numbers greater than normal undermined protective immunity to Listeria. The deleterious outcome may have to do with key differences between intracellular bacterial infection and tumor challenge, as well as differences between immunizing with Ag-pulsed DCs and increasing DCs in vivo. Tumor immunity, similar to immunity for intracellular pathogens, is cell mediated (32). However, tumor cells do not occupy an intracellular compartment, and therefore can be directly recognized and lysed by tumor-specific CD8⁺ T cells induced by DCs. Intracellular Listeria have a reduced exposure to the immune system, but are released after infected cells are lysed by CD8⁺ T cells (14). Under selective pressure of an immune response, and in the presence of DCs that do not kill bacteria and that resist CD8⁺ T cell lysis, Listeria may preferentially replicate in this more favorable cellular niche (48).

Other differences between our studies and those that used DCs as an immunizing vehicle are the numbers of DCs and the period during which DC numbers are increased relative to the time of microbial challenge. In our studies, DC numbers increased during an ongoing infection were available as reservoirs for Listeria. In the previous studies, DC numbers were either relatively low, or DC numbers would have returned to baseline before infectious challenge (21, 47). However, three groups reported that mice treated with Flt3-L before primary infection (days −9 to −1) had better outcomes when challenged with Listeria, Leishmania, or HSV (33, 34, 49). For these studies, a high number of DCs would have been present during the first few days of infection. Still, when we examined Listeria clearance and Listeria-specific CD8⁺ T cell responses with this same regimen and our standard Flt3-L regimen in parallel, both treatment regimens increased bacterial burden at days 5 and 7 postinfection, despite increasing the numbers of Listeria-specific CD8⁺ T cells (data not shown). By contrast, resistance to Listeria in mice treated from days −12 to −4 did not differ from controls, indicating that the deleterious effect of Flt3-L treatment resolved in parallel with the decline in numbers of DCs after treatment was stopped.

In the study by Gregory et al. (33), similar numbers of Listeria were found 6 h after primary infection in Flt3-L-treated and control C57BL/6 mice, but later time points were not assessed. We found that Flt3-L-treated C57BL/6 mice had similar or slightly decreased numbers of Listeria early (days 1 and 3; data not shown), but much higher Listeria numbers at later time points than control C57BL/6 mice (Fig. 1 C). Thus, the deleterious effects of Flt3-L treatment are not apparent early in C57BL/6 mice (unlike BALB/c mice) and would not have been apparent in the study by Gregory et al. Kremer et al. (34) reported that control of cutaneous Leishmania major infection was enhanced in Flt3-L-treated BALB/c mice. Protective immunity against Leishmania infection is dependent on NO and Th1 T cells, which activate macrophages. Unlike Listeria, Leishmania remains within the phagosome after infection, and no direct role for CD8⁺ T cells has been described (50). DCs in our system did produce NO levels as high as macrophages under similar stimuli (data not shown), but did not kill Listeria. However, the role of NO in killing of Listeria remains unclear (43). Therefore, DCs may be able to control Leishmania using NO, and lysis of infected DCs by CD8⁺ T cells may not be required as with Listeria. Similarly, the beneficial effects of Flt3-L treatment in mice infected with HSV (49) most likely reflect the fact that it infects and replicates in epithelial and neural cells rather than in macrophages and DCs. Consistent with this, Vollstedt et al. (51) found that treatment of neonatal C57BL/6 mice with Flt3-L enhanced innate and adaptive immunity to HSV and enhanced innate resistance of neonates to Listeria during the first 5 days of infection, but provided no long-term protection to Listeria, because the mice all subsequently died.

Our finding that Listeria preferentially resides in DCs in vivo is not unexpected because immature DCs reside precisely where they can encounter pathogens from the external environment. During oral Salmonella infection, DCs acquire the bacteria from M cells, or may directly sample the gut lumen for microbes (52, 53). Salmonella preferentially infect and activate DCs, but DCs are unable to kill Salmonella, which persist in vivo for several days (54). Similarly, Mycobacterium bovis bacillus Calmette-Guerin resides in splenic DCs for 2 wk following infection, suggesting that DCs do not kill this bacterium in vivo (41), and Bodnar et al. (55) demonstrated that murine bone marrow-derived DCs do not kill, but are bacteriostatic for M. tuberculosis in vitro, whereas bone marrow macrophages kill M. tuberculosis efficiently. Human DCs are also not microbicidal for M. tuberculosis (56). Thus, the inability of DCs to kill intracellular bacteria in vitro is not likely to reflect infection-induced impairment of DC function, but rather to be a normal feature of this cell type. Listeria-infected DCs did function as potent APC, because CD8⁺ T cell responses were enhanced in Flt3-L-treated mice. In fact, DCs are induced to mature and express high levels of MHC class I and II molecules, costimulatory molecules, and protective cytokines in response to Listeria, M. tuberculosis, or Salmonella spp. (57, 58, 59).

Why are DCs less efficient killers of intracellular bacteria and how might this be beneficial to the host under normal circumstances? Several dominant intracellular bacterial Ags are only made when the bacteria are inside host cells, e.g., LLO from Listeria and α-crystallin from M. tuberculosis (60, 61). The low microbicidal ability of DCs may be an intrinsic mechanism to allow the production, expression, and presentation of the full panel of pathogenic intracellular Ags, which would enable DCs to better access and present these Ags to T cells.

When DCs are present in normal numbers, the system of reduced bacterial killing benefits the host. CD8⁺ T cells need only a short encounter with Ag and APC to divide, and to differentiate into effector cells that are lytic and produce IFN-γ (62, 63). Thus, a few infected DCs are capable of activating hundreds of T cells and would not provide a meaningful niche for an intracellular bacterium to occupy. Wong and Pamer (64) have recently demonstrated that CD8⁺ T cells are primed during the first 72 h of Listeria infection and regulate their expansion by lysing the priming DC in vivo. When DC numbers are artificially increased, their greater presence may recruit and serially stimulate more naive CD8⁺ T cells to give a larger magnitude response (65), as we observed in Flt3-L-treated mice. Bousso and Robey (66) have addressed this by tracking the dynamics of CD8⁺ T cell priming by DCs in intact lymphoid tissue. They have demonstrated that CD8⁺ T cells and DCs have lengthy and stable interactions that, due to limiting DC numbers in vivo, may sterically inhibit further engagement of Ag-specific CD8⁺ T cells. Expanding DC numbers in vivo may overcome this limitation. However, treatment-induced increases in DC numbers may create an “Achilles’ heel” for the host by providing a reservoir in which intracellular bacteria survive in vivo. The relative resistance of DCs to CTL lysis, also observed by others (67, 68), may contribute to Listeria persistence by preventing the release of intracellular bacteria from infected cells, which normally enables more microbicidal phagocytes to acquire and kill the bacteria (14, 48).

The data presented in this work suggest that the numbers of DCs in vivo are delicately balanced for good reason. When DC numbers are increased through Flt3-L or pegGM-CSF in naive mice, strong homeostatic pressure exists in vivo such that DC numbers rapidly return to normal over a period of 2–4 days when the cytokine regimen is discontinued (69) (our unpublished observations). Increased numbers of DCs do enhance the development of Ag-specific T cell responses. Nonetheless, because DCs are poorly microbicidal and relatively resistant to CTL lysis, properties that may be important for their primary role as APC, maintaining DC numbers within tightly controlled limits may be essential for overall host defense to intracellular bacterial pathogens.

We thank Immunex (Amgen) for Flt3-L and pegGM-CSF; Heidi Harowicz for animal husbandry; Sherilyn Smith for assistance with studies of M. tuberculosis; and David Fitzpatrick, Charlie Maliszewski, Tobias Kollmann, Sing Sing Way, and members of the Wilson laboratory for advice and suggestions.