The factors controlling the progression of an immune response to generation of protective memory are poorly understood. We compared the in situ and ex vivo characteristics of CD8 T cells responding to different forms of the same immunogen. Immunization with live Listeria monocytogenes, irradiated L. monocytogenes (IRL), or heat-killed L. monocytogenes (HKL) induced rapid activation of CD8 T cells. However, only IRL and live L. monocytogenes inoculation induced sustained proliferation and supported memory development. Gene and protein expression analysis revealed that the three forms of immunization led to three distinct transcriptional and translational programs. Prior to cell division, CD8 T cell–dendritic cell clusters formed in the spleen after live L. monocytogenes and IRL but not after HKL immunization. Furthermore, HKL immunization induced rapid remodeling of splenic architecture, including loss of marginal zone macrophages, which resulted in impaired bacterial clearance. These results identify initial characteristics of a protective T cell response that have implications for the development of more effective vaccination strategies.
For an intracellular infectious agent, such as Listeria monocytogenes, the CD8 T cell arm of the adaptive immune response is important for providing initial as well as long-term protection (1, 2). The ability of L. monocytogenes to replicate in the cytoplasmic compartment of a cell and the ease with which L. monocytogenes can be genetically modified to express proteins make L. monocytogenes an excellent vaccine candidate (3, 4). Immunization with a live L. monocytogenes-based vaccine results in an effective CD8 T cell response and generation of sterilizing immunity (5–7). In contrast, the CD8 T cell response to heat-killed L. monocytogenes (HKL) vaccination does not lead to protective immunity (8–10). A recent study demonstrated that in contrast to HKL vaccination immunization with irradiated L. monocytogenes (IRL) protected mice from secondary infection (10). However, the protection afforded by IRL immunization was inferior to that achieved after live bacterial infection. The immunological basis for the generation of distinct responses to these forms of the same bacterium has not been fully explored. Moreover, the rules for driving a productive versus a nonproductive immune response in vivo are poorly understood. The protective immunity provided by CD8 T cells to the host depends on the formation of a long-lived memory cell population (11, 12). The differentiation of naive CD8 T cells into effector and memory cells is mediated by a complex series of events, including activation of the innate immune system, effective Ag presentation, CD4 T cell help, secretion of cytokines and growth factors, and modulation of homing receptors.
Early events in T cell activation ultimately determine whether memory is efficiently generated (13–15). The duration of Ag presentation to CD8 T cells also plays a pivotal role in determining the fate of a naive T cell (7, 16–18). Physical and temporal factors are also involved in immune response induction (17, 19–22). Despite these recent advances, little is known about the nature of the developmental program that is initiated in CD8 T cells very early after priming. Therefore, we investigated the parameters that govern the initiation and modulation of primary CD8 T cell immune responses to generate protective memory following immunization with three different types of L. monocytogenes-based recombinant vaccines, focusing on the spleen as the primary site of T cell activation following i.v. inoculation with L. monocytogenes (7, 23). To understand the initial events that determine the eventual outcome of an immunization strategy, we analyzed dendritic cell (DC) activation, Ag presentation, and T cell activation early postimmunization with HKL, IRL, or live L. monocytogenes.
Materials and Methods
C57BL/6J mice were purchased from The Jackson Laboratory (Bar Harbor, ME). The OT-I mouse line was generously provided by Dr. W.R. Heath (Walter and Eliza Hall Institute, Parkville, Australia) and Dr. F. Carbone (Monash Medical School, Prahan, Australia) and was maintained as a C57BL/6-CD45.1-RAG−/− line.
The mCherry gene (24) was shifted from the pRSET-B-mCherry plasmid (generous gift of Dr. Roger Tsien, University of California at San Diego, La Jolla, CA) using the restriction sites BamHI and EcoRI and cloned downstream of the CD11c promoter (25). The linearized construct was injected into C57BL/6 fertilized embryos, and the presence of the transgene was detected by PCR, and mCherry expression by confocal microscopy.
Recombinant L. monocytogenes-producing OVA (rLM-OVA) was produced as previously described (26, 27). The actA− LM-OVA was a generous gift of Dr. John T. Harty (University of Iowa, Iowa City, IA). Mice were immunized i.v. with 1 × 103 CFU rLM-OVA, 1 × 105 or 1 × 107 CFU actA−L. monocytogenes, 1 × 109 CFU HKL, or 1 × 109 CFU IRL. HKL was produced by incubating bacteria from log phase cultures at 70°C for 3 h. Efficiency of killing was assessed by plating undiluted preparations on brain–heart infusion agar for 48 h at 37°C. IRL was generated as previously described (10).
Isolation of lymphocytes
2, 1 mM CaCl2, and 5% FCS at 37°C for 30 min, followed by a 15 min treatment with 1 mM EDTA. Released cells were pooled and then mashed through a cell strainer.
CFSE labeling of cells and adoptive transfer
C57BL/6-CD45.1 OT-I-RAG−/− CD8 cells were resuspended in HBSS at a concentration of 10 × 1066 cells) were resuspended in PBS and adoptively transferred into C57BL/6J-CD45.2 mice by i.v. injection.
At the indicated times postinfection, lymphocytes were isolated and OT-I cells were detected using a CD45.1-specific mAb. For staining, lymphocytes were suspended in PBS/0.2% BSA/0.1% NaN3 at a concentration of 1 × 106 to 1 × 107
Intracellular detection of cytokines
Lymphocytes were isolated from the indicated tissues and cultured for 5 h with 1 μg/ml BD GolgiStop (BD Biosciences), with or without 1 μg/ml of the OVA-derived peptide SIINFEKL. After culture, cells were stained for surface molecules and then fixed, and cell membranes were permeabilized in BD Cytofix/Cytoperm solution (BD Biosciences) and stained with anti–IFN-γ, anti–IL-2, or anti–TNF-α mAbs or the appropriate corresponding isotype control rat IgG (BD Biosciences). Cells were then washed, and the fluorescence intensity was measured on a LSR II.
Whole mount confocal laser microscopy
Mice were sacrificed at the indicated times postimmunization, and the spleen was excised and processed for staining (22
Deep tissue multiphoton microscopy
A total of 5 × 105 CD45.1+ OT-I-RAG−/− cells were CFSE-labeled and transferred to normal CD11c-mCherry mice. One day later, mice were immunized with 109 HKL, 109 IRL, or 105 actA− LM-OVA. A total of 16 h or 24 h later, spleens from mice were excised and thick (1–2 mm) fresh splenic tissue was mounted on chamber slides and imaged by a Prarie Ultima IV microscope (Prairie Technologies, Middleton, WI) fitted with a Spectra Physics Mai Tai DeepSee two-photon laser (Newport Corporation, Santa Clara, CA). To excite CFSE and mCherry, the laser was tuned to 1010 nm. To detect CFSE and second harmonic signal, a 525/550 nm bandpass filter was used, and a 595/550 nm bandpass filter was used to detect mCherry. All of the images were acquired using a water immersion 0.95 numerical aperture ×20 objective. Image analysis was performed using Imaris software (Bitplane).
Microarray gene expression analysis
A total of 5 × 105 CD45.1 OT-I-RAG−/− CD8 cells were adoptively transferred into C57BL/6J-CD45.2 mice by i.v. injection. Twenty-four hours later, mice were left alone or were immunized with 1 × 107 CFU actA– L. monocytogenes, 1 × 109 CFU HKL, or 1 × 109 CFU IRL. Twenty-four hours postimmunization, spleens from five mice in each immunized group were pooled, the cells were stained for expression of CD8, CD45.1, CD69, and MHC class II, and CD8+CD45.1+CD69+MHC class II− cells were sorted on a FACSVantage SE (BD Biosciences). The purity of the sorted population was >94%. Total RNA was extracted from the sorted cells using the Qiagen RNeasy Mini Kit and the QIAshredder columns as per manufacturer’s instructions (Qiagen, Valencia, CA). RNA samples were stored at −80°C in RNase-free water. RNA purity and integrity were analyzed on an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). Total RNA labeling, based on the principles of the Eberwine protocol, was carried out using the Illumina TotalPrep RNA amplification kit (catalog no. IL1791; Ambion, Austin, TX) according to the manufacturer’s instructions (28). Briefly, 300 ng total RNA was reverse-transcribed to cDNA using a T7 oligo(dT) primer. Second-strand cDNA was synthesized, transcribed in vitro, and labeled with biOT-In-16-UTP. Next, 750 ng the biOT-Inylated cRNA was mixed with 10 μl of the Illumina hybridization mix and hybridized to MouseRef-8 Sentrix BeadChip Array (Illumina, San Diego, CA) at 58°C for 16 h. MouseRef-8 targets 25,435 genes with 50-mer probes. Each probe is represented with an average 30-fold redundancy. Hence, with the Illumina bead technology, a single hybridization of RNA from one sample to an array produces ~30 intensity values for each probe. The BeadChip was subsequently washed, stained with streptavidin–Cy3, and scanned on the Illumina BeadArray Reader. The image data files were analyzed using Illumina BeadStudio software package. The Illumina expression data were normalized using the quantile normalization method (29). There were two biological replicates for each CD8 T cell treatment (naive, HKL, IRL, or live L. monocytogenes). Correlation coefficient values (r2) for each biological replicate within each group were: control, 0.9763; HKL, 0.9827; IRL, 0.9843; and live L. monocytogenes, 0.9885. Data were analyzed using the Illumina BeadStudio software or the MultiExperiment Viewer (The Institute for Genomic Research) (30). The data are available through the National Center for Biotechnology Information Gene Expression Omnibus, accession no. GSE20265 (www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE20265).
Mice were immunized with 104 LM-OVA and 109 HKL or 104 LM-OVA alone. Four days postimmunization, bacterial titers within splenic and liver tissues were determined by homogenizing the tissue in PBS containing 1% saponin and plating serial dilutions of the homogenate on brain–heart infusion agar containing 5 μg/ml erythromycin and incubating for 2 d at 37°C.
Division kinetics of OT-I cells after HKL, L. monocytogenes, or IRL immunization
To test the efficacy of activation of the different immunogens, CFSE-labeled CD45.1 TCR transgenic OT-I cells were transferred into naive CD45.2 mice followed by immunization with HKL, actA− LM-OVA (from here on referred to as live L. monocytogenes), or IRL. We used the attenuated form of live L. monocytogenes because this strain is ~1000-fold less virulent than the wild-type strain and therefore allows us to immunize the mice with a high dose (107 CFU) of live bacteria, which may be more comparable to the high initial dose of HKL and IRL immunization. At 24 h postimmunization, no division had occurred regardless of treatment. However, by 48 h postimmunization, most OT-I cells in each group had divided several times, although OT-I cells isolated from HKL-immunized mice clearly lagged behind cells isolated from IRL- or live L. monocytogenes-immunized mice (Fig. 1). Thus, all three types of immunization resulted in sufficient signaling to induce division of Ag-specific CD8 T cells.
CD8 T cell expansion and memory CD8 T cell generation following inactivated versus live L. monocytogenes immunization
In unimmunized mice, OT-I cells made up 0.1–0.2% of splenic CD8 T cells (data not shown). Fig. 2 shows that by 2 d postimmunization OT-I cells had expanded ~10-fold to similar numbers in the spleens of HKL- and IRL-inoculated mice, whereas the response in L. monocytogenes-infected mice was ~1.5-fold higher. However, by day 7 postimmunization, OT-I cells accounted for only 0.2% of CD8 T cells in the spleens of HKL-immunized mice, whereas ~10- to 15-fold more OT-I cells were present in IRL-immunized mice. L. monocytogenes-infected mice mounted a robust response with OT-I cells comprising >60% of splenic CD8 T cells at day 7 postimmunization. By day 32, OT-I cells were barely detectable in HKL-inoculated mice, whereas IRL immunization induced a detectable memory population and live L. monocytogenes infection generated substantial memory. Thus, of the two nonviable bacterial immunizations, IRL was effective in generating memory CD8 T cells. These data explicated recent results showing that IRL but not HKL immunization results in protection from subsequent live bacterial challenge (10).
Because the naive precursor frequency of transferred TCR transgenic CD8 T cells can alter the differentiation patterns of effector T cells, we next compared the expansion of OT-I cells after a low-dose (5 × 103) or high-dose (5 × 105) transfer. Reducing the precursor frequency of naive OTI cells failed to rescue CD8 T cell expansion following HKL immunization (Supplemental Fig. 1).
Although we used an attenuated form of live bacteria (actA-deficient), sustained Ag presentation and inflammatory mediators after live L. monocytogenes immunization may contribute to the superior expansion of OT-I cells. To address this issue, we immunized mice that had previously received OT-I cells with live L. monocytogenes. Two groups of mice were treated with antibiotics 24 h later for 4 or 6 d continuously. Mice then were sacrificed at 5 or 7 d postinfection, and the phenotype and proliferative capacity of OT-I in the spleen were determined. Despite the antibiotic treatment, OT-I expansion (Supplemental Fig. 2) and the activation profile (data not shown) did not change when compared with those of untreated mice. Conversely, when we immunized mice with three doses of 109 HKL every other day for 6 d, OT-I expansion was not rescued (data not shown).
T cell sequestration and CD8 T cell–DC cell cluster formation in response to live L. monocytogenes or IRL but not HKL immunization
Sustained physical interactions between a T cell and a cognate APC may be essential for adequate CD8 T cell differentiation into an effector cell. Therefore, we evaluated the quality of CD8 T cell–DC interactions in the spleen after live L. monocytogenes, IRL, or HKL immunization using flow cytometry and laser scanning confocal microscopy. Interestingly, recovery of splenic OT-I cells at 24 h postimmunization required the use of collagenase-assisted tissue disruption only in those mice that received live L. monocytogenes or IRL but not HKL (data not shown). This finding suggested that strong T cell–APC interactions were induced by live L. monocytogenes and IRL but not by HKL administration. The phenomenon of sequestration of responding T cells early postimmunization occurs in other situations as a result of “conditioning” of DCs and T cells (31–34). To visualize the nature of this sequestration at the earliest stages of an immune response, we performed deep tissue imaging using multiphoton confocal microscopy. Using this approach, we were able to scan large areas of the spleen, and moreover, detection of the second harmonic signal was valuable for differentiating between the splenic compartments being imaged. To visualize DC networks in the spleen, we generated a new transgenic mouse that expresses a variant of the red fluorescent protein mCherry under control of the CD11c promoter (CD11c-mCherry). CFSE-labeled OT-I cells were transferred into naive CD11c-mCherry mice and 1 d later were immunized with HKL, IRL, or live L. monocytogenes. A total of 16 h (Fig. 3A) or 24 h (Fig. 3B) later, spleens from mice were excised and thick (1–2 mm) fresh splenic tissue sections were mounted on chamber slides and imaged by two-photon confocal laser microscopy. Three-dimensional reconstructions of merged z stacks clearly show the formation of OT-I clusters with mCherry+ CD11c DCs as early as 16 h after live L. monocytogenes or IRL immunization (white arrows). In the case of live L. monocytogenes-immunized spleens, the clusters became larger by 24 h. We observed that several follicles of a spleen contained two to four clusters of OT-I cells. At 16 h, these clusters contained ~12–30 OT-I cells. At 24 h, some of the larger clusters contained several hundred OT-I cells and could span large areas of the white pulp T cell zones. However, even after carefully scanning large areas of splenic tissues at great depths (>400 μm) using two-photon microscopy, we did not detect any formation of OT-I clusters after HKL immunization (Fig. 3A, 3B). Moreover, extensive imaging analysis of spleens from HKL-immunized mice at 6, 9, 12, 16, 24, 36, and 48 h postimmunization never revealed OT-I cluster formation (Supplemental Fig. 3A). OT-I–DC clusters were also observed at 24 and 36 h after low-dose live wild-type (actA+) L. monocytogenes infection (Supplemental Fig. 3B, 3C).
To further visualize the nature of the OT-I–DC interaction at 24 h postimmunization, we used the Imaris colocalization program to analyze for close contacts between the transferred cells and APCs in situ. Many large OT-I CD8 T cell clusters closely apposed to DC networks were observed following live L. monocytogenes infection (Fig. 3C), whereas IRL immunization resulted in the formation of fewer and smaller OT-I–DC clusters. In contrast, CD8 T cell–DC clusters were not found at this or at any other time after HKL immunization. Multiple close contacts following live L. monocytogenes and IRL inoculation were observed (the predicted areas of T cell- DC contacts are pseudocolored in white) but were not detected after HKL immunization (Fig. 3A, right panels). These data indicated that the sequestration of CD8 T cells in the spleen after live L. monocytogenes and IRL inoculation was the result of cluster formation produced by strong CD8 T cell–DC interactions.
In addition, we analyzed frozen splenic sections at 24 h postimmunization with HKL (Supplemental Fig. 4). OT-I CD8 T cells were evenly spread throughout the splenic T cell zones, similar to OT-I cells in uninfected spleens. In contrast, OT-I cells in IRL-immunized spleens formed multiple small clusters with CD11c+ DCs that were located primarily at the borders of the T cell–B cell zones. The largest number of clusters of OT-I cells with CD11c+ DCs were observed in live L. monocytogenes-immunized spleens. Furthermore, the OT-I cells in live L. monocytogenes-immunized spleens were the largest in size (insets), indicating the most heightened state of T cell activation.
Gene expression analysis of activated CD8 T cells early after vaccination with live versus inactivated L. monocytogenes
The results indicated that early activation and subsequent differentiation and survival of CD8 T cells following HKL, IRL, or live L. monocytogenes were distinct. T cells have been shown to exhibit significant alteration in gene expression as early as 8 h after polyclonal activation (35). Nevertheless, a comparison of the gene expression pattern of CD8 T cells early after activation using distinct vaccination strategies has not been reported. Therefore, to identify the molecular signature of early T cell activation, we compared the gene expression profiles of sorted OT-I CD8 T cells from either unimmunized mice or mice immunized with HKL, IRL, or live L. monocytogenes 24 h earlier. Gene expression patterns in adoptively transferred activated OT-I cells (CD8+CD45.1+CD69+MHC-II−) were compared with those of naive (CD8+CD45.1+CD69−MHC-II−) OT-I cells isolated from uninfected animals using the Illumina bead microarray containing over 24,000 mouse genes. Sample clustering analysis using the Euclidian distance metrics indicated that the pattern of global gene expression between live L. monocytogenes- and HKL- or IRL-activated CD8 T cells was most dissimilar. In contrast, the gene expression pattern between HKL- and IRL-activated CD8 T cells was closely related (Fig. 4A, dendogram). A synopsis of the overall gene expression profile is presented as dot plots in Fig. 4A, where we compared the pattern of gene expression in OT-I CD8 T cells for each immunization group with naive OT-I cells. Even at 24 h postimmunization and prior to division, expression of a large number of genes was modulated (Fig. 4A, dot plots). The number of genes that showed >2-fold differences in expression (either upregulated or downregulated) over naive OT-I cells correlated with the apparent strength of activation with live L. monocytogenes > IRL > HKL (Fig. 4A). Indeed, the correlation coefficient values (r2) indicated that there were more similarities in gene expression between naive OT-I cells and HKL-activated cells (0.850) when compared with IRL- (0.824) or live L. monocytogenes-activated (0.805) cells.
We compared the gene expression profile among all groups of OT-I cells by performing hierarchical clustering using an Euclidian distance metric on genes that were ≥2-fold upregulated (Fig. 4B) or downregulated (Fig. 4C) between naive and live L. monocytogenes-activated OT-I cells. The heat maps of 10 such clusters are presented in Fig. 4B and 4C. The results of the cluster analysis are presented in Tables I and II, which includes a partial list of relevant genes that were most differentially expressed among the three groups of activated T cells. In most instances, the magnitude of upregulation or downregulation of gene expression was the greatest in live L. monocytogenes-activated OT-I cells followed by IRL- and HKL-activated T cells, respectively (Tables I, II). As expected, expression of genes encoding IL-7Rα, CD3, and CD8 was downregulated 24 h after all three types of immunizations. The expression of genes encoding molecules that regulate homing and migration, such as CD62L, CCR7, platelet/endothelial adhesion molecule 1, and Kruppel-like factor (KLF) 2 and KLF3, were all reduced. In fact, KLF2, which regulates thymocyte and T cell migration (36–38), was the most downregulated gene in the entire array. The expression of mRNA encoding several cytokine receptors was significantly reduced in all three groups of T cells at 24 h postimmunization; however, downregulation was the greatest in live L. monocytogenes-activated OT-I cells. IL-2Rα was one of the few cytokine receptors whose mRNA was highly upregulated in all three groups of T cells; however, the expression was considerably higher in live L. monocytogenes-activated OT-I cells. In contrast to most cytokine receptor genes, the expression of several cytokine genes was greatly increased in OT-I cells isolated from all three immunizations, although expression in each case was significantly higher in live L. monocytogenes-activated T cells. The IFN-γ gene was the most highly upregulated. mRNAs encoding IL-2, IL-3, and IL-10 were all upregulated, and expression was significantly higher in live L. monocytogenes-activated OT-I cells as compared with that in IRL- or HKL-activated cells.
As expected, the mitosis and cell cycle regulator genes were also induced. The degree of upregulation in each group of T cells was proportional to the proliferative capacity of the OT-I cells following HKL, IRL, or live L. monocytogenes immunization (Fig. 2). Intriguingly, as early as 24 h postimmunization, many of the genes that code for inhibitory proteins, such as LAG3 and CTLA-4, were already upregulated, and again, the magnitude of upregulation was the highest in live L. monocytogenes-activated OT-I cells (Table II). Genes encoding certain costimulatory molecules, most notably, OX-40 and 4-1BB, were also substantially induced, whereas others, including CD28 and CD27, were not (data not shown). It is noteworthy that genes encoding transcription factors such as repressor of GATA (ROG) and T cell-specific T-box transcription factor (T-bet) were highly induced in all three groups of activated OT-I cells but particularly in live L. monocytogenes-activated CD8 T cells. Correlated with the high expression of mRNA for ROG was the downregulation of GATA-3 mRNA (Tables I, II). Thus, by 24 h postimmunization with the three different vaccines, a pattern of gene expression was established that would dictate the CD8 T cell effector lineage (e.g., high IFN-γ production and low IL-4 production) as well as the efficient development of memory T cells in response to a requisite threshold of stimulation.
Differential programming of naive CD8 T cells early after vaccination with live versus inactivated L. monocytogenes
Our imaging studies showed that although OT-I cells did not form clusters in the spleen after HKL immunization they were capable of entering cell division. It was important to determine how this difference in activation influenced the programming of CD8 T cells postimmunization. Moreover, the gene-array analyses indicated that as early as 24 h postimmunization the genetic programming of Ag-stimulated T cells is well underway, implying that early interactions help to determine the dwell time and its resultant programming. Thus, using the information obtained from the gene-array study, we attempted to distinguish between the proteome imprint of the T cells activated by the transient OT-I–DC interactions seen after HKL immunization, as opposed to the extended interactions observed after IRL, or live L. monocytogenes immunization early after T cell activation before the onset of cell division. As early as 12 h postimmunization, HKL, IRL, and live L. monocytogenes immunization led to a robust activation of transferred OT-I cells as judged by CD69 and CD11a upregulation, indicating that nearly all of the OT-I cells had encountered Ag. By 24 h postimmunization, the OT-I cells had upregulated expression of CD69, program cell death-1 (PD-1), NKG2D, and CD11a. However, PD-1 and CD69 expression were substantially lower on HKL-activated OT-I cells compared with those on L. monocytogenes- or IRL-activated cells (Fig. 5A). At these time points, CD127 (IL-7R) remained at roughly naive T cell levels. Activation of OT-I cells was Ag-specific because immunization with wild-type live L. monocytogenes (lacking the OVA gene) failed to induce activation markers (left panels). Interestingly, at 12 h postimmunization, live L. monocytogenes and IRL immunization induced substantially more CD3 downregulation than did HKL (Fig. 5A), likely reflecting the enhanced CD8 T cell–DC interactions observed in our imaging studies. In the case of low-dose OT-I transfer, the phenotype of OT-I cells after HKL immunization was not substantially altered. However, NKG2D expression on OT-I cells was increased after low numbers of OT-I cells were transferred (Supplemental Fig. 5A, 5B). Nevertheless, reducing precursor frequency of naive OT-I cells failed to rescue CD8 T cell expansion following HKL immunization (Supplemental Fig. 1). These results demonstrated that immediately after Ag encounter, prior to the onset of cell division, the complex programming of CD8 T cells characterized by rapid gene expression and protein synthesis is already in progress and that the strength of signaling was indicated by the expression of a select set of genes and proteins.
Expression of activational and inhibitory receptors may predict response outcome
To analyze the effects of immunization after the onset of division and during the contraction phase, we examined the phenotype of OT-I cells 2 and 7 d postimmunization (Fig. 5B). Two days postimmunization, CD69 expression had decreased on splenic OT-I cells and CD127 was now downregulated in all three groups (Fig. 5B). Of note, the level of PD-1 expression by OT-I cells continued to increase in live L. monocytogenes- and IRL-immunized mice but was only transiently upregulated following HKL immunization (compare Fig. 5A and Fig. 5B). Of interest was the finding that at day 2 postimmunization PD-1 expression was considerably higher on OT-I cells following live L. monocytogenes versus IRL immunization. Thus, PD-1 expression correlated well with the level of activation as measured by expansion and appeared to predict the level of memory development. During the contraction phase of the response (day 7 postimmunization), OT-I cells isolated from live L. monocytogenes- and IRL-immunized mice expressed high levels of the costimulatory molecule NKG2D, whereas most cells responding to HKL did not. Interestingly, the re-expression of CD127 correlated with the initial level of activation because by day 7 CD127 levels had increased on a larger proportion of OT-I cells responding to live L. monocytogenes or IRL than to HKL. In addition, at day 7 postimmunization, CD62L expression by OT-I cells in live L. monocytogenes-immunized mice was notably lower compared with that by OT-I cells in IRL- or HKL-immunized mice. We have consistently observed a biphasic regulation of CD62L expression by OT-I cells after live L. monocytogenes immunization, such that CD62L expression declined at 12–24 h postimmunization, followed by a significant increase in expression at 2 d postimmunization and downregulation by day 7 postimmunization. In the case of HKL and IRL immunization, CD8 T cells initially downregulated CD62L, but it was promptly re-expressed after the onset of cell division (day 2 postimmunization) without any subsequent reduction in expression.
Distinct patterns of cytokine production in response to live L. monocytogenes and inactivated L. monocytogenes
To determine potential effector functions of OT-I cells early postimmunization, we measured Ag-specific cytokine production. At 48 h postimmunization, spleen cells were stimulated with SIINFEKL peptide for 5 h in vitro followed by intracellular cytokine staining. Nearly all of the OT-I cells from both IRL- and live L. monocytogenes-immunized mice were capable of secreting the cytokines TNF-α, IFN-γ, and IL-2 (Fig. 6). OT-I cells from HKL-immunized mice also produced TNF-α. Although OT-I cells responding to HKL were capable of secreting IFN-γ, only ~70% of the cells did so and the mean fluorescence intensity of IFN-γ+ OT-I cells from HKL-immunized mice was lower than that of IFN-γ+ OT-I cells from IRL- or live L. monocytogenes-immunized mice (Fig. 6). In agreement with the gene-array data (Tables I, II), the major difference observed was in IL-2 production in that OT-I cells isolated from HKL-immunized mice failed to secrete IL-2 (Fig. 6). These results suggested that although effector function was induced by HKL the lack of IL-2 production may have downstream consequences in sustaining the response (39).
HKL immunization leads to reorganization of splenic architecture
Inadequate DC activation has been proposed as a likely reason for the poor CD8 T cell response after HKL immunization (40, 41). To test whether providing inflammatory signals during HKL immunization would reverse the defective response, we coimmunized mice with HKL and 107 CFU of actA− L. monocytogenes (rLM-OVA−) that does not express OVA or with rLM-OVA− alone. Surprisingly, the mice that were coimmunized with HKL and rLM-OVA− did not survive beyond 48 h. Conversely, the mice that were immunized with rLM-OVA− alone survived and appeared to effectively clear the bacteria (data not shown). These data indicated that HKL immunization adversely affected the ability of the animal to control bacterial growth. To investigate this possibility and to ensure the survival of coimmunized animals, we administered HKL and low-dose wild-type LM-OVA (1 × 104 CFU) or LM-OVA alone. The mice were sacrificed at 4 d postimmunization, and the bacterial burdens in spleen and liver were determined. Indeed, the bacterial titers in the livers of animals that were coimmunized were 800-fold higher than those of mice infected with LM-OVA alone (Fig. 7). Coimmunization with IRL and live bacteria yielded similar results (data not shown). Interestingly, bacterial titers in the spleen were not significantly affected by coimmunization with killed and live bacteria.
The lack of early protection following coimmunization suggested that a component of the innate immune system was being negatively affected by HKL treatment. Splenic MZMs have been shown to be important in early control of bacteria and viruses (42–44). Furthermore, both CD11c+ DCs and splenic macrophages have also been implicated in the CD8 T cell response to L. monocytogenes infection (41, 45–47). In addition to DCs and B cells, the splenic MZ contains two types of macrophages called MZMs and MMMs. Therefore, we examined the status of these cell types in the spleen following immunization. Moma-1+ MMMs that line the inner border of the marginal zone (MZ) were not affected by HKL, live L. monocytogenes, or IRL immunization (Fig. 8A; data not shown). However, HKL inoculation resulted in nearly a complete loss of MZMs, and small, rounded ERTR-9+ apparently apoptotic bodies were consistently observed (Fig. 8B). In contrast, after live L. monocytogenes immunization, MZMs were preserved, though a decrease in ERTR-9 staining was observed as compared with that of uninfected controls (Fig. 8B, right panel). Interestingly, IRL immunization only led to a partial destruction of MZMs (data not shown). In addition to the deleterious effect on MZMs, we also observed substantial changes in DC organization in the spleens of HKL-immunized mice (Supplemental Fig. 6). DCs were nearly exclusively localized in the center of the white pulp in the T cell zones. This effect occurred after IRL immunization to a lesser extent but was not evident after live L. monocytogenes infection (Supplemental Fig. 6B, 6C). HKL inoculation resulted in nearly a complete loss of DCs from the MZ and red pulp (Supplemental Fig. 6D). Our data cannot rule out the possibility that the loss of DCs from the red pulp and MZ after HKL immunization was in part due to cell death as well as DC migration into the T cell zones.
TNF-α has previously been shown to be a proapoptotic cytokine that may damage macrophages (48, 49). Therefore, to determine whether TNF was involved in the rapid MZM destruction observed following HKL immunization, we administered anti–TNF-α blocking Ab or isotype control hamster IgG to mice immunized with HKL. Nine hours postimmunization, spleens from mice treated with anti–TNF-α Ab showed no signs of MZM loss, whereas MZMs were depleted in spleens of control animals (Fig. 8C). However, blocking TNF-α early after HKL immunization did not have any effect on OT-I expansion or activation phenotype (Supplemental Fig. 7; data not shown).
As opposed to HKL, other forms of attenuated L. monocytogenes may hold more promise as effective vaccine vehicles. Recent studies have used killed but metabolically active L. monocytogenes (50), conditionally lethal L. monocytogenes mutants (51), and L. monocytogenes deficient in listeriolysin O (52) to decrease infectivity but retain immunogenicity. A very simple approach entails irradiating L. monocytogenes to generate strains unable to replicate, and immunization with IRL provides protective immunity against lethal L. monocytogenes infection (10). Nevertheless, our understanding of how the quality of T cell activation is affected by a particular vaccine inoculation is lacking. Our model system (with three vaccine protocols that yielded three distinct outcomes) provided us an opportunity to identify the pivotal checkpoint(s) during the progression of a T cell immune response that contribute most to the eventual success or failure of a vaccination strategy. We used imaging, cytometric, and gene-array technologies to compare the outcome of inoculation with live L. monocytogenes, HKL, and IRL with respect to early priming events. Our imaging analysis of splenic tissue demonstrated large T cell–DC clusters, found only in spleens of mice immunized with live L. monocytogenes or IRL. Interestingly, compared with live L. monocytogenes these “clusters of activation” were fewer and smaller in size in IRL-immunized mice. In contrast, T cell–DC cluster formation was never observed following HKL inoculation. Similar distinctions have been made in a comparison of CD4 T cells responding to immunogenic versus tolerogenic Ag (53). Thus, cluster formation resulting in sustained TCR signaling (54, 55) appears to be a hallmark of a productive immune response. Furthermore, the size and frequency of early T cell cluster formation was directly proportional to the magnitude of T cell activation and served as a reliable indicator of the quality of T cell differentiation into effector and memory cells. The lack of cluster formation after HKL immunization may be as a result of inefficient TLR activation or other early pattern recognition pathways (10, 56).
Over the past two decades several studies have used microarray analysis to understand the molecular programming of CD8 T cell activation. However, our model provided a unique opportunity to determine the early genetic signatures of CD8 T cells following three distinct immunization strategies. Moreover, extensive analysis of gene expression by CD8 T cells within hours after microbial immunization, prior to cell division, has not been performed. Indeed, microarray analysis of CD8 T cells isolated 24 h after HKL, IRL, or live L. monocytogenes immunization revealed many interesting features that characterized early CD8 T cell activation. As reported previously (35), even resting, naive OT-I CD8 T cells exhibited an active gene expression program. This was particularly illustrated by high signal intensities of genes that code for cytokine receptors, TCR proteins, and signaling molecules. After HKL, IRL, or live L. monocytogenes immunization, changes in gene expression were striking, with a similar number of genes exhibiting increased or decreased expression. Compared to HKL- or IRL-activated OT-I cells, changes in gene expression in live L. monocytogenes-activated T cells were more extensive. Interestingly, as early as 24 h postimmunization, many cytokine genes were highly upregulated and these included IL-3, IL-10, and IL-17F, which were unexpected. The role of IL-10 as a positive or negative regulator of anti-L. monocytogenes CD8 responses remains unclear (57, 58), as are the roles of IL-3 and IL-17 in CD8 T cell activation. With the exception of IL-2R, the extensive downregulation of several other cytokine receptor genes was intriguing and suggested that the inability to respond to certain cytokines, as well as the ability to respond to IL-2 early after Ag stimulation, may play a crucial role in T cell activation and subsequent clonal expansion and survival. Although genes such as CTLA-4, LAG3, and CD225 code for inhibitory proteins, their elevated expression at such an early stage of CD8 T cell activation was surprising. Moreover, many of these genes were most highly expressed in live L. monocytogenes-activated CD8 T cells, suggesting that even very early after Ag encounter T cell activation was influenced by a counterbalance of positive and negative regulation. Differentiation patterns based on the expression of genes known to regulate cytokine expression and T-effector lineage development were also evident very early after activation. Thus, ROG was highly upregulated, as was Tbx21 (T-bet), whereas GATA-3 was substantially downregulated. ROG is a transcription factor that is known as a potent negative regulator of CD4 T cell activation (59–61), and live L. monocytogenes activation resulted in a nearly 200-fold increase in ROG mRNA over naive OT-I cells. Upregulation of ROG leading to downregulation of GATA-3 and the concomitant induction of T-bet (62) correlate well with the “T cytotoxic type-1”-like profile of cytokine expression of L. monocytogenes-activated CD8 T cells. However, the differences in the levels of induction of these genes based on strength of signal also suggested that differentiation of a particular effector lineage is also affected by the nature of the immunogen. Overall, the mRNA expression profiles suggested that each type of L. monocytogenes-based vaccine initiated a unique molecular program of CD8 T cell activation that subsequently led to three distinct outcomes.
Protein expression analysis supported the concept that protracted (live L. monocytogenes or IRL) versus transient (HKL) T cell–DC interactions imprinted a distinct program of activation on responding Ag-specific CD8 T cells. For example, the level of CD3 downregulation correlated with the apparent strength of signaling and is the result of continuous engagement of multiple TCRs with peptide–MHC molecules expressed by APCs (63, 64). Moreover, CD3/TCR internalization may serve to facilitate intracellular signaling that leads to the activation of transcription factors required for T cell activation (64, 65). The lack of CD3 downregulation after HKL immunization was therefore a further indication that the cascade of signaling pathways leading to proper CD8 T cell activation was inadequately initiated. The disparate paths taken by the three differentially programmed responses became increasingly evident when protein expression was analyzed on OT-I cells after the onset of cell division. In particular, the level of PD-1 expression correlated well with the level of activation. Interaction of PD-1 with PD-L1 and PD-L2 is thought to deliver an inhibitory signal to T cells, resulting in lower proliferation, cytolytic activity, and cytokine production (66, 67). In mice, blocking the interaction of PD-1 with its ligands can exacerbate autoimmune diseases (68–70) and rejuvenate exhausted T cells in persistent infections (67). However, PD-1 may also act as a costimulator after Ab-mediated agonism (71) and was recently shown to augment the anti-L. monocytogenes CD8 T cell response (72). Judging from our genetic analysis, in which several other genes encoding negative regulators of T cell expansion were expressed early, it is likely that temporal differences in expression of distinct sets of positive and negative regulators control optimal T cell activation and differentiation.
Unlike the transient expression of PD-1, NKG2D was stably expressed at least until day 7 postimmunization on CD8 T cells from IRL- or live L. monocytogenes- but not HKL-immunized mice. The latter could be a significant contributing factor to the poor outcome of CD8 T cells following HKL immunization. In mice, NKG2D binds to stress-induced ligands, such as retinoic acid early inducible-1 and H60 (73,74), which leads to an interaction with a transmembrane adapter protein DAP10. This coupling of NKG2D with DAP10 in CD8 T cells initiates multiple signaling pathways that augment cytotoxicity, costimulation, and cell survival at least in vitro (74) and thus may play a role in vivo in driving productive T cell responses. The rapidity of re-expression of IL-7Rα also correlated with the differential levels of activation by HKL, IRL, and live L. monocytogenes. Although IL-7Rα downregulation occurred after all three types of immunization, by day 7 postimmunization most CD8 T cells responding to IRL and live L. monocytogenes expressed high levels of IL-7Rα, whereas HKL-activated CD8 T cells were IL-7Rαlow. In general, high levels of IL-7Rα expression correlate with the eventual production of memory cells (75, 76) and our results support this concept. Finally, the ability of responding CD8 T cells to produce IL-2 also predicted efficient priming. Although IL-2 is not required to initiate CD8 T cell blastogenesis, IL-2 is important in maximizing the magnitude of the response (39). As indicated by the lack of adequate mRNA and protein expression in HKL-activated CD8 T cells, early production of IL-2 by CD8 T cells could be important in driving T cell clonal expansion to fruition; thus the inability of HKL-immunized CD8 T cells to secrete IL-2 may contribute to the abortive expansion exhibited by these cells.
The differentiation profile as indicated by the transcriptional and translational signature of OT-I cells after live or killed bacterial immunization can be influenced by the longevity of Ag and inflammation. Sustained Ag presentation and inflammation after live bacterial immunization may lead to increased OT-I cell proliferation and activation. However, despite limiting bacterial survival to 24 h after live L. monocytogenes immunization by antibiotic treatment, the OT-I expansion and activation profile did not change when compared with that of untreated mice. These data agreed with a previous study showing that expansion of Ag-specific CD8 T cells is independent of the duration and severity of in vivo bacterial infection (77). However, differences in inflammatory mediators within the first 24 h (prior to antibiotic treatment) after live or killed bacterial immunization may contribute to the unique phenotypic profile of activated Ag-specific CD8 T cells after each type of immunization.
We have previously shown that naive precursor frequency can alter the differentiation patterns of CD8 T cells with respect to lineage commitment of central and effector memory T cell populations. However, in the current study, our goal was to analyze the earliest events that occur in CD8 T cell activation after the three different types of immunizations. Nevertheless, reducing the precursor frequency of naive OT-I cells failed to rescue CD8 T cell expansion following HKL immunization. Interestingly, the only substantial difference that was noted with low-dose transfer was in the induction of NKG2D, which was enhanced when lower cell numbers were transferred, suggesting that competition and strength of signal may influence expression of this protein. Further studies will be required to investigate this intriguing possibility.
The unexpected death of mice that were coadministered HKL and high-dose attenuated L. monocytogenes (actA−) led us to investigate further the effect of HKL on splenic macrophages and DCs. Macrophage populations in the spleen have long been considered as the first line of defense against blood-borne bacterial or viral pathogens (42, 43, 78, 79). The apparent rapid destruction of ERTR-9+ macrophages that line the outer edge of the MZ following HKL immunization was surprising, although this result helped to explain the inability of the mice to control bacterial growth. Previous studies using clodronate liposomes (43, 78, 79) or osteopetrotic−/− mice (42), where both MMM and MZM populations are deficient, have shown that these macrophage populations are essential for controlling viral and bacterial spread. Our data extend these observations by showing that the initial control of L. monocytogenes infection was dependent primarily on ERTR-9+ MZMs. Macrophages can transfer bacterial Ag to DCs that in turn activate naive T cells in vitro (80) and in vivo (81). Nonetheless, destruction of MZMs by HKL immunization may disrupt the sequence of events leading to adequate activation of DCs, thus resulting in poor CD8 T cell priming; however, blocking TNF-α during HKL immunization did not rescue OT-I expansion. Due to the loss of the MZMs resulting from HKL immunization, the spleen was no longer able to handle the bacterial burden and this in turn resulted in a systemic increase in bacteria as shown for the liver in Fig. 7. This phenomenon has also recently been observed by the Bahjat et al. (82) who show that IL-10 produced by macrophages also plays an inhibitory role during coimmunizations with HKL or other L. monocytogenes mutants that are unable to exit the phagosome. Thus, damage of the MZ and induction of regulatory mechanisms likely work in concert, thus resulting in poor CD8 T cell immunity.
The mechanism of MZM-specific destruction following HKL immunization was mediated by TNF-α. Why HKL and not live L. monocytogenes-induced TNF-α–mediated damage remains unclear. However, splenic macrophages are a potent source of TNF-α, and autocrine TNF action can lead to macrophage apoptosis (48, 83). Our attempt to prevent L. monocytogenes spread after coimmunization by treatment of anti–TNF-α Ab, however, was predictably unsuccessful (data not shown), given the importance of TNF-α in L. monocytogenes clearance postinfection (84). The fact that bacterial burdens were significantly elevated in tertiary tissues after coimmunization of live L. monocytogenes with HKL or IRL suggested that caution must be used in coadministration of multiple vaccines (especially in the face of ongoing or imminent infections), as it is routinely done in infants as young as 2 mo, because deleterious effects may result. Also, a rapid public health response to immediate intentional infection as in bioterrorism could use killed vaccines, which might prove harmful in the face of concomitant infection.
Overall, our results provided evidence for a set of genetic, phenotypic, functional and anatomic events that were predictive of the outcome of the response. Therefore, the analysis of these parameters should aid in rational vaccine design by identifying effective immunization regimens.
We thank Dr. Brenton Gravely and Anupinder Kaur (University of Connecticut Health Center Translational Genomics Core) for assistance and Diane Gran (Flow Cytometry Core, University of Connecticut Health Center, Farmington, CT) for expert cell sorting assistance.
Disclosures The authors have no financial conflicts of interest.
This work was supported by National Institutes of Health Grants P01 AI56172 (to L.L., A.T.V., and S.J.M.), AI76457 (to L.L.), and T32 AI07080 (to D.A.B.); Damon Runyon Cancer Research Foundation Fellowship DRG-1886-05 (to K.M.K.); and in part by the Intramural Research Program of the National Institute of Allergy and Infectious Diseases of the National Institutes of Health (to S.K.D.).
The gene expression data presented in this article have been submitted to the National Center for Biotechnology Information Gene Expression Omnibus under accession number GSE20265.
The online version of this article contains supplemental material.
Abbreviations used in this paper:
- B cell zone
- dendritic cell
- heat-killed Listeria monocytogenes
- IFN regulatory factor
- irradiated L. monocytogenes
- Kruppel-like factor
- L. monocytogenes
- marginal metallophilic macrophage
- marginal zone macrophage
- program cell death-1
- recombinant L. monocytogenes-producing OVA
- repressor of GATA
- second harmonic generation
- T cell zone
- T cell-specific T-box transcription factor.
- Received April 1, 2009.
- Accepted April 18, 2010.
- Copyright © 2010 by The American Association of Immunologists, Inc.