CD8⁺ T Cell Exhaustion, Suppressed Gamma Interferon Production, and Delayed Memory Response Induced by Chronic Brucella melitensis Infection

Abstract

Brucella melitensis is a well-adapted zoonotic pathogen considered a scourge of mankind since recorded history. In some cases, initial infection leads to chronic and reactivating brucellosis, incurring significant morbidity and economic loss. The mechanism by which B. melitensis subverts adaptive immunological memory is poorly understood. Previous work has shown that Brucella-specific CD8⁺ T cells express gamma interferon (IFN-γ) and can transition to long-lived memory cells but are not polyfunctional. In this study, chronic infection of mice with B. melitensis led to CD8⁺ T cell exhaustion, manifested by programmed cell death 1 (PD-1) and lymphocyte activation gene 3 (LAG-3) expression and a lack of IFN-γ production. The B. melitensis-specific CD8⁺ T cells that produced IFN-γ expressed less IFN-γ per cell than did CD8⁺ cells from uninfected mice. Both memory precursor (CD8⁺ LFA1^HI CD127^HI KLRG1^LO) and long-lived memory (CD8⁺ CD27^HI CD127^HI KLRG1^LO) cells were identified during chronic infection. Interestingly, after adoptive transfer, mice receiving cells from chronically infected animals were able to contain infection more rapidly than recipients of cells from acutely infected or uninfected donors, although the proportions of exhausted CD8⁺ T cells increased after adoptive transfer in both challenged and unchallenged recipients. CD8⁺ T cells of challenged recipients initially retained the stunted IFN-γ production found prior to transfer, and cells from acutely infected mice were never seen to transition to either memory subset at all time points tested, up to 30 days post-primary infection, suggesting a delay in the generation of memory. Here we have identified defects in Brucella-responsive CD8⁺ T cells that allow chronic persistence of infection.

INTRODUCTION

Brucellosis caused by Brucella melitensis has a high incidence in developing countries, and the World Health Organization considers brucellosis one of the seven neglected zoonoses, a group of diseases that contribute to the perpetuation of poverty (1, 2). Brucella has many mechanisms to survive and replicate in hostile host cells, including inducing the unfolded-protein response (UPR), hijacking host nutrients, and counteracting the effects of pH changes, among many others (3–6). The chronic, reactivating nature of Brucella infection, along with its stealthy intracellular life-style, makes infections difficult to clear and requires lengthy antibiotic treatment (7–9).

CD8⁺ T cells control intracellular infections by identifying and killing compromised host cells as a part of the adaptive immune response (10, 11). Recognition of nonself antigenic epitopes in the context of major histocompatibility complex (MHC) class I by cytotoxic T cells also leads to the release of effector molecules to increase local inflammation, thereby “raising the alert” of the host in response to intracellular infection (12). A subset of MHC class I-restricted epitopes of Brucella melitensis generated during infection has been characterized and can elicit specific CD8⁺ T cells (13). These T cells have been shown to kill their target cells, release cytokines, and survive into the chronic phase of infection (7). Why, then, in the successful establishment of chronic brucellosis, do we see the highly evolved CD8⁺ T cell arm of adaptive immunity fail to protect the host from long-term infection?

Immunological memory is the ability of the host to mount a fast, effective secondary response to infection. CD8⁺ T cell memory is derived from effectors generated during primary infection or vaccination, a small cohort of which then transitions to a memory precursor phenotype (14–17). Memory precursors, given the right environment, become self-renewing long-lived memory cells (17, 18). CD8⁺ T memory and memory precursors with the CD8⁺ LFA1^HI CD127^HI KLRG1^LO phenotype are distinguished from effector populations by increased levels of surface interleukin-7 (IL-7) receptor (CD127) expression (16, 19–22). Upon binding extracellular IL-7, IL-7 receptor sends an intracellular antiapoptotic signal that the cell needs to sustain the self-renewing state necessary for a long-term antigen-specific memory response (23). In contrast, killer cell lectin-like receptor G1 (KLRG1) expression is decreased in memory precursor and long-lived CD8⁺ T memory populations (18, 24). KLRG1^HI CD8⁺ T cells are characterized as short-lived effectors fated for apoptosis during the T cell contraction phase and those cells that may be transitioning to other states (16). CD27, a tumor necrosis factor (TNF) family receptor, is expressed at high levels in parallel to IL-7 receptor on cells that have survived the antigen-specific CD8⁺ T cell contraction phase to become terminally differentiated, long-lived memory cells (21).

Chronic infections can erode the CD8⁺ memory population by inducing dysfunction via multiple mechanisms, including T cell exhaustion (25). T cell exhaustion is marked by a progressive loss of functionality (i.e., cytokine expression and killing) and fixed surface expression of inhibitory receptors, including programmed cell death 1 (PD-1) and lymphocyte activation gene 3 (LAG-3) (25, 26). Exhausted T cells are inferior to naive T cells at protecting against challenge (27). There are well-documented examples of CD8⁺ T cell failure during other chronic infections (e.g., lymphocytic choriomeningitis virus), including instances of exhaustion, tolerance, and anergy (25). However, the explanation for CD8⁺ T cell failure during chronic brucellosis infection remains unidentified (7, 28).

A very small number of Brucella-specific CD8⁺ T cells exhibiting a T memory phenotype of LFA1^HI CD127^HI KLRG1^LO have been detected up to 9 months postinfection (7). However, Brucella-specific CD8⁺ T cells were not polyfunctional when presented with specific targets in vitro: they were unable to produce TNF-α and IL-2 and possessed only a limited ability to produce gamma interferon (IFN-γ) (7). This lack of polyfunctionality is a trait of low-quality secondary responses that corresponds to the chronic reactivating nature of brucellosis (7). Beyond this lack of polyfunctionality, it is not known whether CD8⁺ T cells present during the chronic phase of Brucella infection can respond by reentering the effector-to-memory transition or by increasing cytokine expression when rechallenged in vivo with antigen. Are these CD8⁺ T cells permanently disabled, or are they capable of mounting a response under appropriate conditions? Discerning whether a cell-intrinsic deficit in functionality or external environmental regulation contributes to the failure of CD8⁺ T cell-mediated immunity will better direct future vaccine design efforts to overcome this dysfunction. Insight into defective memory generation during chronic brucellosis may also have implications for other persistent intracellular infections.

To further investigate the mechanisms underlying chronic infection with Brucella, we sought to identify the breakdown in CD8⁺ T cell immunity. Here we examined CD8⁺ T cells from both the acute and chronic phases of B. melitensis infection for evidence of exhaustion, IFN-γ production, and the development of effector-to-memory transition phenotypes. Using adoptive transfer, we determined the capacity of splenocytes from acutely and chronically infected mice to contribute to Brucella containment in a naive host.

In mice with chronic brucellosis, we identified a small number of memory precursor and long-lived CD8⁺ T cells. We also found that a cohort of CD8⁺ T cells is exhausted during chronic brucellosis and that other responding CD8⁺ T cells have a diminished ability to produce IFN-γ. Although a proportion of the Brucella-specific CD8⁺ T cells were exhausted and had a deficit in IFN-γ production, they were able to participate in the response to infection when challenged in a naive host after adoptive transfer. These findings suggest that CD8⁺ T cells could play a more active role in the host response to infection when the cell-extrinsic Brucella-induced suppressive environment generated by chronic infection is removed or potentially therapeutically altered. This may allow new avenues of vaccine design, taking into account the need to maximize the ability of host CD8⁺ T cells to participate in the response.

MATERIALS AND METHODS

Immunization of mice.

Female BALB/c mice (6 to 8 weeks of age) were obtained from Harlan (Indianapolis, IN) and housed in AAALAC-approved facilities under pathogen-free conditions according to standard protocols. For initial immunization, groups of 16 mice were immunized intraperitoneally (i.p.) with 10⁷ virulent B. melitensis strain GR023 bacteria 6 months prior to sacrifice for chronic infections and 2 weeks prior to sacrifice for acute infections (29). Groups of 4 uninfected age-matched mice were injected i.p. with phosphate-buffered saline (PBS). For challenge experiments, groups of 16 recipient mice were challenged i.p. with 10⁶ B. melitensis strain GR023 bacteria immediately following adoptive transfer of splenocytes from previously infected animals. All animal experiments were conducted upon review and approval from the Institutional Animal Care and Use Committee in compliance with the Guide for the Care and Use of Laboratory Animals (30).

Cell tracking and adoptive transfer.

Spleens from Brucella-infected or naive animals were homogenized, strained through a 70-μm cell strainer, and immediately treated with red blood cell (RBC) lysis buffer. The splenocytes were washed thoroughly and then stained with 10 μm carboxyfluorescein diacetate succinimidyl ester (CFDA-SE/CFSE; Invitrogen) prior to transfer. CFSE-stained cells were adoptively transferred (1 × 10⁷ total cells/mouse) via retro-orbital injection into anesthetized 8-week-old female BALB/c mice within an average of 2 h of the donor splenocytes being collected. The CFSE gating strategy for flow cytometric analysis is shown in Fig. S1 in the supplemental material. Only CFSE-positive (CFSE⁺) cells that were clearly distinguishable from the negative population were used for analysis to minimize any crossover between highly proliferative cells and dead or otherwise CFSE-negative (CFSE⁻) cells.

Determination of bacterial load.

Serial dilution was employed for colony counting to determine the number of Brucella bacteria within tissue samples of infected mice. Splenocytes were lysed in 1 ml water plus 0.1% Triton X-100. Next, 100 μl of lysate was loaded into the first well of each row in a 96-well plate, and 10-fold serial dilutions were made. Eight replicates of each dilution were plated onto brucella broth agar in square plates. Plates were allowed to dry and then incubated at 37°C for 4 days, and CFU were determined.

Flow cytometry and intracellular cytokine assays.

Splenocytes from Brucella- and PBS-immunized mice were passed through a 70-μm strainer and treated with RBC lysis buffer (BioLegend). The phenotype of a portion of the cells was determined by using cell surface staining with anti-CD8 (clone 53-6.7; BioLegend), anti-CD3 (clone 145-2C11; BD Biosciences), anti-LFA1 (leukocyte functional antigen 1, also known as CD11a) (clone M17/4; eBioscience), anti-CD127 (clone A7R34; eBioscience), anti-KLRG1 (clone 2F1; BioLegend), anti-CD27 (clone LG.3A10; BioLegend), anti-PD-1 (clone J43; BD Biosciences), and/or anti-LAG-3 (clone C9B7W; BD Biosciences). For the intracellular cytokine staining assays, a portion of splenocytes was cultured to identify intracellular IFN-γ by using 96-well round-bottom plates (1 × 10⁶ cells/well) in complete medium in the presence of the purified MHC class I Brucella-specific peptide NGSSSMATV or an irrelevant peptide (13) (>90% purity; GenScript, Piscataway, NJ) and 10 μg/ml GolgiPlug (BD Biosciences, San Jose, CA), with or without concanavalin A (Sigma-Aldrich, St. Louis, MO). After 5 h, cells were surface stained with anti-CD8 and anti-CD3 and then fixed and permeabilized according to the manufacturer's protocol for the Cytofix/Cytoperm kit (BD Biosciences), with subsequent intracellular staining with anti-IFN-γ (clone XMG1.2; BD Biosciences). Isotype and positive controls were utilized to confirm the specificity of antibody staining (see Fig. S2 to S4 in the supplemental material). Flow cytometry was performed by using an FC500 instrument (Beckman Coulter, Fullerton, CA). Data were further analyzed by using FlowJo software vX.0.7 (Tree Star, Ashland, OR) and Prism v5.0f (GraphPad Software, La Jolla, CA). Lymphocytes were first gated by using the forward-scatter-versus-side-scatter plot. For the recipients of adoptive transfers, lymphocytes were then determined to be CFSE⁺ or CFSE⁻ by using positive- and negative-control populations and taking into consideration stain fading due to cellular division and time since transfer. Tests for all samples were run in triplicate, and all experiments were duplicated to confirm the presence or absence of rare cell populations.

Statistics.

To determine statistical significance, a paired Student t test and analysis of variance (ANOVA) were performed on the data, with additional analysis using Dunnett's or Bonferroni's multiple-comparison test as needed. All tests were performed by using GraphPad Prism v5.0f (GraphPad Software).

RESULTS

Identification of exhausted CD8⁺ T cells during chronic Brucella infection.

Although memory phenotype CD8⁺ LFA1⁺ CD127^HI KLRG1^LO T cells are present during the chronic phase of Brucella infection, evidence suggests that they do not provide the expected protection from reinfection or reactivation of disease (7, 8). To increase our understanding of why Brucella-specific CD8⁺ T cells fail to resolve infection or mount a successful secondary response, in this study, we analyzed the differences in phenotype and functionality between CD8⁺ T cells found during the acute phase of brucellosis (2 weeks postinfection) and those found during the chronic phase (6 months postinfection). Animals in all groups were 8 weeks of age at the time of primary infection. To control for age-related senescence of the immune response, naive age-matched mice were included for both acute and chronic groups. Thus, acutely infected mice and their age-matched naive counterparts were 10 weeks of age at sacrifice, while chronically infected mice and their age-matched naive counterparts were 8 months of age at sacrifice.

Mice were injected i.p. with PBS (referred to as the naive group) or 1 × 10⁷ B. melitensis bacteria. The dose of 1 × 10⁷ B. melitensis bacteria was chosen as it was previously used for studies of chronic infection where bacteria were eventually cleared from the spleen but extrasplenic infections were established (7). Splenocytes from animals with acute brucellosis (infected for 2 weeks) or chronic brucellosis (infected for 6 months) were analyzed following a 5-h ex vivo peptide pulse by adding the purified Brucella epitope NGSSSMATV to the splenocytes in the presence of brefeldin A. In striking contrast to either acutely infected or naive mice, spleens from chronically infected mice contained a significantly increased number of phenotypically exhausted, CD8⁺ PD-1⁺ LAG-3⁺ cells (Fig. 1A). Further examination of these phenotypically exhausted CD8⁺ T cells revealed a deficit in IFN-γ production (Fig. 1B and C). These results suggest that during chronic brucellosis, a significant portion of the total CD8⁺ T cells become exhausted, thereby limiting their immunological function.

FIG 1.

CD8⁺ T cell exhaustion. Splenocytes were purified from groups of 12 BALB/c mice 2 weeks (Acute) or 6 months (Chronic) after i.p. infection with 10⁷ Brucella melitensis strain GR023 bacteria. Cells were pulsed with the Brucella species-specific peptide NGSSSMATV and then stained for surface CD8, LAG-3, and PD-1 and intracellular IFN-γ. The experiment was done in duplicate. (A) PD-1 and LAG-3 staining profiles of splenic CD8⁺ T cells from acutely infected (left) and chronically infected (right) animals. Splenocytes from 4 uninfected age-matched animals are shown at the bottom. The numbers indicate the percentages of total CD8⁺ T cells that are PD-1⁺ LAG-3⁺ in each experimental group. CD8⁺ PD-1⁺ LAG-3⁺ T cells were then interrogated for intracellular IFN-γ (shown in panel C). (B) Histograms of IFN-γ staining of phenotypically exhausted splenocytes normalized to the mode. IFN-γ expression of total splenocytes is shown in gray, and IFN-γ expression of CD8⁺ PD-1⁺ LAG-3⁺ T cells is shown as a black line. Note that total splenocytes have detectable IFN-γ (gray arrow), while CD8⁺ PD1⁺ LAG3⁺ cells do not. (C) Number of exhausted (CD8⁺ PD-1⁺ LAG-3⁺ IFN-γ⁻) cells in the acutely infected, chronically infected, and age-matched uninfected groups. Statistical significance was determined by ANOVA, with confirmation using Dunnett's multiple-comparison test. A P value of ≤0.05 was considered a significant difference: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

Numbers of IFN-γ-expressing CD8⁺ T cells increase during Brucella infection, but these cells produce less IFN-γ per cell than do those from naive mice.

Although the numbers of total CD8⁺ T cells increased in response to infection with B. melitensis, the IFN-γ contribution from nonexhausted CD8⁺ T cells is thought to be eclipsed by CD4⁺ T cell-derived IFN-γ for the control and clearance of infection (6, 13, 28). We sought to further elucidate why, with this increase in number, IFN-γ-expressing CD8⁺ T cells fail to substantially contribute to the control of infection.

After a 5-h ex vivo peptide pulse as described above, we compared the IFN-γ expression levels of CD8⁺ T cells from naive and acutely and chronically infected mice. Significantly more CD8⁺ IFN-γ⁺ T cells were present in both the acutely and chronically infected groups than in their age-matched naive counterparts. Acutely infected mice also had significantly more CD8⁺ IFN-γ⁺ T cells than did chronically infected mice (Fig. 2A). The mean fluorescence intensity (MFI) of the IFN-γ produced by CD8⁺ T cells was analyzed to determine the amount of cytokine produced on a per-cell basis. The amount of IFN-γ from individual CD8⁺ cells was significantly decreased in both acutely and chronically infected mice compared to the those in age-matched naive mice (Fig. 2B), suggesting that while higher numbers of CD8⁺ T cells express IFN-γ during infection with B. melitensis, CD8⁺ T cells from infected mice produce less IFN-γ per cell than do naive CD8⁺ T cells. In comparison to the findings of exhausted T cells during chronic infection only, diminished IFN-γ production appears relatively early following infection, and this defect apparently persists to at least 6 months.

FIG 2.

IFN-γ expression of CD8⁺ T cells. Splenocytes were purified from groups of 12 BALB/c mice 2 weeks (Acute) or 6 months (Chronic) after i.p. infection with 10⁷ Brucella melitensis strain GR023 bacteria. Groups of 4 age-matched uninfected animals served as controls. Cells were pulsed with the Brucella species-specific peptide NGSSSMATV and then stained for surface CD8 and intracellular IFN-γ expression. The experiment was done in duplicate. (A) Numbers of CD8⁺ T cells expressing IFN-γ in the acutely infected, chronically infected, and age-matched uninfected groups. Statistical significance was tested by using the unpaired t test, with confirmation using ANOVA and Dunnett's multiple-comparison test. (B) MFI of IFN-γ staining in CD8⁺ IFN-γ⁺ T cells. Note that both acutely and chronically infected animals show a lower IFN-γ MFI corresponding to less IFN-γ produced per CD8⁺ IFN-γ⁺ T cell in infected animals. This difference is not related to cell size, as CD8⁺ T cells from infected animals are larger than cells from uninfected age-matched animals (data not shown). Statistical significance was determined by an unpaired t test. A P value of ≤0.05 was considered a significant difference: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

CD8⁺ T memory precursors and long-lived memory cells are found in chronically infected mice but not in acutely infected mice.

We previously identified memory phenotype CD8⁺ T cells during chronic brucellosis (7). Multiple memory phenotypes are now better clarified to distinguish short-lived effectors from memory precursors and terminally differentiated effectors from terminally differentiated long-lived memory CD8⁺ T cells (14, 16–18, 31, 32).

Following a 2-week acute or a 6-month chronic infection with B. melitensis, murine splenocytes were purified and immediately stained for the surface markers CD8, LFA1, CD127, and KLRG1 or CD8, CD27, CD127, and KLRG1. We found significantly higher numbers of both CD8⁺ LFA1^HI CD127^HI KLRG1^LO (Fig. 3A and B) memory/memory precursor and CD8⁺ CD27^HI CD127^HI KLRG1^LO (Fig. 3C and D) long-lived memory cells in chronically infected mice than in acutely infected mice.

FIG 3.

Memory CD8⁺ T cells. Splenocytes were purified from groups of 12 BALB/c mice 2 weeks (Acute) or 6 months (Chronic) after i.p. infection with 10⁷ Brucella melitensis strain GR023 bacteria. Groups of 4 age-matched uninfected animals served as controls. The experiment was done in duplicate. (A) Splenocytes were surfaced stained with anti-CD8, anti-LFA1 (CD11a), anti-CD127, and anti-KLRG1. The number of memory precursor CD8⁺ LFA1^HI CD127^HI KLRG1^LO T cells per 5 × 10⁵ lymphocytes is shown. (B) Representative staining profile of CD8⁺ LFA1^HI T cells. (C) Splenocytes were surface stained with anti-CD8, anti-CD27, anti-CD127, and anti-KLRG1. The number of long-lived memory CD8⁺ CD27^HI CD127^HI KLRG1^LO T cells is shown. (D) Representative staining profile of CD8⁺ CD27^HI T cells. Statistical significance was determined by an unpaired t test. A P value of ≤0.05 was considered a significant difference: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

Here we found significantly fewer CD8⁺ LFA1^HI CD127^HI KLRG1^LO memory precursor T cells in age-matched (“old”) naive mice than in chronically infected mice, but we also found that these cells were significantly enriched in the naive old mice compared to those in either the acutely infected or naive young mice (Fig. 3A and B). Interestingly, there was not a significant increase in the number of CD8⁺ CD27^HI CD127^HI KLRG1^LO long-lived memory cells in chronically infected mice compared to those in naive old mice (P = 0.19), but these long-lived memory T cells were significantly enriched in the naive old mice compared to those in either the acutely infected or naive young mice (Fig. 3C and D).

Control of infectious challenge by adoptively transferred splenocytes.

Data from human brucellosis cases and experimental evidence suggest that CD8⁺ T cells are not highly functional during chronic brucellosis and allow for reactivation of disease as well as reinfections in humans (33). To investigate whether this defect is a cell-intrinsic property or due to external immune environmental regulation, we adoptively transferred splenocytes containing CD8⁺ T cells from acutely and chronically Brucella-infected mice into naive recipient mice with subsequent virulent challenge (Fig. 4A and B). The naive mice, not having developed the Brucella-specific immune environment, provided a “fresh” environment for the Brucella-specific CD8⁺ T cells.

FIG 4.

Virulent challenge of recipients of adoptive transfers. Two hours after receiving either splenocytes from Brucella-infected (acute or chronic recipient) or age-matched naive splenocytes (naive recipient) via adoptive transfer, recipient mice were challenged with 10⁶ B. melitensis strain GR023 bacteria. (A) Schematic representation of the experimental protocol. Briefly, groups of 12 chronically or acutely Brucella melitensis-infected animals were infected with 1 × 10⁷ bacteria 180 days or 14 days prior to adoptive transfer, respectively. Groups of 4 age-matched uninfected animals were injected with PBS 180 days or 14 days prior to adoptive transfer, respectively, to serve as controls (not shown). On day 0, splenocytes were collected from all mice, purified, and then stained with CFSE and adoptively transferred into groups of 4 naive recipient BALB/c mice. Three of the recipient mice were then challenged i.p. with 1 × 10⁶ B. melitensis bacteria, while one animal served as the unchallenged recipient control per time point. Splenocytes were collected at days 5 and 15 posttransfer. Adoptive transfer experiments were done in duplicate. (B) CFU on day 0 of acutely and chronically infected donor mice. (C) Bacterial loads in mice on days 2, 5, 7, 15, and 22 after adoptive transfer and challenge. “NC” indicates groups of recipient mice that received the adoptive transfer of cells but were then not challenged with bacteria. Statistical significance was determined by using the unpaired t test, with confirmation using ANOVA and Dunnett's multiple-comparison test. A P value of ≤0.05 was considered a significant difference: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

The bacterial load was assessed 2, 5, 7, 15, and 22 days following transfer of cells from acutely and chronically infected mice and challenge of the recipient mice (Fig. 4C). Functional immunological memory is anticipated to be more swift and effective than a primary adaptive immune response. Interestingly, at day 2, recipients of cells from chronically infected mice (chronic recipient group) had a higher bacterial load than did recipients of cells from acutely infected mice (acute recipient group) and naive recipient mice, suggesting an initial immunological dysfunction of cells from chronically infected mice. At days 5 and 7 post-adoptive transfer, all groups of challenged recipient mice had very similar bacterial loads. By day 15 after transfer and challenge, the recipients of cells from chronically infected mice had a much-reduced level of infection compared to those of the other challenged groups (Fig. 4C). At day 22 after transfer and challenge, naive cell recipient mice were the only remaining animals with significant bacterial loads (Fig. 4C). Mice that received cells from previously infected or PBS-treated mice via adoptive transfer but that were not challenged are designated “NC” (not challenged). Note that the acute cell recipient group that was not challenged with virulent bacteria demonstrated the presence of a minor infection at 7 days post-adoptive transfer, with a maximum bacterial load of 10¹ CFU (>3 logs lower than that in challenged mice). This suggests that some B. melitensis bacteria were transferred from the acutely infected mice to the recipient mice along with the CFSE-stained splenocytes (Fig. 4A [day 0] and C [day 7]). Recipients of splenocytes from chronically infected mice that were not challenged had no evidence of detectable infection, consistent with data from previous studies using bioluminescent Brucella that revealed intermittent extrasplenic loci during chronic infection (7).

We have shown that CD8⁺ T cells from acutely infected donor mice had not yet transitioned to memory phenotypes at the time of transfer and challenge (Fig. 2A and C [day 0] and 4A). With the transfer of these Brucella-educated splenocytes, the mice in the acute recipient group were not able to control infection better than the mice in the naive recipient group until day 22 after transfer and challenge, consistent with the lack of detected memory during acute primary infection (Fig. 2A and C and 4C).

Exhausted CD8⁺ T cells survive adoptive transfer and challenge.

Exhausted CD8⁺ T cells have not been reported for brucellosis prior to this work, and therefore, it was not known whether these cells would survive adoptive transfer and antigenic challenge. Furthermore, it was not clear if the numbers of these cells would increase or decrease over the course of the 15-day challenge. First, to determine if exhausted CD8⁺ T cells from donor mice could be detected after adoptive transfer, we stained splenocytes from acutely and chronically infected mice with CFSE and adoptively transferred them into naive recipient mice. Recipients were then immediately challenged or not challenged (signified by NC in Fig. 5 to 7). Splenocytes were collected on days 5 and 15 posttransfer and subjected to an in vitro peptide pulse, as described above, followed by intracellular cytokine analysis before fixing the cells and gating on CFSE⁺ cells by using flow cytometry. The CFSE⁺ cells were then interrogated for CD8, PD-1, LAG-3, memory markers, and intracellular IFN-γ (Fig. 5 to 7). The numbers of CD8⁺ PD-1⁺ LAG-3⁺ IFN-γ-negative cells per 10⁴ CFSE⁺ lymphocytes are shown in Fig. 5.

FIG 5.

CD8⁺ T cell exhaustion after adoptive transfer and challenge. Splenocytes were purified from groups of 4 BALB/c mice 5 and 15 days after adoptive transfer and challenge (see Materials and Methods) (Fig. 4). Cells were pulsed with the Brucella species-specific peptide NGSSSMATV and then stained for surface CD8, PD-1, and LAG-3 expression, followed by staining for intracellular IFN-γ expression. “NC” indicates groups of recipient mice that received the adoptive transfer of cells but were then not challenged with bacteria. The experiment was done in duplicate. (A) Comparison of numbers of exhausted CFSE⁺ CD8⁺ PD-1⁺ LAG-3⁺ IFN-γ⁻ T cells in experimental groups on day 5 post-adoptive transfer. (B) Comparison of numbers of exhausted CFSE⁺ CD8⁺ PD-1⁺ LAG-3⁺ IFN-γ⁻ T cells in experimental groups on day 15 post-adoptive transfer. Statistical significance was determined by using ANOVA and Bonferroni's multiple-comparison test. A P value of ≤0.05 was considered a significant difference: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

FIG 7.

Memory phenotypes found in pools of recovered CFSE⁺ CD8⁺ splenocytes. Splenocytes were purified from groups of 3 BALB/c mice 5 and 15 days after adoptive transfer and challenge. Cells were then stained for surface CD8, CD127, KLRG1, and either LFA1 or CD27 and enumerated by using flow cytometry. “NC” and dashed lines indicate groups of recipient mice that received the adoptive transfer of cells but were then not challenged with bacteria. The experiment was done in duplicate. (A) Comparison of numbers of CFSE⁺ CD8⁺ LFA1^HI CD127^HI KLRG1^LO T cells in experimental groups on day 5 post-adoptive transfer. (B) Comparison of numbers of CFSE⁺ CD8⁺ LFA1^HI CD127^HI KLRG1^LO T cells in experimental groups on day 15 post-adoptive transfer. (C) Comparison of numbers of long-lived memory CFSE⁺ CD8⁺ CD27^HI CD127^HI KLRG1^LO T cells in experimental groups on day 5 post-adoptive transfer. (D) Comparison of numbers of CFSE⁺ CD8⁺ CD27^HI CD127^HI KLRG1^LO T cells in experimental groups on day 15 post-adoptive transfer. Statistical significance was determined by using ANOVA and Bonferroni's multiple-comparison test. A P value of ≤0.05 was considered a significant difference: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

We were able to detect exhausted CD8⁺ T cells at day 5 posttransfer from recipient mice that received cells from chronically infected mice and were challenged with Brucella (chronic recipient group) (Fig. 5A). Interestingly, we could not detect this population of cells in animals that received the same cells but were not antigenically challenged (chronic recipient NC group) (Fig. 5A). Although not statistically significant, initial signs of CD8 T cell exhaustion were evident at day 5 in the challenged acute recipient group.

At day 15, the numbers of exhausted cells in mice in the acute and chronic recipient groups remained elevated compared to those in the naive groups (Fig. 5B). By day 15, mice in the acute recipient group subjected to challenge demonstrated an increased cohort of exhausted CD8 T cells compared to those at day 5. As shown in Fig. 1C, a significant cohort of exhausted cells (∼4,000 exhausted CD8⁺ T cells/5 × 10⁵ lymphocytes) existed in the pool of CD8⁺ T cells that were adoptively transferred into the chronic recipient groups from chronically infected donor mice. In comparison, a very small proportion of CD8⁺ T cells was exhausted in the adoptive transfer into the acute recipient mice from the acutely infected donor mice (∼250 exhausted CD8⁺ T cells/5 × 10⁵ lymphocytes); thus, the >400 exhausted CD8⁺ T cells/10⁴ CFSE cells detected at day 15 in the challenged acute recipient group represents an increase.

Numbers of IFN-γ-producing CD8⁺ T cells from infected mice increase after challenge.

To establish whether Brucella-educated CD8⁺ T cells could express IFN-γ upon challenge in a fresh environment, we stained splenocytes from acutely and chronically infected mice with CFSE, adoptively transferred them into naive recipient mice, and analyzed splenocytes at days 5 and 15 posttransfer, as described above. Based on data from previous studies, in a primary infection with B. melitensis, we would expect that numbers of CD8⁺ T cells would be increasing at days 7 to 10 and resolving at 15 days postinfection, during the contraction phase. As shown in Fig. 2, prior to transfer, both acutely and chronically infected mice had a higher number of CD8⁺ T cells expressing IFN-γ than those in their age-matched uninfected counterparts, but this proportion was not maintained 5 days after transfer and challenge: at day 5 posttransfer, more CFSE⁺ CD8⁺ T cells expressed IFN-γ in groups that received cells from acutely or chronically infected mice but were not challenged (acute recipient NC and chronic recipient NC groups) (Fig. 6A). When these cells faced virulent challenge, the numbers of CD8⁺ T cells responding with IFN-γ were significantly decreased (Fig. 6A). These results are consistent with either an expansion of IFN-γ-producing cells in the new environment in the absence of infection or the death of IFN-γ-producing cells if reexposed to infectious challenge.

FIG 6.

IFN-γ expression of recovered CFSE⁺ CD8⁺ T cells. Splenocytes were purified from groups of 3 BALB/c mice 5 and 15 days after adoptive transfer and challenge. Cells were pulsed with the Brucella species-specific peptide NGSSSMATV and then stained for surface CD8 and intracellular IFN-γ. “NC” indicates groups of recipient mice that received the adoptive transfer of cells but were then not challenged with bacteria. The experiment was done in duplicate. (A) Comparison of numbers of CFSE⁺ CD8⁺ IFN-γ⁺ T cells in experimental groups on day 5 post-adoptive transfer. (B) Comparison of numbers of CFSE⁺ CD8⁺ IFN-γ⁺ T cells in experimental groups on day 15 post-adoptive transfer. (C) Comparison of the MFI of IFN-γ from CFSE⁺ CD8⁺ IFN-γ⁺ T cells in experimental groups on day 5 post-adoptive transfer. (D) Comparison of the MFI of IFN-γ from CFSE⁺ CD8⁺ IFN-γ⁺ T cells in experimental groups on day 15 post-adoptive transfer. Statistical significance was determined by using ANOVA and Bonferroni's multiple-comparison test. A P value of ≤0.05 was considered a significant difference: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001.

An interesting change was detected during examination of data from day 15 after transfer and challenge. The numbers of CD8⁺ T cells from acutely and chronically infected mice were still increased, in contrast to cells from the naive recipient and NC groups that were not subject to virulent challenge (Fig. 6B), consistent with expansion of previously exposed CD8⁺ T cells. One of the tenets of adaptive immunological memory is that the secondary response develops faster than the primary response. Interestingly, this prediction is not fulfilled by CD8⁺ T cells from animals acutely or chronically infected with B. melitensis. We show here that numbers of CD8⁺ T cells from acutely or chronically infected animals are still increasing at day 15, and it is only at this time point that CD8⁺ T cells outnumber those found after primary infection (Fig. 6B).

In Fig. 6C and D, the MFI of the IFN-γ staining described above was examined. We show in Fig. 2B that prior to transfer, CD8⁺ T cells from infected mice have a lower MFI, indicative of less IFN-γ being produced per cell, than do cells from naive age-matched mice. At day 5 after transfer and challenge, we see this effect again, and while the acute cells that were challenged had a lower MFI than did NC acute cells, the difference was not quite significant between these two groups (P = 0.07) (Fig. 6C). By day 15, the chronic NC recipient group showed significantly elevated IFN-γ production compared to that in the challenged chronic recipient counterpart (Fig. 6D). Although not statistically significant (P = 0.12) (Fig. 6D), the MFI of IFN-γ expression of the acute recipient NC group was elevated compared to that of the challenged acute recipient group.

Detection of CD8⁺ T memory cells is delayed after adoptive transfer and challenge.

It was unknown not only whether the exceedingly rare CD8⁺ memory cells found in chronically infected mice (Fig. 3A and C) could survive adoptive transfer into naive mice but also how they would respond to subsequent challenge. It was also unknown whether the cells transferred from Brucella-infected mice could reenter the effector-to-memory transition upon virulent challenge in a fresh immune environment. To identify the memory phenotypes that may persist or emerge after transfer, CFSE⁺ cells were interrogated as memory precursors or long-lived memory phenotypes by concurrent staining with anti-CD8, anti-CD127, anti-KLRG1, and either anti-LFA1 or anti-CD27.

The numbers of recovered memory precursor CFSE⁺ CD8⁺ LFA1^HI CD127^HI KLRG1^LO T cells collected per 10⁴ CFSE⁺ lymphocytes at each time point and for each experimental group are shown in Fig. 7. We show in Fig. 3 that chronically infected mice have small but detectable pools of memory precursors and long-lived memory cells. Following adoptive transfer and challenge, numbers of memory precursors were minimal in any group at day 5 (Fig. 7A). Furthermore, there was no significant differences between experimental groups in the numbers of long-lived memory cells found at 5 days posttransfer (Fig. 7C). Notably, at day 15, memory precursors and long-lived memory cells recovered from challenged mice in the chronic recipient group were found at a much higher frequency (roughly 2 to 3% of CFSE⁺ lymphocytes) (Fig. 7B and D) than what was found prior to transfer in the spleens of chronically infected mice (0.1 to 0.12% of lymphocytes) (Fig. 3A and C) but only in challenged mice. Thus, memory precursor and long-lived memory cells expanded in response to challenge posttransfer in the new environment. Of note, numbers of memory precursors and long-lived memory cells did not increase in acute recipients that were challenged, even at day 15 posttransfer, suggesting a delay of at least 30 days in the acquisition of memory during active infection.

DISCUSSION

Here we have identified a cohort of exhausted CD8⁺ T IFN-γ-negative cells expressing LAG-3 and PD-1 in mice with chronic brucellosis. Also, CD8⁺ T cells from chronically infected animals that still expressed IFN-γ produced less IFN-γ on a per-cell basis than did those from uninfected animals, supporting a defective/dysfunctional phenotype. Upon transfer, some of the CD8⁺ T cells derived from B. melitensis-infected mice appear functional and participate in the response to challenge, although these B. melitensis-specific cells from chronically infected animals still make less IFN-γ than do their naive counterparts. Additionally, exhausted CD8⁺ T cells from chronically infected animals survive adoptive transfer and rechallenge. The transferred T memory cells expand in the spleen upon challenge and likely participate in the challenge response, consistent with the decreased numbers of CFU observed at days 15 to 22. Together, our results indicate that CD8⁺ T cells can participate in the response to secondary infection or challenge when removed from Brucella-induced chronic suppressive stimulation and placed into a fresh/naive environment. Both cell-intrinsic (stunted IFN-γ production and delayed increase in the number of IFN-γ-producing cells) and cell-extrinsic factors due to an environment of chronic infection were found to affect the ability of CD8⁺ T cells to control chronic brucellosis.

Current vaccine constructs do not generate a viable CD8⁺ T cell immune response and moreover ignore CD8⁺ T cell-mediated immunological memory because these cells have been considered unimportant in the control of human infection (34). An understanding of how to rescue exhausted CD8⁺ T cells may be critical to decreasing the incidence of reactivation and reinfection, even after vaccination. By blocking PD-1-mediated inhibitory signals and supplementation with additional IL-2, others reported that exhaustion is a reversible phenotype of CD8⁺ T cells (26, 35–37). For instance, a decrease in the production of reactive oxygen intermediates by macrophages mediated by PD-1-bearing CD4⁺ and CD8⁺ T cells allows increased intracellular pathogen load in leishmaniasis. Blocking of the PD-1 pathway can increase the ability of responding phagocytic cells to kill the parasites (38). Interestingly, supplementation with recombinant IL-12 in IRF1^−/− mice greatly enhanced their survival following infection with Brucella (39). Thus, therapeutic enhancement of the Brucella-specific CD8⁺ host response after vaccination or during chronic infection may be possible by targeting exhausted immune cells.

Determination of whether detectable exhausted CD8⁺ T cells are persistent, emergent, or dying after virulent challenge will be important to direct proper therapeutic intervention for vaccine enhancement. In this study, exhausted CD8⁺ T cells were adoptively transferred from chronically infected mice to recipient mice because they were found to be significantly enriched in pretransfer samples (Fig. 1). When the transfer was not followed by challenge, these exhausted cells were undetectable at day 5 posttransfer but were then readily identified at 15 days posttransfer. If these cells persisted in recipient mice after transfer, they may have trafficked to the spleen more slowly in the absence of acute infection than in the chronic recipient group that was challenged with B. melitensis. Alternatively, these exhausted cells present at 15 days posttransfer in the chronic recipient NC group may be generated from nonexhausted or preexhausted CD8⁺ T cells that were adoptively transferred from chronically infected mice. This suggests that when placed into a fresh environment lacking chronic brucellosis-driven stimulation, these “preexhausted” CD8⁺ T cells may then have suffered too much antigenic insult upon challenge to transition to effectors and become rapidly exhausted. It is important to note that because PD-1 and LAG-3 are intimately involved in survival, proliferation, and cell cycle progression, the explanation for the loss of exhausted cells in nonchallenged recipients 5 days after transfer remains to be determined (40, 41). Having now confirmed the presence of a considerable population of exhausted cells in mice with chronic brucellosis, future experiments will address the specific fate of these exhausted cells by magnetic isolation prior to transfer.

In addition to exhaustion as a phenotype of CD8⁺ T cell dysfunction, there are other mechanisms that could be a factor in the lack of a significant contribution of CD8⁺ T cells to the control of brucellosis. Regulatory T cells (Tregs) (42–45), anergy or adaptive tolerance resulting from in vivo hypostimulation (46), and immune senescence (59, 60) and ignorance (25) can lead to an inadequate adaptive response. Furthering these experiments to include older mice as acutely infected animals may give valuable insight into immunosenescence and its role in brucellosis. Also, it is unknown whether certain Brucella-specific T cell clones are more prone to deletion upon the resolution of acute disease leading to a deficit in the secondary response. A crucial next experimental step is to determine the functionality of the specific responding CD8⁺ T cells by optimizing a Brucella-specific MHC class I tetramer (NIH Tetramer Core Facility) using previously defined epitopes (13). Furthermore, congenic mice (i.e., H-2K^b Thy1.1 or Ly5.1 mice) could be used to track adoptively transferred cell populations. Currently, only 2 Brucella melitensis epitopes have been defined, and they are both on the H-2K^d MHC background, which precludes the use of congenic mouse models for tracking adoptively transferred cells.

We also report here that numbers of Brucella-responding IFN-γ-producing CD8⁺ T cells increase in response to infection, as expected, but these cells produce less IFN-γ per cell than do those from mock-immunized mice. This stifled IFN-γ production appears to be actively maintained by infection, as removal of cells to a new environment by adoptive transfer and not challenging cells with infection allow the recovery of per-cell IFN-γ production (Fig. 6C and D). This cellular defect in IFN-γ production may begin to account for the fact that IFN-γ from CD8⁺ T cells does not significantly contribute to the host control of brucellosis (28). The stifled IFN-γ production of Brucella-specific CD8⁺ T cells coupled with the cohort of IFN-γ-negative exhausted CD8⁺ T cells seen in chronic infection may significantly contribute to the persistence and reactivation of brucellosis. Determining the mechanism of this decreased IFN-γ production will be of interest for subsequent studies. Similarly, the rescue of exhausted cells to therapeutically address the limited amount of CD8⁺ T cell-derived IFN-γ during brucellosis is also of interest. Supporting this concept, the addition of recombinant IL-2 to Brucella-vaccinated cattle enhanced IFN-γ production and protection from abortion (47).

The initially higher bacterial load in chronic recipients could be due to dysfunctional CD8⁺ T memory cells but also suggests a lack of compensation by memory CD4⁺ T cells and memory B cells early in the challenge response (48–51). In this study, we focused on the response of splenocytes, and therefore, a greater number of responding cells would be anticipated to appear at the foci of splenic infection following acute infection and challenge than in the spleens of chronically infected mice that may have extrasplenic foci of brucellosis (7, 52). However, the decreased numbers of CFU later during the new challenge and expansion of T memory cells suggest that upon removal from the Brucella-stimulated environment of chronically infected hosts and placement into a fresh, naive host, CD8⁺ T cells respond to challenge with renewed vigor.

Consistent with the decreased number of CFU, transferred cells mounted a vigorous IFN-γ response (Fig. 6B), in which both acute and chronic donors had more IFN-γ-producing cells than did naive infected mice. This increased amount of IFN-γ compared to that in naive mice is suggestive of recruitment of previously activated or memory CD8⁺ T cells. Notably, the responsiveness of chronic donors in a new environment contrasts with the lower percentages of IFN-γ-producing cells directly sampled from chronically infected mice (Fig. 2A). Because Brucella-educated cells were removed from the antigen-specific immune environment of the donor mice, we suggest that this delay in the cytokine response to virulent challenge is a cell-intrinsic defect of CD8⁺ T cells from previously infected animals. In addition to the cell-intrinsic components of constitutively expressed inhibitory receptors on exhausted cells and the defect in IFN-γ production discussed above, this suggests a CD8⁺ T cell-extrinsic environmental component of suppression that exists in animals with chronic brucellosis. Furthermore, removal of this environmental suppression could enhance immune control of brucellosis.

The extracellular immune environment can play a pivotal role in determining the functionality of responding pathogen-specific cells. Components of the immune environment that can suppress or negatively regulate CD8⁺ T cells can include naturally occurring or induced CD4⁺ regulatory T cells. Depletion of regulatory T cells has been found to contribute to the reduction of viral load and reactivation of exhausted CD8⁺ T cells in chronic retroviral infection. Additionally, IL-10 can act on CD8⁺ T cells to induce a regulatory phenotype (53). IL-10 and transforming growth factor β (TGF-β) expressed by Tregs may mediate a generally immune-suppressive environment as well (54). Recent work has just begun to address the role of regulatory T cells and a suppressive immune environment in brucellosis, showing that depletion of IL-10 in vivo can enhance B. abortus clearance in mice (55). Also, Tregs have been found to suppress B. abortus-specific CD4⁺ T cells (56). Future experiments will include the transfer of Brucella-educated cells into naive as well as nonnaive environments for a direct comparison of responses in the investigation of the immune-suppressive environment during acute and chronic brucellosis. More work is needed to determine the effect of the suppressive environment on Brucella-specific CD8⁺ T cells in order to therapeutically exploit this suppression, possibly turning the tables, favoring a stronger CD8⁺ T cell response.

Previous studies presented evidence of CD8⁺ T cells as either pivotal or disposable players in the response to Brucella (13, 28, 57, 58). Our work presented here supports the hypothesis that while CD8⁺ T cells may be deficient in the response to chronic brucellosis, if their function can be therapeutically rescued, they may be the turning point on the road to an anti-Brucella vaccine capable of long-term protection. By either blocking inhibitory receptors such as PD-1 and LAG-3, reducing Treg suppression, or other yet-to-be-described therapeutic interventions, eliciting a more robust Brucella-specific CD8⁺ T cell population via vaccination may very well be possible.