Precursor Abundance Influences Divergent Antigen-Specific CD8⁺ T Cell Responses after Yersinia pseudotuberculosis Foodborne Infection

ABSTRACT

Primary infection of C57BL/6 mice with the bacterial pathogen Yersinia pseudotuberculosis elicits an unusually large H-2K^b-restricted CD8⁺ T cell response to the endogenous and protective bacterial epitope YopE_69–77. To better understand the basis for this large response, the model OVA_257–264 epitope was inserted into YopE in Y. pseudotuberculosis and antigen-specific CD8⁺ T cells in mice were characterized after foodborne infection with the resulting strain. The epitope YopE_69–77 elicited significantly larger CD8⁺ T cell populations in the small intestine, mesenteric lymph nodes (MLNs), spleen, and liver between 7 and 30 days postinfection, despite residing in the same protein and having an affinity for H-2K^b similar to that of OVA_257–264. YopE-specific CD8⁺ T cell precursors were ∼4.6 times as abundant as OVA-specific precursors in the MLNs, spleens, and other lymph nodes of naive mice, explaining the dominance of YopE_69–77 over OVA_257–264 at early infection times. However, other factors contributed to this dominance, as the ratio of YopE-specific to OVA-specific CD8⁺ T cells increased between 7 and 30 days postinfection. We also compared the YopE-specific and OVA-specific CD8⁺ T cells generated during infection for effector and memory phenotypes. Significantly higher percentages of YopE-specific cells were characterized as short-lived effectors, while higher percentages of OVA-specific cells were memory precursor effectors at day 30 postinfection in spleen and liver. Our results suggest that a large precursor number contributes to the dominance and effector and memory functions of CD8⁺ T cells generated in response to the protective YopE_69–77 epitope during Y. pseudotuberculosis infection of C57BL/6 mice.

INTRODUCTION

CD8⁺ T cells are vital for defense against intracellular pathogens and malignant diseases. They are responsible for eliminating altered cells—those infected with intracellular microbes and tumor cells. Antigen-specific CD8⁺ T cells recognize peptide antigens presented on the major histocompatibility class I (MHC-I) molecules on the surfaces of these altered cells. Activated CD8⁺ cells then kill target cells through secretion of cytokines or expression of effector molecules (1). Identification of factors regulating the antigen-specific CD8⁺ T cell response is critical for vaccine design to combat infectious diseases. In addition, bacteria have recently been recognized as a potential vaccine vector to stimulate CD8⁺ T cell response to treat cancer (2). The CD8⁺ T cell response to bacteria is most thoroughly studied using the Gram-positive pathogen Listeria monocytogenes (3). The primary CD8⁺ T cell response to L. monocytogenes can be divided into four phases—activation, expansion, contraction, and memory (3, 4). Following L. monocytogenes infection, antigen-presenting cells (APCs), most notably Batf3-dependent CD8α⁺ dendritic cells (DCs), are required to acquire antigens and present them to specific CD8⁺ T cell precursors in the T cell zone in the secondary lymphoid organs (5, 6). Antigen-specific precursors are very rare, yet the process of antigen presentation to the precursors is surprisingly fast and efficient (7). Activated antigen-specific CD8⁺ T cells then undergo rapid expansion, a process requiring costimulatory factors and inflammatory cytokines, including interleukin 2 (IL-2) and IL-12 (8, 9). Depending on the nature of the inflammatory environment, one activated precursor CD8⁺ T cell can give rise to more than 10,000 daughter cells over the next 5 to 8 days (10–14). Such rapid proliferation was estimated to be near the maximal possible division speed for mammalian cells. This expansion phase ensures the fastest accumulation of CD8⁺ T cells to accomplish their tasks (1). Coupled to this rapid expansion phase, the daughter cells differentiate into either short-lived effector cells or memory precursor effector cells. Local environments, including inflammatory and antigenic stimuli, play an important role in deciding the differentiation path of the daughter cells (14–16). For acute infection, after the rapid proliferation stage and regardless of whether the T cell response is successful in eliminating the pathogen, more than 90 to 95% of the expanded cells undergo apoptosis to result in contraction of the population (10, 17). Inflammation and especially the inflammatory cytokines interferon γ (IFN-γ) and IL-12 influence the contraction phase (4). In addition, IL-7, IL-15, and transforming growth factor β (TGF-β) have all been implicated in the process (3). In the end, the remaining long-lived memory population provides faster and enhanced protection for secondary antigenic challenge (3, 4). Among the memory T cells generated, the tissue-resident memory (T_RM) cells provide immediate local protection against subsequent infections (15).

The Gram-negative bacterial pathogen Yersinia pseudotuberculosis has become a new model system to study CD8⁺ T cell responses to bacteria. As an enteric zoonotic pathogen, Y. pseudotuberculosis is most commonly associated with mesenteric lymphadenitis in humans (18). Experimental infection of mice through the oral route with Y. pseudotuberculosis results in bacterial colonization in Peyer’s patches, the lamina propria, and mesenteric lymph nodes (MLNs). Bacteria also disseminate to the spleen and liver (19, 20). Previously, we showed that sublethal infection of C57BL/6 mice with Y. pseudotuberculosis resulted in a large H-2K^b-restricted CD8⁺ T cell response to YopE_69–77, which is a protective epitope found in one of the major virulence factors of Yersinia (21–26). Y. pseudotuberculosis is an extracellular pathogen, and most of the bacteria have been found extracellularly in vivo during infection, though a small proportion of intracellular bacteria do survive at least temporarily (27–29). This extracellular localization of the bacteria is due to the effects of several antiphagocytic proteins of Yersinia, including YopE. These proteins are translocated into the cytosol of host cells through the bacterial type III secretion system (T3SS) (30). The T3SS is utilized by several Gram-negative pathogens to promote virulence (31). In Yersinia, activation of the T3SS requires close contact between a bacterium and the host cell (32). Evidence suggests that YopE-specific CD8⁺ T cells participate in host protection against Yersinia by the production of tumor necrosis factor alpha (TNF-α) and IFN-γ (22). Alternatively, it was suggested that CD8⁺ T cells use cytotoxic-T-lymphocyte (CTL) activity to kill host cells attached to Yersinia organisms, followed by engulfment and destruction of both the host cell and the attached bacteria by neighboring phagocytes (33). The dominant YopE-specific CD8⁺ T cell response requires CCR2⁺ inflammatory monocytes, which are recruited to foci of Y. pseudotuberculosis infection in lymphoid tissues (25, 26). These monocytes, and presumably their differentiated products, such as inflammatory dendritic cells (DCs), are injected with YopE and are required to function as APCs to directly prime YopE-specific CD8⁺ T cells (25, 26).

To further characterize the CD8⁺ T cell response to Y. pseudotuberculosis, the model CD8⁺ T cell epitope in chicken ovalbumin (OVA_257–264) was chosen to take advantage of existing tools because the OVA antigen has a predicted affinity for H-2K^b similar to that of YopE_69–77 (24). We generated a Y. pseudotuberculosis strain in which the OVA_257–264 sequence was inserted within a linker region of the native YopE protein, ensuring that both epitopes were equally produced during infection. Following foodborne infection of C57BL/6 mice with this strain, we unexpectedly found that the YopE-specific CD8⁺ T cell response was several times higher than that of the OVA-specific response in all tissues analyzed and at different days postinfection. Furthermore, in naive mice, YopE-specific CD8⁺ T cell precursors were ∼4.6 times as abundant as the OVA-specific cells. In addition, YopE-specific and OVA-specific CD8⁺ T cells displayed differences in effector and memory formation. Higher percentages of YopE-specific cells were characterized as short-lived effector cells while higher percentages of OVA-specific cells were memory precursor effector cells. Together, these results support the proposition that the precursor frequency of antigen-specific CD8⁺ T cells contributes to dominance during the primary immune response and may impact effector and memory functions in foodborne Y. pseudotuberculosis infection.

RESULTS

Characterization of 32777-OVA for tissue colonization after foodborne infection.

Previously, Bergsbaken and Bevan measured CD8⁺ T cell responses to OVA_257–264 and YopE_69–77 in mice that were infected with a Y. pseudotuberculosis strain that expressed both native YopE and a YopE-ovalbumin fusion protein under the control of a heterologous promoter (34). In that case, 4 to 5 times more YopE-specific than OVA-specific CD8⁺ T cells were produced in various tissues at the peak (day 9) and memory (day 45) phases of the response (34). Bergsbaken and Bevan reasoned that the higher antigen load of YopE_69–77 compared to OVA_257–264 resulted in this difference in the antigen-specific CD8⁺ T cell response, since ovalbumin was produced at lower levels than YopE (34). Here, to ensure equal antigen load, Y. pseudotuberculosis strain 32777-OVA was generated by inserting the SIINFEKL (OVA_257–264) epitope into a linker region of the YopE protein between the chaperone-binding domain and the GAP domain (Fig. 1A). This placed OVA_257–264 adjacent to YopE_69–77, which was predicted to possess a comparable affinity for MHC-I H-2K^b (50% inhibitory concentrations [IC₅₀] of 17 nM and 20 nM, respectively) (24). C57BL/6 mice were infected by the foodborne route with a sublethal dose of 32777-OVA to characterize tissue colonization. From 4 to 14 days postinfection (dpi), MLNs, livers, and spleens were found to be colonized at variable levels. In general, the mean colonization levels decreased gradually between 7 and 14 dpi and were cleared at 30 dpi (Fig. 1B to D). Similar to what we observed before for the parental wild-type strain 32777 (35), colonization by 32777-OVA ranged greatly from individual to individual, regardless of the tissues analyzed. For example, in spleens at 9 dpi, even from the same experiment, colonization ranged from below the limit of detection to a log of 8 (Fig. 1D). At 14 dpi, when the majority of the animals cleared bacteria from this site, two mice harbored moderate CFU levels in spleens (Fig. 1D). It is possible that insertion of the OVA_257–264 sequence in YopE resulted in attenuation to some degree; however, we reasoned that tissue colonization by 32777-OVA was sufficient to elicit OVA-specific and YopE-specific CD8⁺ T cells after foodborne infection (see below).

FIG 1.

Structure of YopE protein in 32777-OVA and tissue colonization levels of 32777-OVA following foodborne infection. (A) Domain structure of YopE protein, with the signal sequence (Ss) for its secretion through the type 3 secretion system, the chaperone-binding domain (Cb), and the RhoGAP domain indicated. The arginine residue critical for the RhoGAP function is also indicated. The sequence and position of YopE_69–77 and the OVA_257–264 epitope inserted after residue 85 are indicated at the bottom. (B to D) C57BL/6 mice were infected with 5 × 10⁷ CFU of 32777-OVA orally, and the colonization levels were determined by CFU assay at the indicated days postinfection (dpi) from MLNs (B), livers (C), and spleens (D). Each point represents the value obtained from one mouse, and the results shown are pooled from 2 to 6 independent experiments; n = 4 for 4 and 7 dpi, 18 for 9 dpi, 7 for MLNs at 14 dpi, and 8 for the rest at 14 dpi and 30 dpi. The dotted line indicates the limit of detection. Means are indicated with bars.

Higher levels of YopE-specific than OVA-specific CD8⁺ T cells after 32777-OVA infection.

With 32777-OVA foodborne infection, YopE-specific CD8⁺ T cells could be detected above background levels using tetramers and flow cytometry starting at 6 dpi (data not shown), and by 7 dpi, a sizable population was evident in all the tissues analyzed—MLN, liver, and spleen (Fig. 2A). Throughout the time course of the analysis from 7 to 30 dpi, because bacterial colonization impacts inflammatory cytokines and correlates with antigen load, and there was a wide range in bacterial colonization levels from mouse to mouse, the levels of YopE-tetramer positive cells varied widely among the individual animals analyzed (Fig. 2B to E). In general, on average the percentages of YopE-specific CD8⁺ T cells in each tissue increased between 7 and 9 dpi and then remained high over the remaining time course. In different tissues, the number of YopE-specific CD8⁺ T cells peaked at different days postinfection, with spleens harboring the largest numbers of these cells while livers contained the highest percentage (Fig. 2C to E). Unexpectedly, in all tissues analyzed at all the time points, a smaller percentage of OVA-specific CD8⁺ T cells were identified (Fig. 2A to E). Furthermore, the percentage of these cells increased at a smaller magnitude from 7 to 9 dpi. As a result, the ratios of the percentages of YopE-specific/OVA-specific CD8⁺ T cells in MLNs, livers, and spleens increased gradually from 7 to 30 dpi, and in all these tissues, the differences between 7 dpi and 30 dpi were significant (Fig. 2F to H). In conclusion, in all tissues analyzed at all times, there were several times as many YopE-specific CD8⁺ T cells as OVA-specific cells, and the difference between the two populations increased from 7 to 30 dpi.

FIG 2.

Comparison of the levels of OVA-specific and YopE-specific CD8⁺ T cell over time after infection with 32777-OVA. C57BL/6 mice were orally infected as for Fig. 1 with 32777-OVA. At 7, 9, 14, or 30 dpi, single-cell suspensions from indicated tissues were prepared and analyzed by flow cytometry after staining with tetramers and antibodies. (A) Representative contour plots of gated live CD45.2⁺ CD3⁺ CD8⁺ T cells at 7 dpi from MLNs (top), livers (middle), and spleens (bottom) further analyzed for CD44 and OVA-specific (left) or YopE-specific (right) signals. (B) Representative contour plots of live CD45⁺ TCR-β⁺ CD8α⁺ CD44^hi cells from the indicated tissues at 9, 14, or 30 dpi analyzed for OVA and YopE tetramer expression. The percentages (top) and number (bottom) of live tetramer-positive CD8⁺ T cells from MLNs (C), livers (D), and spleens (E) of mice infected for the indicated days. Each point represents the value obtained for one mouse, and the results shown are pooled from 2 independent experiments. For 7 dpi, n = 4; 9 dpi, n = 7; 14 dpi, n = 8; 30 dpi, n = 6. Means and standard deviations are shown. P values were determined with two-way analysis of variance (ANOVA; mixed-effects model) followed by Sidak’s multiple-comparison test. (F to H) Ratio of the percentages of YopE-specific CD8⁺ cells to those of OVA-specific ones at the indicated dpi. Samples in which no OVA-specific cells were detected or the percentage of OVA-specific cells was less than 0.01% were removed from analysis. (F) Ratio of YopE to OVA from MLNs. n = 4, 16, 8, and 3, respectively, for 7, 9, 14, and 30 dpi. (G) Ratio of YopE to OVA from livers. n = 4, 12, 8, and 6, respectively, for 7, 9, 14, and 30 dpi. (H) Ratio of YopE to OVA from spleens. n = 4, 17, 8, and 6, respectively, for 7, 9, 14, and 30 dpi. Results shown are summarized from 2 to 5 experiments. P values were determined with a Brown-Forsythe ANOVA followed by Dunnett’s T3 multiple-comparison test. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Higher levels of YopE-specific than OVA-specific CD8⁺ T cell precursors before 32777-OVA infection.

The number of endogenous CD8⁺ T cell precursors of distinct specificities is known to be different, and the precursor frequency has been indicated to influence the size and progress of both the initial and the memory immune response (36). We consistently detected higher numbers of YopE-specific compared to OVA-specific CD8⁺ T cells at the earliest infection times, raising the possibility that there is a difference in precursor numbers. Therefore, the numbers of YopE- and OVA-specific CD8⁺ T cell precursors in C57BL/6 mice was determined (Fig. 3A to C). Cells from the spleen, MLNs, and other macroscopically identifiable lymph nodes, including inguinal, axillary, brachial, submandibular, cervical, and para-aortic nodes, were pooled from individual naive mice, and precursors were enumerated using the established protocol of peptide–MHC-I (pMHC) tetramer staining in conjunction with magnetic-bead separation (36). The average number of OVA-specific CD8⁺ T cells per mouse was determined to be 302 (Fig. 3A and C), which is higher than what was measured initially, ∼130 (36), but is in range with other measurements reported later on (37). In contrast, the average number of YopE-specific CD8⁺ T cells was determined to be 1,387 per mouse, or 4.6 times the number measured for OVA-specific cells (Fig. 3B and C). Thus, there are significantly more YopE-specific CD8⁺ T precursors than OVA-specific cells in C57BL/6 mice, which likely contributes to the dominance of the former.

FIG 3.

Precursor determination of OVA- and YopE-specific CD8⁺ T cells in naive C57BL/6 mice. Representative dot plots of the enriched OVA (A) or YopE (B) tetramer-positive cells as indicated. (C) Absolute numbers of the naive antigen-specific CD8⁺ T cells from total spleens and lymph nodes of individual naive mice. Each symbol represents the value obtained from an individual mouse (n = 4), and means and standard deviations are indicated. P values were determined by a Mann-Whitney test. *, P < 0.05.

Characterization of YopE-specific and OVA-specific CD8⁺ T cell effector phenotypes generated during 32777-OVA infection.

We next determined if the dominance of YopE-specific over OVA-specific CD8⁺ T cells impacted their effector functions. Initially, the effector phenotypes of YopE- and OVA-specific CD8⁺ T cells were evaluated through accessing the expression levels of the terminal effector marker KLRG1 at 9 dpi. As expected, the levels of KLRG1^high cells varied among the tissues analyzed, and in general, blood contained the highest levels of KLRG1^high antigen-specific CD8⁺ cells and the percentages of KLRG1^high antigen-specific cells were lower in MLNs, liver, and spleen (Fig. 4). Curiously, the YopE-specific CD8⁺ T cells contained lower levels of KLRG^high cells than the OVA-specific ones. In blood, an average of 83% of OVA-specific CD8⁺ T cells were KLRG^high, while in contrast, only an average of 59% of YopE-specific ones were KLRG^high (Fig. 4A, top, and Fig. 4B, first two columns). Significantly lower levels of KLRG^high YopE-specific CD8⁺ T cells were also observed in the spleen and liver (Fig. 4A and B). The levels of KLRG^high YopE-specific CD8⁺ T cells also trended lower in MLNs, but the difference was not significant (P = 0.06) (Fig. 4B). Overall, the difference in the levels of KLRG1^high cells suggests a potential difference in the differentiation of YopE-specific and OVA-specific CD8⁺ T cells at 9 dpi.

FIG 4.

The percentages of OVA-specific or YopE-specific CD8⁺ T cells that were positive for KLRG1 are different at 9 dpi following 32777-OVA infection. C57BL/6 mice were orally infected as described for Fig. 1. Percentages of tetramer positive CD8⁺ T cells from the indicated tissues at 9 dpi that are positive for terminal differentiation marker KLRG1 were determined by flow cytometry analysis. (A) Representative contour plots of CD8⁺ tetramer-positive cells from the indicated tissues were further analyzed for CD44 and KLRG1 signals. (B) The percentages of KLRG1^high cells among OVA (filled circles)- and YopE (open circles)-specific cells were plotted according to tissue source. Each point represents the value obtained from one mouse, and values from the same mouse are connected. The results shown are pooled from 2 to 4 independent experiments. n = 7 for blood, 10 for MLNs, 6 for livers, and 10 for spleens. P values were determined with a ratio paired t test. *, P < 0.05; ***, P < 0.001.

Next, these antigen-specific CD8⁺ T cells at 9 dpi were assessed by measuring the percentages that produced IFN-γ or TNF-α in response to peptide stimulation. Aliquots of cells from MLNs or spleens of mice infected with 32777-OVA 9 days previously were stimulated ex vivo for 4.5 h with either YopE or OVA peptides or left nonstimulated (NS) in the presence of BFA, followed by intracellular cytokine staining and flow cytometry analysis (Fig. 5A). As expected from the dominant presence of YopE-specific CD8⁺ T cells, in comparison to cells treated with OVA peptide or NS cells, a higher percentage of TNF-α-positive CD8⁺ T cells was detected after stimulation with YopE peptide in both MLN and spleen samples (Fig. 5B, left). Higher percentages of IFN-γ-positive CD8⁺ T cells were also observed in spleen samples after treatment with the YopE peptide (Fig. 5C, left). To better compare the percentage of cytokine-producing cells following peptide stimulation, we normalized the value by taking into consideration the background as measured in the corresponding NS sample and the percentage of tetramer-positive cells. Therefore, the normalized value of 1 indicates that the percentage of cytokine-positive cells and the percentage of tetramer-positive cells were the same, while higher values suggest an increased sensitivity of the intracellular cytokine staining over tetramers in detecting responding CD8⁺ T cells (Fig. 5B and C, right). The difference between the OVA- and the YopE-stimulated groups was not significant (Fig. 5B and C, right panels). The result indicates that in response to peptide stimulation, the number of CD8⁺ T cells producing either IFN-γ or TNF-α is proportional to the number of corresponding tetramer-positive CD8⁺ T cells present in the sample.

FIG 5.

Ex vivo production of IFN-γ and TNF-α after stimulation with OVA_257–264 and YopE_69–77 peptides. Mice were orally infected as for Fig. 1. At 9 dpi, total MLN and spleen cells were incubated with the peptide YopE_69–77 or OVA_257–264 or PBS (NS, nonstimulated). (A to C) Intracellular cytokine production from independent wells of cells. Aliquots of cells were incubated as indicated together with BFA for 4.5 h and then stained for indicated intracellular cytokines and surface markers before flow cytometry. (A) Representative histograms of CD8⁺ T cells. (B and C) Percentages of CD8⁺ T cells positive for TNF-α (B) or IFN-γ (C) (left) and the values after normalization using the formula (% cytokine_antigen − % cytokine_NS)/% CD8_antigen (right), where % cytokine_antigen and % cytokine_NS refer to the percentages of TNF-α- and IFN-γ-positive cells in CD8⁺ T cells following treatment with either OVA, YopE peptide, or PBS and % CD8_antigen is the percentage of OVA or YopE tetramer-positive cells in CD8⁺ T cells in the sample. (D) Concentrations of secreted IFN-γ in the medium of cells incubated with peptide or PBS alone for 48 h without BFA were determined with ELISA and reported as absolute values (left) or after normalization using the formula (C_antigen − C_NS)/number_antigen (right), where C_antigen and C_NS refer to the concentration of IFN-γ in femtograms per milliliter following OVA or YopE peptide stimulation or NS and number_antigen refers to the number of antigen-specific CD8⁺ T cells in the particular sample (right). Means and standard deviations are indicated. Data shown are combined from two independent experiments, and each point represents the value obtained from one mouse (n = 4). P values were determined by repeated-measures ANOVA, followed by Sidak’s multiple-comparison test. *, P < 0.05; ***, P < 0.001; ****, P < 0.0001.

To evaluate the abilities of the two types of antigen-specific CD8⁺ T cells to secrete IFN-γ, aliquots of cells from MLN or spleen of mice infected with 32777-OVA 9 days previously were stimulated ex vivo with either YopE or OVA peptides for 48 h or left unstimulated. The concentrations of IFN-γ in the medium were then determined. Higher concentrations of IFN-γ were observed following stimulation with YopE peptide than with the OVA peptide, although the differences were not significant. The difference between YopE-stimulated and NS control splenocytes was statistically significant (Fig. 5D, left). To compare the cytokine secretion efficiency of an average antigen-specific cell, we normalized the concentrations of IFN-γ to the number of tetramer-positive cells in the sample, which was calculated using results of flow cytometry carried out with an additional aliquot of cells. When background IFN-γ concentrations were subtracted, however, two negative values resulted with MLN samples, since the IFN-γ concentrations were higher in NS than in OVA-stimulated cells. These two values were not displayed in the log scale in the figure (Fig. 5D, right, first column). After normalization, the values obtained from YopE peptide stimulation compared to OVA peptide were significantly higher in splenocytes but not MLN cells (Fig. 5D, right, compare the last two columns). Overall, these results indicate that at 9-dpi YopE-specific and OVA-specific CD8⁺ T cells in spleen and MLN have similar capacities to produce IFN-γ or TNF-α in response to peptide stimulation ex vivo, although YopE-specific CD8⁺ T cells in spleen may have a subtle advantage in secretion of IFN-γ under these conditions.

These results prompted us to further evaluate a potential difference in the effector differentiation of these two populations of antigen-specific cells over a longer time course. Following L. monocytogenes infection, through differential expression of CD127 and KLRG1, antigen-specific CD8⁺ T cells can be divided into short-lived effector cells (SLECs; CD127^neg KLRG1⁺), double-positive effector cells (DPECs; CD127⁺ KLRG1⁺), memory precursor effector cells (MPECs; CD127⁺ KLRG1^neg), and early effector cells (EECs; CD127^neg KLRG1^neg) (3). SLECs undergo apoptosis during T cell contraction, while MPECs form long-lived memory cells. A similar process of memory formation has yet to be established following Yersinia infection, but a distinction in SLEC, DPEC, or MPEC populations among OVA- and YopE-specific cells would be consistent with a difference between these two populations of antigen-specific CD8⁺ T cells in differentiation. In either livers or spleens, sufficient numbers of OVA-specific cells were identified that warranted such analysis. For both OVA- and YopE-specific CD8⁺ T cells, at 9 dpi, the percentages of all four effector populations displayed a wide range (Fig. 6). On average, EECs were the largest population at this time point for both tissues (Fig. 6B and C). In spleens, consistent with a later peak time (Fig. 2E, bottom), the YopE-specific CD8⁺ T cells contained higher percentages of EECs than OVA-specific cells (Fig. 6C). With progress of infection, the ranges in the percentages of these effector populations decreased (Fig. 6B and C). Higher levels of SLECs and lower levels of MPECs were identified from YopE-specific than OVA-specific CD8⁺ T cells, and the differences were significant in both livers and spleens at 14 and 30 dpi (Fig. 6B and C). These observations are consistent with a potential difference in memory formation of YopE- and OVA-specific CD8⁺ T cells.

FIG 6.

OVA- and YopE-specific CD8⁺ T cells differ in the composition of SLEC and MPEC populations defined by CD127 and KLRG1 expression following infection with 32777-OVA. Expression of CD127 and KLRG1 in live tetramer-positive CD45⁺ TCRβ⁺ CD8α⁺ CD44^hi T cells was determined at 9, 14, and 30 dpi. (A) Representative contour plots of OVA-specific (top) and YopE-specific (bottom) CD8⁺ T cells from livers (left) and spleens (right) at the indicated days postinfection. The percentages of SLECs (CD127^neg KLRG1⁺), DPECs (CD127⁺ KLRG1⁺), MPECs (CD127⁺ KLRG1^neg), and EECs (CD127^neg KLRG1^neg) in OVA-specific and YopE-specific cells in livers (B) and spleens (C) were plotted according to dpi. Values obtained from the same animal are connected. Data shown are summarized from two experiments for 9 and 30 dpi and one for 14 dpi. For livers, n = 8, 4, and 6 for 9, 14, and 30 dpi, respectively; for spleens, n = 6, 4, and 6 for 9, 14, and 30 dpi, respectively. P values were determined with repeated-measures ANOVA followed by Sidak’s multiple-comparison test. *, P < 0.05; **, P < 0.01.

Higher levels of intestinal YopE-specific than OVA-specific CD8⁺ T cells following 32777-OVA infection.

To evaluate the antigen-specific CD8⁺ T cell response in intestinal tissue, lymphocytes were isolated and analyzed by flow cytometry after staining. Similar to MLNs, livers, and spleens, a wide range was observed in the levels of YopE-specific CD8⁺ T cells among intestinal intraepithelial lymphocytes (IELs) or the lymphocytes from lamina propria (LP) at 9 dpi (Fig. 7A to C). From 7 dpi to 30 dpi, the mean percentage of CD8⁺ T cells positive for the YopE tetramer in IELs remained relatively stable, while the values for OVA-specific cells decreased over this time period (Fig. 7B). Similar results were observed in the LP, although the percentages of YopE-specific CD8⁺ T cells peaked at 14 dpi (Fig. 7C). Additionally, the ratio of YopE-specific cells to OVA-specific CD8⁺ T cells also increased from 9 dpi to 30 dpi, a difference that was significant in the LP (Fig. 7D and E).

FIG 7.

Different levels of OVA-specific and YopE-specific CD8⁺ T cells in IELs and the LP. Tetramer-positive cells were identified from IELs and the LP using flow cytometry analysis following infection. (A) Representative contour plots of live CD45⁺ TCR-β⁺ CD8α⁺ CD44^hi cells from IELs (top) and the LP (bottom) for tetramer signals at the indicated dpi. (B and C) The percentages of OVA tetramer-positive and YopE tetramer-positive CD8⁺ cells (CD45⁺ TCR-β⁺ CD8α⁺) from either IELs (B) or the LP (C) were plotted according to dpi. P values were determined with two-way ANOVA (mixed-effects model) followed by Sidak’s multiple-comparison test. *, P < 0.05; **, P < 0.01; ***, P < 0.001. (D and E) Ratio of YopE-specific CD8⁺ T cells to OVA-specific ones from either IELs (D) or the LP (E), according to dpi. P values were determined with a Brown-Forsythe ANOVA test followed by Dunnett’s T3 multiple-comparison test. *, P < 0.05. Data shown are summarized from one (14 dpi) or two (remainder) experiments. n = 6, 4, and 6 for 9, 14, and 30 dpi, respectively.

To evaluate memory formation, the antigen-specific CD8⁺ T cells were quantified for CD103 and CD69 expression. Bacterial or viral infection results in CD8⁺ T cell effector activation and trafficking into peripheral tissues, such as skin, lung, or intestine, to differentiate into T_RM cells to provide accelerated pathogen clearance upon reencounter (38). Following oral Y. pseudotuberculosis infection, two populations of T_RM cells have been identified—the classical CD69⁺ CD103⁺ T_RM cells and the CD69⁺ CD103^neg cells found in the LP (34, 39). Among IELs, the classical CD69⁺ CD103⁺ T_RM cells were identified from both OVA-specific and YopE-specific CD8⁺ T cells (Fig. 8A and B). As reported before (34), the percentages of CD69⁺ CD103⁺ T_RM cells increased from 9 to 30 dpi among both OVA-specific and YopE-specific CD8⁺ T cells (Fig. 8B, left). Lower percentages of YopE-specific than OVA-specific CD8⁺ T cells were observed in these CD69⁺ CD103⁺ T_RM cell populations throughout the time course analyzed, and this difference was significant at 30 dpi (Fig. 8B, left). In the LP, although both CD69⁺ CD103⁺ and CD69⁺ CD103^neg T_RM cells were identified, their percentages were variable among the two populations of antigen-specific CD8⁺ T cells over time (Fig. 8A and B). In the LP, there was a significantly lower percentage of YopE-specific than OVA-specific CD8⁺ T cells among the CD69⁺ CD103⁺ T_RM cell population at 9 dpi and the CD69⁺ CD103^neg T_RM cell population at 30 dpi (Fig. 8B, middle and right, respectively). Therefore, overall, these results show that there are higher numbers of intestinal YopE-specific than OVA-specific CD8⁺ T cells at 9 dpi following infection with 32777-OVA, and the former cell type dominance increases between 9 and 30 dpi. In addition, both types of antigen-specific cells can become T_RMs, although YopE-specific CD8⁺ T cells may acquire these phenotypes less efficiently during the time analyzed.

FIG 8.

T_RM formation among OVA-specific and YopE-specific CD8⁺ T cells identified from IELs and the LP following 32777-OVA infection. Tetramer-positive cells identified from IELs and the LP were analyzed for expression of CD103 and CD69 using flow cytometry analysis. (A) Representative contour plots of either OVA tetramer-positive (top) or YopE tetramer-positive (bottom) cells isolated from IELs (left) or the LP (right) for CD103 and CD69 signals at the indicated dpi. (B) The percentages of the indicated types of T_RM cells among OVA tetramer-positive and YopE tetramer-positive cells were plotted according to dpi. Data shown are summarized from one (14 dpi) or two (remainder) experiments. n = 6, 4, and 6 for 9, 14, and 30 dpi, respectively. P values were calculated with a ratio paired t test. *, P < 0.05; **, P < 0.01.

DISCUSSION

We set out to characterize the antigen-specific CD8⁺ T cell response in foodborne Y. pseudotuberculosis infection and expressed the model CD8⁺ T cell epitope OVA_257–264 in the same protein as the native Yersinia epitope YopE_69–77. It was initially a surprise that the YopE-specific CD8⁺ T cell response was several times higher than the OVA-specific response. Here, we showed that the YopE-specific CD8⁺ T cells have a high precursor frequency in naive C57BL/6 mice and that this high precursor frequency is associated with the large and relatively more sustained YopE-specific CD8⁺ T cell response than that of OVA-specific cells following Y. pseudotuberculosis infection. In addition, we provide evidence suggesting that these two types of antigen-specific CD8⁺ T cells may be different in additional aspects, including effector differentiation, the ability to secrete cytokines, and memory formation.

The number of T cell precursors specific for different antigens in an individual can vary greatly (reviewed in reference 37). Using tetramer-based cell enrichment, it was determined that the number of naive antigen-specific CD8⁺ T cells ranged from ∼15 to 1,500 cells per mouse for the 25 viral antigenic epitopes studied and the model antigen OVA_257–264 (37). Based on two independent studies, the number of CD8⁺ precursors specific for murine cytomegalovirus M45:D^b is the highest, either 1,500 cells/mouse or 603 cells/mouse, and 4.6 times as abundant as that of OVA-specific precursors (36, 40). Our results here indicated that the average number of YopE-specific CD8⁺ precursor is 1,387/mouse, or 4.6 times as abundant as that of OVA-specific ones (Fig. 3C). Even considering the differences in measurement conducted at different laboratories, this result indicates that the YopE-specific CD8⁺ T cell precursors are among the most abundant precursors known.

The magnitude of a primary T cell response is influenced by the amount and duration of the antigenic peptide presented on MHC molecules of APCs in secondary lymphoid organs. Recently, with the quantification of T precursors, the frequency of the naive precursors was also recognized as an important factor deciding the size of a primary response (37). Here, when the two epitopes of YopE_69–77 and OVA_257–264 were coexpressed in the same protein, theoretically they would be processed and presented equally. In addition, both types of antigen-specific cells are exposed to the same cytokine environment, which impacts T cell activation (4). Therefore, it was initially surprising when we first observed that the OVA-specific CD8⁺ T cell response was much smaller than the YopE-specific response (Fig. 2), especially since the OVA epitope has a slightly stronger calculated affinity for MHC I than the YopE epitope, with IC₅₀ of 17 nM and 20 nM, respectively (24). In this regard, the large number of YopE-specific CD8⁺ T cell precursors in naive C57BL/6 mice correlates well with our observation. Previously, Bergsbaken and Bevan had adoptively transferred 10⁴ OT-I T cells to mice before infection with Y. pseudotuberculosis expressing a YopE-ovalbumin fusion protein (34). This essentially boosted the OVA-specific CD8⁺ T cell “precursor” levels to more than that of native YopE-specific cells. However, the OT-I T cell response they observed was still several times smaller than the YopE-specific response following infection with the OVA-expressing Y. pseudotuberculosis strain. They reasoned that their YopE-ovalbumin protein was expressed from a weaker promoter, while the endogenous YopE was expressed at much higher levels. As a result, a smaller number of OVA peptides may have been presented on MHC I molecules. Consequently, the smaller OVA response was due to the smaller antigen load of OVA than YopE (34). Together, their study and ours presented here indicate that different factors, either the abundance of the antigenic peptides presented or the abundance of the precursors, could differentially impact the magnitude of the primary CD8⁺ T response to even the same kind of pathogen.

Besides the abundance of precursors, the affinity of T cell receptors (TCRs) for pMHC complexes has also been implicated to influence the primary virus-specific cytotoxic T cell response (reviewed in reference 41). The affinity of TCRs for pMHC may dictate the expansion of individual CD8⁺ T cell clones to result in the number of progenies varying over 3 orders of magnitude, as observed in individual T cell transfer experiments (14). Alternatively, the local environment impacts the differential expansion of T cells. Our results of an increased ratio of YopE- to OVA-specific CD8⁺ T cells with the progress of infection from 7 dpi to 30 dpi in multiple tissues, including intestines, are consistent with differential expansion of T cell clones (Fig. 2F to H and 7D and E); i.e., the shorter division time of the YopE-specific CD8⁺ T cells results in faster expansion. In MLNs, livers, and spleens, the YopE/OVA T cell ratios at 7 dpi are close to the YopE/OVA precursor ratios measured in naive mice. Then, through expansion of random combinations of both YopE- and OVA-specific T cell clones, the values of this ratio became variable, but on average, the ratio increased over time. The division speed of antigen-specific CD8⁺ T cells has been shown to impact central memory formation in that central memory precursors divide more slowly (16). This is consistent with our observation that a higher percentage of OVA-specific cells are MPECs at 14 and 30 dpi in both livers and spleens (Fig. 6). Furthermore, when T_RM cells were quantified from IELs or the lymphocytes from the LP, higher percentages of OVA-specific cells were found to bear T_RM markers at 30 dpi (Fig. 8B). In addition, our results suggested that an average YopE-specific CD8⁺ T cell effector may be more efficient in secretion of cytokines such as IFN-γ than an OVA-specific cell. Following ex vivo stimulation with either an OVA or a YopE antigenic peptide, more IFN-γ was secreted per responding cell from the spleen but not MLNs (Fig. 5D).

How could the lower frequency of the OVA-specific precursors be associated with general lower avidity of TCRs for pMHC, and therefore slower division speed during activation? It was shown recently that CD4⁺ T cells cross-reacting with self-peptides may be deleted during development and small naive T cell populations experience extensive clonal deletion (42). Furthermore, the T cells that survived clonal deletion express TCRs with lower affinity (42). A similar process has yet to be demonstrated for CD8⁺ T cells, although it is plausible that the cells from a smaller precursor population also express TCRs with lower affinities than those from a larger precursor population. Although speculative, the possibility that antigen-specific CD8⁺ T cells from a rare precursor population express TCR with lower affinity merits future investigation.

In summary, our finding that there is a high precursor frequency of YopE-specific CD8⁺ T cells partially explains the long-standing mystery behind the exceptional dominance of the protective YopE_69–77 epitope in C57BL/6 mice (21–26). The high precursor frequency of YopE-specific CD8⁺ T cells may contribute to the increased dominance of these cells over OVA-specific cells during the course of 32777-OVA infection, as wells as differences in effector and memory populations of these two cell types. These results have important implications for studying mechanisms of protection afforded by YopE-specific CD8⁺ T cells in Yersinia infection models with C57BL/6 mice. In addition, our findings suggest that 32777-OVA infection of C57BL/6 mice will provide a useful system for better understanding how precursor frequency impacts divergent CD8⁺ T cell responses in general.

MATERIALS AND METHODS

Bacterial strains.

The Y. pseudotuberculosis strains used in this study are serogroup O:1 strain 32777 and its derivative 32777-OVA. To generate strain 32777-OVA, the oligonucleotides 5′-CCGGCTAGCATAATCAACTTTGAAAAACTG-3′ and 5′-CCGGCAGTTTTTCAAAGTTGATTATGCTAG-3′ were synthesized and annealed at room temperature and then inserted into the AgeI site of plasmid pSB890-YopEplus (25). The resulting plasmid, pSB890-YopEOVA, was used in standard allelic exchange procedures to place the SIINFEKL epitope in the linker region of the YopE coding sequence after residue 85 of YopE (Fig. 1A) in the virulence plasmid, together with an NheI site to facilitate identification.

Infection of mice.

C57BL/6 mice were purchased from The Jackson Laboratory and were used within 8 to 16 weeks of age. For foodborne infection, overnight Y. pseudotuberculosis culture grown in Luria-Bertani (LB) medium at 28°C was washed once and resuspended in phosphate-buffered saline (PBS) to achieve the desired number of CFU per milliliter. To infect each mouse, a 50-μl volume of the suspension was applied to a single ∼0.5-cm³ piece of white bread placed on a thin layer of bedding in the bottom of a fresh cage. One mouse was introduced into the cage. The entire cage was kept in the dark and checked periodically until the bread was consumed, usually within 1 to 2 h. Then the mouse was returned to its original cage.

Processing of tissues.

At indicated days postinfection or when death was imminent, mice were euthanized by CO₂ asphyxiation. Blood was collected through heart puncture, and red blood cells were lysed. Mouse MLNs, spleens, and livers were dissected, cut into two pieces aseptically, and weighed. Half of spleens and MLNs were homogenized with a 3-ml syringe plunger in 5 ml of collection medium (RPMI 1640 containing 5% bovine serum, 20 U/ml penicillin, 20 μg/ml streptomycin, and 5 ng/ml gentamicin). Half of the livers were processed to enrich lymphocytes with a gentleMACS dissociator in C tubes according to the manufacturer’s instructions (Miltenyi Biotec). Alternatively, liver pieces were treated with collagenase at 100 U/ml in collagenase buffer (RPMI 1640 medium containing 10% fetal bovine serum, 1 mM CaCl₂, 1 mM MgCl₂, 20 U/ml penicillin, 20 μg/ml streptomycin, and 5 ng/ml gentamicin) at 37°C for 40 min. The homogenates were then mesh filtered through a 70-μm conical filter cap to obtain single-cell suspensions, and lymphocytes were enriched with a Percoll gradient. Small intestine intraepithelial lymphocytes (IELs) and lamina propria (LP) lymphocytes were collected as previously described (43, 44). Briefly, the small intestine was cut open and sliced into short pieces after removal of attached fat, mesentery, Peyer’s patches, and mucus. IELs were collected following two treatments with DTE buffer (1× Hanks balanced salt solution [HBSS] containing 1 mM HEPES, 2.5 mM NaHCO₃, and 1 mM dithioerythritol) at 37°C for 20 min each. Then, after removal of epithelial layers with EDTA buffer, LP lymphocytes were released after treatment with collagenase at 100 U/ml for 40 min at 37°C. IELs and LP lymphocytes were further enriched with a Percoll gradient.

The MLNs, spleens, or the other half of livers were homogenized in fluorescence-activated cell sorting (FACS) buffer (PBS containing 0.2% bovine serum albumin and 2 mM EDTA). Aliquots from the homogenized tissues were serially diluted in LB and plated (100 μl) on LB agar to determine bacterial colonization by CFU assay, and the limit of detection was 100 CFU or a log₁₀ CFU value of 2. All animal procedures were approved by the Stony Brook University Institutional Animal Care and Use Committee.

Flow cytometry.

Viable cells from the single cell suspension or enriched lymphocytes were counted using trypan blue exclusion with a Vi-CELL XR cell viability analyzer (Beckman Coulter). Initially, suspended cells (1 × 10⁶) were blocked using anti-mouse CD16/CD32 (FcγIII/II receptor) clone 2.4G2 (BD) and labeled with the allophycocyanin (APC)-conjugated MHC class I tetramer K^bYopE_69–77, which was provided by the NIH Tetramer Core Facility (Emory University, Atlanta, GA), or OVA_257–264, at room temperature for 1 h and fluorophore-conjugated antibodies on ice for 20 min. The antibodies used were Alexa Fluor 488- or phycoerythrin (PE)-labeled anti-mouse CD8α (53-6.7; BD) and PE/Cy7-labeled anti-mouse CD3e (clone 145-2C11; BD), Brilliant Violet 421- or PE-labeled anti-CD44, Brilliant Violet 421- or peridinin chlorophyll protein (PerCP)/Cy5.5-labeled anti-mouse CD4 (RM4-4), PE-labeled anti-mouse/human KLRG1 (2F1/KLRG1), or V500-labeled anti-mouse CD45.2 (104; BD). CD8⁺ T cells were gated as CD3⁺ CD8⁺ events or live CD45.2⁺ CD3⁺ CD8⁺ populations whenever possible. Labeled cells were analyzed using a Cytek DXP 8 color upgrade. To analyze memory formation, up to 5 × 10⁶ cells were mixed with anti-mouse CD16/CD32 and stained with Brilliant Violet 510-labeled anti-CD45, PerCP/Cy5.5-labeled anti-TCR-β, Brilliant Violet 786-labeled anti-CD8α, APC-eF780-labeled anti-CD44, APC-conjugated OVA tetramer, Brilliant Violet 421-conjugated YopE tetramer, PE-labeled anti-CD127, fluorescein isothiocyanate (FITC)-labeled anti-KLRG1, PE/Dazzle 594-labeled anti-CD103, and PE/Cy7-labeled anti-CD69. Labeled cells were analyzed with LSRFortessa Flow Cytometer using FACSDiva software. Antibodies were from BioLegend unless indicated otherwise. Data were analyzed with FlowJo software (Tree Star).

Secretion of IFN-γ and intracellular cytokine staining following ex vivo stimulation.

To stain for intracellular cytokine, an aliquot of suspended splenocytes or cells from MLNs (1 × 10⁶ cells) were incubated in 200 μl of complete T cell medium (Dulbecco modified Eagle medium supplemented with 10% heat-inactivated fetal bovine serum, 12.5 mM HEPES, 2 mM l-glutamine, 1 mM sodium pyruvate, 1 mM penicillin-streptomycin, and 55 μM β-mercaptoethanol) containing a 10 nM concentration of either OVA_257–264 (H₂N-SIINFEKL-OH) or YopE_69–77 peptide and brefeldin A (BFA; Sigma; 5 μg/ml) for 4.5 h at 37°C. Then the cells were washed and stained first for CD8α, CD4, CD3e, and CD45.2 as described above; then, the cells were fixed and permeabilized with BD Cytofix/Cytoperm kit according to manufacturer’s instructions. Finally, the cells were stained with PE-conjugated anti-mouse IFN-γ (clone XMG1.2) and PerCP/Cy5.5-conjugated anti-mouse TNF-α (clone MP6-XT22; both from BioLegend). Isotype-matched antibodies were used to control for nonspecific binding. To measure the secretion of IFN-γ, cells were incubated for 48 h without BFA. Then, the concentrations of IFN-γ in the supernatant were determined by enzyme-linked immunosorbent assay (ELISA) following the manufacturer’s instructions (R&D Systems, Inc.).

Precursor determination.

Naive antigen-specific CD8⁺ T cells were determined using the MHC-II tetramer pulldown protocol (45) with modifications. In brief, the spleen, inguinal, axillary, brachial, submandibular, cervical, para-aortic, and mesenteric lymph nodes were dissected; then, single-cell suspensions in sorter buffer (PBS with 2% fetal bovine serum) were filtered through a 70-μm nylon cell strainer, pooled, spun down, and resuspended in sorter buffer containing 2% mouse serum and 8 μg Fc blocking antibody 2.4G2 in a final volume about twice that of the pellet. Then, antigen-specific cells were enriched after sequential incubation with either OVA- or YopE-specific tetramers conjugated with APC and then with Miltenyi anti-APC microbeads, followed by standard LS column separation. Eluted cells were further incubated with OVA-specific PE tetramer or YopE-specific BV421 tetramer and then with CD8α-PerCP/Cy5 and CD90.2-PE/Cy7 together with FITC-conjugated dump antibodies (B220, Gr1, CD11c, F4/80, and CD4). Dead cells were disregarded based on staining with Alexa Fluor 700-carboxylic acid-succinimidyl ester. AccuCheck counting beads from Invitrogen were included in the samples of stained cells, and the total number of cells in each sample was determined with the following formula: cell count/bead count × bead stock concentration × bead volume/cell volume × total sample volume.

Statistical analysis.

Statistical analyses were performed in Prism (GraphPad Software). Tests used are listed for each figure. P values smaller than 0.05 were considered significant. Flow data were exported from FlowJo to Microsoft Excel for initial calculation.