Generation of specific effector and memory T cells with gut- and secondary lymphoid tissue- homing potential by oral attenuated CVD 909 typhoid vaccine in humans

Abstract

Induction of effective memory T cells is likely to be critical to the level and duration of protection elicited by novel live oral typhoid vaccines. Using cells from volunteers who ingested Salmonella Typhi vaccine strain CVD 909, we characterized the induction of interferon (IFN)-γ-secreting central (T_CM, CD45RO⁺CD62L⁺) and effector (T_EM, CD45RO⁺CD62L⁻) memory T populations, and their gut-homing potential based on integrin α4/β7 expression. Both CD4⁺ T_EM and T_CM populations secreted IFN-γ. However, although CD4⁺ T_EM expressed, or not, integrin α₄/β₇, CD4⁺ T_CM cells were predominantly integrin α₄/β₇⁺. In contrast, IFN-γ-secreting CD8⁺ cells were predominantly classical T_EM and CD45RA⁺ T_EM (T_EMRA, CD45RO⁻CD62L⁻) subsets. However, although CD8⁺ T_EM expressed, or not, integrin α₄/β₇, CD8⁺ T_EMRA were predominantly integrin α₄/β₇⁺. This is the first demonstration that oral immunization of humans with S. Typhi elicits diverse IFN-γ-secreting CD4⁺ and CD8⁺ T_CM and T_EM subsets able to migrate to the gut and other lymphoid tissues.

Introduction

Typhoid fever is caused by Salmonella enterica serovar Typhi (S. Typhi), a human-restricted pathogen that enters the host through the gut-associated lymphoid tissue. It is estimated that ∼21,650,000 cases and 216,500 deaths occur annually worldwide.¹ Current licensed typhoid vaccines include the purified Vi polysaccharide parenteral vaccine and the live oral attenuated Vi-negative galE mutant S. Typhi strain Ty21a vaccine (henceforth Ty21a).² Ty21a, which does not elicit anti-Vi antibodies, is likely to mediate protection by eliciting antibodies to other S. Typhi antigens^{2, 3} and an array of cell-mediated immune responses (CMIs) at the systemic and mucosal levels.^{4, 5, 6, 7, 8, 9, 10 and 11}

Although Ty21a is known to confer a moderate level of long-lived protection (5–7 years),¹² this requires the administration of three doses. Accordingly, others and we have engineered new attenuated typhoid vaccine strains that aim to be as well-tolerated as Ty21a yet immunogenic and protective following ingestion of just a single dose.^{2, 8, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22 and 23} Strain CVD 908-htrA¹⁹ was well-tolerated and immunogenic in clinical trials, eliciting potent antibody responses to S. Typhi antigens (other than Vi) and CMI.^{8, 18, 19, 21 and 22} In this regard, we have previously shown in volunteers immunized orally with attenuated strains of S. Typhi strains, including Ty21a, CVD 908, CVD 908-htrA, the induction of an array of potent CMI responses that might play an important role in host defense against S. Typhi.^{2, 4, 5, 6, 7, 8, 14, 18, 19, 20, 21, 22 and 23} To develop an attenuated vaccine strain that might also elicit Vi antibodies to enhance protection against typhoid fever, we modified CVD 908-htrA to constitutively express Vi, resulting in strain CVD 909.²⁴ In two clinical trials, CVD 909 was well-tolerated and strongly immunogenic, eliciting robust anti-lipopolysaccharide antibodies. Although serum anti-Vi antibody was not stimulated, IgA antibody-secreting cells to Vi were detected in the majority of vaccinees, indicating mucosal priming to this antigen.¹⁷ We also demonstrated that CVD 909 elicited a wide array of potent CMI, including cytotoxic T cells, interferon (IFN)-γ, tumor necrosis factor-α, and interleukin (IL)-10 (but not IL-2, IL-4, or IL-5) production and proliferation to S. Typhi antigens.²⁵

However, despite this progress, virtually no information is currently available regarding the induction of CD4⁺ and CD8⁺ memory T (T_M)-cell sub-populations by live oral bacterial vaccines and their ability to migrate to appropriate effector and induction sites. Understanding the contribution of these cells to host defense is critical to improve the induction of long-term protection by typhoid and other vaccines administered through mucosal surfaces. It is generally accepted that T_M cells express the CD45RO, but not the CD45RA isoform.^{26, 27, 28, 29, 30, 31, 32 and 33} Moreover, based on the coexpression of CD62L (L-selectin) surface molecules, T_M cells can be subdivided into two main subsets: central memory (T_CM, CD45RO⁺CD62L⁺) and effector memory (T_EM, CD45RO⁺CD62L⁻). Some T_EM, sometimes referred to as “terminal memory” cells, express CD45RA (T_EMRA, CD45RO⁻CD45RA⁺CD62L⁻).^{26, 27, 28, 29 and 30} This novel sub-population (T_EMRA ) of CD4⁺ and CD8⁺ cells recently has been described in humans during viral infections.^{26, 27, 28, 29, 30, 31, 32 and 33} The use of these markers also allows the identification of naive T cells (CD45RO⁻ CD62L⁺).^{26, 27, 28, 29, 30, 31, 32 and 33} In this study, we evaluated the induction of S. Typhi-specific T_CM and T_EM cell populations and their gut-homing potential, using peripheral blood mononuclear cells (PBMCs) isolated from subjects who were orally immunized with one or two doses of CVD 909 in two phase-1 clinical trials.^{17, 24 and 25}

We demonstrate, for the first time, that oral immunization of subjects with an attenuated S. Typhi vaccine elicited diverse specific IFN-γ-secreting CD4⁺ and CD8⁺ T_CM and T_EM T cells. We also provide the first available evidence in humans that oral immunization with an attenuated S. Typhi vaccine candidate strain elicits a subset of specific T_EM that coexpresses CD45RA (T_EMRA). Moreover, we observed that these specific T_M populations are composed of cells that express, or not, the gut-homing molecule integrin α4/β7. These data provide novel and unexpected insights into the homing characteristics and memory-effector T-cell sub-populations elicited by Gram-negative bacteria infection in humans and have wide implications for the development of live attenuated bacterial vaccines.

Results

Characterization of IFN-γ-secreting memory T-cell populations in response to soluble and particulate S. Typhi antigens

A fundamental tenet of the induction of effective and long-lasting immune responses following immunization is the induction of adequate and persistent memory T-cell responses. Thus, we characterized the IFN-γ-secreting cell populations with regard to their memory-effector characteristics following an overnight in vitro stimulation with S. Typhi antigens. After stimulation, PBMCs were simultaneously stained with monoclonal antibodies against CD3, CD4, CD8, CD45RO, CD62L, integrin α₄/β₇, and IFN-γ molecules conjugated to different fluorochromes as detailed in Methods. As expected based on our previous observations,²² CD4⁺ (CD3⁺ CD4⁺ CD8⁻) T cells were the predominant cell population that secreted IFN-γ in response to soluble S. Typhi antigens (e.g., S. Typhi flagella). In fact, in the current study, only four of the 10 volunteers studied showed IFN-γ increases in CD8⁺ (CD3⁺ CD4⁻ CD8⁺) cells, and these increases were of considerably lower magnitude than those of CD4⁺ cells (data not shown). Thus, we limited the analysis of the characteristics of IFN-γ-secreting cells in response to S. Typhi flagella (Figures 1 and 2) and TypVac (data not shown) to CD4⁺ T cells.

Figure 1

Characterization of the IFN-γ-secreting CD4⁺ T cells in response to ex vivo stimulation with purified S. Typhi flagella. Peripheral blood mononuclear cells (PBMCs) from CVD 909-vaccinated volunteers (n=10, from the Ty32004 and Ty35000 studies) 60/91 days after immunization were analyzed by seven-color flow cytometry. (a and c) Representative dot plots from one of the volunteers (subject no. 32004-4). (b) The mean±s.e. of the percentage of T memory cells (CD3⁺ CD4⁺ CD8⁻ CD45RO⁺) from the 10 subjects following stimulation with S. Typhi flagella. (d) The mean±s.e. distribution of naive (CD45RO⁻ CD62L⁺), T central memory cell (T_CM, CD45RO⁺ CD62L⁺), CD45RA-positive T effector memory cell (T_EMRA, CD45RO⁻ CD62L⁻), and T effector memory cell (T_EM, CD45RO⁺ CD62L⁻) subsets of CD4⁺ T cells from all subjects following stimulation. One-tail paired t-tests were used to analyze the data by comparing IFN-γ-secreting (IFN-γ⁺) and the corresponding IFN-γ non-secreting (IFN-γ⁻) sub-populations in b and d. *P<0.05. IFN-γ, interferon-γ.

PowerPoint slide

Figure 2

Expression of the gut-homing receptor integrin α₄/β₇ by IFN-γ-secreting CD4⁺ cells in response to ex vivo stimulation with purified S. Typhi flagella. PBMCs are from the same 10 volunteers described in Figure 1. (a and b) Representative dot plots from one of the volunteers (subject no. 32004-4). (c and d) The mean±s.e. distribution of naive, T_CM, T_EMRA, and T_EM CD4⁺ subsets in integrin α₄/β₇⁻ and integrin α₄/β₇⁺ subsets, respectively, from all subjects following stimulation with S. Typhi flagella. One-tail paired t-tests were used to analyze the data by comparing IFN-γ-secreting (IFN-γ⁺) and the corresponding IFN-γ non-secreting (IFN-γ⁻) sub-populations in c and d. *P<0.05. IFN-γ, interferon-γ; PBMCs, peripheral blood mononuclear cells.

PowerPoint slide

We observed positive significant net increases (mean 0.11±0.03%; range 0.05–0.26) in IFN-γ-secreting cells by total CD4⁺ populations exposed to S. Typhi flagella compared to respective “no antigen” controls (costimulants only) in seven of 10 vaccinees (P=0.007). Representative dot plots showing the gating strategy used for different sub-populations of T cells are shown in Figure 1a and c. As shown in Figure 1b, the percentage of antigen-specific IFN-γ-secreting CD4⁺ T cells coexpressing the T_M marker CD45RO in response to S. Typhi flagella was almost twice as high as that observed in CD4⁺ cells that did not secrete IFN-γ (63.3±3.4 vs. 37.3±2.1, respectively, P<0.001). When we further subdivided these IFN-γ-secreting cells into the various memory T sub-populations described above and compared them with the percentages of the corresponding T populations that did not secrete IFN-γ, we observed marked increases of CD4⁺ T central memory (T_CM, CD45RO⁺CD62L⁺) (P=0.003) and T effector memory cells (T_EM, CD45RO⁺CD62L⁻), (P<0.001) subsets within IFN-γ-secreting cells (Figure 1d). As expected, these changes were accompanied by significantly reduced percentages of the naive (CD45RO⁻CD62L⁺) T subset (P<0.001) within IFN-γ-secreting cells. These results showed that S. Typhi flagella-specific IFN-γ-secreting CD4⁺ T cells are mostly T_M cells with marked increases in both T_CM and T_EM sub-populations. Similar results were observed with TypVac (data not shown). No significant IFN-γ responses were observed in control cultures (absence of stimulation with S. Typhi antigens; data not shown).

Characterization of the homing potential of IFN-γ-secreting CD4⁺ memory T cells in response to soluble antigens

We next evaluated the hypothesis that immunization with CVD 909 elicited CD4⁺S. Typhi-specific T-cell responses able to home to the gut microenvironment. Selective homing of effector/memory cells to the gut is driven, to a large extent, by the expression of integrin α₄/β₇.^{32, 33, 34 and 35} Thus, we studied the expression of the gut-homing molecule integrin α₄/β₇ using an eight-color multichromatic flow cytometric approach to simultaneously assess the expression of integrin α₄/β₇ among T memory-effector sub-populations. Interestingly, results showed that IFN-γ-secreting CD4⁺ cells consisted of populations that express, or not, integrin α₄/β₇ (Figure 2a). However, stimulation with S. Typhi flagella resulted in significant increases in IFN-γ-secreting cells that coexpress integrin α₄/β₇ compared with the IFN-γ-negative populations (34.6±2.6% (range: 23.0–42.6) vs. 20.4±2.3% (range: 12.2–28.2), respectively) (P=0.001). To further characterize the expression of integrin α₄/β₇ in IFN-γ-producing cells, we compared the proportions of T_EM, T_CM, CD45RA⁺ T_EM (T_EMRA, CD45RO⁻CD62L⁻), and naive T subsets that express, or not, integrin α₄/β₇ between IFN-γ⁺ and IFN-γ⁻ subsets. The gating strategy is shown in Figure 2a and b. As shown in Figure 2c, among integrin α₄/β₇-negative cells, IFN-γ-secreting cells were significantly increased (P=0.001) only in the T_EM sub-population. In contrast, among cells expressing integrin α₄/β₇, significant increases in IFN-γ were observed in both T_CM (P=0.011) and T_EM (P=0.004) subsets (Figure 2d).

Characterization of IFN-γ-secreting memory CD8⁺ and CD4⁺ T-cell populations cocultured with S. Typhi-infected cells

We have previously reported that the predominant effector population responding to soluble (S. Typhi flagella) and particulate (TypVac) S. Typhi antigens is composed of CD3⁺ CD4⁺ CD8⁻ T cells. In contrast, the effector subsets responding to S. Typhi-infected stimulators included both CD3⁺ CD8⁺ CD4⁻ (predominant), as well as CD3⁺ CD4⁺ CD8⁻ populations.^{4, 5, 6, 8 and 22} In this study, we confirmed and markedly extended these findings. We used PBMC from six subjects whom we had previously shown to specifically secrete IFN-γ when exposed to S. Typhi-infected targets²⁵ in an eight-color multichromatic flow cytometric approach to characterize in depth these IFN-γ-producing T-cell sub-populations and their homing potential following exposure to S. Typhi-infected blasts. To this end, stimulator cells (S. Typhi-infected or non-infected autologous blasts as controls) were stained with CD45-biotin followed by streptavidin conjugated with pacific blue before coculture with effectors to enable their exclusion by electronic gating during flow cytometric analysis (i.e., only CD45-negative effector populations were analyzed). This approach eliminated an important source of nonspecific background (see Methods for details). Cells incubated in the presence of S. Typhi-infected or non-infected autologous blasts were simultaneously stained with monoclonal antibodies against CD3, CD4, CD8, CD45RO, CD62L, integrin α₄/β_7, and IFN-γ molecules conjugated to different fluorochromes. We then determined by flow cytometry the proportions of T_EM, T_CM, T_EMRA, and T naive subsets in IFN-γ⁺ and IFN-γ⁻ cells, as well as coexpression of integrin α₄/β₇. The gating strategy for these studies is shown in Figure 3a, b, d and e.

Figure 3

Characterization of IFN-γ-secreting CD8⁺ and CD4⁺ T cells following an overnight incubation with S. Typhi-infected autologous blasts. PBMCs from six CVD 909-vaccinated volunteers were analyzed by eight-color flow cytometry as described in Methods. (a, b, d, and e) Representative dot plots from one of the volunteers. (c) The mean±s.e. of the percentage of CD4⁺ and CD8⁺ T memory cells (CD3⁺ CD4⁺ CD8⁻ CD45RO⁺ and CD3⁺ CD4⁻ CD8⁺ CD45RO⁺, respectively) from all subjects following stimulation. (f and g) The mean±s.e. distribution of naive, T_CM, T_EMRA, and T_EM subsets of CD8⁺ and CD4⁺ T cells, respectively, from all subjects following stimulation with S. Typhi-infected autologous blasts. One-tail paired t-tests were used to analyze the data by comparing IFN-γ-secreting (IFN-γ⁺) and the corresponding IFN-γ non-secreting (IFN-γ⁻) sub-population in c, f, and g. *P<0.05. IFN-γ, interferon-γ; PBMCs, peripheral blood mononuclear cells.

PowerPoint slide

The patterns of cell populations that secrete IFN-γ in response to ex vivo stimulation with S. Typhi-infected blasts showed a different profile than that exhibited following stimulation with soluble antigens (see above). CD8⁺ cells were the predominant population that secreted IFN-γ following incubation with S. Typhi-infected blasts. The percentages of CD8⁺ IFN-γ⁺ cells were two- to threefold higher (mean net increases 0.46±0.15%; range: 0.26–0.85%) when cocultured with S. Typhi-infected autologous blasts than in the presence of non-infected blasts (negative controls) (P=0.036). Moreover, a strong trend toward increases in IFN-γ⁺ cells, albeit of lower magnitude, was also observed in CD4⁺ cells from these subjects (mean net increases 0.09±0.04% (range: 0.05–0.21%, P=0.06)).

The percentages of antigen-specific IFN-γ-secreting CD8⁺ T cells coexpressing CD45RO (T_M) following coculture with S. Typhi-infected blasts were higher than those observed in CD8⁺ subsets that did not secrete IFN-γ (42.8±34.9 vs. 22.5±3.1, respectively, P=0.01) (Figure 3c). Increased proportions of CD4⁺ CD45RO⁺ cells were also observed among IFN-γ⁺ cells (71.4±4.8 vs. 44.6±6.0, respectively, P<0.005) (Figure 3c).

When we further subdivided CD8⁺ IFN-γ⁺ cells into the various T sub-populations as described above and compared them with the percentages of the corresponding T populations that did not secrete IFN-γ, we observed marked increases of CD8⁺ T_EM (P=0.02) and a strong trend toward increases in T_EMRA (P=0.12) subsets within IFN-γ-secreting cells. These changes were accompanied by significantly reduced percentages of the naive T subset (P=0.013) within IFN-γ-secreting cells. No changes were observed in the T_CM subset (Figure 3f).

Similar analysis of CD4⁺ cells revealed marked increases of CD4⁺ T_EM (P=0.004) and significantly reduced percentages of the naive CD4⁺ T subset (P=0.007) within IFN-γ-secreting cells. No changes were observed in the CD4⁺ T_CM and T_EMRA subsets (Figure 3g). No significant IFN-γ responses were observed in control cultures (cocultures with non-infected targets; data not shown).

Characterization of the homing potential of IFN-γ-secreting memory CD8⁺ and CD4⁺ T cells cocultured with S. Typhi-infected targets

We then studied integrin α₄/β₇ expression on CD8⁺ and CD4⁺ T cells that specifically secrete IFN-γ following exposure to S. Typhi-infected blasts. Interestingly, IFN-γ-secreting CD8⁺ cells consisted of populations that express, or not, integrin α₄/β₇ (Figure 4a and b). Incubation with S. Typhi-infected blasts resulted in significant decreases in IFN-γ-secreting cells that coexpress integrin α₄/β₇ compared with the IFN-γ-negative populations (35.0±4.6 (range: 27.8–49.6%) vs. 51.2±6.6 (range: 36.8–64.4), respectively) (P=0.012). These results are in contrast with those observed in S. Typhi flagella-simulated cultures.

Figure 4

Expression of the gut-homing receptor integrin α4/β7 by CD8⁺ and CD4⁺ IFN-γ-secreting cells following an overnight incubation with S. Typhi-infected autologous blasts. PBMCs are from the same six volunteers described in Figure 3. Shown are the means±s.e. distributions of naive, T_CM, T_EMRA, and T_EM CD8⁺ (a and b) and CD4⁺ (c and d) populations in integrin α₄/β₇⁻ (a and c) and integrin α₄/β₇⁺ subsets (b and d) from all subjects following stimulation with S. Typhi-infected autologous blasts. One-tail paired t-tests were used to analyze the data by comparing IFN-γ-secreting (IFN-γ⁺) and the corresponding IFN-γ non-secreting (IFN-γ⁻) sub-populations. *P<0.05. IFN-γ, interferon-γ; PBMCs, peripheral blood mononuclear cells.

PowerPoint slide

To characterize the expression of integrin α₄/β₇ in CD8⁺ IFN-γ⁺ cells, we compared the proportions of T_EM, T_CM, T_EMRA, and naive T subsets that express, or not, integrin α₄/β₇ between IFN-γ⁺ and IFN-γ⁻ subsets. As shown in Figure 4a, among integrin α₄/β₇-negative cells, IFN-γ-secreting cells were signi-ficantly increased (P=0.023) only in the T_EM sub-population. In contrast, among cells expressing integrin α₄/β₇, significant increases in IFN-γ⁺ cells were observed in both T_EMRA (P=0.01) and T_EM (P=0.02) subsets (Figure 4b). The proportion of naive cell populations was significantly lower in both integrin α₄/β₇⁻ (P=0.02) and integrin α₄/β₇⁺ (P=0.004) IFN-γ-secreting CD8⁺ cells. No differences were recorded in T_CM subsets regardless of whether they express or not integrin α₄/β₇.

We also characterized the expression of integrin α₄/β₇ in CD4⁺ IFN-γ⁺ cells, by comparing the proportions of T_EM, T_CM, T_EMRA, and naive T subsets that express, or not, integrin α₄/β₇ between IFN-γ⁺ and IFN-γ⁻ subsets. Somewhat increased percentages of cells expressing integrin α₄/β₇ were observed within the CD4⁺ IFN-γ-secreting cell populations (34.4±4.6 vs. 27.4±3.8). Specific CD4⁺ T cells that produce IFN-γ following stimulation with S. Typhi-infected blasts followed a pattern similar to that recorded in response to S. Typhi flagella. As shown in Figure 4c, among integrin α₄/β₇-negative cells, IFN-γ-secreting cells were significantly increased (P=0.005) only in the T_EM sub-population. In contrast, among cells expressing integrin α₄/β₇, significant increases in IFN-γ⁺ cells were observed in both T_CM (P=0.005) and T_EM (P=0.016) subsets (Figure 4d). The proportion of naive cell populations was signi-ficantly lower in both integrin α₄/β₇⁻ (P=0.011) and integrin α₄/β₇⁺ (P=0.001) IFN-γ-secreting CD4⁺ cells. No differences were recorded in T_EMRA subsets regardless of whether they express or not integrin α₄/β₇.

Discussion

Some of the critical challenges being addressed in the deve-lopment of new typhoid vaccines are how to achieve strong immunogenicity without increasing reactogenicity and how to elicit long-term efficacy.³⁶ Enduring protection relies upon the induction of strong, long-lived, broadly based immunological B- and T-cell memory to antigens that correlate with protection, i.e., the ability to recall previous exposures to antigens and to mount enhanced and accelerated effector responses.³⁶ The development of improved typhoid vaccines is a public health priority. In recent years, a handful of promising new attenuated oral typhoid vaccine candidates have been developed by our group and others that exhibit low reactogenicity and broad immunogenicity, encompassing a wide array of CMI responses including increased proliferative responses and cytokine production (particularly IFN-γ, tumor necrosis factor-α, and IL-10), as well as classical and non-classical class I-restricted cytotoxic T cell activity to purified S. Typhi antigens and S. Typhi-infected cells.^{4, 5, 6, 8, 9, 11, 14, 15, 17, 18, 19, 21, 22, 23, 25 and 37} These responses are likely to play an important role in the host defense against S. Typhi by several mechanisms, including enhancement of the bactericidal activity of cells of the innate immune response (e.g., macrophages) and antigen presentation, killing of S. Typhi-infected cells, and providing help for antibody responses to S. Typhi antigens.^{4, 5, 6, 8, 22, 23, 25, 38 and 39}

However, long-term protection depends, to a large extent, on the ability of a vaccine to elicit T_M able to migrate to effector and inductive sites.³⁶ Here, CD4⁺ and CD8⁺ T_M cell populations were evaluated for their expression of CD45RO and the lymph node-homing molecule CD62L (the expression of which has been shown to be highly correlated with that of CCR7).^{30, 35 and 40} The T-cell phenotype based on expression of these markers allows one to define central memory T cells (T_CM, CD45RO⁺ CD62L⁺), effector memory T cells (T_EM, CD45RO⁺ CD62L⁻), and naive T cells (largely CD45RO⁻ CD62L⁺),^{26, 27, 28, 29, 30, 31, 32 and 33} as well as a recently described population of T memory-effectors that express CD45RA (T_EMRA, CD45RO⁻ CD62L⁻), which appears to play a significant role during viral infections in humans.^{26, 27, 28 and 30} It is widely accepted that T_EM populations are ready for imme-diate effector action, whereas T_CM are important in the generation of a new wave of effector T cells.^{30, 31, 32 and 33}

In the present studies, we used anti-CD28 antibody and anti-CD49d monoclonal antibodies (mAbs) as costimulators to characterize the various T_M subsets in response to soluble antigens. Costimulation with these mAbs enhances the detection of low frequencies of antigen-specific memory-effector CD4⁺ and CD8⁺ cell populations with minimal or no increase in the levels of nonspecific stimulation.^{41, 42}

As described in Methods, during the flow cytometric analysis, we followed an electronic-gating strategy designed to eliminate bias in our observations. CD3⁺ CD4⁺ CD8⁻ and CD3⁺ CD4⁻ CD8⁺ cells were analyzed for IFN-γ expression and the expression of integrin α₄/β₇ in each sub-population determined before evaluating the proportions of T_EM, T_CM, T_EMRA, and naive T cells. Using this approach, we first analyzed IFN-γ-secreting cells in response to soluble antigens. In agreement with our previous observations, responses to soluble S. Typhi antigens were predominantly mediated by CD4⁺,^{8, 22} with only a minor component attributable to CD8⁺ cells. IFN-γ-secreting cells were mostly CD4⁺ T_M cells, predominantly T_EM. However, it is noteworthy that a sizable proportion of IFN-γ-secreting cells was T_CM cells and that even some naive T and T_EMRA subsets were found to be IFN-γ⁺. These findings could be explained by the marked heterogeneity observed even within these well-defined populations. For example, it has been shown that T_M are heterogeneous for CD62L expression;^{30, 32 and 33} thus, some of the cells that are defined in this study as T_CM (CD45RO⁺ CD62L⁺) could be T_EM expressing moderate levels of CD62L. Similarly, some of the cells in the naive population (CD45RO⁻ CD62L⁺) may be T_EMRA that express variable levels of CD62L. Further studies in which expression of CCR7, CD27, and other molecules^{30, 32 and 33} are evaluated in conjunction with the molecules described above using more colors in even more complex multichromatic flow cytometry studies will be required to further define the anti-S. Typhi-specific T_M cells elicited by immunization.

It is worth noting that throughout these studies, no significant differences were observed in the proportions of T_EM, T_CM, T_EMRA, and naive T sub-populations among IFN-γ-negative CD4⁺ or CD8⁺ T cells from experimental samples (stimulated with S. Typhi antigens or S. Typhi-infected blasts) and in those observed in negative control cultures (no antigen or non-infected blasts) (data not shown). These results are consistent with the contention that T cells that did not secrete IFN-γ in response to stimulation with S. Typhi flagella or S. Typhi-infected blasts represent non-responding T-cell populations as measured by this effector function.

We next explored the important hypothesis that immunization with live attenuated oral Salmonella vaccines is likely to induce S. Typhi-specific T_M cells with the capability to home to the gut, the site of entry of S. Typhi. This was investigated by determining the expression of integrin α₄/β₇ in the various T-cell subsets. Unexpectedly, although there was an increase in the percentages of CD4⁺ IFN-γ⁺ T cells that express integrin α₄/β₇, a significant proportion of IFN-γ⁺ cells did not express integrin α₄/β₇. Moreover, we observed differences in the proportions of the CD4⁺ T_M subsets present in integrin α₄/β₇⁺ and integrin α₄/β₇⁻ cells. Although IFN-γ⁺ T_CM and T_EM subsets were found to be increased in integrin α₄/β₇⁺ cells, only IFN-γ⁺ T_EM were enriched in integrin α₄/β₇⁻ cells. These results indicate that immunization with CVD 909 elicits a complex pattern of CD4⁺ T_M cells with the capability to home to the lamina propria and Peyer's patches in the gut, as well as to peripheral lymph nodes and perhaps to other sites as well.

We have previously shown that CD8⁺ cytotoxic T-cell responses, which are restricted by classical and non-classical HLA-E major histocompatibility complex molecules, might play a key role by eliminating cells harboring S. Typhi.^{4, 5, 6, 22 and 23} Here, we studied in detail the various T_M subsets of IFN-γ-secreting cells in response to S. Typhi-infected autologous blasts, a response that, although mediated mostly by CD8⁺ cells, also includes a sizable CD4⁺ component.⁸ It is likely that both CD4⁺ and CD8⁺ cells respond to this stimulation because infected targets are able to present S. Typhi antigens in the context of both class I and II molecules. The present studies confirmed and extended previous findings by identifying T_M cells (CD45RO⁺) as the predominant responding populations (Figure 3), although some IFN-γ⁺ cells lack expression of CD45RO. When we looked at T_M subsets composing the IFN-γ⁺ T-cell population, we found differences between the CD8⁺ and CD4⁺ cells. Although increases were observed in T_EM and T_EMRA subsets of CD8⁺ IFN-γ⁺ cells, only increases in T_EM subset were recorded in CD4⁺ populations (Figure 3). To our knowledge, this is the first demonstration that T_EMRA is a significant component of the T_EM memory response to bacteria in humans. This could be very important given that T_EMRA are a distinct subset of CD8⁺ T cells that has been shown to carry the largest amount of perforin and that has been proposed to be among the most active effector T cells.^{26, 27, 28 and 30}

Concerning the expression of homing molecules in CD8⁺ IFN-γ⁺ T_M populations, we have observed marked differences between responses to stimulation with S. Typhi-infected blasts and those measured to soluble S. Typhi antigens. It is important that the proportion of CD8⁺ T_M cells expressing integrin α₄/β₇ was lower in IFN-γ⁺ than in IFN-γ⁻ subsets. This could be the result of a depletion in peripheral blood of integrin α₄/β₇⁺ CD8⁺ T_M cells, which peaked at earlier times, and were retained in the gut. As a result, when measured 60–90 days after immunization, most recirculating T_M cells might not express integrin α₄/β₇. However, the CD8⁺ T_M subsets that do persist in circulation showed different compositions among integrin α₄/β₇⁺ and integrin α₄/β₇⁻ cells. Although increased proportions of T_EM were observed in integrin α₄/β₇⁻ cells, both T_EMRA and T_EM were increased among integrin α₄/β₇⁺ populations. In contrast, the CD4⁺ T_M subset distribution was similar to that observed with soluble antigens, i.e., increased T_EM in integrin α₄/β₇⁻ and increased T_CM and T_EM responses in integrin α₄/β₇⁺ populations. Other homing molecules, e.g., the CCR9 receptor, expressed by subsets of integrin β₇⁺ T cells, and its ligand, the chemokine CCL25, have also been shown to play a selective role in effector T-cell homing to the small intestinal mucosa.³⁵ The expression of this and other important homing molecules will be investigated in future studies to further explore the gut-homing potential of specific effector and memory T cells elicited by live oral S. Typhi vaccines in humans.

Taken together, our results in CVD 909 vaccinees showed that the responses elicited are heterogenous with regard to the S. Typhi-specific T_M subsets stimulated, as well as their capability to migrate both to the gut (lamina propria as well as Peyer's patches) and to the peripheral lymph nodes. However, we observed distinct patterns in CD4⁺ and CD8⁺ T cells. Although increases in the proportions of S. Typhi-specific CD4⁺ integrin α₄/β₇⁺ T_M cells were largely in T_EM and T_CM subsets, the CD8⁺ integrin α₄/β₇⁺ T_M subsets preferentially expanded were predominantly T_EM and T_EMRA. In contrast, increases in integrin α₄/β₇⁻ CD4⁺ and CD8⁺ T_M cells were exclusively on T_EM cells. These observations are in contrast to the only other study in the literature, of which we are aware, that has recently reported that immunization with Ty21a induced CD4⁺ and CD8⁺ T_M cells that almost exclusively express integrin β₇.¹¹ However, these disparities could be attributed to several factors, including the use of different vaccines, that responses were observed at days 7 and 14 but not at later time points, that only CD45RA was used to identify T_M cells (an inaccurate marker of T_M in the absence of coexpression of other molecules such as CD62L or CCR7), that an mAb to integrin β₇ instead of integrin α₄/β₇ was used, or to differences in antigenic preparations. Among these variables, the one that is likely to be of particular significance is that only cells that are homing to the gut could be identified at early time points, whereas a more complex mix of T_M subsets homing to various tissues can be found in circulation at later times. We are directly addressing this possibility by carefully studying the kinetics of induction following immunization.

In sum, we have provided novel information concerning specific T_M-cell subsets elicited by immunization with an attenuated S. Typhi oral vaccine, including the first demonstration that bacterial infection elicits the appearance of strong CD8⁺ T_EMRA responses in humans. Moreover, we showed that immunization elicited T_M subsets with the potential to migrate to the gut as well as to peripheral lymph nodes that persist for at least 90 days in circulation. These observations have broad implications for the development of attenuated bacterial vaccines, as well as for their use as live vectors to deliver prokaryotic and eukaryotic expression vectors in humans.

Methods

Subjects, vaccinations, and isolation of PBMC. Peripheral blood mononuclear cells from subjects recruited from the Baltimore metropolitan area participating in two separate CVD 909 phase-1 vaccine trials were used in these studies.^{17, 25} One study (hereafter Ty32004) consisted of six subjects (median age 28.3 years, range: 20–39 years) who received a single dose of CVD 909 (2.5×10⁹ colony-forming units), and blood samples were collected on days 0, 28, and 91 postimmunization. The second study (hereafter Ty35000) included eight healthy subjects (median age 24 years, range 19–39 years) who were immunized with two oral doses of CVD 909 (6.2×10⁹ colony-forming units per dose), one dose on day 0, and one dose on day 14. PBMCs from 12 of these volunteers, who have previously shown to specifically secrete IFN-γ in response to S. Typhi flagella, S. Typhi particulate, or S. Typhi-infected targets following immunization²⁵ were used in these studies. Blood samples from these subjects were obtained on days 0, 28, and 60 postvaccination.²⁵ PBMCs were isolated immediately after the blood draws by density-gradient centrifugation and cryopreserved in liquid N₂ following standard techniques until used in the assays.^{14, 22} Volunteers gave signed informed consent. The human experimentation guidelines of the US Department of Health and Human Services and those of the University of Maryland, Baltimore, were followed in the conduct of the present clinical research.

Antigens. S. Typhi flagella H-antigen: Flagellar antigen was purified from the rough S. Typhi strain Ty2R following standard procedures and used at a concentration of 10 μg ml⁻¹.^{7, 14}

Heat-phenolyzed whole-cell S. Typhi (TypVac): The particulate S. Typhi antigen consisted of Ty2 heat-killed, phenol-preserved, phenol-killed, whole-cell bacteria (typhoid vaccine; Wyeth Laboratories, Marietta, PA, lot no. 4978038, ∼2.5×10⁸ organisms per ml) and was used in the assays as previously described.¹⁴

Preparation of autologous blasts for use as stimulators in intracellular IFN-γ production assays by flow cytometry: Autologous phytohemagglutinin-stimulated PBMCs (henceforth called “blasts”) were used in this study as stimulator cells. Blasts were prepared by stimulation with 1 μg ml⁻¹ phytohemagglutinin-L for 24 h, washed, and cultured with 20 U ml⁻¹ human recombinant interleukin-2 (Boehringer Gmbh, Mannheim, Germany) for 5–6 days as previously described.^{6, 25}

Infection of stimulator cells with S. Typhi for use in intracellular IFN-γ production assays by flow cytometry: Blasts were incubated in RPMI-1640 (without antibiotics) for 3 h at 37 °C in the absence or presence of wild-type S. Typhi strain ISP1820 (wild-type S. Typhi) (obtained from J. Nataro, Center for Vaccine Development) at different multiplicities of infection as previously described.²² After exposure to S. Typhi, cells were washed and incubated overnight at 37 °C in RPMI-1640 supplemented with 100 U ml⁻¹ penicillin, 100 μg ml⁻¹ streptomycin, 50 μg ml⁻¹ gentamicin, 2 mM L-glutamine, 2.5 mM sodium pyruvate, 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid buffer, and 10% heat-inactivated fetal bovine serum (complete RPMI) before being used to stimulate effector PBMC.

Ex vivo stimulation for IFN-γ secretion studies by multichromatic flow cytometry intracellular cytokine analysis: PBMCs resuspended in complete RPMI were stimulated with S. Typhi flagella (10 μg ml⁻¹) or TypVac (2×10⁵ particles per well) in the presence of mAbs to costimulatory molecules CD28 (clone 37.51) and CD49d (clone 9F10) (1.5 μg ml⁻¹ each; BD PharMingen, San Jose, CA), or with S. Typhi-infected or uninfected autologous blasts (1:7 to effector PBMC). Unstimulated PBMCs or stimulation with phytohemagglutinin-L (1 μg ml⁻¹), or PMA (1 ng ml⁻¹) plus ionomycin (500 ng ml⁻¹) were used as negative and positive controls, respectively. After an overnight incubation, Brefeldin A (Sigma-Aldrich, St Louis, MO) was added at a final concentration of 10 μg ml⁻¹ and cultures were continued for an additional 5–6 h. Cells were then harvested, washed, and stained for seven- or eight-color multichromatic flow cytometric analysis as described below.

Staining procedures and seven- or eight-color multichromatic flow cytometry analysis strategy: Surface staining with monoclonal antibodies to identify T-cell sub-populations and intracellular staining to determine IFN-γ content were performed according to standard methods.^{4, 5, 6 and 22} For the experiments done with antigens (S. Typhi flagella and TypVac), cells were surface-stained with CD3-PE-Cy7 (clone UCHT1; Beckman-Coulter, Fullerton, CA), CD4-FITC (clone RPA-T4; PharMingen, San Jose, CA), CD8-PerCP-Cy5.5 (Clone SK1; BD PharMingen), CD45RO-ECD (PE-Texas red, Clone UCHL-1; Immunotech, Fullerton, CA), CD62L-APC-Cy7 (Clone DREG-56; Caltag, Carlsbad, CA), and integrin α4β7-Alexa 647 conjugate (the ACT-1mAb was kindly provided by W. Newman, Leukosite, Cambridge, MA), and conjugated to Alexa 647 using an Alexa 647-labeling kit (Molecular probes, Eugene, OR). In some experiments, infected and uninfected autologous blasts were used as stimulators. In the latter studies, target cells were stained with CD45-biotin (a leukocyte common surface antigen, Clone H130; BD PharMingen) and streptavidin-conjugated with pacific blue (Molecular Probes) before exposure to the effectors to enable their exclusion from analysis as previously described.⁶ Following surface staining, cells were fixed and permeabilized for intracellular staining with IFN-γ-PE (Clone B27; PharMingen). Samples were run in a MoFlo flow cytometer/cell sorter system (Coulter-Cytomation, Fort Collins, CO) and analyzed using the WinList flow cytometry software analysis package (Verity Software House, Topsham, ME). Cells present on the lymph gate (defined in forward vs. side scatter cytograms) were gated on CD3⁺ cells and then on either CD4⁺CD8⁻ or CD4⁻CD8⁺ subsets. Cells in each population were subsequently analyzed for IFN-γ expression and the expression of integrin α₄/β₇ in each sub-population was determined. Finally, the proportions of T_EM, T_CM, T_EMRA, and naive T cells were determined on IFN-γ⁺ and IFN-γ⁻ subsets that express, or not, integrin α₄/β₇.

Statistical analysis. All tests were performed using SigmaStat software (version 3.1; Systat Software, Point Richmond, CA). P-values <0.05 were considered significant.

Disclosure

Drs Sztein and Levine are coinventors of a patent for attenuated Salmonella enterica serovar Typhi strains that constitutively express Vi capsular polysaccharide antigen. However, no company has licensed this technology. Drs Wahid, Salerno-Gonçalves, and Tacket declare that they do not have a commercial or other association that might pose a conflict of interest.