Immunogenicity and Efficacy of Live-Attenuated Salmonella Typhimurium Vaccine Candidate CVD 1926 in a Rhesus Macaque Model of Gastroenteritis

ABSTRACT

Salmonella Typhimurium is a common cause of foodborne gastroenteritis and a less frequent but important cause of invasive disease, especially in developing countries. In our previous work, we showed that a live-attenuated S. Typhimurium vaccine (CVD 1921) was safe and immunogenic in rhesus macaques, although shed for an unacceptably long period (10 days) postimmunization. Consequently, we engineered a new strain, CVD 1926, which was shown to be safe and immunogenic in mice, as well as less reactogenic in mice and human cell-derived organoids than CVD 1921. In this study, we assessed the reactogenicity and efficacy of CVD 1926 in rhesus macaques. Animals were given two doses of either CVD 1926 or saline perorally. The vaccine was well-tolerated, with shedding in stool limited to a mean of 5 days. All CVD 1926-immunized animals had both a serological and a T cell response to vaccination. At 4 weeks postimmunization, animals were challenged with wild-type S. Typhimurium I77. Unvaccinated (saline) animals had severe diarrhea, with two animals succumbing to infection. Animals receiving CVD 1926 were largely protected, with only one animal having moderate diarrhea. Vaccine efficacy in this gastroenteritis model was 80%. S. Typhimurium vaccine strain CVD 1926 was safe and effective in rhesus macaques and shed for a shorter period than other previously tested live-attenuated vaccine strains. This strain could be combined with other live-attenuated Salmonella vaccine strains to create a pan-Salmonella vaccine.

INTRODUCTION

Nontyphoidal Salmonella (NTS) is an important cause of gastroenteritis worldwide. In the United States, foodborne outbreaks of Salmonella gastroenteritis are a regular occurrence. The economic burden of NTS gastroenteritis in 2013 was measured to be $3.6 billion USD (1). While most people will suffer only a self-limiting gastroenteritis from NTS infections, these outbreaks can be deadly in older adults and are a major issue for long-term assisted care facilities (2–4). In addition, the spread of antibiotic resistance among NTS strains has resulted in up to 100,000 antibiotic-resistant infections per year in the United States (5). This has led the Centers for Disease Control and Prevention to list NTS as a serious threat, requiring prompt and sustained action (5).

Of the nontyphoidal strains that are actively circulating in the United States and Europe, Salmonella enterica serovar Typhimurium has consistently been one of the most highly prevalent (6, 7). There are currently no licensed vaccines for S. Typhimurium. One candidate, live-attenuated vaccine strain WT05, has been assessed for safety and immunogenicity in a human clinical trial (8). Immunization with WT05 induced a serum antibody response against S. Typhimurium lipopolysaccharide (LPS) in adults receiving the highest doses (10⁸ to 10⁹ CFU). The vaccine was well tolerated, but the strain was shed in stool for up to 23 days postimmunization. This is an issue for community safety, since the longer a vaccine is shed, the more likely it is to be transmitted to immunocompromised individuals.

In our previous work, we created a novel S. Typhimurium live-attenuated vaccine strain, CVD 1921, which contained mutations in the guaBA and clpP loci (9). Mutation of guaBA leads to guanine auxotrophy (10), while deletion of clpP has multiple effects on Salmonella, including attenuation as well as upregulation of cell-associated flagella (11, 12). This vaccine strain was shown to be safe and immunogenic in mice and rhesus macaques (9, 13). Both healthy (simian immunodeficiency virus [SIV]-uninfected) and SIV-infected macaques were used for safety studies, with the vaccine being well tolerated in both populations (13). Serum antibody responses against the vaccine strain were detected in three of three healthy and one of three SIV⁺ rhesus macaques after a single vaccine dose. However, the vaccine strain was shed in stool for 10 days postimmunization.

To reduce shedding, we added two additional mutations to CVD 1921: ΔpipA and ΔhtrA. The new vaccine strain is called CVD 1926 (14). The pipA gene has multiple functions, one of which is its ability to induce intestinal fluid accumulation (15). Mutation of pipA in both S. Dublin and S. Typhimurium caused a decrease in the ability of the bacteria to induce fluid accumulation in ileal loops, a proxy for diarrhea (14, 15). The htrA gene encodes a heat shock protein. The loss of this gene is highly attenuating and has been used in other live-attenuated vaccine candidates, such as Salmonella Typhi CVD 908-htrA (16), to prevent vaccinemia and thereby improve vaccine safety.

The combination of these four mutations resulted in a vaccine strain that was safe in mice and also strongly immunogenic. In a human intestinal organoid model, vaccine strain CVD 1926 was less inflammatory than previous vaccine iterations, inducing less proinflammatory cytokines (14). Despite this decreased reactogenicity, CVD 1926 was highly immunogenic in mice, producing anti-LPS titers which lasted for at least 3 months postimmunization. Mice receiving three doses of CVD 1926 and challenged with wild-type S. Typhimurium at either 1 or 3 months postimmunization were significantly protected from mortality (14).

The aim of this study was to determine the reactogenicity and immunogenicity of the optimized CVD 1926 vaccine strain in rhesus macaques and efficacy against Salmonella gastroenteritis. The rhesus macaque model is the gold standard animal model for S. Typhimurium gastroenteritis, because infected macaques show many of the clinical signs observed in humans (17–19). This study provides crucial data regarding vaccine efficacy of CVD 1926, shedding, and vaccine-induced immune responses for future clinical trials, thereby informing vaccine development efforts to prevent NTS gastroenteritis.

RESULTS

Immunization of rhesus macaques with CVD 1926.

Rhesus macaques were immunized with either saline or CVD 1926 (S. Typhimurium I77 ΔguaBA ΔclpP ΔpipA ΔhtrA) intragastrically. Two doses of vaccine were given 4 weeks apart. Immunization was well tolerated, with no animals showing signs of illness or dehydration. After the first immunization, bacteria were shed for between 4 and 8 days postimmunization, with the average being 5 days of shedding (Fig. 1). The peak of shedding was observed on day 1 postimmunization, with animals shedding 1.5% of the vaccine inoculum on average. After the second vaccine dose, an average of 0.12% of the inoculum was shed on day 1 postimmunization.

FIG 1.

Shedding of vaccine strain CVD 1926 in stool postimmunization. The vaccine strain CVD 1926 was detected in stool after the first vaccine dose by direct colony counts on Salmonella-Shigella selective agar plates. The limit of detection (dashed line) was 100 CFU/g.

Immune response to vaccination.

To determine the antibody responses to vaccination, serum anti-COPS and anti-flagellin IgG titers were measured before immunization and at days 7, 21, 35, and 49 postimmunization. The anti-COPS IgG titers prevaccination were all below the limit of detection (Fig. 2A). After the first dose of CVD 1926, three of five rhesus macaques seroconverted and after the second dose, all five animals seroconverted. Anti-COPS serum IgG titers in animals receiving CVD 1926 were significantly higher than those receiving saline at both days 35 and 49 postimmunization (P = 0.01).

FIG 2.

Serum anti-COPS and anti-FliC IgG titers in rhesus macaques postimmunization. Sera were collected from animals preimmunization and on days 7, 21, 35, and 49 postimmunization. Immunizations occurred on days 0 and 28, as indicated by arrows. Anti-COPS (A) and anti-FliC (B) IgG titers were determined by ELISA. Seroconversion was defined as a 4-fold increase over preimmunization titers. (C) Antibody functionality was measured by comparing opsonization and uptake by J774 macrophages. The data are shown as the fold uptake (uptake postimmunization/uptake preimmunization). Animals administered saline are represented by gray closed symbols, and animals administered CVD 1926 are represented by black open symbols. Limits of detection are shown as dashed lines. *, P < 0.05; **, P < 0.001 (Student t test).

In contrast, anti-flagellin serum IgG titers preimmunization were variable between animals (Fig. 2B). Despite the higher baseline titers, four of five animals receiving CVD 1926 seroconverted for anti-flagellin serum IgG after a single dose of vaccine, and all animals seroconverted after the second dose. No animals in the saline group seroconverted. However, there was no significant difference in the anti-flagellin titers between animals receiving saline and those receiving CVD 1926.

To determine the functional capacity of the anti-Salmonella antibodies, we measured their ability to promote uptake by macrophages. There was no uptake in prevaccination (t = −21) serum samples. Sera from animals receiving two doses of CVD 1926 was able to significantly increase S. Typhimurium macrophage uptake over unvaccinated animals (Fig. 2C).

T cell-mediated immunity (T-CMI) induced by vaccination was evaluated in peripheral blood mononuclear cells (PBMC) collected at days 0, 28, and 57 by flow cytometry. Similar to humans, CD4 and CD8 T cell compartments in rhesus macaques can be divided into subsets using the CCR7 and CD45RA markers (20) (see Fig. S1 in the supplemental material). Cutoffs to define positive responses for each cytokine/CD107a were established based on the background responses identified using the saline group (see Materials and Methods and Fig. S2 for examples). After two CVD 1926 vaccine doses, T-CMI was evident in CD8 T effector memory (T_EM), T_EM CD45RA⁺ (T_EMRA which are also part of the T effector compartment), and T central memory (T_CM) cells in 60, 80, and 40% of the animals, respectively (Table 1). These animals showed either upregulation of the cytolytic marker CD107a (21, 22) or the production of one or more cytokines. We evaluated whether cells upregulating one cytokine/CD107a (henceforth called function) were more abundant than those producing two, or more functions (multifunctional [MF] cells). MF analyses were focalized on PBMC from day 57, since this time point reflects the cytokine production potential of the animals right before challenge. All three CD8 T_EM responders showed a predominance of cells producing only one function (Fig. 3A). In two of these animals (14C143 and 14C148), the predominant marker upregulated was CD107a. In the other animal the predominant single marker was gamma interferon (IFN-γ). At day 57, no MF cells were present in the CD8 T_EMRA or T_CM compartments.

TABLE 1.

CD8 T cell-mediated immunity induced by CVD 1926^a

Animal	CD8 TEM (3/5 [60%])					CD8 TEMRA (4/5 [80%])					CD8 TCM (2/5 [40%])
	CD107a	IFN-γ	IL-2	IL-17A	TNF-α	CD107a	IFN-γ	IL-2	IL-17A	TNF-α	CD107a	IFN-γ	IL-2	IL-17A	TNF-α
14C143	+b	−	+c	+d	−	−	−	−	+c	−	+d	+d	+d	−	+d
14C148	+c	−	−	+c	−	−	+c	−	−	−	−	−	−	−	−
14C150	−e	−	−	−	−	−	−	+b	−	−	−	−	−	−	−
14C171	−	−	−	−	−	−	−	−	−	−	+d	+d	+b	−	+d
14C193	−	+b	+b	−	−	−	−	+b	−	−	−	−	−	−	−

^aT-CMI responses were assessed after vaccination in multiple CD8 memory T cell subsets, including T_EM (CCR7⁻ CD45RA⁻), T_EMRA (CCR7⁻ CD45RA⁺), and T_CM (CCR7⁺ CD45RA⁻) cells.

^bAfter first and second immunizations.

^cAfter second immunization.

^dAfter first immunization.

^e−, no change.

FIG 3.

Multifunctional (MF) T-CMI induced by CVD 1926. On day 57 (28 days after the second immunization and the day of challenge), the presence of CD8 and CD4 T cells able to produce more than one cytokine/CD107a was assessed. (A and B) NHPs with MF cells in CD8 and CD4 T_EM compartments, respectively. (C) Animals with MF cells in the CD4 T_EMRA compartment. (D) MF cells in the CD4 T_CM compartment. Black bars, single cytokine/CD107a upregulation; white bars, more than one cytokine/CD107a (MF) upregulated.

In the CD4 compartment, 60, 100, and 20% of the animals showed T-CMI to S. Typhimurium I77 in the T_EM, T_EMRA, and T_CM subsets, respectively, after two CVD 1926 doses (Table 2). In CD4 T_EM cells of NHP 14C143, single functionality (CD107a upregulation) was the predominant feature at day 57 (Fig. 3B). In contrast, animal 14C150 showed almost similar percentages of cells producing one, two, and three cytokines. In the same animal (14C150), CD4 T_EMRA cells produced predominantly interleukin-2 (IL-2; single cytokine) (Fig. 3C), which highlights the differences between cells from diverse T compartments, even in the same animal. Finally, CD4 T_CM cells in animal 14C143 had some MF capacity, but single functionality was predominant (Fig. 3D).

TABLE 2.

CD4 T cell-mediated immunity induced by CVD 1926^a

Animal	CD4 TEM (3/5 [60%])					CD4 TEMRA (5/5 [100%])					CD4 TCM (1/5 [20%])
	CD107a	IFN-γ	IL-2	IL-17A	TNF-α	CD107a	IFN-γ	IL-2	IL-17A	TNF-α	CD107a	IFN-γ	IL-2	IL-17A	TNF-α
14C143	+b	−	+d	+c	+d	+d	+b	−	−	−	+b	+b	+d	+c	−
14C148	−e	−	−	−	−	−	+c	−	−	−	−	−	−	−	−
14C150	−	+c	+b	−	+c	−	−	+b	+c	−	−	−	−	−	−
14C171	−	−	−	−	−	−	−	−	+d	−	−	−	−	−	−
14C193	−	+d	+d	+d	+d	−	−	−	+b	−	−	−	−	−	−

^aT-CMI responses were assessed after vaccination in multiple CD4 memory T cell subsets, including T_EM (CCR7⁻ CD45RA⁻), T_EMRA (CCR7⁻ CD45RA+), and T_CM (CCR7⁺ CD45RA⁻) cells.

^bAfter first and second immunizations.

^cAfter second immunization.

^dAfter first immunization.

^e−, no change.

Challenge of rhesus macaques with wild-type S. Typhimurium I77.

Rhesus macaques were challenged intragastrically with wild-type S. Typhimurium I77 on day 57 of the study. Animals were monitored twice daily and received intravenous rehydration and analgesics if they showed clinical signs of illness, such as diarrhea, dehydration, and lethargy. Two animals in the unvaccinated group were euthanized due to overwhelming infection on days 59 and 60 (days 2 and 3 postchallenge). Both animals showed signs of systemic infection, with high bacterial counts in the spleen, liver, intestines, colon, and mesenteric lymph nodes (MLN) at necropsy (Table 3). Intestinal tissues exhibited extensive histopathology, with severe erosive enteritis found in the ileum and acute colitis being present in both animals.

TABLE 3.

Bacterial counts in organs at necropsy^a

Group	Animal	Tissue (CFU/g)						Day of euthanasia
		Spleen	Liver	Terminal ileum	Proximal ileum	Colon	MLN
Saline	14C178	–	–	1,100	–	300	1,800	80
	14C187	–	–	–	–	100	–	78
	14C280	–	–	–	100	–	–	73
	14C301	5,100	600	3.25 × 105	6 × 106	4.75 × 105	2.13 × 104	60*
	14C117	600	700	1.13 × 106	6.75 × 106	7 × 106	9,250	59*
CVD 1926	14C143	–	–	–	–	–	–	74
	14C148	–	–	1,400	–	–	–	77
	14C150	–	–	–	–	–	400	79
	14C171	–	–	100	–	200	200	72
	14C193	–	–	–	–	–	2,700	81

^a–, not detected (limit of detection = 100 CFU/g). *, Early euthanasia following challenge.

Intestinal colonization was measured by determining bacterial counts in stool. All animals had high bacterial counts on day 1 postchallenge (experiment day 58, Fig. 4). Animals that developed clinical diarrhea had their counts increase over time, peaking at day 3 postchallenge (day 60). In contrast, animals that did not develop clinical symptoms had their counts decrease each day postchallenge. The only day in which there was a significant difference in bacterial counts between vaccinated and nonvaccinated individuals was day 6 postchallenge (day 63, P = 0.0006). By using area under the curve analysis, we determined that animals vaccinated with CVD 1926 had significantly less bacterial burden than unvaccinated animals (P = 0.029, one tailed).

FIG 4.

Bacterial shedding in stool after challenge with wild-type Salmonella Typhimurium strain I77. Salmonella Typhimurium I77 was detected in stool postchallenge by plating on Salmonella-Shigella selective media. The limit of detection (dashed line) was 100 CFU/g. *, P < 0.001; †, euthanized.

Diarrhea severity was the primary endpoint used for this gastroenteritis model. All unvaccinated animals had diarrhea following challenge (Table 4). The diarrhea severity for all unvaccinated animals peaked on days 2 to 3 postchallenge (days 59 to 60) and was classified as severe. Excluding the two animals that succumbed to illness, the remaining animals showed signs of moderate-to-severe diarrhea (MSD) for a total of 6 to 7 days. In contrast, only one of the five vaccinated animals exhibited signs of diarrhea. This animal, 14C150, had moderate diarrhea for 5 days total, but unlike unvaccinated animals, it did not exhibit signs of severe diarrhea. None of the other vaccinated animals showed any signs of abnormal stools. Collectively, unvaccinated animals had a significantly greater number of days with diarrhea compared to unvaccinated animals (P = 0.05). The overall vaccine efficacy for CVD 1926 was calculated using MSD as an endpoint. The efficacy of the vaccine in the rhesus macaque gastroenteritis model was determined to be 80% (P = 0.024, Fisher exact test, one tailed). Physiological measures of infection, including temperature and weight loss, were also obtained but no differences were observed between animals that received saline versus CVD 1926 (see the supplemental results and Fig. S3).

TABLE 4.

Stool grading of rhesus macaques challenged with Salmonella Typhimurium I77^a

Group	Animal	Day postchallenge (study day)
		0 (57)	1 (58)	2 (59)	3 (60)	4 (61)	5 (62)	6 (63)	7 (64)	8 (65)
Saline	14C178			+++	+++	++	++	++	++
	14C187			++	+++	++	++	++	++
	14C280			+++	+++	++	++	++	++	++
	14C301			++	+++*
	14C117			+++*
CVD 1926	14C143
	14C148
	14C150			++	++	++	++	++	+
	14C171
	14C193

^a+, loose stools; ++, moderate diarrhea; +++, severe diarrhea/dysentery. *, euthanized due to severe disease.

At necropsy, samples of the ileum, colon, spleen, liver, and MLN were taken to assess bacterial burden and inflammation. Of the animals that survived challenge, all three saline-vaccinated animals had S. Typhimurium counts in the gut (proximal ileum, terminal ileum, or colon; Table 3). Only two of the five CVD 1926-vaccinated animals had S. Typhimurium counts in the gut. For the mesenteric lymph nodes, bacteria were found in one of three saline-vaccinated animals versus three of five CVD 1926-vaccinated animals. No S. Typhimurium were detected in either the spleen or liver. Histopathological analysis of tissues taken at necropsy showed little or no pathology. In some animals, signs of recent infection such as reactive macrophages at the tips of intestinal villi were observed.

The T-CMI induced by challenge in CD8 and CD4 T_EM cells was studied in PBMC collected at euthanasia from animals that survived challenge (days 72 to 81) (see Tables S1 to S5 and Fig. S4 in the supplemental material). Responses were more robust in vaccinated animals compared to prechallenge.

DISCUSSION

Despite numerous S. Typhimurium live-attenuated vaccines being reported as being nonreactogenic in preclinical animal models, few have progressed into clinical trials. This may be due in part to results from an early clinical trial with live-attenuated S. Typhimurium strain WT05, in which the vaccine was shed for up to 23 days postimmunization (8). Solving this issue has been complicated from a preclinical standpoint due to the lack of an appropriate small animal model for gastroenteritis. To address this, we have utilized a rhesus macaque model of S. Typhimurium gastroenteritis, which more closely mimics human disease (13, 18). From our earlier experiment in rhesus macaques, strain CVD 1921 was shed for up to 10 days (13). This vaccine was well tolerated in both SIV-positive and SIV-negative rhesus macaques but did induce some adverse reactions in mice (9, 13, 14). By adding pipA and htrA mutations in strain CVD 1926, we were able to improve the safety of CVD 1926 in mice, while maintaining immunogenicity and vaccine efficacy (14).

In this study, we showed that vaccine CVD 1926 is well tolerated in rhesus macaques, with shedding postimmunization reduced to a mean of 5 days (range of 4 to 7 days). This is similar to that observed for two candidate live-attenuated S. Typhi vaccines tested in clinical trials, CVD 908-htrA and CVD 909, which were shed in feces of volunteers for a short window postimmunization (up to 4 days and 8 days, respectively) (23, 24). Shedding of CVD 1926 peaked on day 1, suggesting that most of the inoculum passed through the animal, without significant persistence or growth. This represents a significant improvement over the earlier vaccine candidates.

Animals vaccinated with CVD 1926 seroconverted for both S. Typhimurium anti-LPS and anti-flagellum IgG antibodies. These antibodies were also functional in promoting uptake of the S. Typhimurium challenge strain by murine macrophages. Although prevaccine anti-flagellum titers were elevated, these antibodies were not able to promote macrophage uptake. As the animals were all negative for S. Typhimurium anti-LPS antibodies, it is postulated that the measured anti-flagellum response was against flagellum epitopes shared by other Salmonella species.

While the vaccine was able to induce a functional antibody response in all animals, this did not translate to protection of all animals against diarrhea. Rhesus macaque 14C150 produced the highest functional antibody titer and yet succumbed to moderate diarrheal disease of only slightly less severity than unvaccinated animals. While we were unable to identify any other human or nonhuman primate S. Typhimurium studies to put this result in context, this is consistent with data from the S. Typhi controlled human infection model (CHIM), in which protection was not correlated with the functional ability of antibodies to mediate serum bactericidal activity, although there was some decrease in severity associated with increased antibody titer (25). This observation was confirmed in a recent paper by Jin et al. (26), which found that antibody-dependent complement deposition induced by Vi-polysaccharide and Vi-tetanus toxoid conjugate vaccines was not correlated with protection against S. Typhi. In this CHIM model, antibody-dependent neutrophil and monocyte phagocytosis were broadly associated with protection against symptomatic infection; however, the multifunctionality of the innate immune cell response was also highlighted as important. Whether these findings for S. Typhi are generalizable to NTS will require further investigation.

Due to the intracellular nature of Salmonella spp., T-CMI is expected to play an important role in protection, as well as in the resolution of disease. Defined T-CMI responses have been associated with disease outcome after challenge with S. Typhi in humans, suggesting a major role for T cells in protection induced by live-attenuated Salmonella vaccines (27, 28). T_CM and T_EM represent different stages in the differentiation of naive cells into memory cells. T_CM cells are the first stage of memory differentiation, and these cells can become T_EM cells (20, 29). T_EM cells express integrins and chemokine receptors that facilitate localization to inflamed tissues and thus are likely involved in protection at the local site (20). In humans, T_EMRA represent the final differentiation stage of T_EM cells (hence, these are considered effector memory cells, too); however, whether this also applies to T_EMRA cells in rhesus macaques is still unclear (29, 30). In sum, after two vaccine doses of CVD 1926, T-CMI responses were identified in all animals. These responses were identified in the various memory compartments and might represent diverse stages of the development of memory T-CMI, which most likely is individual dependent.

In an effort to identify a potential correlate of protection, we compared the immune responses elicited in the protected animals versus the unprotected animal (see Table S6). Interestingly, the single unprotected animal only possessed CD8 cells capable of releasing IL-2 following immunization but none of the other cytokines examined. In contrast, CD8 cells from all of the protected animals released at least two cytokines (either as single functional or multifunctional cells). None of the other immune responses examined correlated with protection. Due to small animal numbers, it is difficult to draw conclusions from these data, and further work will be required to tease out the relative importance of different arms of the immune system in the response to S. Typhimurium infection. However, our data show that CVD 1926 is able to induce T-CMI in CD4 and CD8 T cells, and these responses, most likely in conjunction with humoral immunity, protected animals from challenge.

The benefit of live-attenuated vaccines over subunit or conjugate vaccines is their ability to induce robust mucosal immune responses. In this study, we were unable to measure mucosal responses due to technical limitations in using the rhesus macaque model. Profiling the immune response to immunization within the intestinal mucosa should be a priority for future studies, both to understand how the mucosa responds to immunization and challenge and to identify putative correlates of protection. In the absence of a modern experimental human challenge model for NTS, having a thorough understanding of what constitutes an effective immune response to vaccination in this model will be invaluable for moving the CVD 1926 vaccine forward to licensure.

To conclude, S. Typhimurium strain CVD 1926 is safe and protective against gastroenteritis in a rhesus macaque model, with a vaccine efficacy of 80%. It is also capable of inducing both antibody and T-CMI. This vaccine strain could be used in combination with other Salmonella live-attenuated vaccines to combat antibiotic resistance and in populations with increased risk of severe disease, such as the elderly.

MATERIALS AND METHODS

Animal ethics statements.

Rhesus macaques of Indian origin (Macaca mulatta; 2 to 3 years old; females) were purchased from Covance Research Products, Inc. (Denver, PA). Animals were prescreened for anti-Salmonella LPS serum IgG antibodies, and only seronegative animals were used in the study. The study was carried out in the animal facility of the Program of Comparative Medicine at University of Maryland School of Medicine, which is fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International. All procedures were approved by the Institutional Animal Care and Use Committee, conformed to guidelines in the Guide for the Care and Use of Laboratory Animals and the Animal Welfare Act, and were fully compliant with recommendations in the Biosafety in Microbiological and Biomedical Laboratories Guide. In-depth descriptions of animal care and technical procedures can be found in Ramachandran et al. (18).

Immunization of rhesus macaques with CVD 1926.

Live-attenuated S. Typhimurium vaccine strain CVD 1926 (ΔguaBA ΔclpP ΔpipA ΔhtrA) was prepared fresh for each immunization by harvesting bacterial growth from HS plates grown overnight and resuspended in phosphate-buffered saline (PBS) at a concentration of 6 × 10⁹ CFU/ml. Rhesus macaques were immunized intragastrically (i.g.) on days 0 and 28 of the study. Animals were given either 1 ml of saline (animals 14C117, 14C187, 14C280, 14C301, and 14C178) or 1 ml of bacterial suspension (6 × 10⁹ CFU; animals 14C150, 14C171, 14C193, 14C148, and 14C143) in 14 ml of 8.4% sodium bicarbonate. To determine vaccine shedding, fecal samples were collected on days 0 to 7, 9, 11, 14, 21, 24, 28, 29, 35, 38, 42, and 49. Serum samples were collected on days −21, 7, 21, 35, and 49 to assess serum antibody levels. As detailed below, blood specimens for T cell studies were collected on days 0, 28, and 57 and at the day of euthanasia (days 72 to 81).

Quantitative analysis of S. Typhimurium in stool.

The S. Typhimurium burden in stool was quantified by colony counts. Stool samples were collected from each animal, and then 1 g of stool was resuspended in 9 ml of sterile PBS. Samples were serially diluted in PBS, and 100-μl aliquots were plated on Salmonella-Shigella (SS) agar plates (BD, Sparks MD).

Assessment of antibody responses.

Serum anti-Salmonella core and O polysaccharide (COPS) and FliC IgG titers were measured by enzyme-linked immunosorbent assay (ELISA) as described previously (9). Opsonophagocytic uptake by J774 macrophages was assessed as described by Tennant et al. (9). Details of each assay are shown in the supplemental methods.

Challenge with S. Typhimurium I77.

Rhesus macaques were infected i.g. on day 57 with 9 × 10⁹ CFU of wild-type S. Typhimurium strain I77, employing the same method used for the immunization. Fecal samples were collected and scored for stool consistency (–, normal; +, loose stools; ++, moderate diarrhea; +++, severe diarrhea/dysentery) on days 57 to 63, 65, 67, 70, and 73 and at necropsy. Animals were weighed on days 57 and 60, as well as at necropsy, and at any time they were sedated for rehydration. Blood was collected for blood chemistry on day 60. Rhesus macaques were monitored daily for clinical signs of diarrhea, lethargy, dysentery, weight loss, and fever. Animals were treated with 10 to 20 ml/kg of lactated Ringer’s solution given every 12 h if they showed signs of lethargy or dehydration, until clinical signs resolved. Animals meeting alternative endpoints (i.e., displaying more than 20% body weight loss compared to body weight at the beginning of the study, animals not responding to rehydration and/or pain relief treatment after 48 to 72 h of starting treatment, and/or animals displaying hunched posture or lack of movement when stimulated in the cage) were euthanized with sodium pentobarbital (100 mg/kg) delivered intravenously. Animals surviving bacterial challenge were euthanized between days 72 and 81.

Quantitative analysis of S. Typhimurium in organs at necropsy.

Samples of spleen, liver, ileum, colon, and mesenteric lymph nodes were collected at necropsy and homogenized using a tissue homogenizer (Omni International, Kennesaw, GA). Homogenized samples were plated onto SS agar to determine bacterial burden in organs.

Histopathological analysis of tissues by hematoxylin-eosin staining.

Tissue samples from the ileum, colon, spleen, liver, and mesenteric lymph nodes were collected from each animal at necropsy. Samples were formalin fixed, paraffin embedded, and stained with hematoxylin and eosin to determine gross inflammation, polymorphonuclear leukocyte infiltration, and epithelial barrier integrity. Slides were scored for inflammation by a pathologist who was blinded to the sample’s identity.

Isolation of peripheral blood mononuclear cells.

PBMC were isolated immediately after blood draws by density gradient centrifugation and cryopreserved in liquid nitrogen as previously described (31, 32). For T cell studies, blood specimens were collected on days 0, 28, and 57. A final blood sample was collected from surviving animals the day of euthanasia (days 72 to 81 of the experiment).

Ex vivo stimulation of effector cells and flow cytometry staining.

PBMC from days 0, 28, 57, and 72 to 81 were thawed and rested overnight. One million cells from each time point were stimulated with (i) media, (ii) Staphylococcus enterotoxin B (SEB; 5 μg/ml), or (iii) S. Typhimurium I77 cell lysate (5 μg/ml) for 16 to 18 h. Cells were also incubated for 2 h in the presence of anti-CD107a-FITC (clone H4A3; BD Biosciences) before overnight incubation with the protein transport blockers monensin (1 μg/ml; Sigma) and brefeldin A (2 μg/ml; Sigma). The next day, PBMC were harvested, washed in PBS, stained extracellularly, permeabilized, and stained intracellularly, using a 13-color panel (listed in the supplemental methods) as previously described (32). Cell viability was assessed using a Violet Live/Dead viability kit (Invitrogen, Carlsbad, CA). Stained cells were fixed with 1% paraformaldehyde in PBS. Samples were acquired using a customized LSRII flow cytometer (BD Biosciences) and analyzed using FlowJo v10 (FlowJo, LLC, San Francisco, CA). Responses against S. Typhimurium I77 were expressed as a net percentage of positive cells (i.e., total percentage of positive cells in the presence of S. Typhimurium-stimulated cells minus percentage of positive cells in cultures with media only). For each cytokine and CD107a, a cutoff to determine positive responses was included. The cutoff was the highest value for the cytokine/CD107a identified in the saline group at any time point postvaccination (for examples, see Fig. S2 in the supplemental material). By defining a cutoff that used the background responses in the saline group, we were able to identify the “true” responders in the vaccine group. Multifunctional (MF) cells were analyzed using the Boolean gate function in FlowJo and involved activated cells (CD69⁺) that produced cytokines and/or upregulated CD107a within each of the T memory subsets. In Fig. S5, examples of upregulation of cells producing more than one cytokine/CD107a after immunization are displayed from a representative animal. Note that the displayed data do not include CD69, since a third dimension is not possible in two-dimensional plots, and are therefore used only to demonstrate the presence of MF cells and frequency changes over time in a simple visualization format.

Statistical analysis.

Statistical analyses were carried out using GraphPad Prism Software. A Student t test was used to compare ELISA titers and stool counts. Vaccine efficacy and the number of days with diarrhea were analyzed using the Fisher exact test. For all analyses, a P value of ≤0.05 was considered significant.