A protein-free vaccine stimulates innate immunity and protects against nosocomial pathogens

Abstract

Traditional vaccines are difficult to deploy against the diverse antimicrobial-resistant, nosocomial pathogens that cause health care–associated infections. We developed a protein-free vaccine composed of aluminum hydroxide, monophosphoryl lipid A, and fungal mannan that improved survival and reduced bacterial burden of mice with invasive blood or lung infections caused by methicillin-resistant Staphylococcus aureus, vancomycinresistant Enterococcus faecalis, extended-spectrum beta-lactamase–expressing Escherichia coli, and carbapenem-resistant strains of Acinetobacter baumannii, Klebsiella pneumoniae, and Pseudomonas aeruginosa. The vaccine also conferred protection against the fungi Rhizopus delemar and Candida albicans. Efficacy was apparent by 24 hours and lasted for up to 28 days after a single vaccine dose, with a second dose restoring efficacy. The vaccine acted through stimulation of the innate, rather than the adaptive, immune system, as demonstrated by efficacy in the absence of lymphocytes that were abrogated by macrophage depletion. A role for macrophages was further supported by the finding that vaccination induced macrophage epigenetic alterations that modulated phagocytosis and the inflammatory response to infection. Together, these data show that this protein-free vaccine is a promising strategy to prevent deadly antimicrobial-resistant health care–associated infections.

INTRODUCTION

Every year, health care–associated infections (HAIs) afflict more than 722,000 people, result in more than 90,000 deaths, and generate a financial burden between $28 billion and $45 billion in the United States alone (1, 2). On any given day, more than 3% of hospitalized patients in the United States (3), and 4% in the European Union (4), have an HAI. In 2022, the World Health Organization reported that 7% of all hospital admissions in high-income countries, and 15% in low- and middle-income countries, will develop a HAI, including up to 30% of patients in intensive care units (5). In most cases, HAIs are caused by antimicrobial-resistant (AMR) bacterial and fungal pathogens, which are associated with worse mortality and morbidity than antimicrobial-susceptible pathogens (6). Despite the high incidence of HAIs, there are currently no U.S. Food and Drug Administration–approved vaccines against the most commonly encountered and antibiotic-resistant pathogens, such as methicillin-resistant Staphylococcus aureus (MRSA), vancomycinresistant Enterococcus (VRE), carbapenem-resistant Acinetobacter baumannii, extended-spectrum beta-lactamase (ESBL)–expressing or carbapenem-resistant Enterobacterales, and carbapenem-resistant Pseudomonas aeruginosa.

Broadly speaking, there are two approaches to preventing infections: horizontal and vertical (7). Vertical approaches to infection prevention, such as traditional protein-based vaccines, target one pathogen at a time. The challenge of implementing such an approach in the hospital setting is that multiple vaccines would have to be deployed simultaneously to protect against the myriad AMR pathogens that cause HAIs. In contrast, horizontal approaches protect against a wide range of pathogens from one intervention; examples include terminal disinfecting of hospital rooms, hand washing, and personal protective equipment. Although the merits of both approaches are often debated, horizontal approaches have been shown to have a broader impact at lower costs for HAIs (8).

While developing a vaccine to protect against MRSA infections, we found that combining the adjuvant aluminum hydroxide [Al(OH)₃] with the protein antigen α-hemolysin markedly decreased the severity of skin and soft-tissue infections and dermonecrosis in mice, yet it offered no protection against bacteremia (9). We subsequently sought to enhance the immunogenicity of the vaccine by incorporating additional adjuvants into the vaccine. A specific combination of Al(OH)₃, monophosphoryl lipid A (MPL), and whole glucan particles (WGP) led to enhanced survival in our murine bacteremia sepsis model but did so in a manner independent of protein antigens. We were intrigued by the possibility of a multivalent, protein-free vaccine that protects against infection. With no pathogen-specific components, we hypothesized that the vaccine could potentially prevent a variety of nosocomial pathogens across taxonomic kingdoms, including Gram-positive bacteria, Gram-negative bacteria, and fungi. We also hypothesized that this broad-spectrum protein-free vaccine could improve survival by modulating the innate immune response through a phenomenon known as trained immunity: the ability of the innate immune system to mount a more effective host response after exposure to prior infection or vaccination.

RESULTS

A combination of Al(OH)₃, MPL, and WGP provided protection against multiple AMR pathogens in mice

We first explored the benefit of the individual components of the tripartite vaccine against S. aureus bacteremia. We immunized mice with each individual component or dual or triple combinations of Al(OH)₃, MPL, and WGP. All mice were then challenged intravenously (iv) with a USA300 clinical blood isolate of MRSA, LAC. The triple combination of Al(OH)₃, MPL, and WGP (A + M + W) was the regimen that mediated the most improvement in survival time compared with the Al(OH)₃-only control (fig. S1A).

We next sought to evaluate the duration of protection from the initial triple vaccine and the potential for a second dose to add additional benefit. We immunized mice with phosphate-buffered saline (PBS) or A + M + W, and 3 weeks later, half of the mice were immunized again. Mice were then challenged intravenously with S. aureus LAC at 3, 7, or 21 days after the final immunization. For the single-dose recipients, the vaccine improved survival time at 3 (log-rank, P = 0.02) and 7 days (log-rank, P = 0.05) after the dose but not at 21 days after immunization (fig. S1B). The second dose restored the vaccine’s improvement in survival time for a further 3 days (log-rank, P = 0.02) but not a further 7 or 21 days (fig. S1C).

Because the immunostimulatory components of the tripartite vaccine were not specific to S. aureus, we investigated the ability of the vaccine to protect against bacteremia caused by the Gram-negative bacterium A. baumannii. We immunized mice with PBS or A + M + W and infected them intravenously either 3 or 7 days after immunization with an extremely drug-resistant (XDR), clinical lung and blood isolate of A. baumannii, HUMC1 (Fig. 1A). Groups immunized with A + M + W were fully protected against otherwise lethal bacteremia (log-rank, P = 0.0009) (Fig. 1B).

Fig. 1. A protein-free vaccine provides short-term protection against lethal bloodstream infections.

(A) Shown are the immunization timelines used for this figure. (B) Male C3HeB/Fe mice (N = 6 per group) were immunized and 3 or 7 days later infected intravenously with 1.7 × 10⁷ CFU XDR A. baumannii HUMC1. (C) Male C3HeB/Fe mice (N = 6 per group) were immunized with the indicated combinations and 3 days later infected intravenously with 2.5 × 10⁷ CFU XDR A. baumannii HUMC1. (D) Male C3HeB/Fe mice (N = 8 per group) were immunized and 3 days later infected by oropharyngeal aspiration (OA) with 1.5 × 10⁸ CFU XDR A. baumannii HUMC1 as a pneumonia model. (E) Male C57BL/6 mice (N = 6 per group) were immunized and 3 or 7 days later infected intravenously with 3.8 × 10⁸ CFU K. pneumoniae KP3. (F) Female BALB/c mice (N = 10 per group) were immunized with the previous dose of vaccine or with doses 3- or 10-fold lower (33 or 10%), and 3 days later infected intravenously with 8.4 × 10⁷ CFU MRSA LAC. (G) Male C3HeB/Fe mice (N = 5 per group) were immunized with the previous dose of vaccine or with doses 3- or 10-fold lower (33 or 10%), and 3 days later infected intravenously with 4.6 × 10⁷ CFU XDR A. baumannii HUMC1. Differences in survival were determined by log-rank test (α = 0.05). *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001 versus PBS; n.s., not significant.

Similar to S. aureus, we tested whether all three components were necessary to protect against A. baumannii bacteremia. We immunized mice with zero, one, two, or all three components and challenged them 3 days later with a larger inoculum of A. baumannii, enabling differentiation between the survival benefits conferred on each immunization group. Compared with other groups, A + M + W provided the greatest improvement in survival time (log-rank, P = 0.0001) (Fig. 1C).

We next considered whether the tripartite vaccine could improve the survival of mice infected through another common route of infection in hospitalized patients—pneumonia—using our previously described oropharyngeal aspiration (OA) pneumonia model (10). Despite protecting against bacteremia, A + M + W did not protect against A. baumannii pneumonia infection (Fig. 1D). We also assessed the ability of the vaccine to protect against lethal bloodstream infection from Klebsiella pneumoniae, another Gram-negative bacterium. Although no mice survived, immunized groups experienced significantly delayed mortality (log-rank, P = 0.001) (Fig. 1E). Last, we explored the effect of lower doses against S. aureus and A. baumannii infection and found dose-dependent reductions in efficacy (Fig. 1, F and G).

Replacement of whole glucan particles with mannan improved efficacy and conferred protection against pneumonia

On the basis of prior experience working with a mannosylated protein vaccine (11), we sought to enhance vaccine efficacy by incorporating fungal mannan, an oligosaccharide shown to have immunomodulating properties. Mice were immunized with either a quadruple regimen in which mannan was added to A + M + W (A + M + W + MA) or a triple regimen with mannan replacing WGP (A + M + MA). Three days later, mice were challenged with a normally lethal inoculum of MRSA, carbapenem-resistant A. baumannii, carbapenem-resistant K. pneumoniae, ESBL-expressing Escherichia coli, Rhizopus delemar (in neutropenic mice), or Candida albicans. Overall, the vaccine in which mannan substituted for WGP (A + M + MA) mediated significantly improved survival time against lethal bloodstream infection caused by all of the pathogens, including the neutropenic mouse model of R. delemar mucormycosis (log-rank, P < 0.05 versus PBS for all pathogens) (Fig. 2, A to G). Furthermore, in contrast to the vaccine with WGP (Fig. 1D), which did not protect against A. baumannii pneumonia, inclusion of mannan helped confer a marked survival advantage in mice infected with A. baumannii pneumonia (log-rank, P = 0.003 versus PBS) (Fig. 2H) and also improved survival time in mice with carbapenem-resistant P. aeruginosa pneumonia (Fig. 2I). The quadruple vaccine was either no more effective than or less effective than the A + M + MA triple vaccine when compared head to head.

Fig. 2. Replacing whole glucan particles in the protein-free vaccine with mannan conferred superior protection.

(A) Shown is the immunization timeline used for this figure. Mice were immunized with PBS, A + M + MA, or A + M + W + MA. Three days later, mice were infected as indicated, and survival (B to I) or bacterial burden (J) was measured. (B) Female BALB/c mice (N = 8 to 16 per group) were infected intravenously with 5.1 × 10⁷ to 6.9 × 10⁷ CFU MRSA LAC. (C) Male C3HeB/Fe mice (N = 5 per group) were infected intravenously with 1.4 × 10⁷ to 2.9 × 10⁷ CFU XDR A. baumannii HUM1. (D) Male C57BL/6 mice (N = 5 per group) were infected intravenously with 2.4 × 10⁸ CFU carbapenem-resistant K. pneumoniae KPC-KP1. (E) Male C57BL/6 mice (N = 5 to 7 per group) were infected intravenously with 7.8 × 10⁷ to 9 × 10⁷ CFU ESBL E. coli JJ1886. (F) Neutrophil-depleted (−NΦ) female BALB/c mice (N = 5 per group) were infected intravenously with 3.0 × 10³ CFU R. delemar 99–880. (G) Male BALB/c mice (N = 5 per group) were infected intravenously with 1.7 × 10⁵ C. albicans. (H) Male C3HeB/Fe mice (N = 5 per group) were infected by OA with 1.6 × 10⁸ CFU XDR A. baumannii HUMC1. (I) Male C3HeB/Fe mice (N = 7 or 8 per group) were infected by OA with 5.7 × 10⁵ to 1.0 × 10⁶ CFU XDR P. aeruginosa PA9019. (J) Female C3HeB/Fe mice (N = 6 per group) were infected intravenously with 1.6 × 10⁸ to 2.8 × 10⁸ CFU VRE 51299; blood bacterial burden was analyzed 1 hour later. Survival was compared by the log-rank test (α = 0.05). Bacterial burden was compared by the Wilcoxon-Mann-Whitney test (α = 0.05). *P ≤ 0.05 and **P ≤ 0.01 versus PBS. Data in (J) are presented as median with interquartile range (IQR).

The efficacy of the A + M + MAvaccine was also tested by a team of scientists at another organization, which independently prepared and immunized BALB/c mice and challenged them intranasally with invasive P. aeruginosa 6294 or cytotoxic P. aeruginosa 6077 or intravenously with MRSA SF8300 5 days after immunization. All models demonstrated significant improvement in survival time in vaccinated mice as compared with control mice 5 days after immunization (log-rank, P < 0.001 P. aeruginosa 6294, P < 0.001 P. aeruginosa 6077, P = 0.005 S. aureus) (fig. S2, A to C).

Last, we tested the vaccine against VRE. Because VRE tends to be of lower virulence, an unrealistically high inoculum, greater than 5 × 10⁸ colony-forming units (CFU), is required to achieve a lethal dose in immunocompetent mice. We therefore evaluated bacterial blood density rather than survival. Mice immunized with A + M + MA had a significantly lower blood bacterial burden than control mice 1 hour after infection (Mann-Whitney test, P = 0.04) (Fig. 2J).

Vaccine efficacy lasted up to 28 days at higher doses

Having established A + M + MA as the lead formulation, we sought to evaluate the duration of the protective effect and whether increased doses could improve survival or duration of efficacy. We immunized mice with 3× or 10× as much mannan and MPL and challenged them 1, 2, 3, or 4 weeks later with a lethal inoculum of P. aeruginosa, S. aureus, or A. baumannii. For P. aeruginosa infection, 10× enabled persistent efficacy for 28 days (log-rank, P = 0.04) (fig. S2D). For S. aureus infection, dose increase to either 3× or 10× enabled persistent efficacy for 21 days (Fig. 3A) but not 28 days after immunization (fig. S2E). For A. baumannii, only the largest dose (10×) provided survival benefit when mice were challenged 2 or 3 weeks after immunization (Fig. 3B).

Fig. 3. Larger vaccine doses extended the duration of protection.

(A) Female BALB/c mice (N = 8 per group) were immunized with PBS; A + M + MA 1× [0.1% Al(OH)₃, 10 μg of MPL, and 100 μg of mannan]; A + M + MA 3× [0.1% Al(OH)₃, 30 μg of MPL, and 300 μg of mannan]; or A + M + MA 10× [0.1% Al(OH)₃, 100 μg of MPL, and 1000 μg of mannan]. Seven, 14, or 21 days later, mice were infected intravenously with 9.7 × 10⁷ CFU MRSA LAC, and survival was measured. (B) Male C3HeB/Fe mice (N = 5 per group) were immunized as in (A). Seven, 14, or 21 days later, mice were infected intravenously with 2.9 × 10⁷ CFU XDR A. baumannii HUMC1, and survival was measured. (C) Female BALB/c mice (N = 10 per group) were immunized with PBS or A + M + MA. One day later, mice were infected intravenously with 1.2 × 10⁸ CFU MRSA LAC, and survival was measured. (D) Male C3HeB/Fe mice (N = 5 per group) were immunized as in (C). One day later, mice were infected intravenously with 2.8 × 10⁷ CFU XDR A. baumannii HUMC1, and survival was measured. (E) Female NSG mice with human CD34⁺ hematopoietic stem cells (N = 4 per group) were immunized as in (C). Three days later, mice were infected intravenously with 1.4 × 10⁷ CFU XDR A. baumannii HUMC1, and survival was measured. (F) Female BALB/c mice (N = 8 per group) were immunized with PBS, A + M + MA, or GMP/GLP grade A + M + MA. Three days later, mice were infected intravenously with 3 × 10⁸ CFU MRSA LAC, and survival was measured. (G) Male C3HeB/Fe mice (N = 5 per group) were immunized as in (F). Three days later, mice were infected intravenously with 2.4 × 10⁷ CFU XDR A. baumannii HUMC1, and survival was measured. (H) Male C3HeB/Fe mice (N = 3 per group) were immunized with PBS or premixed GMP/GLP grade A + M + MA stored at room temperature (RT) or 4°C for 3 or 8 months. Three days later, mice were infected intravenously with 1.5 × 10⁷ CFU XDR A. baumannii HUMC1, and survival was measured. Survival was compared by the log-rank test (α = 0.05). *P ≤ 0.05 and **P ≤ 0.01 versus PBS.

We next sought to define how rapidly vaccine-mediated efficacy began. We immunized mice with the original (1×) dose of A + M + MA and challenged them with a lethal inoculum of S. aureus or A. baumannii 24 hours later, resulting in improved protection as compared with controls (log-rank, P = 0.005 S. aureus and P = 0.003 A. baumannii) (Fig. 3, C and D). This was similar to results seen during previous studies where mice were infected 3 days (Fig. 2, B and C) after immunization.

To assess its translatability to humans, we immunized mice that had undergone whole-body irradiation followed by CD34⁺ human stem cell transplantation with A + M + MA 3 days before infection (12, 13). Despite being severely immunocompromised, the vaccine still increased the survival time considerably against A. baumannii (log-rank, P = 0.008) (Fig. 3E). To support potential future clinical development, we also tested Good Manufacturing Practice (GMP)–compliant Al(OH)₃ and MPL and Good Laboratory Practice (GLP)–compliant mannan by immunizing mice with GMP/GLP-grade A + M + MA (G-A + M + MA) 3 days before infection. The GMP/GLP-grade material provided the same survival benefit as the non-GMP/GLP-grade material (log-rank, P = 0.004 S. aureus and P = 0.003 A. baumannii) (Fig. 3, F to G). To further aid future clinical translation, we also tested the stability of the vaccine that had been prepared in advance and stored at ambient room temperature or 4°C for 3 months. In both storage conditions, GMP/GLP-grade material retained the same survival benefit as the freshly prepared GMP/GLP-grade material (log-rank, P = 0.01) (Fig. 3H).

Macrophages were required for vaccine-mediated protection

To begin to define the mechanism of protection, we analyzed the plasma of vaccinated mice and confirmed that immunizations did not result in antibodies specific to A. baumannii or S. aureus (table S1). We then evaluated changes over a 21-day period after immunization in eight immune cell populations sourced from the spleens: T cells, B cells, dendritic cells, natural killer (NK) cells, myeloid cells, neutrophils, monocytes, and macrophages. We also conducted complete blood counts with differentials in immunized and non-immunized mice.

Spleen analysis demonstrated no change in the proportion of CD45⁺ cells that were T cells (figs. S3 and S4). Conversely, the proportion of CD45⁺ cells that were B cells was elevated by the first day after immunization and persisted for at least 3 weeks (fig. S3). Dendritic cell and NK cell populations decreased proportionately by the first day after immunization as a proportion of CD45⁺ cells, beginning to increase by 3 to 7 days after immunization (fig. S3). No trend was observed for the change in the proportion of CD45⁺ cells that were of the myeloid lineage (fig. S3). However, by the first day after immunization, there was a change in the proportion of myeloid cell types, including neutrophils, monocytes, and macrophages (fig. S3). Specifically, the proportion of neutrophils decreased, remaining below baseline for at least 3 weeks, as the proportion of monocytes and macrophages increased (fig. S3).

Complete blood count with differential in vaccinated versus control mice demonstrated no change in the number of total white blood cells (WBCs), platelet count, total red blood cells, or lymphocytes over time (fig. S5). Although the absolute number of lymphocytes did not change, the percentage of lymphocytes decreased by the first day after immunization, which lasted until 3 days after immunization, suggesting an increase in circulating myeloid cells shortly after immunization. Both the neutrophil cell count and the percentage of WBCs increased during the same period after immunization. Monocyte numbers and percentages noticeably increased and remained above baseline up to 21 days after immunization (fig. S5).

The lack of antibody-mediated immunity (table S1) and the rapid induction of protective immunity (Fig. 3, C and D) suggested that the vaccine provided benefit through the innate immune system, rather than through adaptive immunity. We thus hypothesized that only the innate immune system was necessary to provide survival benefit. Recombination activating gene 1 (RAG1) is critical for V(D)J recombination, and mice with nonfunctional RAG1 lack mature B and T lymphocytes, consistent with severe combined immunodeficiency disorders (14–17). To determine whether lymphocytes played a role in protection mediated by A + M + MA, we immunized wild-type and RAG1-knockout (RAG1-KO) mice with A + M + MA and challenged them with A. baumannii 3 days after immunization. Even in the absence of mature B and T lymphocytes in the RAG1-KO mice, the A + M + MA vaccine successfully improved survival time during an otherwise lethal infection (log-rank, P = 0.0009 RAG1-KO, PBS versus RAG1-KO, A + M + MA; Fig. 4A).

Fig. 4. Monocytes and macrophages are key effectors of A + M + MA-mediated protection.

(A) Male C57BL/6 wild-type mice (N = 5 per group) and RAG1-KO mice (N = 6 per group) were immunized with A + M + MA and infected intravenously with 3.9 × 10⁷ to 7.8 × 10⁷ CFU XDR A. baumannii HUMC1. Survival was measured after infection. (B to E) Male C3HeB/Fe mice (N = 5 per group) and female BALB/c mice (N = 8 per group) were depleted of natural killer (NK) cells (B and C) or macrophages and monocytes (MΦ) (D and E). Mice were then immunized with A + M + MA and infected intravenously with 2.3 × 10⁷ to 2.7 × 10⁷ CFU XDR A. baumannii (B and D) HUMC1 or 1.4 × 10⁸ to 3.0 × 10⁸ CFU MRSA LAC (C and E). (F and G) Primary human monocytes (F) and RAW 264.7 macrophages (G) were stimulated with A + M + MA or IFN-γ for 3 days and evaluated for their ability to take up A. baumannii ATCC17978. (H) Primary human monocytes were stimulated with A + M + MA for 1 day, rested for 2 or 5 days, and evaluated for their ability to take up A. baumannii ATCC17978. Naïve macrophages without A + M + MA stimulation were used as a negative control. Data in (F) to (H) are presented as median with IQR. Survival was compared by log-rank test, and all comparisons were made versus the PBS group (α = 0.05). Macrophage uptake was compared by the Wilcoxon rank sum test (α = 0.05). *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001.

We therefore hypothesized that innate immune cells were essential in vaccine-mediated efficacy. To identify the innate immune cell type(s) essential in A + M + MA–mediated efficacy, we immunized mice with A + M + MA and selectively depleted mice of their NK cells or monocytes and macrophages, because studies have shown that NK cells, monocytes, and macrophages are key to inducing trained immunity (18–20). Depletion of NK cells partially ablated the protective efficacy of the vaccine against S. aureus but not A. baumannii (log-rank, P = 0.003) (Fig. 4, B and C). However, the depletion of monocytes and macrophages completely ablated vaccine-mediated efficacy against both S. aureus and A. baumannii infection (Fig. 4, D and E).

We then investigated whether A + M + MA could directly enhance macrophage phagocytosis. Ex vivo human macrophages differentiated from freshly harvested peripheral blood mononuclear cells (PBMCs) or murine macrophage-like RAW 264.7 cells exposed to A + M + MA for 3 days took up significantly more A. baumannii compared with naïve macrophages not exposed to A + M + MA (Kruskal-Wallis, P < 0.0001 for human macrophages and murine RAW 264.7 cells; Fig. 4, F and G). The human macrophage phagocytosis experiment was then repeated by stimulating with a vaccine for 1 day, followed by a 5-day rest period (6 days total) before bacterial coculture; vaccine stimulation again resulted in a significant increase in phagocytosis of bacteria (Kruskal-Wallis, P < 0.0001; Fig. 4H).

The vaccine induced a net anti-inflammatory cytokine profile

Having established the importance of monocytes and macrophages, we investigated their immunomodulatory properties by evaluating cytokine-related gene expression, bacterial burden, and cytokine profiles in mice immunized with A + M + MA before and after being infected with S. aureus or A. baumannii. Initially, we attempted to analyze cytokine concentrations in the plasma of immunized, uninfected mice. However, the cytokines evaluated, including interferon-γ (IFN-γ), tumor necrosis factor (TNF), interleukin-1β (IL-1β), IL-4, IL-6, IL-10, IL-12p70, IL-13, and IL-17A, were below the limit of detection (table S2). We then assessed cytokine-associated gene expression in macrophages from spleens excised from vaccinated or control uninfected mice and found that, at 3 days after immunization, macrophages had lower expression of the genes encoding the pro-inflammatory cytokines Il12a (Kruskal-Wallis, P = 0.02), Il17rb (Kruskal-Wallis, P = 0.03), and Tnfrsf13c (Kruskal-Wallis, P = 0.02), as well as higher expression of the anti-inflammatory cytokine-associated genes Il10 (Kruskal-Wallis, P = 0.02) and Il18bp (Kruskal-Wallis, P = 0.03) (Fig. 5A). Expression of genes encoding Il1ɑ and Il1r2 were not significantly different (Kruskal-Wallis, P > 0.05; Fig. 5A).

Fig. 5. Pro-inflammatory cytokine-encoding genes are expressed at lower abundance in mice immunized with A + M+ MA.

Mice were immunized with A + M + MA 1, 3, 7, 14, or 21 days before infection. Naïve mice (zero days after immunization) were included as a control. Gene expression was measured by RNA-seq. (A) Cytokine-related gene expression changes are shown in mouse splenic macrophages (N = 3). (B and C) Female BALB/c mice (N = 5 per group) and male C3HeB/Fe mice (N = 5) were infected intravenously through the tail vein with 1.5 × 10⁸ CFU MRSA LAC (B) or 1.8 × 10⁷ CFU XDR A. baumannii HUMC1 (C), respectively. Plasma from each mouse was then analyzed by Luminex for cytokines: IL-4, IL-6, IL-10, IL-12p70, and TNF. Dashed lines connect median values. The ratio between IL-10 and TNF is also shown. Data were analyzed by the Kruskal-Wallis test (α = 0.05). *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001 versus naïve mice (zero days after immunization).

We then analyzed 4 hours after infection bacterial burden in immunized mice challenged with lethal S. aureus or A. baumannii infection. In mice infected with S. aureus at 3 days after immunization, vaccinated mice had a significant reduction in blood bacterial density at 4 hours after infection (Kruskal-Wallis, P = 0.01) (fig. S6A). However, in mice infected with A. baumannii, the reduction in bacterial burden was much greater (1000-fold) compared with non-immunized mice (fig. S6B).

We have previously observed that cytokine profiles may be a better indicator of survival outcomes than changes in bacterial burden (21–23). Hence, we analyzed cytokine concentrations in the plasma of the same immunized mice challenged with a lethal dose of S. aureus or A. baumannii. At 4 hours after S. aureus infection, immunized mice had lower plasma concentrations of the proinflammatory cytokines IL-6 and TNF. They also had an elevated IL-10:TNF ratio, which indicates a net anti-inflammatory response (Fig. 5B). IL-10:TNF is a specific biomarker that has been associated with changes in clinical outcome during bacterial infection in patients (24–26). Similarly, at 4 hours after A. baumannii infection, immunized mice had lower plasma concentrations of the pro-inflammatory cytokines IL-6, IL-12, and TNF, as well as higher IL-10 resulting in an even more drastic increase in IL-10:TNF ratio (Fig. 5C). Concentrations of several cytokines were below the limit of detection (IFN-γ, IL-13, and IL-17A) or did not change (IL-1β and IL-4) (Fig. 5, B and C, fig. S7, and table S3). Thus, immunization decreased pro-inflammatory cytokines and increased anti-inflammatory cytokines as compared with naïve mice.

A + M + MA vaccination induces epigenetic and gene expression changes in macrophages

To further understand the mechanism of efficacy mediated by A + M + MA and evaluate whether the vaccine induces modifications in epigenetic biomarkers, we first studied histone reprogramming and transcription changes in splenic macrophages isolated from immunized mice. Epigenetic reprogramming has been shown to be a key characteristic of trained immunity (19, 27), with acetylation of histone 3 lysine 27 (H3K27ac) as a key marker for activation of promoters and enhancers (28). We first isolated splenic macrophages at 3 (vaccine effective) or 21 days (vaccine no longer effective) after immunization with the original 1× dose of the A + M + MA vaccine and then performed H3K27ac chromatin immunoprecipitation sequencing (ChIP-seq) and RNA sequencing (RNA-seq). H3K27ac ChIP-seq from macrophages isolated from mice immunized for 3 days demonstrated 1080 differential peaks compared with naïve mice and 1039 differential peaks compared with mice immunized for 21 days (Fig. 6A). Naïve control mice and mice immunized for 21 days showed few differential peaks (Fig. 6A).

Fig. 6. A + M + MA immunization induces epigenetic reprogramming and gene expression changes in murine splenic macrophages 3 days after vaccination.

H3K27ac ChIP-seq and RNA-seq studies were performed using splenic macrophages from isolated mice immunized with A + M + MA 3 or 21 days earlier or from unimmunized naïve control mice (N = 3 per group). (A) Shown is a summary of H3K27ac differential peaks (reflecting changes in chromatin acetylation from splenic macrophages) and PCA between 3- or 21-day immunized mice versus naïve control mice. Differential peaks were identified as adjusted P < 0.05, fold change (FC) > 2, and reads per peak > 50. (B) Shown is a summary of differentially expressed genes (DEGs) and PCA between 3- or 21-day immunized versus naïve control mice by RNA-seq. DEGs were identified as those showing P < 0.05, FC > 2, and reads per kilobase of transcript per million mapped reads (RPKM) > 1. (C) Heatmaps of H3K27ac differential peaks (left) and DEGs (right) are shown. The top half shows gene sites with increased acetylation or expression for macrophages isolated from 3-day vaccinated mice (more yellow and orange) versus macrophages from naïve mice or versus macrophages from 21-day vaccinated mice. The bottom half shows gene sites where acetylation or gene expression went down (more blue and cyan). A, B, and C represent individual mice. (D) Shown is a pathway analysis for H3K27ac differential peaks and DEGs using genes expressed higher in macrophages isolated from 3-day immunized mice compared with naïve control mice.

Similarly, RNA-seq identified 779 differentially expressed genes (DEGs) between naïve mice and mice immunized 3 days prior, 652 DEGs between mice immunized 3 days prior and mice immunized 21 days prior, but only 170 DEGs between naïve mice and mice immunized 21 days prior (Fig. 6B). The majority of highly acetylated H3K27 regions in mice immunized 3 days prior returned to baseline by 21 days after immunization. Most regions with low H3K27 acetylation in mice immunized 3 days prior had increased by 21 days after immunization, although not all returned to baseline (Fig. 6C). Similar gene expression changes were observed by RNA-seq (Fig. 6C), suggesting that changes in H3K27ac resulted in changes in gene expression. To understand the biological processes influenced by A + M + MA immunization, we conducted pathway analyses on DEGs. The analysis found that many genes highly up-regulated at 3 days after immunization were associated with host defense against various infections, such as those caused by S. aureus, Bordetella pertussis, Plasmodium species, and prions, as well as pathways known to be related to activation of the innate immune system, such as lysosome formation and the complement cascade (Fig. 6D).

Epigenetic and transcriptome changes are induced in human macrophages exposed to A + M + MA ex vivo

Since trained immunity was first found, studies confirming trained immunity have often focused on the Bacillus Calmette-Guérin (BCG) vaccine–induced human PBMCs (19, 20, 29, 30). Thus, to study the mechanism of the vaccine in a well-established trained immunity model, we used human macrophages differentiated from monocytes isolated from PBMCs to see whether the epigenetic and transcriptome changes we observed in mouse splenic macrophages also occurred in primary human macrophages. Macrophages were differentiated from PBMC-isolated human monocytes by culturing in the presence of human serum for a total of 6 days. To complete the differentiation process, 3-day stimulated macrophages were cultured in vitro for 3 days before the stimulation. The macrophages were exposed to A + M + MA for 24 hours and rested for 2 (3 days total) or 5 days (6 days total) before exposure to bacteria; this was done because an ex vivo model with 6 days of stimulation and resting has been used in numerous studies to study trained immunity phenotypes (19, 27, 29, 31, 32). Three days of stimulation and resting were included to represent the 3-day immunization time point in mice.

We first analyzed the cytokine production by ex vivo differentiated primary human macrophages stimulated with A + M + MA, rested for a total of 3 or 6 days, and then exposed to S. aureus or A. baumannii for 24 hours (fig. S8A). When cocultured with S. aureus or A. baumannii, compared with naïve human macrophages, human macrophages that had been stimulated and rested with A + M + MA for 6 days secreted less IL-6 and IL-10 and had lower IL-10:TNF ratio (Mann-Whitney test, P < 0.05 versus naïve for IL-6, IL-10, and IL-10:TNF ratio) (fig. S8B). Several other cytokines were not changed (IL-4) or below the limit of detection (IFN-γ, IL-12p70, IL-13, and IL-17A) (table S4).

Last, we also studied epigenetic and transcription changes in the same A + M + MA–stimulated human macrophages. H3K27ac ChIP-seq revealed 1622 or 1482 differential peaks comparing naïve, nonstimulated human macrophages with 3- or 6-day stimulated and rested macrophages, respectively (Fig. 7A). Similarly, RNA-seq identified 307 or 171 DEGs between naïve and 3- or 6-day stimulated and rested macrophages, respectively (Fig. 7B). Principal components analysis (PCA) of H3K27ac analysis showed a separation between A + M + MA–stimulated and naïve macrophages differentiated from human monocytes (Fig. 7, A and B). Heatmap analysis demonstrated epigenetic and transcriptome changes for human macrophages stimulated and rested 3 or 6 days after stimulation with the vaccine compared with naïve, unstimulated macrophages (Fig. 7C).

Fig. 7. A + M + MA immunization induces epigenetic reprogramming and gene expression changes in human macrophages.

H3K27ac ChIP-seq and RNA-seq studies were performed using human primary macrophages differentiated from monocytes exposed to A + M + MA for 24 hours and rested 2 or 5 days for a total of 3 or 6 days in culture, respectively (N = 3 per group). Three-day stimulated macrophages were also cultured in vitro for 3 days before the stimulation. (A) Shown is a summary of H3K27ac differential peaks and PCA analysis between 3-day stimulated and rested macrophages, 6-day stimulated and rested macrophages, naïve macrophages, and naïve, undifferentiated monocytes. Differential peaks were identified as adjusted P < 0.05, FC > 2, and reads per peak > 50. (B) Shown is a summary of DEGs between 3-day stimulated and rested macrophages, 6-day stimulated and rested macrophages, naïve macrophages, and naïve, undifferentiated monocytes. DEGs were identified as those showing P < 0.05, FC > 2, and RPKM > 1. (C) The heatmaps show differences in H3K27ac and RNA expression between stimulated macrophages stimulated and rested for 3 or 6 days after stimulation as compared with unstimulated macrophages. H3K27ac differential peaks are shown on the left. Three days versus naïve shows 1622 differential peaks between 3 days and naïve macrophages. Six days versus naïve shows 1482 differential peaks between 6 days and naïve macrophages. DEGs are shown on the right. Three days versus naïve shows 307 DEGs between 3 days and naïve macrophages. Six days versus naïve shows 171 DEGs between 6 days and naïve macrophages. A, B, and C represent independent replicates. (D) Shown is a pathway analysis of H3K27ac differential peaks and DEGs that were shown to be higher in macrophages stimulated and rested for 6 days compared with naïve control macrophages. MAPK, mitogen-activated protein kinase; JNK, c-Jun N-terminal kinase.

To understand the biological processes influenced by A + M + MA in human macrophages, we conducted a pathway analysis comparing 6-day stimulated with naïve macrophages. Multiple metabolic pathways were up-regulated among 6-day stimulated macrophages, including pantothenate and coenzyme A (CoA) biosynthesis, fatty acids biosynthesis, fatty acid elongation, coenzyme, nucleoside bisphosphate, ribonucleotide bisphosphate, purine nucleoside bisphosphate, and cellular protein metabolic processes. Immune pathways, such as mitogen-activated protein kinase and TNF signaling, were also up-regulated (Fig. 7D and table S5). A pathway analysis of transcriptome changes highlighted differences in metabolic pathways, such as arachidonic acid metabolism, as well as immune pathways, such as the hematopoietic cell lineage pathway (Fig. 7D and table S5).

DISCUSSION

The global crisis of AMR infections continues to expand, and the pipeline of new antibiotics to combat the threat cannot keep pace (33). Our study used models for two common routes of infections caused by AMR nosocomial pathogens: central line–associated bloodstream infections and ventilator-associated pneumonia (34). Many such infections are XDR with few effective prevention or treatment options. Most AMR pathogens are found in hospitals, and new strategies are critically needed to prevent such infections (35).

We found that a protein-free, tripartite vaccine improved the survival time in mice infected with varied, high-priority AMR pathogens, including Gram-positive bacteria, Gram-negative bacteria, and fungi, in both models of bacteremia and pneumonia, including in CD34⁺-humanized NSG mice. Although the magnitude of efficacy did vary by the pathogen, inoculum, and route of infection, we observed vaccine efficacy (improved survival time or reduced bacterial burden) in two independent laboratories, testing distinct strains of mice, bacteria, and fungi. Efficacy began by 24 hours after vaccination and lasted up to 3 weeks, and the effect was mediated by the innate rather than the adaptive immune system. In vivo efficacy required macrophages for protection against both the Gram-positive pathogen S. aureus and the Gram-negative pathogen A. baumannii. NK cells were required for optimal benefit against S. aureus infection but not for the Gram-negative pathogen A. baumannii. Neither the abrogation of neutrophils nor the absence of lymphocytes disrupted efficacy. Although we established that neutrophils were not required for protection, the results do not exclude the possibility of an adjunctive role in efficacy. Furthermore, vaccination increased the number of neutrophils, monocytes, and macrophages in the blood, and the vaccine directly induced monocyteto-macrophage differentiation and markedly enhanced phagocytosis in vitro and ex vivo.

Macrophages were shown to mediate improved survival in a manner that involved cytokine modulation and epigenetic reprogramming. The data demonstrate an overall shift to a net anti-inflammatory response to infection, even while up-regulating macrophage phagocytosis and enhancing clearance of bacteria in vivo. The shift likely helped to avoid sepsis-induced immune paralysis, enabling a superior survival outcome. The same trend was observed in mouse and human macrophages as well as the immune system as a whole.

We confirmed that the tripartite vaccine induced H3K27ac epigenetic changes and gene expression changes in mouse splenic macrophages as well as human macrophages derived from PBMC-isolated monocytes. We used a well-established ex vivo model used for a trained immunity study where human monocytes were stimulated for 24 hours and rested for 2 or 5 days to allow the functional program of cells to return to the study state; using this system, we demonstrated that the vaccine induced long-term epigenetic and transcriptome changes in human macrophages (19, 36). Specifically, priming has been described to be typically short term, wearing off after several days (37). In contrast, we observed up to 4 weeks of protective benefit by the vaccine in vivo. This correlated with continued epigenetic and gene expression changes, enhanced phagocytosis, and altered cytokine production by macrophages even after a 5-day rest period after vaccine exposure, collectively indicating that the effects of the vaccine cannot be explained by priming. The observed decrease in pro-inflammatory cytokines, such as IL-6 and TNF, and increase in the anti-inflammatory cytokine IL-10 (in the context of A. baumannii) after infection raise the potential for immune tolerance. However, tolerance is inconsistent with the markedly enhanced phagocytosis phenotype ex vivo and the decreased in vivo bacterial burden after infection in mice (37). Rather, collectively, these data are consistent with a trained immunity phenotype characterized by increased antimicrobial activity of innate immune cells (37, 38). Further study of other potential epigenetic changes is warranted to delineate the full scope of innate immune changes mediated by the vaccine.

Numerous studies have demonstrated that epigenetic changes are integral to inducing innate trained immunity (19, 31, 39) in NK cells, monocytes, and macrophages (19, 20, 31, 39, 40). Another vaccine, the BCG vaccine for tuberculosis, was found to provide generalized protection against a variety of other infections in children in an antigen-independent manner that was consistent with trained immunity (38). A phase 3, double-blinded, randomized clinical trial demonstrated that trained immunity induced by the BCG vaccine provided protection against respiratory tract infections in elderly patients (29). Furthermore, researchers have proposed the development of a trained immunity-based vaccine (41). In addition to our findings, the findings from the BCG vaccine and others demonstrate the potential of trained immunity to induce protection against a broad spectrum of infections in patients (20, 32, 36). However, a live vaccine, like BCG, cannot be safely administered to acutely ill patients going to hospitals, given the risk of “BCG-osis” (dissemination of the live bacteria), which has been described to occur in patients with decreased immunity (42). A vaccine consisting of nonliving components is necessary to administer to acutely hospitalized patients to prevent HAIs.

Metabolic change is another characteristic often studied with trained immunity-induced innate immune cells. We found epigenetic and transcriptional changes in multiple metabolic pathways, including those affecting nucleotides, ribonucleotides, and the tricarboxylic acid (TCA) cycle, as well as lipid and fatty acid metabolism. Changes in the TCA cycle and fatty acid metabolism have also been shown to be hallmark changes in BCG-induced trained immunity (43–46). The most substantial pathways induced by H3K27ac changes, the pantothenate and CoA biosynthesis, have also been shown to be particularly up-regulated by BCG vaccine–induced trained immunity, and the TCA cycle has been suggested to be one of the key metabolic pathways induced by trained immunity (43, 44). The pantothenate and CoA biosynthesis and biosynthesis of unsaturated fatty acids are also linked to lipid metabolism. Our results suggest that multiple key metabolism pathways are up-regulated by the vaccine, which is consistent with trained immunity phenotype. Further research into the functional metabolic effects of the vaccine on macrophages is warranted.

Our study has limitations. Although our vaccine was effective against a diverse array of nosocomial, AMR pathogens, it is possible that the vaccine is not protective against some other nosocomial pathogens. It is conceivable that infection could be worsened in some cases if a pro-inflammatory response is required for pathogen clearance. For example, multiple clinical studies have shown that BCG vaccine–induced trained immunity is not protective against coronavirus disease 2019 infection, although it has been shown to be protective against other viral infections (47, 48). In addition, although we tested the vaccine in a humanized mouse model, the model cannot completely recapitulate the human immune system. Therefore, we will not know whether the vaccine is effective in humans until clinical trials are conducted (49).

In conclusion, we describe a protein-free vaccine that has shown promise for preventing blood and lung infections caused by the highest-priority AMR pathogens that cause HAIs. The breadth and robustness of protection offered by this vaccine suggest that it carries the promise to reduce antimicrobial usage and improve outcomes for hospitalized patients. Traditional vaccines are “vertical” infection prevention approaches (7), which activate antigen-specific lymphocytes that target one pathogen at a time. This single-pathogen targeting makes such vaccines difficult to develop or deploy for the prevention of HAIs, which are caused by myriad bacterial and fungal pathogens. In contrast, trained immunity offers a much broader targeting of pathogens and is reflective of a “horizontal” infection prevention approach, rather than being pathogen specific. The cost of the horizontal targeting of many pathogens from one vaccine through trained immunity is that protection lasts for a shorter period of time than it does for adaptive immune-mediated traditional, pathogen-specific vaccines. However, given that the average duration of acute care hospitalization is 5 days and that 95% of hospitalizations last less than 21 days (50), the shorter duration of protective effect offered by trained immunity is not a detriment to vaccine efficacy or deployment.

MATERIALS AND METHODS

Study design

The objective of this study was to test whether our A + M + MA vaccine can provide protection against diverse AMR nosocomial pathogens in different mouse strains, as well as to study the mechanism of protection by identifying the stimulated innate immune cell populations and determining epigenetic and transcriptome changes within those populations. The number of replicates and independent experiments for each assay is indicated in the figure legends. All animal work was approved by either the Institutional Animal Care and Use Committee at University of Southern California or the AstraZeneca Institutional Animal Care and Use Committee in an Association for Assessment and Accreditation of Laboratory Animal Care International–accredited facility, in compliance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. Animals were randomly assigned to different vaccination groups, and personnel were not blinded to these assignments. Infected mice developed weight loss, ruffled fur, poor appetite, decreased ambulation, huddling behavior, and low body temperature. Weight loss of greater than 15% of body weight triggered euthanasia. Mice were monitored at least twice daily for up to 28 days. Soft bedding and other enrichment devices were provided as recommended by the veterinary staff.

Mouse strains

RAG1-KO (strain 034159), C3HeB/Fe (strain 000658), BALB/c (strain 000651), C57BL/6 (strain 000664), and human CD34⁺ hematopoietic stem cell–engrafted NSG mice were purchased from Jackson Laboratories. The mouse strain for each infection model was selected on the basis of prior experience, as well as other published studies. Multiple mice strains were also used to ensure that efficacy is not specific to any unique mouse strain. C3HeB/Fe mice were used for A. baumannii (10, 51), P. aeruginosa, and E. faecalis infections. BALB/c mice were used for S. aureus (9, 52), R. delemar (53), P. aeruginosa (54), and C. albicans (55) infections. C57BL/6 mice were used for K. pneumoniae (56) infection.

Vaccine immunization

Mannan (Sigma-Aldrich, M3640), MPL (InvivoGen, tlrl-mpls), WGP (InvivoGen, tlrl-wgp), and 2% Al(OH)₃ (Accurate Chemical & Scientific Corporation, A1090S) were prepared and stored according to the manufacturer’s protocol. Mannan (MedicaPharma, mannan), PHAD (Avanti Polar Lipids, 699800P), and 2% Al(OH)₃ (Croda, AJV3012) were used in GMP/GLP-grade A + M + MA. Unless specified, the vaccine was freshly prepared each time. For stability testing, GMP/GLP-grade A + M + MA was mixed as mentioned above and stored at ambient room temperature or 4°C avoiding light. Mice were immunized with 200 μl administered subcutaneously in the scruff of the neck with the premixed vaccine in PBS. Unless specified otherwise, each mouse was given single or combinations of 0.1% Al(OH)₃, 100 μg of WGP, 10 μg of MPL, and 100 μg of mannan. A + M + W included 0.1% Al(OH)₃, 10 μg of MPL, and 100 μg of WGP. A + M + W 30% included 0.1% Al(OH)₃, 3 μg of MPL, and 30 μg of WGP. A + M + W 10% included 0.1% Al(OH)₃, 1 μg of MPL, and 10 μg of WGP. A + M + W 1% included 0.1% Al(OH)₃, 0.1 μg of MPL, and 1 μg of WGP. A + M + MA (or A + M + MA 1×) included 0.1% Al(OH)₃, 10 μg of MPL, and 100 μg of mannan. A + M + MA 3× included 0.1% Al(OH)₃, 30 μg of MPL, and 300 μg of mannan. A + M + MA 10× included 0.1% Al(OH)₃, 100 μg of MPL, and 1000 μg of mannan.

Statistical analysis

All raw, individual-level data are presented in data file S1. Survival was compared by the nonparametric log-rank test with α = 0.05. Bacterial burden, cell population frequencies, macrophage assays, and cytokine concentrations were compared by the Wilcoxon-Mann-Whitney test or the Kruskal-Wallis test with α = 0.05.

Data and materials availability:

All data associated with this study are present in the paper or the Supplementary Materials. RNA-seq and H3K27ac ChIP-seq data are available at GEO under accession numbers GSE216061 and GSE216062 (mouse macrophages) and GSE235894 and GSE235897 (human macrophages).