Immunisation with the BCG and DTPw vaccines induces different programs of trained immunity in mice

Abstract

In addition to providing pathogen-specific immunity, vaccines can also confer nonspecific effects (NSEs) on mortality and morbidity unrelated to the targeted disease. Immunisation with live vaccines, such as the BCG vaccine, has generally been associated with significantly reduced all-cause infant mortality. In contrast, some inactivated vaccines, such as the diphtheria, tetanus, whole-cell pertussis (DTPw) vaccine, have been controversially associated with increased all-cause mortality especially in female infants in high-mortality settings. The NSEs associated with BCG have been attributed, in part, to the induction of trained immunity, an epigenetic and metabolic reprograming of innate immune cells, increasing their responsiveness to subsequent microbial encounters. Whether non-live vaccines such as DTPw induce trained immunity is currently poorly understood. Here, we report that immunisation of mice with DTPw induced a unique program of trained immunity in comparison to BCG immunised mice. Altered monocyte and DC cytokine responses were evident in DTPw immunised mice even months after vaccination. Furthermore, splenic cDCs from DTPw immunised mice had altered chromatin accessibility at loci involved in immunity and metabolism, suggesting that these changes were epigenetically mediated. Interestingly, changing the order in which the BCG and DTPw vaccines were co-administered to mice altered subsequent trained immune responses. Given these differences in trained immunity, we also assessed whether administration of these vaccines altered susceptibility to sepsis in two different mouse models. Immunisation with either BCG or a DTPw-containing vaccine prior to the induction of sepsis did not significantly alter survival. Further studies are now needed to more fully investigate the potential consequences of DTPw induced trained immunity in different contexts and to assess whether other non-live vaccines also induce similar changes.

1. Introduction

Vaccination is one of the most effective public health interventions to reduce infectious disease incidence and severity and prevents >2 million deaths annually [1]. Vaccines confer pathogen-specific protection via the induction of antigen-specific adaptive immune responses. A growing body of evidence, however, suggests that some vaccines can also elicit nonspecific effects (NSEs) on morbidity and mortality unrelated to the targeted disease [2,3]. Live vaccines - including those against tuberculosis, polio, measles, and smallpox - have been associated with greater reductions in all-cause mortality than can be explained by protection against the targeted pathogens [4–8]. Multiple epidemiological studies in Guinea Bissau, India, Papua New Guinea and Malawi and three randomised controlled trials (RCT) have shown that the BCG vaccine, a live-attenuated vaccine against tuberculosis, is associated with reductions in neonatal mortality of up to 50% [9,10]. Studies in adults have also shown that BCG reduces the risk of hospitalisation for respiratory viral infections [7,11] and a RCT found that BCG-immunised adults had significantly reduced viral titres following challenge with an attenuated yellow fever virus [12]. Based on these data, international clinical trials are currently underway to investigate whether BCG provides nonspecific protection against COVID19 in healthcare workers [13].

In contrast, several non-live vaccines including the diphtheria, tetanus, whole-cell pertussis (DTPw) vaccine have been controversially associated with a net increase in all-cause infant mortality in low income countries, despite providing effective disease-specific protective immunity [2]. These effects appear to be particularly evident in girls, suggesting sex-differential effects [14,15]. All-cause mortality has, for example, been reported to be 84% higher in girls in Guinea Bissau immunised with the DTPw vaccine [16]. Further observational studies in >10 countries have supported these findings [17–20]. These effects appear to be predominantly due to increased mortality from illness with fever and other infections (i.e. unrelated to pathogens targeted by the vaccine) including respiratory infections and diarrheal disease [18]. One possibility that has been suggested is that immunisation with DTPw enhances susceptibility to sepsis, which is a significant cause of infant mortality in low-income countries [21]. An increasing number of studies have now reported similar associations for other non-live vaccines [2,9,22] and a better understanding of the effects of non-live vaccines on the innate immune system is therefore urgently needed.

While the mechanisms underlying vaccine NSEs are incompletely understood, there is now substantial evidence that the NSEs associated with BCG are mediated at least in part via trained immunity – a microbially induced epigenetic and metabolic reprogramming of innate immune cells that leads to heightened responses to subsequent unrelated infections [23,24]. BCG vaccination induces genome-wide epigenetic changes and immunometabolic alterations in monocytes [25], neutrophils [26] and NK cells [27], priming them for increased proinflammatory cytokine responses (e.g. TNFα and IL6) [25–31] and conferring increased protection from experimental infection [23]. These changes have been shown to persist in circulating monocytes for at least 12 months [24] and in neutrophils for at least 3 months [26]. How BCG could reprogram these short-lived cells for such a long time remained a puzzle until recently, when it was demonstrated that BCG reprograms haematopoietic stem cells (HSC) in the bone marrow [32,33].

Whether non-live vaccines, such as DTPw, induce trained immunity is very poorly understood. Here, we used a mouse model to investigate whether live (BCG) and non-live (DTPw) vaccines differentially reprogram trained immune responses and investigated whether immunisation with DTPw-containing vaccines altered susceptibility to sepsis. We found that immunisation with DTPw led to altered monocyte and DC cytokine responses to subsequent ex vivo restimulation compared to BCG immunised mice. Furthermore, immunisation with either BCG or DTPw differentially altered chromatin accessibility in cDCs, suggesting that these effects were epigenetically mediated. We also found that immunisation with DTPw led to an increased frequency of DCs and their precursors in the bone marrow. These effects were evident months after immunisation and indicate that both live and non-live vaccines induce unique programs of trained immunity. Interestingly, these effects on trained immunity were dependent on the order in which vaccines were administered. Immunisation with a pentavalent DTPw-Hib-HepB vaccine was not associated with significantly increased mortality or morbidity in two different mouse models of sepsis.

2. Methods

2.1. Mice

For adult/weaning immunisation studies, C57BL/6J (RRID: IMSR_JAX:000664) mice were bred and maintained under specific and opportunistic pathogen free conditions at the South Australian Health and Medical Research Institute (SAHMRI) in individually-ventilated cages. For adult sepsis survival studies, mice were housed in a conventional PC2 facility (SAHMRI Preclinical, Imaging and Research Laboratories). Mice were housed under standardised conditions with regulated daylight, humidity, and temperature with free access to food and water. Experiments were approved by the SAHMRI Animal Ethics Committee according to national guidelines. Female mice were used for all experiments, as epidemiological evidence suggests that vaccine nonspecific effects are more apparent in females [19,34,35], except for the neonatal sepsis experiment, where both males and females were included.

For neonatal sepsis studies, C57BL/6J (B6, stock number 000644) mice were purchased from Jackson Laboratories and maintained under specific pathogen free conditions at the British Columbia Children’s Hospital Research Centre animal facility. Mating pairs were established a minimum of one week after arrival to the animal facility and placed on a high-fat diet. Dams remained on the high-fat for the duration of the experiment. Animal procedures were approved by the University of British Columbia Animal Care Committee.

2.2. Immunisations

Mice were immunised as weaners or as 4–5 day old pups as indicated in the figure legends. For immunisation of weaners, litters were randomly mixed among cages and full cages of mice were immunised at the same time. Mice were mock-immunised subcutaneously with 1x phosphate buffered saline (PBS; 100 μl; Sigma-Aldrich) or subcutaneously immunised with the BCG SSI vaccine (1/10th vial diluted in 100 μl PBS; BCG SSI, Danish 1331 sub-strain, 2–8 × 10⁶ CFU/vial) or BCG 10 vaccine (1/10th vial diluted in 100 μl PBS; BCG 10, Brazilian Moreau sub-strain 1.5–6 × 10⁶ CFU/vial), DTPw SII vaccine (1/10th vial diluted in 100 μl PBS; Serum Institute of India PVT. LTD), Tritanrix-HepB DTPw vaccine (1/5th – 1/10th vial diluted in 100 μl PBS; Tritanrix HepB; GSK), EasyFive pentavalent vaccine (1/10th vial; EasyFive-TT; Panacea Biotech) or Pentavac SII vaccine (DTPw-HepB-HiB; Serum Institute of India) as stated in the figure legends.

2.3. Tissue processing and FACS

Aseptically isolated spleens were homogenized in RPMI (RPMI-1640; Sigma-Aldrich) containing 1% (v/v) heat inactivated fetal calf serum (FCS; Assay Matrix) and filtered through an 80 μm nylon filter (Millipore). Cells were centrifuged (5 min, 350×g, 4 °C) and erythrocytes lysed with cold 1x PharmLyse (BD Biosciences) prior to 2x washes in RPMI 1% FCS. Bone marrow was isolated from aseptically harvested femurs and tibia of the hind legs. Cells were extracted via centrifugation [36] and erythrocytes lysed (Pharmlyse) prior to staining. For staining, cells (1–5 million/sample) were mixed with the relevant antibody cocktail (Tables S1–2) in 30 μl FACS buffer (PBS with 0.1% BSA and 0.4% EDTA) and incubated on ice (30 min). Cells were washed 2x in FACS buffer and acquired on a 5-laser BD LSRFortessa X-20 or a BD FACSymphony flow cytometer. Fluorescence spillover was manually compensated using single-stained controls; data were analyzed using FlowJo v10.5–10.7 software.

2.4. Splenocyte stimulation

For stimulations for intracellular staining, 0.5–1 × 10⁷ cells/ml splenocytes were stimulated with heat killed Candida albicans (HKCA; 10⁶ cells/ml; InvivoGen, cat: tlrl-hkca), resiquimod (R848; 1 μg/ml; InvivoGen, cat: tlr-r848) with polyinosinic–polycy tidylic acid sodium salt (Poly(I:C); 10 μg/ml; Sigma-Aldrich) or lipopolysaccharide (LPS; 10–100 ng/ml; E. coli serotype 055:B5, Sigma-Aldrich) for 4 h. 1x Golgi Plug Brefeldin A (BD Biosciences) was added for the final three hours of the stimulation. Cells were centrifuged (5 min, 350×g, 4 °C) and washed with 200 μl FACS buffer prior to ICS.

2.5. ICS

Samples were stained with 30 μl ICS surface staining antibody cocktail (Table S3) in FACS buffer for (30 min) on ice, washed with FACS buffer (200 μl) and fixed with 80 μl Cytofix Fix/Perm buffer (BD Biosciences) or 4% formaldehyde in PBS for 20 min. Fixed cells were permeabilised with 1x Cytofix wash with perm/wash buffer (BD Biosciences) and stained with an ICS cytokine cocktail (Table S4) for 30 min.

2.6. Blood stimulation assays

EDTA-anticoagulated blood (100 μl) was washed with 1 mL RPMI supplemented with 1% FCS and centrifuged (400×g, 5 min). Cell pellets were resuspended in 1000 μl lysis buffer (Pharmlyse, BD Biosciences) and incubated for 2 min at room temperature. Cell pellets were washed (9 mL RMPI 1% FCS) and resuspended in 100 μl complete RPMI (cRPMI) supplemented with 10% (v/v) heat inactivated FCS, non-essential amino acids, penicillin/streptomycin, glutamine and 2-mercaptoethanol. Cells (50 μl) were cultured in 96-well plates in the presence of 50 μl cRPMI and Golgi Plug (BD Biosciences) with or without 100 ng/mL LPS for 3 h (37 °C, 5% CO₂). Cells were harvested by centrifugation and washed twice with 200 μl FACS buffer prior to ICS.

2.7. Cell sorting

For sorting, whole spleens were homogenized, filtered through an 80 μm nylon filter and washed in FACS buffer. Pellets were resuspended in 100 μl sorting antibody cocktail (Table S5) in FACS buffer and stained in the dark for 30 min. Samples were washed and incubated with 25 μl anti-biotin microbeads (Miltenyi) in 300 μl AutoMACS buffer for 30 min, prior to magnetic depletion of CD3/CD19⁺ lymphocytes (Miltenyi AutoMACS). For ATAC-seq experiments, cDC (NK1.1⁻Ly6G⁻B220⁻CD11c⁺MHCII⁺) were sorted from depleted samples on a BD FACSMelody Cell Sorter. For cytokine stimulation experiments, monocytes (NK1.1⁻Ly6G⁻CD11c⁻-CD11b⁺) and DC (NK1.1⁻Ly6G⁻CD11c⁺) were sorted from depleted samples on a BD FACSAria Fusion or BD FACSMelody Cell Sorter. Sorted cells were cultured in 96 well plates (10⁵/well) with HKCA (10⁶ cells/ml), resiquimod (1 μg/m) or LPS (100 ng/m) for 18 h and cell-free supernatants collected for cytokine analysis.

2.8. Cytokine measurement

Cytokine secretion from cells was measured using commercial TNFα or IL6 ELISA kits (Life Technologies) according to the manufacturer’s instructions. Cytokine concentrations in serum were measured using a flow cytometric multiplex bead array (Biolegend Legendplex), acquired on a BD LSR Fortessa X20 flow cytometer.

2.9. Antibody ELISAs

Specific antibody responses were assessed via ELISA. Plates were coated with trivalent DTPw vaccine (1/200 dilution), Diphtheria toxoid (0.5 μg/mL, List Labs) or BCG vaccine (20 μg/mL) and blocked with ELISA buffer (Life Technologies). Diluted sera were added and incubated overnight at 4 °C. Bound antibodies were detected with HRP-linked Goat anti-mouse Total IgG detection antibody (1:2000, Life Technologies) and developed with TMB substrate.

2.10. Assay for Transposase Accessible chromatin (ATAC)-seq

ATAC-seq was performed on freshly sorted cDC using the OMNI-ATAC method [37]. Briefly, 50,000 cells were lysed in 50 μl ATAC-seq lysis buffer containing 0.1% NP-40, 0.1% Tween-20 and 0.001% digitonin (Promega) in 0.5 mL tubes and nuclei pelleted via centrifugation. Nuclei were transposed using 2 μl Tn5 transposase (Illumina) in 50 μl 1 × TD buffer for 30 min at 37 °C. Transposed DNA samples were purified using DNA Clean and Concentrator 5 columns (Zymo Research) prior to PCR amplification with adaptor-specific primers containing Illumina index barcode sequences. Amplified libraries were cleaned up with Zymo DNA Clean and Concentrator 5 columns and libraries were quantified using a 4150 TapeStation system (Agilent) with High Sensitivity D1000 reagents. Following QC, libraries were sequenced on a NextSeq 500 Sequencer (Illumina) using a High-Output Flow Cell (2 × 75bp PE reads; 23 m PE reads/sample).

2.11. ATAC-Seq data analysis

Fastq read quality was assessed using FastQC version v0.11.3 and summarized with MultiQC [38]. Adapter sequences were removed using Cutadapt v2.8 [39], and remaining sequences were quality filtered with Trimmomatic v0.38 [40] (sliding (4nt) window with a minimum PHRED score of 20, together with average score of 30). Reads were then aligned to the GRCm38 mouse genome using Bowtie v2.3.5 “–very-sensitive” mode [41]. SAMtools was used to convert to BAM format and sort alignments [42]. The Bioconductor package ATACseqQC was used to assess transcription start site enrichment [43], and peaks were called using Genrich (available from https://github.com/jsh58/Genrich) in the ‘ATAC-seq mode’ with a q value threshold of 0.05, excluding mitochondrial peaks and removing PCR duplicates. Feature Counts v1.5.0-p2 was used to count aligned reads to generate a count per sample for each peak [44]. These counts were then imported into R v3.6.3 for further statistical analysis. Peaks were filtered so as to only include those with at least 10 reads uniquely aligned in at least 2 samples and unknown sources of variation in the data were removed using SVAseq [45]. EdgeR was used to normalize the data (using trimmed mean of M-values method) and perform differential accessibility analyses (with the glmLRT function) [46]. Gene set enrichment analysis was performed using the MSigDB v7.1 gene sets [47] and the Camera function in the EdgeR package. Alignments were visualized in R v3.6.3 using the GenomicAlignments v1.20.1 and ggbio v1.32.1 libraries.

2.12. Cecal ligation and puncture sepsis model

CLP sepsis was performed on immunised mice six weeks following vaccination, as previously described [48]. Briefly, mice were anaesthetised (1 L/min 2.5–3.5% isoflurane in O₂) and the abdomen shaved and sterilised prior to laparotomy. For the 24hr pilot study, 7 mm of the tip of the caecum was ligated and pierced through with a 21G needle to produce two punctures. For the sepsis survival study, a 15 mm ligation was used with two 21G needle punctures. The laparotomy was closed with non-absorbable sutures and mice were administered subcutaneous buprenorphine (0.05 mg/kg) and normal saline (1 mL) and allowed to recover from anaesthetic in a pre-warmed cage before transfer to a group cage. Mice were injected with saline (1 mL; SC) and provided with moistened food daily. All animals were monitored 6 hourly for temperature and clinical scoring (Table S6). Blood samples for cytokine analysis (50 μl) were collected via facial vein bleed 24 h following surgery. Mice were provided subcutaneous buprenorphine every 6–12 h if exhibiting signs of pain and monitored for survival using death as an endpoint.

2.13. Neonatal sepsis model

Mice were randomized to vaccination with the Pentavac SII vaccine (DTPw-HepB-HiB; Serum Institute of India) (50 μl vaccine via subcutaneous injection) or no injection on day of life (DOL) 4–5, 9–10, and 14, balancing the number of males and females in each treatment. The cecal slurry challenge was performed as previously described [49–51]. Briefly, cecal contents from 6 to 12 week-old male mice were collected and resuspended at 160 mg/mL in sterile dextrose water. The mixture was filtered through a 70 μm filter and stored in 0.5 mL aliquots (−80 °C). At DOL 14, mice were challenged with 2.25 mg/g cecal slurry via intraperitoneal injection, a dose previously titrated to achieve 70% lethality. Mice were monitored at 12–14 h post challenge (HPC), 16–19 HPC, 21–25 HPC, 36–40 HPC, 86–94 HPC, and 116–160 HPC for disease morbidity and body weight. Clinical scores were assigned based on body weight, attitude and behaviour, hydration, physical appearance, facial grimace, and other clinical signs such as soft stools or increased respiratory effort (Table S7). Mice that were immobile or scored >3 at >24 HPC were considered to have reached humane endpoint and were euthanized.

3. Results

3.1. Immunisation with BCG and DTPw leads to altered monocyte and DC cytokine responses to subsequent microbial stimuli

Immunisation with the live-attenuated BCG vaccine is now well-established to induce trained immunity in mice and humans [25,27,32]. To investigate whether immunisation with the non-live DTPw vaccine differentially reprograms trained immune responses compared to immunisation with BCG, we compared cytokine responses in splenic monocytes and DCs from mice immunised 10 weeks earlier with either the DTPw SII or BCG SSI vaccines. Cytokine responses were assessed by intracellular cytokine staining (ICS) following ex vivo stimulation of splenocytes with either heat-killed fungal pathogen C. albicans (HKCA), bacterial lipopolysaccharide (LPS) or resiquimod plus polyinosinic–poly cytidylic acid (R848 + PolyI:C) to mimic TLR sensing of a viral pathogen (Fig. 1A). Enhanced innate immune cell cytokine responses in immunised animals/individuals is a widely accepted hallmark of trained immunity [23]. We focused our assessment on the production of TNFα and IL6, as these cytokines have previously been used in other studies as readouts of vaccine-induced trained immunity [25,52].

Fig. 1. The DTPw and BCG vaccines induce long-lasting alterations in monocyte and DC cytokine responses.

Female 3–4 week old (n = 5) C57BL/6J mice were subcutaneously immunised with BCG (SSI), DTPw (SII) or saline (mock) and boosted two weeks later (A). Splenocytes were harvested 10 weeks following prime immunisation and cultured with media alone (unstim) or with heat-killed C. albicans (1 × 10⁶ cells/mL; HKCA), LPS (10 ng/mL), or resiquimod (1 μg/ml; R848) with Poly I:C (10 μg/ml), for 4 h. TNFα and IL6 were quantified via intracellular cytokine staining (ICS) in monocytes (B-C) and DCs (D-E). (F-G) Blood samples from BCG (SSI), DTPw (SII) or saline (mock) immunised mice were collected at V + 6 weeks, erythrocyte depleted and stimulated with LPS (100 ng/mL; 3 h). TNFα and IL6 production in monocytes (F) and DCs (G) were assessed via ICS. Flow cytometry analysis of spleen (H) and bone marrow (I&J) collected from mice 10 weeks after prime immunisation with BCG (SSI), DTPw (SII) or saline (mock). Data are represented as the mean ± SEM. Data in B-E represent a single independent experiment, and data from F-J represent a separate independent experiment. A one-way ANOVA was used to assess statistical significance. * = P > 0.005, ** = P > 0.05, *** = P > 0.001, ns = not significant.

Consistent with previous reports [25,30], compared to mock-immunised controls there was a significantly higher proportion of TNFα⁺ splenic monocytes (CD19⁻CD3⁻Ly6G⁻NK1.1⁻CD11c⁻-CD11b⁺) following stimulation of splenocytes collected from BCG-immunised mice with HKCA, LPS or R848 + PolyI:C (Fig. 1B). In comparison, there was a significantly higher proportion of TNFα⁺ monocytes following stimulation of splenocytes from DTPw-immunised mice with LPS and R848 + PolyI:C, but not HKCA. Interestingly, we found that DTPw-immunised mice had a significantly higher proportion of IL6⁺ monocytes compared to mock-immunised controls in response to LPS and R848 + PolyI:C stimulation, but again not HKCA (Fig. 1C), while this was not observed following stimulation of splenocytes from BCG-immunised mice.

While BCG-induced trained immune responses have previously been reported in monocytes, neutrophils, [26] and NK cells, [25,27] it is poorly understood whether trained immune responses can also be elicited in DCs following vaccination. To investigate this, we also assessed cytokine responses in splenic DCs (CD19⁻CD3⁻-Ly6G⁻NK1.1⁻CD11c⁺) following stimulation of splenocytes from BCG or DTPw-immunised mice. We found that a significantly higher proportion of DCs from BCG or DTPw-immunised mice produced TNFα following stimulation with HKCA, LPS or R848 + PolyI: C compared to mock-immunised controls (Fig. 1D). Interestingly, we observed a marked increase in the proportion of IL6⁺ DCs following LPS stimulation of splenocytes from DTPw-immunised mice compared to mock-immunised controls, and this was not observed in DCs from BCG-immunised mice (Fig. 1E). Compared to mock-immunised controls, there was a significantly higher proportion of IL6⁺ DCs following stimulation of splenocytes from BCG and DTPw-immunised mice with R848 + PolyI:C. A significantly higher proportion of IL6⁺ DCs following HKCA stimulation was only observed in BCG-immunised mice (Fig. 1E). Immunisation with either the DTPw or BCG vaccines thus differentially altered splenic monocyte and DC responses to subsequent microbial stimuli, which was dependant on the vaccine administered and the stimulant used. These data suggest that the BCG and DTPw vaccines each induce different programs of trained immunity in mice.

To determine if trained immune responses were also evident in circulating myeloid cells, in another independent experiment, we collected blood samples from mice six weeks following the first (prime) immunisation with BCG, DTPw or saline (mock). Blood samples were depleted of erythrocytes and stimulated with LPS and cytokine production assessed via ICS. LPS stimulation led to almost all circulating monocytes and DCs to produce TNFα and no differences between mock, BCG or DTPw-immunised mice were evident (Fig. 1F–G). The proportion of IL6⁺ monocytes was also not significantly different between the groups (Fig. 1F). Interestingly, and consistent with the data in splenic DCs, the proportion of IL6⁺ DCs was significantly higher in DTPw-immunised mice in response to ex vivo stimulation with LPS, compared to mock-immunised mice (Fig. 1G). This was not observed in BCG-immunised mice.

It has previously been demonstrated that following BCG vaccination increased numbers of monocytes are evident in peripheral blood in humans and in the lungs of mice [25]. To determine if immunisation with DTPw altered the proportion of monocytes or DCs in the spleen or bone marrow, we performed flow cytometry analysis of spleen and bone marrow isolated from mice immunised 10 weeks prior with BCG or trivalent DTPw. We found significantly more monocytes (as a proportion of live cells) in the spleen of BCG-immunised mice compared to mock- or DTPw-immunised mice (Fig. 1H). The proportion of splenic DCs in mice administered either vaccine was significantly reduced compared to mock-immunised controls (Fig. 1H), which may be due to trafficking of these cells to the site of immunisation or draining lymph node.

A key feature of BCG-induced trained immunity, which explains its long-lasting influence on innate immune cells, is the reprogramming of haemopoietic stem cell (HSC) precursors in the bone marrow resulting in functionally altered progeny [32,33]. We observed a higher proportion of CD11c⁺ cells (DCs and their precursors) in the bone marrow of DTPw-immunised mice compared to a higher proportion of Ly6G⁺ cells (neutrophils and their precursors) in bone marrow of BCG-immunised mice (Fig. 1I). The proportion of Ly6G⁻CD11c⁻CD11b⁺ cells in the bone marrow was not significantly different between the groups. A higher frequency of CD11c⁺ cells in the bone marrow suggested that DTPw-immunised mice had an increased capacity for generating DCs. To further assess this, we used flow cytometry analysis to quantify Lin⁻cKit⁺Sca1⁺ (LKS) HSCs in bone marrow collected 10 weeks following prime immunisation. We found that DTPw-immunised mice had a higher frequency of CD150⁻CD34⁺ short-term HSCs (ST-HSCs P = 0.06) compared to mock-immunised mice and that common myeloid progenitors (CMP; LK Sca1⁻CD34⁺) were significantly increased in these mice (Fig. 1J). Together, our results suggest that vaccination with DTPw induces functional changes in monocytes and DCs in the periphery and spleen and promotes the generation of DCs in the bone marrow. Further analysis is required to determine whether DTPw-containing vaccines induce functional alterations to HSCs, as is associated with BCG-induced trained immunity [32].

3.2. Immunisation with BCG or DTPw differentially reprograms chromatin accessibility in DCs

Previous studies have demonstrated that the trained immunity induced by BCG is mediated in part by genome-wide epigenetic reprogramming of monocytes, NK cells and neutrophils, including changes in chromatin accessibility, which leaves cells primed for increased cytokine responses to subsequent microbial encounters [25]. Given that immunisation with DTPw and BCG led to altered DC cytokine responses and that DCs have not been extensively investigated in the context of vaccine-induced trained immunity, we focused our analyses on DCs. We profiled genome-wide chromatin accessibility using Assay for Transposase-Accessible Chromatin sequencing (ATAC-seq) [53] of conventional DCs (cDCs) isolated from the spleens of mice immunised with BCG, DTPw, or saline. Spleens were harvested 12 weeks after the prime immunisation to assess long-term changes in the cDC epigenome, and viable CD11c⁺MHCII⁺B220⁻ cDCs were isolated by FACS. Importantly, flow cytometry analysis revealed that there were no significant differences in the proportion of CD11b⁺ cDCs in the cells sorted from BCG or DTPw-immunised mice (Fig. S1), which could otherwise confound the detection of differences in chromatin accessibility. ATAC-seq reads preferentially aligned to the genome in close proximity to Transcription Start Sites (TSS) (Fig. 2A), consistent with high-quality libraries [40]. Following adjustment for batch effects (i.e. day of cell sorting and transposition), MDS analysis revealed that samples clustered by the vaccine administered, consistent with each vaccine inducing different patterns of chromatin accessibility (Fig. 2B). Gene set enrichment analysis (GSEA) revealed that genes involved in a number of key immune and metabolic pathways including the IFNa and IFNc responses, IL6/JAK/STAT3 signalling, and oxidative phosphorylation, were statistically enriched among genes with altered chromatin accessibility in BCG or DTPw-immunised mice (Fig. 2C; Table S8–10). Interestingly, genes involved in the complement and TNFα signalling pathways were enriched in samples from BCG but not DTPw-immunised mice, suggesting that immunisation with DTPw has a different effect on DC chromatin accessibility compared to BCG-immunised mice. We statistically assessed differences between DTPw and BCG-immunised mice via GSEA (Fig. 2D) and found an enrichment for genes involved in the inflammatory response and other immune related pathways in cDCs from BCG compared to DTPw-immunised mice (Fig. 2D–E). Loci associated with higher chromatin accessibility in DCs from BCG-immunised mice included Stat1 and Acod1/Irg1 (Fig. 2F–G). These genes have previously been associated with protective anti-mycobacterial immune responses with roles in IFN signalling and immunometabolism, respectively [54,55]. Loci associated with increased chromatin accessibility in DCs from DTPw-immunised mice included Arrb1 and Cxcr5, which have been associated with DC maturation and promotion of Th2 responses, respectively [56,57]. These results suggest that both BCG and DTPw induce changes in chromatin accessibility in cDCs, and that the epigenetic remodelling driven by Depew is distinct to the remodelling driven by BCG vaccination.

Fig. 2. Immunisation with DTPw and BCG differentially alters chromatin accessibility in conventional DCs.

ATAC-seq was performed on 50,000 sorted cDCs from n = 4 DTPw (SII), n = 4 BCG (SSI) or n = 2 mock immunised female C57BL/6J mice. (A) Analysis with the ATACseqQC Bioconductor package revealed a strong enrichment of ATAC-seq reads at transcription start sites (TSS) (enrichment score > 15) indicating high-quality ATAC-seq data. (B) MDS analysis of ATAC-seq data. (C) Gene set enrichment analysis (GSEA) revealed an enrichment of genes involved in key immune and metabolic pathways among loci with altered chromatin accessibility in BCG and DTPw-immunised mice. (D) GSEA comparing chromatin accessibility in DCs from BCG and DTPw-immunised mice. (E) Heatmap of normalised read counts at loci identified in the GSEA analysis to have altered chromatin accessibility. (F) Boxplot of normalised read counts at selected loci relative to GAPDH. CPM = counts per million. (G) ATAC-seq read alignments at selected peaks of interest. Statistical significance was assessed via a one-way ANOVA with multiple comparisons. * = P < 0.05, ** = P < 0.01. Data are from a single experiment.

3.3. Trained immune responses induced by BCG and DTPw are influenced by both previous and subsequent immunisations

Mounting epidemiological evidence now supports that a ‘live vaccine last’ schedule could substantially mitigate any deleterious NSEs associated with non-live vaccines [3,52]. We hypothesised that changes to the order in which vaccines are administered could alter trained immunity, providing an immunological basis for this epidemiological evidence. To assess this hypothesis preclinically, we immunised (prime + boost) mice with BCG or DTPw alone, or with both vaccines administered in different orders (Fig. 3A). Spleens were collected 12 weeks after the prime immunization and monocyte and DC cytokine responses were assessed via ICS following ex vivo stimulation of splenocytes with a panel of microbial stimuli (HKCA, LPS and R848 + PolyI:C).

Fig. 3. The order in which vaccines are administered alters monocyte and DC trained immune responses.

(A) Female C57BL/6J mice (10 mice/group) were subcutaneously immunised at 3–4 weeks of age with saline (mock), BCG (SSI)), or DTPw (SII) and boosted 2 weeks later or were immunised with both vaccines administered in different orders (DTPw + boost + BCG or BCG + DTPw + boost). Splenocytes collected at 12 weeks post the prime immunisation were either unstimulated (media) or were stimulated ex vivo with LPS (100 ng/mL), HKCA (1 × 10⁶ cells/mL) or R848 (1 μg/ml) + Poly I:C (10 μg/ml) prior to the assessment of the proportion of TNFα⁺ and IL6⁺ monocytes (B-C) and DCs (D-E) by intracellular cytokine staining. (F) Fold-enrichment of selected loci normalised to GAPDH was assessed by ATAC-qPCR. Data are represented as the mean ± SEM and are from a single experiment. Statistical significance was assessed via a one-way ANOVA with multiple comparisons. * = P < 0.05, ** = P < 0.01, *** = P < 0.001.

We noted that in some cases the trained cytokine responses in this experiment (Fig. 3B–E) differed to those we previously observed (in Fig. 1B–E). For example, we noted that in this experiment monocytes from mice immunised with DTPw alone did not have enhanced TNFα or IL6 responses following ex vivo stimulation, compared to mock-immunised controls. Despite this variation between experiments, there were several effects that were consistently identified across multiple different experiments. For example, consistent with our previous data (Fig. 1B), and with another independent experiment administering a different BCG vaccine, BCG10 (Fig. S2A), the proportion of splenic TNFα⁺ monocytes was significantly higher in mice immunised with BCG following ex vivo stimulation of splenocytes with HKCA, compared to mock- or DTPw-immunised mice (Fig. 3B). Interestingly, mice immunised with the DTPw vaccine, before or after immunisation with BCG, had a significantly reduced proportion of TNFα⁺ monocytes following HKCA stimulation compared to mice immunised with BCG alone.

Another consistent effect that we observed across multiple different experiments (Fig. 1E & Fig. S2D), was a significantly higher proportion of IL6⁺ splenic and blood DCs in response to LPS stimulation of cells collected from DTPw-immunised mice (Fig. 3E). In BCG-vaccinated mice, DC IL6 responses were mildly increased, and this was only significant following LPS stimulation. Interestingly, administering BCG to mice before or after DTPw led to a significant reduction in the proportion of IL6⁺ DCs (Fig. 3E). Furthermore, immunisation with BCG following DTPw (‘live vaccine last’) resulted in a significantly higher proportion of IL6⁺ DCs following HKCA or LPS stimulation compared to mice administered DTPw following BCG (‘nonlive vaccine last’), indicating that the order that vaccines are administered appears to significantly influence trained DC cytokine responses. These effects were not explained by differences in cell viability, as the viability of monocytes and DCs from the different immunisation groups was not significantly different (Fig. S3). Together, these data suggest that the order in which vaccines are administered significantly influences trained immunity, which could provide a potential explanation for associations between vaccine administration order and vaccine NSEs. Importantly, changes to the order in which vaccines were administered did not alter adaptive immune responses, as antibody responses to the BCG and DTPw vaccines were not significantly different between groups (Fig. S4).

To assess whether changes to the order in which vaccines were administered altered chromatin accessibility, we performed ATAC-qPCR using custom primers corresponding to loci identified in our previous ATAC-seq data as having altered chromatin accessibility in BCG or DTPw-immunised mice (Fig. 2F). Consistent with the ATAC-seq data, the ATAC-qPCR data indicated that chromatin accessibility at regions associated with Acod1/Irg1 were significantly higher in DCs from BCG-immunised mice and at Arrb1 and Cxcr5 in DTPw-immunised mice (Fig. 3H). Interestingly, mice immunised with both BCG and DTPw had increased chromatin accessibility near Acod1/Irg1 compared to mice immunised with DTPw alone. These data suggest that changes to the vaccination schedule could alter chromatin accessibility at some loci. Further work is needed to assess the effects of vaccine administration order more fully.

3.4. DTPw vaccination alters acute illness severity and cytokine responses in a 24hr period following the induction of polymicrobial sepsis

Epidemiological data has suggested that DTPw vaccination may increase the risk of death from unrelated infections including pneumonia, infections presenting with fever [18] and sepsis, [34,58] a significant cause of infant mortality in regions where detrimental effects of DTPw vaccination have been suggested [21]. These observational studies are, however, unable to determine causal associations and may be subject to bias and confounding. Furthermore, a biological mechanism explaining these observations is currently lacking. Trained immunity induced by BCG has been associated with protection against sepsis and other infections [7,8,25], but whether immunisation with DTPw alters susceptibility to sepsis is currently unknown. To investigate this, we first conducted a short pilot experiment in which sepsis was induced by cecal ligation and puncture (CLP) in mice (n = 8/group) immunised six weeks prior with saline (mock), BCG or DTPw (Fig. 4A). A group of mock-immmunised sham-surgery mice were included as controls. Interestingly, we found that DTPw-immunised mice exhibited significantly higher clinical scores (Fig. 4B–C) and significantly lower temperatures (Fig. 4D) over a 24 h period following the induction of sepsis compared to mock-immunised mice, suggesting more severe disease. Bacteraemia in DTPw-immunised mice was variable, with two DTPw-immunised animals exhibiting high bacterial loads 24 h post-surgery (Fig. 4E). Interestingly, bacteraemia was significantly higher in BCG-immunised mice compared to mock-immunised mice. Cytokines were quantified in peritoneal lavage fluid collected 24 h post-surgery and we found that DTPw-immunised mice had significantly higher levels TNFα and IL1β, but not IL6, in the peritoneal lavage fluid compared to mock-immunised mice (Fig. 4F). While elevated cytokine levels may be due to trained cytokine responses, they may also be secondary to bacteraemia and an inability to control infection. However, while BCG immunised mice had a significantly higher bacterial load, they did not exhibit significantly higher levels of TNFα, IL1β or IL6 compared to mock-immunised mice. These results suggest that DTPw vaccination could increase the severity of sepsis in mice, likely by amplifying cytokine responses during the acute phase.

Fig. 4. Immunisation with DTPw alters cytokine responses and acute illness severity in mice 24 h after the induction of polymicrobial sepsis.

(A) Female C57BL/6J mice were subcutaneously immunised with saline (mock), BCG (SSI) or DTPw (SII) and responses were boosted 2 weeks later. 6 weeks after the prime immunisation, sepsis was induced by cecal ligation and puncture. Clinical scores (B-C) and surface temperature (D) were assessed over a 24-hour period following the induction of sepsis, and area under the curve (AUC) was compared between groups. (E) Bacterial load in blood was quantified via aerobic culture. (F) Cytokine concentrations in peritoneal lavage fluid were assessed via multiplex immunoassay. Data are represented as the mean ± SEM and are from a single experiment. Statistical significance was assessed using a Kruskal-Wallis test with multiple comparisons. * = P < 0.05.

3.5. Immunisation with a pentavalent DTPw-Hep-Hib vaccine does not alter survival in a model of polymicrobial sepsis

Given the poorer clinical scores observed in DTPw-immunised mice in our pilot sepsis model, we next investigated whether mice immunised with a pentavalent DTPw-Hep-Hib vaccine had increased mortality over a 7 day period following the induction of sepsis by CLP. Due to the limited availability of trivalent DTPw, female mice (n = 25/group) were immunised at 4 weeks of age with pentavalent DTPw-Hep-Hib (Penta; EasyFive-TT) and compared to BCG or mock-immunised mice (Fig. 5A). Polymicrobial sepsis was induced by CLP at six weeks following the prime immunisations and clinical score and survival were monitored for 7 days following the induction of sepsis. Clinical scores (Fig. 5B) and temperature (Fig. 5C) were not significantly different between the groups over the assessment period. Induction of sepsis resulted in significantly increased plasma TNFα and IL6 (Fig. 5D) compared to sham-surgery mice. Serum TNFα was highest in the Penta-immunised group, however this was not significantly different from mock-immunised mice (Fig. 5D). The overall mortality rate was 20%, however, mortality was not significantly different between the groups (Fig. 5E). These data indicate that immunisation with either BCG or pentavalent DTPw-Hep-Hib does not alter susceptibility to sepsis induced by CLP in adult mice.

Fig. 5. Immunisation with the pentavalent DTPw-Hep-Hib vaccine does not alter survival in a cecal ligation and puncture model of polymicrobial sepsis.

(A) Female C57BL/6J mice (n = 25) were immunised and with saline (mock), BCG (SSI) or EasyFive® pentavalent DTPw-HepB-HiB vaccine (Penta) and sepsis induced 6 weeks later via cecal ligation and puncture surgery. (B) Clinical scores and (C) temperature were assessed for 7 days and the area under the curve (AUC) was compared between groups. (D) Cytokine concentrations in peripheral blood collected 24 h post-surgery were assessed via multiplex immunoassay. (E) Survival was recorded over 7 days. Data are represented as the mean ± SEM and are from a single experiment. Clinical scores, temperature and cytokine data were compared via Kruskal-Wallis test with multiple comparisons. Survival data were compared via a Mantel-Cox survival test. * = P > 0.05, ** = P > 0.01, *** = P > 0.001.

Previous studies have revealed that BCG protects neonatal, but not adult, mice from sepsis though emergency granulopoiesis [51]. To determine whether a pentavalent DTPw-Hep-Hib vaccine could influence susceptibility to sepsis in early life, we induced sepsis in immunised or mock-immunised 14 day old mice via a cecal-slurry injection. Neonatal (4 day old) mice were immunised 10 days prior to the induction of sepsis with saline (mock) or the Pentavac SII vaccine and boosted twice, in accordance with infant vaccination schedules (Fig. 6A). Clinical scores were elevated for 1–4 days following injection of the cecal-slurry, however clinical scores were not significantly different between Penta-immunised and unimmunised mice (Fig. 6B). Survival was not altered between the vaccine groups overall or in male or female mice when these were analysed separately (Fig. 6C–E). These data suggest that in adult or neonatal mice susceptibility to sepsis is not significantly altered by prior vaccination with Pentavalent DTPw-containing vaccines.

Fig. 6. Immunisation with a pentavalent DTPw-HepB-HiB vaccine does not alter survival in a cecal slurry injection model of neonatal sepsis.

(A) Neonatal mice (4–5 days old) were immunised with saline (mock) or Pentavac® SII DTPw-HepB-HiB vaccine (Penta) and similarly boosted 5 and 10 days later. Sepsis was induced via intraperitoneal cecal slurry injection 10 days following the prime immunisation, at 14 days old. (B) Clinical scores were monitored every 6–12 h for 7 days. Survival curves of all mice (C), and male (D) and female (E) mice are shown. Data are represented as the mean ± SEM and are from a single experiment. Survival data were compared via Mantel-Cox survival test. Clinical scores were compared via a Wilcoxon Rank-Sum Test with multiple comparisons. ns = not significant.

4. Discussion

Increasing evidence suggests that in addition to providing specific immunity against targeted pathogens, vaccines can confer non-specific effects (NSEs) - effects on morbidity and mortality unrelated to the targeted disease. The NSEs conferred by BCG are explained in part by the induction of trained immunity, which reprograms innate immune cells for increased responses to subsequent microbial encounters, however the potential mechanisms underpinning NSEs conferred by the DTPw vaccine are poorly understood.

Here, we report that the BCG SSI and DTPw vaccines each induce unique programs of trained immunity in monocytes and DCs in mice, resulting in heightened cytokine responses to a diverse panel of microbial stimuli even months after vaccination. We found that immunisation with BCG was associated with increased TNFα production by monocytes, while DTPw tended to lead to increased production of IL6 in DCs, following stimulation. These different programs of trained immunity are likely induced through differential sensing of the BCG and DTPw vaccines by pattern recognition receptors (PRR) on innate immune cells. BCG-induced trained innate immunity is dependent on the NOD2 receptor [25], which senses bacterial peptidoglycans. Though it is unknown how DTP vaccines may induce trained immunity, B. pertussis contains a host of immunomodulatory virulence factors and ligands capable of signalling through multiple PRRs including TLR2, TLR4, NOD2 and CR3 [59].

Trained immune cytokine responses were also highly dependent on the microbial stimulus used to stimulate splenocytes. For example, increased TNFα responses in BCG immunised mice were particularly apparent following stimulation with HKCA, a fungal pathogen, but less apparent or not significantly different compared to mock-immunised mice following stimulation with LPS or R848 + PolyI:C. This mirrors reports in human studies, which have found that PBMC trained immune responses are stimulant-dependant [52], and is likely due to different PRR agonists inducing cytokine release through different cell signalling pathways. HKCA contains β-glucans, which engage the Dectin-1 receptor to stimulate innate cell cytokine responses via NF-κB activation in a MyD88-independent manner [60]. LPS is detected via the TLR4 receptor, whereas R848 and PolyI:C are ligands for endosomal TLR7/8 and TLR3, respectively, inducing cytokine expression via a MyD88 dependent pathway [61]. Vaccine-induced trained immunity may influence one or several mediators of these pathways to alter immune responses in a stimulant-dependant manner.

We also uncovered evidence of epigenetic remodelling of DCs in immunised mice three months following vaccination suggesting that altered responses are epigenetically programmed. In addition, DCs and their precursors were more frequent in bone marrow of DTPw-immunised mice compared to mock-immunised controls, indicating that DTPw-induced trained immunity in DCs involves a remodelling of the haematopoietic compartment. Importantly, our data show that DCs can undergo training following vaccination, resulting in altered cytokine responses in a similar manner to that previously described in monocytes, NK cells and neutrophils [26]. As DCs are important for integrating innate and adaptive immune responses, vaccine-induced DC training could be potentially harnessed in future to optimise adaptive responses [25,29].

Given epidemiological data suggesting that the most recent vaccine administered can alter vaccine NSEs³, we assessed whether changing the order in which vaccines are administered alters trained immunity and found that vaccine administration order had a significant effect on trained cytokine responses. For example, administering both BCG and DTPw vaccines (2 weeks apart) resulted in trained immune responses that were different compared to when either vaccine was administered on its own. Mice immunised with BCG two weeks before or after DTPw had a significantly lower proportion of DCs producing IL6 compared to mice immunised with DTPw alone. These data are consistent with a recent clinical study demonstrating that simultaneous administration of BCG with the acellular DTP (DTaP) vaccine led to reduced IL6 responses compared to DTaP administration alone [52]. These findings suggest that vaccine-induced trained immunity is a complex process where innate immune cells are educated by vaccine history, and presumably encounters with other pathogenic [62] or commensal [63] microbes, determining how cells will ultimately respond to subsequent microbial encounters.

Mounting epidemiological data suggest that immunisation with DTPw vaccine may be associated with increased all-cause mortality in specific populations for reasons that are poorly understood. We hypothesised that DTPw-induced trained immunity influenced susceptibility to sepsis, a leading cause of infant mortality globally. To assess this, we first conducted a pilot study of polymicrobial sepsis in immunised mice and found increased disease severity and cytokine responses associated with DTPw vaccination. Subsequent experiments, however, using a pentavalent DTPw-Hep-Hib (Penta) vaccine did not reveal any significant impact of prior immunisation on mortality in a lethal polymicrobial sepsis model. While our findings suggest that DTPw-induced trained immunity does not enhance susceptibility to sepsis in mice, we cannot definitively rule out that the DTPw vaccine could detrimentally reprogram the immune response in human infants with sepsis. It is also possible that immunisation with DTPw could influence susceptibility to other infections that represent significant causes of mortality in infant populations, such as respiratory infections. Further studies are needed to investigate whether DTPw-induced trained immunity influences disease susceptibility in other contexts. Given that we observed some differences between mice immunised with trivalent versus pentavalent DTPw vaccines, further studies are needed to investigate if pentavalent DTPw-containing vaccines induce trained immunity in a similar manner to trivalent ones.

In conclusion, our data suggest that DTPw can induce trained immunity in murine monocytes and DCs, resulting in altered cytokine responses that last at least three months, providing a possible immunological rationale for the divergent NSEs associated with the BCG and DTPw vaccines. Importantly, these effects were partially negated by immunisation with BCG two weeks before or after DTPw, supporting the case for inclusion of live vaccines in vaccine schedules to mitigate against any potentially detrimental responses induced by non-live vaccines, such as DTPw.

Abbreviations:

ATAC

Assay for Transposase-Accessible Chromatin

BCG

Bacillus Calmette–Guérin

CMP

Common Myeloid Progenitor

DOL

Day of life

DTPw

Diphtheria-Tetanus-Whole Cell Pertussis

HKCA

Heat-killed Candida albicans

HKSA

Heat-killed Staphylococcus aureus

HPC

Hours post challenge

HSC

Haematopoietic Stem Cell

ICS

Intracellular cytokine staining

IL6

Interleukin 6

LPS

lipopolysaccharide

NSE

Nonspecific effects

R848

Resiquimod

TNFα

Tumour necrosis factor alpha