Adjuvant pretreatment with alum protects neonatal mice in sepsis through myeloid cell activation

Summary

The high mortality in neonatal sepsis has been related to both quantitative and qualitative differences in host protective immunity. Pretreatment strategies to prevent sepsis have received inadequate consideration, especially in the premature neonate, where outcomes from sepsis are so dismal. Aluminium salts‐based adjuvants (alum) are used currently in many paediatric vaccines, but their use as an innate immune stimulant alone has not been well studied. We asked whether pretreatment with alum adjuvant alone could improve outcome and host innate immunity in neonatal mice given polymicrobial sepsis. Subcutaneous alum pretreatment improves survival to polymicrobial sepsis in both wild‐type and T and B cell‐deficient neonatal mice, but not in caspase‐1/11 null mice. Moreover, alum increases peritoneal macrophage and neutrophil phagocytosis, and decreases bacterial colonization in the peritoneum. Bone marrow‐derived neutrophils from alum‐pretreated neonates produce more neutrophil extracellular traps (NETs) and exhibit increased expression of neutrophil elastase (NE) after in‐vitro stimulation with phorbol esters. In addition, alum pretreatment increases bone marrow and splenic haematopoietic stem cell expansion following sepsis. Pretreatment of neonatal mice with an alum‐based adjuvant can stimulate multiple innate immune cell functions and improve survival. These novel findings suggest a therapeutic pathway for the use of existing alum‐based adjuvants for preventing sepsis in premature infants.

Introduction

Despite significant advances in critical care medicine, morbidity and mortality from neonatal sepsis remains high. Premature (< 37 weeks gestation) and low and very low birth weight (LBW/VLBW) neonates are most susceptible to adverse outcomes from neonatal sepsis, which represents a significant public health problem costing in excess of US$3.2 billion per year 1, 2. Neonates, in contrast to adults, are heavily reliant upon their innate immune system and less so the adaptive immune response for protective immunity 3. Furthermore, the neonate's innate immune response is qualitatively different from either the adolescent or adult 4. This reliance upon innate immunity and well‐documented defects in adaptive immunity, which extend beyond the neonatal period, contributes to the increased mortality from sepsis in this vulnerable population 5, 6.

Polymorphonuclear leucocytes (PMNs) are the main effector cells of the innate immune system using phagocytosis and degranulation of anti‐microbial enzymes to clear pathogens at the site of infection or inflammation. Neutrophils, the most abundant PMN, migrate from blood to the site of infection/inflammation, where they phagocytize rapidly and kill bacteria. More recently, neutrophil extracellular traps (NETs) have been recognized as an important anti‐microbial mechanism in host defence 7, and their formation and function require extended stimulation in neonatal PMNs 8, 9. Neonates have reduced production of neutrophil progenitor cells and neutrophil storage pools compared to adults 10. This reduction in granulopoiesis is often characterized clinically in human preterm neonates by a neutropenic response to infection and sepsis 11. In addition to quantitative deficiencies, neonatal PMNs are characterized further by functional defects including reduced chemotaxis and phagocytosis, and decreased production of reactive oxygen species (ROS) 8, 11. Conversely, monocytes and macrophages are also involved in the host protective response to sepsis, and these cells are immature in newborns 12. Recently, it has been shown that the transcription pattern in monocytes from neonates during the first few days of life is opposite to the adult monocyte response after stimulation, with a strong induction of the Toll/interleukin (IL)‐1R domain‐containing adaptor inducing interferon (IFN)‐β (TRIF)‐dependent regulatory genes, and a weak myeloid differentiation primary response 88 (MyD88)‐dependent proinflammatory response induced mainly by the over‐production of S100 alarmins 13.

As mortality from sepsis in neonates, especially the premature, is so dismal, we have sought practical means to stimulate host‐protective immunity and prevent and/or reduce the severity of sepsis in the neonate, the concept being that prophylactic exposure of the neonate to innate immune adjuvants would improve protective immunity and reduce the likelihood of infections. We have demonstrated that innate immune function and host survival can be improved significantly when neonatal mice are pretreated with the Toll‐like receptor (TLR) agonists lipopolysaccharide or resiquimod 14. Although other TLR agonists, such as complete Freund's adjuvant and monophosphoryl lipid A (MPLA), have also been employed as adjuvants, both are used infrequently in paediatric vaccines. In contrast, aluminium salts (alum), alone or combined with TLR agonists, are both Food and Drug Administration (FDA)‐approved and used commonly in paediatric vaccines.

For almost a century alum has been used as adjuvant to enhance the adaptive immune response to antigens contained in human vaccines. Despite a significant amount of published literature describing the inflammatory effects of alum, controversy remains regarding the mechanisms of alum‐induced protective immunity, as alum appears to have little effect on T helper type 1 (Th1) responses 15. Some studies suggest that alum enhances antigen uptake by dendritic cells (DCs), enhances cell recruitment to the injection site and promotes immune cell function mainly via an inflammasome‐dependent mechanism 16. Nevertheless, alum has also been shown to promote the innate immune response via the phosphatidylinositol‐4,5‐bisphosphate 3‐kinase (PI3k) signalling pathway by promoting cell death and the release of alarmins (DAMPs) and by stimulation of IL‐1 production via caspase‐1 17, 18.

To develop alum further as a potential immune stimulant for the neonate at risk of sepsis, we used a murine model of polymicrobial neonatal sepsis 19. We demonstrate here for the first time, to our knowledge, that subcutaneous alum pretreatment alone improved survival in neonatal sepsis by promoting innate immune effector cell function. Following alum pretreatment, improved macrophage and neutrophil phagocytosis and increased formation of NETs were seen ex vivo following neonatal intra‐abdominal sepsis. The survival advantage conferred by alum pretreatment was independent of the adaptive immune response in neonates, but was caspase 1/11‐dependent. Finally, bone marrow and extra‐medullary myelopoiesis following neonatal sepsis was also improved following alum administration. Taken together, these results suggest that alum offers a clinically applicable, safe and well‐accepted prophylactic modality to prevent sepsis in a highly vulnerable population.

Materials and methods

Murine models

All studies were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Florida. Six to 8‐week‐old male and female pathogen‐free C57BL/6 [B6; wild‐type (WT)], Rag‐1 knock‐out (B6.129S7‐Rag1^tm1Mom/J back‐crossed onto a B6 background) and caspase‐1/11 null (B6N.129S2‐Casp1^tm1Flv/J back‐crossed onto a B6 background) were purchased from Jackson Laboratory (Bar Harbor, ME, USA). To generate neonatal mice, harem‐breeding schemes were established with separation of pregnant females once pregnancy was identified visually. Pregnant females were monitored daily to record an accurate date of birth. Mixed‐gender litters aged 5–7 days were used for experimental procedures and defined as neonates. In every case, individual neonates from within the same litters were assigned to the different treatment groups, including controls. This was performed to control variation due to differences in maternal care and the administration of slurry from different donors. Results obtained and presented were from at least two and commonly more litters. The actual number of neonates in each treatment group is presented.

Polymicrobial intra‐abdominal neonatal sepsis was induced using the cecal slurry (CS) model, as described previously 19. Briefly, fresh cecal contents were harvested from euthanized 6–8‐week‐old adult female B6 ‘donor’ mice, and were suspended in 5% dextrose solution to achieve a final concentration of 80 mg/ml. Subsequently, 1·1 mg/g BW of CS was administered in WT mice and 1·7 mg/g BW in caspase 1/11 null mice via intraperitoneal (i.p.) injection within 15 min of harvest to achieve the desired lethal dose (LD), as determined previously 19. Caspase‐1/11 null mice required a higher dose of CS to produce the same lethal dose seen in WT mice due to their well‐known resistance to polymicrobial sepsis 20, 21. Neonatal mice were pretreated 24 h prior to CS challenge with 100 μg of sterile, injection‐grade aluminium hydroxide and magnesium (alum) (Imject®; Thermo Fisher Scientific, Rockford, IL, USA) administered subcutaneously (s.c.).

Survival was assessed daily for 7 days. Neonates were evaluated twice daily, and moribund animals were euthanized. Animals that required euthanasia were deemed to be non‐survivors.

Peritoneal lavage and spleen collection for phenotypic analysis

Mouse peritoneal wash and spleen were collected at 0 (no treatment: NT or alum pretreated: A‐PT) 6, 18 and 24 h following CS administration. Peritoneal cells were isolated as described previously 14. Spleens were harvested and dissociated using a 70 µm sterile cell strainer (BD Falcon, Durham, NC, USA), red blood cells were lysed using ammonium chloride lysis buffer. Cell suspensions of peritoneal wash and splenocytes were stained with anti‐lymphocyte antigen 6 complex locus C1 (Ly6C) fluorescein isothiocyanate (FITC), anti‐CD11b phycoerythrin cyanin 7 (PE‐Cγ7), anti‐Ly6G allophycocyanin (APC), anti‐CD11c APC 780 and anti‐F4/80 PE (eBioscience, San Diego, CA, USA). Expression of co‐stimulatory molecules CD80 (anti‐CD80 FITC) and CD86 (anti‐CD86 PE) in monocytes and neutrophils was analysed on CD11b‐positive and Ly6G‐positive cells, respectively. Antibodies were purchased from Becton Dickinson (BD, Franklin Lakes, NJ, USA) unless indicated otherwise. Samples were acquired and analysed on an LSRII flow cytometer (BD) and FACSDiva™ software (BD). A minimum of 1 × 10⁴ live cells were collected for analysis. Sytox Blue (Invitrogen, Carlsbad, CA, USA) was used for cell viability analysis. Absolute numbers of cells were determined by multiplying the percentage of monocytes (Ly6G–Ly6C+CD11b+), neutrophils (Ly6G+CD11b+) and macrophages (F4/80+CD11b+) within the total sample population by the total sample cell number.

Functional analysis (phagocytosis) of peritoneal macrophages and neutrophils

At time‐points 0, 6, 18 and 24 h following sepsis, peritoneal cell suspensions containing 1 × 10⁵ cells were incubated with pHrodo green Escherichia coli BioParticles™ (Invitrogen) and incubated at 37°C in a water bath for 1·5 h. Cells were washed with cold phosphate‐buffered saline (PBS) and stained for macrophages (anti‐F4/80 PE and anti‐CD11b PE‐Cγ7) and neutrophils (anti‐Ly6G APC and anti‐CD11b PE‐Cγ7). Phagocytosing cells were identified as FITC‐positive and quantified by flow cytometry using a LSRII flow cytometer.

Bacterial count determination

Peritoneal lavage was harvested to evaluate bacteria colonization as described previously 19. Briefly, 100 µl of serially diluted peritoneal wash was plated on sheep's blood agar plates (Thermo Fisher Scientific, Waltham, MA, USA) and incubated at 37°C for 24 h following sepsis. Plates were counted following 24 h incubation.

Determination of circulating cytokine response following polymicrobial sepsis

Blood samples were collected at 0, 6, 18 and 24 h following sepsis. Circulating cytokine concentrations were determined in plasma by a customized mouse cytokine magnetic bead panel Milliplex^® map kit (EMD Millipore, Billerica, MA, USA), including IL‐1α, IL‐1β and tumour necrosis factor (TNF)‐α magnetic bead panels following the instructions of the manufacturer.

Isolation of bone marrow‐derived neutrophils

Bone marrow was harvested from bilateral femurs and tibias, pooled and disassociated through a 70‐µm sterile filter into RPMI‐1640 medium supplemented with 10% fetal calf sera (FCS) and 1% penicillin/streptomycin. After osmotic haemolysis and wash, neutrophils were isolated using the Mouse Neutrophil Enrichment kit (Easysep, Stemcell Tech., Vancouver, Canada). Purity was assessed by flow cytometry using anti‐Ly6G APC and anti‐CD11b PE‐Cγ7 antibodies.

Ex‐vivo stimulation of PMN and NETs release

Freshly isolated neutrophils (3·5 × 10⁵/well) were seeded on round coverslips in 24‐well plates for 30 min at 37°C and 5% CO₂. Neutrophils were stimulated with 75 nM phorbol 12‐myristate 13‐acetate (PMA) (Sigma‐Aldrich, St Louis, MO, USA) for 3·5 h at 37°C and 5% CO₂. Following incubation, cells were fixed with methanol at −20°C and permeabilized with 0·3% Triton X‐100. Blocking was performed with normal goat serum (Life Tech Corp., Carlsbad, CA, USA). Coverslips were incubated overnight with anti‐neutrophil elastase antibody (Abcam, Cambridge, MA, USA), washed with PBS and incubated with a goat anti‐rabbit immunoglobulin (Ig)G Alexa Fluor 488 conjugated antibody (Life Tech Corp., Carlsbad, CA, USA) at room temperature. Cells were counterstained with 1 µg/ml PureBlu™ Hoechst 33342 (Bio‐Rad, Richmond, CA, USA) and slides were mounted in 5% n‐Propyl Gallate/PBS. Images were captured using a fluorescence microscope equipped with an AxioCam MRm3 camera using AxioVision^® version 4.8 software (Zeiss Axioscop 2 Plus Fluorescence; Carl Zeiss International, Oberkochen, Germany). A minimum of 600 cells were analysed using ImageJ^® software (NIH, Bethesda, MD, USA) and multiple images captured per slide. NETs were characterized as linear and/or reticular structures anchored to neutrophils, as described previously 22. The percentage of neutrophil elastase‐positive cells (NE+) and neutrophil extracellular traps producing cells (NETs+ cells) was also calculated.

Isolation of splenocytes and bone marrow for Lin^–Sca‐1⁺ c‐kit⁺ (LSK) progenitor cell analysis

Spleen and bone marrow were harvested as described. Cells were stained with mouse biotin lineage antibody cocktail (anti‐Gr1, anti‐B220, anti‐Ter119, anti‐CD3e and anti‐CD11b) and labelled with anti‐streptavidin peridinin chlorophyll (PerCP)‐Cy5.5, anti‐C‐kit APC and anti‐Sca‐1 PE‐Cγ7. A minimum of 1 × 10⁴ non‐debris living cells were used and analysis was performed by flow cytometry.

Statistical analysis

All statistical analyses were assessed using GraphPad Prism software version 6 (GraphPad Software Inc., La Jolla, CA, USA). Differences in 7‐day survival were calculated using the log‐rank (Mantel–Cox) test. The means for three or more groups were compared using one‐way analysis of variance (anova) with either Tukey's post‐hoc analysis or Student's t‐test. Values are presented as mean ± standard error of the mean (s.e.m.) for parametric testing. For plasma cytokine measurements that fail tests of normality, values are presented as medians and intraquartile ranges. Differences among groups were determined using Wilcoxon's and Spearman's ranked tests. However, in all cases, significance was determined at the 95% confidence level.

Results

Outcomes

Alum pretreatment improves survival in neonatal intra‐abdominal sepsis

Neonatal mice were treated with alum (100 µg s.c. on the midscapular dorsum) 24 h prior to induction of polymicrobial sepsis (CS). Neonatal mice that received 5% dextrose i.p. did not show any mortality. Alum pretreated WT neonatal mice had significantly improved survival compared to their non‐treated littermates following sepsis (P < 0·05) (Fig. 1). To investigate further whether or not improved survival following alum pretreatment in neonatal sepsis required the adaptive immune response, we repeated survival studies using Rag‐1 null neonatal mice. Rag‐1 null mice do not develop functional B or T lymphocytes. Mortality in Rag‐1 null neonates following CS administration was markedly but not significantly greater than observed in WT mice. These findings differ somewhat from earlier studies, which showed no difference in survival between WT and Rag‐1 null mice given CS 14. Although an explanation for the differences is not easily forthcoming, differences in the microbiome can affect both the microbial pattern of the CS as well as the microbiome‐induced maturation of the neonatal immunity. Differences in diet, housing and seasonal variations may all contribute. Importantly, alum pretreatment also improved 7‐day survival significantly in Rag‐1 null mice following sepsis (P < 0·01) (Fig. 1).

Figure 1.

Aluminium salts (alum) pretreatment enhances survival in neonatal mice with sepsis induced by cecal slurry (CS). Sepsis‐survival was determined in wild‐type (WT), Rag‐1 null and caspase‐1/11 null neonatal mice (5–7 days). Neonates received either alum [100 µg subcutaneously (s.c.)] or no treatment (NT) 24 h prior to being challenged with an intraperitoneal (i.p.) injection of CS [lethal dose (LD)_40–60] to induce polymicrobial sepsis. Survival was assessed for 7 days after sepsis. Results shown are combined data from three independent experiments with different litters. *P < 0·05; **P < 0·01 versus CS (log‐rank Mantel–Cox test). Dextrose controls (white circle) did not show mortality. WT + CS (white square) and WT + aluminium (alum) + CS (black square), n = 30 each group; Rag‐1 null + CS (up‐pointing triangle) and Rag‐1 null + alum + CS (down‐pointing triangle), n = 26 in each group; caspase‐1/11 null + CS (black diamond) and caspase‐1/11 null + alum + CS (white diamond), n = 21 in each group.

Alum activates the inflammasome in macrophages and bone marrow‐derived DC in vitro 23. The inflammasome is an intracellular group of large multi‐protein complexes that detect pathogen‐associated and damage‐associated molecular patterns (PAMPs and DAMPs, respectively) 24. Therefore, in order to elucidate whether the survival advantage in alum pretreated neonatal mice is dependent upon the inflammasome, we repeated our survival studies in caspase‐1/11 null mice. Interestingly, caspase‐1/11 null mice are more resistant to polymicrobial sepsis than WT animals 20. The same dose of CS used in WT mice (1·1 mg/kg BW) produced no mortality in caspase‐1/11 null mice (data not shown), and the dose of CS had to be increased to 1·7 mg/kg to produce the same degree of mortality as seen in WT mice administered 1·1 mg/kg BW CS. As shown in Fig. 1, alum pretreatment at this dose of CS did not improve survival in the caspase‐1/11 null neonatal mice following sepsis when compared with caspase‐1/11 null littermates not pretreated with alum (CS only).

As alum did not improve survival in the caspase‐1/11‐deficient neonatal mice, we sought to determine whether the inflammasome was required for production of the cytokines IL‐1α and IL‐1β in WT neonatal mice. Alum pretreatment increased IL‐1α significantly 24 h after CS (Fig. 2a; P < 0·05 versus CS 24 h). Conversely, IL‐1β peaked 6 h after sepsis in the alum pretreated group when compared with controls (NT, prior to the induction of sepsis) (Fig. 2b; P < 0.01 versus NT, time 0) while no change was observed in the CS group at 6 h. At 18 and 24 h after sepsis, IL‐1β concentration decreased to control levels in both groups, but the magnitude of the differences with alum treatment was modest, at best. Whether these changes in IL‐1 concentrations have any biological relevance is unclear. TNF‐α production increased significantly 18 h after sepsis (Fig. 2c; P < 0·01 versus NT). Although alum pretreatment increased the production of TNF‐α significantly at 6, 18 and 24 h after sepsis when compared with controls (NT, time 0), it did not reach statistical difference when compared with septic mice at different time‐points (Fig. 2c). These findings suggest that alum pretreatment increases the magnitude of the cytokine response only modestly, and it is unlikely that the improvements in outcome are dependent upon the cytokine inflammatory response.

Figure 2.

Aluminium salts (alum) promote cytokine production following polymicrobial sepsis. Plasma was collected at 0 [no treatment (NT)], prior to sepsis or 24 h after alum pretreatment (A‐PT), 6, 18 and 24 h after intraperitoneal (i.p.) cecal slurry (CS) injection [lethal dose (LD)_40–60] and assayed for (a) interleukin (IL)‐1α, (b) IL‐1β and (c) tumour necrosis factor (TNF)‐α using a mouse cytokine magnetic bead panel Milliplex^® map kit. Results are presented as median and interquartile range; n = 8–12 per group from four different experiments. *P < 0·05 versus CS 24 h; **P < 0·01 and ***P < 0·001 versus NT).

Alum pretreatment improves phagocytosis in peritoneal macrophages and neutrophils during neonatal sepsis

Because alum improved survival in mice lacking a functional adaptive immune system, we focused upon aspects of innate immunity required for host protection. Subcutaneous alum pretreatment had no effect on the number of peritoneal macrophages (F4/80+CD11b+ cells) recruited at 0, 6, 18 and 24 h following polymicrobial sepsis (Fig. 3a). However, their functional ability was improved, as the percentage of recruited macrophages capable of phagocytosis increased significantly following alum treatment immediately prior to sepsis (Fig. 3b; P < 0·01) and 6 h after sepsis (Fig. 3b; P < 0·05). Alum pretreatment induced less recruitment of peritoneal neutrophils (Ly6G+CD11b+ cells) 24 h after sepsis (P < 0·05; Fig. 4a); however, the percentage of peritoneal neutrophils phagocytosing also increased in the alum pretreated group following polymicrobial sepsis at 18 and 24 h after CS (P < 0·001; Fig. 4b). As expected, with increased phagocytic activity in peritoneal cells following alum pretreatment we observed significantly fewer numbers of bacterial colony‐forming units (CFUs) in the peritoneal lavage of alum‐pretreated neonatal mice 24 h following CS challenge (P < 0·05; Fig. 5).

Figure 3.

Aluminium salts (alum) pretreatment (A‐PT) improve phagocytosis in peritoneal macrophages. Peritoneal lavage cells were harvested at 0 [no treatment (NT) or A‐PT] 6, 18 and 24 h after the induction of intra‐abdominal polymicrobial sepsis. Cells were stained with anti‐F4/80 and anti‐CD11b antibodies, and quantified by flow cytometry (a). The percentage of macrophages phagocytosing (b) was determined by incubating 1 × 10⁵ peritoneal live cells with pHrodo Green Escherichia coli BioParticlesTM for 1·5 h at 37°C and then analysed by flow cytometry. Values represent mean cells ± standard error of the mean (s.e.m.); n = 6 at each time‐point from four different experiments. **P < 0·01 at 0 h (NT or A‐PT); *P < 0·05 at 6 h following sepsis.

Figure 4.

Aluminium salts (pretreatment) (A‐PT) improve peritoneal neutrophil phagocytosis following polymicrobial sepsis. Peritoneal wash was harvested at 0 [no treatment (NT) or A‐PT] 6, 18 and 24 h after the induction of polymicrobial intra‐abdominal sepsis. Cells were stained for neutrophils with anti‐lymphocyte antigen 6 complex locus G6D (Ly6G) and anti‐CD11b antibodies and quantified using flow cytometry (a). Phagocytic activity (% phagocytic neutrophils) was assayed by incubating 1 × 10⁵ peritoneal live cells with pHrodo Green Escherichia coli BioParticlesTM for 1·5 h at 37°C and then analysed by flow cytometry (b). Values represent mean ± standard error of the mean (s.e.m.) from four independent experiments; n = 6 (0 and 6 h after sepsis) and n = 10 at 18 and 24 h after sepsis. ***P < 0·001; *P < 0·05.

Figure 5.

Reduction in bacterial colonization 24 h in the aluminium salts (alum) pretreated neonatal mice compared with cecal slurry (CS). Neonatal mice were treated with 100 μg of alum subcutaneously (s.c.) 24 h before being challenged with intra‐abdominal polymicrobial sepsis (CS). Peritoneal lavage was collected 24 h following sepsis, serially diluted and plated on sheep's blood agar. Plates were counted after 24 h of incubation at 37°C. Colony‐forming units (CFUs) were quantified and analysed by Mann–Whitney t‐test; n = 12 (CS) and n = 8 (alum + CS) from two different experiments. *P < 0·05.

Alum pretreatment improves production of NETs

As alum pretreated mice exhibited increased phagocytic activity and improved bacterial clearance, we subsequently evaluated the role of alum pretreatment on NETs production ex vivo, as these nuclear material extrusions both capture and kill extracellular pathogens. Bone marrow‐derived neutrophils from alum pretreated and non‐treated neonatal mice were harvested and stimulated ex vivo using phorbol ester (PMA). After 3·5 h, we observed an increased number of NET‐positive and neutrophil elastase (NE)‐positive PMNs isolated from the alum pretreated mice when compared with the PMNs isolated from untreated neonatal mice (P < 0·05; Fig. 6).

Figure 6.

Effect of aluminium salts (alum) pretreatment on neutrophil extracellular traps (NETs) production (a) and neutrophil elastase expression (b) ex vivo. Bone marrow neutrophils were isolated through negative selection using the mouse neutrophil enrichment kit (EasySep, StemCell Tech) and pooled into two groups (n = 5 each): alum pretreated (A‐PT) and non‐treated (NT) neonatal mice (7 days old). Neutrophils were seeded on poly‐L‐lysine coated coverslips (3·5 × 10⁵ cells per well) and stimulated with 75 nM phorbol myristate acetate (PMA) for 3.5 h at 37°C. Indirect immunostaining was performed to detect NETs using anti‐neutrophil elastase (NE) expression [fluorescein isothiocyanate (FITC)]. DNA was stained with Hoechst 33342. Ex‐vivo NETs were characterized as linear and/or reticular structures anchored to neutrophils using a fluorescence microscope Zeiss Axioscop 2 Plus fluorescence. Representative fluorescence microscopy images of NT (c) or A‐PT (d) PMA‐stimulated neutrophils showing NETs production (arrows); ×40 objective. The data were obtained from three independent experiments and are presented as mean ± standard error of the mean (s.e.m.). *P < 0·05.

Alum modulates the expression of co‐stimulatory molecules in polymicrobial neonatal sepsis

Although adaptive immunity is not a significant contributor to outcome in neonates, up‐regulation of the co‐stimulatory receptors CD80 and CD86 on antigen‐presenting cells can be used as a biomarker of the activation status of myeloid populations. As documented previously in sepsis induced by cecal ligation and puncture 25, CD80 was up‐regulated on neonatal splenic monocytes and neutrophils 18 and 24 h after sepsis (P < 0·01; Fig. 7a and b, respectively). Similarly, CD86 was up‐regulated at 24 h on monocytes and neutrophils in the spleen (P < 0·01 and P < 0·0001, respectively; Fig. 7c,d) following neonatal sepsis. Alum pretreatment induced a marked over‐expression of CD80 on splenic monocytes and neutrophils (Fig. 7a and b); conversely, alum induced an early increase in CD86 expression on splenic monocytes (P < 0·05, Fig. 7c) and 24 h after sepsis in neutrophils (P < 0·01, Fig. 7d). Similarly, in whole blood, alum pretreatment induced increased expression of CD80 and CD86 in monocytes and neutrophils 24 h after sepsis while peritoneal neutrophils showed increased CD86 expression at 24 h following sepsis (data not shown).

Figure 7.

Expression of CD80 and CD86 after induction of abdominal polymicrobial sepsis. Five to 7‐day‐old wild‐type (WT) mice received aluminium salts (alum) subcutaneously (s.c.) 24 h prior to intraperitoneal (i.p.) challenge with cecal slurry (CS). Mice were euthanized at 0 (prior to CS) 6, 18 and 24 h after sepsis and spleens were harvested for the analysis of the splenic expression of the co‐stimulatory molecules CD80 and CD86 on CD11b+ cells and lymphocyte antigen 6 complex locus G6D (Ly6G)+ cells using flow cytometry. (a) CD80 expression on splenic CD11b⁺ cells and (b) Ly6G⁺ cells. ###P < 0·001; ##P < 0·01 versus CS at 0 h; ***P < 0·001; **P < 0·01; *P < 0·05. (c) CD86 expression on splenic CD11b⁺ cells and (d) Ly6G⁺ cells. ###P < 0·001 versus CS at 0 h; **P < 0·001; *P < 0·05; n = 5 per time‐point from four independent experiments.

Alum pretreatment improves neonatal emergency myelopoiesis

Emergency myelopoiesis is critical for sustaining the host‐protective response following an inflammatory insult or sepsis. We have demonstrated previously that neonatal mice have impaired emergency myelopoiesis following polymicrobial sepsis compared to young adults, and this probably contributes to worse survival 26. As we demonstrated that alum could improve phagocytosis, we subsequently evaluated the effect of subcutaneous alum treatment on the number of progenitor haematopoietic stem cells (HSCs or LSKs) prior to sepsis (24 h after adjuvant administration) and at 24 and 36 h following polymicrobial sepsis. Alum pretreatment increased the percentage of LSKs in bone marrow (BM) significantly at baseline and 24 h after the induction of sepsis (P < 0·05; Fig. 8a). As the spleen has a primary haematopoietic function in neonates, we also evaluated alum pretreatment followed by septic challenge on extramedullary LSK production. Similar to our results in bone marrow, neonates pretreated with alum demonstrated increased extramedullary haematopoiesis at baseline and 36 h following polymicrobial sepsis (P < 0·05; Fig. 8b).

Figure 8.

Aluminium salts (alum) increased expansion of neonatal haematopoietic stem cells (HSCs) following polymicrobial sepsis. (a) Bone marrow (BM) and (b) splenocytes were harvested, stained and analysed for Lin^–Sca1^– c‐kit⁺ cells (LSKs or primitive HSCs) by flow cytometry at baseline [0 h, no treatment (NT) or alum pretreatment (A‐PT)] 24 and 36 h after cecal slurry (CS). Alum improved the expansion of bone marrow LSKs significantly at baseline and 24 h after CS challenge. Expansion of LSKs in the spleen was also improved in the alum pretreated group, at baseline (0) and 36 h after polymicrobial sepsis. All values are expressed as mean ± standard error of the mean (s.e.m.). *P < 0·05; n = 13 per group at 0 and 24 h; n = 8 per group at 36 h (four independent experiments).

Discussion

Aluminium hydroxide (alum) remains the most common and widely accepted adjuvant used today in the United States 16. Alum salts adjuvants have been used since the 1920s, although their mechanisms of action are still not well understood. When administered s.c. as part of a vaccine, alum has multiple properties. Alum has a depot effect where it prolongs the exposure of the primary antigen to host tissues. This property is of little interest to our studies. Alum is also phagocytosed by resident macrophages and DCs 23. As shown in this study, alum administered s.c. as in a vaccination protocol has dramatic stimulatory effects on multiple systemic components of innate immunity, making it a potential prophylactic therapy with minimal adverse potential. In fact, these studies argue strongly that current FDA‐approved alum adjuvants could become a primary therapy for preventing sepsis in very low birth weight neonates at increased risk of developing sepsis. Although optimal dosing and timing were not determined, the studies clearly show the potential therapeutic opportunities for this clinical approach.

In this study, we demonstrated that the benefit of alum pretreatment was independent of T and B cells. This finding suggests that alum enhances innate immune responses preferentially in neonates. One of the primary mechanisms of action of particulate adjuvants, including alum, is via activation of the nucleotide‐binding domain (NOD)‐like receptor protein 3 (NLRP3) inflammasome, promoting the release of the caspase‐1‐dependent proinflammatory cytokines IL‐1β, IL‐18 and IL‐33 23, 27. These members of the IL‐1 superfamily are expressed mainly by macrophages, as neutrophils are not a significant source of IL‐1β in response to alum, due to the inability of the NLRP3 inflammasome to respond to insoluble agents in neutrophils 28. Conversely, alum can activate the Syk tyrosine kinase pathway and modulate the immune response by inducing cytotoxic effects 29; first, it promotes an influx of innate immune effector cells to the site of the injection, and then the release of endogenous alarmins or DAMPs due to cell death 30, 31. Here we show that alum pretreatment does not improve survival further in caspase‐1/11 null neonatal mice; thereby, the beneficial effect of alum pretreatment is most probably dependent upon assembly of the inflammasome and activation of caspase‐1/11 32. This, of course, raises an interesting contradiction: if deletion of caspase‐1/11 improves survival to sepsis by itself, but blocks the improved survival to alum pretreatment, then the mechanism(s) of survival advantage to both caspase‐1/11 deletion alone and caspase 1/11‐dependent alum pretreatment are probably independent.

Although a complete resolution of this conundrum is beyond the scope of the current report, there is strong support for the conclusion that the inflammasome is required for the alum‐induced beneficial contribution to the host response to sepsis. Alum promotes the recruitment of monocytes, DCs and PMNs, antigen uptake, maturation and T cell stimulatory activity of DCs 33. In our model, where alum was administered s.c., alum pretreatment did not increase macrophage and neutrophil recruitment into the peritoneal cavity following polymicrobial sepsis. However, it altered the phenotype of the recruited cells, including increased activation status, phagocytosis and NET formation, with a significant reduction in the bacterial colonization. Recently, the global proteomic response of human neonatal monocytes to alum demonstrated the activation of phagocytosis‐related signalling (caveolar‐mediated endocytosis, MSP‐RON) when monocytes were stimulated with alum in vitro 34. Professional phagocytic cells such as macrophages are able to sense alum crystals in vitro via the activation of the NLRP3 inflammasome 35.

Another essential cell population required for protective immunity is the PMN, the first specialized phagocytic cell to migrate to the site of infection. In human neonates, multiple deficiencies have been described. First, neonates have decreased bone marrow neutrophil storage pool, which results in fewer neutrophils in response to an acute infection 10; secondly, the chemotactic response to the site of infection is decreased as a result of a lower expression of adhesion molecules such as Mac1 and L‐selectin, and finally the anti‐bacterial function has been shown to be deficient 11, 36.

After 18 and 24 h following polymicrobial sepsis, neonatal mice showed a significant recruitment of peritoneal neutrophils. Although there was no effect in the total number of neutrophils due to alum pretreatment, the number of peritoneal neutrophils phagocytosing increased significantly. In addition to an increased phagocytic activity in neutrophils in vivo in the alum pretreated mice, our ex‐vivo studies revealed that alum pretreated neutrophils produce more NETs after PMA stimulation when compared to untreated PMNs. Neutrophil extracellular traps are an important component of extracellular anti‐bacterial activity during an innate immune response 7. In human preterm or term neonates the neutrophil response is deficient, as it has been shown that human cord blood cells do not have the ability to produce NETs following stimulation with lipopolysaccharide (LPS) or PMA, contributing to decreased bacterial killing in vitro 8. Our ex‐vivo results suggest that alum potentiates the neutrophil response by increasing their ability to produce NETs and anti‐microbial enzymes, such as elastase, and also by increasing their phagocytic activity. An indirect response to alum is supported by the observation that neutrophil phagocytosis peaks after 18 h following sepsis once macrophage phagocytosis has peaked at 0 and 6 h.

In addition, neonates have delayed and impaired emergency granulopoiesis in response to sepsis compared with adults 26. In sepsis, neonatal mice have delayed myelopoiesis related probably to an increased expression of C‐X‐C motif chemokine ligand 12 (CXCL12), which has been reported previously in human septic neonates 20, 26, 37. The use of aluminium adjuvants suppresses CXCL12 expression in bone marrow, which correlates with the promotion of granulopoiesis 38. In addition, alum administration induces a local inflammatory response in the spleen 39, which could explain the increased extramedullary myelopoiesis shown in the alum‐pretreated mice. Once the mature granulocytes are activated their lifespan is short (hours to days); therefore, increasing granulopoiesis during the inflammatory response could be advantageous to the host as persistent neutropenia leads to death from overwhelming infection 40, 41.

Clinical significance

There are, at present, very few options for preventing infections in neonates who are born prematurely. Antibiotics are commonly over‐used because they represent the only accepted approach to preventing early and late septic events. The findings reported here suggest that administration of vaccine adjuvants containing alum alone may stimulate host‐protective immunity in premature infants. Importantly, our findings reveal that alum, one of the most commonly used and accepted adjuvants used currently in paediatric vaccines, can not only activate the innate immune cell effector functions in neonatal mice, but can also improve emergency myelopoiesis, which results in improved survival in sepsis. The current recommendation for alum in humans as a vaccine adjuvant is approximately 0·25 mg alum per kg BW. The dose of alum used in these studies was approximately 35 mg/kg BW, or 135 times the human dose. It will be interesting to determine whether the doses of alum used in current human vaccines contain sufficient adjuvant to activate innate immunity fully, or whether quantities in humans could be used to determine maximal innate immune functions. However, further exploration of the benefit of alum adjuvants for the prevention of sepsis in neonates appears warranted.

Disclosure

The authors declare no conflicts of interest.

Author contributions

J. C. R. designed and conducted experiments, data collection and interpretation, performed statistical analysis and manuscript preparation (drafting, editing and revisions). A. L. C. designed and conducted experiments, data collection and interpretation and manuscript review. S. L. R. and B. M. assisted with conducting experiments, data collection, statistical analysis and manuscript review. D. C. N. and R. U. assisted with conducting experiments and data collection, manuscript review. P. E. assisted with designing experiments, data interpretation and manuscript review. J. L. W. assisted with data interpretation and manuscript review. L. L. M. and S. D. L. designed experiments, performed data interpretation and statistical analysis and manuscript preparation (including drafting, editing and revisions).