Although Streptococcus pneumoniae is usually found as a commensal in healthy individuals, it can act as a pathogen in trauma patients, causing such complications as early-onset pneumonia and sepsis. We discovered that treating mice with an A-class CpG-oligodeoxynucleotide (ODN) at 2 h after traumatic injury significantly improved mouse survival following early-onset secondary lung infection with S. pneumoniae. This study used mass cytometry (cytometry by time-of-flight) and Luminex technologies to characterize the cellular immune response to secondary S. pneumoniae lung infection at 1 and 3 d postinfection. We found increased expression of CD14, CD64, and PD-L1 on F4-80+ and F4-80+CD11c+ macrophages, CD11c+ dendritic cells, and CD14+CD172a+ cells after burn-injury and infection, supporting previous reports of innate immune cell activation in sepsis. CpG-ODN treatment at 2 h after burn-injury reversed these effects; improved pathogen clearance; and led to an increased expression of CD25, CD27, MHCII, and IL-17 on or in TCRγδ cells at 1 d postinfection. At 3 d postinfection, CpG-ODN treatment increased the expression of PD-L1 on innate cell subsets. Furthermore, we analyzed cytokine levels in lung-washout samples of TCRγδ cell–depleted (TCRγδ−) mice to demonstrate that the effects of CpG-ODN on cytokine expression after burn-injury and S. pneumoniae infection rely on functional TCRγδ cells. In summary, we demonstrate that cytometry by time-of-flight provides an effective strategy to systematically identify specific cellular phenotypic responses to trauma and bacterial pneumonia and to discover changes in immune system phenotypes associated with beneficial immunotherapy.
This article is featured in In This Issue, p.519
Despite improvements in emergency care medicine, traumatic injuries remain the leading cause of death in people under the age of 45, and the third leading cause of death in all populations (1). A major complication of trauma is nosocomial infection, the most common cause of late death after trauma, occurring between 3 days and 3 weeks after severe traumatic injury (2). Spread of microorganisms to the lower respiratory tract by aspiration or during emergent intubation after traumatic injuries contributes to the development of early-onset pneumonia in trauma patients, increasing morbidity and treatment costs. Although late-onset nosocomial pneumonia is often caused by Pseudomonas aeruginosa or Staphylococcus aureus, pathogens more commonly found in early-onset pneumonia are Streptococcus pneumoniae and Haemophilus influenzae (3, 4).
To study potential beneficial effects of immune response–modifying drugs on survival after trauma and early-onset pneumonia, we developed and used a two-hit mouse model of burn-injury followed by secondary S. pneumoniae lung infection 1 d after injury. Using this model, we found that a single treatment with an A-class CpG-oligodeoxynucleotide (ODN) 2336 2 h after the injury significantly improved survival of mice given a secondary S. pneumoniae lung infection. Many studies have demonstrated that pattern recognition receptors like TLRs on innate immune cells are critical mediators of the early response to bacterial infection (5, 6). CpG-ODNs act as TLR9 agonists and were shown to have different immunostimulatory effects on innate cell subsets (7). Interest in their immune-modulatory behavior led to the synthesis, testing, and classification of four structurally different classes of CpG-ODNs. A-class CpG-ODNs, as used in our experiments, were shown to potentially elicit strong IFN-α production by plasmacytoid DCs (7, 8).
To determine how CpG-ODN mediates its beneficial effects in this two-hit response model, we decided to apply an inductive scientific approach, as described by D. Glass and N. Hall (9). Initially, we took advantage of the vast data collection capability of cytometry by time-of-flight (CyTOF) mass cytometry to phenotype the immune response at the site of infection. This approach seemed feasible because no meaningful hypothesis could be initially phrased to answer how CpG-ODN might induce its beneficial effects. By this strategy, we reasoned that a hypothesis would evolve from observations made during CyTOF experiments. Using CyTOF technology, we were able to quantify frequencies of immune cell subsets at the site of infection and also simultaneously analyze functional marker profiles on and in distinct cell subsets. As such, our CyTOF panels included Abs specific for cell-surface markers, transcription factors, and cytokines. We decided to take advantage of CyTOF for two major reasons. First, CyTOF technology uses metal isotope–labeled Abs instead of fluorochrome-labeled Abs. Sample analysis, therefore, is not affected by autofluorescent signal interference emanating from bacterial pathogens as well as activated macrophages and neutrophils, both of which are abundantly present in lung-washouts postinfection (10). Second, CyTOF allows for using larger Ab staining panels than does conventional flow cytometry, enabling us to screen for multiple cell phenotypical changes in a largely unbiased way (10).
The CyTOF panel we used for characterizing the cellular immune response to S. pneumoniae infection in lung-washout samples is shown in Table I. Again, it includes Abs used as cell subset–defining markers as well as functional markers, transcription factors, and cytokines. To visualize immune cell subsets involved in the systemic response to S. pneumoniae infection, we used visualization algorithm for stochastic neighbor embedding (viSNE) analysis. As previously described, viSNE allows for high-dimensional cytometry data to be displayed in two dimensions, while conserving the high-dimensional structure of the data (11). Color is used as a third dimension, allowing us to interactively visualize features of the two-dimensionally displayed cell populations. Because only basic gating steps are required to exclude cell-doublets and dead cells, this type of analysis is largely unbiased. On the basis of viSNE analysis, we developed a comprehensive gating strategy to further evaluate differences in functional marker and cytokine expression on immune cell subsets after CpG-ODN treatment at 1 d and 3 d after lung infection (Fig. 1).
|Marker .||Clone .||Isotope Label .|
|Marker .||Clone .||Isotope Label .|
To confirm and expand the findings from CyTOF analysis, we determined free cytokine levels in lung-washout samples at 1 d and 3 d postinfection, using Luminex assays. Consistent with previous reports, we were able to show that CpG-ODN increased the early inflammatory response of innate immune cell subsets to S. pneumoniae infection after burn-injury. Furthermore, we also detected increased activation of TCRγδ cells in the lung after CpG-ODN treatment. Because it is known that TCRγδ cells play a major role in immune surveillance and regulation of the immune response in microbial-exposed epithelia such as the lung (12), we used a model of TCRγδ-depleted mice (TCRγδ−) to evaluate whether beneficial effects of CpG-ODN treatment relied on the presence of TCRγδ cells. TCRγδ cells were found to be required for mediating CpG-ODN–induced, beneficial effects on cytokine production and pathogen clearance after trauma and secondary infection. Thus, this article identifies TCRγδ cells as central mediators of the beneficial activity of CpG-ODN treatment.
Materials and Methods
Outbred, male CD-1 mice were purchased from Charles River Laboratories (Wilmington, MA). All mice were maintained in our full-barrier animal facility under controlled temperature, humidity, and 12-h light–dark regimen, and were provided with standard chow and water ad libitum. Mice were acclimatized for at least 1 wk before being used in experiments. All animal protocols performed in this study were approved by the Harvard Medical School Standing Committee on Animal Research and were found to be in accordance with guidelines set by the US Department of Agriculture and the National Institutes of Health (Bethesda, MD).
Mouse burn-injury model
The burn-injury protocol used in these studies was approved by the Harvard Medical School Standing Committee on Animal Research and was performed as previously described (13). Briefly, mice were anesthetized via intraperitoneal injection of ketamine/xylazine 125/20 mg/kg. The dorsal fur was shaved, and the animal was placed in an insulated plastic mold to expose 25% of its total body surface area. The exposed dorsum was then immersed in 90°C water for 9 s. This approach causes a well-demarcated, full-thickness (third-degree) burn-injury. Owing to nerve cell damage, this level of injury has been defined as being painless, and analgesia was not required postinjury. Immediately after the procedure, sham and burn-injured mice were resuscitated by an i.p. injection of 1 ml pyrogen-free normal saline. This is a nonlethal injury because mortality is <5%. Groups of mice receiving A-class CpG-ODN 2336 treatment were injected with 0.2 mg/kg CpG-ODN in saline i.p. at 2 h after burn-injury. Control mice were treated with saline i.p. Mice were subjected to S. pneumoniae lung infection at 1 d after burn-injury.
S. pneumoniae lung infection model
S. pneumoniae (strain 99.55, capsular subtype 6A) organisms were maintained as frozen stocks at −80°C. Bacteria were grown for 16 h with gentle agitation in brain-heart infusion broth medium at 37°C. Bacteria were harvested at a midlog phase by centrifugation at 450 × g and washed twice by centrifugation with Dulbecco’s PBS. On the basis of absorption spectroscopy measurements, the bacteria were diluted to contain 3–5 × 107 CFU/ml in PBS. This inoculum dose was found to cause LD70 responses in normal male CD-1 mice over a 14-d period, with deaths first occurring at days 4–5 postinfection (Fig. 2A). To infect, mice were anesthetized by i.p. injection using ketamine/xylazine at 125/10 mg/kg. Mice were then held with their nares upright, and 40 μl bacteria suspension was administered intranasally (1.2–2 × 106 CFU). Bacterial CFUs were quantified by drop-plating of serial dilutions on Luria–Bertani agar plates, and colonies were counted the following day after incubating Luria–Bertani agar plates at 37°C.
TCRγδ cell depletion model
Outbred CD-1 mice received anti-mouse TCR Vgamma1.1/Cr4 and anti-mouse TCR Vgamma2 Abs at 1 mg/kg each i.p. 3 d before undergoing burn-injury to deplete TCRγδ cells. By flow cytometry, we confirmed >95% depletion of TCRγδ cells in blood samples of mice 3 d after anti-TCRγδ Abs administration (Fig. 8A).
Cells were prepared in culture medium (C5); RPMI 1640 was supplemented with 5% heat-inactivated FCS, 1 mM glutamine, 10 mM HEPES, 100× 2M nonessential amino acids, penicillin/streptomycin/fungizone, and 2.5 × 10−5 M 2-ME, all purchased from Life Technologies–Invitrogen (Grand Island, NY).
Cells were fixed and permeabilized for conventional flow cytometry using PBS with paraformaldehyde (PFA) purchased from Alfa Aesar (Ward Hill, MA) and HPLC-grade methanol purchased from Sigma-Aldrich (St Louis, MO). Fluorescent-labeled Abs specific for mouse CD3 and TCRγδ for flow cytometry stains were purchased from BioLegend (San Diego, CA). TruStain FcX Fc receptor blocking reagent from BioLegend was used in all stains to prevent nonspecific staining signals. For CyTOF mass cytometry, Abs were strategically labeled with rare-earth metal isotopes using MaxPar reagent kits from Fluidigm Sciences/DVS Sciences (Sunnyvale, CA). The CyTOF staining panels used in this study are shown in Table I.
Lung-washout cell preparations
Cells were prepared from mice at 1 d or 3 d after S. pneumoniae infection. Mice were euthanized by CO2 asphyxiation. To collect lung-washout fluid, the trachea was dissected and then cannulated with the outer sheath of a 22-gauge i.v. catheter (Exel International, Los Angeles, CA). The lung was rinsed repeatedly with a total of 3 ml ice-cold calcium/magnesium-free PBS containing 2 mM EDTA. This method allowed us to measure immune cells that specifically entered the site of infection rather than the whole organ. To prepare single-cell suspensions, cells from lung-washout fluid were harvested and washed twice in C5 medium by centrifugation at 100 × g for 10 min. RBCs were lysed using our own formulated ammonium chloride–based mouse RBC lysis buffer (JL-Buffer) that was developed for gentle lysis to preserve fragile cells, such as neutrophils. Cell preparations were washed twice by centrifugation in C5 medium and then filtered in 70-μm cell strainers.
For blood cell preparation, 0.5 ml blood was harvested by cardiac puncture in 2 mM EDTA to prevent clotting. RBCs were lysed using JL-Buffer. Blood cells were washed twice in C5 medium by centrifugation at 100 × g for 10 min and resuspended in C5 medium. For flow cytometry and CyTOF stains, cells were plated in 96-well round-bottom plates at a density of 1 × 106 cells per well. For CyTOF intracellular cytokine staining, cells were incubated for 4 h at 37°C with brefeldin A, 10 μg/ml, to allow for intracellular accumulation of cytokines.
All CyTOF staining was performed at room temperature. For all stains, rhodium viability staining reagent (Fluidigm Sciences) was added to plated cells for 10 min prior to staining. After washing by centrifugation, Fc-block reagent was added for 10 min to the cells before adding CyTOF Ab staining cocktails. Cells were stained for 30 min, then washed once in CyTOF staining buffer (cell staining buffer, calcium/magnesium-free PBS, 0.2% BSA, 0.05% sodium azide). Cells were fixed and permeabilized using Fix/Perm Buffer (PBS with 1.6% PFA and 0.3% saponin). Subsequently, intracellular Abs were added for 30 min before cells were washed and fixed with 1.6% PFA. Iridium-Intercalator solution (MaxPar Intercalator–Ir, 500 μM; Fluidigm Sciences) was added to cells in Fix/Perm Buffer according to the manufacturer’s protocol. After a final washing step, cells were reconstituted in Milli-Q–filtered distilled water at a concentration of 5 × 105 cells/ml containing EQ calibration beads according to the manufacturer’s protocol (EQ Four Element Calibration Beads; Fluidigm Sciences). The gating strategy is described in Fig. 1. Cells were analyzed on a CyTOF 2 Mass Cytometer (Fluidigm Sciences). Data analysis was conducted using Cytobank (Mountain View, CA) and the FlowJo software program (TreeStar, Ashland, OR).
Luminex multiplex cytokine detection assay
For multiplex cytokine analysis of the lung-washout fluid, samples were collected directly after the first centrifugation step and either directly stained for Luminex assay or frozen at −80°C immediately. The concentrations of IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-12p40, IL-12p70, IL-13, IL-17A, IL-33, IFN-γ, TNF-α, MCP-1, and GM-CSF in lung-washout samples at 1 d and 3 d after S. pneumoniae infection were assessed using the cytokine multiplex assay technology by Luminex. The assay was conducted using 20 μl of sample, and cytokine levels were determined by standard curve analysis. The plate was read on a Luminex 200TM instrument (Luminex, Austin, TX). Data acquisition and analysis were conducted using StarStation software v2.3 (Applied Cytometry Systems, Dinnington, U.K.).
GraphPad Prism 6.0 software (GraphPad, San Diego, CA) was used for statistical calculations. One-way ANOVA with the Tukey multiple comparison test or one-tailed Student t tests were used to analyze these data, as indicated in the figure legends. For all data, p < 0.05 was considered statistically significant with a 95% confidence interval.
CpG-ODN treatment after burn-injury improves survival of mice after a secondary S. pneumoniae lung infection
To evaluate potential beneficial effects of immune response–modifying drugs, we developed a two-hit injury/infection model by subjecting burn-injured outbred, male CD-1 mice to a secondary S. pneumoniae lung infection (Fig. 1). This model is clinically relevant because early-onset pneumonia is a major complication occurring in patients suffering from traumatic injuries (1, 2). We started by establishing a severe model of S. pneumoniae lung infection in which 70% of uninjured mice would succumb to infection within 7 d after intranasal bacterial administration (CTL-Infected group) (Fig. 1A). We then introduced our two-hit infection model subjecting mice to a third-degree burn-injury before intranasal S. pneumoniae infection at 1 d after burn-injury (Burn-Infected group). Mortality in burn-injured mice undergoing secondary lung infection significantly increased to 92%, compared with the CTL-Infected mice (p < 0.05, log-rank test).
We discovered that treating mice with CpG-ODN at 2 h after burn-injury significantly reduced mortality after a secondary S. pneumoniae infection to 70% (Burn-CpG-Infected group), to the level observed in uninjured mice (Fig. 1A). In addition, we found significantly improved lung bacterial clearance in Burn-CpG-Infected compared with Burn-Infected groups of mice (Fig. 1B). Although it is known that CpG-ODN has immune-modulatory activity as a TLR9 agonist (14), to our knowledge its beneficial activity after trauma and secondary infection has not yet been described.
Visualization of the immune response in lung-washout samples at 1 d after S. pneumoniae infection
To evaluate CpG-ODN–induced changes of immune phenotype after trauma and secondary lung infection, we used CyTOF technology as an approach for systems immunology analysis. As a first step, we characterized the immune response phenotype in the lung following S. pneumoniae infection in (noninjured) CTL-Infected mice. We used viSNE analysis (11) to interpret CyTOF staining data of immune cell subsets found in lung-washout samples of mice at 1 d postinfection (Fig. 3). To run viSNE analysis, we focused on cell subsets responsible for the regulation of the immune response, by including CD45+Gr-1lo and CD45+Gr-1− immune cell subsets, but excluded Gr-1hi granulocytes by a gating strategy displayed in Fig. 2. CD11c, F4-80, CD14, I-A/I-E (MHCII), and CD172a were markers used to identify innate cell subsets; TCRβ, TCRγδ, CD3ε, CD4, CD8α, and Foxp3 identified T cell subsets; NK1.1 and CD49b identified NK cells. The distribution of major immune cell subsets found in lung-washout samples at 1 d after S. pneumoniae infection is displayed as viSNE plots in Fig. 3.
Besides F4-80+CD11c−MHCII+CD172a+ macrophages (Mac) and F4-80−CD11c+MHCII+CD172a+ dendritic cells (DCs), we also observed a group of F4-80+CD11c+CD172a+MHCII− macrophages (CD11c+ Mac). In addition, we identified a subset of CD14+CD172a+MHCIIlo innate cells that did not express F4-80 or CD11c. The expression of NK1.1 and CD49b (not shown) allowed for the identification of NK cells. Expression patterns of CD3, CD4, CD8α, and transcription factor Foxp3 allowed for the identification of CD3+CD4+ and CD3+CD8+ T cells as well as CD3+CD4+Foxp3+ regulatory T cells (Tregs). TCRγδ was found to be expressed on two different immune cell subsets in lung-washout samples postinfection: one subset of CD3+TCRγδ+ cells was located in the vicinity of CD3+TCRβ+ cells in viSNE plots, and the other TCRγδ cell subset was CD3−TCRγδ+CD11c+.
Influence of burn-injury and CpG-ODN treatment on the immune cell phenotype after S. pneumoniae infection
As described in this article, administration of CpG-ODN 2 h after burn-injury significantly increased the survival of mice after a secondary S. pneumoniae lung infection. Therefore, we were interested in using CyTOF to screen for potential differences in immune cell subset percentages found in lung-washouts of CTL-Infected, Burn-Infected, and Burn-CpG-Infected groups of mice at 1 d postinfection. On the basis of our findings in viSNE, we designed a gating strategy displayed in Fig. 2. Although we found that burn-injury significantly increased relative expression of Gr-1+ granulocytes and innate cell subsets in Burn-Infected compared with CTL-Infected mice, no significant differences in relative immune cell types were observed in immune cell subset expression between Burn-Infected and Burn-CpG-Infected mice (Fig. 4A and 4B). In a next step, we used CyTOF for simultaneous analysis of functional marker expression on innate and adaptive immune cell subsets defined above.
To visualize differences between burn-injury alone and in combination with CpG-ODN treatment after S. pneumoniae lung infection, we displayed median functional marker expression on cell subsets in heatmap plots (Fig. 5). Fig. 5A displays median expression values of functional surface markers, transcription factors, and cytokines as observed 1 d after S. pneumoniae infection (CTL-Infected) in lung-washout samples. Fig. 5B displays median functional marker expression on cells from Burn-Infected mice at 1 d postinfection. The heatmap shows increased expression of proinflammatory functional markers CD14 on F4-80+ Macs and CD64 on CD11c+ Macs. In addition, the expression of proinflammatory cytokines, IFN-γ, and, to a minor extent, TNF-α was increased in Burn-Infected compared with CTL-Infected mice. As shown in Fig. 5C, CpG-ODN administration after burn-injury was able to partially reverse these proinflammatory changes of burn-injury after secondary S. pneumoniae infection.
Of interest, expression of the immune-inhibitory molecule PD-L1 was found to be upregulated on CD11c+ Macs and other innate immune cell subsets in Burn-Infected versus CTL-Infected mice. Burn-injury also induced ICOS expression on Tregs at 1 d after lung infection, compared with CTL-Infected mice. Furthermore, increased expression of the counterinflammatory cytokine IL-10 was specifically observed in CD14+CD172+ innate cells of Burn-Infected versus CTL-Infected mice. CpG-ODN administration after burn-injury was found to reverse many of these phenotypic changes measured at 1 d after secondary lung infection, whereas IL-17 expression in TCRγδ cells was increased, compared with that in Burn-Infected mice. In summary, we found that the pattern of functional marker expression on or in multiple immune cell subsets in Burn-CpG-Infected mice (Fig. 5C) visually resembled the expression pattern of CTL-Infected mice.
As described in this article, CpG-ODN administration 2 h after burn-injury reduced mortality after secondary S. pneumoniae infection, compared with mortality in Burn-Injured mice. Heatmap analysis of functional marker expression allowed us to identify innate immune cell subsets, TCRγδ cells, and Tregs as populations most affected by CpG-ODN treatment. Next, we were interested in studying changes in functional marker expression on lung-washout cells over time.
CpG-ODN induced changes of functional marker expression in burn-injured mice at 1 and 3 d after S. pneumoniae infection
Analysis of the immune response at day 1 postinfection identified immune cell subsets and functional markers affected by burn-injury, CpG-ODN treatment, and infection. To evaluate beneficial effects of CpG-ODN administration in our model over time, we analyzed relative changes of subset-specific markers on innate and adaptive immune cell subsets in Burn-Infected and Burn-CpG-Infected mice at 1 d and 3 d after S. pneumoniae infection (Fig. 6). To display relative changes, we defined the level of functional marker expression of Burn-Infected mice at 1 d postinfection as baseline (first column). The relative increase or decrease of expression compared with the baseline is displayed horizontally in yellow or blue, respectively. We observed that PD-L1 was highly expressed on innate cell subsets at 1 d after burn-injury and S. pneumoniae infection. The data in Fig. 6A display a CpG-ODN–induced relative decrease in PD-L1 expression level on F4-80+CD11c− macrophages (Mac) and CD11c+F4-80+ macrophages (CD11c+ Mac), as well as on CD14+CD172a+ innate cells, at day 1 postinfection. Conversely, at 3 d after burn-injury and infection, CpG-ODN treatment caused an increased expression of PD-L1 on CD11c+ Macs, CD11c+ DCs, and CD14+CD172a+ cells compared with Burn-Infected mice. Expression of MHCII on CD11c+ DCs was increased by CpG-ODN at 1 d postinfection but slightly decreased at 3 d postinfection compared with nontreated, Burn-Infected mice. In addition, expression of the chemokine receptor CCR6 was increased by CpG-ODN administration at both 1 d and 3 d postinfection on innate and adaptive immune cell subsets, compared with that in Burn-Infected mice.
Fig. 6B displays changes in functional marker expression in adaptive immune cell subsets. TCRγδ cells showed an abundance of functional marker changes induced by CpG-ODN administration at day 1 postinfection. These included increased expression of CD25, CD27, and MHCII on TCRγδ cells in CpG-ODN–treated mice at 1 d, but not 3 d, postinfection. Consistent with these observed influences of CpG-ODN on TCRγδ cell activation, we found that IL-17A was increased by CpG-ODN at 1 d but decreased at 3 d after S. pneumoniae infection. Following heatmap analysis, data were analyzed for significant differences in functional marker expression in distinct immune cell subsets. Fig. 6C displays two-dimensional plots of concatenated data to demonstrate how median PD-L1 expression was analyzed on innate immune cell subsets of Burn-Infected and Burn-CpG-Infected mice at 1 d and 3 d after S. pneumoniae infection.
The results shown in Fig. 6D demonstrate that CpG-ODN treatment significantly reduced PD-L1 expression on F4-80+ Macs, F4-80+CD11c+ Macs, and CD14+CD172a+ innate cells at 1 d postinfection. However, at 3 d postinfection, CpG-ODN–treated mice displayed significant upregulation of PD-L1 on CD11c+ Macs and CD14+CD172a+ cells. MHCII expression was significantly upregulated on F4-80+ Macs and F4-80+CD11c+ Macs in CpG-ODN–treated mice at 1 d postinfection. No significant differences were observed at 3 d postinfection between CpG-ODN–treated and nontreated groups. IL-17A expression was increased at 1 d postinfection in TCRγδ cells and CD11c+ Macs, although the increase was not statistically significant. At 3 d postinfection however, CpG-ODN–treated mice showed a significant decrease of IL-17A expression in TCRγδ cells and CD8+ T cells. Finally, levels of TNF-α were significantly reduced in CD11c+ Macs after CpG-ODN administration at 1 d postinfection.
Influence of burn-injury and CpG-ODN treatment on lung-washout cytokine levels at 1 d and 3 d after S. pneumoniae infection
We used Luminex technology to evaluate cytokine levels in lung-washout samples of CTL-Infected, Burn-Infected, and Burn-CpG-Infected mice at 1 d and 3 d after S. pneumoniae lung infection. Although CyTOF was used to determine intracellular cytokine expression of different immune cell subsets, Luminex analysis of cytokine levels in lung-washouts provided information about released cytokines during the immune response at the focus of infection. As shown in Fig. 7A, cytokine levels of IFN-γ and IL-12p70 were significantly reduced in Burn-Infected compared with CTL-Infected mice at 1 d postinfection. In addition, levels of IL-1α, IL-12p40, and TNF-α were reduced in Burn-Infected mice, although the reductions were not statistically significant. However, CpG-ODN treatment 2 h after burn-injury increased lung IL-1α, IL-12p70, IL-12p40, and IFN-γ levels at 1 d postinfection. Of note, CpG-ODN treatment after burn-injury significantly increased lung IL-6 levels at 1 d postinfection, whereas burn-injury alone did not have any significant effect on IL-6 levels. TNF-α levels, in contrast, were reduced in Burn-CpG-Infected mice at 1 d postinfection. At 3 d postinfection, CpG-ODN treatment of burn-injured mice led to a significant reduction of lung IL-6 and IFN-γ levels compared with both other groups and Burn-Injured mice, respectively (Fig. 7B).
TCRγδ cells mediate CpG-ODN–induced effects during the early immune response after burn-injury and secondary S. pneumoniae lung infection
CyTOF analysis showed increased expression of functional activation markers on TCRγδ cells. To determine whether functional TCRγδ cells were required to mediate the beneficial effects of CpG-ODN, we depleted mice of TCRγδ-expressing cells (TCRγδ−) before subjecting them to burn-injury, CpG-ODN treatment, and secondary lung infection. Fig. 8A displays representative flow cytometry plots showing >95% reduced percentages of TCRγδ cells in the blood of TCRγδ− compared with nondepleted [wild-type (WT)] CD-1 mice. We then compared cytokine levels in lung-washouts at 1 d after S. pneumoniae infection between WT Burn-Infected and WT Burn-CpG-Infected mice with those of TCRγδ− Burn-CpG-Infected mice (Fig. 8B). Clearly, TCRγδ− Burn-CpG-Infected mice displayed significantly lower levels of IL-6, IL-12p40, IL-17A, and IFN-γ compared with WT Burn-CpG-Infected mice. However, in comparing TCRγδ− Burn-CpG-Infected with WT Burn-Infected mice, only levels of IL-6 and IL-17A were found to be significantly decreased, indicating two important findings: First, TCRγδ depletion, as expected, resulted in a reduced IL-17A production, indicating a lack of functional TCRγδ cells. Second, other cytokine levels were not significantly decreased in TCRγδ− Burn-CpG-Infected compared with WT Burn-Infected mice. This finding demonstrates that TCRγδ cells are required for mediating the CpG-ODN–induced beneficial effects on cytokine production after trauma and secondary infection. Finally, improved pathogen clearance after CpG-ODN treatment was not observed in TCRγδ− mice compared with WT mice after burn-injury and infection (Fig. 8C).
Despite advances in emergency medicine and supportive care, morbidity and mortality after traumatic injury remain high. For several reasons, patients who initially survive trauma are at high risk for developing secondary pneumonia. First, invasive procedures during resuscitation can disrupt natural barriers and initiate subsequent infection with commensal bacteria. Second, massive blood loss results in a significant decrease of immune defensive proteins (15), whereas high-volume blood transfusion increases the risk of secondary infections after trauma (16, 17). In addition, factors like injury pattern, inevitable multiple surgeries, and administration of sedatives increase the risk of developing secondary infections after trauma (15). Although S. pneumoniae or H. influenzae organisms are found in healthy individuals as commensals, they can act as pathogens, causing early-onset pneumonia in severely injured patients (3, 4). In patients who died of sepsis as a complication of infection, functional and phenotypical changes consistent with suppressed immune function were found on spleen and lung immune cell subsets (18). In that report, the expression of immune-inhibitory markers, PD-L1 and PD-L2, were increased on resident lung DCs and airway epithelial cells. Furthermore, T cells in the spleens and lungs of sepsis patients showed increased expression of the PD-1 receptor (18). Other groups described expanded Treg and myeloid-derived suppressor cell populations during trauma and sepsis (19, 20). Although some studies suggested that patients who died of sepsis had marked immunosuppression, other reports described a model of protracted and unabated inflammation in patients who died of sepsis. Both models acknowledge the impairment of the adaptive immune response in the course of sepsis, but the latter model theorized that the uncontrolled innate cell-driven inflammation would lead to organ dysfunction and death (21).
In this article, we used a severe model of trauma in combination with secondary S. pneumoniae infection to help better understand the immunological influences of trauma on early-onset lung infection. We also used this model to uncover the beneficial effects of CpG-ODN treatment on infection survival response. Specifically, we wanted to determine how CpG-ODN treatment would influence cell-mediated immune reactivity at the site of infection. Accordingly, we decided to analyze immune cells entering the lung airway postinfection by studying lung-washout samples rather than lung tissue or other peripheral immune organs. Functional phenotyping of immune cell subsets revealed increased expression of proinflammatory markers, CD14 and CD64, on certain innate cell subsets in burn-injured mice at 1 d after secondary lung infection (Fig. 5). At the same time point, we showed increased expression of the immunoinhibitory marker PD-L1 on most innate cell subsets and increased expression of the activation marker ICOS on lung Tregs in Burn-Infected as compared with CTL-Infected mice. Importantly, we found that CpG-ODN administration affected the expression of these same proinflammatory and immunosuppressive functional cell markers (Fig. 5). Furthermore, we demonstrated that Burn-CpG-Infected mice showed significantly better antimicrobial immune function, as measured by increased bacterial clearance in the lungs (Fig. 2B). In combination, these data suggest that CpG-ODN treatment boosts antimicrobial immunity by altering the balance between innate inflammatory responses and immune suppression. On the basis of observations from these initial CyTOF experiments, we hypothesized that CpG-ODN treatment would lead to a more balanced immune response after secondary lung infection, preventing a protracted inflammatory phenotype and suppressed adaptive immune function.
To test this hypothesis, we used CyTOF to compare CpG-ODN–induced functional phenotype changes in burn-injured mice at 1 and 3 d after S. pneumoniae infection (Fig. 6). In addition, we used Luminex multiplex assays to evaluate cytokine levels in lung-washouts of Burn-Infected and Burn-CpG-Infected mice after lung infection (Fig. 7). We observed that mice undergoing burn-injury before secondary lung infection displayed decreased levels of multiple proinflammatory cytokines (IL-1α, IL-12p70, IFN-γ, and TNF-α) in their lung-washouts at 1 d postinfection compared with CTL-Infected mice. This injury-specific effect was reversed when CpG-ODN was administered 2 h after burn-injury. In addition, CpG-ODN treatment significantly increased IL-6 levels as compared with CTL-Infected and Burn-Infected groups of mice at 1 d postinfection. Increased levels of proinflammatory cytokines after CpG-ODN administration have been described before in different models (7, 22), supporting the idea that CpG-ODN may increase the kinetics of antimicrobial immune reactivity. In support of this idea, we found lower TNF-α levels in lung-washouts from CpG-ODN–treated burn-injured mice, which could be interpreted as CpG-ODN causing rapid lung bacterial clearance to reduce TNF-α production. Although TNF-α is one of the key cytokines responsible for mediating septic shock, clinical trials using anti–TNF-α in sepsis patients have not been successful (23), likely owing to the complexity of clinical sepsis. TNF-α was also described to induce strong Treg activation and proliferation via TNFR2 (24). In accordance with this, we found decreased expression of an activation marker, ICOS, on Tregs after CpG-ODN treatment (Fig. 6B).
Luminex cytokine analysis of lung-washout samples at 3 d after S. pneumoniae infection revealed significantly lower levels of proinflammatory IL-6 and IFN-γ in burn-injured mice after CpG-ODN administration compared with CTL-Infected and Burn-Infected mice (Fig. 7). Similarly, CyTOF analysis showed a significant decrease in proinflammatory IL-17A expression in TCRγδ and CD8+ T cells in CpG-ODN–treated, burn-injured mice. Of interest, at 3 d postinfection, we found that PD-L1 expression was increased on CD11c+ Macs and CD14+CD172a+ innate cells in burn-injured, CpG-ODN–treated mice. The role of PD-L1 and its receptor, PD-1, during disease and sepsis is controversial. As described above, activation of PD-1 by its ligands PD-L1 and PD-L2 has been shown to contribute to immunosuppression and increased mortality in sepsis patients (18, 25, 26). In contrast, beneficial effects of PD-1/PD-L interactions were demonstrated by showing that PD-L expression on APCs inhibits T cell activation and induces peripheral tolerance (27, 28). Our findings support that increased PD-L1 expression on CD11c+ Macs and CD14+172a+ cells at 1 d after burn-injury and secondary infection contributes to higher mortality from infections. However, at 3 d postinfection, we observed increased PD-L1 expression on these same cell types, which suggests PD-L1 may also have protective functions at later time points postinfection, perhaps by promoting resolution of inflammation and a healing response. Thus, CpG-ODN treatment given after traumatic injury appears to induce heightened acute antimicrobial immune function, but may also make the host more tolerant to infection by reducing immune-mediated tissue damage and systemic inflammation (29, 30). Moreover, these findings emphasize the importance of understanding the timing of functional or regulatory molecule expression to help better explain contradictory findings reported on the inflammatory and immunosuppressive phenotypes found in sepsis patients (18, 21).
Although CyTOF analysis demonstrated different effects of CpG-ODN on innate immune cell subsets, we also observed strong, CpG-ODN–dependent activation of TCRγδ cells in burn-injured mice at day 1 after S. pneumoniae infection. TCRγδ cells are known to be involved in the initial immune response to pathogens, especially at mucosal surfaces and skin (31–33). It is well established that γδ T cells are required for effective antimicrobial immune function because TCRδ−/− mice were shown to be highly susceptible to infections caused by S. pneumoniae, Klebsiella pneumoniae, Escherichia coli, Listeria monocytogenes, and Toxoplasma gondii (31, 34, 35). In a model of influenza virus infection, IL-17 production by TCRγδ cells has also been shown to promote host defense to secondary S. pneumoniae and S. aureus lung infection (36, 37). We observed that CpG-ODN induced the upregulation of MHCII (I-A/I-E), CD25, and CD27 on TCRγδ cells (Fig. 6B) in burn-injured mice at 1 d after S. pneumoniae infection, suggesting increased activation (38–40). At the same time, CyTOF analysis showed increased expression of proinflammatory IL-17 in TCRγδ cells after CpG-ODN administration. These data suggest that TCRγδ cells might mediate some of the beneficial effects of CpG-ODN in this model of trauma and infection. To evaluate the dependence of CpG-ODN–mediated effects on functional TCRγδ cells, we depleted mice of TCRγδ cells (TCRγδ−) by anti-TCRγδ Ab treatment mixture before subjecting them to burn-injury, CpG-ODN treatment, and secondary lung infection. We confirmed TCRγδ cell depletion by flow cytometry of lung-washout samples from S. pneumoniae–infected mice. Luminex analysis comparing cytokine levels in TCRγδ-depleted, Burn-CpG-Infected mice compared with WT Burn-Infected and WT Burn-CpG-Infected mice indicated that the CpG-ODN–induced increase in IL-6, IL-12p40, IL-17, and IFN-γ at 1 d after S. pneumoniae lung infection did not occur. In addition, pathogen clearance appeared to be impaired in TCRγδ− CpG-ODN–treated mice. Thus, these results provide strong evidence to support a contribution of TCRγδ T cells to the beneficial activity induced by CpG-ODN treatment.
This study has some limitations. First, we focus on the early cell-mediated immune response to trauma and lung infection with regard to CpG-ODN administration. Future studies could examine long-term effects of CpG-ODN treatment on resolution and healing using a milder S. pneumoniae infection model. Second, owing to a limitation in panel size, we did not include B cell marker Abs in our CyTOF staining panel. This decision was made because B cells are a minor cell population found in the lung airways postinfection. For the same reason, we were not able to include Abs to determine levels of type I IFN, which is known to be induced by other A-classs CpG-ODNs (7, 8). Finally, although it would be interesting to determine PD-1 expression on T cells, PD-1 was not included in our original CyTOF panel because we did not anticipate that PD-L1 would be significantly changed by CpG-ODN treatment.
An appreciation for the multitude of cell subsets, signaling pathways, and cytokines involved in the complex host response to trauma and infection has grown. As a consequence, the importance of a systems biology/immunology approach to better understand processes underlying complex physiological responses like trauma and infections was discussed by other investigators (9, 41). Systems biology relies on efficient methods for generating unbiased data that can then be used to form testable scientific questions or hypotheses. Newly established methods like RNA-Seq and CyTOF now make this type of research strategy possible for approaching scientific questions with multiple physiological inputs. In this article, we demonstrate how CyTOF helped us determine the cells and mediators involved in beneficial CpG-ODN effects on the response to trauma and secondary infection. We then used the insight gained from our data to specifically test the hypothesis that TCRγδ T cells act as mediators of the beneficial activity induced by CpG-ODN treatment.
In summary, we used a systems immunology approach to generate data to help identify cellular processes occurring in response to trauma and infection. By comparing immune responses in CpG-ODN–treated and untreated mice, we demonstrate how this strategy worked to discover beneficial changes associated with improved outcome. We demonstrate that early CpG-ODN treatment of burn-injured mice enhanced the response in the lung to bacterial challenge in a TCRγδ cell–dependent manner. The enhanced immune response resulted in improved pathogen clearance and increased the survival rate of burn-injured mice after secondary S. pneumoniae infection. Importantly, we also identified that F4-80+ and CD11c+F4-80+ macrophages, CD11c+ DCs, CD14+CD172a+ cells, and TCRγδ cells are the major cellular targets for CpG-ODN. For future experiments, similar systems biology approaches will be used by us to help gain valuable insights into complex pathophysiological processes caused by trauma and severe infections. We also will use similar strategies to test or optimize preventive treatment approaches to improve outcomes for trauma and sepsis.
We thank Nicole Paul and John Daley II at the Dana Farber Cancer Institute Flow Cytometry Facility for operating the CyTOF2 instruments and for helpful technical inputs, and we thank the Harvard Medical Area CyTOF Consortium, who made it possible to collaboratively purchase the CyTOF2 instruments for access to Harvard Medical Area CyTOF Consortium members. The Harvard Medical Area CyTOF Consortium was established with combined support from the following six institutions: Beth Israel Deaconess Medical Center, Brigham and Women’s Hospital, Dana Farber Cancer Institute, Harvard Medical School, Harvard Stem Cell Institute, and Ragon Institute of MGH, MIT and Harvard.
This work was supported by National Institutes of Health Grants AI092905-04 and AI107360-02, Deutsche Forschungsgemeinschaft Grant WA 3426/1-1, the Brigham and Women’s Hospital Biomedical Research Institute, and the Harvard Medical Area CyTOF Consortium.
Abbreviations used in this article:
cytometry by time-of-flight
regulatory T cell
visualization algorithm for stochastic neighbor embedding
The authors have no financial conflicts of interest.