Activation-dependent modulation of Streptococcus pneumoniae-mediated death in human lymphocytes

Abstract

Streptococcus pneumoniae (Spn) is a leading cause of community-acquired pneumonia, with infants and the elderly exhibiting significant susceptibility to the development of severe disease. A growing body of evidence supports the ability of Spn to negatively regulate the host response to infection, e.g. the capacity to induce death in numerous cell types. However, our understanding of the ability of Spn to directly impact lymphocytes remains limited. In this study, we tested the hypothesis that lymphocyte type and activation state influences the susceptibility to pneumococcus-mediated death. We show that in the resting state, CD4⁺ T cells exhibit a modestly increased susceptibility to Spn-induced death compared to CD8⁺ T cells or NK cells. In the presence of activating stimuli, the situation most reflective of what would occur in vivo during infection, all subsets demonstrated a significant increase in sensitivity to Spn-mediated death. Importantly, the activated subsets diverged dramatically in susceptibility with natural killer cells exhibiting an 8.6-fold greater sensitivity to pneumococcal components compared to the T-cell subsets. These results significantly expand our understanding of the capacity for pneumococcus to negatively regulate lymphocytes.

These studies show that the susceptibility to Streptococcus pneumoniae-mediated death in human lymphocytes increases following activation, with NK cells exhibiting the greatest sensitivity.

INTRODUCTION

The Centers for Disease Control and Prevention (CDC) reported pneumonia as the leading cause of mortality from an infectious agent in adults (Xu et al.2013). Streptococcus pneumoniae (Spn) is responsible for nearly 50% of diagnosed pneumonia cases in North America, representing a much larger proportion of cases than other associated bacteria, e.g. Haemophilus influenzae, Staphylococcus aureus, Moraxella catarrhalis and Legionella pneumophila (Brown 2012). The World Health Organization (WHO) estimates that 1.6 million people die from Spn infections annually worldwide. Of those, individuals younger than two or older than 65 years of age are at highest risk for invasive disease (Roush and Baldy 2008). In addition to potential life-threatening pneumonia, Spn infection can cause sinusitis, otitis media and invasive pneumococcal diseases such as bacteremia, sepsis and meningitis.

More than 97 unique serotypes of Spn have been characterized (Geno et al.2015). The serotypes vary greatly in pathogenicity, in part due to differences in the capsular polysaccharide, which impacts the ability to evade opsonophagocytosis and colonize the nasopharynx. Severe disease occurs when colonizing pneumococci escape the upper airway and invade deeper tissue compartments. Death associated with invasive disease is highest following infection with serotypes 3, 6A, 6B, 9N and 19F (Weinberger et al.2010). Current vaccines have significantly decreased the rates of invasive disease due to virulent serotypes (Keller, Robinson and McDaniel 2016), but rates of otitis media and population colonization have been only minimally impacted (Keller, Robinson and McDaniel 2016).

Spn has an array of virulence factors that contribute to its potential to cause disease, including the ability to induce cell death in a large number of cell types (Cockeran, Anderson and Feldman 2002; Mitchell and Dalziel 2014). Cell death is largely attributed to expression of the pore-forming cytolysin pneumolysin (PLY) (Cockeran, Anderson and Feldman 2002; Mitchell and Dalziel 2014). PLY has a number of properties that can contribute to Spn pathogenesis (Cockeran, Anderson and Feldman 2002), including hemolytic activity (Sanders et al.2008). A number of ply alleles have been identified and in vitro analyses demonstrate that the PLY proteins produced vary extensively in their hemolytic activity (Morales et al.2015). Although all cells appear susceptible to Spn-mediated killing, sensitivity varies significantly across cell types (Husmann et al.2006). This may in part be to differences in the ability of host cells to actively combat the effects of cytolysins via endocytic removal of toxin-containing areas of the membrane (Idone et al.2008) and subsequent degradation of the toxin (Corrotte et al.2012). The efficiency of toxin removal would be predicted to be an important determinant of cell-type-dependent differences in the susceptibility to the cytolysin-mediated killing. There are also some data that suggest activation state can impact the sensitivity to the lytic effects of Spn. Treatment of the human monocyte cell line U937 with exogenous interferon γ (IFNγ) increased the resistance to death by nearly 2-fold (Hirst et al.2002). However, this was not observed for an alternative monocyte line, THP-1 or for lung epithelial cell lines (Hirst et al.2002). Thus, whether and how activation alters the sensitivity of individual cell types, e.g. lymphocytes, remains unclear.

Spn-mediated killing of lymphocytes has been reported (Daigneault et al.2012). Interestingly, in this previous study both caspase-dependent and independent death were detected and the death pathway employed was controlled by the presence of monocytes in the culture (Daigneault et al.2012). In the current study, we used a panel of Spn strains with differing hemolytic potentials to evaluate the relative sensitivity to death of individual human lymphocyte subsets. We analyzed CD8⁺ T cells, CD4⁺ T cells and natural killer (NK) cells. Furthermore, we assessed the effect of activation on the sensitivity to Spn-induced death for each cell type.

MATERIALS AND METHODS

Ethics statement

All research using donated blood from healthy donors complied with federal and institutional guidelines set forth by Wake Forest University and the Human Subjects Research community. The Internal Review Board (IRB) at Wake Forest University approved all procedures and methods [IRB 00035037-SPID816].

Human PBMC

Human blood was collected from healthy subjects by a certified phlebotomist in the Clinical Research Unit (CRU) at Wake Forest Baptist Medical Center. A volume of 80–100 mL of blood was collected into 10 mL heparinized vacutainers. Mononuclear cells were isolated with Lymphoprep as per the manufacturer's instructions.

Spn lysate preparation

Spn strains used in this study are listed in Table 1. Spn were cultured in brain–heart infusion (BHI) broth (Difco, BD Diagnostics, Franklin Lakes, NJ) supplemented with 10% heat-inactivated horse serum (Life Technologies, Waltham, MA) and catalase (2500 U/mL) mL at 37°C to mid-log phase (OD₆₀₀ 0.4-0.8) and freezer stocks were prepared in 18% glycerol/mL. Aliquots were stored at –80°C. Thawed-frozen aliquots were seeded into 1 L of BHI broth supplemented with 1% choline chloride (to prevent autolysis) and grown overnight at 37°C. Cultures were centrifuged and bacterial pellets were washed three times with phosphate-buffered saline (PBS). Washed pellets were resuspended in 30–50 mL of PBS, and the bacteria were mechanically disrupted using an Emusliflex C3 (Avestin, Inc., Ottawa, ON, Canada). Lysed bacteria were centrifuged at 12 000 x g for 20 min at 22°C to pellet any remaining intact bacteria and insoluble components. The protein concentration of the supernatant was determined using a bicinchoninic acid protein assay kit (Thermo Scientific, Waltham, MA). Lysates were aliquoted in 200 μL volumes and stored at –80°C.

Table 1.

Strains used in this study.

Strain	Serotype	Disease	Reference
D39	2	Pneumonia	Gingles et al. (2001)
EF6796	6A	Pneumonia, bacteremia	Briles et al. (1992)
EF3030	19F	Colonizes nasopharynx, otitis media	Briles et al. (1992)
16654	23F	Otitis media	D. Briles Collection
MNZ1113	Null	Colonizes nasopharynx, otitis media	Hiller et al. (2010)
PLNA	2		Berry et al. (1989)

Ex vivo stimulation and lysate treatment

PBMC were thawed and rested overnight. A total of 3×10⁵ cells were added per well of a 96-well plate. For analysis of activated cells, PBMC were cultured in the presence of 100 ng/mL phorbol 12-myristate 13-acetate (PMA) and 1 μg/mL ionomycin. Lysate-mediated killing was assessed by addition of the indicated amounts followed by culture at 37°C for 5 h.

Flow cytometry

Cell viability was determined by Zombie Violet staining (BioLegend). The following antibodies were used: APC anti-human CD56, APC-eFluor780 anti-human CD3, FITC anti-human CD4, PE anti-human CD4, PerCP-Cy5.5 anti-human CD4, PE-Cy7 anti-human CD8α and PerCP-Cy5.5 anti-human CD8α. APC-eFluor780 anti-human CD3 was purchased from eBioscience, PE anti-human CD4 was purchased from BD Biosciences, San Jose, CA and all other antibodies were purchased from BioLegend, San Diego, CA. Samples were acquired on a BD Biosciences Canto II instrument. Data were analyzed using BD FACSDIVA (BD Biosciences, San Jose, CA) and FlowJo (TreeStar, Ashland, OR) software.

Hemolysin assay

This protocol was adapted from previously published methods (Baba et al.2001; Sanders et al.2008). An aliquot of Spn lysate was thawed, warmed to 37°C and serially diluted into 100 μL of washed sheep red blood cells (RBC) (Rockland, Limerick, PA). The mixtures were incubated in a 37°C water bath for 30 min. Lysis was measured by measuring the OD of the transferred supernatant. Triton X-100-lysed samples served as a positive control.

Statistical analyses

Lymphocyte survival and hemolysin activity were analyzed by non-linear regression using the following four-parameter logistic equation,

where upr and lwr represent the upper and lower asymptote, x is the log of the amount of lysate, ED₅₀ is the log of the midpoint or 50% effective dose and scale represents the slope of the curve at the midpoint. For lymphocyte survival, y represents the fraction of viable cells, which was scaled to a maximum of 100; the upper and lower asymptotes were constrained to 100 and 0, respectively, and the ED₅₀ and scale parameters were obtained by non-linear least squares regression. For the analysis of hemolysin activity, y represents the increased absorbance at 415 nm. Here, the lower asymptote was constrained to 0 and non-linear least squares regression was used to determine the upper asymptote, ED₅₀ and scale parameters.

In order to compare non-linear regression parameters within an experiment, the most parsimonious logistic regression model was identified by comparing full and reduced models. In each case, the full model allowed the ED₅₀ and scale parameter to vary with each curve. Reduced models held either or both the ED₅₀ and scale parameter fixed for each treatment. The likelihood ratio (F) test was used determine if the data could be described by the reduced model and the full model where a p-value > 0.05 indicated no difference between full and reduced models. ED₅₀ values obtained from reduced models were analyzed further as required. The Shapiro-Wilk test was used to determine if the distribution of log-transformed ED₅₀ values could be considered normal. The Fligner-Killeen test was used to evaluate the homogeneity of variance between the groups. Data that satisfied both conditions were evaluated by ANOVA followed by the Tukey HSD post hoc test to identify differences between groups. Data that were not normally distributed were evaluated by the Kruskal-Wallis rank test for the possibility of significant differences between two or more groups. In this case, p-values obtained for each pair of values with the Mann-Whitney test were corrected for multiple comparisons by conservative Bonferroni method. Non-linear regression and statistical analysis were performed with the suite of open source tools available in R (Team RC 2015).

RESULTS

Strain-dependent differences in Spn lysate induced death of PBMC

In our studies, we performed an analysis of the susceptibility of human lymphocytes to the death induced by a panel of Spn strains (Table 1). These strains differ with regard to serotype as well as site and disease associated with their original isolation. D39 is a well-studied strain that causes invasive disease in animal models (Gingles et al.2001). EF3030 was originally isolated from the nasopharynx of an adult with otitis media (Andersson et al.1981; Briles et al.1992). In animal models, it predominantly colonizes the nasopharynx and on its own causes does not cause fatal invasive disease (Briles et al.2003). EF6796 (Briles et al.1992) and 16654 (R. Dagan collection) were isolated from the blood of a patient with pneumonia and the ear effusion of a patient with otitis media, respectively. MNZ113 is a non-encapsulated strain originally isolated from the nasopharynx of a patient with otitis media (Hiller et al.2010).

To test the potential of Spn components to induce death, we prepared bacterial lysates from overnight cultures via mechanical disruption followed by centrifugation to remove non-soluble factors. This approach allowed PBMC exposure to a large variety of pneumococcal components. PBMC purified from healthy adult human donors were cultured for 5 h in the presence of titrated amounts of lysate. Survival in the absence of lysate was set at 100% for each donor. Under this condition, survival across donors was 88±4.1%. The fraction of surviving PBMC following treatment with Spn was fit to a four-parameter logistic function with minimum and maximum values constrained to 0 and 100. Figure 1 shows the measured values and best-fit regression for each of the four donors tested on the panel of strains. Overall, donors exhibited similar death as a result of exposure to the lysates, with some variability noted, e.g. donor 2 treated with D39 lysate (Fig. 1E).

Figure 1.

Strain-dependent differences in susceptibility to Spn-mediated death in human PBMC. Lysates from D39, EF6796, MNZ1113, 16654 and EF3030 were produced by growing bacteria overnight in the presence of choline. Mechanically disrupted bacterial lysates were added at the indicated amounts. The survival of PBMC was quantified based on Zombie Violet intensity with survival in the untreated sample set at 100%. Data from individual donors (n = 4 except for MNZ1113 where n = 3) and best-fit regression for each combination of donor and bacterial strain are shown in panels A–E. Because the survival curve of MNZ1113-treated cells could be described by a constant as well as by logistic regression, MNZ1113-treated cells were excluded from the analysis in panel F showing the mean±SEM of the ED₅₀. ANOVA identified a statistically significant difference for the log of ED₅₀ values between groups [F(4,12) = 82.1, P < 10⁻⁶]. Specific differences between groups were identified by the Tukey HSD post hoc test and are indicated.

The ED₅₀ was determined by non-linear least squares regression, with the exception of MNZ1113 for which a meaningful ED₅₀ value could not be calculated given the minimal death observed. Thus, MNZ1113-challenged cells were excluded from the following analysis. Comparison of ED₅₀ values from the remaining strains (Fig. 1F) revealed statistically significant differences among groups (p < 10⁻⁶). Specifically, the ED₅₀ for PBMC exposed to 16654 lysate was significantly higher than that for EF6796 or D39 challenge (>2-fold, P < 0.03). No other pairwise comparisons (range of 0.9 to 2.1-fold) were statistically significant. Death following exposure D39 or EF6796 lysate was robust with 70%–75% death observed at the 6 μg dose on average.

The ability of Spn lysate preparations to induce PBMC death correlates with hemolytic activity

We predicted that the differences observed in PBMC death following treatment with the panel of lysates would correlate with differences in hemolytic activity. Sheep RBC were exposed to bacterial lysates, and hemolysis was measured as the change in absorbance at 415 nm. Data were fit to a four-parameter logistic function with the minimum value constrained to zero. The maximum value, ED₅₀ and scale parameter were determined by non-linear least squares regression for data from two (MNZ1113) or four (all others) independent replicates. The pneumolysin-deficient strain of D39, PLNA (Berry et al.1989), served as a negative control for these analyses.

D39 and EF6796 showed the greatest lytic activity, followed by EF3030, 16654 and finally MNZ1113 (Fig. 2A). As predicted, treatment with lysate derived from the PLY-deficient strain of D39 did not result in appreciable RBC lysis (data not shown). These data revealed a correlation between lysate-mediated hemolysis and lymphocyte death, e.g. MNZ1113 exhibited little cell death or hemolysis, 16654 induced intermediate death and hemolysis and EF6796 and D39 resulted in the greatest lymphocyte death and hemolysis. These data support a model wherein PLY contributes to different capacities to induce lymphocyte death observed across the strains.

Figure 2.

The differential death induced by Spn strains correlates with hemolytic activity. (A) Sheep RBC were treated for 30 min at 37°C with the indicated amounts of Spn lysate. The maximum value, ED₅₀ and scale parameter were determined by non-linear least squares regression for data from two (MNZ1113) or four (all others) independent replicates. The maximum value used for MNZ1113-treated samples was set to the average absorbance after Triton X-100 lysis. ED₅₀ values are represented by the box and whiskers plot. The Kruskal-Wallis test identified a statistically significant difference in ED₅₀ elicited by the different bacterial lysates, Χ²(4) = 16.0, P = 0.003. A pairwise t-test with Bonferroni post hoc correction for multiple comparisons showed each lysate to be statistically distinct (P < 0.025) with the exception of D39 and EF6796 (P = 0.22). (B) PBMC were treated for 5 h with lysate from D39 or its PLY-deficient counterpart (PLNA). PLY is required for PBMC death because the slope of the PLNA line in Fig. 2B was not statistically different from zero (P = 0.34).

To directly test this possibility, PBMC were treated with lysate derived from PLNA or D39. While PBMC exposed to D39 lysate underwent a high amount of death, they were minimally affected by treatment with the PLNA lysate (Fig. 2B). Thus, for D39, PBMC death was dependent on PLY activity.

CD4⁺ T cells exhibit a modest increase in sensitivity to death induced by Spn lysates

PBMC are a heterogeneous mixture of cell types including T cells, B cells, monocytes and others. Given a report that the efficiency of death induced by Spn can vary with cell type (Husmann et al.2006), we asked whether CD4⁺ T cells, CD8⁺ T cells and NK cells differed in their sensitivity to lysate treatment. These subsets were chosen based on their reported role as effectors in Spn infection (Kawakami et al.2003; Malley et al.2005; Zhang et al.2007; Elhaik-Goldman et al.2011; Weber, Tian and Pirofski 2011). PBMC were incubated with titrated amounts of each Spn lysate. Following the 5-h stimulation period, CD8⁺ T-cell (CD3⁺CD8⁺CD4⁻), CD4⁺ T-cell (CD3⁺CD4⁺CD8⁻) and NK-cell (CD56⁺CD3⁻) populations were analyzed (Fig. 3). The fraction of lymphocytes surviving exposure to bacterial lysate was fit to a four-parameter logistic function with minimum and maximum values constrained to 0 and 100 to determine the amount of lysate required for 50% death. Because MNZ1113-treated cells were unaffected over the range of lysate used, they were excluded from this analysis. This calculation showed that lysate derived from strains 16654, EF3030 and EF6796 induced different patterns of death among CD4⁺ T cells, CD8⁺ T cells and NK cells. CD8⁺ T cells and NK cells required approximately 1.5-fold more lysate for 50% cell death compared to CD4⁺ T cells. Similarly, the change in cell death per μg of lysate near the inflection point (scale parameter) was greatest for CD4⁺ T cells. Nonetheless, differences between cell types were not as substantial as the differences between lysates. Curiously, although more lysate was required to kill 50% of the NK cells, a greater fraction (2.5% to 9.6%) of NK cells were killed at the lower amount of lysate tested (1.5 μg) compared to other cell types (Fig. 3). This difference was significant compared to CD4⁺ and CD8⁺ T cells (P < 0.003, Tukey HSD test). Although the basis for this effect is unknown, it could reflect a mixed population of NK cells that differ in susceptibility to lysate-induced death.

Figure 3.

CD4⁺ T cells are modestly more sensitive to Spn-mediated death compared to CD8⁺ T cells or NK cells. Cells were cultured for 5 h in the presence or absence of titrated amounts of the indicated lysate. Cells were stained with antibodies to CD4, CD8, CD3 and CD56 to identify CD8⁺ T-cell, CD4⁺ T-cell and NK-cell subsets. Death was assessed by staining with Zombie Violet. The mean ± SEM for the survival of each subset is shown along with the best-fit two-parameter logistic regression.

Activation results in a robust increase in sensitivity to lysate-mediated killing

Lymphocytes recruited to the site of infection will be activated in response to antigen presented on surrounding cells or by cytokine signals present in the environment (for review, see Cox, Kahan and Zajac 2013). Thus, we explored the hypothesis that activation would alter the sensitivity of the cells to Spn-induced death. PBMC were activated by treatment with PMA and ionomycin at the time of addition of the lysate. The fraction of stimulated and non-stimulated cells surviving exposure to varying amounts of lysate was fit to a four-parameter logistic function with minimum and maximum values constrained to 0 and 100 and the scale parameter fixed for each lysate. As for the analysis of resting PBMC, a meaningful ED₅₀ value could not be determined for the MNZ1113-treated sample (Fig. 4A). For all other lysates, activation resulted in a significant shift in the sensitivity to death [F(20,22), P < 0.018]. The increased sensitivity for each treatment following stimulation with PMA and ionomycin is evident in Fig. 4F, where the 95% confidence intervals for the ED₅₀ within each treatment are non-overlapping. On average, activation decreased the amount of lysate required to induce death by approximately 2.3-fold (range 1.6- to 3.8-fold). Thus, changes associated with the activation of lymphocytes increased their susceptibility to the killing effects of Spn.

Figure 4.

Activation increases the sensitivity of PBMC to Spn-mediated death. PBMC were cultured for 5 h in the presence or absence of PMA and ionomycin together with titrated amounts of Spn lysate. Death was assessed by staining with Zombie Violet for three (A) or four (B-E) donors. A reduced regression model in which the scale parameter was held constant for each lysate, but the ED₅₀ value was allowed to vary described the data as well as the full model [F(20,21), P > 0.11]. Therefore, results shown in panels A–E represent the mean ± SEM with the best-fit regression line from this reduced model. Because survival of non-stimulated MNZ1113-treated cells could be described by a constant value as well as by logistic regression, MNZ1113 was excluded the analysis shown in panel F which shows the estimated ED₅₀ value and 95% confidence interval obtained by the profile likelihood method.

Activated NK cells exhibit the greatest sensitivity to Spn-mediated death

Given the activation induced increase in susceptibility to death observed in total PBMC, it was critical to determine whether this shift in sensitivity occurred similarly among the lymphocyte subsets. The results are best described by a full regression model in which the ED₅₀ and scale were allowed to vary with cell type within each treatment. However, because most of the deviation from a reduced model was limited to cells treated with EF3030 and EF6796 lysate and to facilitate comparison with the results obtained using whole PBMC (Fig. 4), we chose to analyze results from the reduced model holding the scale parameter constant for each lysate. Figure 5(A–E) shows the mean and standard error with the best-fit regression using this reduced model.

Figure 5.

Activated NK cells exhibit a highly increased sensitivity to Spn-mediated death compared to activated T cells. PBMC were cultured for 5 h in the presence of PMA and ionomycin together with titrated amounts of Spn lysate. Survival of CD4⁺ T cells, CD8⁺ T cells and NK cells at each dose of lysate was assessed by staining with Zombie Violet and fit to a two-parameter logistic regression model where the scale parameter was held constant for each lysate and the ED₅₀ value was allowed to vary with cell subset. Results are shown as the mean ± SEM for three (A) or four (B–E) donors with best-fit regression line. Panel F shows the estimated ED₅₀ value and 95% confidence interval from the profile likelihood method for each combination of Spn lysate and lymphocyte subset.

The ED₅₀ value and 95% confidence intervals for each combination of bacterial treatment and lymphocyte subset (Fig. 5F) show that comparable amounts of bacterial lysate were required to kill 50% of the stimulated CD4⁺ and CD8⁺ T cells. Unexpectedly, activated NK cells were significantly more sensitive than activated CD4⁺ or CD8⁺ T cells, on average exhibiting an 8.6-fold (range 5.2- to 12.8-fold) decrease in the amount of lysate necessary to induce death in 50% of cells. These data show that in stark contrast to what was observed for resting cells, activated NK cells exhibit a profoundly increased sensitivity to death compared to their activated counterpart.

DISCUSSION

The ability to cause death in a broad range of cell types is a well-documented property of Spn (for review, see Mitchell and Dalziel 2014). Susceptible cell types include a number of immune cells, i.e. dendritic cells (Littmann et al.2009), B lymphocytes (Daigneault et al.2012) and T lymphocytes (Daigneault et al.2012). Given the evidence for the immunoregulatory effects of Spn, we set out to perform an in depth analysis of Spn-mediated death of human T-cell and NK subsets as these cells are important contributors to the clearance of pneumococcus (Kawakami et al.2003; Malley et al.2005; Zhang et al.2007; Elhaik-Goldman et al.2011). NK cells are recruited to the lungs within 6 h of infection with Spn (Kawakami et al.2003) and serve a critical role in early clearance, primarily through the production of IFN γ (Elhaik-Goldman et al.2011). In addition, an effective adaptive immune response requires generation of protective CD4⁺ T effectors (Malley et al.2005; Zhang et al.2007). Specifically, Th17 cells are reported to be critical contributors of Spn clearance (Zhang et al.2007). Finally, in the event of coinfection with a viral pathogen, NK cells as well as anti-viral CD4⁺ and CD8⁺ T cells would be called into the lung to combat infection.

To broaden our study, we used a panel of Spn strains representing a number of serotypes and derived from individuals with differences in disease states (Andersson et al.1981; Briles et al.1992, 2003; Gingles et al.2001; Hiller et al.2010). We found that, as expected, strains varied in their ability to induce death in PBMC. Interestingly, there is some correlation with the disease state and hemolytic potential of the strains assessed in our study, although admittedly whether differences in cell killing potential contributes to this requires further study. In general, individual donors exhibited a similar hierarchy with regard to the relative sensitivity to distinct strains. Evaluation of CD4⁺ T-cell, CD8⁺ T-cell and NK-cell populations within these donors revealed relatively comparable susceptibility to death induced by Spn, although CD4⁺ T cells were found to have a modest increase. However, when death was assessed in the presence of an activation signal, a previously unknown modulation was observed. For all subsets, exposure to pneumococcal components at the time of activation resulted in a significant increase in sensitivity to the death. Surprisingly, the shift in the dose required differed among the subsets, with NK cells displaying a much greater sensitivity to Spn-induced death compared to T cells. Our results with a PLY-deficient D39 provide evidence that death is dependent on pneumolysin. While not directly tested for the other strains, based on previous data in the literature demonstrating the ability of pneumolysin to kill cells (Cockeran, Anderson and Feldman 2002; Mitchell and Dalziel 2014) and our finding with the pneumolysin deficient strain of D39, we would speculate that the differences in PBMC death observed following treatment with each of the lysates are impacted by PLY quantity and/or quality. Alternatively, it is possible that another factor contributes to the death mediated by PLY and that differential expression or activity of this factor is responsible for the differences among the strains.

Our finding that activated lymphocytes increased their sensitivity to death induced by Spn components is intriguing and has important implications for immune cells that enter the infected tissue where high levels of Spn components could impede pathogen clearance through the death of immune effectors. Given the proposed role of PLY, it is tempting to speculate that a difference in the interaction of this molecule with cells as a result of activation regulates the change in sensitivity. The cytolytic activity of PLY is dependent on binding to cholesterol (Tweten 2005) with oligomerization of up to 50 monomers resulting in pores that can reach 350Å in diameter (Mitchell and Dalziel 2014). Studies in human corneal epithelial cells have shown that PLY is associated with lipid rafts, which are membrane structures that are highly enriched in cholesterol (Taylor et al.2013). Lipid rafts are known to play a critical role in the activation of lymphocytes and disruption of these structures strongly inhibits this process (Jury, Flores-Borja and Kabouridis 2007). Furthermore, lymphocyte activation results in significant changes in the distribution of cholesterol, with the vast majority of the predominantly intracellularly distributed raft component ganglioside GM1 redistributing to the cell membrane of naive T cells within 4 h (Tuosto et al.2001). The known change in cholesterol distribution that occurs as a result of activation is an attractive possibility to explain the increased sensitivity of lymphocytes to Spn-mediated death following stimulation.

Another potential regulator of sensitivity is the ability of a cell to remove toxin-containing areas of membrane through endocytosis (Idone et al.2008; Corrotte et al.2012). Previous studies showed that membrane repair following exposure to the bacterial toxin SLO occurred through a novel pathway that resulted in the targeting of damaged membrane to lysosomes, where they are degraded. Interestingly, activated lymphocytes have been reported to decrease membrane turnover (Lewis and Pegrum 1977), a change which would be predicted to increase their sensitivity to death. Additional studies are required to understand the control of the susceptibility to Spn-mediated death in the distinct immune populations and how it is regulated with activation.

In spite of an available vaccine, Spn-mediated disease remains a serious public health concern, in part due to the emergence of previously rare serotypes as a result of vaccine-mediated pressures. This, together with the increasing antibiotic resistance present in circulating strains, makes a greater understanding of Spn-mediated modulation of the host response a pressing concern. Our data show that activation of lymphocytes greatly increased their susceptibility to death mediated by Spn. Among these activated populations, NK cells are by far the most sensitive. These results lead to a model whereby activation of cells that enter the tissue to combat Spn infection makes them especially sensitive to death induced by Spn. Given the important role for NK cells in the early control of infection and T cells in the late control, this would serve as a potent immune evasion mechanism allowing for progression of infection.

FUNDING

LKB was supported in part by National Institutes of Health [T32 AI007401].

Conflict of interest. None declared.