Streptococcus pneumoniae is both a frequent colonizer of the human nasopharynx and a major cause of invasive diseases. Depending on preceding viral infections, the immune status of the host, and on the pneumococcal strain, an asymptomatic colonization can establish and progress to pneumonia, sepsis, or meningitis [1,2]. These infections are associated with high morbidity and mortality. It is estimated that S. pneumoniae causes over 1 million infant deaths every year worldwide and probably even more in elderly people and immunocompromised patients [3].

The more than 90 capsular serotypes of S. pneumoniae vary markedly in their ability to cause invasive infection. For example, serotypes 3, 6B, 9N and others are associated with nasopharyngeal colonization but also with infections in patients with underlying diseases, and with higher mortality when causing pneumonia. In contrast, dominating clones of serotypes 1 and 8 are frequently found in invasive diseases that, however, show lower case-fatality rates [4–7]. Among serotype 1 pneumococci, the recently emerging MLST306 clone now dominates by over 80% in many parts of the world [8–13]. While the degree of encapsulation positively associates with colonization prevalence and virulence [5], other bacterial factors that affect the interaction with the host´s innate immune system might also play a role.

In addition to the capsule, pneumolysin (PLY) is a major virulence factor of S. pneumoniae [1,14]. PLY is a member of the cholesterol-dependent cytolysins expressed by various Gram-positive bacteria. PLY of most pneumococci binds to cholesterol-containing membranes, forms pores upon oligomerization, and thereby causes cell lysis [15,16]. At sublytic concentrations, PLY has been described to activate the complement system and to stimulate cytokine production in monocytes and macrophages [1,14]. Moreover, PLY has been implicated in biofilm formation, independent of its haemolytic activity [17]. Studies in mouse models of primary pneumococcal pneumonia demonstrated that PLY-deficient strains are more rapidly cleared from the lungs and induce less inflammation compared to isogenic toxin-producing bacteria [18–20]. Interestingly, PLY exists in at least 16 different protein variants with variable haemolytic activity. For example, allele 5 PLY expressed by serotype 1 MLST306 and some serotype 8 pneumococci is non-haemolytic in contrast to the allele 1 PLY of the D39 strain [9,21,22].

Recent studies by us and others showed that the innate immune system can sense haemolytic PLY through the canonical NLRP3 inflammasome [23–26]. Canonical inflammasomes are multiprotein complexes mainly expressed in monocytes and macrophages. They are composed of a receptor molecule belonging to the NOD-like receptor family (such as NLRP3) or the PYHIN protein family (e.g. AIM2), the adapter molecule ASC and caspase-1 [27,28]. Inflammasomes post-transcriptionally regulate the production of IL-1β and IL-18 by a caspase-1-dependent processing of the cytokine pro-forms into mature cytokines, and induce an inflammatory cell death called pyroptosis. Mice lacking inflammasome components or receptors for IL-1 or IL-18 show an altered susceptibility towards S. pneumoniae [23–26,29,30], indicating that inflammasomes are critical components of the innate defence system during pneumococcal infection.

Here we demonstrate that different pneumococcal strains are differentially recognized by the innate immune system in human lung tissue. The NLRP3 inflammasome senses S. pneumoniae expressing haemolytic PLY in human lung tissue, whereas the pneumococcal serotypes 1 and 8 expressing non-haemolytic toxins are poor inflammasome activators.

S. pneumoniae serotypes with variable haemolytic activities differentially stimulate caspase-1-dependent production of IL-1β by human cells.

To confirm that production of IL-1ß by human cells is dependent on PLY and on conventional inflammasomes, we infected human PBMCs with S. pneumoniae serotype 2 expressing haemolytic allele 1 PLY (D39) or with the isogenic mutant lacking PLY (D39Δply) and tested the effect of a caspase-1 inhibitor. We found that only S. pneumoniae expressing PLY stimulated strong IL-1β production, whereas the PLY-negative mutant induced only small amounts of IL-1β (Fig 1A). In contrast, transcriptional regulation of Il1b mRNA and production of the inflammasome-independent chemokine IL-8 was not affected by the lack of PLY (Fig 1B and 1C). Moreover, IL-1β but not IL-8 secretion was strongly reduced by treatment with the caspase-1 inhibitor Z-YVAD (Fig 1D and 1E). These data are in line with previously published studies and show that pneumococcal PLY is recognized by a conventional caspase-1-dependent inflammasome to stimulate IL-1β production [23,24].

Next, we investigated the ability of different pneumococcal strains to induce haemolysis and IL-1β production. S. pneumoniae serotypes 3, 6B, and 9N induced red blood cell lysis, indicating expression of a haemolytic PLY (Fig 1F). In contrast, serotypes 1 and 8, expressing a non-haemolytic PLY, exhibited no haemolytic activities, as shown before (Fig 1F) [21]. Sequencing of the ply gene confirmed the expression of an allele 5 PLY in serotype 1 and 8 bacteria, and demonstrated that serotype 3, 6B and 9N express haemolytic allele 1 or 2 PLY (Table 1). Allele 5 PLY possesses several mutations in the protein domain D3 (Fig 1G), which is suggested to insert into the host cell membrane after conformational changes and is therefore crucial for the pore-forming ability of PLY [41]. In accordance, only serotype 6B and 9N stimulated a strong IL-1β production in human PBMCs, whereas serotypes 1 and 8 did not (Fig 1H). Unexpectedly, serotype 3 was also a rather weak inducer of IL-1β production. This suggests that other factors than PLY, such as perhaps an extensive capsule production of serotype 3 pneumococci, affect the innate sensing of the bacteria. All tested bacteria induced the transcriptional regulation of Il1b mRNA (Fig 1I) and equally stimulated production of the inflammasome-independent chemokine IL-8 (Fig 1J). Moreover, cell death induced by S. pneumoniae D39 was dependent on PLY (Fig 1K), and correlated with expression of haemolytic PLY (Fig 1L). Accordingly, purified allele 1 PLY induced a much stronger cell death as compared to allele 5 PLY in humans cells (Fig 1M). Collectively, the data confirm and extend previously published results [24], and demonstrate that pneumococcal strains expressing diverse toxins induce different amounts of IL-1β production and cell death.

The NLRP3 inflammasome is an important mediator of innate immunity during pneumococcal pneumonia. It senses pneumococcal PLY in murine macrophages and mediates production of the key inflammatory cytokines IL-1β and IL-18 [23,24]. Accordingly, mice deficient in inflammasome components or receptors for IL-1 or IL-18 show an altered susceptibility towards S. pneumoniae [23–25,29,30]. Although rodents, and in particular mice, are widely used in biomedical research, increasing evidence suggests that some mechanisms and pathways of the immune system differ between rodents and humans. It is therefore important to prove the relevance of mechanisms that have been identified in murine models in human-relevant systems. In this study, we used an ex vivo explant model allowing for the investigation of bacterial [34] and viral infections in human lung tissue [44,45]. We demonstrate that the detection of PLY by the NLRP3 inflammasome is also critically involved in the production of IL-1ß in human lung tissue. Furthermore, various pneumococcal strains expressing natural PLY variants differentially induce inflammasome-dependent IL-1β production by human cells and lung tissue.

PLY is a major virulence factor of S. pneumoniae that exhibits a wide variety of activities consistent with virulence [14–16,46,47]. It is cytotoxic to host cells, inhibits ciliary beating on the respiratory epithelium, activates the classical complement system, and stimulates inflammasomes [14]. Accordingly, bacterial mutants lacking PLY have been shown to be more rapidly cleared from the lungs compared to wild-type bacteria [18–20]. It is, however, less clear which of the different activities of PLY contribute most to pneumococcal virulence.

The MLST306 serotype 1 clone has emerged as being responsible for over 80% of serotype 1 diseases in many parts of the world, and has been associated with invasive pneumococcal disease outbreaks [9,10,12,13]. MLST306 pneumococci express allele 5 PLY which differs in a few amino acids from haemolytic allele 1 PLY and lacks any haemolytic activity but retains the ability to bind to cholesterol-containing membranes [9,21]. The expression of a non-haemolytic PLY might make serotype 1 bacteria less damaging and harmful for the host, and possibly helps to explain why serotype 1 diseases are associated with lower case fatality rates compared to other pneumococcal infections [5]. However, not all serotype 1 pneumococci carry the non-haemolytic PLY [21], and further studies are clearly required to prove or disprove this hypothesis.

A recent study compared isogenic mutants in the D39 background and showed that mutants expressing allele 5 PLY exhibited a medium-level virulence compared to bacteria expressing haemolytic PLY or lacking any toxin expression [48]. Taken this fact together with epidemiological data might indicate that cytolytic activity is not an essential characteristic of pneumococci for being virulent. Non-cytolytic properties of PLY are at least similarly important. Moreover, the different activities of PLY might be even less essential for secondary pneumococcal pneumonia following influenza virus infection [49].

We suggest that MLST306 serotype 1 might actually take advantage of expressing a non-haemolytic toxin variant. The expression of allele 5 PLY enables serotype 1 pneumococci to avoid the detection by inflammasomes and to minimize the production of key pro-inflammatory cytokines. We speculate that the evasion of inflammasome-dependent detection by serotype 1 contributes to its ability to invade sterile compartments. Of note, expression of haemolytic PLY has additionally been shown to be required for the detection of pneumococcal peptidoglycan by NOD2 and DNA by a STING dependent pathway, respectively [47,50–52]. Avoiding recognition by those pathways might thus also add to the relative particularity of serotype 1 pneumococci. Potentially in line with this assumption, serotype 1 pneumococci appear to be weak inducers of Il1b mRNA (Fig 1I). Moreover, it is likely that other PLY-independent characteristics of this strain such as e.g. capsule or cell wall composition influence the interaction with the host´s innate immune system. For example, we recently showed that serotype 1 bacteria differ to other serotypes in their binding to the C-type lectin receptor Mincle [53].

Taken together, we demonstrate that different pneumococcal serotypes differentially activate the NLRP3 inflammasome in human lung tissue, and that important serotype 1 and 8 clones evade recognition by this central innate immune sensing mechanism. It will be important to further dissect the contribution of this innate immune evasion to the pathogenic potential of these globally emerging bacterial strains.