Structural insights into loss of function of a pore forming toxin and its role in pneumococcal adaptation to an intracellular lifestyle

The opportunistic pathogen Streptococcus pneumoniae has dual lifestyles: one of an asymptomatic colonizer in the human nasopharynx and the other of a deadly pathogen invading sterile host compartments. The latter triggers an overwhelming inflammatory response, partly driven via pore forming activity of the cholesterol dependent cytolysin (CDC), pneumolysin. Although pneumolysin-induced inflammation drives person-to-person transmission from nasopharynx, the primary reservoir for pneumococcus, it also contributes to high mortality rates, creating a bottleneck that hampers widespread bacterial dissemination, thus acting as a double-edged sword. Serotype 1 ST306, a widespread pneumococcal clone, harbours a non-hemolytic variant of pneumolysin (Ply-NH). Performing crystal structure analysis of Ply-NH, we identified Y150H and T172I as key substitutions responsible for loss of its pore forming activity. We uncovered a novel inter-molecular cation-π interaction, governing formation of the transmembrane β-hairpins (TMH) in the pore state of Ply, which can be extended to other CDCs. H150 in Ply-NH disrupts this interaction, while I172 provides structural rigidity to domain-3, through hydrophobic interactions, inhibiting TMH formation. Loss of pore forming activity enabled improved cellular invasion and autophagy evasion, promoting an atypical intracellular lifestyle for pneumococcus, a finding that was corroborated in in vivo infection models. Attenuation of inflammatory responses and tissue damage promoted tolerance of Ply-NH-expressing pneumococcus in the lower respiratory tract. Adoption of this altered lifestyle may be necessary for ST306 due to its limited nasopharyngeal carriage, with Ply-NH, aided partly by loss of its pore forming ability, facilitating a benign association of SPN in an alternative, intracellular host niche.

1. The authors provide a crystal structure of the pore-forming toxin, pneumolysin, from strain ST306, a strain Streptococcus pneumoniae that produces a non-hemolytic (NH) variant of pneumolysin. They go on to present data to suggest that this NH variant leads to increased phagocytosis of S. pneumoniae and facilitates intracellular persistence. In vivo, the NH variant is more rapidly cleared but also invokes a reduced inflammatory response. The study presents the new crystal structure for this non-hemolytic variant of pneumolysin, which is of interest to the field. Where this study falls short is the inconsistencies with the in vitro and in vivo data. Many of the observations seen are largely what is observed with pneumolysin knockouts or toxoid variants, and thus do not add significantly to the field. New data shows that the NH version can persist in a low dose model, the mechanism for which is not known, which would be the major shortfall of the study, when other data is limited as far as comparisons to toxoid or clean deletion strains. The major mechanistic advance would be to understand why this pneumolysin expressing versions persists better than the other variants.

Response:
We thank the reviewer for positive comments and constructive criticism of our work. We are appreciative of this process which has helped to improve the manuscript. In the revised manuscript we have attempted to address the concerns raised by the reviewer and have added new data with ply knockout strains and the W433F toxoid variant. These new data demonstrate that possession of Ply-NH is not comparable to either loss of Ply (see comparisons with the knockout strain) or presence of a non-hemolytic Ply (see comparisons with W433F strain), in terms of infection outcomes.
It has been demonstrated previously (Reference # 7) that, at low concentration, Ply interacts with MRC-1 on dendritic cells (DCs) to impede secretion of pro-inflammatory cytokines. We now show that DCs secrete higher amounts of pro-inflammatory cytokine TNF-α when infected with a Ply deletion mutant compared to an isogenic Ply-NH-bearing SPN strain ( Fig S12A). This finding offers some insight into why D39:Ply-NH derives an advantage relative to D39ply in vivo, as it would be subjected to reduced downstream inflammatory effects, such as macrophage killing.
SPN strains harbouring different Ply variants are differentially susceptible to intracellular killing mechanisms upon invasion of lung epithelial cells. A Ply W433F -bearing strain associated more readily with autophagy markers than did D39:Ply-NH , and this finding may explain the relative in vivo advantage of D39:Ply-NH vs D39:Ply W433F .

Reviewer # 2
1. The article submitted by Badgudgar et al represent a highly ambitious study on the impact of non-hemolytic pneumolysin on serotype 1 disease. The authors go on to crystalize the non-hemolytic version of pneumolysin found in serotype 1 ST306, determine that it binds to the cell, forms oligomeric rings but does not insert into the pore due to a mutation in the area that forms transmembrane b-hairpins (Y150H and T172I, that this mutation (modestly) prolongs the ability of pneumococci to survive within cells, and despite being attenuated overall for disease pneumococci that express a non-hemolytic toxin are able to persist within what appears to be an intracellular state. The conclusions reached by the authors are that this represents an evolutionary move by the bacterium to co-exist with the host as otherwise it causes death which is not optimal for the bacterium. The authors are to be commended in what is truly a herculean effort in regards to the structural aspects of the work. Unfortunately the rigor drops off as they move towards the evidence that suggests this intracellular state is vital for the disease process and that this is a major aspect in the lifestyle of serotype 1, moreover that the ability to persist in vivo is driving type 1 disease.

Response:
We thank the reviewer for appreciating our efforts and have performed additional experiments, both in vitro and in vivo, to strengthen our claims. We believe with these additional experiments can address the concerns raised by the reviewer. Comparison of interaction of Ply-NH harbouring SPN with THP1 macrophages versus A549 alveolar epithelial cells indicate that though the invasion is higher in the macrophages (presumably due to phagocytosis), the survival and recycling efficiencies are significantly less compared to A549, particularly at later time points ( Fig S13). Thus, a niche inside non-professional phagocytes, such as alveolar epithelial cells, would ensure the improved survival of SPN and additionally enable its egression out of the cells for the purpose of transmission.
2. One other major concern is that the authors repeatedly simplify the complexity that is pneumococcal disease. A much more nuanced statement of how pneumococcal disease arise, the role of colonization and inflammation, and disease is required. 3. Overall the data that suggests intracellular persistence of type 1 ST306 is vital for prolonging the bacterium state and this confers an advantage to the bacterium is insufficient. I say this not to detract from what in many ways is an excellent work effort, but to make sure that the conclusion reached by the authors are well supported and therefore find the same support by future readers. As is, I remain skeptical and I would dismiss the papers conclusions as being unsubstantiated.

Response:
We thank the reviewer for appreciating our efforts and trust that the revised version provides sufficient evidence to convince that the intracellular residence of serotype 1, aided by Ply-NH, plays a significant role in its persistence. Specifically, we have added the following data: 1. New acute infection model data to demonstrate that infection progresses differently in D39:Ply-NH infected mice than in those infected with wither D39ply or D39:Ply W433F , suggesting that harbouring allele 5 pneumolysin is not functionally equivalent to either loss of pneumolysin or possession of a non-hemolytic toxoid variant (Fig 6A, B).
Reviewer # 3 1. The current study by Badgujar et. al concerns the pore forming toxin pneumolysin (Ply) and the structural difference between traditional Spn strain and serotype 1 isolates that are non-hemolytic. This is a resubmission of the original submission. The first part of the study is based on the solved crystal structure of the NH Ply by the authors, and provides interesting contrasts with the hemolytic-Ply variants. One of the distinguishing features that makes Ply-NH unique is its ability to bind and oligomerize on cell membranes but not to form pores. Next, the authors test the differences between the two Ply in different cell culture models and establish that the loss of pore formation leads to increased internalization and persistence of Spn in host cells. Lastly, the authors test the difference in ply in a murine model, and suggest that defective NH ply strains are less virulent and persist longer in the host.

Response:
We thank the reviewer for positive comments regarding our work and for providing a summary of our findings.

Part II -Major Issues
Reviewer # 1 1. Phagocytosis was performed with R6-Ply-NH and -H. How does this compare with a Ply null or toxoid derivative? The authors also mentioned differences were not due to differential pneumolysin production. Isogenically derived strains of S. pneumoniae do exist with altered Ply production, if this were to be tested.

Response:
As suggested by the reviewer, we have performed phagocytosis assay with the Ply null mutant. We observed that at extended time points post infection, both ply mutant and Ply-NH bearing strains, showed improved persistence compared to SPN strain bearing Ply-H (Fig S9F).
We have shown by western blotting and ELISA that isogenic SPN strains harbouring various forms of Ply express Ply in comparable amounts (Fig S6A-B). As our study focusses on how structurally different form of Ply influence interactions with the host, we believe a comparison of different levels of Ply production would be beyond the scope of this manuscript.
2. In vivo the authors show there is improved survival of mice with the NH variant, lack of Ply is known to influence virulence. The capacity to persist was seen when bacterial counts were determined at later times ( Figure 6B)-no statistics is shown in this, is NH vs H statistically different? Is this result any different to a Ply null strain? These mice no doubt have a reduced host response.

Response:
As only 1 mouse infected with Ply-H bearing SPN strain survived till later time points, we are not able do a statistical comparison with Ply-H. Instead, we have now added data for both D39ply and D39:Ply W433F and show statistical comparisons vs D39:Ply-NH (Fig 6B). In this model, Ply-H infected animals succumb early, likely due to overwhelming inflammation. However, animals infected with Ply-NH, ply or Ply W433F -expressing strains survive longer, with only 0-20% of mice developing invasive disease. At one-week post infection, surviving mice infected with D39:Ply-NH had significantly higher lung CFU than those infected with either D39ply or D39:Ply W433F . We believe that the explanation for this finding is as follows: 1. D39ply cannot interact with MRC-1 and so stimulates cytokine secretion from DCs, whilst D39:Ply-NH retains residual MRC-1 binding activity (Fig S12A). Ply's interaction with MRC-1 have been shown to impede fusion with lysosomes and thwart pneumococcal killing (Reference # 7). 2. D39:Ply W433F is less able to avoid autophagy pathways in airway epithelial cells than D39:Ply-NH (Fig S9).
3. Figure 6E would appear to be an important piece of data that delineates Ply-NH vs the toxoid mutant and a Ply null strain. The question is why does this behave differently?

Response:
We hope our answer above goes some way to addressing this question. We believe that Ply-NH has an autophagy-evasion advantage relative to D39:Ply W433F , enabling persistence intracellularly, and an immune response dampening advantage relative to D39ply, by virtue of interactions with MRC-1.

Reviewer # 2
1. Introduction, The authors paint a black and white picture of pneumococcal disease the bacterium is a killer. This is a simplification that is not true, the vast majority of individuals who are colonized are asymptomatic. Severe disease is in large part due to the state of the host, which is why Spn is generally considered to be an opportunistic infection. Most serotypes of Spn, the exception being type 1, have very low attack rates and do not with any frequently cause disease. Thus despite producing a hemolytic version of pneumolysin the some versions of this bacterium have adapted for long-term asymptomatic colonization. The idea that the bacteria has an attenuated toxin to reduce disease severity and prolong duration is therefore perhaps one that is restricted to serotype 1 versus being generally applicable to all of Spn.

Response:
We apologise for this unintended oversimplification. We had altered the emphasis within the introduction, making changes at lines 71-73 and adding new text (lines 92-97) that highlights the key role played by hemolytic Ply in ensuring onward transmission. This concept is revisited in the discussion (Page no. 24, lines 493-496) and we have added text here to indicate that we are referring specifically to serotype 1, with regards to the advantages of Ply-NH.
2. Lines 80-87, Pneumolysin triggers lung epithelial cell death via necroptosis, this has been demonstrated to cause the release IL-1α from necrotic cells which is inflammatory. This should be added to the text.

Response:
As suggested by the reviewer, Ply mediated necroptosis of lung epithelial cells and subsequent release of IL-1α has now been described in the text (Page no. 5, lines 86-88).
3. I found the structural data to be strong and have no concerns. The authors should be commended on a very detailed and robust bit of work. The supplemental data provided by the investigator is also excellent and I congratulate the authors on their efforts to provide clarity and transparency in this manner.

Response:
We are glad to know that the reviewer liked our work and thank him/her for appreciating our efforts.
4. Line 243, The authors hypothesize that it is the targeting of PLY-H on lipid rafts that is the reason why PLY-H impairs endocytotic processes whereas PLY-NH does not. Indeed PLY is likely to be found on lipid rafts as these are cholesterol rich and support the oligomerization of the toxinultimately causing them to be dispersed. However, it is seems far more likely that the iondysregulation caused by the toxin-formed pores formed by PLY-H is responsible for altering cell signaling within the cell, both due to the ion dysregulation but also energy loss in attempt to resotore the ion gradient. There is also a patch/plug response that will disrupt other normal signaling and endocytosis events -as pore-complexes are expelled. Thus the impact of hemolytic pneumolysin on a cell is very broad with lipid raft formation being only one aspect. This is not adequately recognized.

Response:
We agree with the reviewer that the impact of the pore forming ability of Ply-H on host cell is multifaceted. Apart from its effect on the lipid raft, pores formed by Ply-H on the plasma membrane can cause ion imbalance and disrupt other normal cellular signalling mechanisms which in turn could hinder endocytosis events. This has been now been acknowledged in the text (Page no. 21, lines 419-422). However, disruption of lipid rafts by Ply mediated pores appears to be a key reason for the reduced invasive ability of Ply-H harbouring SPN which we investigated in the present study.
5. The results on prolonged intracellular survival in A549 cells shown in Fig5A (a difference at 16h of 9.5% to 13.5%; at 24h 2.5% to 6%) (are less than compelling). ON the other hand the data shown in Fig 5B with macrophages are more so. With data from the Oggioni lab now demonstrating that Spn replicate intracellularly and emerge, what is the fate of the bacterium and the cells that are hosting them? I think its important for the authors to consider that intracellular persistence within epithelial cells may not be why serotype1 does this and the effect may be principal in other cell types and affecting measures not examined. The fate of the hosting cells may provide an important clue.

Response:
As suggested by the reviewer, we have now performed invasion, intracellular survival and recycling assay on both A549 and THP-1 cells following infection with SPN strain bearing Ply-NH (Fig S13 A, C-D). Our results suggest that though the invasion is higher in THP-1s (due, presumably, to its higher phagocytic uptake ability), the survival and recycling efficiencies are significantly less compared to A549, particularly at later time points. This demonstrates that though there may not be a preference in terms of invasion in a particular cell type, SPN invading the epithelial cells have higher chances of survival and egression as compared to those invading macrophages. We also tested the fate of both the host cell types following pneumococcal infection and found that there is no significant difference in the host cell viability before and after infection with a Ply-NH bearing SPN strain (Fig S13B).
6. Figure 6F. Given the importance of the finding to the authors claim, the authors need to provide quantitative data on the number of intracellular pneumococci they detect within infected lungs. As