Structural basis underlying the synergism of NADase and SLO during group A Streptococcus infection

Group A Streptococcus (GAS) is a strict human pathogen possessing a unique pathogenic trait that utilizes the cooperative activity of NAD+-glycohydrolase (NADase) and Streptolysin O (SLO) to enhance its virulence. How NADase interacts with SLO to synergistically promote GAS cytotoxicity and intracellular survival is a long-standing question. Here, the structure and dynamic nature of the NADase/SLO complex are elucidated by X-ray crystallography and small-angle scattering, illustrating atomic details of the complex interface and functionally relevant conformations. Structure-guided studies reveal a salt-bridge interaction between NADase and SLO is important to cytotoxicity and resistance to phagocytic killing during GAS infection. Furthermore, the biological significance of the NADase/SLO complex in GAS virulence is demonstrated in a murine infection model. Overall, this work delivers the structure-functional relationship of the NADase/SLO complex and pinpoints the key interacting residues that are central to the coordinated actions of NADase and SLO in the pathogenesis of GAS infection.

In "Structural basis underlying the synergism of NADase and SLO in the pathogenesis of Group A Streptococcus infection", Tsai et al examine the interaction of two GAS toxins. Their structural studies appear reasonable and justified. Based on these observations, they generate a point mutant in order to disrupt association between these toxins and to discover the importance of this in pathogenesis. It is the activity of this mutant on which a large portion of their argument lies, and which could use additional validation. The writing is overall quite strong and clear, but in specific areas is lacking in method and statistical detail. It would further benefit from additional analysis and discussion for a larger audience outside specialists intreated in these toxins. For a broader point, this work in part shows these toxins binding is not required for their activity, nor virulence. While this is not a negative finding per se, it is deserving of greater discussion as it somewhat goes against the argument for the study of these proteins. Please see below for more detailed specific comments The animal model methods are unclear, and potentially problematic. It was performed "as previously described" with a reference to Liu et al Front Micro, which does not describe similar methods. It also appears to be an underpowered study. Independent of concerns on how this experiment was done, it is not clear that it has any biological significance. There is no difference in bacterial replication, and the difference in lesion area is slight, and subject to interpretation. It appears they believe 'necrotic skin lesion' (which may just be the red scabbed area) is a specific indicator of disease, but no indication for this is given. Certainly, non-scabbed tissue seems to play a role in disease, so saying that specifically is important does not advance a useful argument. The lack of effect on bacterial replication or mouse survival does not support "enhanced pathogenicity of GAS is demonstrated in vitro and in vivo" (line 335). This rather argues the counter, that this mechanism has little to no effect, and is certainly neither essential nor sufficient. Figure 4 is lacking in controls. First, negative controls of SLO-mutant and NADase mutant are essential to support the specificity of this system. Second, it is not clear that the small number of bacteria in 4C is because they are intracellular, as they could be persisters, adherent on the sides of wells, etc, as their specific location is not shown, just a failure to kill. Third, Translocation should be shown with their proteins, not just bacteria that may have additional differences. Fourth, this should include the NADase(D315R)/SLO(R531D) complex to show functional restoration to support their observations in other experiments (as in 2B). Last, the number of samples (unclear whether biological or technical replicates, and how many experiments) is variable for no clear reason.
The argument of line 245 "discrepant effect of cytochalasin D on bacterial invasion…" does not appear to make sense. In the protocol of 4D, rapid gent treatment means few living bacteria are present most of the time, where 4a has late gent treatment, so most of the activity is present throughout the experiment, but the kill at the end means a focus on a tiny subpopulation that likely has a negligible contribution to total activity. The framing of these experiments makes it unclear what specific argument is being advanced, but with some many different variables, and lots of experimental biases being introduced, its not clear that the mechanisms the authors are trying to connect are causal to their observations. Methods are not described for much of the supplemental materials. S4 for example, appears to be describing activity of purified protein. If so, it is not clear that there is differences in expression or activity when expressed by GAS. S9 appears to be a Western blot, but the conditions are unclear. I'm assuming it could be GAS cultures treated as during an infection. The text claims these are comparable bands, but they look different, theres no loading or positive or negative controls, repeats, or quantification. If all this is controlled for, an effect on protein secretion can then be excluded, but this would be better examined during infection, where effects on protein stability and cell targeting can be examined. Binding of SLO and SPN to the host cell membrane (as in ref 25 and other papers) is ultimately the measure required to be sure differences from the mutation are solely interaction between these proteins and not other, undercharacterized, activities.
NADase activity is a virulence factor and is cytotoxic -if the difference in activity in Fig 4B is biologically significant, why is there no effect on growth in vitro? Cell viability is an essential control for the experiments of Fig 4, as SLO and NADase are both cytotoxic.
Regarding Trp81 (discussed on line 22): Their model appears at odds with known biochemical observations. The speculation provided that dynamics may reveal this residue is not apparent in their structure. Further characterization of this region to support their model is needed.
Presumably, the NADase mutation does not impact IFS inhibition. Have the authors excluded effects from this interaction on bacterial viability and bacterial NAD levels?
Throughout, the numbers and statistics are completely missing or inadequate. Most measures in the reporting summary checklist are absent, including sample size, statistical tests, biological/technical replicates, etc, are absent.
Reviewer #2 (Remarks to the Author): In the manuscript Tsia et. al. solves the structure of a truncated NADase/SLO complex and uncover how they interact with each other. In particular a salt bridge between NADase D315 and SLO R531 is important for the interaction. Further analysis reveals that the interaction and conformational dynamics between NADase and SLO is important for the function of the complex. In addition the authors demonstrate that disrupting the interaction between NADase and SLO affects GAS infection in cells and a mouse model. The manuscript presents novel, important results and is of interest to people involved in the field, but certain issues need to be resolved before it can be considered for publication.
The writing in the manuscript is not clear and it is difficult to follow what the authors are trying to communicate. The manuscript would greatly benefit if a native speaker worked on the writing and sentence formulation.
The X-ray data is fine, and the structure refinement has been performed in a satisfactory manner. The Rmerge value for the high resolution shell is a bit high at 0.54, ideally it should be under 0.5. The data analysis agrees with the solved structure.
The SAXS data is of high quality and analyzed correctly. The modeling based on the data also fits the data well.
The SANS data is also of high quality, but I am uncertain what they are trying to demonstrate with it. A better explanation of why the experiment was performed and what it shows would improve the manuscript.
In Supp fig 7. Legend states "The modelling indicates that 66 % is extended and 34 % compact" I do not know what the authors mean by modelling here, maybe they mean analysis. The experiments on A549 and U937 were carried out with very different MOIs. For the A549 cells the MOI was 5 and for the U937 cells the MOI was 50. Why is there such a huge difference between the MOIs?
The cytotoxicity experiments on U937 cells should be repeated with uninfected cells to determine what the baseline cytotoxicity level is, since the lactate dehydrogenase release will happen with any membrane damage.
For the mouse experiments the age is reported to be 8-10 weeks while in the reporting summary states that the age is 12 weeks.
For figure 7B the measurement of the lesion area is missing.
For supp fig 9 where the authors look at the secreted level of NADase and SLO between WT and mutated GAS should be repeated with a loading control.

Reviewer #3 (Remarks to the Author):
This manuscript described the critical role of D315 of NADase in forming the complex with SLO through X-ray crystal structure by forming the salt bridge with R531 of SLO. The mutation of D315 to Arg disrupted the complex formation. The in vitro and in vivo studies also supported the role of the complex formation of NADase/SLO complex in the pathogenicity of the group A Streptococcus (GAS). The X-ray crystal structure was only from the truncated NADase and SLO with 2.45 angstrom. The authors did study the interaction between full length NADase and SLO using SAXS and SANS. The authors demonstrated the formation of the complex between NADaseD315R/SLOR531D, it would be interesting to see if a double mutant of GAS genome to NADaseD315R/SLOR531D can restore the pathogenicity of GAS (or at least make some discussions). It is not clear why the authors chose G330D mutant for crystallization. It would be interesting to see the crystal structure of both wildtype and G330D mutant complex rather than through superimposing in Fig. S2. Some other minor issues: (1) it would be clearer if the authors define the symbol of Δ as the truncation or use NADase193-451 (same for SLO) in the manuscript; (2) give the full name of MALS (multi-angle light scatting?).

Reviewers' comments:
Reviewer #1 (Remarks to the Author): In "Structural basis underlying the synergism of NADase and SLO in the pathogenesis of Group A Streptococcus infection", Tsai et al examine the interaction of two GAS toxins. Their structural studies appear reasonable and justified. Based on these observations, they generate a point mutant in order to disrupt association between these toxins and to discover the importance of this in pathogenesis. It is the activity of this mutant on which a large portion of their argument lies, and which could use additional validation. The writing is overall quite strong and clear, but in specific areas is lacking in method and statistical detail. It would further benefit from additional analysis and discussion for a larger audience outside specialists intreated in these toxins. For a broader point, this work in part shows these toxins binding is not required for their activity, nor virulence. While this is not a negative finding per se, it is deserving of greater discussion as it somewhat goes against the argument for the study of these proteins. Please see below for more detailed specific comments The animal model methods are unclear, and potentially problematic. It was performed "as previously described" with a reference to Liu et al Front Micro, which does not describe similar methods. It also appears to be an underpowered study. Independent of concerns on how this experiment was done, it is not clear that it has any biological significance. There is no difference in bacterial replication, and the difference in lesion area is slight, and subject to interpretation. It appears they believe 'necrotic skin lesion' (which may just be the red scabbed area) is a specific indicator of disease, but no indication for this is given. Certainly, non-scabbed tissue seems to play a role in disease, so saying that specifically is important does not advance a useful argument. The lack of effect on bacterial replication or mouse survival does not support "enhanced pathogenicity of GAS is demonstrated in vitro and in vivo" (line 335). This rather argues the counter, that this mechanism has little to no effect, and is certainly neither essential nor sufficient. Figure 4 is lacking in controls. First, negative controls of SLO-mutant and NADase mutant are essential to support the specificity of this system. Second, it is not clear that the small number of bacteria in 4C is because they are intracellular, as they could be persisters, adherent on the sides of wells, etc, as their specific location is not shown, just a failure to kill. Third, Translocation should be shown with their proteins, not just bacteria that may have additional differences. Fourth, this should include the NADase(D315R)/SLO(R531D) complex to show functional restoration to support their observations in other experiments (as in 2B). Last, the number of samples (unclear whether biological or technical replicates, and how many experiments) is variable for no clear reason.
The argument of line 245 "discrepant effect of cytochalasin D on bacterial invasion…" does not appear to make sense. In the protocol of 4D, rapid gent treatment means few living bacteria are present most of the time, where 4a has late gent treatment, so most of the activity is present throughout the experiment, but the kill at the end means a focus on a tiny subpopulation that likely has a negligible contribution to total activity. The framing of these experiments makes it unclear what specific argument is being advanced, but with some many different variables, and lots of experimental biases being introduced, its not clear that the mechanisms the authors are trying to connect are causal to their observations. Methods are not described for much of the supplemental materials. S4 for example, appears to be describing activity of purified protein. If so, it is not clear that there is differences in expression or activity when expressed by GAS. S9 appears to be a Western blot, but the conditions are unclear. I'm assuming it could be GAS cultures treated as during an infection. The text claims these are comparable bands, but they look different, theres no loading or positive or negative controls, repeats, or quantification. If all this is controlled for, an effect on protein secretion can then be excluded, but this would be better examined during infection, where effects on protein stability and cell targeting can be examined. Binding of SLO and SPN to the host cell membrane (as in ref 25 and other papers) is ultimately the measure required to be sure differences from the mutation are solely interaction between these proteins and not other, undercharacterized, activities.
NADase activity is a virulence factor and is cytotoxic -if the difference in activity in Fig 4B is biologically significant, why is there no effect on growth in vitro? Cell viability is an essential control for the experiments of Fig 4, as SLO and NADase are both cytotoxic.
Regarding Trp81 (discussed on line 22): Their model appears at odds with known biochemical observations. The speculation provided that dynamics may reveal this residue is not apparent in their structure. Further characterization of this region to support their model is needed.
Presumably, the NADase mutation does not impact IFS inhibition. Have the authors excluded effects from this interaction on bacterial viability and bacterial NAD levels?
Throughout, the numbers and statistics are completely missing or inadequate. Most measures in the reporting summary checklist are absent, including sample size, statistical tests, biological/technical replicates, etc, are absent.
Comment 1: The animal model methods are unclear, and potentially problematic. It was performed "as previously described" with a reference to Liu et al Front Micro, which does not describe similar methods.

Response:
We apologize to the reviewer that the previously cited reference regarding the airpouch animal model is overly brief that causes the confusion. In the revised manuscript, we have included two more references (Kuo et al. 1998;Lu et al. 2013) and added a more detailed description of the procedures in the Materials and Methods section (pages 24-25, lines 539-556). Comment 2: It also appears to be an underpowered study. Independent of concerns on how this experiment was done, it is not clear that it has any biological significance. There is no difference in bacterial replication, and the difference in lesion area is slight, and subject to interpretation. It appears they believe 'necrotic skin lesion' (which may just be the red scabbed area) is a specific indicator of disease, but no indication for this is given. Certainly, non-scabbed tissue seems to play a role in disease, so saying that specifically is important does not advance a useful argument. The lack of effect on bacterial replication or mouse survival does not support "enhanced pathogenicity of GAS is demonstrated in vitro and in vivo" (line 335). This rather argues the counter, that this mechanism has little to no effect, and is certainly neither essential nor sufficient.

Response:
We agree with the reviewer that the animal experiments were underpowered. Thus, we repeated the air pouch animal experiments with increased animal numbers (N=8) to confirm the biological significance of the NADase/SLO complex in GAS virulence. At 24-hour post-infection, the A20 D315R mutant induced significantly smaller sizes of lesions than A20 ( Fig. 5a and 5b, Supplementary Fig. 11). Consistent with the smaller lesion caused by A20 D315R , lower levels of bacterial counts were recovered from A20 D315R than those from A20 (Fig. 5c). Moreover, A20 D315R -infected animals produced lower levels of the inflammatory cytokine IL-1b compared to A20-infected mice (Fig. 5d), suggesting that A20 D315R -infected animals exhibited lower levels of inflammation than the A20-infected ones. Taken together, these findings support that the formation of the NADase/SLO complex contributes to the pathogenesis of GAS. The updated results have been incorporated into the revised manuscript (page 15, lines 316-328). Previously, we showed no significant difference in the bacterial burdens recovered from A20 D315R -and A20-infected mice (Fig. 5C in the previous version), while in the revised version we showed that the bacterial burden in A20 D315R -infected mice was significantly lower than those in A20-infected mice at 24-hour post-infection (Fig. 5c). We apologize that the bacterial counts shown in the previous version was actually recovered from 48-hour post-infection that was mislabeled as 24-hour post-infection. In the previous animal experiments, the images of skin lesions were photographed at 24-hour and 48-hour post-infection (please see the Figure below) and the bacterial burdens were determined at 48-hour post-infection before animal sacrificing. The infected mice may reach the recovered stage at 48-hour post-infection (Figure below), and thus the bacterial burdens and skin lesions in A20 D315R -and A20-infected mice showed no significant difference. Therefore, 24-hour post-infection is a better time window to compare the virulence of A20 D315R and A20. 4 Comment 3: First, negative controls of SLO-mutant and NADase mutant are essential to support the specificity of this system.

Response:
We agree. In the revised manuscript, we have included the SLO-mutant of A20 (A20 Dslo ), NADase mutant (A20 D315R ), A20 and mock in the cytotoxicity and intracellular survival assays. As expected, A20 Dslo showed no significant NADase translocation into A549 cells after the bacterium-host interaction, in comparison to the mock-treated group ( Fig. 4b and 4c). Also, A20 Dslo showed significantly lower survival within the macrophage cell line U937 than A20 and the NADase mutant A20 D315R (Fig. 4g).

Comment 4:
Second, it is not clear that the small number of bacteria in 4C is because they are intracellular, as they could be persisters, adherent on the sides of wells, etc, as their specific location is not shown, just a failure to kill.

Response:
We agree with the reviewer that the small number of bacteria are not necessarily located within the host cells. However, the low bacterial counts were not likely to affect the results of the translocation of NADase because A20 Dslo also showed a similar level of residual live bacteria in the same experiment settings, while no significant NADase translocation was observed ( Fig. 4b  and 4c). This indicates that the residual live bacteria did not play a significant role in affecting the results of the NADase translocation. Given that the Fig. 4C in the previous manuscript causes confusion and is not the main theme of the manuscript, we have removed it from the Result section.
Comment 5: Third, Translocation should be shown with their proteins, not just bacteria that may have additional differences.

Response:
We agree and we have investigated the protein levels of the intracellular NADase after A549 cells were incubated with A20, A20 D315R , and A20 Dslo by immunoblots. Consistent with the results of the intracellular NADase activity, the A20-infected cells showed significantly higher levels of intracellular NADase than A20 D315R , while A20 Dslo -infected cells showed no significant NADase translocation into the cells. The results are included in the revised manuscript (Fig. 4b, page 13, lines 271-274) and the experimental procedures are described in the Materials and Methods section (pages 22-23, lines 493-505). Response: Thank you for the suggestion. To perform the restoration experiments, we have constructed the SLOR531D mutant (A20 R531D ) and NADaseD315R/SLOR531D double mutant (A20 R531D,D315R ) strains. Unexpectedly, significantly lower levels of SLO protein secretion were observed in A20 R531D (Supplementary Fig. 9d and 9e). Given that the translocation of NADase is SLO-dependent, the discrepancy in the SLO secretion levels between A20 and A20 R531D,D315R did not allow us to properly evaluate whether rescuing the salt bridge could restore the ability of NADase translocation in A20 R531D,D315R . On the other hand, A20 R531D and A20 R531D,D315R showed comparable levels of SLO secretion ( Supplementary Fig. 9d and 9e) that permits us to investigate the correlation of salt bridge formation with NADase translocation. The results show A20 R531D,D315R exhibited a higher level of NADase translocation than A20 R531D during GAS-A549 infection (Fig. 4e), demonstrating the regeneration of salt bridge at NADase/SLO complex interface (Fig. 2b) correlates with functional restoration. The new results are incorporated into the Results section (Page 14, lines 289-305).
Comment 7: Last, the number of samples (unclear whether biological or technical replicates, and how many experiments) is variable for no clear reason.

Response:
In the revised manuscript, the results in Figure 4 have been repeated. The data are the representative of three biologically independent experiments performed in quadruplicate.

Comment 8:
The argument of line 245 "discrepant effect of cytochalasin D on bacterial invasion…" does not appear to make sense. In the protocol of 4D, rapid gent treatment means few living bacteria are present most of the time, where 4a has late gent treatment, so most of the activity is present throughout the experiment, but the kill at the end means a focus on a tiny subpopulation that likely has a negligible contribution to total activity. The framing of these experiments makes it unclear what specific argument is being advanced, but with some many different variables, and lots of experimental biases being introduced, its not clear that the mechanisms the authors are trying to connect are causal to their observations.

Response:
We apologize for the confusion. Please allow us to explain in details. The purpose of the experiments is to reveal the biological significance of NADase/SLO complex formation in GAS cytotoxicity (Fig. 4A in the previous manuscript) and in intracellular survival in immune cells (Fig. 4D in the previous manuscript). Therefore, the timings of gentamicin treatment in the two protocols have their respective purpose.
The protocol described in Fig. 4A is to measure the SLO-mediated NADase translocation into host cells and the consequent cytotoxic effects. The reason for the late gentamicin treatment (Fig. 4A in the previous manuscript) is to allow the adequate period of time for the live bacteria to produce and secrete sufficient amount of NADase and SLO. Since the live bacteria may interfere with the measurement of NADase translocation, the late treatment of gentamicin is to kill the live 6 GAS that stay outside the host cells, while cytochalasin D treatment was to block the bacterium internalization by the host cells. Although after the retreatment of both gentamicin and cytochalasin D, some live bacteria still remained (Fig. 4C in the previous manuscript). However, the amounts of the remaining live bacteria were neglectable which did not interfere with the results and the Fig. 4C has been removed from the revised manuscript (response to comment 4). To avoid the confusion, the sentence "discrepant effect of cytochalasin D on bacterial invasion…" has also been removed in the revised version.
The protocol shown in the Fig. 4D of the previous version (Fig. 4f in the revised manuscript) is to measure the intracellular survival of GAS after the bacteria are phagocytosed by macrophages. The early gentamicin treatment was to kill the bacteria not phagocytosed by the macrophages, which allows the measurement of live bacterial amounts within macrophages (Fig. 4g in the revised manuscript).
Comment 9: Methods are not described for much of the supplemental materials. S4 for example, appears to be describing activity of purified protein. If so, it is not clear that there is differences in expression or activity when expressed by GAS. S9 appears to be a Western blot, but the conditions are unclear. I'm assuming it could be GAS cultures treated as during an infection. The text claims these are comparable bands, but they look different, theres no loading or positive or negative controls, repeats, or quantification. If all this is controlled for, an effect on protein secretion can then be excluded, but this would be better examined during infection, where effects on protein stability and cell targeting can be examined. Binding of SLO and SPN to the host cell membrane (as in ref 25 and other papers) is ultimately the measure required to be sure differences from the mutation are solely interaction between these proteins and not other, undercharacterized, activities.

Response:
We apologize and we have added the missing materials and methods in the revised Supplementary Information, including recombinant NADase activity assay, activity of NADase in GAS culture supernatant and the immunoblotting of NADase secretion (pages 2-4, lines 24-77 in Supplementary Information). Regarding the NADase secretion, we have repeated the experiment ( Supplementary Fig. 9a in the revised manuscript) to detect and quantitate the secretion level by immunoblots with a loading control (the total protein of the bacterial lysates in the corresponding culture). Quantification of three independent immunoblots shows A20 and A20 D315R express similar levels of secreted NADase in their culture supernatant ( Supplementary Fig. 9b in the revised manuscript). In addition, we found similar levels of NADase in A20-and A20 D315Rinfected cells were bound to host cell membranes (Fig. 4b in the revised manuscript), suggesting D315R mutation on NADase did not affect the membrane-binding ability of NADase. We also found a similar membrane-binding level of SLO in A20-and A20 D315R -infected cells (Fig. 4b in the revised manuscript). Thus, we believe that the difference in NADase translocation into host cytosol between A20 and A20 D315R is mainly due to interference of the NADase-SLO interaction caused by the D315R mutation.
Comment 10: NADase activity is a virulence factor and is cytotoxic -if the difference in activity in Fig 4B is biologically significant, why is there no effect on growth in vitro? Cell viability is an essential control for the experiments of Fig 4, as SLO and NADase are both cytotoxic.

Reviewer #2 (Remarks to the Author):
In the manuscript Tsia et. al. solves the structure of a truncated NADase/SLO complex and uncover how they interact with each other. In particular a salt bridge between NADase D315 and SLO R531 is important for the interaction. Further analysis reveals that the interaction and conformational dynamics between NADase and SLO is important for the function of the complex. In addition the authors demonstrate that disrupting the interaction between NADase and SLO affects GAS infection in cells and a mouse model. The manuscript presents novel, important results and is of interest to people involved in the field, but certain issues need to be resolved before it can be considered for publication.
The writing in the manuscript is not clear and it is difficult to follow what the authors are trying to communicate. The manuscript would greatly benefit if a native speaker worked on the writing and sentence formulation.
The X-ray data is fine, and the structure refinement has been performed in a satisfactory manner. The Rmerge value for the high resolution shell is a bit high at 0.54, ideally it should be under 0.5. The data analysis agrees with the solved structure.
The SAXS data is of high quality and analyzed correctly. The modeling based on the data also fits the data well.
The SANS data is also of high quality, but I am uncertain what they are trying to demonstrate with it. A better explanation of why the experiment was performed and what it shows would improve the manuscript.
In Supp fig 7. Legend states "The modelling indicates that 66 % is extended and 34 % compact" I do not know what the authors mean by modelling here, maybe they mean analysis. The experiments on A549 and U937 were carried out with very different MOIs. For the A549 cells the MOI was 5 and for the U937 cells the MOI was 50. Why is there such a huge difference between the MOIs?
The cytotoxicity experiments on U937 cells should be repeated with uninfected cells to determine what the baseline cytotoxicity level is, since the lactate dehydrogenase release will happen with any membrane damage.
For the mouse experiments the age is reported to be 8-10 weeks while in the reporting summary states that the age is 12 weeks.