Inflammasome-mediated GSDMD activation facilitates escape of Candida albicans from macrophages

Candida albicans is the most common cause of fungal sepsis. Inhibition of inflammasome activity confers resistance to polymicrobial and LPS-induced sepsis; however, inflammasome signaling appears to protect against C. albicans infection, so inflammasome inhibitors are not clinically useful for candidiasis. Here we show disruption of GSDMD, a known inflammasome target and key pyroptotic cell death mediator, paradoxically alleviates candidiasis, improving outcomes and survival of Candida-infected mice. Mechanistically, C. albicans hijacked the canonical inflammasome-GSDMD axis-mediated pyroptosis to promote their escape from macrophages, deploying hyphae and candidalysin, a pore-forming toxin expressed by hyphae. GSDMD inhibition alleviated candidiasis by preventing C. albicans escape from macrophages while maintaining inflammasome-dependent but GSDMD-independent IL-1β production for anti-fungal host defenses. This study demonstrates key functions for GSDMD in Candida’s escape from host immunity in vitro and in vivo and suggests that GSDMD may be a potential therapeutic target in C. albicans-induced sepsis.

highlighting to make the message clearer.
The authors show that IL-1b processing and GSDMD cleavage are dependent on caspase-1/11 using caspase-1/11-/-mice and the caspase-1 inhibitor VX765. However, they do no show caspase-1/11 expression and activation (for instance, in WT vs. KO macrophages). Although gsdmd-/macrophages show an increase in the intracellular level of mature IL-1b, suggesting that IL-1b cleavage still occurs, the detection of caspase-1/11 activity or activation would be helpful in both Figure 1 and 2. Also, IL-1 release was impaired in caspase-1/11-/-, gsdmd-/-macrophages or when the VX765 chemical inhibitors was used. The specificity or the global effect of the treatment should be addressed, for instance by looking at other cytokines.
The authors show that KCl treatment (Fig.6) inhibits GSDMD cleavage and cell death. Does KCl inhibit caspase-1 activation or IL-1b processing and release? Is the effect specific for IL-1b? More data are required to interpret the findings properly.
Dynamics (time course data) are required. Previous, initial studies have simplified the message that inflammasome activation = pyroptosis. More recent work including this submission indicates that this message is clearly wrong. The relationship between inflammasome activation, pyroptosis and GSDMD activation in Candida infection is complex. It appears that inflammasome activation can be independent of pyroptosis but that a key event of pyroptosis is GSDMD activation. There is also a combination of two pore forming events: GSDMD and candidalysin. These may act separately or together. However, all the above depends on the time point of analysis (as demonstrated by the Hube and Traven groups) and what is knocked out. The dynamics studies show that there is a defined order of inflammasome/pyroptosis/cell death during C.alb-macrophage interactions (early/late). Thus, undertaking time course assays are important in data interpretation. Note: two publications from the Traven group have not been cited or discussed -see below.
Conceptual. A schematic diagram is also required. Regarding this, the authors should consider that the lack of pyroptosis induction via GSDMD may relate to the early phase escape route by C. albicans (see Kasper et al). The two later phase escape routes may be mediated by candidalysin and physical forces of hyphae (associated with metabolism re: Traven group data). Thus, if you inhibit/KO GSDMD, you may remove one escape route, while at the same time the macrophage can better deal with candidalysin-mediated damage, while protective IL-1b driven inflammation is still induced by candidalysin. Note: candidalysin can induce the inflammasome and pyroptosis-independent cell lysis, and pyroptosis can also be inflammasome-independent (Kasper et al and others). Also note that if you KO the pyroptosis axis by knocking down the inflammasome, you will still see lysis -this means that lysis induced by candidalysin will happen independent of pyroptosis. The Traven lab stressed that this depends on the time point: early phase = pyroptosis-mediated cell death and inflammasome activation; later phase = candidalysin-mediated cell death and inflammasome activation; very late phase = escape by physical forces/cell death and cell death by glucose consumption. Thus, as mentioned, this is complex, and the authors need to discuss their findings more fully and in context regarding these different processes. This is not possible unless they undertake time course studies (see dynamics issue above).
experiments with unopsonised C. albicans to confirm they obtain the same data as with opsonised C. albicans to address the role of complement. A better discussion of this is also required.

Mechanism and novelty
Most of the non-GSDMD data has previously been published (for C. albicans and candidalysin). Regarding the GSDMD data, the authors need to discuss/determine how GSDMD fits functionally and mechanistically in the recognised inflammasome/pyroptosis processes (see above).
Although K+ efflux has been demonstrated as an important event for GSDMD activation, the manuscript lacks depth with respect to possible mechanisms of inflammasome/Caspase-1/GSDMD activation in response to C. albicans. As the mechanisms by which C. albicans regulate inflammasome and GSDMD activation in their model is not described, the manuscript will greatly benefit from at least discussing how this might happen (see Point 1 comments).
In vitro stimulation of macrophages with the GSDMD inhibitor NSA specifically inhibits IL-1b release and pyroptosis following LPS+nigericin stimulation. Does NSA inhibit macrophage cell death and Candida escape from macrophages? What about the inflammatory response following Candida stimulation in the presence of NSA (does NSA specifically control IL-1b? What about other inflammatory cytokines?). Also, the authors show that NSA contributes to increased resistance to Candida infection in vivo. A discussion on this finding is missing and the overall conclusion based on the use of NSA lacks a mechanism. This needs to be addressed.
Work is with GDSMD. What about the role of other Gasdermins? Discussion.
The authors use a full gsdmd KO mouse. Therefore, apart from macrophages, other cells (immune or otherwise) may well be important for the readouts observed in the in vivo work, given that most hematopoietic and non-hematopoietic cells tested have a role in C. albicans infection in vivo. Discussion.
Authors show that significant macrophage death still occurred in the absence of GSDMD and conclude that C. albicans-induced macrophage death could be mediated by both GSDMD-dependent pyroptosis and GSDMD-independent mechanisms. OK, this is new, but no real mechanism is provided.
Candidalysin disruption in C. albicans did not completely abolish IL-1b release from BMDMs, indicating the presence of candidalysin independent IL-1b production. Yes, but no real mechanism is provided.
They indicate that GSDMD disruption paradoxically improved host anti-C. albicans responses, suggesting that C. albicans may have hijacked this defense mechanism to improve its survival in infected hosts. OK, but no mechanism, rather speculation.
The main novelty of the ms is the notion that GSDMD facilitates escape from macrophages. However, C. albicans escape has previously been quantified (Kasper et al) and no difference in piercing or hyphal length was observed in human primary macrophages. All data in the current submission is based on mouse macrophages (where differences in hyphal length was observed), but mouse and human macrophages behave differently. Conclusions/findings should be supported using human macrophages. Furthermore, the authors indicate that candidalysin mediates escape via GSDMD activation and pyroptosis, but that candidalysin also does this via direct membrane lysis. Are the authors indicating these processes are connected or independent? More discussion required.
3. References and lack of discussion One of the authors (Kanneganti) has previously published data on GSDMD cleavage by C. albicans (JBC PMID: 33109609). However, this publication is not cited or discussed in the current submission, which reduces novelty somewhat and raises the question of why wasn't this publication cited? Irrespective, in Fig.2 of the JBC paper, it appears that activation of Casp-1 (p20) is slightly reduced in gsdmd KO mice (however, no quantification is available). Therefore, in the current submission, the authors should also show Caspase-1 activation where relevant as controls (also relevant for Point 1 above).
The authors show that KCl supplementation reduces GSDMD activation and macrophage cell death (Fig.6), but inhibition of macrophage pyroptosis following KCl supplementation has been already shown following Candida infection (PMID: 30131363). Importantly, treatment of KCl also reduces ASC speck formation and the number of CFUs, thus suggesting fungi can use pyroptosis to evade killing by macrophages. However, the authors do not cite or discuss this paper. Needs including with proper discussion.
Important publications and concepts (dynamics) from the Traven Lab have been omitted (see above). "Metabolic competition between host and pathogen dictates inflammasome responses to fungal infection" (PMID: 32750090), and "Glucose homeostasis is important for immune cell viability during candida challenge and host survival of systemic fungal infection" (PMID: 29719235). Should be included in context and discussed.
In the mouse systemic candidiasis model, the authors found that mice infected with the ece1 deficient C. albicans displayed higher survival rates and decreased fungal burden compared with WT C. albicans. While the Swidergall reference is cited, the data are in contract with that ms, which found increased fungal burden with the ece1 deficient C. albicans. This is not discussed. How do the authors explain this?
C. albicans-dependent activation of the inflammasome/pyroptosis is complex. The authors should consider recent reviews on the topic and determine how their work fits into this (e.g.: PMID: 33293167; PMID: 24967821; PMID: 33119702), none of which have been cited.

Minor
The title is rather misleading and needs modifying, as they find that cell death is promoted by both inflammasome-dependent and -independent mechanisms. Also, significant macrophage death occurred in the absence of GSDMD. Furthermore, they find candidalysin-dependent and independent mechanisms, suggesting GSDMD is not the factor 'dictating' escape of C. albicans.
Abstract. Candidalysin is not ON hyphae, it's secreted. Change to "expressed by hyphae".
The ms states: "For in vitro experiments, the GSDMD antagonist necrosulfonamide (NSA) was dissolved in dimethyl sulfoxide (DMSO), with DMSO alone used as control (lanes 554-555)". Did the authors mean in vivo? Consider amending.
The authors suggest that pharmacologic inhibition of GSDMD may offer a new therapeutic strategy for C. albicans infection. This is way too early to state. Given the lack of mechanism and the significant experimentation the authors clearly need to undertake, all such comments should to be removed throughout the ms. 5. Direct comment to sentences of the paper "Mice lacking key inflammasome components are hypersusceptible to Candida infection" That depends on the type of Candida infection. Inflammasome activation can be beneficial in candidiasis, and detrimental i.e. during VVC (immunopathology). The authors need to be more precise. Furthermore, every candidemia patient is different as will be the 'mechanisms' of immunity, thus personalized medicine is key.
"Our results have shown that inflammasome-mediated GSDMD activation dictated C. albicans escape from macrophages" Disagree, and it contradicts some of the data and many statements in the ms. This conclusion should be altered as does the title.
"IL-1b release was significantly reduced when macrophages were treated with candidalysin-deficient C. albicans" and "However, candidalysin disruption in C. albicans did not completely abolish IL-1b release from BMDMs, indicating the presence of candidalysin-independent IL-1b production" That should be pyroptosis. This is likely to be mediated by any other factors of the fungus that provide both signals to the inflammasome.
Many of the key findings/statements in the ms have been shown before, predominantly in the Kasper et al paper. This is not always recognized by the authors. These include: "Infection with candidalysin-deficient C. albicans also induced significantly less LDH release, again suggesting that candidalysin partially contributed to macrophage death" "Similarly, C. albicans escape from macrophages and C. albicans-induced macrophage death were mediated by both candidalysin-independent and -dependent mechanisms ..." "As a pore-forming toxin, candidalysin may facilitate C. albicans-induced rupture of the plasma membrane and macrophage lytic death, which enable C. albicans to escape from host cells " "The hyphae from candidalysin-deficient C. albicans could push and stretch the macrophage plasma membranes, and yet these infected macrophages were PI negative, suggesting that candidalysin was indeed critical for facilitating disruption of host plasma membrane integrity" "It is noteworthy that candidalysin-deficient C. albicans could eventually break the plasma membrane, contributing to candidalysin-independent C. albicans escape. This process may be mediated by mechanical piercing or other hyphal proteins, ..." "C. albicans-induced macrophage death likely mediated by both inflammasome-dependent pyroptosis and inflammasome-independent mechanisms" "...inflammasome-independent macrophage death was also facilitated by candidalysin and hyphae formation" Regarding "This process was also suppressed in Gsdmd -/-BMDMs, demonstrating that GSDMD was important for both candidalysin-independent and -dependent IL-1b release (Fig.5D). " There maybe candidalysin-mediated, GSDMD-independent IL-1b release due to candidalysin pore formation, but if you have both, IL-1β release is much higher likely because the other stimuli provided by C. albicans are poor inducers of inflammasome activation while candidalysin will cause rapid K efflux. Note: escape due to physical forces are independent of GSDMD.
" however, macrophages infected with candidalysin-deficient mutants still released significant amounts of IL-1b, suggesting that other fungal components (e.g., b-glucans) could activate inflammasomes and process IL-1b during C. albicans infection." See above: the time point is important. Early events = pyroptosis = candidalysin independent.

Reviewer #2 (Remarks to the Author):
This is a great manuscript from Kambara and colleagues showing a novel link between GSDMD activation and the intracellular escape of Candida albicans. The data are very clear and convincing, the observation novel and important and I have almost no comments. However, my only reservation is the in vivo relevance of this phenotype. The authors show a clear in vivo phenotype in the GSDMD mice and then go onto to reveal the mechanism using BMDMS. In vivo support for their hypothesis is given at the end with the GSDMD inhibitors (but essentially this is reproducing the the KO phenotype). What is not clear is whether the mechanism derived from the in vitro studies explains the in vivo phenotype. For example, there is evidence from several models (including zebrafish..see work from Rob Wheeler) that Candida hyphenation inside macrophages in vivo is a rare event. Also the impact of tissue matrix on macrophages responses etc. (also shown in in vitro work using 2d vs 3d systems). Thus I would like to see some formal in vivo evidence supporting the hyphae, candidalysin, GSDMD model. Also, the survival of GSDMD deficient mice infected with the ece mutant should be shown (surprising this was not, as this would also help validate their system),

Reviewer #3 (Remarks to the Author):
This study reports that unlike caspase1/11 loss, deletion of the inflammatory caspase substrate GSDMD protects against Candida-induced sepsis and this likely occurs as a consequence of reduced host cell death facilitating increased killing of intracellular Candida and subsequent decreased fungal dissemination. This finding is unexpected as deletion of caspase-1/11, which cleave GSDMD, has the opposite effect and increases animal susceptibility to Candida infection. It is also of significant pharmaceutical interest, as it points towards GSDMD as being a bona fide therapeutic target in this condition. The experiments are well performed using relevant genetically targeted mice or GSDMD inhibitor (NSA), the results clear, and the conclusions supported by the data presented.
I only have a few relatively minor questions for the authors; 1. Figure 1 and Figure 7. The number of times the in vivo experiments were independently repeated to confirm the results presented should be clearly noted in the figure legend for readers.
2. Does the reduced killing capacity of GSDMD deficient macrophages allow them to generate more inflammatory cytokines (e.g. TNF, IL-6), and if so, could this also contribute to anti-fungal responses? 1

Point-by-point responses NCOMMS-21-03664-T
We appreciate the conceptual recognition of the importance of our work by the reviewers and are pleased with the conclusions that "the work is interesting and well written","this is the first work that looks at GSDMD in detail" (Reviewer #1), "this is a great manuscript" "the data are very clear and convincing, the observation novel and important and I have almost no comments" (Reviewer #2), "The experiments are well performed, the results clear, and the conclusions supported by the data presented" (Reviewer #3). We thank the reviewers for raising insightful comments that helped us improve the study and the manuscript substantially. We revised our manuscript by closely following these suggestions. We performed additional experiments/analyses in the past 5 months and added a significant number of new results (Fig.1a, Fig.2h-j, Fig.4c

Reviewer 1 Kambara et al. describe an interesting observation that GSDMD-dependent pyroptosis may promote Candida escape from macrophages and contributes to disseminated candidiasis. Incubation of murine macrophages with C. albicans leads to IL-1b release in a GSDMD-dependent and -independent fashion, and disruption of GSDMD prevents fungal escape while maintaining IL-1b-mediated anti-Candida immune defences. The yeast-to-hypha morphogenetic transition and K+ efflux is crucial for GSDMD activation and cell death. They also address the role of the toxin candidalysin in driving both GSDMD-dependent pyroptosis and GSDMD-independent lytic cell death. Finally, they show that administration of the GSDMD inhibitor necrosulfonamide (NSA) alleviates C. albicans infections in mice. Overall, the work is interesting and well written. While most of the approaches/findings have been published elsewhere regarding macrophage pyroptosis in the context of C. albicans (and candidalysin)
infections, this is the first work that looks at GSDMD in detail. However, the ms lacks several important controls, lacks a coherent/complete mechanism, is inadequately discussed in places, and is speculative in several areas. As such, while the work has great promise, the ms currently provides only an incremental advance rather than a stepwise change in our understanding of C. albicans-induced inflammasomes/pyroptosis/cell death. One could also summarise by saying the data do not currently provide a clear, new picture but do provide the foundations to obtain a clear, new picture. The detailed review that follows is so the authors can convert this interesting paper into something more solid and substantial. I hope the authors take this on board and undertake the necessary work and modifications.

Controls and data interpretation
The ece1 null from Bill Fonzi was used. Candidalysin is generated from the processing of the parent protein Ece1 into 8 (or more) peptides (including candidalysin). Thus, some of the responses observed could be driven by these peptides and not candidalysin. The authors need to undertake experiments using both the candidalysin-only null C. albicans strain and with synthetic candidalysin to confirm that all these responses are due to candidalysin. The authors also do not fulfil Koch's postulates and should also undertake experiments using the ECE1 revertant strain. All these mutants/strains and synthetic candidalysin are readily available from the Hube and Naglik labs. This alone requires a considerable amount of work but is essential to verify many of the ms's conclusions. Furthermore, the authors themselves state that "hyphal proteins other than candidalysin may facilitate GSDMD cleavage and plasma membrane perforation, leading to host cell death etc." Hence, the above experiments will also exclude (or more excitingly, identify) a role for the other Ece1 peptides in the observed responses.
-We appreciate these great suggestions. However, we feel this set of experiments is outside the scope of this study, which focuses on the unexpected role of GSDMD (compared to caspases1/11) in candidiasis. The editor agreed and instructed us not to pursue these experiments at this stage.
The ms states: "Yeast-locked C. albicans failed to trigger GSDMD cleavage and IL-1b production by BMDMs (Fig.4A-B) (lanes 229-230)". Again, in Table Fig. S4 (Fig.S10 in the revised manuscript): "Pro-IL-1b is expressed but cannot be processed". Two concerns here. The data in Figure 4A show a band corresponding to the mature IL-1b in both cell lysate and supernatants. Perhaps, the total amount of processed IL-1b (intracellular and extracellular) in a) It is a good observation by the reviewer. Thanks! We modified the related sentences in the revised manuscript (Page 9 and Fig.S10). Pro-IL-1β was not expressed in untreated macrophages, thus we could not detect any m-IL-1β (intracellular or extracellular). We agree that, based on our results, yeast-locked C. albicans can still release a very small amount of m-IL-1β ( Fig.4a-b and Fig.4c) and induce a small level of cell death (Fig.4G) in macrophages.
b) Sorry for the omission. We added the "untreated macrophages" controls in Fig.4c and Fig.4g. Untreated WT and Gsdmd-/-macrophages did not produce pro-IL-1 and thus the secretion of m-IL-1β was almost undetectable. c) Although C. albicans-induced LDH release was largely hyphae dependent. Yeast-locked C.
albicans could still trigger a small LDH release. We reanalysed the data in Fig.8e as suggested. Indeed, KCl (50 mM) also decreased LDH release in macrophages stimulated with the yeastlocked strain (# p<0.05 vs. untreated macrophages). c) Based on these results, we conclude that candidalysin can trigger both caspase-1-dependent andindependent cell death. The "candidalysin-induced macrophage cell death" described Kasper et al. is also partially mediated by the "GSDMD-dependent pyroptosis" elucidated in our study. We further discuss and clarify this point in the revised manuscript (Page 19).
The authors show that IL-1b processing and GSDMD cleavage are dependent on caspase-1/11 using caspase-1/11-/mice and the caspase-1 inhibitor VX765. However, they do no show caspase-1/11 expression and activation (for instance, in WT vs. KO macrophages). Although gsdmd-/-macrophages show an increase in the intracellular level of mature IL-1b, suggesting that IL-1b cleavage still occurs, the detection of caspase-1/11 activity or activation would be helpful in both Figure 1 and 2. Also, IL-1b release was impaired in caspase-1/11-/-, gsdmd-/-macrophages or when the VX765 chemical inhibitors was used. The specificity or the global effect of the treatment should be addressed, for instance by looking at other cytokines.
We conducted these experiments as suggested: a) We confirmed that GSDMD disruption does not affect the cleavage and activation of caspase 1 ( Fig.1a and Page 6) b) The Caspase-1/11-/-mice and the caspase-1 inhibitor VX765 are commonly used, and the resulting caspase-1 inhibition has been demonstrated by multiple labs including my own lab. We added the related results in Fig.S1a and Fig.S6a in the revised manuscript.
c) As suggested, we measured the amount of TNFα and IL6 secreted by stimulated macrophages and confirmed that the inhibitory effect elicited by caspase-1/11 inhibition ( Fig.S6c and Page 12) or GSDMD disruption ( Fig.S3c and Page 9) is specific for IL-1β secretion.
The authors show that KCl treatment (Fig.6) (Fig.8 in the revised manuscript) inhibits GSDMD cleavage and cell death. Does KCl inhibit caspase-1 activation or IL-1b processing and release? Is the effect specific for IL-1b? More data are required to interpret the findings properly.
-The role of K+ efflux in inflammation/caspase-1 activation is well-documented. We conducted the requested experiments and integrated them in the revised manuscript: c) KCl did not suppress TNFα or IL6 secretion ( Fig.S7c and Page 12).

Dynamics (time course data) are required. Previous, initial studies have simplified the message that inflammasome activation = pyroptosis. More recent work including this submission indicates that this message is clearly wrong. The relationship between inflammasome activation, pyroptosis and GSDMD activation in Candida infection is
complex. It appears that inflammasome activation can be independent of pyroptosis but that a key event of pyroptosis is GSDMD activation. There is also a combination of two pore forming events: GSDMD and candidalysin. These may act  Fig.S8e-g). However, as acknowledged by the reviewer, the relationship between inflammasome activation, candidalysin, pyroptosis, and GSDMD activation in Candida infection is complex. We tried not to overinterpret these results. In the discussion, we focused on the main findings of this study 1) the unexpected role of GSDMD in anti-candida host defense and 2) the interactions between GSDMD and the known factors/pathways/mechanisms (Pages 20-21). We also discussed the potential "defined order of inflammasome/pyroptosis/cell death" as suggested (Page 21-22). b) We defined C. albicans-induced lytic cell death as "macrophage lysis". We adopted this term from Wellington et al (Wellington et al., 2014). When inflammasome and pyroptosis axis is blocked, C. albicans-induced macrophage lysis can not be completely inhibited, suggesting that C. albicans can escape from macrophages via both inflammasome-dependent and -independent mechanisms. Inflammasome-independent macrophage lysis may occur at later time points and/or higher organism burdens.
c) We are familiar with the seminal work reported by the Traven lab. Their two recent publications on the role of metabolic interactions between host and pathogen in modulating Candida-induced inflammasome activation and macrophage viability have now been cited (Tucey et al., 2018;Tucey et al., 2020) and discussed (Page 21-22).

Conceptual. A schematic diagram is also required. Regarding this, the authors should consider that the lack of pyroptosis induction via GSDMD may relate to the early phase escape route by C. albicans (see Kasper et al). The two later phase escape routes may be mediated by candidalysin and physical forces of hyphae (associated with metabolism re: Traven group data). Thus, if you inhibit/KO GSDMD, you may remove one escape route, while at the same time the macrophage can better deal with candidalysin-mediated damage, while protective IL-1b driven inflammation is still induced by candidalysin. Note: candidalysin can induce the inflammasome and pyroptosisindependent cell lysis, and pyroptosis can also be inflammasome-independent (Kasper et al and others)
. Also note that if you KO the pyroptosis axis by knocking down the inflammasome, you will still see lysis -this means that lysis induced by candidalysin will happen independent of pyroptosis. The Traven lab stressed that this depends on the time point: early phase = pyroptosis-mediated cell death and inflammasome activation; later phase = candidalysinmediated cell death and inflammasome activation; very late phase = escape by physical forces/cell death and cell death by glucose consumption. Thus, as mentioned, this is complex, and the authors need to discuss their findings more fully and in context regarding these different processes. This is not possible unless they undertake time course studies (see dynamics issue above).
-We appreciate all these suggestions and modified Fig.S10 (a schematic diagram) accordingly. The model and the related results from other labs were further discussed on Pages 20-22. We focused on the main findings of this study -1) the unexpected role of GSDMD in anti-candida host defense and 2) the interactions between GSDMD and the known factors/pathways/mechanisms. b) NSA directly inhibits GSDMD as reported by Rathkey et al. In this study, we used NSA to demonstrate that pharmacological inhibition of GSDMD is an effective therapeutic strategy for treating candidiasis (although the inhibitor may not be completely specific). Currently, there is no absolutely GSDMD-specific inhibitor.

Although the authors show that NSA treatment alleviates Candida infection and therefore the inflammasome response should not be dramatically impaired, a recent publication (PMID: 31209100, Rashidi et al, -also not cited in this submission) has demonstrated that NSA pre-treatment blocks inflammasome priming and caspase-1 activation independent of GSDMD. The Rashidi findings conflict with the original paper (Rathkey et al. 2018), which is the reference cited in this ms as the main justification for using NSA. Perhaps, the authors should address this point and verify whether NSA is affecting any priming and/or activation event in their system following
c) We added a control as suggested. As observed in Gsdmd -/mice, NSA-treated mice produced less IL-1β after C. albicans infection (Fig.S9g). Similarly, this modest IL-1β production appeared to be sufficient for driving anti-fungal immunity (Fig.9a-f). NSA did not affect priming (NLRP3 expression) or activation (caspase-1 cleavage) event ( Fig.S8a-b). The effect on IL-1β production was specific; C. albicans-induced production of TNFα and IL6 was unaltered (Fig.S8c-d). These results were discussed on Pages 13-14 in the revised manuscript.
The authors used opsonised C. albicans for their experiments. This is different to other publications since opsonised and unopsonised C. albicans can give different results due to the role of complement in fungal immunity (see multiple publications by the Zipfel lab). The authors should undertake multiple experiments with unopsonised C. albicans to confirm they obtain the same data as with opsonised C. albicans to address the role of complement. A better discussion of this is also required. a) In this study, we focused on the candida cells engulfed by macrophages. Compared to unopsonized C. albicans, opsonized C. albicans were engulfed by the macrophages more efficiently, which is consistent with the significance of opsonization in phagocytosis (Kagaya and Fukazawa, 1981;Luo et al., 2013;Marodi et al., 1991;Pereira and Hosking, 1984;Wellington et al., 2003). We added this result in the manuscript (Fig.S3a-b). Thus, to facilitate phagocytosis and most importantly to mimic the serum-containing physiological condition, we used opsonised C. albicans in our experiments. This discussion was included in the revised manuscript on Page 8. b) We appreciate the studies conducted by the Zipfel lab and agree that complements play critical roles in host defense again Candida infection. However, the focus of current study is GSDMD. We feel the experiments aiming to "address the role of complement" are outside the scope of this study.

Mechanism and novelty
Most of the non-GSDMD data has previously been published (for C. albicans and candidalysin). Regarding the GSDMD data, the authors need to discuss/determine how GSDMD fits functionally and mechanistically in the recognised inflammasome/pyroptosis processes (see above).
Although K+ efflux has been demonstrated as an important event for GSDMD activation, the manuscript lacks depth with respect to possible mechanisms of inflammasome/Caspase-1/GSDMD activation in response to C. albicans. As the mechanisms by which C. albicans regulate inflammasome and GSDMD activation in their model is not described, the manuscript will greatly benefit from at least discussing how this might happen (see Point 1 comments).
a) As correctly pointed by this reviewer, the key finding of our study is the unexpected effect of GSDMD disruption. The role of K+ efflux, inflammasome, and Caspase-1 in candida infection has been investigated in previous studies. In the revised manuscript, we further clarified the significance of our study and emphasized the role of GSDMD in C. albicans host defense (Pages 17-18).
b) As suggested, we included a paragraph to discuss the complex mechanisms by which C. albicans regulate inflammasome and GSDMD activation in macrophages (Pages 20-21) and also included a schematic diagram in the revised manuscript (Fig.S10). c) We further discussed the role of K+ efflux in inflammasome/Caspase-1 activation. Studies have shown that the decrease in intracellular K+, which is likely mediated by the TWIK2 potassium efflux channel (Di et al., 2018), is an essential trigger for NLRP3 activation induced by ATP and other DAMPs (Franchi et al., 2007;Muñoz-Planillo et al., 2013;Pétrilli et al., 2007) (Page 12).

In vitro stimulation of macrophages with the GSDMD inhibitor NSA specifically inhibits IL-1b release and pyroptosis following LPS+nigericin stimulation. Does NSA inhibit macrophage cell death and Candida escape from macrophages? What about the inflammatory response following Candida stimulation in the presence of NSA (does NSA specifically control IL-1b? What about other inflammatory cytokines?). Also, the authors show that NSA contributes to increased resistance to Candida infection in vivo. A discussion on this finding is missing and the overall conclusion based on the use of NSA lacks a mechanism. This needs to be addressed.
a) We further discussed the effect of NSA on inflammasome and caspase 1 in the revised manuscript (Page 13). In nigericin-stimulated LPS-primed BMDM cells, NSA does not inhibit other innate immune pathways. Cleavage of GSDMD occurred normally in NSA-treated BMDMs, indicating that NSA does not inhibit caspase-1 under this condition (Fig.5A in Rathkey et al., 2018). b) We conducted the requested experiments and showed that NSA could indeed inhibit macrophage cell death and Candida escape from macrophages ( Fig.S8e-g). NSA treatment led to reduced production of IL-1β, but not other cytokines (Fig.S8d). c) We show that NSA contributes to increased resistance to Candida infection in vivo. A discussion on this finding was added in the revised manuscript (Pages 13-14). It is well established that NSA can efficiently inhibit GSDMD-mediated pyroptosis. Thus, our results confirmed that inhibition of GSDMD can induce resistance to Candida infection. As observed in Gsdmd -/mice, NSA-treated mice produced less IL-1β after C. albicans infection (Fig.9g). Similarly, this modest IL-1β production appeared to be sufficient for initiating anti-fungal immunity.
Work is with GDSMD. What about the role of other Gasdermins? Discussion.
-As suggested, we discussed the role of other Gasdermins in the revised manuscript (Page 18).
The authors use a full gsdmd KO mouse. Therefore, apart from macrophages, other cells (immune or otherwise) may well be important for the readouts observed in the in vivo work, given that most hematopoietic and nonhematopoietic cells tested have a role in C. albicans infection in vivo. Discussion.
-As suggested, we discussed this point in the revised manuscript (Page 18).
Authors show that significant macrophage death still occurred in the absence of GSDMD and conclude that C. albicans-induced macrophage death could be mediated by both GSDMD-dependent pyroptosis and GSDMDindependent mechanisms. OK, this is new, but no real mechanism is provided.
Candidalysin disruption in C. albicans did not completely abolish IL-1b release from BMDMs, indicating the presence of candidalysin independent IL-1b production. Yes, but no real mechanism is provided.
They indicate that GSDMD disruption paradoxically improved host anti-C. albicans responses, suggesting that C. albicans may have hijacked this defense mechanism to improve its survival in infected hosts. OK, but no mechanism, rather speculation.
-(Also see above) We agree. As pointed out by this reviewer, the relationship between inflammasome activation, candidalysin, pyroptosis, and GSDMD activation in Candida infection is complex. Several labs have made major contributions to the growth of understanding of the underlying mechanisms. It took many years and tremendous efforts to accomplish this. Here we do not intend to reveal all the remaining mechanisms and/or to establish a unified model. The major discovery of this study is that GSDMD disruption protects against C. albicans-induced sepsis, contrary to observations in inflammasome-deficient (e.g., Casp1/11 -/-) mice. We further emphasized the significance of this finding in the revised manuscript (Pages 17-18). Uncovering the GSDMD-and/or candidalysin-independent mechanisms will extend beyond the current study.
-As suggested by the reviewer, we added a paragraph to further discuss these potential mechanisms (Pages 20-21) and also included a schematic diagram in the revised manuscript (Fig.S10). After discussing with the co-authors and colleagues, we all feel that we should focus on what we did and avoid overinterpreting our results. As correctly pointed out by this reviewer, the main finding of this study is the unexpected role of GSDMD in anti-candida host defense. The role of inflammasome, macrophage lysis, candidalysin, and the "two-phased model" have all been previously reported. Thus, instead of trying to establish a unified model, we focused on the interactions between GSDMD and the known factors/pathways/mechanisms (Pages 20-21 and Fig.S10).
The main novelty of the ms is the notion that GSDMD facilitates escape from macrophages. However, C. albicans escape has previously been quantified (Kasper et al) and no difference in piercing or hyphal length was observed in human primary macrophages. All data in the current submission is based on mouse macrophages (where differences in hyphal length was observed), but mouse and human macrophages behave differently. Conclusions/findings should be supported using human macrophages. Furthermore, the authors indicate that candidalysin mediates escape via GSDMD activation and pyroptosis, but that candidalysin also does this via direct membrane lysis. Are the authors indicating these processes are connected or independent? More discussion required.
a) Thanks for the comments. We conducted the suggested experiments using human monocytederived macrophages (hMDMs). Both WT and candidalysin-deficient C. albicans were used. The human GSDMD and caspase-1 were pharmacologically inhibited by NSA (Fig.10) and VX765 (Fig.S9), respectively. We confirmed that GSDMD and candidalysin also played a critical role in clearance of C. albicans by human macrophages (Pages 14-15). b) Kasper et al previously showed that "there was no difference (between WT and candidalysindeficient C. albicans) in piercing or hyphal length in human primary macrophages" (Fig.2f in Kasper et al PMID: 30323213). The purpose of these experiments was to demonstrate that candidalysin did not affect the hyphal growth. Thus, what they measured was the average hyphal length in the cells (10-20 μm) and they measured it at the relatively early stage of infection (3 h p.i.) when hyphae were still growing. We were not able to find other details (e.g the number of cells/data point and the number of hyphal/cell examined) regarding these experiments. The conclusion of this experiment is that the decreased inflammasome activation in the absence of ECE1 was not due to reduced uptake of fungal cells or hyphal defects. In contrast, the purpose of our experiments is to assess the constraint of Candida growth in intact macrophages. Thus, we measured the maximum length of intracellular hyphae in WT and Gsdmd-/-macrophages (>25 μm), and we measured it at 6 h after the infection. Under this condition, due to the impaired escape, candidalysin-deficient C. albicans grew longer hyphae (>50 μm) inside macrophages (Fig.S4b). These hyphae from candidalysin-deficient C. albicans could push and stretch the macrophage plasma membranes, and yet the infected macrophages were PI negative, suggesting that candidalysin was indeed critical for facilitating disruption of host plasma membrane integrity (Fig.S4a). In this experiment, at least 100 cells were assessed for each sample. We also counted the number of escaped hyphae per macrophage (Fig.3d). After 6 h, there were too many hyphae to be counted accurately.
c) Our results suggest that candidalysin can trigger both GSDMD-dependent and -independent macrophage cell death. Candidalysin can induce K + efflux through plasma membrane perforation, which in turn leads to canonical inflammasome activation, GSDMD cleavage, and macrophage pyroptosis. As a pore-forming toxin, candidalysin can also induce macrophage lysis in the absence of GSDMD. These are two parallel and independent mechanisms. We discussed this point in the revised manuscript (Page 19).

References and lack of discussion
One of the authors (Kanneganti) has previously published data on GSDMD cleavage by C. albicans (JBC PMID: 33109609). However, this publication is not cited or discussed in the current submission, which reduces novelty somewhat and raises the question of why wasn't this publication cited? Irrespective, in Fig.2 of the JBC paper, it appears that activation of Casp-1 (p20) is slightly reduced in gsdmd KO mice (however, no quantification is available). Therefore, in the current submission, the authors should also show Caspase-1 activation where relevant as controls (also relevant for Point 1 above).
-Sorry for the confusion. Dr. Thirumala-Devi Kanneganti, the corresponding author of the abovementioned paper (JBC PMID: 33109609), is a professor at the St. Jude Children's Research Hospital. Apurva Kanneganti, a Harvard undergraduate student and a co-author of this paper, is a different person. The paper from the Kanneganti lab had not been published when we prepared our manuscript. We now cited this paper in the revised manuscript (Page 17).
-(Also see above) As suggested, we assessed Caspase-1 activation in the in the revised manuscript.
The authors show that KCl supplementation reduces GSDMD activation and macrophage cell death (Fig.6), but inhibition of macrophage pyroptosis following KCl supplementation has been already shown following Candida infection (PMID: 30131363). Importantly, treatment of KCl also reduces ASC speck formation and the number of CFUs, thus suggesting fungi can use pyroptosis to evade killing by macrophages. However, the authors do not cite or discuss this paper. Needs including with proper discussion.
-This is a great point! We cited the paper (O'Meara et al, 2018) and discussed the related results in the revised manuscript (Page 12).

Important publications and concepts (dynamics) from the Traven Lab have been omitted (see above). "Metabolic competition between host and pathogen dictates inflammasome responses to fungal infection" (PMID: 32750090),
and "Glucose homeostasis is important for immune cell viability during candida challenge and host survival of systemic fungal infection" (PMID: 29719235). Should be included in context and discussed.
-(Also see above) We cited and discussed these studies in the revised manuscript.
In the mouse systemic candidiasis model, the authors found that mice infected with the ece1 deficient C. albicans displayed higher survival rates and decreased fungal burden compared with WT C. albicans. While the Swidergall reference is cited, the data are in contract with that ms, which found increased fungal burden with the ece1 deficient C. albicans. This is not discussed. How do the authors explain this?
-This is a great observation. Thanks for pointing this out. Both my lab and Naglik lab (Swidergall et al., 2019) reported that candidalysin promotes mortality in murine models of systemic fungal infection. However, Naglik lab focused on neutrophil recruitment. Their data indicated that candidalysin was required for neutrophil recruitment; thus when ece1 deficient C. albicans (2 × 10 5 ) were used to infect mice, fungal burden of the kidney at 1 day increased significantly due to reduced neutrophil accumulation. This effect was diminished 4 days post-infection. Our study focuses on the escape of C. albicans from macrophages, and we inoculated mice with more Candida albicans cells (1 × 10 6 ) and assessed fungal burden 2 days post-infection. In this setup, mice infected with the ece1 deficient C. albicans displayed decreased fungal burden compared with WT C. albicans.
-Thus, the discrepancy is likely caused by different experimental conditions in their compared with our experiments. Swidengall et al sacrificed the mice at 1 and 4 days, whereas we sacrificed them at 2 days. Our inoculum was 1x10 6 cells, theirs was 2x10 5 cells for immunocompetent mice and 5x10 4 for neutropenic mice. Additionally, experiments of Swidergall et al and ours were performed using different strains. We used a strain received from the lab who first cloned and deleted ECE1 (Birse et al., 1993). Swidergall et al used different strains, that were differently constructed, first published in Moyes et al paper (Moyes et al., 2016).
-Of note, in an oral candidiasis model (Moyes et al., 2016)(another paper published by the Naglik lab), the fungal burden was also lower in ece Δ/Δ infection compared with wild type, in line with our own findings. In the Moyes et al study, it was revealed that "tongue tissue from ece1Δ/Δinfected animals (n = 17/20) showed no invasive fungi and no inflammatory infiltrates or damage (Fig.1g)" and "very low numbers of ece1Δ/Δ cells in only 3/20 mice (Extended Data Fig. 2a) were detected". This discrepancy was not explained or mentioned in the 2019 Swidengall et al paper (Swidergall et al., 2019).
-In summary, Swidergall et al. saw higher organ fungal burdens, while Moyes et al and we saw concordant results of lower fungal burdens and decreased virulence in ece Δ/Δ infection compared with wild type. We discussed these results in the revised manuscript (Page 21).
higher likely because the other stimuli provided by C. albicans are poor inducers of inflammasome activation while candidalysin will cause rapid K efflux. Note: escape due to physical forces are independent of GSDMD.
" however, macrophages infected with candidalysin-deficient mutants still released significant amounts of IL-1b, suggesting that other fungal components (e.g., b-glucans) could activate inflammasomes and process IL-1b during C. albicans infection." See above: the time point is important. Early events = pyroptosis = candidalysin independent.
-(Also see above) The relationship between inflammasome activation, candidalysin, pyroptosis, and GSDMD activation in Candida infection is complex. We added a paragraph to further discuss these potential mechanisms (Pages 20-21) and also included a schematic diagram in the revised manuscript (Fig.S10).

Reviewer 2
This is a great manuscript from Kambara and colleagues showing a novel link between GSDMD activation and the intracellular escape of Candida albicans. The data are very clear and convincing, the observation novel and important and I have almost no comments. However, my only reservation is the in vivo relevance of this phenotype. The authors show a clear in vivo phenotype in the GSDMD mice and then go onto to reveal the mechanism using BMDMS. In vivo support for their hypothesis is given at the end with the GSDMD inhibitors (but essentially this is reproducing the the KO phenotype). What is not clear is whether the mechanism derived from the in vitro studies explains the in vivo phenotype. For example, there is evidence from several models (including zebrafish..see work from Rob Wheeler) that Candida hyphenation inside macrophages in vivo is a rare event. Also the impact of tissue matrix on macrophages responses etc. (also shown in in vitro work using 2d vs 3d systems). Thus I would like to see some formal in vivo evidence supporting the hyphae, candidalysin, GSDMD model. Also, the survival of GSDMD deficient mice infected with the ece mutant should be shown (surprising this was not, as this would also help validate their system).
-We are pleased with the complimentary comment of the reviewer "This is a great manuscript …The data are very clear and convincing, the observation novel and important and I have almost no comments".
-As suggested by this reviewer, we conducted several additional experiments to provide more in vivo evidence supporting our hyphae-candidalysin-GSDMD model.

a)
We examined candida hyphaenation inside macrophages in vivo in WT and GSDMD KO mice challenged with the either WT or candidalysin-deficient (ece1Δ/Δ) C. albicans. These data are included in Fig.S5 in the revised manuscript. The results are described and discussed on Page 11. The related references were also discussed.

b)
As suggested, we measured the survival, body weight change, and clinical score of GSDMDdeficient mice infected with ece1Δ/Δ C. albicans. These data are included in Fig.S4c-i. The related results are described and discussed on Page 11.

Reviewer 3
This study reports that unlike caspase1/11 loss, deletion of the inflammatory caspase substrate GSDMD protects against Candida-induced sepsis and this likely occurs as a consequence of reduced host cell death facilitating increased killing of intracellular Candida and subsequent decreased fungal dissemination. This finding is unexpected as deletion of caspase-1/11, which cleave GSDMD, has the opposite effect and increases animal susceptibility to Candida infection. It is also of significant pharmaceutical interest, as it points towards GSDMD as being a bona fide therapeutic target in this condition. The experiments are well performed using relevant genetically targeted mice or GSDMD inhibitor (NSA), the results clear, and the conclusions supported by the data presented.
-We are pleased with the conclusion of this reviewer that "The experiments are well performed using relevant genetically targeted mice or GSDMD inhibitor (NSA), the results clear, and the conclusions supported by the data presented". The reviewer only has "a few relatively minor questions for the authors": -Thanks for pointing this out. We clarified these details in the figure and figure legends ( Fig.1 and  Fig.9).
2. Does the increased killing capacity of GSDMD deficient macrophages allow them to generate more inflammatory cytokines (e.g. TNF, IL-6), and if so, could this also contribute to anti-fungal responses?
-We measured the levels of TNFα and IL-6 produced by infected WT and GSDMD deficient macrophages as suggested. Candida infection could directly trigger TNFα and IL-6 production in macrophages. However, WT and GSDMD deficient macrophages generated the same amount of TNFα and IL-6 after Candida infection, supporting the notion that expression and secretion of these two inflammatory cytokines are GSDMD-independent. Thus, the elevated anti-fungal response observed in the GSDMD-deficiency mice was unlikely mediated by overall upregulation of inflammatory cytokine production These data are included in Fig.S3c in the revised manuscript. The results are described and discussed on Page 9.
3. The authors suggest that caspase-1 mediated IL-1beta activation and release, which still occurs in GSDMD deficient cells, explains why caspase-1/11 and GSDMD loss have the opposite phenotype, despite both being required for pyroptosis. Can the authors prove this by, for example, using neutralising IL-1beta antibodies in infected GSDMD KO mice to reverse the protective phenotype?
-This is a great suggestion! We performed this experiment using anakinra, a commonly used interleukin 1 receptor antagonist. In the presence of anakinra, the protective effect elicited by GSDMD disruption was completely inhibited. Anakinra-treated WT and GSDMD-deficiency mice displayed worse clinical score and reduced survival after C. albicans infection compared to untreated GSDMDdeficient mice (Fig.2h-j, Fig.S2, and Page 7).