Article activity feed

  1. eLife assessment

    This paper will be of interest to cell biologists, structural biologists, and biophysicists studying programmed cell death, membrane transport, and protein-lipid interactions. The simulation data presented offers atomistic detail of how gasdermin-D N-terminal domains assemble on the plasma membrane and trigger the formation of membrane pores which lead to pyroptosis. The study is well designed and the resulting data are rigorously analyzed; however, some clarifications and additional data are required to fully justify the conclusions.

  2. Reviewer #1 (Public Review):

    Generally, the strength of this work is the submolecular resolution provided by the MD simulations, while at the same time the weakness is that the results of such simulations are only as good as the force fields used to describe the interactions between gasdermin-D subunits within an assembly and between these subunits and the lipids in the membrane. These simulations yield several interesting results, while also raising various questions, as follows.

    The MD simulations are consistent with previous results (e.g., Ding et al., 2016 and Mulvihill et al., 2018, cited in manuscript that gasdermins preferentially bind to anionic lipids, which is not new, but the results are novel here in identifying these interactions at submolecular scale. However, by only showing results for interactions with PI(4,5)P2, without any results for other lipids (if only as a negative control), it remains hard for a reader to assess the relevance and strength of these interactions.

    The next result is that the 33-mer "prepore" gasdermin assembly deforms the membrane by just binding - and not inserting into - the membrane. It may seem surprising that such an effect of the membrane may occur without membrane insertion, but it is consistent with previous (experimental) observations for prepore assemblies of the cholesterol dependent cytolysins pneumolysin (Tilley et al., 2005; Faraj et al., Sci Rep 2020) and suilysin (Leung et al., 2014, cited in manuscript). The authors also present data on the 33-mer ring-shaped pore confirmation, not surprisingly finding this pore to be stable.

    The more novel results emerge when considering monomers and smaller oligomers. To assess their potential role in pore formation, MD simulations are shown that demonstrate stability of inserted monomers, dimers, etc. of gasdermin-D. Although, as noted by the authors, arc-shaped pores are a common feature for pore forming proteins, it is quite remarkable that a monomer is enough to provide a stable membrane-inserted configuration. The unanswered question, however, is if such smaller gasdermin assemblies will be able to insert into the membrane, presuming that there may be an activation barrier to overcome between prepore (membrane-bound) and pore (membrane-inserted) configuration. That is, while the MD results how that such small oligomers can adopt stable membrane-inserted configurations, they do not justify the authors' claim that such oligomers "create" membrane pores.

    The final main and valuable result is about the fate of the lipids in arc-shaped gasdermin assemblies, although the comparison with the ring-shaped pore is lacking (e.g., by initiating the pore assembly with lipids still embedded within the ring). For the arc-shaped pores, the lipids are shown to recede from the inside of the arcs, providing new insight into how the membrane is locally removed. Most intriguingly, the line tension of the lipids appear to "crack" the 16-mer assembly, resulting in a smaller-aperture slit-shaped pores (as have observed by AFM previously). One weakness here is that only a single such cracking event (N=1) is shown to result in the slit-shaped pore.

    Another question is how this observation relates to previous MD simulations (by the same lab, Vogele et al, 2019) of pneumolysin pores. Based on MD results and structural details, how do gasdermin-D and pneumolysin compare when viewed through the lens of MD?

    Finally, the authors conclude that there are two distinct pathways of membrane pore formation by gasdermin-D (Fig. 5), but do not explain why they exclude formation of larger arc-shaped "adhered prepores" as a pathway of pore formation. Why would larger adhered prepores only insert into the membrane as full rings and not as larger arc-shaped assemblies? That conclusion does not seem justified by the data.

  3. Reviewer #2 (Public Review):

    Schaefer and Hummer have performed all-atom molecular dynamics (MD) simulations to study the mechanism of GSDMDNT assembly in membranes closely resembling human plasma membranes. Poses of GSDMDNT-lipid interaction were analyzed. Comparing the assemblies of different GSDMDNT oligomeric states reveals key steps in the membrane pore formation by GSDMDNT, resulting in a model with two GSDMDNT concentration-dependent pathways. That is, low concentration favors monomer insertion followed by assembly in the membranes, whereas high concentration promotes prepore formation at the membrane surface followed by membrane insertion to mature into pore. This model is valuable since it reconciles different experiments that cast doubt on the exact order and mechanism with which GSDMDNT binds the plasma membranes. With comparisons against the existing studies, this paper has provided a better understanding of how various factors such as GSDMDNT concentration and, in particular, the membrane composition may influence the process. The study was well carried out. Given the system size, complexity of the membrane composition, and abundance of cholesterol, the simulations were conducted with strong physical rigor (e.g., long all-atom equilibration with tensionless membranes and with cholesterol flip-flop in equilibrium). The paper was well-organized and nicely written.