Cell junction disruption drives translocation of gasdermin A and gasdermin B from cytoskeleton to plasma membrane during acantholysis

Reviewer #1 (Public review):

The manuscript by Kang et al. examines Gasdermin family member localization in keratinocytes and further reports that the PP2A inhibitor caliculyn A induces a pyroptosis-like morphology, presumably due to cytoskeletal disruption. Gasdermin A and B then translocate from the cytoskeleton to the membrane, where they enable keratin secretion without cleavage. The manuscript is well-controlled, and the data are convincing. The findings could be triangulated more effectively with additional microscopic studies to complement the fractionation experiments.

My specific comments are as follows:

(1) Nitrogen cavitation experiments and fluorescence microscopy have shown that the N-terminus of Gasdermin A first localizes to the mitochondrial membrane and then to the plasma membrane (Kondolf et al., JBC, 2024). The same was shown for GSDMD by Judy Lieberman's group in a 2024 Immunity paper. It is surprising that the fractionation experiments didn't include mitochondrial localization.

(2) It appears that the blots in Figure 2 showing purity of fractionation (Lamin, LDHA, KRT19) are the same as those in Figure 1. I am assuming the experiment was performed at the same time. I think this should be made crystal clear, and it also raises the question of why not just include GSDMC in Figure 1, since they already have the other Gasdermins in that figure.

(3) There's negative data in this manuscript, but I consider it important. Gasdermin antibodies are problematic, and caspase cleavage of GSDMC has been inconsistent. This negative data is important to share.

(4) In Figure 1, why do GSDMD and GSDME show cleavage products without stimulation (minor point for curiosity)?

(5) The fractionation experiments would be more convincing if they were complemented by microscopic localization. Antibodies might be an issue, but others have used fluorescent tags successfully (need to tag the Gasdermins internally).

(6) PP2A is a broad serine-threonine phosphatase with hundreds of substrates. It's hard to rule in a cytoskeletal role to the exclusion of all the other changes caliculyn will cause - especially after 3 hours of exposure. A paragraph discussing this could be added to the Discussion. It's surprising that the authors don't disrupt the cytoskeleton in other ways.

(7) I'm confused by the data in Figure 5C. Knocking down GSDMA and GSDMB causes no change in keratin release (and may even increase it). If the point of the manuscript is that these cause protein secretion, shouldn't there be a change? Do they affect secretion?

(8) Along these lines, which GSDMB isoforms are present in the cell? It's well established that exon structure affects function.

(9) Minor point for curiosity: N/TERT cells can be differentiated and can form organoids. Have the authors attempted to determine if differentiation affects their findings? This would be especially interesting given the different localizations of GSDMA and GSDMB in the layers of the skin.

Overall, this manuscript provides valuable data for the field. Although it includes negative results, I believe sharing them is crucial. The experiments are well-executed and mostly convincing, though some aspects, particularly Figure 5C, appear contradictory. The manuscript would benefit from this reputable lab addressing the earlier points, especially the microscopic testing of the hypothesis.

Reviewer #2 (Public review):

Summary:

This manuscript, by Kang and colleagues, examines the activity of the Gasdermin (GSDM) family of proteins in epidermal skin cells called keratinocytes. Gasdermins are a relatively recently discovered family of pore-forming proteins that are generally studied for their role in inflammasome activation and pyroptosis. The role and functions of GSDMs in keratinocytes are not fully understood, but programmed cell death is important to keratinocytes as part of the cornification process. Keratinocyte pyroptosis associated with GSDMD and GSDME has been linked to the pathogenesis of psoriasis, and GSDMA-mediated pyroptosis in keratinocytes is responsive to Staphylococcus aureus infection. These data suggest that defining the role of GSDM in keratinocytes could provide valuable insight into the treatment of skin pathologies, particularly those involving inflammation.

Strengths:

In this study, the authors use two human keratinocyte cell lines to focus primarily on the biology of GSDM in keratinocytes, particularly A, B and C, both during homeostatic conditions and in response to some pathogenic stimuli. The data indicate that the full-length GSDMA and B proteins are colocalizing not to the plasma membrane, but to the cytoskeletal, cytoplasmic or membrane fractions, in direct contrast to a number of other studies. Further, there are data showing the GSDM are not cleaved by the same proteases as in other studies, and that GSDMA is not involved control of Vaccinia virus skin infection. The controls for cleavage done in HeLa cells with a number of inhibitors were particularly strong, and the overall finding is interesting and potentially important, providing novel insight in the biology of GSDM proteins in keratinocytes. The second half of the paper focuses on treatment with calyculin A, a protein phosphatase inhibitor that the authors report recapitulates acantholysis, i.e. the loss of cell-cell junctions between keratinocytes. Calyculin A altered the localization of GSDM but increased cell death and keratin release irrespective of GSDM activity, suggesting that acantholytic cell death does not involve GSDM. Overall, these data are an interesting study showing differences in GSDM biology in keratinocytes and indicating that GSDM are not involved in a number of pathogenic processes associated with skin diseases.

Weaknesses:

Unfortunately, the data supporting these findings are not particularly strong and are somewhat confusing, and the study is not well organized. One of the major weaknesses is that most work was done in two human keratinocyte cell lines, and subcellular localization subcellular localization results were not consistent between the two lines. This is not hugely surprising, as cell lines even from the same cell type are very different, and the authors note that there may be some variability associated with the model used. However, the inconsistency between the two lines, alongside the lack of comparison to primary cells or patient data, makes it very difficult to determine which, if any, result accurately recapitulates the biology in human cells. The authors do suggest that the N-TERT line is more similar to primary cells but rely on a single study and no other research or data to support this. Several other experiments compare the keratinocyte lines to a colorectal adenocarcinoma cell line, showing the keratinocytes are also different from these cells, but it is not clear that they support the information about GSDM in either keratinocyte line other than showing that keratinocytes are distinct from colorectal cells.

There are several specific concerns regarding the data, particularly the western blotting. In a number of the blots, there are bands which are not mentioned. For example, in Figure 3A, "GSDMA and GSDMB translocated from the cytoskeleton equally to both the cytoplasmic and membrane fractions upon calyculin A treatment". But in the western blot shown, GSDMA moves from being entirely in the cytoskeleton to having lighter bands in the cytoskeleton and chromatin-bound regions and heavier bands in the membrane and cytoplasm. GSDMB shows heavier bands in the membrane and cytoplasm after Calyculin A, but the heaviest band remains in the chromatin-bound region. In addition, the western blots have no associated N or pooled data graph with statistics, so it is not clear how many studies it represents or if there were significant changes. A lack of statistical analysis was also seen in the Yoyo-1, membrane ballooning and cell death analyses in 3A, 3C, 4B and 4C, with the membrane ballooning finding being entirely qualitative and not using any image analysis - comparison of P2A or cell area could resolve this. The methods do note that unpaired T-tests and 2-way ANOVAs were performed in some places, so the results of these should be clearly shown on the graphs. Beyond this, all of the localization studies rely on a single modality, western blotting. This finding should be supported by orthogonal approaches, ideally immunofluorescence. Finally, the presence of GAPDH in the supernatant (5A) raises an important issue regarding nonspecific intracellular release. If the GAPDH is present in the sup that is representative of cell damage and does not correlate with the absence of other intracellular markers - such as those found in lysates - in the supernatant.

The final weakness of the paper was that it is not really clear what the study is trying to say, and it was difficult to follow the logic and data, with several figures being out of order. There are a number of statements that feel overinterpreted, such as stating that clathrin-mediated endocytosis is required for calyculin A mediated acantholysis. While it is clear that Calyculin A recapitulates features similar to acantholysis, it is a broad perturbation of the cell, and it is not clear that it is the same process. Further, the pharmacological evidence is not strong enough to make such a clear claim about the role of clathrin-mediated endocytosis. Similarly, the statement that "GSDMs can induce pyroptosis not only in their cleaved forms but also as full-length proteins through specific post-translational modifications" is much too strong for these data. The terminology surrounding "pyroptosis" should be used more carefully, as yoyo-1 uptake and membrane ballooning demonstrate membrane permeability and morphological changes but are not sufficient for cell death.