Protein phase change batteries drive innate immune signaling and cell fate
Curation statements for this article:-
Curated by eLife
eLife Assessment
This valuable study investigates the self-assembly activity of death-fold domains. The data collected using advanced microscopy and distributed amphifluoric FRET-based flow cytometry methods provide solid evidence for the conclusions, although the interpretations based on these conclusions remain speculative in some cases. This paper is broad interest to those studying a variety of biological pathways involved in inflammatory responses and various forms of cell death.
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
How minute pathogenic signals trigger decisive immune responses is a fundamental question in biology. Classical signaling often relies on ATP-driven enzymatic cascades, but innate immunity frequently employs death fold domain (DFD) self-assembly. The energetic basis of this assembly is unknown. Here, we show that specific DFDs function as energy reservoirs through metastable supersaturation. Characterizing all 109 human DFDs, we identified sequence-encoded nucleation barriers specifically in the central adaptors of inflammatory signalosomes, allowing them to accumulate far above their saturation concentration while remaining soluble and poised for activation. We demonstrate that the inflammasome adaptor ASC is constitutively supersaturated in vivo , retaining energy to power on-demand cell death. Swapping a non-supersaturable DFD in the apoptosome with a supersaturable one sensitized cells to sublethal stimuli. Mapping all DFD nucleating interactions revealed that supersaturated adaptors are specifically templated by other DFDs in their respective pathways, limiting deleterious crosstalk. Across human cell types, adaptor supersaturation strongly correlates with cell turnover, implicating this thermodynamic principle in the trade-off between immunity and longevity. Profiling homologues from fish, sponge, and bacteria, we find nucleation barriers to be ancestrally conserved. These findings reveal DFD adaptors as biological phase-change materials that function like batteries, storing and privatizing energy for life-or-death decisions.
Article activity feed
-
eLife Assessment
This valuable study investigates the self-assembly activity of death-fold domains. The data collected using advanced microscopy and distributed amphifluoric FRET-based flow cytometry methods provide solid evidence for the conclusions, although the interpretations based on these conclusions remain speculative in some cases. This paper is broad interest to those studying a variety of biological pathways involved in inflammatory responses and various forms of cell death.
-
Reviewer #1 (Public review):
Summary:
This is a high-quality and extensive study that reveals differences in the self-assembly properties of the full set of 109 human death fold domains (DFDs). Distributed amphifluoric FRET (DAmFRET) is a powerful tool that reveals the self-assembly behaviour of the DFDs, in non-seeded and seeded contexts, and allows comparison of the nature and extent of self-assembly. The nature of the barriers to nucleation is revealed in the transition from low to high AmFRET. Alongside analysis of the saturation concentration and protein concentration in the absence of seed, the subset of proteins that exhibited discontinuous transitions to higher-order assemblies was observed to have higher concentrations than DFDs that exhibited continuous transitions. The experiments probing the ~20% of DFDs that exhibit …
Reviewer #1 (Public review):
Summary:
This is a high-quality and extensive study that reveals differences in the self-assembly properties of the full set of 109 human death fold domains (DFDs). Distributed amphifluoric FRET (DAmFRET) is a powerful tool that reveals the self-assembly behaviour of the DFDs, in non-seeded and seeded contexts, and allows comparison of the nature and extent of self-assembly. The nature of the barriers to nucleation is revealed in the transition from low to high AmFRET. Alongside analysis of the saturation concentration and protein concentration in the absence of seed, the subset of proteins that exhibited discontinuous transitions to higher-order assemblies was observed to have higher concentrations than DFDs that exhibited continuous transitions. The experiments probing the ~20% of DFDs that exhibit discontinuous transition to polymeric form suggest that they populate a metastable, supersaturated form in the absence of cognate signal. This is suggestive of a high intrinsic barrier to nucleation.
Strengths:
The differences in self-assembly behaviour are significant and likely identify mechanistic differences across this large family of signalling adapter domains. The work is of high quality, and the evidence for a range of behaviours is strong. This is an important and useful starting point since the different assembly mechanisms point towards specific cellular roles. However, understanding the molecular basis for these differences will require further analysis.
An impressive optogenetic approach was engineered and applied to initiate self-assembly of CASP1 and CASP9 DFDs, as a model for apoptosome initiation in these two DFDs with differing continuous or discontinuous assembly properties. This comparison revealed clear differences in the stability and reversibility of the assemblies, supporting the hypothesis that supersaturation-mediated DFD assembly underlies signal amplification in at least some of the DFDs.
The study reveals interesting correlations between supersaturation of DFD adapters in short- and long-lived cells, suggestive of a relationship between the mechanism of assembly and cellular context. Additionally, the comprehensive nature of the study provides strong evidence that the interactions are almost all homomeric or limited to members of the same DFD subfamily or interaction network. Similar approaches with bacterial proteins from innate immunity operons suggest that their polymerisation may be driven by similar mechanisms.
Weaknesses:
Only a limited investigation of assembly morphology was conducted by microscopy. There was a tendency for discontinuous structures to form fibrillar structures and continuous to populate diffuse or punctate structures, but there was overlap across all categories, which is not fully explored. The methodology used to probe oligomeric assembly and stability (SDD-AGE) does not justify the conclusions drawn regarding stability and native structure within the assemblies.
The work identifies important differences between DFDs and clearly different patterns of association. However, most of the detailed analysis is of the DFDs that exhibit a discontinuous transition, and important questions remain about the majority of other DFDs and why some assemblies should be reversible and others not, and about the nature of signalling arising from a continuous transition to polymeric form.
Some key examples of well-studied DFDs, such as MyD88 and RIPK,1 deserve more discussion, since they display somewhat surprising results. More detailed exploration of these candidates, where much is known about their structures and the nature of the assemblies from other work, could substantiate the conclusions here and transform some of the conclusions from speculative to convincing.
The study concludes with general statements about the relationship between stochastic nucleation and mortality, which provide food for thought and discussion but which, as they concede, are highly speculative. The analogies that are drawn with batteries and privatisation will likely not be clearly understood by all readers. The authors do not discuss limitations of the study or elaborate on further experiments that could interrogate the model.
-
Reviewer #2 (Public review):
Summary:
The manuscript from Rodriguez Gama et al. proposes several interesting conclusions based on different oligomerization properties of Death-Fold Domains (DFDs) in cells, their natural abundance, and supersaturation properties. These ideas are:
(1) DFDs broadly store the cell's energy by remaining in a supersaturated state;
(2) Cells are constantly in a vulnerable state that could lead to cell death;
(3) The cell's lifespan depends on the supersaturation levels of certain DFDs.Overall, the evidence supporting these claims is not completely solid. Some concerns were noted.
Strengths:
Systematic analysis of DFD self-assembly and its relationship with protein abundance, supersaturation, cell longevity, and evolution.
Weaknesses
(1) On page 2, it is stated, "Nucleation barriers increase with the entropic …
Reviewer #2 (Public review):
Summary:
The manuscript from Rodriguez Gama et al. proposes several interesting conclusions based on different oligomerization properties of Death-Fold Domains (DFDs) in cells, their natural abundance, and supersaturation properties. These ideas are:
(1) DFDs broadly store the cell's energy by remaining in a supersaturated state;
(2) Cells are constantly in a vulnerable state that could lead to cell death;
(3) The cell's lifespan depends on the supersaturation levels of certain DFDs.Overall, the evidence supporting these claims is not completely solid. Some concerns were noted.
Strengths:
Systematic analysis of DFD self-assembly and its relationship with protein abundance, supersaturation, cell longevity, and evolution.
Weaknesses
(1) On page 2, it is stated, "Nucleation barriers increase with the entropic cost of assembly. Assemblies with large barriers, therefore, tend to be more ordered than those without. Ordered assembly often manifests as long filaments in cells," as a way to explain the observed results that DFDs assemblies that transitioned discontinuously form fibrils, whereas those that transitioned continuously (low-to-high) formed spherical or amorphous puncta. It is unlikely to be able to differentiate between amorphous and structured puncta by conventional confocal microscopy. Some DFDs self-assemble into structured puncta formed by intertwined fibrils. Such fibril nets are more structured and thus should be associated with a higher entropic cost. Therefore, the results in Figure 1B do not seem to agree with the reasoning described.
(2) Errors for the data shown in Figure 1B would have been very useful to determine whether the population differences between diffuse, punctate, and fibrillar for the continuous (low-to-high) transition are meaningful.
(3) A main concern in the data shown in Figure 1B and F is that the number of counts for discontinuous compared to continuous is small. Thus, the significance of the results is difficult to evaluate in the context of the broad function of DFDs as batteries, as stated at the beginning of the manuscript.
(4) The proteins or domains that are self-seeded (Figure 1F) should be listed such that the reader has a better understanding of whether domains or full-length proteins are considered, whether other domains have an effect on self-seeding (which is not discussed), and whether there is repetition.
(5) The authors indicate an anticorrelation between transcript abundance and Csat based on the data shown in Figure 2B; however, the data are scattered. It is not clear why an anticorrelation is inferred.
(6) It would be useful to indicate the expected range of degree centrality. The differences observed are very small. This is specifically the case for the BC values. The lack of context and the small differences cast doubts on their significance. It would be beneficial to describe these data in the context of the centrality values of other proteins.
(7) Page 3 section title: "Nucleation barriers are a characteristic feature of inflammatory signalosome adaptors." This title seems to contradict the results shown in Figure 2D, where full-length CARD9 and CARD11 are classified as sensors, but it has been reported that they are adaptor proteins with key roles in the inflammatory response. Please see the following references as examples: The adaptor protein CARD9 is essential for the activation of myeloid cells through ITAM-associated and Toll-like receptors. Nat Immunol 8, 619-629 (2007), and Mechanisms of Regulated and Dysregulated CARD11 Signaling in Adaptive Immunity and Disease. Front Immunol. 2018 Sep 19;9:2105.
However, both CARD9 and CARD11 show discontinuous to continuous behavior for the individual DFDs versus full-length proteins, respectively, in contrast to the results obtained for ASC, FADD, etc. FADD plays a key role in apoptosis but shows the same behavior as BCL10 and ASC. However, the manuscript indicates that this behavior is characteristic of inflammatory signalosomes. What is the explanation for adaptor proteins behaving in different ways? This casts doubts about the possibility of deriving general conclusions on the significance of these observations, or the subtitles in the results section seem to be oversimplifications.
(8) IFI16-PYD displays discontinuous behavior according to Figure S1H; however, it is not included in Figure 2D, but AIM 2 is.
(9) To demonstrate that "Nucleation barriers facilitate signal amplification in human cells," constructs using APAF1 CARD, NLRC4 CARD, caspase-9 CARD, and a chimera of the latter are used to create what the authors refer to as apoptsomes. Even though puncta are observed, referring to these assemblies as apoptosomes seems somewhat misleading. In addition, it is not clear why the activity of caspase-9 was not measured directly, instead of that of capsae-3 and 7, which could be activated by other means. The polymerization of caspase-1 CARD with NLRC4 CARD, leading to irreversible puncta, could just mean that the polymers are more stable. In fact, not all DFDs form equally stable or identical complexes, which does not necessarily imply that a nucleation barrier facilitates signal amplification. Could this conclusion be an overstatement?
(10) To demonstrate that "Innate immune adaptors are endogenously supersaturated," it is stated on page 5 that ASC clusters continue to grow for the full duration of the time course and that AIM2-PYD stops growing after 5 min. The data shown in Figure 4F indicate that AIM2-PYD grows after 5 mins, although slowly, and ASC starts to slow down at ~ 13 min. Because ASC has two DFDs, assemblies can grow faster and become bigger. How is this related to supersaturation?
-
Author response:
We appreciate constructive feedback from both reviewers. Reviewer 1 provided a very positive assessment and helpful suggestions for clarity, which we will incorporate.
We also thank Reviewer 2 for their detailed comments. In some instances, their public review raised concerns about specific data or interpretations that are, in fact, already presented and justified in the original manuscript. This feedback has highlighted a need to improve the clarity of our presentation.
In our revised manuscript, we will make key information more prominent to prevent further misunderstandings. We will also provide additional statistical validation for our conclusions, additional data from the optogenetic experiments and high throughput imaging, and further elaborate on the behaviors of specific proteins (FADD, MyD88, and RIPK1). We are …
Author response:
We appreciate constructive feedback from both reviewers. Reviewer 1 provided a very positive assessment and helpful suggestions for clarity, which we will incorporate.
We also thank Reviewer 2 for their detailed comments. In some instances, their public review raised concerns about specific data or interpretations that are, in fact, already presented and justified in the original manuscript. This feedback has highlighted a need to improve the clarity of our presentation.
In our revised manuscript, we will make key information more prominent to prevent further misunderstandings. We will also provide additional statistical validation for our conclusions, additional data from the optogenetic experiments and high throughput imaging, and further elaborate on the behaviors of specific proteins (FADD, MyD88, and RIPK1). We are confident that these revisions will make our findings more transparent and accessible to readers, and we look forward to submitting our revised manuscript.
-
