Molecular mechanisms of NLR proteins in inflammasome activation

Abstract

Nucleotide-binding domain and leucine-rich repeat-containing proteins (NLRs) are central mediators of inflammasome signaling, activating caspase-1 to drive pyroptotic cell death and pro-inflammatory responses. Although NLR activation is generally thought to involve conformational rearrangements that oligomerize their N- or C-terminal death domains to nucleate ASC or pro-caspase-1 oligomerization, the molecular mechanisms that enable these transitions vary substantially among family members. Here, we highlight recent structural advances that reveal how distinct NLRs are activated and assemble into higher-order signaling platforms, providing mechanistic insights into the diverse activation mechanisms that converge on a common inflammatory outcome.

NLR family members are pattern-recognition receptors of the innate immune system. Here, the authors reviewed the NLR family members, their structural insights and signaling repertoire.

Introduction

Nucleotide-binding domain (NBD) and leucine-rich repeat (LRR)-containing (NLR) proteins constitute a fantastic family of intracellular proteins that play essential roles in diverse biological processes, ranging from immunoregulation to embryonic development¹. The first NLR was characterized over two decades ago as the human MHC (major histocompatibility complex) class II transactivator (CIITA)^2,3. Subsequently, 22 NLRs were identified in the human genome that share a conserved NBD-LRR architecture⁴. Among them, the inflammasome-forming NLRs represent a key subset of pattern recognition receptors (PRRs) that detect cytoplasmic pathogen-associated molecular patterns (PAMPs) or danger-associated molecular patterns (DAMPs). Upon activation, these NLRs promote caspase-1 activation and initiate pro-inflammatory responses. Throughout this review, we define the inflammasome platform as the molecular apparatus responsible for recruiting and activating caspase-1.

Structurally, most inflammasomal NLRs comprise a variable N-terminal death domain (DD), a middle NBD-containing ATPase (NACHT) domain, and a C-terminal LRR domain, with some members also possessing additional flanking C-terminal regions (Fig. 1A). The NACHT domain usually mediates ATP-dependent oligomerization for the formation of higher-order signaling complexes⁵, whereas the LRR domain is involved in ligand recognition and autoregulation⁶.

Fig. 1. Domain structures of NLRs and death domain interactions.

A Domain organizations of the representative inflammasomal NLR proteins, ASC adapter, and pro-caspase-1. B PYD-PYD and CARD-CARD mediated death domain interactions in inflammasome assembly. C The cryo-EM structure of full-length ASCb filament (PDB: 9WZ7) showing an inner core of PYD and an outer layer of CARD domains. One ASCb subunit is zoomed in and indicated by a dashed box. D Hypothetical ASC-dependent hNLRP1 inflammasome NLRP1^CARD filament:ASC^CARD-CASP1^CARD filament:CASP1-pro-IL-18 (PDB: 6XKK, 7KEU, 5FNA, 8SV1) upon activation.

The N-terminal DD of NLRs functions as an effector domain and mediates specific interactions with either the adapter protein apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC) or directly with pro-caspase-1, the inactive precursor of caspase-1 (Fig. 1A, B)⁷. Except for NAIP (NLR family apoptosis inhibitory protein), which contains three baculovirus inhibitor of apoptosis repeats (BIRs), the N-terminal DDs of most inflammasomal NLRs are either a pyrin domain (PYD) or caspase recruitment domain (CARD) (Fig. 1A). For example, NLR family CARD domain containing 4 (NLRC4) recruits both ASC and pro-caspase-1 through its N-terminal CARD^8–12, whereas NLR family pyrin domain containing 3 (NLRP3) and NLR family pyrin domain containing 6 (NLRP6) form inflammasomes via their N-terminal PYDs and the adapter ASC¹³. The NLR family pyrin domain containing 1 (NLRP1) and CARD containing protein 8 (CARD8) inflammasomes are both formed via their C-terminal CARDs (Fig. 1B)^14–18. The ASC-ASC and CARD-CARD interactions primarily occur through an induced polymerization mechanism, in which an upstream ASC or CARD oligomer functions as a seed to nucleate the filamentation of downstream ASC- or CARD- containing molecules (Fig. 1C, D)^19–21. This induced polymerization brings pro-caspase-1 into proximity, thereby promoting its activation and cleavage of pro-interleukin (IL)-1β, pro-IL-18, and Gasdermin D (GSDMD) (Fig. 1B, D).

Although inflammasomal NLRs converge on pro-caspase-1 activation and the induction of pyroptosis, the upstream signals they detect and their mechanisms of activation are highly distinct. For example, the NAIP/NLRC4 inflammasome senses effector proteins derived from Gram-negative bacteria²², whereas NLRP3 responds to a broad spectrum of cellular stress signals, including K⁺ efflux, extracellular ATP, mitochondrial DNA (mtDNA), and reactive oxygen species (ROS)^13,23. In this review, we discuss the currently characterized mechanisms of inflammasomal NLR activation from a structural perspective and propose directions for further understanding of inflammasome machineries in mammals.

Ligand-induced oligomerization of NAIP/NLRC4

Among the NLR inflammasomes, NAIP/NLRC4 is one of the best characterized NLR inflammasomes and has provided the first model of ligand-induced inflammasome activation. Notably, the system differs between mice and humans in the number of NAIP paralogs. In mice, multiple NAIP paralogs act as ligand-specific sensors that recognize distinct bacterial components, such as type III secretion system (T3SS) needle and inner rod proteins, as well as flagellin^22,24,25 (Fig. 2A). Upon ligand binding, these NAIPs recruit and activate NLRC4, leading to the assembly of the NAIP/NLRC4 inflammasome complex. In contrast, humans possess a single NAIP with broader ligand specificity. Although the structural basis of needle-induced human NAIP/NLRC4 activation closely resembles that of the mouse system²⁶, the structural mechanisms by which human NAIP recognizes other ligands remain unknown. This gap highlights the need for further investigation into the species-specific diversification of NAIP-mediated pathogen recognition^25,27,28.

Fig. 2. Ligand-induced NLRC4 oligomerization.

A Distinct bacterial ligands recognized by different NAIPs in NAIP/NLRC4 pathway. B Domain organization of NLRC4. C Crystal structure of inactive NLRC4. (PDB: 4KXF) D Domain organizations of NAIP and bacterial ligand FliC. E,F Superposition of the unliganded NAIP5 (gray, PDB: 7RAV) and liganded NAIP5 (color, PDB: 8FML). Close-up views of the nucleotide-binding pocket, WHD and HD2-LRR domain changes. G Superposition of the inactive NLRC4 (gray, PDB: 4KXF) and active NLRC4 (color, PDB: 3JBL). H A structure-derived model of NLRC4 inflammasome activation.

Both NAIP and NLRC4 are expressed at basal levels in resting cells, and it is essential that they remain inactive in the absence of ligand to prevent spontaneous inflammasome activation and autoinflammation. The structural basis of NLRC4 autoinhibition is revealed by a crystal structure capturing NLRC4 in a closed, monomeric conformation²⁹. In this inactive state, ADP is bound within the NACHT module and stabilizes a compact arrangement of its core subdomains, thereby locking NLRC4 in an assembly-incompetent configuration. This autoinhibited mechanism is demonstrated by the H433A mutation, which destabilizes ADP binding and results in constitutive activation (Fig. 2B, C)²⁹. In addition to this nucleotide-dependent restraint, peripheral regulatory domains further suppress oligomerization. For example, the helical domain 2 (HD2) packs against the NBD core and masks structural elements required for intermolecular association, whereas the C-terminal LRR domain lies along the lateral surface of the NACHT module and sterically occludes the protomer-protomer interface (Fig. 2C)²⁹.

Overcoming NLRC4 auto-inhibition is achieved by the ligand-bound NAIP. In contrast to the closed, autoinhibited conformation of NLRC4, cryo-EM studies show that unliganded NAIP5 (a murine NAIP that senses flagellin) adopts a wide-open, monomeric architecture³⁰ (Fig. 2D–F). In this resting state, the NBD-HD1 core remains relatively stable, whereas the HD2 and LRR domains are positioned away from the NACHT core and exhibit high flexibility. Notably, unliganded NAIP5 is bound to ATP, indicating that ATP binding alone is insufficient to drive NAIP into an active conformation³⁰.

Structural comparisons between unliganded and ligand-bound NAIP5 further demonstrate that NAIP5 activation is primarily driven by rearrangements within the winged-helix domain (WHD)-HD2-LRR module³⁰. Flagellin binds NAIP through two major interfaces: (1) the extreme C-terminal end of flagellin inserts into a hydrophobic pocket formed by the NTD, BIR1, and HD1 domains; and (2) extensive interactions between the D0 domain of flagellin and the HD2, ID, and LRR domains of NAIP (Fig. 2E)^30,31. Upon flagellin binding, the WHD undergoes a local conformational change, accompanied by coordinated repositioning of the HD2 and LRR domains, resulting in a more compact and ordered architecture (Fig. 2E, F)^10,12,31,32. These movements reorganize two surface loops within the WHD, which form part of the interface responsible for engaging NLRC4. In the ligand-bound structure, these WHD loops and the repositioned HD2-LRR region together generate the structural configuration required for productive interaction with NLRC4 and subsequent activation³¹. The functions of the NAIP BIR2 and BIR3 domains remain incompletely defined.

Following NAIP-mediated recruitment, NLRC4 undergoes a pronounced rigid-body conformational transition in which the WHD-HD2-LRR block rotates by ~87° relative to the NBD-HD1 core (Fig. 2G)^9,11. Structural comparison of inactive and activated NAIP/NLRC4 reveals that NAIP-mediated NLRC4 activation requires the interaction of two functionally distinct interfaces. One interface, termed the nucleation surface on NAIP, recruits a largely preserved receptor surface in inactive NLRC4, allowing NAIP5 to convert the first NLRC4 protomer into the active conformation^9,11. The role of the nucleotide in NLRC4 activation remains unresolved. Although ADP is essential for maintaining NLRC4 in its autoinhibited conformation²⁹, no nucleotide density was observed in the active NLRC4 structure¹¹. This absence may reflect limitations in the cryo-EM resolution or indicate that nucleotide release actively contributes to NLRC4 activation. Once the first NLRC4 protomer is activated, it no longer depends on NAIP5 for further assembly. Instead, the newly exposed nucleation surface of activated NLRC4 engages the receptor surface of a neighboring inactive NLRC4 and induces the same WHD-centered conformational rearrangement. This templated conversion generates another activated NLRC4 protomer with its own nucleation surface (Fig. 2H)⁹. Through repeated rounds of this receptor-nucleation surface interaction, NLRC4 molecules are recruited and converted one after another, driving the self-propagating growth of the disk-like inflammasome assembly, followed by caspase-1 oligomerization and activation (Fig. 2H). It is worth noting that one ligand-bound NAIP acts as the seed for NLRC4 polymerization and typically occupies only one position in the disk, rather than propagating as repeated NAIP subunits. This mechanism ensures high sensitivity of the inflammasome pathway, as a single ligand-bound NAIP can initiate the self-propagating recruitment of NLRC4.

Transition from inactive oligomers to active oligomers of NLRP3

Unlike the NAIP/NLRC4 inflammasome, the activation of NLRP3 is generally considered a two-step process, with an initial priming signal sensed by Toll-like receptors (TLRs) that increases the cellular abundance of NLRP3^33–36. Following priming, NLRP3 is activated by diverse forms of cellular stress such as ATP, potassium efflux, reactive oxygen species (ROS), mitochondrial DNA, and microbial toxins^37–39. Because a unifying direct ligand for NLRP3 has not been identified, it remains unclear how diverse upstream stimuli converge to activate NLRP3. However, recent cryo-EM studies have revealed a series of unexpected mechanisms of NLRP3 activation.

First, cryo-EM studies have fundamentally revised the view that inactive NLRP3 exists primarily as a monomer. Instead, full-length mouse and human NLRP3 assemble into large oligomeric “cage”-like structures composed of two stacked rings (Fig. 3A–C), with each ring containing five to eight subunits, yielding 12-16mer assemblies in the mouse protein and predominantly decameric assemblies in the human counterpart^40–42. These assemblies form barrel-shaped dodecameric complexes in which the NACHT domains remain spatially separated and inactive, while oligomerization is driven almost entirely by LRR-LRR interactions arranged in face-to-face and back-to-back interfaces (Fig. 3B). Disruption of these contacts destabilizes the assembly and promotes spontaneous activation, supporting the cage as a defined autoinhibitory configuration. Although the PYD domains are poorly resolved, weak densities within the central cavity suggest that they are sequestered inside the oligomeric shell, rendering the assemblies incapable of nucleating ASC-PYD filaments due to the narrow opening of the cage⁴¹.

Fig. 3. NLRP3 transition from inactive to active oligomers.

A Domain organizations of NLRP3 and NEK7. B Close-up views of the NLRP3 cage “face-to-face” and “back-to-back” interface. (PDB: 7LFH) C Close-up views of the NLRP3 cage inhibitor binding pocket. ADP and MCC950 molecules are shown as green and blue, respectively. (PDB: 7VTQ) D Left: Cryo-EM structure of NLRP3-NEK7 complex. Right: superposition of the NLRP3-NEK7 complex (PDB: 6NPY) onto one protomer in the NLRP3 dodecamer. E Cryo-EM structure of NLRP3 inflammasome. (PDB: 8EJ4) F Superposition of active NLRP3 (color) and inactive NLRP3 (PDB: 6NPY in gray). G A structure-derived model of NLRP3 inflammasome activation. The dotted line indicates that the mechanism has not yet been clarified.

A defining feature of the inactive NLRP3 cage is its close association with cellular membranes. Full-length NLRP3 is enriched in membrane fractions in mammalian cells, and purified oligomeric assemblies bind phosphorylated phosphoinositides through a conserved polybasic region within the PYD-NACHT linker. Disruption of this region weakens membrane association and markedly reduces formation of the double-ring cage, whereas restoration of membrane targeting rescues cage assembly, indicating that membrane binding actively promotes and stabilizes the autoinhibited oligomeric state^41,43.

Structural studies also reveal that this inactive architecture can be reinforced by nucleotide and small-molecule binding. In the presence of ADP and the NLRP3 inhibitor MCC950^44–46, both human and mouse NLRP3 adopt stable oligomeric assemblies in which the NACHT domain remains in a closed conformation. MCC950 occupies a cavity that is spatially distinct from the nucleotide-binding pocket. By engaging residues across multiple NACHT subdomains, it cooperates with ADP to stabilize the inhibited conformation, effectively acting as a conformational lock that maintains NLRP3 in its inactive assembly. (Fig. 3C)^45,47,48.

A critical regulator for NLRP3 activation is NIMA-related kinase 7 (NEK7), which was identified through genetic screening and biochemical purification as an essential and specific cofactor of the NLRP3 inflammasome^49,50. Cryo-EM structures of the NLRP3-NEK7 complex show that NEK7 binds NLRP3 through its C-terminal kinase lobe, docking onto a composite surface formed by the LRR domain and the HD2-NBD regions of the NACHT domain (Fig. 3D)^50,51. This interface is dominated by electrostatic and shape-complementary contacts, and its disruption abolishes inflammasome activation without affecting NEK7 kinase activity, indicating a structural adapter role. Importantly, the NEK7-binding surface overlaps with the LRR-LRR interfaces that stabilize the autoinhibited cage, rendering NEK7 binding sterically incompatible with the double-ring assembly and thereby promoting remodeling of the inactive oligomeric state (Fig. 3D)^41,51.

The cryo-EM structure of the active NLRP3 inflammasome further support a model in which NEK7-associated remodeling contributes to the transition of NLRP3 from the autoinhibited, membrane-associated cage to an activation-competent oligomeric platform (Fig. 3E)⁵². At the core of this transition, the NACHT domain undergoes a major conformational rearrangement characterized by coordinated rotation of WHD-HD2-LRR domains (Fig. 3F), exposing oligomerization surfaces that are inaccessible in the inactive state⁵². The prominent structural feature of this transition is the ordering of the Fish-Specific NACHT-Associated region (FISNA, AA 135-218). Whereas FISNA is partially disordered in inactive NLRP3, it becomes fully structured in the active complex and directly participates in inter-protomer contacts within the disk, thereby contributing to stabilization of the oligomeric assembly⁵². In parallel, the pyrin domains, which are sequestered within the cage in the resting state, become released and spatially reorganized. Their alignment provides a nucleation template for ASC PYD polymerization, initiating formation of filamentous adapter scaffolds.

Interestingly, the active NLRP3 inflammasome disk was captured with ATP-γ-S bound to the NACHT domain, consistent with the idea that ATP-binding stabilizes the active NLRP3 conformation. However, the nucleotide-dependent activation mechanism remains incompletely understood. Biochemical and cellular studies show that Walker A mutations that impair nucleotide binding strongly compromise NLRP3 activation. In contrast, mutations in or near Walker B region can reduce ATP hydrolysis and, in some disease-associated contexts, promote NLRP3 hyperactivity⁴⁸. These observations support a model in which ATP binding favors the active state, whereas ATP hydrolysis may contribute to resetting or restraining NLRP3 activity. However, whether ATP binding alone is sufficient to drive productive inflammasome assembly remains unresolved, and the temporal order of ADP-to-ATP exchange, inactive cage opening, NEK7 engagement, and active-disk assembly has not yet been directly measured. Further kinetic, biochemical, and intermediate-state structural studies will be needed to define how nucleotide exchange is coupled to NLRP3 cage disassembly and inflammasome assembly (Fig. 3G).

Proteolytic activation of NLRP1

NLRP1 was the first inflammasome-forming protein to be described⁵³. It exhibits extensive species-specific diversification, with one human NLRP1 gene encoding multiple isoforms and several rodent Nlrp1 paralogs or alleles. Human NLRP1 isoforms contain an N-terminal PYD followed by a disordered linker preceding the NACHT domain, whereas rodent NLRP1 proteins lack the N-terminal PYD and contain a significantly shorter linker before NACHT. Whether the PYD of human NLRP1 contributes to inflammasome assembly remains unclear. All NLRP1 isoforms harbor an auto-processing FIIND domain, which undergoes autoproteolytic cleavage to generate NLRP1-NT (N-terminal fragment) and NLRP1-CT (C-terminal fragment) (Fig. 4A). The FIIND consists of two structural subdomains, ZU5 (named after ZO-1 and UNC5) and UPA (shared by UNC5, PIDD, and ankyrins), and the cleavage site within the FIIND is located at the junction between the ZU5 and UPA subdomains. Following cleavage, the UPA β13 forms two antiparallel β-sheets with the cleaved ZU5, thereby preventing the release of UPA-CARD (Fig. 4B)⁵⁴.

Fig. 4. The activation and regulation of NLRP1 and CARD8 inflammasomes.

A Schematics of NLRP1 showing the proteolysis within the disordered region between PYD and NACHT domains, and autoproteolysis within the FIIND domain. ZU5 and UPA stay non-covalently attached. B Crystal structure of FIIND (PDB: 7CRV) showing intramolecular interaction of UPA-β13 with ZU5. C Cryo-EM structure of NLRP1^93-994-sfTRX (PDB: 7WGE) showing that oxidized TRX1 was associated with NACHT domain of hNLRP1 via a disulfide bond to stabilize its inhibited state. D The LRR domains of hNLRP1 senses viral dsRNA via direct interaction. E Phosphorylation, ubiquitination and proteasomal degradation happen between PYD-NACHT domains and within PYD, respectively, all destabilize NLRP1 and promote activation of the NLRP1 inflammatory pathway. F Domain organization of human DPP9. G Cryo-EM structure of the ternary NLRP1^A-NLRP1^B-DPP9 complex (PDB: 6X6A, left panel) showing the association of full-length FIINDs (UPA^A) and auto-cleaved (UPA^B) in the inactive state of NLRP1 mediated by DPP9. A close-up view of the N-terminal peptide of UPA^B into the DPP9 active site in DPP9-NLRP1 (right panel). H Cryo-EM structure of DPP9^VbP-NLRP1 showing assembly of autoprocessed and full-length FIINDs (PDB: 6X6C, left panel) and a close-up view of UPA N-terminus replaced by VbP in the DPP9 pocket (right panel). I Domain organization of human CARD8. J Cryo-EM structure of DPP9-CARD8 (PDB: 7JKQ) showing assembly of autoprocessed and full-length FIINDs similar to (G). K Close-up view of the DPP9 pocket in the inactive state of DPP9-CARD8 (PDB: 7JKQ) showing the absence of UPA. L Close-up view of DPP9 pocket taken by VbP in DPP9^VbP-CARD8 (PDB: 7JN7).

Multiple signals have been identified that regulate the NLRP1 inflammasome. For example, bacterial anthrax lethal factor (LF, a Zn²⁺-dependent metalloprotease that cleaves NACHT)⁵⁵ or viral proteases⁵⁶ cleave NLRP1 before the NACHT domain, generating a destabilized neo-N-terminus that is targeted for proteasomal degradation, thereby releasing the UPA-CARD fragment (Fig. 4A). NLRP1 can also sense changes in the cytosolic redox environment through direct interaction with oxidized thioredoxin-1(TRX1), which stabilizes NLRP1 and protects it from proteasomal degradation^57,58 (Fig. 4C). In addition, the LRR domain of hNLRP1 senses viral double-stranded RNA (dsRNA), although the molecular mechanism of this interaction needs further investigation (Fig. 4D)⁵⁹. Furthermore, the disordered linker region upstream of NACHT is subject to post-translational modifications. For instance, it can undergo ubiquitination under conditions such as exposure to LF^60–62, and phosphorylation via the ZAKα-p38 pathway under ribotoxic stress (Fig. 4E)^60–62, thereby activating the inflammasome.

Controlled release of NLRP1 and CARD8

In addition to the diverse regulatory mechanisms that activate NLRP1, dipeptidyl peptidases 9 and 8 (DPP9 and DPP8) function as key negative regulators (Fig. 4F). Structural and biochemical studies have demonstrated that DPP9 restrains spontaneous NLRP1 activation by sequestering NLRP1-CT in an inhibitory complex^15,54,63. Specifically, cryo-EM structures of the human DPP9-NLRP1 complex revealed a ternary inhibitory assembly comprising DPP9, one NLRP1-CT (UPA-CARD), and a second NLRP1 molecule resolved at the level of its FIIND (ZU5-UPA). Although only the FIIND region of the second NLRP1 is visualized in the EM density, the intact ZU5-UPA architecture indicates that this species represents full-length NLRP1, with the remaining N-terminal domains conformationally flexible¹⁵ (Fig. 4G, left panel). Importantly, DPP9 accommodates NLRP1-CT only in the presence of full-length FIIND, suggesting that inflammasome activation is governed by the cytosolic ratio of NLRP1-CT to NLRP1-FL.

Mechanistically, the unfolded N-terminus of NLRP1-CT inserts into the catalytic pocket of DPP9, forming hydrogen bonds and salt-bridge interactions (Fig. 4G, right panel). Binding of the DPP9 inhibitor, Val-boroPro (VbP) displaces this interaction by occupying the catalytic site, thereby releasing NLRP1-CT and promoting inflammasome activation (Fig. 4H)⁵⁴. Together, these structural insights establish DPP9 as a critical checkpoint in NLRP1 inflammasome activation.

Another inflammasome regulated by DDP9/8 is CARD8, which is a primate-specific inflammasome sensor of HIV-1 protease activity⁶⁴ that is primarily expressed in human lymphocytes⁶⁵, and is up-regulated in cancer cells⁶⁶. Structurally, CARD8 possesses a disordered N-terminal region, but its carboxy-terminus resembles NLRP1-CT with its FIIND and CARD (Fig. 4I). CARD8 undergoes similar autoproteolytic cleavage within FIIND¹⁷ and forms a stable inhibitory complex with DPP9⁶⁷. Unlike NLRP1-CT, CARD8-CT binds DPP9 outside its catalytic site (Fig. 4J, K) and is not directly displaced by VbP (Fig. 4L)⁶⁷. Nevertheless, VbP disrupts this complex in cells through a mechanism that remains unclear, and a larger DPP9 active site probe, a fluorophosphonate (FP)-biotin with a polyethylene glycol linker, weakens DPP9-CARD8 interaction in vitro⁶⁷. Despite the differences between NLRP1 and CARD8 activation, both sensors converge on a shared principle: DPP9 binding defines a molecular gatekeeper that must be overcome before inflammasome assembly proceeds.

Liquid-liquid phase separation of NLRP6

In addition to ordered oligomerization, liquid-liquid phase separation (LLPS) also promotes inflammasome assembly by increasing the local concentration of NLR proteins. A prominent example is NLRP6, originally identified as pyrin-containing Apaf-1-like protein 5 (PYPAF5)⁶⁸. NLRP6 is highly expressed in intestinal colonic myofibroblasts and epithelial cells^69,70 and exhibits the conserved PYD-NACHT-LRR domain organization characteristic of inflammasomal NLRs. NLRP6 has been proposed to sense a wide range of PAMPs and DAMPs, including microbial lipoteichoic acid (LTA)⁷¹, microbiota-derived metabolites such as taurine⁷², and viral dsRNA⁷³. Although cryo-EM structures of NLRP6 PYD filaments support its capability to form inflammasomes in a manner similar to other NLRs⁷⁴, the mechanism driving NLRP6 PYD oligomerization appears distinct. Notably, viral dsRNA can induce NLRP6 LLPS through a disordered poly-lysine sequence within the NACHT domain, generating dynamic, membraneless condensates that concentrate ASC via PYD-PYD interactions, thereby facilitating inflammasome assembly⁷⁵.

Future perspectives

The diversity of NLR inflammasome assembly highlights how evolution repurposes conserved structural modules to implement distinct activation mechanisms that converge on a shared inflammatory outcome. Here, we have discussed several of the better-characterized NLR inflammasomes; however, for other NLRs, such as NLRP10 and NLRP12, the signals they sense and the molecular basis of their activation remain unresolved⁷⁶. Even for the NLRs described above, the dynamic sequence of conformational transitions from autoinhibition to inflammasome assembly is not yet fully defined. Crosstalk among different inflammasome sensors represents exciting new frontiers in inflammasome research. However, structural information on the underlying molecular mechanisms remains limited. Addressing these gaps will require integrative structural, biochemical, and in situ approaches capable of capturing transient intermediates. Ultimately, structure-guided therapeutic strategies that selectively stabilize specific conformational states may enable precise modulation of inflammasome activity in inflammatory and autoinflammatory diseases.

A table of determined structures referred to in this review is listed in Supp. Table 1.

Author contributions

Y.L. and K.J. contributed equally to this work and drafted the manuscript. L.Z. supervised the work and critically revised the manuscript. All authors approved the final version.

Peer review

Peer review information

Communications Biology thanks all reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Si Ming Man, Kaliya Georgieva and Christina Karlsson Rosenthal.

Funding

This work was supported by the National Institutes of Health grant R01AI165580 (L.Z.).

Data availability

No new datasets were generated or analyzed during the preparation of this review.

Competing interests

The authors declare no conflict of interest.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s42003-026-11169-w.