Search

Nucleotide immune signal activates bacterial caspase by filament depolymerization

Xuzichao Li , Qiuqiu He , Bin Liu , Zhikun Liu , Qian Zhang , Xiaoshen Wang , Tongyu Mao , Jie Han , Shuqin Zhang , Fumeng Liao , Yingcan Liu , Zhengda Shan , Zhenning Yin , Yong Wei , Zhiyong Yuan , Hang Yin , Heng Zhang

Vita ››

Vita > Correspondence > DOI: 10.15302/vita.2026.10.0084
Vita Open AccessPublished:

Nucleotide immune signal activates bacterial caspase by filament depolymerization

Author information +
History +
Vita () Cite this article
PDF (2402KB)

登录浏览全文

4963

注册一个新账户 忘记密码

Dear Editor,
Nucleotide immune signaling pathways rely on enzymatically generated second messengers to activate downstream effectors1. In bacterial type IV Thoeris systems, TIR enzymes convert NAD+ into 5′-diphosphate-cyclo[N7:1′′]-ribose (N7-cADPR), which activates the caspase-like protease to confer population-level immunity during phage infection2. In striking contrast to human caspases, whose activation depends on proteolytic maturation, type IV Thoeris caspases are controlled by a nucleotide second messenger-mediated mechanism2. However, the molecular basis of N7-cADPR-dependent TIR-caspase activation in antiphage defense remains unknown. Here, we biochemically and structurally characterize the type IV Thoeris system from Escherichia coli. Our study reveals how nucleotide immune signal activates bacterial caspase through filament depolymerization, expanding our understanding of nucleotide immune pathways and prokaryotic defense strategies.
Type IV Thoeris locus from Escherichia coli encodes two protein components, including a TIR-domain protein (EcTIR: aa 1–133) and a caspase-like protease (EcCaspase: aa 1–321) (Fig. 1a). The protease harbors an N-terminal caspase-like domain (NTD: aa 1–231) that is naturally fused to a C-terminal domain of unknown function (CTD: aa 232–321). We first cloned the complete type IV Thoeris operon under a pBAD promoter and assessed its antiviral activity against a panel of phages in Escherichia coli. This system conferred strong resistance to various types of phages, with the strongest resistance against the T6 phage (Fig. 1b). RT-qPCR analysis revealed that EcCaspase expression was upregulated upon phage infection (Supplementary Fig. S1). Catalytic mutants in either EcTIR (E84Q) or the EcCaspase (C132A) abolished protection, demonstrating that the activities of both enzymes are critical for anti-phage defense (Fig. 1c). Consistent with previous reports that the EcCaspase exhibits proteolysis activity2, our fluorescence-based peptide assays confirmed this activity, supporting the notion that EcCaspase-mediated proteolytic cleavage underlies antiphage defense (Fig. 1d, e; Supplementary Fig. S2). Strikingly, negative-stain analyses of both MBP-fused and tag-free purified wild-type (WT) EcCaspase revealed the formation of long, continuous filaments ranging from several tens to hundreds of nanometers in length (Supplementary Fig. S3a, b). Moreover, EcCaspase readily forms filaments even at low concentrations (Supplementary Fig. S3c), whereas higher concentrations markedly increase filament length, consistent with a concentration-regulated assembly process. In contrast, these assemblies dissociate into low-order oligomers in the presence of N7-cADPR, regardless of the MBP-fused or tag-free EcCaspase (Supplementary Fig. S3a, b). Consistently, native PAGE analysis of GFP-tagged EcCaspase in E. coli showed that phage infection shifted EcCaspase from high-molecular-weight assemblies retained on the top of the gel to migrating species, suggesting an assembly–to–disassembly transition in vivo (Supplementary Fig. S3d). Thus, it seems that the type IV Thoeris EcCaspase follows a ligand-triggered depolymerization-activation mechanism, switching from a filamentous, likely inhibitory state to an active state upon N7-cADPR binding.
To elucidate the molecular basis of filament formation, we performed cryo-EM analysis of both MBP-fused and tag-free EcCaspase in the absence of N7-cADPR (Supplementary Figs. S4, S5). Both structures exhibit highly similar overall architectures and protomer conformations (Supplementary Fig. S5e). Due to the well-defined density, we focused subsequent structural analyses on the MBP-fused EcCaspase structure unless otherwise noted. The filament displays a pseudo-C4 point-group symmetry: the CTDs stack to form a compact central core, while the NTDs wrap helically around this core, resembling a spiraling dragon coiling around a pillar (Fig. 1f, g; Supplementary Fig. S6a). EcCaspase protomer adopts a bilobed, dumbbell-like architecture comprising an NTD part and a CTD region, likely to mediate ligand engagement (Supplementary Fig. S6b–d). In the filament structure, two EcCaspase protomers constitute a dimeric asymmetric unit (ASU) (Fig. 1h). These dimer units assemble into a right-handed helix with a rotation of 90° per ASU and an axial rise of 22.9 Å, yielding a helical pitch of ~91.6 Å per turn (Fig. 1f). This arrangement produces a well-ordered, spine-and-sheath-like organization of the filament. Within the dimeric asymmetric unit, the two EcCaspase protomers (termed A1 and A2) associate into a domain-swapped dimer, forming three intra-dimer interfaces with a combined buried surface area of ~2,100 Å2 (Supplementary Fig. S7a–c). The EcCaspase dimers pack in a tightly arranged, side-by-side configuration to assemble into the filament. Both the NTDs and CTDs contribute to filament formation, generating two major interfaces (NTD–NTD and CTD–CTD interfaces) that together bury ~1,700 Å2 of surface area (Supplementary Fig. S7d, e). Mutations disrupting these interfaces prevented filament formation, markedly reduced N7-cADPR binding, abolished EcCaspase-mediated peptide cleavage, and completely eliminated type IV Thoeris-mediated antiphage defense, implying that filament assembly is essential for the antiviral function of the system (Fig. 1e, i; Supplementary Figs. S3e, S8).
We next determined the cryo-EM structure of MBP-fused EcCaspase bound to N7-cADPR (Supplementary Fig. S9). The two-dimensional class averages revealed oligomeric particles, distinct from the supramolecular assembly observed in the ligand-free sample. The majority of particles (~60%) represent a dimeric species, whereas two additional classes corresponding to tetrameric and hexameric assemblies (each ~20%) were also observed (Fig. 1j; Supplementary Fig. S9c). However, the dimeric units among the three oligomeric states were essentially identical (Fig. 1k). It is plausible that the tetrameric and hexameric species likely represent intermediates upon N7-cADPR binding. In the N7-cADPR-bound dimeric unit, an N7-cADPR molecule nestles into an inter-protomer pocket created by the two CTDs (Fig. 1l; Supplementary Fig. S10a, b). The ligand is recognized through extensive π-stacking, hydrogen-bonding, and electrostatic interactions. Mutations disrupting the N7-cADPR-binding pocket markedly reduced EcCaspase-mediated peptide cleavage and completely abolished antiphage activity, demonstrating that N7-cADPR recognition by the CTD pocket is indispensable for EcCaspase-dependent peptide degradation and the antiviral defense of the type IV Thoeris system (Fig. 1e, i).
While the overall structure of the dimeric EcCaspase unit in the absence of N7-cADPR resembles that of the N7-cADPR-bound dimer, the CTDs rotate outward by approximately 7 Å to accommodate N7-cADPR (Fig. 1m). Remarkably, structural superposition of the N7-cADPR-bound dimer onto the dimeric unit in the context of filamentous form reveals that N7-cADPR binding induces an outward displacement of the CTD from the A1 protomer toward the CTD from the C2 protomer, thereby leading to a steric clash between the two CTDs (Fig. 1n, o). At the same time, the CTD of A1 moves away from the CTD of the neighboring B1 protomer, disrupting the original CTD–CTD interface, including the interaction between Asp289 and Leu235, which has been shown to be critical for filament formation (Fig. 1o; Supplementary Fig. S7e). Similarly, the CTD–CTD interface is also impaired in the ligand-bound tetrameric and hexameric assemblies, with only partial NTD-mediated contacts maintaining these intermediate states (Supplementary Fig. S10c–f). Notably, Tyr295 in the CTD of A1 reorients forms a polar contact with Arg239 in the CTD of B1 to directly bind with N7-cADPR, which stabilizes its binding (Fig. 1l, o). Together, it can be concluded that N7-cADPR binding not only imposes steric hindrances that interfere with the filament formation, but also weakens the assembly interactions. These ligand-initiated conformational rearrangements may propagate throughout the filament structure and finally be coupled with the structural destabilization and subsequent depolymerization.
To test whether the CTD-mediated clash can act as a trigger for activation, we introduced the D267R/E268R double mutation to mimic the steric conflict (Supplementary Fig. S11a). The mutant exhibited almost no detectable substrate cleavage activity even in the presence of N7-cADPR (Fig. 1e), suggesting that the CTD conformational change is unlikely to serve as the primary trigger for protease activation. Instead, it more likely reflects a ligand-induced structural rearrangement within the assembled state, with ligand binding acting as the principal activation cue. Alternatively, full allosteric activation of the NTD may require both ligand engagement and disassembly.
In human caspases, four conserved loops (L1–L4) surrounding the catalytic pocket undergo rearrangement upon cleavage to form a catalytically competent enzyme3,4. Structural comparison revealed that EcCaspase contains four corresponding loops, prompting us to compare their conformations between the ligand-free (auto-inhibited) and N7-cADPR-bound (active) states (Fig. 1p; Supplementary Fig. S11b). The catalytic dyad adopts a conformation reminiscent of that in human caspase and is not altered upon N7-cADPR binding. Likewise, the L1, L2, and L4 loops also exhibit slight shifts in response to N7-cADPR binding, contrasting with the substantial conformation changes in human caspase. However, the L3 loop undergoes a pronounced order-to-disorder transition. In particular, L3 folds into a short helix that packs over the catalytic dyad in the filament structure. In comparison with the inhibitor/substrate-bound human caspase structure (Supplementary Fig. S11c), the helical L3 in EcCaspase sterically clashes with the inhibitor/substrate. Thus, it appears that the L3 region of EcCaspase occupies the inhibitor/substrate-binding site, thereby blocking substrate access and keeping EcCaspase in an auto-inhibited state. Upon activation by N7-cADPR, however, the L3 region becomes highly flexible (Fig. 1p), allowing protein or peptide substrates to enter the catalytic pocket. Consistently, L3 deletion did not affect filament assembly or N7-cADPR binding but abolished N7-cADPR-dependent peptide cleavage and impaired antiphage defense (Fig. 1e, i; Supplementary Figs. S3e, S8). These observations indicate that the L3 region may function as a gating element that regulates the activation of EcCaspase.
To identify phage activators, we performed a Strep pull-down assay followed by LC-MS/MS using E. coli expressing Strep-tagged EcTIR after T6 infection (Supplementary Fig. S12a). The T6 major head protein (Mhp, a capsid protein) was identified as the most enriched EcTIR-associated factor (Supplementary Fig. S12b). The EcTIR-Mhp association was further validated by the in vivo co-immunoprecipitation assay, supporting Mhp as a strong candidate for a phage-derived activator of the type IV Thoeris system (Fig. 1q). Moreover, we also isolated T6 escape mutants on cells expressing the type IV Thoeris system and sequenced their genomes to identify the functional activators (Supplementary Fig. S13). Strikingly, all four independent escape isolates carried nonsense mutations in the mhp gene that result in premature termination of the major head protein. These data suggest that mutations in Mhp allow phage to bypass the Thoeris activation, providing genetic evidence that the major capsid protein is the bona fide trigger of this TIR-caspase defense pathway. Together with previous studies of type I Thoeris5, these findings support a conserved “capsid-sensing” strategy for TIR activation across Thoeris systems. To further evaluate whether Mhp indeed activates this system, we co-expressed Mhp together with the Thoeris operon in E. coli (Fig. 1r). Whereas expression of either component alone was not toxic, co-expression of Mhp with the Thoeris system resulted in pronounced growth inhibition. Importantly, this toxicity was reduced when the catalytic residues of either EcTIR or EcCaspase were mutated, demonstrating that Mhp-induced toxicity is dependent on EcTIR-catalyzed N7-cADPR synthesis and subsequent EcCaspase activation.
Unlike human inflammatory and initiator caspases, which are activated through higher-order assembly6-8, EcCaspase adopts an alternative activation mechanism involving ligand-triggered assembly remodeling. Our findings suggest that the assembled filament may pre-organize neighboring CTDs into a composite pocket to enable N7-cADPR-binding, priming the complex for ligand-triggered disassembly and activation. By contrast, the isolated CTD-containing protein remains monomeric, likely because of weak intrinsic CTD–CTD interactions (Supplementary Figs. S7b, S14). Binding of a single N7-cADPR molecule at the CTD dimer interface triggers the rearrangement of the CTDs for filament depolymerization. Thus, filament formation is crucial for positioning the CTDs to accommodate N7-cADPR. Ligand binding may promote filament disassembly, releasing conformational constraints imposed by the filament architecture and enabling the structural flexibility and active-site remodeling required for protease activation. The resulting lower-order assemblies may further provide increased conformational adaptability for substrate engagement within the crowded bacterial cytoplasm.
An L3-helical element gates the catalytic core for inhibition in the EcCaspase filament. Ligand-triggered depolymerization displaces L3 from the catalytic dyad and opens the catalytic pocket for substrate access, thereby activating the protease activity. Although L3 destabilization may be partly coupled to filament disassembly, disassembled oligomers alone are insufficient to support proteolytic activity (Fig. 1e), indicating that filament disassembly is not the sole determinant of L3 displacement. The precise allosteric mechanism linking ligand binding to L3 destabilization therefore remains to be fully elucidated. Assembly-dependent activation has been described for bacterial immune proteases such as the CRISPR-associated Lon-SAVED protease CalpL9,10, whereas filament depolymerization activation remains largely unexplored. Notably, the filament depolymerization activation mechanism described here has so far been established only for EcCaspase, and whether it is conserved across other type IV Thoeris systems remains unknown. Together with the recently characterized Pseudomonas caspase, which undergoes ligand-induced filament remodeling11, our findings suggest that Thoeris systems employ diverse ligand-driven assembly transitions to regulate protease activation. Although native PAGE supports phage-induced EcCaspase disassembly (Supplementary Fig. S3d), direct visualization of this transition in vivo remains challenging and will require future in situ approaches such as cryo-ET. Together, our findings uncover a ligand-triggered depolymerization-activation mechanism for a bacterial caspase-like protease and establish type IV Thoeris as a prototypical nucleotide immune signaling pathway that couples TIR-generated small molecules to proteolytic effector activation (Fig. 1s).

DATA AVAILABILITY

The atomic coordinates have been deposited in the Protein Data Bank under accession codes 9X0G (N7-cADPR-bound EcCaspase dimer), 9X0T (N7-cADPR-bound EcCaspase tetramer), 9X0U (N7-cADPR-bound EcCaspase hexamer), 9X0S (MBP-fused EcCaspase filament) and 27XF (EcCaspase filament). Cryo-EM maps have been deposited in the Electron Microscopy Data Bank under corresponding accession codes EMD-66434, EMD-66442, EMD-66443, EMD-66441 and EMD-81533.

[1]

Hobbs, S.J. & Kranzusch, P.J. Annu. Rev. Microbiol. 78, 255–276 (2024).

[2]

Rousset, F. et al. Science 387, 510–516 (2025).

[3]

Shi, Y.G. Cell 117, 855–858 (2004).

[4]

Riedl, S.J. et al. Proc. Natl. Acad. Sci. USA 98, 14790–14795 (2001).

[5]

Roberts, C.G. et al. Nat. Microbiol. 10, 2892–2902 (2025).

[6]

Park, H.H. Int. J. Mol. Med. 43, 1119–1127 (2019).

[7]

Wein, T. et al. Nature 639, 727–734 (2025).

[8]

Park, H.H. Cell Death Differ. 26, 1213–1220 (2019).

[9]

Huang, P. et al. Nat. Chem. Biol. (2026). DOI: 10.1038/s41589-026-02252-8.

[10]

Smalakyte, D., Ruksenaite, A., Sasnauskas, G., Tamulaitiene, G. & Tamulaitis, G. Mol. Cell 84, 4239–4255.e8 (2024).

[11]

Wei, L. et al. Vita (2026). DOI: 10.15302/vita.2026.06.0047.

RIGHTS & PERMISSIONS

The Author(s) 2026. Published by Higher Education Press. This is an Open Access article distributed under the terms of the CC BY license (https://creativecommons.org/licenses/by/4.0/).

Cite this article

Download citation ▾
Li, X. et al.  Nucleotide immune signal activates bacterial caspase by filament depolymerization  Vita https://doi.org/10.15302/vita.2026.10.0084 ()
AI Summary AI Mindmap
PDF (2402KB)

Supplementary Files

Supplementary Materials

Sections
Figures
References

706

Accesses

0

Citation

Detail

Recommended

AI思维导图

/

〈 〉