Dynamic balance of CRISPR-Cas immunity and resistance plasmid anti-immunity mediated by a bifunctional protein AcrIE10

Abstract

Despite targeting by CRISPR-Cas system, antimicrobial resistance plasmids are prevalent in clinical isolates of carbapenem-resistant Klebsiella pneumoniae which represent a major public health threat. A stable co-existence of plasmids and CRISPR-Cas systems is mediated by anti-CRISPR (Acr) proteins. Here, we report that previously identified AcrIE10 encoded by a resistance plasmid combines two functions: it inhibits CRISPR immunity by directly binding Cas7* subunit through its Acr domain, and acts as an Acr-associated (Aca) protein that self-represses the transcription of Acr locus. AcrIE10 is an example of an Aca protein that utilizes N-terminal ribbon-helix-helix (RHH) domain to specifically recognize the inverted repeat (IR) region in its own promoter. Crucially, a dimerization of AcrIE10 dimers is required for the effective binding to the IR and self-repression, while stoichiometry-dependent interaction with Cas7* facilitates transition to de-repressed state. These findings elucidate molecular mechanisms by which AcrIE10 operates as a dual functionAcr-Aca protein to achieve a delicate balance between host CRISPR-Cas immunity and plasmid anti-defense.

The transmission of plasmids represents a key mechanism for antimicrobial resistance. Here, authors show AcrIE10 inhibits bacterial CRISPR immunity and dynamically regulates self-expression, balancing host immunity and plasmid anti-defense.

Introduction

Bacteria and their competitors, including bacteriophages, plasmids, and other mobile genetic elements (MGEs), are engaged in a perpetual arms race, forcing their co-evolution¹. Bacteria employ diverse defensive strategies that limit the spread of MGEs to lower metabolic burden associated with their replication. However, often co-existence with MGEs provides evolutionary benefits, such as acquisition of antibiotic resistance genes². CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats)-Cas (CRISPR-associated genes) adaptive immunity is one of the most effective strategies bacteria have developed to combat MGEs³. As a drawback, the barrier to the horizontal gene transfer imposed by CRISPR-Cas systems may also limit the evolutionary potential of bacterial population^4,5.

Carbapenem-resistant Klebsiella pneumoniae (CRKP), recognized as one of the highest-priority pathogens by the World Health Organization, is an outstanding example of bacterial evolution⁶. CRKP carbapenemase genes (predominantly bla_KPC) are usually carried by conjugative plasmids, which promote their horizontal transfer and contribute to the poor outcome of clinical infections⁷. The spread of bla_KPC plasmids in natural populations of K. pneumoniae could be countered by the type I-E CRISPR-Cas system^8,9. Notably, sequence type (ST) 15 has become an emerging clone of carbapenem-resistant K. pneumoniae in clinical settings, despite carrying Type I-E* CRISPR-Cas (a subtype of Type I-E CRISPR-Cas) with spacers against bla_KPC plasmids. This indicates that the CRISPR-Cas system is actively silenced by the bacterial host or suppressed by the plasmid invader, which leads to the spread of bla_KPC resistance genes^9,10. This phenomenon reveals a critical trade-off in the bacteria-plasmid co-evolution under antibiotic selection pressure.

Anti-CRISPR (Acr) proteins are natural inhibitors of bacterial immune systems, encoded by phages and other MGEs^11–13. acr genes often form anti-defense clusters guarding MGE against diverse CRISPR-Cas systems¹⁴, and these operons are controlled by conserved anti-CRISPR associated (Aca) proteins. To date, 13 types of Aca have been predicted, although only 10 have been experimentally verified^15,16. All of them contain a helix-turn-helix (HTH) domain that binds to the inverted repeats in the acr promoter and functions as a transcriptional repressor, although Aca2 also regulates translation through binding to the additional repeat in the mRNA^17–19. The model of Aca-mediated regulation suggests a spike-in of Acr expression at the early stages of MGE infection, required to shut down pre-existing CRISPR-Cas complexes, followed by a repression of Acr synthesis²⁰. Overproduction of Acrs could have detrimental effects on the MGE replication^19,21, although the universality of this model has not been established. Negative-feedback regulation with Aca repressor eliminates toxic effects of Acr expression while supporting CRISPR-Cas inhibition at the onset of infection.

In contrast to the typical standalone HTH-domain Acas, some acr genes encode fusions that integrate both Acr and Aca components, with a limited number of bifunctional Acr-Aca proteins described so far^22–26. In a previous study, we identified the acrIE9.2 gene encoded by Type I-E* CRISPR-Cas targeted plasmids, which is a potential factor leading to the dissemination of bla_KPC plasmids in ST15 K. pneumoniae⁹. AcrIE9.2 was found in 82.5% (495 of 600) of ST15 strains encoding CRISPR-Cas systems, and in the majority of loci, co-occurs with the AcrIE10 protein. Recently, Jiang et al demonstrated that AcrIE10 inhibits the I-E* CRISPR-Cas system by specifically binding to Cas7 to prevent its hexamerization and preventing the surveillance complex assembly and crRNA loading²⁷. While searching for the potential Aca regulation of this widespread loci, we noticed that AcrIE10 encodes an N-terminal ribbon-helix-helix (RHH) domain, which is involved in DNA binding in diverse transcriptional factors²⁸. This raised a plausible hypothesis about bifunctional Acr-Aca activity of the AcrIE10 protein and the role of the RHH domain in acr transcriptional silencing.

Combining cryo-electron microscopy, X-ray crystallography, in vivo CRISPR interference experiments, and in vitro biochemistry, we elucidated the bifunctional behavior of AcrIE10 and revealed the dynamic nature of acrIE9.2-acrIE10 loci regulation, allowing MGEs to match Acr expression to CRISPR-Cas abundance. Via its RHH domain, AcrIE10 recognizes inverted repeats in the promoter region and acquires a dimer-of-dimers architecture in complex with DNA. AcrIE10 Acr domain is responsible for Cas7* inhibition, yet is also involved in bridging two AcrIE10 dimers, which is essential for DNA binding. This competition for the same binding interface allows for modulating protein interaction with DNA in the presence of a Type I–E* CRISPR-Cas system. An AcrIE10:Cas7*:DNA complex is stable when only one Cas7* subunit is bound, but loses interaction to DNA upon binding of more Cas7* subunits. Rapid production of AcrIE10 shifts the stoichiometry of the AcrIE10-Cas7* complex to a 2:1 ratio, re-establishing transcriptional repression. This reversible conformational reprogramming, driven by a Cas7* concentration, reveals how Acr expression can be finetuned to achieve a stable co-existence of antibiotic resistance plasmids with the host CRISPR-Cas immunity in clinical isolates of K. pneumoniae.

Results

AcrIE10 functions as an Aca protein

We previously identified that AcrIE9.2 of bla_KPC plasmids is responsible for the suppression of the Type I-E* CRISPR-Cas system in K. pneumoniae. However, the same acr locus is encoding a recently identified AcrIE10 protein²⁸ (Fig. 1a). Gene neighborhood analysis tool webFlaGs was used to explore the genetic environment of acr locus revealing that acrIE9.2 and acrIE10 form a highly abundant gene pair (Fig. 1b). To investigate the association between the acrIE10 locus and plasmid transfer, we conducted a BLAST search against a public database containing 9358 Enterobacteriaceae genomes and 88 clinical isolated ST15 K. pneumoniae (Supplementary Data 1). The results revealed that the acrIE10 was detected in 13.7% (1,277/9,358) of Enterobacteriaceae genomes, with a notably high prevalence in ST15 K. pneumoniae (94.3%, 83/88) (Fig. 1c). Remarkably, over 97.4% of acrIE10 genes were located on plasmids (Fig. 1d). Further analysis revealed that 29.8% (380/1277) of the acrIE10 carrying strains had CRISPR-Cas systems, of which 41.3% (157/380) strains harbored CRISPR-targeted plasmids (Supplementary Data 2).

Fig. 1. AcrIE10 functions as an Aca protein.

a Schematic of AcrIE10 and AcrIE9.2 reported Acr function. b Schematic of AcrIE10 neighborhood structure. Gene neighborhood analysis tool webFlaGs was used to explore the genetic environment of the Acr locus, resulting in the finding that acrIE9.2 and acrIE10 form a highly abundant genetic module. A conserved Hok family toxin, belonging to the type I toxin-antitoxin system, located upstream of acrIE10 locus while a hypothetical protein with unknown function located downstream of locus among some clusters. c The distribution of AcrIE10 in 9357 Enterobacteriaceae isolates and 88 ST15 clinical isolated K. pneumoniae. d The location of AcrIE10 in public available Enterobacteriaceae isolates. e K. pneumonia I–E* CRISPR-Cas system promotes targeted plasmid loss, which can be counteracted with acrIE9.2-acrIE10 expression. Plasmid retention rate after serial passages is indicated. EV - Empty Vector; pRSF-protospacer represents plasmid carrying I–E*-targeted protospacer with an optimal AAT protospacer adjacent motif (PAM); acr genes were expressed from I–E*-targeted pRSF vector using their native promoters, while I-E* CRISPR-Cas system and CRISPR cassette encoding targeting spacer were expressed from pHSG398 using native promoters. All data are presented as mean ± SD of three biological replicates. f, g Relative expression of gene acrIE9.2 in clinical isolate R41, in the presence of differing levels of AcrIE10:wild type and its acrIE10 knock out strain in R41-ΔAcrIE10, empty vector (EV), Apr-pHSG398 vector for overexpression of acrIE10. Data are shown as the mean relative acrIE9.2 expression of three biological replicates ± SEM (error bars). Data were analyzed using a two-sided Welch’s t-test. No adjustment was made for multiple comparisons. p-values:1f p = 0.0001, 1g p < 0.0001. h AcrIE10 suppresses eGFP expression with acrIE9.2 promoter: 1. pUC plasmid with inducible lac promoter; 2. pUC plasmid without lac promoter; 3. pUC plasmid with the acrIE9.2 promoter; 4. pUC plasmid with acrIE9.2 promoter, in the presence of an empty pACYC plasmid; 5. pUC plasmid with the acrIE9.2 promoter, in conditions of AcrIE10 expression from pACYC plasmid. The schematic diagrams of the construction for each BL21 model strain are shown.

To further dissect the role of this locus in the coexistence of CRISPR-targeted carbapenem-resistant plasmids with the resident CRISPR-Cas system, we transferred the I–E*CRISPR-Cas system of K. pneumoniae with native CRISPR array in Escherichia coli BL21 (DE3), lacking its native cas genes, and challenged the system with a pRSF plasmid carrying a targeted protospacer. The results showed that while the protospacer-bearing plasmid was eliminated by the I–E* CRISPR-Cas system, the acrIE9.2-acrIE10 locus contributed to its stabilization, confirming its anti-CRISPR function (Fig. 1e). Plasmid stability in the presence of AcrIE9.2 alone is similar to that observed when both AcrIE9.2 and AcrIE10 are co-expressed, indicating that AcrIE9.2 acts as the primary anti-CRISPR protein within this acr operon.

To study how the acrIE9.2-acrIE10 locus is regulated, we first verified that acrIE9.2 and acrIE10 genes are co-transcribed (Supplementary Fig. 1a), and thus used acrIE9.2 expression levels as a readout of the whole locus transcription in the downstream experiments. In the native background of K. pneumoniae R41 clinical isolate, acrIE10 knockout resulted in approximately 80-fold upregulation of the acrIE9.2 transcription (Fig. 1f) with no significant changes in plasmid copy number in the mutant strain (Supplementary Fig. 1b). Conversely, overexpression of acrIE10 strongly suppressed the transcription of the acrIE9.2-acrIE10 in ST15 K. pneumoniae R41 (Fig. 1g), suggesting that AcrIE10 represents an Aca transcriptional repressor.

To directly validate this effect, we cloned the eGFP reporter gene downstream of the acrIE9.2-acrIE10 promoter sequence and monitored fluorescence in the presence or absence of AcrIE10. While eGFP was actively produced in the control culture, AcrIE10 expression resulted in a fluorescence intensity matching that of a promoterless construct, confirming that AcrIE10 functions as an effective transcriptional repressor of the acrIE9.2-acrIE10 locus (Fig. 1h). Phylogenetic analysis revealed the evolutionary conservation of AcrIE10 (Supplementary Fig.1c, d), which motivated further investigation.

Structure of AcrIE10 in the apo- and DNA-bound states

To understand the basis of AcrIE10 repression activity we resolved the crystal structure of the apo AcrIE10 protein with 2.39 Å resolution (Supplementary Table 1), and reveal that AcrIE10 forms a dimer and is composed of a Acr domain made up of five short helices and an N-terminal RHH domain typical for transcription factors²⁸(Fig. 2a). Analysis of the acrIE9.2-acrIE10 promoter region identified an inverted repeat (IR) sequence overlapping with −10 and −35 RNA-polymerase binding sites and we hypothesized that these elements interact with AcrIE10 RHH domain. We synthesized a 50-bp double-stranded DNA probe, matching the promoter region, and using EMSA demonstrated that AcrIE10 site-specifically binds to the promoter DNA(Supplementary Fig. 2a), whereas mutations in the IR sequences completely abolished the DNA-binding capacity of AcrIE10, as no shifted bands were observed in EMSA even at the highest protein concentration (Fig. 2b, c).

Fig. 2. Structure of AcrIE10-IR complex.

a The crystal structure of apo AcrIE10. The N-terminal RHH domain is colored yellow, and the C-terminal Acr domain is colored light green. The AcrIE10 is in a dimeric form, with one monomer colored according to the above color scheme, and the other monomer shown in silver. The electrostatic surface potential distribution of one monomer is generated using PyMOL. Red represents negative charges, and blue represents positive charges. b Sequence of the acrIE9.2 promoter DNA region with inverted repeats, −10 and −35 promoter elements. A mutated variant of IR sequences is shown. c EMSA results demonstrating AcrIE10 binding to the acrIE9.2-acrIE10 promoter DNA and the lack of interaction with the disrupted-IR promoter. DNA (WT and Mutant) binding was analyzed by EMSA using 0.1 μM DNA and AcrIE10 at increasing concentrations of 0, 0.001, 0.002, 0.004, 0.008, 0.1, 0.2, 0.4, 0.8, and 0.8 μM. The right-most lane in each gel contains only the highest concentration of AcrIE10 (0.8 μM) without DNA, serving as a negative control to ensure the purified protein is free of nucleic acid contamination. d The crystal structure of AcrIF10-DNA complex is shown in the cartoon model, together with its DNA sequence. The N-terminal RHH domain and C-terminal Acr domain are colored in cyan and red, respectively. The 2|Fo|–|Fc| σ-weighted map is contoured at 1.0 σ. e The refined model of the AcrIE10-DNA complex structure is shown in the cartoon model. The structure within each asymmetric unit is delineated by a black dashed box; the RHH domain and Acr d are colored in different colors, respectively. f Details of AcrIE10’s specific recognition of DNA, with amino acid residues involved in recognition and their corresponding base interactions labeled. Hydrogen bonds are represented by gray lines. g Details view of IR DNA bound by AcrIE10 dimer-dimer. The 2|Fo|–|Fc| σ-weighted map is contoured at 1.0 σ. h Schematic representation of the polar interactions between the three amino acid side chains of each β-strand of an RHH2 and the nucleotide bases within a DNA operator sub-site, based on the structures of AcrIE10-IR complexes. i The dimer-of-dimers interface of AcrIE10 is shown in the surface model, consistent with the colors in (e). j Impact of IR DNA spacing on AcrIE10 dimer-of-dimers assembly. EMSA analysis showing the binding of AcrIE10 to IR DNA substrates with varying spacing lengths. Reactions contained 0.1 μM DNA and increasing concentrations of AcrIE10 (0, 0.01, 0.05, 0.1, 0.2, and 0.4 μM). The results demonstrate that a specific spacing length is required to facilitate the higher-order dimer-of-dimers assembly. In the summary table, “Yes” indicates the presence of an observable shifted band, while “No” indicates no detectable complex formation, highlighting the structural constraints of the AcrIE10-DNA interaction. dimer-of-dimers.

To unravel the molecular mechanism underlying AcrIE10-mediated transcriptional suppression, we conducted X-ray crystallographic studies on stable complexes formed by incubating AcrIE10 with double-stranded DNA (dsDNA) fragments containing inverted repeat (IR) sequences (Supplementary Table 1). We successfully obtained crystals of AcrIE10 bound to a 20-bp dsDNA with a 5’-overhang of A nucleotide through optimized crystallization conditions. Using an AlphaFold3-predicted model as the molecular replacement template, we resolved the complex structure at 3.6 Å resolution (Fig. 2d, e, Supplementary Fig. 2b and Supplementary Table 1). It is important to note that the structures resolved within each asymmetric unit contained only a single AcrIE10 dimer and half of the IR DNA sequence (Fig. 2d). Leveraging the Alphafold-predicted structure as an initial model, we performed refinements using the electron density map (Fig. 2e). This approach ultimately yielded a refined model that represents the AcrIE10-DNA interaction.

Structural analysis revealed that the two AcrIE10 dimers need to further dimerize to bind to a 20-bp dsDNA molecule containing a complete IR sequence (Fig. 2e). The inherent spacer region between the IR sequence conferred pseudosymmetry to the overall complex in the crystal (Fig. 2e). Crucially, the positively charged RHH domain mediated protein-DNA interaction through electrostatic complementarity (Supplementary Fig. 2c). Similar to the DNA bending observed in HTH-containing Aca proteins²⁹, the RHH domain of AcrIE10 also induces conformational bending of the canonical B-form DNA duplex upon binding to IR DNA (Fig. 2g).

The functional unit of the classical RHH domain is a dimer (RHH2), which forms a stable, symmetrically structured domain through the tight winding of two RHH motifs, enabling DNA binding³⁰. The N-terminal structure of AcrIE10 follows this pattern (Fig. 2f). The short β-strand at the N-terminus of each RHH monomer forms an anti-parallel β-sheet in RHH2 that inserts into the major groove of the core region of the IR dsDNA. Each AcrIE10 monomer engages in specific interactions through the side chains of R7, N8, and S10, in which R7 primarily mediates sequence-specific nucleotide recognition (Fig. 2f–h). Disruption of this interaction via R7A mutation completely abolished AcrIE10 binding to the promoter region (Supplementary Fig. 3a). The second α-helix (α2) of RHH2 interacts with the DNA phosphate backbone through conserved interactions (T29 and V28) at its N-terminal region (Fig. 2h), anchoring the RHH2 domain in its precise positioning. Notably, the formation of the anti-parallel β-sheet is critical for the DNA binding by RHH2, since the F11P mutation, predicted to disrupt the β-sheet structure, significantly weakens the interaction between AcrIE10 and DNA (Supplementary Fig. 3a).

In the AcrIE10-DNA complex structure, we observed that two AcrIE10 dimers cooperatively promote DNA binding through two symmetric dimer-of-dimers interaction interfaces formed between the dimers (Fig. 2i). Although the resolution of the AcrIE10-IR complex limits the precise assignment of side-chain interactions at the dimer-of-dimers interface, our structural analysis reveals a predominantly electrostatic nature of this interaction. The interface is characterized by complementary electrostatic surface potentials rather than high-specificity residue-to-residue contacts. This suggests a model where DNA binding positions two AcrIE10 dimers in a manner that favors their spontaneous stabilization via charge complementarity (Supplementary Fig. 2d). While faint higher-order bands are detectable in the absence of DNA, the presence of the IR sequence significantly promotes and stabilizes the dimer-of-dimers assembly. Thus, DNA likely functions by positioning the AcrIE10 molecules at favorable angles and distances to facilitate the formation of a stable complex. (Supplementary Fig. 3b). The stability of the AcrIE10 dimer-of-dimers interface is precisely regulated by the length of the IR spacer region: a 2-bp spacer allows stable interface formation, whereas shortening (0 bp or 1 bp) or extending (4 bp, 8 bp, or 16 bp) the spacer disrupts the interface and significantly reduces the DNA-binding capacity of AcrIE10. (Fig. 2j).

In summary, our study elucidates the structural basis for AcrIE10 Aca function: the N-terminal RHH domain of AcrIE10 mediates specific recognition of the IR sequence in the promoter region, while the IR spacer region precisely modulates the spatial distance between two AcrIE10 dimers to stabilize the overall complex. This cooperative mechanism enables the dimer of AcrIE10 to efficiently bind to the IR dsDNA, forming a strong and stable repressor complex at the acr promoter.

Cas7* binding modulates AcrIE10 affinity to DNA

AcrIE10 was previously identified as an anti-CRISPR binding to the Cas7* subunit of the type I-E* CRISPR-Cas system, and the crystal structure of the complex has been resolved²⁷. In light of the current study, AcrIE10 can be classified as a bifunctional Acr-Aca protein, sharing similar characteristics with AcrIF24 and AcrIIA1^12,22,31. Notably, among the reported bifunctional Acr-Aca proteins, their anti-CRISPR (Acr) and transcriptional repression (Aca) functions are typically independent of each other^{22,23,25,31–34}. However, when we superimposed the AcrIE10-DNA complex with the AcrIE10-Cas7* complex, there is a conflict between the Cas7* binding site and the dimer-of-dimers interaction mode of AcrIE10 (Fig. 3a).

Fig. 3. Structural insights into the AcrIE10:IR:Cas7* complex.

a Structural basis for the steric clash between AcrIE10 assembly states. Superimposition of the DNA-bound state and the Cas7*-bound state (PDB: 8JI9) reveals a significant interface conflict. The AcrIE10 dimer-of-dimers is shown with its two constituent dimers colored cyan and gray-black (DNA is omitted for clarity). The AcrIE10:Cas7* inhibitory complex is superposed, with the AcrIE10 subunits colored pink and the Cas7* subunit colored magenta. b Biphasic effect of Cas7*on AcrIE10-DNA complex assembly. EMSA analysis showing that the addition of Cas7* initially promotes the formation of a higher-order AcrIE10: Cas7*: IR DNA ternary complex* (indicated by the synergistic shift). However, as the Cas7* concentration increases further, a transition in the assembly state is observed, characterized by the dissociation of the complex and the re-emergence of free DNA. Note that while some DNA remains sequestered in high-molecular-weight (HMW) species at the top of the gel, the increase in free DNA at the highest Cas7* concentrations indicates a destabilization of the promoter-bound assembly. 2-dimensional (2D) classification results of two classes of particles from cryo-EM analysis of the AcrIE10:Cas7*:IR complex, with corresponding schematics showing the assembly process. Arrows indicate AcrIE10 molecules (light blue) and Cas7* molecules (bright pink), consistent with the colors in (c). Detailed analysis can be found in Supplementary Fig 4. Scale bar: 10 nm. Each experiment was repeated at least three times independently, yielding consistent results. c, d Cryo-EM density map and cartoon representation of the AcrIE10:Cas7*:IR complex with domain schematic of AcrIE10 and Cas7*, along with the secondary structure of the IR sequence used for complex assembly. e A schematic of the AcrIE10:Cas7*:IR complex formation. f Subsection of the structure in c, showing RHH2 structure of AcrIE10 dimer bound to DNA. g Comparison between the AcrIE10:Cas7*:IR complex and the AcrIE10-IR complex (gray). The molecules in both complexes are shown in a cartoon. h Subsection of the structure in c, showing AcrIE10:Cas7* interaction. i Comparison between the AcrIE10:Cas7*:IR complex and AcrIE10:Cas7* (PDB:8JI9, yellow and purple).

To investigate how the interaction between AcrIE10 and Cas7* modulates the Aca function of AcrIE10, we first employed EMSA experiments. By gradually increasing the concentration of Cas7* in the AcrIE10-IR transcriptional repression complex system, we simulated the CRISPR-mediated clearance pressure on foreign mobile genetic elements (MGEs). Surprisingly, the addition of Cas7* exerts a biphasic effect on the DNA-binding activity of AcrIE10. Initially, low concentrations of Cas7* significantly enhance the mobility shift, indicating the formation of a synergistic AcrIE10-Cas7*-DNA ternary complex. However, as Cas7* levels increase further, this complex gradually diminishes, accompanied by the re-emergence of free DNA. This suggests that excessive Cas7* may destabilize the AcrIE10-DNA interaction, eventually leading to the release of the promoter DNA. (Fig. 3b). To further validate that AcrIE10 DNA binding activity was modulated by Cas7*, we constructed a Cas7*-binding-deficient mutant AcrIE10-E78A (Supplementary Fig 3c) and repeated the experiment. The results confirmed that the band shifts were entirely dependent on Cas7* binding (Supplementary Fig 3d). These results demonstrate that multiple forms of AcrIE10 complexes could exist: at low Cas7* concentrations, AcrIE10 simultaneously binds to DNA and Cas7*, while a further increase of Cas7* concentration results in destabilization of promoter-bound AcrIE10 complexes.

Structural basis for AcrIE10 allosteric regulation by Cas7*

To resolve the molecular details of the ternary complex observed at low Cas7* concentration, cryo-electron microscopy single-particle analysis (cryo-EM SPA) was performed on the AcrIE10:Cas7*:IR dsDNA assembly (Red asterisks indicate conditions used for cryo-EM sample preparation, Fig. 3b, Supplementary Fig. 4a–d and Supplementary Table 2.). Two-dimensional classification revealed two distinct conformational states: (1) an asymmetric complex comprising two AcrIE10 dimers, one Cas7* subunit, and one dsDNA molecule; (2) a symmetric complex incorporating an additional Cas7* subunit (Fig. 3c–e). Both complexes were resolved to a resolution of 3.4 Å, with refined density maps and atomic models exhibiting perfect agreement with the state transitions observed in EMSA experiments (Fig. 3b).

To exclude the possibility that flanking DNA sequences might sterically interfere with the formation of the AcrIE10-Cas7* assembly, EMSA experiments were performed using a longer 50-bp DNA substrate. The results confirmed that the ternary complex forms effectively regardless of the DNA length (Supplementary Fig. 5e). Structural modeling colored by AlphaFold confidence (pLDDT) suggests that the Cas7* loop regions possess sufficient structural plausibility to accommodate longer DNA (Supplementary Fig. 5a, b). These findings indicate that the Cas7* sensing mechanism is robust and not restricted by the flanking DNA scaffold. Further comparison between the AcrIE10-DNA crystal structure and cryo-EM model revealed no significant conformational changes in the RHH2 domain upon DNA binding (RMSD = 0.763 Å, Fig. 3f–g). This suggests that Cas7* binding to the external interface of AcrIE10 dimer does not interfere with AcrIE10-DNA interaction. Similarly, structural comparisons with reported DNA-free AcrIE10:Cas7* complexes (PDB:8JI9) reveal that the AcrIE10:Cas7* interface does not change upon DNA binding(RMSD = 0.956 Å, Fig. 3h, i).

AcrIE10 alleviates transcriptional repression in the presence of CRISPR-Cas system

Structural analysis demonstrated that AcrIE10 forms two symmetric dimer-of-dimers interfaces when bound to DNA. For clarity, we defined the negatively charged surface as Surface1 and the opposing binding surface as Surface2 (Fig. 4a). Specifically, Surface1 is formed by the outermost helix at the C-terminus of one AcrIE10 molecule, where the negatively charged residues E78, N77, and D81 engage in critical interactions with R113 and R43 from the opposing AcrIE10 molecule (Fig. 4a). Experimental evidence suggests that this dimer-of-dimers interaction is only stable at specific distances and angles, which is essential for maintaining the DNA-binding capability of the RHH2 domain (Fig. 2j). Intriguingly, a structural comparison of the AcrIE10-DNA-Cas7* complex with the AcrIE10-Cas7* complex indicates that the Cas7*-binding interface coincides with Surface1 (Fig. 4b). In particular, E78 and D81, which are critical for Cas7* binding, are also conserved in the dimer-of-dimers interactions (Fig. 4b).

Fig. 4. AcrIE10 alleviates its self-repression upon CRISPR immune activation.

a Cartoon representation of the AcrIE10 dimer-of-dimers interface in the AcrIE10-IR-Cas7* complex. For clarity, the DNA structure is not shown. The dimer-of-dimers interface is highlighted with a blue box, with the detailed interactions presented. AcrIE10 molecules are colored in light blue, blue, green, and gray-green, while Cas7* is shown in pink and light pink. Detailed view (bottom) of one of the dimer-of-dimers interfaces, with amino acid residues involved in interface formation displayed as sticks. The electrostatic potential surface of surface1 and surface2 is maintained in a cartoon. The surface was generated using ChimeraX, with red representing −1.0 kT and blue representing +1.0 kT. b Cartoon representation of the AcrIE10:Cas7* interface in the AcrIE10- Cas7* complex (PDB: 8JI9). The interface is highlighted with a red box, with the detailed interactions presented. Detailed view (bottom) of one of the dimer-of-dimers interfaces, with amino acid residues involved in interface formation displayed as sticks. The electrostatic potential surface of surface1 and surface2 is maintained in a cartoon. The surface was generated using ChimeraX, with red representing −1.0 kT and blue representing +1.0 kT. c Schematic diagram of the AcrIE10 transition from Aca to Acr in the presence of the CRISPR-Cas system. d EMSA demonstrates the re-establishment of transcriptional repression by excess AcrIE10. AcrIE10 initially recruits Cas7* for synergistic binding; higher concentrations of AcrIE10 lead to the formation of a direct AcrIE10-DNA complex, thereby maintaining the repressed state of the promoter. The uppermost band indicates the formation of a transient complex at low Cas7* concentrations (0.1 μM). Each experiment was repeated at least three times independently, yielding consistent results. e Size-exclusion chromatography (SEC) analysis of AcrIE10–Cas7* complex formation. The interaction was analyzed at different molar ratios of the components, with Cas7* (50 μM) incubated with AcrIE10 at either 50 μM (2:2 molar ratio) or 100 μM (2:1 molar ratio). Please see Supplementary Fig. 6b for control profiles. The data are representative of two independent experiments. AcrIE10 monomers are labeled in light blue or green, and Cas7* in pink. f A model explaining the differential binding of AcrIE10:Cas7* complexes to DNA depending on their stoichiometry. g ITC results demonstrating titration of free AcrIE10 (10 μM) or AcrIE10:Cas7* complexes (15 μM) at (2:1) and (2:2) molar ratios against IR DNA (100 μM).

Based on these findings, we propose that after the external interface of the AcrIE10 dimer is occupied by Cas7*, further addition of Cas7* will result in its binding to the Surface1, competitively disrupting the dimer-of-dimers interface of AcrIE10. This weakens the binding of the RHH2 domain to DNA, thereby initiating the anti-CRISPR expression (Fig. 4c). To confirm this model in vivo, we used qPCR to measure acrIE9.2 and acrIE10 expression levels in a clinical strain of K. pneumoniae R41 and demonstrated that Cas7* overexpression results in transcriptional activation of the acr operon (Supplementary Fig. 6a). This mechanism allows us to fine-tune Acr expression to the CRISPR-Cas abundance in the host cell, via direct sensing of Cas7* subunit. We suppose that at the early stages of expression, AcrIE10 will preferentially form 2:2 stoichiometry complexes with Cas7* that are unable to bind DNA; however, when the free Cas7* subunit is completely bound, further production of AcrIE10 will result in the assembly of AcrIE10 dimers of dimers on DNA, resulting in the repression of the acr loci. Conversely, a gradual increase in Cas7* level is predicted to result in the destabilization of the AcrIE10: DNA complexes. This supports a model where AcrIE10 facilitates a functional transition from Aca-mediated repression to Acr-mediated CRISPR inhibition (Fig. 4c).

Stoichiometry of the Cas7*:AcrIE10 complexes determines their DNA-binding ability

To understand the mechanistic details of AcrIE10 balancing between Acr and Aca functions and confirm the stoichiometry of the mutually exclusive AcrIE10: DNA and AcrIE10:Cas7* complexes, we performed EMSA with increasing ratios of AcrIE10 -to-Cas7* to simulate this process (Fig. 4d, Supplementary Fig. 3e). The results showed that upon increase of AcrIE10 concentration, ternary Cas7*:AcrIE10:DNA complex emerged at low concentrations of Cas7* (0.1 μM) before AcrIE10:DNA complex, suggesting that the inhibition-relieved and inhibition-restored states of AcrIE10 are structurally indistinguishable. Moreover, in size-exclusion chromatography experiments, we observed different peaks when incubating AcrIE10 and Cas7* proteins at different stoichiometric ratios (Fig. 4e). When we incubated AcrIE10 and Cas7* at a 1:1 ratio, a unique peak corresponding to a 2:2 AcrIE10:Cas7* complex was obtained, as previously described²⁷. However, when we incubated AcrIE10 and Cas7* with excess of AcrIE10, a new peak appeared and shifted to a later position, indicating formation of a lower molecular weight species (Fig. 4e, Supplementary Fig. 6b). This peak also contained AcrIE10 and Cas7*, likely representing a 2:1 AcrIE10:Cas7* stoichiometry (Fig. 4e). Combining the structural information and molecular weight analysis, we propose that the complex within this peak represents a 2:1 AcrIE10:Cas7* stoichiometry. Next, we compared the affinity to the IR DNA for the AcrIE10:Cas7* complexes of the 2:2 and 2:1 stoichiometry. ITC results demonstrated that while the 2:1 complex still binds to the IR sequence, with a binding affinity (K_D value) similar to that of the AcrIE10 dimer (Fig. 4g), a 2:2 complex completely lacked affinity to DNA.

These results confirm that AcrIE10 can form multiple complexes with Cas7* that have different DNA-binding activity: an excess of Cas7* results in a saturation of both binding interfaces of the AcrIE10 dimer, and formation of 2:2 AcrIE10:Cas7* complexes unable to interact with DNA; in contrast, excess of AcrIE10 results in the formation of 2:1 AcrIE10:Cas7* complexes, leaving the second Cas7*-binding interface available for interaction between AcrIE10 dimers, this will ultimately result in the binding of the AcrIE10 dimer of dimers to the acr promoter and transcriptional repression (Fig. 4f).

AcrIE10-mediated dynamic equilibrium between CRISPR immunity and anti-immunity

Our study reveals that the dynamic equilibrium between plasmid-encoded anti-immunity and host CRISPR-Cas immunity is intrinsically linked to the unique bifunctional domain organization of AcrIE10. The coordinated interplay between its anti-CRISPR (Acr) and autoregulatory (Aca) functions establishes a responsive yet self-limiting regulatory circuit. Despite being equipped with a robust CRISPR-Cas immune system, the ST15 CRKP strains investigated in this study stably retain their CRISPR-Cas-targeted antimicrobial-resistant plasmid through the anti-CRISPR activity of the acrIE9.2-acrIE10 locus.

Under normal conditions, bacterial CRISPR immune systems remain in a silent or low-activity state to minimize fitness cost. Meanwhile, our structural and biochemical data suggest that the AcrIE10 protein likely adapts to the CRISPR system’s immunologically silent state by suppressing its own expression via the N-terminal RHH domain (functions as Aca, Supplementary Fig. 6c). Upon plasmid invasion, activation of the CRISPR-Cas system disrupts the functional equilibrium between CRISPR-mediated immunity and Acr counter-defense, shifting the balance toward CRISPR-dominant immune protection (Fig. 5, The pink phase). Immune effector proteins such as Cas7* increasingly accumulate and competitively bind the Acr domain of AcrIE10, resulting in the disruption of the AcrIE10 dimer–dimer interface, thereby alleviating Aca-mediated repression. Consequently, it triggers robust acrIE10 transcription, boosting anti-CRISPR activity in response to host immunity. Overexpression of AcrIE9.2 and AcrIE10 inhibits CRISPR-mediated immune defense, ultimately shifting the CRISPR-Acr balance toward anti-CRISPR dominance (Fig. 5 cyan phase). Acr-deficient AcrIE10 fails to respond to CRISPR-Cas activation, allowing unrestrained host immunity to clear the plasmid (Supplementary Fig. 6c, d).

Fig. 5. Mechanistic insights into AcrIE10-mediated dynamic regulation.

A dynamic equilibrium model illustrating the AcrIE10-mediated mechanism for maintaining plasmid–host coexistence. Upon host immune activation (pink), high CRISPR-Cas activity in the bacterium triggers the release of AcrIE10’s self-repression (Aca function), initiating its anti-CRISPR activity to protect the plasmid from host immunity (cyan). Once the anti-immunity response reaches a threshold, AcrIE10 rebinds to the upstream promoter to repress its own transcription, preventing toxic overexpression and sustaining plasmid–host coexistence.

As AcrIE10 accumulates, a stoichiometric shift promotes reformation of the DNA-binding competent dimer. The RHH domain re-engages with the promoter, restoring repression while anti-CRISPR activity remains sufficient to suppress immunity (functions as Aca again). The CRISPR system and Acr can return to a basal state of functional equilibrium. This co-functional state of Acr and Aca establishes a tunable, self-limiting feedback loop that supports plasmid-host coexistence (Fig. 5).

Discussion

Antibiotic resistance of clinically significant bacteria, including K. pneumoniae and other pathogens of the ESKAPE group, represents one of the most important challenges to public health³⁵. In the globally prevalent ST15 CRKP (carbapenem-resistant K. pneumoniae) strain, the presence of the anti-CRISPR proteins may be the reason for the stable coexistence of the resident Type I-E* CRISPR-Cas systems and CRISPR-targeted bla_KPC plasmids⁹. This study unveiled the complex mechanisms underlying transcriptional regulation of acr loci in CRKP bla_KPC plasmids, and reports an instance of an RHH-mediated Acr repression.

AcrIE10 was discovered as an Anti-CRISPR that interacts with the Cas7* subunit of K. pneumoniae Type I-E* CRISPR-Cas system²⁷. In this study, we identified that AcrIE10 is a bifunctional Acr-Aca protein that encompasses a C-terminal Acr domain and an N-terminal RHH Aca domain and auto-regulates transcription of the acrIE9.2-acrIE10 loci. Building upon the structural framework established by Jiang et al.²⁷, which identified AcrIE10 as a dimeric protein, we further elucidate that AcrIE10 functions as a dimer-of-dimers to exert its regulatory roles. Our findings also support a model in which the promoter DNA serves as a molecular scaffold to facilitate and stabilize the higher-order oligomerization of AcrIE10. Ten experimentally verified Aca proteins (Aca1-10) have been identified as conserved HTH domain proteins encoded in the vicinity of acr clusters^15,16. In addition, a limited number of Acr proteins (AcrIIA1, AcrIIA6, AcrIIA13, AcrIIA14, AcrIIA15, AcrIF24) were shown to contain HTH domain fusions involved in transcriptional autoregulation and thus represent a bifunctional Acr-Aca protein. AcrIE10 is an example of a non-HTH domain Aca protein, which functions as the transcriptional regulator through an RHH domain. AcrIE10 engages with promoter DNA through a unique dimerization interface, and the antiparallel β-sheet of the RHH domain mediates sequence-specific recognition within the DNA major groove, similar to other RHH transcription factors²⁸. AcrIE10 forms a stable dimer-of-dimers complex independently (Fig. 4a). This unique feature likely enables rapid adaptation to fluctuating CRISPR-Cas pressure within the host. Similar structural features have been reported in prior studies of AcrIF25, though detailed characterization remains absent¹³. Our study redefines the boundaries of Aca proteins, previously regarded as landmarks for locating and identifying novel Acrs, paving the way for the identification of novel RHH repressors and anti-CRISPR proteins under their regulation.

Aca and Acr-Aca proteins have been identified mainly in the genomes of phages and archaeal viruses³⁶. AcrIE10 is an example of Acr-Aca predominantly found on plasmids, which should affect its interaction with the CRISPR-Cas systems, due to the different copy number of the acr gene compared to prophages, and different infection dynamics and outcomes, compared to the phage infection. It was proposed that uncontrolled Acr expression could have detrimental effects on the MGE replication, thus after a burst of Acr production at the onset of infection, further expression of the acr genes should be downregulated to mitigate toxic effects²⁰. Negative feedback loop is the predominant regulatory mechanism used by bacteria and phages to support this type of regulation, and Aca proteins are co-transcribed with the acr genes they regulate³⁷.

Here, we reveal that negative-feedback regulation of AcrIE10 is controlled by its interaction with Cas7*, which should allow MGE to fine-tune acr expression level to the CRISPR-Cas abundance. Acr proteins need a rapid and sufficient production to inactivate or evade the host CRISPR-Cas immune mechanism. Many CRISPR-Cas systems^38–40 can upregulate their expression levels or even activate accessory immunity systems in the presence of inhibitory Acr proteins. Strikingly, Acr proteins also de-repress CRISPR-Cas transcription. It could be assumed that dual activity Acr-Aca proteins that sense CRISPR-Cas abundance serve as a countermeasure to these regulatory circuits, allowing MGE to adjust Acr expression even when the CRISPR-Cas system is upregulated. Previous study found the phenomenon that AcrIIA1 acts as a “Cas9 sensor”³¹, our study revealed the mechanism behind the transcriptional activation of Acr in response to increased Cas expression and self-repression with the decrease of Cas7* level.

Based on our structural and biochemical findings, we propose a model where AcrIE10 regulates transcription through stoichiometry-dependent conformational remodeling triggered by interaction with Cas7*. The interaction interface between AcrIE10 dimers and between AcrIE10 and Cas7* overlaps; therefore, the AcrIE10:Cas7* complex retains DNA-binding activity at 2:1 stoichiometry but loses the ability to dimerize on DNA at 2:2 stoichiometry. While our in vivo data demonstrate that Cas7* overexpression leads to acr derepression (Supplementary Fig. 6a), directly capturing the real-time dynamics of Cas7* fluctuation and promoter activity during rapid physiological events, such as plasmid conjugation or CRISPR activation, remains a challenge for the upcoming studies. We also noticed that binding of Cas7* to AcrIE10 increases its sensitivity to proteolysis, which can ensure clearance of Cas7* from inactive AcrIE10:Cas7* (2:2) complexes and restoration of the Aca function. This finely tuned regulatory mechanism not only explains the persistence of bla_KPC-harboring plasmids in ST15 K. pneumoniae strains but also provides a general framework for understanding how various MGEs adapt to coexistence with their hosts.

Previous studies identified that aca genes can be widely co-opted by different acr clusters¹⁶. However, bioinformatics analysis demonstrates that acrIE9.2 and acrIE10 genes likely form a functional unit, suggesting that the Acr-Aca function of AcrIE10 is deeply connected with AcrIE9.2. Of interest, both proteins target the Cas7 subunit of the CRISPR-Cas system^27,41, although AcrIE9.2 interacts with Cas7 as part of a Cascade complex, while AcrIE10 binds to a standalone Cas7 subunit. Potentially, both proteins could act cooperatively or even synergistically to disarm CRISPR-Cas systems, a mechanism that warrants further studies. Alternatively, the activity of two Acrs can be mediated at the different stages of the plasmid lifecycle, i.e., one protein could disarm existing Cascade complexes at the onset of conjugation, while the second could control assembly of novel complexes during conjugative plasmid maintenance. Our plasmid stability assay indicates that AcrIE9.2 acts as the primary anti-CRISPR protein within the acr operon, while the anti-CRISPR activity of AcrIE10 is weaker. Therefore, AcrIE10 may function as a sensor-regulator that monitors Cas7* levels to modulate the operon’s transcription at the maintenance stage, ensuring a balanced anti-CRISPR response and mitigating Acr effects on bacterial fitness. It remains to be determined whether Cas7* binding to AcrIE10 occurs through the depletion of the free Cas7* pool prior to Cascade assembly, or by sequestering it from the pre-formed Cascade complex and whether AcrIE9.2 activity contributes to Cas7 release. Our further study will try to explore the molecular interplay among those two Acr proteins and Cas7*.

In summary, our results elucidate molecular mechanisms allowing AcrIE10 to operate as a bifunctional Acr-Aca protein and report an example of RHH-based regulation of acr expression. We demonstrate that AcrIE10 Aca activity is controlled by Cas7* abundance. This allows to dynamically balance of acr expression and ensures the compatibility of CRISPR-targeted antibiotic resistance plasmids with the pathogenic K. pneumoniae strains. This study provides an insight into the emergence and dissemination of clinically-significant pathogenic strains from the perspective of the interaction between bacterial immunity systems and antimicrobial-resistant plasmids.

Methods

Bacterial strains

K. pneumoniae RJKP41(R41) is a clinical isolate that contains an acrIE9.2-acrIE10 operon (GenBank accession number GCA_025884075.1). Competent E. coli strains of DH5α and BL21(DE3) were purchased from Sangon Biotech. If not specified, all bacterial strains were grown at 37 °C in Luria-Bertani (LB) media with 180 rpm shaking for liquid cultures, and with agar (15 g/L) for solid plates. For maintaining engineered plasmids, media were supplemented with kanamycin (50 μg/ml), apramycin sulfate (50 μg/ml), chloramphenicol (200 μg/ml), ampicillin (50 μg/ml) or meropenem (5 μg/ml). The publicly available Enterobacteriaceae strain (9358 in total) were downloaded from BACTERIAL BIOINFORMATICS RESOURCE CENTER, PATRIC. ST15 K. pneumoniae was isolated from patients at a teaching hospital in Shanghai, China.

Bioinformatic analysis of AcrIE10

Gene neighborhood analysis tool webFlaGs⁴² was used to explore the genetic environment of this acr locus with default parameters. The acrIE10 locus was searched using BLAST against a database downloaded from publicly available Enterobacteriaceae family with complete genome sequences (total 9358, Supplementary Data 1) and 88 clinical isolates of ST15 K. pneumoniae (Supplementary Data 1) with a query of identities >90% and coverage >80% to explore the correlation between the presence of the acrIE10 locus and transfer of plasmids. Molecular Evolutionary Genetics Analysis (Mega) software was used to construct the phylogenetic tree of AcrIE10 and its homologous proteins. Interactive Tree Of Life (iTOL) was employed for the visualization of the phylogenetic tree. CRISPR loci in AcrIE10-positive isolates were identified using CRISPRCas-Finder software with default parameters. Only isolates with CRISPR-Cas arrays classified as evidence level 4 were retained as CRISPR-Cas-positive. For isolates co-harboring CRISPR-Cas systems and AcrIE10, BLAST was employed to detect CRISPR-Cas targeted plasmids with 100% sequence identity to spacers.

Plasmid construction

The expression plasmid of acrIE10 was used as an example to illustrate the method of plasmid construction. The encoding gene of AcrIE10 protein (GenBank accession number: OQ389520) was PCR-amplified from genomic of a clinical isolate K. pneumoniae R41, cleaved with EcoRI and HindIII (Thermo Fisher Scientific) and ligated to pET28a. Subsequently, the amplified fragment and the linearized plasmid were recombined using Hieff Clone® Plus One Step Cloning Kit purchased from Yeasen Biotech according to the manufacturer’s instructions. Other engineered plasmids were constructed using the same method. Construction of these plasmids was confirmed by sequencing using the universal primers, respectively. Primers used for plasmid construction were listed in Supplementary Table 3.

Knockout and mutation of acrIE10 in clinical isolate R41

To construct N-terminal-domain and C-terminal-domain mutations of AcrIE10 in K. pneumoniae R41, we generated vectors for each mutation using pCONJ5H, in which the hygromycin phosphotransferase gene (hph) was employed to replace the resistance genes in pCONJ working vectors⁴³. Upstream and downstream flanking regions (approximately 500 bp) of the target gene were amplified by PCR with the relevant primers and assembled into pCONJ5H, with NEBuilder HiFi DNA Assembly Cloning Kit (New England Biolabs). K. pneumoniae mutants were then generated by conjugation. A single colony of the transconjugant was used to inoculate liquid LB with 200 μg/ml hygromycin. Bacteria were collected by centrifugation, washed with phosphate-buffered saline (PBS) and diluted five-fold into 5 ml LB, and then grown for 4 h at 37 °C. 5 μl of the culture was then plated on salt-free LB agar containing 40% (w/v) sucrose to select for loss of the integrated plasmid. Colonies were analyzed by PCR to identify mutants and confirmed by sequencing.

Plasmid containing reporter assays

The reporter plasmid (pUC-EGFP) encodes the green fluorescent protein gene (egfp) under the control of the lac promoter. A negative control was constructed by removing the lac promoter. To determine the activity of the acrIE9.2-acrIE10 promoter, the lac promoter in the reporter plasmid was replaced by the −35 and −10 region of the acrIE9.2-acrIE10 promoter (pUC-pro-eGFP). Each eGFP reporter plasmid was tested with an acrIE10 expression plasmid (pACYC-AcrIE10) as well as the corresponding empty vector (pACYC-EV) in E. coli BL21. Overnight cultures of BL21 model strains containing plasmids were grown on plates of LB with 1.5% agar at 37 °C. After 20 h of growth, the eGFP expression of the strain was visualized by a Blue LED transilluminator (Monad Biotech Co., Ltd, Shanghai, China).

RNA extraction and RT-qPCR

Overnight cultures of clinical isolates were diluted 1:100 into LB broth and grown with shaking at 37 °C to an OD600 of 0.6. Cell pellets were resuspended in 5 mg/mL lysozyme solution and incubated at 37 °C for 30 min, then centrifuged at 12,000 rpm for 5 min. After discarding the supernatant and collecting the pellet, total RNA was extracted with SteadyPure RNA Extraction Kit (Accurate Biology) according to the manufacturer’s instructions. cDNA was synthesized using Evo M-MLV RT Kit (Accurate Biology) and quantified using SYBR Green Premix Pro Taq HS qPCR Kit (Accurate Biology) with primers listed in Supplementary Table 3. Gene repA was used as an endogenous reference gene. The experiment was repeated for three times. Data were analyzed using Applied Biosystems by Thermo Fisher Scientific. A two-sided Welch’s t-test was implemented to evaluate statistical significance, without adjustment for multiple comparisons.

Plasmid stability analysis

E. coli BL21(DE3) (Sangon Biotech) carrying engineer plasmid pHSG398 encoding the type I-E* CRISPR-Cas system (pCRISPR) was constructed as described in our previous study (BL21-pCRISPR)⁹. Then, pRSF-duet plasmids carrying the wild-type acrIE10 gene and its mutant variants was constructed and transformed into BL21-pCRISPR. The transformants were passaged every 12 h in LB medium containing chloramphenicol (15 μg/mL) in order to maintain the stable presence of the pCRISPR. For each passage, serially diluted cultures were plated on LB agar with chloramphenicol (15 μg/mL) and grown at 37 °C for 12 h. One hundred colonies were collected from each plate and spotted on chloramphenicol-containing LB plates in the presence or absence of Kanamycin (50 μg/mL). The relative plasmid stability was calculated by comparing the number of colonies on the LB plate with or without Kanamycin.

Protein purification

The DNA sequence encoding AcrIE10 was cloned from R41 and inserted into the pET His6 Sumo cloning vector. After sequence verification, the plasmids were transformed into Escherichia coli BL21(DE3) cells for expression in Luria-Bertani (LB) medium. When the optical density at 600 nm (OD600) reached 0.6, protein expression was induced by adding 0.5 mM isopropyl-β-d-1-thiogalactopyranoside (IPTG), followed by incubation at 16 °C overnight. Harvested cells were resuspended in lysis buffer (50 mM Tris, pH 8.0, 500 mM NaCl, 25 mM imidazole) and lysed using high-pressure homogenization. After centrifugation, the supernatant was applied to prepacked Ni-NTA columns (HisTrap, Cytiva) and extensively washed with lysis buffer. AcrIE10 was eluted using lysis buffer with a gradient concentration of imidazole. To remove the His6-SUMO tag, the protein was digested with ULP protease overnight and further purified using Ni-NTA columns to obtain tag-free AcrIE10. The purified AcrIE10 protein was concentrated and subjected to a Superdex 75 Increase 10/300 GL column (Cytiva) for final purification in a buffer containing 20 mM Tris (pH 8.0) and 150 mM NaCl in 0.5 ml/min flow rates. All purification steps were performed at 24 °C.

The purification of Cas7* followed a similar procedure. Plasmids encoding His6-Sumo-Cas7* were transformed into BL21(DE3) cells, and protein expression was induced under the same conditions as AcrIE10. Harvested cells were resuspended in lysis buffer (20 mM HEPES, pH 7.5, 500 mM NaCl, 5% glycerol, 1 mM TCEP) supplemented with 0.2 mM phenylmethylsulfonyl fluoride (PMSF) and lysed by high-pressure homogenization. After centrifugation, the supernatant was loaded onto prepacked Ni-NTA columns and eluted in lysis buffer with a gradient concentration of imidazole. The His6-SUMO tag was removed via ULP protease digestion for 3 h followed by further purification using a HiTrap QHP ion-exchange column (Cytiva). The protein was concentrated and subjected to a Superdex 75 Increase 10/300 GL column (Cytiva) for final purification in a buffer containing 20 mM HEPES (pH 7.5), 150 mM NaCl, and 5% glycerol. All purification steps were performed at 4 °C.

Mutagenesis

All AcrIE10 mutations were introduced using PCR-based QuickChange site-directed mutagenesis and confirmed by DNA sequencing. The primers used for mutagenesis are listed in Supplementary Table 3. Mutant proteins were expressed and purified following the same protocol as the wild-type proteins.

Electrophoretic Mobility Shift Assays (EMSAs)

EMSAs were performed in a reaction volume of 20 μl containing 20 mM HEPES (pH 7.5), 100 mM NaCl, 5% glycerol, 5 mM MgCl2, 0.1 μg/μl BSA, and purified AcrIE10 protein at the concentrations indicated in the figure legends. Binding reactions were incubated at room temperature for 15 min. For competition experiments, dsDNA (final concentration 200 nM) was pre-incubated with AcrIE10 for 15 min before adding Cas7*, followed by a further 15-min incubation. After incubation, 2.5 μl of loading dye (0.5 × TBE: 45 mM Tris, pH 8.3, 45 mM boric acid, 1 mM EDTA, 34% glycerol, 0.2% bromophenol blue) was added, and samples were loaded onto an 8% polyacrylamide gel (19:1 acrylamide/bis-acrylamide, 1 × TBE, 2.5% glycerol, 0.6 mg/ml ammonium persulfate, 0.05% TEMED). Gels were pre-cooled to 4 °C and run at 110 V for approximately 1.5 h at 4 °C. DNA bands were visualized using SYBR Gold (Invitrogen) staining and imaged at 488 nm with a Typhoon imaging system (GE).

Preparation of substrate DNA

Double-stranded DNA (dsDNA) substrates for crystallization and electrophoretic mobility shift assays (EMSAs) were prepared by annealing single-stranded oligonucleotides (synthesized by Sangon Biotech) designed based on the promoter region of acrIE10. Oligonucleotides were denatured at 95 °C for 5 min and annealed at room temperature. DNA sequences are provided in the Supplementary Table 3.

Crystallization and structure determination

For AcrIE10 alone, crystallization drops contained 1 μl of purified protein (5 mg/ml) mixed with 1 μl of reservoir solution A (4% v/v Tacsimate, pH 7.0, 12% w/v polyethylene glycol 3350) in hanging-drop vapor diffusion plates equilibrated against 500 μl of reservoir solution A. Crystals grew at 16 °C. The resulting rod-shaped crystals were transferred to reservoir solution A supplemented with 20% glycerol and flash-cooled in liquid nitrogen.

For the AcrIE10-IR nucleic acid complex, crystallization drops contained 1 μl of complex (8 mg/ml, prepared by mixing DNA and protein at a 1.5:1 molar ratio and incubating at room temperature for 30 min) and 1 μl of reservoir solution A. Subsequent steps were identical to those for AcrIE10 alone.

Diffraction data were collected at beamlines BL10U2 and BL18U at the Shanghai Synchrotron Radiation Facility (SSRF). The 3D models of AcrIE10 and AcrIE10-IR complexes were generated using AlphaFold2, and molecular replacement was performed with CCP4i⁴⁴ and PHENIX⁴⁵. Structures were refined using PHENIX⁴⁵ and iterative manual fitting in COOT⁴⁶. Crystallographic data collection and refinement statistics are summarized in the Supplementary Table 1.

Cryo-EM sample preparation and data collection

For cryo-EM sample preparation, annealed DNA (10 μl) was mixed with 10 μl of AcrIE10 protein (1 mg/ml) and 10 μl of Cas7* protein (prepared as described above) and incubated at 22 °C for 15 min. The final buffer contained 20 mM Tris-HCl (pH 8.0), 150 mM NaCl, and 5 mM MgCl2 with a DNA: protein molar ratio of 1:4:2. For grid preparation, glow-discharged Cu300 R1.2/1.3 holey carbon grids (Quantifoil) were mounted in a Vitrobot Mark IV chamber (Thermo Fisher Scientific) at 4 °C and 100% humidity. A 3.5-μl aliquot of the complex was applied, blotted with grade 595 filter paper (Ted Pella), and immediately plunged into liquid ethane.

Cryo-EM data were collected using a Thermo Scientific Titan Krios G3i cryo-TEM at MIT.nano, equipped with a K3 direct detector (Gatan) operated in super-resolution mode with 2 × binning. Micrographs were acquired using EPU software in AFIS mode, resulting in 4178 movies at a nominal pixel size of 0.959 Å. Defocus ranged from −0.8 μm to −1.2 μm, with an exposure time of 1.33 s fractionated into 40 frames, yielding a total dose of 42.2 e⁻ Å⁻² per micrograph.

Cryo-EM data processing

The cryo-EM data processing workflow is summarized in the Supplementary Figs. Movies were motion-corrected using the Patch Motion Corr module in CryoSPARC⁴⁷, and contrast transfer function (CTF) parameters were estimated using the Patch CTF tool. Particles were initially picked using the Blob Picker module (size range 100–220 Å) in CryoSPARC, extracting 1,413,197 particles. To further enhance map quality and resolve rarer particle views, template-based picking was performed using the best 2D class averages from the initial round. Particles were extracted with a box size of 180 pixels, corresponding to a physical size of ~173 Å, which is more than twice the maximum dimension of the complex to ensure accurate CTF correction. After multiple rounds of 2D classification, 289,428 high-quality particles were selected. Initial models constructed from these particles exhibited anisotropy. Among the 3D classes obtained, two classes containing clear dsDNA and protein densities (106,123 and 108,754 particles, respectively) were subjected to CTF refinement, reaching resolutions of 3.46 Å and 3.45 Å. One class with improved density was further refined, yielding a final resolution of 3.76 Å, validated with a gold-standard Fourier shell correlation (FSC) cutoff of 0.143. Data collection and refinement statistics are detailed in Supplementary Table 2.

The final atomic models were generated by rigid-body docking of AlphaFold3 (https://alphafoldserver.com) templates, followed by multiple rounds of real-space refinement in PHENIX⁴⁵ and manual adjustment in COOT⁴⁶ to resolve side-chain positions based on the experimental density. Figures, including those of the crystal structures, were created using UCSF ChimeraX⁴⁸ and PyMOL (http://www.pymol.org).

Isothermal titration calorimetry (ITC)

ITC experiments were performed using a MicroCal PEAQ-ITC system (Malvern Panalytical). AcrIE10 (10 μM) or AcrIE10:Cas7* (15 μM) was titrated against dsDNA (100–200 μM) in a buffer containing 150 mM NaCl, 20 mM HEPES (pH 7.5), and 1 mM MgCl₂ at 25 °C. The titration included a single 0.8 µL injection followed by 15 injections of 2 µL each, with 150-second intervals. Data were analyzed using GraphPad Prism 8.

Chemical cross-linking

To investigate the oligomeric state of AcrIE10 in the presence or absence of DNA, chemical cross-linking assays were performed using 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC, Sigma-Aldrich). AcrIE10 (0.5 μM) was incubated with or without the 20-bp IR DNA (0.1 μM) in a reaction buffer containing 20 mM HEPES (pH 7.5) and 150 mM NaCl for 15 min at room temperature to allow complex formation. Subsequently, EDC was added to a final concentration of 0.1 mM (freshly prepared). The reaction mixtures were incubated at room temperature for 45 min. To stop the cross-linking, the reactions were quenched by the addition of Tris-HCl (pH 8.0) to a final concentration of 10 mM and incubated for an additional 15 min. The samples were then mixed with SDS loading buffer, boiled at 95 °C for 10 min, and analyzed by SDS-PAGE using 4–20% gradient gels. Protein bands were visualized by Coomassie Brilliant Blue staining to detect the formation of dimer-of-dimers (tetramers).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary information

Supplementary Information ^{(177.4MB, pdf)}

41467_2026_76353_MOESM2_ESM.pdf ^{(105.2KB, pdf)}

Description of Additional Supplementary Files

Supplementary Data 1 ^{(160.2KB, xlsx)}

Supplementary Data 2 ^{(193.7KB, xlsx)}

Reporting Summary ^{(1.7MB, pdf)}

Transparent Peer Review file ^{(1.1MB, pdf)}

Source data

Source Data ^{(51.7MB, zip)}

Author contributions

W.T., Y.Y., and C.W. conceived the study. Experimental design was supervised by W.T., D.L., J.M., and M.W. Bioinformatic analyses were performed by D.L., X.C., O.K., and Y.Y. Plasmid construction and bacterial cultivation were carried out by W.T. and Y.Y. In vivo plasmid stability assays and qPCR experiments were conducted by Y.Y. and X.Z. Protein expression and purification were carried out by Y.Y. and Y.Z. Biochemical characterization of AcrIE10 was performed by W.T. and Y.Y. under the supervision of J.M. and M.W. The X-ray crystal structures of AcrIE10 and AcrIE10-IR, as well as the cryo-EM structure of the AcrIE10–Cas7* complex, were determined by W.T. with guidance from M.J. J.W., J.G., and Y.W. participated in X-ray crystal data analysis. In vitro validation of AcrIE10 structural models via mutational analysis was performed by W.T. and Y.Y., and in vivo validation was conducted by Y.Y., X.L., and C.W. W.T., Y.Y., C.W., and D.L.wrote the manuscript. J.M., A.I., and M.W. were in charge of the manuscript revisions. All authors approved the final draft of the manuscript.

Peer review

Peer review information

Nature Communications thanks Yue Feng, Hiroshi Nishimasu and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by the National Natural Science Foundation of China (grant numbers 82402671, W2512095, 81991531, and 32471347), Shanghai Municipal Science and Technology Commission (grant number 22410710300), and Russian Scientific Foundation grant 25-44-02137.

Data availability

The cryo-EM maps of the AcrIE10-Cas7*-DNA have been deposited in the Electron Microscopy Data Bank (https://www.ebi.ac.uk/pdbe/emdb/) under the accession numbers EMD-64090. The crystal structure of AcrIE10-DNA and AcrIE10-Cas7*-DNA has been deposited in the PDB database under the accession numbers 9M7T and 9UER, respectively. Structure of the AcrIE10-Cas7 complex under the accession number 8JI9 was referenced in the manuscript. All data generated or analyzed during this study are included in this manuscript. All other data supporting the findings of this study, including source data for figures, are available within the Article and its Supplementary Information files. Any additional requests can be directed to the corresponding author, Minggui Wang(mgwang@fudan.edu.cn) upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76353-2.