A bacterial-killing type IV secretion system as a hidden antifungal device

Abstract

Interkingdom competition between bacteria and fungi in nutrient-limited environments drives microbial community dynamics, yet the molecular mechanisms enabling bacterial dominance during bacterial-fungal interaction remain poorly characterized. This study reveals a paradigm-shifting role for bacterial type IV secretion system (T4SS) by demonstrating that a specialized bacterial-killing T4SS (T4SS_BK) in Lysobacter enzymogenes serves as a universal cross-kingdom antifungal weapon. Combining genetic, biochemical, and bioinformatical approaches, we show that T4SS_BK mediates direct contact-dependent suppression of phylogenetically diverse fungi—including unicellular yeasts and filamentous species—through delivery of a conserved deoxyribonuclease effector, LtdD. Structural characterization reveals that LtdD homologs are evolutionarily conserved across T4SS_BK-harboring bacterial lineages, suggesting widespread ecological deployment of this antifungal strategy. These findings overturn the classical dogma of T4SS_BK as an antibacterial apparatus and establish it as a multifunctional microbial combat system capable of trans-kingdom interference. By elucidating a DNA-targeting mechanism of fungal growth inhibition, our work not only redefines the ecological role of T4SS_BK in microbial warfare but also provides a molecular blueprint for engineering DNA-based antifungal therapies.

INTRODUCTION

Bacterial-fungal interactions (BFIs) are crucial in shaping microbial ecosystems, notably influencing nutrient cycling, disease dynamics, and overall ecosystem stability (1). Within these interactions, antagonism occurs as bacteria compete for limited resources by suppressing fungal growth. Historically, this suppression has been attributed to diffusible chemical compounds, including extracellular signaling molecules and secondary metabolites (2). For instance, the quorum-sensing signal homoserine lactones produced by Pseudomonas aeruginosa blocks hyphal development of Candida albicans by affecting adenosine 3′,5′-monophosphate (cAMP) signaling (3). Secondary metabolites such as lipopeptides secreted by Bacillus disrupt fungal membranes, while phenazines from Pseudomonas complicate interkingdom interactions through inducing hydroxyl radical production in fungi (4, 5). These findings have propelled the search for previously unknown antifungal agents, which are critical for combating fungal pathogens resistant to existing agrochemicals and medical antifungals.

In addition to chemical warfare, physical interactions—such as direct contact, biofilm formation, and secretion system-mediated attacks—play equally vital roles in BFIs (6). The type VI secretion system (T6SS), a contact-dependent antibacterial mechanism, has recently emerged as an antifungal device (7). Although T6SS typically mediates interbacterial competition (8), studies have shown its ability to deliver effectors that inhibit fungal DNA accumulation and disrupt metabolic pathways or plasma membrane, thereby suppressing fungal growth independently of diffusible antifungal metabolites (9). However, the absence of T6SS in many bacteria raises a critical question: What alternative bacterial protein secretion systems enable broad-spectrum antifungal competition?

The type IV secretion system (T4SS), a multifunctional molecular transporter present in many bacteria, represents a promising candidate. T4SS is traditionally known for its role in horizontal gene transfer (e.g., Agrobacterium tumefaciens T4SS transferring tumor-inducing DNA to plants) or effector delivery into host cells (e.g., Legionella pneumophila T4SS manipulating macrophages) (10, 11). Although several bacterial T4SS effectors (T4Es) have been reported to inhibit fungal growth when artificially expressed in yeast (12), experimental evidence of bacterial T4SS-mediated fungal inhibition has not yet been reported.

A specialized subgroup of T4SS, known as the widespread bacterial-killing T4SS (T4SS_BK), was initially identified in pathogens such as Xanthomonas citri and Stenotrophomonas maltophilia, serving as a contact-dependent antibacterial mechanism (13, 14). The T4SS_BK system delivers effectors containing the XVIPCD domain (Xanthomonas VirD4-interacting protein conserved domain) to lyse competitor bacteria, thereby aiding pathogens in manipulating the microbiome during infection. Translocation of these effectors requires their interaction with the T4SS_BK-specific adenosine triphosphatase (ATPase) VirD4 (15). Genomic surveys reveal the presence of T4SS_BK in beneficial bacteria, such as the plant-beneficial Lysobacter enzymogenes, which uses this system for interspecies bacterial competition while also producing the antifungal antibiotic known as heat-stable antifungal factor (HSAF or ningrongmycin) (16, 17). Although HSAF can inhibit various filamentous fungi and the yeast-like fungus C. albicans in vitro (18), it has been reported that Saccharomyces cerevisiae is not sensitive to HSAF (19). Despite the potential ecological versatility of T4SS_BK, this system has been solely regarded as an antibacterial tool—an established paradigm that warrants reconsideration.

Here, using L. enzymogenes as a model, we report unprecedented role of T4SS_BK in fungal suppression. We identify a T4E, LtdD (Lysobacter T4E potentially triggering DNA digestion), which degrades fungal DNA and is structurally and functionally conserved across T4SS_BK lineages. This work redefines T4SS_BK as a previously unidentified layer of interkingdom weaponry, offering alternative strategies to combat antifungal resistance.

RESULTS

L. enzymogenes inhibited the growth of unicellular fungi through T4SS_BK during BFIs

Our previous work demonstrated that L. enzymogenes OH11 inhibits growth of both yeast and filamentous fungi in a contact-dependent pattern. This antifungal activity was independent of the antifungal metabolite HSAF, but dependent on T6SS (20). Notably, the HSAF-T6SS double mutant still exhibited partial antifungal activity, suggesting that L. enzymogenes may use other contact-dependent antifungal mechanism (20). We propose that T4SS_BK could serve as one such mechanism, which has been previously overlooked in BFI. The genome of OH11 encodes an intact T4SS_BK gene cluster, and virD4, encoding a T4SS_BK-specific ATPase, is essential for T4SS_BK function by facilitating the translocation of effector proteins into the target cells (16, 21). We knocked out virD4 in an HSAF-deficient mutant background (ΔlafB) to generate a ΔlafBΔvirD4 double mutant. We first tested whether L. enzymogenes uses T4SS_BK to target unicellular model fungi, using the green fluorescent protein (GFP)–labeled S. cerevisiae W303 as a model. Bacterial and yeast cells were both normalized to an optical density of 600 nm (OD₆₀₀) of 1.0 and then cocultured at various volume ratios. Using fluorescence microscopy and viable S. cerevisiae cell counts, we found that ΔlafB could completely kill S. cerevisiae at a volume ratio of 1:1. In contrast, blockade of T4SS_BK by the virD4 mutation in ΔlafB reduced its antifungal activity against S. cerevisiae growth (Fig. 1, A and B), indicating that T4SS_BK played a key role in cross-kingdom inhibition of fungal growth by L. enzymogenes. The antifungal defect observed in ΔlafBΔvirD4 was rescued by virD4 complementation (ΔlafBΔvirD4-c) via a chromosomal knockin approach (Fig. 1, A and B). All the L. enzymogenes strains described above and in the following sections showed similar growth rates to the wild type (fig. S1).

Fig. 1. L. enzymogenes inhibits the growth of unicellular fungi through T4SS_BK.

(A) L. enzymogenes inhibited growth of S. cerevisiae in a T4SS_BK-dependent manner. Compared with ΔlafB, T4SS_BK inactivated mutant ΔlafBΔvirD4 exhibited reduced antifungal activity against S. cerevisiae. L. enzymogenes strains and S. cerevisiae strain W303 (GFP) were normalized to an OD₆₀₀ of 1.0, mixed in various volumetric ratios, and cultured for 24 hours. (B) The viable count of S. cerevisiae (L. enzymogenes:S. cerevisiae = 1:1) in (A) was quantified and expressed as log₁₀ colony-forming units (CFU)/ml. The dashed line represented the limit of detection (LOD; 3 log₁₀ CFU/ml). N.D., indicated not detected. (C) Representative images of S. cerevisiae budding (L. enzymogenes:S. cerevisiae = 1:1) in (A). Red arrows indicated the nascent buds (≤50% the size of the mother cell). (D) Percentage of cells with small bud in the budding cell population in (C) was quantified. (E) SEM analysis demonstrated that both ΔlafB and ΔlafBΔvirD4 attached on the surface of S. cerevisiae during BFI. (F) TEM analysis demonstrated that ΔlafB caused cell deformation of S. cerevisiae, while this deformation was partially restored during treatment by ΔlafBΔvirD4. Red arrows indicated cell deformation of S. cerevisiae. (G) Antagonism of L. enzymogenes to S. cerevisiae was restricted to the areas where their colonies were in direct contact. mCherry-labeled L. enzymogenes strains were cocultured in an overlap pattern with GFP-labeled S. cerevisiae. ΔlafB killed S. cerevisiae in the contact region, while ΔlafBΔvirD4 lost this ability. (H) Compared to ΔlafB, ΔlafBΔvirD4 showed reduced antifungal activity to C. albicans. (I) The viable count of C. albicans (L. enzymogenes:C. albicans = 5:1) in (H) was quantified and expressed as log₁₀ CFU/ml. The dashed line represented the LOD (3 log₁₀ CFU/ml).

To explore how T4SS_BK is involved in fungal inhibition by L. enzymogenes, we carried out microscopy examinations to observe the interaction between L. enzymogenes and S. cerevisiae. After 24-hour coculture, ΔlafB-treated S. cerevisiae cells showed defect in budding, that the ratio of small budding in budded cells was significantly increased. Under similar conditions, most budded cells of nontreated and ΔlafBΔvirD4-treated S. cerevisiae cells showed normal sizes (Fig. 1, C and D). By scanning electron microscopy (SEM), we found that both ΔlafB and ΔlafBΔvirD4 could attach on the surface of S. cerevisiae (Fig. 1E). Further analysis using transmission electronic microscopy (TEM) confirmed that the cells of S. cerevisiae were deformed after contacting with ΔlafB, while the cell deformity of ΔlafBΔvirD4-treated S. cerevisiae was recovered (Fig. 1F). Additionally, coculturing mCherry-labeled L. enzymogenes and GFP-labeled S. cerevisiae in an overlapping pattern demonstrated that ΔlafB killed S. cerevisiae in the overlapping zone, while ΔlafBΔvirD4 lost this activity (Fig. 1G). These results indicate that during intercellular contact, L. enzymogenes kills S. cerevisiae in a T4SS_BK-dependent manner.

We also investigated the T4SS_BK-mediated antifungal activity of L. enzymogenes using another unicellular fungus, the human pathogenic C. albicans Sc5314. The ΔlafBΔvirD4 mutant exhibited reduced antifungal activity against the yeast form of C. albicans compared to ΔlafB (Fig. 1, H and I). Together, these findings suggest that T4SS_BK in L. enzymogenes is involved in interkingdom antifungal activity against unicellular fungi.

T4SS_BK mediates the growth inhibition of filamentous fungi by L. enzymogenes

To explore whether T4SS_BK also targets filamentous fungi, we used Fusarium graminearum PH-1, a prominent model in molecular plant pathology (22), as a filamentous fungal system. Bacterial cells (normalized to an OD₆₀₀ of 1.0) and fungal conidia (2 × 10⁵ conidia/ml) were cocultured at a volume ratio of 1:1. The ΔlafBΔvirD4 mutant showed significant reduced antifungal activity compared with ΔlafB, but this defect was rescued by the T4SS_BK-complemented strain (Fig. 2, A and B).

Fig. 2. T4SS_BK mediates the antifungal activity of L. enzymogenes against filamentous fungi.

(A) L. enzymogenes inhibited growth of F. graminearum in a T4SS_BK- dependent pattern. Bacterial cells adjusted to OD₆₀₀ of 1.0 were mixed with F. graminearum conidia (2 × 10⁵ conidia/ml) at a 1:1 volume ratio and cultured on PDA medium. ΔlafBΔvirD4 exhibited significantly reduced inhibitory activity to F. graminearum compared with ΔlafB. WT, wild type. (B) The fungal colony sizes quantified from (A). (C) Elongation of F. graminearum germinating hyphae was inhibited by ΔlafB, whereas the inhibitory ability of ΔlafBΔvirD4 was significantly diminished. The hyphae were stained by calcofluor white to enhance visibility. (D) Quantification of hyphae length from (C). (E) SEM analysis demonstrated that ΔlafB attachment resulted in partial hyphal degradation of F. graminearum, while this defect was less pronounced in the ΔlafBΔvirD4. Red arrows indicated the partial hyphal degradation. (F) TEM analysis demonstrated that F. graminearum cell wall and plasma membrane were damaged upon contacting with ΔlafB, while this deformation was restored during interactions with ΔlafBΔvirD4. Red arrows indicated the damage of cell wall and cell membrane. (G) L. enzymogenes inhibited fungal biomass in natural soil via T4SS_BK. ΔlafB and ΔlafBΔvirD4 were individually inoculated into the natural soil planted with banana for 7 days. The relative biomass of total fungi and bacteria was measured by qPCR using total soil DNA. (H) PCoA of microbiome sequencing indicated that the fungal communities derived from the ΔlafB-treated soil were clustered separately from those derived from the control (LB-treated soil) and the ΔlafBΔvirD4-treated soil. (I) L. enzymogenes inhibited growth of Trichoderma sp. XJ-A, which was isolated from the untreated natural banana field soil, in a T4SS_BK-dependent pattern. (J) The fungal colony sizes quantified from (I).

Microscopic observations showed that the elongation of F. graminearum hyphae during germination was inhibited by ΔlafB, but this inhibition was significantly reduced in the ΔlafBΔvirD4 mutant (Fig. 2, C and D). SEM revealed that ΔlafB could contact the F. graminearum surface and induce partial hyphal degradation after conidia gemination, whereas ΔlafBΔvirD4 failed to elicit such effects (Fig. 2E). Further analysis with TEM showed that the cell wall and plasma membrane of F. graminearum were damaged upon contact with ΔlafB, while cell deformation in ΔlafBΔvirD4-treated F. graminearum was partially restored (Fig. 2F).

We further tested whether L. enzymogenes could inhibit mycelial growth by soaking hyphal plugs in bacterial suspensions and then inoculating them onto agar plates. The wild-type strain OH11 almost completely inhibited hyphal growth, whereas the HSAF-deficient mutant ΔlafB caused only about 15% inhibition, as reflected by fungal colony size. The ΔlafBΔvirD4 double mutant exhibited no significant inhibition of fungal growth (fig. S2), indicating that T4SS_BK has limited efficacy against mature hyphae. Therefore, we focused on the more visible phenotype—fungal inhibition from conidia germination to hyphal development—in subsequent experiments.

To investigate whether L. enzymogenes OH11, isolated from soil (23), can shape the soil fungal microbiome in the field through T4SS_BK, independent of the known antifungal metabolite HSAF, we inoculated the ΔlafB and ΔlafBΔvirD4 mutants into natural soil planted with bananas. After 7 days, total fungal biomass was reduced in the ΔlafB-treated soil, while it increased in the ΔlafBΔvirD4-treated soil (Fig. 2G). Microbiome sequencing and principal coordinates analysis (PCoA), based on Bray-Curtis dissimilarity, showed that the fungal communities from ΔlafB-treated soil clustered separately from those in LB-treated and ΔlafBΔvirD4-treated soils (Fig. 2H), indicating that L. enzymogenes could shape the fungal community in the field via T4SS_BK. Several fungi were isolated from the untreated banana soil samples, including a Trichoderma sp. strain XJ-A (table S1). Coculturing Trichoderma conidia with L. enzymogenes strains revealed that L. enzymogenes could inhibit the growth of Trichoderma sp. XJ-A in a T4SS_BK-dependent manner (Fig. 2, I and J). These findings demonstrate that L. enzymogenes can target filamentous fungi and reshape the plant rhizosphere fungal community through T4SS_BK.

T4SS_BK-mediated antifungal functions independently of T6SS

RNA sequencing (RNA-seq) analysis identified 650 up-regulated and 605 down-regulated genes in F. graminearum conidia treated with ΔlafBΔvirD4 compared to those treated with ΔlafB (table S2). In L. enzymogenes, the T6SS also mediates contact-dependent antifungal activity. We asked whether the T4SS-mediated antifungal effect is an indirect result of impaired T6SS activity due to regulatory cross-talk. To address this, we compared the T4SS_BK-regulated genes to those previously identified in F. graminearum conidia in response to the T6SS of L. enzymogenes (20). We found 410 up-regulated and 431 down-regulated genes specific to T4SS_BK (Fig. 3A), suggesting that the antifungal mechanisms of T4SS_BK differ from those of T6SS. Gene ontology (GO) enrichment analysis further revealed that the T4SS_BK-specific responsive genes in F. graminearum were primarily involved in carbohydrate metabolic process, detoxification, and amino acid metabolism (specifically threonine, leucine, and lysine) (Fig. 3B). These results indicate that T4SS_BK-mediated antifungal activity appears to be independent of T6SS in L. enzymogenes.

Fig. 3. T4SS_BK mediated antifungal activity operates independent of T6SS.

(A) A Venn diagram illustrated the differentially expressed genes (DEGs) of F. graminearum treated with the ΔlafBΔvirD4 and ΔlafBΔtssM mutants, compared to treatment with the ΔlafB strain; ΔlafB, the HSAF-deficient mutant of L. enzymogenes; ΔT4SS indicated ΔlafBΔvirD4, a mutant with in-frame deletion of virD4, resulting in the inactivation of the T4SS_BK in the ΔlafB background; ΔT6SS indicated ΔlafBΔtssM, a mutant with in-frame deletion of tssM, resulting in the inactivation of the T6SS in the ΔlafB background. (B) The GO enrichment analysis displays DEGs in F. graminearum in response to T4SS_BK and T6SS. The GO terms with P values <0.05 are shown. (C) T4SS_BK inactivation does not affect the secretion of the inner tube Hcp of the T6SS in L. enzymogenes. RNAP (RNA polymerase beta) acted as an intracellular control. (D) Compared to ΔlafBΔvirD4 and ΔlafBΔtssM, ΔlafBΔvirD4ΔtssM exhibited reduced antifungal activity against S. cerevisiae in mixed colonies. ΔlafBΔvirD4ΔtssM is a mutant of the T4SS_BK and T6SS in the ΔlafB background. (E) The viable count of S. cerevisiae (L. enzymogenes:S. cerevisiae = 1:3) in (D) was quantified and expressed as log₁₀ CFU/ml. The dashed line represented the LOD (3 log₁₀ CFU/ml). (F) Compared to ΔlafBΔvirD4 and ΔlafBΔtssM, ΔlafBΔvirD4ΔtssM showed reduced antifungal activity against F. graminearum. (G) The fungal colony sizes quantified from (F).

To test this hypothesis, we conducted a series of assays. The first assay involved the secretion of the inner tube protein Hcp in the T4SS_BK-defective mutant, as Hcp secretion is the hallmark of functional T6SS (24). For this purpose, the L. enzymogenes strains containing an hcp-FLAG overexpression vector were cultured in LB medium, and then, the proteins of the culture supernatants and bacterial cells were subsequently collected. Western blot analysis revealed that the cellular Hcp-FLAG protein was detected in all the strains; however, Hcp-FLAG secretion was observed only in the ΔlafB and ΔlafBΔvirD4 mutants, but not the T6SS-defective mutant, ΔlafBΔtssM (Fig. 3C). This result indicates that the blockage of T4SS_BK via the virD4 mutation did not affect T6SS function, supporting the notion that T4SS_BK-mediated antifungal activity is independent of T6SS. The second assay involved genetic validation. We knocked out tssM in ΔlafBΔvirD4 background to create a T4SS_BK-T6SS double mutant, designated as ΔlafBΔvirD4ΔtssM. Coculture assays demonstrated that this double mutant exhibited further reduced antifungal activity to both S. cerevisiae (Fig. 3, D and E) and F. graminearum (Fig. 3, F and G) compared to the ΔlafBΔvirD4 mutant alone. These results indicate that T4SS_BK may function largely independently of T6SS in fungal inhibition.

L. enzymogenes is capable of translocating an antifungal T4SS_BK effector, LtdD, into fungal cells

Considering that T4SS_BK functions in interbacterial competition by translocating antibacterial effector proteins (T4Es) into prey cells (25), we hypothesized that L. enzymogenes might translocate toxic T4Es into fungal cells to inhibit fungal growth. In our previous work, we identified 16 T4Es containing the XVIPCD domain in the L. enzymogenes genome (16). Using an inducible expression system, we expressed these effector genes in S. cerevisiae and found that one of them, Le4579, exhibited antifungal activity against S. cerevisiae similar to that of the characterized T6SS effector Le1893 from L. enzymogenes (Fig. 4A and fig. S3) (20). Microscopic observations with TEM showed that in vivo expression of Le4579 resulted in morphological abnormalities in S. cerevisiae cells (Fig. 4B). We knocked out Le4579 in the ΔlafB background and found that this double mutant (ΔlafBΔLe4579), under coculture conditions, exhibited reduced antifungal activity to both S. cerevisiae and F. graminearum (Fig. 4, C to F), suggesting that Le4579 was a key effector for T4SS_BK-mediated antifungal function in L. enzymogenes.

Fig. 4. L. enzymogenes encodes an antifungal T4SS_BK effector.

(A) Le4579 displayed fungal toxicity when expressed in S. cerevisiae. S. cerevisiae strain W303 harboring different effectors were diluted to a series of optical densities (ODs) and cultured for 48 hours on noninducing (glucose) or inducing (galactose) medium. The EV and the characterized antifungal effector Le1893 (20) served as negative and positive controls, respectively. (B) TEM observation revealed that expression of Le4579 induced deformation of yeast cells. (C) Compared to ΔlafB, ΔlafBΔLe4579 exhibited partially reduced antifungal activity against S. cerevisiae in mixed colonies. ΔlafBΔLe4579 is a mutant of the Le4579 in the ΔlafB background. (D) The viable count of S. cerevisiae (L. enzymogenes:S. cerevisiae = 1:1) in (C) was quantified and expressed as log₁₀ CFU/ml. The dashed line represented the LOD (3 log₁₀ CFU/ml). (E) Compared to ΔlafB, ΔlafBΔLe4579 showed reduced antifungal activity against F. graminearum. (F) The fungal colony sizes quantified from (E). (G) Structure alignment of Le4579 (green) with the XVIPCD domain of the reported structure of X-Tfe (gray, 7mu9), a T4SS_BK effector from Xanthomonas campestris (13), was performed. (H) Structure alignment of the N-terminal region of Le4579 (1 to 394 amino acids, green) with that of UZH43848.1 (1 to 367 amino acids, purple), the latter of which contains an HNH/ENDO VII nuclease superfamily domain. (I) Predicted domains of Le4579 and truncated mutants used in this study. (J) The induced fungal toxicity of Le4579 in S. cerevisiae required the presence of the HNH/ENDO VII domain.

The predicted structure of Le4579, obtained from AlphaFold DB (26), revealed that its C-terminal region shared the highest similarity with the conserved XVIPCD domain of X-Tfe, a T4E from X. citri (Fig. 4G) (13). However, the N-terminal structure of Le4579 did not resemble reported T4Es. We identified homologs of Le4579 in the National Center for Biotechnology Information (NCBI) and found that most contained an HNH/ENDO VII nuclease superfamily domain or a domain of unknown function (DUF6696). Structure alignment showed that the N-terminal region of Le4579 was similar to proteins with an HNH/ENDO VII domain (Fig. 4H). The HNH/ENDO VII nuclease superfamily, also known as the His-Me finger endonuclease superfamily, includes 38 distinct families sharing limited sequence homology but exhibiting a highly similar catalytic motif with a ββα topology (27, 28). Several bacterial T6SS effectors belonging to this superfamily could cause DNA degradation and growth inhibition during heterologous expression in bacteria (29). Since Le4579 contains an N-terminal HNH/ENDO VII domain likely involved in its toxic function and a C-terminal XVIPCD domain involved in translocation, we designed truncated mutants of Le4579 for further study (Fig. 4I). We expressed truncated Le4579 in S. cerevisiae and found that its antifungal activity was dependent on the HNH/ENDO VII domain (Fig. 4J). We thus reasonably designated Le4579 as LtdD and use it in the following contexts. Through bacterial two-hybrid and pull-down assays, we found that LtdD interacted with the intracellular domain of VirD4, and this interaction depends on the XVIPCD domain of LtdD (Fig. 5A and fig. S4), supporting the finding LtdD is a T4E.

Fig. 5. The T4SS_BK effector LtdD inhibits fungal growth by digesting fungal DNA.

(A) Pull-down assay reveals the full-length LtdD directly interacted with the cytoplasmic domain of VirD4 (VirD4-His), and the XVIPCD domain of LtdD is required for this interaction. (B) Schematic diagram of the LtdD translocation assay. L. enzymogenes OH11 derivative strains carrying the Cre-LtdD(M3) hybrid protein served as a donor strains, and S. cerevisiae strain W303 harboring a pGADT7 carrying the loxP-ura3-loxP cassette is a recipient strain. Created in BioRender: L. Lin (2026); https://BioRender.com/c38ol7i. (C) LtdD translocation depended on T4SS_BK. After coculture, the donor strains and recipient strain were cultured and smeared on SD-Leu plates containing 5-FOA (0.1 mg/ml). ΔlafB and ΔlafBΔtssM, but not ΔlafBΔvirD4, were able to translocate a truncated version of LtdD (M3), which lacks antifungal activity, into S. cerevisiae. Cre recombination of ura3 was confirmed by PCR. (D) The data demonstrate the efficiency of translocation, calculated as the ratio of recombined recipient cells to total recipient cells. (E) LtdD exhibited in vitro DNase activity to genomic DNA of S. cerevisiae and F. graminearum, which was dependent on HNH/ENDO VII domain. (F) Representative PI signal peaks of S. cerevisiae expressing LtdD and its truncated M3. S. cerevisiae strains were stained with PI, and their relative DNA abundance was analyzed with flow cytometry. (G) PI signal strength quantified from (F). (H) Plasmid-based expression of LtdD induced DNA damage in S. cerevisiae. DNA damage was assessed by Western blot analysis using the anti-γH2A antibody. The anti-H2B antibody served as a loading control. (H) LtdD digested essential genes of S. cerevisiae in vitro. Act1 was the actin gene. lcp5, nop7, utp18p, rrp14, and ade13 are involved in DNA or RNA biosynthesis, while gls2, pkc1, and chs3 are related to cell wall biosynthesis.

Next, we used a modified Cre/loxP system (30, 31) to test whether LtdD could be naturally translocated from L. enzymogenes to fungi, using the yeast S. cerevisiae as a model organism to facilitate this investigation. To achieve this, we constructed a pGADT7 vector designed to express the ura3 gene, with loxP sites introduced on both sides of the ura3 gene. The yeast strain containing the loxP-ura3-loxP cassette served as the recipient strain. Given that the full-length LtdD exhibited strong antifungal activity, we opted to use a truncated version of LtdD, retaining only the XVIPCD domain (designated as M3 in Fig. 4I), to investigate effector translocation. We constructed a pBBR-MCS5 vector expressing Cre-M3 and subsequently transformed it into the ΔlafB, ΔlafBΔvirD4, and ΔlafBΔtssM strains to create the donor strains. The donor strains and the recipient strain were cocultured on yeast extract, peptone, and dextrose (YPD) plates. Upon successful translocation of Cre-M3 to the recipient strain, the loxP element could be specifically cleaved, resulting in the removal of the ura3 gene. This cleavage enabled the growth of the recombined recipient strain under selective conditions supplemented with 5-FOA (Fig. 5B). Using this strategy, we found that the ΔlafB and ΔlafBΔtssM derivative donor strains, but not the ΔlafBΔvirD4 strain, were capable of translocating Cre-M3 into the recipient strain (Fig. 5, C and D), with DNA sizes observed ranging from the noncleaved fragment of 1556 base pairs (bp) to the cleaved fragment of 502 bp (Fig. 5C). These results demonstrate that L. enzymogenes can naturally translocate LtdD into fungal cells via the T4SS_BK system (Fig. 5D).

Translocated LtdD inhibited fungal growth through digesting DNA

To investigate whether LtdD exhibits deoxyribonuclease (DNase) activity, we purified both full-length LtdD and a truncated mutant lacking the HNH/ENDO VII domain (known as M3). Incubation of these proteins with various fungal genomic DNA, including that from unicellular yeasts and filamentous species, revealed that LtdD effectively degraded all tested fungal DNA. Furthermore, its DNase activity was found to be dependent on the presence of the HNH/ENDO VII domain (Fig. 5E and fig. S5). These results indicate that LtdD, following translocation, inhibits fungal growth by digesting genomic DNA.

To further assess whether LtdD also degrades DNA within fungal cells, we analyzed the DNA content of the yeast S. cerevisiae expressing LtdD, M3, and an empty vector control (EV) using flow cytometry. The propidium iodide (PI) signal in the LtdD-expressing strain was significantly reduced compared to the EV and M3-expressing strains (Fig. 5F), suggesting that LtdD functions as a DNase in vivo, targeting the yeast genome. Additionally, Western blot analysis using an anti-γH2A antibody—a well-established marker for DNA damage in eukaryotes (32)—revealed that LtdD, but not M3, induced DNA damage in S. cerevisiae cells (Fig. 5G). These findings suggest that LtdD may degrade genomic DNA in fungal cells, leading to DNA damage and subsequent growth inhibition. We cloned several essential yeast genes involved in DNA/RNA biosynthesis or cell wall biosynthesis, including lcp5, nop7, utp18p, rrp14, ade13, gls2, pkc1, and chs3 (33–40), and found that LtdD also digested these specific essential genes (Fig. 5H). Collectively, these results demonstrate that L. enzymogenes can translocate the DNase effector LtdD into fungal cells, triggering degradation of essential gene DNA and resulting in fungal growth inhibition. Although LtdD can digest Escherichia coli genomic DNA in vitro (fig. S6), it neither inhibited E. coli growth during heterologous expression nor contributed to T4SS_BK-mediated antibacterial activity against E. coli in coculture assays (fig. S6). These results suggest that LtdD likely targets fungi specifically.

LtdD homologs also exhibited coordinated antifungal and DNase activity

Given the widespread presence of T4SS_BK in bacteria (15), we next investigated whether the antifungal activity mediated by LtdD also occurs in other bacteria. We predicted T4Es (XVIPCD domain-containing proteins) by performing a hidden Markov model (HMM) search (41) against the total protein datasets of Lysobacter strains we collected, as well as previously reported T4SS_BK-encoding bacterial strains, using the XVIPCD domain sequences of 13 X. citri XVIPCD proteins as queries (13, 15). We then predicted the structures of these effectors using ColabFold and used US-align to identify protein structurally similar to LtdD (42, 43). We also incorporated members from the AFDB cluster containing LtdD due to their structural similarity (44). In total, we identified 57 proteins with a similar structure from the Pseudomonadota phylum, including 45 proteins from Lysobacter and 8 proteins from Pseudoxanthomonas (Fig. 6A, fig. S7, and table S3). We cloned La1212 from biocontrol bacterium Lysobacter antibioticus OH13 (LAOH13_1212) (45) and artificially synthesized Pw2805 (A0A4Q8LHM9) from human pathogenic bacterium Pseudoxanthomonas winnipegensis NML171200 (46), both of which showed high structural similarity to LtdD (Fig. 6, B and C). Upon heterologous expression, we found that both proteins could inhibit S. cerevisiae growth in vivo (Fig. 6D). The purified La1212 and Pw2805 proteins also exhibited DNase activities in vitro (Fig. 6E). These results indicate that LtdD homologs with DNase and antifungal activity are widely spread among bacteria in the phylum Pseudomonadota.

Fig. 6. LtdD homologs exhibit coordinated antifungal and DNase activities.

(A) 57 T4SS_BK effectors from Pseudomonadota showed similar structure with LtdD. The total proteins of Lysobacter and other T4SS_BK-encoding bacteria were used to predict T4SS_BK effectors for further structure analyses. The structures of these predicted T4SEs were predicted by Colabfold, and the structural similarity with LtdD of above T4SEs and T4SEs from AFDB cluster was analyzed using US-align. (B) Predicted domains of La1212 from L. antibioticus OH13 and structure alignment of La1212 (orange) with LtdD (green). (C) Predicted domains of Pw2805 from Physella winnipegensis NML171200 and structure alignment of Pw2805 (blue) with LtdD (green). (D) La1212 and Pw2805 exhibited fungal toxicity when expressed in S. cerevisiae. (E) In vitro DNase activity assay showed that La1212 and Pw2805 could digest S. cerevisiae genomic DNA. (F) A working model illustrates T4SS_BK-mediated cross-kingdom fungal inhibition by L. enzymogenes. L. enzymogenes can inhibit both unicellular and filamentous fungi through T4SS_BK during BFI. L. enzymogenes encodes a series of T4SS_BK effectors, including LtdD, which has DNase activity. During intercellular contact with fungi, LtdD is likely translocated into fungal cells via T4SS_BK. Once inside the fungal cells, LtdD inhibits fungal growth by degrading fungal genomic DNA. Created in BioRender: L. Lin (2026); https://BioRender.com/c38ol7i.

DISCUSSION

Our study establishes T4SS_BK as a versatile molecular weapon capable of transcending kingdom boundaries, highlighting its role as a widespread bacterial protein secretion system in driving BFIs. The discovery of LtdD, a DNase effector deployed by L. enzymogenes to degrade fungal DNA, expands the functional repertoire of T4SS_BK beyond its traditional role in antibacterial competition. The structural conservation of LtdD homologs and their validated antifungal activity across various bacterial lineages underscore a widespread evolutionary strategy, positioning T4SS_BK as a universal mediator of bacterial-fungal antagonism (Fig. 6F). Our findings not only highlight the ecological significance of T4SS_BK, but also propose that bacterial interactions with fungi may extend beyond mechanisms predominantly attributed to diffusible chemical compounds.

While the antifungal activities of T4SS_BK are noteworthy, it is essential to contextualize these findings within the broader landscape of bacterial secretion system–mediated antifungal strategies. Various bacteria use the type II secretion system to secrete extracellular lytic enzymes that disrupt the fungal cell wall (47). These enzymes act without direct contact between bacteria and fungi, as they do not require entry into fungal cells. In contrast, the T6SS necessitates direct contact to inhibit fungal growth through the injection of toxic effectors. These effectors function by disrupting cellular integrity or interfering with metabolic pathways (9). Here, we show that the role of T4SS_BK as an antifungal weapon is characterized by its ability to target phylogenetically diverse fungi through the degradation of fungal DNA via the effector LtdD. The structural conservation of LtdD across T4SS_BK-harboring lineages suggests that this antifungal strategy may be evolutionarily conserved, allowing bacteria such as L. enzymogenes to thrive in environments rich in fungi.

LtdD could degrade fungal DNA substrates in vitro without apparent sequence specificity, suggesting that LtdD-mediated genomic DNA degradation is a key mechanism underlying the defects in fungal growth and cell morphology caused by T4SS_BK. On the other hand, the dual functionality of LtdD—nonspecific DNase activity against both fungal and bacterial (e.g., E. coli) DNA in vitro, yet selective cytotoxicity toward fungi but not E. coli—raises important questions about bacterial protection mechanisms. We propose that E. coli may have unknown protective mechanisms that block the DNase activity of LtdD, as previous studies have demonstrated immunity-independent protection mediated by envelope stress responses in E. coli against the Vibrio cholerae T6SS effector TseH (48).

Notably, effector-driven BFI mechanisms are not limited to LtdD from L. enzymogenes; similar phenomena have also been observed in other bacteria, including the recently reported Acidovorax citrulli, a plant-pathogenic bacterium, and Salmonella enterica, a human pathogenic bacterium. The former uses a DNase effector, TseN, which exhibits broad-spectrum antifungal activity (49), while the latter delivers the effector protein SopB via a type III secretion system (T3SS) into the fungus C. albicans to manipulate fungal metabolism, thereby enhancing bacterial virulence (50). These results uncover that bacteria use at least three specialized protein secretion systems (T4SS_BK, T6SS, and T3SS) to deliver diverse effector proteins that interact with fungi, resulting in differential outputs. This highlights a conserved BFI pattern mediated by bacterial specialized protein secretion systems across both biocontrol and pathogenic bacteria.

Our findings, in conjunction with previous research, reveal at least four distinct strategies used by L. enzymogenes in its interactions with fungi. First, the production of the antifungal antibiotic HSAF (ningrongmycin) enables the bacterium to exert antifungal effects without requiring direct cell-to-cell contact, thereby enhancing its ecological dominance (51, 52). Second, antifungal protein toxins are secreted through the flagellar type III secretion system, demonstrating the bacterium’s ability to use various secretion pathways despite being nonflagellated (53). Third, our recent study showed that T6SS can protect crop plants from filamentous fungal infections by potentially injecting Rhs-like fungal toxin effectors (18). Last, the T4SS_BK, as elucidated in this study, represents a previously unknown antifungal strategy that further underscores the remarkable adaptive capabilities of L. enzymogenes. Collectively, these four mechanisms illustrate the diverse arsenal of antifungal strategies used by L. enzymogenes, positioning it as a formidable player in microbial ecosystems and a potential key to innovative solutions in agricultural biotechnology.

In conclusion, our findings reveal a previously unrecognized layer of microbial warfare, positioning T4SS_BK as a significant player in the continual battle between bacteria and fungi. From an application perspective, LtdD and its homologs represent promising candidates for engineered biocontrol agents. Their targeted DNA degradation mechanism could circumvent fungal resistance to conventional antifungals, offering sustainable solutions for crop protection and clinical antifungal therapies. Future research should explore effector delivery mechanisms, host specificity, and ecological tradeoffs to fully harness the potential of this system.

MATERIALS AND METHODS

Strains and growth conditions

The details of the bacterial and fungal strains used in this study are presented in table S4. Unless otherwise specified, bacterial strains were grown in LB medium at 28°C, while S. cerevisiae and C. albicans were cultured in YPD medium at the same temperature. Filamentous fungi were cultivated on PDA medium at 25°C. The conidia of F. graminearum were induced by inoculating five 5-mm hyphal plugs in a 50-ml flask containing 20 ml of carboxymethylcellulose (CMC) medium at 25°C for 3 days (54). The conidia of Trichoderma sp. were induced by inoculating on PDA for 7 days at 25°C.

Genetic approaches

The plasmids and primers used in this study are listed in tables S5 and S6, respectively. The in-frame deletion of virD4 in the ΔlafB strain was achieved using a double-crossover homologous recombination technique, as previously described (16). Briefly, the two flanking regions of virD4 were amplified by polymerase chain reaction (PCR) and cloned into the suicide vector pEX18Gm, which was then transformed into the ΔlafB strain via electroporation. Single-crossover recombinants were selected on LB plates supplemented with kanamycin (25 μg/ml) and gentamicin (150 μg/ml). The transformants were subsequently cultivated on LB plates containing 10% (w/v) sucrose and kanamycin (100 μg/ml) to select for double crossovers. The resulting virD4 mutants were verified by PCR using the corresponding primers. To complement the virD4 gene using a chromosomal knockin approach, a fragment containing the two flanking regions of virD4 along with the entire virD4 genes was inserted into pEX18Gm and transformed into the ΔlafBΔvirD4 double mutant by electroporation. The complementation strains were verified using the same method for the virD4 mutants. The in-frame deletion and complementation of other genes was achieved with the same method as virD4.

Antifungal activity assay of L. enzymogenes

To investigate contact-dependent inhibition to growth of S. cerevisiae and C. albicans, GFP-labeled W303 was grown overnight at 28°C in SC-U (SC minimal medium lacking uracil) supplemented with 2% glucose, while C. albicans was cultivated overnight at 28°C in YPD. Yeast and Lysobacter cells were collected by centrifugation (6000 rpm for 3 min at room temperature), washed, and diluted with sterilized ddH₂O to an OD₆₀₀ = 1.0. The yeast (S. cerevisiae or C. albicans) and Lysobacter cells were then mixed in volume ratios of 1:15, 1:5, 1:1, 5:1, and 15:1. Following this, 2 μl of the mixed culture was spotted onto YPD plates and incubated at 28°C for 24 hours before observation. A FUSION FX Spectra (Vilber Fusion Fx7) was used to observe GFP signal of S. cerevisiae. Additionally, The GFP-labeled S. cerevisiae and mCherry-labeled Lysobacter cells were inoculated in an overlapping pattern to determine whether the antifungal activity was dependent on intercellular contact.

To investigate the antifungal activity against conidia of F. graminearum and Trichoderma sp., L. enzymogenes cells washed by ddH₂O were diluted to OD₆₀₀ of 1.0. Fungal conidia were induced and filtered through two layers of Mirocloth to remove mycelia. The conidia were then diluted to 200 conidia/μl in sterilized ddH₂O. Subsequently, the bacterial cells and fungal conidia were mixed in a 1:1 volume ratio and spotted (2 μl) onto PDA medium. The cultures were incubated at 25°C for 48 hours before imaging. To evaluate inhibitory activities against mature mycelia of F. graminearum, 5-mm-diameter hyphal plugs were soaked in 5 ml of L. enzymogenes culture (OD₆₀₀ = 1.0) and then incubated at 28°C for 1 hour with shaking at 150 rpm. After that, the plugs were transferred and inoculated onto PDA for 3 days before imaging.

Microscope observation

To investigate the inhibition process of F. graminearum, mixtures of L. enzymogenes cells (OD₆₀₀ = 1.0) and F. graminearum conidia (10⁵ conidia/ml) at a 1:1 volume ratio were inoculated into 24-well cell culture plates. Then, 200 μl of liquid PDB was added, and the plates were statically incubated to mimic intercellular contact. The samples were cultured for 24 hours and examined using a fluorescence microscope (Zeiss Axio Observer 3). The culture was stained with calcofluor white (CFW; 1 μg/ml) for 5 min to highlight the hyphae.

For SEM observations of the interaction between L. enzymogenes and fungi, normalized cells of L. enzymogenes and fungi were co-inoculated onto 25-μm nylon membranes on YPD (for S. cerevisiae) or PDA (for F. graminearum). Following a 24-hour culture, the nylon membranes were collected and fixed in phosphate-buffered saline (PBS) containing 3% glutaraldehyde and 1% paraformaldehyde. The samples were then observed according to previously established protocols (55). In brief, the samples were fixed in 2.5% glutaraldehyde at 4°C for 12 hours, rinsed three times with 0.1 M phosphate buffer (pH 7.0), dehydrated through a graded ethanol series (30 to 100%), dried using a critical point dryer, mounted onto aluminum stubs with conductive carbon tape, sputter-coated with 10 nm of gold/platinum, and observed with a Hitachi Regulus 8100 SEM.

TEM was used to examine the detailed morphology of fungi. For the interaction between fungi and L. enzymogenes, the fungal and bacterial mixtures were cultured for 24 hours in liquid YPD (for S. cerevisiae) or PDA (for F. graminearum) and subsequently collected by centrifugation (6000 rpm for 3 min at 4°C). For S. cerevisiae expressing either the EV or LetdD, individual transformant was grown overnight at 28°C in SC-U supplemented with 2% glucose. Yeast cells were washed with sterilized ddH₂O and then diluted in SC-U medium supplemented with 2% galactose to induce effector expression. The yeast samples were collected by centrifugation (6000 rpm for 3 min at 4°C) after 6 hours. The samples were fixed in PBS containing 3% glutaraldehyde and 1% paraformaldehyde and subsequently observed following established procedures (56). Briefly, the samples were rinsed three times with SPB (0.1 M, pH 7.0), dehydrated in a graded ethanol series (30 to 95%), and polymerized overnight in Lowicryl K4M embedding medium. The blocks were then cut into 70-nm sections using an ultramicrotome, stained with lead citrate and uranyl acetate for 5 min, and observed with a Hitachi H-7800 TEM.

Transcriptome analysis

For RNA-seq sampling, bacterial cells of the L. enzymogenes strain were washed and diluted to OD₆₀₀ of 1.0. The conidia of F. graminearum were diluted to 1000 conidia/μl using sterilized ddH₂O. L. enzymogenes cells and conidia were mixed at a 1:1 volume ratio and spotted onto a 0.22-μM polyvinylidene difluoride filter membrane on PDA medium. Following 24 hours of coculturing, the mixtures were collected together with the filter. Three replicates of each sample were then sent to BGI Genomics Corporation for RNA-seq. Fungal RNA was extracted using the TRIzol method in accordance with the manufacturer’s protocol. RNA-seq was performed using the DNBseq platform along with a 100-bp paired-end module. Raw reads were filtered for subsequent analysis by removing reads containing adapter, poly-N, or low-quality reads. The clean reads were aligned to the genome of F. graminearum PH-1 (EnsemblFungi RR1 gene annotation RR1.41) using Tophat (ccb.jhu.edu). The mapped reads were quantified with the Cufflinks program (https://cole-trapnell-lab.github.io/projects/cufflinks), and transcript levels for each gene were calculated as RPKM (reads per kilobase transcript length per million reads mapped). Differentially expressed genes (DEGs) were identified using featureCounts software (bioinf.wehi.edu.au), and the log 2 fold change (log2FC) values along with adjusted P values were determined using DESeq2 software (www.bioconductor.org). Genes were classified as differentially expressed if they exhibited an adjusted P value of <0.05 and an absolute log2FC ≥ 1. GO analysis of the DEGs was conducted using the GOseq R package (www.bioconductor.org), and GO terms were designated as significantly enriched if they had an adjusted P value of <0.05.

Microbiome analysis

To investigate whether L. enzymogenes influenced fungal community composition in the field, we collected soil samples from a banana plantation and introduced fresh banana plants into these samples to activate the soil microbiome. After 7 days, cultures of ΔlafB and ΔlafBΔvirD4 were introduced to the soil samples, with LB serving as a negative control. Subsequently, the DNA from rhizosphere soil was extracted 7 days after root irrigation using the QIAGEN DNeasy PowerSoil Kit (QIAGEN, Germany). The PCR amplicons were pooled in equimolar ratios and sequenced by Magigene Biotechnology Co., Ltd. on an Illumina Nova 6000 platform. Raw sequences were assembled with a minimum overlap of 35 bp between forward and reverse reads, allowing for fewer than two mismatches. These sequences were then subjected to quality control (quality score 30), as well as dereplication and removal of singleton, chimeric, mitochondrial, and chloroplast sequences. The remaining sequences were classified into operational taxonomic units (OTUs) with 97% similarity and subsequently assigned taxonomy using the UNITE database for the ITS ribosomal RNA genes. A Bray-Curtis distance-based PCoA was conducted on the rarefied OTU table using the vegan R package (https://cran.r-project.org/web/packages/vegan/index.html). The relative biomass of fungi and bacteria was measured by quantitative PCR (qPCR) using the BD-ITS1F/BD-ITS2 and L-V3/R-V3 primers. Reverse transcription qPCR was performed in a reaction mixture of 20 ml of SYBR Premix ExTaq (TaKaRa Bio Inc.) following the manufacturer’s protocol.

Isolation of fungi from banana field soil

Soil samples (2 g) were diluted 10-fold with sterilized ddH₂O (20 ml). The resulting mixtures were incubated on a shaker at 25°C for 24 hours. Subsequently, the mixtures were further diluted 10 to 1000 folds with sterilized ddH₂O and smeared on PDA plates. Individual fungal colony was isolated after 5 days. For molecular identification, the genomic DNA was isolated, and the ITS region of each strain was amplified with corresponding primers. The amplified DNA fragments were purified, sequenced, and compared to sequences available in the NCBI database using the BLASTN program for species identification.

Secretion assay of the T6SS inner-tube Hcp

L. enzymogenes strains harboring plasmid pBBR-Hcp-FLAG were grown in 40 ml of LB until the OD₆₀₀ reached 0.6. Culture supernatants were separated from cells by centrifugation at 6000 rpm at 4°C for 30 min. The supernatants were filtered through a 0.22-μm Millipore filter (Merck Millipore) and precipitated with ice-cold 10% (v/v) trichloroacetic acid. After overnight incubation at 4°C, proteins were pelleted by centrifugation at 14,000 rpm at 4°C for 30 min and washed three times with ice-cold acetone. The resulting precipitates of secreted proteins were resuspended in 1× PBS. Intracellular proteins were prepared by lysing whole bacterial cells in 2× SDS buffer. Both secreted and intracellular proteins were subjected to SDS–polyacrylamide gel electrophoresis. The secretion of Hcp was detected by Western blot with anti-FLAG antibody (M20008, Abmart). RNA polymerase (RNAP) was an intracellular control detected with anti-RNA polymerase beta (EPR18704, Abcam).

Structure prediction of T4SS_BK effectors

The total protein sequences of Lysobacter and other T4SS_BK-encoding bacteria (15, 45) were used to predict T4SEs containing conserved XVIPCD domain with the same method as our previous report (16). The XVIPCD domain sequences of 13 X. citri XVIPCD proteins (13) were first aligned using the MUSCLE tool first and then used to construct a profile HMM. Subsequently, an HMM search was done against proteins of Lysobacter and other T4SS_BK-encoding bacteria using the hmmsearch program implemented in HMMER (41). A XVIPCD domain was considered present when the HMM search E value was lower than 10⁻⁵. The structures of these predicted T4SEs were predicted by Colabfold (42), and the structural similarity of above T4SEs and T4SEs from AFDB cluster was analyzed using US-align (43, 44, 57). The structures were visualized with PyMOL software (https://pymol.org/).

Antifungal activity of T4SS_BK effectors

To identify T4SEs with antifungal activity, individual T4SE genes were inserted into pYES-NTA vector and transformed into the yeast strain W303. Each yeast transformant was cultured in SC-U medium overnight at 28°C, supplemented with 2% glucose. After centrifugation (6000 rpm for 3 min at room temperature), the yeast cells were washed with sterilized ddH₂O and diluted to OD₆₀₀ of 1.0. Each strain was then subjected to a series of dilutions, spotted (5 μl) onto repressing (2% glucose) or inducing (2% galactose) medium, and incubated at 28°C for 48 hours before photography. T6SS effector Le1893 from L. enzymogenes (20) served as the antifungal control.

To investigate whether LtdD can induce DNA damage, S. cerevisiae transformants expressing different effectors were individually grown overnight at 28°C in SC-U supplemented with 2% glucose. Yeast cells were washed with sterilized ddH₂O and then diluted in SC-U medium supplemented with 2% galactose to induce effector expression. After 8 hours, the yeast samples were collected by centrifugation (6000 rpm for 3 min at room temperature), and the total proteins were extracted with Yeast Total Protein Extraction Kit (Sangon, China) and detected with antibodies specific for the histone H2B and γH2A (Abcam, UK) to check whether LtdD caused DNA damage in S. cerevisiae.

Interaction between LtdD and VirD4

For pull-down assay, pBAD containing LtdD-FLAG and pBADGM containing VirD4-His (cytoplasmic domain of VirD4) were transformed to E. coli Top10 by electroporation. After growing the transformed E. coli Top in LB (20 ml) at 37°C with OD₆₀₀ of 0.4, 100 μl of 20% arabinose was added and induced for 12 hours at 16°C. The cells were then centrifuged at 12,000 rpm for 10 min at 4°C. Harvested cells were resuspended in 4 ml of 10 mM PBS buffer and then sonicated. After centrifugation (12,000 rpm for 10 min at 4°C), 1 ml of soluble protein was mixed with 15 μl of anti-FLAG magnetic beads (Bimake, China). Following incubation at 4°C overnight, beads were washed five times for 5 min each with 1 ml of 10 mM PBS buffer containing 1% Triton X-100. Then, the proteins were boiled in protein loading buffer and identified by Western blot using specific anti-His and anti-FLAG monoclonal antibodies (Abmart, China).

We also used the BacterioMatch II two-hybrid (B2H) system (Agilent Technologies, USA) to detect LtdD-VirD4 interaction. In brief, ltdD was cloned into the pBT plasmid, whereas virD4-T, which encodes the cytoplasmic domain of VirD4, was inserted into the pTRG plasmid. These constructs were transformed into the E. coli XL1-Blue MRF’Kan strain. Positive controls included plasmids pBT-GacS and pTRG-GacS, whereas negative controls comprised transformants with empty pTRG and pBT vectors. The interaction was checked on the minimal medium (M9) supplemented with 5 mM 3-AT, streptomycin (2 μg/ml), tetracycline (12.5 μg/ml), chloramphenicol (34 μg/ml), and kanamycin (30 μg/ml), as described in a previous work (30).

Cross-kingdom translocation of LtdD

A modified Cre/loxP system was used to test the natural translocation of LtdD from L. enzymogenes to S. cerevisiae (30, 31). To achieve this, we construct a pGADT7 vector to express the ura3 gene, with loxP sites introduced on both sides of the ura3 gene. The yeast strain containing loxP-ura3-loxP cassette was used as the recipient strain. We constructed a pBBR vecotor expressing Cre-M3 and then transformed it into ΔlafB, ΔlafBΔvirD4, and ΔlafBΔtssM to make the donor strains. The donor strains and recipient strain were cocultured on YPD plates for 24 hours. After that, the colonies were eluted from YPD plates and cultured in SD-Leu medium containing chloramphenicol for 24 hours. The culture was then smeared on SD-Leu medium containing 5-FOA. Once Cre-M3 was translocated to the recipient strain, the loxP element could be specifically cleaved to remove ura3 gene, which would result in growth under 5-FOA stress.

DNase activity assay of T4SS_BK effectors

The sequences of T4SEs were cloned into pET30a vector and transformed into BL21 (DE3) for IPTG-induced expression. The proteins were purified with using Ni-NTA resin (GE Healthcare, China) and eluted by PBS buffer with 250 mM imidazole; after that, the proteins were concentrated with ultrafiltration tubes using PBS buffer. A 40-μl reaction mixture containing 200 ng of DNA and 1× CutSmart Buffer (New England Biolabs, USA) with partially purified proteins were incubated at 37°C for 30 min. DNase (New England Biolabs, USA; diluted to 0.01 U/μl) was used as positive control. The digestion of DNA was determined by agarose gel electrophoresis.

The DNA content was quantified using flow cytometry, as previously reported (58). Briefly, 1 × 10⁶ yeast cells cultivated 12 hours in inducing medium were harvested and fixed by incubation in 70% ethanol for 12 hours. The cells were subsequently washed with PBS and treated with 0.1% RNase A for 1 hour at 25°C. For DNA staining, the treated cells were incubated for 1 hour in PBS containing PI (50 μg/ml) at 25°C. DNA content was detected by measuring the PE-A value via flow cytometry (FACSCanto II). Unstained S. cerevisiae cells (without PI) were used as a negative control to subtract background signals.

Statistical analysis

Statistical analyses were performed using GraphPad 8.0 software. Three replicates of each sample were analyzed using a one-way analysis of variance (ANOVA) test to determine significant differences between groups. A significance level of P < 0.01 was considered statistically significant. Different letters indicate significant differences among groups. This study followed MIBBI guidelines for reporting microbiology research.

Supplementary Materials

The PDF file includes:

Figs. S1 to S7

Legends for tables S1 to S6

References

Other Supplementary Material for this manuscript includes the following:

Tables S1 to S6