The Lsm1–7 complex couples 3′-end protection to histone H4 acetylation to maintain mRNA homeostasis in Fusarium graminearum

Abstract

Eukaryotic messenger RNA (mRNA) homeostasis requires precise coordination between synthesis and decay, yet the mechanisms governing this balance in fungal pathogens remain elusive. Here we provide a comprehensive characterization of the Lsm1–7 complex in the cereal pathogen Fusarium graminearum. We show that Lsm1–7 assembles into a conserved hetero-heptameric module that localizes to processing bodies (P-bodies) and is required for fungal growth, virulence, and mycotoxin biosynthesis. Mechanistically, Lsm1–7/Pat1 binds U/A-rich 3′ termini of a defined set of transcripts enriched for central metabolism and restrains their 3′–5′ decay. Genetic suppressor analyses and mechanistic dissection identify two parallel decay routes antagonized by Lsm1–7, including the exosome recruited by the uridyltransferase Cid1 and the Ski–exosome complex mediated by the newly identified scaffold protein Lsp1. Moreover, loss of Lsm1–7 elicits a compensatory transcriptional response involving the Rpd3L (Sin3) histone deacetylase complex, in which elevated histone H4 acetylation at affected loci partially restores transcript output. Together, our results define an integrated cytoplasmic–nuclear regulatory axis in F. graminearum that couples 3′-end protection to chromatin-based transcriptional buffering to maintain mRNA homeostasis. While Lsp1 appears lineage-adapted, the underlying logic may reflect a broader principle of gene-expression buffering that supports fungal fitness and pathogenicity.

Graphical Abstract

Graphical Abstract.

Introduction

Eukaryotic messenger RNA (mRNA) homeostasis is maintained by coordinated control of cytoplasmic translation, storage, and decay, which together determine mRNA half-lives and gene expression outputs [1]. This coordination buffers transcript levels while enabling rapid gene-expression changes in response to environmental and intracellular signals [2–4]. A major part of this control is spatially organized in the cytoplasm in processing bodies (P-bodies). P-bodies are membraneless granules formed via liquid–liquid phase separation and are enriched for untranslated mRNAs together with RNA-binding proteins that repress translation, promote mRNA storage, or facilitate mRNA degradation [5, 6]. Although P-bodies are evolutionarily conserved from yeast to mammals, they are highly dynamic and remodel in response to diverse cues [7, 8]. In yeast, nutrient signaling pathways such as Ras/PKA modulate P-body assembly during glucose limitation [9]. Genotoxic stress can drive recruitment of specific DNA damage response mRNAs into P-bodies, contributing to granule remodeling [10, 11]. In Arabidopsis thaliana, P-bodies participate in developmental transitions by releasing stored transcripts during seed germination and are rapidly reorganized during innate immune responses triggered by microbe-associated molecular patterns [12, 13]. In mammalian cells, P-bodies have been linked to disease-associated programs, including tumor contexts in which YAP/TAZ promotes P-body formation [14]. Collectively, these findings highlight that although P-bodies are structurally conserved across eukaryotes, their molecular composition, dynamics, and biological functions exhibit substantial organism- and context-specific differences.

P-body assembly and function rely on a conserved set of proteins that coordinate major steps of cytoplasmic mRNA metabolism [15]. P-bodies concentrate factors involved in deadenylation, decapping, exonucleolytic degradation and translational repression. These include the Ccr4–Not and Pan2/3 deadenylases, the Dcp1/2 decapping enzyme together with decapping-associated factors such as Edc3/4, Pat1 and the Lsm1–7 complex, the 5′–3′ exonuclease Xrn1, and translational repressors such as Dhh1 and Scd6 [16–18]. Among these components, the Lsm1–7 complex stands out as a core P-body module that can connect events at the mRNA 3′ end to downstream decapping and decay decisions. In cooperation with Pat1, Lsm1–7 preferentially engages deadenylated mRNAs and promotes decapping by stimulating the Dcp1/2 machinery [19–22]. Beyond decapping, Lsm1–7 binds U-rich 3′ termini and has been implicated in the turnover of transcripts bearing terminal uridylation marks, linking 3′-end remodeling to mRNA decay [23–25]. In yeast, Lsm1–7 protects autophagy-related mRNAs from 3′–5′ degradation during nitrogen starvation [26]. In A. thaliana, it regulates development and abiotic stress responses by modulating specific targets, including the ABA-related transcript NCED3 during cold stress [27, 28]. In mammals, Lsm1 dysregulation is linked to tumorigenesis and immune programs [29–31]. Despite this broad biological significance, the molecular mechanisms by which the Lsm1–7 complex controls mRNA homeostasis in vivo are not fully understood.

These context-dependent functions of P-body modules suggest a functional communication between cytoplasmic mRNA fate decisions and nuclear gene regulation. Increasing evidence links P-body components to chromatin states and histone modifications. In stem cells, silencing the P-body component DDX6 triggers epigenetic remodeling at enhancers and heterochromatin, accompanied by increased H3K27 acetylation and altered H3K9me3 patterns, which lead to P-body dissolution and retranslation of stored mRNAs [32]. In oocytes, loss of LSM14 reduces H3K9me3 and leads to aberrant transcriptional activation [33]. In early embryos, LSM1 knockdown disrupts incorporation of the histone variant H3.3 into the male pronucleus and leads to increased H3K9me3 of undegraded major satellite RNAs [31]. These findings support a functional link between cytoplasmic P-bodies and nuclear histone modification dynamics. However, how P-body-associated mRNA decay or protection is coupled to transcriptional compensation, especially through histone acetylation, is unclear.

To define how a conserved P-body module specifies mRNA fate and engages nuclear feedback, we investigated these mechanisms in the filamentous fungal pathogen Fusarium graminearum, which causes major yield losses and contaminates grain with mycotoxins such as deoxynivalenol (DON) [34]. During DON biosynthesis and plant infection, we observed strong P-body induction in this fungus (Supplementary Fig. S1A–C), suggesting that P-bodies are engaged during virulence-associated programs. We therefore focused on the Lsm1–7/Pat1 complex, a core P-body module linking 3′ ends to mRNA fate decisions, which is highly expressed during these stages (Supplementary Fig. S1D). Our findings show that Lsm1–7 is required for fungal development, virulence, and DON biosynthesis. Mechanistically, the Lsm1–7/Pat1 complex binds oligo(U/A)-tailed mRNA 3′ ends to limit 3′–5′ decay by restricting access of exonucleolytic decay machineries, including the Cid1-recruited exosome and the Ski–exosome complex recruited by a newly identified scaffold, Lsp1. Loss of this protection elicits transcriptional compensation through the Rpd3L histone deacetylase complex, linking cytoplasmic mRNA turnover to chromatin regulation. Together, our findings delineate an integrated regulatory axis whereby the Lsm1–7/Pat1 complex safeguards oligo(U/A)-tailed mRNAs from 3′–5′ decay and couples this cytoplasmic protection to histone acetylation-mediated transcriptional compensation in a filamentous fungal pathogen.

Materials and methods

Fungal strain culture conditions

Fusarium graminearum strain PH-1 (NRRL 31084), originally isolated from maize in Michigan, USA, was used as the wild type (WT). All mutant strains, including gene deletions, complementations, and site-directed point mutants, were generated in the PH-1 background. For vegetative growth assays, strains were cultured on complete medium (CM; 10 g of glucose, 2 g peptone, 1 g yeast extract, 1 g casamino acids, 6 g NaNO₃, 0.52 g KCl, 0.52 g MgSO₄·7H₂O, 1.52 g KH₂PO₄, 0.01% trace elements, 0.01% of vitamins, 1% agar per liter, pH 6.5) plates. For liquid culture, CM was used to obtain vegetative hyphae for subsequent extraction of genomic DNA, total RNA, or proteins. Trichothecene biosynthesis induction (TBI; 30 g sucrose, 1 g KH₂PO₄, 0.5 g MgSO₄·7H₂O, 0.5 g KCl, 0.01 g FeSO₄·7H₂O, 1.47 g putrescine hydrochloride, 0.05% trace element per liter, pH 4.5) medium was used to assess TRI gene expression and deoxynivalenol (DON) production [35]. All experiments were independently repeated three times.

Strain constructions

Gene deletion, gene knockdown and in/ex situ complementation strains were generated using the double-joint (DJ) PCR method, as previously described [36]. For gene deletion, the 5′ and 3′ flanking sequences of each target gene were amplified using gene-specific primers (Supplementary Table S1) and fused with the hygromycin B resistance cassette (hph) via overlap PCR. The resulting constructs were introduced into F. graminearum PH-1 protoplasts using polyethylene glycol (PEG)-mediated transformation [37]. For gene knockdown, the promoter was replaced with P_zear, and the inducer β-estradiol (30 μM) was added to the medium during mutant selection [38]. For in situ complementation, open reading frames (ORFs) were fused in-frame with GFP or RFP at either terminus, flanked by their native 5′ and 3′ untranslated regions, and co-amplified with the geneticin resistance cassette (neo) using DJ-PCR. For ex situ tag-fusion constructs, the ORFs of target genes were amplified and co-transformed with the restriction enzyme-digested pYF11 plasmid into yeast strain XK1-25 [39] using the yeast homologous recombination system. Transformants were selected on SD-Leu medium (26.7 g Minimal SD base, 0.74 g -Trp DO supplement, 1% agar per liter, pH 6.7) at 30°C for 3 days. Verified recombinant plasmids were recovered from yeast, propagated in Escherichia coli DH5α, and subsequently used for PEG-mediated transformation into F. graminearum protoplasts.

Phenotype of hyphal growth, pathogenicity, and DON biosynthesis assays

Vegetative growth was assessed by inoculating 5-mm mycelial plugs from the actively growing colony margin onto CM plates, followed by incubation at 25°C for 3 days. The virulence of F. graminearum strains was assessed in a head blight assay by injecting conidial suspensions (1 × 10⁵ conidia/ml) into a central floret of flowering wheat heads (Triticum aestivum, cv. Jimai 22), with symptoms observed 14 days post-inoculation. To evaluate DON biosynthesis, strains were cultured in TBI liquid medium at 28°C in the dark with shaking at 150 rpm. After 7 days, culture supernatants were collected for DON quantification using a commercial enzyme-linked immunosorbent assay (ELISA)-based DON detection kit (Wis008, Wise Science, Zhenjiang, China). All assays were performed with at least three biological replicates.

Phylogenetic analysis

To explore the evolutionary conservation of Lsm1–7 complex subunits, a local BLAST database was constructed using 1642 publicly available fungal genomes. Protein sequences of the Lsm1–7 complex from Saccharomyces cerevisiae served as query sequences for homology searches across the dataset. Phylogenetic analysis of Lsp1, Ski2, and Cid1 was performed by retrieving homologous amino acid sequences from other species using BLAST. Multiple sequence alignments were conducted with ClustalW, and phylogenetic trees were constructed using the Neighbor-Joining method in MEGA 11 with 1000 bootstrap replicates [40]. The resulting alignments were visualized in Jalview to evaluate sequence conservation across species [41].

Yeast two-hybrid (Y2H) assays

Full-length complementary DNA (cDNA) sequences were amplified from PH-1 total RNA using gene-specific primers (Supplementary Table S1) and cloned into the GAL4 DNA-binding domain vector pGBKT7 and the GAL4 activation domain vector pGADT7 (Clontech, Mountain View, CA, USA). Cloning was performed using the ClonExpress One Step Cloning Kit (C112-01, Vazyme, Nanjing, China) following vector digestion with appropriate restriction enzymes. The recombinant plasmids were co-transformed into S. cerevisiae strain Y2H Gold using the standard lithium acetate method. The pGBKT7-53/pGADT7-T pair was used as a positive control, while pGBKT7-Lam/pGADT7-T served as a negative control. To test for self-activation, pGBKT7-target gene constructs were co-transformed with empty pGADT7. Transformants were selected on SD/-Leu/-Trp/-His/-Ade (26.7 g Minimal SD base, 0.62 g -Leu/-Trp DO supplement, 1% agar per liter, pH 6.7) plates using serial dilutions and incubated at 30°C for 4 days. Each assay was independently repeated three times to ensure reproducibility.

Western blotting assays

Western blotting was performed as previously described [42]. Primary antibodies used for immunoblotting included anti-GFP (ab32146, Abcam, Cambridge, UK), anti-Flag (AE092, ABclonal, Wuhan, China), anti-H4ac (39243, Active Motif, Carlsbad, CA, USA), anti-H4K5ac (A20397, ABclonal, Wuhan, China), anti-H4K8ac (A7258, ABclonal, Wuhan, China), and anti-H4K12ac (A23683, ABclonal, Wuhan, China) to detect GFP, Flag, and acetylated histone H4 modifications (H4ac, H4K5ac, H4K8ac, H4K12ac). Anti-H4 (ET1612-43, HuaAn Biotechnology, Hangzhou, China) and anti-GAPDH (EM1101, HuaAn Biotechnology, Hangzhou, China) were used as loading controls [43]. Signal intensities were quantified using ImageJ software (NIH, Bethesda, MD, USA).

Affinity capture-mass spectrometry analysis

To identify protein interaction partners, Lsm1, Lsp1, and Cid1 were individually GFP-tagged and expressed in the wild-type F. graminearum background. For each sample, 1 ml of total protein lysate was incubated with 40 μl of anti-GFP magnetic beads (SA070001, Smart-Life, Changzhou, China) at 4°C overnight with gentle rotation. Beads were washed 5–8 times with 1 ml Tris-buffered saline (TBS; 20 mM Tris–HCl, 500 mM NaCl, pH 7.5). Bound proteins were eluted by boiling the beads in 50 μl TBS containing 10 μl of 10% sodium dodecyl sulfate (SDS) for 10 min. After centrifugation at 12 000 × g for 5 min at room temperature, the supernatants were subjected to mass spectrometry analysis by QLbio (Shanghai, China).

Co-immunoprecipitation assay

To validate protein–protein interactions, candidate genes were fused with GFP or Flag tags and co-expressed in PH-1. For each co-immunoprecipitation (co-IP) assay, 1 ml of total protein lysate was incubated with 25 μl of anti-GFP or anti-Flag magnetic beads (SA070001 or SA042001, Smart-Life, Changzhou, China) for 6 h at 4°C with gentle rotation. A 200 μl aliquot of lysate was retained as input and stored at −80°C. Bound proteins were eluted by boiling in 50 μl TBS supplemented with 10 μl of 10% SDS for 10 min. Both input and immunoprecipitated (IP) fractions were analyzed by western blotting.

Bimolecular fluorescence complementation assays

The pHZ65 vector harboring NYFP (hph resistance) and CYFP (zeocin resistance) cassettes was digested with restriction enzymes and ligated with target gene fragments amplified using the primers listed in Supplementary Table S1. Recombinant constructs were verified by Sanger sequencing and co-transformed into PH-1. Transformants were selected using hph and zeocin. YFP fluorescence in hyphae was visualized using a Zeiss LSM880 confocal microscope (Göttingen, Niedersachsen, Germany).

RNA-seq and quantitative reverse transcriptase-polymerase chain reaction analysis

For transcriptomic analysis, the PH-1 and mutant strains Δlsm1, Δlsp1, and Δlsp1/lsm1, Δcid1, Δcid1/lsm1, Δsin3, and Δsin3/lsm1, were cultured in CM at 25°C with shaking for 24 h. Three independent biological replicates were prepared for each strain. Total RNA was extracted, and library construction and high-throughput sequencing were performed by Illumina NovaSeq 6000 platform with paired-end reads. Genes with a |log₂fold change| > 1 and a false discovery rate (FDR) < 0.05 were considered significantly differentially expressed. Functional enrichment and pathway analyzes were performed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database via the DAVID Bioinformatics Resources (https://david.ncifcrf.gov/). For quantitative reverse transcriptase-polymerase chain reaction (RT-qPCR), primer sequences are listed in Supplementary Table S1. ACTIN was used as the internal reference gene [44].

Transcriptional shutoff

Mycelia were grown in CM for 24 h at 25°C with shaking at 180 rpm. Cultures were then treated with either dimethyl sulfoxide (DMSO) or 10 μg/ml Thiolutin (HY-N6712, MedChemExpress, Monmouth Junction, NJ, USA). Samples were collected at 0, 15, and 30 min after treatment. Total RNA was extracted, and RT-qPCR was performed. Relative expression levels were calculated using the ΔΔCt method, with ACTIN mRNA and the 0 min time point used as references [45].

Chromatin immunoprecipitation sequencing and chromatin immunoprecipitation-qPCR analysis

Chromatin immunoprecipitation (ChIP) was performed following a previously published protocol [46]. The F. graminearum strains PH-1, Δlsm1, and Δsin3 were cultured in CM at 25°C with shaking for 24 h. IP samples were incubated overnight with a rabbit polyclonal anti-H4 antibody, while Mock samples were incubated with anti-IgG1 antibody (MA1-10406, Invitrogen, Carlsbad, CA, USA), followed by incubation with protein A agarose beads for 4 h at 4°C. For ChIP-seq, IP and Mock DNA samples were sequenced on NovaSeq 6000 platform. For ChIP-qPCR, purified IP and Mock DNA were subjected to qPCR using the ChamQ Universal SYBR qPCR Master Mix (Q711-02, Vazyme, Nanjing, China). ACTIN was used as the internal reference gene.

RIP-seq and RIP-qPCR analysis

RNA immunoprecipitation followed by sequencing (RIP-seq) and quantitative PCR (RIP-qPCR) were performed as previously described [42]. A Flag-tagged Lsm1 complementation strain (Δlsm1::Lsm1-Flag) and PH-1 were used for immunoprecipitation (IP) and mock control, respectively. Strains were cultured in CM liquid medium at 25°C with shaking for 24 h. For RIP-seq, RNA from IP (Δlsm1::Lsm1-Flag) and mock (PH-1) samples was used for cDNA library construction and sequenced on Illumina NovaSeq 6000. For RIP-qPCR, 1 μl of RNA from Input, IP, and Mock samples was reverse transcribed according to the standard protocol. ACTIN was used as a reference gene. mRNA enrichment was calculated using the 2^ΔCt method: ΔCt = Ct (Mock) – Ct (IP).

Poly(A)-seq and data analysis

Poly(A)-seq was performed as previously described with minor modifications [47]. F. graminearum wild-type strain PH-1 and the deletion mutants Δcid1 and Δlsm1 were cultured in CM at 25°C with shaking for 24 h. Total RNA was extracted using standard protocols and treated with RQ1 DNase (M6101, Promega, Madison, WI, USA) to remove genomic DNA. Partial digestion was performed with RNase T1 (EN0541, Thermo Fisher Scientific, Waltham, MA, USA) to fragment RNA. Polyadenylated RNA was isolated using oligo(dT)-conjugated magnetic beads (61005, Invitrogen, Carlsbad, CA, USA). A 3′ RNA adaptor (5′-TGGAATTCTCGGGTGCCAAGG-3′) was ligated to the poly(A) tails, followed by cDNA library construction using the ScriptSeq v2 RNA-Seq Library Preparation Kit (SSV21124, Illumina, San Diego, CA, USA). PCR-amplified fragments of 200–500 bp were gel-purified and sequenced using150-bp paired-end sequencing on the Illumina NovaSeq 6000 platform. High-quality 5′ end sequencing reads containing identifiable 3′ adaptor sequences were retained. A 3-nt random sequence at the 5′ end and the 3′ Illumina adaptor were trimmed. Reads of ≥ 18 nt in length, with poly(A) tails removed, were aligned to the reference genome using Bowtie2. This approach enabled simultaneous identification of poly(A) tail addition sites and their corresponding genes.

SweAMI FISH assay

The Saiweier Signal Amplification Multiplex Isothermal (SweAMI) fluorescence in situ hybridization (FISH) assay was performed using gene-specific probes (Servicebio, Wuhan, China); probe sequences are listed in Supplementary Table S1. Mycelia were cultured in CM for 24 h, fixed in situ hybridization fixative for ≥ 12 h, embedded in paraffin, and sectioned at a thickness of 5 μm. Hybridization procedures followed previously established protocols [48, 49]. Fluorescent signals were visualized using a Zeiss LSM980 confocal microscope (Göttingen, Niedersachsen, Germany).

Statistical analysis

All quantitative data are presented as mean ± standard deviation (SD) from at least three independent experiments. Statistical analyzes were performed using GraphPad Prism version 10.0 (GraphPad Software, San Diego, CA, USA). Differences between two groups were assessed using the unpaired Student’s t-test. For comparisons among multiple groups, one-way analysis of variance (ANOVA) was conducted, followed by Fisher’s least significant difference (LSD) post hoc test. Statistical significance was defined at a threshold of P <.05. Differences among groups are indicated by distinct lowercase letters, with different letters denoting statistically significant differences.

Results

Lsm1–7 complex is a core P-body component required for F. graminearum growth and virulence

Comparative analyzes across representative eukaryotic lineages identified Lsm1–Lsm7 in major clades, consistent with broad conservation of the Lsm1–7 complex in eukaryotes (Supplementary Fig. S2A). We next examined conservation within fungi by analyzing 1642 fungal genomes and found that all seven subunits are broadly retained, with particularly strong conservation in Saccharomycotina; homologs share ∼40% average amino-acid identity (Fig. 1A and Supplementary Fig. S2B). In F. graminearum, we identified orthologs of Lsm1–Lsm7, each encoding a canonical Sm-like domain like yeast and human proteins (Supplementary Fig. S2C). We then determined subcellular localization by expressing GFP-tagged Lsm subunits in the wild-type strain PH-1. Confocal microscopy showed that Lsm1-GFP formed distinct cytoplasmic puncta that co-localized with the P-body marker Scd6-RFP (Fig. 1B). In contrast, Lsm2–Lsm7 displayed dual localization to cytoplasmic puncta and the nucleus, as indicated by Histone H1-RFP (Fig. 1C). Formation of Lsm1-GFP puncta depended on the integrity of the complex, as deletion of individual components (for example, Δlsm3 or Δlsm5) abolished Lsm1-GFP foci (Fig. 1D). To define complex assembly, we mapped interactions among Lsm proteins using yeast two-hybrid (Y2H), bimolecular fluorescence complementation (BiFC), and affinity capture-mass spectrometry (AC-MS). Y2H and BiFC detected nine robust pairwise interactions, with Lsm3 displaying the highest connectivity and interacting with four other subunits (Fig. 1E and F). Consistently, AC-MS using Lsm1-GFP as bait recovered all other Lsm proteins (Lsm2–Lsm7) (Supplementary Table S2). Co-immunoprecipitation (Co-IP) further validated representative interactions, including Lsm1–Lsm6 and Lsm1–Lsm7 (Fig. 1G). Together, these results demonstrate that Lsm1–7 assembles into a conserved heteroheptamer in this fungus (Fig. 1H).

Figure 1.

Identification, interaction architecture, and biological essentiality of the Lsm1–7 complex in F. graminearum. (A) Conservation analysis of the seven Lsm subunits across fungi using S. cerevisiae homologs as queries. (B) Co-localization of Lsm1-GFP with the conserved P-body marker Scd6-RFP. (C) Subcellular localization of individual GFP-tagged Lsm subunits in hyphae; H1-RFP serves as a nuclear marker. Scale bar = 5 µm. (D) Subcellular localization of Lsm3-GFP in WT, Δlsm1 and Δlsm5. Scale bar = 5 µm. (E) Interaction network among Lsm subunits determined by yeast two-hybrid (Y2H) assays. Yeast transformants were spotted on selective medium lacking Leu/Trp/His/Ade. (F) BiFC validation of representative interactions. Scale bar = 5 µm. (G) Co-immunoprecipitation (Co-IP) confirming Lsm1 association with Lsm6 and Lsm7. (H) Interaction model of the hetero-heptameric Lsm1–7 complex; edge colors indicate supporting assays (Y2H, BiFC, Co-IP, or AC-MS). (I) Colony growth of PH-1 and Δlsm mutants on CM after 3 days. (J) Virulence assay on flowering wheat heads at 14 days post-inoculation. (K) Deoxynivalenol (DON) production after 7 days in TBI medium. (L) Visualization of DON toxisome assembly using Tri1-GFP as a reporter in mycelia grown in TBI medium for 48 h. (M) Western blot analysis of Tri1-GFP protein abundance; GAPDH was used as an internal loading control.

We next assessed biological function by generating single-gene deletion mutants for each Lsm subunit. All Δlsm mutants exhibited severe vegetative growth defects compared with PH-1 (Fig. 1I). Virulence assays on flowering wheat heads showed that Δlsm5 caused symptoms largely restricted to inoculated spikelets, whereas the remaining mutants were non-pathogenic (Fig. 1J). Because DON promotes fungal spread within a spike, we quantified DON production after growth in TBI medium. All Δlsm mutants produced markedly less DON than PH-1 (Fig. 1K). Consistent with impaired toxin biosynthesis, DON toxisomes, visualized using Tri1-GFP, were absent in the mutants but readily detected in PH-1 (Fig. 1L). Immunoblotting showed that Tri1-GFP accumulated to near-undetectable levels in the Δlsm strains (Fig. 1M). Collectively, these data establish that the Lsm1–7 complex assembles in P-bodies and is required for growth, virulence, and DON biosynthesis in F. graminearum.

Lsm1–7 complex regulates mRNA turnover of central metabolic genes

Lsm1–7 complex is an RNA-binding complex implicated in mRNA turnover, prompting us to identify its direct RNA substrates in F. graminearum. We performed RNA immunoprecipitation sequencing (RIP-seq) using a Δlsm1 strain complemented with LSM1-Flag under its native promoter at the original locus (Fig. 2A). RIP-seq identified 294 significant Lsm1-Flag binding peaks (P <.05, fold change > 2), mapping to 271 protein-coding genes and a small set of noncoding RNAs (Supplementary Table S3). KEGG analysis showed that Lsm1-bound transcripts are enriched for ribosome biogenesis and primary metabolic pathways, including carbon metabolism and amino acid biosynthesis (Fig. 2B). We next assessed transcriptome changes upon loss of Lsm1 by RNA-seq of Δlsm1 and WT. Differential expression analysis identified 1464 upregulated genes (log₂FC > 1) and 1538 downregulated genes (log₂FC < −1) in Δlsm1 (Supplementary Fig. S3A and Supplementary Table S4). KEGG analysis showed that downregulated genes were primarily enriched in carbon metabolism and amino acid biosynthesis (Fig. 2C), while upregulated genes were associated with DNA damage repair, ABC transporters, and autophagy pathways (Supplementary Fig. S3B–D). When intersecting RIP-seq targets with the RNA-seq dataset, 110 of the 271 Lsm1-bound mRNAs were significantly downregulated in Δlsm1 (P <.05, log₂FC < −1), while only three were upregulated; the remaining 159 showed no significant change (Fig. 2D). The downregulated, Lsm1-bound subset was again enriched in carbon metabolism and amino acid biosynthesis, consistent with a direct role of Lsm1–7 complex in maintaining mRNA abundance for primary metabolic pathways.

Figure 2.

Lsm1 binds mRNAs enriched for central metabolism and supports their steady-state abundance. (A) Schematic of RIP-seq using Δlsm1::Lsm1-Flag to identify Lsm1-bound RNAs. (B) KEGG enrichment of genes associated with Lsm1-bound mRNAs identified by RIP-seq. (C) KEGG enrichment of genes downregulated in Δlsm1 relative to PH-1. (D) Differential expression of Lsm1-bound mRNAs in Δlsm1 relative to PH-1 (RNA-seq); significantly upregulated and downregulated targets are indicated (P <.05; log₂FC > 1). (E) Glycolysis pathway map highlighting Lsm1-bound genes that are reduced in Δlsm1 (P <.05; log₂FC < −1). (F) RIP-seq profiles for representative glycolysis genes; values in parentheses indicate log₂FC (Δlsm1 versus PH-1). The y-axis represents the normalized read count per nucleotide. (G) RIP-qPCR validation of enrichment of selected glycolysis mRNAs in Lsm1-Flag immunoprecipitates. (H) RT-qPCR measurement of selected glycolysis mRNAs in Δlsm1 relative to PH-1. (I) Time-course expression of selected glycolysis mRNAs in PH-1 and Δlsm1 grown in CM or TBI (24–48 h); PH-1 at 24 h in CM was set to 1. (J) RNA in situ hybridization showing localization of selected Lsm1-bound mRNAs relative to Scd6-marked P-bodies. A line scan graph illustrates the relative localization of RFP-Scd6, RNA-FBA1, and 4′,6-diamidino-2-phenylindole (DAPI). Scale bar = 5 μm. (K) Growth phenotypes of knockdown strains of selected glycolysis genes. Data are mean ± SD; (G, H) n = 4 biological replicates, (I) n = 3 biological replicates. Statistical significance was assessed by one-way ANOVA followed by Fisher’s LSD test; different letters indicate significant differences (P <.05).

Glycolysis emerged as a prominent example of this regulation. RIP-seq identified 13 Lsm1-bound genes in the glycolysis pathway (Fig. 2E and F, and Supplementary Fig. S3E), all of which showed reduced abundance in Δlsm1 (Supplementary Table S3). RIP-qPCR validated enrichment of eight representative glycolysis mRNAs in the Δlsm1::Lsm1-Flag strain (Fig. 2G), and RT-qPCR confirmed their decreased steady-state levels in Δlsm1 (Fig. 2H). To directly test whether loss of Lsm1 alters decay kinetics, we performed transcriptional shut-off assays and observed accelerated decay of glycolytic mRNAs in Δlsm1 relative to PH-1 (Supplementary Fig. S3F). Consistent with condition-dependent regulation, several Lsm1-bound glycolysis transcripts (PGK1, FBA1, PGI1, and THD3) remained lower in Δlsm1 across 24–48 h growth and were most strongly reduced in TBI medium (Fig. 2I). RNA in situ hybridization further showed that these transcripts preferentially localized to P-bodies (Fig. 2J and Supplementary Fig. S3G). Finally, knockdown of selected metabolic targets (e.g. FBA1 and ADH2) caused severe growth defects (Fig. 2K), linking the Lsm1-bound metabolic regulon to fungal fitness. Together, these results indicate that Lsm1–7 binds a defined set of mRNAs enriched for central metabolism and limits their accelerated decay to support core biological processes.

A lineage-specific scaffold Lsp1 recruits Ski–exosome to P-bodies and modulates Lsm1-dependent transcripts

The severe growth defects of Δlsm mutants were frequently bypassed by spontaneous fast-growing suppressor sectors (Fig. 3A). We isolated four independent suppressors (S1–S4) from the Δlsm1 background and performed whole-genome sequencing. Comparative analysis identified disruptive mutations in a single locus, FGSG_01632, including two frameshift and two nonsense alleles predicted to truncate the encoded protein (Fig. 3B and Supplementary Fig. S4A). FGSG_01632 encodes a WD40-repeat protein that we named Lsp1 (Lsm suppressor protein 1). Consistent with a causal role, deletion of LSP1 partially rescued the growth defects of Δlsm1 and Δlsm2, and the resulting double mutants, Δlsp1/lsm1 and Δlsp1/lsm2, phenocopied the spontaneous suppressors (Fig. 3C). These genetic data place Lsp1 in the Lsm1–7 regulatory pathway.

Figure 3.

Lsp1 recruits RNA exosome/SKI complexes into P-bodies to modulate mRNA abundance. (A) Δlsm1 colonies frequently produced fast-growth suppressor sectors (arrows); four representative suppressors were shown after growth on CM. (B) Suppressor mutations mapped to Lsp1 (FGSG_01632); frameshift and nonsense alleles are indicated. (C) Growth phenotypes of PH-1 and indicated mutants on CM. (D) Y2H interaction analysis between Lsp1 and decapping proteins (Dcp1, Dcp2, Edc3). (E) Co-IP validating the Lsp1-Dcp2 association in vivo. (F) Co-localization of RFP-Lsp1 and GFP-Dcp2 in hyphae grown in liquid CM for 16 h. The line-scan profile (arrow) reports relative fluorescence intensities. Scale bar = 5 µm. (G) Y2H interaction analysis between Lsp1 and components of the RNA exosome and Ski complexes. (H) Co-IP validation of representative interactions between Lsp1 and Csl4 (exosome) or Ski2 (Ski complex). (I) Fluorescence microscopy showing RFP-Lsp1 with GFP-tagged Csl4 or Ski2 in CM-grown hyphae (16 h), with line-scan profiles. Scale bar = 5 µm. (J) Localization of GFP-Csl4 or GFP-Ski2 in Δlsp1 strain. Scale bar = 5 µm. (K) RNA-seq heatmap of log₂FC for 110 Lsm1-bound transcripts in Δlsm1, Δlsp1/lsm1, and Δlsp1 relative to WT. (L) RT-qPCR validation of representative substrates in the indicated strains. Error bars represent standard deviation from three replicates. Different letters indicate significant differences (P <.05; one-way ANOVA with Fisher’s LSD). (M) Working model: Lsm1–7 occupancy at mRNA 3′ ends limits access of Lsp1-recruited Ski–exosome decay factors; loss of Lsp1 dampens decay-factor recruitment and suppresses Δlsm1 defects.

We next examined Lsp1 conservation and potential functional analogs. Sequence-based phylogeny identified Lsp1 homologs largely restricted to Pezizomycotina, and absent or highly divergent in other fungal lineages (Supplementary Fig. S4B). Despite low sequence identity (<15%), structural modeling using Template Modeling alignment (TM-align) revealed similarity between Lsp1 WD40 domain and the metazoan decapping scaffold Edc4 (TM-score 0.64–0.74) (Supplementary Fig. S4C and D), suggesting that Lsp1 may represent a Pezizomycotina-specific structural analog of Edc4, consistent with convergent evolution of decapping scaffolds. To test whether Lsp1 associates with the decapping machinery, we assessed physical interactions and subcellular localization with known decapping components. Y2H assays detected direct interactions between Dcp2 and Dcp1, Edc3, or Lsp1 (Fig. 3D). Co-IP further supported an association between Lsp1 and Dcp2 in vivo (Fig. 3E). Fluorescence microscopy showed that Lsp1 co-localizes with Dcp1, Dcp2, and Edc3 in cytoplasmic P-bodies (Fig. 3F and Supplementary Fig. S4E). Together, these results link Lsp1 to P-body–associated decapping factors.

To define the mechanism underlying Δlsp1-mediated suppression of Δlsm1, we performed AC-MS using GFP-Lsp1 as bait. AC-MS recovered multiple subunits of the RNA exosome, a central 3′–5′ RNA decay/processing machine, as well as components of the Ski complex (Supplementary Table S5). Y2H and Co-IP validated interactions between Lsp1 and exosome factors (Csl4, Mtr3, Rrp4) and Ski proteins (Ski2, Ski3) (Fig. 3G and H, and Supplementary Fig. S5A). Fluorescence localization experiments further showed that Csl4 and Mtr3 frequently formed ring-like signals surrounding Lsp1 in P-bodies, while other interactors largely overlapped with Lsp1 (Fig. 3I, and Supplementary Fig. S5B and C). We next tested whether Lsp1 is required for P-body association of these decay factors. In Δlsp1, Csl4-GFP, Rrp4-GFP became predominantly nuclear, Mtr3-GFP showed reduced P-body localization, and Ski2-GFP/Ski3-GFP showed reduced accumulation in P-bodies (Fig. 3J and Supplementary Fig. S5D). These results indicate that Lsp1 is required for efficient P-body association of Ski–exosome components.

We next tested whether Lsp1 influences Lsm1-dependent transcript abundance. RNA-seq of Δlsp1 identified 1014 upregulated and 702 downregulated genes (|log₂FC| >1, FDR < 0.05) (Supplementary Fig. S6A), consistent with a broad role of Lsp1 in limiting steady-state mRNA abundance. Notably, glycolysis genes were enriched among the upregulated transcripts in Δlsp1 (Supplementary Fig. S6B). Remarkably, ∼35% (1062/3002) of differentially expressed genes in Δlsm1 were restored toward wild-type levels in Δlsp1/lsm1 (Fig. 3K, Supplementary Fig. S6C–E, and Supplementary Tables S4, S6, and S7). Notably, 110 Lsm1-bound transcripts that were reduced in Δlsm1 showed partial recovery in the double mutant. RT-qPCR further confirmed that representative Lsm1-bound glycolysis mRNAs (PGK1, FBA1, PGI1, and THD3) were increased or unchanged in Δlsp1 and recovered in Δlsp1/lsm1, relative to Δlsm1 (Fig. 3L). Together, these results identified Lsp1 as a lineage-specific P-body scaffold that recruits Ski-exosome complexes and modulates Lsm1–7 complex dependent transcript abundance. These data support a model in which Lsm1–7 protects mRNA 3′ ends from 3′–5′ decay, whereas Lsp1 promotes Ski-exosome recruitment to P-bodies; loss of Lsp1 reduces decay-factor recruitment and partially restores mRNA levels and growth in Δlsm1 (Fig. 3M).

Pat1 bridges Lsm1–7 and Lsp1 to couple mRNA protection to 3′–5′ decay

Lsp1 showed no detectable interaction with Lsm1–7 subunits, and Δlsm1 did not alter Lsp1 co-localization with Ski–exosome core components (e.g. Csl4, Ski2) in P-bodies (Supplementary Fig. S7A and B). These observations suggested that an additional factor links the Lsp1–Ski–exosome module to Lsm1–7 at target mRNA 3′ ends. To identify such a mediator, we performed AC-MS using Lsp1-GFP as bait and recovered multiple P-body proteins among which the conserved scaffold Pat1 was a prominent candidate (Supplementary Table S5). Y2H and Co-IP assays confirmed that Pat1 associates with Lsp1 and interacts broadly with the Lsm1–7 complex (Fig. 4A–C). Pat1-GFP localized to P-bodies and the nucleus, and co-localized with Lsp1 in P-bodies (Fig. 4D). Loss of Pat1 reduced Lsp1-RFP foci size (Fig. 4E, 28 ± 10% decrease versus WT) and caused Lsm1-GFP to become diffuse in the cytoplasm rather than accumulating in P-bodies (Fig. 4F). Conversely, Pat1 P-body localization was maintained in Δlsm1 and Δlsp1 (Fig. 4G). Together, these data identify Pat1 as a bridging scaffold that supports assembly of an Lsm1–7/Pat1/Lsp1 regulatory hub in P-bodies.

Figure 4.

Pat1 links Lsp1 to the Lsm1–7 complex and constrains Ski2-dependent 3′–5′ decay. (A) Y2H analysis of the Lsp1–Pat1 interaction. (B) Co-IP assay validating the Lsp1-Pat1 interaction in vivo, detected by immunoblotting. (C) Y2H analysis of interactions between Pat1 and individual Lsm1–7 subunits. (D) Subcellular localization of GFP-Pat1 with RFP-H1 or RFP-Lsp1 in hyphae grown in liquid CM for 16 h. Scale bar = 5 µm. Line scan profiles show relative localization of RFP-Lsp1 or RFP-H1 and GFP-Pat1 along the indicated paths (arrows). (E) Localization of Lsp1-RFP in PH-1 and Δpat1, with quantification of P-body size based on Lsp1-RFP foci. (F) Localization of GFP-Lsm1 in PH-1 and Δpat1. (G) Localization of GFP-Pat1 in PH-1, Δlsm1, and Δlsp1. Scale bar = 5 µm. (H) Colony growth of PH-1 and indicated mutants on CM. (I) RT-qPCR analysis of selected target transcripts in PH-1 and the indicated mutants. (J) Δpat1 colonies producing fast-growing suppressor sectors (arrows) and the isolated suppressor S1. (K) Ski2 mutation identified in the Δpat1 suppressor. (L) Growth phenotypes of PH-1 and the indicated mutants. (M) RT-qPCR analysis of selected target genes in the indicated strains. (N) Proposed model for the mediator function of Pat1. Pat1 recruits Lsm1–7 to mRNA 3′ ends and, via its interaction with Lsp1, limits access of the Ski–exosome machinery in P-bodies, thereby restraining 3′–5′ decay. Error bars indicate standard deviation from three biological replicates. Different letters indicate statistically significant differences (P <.05; one-way ANOVA followed by Fisher’s LSD test).

Genetic interactions further supported functional interdependence among these components. The Δpat1 mutant displayed severe growth defects like Δlsm1, and the Δpat1/lsm1 double mutant was further impaired (Fig. 4H). In contrast, deleting LSP1 partially rescued the growth phenotype of Δpat1, and Pat1 overexpression in the Δlsm1 (Δlsm1::Pat1^OE) partially rescued its growth defect (Fig. 4H). Consistently, RT-qPCR showed that Pat1 overexpression partially restored these Lsm1-bound glycolysis mRNAs in Δlsm1, and that LSP1 deletion similarly increased their abundance in the Δpat1 background (Fig. 4I).

To test whether Ski–exosome mediates the Δpat1 defects, we isolated fast-growing Δpat1 suppressor sectors on agar plates (Fig. 4J). Whole-genome sequencing of one suppressor (S1) identified a missense mutation in FGSG_02781 encoding Ski2, the helicase subunit of the Ski complex (Fig. 4K and Supplementary Fig. S8). The substitution mapped to the conserved DHSCT domain of Ski2, which is required for ATP-dependent helicase activity and Ski–exosome-mediated RNA decay. Deleting SKI2 in the Δpat1 background largely restored growth and rescued the reduced abundance of representative glycolysis mRNAs (Fig. 4L and M), indicating that the Δpat1 defects depend on Ski2-driven decay activity. Together, these results support a tripartite model in which Pat1 recruits Lsm1–7 to target mRNA 3′ ends while also engaging Lsp1, thereby restricting Ski–exosome access to these substrates within P-bodies (Fig. 4N). This organization links mRNA protection to 3′–5′ decay control and supports fungal growth and primary metabolism.

Lsm1–7 counteracts Cid1-dependent 3′ uridylation to restrain 3′–5′ decay

Δlsm1 colonies were unstable on agar plates and frequently produced a second class of suppressor sectors with partial restoration of hyphal growth (Fig. 5A). Whole-genome sequencing of three independent suppressors identified recurrent disruptive mutations in FGSG_04263, including frameshift alleles and a missense substitution (Cid1^G903N) (Fig. 5B and Supplementary Fig. S9A). FGSG_04263 encodes a terminal uridyltransferase (TUTase) with a PAP/25A-associated domain and is conserved among filamentous fungi; phylogenetic analysis placed it with the Schizosaccharomyces pombe uridyltransferase Cid1 (Supplementary Fig. S9B). We therefore designated this gene as CID1 in this fungus. Consistent with a causal role, deletion of CID1 partially suppressed the growth defects of Δlsm1 and Δlsm2, and the Δcid1/lsm1 and Δcid1/lsm2 double mutants phenocopied the suppressor sectors (Fig. 5C). The Cid1^G903 residue is conserved among fungal homologs, and the Cid1^G903N allele partially rescued the Δlsm1 growth defect (Supplementary Fig. S9B and C). Molecular dynamics simulations further suggested that the G903N substitution altered Cid1 conformation and surface electrostatic potential, potentially affecting its binding to substrates and/or catalytic activity (Supplementary Fig. S9D and E).

Figure 5.

Cid1-mediated uridylation regulates the abundance and 3′-end features of Lsm1-bound mRNAs. (A) Δlsm1 colonies producing suppressor sectors (arrows), and three representative suppressors on CM. (B) Mutation sites in Cid1 associated with suppressors. (C) Growth phenotypes of PH-1 and the indicated mutants. (D) RNA-seq heatmap (log₂FC) for 110 Lsm1-bound transcripts in Δlsm1, Δcid1, and Δcid1/lsm1 relative to WT. (E) RT-qPCR validation of representative Lsm1-bound targets in WT and the indicated strains. Error bars indicate standard deviation from three biological replicates. Different letters indicate statistically significant differences (P <.05; one-way ANOVA followed by Fisher’s LSD test). (F) Cid1-GFP co-localizes with the P-body marker Scd6-RFP. (G) Schematic illustration of the Poly(A)-seq protocol. (H) KEGG enrichment of uridylated transcripts. (I) Fractions of Poly(A)U versus non-Poly(A)U genes among all detected transcripts (n = 10 444) and among Lsm1-bound genes (n = 271) in PH-1. (J) Mean fraction of poly(A), poly(A)U, and oligoU extensions mapped to mRNAs in different strains. (K) Clustered heatmap showing terminal U frequency in Lsm1-bound mRNAs across different strains. (L–N) Tail-length distributions and summary statistics (average/median) for poly(A), poly(A)U, and oligo(U) extensions on Lsm1-bound mRNAs in each strain. (O) Y2H interaction between Cid1 and core components of the RNA exosome (Csl4, Mtr3) and Pat1. (P) Co-IP validation of the Csl4–Cid1 association. (Q) Localization of GFP-Csl4 in Δcid1. Scale bar = 5 µm. (R) Growth phenotypes of PH-1 and the indicated mutants.

To define how Cid1 impacts the Lsm1-dependent transcript program, we profiled Δcid1 by RNA-seq. Δcid1 showed a largely derepressive signature, with 338 genes upregulated and 31 genes downregulated (|log₂FC| >1, FDR < 0.05) (Supplementary Fig. S10A–C and Supplementary Tables S4, S6, and S7), and upregulated transcripts were enriched for amino acid metabolism and glycolysis (Supplementary Fig. S10D). Notably, ∼30% (821/3002) of differentially expressed genes in Δlsm1 reverted toward wild-type levels in Δcid1/lsm1 (Supplementary Fig. S10E). Among Lsm1-bound mRNAs, 110 transcripts increased in abundance in Δcid1/lsm1 compared to Δlsm1 (Fig. 5D). RT-qPCR confirmed that representative Lsm1-bound glycolysis mRNAs (PGK1, FBA1, PGI1 and THD3), reduced in Δlsm1, were restored in Δcid1/lsm1 and in Δlsm1::Cid1^G903N (Fig. 5E and Supplementary Fig. S9F), linking transcript recovery to phenotypic suppression. Therefore, reduced Cid1 activity partially mitigates the accelerated transcript loss in Δlsm1.

Cid1-GFP co-localized with the P-body marker Scd6-RFP, indicating that Cid1 associated with P-bodies (Fig. 5F). Since Cid1 homologs mediate 3′ RNA uridylation to promote mRNA degradation, we next tested whether Cid1 controls 3′ uridylation of Lsm1-regulated transcripts. Poly(A)-seq revealed widespread 3′ uridylation in WT, affecting 34% (3574/10 444) of detected transcripts and enriched in primary metabolic pathways (Fig. 5G–I). Strikingly, 81% (220/271) of Lsm1-bound mRNAs carried uridylation signatures (Fig. 5H). Deletion of Cid1 reduced 3′ uridylation on Lsm1-bound transcripts, consistent with Cid1 being a major mRNA uridyltransferase in this fungus. In contrast, uridylation of the same Lsm1-bound set increased in Δlsm1 (Fig. 5J and K), consistent with Lsm1–7 normally limiting exposure of substrate 3′ ends to Cid1-dependent tailing. Poly(A) tail profiling showed longer tails in Δcid1 (average ∼48 nt/medium ∼41 nt) and shorter tails in Δlsm1 (∼42 nt/∼36 nt), compared to WT (∼43 nt/∼38 nt) (Fig. 5L–N), supporting distinct 3′-end remodeling states in these backgrounds.

To connect Cid1-dependent tailing to 3′–5′ decay machinery, we examined the Cid1 interaction network. AC-MS using GFP-Cid1 recovered multiple P-body proteins (Supplementary Table S8). Y2H and Co-IP confirmed the interactions between Cid1 and RNA exosome components Csl4, Mtr3 and Pat1 (Fig. 5O–P and Supplementary Fig. S11A). Consistent with a role in P-body recruitment, Csl4-GFP and Mtr3-GFP failed to accumulate in P-bodies in Δcid1, whereas Pat1-GFP localization was maintained (Fig. 5Q and Supplementary Fig. S11B). Moreover, deleting CID1 partially rescued the growth phenotype of Δpat1 (Fig. 5R), indicating that Cid1 activity contributes to the Δpat1 phenotype. Taken together, these results support a model in which Cid1-dependent 3′ uridylation promotes Ski–exosome-mediated 3′–5′ decay of susceptible transcripts, whereas Lsm1–7/Pat1 counteracts this pathway by protecting U/A-rich 3′ ends. Loss of CID1 reduces uridylation and decay-factor recruitment, thereby partially restoring Lsm1-bound metabolic mRNAs and alleviating Δlsm1 growth defects.

Lsm1–7 protects mRNA 3′ ends from parallel Cid1- and Lsp1-mediated 3′–5′ decay

The partial restoration of hyphal growth defects in Δlsm1 by individual mutations in CID1 or LSP1 suggested that these two regulators might function in parallel. To test this, we generated a triple mutant strain (Δlsp1/cid1/lsm1) and assessed its phenotypic rescue. Remarkably, the triple mutant exhibited ∼70% recovery in hyphal growth relative to WT, significantly exceeding the rescue observed in either Δlsp1/lsm1 or Δcid1/lsm1 double mutants (Fig. 6A and B). This additive genetic interaction suggests that Cid1 and Lsp1 function through independent yet converging pathways to regulate mRNA turnover in the absence of Lsm1–7. Consistent with this genetic interaction, transcriptomic analysis showed that ∼41% (45/110) of Lsm1-bound transcripts were restored toward wild-type levels in the two double mutants (Fig. 6C). RT-qPCR further confirmed that representative Lsm1-bound glycolysis mRNAs (PGK1, FBA1, PGI1, and THD3) were fully restored to wild-type levels in Δlsp1/cid1/lsm1 (Fig. 6D). Importantly, transcriptional shut-off assays revealed that the accelerated degradation rates of these substrate mRNAs in Δlsm1 were restored to near wild-type levels in the triple and double mutants (Supplementary Fig. S12A). In addition, the triple mutant partially restored mycotoxin production, expression of selected DON biosynthetic genes, and virulence on wheat heads (Fig. 6E–G).

Figure 6.

Cid1 and Lsp1 act additively to suppress accelerated decay in Δlsm1 and support fungal growth and virulence. (A) Growth phenotypes of PH-1 and the indicated mutants on CM. (B) Quantification of colony diameters from panel (A). (C) Venn diagram showing the overlap between the 110 Lsm1-bound mRNAs and transcripts significantly increased in Δlsp1/lsm1 and Δcid1/lsm1 relative to Δlsm1. (D) RT-qPCR validation of representative Lsm1-bound targets in WT and the Δlsp1/cid1/lsm1 triple mutant. (E) DON production of each strain after 7 days of culture in TBI medium. (F) RT-qPCR of TRI genes (TRI1, TRI5, TRI6, TRI10, TRI101) in the indicated strains. Fungal mycelia were harvested after 3 days of grown in TBI medium. (G) Virulence of the indicated strains on wheat heads at 14 days post-inoculation. Error bars indicate standard deviation from three biological replicates. Different letters indicate statistically significant differences (P <.05; one-way ANOVA followed by Fisher’s LSD test).

Integrating these genetic, omics, and phenotypic data, we propose a two-route model for Lsm1–7-mediated mRNA 3′ ends homeostasis (Supplementary Fig. S12B). In one route, a subset of transcripts acquires terminal uridylation by Cid1; binding of Lsm1–7/Pat1 to U-tailed 3′ termini limits access of the Cid1-associated exosome, thereby restraining 3′–5′ decay. In the other route, Lsm1–7/Pat1 engages transcripts with oligo(A)-rich 3′ ends and antagonizes Lsp1-dependent recruitment of the Ski–exosome complex. By protecting both U- and A-rich 3′ termini, Lsm1–7/Pat1 suppresses parallel 3′–5′ decay pathways and maintains the stability of metabolic transcripts required for fungal growth, development, and virulence.

Loss of Lsm1–7 engages a histone H4 acetylation-linked transcriptional compensation pathway

Beyond components of the mRNA decay machinery, our suppressor screen identified a third category of Δlsm1 suppressors with partially restored growth (Fig. 7A). Whole-genome sequencing identified a nonsense mutation (^G2709^A) in SIN3 (FGSG_09306), which encodes a core subunit of the evolutionarily conserved Rpd3L histone deacetylase (HDAC) complex [50] (Fig. 7B). Targeted deletion of SIN3 in the Δlsm1 background (Δsin3/lsm1) confirmed its role in rescuing hyphal growth defects (Fig. 7C), indicating that reduced Rpd3L/Sin3 function can buffer defects caused by loss of Lsm1–7.

Figure 7.

Rpd3L/Sin3 disruption partially suppresses Δlsm1 defects by elevating histone H4 acetylation and restoring Lsm1-target expression. (A) Δlsm1 colonies showing spontaneous fast-growing suppressor sectors (arrows) and two representative suppressors on CM. (B) Mutation site in Sin3 associated with suppressor formation. (C) Growth phenotypes of PH-1 and the indicated mutants on CM. (D) Venn diagram illustrating differentially expressed genes between Δlsm1 and Δsin3/lsm1 after 24 h in CM (upregulated, light red; downregulated, light purple). (E) RNA-seq heatmap of log₂FC values for 110 Lsm1-bound mRNAs in Δlsm1, Δsin3, and Δsin3/lsm1 mutants relative to WT. (F) RT-qPCR validation of representative Lsm1-bound substrates in PH-1 and the indicated mutants. (G) Immunoblot analysis of global histone H4 acetylation (pan-H4ac) and site-specific marks (H4K5ac, H4K8ac, H4K12ac) in strains cultured as in (D); total H4 serves as a loading control. (H) Overlap of H4ac ChIP-seq peaks identified in PH-1, Δsin3, and Δlsm1 mutants. (I) Metagene profiles of mean H4ac signal across genes (±2 kb from peak summits); TSS, transcription start site; TTS, transcription termination site. (J) RT-qPCR analysis of BIOF (FGSG_09048) gene expression in PH-1 and indicated mutants. (K) ChIP-qPCR showing H4ac enrichment at the BIOF promoter and gene body. Error bars indicate standard deviation from three biological replicates. Different letters indicate statistically significant differences (P <.05; one-way ANOVA followed by Fisher’s LSD test).

Transcriptomic analysis revealed that Sin3 primarily functions as a transcriptional repressor (Supplementary Fig. S13A–C and Supplementary Tables S4, S6, and S7). Notably, its loss in Δlsm1 restored the expression of ∼40% of differentially expressed genes toward wild-type levels, and the restored set was enriched for primary metabolic pathways (Fig. 7D and E, and Supplementary Fig. S13D and E). This compensation preferentially affected direct Lsm1 targets: among the 110 Lsm1-bound mRNAs, 85 showed partial or near-complete restoration in Δsin3/lsm1 (Fig. 7E). RT-qPCR confirmed recovery of representative glycolysis-related Lsm1 targets, including FBA1 and PGI1 (Fig. 7F). These data suggest that relieving Sin3-dependent repression can partially bypass the requirement for Lsm1–7-dependent maintenance of metabolic mRNA abundance.

Given that Sin3 is a key scaffold for histone deacetylation, we investigated whether the loss of Lsm1–7 triggers an epigenetic response. Western blot analysis showed that both Δsin3 and Δlsm1 mutants exhibited elevated global levels of histone H4 acetylation (H4ac), including pan-H4 acetylation and site-specific marks (H4K5ac, H4K8ac, H4K12ac) (Fig. 7G). We then mapped H4ac genome-wide by ChIP-seq. WT contained 9830 significant H4ac peaks (q < 0.05), primarily at promoters and gene bodies. Δlsm1 showed 7515 peaks and moderate, promoter-biased H4ac gains at a subset of loci, consistent with an endogenous but limited transcriptional compensation response. In contrast, Δsin3 exhibited a strong expansion to 16 838 peaks (q < 0.05) with broadly elevated H4ac across promoters and gene bodies, a pattern retained in Δsin3/lsm1 (Fig. 7H and I, Supplementary Fig. S14, and Supplementary Table S9). At representative Lsm1 target loci, restoration of transcript abundance in Δsin3/lsm1 coincided with marked H4ac enrichment. For example, the biotin biosynthesis gene BIOF (FGSG_09048) was downregulated in Δlsm1 but restored toward wild-type levels in Δsin3/lsm1 (Fig. 7J), accompanied by pronounced H4ac increases at both promoter and gene body in Δsin3 and Δsin3/lsm1 (Fig. 7K). Together, these results define a chromatin-based transcriptional buffering route that partially compensates for accelerated loss of Lsm1-regulated transcripts. The genetic suppression by SIN3 and the accompanying increase in H4 acetylation support a model in which reduced Rpd3L/Sin3-mediated deacetylation enhances transcriptional output of key metabolic genes, thereby offsetting mRNA depletion when Lsm1–7-dependent 3′-end protection is compromised (Fig. 8).

Figure 8.

Integrated model of Lsm1–7-mediated mRNA homeostasis and nucleocytoplasmic feedback in F. graminearum. In wild-type cells, the Lsm1–7/Pat1 complex acts as a physical shield at mRNA 3′ ends within P-bodies. This occupancy provides steric occlusion against two parallel 3′–5′ decay routes: the RNA exosome recruited by Cid1 and the Ski–exosome complex scaffolded by Lsp1. In this protected state, nuclear Rpd3L deacetylase activity maintains basal histone H4 acetylation (H4ac) and transcriptional output. Conversely, loss of Lsm1–7 protection exposes metabolic transcripts to rapid exonucleolytic degradation through these dual routes. This accelerated cytoplasmic decay triggers a compensatory “transcriptional buffering” response in the nucleus. The resulting feedback loop leads to reduced Rpd3L-mediated deacetylation and enhanced H4ac at affected promoters, boosting transcription to offset the loss of post-transcriptional stability. This integrated axis, coupling 3′-end protection to nuclear chromatin remodeling, is indispensable for the mRNA homeostasis required for F. graminearum growth, mycotoxin biosynthesis, and pathogenesis. 

Discussion

P-bodies are traditionally regarded as cytoplasmic granules that are enriched for mRNA decay factors and associated with 5′–3′ mRNA turnover [7, 8]. The Lsm1–7 complex is a core structural and functional component of these granules [51]. Lsm1–7 canonically binds deadenylated mRNAs, often with short U/A-rich 3′ tails, and promotes decapping followed by Xrn1-mediated degradation [24, 25]. Recent evidence suggests that P-bodies can also serve as mRNA reservoirs, and that Lsm1–7 can protect subsets of transcripts by limiting exosome access to 3′ ends [52, 53]. However, how Lsm1–7 stabilizes its substrates in vivo has remained unclear. Our results refine this model by indicating that Lsm1–7 stabilizes substrates, at least in part, through physical protection at the 3′ end. Lsm1-bound mRNAs are rapidly depleted in Δlsm1, consistent with accelerated decay. Mechanistically, Lsm1–7, via Pat1, restricts 3′–5′ degradation by limiting access of (i) a Cid1-associated exosome pathway and (ii) an Lsp1-associated Ski–exosome pathway. These findings add a post-transcriptional layer of regulation within P-bodies.

The identification of Lsp1 provides an evolutionary perspective on the organization of decay machinery in filamentous fungi. While Cid1-like uridyltransferases are conserved across eukaryotes [24, 25, 54, 55, 56], Lsp1 exhibits sequence-level conservation within the Pezizomycotina. Structural modeling nevertheless suggested similarity between Lsp1 and eukaryotic Edc4 orthologs, particularly within the WD40 domain, despite low overall sequence identity (<15%). This pattern is consistent with structural convergence and suggests that Lsp1 may act as a Pezizomycotina-specific Edc4-like scaffold [57–59]. Since both Cid1 and Lsp1 are P-body–associated factors linked to recruitment of 3′–5′ decay machinery, our findings support a model in which Lsm1–7-mediated 3′-end protection is coordinated with local decay-factor recruitment in P-bodies, shaping mRNA fate in filamentous fungi.

The Lsm1–7 complex is broadly conserved across eukaryotes, although its physiological roles vary across lineages [60]. In F. graminearum, Lsm1–7 binds mRNAs encoding key enzymes in glycolysis and the TCA cycle and supports their abundance. This points to a conserved link between Lsm1–7 and metabolic mRNA regulation, consistent with observations in S. cerevisiae, where glycolytic mRNAs associate with Lsm1–7-containing mRNPs to form the structural platform for P-body assembly [61–63]. Our data are also consistent with mammalian studies showing that Lsm1 promotes metabolic reprogramming by stabilizing glycolytic transcripts [30, 31]. The functional conservation of Lsm1–7 across such diverse lineages, ranging from fungal development and pathogenesis [26, 64] to abiotic stress responses in Arabidopsis [27, 28], highlights its broad role in coordinating mRNA fate decisions. In F. graminearum, this specialized stabilization is essential for meeting the energetic demands of growth, virulence, and mycotoxin biosynthesis, suggesting that perturbing this axis could be explored for antifungal intervention.

mRNA homeostasis depends on a dynamic balance between synthesis and degradation. Emerging evidence suggests that cytoplasmic RNA decay feeds back to the nucleus to influence transcription through a homeostatic feedback loop [65, 66]. On the one hand, nuclear mRNA output can influence cytoplasmic mRNA homeostasis. When transcription is impaired, mRNA stability can increase to buffer reduced synthesis. For example, in mouse embryonic fibroblasts, deletion of the TFIIH kinase subunit MAT1 strongly reduces Pol II CTD Ser5 phosphorylation and nascent transcription, yet steady-state mRNA levels decline only partially due to globally slowed decay and extended transcript half-lives [67]. In S. cerevisiae, cytoplasmic RNA-binding protein Puf5 slows the degradation of its target mRNAs, maintaining their steady-state levels in the H3K56A background [45]. On the other hand, cytoplasmic mRNA decay can feed back to the nucleus, as certain P-body RNA decay factors shuttle between compartments and coordinate transcription and mRNA stability, contributing to mRNA buffering [68]. For example, in S. cerevisiae, deletion of the 5′–3′ exonuclease Xrn1 slows RNA polymerase II elongation and reduces transcriptional output, forming compensatory feedback that maintains global mRNA homeostasis [69–72]. Also, Puf3, Upf2, and Upf3 stimulate the decay of specific mRNAs and repress transcription, forming an additive crosstalk that promotes global control of mRNA levels [73]. Our data support a nucleocytoplasmic buffering response upon loss of Lsm1–7. In Δlsm1, accelerated 3′–5′ decay is accompanied by increased H4 acetylation at promoters, and genetic disruption of the Rpd3L component Sin3 partially restores transcript levels. This chromatin-based response is consistent with transcriptional buffering that offsets reduced mRNA abundance, which aligns with recent models of “mRNA buffering”. This transcriptional buffering response provides a crucial compensatory mechanism to counteract the effects of compromised post-transcriptional regulation. Our study extends these ideas by showing that chromatin modification and mRNA decay are functionally coupled, with Sin3 loss increasing H4 acetylation and partially restoring expression of key targets in Δlsm1.

In summary, our findings define a P-body-centered module in which Lsm1–7/Pat1 limits 3′–5′ decay through Cid1- and Lsp1-associated pathways and reveal a chromatin-based buffering response involving Rpd3L/Sin3. These findings in F. graminearum inform conserved principles of mRNA homeostasis and highlight lineage-specific adaptations of P-body scaffolds. Future work should address how cytoplasmic mRNA turnover is sensed and coupled to chromatin regulation.

Supplementary data

Supplementary data is available at NAR Online.

Conflict of interest

None declared.

Funding

This work was funded by supported by the National Natural Science Foundation (32172356), the Natural Science Foundation of Zhejiang Province (LZ23C140004), National Key R&D Program of China (2022YFD1400100), “Pioneer” and “Leading Goose” R&D Program of Zhejiang (2023C02030), China Agriculture Research System (CARS-3-1-15) and Fundamental Research Funds for the Central Universities (226-2024-00213, 226-2024-00070). Funding to pay the Open Access publication charges for this article was provided by National Natural Science Foundation (32172356).

Data availability

The raw sequencing data generated in this study have been deposited in the Genome Sequence Archive (GSA) hosted by the National Genomics Data Center (NGDC). The RNA-seq, RIP-seq, ChIP-seq and Poly(A)-seq raw data, along with processed information, are available under the BioProject ID PRJCA038704 (https://ngdc.cncb.ac.cn/bioproject/browse/PRJCA038704).