Telomere-to-telomere genome assembly of Metarhizium acridum provides insights into Metarhizium genome evolution Authors: Qian Zhao, Xin Sun, Xi Li, Chunmei Yang, Guangjun Wang, Zehua Zhang, Shiqian Feng, Hui Dong, Xiongbing Tu Published: 2026 ABSTRACT Metarhizium acridum is a highly specialized entomopathogenic fungus widely used as a biocontrol agent against locusts. Despite its ecological and agricultural importance, the genomic basis of its host specificity and pathogenicity remains poorly understood. We present the first telomere-to-telomere (T2T) genome assembly of M. acridum, using PacBio HiFi sequencing and Hi-C scaffolding data, for seven chromosomes (45.1 Mb, N50 = 5.9 Mb) with 96.7% BUSCO completeness, which is the most contiguous Metarhizium genome to date. Repeat elements constitute 19.7% of the genome, dominated by long terminal repeat (LTR) retrotransposons (8.6%) while genome annotation revealed 11,623 protein-coding genes, including 716 carbohydrate-active enzymes and 254 effector proteins. Comparative genomics analysis identified 257 genes shared between M. acridum and M. anisopliae, 465 genes shared between M. acridum and M. brunneum. Synteny analysis indicated frequent chromosomal rearrangements between M. acridum and M. brunneum, with fragments from one chromosome of M. acridum distributed to 2–6 chromosomes of M. brunneum. The aegerolysin-domain toxin gene MaAh-1A showed approximately 10-fold upregulation during locust infection. After trypsin-activated of MaAh-1A, an approximately 16% decline of survival rate was observed. Our study provides an ideal reference T2T genome as the foundational resource for exploring genomic resources of biological control and advancing the application of Metarhizium fungi in sustainable pest management.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12864-026-12599-w. Introduction The genus Metarhizium (Hypocreales: Clavicipitaceae) is widely recognized for its insect-pathogenic properties and has been extensively used in biological pest control [1]. Historically, species in this genus were grouped into a broad species complex known as M. anisopliae based on morphological traits. However, advances in molecular phylogenetics and genomics have revealed substantial genetic divergence within this complex, leading to its reclassification into several distinct species including specialists (M. acridum), which infect only locusts [2] (Fig. 1), and generalists (M. anisopliae, M. robertsii and M. brunneum etc.), which infect a broad range of insects [3]. The divergence of Metarhizium species and their adaptation to different host has been an interesting case study for understanding fungal evolution and pathogen-host interactions [4, 5]. It has revealed that broad-host-range Metarhizium species evolved stepwise from specialized ancestors to a wider host-range species, gradually losing their sexual reproductive capability to rapidly adapt to diverse host environments. In generalists, protein families associated with virulence and environmental sensing have undergone significant expansion, while genome defence mechanisms have weakened, facilitating genomic rearrangements and the acquisition of novel genes [6, 7]. Fig. 1M. acridum infection in Locusta migratoria at 12 days post-inoculation. a Dorsal view. b Lateral view. Scale bar = 10 mm (calibrated) M. acridum infection in Locusta migratoria at 12 days post-inoculation. a Dorsal view. b Lateral view. Scale bar = 10 mm (calibrated) However, fragmentation at the assembly level can reduce the accuracy of the analysis. Chromosome-level assembly can greatly reduce assembly fragmentation (contigs/scaffolds), minimizing gene loss or misplacement [8, 9]. It enables precise localization of genes, regulatory elements (such as promoters and enhancers), and structural variations (such as insertions, deletions, and inversions) [10–12]. This is of great significance for revealing chromosomal structural changes and mechanisms of species differentiation. To date, a total of twelve Metarhizium genomes have been published, among which only two (M. brunneum and M. pinghaense) have been assembled to the chromosome level [13]. M. acridum is currently the only known obligate entomopathogenic fungus to Orthoptera insects. However, its genome sequencing is available only at draft level, which hampers the accurate identification and localization of gene structures and regulatory elements. This limitation restricts in-depth studies on gene function annotation, comparative genomics, and structural variation analysis. To elucidate the genomic basis of host adaptation in the locust-specific pathogen M. acridum, we generated a complete telomere-to-telomere (T2T) genome assembly through PacBio sequencing, Hi-C technology and conducted comparative genomics across eight broad-host-range Metarhizium species (Table 1). Our integrated analyses revealed structural variations and evolutionary signatures in the host. Through functional characterization, we demonstrated that the asp-hemolysin protein MaAh-1A, Ah-1A from M. acridum, helps to kill the locusts. This study provides genomic resources and mechanistic insights into fungal entomopathogenicity, facilitating the development of improved biocontrol strategies. Table 1Genome statistics in genus MetarhiziumGenome Features M. acridum M. brunneum M. anisopliae M. album M. guizhouense M. humberi M. hybridum M. rileyi M. robertsii Genbank No.the present studyGCA_013426205.1GCA_039654215.1GCA_000804445.1GCA_000814955.1GCA_020102295.1GCA_000814975.1GCA_007866325.1GCA_000591435.1Genome Size (MB)45.137.839.530.443.538.638.531.840.3Chromosomes/scaffolds/contigs No.7828257563107274249367MethodsPacBio+Hi-CNanopore+IlluminaPacBioIlluminaIlluminaIlluminaIlluminaIllumina454Chromosomes/scaffolds/contigs N505.9 MB4.8 MB4.2 MB1.1 MB554 KB1.4 MB2.0 MB815 KB566 KBGenes11,62311,42012,7468,47211,78710,54610,89189,4412,369BUSCO Evaluation98.9%99.1%90.3%96.2%96.4%95.4%97.1%97.7%99.5% Table 1Genome statistics in genus MetarhiziumGenome Features M. acridum M. brunneum M. anisopliae M. album M. guizhouense M. humberi M. hybridum M. rileyi M. robertsii Genbank No.the present studyGCA_013426205.1GCA_039654215.1GCA_000804445.1GCA_000814955.1GCA_020102295.1GCA_000814975.1GCA_007866325.1GCA_000591435.1Genome Size (MB)45.137.839.530.443.538.638.531.840.3Chromosomes/scaffolds/contigs No.7828257563107274249367MethodsPacBio+Hi-CNanopore+IlluminaPacBioIlluminaIlluminaIlluminaIlluminaIllumina454Chromosomes/scaffolds/contigs N505.9 MB4.8 MB4.2 MB1.1 MB554 KB1.4 MB2.0 MB815 KB566 KBGenes11,62311,42012,7468,47211,78710,54610,89189,4412,369BUSCO Evaluation98.9%99.1%90.3%96.2%96.4%95.4%97.1%97.7%99.5% Genome statistics in genus Metarhizium Chromosomes/scaffolds/ contigs No. PacBio+ Hi-C Nanopore+ Illumina Chromosomes/scaffolds/ contigs N50 Methods Samples collection Sample of M. acridum (IPPC024 strain) was collected in Xilinhot, Inner Mongolia Autonomous Region, China (Latitude 43°40.55’ N; Longitude 116°22.13’ E). The strain was placed on the Potato Dextrose Agar Yeast Extract (PDAY) medium under dark environment. DNA extraction, RNA extraction, library construction, and sequencing Genomic DNA of 0.6 g hyphae was extracted using FastPure Microbiome DNA Isolation Kit (Vazyme, DC502). Following DNA quality assessment, a 15 kb HiFi library was constructed according to the recommended pipeline, involving DNA damage and end repair, hairpin adapter ligation and purification of the library. After quality control verification, a SMRTbell library was sequenced on the PacBio Revio platform (Pacific Biosciences, CA, USA). After sequencing, the raw data (polymerase reads) was transferred into HiFi reads through SMRT Link functions. For transcriptome analysis, total RNA was isolated from 0.4 g mixture of hyphae using FastPure® Universal Plant Total RNA Isolation Kit (Vazyme, RC411). mRNA was reverse transcribed into cDNA using the FastKing gDNA Dispelling RT SuperMix Kit (TIANGEN, KR118). The Oligo dT primer consists of a series of consecutive thymine nucleotides that specifically anneal to the poly-A tail at the 3’ end of mRNA. During reverse transcription, only mRNAs with a poly-A tail are targeted to generate cDNA from the 3’ end toward the upstream. Followed by library construction with the BGI DNBSEQ-T7 platform according to the manufacturer’s instructions. Sequencing was performed in paired-end 150 bp mode (insert size 350 bp), generating approximately 6 GB of data. All sequencing services were provided by Berry Genomics Company. Hi-C sequencing Approximately 0.6 g hyphae were processed for Hi-C analysis. The experimental procedure included: formaldehyde cross-linking of homogenized samples, DNA digestion with MboI restriction enzyme, biotin labeling of fragment ends, blunt-end ligation, and random shearing DNA into fragments (300–500 bp). The prepared library was sequenced on the BGI DNBSEQ-T7 platform, generating ~ 10 GB of 150 bp paired-end reads. Genome assembly The PacBio HiFi data was processed and assembled using hifiasm v0.19.6 [14] using default parameters, with the primary mode selected. The PacBio sequencing data was mapped to the draft genome using pbmm2 v1.4.0 and arrow algorithm was used for assembly polish using gcpp v1.9.0 (Supplementary File 1) (Pacific Biosciences official toolkits, https://github.com/PacificBiosciences). Redundant sequences were removed with purge_dups v1.2.5 (Supplementary File 2) [15]. The purge_dups was applied only to the primary assembly. Indeed, no additional polishing steps were included at this stage. Filtered Hi-C sequencing reads were aligned against the polished genome reference using BWA v0.7.17 [16], with alignment processing performed through the integrated Juicer workflow. 3D-DNA v180114 [17] was employed for chromosomal scaffolding using exclusively uniquely mapped and valid paired-end reads, followed by manual order the scaffold with Juicebox [18] to achieve the final chromosome-level assembly. The completeness of the final genome assembly was evaluated using BUSCO v5.1.3 [19] with hypocreales_odb10 database. Genome annotation To confirm the completeness of the telomere-to-telomere (T2T) assembly, telomeric regions were identified by scanning tandem repeats using TRF v4.09 [20] with the following parameters: match score = 2, mismatch penalty = 5, indel penalty = 7, minimum alignment score = 80, maximum period size = 10, and maximum repeat length = 2000. The canonical telomeric repeat motif (TTAGGG) was searched in the TRF output, and regions within 1 kb of scaffold ends were examined. All seven chromosomes exhibited terminal (TTAGGG)n repeats at both 5’ and 3’ ends, confirming the T2T assembly completeness. The length of telomeric repeats ranged from 123 bp to 214 bp (Table 2). Additionally, PacBio HiFi reads aligned uniformly across the termini with no gaps (visualized in IGV), further supporting the integrity of telomeric regions. Table 2Telomere locations across seven chromosomes with TTAGGG as typical repeat elementsChromosome5’ ends3’ endsStartEnd(TTAGGG)nStartEnd(TTAGGG)n 1 1139227,891,0547,891,19523 2 1123217,530,3307,530,47123 3 1136236,911,6736,911,79620 4 1151266,760,1156,760,24321 5 1127205,928,2505,928,39224 6 1125205,673,4485,673,58023 7 1137224,350,5994,350,81232 Table 2Telomere locations across seven chromosomes with TTAGGG as typical repeat elementsChromosome5’ ends3’ endsStartEnd(TTAGGG)nStartEnd(TTAGGG)n 1 1139227,891,0547,891,19523 2 1123217,530,3307,530,47123 3 1136236,911,6736,911,79620 4 1151266,760,1156,760,24321 5 1127205,928,2505,928,39224 6 1125205,673,4485,673,58023 7 1137224,350,5994,350,81232 Telomere locations across seven chromosomes with TTAGGG as typical repeat elements We employed a comprehensive repeat annotation pipeline. MITE elements were identified using MITE-Hunter [21] with parameters: -n 5 -c 20. LTR retrotransposons were predicted using LTRharvest [22] (similarity 90%, vicinal distance 10 bp, seed length 20 bp, LTR length 100–7000 bp, TSD 4–6 bp) and LTR_Finder (-D 15000 -d 1000 -L 7000 -l 100 -p 20 -C -M 0.9), with results integrated using LTR_retriever v2.9.5 [23]. Known repeats were identified using RepeatMasker v4.1.5 [24] against RepBase. The de novo repeat prediction was performed using RepeatModeler v2.0.3 [25] with NCBI engine. All repeat libraries were combined for final masking using RepeatMasker. Moreover, tRNAscan‑SE 2.0 (tRNA) [26], and Infernal + Rfam (snRNA, snoRNA, rRNA) [27] were used to identify these regions, merge and de‑duplicate the outputs to obtain a complete non‑coding RNA annotation (Supplementary File 3). We employed an evidence-based prediction pipeline. Homology-based prediction was conducted using GeMoMa v1.9 [28] using M. acridum ARSEF 324 as reference. RNA-seq alignments were processed using PASA v2.5.2 [29] with stringent alignment overlap (30%) using both BLAT [30] and GMAP [31] aligners. Moreover, three predictors were applied for Ab initio prediction: Augustus v3.5.0 [32] with species-specific parameters, GlimmerHMM v3.0.4 [33] with fungal parameters and GeneMark-ES v4 [34] with fungal parameters and intron hints. All predicting results were integrated using EVM r2012-06–25 [35] with weights optimized for each evidence type (Supplementary Table 1). Functional annotations were also conducted. Antibiotic resistance was analysed with CARD database using RGI v6.0.3 [36]. Carbohydrate-active enzymes was detected using run_dbCAN v4.1.4 [37] with default parameters. Nonetheless, general protein annotation was conducted by DIAMOND v2.1.8 [38] searches against MEROPS (peptidases), PHI-base (pathogen-host interactions), SwissProt (curated proteins) and TCDB (transporters) with an e-value cut-off value 1e−5 and max 10 hits per query. The above analysis results are presented in Supplementary Table 2. Effector proteins was predicted using SignalP 6.0 [39] to predict 987 proteins containing signal peptides (Supplementary Table 3), and TMHMM 2.0 [40] to predict 9,471 proteins without transmembrane domains (Supplementary Table 4). Finally, 793 proteins in the intersection were further analyzed using EffectorP v3.0 [41], resulting in 254 predicted effectors (Supplementary Table 5). Synteny detection, orthology prediction and phylogenetic analyses JCVI [42] was used to detect syntenic blocks between M. acridum (the present study) and M. brunneum (the only published and available Metarhizium genome in chromosome level). Those blocks with ≥ 8 genes were used (--minspan = 8). Metarhizium species with a published genome including M. album (GCA_000804445.1), M. anisopliae (GCA_039654215.1), M. brunneum (GCA_013426205.1), M. guizhouense (GCA_000814955.1), M. humberi (GCA_020102295.1), M. hybridum (GCA_000814975.1), M. rileyi (GCA_007866325.1) and M. robertsii (GCA_000591435.1) were selected for orthology analysis with Pochonia chlamydosporia (GCA_001653235.2) from the same family as the outgroup. Gene family classification (including orthologous and paralogous) was detected with OrthoFinder 2.5.5 [43] using default settings. Following MAFFT [44] alignment and concatenation of all single-copy orthologous protein sequences, phylogenetic reconstruction was performed with FastTree [45], MCMCTREE (PAML package v4.9 [46]) was used to date this phylogenetic tree. We retrieved the divergence time between (1) M. brunneum and M. anisopliae (10 MYA, million years ago), (2) M. acridum and M. brunneum (7–20 MYA) from TimeTree database [47]. Domain architecture prediction for the curated non-redundant proteomes of M. acridum, M. brunneum, and M. anisopliae was conducted with HMMER v3.1b2 [48]. With an e-value threshold set at 1e−10, the protein domains of M. acridum were systematically compared against those of M. brunneum and M. anisopliae. Domain counting and comparison was performed using custom scripts, with subsequent visualization generated by the ggplot2 package [49] in R v4.0.4 [50] (Supplementary File 4). Ah-1A sequence and phylogeny BLAST 2.14.1+ [51] was employed to detect Ah-1A proteins in Metarhizium species using a dataset including Ah-1A proteins from M. anisopliae, M. brunneum and Aspergillus awamori, respectively. The cut-off e-value was set to 1e−10. Metarhizium species with the Ah-1A proteins were used for phylogenetic analyses. IQ-TREE 2.2.4 [52] was used for phylogenetic reconstruction under automatic model searching mode with bootstrap numbers of 1,000 and maximum iterations of 1,000. Domains were figured using TBtools-II [53]. qRT-PCR of MaAh-1A during infesting locusts Conidiospores of M. acridum were harvested from PDAY medium and diluted to a concentration of 1 × 106 conidiospores mL−1 in 0.01% Tween-80 solution. Conidiospores (5 µL) were injected into the hemolymph of fifth-instar nymphs of L. migratoria using a microsyringe. The midguts were collected at 0, 2, and 4 days post-injection during the course of infection. Before dissection, the locusts were soaked in 75% ethanol for 3 min to disinfect the cuticle, then thoroughly rinsed with sterile water to remove any residual ethanol. Midguts infested by M. acridum were isolated from fifth-instar nymphs of L. migratoria. For each sample, 10 locusts were used for collecting midguts. The methods for RNA extraction and reverse transcription to cDNA were performed as previously described. Through the 2−ΔΔCt method with triplicate biological and technical replicates, we systematically quantified relative expression changes of target genes [54]. For quantification, the Rp49 gene [55] was used as the endogenous control. The qPCR primers are listed in Supplementary Table 6. Then, qPCR was performed using Taq Pro Universal SYBR qPCR Master Mix (Vazyme, Q712) and the QuantStudio™ 5 Real-Time PCR Instrument (ThermoFisher Scientific, Shanghai, China) under the following conditions: initial denaturation (95℃, 30 s); 40 cycles of 95℃ for 10 s and 60℃ for 30 sec. MaAh-1A protein expression and functional validation Fragments of MaAh-1A and MaAh-1B were cloned into the expression vector pET30a (NovoPro, China) containing Hexa-histidine (His) fragment. Primers were listed in Supplementary Table 6. The pET30a-MaAh-1A and pET30a-MaAh-1B plasmids transformation were performed using chemically competent Escherichia coli strain BL-21(DE3) cells following standard heat-shock protocols (TSINGKE, DLC201). E. coli cultures were maintained in LB broth (1% NaCl, 1% tryptone, 0.5% yeast extract, pH 7.0) containing 50 µg/mL kanamycin at 37℃ with 200 rpm until reaching an OD600 of 0.6. Expression of target proteins (MaAh-1A and MaAh-1B) was induced with 0.1 mM Isopropyl β-D-thiogalactosid (IPTG) under low-temperature conditions (16℃, 20 h) to enhance soluble expression. Bacterial cells were then collected by low-speed centrifugation (5,000 × g, 10 min) at 4℃ to maintain protein stability. Cell lysates were resuspended in phosphate-buffered saline (PBS; 10 mM Na2HPO4, 2 mM NaH2PO4, 135 mM NaCl, 4.7 mM KCl, pH 7.4) and then sonicated (SCIENTZ, JY96-IIN, China). The soluble fraction was separated by centrifugation at 8,000 × g for 10 min at 4℃ to remove insoluble aggregates and cellular debris. The proteins were purified using a protein purification kit, HyPur T Ni-NTA 6FF (His-Tag) PrePacked Gravity Column (Sangon Biotech, C600793). The MaAh-1A protein was eluted using 100 mM imidazole, while the MaAh-1B protein was eluted with 50 mM imidazole. The eluted fractions were concentrated and subjected to buffer exchange with PBS (pH 7.4) using Amicon Ultra-15 centrifugal filter units: Ultracel®−10 K (Millipore, UFC901024) for MaAh-1A and Ultracel®−30 K (Millipore, UFC903096) for MaAh-1B. Protein concentration was determined using the BCA Protein Assay Kit (Solarbio, PC0020). Trypsin activity in insect samples was quantified using an Insect Trypsin ELISA Detection Kit (MLBio, YJ036260). Midgut fluid was extracted from ten fifth-instar nymphs using 100 µL of ice-cold PBS [56]. Hemolymph was also collected from ten fifth-instar nymphs, followed by centrifugation at 13,000 x g for 15 min at 4℃. The ELISA kit employs a sandwich-based method to specifically detect trypsin in insect extracts. To assess trypsin processing of MaAh-1A and MaAh-1B proteins, working solutions of trypsin (1.0 mg/mL in PBS) (MCE, HY-129047) were prepared. The proteins were incubated with trypsin at a mass ratio of 1:100 (trypsin: MaAh-1A or MaAh-1B) at 37℃ for 1 h. The extent of protein processing was evaluated by SDS-PAGE, followed by in-gel staining using Coomassie Brilliant Blue G-250 Staining Buffer (GenStar, E154). To evaluate the virulence of MaAh-1A and MaAh-1B proteins, bioassays were conducted via direct injection into fifth-instar nymphs. Proteins, either untreated or pre-digested with trypsin, were prepared at a concentration of 2.5 mg/mL, and 5 µL of MaAh-1A, MaAh-1B, or a mixture of MaAh-1A and MaAh-1B was injected into each nymph using a microsyringe. Nymphs injected with sterile PBS (pH 7.4) served as controls. Each treatment group consisted of 45 nymphs. Following injection, the insects were maintained at 28 ± 1℃, 60 ± 5% relative humidity, and a 16:8 h light/dark cycle for 6 days. Nymph mortality was recorded every 12 h. Statistical analysis Statistical processing was implemented through DPS 7.05 [57] for basic analyses and GraphPad Prism 10.1.2 [58] for advanced tests. Quantitative comparisons of gene expression employed one-way ANOVA with least significant difference (LSD) post hoc testing for multiple comparisons. For survival data, the Mantel-Cox version of the log-rank test was applied. In all analyses, results with P-values below 0.05 were deemed statistically significant. Results Genome sequencing and assembly In total 4.4 GB PacBio HiFi data (247,437 reads) was obtained which is about 98-fold coverage with the N50 at 17.8 KB (Supplementary Fig. 1). The hifiasm assembler produced a draft genome comprising 108 contigs totaling 49.4 MB, with an N50 contig size of 5.7 MB. After correcting the CCS sequences with the HiFi sequencing mode, we retrieved 8 contigs with a total size of 45.1 MB and N50 at 5.7 MB with BUSCO completeness at 96.7% [S: 96.4%, D: 0.3%], F: 0.8%, M: 2.5%, n: 4494. This version of genome was used in the subsequent Hi-C analysis. We generated 31,774,276 Hi-C read pairs (9.5 Gb) for chromosome-level genome assembly. Following alignment to the purged assembly, 96.4% (24,088,436/25,000,712) of uniquely mapped read pairs showed valid interaction patterns, the cis/trans ratio is 2.4 and cis/(cis+trans) proportion is 70.59%, confirming a good quality of Hi-C data (Fig. 2a, Supplementary Tables 7 and 8). After manually checking, we obtained a genome assembly with the 7 chromosomes covering 45.1 Mb (> 99% genome bases). These chromosomes ranged from 4.4 Mb to 7.9 Mb in lengths (Fig. 2b) and reached a scaffold N50 of 5.9 Mb. The BUSCO evaluation result (C: 96.7% [S: 96.5%, D: 0.2%], F: 0.8%, M: 2.5%), together with consistent GC percentages and stable coverage (Supplementary Fig. 2), suggested minimal contamination and high assembly quality. Compared to genomes available in Metarhizium (Table 1), M. brunneum showed similar quality to M. acridum in N50 and gene comparisons. M. acridum possessed the biggest genome size and N50 across Metarhizium species, as well as a high BUSCO completeness, which indicated a high-quality reference genome for the genus Metarhizium. Fig. 2Genome assembly, Hi-C and synteny. a Hi-C interaction map produced by HiC-Pro. b Circos plot of 7 chromosomes with the outer 7 bars representing chromosomes, while the orange, blue, purple colors represented GC content, gene counts and repeat sequences in 50 kb windows. c Synteny between two Materhizium species with chromosomal level genomes: M. acridum and M. brunneum, while inversions and translocations in the right line chart Genome assembly, Hi-C and synteny. a Hi-C interaction map produced by HiC-Pro. b Circos plot of 7 chromosomes with the outer 7 bars representing chromosomes, while the orange, blue, purple colors represented GC content, gene counts and repeat sequences in 50 kb windows. c Synteny between two Materhizium species with chromosomal level genomes: M. acridum and M. brunneum, while inversions and translocations in the right line chart Genome annotations Genome structural annotation diagnosed 8,853,299 bp of repetitive sequences (19.7% of the M. acridum genome), with retrotransposons (8.6%) and DNA transposons (0.8%) constituting the major repeat categories (Supplementary Table 9). Among retroelements, 8.57% genome sequences are long terminal repeat (LTR) with only 0.01% as short interspersed elements (SINEs). For DNA transposons, 0.75% are typical sequences and 0.06% are PIF-Harbinger. There are also unclassified repeat sequences (10.23%). Nonetheless, through non-coding RNA detection, we found 33 rRNAs, 48 tRNAs, 5 snRNAs and 18 snoRNAs in the genome sequences (Supplementary Table 10). Genome annotation identified 11,623 genes in M. acridum, with an average gene length of 2,108 bp (Supplementary Table 11). The BUSCO result of this gene set is C: 98.9% [S: 98.7%, D: 0.2%], F: 0.7%, M: 0.4%, indicating a good quality of structural annotation. Among all genes, there are 1 antibiotic resistance (to glycopeptide), 716 Carbohydrate-active enzymes, 254 effector proteins found in M. acridum genome. During general function annotation, we identified 304, 3,340, 6,276 and 1,782 hits/genes from MEROPS (Supplementary Table 12), PHI-base (Supplementary Table 13), Swiss-Prot (Supplementary Table 14) and TCDB (Supplementary Table 15) databases, respectively. Genome structures Typical fungi telomere usually contains “TTAGGG” repeats which are detected in both sides for all seven chromosomes in M. acridum genome. Even though, there’s not clear AT-rich phenomenon, such observations are normal in fungi [59]. Telomeric repeat analysis (Table 2) revealed scaffold-terminal repeats ranging from 20 to 32 “TTAGGG” units across 14 telomere region (from 7 chromosomes). Moreover, M. acridum exhibits a good syntenic relationship with M. brunneum (Fig. 2c) and in total 67 syntenic blocks (> 8 continuous genes) are detected, covering 9,239 genes (79.5% of total genes). Synteny analysis indicated frequent chromosomal rearrangements between M. acridum and M. brunneum, with fragments from one chromosome of M. acridum distributed across 2–6 chromosomes of M. brunneum (Fig. 2c). These observations indicate extensive chromosomal reorganization, consistent at a descriptive level with previous reports of chromosome transfer between M. robertsii and M. guizhouense [60]. Gene orthology analysis and phylogeny reconstruction In total 12,353 gene families (orthogroups) were identified in ortholog analysis (Fig. 3a, Supplementary Table 16), and we found that the 257 genes shared between M. acridum and M. anisopliae, 465 genes shared between M. acridum and M. brunneum, which is fewer than the 959 genes shared between M. anisopliae and M. brunneum. Nonetheless, we detected 7,560−10,045 single copy genes and 583–2166 multiple copy genes across 9 Metarhizium species (Fig. 3b, Supplementary Table 17). Moreover, we identified ~ 1,000 genes specific to the group including M. anisopliae, M. brunneum, M. guizhouense, M. humberi, M. hybridum, and M. robertsii whose host came from more than 2 insect orders, these patterns reflect their special evolutionary roads, which will be elucidated through detailed gene family analyses. In the examination of gene architectures (Fig. 3c), most domain presented similar numbers between M. acridum and M. brunneum/M. anisopliae, indicating of similar gene composition. Based on the results of MCMCTREE (Fig. 3a), Metarhizium clade formed about 222 MYA, while the divergence time between M. acridum and its sister group was about 164 MYA. Fig. 3Phylogeny and gene features of Metarhizium. a Phylogenetic trees inferred from concatenated single copy genes using Fasttree. MCMCTREE was used for molecular dating. The venn plot exhibited two of M. acridum, M. anisopliae and M. brunneum. b The single copy genes, multiple copy genes and group-specific genes were analysed based on the results from OrthoFinder. The “group-specific” indicates that Metarhizium possesses dtx genes, which include six Metarhizium species: Mhyb, Mani, Mhum, Mrob, Mbru and Mgui. c BLAST based protein domain comparisons. The non-redundant proteins of M. acridum were compared against M. anisopliae and M. brunneum Phylogeny and gene features of Metarhizium. a Phylogenetic trees inferred from concatenated single copy genes using Fasttree. MCMCTREE was used for molecular dating. The venn plot exhibited two of M. acridum, M. anisopliae and M. brunneum. b The single copy genes, multiple copy genes and group-specific genes were analysed based on the results from OrthoFinder. The “group-specific” indicates that Metarhizium possesses dtx genes, which include six Metarhizium species: Mhyb, Mani, Mhum, Mrob, Mbru and Mgui. c BLAST based protein domain comparisons. The non-redundant proteins of M. acridum were compared against M. anisopliae and M. brunneum Phylogeny and expression of Ah-1A As is shown in Fig. 3a, the phylogenetic analysis indicated a clade of Metarhizium genus acquired Ah-1A. After HMMER searching, we found Ah-1A proteins in 6 Metarhizium species (Fig. 4a) possessing an aegerolysin domain. Ah-1A is a cytolytic protein toxin produced by the pathogenic fungus and plays a significant role in fungal host-pathogen interactions [61–63]. Considering these 6 species clustered together and prone to possess a high toxicity to their host, we predicted Ah-1A could help infest the host (Fig. 3a). For M. acridum, we predict a potential role of Ah-1A during the infestation to locust. Hence, qPCR verification was conducted and we found about 10-fold high expression of MaAh-1A two/four days after M. acridum’s infestation to the migratory locust, L. migratoria (Fig. 4b). Fig. 4Phylogeny and infestation expression of Ah-1A. a Ah-1A was detected in six Metarhizium species with aegerolysin doamin labled in green. b Higher expression of MaAh-1A gene was detected two and four days after M. acridum infestation Phylogeny and infestation expression of Ah-1A. a Ah-1A was detected in six Metarhizium species with aegerolysin doamin labled in green. b Higher expression of MaAh-1A gene was detected two and four days after M. acridum infestation Functional validation of MaAh-1A The MaAh-1A protein from M. acridum has a molecular weight of 17.02 kDa (including an N-terminal methionine and a 6 × His tag). Following trypsin digestion, two major fragments of MaAh-1A were observed (Fig. 5a). The MaAh-1B protein, which functions as a molecular chaperone of MaAh-1A, has a molecular weight of 78.18 kDa (also containing a methionine and 6 × His tag). Upon trypsin treatment, MaAh-1B was degraded into small fragments without the appearance of any major protein fragments (Fig. 5a). The trypsin activity of the midgut fluid extract was significantly higher than that of the control group (One-way ANOVA, P < 0.05), whereas no significant differences were observed between the hemolymph and the control group (Supplementary Fig. 3). Furthermore, the digestion products of MaAh-1A produced by midgut fluid extract were identical to those generated by commercial trypsin, while MaAh-1A and MaAh-1B remained intact when incubated with hemolymph (Supplementary Figs. 4 and 5). These results indicate that proteins injected into the hemolymph remain unhydrolyzed due to insufficient proteolytic activity. Injection of different protein combinations revealed that the toxicity of the trypsin-digested mixture of MaAh-1A and MaAh-1B was comparable to that of trypsin-digested MaAh-1A alone (Log-rank test, P = 0.85). Moreover, the trypsin-treated MaAh-1A and MaAh-1B mixture induced significantly higher mortality than the untreated mixture (Log-rank test, P < 0.05). Similarly, trypsin-treated MaAh-1A alone resulted in significantly greater mortality compared to untreated MaAh-1A (Log-rank test, P < 0.05) (Fig. 5b). In contrast, injection of untreated MaAh-1A, MaAh-1B, or their mixture showed no significant differences in survival compared to the eGFP control group (Log-rank test: MaAh-1A vs. eGFP, P = 0.99; MaAh-1B vs. eGFP, P = 0.99; MaAh-1A + MaAh-1B mixture vs. eGFP, P = 0.68) (Fig. 5b). Additionally, there was no significant difference between trypsin-treated and untreated MaAh-1B (Log-rank test, P = 0.75), nor between the trypsin-treated and untreated mixtures of MaAh-1A and MaAh-1B (Log-rank test, P = 0.85). These findings indicate that MaAh-1A helps to kill the locusts. Fig. 5Proteolysis and functional validation of MaAh-1A and MaAh-1B. a Trypsin hydrolyzes MaAh-1A and MaAh-1B protein. M, protein marker; 1, Purified MaAh-1A; 2, Purified MaAh-1B; 3, Mixed of purified MaAh-1A and MaAh-1B; 4, MaAh-1A incubated at 37℃ for 1 h; 5, MaAh-1B incubated at 37℃ for 1 h; 6, Mixed of MaAh-1A and MaAh-1B incubated at 37℃ for 1 h; 7, MaAh-1A incubated with trypsin at 37℃ for 1 h; 8, MaAh-1B incubated with trypsin at 37℃ for 1 h; 9, Mixed of MaAh-1A and MaAh-1B incubated with trypsin at 37℃ for 1 h. b, Survival of locusts (n = 45) after infection of 2.5 mg/mL MaAh-1A, MaAh-1B and mixed of MaAh-1A and MaAh-1B treated with trypsin. Locusts injected with PBS and eGFP protein served as the control. Differences between the control and protein-injected groups are analyzed using the Log-rank test Proteolysis and functional validation of MaAh-1A and MaAh-1B. a Trypsin hydrolyzes MaAh-1A and MaAh-1B protein. M, protein marker; 1, Purified MaAh-1A; 2, Purified MaAh-1B; 3, Mixed of purified MaAh-1A and MaAh-1B; 4, MaAh-1A incubated at 37℃ for 1 h; 5, MaAh-1B incubated at 37℃ for 1 h; 6, Mixed of MaAh-1A and MaAh-1B incubated at 37℃ for 1 h; 7, MaAh-1A incubated with trypsin at 37℃ for 1 h; 8, MaAh-1B incubated with trypsin at 37℃ for 1 h; 9, Mixed of MaAh-1A and MaAh-1B incubated with trypsin at 37℃ for 1 h. b, Survival of locusts (n = 45) after infection of 2.5 mg/mL MaAh-1A, MaAh-1B and mixed of MaAh-1A and MaAh-1B treated with trypsin. Locusts injected with PBS and eGFP protein served as the control. Differences between the control and protein-injected groups are analyzed using the Log-rank test Discussion Our telomere-to-telomere assembly of M. acridum provides the first T2T reference genome for this locust-specific pathogen, as well as for the important entomopathogen genus Metarhizium. This genome achieves full continuity, closing gaps in highly repetitive regions (e.g., LTR retrotransposons, 8.6% of the genome), providing complete telomere-to-telomere coverage for all chromosomes and enabling accurate annotation of gene clusters (e.g., secondary metabolite genes, effector proteins) that are often misassembled in draft genomes. The presence of “TTAGGG” repeats (20–32 copies per telomere) across all chromosomes confirms a fungal-typical telomere structure. Based on our results, we have a significant higher N50 than our assemblies. Moreover, we identified all telomere regions of 7 chromosomes. All these results indicated of a good quality of newly assembled genome in Meterizium. Cordyceps militaris has a T2T genome of 32.67 Mb with an N50 of 4.70 Mb and no gaps [64]. In this study, M. acridum T2T genome is 45.1 Mb, has an N50 of 5.90 Mb, and is also gap‑free. We likewise produced T2T assemblies of comparable or higher quality for M. acridum. The high proportion of unclassified repeats likely results from two factors. On one hand, there are genuine novel repeats—Metarhizium has accumulated lineage-specific transposons and low-complexity sequences not present in public repeat databases. On the other hand, technical limitations also contribute to the high proportion of unclassified repeats, as short-read assemblies are often fragmented and generate many short, incomplete repeat families that cannot be fully assembled or classified. For comparison, the proportion of unclassified repeats was 13.24% in cf. Sphaerocystis sp [65]., 12.31% in Puccinia silphii [66]. It is speculated that unclassified repeats may be a common phenomenon. Although no specific values are provided, it can still be seen that M. acridum, M. anisopliae, and M. robertsii have a certain proportion of unclassified repeat sequences. In this study, 10.23% unclassified repeats may reflect Metarhizium-specific novel repeats, or may be limited by the current assembly and annotation process. Comparative genomic analysis revealed striking patterns of chromosomal evolution in Metarhizium species. The chromosome-scale synteny between M. acridum and its generalist relative M. brunneum (Fig. 1c) uncovered extensive genome restructuring. Synteny analysis revealed extensive chromosomal rearrangements between M. acridum and M. brunneum, with fragments from a single M. acridum chromosome distributed across 2–6 M. brunneum chromosomes (Fig. 3c). Demonstrating remarkable genome plasticity within this fungal genus. The observed structural variations are particularly concentrated in subtelomeric regions [67], known hotspots for fungal genome evolution, where they may facilitate rapid adaptation through modification of virulence-related genes [68]. The subtelomeric region lies immediately adjacent to the telomere, gene-poor but rich in transposable elements [69, 70]. In Candida albicans, the subtelomeric regions subjected to TPE (telomere position effect) can reach up to 25 kb, much larger than in other known fungi. In M. anisopliae, monomorphic subtelomeric DNA in the filamentous fungus within the subtelomeric range of 8.1 kb [71]. Based on these cases, we tested whether the predicted effectors were included in the range of 30 kb at both ends of seven chromosomes in M. acridum, and found that these genes were not distributed in this range (Supplementary File 5). Aegerolysins belong to the Aegerolysin family (PF06355) and can function as membrane lipid-binding proteins, playing roles in recognition, binding, or synergistic membrane disruption [72, 73]. Aegerolysins act alone or in conjunction with membrane-attack-complex/perforin (MACPF)-like proteins to form transmembrane pores that result in cell lysis [74–76]. These proteins have been found encoded in the genomes of entomopathogenic fungi such as Beauveria [77], Cordyceps [78], Ophiocordyceps [79], and Metarhizium [4, 6]. Previous study suggests that aegerolysins in Beauveria bassiana may bind to insect ceramide-phosphoethanolamine [73, 76]. However, the roles of aegerolysins in invertebrate-pathogenic fungi of the order Hypocreales remain largely unknown. In this study, we detected genes encoding aegerolysins in various species of Metarhizium. Additionally, we identified another protein MaAh-1B, containing MACPF domain PF01823. Based on expression and structural prediction analyses, we selected MaAh-1B as a candidate molecular partner for trypsin digestion experiments (Supplementary Fig. 5). We also measured trypsin activity in the gut and hemolymph of locusts. Our results showed significantly higher trypsin activity in the gut compared to the hemolymph, where activity was nearly undetectable (Supplementary Fig. 3). Upon incubation with trypsin and intestinal fluid, MaAh-1B was cleaved into small fragments (Supplementary Fig. 4), and the protein exhibited no insecticidal activity against locusts either before or after hydrolysis (Supplementary Fig. 5). Moreover, co-application with MaAh-1A did not enhance virulence. Whether MaAh-1A interacts with other molecular partners remains to be clarified, assays such as yeast two-hybrid and GST pull-down are necessary to verify interacting proteins [80–82]. Although aegerolysin-MACPF proteins commonly function as binary pore-forming partners, our current experimental results do not show detectable synergistic activity between MaAh-1A and MaAh-1B. Several biological and mechanistic factors can explain this observation without contradicting the expected relationship between these two protein types. Firstly, functional dependence may require specific activation or conditions not replicated in our assays. MACPF partners in other systems often depend on precise biochemical or membrane conditions to interact productively with their aegerolysin counterpart. The absence of synergy in our assays may therefore reflect limitations in the in vitro environment rather than a lack of biological relevance [83]. Secondly, MaAh-1B may play a supportive or structural role [84]. Thirdly, MaAh-1A may exhibit basal activity independent of its MACPF partner. Some aegerolysin-domain proteins can display weak or conditional activity on their own [85, 86]. Finally, Functional divergence within the aegerolysin-MACPF family. Not all predicted binary pairs show identical interaction patterns across species. MaAh-1B may be a MACPF-domain protein that has undergone functional specialization, resulting in a role that is not primarily synergistic in toxicity, or that acts only under specific physiological contexts [87, 88]. Overall interpretation: Based on our results, MaAh-1B does not demonstrate measurable synergy with MaAh-1A under the conditions tested. Therefore, its contribution is more consistent with a supportive or context-dependent structural role rather than a direct enhancement of toxicity. The complete chromosome architecture provides accurate information on the positions, copy numbers, and spatial relationships of genes and their regulatory elements, improving functional predictions for protein‑coding genes, including those in repetitive or highly variable regions. T2T genomes, generated with PacBio HiFi sequencing and Hi-C scaffolding data, yield gap‑free assemblies of entire chromosomes. This allows precise localization of large structural variations, gene clusters, and regulatory elements, enabling the full sequences of virulence factors, host‑adaptation genes, and biocontrol targets to be resolved. Consequently, functional annotation and comparative genomics can reveal how these genes contribute to pathogenicity and adaptation through structural variation, gene amplification, or regulatory networks. Conclusion This study presents the first T2T genome assembly of the locust-specific fungal pathogen M. acridum. Through comprehensive genome structure analysis, synteny comparisons, and gene family evolution studies, we reveal key features during Metarhizium genome evolution. Functional characterization demonstrates that MaAh-1A helps to kill the locusts. These findings provide fundamental insights into fungal pathogen evolution while offering new opportunities for developing precision biocontrol strategies against orthopteran pests. Supplementary Information Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2. Supplementary Material 3. Supplementary Material 3. Supplementary Material 4. Supplementary Material 4. Supplementary Material 5. Supplementary Material 5. Supplementary Material 6. Supplementary Material 6. Supplementary Material 7. Supplementary Material 7. Supplementary Material 8. Supplementary Material 8. Supplementary Material 9. Supplementary Material 9. Supplementary Material 10. Supplementary Material 10. Supplementary Material 11. Supplementary Material 11. Supplementary Material 12. Supplementary Material 12. Supplementary Material 13. Supplementary Material 13. Supplementary Material 14. Supplementary Material 14. Supplementary Material 15. Supplementary Material 15. Supplementary Material 16. Supplementary Material 16. Supplementary Material 17. Supplementary Material 17. Supplementary Material 18. Supplementary Material 18. Supplementary Material 19. Supplementary Material 19. Supplementary Material 20. Supplementary Material 20. Supplementary Material 21. Supplementary Material 21. Supplementary Material 22. Supplementary Material 22. Supplementary Material 23. Supplementary Material 23. Supplementary Material 24. Supplementary Material 24. Supplementary Material 25. Supplementary Material 25. Supplementary Material 26. Supplementary Material 26. Supplementary Material 27. Supplementary Material 27. Supplementary Material 28. Supplementary Material 28. Supplementary Material 29. Supplementary Material 29. Supplementary Material 30. Supplementary Material 30. Supplementary Material 31. Supplementary Material 31.