The telomere-to-telomere chromosome-scale genome assembly of Acremonium chrysogenum

Abstract

Acremonium chrysogenum is a notable filamentous fungus recognized for its essential contribution to the pharmaceutical sector through the biosynthesis of cephalosporin C (CPC). CPC functions as a key intermediate in the biosynthesis of β-lactam antibiotics, which are employed to combat bacterial infections. This study successfully generated a telomere-to-telomere (T2T) chromosome-scale genome sequence for A. chrysogenum, combining BGI short reads, PacBio HiFi long reads, and Hi-C technology. This genome sequence contained eight complete chromosomes (29.00 Mb) and a circular mitochondrial genome (27.27 kb), featuring an N50 length of 3.87 Mb. Repetitive elements accounted for 9.65% of genomic content, and a total of 7,745 genes involved in protein coding were annotated. This well-assembled reference genome of A. chrysogenum serves as an important foundation for elucidating the biosynthetic pathway of cephalosporin C and for molecular breeding. Furthermore, it offers valuable insights into chromosome organization, genome evolution, and regulatory mechanisms, facilitating future advancements in antibiotic research and fungal biotechnology.

Background & Summary

Acremonium chrysogenum, recently reclassified as Hapsidospora chrysogena, is a critical filamentous fungus in industrial production, best known as the natural producer of cephalosporin C, a key precursor in the synthesis of cephalosporin antibiotics. Antibiotics play an indispensable role in clinical therapy, with cephalosporins accounting for 20% of the antibiotic market¹. Most antibiotics derive from microbial secondary metabolites and their derivatives². In 1954, Professor Giuseppe Brotzu discovered a fungus capable of producing cephalosporin C³. Initially named Cephalosporium acremonium, it was later reclassified as Acremonium chrysogenum and, most recently in June 2023, renamed Hapsidospora chrysogena⁴. Since the name A. chrysogenum has been widely adopted in scientific literature, we continue to use this designation in the present study, in accordance with other researchers, to avoid any potential confusion. The discovery revealed the fungus’s broad antibacterial properties against diverse bacterial types and marked a critical milestone in antibiotic research enabling large-scale production. Over time, Acremonium chrysogenum has emerged as a major producer of cephalosporin antibiotics. As the primary raw material for synthesizing 7-aminocephalosporanic acid (7-ACA), a crucial intermediate, its output and cost of cephalosporin C significantly affect the supply and pricing of cephalosporin antibiotics. Given their broad antimicrobial efficacy and relative safety for both humans and animals, cephalosporin antibiotics are widely used to treat bacterial infections, establishing them as a cornerstone of clinical anti-infection therapy⁵.

In 2014, Terfehr and Dominik’s team constructed the first draft genome assembly of Acremonium chrysogenum, generating 541 scaffolds (N50: 16.69 kb) based on second-generation sequencing data⁶. However, this assembly contained 1,189 sequence gaps, resulting in a highly fragmented genome structure. Although subsequent studies have systematically elucidated the cephalosporin C (CPC) biosynthetic pathway⁷, the incomplete genome assembly has significantly hindered fundamental research into the species’ genetic regulatory mechanisms and secondary metabolite networks. To advance genomic studies of Hapsidospora species, the establishment of a high-resolution reference genome has become an urgent necessity. Compared to previous fragmented assemblies, this more complete and contiguous genome assembly provides a molecular foundation for investigating cephalosporin C biosynthesis while also being regarded as an important genetic asset for strain engineering and the development of efficient breeding strategies.

In the present work, we generated the telomere-to-telomere (T2T) chromosome-scale genome (Fig. 1) for Acremonium chrysogenum, achieved through the integration of cutting-edge sequencing methods such as BGI-generated short reads, PacBio HiFi reads, and Hi-C scaffolding. The final genome measured 29.00 Mb, featuring an N50 of 3.87 Mb, with the genome sequences organized into eight chromosomes. Annotation identified 7,745 protein-coding genes. Additionally, a circular mitochondrial genome of 27.73 kb in length was successfully assembled. The quality of our assembled genome was significantly improved compared to previous versions (Table 1)^6,8. The presence of a high-quality, complete genome assembly is crucial for fundamental biological research. The genomic datas generated in this research not only establish a basis for subsequent studies into the molecular mechanisms of cephalosporin biosynthesis in Acremonium chrysogenum, but also pave the way for innovative applications in strain improvement, metabolic pathway optimization, and antibiotic development.

Fig. 1.

Chromosomal distribution of different elements in A. chrysogenum. Displayed from the outermost to the innermost ring: (a) Chromosomes karyotype (The black area in the chromosome is the location of the centromere). (b) Copia denisity. (c) TE denisity. (d) GC denisity. (e) Gene density.

Table 1.

Summary of the A. chrysogenum genome assembly comparisons.

Feature	A. chrysogenum CGMCC3.3795	A. chrysogenum ATCC11550	A. chrysogenum RNCM 408D
Genome size	29.0 Mb	28.6 Mb	28.9 Mb
Number of Contigs	10	2,799	2,258
N50 of Contigs	3.8 Mb	149.5 kb	33.9 kb
Number of Scaffolds	8	541	2,258
N50 of Scaffolds	3.9 Mb	166.9 kb	33.9 kb
GC content of the genome (%)	55	55	55
Number of gaps	2	1,189	not provided
Assembly level	Chromosome	Scaffold	Contig

Methods

Sample collection and sequencing

Collection and extraction of DNA samples

The A. chrysogenum strain CGMCC3.3795 was routinely cultivated and maintained on slant culture medium. For genomic DNA extraction, hyphae grown for 7 days at 30 °C and 50% humidity were harvested from these slant cultures. These harvested hyphae were then inoculated into Gelrite Yeast Extract and Malt Extract (GYM) liquid medium incubated with shaking (180 rpm) for 42 hours at 30 °C before collection. Genomic DNA used for sequencing was subsequently isolated from collected hyphae by applying an adapted CTAB protocol⁹.

Illumina, PacBio and Hi-C sequencing

A combined sequencing strategy was used to build a high-quality genome, drawing on BGI short reads, PacBio long reads, and Hi-C methods. Short-read data were produced by constructing a 150 bp insert-size library performed using DNBSEQ-T7 (BGI, Shenzhen, China), yielding 2.33 million reads and 3.50 Gb of raw sequences (Table 2). Using the Template Prep Kit 2.0, a SMRTbell library was generated to acquire PacBio long-read data, which was sequenced via the Sequel II platform (Pacific Biosciences, Menlo Park, USA). This process yielded 1.78 Gb of HiFi reads with an N50 length of 13.88 kb (Table 2). The Hi-C protocol followed standard procedures, including formaldehyde crosslinking, DNA cleavage and biotinylation, fragment ligation, and purification. After random fragmentation of the DNA into 300–500 bp fragments, libraries were prepared and sequenced using the DNBSEQ-T7 platform, resulting in 52.09 Gb of sequencing output (Table 2)¹⁰.

Table 2.

Summary of sequencing data of A. chrysogenum for telomere-to-telomere assembly and genome annotation.

Sequencing	Raw bases (Gb)	Raw reads	N50 length (bp)	Application
HiFi	1.78	143,530	13,877	Assembly
Hi-C	52.09	173,638,290	2 × 150	Chromosome construction
BGI DNBSEQ-T7	3.50	23,363,760	2 × 150	Genome evaluation
RNA-seq	6.92	46,140,436	2 × 150	Genome annotation

RNA sequencing and analysis

Hyphal RNA was extracted using TRIzol reagent (Invitrogen, CA, USA). Poly-T oligo-bound magnetic beads were used to purify mRNA from total RNA. Following mRNA extraction, RNA-seq libraries were constructed using the VAHTS Universal V6 Kit for MGI (Vazyme, Nanjing, China), with unique index codes assigned as per the manufacturer’s protocol. These libraries were then sequenced on the DNBSEQ-T7 platform, producing 6.92 Gb of raw RNA-seq output (Table 2).

Genome sequence assembly

Genome survey

We used SOAPnuke software¹¹ (v2.1.0) to clean the raw short reads and obtain clean reads. The clean reads served to ascertain the entire genomic scale. A k-mer analysis (17-mer) was performed on the short-read data to generate a frequency distribution. Based on this analysis, the genome size was inferred to be roughly 31.98 Mb. Additionally, the content of repetitive elements was approximately 16.11%, while genomic heterozygosity measured about 0.21%, both determined using GCE software¹² (v1.0.2) (Table 3).

Table 3.

Summary of the A. chrysogenum genome size estimation by 17-mer analysis.

K-mer	Number	Depth	Genome Size (Mb)	Heterozygous Ratio (%)	Duplication Ratio (%)
17	3,108,257,032	95	31.98	0.21	16.11

Genome assembly

Sequencing of SMRTbell libraries took place on the PacBio Sequel II platform, and HiFi reads resulted from ccs processing(v6.3.0; https://github.com/pacificbiosciences/unanimity), applying the option ‘-minPasses 3′. Highly precise HiFi reads, approximately 15 kilobases long and with over 99% accuracy, served as input for genome assembly using Flye¹³ (v2.9) with default settings. To process the Hi-C sequencing data, Trimmomatic¹⁴ (v0.39) was used for adapter trimming and quality filtering. Subsequently, the filtered reads were mapped to the 10 contigs using Juicer¹⁵ (v1.6; https://github.com/aidenlab/juicer) to compute interaction frequencies. Next, 3D-DNA¹⁶ (v180922) was employed to perform misjoin adjustments in two iterations (-r1) under default configuration. After contig orientation, Juicer was used to construct interaction matrices, and manual refinement was carried out using Juicebox Assembly Tools¹⁶ (v1.11.08). These contigs were anchored onto eight pseudochromosomes (Fig. 2), consistent with previous karyotype analysis of A. chrysogenum, which confirmed the presence of eight chromosomes in this species^17,18. In total, the assembled genome of A. chrysogenum CGMCC3.3795 is approximately 29.00 Mb achieving an N50 scaffold of 3.87 Mb (Table 1). These sequences were successfully arranged into eight pseudochromosomes.

Fig. 2.

A visualization of the genome-wide Hi-C interaction matrix at a resolution of 250 kb.

The mitochondrial genome was assembled from short-read data using GetOrganelle software¹⁹, with the mitochondrial genome sequence of A. chrysogenum ATCC11550 (accession number KF757229.1) serving as the reference. The resulting mitochondrial genome is 27.73 kb in size.

Genome annotation

Detection of repetitive sequences within the A. chrysogenum genome involved both de novo and homology-based approaches. A de novo repeat library was constructed using RepeatModeler²⁰ (v2.0.1) (http://www.repeatmasker.org/RepeatModeler/) for initial identification of repetitive sequences. Homology-based tools RepeatMasker²¹ and RepeatProteinMask (v4.1.2), along with Repbase library^22,23 (https://www.girinst.org/repbase/), were subsequently applied to annotate characterized transposable elements (TEs). LTR_FINDER²⁴ (v1.0.7) was utilized to detect long terminal repeat (LTR) retrotransposons. The Tandem Repeat Finder (TRF) package²⁵ was applied to locate tandemly repeated regions, whereas RepeatMasker was utilized to uncover non-interspersed repetitive elements, such as simple sequences, satellite DNA, and regions of low complexity. Finally, repetitive element datasets obtained from the two methods were integrated into a unified library, and RepeatMasker was applied to annotate the overall repeat content. In total, 2.83 Mb of repetitive elements, comprising 9.76% of the A. chrysogenum genome, were identified. Of the identified repeats, LTR elements accounted for the largest portion, comprising 7.61% of the genome (Table 4).

Table 4.

Summary of repeat elements.

Repetitive sequence	Number/% in genome
Tandem repeats	0.71
DNA	0.63
LINE	0.22
SINE	0.00122
LTR	7.61
Unknown	0.88
Total	9.76%
Protein-coding genes
the total number of genes	7,745
the average gene length (bp)	2,455.11
the average cds_length (bp)	1,570.14
the average exon_number of per gene	3.02
the average exon_length (bp)	748.7
the average intron_length (bp)	96.16
Function annotation
InterPro	5,827 (75.24%)
GO	5,856 (75.61%)
KEGG_ALL	7,312 (94.41%)
KEGG_KO	3,383 (43.68%)
Swissprot	5,211 (67.28%)
TrEMBL	7,713 (99.59%)
NR	7,714 (99.60%)
Annotated	7,714 (99.60%)
Unannotated	31 (0.40%)

Inferring architecture involved integrating ab initio, homology-based, along with transcriptome-based strategies. Before annotating, RepeatMasker masked A. chrysogenum’s repetitive regions, performing both soft and hard masking. For the ab initio-based approach, we adopted Augustus²⁶ (v3.4.0) and GlimmerHMM²⁷ (v3.0.4). For the homology-guided approach, protein datasets from A. chrysogenum ATCC11550, Penicillium chrysogenum, and Aspergillus luchuensis were obtained via the National Center for Biotechnology Information (NCBI) database and aligned with the A. chrysogenum CGMCC3.3795 genome using TBLASTN²⁸ (NCBI BLAST v2.11.0+). Subsequently, gene models were predicted through homology using Exonerate²⁹ (v2.4.0). For the transcriptome-based method, high-quality RNA-Seq data were first assembled into transcript sequences with Trinity³⁰ (v2.8.5), followed by gene structure annotation using PASA³¹ (v2.4.1). MAKER³² (v3.01.03) combined outcomes from all methods, yielding consensus models. This ultimately provided consistent, separate structures representing the gene architecture of A. chrysogenum.

Gene functions were predicted by identifying top hits through comparisons against several functional databases, including NCBI Non-Redundant (NR)³³, Gene Ontology (GO)^34,35, Translation of European Molecular Biology Laboratory (TrEMBL)³⁶, InterPro³⁷, Swiss-Prot³⁶, and the Kyoto Encyclopedia of Genes and Genomes (KEGG)³⁸, utilizing BLASTP²⁸ (NCBI BLAST v2.11.0+) setting the E-value to 1E-5. Our analysis identified 7,745 protein-coding genes, with 7,714 (99.60%) of these successfully assigned functional annotations in a minimum of one database (Table 4).

Identification of non-coding RNA genes

TRNA genes were detected using tRNAscan-SE³⁹ (v2.0.9) under default settings. For rRNA identification, RNAmmer⁴⁰ (v1.2) was employed. Infernal⁴¹ (v1.1.4) was employed to detect miRNAs and snRNAs by aligning sequences to the Rfam database⁴² (v14.1) using standard parameters. In total, the A. chrysogenum genome contains 111 tRNAs, 28 rRNAs, 19 snRNAs, and no miRNAs (Table 5).

Table 5.

Summary statistics of noncoding RNA.

Type		Copy	Total length(bp)	% of genome
miRNA		0	0	0.00
tRNA		111	9,969	0.03
rRNA	rRNA	28	18,559	0.06
	18S	3	5,385	0.02
	28S	3	10,617	0.04
	5S	22	2,557	0.01
snRNA	snRNA	19	2,381	0.01
	CD-box	12	1,288	0.00
	HACA-box	1	179	0.00
	splicing	6	914	0.00
	scaRNA	0	0	0.00

Telomere and centromere identification

Telomere identification was carried out by detecting characteristic telomeric repeat sequences (CCCTAA at the 5′ end and TTAGGG at the 3′ end). Of the 16 expected telomeres (two per chromosome across eight chromosomes), 15 were successfully identified. HiFi telomeric reads were manually inspected, and the missing telomeric sequences on the final chromosome were patched accordingly (Table 6, Fig. 3). Centromere identification was performed using Tandem Repeats Finder²⁵ to detect clusters of candidate centromeric tandem repeats. These repeats were found to be continuous and occupied the majority of each chromosome. To determine precise centromere positions, the candidate centromeric tandem repeats were integrated with Hi-C interaction heatmaps (Table 7, Fig. 3).

Table 6.

Summary of telomere information of the A. chrysogenum genome.

Chr	upstream_start	upstream_end	downstream_start	downstream_end
Superscaffold1	1	135	6,741,395	6,741,520
Superscaffold2	1	238	4,610,212	4,610,372
Superscaffold3	1	127	3,877,092	3,877,216
Superscaffold4	1	120	3,821,661	3,821,780
Superscaffold5	1	134	3,309,648	3,309,767
Superscaffold6	1	133	2,539,188	2,539,301
Superscaffold7	1	122	2,332,562	2,332,690
Superscaffold8	1	120	1,768,644	1,768,666

Fig. 3.

Distribution map of centromeres and telomeres on the assembled A. chrysogenum genome. The telomeric sequences are represented by the green termini, while the central region of crossover corresponds to the centromeric sequence.

Table 7.

Summary of predicted centromere information of the A. chrysogenum genome.

chr	start	end
Superscaffold1	4,304,000	4,389,000
Superscaffold2	291,000	660,000
Superscaffold3	2,762,000	2,876,000
Superscaffold4	1,393,000	1,423,000
Superscaffold5	2,469,000	2,608,000
Superscaffold6	1,300,000	1,335,000
Superscaffold7	1,533,000	1,587,000
Superscaffold8	753,000	781,000

Data Records

The genome assembly data is available on GenBank under accession number GCA_965197235⁴³.

Raw sequencing data can be retrieved from the Genome Sequence Archive (GSA) at NGDC under accession number CRA023113⁴⁴.

The genome annotation files, gene CDS, and protein data have been submitted to Figshare⁴⁵.

Technical Validation

DNA sample quality

We evaluated the extracted DNA’s quality and quantity through a combination of methods: a NanoDrop 2000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA), a Qubit 3.0 Fluorometer with a dsDNA HS Assay Kit (Life Technologies, Carlsbad, CA, USA), and electrophoresis using 0.8% agarose gel.

Sequencing data assessment

Short-read data were processed using SOAPnuke¹¹ (v2.1.0), revealing a GC content of 52.9% and Q20/Q30 accuracy rates of 98.7% and 95.1%. The Hi-C dataset had a GC content of 55.15%, with corresponding Q20 and Q30 of 97.31% and 91.21% (Table 8).

Table 8.

Statistics of genomic of the BGI short-read sequencing data.

Application	ReadNum	BaseCount(Gb)	ReadLength(bp)	Q20(%)	Q30(%)	GC Content(%)
survey	23,363,760	3.50	150;150	98.70	95.10	52.90
Hi-C	173,638,290	52.09	150;150	97.31	91.21	55.15

Assessment of genome assembly quality

Multiple complementary methods were applied for assessing the A. chrysogenum assembly. As a primary step, BUSCO⁴⁶ (v5.2.2) was run with the fungi_odb10 database (758 genes) to gauge genome completeness, resulting in a robust 99.3% completeness score. Next, BGI short reads were mapped onto the assembled genome via BWA⁴⁷ (v0.7.17), resulting in a 99.99% coverage rate. Third, mapping of TGS long reads was performed with Minimap2⁴⁸ (v2.24), resulting in 100% genome coverage. Fourth, LTR_retriever⁴⁹ estimated assembly quality using the LTR Assembly Index (LAI), producing a 30.06 score, which meets the threshold for a golden reference genome. Fifth, Mequery⁵⁰ (v1.3) evaluated overall base-level quality and completeness, producing metrics of 53.51 QV and 98% completeness. The results collectively confirm the A. chrysogenum assembly’s high accuracy, completeness, and stability.

Author contributions

Z.H. supervised the research. C.H., Y.Z. and H.L. contributed to research design, C.H., Y.Z., H.L. and M.C. analyzed the data. C.H., Y.Z., J.L. and Z.H. drafted and revised the manuscript. All authors have reviewed and approved the final version of the manuscript.

Code availability

All experimental procedures were performed using standard, off-the-shelf tools, eliminating the need for custom scripting. Furthermore, every bioinformatics tool and pipeline employed is publicly accessible. Unless specific parameters are otherwise noted, the default settings recommended by their developers were consistently applied.

Competing interests

The authors declare no competing interests.