The T. castaneum specimens used in this study were collected in 2022 from infested sauce-flavor Daqu in Moutai Town, Guizhou Province, China (27°51′ N, 106°22′ E). These beetles have been maintained on sauce-flavor Daqu in the laboratory since collection. We pooled thirty adult beetles (15 males, 15 females) for analysis and stored them at -80 °C in the Entomology Laboratory of Moutai Institute.

Total genomic DNA was extracted from insects using the TIANamp Genomic DNA Kit (Tiangen, Beijing, China) following the manufacturer’s protocol. DNA concentrations were standardized to 0.3 ng/μL for all samples. Whole genome shotgun (WGS) sequencing was conducted through next-generation sequencing (NGS) technology on the Illumina NovaSeq platform. A library with 400 bp inserts was constructed and sequenced using paired-end methodology at Personalbio (Nanjing, China). Raw sequencing data were transferred to a computer workstation for processing. Initial quality assessment was performed using FastQC v.0.11.9, followed by adapter removal with Trim Galore v.0.6.5. Final data validation was conducted through repeat FastQC analysis.

High-quality processed sequencing data were first assembled into Contigs using A5 miseq v20150522 (7), followed by Scaffold construction with SPAdes v3.9.0 (8). Gap filling between contigs was performed through collinearity analysis using Mummer v3.1 (9). Final mitochondrial sequence correction was completed using Pilon v1.18 (10). Assembly accuracy was validated through: (1) BLASTn-confirmed terminal overlaps for circularization, (2) uniform depth distribution averaging 1,580.97× ( Supplementary Figure S1 ), and (3) >95% read concordance rate with Q≥30 scores derived from high-quality data (96.97% HQ reads).

The assembled mitochondrial genome was annotated using the MITOS web server (http://mitos.bioinf.uni-leipzig.de) (11). Sequence alignment was performed with SnapGene software using sequences detailed in Table 1 . The nad3 protein homology model was built with SWISS-MODEL (https://swissmodel.expasy.org/; accessed on 28 July 2025). Bar graphs were drawn using GraphPad Prism 8.0.

We sequenced the mitochondrial genome of the T. castaneum population from Moutai Town, Guizhou, China, reared on Sauce-flavor Daqu. The read sequencing produced 2.9 billion bases (Gb). MiSeq reads corresponding to the mitochondrial genome were separated from nuclear genomic reads (see Methods), with approximately 0.41% of the 19,670,196 total reads representing putative mitochondrial reads. The average sequencing coverage depth was 1580.97×(median = 1580) across the mitochondrial genome ( Supplementary Figure S1 ). These reads produced a mitochondrial genome assembly (gMT) of the China isolate of T. castaneum with a total length of 15,885 bp (GenBank PV563855), sharing 98.86% identity with the California strain (5) and 98.87% with the Jiangsu strain (6). Features of the mitochondrial genome were annotated to identify 22 tRNA genes, 2 rRNA genes, 13 protein coding genes (7 nad subunits, 3 cox, 2 atp, and 1 cob), and a non-coding AT-rich control region (1238 bp) ( Figure 1 ). The overall mitogenome exhibited a nucleotide composition of 39.80% A, 9.81% G, 18.38% C, and 32.01% T, with a markedly biased A+T content of 71.81% ( Table 2 ). This A+T value slightly exceeds those reported for the California (5) population (71.68%) and Jiangsu (6) laboratory strain (71.00%), suggesting potential lineage-specific genomic adaptations.

In PCGs, four (nad4, nad4l, nad5, nad1) of the 13 PCGs were coded on the N-strand, with the other nine genes (cox1, cox2, cox3, atp8, atp6, nad2, nad3, nad6, and cob) were coded on the C-strand. Among the 13 PCGs, the longest was the nad5 gene (1,713 bp) and the shortest was the atp8 gene (159 bp). The start codons of cox1 (CTG) and nad1 (TTG) in this study exhibit striking contrast with those of the Jiangsu and California populations (5, 6). Specifically, the Jiangsu population demonstrates AAT for cox1 and ATT for nad1, while the California population (5) utilizes AAA and ATA as start codons for cox1 and nad1, respectively ( Table 3 ). Notably, the remaining 11 PCGs uniformly employ ATN start codons across all analyzed populations. Among the 13 PCGs, ten terminated with standard stop codons (TAA/TAG), while nad5 used ATT, and the remaining two ended with incomplete stops (single T) ( Table 3 ). It is generally accepted that incomplete codon structures signal a halt of protein translation in insects and other invertebrates (12). With the exception of atp8, cox2, and cox3, all single-nucleotide variants (SNVs) resulted in nonsynonymous mutations ( Figure 2 ). Notably, higher frequencies of nonsynonymous mutations were observed in nad5, nad4, nad4l, and nad1 ( Figure 2 ).

In the present study, the nad3 gene comprises 354 nucleotides encoding 117 amino acids, contrasting with previously reported 360-nucleotide sequences encoding 119 amino acids in other populations (5, 6). Sequence alignment revealed a thymine insertion at position 5,594 within the nad3 locus ( Figure 3 ). This single-nucleotide insertion induces a 7-bp frameshift (original initiation codon ATA at position 5,529 → revised ATT at position 5,536), resulting in complete divergence of the first 20 amino acid residues compared to published sequences. Crucially, this frameshift modifies the open reading frame (ORF), altering the protein’s tertiary structure at the N-terminus while preserving downstream structural domains ( Figure 3C ), suggesting potential compensatory mechanisms in mitochondrial translation.

All 22 tRNA genes were identified, with a combined length of 1,435 bp (range: 59–71 bp; Table 3 ). Notably, only trnS1 (AGN) deviated structurally, lacking the dihydrouridine (DHU) arm characteristic of canonical cloverleaf folds ( Supplementary Figure S2 ). The remaining 21 tRNAs exhibited standard cloverleaf secondary structures ( Supplementary Figure S2 ). Gene boundaries were determined primarily by homology-based alignment using MITOS, with secondary-structure prediction (tRNAscan-SE) serving as validation. Eight tRNA genes were located on the minor strand, and fourteen on the major strand. The rrnL and rrnS genes measured 1,261 bp and 757 bp, respectively ( Table 3 ).

The control region of T. castaneum, a highly variable segment critical for mitogenome replication and transcription initiation, spans 1,238 bp between trnI and rrnS (A+T content: 82.80%; AT-skew: -0.078; GC-skew: -0.001), with population-specific variations: 1,237 bp/82.30% A+T in the Jiangsu population and 1,239 bp/82.50% A+T in the California population (4). Sequence alignment revealed 26 single nucleotide variants (SNVs) between the reported mitochondrial genome control region and the Jiangsu population (6), 20 SNVs compared to the California population (5), including 7 shared SNVs ( Supplementary Figure S3 ).