Gapless and phased genome assembly reveals post-allopolyploidization subgenomic diversification of tobacco centromeres Authors: Weikai Chen, Shaoying Chen, Jingxuan Wang, Dian Meng, Jun Li, Li Guo Published: 2026 ABSTRACT Common tobacco (Nicotiana tabacum) is an economically important crop worldwide whose allotetraploid genome remains incompletely assembled. We present a 4.2-Gb telomere-to-telomere, gap-free, and phased genome assembly. This assembly reveals distinct evolutionary trajectories between the two subgenomes, particularly within heterochromatic centromeric regions. Consistent with the rapid sequence divergence of the T subgenome, we observe pronounced differences in DNA methylation landscapes between subgenomes. To understand centromere evolution following polyploidization, we characterize centromere architecture in tetraploid Nicotiana tabacum and its diploid progenitors, uncovering asymmetric subgenomic diversification and frequent centromere repositioning. These findings provide new insights into dynamic evolution of centromeres in polyploid genomes.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13059-026-04163-2. Background Common tobacco (N. tabacum, SSTT, 2n = 4x = 48) is an economically important crop cultivated worldwide and a model organism for plant research. The tetraploid tobacco originated from an interspecific hybridization between its diploid ancestors, N. sylvestris (SS, 2n = 2x = 24) and N. tomentosiformis (TT, 2n = 2x = 24), approximately 0.2 million years ago (Mya) [1]. The draft genome of tobacco was first released in 2014 [2] and subsequently improved using long-read sequencing technology [3–5]. However, hundreds of gaps remain unresolved in these assemblies, particularly in heterochromatic regions harboring long arrays of satellite DNA. Generating a telomere-to-telomere (T2T) gap-free genome assembly is essential not only for accelerating tobacco genetics and breeding, but also for deciphering the dynamics and evolution of complex genomic regions such as centromeres. Centromeres are critical genomic regions for kinetochore assembly, ensuring the accurate chromosome segregation during cell division. Despite its conserved function, the underlying genomic sequences demonstrate remarkable diversity [6]. For example, Arabidopsis thaliana centromeres are primarily composed of megabase-scale CEN180 satellite arrays [7], whereas Oryza sativa centromeres display a mosaic of CentO satellites and centromere-specific retrotransposons of rice (CRRs) [8]. The Solanaceae centromeres are mainly featured by Ty3/Gypsy retrotransposons [9–11], similar with cotton [12] and wheat [13], while two distinct types of satellite-based and Gypsy-based centromeres are present in N. benthamiana [13], illustrating that the genomic architecture of centromeres is highly plastic and lineage-specific. The structure of N. tabacum centromeres has been a subject of controversy, with conflicting reports suggesting either tandem repeats [5, 14] or retrotransposons [15, 16] as the dominant feature. Nevertheless, the centromere structures of N. tabacum exhibit considerable ‌variation with its wild relative N. benthamiana. To characterize the overall structure of N. tabacum centromeres, we assembled its T2T gap-free phased genome, and characterized its centromere structure by conducting chromatin immunoprecipitation sequencing (ChIP-seq) using centromere-specific histone H3 (CENH3) antibody. Our analyses revealed the mysterious genomic landscape of N. tabacum centromeres, and unraveled the evolutionary history of centromere dynamics during diploid species divergence and allopolyploidization process. Results and discussion In this study, we achieved the first T2T gap-free genome assembly of the tobacco cultivar ‘Yabuli’ (YBL), through a combination of 116 × PacBio HiFi, 40 × Oxford Nanopore Technology (ONT) ultra-long sequencing and 109 × high-throughput chromatin conformation capture and sequencing (Hi-C) data (Additional file 1: Table S1). Hifiasm pipeline [17] integrating both HiFi and ONT reads generated an initial N. tabacum genome assembly which contained 13 contigs representing single T2T gap-free chromosomes. The remaining contigs were successfully scaffolded into another 11 chromosomes using Hi-C reads, followed by a comprehensive closure of existing 18 assembly gaps using long-reads. After telomere patching and assembly polishing, we finally obtained a gap-free genome assembly of the N. tabacum (NtT2T) that sized 4,199.03 Mb with contig N50 of 180.27 Mb (Fig. 1A; Additional file 2: Fig. S1).Fig. 1Overall features and centromere landscape of T2T tobacco genome. A Genomic synteny between phased tobacco NtT2T and NtaSR1 assemblies. The collinear regions were linked by gray lines, while genomic rearrangements were shown by ribbons with different colors. The telomeres and centromeres in the NtT2T assembly and the unclosed gaps in NtaSR1 assembly are labeled. The reference genome of N. tabacum cv. Petite Havana SR1 (NtaSR1) was used. B Composition of tobacco centromere sequences on each chromosome. ‌NUMTs, nuclear mitochondrial DNA segments; Non-TE, non-repetitive sequences. C Genomic landscape of two representative centromeres in each subgenome. The tracks show CENH3 ChIP-seq signals, Gypsy density, satellite density and distribution of Nt2-7, CRM and Tekay elements Overall features and centromere landscape of T2T tobacco genome. A Genomic synteny between phased tobacco NtT2T and NtaSR1 assemblies. The collinear regions were linked by gray lines, while genomic rearrangements were shown by ribbons with different colors. The telomeres and centromeres in the NtT2T assembly and the unclosed gaps in NtaSR1 assembly are labeled. The reference genome of N. tabacum cv. Petite Havana SR1 (NtaSR1) was used. B Composition of tobacco centromere sequences on each chromosome. ‌NUMTs, nuclear mitochondrial DNA segments; Non-TE, non-repetitive sequences. C Genomic landscape of two representative centromeres in each subgenome. The tracks show CENH3 ChIP-seq signals, Gypsy density, satellite density and distribution of Nt2-7, CRM and Tekay elements Then multiple strategies were applied to evaluate the accuracy and completeness of the NtT2T. First, Hi-C chromatin matrix (Additional file 2: Fig. S2), and the whole-genome alignment against Petite Havana SR1 (NtaSR1) genome (Fig. 1A) both confirmed the large-scale assembly accuracy of NtT2T. Second, the mapped HiFi reads to NtT2T showed a nearly uniform coverage across the whole genome (Additional file 2: Fig. S3). Third, BUSCO evaluation captured 99.6% of the complete gene set, the quality value (QV) reached 64.07, and LTR assembly index (LAI) scored 14.55, together indicating substantial improvement of NtT2T over previous versions (Additional file 1: Table S2). Noteworthy, 40 out of 48 telomeres were detected in NtT2T (Additional file 1: Table S3), missing eight telomeres mainly due to the incomplete assembly of nucleolus organizer regions (Additional file 2: Fig. S4) and subterminal satellite arrays (Additional file 1: Table S3). Genome annotation predicted 65,500 gene models in the NtT2T with completeness and consistency comparable to other recently published annotations (Additional file 2: Fig. S5). Overall, these evaluations demonstrated NtT2T as a reference genome for N. tabacum with the highest quality of all versions. We then phased the NtT2T into two subgenomes (subS and subT) by distinguishing subgenome differential k-mers (Additional file 2: Fig. S6), and identified the exchange regions between two subgenomes (Additional file 1: Table S4). Comparing with the recent NtaSR1 assembly, a total of 211.2 Mb sequences were newly assembled (Fig. 1A), the majority of which was located in subtelomeric heterochromatic caps (56.8%) and highly diverged regions (30.9%). The subterminal repeats were featured by specific 183-bp satellites (HRS60), which mostly resided in the subS (Additional file 1: Table S3). Collinearity analysis of NtT2T and the near-T2T assemblies of two wild diploid ancestors revealed asymmetric evolution of subT and subS (Additional file 2: Fig. S7). Interestingly, the subT showed less sequence similarity with N. tomentosiformis than that between subS and N. sylvestris, suggestive of rapid divergence of paternal T-genome. Comparative genomics analysis also revealed multiple events associated with allopolyploid divergence, such as intergenomic translocation (Additional file 2: Fig. S8), genome downsizing and satellite repeat divergence (Additional file 2: Figs. S9-S11). Satellite DNA sequences are excellent markers of rapid and subtle evolutionary changes. Particularly, NTRS, the molecular maker of tobacco T04 chromosome [18], lost high copy of satellite repeats during the transition from the wild diploid ancestors to tetraploid tobacco; while RETS, the S-genome specific satellites [18], showed the opposite trend (Additional file 1: Table S5; Additional file 2: Fig. S11). Consistent with previous finding [5], the subT underwent more pronounced contraction than the counterpart of subS, particularly in the heterochromatic regions. The divergent evolutionary trajectories between two subgenomes likely stem from the genomic shock following allopolyploidization. While the paired Wilcoxon test revealed a significant global expression asymmetry between the 22,018 homeologous gene pairs, the magnitude of this bias remains notably subtle (Additional file 2: Fig. S12). This tempered transcriptional divergence aligns with the modest but pronounced epigenetic asymmetry in DNA methylation, suggesting that the subS genome is under stricter heterochromatic constraints. It is crucial to note that the epigenetic variations are likely a footprint of uneven subgenomic evolution, potentially linked to transposon bursts or underlying sequence variations between the two subgenomes. Centromere is a specialized chromosomal region that responsible for sister chromatid cohesion during cell division, epigenetically defined by the presence of centromere-specific histone H3 (CENH3). We determined the tobacco centromeres by ChIP-seq using tobacco-specific CENH3 antibody (Additional file 2: Fig. S13), which ranged from 0.88 to 2.7 Mb in size, with an average length of 1.61 Mb (Additional file 1: Table S6). Tobacco centromeres were dominated by 71.5% of Ty3/Gypsy retrotransposons (Additional file 1: Table S7), except for centromeres of Chr08 (CenS08) and Chr22 (CenS22) where specific satellite arrays covered about one-third of the regions (Fig. 1B; Additional file 2: Fig. S14), distinctly different from its close relative N. benthamiana [19]. Although Tekay lineages were predominant across the tobacco genome with comparable levels between the arms and centromeres, the CRM family was enriched markedly in the peri-/centromeres (Fig. 1C; Additional file 2: Fig. S15). The comparative profiling revealed a functional divergence between retrotransposon families, where CENH3 signals were enriched on Tekay, but depleted within CRM elements (Additional file 2: Fig. S16). This suggests that Tekay has been functionally recruited as the primary scaffold for kinetochore assembly in N. tabacum. Estimation of the LTR insertion time revealed that most retrotransposons were inherited from diploid parents, far before the interspecific hybridization at ~ 0.2 Mya (Additional file 2: Fig. S15). Phylogenetic analysis classified Tekay elements into six branches mainly based on their subgenome assignment, regardless of their chromosomal positions (Additional file 2: Fig. S17). However, the Tekay elements inside and outside the centromeres showed distinct DNA methylations, especially on the flanking LTR motifs, indicative of epigenetic modification variations (Additional file 2: Fig. S18). Notably, five groups of centromeric Tekay elements originated from both parental diploids, while the last group was specific to subT (Fig. 2A), including the previously identified T-genome-specific NtoCR [16]. In addition, the Nt2-7, a LTR fragment of Tekay, has been shown to localize in the centromeric regions of tobacco [15, 16], in agreement with our observation especially in the subT (Fig. 1C). Although numerous centromeric transposons contained Nt2-7 sequences (Fig. 2A and B), majority of them have lost their internal domains (Additional file 2: Fig. S19), suggesting that Nt2-7 have evolved autonomous amplification mechanisms or have been differentially retained during subgenome evolution. The phylogenetic tree placed the tobacco Nt2-7 with tomato TGR4 [10] together, indicating the centromeric retrotransposons were relatively conserved in some Solanaceae species. Furthermore, the Ogre element preferentially inserted in the subS (Additional file 2: Fig. S20), which might have aggravated the centromere diversity between two subgenomes. Interestingly, we observed that the centromeric 48-bp satellite (CEN48) within CenS22 probably originated from the repetitive array of specific Ogre retrotransposons (Additional file 2: Fig. S21).Fig. 2Post-allotetraploidization subgenomic diversification and repositioning of tobacco centromeres. A Maximum likelihood phylogenetic tree of 356 full-length centromeric Tekay elements, color coded according to subgenome assignment. The bootstrap values of key nodes are labelled. B Sequence alignments of five Tekay retrotransposons highlighted in (A) panel. The yellow rectangle represents terminal LTRs, while the blue ones represent internal protein-coding domains, including group-specific antigen (GAG), protease (PROT), reverse transcriptase (RT), ribonuclease H (RH), integrase (INT), and a chromodomain (CHD). The grey boxes show the sequence identity between two elements. C The relative distance of tetraploid tobacco centromeres to CRM and Nt2-7 enriched regions, and diploid ancestry centromeres. D The recent centromere repositioning occurred from diploid ancestors to tetraploid tobacco. The blue and red lines represent the location of Nt2-7 and CRM. E Summary of evolutionary trajectories of tobacco centromeres. The common ancestor of tobacco and tomato shared an ancestral centromere dominated by Tekay. Following species divergence, ancient centromere shift occurred between the diploid N. sylvestris and N. tomentosiformis lineages, accompanied by the invasion of CRM and subgenome-specific Nt2-7 elements. After the allopolyploidization event, the tetraploid genome underwent asymmetric subgenomic evolution and frequent centromere repositioning. The sequence identity is indicated by gray shaded blocks: deep gray > medium gray > light gray Post-allotetraploidization subgenomic diversification and repositioning of tobacco centromeres. A Maximum likelihood phylogenetic tree of 356 full-length centromeric Tekay elements, color coded according to subgenome assignment. The bootstrap values of key nodes are labelled. B Sequence alignments of five Tekay retrotransposons highlighted in (A) panel. The yellow rectangle represents terminal LTRs, while the blue ones represent internal protein-coding domains, including group-specific antigen (GAG), protease (PROT), reverse transcriptase (RT), ribonuclease H (RH), integrase (INT), and a chromodomain (CHD). The grey boxes show the sequence identity between two elements. C The relative distance of tetraploid tobacco centromeres to CRM and Nt2-7 enriched regions, and diploid ancestry centromeres. D The recent centromere repositioning occurred from diploid ancestors to tetraploid tobacco. The blue and red lines represent the location of Nt2-7 and CRM. E Summary of evolutionary trajectories of tobacco centromeres. The common ancestor of tobacco and tomato shared an ancestral centromere dominated by Tekay. Following species divergence, ancient centromere shift occurred between the diploid N. sylvestris and N. tomentosiformis lineages, accompanied by the invasion of CRM and subgenome-specific Nt2-7 elements. After the allopolyploidization event, the tetraploid genome underwent asymmetric subgenomic evolution and frequent centromere repositioning. The sequence identity is indicated by gray shaded blocks: deep gray > medium gray > light gray Polyploidization via genomic duplication and hybridization is a key driver of genomic evolution in plants, causing a genomic shock and instability to centromeres. To illustrate the centromere dynamics switching from diploid to tetraploid in Nicotiana genus, we predicted the centromere positions from diploid parental species based on the strong inter-chromosomal Hi-C interaction signals [20] (Additional file 1: Table S8; Additional file 2: Figs. S22-S24). Syntenic analysis revealed a substantial number of centromere repositioning and rearrangement after allopolyploidization (Additional file 2: Figs. S25-S26). We observed a strong spatial congruence between Nt2-7- and CRM-enriched regions, particularly within the subT (Fig. 2C), which either overlapped with the tetraploid or diploid peri-/centromeres, or were positioned within neighboring zones to them. Our model suggests these co-localized regions resemble ancient centromeric footprints with similar structure, although the enrichment of Nt2-7 exhibits a distinct subgenome bias. These sites subsequently underwent centromere repositioning, and gave rise to the current centromere landscapes. For instance, despite the well alignments of peri-/centromeric regions between subS and N. sylvestris, five centromeres showed megabase-scale shifts. The most notable example was found on CenS08, which was devoid of satellites in N. sylvestris, but shifted to 90-bp NTS9 satellite arrays at an ectopic location without sequence rearrangement in N. tabacum (Fig. 2D; Additional file 2: Fig. S11). On the contrary, sequence collinearity between subT and N. tomentosiformis was relatively low across centromeres, further supporting the rapid evolution of T-genome (Additional file 2: Fig. S26). We observed multiple inversions inside or in vicinity of the functional centromeres, which might contribute to centromere shift and evolution, consistent with previous research in einkorn [21]. For example, the large inversion displaced parts of functional centromeres, which might then drive the repositioning of CenT14 in tobacco genome (Fig. 2D). We also noticed that the centromeres were highly dynamic among Nicotiana species (Additional file 2: Fig. S27). The relatively distant position of centromeres between homeologous chromosome pairs, both in diploids and tetraploids, suggested ancient centromere shifts between T- and S- genomes (Additional file 2: Figs. S15, S27). This is in agreement with our previous observation in N. benthamiana [19], indicative of extensive centromere rearrangements following divergence of Nicotiana species. Intriguingly, N. tabacum is a young polyploid (~ 0.2 Mya), while N. benthamiana is a much older polyploid (~ 6 Mya) [1]. Thus, we hypothesize that the species divergence first drove the centromere shifts among diploid Nicotiana progenitors, which then repositioned during transition from diploid to tetraploid (Fig. 2E). Notably, the satellite-based centromeres in N. benthamiana likely de novo emerged during allopolyploidization, characterized by rDNA-derived 33-/43-bp microsatellites [19], whereas those in N. tabacum arose from the displacement on pre-existing satellite arrays, suggesting that the temporal divergence in polyploid speciation might account for this distinct evolutionary trajectory. Conclusions In summary, we achieve the first T2T gap-free and phased polyploid genome assembly for the cultivated tobacco, providing an essential genomic resource for future precise breeding and functional genomics studies. We also reveal the highly dynamic and rapidly evolving tobacco centromeres during polyploidization process, coupled with ancient and recent centromere shifts, which together enrich our knowledge on centromere architecture, diversity and evolution in Nicotiana species. Methods Plant materials and genome sequencing The Nicotiana tabacum cv. Yabuli plants were routinely grown in a custom soil mix with a day/night cycle of 16 h/8 h at a constant temperature of 25 °C in greenhouse of Peking University Institute of Advanced Agricultural Sciences (PKU-IAAS) (Weifang, China). After two weeks, the fresh young leaves were collected and used for genome sequencing and ChIP-seq. Tissues of roots, leaves, stems, flowers and seeds at two days post anthesis were used to perform Illumina RNA-sequencing on Novaseq 6000 platform and full-length transcriptome sequencing on PacBio Sequal IIe instrument. Whole genome sequencing High molecular weight (HMW) genomic DNA was isolated using CTAB method, followed by purification using a QIAGEN® Genomic Kit (QIAGEN, CA, USA). The genomic DNA was checked by Qubit (Thermo Fisher Inc.) and used for library construction. Illumina paired-end sequencing library was prepared using NEB Next® Ultra™ DNA Library Prep Kit for Illumina (NEB, USA) following its standard protocol, and sequenced on Illumina Novoseq 6000 platform. Standard PacBio SMRTbell library was constructed using PacBio SMRTbell Express Template Prep Kit 2.0 (PacBio, CA, USA), and sequenced using the PacBio Sequel II system at Biomarker Technologies Corporation (Qingdao, China). HMW genomic DNA used for the construction of ONT ultralong sequencing library was prepared by the nuclei method [22]. The quality of HMW genomic DNA was checked using a pulse field gel electrophoresis apparatus (BioRad). Then the long DNA fragments were size-selected and processed following the Ligation Sequencing SQK-LSK109 Kit (ONT, Oxford, UK) protocol. The final DNA library was sequenced using the GridION X5/PromethION sequencer (ONT, Oxford, UK) via the Anoroad Gene Technology (Beijing, China) and the PKU-IAAS Single-Cell and Single-Molecule Core Facility (Weifang, China). The Hi-C library was prepared from cross-linked chromatins of young fresh leaves according to the standard protocol [23]. The cells were lysed and subjected to DpnII restriction enzyme digestion, followed by ligation with biotin. Then the final library was sequenced using Illumina NovaSeq 6000 to obtain 2 × 150-bp paired-end reads at Biomarker Technologies Corporation (Qingdao, China). Transcriptome sequencing The isolation of total RNA was conducted using Trizol RNA extraction reagent (Thermo Fisher Inc.) following standard protocol. The extracted RNA was assessed using RNA Nano 6000 Assay Kit of the Bioanalyzer 2100 system (Agilent Technologies, CA). RNA samples with a RIN (RNA integrity number) > 6.0 were proceeded to downstream library construction using Illumina True-seq Transcriptome Kit (Illumina, CA). The libraries were then sequenced using Illumina Novaseq 6000 platform to generate 150-bp paired-end reads. For full-length transcriptome sequencing, about 5 µg mRNA was reverse-transcribed into full-length cDNA molecules with SMARTer™ PCR cDNA Synthesis Kit (Clontech, CA, USA), and the cDNA was then PCR amplified, end repaired and adapter ligated. The ligation products were further treated by exonuclease to degrade failed ones. Finally, the Iso-Seq library was sequenced using PacBio Sequal IIe instrument at Biomarker Technologies Corporation (Qingdao, China). Full-length transcripts were assembled using the custom SMRT Analysis software Isoseq3 pipeline (https://github.com/PacificBiosciences/IsoSeq) and used for guiding gene annotation. CENH3 ChIP-seq ChIP experiment was performed using the leaves of N. tabacum and an anti-CENH3 antibody [19]. In brief, young leaves were fixed with fixing buffer (1% formaldehyde, 10 mM potassium phosphate pH 7.0, 50 mM NaCl) at room temperature for 15 min under vacuum, and then incubated with addition of 0.15 M glycine for another 5 min. Then, 1 g fixed tissue was homogenized with liquid nitrogen, resuspend in 1 ml of cell lysis buffer (10 mM Tris, 10 mM NaCl, 0.2% NP-40 pH 8.0, 1 × protease inhibitors), incubated for 10 min on ice and centrifuged at 1,500 rpm (RC-3B, 600 × g) for 5 min. After that, the nuclei were resuspended in 1 ml of nuclei lysis buffer (50 mM Tris, 10 mM EDTA, 1% SDS, 1 × protease inhibitors) and incubated for 10 min on ice. The resuspended chromatin solution was sonicated for five cycles of 15 s on and 15 s off, and then centrifuged at 13,000 rpm for 10 min at 4 °C. The sonicated chromatin supernatant was incubated with 2.5 µg of anti-CENH3 for 12 h at 4 °C, and then incubated with 50 µl protein A/G beads for 4 h at 4 °C. The incubated beads were washed twice with each of the following buffers: wash buffer A (50 mM HEPES–KOH pH 7.5, 140 mM NaCl, 1 mM EDTA pH 8.0, 0.1% Na-Deoxycholate, 1% Triton X-100, 0.1% SDS), wash buffer B (50 mM HEPES–KOH pH 7.9, 500 mM NaCl, 1 mM EDTA pH 8.0, 0.1% Na-Deoxycholate, 1% Triton X-100, 0.1% SDS), wash buffer C (20 mM Tris–HCl pH8.0, 250 mM LiCl, 1 mM EDTA pH 8.0, 0.5% Na-Deoxycholate, 0.5% IGEPAL C-630, 0.1% SDS), wash buffer D (TE with 0.2% Triton X-100), and TE buffer. DNA was eluted off the beads in 200 μl elution buffer (50 mM Tris–HCl pH 8.0, 10 mM EDTA, 1% SDS) and incubated at 65 °C overnight. To purify eluted DNA, 200 μl TE was added, and then RNA was degraded by incubation with 80 μg RNase A (Sigma, R4642) at 37℃ for 2 h, protein was degraded by incubation with 200 mg proteinase K (Invitrogen, 25,530,049) at 55℃ for 2 h. The phenol: chloroform: isoamyl alcohol extraction was performed followed by an ethanol precipitation. The DNA was then resuspended in 50 μl TE and the final ChIP library was amplified with the VAHTS® Universal DNA Library Prep Kit for Illumina V3 (Vazyme, ND607), and sequenced on the Illumina Novaseq 6000 platform to produce 150-bp paired-end reads at Shandong Xiuyue Biotechnology Co., Ltd (Jinan, China). Genome assembly To obtain a high-contiguity assembly, hifiasm (v0.19.5) [17] was employed combining the HiFi and ONT reads with the parameters of “-l 0 –n-hap 4”. The draft HiFi & ONT assembly sized 4.26 Gb with contig N50 of 162.2 Mb. To close any potential gaps, we also assembled the ultra-long ONT reads using Canu (v2.2) [24], and obtained an ONT assembly in 4.18 Gb with contig N50 of 44.1 Mb. Then Hi-C reads were used to anchor all assembly contigs through the pipeline of BWA (v0.7.17) [25], Juicer (v1.5) [26], 3D-DNA (v180419) [27] and Juicebox (v1.11.08) [28]. We compared these two assemblies through minimap2 (v2.24) [29] and D-Genies [30], and manually filled the gaps in HiFi & ONT assembly using sequences from the ONT assembly. The remaining gaps were closed using TGS-GapCloser (v1.2.1) [31] with corrected ONT reads. Then, the 24 chromosomal-level contigs was polished using the strategy previously developed in human T2T genome project [32]. Finally, we obtained a T2T gap-free assembly of N. tabacum with total length of 4.20 Gb and contig N50 of 179.6 Mb. Genome evaluation To validate the large-scale assembly quality, we aligned against the published reference assembly using SyRI [33], and manually checked the Hi-C interaction matrix. Assembly and annotation completeness were evaluated via both BUSCO (v5.4.3) [34] and OMARK (v2.0.3) [35] analysis. Specifically, BUSCO was run against the solanales_odb10 database (n = 5,950), while OMARK utilized the LUCA.h5 database to assess proteome consistency. LTR Assembly Index (LAI) calculated using LTR_retriever (v2.9.0) [36] was applied to evaluate the assembly continuity of LTR retrotransposons. Telomeres were identified by screening high copies of canonical telomeric repeat motif ‘TTTAGGG’ at the distal ends of each chromosome. To assess assembly correctness, we estimated quality value (QV) using Merqury (v1.3) [37] from HiFi reads. We also mapped HiFi reads against the T2T assembly to check any abnormal coverage signals. Repeat sequence annotation Annotation of whole genome repeats was performed using pipeline of RepeatModeler (https://github.com/Dfam-consortium/RepeatModeler) and RepeatMasker (v4.1.2) [38] with default setting. The nonredundant repetitive library included the universal Repbase database (version 20,181,026) and the N. tabacum specific repeat library constructed by RepeatModeler. Then RepeatMasker was used to annotate the repetitive elements and soft-mask the genome. The intact LTR retrotransposons were identified combining the tools of LTR_Finder (v1.2) [39], LTRharvest (v1.6.2) [40], and LTR_retriever (v2.9.0) [36], which were then passed to TEsorter (v1.3) [41] to classify their subfamily. The parameters of LTRharvest were as follows: “-seqids yes -similar 80 -vic 10 -seed 20 -minlenltr 100 -maxlenltr 7000 -mintsd 4 -maxtsd 6”. The Ty3-Gypsy elements were classified into seven clades as Athila, CRM, Galadriel, Ogre, Reina, Retand, and Tekay. The insertion time of intact LTR retrotransposons were calculated via an average substitution rate of 1.3 × 10−8 per site per year using LTR_retriever [36]. Phylogenetic analysis of Tekay For the phylogenetic analysis, we extracted the subset (33,357) of Tekay elements that in correct order of five core domains (gag, protease, reverse transcriptase, RNaseH and integrase) according to TEsorter [41]. Then we aligned the concatenated full domains with MAFFT (v7.505) [42], and constructed the maximum-likelihood phylogenetic tree using IQ-TREE (v2.0.3) [43]. We also conducted same analysis on the subset (391) of full-length centromeric Tekay retrotransposons. Fluorescence in situ hybridization (FISH) Two DNA probes were used for in situ hybridization: (1) clone pTa71, containing the 18S-5.8S-26S rDNA repeat unit from Triticum aestivum, and (2) clone pTa794, that includes the complete 5S rRNA gene unit from Triticum aestivum, which were labelled with Fluorescein-12-dUTP (green) and Texas Red-5-dUTP (red) using nick translation, respectively. FISH experiment of mitotic metaphase chromosomes was performed as previously reported with minor modifications [18]. Briefly, young root tips were first incubated in a pressure-tolerant cylinder, with nitrous oxide gas of 1.0 MPa applied for 2 h, and then fixed in a 3:1 (v/v) methanol: acetic acid solution for 1 h in an ice bath. After this, the fixed samples were washed twice with distilled water, and digested in an enzyme mixture of 0.3% Pectinase, 0.3% Driselase and 0.3% Cellulase at 37 °C for 1 h. The digested root tips were crushed with tweezers gently to form a fine suspension, and 10 µL of 60% acetic acid was added twice to a 10 µL of suspension on the slide to facilitate chromosomal spreading, Then, the slides were incubated at 45 °C for 2 min, flooded with Carnoy’s fixative and air-dried. After probe hybridization, chromosomes were counterstained with 4',6-diamidino-2-phenylindole (DAPI), and fluorescence signals were captured using a Nikon Ni-U fluorescence microscope (Nikon, Tokyo). Gene model annotation The structure of the protein-coding genes was predicted using MAKER (v2.31.11) [44] and BRAKER2 (v2.1.6) [45] pipeline combining evidence from homology protein, RNA-seq transcript, and ab initio gene prediction. The proteins used for homology-based prediction were from N. tabacum NtaSR1 [4], N. benthamiana NbT2T [19], S. lycopersicum SL6.0 [46], S. tuberosum DM6.0 [47], and universal Swiss-Prot proteins. RNA-seq reads were assembled into transcripts using HISAT2 (v2.2.1) [48] and StringTie (v1.13) [49]. The full-length transcriptome reads were processed using the SMRT Analysis software Isoseq3 (https://github.com/PacificBiosciences/IsoSeq). The SNAP [50], GeneMark-ET [51], and AUGUSTUS [52] models were trained using MAKER and BRAKER2 pipeline as previously reported [11]. Finally, the trained models, protein and transcript sequences were integrated into MAKER to predict credible gene structures. Gene models with complete open reading frame were retained, and manually corrected in IGV-GSAman (https://gitee.com/CJchen/IGV-sRNA) with the support of transcript coverage and previous annotations [4, 5]. Subgenome assignment We used SubPhaser (v1.2.5) [53] to phase and partition the subgenomes of allotetraploid N. tabacum based on differential k-mers among homoeologous chromosome sets. Briefly, the syntenic analysis was firstly performed by JCVI (v1.1.19) [54], and genes in syntenic blocks were identified as homeologues. Then SubPhaser searches for the subgenome specific k-mers, and assigns homoeologous chromosomes into two subgenomes. Homoeologous exchange regions were identified using aforementioned specific k-mers. If a region on the S subgenome is dominated by T-specific k-mers, it is classified as an exchange region. The whole-genome synteny plot between two subgenomes was generated and visualized using JCVI package [54] and GENESPACE (v0.9.3) [55]. Transcriptome analysis RNA-seq data reads (NCBI accession, PRJNA940510) were trimmed using fastp [56] and then aligned to NtT2T transcripts using kallisto program [57]. The expression level was quantified as transcripts per million (TPM). To compare homologous gene expression between subgenomes, expression analysis was restricted to 22,018 homologous gene pairs as described in Ref. [5]. Paired Wilcoxon signed-rank test was used to calculate P values for all the comparisons. Epigenomic data processing Paired-end ChIP-seq reads were preprocessed with fastp (v0.23.2) [56] to remove adaptors and low-quality bases, and then aligned to the assembly using bowtie2 (v2.5.1) [58] with default setting. Alignments with mapping quality of < 30 were discarded, and all PCR duplicates were removed using SAMtools (v1.10) [59] with the command SAMtools rmdup. Then, the bamCompare tool from deepTools (v3.5.1) [60] was employed to quantify CENH3 enrichment expressing as the log2 of the ratio (ChIP/control). The centromere positions of two diploid ancestors of N. sylvestris and N. tomentosiformis were predicted by Hi-C contact maps (Additional file 2: Fig. S21), according to previous report [20]. These two near-T2T assemblies and corresponding Hi-C sequencing data were downloaded from a recent study [5]. To detect DNA methylation state, we used a deep-learning method by DeepSignal-plant (v0.1.6) to call CG, CHG and CHH contexts from ONT reads [61]. Briefly, raw fast5 files were first basecalled using Guppy (v6.3.7, https://nanoporetech.com) with the config of ‘dna_r9.4.1_450bps_hac_prom.cfg’ and then re-squiggled using tombo (v1.5.1, https://github.com/nanoporetech/tombo). Then, DNA methylation for CG, CHG, and CHH contexts were called with three respective models, and methylation frequencies at each site were generated by call_freq function. DNA methylation proportions for cytosines covered by less than five reads were excluded. To evaluate gene methylation levels, we computed the mean methylation levels across the gene body and its adjacent 2-kb regions. No other scripts and software were used other than those mentioned in the “Methods” section. Supplementary Information Additional file 1: This file contains Tables S1-S8.Additional file 2: This file contains Figures S1-S27. Additional file 1: This file contains Tables S1-S8. Additional file 2: This file contains Figures S1-S27.