Combined SMRT and Illumina RNA Sequencing Reveals the Alternative Splicing-Mediated Regulation of Anthocyanin Accumulation in Potato (Solanum tuberosum L.)

TL;DR

This study shows how alternative splicing of a specific gene affects anthocyanin production in potatoes.

Contribution

The study reveals how alternative splicing of the StB3-like gene regulates anthocyanin accumulation in potatoes.

Findings

Alternative splicing events were identified in anthocyanin-related genes and transcription factors in potato.

StB3-like gene produces three spliced variants, two of which enhance anthocyanin accumulation when co-expressed with StAN1.

The regulatory role of alternative splicing in anthocyanin biosynthesis is confirmed in a StAN1-dependent manner.

Abstract

Anthocyanins, as natural pigments with high nutritional value, have been extensively studied in terms of their biosynthetic pathways. However, the specific impact of alternative splicing (AS) on anthocyanin biosynthesis in potatoes and its potential regulatory mechanisms remain unclear. This study analyzed full-length transcriptome (SMRT) and Illumina RNA-seq datasets from three developmental stages of multiple potato (Solanum tuberosum L.) cultivars. After RNA-seq correction, numerous AS events were identified in both white and colored samples. To dissect the regulatory role of AS in anthocyanin biosynthesis, we further analyzed the AS patterns of anthocyanin biosynthesis-related structural genes and transcription factors (TFs). Through this targeted analysis, we found that a subset of these structural genes and TFs exhibited AS, generating functionally diverse transcript variants. Among these, the candidate gene StB3-like (Soltu.DM.04G010530), a B3 family TF associated with anthocyanin synthesis, was selected for preliminary functional validation. This gene produced three alternatively spliced transcripts (StB3-like-1, -2, -3). Transient co-expression with StAN1 (Soltu.DM.10G020850) in tobacco showed that StB3-like-2 and StB3-like-3 significantly increased anthocyanin accumulation, whereas StB3-like-1 had no effect. These results confirm that different transcripts produced by the alternative splicing of the same gene differentially regulate anthocyanin accumulation in a StAN1-dependent manner, laying the groundwork for further investigation into the role of alternative splicing in potato anthocyanin accumulation.

Figures

Full text

1. Introduction

Originating from the Andean regions of Chile and Peru, potato (Solanum tuberosum L.) is an annual tuber crop of the Solanaceae family. It is widely cultivated in over 150 countries and regions globally, owing to its excellent stress resistance, low resource requirements, and high yield potential [1]. Its production scale is second only to that of wheat (Triticum aestivum) and rice (Oryza sativa), and it is the most productive non-cereal food crop globally [2]. Potato cultivation in China began in the early 19th century, initially introduced as a staple food and disaster relief crop in southwest China [3,4]. Since the 21st century, potato production has continued to grow. By 2024, China’s potato harvest area had reached over 4.5 million hectares, with production exceeding 94 million tons [5]. Special attention has been given to the colored potato germplasm resources formed by the natural mutation of common cultivars, whose periderm (tuber skin) and perimedullary tissue (tuber flesh) exhibit characteristic colors such as red, blue, purple, and black [6]. Colored potatoes not only exhibit phenotypic diversity but also contain anthocyanin levels ranging from 10 to 70 mg per 100 g FW—3 to 4 times higher than those of traditional varieties. While traditional varieties possess antioxidant capacities equivalent to 930 to 1380 mg Trolox equivalents per kg FW, colored varieties surpass these values by 2.5 to 3 times [7].

Anthocyanins are a class of water-soluble phenolic pigments belonging to the flavonoid group, widely distributed in the cell sap of various plant organs [8,9]. Six of the most common anthocyanins have been identified in potatoes. Pelargonidin-3-(p-coumaroyl-rutinoside)-5-glucoside is the primary anthocyanin in red-fleshed potatoes, while petunidin-3-(p-coumaroyl-rutinoside)-5-glucoside is the predominant anthocyanin in purple-fleshed potatoes [10]. Their biosynthetic pathway begins with phenylalanine, which is converted into 4-coumaroyl-CoA via the phenylpropanoid pathway [11], followed by the sequential generation of intermediate products through the flavonoid synthesis pathway, and finally the formation of anthocyanins via specific modifications [12]. In potatoes, this process is directly regulated by structural genes such as PAL, CHS, CHI, F3H, DFR, F3′H, F3′5′H, ANS, and UFGT while also being coregulated by modifying enzymes like acyltransferases [13]. Additionally, numerous transcription factors associated with potato anthocyanin biosynthesis have been identified, including MYB, bHLH, WKRY, WD40, and others. For example, StMYBATV competitively binds to StbHLH1 to regulate potato anthocyanin biosynthesis, while StMYB3 negatively regulates tuber anthocyanin biosynthesis [14]. StWRKY13 activates the activity of StCHS, StF3H, StDFR, and StANS, thereby promoting anthocyanin biosynthesis in potato tubers [15]. Although the biosynthetic pathway of anthocyanin glycosides has been studied in depth, its regulatory mechanisms in different plants still require further clarification. Especially in herbaceous crops like potatoes, genetic research is limited due to the lack of sufficient mutant resources.

As a dual-purpose crop serving as both a staple food and vegetable [16], potatoes not only provide traditional nutrients, such as starch (approximately 18–21%), protein (2–3%), and various minerals and vitamins, but are also rich in bioactive compounds including phenolic compounds [17], glycoside alkaloids [18], and dietary fiber [19]. These components confer multiple physiological benefits, including antioxidant and anti-inflammatory effects, the regulation of gut microbiota, and improvement in metabolic health [16,19,20]. Research confirmed that anthocyanins in colored potatoes hold significant potential for preventing alcoholic liver damage. One study established a chronic alcoholic liver injury model using 64 three-week-old Kunming mice (equal numbers of males and females). Mice were administered anthocyanins extracted from “Heijingang” potatoes (primarily containing petunidin-3-coumarosyl-5-glucoside) via oral gavage combined with three anthocyanin doses (2, 5, 10 mg/kg/day) over a 5-week period. The results demonstrated that this anthocyanin effectively alleviated alcohol-induced oxidative stress and liver damage [21]. Field observations revealed that colored potatoes were significantly more resistant to late blight than common varieties, which may be closely associated with the bacteriostatic activity of anthocyanin glycosides [22]. Phenylpropanoid compounds belong to the class of plant-induced antimicrobial substances. When plants encounter biotic stresses such as pathogen infection, the phenylpropanoid biosynthesis pathway within them is rapidly activated. PAL serves as the central key enzyme in this metabolic pathway. It catalyzes the deamination of phenylalanine to produce cinnamic acid, thereby mediating the biosynthesis of downstream phenolic compounds and ultimately enhancing the plant’s inherent antimicrobial defense capabilities [23]. Colored potatoes exhibit excellent light and heat stability, high yields, environmental adaptability, and low production costs. Their anthocyanins can be used as natural colorants in baked goods and beverages and also serve as natural preservatives in food applications [24]. Among these types of potatoes, Purple Majesty exhibited the highest anthocyanin content at 113.19 mg/100 g FW [25]. Given the coregulatory relationship between the anthocyanin biosynthesis pathway and defense/metabolic pathways, anthocyanin phenotypes can assist in the indirect screening of resistance-related traits (such as late blight resistance) [26].

Transcriptomes reflect the types and quantities of genes in cells, revealing the molecular mechanisms underlying physiological and biochemical processes [27]. RNA-seq, based on the Illumina platform, is widely used for transcript structure analysis, differential gene expression detection, alternative splicing (AS) identification, and transcript abundance quantification [28]. However, to overcome the limitations of short read lengths and improve sequencing accuracy, the third-generation sequencing technology single-molecule real-time (SMRT) sequencing has emerged [29,30]. SMRT sequencing achieves an average read length of over 10 kb (up to 60 kb) [31], enabling full-length transcriptome analysis that compensates for the shortcomings of second-generation sequencing [32]. Currently, SMRT sequencing is widely applied in the full-length transcriptome analysis of plants, particularly in AS event detection, significantly improving isoform identification sensitivity [33,34,35]. Thus, SMRT sequencing has distinct advantages over RNA-seq in AS analysis.

AS is the process by which a single precursor mRNA is spliced into multiple mature mRNAs so that the same gene produces multiple different mature mRNAs, which ultimately encode different proteins [36]. In organisms, five main types of AS exist, which are exon skipping (ES), alternative 5′ splice site (A5), alternative 3′ splice site (A3), intron retention (IR), and mutually exclusive exon (MEE) [37]. Studies have shown that alternative splicing plays a key role in many processes such as stress response [38,39,40], circadian regulation [41], and growth and development in plants. For example, the interaction between CcMYB6-1 and CcbHLH1 enhances the activity of the CcF3H and CcDFR promoters. The AS of CcbHLH1 prevents it from co-activating the anthocyanin biosynthesis pathway with CcMYB6-1, leading to white cornflowers [42,43,44]. Different splicing variants of CsbHLH133 respond to multiple stress conditions through specific regulatory patterns, thereby modulating the biosynthesis of geraniol in tea plants [45]. The AS of CA10g11690 in chili peppers contributes to purple exocarp development [46], and the AS of CsTT8 in the fourth exon in blood oranges forms Δ15-TT8, which may downregulate TT8 expression and affect fruit pigmentation [47]. In tea plants, the expression levels of genes including C4H, CHS, FLS, F3′5′H, PAL, UFGT, ANR, ANS, F3H, 4CL1, C3H, CHI, and DFR and their AS transcripts were validated. The results indicate that the AS transcripts of C4H1, FLS1, and PAL2 are key transcripts regulating anthocyanin biosynthesis in tea plants [48]. These studies strongly suggest that AS events are critical for the accumulation of anthocyanins in plants. Although some AS events have been identified during plant developmental stages, the regulatory mechanisms behind these events still need to be explored in depth.

Colored potatoes hold significant research value, yet the post-transcriptional regulatory mechanisms governing anthocyanin accumulation remain incompletely elucidated. This study’s significance resides in integrating SMRT sequencing and RNA-seq to analyze three different-colored potatoes across three developmental stages (tuber development stage, tuber enlargement stage, tuber harvest stage). Through SSR locus analysis, lncRNA characterization, and open reading frame (ORF) prediction, we identify differentially expressed genes and AS-related genes in anthocyanin biosynthesis—notably validating that StB3-like’s three AS-derived transcripts differentially regulate anthocyanin accumulation. This study fills a research gap in the role of alternative splicing within the potato anthocyanin biosynthesis regulatory pathway, clarifies a key post-transcriptional regulatory mechanism, and lays a solid foundation for further research on potato anthocyanin molecular mechanisms.

2. Results

2.1. Construction and Annotation of Full-Length Transcriptome of Potato

SMRT sequencing was performed separately on white (F01) and colored (F02) potato cultivars. After quality control, clean data volumes of 17.07 GB and 23.36 GB were obtained for F01 and F02 (Table S1). Based on cyclic common sequence (CCS) analysis, 365,197 and 566,802 CCSs were identified in the F01 and F02 samples, respectively (Table S2). The CCSs were categorized into full-length sequences and non-full-length sequences by detecting whether they contained 5′ primers, 3′ primers, and poly(A) tail structures. Among them, 309,043 and 481,798 full-length non-chimeric (FLNC) sequences were identified in F01 and F02, which accounted for 84.62% and 85.00% of the total number of corresponding CCSs, respectively (Table 1). After correcting the obtained FLNC sequences, 75,864 high-quality transcripts and 4330 low-quality transcripts were obtained in F01, and 102,209 high-quality transcripts and 3378 low-quality transcripts were obtained in F02 (Table 2). Subsequently, the corrected consensus sequences were aligned with the reference genome using GMAP (Genomic Mapping and Alignment Program), redundant transcripts were removed using cDNA_Cupcake (https://github.com/Magdoll/cDNA_Cupcake/wiki, accessed on 25 November 2025), and finally 38,663 and 46,249 non-redundant transcript sequences were obtained in the F01 and F02 samples, respectively. Alternative splicing analysis was performed on the de-redundant transcripts, and 15,270 and 16,852 loci were detected in the F01 and F02 samples, respectively.

2.2. Functional Annotations for New Transcripts

We utilized Clusters of Orthologous Groups (COG), Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), the euKaryotic Ortholog Groups (KOG), Protein family (Pfam), Swiss-Prot Protein Sequence (Swiss-Prot), eggNOG, and NCBI Non-Redundant Protein Database (NR) for the functional annotation of proteins encoded by new transcripts. The annotation results showed that 17,840 (42.35%) new transcripts were annotated in the COG database, 33,582 (79.72%) in the GO database, 19,959 (47.38%) in the KEGG database, 27,266 (64.72%) in the KOG database, 30,703 (72.88%) in the Swiss-Prot database, 39,755 (94.37%) in the eggNOG database, and 42,065 (99.86%) in the NR database (Table S3). Combining the eight databases, a total of 8395 novel transcripts were co-annotated (Figure S1). KOG analysis showed that the 27,266 annotated novel transcripts were classified into 25 functional categories. Among them, the three categories with the highest percentage were General function prediction only; Signal transduction mechanisms; and Posttranslational modification, protein turnover, and chaperones (Figure S2). A total of 33,582 annotations were found in the GO enrichment analysis. The 33,582 annotated novel transcripts in the GO enrichment analysis were mainly enriched in different cellular components (CCs), molecular functions (MFs), and biological processes (BPs). Specifically, at the cellular component level, new transcripts were mainly related to cells and organelles; at the biological process level, they were mainly related to metabolic processes and cellular processes; at the molecular function level, the pathways were closely related to protein binding and catalytic activity (Figure S3). KEGG pathway enrichment analysis indicated that the most significantly enriched pathways for the new transcripts were Starch and sucrose metabolism, Protein processing in the endoplasmic reticulum, and the Spliceosome (Figure S4).

2.3. Prediction of CDSs, lncRNAs, and SSRs

TransDecoder software (v.3.0.0) was used to predict potential coding regions based on transcriptome sequences, and open reading frame (ORF) annotations were predicted for novel transcripts identified in the variable splicing analysis. This study predicted a total of 42,026 ORFs, including 36,002 complete ORFs. The protein sequences encoded by the predicted complete ORFs ranged in length from 100 to 2300 bp (Figure S5). Among them, proteins with lengths of 100–200 amino acids were the most abundant (8586, 23.85%), followed by proteins with lengths of 200–300 amino acids (7029, 19.52%) (Table S4).

Long-stranded noncoding RNA (lncRNA) usually refers to RNA molecules that are greater than 200 nt in length and similar in structure to mRNA but do not have the ability to code for proteins [49]. Studies have shown that lncRNAs play important roles in various biological processes such as epigenetic regulation, transcriptional regulation, and post-transcriptional regulation. In this study, four methods, namely CPC, CNCI, Pfam, and CPAT, were combined for lncRNA prediction, and a total of 1544 lncRNA transcripts were identified (Figure 1A). Based on their genomic localization, these lncRNAs were classified into the following: intergenic lncRNAs (840), antisense lncRNAs (103), intronic lncRNAs (80), and sense lncRNAs (461) (Figure S6).

Simple sequence repeat (SSR) markers are widely used in the fields of plant genetic map construction, variety identification, genetic diversity analysis, and molecular marker-assisted selection because of their strong polymorphism, good repeatability, and high stability. In this study, 67,025 sequences (total length 148,382,563 bp) with a length greater than 500 bp were analyzed in terms of SSRs, and a total of 22,777 SSR sites were detected. Among them, 17,540 sequences contained at least one SSR site; among these 17,540 sequences, 3920 sequences contained more than one SSR site. The detected SSRs can be categorized into seven groups according to the type of repeat unit: Mono-nucleotide (13,270), Di-nucleotide (3804), Tri-nucleotide (5316), Tetra-nucleotide (201), Penta-nucleotide (52), Hexa-nucleotide (134), and compound SSRs (1560). The statistics showed that single-base repeat types were the most abundant, followed by three- and two-base repeat types, and five-base repeat types were the least abundant (Figure 1B, Table S5).

2.4. Identification of Alternative Polyadenylation (APA) and AS

The post-transcriptional processing of pre-mRNA into mature mRNA primarily includes 5′ end capping, 3′ end poly(A) tail addition, and intron splicing. Notably, the positional heterogeneity of the 3′ end poly(A) tail (variable poly(A) sites) can influence the binding of RNA-binding proteins to mRNA, thereby regulating RNA splicing and translation processes. This study analyzed the 3′ ends of transcripts based on PacBio sequencing data. The results showed that in the F01 sample, there were 3666 transcripts containing a single poly(A) site, while there were 279 transcripts containing five or more poly(A) sites (Figure 1C). In the F02 sample, there were 3730 transcripts with a single poly(A) site and 512 transcripts with five or more poly(A) sites (Figure 1D).

To investigate the effect of AS on anthocyanin accumulation in potatoes, this study performed alternative splicing analysis on 38,663 and 46,249 non-redundant transcripts. The results showed that the two groups of samples detected 15,270 and 16,852 gene loci, respectively, including 969 and 1270 new gene loci and 22,743 and 26,751 new transcripts, respectively. The alternative splicing types in the two samples were consistent, ranked in descending order by frequency as follows: intron retention (IR), alternative 3′ splice site (A3), exon skipping (ES), alternative 5′ splice site (A5), and mutually exclusive exon (MEE). Among these, the white variety detected a total of 12,524 AS events, including 132 MEE, 7548 IR, 1321 ES, 1711 A5, and 2353 A3 events (Figure 1E). The colored variety detected 14,287 AS events, including 170 MEE, 8064 IR, 1739 ES, 1447 A5, and 2867 A3 events (Figure 1F).

2.5. Transcription Factor Prediction

Using iTAK software (v.1.5) for transcription factor prediction, a total of 2767 transcription factors (TFs) were identified, belonging to 68 different families. Among these, the bHLH family had the largest number of members (207), followed by the MYB family (173), C2H2 family (165), AP2/ERF-ERF family (160), NAC family (139), WRKY family (123), and bZIP family (122) (Figure 2A). Additionally, nine TF families had member counts between 50 and 120, specifically including the B3, MADS-M-type, MYB-related, C3H, GARP-G2-like, GRAS, HB-HD-ZIP, C2C2-GATA, and LOB families.

2.6. Short-Read Transcriptomic Correction for Full-Length Transcriptomics

RNA-seq was performed on 27 samples from three different-colored cultivars at three developmental stages using the Illumina platform. After quality control filtering to remove low-quality sequences, high-quality clean reads were obtained. The Q30 score was approximately 93%, and the GC content was approximately 44%, indicating good sequencing data quality (Table S6). Subsequently, the clean reads were aligned to the potato full-length transcriptome (FL transcriptome) reference sequence, and gene expression levels were estimated using the FPKM (Fragments Per Kilobase of transcript per Million mapped reads) method. To assess the relationships between samples, we performed principal component analysis (PCA) based on the FPKM expression data. The PCA results showed that all biological replicate samples clustered closely together, indicating high correlation among them and good experimental reproducibility. Samples from the three cultivars and three different developmental stages exhibited a clear separation trend (Figure 2B). These results suggest that developmental stage is the primary factor influencing transcriptomic changes associated with potato tuber color.

2.7. Differential Gene Analysis

To validate the reliability of the differentially expressed gene (DEG) (log_2_FC > 1, FDR < 0.05) analysis results, Pearson correlation analysis was performed on each sample in this study. The results showed that the coefficient of determination (R^2^) between the three biological replicate samples under each experimental condition was greater than 0.7, indicating high consistency among biological replicates and providing a stable and reliable data foundation for subsequent DEG analysis (Figure 2C). Through systematic analysis, a total of 9350 DEGs were identified in the three potato clones. This study found that the number of DEGs showed a significant increasing trend as tuber development progressed. Specifically, the number of upregulated differentially expressed genes in the three clonal lines at the three developmental stages was 1336–1568, 1551–2045, and 1972–2323, while the corresponding numbers of downregulated differentially expressed genes were 1553–1660, 1165–1641, and 1410–1749, respectively (Figure 2D). Additionally, when comparing yellow and colored potato clones, a total of 4904 genes showed significant differential expression, with 1081, 1263, and 1380 differentially expressed genes in the S1, S2, and S3 developmental stages, respectively (Figure 3A–C). Intergroup comparisons showed that the differences between the XDS3 and HMS3 sample groups were the most significant, with a total of 3748 differentially expressed genes detected (including 1999 upregulated genes and 1749 downregulated genes), followed by the XDS3 and LTS3 sample groups, with 3744 differentially expressed genes identified (including 2323 upregulated genes and 1421 downregulated genes) (Figure 2D). The KEGG pathway enrichment analysis of differentially expressed genes showed that, in the S1 developmental stage, differentially expressed genes were significantly enriched in the cytochrome P450 metabolic pathway, plant MAPK signaling pathway, and phenylpropanoid biosynthesis pathway (Figure 3D); at the S2 stage, they were primarily enriched in the cytochrome P450 metabolic pathway and the phenylpropanoid biosynthesis pathway (Figure 3E); by the S3 stage, the plant MAPK signaling pathway and the phenylpropanoid biosynthesis pathway became the most significantly enriched pathways (Figure 3F).

2.8. Expression Patterns of Flavonoid Biosynthesis Genes in Different Cultivars

To further elucidate the expression patterns of key genes in the anthocyanin biosynthesis pathway during different developmental stages of three potato cultivars, this study analyzed the expression profiles of genes associated with this pathway (Figure 4, Table S7). In the phenylpropanoid pathway upstream of anthocyanin synthesis, 11 differentially expressed phenylalanine ammonia-lyase (PAL) genes were detected, all of which exhibited high expression levels in colored potato cultivars. Cinnamic acid is converted into 4-coumaroyl-CoA via catalysis by cinnamic acid 4-hydroxylase (C4H) and 4-coumaroyl-CoA ligase (4CL). One C4H gene and three 4CL genes showed significantly upregulated expression in colored potatoes. After entering the flavonoid synthesis pathway, two chalcone synthase genes (CHS) and one chalcone isomerase gene (CHI) associated with chalcone synthesis were all upregulated in colored potatoes. Concurrently, the high expression level of the flavanone 3-hydroxylase gene (F3H) in colored potatoes is presumed to promote the accumulation of dihydrokaempferol in their tubers. In the core pathway of anthocyanin biosynthesis, dihydroflavonol 4-reductase (DFR) reduces dihydrokaempferol, dihydroquercetin, and dihydromyricetin to colorless anthocyanins, which are subsequently catalyzed by anthocyanin synthase (ANS) to form colored anthocyanin aglycones. Flavonol synthase (FLS) converts dihydrokaempferol into kaempferol (a precursor diverging from anthocyanin biosynthesis). Analysis revealed that two DFR genes and the UDP-glucose flavonoid 3-O glucosyltransferase (UFGT) gene were highly expressed in colored potatoes. This upregulation is presumed to drive the increased accumulation of cyanidin, delphinidin, and pelargonidin in tubers. Additionally, the high expression level of the F3H, flavonoid 3′-hydroxylase (F3′H), and flavonoid 3′,5′-hydroxylase (F3′5′H) genes in colored potatoes likely contributes significantly to the red coloration of their tuber flesh. Among anthocyanin modification- and transport-related genes, six multidrug and toxin extrusion transporter (MATE) genes, seven glutathione S-transferase (GST) genes, one flavonoid 3′-O-methyltransferase (AOMT) gene, two O-O-methyltransferase (OMT) genes, and two ABC transporter genes were identified. Among these, the expression levels of the MATE-3, GST-3, and AOMT-1 genes were significantly higher in LT and HM than in XD. Specifically, MATE-3 (Soltu.DM.03G018250) exhibited expression ranges of 250–400 in LT and 170–580 in HM, and those of GST-3 (Soltu.DM.02G020850) ranged from 440 to 510 (LT) and 320 to 860 (HM), while AOMT-1 (Soltu.DM.09G025040) showed the most pronounced upregulation, with expression ranges reaching 1260–1441 in LT and 720–1750 in HM.

To validate the reliability of RNA-seq data, five genes (DFR, F3H, GST, F3′5′H, AOMT) associated with anthocyanin biosynthesis were selected for qRT-PCR analysis. The results showed that although the expression trends in the qRT-PCR results differed from those in RNA-seq data, the overall expression trends were consistent. This indicates that the qRT-PCR results are significantly correlated with RNA-seq data, demonstrating the reliability of RNA-seq data (Figure 5).

2.9. Analysis of Alternative Splicing in Potato Anthocyanin Biosynthesis-Related Genes and Transcription Factors

2.9.1. AS in Structural Genes

To clarify the role of AS in anthocyanin accumulation during potato growth and development, we first analyzed AS events in structural genes involved in anthocyanin biosynthesis (Table S8). The results revealed that 17 out of 88 structural genes exhibited alternative splicing, accounting for 19% of the total genes. Among these, F3H (Soltu.DM.02G023850), DFR-1 (Soltu.DM.02G024900), MATE-3 (Soltu.DM.03G018250), CHI (Soltu.DM.05G022280), CHS-2 (Soltu.DM.09G028560), and F3′5′H-1 (Soltu.DM.11G020990) each produced two transcripts. Expression trends showed lower levels in XD cultivars and increased expression in HM and LT cultivars. OMT1-2 (Soltu.DM.10G000070) also produces two transcripts with distinct expression patterns: Soltu.DM.10G000070.1 shows low expression in the XD cultivar but high expression in HM and LT, while Soltu.DM.10G000070.2 exhibits no significant differences across the three cultivars. For 4CL-3 (Soltu.DM.06G024540), Soltu.DM.06G024540.1 exhibited low expression in the LT cultivar and high expression in the XD and HM cultivars during the S1 stage. During the S3 stage, it showed low expression in the HM cultivar and high expression in the XD and LT cultivars. while no significant differences were observed among the three cultivars during the S2 stage. However, Soltu.DM.06G024540.2 exhibited the same expression trend across all cultivars. No obvious trends were observed for the remaining structural genes.

2.9.2. AS in Transcription Factors

Transcription factors (TFs) play a key regulatory role in gene expression regulatory networks. In this study, we identified a total of 403 differentially expressed TFs (log_2_FC > 2, FDR < 0.01), belonging to 51 different families. Among these, the most abundant family was the bHLH family (39 members), followed by the AP2/ERF-ERF family (38 members), the MYB family (33 members), and the C2H2 and WRKY families, which were tied for fourth place (23 members each) (Figure 6A). Further analysis revealed that the FPKM values of most transcription factors were higher in LT and HM tuber flesh. Notably, the NF-YB and SBP families exhibited high expression levels in XL at all three developmental stages while showing low expression levels in the tuber flesh of both LT and HM; HB-other TFs were specifically highly expressed in LT tuber flesh across all three stages while showing low expression in both the XD and HM cultivars; C2C2-GATA, C2C2-LSD, HB-HD-ZIP, HB-KNOX, and MADS-MIKC TFs maintained high expression levels in the tuber flesh of HM. Additionally, SRS and Tify TFs were highly expressed only in tuber flesh tissue at the LTS2 stage, while GRAS and WRKY TFs peaked at the S1 stage in XD tuber flesh. Notably, MYB TFs maintained stable high expression levels in the tuber flesh of the LT and HM cultivars across all three stages (Figure 6B).

To further explore the contribution of AS to anthocyanin regulation, we analyzed AS patterns among TFs. The results revealed that 33 out of 360 transcription factors belonging to 51 families exhibited alternative splicing, with 29 of these being differentially expressed (FPKM > 1, log_2_FC > 2). Several TF families showed significant expression divergence among alternative transcripts (Table S9). The AP2/ERF-ERF family member Soltu.DM.03G037290 produced three transcripts with distinct expression patterns: Soltu.DM.03G037290.2 showed higher expression in the HM cultivar, and Soltu.DM.03G037290.4 showed progressively increasing expression in the S3 stage across cultivars XD, LT, and HM, while Soltu.DM.03G037290.5 exhibited rising expression levels in the order XD, LT, HM across the three stages. The B3 family Soltu.DM.04G010530 (StB3-like) produced three transcripts: Soltu.DM.04G010530.1 exhibited low expression in the XD cultivar but significantly increased in the LT and HM cultivars, and Soltu.DM.04G010530.2 and Soltu.DM.04G010530.3 exhibited the highest expression in the XD cultivar, followed by the HM cultivar, and the lowest in the LT cultivar. Within the HB-HD-ZIP family, Soltu.DM.06G018130 comprises two transcripts. Soltu.DM.06G018130.1 exhibited divergent expression trends across developmental stages in the three cultivars: expression increased in the order LT, XD, HM during the S1 and S2 stages but reversed to XD, LT, HM during the S3 stage. Conversely, Soltu.DM.06G018130.2 exhibited higher expression levels in the LT cultivar during the S1 stage, with lower expression in the XD and HM cultivars. During the S2 and S3 stages, its expression increased in the XD cultivar while decreasing in the other two cultivars (Figure 6C). These results indicate that alternative splicing-generated transcripts regulate anthocyanin accumulation across different cultivars and developmental stages.

2.10. Different Transcripts Result in Varying Anthocyanin Content in Tobacco Leaves

The StB3-like gene was selected for further functional validation, and StB3-like-1 represents the full-length transcript of StB3-like, which is predicted to encode 246 amino acids. The first exon of StB3-like-1 undergoes alternative splicing to form StB3-like-2, encoding 234 amino acids. StB3-like-2 then undergoes exon-specific splicing to generate StB3-like-3, which encodes 229 amino acids (Figure 7A). Previous research has confirmed that StAN1 (Soltu.DM.10G020850), as a positive regulator of anthocyanin biosynthesis, significantly promoted anthocyanin accumulation when overexpressed in tobacco [50]. To validate the expression levels of the selected transcripts, qRT-PCR analysis was performed to determine the expression levels of the three StB3-like transcripts in the cultivars XD, LT, and HM. The results confirmed that expression trends aligned closely with FPKM values, supporting the reliability of the identified transcripts (Figure 7B). To investigate the functional specificity of each StB3-like transcript, transient overexpression assays were conducted in tobacco (variety Nc89). First, we individually injected StB3-like-1, StB3-like-2, and StB3-like-3 into tobacco plants and found that they did not affect anthocyanin accumulation (Figure S7). Four combinations were tested, including StAN1 + EV, StAN1 + StB3-like-1, StAN1 + StB3-like-2, and StAN1 + StB3-like-3 coexpression combinations. StAN1 + EV was injected into one side of the tobacco leaf as a control, while StAN1 and each coexpression combination were injected into the other side (Figure 7C). After culturing the injected bacterial solution for 3~5 days, it was observed that compared to the StAN1 + EV treatment group, the leaf color of the StAN1 + StB3-like-1 combination showed visibly lighter pigmentation compared with the control, while the leaf color of the StAN1 + StB3-like-2 and StAN1 + StB3-like-3 combinations significantly darkened. A quantitative analysis of anthocyanins confirmed that leaves injected with the StAN1 + StB3-like-2 and StAN1 + StB3-like-3 combinations exhibited significantly higher anthocyanin glycoside content compared to leaves injected with the StAN1 + StB3-like-1 combination (Figure 7D). These results indicate that StB3-like alternative transcripts individually have no effect on anthocyanin biosynthesis. However, when co-injected with AN1, StB3-like-2 and StB3-like-3 function as positive regulators, while StB3-like-1 appears to exert only a weak or neutral influence.

To further characterize the anthocyanin profiles associated with each transcript, we performed a targeted quantification of anthocyanin components in the infiltrated tobacco leaves using UPLC-MS/MS. Orthogonal partial least squares discriminant analysis (OPLS-DA) confirmed high data quality and reproducibility across biological replicates (Figure S8). A total of eight anthocyanin classes were detected, with cyanidin and peonidin derivatives being the most abundant, while delphinidin, pelargonidin, and other classes were present at relatively lower levels (Figure 8A). Following co-injection with StAN1, comparative analysis revealed distinct differential metabolite profiles for each StB3-like transcript compared to the EV control (Figure 8B). Specifically, five differential metabolites were identified for both StB3-like-1 and StB3-like-3, whereas only three were altered in StB3-like-2. Notably, Peonidin-3-O-sophoroside-5-O-glucoside was a common differential metabolite across all three transcripts. In summary, the three StB3-like transcripts differentially regulate anthocyanin metabolism in tobacco by distinctly influencing the biosynthesis and accumulation of specific anthocyanin compounds.

To investigate the effects of different transcripts on anthocyanin synthesis, we analyzed changes in anthocyanin metabolite levels (Figure 9). Among the detected metabolites, Delphinidin-3-O-rutinoside, Cyanidin-3-O-glucoside, Pelargonidin-3-O-coumaroyl-5-O-galactoside, Cyanidin-3-[6″-(rhamnosyl) glucoside, and Peonidin-3-O-(caffeoyl) rhamnoside exceeded 100 μg/g. Among these, Delphinidin-3-O-rutinoside, Cyanidin-3-O-glucoside, and Peonidin-3-O-(caffeoyl) rhamnoside exhibited the highest concentrations in StB3-like-3, followed by StB3-like-2, while StB3-like-1 showed the lowest levels. The Cyanidin-3-[6″-(rhamnosyl) glucoside content in StB3-like-2 and StB3-like-3 reached 1.2-fold and 1.3-fold that of StB3-like-1, respectively. Two metabolites (Cyanidin-3-O-sambubioside and Cyanidin-3-xylosyl-glucoside) showed high expression levels in StB3-like-3, with concentrations ranging from 16 to 30 μg/g. Eleven metabolites ranged in concentration from 1 to 10 μg/g: eight of these (including Delphinidin-3-O-glucoside, Delphinidin-3-O-galactoside, Cyanidin-3-O-sophoroside, Petunidin-3-O-(6-O-p-coumaroyl)-glucoside, Peonidin-3-O-glucoside, Pelargonidin-3-O-glucoside, Peonidin-3,5-O-diglucoside, Delphinidin-3-O-sambubioside) exhibited the highest content in StB3-like-3; Malvidin-3-O-rutinoside was highly expressed in both StB3-like-2 and StB3-like-3, whereas Cyanidin-3-O-(6″-O-caffeoyl) rhamnoside showed the highest expression level in StB3-like-1. Twenty-two metabolites had concentrations below 1 μg/g, with one-third exhibiting relatively high levels in StB3-like-3.

3. Discussion

Colorful potatoes are widely cultivated due to their exceptionally high nutritional value. Investigating the accumulation mechanism of anthocyanins in potatoes is of significant importance for the practical application of colorful potatoes in production and daily life. Currently, existing studies utilize SMRT sequencing combined with RNA-Seq data analysis to investigate the regulatory gene network governing the biosynthesis of anthocyanin compounds in potatoes. Potato tubers serve as crucial organs for nutrient absorption, storage, and transport in plants, playing an indispensable role in plant growth, development, and stress responses. Under potato heat stress conditions, SMRT sequencing yielded 245,298 FLNC reads and 58,747 high-quality full-length transcripts [51]. Through PacBio sequencing, a total of 11,665 complete open reading frames were predicted in potato seedlings [52]. Based on SMRT sequencing, 5955 long noncoding RNAs and 700 AS events were identified in Populus plants [53]. In this study, the PacBio sequencing of white and colored potato varieties yielded over 40 GB of clean data and 178,073 FLNC reads, enabling the comprehensive identification of 42,065 transcripts, 36,002 complete CDSs, 2061 TFs, 1544 lncRNAs, and extensive structural features including SSRs and poly-A sites. Among them, lncRNAs may regulate anthocyanin biosynthesis through cis/trans-acting mechanisms, while SSRs can serve as molecular markers linked to anthocyanin-related genes, providing valuable genomic resources for further exploring their functional implications in potato anthocyanin metabolism.

This study integrated RNA-seq data from three developmental stages of different potato cultivars with the FL transcriptome of this study to investigate the DEGs and their enriched pathways associated with anthocyanin accumulation in potato tubers. The results showed that the number of DEGs was higher in the XDS3 vs. HMS3 and XDS3 vs. LTS3 comparison groups compared to the first and second stages. During potato development, biological pathways are primarily enriched in the cytochrome P450 pathway, MAPK signaling pathway—plant, phenylpropanoid biosynthesis pathway, DNA replication protein pathway, DNA replication pathway, and transporter pathway. Relevant studies have confirmed that the phenylpropanoid biosynthesis pathway participates in the accumulation process of anthocyanins [54,55]. Furthermore, the enrichment in the transporter pathway aligns with the conclusion that “ATP-binding cassette transporters are key regulators of anthocyanin accumulation in potato tubers” [56]. The cytochrome P450 pathway, meanwhile, is closely associated with the modification of secondary metabolites during anthocyanin synthesis [57]. Pathway enrichment patterns varied across developmental stages. Stage S1 showed enrichment in the cytochrome P450 pathway, MAPK signaling pathway—plant pathway, phenylpropanoid biosynthesis pathway, DNA replication protein pathway, and DNA replication pathway. Stage S2 showed enrichment only in the cytochrome P450 pathway, phenylpropanoid biosynthesis pathway, DNA replication protein pathway, and DNA replication pathway. Stage S3 demonstrated enrichment in all six pathways. This likely reflects the coordinated regulatory mechanism governing anthocyanin synthesis and transport during tuber maturation [58].

AS enhances the coding capacity of the genome by generating multiple transcripts through the utilization of multiple splicing sites within genes. Currently, research on AS in plants primarily focuses on areas such as salt stress [59], light signal transduction [60], fruit development [61], high-temperature stress, and drought stress [62]. Studies have shown that mutations in the 5′ splicing site of the second intron of the DFR (dihydroflavonol-4-reductase) gene in Solanaceae plants affect anthocyanin accumulation [63]. In the Sichuan pepper varieties “Rongchang Wucai” (WC) and “Zhuye” (ZY), there are 143,122 unique coding sequences, 105,465 SSRs, 20,145 TFs, and 4719 lncRNAs, with 142,829 transcripts successfully annotated [64]. In grape berries, the AS variant VvMYBA1-L of the VvMYBA1 gene causes the premature termination of the encoded protein; the overexpression of VvMYBA1-L delays the accumulation of flesh color, while the overexpression of VvMYBA1 promotes the accumulation of flesh color [65]. In this study, 12,524 AS events were identified in F01, and 14,287 AS events were identified in F02. After screening for differentially expressed transcripts from the transcripts obtained from the alternative splicing analysis, three groups of genes were found to have corresponding transcripts. These differentially expressed genes associated with anthocyanin synthesis via AS events warrant further investigation in future studies.

Transcription factors, as key components of gene regulatory networks, play a central role in plants’ responses to both biotic and abiotic stresses. During their functional processes, transcription factors undergo post-transcriptional modifications, with alternative splicing mechanisms being one of the most extensively studied. In soybean, the overexpression of GmNTL9.m1 resulted in wilting and chlorotic symptoms under sodium bicarbonate stress, whereas the overexpression of GmNTL9.m2 conferred significant tolerance [66]. In tea plants, the full-length CsJAZ1-1 and variant CsJAZ1-2 interact with CsMYC2 to inhibit anthocyanin biosynthesis, while CsJAZ1-3 relieves this inhibition through a competitive mechanism [67]. In rapeseed, BnaPAP2.A7-744 positively regulates purple leaf pigmentation, whereas BnaPAP2.A7-910 and BnaPAP2.A7-395 have no effect on it [68]. The AS of CiFD in citrus forms two splicing variants (CiFDα and CiFDβ); CiFDβ directly binds to the promoter of CiAP1 to promote flowering, while CiFDα exerts its function by forming an FAC complex that binds to the CiAP1 promoter. The B3 transcription factor family comprises four subfamilies: LAV, RAV, ARE, and REM [69]. Strawberry FaRAV1 promotes anthocyanin accumulation by activating the FaMYB10 promoter and structural genes involved in anthocyanin biosynthesis [70]. Apple MdARF13 inhibits anthocyanin accumulation by directly binding to the MdDFR promoter [71]. The Arabidopsis REM subfamily gene VRN1 (vernalization 1) participates in regulating plant flowering [72,73]. In this study, we focused on investigating the roles of different transcripts of the B3 family transcription factor StB3-like in anthocyanin biosynthesis. The results showed that the co-injection of StB3-like-2 and StB3-like-3 with StAN1 markedly promoted anthocyanin accumulation compared with the control, whereas StB3-like-1 exerted no apparent effect. Metabolomic analysis further revealed that different transcripts specifically regulate the accumulation of multiple anthocyanin-related metabolites. For instance, StB3-like-3 specifically accumulates compounds such as Petunidin-3-O-(6-O-p-coumaroyl)-glucoside. Although some transcripts exhibited primer cross-reactivity in qRT-PCR detection, the distinct metabolite profiles clearly support StB3-like-2 and StB3-like-3 as independent transcripts. These findings suggest that different StB3-like transcripts may participate in regulating the anthocyanin biosynthesis pathway by specifically controlling metabolite accumulation.

Beyond the B3 family, we also identified AS events in transcription factors such as bHLH, MYB, and AP2/ERF-ERF. For example, the AS of CmbHLH2 in chrysanthemum generates two variants: only the full-length CmbHLH2^Full^ interacts with CmMYB6 to promote anthocyanin biosynthesis, while the CmbHLH2^Short^ variant is non-functional [74]. The AS of VaMYB108 in grape produces VaMYB108^normal^ and VaMYB108^short^, both induced by low temperature and capable of mutually upregulating each other’s expression [28]. However, whether the AS of these transcription factors affects anthocyanin accumulation in potato requires further verification.

Scope and Limitations

Functional validation was performed in tobacco, a heterologous system. Although the results provide strong evidence for the regulatory role of StB3-like transcripts, further potato-specific validation (e.g., stable transformation, gene silencing) is required to confirm their causal roles in tuber anthocyanin accumulation. Additionally, the mechanisms underlying AS regulation remain to be explored.

4. Materials and Methods

4.1. Preparation of Plant Materials

The potato clonal varieties used in this study included the yellow-skinned Xindaping, the red-skinned Lingtianhongmei, and the purple-skinned Heimeiren, all cultivated at the Dingxi Academy of Agricultural Sciences in Gansu Province, where the annual average temperature is 6.3 °C, the annual average sunshine duration is 2500.1 h, the soil type is yellow loam, and irrigation was maintained at 75–80% of field capacity (Figure 10). Potato tubers were harvested at 100 days after planting (S1, tuber development stage), 130 days (S2, tuber enlargement stage), and 150 days (S3, tuber harvest stage). An average of five tubers per plant were collected for each cultivar at each stage, with three biological replicates for all samples. The collected samples were immediately frozen in liquid nitrogen and then stored at −80 °C for subsequent use.

4.2. RNA Isolation, Library Preparation, and SMRT Sequencing

RNA was extracted from potato tubers using the TaKaRa Polysaccharide-Polyphenol RNA Extraction Kit (TaRaKa, Beijing, China). Its integrity and concentration were assessed via 1% agarose gel electrophoresis and Nanodrop ND-2000 (Thermo Fisher Scientific Inc., Shanghai, China), respectively. Subsequently, the SMARTer™ PCR cDNA Synthesis Kit (TaRaKa, Beijing, China) was used to synthesize full-length cDNA from mRNA. The full-length cDNA was then amplified by PCR (20 cycles). The full-length cDNA underwent end repair, followed by ligation with SMRT dumbbell-shaped adapters. Nucleic acid exoenzymes were used to digest the cDNA, resulting in a sequencing library. The obtained clean reads (Q20 > 90%) were mapped to the S.tuberosum reference genome (PGSC_DM_v6.1, http://solanaceae.plantbiology.msu.edu/pgsc_download.shtml, accessed on 16 November 2025) using HISAT2 (v2.0.1, https://daehwankimlab.github.io/hisat2/, accessed on 18 November 2025). DESeq2 was used for differential expression analysis between samples. Genes with log_2_FC > 1 and FDR < 0.05 were considered significantly differentially expressed genes (DEGs). DEGs involved in anthocyanin accumulation were screened based on KEGG functional annotation [75].

4.3. Identification of Novel Transcripts (SSR), Prediction of Long Noncoding RNAs (lncRNAs), and Sequence Annotation

New transcripts longer than 500 bp were screened and analyzed using MISA (http://pgrc.ipk-gatersleben.de/misa/, accessed on 20 November 2025) for SSRs [76]. MISA identifies seven types of SSRs by analyzing transcript sequences: single-base repeats, double-base repeats, triple-base repeats, quadruple-base repeats, quintuple-base repeats, sextuple-base repeats, and mixed SSRs.

TransDecoder (v.3.0.0) identifies reliable potential coding regions from transcript sequences based on open reading frame (ORF) length, log-likelihood score, and the alignment of amino acid sequences with protein domains in the Pfam database [77]. Four encoding potential analysis methods—CPC analysis, CNCI analysis, Pfam protein domain analysis, and CPAT analysis—were used to predict lncRNAs in newly discovered transcripts [78]. Astalavista (v.3.2) was used to identify five types of alternative splicing in the samples: ES, A3, A5, IR, and MEE.

The sequences of the newly identified transcripts were aligned with the NR, Swiss-Prot, GO, COG, KOG, Pfam, and KEGG databases using BLAST (v.2.2.26), resulting in the annotation of transcripts. Transcription factor prediction was performed using iTAK (v.1.5).

4.4. Quantitative Real-Time PCR (qRT-PCR) Analysis

RNA was extracted from the samples using an RNA extraction kit (Cat. No. 9769, TaKaRa, Beijing, China). The RNA samples were reverse-transcribed into cDNA using the FastKing gDNA Dispelling RT SuperMix kit (TianGen, Beijing, China). The SuperReal PreMix Plus kit (TianGen, Beijing, China) was used according to the manufacturer’s instructions, followed by qRT-PCR on the Bio-Rad CFX 96 system (BioRad Laboratories, Hercules, CA, USA). The StEF-1α (AB061263) gene was used as an internal control. Each assay was performed in triplicate, and gene expression levels were analyzed using the 2^−ΔΔCt^ method [79]. The specific primers used in this experiment are detailed in Table S10.

4.5. Functional Validation of Candidate Genes via Transient Expression in Tobacco

In this study, the three transcripts of Soltu.DM.04G010530 were then cloned into the pNC-Green-SK vector and transformed into the Agrobacterium strain GV3101. For the transient coloration experiment, Nc89 was used as the experimental material. After 3–5 days of injection, the coloration of the tobacco leaves was observed, and samples were collected for subsequent experiments.

4.6. Anthocyanin Content Determination

The anthocyanin content in tobacco was determined based on the structural properties of anthocyanins under different pH conditions. Anthocyanins exhibit a maximum absorption peak at 530 nm when pH = 1, while at pH = 4.5, anthocyanins convert to colorless chalcones, resulting in no absorption peak at 530 nm. The determination was conducted as follows: Tobacco leaves were ground into powder. A total of 0.1 g of the sample was weighed and mixed with 1 mL of extraction solution. After thorough homogenization, it was transferred to an EP tube. The extraction solution was diluted to 1 mL. The solution was extracted at 60 °C for 30 min, then centrifuged at 12,000 rpm for 10 min at room temperature. The supernatant was aspirated for analysis [80]. The anthocyanin content in the sample was determined by measuring the absorbance values at 530 nm and 700 nm under different pH conditions (BC1385, Solarbio, Beijing, China).

4.7. Metabolite Analysis

Metware Biotechnology (Wuhan, China) performed the metabolomic analysis for this study. Tobacco leaves were freeze-dried using a vacuum freeze dryer (Scientz-100F, Ningbo, China) and ground into powder using a ball mill (MM 400, Retsch, Haan, North Rhine-Westphalia, Germany). A total of 50 mg of each sample was weighed and dissolved in 500 μL of 50% methanol aqueous solution (containing 0.1% hydrochloric acid). Each sample was weighed three times as biological replicates. The mixture was vortexed for 5 min, sonicated for 5 min, and centrifuged at 4 °C and 12,000 rpm for 3 min. The supernatant was aspirated, and the procedure was repeated once. The two supernatants were combined and filtered using a 0.22 μm microporous membrane (Millipore, Billerica, MA, USA) for filtration. Metabolites were identified using ultra-performance liquid chromatography (UPLC, ExionLC™ AD, SCIEX, Framingham, MA, USA) coupled with tandem mass spectrometry (MS/MS, QTRAP^®^ 6500+, SCIEX, Framingham, MA, USA) (UPLC-MS/MS).

Orthogonal partial least squares discriminant analysis (OPLS-DA) was performed on identified metabolites using R (v4.2.2). Significant differences were determined in metabolites between groups based on an absolute log_2_FC (fold change) > 2 and variable importance in projection (VIP) ≥ 1 [81].

4.8. Statistical Analysis

All experiments were performed in triplicate, with the results presented as the mean ± standard error (SE). Data analysis was conducted using SPSS (v.26.0) for a one-way analysis of variance (ANOVA) and Duncan’s multiple range test with p < 0.05.

5. Conclusions

This study aimed to elucidate the molecular mechanisms underlying pigment biosynthesis in potato tubers by integrating second- and third-generation transcriptomic sequencing technologies across different developmental stages. Transcriptomic analysis revealed abundant AS events and numerous novel transcripts, highlighting the complexity of gene regulation in potatoes. Notably, the StB3-like gene family exhibits distinct functional differentiation. Transient expression experiments confirmed that only StB3-like-2 and StB3-like-3 (but not StB3-like-1) significantly enhance anthocyanin accumulation when coexpressed with StAN1, validating their specific regulatory roles in pigment deposition. Collectively, this study deepens our understanding of transcriptional mechanisms linking disease resistance and pigmentation, providing valuable molecular markers and functional candidate genes for breeding nutritionally rich, visually appealing, and disease-resistant potato.