Genome Wide Identification of Terpenoid Metabolism Pathway Genes in Chili and Screening of Key Regulatory Genes for Fruit Terpenoid Aroma Components

Abstract

Aroma is an important processing and consumption quality trait of fruits and vegetables, and terpenes produced from the terpenoid metabolic pathway are a critical component of chili fruit flavor. This pathway involves the participation of at least eighteen enzymes, such as AACT, HMGS, HMGR, MVK, PMK, MVD, FPPS, GGPPS, DXS, DXR, MCT, CMK, MECPS, HDS, HDR, GPPS, IDI, and TPS. In this study, the genome wide information, expression characteristics, and relationship with terpenoids of terpenoid pathway genes are analyzed in C. annuum. The results showed that C. annuum has sixty-seven genes related to terpene metabolic pathways. Non-targeted metabolomics studies found that the content of aromatic terpenoids α-calacorene, α-cubene, and cis-β-farnesene increased with fruit development in HDL fruits, while linalool and nerolidol were much higher in GLD608. Correlation analyses between qRT-PCR and metabolome data showed that the expression levels of CaHMGS-3, CaMVD-1, CaCMK-1, and CaGGPPS-2 were positively correlated with the content of linalool, a flavor monoterpene alcohol. CaMECPS-1 was positively correlated with cis-β-farnesene, and there was also a significant positive regulatory relationship between CaTPS-5 and nerolidol relationship. In conclusion, the present study provides genetic resources for further studies on the gene regulatory mechanisms of flavor synthesis and terpenoid metabolic pathways in chili.

1. Introduction

Chili is an annual or perennial vegetable of the Solanaceae family, known worldwide for its wide variety, with high nutritional quality, and unique spicy flavor. Currently, thirty-six species of Capsicum have been identified, of which five species, namely C. annuum, C. chinense, C. frutescens, C. baccatum, and C. pubescens, have undergone domestication and are now widely cultivated [1]. Chili is widely used for fresh consumption, providing essential nutrients such as vitamins C and E, minerals, and carotenoids to people. Additionally, its unique aroma and spiciness make it ideal for the development of condiments and functional products [2]. The sensory appeal of volatile aroma components (VACs) in chili is also a critical attribute for assessing the quality of chili fruits [3].

So far, at least three pathways for the biosynthesis of VACs have been found in plants, including phenylpropanoid, alkaloid, and terpene metabolic pathways [4]. The terpene metabolism pathway, also known as the isoprene metabolism pathway, includes the mevalonate pathway (MVA pathway) in eukaryotic cytoplasm and the methylerythritol 4-phosphate pathway (MEP pathway) in plant chloroplasts or other plastids. In the MVA pathway, acetyl CoA produced by the breakdown of sugars and fatty acids are condensed into acetyl CoA carboxylase (AcAc CoA) under the action of acetyl CoA C-acetyl transferase (AACT). AcAc CoA is catalyzed by 3-hydroxy-3-methylglutaryl-CoA synthase (HMGS) to form HMG CoA, which is then reacted by 3-hydroxy-3-methylglutaryl-CoA reductase (HMGR) to form mevalonate. Mevalonate is further phosphorylated and decarboxylated under the action of mevalonate kinase (MVK), 5-phosphate mevalonate enzyme (MPK), and mevalonate pyrophosphate decarboxylase (MVD), forming isopentenyl diphosphate (IPP). Under the catalysis of isopentenyl diphosphate isomerase (IDI), IPP isomerizes to form dimethylallyl diphosphate (DMAPP). IPP and DMAPP undergo a series of addition reactions under the action of geranylgeranyl pyrophosphate synthase (GGPPS) and farnesyl pyrophosphate synthase (FPPS) to form geranylgeranyl diphosphate (GGPP) and farnesyl diphosphate (FPP), respectively, and further produce diterpenes and sesquiterpenes under terpene synthase (TPS) catalysis. In the MEP pathway, 1-deoxy-D-xylulose-5-phosphate synthase (DXS) catalyzes the condensation of glyceraldehyde-3-phosphate dehydrogenase (G3P) produced by photosynthesis with pyruvic acid to form 1-deoxy-D/xylulose-5-phosphate (DXP), which is reduced to MEP under the action of 1-deoxy-D-xylulose-5-phosphate reductase (DXR). MEP successively produces 4-hydroxy-3-methyl-2-butenyl pyrophosphate (HMBPP) under the catalysis of MEP cytidine transferase (MCT), 4-diphosphocytidyl-2-C-methyl-D-erythritol kinase (CMK), 2C-methyl-D-erythroid alcohol 2,4-dichlorophosphate synthase (MDS), and hydroxymethylbutynyl diphosphate synthase (HDS), which further catalyzes the formation of IPP through 4-hydroxy-3-methylbut-2-enyl diphosphate reductase (HDR). IPP and its isomerized product DMAPP further form aromatic monoterpenes, diterpenes, and tetraterpenes under the action of GGPPS, geranyl pyrophosphate synthase (GPPS), and TPS (Figure 1). It is worth noting that the MVA and MEP pathways are connected by IPP, which can move between the cytoplasm and the plastids. Retaining the metabolism of the two IPP/DMAP pathways is believed to better control the compartment-specific isoprene pool, allowing it to transform into MEP-derived monoterpenes and diterpenes, carotenoids, plastids containing quinones, and chlorophyll, as well as MVA-derived sesquiterpenes, sterols, brassinosteroids, and triterpenes [5].

In recent years, genes encoding enzymes of the MVA and MEP pathways, especially those with important regulatory roles, have been the subject of considerable research. As the first enzyme in the MVA pathway, the overexpression of GlAACT in ganoderma lucidum led to an increase in triterpenoid content [6], indicating a positive correlation between GlAACT expression and triterpenoid content. HMGR is an endoplasmic reticulum membrane binding enzyme and is widely recognized as the rate-limiting enzyme of the MVP pathway. HMGR is encoded by a small gene family, and although some isomers engage in the synthesis of isoprenoids, the roles of other isomers are still unclear [7]. MVK is the first ATP-dependent enzyme in the MVP pathway and is one of the rate-limiting enzymes that control the entire metabolic pathway. It is highly expressed in Arabidopsis and promotes a significant increase in the content of intermediate metabolites such as dienes in triterpenoid synthesis [8]. The PMK gene in the fruit of the silenced Mangifera indica showed meaningful changes in terpene metabolites, with a decrease or disappearance of β-caryophyllene, β-pinene, bisabolene, and β-guaiene. However, d-limonene was newly observed in the silenced fruit. Due to its role in the formation of sesquiterpenes and triterpenes, research on FPPS has been widely studied. For example, when cotton AaFPPS cDNA was transferred to Artemisia annua, it was found that the content of sesquiterpene lactone artemisinin in transgenic plants increased by 2 to 5 times [9,10]. Related results also appeared in Rosa rugosa overexpressing RrFPPS [11].

DXS and DXR are considered important rate-limiting enzymes in the MEP pathway. Transient overexpression of AcDXS1 leads to a significant increase in monoterpenes in Nicotiana tabacum leaves [12]. Silencing of tomato SIDXS leads to a decrease in sesquiterpene content, and the overexpression of DXR increases paclitaxel synthesis in Arabidopsis, all indicating a positive correlation between DXS and DXR expression levels and terpenes. IDI, GGPPS, and TPS are three enzymes that appear in the cytoplasm and plastids and affect the synthesis of terpenoids. GGPPS can catalyze the formation of GGPP from three molecules of IPP and DMAPP, thereby affecting the content of diterpenes. The decrease in AtGGPPS11 expression in Arabidopsis not only leads to pale phenotype and developmental delay but also accompanies a decrease in the content of isoprenoid substances derived from plastid GGPP [13]. Research has found that the expression patterns of two TPS genes in lilium brownie var. viridulum baker are significantly positively correlated with the release of ocimene and myrcene [14], the overexpression of CpTPS4, 6, 7, 13, 14, 15, and 16 confirmed obvious upregulation of terpene content in Chimonanthus praecox L. [15], while in Freesia × hybrida, FhTPS1 is responsible for the formation of linalool [16]. Compared with the control group, transient convey of SoTPS2 and SoTPS3 in the petals of Antirrhinum majus increased the release of ocimene by 10.91 times and 23.67 times. However, GGPPS and TPS belong to a family of proteins, and more expression characteristics and their relationship with terpenes still need to be clarified. By overexpressing VvTPS59 in V. quinquangularis, a significant increase in the levels of linalool, linalool oxide, geraniol, and citronellol was found, suggesting that TPS59 plays a role in the synthesis of linalool and its derivatives in Vitis vinifera [17]. Jiang showed [18] that the overexpression of LiMCT in Arabidopsis thaliana affected the expression levels of genes of the MEP and MVA pathways, suggesting that the overexpression of LiMCT in Arabidopsis thaliana affects the metabolic flow of the C5 precursor of two different terpene synthesis pathways. The expression of the monoterpene synthase AtTPS14 was nearly four-fold higher in transgenic Arabidopsis thaliana flowers compared with the control, suggesting that LiMCT plays an important role in regulating monoterpene synthesis and synthesis of other isoprenoid-like precursors in transgenic Arabidopsis thaliana flowers.

Many terpenoids are also important components of chili fruit aroma. An analysis of the components of red pepper fruit revealed that terpenoids, including carvacene, (4aS-cis)-2,4a,5,6,7,8,9,9a-octahydro-3,5,5-trimethyl-9-methylene-1H-benzylidene, and cis-(-)-2, 4a,5,6,9a-hexahydro-3,5,5,9-tetramethylene-1H-benzocycloheptene, were the main aroma components [19]. HS-SPME-GC-MS analysis of volatile aroma components in the fruits of four different chili peppers at different developmental stages showed that terpenes such as nerolidol, linalool, α-murolene, ar-himachalene, and cis-α-bergamot were higher in content. Among them, the contents of nerolidol and linalool are higher in the fruits of C. annuum, while the contents of α-muurolene, ar-himachalene, and cis-α-bergamotene were higher in the fruits of C. chinense, showing significant inconsistency [20].

However, there is little research on the genes and regulatory mechanisms involved in the formation of terpenes in chili fruits, which hinders the revelation of differential mechanisms in terpenoid synthesis and molecular breeding for chili fragrance. Early transcriptome sequencing combined with untargeted metabolomics analysis revealed three genes encoding DXS, three genes encoding HMGS, and sixteen genes encoding TPS, which may be related to the formation of terpenoids. The aim of this study is to further analyze the bioinformatics characteristics of terpenoid pathway genes within the genome, their expression levels during pepper fruit development, and their relationship with the content of major terpenoids, laying the foundation for discovering key genes and their applications.

2. Materials and Methods

2.1. Plant Material and Treatment

In this study, we selected a total of three chili materials, namely Guizhou ChaotianjiaoD608 (GLD608, Capsicum annuum L., an annual dried pepper variety), Guijiao (GJ, Capsicum chinense L., a highly hot variety originating from India), and Hainan huangdenglong (HDL, Capsicum chinense L., local varieties unique to Hainan Province, China). All three materials were preserved by the School of Breeding and Propagation, Hainan University (Sanya Institute of Breeding and Propagation). High-quality seeds were selected, germinated at 28 °C, spot-sown in 32-well seedling trays with substrate (triton/vermiculite/perlite = 2:1:1), and cultured under 16/h of light/8 h of darkness and 28 °C of light/22 °C of darkness, and then transplanted into 5L pots when they were cultivated to about 6 true leaves. From December 2023 to February 2024, fruits of three pepper varieties were collected at the green, breaking, and maturation stages based on fruit size and color. Thirty fruits were randomly selected from each stage and divided into three biological replicates, totaling 26 fruit samples. Each fruit sample was homogenized by a mixing method, quickly frozen in liquid nitrogen, and stored in an ultra-low temperature refrigerator at −80 °C for subsequent transcriptome and metabolomics analysis.

2.2. Selection of Genes Related to Biosynthesis of Terpene Metabolic Pathway

The C. annuum genome sequence and annotation files (GCF-000710875.1) were obtained from NCBI. The Arabidopsis thaliana L.-related enzyme protein sequences [21] were downloaded from NCBI (https://www.ncbi.nlm.nih.gov/, accessed on 11 June 2024) and searched in the C. annuum genome with Blast, the filtering condition of E-value < 0.001, then the preliminary terpenoid metabolic pathway family members were obtained. Protein sequences were obtained from these candidate gene IDs and submitted to sites such as SMART (http://smart.embl-heidelberg.de/, accessed on 11 June 2024) and PFAM (https://pfam.xfam.org/search, accessed on 11 June 2024), for verification, eliminating sequences without complete structural domains.

2.3. Sequence Analysis of Genes Related to Terpene Metabolic Pathway Biosynthesis

Protein sequences and CDS sequences of genes were extracted from C. annuum genome sequence. The physical and chemical properties of related gene proteins were predicted by using Protpara4.0 software. We used Protcomp 9.0 (https://web.expasy.org/protparam/, accessed on 11 June 2024) to predict the subcellular localization of genes. Protpara 4.0 software and Protcomp 9.0 selects default parameters.

2.4. Chromosome Mapping, Phylogenetic Tree Analysis, Intron–Exon Analysis, and Promoter Analysis of Related Genes

We used Tbtools-IIo visualize the position of genes on chromosomes. Tbtools was used to blast the protein sequences of related genes, and the number of introns and exons was combined to determine gene duplication events. MeMe 5.5.8 software was used to analyze the conserved motifs of related genes, and the motif number was set to 25. In Mega X software, maximum likelihood was used to construct phylogenetic trees. The 2000 bp region upstream of ATG was selected as the promoter region. Cis-regulatory elements in the promoter region were predicted using PlantCare (http://bioinformatics.psb.ugent, accessed on 11 June 2024), and the predictions were displayed using Tbtools. Tbtools, MeMe, PlantCare select default parameters, “Model” in ML phylogeny select LG model, “Rates among Sites” select Gamma Distribute.

2.5. GC-MS Experiments and Transcriptome Experiments

The expression characteristics of terpene metabolism-related genes in GLD608, GJ, and HDL were analyzed by RNA-seq technology and non-targeted metabolome technology during fruit green stage, color transition stage, and red stage. The heatmap was made by Tbtools. In the RNA-seq and GC-MS experiment analyses, 10 plants of each chili variety were planted with the same cultivation and management practices except for varietal differences. Chili fruits of uniform color, similar size, uniform maturity, and free of pests, diseases and mechanical damage were harvested from 5 to 10 chili plants of similar growth according to fruit size and fruit color, respectively, and 30 fruits of similar size were randomly selected in each growth period, for a total of 36 fruit samples (4 varieties × 3 developmental periods × 3 biological replicates).

2.5.1. Sample Preparation and Extraction

Shanghai OE Biotech Co., Ltd. completed untargeted metabolome sequencing. HS-SPME extracted VACs from pepper fruit samples: fruit samples were frozen fresh fruit and were immediately powdered with a grinder (MM400,Retsch, Berlin, Germany), then defrosted at room temperature prior to analysis. Accurately weigh the 1 g of sample powder and add to an ethanol solution containing 20 μg/mL n-alkanes (C7–C40) as an internal standard (configuration method: 980 μL ethanol solution (99.7%) + 20 μL n-alkanes mother liquor (1 mg/mL)) and then transfer to a sealed headspace injection bottle (20 mL, Agilent, Santa Clara, CA, USA) for the release of volatiles. The headspace bottle was equilibrated at a temperature of 60 °C and a shaking speed of 450 rpm for 10 min. Then, the extraction head (50/30 μm DVB/CAR/PDMS, Sigma, Shanghai, China) was inserted into the headspace part of the sample and extracted for 60 min.

2.5.2. GC-MS Analysis of VACs

Agilent 7890B chromatography and a 5977B mass spectrometer equipped with a DB-5MS capillary column (30 m × 0.25 mm × 0.25 μm, Agilent J&W Scientific, Folsom, CA, USA) were used to carry out the qualitative and quantitative analysis of volatile aromatic compounds. The extracted sample was directly injected into the injection port of the gas chromatograph–mass spectrometer, desorbed at 250 °C for 5 min, and the injection port temperature was set at 230 °C. High-purity helium (Purity ≥ 99.999%) was used as the carrier gas. The initial oven temperature was set at 40 °C for 1 min, increased to 230 °C at 4 °C/min and held for 1.5 min, then increased to 250 °C at a rate of 10 °C/min and kept for 2 min. Mass spectrometry was recorded in the ionization mode of electron impact ion source (EI) energy of 70 eV. The ion source temperature was set to 230 °C, and the quadrupole temperature was 150 °C. Mass spectrometry data were extracted by full scan mode (SCAN) with a mass scanning range of 40–500 m/z. In the process of mass spectrometry analysis, all samples were mixed as quality control (QC) samples to test the stability of the system mass spectrometry platform during the whole experiment. MS-DIAL 2.74 software (accessed on 3 August 2024) analyzed the GC-MS raw data for peak detection, peak recognition, MS2 Dec deconvolution, and peak alignment. They then compared with MS in the NIST database (https://webbook.nist.gov/chemistry/, accessed on 5 August 2024) to determine the species of VACs. VACs with MS matching scores greater than 80% were retained, and the final substance type was determined after manual identification and comparison. The relative content of volatiles was calculated by the standard internal method. Detailed information is provided in the Supplementary Information Table S1. The formula is as follows:

f[relative content of volatiles (μg/kg)] = peak area of target × 0.1 μg/(peak area of internal standard × 1000 g).

Retention index formula: RI = n × 100 + 100 × (Ti − Tn)/(Tn + 1 − Tn)

2.5.3. Screening of Differential VACs and Marker Aroma Compounds

Multivariate statistical analysis was used to screen the differential VACs in the samples. The relative content of VACs was used as the analysis data, and the PLS-DA model of VACs in different pepper fruit samples was constructed by (PLS-DA). The VACs in the model were screened by variable weight value (VIP) > 1, and a t-test was performed to verify its significance.

2.5.4. RNA Sequencing Analysis

The transcriptome sequencing and DEGs expression analysis were performed by Shanghai OE Biotech Co., Ltd. (Shanghai, China). Pre-process the raw data to remove low-quality reads containing Ploy-N. The obtained clean reads were mapped to the pepper genome Pepper Zunla_1_Ref_v1.0 (https://www.ncbi.nlm.nih.gov/data-hub/genome/GCF_000710875.1/ (accessed on 30 March 2022)) using the default parameter HISAT 2.2.1 software. Then, the transcript assembly was performed, and the transcript abundance was quantified. The calculation formula of transcript abundance FPKM value is as follows:

f(FPKM) = cDNA Fragments/[Mapped Fragments (Millions) × Transcript Length (kb)]

2.6. RNA Extraction and qRT-PCR Analysis

According to the instructions of the RNA extraction kit (Vazyme, Nanjing, China), total RNA was isolated from fruit samples of GLD608, GJ, and HDL at different stages. The concentration of extracted RNA was measured using a spectrophotometer (Thermo Scientific, Wilmington, DE, USA), and total RNA was reverse transcribed into cDNA using the HiScript III 1st Stand cDNA synthesis kit (Vazyme, Nanjing, China). Design primer sequences using Premier 5.0 software. For specific details primer sequences, see Table S2. Using actin (accession number: AY486137.1) as a reference gene, the reaction enzyme for qRT-PCR was 2× ChamQ Universal SYBR qPCR Master Mix (Vazyme, Nanjing, China). Finally, the data were analyzed using the 2^−ΔΔCT method.

2.7. Screening of Enzyme Genes Related to Terpenoid Metabolic Pathway in C. annuum

The expression of terpenoid metabolic pathway-related enzyme genes was investigated in three chili species at different developmental stages in combination with qRT-PCR. The contents of major flavor substances of terpenoid metabolic pathways in the non-targeted metabolomics of the three chili species were analyzed, and the linear relationship between the expression levels and flavor substance contents was determined in C. annuum.

2.8. Statistical Analysis

The experiment was conducted in a completely randomized manner, with three replications set up for each experiment, and the data are expressed as mean ± standard deviation (SD). Statistical analyses of all data in this study were performed using SPSS 22.0 software (Statistical Package for the Social Sciences, Chicago, IL, USA). Significant differences were determined by Duncan’s multiple range test at a level of p < 0.05. Histograms were generated by GraphPad Prism (version 5, GraphPad Software, Washington, DC, USA).

3. Results

3.1. Physicochemical Properties of Terpene Metabolic Pathway Gene Proteins

A total of sixty-seven genes related to terpene metabolism were identified from the C. annuum genome. Specific information is provided in the Supplementary Material, which is detailed in the Supplementary Material Tables S3 and S4. The molecular weights of the proteins of the genes varied widely, with the largest being CaTPS-12 (98.5 KD) and the smallest being CaMECPS-1 (24.48 KD). The isoelectric points ranged from 4.98 (CaTPS-26) to 9.04 (CaAACT-1), and the molecular weights of the genes in each gene family did not differ much from the isoelectric points. There are 29 stable proteins and seven hydrophilic proteins, CaAACT-1-2, CaHMGR-1-2-3, CaMVK-1, and CaDXS-1. Subcellular localization prediction shows that 34 proteins, including CaAACT-1, CaPMK-1, CaIDI-1, etc., are localized in chloroplasts, while the remaining 33 genes are mostly located in cytoplasm or mitochondria. Among the 33 CaTPSs, 13 members such as CaTPS-2-3-4-10-23 are in chloroplasts and may participate in MEP metabolism, while 15 members such as CaTPS-1-5-6-18 are localized in cytoplasm and are likely to regulate MVA metabolism. CaHMGR-1-3 are localized in peroxisomes, and CaHMGR-2 and CaTPS-20 are localized in mitochondria. Detailed information is provided in the supplementary information Table S5.

3.2. Chromosomal Localization Analysis of Terpene Metabolic Pathway Genes

To accurately understand the localization of 67 genes on chromosomes, we constructed a chromosome map based on the location information retrieved from the C. annuum genome database (Figure 2). Sixty-one genes were distributed across 12 chromosomes; CaDXS-1, CaDXS-2, CaGGPPS-4, CaGGPPS-5, CaTPS-7, and CaTPS-33 have not yet been positioned on any chromosome. The highest number of genes were found on chromosome 8 where 13 genes were distributed. The number of genes distributed on chromosomes 3, 7, 9, and 11 ranged from 2 to 7. The minimum number of genes contained on chromosome 10 was only one. Tandem duplication of genes was also found on several chromosomes. For example, two groups of tandemly duplicated genes were present on chromosome 3, CaTPS-30 and CaTPS-27, and CaTPS-2 and CaTPS-21. It was postulated that there were clusters of metabolic genes on chromosome 8. Concerning CaTPS-24, CaTPS-26, CaTPS-3, and CaTPS-4, all four of these genes are engaged in the MEP pathway and have a spacer length of less than 3.5 Mb, which is consistent with the general characterization of gene clusters.

3.3. Phylogenetic, Gene Structure, and Promoter Analysis of Terpene Metabolism Genes

To reveal the potential functions of related proteins in the evolutionary process, we constructed the phylogenetic relationships of enzyme proteins involved in the terpenoid metabolism pathway. The phylogenetic trees of 18 related enzymes are shown in detail in the attached image 1. Most of the genes are more closely related to Solanum lycopersicum, e.g., CaGPPS-1 and CaMECPS-1. But interestingly, CaHMGR-2 is more closely related to Artemisia annua, and CaFPPS-1 is more closely related to Gossypium hirsutum, which may be related to the retention of relative protein sequences during the evolutionary process. CaPMK-2 is more closely related to Nicotiana tabacum, and the same is also true for CaIDI-1-3 and CaDXS-2, which is consistent with the fact that chili and tobacco belong to the same family, Solanaceae. Moreover, the CaTPSs family was further divided into four subclasses: TPS-a, TPS-b, TPS-e/f, and TPS-g. Among them, the TPS-a subclass contained twenty-two genes, the TPS-e/f subclass contained CaTPS-5 and CaTPS-12, with the former being more closely related to Arabidopsis KS1, and the latter being more closely related to Arabidopsis TPS4. The TPS-g subclass contained only the CaTPS-31 gene.

To further explore the structure, function, characteristics, and conserved motifs of terpene metabolic pathway-related genes and their amino acid sequences, the number and location of introns and exons and conserved motifs of sixty-seven members were analyzed (Figure 3). Of the 67 enzyme genes identified, 53 genes contained conserved motifs, while 14 genes did not retrieve any motifs, such as CaHDS-1, CaIDI-1-2-3, CaDXR-1, CaAACT-1-2, CaMVK-1, CaMECPS-1, CaDXS-1-3, CaHDR-1, and CaCMK-1. Genes in the same gene family are highly conservative and share many of the same motifs. CaTPSs all contain motifs 5, 2, 3, and 6. All CaGGPPS genes have motifs 18 and 21 (Figure 3B). Yet, there is still some variation in the type of motifs of different genes within the same family. For example, in the CaPMK family, only CaPMK-2 contains motif 23. In the CaTPS family, CaTPS-12 and CaTPS-5 lack motif 9, CaTPS-12 and CaTPS-24 lack motif 8, and only CaTPS-5 has motif 23. In addition, CaTPS-2 has two instances of motif 9. The presence of these unique motifs suggests that these genes may have specific functions in the family. Conservative domain analysis shows that in most cases, the same gene family has the same domain characteristics. Members of CaTPSs include both N-terminal and C-terminal domains (Figure 3C). However, CaHMGR-4 differs from other CaHMGRs in that it possesses a unique MARVEL quadruple transmembrane domain. CaDXS-1 lacks an N-terminal domain but has an N-terminal gene superfamily domain. These differences may lead to some functional differences in gene expression products. In addition, CaFPPSs, CaGGPPSs, and CaGPPS all contain polyprenyl-synt structural domains, suggesting that these genes may exercise the same or similar functions. According to the analysis of the gene structure (Figure 3D), the number of introns ranged from 0 to 19, with CaGGPPS-2, CaGGPPS-3, and CaGGPPS-4 containing no introns, whereas CaHDS-1 had 19 introns. The number of exons ranged from 1 to 19, with CaGGPPS-2, CaGGPPS-3, and CaGGPPS-4 containing only 1 exon, whereas CaHDS-1 had 19 exons. Different gene families showed significant differences in the number of introns and exons. Exon numbers in CaGGPPSs ranged from 1 to 6 predominantly, whereas exon numbers in CaTPSs were clustered between 7 and 8. In addition, members of the same gene family usually have similar numbers of introns and exons. For example, CaIDI-1-2-3 all have 6 exons, while 10 introns appear in CaPMK-1 and CaPMK-2. Further analysis reveals that CaIDI-2, CaPMK-2, CaDXS-1, and CaTPS-8 have significantly longer intron lengths than other genes within the same family, suggesting that these genes have more expression regulatory elements.

Furthermore, the upstream 2000 bp sequences of the start codons of 67 terpenoid-related genes were analyzed to reveal the expression characteristics of terpenoid genes and explore possible regulatory mechanisms. The results showed that the promoters of terpenoid genes all contain elements related to abiotic stress, biotic stress, light induction, and plant growth and development. Among them, the number of elements related to abiotic stress and plant growth and development accounted for more than 85%. Except for CaGGPPS-2, CaHMGR-2, and CaTPS-9, all other gene promoters contain MYB binding components (Figure 4). CaTPSs promoter elements are mainly involved in abiotic and biotic stresses as well as in phytohormone response factors. Stress and hormone response elements were also commonly distributed in the gene promoters. The extracted stress-related cis-elements (Myb, Myc, ARE, and STRE) indicated that CaIDI-3, CaTPS-8, CaTPS-32, and CaHDR-1 might play a key role in responding to unfavorable conditions. In addition, the drought-inducible element MBS was found among the abiotic stress elements, and this type of response element could be involved in transcriptional regulation in plants in synergy with the MYB element. Among the light-responsive elements, CaFPPS-2 may play a key role in the light-controlled development of plants. In addition, some promoters are rich in ABRE (involved in ABA response), such as CaFPPS-2, CaTPS-2, and CaTPS-25, and these genes may be responsive to ABA hormones. Further analysis of the number of cis-acting elements in the promoter region shows that CaFPPS-2 contains the most cis-acting elements, while CaHDR-1 contains the most nonbiological stress response elements. CaMECPS-1, CaTPS-1, CaTPS-10, and CaTPS-20 do not contain cis-acting elements related to photoreaction.

3.4. Expression Patterns of Enzyme Genes for Terpene Metabolism and Synthesis During Chili Fruit Development

The expressions of 67 genes were analyzed by transcriptome data at the fruit green, breaking, and red stages in GLD608, GJ, and HDL (Figure 5). The results showed that AACT-1, IDI-2, GGPPS-1, HDR-1, and HDS-1 exhibited high expression levels at various stages of pepper fruit development, indicating that these genes were involved in the biosynthesis of terpenoid flavor compounds throughout the entire development process. GGPPS-3, GGPPS-5, TPS-2, TPS-4, TPS-7, TPS-9, TPS-11, TPS-13, TPS-15, TPS-16, TPS-18, TPS-19, TPS-22, TPS-24, TPS-26, TPS-27, TPS-29, TPS-31, TPS-32, and TPS-33, a total of 20 genes, were not detected to be expressed. In addition, there are 16 genes with differential expressions, including AACT-2, HMGS-3, HMGR-2, MVK-1, PMK-1, MVD-1, IDI-3, FPPS-1, GGPPS-2, DXR-1, DXS-2, MCT-1, CMK-1, MECPS-1, GPPS-1, and TPS-5. The remaining twenty-six genes showed differential expressions among varieties or developmental stages. For example, HMGS-2 expression was higher at the green fruit stage in GLD608. TPS-1 and TPS-3 had high expression at the green fruit stage in HDL. TPS-12 showed high countenance at the green fruit stage in all three varieties. There are also some genes with differential expressions in a variety of types and developmental stages. FPPS-1, IDI-1, DXS-3, TPS-6, TPS-10, TPS-16, TPS-26, TPS-28, TPS-30 showed high expression levels during the green fruit stage of GLD608, while it was not detected or expressed very low in other varieties and growth stages. TPS-23 showed high expression levels in GJ and HDL at the breaking and red stages. Additionally, HMGS-1, FPPS-2, GGPPS-6, DXS-1, TPS-17, and TPS-18 were detected in tiny amounts at different growth stages of the three varieties.

Furthermore, we validated the expression levels of 16 differentially expressed genes (AACT-2, HMGS-3, HMGR-2, MVK-1, PMK-1, MVD-1, IDI-3, FPPS-1, GGPPS-2, DXR-1, DXS-2, MCT-1, CMK-1, MECPS-1, GPPS-1, TPS-5) at different developmental stages of GLD608, GJ, and HDL fruits using qRT-PCR. (Figure 6). In GLD608, the expression levels of most genes first decreased and then increased. For illustration, compared with the green fruit stage, the expression levels of the AACT-2 and HMGS-3 genes, respectively, increased by 1.5 times and 3 times in the mature stage. It is worth noting that there are also some gene expression levels that show continuous changes over time. Concerning the IDI-3 gene, its expression level continues to increase, and the mature stage is about thirteen times that of the green fruit stage.

The expression levels of some genes in GJ, such as FPPS-1, CMK-1, MECPS-1, GGPPS-2, and TPS-5, are significantly higher during the color transformation and mature stages than during the green fruit stage, about 8 to 40 times higher. Also, only the AACT-1 gene had the lowest expression level during the mature stage, decreasing by 2.5 times compared to the green stage and 4.5 times compared to the color transition stage. The expression level of the MCT-1 gene in HDL during the green fruit stage is about 14 times that during the color transition stage and 50 times that during the mature stage. In addition, most genes (such as AACT-1, FPPS-1, and IDI-3) have much higher expression levels during the green fruit stage and color transition stage than during the mature stage. The most noteworthy aspect is that the expression levels of the TPS-5 genes in the three stages are the same, indicating that certain aroma substances in HDL may continue to be synthesized and volatilized.

By comparing the changes in the gene expression levels of GLD608, GJ, and HDL, it was found that the expression levels of GLD608 showed an opposite trend to those of GJ and HDL. Also, the gene expression levels of GJ and HDL are higher than GLD608, especially the AACT-1, HMGR-2, FPPS-1, GPPS-1, and TPS-5 genes. This may explain the differences in aroma characteristics among different chili varieties.

3.5. Changes in Expression of Enzymes and Correlation with Content of Corresponding Metabolites

To analyze the relationship between the gene expression of the terpene metabolic pathways and changes in aroma substance content, we determined the changes in terpene content at three stages of fruit development in three varieties, GLD608, GJ, and HDL, by using an untargeted metabolomics approach, and selected key volatiles for analysis, including α-calacorene (C15H20), α-cubebene (C15H24), cis-β-farnesene (C15H24), himachala-2,4-diene (C15H24), nerolidol (C15H26O), and linalool (C10H18O).

Table 1. Qualitative and quantitative table of volatile aroma compounds in chili fruits of three varieties and three growth stages.

Retention Indices	Retention Time	Formula	Stage	Green Stage
			Aroma Volatile Profile (μg/kg)	GLD608					GJ	HDL
1540.658602	29.957	C15H20	α-calacorene	0.189501185	0.169392939	0.161516859	2.026104808	1.83741121	1.256568456	9.08122777	12.42851114	13.34401757
1346.721064	23.854	C15H24	α-cubebene	0.189501185	0.169392939	0.161516859	6.750840996	4.859400783	4.038638958	24.42060494	34.12703405	34.63315802
1450.990415	27.213	C15H24	cis-β-farnesene	9.914039453	6.998453542	9.900740257	4.282773783	3.054284601	2.100333178	29.31950673	44.11500031	0.566948561
1481.916933	28.181	C15H24	himachala-2,4-diene	0.672396897	0.728896278	0.890786308	0.46487621	0.375202389	0.295370111	1.781478646	81.0652433	89.05190136
1560.719086	30.554	C15H26O	nerolidol	5.100512268	5.275688126	6.298854821	0.052066552	0.059313761	0.064776714	0.105807802	0.108882511	0.114321495
1098.606177	15.15	C10H28O	linalool	17.72278692	29.74326537	21.83802079	5.307310451	5.187160583	8.333778717	3.449577756	2.020037205	2.212748445
Retention Indices	Retention Time	Formula	Stage	Transition Stage
			Aroma Volatile Profile (μg/kg)	GLD608					GJ	HDL
1540.658602	29.957	C15H20	α-calacorene	0.290480591	0.307122624	0.261394513	14.57378854	14.76637357	12.91119967	17.3293049	19.20997281	18.01882563
1346.721064	23.854	C15H24	α-cubebene	0.032606452	0.03005341	0.037267364	22.99147365	0.257943059	0.588673996	32.22838886	35.82612365	34.24461297
1450.990415	27.213	C15H24	cis-β-farnesene	6.203308506	6.999444129	6.297113557	111.2381479	134.2963163	115.9745627	118.6495515	98.17047579	138.5707145
1481.916933	28.181	C15H24	himachala-2,4-diene	1.519322631	1.519458133	1.559081741	92.49001943	102.9068201	88.94440934	19.41318347	5.577432573	13.49877299
1560.719086	30.554	C15H26O	nerolidol	4.725287206	4.891903877	4.684003949	0.343238587	0.336734435	0.336889952	0.195508473	0.192212047	0.257280851
1098.606177	15.15	C10H28O	linalool	13.28365092	13.26640641	13.47729543	5.776609437	8.67423761	6.456407814	2.099548198	4.399658607	1.901405828
Retention Indices	Retention Time	Formula	Stage	Red stage
			Aroma Volatile Profile (μg/kg)	GLD608					GJ	HDL
1540.658602	29.957	C15H20	α-calacorene	0.196212603	0.250821045	0.204236846	11.19527999	9.958240804	6.920142137	18.75339771	20.40489899	22.11276696
1346.721064	23.854	C15H24	α-cubebene	0.007219878	0.018750759	0.020920522	26.13886164	27.64447898	23.07029552	32.35776509	33.17996989	34.38264749
1450.990415	27.213	C15H24	cis-β-farnesene	7.369725556	9.761146872	6.212745679	102.0934501	85.58305552	70.58480417	137.6160577	152.7423527	158.0275569
1481.916933	28.181	C15H24	himachala-2,4-diene	0.704335791	0.82826117	1.036842206	60.51347448	84.83262687	49.36515377	14.03306576	3.924964995	10.53879978
1560.719086	30.554	C15H26O	nerolidol	5.001703282	7.459903041	8.821545182	0.386440769	0.409054335	0.423033271	0.262315722	0.3180631	0.297885325
1098.606177	15.15	C10H28O	linalool	10.41722888	8.321212038	9.494170242	5.968380794	6.227085797	5.889582503	2.839207021	2.164262988	2.674034191

demonstrates the retention indices, retention times, and concentrations of the six major compounds for GLD608, GJ, and HDL at the three stages of green, transition, and red, where the concentrations provide the values for three replicate treatments. The results showed that there was no significant difference in the content of cis-β-farnesene during the green fruit stage of the variety, while the volatilization of other compounds took meaningful changes during the fruit development stage. Among them, the three developmental stages of HDL, the content of α-calacorene (Figure 7A) and α-cabebene (Figure 7B), is about 100 times higher than GLD608 during the green fruit stage, and about 300 times higher during the color transformation and maturity stages, indicating that HDL varieties have stronger woody aroma characteristics. Cis-β-farnesene (Figure 7C) and himachala-2,4-diene (Figure 7D) showed remarkable changes in their levels of expression at mid- to late fruit development in GJ and HDL. Cis-β-farnesene content in HDL was about 40-fold higher than that of GLD608 at the color-turning stage, and 60-fold higher at the mature stage. Himachala-2,4-diene volatilization was 600-fold higher in GJ than in GLD608 at the turn-color stage and about 500-fold higher at the mature stage, showing that GJ and HDL have the strongest pungent aroma at mid- to late fruit development. The higher volatility of linalool (Figure 7E) and nerolidol (Figure 7F) in GLD608 indicates that the variety possesses a stronger herbal aroma.

Combined with analyses of distinct stages of expression in the same species, it was found that the content of α-cubebene (Figure 7H) declined consistently in GLD608, whereas the content of cis-β-farnesene (Figure 7I) and himachala-2,4-diene (Figure 7J) increased notably in GJ. Of interest, the levels of key volatile substances were equivalent at different developmental stages of HDL, indicating stable synthesis as well as volatilization of the compounds.

Gene expression plays a vital role in the plant, which directly or indirectly affects the strength of metabolites. It is therefore reasonable to assume that these genes have a direct relationship with the formation of the corresponding compounds. CaHMGS-3 and CaMVD-1 are involved in the MVA metabolic pathway for the formation of sesquiterpene and diterpene compounds, while CaCMK-1 and CaMECPS-1 are involved in the MEP metabolic pathway for monoterpene compounds, and CaGGPPS-2 as well as CaTPS-5 are involved in the synthesis of downstream metabolites. By analyzing the correlation between gene expression and terpene content during ripening of C. annuum fruits (Figure 8), it was shown that the expression levels of CaHMGS-3 (Figure 8A), CaMVD-1 (Figure 8B), CaCMK-1 (Figure 8C), and CaGGPPS-2 (Figure 8E) were positively correlated with linalool, and the expression level of CaMECPS-1 (Figure 8D) was negatively correlated with the expression level of cis-β-farnesene showed a significant positive correlation. While the expression level of CaTPS-5 (Figure 8F) showed a strong positive linear relationship with nerolidol. The above results suggest that these genes may play a key regulatory role in the production of different terpenoids and are directly involved in the synthesis of various volatile compounds. The relationships between other genes and compounds are shown in Supplemental Information Figure S2.

4. Discussion

Many genes involved in terpene metabolism and biosynthesis were identified, cloned, and are well characterized. In this study, we identified 67 genes related to enzymes involved in flavor synthesis, analyzed their physicochemical properties, and performed subcellular localization predictions. Among these 67 genes, nearly all encode hydrophilic proteins, except for CaAACT-1-2, CaHMGR-1-2-3, CaMVK-1, and CaDXS-1, which are hydrophobic. However, the influence of pH within distinct cellular microenvironments on the synthesis, stability, and function of these proteins remains to be further investigated.

Chromosome localization shows that most genes are distributed on Chr8, with only CaFPPS-2 located on Chr10. Tandem and fragment duplication analysis revealed multiple gene clusters in the duplicated regions, indicating that some terpenoid synthesis genes may be generated through gene duplication events. Multiple copies of these genes may buffer mutation effects [22], thereby enhancing adaptability under adverse environmental conditions. Tandem genes provide genetic redundancy, maintaining functional stability in the face of gene-specific mutations or enabling novel functions through mutational changes [23]. Additionally, environmental stress and adaptive evolution are likely to increase the probability of tandem duplication within populations. Identifying these duplication events provides valuable insights into evolutionary relationships among genes and aids in predicting their functions. We further hypothesize the presence of metabolic cluster genes among them. These metabolic cluster genes, which are linked closely in clusters, synergistically interact to regulate multiple secondary metabolic pathways [24], acting as co-regulators within biological systems [25]. Gene clusters for various compounds, including terpenoids, have already been identified in species such as Zea mays L. [26], Avena sativa L. [27], and Oryza sativa L. [28]. The existence of such clusters may elucidate genome evolution and offer insights into the expansion and functional diversification of terpenoid metabolic pathways, providing clues about species’ evolutionary history. Studying the transcriptional regulatory mechanisms of these clusters, including the ways co-regulated genes achieve synergistic expression via shared promoters, enhancers, and other regulatory elements, holds significant potential for advancing our understanding of metabolic pathway regulation.

The phylogenetic tree revealed complex evolutionary relationships among terpenoid metabolism genes in peppers and those in other species. A total of 67 enzyme-encoding genes exhibited high sequence similarity to terpenoid pathway-related genes in the model plants Arabidopsis thaliana L., Solanum lycopersicum L., and Oryza sativa L., which clustered within the same branches. We hypothesize that these genes may perform functions analogous to those in other species, providing a basis for future functional verification experiments. An analysis of the gene aggregation patterns and branch length patterns suggests that these genes experienced gene duplication, functional divergence, and other evolutionary events. The overexpression of AACT, HMGR in Arabidopsis confirmed an increase in the content of triterpenoid compounds [29]. In tomato, Ezquerro showed [30] that the GGPPS enzyme is involved in the increase in diterpenoid content in leaves as well as roots during stress. The KS gene encodes a protein with ent-kaurene synthase activity, which catalyzes the second step in the gibberellin biosynthesis pathway in the cyclisation of GGPP to ent-kaurene. This gene has been shown to be involved in the Oryza gibberellin anabolic pathway [31] but has been poorly studied in flavor. This information is crucial not only for understanding the roles of terpenoid metabolism genes in peppers and their functions within metabolic pathways but also for gaining insights into the evolution of gene families.

Introns are essential components of eukaryotic genes, exhibiting both conservation and diversity in their structure. Among the 67 genes analyzed, intron-free genes were discovered (CaGGPPS-2, CaGGPPS-3, CaGGPPS-4). Introns confer evolutionary advantages by increasing gene length, enhancing intergenic recombination, and contributing to gene expression regulation [32]. Intron less genes, while lacking the recombinational advantages associated with introns, may serve a unique function in rapid response regulation due to their swift transcription into proteins. We hypothesize that CaGGPPS-2, CaGGPPS-3, and CaGGPPS-4 may contribute to faster plant growth and development, potentially shortening breeding cycles. Gene expression patterns correlate significantly with intron length, which tends to be shorter in high-abundance genes compared to low-abundance genes in animals and Arabidopsis [33]. Thus, quantifying low-abundance genes may require highly sensitive assays for accurate measurement.

Cis-acting elements are pivotal in gene regulation. These DNA sequences, located in or near gene promoter regions, specifically regulate transcription levels by binding to transcription factors [34]. Understanding these elements and their interactions with promoters is essential for elucidating gene expression mechanisms and regulatory networks. An analysis of the cis-acting elements indicated that the more numerous and conserved CaGGPPSs and CaTPSs are critical to terpene metabolism and synthesis, providing a foundation for future research into downstream product mechanisms and functions in pepper terpene pathways. Further analysis revealed that the 34 identified cis-regulatory elements were predominantly associated with phytohormone regulatory pathways, especially those involved in methyl jasmone and salicylic acid responses, such as the CGTCA-motif (methyl jasmone response) and TCA-element (salicylic acid response). Methyl jasmone and salicylic acid are two major resistance signaling pathways in plants that play distinct roles in regulating the synthesis and metabolism of flavor compounds. These findings offer new perspectives for investigating the mechanisms underlying chili fruit flavor compound synthesis.

Heatmap analysis revealed that some genes exhibited a darker red color, indicating a significant increase in expression and high activation of genes related to the synthesis of chili aroma compounds. This pattern suggests that these conditions may be associated with the initiation or active phase of aroma synthesis pathways. In contrast, other genes displayed lighter-colored or blank regions, suggesting no strong correlation with aroma pathway activation. These differences indicate that specific conditions may favor the activation of flavor compound synthesis pathways, especially those involving TPSs, GGPPSs, DXRs, and DHSs. Combined with the heatmap, we can see that certain genes are highly expressed under specific conditions, such as AACT-1 and HDR-1, which are consistently highly expressed under multiple conditions, suggesting that they may be “universal” synthetic genes that are always involved in the synthesis of fragrance substances. And the enhanced expression of other genes (TPS-5-26) under specific conditions may be involved in the synthesis of specific compounds, which reflects the complexity and diversity of the synthesis pathway of pepper flavor substances.

Volatile terpene components serve as crucial indicators for aroma identification [35]. C. annuum and C. chinense are among the most widely cultivated peppers in China. Through a combination of expression and volatile content analysis, we selected six volatiles: α-calacorene (C15H20), which has a woody aroma, α-cubebene (C15H24), rose-oil scented cis-β-farnesene (C15H24), pungent himachala-2,4-diene (C15H24), apple blossom scented nerolidol (C15H26O), and slightly herbaceous linalool (C10H18O). Alpha.-calacorene and alpha.-cubebene are currently less studied. It is known that the former can be obtained from the rhizomes of Acorus calamus L., while the latter can be isolated from Mentha canadensis L., and both are commonly used as flavoring constituents. Nerolidol, a naturally occurring sesquiterpene, serves as a volatile signal that induces plants to defend themselves against insects and pathogens [36]. Additionally, nerolidol can enhance the activity of enzymes responsible for degrading juvenile-protecting hormones in plants and inhibit the growth and development of insects, providing a new perspective for the development of novel pesticides with potent antimicrobial activity [37]. Linalool is a natural compound resistant to microorganisms [38], which can boost the supply of GPP as a precursor to enhance the efficiency of GPP in linalool biosynthesis [39] and achieve potent antibacterial activity [40]. A comparative study of volatile content revealed that the GJ and HDL varieties possess higher levels of α-calacorene, α-cubebene, cis-β-farnesene, and himachala-2,4-diene compared to GLD608. Thus, they exhibit stronger herbal and pungent flavors. A study by Moreno et al. [41] highlighted that the most notable difference between C. annuum and C. chinense is that the latter has a volatile pungent odor and a distinctive spicy flavor [42], which is consistent with our findings. Furthermore, it was observed that the less spicy varieties exhibit a more pronounced aromatic odor, and the content of α-cubebene in HDL varieties is significantly higher than that in GJ varieties. Previously known differences were observed in the expression of enzyme genes related to metabolic pathways across various varieties and developmental periods. Furthermore, the content of volatile substances exhibited differing trends, suggesting a potential delayed effect on the synthesis of chili flavor compounds at various developmental stages. Negative correlations between volatiles and gene expression levels suggest that these genes may play a balancing or regulatory role in terpenoid synthesis. When gene expression levels are elevated, the synthesis of the final compound may be reduced by inhibiting the flow of specific precursor materials. Conversely, the positive correlation between gene expression and volatile content implies that these genes may be directly or indirectly involved in the synthesis of target compounds, either by promoting the generation of key precursor substances or by stimulating the accumulation of downstream metabolites, thereby enhancing metabolic pathway activity. Several potential reasons for the inconsistency between some genes and volatile organic compounds include, but are not limited to, the insufficient expression level of a single gene to significantly affect the concentration of final metabolites, inadequate substrate supply, shunting in metabolic pathways, or the necessity for constructive interaction among multiple genes. Thus, to assess these hypotheses, it is essential to further investigate the specific functions and regulatory mechanisms of these genes in the synthesis of aroma compounds through gene silencing or knockout experiments.

Substantial progress has been made in understanding gene expressions linked to plant terpenoid metabolic pathways, including biosynthetic processes and molecular markers associated with terpene content [43]. However, the regulation of secondary metabolite synthesis through the synergistic control of related enzymes is still underexplored. This study not only sheds light on the evolutionary development of terpenoid biosynthetic pathways in annual chili peppers but also offers insights into the evolutionary pathways of terpenoid metabolism. These findings are anticipated to support research aimed at enhancing chili fruit flavor and provide a theoretical foundation for breeding chili varieties with intensified flavor profiles. In this study, gene family screening, RNA-seq analysis, and GC-MS linkage detection were used to finally screen candidate genes that may be associated with the synthesis of specific volatile compounds. Our results are more scientific and reliable than the studies on gene families only.

Overall, terpene metabolism-related genes play a crucial role in the growth and development of peppers. Comparative genomic analysis provides essential insights into terpene metabolic pathways and serves as a valuable reference for aroma enhancement and pepper breeding. This work lays a foundation for future research into genetic improvement and the precise regulation of functional genes in chili varieties.