GC-MS Metabolite and Transcriptome Analyses Reveal the Differences of Volatile Synthesis and Gene Expression Profiling between Two Apple Varieties

Aroma is a key quality attribute of apples, making major contributions to commercial value and consumer choice. However, the mechanism underlying molecular regulation of aroma formation genes and transcription factors remains poorly understood in apples. Here, we investigated the aroma volatile profiles of two apple varieties with distinctive flavors using headspace solid-phase microextraction (HS-SPME) combined with gas chromatography–mass spectrometry (GC-MS). A total of 35 volatile compounds were identified in Granny Smith and Jonagold apples. Aldehydes were the most abundant volatiles contributing to the aroma in Granny Smith apple while esters were the dominant volatile compounds in Jonagold apple. In order to know more about the expression levels of aroma-related genes involved in the metabolic pathways, transcriptome sequencing of these two different apple varieties was conducted utilizing the Illumina platform. In total, 94 differentially expressed genes (DEGs) were found in the fatty acid metabolism, amino acid metabolism, the mevalonate pathway and phenylpropanoid pathway. Furthermore, compared to the Granny Smith apple, the expression of multiple genes and transcription factors were upregulated in the Jonagold apple, which might play important roles in the synthesis of aroma volatile compounds. Our study contributes toward better understanding on the molecular mechanism of aroma synthesis in apples and provides a valuable reference for metabolic engineering and flavor improvement in the future.


Introduction
Apple (Malus × domestica Borkh.) is one of the most widely cultivated fruits in temperate regions of the world [1]. The characteristic flavor of apple fruits depends upon taste and aroma, while aroma is often considered the dominant component and greatly affects consumer acceptance [2][3][4]. The aroma profile in apple results from the interaction of many volatile compounds, including esters, alcohols, aldehydes, ketones, acids and terpenoids, with only a few that contribute significantly to the apple sensory quality [5][6][7]. The esters, particularly those with six to ten numbered carbon chains including combinations of acetic, butanoic, and hexanoic acids with ethyl, butyl, and hexyl alcohols, have been considered as the primary contributors to apple fruity aroma [8]. Alcohols are the second most important compounds in ripe apples after esters, the most abundant being 2-methyl-1-butanol, 1-butanol, 1-hexanol and 1-propanol [9]. Aldehydes are produced at relatively high amounts in immature apples, such as hexanal, trans-2-hexenal and butanal [10]. Nowadays, more than 300 volatile compounds have been measured in apple fruits [11]. All the volatiles are of great importance for the complete characteristic aroma profile of apples.

Phenotype and Physiological Traits of Granny Smith and Jonagold Apples
The phenotypes of Granny Smith and Jonagold apples were observed. As shown in Figure 1, the apple variety of Granny Smith was characterized with the green skin, in comparison with the red stripe skin in Jonagold. Color perception is the result of three parameters L* (lightness), a* (red-green), and b* (yellow-blue). The values of color parameters L*, a* and b* were 63.52, −17.62 and 37.95 in Granny Smith, while these values in Jonagold were 58.62, 13.38 and 29.13. Moreover, there are some other differences in physiological traits between these two apple varieties ( Table 1). The fresh weight (FW) of Granny Smith and Jonagold apples was 205 and 240 g, respectively. The total soluble solid (TSS) of Granny Smith (13.7 • Brix) was lower than that of Jonagold (14.2 • Brix). On the contrary, the titratable acidity (TA) of Granny Smith (0.40%) was higher than that of Jonagold (0.32%). Firmness is another important factor to evaluate the quality of apple fruits [25]. The firmness of Granny Smith was 7.05 kg/cm 2 , higher than that in Jonagold (6.58 kg/cm 2 ) (Table 1). Notably, the internal ethylene of the Granny Smith apple was 5 µL/L, lower than that of Jonagold (62 µL/L). However, the CO 2 production rate in Granny Smith (70 µmol/kg·s) was higher than the 48 µmol/kg·s in Jonagold apple (Table 1). [25]. The firmness of Granny Smith was 7.05 kg/cm 2 , higher than that in Jona kg/cm 2 ) (Table 1). Notably, the internal ethylene of the Granny Smith apple w lower than that of Jonagold (62 μL/L). However, the CO2 production rate in Gran (70 μmol/kg·s) was higher than the 48 μmol/kg·s in Jonagold apple (Table 1).  Values are means ± standard deviation; TSS-total soluble solid; TA-titratable acidity.

Volatile Compound Profiles in Granny Smith and Jonagold Apples
Aroma is an essential sensorial property, which determined by the vola  Values are means ± standard deviation; TSS-total soluble solid; TA-titratable acidity.

RNA-seq, Assembly and DEGs Functional Annotation
To explore the molecular mechanisms underlying differences in aroma volatile compounds between Granny Smith and Jonagold apples, six cDNA libraries (three biological replicates) were constructed from these two apple varieties and sequenced utilizing the Illumina sequencing platform. Approximately 6.15-6.94 G of clean bases were produced from Granny Smith and Jonagold apples (Table S3). After data filtering and quality trimming, 41.01-46.29 million clean reads were obtained through sequencing of the cDNA libraries prepared from these two apple varieties. The Q30 percentage was over 93.01% for each sample and the average GC content was 47.03% for all six libraries ( Figure S1; Table S3). Furthermore, average error rate of the sequenced bases was less than 0.03% ( Figure S2). A total of 33.21-40.12 million reads were mapped to the apple reference genome, with match ratios in the range of 78.29-92.30% ( Figure S3; Table S4).
FPKM methods were used to analyze gene expression patterns in Granny Smith and Jonagold apple libraries ( Figure S4). Comparing these two types of libraries with respect to the FPKM calculation, 20,481 and 20,794 genes were identified in the cDNA libraries, while 1954 and 2267 genes were expressed specifically in Granny Smith and Jonagold apples, respectively ( Figure S5). A total of 8117 DEGs were detected, including 3682 upregulated genes and 4435 downregulated genes ( Figure 3a). To investigate the differences in gene expression patterns between Granny Smith and Jonagold, DEGs were subjected to hierarchical clustering analysis of transcript abundances using the FPKM values, and these DEGs had different transcriptome profiles in the two apple varieties (Figure 3b).
In order to know more about the biological function of the DEGs, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed. According to the GO enrichment analysis, these DEGs were classified into three categories including biological process, cellular component and molecular function. The most abundant terms of each category were ADP binding, single-organism carbohydrate metabolic and cell wall process (Figure 3c). Based on the KEGG enrichment analysis, DEGs were matched and assigned to 122 pathways. The most enriched pathways were carbon metabolism, biosynthesis of amino acids, plant hormone signal transduction and fatty acid metabolism. Moreover, the pathways of phenylalanine metabolism and linoleic acid metabolism were also enriched ( Figure 3d).

Expression Levels of Genes Involved in Fatty Acid and Isoleucine Metabolism Pathways
Fatty acid and isoleucine metabolism are the two main pathways for synthesis of aldehydes, alcohols and esters in apples [6]. In this study, we identified 50 DEGs related to fatty acid metabolism and 24 DEGs associated with isoleucine metabolism (Figure 4). Among them, most of the DEGs showed higher expression levels in the Jonagold apple than that in the Granny Smith apple. In the pathway of fatty acid biosynthesis, the expression levels of ACC, MAT, KAS and FAD genes exhibited higher expression levels in the Jonagold apple in comparison with in the Granny Smith apple. Similarly, the DEGs KAR, DH, ER and OTE showed the same expression patterns. Lipoxygenase (LOX) is the initial step of ester biosynthesis in fatty acid metabolization [26]. Twelve LOX DEGs were identified in our study, most of which were significantly upregulated in the Jonagold apple. Hydroperoxide lyase (HPL) is the last step toward aldehyde biosynthesis [27]. Compared to the Granny Smith apple, the expression levels of HPL gene showed higher in the Jonagold apple (Table S5). Alcohol dehydrogenases (ADH) can catalyze the aldehydes to the

Expression Levels of Genes Involved in Fatty Acid and Isoleucine Metabolism Pathways
Fatty acid and isoleucine metabolism are the two main pathways for synthesis of aldehydes, alcohols and esters in apples [6]. In this study, we identified 50 DEGs related to fatty acid metabolism and 24 DEGs associated with isoleucine metabolism (Figure 4). Among them, most of the DEGs showed higher expression levels in the Jonagold apple than that in the Granny Smith apple. In the pathway of fatty acid biosynthesis, the expression levels of ACC, MAT, KAS and FAD genes exhibited higher expression levels in the Jonagold apple in comparison with in the Granny Smith apple. Similarly, the DEGs KAR, DH, ER and OTE showed the same expression patterns. Lipoxygenase (LOX) is the initial step of ester biosynthesis in fatty acid metabolization [26]. Twelve LOX DEGs were identified in our study, most of which were significantly upregulated in the Jonagold apple. Hydroperoxide lyase (HPL) is the last step toward aldehyde biosynthesis [27]. Compared to the Granny Smith apple, the expression levels of HPL gene showed higher in the Jonagold apple (Table S5). Alcohol dehydrogenases (ADH) can catalyze the aldehydes to the corresponding alcohols. Seven DEGs encoding ADH were detected in our dataset, most of which were significantly upregulated in the Granny Smith apple. Alcohol acyltransferase (AAT) is the key enzyme for the biosynthesis of esters [9]. Ten AAT DEGs exhibited higher expression levels in the Jonagold than that in the Granny Smith apple. corresponding alcohols. Seven DEGs encoding ADH were detected in our dataset, most of which were significantly upregulated in the Granny Smith apple. Alcohol acyltransferase (AAT) is the key enzyme for the biosynthesis of esters [9]. Ten AAT DEGs exhibited higher expression levels in the Jonagold than that in the Granny Smith apple.
Straight-chain esters are synthesized from the fatty acid metabolism pathway while branched-chain esters are produced by the pathway of isoleucine metabolism [13]. Aldehyde dehydrogenase (ALDH) and carboxylesterase (CXE) are the two important enzymes for the synthesis of branched-chain esters. In our study, four DEGs encoding ALDH and eight DEGs encoding CXE were identified, most of which were upregulated in the Jonagold apple compared to the Granny Smith apple. In addition, some DEGs encoding acetolactic synthetase (ALS), acetohydroacid isomeroreductase (AHIR) and branched-chain aminotransferase (BCAT) also had enhanced mRNA levels in the Jonagold apple ( Figure  4; Table S5).  Straight-chain esters are synthesized from the fatty acid metabolism pathway while branched-chain esters are produced by the pathway of isoleucine metabolism [13]. Aldehyde dehydrogenase (ALDH) and carboxylesterase (CXE) are the two important enzymes for the synthesis of branched-chain esters. In our study, four DEGs encoding ALDH and eight DEGs encoding CXE were identified, most of which were upregulated in the Jonagold apple compared to the Granny Smith apple. In addition, some DEGs encoding acetolactic synthetase (ALS), acetohydroacid isomeroreductase (AHIR) and branched-chain aminotransferase (BCAT) also had enhanced mRNA levels in the Jonagold apple ( Figure 4; Table S5).

Expression Levels of Genes Related to Sesquiterpene and Phenylpropanoid Metabolism
Sesquiterpenes are the most prominent terpenoids in apples [28]. In our study, αfarnesene was the most abundant sesquiterpene accumulated in Granny Smith and Jonagold apples. A total of nine enzymatic steps are involved in the biosynthetic pathway of sesquiterpene (Figure 5a). Compared to the Granny Smith apple, two DEGs encoding HMG-CoA synthase (HMGS) and HMG-CoA reductase (HMGR) exhibited higher expression levels in the Jonagold apple (Table S6). In cytosol, isopentenyl diphosphate (IPP) and its allylic isomer dimethylallyl diphosphate (DMAPP) catalyzed by the enzyme farnesyl pyrophosphate synthase (FPPS) results in the formation of farnesyl pyrophosphate. The FPPS genes were significantly downregulated in the Granny Smith apple in comparison with the Jonagold apple. Additionally, the final step in sesquiterpene biosynthesis in apple is catalyzed by α-farnesene synthase (AFS). In this study, one DEG was identified as the AFS gene and showed a higher expression level in the Jonagold apple than that in the Granny Smith apple (Figure 5a).

Expression Levels of Genes Related to Sesquiterpene and Phenylpropanoid Metabolism
Sesquiterpenes are the most prominent terpenoids in apples [28]. In our study, αfarnesene was the most abundant sesquiterpene accumulated in Granny Smith and Jonagold apples. A total of nine enzymatic steps are involved in the biosynthetic pathway of sesquiterpene (Figure 5a). Compared to the Granny Smith apple, two DEGs encoding HMG-CoA synthase (HMGS) and HMG-CoA reductase (HMGR) exhibited higher expression levels in the Jonagold apple (Table S6). In cytosol, isopentenyl diphosphate (IPP) and its allylic isomer dimethylallyl diphosphate (DMAPP) catalyzed by the enzyme farnesyl pyrophosphate synthase (FPPS) results in the formation of farnesyl pyrophosphate. The FPPS genes were significantly downregulated in the Granny Smith apple in comparison with the Jonagold apple. Additionally, the final step in sesquiterpene biosynthesis in apple is catalyzed by α-farnesene synthase (AFS). In this study, one DEG was identified as the AFS gene and showed a higher expression level in the Jonagold apple than that in the Granny Smith apple (Figure 5a).
Estragole, which belongs to the phenylpropenes, is synthesized from the phenylpropanoid pathway in apples. In our study, a total of 14 DEGs were obtained in this pathway (Figure 5b; Table S6). At the beginning of this pathway, four DEGs encoding phenylalanine ammonia lyase (PAL) were upregulated in the Jonagold apple (Figure 5b). However, compared to the Granny Smith apple, the C4H, 4CL and CCoAR genes showed lower expression levels in the Jonagold apple. The final step in estragole biosynthetic pathway involves a class of methyltransferase. Three DEGs encoding O-methyltransferase (OMT) were identified in the dataset and all of them exhibited higher expression levels in the Jonagold apple compared with the Granny Smith apple.  Estragole, which belongs to the phenylpropenes, is synthesized from the phenylpropanoid pathway in apples. In our study, a total of 14 DEGs were obtained in this pathway (Figure 5b; Table S6). At the beginning of this pathway, four DEGs encoding phenylalanine ammonia lyase (PAL) were upregulated in the Jonagold apple (Figure 5b). However, compared to the Granny Smith apple, the C4H, 4CL and CCoAR genes showed lower expression levels in the Jonagold apple. The final step in estragole biosynthetic pathway involves a class of methyltransferase. Three DEGs encoding O-methyltransferase (OMT) were identified in the dataset and all of them exhibited higher expression levels in the Jonagold apple compared with the Granny Smith apple.

Expression Levels of Transcription Factors Associated with Volatile Compounds
Transcription factors (TFs) play many important roles in regulating gene expression in various plant biological processes including the production of volatile compounds. In this study, we found that 30 DEGs were putatively identified as TFs associated with NAC, MYB, bHLH, ERF, WRKY, bZIP and MADS families ( Figure S6). The NAC family was represented by five DEGs that were upregulated in the Jonagold apple. Moreover, five DEGs, belonging to the MYB family, were obtained and exhibited higher expression levels in the Jonagold apple than in the Granny Smith apple. In addition, some DEGs encoding ERF, WRKY and bZIP family TFs also showed higher expression levels in the Jonagold apple compared to the Granny Smith apple ( Figure S6).

Validation of Gene Expression Levels Using qRT-PCR
To validate the expression patterns of DEGs observed in RNA-seq datasets, 12 key genes were selected for analysis of their expression levels using qRT-PCR. These DEGs included LOX, HPL, ADH, AAT, BCAT, ALDH and CXE involved in fatty acid and isoleucine metabolism, AFS and OMT related to sesquiterpene and phenylpropanoid metabolism, transcription factors NAC1, MYB5 and bZIP44 that could regulate the production of volatile compounds. Except for ADH and ALDH, these DEGs all showed higher expression levels in the Jonagold apple than those in the Granny Smith apple, which were consistent with the RNA-Seq data ( Figure 6). These results confirmed that the transcriptomic data in our study were accurate and reliable.

Expression Levels of Transcription Factors Associated with Volatile Compounds
Transcription factors (TFs) play many important roles in regulating gene expression in various plant biological processes including the production of volatile compounds. In this study, we found that 30 DEGs were putatively identified as TFs associated with NAC, MYB, bHLH, ERF, WRKY, bZIP and MADS families ( Figure S6). The NAC family was represented by five DEGs that were upregulated in the Jonagold apple. Moreover, five DEGs, belonging to the MYB family, were obtained and exhibited higher expression levels in the Jonagold apple than in the Granny Smith apple. In addition, some DEGs encoding ERF, WRKY and bZIP family TFs also showed higher expression levels in the Jonagold apple compared to the Granny Smith apple ( Figure S6).

Validation of Gene Expression Levels Using qRT-PCR
To validate the expression patterns of DEGs observed in RNA-seq datasets, 12 key genes were selected for analysis of their expression levels using qRT-PCR. These DEGs included LOX, HPL, ADH, AAT, BCAT, ALDH and CXE involved in fatty acid and isoleucine metabolism, AFS and OMT related to sesquiterpene and phenylpropanoid metabolism, transcription factors NAC1, MYB5 and bZIP44 that could regulate the production of volatile compounds. Except for ADH and ALDH, these DEGs all showed higher expression levels in the Jonagold apple than those in the Granny Smith apple, which were consistent with the RNA-Seq data ( Figure 6). These results confirmed that the transcriptomic data in our study were accurate and reliable.

Discussion
Aroma is an important quality indicator of fruit flavor that can significantly influence the consumer choices [29]. Many factors can affect the aroma composition of apple fruit, among which cultivar is the most important [6]. In this study, volatile compounds of two famous apple varieties 'Granny Smith' and 'Jonagold' were determined using HS-SPME

Discussion
Aroma is an important quality indicator of fruit flavor that can significantly influence the consumer choices [29]. Many factors can affect the aroma composition of apple fruit, among which cultivar is the most important [6]. In this study, volatile compounds of two famous apple varieties 'Granny Smith' and 'Jonagold' were determined using HS-SPME with GC-MS. The types of volatile compounds in the Jonagold apple were more than that in the Granny Smith apple. Among these volatiles, esters were the predominant compounds in the Jonagold apple, accounting for more than 60% of the total volatiles, which was consistent with previous report that esters were the most abundant group that contribute mainly to the sweet and fruity odors of apple fruits [11,30]. Moreover, hexyl acetate, confers a sweet and fruity odor with floral notes, was the most plentiful ester in the Jonagold apple, which was in accordance with the results of previous studies [31]. The Granny Smith apple is known to have a very low volatile emission [1]. In the present study, despite the fact that the total content of volatile compounds in the Granny Smith apple was significantly lower than that in the Jonagold apple, Granny Smith possessed more types of alcohols and aldehyde, occupied over 90% of the total volatiles. For the Granny Smith apple, 1-hexanol was the most abundant compound among alcohols and (E)-2-hexenal was the major aldehyde. Additionally, a range of volatiles such as α-farnesene and estragole have been detected in the Granny Smith apple, but accumulated more in Jonagold apples. The aroma volatiles of various apple varieties were different in class and quantity, and gave a characteristic sensorial perception for each type of apple fruit [32]. Previous studies have classified Jonagold into the 'ester type' apple based on its high ester production [5,33]. According to this classification, the Granny Smith studied here would be the 'aldehyde type' apple because the aldehydes in its fruit represent more than 85% of the total aromatic components. The differences in composition and concentration of volatile compounds between Granny Smith and Jonagold apples may determine diverse aroma profiles of the fruits.
In order to reveal the molecular mechanisms underlying the synthesis of volatile compounds, transcriptome sequencing of the Granny Smith and Jonagold apples was performed using the Illumina Novaseq 6000. It has been reported that at least four pathways are responsible for the production of volatile compounds in apples [6]: Straight-chain aldehydes, alcohols and esters are synthesized from fatty acid metabolism pathway [13]; Branched-chain aldehydes, alcohols and esters are derived from isoleucine pathway [34]; Sesquiterpenes are synthesized via the mevalonate pathway and phenylpropenes are synthesized from the phenylpropanoid pathway [12,35]. In this study, a total of 94 DEGs were detected in these four pathways, indicating the dramatical differences in volatile synthesis between the Granny Smith and Jonagold apples. For the formation of aldehydes, alcohols and esters in apples, β-oxidation and the lipoxygenase pathway are the two essential enzymatic systems in the catabolism of fatty acids [36,37]. The β-oxidation of long-chain fatty acids produces shorter acids such as acetic, butanoic and hexanoic acids, which can be reduced to their corresponding alcohols before being esterified with acyl-CoA by the AAT enzyme [9]. The lipoxygenase pathway contains four vital enzymes, inclusive of LOX, HPL, ADH and AAT [36]. Lipoxygenase (LOX) is a dioxygenase that catalyzes the oxygenation of polyunsaturated fatty acids. In 'Royal Gala' and 'Prima' apples, the genes MdLOX1a and MdLOX5e were found to be involved in the production of volatiles [38]. In our study, the twelve MdLOX genes showed significantly higher expression levels in the Jonagold apple than that in the Granny Smith apple. Combined with the measurement results of volatile profiles, the upregulated expression pattern of MdLOX genes in the Jonagold apple may play a key role in the volatile compounds accumulation. Products of the LOX reaction can be converted to aldehydes by the hydroperoxide lyase (HPL) enzyme and aldehydes are subsequently reduced to the corresponding alcohol by the alcohol dehydrogenase (ADH) [12]. The alcohols resulting from ADH enzymatic activity are natural substrates for the alcohol acyltransferase (AAT), which transfers an acyl group through an oxygen-dependent reaction from acyl-CoA to the OH group of an alcohol forming an ester [39]. In our study, compared to the Granny Smith apple, the MdAAT gene exhibited significantly higher expression levels in the Jonagold apple, accounting for the high ester production. Amino acids are the second most important source of volatile compounds in the aroma of apples. A previous study has reported that the branched amino acids, especially isoleucine, can produce the branched-chain esters in apples [34]. In our dataset, the key gene MdBCAT, which catalyzes the last step in the synthesis of branched amino acids, was higher expressed in the Jonagold apple than that in the Granny Smith apple, and might provide more precursors for the synthesis of branched-chain aldehydes, alcohols, acids and esters. Sesquiterpenes, which synthesized via the mevalonate pathway, are the main aroma volatiles from the isoprenoid family in the apple fruits [9]. In our study, there was only one terpenoid compound α-farnesene (sesquiterpene) detected in Granny Smith and Jonagold apples. α-Farnesene synthase (AFS) catalyzed the final step in sesquiterpene biosynthesis in apples [40]. The MdAFS as well as the MdHMGS, MdHMGR and MdFPPS genes in this pathway were all upregulated in Jonagold apples compared with the Granny Smith apple, might be responsible for high production of α-farnesene. In addition, phenylpropenes, such as eugenol, estragole and isoestragole, also contribute to apple aroma and these compounds are derived from the phenylpropanoid pathway [17,30]. It has been reported that estragole imparts a spicy and aniseed note to 'Ellison Orange' and 'Royal Gala' apples [17]. In this study, the compound estragole was detected in both two apple varieties, the key genes MdPAL and MdOMT exhibited high expression levels in Jonagold apples, possibly accelerating the reaction of enzymes in estragole biosynthesis and resulting in the accumulation of estragole.
Transcriptional regulation mediated by transcription factors (TFs) is a pivotal control hierarchy for various plant biological processes, including the synthesis of volatile compounds [41,42]. A previous study has reported that FaMYB9 is positively involved in volatile biosynthesis in strawberries, transient silencing of FaMYB9 delayed the fruit development, resulting in a significant decrease in the contents of C6 volatiles [43]. Moreover, in banana, MabZIP4 and MabZIP5 were capable of binding directly to BanAAT promoter, suggesting the function of MabZIPs in controlling aroma production [42]. Additionally, the transcription factor PpNAC1 was proved to transcriptionally activate the PpAAT1 gene and formation of volatile esters in peaches [44]. AP2/ERF transcription factors have been shown to be associated with the production of volatile terpenes in sweet orange fruits [45]. In order to clarify potential transcription factors that may be involved in regulating volatile biosynthesis, expression patterns of TFs such as NAC, MYB, bZIP and ERF were analyzed. In the Jonagold apple, the MdNAC1, MdMYB5, MdERF8 and MdbZIP44 TFs had higher expression levels than in the Granny Smith apple. Combined with the aroma profiles of these two apple varieties, we speculate that these TFs can bind to the promoters of some key genes and may play important roles in regulating the synthesis of volatile compounds. Therefore, further functional studies will focus on clarifying the regulatory roles of these transcription factors in the biosynthesis of volatiles in apple.

Plant Materials and Sample Collection
'Granny Smith' and 'Jonagold' apple fruit samples were collected at the Baishui Apple Experimental Station of Northwest Agriculture and Forestry University in Shaanxi, China (35 • 21 N, 109 • 55 E). The two apple varieties were grafted onto M26 rootstock (Malus domestica) and planted at a density of 4 × 2 m. Unbagged apple fruits were harvested at the full ripening stage in 2020. Three biological replicates for each apple variety were prepared, with each biological replicate consisting of 8-10 fruits. The pulps of two apple varieties were carefully separated with the peels, pooled, immediately frozen in liquid nitrogen and then stored at −80 • C until use.

Determination of Fruit Physiological Characteristics
Measurements of apple fruit height and diameter were taken using a digital vernier caliper (Meinaite, Chengdu, China). Determinations of single fruit weight and skin color were performed by an electronic balance (MettlerToledo Inc., Greifensee, Switzerland) and a colorimeter (CR-400, Minolta, Japan), respectively. Test of apple flesh texture was conducted using a texture analyzer (FTA GS-15, Berlin, Germany; test depth 8 mm). Total soluble solid (TSS) and titratable acidity (TA) of apple fruits were measured with a hand refractometer (Atago, Tokyo, Japan) and a digital fruit acidity meter (GMK-835F Perfect, Berlin, Germany), respectively. The internal ethylene of the apples was evaluated according to the method described by Cai et al. [46] using a gas chromatograph (Agilent Technologies 7890A, Palo Alto, CA, USA) equipped with flame ionization detector and a HP-AL/S column. Measurement of CO 2 production was performed by the method described by Contreras and Beaudry [47].

Volatile Compounds Extraction and GC-MS Analysis
HS-SPME was used for the extraction of volatile compounds in apples. The pulps of apple fruits were ground into powder in liquid nitrogen. For each extraction, 5 g of powder was placed into a 50 mL screwcap vial containing 1.0 g of NaCl and a magnetic stirring rotor to facilitate the release of volatiles. Prior to sealing of the vials, 10 µL of 0.4 mg/mL 3-nonanone was added as internal standard. After the headspace vial equilibrated at 50 • C for 10 min on a metal heating agitation platform, the SPME fiber coated with 50/30 µm thickness of divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS, Supelco, Bellefonte, PA, USA) was inserted into the headspace with continuous heating and agitation (200 rpm) for 30 min to adsorb volatile compounds. Then, the fiber was introduced into the heated injector port of the chromatograph for desorption at 250 • C for 2.5 min.
The volatile compounds were analyzed with a Thermo Trace GC Ultra gas chromatograph (Thermo Fisher Scientific, Waltham, MA, USA), equipped with a 60 m × 0.25 mm × 0.25 µm HP-INNOWax capillary column. Helium was circulated as the carrier gas with a flow of 1.0 mL/min in a splitless mode. The initial oven temperature was kept at 40 • C for 3 min and increased to 150 • C at a rate of 5 • C/min, then increased at 10 • C/min to 220 • C and held for 5 min. The temperature of ion source and transfer line were both 240 • C. Mass spectra were operated in electron ionization mode of 70 eV with the scan range from m/z 35 to 450.
Identification of volatile compounds was based on mass spectra matching with the database of NIST/EPA/NIH Mass Spectral Library (NIST 2014), the retention times with an in-house developed retention time library based on commercial standards, and the retention indices with those of authentic compounds or literature data, which calculated under the same chromatographic conditions after the injection of a C7-C30 n-alkane series (Supelco, Bellefonte, PA, USA). Quantification was carried out by the internal standard method, as illustrated by Qin et al. [48], where the concentration of each volatile compound was normalized to that of 3-nonanone.

RNA Extraction, Library Preparation and Transcriptome Sequencing
Total RNA was isolated from the pulp of the two apple varieties (180 days after flowering) using the RNAprep pure Plant Kit (Tiangen, Beijing, China) according to the manufacturer's instructions. Quality and quantity of the total RNA were assessed using a Qubit 2.0 fluorometer RNA Assay Kit (Invitrogen Inc., Carlsbad, CA, USA) and Agilent 2100 Bioanalyzer (Agilent, Palo Alto, CA, USA). A total amount of 3 µg of the high-quality RNA per sample was used for the cDNA library preparation and 6 libraries (3 biological replicates) were constructed by NEBNext ® UltraTM RNA Library Prep Kit for Illumina ® (NEB, Ipswich, MA, USA). Transcriptome sequencing was performed using Illumina Novaseq platform (Illumina, San Diego, CA, USA) to paired-end reads by Novogene (Beijing, China). Quantitative real-time reverse transcription PCR (qRT-PCR) was performed with an iQ5 Multicolor Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA) using SYBR ® Green Master Mix (TaKaRa, Kyoto, Japan). Each sample was assessed in three biological repeats and normalized using apple MdActin and MdEF-1α genes as the internal control. The relative expression of the genes was analyzed by the 2 −∆∆CT method and the primers used for qRT-PCR were presented in Table S1.

Differential Expression Analysis and Functional Annotations
In order to obtain the clean reads, the raw reads were filtered by removing the adaptor, low-quality reads (Qphred ≤ 20 bases) and unknown sequences. Then, clean reads were mapped to apple reference genome (https://iris.angers.inra.fr/gddh13, accessed on 1 July 2017). Quantification of the gene expression levels was estimated as fragments per kilobase of transcript per million mapped reads (FPKM). Cufflinks (version 2.2.1) software was used in the process of quantifying gene expression, and the counts were then normalized to the FPKM values. DESeq R package (version 1.18.0) was used to identify the differentially expressed genes (DEGs) between the two apple varieties. A fold change (FC) value of ≥2 and a false discovery rate (FDR) of <0.05 were used as criteria for selecting DEGs. GOseq R package was used to analyze Gene Ontology (GO) enrichment. GO terms with corrected p-values < 0.05 were assigned as significantly enriched by DEGs. Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of DEGs was conducted using the KOBAS software (version 3.0).

Statistical Analysis
Data were presented as mean values of three biological replicates ± standard error (SE). Statistical analysis and one-way analysis of variance (ANOVA) were performed by SPSS v19.0 software (SPSS Ins., Chicago, IL, USA), and significant differences between the samples were assessed by the Duncan test (p < 0.05).

Conclusions
In this study, we investigated the volatile profiles of the Granny Smith and Jonagold apples and found that aldehydes were the most abundant types of volatile compounds contributing to the apple aroma in Granny Smith while esters were the main types of volatile compounds in the Jonagold apple. The analysis of transcriptome sequencing revealed 94 DEGs in the pathways of fatty acid metabolism, amino acid metabolism, mevalonate and phenylpropanoid. Compared with the Granny Smith apple, the key genes such as MdLOX, MdAAT, MdAFS, MdPAL and MdOMT, and the important transcription factors such as MdNAC, MdMYB, MdERF and MdbZIP all showed higher expression levels in the Jonagold apple, resulting in the accumulation of some esters as well as the α-farnesene and estragole, and might be responsible for the high production of volatiles. Furthermore, qRT-PCR experiments verified that those genes were differentially expressed between the two apple varieties. Production of aroma in apple fruits was the result of a combination of complex metabolic pathways, including the diverse physiological processes and molecular regulation mechanisms. Our study provides valuable clues to gain better insight into the aroma formation process of fruits.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The transcriptomic data generated in this study are submitted to NCBI Gene Expression Omnibus repository (http://www.ncbi.nlm.nih.gov/geo/, accessed on 7 March 2022) under accession number GSE176136.