Relationship Between Volatile Aroma Components and Amino Acid Metabolism in Crabapple (Malus spp.) Flowers, and Development of a Cultivar Classification Model

Abstract

The integration of HS-SPME-GC/MS and UPLC-MS/MS techniques enabled the profiling of volatile organic compounds (VOCs) and amino acids (AAs) in 18 crabapple flower cultivars, facilitating the development of a novel VOC–AA model. Among the 51 identified VOCs, benzyl alcohol, benzaldehyde, and ethyl benzoate were predominant, categorizing cultivars into fruit-almond, fruit-sweet, and mixed types. The amino acids, namely glutamic acid (Glu), asparagine (Asn), aspartic acid (Asp), serine (Ser), and alanine (Ala) constituted 83.6% of the total AAs identified. Notably, specific amino acids showed positive correlations with key VOCs, suggesting a metabolic regulatory mechanism. The Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA) model, when combined with volatile organic compounds (VOCs) and amino acid profiles, enabled more effective aroma type classification, providing a robust foundation for further studies on aroma mechanisms and targeted breeding.

1. Introduction

The genus Malus (crabapple) is an important ornamental group in the Rosaceae family, widely cultivated for its floral fragrance [1]. Floral scent is a key trait influencing the ornamental value and ecological interactions of flowering plants [2]. Among the biochemical precursors of floral volatile organic compounds (VOCs), amino acids (AAs) are known to play essential roles in the biosynthesis of various aromatic compounds [3,4]. However, the relationship between floral scent profiles and amino acid metabolism in Malus remains poorly understood. This study examines the association between flower scent and amino acid composition in various Malus cultivars, offering preliminary insights into potential biochemical influences on floral scent differences. In recent years, the rapid advancements in metabolomics have increasingly illuminated the mechanisms underlying floral scent production and their interactions with primary metabolites, positioning these as burgeoning areas of scientific inquiry [5]. Floral scents are predominantly composed of VOCs, with petals serving as the principal source of emission, thereby contributing substantially to the VOC profile [6]. These aromatic components are not only pivotal in plant ecology by attracting pollinators such as bees, butterflies, and bird vectors, thereby enhancing pollination success and reproductive fitness [7], but also hold significant value in the horticulture, food, and fragrance industries [4]. In particular, their unique fragrances are exploited in the breeding of ornamental plants, the manufacture of high-end perfumes, the enhancement of food flavors, and the development of natural medicines [8,9,10]. Regarding their biosynthetic pathways, floral VOCs are categorized into three major classes: terpenoids, benzenoids/phenylpropanoids, and fatty acid derivatives [11]. Terpenoids, as the most abundant volatile component, play a critical role in plant-environment interactions, particularly in attracting pollinators [12]. Benzenoid/phenylpropanoid compounds, representing the second largest class of volatiles, are vital secondary metabolites [4]. Meanwhile, volatile fatty acid derivatives contribute distinct scent characteristics and perform multiple roles in plant–environment interactions, often in response to biotic stress [13]. Thus, floral VOCs demonstrate complex ecological interactions, providing both adaptive functions, such as stress resistance and defense, and a wide range of applications in fragrances, food products, and pharmaceuticals. Nonetheless, the metabolic regulatory networks and coordinated functions that underpin these volatiles require further elucidation.

The crabapple is celebrated for its splendid blossoms, and numerous fragrant varieties have been cultivated successfully through both natural selection and artificial breeding [14]. Breeding programs have largely concentrated on ornamental characteristics, including flower shape, color, and fragrance [15]. In response to the increasing demand for fragrant ornamentals, breeders have given priority to cultivars with pronounced aromas during the selection process [16,17]. Current research suggests that the diversity of VOCs in crabapple flowers forms the basis of their unique scent profiles, predominantly comprising alcohols, aldehydes, esters, and terpenoids [18]. Significant variances in VOC composition are evident among different cultivars. For instance, in the ornamental crabapple cultivar Malus ‘Lollipop’, esters, benzenoids, and alkanes are the primary components [17], while in Malus ioensis ‘Prairie Rose’, benzyl alcohol, lilac aldehyde, and benzaldehyde contribute distinct fragrances to different crabapple varieties [14]. Although these studies have elucidated the chemical foundations of the crabapple scent, they have generally focused on individual VOC components and lacked comprehensive multi-metabolite comparative analyses, such as those integrating AAs or anthocyanins with VOCs, across different cultivars. This gap has resulted in a partial understanding of the aroma germplasm diversity in crabapple, thus limiting the targeted enhancement of fragrance traits and the development of innovative cultivars.

Previous studies have documented associations between AA levels and the intensity of floral scent. For example, alanine (Ala) and serine (Ser) are known to produce 2-methylpropanal, noted for its sweet aroma, via Strecker degradation [19], while aspartic acid (Asp) and glutamic acid (Glu) result in acidic and umami substances through decarboxylation reactions [20]. Recently, an advanced analytical platform utilizing ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) equipped with multiple reaction monitoring (MRM) technology has been developed. This platform has enabled researchers to quantitatively determine the dynamic interplay between AA variations and the accumulation of ester aroma compounds during the fermentation of rice wine, revealing that valine (Val) enhances the production of ethyl acetate [21]. In the daylily (Hemerocallis spp.), a marked correlation between the metabolic flux of phenylalanine (Phe) and the synthesis of benzaldehyde has underscored the regulatory role of AA metabolic networks in defining characteristic floral scents [1]. Despite these advancements, research on the interaction mechanisms between floral VOCs and AAs is still limited, particularly in systematic multi-cultivar studies of woody ornamental plants such as crabapple, which remain relatively rare.

In this study, we constructed a predictive interaction network model between AAs and VOCs in crabapple flowers. By integrating headspace solid-phase microextraction coupled with gas chromatography-mass spectrometry (HS-SPME-GC/MS) and UPLC-MS/MS, we systematically analyzed the volatile compound and AA profiles of 18 crabapple cultivars. We established an aroma phenotype classification model based on odor activity values (OAVs) and validated metabolic differences between groups through hierarchical cluster analysis (HCA). A VOC–AA network model was developed for crabapple flowers using Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA). This model elucidated the metabolic basis of crabapple floral scent formation and yielded an aroma-type prediction model based on VOCs and AA data. This multi-omics framework provides new insights into dissecting the molecular mechanisms of crabapple flower fragrance, facilitates directed breeding for scent, and offers a basis for developing functional horticultural products (e.g., floral teas, aromatic essential oils).

2. Materials and Methods

2.1. Plant Materials

Eighteen cultivars of crabapple (Malus spp.) were selected for this study, all of which were cultivated at the Hebei Forestry Research Institute (Hebei Academy of Forestry Sciences, Shijiazhuang, China) (Figure 1).

2.2. Chemicals and Standards

The chemicals used included acetonitrile (HPLC grade), tryptophan (Trp), tyrosine (Tyr), citrulline (Cit), arginine (Arg), Phe, histidine (His), methionine (Met), Glu, glutamine (Gln), lysine (Lys), Asp, asparagine (Asn), leucine (Leu), isoleucine (Ile), cysteine (Cys), threonine (Thr), Val, proline (Pro), Ser, gamma-aminobutyric acid (GABA), Ala, and glycine (Gly). These were obtained from the Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). The AA solutions were prepared using ultrapure water (Wahaha Group Co., Ltd., Hangzhou, China). Both C8-C30 n-alkane standards and 4-methyl-2-pentanol standards were purchased from Merck KGaA, Darmstadt, Germany, and prepared in ethanol (Sinopharm Chemical Reagent Co., Ltd., Shanghai, China) solutions.

2.3. Sensory Evaluation

During the flowering period of the crabapple, samples were collected on clear days from 8:00 AM to 11:00 AM between March and April 2024. Flower samples from each cultivar were randomly labeled and transported to the laboratory, where they were placed in identical containers filled with room-temperature water for further processing. Ten adult panelists (aged 20–40, healthy, without adverse habits or history of pollen allergy) participated in the sensory evaluation. All participants agreed to refrain from exposure to strong odors and the use of scented products during the assessment period and to maintain a calm state before evaluation.

Each sample’s fragrance was scored on a scale of three intensity levels: no/light fragrance (1 < score ≤ 5), medium fragrance (5 < score ≤ 8), and strong fragrance (8 < score ≤ 10). The panelists evaluated the samples sequentially; after assessing three samples, each participant took a 1–2 min break to avoid sensory fatigue. The final score for each cultivar was the average of all panelists’ scores, with outliers being excluded using Dixon’s Q test.

2.4. Collection and Determination of Volatile Compounds

The analysis of volatile compounds was conducted employing HS-SPME-GC/MS, adhering to the method described by Qin et al. [22] with slight modifications.

Volatile compounds in crabapple cultivar samples were extracted using headspace solid-phase microextraction (HS-SPME) with a divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS) fiber assembly (Agilent, Santa Clara, CA, USA), followed by analysis with GC–MS. The analyses were carried out using an Agilent 7890B gas chromatograph equipped with a 5977A mass selective detector, a CTC PAL autosampler, and an HP-Innowax column (30 m × 0.250 mm I.D., 0.5 μm film thickness). The oven temperature program was set as follows: 50 °C for 2 min, then increased at 3 °C/min to 220 °C, followed by 8 °C/min to a final temperature of 235 °C. Helium was used as the carrier gas at a flow rate of 1 mL/min.

Sample preparation was performed as follows. A total of 0.5 g of fresh flower sample was placed into a 20 mL headspace vial, followed by the addition of 2 g of sodium chloride and 5 mL of 10% (v/v) anhydrous ethanol. Then, 5 μL of 2000 ppm 4-methyl-2-pentanol solution was added as the internal standard, resulting in a final concentration of 2 mg/L in the vial. In this study, 4-methyl-2-pentanol was selected as the internal standard for GC–MS analysis. This compound exhibits good chemical and thermal stability, moderate volatility, and stable mass spectrometric response. Preliminary tests confirmed that it was not detected in the volatile components of crabapple flowers, and its retention time did not overlap with the main aroma compounds. Therefore, it was considered suitable as an internal standard for this analysis.

Volatile compounds were identified by matching mass spectra with the NIST14.0 library and comparing retention indices (RI). Semiquantitative analysis was conducted using 4-methyl-2-pentanol as the internal standard.

To eliminate potential background VOCs originating from solvents or instrumentation, a blank control was included in the analysis. The blank consisted of 2 g of sodium chloride and 5 mL of 10% (v/v) anhydrous ethanol, with 5 μL of 2000 ppm 4-methyl-2-pentanol added as internal standard. The same HS-SPME–GC–MS procedure was applied to the blank to subtract any interfering signals from the final dataset.

The odor descriptions of the volatile compounds were based on references [23,24]. The odor thresholds in water primarily referenced Van Gemert and the online database (http://www.vcf-online.nl/VcfHome.cfm, accessed on 20 August 2024). OAVs were calculated as follows: OAVs = concentration of each compound (mg/L) divided by the odor threshold for that compound in water (mg/L).

2.5. Determination of AA Content

For the analysis of AAs, 1 g of fresh samples was transferred to a 50 mL centrifuge tube containing 15 mL of a 70% ethanol extraction solution. The mixture underwent ultrasonication for 5 min at 210 W, followed by centrifugation at 10,000× g rpm for 10 min. Subsequently, 1 mL of the supernatant was filtered through a 0.22 μm microporous membrane. AAs were separated through pre-column derivatization and analyzed using HPLC on an Agilent 1290 system, fitted with an Agilent Zorbax SB-C18 Rapid Resolution HT column (3.0 mm × 50 mm, 1.8 μm). This analysis was conducted in accordance with the method outlined by Wang et al. [25]. The mobile phase comprised a 25 mM acetate buffer (pH 5.8) containing 0.02% sodium azide (Phase A) and 100% acetonitrile (Phase B). The flow rate was maintained at 0.6 mL/min. The gradient elution program was meticulously structured as follows: an initial 6% B was held for 3 min; B was then increased from 6% to 14% over 7 min and held at 14% for 3.5 min; subsequently, B was raised from 14% to 19.5% over 2 min, then to 20% over 2.5 min, and held for 1 min. The gradient continued with an increase from 20% to 26% B over 1 min, from 26% to 30% B over 3 min, and from 30% to 50% B over 1 min. Afterward, B was raised from 50% to 63% over 1 min and held for 1.5 min; it was then increased to 100% B in 1 min and maintained for an additional 1 min. Following this, the column underwent washing and reconditioning processes. The detailed gradient profile was as follows: 0–3 min, 5% B; 3–6 min, 20% B; 6–8 min, 30% B; 8–10 min, 40% B; 10–15 min, 50% B; 15–17 min, 5% B. The injection volume was set at 2 μL. Detection was performed using a photodiode array at 280 nm, and the column temperature was controlled at 16 °C. AAs were identified by comparing their retention times with those of reference standards. Standard curves were prepared, and quantitative analysis of the samples was executed following the methodology proposed by Wang et al. [25]. All samples were analyzed in triplicate.

2.6. Statistical Analysis

Data management and analysis were facilitated using Agilent ChemStation software (version G1701FA) for VOC peak integration and quantification, while Microsoft Excel 2010 served to organize and classify aroma components. Statistical procedures and graphical representations were conducted using SPSS 26.0 and Origin 2022, respectively. Heatmap clustering of AA data was performed using Heml 1.0, and OPLS-DA was executed with SIMCA 14.1. All experiments were conducted in triplicate to minimize experimental variability. Results are expressed as mean ± standard deviation (SD).

3. Results

3.1. Classification of Volatile Component

Sensory evaluation of 18 crabapple cultivars facilitated their classification into three groups based on aroma intensity: strong fragrance (five cultivars, scores ranging from 8 to less than 10), medium fragrance (seven cultivars, scores ranging from 5 to less than 8), and light or no fragrance (six cultivars, scores ranging from 1 to less than 5) (see Figure 2A and Table S1).

To elucidate the metabolic foundations underlying these differences in aroma, the VOCs present in the 18 crabapple cultivars were analyzed using HS-SPME-GC/MS. This analysis led to the identification of 51 VOCs, as detailed in Table S2. Notably, there were significant variations in the distribution of compound classes among the cultivars. Esters constituted the most prevalent class of volatiles, comprising 21 compounds and accounting for 43.14% of the total. This was followed by aldehydes (six compounds, 11.76%), alcohols (five compounds, 9.80%), alkanes (five compounds, 9.80%), terpenoids (three compounds, 5.88%), acids (four compounds, 7.84%), and other unclassified compounds (seven compounds, 13.73%). These findings offer a comprehensive overview of the types and relative abundances of floral scent components in crabapples. The sulfur-containing compound timonacic was identified in crabapple species, and is known for its potential antioxidant, stress-resistance, anti-tumor, and anti-inflammatory properties [26]. These attributes render timonacic a promising candidate for natural drug development, thereby expanding the chemical diversity of secondary metabolites in crabapple species and furnishing novel insights for future research into functional compounds.

Figure 2B illustrates significant variations in the types of volatile components among the 18 crabapple flowers. Specifically, ‘Royalty’ (27 types), ‘Spring Snow’ (23 types), and ‘Harvest Gold’ (23 types) exhibited the highest counts of volatile components, surpassing the group average (21.4 ± 3.2 types) by 26.2%, 7.5%, and 7.5%, respectively. Notably, an acid compound was exclusively detected in ‘Harvest Gold’, ‘Flame’, and ‘Royalty’, while absent in the other 15 cultivars. In contrast, ‘Sparkler’ and ‘Indian Magic’ presented the fewest volatile components, each with only 17 types, which was 20.6% lower than the group average. Further analysis revealed that although ‘Sparkler’ and ‘Indian Magic’ had fewer ester components (six to seven types), their relative proportions (35.3–41.2%) were comparable to those in cultivars with a higher diversity of volatiles (33.3–39.1%). This suggests that the core metabolic pathways for ester synthesis might be conservatively regulated across different cultivars.

3.2. Analysis of Major Volatile Components in Different Cultivars

In this study, only 25 compounds, each with a relative content exceeding 0.1 mg/L, were selected for detailed analysis. Collectively, these compounds represented 99.11% of the total volatile profile, underscoring their significance as the principal constituents. The major compounds included nine esters, five alcohols, four alkanes, three aldehydes, and three terpenoids, as detailed in Table S3. Quantitative assessments (refer to Figure 3A,B) revealed that the cultivars ‘Royal Raindrops’, ‘Profusion’, and ‘Prairifire’ exhibited the highest total volatile contents, measuring 250.06 mg/L, 86.50 mg/L, and 80.93 mg/L, respectively. These elevated levels correlate with the cultivars being perceived as highly fragrant, suggesting that variations in volatile content significantly influence the olfactory distinctions among different cultivars.

The distribution of aldehydes showed considerable variability among the cultivars, with over half (10 out of 18) displaying a substantially higher total aldehyde content compared to others. Notably, ‘Royal Raindrops’ demonstrated the peak aldehyde concentration at 197.02 mg/L. Similarly, the ester content varied significantly across the cultivars, with ‘Royal Raindrops’ also presenting the highest ester concentration at 40.82 mg/L, while ‘Sparkler’ recorded the lowest at 5.79 mg/L. All cultivars exhibited detectable levels of alcohol, maintaining a 100% detection rate. The coefficient of variation for total alcohol content was significantly lower (CV = 36.20%) than that for esters (CV = 67.67%), indicating a more conserved nature of alcohol biosynthesis pathways among the cultivars. A key validation step confirmed that the overall trends in total volatile compound contents were generally consistent with the sensory evaluation scores as illustrated in Figure 3C.

3.3. Odor-Based Classification of Volatiles and OAV Analysis

The odor characteristics of the major volatile compounds facilitated the classification of the aromas from 18 crabapple cultivars into six scent note groups: floral, fruity-almond, fatty, fruity-sweet, woody, and spicy. Figure 4A,B depict the total content and relative proportions of each volatile class across the cultivars. The scent categories of fruity-almond, fruity-sweet, and woody emerged as the most dominant, collectively comprising over 77% of the total volatile composition among the cultivars.

Heatmap analysis (Figure 4C) revealed distinct, cultivar-specific patterns in the concentrations of 25 principal aroma compounds. For instance, benzaldehyde and ethyl benzoate were found in significantly higher concentrations in ‘Royal Raindrops’, whereas nonadecane was notably enriched in ‘Royalty’. Linalool was present at elevated levels in ‘Purple Classic’, and benzyl benzoate was detected exclusively in ‘Profusion’ and ‘Royalty’. Ethyl hexadecanoate was ubiquitous across all cultivars, showing no significant variation in concentration. It is crucial to recognize that the presence of a volatile compound does not necessarily correlate with its olfactory impact, as the odor threshold plays a pivotal role in scent perception. The Odor Activity Value (OAV) is commonly employed to quantify the contribution of aromas; typically, an OAV greater than 10 signifies a major aroma contributor, while an OAV above 1 suggests a contributing component [27]. Table S4 indicates that benzaldehyde displayed an exceptionally high OAV in certain cultivars. In ‘Royal Raindrops’, the OAV for benzaldehyde reached 39,387.75, with similarly high values in ‘Prairifire’ (11,718.42) and ‘Profusion’ (13,887.90), indicating a substantial influence on the aromas of these cultivars. Similarly, ethyl benzoate demonstrated high OAVs across multiple cultivars, peaking at 1917.22 in ‘Royal Raindrops’. Although benzyl alcohol displayed a lower OAV in comparison, it still achieved 171.94 in ‘Royal Raindrops’, significantly surpassing the threshold for a key aromatic contribution (OAV > 10). These three aromatic compounds—benzaldehyde, ethyl benzoate, and benzyl alcohol—are notably prevalent in the most fragrant cultivars, highlighting their essential role in differentiating aroma profiles and establishing them as crucial chemical markers of crabapple floral scent. In contrast, compounds such as alkanes (e.g., heneicosane) are commonly found among cultivars (Figure 4C); however, their extremely high odor thresholds result in minimal contributions to the overall scent [28]. Consequently, they are not considered key aroma constituents.

3.4. Classification of Crabapple Cultivars Based on Fragrance Profiles

Utilizing the OAVs of major volatile compounds and the HCA applied to these OAVs (as depicted in Figure 5A,B), the fragrance phenotypes of 18 crabapple cultivars were categorized into three primary scent types: fruit-almond, fruit-sweet, and mixed. Below, we describe the key chemical markers and representative sensory attributes associated with each aroma type.

• Fruit-Almond Type: This classification includes the cultivars ‘Tamlin’, ‘Harvest Gold’, ‘Donald Wyman’, ‘Prairifire’, ‘Royal Raindrops’, ‘Indian Magic’, ‘Profusion’, ‘Red Jewel’, and ‘Adams’. The predominant aromatic driver in these cultivars is benzaldehyde, which constitutes over 80% of the total OAV. Benzaldehyde imparts an almond-like scent, stemming from its molecular structure that combines a benzene ring with a formyl substituent. This compound is prevalent in plants of the Rosaceae family and is typically produced through the deamination of Phe via the shikimate pathway [20]. The scent profile of this group aligns closely with that of the tropical water lily Nymphaea ‘Eldorado’ at full bloom, which is also dominated by benzaldehyde [29].
• Fruit-Sweet Type: This group comprises ‘Kelsey’, ‘Radiant’, and ‘Purple Prince’. Their fragrance is primarily characterized by ethyl benzoate, accounting for over 65% of the total OAV. Ethyl benzoate delivers a distinct fruity sweet aroma. Biologically, the synthesis of ethyl benzoate is generally facilitated by BAHD acyltransferases, such as LoAAT1, which utilize benzoyl-CoA and ethanol as substrates to form this ester aroma molecule [30].
• Mixed Type: The cultivars ‘Purple Classic’, ‘Flame’, ‘Toringo’, ‘Tang-2’, ‘Sparkler’, ‘Spring Snow’, and ‘Royalty’ fall under the mixed type category. The fragrance of these cultivars is defined by a co-dominance of benzaldehyde, ethyl benzoate, and benzyl alcohol, with their contributions to the OAV being relatively balanced. This results in a complex and layered aroma profile. These cultivars are often referred to as having a “mixed” fragrance type [31], displaying diverse and challenging-to-classify sensory characteristics during evaluations.

To explore variations in aroma characteristics among three distinct aroma types: fruit-almond, fruit-sweet, and mixed, OPLS-DA was applied to the VOC composition data derived from 18 cultivars. The analysis employed the relative concentrations of 11 key aroma compounds (VIP > 1) as predictor variables, with aroma type serving as the response variable. The model demonstrated robust goodness-of-fit and predictive accuracy, evidenced by an R²Y of 0.608 and a Q² of 0.423 (Figure 6A). To ensure model reliability, 200 permutation tests were conducted. The results, as depicted in Figure 6B, showed that the Q² regression line’s intercept on the vertical axis was below zero, confirming that the model was not subject to overfitting and thus validating its utility for discriminating among aroma types.

The OPLS-DA score plot (Figure 6A) delineated a clear demarcation among the three groups of aroma types within the principal component space, highlighting systematic differences in VOC profiles across the cultivars. Cultivars characterized by the fruit-almond aroma type generally occupied the positive side of the first principal component (t [1] > 0). This clustering pattern indicates their association with higher benzaldehyde content. Cultivars characterized by the fruit-sweet aroma type were primarily located in the upper region of the second principal component (t [2] > 1), suggesting their aroma profiles were more strongly influenced by ethyl benzoate. Notably, cultivar 17 (‘Adams’) exhibited an intermediate position near the boundary between fruit-almond and mixed aroma types. Although it contained a relatively high level of ethyl benzoate (17.81 mg/L), its benzaldehyde content (33.52 mg/L) was substantially higher, leading to its initial classification into the fruit-almond aroma type. The intermediate placement of cultivar 17 suggests potential metabolic characteristics shared with the mixed aroma type, warranting further analysis. Cultivars of the mixed aroma type were primarily located on the negative side of the first principal component (t [1] < −0.5) and were characterized by the concurrent presence of benzyl alcohol, ethyl benzoate, and benzaldehyde. This distribution pattern suggests possible substrate competition or metabolic crosstalk among these compounds within mixed-type cultivars, indicating the simultaneous activity of multiple biosynthetic pathways [32].

The loading plot (Figure 6C) clearly illustrates the contributions of individual volatile components to aroma-type discrimination among fruit-almond (F-A), fruit-sweet (F-S), and mixed cultivars. Specifically, benzaldehyde exhibited a positive loading along the first predictive component (p [1] = 0.25), indicating a significant contribution to the classification of fruit-almond type cultivars. In contrast, ethyl benzoate showed a similar directional trend along p [1] but with a lower loading value, suggesting a relatively weaker influence. Compounds such as nonadecane and ethyl myristate, positioned near the center along the p [1] axis but displaced vertically, appear to have stronger contributions to differentiation along the second predictive component (p [2]). Moreover, volatile components such as diisoamyl, hexacosane, and (E)-3-hexen-1-ol, which cluster towards negative values along p [1], primarily contributed to the aroma profiles characteristic of mixed-type cultivars. These results indicate that benzaldehyde plays an essential role in defining the fruit-almond aroma, while different sets of volatile compounds drive the aroma characteristics of fruit-sweet and mixed types.

3.5. AA Analysis

Quantitative profiling of AAs in 18 crabapple flower cultivars using UPLC-MS/MS revealed variations in total AA content across cultivars, as depicted in Figure 7A,B. The cultivar ‘Kelsey’ exhibited the highest total AA concentration at 110.00 mg/L, whereas ‘Radiant’ reported the lowest at 30.90 mg/L. A detailed component analysis identified Glu, asparagine Asn, Asp, Ser, and Ala as the five predominant AAs, which collectively constituted 83.57% of the total AA content (refer to Table S5).

To investigate the differential AA profiles across various aroma types: fruit-almond, fruit-sweet, and mixed, an OPLS-DA model was developed. This model employed the relative AA contents as predictor variables and aroma type as the response variable, as illustrated in Figure 7C. The classification model achieved a reasonable fit (R²Y = 0.631) and predictive capability (Q² = 0.520). The reliability of the model was further affirmed through 200 permutation tests, which indicated no evidence of overfitting as the Q² intercept in the permutation tests was below zero, thereby validating the model’s efficacy in distinguishing between aroma types based on AA profiles, as shown in Figure 7D.

In the corresponding OPLS-DA score plot for AAs, distinct distribution patterns were observed among the three aroma-type groups within the principal component space. Samples characterized by a fruit-almond aroma were dispersed across multiple clusters ranging from t [1] (−0.5 to 4) and t [2] (−2 to 11), reflecting a high degree of metabolic heterogeneity within this group. Conversely, fruit-sweet samples predominantly clustered towards the lower end of t [1] (−2 to 0) and spanned a moderate to high range on t [2] (1 to 4), indicating greater homogeneity within this group. ‘Tamlin’ and ‘Radiant’ exhibited similar t [1] values but showed some separation along the t [2] axis. Although they belong to different aroma types, this distribution pattern may reflect partial metabolic similarities. However, the possibility of sample size effects cannot be excluded.

3.6. Correlation Analysis

To systematically explore the potential synergistic relationships between AAs and volatile aroma compounds, a correlation heatmap was constructed. This heatmap included 11 aroma compounds identified with VIP scores greater than 1, and 21 AAs, as depicted in Figure 8A. This visual representation facilitated an understanding of the interrelationships among these compounds. The analysis revealed significant correlations between specific metabolites. For instance, diisoamyl demonstrated an extremely significant positive correlation with Cit (p < 0.001). Similarly, benzaldehyde was significantly positively correlated with Leu (p < 0.01) and showed positive correlations with Phe, Lys, and His (p < 0.05). Nonadecane exhibited positive correlations with Gln and Asn (p < 0.05). Ethyl benzoate displayed positive correlations with both Phe and Leu (p < 0.05). Additionally, ethyl palmitate was positively correlated with Asn (p < 0.05) but negatively correlated with Lys and Pro (p < 0.05). Conversely, (E)-3-hexen-1-ol showed a significant negative correlation with Asp (p < 0.01). These findings imply that certain AAs may act either as substrates or as regulatory factors in the biosynthesis of characteristic aroma compounds, thus influencing the formation of distinct aroma profiles.

The differences among the three aroma types—fruit-almond, fruit-sweet, and mixed—were further examined through the integration of AA and aroma compound profiles using OPLS-DA-based multivariate modeling (Figure 8B). In this model, aroma compounds with VIP values greater than 1 and the relative content of various AAs served as independent variables, while aroma type was the dependent variable. The integrated model exhibited excellent fit and predictive capabilities (R²Y = 0.942, Q² = 0.732). The OPLS-DA score plot demonstrated a clear separation of the three aroma types within the principal component space, illustrating classification boundaries that were notably superior to those of single-variable models. The stability of the model was further confirmed by 200 permutation tests (Figure 8C), which indicated that the Q² regression line of the permutation model intersected the vertical axis at a negative value, thereby confirming the absence of overfitting and validating the model’s reliability for discriminating aroma types in crabapple.

The incorporation of AA variables did not substantially alter the distribution boundaries among the three aroma types within the principal component space; however, it did enhance the intra-group clustering of samples. These results suggest that AAs may play a role in differentiating aroma types by enhancing metabolic stability. The findings underscore a potential cooperative regulatory role of AAs in the formation of crabapple flower aroma profiles.

4. Discussion

In this study, we systematically analyzed the underlying mechanisms responsible for differences in aroma formation among crabapple flowers. Employing a combination of HS-SPME-GC/MS and UPLC-MS/MS technologies, and focusing on both VOCs and AAs, we constructed an OPLS-DA-based model that enabled the precise classification of aroma phenotypes. Crabapple cultivars were categorized into three primary aroma types—fruit-almond, fruit-sweet, and mixed—based on their volatile compound profiles.

HS-SPME-GC/MS, renowned for its high sensitivity and resolution, is extensively utilized for capturing and analyzing trace floral VOCs [33,34]. Utilizing this technique, we detected a total of 51 volatile compounds in crabapple flowers, encompassing six classes of metabolites, including monoterpenoids, phenylpropanoids, and fatty acid derivatives. This diversity indicates a rich chemical foundation for the scent of crabapple flowers. Compared to some highly fragrant Rosaceous plants, such as roses [35], the aroma intensity of crabapple flowers was significantly lower, likely due to a lesser proportion of high-odor-impact components in their VOC profiles. Notable differences in VOC composition were observed among the cultivars: ‘Royalty’ exhibited the highest diversity, with 27 identified compounds, while ‘Sparkler’ and ‘Indian Magic’ each had the fewest, with 17 compounds. This variability suggests a considerable degree of diversity in the secondary metabolic networks among crabapple cultivars. Notably, a sulfur-containing metabolite, timonacic, was consistently detected across all crabapple samples. Despite its relatively low content among the VOCs, timonacic was found in 100% of the samples. This consistency suggests that timonacic may be derived from a conserved sulfur metabolic pathway in crabapple species, pointing to new avenues for future research on its ecological adaptations and potential functions [36]. Previous studies have indicated that timonacic possesses strong antioxidant and anti-tumor properties [37], suggesting its involvement in the environmental stress responses of crabapples and its value in research for development as a potential functional ingredient [38]. This discovery provides new insights into the relationship between characteristic metabolites and plant adaptability, as well as the multifunctional utilization of natural product resources.

The odor-active compounds are pivotal in determining the scent profile of a plant, and the approach based on odor thresholds is a fundamental strategy for identifying these compounds. This approach finds extensive application in the realms of food aroma research and the study of floral scents [39]. In the current study, employing a screening criterion of a relative concentration greater than or equal to 0.1 mg/L, a total of 25 characteristic VOCs were identified, which accounted for 99.11% of the total VOC content. Through a systematic analysis using OAVs, 19 key aroma-active compounds were identified (OAV > 1). Among these, benzyl alcohol, benzaldehyde, and ethyl benzoate, with OAVs surpassing 100, collectively contributed to 98% of the total aroma intensity and were recognized as the primary contributors to the characteristic scent of crabapple flowers. Notably, some components, though present in minor quantities, demonstrated significant OAVs [40], such as ethyl cinnamate (OAV = 25.53) and ethyl caprylate (OAV = 10.40), indicating that a “low concentration–high contribution” mechanism plays a critical role in the aroma formation of crabapple flowers, particularly in imparting sweet notes [41]. Previous research has shown that the characteristic aroma components of the chrysanthemum included isocyclocitral, eucalyptus alcohol, α-pinene, β-farnesene, and caryophyllene [42], while Rhododendron fortunei was predominantly characterized by eucalyptol and β-myrcene [43]. In contrast, the aroma profile of spp. is markedly different, being primarily dominated by phenylpropanoid-derived compounds such as ethyl benzoate and benzaldehyde. These aromatic substances are widely distributed in plants of the Rosaceae family, such as lilac [44] and peony [45], and are considered key aroma compounds that contribute to sweet and almond-like notes. This observation provides valuable insights into the comparative study of aroma evolution across different plant families and genera. The analysis revealed that certain cultivars exhibited unique olfactory traits. For instance, ‘Purple Classic’ was characterized by a lavender-like scent, largely due to the dominance of geraniol, whose biosynthetic pathway might be closely associated with the phylogenetic conservation of the species [28]. Building on these findings, this study developed an aroma classification system based on the sensory attributes of VOCs. The 25 core VOCs were categorized into six aroma groups: floral, almond-fruity, sweet-fruity, fatty, woody, and spicy. A weighted aroma model was constructed using contributions from OAVs as weighting factors. Ultimately, the 18 cultivars were classified into three main aroma types: almond-fruity, sweet-fruity-floral, and multi-component blended.

During the process of secondary metabolism in plants, AAs function as crucial precursor molecules [46]. They are transformed into various volatile aromatic compounds, including aldehydes, alcohols, and esters, through reactions such as decarboxylation, deamination, and transamination. The pivotal role of AAs in the biosynthesis of flavor compounds is well-documented in food chemistry research [47]. In this study, we employed UPLC-MS/MS to systematically quantify free AAs in the flowers of 18 crabapple cultivars. Our findings revealed a pronounced enrichment of Glu, Asn, Asp, Ser, and Ala, which collectively accounted for 83.6% of the total AA content. These AAs are engaged in key physiological processes such as nitrogen transport, carbon flux regulation, and precursor supply, which are essential for the metabolic underpinnings of floral scent production [48]. Flowers, being metabolically active tissues with robust glycolytic activity, are likely to enhance the synthesis of specific AAs [49]. Notably, Glu serves as the primary donor of amino groups in plants, participating in various transamination processes and the regulation of nitrogen balance [50]. Both Asn and Asp not only act as forms of nitrogen storage and transport but also serve as precursors for aromatic AAs like Phe and Tyr, which are integral to the phenylpropanoid metabolic pathways [51]. Additionally, Ser, formed from glycolytic intermediates, can be converted into Cys and participate in the synthesis of thiol-based aroma compounds [52]. The consistent detection of the sulfur-containing compound timonacic across all samples suggests that its biosynthesis may be intimately linked to Ser enrichment in sulfur metabolism. A correlation analysis between VOCs and AAs revealed significant positive correlations between Phe and benzaldehyde, Leu and ethyl benzoate, and Cit with fatty acid-derived VOCs. This analysis suggests that certain AAs may influence the differential accumulation of characteristic aroma compounds through the regulation of substrate flux, enzyme activities, or metabolic network orientation [53]. Collectively, these AAs not only act as precursors for the formation of crabapple floral scents but also offer potential targets for metabolic regulation and targeted aroma breeding.

In analyzing the aroma type differences among crabapple flowers, a chemometric network approach was utilized, employing a multiple-validation system. Initially, separate OPLS-DA models were constructed based on VOCs data (R²Y = 0.608, Q² = 423) and AA profiles (R²Y = 0.631, Q² = 0.520) from 18 crabapple cultivars, which demonstrated limited predictive capabilities when using either VOCs or AAs alone for aroma classification. Subsequently, an innovative combined VOC–AA model was developed, achieving a significant improvement in predictive performance (R²Y = 0.942, Q² = 0.732). An optimization of this model revealed that variations in aroma phenotypes arise from the synergistic interactions between VOC synthesis and AA metabolic pathways. This finding serves as a methodological reference for future multi-omics studies investigating the intricate aromatic mechanisms in flowers.

5. Conclusions

This study comprehensively elucidated the complex regulatory mechanisms governing aroma formation in crabapple flowers and successfully classified crabapple aroma types through a novel predictive model that integrates VOC and AA data. Employing chemometric methods, this research unveiled intricate relationships between aroma compounds and AAs. Volatile aroma components from 18 crabapple flowers were extracted and analyzed using HS-SPME-GC/MS, while AA profiles were determined via UPLC-MS/MS technology. The analysis revealed that Glu, Asp, Asn, Ser, and Ala were particularly abundant across the cultivars and played significant roles in the synthesis of aroma compounds. Major aroma compounds identified included benzyl alcohol, benzaldehyde, and ethyl benzoate, which possessed notably higher OAVs than other components, underscoring their critical influence on the aroma profile. The correlation analysis identified significant positive associations between Phe and benzaldehyde, Leu and ethyl benzoate, and Cit with fatty acid-derived VOCs, indicating potential synergistic interactions. Future research could expand this methodology to explore the interactions between aroma compounds and AAs in a broader array of crabapple cultivars. This could provide a theoretical basis for the molecular mechanism analysis and varietal improvement of crabapple flower aromas. Additionally, the detection of timonacic, noted for its potential antioxidant and anti-tumor properties, underscores its promising applications as a functional ingredient for both medicinal and edible uses.