Metabolomics and sensory evaluation combined analysis reveal the effect of processing methods and different forms on the flavor of rose herbal tea

Table of Contents

Sensory evaluation of rose herbal tea

The sensory quality of rose herbal tea was evaluated using traditional sensory evaluations. The color of LTD roses and their aqueous tea infusions were darker, which results from ascorbic acid browning, the Maillard reaction, and the caramel reaction¹⁶. VFD roses have the freshest red color, while the color of rose tea infusions is lighter (Fig. 1A). The color intensity of rose petals and their tea infusion depends on the type and content of metabolic compounds contained. Vacuum and low temperature environment suppresses the oxidation reactions and enzymatic browning of roses, thereby better maintaining the original color of the sample¹⁷.

**Fig. 1: Sensory quality analysis of four rose herbal teas.**

Similarly, those rose herbal teas have distinct sensory characteristics (Supplementary Data 1). The results demonstrated that LTD rose tea has higher floral and sweet intensity than other aromas, while VFD rose tea has higher intensity of fruity and woody aromas than other aromas. In addition, the RCs of rose tea (LTD-RC and VFD-RC) had distinct milky aromas in aroma (Fig. 1B). The taste characteristics of rose tea vary greatly among different forms and processes. The RCs (LTD-RC and VFD-RC) exhibited distinct sweetness, especially LTD-RC. LTD-RP exhibited distinct flavor characteristics, including strong sourness and astringency, together with low-intensity freshness, smoothness, and sweetness. This may be due to the fact that polyphenolic compounds in LTD-RP, especially anthocyanins, are more easily dissolved in rose tea infusion¹⁵. Furthermore, the drying process of the VFD results in a smoother and fresher taste of the rose infusion (Fig. 1C). In summary, different forms and dried processed rose teas have specific flavor properties, it is crucial to investigate the flavor characteristics components of rose tea.

AAs differential accumulation in different forms of rose herbal tea using LTD and VFD

Amino acids and their derivatives (AAs) are the primary participants in the flavor of rose tea. They have unique sensory characteristics and physiological functions, and play crucial roles in aroma, taste, and even health benefits¹⁸. In this study, there were 78 kinds of AAs were identified. In four rose herbal teas, γ-aminobutyric acid was the most prominent amino acid, followed by L-glutamic acid, L-asparagine anhydrous, L-glutamine, and N-acetylneuraminic acid (Supplementary Data 2a).

To examine the variations in the accumulation profile of AAs in the four types of rose tea, principal component analysis (PCA) was performed. The scatter plot of the PCA scores showed that the four types of rose tea were divided into three categorie: the LTD samples were grouped together, and the VFD-RB and VFD-RC samples were separated (Fig. 2A). This indicates that the degree of flower development and the drying procedure have an impact on the amount of amino acids that accumulate in rose tea. Trend clustering analysis (TCA) revealed all the AAs in the four rose herbal tea could clustered into 10 classes (Supplementary Fig. 1A). The AAs in classes 1, 5, and 7 were the key AAs affecting the distribution of the LTD samples, while the AAs in classes 4, 6, 8, and 10 were the key AAs affecting the grouping of the VFD samples (Fig. 2B).

**Fig. 2: Amino acids (AAs) and organic acids (OAs) profiles of four rose samples and drying techniques.**

There are differences in the 78 amino acid accumulation patterns among the four varieties of rose tea, although the total amount of amino acids has no significant difference (Fig. 2C). Among the four rose herbal teas, VFD-RC had the most amino acids (51.97 mg/g), followed by LTD-RC (48.69 mg/g), LTD-RP (48.53 mg/g), and VFD-RB (46.05 mg/g). The results of hierarchical clustering analysis (HCA) showed that the AAs in class 1 (22 AAs), such as N-acetylneuraminic acid, L-tryptophan, glycine, and L-alanine, mainly accumulate in the LTD samples. L-proline and L-serine were more abundant in LTD-RP, while L-phenylalanine and Nα-acetyl-L-arginine were more abundant in LTD-RC. Moreover, rose tea processed by VFD accumulated more L-tryptophyl-L-glutamic acid, ethanolamine, and L-α-aspartyl-L-phenylalanine. In general, all those four types of rose herbal tea have accumulated characteristic amino acids, although there was no significant difference in the total amount of amino acids between them.

Differential accumulation of OAs in different forms of rose herbal tea using LTD and VFD

Sour taste is usually tasted in rose tea brewing and has a negative impact on sensory quality. Acidic metabolites can affect the flavor of tea, and organic acids (OAs) are considered one of the main factors¹⁹. The organic acid composition of different types of tea varies, and the primary OAs in black tea are citric acid and quinic acid²⁰. The most prominent organic acids detected in rose herbal tea were L-malic acid, maslinic acid, tartaric acid, and cis-aconitic acid (Supplementary Data 2b). Among them, L-malic acid was dominant. Malic acid improves oral health in patients with dry mouth syndrome²¹.

To explore the effects of maturity and drying methods of rose tea on the accumulation of OAs, PCA was conducted. The four finished rose teas were clustered separately, and the LTD samples were clustered together, and separated from VDD-RB and VFD-RC. This indicates a significant impact of flower maturity and drying method on organic acid content (Fig. 2D). Based on TCA, the compound loading graph showed that these 52 OAs could be divided into 10 classes (Fig. 2E, Supplementary Fig. 1B). The OAs in classes 3, 4, 6, and 10 were the key OAs that mainly accumulated in the VFD samples, while the OAs in classes 1, 5, 7, and 8 were the key OAs that mainly accumulated in the LTD samples.

The overall quantity of OAs did not significantly differ between the LTD samples, but it did significantly differ between the VFD samples and between the VFD and LTD samples. The OAs content of VFD-RC was 2.10 mg/g, followed by those of VFD-RB, LTD-RP, and LTD-RC (Fig. 2F). The differential accumulation of several organic acids in rose tea was further examined using HCA. The 6 OAs in Group 6, including L-malic acid, oxoglutaric acid, pyruvic acid, cryptochlorogenic acid, 2-methylsuccinic acid, and kynurenine, were more abundant in VFD-RC than in the other groups. In VFD-RP, gallic acid, shikimic acid, succinic acid, aminobenzoic acid, 4-hydroxyphenylacetic acid, 2-hydroxy-2-methylbutyric acid, azelaic acid, sebacic acid, and tartaric acid accumulate more. Specifically, azelaic acid, sebacic acid, and 2-hydroxy-2-methylbutyric acid accumulate at high levels in VFD-RP. Gallic acid (GA) and succinic acid (SA) were discovered to strengthened the umami taste²². Further explanation showed that flower maturity and drying methods significantly affect the accumulation of OAs.

Differential accumulation of SSs in different forms of rose herbal tea using LTD and VFD

SSs can conduce to the formation of aroma and sweetness in rose herbal tea²³. In this study, there were 29 kinds of SSs identified in the four rose herbal teas. The most abundant soluble sugar among the four types of rose tea was glucose, which reached 95.61 mg/g in VFD-RC, followed by fructose and sucrose, respectively (Supplementary Data 2c). The PCA score plots indicated that the four rose teas could be well distinguished based on these 29 SSs (Fig. 3A). The total amounts of these SSs in LTD-RP and VFD-RC were relatively high, but the difference was not significant. The SSs content in the VFD-RB was the lowest (Fig. 3B). This indicates that the content of SSs in rose tea is influenced mainly by the maturity of the flowers.

**Fig. 3: Soluble sugars profiles in four rose samples and drying techniques.**

To better distinguish the types of SSs that accumulated in the four types of rose tea, TCA assays were conducted on the SSs mentioned above. The TCA results indicated that SSs can be divided into 6 classes based on differences in accumulation among the four types of rose tea (Fig. 3C). The loading diagram and HCA showed that 12 SSs in class 1, mainly sucrose, raffinose, and D-arabinose, were highly accumulated in LTD rose teas. The 6 SSs in class 6, mainly maltose, D-mannose-6-phosphate, and D-glucuronic acid, were highly accumulated in rose tea processed by VFD. In addition, the content of the 6 SSs in class 4 did not differ significantly among the four types of rose tea, especially L-rhamnose, D-sorbitol, and D-ribono-1,4-lactone (Fig. 3D, E). This indicates that compared to the drying methods, the open state of roses have more effects on the soluble sugar type and content of rose tea.

Flavonoids differential accumulation in different forms of rose herbal tea using LTD and VFD

Phenolic compounds such as flavonoids and anthocyanins give roses strong antioxidant activity²⁴. In this study, 69 kinds of flavonoids with contents greater than 1 μg/g were detected in four rose herbal teas, including 46 flavones and flavonols, 3 flavanols, 2 procyanidins, and 18 anthocyanins (Supplementary Data 2d). The 18 main anthocyanins were divided into 5 categories: petunidin (1), delphinidin (3), pelargonidin (2), cyanidin (6), and peonidin (6). Among the rose samples, the flavonoids content was highest in the VFD-RC (11.68 mg/g) and lowest in the LTD-RP (8.67 mg/g). During the fully open flowering period, the content of flavonoids increased²⁵. This suggested that the accumulation of flavonoids was influenced by the flowering period in rose tea.

However, the scatter plot of the PCA scores revealed that it is difficult to distinguish between the four rose herbal tea (Fig. 4A). The compound loading diagram showed that the most diverse compounds in rose tea are flavonoids and flavonol glycosides (Fig. 4B). There was no appreciable difference in the total content of flavonoids amongst rose samples handled with the same drying technique, while the total flavonoids content of rose herbal tea processed by LTD and those by VFD differed significantly (Fig. 4C). Furthermore, VFD roses had a greater overall flavonoid content than LTD roses. Flavonoids in LTD samples may degrade as a result of prolonged drying and high temperatures²⁴. This indicated that drying techniques have a greater impact on the flavonoid accumulation in rose herbal tea.

**Fig. 4: Flavonoids profiles of the four rose samples and drying techniques.**

Anthocyanins are the main colorants in rose petals. The content of anthocyanins in LTD-RP was the highest, mainly cyanidin and peonidin, while the contents of petunidin, delphinidin, and pelargonidin were relatively low. Furthermore, LTD-RP contained the lowest amounts of procyanidins, flavanols, flavones, and flavonols (Fig. 4C). LTD-RP does not contain stamens, but it is precisely the stamens that contain more flavones and flavonols, which have a high antioxidant capacity²⁶.

HCA was used to comprehensively analyze various types of flavonoids in rose tea (Fig. 4D). The most abundant anthocyanins detected in rose herbal tea were peonidin-3,5-O-diglucoside (Pn3G5G), and cyanidin-3,5-O-diglucoside (Cn3G5G). This determines the color tone of rose petals²⁷. VFD-RC had the highest content of Cn3G5G and Pn3G5G, resulting in a color consistent with that of fresh flowers. The accumulation of anthocyanins in LTD-RP is relatively high, except for acylated anthocyanin glycosides and cyanidin-3-O-rutinoside. The contents of delphinidin-3,5-O-diglucoside, Pn3G5G, cyanidin-3-O-sambubioside, and cyanidin-3-O-glucoside, were generally lower in the RC stage, especially in the LTD-RC sample. Flavonoid glycosides in rose teas were mainly derivatives of glycosylated, acylated, and methylated quercetin and kaempferol. The VFD rose samples contained more catechins and anthocyanin B3. More quercetin and catechins are present in rose buds¹³. Based on the above results, according to the types and contents of flavonoids, the rose samples from the LTD and VFD groups did not aggregate well.

Therefore, the total accumulation of flavonoids in rose tea is related to the drying method, while the kinds of metabolites and their contents are related to the developmental stage of the flowers.

VOCs profiles of different forms of rose herbal tea

The flavor of rose herbal tea is significantly influenced by aroma, which is the outcome of multiple volatile components working together. In recent years, the relative odor activity value (rOAV) has been increasingly applied by scholars to identify key flavor components in various types of food and tea²⁸. Generally, an rOAV ≥ 1 signifies that the compound directly assists contribution to the flavor of the rose herbal tea. This section explores the differences in VOCs in rose corollas, buds, and petals under LTD and VFD processes.

A total of 88 aromatic active substances (rOAV ≥ 1) were identified in the LTD-RP, LTD-RC, VFD-RC, and VFD-RB samples through GC-MS and rOAVs analysis. These substances can be divided into 9 categories, namely 11 alcohols, 6 ketones, 15 aldehydes, 4 phenols, 13 esters, 7 aromatics, 4 hydrocarbons, 17 terpenoids, and 9 heterocyclic compounds (Supplementary Data 3a). The contents of VOCs were greatest in the VFD-RC (99.18 μg/g), and lowest in the LTD-RC (59.99 μg/g) (Fig. 5A). According to recent studies, alcohol is the primary aroma component of roses^6,29. However, our research revealed that among the 88 key VOCs, heterocyclic compounds accounted for 23–29% (9 categories) of the total VOCs the four types of rose tea. LTD samples contain more esters, while VFD samples contain more aldehydes (Fig. 5B). Notably, 3-(2-furanyl)-2-propenal, eugenol, (Z)-2-decenal, n-valeric acid cis-3-hexenyl ester, acetic acid, 2-phenylethyl ester, and trans-carveol were the abundant VOCs in Pingyin rose samples⁶.

**Fig. 5: Analysis of key volatile organic compounds (VOCs) in four rose samples through GC-MS and rOAVs (rOAV ≥ 1).**

The accumulation patterns of 88 VOCs in the four types of rose tea were examined using PCA. The scatter plot of the PCA scores showed that the four samples were dispersed into three groups, among which the LTD samples could be clearly separated from the VFD-RC and VFD-RB samples (Fig. 5C). The loading scatter plot showed that VOCs are dispersed in three different regions. Combined with the PCA score scatter plot, these three groups of VOCs were named as the LTD type, VFD-RC type, and VFD-RB type, respectively (Fig. 5D; Supplementary Data 3b).

HCA revealed the characteristic accumulation of VOCs in different rose herbal teas (Fig. 5E–G). There are 48 VOCs in the VFD-RB type, among which 19 are characteristic VOCs of the VFD-RB, including benzaldehyde, 1-methyl-naphthalene, (E)-4-decenal, 2-pentyl-furan, 1-octene, hexanal, biphenyl, and (E, E)-3,5-octadien-2-one. The remaining 29 VOCs, including eugenol, (isothiocyanatomethyl)-benzene, (Z)-4-heptenal, o-cymene, and (E)-2-nonenal are characteristic accumulated VOCs of VFD processing methods (Fig. 5E). There are 13 VOCs in the LTD type, acetic acid, 2-phenylethyl ester, trans-carveol, and n-valeric acid cis-3-hexenyl ester accumulate more in LTD processed rose tea. Among them, α-Farnesene and geranyl isobutyrate are characteristic accumulation of LTD-RC (Fig. 5F). VFD-RC type conteins 27 VOCs, such as 2-undecanone, methyl benzoate, n-butylbenzene, and 4-phenyl-2-butanol (Fig. 5G). The differences in VOCs among the different drying processes can be attributed to differences in drying mechanisms and dehydration behaviors.

Screening of characteristic flavor metabolites of rose tea

To further identify the characteristic flavor metabolites of rose tea, partial least squares discriminant analysis (PLS-DA) was conducted on VOCs (Supplementary Fig. 2) and non-volatile compounds (Supplementary Fig. 3). The differences of volatile compounds between the LTD-RC and LTD-RP samples were not clear, while the differences between the LTD samples and the VFD-RC and VFD-RB samples were significant (Fig. 5C, D; Supplementary Fig. 2C, D). This indicates that drying method and flowering degree have a significant effect on VOCs of rose tea. Additionally, 24 compounds with rOAV and variable importance in projection (VIP) greater than 1 were selected (Supplementary Data 3c). Among them, trans-carveol, n-valeric acid cis-3-hexenyl ester, tetrahydro-4-methyl-2-(2-methyl-1-propenyl)-2H-pyran, and 2-phenylethyl ester acetic acid, were the main VOCs in the LTD rose samples. Hexanal, 2-pentyl-furan, (E)-2-hexenoic acid, and 2-hexanol played pivotal roles in VFD-RB. In addition, trans-β-ocimene, 2-undecanone, (Z)-3,7-dimethyl-1,3,6-octatriene, citral, and β-myrcene played pivotal roles in VFD-RC.

Potential link between taste metabolites and sensory characteristics was found using network analysis (Fig. 6A, B). The correlation analysis revealed a positively link between 1-octene, 2-hexanol, and hexanal and the milky aroma. Most of the milky aroma comes from lipid degradation pathways. For example, 2-hexanol is derived from the oxidation of linoleic acid, which is facilitated by lipoxygenase. Hexanal is considered one of the key aromas in milky-white tea³⁰. The floral and sweet aromas were positively correlated with 2-phenylethyl acetate, cis-3-hexenyl n-valerate, and trans-carveol. Phenylethyl acetate has a rose like odor and is widely used to add fragrance or flavor to cosmetics and beverages³¹. Cis-3-hexenyl n-valerate is considered one of the characteristic aroma components of matcha³². The fruity and woody aromas were positively correlated with eugenol, 2-methoxy-4-propyl-phenol, and (isothiocyanatomethyl)-benzene. It has been established that eugenol is a crucial aroma component in rose-based products³³. 2-Undecanone, α-pinene, β-myrcene, citral, (Z)-2-decenal, trans-β-ocimene and 3-(2-furanyl)-2-propenal had greater effects on the grass aroma. The majority of odorants that have a positive correlation with the aroma of grass have a negative correlation with the aroma of milky, and they are more concentrated in the VFD rose samples.

**Fig. 6: Network correlation analysis between key VOCs and nonvolatile compounds and aroma and taste of sensory quality analysis results.**

A similar situation occurs in the contribution of nonvolatile compounds to taste. Considering the significant differences in the contents of AAs, OAs, SSs, and flavonoids, conducting PLS-DA separately can provide a more comprehensive analysis of key nonvolatile compounds (Supplementary Fig. 3–6). Fifty-one compounds with a VIP greater than 1 were selected, including 21 amino acids, 12 organic acids, 4 SSs, and 14 flavonoids (Supplementary Data 4). Notably, for substances with strong positive correlation with astringent and sour, including γ-glutamate-cysteine, Nα-acetyl-L-glutamine, L-serine, 2”-O-galloylhyperin, L-proline, α-aminoadipic acid, L-alanine, glutathione oxidized, N-acetylneuraminic acid, β-alanine, phosphorylethanolamine, 3-hydroxymethylglutaric acid, glucose, glycine, tartaric acid, naringenin-7-O-glucoside, and peonidin-3-O-galactoside, had strong negative correlation with smooth and fresh. Additionally, L-arginine, raffinose, and L-threonine contributed more to the sweet aroma, while the most prevalent SSs in rose tea, glucose, and D-fructose, were negatively correlated with sweetness (Fig. 6C). Sensory studies have shown that sourness suppresses sweetness, and the inhibitory effect of sourness on sweetness is achieved by increasing the concentration of sourness³⁴. In this study, the contribution of sucrose to sourness was much greater than that of sweetness, suggesting that a high concentration of sourness weakened the correlation between sucrose and sweetness.

link