Three varieties of blue honeysuckle, namely Chaoxian (CX), Luohuotan (LHT), and Lanjingling (LJL), were selected for the experiment. All blue honeysuckle samples were harvested at full ripeness in mid-June from the same experimental orchard (Harbin, Heilongjiang, China) under consistent cultivation conditions. The three cultivars of blue honeysuckle exhibited distinct morphological differences (Figure 1). “CX” berries exhibitd a slender cylindrical shape, with average fruit length and width of approximately 2.3 cm and 1.3 cm, respectively; ‘LHT’ fruits were shorter, rounder, and oval-shaped, averaging approximately 1.5 cm in length and 0.9 cm in width; “LJL” fruits displayed intermediate length (about 2.1 cm long and 1.0 cm wide) with irregular morphology and slight flattening. Methanol, acetonitrile, and formic acid (chromatographically pure) were purchased from Kemio Chemical Tengda Biotechnology Co. (Xi’an, China). Potassium chloride and tartaric acid (analytical grade) were supplied by Merck KGaA (Darmstadt, Germany). Other reagents used were of analytical grade.

Taste attributes of the three different varieties of blue honeysuckle samples were determined using a TS-5000Z electronic tongue system (INSENT Corporation, Japan). Fresh blue honeysuckle fruit samples were squeezed to obtain the juice. For the analysis of sourness, bitterness, astringency, umami, and saltiness, the juice was diluted three times with distilled water. For sweetness analysis, a separate aliquot of the juice was diluted 100 times with distilled water.

The sensory panel was composed of 15 members (8 females and 7 males, aged 25–45 years) and underwent specialized bitterness recognition training with quinine solutions (0.01, 0.02, 0.03, and 0.05 g/L). Fresh blue honeysuckle samples from three varieties (CX, LHT, and LJL) were randomly coded and presented to sensory panelists at room temperature (25 ± 1°C). The intensity of bitterness was rated using a structured 9-point scale, where 1 represented “not bitter at all” and 9 indicated “extremely bitter.” Between samples, panelists rinsed their mouths with purified water and waited for 1 min to avoid taste interference.

The results of three parallel tests were averaged and expressed as mean ± standard deviation. Statistical analyses were conducted using IBM SPSS Statistics 22.0 (SPSS Inc., Chicago, IL, United States), with significant differences among groups determined by one-way analysis of variance (ANOVA) followed by Duncan’s multiple range test for post hoc pairwise comparisons at p < 0.05. Principal component analysis (PCA) and hierarchical cluster analysis cluster were conducted using Origin 2023 (OriginLab Corporation, Northampton, Massachusetts, United States) software, and scientific graphs were generated.

Flavonoids, sugars, and polyphenols were widely present in blue honeysuckle, as reported in previous studies (2, 8). The contents of TF, TS, TA, and TP of the three varieties of blue honeysuckle were shown in Table 1. Among the three cultivars, CX exhibited the highest contents of TF and TP, followed by LHT with intermediate values, while LJL showed the lowest contents of TF and TP. CX and LHT showed no significant difference in TS content, but both were significantly higher than LJL. In terms of TA content, the level in LHT was significantly higher than those in CX and LJL. These compositional differences are likely to influence the sensory characteristics of the cultivars, particularly the bitterness, as higher TF and TP contents have been associated with increased bitter taste intensity (24).

Free amino acids act as key contributors to the sensory characteristics of natural foods, with diverse taste profiles including sourness, sweetness, bitterness, and umami (10). The free amino acid contents of different varieties in blue honeysuckle were shown in Table 2. A complete set of 22 amino acids was measured, consisting of 9 essential amino acids, and 13 non-essential amino acids. Among them, l-glutamine, l-glutamic acid, and l-aspartic acid collectively account for over 80% of the total amino acid content in blue honeysuckle, constituting the primary components of its free amino acid profile. These free amino acids serve as the primary contributors to the umami taste of blue honeysuckle. The contents of l-arginine, l-lysine, l-valine, l-leucine, and l-phenylalanine reached the maximum levels in CX. Based on sensory attributes, the aforementioned amino acids are generally associated with bitterness (22, 25).

The taste attributes of three different varieties of blue honeysuckle were measured by the electronic tongue system. Since the reference solution was prepared using 30 mM potassium chloride and 0.3 mM tartaric acid, which contain small amounts of acid and salt, the tasteless points of the sourness and saltiness were −13 and −6, respectively. As shown in Table 3, blue honeysuckle exhibited diverse taste profiles, and all other taste indicators except for richness are effective indicators (with response values higher than the tasteless point). Sourness was the prominent flavour indicator of all three blue honeysuckle fruits, with LHT being the most acidic, which is consistent with the results of the TA measurements. Blue honeysuckle fruits had a certain degree of bitterness and astringency, and there were some differences in bitterness and astringency among the three varieties, with CX having the greatest bitterness, aftertaste-bitterness (B), and aftertaste-astringency (A), LHT being in the middle, and LJL having the least. The bitterness intensity of the three blue honeysuckle varieties (CX, LHT, and LJL) aligns consistently with sensory evaluation test results.

PCA is a multivariate analytical tool designed to uncover underlying patterns and correlations within complex datasets (26). Based on the PCA results (Figure 2A), the first principal component (PC1) and the second principal component (PC2) explained 75.5 and 20.5% of the variation, respectively. The proximity of the sample points within the group indicated high intra-group reproducibility, while the considerable distance between the sample points of various groups suggested differences in flavor among the three blue honeysuckle groups (27). The taste attribute contribution plot (Figure 2B) demonstrated that bitterness exhibited the highest contribution along the PC1, followed by astringent-A and aftertaste-B. On the PC2, astringency, saltiness, and sourness demonstrated significant contributions. These findings indicated that the flavor differentiation among the three blue honeysuckle varieties was primarily attributed to the aforementioned taste attributes.

Results from sensory evaluation confirmed significant differences in bitter taste intensity among the three varieties (Figure 3), with CX showing the highest average bitterness score of 7.8 ± 0.52, followed by LHT with a moderate bitterness score of 6.0 ± 0.34, and LJL recording the lowest bitterness score of 3.1 ± 0.25.

A total of 692 metabolites were identified, including 71 amino acids and their derivatives, 105 phenolic acids, 36 nucleotides and their derivatives, 176 flavonoids, 2 quinones, 26 lignans and coumarins, 11 tannins, 9 alkaloids, 23 terpenes, 65 organic acids, 104 lipids, and 64 other metabolites. Among them, flavonoids, phenolic acids, and lipids were established as the principal constituents of CX, LHT, and LJL. These functional compounds demonstrated antioxidative properties by mitigating oxidative stress and suppressing β-amyloid production, thereby indirectly attenuating Alzheimer’s disease (14).

PCA of metabolite data was carried out for three varieties of blue honeysuckle samples. As shown in Figure 4A, the contribution rates of PC1 and PC2 were 43.7 and 28.1%, respectively, with a combined total of 71.8%, suggesting that PC1 and PC2 contained most of the metabolite information of the blue honeysuckle samples (26). The three blue honeysuckle varieties (CX, LHT, and LJL) were clearly separated and the three biological replicate samples of each variety were clustered together, demonstrating the reliability and good repeatability of the experiment, as well as significant differences among the three varieties (28). Moreover, distinct clustering into three groups was evident on the hierarchical clustering heatmap (Figure 4B), indicating significant differences in metabolites among the three varieties.

To identify metabolites with significant differences in accumulation between blue honeysuckle varieties, a screening threshold of |log₂Fold change (FC)| ≥ 1 and p ≤ 0.05 was applied. As shown in Figure 4C, a total of 221 differential metabolites were detected between CX and LHT, with 151 up-regulated and 70 down-regulated metabolites in CX. Among the 185 differential metabolites identified between CX and LJL, 119 and 66 metabolites were up-regulated and down-regulated in CX, respectively (Figure 4D). In addition, 154 differential metabolites were present between LHT and LJL, with 84 and 70 metabolites up- and down-regulated in LHT, respectively (Figure 4E). The differential metabolites in various blue honeysuckle samples could be categorized into more than 11 classes (Table 4), and the majority of differential metabolites were categorized into four classes, including flavonoids, amino acids and derivatives, lipids, and phenolic acids. Meanwhile, multiple comparative analysis (Figure 4F) revealed that three different varieties of blue honeysuckle had 24 shared differential metabolites. These metabolites covered multiple biochemical classes, mainly flavonoids (e.g., isohyperoside, pinocembrin-7-O-neohesperidoside, and kaempferol-3,7-di-O-glucoside), amino acids and derivatives (e.g., 5-hydroxy-l-tryptophan, l-phenylalanine, and l-isoleucine), phenolic acids (e.g., brevifolin carboxylic acid and 3,4,5-tricaffeoylquinic acid), and organic acids (e.g., fumaric acid and succinic anhydride).

Overall, the differentially accumulated metabolites could be broadly divided into two groups (Table 4). The first group of differential metabolites constituted antioxidant components such as flavonoids, alkaloids, phenolic acids, and terpenoids, while the second group included taste-related components such as organic acids, and amino acids and derivatives (29). Most differentially accumulated metabolites of the antioxidant components were found at higher concentrations in CX sample. CX, LHT, and LJL displayed similar metabolite accumulation patterns. There were significant differences in about 20–30% of the metabolites between the various varieties of blue honeysuckle samples, with flavonoids being the major differential metabolites.

The metabolites identified in the blue honeysuckle samples were further analyzed in comparison with the BitterDB database (30) and relevant literature for further screening of potential bitter metabolites. As shown in Table 5, a total of 73 were screened as potentially associated with the bitterness of blue honeysuckle, including 11 amino acids and derivatives, 11 phenolic acids, 1 nucleotide and derivative, 30 flavonoids, 6 tannins, 2 terpenoids, 1 organic acid, 6 lipids, and 5 other categories. Flavonoids, amino acids and derivatives, phenolic acids, and lipids constituted the most abundant classes of potential bitterness-associated metabolites identified in blue honeysuckle.

Flavonoids have been systematically characterized as the primary bitter principles in fruits and vegetables (13). Catechin and gallic acid are the main bitter substances in green tea, and catechin, epicatechin, gallic acid, and quercetin serve as key contributors to the increased bitterness in wine (10). According to Table 5, the most diverse types of bitter substances screened from blue honeysuckle were flavonoids (30 species), primarily including chrysin, pinocembrin, naringenin, phloretin, luteolin, eriodictyol, catechin, epicatechin, epigallocatechin, isorhamnetin, apigenin-O-glycosides, naringenin-O-glycosides, kaempferol-O-glycosides, quercetin-O-glycosides, luteolin-O-glycosides, cyanidin-O-glycosides, eriodictyol-O-glycosides, diosmetin-O-glycosides, and isorhamnetin-O-glycosides. Xia et al. (1) similarly identified various flavonoid bitter substances such as anthocyanins, flavan-3-ols, flavanonols, flavones, and flavonols in blue honeysuckle. Epigallocatechin is a common catechin compound found in plant-based foods such as tea, apples, and grapes (31). Catechin is the main component of tea polyphenols, accounting for approximately 75 to 80% of the total polyphenol content in tea (32). The above two substances are one of the primary sources of the bitter and astringent taste of tea. Compared with LJL, epicatechin was significantly down-regulated in CX and LHT (Supplementary Table S1). It has been reported that phenolic acids exhibit a sour and astringent taste, whereas quercetin-3-O-rutinoside has a mild astringent taste. Phenolic acids and quercetin-3-O-rutinoside could enhance the bitterness of catechins (33). Most of the flavonoid bitter metabolites exhibited high bitter intensity. For instance, kaempferol-3-O-galactoside and kaempferol-3-O-rutinoside demonstrated remarkably low bitterness thresholds of 0.00067 mmol/L and 0.00043 mmol/L, respectively, both significantly lower than the widely used bitter reference compound quinine hydrochloride (0.3 mmol/L) (22). The content of eriodictyol, isorhamnetin, kaempferol-3-O-galactoside, luteolin-6-C-glucoside, naringenin-7-O-neohesperidoside, and eriodictyol-7-O-rutinoside showed up-regulated expression in CX (Supplementary Table S1). Thus, these flavonoids may be one of the contributing factors to the intensified bitter taste in the varieties of blue honeysuckle.

Amino acids, particularly essential amino acids, frequently exhibit bitter taste profiles and serve as biosynthetic precursors for bitter constituents in food systems (34). Studies have revealed that l-phenylalanine is a key contributor to the bitter taste in bamboo shoots, and l-valine, l-isoleucine, and l-phenylalanine are identified as primary bitter components in Zanthoxylum Bungeanum Maxim (22, 35). In blue honeysuckle, several bitter amino acids have been identified, including l-valine, l-leucine, l-isoleucine, l-tyrosine, and l-tryptophan. These amino acids possess relatively high bitterness intensities, with reported taste thresholds of 30, 12, 10, 4, and 5 mmol/L, respectively (36, 37). Compared to the LJL group, l-histidine, l-leucine, l-valine, and l-phenylalanine exhibited significant up-regulation in both CX and LHT samples (FC ≥ 2). The shared differential metabolites between CX, LHT, and LJL included two bitter compounds, l-phenylalanine and l-isoleucine (Figure 4F), which might be the primary contributors to the bitterness of the blue honeysuckle. The potential role of amino acid bitter metabolites in blue honeysuckle bitterness needs to be further explored.

Phenolic acids contribute to bitterness and astringency in wine, peas, and corn germ protein flour (38). In the present study, phenolic acids like benzamide, salicylic acid, gentisic acid, caffeic acid, and arbutin contributed to the bitter flavor of blue honeysuckle. Salicylic acid frequently accumulates in bitter vegetables and fruits such as bitter orange and bitter gourd, but it typically functions as a precursor metabolite or byproduct of the primary bitter components, rather than serving as a direct contributor to bitterness (39). Coffee acid contributes to both the bitterness and astringency of peas (38). Arbutin is a phenolic glycoside compound with a reported bitter taste threshold of only 0.9 mmol/L, which produces a bitter taste by stimulating the hTAS2R16 receptor (25). Fatty acids have recently been identified as the primary bitter compounds in oat and pea protein isolates (37, 40). Blue honeysuckle contained free fatty acids with strong bitterness (bitter recognition threshold < 1 mmol/L), such as palmitic acid, α-linolenic acid, linoleic acid, stearic acid, and (9Z,11E) − 13-hydroxyoctadeca-9,11-dienoic acid. Among them, α-linolenic acid, linoleic acid, and stearic acid were more abundant in blue honeysuckle (peak area > 1 × 10⁷). The phenolic acid and fatty acid metabolites showed no significant differences among CX, LHT, and LJL (Supplementary Table S1).

Tannins are oligomers or polymers composed of derivates from (+)-catechin and its isomers, which may be responsible for bitter and astringent flavors in grapes and wine (38). Six bitter tannins, namely procyanidin B1, B2, B3, B4, C1, and C2, were identified in blue honeysuckle. Procyanidin B4 and C2 are characterized with astringent and bitter taste in an aqueous ethanol solution at a concentration of 0.9 g/L. Tannin-derived bitter metabolites identified in blue honeysuckle possess exceedingly low bitterness thresholds (< 0.5 mmol/L). Procyanidin B2, B4, and C1 showed down-regulation in CX and LHT compared to the low-bitter LJL (Supplementary Table S1). In addition to the above substances, a number of terpenoids (sweroside, swertiamarin), vitamins (D-pantothenic acid, vitamin B1, B2, B3), resveratrol, adenosine, and quinic acid with a bitter flavor have been identified in blue honeysuckle. Compared with LHT, vitamin B1 was significantly up-regulated in CX, while sweroside was significantly down-regulated in CX.

The KEGG database is widely utilized for investigating signal transduction pathways and metabolite accumulation, serving as a principal public repository for pathway analysis (60). As illustrated in Figure 5, the differential bitter metabolites among various blue honeysuckle samples are predominantly accumulated in biosynthesis of secondary metabolites (ko01110), metabolic pathways (ko01100), biosynthesis of amino acids (ko01230), flavonoid biosynthesis (ko00941), aminoacyl-tRNA biosynthesis (ko00970), flavone and flavonol biosynthesis (ko00944), and ABC transporters (ko02010). These enriched KEGG pathways were mostly closely related to fruit growth, development, and flavor formation of blue honeysuckle (13). The biosynthesis of amino acids and flavonoids biosynthesis were identified as key pathways associated with bitter compound accumulation in blue honeysuckle (Figure 6). Within the biosynthesis of amino acids pathway, significant accumulation of branched-chain amino acids (BCAAs), such as l-valine, l-leucine, and l-isoleucine, was observed in the high-bitter cultivar CX compared to LHT and LJL. These amino acids and their metabolic intermediates (e.g., 2-oxoisovalerate, 3-isopropylmalate) are known precursors for various secondary metabolites, including bitter-tasting compounds. The elevated levels of l-histidine in CX also suggest its potential role in bitterness perception. More importantly, the flavonoid biosynthesis pathway showed a strong association with bitterness differentiation. In particular, naringenin, prunin, naringin, and vitexin were significantly up-regulated in CX and LHT, with much lower levels detected in LJL. Epicatechin was found enriched in all three cultivars (peak area > 1 × 10⁶, Table 5) but showed the highest accumulation in LJL, implying that it may contribute to overall taste complexity but is not the primary driver of bitterness. Collectively, the accumulation patterns of l-valine, l-leucine, l-isoleucine, l-histidine, naringin, and vitexin were consistent with the bitterness gradient (CX > LHT > LJL), indicating that they are potential key factors contributing to the bitterness of blue honeysuckle. These compounds might synergistically cause bitter taste variations among blue honeysuckle varieties and warrant further validation as biomarkers of bitterness in future breeding and quality control strategies.