LC-MS-Based Screening for Colchicine and Characterization of Major Bitter Constituents in Lily

Abstract

Lilies (Lilium spp.) are highly valued in China for their edible and medicinal properties; however, bitterness in certain varieties limits consumer acceptance. Although historically attributed to colchicine, the presence of alkaloids in lilies remains a subject of debate. This research screened five lily species for colchicine and its 15 biosynthetic precursors, using Gloriosa superba and Colchicum autumnale as positive controls. While detected in the controls, none were detected in any tissues (bulbs, roots, stems, flowers, and leaves) of the five lilies. A comparative analysis of five lily varieties—Longyahong, Lanzhou, Lilium lancifolium, Longya, and Guiyanghong—revealed that Longyahong exhibited the strongest bitterness, which was localized exclusively in the bulb peels. Based on comparative LC-MS profiling between bitter and non-bitter varieties, three high-abundance compounds were selected for isolation and subsequent sensory evaluation. Two monomeric compounds were isolated and confirmed via chromatographic methods as the primary bitter components. This study provides compelling chemical and biochemical evidence of the presence of colchicine in the examined lilies. By identifying two specific bitter components in Longyahong bulb peels, these findings refute the long-standing misconception regarding colchicine in lilies and provide a chemical foundation for improving the palatability and commercial value of bitter lily varieties.

1. Introduction

Lilies are perennial bulbous plants belonging to the genus Lilium in the family Liliaceae [1]. Originating in China, they are widely cultivated as ornamentals across East Asia, Europe, and North America [2]. The Liliaceae family encompasses approximately 175 genera, with roughly 100 wild Lilum species distributed globally. China is home to 55 of these species, predominantly distributed in the southwest and central regions, 10 of which are known to be edible [3,4]. According to the Chinese Pharmacopoeia, the term “Lily bulbs”(baihe) refers to the dried fleshy scales of Lilium lancifolium Thunb., Lilium brownii F. E. Brown var. viridulum Baker, or Lilium pumilum DC. [5]. Valued as both a functional food and a traditional medicine, lily bulbs are extensively utilized in the food and pharmaceutical industries, demonstrating significant market potential [6]. First documented in the Shennong’s Materia Medica, medicinal lily is recognized for its ability to nourish yin, moisten the lungs, clear heart-fire, and tranquilize the mind. Clinically, it is prescribed for conditions such as yin deficiency-related chronic cough, as well as anxiety, insomnia, frequent dreaming, and mental cloudiness. Edible varieties are versatile in culinary applications, commonly prepared through stewing, boiling, stir-frying or incorporation into soups [3,7]. Phytochemical investigations have revealed that the genus Lilium is rich in steroidal saponins, polysaccharides, alkaloids, and flavonoids. These compounds exhibit a broad range of pharmacological activities, including antitumor, hypoglycemic, antibacterial, antioxidant, antidepressant, and anti-inflammatory activities [8].

Colchicine, first isolated from the genus Colchicum in 1820, is characterized by its bitter taste and highly toxic nature [9]. Research indicates that colchicine is prevalent throughout the Colchicaceae family [10], distributed across several genera such as Wurmbea [11] and Iphigenia [12]. Clinically, colchicine is standardly administered to treat gout, familial Mediterranean fever [13], and specific cardiovascular diseases [14]. However, it possesses a narrow therapeutic window and frequently induces adverse effects involving multiple organ systems, including neuropathy, gastrointestinal distress, hepatotoxicity, and hematological toxicity [15]. From 1985 to 2025, over 140 publications attributed the bitterness of lilies to the presence of colchicine. Among them, 18 studies specifically performed chemical testing, comprising 4 qualitative and 14 quantitative analyses. Nevertheless, some reports suggest that these identifications may be confounded by overlapping retention times in chromatographic analyses, potentially leading to misidentification [16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33]. Furthermore, key gene clusters associated with colchicine biosynthesis have not been identified in the lily genome [4]. Notably, there are no documented cases of colchicine poisoning resulting from the consumption of edible lilies. Consequently, the presence of colchicine in Lilium species warrants rigorous re-investigation.

Bitterness is one of the five primary tastes recognized in traditional Chinese medicine, characterized by an exceptionally low detection threshold. It can be perceived at concentrations as low as 0.0016%, rendering it highly detectable and capable of significantly compromising palatability [34,35,36]. Research indicates that while most edible lilies possess a sweet taste, medicinal varieties are predominantly bitter [37]. Historically, the bitterness has been attributed to colchicine, based on its reported presence and inherently bitter profile [23]. Current studies on lilies have largely focused on the chemical composition and pharmacological efficacy of the bulbs, yet investigations into specific bitter constituents remain limited. Among the known constituents, compounds such as glycosides and alkaloids have been identified; however, the precise substances responsible for the bitterness remain elusive. Furthermore, as many natural toxins or spoiled foods in nature are bitter, humans have evolved an innate aversion to bitterness as a protective mechanism against ingesting harmful substances [38,39]. With rising living standards and health consciousness, consumers may limit their intake due to concerns over potential toxicity associated with bitter components. Consequently, bitterness in lilies exerts a dual negative effect: it directly compromises palatability and indirectly raises safety concerns, together substantially restricting market demand and the broader application of lilies.

While lilies contain bitter compounds, their specific chemical identities remain poorly understood, thereby restricting their broader application in the food and pharmaceutical sectors. Furthermore, persistent consumer misconceptions regarding the potential presence of colchicine have raised unwarranted safety concerns. Therefore, this study aimed to characterize the bitter constituents, identify the specific tissues responsible for bitterness, and definitively clarify the safety profile regarding colchicine. To achieve these objectives, an integrated approach was employed, combining sensory evaluation, high-performance liquid chromatography coupled with quadrupole time-of-flight mass spectrometry (HPLC-Q-TOF-MS), and chromatographic isolation techniques. The findings are expected to provide a robust scientific foundation for developing effective debittering processes and promoting the safe, expanded utilization of lily-based products.

2. Results

2.1. Screening for Colchicine in Lilium

In this study, HPLC-Q-TOF-MS was employed to analyze the presence of colchicine, using a certified standard (m/z 400.1755, [M + H]⁺) as a reference. The exact mass of colchicine was extracted from the extracted ion chromatograms (EICs) of G. superba and C. autumnale bulbs, as well as from the bulbs, roots, stems, leaves, and flowers of the five lily varieties. By comparing retention times, exact masses, and MS/MS fragment ions with those of the standard, colchicine was definitively identified in the positive controls (G. superba and C. autumnale). In contrast, no signals corresponding to the exact mass or fragmentation pattern of colchicine were detected in any of the examined lily tissues. These results unequivocally demonstrate that colchicine is absent in the investigated Lilium species (Figure 1).

The biosynthetic pathway of colchicine in G. superba has been well-characterized, encompassing 15 precursor compounds. In this study, these 15 precursors were screened by analyzing the extracted ion chromatograms (EICs) of bulbs, roots, stems, leaves, and flowers from lily varieties, as well as bulbs of G. superba and C. autumnale. Utilizing full-scan MS and MS/MS data, all 15 precursors were confirmed in the G. superba and C. autumnale bulbs yet remain undetected in all examined lily tissues. Taking the identification of Cp-9, a key intermediate in the colchicine biosynthetic pathway, as a representative example (Figure 2): the theoretical m/z of Cp-9 is 360.1805 in ESI⁺ mode. EICs of m/z 360.1805 were generated from 20 different tissue samples across the five lily varieties (LYH, LZ, L. lancifolium, LY, and GYH). The results showed distinct peaks in G. superba and C. autumnale, whereas no corresponding signals were observed in any of the five lily species (Figure 2a).

Subsequently, MS/MS analysis of the precursor ion at m/z 360.1805 (Figure 2b) revealed characteristic fragment ions at m/z 179.0703 and 153.054. These fragments showed high consistency with the reference data for Cp-9 (Figure 2c,d), thereby confirming its identity in the positive controls. Following the established colchicine biosynthetic pathway (Figure 2e), this methodology was applied to screen for all 15 precursors (Cp-1 to Cp-15). While all 15 compounds were readily detectable in G. superba and C. autumnale bulbs, none were identified in the 20 lily tissue samples, including fresh flower buds, roots, stems, and leaves (Table S1; Figures S3–S17). These findings indicate the entire absence of the colchicine biosynthetic pathway in lilies, providing conclusive evidence that Lilium species do not contain colchicine.

2.2. Identification of Bitter Compounds in Lilies

2.2.1. Sensory Evaluation of Tissue Localization of Bitterness

Aqueous solutions (1.0 mg/mL) were prepared from the bulb extracts of five lily varieties subjected to sensory evaluation by a panel of ten trained assessors. The results revealed that the Longyahong (LYH) variety exhibited the most pronounced bitterness, with an intensity score significantly higher than that of the other varieties. In comparison, L. lancifolium, Longya (LY), and Guiyanghong (GYH) presented only mild bitterness, whereas the Lanzhou (LZ) lily was found to be devoid of detectable bitterness (Figure 3a). To determine the tissue-specific distribution of the bitter constituents, extracts were separately prepared from the peel and pulp of LYH bulbs. Sensory analysis indicated that the peel extract was intensely bitter, whereas the pulp extract was non-bitter and possessed a mild sweet taste (Figure 3b). These findings collectively demonstrate that the bitterness in the LYH variety is predominantly localized in the bulb peel.

2.2.2. Identifying the Potential Bitter Compounds via HPLC-Q-TOF-MS

Comparative analysis of the total ion chromatograms (TICs) between the most bitter variety (LYH) and the non-bitter variety (LZ) revealed three prominent peaks (compounds 1–3). These constituents were highly abundant in LYH peel but were virtually undetectable in LZ (Figure 4a,c). Given the strong positive correlation between their abundance and the observed bitterness intensity across varieties, these three compounds were prioritized as candidate bitter principles. Consistent with this hypothesis, compounds 1–3 were also detected in L. lancifolium (Figure 4d), LY (Figure 4e), and GYH (Figure 4f), all of which exhibit mild bitterness. Further investigation into the tissue-specific distribution of these metabolites confirmed their localization, supporting their role as potential bitter ingredients. TIC comparison demonstrated that the majority of metabolites, including the candidate compounds 1–3, were concentrated in the peel, with only trace levels detected in the pulp (Figure 4a,b). This chemical profile aligns perfectly with the sensory evaluation, which indicated pronounced bitterness in the LYH peel extract but a non-bitter taste in the pulp. Consequently, the bulb peel of LYH was selected as the raw material for the subsequent targeted isolation and structural characterization of these potential bitter compounds.

2.2.3. Structural Characterization of Compounds 1 and 3

Compound 1 (Figure 5a) was isolated as a white powder. High-resolution ESI-MS analysis in positive ion mode displayed a protonated molecular ion at m/z 576.3895, [M + H]⁺, corresponding to the molecular formula of C₃₃H₅₄NO₇. The MS/MS spectrum exhibited a characteristic fragment ion at m/z 414.3358, indicating the loss of a glucose moiety. Subsequent fragment ions at m/z 396.3261, 147.1167, and 98.0965 further supported its identification as a terpenoid alkaloid (Figure S1a). In the ¹³C NMR spectrum (Figure S1b), signals at δ 140.7 and 121.7 confirmed the presence of a carbon–carbon double bond. The ¹H NMR spectrum (Figure S1c) displayed four characteristic methyl signals δ 1.09 (d, 3H), 0.94 (s, 3H), 0.90 (d, 3H), and 0.72 (s, 3H). Based on comprehensive spectroscopic analysis and comparison with previously reported data, compound 1 was identified as (22R, 25R)-spiro-5-en-3β-ol-O-β-D-glucopyranoside [40]. Although previously reported in Lilium candidum, this is the first documented occurrence of this compound in the LYH variety.

Compound 3 (Figure 5b) was also obtained as a white powder. ESI-MS analysis yielded a protonated molecular ion at m/z 883.4763, [M + H]⁺, consistent with the molecular formula C₄₅H₇₁O₁₇. MS/MS analysis showed the successive loss of one rhamnose and two glucose units, yielding a key aglycone fragment ion at m/z 413.3042. Additional characteristic fragment ions at m/z 395.2941, 271.2053, and 253.1950 supported the classification of compound 3 as a diosgenin-type saponin (Figure S2). By combining the NMR data with literature values, compound 3 was identified as (25S)-spirostane-5-en-3β-27-tetramethyl-3-O-[(1→2)-α-L-rhamnopyranosyl-(1→6)-β-D-glucopyranosyl]-β-D-glucopyranoside [40]. This represents the first report of this compound in the LYH variety.

Notably, while compound 2 was identified as a potential candidate based on MS profiling and presumed to be an analog of 1, it could not be isolated with sufficient purity for full structural characterization during this study.

2.2.4. Sensory Evaluation of Compounds 1 and 3

Sensory evaluation of the two isolated constituents confirmed that both compounds 1 and 3 exhibited distinct bitterness at a concentration of 1.0 mg/mL. Compound 1 elicited the highest intensity bitterness with a mean score of 4.6, characterized as a moderate bitterness level, yet still perceptible bitterness intensity. These findings validate that both compounds are key bitter-tasting principles in lilies (Figure 3c), with compound 1 likely being a primary contributor to the overall bitter profile of the LYH variety.

3. Discussion

Historically, both the scientific literature and prevalent public perception have attributed the bitterness of lilies to the presence of colchicine. However, our comprehensive analysis provides definitive evidence refuting this long-standing association. Colchicine was not detected in any of the bulbs, stems, leaves, or flowers of the five lily species examined. This finding aligns with toxicological expectations; colchicine is a potent alkaloid with a narrow therapeutic index, where therapeutic and toxic dosages overlap significantly. Acute ingestion exceeding 0.5 mg/kg is associated with high fatality rates [41]. Given that no cases of colchicine poisoning from lily consumption have been documented, its natural occurrence in these plants is highly improbable. Our systematic review of 140 publications (spanning 1985–2025) further elucidates the origin of this misconception. Over 87% of these studies cited the presence of colchicine without providing original analytical confirmation. Among the 18 studies that did perform empirical testing (4 qualitative and 14 quantitative), the majority relied solely on HPLC-UV. This method is notoriously susceptible to false positives due to the chromatographic co-eluting of colchicine with other structurally similar compounds. Notably, even when advanced LC-MS/MS failed to detect the compound, some researchers still erroneously concluded that “trace amounts” were present [42]. Most decisively, recent genomic evidence confirms that lilies lack the essential biosynthesis gene cluster required for colchicine production [4]. Collectively, these multidimensional lines of evidence—analytical, toxicological, bibliographic, and genomic—conclusively demonstrate that colchicine is absent in lilies, thereby correcting a significant and persistent error in the botanical literature.

Having demonstrated that colchicine is not responsible for the bitterness in lilies, we sought to identify the actual chemical constituents underlying this trait. Our study revealed that the LYH variety exhibits bitterness—substantially higher than other examined varieties—which is primarily localized in the bulb peel. Comparative UPLC-Q-TOF-MS profiling successfully identified three candidate bitter components. Targeted isolation and structural characterization yielded two high-abundance constituents: compounds 1 and 3. Sensory evaluation confirmed both as bitter principles, with compound 1 eliciting a significantly stronger bitter response than compound 3. Although compound 2 was identified via MS/MS as a structural analog of 1, its insufficient purity precluded definitive sensory testing. Consistent with the established bitter nature of most alkaloids [43,44], our findings suggest that alkaloid content directly dictates the bitterness intensity in Lilium species. This research establishes, for the first time, that compounds 1 and 3—rather than colchicine—are the authentic bitter principles in lilies. These results provide a definitive molecular basis for quality evaluation, sensory control, and targeted breeding of edible lily cultivars.

Our conclusion that colchicine is not a natural constituent of Lilium is predicated on multi-dimensional converging evidence, extending beyond its mere absence in the five cultivated varieties. Our systematic analysis encompassed multiple tissues—including bulbs, roots, leaves, and flowers—to ensure spatial representation. Critically, we expanded our screening to include all 15 known biosynthetic precursors within the colchicine pathway. Their consistent absence of both the end-product and its essential intermediates across all samples strongly indicates that the colchicine biosynthetic pathway is non-functional, if not entirely absent, in these species. This chemical evidence is further corroborated by recent genomic data, which confirms the absence of key biosynthetic gene clusters in the Lilium genome [4]. Collectively, these analytical and genetic findings provide a robust foundation for reclassifying the chemical profile of the genus Lilium.

Regarding prior reports of colchicine in Lilium, we propose that methodological limitations likely account for the observed discrepancies. Many early studies relied on HPLC-UV co-elution for identification [19,27], a technique notoriously prone to false positives within complex plant matrices. In contrast, our study employed high-resolution UPLC-Q-TOF-MS, providing accurate mass measurements and MS/MS fragmentation patterns that offer superior chemical specificity, thereby minimizing the risk of misidentification. Consequently, we contend that these historical discrepancies stem from analytical artifacts rather than biological variability.

While plant secondary metabolism is inherently influenced by environmental factors, the consistent absence of the entire colchicine biosynthetic pathway—corroborated by genomic data—strongly argues against its natural occurrence in the genus Lilium. Although specific spike-recovery experiments for colchicine were not conducted, the robustness of our analytical pipeline is supported by several factors: (i) the employment of a validated alkaloid extraction protocol (70% ethanol, ultrasonic); (ii) the simultaneous absence of all 15 biosynthetic precursors, which precludes the possibility of a selective extraction failure; and (iii) the successful identification of a diverse array of other secondary metabolites, including steroidal alkaloids and saponins, from the same extracts. This broad-spectrum metabolic coverage demonstrates general extraction efficacy. Collectively, these findings suggest that the absence of colchicine is a fundamental biological trait of the genus, rather than a methodological artifact.

4. Materials and Methods

4.1. Plant Materials, Chemicals, and Reagents

Five Lilium varieties were collected from their representative cultivation regions in China: “Longyahong “(Lukou District, Zhuzhou City, Hunan Province), “Lanzhou” (Lanzhou City, Gansu Province), Lilium lancifolium (Longshan District, Hunan Province), “Guiyanghong” (Guiyang City, Guizhou Province), and “Longya” (Yangqiao Town, Hengdong County, Hunan Province). Additionally, Gloriosa superba and Colchicum autumnale (supplied by Nanjing University of Chinese Medicine) were utilized as positive controls for colchicine detection. All botanical samples were authenticated by Professor Zhixing Qing (Hunan Agricultural University). Voucher specimens were deposited at the Herbarium of Hunan Agricultural University (HNAU) with the following accession numbers: HNAU-LILY-001–005, HNAU-GS-001, and HNAU-CA-001.

The colchicine reference standard was purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). Analytical-grade ethanol, methanol, and dichloromethane were procured from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). LC-MS-grade acetonitrile, formic acid, and methanol were obtained from Merck KGaA (Darmstadt, Germany). Silica gel (100–200 and 300–400 mesh) was sourced from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China)

4.2. Instrumentation

Liquid chromatography–mass spectrometry was performed using an Agilent 1290 Infinity II UHPLC system coupled to an Agilent 6530 Quadrupole Time-of-Flight (Q-TOF) mass spectrometer (Agilent Technologies, Santa Clara, CA, USA). Ultrapure water was prepared using a Milli-Q Advantage A10 system (Millipore, Billerica, MA, USA). Sample preparation and processing were conducted using the following equipment: an ultrasonic cleaner (Kunshan Ultrasonic Instrument Co., Ltd., Kunshan, China); an RV 10 rotary evaporator (IKA-Werke GmbH & Co. KG, Staufen, Germany) equipped with an SHZ-D (III) water-circulating vacuum pump (Gongyi Yuhua Instrument Co., Ltd., Gongyi, China); a Scientz-10 N freeze dryer (Ningbo Scientz Biotechnology Co., Ltd., Ningbo, China); and a PL230 analytical balance (Mettler-Toledo, Greifensee, Switzerland).

4.3. Sample Preparation and Extraction

4.3.1. Pre-Treatment of Lily Tissues

Fresh plant materials (bulbs, roots, stems, flowers, and leaves of the five studied varieties) were transported to the laboratory on ice immediately following collection. After removing damaged or soiled tissues, samples were thoroughly rinsed with running tap water and subsequently washed with distilled water. The materials were dissected into uniform pieces (approximately 0.5 cm³ for bulbs, roots, and stems; leaves and flowers remained intact), flash-frozen in liquid nitrogen, and stored at −80 °C. The positive control materials (G. superba and C. autumnale) were processed using the identical extraction protocol to ensure methodological consistency.

The frozen tissues were lyophilized at −65 °C under high vacuum (20 Pa) for 48 h and ground into a fine powder. For extraction, 0.1000 g of each powdered sample was accurately weighed and suspended in 10 mL of 70% ethanol (v/v). Ultrasonic-assisted extraction was performed for 1 h. The resulting extracts (approximately 1.5 mL) were collected via syringe and filtered through a 0.22 μm organic membrane filter; the initial 2–3 drops of filtrate were discarded to ensure equilibrium. The final extracts were transferred into autosampler vials, sealed, and stored for UPLC-Q-TOF-MS analysis.

4.3.2. HPLC-Q-TOF-MS Conditions

Chromatographic separation was performed on an Agilent 1290 Infinity II UHPLC system (Agilent Technologies, Aanta Clara, CA, USA). Separation was achieved using an XAqua-C18 column (150 mm × 2.1 mm, 2.8 μm; Agilent Technologies) maintained at 30 °C. The mobile phase consisted of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B). The gradient elution program was set as follows: 0–15 min, 5–25% B; 15–27 min, 25–38% B; 27–35 min, 38–90% B. The flow rate was set at 0.3 mL/min, with an injection volume of 1 μL. Mass spectrometry detection was carried out on an Agilent 6530 Q-TOF mass spectrometer equipped with an electrospray ionization (ESI) source operated in positive ion mode. Data acquisition was performed in centroid mode across a mass range of m/z 100–1000. The optimized source parameters were as follows: gas temperature, 350 °C; drying gas flow, 12 L/min; nebulizer gas flow, 10 L/min; capillary voltage, 4000 V; fragmentor voltage, 135 V; skimmer voltage, 65 V; and the collision energies, 10 V, 20 V, and 30 V. To ensure high mass accuracy (<5 ppm), internal reference mass correction was applied using continuously infused reference ions at m/z 112.9855 and 966.0007

4.3.3. Targeted Screening Strategy for Colchicine and Its Biosynthetic Precursors

A dual-targeted strategy was employed to rigorously verify the presence or absence of colchicine in the studied samples. First, direct detection of colchicine was performed. The chromatographic retention time, accurate mass (m/z = 400.1755, [M + H]⁺), and MS/MS fragmentation pattern were first established using an authentic colchicine standard. These parameters were then applied to screen the extracted ion chromatograms (EICs) of all Lilium tissues (bulbs, stems, leaves, and flowers). The corms of C. autumnale and tubers of G. superba were analyzed in parallel as positive controls. Any peaks matching the theoretical mass of colchicine were further verified by comparing their retention times and MS/MS spectra against the standard. Second, to corroborate the chemical profiling, the screening was expanded to include the entire biosynthetic pathway. Based on the established pathway in C. autumnale, we systematically screened for 15 known biosynthetic precursors (listed in Figures S3–S17). EICs were generated for the theoretically accurate masses of these intermediates across all Lilium samples and positive controls. Identification was achieved by matching accurate mass and, where available, MS/MS fragmentation patterns. This comprehensive approach ensured a robust evaluation of colchicine metabolism in the genus Lilium.

4.4. Preparation of Crude Extracts for Bitter Compound Identification

4.4.1. Extraction of Lilium Bulbs and Positive Controls

Fresh bulbs (2.0 kg) were collected for each of the five Lilium varieties (LYH, LZ, L. lancifolium, LY, and GYH) and G. superba and C. autumnale. Additionally, underground storage organs of G. superba and C. autumnale were processed for comparison. The scales were peeled, dried at 65 °C, and pulverized using a mechanical grinder. The resulting powder was suspended in 4 L of 70% ethanol (v/v). The mixture was subjected to ultrasonic-assisted extraction for 1 h, followed by a 4 h static maceration period at room temperature. The supernatant was separated from the residue by means of filtration. The solvent was subsequently removed using a rotary evaporator, and the remaining aqueous phase was lyophilized to yield crude extracts. The extraction yields (w/w; dry weight basis) were as follows: LYH (1.28%), LZ (5.65%), L. lancifolium (1.59%), LY (1.05%), GYH (2.41%), G. superba (12.18%), and C. autumnale (11.49%). Mass spectrometric profiling of these extracts was performed using the UPLC-Q-TOF-MS conditions described in Section 4.3.2.

4.4.2. Preparation of Peel and Pulp Extracts of Longyahong Bulbs

A total of 20 kg of fresh LYH bulbs were cleaned, and all damaged tissues were removed. To facilitate the separation of tissues, the bulbs were subjected to a freeze–thaw cycle, being frozen at −20 °C for 12 h and subsequently thawed at room temperature. The peel (457 g) and pulp (19.35 kg) were separated and vacuum dried. The dried materials were pulverized, and extraction was performed using 70% ethanol at a solid-to-liquid ratio of 1:10 (w/v). The mixture underwent ultrasonic-assisted extraction for 1 h, and the process was repeated three times to ensure exhaustive extraction. The resulting filtrates were pooled and concentrated under reduced pressure. The yield for the peel extract was approximately 23.84% (resulting in ~108.9 g of extract), and for the pulp extract, it was approximately 0.82% (resulting in ~158.6 g of extract). Chemical profiling of both extracts was conducted via UPLC-Q-TOF-MS as described in Section 4.3.2.

4.5. Sensory Evaluation

A trained sensory panel comprising 10 assessors (5 males and 5 females) was established following a sensitivity screening for bitter and sweet taste perception. To minimize carry-over effects and assessment errors, panelists were provided with mineral water for palate cleansing prior to and between evaluations. A mandatory 5-min rest interval was enforced between samples to prevent sensory fatigue. The evaluation protocol strictly adhered to established methodologies [45,46,47]. Test solutions were prepared by dissolving the crude extracts and isolated compounds (Compounds 1 and 3) in potable water to a final concentration of 1.0 mg/mL. The samples were presented in randomized aliquots to eliminate order bias. Each sample was evaluated in triplicate by each panelist.

The bitterness intensity was scored using the following scale: 0–2 points: no bitterness or off-flavor; >2–4 points: slight bitterness, acceptable; >4–6 points: moderate bitterness, barely acceptable; >6–8 points: noticeable bitterness, somewhat unacceptable; >8–10 points: intense bitterness, unacceptable.

4.6. Targeted Isolation of Potential Bitter Compounds

Preliminary sensory evaluation confirmed LYH as the variety with the highest bitterness intensity, whereas LZ was identified as non-bitter. A comparative analysis of their HPLC-MS total ion chromatograms (TICs) revealed that compounds 1–3 were highly abundant in the bitter LYH variety but were negligible or absent in the non-bitter LZ variety (Figure 4a,c). Consequently, these three differential constituents were prioritized as targets for isolation to verify their contribution to the observed bitterness.

4.6.1. Preliminary Fractionation by Means of Silica Gel Chromatography

The peel extract of LYH (100 g) was fractionated on a silica gel column (100 mm i.d., 100–200 mesh). Gradient elution was performed using a dichloromethane–methanol system (ratios ranging from 10:1 to 0:1, v/v), with each approximately 1.5 L of eluent collected for each gradient step. The eluates were monitored by means of thin-layer chromatography (TLC) and visualized by spraying with 10% H₂SO₄ in ethanol followed by heating. Fractions exhibiting similar TLC patterns were combined, yielding 20 sub-fractions (Fr. A1–A20). Subsequent MS screening indicated that the target bitter candidates (Compounds 1–3) were predominantly enriched in A10, A15, and A16.

4.6.2. Isolation of Compound 1

Target Compound 1, which was initially localized in Fr. A15 based on MS screening, was subject to further purification. The fraction was chromatographed on a silica gel column (50 mm i.d., 300–400 mesh) using an isocratic elution system of dichloromethane–methanol (2:1, v/v). Eluates exhibiting identical TLC profiles (visualized with 10% H₂SO₄ in ethanol) were combined to yield five sub-fractions (Fr. B1–B5). MS monitoring confirmed that Compound 1 was concentrated in Fr. B3. This sub-fraction was then re-chromatographed under the same conditions (silica gel, 300–400 mesh; DCM–MeOH 2:1) to afford 57.23 mg of pure compound 1.

The structure of compound 1 was elucidated through comprehensive spectroscopic analysis. The complete ¹H and ¹³C spectrum data for compound 1 are summarized as follows: ¹H-NMR (400 MHz, DMSO): δ 5.33 (s, 1H), δ 5.02 (s, 1H), δ 4.36–4.34 (d, J = 8, 1H), δ 4.30–4.23 (m, 2H), δ 4.03–3.96 (m, 2H), δ 3.65–3.62 (m, J = 12, 6H), δ 3.21 (s, 1H), δ 2.92–2.87 (m, 1H), δ 2.41–2.38 (d, J = 12, 1H), δ 2.18–2.12 (t, J = 24, 2H), δ2.01 (s, 1H), δ 1.92–1.89 (t, J = 12, 2H), δ 1.62–1.52 (m, 8H), δ 1.46–1.42 (d, J = 12, 2H), δ 1.42–1.30 (m, 5H), δ 1.23 (s, 2H), δ 1.17 (s, 1H), δ 1.15–1.13 (m, 2H), δ 1.00–0.90 (t, 12H). ¹³C-NMR (100 MHz, DMSO): δ 140.7, 121.7, 109.3, 100.6, 80.7, 77.9, 77.0, 76.8, 72.3, 70.8, 62.1, 61.3, 56.2, 50.0, 41.6, 40.2, 38.1, 38.0, 37.2, 36.8, 31.9, 31.9, 31.8, 31.8, 31.8, 31.4, 30.9, 29.4, 20.8, 19.3, 18.1, 16.4, 15.0.

4.6.3. Isolation of Compound 3

Compound 3, also detected in Fr. A16, was isolated using a similar gel column chromatography setup (50 mm i.d., 300–400 mesh). The column was eluted isocratically with a dichloromethane–methanol system (5:1, v/v). Eluates showing similar TLC characteristics were combined to yield eight sub-fractions (Fr. C1–C8). MS analysis localized the target compound 3 in Fr. C8. This fraction was further purified on a smaller silica gel column (24 mm i.d.), using the same isocratic conditions (DCM–MeOH 5:1). This final purification step yielded 76.4 mg of compound 3 as a pure isolate.

The structure of compound 3 was elucidated through comprehensive spectroscopic analysis, and its ¹³C NMR and ¹H NMR spectral data are as follows: 1H-NMR (400 MHz, DMSO): δ 5.33 (s, 1H), δ 5.04 (s, 1H), δ 4.36–4.24 (s, 2H), δ 4.03–3.96 (s, 1H), δ 3.85–3.74 (m, 2H), δ 3.66–3.55 (m, 5H), δ 3.49–3.35 (m, 9H), δ 3.22–3.14 (m, 3H), δ 3.10–3.05 (m, 2H), δ 2.43–2,39 (m, 3H), δ 2.30–2.26 (m, 1H), δ 2.19–2.09 (m, 2H), δ 2.01–1.33 (m, 22H), δ 1.17–1.09 (m, 10H), δ 1.01–0.87 (m, 8H). 13C-NMR (100 MHz, DMSO): δ 140.8, 121.7, 121.7, 80.8, 80.8, 65.4, 62.6, 56.2, 50, 41.6, 38.1, 38.1, 38, 37.3, 37.3, 32, 32, 31.9, 31.4, 30.7, 29.5, 28, 20.9, 19.4, 18.2, 16.5, 15.1.

5. Conclusions

For nearly four decades, it has been a prevailing hypothesis that colchicine is present in Lilium species and serves as the source of their bitterness. In contrast, our study provides compelling evidence to refute this belief. We detected neither colchicine nor any of its 15 biosynthetic precursors in the examined lily tissues, confirming that this alkaloid does not naturally occur in these species and is unrelated to their bitter taste. To elucidate the true chemical basis of this bitterness, we focused on the LYH variety, which exhibited pronounced bitterness localized predominantly in the bulb peel. Through comparative LC-MS profiling, three high-abundance candidate compounds were identified, two of which were successfully purified via chromatographic isolation. This study represents the first systematic evidence demonstrating the absence of colchicine in edible lilies and identifies, for the first time, the primary compounds responsible for the bitterness in the LYH variety.