The following alkaloids were purchased from Cayman Chemical: scoulerine (item no. 35140; batch no. 0626850-3), isocorypalmine (item no. 29886; batch no. 0694012-1), glaucine (item no. 17338; batch no. 0481648-13), tetrahydropalmatine (item no. 20535; batch no. 0491304-7), tetrahydroberberine (item no. 33157; batch no. 0605003-10), corydaline (item no. 27654; batch no. 0570888-1), tetrahydrocoptisine (item no. 34603; batch no. 0620069-2), protopine (hydrochloride) (item no. 23366; batch no. 0514853-5), columbamine (item no. 35032; batch no. 0626484-3), coptisine (chloride) (item no. 28424; batch no. 0568598-14), palmatine (chloride) (item no. 30318; batch no. 0585590-3) and dehydrocorydaline (chloride) (item no. 30972; batch no. 0594357-5). CY dietary supplements and the CY rhizome sample (Plum Dragon Herbs) were ordered from online retail markets. Chemicals and HPLC solvents were from Sigma Aldrich or Fisher.

CY dietary supplement sample 7 (2 g) was extracted with hot water (20 mL) by vortex mixing followed by centrifugation at 4,650 x g for 10 min. The supernatant was collected and the pellet was re-extracted with 10 mL of hot water. The supernatants were combined, centrifuged again (4,650 x g for 12 min) and loaded on a Waters Oasis WCX 6 cc cartridge (500 mg, 60 μm) that had been preconditioned with methanol and water. Two 2 g-aliquots of CY powder were extracted and processed in parallel. After loading the WCX cartridges were washed with 5% NH₄OH (6 mL) and eluted with methanol (6 mL) followed by a solvent of acetonitrile/methanol/formic acid (80/20/2; by vol.; 6 mL) to give the WCX “methanol” and “formic acid” eluates, respectively. Preparation and elution of the WCX cartridge was performed according to the manufacturer’s instructions. The eluates were evaporated under a stream of N₂ and dissolved in ca. 3 mL of methanol for the methanol eluate and ca. 150 μL of methanol for the formic acid eluate.

Semi-preparative RP-HPLC used an Agilent 1,200 Series HPLC system with a diode array detector and a Waters Symmetry C18 7 μm column (7.8 × 150 mm) eluted at 2 mL/min flow rate. For isolation of alkaloids of the methanol eluate the gradient used a solvent of water/acetonitrile/acetic acid 90/10/0.01 (by vol.) to 75/25/0.01 (by vol.) in 30 min, then changed to 20/80/0.01 (by vol.) in 2 min, hold 2 min, and return to starting solvent in 2 min. For the formic acid eluate the gradient used a solvent of 50 mM NH₄OAc pH 8/acetonitrile 90/10 (by vol.) changed to 60/40 (by vol.) in 40 min, then changed to 30/70 (by vol.) in 2 min, hold 2 min, and return to starting solvent in 2 min. Use of a basic HPLC solvent for alkaloid purification was suggested by ref. (Zhang et al., 2009). Peaks from the formic acid eluate were collected in tubes that contained 50 μL of a mixture of water/acetic acid (4/1, by vol.) in order to prevent isolation artifacts occurring at basic pH (Marek et al., 2003).

The alkaloids collected from the methanol eluate were extracted from RP-HPLC solvent by evaporating acetonitrile under a stream of N₂, adding water, and loading on a pre-conditioned Waters Oasis HLB 3 cc cartridge (60 mg). The cartridge was washed with water and alkaloids were eluted with methanol. Alkaloids from the formic acid eluate were extracted from RP-HPLC solvent by evaporation of acetonitrile and adding water and loading on a pre-conditioned Waters Oasis WCX 3 cc cartridge (60 mg). The cartridge was washed with water, and alkaloids were eluted with acetonitrile containing 1% acetic acid (750 μL) followed by acetonitrile (750 μL).

Collected alkaloids were checked for purity using a Waters Symmetry C18 5 μm column (4.6 × 250 mm) eluted at 1 mL/min flow rate with a gradient of water/acetonitrile/acetic acid 90/10/0.01 (by vol.) to 60/40/0.01 (by vol.) in 18 min. When isolated compounds had a purity (RP-HPLC UV 205 nm) less than 95% they were repurified. Re-purification was achieved using the same column and gradient as in the original purification or isocratic elution with a solvent of water/acetonitrile/acetic acid 85/15/0.01 (by vol.). Dehydrocorybulbine was re-purified using a gradient of 50 mM NH₄OAc pH 8/acetonitrile 70/30 (by vol.) to 60/40 (by vol.) in 20 min; collected fractions were acidified using a mixture of water/acetic acid (4/1, by vol.) as described above.

CY dietary supplements (25 mg) were weighed into a 2 mL Eppendorf tube. Hot water (ca 80°C; 1 mL) was added, spiked with internal standards (methyl nicotinate (MeN): 40 μg; nitidine chloride: 20 μg), and samples were vortex mixed for 1–2 min. After centrifugation (2 min at 17,000 x g) the supernatant was loaded on a pre-conditioned (3 mL of methanol followed by 3 mL of water) Waters Oasis WCX 3cc cartridge (60 mg, 30 μm). The cartridge was washed with 1 mL of 5% NH₄OH and eluted with 1 mL of methanol followed by 1 mL of acetonitrile/methanol/formic acid (80/20/2, by vol.). Each supplement sample was extracted in two independent repeats. For the generation of calibration curves, different amounts of alkaloids were mixed with a fixed amount of standard (MeN or nitidine), and injected on RP-HPLC. Glaucine, tetrahydropalmatine, scoulerine, and corydaline standard curves were used to quantify alkaloids in the methanol eluate. Protopine, coptisine, and palmatine standard curves were used to quantify the alkaloids in the formic acid eluate.

For quantification an aliquot of 20 μL of each eluate (≈1 mL) was injected on RP-HPLC using an Agilent 1200 SL HPLC system equipped with a Water Symmetry Shield C18 5 μm column (4.6 × 250 mm) and a diode array detector with monitoring at 205, 220, 235, 270, 340, and 430 nm wavelength. The flow rate was 1 mL/min and compounds were eluted using a linear gradient of water/acetonitrile/acetic acid from 90/10/0.01 (by vol.) to 70/30/0.01 (by vol.) within 28 min followed by a wash step using water/acetonitrile/acetic acid 20/80/0.01 (by vol.). Alkaloids were quantified based on their peak area at 205, 220, or 340 nm.

CY rhizome was obtained in slices of 2–3 mm thickness and about 1 cm in diameter. Three slices were ground into a fine powder using a mortar and pestle. The powder (25 mg) was weighed into a 2 mL Eppendorf tube, spiked with internal standards, and alkaloids were extracted using 1 mL of hot water followed by WCX cartridge fractionation into methanol and formic acid eluates as described above.

High-resolution LC-MS of alkaloids was performed using a Q Exactive HF Hybrid Quadrupole-Orbitrap instrument (Thermo Fisher Scientific) operated in the electrospray ionization (ESI) positive mode. Prior to analysis, the instrument was calibrated with an ESI-positive ion calibration solution. A Waters Symmetry C18 column (2.1 × 50 mm, 1.8 µm) was eluted at room temperature with a gradient of acetonitrile in water/0.1% formic acid changed from 10% acetonitrile to 95% in 3 min at a flow rate of 0.4 mL/min.

Purified alkaloids were dissolved in deuterated solvents (CD₃OD, CD₃CN, CDCl₃, or mixtures thereof) for NMR analysis. NMR spectra were recorded using a Bruker AV-II 600 MHz spectrometer equipped with a cryoprobe. Chemical shifts (δ value) are given relative to the residual non-deuterated solvent and are reported in parts per million (ppm). Coupling constants (J) are given in Hertz (Hz). Pulse frequencies were taken from the Bruker library.

A survey of three local health food and supplement stores in Nashville, TN conducted in March 2024 did not find any CY products available. Therefore, all CY dietary supplements were purchased online. Products containing CY in combination with other extracts or bioactive ingredients were excluded from the analysis. Fourteen different products were selected, capturing a majority of the products available. By comparison, the NIH/Office of Dietary Supplements Dietary Supplement Label Database (https://dsld.od.nih.gov/, accessed March 2024) listed about 13 different products on the market that contain CY as the only bioactive ingredient. Thus, the selected products provided a meaningful survey of CY dietary supplements available to consumers in the United States at the time.

The 14 selected products included open powders or granules (4), capsules (8), and two liquid formulations. Table 1 provides an overview of the product names, serving sizes, and supplement facts information retrieved from the labels. The “supplement facts” labels of samples 8, 10, and 12 identified the products as “Corydalis” yet not explicitly as “Corydalis yanhusuo”. Sample 8 was referred to as “Yan Hu Suo” elsewhere on the label, implying that the product contained an extract of CY. The two other samples (10 and 12) were assumed to contain CY, or might be mistaken by consumers to do so, and thus were included in the analysis.

We isolated and identified 19 alkaloids from one of the CY dietary supplements (sample 7 in Table 1) that had shown high alkaloid content in a preliminary analysis. Alkaloids were solubilized using hot water and separated from other material by binding to a weak cation exchange (WCX) cartridge. Alkaloids were eluted from the WCX cartridge using methanol to obtain tertiary amine alkaloids (i.e., tetrahydroprotoberberines; “methanol eluate”) followed by elution with a solvent of acetonitrile/methanol/formic acid (80/20/2, by vol.) to provide mostly quaternary amine alkaloids (i.e., protoberberines and others; “formic acid eluate”). Alkaloids were purified using semi-preparative RP-HPLC; structural identification of isolated alkaloids was achieved using NMR and HR-MS analyses and/or comparison to authentic standards when available (see Supplementary Material). The structures of nine alkaloids (M1-M9) isolated from the methanol and 10 alkaloids (F1-F10) from the formic acid WCX eluates, respectively, are shown in Figure 1.

Alkaloids were solubilized from the dietary supplements using hot water and applied to a WCX column to provide a “methanol” and “formic acid” eluate for separate RP-HPLC analysis. The method used for quantification of alkaloids is illustrated in Figure 2, showing RP-HPLC analysis with UV/Vis diode array detection of sample 7 as example. Both the tertiary (methanol eluate; Figure 2A) and quaternary (formic acid eluate; Figure 2B) amine alkaloids obtained from the WCX cartridge were well resolved. Resolution and peak shape of alkaloids was enhanced by using a Symmetry Shield C18 column (Waters) that provides embedded polar carbamate groups at the base of each ligand. The embedded polar groups bind water molecules that “shield” charged silanol groups and therefore decrease the amount of tailing with charged basic compounds. Regardless of what HPLC column was used we noticed that alkaloids showed a tendency to switch elution order and change retention times beyond what was expected upon variations in chromatographic conditions, even when variations were minor. Similar effects were observed when different amounts of alkaloids were injected. Thus, caution should be used in peak identification especially since the UV spectra of many of the alkaloids are similar. Two internal standards, methyl nicotinate and nitidine, were added to the supplements for quantification of alkaloids in the WCX methanol and formic acid eluates, respectively. The standards were present only in the intended WCX eluates, were well resolved from the alkaloids of interest, had suitable UV/Vis spectra, and were absent from a list of known alkaloids of CY (Feng et al., 2023).

Alkaloids were grouped by their UV/Vis spectra for quantification. Alkaloids had one of two general types of UV/Vis spectra, based on the degree of saturation in the third ring. Protoberberines are yellow in color with several maxima around 230 nm, 270 nm, 340 nm, and 430 nm whereas the saturated tetrahydroprotoberberines have mostly end absorbance in the UV and are lacking color (Supplementary Figures 1, 2). Accordingly, alkaloids were quantified using the absorbance of the chromatographic peaks at 205 nm, 220 nm, or 340 nm as appropriate. For the methanol eluate, a calibration curve for scoulerine (M1) was used to quantify M1, M2, M3, and M4 at 205 nm. Glaucine (M5) and tetrahydropalmatine (M6) calibration curves were used to quantify M5 and M6, respectively, at 220 nm. A calibration curve for corydaline (M8) was used to quantify M7, M8, and M9 at 205 nm. For alkaloids in the formic acid eluate, a calibration curve for protopine (F4) was used to quantify F1, F2, F3, F4, and F5 at 205 nm. A calibration curve for coptisine (F6) was used to quantify F6, F7, and F8 at 340 nm. A calibration curve for palmatine (F9) was used to quantify F9 and F10 at 340 nm (see Supplementary Material). The LOQ was between 5 and 6 ng per injection (20 μL) on HPLC.

A graphical summary of the quantification results is shown in Figure 3. The content of alkaloids in the methanol and formic acid eluates of the CY dietary supplements is listed in Tables 2, 3, respectively. The most abundant alkaloids in the CY dietary supplements were tetrahydrocolumbamine (M4), glaucine (M5), tetrahydropalmatine (M6), corydaline (M8), protopine (F4), 14-hydroxy-N-methylcanadine (F5), coptisine (F6), and dehydrocorydaline (F10) (Figures 3A, B). Based on their content of alkaloids the CY dietary supplements roughly fell into three groups (Figure 3C). A group of five supplements (samples 4, 5, 7, 11, 12) had an overall high and similar content of alkaloids (9.5 ± 1.6 mg/g). The remaining samples either had a much lower content of alkaloids (samples 1, 3, 6, 13, 14; 1.8 ± 0.9 mg/g), or the alkaloids were near or below the lower limit of quantification (samples 8, 9, 10). When a 4-fold higher amount of the latter three samples was extracted the results were unchanged.

The near absence of the alkaloids of interest is illustrated for sample 9 (Figures 4A, B), showing RP-HPLC chromatograms of the methanol and formic acid eluates of the WCX column. No other peaks with UV/Vis spectra characteristic of isoquinoline alkaloids were detected, indicating that the sample was largely devoid of this type of bioactive alkaloids. For sample 2 the unusually high content of tetrahydropalmatine together with an overall low to absent content of other alkaloids is shown in Figures 4C, D. Tetrahydropalmatine was present in sample 2 at about 5 mg/g (i.e., per suggested serving size), which was >10-times the amount of tetrahydropalmatine in the next highest sample. The fact that only one alkaloid was of high abundance in the sample suggested it was added as a pure compound (of synthetic or natural origin) and was not included as an extract from another botanical which would likely have provided additional alkaloids.

Three of the four samples with unusual or low to absent alkaloid content (samples 2, 8, and 9; sample 10 was no longer available) were repurchased in order to test a different batch for consistency of the findings. In repurchased sample 2 the excessive content of tetrahydropalmatine was no longer present while other alkaloids were increased. Sample 8, which did not contain any alkaloids in the original sample, showed an increased though low amount in the repurchased sample. For sample 9, which had very low content of alkaloids, the amount was even further decreased, and the repurchased sample contained only three alkaloids that were barely above the LOQ. Even though the product label explicitly mentioned “DHCB” (i.e., dehydrocorybulbine) sample 9 had the lowest content of dehydrocorybulbine of all samples, and the amount was decreased to below LOQ in the repurchased sample.

The alkaloids of the CY dietary supplements were compared to an alkaloid extract prepared from the rhizome of CY. The alkaloids in the supplements with high alkaloid content and in the CY rhizome were similar regarding type and their relative abundance (Figure 3). Quantitative analysis showed that the absolute amount of alkaloids in the supplements with high content of alkaloids (9.5 ± 1.6 mg/g) was similar to the rhizome (12.7 mg/g) (Tables 2, 3).