Differentiation of Lepidium meyenii (Maca) from Different Origins by Electrospray Ionization Mass Spectrometry with Principal Component Analysis.

In this paper, we report a high-speed, high-sensitivity, and economic method to authenticate Maca. After being extracted by ethanol, nutritional components of a dozen kinds of Maca were detected by electrospray ionization mass spectrometry. Combined with principal component analysis (PCA), these samples can be rapidly differentiated after selecting the origins and principal components in the projection of components 1 and 2. The result suggests that sample 2 from Lijiang gets the highest comprehensive score among the samples and sample 1 from Huize gets the lowest comprehensive score among the samples in positive-ion mass spectra. However, sample 2 from Peru gets the highest comprehensive score among the samples and sample 3 from Lijiang gets the lowest comprehensive score among the samples in negative-ion mass spectra. Compared with the PCA results, the data of negative-ion mass spectra can better differentiate these samples than those of positive-ion mass spectra. This method has the advantages of easy operation and high efficiency, which make it a differential tool in the fields of food safety, medicinal chemistry, and materials science.

Introduction

Maca (Lepidium meyenii), one of the Cruciferae family, contains some active components such as alkaloid, glucosinolate, sterol, and isothiocyanate, apart from macronutrients. Alkaloids in Maca contain macamide,[1] macaridine,[2] β-carbazoline,[3] imidazole alkaloid,[4−6] and macahydantoin.[7] Macamides, the representative component of Maca, are secondary amides formed by benzylamine and a fatty acid moiety with different hydrocarbon chain lengths and degrees of unsaturation.[8,9] Glucosinolate is an important secondary metabolite in Maca. Glucotropaeolin is considered to be the most abundant glucosinolate in Maca.[10]

The rich components give Maca several remarkable biological activities.[11] In fact, Maca has been used as a functional food to strengthen the human body, increase fertility, improve sexual function, prevent depression and anemia in the Andes for centuries. The pentane extract of Maca contains a number of macamides that may act on the endocannabinoid system by inhibitory activity on fatty acid amide hydrolase.[12] Sterols or estrogens in Maca have a positive effect on sexual behavior and spermatogenesis in mice and men.[1,13−15] Lep B in Maca can inhibit nitrite production in macrophages.[16] In conclusion, the promising pharmacological activity of Maca has gradually won the approval of the customer.

Nowadays, Maca is recommended as a food supplement and widely cultivated in many countries and areas, especially China because it is efficacious in treating cancer, anemia, gastritis, high blood pressure, erectile dysfunction, infertility, stress, and depression.[17−19] However, there are different climate and planting environments of Maca between China and Peru. Such differences may cause the different nutritional components of Maca. Hence, it is important to study the effect of planting environments on the nutritional components of Maca. Due to its high sensitivity, excellent analytical efficiency, and simple sample preparation, electrospray ionization mass spectrometry (ESIMS) was used to differentiate Maca from different origins.

Partial least squares discriminant analysis (PLS-DA) is a supervised analytical method which can achieve a classification performance by reducing the dimension of complex data.[20] It was reported that based on UV data of macamides, PLS-DA was designed to discriminate Maca from different origins.[21] Principal component analysis (PCA) uses the method of reduction of dimensionality to find a data exploration relationship between objects, to estimate the correlation structure of the variables, and to investigate how many components (a linear combination of original features) are necessary to explain the greater part of variance with a minimum loss of information.[22,23] In this paper, combined with the mass spectral data, PCA was applied to successfully differentiate a dozen kinds of samples with a simple extract technique and to study the effect of origins on the constituents of Maca.

Results and Discussion

A dozen kinds of Maca, including Peru 1, 2, 3, Huize 1, 2, 3, Lijiang 1, 2, 3, Shangri-La 1, 2, and 3 were selected as typical samples to analyze the nutritional components by ESIMS with PCA. Herein, 1, 2, and 3 indicate three color types of Maca, namely purple, yellow, and white, respectively. Peru, Huize, Lijiang, and Shangri-La show the origins of Maca.

Mass Spectrometry Analysis

The mass spectra for Peru samples in different ion modes are shown in Figure . Compared with methods of mass spectrometry (MS), ESIMS tends to form multiple adducts of analytes. We identify the compounds by comparison of the detected m/z to the calculated theoretical m/z, gained fragment ion information, and that reported in the published literature (Table S1 in the Supporting Information). The dominant peak at m/z 116.0706 corresponds to [Pro + H]+ in positive-ion ESIMS. The peaks at m/z 104.0496, 175.1190, and 381.0794 correspond to [Bn + H]+, [Arg + H]+, and [Suc + K]+, respectively. The peaks at m/z 252.1969 and 270.2074 correspond to [NBoa + H2O + H]+ and [NBoa +2H2O + H]+, which are called macamide in Maca by high-performance liquid chromatography (HPLC)–UV–MS/MS.[8] The peaks at m/z 277.1709 and 291.1855 correspond to [Lep A – Cl]+ and [Lep B – Cl]+, which are known to be imidazole alkaloids from the root of Maca by spectroscopic methods.[24] The peak at m/z 408.2663 corresponds to [NBD + K]+, which belongs to macamide in Maca by HPLC–UV–MS/MS.[8]

Figure 1

ESI mass spectra of ethanol extracts of Peru samples: (A) sample 1 in positive-ion mode, (B) sample 1 in negative-ion mode, (C) sample 2 in positive-ion mode, (D) sample 2 in negative-ion mode, (E) sample 3 in positive-ion mode, and (F) sample 3 in negative-ion mode.

In negative-ion ESIMS, the dominant peak at m/z 133.0142 corresponds to [HA – H]− in Figure B, the dominant peak at m/z 376.0366 corresponds to [HPGlu – H]− in Figure D, and the dominant peak at m/z 408.0428 corresponds to [BGlu – H]− in Figure F.[25,26] The peaks at m/z 179.0561, 255.2330, and 279.2320 correspond to [DA – H]−, [PA – H]−, and [LA – H]−, respectively, which were analyzed as methyl esters after hydrolysis in potassium hydroxide solution by gas chromatography (GC).[27,28] The peaks at m/z 341.1090 and 683.2241 correspond to [Suc – H]− and [Tetra + H2O – H]−. The peak at m/z 431.3086 corresponds to [Erg + Cl]−, which is determined to be sterol in Maca by GC.[11] The peaks at m/z 456.0629 and 474.0735 may correspond to [MBGlu + H2O – H]− and [MBGlu + 2H2O – H]−, which belong to glucosinolates by GC-MS.[26,29]

In conclusion, samples 1 and 2 from Peru contain the same nutritional components, including Bn, Pro, Arg, NBoa, Lep A, Lep B, Suc, HA, DA, PA, LA, HPGlu, BGlu, Erg, MBGlu, Tetra, and so on. Sample 3 from Peru contains the same nutritional components as those of samples 1 and 2, except NBD.

Figure shows the mass spectra for Huize samples by ESIMS in different modes. The peak of protonated Pro is observed as a dominant signal in positive-ion MS of Huize samples. The peak at m/z 370.3104 corresponds to [NBD + H]+, which is called macamides in Maca by HPLC–UV–MS/MS.[8] The peaks at m/z 346.3104 and 384.2897 may correspond to [NBha + H]+ and [NBOO + H]+, which are benzylated alkamides (macamides) from the tuber of Maca by one-dimensional and two-dimensional nuclear magnetic resonance spectroscopic analyses.[2]

Figure 2

ESI mass spectra of ethanol extracts of Huize samples: (A) sample 1 in positive-ion mode, (B) sample 1 in negative-ion mode, (C) sample 2 in positive-ion mode, (D) sample 2 in negative-ion mode, (E) sample 3 in positive-ion mode, and (F) sample 3 in negative-ion mode.

In negative-ion ESIMS, the signal of deprotonated HA is the base peak in Figure B,D, and the signal of deprotonated BGlu is the base peak in Figure F, respectively.

The nutritional components of sample 1 from Huize are Pro, Arg, HA, NBoa, Lep A, NBD, Suc, HA, DA, PA, LA, HPGlu, MBGlu, and Tetra. The nutritional components of sample 2 from Huize are Pro, Arg, HA, NBoa, Lep A, Lep B, Suc, NBOO, HA, DA, PA, LA, HPGlu, MBGlu, and Tetra. The nutritional components of sample 3 from Huize are Pro, Arg, HA, NBoa, Lep A, NBha, NBD, Suc, NBOO, HA, DA, PA, LA, HPGlu, MBGlu, and Tetra.

The peak of protonated Pro is observed as a dominant signal in positive-ion MS of Lijiang samples. The peak of deprotonated DA is observed as a dominant signal in Figure B, and the peak of deprotonated Suc is observed as a dominant signal in Figure D,F, respectively. The nutritional components of samples 1 and 2 from Lijiang are Pro, Arg, DA, NBoa, Lep B, Suc, HA, PA, LA, HPGlu, BGlu, MBGlu, and Tetra; however, those of sample 3 from Lijiang are Arg, Suc, HA, PA, LA, HPGlu, BGlu, MBGlu, and Tetra.

Figure 3

ESI mass spectra of ethanol extracts of Lijiang samples: (A) sample 1 in positive-ion mode, (B) sample 1 in negative-ion mode, (C) sample 2 in positive-ion mode, (D) sample 2 in negative-ion mode, (E) sample 3 in positive-ion mode, and (F) sample 3 in negative-ion mode.

The mass spectra for Shangri-La samples in different ion modes are given in Figure . The signal of protonated Pro is shown as the base peak in Figure A,C, and the signal of dechlorided Lep B is shown as the base peak in Figure E, respectively. The signal of deprotonated BGlu is shown as base peak in Figure B,D, and the signal of deprotonated Suc is shown as the base peak in Figure F, respectively. The result shows that these samples from Shangri-La contain the same nutritional components.

Figure 4

ESI mass spectra of ethanol extracts of Shangri-La samples: (A) sample 1 in positive-ion mode, (B) sample 1 in negative-ion mode, (C) sample 2 in positive-ion mode, (D) sample 2 in negative-ion mode, (E) sample 3 in positive-ion mode, and (F) sample 3 in negative-ion mode.

Principal Component Analysis

The nutritional components including amino acid, macamide, imidazole alkaloid, glucosinolate, oligosaccharide, and aliphatic acid in these samples were detected by ESIMS. These samples were extracted and analyzed six times at the same experimental conditions. Means and standard deviation values of the relative intensity of several peaks in the mass spectra of these samples are listed in Tables and 2. Data are shown as mean ± standard deviation. It was observed that the relative intensity of Pro is the greatest in the samples, except that in sample 1 from Shangri-La in Table . However, there is much difference among the analytical data of nutritional components in these samples in Table .

Table 1

Means and Standard Deviation Values of the Relative Intensity of the Peaks for Principal Components in the Positive-Ion Mass Spectra of Samples According to Their Origins

sample	Bn	Pro	Arg	NBoa	Lep A	Lep B	NBha	Suc
Peru 1	46.52 ± 0.29	100	61.95 ± 1.68	13.62 ± 0.21	36.49 ± 0.45	38.46 ± 1.01	25.42 ± 0.09	17.13 ± 0.71
Peru 2	27.62 ± 2.43	100	33.71 ± 0.79	23.39 ± 0.41	33.30 ± 0.98	37.10 ± 1.24	10.66 ± 1.22	27.73 ± 0.86
Peru 3	32.94 ± 1.21	100	44.09 ± 1.02	30.68 ± 0.32	30.79 ± 2.75	54.27 ± 0.19	20.17 ± 0.10	27.00 ± 1.42
Huize 1	15.64 ± 1.40	100	17.63 ± 0.72	14.25 ± 1.83	17.81 ± 1.05	8.88 ± 0.52	13.81 ± 1.45	19.49 ± 2.93
Huize 2	20.64 ± 1.03	100	16.07 ± 1.20	23.68 ± 1.11	34.97 ± 2.65	16.28 ± 1.03	6.93 ± 3.28	18.70±1.83
Huize 3	16.55 ± 0.94	100	16.61 ± 0.93	14.60 ± 2.37	31.00 ± 1.23	15.50 ± 0.55	14.30 ± 1.67	19.31 ± 2.01
Lijiang 1	30.68 ± 1.34	100	38.71 ± 3.38	11.79 ± 2.99	7.50 ± 2.91	16.46 ± 1.57	15.93 ± 2.21	27.45 ± 1.59
Lijiang 2	29.86 ± 2.08	100	35.38 ± 1.53	30.28 ± 0.67	35.95 ± 1.00	85.17 ± 2.71	10.64 ± 1.63	32.63 ± 2.30
Lijiang 3	24.94 ± 0.69	100	24.23 ± 2.11	32.58 ± 0.96	32.40 ± 1.51	65.23 ± 2.10	15.27 ± 1.74	16.32 ± 3.01
Shangri-La 1	21.98 ± 1.65	67.92 ± 0.95	30.38 ± 2.70	18.02 ± 3.03	50.58 ± 0.80	100	9.19 ± 0.37	18.91 ± 2.47
Shangri-La 2	29.51 ± 1.55	100	35.70 ± 2.19	14.66 ± 3.01	17.93 ± 0.45	23.34 ± 2.49	13.28 ± 1.54	33.36 ± 1.31
Shangri-La 3	30.63 ± 2.31	100	40.31 ± 2.11	14.79 ± 2.63	18.57 ± 0.23	29.42 ± 2.14	14.93 ± 1.81	30.25 ± 1.06

Table 2

Means and Standard Deviation Values of the Relative Intensity of the Peaks for Principal Components in the Negative-Ion Mass Spectra of Samples According to Their Origins

sample	HA	DA	PA	LA	Suc	HPGlu	BGlu	MBGlu	Tetra
Peru 1	54.37 ± 0.72	21.28 ± 1.66	28.32 ± 2.72	60.31 ± 1.85	80.52 ± 2.33	19.18 ± 2.11	100	34.10 ± 1.08	18.77 ± 0.76
Peru 2	79.75 ± 2.71	42.46 ± 2.22	45.23 ± 2.20	68.04 ± 2.79	68.20 ± 1.84	100	57.60 ± 0.85	47.38 ± 3.83	16.11 ± 1.79
Peru 3	100	33.31 ± 3.37	31.01 ± 2.46	54.34 ± 1.30	87.92 ± 1.29	27.81 ± 3.06	86.88 ± 3.04	35.64 ± 1.29	22.54 ± 1.34
Huize 1	66.19 ± 0.98	17.55 ± 2.80	28.07 ± 2.91	72.27 ± 3.74	69.29 ± 0.97	23.41 ± 2.27	100	39.51 ± 2.40	18.19 ± 1.53
Huize 2	100	27.62 ± 1.51	25.49 ± 3.16	75.70 ± 2.66	96.62 ± 1.25	24.58 ± 1.52	49.55 ± 2.17	48.85 ± 0.79	22.12 ± 1.19
Huize 3	100	28.67 ± 1.79	31.31 ± 3.81	92.79 ± 2.14	95.14 ± 1.38	17.51 ± 1.01	69.13 ± 1.51	44.02 ± 1.30	19.40 ± 3.21
Lijiang 1	81.32 ± 2.33	47.37 ± 2.65	27.22 ± 1.59	38.49 ± 1.13	16.97 ± 1.40	39.91 ± 1.39	100	54.35 ± 1.85	29.91 ± 0.75
Lijiang 2	90.90 ± 2.14	49.28 ± 2.23	26.44 ± 1.34	39.22 ± 3.30	100	32.05 ± 0.79	80.49 ± 1.68	55.55 ± 3.36	22.84 ± 1.87
Lijiang 3	93.24 ± 3.01	100	31.16 ± 1.69	36.78 ± 1.88	51.66 ± 0.73	17.93 ± 1.53	31.36 ± 2.50	36.92 ± 4.53	11.04 ± 2.06
Shangri-La 1	44.25 ± 2.07	21.77 ± 2.70	29.00 ± 3.65	52.53 ± 3.92	83.55 ± 2.97	22.86 ± 3.08	100	37.73 ± 3.39	13.72 ± 1.50
Shangri-La 2	48.56 ± 3.33	23.94 ± 2.36	41.97 ± 1.23	73.68 ± 3.13	89.77 ± 1.79	28.66 ± 0.60	100	44.81 ± 2.28	15.09 ± 2.54
Shangri-La 3	72.09 ± 1.82	45.56 ± 1.38	36.67 ± 1.33	38.86 ± 2.98	100	25.03 ± 2.72	42.14 ± 1.09	53.57 ± 1.67	18.99 ± 2.47

PCA was applied to analyze and process the mass spectral data of Maca to further discuss the difference of the nutritional components among these samples.[23,30] Four principal components with eigenvalues higher than one accounting for 94.71 and 78.90% of total variance are obtained in Tables and 4. From the loadings of the variables (Table ), mainly Bn and Arg are the dominant features in the first principal component, accounting for 37.89% of the total variability, Lep A and B dominate in the second principal component, representing 31.10% of the total variance, and Suc and NBoa in the third and fourth principal component, representing 14.17 and 11.55% of the total variance, respectively. In Table , DA is the dominant feature in the first principal component, accounting for 25.05% of the total variability, Tetra dominates in the second principal component, representing 21.19% of the total variance, and HPGlu and MBGlu dominate in the third and fourth principal component, representing 16.62 and 16.04% of the total variance.

Table 3

Loadings of the Features in the First Four Principal Components of the Positive-Ion Mass Spectra

compounds	PC1	PC2	PC3	PC4
Bn	0.946	–0.256	0.051	0.188
Arg	0.937	–0.236	–0.005	0.234
NBha	0.726	–0.431	–0.340	–0.246
Lep A	0.232	0.896	–0.142	0.110
Lep B	0.380	0.822	0.243	0.105
Pro	0.033	–0.729	0.321	–0.383
Suc	0.093	–0.364	0.784	0.021
NBoa	0.161	0.538	0.598	0.978
eigenvalue	3.410	2.799	1.275	1.039
cumulative % of variation	37.89	68.99	83.16	94.71

Table 4

Loadings of the Features in the First Four Principal Components of the Negative-Ion Mass Spectra

compounds	PC1	PC2	PC3	PC4
DA	0.862	–0.078	–0.037	–0.453
BGlu	–0.763	0.483	0.317	0.008
HA	0.680	0.122	–0.380	0.321
LA	–0.518	–0.446	–0.220	0.502
Tetra	0.153	0.769	0.114	0.584
PA	0.024	–0.733	0.599	0.034
HPGlu	0.247	–0.326	0.774	0.279
Suc	–0.329	–0.391	–0.450	0.279
MBGlu	0.481	0.112	0.222	0.675
eigenvalue	2.505	2.118	1.662	1.604
cumulative % of variation	25.05	46.24	62.86	78.90

The scores and ranks of the ingredients of Maca (Tables S2 and S3 in the Supporting Information) were used to evaluate the quality of the samples from different origins and colors. The result in positive-ion mass spectra shows that sample 2 from Lijiang gets the highest comprehensive score among the samples but sample 1 from Huize gets the lowest (Table S2 in the Supporting Information). The result in negative-ion mass spectra shows that sample 2 from Peru gets the highest comprehensive score among the samples but sample 3 from Lijiang gets the lowest (Table S3 in the Supporting Information).

The PCA plots were applied to further differentiate these samples. Scores (PC1 and PC2) corresponding to different origins of samples are also plotted in Figure . Samples from Huize are distributed in the lower left quadrant, and samples 2 and 3 from Huize are difficult to differentiate in Figure A. In addition, it is also difficult to differentiate among sample 3 from Lijiang and samples 2 and 3 from Shangri-La. However, in Figure B, except sample 1 from Peru and sample 3 from Shangri-La, other samples differentiate well. In conclusion, for these samples, each data item distributes in special regions and the center of each region is not overlapping in the PCA map. The result suggests that these samples can be rapidly differentiated by ESIMS combined with PCA.

Figure 5

Principal component analysis maps for a dozen kinds of samples. Correlations between the growing regions and principal components in the projection of principal components 1 and 2: (A) in the positive-ion mass spectra and (B) in the negative-ion mass spectra.

Compared with the PCA results, the data of negative-ion MS can better differentiate these samples than those of positive-ion MS. This method can also be applied to classify and distinguish food, drug, and material. The study of a novel extraction method of macamide and differentiation of Maca from different origins based on macamide is currently in progress in our laboratories.

Materials and Methods

All kinds of Maca were purchased from Fengxi Tea Shop (Kunming, China). Methanol was obtained from Thermo Fisher Scientific Inc. (Massachusetts). Other chemicals were from Tianjin Fengchuan Chemical Reagent Factory (Tianjin, China). All chemicals used were of highest purity and without further purification.

Maca samples were gained from high altitudes (2500–3500 m) and dried in the sun with less than 5% of water content. Ten individuals in the same batch were mixed together and then pulverized. Maca powder was sieved through a 60-mesh stainless steel sieve. To maximize the dissolution of nutritional components of Maca, 75% ethanol (3 × 10 mL) is used for extraction at room temperature by an ultrasonic method. Then, the extract is filtered and the filtrate is collected. The filtrate was diluted to 50 mL with 75% ethanol and filtered through a 0.22 μm membrane filter before analysis by ESIMS.

Mass spectral experiments were performed with a time-of-flight mass spectrometer (Bruker Daltonics, micrOTOF II). The experimental parameters were as follows: flow rate of the sample solution: 2 × 10–3 mL·min–1; flow rate of nitrogen: 4 L·min–1; temperature of dry gas: 200 °C; voltage of capillary: 2.6 kV in both positive and negative-ion modes; voltage off end plate offset: 0.5 kV. The MicrOTOF control software of the ESIMS instrument was used to record the full scan mass spectra at the mass range of 50–800 Da.

The software SPSS 24.0 was applied to differentiate Maca from different origins based on the nutritional components.