Comparative Analysis of the Physicochemical Properties and Metabolites of Farinose and Crisp Lotus Roots (Nelumbo nucifera Gaertn.) with Different Geographical Origins

Lotus roots are widely consumed vegetables because of their great taste and abundant nutrients, but their quality varies with the environments and cultivar. This study systematically compared farinose (Elian No. 5) and crisp (Elian No. 6) lotus root cultivars from three geographical origins. Pasting and texture characteristics verified that Elian No. 5 possessed lower hardness and lower ability to withstand shear stress and heating during cooking compared with Elian No. 6. Untargeted metabolite profiling was first performed using ultrahigh-performance liquid chromatography coupled with electrospray ionization quadrupole time-of-flight mass spectrometry (UPLC-Q-TOF-MS) combined with a Zeno trap. In total, 188 metabolites were identified based on the matching chemistry database. Multivariate analysis demonstrated that lotus roots from different cultivars and origins could be adequately distinguished. Sixty-one differential metabolites were identified among three Elian No. 5 samples, and 28 were identified among three Elian No. 6 samples. Isoscopoletin, scopoletin, and paprazine were the most differential metabolites between Elian No. 5 and Elian No. 6. These results can inform future research on the discrimination and utilization of lotus roots.


Introduction
Nelumbo nucifera Gaertn., a perennial aquatic crop belonging to the family Nelumbonaceae, is consumed worldwide, including its rhizomes, seeds, leaves, flowers, and stamens [1,2]. Among them, the edible rhizome, also called the lotus root, is rich in starch, protein, vitamins, and some bioactive substances, such as beta carotene, polysaccharides, flavonoids, phenols, and triterpenes [3][4][5]. Coupled with its fascinating color, great taste, and potential antioxidant capacity, lotus root is accepted as a popular vegetable and fruit across the globe [6].
In recent decades, numerous studies have focused on increasing the shelf life of lotus root, and functional constituent analyses of lotus root have made good progress [7,8]. However, the nutrient concentrations in lotus roots vary among cultivars and are also affected by geographical locations, growing conditions, etc. [9]. To date, more than 300 kinds of lotus root cultivars are available from China and are distributed in the regions surrounding the mid-down Yangtse River, including Hubei, Jiangsu, Zhejiang, Anhui, Jiangxi, and Hunan provinces [2,10]. According to their taste and texture qualities, lotus roots can be broadly divided into farinose and crisp lotus roots. Generally, farinose lotus roots are cooked as filtered by deionized water at least four times until the supernatant was transparent. The mixture was repeatedly precipitated for 1 h, and the residue was collected and freeze-dried after removing the surface impurities. Finally, the freeze-dried pieces and starches were ground into powder using a mill (Tube-Mill 100, IKA, Staufen, Germany) and passed through a 100-mesh sieve. Then, the lotus root flours and isolated starches were stored at −20 • C for subsequent analysis.

Lotus Root Flours and Isolated Starch Preparation
Lotus root samples were washed with tap water to remove the attached soil. Afte drying the surface water, the peeled lotus roots were cut into pieces, and some of the sam ples were immediately stored at −20 °C for 48 h and then freeze-dried (FD5-3, Gold SIM (China), Beijing, China). Some were homogenized for starch preparation. In detail, th homogenate solution after filtration was precipitated for 1 h, and the separated residu was filtered by deionized water at least four times until the supernatant was transparen The mixture was repeatedly precipitated for 1 h, and the residue was collected and freeze dried after removing the surface impurities. Finally, the freeze-dried pieces and starche were ground into powder using a mill (Tube-Mill 100, IKA, Staufen, Germany) and passe through a 100-mesh sieve. Then, the lotus root flours and isolated starches were stored a −20 °C for subsequent analysis.

Pasting Properties Analysis
The pasting properties of lotus root flours and isolated starches were investigate with a Rapid Visco-Analyzer (RVA) (Perten, RVA4500, Macquarie Park, Australia). Sam ples (1.5 g) were weighed accurately, and 25 mL deionized water was added to prepar 6% (w/w) suspensions. The evaluation program was as follows: the suspensions were fir held at 50 °C for 2 min, heated to 95 °C at an increase rate of 12 °C/min, held for 2.5 min and subsequently cooled to 50 °C at the same rate. Seven pasting properties were chara terized, including peak viscosity (PV), trough viscosity (TV), final viscosity (FV), break down viscosity (BV), setback viscosity (SV), peak time (PT), and pasting temperatur

Lotus Root Flours and Isolated Starch Preparation
Lotus root samples were washed with tap water to remove the attached soil. Afte drying the surface water, the peeled lotus roots were cut into pieces, and some of the sam ples were immediately stored at −20 °C for 48 h and then freeze-dried (FD5-3, Gold SIM (China), Beijing, China). Some were homogenized for starch preparation. In detail, th homogenate solution after filtration was precipitated for 1 h, and the separated residu was filtered by deionized water at least four times until the supernatant was transparen The mixture was repeatedly precipitated for 1 h, and the residue was collected and freez dried after removing the surface impurities. Finally, the freeze-dried pieces and starche were ground into powder using a mill (Tube-Mill 100, IKA, Staufen, Germany) and passe through a 100-mesh sieve. Then, the lotus root flours and isolated starches were stored a −20 °C for subsequent analysis.

Pasting Properties Analysis
The pasting properties of lotus root flours and isolated starches were investigate with a Rapid Visco-Analyzer (RVA) (Perten, RVA4500, Macquarie Park, Australia). Sam ples (1.5 g) were weighed accurately, and 25 mL deionized water was added to prepar 6% (w/w) suspensions. The evaluation program was as follows: the suspensions were fir held at 50 °C for 2 min, heated to 95 °C at an increase rate of 12 °C/min, held for 2.5 min and subsequently cooled to 50 °C at the same rate. Seven pasting properties were chara terized, including peak viscosity (PV), trough viscosity (TV), final viscosity (FV), break down viscosity (BV), setback viscosity (SV), peak time (PT), and pasting temperatur

Lotus Root Flours and Isolated Starch Preparation
Lotus root samples were washed with tap water to remove the attached soil. Afte drying the surface water, the peeled lotus roots were cut into pieces, and some of the sam ples were immediately stored at −20 °C for 48 h and then freeze-dried (FD5-3, Gold SIM (China), Beijing, China). Some were homogenized for starch preparation. In detail, th homogenate solution after filtration was precipitated for 1 h, and the separated residu was filtered by deionized water at least four times until the supernatant was transparen The mixture was repeatedly precipitated for 1 h, and the residue was collected and freez dried after removing the surface impurities. Finally, the freeze-dried pieces and starche were ground into powder using a mill (Tube-Mill 100, IKA, Staufen, Germany) and passe through a 100-mesh sieve. Then, the lotus root flours and isolated starches were stored −20 °C for subsequent analysis.

Pasting Properties Analysis
The pasting properties of lotus root flours and isolated starches were investigate with a Rapid Visco-Analyzer (RVA) (Perten, RVA4500, Macquarie Park, Australia). Sam ples (1.5 g) were weighed accurately, and 25 mL deionized water was added to prepar 6% (w/w) suspensions. The evaluation program was as follows: the suspensions were fir held at 50 °C for 2 min, heated to 95 °C at an increase rate of 12 °C/min, held for 2.5 mi and subsequently cooled to 50 °C at the same rate. Seven pasting properties were chara terized, including peak viscosity (PV), trough viscosity (TV), final viscosity (FV), break down viscosity (BV), setback viscosity (SV), peak time (PT), and pasting temperatur

Lotus Root Flours and Isolated Starch Preparation
Lotus root samples were washed with tap water to remove the attached soil. Afte drying the surface water, the peeled lotus roots were cut into pieces, and some of the sam ples were immediately stored at −20 °C for 48 h and then freeze-dried (FD5-3, Gold SIM (China), Beijing, China). Some were homogenized for starch preparation. In detail, th homogenate solution after filtration was precipitated for 1 h, and the separated residu was filtered by deionized water at least four times until the supernatant was transparen The mixture was repeatedly precipitated for 1 h, and the residue was collected and freez dried after removing the surface impurities. Finally, the freeze-dried pieces and starche were ground into powder using a mill (Tube-Mill 100, IKA, Staufen, Germany) and passe through a 100-mesh sieve. Then, the lotus root flours and isolated starches were stored a −20 °C for subsequent analysis.

Pasting Properties Analysis
The pasting properties of lotus root flours and isolated starches were investigate with a Rapid Visco-Analyzer (RVA) (Perten, RVA4500, Macquarie Park, Australia). Sam ples (1.5 g) were weighed accurately, and 25 mL deionized water was added to prepar 6% (w/w) suspensions. The evaluation program was as follows: the suspensions were fir held at 50 °C for 2 min, heated to 95 °C at an increase rate of 12 °C/min, held for 2.5 min and subsequently cooled to 50 °C at the same rate. Seven pasting properties were chara terized, including peak viscosity (PV), trough viscosity (TV), final viscosity (FV), break down viscosity (BV), setback viscosity (SV), peak time (PT), and pasting temperatur

Lotus Root Flours and Isolated Starch Preparation
Lotus root samples were washed with tap water to remove the attached soil. Afte drying the surface water, the peeled lotus roots were cut into pieces, and some of the sam ples were immediately stored at −20 °C for 48 h and then freeze-dried (FD5-3, Gold SIM (China), Beijing, China). Some were homogenized for starch preparation. In detail, th homogenate solution after filtration was precipitated for 1 h, and the separated residu was filtered by deionized water at least four times until the supernatant was transparen The mixture was repeatedly precipitated for 1 h, and the residue was collected and freez dried after removing the surface impurities. Finally, the freeze-dried pieces and starche were ground into powder using a mill (Tube-Mill 100, IKA, Staufen, Germany) and passe through a 100-mesh sieve. Then, the lotus root flours and isolated starches were stored a −20 °C for subsequent analysis.

Pasting Properties Analysis
The pasting properties of lotus root flours and isolated starches were investigate with a Rapid Visco-Analyzer (RVA) (Perten, RVA4500, Macquarie Park, Australia). Sam ples (1.5 g) were weighed accurately, and 25 mL deionized water was added to prepar 6% (w/w) suspensions. The evaluation program was as follows: the suspensions were fir held at 50 °C for 2 min, heated to 95 °C at an increase rate of 12 °C/min, held for 2.5 min and subsequently cooled to 50 °C at the same rate. Seven pasting properties were chara terized, including peak viscosity (PV), trough viscosity (TV), final viscosity (FV), break down viscosity (BV), setback viscosity (SV), peak time (PT), and pasting temperatur (PTP). Each sample was analyzed in triplicate.

Lotus Root Flours and Isolated Starch Preparation
Lotus root samples were washed with tap water to remove the attached soil. Afte drying the surface water, the peeled lotus roots were cut into pieces, and some of the sam ples were immediately stored at −20 °C for 48 h and then freeze-dried (FD5-3, Gold SIM (China), Beijing, China). Some were homogenized for starch preparation. In detail, th homogenate solution after filtration was precipitated for 1 h, and the separated residu was filtered by deionized water at least four times until the supernatant was transparen The mixture was repeatedly precipitated for 1 h, and the residue was collected and freeze dried after removing the surface impurities. Finally, the freeze-dried pieces and starche were ground into powder using a mill (Tube-Mill 100, IKA, Staufen, Germany) and passe through a 100-mesh sieve. Then, the lotus root flours and isolated starches were stored a −20 °C for subsequent analysis.

Pasting Properties Analysis
The pasting properties of lotus root flours and isolated starches were investigate with a Rapid Visco-Analyzer (RVA) (Perten, RVA4500, Macquarie Park, Australia). Sam ples (1.5 g) were weighed accurately, and 25 mL deionized water was added to prepar 6% (w/w) suspensions. The evaluation program was as follows: the suspensions were fir held at 50 °C for 2 min, heated to 95 °C at an increase rate of 12 °C/min, held for 2.5 min and subsequently cooled to 50 °C at the same rate. Seven pasting properties were chara terized, including peak viscosity (PV), trough viscosity (TV), final viscosity (FV), break down viscosity (BV), setback viscosity (SV), peak time (PT), and pasting temperatur (PTP). Each sample was analyzed in triplicate.

Pasting Properties Analysis
The pasting properties of lotus root flours and isolated starches were investigated with a Rapid Visco-Analyzer (RVA) (Perten, RVA4500, Macquarie Park, Australia). Samples (1.5 g) were weighed accurately, and 25 mL deionized water was added to prepare 6% (w/w) suspensions. The evaluation program was as follows: the suspensions were first held at 50 • C for 2 min, heated to 95 • C at an increase rate of 12 • C/min, held for 2.5 min, and subsequently cooled to 50 • C at the same rate. Seven pasting properties were characterized, including peak viscosity (PV), trough viscosity (TV), final viscosity (FV), breakdown viscosity (BV), setback viscosity (SV), peak time (PT), and pasting temperature (PTP). Each sample was analyzed in triplicate.

Texture Characteristics Analysis
The clean and fresh lotus roots were cut into 1.0 ± 0.1 cm thick slices, and then the texture characteristics of slices of raw and cooked (steamed for 1 h) lotus roots were analyzed. The texture properties, including hardness, toughness, stickiness, and chewiness, of the tested samples were compared by using a Tensipresser My Boy II texture analyzer (Taketomo Electric, Tokyo, Japan) equipped with Texture Expert software (version 2.26 E). A 1 mm puncture probe at a 15 mm distance at a pretest was used for texture detection. The pretest, test, and posttest speeds were 2.0 mm/s, 1.0 mm/s, and 5 mm/s, respectively. In two cycles, the compression deformation was 35%. Experiments on each sample were repeated five times.
Moreover, the lotus root flour and starch gels obtained from RVA experiments were stored at 4 • C for 24 h and then incubated for 2 h at room temperature to test the gel texture properties by the same texture analyzer. A two-cycle compression test was implemented with a 2.5 cm diameter flat-ended cylinder probe for a 30 mm initial height at a 2.0 mm/s pretest speed and 35% compression deformation. Experiments on each sample were tested in triplicate.

Metabolomics Analysis by UPLC-Q-TOF-MS
Freeze-dried lotus root flours of 40 mg were weighed and dissolved in 1.8 mL precooled 50% methanol in water. The mixtures were vortexed for 1 min and oscillated for 10 min. The suspensions were centrifuged at 15,000 rpm for 10 min at 4 • C. The supernatant was collected and mixed with 0.9 mL precooled 50% methanol in water to repeat the extraction. Then, the combined supernatants were filtered for subsequent analysis. Each sample was conducted with seven replications. Quality control (QC) samples were prepared by combining equal aliquots from all extracts of the lotus freeze-dried flour and were injected into every six specimens during the whole analysis.
Chromatographic separation was achieved on a SCIEX ExionLC AD system equipped with an ACQUITY UPLC HSS T3 column (2.1 × 100 mm, 1.8 µm) and ACQUITY UPLC BEH Amide column (2.1 × 100 mm, 1.7 µm) separately maintained at 40°C. The injection volume was 5 µL, and the sample plate was maintained at 4°C. When analyzed with the HSS T3 column, a gradient elution program consisting of mobile phases A (0.05% formic acid in water) and B (acetonitrile) was used at a flow rate of 0.4 mL/min, and the gradient program was as follows The mass spectrometric data of all samples were acquired with an informationdependent acquisition (IDA) method using the ZenoTOF 7600 mass spectrometer (AB SCIEX, Framingham, MA, USA) in electrospray ionization (ESI) positive and negative ion modes. The parameters were set as follows: the ion spray voltage floating (ISVF) was set at 5500 V for positive mode and −4500 V for negative mode, the nebulizing gas (GS1) and drying gas (GS2) were both nitrogen set at 50 psi; turbo V source temperature (TEM) was set at 500 • C; collision gas (CAD) was nitrogen set at 7 psi; curtain gas (CUR) was nitrogen set at 35 psi; mass range from 60 to 1200 Da. Mass spectral data were acquired in TOFMS-IDA-TOFMSMS mode and analyzed using SCIEX OS software (version 2.2, AB SCIEX, USA). Compound identification was performed by using empirical formula finding, natural product secondary standard spectrum library, and online database searching. In addition, a Zeno trap was employed to increase the intensity of secondary iron mass spectrometry; IDA combined with dynamic background subtraction (DBS) was used to trigger acquisition of the MS/MS information of low-level constituents.

Data Processing and Statistical Analysis
The results from pasting and texture properties were expressed as the mean ± standard deviation (SD). Statistically significant differences were analyzed by one-way analysis of variance (ANOVA), and the differences between the means of samples were carried out by Duncan's test using SPSS 16.0. The MS raw data (wiff.scan files) were processed with SCIEX OS software (Version 2.1). Metabolite identification was performed based on the precise mass of the molecules (<5 ppm) and the MS/MS spectra with an in-house database. All data were normalized, centered on the mean, and divided by the square root of the standard deviation of each variable (Pareto scaling). After normalization, the data were subjected to principal component analysis (PCA) and orthogonal partial least-squares discriminant analysis (OPLS-DA) by SIMCA 13.0 (Umetrics, Umea, Sweden). The measure of fit of the Model (R 2 ) and the measure of predictive ability of the Model (Q 2 ) were used to evaluate the models. The variable importance in the projection (VIP) value (VIP > 1) from the OPLS-DA model contributed to the classification of metabolites. Multivariate analyses including fold change (FC, Elian No. 5/Elian No. 6) and p-value calculation were performed by MetaboAnalyst version 5.0 (https://www.metaboanalyst.ca).

Pasting Behavior of Lotus Root Flours and Starches
The pasting properties of flour and starch samples from Elian No. 5 and Elian No. 6 are presented in Table 2. Generally, the pasting viscosities, including PV, TV, BV, FV, and SV, of starch samples were 1.3~8.0 times those of the whole flours, which was consistent with several previous studies [27,28]. The differences between flours and starches were related to the presence of lipids, proteins, fibers, and lower starch in flours. Most pasting viscosities of Elian No. 5 flours were significantly higher (p < 0.05) than those of Elian No. 6 flours, and the coefficients of variation were all over 30%. Notably, the BVs of Elian No. 5 flours and starches were both higher than those of Elian No. 6 flours and starches, indicating that Elian No. 5 possessed a lower ability to withstand shear stress and heating during cooking [29,30]. The SVs of the flours and starches of Elian No. 6 were both lower than those of Elian No. 5; thus, Elian No. 5 might have a remarkable tendency to retrograde compared to Elian No. 6 during the process. These results further confirmed the difference between the Elian No. 5 and Elian No. 6 cultivars. The pasting property trends of Elian No. 5 from the three planting areas were similar to those of Elian No. 6.

Texture Properties of Lotus Root Slices and Gels
The results of the texture properties of raw and cooked lotus root slices, as well as the flour and starch gels, are presented in Figure 1. It was evident that the hardness of the raw slices and flour gel from Elian No. 5 were higher than that of Elian No. 6, but the hardness of cooked lotus root slices and starch gels were significantly lower (p < 0.05, Figure 1a). A similar trend was observed in the chewiness value, a parameter that was related to hardness (Figure 1d). These results further proved that the texture of Elian No. 5 became farinose, while Elian No. 6 remained crisp during cooking, since the hardness of Elian No. 5 dropped more. It is well known that cooking heat treatment results in the decomposition of intercellular mucus and softening of the texture [31]. Liu et al. [32] indicated that cell wall polyacrylamide, especially the ratio of chelate-soluble fraction to alcohol-insoluble residue, was the major factor affecting the texture of lotus root after cooking. Beyond that, our results showed that the retrogradation behavior of starch also affected hardness, while the hardness of the flour gel of Elian No. 5 was still slightly higher than that of Elian No. 6, and the additional analysis of composition differences between flours and isolated starches may provide a more detailed explanation [33,34]. The springiness of raw and cooked slices from Elian No. 5 was higher than that of Elian No. 6, but the springiness of flour and starch samples had no obvious regularity (Figure 1b). The cohesiveness (Figure 1c) of raw and cooked slices was quite the opposite between Elian No. 5 and Elian No. 6, while it was disorderly in flour and starch gels. The texture properties strongly suggested the opposite texture of Elian No. 5 and Elian No. 6, but this feature was consistent in the three planting areas. raw slices and flour gel from Elian No. 5 were higher than that of Elian No. 6, but the hardness of cooked lotus root slices and starch gels were significantly lower (p < 0.05, Figure 1a). A similar trend was observed in the chewiness value, a parameter that was related to hardness (Figure 1d). These results further proved that the texture of Elian No. 5 became farinose, while Elian No. 6 remained crisp during cooking, since the hardness of Elian No. 5 dropped more. It is well known that cooking heat treatment results in the decomposition of intercellular mucus and softening of the texture [31]. Liu et al. [32] indicated that cell wall polyacrylamide, especially the ratio of chelate-soluble fraction to alcohol-insoluble residue, was the major factor affecting the texture of lotus root after cooking. Beyond that, our results showed that the retrogradation behavior of starch also affected hardness, while the hardness of the flour gel of Elian No. 5 was still slightly higher than that of Elian No. 6, and the additional analysis of composition differences between flours and isolated starches may provide a more detailed explanation [33,34]. The springiness of raw and cooked slices from Elian No. 5 was higher than that of Elian No. 6, but the springiness of flour and starch samples had no obvious regularity (Figure 1b). The cohesiveness (Figure 1c) of raw and cooked slices was quite the opposite between Elian No. 5 and Elian No. 6, while it was disorderly in flour and starch gels. The texture properties strongly suggested the opposite texture of Elian No. 5 and Elian No. 6, but this feature was consistent in the three planting areas.

Interpretation of Zeno-TOF-MS Mass Spectra and Metabolite Profiles of Lotus Root Samples
Accurate identification of metabolites is particularly important in metabolomics studies, which are dependent on high-quality secondary mass spectrometry. In this study, a Zeno trap performed on a ZenoTOF™ 7600 system [24] was first implemented for the systematic comparison of the metabolites of lotus root cultivars with distinct textures. The potential compounds were identified using an in-house MS/MS high-resolution library

Interpretation of Zeno-TOF-MS Mass Spectra and Metabolite Profiles of Lotus Root Samples
Accurate identification of metabolites is particularly important in metabolomics studies, which are dependent on high-quality secondary mass spectrometry. In this study, a Zeno trap performed on a ZenoTOF™ 7600 system [24] was first implemented for the systematic comparison of the metabolites of lotus root cultivars with distinct textures. The potential compounds were identified using an in-house MS/MS high-resolution library containing over 700 common metabolites. The metabolites with mass error < 5 ppm, isotope difference < 10%, and library hit score > 80 were classified as identified. Compared with the Zeno trap-off model, the intensity of the spectrum from the Zeno trap-on model increased 5 to 20 times ( Figure S1a); meanwhile, a higher matching score of the secondary mass spectrometry library was obtained with a low injection concentration ( Figure S1b). Thus, the Zeno trap contributed to the accurate identification of samples with low-abundance metabolites. Based on this method, the metabolite profile of freeze-dried lotus root samples was represented, and a total of 124 and 64 kinds of compounds were identified by positive and negative ion modes, respectively (Table S1). Consistent with previous findings, lotus roots were rich in bioactive substances [8,35]. In this study, all identified compounds mainly included 3 kinds of alkaloids, 36 kinds of amino acids and their derivatives, 4 kinds of coumarins, 3 kinds of flavone glycosides, 26 kinds of flavonoids, 2 kinds of isoquinolines, 1 kind of lignan, 23 kinds of lipids, 13 kinds of nucleotides, 10 kinds of organic acids, 9 kinds of polyphenols, 24 kinds of saccharides and their derivatives, 7 kinds of vitamins, 3 kinds of terpenoids, and 24 other compounds.

Principal Component Analysis of Metabolites in Lotus Roots
Unsupervised PCA models were constructed to illustrate the differences in the metabolites of different lotus root species from three geographical origins, Hanchuan, Jiangxia, and Honghu cities. The PCA score plot for all lotus root samples and QC samples ( Figure S2) could be clearly distinguished with a total variance of 60.9%, in which PC1 and PC2 explained 43.9% and 17% of the total variance, respectively. It was obvious that the metabolites of the same lotus root cultivar from the three producing areas were significantly different and could be effectively distinguished. As shown in the PCA score plot (Figure 2a) obtained from the metabolites of Elian No. 5, the principal components explained 58.5% and 17.2% of the total variation for three geographical origins. Then, a cluster analysis was performed to investigate the relative abundances of identified metabolites, as visualized using a heatmap (Figure 2b). The metabolites identified from different origins were clearly divided into 3 groups. There were 61 differential metabolites identified with VIP > 1, including isoscopoletin, naringin, vitamin C, and coumarin. In addition, the relative abundance of differential metabolites in lotus samples from Honghu was higher than those from Hanchuan and Jiangxia. Similarly, the PCA score plot obtained from the metabolites of Elian No. 6 is presented in Figure 2c, and the principal components explained 47.9% and 18.9% of the total variation for the three geographical origins. The heatmap (Figure 2d) showed only 28 differential metabolites among the three groups. Interestingly, gallocatechin, esculin hydrate, and salidroside were richer in Elian No. 6 from Hanchuan and Jiangxia, while some other functional compounds, such as argininosuccinic acid, levodopa, isoferulic acid, and pinocembrin, possessed significantly higher abundance in the Honghu group. Recently, non-target approaches for food authenticity have been widely used [36], LC-MS-based metabolomics was effective for the authentication of narrow-geographic samples [37], while the geographical origin analysis of lotus root was limited [38]. This study further indicated that the metabolism of lotus roots was significantly affected by growing regions [39], but there is very little known about the application of most of these metabolites. Future studies focused on these functional metabolites could help the discrimination of lotus roots, as well as boost the consumption of lotus root products with regional characteristics.

Orthogonal Projections to Latent Structures-Discriminant Analysis of the Metabolites in Lotus Roots
Orthogonal projections to latent structures-discriminant analysis (OPLS-DA) was performed to distinguish Elian No. 5 and Elian No. 6 cultivars (Figure 3) because of its powerful ability to classify samples that contain only two groups [40]. In this study, the OPLS-DA models after the pretreatment method (Par-scaling) revealed the best classification result [41]. Generally, three OPLS-DA score plots of the tested samples from three regions were sufficiently fit and predictive, since both of them had satisfied the conditions that the R 2 Y(cum) and Q 2 (cum) values were greater than 0.5 [42]. Specifically  that the y-intercepts of the three models were all less than 0.05 for the Q 2 intercept [43]. As shown in Figure 3, the Q2 values were −0.138, −0.214, and −0.195 for the above models, further demonstrating the reliability of the three models without any overlap.

Orthogonal Projections to Latent Structures-Discriminant Analysis of the Metabolites in Lotus Roots
Orthogonal projections to latent structures-discriminant analysis (OPLS-DA) was performed to distinguish Elian No. 5 and Elian No. 6 cultivars (Figure 3) because of its powerful ability to classify samples that contain only two groups [40]. In this study, the OPLS-DA models after the pretreatment method (Par-scaling) revealed the best classification result [41]. Generally, three OPLS-DA score plots of the tested samples from three regions were sufficiently fit and predictive, since both of them had satisfied the conditions that the R 2 Y(cum) and Q 2 (cum) values were greater than 0.5 [42].  To further compare the differences in metabolites between the two types of lotus root cultivars, the variables with VIP > 1, p value < 0.05, FC value >2 or <0.5 were used to identify potential markers in each region [41]. First, 22 significantly differential metabolites were identified between the Elian No. 5 and Elian No. 6 samples from Hanchuan (Table 3a). Obviously, the relative amounts of 6-methoxyquinoline and serotonin in Elian No. 5 were higher than those in Elian No. 6, but the remaining 22 metabolites, including pipecolic acid, coumarin, scopoletin, and paprazine, were lower. Second, 12 significantly differential metabolites were identified between Elian No. 5 and Elian No. 6 samples from Jiangxia (Table 3b). The relative amounts of naringin, pinocembrin, α-phenylglycine, 6-methoxyquinoline, and levodopa in Elian No. 5 were higher, while those of isoscopoletin, paprazine, scopoletin, 5 -s-methyl-5 -thioadenosine, 5-methylthioadenosine, isoferulic acid, and γ-aminobutyric acid in Elian No. 5 were lower. Third, 15 significantly differential metabolites were identified between the Elian No. 5 and Elian No. 6 samples from Honghu (Table 3c). Unlike in the Hanchuan and Jiangxia groups, the levels of 10 differential metabolites, including hederagenin, corosolic acid, and esculin hydratein, in Elian No. 5 were higher than those in Elian No. 6, and only four compounds (isoscopoletin, paprazine, scopoletin, and tyramine) were lower. Notably, the levels of isoscopoletin, scopoletin, and paprazine in all tested Elian No. 5 samples were significantly lower than those in all tested Elian No. 6 samples, indicating their potential for discriminating Elian No. 5 and Elian No. 6 cultivars, as well as farinose and crisp lotus roots. Many studies have comparatively analyzed the metabolites of different lotus root cultivars or different parts of the lotus [21,35,44], while the application for the classification of different lotus genotypes is immature, which need to be investigated in future research. between the two lotus root cultivars. Moreover, 200 permutation test results showed that the y-intercepts of the three models were all less than 0.05 for the Q 2 intercept [43]. As shown in Figure 3, the Q2 values were −0.138, −0.214, and −0.195 for the above models, further demonstrating the reliability of the three models without any overlap. To further compare the differences in metabolites between the two types of lotus root cultivars, the variables with VIP > 1, p value < 0.05, FC value >2 or <0.5 were used to identify potential markers in each region [41]. First, 22 significantly differential metabolites were identified between the Elian No. 5 and Elian No. 6 samples from Hanchuan (Table  3a). Obviously, the relative amounts of 6-methoxyquinoline and serotonin in Elian No. 5 were higher than those in Elian No. 6, but the remaining 22 metabolites, including pipecolic acid, coumarin, scopoletin, and paprazine, were lower. Second, 12 significantly differential metabolites were identified between Elian No. 5 and Elian No. 6 samples from Jiangxia (Table 3b). The relative amounts of naringin, pinocembrin, α-phenylglycine, 6methoxyquinoline, and levodopa in Elian No. 5 were higher, while those of isoscopoletin,

Conclusions
Pasting and texture properties were analyzed between two farinose and crisp lotus root cultivars, Elian No. 5 and Elian No. 6, and the results strongly proved the opposite textures of Elian No. 5 and Elian No. 6, but this feature was consistent in different planting areas. For the first time, untargeted metabolite profiling of Elian No. 5 and Elian No. 6 from three geographical regions was performed by UPLC-Q-TOF-MS with a ZenoTOF™ 7600 system equipped with a Zeno trap approach. In total, 188 metabolites were identified from the chemical database, and these metabolites varied significantly among lotus roots from different cultivars and origins according to multivariate analysis. Isoscopoletin, scopoletin, and paprazine showed potential for discriminating between the Elian No. 5 and Elian No. 6 cultivars. In conclusion, the UPLC-Q-TOF-MS method described in this study can be useful for lotus root classification. Further research based on a large sample set including more cultivars and crop seasons must be conducted to broaden the knowledge and support further utilization of lotus roots in food industries.

Supplementary Materials:
The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/foods12132493/s1, Figure S1: the intensity (a) and matching score (b) of spectrum from Zeno trap-on and Zeno trap-off model; Figure S2: the PCA score plot for all lotus root samples and QC samples; Table S1: the identified metabolites in lotus root.

Data Availability Statement:
The data used to support the findings of this study can be made available by the corresponding author upon request.