Anatomical and Chemical Characterization of Ulmus Species from South Korea

Ulmus species (Ulmaceae) are large deciduous trees distributed throughout Korea. Although their root and stem bark have been used to treat gastrointestinal diseases and wounds in folk medicine, commercial products are consumed without any standardization. Therefore, we examined anatomical and chemical differences among five Ulmus species in South Korea. Transverse sections of leaf, stem, and root barks were examined under a microscope to elucidate anatomical differences. Stem and root bark exhibited characteristic medullary ray and secretary canal size. Leaf surface, petiole, and midrib exhibited characteristic inner morphologies including stomatal size, parenchyma, and epidermal cell diameter, as well as ratio of vascular bundle thickness to diameter among the samples. Orthogonal projections to latent structures discriminant analysis of anatomical data efficiently differentiated the five species. To evaluate chemical differences among the five species, we quantified (-)-catechin, (-)-catechin-7-O-β-D-apiofuranoside, (-)-catechin-7-O-α-L-rhamnopyranoside, (-)-catechin-7-O-β-D-xylopyranoside, (-)-catechin-7-O-β-D-glucopyranoside, and (-)-catechin-5-O-β-D-apiofuranoside using high-performance liquid chromatography with a diode-array detector. (-)-Catechin-7-O-β-D-apiofuranoside content was the highest among all compounds in all species, and (-)-catechin-7-O-α-L-rhamnopyranoside content was characteristically the highest in Ulmus parvifolia among the five species. Overall, the Ulmus species tested was able to be clearly distinguished on the basis of anatomy and chemical composition, which may be used as scientific criteria for appropriate identification and standard establishment for commercialization of these species

While U. parvifolia belongs to Section Microptelea, the other four species belong to Section Ulmus. It is known that Ulmus species produce mucilage, and solitary pores or pore clusters with multiseriate medullary rays exist in the wood specimens of this genus [21,22]. The wood of Ulmus species has diffuse or semi-ring or ring porosity, one to eight seriate and homocellular rays, and prismatic crystals in axial parenchyma or enlarged cells [21]. The wood of the five Ulmus species in South Korea has ring porosity. U. parvifolia has 2-3 deep pores in earlywood zone and thick-walled fibers in latewood zone. U. macrocarpa has one deep pore in earywood zone. While more than three rows of earlywood vessels are found in U. parvifolia and U. pumila, one to three rows in the other three species. In an anatomical study on the bark of Ulmus species, it was reported that orderly arranged phloem fibers and parenchyma strands were found in the bark of U. americana seedlings, but significant intracellular space was not observed [23].
While previous anatomical studies such as comparative anatomy of Ulmaceae, wood anatomy of extant Ulmaceae, and that of 12 species and two varieties from Ulmus from China have been reported, along with their phytochemicals and biological activities, anatomical and chemical differences among the root and stem barks of the five Ulmus species in South Korea remain unexplored [21][22][23][24]. Therefore, the present study aimed to establish a standardization method to identify the botanical origin of biomedical materials on the basis of differences in the inner morphological and phytochemical characteristics of the five Ulmus species in Korea.

Anatomical Characteristics of the Leaf Surface
The adaxial and abaxial leaf surfaces of the five Ulmus species were subjected to scanning electron microscopy (SEM) (Figure 1a,b). Hairs were observed on both the leaf surfaces in all species, except on the abaxial leaf surface in U. parvifolia. They were unicellular pointed trichomes. Short uniseriate stalked trichomes with a multicellular head were also found in all species, except in U. parvifolia. This result is consistent with the previous report that multicellular trichomes were not founded in U. americana, U. crassiflola, U. glabra, U. minor, U. laevis, U. parvifolia, and U. thomasii [22]. There is no report on U. davidiana var. japonica, and U. laciniata showed longer hair and higher hair density on the adaxial leaf surface compared to the others. U. parvifolia showed short hair and the lowest hair density among the five species. Anomocytic stomata were observed only on the abaxial leaf surface of all species (Figure 1c). The number of stomata, but not the stomatal index, is significantly affected by factors such as growth, environment, maturity, veins, and hair [25,26]. The longest stomatal apparatus was found in U. pumila (30.0 ± 1.1 µm) and the shortest in U. macrocarpa (18.8 ± 1.4 µm). Likewise, the widest stomatal width was found in U. pumila (25.7 ± 1.2 µm), while the narrowest in U. macrocarpa (14.4 ± 1.5 µm). In this study, the stomatal index of U. pumila and U. macrocarpa was the highest (19.2 ± 0.2 and 18.8 ± 0.1, respectively) and the lowest was of U. davidiana var. japonica (11.4 ± 0.2)

Anatomical Characteristics of the Leaf Midrib
The vascular bundle of the midrib was U-shaped in all Ulmus species (Figure 2). The vascular bundle width was similar to vascular bundle height ( Table 2). The widest epidermal cells in the adaxial part were found in U. parvifolia (20.1 ± 0.8 μm) and the narrowest were in U. davidiana var. japonica (11.5 ± 0.7 μm). The longest epidermal cells were observed in U. parvifolia (15.3 ± 1.5 μm) and the shortest in UPU and U. davidiana var. japonica (10.5 ± 1.4 and 9.9 ± 0.9 μm, respectively). The widest epidermal cells in the abaxial part were observed in U. laciniata (18.9 ± 1.4 μm) and the narrowest in U. davidiana var. japonica and U. parvifolia (14.0 ± 3.3 and 13.8 ± 1.3 μm, respectively). The longest epidermal cells

Anatomical Characteristics of the Leaf Midrib
The vascular bundle of the midrib was U-shaped in all Ulmus species (Figure 2). The vascular bundle width was similar to vascular bundle height ( Table 2). The widest epidermal cells in the adaxial part were found in U. parvifolia (20.1 ± 0.8 µm) and the narrowest were in U. davidiana var. japonica (11.5 ± 0.7 µm). The longest epidermal cells were observed in U. parvifolia (15.3 ± 1.5 µm) and the shortest in UPU and U. davidiana var. japonica (10.5 ± 1.4 and 9.9 ± 0.9 µm, respectively). The widest epidermal cells in the abaxial part were observed in U. laciniata (18.9 ± 1.4 µm) and the narrowest in U. davidiana var. japonica and U. parvifolia (14.0 ± 3.3 and 13.8 ± 1.3 µm, respectively). The longest epidermal cells were observed in U. laciniata and U. macrocarpa (14.9 ± 0.6 and 14.2 ± 3.5 µm, respectively). The longest diameter of collenchyma cells was observed in U. laciniata (38.6 ± 4.0 µm), and the shortest in U. davidiana var. japonica (15.0 ± 1.6 µm). Collenchyma cells support the stem near the epidermis and the parenchyma cells, helping the plant to grow straight [27]. The diameter of parenchyma cells in the cortex and pith was the longest in U. laciniata Plants 2021, 10, 2617 4 of 15 (29.8 ± 2.8 and 25.6 ± 2.2 µm, respectively). Higher ratio of vascular bundle thickness to midrib diameter was observed in U. parvifolia and U. pumila (0.17 ± 0.01 and 0.16 ± 0.01, respectively) than the others (Table 2). No significant differences in the ratio of vascular bundle width to vascular bundle height was found among the samples with the value of about one. Calcium oxalate or calcium carbonate crystals were usually present in plant tissues. Prismatic calcium oxalate crystals in the midrib of the Ulmus species exhibited a U-shaped distribution pattern along the collenchyma cells, and these crystals were the most abundant in U. pumila and the least abundant in U. laciniata. In U. laciniata, both prismatic crystals and druses were detected in parenchyma cells ( Figure 2).
were observed in U. laciniata and U. macrocarpa (14.9 ± 0.6 and 14.2 ± 3.5 μm, respectively). The longest diameter of collenchyma cells was observed in U. laciniata (38.6 ± 4.0 μm), and the shortest in U. davidiana var. japonica (15.0 ± 1.6 μm). Collenchyma cells support the stem near the epidermis and the parenchyma cells, helping the plant to grow straight [27]. The diameter of parenchyma cells in the cortex and pith was the longest in U. laciniata (29.8 ± 2.8 and 25.6 ± 2.2 μm, respectively). Higher ratio of vascular bundle thickness to midrib diameter was observed in U. parvifolia and U. pumila (0.17 ± 0.01 and 0.16 ± 0.01, respectively) than the others (Table 2). No significant differences in the ratio of vascular bundle width to vascular bundle height was found among the samples with the value of about one. Calcium oxalate or calcium carbonate crystals were usually present in plant tissues. Prismatic calcium oxalate crystals in the midrib of the Ulmus species exhibited a U-shaped distribution pattern along the collenchyma cells, and these crystals were the most abundant in U. pumila and the least abundant in U. laciniata. In U. laciniata, both prismatic crystals and druses were detected in parenchyma cells (Figure 2). .  Data are expressed as mean ± SD (n > 3) of five independent experiments. Different upper letters in the same line indicate a significant difference (p < 0.05) among samples. * Ratio of vascular bundle thickness to midrib diameter, and ratio of vascular bundle width to vascular bundle height were calculated from a/b and c/d, respectively, in Figure 2.

Anatomical Characteristics of the Petiole
The petiole was round in all species. The widest epidermal cells in the adaxial part were observed in U. macrocarpa and U. laciniata (18.8 ± 3.2 and 18.3 ± 5.0 µm, respectively) and the narrowest in U. pumila (13.5 ± 1.0 µm). The longest epidermal cells in the adaxial part were observed in U. macrocarpa and U. laciniata (16.6 ± 3.5 and 13.9 ± 1.3 µm, respectively). The widest epidermal cells in the abaxial part were observed in U. macrocarpa and U. laciniata (24.5 ± 3.5 and 21.5 ± 3.2 µm, respectively), and the longest epidermal cells in this part were observed in U. laciniata and the shortest in U. pumila. Collenchyma cells were detected in all samples. A significant difference among the species was found in the diameter of collenchyma cells. While the values of U. macrocarpa and U. laciniata were more than 30 µm, they were less than 20 µm in U. parvifolia and U. pumila. The diameter of parenchyma cells in the cortex was the longest in U. parvifolia (23.5 ± 1.6 µm), and the diameter of parenchyma cells in the pith was the longest in U. davidiana var. japonica and U. laciniata (22.3 ± 5.5 and 20.3 ± 1.5 µm, respectively). The vascular bundle in the petiole was U-shaped in all species. The ratio of vascular bundle thickness to petiole diameter was the highest in U. parvifolia and U. pumila (0.10 ± 0.01 and 0.11 ± 0.02, respectively). Unlike leaf midrib, petiole showed significant difference in the ratio of vascular bundle width to vascular bundle height among the samples. While U. parvifolia showed square type of vascular bundle, oblong type of vascular bundle was observed in the others. U. davidiana var. japonica and U. laciniata showed the lowest ratio. Calcium oxalate crystals were detected, exhibiting a U-shape distribution pattern along collenchyma cells, in all species, except in U. pumila, in which prismatic crystals exhibited an O-shape distribution pattern. Interestingly, calcium oxalate crystals were mainly concentrated in the abaxial part of U. parvifolia but distributed throughout the parenchyma cells in U. macrocarpa. The abundance of druses was the highest in U. pumila and the lowest in U. laciniata ( Figure 3 and Table 3).
Ratio of vascular bundle width to vascular bundle height * 1.02 ± 0.18 a 1.05 ± 0.02 a 1.12 ± 0.27 a 1.14 ± 0.18 a 0.94 ± 0.07 a Data are expressed as mean ± SD (n > 3) of five independent experiments. Different upper letters in the same line indicate a significant difference (p < 0.05) among samples. * Ratio of vascular bundle thickness to midrib diameter, and ratio of vascular bundle width to vascular bundle height were calculated from a/b and c/d, respectively, in Figure 2.

Anatomical Characteristics of the Petiole
The petiole was round in all species. The widest epidermal cells in the adaxial part were observed in U. macrocarpa and U. laciniata (18.8 ± 3.2 and 18.3 ± 5.0 μm, respectively) and the narrowest in U. pumila (13.5 ± 1.0 μm). The longest epidermal cells in the adaxial part were observed in U. macrocarpa and U. laciniata (16.6 ± 3.5 and 13.9 ± 1.3 μm, respectively). The widest epidermal cells in the abaxial part were observed in U. macrocarpa and U. laciniata (24.5 ± 3.5 and 21.5 ± 3.2 μm, respectively), and the longest epidermal cells in this part were observed in U. laciniata and the shortest in U. pumila. Collenchyma cells were detected in all samples. A significant difference among the species was found in the diameter of collenchyma cells. While the values of U. macrocarpa and U. laciniata were more than 30 μm, they were less than 20 μm in U. parvifolia and U. pumila. The diameter of parenchyma cells in the cortex was the longest in U. parvifolia (23.5 ± 1.6 μm), and the diameter of parenchyma cells in the pith was the longest in U. davidiana var. japonica and U. laciniata (22.3 ± 5.5 and 20.3 ± 1.5 μm, respectively). The vascular bundle in the petiole was U-shaped in all species. The ratio of vascular bundle thickness to petiole diameter was the highest in U. parvifolia and U. pumila (0.10 ± 0.01 and 0.11 ± 0.02, respectively). Unlike leaf midrib, petiole showed significant difference in the ratio of vascular bundle width to vascular bundle height among the samples. While U. parvifolia showed square type of vascular bundle, oblong type of vascular bundle was observed in the others. U. davidiana var. japonica and U. laciniata showed the lowest ratio. Calcium oxalate crystals were detected, exhibiting a U-shape distribution pattern along collenchyma cells, in all species, except in U. pumila, in which prismatic crystals exhibited an O-shape distribution pattern. Interestingly, calcium oxalate crystals were mainly concentrated in the abaxial part of U. parvifolia but distributed throughout the parenchyma cells in U. macrocarpa. The abundance of druses was the highest in U. pumila and the lowest in U. laciniata ( Figure 3 and Table 3).

Anatomical Characteristics of the Stem Bark
The medullary rays of the stem bark were straight in U. davidiana var. japonica and U. parvifolia, and were curved and closer to the epidermis in U. macrocarpa, U. laciniata, and U. pumila. The medullary ray was 2-4 cells thick in all species, with the highest frequency in U. davidiana var. japonica (4.3 ± 0.7 per 1 mm 2 ). The length of the medullary ray cells was the longest in U. laciniata and U. parvifolia (51.4 ± 6.6 and 47.0 ± 7.2 µm, respectively) and the shortest in U. macrocarpa (20.0 ± 7.0 µm). The width of the medullary ray cells was the widest in U. macrocarpa (17.0 ± 1.8 µm), followed by U. laciniata (15.5 ± 0.5 µm). Secretory canals, which synthesize and store chemicals for defense against herbivores and pathogens [28], were uniformly distributed throughout the stem bark in all species, except in U. macrocarpa. It is known that Ulmus species produce mucilage, and solitary pores or pore clusters with multiseriate medullary rays exist in the wood specimens of this genus [21,22]. The frequency of secretary canals was the highest in U. parvifolia (23.4 ± 4.0%) and the lowest in U. macrocarpa (2.5 ± 4.0% in 1 mm 2 ). The number of secretary canals was the highest in U. parvifolia and U. davidiana var. japonica (20.5 ± 2.7 and 18.4 ± 2.5, respectively). The widest the secretary canals were observed in U. parvifolia (167.9 ± 11.2 µm), while the longest secretary canals were observed in U. parvifolia and U. pumila (237.4 ± 28.8 and 224.5 ± 27.3 µm, respectively) ( Figure 4 and Table 4).

Anatomical Characteristics of the Root Bark
The medullary rays of the root bark, unlike those of the stem bark, were curved in all species, except in U. laciniata, in which they were disconnected ( Figure 5). The medullary ray was 2-4 cells thick in all species, with the highest frequency in U. laciniata (3.6 ± 0.5 per 1 mm 2 ). The longest medullary ray cells were observed in U. pumila (54.5 ± 5.7 µm), while the widest medullary ray cells were observed in U. parvifolia (19.0 ± 0.6 µm) and the narrowest in U. laciniata. Secretary canals were unevenly distributed and overlapped in the root bark, unlike those in the stem bark. The frequency of secretary canals was the highest in U. pumila (30.8 ± 0.2% per 1 mm 2 ) and the lowest in U. laciniata and U. davidiana var. japonica (16.2 ± 1.4 and 16.8 ± 0.4% per 1 mm 2 , respectively). The number of secretary canals per square millimeter was the highest in U. macrocarpa (36.6 ± 8.1) and the lowest in U. davidiana var. japonica (13.9 ± 0.9). The longest secretary canals were observed in U. pumila (171.5 ± 1.7 µm) and the shortest in U. davidiana var. japonica. The widest secretary canals were observed in U. pumila (248.4 ± 15.5 µm) and the narrowest in U. macrocarpa and U. laciniata (Table 5).

Multivariate Statistical Analysis of Anatomical Data
Orthogonal projections to latent structures-discriminant analysis (OPLS-DA) was performed to classify the five species on the basis of their anatomical data. Regarding leaf, the midrib and petiole presented characteristic sizes of the epidermal, parenchyma, and collenchyma cells on the abaxial and adaxial parts. Among the five Ulmus species, U. davidiana var. japonica, U. macrocarpa, and U. laciniata could be successfully distinguished, but U. parvifolia and U. pumila could not be distinguished and appeared to overlap with each other (Figure 6a). The leaves of U. parvifolia and U. pumila could not be distinguished from each other due to the similar ratio of vascular bundle thickness to diameter in the midrib and petiole. Regarding the stem and root bark, the five species presented similar values of frequency and size of the medullary ray cells but different values of the ratio, frequency, and size of the secretary canals. Therefore, Ulmus species could be distinguished using OPLS-DA of stem and root bark anatomical data (Figure 6b-d). Overall, the five Ulmus species could be clearly distinguished and classified on the basis of the anatomical features of the leaf, stem bark, and root bark (Figure 6e).
In the root bark, the content of compound 1 was the highest in U. davidiana var. japonica and U. laciniata (2.96 and 3.21 mg g −1 •DW −1 , respectively), followed by U. parvifolia and U. pumila (1.15 and 1.26 mg g −1 •DW −1 , respectively). The content of compound 2 was the highest among the six compounds in the root bark of all species, as observed in the Considerable differences in the content of compounds 1-6 were observed among the five species. In the stem bark, the content of compound 1 was the highest in U. pumila (9.72 mg g −1 ·dry weight (DW) −1 ), followed by U. parvifolia (2.78 mg g −1 ·DW −1 ). The content of compound 2 was the highest among the six compounds in all species (23.62 to 21.21 mg g −1 ·DW −1 ) in U. pumila, U. davidiana var. japonica, and U. macrocarpa; 15.95 mg g −1 ·DW −1 in U. laciniata; and 14.08 mg g −1 ·DW −1 in U. parvifolia. The content of compound 3 was the highest in U. parvifolia (6.81 mg g −1 ·DW −1 ). The content of compound 4 was the highest in U. pumila (2.92 ± 0.34 mg g −1 ·DW −1 ) and the lowest in U. parvifolia (0.49 ± 0.04 mg g −1 ·DW −1 ). The content of compound 6 was the lowest among the six compounds in all species, and the content was the highest in U. pumila among the five species.

Plant Materials and Reagents
Five  (Table 7).
Glycerin (Junsei Chemicals Co., Ltd., Tokyo, Japan) was used to prepare specimens for anatomical examination. Ethanol (Daejung Chemicals and Metals Co., Ltd., Shiheung, Korea) was used for sample extraction. HPLC was performed using water and MeOH In the root bark, the content of compound 1 was the highest in U. davidiana var. japonica and U. laciniata (2.96 and 3.21 mg g −1 ·DW −1 , respectively), followed by U. parvifolia and U. pumila (1.15 and 1.26 mg g −1 ·DW −1 , respectively). The content of compound 2 was the highest among the six compounds in the root bark of all species, as observed in the stem bark. The content of compound 2 in the root bark was the highest in U. macrocarpa and U. laciniata (26.61 and 24.89 mg g −1 ·DW −1 , respectively). While the content of compound 2 was the lowest in U. parvifolia, the content of compound 3 was the highest in U. parvifolia (1.63 mg g −1 ·DW −1 ), as observed in the stem bark. The content of compound 3 was similar among U. davidiana var. japonica, U. laciniata, and U. macrocarpa (0.51-0.57 mg g −1 ·DW −1 ). The content of compound 4 was the highest in U. laciniata, followed by U. macrocarpa, and the content of compound 5 was the highest in U. macrocarpa (1.83 mg g −1 ·DW −1 ), followed by U. laciniata (1.50 mg g −1 ·DW −1 ). The content of compound 6 was the lowest in the root bark among the six compounds in all species, as observed in the stem bark. The content of compound 6 was the highest in U. davidiana var. japonica (0.46 mg g −1 ·DW −1 ). In summary, the content of compound 2 was the highest among the six compounds in the stem and root bark of all species. Meanwhile, the content of compound 3 was the highest in the stem and root bark of U. parvifolia among the five species (Table 6). Although chemical taxonomy with popular flavonoids such myricetin and quercetin by classical paper chromatography has been accomplished on Ulmus species without U. laciniata and U. davidiana var. japonica [39], this is the first report on chemical differences by HPLC on the major secondary metabolites of five Ulmus species in South Korea.    (Table 7).

Plant Materials and Reagents
Glycerin (Junsei Chemicals Co., Ltd., Tokyo, Japan) was used to prepare specimens for anatomical examination. Ethanol (Daejung Chemicals and Metals Co., Ltd., Shiheung, Korea) was used for sample extraction. HPLC was performed using water and MeOH (Thermo Fisher Scientific Korea Ltd., Seoul, Korea). NMR solvents were purchased from Cambridge Isotope Laboratories, Inc. (Andover, MA, USA). Other reagents used were of high analytical grade.

Anatomical Study
Healthy and well-acclimatized samples (stem bark, root bark, midrib, and petiole of leaves) were obtained from the five Ulmus species and were preserved in 50% ethanol solution. Free hand or hand microtome sections with 20 to 40 µm-thickness were prepared using razor blades or a hand-held microtome (Euromex MT.5500, Arnhem, the Netherlands). Four to five sections were obtained from the middle part of the collected stem and root bark, lower part of midrib, and central part of petiole. The adaxial and abaxial leaf surfaces were analyzed using SEM (JSM-6380LV, Jeol, Tokyo, Japan) [38]. Eau de Javelle solution (Sigma, Minneapolis, MN, USA) was used to bleach the samples. Then, the samples were mounted in 100% glycerin or 50% glycerinated water on glass slides. All samples were observed under a light microscope (BX53F, Olympus, Tokyo, Japan), and photomicrographs were obtained using image processing software (IMT i-Solution Inc., Vancouver, BC, Canada) coupled to a video camera (PixeLINK, Ottawa, ON, Canada). Over five specimens of each species were analyzed to obtain representative characteristics, and five regions were measured on each photomicrograph.
Transverse sections of the stem and root bark were used to count the frequency of medullary rays and secretary canals in an area of 1 mm 2 . A range of 200 × 200 µm was selected to count the frequency of stomata and stomatal index on the abaxial leaf surface.  Figure S1, The EI-MS spectrum of compound 1. Figure S2, The 1 H-NMR spectrum of compound 1. Figure S3, The 13 C-NMR spectrum of compound 1. Figure S4, The FAB-MS spectrum of compound 2. Figure S5, The 1 H-NMR spectrum of compound 2. Figure S6, The 13 C-NMR spectrum of compound 2. Figure S7, The FAB-MS spectrum of compound 3. Figure S8, The 1 H-NMR spectrum of compound 3. Figure S9, The 13 C-NMR spectrum of compound 3. Figure S10, The ESI-MS spectrum of compound 4. Figure S11, The 1 H-NMR spectrum of compound 4. Figure S12, The 13 C-NMR spectrum of compound 4. Figure S13, The ESI-MS spectrum of compound 5. Figure S14, The 1 H-NMR spectrum of compound 5. Figure S15, The 13 C-NMR spectrum of compound 5. Figure  S16, The ESI-MS spectrum of compound 6. Figure S17, The 1 H-NMR spectrum of compound 6. Figure  S18, The 13 C-NMR spectrum of compound 6. 1 H and 13 C NMR assign data of isolated compounds.