Hydrolysable Tannins and Biological Activities of Meriania hernandoi and Meriania nobilis (Melastomataceae)

A bio-guided study of leaf extracts allowed the isolation of two new macrobicyclic hydrolysable tannins, namely merianin A (1) and merianin B (2), and oct-1-en-3-yl β-xylopyranosyl-(1”-6’)-β-glucopyranoside (3) from Meriania hernandoi, in addition to 11 known compounds reported for the first time in the Meriania genus. The structures were elucidated by spectroscopic analyses including one- and two-dimensional NMR techniques and mass spectrometry. The bioactivities of the compounds were determined by measuring the DPPH radical scavenging activity and by carrying out antioxidant power assays (FRAP), etiolated wheat coleoptile assays and phytotoxicity assays on the standard target species Lycopersicum esculentum W. (tomato). Compounds 1 and 2 exhibited the best free radical scavenging activities, with FRS50 values of 2.0 and 1.9 µM, respectively.


Introduction
Melastomataceae is a woody dicotyledonous family with approximately 4200 to 4500 species and 166 genera distributed mainly in the South American neotropics [1,2]. Meriania Swartz (Melastomataceae) is a neotropical genus of shrubs and trees with around 93 species distributed through southern Mexico, Central America, the Greater Antilles, Andean South America, the Guaiana Highlands, and south to southeastern Brazil [3]. In Colombia, 37 species of this genus have been found and these include Meriania nobilis Triana and Meriania hernandoi L. Uribe. M. nobilis is also called wax flower because its flowers are melliferous. M. nobilis is an endemic tree species in the Colombian Andes, which is its native habitat. The tree is mainly used for ornamental purposes due to its striking intense violet flowers. Similarly, M. hernandoi is mainly ornamental due to the intense orange color of its flowers, which is an unusual characteristic amongst the Melastomataceae family, for which violet, fuchsia and white tones prevail in flowers. These plants grow in the humid and foggy forests of southern Colombia and northern Ecuador. These two species are members of the Melastomataceae family, which is rich in phenolic constituents such as flavonoids, terpenoids, quinones, lignans and tannins [4]. These constituents are responsible for analgesic, anti-inflammatory [5], antioxidant [6], The initial extracts were tested in a general activity bioassay, namely the etiolated wheat coleoptile assay. This is an initial screening method to evaluate activity and it is highly sensitive to pharmacological activities [10] and also shows a good correlation with phytotoxic activity. The results obtained were similar to those for antioxidant activity. The extracts Mh 2.1 and Mh 2.   9 19 ± 1 252 ± 2 11 19 ± 1 263 ± 1 12 48 ± 1 166 ± 3 13 19 ± 1 283 ± 3 14 473 ± 3 44 ± 1 Data are expressed as the mean ± SD. a Concentration of the sample required to scavenge 50% of the DPPH free radicals. b Total phenolic content measured using the Folin-Ciocalteu method. c Not tested.
The initial extracts were tested in a general activity bioassay, namely the etiolated wheat coleoptile assay. This is an initial screening method to evaluate activity and it is highly sensitive to pharmacological activities [10] and also shows a good correlation with phytotoxic activity. The results obtained were similar to those for antioxidant activity. The extracts Mh 2.1 and Mh 2.2 exhibited a high inhibitory effect on coleoptiles, with 58% and 57% inhibition, respectively, at the highest concentrations ( Figure 2a). The extracts that showed inhibitory activity on coleoptiles were evaluated for phytotoxic activity. The assay was performed using L. sativa L. (lettuce), Lycopersicum esculentum Will. (tomato), Lepidium sativum L. (cress), and Allium cepa L. (onion) as standard target species (STS) [11]. Moreover, Lolium perenne and Lolium rigidum were used as weeds and significant effects were only observed on the root growth of STS in tomato (Figure 2b). The inhibition values for extracts Mh 2.1 and Mh 2.2 on the root growth of tomato at 800 ppm were 37% and 33%, respectively. These values are lower than those obtained with Logran®(82%), a commercial herbicide used as a positive control. According to the biological activity results, the extracts Mh 2.1, Mh 2.2 and Mn 2.2 were selected for bio-guided fractionation and isolation of the major components.

Characterization of the Compounds
Compound 1 was obtained as a brown amorphous solid with a molecular weight of 1872.1738 and molecular formula C82H56O52, as established by the EI-MS spectrum with an ion peak at m/z 1871.1674 [M − H] -and doubly charged ion at m/z 935.0801 [M − 2H] 2-in negative mode. The 1 H-NMR spectrum revealed the presence of three galloyl groups by three signals that integrate to two protons each at δ 7.23, 7.09 and 6.96, which were assigned as shown in Table 2 according to long range correlations. A valoneoyl group and two hexahydroxydiphenyl (HHDP) groups were evidenced by seven 1H-singlets at δ 7.12, 6.62, 6.54, 6.48, 6.41 6.39 and 5.91, which were also assigned from the HMBC spectrum. The (S) configuration of both the HHDP and valoneoyl groups was confirmed by the circular dichroism (CD) spectrum, which exhibited a negative Cotton effect at around 253 nm and a positive effect near to 276 nm, i.e., similar to that of the lignan gomisin D, whose configuration is (S) [20], as depicted in supplementary material S-13. The dimeric nature of 1 was demonstrated by the presence of two anomeric signals at δH 6.10 [d, J = 8.5 Hz, glucose (Glc) I, H-1] and 6.24 [d, J = 8.4 Hz, Glc II, H-1]. The other glucose proton signals were all assigned based on 1 H-1 H shift correlation spectroscopy (COSY) and J-resolved spectra, which indicated that the conformation adopted by the two glucoses is 4 C1 and that they are fully acylated. A large difference between the chemical shifts of the H-6 proton signals on Glc II (∆H6 = 1.45) suggested the presence of a biphenyl moiety with  The 1 H-NMR spectrum revealed the presence of three galloyl groups by three signals that integrate to two protons each at δ 7.23, 7.09 and 6.96, which were assigned as shown in Table 2 according to long range correlations. A valoneoyl group and two hexahydroxydiphenyl (HHDP) groups were evidenced by seven 1H-singlets at δ 7.12, 6.62, 6.54, 6.48, 6.41 6.39 and 5.91, which were also assigned from the HMBC spectrum. The (S) configuration of both the HHDP and valoneoyl groups was confirmed by the circular dichroism (CD) spectrum, which exhibited a negative Cotton effect at around 253 nm and a positive effect near to 276 nm, i.e., similar to that of the lignan gomisin D, whose configuration is (S) [20], as depicted in Supplementary Materials Figure S13. The dimeric nature of 1 was demonstrated by the presence of two anomeric signals at δ H 6.10 [d, J = 8.5 Hz, glucose (Glc) I, H-1] and 6.24 [d, J = 8.4 Hz, Glc II, H-1]. The other glucose proton signals were all assigned based on 1 H-1 H shift correlation spectroscopy (COSY) and J-resolved spectra, which indicated that the conformation adopted by the two glucoses is 4 C 1 and that they are fully acylated. A large difference between the chemical shifts of the H-6 proton signals on Glc II (∆H6 = 1.45) suggested the presence of a biphenyl moiety with bridged ester linkages between O-4/O-6 [21] as in telimagrandin II [22]. The location of this unit on the glucose core and its exact attachment mode were determined based on the HMBC three-bond correlations from the corresponding protons of the glucose units and from the aromatic protons to the carbonyl carbons, as summarized in Table 3 and Figure 3. The signal at δ 7.12 (valoneoyl E-6) showed connectivity to the glucose I H-4 (5.92) at three bonds with an ester carbonyl at 165.7 ppm. Similarly, the signal at δ 5.91 (valoneoyl C'-3) showed coupling to the glucose II H-2 (5.09) and to an ester carbonyl at 169.5 ppm and the signal at δ 6.48 (valoneoyl D-3) correlated to the glucose I H-3 (5.54) and with an ester carbonyl at δ 170.5. These spectroscopic data for 1 were different to those of the isomer nobotanin B [23], especially for the valoneoyl signal at 5.91 ppm in 1, which is shifted downfield with respect to the signal at 6.12 ppm in nobotanin B, and also the sugar signals H-3, H-4 and H-6. The evidence outlined above is consistent with compound 1 being a dimer of pentagalloyl glucose [24] and tellimagrandin II; which, in a biogenetic approach, is initially linked by oxidative C-C coupling between the galloyl units on C-2 of the pentagalloyl glucose and that on C-3 of the tellimagandin II to give intermediate I ( Figure 3). In the next step, the macrocycle is generated by another oxidative coupling between the galloyl units from C-2 of telimagrandin II to C-3 of pentagalloyl glucose to generate intermediate II. An oxidative C-O-C coupling between rings D and E of positions 3 and 4 of pentagalloyl glucose leads to the formation of merianin A (1); while the oxidative C-O-C coupling between the 1 and 2 of telimagrandin II produces the merianin B isomer (2). The letters of the aromatic rings were assigned according to the sequence of galloylation in the biosynthesis of the pentagalloyl glucose. This biogenetic proposal is reinforced by the downfield shift of carbons C-2 and C-3 in both glucose units, with respect to pentagalloyl glucose and tellimagrandin II, as shown in Table 3. Acid hydrolysis followed by methylation and silylation of 1 confirmed the identity of the sugars in the molecule as D-glucose. The structure of compound 1 was therefore established as the macrobicyclic hydrolysable tannin merianin A (1), as shown in Figure 4. the carbonyl carbons, as summarized in Table 3 and Figure 3. The signal at δ 7.12 (valoneoyl E-6) showed connectivity to the glucose I H-4 (5.92) at three bonds with an ester carbonyl at 165.7 ppm. Similarly, the signal at δ 5.91 (valoneoyl C'-3) showed coupling to the glucose II H-2 (5.09) and to an ester carbonyl at 169.5 ppm and the signal at δ 6.48 (valoneoyl D-3) correlated to the glucose I H-3 (5.54) and with an ester carbonyl at δ 170.5. These spectroscopic data for 1 were different to those of the isomer nobotanin B [23], especially for the valoneoyl signal at 5.91 ppm in 1, which is shifted downfield with respect to the signal at 6.12 ppm in nobotanin B, and also the sugar signals H-3, H-4 and H-6. The evidence outlined above is consistent with compound 1 being a dimer of pentagalloyl glucose [24] and tellimagrandin II; which, in a biogenetic approach, is initially linked by oxidative C-C coupling between the galloyl units on C-2 of the pentagalloyl glucose and that on C-3 of the tellimagandin II to give intermediate I ( Figure 3). In the next step, the macrocycle is generated by another oxidative coupling between the galloyl units from C-2 of telimagrandin II to C-3 of pentagalloyl glucose to generate intermediate II. An oxidative C-O-C coupling between rings D and E of positions 3 and 4 of pentagalloyl glucose leads to the formation of merianin A (1); while the oxidative C-O-C coupling between the 1 and 2 of telimagrandin II produces the merianin B isomer (2). The letters of the aromatic rings were assigned according to the sequence of galloylation in the biosynthesis of the pentagalloyl glucose. This biogenetic proposal is reinforced by the downfield shift of carbons C-2 and C-3 in both glucose units, with respect to pentagalloyl glucose and tellimagrandin II, as shown in Table 3. Acid hydrolysis followed by methylation and silylation of 1 confirmed the identity of the sugars in the molecule as D-glucose. The structure of compound 1 was therefore established as the macrobicyclic hydrolysable tannin merianin A (1), as shown in Figure 4.     Figure S14). The 1 H-NMR and 13 C-NMR spectra were virtually identical to those obtained for 1. A subtle difference between 2 and 1 was evidenced by the HMBC correlation. The three-bond correlation from proton signals at δ 7.12 (valoneoyl A'-6) and glucose II H-1 (6.24) to the ester carbonyl at δ 165.7 provided evidence for the C-O-C oxidative coupling between the galloyl moieties on C-1 and C-2 of telimagrandin II monomeric unit (Figures 3 and 4). Furthermore, the downfield shifts of C-2 and C-3 with respect to the corresponding signals of telimagrandin II, as shown in Table 3, support the presence of the large macrocycle. Therefore, the signals at δ 6.96 (galloyl E) and the glucose I H-4 (5.92) exhibited three-bond coupling to the ester carbonyl at δ 165.7 (Figure 3). Acid hydrolysis followed by methylation and silylation of 2 confirmed the identity of the sugars as D-glucose. Consequently 2 is an isomer of 1 and it was named merianin B. reflect different aspects of entity, and our experimental results show that the entity linking system could obtain better performance with an entity embedding framework which can capture a larger range of different entity information aspects. Some earlier methods [2,6] only take the entity description into consideration, with heuristic methods like BOW or TF-IDF. However, Sun [1] argues that these methods are insufficient to disentangle the underlying explanatory factors of the data and proposes a method which employs a convolutional neural network (CNN) to encode the entity description. Some other methods [3,[7][8][9] try to encode the entity, based on the idea of word embedding, which only takes entity context into consideration. However, all the methods above fail to capture different information aspects of entity, which could result in a loss of information. Inspired by Nitish Gupta's work [4], we design an entity embedding framework which can capture different information aspects of entity. Other entity information aspects may be lacking, like entity type, while the entity description and entity context are common for most knowledge bases. So, in our work, we mainly investigate how to effectively encode and combine entity description and entity context, to generate dense unified entity embeddings. It is also worth noting that our entity embedding framework could be easily extended to the case in which there are more entity aspects. We use long short term memory (LSTM) [10] and CNN [11] to encode the entity context and entity description, respectively, and afterwards we design a function to encourage the entity embedding's similarity to all of the encoded representations (e.g., representation of entity context and representation of entity description). Compared to previous methods, our approach can capture different aspects of entity information, and our experimental results show that our global model could obtain a better performance, with entity embeddings which can capture richer entity information aspects.  H-NMR spectrum revealed two anomeric protons at δ 4.35 (d, J = 7.9) and 4.34 (d, J = 7.8). The other sugar proton signals were all assigned based on 1 H-1 H shift correlation spectroscopy (COSY), total correlated spectroscopy (TOCSY) and J-resolved spectroscopy (J-res), which indicated a sequential trans diaxial relationship between H-1' (4.35) and H-5' (J = 7-9 Hz) corresponding to β-glucopyranose and a sequential trans diaxial relationship between H-1" (4.34) and H-4" (J = 7-9 Hz) indicated a β-xylopyranose ( Table 4). The spectra contained signals that evidence a linear structure of an octenyl moiety, with three signals at 5.05 (dd, J = 3.7, 10.5), 5.18 (dd, J = 3.7, 17.3) and 5.82 (ddd, J = 7.1, 10.5, 17.3) (Table 4) indicating the presence a double bond in the molecule. The linkage between the aglycon and sugar moiety was determined based on the HMBC correlations, which showed coupling of glucose H-1' and C-3 to three bonds, with evidence of coupling for H-6' with C-1"to three bonds indicating interglycosidic linkage. Silylation of 3 confirmed the identity of the sugars in the molecule. The spectroscopic evidence is consistent with this compound being oct-1-en-3-yl β-xylopyranosyl-(1"-6')-β-glucopyranoside (3) (Figure 4). A similar compound has been isolated from the Melastomataceae family and this was designated as a matsutake alcohol derivative [25].

Bioactivity of Compounds
The antioxidant activities of the isolated compounds were determined using the DPPH assay and values are expressed as the concentration of the sample required to scavenge 50% of the DPPH free radicals (FRS 50 ) ( Table 1). The results show that all of the compounds isolated from M. hernandoi exhibited low FRS 50 values and these are comparable to the value obtained for quercetin. These compounds possess a high scavenging capacity for free radicals. The hydrolysable tannins showed the highest scavenging capacity for free radicals and this is due to the fact that a hydrolysable tannin can donate a hydrogen atom and form a stable quinone [26,27]. In addition, compounds 1 and 2 exhibited the best free radical scavenging activity, with FRS 50 values of 2.0 and 1.9 µM, respectively, since these compounds have a dimeric structure with more catechol-type hydroxyl groups.
The results obtained for the antioxidant power (FRAP) show that all of the compounds isolated from M. hernandoi have high values. This finding correlates with the FRS 50 results. The hydrolysable tannins tested had a high antioxidant power-especially the dimeric tannins. In the FRAP assay, the activity values for phenolic compounds seem to depend on the degree of hydroxylation and the extent of conjugation of the phenolic compounds [28]. For example, compounds 1 and 2, which are macrocyclic, and 7, which has a dimeric structure, presented the highest FRAP values (Table 1).
Compounds 1, 2, 4, 7, 10 and 11 were available in larger quantities and these were tested in the etiolated wheat coleoptile assay. The values obtained were high for the macrocyclic tannins and the dimeric hydrolysable tannin, with values of 77% for merianin A (1), 70% for merianin B (2) and 70% for nobotanin F (7) at a concentration of 1000 µM. Casuarinin (4) is a monomeric tannin and this showed only moderate activity, with a value of 43% at 1000 µM. In contrast, the compounds quercitrin (10) and quercetin (11) showed low activity, with values of 24% and 18% (Figure 6), respectively. Similar results were obtained in the phytotoxicity assay; the strongest inhibition was observed for merianin A (1), merianin B (2) and nobotanin F (7), with values in the range 40-50% at a concentration of 1000 µM. The monomeric tannin casuarinin (4) showed low phytotoxicity, with a value of only 27% at 1000 µM (Figure 4b). Significant effects were only observed on the root growth in Lycopersicum esculentum Will (tomato). The results for the inhibition of coleoptile growth and phytotoxicity show that activity increases with the molecular size of the tannin, as the compounds with a dimeric structure or macrocycles showed higher activity than the monomer casuarinin. Likewise, the activity seems to be related to the number of galloyl groups and HHDP groups in the molecule. This trend has also been reported for other types of activity, such as cytotoxic activity against human tumor cell lines [29] and neuraminidase inhibitory activity in anti-influence therapy [30], where the activity is enhanced by increasing the number of free galloyl groups or HHDP. A total of 14 compounds have been isolated from M. hernandoi and M. nobilis and three of these are described for the first time. Among the major compounds, the new hydrolysable tannins 1 and 2, along with compound 7, show higher activities in the etiolated wheat coleoptile bioassay and against the STS tomato seeds. Furthermore, together with the antioxidant activities, compound 4 showed different biological activities such as anti-herpes [31], neuroprotective effects by suppressing glutamate-mediated apoptosis in human cells [32] and therapeutic properties against inflammatory skin diseases [33]. Compound 8 also proved to be an α-glucosidase inhibitor [34]. The results obtained show that the species M. hernandoi is a good source of promising bioactive metabolites that include pharmacological activities.

General Experimental Procedures
Optical rotations were measured with a JASCO (Tokyo, Japan) model P-2000 series digital polarimeter. UV-visible spectra were recorded using a JASCO (Tokyo, Japan) UV-730 spectrophotometer. 1 H and 13 C-NMR spectra were recorded in methanol-d4 or acetone-d6 + D2O with a Bruker (Bruker-Biospin GmbH, Rheinstetten, Germany) Advance III 400 spectrometer at 400 MHz and 100 MHz and Agilent Technologies (Santa Clara, CA, USA) spectrometers at 500 MHz and 125 MHz or at 600 MHz and 200 MHz. A combination of 1D spectra such as 1 H-NMR, 13 C-NMR, DEPT135, and 2D spectra such as 1 H-1 H-COSY, 1 H-1 H-TOCSY, ROESY, J-resolved, HSQC and HMBC were used to determine the chemical structures. A selection of these spectra is provided in the supplementary file.

Extraction and Isolation
Dried and powdered leaves (500 g) of each species were successively extracted three times with n-hexane followed by acetone/water 70% in a Branson Scientific ultrasonic bath at room temperature for 15 minutes. The solvent was evaporated under reduced pressure. The extract was liquid-liquid partitioned in ethyl acetate to obtain extracts Mn 2.1 (5.14 g, 1.1%) and Mh 2.1 (8.1 g, 1.6%) and in water-saturated n-BuOH to obtain extracts Mn 2.2 (2.01 g, 0.4%) and Mh 2.2 (28.3 g, 5.7%) and aqueous fractions were denoted as Mn 2.3 (13.6 g, 2.7%) and Mh 2.3 (15.1 g, 3.0%). The extracts Mn 2.1, Mh 2.1 and Mh 2.2 were fractionated using a water/MeOH stepwise gradient (10-100% MeOH, increment of 10% in each step) on a DIAION HP-20 (30 cm × 5 cm I.D) to obtain eleven fractions. The fractions that exhibited activity were chromatographed on an MCI-Gel CHP20P (460 mm length × 26 mm I.D) with a water/MeOH gradient (10-100% MeOH, increments of 5% in each step) to obtain eleven pure compounds from M. hernandoi and one pure compound from M. nobilis. The compounds were characterized as follows: where A C is the absorbance of the DPPH radicals without the sample or positive control and as is the absorbance of DPPH radicals with sample or positive control. The efficient concentrations of the samples and positive controls that inhibit FRS 50 were calculated and are expressed as mg/L.

Measurement of Ferric Reducing Antioxidant Power (FRAP)
The reducing power from Fe 3+ to Fe 2+ by an antioxidant was determined using the method of Benzie and Strain [36] adjusting for use in a 96-well microplate. The extracts and the standard were prepared at 1024 mg/L, the pure compounds and standard were prepared at 250 µM in methanol or water. The FRAP reagent was prepared by mixing acetate buffer (300 mM, pH 3.6), 10 mM TPTZ (2,4,6-tripyridyl-s-triazine) in 40 mM acidified methanol, and 20 mM FeCl 3. 6H 2 O at a 10:1:1 ratio (v/v/v) in this order. The mixture was then heated for 1 h at 37 • C and allowed to stand at room temperature. Immediately, two-fold serial dilutions (1-512 mg/L to extracts and 15.6-250 µM to pure compounds) of the sample (30 mL) were mixed with 140 mL of FRAP reagent and 30 mL of water on the microplate. Finally, the absorbance was measured at 600 nm in a microplate reader (Metertech, AccuReader M 965+) after 1 h of reaction. Quercetin was used as the positive control and was examined in parallel experiments. All samples were measured in triplicate. A standard calibration curve for iron(II) sulfate heptahydrate (FeSO 4 .7H 2 O) was plotted. All results are expressed as mg FeSO 4 .7H 2 O (100 g) −1 dry extract and as µM FeSO 4 .7H 2 O

Total Phenolic Content Measured Using the Folin-Ciocalteu Method (TPC)
Total phenolic content was measured using the Folin-Ciocalteu method described by Sdiri and co-workers [35]. This method was applied in the 96-well microplate format. The samples (extract, fraction or standard) were prepared at 1024 mg/L in 2-propanol or methanol. The sample (100 mL) at two-fold serial dilutions (512-15.6 mg/L) was mixed with 50 mL of 20% v/v Folin-Ciocalteu reagent and 50 mL of sodium carbonate solution (1.6% w/v) on the microplate. The mixture was heated for 1 h at 60 • C and then allowed to cool down to room temperature. The absorbance was measured at 650 nm in a microplate reader (Metertech, AccuReader M 965+). The samples were measured in quadruplicate. Gallic acid (GA) (1-512 mg/L) was used as the standard in the calibration curve, with the optimal range of 32 to 1.0 mg/L. Average results are expressed as mg GA (100 g) −1 dry extract.

Etiolated Wheat Coleoptile Assays
All extracts were tested in this general activity bioassay and this initial screening was employed to evaluate activity. Wheat seeds (Triticum aestivum L.) were used in this bioassay according to the methodology previously described in the literature [35]. The extracts were prepared at 800, 400 and 200 mg L −1 and the pure compounds at 1000, 333, 100, 33 and 10 µM. A buffered nutritive aqueous solution with DMSO (0.5% v/v) without any tested extract was used as negative control. The commercial herbicide Logran®was used as positive control. Coleoptile elongation was measured using the Photomed© system. The results are presented in bar charts and are shown as percentage differences from the control. Thus, zero represents the control, negative values represent inhibition and positive values denote stimulation of the evaluated parameter.

Standard Target Species Bioassay (STS) and Weeds
The extracts that exhibited activity in the general bioassay were tested in this bioassay. L. sativa L. (lettuce), Lycopersicum esculentum Will. (tomato), Lepidium sativum L. (cress), and Allium cepa L. (onion) were used as standard target species (STS). The weeds Lolium perenne and Lolium rigidum were also used and this bioassay was carried out according to the methodology previously described in the literature [37]. The samples were prepared with aqueous buffer solution at 10 −2 M of 2-[N-morpholino]ethanesulfonic acid (MES) and 1 M NaOH and were dissolved in DMSO to obtain working concentrations of 800, 400 and 200 mg L −1 for extracts and 1000, 333, 100, 33 and 10 µM for pure compounds. Each sample contained a constant quantity of 0.5% DMSO. The negative controls were the buffer containing DMSO without the tested compounds and the positive control was Logran®. The measurements (shoot and root length, and germination) were made using a Fitomed© system, which allowed automatic data acquisition and statistical analysis with the associated software. Results are presented as percentage differences from the control. Zero represents control, positive values represent stimulation, and negative values represent inhibition.