Multivariate Analysis of Amino Acids and Health Beneficial Properties of Cantaloupe Varieties Grown in Six Locations in the United States

Cantaloupe is a good dietary source of amino acids, including γ-aminobutyric acid (GABA), glutamine, and citrulline. However, the levels of these amino acids vary among different cantaloupe varieties grown in different locations. Understanding the variation in amino acid contents provides fundamentally important information for quality control and improving melon varieties. To examine this variation, we measured the amino acid contents in cantaloupes grown in six locations in the United States (Texas, Georgia, North Carolina, California, Indiana, and Arizona). Principal component analyses were applied to analyze the effect of growing location on the amino acid profiles in different varieties. The GABA content ranged from 1006.14 ± 64.77 to 3187.12 ± 64.96 µg/g and citrulline ranged from 92.65 ± 9.52 to 464.75 ± 34.97 µg/g depending on the variety and location. Total phenolic contents, α-amylase inhibition, and antioxidant activities were also measured. Tuscan type Da Vinci had significantly higher phenolic contents in Arizona (381.99 ± 16.21 µg/g) but had the lowest level when grown in California (224.56 ± 14.62 µg/g). Our analyses showed significant differences in amino acid levels, phenolics contents, and antioxidant activity in the cantaloupe varieties based on the growing location. These findings underline the importance of considering growing location in the selection and improvement of cantaloupe varieties.


Introduction
Melons (Cucumis melo L.) such as cantaloupe are important commercial crops and are consumed all over the world. Cantaloupe is favored by consumers due to its pleasant aroma, refreshing flavor qualities, nutrition, and phytochemical composition [1,2]. Melon fruits have high levels of the GABA [3] and other amino acids, particularly valine and alanine [4]. Amino acid identification and quantification of melon can be performed by reverse phase liquid chromatography, gas chromatography, and nuclear magnetic resonance [5][6][7]. Liquid chromatography is a commonly used technique for analysis of amino acids from plants. Analysis is mainly performed by liquid chromatography coupled with ultraviolet, diode-array, fluorescent, or mass detectors after derivatization. Commonly used derivatization reagents are dansyl-chloride, O-phthalaldehyde, 9-fluorenylmethyloxycarbonyl, and 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate [5,[8][9][10]. The combined free amino acid contents of cantaloupe varieties grown in different locations are presented in Figure 2. Combined free amino acids content in melon F-39 ranged from 5035.73 to 6318.7195 µg/g with the highest in IN and lowest in NC. In HT-IG, the amino acid content ranged from 5420.01 to 7682.76 µg/g and was higher in CA as compared with other locations. Similarly, in TT-DV melons, combined free amino acids ranged from 5011.22 to 8421.68 µg/g depending on the growing location. Among the commercial local varieties, CAR from CA had the highest (7086.10 µg/g) combined free amino acids followed by CRU (6907.68 µg/g) from IN but ATH fruits (4741.53 µg/g) from GA had the lowest value. The combined free amino acid contents of cantaloupe varieties grown in different locations are presented in Figure 2. Combined free amino acids content in melon F-39 ranged from 5035.73 to 6318.7195 µg/g with the highest in IN and lowest in NC. In HT-IG, the amino acid content ranged from 5420.01 to 7682.76 µg/g and was higher in CA as compared with other locations. Similarly, in TT-DV melons, combined free amino acids ranged from 5011.22 to 8421.68 µg/g depending on the growing location. Among the commercial local varieties, CAR from CA had the highest (7086.10 µg/g) combined free amino acids followed by CRU (6907.68 µg/g) from IN but ATH fruits (4741.53 µg/g) from GA had the lowest value.
The comparative amino acids chromatograms of cantaloupe variety TT-DV grown in different locations are shown in Figure 3. Variation in the HPLC-FLD peak height of different amino acids in the TT-DV variety provides evidence of differences in amino acids content among the states. For example, the highest peak intensity for GABA (peak 12) was detected in NC and lowest in CA and GA. Figure S1 depicts the HPLC-FLD chromatograms of amino acids from different cantaloupe varieties grown in Texas. The amino acid levels in the varieties and growing locations were compared for significance at α = 0.05 (Table S1). Comparing fruits grown in the different locations, glutamine (Gln) was significantly higher in F-39 fruit harvested from TX (1227.04 ± 93.65 µg/g), followed by AZ (1092 ± 57.02 µg/g) compared to other locations ( Figure 4). Compared also with other locations, TT-DV melons grown in IN had significantly higher levels of Gln (2034.94 ± 158.74 µg/g) followed by TX (1453.68 ± 57.25 µg/g) and NC (1406.29 ± 19.67 µg/g). For the commercial local varieties, Gln was significantly higher in fruit grown in TX (PRI) (1637.61 ± 60.04 µg/g) followed by IN (CRU) (1454.65 ± 58.61 µg/g) and NC (ATH) (1278.52 ± 120.43 µg/g) than in other locations ( Figure 4). Our results are consistent with other studies that reported higher levels of Gln (2060.3 µg/g) in melon fruit compared to other amino acids [4].
Among the commercial local varieties (PRI, ATH, CAR, ALG, and CRU), there was no significant difference in GABA content. Another published study evaluated the GABA content in melon harvested in different seasons-autumn, winter, spring, and summer. They reported the highest content of GABA in melons harvested in summer (2010 µg/g) and the lowest (880 µg/g) in winter season, consistent with a strong effect of environment on the accumulation of GABA in melon fruits [3]. The Asp content was significantly higher in CA and AZ at 2388.45 ± 147.30 and 1500.88 ± 47.58 µg/g, respectively, than in other locations. Asp was higher in fruits grown in CA (1580.12 ± 116.98 µg/g) and AZ (1075.38 ± 107.70 µg/g) compared to other locations. The comparative amino acids chromatograms of cantaloupe variety TT-DV grown in different locations are shown in Figure 3. Variation in the HPLC-FLD peak height of different amino acids in the TT-DV variety provides evidence of differences in amino acids content among the states. For example, the highest peak intensity for GABA (peak 12) was detected in NC and lowest in CA and GA. Figure S1 depicts the HPLC-FLD chromatograms of amino acids from different cantaloupe varieties grown in Texas. The amino acid levels in the varieties and growing locations were compared for significance at α = 0.05 (Table S1). Comparing fruits grown in the different locations, glutamine (Gln) was significantly higher in F-39 fruit harvested from TX (1227.04 ± 93.65 µg/g), followed by AZ (1092 ± 57.02 µg/g) compared to other locations ( Figure 4). Compared also with other locations, TT-DV melons grown in IN had significantly higher levels of Gln (2034.94 ± 158.74 µg/g) followed by TX (1453.68 ± 57.25 µg/g) and NC (1406.29 ± 19.67 µg/g). For the commercial local varieties, Gln was significantly higher in fruit grown in TX (PRI) (1637.61 ± 60.04 µg/g) followed by IN (CRU) (1454.65 ± 58.61 µg/g) and NC (ATH) (1278.52 ± 120.43 µg/g) than in other locations ( Figure 4). Our results are consistent with other studies that reported higher levels of Gln (2060.3 µg/g) in melon fruit compared to other amino acids [4].
In melon fruit, GABA levels ranged from 1006.14 ± 64.77 to 3187.12 ± 64.96 µg/g depending on variety and location. GABA content in F-39 was significantly higher in fruits grown in NC (1728. 35   Levels of the nitric oxide precursors arginine (Arg) and citrulline (Cit) were significantly influenced by the growing conditions ( Figure 5). Arg ranged from 38 Figure 5). Similarly, TT-DV fruit had significantly higher Arg in AZ (123.85 ± 11.65 µg/g), while Cit was highest in IN (297.97 ± 21.33 µg/g) and TX (243.01 ± 10.50 µg/g) compared to other locations. Among the commercial local varieties, Arg content in ATH from NC (50.88 ± 5.99 µg/g) and GA (38.17 ± 2.11 µg/g) was the lowest and varieties CRU (CA) and ALG (AZ) had higher levels, 145.77 ± 12.5 and 114.25 ± 8.29 µg/g, respectively as compared to other varieties. Melon varieties CRU (IN) and ALG (AZ) had significantly higher levels of Cit at 360.52 ± 23.49 and 307.33 ± 58.84 µg/g, respectively, than the other varieties harvested from different locations. Citrulline, a non-essential amino acid and precursor to the arginine, is mainly present in Cucurbitaceae family members such as Plants 2020, 9, 1058 5 of 25 watermelon, squash, and cucumber [5]. A recent study reported the effect of two growing locations in NC on Cit and Arg levels, using different cultivars from those we studied here. They observed significant variation in the Cit and Arg levels within Cucumis melo cultivars. The Cit content ranged from 163 to 863 µg/g fresh weight and Arg ranged from 94 to 180 µg/g fresh weight [47]. These findings showed that amino acids contents are strongly influenced by growing locations and cultivars.    Levels of the nitric oxide precursors arginine (Arg) and citrulline (Cit) were significantly influenced by the growing conditions ( Figure 5). Arg ranged from 38.17 ± 2.11 to 184.51 ± 18.18 µg/g and Cit ranged from 92. 65 Figure 5). Similarly, TT-DV fruit had significantly higher Arg in AZ (123.85 ± 11.65 µg/g), while Cit was highest in IN (297.97 ± 21.33 µg/g) and TX (243.01 ± 10.50 µg/g) compared to other locations. Among the commercial local varieties, Arg content in ATH from NC (50.88 ± 5.99 µg/g) and GA (38.17 ± 2.11 µg/g) was the lowest and varieties CRU (CA) and ALG (AZ) had higher levels, 145.77 ± 12.5 and 114.25 ± 8.29 µg/g, respectively as compared to other varieties. Melon varieties CRU (IN) and ALG (AZ) had significantly higher levels of Cit at 360.52 ± 23.49 and 307.33 ± 58.84 µg/g, respectively, than the other varieties harvested from different locations. Citrulline, a non-essential amino acid and precursor to the arginine, is mainly present in Cucurbitaceae family members such as watermelon, squash, and cucumber [5]. A recent study reported the effect of two growing locations in NC on Cit and Arg levels, using different cultivars from those we studied here. They observed significant variation in the Cit and Arg levels within Cucumis melo cultivars. The Cit content ranged from 163 to 863 µg/g fresh weight and Arg ranged from 94 to 180 µg/g fresh weight [47]. These findings showed that amino acids contents are strongly influenced by growing locations and cultivars.  Melons are also rich in essential amino acids such as Val, Thr, Met, Trp, Phe, Ile, and Leu, which play a major role in fruit aroma, nutritional value, and health-promoting properties. Val content was high in all varieties grown in different locations and ranged from 243.75 ± 18.72 to 671.41 ± 4.50 µg/g ( Table 1). Compared with the other locations, F-39 fruit grown in IN (671.41 ± 23.84 µg/g) had significantly high levels of Val, followed by NC (455.75 ± 3.54 µg/g) and TX (340.15 ± 14.09 µg/g). HT-IG showed no significant variation in Val content among the locations. TT-DV fruit had a significantly higher level of Val in AZ (541.50 ± 41.10 µg/g) and NC (509.08 ± 7.11 µg/g). According to the Food and Agriculture Organization (FAO) and the World Health Organization (WHO, 1985), daily requirements of amino acids are Trp (3.5 µg/g), Val (10 µg/g), Ile (10 µg/g), and Leu (14 µg/g) [48]. These show that melon is an excellent dietary source for essential amino acids.    Among the commercial local varieties, CRU grown in IN had higher Val content (510.73 ± 4.5 µg/g) and varieties harvested from TX (PRI), GA (ATH), CA (CAR) and AZ (ALG) had no significant difference. Thr content was significantly higher in F-39 fruit grown in AZ (132.49 ± 14.01µg/g) and GA (106.76 ± 21.28 µg/g), while in commercial local varieties, higher levels of Thr were measured in fruit grown in AZ (99.01 ± 5.55 µg/g) and CA (90.79 ± 11.25 µg/g). Similarly, HT-IG (100.28 ± 5.35 µg/g) and TT-DV (109.29 ± 11.08 µg/g) fruits showed higher Thr contents in AZ samples compared to other states. Trp is the precursor of serotonins, which regulate human mood, anxiety, appetite, and sleep [49]. Among the analyzed samples, HT-IG and the commercial local variety (CAR) harvested from CA showed significant higher contents of Trp at 71.82 ± 1.73 µg/g and 58.82 ± 2.04 µg/g, respectively, while TT-DV fruits grown in AZ had higher levels of Trp (68.67 ± 10.75 µg/g) compared to other locations and varieties. Melons are also rich in essential amino acids such as Val, Thr, Met, Trp, Phe, Ile, and Leu, which play a major role in fruit aroma, nutritional value, and health-promoting properties. Val content was high in all varieties grown in different locations and ranged from 243.75 ± 18.72 to 671.41 ± 4.50 µg/g ( Table 1). Compared with the other locations, F-39 fruit grown in IN (671.41 ± 23.84 µg/g) had significantly high levels of Val, followed by NC (455.75 ± 3.54 µg/g) and TX (340.15 ± 14.09 µg/g). HT-IG showed no significant variation in Val content among the locations. TT-DV fruit had a significantly higher level of Val in AZ (541.50 ± 41.10 µg/g) and NC (509.08 ± 7.11 µg/g). According to the Food and Agriculture Organization (FAO) and the World Health Organization (WHO, 1985), daily requirements of amino acids are Trp (3.5 µg/g), Val (10 µg/g), Ile (10 µg/g), and Leu (14 µg/g) [48]. These show that melon is an excellent dietary source for essential amino acids.
Among the commercial local varieties, CRU grown in IN had higher Val content (510.73 ± 4.5 µg/g) and varieties harvested from TX (PRI), GA (ATH), CA (CAR) and AZ (ALG) had no significant difference. Thr content was significantly higher in F-39 fruit grown in AZ (132.49 ± 14.01µg/g) and GA (106.76 ± 21.28 µg/g), while in commercial local varieties, higher levels of Thr were measured in fruit grown in AZ (99.01 ± 5.55 µg/g) and CA (90.79 ± 11.25 µg/g). Similarly, HT-IG (100.28 ± 5.35 µg/g) and TT-DV (109.29 ± 11.08 µg/g) fruits showed higher Thr contents in AZ samples compared to other states. Trp is the precursor of serotonins, which regulate human mood, anxiety, appetite, and sleep [49]. Among the analyzed samples, HT-IG and the commercial local variety (CAR) harvested from CA showed significant higher contents of Trp at 71.82 ± 1.73 µg/g and 58.82 ± 2.04 µg/g, respectively, while TT-DV fruits grown in AZ had higher levels of Trp (68.67 ± 10.75 µg/g) compared to other locations and varieties. The non-essential amino acid Ala was also identified in melon fruit, and was significantly higher in F-39 (494.13 ± 53.30 µg/g), HT-IG (477.01 ± 52.75 µg/g), and commercial local varieties (429.43 ± 46.75 µg/g) harvested from CA. Alanine was reported to be a major amino acid in melon and imparts sweetness to melon fruits [4]. Plants are exposed to various environmental stress factors such as drought, extreme temperature, chilling, and salinity. These factors significantly affect the numerous metabolites and metabolic pathways which influence plant growth and development [50,51]. Accumulation of amino acids under stress conditions has been reported in published literature [52][53][54]. Increases in the levels of specific amino acids have shown beneficial effect in a stressful environment. Accumulation of proline and citrulline has been correlated with plant stress tolerance. They act as a reactive oxygen species scavenger and provide protection to the plant under different environmental stresses [13,55]. Amino acids Asp and Glu are both involved in nitrogen remobilization in plants. Increases in the levels of Ser and Gly act as a biomarker for photorespiration [52][53][54][55][56]. Plants also accumulate GABA under various environmental stresses and this plays a vital role in osmotic and pH regulation, plant development, and carbon-nitrogen balance in plants [53][54][55][56][57]. A greenhouse study demonstrated that cold stress or low temperature (2 to 8 • C) exposure of spinach considerably increased the GABA content [58]. Similarly, in another study, cold exposure (−8 • C) induced significant increases in GABA content in wheat and barley [59]. In our study, higher levels of GABA were observed among the melons grown in the NC state ( Figure 4). According to the weather data collected at the NC state, the minimum average temperatures in the months of March and April were −4.4 and 0 • C, respectively ( Figure S2). Similarly, IN also had lower minimum temperatures, −5.5 • C for March and −2.7 • C for April, which might have been responsible for the higher level of GABA in melons harvested from these locations. In contrast, the average temperatures in AZ were 6.6 and 10 • C for the months of March and April, respectively, and low levels of GABA were observed in F-39 and HT-IG varieties. Based on these results, it can be inferred that a low temperature exposure during the early growth phase of melons may be an important contributing factor to enhance the GABA content in melon cultivars. There was also variation in the precipitation among the growing locations, but additional irrigation was applied in all locations, therefore, the amount of precipitation was probably not a contributing factor towards amino acids accumulation in melons.
These data suggest that no one variety-location combination produced the highest levels of amino acids, suggesting that breeding programs could produce cultivars with improved amino acid contents, but they must consider the growing locations and environmental stress factors. As observed in our data, GABA was high in all analyzed varieties grown in NC and IN locations, while F-39 and TT-DV grown in CA had low GABA content. The Gln content in TT-DV, HT-IG, and commercial local varieties was high in the IN location, while F-39 showed low content in the same location. Similarly, Cit was high in the IN location in all varieties except F-39. Moreover, another study reported that the variation in melon metabolites (volatiles, sugars, and amino acids) was affected by various factors such as the variety, growing season, planting dates, locations, and agricultural management practices [60].

PCA of Each Variety Grown in Different Locations
To further explore the interaction of variety and location, we conducted principal component analysis (PCA), which enables visualization of the location-amino acid contents space. Figure 6A-D shows the PCA of variation in amino acid profiles of each cantaloupe variety grown in different locations. The first (F1) and the second (F2) components explained 43.13% and 21.15% of the total variance, respectively. The effect of growing location on cantaloupe F-39 variety was distinguished based on the effect of Arg, Asn, Gln, Cit, Thr, Gly, Ala, GABA, Pro, and Val contents on component F1 and Ser, Asp, and Trp contents on component F2 ( Figure 6A). Generally, the PCA indicated that melons grown in AZ had higher levels of Cit, Arg, Thr, and Asn, whereas melons grown in IN and NC had higher levels of Gly, Val, Pro, and GABA. Some differentiation among the states was also detected on component F2. CA-grown fruit had higher amounts of Asp, Ser, and Trp. Similarly, in variety HT-IG, states were differentiated based on the effect of Arg, Asn, Cit, Ser, Asp, Gly, Ala, Pro, Val, Trp, and Phe Plants 2020, 9, 1058 11 of 25 on component F1, and Hyp, Thr, β-Ala, GABA, Ile, and Leu on component F2 ( Figure 6B). Overall, the PCA showed that HT-IG grown in GA had low levels of Arg, Asn, Cit, Gly, Ala, Pro, Trp, and Phe but HT-IG grown in CA had high levels of these amino acids. Likewise, HT-IG grown in NC had high amounts of Hyp, β-Ala, and GABA but HT-IG grown in TX had less of these amino acids. Overall, the first two principal components explained 71.92% of the total variation. variance, respectively. The effect of growing location on cantaloupe F-39 variety was distinguished based on the effect of Arg, Asn, Gln, Cit, Thr, Gly, Ala, GABA, Pro, and Val contents on component F1 and Ser, Asp, and Trp contents on component F2 ( Figure 6A). Generally, the PCA indicated that melons grown in AZ had higher levels of Cit, Arg, Thr, and Asn, whereas melons grown in IN and NC had higher levels of Gly, Val, Pro, and GABA. Some differentiation among the states was also detected on component F2. CA-grown fruit had higher amounts of Asp, Ser, and Trp. Similarly, in variety HT-IG, states were differentiated based on the effect of Arg, Asn, Cit, Ser, Asp, Gly, Ala, Pro, Val, Trp, and Phe on component F1, and Hyp, Thr, β-Ala, GABA, Ile, and Leu on component F2 ( Figure 6B). Overall, the PCA showed that HT-IG grown in GA had low levels of Arg, Asn, Cit, Gly, Ala, Pro, Trp, and Phe but HT-IG grown in CA had high levels of these amino acids. Likewise, HT-IG grown in NC had high amounts of Hyp, β-Ala, and GABA but HT-IG grown in TX had less of these amino acids. Overall, the first two principal components explained 71.92% of the total variation. The growing locations for TT-DV were differentiated based on Ser, Thr, Gly, GABA, Pro, and Val on component F1 and Arg, Gln, Cit, Asp, Ala, Met, Trp, and Leu on component F2 ( Figure 6C). Overall, examination of the location-amino acid content space indicated that the variety TT-DV grown in AZ and CA had higher levels of Arg and Asp. The first two principal components explained 66.05% of the total variation among the state average scores. Based on component F1, GA had lower levels of Ser, Gly, and Pro, while IN had higher levels of these amino acids. The commercial local varieties grown in different states were distinguished based on the effect of Arg, Asn, Cit, Ser, Thr, Gly, β-Ala, Ala, Pro, Val, Trp, Phe, and Ile on component F1 and Met and Leu on component F2 ( Figure 6D). Examination of the location-amino acid level space indicated that the melons grown in IN had higher levels of Cit, Ser, Hyp, β-Ala, Val, and Ile. Melons grown in NC showed higher levels of Met and melons grown in CA had higher levels of Arg, Asn, Ala, and Trp. The first two principal components explained 64.63% of the total variation among the state average scores. The PCA results were supported by grouping of similar states in hierarchical cluster analysis (HCA) ( Figure S3). States within each HCA cluster were also nearby in PCA, indicating that these states have similar amino acid profiles.

PCA of Different Varieties with a Single Growing Location
PCA also was conducted to visualize the cantaloupe variety-amino acid levels space for each growing location (Figure 7). To support PCA, HCA was performed to cluster cantaloupe varieties in each growing location ( Figure S4).
The growing locations for TT-DV were differentiated based on Ser, Thr, Gly, GABA, Pro, and Val on component F1 and Arg, Gln, Cit, Asp, Ala, Met, Trp, and Leu on component F2 ( Figure 6C). Overall, examination of the location-amino acid content space indicated that the variety TT-DV grown in AZ and CA had higher levels of Arg and Asp. The first two principal components explained 66.05% of the total variation among the state average scores. Based on component F1, GA had lower levels of Ser, Gly, and Pro, while IN had higher levels of these amino acids. The commercial local varieties grown in different states were distinguished based on the effect of Arg, Asn, Cit, Ser, Thr, Gly, β-Ala, Ala, Pro, Val, Trp, Phe, and Ile on component F1 and Met and Leu on component F2 ( Figure 6D). Examination of the location-amino acid level space indicated that the melons grown in IN had higher levels of Cit, Ser, Hyp, β-Ala, Val, and Ile. Melons grown in NC showed higher levels of Met and melons grown in CA had higher levels of Arg, Asn, Ala, and Trp. The first two principal components explained 64.63% of the total variation among the state average scores. The PCA results were supported by grouping of similar states in hierarchical cluster analysis (HCA) ( Figure S3). States within each HCA cluster were also nearby in PCA, indicating that these states have similar amino acid profiles.

PCA of Different Varieties with a Single Growing Location
PCA also was conducted to visualize the cantaloupe variety-amino acid levels space for each growing location (Figure 7). To support PCA, HCA was performed to cluster cantaloupe varieties in each growing location ( Figure S4). The PCA and HCA showed that melon varieties were distinguished differently in each state. For example, melons grown in TX were differentiated based on the effect of Arg, Cit, Hyp, Gly, GABA, Val, Trp, Phe, lle, and Leu on component F1, and Asn, Ser, Asp, Thr, Ala, Pro, and Met on component F2 ( Figure 7A). HT-IG and TT-DV differed from each other by amino acids that affected component F1. HT-IG had lower amounts of GABA, Leu, Val, and Hyp and TT-DV had higher levels of these amino acids. The first and second components jointly explained 79.04% of the total variance. Likewise, in GA-grown melon varieties, F-39 and ATH differed from each other based on the effects of Arg, Asn, Asp, Hyp, Thr, Pro, Met, and Val on component F1 ( Figure 7B). F-39 melons had higher amounts of Arg, Cit, Gln, Ser, Pro, and Hyp and ATH melons had higher amounts of Asp, Gly, β-Ala, and Val. Based on component F2, TT-DV had higher amounts of Trp and GABA than other melon varieties. For the melon varieties grown in NC, the first and second components explained 81.98% of the total variance ( Figure 7C). The melon varieties were differentiated based on the effects of Asn, Ser, Asp, Hyp, Gly, β-Ala, Pro, Val, and Leu on component F1, and Gln, Cit, Ala, Met, and Phe on component F2. The HT-IG variety had more amino acids affecting component F1, except Leu and Hyp, which were higher in variety ATH. F-39 had lower levels of amino acids (except Ala) driving component F2.
Melons grown in AZ were differentiated by PCA based on the effects of amino acids on component F1, except Arg, Thr, Met, lle, and Leu, which affected component F2 ( Figure 7D). The first two components accounted for 88.88% of the total variance. TT-DV melons had higher amino acid contents driving component F1, except Asn, Gln, Cit, and β-Ala. This indicated that TT-DV was distinct from other melon varieties grown in AZ, with higher amounts of 10 amino acids. The first two components accounted for 88.04% of total variance in melons grown in CA ( Figure 7E). Excluding Thr, GABA, Val, and Trp, melon varieties were differentiated based on the effect of the other amino acids on component F1. The first component accounted for 64.69% of the total variance and HT-IG differed from other varieties since it contained higher amounts of amino acids that were driving component F1, except Phe, lle, Leu, Asp, and Hyp, which were higher in TT-DV melons.
Similarly, the first two components accounted for 73.84% of the total variance in melon varieties grown in IN, based on the effects of Asn, Gln, Cit, Ser, Asp, Hyp, Ala, GABA, and Val on component F1, and Arg, Thr, Gly, β-Ala, Pro, lle, and Leu on component F2 ( Figure 7F). TT-DV and F-39 differ from each other based on amino acids that affect component F1: if the amino acids that affect component F1 were high in TT-DV, then, they were low in F-39 and vice versa. Likewise, varieties APH and HT-IG were dissimilar to each other based on the amino acid contents driving component F2. The PCA results demonstrated that amino acids level in melon varieties grown in each location were significantly influenced by the cultivar. This suggested that the selection of cultivars that are suitable for the specific location is essential for achieving higher levels of amino acids.

Total Phenolics, α-Amylase and Antioxidant Activities
Cantaloupe varieties grown at different locations showed significant differences in total phenolics content and antioxidant activities ( Figure 8A-D). The Folin-Ciocalteu (FC) method was used to estimate the total content of phenolics in cantaloupes. F-39 showed significantly higher total phenolics content in IN (489.76 ± 15.49 µg/g) followed by NC (419.72 ± 15.09 µg/g) and GA (387.10 ± 30.52 µg/g) compared to other locations ( Figure 8A). There were no significant differences in HT-IG melons grown in all locations. TT-DV had higher total phenolic contents in NC (381.91 ± 33.82 µg/g) and AZ (381.99 ± 16.21 µg/g) and had the lowest level in CA (224.56 ± 14.62 µg/g). Among the commercial varieties, PRI had a significantly lower amount of phenolics (218.78 ± 13.09 µg/g) compared to other varieties. Antioxidant properties of phenolic compounds protect the body from free radical damage. Phenolic compounds exhibit several physiological properties such as antidiabetic, anti-inflammatory, antioxidant, cardio-protective, and antitumor activities [33,61].
High concentrations of phenolic compounds are reported in melons with the highest amounts in the rind, followed by seeds then pulp [37]. The present results are in agreement with previously reported total phenolics contents (243.8 µg/g) in cantaloupe [62], although another study reported lower total phenolics in cantaloupe fruit juice (95.35 ± 9.23 µg/g) and pulp (101.90 ± 14.99 µg/g) [63]. The variation in the total phenolics content may be due to cantaloupe cultivars and growing locations. The prominent phenolic compounds reported in melon have been isovanillic acid, coumaric acid-hexoside, hydroxyl-benzoic acid hexoside, chlorogenic acid, apigenin-7-α-glucoside, ferulic acid, and luteolin-7-O-glucoside [36,44]. The antidiabetic potential of cantaloupe was evaluated by determining their capacity to inhibit the α-amylase activity that is responsible for breakdown of polysaccharides to monosaccharides. The α-amylase activity of cantaloupe extracts of different varieties was presented in Figure 8B. Results demonstrated cantaloupe have hypoglycemic properties and there was no significant difference in α-amylase inhibition activity among the varieties and growing locations.
Plants 2020, 9, x FOR PEER REVIEW 14 of 25 free radical damage. Phenolic compounds exhibit several physiological properties such as antidiabetic, anti-inflammatory, antioxidant, cardio-protective, and antitumor activities [33,61]. High concentrations of phenolic compounds are reported in melons with the highest amounts in the rind, followed by seeds then pulp [37]. The present results are in agreement with previously reported total phenolics contents (243.8 µg/g) in cantaloupe [62], although another study reported lower total phenolics in cantaloupe fruit juice (95.35 ± 9.23 µg/g) and pulp (101.90 ± 14.99 µg/g) [63]. The variation in the total phenolics content may be due to cantaloupe cultivars and growing locations. The prominent phenolic compounds reported in melon have been isovanillic acid, coumaric acidhexoside, hydroxyl-benzoic acid hexoside, chlorogenic acid, apigenin-7-α-glucoside, ferulic acid, and luteolin-7-O-glucoside [36,44]. The antidiabetic potential of cantaloupe was evaluated by determining their capacity to inhibit the α-amylase activity that is responsible for breakdown of polysaccharides to monosaccharides. The α-amylase activity of cantaloupe extracts of different varieties was presented in Figure 8B. Results demonstrated cantaloupe have hypoglycemic properties and there was no significant difference in α-amylase inhibition activity among the varieties and growing locations.
Previous studies have revealed that a strong correlation existed between phenolic compounds and inhibition of α-amylase enzyme. Phenolic compounds such as cinnamic acid, ferulic acid, chlorogenic acid, and caffeic acid were reported as potent inhibitors of α-amylase enzyme [64,65]. In a published study, polar extracts such as methanol and water had shown α-amylase inhibition, but their inhibition potential was lower than that of the hexane and chloroform extracts [66]. Similarly, inhibition of α-amylase activity of hexane and ethanol extracts of melon seeds was reported. It was found that although the ethanolic extract had a higher phenolics content as compared to the hexane extract, the α-amylase inhibiting activity was lower [46], suggesting that phenolic content is not the only factor determining α-amylase inhibiting activity and that other lipophilic or hydrophilic compounds may also play a role.
Free radical scavenging activity affects the synergistic or antagonistic behavior of phytonutrients, which define the nutritional and health-promoting functional qualities of a particular  Previous studies have revealed that a strong correlation existed between phenolic compounds and inhibition of α-amylase enzyme. Phenolic compounds such as cinnamic acid, ferulic acid, chlorogenic acid, and caffeic acid were reported as potent inhibitors of α-amylase enzyme [64,65]. In a published study, polar extracts such as methanol and water had shown α-amylase inhibition, but their inhibition potential was lower than that of the hexane and chloroform extracts [66]. Similarly, inhibition of α-amylase activity of hexane and ethanol extracts of melon seeds was reported. It was found that although the ethanolic extract had a higher phenolics content as compared to the hexane extract, the α-amylase inhibiting activity was lower [46], suggesting that phenolic content is not the only factor determining α-amylase inhibiting activity and that other lipophilic or hydrophilic compounds may also play a role.
Free radical scavenging activity affects the synergistic or antagonistic behavior of phytonutrients, which define the nutritional and health-promoting functional qualities of a particular food. Reactive free radicals are associated with several chronic conditions such as inflammation, diabetes, cancer, and cardiovascular disease. Therefore, several recent studies supported the utilization of antioxidants as tools for disease management [32,67]. In the present study, extracts from cantaloupes from different growing locations and varieties were made and the free radical scavenging activity was determined by DPPH and ABTS assays. The DPPH assay is based on the reactivity of free radical and hydrogen donors. Antioxidants react with DPPH and form a 1,1,-diphenyl-2-picryl hydrazine adduct with a characteristic absorption at 517 nm [68]. DPPH free radical scavenging activity of F-39 from AZ (115.46 ± 10.60 µg/g) was significantly higher followed by IN (100.49 ± 2.65 µg/g) and CA (89.15 ± 7.72 µg/g) compared to other locations ( Figure 8B). Similarly, TT-DV variety harvested from AZ (88.11 ± 2.85 µg/g) and TX (73.61 ± 8.77 µg/g) had significantly higher, while GA (51.85 ± 2.72 µg/g) had the lowest free radical scavenging activity. Among the commercial local varieties, higher DPPH activity was observed in ALG (91.10 ± 1.92 µg/g) as compared to varieties grown in different locations. Results of the ABTS assay showed F-39 harvested from TX (231.01 ± 56.85 µg/g) and IN (170.77 ± 7.56 µg/g) had significantly higher activity compared to other locations. Similarly, higher ABTS activity was found in TT-DV from TX (290.29 ± 53.57 µg/g), while in commercial local varieties, PRI from TX and ATH from GA had significantly higher activity 230.81 ± 46.16 and 169.73 ± 5.23 µg/g, respectively, than other varieties grown in different locations ( Figure 8C,D).
There was variation in the activities obtained from DPPH and ABTS assays, possibly due to the type of reaction mechanisms, hydrogen atom transfer, and single electron transfer, respectively [68,69]. Moreover, the differences may reflect differences in the type of antioxidants present in the fruit, as DPPH · is scavenged by hydrophobic antioxidants and ABTS · is scavenged by hydrophilic and hydrophobic antioxidants. Our results are in agreement with previously reported ABTS free radical scavenging activity (153.3 ± 18.0 µg/g) in fresh-cut cantaloupe [62]. Similarly, in another study, ABTS activities reported for cantaloupe fruit juice and pulp were 323.21 ± 53.80 and 220.38 ± 40.50 µg/g, respectively [63]. Overall, higher antioxidant activities were reported in melon due to the presence of total phenolics [37,70]. Correlation between total phenolics and antioxidant activities were carried out with Pearson's correlation coefficient (Table S2). The total phenolics content of all melon varieties is positively correlated with antioxidant activity. Likewise, DPPH and ABTS are positively correlated in all varieties except commercial local varieties.
Antioxidant activity is also influenced by the structure of amino acids. Amino acids with strong reducing functional groups, such as hydroxy groups, attached with phenolic groups (tyrosine) and thiols (methionine and cysteine), have higher antioxidant activity than acidic (glutamic and aspartic) and neutral amino acids (proline, and serine) [71]. Several mechanisms, such as radical scavenging ability or chelation of metals which can quench singlet oxygen or breakdown hydroperoxide into free radicals, are proposed to explain antioxidant activity of amino acids [72,73]. For example, amino acids with thiol groups (methionine and cysteine) act as a metal chelator and represent 40%-80% of the total antioxidant activity of human serum albumin [74]. Methionine and cysteine also have the ability to reduce oxidative stress caused by lead [75]. Amino acids have also been shown to produce a synergistic effect with tocopherol which improves antioxidant activity [76]. Synergistic effects of methionine with epigallocatechin gallate (EGCG) and ascorbic acid have also been reported [77]. GABA can act as signaling molecules and exhibits reactive oxygen species (ROS) scavenging activity [78]. GABA activates the transcription of enzymes involved in phenolic biosynthesis, which promotes the accumulation of phenolic compounds and enhances antioxidant activity [79]. Yen et al. reported that GABA and total phenolic content were positively correlated during germination of hull-less barley. Moreover, phenolic, ferulic, and p-coumaric acids had a significantly positive correlation with GABA as compared to vanillic and caffeic acids [79]. Similarly, in tomato plants, GABA enhanced the level of gallic and coumaric acids under salinity conditions [80]. Our results show that variety F-39 had high GABA ( Figure 4) and total phenolics content ( Figure 8A) in NC and IN; this may be due to accumulation of phenolic acids such as ferulic and p-coumaric acids, which are known as prominent phenolic compounds in cantaloupe [36,44]. Similarly, the TT-DV variety also had high levels of GABA and total phenolics content in NC, IN, and AZ. Phenylalanine (aromatic amino acid) plays a significant role in plant metabolism through the phenylalanine ammonia-lyase (PAL) pathway and is involved in the formation and accumulation of phenolic compounds [81,82]. Our results show that Phe content in F-39 and TT-DV was higher in TX, IN, and AZ compared to other locations, which correlates with higher antioxidant activity of the fruits harvested from these locations. This may be due to the formation of phenolic compounds through the PAL pathway ( Figure 8C,D).
The data were also analyzed in order to determine if the amino acid levels correlate with the antioxidant assays results (Figure 9, Figure 10, Figures S5 and S6). Amino acids Ser, Trp, and Met were positively correlated with antioxidant assays which vary with the variety. This suggested that amino acids play a significant role in antioxidant activities. Phongthai et al. reported that phenylalanine content in rice bran hydrolysate was positively correlated with DPPH and ABTS assays [83]. Figure 11 represents the tentative radical scavenging mechanism of selected amino acids. DPPH and ABTS assays usually measure the antioxidant behavior of the compounds through single electron transfer (SET) and also by hydrogen atom transfer (HAT) mechanism [84,85]. Aromatic amino acid, Trp, can act mainly through SET mechanism while hydroxy-containing amino acids such as Ser may act through HAT mechanism. Generally, both mechanisms occur in parallel, but depending on the amino acid structure and type of assays, one can dominate [84]. Ketnawa et al. reported that protein hydrolysates and peptides comprising a higher content of aromatic amino acids such as Tyr, Trp, and His have been linked with strong antioxidant activities via electron transfer mechanism [86].
Overall, similar to the results of the amino acids, the results of total phenolics and antioxidant activities showed significant variation among the varieties and growing locations, which may be considered as a factor for future melon breeding. Indeed, our results emphasize the need to consider location in breeding programs and suggest that there is ample room for improvement of the local adaptation of specific cultivars for growth in different regions. Therefore, our study will inform the selection of germplasm and location for future breeding efforts to improve the nutritional value of this popular, healthful fruit.
Plants 2020, 9, x FOR PEER REVIEW 16 of 25 significant role in plant metabolism through the phenylalanine ammonia-lyase (PAL) pathway and is involved in the formation and accumulation of phenolic compounds [81,82]. Our results show that Phe content in F-39 and TT-DV was higher in TX, IN, and AZ compared to other locations, which correlates with higher antioxidant activity of the fruits harvested from these locations. This may be due to the formation of phenolic compounds through the PAL pathway ( Figure 8C,D). The data were also analyzed in order to determine if the amino acid levels correlate with the antioxidant assays results (Figures 9,10,S5 and S6). Amino acids Ser, Trp, and Met were positively correlated with antioxidant assays which vary with the variety. This suggested that amino acids play a significant role in antioxidant activities. Phongthai et al. reported that phenylalanine content in rice bran hydrolysate was positively correlated with DPPH and ABTS assays [83]. Figure 11 represents the tentative radical scavenging mechanism of selected amino acids. DPPH and ABTS assays usually measure the antioxidant behavior of the compounds through single electron transfer (SET) and also by hydrogen atom transfer (HAT) mechanism [84,85]. Aromatic amino acid, Trp, can act mainly through SET mechanism while hydroxy-containing amino acids such as Ser may act through HAT mechanism. Generally, both mechanisms occur in parallel, but depending on the amino acid structure and type of assays, one can dominate [84]. Ketnawa et al. reported that protein hydrolysates and peptides comprising a higher content of aromatic amino acids such as Tyr, Trp, and His have been linked with strong antioxidant activities via electron transfer mechanism [86].    Overall, similar to the results of the amino acids, the results of total phenolics and antioxidant activities showed significant variation among the varieties and growing locations, which may be  Overall, similar to the results of the amino acids, the results of total phenolics and antioxidant activities showed significant variation among the varieties and growing locations, which may be

Fruit Growing Locations
The  Figure 1). Weather data from all locations are given in Figure S2. Other growing conditions such as soil and irrigation are presented in Table S3. Harvested fruits were shipped by carrier in refrigerated conditions to the Vegetable and Fruit Improvement Center (Texas A & M University, College Station, USA) for analysis.

Sample Preparation
Cantaloupes were cut horizontally into two equal halves. The pulp was removed from the rind and cut into small pieces. Samples were blended for 1 min at maximum speed (Oster Blender with 12 speeds, 450 W). The blended samples (15 g) were mixed with 10 mL of methanol. The contents were homogenized for 2 min at 10,000 rpm (850 Homogenizer, Thermo Fisher Scientific, Waltham, MA, USA) and sonicated for 30 min. The extracts were centrifuged at 7826× g for 10 min and supernatants were filtered through Whatman filter paper No. 1. The residues were re-extracted twice with methanol (2 × 7 mL) to ensure complete extraction. The resultant supernatants were pooled and the final volume was recorded.

Amino Acid Derivatization
A 350-µL sample of the extracted supernatant was used for dansyl-chloride (DNS-CL) derivatization. To the above aliquot of the supernatant, 125 µL DNS-CL (0.01% in acetone), 415 µL of 5 mM sodium borate buffer, and 50 µL diamino heptane (internal standard 500 ppm) were added. Reaction mixtures were vortexed and incubated for 30 min in a water bath (60 ± 1 • C). Derivatization was stopped by adding 2 N acetic acid (50 µL). Later, the reaction mixture was centrifuged at 10,621× g for 5 min and the supernatant was analyzed by HPLC-FLD.
3.6. Total Phenolic Content, α-Amylase, and Antioxidant Activities 3.6.1. Total Phenolic Content The Folin-Ciocalteu (FC) reagent was used for the determination of total phenolic content from melon samples as described in our previous publication [87]. Briefly, extracted samples (10 µL) were added to different wells of a microplate and adjusted to 100 µL with methanol. An amount of 40 µL FC reagent (25%) was added and incubated for 10 min, then sodium carbonate (50 µL) was added to each well and incubated for further 20 min. The absorbance at 760 nm was read using a KC-4 Microplate Reader (BioTek Instruments, Winooski, VT, U.S.A.). A standard curve was developed using various concentration of gallic acid (0, 10,20,30,40,50,75, and 100 µg/mL). Results were expressed as gallic acid equivalents.

α-Amylase Activity
Inhibition of α-amylase was performed according to published protocol [66]. Aliquots (10 µL) of melon extract were pipetted into a 96-well micro plate and final volume was adjusted to 140 µL with a 1% saline solution. To each well, 45 µL of starch (1%) were added followed by 45 µL of amylase (1 unit/mL) to initiate the reaction. The reaction mixture was incubated at 25 • C for 60 min. Fifty µL of 3,5-dinitrosalicylic acid was added to the sample and standard wells followed by incubation for 60 min at 50 • C. Change in color was measured at 540 nm. Acarbose (80 µg/mL) and methanol were used as positive and negative controls, respectively. Different concentrations of dextrose (equivalent to 25, 50, 75, 100, 125, and 150 µg) were prepared to develop a standard curve.
3.6.3. 2,2-Diphenyl-1-Picrylhydrazyl (DPPH) Assay DPPH · radical scavenging activity was measured according to our published protocols [88]. Briefly, 10 µL aliquots of methanol extract were added to 96-well micro plates (KC-4 Microplate Reader) and adjusted to 100 µL with methanol. An aliquot of 180 µL of methanolic DPPH solution (0.1 mM) was added to each well. A reaction mixture was incubated for 20 min in the dark and the absorbance was recorded at 515 nm. Standard ascorbic acid was used to prepare the calibration curve and results were expressed as µg/g ascorbic acid equivalent.
3.6.4. 2,2 -Azino-Bis (3-ethylbenzothiazoline-6-sulfonic acid) (ABTS·) Assay ABTS· assay was performed according to the methodology reported in our published method [68]. Briefly, ABTS · was prepared by mixing 7 mmol/L aqueous ABTS solution and 2.45 mmol/L potassium persulfate (K2S2O8) solution. The resultant solution was kept in the dark for 16 h. Fresh standard ascorbic acid and methanol extract were prepared, aliquots (10 µL) were pipetted into 96-well micro plates, and the total volume of each well was adjusted to 40 µL with methanol. Freshly prepared ABTS solution (initial absorbance was 0.7 at 734 nm) was added to each well and absorbance was measured at 734 nm using a KC-4 Microplate Reader (BioTek Instruments, Inc., Winooski, VT, USA).

Statistical Analysis
Data were analyzed by principal component analysis (PCA) using Pearson's correlation method, Hierarchical Cluster analysis, and one-way analysis of variance (ANOVA) using XLSTAT software (Addinsoft, Paris, France). Five cantaloupe samples were analyzed from each variety and duplicate measurements (n = 5 × 2) were performed for each sample. Significant differences between means were observed by Tukey's HSD test at a 5% probability level (p ≤ 0.05).

Conclusions
In the present study, we performed comparative profiling of amino acids, phenolic contents α-amylase, and antioxidant activities in melons harvested from six different growing locations (Texas, Georgia, North Carolina, California, Indiana, and Arizona). Statistical analyses showed significant differences in amino acid levels, phenolics contents, and antioxidant activity in the melon varieties based on the growing location. Cantaloupes are good sources of amino acids, especially the neurotransmitters, nitric oxide precursors, and essential amino acids. PCA differentiated the melon varieties in all states by amino acid content. In addition, PCA results were supported by HCA grouping of the states based on the similarity in amino acid content. The findings from this study provide important insight into the influence of growing location on the accumulation of amino acids and phenolic compounds among the different cantaloupe varieties and could prove valuable for melon improvement.
Supplementary Materials: The following are available online at http://www.mdpi.com/2223-7747/9/9/1058/s1: Figure S1. Comparative HPLC-FLD chromatograms of amino acids from different cantaloupe varieties grown in Texas, Uvalde. Figure S2. Precipitation and temperature of cantaloupe growing locations. Figure S3. Results of hierarchical cluster analysis of melon varieties harvested from different locations (dendrograms using average linkage between groups). Figure S4. Results of hierarchical cluster analysis of four different melon varieties harvested from same locations (dendrogram using average linkage between groups). Figure S5. The relationship between selected amino acids and DPPH assay. Figure S6. The relationship between selected amino acids and ABTS assay. Table S1 p-value of amino acids. Table S2. Pearson's correlation coefficients (r) of total phenolics and antioxidant activities in different cantaloupe varieties. Table S3. Location and growing conditions for melon variety.