Plant material.

Fifty-six (Table 1) watermelon cultigens (open-pollinated and hybrids) grown in Lane, OK (Bernow fine-loamy, siliceous, thermic, Glossic Paleudalf soil) were arranged in a randomized complete block design with four replications. Plant management provided was as outlined in Motes and Cuperus (1995). Plants were transplanted onto bare, raised beds, were irrigated as needed (at the first signs of water stress), and fertigated bimonthly. To assess environmental effects, 10 watermelon cultigens (Table 2) were also grown at College Station, TX [Zack very fine sandy loam (fine, smectitic, thermic Udertic Paleustalfs)] using the same planting scheme. Pre-plant fertilizer was added according to soil test results and plant care provided as outlined in Masabni et al. (2011). The plants were transplanted on 21 May onto black plastic mulched beds that were drip-irrigated using a timer to irrigate daily based on evapotranspiration rates. All plants were grown during the summer of 2010 and were harvested throughout the season. Flesh from all fruit harvested from Lane, OK, and from a representative collection of fruit from College Station, TX, were collected from ripe fruit only. Maturity was assessed by external and internal characteristics (i.e., waxiness, tendril death, Brix, firmness, seed maturity). A digital refractometer was used to determine Brix. Samples were collected from the center of each fruit, pureed for 3 min with a Polytron PT 10-35 grinder (Kinematica AB, Lucerne, Switzerland) set at medium speed and stored at –80 °C until analyzed for L-citrulline.

Table 1.L-citrulline concentrations averaged by cultigen across replications at Lane, OK.^z

View Fullscreen

Table 2.L-citrulline concentration for cultigens produced at two locations and averaged across locations.^z

View Fullscreen

Citrulline quantification by thin-layer chromatography plates.

L-citrulline was analyzed using a thin-layer chromatography (TLC) plate method, which was a slight modification of a Brenner and Niederwieser (1960) method. L-citrulline was quantified on the basis of an L-citrulline standard (Sigma-Aldrich, St. Louis, MO). Briefly, 40-g samples were pureed using a Brinkmann Polytron Homogenizer (Brinkmann Instruments, Inc., Westbury, NY) with a 20 mm O.D. blade. One milliliter of the liquid puree was centrifuged at 15,800 g_n for 10 min to remove debris. Supernatants were diluted to make a 10% and a 20% solution in deionized water. Ten microliters of the diluents were loaded on a 20 × 20-cm silica gel matrix (200-μm layer thickness, 5- to 17-μm particle size) TLC plate (Sigma-Aldrich). The spots were air-dried and amino acids were resolved using a solvent (2:1:1 n-butanol:acetic acid:deionized water). Plates were developed with 0.2% ninhydrin in ethanol by baking at 95 °C for 5 to 10 min. Densitometric scans of the citrulline spots were visualized and calculated against standards using a Kodak Image station (Model 440CF; Eastman Kodak, Rochester, NY). Data are in mg·g⁻¹ fresh weight, because 1 mL of watermelon puree is very nearly and consistently 1 g.

Citrulline quantification by high-performance liquid chromatography.

Frozen tissue was thawed and then centrifuged at 10,000 g_n for 10 min at 15 °C to remove insoluble components. L-citrulline was quantified by a method first reported in Chang et al. (1983) and subsequently modified by Krause et al. (1995) and by Sethuraman et al. (2004) for the determination of physiological amino acids and biogenic amines by reversed-phase high-performance liquid chromatography (HPLC) separation of their dabsyl derivatives. The Sethuraman et al. (2004) method was recently modified by Fish and Bruton (2010) for extraction and quantification of physiological amino acids from cucurbits. The gradient program of Sethuraman et al. (2004) was used. Separation/quantification was performed on a Varian ProStar tenary solvent HPLC system equipped with an autosampler and diode array detector (Varian, Walnut Creek, CA). A 250 mm × 4 mm, 5-μm Luna™ C18 reversed-phase column was used (Phenomenex, Torrance, CA) at 50 °C and at a flow rate of 1.0 mL·min⁻¹. The dabsyl derivatives eluting from the column were monitored by absorbance at 468 nm. Dabsyl chloride [4-(4-dimethylaminophenylazo) benzenesulfonyl chloride] was purchased from Pierce (Pierce Biotechnology, Rockford, IL), and the amino acids used for calibration of the method were purchased from Sigma-Aldrich as a commercial 18 amino acid calibration mixture. The physiological amino acids, L-citrulline, L-glutamine, L-asparagine, and γ-aminobutyric acid (all purchased from Sigma-Aldrich), were quantitatively prepared individually and as a mixture. The suspected identity of L-citrulline in watermelon was confirmed by cochromatography of the extract with the authentic individual amino acid. Calibration curves for each amino acid ranged between 0.2 and 1.6 nM of the amino acid on the column, and the unknowns were diluted to this concentration range for analysis.

Data analysis.

Data were subjected to analysis of variance using PROC GLM (Version 9.1.3; SAS, Cary, NC). Responses to main effects were determined and means separated using the Ryan-Einot-Gabriel-Welsch multiple F test.

Comparison of high-performance liquid chromatography to the modified thin-layer chromatography plate method.

Amino acid analysis of protein hydrolysates is generally considered a routine procedure; however, quantification of physiological L-citrulline is often complicated by the presence of L-glutamine. The method used here estimates L-citrulline and L-glutamine together. Because ripe watermelon contains from three to 10 times more L-citrulline than L-glutamine (Fish and Bruton, 2010), the amount of L-glutamine in the samples should not adversely affect the reliability of comparisons of citrulline levels among samples. To verify this, 14 watermelon samples were compared using the TLC plate method and HPLC methods mentioned previously (Fig. 1). Although the TLC plate method used is not as precise as HPLC, and although L-glutamine might increase the estimated L-citrulline when using the TLC plate method, this method gave L-citrulline concentrations in the same general range as HPLC. This range was adequate for estimating L-citrulline within a set of watermelon samples when compared with another set of watermelon samples. As a practical estimate of the average deviation from the HPLC values, the mean absolute deviation of the 14 samples from its HPLC value were determined to be 0.35 mg·g⁻¹. Considering a 0.35 mg·g⁻¹ deviation in the 14 samples, which ranged from no detectable citrulline to samples containing 1.8 mg·mL⁻¹, this suggests an inaccuracy of ≈20%. This is an acceptable amount of error for use as a screening tool to find large changes in citrulline concentrations resulting from genetics or environmental effects. However, for exact values of a small number of samples, HPLC should be used.

View Figure

• Download as Powerpoint
• Download Figure

Environment effect.

L-citrulline concentration was affected by environment (average of all fruit tested in Lane, OK = 3.10 mg·g⁻¹ fresh weight and in College Station, TX = 1.67 mg·g⁻¹ fresh weight) (Table 3). All cultigens had higher L-citrulline concentration in Lane, OK, than College Station, TX (Table 2). When combined across locations for the 10 cultigens, there was not a significant environment by cultigens interaction (P = 0.475), but there was a significant difference between locations (P = 0.001 level) and cultigens (P = 0.05 level, discussed in the next section). Our previous study (Davis et al., 2010) did not show a statistical difference for L-citrulline among three locations; however, fewer samples were tested in that study. Additionally, Fish and Bruton (2010) reported no difference in flesh L-citrulline levels in two locations in the same year, but again few samples (five per location) were tested in their study. The amount of environmental effect varies among cultigens. ‘Charleston Gray’ had the highest between environment mean difference (2.50 mg·g⁻¹ fresh weight) and ‘Starbrite F1’ and ‘Minilee’ had the most stable between environment mean difference (0.85 and 0.86 mg·g⁻¹ fresh weight, respectively).

Table 3.L-citrulline concentration differences between locations when comparing the 10 cultigens common to both locations.^z

View Fullscreen

Within-cultigen differences were high in some cultigens. For example, ‘Congo’ had a 1.26 to 7.21 mg·g⁻¹ fresh sample deviation at one location. Of the lines in which 10 or more fruit were tested, ‘AU-Jubilant’ had the most stable L-citrulline expression, 2.23 to 4.03 mg·g⁻¹ fresh weight deviation at one location (data not shown). It appears that environment does not significantly increase within-cultigen variation (relative se of the 10 cultigens tested at each location was ± 35.3% for College Station, TX, and ± 32.9% for Lane, OK) (data not shown). ‘AU-Jubilant’ would be a likely candidate for development of cultivars having stable L-citrulline expression. However, ‘AU-Jubilant’ should be tested in more environments over several years to verify that conclusion.

The notable production differences between the two locations were (in Oklahoma) bare ground, fertigation, and watering when plants start to show water stress vs. (in Texas) black plastic mulch, pre-plant fertilization, and watering daily as needed based on evapotranspiration rates. Environmental conditions (data not shown) averaged over a 2-week period before harvesting through the last day of harvest [main growing period for L-citrulline accumulation according to Fish and Bruton (2010)] at each location revealed the following: the Texas location had an average air temperature of 3 °C higher than the Oklahoma production location. Average and total rainfall in Texas was approximately half that received in Oklahoma but because additional irrigation was applied at both locations, the amount of rain was probably not a factor in L-citrulline accumulation. Relative humidity and average total daily solar radiation were negligibly different between both locations and therefore most likely did not play a role in L-citrulline accumulation. Looking at these different environmental and production environments, it is likely the following have an effect on L-citrulline accumulation: soil temperature, moisture, and fertility and air temperature. Because previous reports (Akashi et al., 2001; Yokota et al., 2002) have shown that drought stress increases watermelon citrulline concentration in leaf tissue, it is provocative to speculate that drought stress can also increase the concentration of this healthful amino acid in fruit itself.

Because environment had such a noticeable effect on L-citrulline concentration, watermelon cultigens should be tested in multiple environments over multiple years to verify location and/or environmental effects and to determine which cultigens perform well under many different environmental conditions.

Cultigen differences.

Because there was high within-cultigen variance for L-citrulline (discussed previously), the cultigens could not statistically be separated into low-, medium-, and high-expressing fruit. However, there was a statistical difference between the cultigens with the lowest and highest values (Table 1). Cultigens with the highest L-citrulline concentration (from 3.57 to 4.52 mg·g⁻¹ fresh weight) were ‘Tom Watson’, PI 306364, and ‘Jubilee’. Cultigens with the lowest L-citrulline concentration were PI 189317, PI 161373, PI 164993, PI 181742, PI 162667, ‘Medium Sugar Watermelon 28’, ‘Low Sugar Watermelon 194’, PI 164992, and ‘Low Sugar Watermelon 177’ (from 1.09 to 2.18 mg·g⁻¹ fresh weight). Although the PI accessions tended to have lower L-citrulline than the elite cultivars, there was no correlation between amount of this amino acid and watermelon type (open-pollinated vs. F1 hybrid), color (red, orange, salmon yellow, vs. white), fruit size, fruit shape, rind color or pattern, or fruit earliness (data not shown). Interestingly, cultigens with low soluble solids (° Brix) tended to have low L-citrulline concentration. However, cultigens with high soluble solids did not necessarily have higher L-citrulline (data not shown).

The L-citrulline concentrations for the watermelon in this study (2.97 mg·g⁻¹ fresh weight) were in the same range as previously published studies. Davis et al. (2010) reported an overall value of 2.4 mg·g⁻¹ fresh weight. Liu et al. (2010) reported an overall value of 2.27 mg·g⁻¹ fresh weight for mature diploid cultivars. Rimando and Perkins-Veazie (2005) reported an overall average of 2.4 mg·g⁻¹ fresh weight from 14 cultivars.

Rimando and Perkins-Veazie (2005) suggested that canary yellow and orange-fleshed fruit have higher L-citrulline values. In this study, there was no correlation between color and L-citrulline concentrations. ‘Tendersweet Orange Flesh’ was tested in this trial and in the Rimando and Perkins-Veazie trial and their average L-citrulline value for this cultigens was 1 mg·g⁻¹ fresh weight higher than our average value and is likely the result of environmental effects.