Red skin radish (Raphanus savitus L. var. “Cherry Belle”) seeds were soaked in deionized water for about 12 h, and then put in moist gauze to germinate. One-day-old uniform seeds were selected and laid on gauze in plastic containers containing deionized water. Containers were maintained in an incubator (Zhejiang United Saifu Laboratory Instrument Co., Ltd., Ningbo, China) in dark at 25°C for another 48 h. Then the sprouts were treated with different solutions and transferred into another incubator with white light (100 μmol⋅m^-2⋅s^-1) for another 24 or 48 h.

One liter of Hoagland nutrition solution contains 945 mg Ca(NO₃)₂⋅4H₂O, 506 mg KNO₃, 80 mg NH₄NO₃, 136 mg KH₂PO₄, 493 mg MgSO₄⋅7H₂O, 13.9 mg FeSO₄⋅7H₂O, 18.65 mg EDTA-Na, 2.86 mg H₃BO₃, 1.81 mg MnCl₂⋅4H₂O, 0.22 mg ZnSO₄⋅7H₂O, 0.051 mg CuSO₄⋅5H₂O and 0.12 mg Na₂MoO₄⋅2H₂O. For N deficiency, when nutrition solution was prepared, KNO₃, Ca(NO₃)₂4⋅H₂O and NH₄NO₃ were not used, while KCl, CaCl₂ were added to keep the constant concentration of K and Ca, and others were the same as normal Hoagland nutrient solution. For P deficiency, KH₂PO₄ was replaced by KCl, so that P was deficiency in the solution but K was constant. For S deficiency, MgSO₄ was replaced by MgCl₂. For K deficiency, KNO₃ and KH₂PO₄ was not used, while more NH₄NO₃ and NaH₂PO₄ were added to supplement N and P. For Ca deficiency, Ca(NO₃)₂4H₂O was not used, while more NH₄NO₃ was add to keep the constant N concentration. For Mg deficiency, MgSO₄ was not used. After preparation, the pH value of all these solutions was adjusted to 6.0.

The determination of anthocyanin content in the radish hypocotyls was according to the method developed by Su et al. (2014), which involves measuring the absorbance (530) of extracts.

Hypocotyls of radish sprouts were transected by a blade and observed under a stereoscopic microscope (Stemi 2000-C; Carl Zeiss, Germany). Pictures were photographed on a color film (Powershot A620, Canon Photo Film, Japan).

Total RNA was extracted from radish hypocotyl samples using Trizol extraction reagent (Invitrogen, Gaithersburg, MD, USA) and high purity of RNA with ratio of 260/280 nm > 1.9 was used. First-strand cDNA was synthesized in a 20 μL reaction volume (Thermo Scientific, MD, Lithuania) containing 1 μL of RevertAid M-MuLV reverse transcriptase and 1 μL of oligo (dT)₁₈ primer according to the manufacturer’s instructions. A Mastercycler^® ep realplex real-time PCR system (ABI7500, MD, USA) with Bestar^® SybrGreen qPCR mastermix (DBI, Bioscience Inc., Germany) in a 20 μL reaction volume was used to perform the real-time quantitative PCR reactions according to user manual.

Primer Expressversion 3.0 (Applied Biosystems) was used to design all PCR primers targeting actin, PAL, CHS, CHI, F3H, DFR, LDOX, ANS, and UFGT (Su et al., 2014). All primers (Supplementary Table 1) were synthesized by Genewiz Bio-engineering Ltd. Company (Suzhou, China). The identification of PAL, F3H, ANS and UFGT in radish genes was based on using their Arabidopsis orthologs for homology search in databank of R. sativus available at http://bioinfo.bti.cornell.edu/radish (Shen et al., 2013). The transcription levels were presented as values compared to those of corresponding control samples, after normalization to actin expression levels.

Hypocotyl samples (0.05 g FW) were ground in a mortar with liquid nitrogen, and the powder transferred with 3 mL of a phosphate buffered solution (pH 7.0) into centrifuge tubes. After 15 min centrifugation at 13,000 g (4°C), 0.1 mL of the supernatant was combined with 5 mL of Coomassie brilliant blue G-250 solution. Two minutes later, the soluble protein content (mg g^-1 FW) was determined at a wavelength of 595 nm.

Fresh samples of the radish hypocotyls (1 g) were ground in a mortar with 2.5 mL of distilled water. The homogenates were centrifuged at 10,000 rpm for 15 min, and the supernatant was used to analyze the contents of soluble sugar, free amino acid, glucose (Glu), fructose (Fru) and sucrose (Suc). The soluble sugar content was determined using the sulfuric acid anthrone method with measurements conducted on a spectrophotometer (UV-5200 spectrophotometer, Shanghai Metash Instruments Co., Ltd, Shanghai, China) at a wavelength of 630 nm (Morris, 1948). Free amino acids content was determined using the ninhydrin method with measurements done at a wavelength of 570 nm (Moore and Stein, 1948).

Glc, Fru, and Suc levels were determined by the method from Gordon et al. (1997). Briefly, for the determination of Glc, the extract was incubated with 200 μL of buffer (50 mM imidazole, 1 mM MgCl₂, 0.75 mM NAD, 0.85 mM ATP) containing 0.04 unit of Glc-6-P dehydrogenase and 0.1 unit of hexokinase, and then 50 μL samples were assayed in 96-well plate. For Fru and Suc, phosphoglucose isomerase (0.4 unit/well) and acid invertase (20 units/well), respectively, were added in the mix before measurement using a plate reader (TECAN Infinite M 200) at 340 nm.

Hypocotyl samples (0.05 g FW) were ground in a mortar with liquid nitrogen and then the powder was transferred together with 3 mL of a phosphate buffered solution (pH 7.0) into centrifuge tubes. After 15 min centrifugation at 13,000 g (4°C), the supernatant was collected for the analysis of enzymes activities. Enzyme activities are expressed as moles of metabolite generated/consumed per milligram of protein per unit of time.

SPS activity measurements were based on the method from Klann et al. (1993), and 1 U = 0.5 μmol h^-1.

SS activity was determined according to Klann et al. (1993). Absorbance was measured at 540 nm using Fru as a standard with 1 U equal to 0.5 μmol h^-1.

For NR, radish hypocotyl tissue (0.1 g) was ground in 1 mL of buffer containing 50 mM KH₂PO₄-KOH buffer, pH 7.5, 2 mM EDTA, 2 mM dithiothreitol, and 1% polyvinylpolypyrrolidone. NR activity was measured according to the method from Reguera et al. (2013). The activity of NR was expressed as the amount of NO₂^- produced per unit of fresh weight per hour and 1 U was 0.5 μmol h^-1.

GS were determined according to O’Neal and Joy (1973), 1 U was 1 μmol h^-1.

Values presented are means ± standard deviation (SD) of three replicates. Data was subjected to analysis of variance (ANOVA), and mean values were compared by Duncan’s and Tukey’s multiple range test (p < 0.05). All the statistical analyses were performed using SPSS 19.0 for Windows.

In our preliminary experiments, an interesting phenomenon was observed in which the color of radish hypocotyls varied depending on the composition of the culture solution used. As the main pigment responsible for the red color in radish sprouts are anthocyanins, the anthocyanin contents in the hypocotyls of radish sprouts grown in Hoagland solutions of different strengths were measured (Figure 1A). The results showed that anthocyanin contents increased with duration of the treatment, peaking at 32 h, after which anthocyanin contents remained at relative steady levels. In addition, amount of anthocyanins decreased with increase of the strength of Hoagland solution, with almost twofold higher level being measured in radish sprouts grown in deionized water compared with those grown in full strength Hoagland solution.

The above finding was used to further investigate whether a specific component of the Hoagland solution effected the anthocyanin level. A single-factorial experiments were designed with one of the nutrients [nitrogen (N), phosphorus (P), sulfur (S), potassium (K), calcium (Ca) and magnesium (Mg)] being removed at a time from the Hogland solution. Radish seedlings were grown in each of the modified Hogland solutions and assessed for anthocyanin contents (Figures 1B–D). As shown in Figure 1B, the morphological characteristics of radish sprouts grown in various conditions were similar, while a considerable increase of anthocyanin content was observed only in the absence of N in the nutrient solution (Figures 1C,D), suggesting that N element is one of the major contributing factors to the observed phenomenon.

To investigate the effects of N in the regulation of anthocyanin biosynthesis, radish seedlings were grown in Hoagland solutions containing various amounts of N (from 0 to 8.0 mM) and anthocyanin contents were assessed in the hypocotyl tissues (Figure 2A). As expected, the contents of anthocyanin accumulated depended on the N concentration in the nutrient solution with a progressive decrease of the anthocyanin levels according to the increase of N concentration. Additionally, hypocotyls of sprouts grown under excessive N showed more soluble proteins but less soluble sugars (Figures 2B,C).

Suc, Fru and Glu are three main forms of soluble sugars in plants (Rosa et al., 2009). In the hypocotyls of radish sprouts, the concentration of Suc was the highest (approximate 13 mg/g FW), followed by Fru (approximate 10 mg/g FW), with relatively low concentration in Glu (approximate 5 mg/g FW) detected (Figure 3A). When subjected to N deficiency, levels of all these soluble sugars were considerably increased, with contents being 40, 20, and 40% higher for Suc, Fru and Glu, respectively (Figure 3A). Addition of these soluble sugars enhanced accumulation of anthocyanins in the hypocotyls. The biggest changes were observed on addition of Suc with over twofold increase under 10 mM Suc concentration. The response was dose-dependent, with increase in anthocyanin contents in response to increase of sugar concentrations (Figure 3B). Exogenous addition of 50 mM soluble sugars inhibited the growth of radish sprouts, thus a concentration of 10 mM was selected for the following experiments.

To further determine the relationship between N deficiency and soluble sugars in anthocyanin accumulation, sprouts were grown in N-deficient conditions with addition of different soluble sugars. Addition of 10 mM Suc significantly increased the anthocyanin accumulation in the hypocotyls compared to control, whereas only slight enhancement of anthocyanin content was observed under addition of 10 mM Glu and Fru (Figure 3C). Therefore, effect of Suc on the anthocyanin biosynthesis was explored further. In addition, we observed that sprouts subjected to addition of Glu or Suc under N deficiency showed much higher level of anthocyanins as those grown under Glu or Suc with presence of N (Figure 3C).

The results of cross section showed that anthocyanins mainly accumulated in the epidermis of the hypocotyls of radish sprouts (Figure 4A), and compared with control, addition of Suc markedly increased the content of anthocyanins in the hypocotyls. DCMU [3-(3,4-dichlorophenyl)-1,1-dimethylurea] inhibits photosynthetic electron transport, and consequently reduce the production of chemical energy (ATP) and reducing power (NADPH), resulting in inhibition of carbon fixation process and, eventually, sugar biosynthesis (Jeong et al., 2010). Addition of DCMU decreased the anthocyanin accumulation significantly, which was reverted by Suc supplementation (Figures 4A,B). N deficiency also induced observable increase in anthocyanin content compared to control. Addition of 10 mM Suc to N-deficient solution enhanced anthocyanin content by 40%, whereas addition of DCMU completely eliminated the effect (Figure 4B). Supplementing Suc and DCMU together to growth solution negated each other leading to anthocyanin content being similar to those in N-deficient solution. Change of growth conditions also affected amounts of soluble sugars accumulated in the hypocotyls of radish sprouts, with addition of Suc causing their substantial increase under both growth conditions (the presence and absence of N in the nutrient solutions, Figure 4C). The trend observed for soluble sugars was similar to changes in anthocyanin levels described above (Figures 4B,C, respectively).

In addition to the content of anthocyanins, the transcript levels of anthocyanin biosynthesis-related genes (PAL, CHS, CHI, F3H, DFR, UFGT, LDOX, and ANS) were measured under different treatments. As shown in Figure 5, the trends of changes in gene transcriptions were in agreement with the changes of anthocyanin content under the same growth conditions. Indeed, sprouts had substantially higher expression of anthocyanin biosynthesis-related genes under N deficiency compared to genes expressed under control conditions (full strength Hoagland solution) either in the presence or absence of Suc, DCMU and their combination. The highest transcript levels of the genes were observed in the hypocotyls of sprouts subjected to N deficiency with addition of Suc. The latter phenomenon, however, disappeared when sprouts were treated with DCMU.

The activity of SS, sucrose phosphate synthase (SPS), GS and NR were determined to further investigate the effects of N deficiency on Suc biosynthesis and N metabolism (Figures 6A–D). Results showed that N deficiency positively regulated SS activity (Figure 6A) and negatively regulated NR activity with the latter being reduced by nearly twofold (Figure 6D). No effects on the activities of SPS and GS were found (Figure 6B,C). Nitrogen deficiency also negatively affected free amino acids, soluble proteins levels, and NR activity but not activity of GS (Figures 6E–H).

We also assessed effects of Suc presence in the growth solutions on levels of amino acids and soluble proteins and enzymes activities. Addition of Suc to full strength Hoagland solution (control) significantly reduced the content of soluble proteins and activities of GS and NR (Figures 6F–H). Suc did not affect levels of free amino acids in either control or N-deficient solution (Figure 6E). No change was also found in GS activity under N-deficient condition (Figure 6F). At the same time addition of Suc to N deficient solution led to a decrease of NR activity and to amounts of soluble proteins, similar to the trends observed under control conditions (Figures 6F,H).