Twelve cultivars of rice (Oryza sativa L.) commonly grown in China were used, six each of japonica (HY1, WLJ1, WYJ7, LJ9, WYJ21, and NJ9108) and indica (YLY2, YD6, YLY800, LY808, SLY862, and LY1259). The rice seeds were surface-sterilized with 30% H₂O₂ for 15 min, washed thoroughly with deionized water, soaked in deionized water for 4 h, and germinated at 25°C in darkness on a piece of gauze placed inside a polythene container. When the seedlings had grown about 2 cm tall, they were moved to a growth chamber with day/night temperatures of 27/20°C, a day length of 14 h, a light intensity of 375 μmol photon m^-2 s^-1, and a relative humidity of 70% and grown in a modified nutrient solution recommended by the International Rice Research Institute without being aerated (Fan et al., 2007). The composition of the nutrient solution (all figures in millimoles) was as follows: 0.75 (NH₄)₂SO₄, 1.5 NaNO₃, 0.32 NaH₂PO₄⋅2H₂O, 0.5 K₂SO₄, 1.7 MgSO₄7H₂O, and 1.0 CaCl₂, supplemented with (all figures in micromoles) 9.1 MnCl₂⋅4H₂O, 0.16 CuSO₄⋅5H₂O, 0.15 ZnSO₄⋅7H₂O, 0.07 (NH₄)₆Mo₇O₂₄⋅4H₂O, 18 H₃BO₃, and 40 FeSO₄⋅7H₂O-EDTA. The nutrient solution (pH 5.5) was renewed every 3 days. After 15 days, uniform-looking seedlings were selected and transplanted into PVC pots (14.5 cm tall and 10 cm in diameter, each holding 12 plants). The plants were carefully removed after 25 days and their roots, washed in deionized water, were used for the subsequent experiments.

Before the adsorption experiments, the plants were placed in deionized water for 12 h to remove excess nutrient ions from the root surfaces. The roots, washed three times in deionized water and then carefully blotted, were immersed in 1 L solution of Cu or Cd, and magnetically stirred. There are three initial concentrations of 10, 40, and 100 μM for both metals. After 2 h, the roots were removed, washed twice in deionized water, blotted dry with filter paper, and placed sequentially in 1 L each of 0.1 M KNO₃, 0.05 M EDTA-2Na (pH 6), and 0.01 M HCl, 1 h in each solution, to extract the exchangeable, complexed, and precipitated Cu or Cd from the root surfaces, respectively (Liu and Xu, 2015; Liu et al., 2016), and their contents in the extractants were determined using atomic absorption spectrophotometry (nov AA350, Analytik Jena AG, Jena, Germany). After the rice roots were exposed to metal solution for 2 h, the amount of Cu and Cd entered the symplasm during the experiment should be much lower, thus they were not measured and discussed in this study. Three replicates were maintained of each cultivar for each treatment.

The adsorption of Cu and Cd by rice roots was performed without oxygenation. It was possible that some of Cu(II) and Cd(II) were reduced to Cu(I) and Cd(I) by root exudates during the adsorption experiments. However, the amount of Cu(II) and Cd(II) reduced should be low due to weak reducing condition and short reaction time of 2 h.

The zeta potential of rice roots was measured using the streaming potential equipment developed by our group (Li et al., 2015). The roots were placed into the measuring cells (length, 3 cm; internal diameter, 1.4 cm) of the streaming potential equipment. A multimeter was used for measuring the streaming potential (ΔE, mV) through a pair of non-polarizable Ag/AgCl electrodes. A conductivity meter was used for measuring the electrical resistance of the measuring cell with a pair of platinum electrodes. All the electrodes were attached to the ends of the measuring cell. The difference in the pressure of liquid (ΔP, Pa) between the two ends of the measuring cell was recorded using a liquid manometer. Electrolyte solutions were pumped into the measuring cell, and the streaming potential was varied along a hydraulic gradient (ΔP, Pa), achieved by adjusting a valve, and the ratio of ΔE to ΔP was calculated. The zeta potential (ζ) was calculated using the Helmholtz–Smoluchowski equation (Childress and Elimelech, 1996) as follows:

where ΔE is the streaming potential (mV); ΔP is the difference between the liquid pressures (Pa); μ is the dynamic viscosity of the solution (Pa S); ε is the permittivity of the test solution (F/m); ε₀ is the permittivity of free space (F/m), and κ is the conductivity of the solution (S/m).

The ATR-FTIR spectra of rice roots exposed to 100 μM Cu or 100 μM Cd were measured using a Nicolet iS10 FTIR spectrometer (Nicolet Analytical Instruments, Madison, WI, United States) equipped with an ATR diamond crystal (the roots were placed on the crystal). The spectral range was 650–4000 cm^-1 with a resolution of 4 cm^-1 and 64 scans.

The average values of Cu and Cd adsorbed by the roots of each of the 12 cultivars, six japonica and six indica, and their average zeta potentials were calculated and the results were expressed as means ± standard deviation. Data processing and statistical analyses were carried out using SPSS ver. 20.0 for Windows (Chicago, IL, United States). One-way analysis of variance (ANOVA) was used in all experiments to ascertain whether the differences between treatments were significant at P < 0.05 (Pearson’s correlation).

After the roots had been exposed to Cu or Cd, the metal cations on the roots could be differentiated into their respective exchangeable, complexed, and precipitated forms (Figure 1). The exchangeable form of both metal cations accounted for the largest share, followed, in that order, by the complexed and the precipitated forms, the share of exchangeable Cd being greater than that of Cu. For example, when the concentration of metals was 100 μM, exchangeable Cu on the root surface accounted for 50.9% of the total Cu in indica and 48.7% in japonica, whereas the corresponding values for exchangeable Cd were 69.1 and 71.2%. However, complexed Cu accounted for a greater proportion in the total than complexed Cd did. At the initial concentration of 100 μM, complexed Cu contributed 37.0% of the total in indica and 38.7% in japonica, the corresponding figures for Cd being 15.2 and 15.6%. Exchangeable Cu and Cd on the roots of indica were significantly higher than those on the roots of japonica (P < 0.05) (Figure 1). The complexed forms of Cu and Cd were also higher in indica but the differences between the two types of rice were not significant except for complexed Cu at 100 μM (Figure 1). The differences between the two types of rice with respect to the precipitated forms of Cu and Cd were not significant.

The zeta potential of roots of 40-day-old plants of different rice varieties is shown in Figure 2. The means of zeta potential in Figure 2 were calculated from six individual indica varieties or six individual japonica varieties (Supplementary Table S1). The zeta potential became more negative as the pH of the nutrient solution increased because of the increasing dissociation of functional groups on the roots, as was also observed earlier (Li et al., 2015; Liu et al., 2016), and was more negative in indica than in japonica (Figure 2): the difference was significant (P < 0.05), suggesting that root surfaces of indica rice carry a greater negative charge and thus have greater electrostatic attraction for metal cations.

The functional groups on rice roots are the major sources of the negative charge and binding sites for heavy metal cations. The ATR-FTIR spectra for the roots of indica and japonica varieties are shown in Figure 3. The absorption bands at 1635, 1543, and 1247 cm^-1 were attributed to amide I (antisymmetric stretching vibrations of carboxyl), amide II (N–H bending vibrations), and amide III (C–N stretching and N–H bending vibrations) (Cai et al., 2014; Wang et al., 2015). The peak near 1417 cm^-1 was assigned to symmetric -COO- stretching (Wang and Nielsen, 1992); the peaks at 1543, 1370, and 1318 cm^-1 were assigned to the CH₂ stretch of cellulose (Cai et al., 2014; Lv et al., 2016); and the 897 cm^-1 peak was assigned to the β-anomeric bond of cellulose (Kačuráková et al., 2000; Zhong et al., 2011). The peaks at 1150 and 1031 cm^-1 were assigned to C–O–C stretching or the skeletal vibration of lipids or cellulose and C–OH bending vibrations in carbohydrates, respectively (Kačuráková et al., 2000; Cai et al., 2014). No obvious differences in the locations of the absorption peaks were observed between the spectra of indica and japonica. However, the intensity of absorption peaks in the ATR-FTIR spectra of indica was higher than that of japonica (Figure 3). These results indicate that although both indica and japonica have the same surface functional groups on their roots, the groups are present in larger numbers on the roots of indica, which explains the greater negative charge on its roots.

Rice varieties of the japonica group tolerate Al toxicity better than those of the indica group do (Ma et al., 2002; Watanabe and Okada, 2005; Yang et al., 2008; Famoso et al., 2010; Zhao et al., 2013). The varieties sensitive to Al carry a greater negative charge on their roots than those tolerant to Al (Yermiyahu et al., 1997; Wang et al., 2015), which is one of reasons for the lower tolerance of the sensitive varieties (Wang et al., 2015; Liu et al., 2016). In the present study, roots of six indica varieties were observed to carry a greater negative charge than that carried by roots of the six japonica varieties (Figure 2). When the roots were exposed to Cu or Cd, the exchangeable forms of adsorbed Cu and Cd on the roots of indica were significantly greater than those on the roots of japonica (Figure 1). The exchangeable forms of Cu and Cd were adsorbed on the negatively charged roots through electrostatic attraction between the heavy metal cations and the root surface. The greater negative charge on the roots of indica varieties adsorbed greater exchangeable Cu and Cd. The cations of Cu and Cd adsorbed electrostatically on rice roots were located in the diffuse layers of the electric double layers on the negatively charged surfaces of the roots. These cations are highly reactive and easily absorbed by plants. Therefore, the greater electrostatic adsorption of Cu and Cd on the roots of indica rice caused by the greater negative charge on the roots may be one of reasons for the higher uptake and accumulation of Cu and Cd by the indica group. Liu et al. (2010) discovered that Cd concentrations in root tissues and shoots of Shanyou 63 (indica) were higher than that of Wuyunjing7 (japonica). Yoshihara et al. (2010) also found that the Cu concentration in roots was higher in indica than in japonica when the rice was exposed to the 10 μM Cu.

Of its three binding forms, Cd existed mainly as the exchangeable form on rice roots; the proportions of the complexed and the precipitated forms of Cd were much lower (Figure 1). On the other hand, Cu existed mainly in exchangeable and complexed forms; the proportion of precipitated Cu was much lower (Figure 1). Therefore, the surface charge on rice roots plays an important role in the adsorption of both heavy metal cations, especially that of Cd.

Correlation analysis was performed between the zeta potential of the roots of each of the twelve rice cultivars tested in this study and amount of Cu and Cd adsorbed on the roots of these rice cultivars. Based on Pearson’s correlation analysis, there was a significant correlation between zeta potential of different rice roots at pH5.5 and amount of Cu and Cd adsorbed on the roots of these rice cultivars (Figure 4), which confirmed that surface charge was one of important factors determining the adsorption of Cu and Cd on the roots of different cultivars of rice. Compared with the japonica varieties, the indica varieties carried greater negative charge on their root surfaces, which may be one of the reasons why the indica varieties were more sensitive to Cu and Cd and adsorbed more Cu and Cd on their root surfaces than japonica varieties.

The functional groups on rice roots are major sources of the negative charge and of the binding sites for heavy metal cations. Plant roots are negatively charged owing to the deprotonation and protonation of the surface functional groups (-COOH, -OH, -NH₂, and -H₂PO₄) on the cell walls and cell membranes of roots (Kinraide et al., 1992; Meychik et al., 2005; Wu and Hendershot, 2008). The intensity of absorption peaks in the ATR-FTIR spectra of the roots of indica was greater than that of the roots of japonica (Figure 3), suggesting that the functional groups were present in larger numbers on the roots of indica and were responsible for the greater negative charge and greater exchangeable Cu and Cd on its roots.

Wave number separation between the absorption peaks of antisymmetric and symmetric stretching vibrations has been used for determining the binding strength of carboxyl-containing organometallic complexes (Nakamoto, 1970). The spectra of the roots of different rice varieties with adsorbed Cu and Cd are shown in Figures 5, 6. It is evident that adsorption of Cu and Cd changes the absorption peaks for antisymmetric and symmetric stretching vibrations of carboxyl. For example, the absorption peak of carboxyl on the rice roots of YL1259 at 1635 cm^-1 was shifted to 1640 and 1646 cm^-1 after adsorption of Cu and Cd on the rice roots (Figure 7). So were the directions of the shifts for the two types of stretching: the peak for the antisymmetric stretching vibration shifted to higher wave numbers as rice roots were saturated with Cu or Cd, whereas the peak for the symmetric stretching vibration shifted to lower wave numbers.

The average wave number separation between the peaks of antisymmetric and symmetric stretching vibrations of carboxyl was calculated from the spectra in Figure 5 (Cu) and Figure 6 (Cd), and the results are shown in Figure 8. The data of the separations between the peaks of antisymmetric and symmetric stretching vibrations of carboxyl for six individual indica or japonica varieties were presented in Supplementary Table S2. After the roots were exposed to Cu or Cd, the wave number separation of indica between the peaks of antisymmetric and symmetric stretching vibrations was significantly greater than that of japonica (P < 0.05), suggesting that the strength of binding of Cu and Cd on roots of indica rice was greater than that of japonica rice. Therefore, the preference for Cu and Cd of the roots of indica was stronger than that of the roots of japonica. The results in Figure 8 also indicate that the wave number separation of roots that had adsorbed Cu was greater than that of the roots that had adsorbed Cd, consistent with the greater amount of complexed Cu on rice roots than that of complexed Cd (Figure 1).

It is evident from the results shown in Figure 2, 3 that the roots of the two types differ in their surface chemical characteristics. Wave number separation between the absorption peaks of antisymmetric and symmetric stretching vibrations has been used for determining the binding strength of carboxyl-containing organometallic complexes (Nakamoto, 1970). Such a wave number separation of the two carboxyl bands can be explained as the result of an increasingly covalent metal-carboxyl bonding: the stronger the covalent bonding, the more asymmetric the structure of the carboxyl group. Any increase in the asymmetry of the carboxyl structure is reflected in greater separation of the two carboxyl vibrational bands (Nakamoto et al., 1961). Based on the above explanation, the bonding strength of Cu and Cd with the carboxyl groups on rice roots can be directly evaluated by comparing the wave number separations between the bands of antisymmetric and symmetric stretching vibrations. Indica varieties exhibited stronger interactions between their roots and Cu and Cd than japonica varieties did, as indicated by the greater wave number separation (Figure 8). This is to be expected, because indica varieties also carried a greater negative charge (Figure 2) and more adsorption sites for heavy metal cations (Figure 3) on their roots. Wang and Nielsen (1992) investigated the binding strength of Mn(II) to the cell walls of tobacco roots using wave number separation and found that the Mn-sensitive variety KY 14 exhibited a stronger interaction between the cell wall and the metal cation than the Mn-tolerant variety T.I. 1112 did, an observation consistent with the results reported in the present study.

The proportion of complexed Cu on rice roots was much greater than that of complexed Cd on the roots of the same type of rice (Figure 1). Soil organic matter and crop straw biochars behave similarly (Zhong et al., 2010; Xu and Zhao, 2013), because of the greater ability of organic functional groups to complex with Cu than with Cd (McBride et al., 1997). Nishizono et al. (1987) also found that the affinity of the cell wall of plant roots for Cu was markedly greater than that for Cd and Zn. These interpretations of the differences between the abilities of rice roots to form complexes with Cu and Cd are also supported by the results of wave number separations presented in Figure 8.