4 DISCUSSION
Although biuret toxicity in crops is a well-known issue, little is known about the physiology underlying biuret injury. Here, we analyzed for the first time biuret uptake in rice plants quantitatively using15N-labelled biuret and revealed that a considerable amount of biuret was taken up by wild-type rice (Table 1). As wild-type rice plants did not show any biuret decomposing activity (Figure 3), biuret-derived 15N is considered as biuret in plants. Therefore, the shoot biuret concentration is approximately equal to 0.4 mmol L-1 when it is expressed on the basis of the tissue water content. The concentration of biuret was higher in shoots than in roots, which indicates that the amount of biuret retained in roots was small, and that biuret was accumulated in shoots through the transpiration stream. As biuret is a small polar molecule without lipophilic parts, cellular membranes may be slightly permeable to biuret. The rate of biuret uptake was calculated from the15N content in whole seedlings, and was found to be equivalent to 0.5 µmol g-1 root dw h-1. For comparisons, the rate of urea influx into roots, which is largely mediated by channels and a high-affinity transporter, were about 20 µmol g-1 root dw h-1 in Arabidopsis thaliana (Kojima, Bhner, Gassert, Yuan & von Wirén, 2007) and about 6 µmol g-1root dw h-1 in rice (Wang et al., 2012), when 0.3 mmol L-1 urea was supplied as a sole N source. The observed uptake rate of biuret was one to two orders of magnitudes lower than that of urea. Similarly, the permeability of biuret could not be detected at 10ºC in mouse erythrocytes that were permeable to urea (Zhao, Sonawane, Levin & Yang, 2007). Biuret could possibly move across membranes via simple diffusion. To evaluate biuret accumulation over a prolonged period, we need to develop a method to detect biuret directly. We are currently modifying HPLC methods, to separate biuret from other UV-absorbing compounds in plants.
The overexpression of bacterial biuret hydrolase conferred biuret tolerance to rice plants (Figure 4, Table 2). Conversely, biuret was seemingly not metabolized, or very slowly metabolized in wild-type rice plants. This is consistent with our enzyme assay results, obtained using leaf crude extracts (Figure 3), and with the previous report on biuret in orange leaves, in which biuret was detected by the eight months after foliar application by a qualitative analysis (Impey & Jones, 1960). The lack of an efficient decomposition pathway is probably responsible for biuret accumulation and toxicity in rice plants. Besides, biuret tolerance conferred by the biuret hydrolase suggested that an injury in rice plants occurred because of the direct effects of biuret within plants, but not from the indirect effects of biuret outside roots.
Additionally, our results on the biuret injury in wild-type rice plants gave some indications of mechanisms underlying biuret toxicity. In wild-type rice seedlings, a biuret concentration of 0.1 mmol L-1 and above in the culture solution caused a significant reduction in the growth (Figure 1). It was roughly consistent with the toxic concentration of biuret reported for hydroponically grown naked barley (Funabiki Ogata & Sakamoto, 1956) and pot cultured young citrus and avocado plants (Haas & Brusca, 1954). The rather high dose, together with the significant accumulation of biuret in rice shoots, suggests that biuret is moderately toxic and that biuret might have a weak affinity with its target. The occurrence of leaf chlorosis was observed, along with growth inhibition, in biuret-injured rice seedlings (Figure 1). The colorless appearance of elongating young leaves indicates that chloroplast development was impaired by excessive biuret levels. Closely similar patterns of chlorosis were often observed in rice seedlings exposed to cold stress (Yoshida, Kanno, Sato & Kameya, 1996). It has been shown that cold stress especially impairs the establishment of the plastid genetic system, during chloroplast development in rice seedlings (Kusumi et al., 2011). Although the mechanism by which it occurs is yet to be verified, biuret in leaves might trigger similar downstream cellular responses. On the other hand, we have found that biuret causes growth inhibition even in heterotrophic suspension cells of rice (unpublished data). Hence, biuret probably inhibits other basic metabolic processes as well. To have a better understanding of the mechanism of biuret injury, we are planning to analyze changes in transcript and metabolite levels in rice cells under biuret toxicity.
The findings reported here clearly demonstrate that it is possible to confer biuret detoxification ability on rice plants by introducing the microbial biuret hydrolase (Figure 4, Table 2). Moreover, rice plants overexpressing biuret hydrolase utilize ammonium-N produced by the hydrolysis of biuret in plant cells as an additional N source (Table 3). In soil, the decomposition rate, or the mineralization rate, of biuret is slower than that of urea (Ogata & Funabiki, 1956; Sahrawat, 1981). Taken together, when biuret is applied as a N fertilizer to the transgenic rice lines that were generated here, the fertilizer use efficiency would possibly be improved compared with that of urea fertilization. We are currently working on soil-culture experiments to evaluate the effect of biuret as a N fertilizer.