Abstract

   Biuret, a common impurity in urea fertilizers, is toxic to plants, but
   little is known about the physiological mechanisms underlying its
   toxicity. Here, we analyzed biuret toxicity in rice (Oryza sativa)
   plants. We carried out uptake experiments using ^15N‐labelled biuret
   and demonstrated that biuret could reach sub millimolar concentrations
   in rice plants. We also demonstrated that the hydrolysis of biuret in
   plant cells could confer biuret tolerance to rice plants. This occurred
   because transgenic rice plants that overexpressed an exogenous biuret
   hydrolase cloned from a soil bacterium gained improved tolerance to
   biuret toxicity. Our results indicate that biuret toxicity is not an
   indirect toxicity caused by the presence of biuret outside the roots,
   and that biuret is not quickly metabolized in wild‐type rice plants.
   Additionally, it was suggested that biuret was used as an additional
   nitrogen source in transgenic rice plants, because biuret
   hydrolase‐overexpressing rice plants accumulated more biuret‐derived N,
   as compared to wild‐type rice.

   Keywords: biuret toxicity, fertilizer, nitrogen, rice, urea

1. INTRODUCTION

   Urea, currently the most widely used nitrogen (N) fertilizer worldwide
   (IFASTAT, [32]https://www.ifastat.org/databases), might contain biuret
   [(CONH[2])[2]NH], as a common impurity. Biuret is formed by the thermal
   condensation of urea. It has been known since the 1950s that excessive
   amounts of biuret in urea fertilizers cause injury in crops
   (Jones, [33]1954; Sanford et al., [34]1954). A wide range of crops can
   be potentially affected by biuret toxicity, which often manifests as
   leaf chlorosis and stunted growth, especially in the young seedling
   stage (Mikkelsen, [35]1990). Earlier studies indicated that biuret
   inhibited protein synthesis in Xanthium pensylvanicum leaves (Webster
   et al., [36]1957) and wheat (Triticum aestivum) germplasms (Ogata &
   Yamamoto, [37]1959). The protein content, however, did not so much
   decrease in biuret‐injured orange (Citrus sinensis) leaves (Impey &
   Jones, [38]1960). It remains uncertain whether biuret has a direct
   effect on the protein synthetic machinery. Additionally,
   ultrastructural analyses showed that changes in chloroplast structure
   in biuret‐injured leaves were similar to those in senescent leaves in
   grapefruit (Citrus paradise) and orange plants (Achor &
   Albrigo, [39]2005). Moreover biuret seems to remain unmetabolized in
   plants, because it was still detected in orange leaves, eight months
   after foliar spraying was performed (Impey & Jones, [40]1960). The
   exact mechanism underlying biuret toxicity in plants, however, is still
   far from being understood.

   To avoid this hazard, the biuret content in fertilizers is regulated;
   for example, the upper limit of biuret‐N in urea fertilizer is set at
   2% of the total N content in Japanese law. Currently, biuret injury has
   become less frequent in farmers’ fields, owing to advances in the
   technology used for manufacturing urea fertilizers. One method for the
   fertilization of rice (Oryza sativa) crops involves a single basal
   application of polymer‐coated urea into seedling trays, which improves
   N use efficiency and labor efficiency. The extremely high density of
   coated urea fertilizer adjacent to the roots resulted in biuret
   toxicity, even though fertilizers that met the official standards were
   used (Tanahashi et al., [41]2003). This illustrates that the risk of
   biuret toxicity in crops remains latent.

   Certain soil bacteria decompose biuret (Aukema et al., [42]2020;
   Cameron et al., [43]2011; Esquirol et al., [44]2018, [45]2020; Jensen &
   Schrøder, [46]1965; Martinez et al., [47]2001; Robinson
   et al., [48]2018). Biuret in soil occurs as an impurity of urea
   fertilizer. It is also produced by the degradation of cyanuric acid, an
   intermediate metabolite of s‐triazine compounds such as melamine and
   atrazine (Aukema et al., [49]2020). The evolution of biuret
   catabolizing enzymes in bacteria is considered to be associated with
   the use of urea fertilizer and s‐triazine herbicide (Esquirol
   et al., [50]2020; Robinson et al., [51]2018). One molecule of biuret is
   converted into three ammonium and two bicarbonate ions via the
   biodegradation pathway. The first step of biuret degradation is the
   hydrolysis of biuret into ammonium and allophanate, which is catalyzed
   by biuret hydrolase (Cameron et al., [52]2011). Then, allophanate could
   either undergo spontaneous decarboxylation to form urea under neutral
   and acidic conditions, or be hydrolyzed further by allophanate
   hydrolase into ammonium and bicarbonate (Cheng et al., [53]2005).

   Homologs of biuret hydrolase were detected in a broad range of
   microorganisms but remained undetected in animals and land plants
   (Robinson et al., [54]2018). Although it is not known yet how much
   biuret plants would take up and accumulate, introduction of biuret
   hydrolase from soil bacteria might contribute to biuret detoxification
   in plant cells. Furthermore, biuret could act as a slow‐release N
   fertilizer as researchers have suggested (Esquirol et al., [55]2020;
   Sahrawat, [56]1981) and as a weed controller (Figure [57]S1) if we
   could confer the biuret‐detoxifying ability to crop plants.

   Here, we investigated biuret injury in rice plants. We first evaluated
   biuret uptake in rice plants quantitatively using ^15N‐labelled biuret.
   Furthermore, we generated transgenic rice plants that overexpressed
   bacterial biuret hydrolase and examined their biuret tolerance.
   Additionally, we performed a microarray‐based transcriptome analysis
   using suspension rice cells to investigate the general effects of
   biuret on rice cells.

2. MATERIALS AND METHODS

2.1. Plant materials and growth conditions

   Seeds of Nipponbare, a japonica rice (Oryza sativa) cultivar, were
   purchased from Nouken Inc. (Kyoto, Japan). Transgenic rice lines
   overexpressing bacterial biuret hydrolase, which were generated from
   Nipponbare in this study, were self‐pollinated to obtain T[1]
   progenies. T[1] and T[2] seeds were used for experiments.

   Rice plants were grown under hydroponic conditions in a growth chamber
   (NS‐280 FHW; Takayama Seisakusyo, Kyoto, Japan) under a temperature,
   photo period, and light intensity of 30ºC, 12 hr, and
   350 mmol/m^2 s^−1, respectively. Rice seeds were soaked in distilled
   water supplemented with fungicide (Trifumin; Nippon Soda Co., Ltd.,
   Tokyo, Japan) for three days at 30ºC. Ten or twelve of the imbibed
   seeds were sown on a nylon‐mesh (18 mesh, 24 × 36 mm) supported by a
   plastic frame floating on the culture solution. The culture solution
   contained 1 mmol/L (NH[4])[2]SO[4], 0.25 mmol/L KH[2]PO[4], 0.5 mmol/L
   KCl, 0.5 mmol/L CaCl[2], 0.5 mmol/L MgCl[2], 0.09 mmol/L
   ethylenediamine‐N,N,N',N'‐tetra acetic acid, iron(III), sodium salt,
   trihydrate (FeNa‐EDTA) and Arnon's micronutrient (cited by
   Hewitt, [58]1966). The culture solutions were not aerated, and the
   solutions were renewed once a week. Seeds that did not germinate were
   removed from the mesh at an appropriate time. Biuret was included in
   the culture solution whenever necessary.

   Rice Oc cell suspension culture line (Baba et al., [59]1986) was
   provided by the RIKEN BRC, participating in the National BioResource
   Project of the MEXT/AMED, Japan. The suspension cells were maintained
   shaking in the dark at 25ºC in 80 ml of Murashige and Skoog medium with
   3% sucrose and 1 mg/L 2,4‐dichlorophenoxyacetic acid, pH 5.7 in a
   300‐mL conical flask and subcultured weekly. When necessary,
   filter‐sterilized biuret solution was added to the autoclaved medium.
   Suspension cells were exposed to biuret toxicity by subculturing 2 ml
   of seven‐day‐old cell suspension into 80 ml of the medium supplemented
   with biuret at desired concentrations. At harvest, cells in a flask
   were collected by suction filtration.

2.2. Measurement of biuret in culture solutions

   Aliquots (0.94 ml) of the culture solution were mixed with 10 µl of
   0.5 mol/L potassium phosphate buffer (pH 6.8) and 50 µl of methanol,
   and filtered using the Cosmospin Filter G (0.2 µm; Nacalai tesque,
   Kyoto, Japan). A 20‐µl aliquot of the sample was injected into an HPLC
   system (LC‐10AS, Shimadzu, Kyoto, Japan) equipped with a COSMOSIL
   5C18‐PAQ column (5 µm, 4.6 mm I.D. × 250 mm; Nacalai tesque, Kyoto,
   Japan). The eluent consisted of 5 mmol/L potassium phosphate buffer (pH
   6.8) and 5% (v/v) methanol, and the flow rate was 0.2 ml/min. The
   absorbance of the eluate was monitored at 210 nm using a UV detector
   (SPD‐10A, Shimadzu, Kyoto, Japan).

2.3. Determination of biuret uptake in rice plants

   ^15N‐labelled biuret (biuret‐^15N[3], ≥98 atom% ^15N) was purchased
   from Sigma‐Aldrich Co (St. Louis, MO, USA). Rice seedlings were raised
   in the absence of biuret for 17 days. Then, seedlings with uniform
   sizes were picked from the mesh, and two plants were transferred to a
   50‐mL glass vial with 40 ml of culture solution, one day before
   performing the uptake experiments, to eliminate the possible effects of
   damaged roots. When we used transgenic lines, the presence of the
   transgene was confirmed beforehand by PCR, using the DNA extracted from
   leaf blades of the second leaf. Uptake experiments were initiated by
   the addition of an aliquot of ^15N‐labeled biuret solution into the
   vial, to achieve a solution with a final biuret concentration of
   0.3 mmol/L. Plants were allowed to take up ^15N‐labeled biuret for
   48 hr. Control plants, to which biuret was not applied, were also grown
   similarly.

   At harvest, roots were rinsed twice in 100 ml of distilled water (3 min
   each). Plants were separated into shoots and roots, dried in an oven at
   70ºC for two days, and ground into a powder using a ball mill. The N
   concentration in plant samples was determined using a CN analyzer
   (SUMIGRAPH NC‐22F, Sumika Chemical Analysis Service, Osaka, Japan). The
   atom % of ^15N was determined under contract (Shoko Science Co Ltd,
   Kanagawa, Japan). Biuret‐derived N content in plant parts were
   calculated from the ^15N content in samples.

2.4. Generation of transgenic rice plants overexpressing biuret hydrolase

   Transgenic Nipponbare plants that over‐expressed bacterial biuret
   hydrolase under the control of the cauliflower mosaic virus 35S
   promoter were generated. A biuret decomposing soil bacterium, Rhizobium
   sp. KaB01, which was isolated in this study, was used as a biuret
   hydrolase donor (see Supplemental methods). Insert DNA was amplified
   from the genomic DNA of KaB01 via PCR, using Prime Star polymerase
   (Takara Bio, Shiga, Japan) and primers 5′‐CACCATGAAGACACTTTCCAGCGC‐3′
   and 5′‐TGGCAAATGCCTCTCAAGG‐3′. Subcloned PCR products in the vector
   pENTR/D‐TOPO (Life Technologies, Carlsbad, CA) were then transferred to
   a binary vector pGWB502omega (Nakagawa et al., [60]2007), via the LR
   reaction. The transformation of rice was mediated by an agrobacterium,
   as described by Toki et al. ([61]2006), using the Agrobacterium
   tumefaciens strain EHA105. The presence of the transgene in regenerated
   T[0] plants was confirmed by PCR, using the Blend Taq polymerase
   (Toyobo, Osaka, Japan) and primers 5′‐ATGAAGACACTTTCCAGCGC‐3′ and
   5′‐TGGCAAATGCCTCTCAAGG‐3′.

2.5. Transgene expression analysis

   Total RNA was extracted from the youngest leaf blade of regenerated
   T[0]plants at their vegetative growth stage, using the Plant Total RNA
   Extraction Miniprep System (Viogene, Taipei, Taiwan). First‐strand cDNA
   was synthesized from total RNA using an oligo dT primer and ReverTra
   Ace polymerase (Toyobo, Osaka, Japan). Quantitative real‐time RT‐PCR
   was performed in duplicate with the TP850 thermal cycler dice real time
   system single (Takara Bio, Shiga, Japan), using the THUNDERBIRD^® SYBR
   qPCR Mix (Toyobo, Osaka, Japan) and primers
   5′‐AGCCGATCAAAAAGGTGCTGTC‐3′ and 5′‐AATGATATCCCAGCCAGGTTCTCC‐3′. The
   relative expression level was calculated as a ratio to a geometric mean
   of the expression of ubiquitin and actin. The sequences of primers were
   5′‐AGAAGGAGTCCACCCTCCACC‐3′ and 5′‐GCATCCAGCACAGTAAAACACG‐3′ for
   ubiquitin and 5′‐ATCCTTGTATGCTAGCGGTCGA‐3′ and
   5′‐ATCCAACCGGAGGATAGCATG‐3′ for actin.

2.6. Assay for biuret hydrolase activity

   Crude plant extracts were prepared as described below. Shoots of
   individual 19‐day‐old wild‐type Nipponbare and transgenic B3‐9‐1 plants
   and 16‐day‐old B2‐3‐3‐3 plants were weighed, and ground into powders
   under liquid nitrogen using a mortar and pestle. Then, the powdered
   tissue was homogenized in ten volumes of 25 mmol/L 3‐(N‐morpholino)
   propanesulfonic acid (MOPS) buffer (pH 8.0). After centrifugation, the
   supernatant was used for the enzyme assay. Extractions were performed
   in duplicate for transgenic lines, and in triplicate for wild‐type
   plants.

   The biuret hydrolase assay was carried out as described by Martinez
   et al. ([62]2001). The assay solution contained 50 mmol/L sodium
   phosphate buffer (pH 8), 3 mmol/L biuret, and crude cell extracts. The
   assay mix was incubated at 30ºC, and the reaction was stopped by the
   addition of 0.5 mol/L H[2]SO[4]. The amount of ammonia released from
   biuret was colorimetrically determined by indophenol blue method
   (Weatherburn, [63]1967). Protein concentrations were determined by the
   Bradford method, using Protein Assay CBB Solution (Nacalai tesque,
   Kyoto, Japan). Measurements were performed in duplicate.

2.7. Evaluation of biuret tolerance in transgenic rice plants over expressing
biuret hydrolase

   Seeds from each line were sown onto two nylon‐mesh floats on the
   culture solution in the absence of biuret and grown for two days, to
   achieve uniform germination. Then, one of the two floats were
   transferred into a new 2‐L container that did not contain biuret and
   the other was transferred into another 2‐L container containing
   0.3 mmol/L biuret. Plants were harvested 7 days after the onset of the
   biuret treatment. After determining plant heights, leaf blades from the
   third leaf were used to determine the chlorophyll content, and leaf
   blades from the second leaf were used for DNA extraction. Chlorophyll
   was extracted from tissues using 80% aqueous acetone buffered with
   2.5 mmol/L sodium phosphate buffer (pH 7.8), and its levels were
   determined according to the method described by Porra
   et al. ([64]1989). The presence of the transgene was confirmed by the
   PCR method, using Blend Taq polymerase (Toyobo, Osaka, Japan) and the
   primers 5′‐ATGAAGACACTTTCCAGCGC‐3′ and 5′‐TGGCAAATGCCTCTCAAGG‐3′, and
   the data for plants not carrying the transgene were omitted from the
   analysis.

2.8. DNA microarray‐based transcriptome analysis

   Total RNA was extracted from 3‐ and 5‐day‐old rice suspension cells
   grown in the media supplemented with 0 or 0.3 mmol/L biuret using
   RNeasy Plant Mini Kit (Qiagen, Hilden, Germany). Two‐color microarray
   analysis with two biological replicates was performed using Agilent
   Rice Oligo DNA Microarray 4x44K slide (Agilent, Santa Clara, CA, USA)
   according to the manufacturer's instructions to estimate the ratio of
   transcript abundance between the treatments at each culture period.
   Cyanine 3 or cyanine 5 labeled antisense cRNA was synthesized from the
   total RNA sample using Agilent Quick Amp Labeling Kit and hybridized to
   the microarray slide at 65°C for 17 hr using Agilent Gene Expression
   Hybridization Kit. The slide was scanned using Agilent DNA microarray
   scanner. Data were extracted with Agilent Feature Extraction software.
   Differentially expressed genes (DEGs) were detected using an R software
   package limma (Ritche et al., [65]2015). Genes were considered to be
   differentially expressed if the p‐value was less than 0.005.

2.9. Functional classification of differentially expressed genes

   Rice loci corresponding to the probes were confirmed using blastn
   searches against IRGSP‐1.0 genes_2020‐06‐03 that was downloaded from
   the Rice Annotation Project (RAP) database
   ([66]https://rapdb.dna.affrc.go.jp). Probes whose sequence was not
   mapped on rice locus or mapped on multiple loci were omitted from the
   subsequent analysis. When there were several probes corresponding to a
   locus, and when there was a marked inconsistency in the expression
   data, such a gene was omitted from DEGs' lists.

   Gene ontology (GO) terms associated with the rice locus were obtained
   through the PANTHER Classification System (Mi et al., [67]2019) using
   the RAP gene identifiers as queries. Kyoto Encyclopedia of Genes and
   Genomes (KEGG) pathway identifiers for rice genes (KEGG organism code:
   dosa) were obtained from the KEGG database (Kanehisa & Goto, [68]2000;
   [69]https://www.genome.jp/kegg/). Overrepresentation of these
   annotations in DEGs was tested by fisher's exact test, and
   Benjamini‐Hochberg adjusted p‐values < 0.05 was set as a threshold.
   Transcription factor and transcriptional regulator genes were
   identified according to the classification in the Plant Transcription
   Factor Database version 3.0 (Pérez‐Rodríguez et al., [70]2010;
   [71]http://plntfdb.bio.uni‐potsdam.de/v3.0/).

2.10. Accession

   The sequenced data of Rhizobium sp. KaB01 biuret hydrolase has been
   submitted to the DDBJ/EMBL/GenBank databases under the accession number
   [72]LC532383.

3. RESULTS

3.1. Biuret injury in hydroponically grown rice seedlings

   We first investigated the biuret sensitivity of wild‐type rice plants.
   When wild‐type Nipponbare seeds were sown in culture solutions
   supplemented with varying levels of biuret (0, 0.1, 0.3, and 1.0 mmol/L
   biuret), germination was not inhibited, regardless of the biuret
   levels. The plant height of 7‐day‐old seedlings decreased with
   increasing concentrations of biuret (Figure [73]1a). Chlorosis was
   observed in some plants exposed to 0.1 mmol/L biuret and all plants
   exposed to 0.3 mmol/L biuret. It was most markedly observed in leaf
   blades of the emerging third leaf. In plants exposed to 1.0 mmol/L
   biuret, chlorosis was less prominent, probably because of the severe
   reduction in the growth (Figure [74]1b).

FIGURE 1.

   FIGURE 1
   [75]Open in a new tab

   Symptoms of biuret toxicity in rice plants. Rice seedlings were
   hydroponically grown in a culture solution supplemented with 0, 0.1,
   0.3, and 1.0 mmol/L biuret. (a) Plant heights of 7‐day‐old seedlings.
   The data represent means ± SD (n = 22–24). Different alphabets indicate
   significant differences between groups (p < 0.05, Tukey test). (b)
   Representative images of 7‐day‐old seedlings

3.2. Biuret uptake by rice seedlings

   Prior to the evaluation of biuret uptake by rice plants, we first
   examined biuret decomposition in culture solutions. Containers filled
   with 1 L of culture solution supplemented with 0.3 or 1.0 mmol/L biuret
   were placed in a growth chamber, and aliquots of culture solutions were
   collected from the containers at 0, 12, 24, 48, and 72 hr. The biuret
   concentrations were allowed to remain constant for up to 72 hr. This
   result indicated that biuret was not hydrolyzed in the culture solution
   within this period.

   Then, we carried out 48‐hr uptake experiments using ^15N‐labelled
   biuret. Biuret can be measured as a colored chelation complex with
   cupric ions or by HPLC analysis combined with UV‐detection, which we
   used for the determination of biuret concentrations in the culture
   solution. However, the sensitivity of the colorimetry process was too
   low to determine biuret concentrations in plant samples. Additionally,
   the peak of biuret could not be separated from UV‐absorbing metabolites
   of rice plants in our system.

   When 19‐day‐old Nipponbare seedlings were allowed to take up
   ^15N‐labeled biuret for 48 hr, biuret‐derived ^15N was detected both in
   the shoots and roots of the seedlings (Table [76]1). This result
   indicated that biuret was taken up by rice roots and possibly
   translocated into rice shoots. The biuret‐derived ^15N concentrations
   in the shoots and roots were equal to 4.5 and 1.9 µmol biuret g^−1 dw,
   respectively. Then, based on the amount of biuret in whole seedlings,
   the uptake rate of biuret with an external supply of 0.3 mmol/L was
   calculated at 0.5 µmol/g root dw h^−1.

Table 1.

   ^15N‐Biuret uptake in rice plants. Nineteen‐day‐old Nipponbare
   seedlings were exposed to 0.3 mmol/L ^15N‐biuret for 48 hr. Two
   seedlings were incubated in each of three 40‐mL vials, and plants in a
   vial were analyzed as one sample. Values are expressed as means ± SD
   (n = 3)
          Dry weight   Total‐N     Biuret‐derived N
          (mg vial^−1) (mmol/g DW) (µmol/g DW) (µmol vial^−1)
   Shoots  124 ± 3.88  2.43 ± 0.09 13.5 ± 1.02 1.67 ± 0.18
   Roots  23.8 ± 1.06  1.75 ± 0.01 5.83 ± 0.55 0.13 ± 0.01
   [77]Open in a new tab

3.3. Transgene expression and biuret decomposing activity in transgenic rice
plants

   To examine the effect of biuret hydrolysis in plant cells on the biuret
   tolerance of rice plants, we generated transgenic rice plants
   overexpressing biuret hydrolase that were cloned from Rhizobium sp.
   KaB01. The inserted biuret hydrolase encoded a protein consisting of
   238 amino acid residues (Figure [78]S2), whose amino acid sequence
   showed 92% similarity to a known biuret hydrolase of R. leguminosarum
   bv. viciae 3841 ([79]WP_011654379.1; Cameron et al., [80]2011). In
   addition, purified recombinant maltose‐binding protein fusion proteins
   expressed in Escherichia coli showed the biuret decomposing activity
   (Figure [81]S3). Signal peptides were not detected via the SignalP‐5.0
   program (Almagro Armenteros et al., [82]2019), and the protein was
   predicted to be cytoplasmic localized, using the program PSORTb v.3.0
   (Yu et al., [83]2010).

   The b iuret hydrolase transgene was expressed in varying levels in
   regenerated T[0] rice leaves, while its expression was not detected in
   wild‐type Nipponbare (Figure [84]2). Self‐pollinated progenies of two
   independent transgenic lines, B2‐3‐3 and B3‐9‐1, which showed high
   expression levels, were used for further examination.

FIGURE 2.

   FIGURE 2
   [85]Open in a new tab

   Relative expression levels of biuret hydrolase in transgenic rice
   plants. The expression in the youngest leaf at the vegetative growth
   stage was determined via real‐time PCR. Ubiquitin and Actin1 were used
   as internal standards for normalization. The values are means of two
   technical replicates. Error bars indicate ranges of relative expression
   levels calculated from standard deviations of ∆Ct value. In names of
   transgenic lines, the combination of the first and second numbers
   indicates a callus derived from a single seed, and the third number
   indicates an individual plant regenerated from the callus

   Crude extracts prepared from 19‐day‐old seedlings of B2‐3‐3‐3 (T[2]) or
   16‐day‐old seedlings of B3‐9‐1 (T[1]) showed the biuret decomposing
   activity. In both lines, the amount of ammonia released from biuret
   increased linearly with time (Figure [86]3). The line B3‐9‐1 showed
   higher specific activity than B2‐3‐3‐3. The specific activities of
   extracts from the B3‐9‐1 and B2‐3‐3‐3 plants were 8.6 and
   0.67 nmol/min mg^−1 protein. The result was inconsistent with the
   similar transgene expression levels in these lines. It might be
   affected by the relationship between gene expression and protein
   expression. Wild‐type Nipponbare plants did not exhibit
   ammonia‐releasing activity (Figure [87]3). Extracts heated at 100ºC for
   5 min were inactive. In addition, extracts prepared from a null
   segregant of B3‐9‐1 did not exhibit ammonia‐releasing activity. These
   results indicated that the transgenic rice lines were conferred with
   biuret decomposing ability by the exogenous biuret hydrolase.

FIGURE 3.

   FIGURE 3
   [88]Open in a new tab

   Biuret‐decomposing activities of crude extracts of rice plants. Crude
   extracts were prepared from shoots of individual 19‐day‐old wild‐type
   Nipponbare (solid boxes, n = 3) and transgenic B3‐9‐1 (solid circles,
   n = 2) plants and 16‐day‐old B2‐3‐3‐3 (open circles, n = 2). Extracts
   were incubated at 30ºC with 50 mmol/L sodium phosphate buffer (pH 8)
   and 3 mmol/L biuret. The reaction was stopped by adding 0.5 mol/L
   H[2]SO[4]. The amount of ammonia released from biuret was determined
   using indophenol blue. The data represent means ± SD

3.4. Biuret tolerance in transgenic rice plants overexpressing biuret
hydrolase

   Rice plants overexpressing biuret hydrolase showed improved tolerance
   to biuret (Figure [89]4). The plant height of 9‐day‐old wild‐type rice
   seedlings was reduced or remained unchanged with the addition of
   0.3 mmol/L biuret. It was slightly increased in two transgenic lines,
   although the difference was not statically significant (Figure [90]5a).
   Similar results were also obtained for the chlorophyll content observed
   in leaf blades of the third leaf. The chlorophyll content in wild‐type
   plants decreased significantly in the presence of excessive biuret
   levels but slightly and not significantly increased in B3‐9‐1 and
   B2‐3‐3‐3 plants (Figure [91]5b). These results clearly showed that rice
   plants overexpressing biuret hydrolase gained tolerance to biuret
   toxicity through the decomposition of biuret in plants.

FIGURE 4.

   FIGURE 4
   [92]Open in a new tab

   Biuret tolerance of a transgenic rice line overexpressing exogenous
   biuret hydrolase. Biuret treatment (0 and 0.3 mmol/L) was started two
   days after the sowing process. Photos were taken 9 days after sowing.
   Wild‐type rice plants (left panel) and transgenic rice plants (line
   B2‐3‐3‐3, right panel) grown without and with biuret

FIGURE 5.

   FIGURE 5
   [93]Open in a new tab

   Plant height and chlorophyll content of wild‐type (WT) and biuret
   hydrolase‐overexpressing rice plants under the effect of biuret
   toxicity. (a) Plant height, and (b) chlorophyll content in third leaf.
   Biuret treatment (0.3 mmol/L) was started two days after sowing, and
   nine plants were harvested. Two transgenic lines, B2‐3‐3‐3 and B3‐9‐1,
   were grown in separate experiments. Light gray boxes indicate control
   plants, and dark gray boxes indicate biuret‐treated plants. Values are
   expressed means ± SD (n = 12–16). Asterisks on the top of panels
   indicate the significance of the interaction between the line and
   treatment (Two‐way ANOVA, *p < 0.05; ***p <0 .001). Different alphabets
   indicate significant differences among groups (p < 0.05, Tukey's test)

3.5. Biuret uptake in transgenic rice plants

   The slight increment in plant heights of biuret‐treated transgenic
   lines suggested that biuret was utilized as an additional N source.
   Therefore, we also determined biuret uptake in biuret
   hydrolase‐overexpressing lines. The concentration of biuret‐derived N
   was 2.6 and 5.6 times higher in two transgenic lines, B2‐3‐3‐11 (T[2])
   and B3‐9‐1‐5 (T[2]), as compared to that in wild‐type plants
   (Table [94]2). This suggested that biuret was decomposed in the roots
   and that ^15N derived from biuret was built into more mobile compounds,
   such as amino acids, and transported into shoots. Taken together,
   biuret hydrolase‐overexpressing plants seemingly used biuret as an
   extra N source.

Table 2.

   Biuret‐derived ^15N in shoots of biuret hydrolase‐overexpressing rice
   plants. Nineteen‐day‐old seedlings were exposed to
   0.3 mmol/L^15N‐biuret for 48 hr. Two seedlings were incubated in each
   of 40‐mL vials, Values are expressed means ± SD (WT: n = 2, B2‐3‐3‐11
   and B3‐9‐1‐5: n = 3)
   Line      Dry weight    Total‐N       Biuret‐derived N
             (mg vial^−1)  (mmol/g DW)   (µmol/g DW)   (µmol vial^−1)
   WT        109^b ± 11.2  2.30^a ± 0.13 16.4^a ± 1.76 1.80^a ± 0.38
   B2‐3‐3‐11 76.2^a ± 6.67 2.35^a ± 0.21 42.3^b ± 5.39 3.20^b ± 0.13
   B3‐9‐1‐5  77.6^a ± 5.79 2.42^a ± 0.12 91.4^c ± 1.98 7.09^c ± 0.59
   [95]Open in a new tab

   Different alphabets indicate significant differences between rice lines
   (p < 0.05, Tukey's test).

3.6. Transcriptome analysis of rice suspension cells under biuret toxicity

   Lastly, gene expression changes in response to biuret toxicity were
   examined in rice suspension cells. When rice suspension cells were
   subcultured into the culture medium supplemented with 0–1 mmol/L
   biuret, the fresh weight of 7‐day‐old cells decreased with increasing
   biuret concentration in the culture medium (Figure [96]6a). The
   reduction in growth was significant at biuret concentrations above
   0.3 mmol/L. Rice cells hardly proliferated under 1 mmol/L biuret
   toxicity. Figure [97]6b shows representative data for the changes in
   the fresh weights over time for the control and 0.3 mmol/L
   biuret‐treated cells. Cells kept growing until seven days after
   subculturing. The difference between the treatments in cell fresh
   weight was not significant until day 5 but was significant by day 7
   (Figure [98]6b).

FIGURE 6.

   FIGURE 6
   [99]Open in a new tab

   Microarray‐based transcriptome analysis in rice suspension cells under
   biuret toxicity. (a) Biuret toxicity in rice suspension cells. Cells
   were subcultured into a culture medium supplemented with 0, 0.1, 0.3,
   and 1.0 mmol/L biuret. Fresh cell weight per culture flask was measured
   seven days after subcloning. Values are expressed means ± SD (n = 3).
   Different alphabets indicate significant differences among groups (p <
   0.05, Tukey's test). (b) Changes in the fresh cell weight over time.
   Rice cells were subcultured into the culture medium supplemented with 0
   and 0.3 mmol/L biuret on day 0. The cells on day three and day 5 were
   used for subsequent microarray analyses. White circles indicate the
   control cells, and black circles indicate biuret‐treated cells. Values
   are expressed means ± SD (n = 2). An asterisk indicates a significant
   difference between the treatments (p < 0.05, t test). (c) Relationship
   between the normalized average gene expression levels and fold changes
   of gene expression of the 3‐day‐old biuret‐treated cells to the control
   cells. Each symbol denotes each microarray spot. Values are geometrical
   means of two arrays (n = 2). (d) Relationship between the normalized
   average gene expression levels and fold changes of gene expression of
   the 5‐day‐old biuret‐treated cells to the control cells. Each symbol
   denotes each microarray spot. Eleven points with high non‐significant
   fold‐change are omitted from the figure. Values are geometrical means
   of two arrays (n = 2). (e) Numbers of differentially expressed genes

   The cells on days 3 and 5 (Figure [100]6b) were used for the subsequent
   microarray‐based transcriptome analysis in which gene expression was
   compared between the biuret treatments at each culture period. The gene
   expression levels on 436 and 1,743 microarray spots were significantly
   different between the treatments in cells on days 3 and 5, respectively
   (Figure [101]6c,d). As genes, 246 (204 up‐ and 42 down‐regulated) and
   966 (736 up‐ and 230‐down‐regulated) loci of the total 22,794 loci were
   considered as differentially expressed genes (DEGs) for day 3 and day 5
   datasets (Figure [102]6e, Dataset S1 and DatasetS2). Forty‐seven of
   these genes were differentially expressed in both the culture period.

   Then, we performed GO enrichment analysis and KEGG pathway enrichment
   analysis to know the biological significance of the DEGs.
   One‐hundred‐forty‐five DEGs on day 3 and 704 DEGs on day 5 were
   annotated to one or more GO terms. Similarly, 23 DEGs on day 3 and 165
   DEGs on day 5 were mapped onto KEGG pathways. Any significant
   (adjusted‐p < 0.05) overrepresentation of GO terms or KEGG pathways was
   not detected in DEGs on day 3.

   Seventeen GO terms and three KEGG pathways were overrepresented in DEGs
   on day 5 (Figure [103]7, Figure [104]S4). The overrepresented terms
   indicated that metabolic reactions involved in cellular redox
   regulation were active in the biuret‐treated cells and thus suggested
   the accumulation of reactive oxygen species (ROS) in the cells. Terms
   such as “Oxidation‐reduction process” (GO: 0055114), “hydrogen peroxide
   catabolic process” (GO: 0042744), “cellular oxidant detoxification”
   (GO: 0098869), “oxidoreductase activity” (GO: 0016491), “heme binding”
   (GO: 002037), “peroxidase activity” (GO: 0004601), and “electron
   transfer activity” (GO: 0009055) were directly related to cellular
   redox conditions. Two other terms, “copper ion transport” (GO: 0006825)
   and “copper ion transmembrane transporter activity” (GO: 0005375), were
   also relevant to redox regulations, for copper is an essential
   redox‐active transition metal. As well‐known, biuret forms chelation
   complexes with copper ions in a strongly alkaline solution; however, it
   is uncertain whether biuret directly interacts with copper ions in rice
   cells.

FIGURE 7.

   FIGURE 7
   [105]Open in a new tab

   GO term and KEGG pathway enrichment analysis for differentially
   expressed genes in 5‐day‐old rice suspension cells under biuret
   toxicity. The overrepresentation was tested by Fisher's exact test, and
   terms with an adjusted p‐value below 0.05 are listed

   The enrichment analyses also indicated active adaptive responses those
   similar to responses to a broad range of abiotic and biotic stresses.
   “Phenylpropanoid biosynthesis” (dosa00940) pathway starts with
   phenylalanine and ends with phenolic compounds, including flavonoids
   and lignin, has been known to be activated under stresses (Dixon &
   Paiva, [106]1995). “Glutathione metabolism pathway” (dosa009480) and
   “glutathione transferase activity” (GO: 0004364) are essential for
   cellular redox control (Dixon et al., [107]2002). “Pectinesterase
   inhibitor activity” (GO: 0046910) and some genes categorized under
   “hydrolase activity, acting on glycosyl bonds” (GO: 0016798) relates
   stiffening or loosening of cell walls (Hong et al., [108]2010; Sharma
   et al., [109]2013). “Lipid transport” (GO: 0006869) is also involved in
   stress adaptation (Guo et al., [110]2013; Zhao et al., [111]2020).

   Besides, approximately 8% of DEGs consisted of transcription factor
   genes and other transcription regulator genes (Dataset S1 and Dataset
   S2). Only two genes, one MYB gene, Os08g0151300, and one TCP gene,
   Os02g0747400, were shared between the culture periods. Both of them
   were up‐regulated. Os08g0151300 (CEF1/OsMYB103L) was reported to
   regulate cellulose synthesis (Yang et al., [112]2014).

   Transcription factor genes in DEGs included genes known to be
   regulating stress responses besides genes known to be regulating growth
   and development. In DEGs on day 3, there were five stress‐responsive
   genes: two AP2‐EREBP genes Os09g0522000 and Os09g0522200 (OsDREB1B and
   OsDREB1A; Dubouzet et al., [113]2003), a bHLH gene Os04g0301500
   (OsbHLH6; Meng et al., [114]2020), a C3H gene Os05g0128200 (OsTZ5;
   Selvaraj et al., [115]2020), and a NAC gene Os08g0436700 (ONAC063;
   Yokotani et al., [116]2009). These genes could be induced by cold,
   drought, salt, or fungal infection. Additionally, we investigated GO
   terms associated with DEGs and started with response to or defense
   response to and found water deprivation, salt, cold, wounding, fungus,
   bacterium, and oomycetes after these words (Table [117]S1).

   Taken together, results of microarray analysis suggested that ROS
   accumulated in 5‐day‐old rice cells under 0.3 mmol/L biuret toxicity to
   cause oxidative stress. The molecules involved in response to biuret
   stress had some similarities with those for other environmental
   stresses. Furthermore, two overrepresented terms in the cell component
   category suggested metabolisms in the apoplast were active.

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 using
   ^15N‐labelled biuret and revealed that a considerable amount of biuret
   was taken up by wild‐type rice (Table [118]1). As wild‐type rice plants
   did not show any biuret decomposing activity (Figure [119]3),
   biuret‐derived ^15N is considered as biuret in plants. Therefore, the
   shoot biuret concentration is approximately equal to 0.4 mmol/L 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 the ^15N content in whole seedlings, and was
   found to be equivalent to 0.5 µmol/g 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 root dw h^−1 in
   Arabidopsis thaliana (Kojima et al., [120]2007) and about 6 µmol/g root
   dw h^−1 in rice (Wang et al., [121]2012), when 0.3 mmol/L 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 et al., [122]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 [123]4, Figure 5). 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 [124]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, [125]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
   and rice suspension cells gave some indications of mechanisms
   underlying biuret toxicity. In wild‐type rice seedlings, a biuret
   concentration of 0.1 mmol/L and above in the culture solution caused a
   significant reduction in the growth (Figure [126]1). It was roughly
   consistent with the toxic concentration of biuret reported for
   hydroponically grown naked barley (Funabiki et al., [127]1956) and pot
   cultured young citrus and avocado plants (Haas & Brusca, [128]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 [129]1). The
   colorless appearance of elongating young leaves indicates that
   excessive biuret impaired chloroplast development. Closely similar
   chlorosis was often observed in rice seedlings exposed to cold stress
   (Yoshida et al., [130]1996). It also occurs in seedlings suddenly
   exposed to direct sunlight after germination in the darkness during
   preparation for rice seedlings for machine transplanting. 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., [131]2011). Biuret might trigger similar downstream
   cellular responses.

   The chlorosis and reduced growth in rice seedlings are possibly not in
   a causal relationship. They may be parallel since biuret causes growth
   inhibition even in heterotrophic suspension cells of rice
   (Figure [132]6a). It is unlikely that decreased photosynthetic activity
   caused reduced growth in suspension cells.

   ROS seem to be involved in the upstream of such biuret injury
   (Figure [133]7). As well‐known, excess ROS cause severe cellular
   damages. Additionally, ROS under proper regulation are necessary for
   cellular and systemic signal transduction in plants (Fichman &
   Mittler, [134]2020; Miller et al., [135]2009). Plasma
   membrane‐localized NADPH oxidases, or respiratory burst homolog (Rboh),
   produce ROS. The produced ROS first accumulate in the apoplast, and
   then go into the cytosol or diffuse toward neighboring cells for
   signaling (Fichman & Mittler, [136]2020). Apoplastic ROS are also used
   for cell wall stiffening and loosening (Kärkönen &
   Kuchitsu, [137]2015). In our GO enrichment analysis, the term
   “peroxidase activity” is overrepresented (Figure [138]7,
   Figure [139]S4). This suggests the enhanced activity of peroxidases
   under biuret toxicity. Cell wall‐localized peroxidases could act in
   cross‐linking phenolic compounds using H[2]O[2] as an oxidant and thus
   tightening cell walls (Cosio & Dunand, [140]2009). At the same time,
   they could act in the cleavage of cell wall polysaccharides through the
   production of ·OH and thus loosening cell walls (Liszkay
   et al., [141]2003). Moreover, two overrepresented terms,
   “Pectinesterase inhibitor activity” and “hydrolase activity, acting on
   glycosyl bonds” were also related to cell wall polysaccharide
   modification. Therefore, it is likely that modifications of cell walls
   that are often found in plants under abiotic stresses
   (Tenhaken, [142]2015) also occur under biuret toxicity. The rigidity of
   cell walls restricts cell elongation. Our results do not show whether
   the cell wall rigidity changes or not. Enzyme activity and metabolites
   should also be examined. However, it may be a cause of retarded growth
   of rice plants under biuret toxicity.

   The gene expression change of rice suspension cells under biuret
   toxicity showed some similarities with various environmental stresses
   and plant hormones (Figure [143]7, Table S1). Overall responses for
   stress in plants result from a combination of signal transductions
   through pathways specific to the stress and common to other stresses
   (Knight & Knight, [144]2001; Sewelam et al., [145]2016). The pathway of
   the response to biuret excess in rice cells also appears to be
   complicated. Biuret is commercially used as non‐protein nitrogen to
   feed ruminants; that is, there is no known toxicity to animals.
   Therefore, we have postulated that biuret would inhibit some
   plant‐specific metabolic reactions. However, plant‐specific elaborated
   protection systems for stresses might be the cause of plant‐specific
   toxicity of biuret. The site of action of biuret is still unknown.
   Further research should be focused on this point.

   In conclusion, 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 [146]4, Figure 5).
   Moreover, rice plants overexpressing biuret hydrolase utilize
   ammonium‐N produced by the hydrolysis of biuret in plant cells as an
   additional N source (Table [147]2). In soil, the decomposition rate, or
   the mineralization rate, of biuret is slower than that of urea (Ogata &
   Funabiki, [148]1956; Sahrawat, [149]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.

AUTHOR CONTRIBUTIONS

   K.O. and T.M. conceived and designed the study and supervised the
   experiments. A.U. performed the experiments on intact plants. Y.M.
   isolated the soil bacterium. M.N. performed the experiments on
   suspension cells. K.O. contributed to the preliminary experiments,
   analyzed microarray data, and wrote the manuscript. All authors
   contributed to the writing of the manuscript.

Supporting information

   Supinfo S1
   [150]Click here for additional data file.^ (227.9KB, pdf)

   Supinfo S2
   [151]Click here for additional data file.^ (854.5KB, pdf)

   Supplemental Data S1
   [152]Click here for additional data file.^ (15.4KB, txt)

   Supplemental Data S2
   [153]Click here for additional data file.^ (63.8KB, txt)

ACKNOWLEDGEMENTS