OsNRAMP3 Is a Vascular Bundles-Specific Manganese Transporter That Is Responsible for Manganese Distribution in Rice

Manganese (Mn) is an essential trace element for plants. Recently, the genes responsible for uptake of Mn in plants were identified in Arabidopsis and rice. However, the mechanism of Mn distribution in plants has not been clarified. In the present study we identified a natural resistance-associated macrophage protein (NRAMP) family gene in rice, OsNRAMP3, involved in Mn distribution. OsNRAMP3 encodes a plasma membrane-localized protein and was specifically expressed in vascular bundles, especially in phloem cells. Yeast complementation assay showed that OsNRAMP3 is a functional Mn-influx transporter. When OsNRAMP3 was absent, rice plants showed high sensitivity to Mn deficiency. Serious necrosis appeared on young leaves and root tips of the OsNRAMP3 knockout line cultivated under low Mn conditions, and high Mn supplies could rescue this phenotype. However, the necrotic young leaves of the knockout line possessed similar levels of Mn to the wild type, suggesting that the necrotic appearance was caused by disturbed distribution of Mn but not a general Mn shortage. Additionally, compared with wild type, leaf Mn content in osnramp3 plants was mostly in older leaves. We conclude that OsNRAMP3 is a vascular bundle-localized Mn-influx transporter involved in Mn distribution and contributes to remobilization of Mn from old to young leaves.


Introduction
Manganese (Mn) is an essential metal nutrient in most organisms. In plants, Mn plays an important role in photosystem II and is a required cofactor for a variety of enzymes [1]. Mn deficiency can reduce plant growth and increase susceptibility to low temperature and pathogen infection [2]. Despite its importance, the amount of Mn required by a plant is relatively low; however, the capacity for Mn uptake always exceeds this requirement and excess Mn can be particularly toxic to plant growth [3]. In general, the uptake and detoxification of Mn is well balanced in plants. Many gene families have been identified as involved in Mn uptake or detoxification of excess Mn.
Much of our understanding on Mn uptake in plants comes from the complementation tests on yeast mutant strain Dsmf1, which is deficient in Mn uptake [4]. SMF1, a gene from natural resistanceassociated macrophage protein (NRAMP) family in yeast (Saccharomyces cerevisiae), is responsive to the high-affinity Mn 2+ accumulation into yeast cell. In plants, NRAMP family genes also contribute to Mn 2+ transportation or translocation. AtNRAMP1 and OsNRAMP5 encode plasma-located proteins that are major high-affinity Mn transporters in Arabidopsis and rice (Oryza sativa L.), respectively [5,6]. AtNRAMP3 and AtNRAMP4, two other NRAMP members in Arabidopsis, are targeted to vacuolar membrane and operate in the retrieval of Mn 2+ from vacuoles in leaf mesophyll cells under Mn deficiency [7]. Shown to be functional orthologs of AtNRAMP3 and AtNRAMP4, TcNRAMP3 and TcNRAMP4 from the metal hyperaccumulator Thlaspi caerulescens are both tonoplast proteins and can complement the Dsmf1 yeast mutant phenotype [8]. When expressed in yeast strains, LeNRAMP1 and LeNRAMP3 from tomato (Lycopersicon esculentum) are targeted to vesicle and tonoplast, respectively, and can retrieve the phenotype of yeast mutant Dsmf1 [9].
Mn always shares the same transporters with iron (Fe) in plants [3,5,10]. ZIP [zinc-regulated transporter/iron-regulated transporter (ZRT/IRT)-related proteins] genes and YSL (yellow stripe 1-like) genes are two well-known transporter families of Fe in plants. In recent years, many genes from these two families have been shown capable of transporting Mn. For example, AtIRT1 is an Arabidopsis ZIP transporter with a broad substrate range, including Mn 2+ [11]. Besides AtIRT1, six other ZIP genes (AtZIP1, AtZIP2, AtZIP3, AtZIP5, AtZIP6 and AtZIP9) from Arabidopsis were recently shown to be functional in transporting Mn 2+ in yeast [12]. Several members of the ZIP family from plant species other than Arabidopsis also showed the ability to transport Mn 2+ , including LeIRT1 and LeIRT2 from tomato, MtZIP4 and MtZIP7 from Medicago truncatula, and HvIRT1 from barley (Hordeum vulgare) [13][14][15]. In rice, no ZIP genes have been shown to transport Mn 2+ , but two members from the YSL family (OsYSL2 and OsYSL6) showed activity in Mn 2+ transport [16,17].
Unlike NRAMP and ZIP gene families, cation exchanger (CAX) genes in plants are always involved in tolerance of Mn toxicity. In Arabidopsis, six CAX genes have been identified. AtCAX1 and AtCAX2 were the first CAX genes reported in plants and are known calcium (Ca) transporters [18]. However, expression studies in tobacco (Nicotiana tabacum) and yeast showed that AtCAX2 is also involved in Mn 2+ transport [19]. The protein encoded by AtCAX2 is targeted to vacuolar membrane and confers Mn 2+ tolerance in Arabidopsis. Furthermore, in a yeast Mntolerance screen with an Arabidopsis cDNA library, AtCAX2 was the only gene identified out of the 10 5 transformants that could suppress the Mn toxicity phenotype [20]. However, other CAX genes in Arabidopsis did not show Mn 2+ transport activity. But interestingly, when expressed in yeast with an N-terminally truncated form, AtCAX5 could mediate Mn 2+ transport, suggesting that CAX genes in Arabidopsis may control the Mn 2+ transport activity by an autoregulatory region at the N-terminus [21]. Five CAX genes have been identified in rice. The OsCAX3 and Nterminal truncated OsCAX1a from rice could confer tolerance to Mn when expressed in yeast [22]. In addition to the CAX gene family, the endomembrane-type Ca-ATPase (ECA) gene family is another one involved in both Ca 2+ and Mn 2+ transport. In Arabidopsis, AtECA1 and AtECA3 showed Mn transport activity when expressed in yeast [23,24]. Interestingly, protein encoded by AtECA1 is mainly targeted on endoplasmic reticulum membrane and increases tolerance of plants to Mn toxicity; however, AtECA3 encodes a Golgi-localized protein and is required in Arabidopsis under Mn-deficient conditions [25].
The cation diffusion facilitator (CDF) gene family also contributes to Mn tolerance. Four CDF genes (ShMTP1-ShMTP4), the first CDF genes identified in plants, were screened from Stylosanthes hamate for enhancing Mn 2+ tolerance by expressing a cDNA library in yeast [26]. Further studies on ShMTP1 suggested that it encoded a tonoplast protein in Arabidopsis and conferred Mn 2+ tolerance through internal sequestration. There are 12 members in the CDF family in Arabidopsis, and proteins encoded by four of these genes (AtMTP8-AtMTP11) are closely related to the protein encoded by ShMTP1 [27]. Furthermore, knockout of AtMTP11 in Arabidopsis resulted in hypersensitivity to high Mn 2+ , suggesting that AtMTP11 can confer plant Mn tolerance [27,28].
Most knowledge concerning Mn in plants was obtained from studies focused on uptake and tolerance, but little is known about the mechanisms of distribution and translocation of Mn in plants.
In this study, we identified another NRAMP gene from rice, OsNRAMP3, involved in Mn distribution. OsNRAMP3 encoded a plasma membrane-localized protein with activity in transporting Mn and was expressed specifically in vascular bundles, especially phloem cells. Knocking out of OsNRAMP3 in rice resulted in high sensitivity to Mn deficiency and disturbed Mn distribution in leaves. These data suggested that OsNRAMP3 played an important role in Mn distribution in rice.

Plant Materials
The knockout line of OsNRAMP3 and wild type were based on O. sativa L. ssp. japonica cv. DongJin background. The osnramp3 mutant is a T-DNA insert mutant ordered from the POSTECH RISD database (http://www.postech.ac.kr/life/pfg/risd/) [29]. Promoter analysis and expression profile analyzed by real-time PCR were based on O. sativa L. ssp. japonica cv. Zhonghua 11.

Subcellular Localization of OsNRAMP3
The coding sequence of OsNRAMP3 for subcellular localization was amplified using cv. Zhonghua11 total cDNA as template and primers 59-gactgaattcatgagcggcccaatgcaa-39 (forward) and 59gactggtaccatcgagatcagaagcagttcgct-39 (reverse), which were designed based on the sequence of LOC_Os06g46310.1 from the RGAP database (http://rice.plantbiology.msu.edu/). The PCR product was cleaved using EcoR1 and Kpn1 and ligated into pM999-GFP with correct direction. The construct was transformed into Arabidopsis mesophyll protoplasts and observed using a confocal laser scanning microscope (TCS SP2; Leica) after incubation at 22uC for 12-24 h [30].

GUS Staining Assay
To construct the OsNRAMP3-promoter:GUS plasmid, 2.1 kb of genomic sequence located upstream of the OsNRAMP3 initiation codon was amplified by PCR from cv. Zhonghua 11 genomic DNA using primers Pnr5-F (59-agctgcagatgcgccaaaatactgaat-39) and Pnr5-R (59-tcggatcctgcaagaaccctcaagact-39). The amplified promoter fragment was digested by BamH1 and Pst1 and introduced into vector pDX2181 in the correct direction [31]. The construct was transformed into cv. Zhonghua 11 by Agrobacterium tumefaciens-mediated transformation [32]. The transgenic plant tissues were incubated in X-Gluc strain buffer at 37uC for 4 h [33].

RNA Extraction and Real-time PCR
Total RNA was extracted using TRizol reagent (Invitrogen). Of total RNA, 3 mg was used to synthesize the first-strand cDNAs in 20 ml of reaction mixture using M-MLV reverse transcriptase (Invitrogen) according to the manufacturer's instructions. Realtime PCR was performed using the SYBR Premix Ex Taq TM (TaKaRa) with the following gene-specific primers for OsN-RAMP3: rqNR3-F (59-tcagcagcgaactgcttctgatct-39) and rqNR3-R (59-atcagctggctaactctttgggct-39). The rice Ubiquitin 5 gene was used as the internal control with the following primers: qUbq-F (59aaccagctgaggcccaaga-39) and qUbq-R (59-acgattgatttaaccagtccatga-39). The real-time PCR reaction was performed on an Applied Biosystems 7500 PCR instrument.

Elemental Analysis
Shoots and roots were harvested separately and roots were washed with distilled water twice before sampling. After drying at 80uC for 3 d, all samples were digested in 65% nitric acid in a MARS6 microwave (CEM) at a temperature gradient of 120-180uC for 45 min, and then diluted in deionized water. The metal contents of the samples were determined by inductively coupled plasma-mass spectrometry (ICP-MS; Agilent 7700 series, USA).
For metal determination, yeast cells transformed with different vectors were grown for 30 h starting from OD600 of 0.01 on SD-Ura liquid medium supplemented with 0.02 mM Mn at pH 6.0. Galactose instead of glucose was used here for induction of the GAL promoter. Cells were harvested by centrifugation and washed three times with deionized water (MilliQ; Millipore). All samples were dried at 80uC for 3 d, and then used to determine metal contents by ICP-MS.

Expression Profiles of OsNRAMP3 in Different Tissues of Rice
Seven different tissues from O. sativa L. ssp. japonica cv. Zhonghua 11 were used to investigate the expression pattern of OsNRAMP3. We found OsNRAMP3 transcript levels to be similar in most tissues and slightly higher in panicles and the first culm ( Figure 1A). We next detected transcript levels of OsNRAMP3 in different leaves ( Figure 1B). Four leaves from the same tiller were analyzed here, in which the first leaf was the oldest and the fourth leaf was youngest and partly wrapped in sheath. Interestingly, the expression level of OsNRAMP3 increased with leaf age under both Mn-sufficient and Mn-deficient conditions. In addition, the fourth leaf showed a differential expression pattern of clearly lower OsNRAMP3 expression in unexpanded compared to expanded regions ( Figure 1C).
The NRAMP family genes in Arabidopsis are always induced by Fe or Mn deficiency [5,35]. Time-course experiments under Mn and Fe deficiency were used to test the response of OsNRAMP3 to Mn and Fe deficiency, respectively -real-time PCR revealed that OsNRAMP3 was little affected by Mn or Fe deficiency ( Figure 1D and 1E). In addition, expression did not respond to different Mn supplies ( Figure S1).

Identifying the OsNRAMP3 Knockout Line
OsNRAMP3 contains 13 exons and 12 introns (Figure 2A), encoding a protein with 550 amino acids. To investigate the biological function of OsNRAMP3, we obtained a T-DNA insertion line from POSTECH [29]. The T-DNA was inserted into the twelfth exon. Transcripts of OsNRAMP3 could not be detected in the line ( Figure 2B), suggesting it was a knockout mutant of OsNRAMP3.
As NRAMP genes always play important roles in transporting Mn, we firstly investigated the response of osnramp3 plants (the knockout line of OsNRAMP3) to different Mn supplies ( Figure 2C, panels a-c). Interestingly, when plants were cultivated at low Mn supply (0.08 mM), osnramp3 plants showed obviously reduced growth compared with wild type. Careful observation revealed serious necrosis in young leaves and roots tips of osnramp3 plants, but not in older leaves ( Figure 2C, panels d-f).
After harvest, the plant height, root length and dry weight of samples were recorded. At low Mn (0.08 mM) condition, the plant heights and root lengths of osnramp3 showed obvious decreases ( Figure 3A and 3B); and dry weights of the roots and shoots in osnramp3 were 54 and 70% of wild type, respectively ( Figure 3C and 3D), which confirmed the observed phenotype ( Figure 2). When supplied with medium (8 mM) or high (800 mM) Mn, root of osnramp3 plants were shorter than that of wild type but there were no differences in plant height or dry weight.

Metal Determination in Roots and Shoots
Since the knockout line of OsNRAMP3 was sensitive to low Mn, we determined the Mn contents of roots and shoots of osnramp3 and wild-type plants. At a 0.08 mM Mn supply, both roots and shoots of the knockout line accumulated significantly more Mn than the wild type ( Figure 4A and 4B). There was no apparent difference in Mn contents between osnramp3 and wild-type plants at higher Mn supplies. Absence of OsNRAMP3 also did not affect magnesium (Mg), Ca, Fe, copper (Cu), zinc (Zn) and cadmium (Cd) contents of roots and shoots of rice plants cultivated under normal conditions ( Figure S2).

Knockout of OsNRAMP3 Resulted in Disturbed Distribution of Mn in Leaves
It seemed contradictory that osnramp3 mutant plants possessed higher Mn contents in roots and shoots but showed reduced growth under low Mn conditions compared with wild-type plants.
To determine how this occurred, we next examined the distribution of Mn contents in different leaves for 0.08 or 8 mM Mn supplies. In general, Mn contents increased with leaf age in the treated and control plants. However, at a 0.08 mM Mn supply, osnramp3 plants showed significantly higher Mn contents in the first and second leaves than wild-type plants; but there were no significant differences in third and fourth leaves ( Figure 5A). Under 8 mM Mn conditions, higher Mn contents were also observed in first leaves of osnramp3 compared with wild-type plants ( Figure 5B); however, this difference was not observed in other leaves. The results here revealed that the absence of OsNRAMP3 in rice caused the Mn accumulation in older leaves, suggesting that OsNRAMP3 was important for Mn distribution in leaves.
In addition, the contents of Ca, Fe and Cd in older leaves were slightly, but not significantly, higher than that of wild type ( Figure  S3). There were no differences in Mg, Cu and Zn contents in corresponding leaves between wild type and osnramp3 plants.

Expression of OsNRAMP3 was Specific to Vascular Bundles
To further investigate the expression pattern of OsNRAMP3 in different tissues, a 2106-bp promoter region of OsNRAMP3 was used to direct beta-glucuronidase (GUS) expression in rice plants. GUS activity was analyzed for different rice tissues. Although there was high GUS activity in various tissues, such as roots ( Figure 6A, 6B and 6C), ligules and sheaths of leaves ( Figure 6D), leaf blades ( Figure 6E and 6F) and grain hulls ( Figure 6G), it was specific to vascular bundles of these tissues. In addition, GUS activity was also found in root tips and dorsal vascular bundles of endosperm ( Figure 6A and 6H). To determine the exact location of GUS activity, we made transverse sections of roots and leaves after GUS staining. High-magnification observation of roots showed that GUS activity was specific in phloem cells ( Figure 6I and 6J). In leaves, GUS activity was mainly detected in phloem cells as also for roots ( Figure 6K, 6L, 6M and 6N). An enlarged view of the phloem region revealed particularly strong GUS activity in companion cells ( Figure 6M). In addition, some parenchyma cells between phloem and xylem also showed GUS activity in large vascular bundles ( Figure 6L).

OsNRAMP3 Encodes a Plasma Membrane-localized Protein
To study the subcellular localization of the OsNRAMP3 protein, a vector expressing an OsNRAMP3-GFP fusion under the control of cauliflower mosaic virus (CaMV) 35S promoter was constructed and transformed into Arabidopsis protoplasts. The OsNRAMP3-GFP expression in the protoplasts was examined with confocal microscopy 16 h after transformation. The fluorescence showed that OsNRAMP3 was a plasma membranelocalized protein (Figure 7).

OsNRAMP3 is a Functional Mn-influx Transporter
To determine the Mn transport activity of OsNRAMP3, we expressed OsNRAMP3 in yeast mutant strain Dsmf1. Here AtNRAMP1 was used as a positive control, as AtNRAMP1 has been shown to be a functional Mn transporter and can retrieve the growth of Dsmf1 under Mn-limited conditions [5,36]. The growth of yeast strains Dsmf1 and its wild type transformed with pYES2 (negative control), OsNRAMP3 or AtNRAMP1 were analyzed on Mn-limited medium which controlled by different concentration of ethylene glycol tetraacetic acid (EGTA). The results showed that OsNRAMP3 significantly rescued the growth of Dsmf1 when Mn was limited ( Figure 8A), suggesting that OsNRAMP3 possessed high activity for transporting Mn. To further confirm this result, metal concentrations were analyzed in these strains. When OsNRAMP3 was expressed in yeast strain Dsmf1, the Mn contents were almost retrieved to the level of wild type ( Figure 8B), strongly indicating that OsNRAMP3 was a Mn-influx transporter.

OsNRAMP3 is a Vascular Bundle-located Mn-influx Transporter Involved in Mn Translocation
NRAMP family genes were well known to be related to Mn transport and many of them which encoded plasma-membrane targeted proteins acted as high-affinity transporters to absorb Mn from outside: such as SMF1 from yeast, AtNRAMP1 from Arabidopsis and OsNRAMP5 from rice [4][5][6]. In the present study, we identified OsNRAMP3, another NRAMP gene from rice, with the function of Mn transport. We demonstrated that knockout of OsNRAMP3 resulted in serious reduction in growth of rice plants under low Mn conditions and this reduction could be rescued by higher Mn supply (Figure 2 and 3). Indeed, the contents and translocation of Mn in rice plants were obviously affected by the knockout of OsNRAMP3 (Figure 4 and 5). Consistent with a role of OsNRAMP3 in the Mn translocation are the following findings: (1)

Possible Mechanisms of OsNRAMP3 Involvement in Mn Translocation
In general, proteins encoded by NRAMP genes can be divided into two groups: plasma membrane-targeted and intracellular membrane-located proteins. The plasma membrane-targeted proteins from the NRAMP family contribute to uptake of Mn from medium as described above [5,6]. The intracellular membrane-located proteins always act to retrieve Mn from vacuoles or other organelles: such as AtNRAMP3 and AtN-RAMP4 from Arabidopsis [7], TcNRAMP3 and TcNRAMP4 from Thlaspi caerulescens [8], and LeNRAMP1 and LeNRAMP3 from tomato [9]. In this study, OsNRAMP3 showed difference: OsNRAMP3 encoded a plasma membrane protein (Figure 7) but was not involved in Mn uptake. The knockout mutant of OsNRAMP3 was highly sensitive to Mn deficiency ( Figure 2C), which were similar to the characteristics of OsNRAMP5 described by Sasaki et al. [6]. However, OsNRAMP3 expression was restricted to vascular bundles rather than the exodermis and endodermis of roots (Figure 6), indicating a different function from OsNRAMP5. In addition, knockout of OsNRAMP3 resulted in Mn contents increasing in roots and shoots of osnramp3 mutant plants     withered appearance could be retrieved by higher Mn supply ( Figure 2C), suggesting that the withered leaves of osnramp3 were deficient in the Mn required to maintain growth. However, osnramp3 plants accumulated more Mn in both roots and shoots compared with wild-type plants (Figure 4), which seemed contradictory. Most probably as feedback of signals of lacking Mn in withered leaves, the Mn uptake system of osnramp3 plants was enhanced to absorb more Mn from outside, which caused increased Mn contents in roots and shoots of osnramp3 compared with wild type. Consistent with that, further studies showed that the Mn content of each leaf of osnramp3 plants was not lower than corresponding leaves of wild-type plants ( Figure 5A). This suggested that the symptoms of lacking Mn in withered leaves were caused by disturbed distribution of Mn rather than general Mn shortage. GUS analysis in leaves showed that OsNRAMP3 was specifically expressed in phloem cells and some parenchyma cells between phloem and xylem ( Figure 6M and 6N), implying that the disturbed distribution of Mn was probably due to sequestrating  Mn into vascular bundles of osnramp3 plants. As OsNRAMP3 is an influx Mn transporter, and parenchyma cells in vascular bundles always operate the loading or unloading of ion for xylem [37,38], the Mn transported via xylem was probably unable to leave the xylem due to the absence of OsNRAMP3 in knockout lines. However, this suggestion requires more supporting evidence.
Phloem always plays important roles in remobilization of various ions in shoots, such as Zn, Fe, Cu and Mn [39][40][41]. In barley, it was reported that the transpiration stream has little effect on Mn translocation to the youngest leaf, whereas the strong effect on Mn translocation in older leaves [42], implying that major Mn concentration in young leaf is probably transported through the phloem. In the present study, OsNRAMP3 encoded a Mn-influx transporter and was mainly expressed in phloem cells, suggesting that OsNRAMP3 may contribute to remobilization of Mn in shoots. The increased Mn contents in osnramp3 plants mainly occurred in older not younger leaves under low Mn conditions ( Figure 5A). Coincidentally, in rice plants, the expression of OsNRAMP3 in leaves slightly increased with leaf aging ( Figure 1B). This implied that OsNRAMP3 played an important role in leaf Mn distribution. Together with the phenotype that only young leaves of osnramp3 plants were sensitive to Mn deficiency, the results here strongly suggested that OsNRAMP3 operated the Mn remobilization from old to young leaves via phloem cells.
In rice, OsYSL2 was reported to be a phloem-located metal transporter responsible for long-distance transport of Fe and Mn, especially in seeds [16,41]. Similarly to OsYSL2, the expression of OsNRAMP3 was also detectable in vascular bundles of endosperm and hulls, suggesting that OsNRAMP3 was required in these tissues. Confirming whether OsNRAMP3 is involved in transporting Mn into seeds requires more detailed study. In roots, the expression of OsNRAMP3 was specific to phloem cells and root tip cells ( Figure 6A and 6J). Coincidently, the root tip of knockout plants of OsNRAMP3 became necrotic under low Mn conditions (Fig. 2C, panel e). We investigated the Mn contents in different root regions at different distances from the root tip in rice cv. Zhonghua 11 (data not shown). The root tip possessed the highest Mn contents of the different root regions, suggesting that Mn played an important role in growth of root tips. However, because of serious necrosis in root tips of osnramp3 plants under low Mn conditions, it is difficult to determine the Mn contents in osnramp3 root tips. To determine whether OsNRAMP3 is related to root tip Mn contents and how OsNRAMP3 affects growth of root tips will require further study.
In conclusion, OsNRAMP3 is a vascular bundle-localized influx Mn transporter that is involved in Mn translocation in rice plants. Knockout of OsNRAMP3 disturbed the transport of Mn into young leaves from old leaves and vascular bundle cells, resulting in high sensitivity to low Mn conditions. It is worth noting that when we finished our experiments, we noticed a related but independent work by Yamaji et al. [43], in which OsNRAMP3 was also described as a Mn transporter contributing to Mn distribution in shoots.