The presence of microRNA species in plant phloem sap suggests potential signaling roles by long-distance regulation of gene expression. Proof for such a role for a phloem-mobile microRNA is lacking. Here we show that phosphate (Pi) starvation-induced microRNA399 (miR399) is present in the phloem sap of two diverse plant species, rapeseed and pumpkin, and levels are strongly and specifically increased in phloem sap during Pi deprivation. By performing micro-grafting experiments using Arabidopsis, we further show that chimeric plants constitutively over-expressing miR399 in the shoot accumulate mature miR399 species to very high levels in their wild-type roots, while corresponding primary transcripts are virtually absent in roots, demonstrating shoot-to-root transport. The chimeric plants exhibit (i) down-regulation of the miR399 target transcript (PHO2), which encodes a critical component for maintenance of Pi homeostasis, in the wild-type root, and (ii) Pi accumulation in the shoot, which is the phenotype of pho2 mutants, miR399 over-expressers or chimeric plants with a genetic knock-out of PHO2 in the root. Hence the transported miR399 molecules retain biological activity. This is a demonstration of systemic control of a biological process, i.e. maintenance of plant Pi homeostasis, by a phloem-mobile microRNA.
Correct allocation of the available Pi in the plant, i.e. maintenance of Pi homeostasis by supplying growing organs and tissues such as young leaves and inflorescences with the spare resources taken up by roots, is also important for plant productivity and survival during Pi limitation. The existence of mutants with altered levels and distribution of Pi (pho1, pho2 and pup1; see Delhaize and Randall, 1995; Poirier et al., 1991; Trull and Deikman, 1998) suggests that plants regulate the long-distance transport and allocation of Pi between shoot and root. PHO1 encodes the founder member of a novel transporter family (Hamburger et al., 2002) and probably is involved in Pi transport into the xylem (Wang et al., 2004), and PHO2 encodes an E2 ubiquitin-conjugating enzyme (Aung et al., 2006; Bari et al., 2006). The PHO2 E2 conjugase is a major component for the maintenance of Pi homeostasis in Arabidopsis. pho2 mutants grown in Pi-replete conditions display continuous induction of a subset of phosphate starvation-induced (PSI) genes, including some encoding phosphate transporters (Aung et al., 2006; Bari et al., 2006), and accumulate Pi to high levels (three- to fivefold increased) in shoots but not roots (Bari et al., 2006; Delhaize and Randall, 1995).
PHO2 expression is post-transcriptionally regulated by microRNA399 (miR399). miR399 acts through transcript cleavage (Allen et al., 2005) and probably also translational repression (Bari et al., 2006). Consistently, over-expression of miR399 leads to high shoot Pi levels, thus phenocopying pho2 mutants (Aung et al., 2006; Bari et al., 2006; Fujii et al., 2005). The PHO2 gene harbors five binding sites complementary to miR399 in its 5′ UTR approximately 200–400 nucleotides upstream of the start codon, and miR399-dependent PHO2 transcript cleavage at these sites has been experimentally verified (Allen et al., 2005). Six miR399 species (a–f) that arise from five primary transcripts (PTs) have been identified in Arabidopsis (http://microrna.sanger.ac.uk). miR399d is by far the most prominent one during Pi limitation, based on the abundance of its primary transcript (Bari et al., 2006). The levels of mature miR399 and the five miR399 PTs are highly and specifically increased during Pi limitation (Bari et al., 2006; Fujii et al., 2005), suggesting inhibition of PHO2 under these conditions. Consistently, removal of the miR399 binding sites results in high and stable PHO2 transcript levels in Pi-limited plants (Fujii et al., 2005). Micrografting experiments revealed that a pho2 mutant root genotype is sufficient to result in shoot Pi accumulation (Bari et al., 2006). On the other hand, MIR399 genes, and especially MIR399d, display particularly strong induction in shoots (as compared to roots) during Pi limitation, with the shoot miR399d PT level exceeding the root PHO2 transcript level (Bari et al., 2006). This led us to hypothesize that miR399 itself might act as a long-distance Pi starvation signal in Arabidopsis.
Presence and Pi-specific induction of miR399 in phloem sap
To demonstrate the presence of miR399 and its Pi-status-dependent changes in phloem sap, we grew rapeseed (Brassica napus) and pumpkin (Cucurbita maxima) plants under abundant and limiting Pi conditions, and collected phloem sap for the determination of microRNA levels. Doing this type of experiment with Arabidopsis is impracticable because of (i) major difficulties associated with the collection of sufficiently large quantities of phloem sap from this small plant species, (ii) problems with RNase contamination and/or wounding-inducible RNase activity (Köck et al., 2004; LeBrasseur et al., 2002) in phloem sap exudates collected by decapitation or incision methods, making isolation and quantification of RNA impossible (A.B. and J.K., unpublished observations), and (iii) contamination of the sap with content (e.g. RNA) from other cells. Rapeseed plants are a useful alternative to Arabidopsis due to the close genetic relatedness of the two species, as well as their similar phloem sap composition, as judged by their protein profiles (J.K., unpublished data). In addition, phloem sap exudation from incised rapeseed inflorescence stems is quite strong, thus enabling rapid sampling and minimization of problems involving contaminating/induced RNase activity.
Mature miR399 was detected by quantitative real-time PCR and RNA gel blotting in rapeseed and pumpkin phloem sap (Figure 1), but miR399 precursors or primary transcripts were not detectable (results not shown). Mature miR399 levels increased very strongly in the phloem sap of Pi-limited plants (a difference of 7–10 quantitative real-time PCR CT value units, equivalent to a >100-fold change) (Figure 1a–c,e). The increase was even stronger and the abundance higher in phloem sap than in leaves, for example (Figure 1c), suggesting efficient transport of miR399 into the sap and/or strong induction of miR399 expression in phloem companion cells. The increase/accumulation of miR399 in phloem sap or other plant organs such as roots was also specific, as levels for other microRNA species, i.e. miR164 (Figure 1d) and miR172 (Figure 1f), did not change during Pi limitation. These results demonstrate that miR399 is present and highly abundant in the phloem, and indicate that the increase during Pi limitation is not related to decreased phloem mass flow. They also underscore the conserved Pi-responsiveness of miR399 in distantly related plant species (Bari et al., 2006; Chiou et al., 2006).
We further investigated the Pi specificity of this response by investigating miR399 levels in phloem sap of plants starved of sulfate. RNA gel blotting showed that mature miR399 abundance was unaffected by sulfate deprivation in phloem sap (Figure 1g) and leaves and roots (results not shown) of rapeseed plants. This is in accordance with previous results showing that miR399 primary transcripts are unaffected by limitation of sulfate, nitrogen or carbohydrates (Bari et al., 2006). It was also demonstrated that mature miR399 is unaffected by nitrogen (Bari et al., 2006) or potassium limitation (Fujii et al., 2005). Taken together, these results show that the responsiveness of miR399 and its primary transcripts is Pi-specific in all plant organs/samples, including phloem sap.
Micrografting reveals shoot-to-root transport and biological activity of shoot-derived miR399 in roots
We next investigated whether miR399 produced in the shoot can be detected and consequently alter expression of its target gene (PHO2) in the root. For this purpose, we used micro-grafting (Turnbull et al., 2002) to generate chimeric Arabidopsis plants that over-express miR399 in either shoots or roots. Six miR399 species (a–f) that arise from five primary transcripts have been identified in Arabidopsis. miR399d is the most prominent one during Pi limitation, based on the abundance of its primary transcript (Bari et al., 2006) (Figures 2 and 3). Transgenic plants strongly over-expressing miR399d (OX) under control of the 35S promoter were previously shown to have approximately 10-fold lower levels of the target PHO2 transcript (Bari et al., 2006), and to accumulate Pi in leaves to levels as high as in pho2 mutants or in micro-grafted chimeric plants with a pho2 root and a wild-type (WT) shoot scion (Bari et al., 2006).
As expected, the primary transcripts of miR399d (miR399d PT) and mature miR399 were strongly increased (threshold fluorescence was reached 12–13 cycles earlier during quantitative real-time PCR, equivalent to a >1000-fold increase) in root and shoot material from OX plants compared to the same material from WT plants (Figures 2 and 3 and Table S1). The levels of miR399d PT and mature miR399 in OX shoots matched the high levels found in Pi-starved WT shoots (Figure 2d,f), but the levels of the other miR399 primary transcripts (a, b, c, e/f) remained at barely detectable or undetectable levels, as in Pi-replete WT (see Table S1; note that in Figures 2 and 3 values for 40-deltaCT of less than approximately 30 correspond to transcript copy numbers of far less than 1 per cell in our quantitative real-time PCR system (Czechowski et al., 2004; also see Experimental Procedures)). In Figures 2 and 3, values for 40 −ΔCT of less than approximately 30 correspond to transcript copy numbers of much less than 1 per cell in our quantitative real-time PCR system (see Experimental procedures). The level of PHO2 transcript, measured using two quantitative real-time PCR primer pairs amplifying near the 3′ and 5′ ends of the PHO2 coding sequence (CDS), was approximately 10-fold decreased in OX plants (equivalent to 3 or 4 quantitative real-time PCR cycles) (Figure 2g,h and Table S1), and the leaf Pi level was four- to fivefold increased (Figure 4) as previously reported (Bari et al., 2006).
When shoot material from chimeras with an OX shoot and WT root (OX/WT) was investigated, it showed the same behavior as OX shoots of non-chimeric plants with respect to the levels of miR399 PTs, mature miR399 and PHO2 transcript (Figure 2). The root material of the same OX/WT chimeras exhibited WT-like miR399 PT levels (Figure 3a–e and Table S1), which were all below the expression level of some of the rarest transcription factor transcripts known (e.g. WUSCHEL, At2g17950), equivalent to an abundance of approximately 1 copy per 100 cells or less for each miR399 PT (see Czechowski et al., 2004). In contrast, the same WT roots had very high, OX-like, mature miR399 levels (Figure 3f), equivalent to a copy number of > 500 per cell, which is similar to that of one of the most abundant plant transcripts, UBQ10 (see Czechowski et al., 2005) and OX-like low PHO2 transcript level (Figure 3g,h). The shoot material had high Pi levels in the leaves, similar to those of genotypes in which PHO2 is inhibited in roots (Figure 4). We have previously shown by micro-grafting of pho2 and WT that decreased PHO2 in the shoot alone does not lead to increased Pi in the shoot (Bari et al., 2006).
A theoretical explanation for the virtually absent miR399 PTs and the simultaneous massive accumulation of mature miR399 in roots of OX/WT chimeras could be highly efficient cleavage of miR399 PTs into shorter fragments that escape oligo(dT)-primed reverse transcription and subsequent quantitative real-time PCR detection. Although there is no indication of highly efficient cleavage in roots of OX plants or WT/OX chimeras (i.e. the level of miR399 PTs mirrors the mature miR399 level), we analyzed whether there is an increase of corresponding fragments by reverse-transcribing total RNA with a primer binding to the 3′ region of the miR399d precursor sequence (Figure 5) (see Experimental procedures). However, there was no evidence for accumulation of such fragments in the roots of the OX/WT chimeras, as the level was as low as in Pi-replete WT. Similarly, the miR399 complementary duplex strand (miR399*) could not be detected in roots of OX/WT chimeras or roots of Pi-replete WT, but was clearly present in Pi-limited WT roots and miR399d PT over-expresser roots (results not shown). These results further suggest that the MIR399 genes are truly inactive in the roots of the OX/WT chimeras. The explanation for the massive accumulation of mature miR399 found in roots of OX/WT chimeras therefore must be transport of miR399 molecules from the OX scion to the WT root, where they exert their biological function, i.e. suppress PHO2. Co-localization of miR399 and the PHO2 target transcript in the root vascular cylinder has been reported previously (Aung et al., 2006), and further supports their interaction in vivo. The inactivity of the MIR399 genes in the roots of OX/WT chimeras also excludes the possibility that another phloem-mobile molecular species (e.g. a hormone or metabolite) acts as an intermediate signal to trigger miR399 synthesis.
A situation similar to that in the OX/WT chimera was also observed in Pi-deprived WT plants. During Pi starvation, levels of the miR399 PT rise in WT roots, but only to levels that are approximately 1% of the level of mature miR399 (Figure 3 and Figure S1a), while miR399 PT levels in WT shoots equal or exceed the mature miR399 level (Figure 2 and Figure S1a). This result indicates that miR399 phloem transport is relevant in the normal biological situation. In addition, given the relatively long half-life time of mature miR399 (approximately 12 h, Figure S1b) (Bari et al., 2006), we estimated that phloem transport rates are sufficiently high to achieve and maintain very high (i.e. PHO-inhibitory) levels of mature miR399 in roots that do not express miR399 PTs within a physiological response time of 2–3 h (see Appendix S1). In summary, Pi deprivation in the shoot can be rapidly signaled to the roots through miR399. Subsequent inhibition of PHO2 by miR399 in the root will lead to increased Pi uptake and transport to the shoot (Aung et al., 2006; Bari et al., 2006; Delhaize and Randall, 1995). This elegant mechanism is capable of maintaining shoot Pi homeostasis and optimizing shoot growth during Pi limitation.
No evidence for root-to-shoot transport of miR399
Reciprocal chimeras (WT/OX) that over-express miR399d PT and mature miR399 in roots were also investigated. The root material (Figure 3) resembled that of normal OX plants, while the shoot material (Figure 2) resembled WT material, with low miR399d PT and low mature miR399. This indicates that root-to-shoot transport of miR399, which would occur via the xylem, does not take place or occurs to a negligible extent. Unlike water or ions, the transport of larger molecules, including miR399, is not circulatory between shoot and root (Lough and Lucas, 2006). Accordingly, miR399 was not detectable in xylem sap from oilseed rape plants grown under normal or phosphate-deficient conditions (Buhtz et al., in press). The level of PHO2 transcript in the shoots of WT/OX chimeras was high, resembling Pi-replete WT shoots (Figure 2g,h). Leaf Pi levels were also high (Figure 4), resembling genotypes in which the level of PHO2 transcript is low in roots (Bari et al., 2006).
Taken together, these results pinpoint the mature miR399 as a phloem-mobile molecule that increases in concentration in phloem sap during Pi limitation, and that can be transported between plant organs and exert its biological role at its destination. Hence, miR399 fulfills the criteria for a long-distance signaling molecule. The existence of multiple microRNA species in phloem sap of diverse plant species (Buhtz et al., in press; Yoo et al., 2004) indicates that this might only be one example of systemic control of a biological process by a microRNA.
Despite the demonstration that miR399 is transported in the phloem conduit, and thereby plays a central role in the maintenance of whole-plant Pi homeostasis, several questions remain to be addressed with respect to the transport mechanism. These include the mechanism and proteins required for miR399 entry into and exit from the sieve tubes, and the requirement and nature of RNA binding proteins to allow miR399 phloem transport, such as the 27 kDa phloem small RNA binding protein CmPSRP1 (Yoo et al., 2004) or the 16 kDa phloem protein 16 (CmPP16) (Xoconostle-Cázares et al., 1999) in Cucurbita maxima. With respect to translocation in the phloem, the current data indicate that miR399, as already suggested for the whole miRNA population (Yoo et al., 2004), travels as a single-stranded molecule, as, in contrast to the miR399 sense strand, the level of the miR399* strand does not show any increase under phosphate starvation (Buhtz et al., in press). miR399 is efficiently translocated from Pi-replete OX shoots to Pi-replete WT roots. This suggests that the components of miR399 phloem uptake, translocation and exit are either constitutively present or induced during Pi limitation in a miR399-dependent manner.
Long-distance signaling in biological systems was put in place when multi-cellular organisms evolved. In this regard, it is interesting to note that miR399 and its target PHO2 gene are found in higher multi-cellular plants only, including rice, poplar and Medicago, but not in single-celled algae (Bari et al., 2006). The Pi-dependence of miR399 expression is strongly conserved between dicots and monocots, as is the target gene with respect to gene structure, encoded protein sequence, and the number and positions of miR399 binding sites in its 5′ UTR (Bari et al., 2006). This suggests co-evolution and conservation of the molecular components required for Pi homeostasis in higher multi-cellular plants. Long-distance transport of miR399 further cements this view.
Growth of rapeseed and pumpkin, and phloem sap sampling
Rapeseed (Brassica napus cv. Drakkar) and pumpkin (Cucurbita maxima) plants were grown in 18 cm pots containing a low-Pi soil mix in a phytotron at 25°C (day)/20°C (night), with a 16 h light period at an intensity of 550 μE. Plants were divided into two sets; one set was supplied with a full nutrient solution containing sufficient (3 mm) Pi, and the other one was supplied with the same nutrient solution free of Pi. Rapeseed plants were grown until flowering (7–8 weeks) and phloem sap was then collected by incising inflorescence stems (Giavalisco et al., 2006). Phloem sap from pumpkin plants was collected by stem incision 3 weeks after germination. After discarding the first droplets, the exuded phloem sap was collected into four volumes of ice-cold Trizol reagent (Invitrogen, http://www.invitrogen.com/).
Hydroponically grown rapeseed plants were germinated on filter paper for 1 week before transferring them to a hydroponics system containing nutrient medium (Buhtz et al., in press) for 7–8 weeks. Nutrient solutions were changed after 3 weeks, and subsequently renewed once a week. Phosphate and sulfur starvation were initiated after 3 weeks by changing to medium containing no phosphate or sulfur, respectively, and phloem sap was collected from inflorescence stems as described above.
Micrografting of Arabidopsis seedlings
Micrografting of 6-day-old seedlings grown on sterile vertical agar plates was performed as described previously (Turnbull et al., 2002). Seeds of the miR399d over-expressers, wild-type (ecotype Col-0) and pho2 mutant were surface-sterilized, and stratified at 4°C for 3 days. The seed were then laid onto sterile half-strength MS agar (0.7%) plates. The plates were kept vertically in constant light (approximately 120 μE) at 20°C for 3 days in an Arabidopsis growth chamber (Percival Scientific Inc., http://www.percival-scientific.com/), and then kept at 28°C under an 8 h photoperiod (60 μE) for another 3 days. The seedlings were grafted using silicon tubing (0.3 mm internal diameter) as the collar, and grafted seedlings were kept on identical sterile agar plates and under the same growth conditions for another 6–8 days until the graft junction had healed. Successfully grafted plantlets (approximately 14 days old) were transferred to a hydroponic culture system with half-strength nutrient solution containing 1.5 mm or no Pi (Scheible et al., 2004), and grown for another 3 weeks before harvesting of leaf and root material into liquid nitrogen. This material was used for analysis of gene expression and Pi content. Special care was taken to avoid any plants with adventitious roots.
Pi levels were measured using a colorimetric micromethod (Itaya and Ui, 1966). Discs were taken from leaves of the same age using a 6 mm cork borer, and individually ground in 50 μl distilled water using plastic rods (Sarstedt, http://www.sarstedt.com) in 96-well plates (Greiner Bio-One, http://www.greinerbioone.com). Extracts were diluted with water to 125 μl, and then the microtiter plate was centrifuged at 3000 gfor 4 min. Aliquots (10 μl) of the supernatant were mixed with 15 μl water, and 100 μl 1 m HCl and 100 μl color reagent (one volume of 4.2% (NH4)6Mo7O24·H2O in 5 N HCl, three volumes of 0.2% malachite green dye in water) were added. The plate was incubated at room temperature for 15 min before adding 100 μl 1.5% Tween-20. The absorbance at 660 nm was measured another 15 min later. The Pi concentration in the samples was determined against a calibration curve.
RNA isolation, cDNA preparation, and quantitative real-time PCR
RNA isolation of total RNA from Brassica napus and Curcurbita maxima was achieved using Trizol reagent, and isolation of RNA from roots and shoots of micrografted Arabidopsis plants was performed using RNeasy mini kits (Qiagen; http://www.qiagen.com). cDNA preparation and quantitative real-time PCR were performed using the primer pairs previously described (Bari et al., 2006; Czechowski et al., 2004, 2005). cDNA synthesis for the detection of miR399d precursors/primary transcript fragments was primed using 5′- GGCAAATCTCCTTTGGCAGAG-3′ (RT2 in Figure 5). Typically, approximately 5–10 ng of cDNA, the amount produced from the RNA of several thousand cells (Czechowski et al., 2004), was used per assay, yielding a CT value of 16–17 for UBQ10 (see Table S1), which is equivalent to 1–2 million single-stranded UBQ10 template copies in the PCR reaction vessel.
For quantification of mature miR399 and miR164, total RNA was reverse-transcribed using stem-loop primers (5′-GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACCGGGGCAAA-3′ and 5′-GTCGTATCCAGTGCAGGGTCCGAzGGTATTCGCACTGGAT ACGACCGCACG-3′, respectively) and MultiScribe reverse transcriptase (Applied Biosystems; http://www.appliedbiosystems.com/) (Chen et al., 2005). In brief, 1 μg total RNA was mixed with 1 μl 10 mm dNTPs, 1 μl 2.5 μm miRNA-specific stem-loop RT primer and RNase-free water to a final volume of 36.5 μl, heated to 65°C for 5 min, and then chilled on ice. Then 10 μl of 5x first-strand buffer, 2 μl 0.1 m DTT, 0.5 μl RNase inhibitor and 1 μl MultiScribe reverse transcriptase were added. The reaction mixture was incubated for 30 min at 16°C, followed by 30 min at 42°C and 5 min at 85°C, and then held at 4°C until quantitative real-time PCR amplification using primers miR399fwd (5′-CGACGTGCCAAAGGAGATTTG-3′), miR164fwd (5′-CACGTGGAGAAGCAG GGCA-3′) and miRrev (5′-CCAGTGCAGGGTCCGAGGT-3′). Rather than TaqMan probes (Chen et al., 2005), SYBR® Green (Applied Biosystems; http://www.appliedbiosystems.com) was used as the fluorescence dye for real-time monitoring of DNA amplification.
Primer efficiency (E) was determined for all amplicons using the freeware program LinRegPCR (Ramakers et al., 2003). The value did not change significantly between different cDNA samples, and was always higher than 0.80.
RNA gel blot analysis of microRNA species
Northern analysis was performed as described by Bari et al. (2006). Aliquots (20 μg) of total RNA were resolved on denaturing 17% polyacrylamide gels containing 7 m urea in 0.5x Tris-Borate/EDTA. RNA was blotted to positively charged nylon membranes (Biodyne B, Pall Europe Ltd; http://www.pall.com/) using a semi-dry transfer cell (Bio-Rad; http://www.bio-rad.com/) and auto-crosslinked at 0.12 mJ in a Stratalinker 1800 (Stratagene; http://www.stratagene.com/). The membrane was hybridized at 42°C with 32P end-labeled probes (5′-TTACAGGGCAAATCTCCTTTGGCA-3′ and 5′-ATGCAGCATCATCAAGATTCT-3′) complementary to miR399 and miR172, respectively, and visualized using a phosphorimager BAS-1800 II (Fuji Photo Film Co. Ltd; http://www.fujifilm.com/).
Materials not commercially available and used in the experiments reported will be made available for non-profit research on request.
We thank Professor Mark Stitt (Max-Planck Institute for Molecular Plant Physiology) for his comments on the manuscript, and Dr Tzyy-Jen Chiou (Academia Sinica, Taipei, Taiwan) for discussions on the results. The work was supported by the Max-Planck Society.