Distinct sensitivities to phosphate deprivation suggest that RGF peptides play disparate roles in Arabidopsis thaliana root development

Summary

• Growing agricultural demands in the face of impending phosphate (Pi) shortages underscore a need for a better understanding of plant development under conditions of Pi-deprivation. Pi is an essential nutrient that is a major component of fertilizer. Plants have evolved strategies to improve acquisition of this nutrient by altering root development under shortage conditions. We show that signaling peptides thought to act redundantly in Arabidopsis thaliana development have distinct functions in response to Pi deprivation.

• Using microscopy and confocal imaging, roots were analyzed for growth rate and cellular composition. Using expression microarrays, genes influencing development in response to phosphate deprivation were identified.

• ROOT GROWTH FACTOR1 (RGF1) and RGF2 influenced different aspects of root development under conditions of Pi-deprivation. We found that RGF2 influenced the longitudinal growth rate in the primary root in response to Pi-deprivation, whereas RGF1 affected circumferential cell number in the root meristem.

• These data suggest that mechanisms controlling adaptive development can depend on disparate functions of genes thought to act redundantly thus elucidating new functions for important developmental regulators.

Introduction

Plants cannot move when conditions become unfavorable, so they have evolved a plastic developmental program that facilitates access to water, nutrients, and light. Acquisition of the essential nutrient, inorganic phosphate (Pi), is particularly challenging. Because Pi is taken in at the root–soil interface, conditions of low availability are managed by modulating root development. By slowing primary root growth, increasing circumferential cell number, and increasing lateral root and root hair outgrowth, plants are thought to increase Pi-acquisition potential (Ma et al., 2001; Lopez-Bucio et al., 2002; Sanchez-Calderon et al., 2005). Mechanisms underlying these developmental changes remain elusive. A clearer understanding should enable engineering or breeding of plants that access and use Pi more efficiently. Furthermore, decreasing waste of this nonrenewable resource is of paramount importance due to the high fiscal and environmental costs associated with excess Pi-fertilizer use (Elser & Bennett, 2011).

To identify genes that control responses to Pi availability, many groups have paired transcriptomics with isolation of root, shoot or whole plant tissues (Hammond et al., 2003; Wu et al., 2003; Misson et al., 2005; Muller et al., 2007). Roles for transcription factors of several families, a SUMO-ligase, secreted phosphatases, high-affinity transporters, and a vesicular trafficking protein have been demonstrated, but with little understanding as to which tissues or cell types the activities occur in (Miller et al., 2001; Miura et al., 2005; Bari et al., 2006; Bustos et al., 2010; Bayle et al., 2011; Wang et al., 2011). Due to homogenization of many tissues in these studies, information about localization of expression was lost. Additionally, dilution of transcripts over entire organs limits identification of genes expressed within specific tissues. Therefore, little agreement exists between these studies, demonstrating that transcriptomic responses vary widely depending on growth conditions (e.g. hydroponics vs solid media), plant age at transfer, low Pi-exposure timing, and lighting conditions (i.e. due to carbon fixation). Types of tissues profiled will affect interpretation of data as well. In fact, the tissues assayed included whole plant or whole roots and leaves for expression analysis, leaving open questions about responses within individual tissues. All of these factors have led to a poor understanding of underlying responses to Pi-deprivation, leaving many open mechanistic questions.

While little tissue-specific information is available, morphological data have identified the root tip as the location of developmental reprogramming with respect to primary root growth under conditions of Pi deprivation (Svistoonoff et al., 2007). In this study, we paired transcriptional profiling with isolation of specific tissues in the root tip to clarify our understanding of how development is modulated in response to Pi-deprivation.

One class of genes known to modulate development is a family of secreted peptides, which act as intercellular signaling molecules. Several members of the ROOT GROWTH FACTOR/GOLVEN/CLE-like (RGF/GLV/CLEL) family of plant peptides have been identified as modulators of root growth. RGF1/GLV11, RGF2/GLV5, and RGF3/GLV7 were found to act redundantly, following post-translational tyrosine sulfation, to promote longitudinal growth of the root apical meristem via modulation of PLETHORA transcription factor expression (Matzusaki et al., 2010). Additionally, Fernandez et al. (2013) found that each of these genes repressed lateral root outgrowth when individually overexpressed. Due to the phenotypic similarity between mutants for these genes and that of roots exposed to low Pi conditions, we hypothesized that these genes modulate root development in response to low Pi. Our findings indicate that meristem morphology depends on Pi availability, and that novel roles for RGFs include developmental modulation of meristematic epidermis and cortex tissue. Furthermore, knock-out mutants of different RGF peptides exhibit different sensitivities to low Pi exposure indicating differential use of these signaling molecules in response to low Pi.

Materials and Methods

Plant growth conditions

Arabidopsis thaliana (L.) Heynh. seeds were surface sterilized for 1–3 h in an air-tight chamber with a reaction mixture of 60 ml bleach and 2 ml 12 M HCl. Seeds were allowed to imbibe in the dark for 1–3 nights at 4°C. After sterile plating on 1× MS, 1% sucrose, 1% agar media (control plates; Caisson catalog #MSP01, BD Difco catalog #214510) and sealing with porous tape, seedlings grew vertically on square agar plates in a Percival growth chamber set for 16 h : 8 h, day : night unless otherwise specified.

Root length measurements and circumferential cell quantification

Seedlings were transferred to control or 1× low Pi MS (low Pi plates; Caisson catalog #MSP11, 1% sucrose, 1% agar) plates at 5 d post imbibition. Root tips were marked daily and plates scanned after 4 d post transfer (DPT). Roots were measured in mm using ImageJ software. For circumferential cell quantification, transverse optical sections were obtained and cells were counted around the root circumference for each tissue type. Student’s t-tests were performed for all statistical comparisons.

Confocal microscopy and RootArray imaging

Using a Zeiss 510 upright confocal microscope, live roots were stained with 10 ug ml⁻¹ propidium iodide in water for 2 min and imaged for medial and longitudinal optical sections of the root apical meristem in a plane positioned 5–6 cortex cells above the quiescent center (QC).

The RootArray platform was used for whole meristem, three-dimensional imaging. pCO2:YFP-H2A/Col-0 seeds (Heidstra et al., 2004) were soaked for 2–3 nights in water at 4°C, and sown and sterilized within a microfluidic growth chamber called the RootArray (Busch et al., 2012). Grown under continuous light at room temperature, seedlings grew for 5 d in liquid 1× MS, 1% sucrose media before switching to 1× MS without Pi, 1% sucrose. Roots were imaged over time at an interval of 1–1.5 h following addition of 10 ug ml⁻¹ propidium iodide to growth media.

High resolution microarray for transcriptional profiling

Green fluorescent protein (GFP) reporter line seeds were surface sterilized with 50% bleach, 10% Triton-X100 for 10 min and 70% ethanol for 1 min. Seeds were sown on control media with mesh in two rows with a depth of 1 seed and a total of 60–80 seeds per plate, and sealed with porous tape. Seedlings were grown under 16 h : 8 h, day : night or continuous light. Seedlings were transferred to control or low Pi plates at 5 d post imbibition, and exposed for 24 (control and low Pi), 36 (low Pi), and 48 (low Pi) h before tissue collection. Cells of root meristems were enriched by cutting c. 1–2 mm from the root tip and enzymatically digesting root tip tissue to yield protoplasts. Then, fluorescence activated cell-sorting (FAC-sorting) yielded GFP-marker protoplasts for lines marking the cortex (pCO2:YFP) and epidermis (pWER:GFP) cell lineages. Following preparation of total RNA (Qiagen Micro RNeasy kit, Valencia, CA, USA), RNA was precipitated with 0.5 M ammonium acetate, 50 ug ml⁻¹ Glycoblue, and 2 volumes of 100% ethanol for purification. Then, transcripts were prepared for expression microarrays using the Affymetrix 3’ IVT Expression system and no less than 8.5 ug aRNA was hybridized to ATH1 microarrays. Three replicate experiments were carried out on separate days (except for sample CO2, which had two replicates due to poor RNA yield on the third sample). Global normalization and mixed-model ANOVA statistical analyses were performed as described (Levesque et al., 2006).

Quantitative reverse transcription (qRT)-PCR

Purified RNA (500 ng) that was left-over from cell-sorted root meristems was used for qRT-PCR using the Applied Biosystems High-Capacity cDNA Reverse Transcription kit (Catalog No. 4368814; Grand Island, NY, USA) and Sybr Green Real-Time PCR Master Mix (Life Technologies, Carlsbad, CA, USA; Catalog No. 4367659). Each experiment was performed in triplicate to correct for experimental error (technical replicates). Three biological replicates were run for each sample and relative expression values were found using the ΔΔCt method.

Peptide treatment

Seedlings were transferred to low Pi or low Pi + RGF1 MS media at 5 d post imbibition. Low Pi + RGF1 media contained a final concentration of 10 nM RGF1 peptide (diluted in water). Seedlings grew vertically for up to 4 d on a 16 h : 8 h, day : night cycle. Due to occasional root growth away from media surface, only roots lying flat were included in the analyses. The Student’s t-test was used for statistical comparisons between wild type and each mutant and no less than 20 seedlings per time point were included in the analysis.

Results

Root meristem morphology is altered in response to phosphate deprivation

To elucidate mechanisms controlling root development under low Pi, we monitored developmental responses with high spatial and temporal resolution. Using transfer experiments (see the Materials and Methods section) we found that primary root growth rate decreased within three days of exposure to low P_i (Fig. 1a). Reduction in growth rate continued after 4 and 5 d, ultimately leading to growth arrest (not shown). Reasoning that developmental events responsible for these changes must begin before 3 d of exposure, we focused on this early time-frame and found that aspects of meristem morphology are affected by Pi-deprivation. Within 2 d of exposure, we observed changes in the cortex tissue. This layer of cells changed from a V-shape to a U-shape when observed in an optical median longitudinal section, suggesting a broadening of the root meristem (Fig. 1b). Upon further examination, we found that this feature was attributable to increased radial cell divisions (Fig. 1c). By 2 d after transfer, roots on low Pi exhibited a greater number of cells around the central axis in the epidermis, cortex, and endodermis (Fig. 1d). Similar morphological changes had been reported within differentiating root tissues, but not within the undifferentiated root meristem (Ma et al., 2002). Thus, these cell divisions mark some of the earliest observable changes to development in response to Pi-deprivation.

Fig. 1.

Phosphate (Pi) deprivation modulates Arabidopsis thaliana primary root growth and meristem morphology. (a)Wild type seedlings were transferred from control to low Pi media at 5 d post imbibition. Primary root growth slows by 3 d post transfer (DPT). Error bars indicate ± SD. (b) Longitudinal medial optical sections of meristem show increasing disorganization over time (DPT). Cortex tissue layers (C) change from V-shaped in control conditions to U-shaped by 2 DPT. (c) Transverse optical sections show increase in circumferential epidermis (ep), cortex (c), endodermis (en), and pericycle (p) cell numbers (numbers represent DPT). (d) Chart summarizing change in circumferential cell number after 1 (dark gray) and 2 (light gray) DPT for several tissue layers. (e) Reconstructed image of pCO2:YFP-H2A reporter line collected 40 h after Pi restriction on the RootArray imaging platform. White arrows highlight bifurcation points for additional files of cortex cell nuclei. #, P < 0.001; *, P < 0.05.

To determine the location of these divisions within the meristem we used the RootArray imaging platform and visualized roots growing over time in Pi-free conditions (Busch et al., 2012). Observation of growth of the pCO2:H2A-YFP reporter line, which marks the cortex lineage, revealed files of ectopically divided cortex cells throughout the meristem by 40 h of exposure to Pi-free media (Fig. 1e; Heidstra et al., 2004). Interestingly, these ‘files’ of divided cortex cells were not always contiguous suggesting that these divisions may be stochastically controlled, or that the local Pi concentration may differ around the meristem.

Under normal growth conditions, root hair specification depends on interactions between epidermal cells and underlying cortical cell junctions (Kwak & Schiefelbein, 2007). A positional cue specific to the junction between cortical cells is thought to provide the hair cell fate signal, resulting in an alternating pattern around the circumference of the root. SCRAMBLED (SCM), which encodes a leucine-rich-repeat receptor-like kinase (LRR-RLK), is expressed throughout the developing root, and may transduce this hypothetical signal to the overlaying epidermal tissue (Kwak et al., 2008). Furthermore, roots have been found to increase the number of cortical and root hair cell files within the differentiation zone in response to low Pi (Ma et al., 2002). Although root hairs can emerge independently of underlying cortex cell junctions under conditions of Pi-deprivation (Muller & Schmidt, 2004; Supporting Information Fig. S1b), we hypothesized that cortical cells in the meristem divide radially as a means of generating additional sites for root hair formation. Using the root-hair marker line pCOBL9:GFP, we found that roots exposed to low Pi for at least 3 d produced extra root hairs, with most hairs overlying cortex cell junctions (Fig. S1c). This result supports a model in which radial divisions in the meristematic cortex occurring in response to low Pi lead to the formation of additional root hairs, which aid in Pi acquisition.

In addition to ectopic divisions in the meristem, we observed disorganization of the QC and surrounding stem cell niche upon transfer to low Pi media. Disorganization of the niche began by 2 DPT and increased in severity over time, with the QC becoming difficult to recognize after 4 d on low P_i (Fig. S2a,b). Interestingly, we observed expansion of reporter expression for markers of the QC by 4 DPT, suggesting that low P_i induces expansion of the QC (Fig. S2a). The proximal causes of these divisions are unclear, but they do appear to involve redistribution of auxin. Using the auxin reporter pDR5:GFP as a proxy for QC size, we found that QC expansion precedes the slowing of root growth, occurring by 2 DPT (Fig. S2b). These data support a model that includes QC expansion concurrent with radial division of meristematic epidermis, cortex, and endodermis cells, followed by inhibition of primary root growth under conditions of Pi-deprivation.

High-resolution expression analysis identifies genes coregulated in meristematic cortex and epidermis

Using confocal microscopy and live imaging with the RootArray platform, we found that meristematic cortex cells divide radially in response to low Pi by 40 h after transfer. To identify regulators of these divisions we performed a transcriptomic analysis of meristematic cortex following low Pi exposure. We performed a similar analysis for meristematic epidermis, as it lies directly between the environment and the cortex. We used GFP reporters and cell sorting to isolate cortex (pCO2:H2A:YFP; Heidstra et al., 2004) and epidermis cells (pWER:GFP; Lee & Schiefelbein, 2002) from root meristems following 24, 36, and 48 h of low Pi exposure and performed expression microarray analysis. Reporter expression within these lines did not change in response to low Pi within the time-frame profiled (data not shown). For cortex tissue, 197 genes were induced at least two-fold between 24–48 h of exposure (Table S1) and 54 genes were induced two-fold within epidermis tissue (Table S2). Reasoning that transcripts induced in both tissues might be particularly informative, we intersected the two gene lists and identified six genes (Fig. 2a,b).

Fig. 2.

Tissue-specific expression modulation within the Arabidopsis thaliana meristem reveals several candidate regulators of cortex and epidermis anticlinal divisions occurring in response to phosphate (Pi)-deprivation. (a) Venn diagram showing total number of genes induced within each tissue by two-fold or greater. (b) Normalized expression over time for each of six genes induced in cortex (diamonds) and epidermis (squares). (c) Quantitative reverse transcription (qRT)-PCR reveals that RGF1, RGF2, and RGF3 are induced in epidermis at each time-point and, to a lesser degree, after 48 h in cortex tissue.

As we had tightly controlled low Pi exposure times and tissues isolated, we thought this group of genes might play a role regulating circumferential divisions. Among the six genes were two that had previously been identified as phosphate-responsive, PHOSPHATE STARVATION INDUCED-3 (At3g47420) and SPX domain gene-1 (At5g20150), there were two associated with root hair development and differentiation, ROOT HAIR SPECIFIC-7 (At1g54970) and XYLOGLUCAN ENDOTRANSGLUCOSYNTHASE/HYDROLASE-12 (At5g57530), a chitinase-family gene (At1g56680), and a root meristem growth factor, ROOT MERISTEM GROWTH FACTOR 3 (RGF3; Wang et al., 2004; Miura et al., 2005; Bari et al., 2006; Won et al., 2009; Diet et al., 2009). Due to a previously described role in meristem development, we focused on RGF3 (At2g04025). This transcript showed two-fold induction in the cortex after 24 h and after 36 h in the epidermis. RGF3 was previously shown to act redundantly with RGF1 and RGF2 to influence meristem length (Matsuzaki et al., 2010). Because RGF1 and RGF2 are not represented on the ATH1 microarray, we used qRT-PCR and found that both genes are induced in epidermis after 24, 36, and 48 h of Pi-deprivation, as well as by 48 h in cortex tissue (Fig. 2c). Due to limited sample availability, cortex samples for 24 and 36 h exposures were not tested for RGF1 or RGF2 transcript abundances.

RGF mutants exhibit distinct sensitivities in response to Pi-deprivation

RGF2 modulates longitudinal root growth

Under normal nutrient conditions, Matsuzaki et al. (2010) found that rgf1rgf2rgf3 (Salk_132484, Salk_145834, Salk_053439) triple mutant meristems were significantly shorter than wild type. Interestingly, this phenotype was rescued by exogenous application of tyrosine-sulfated RGF1 peptide, however single mutants of rgf1, rgf2 or rgf3 had no significant difference in meristem length. Because we found that RGF1, RGF2, and RGF3 transcripts are induced in the meristem by 48 h of exposure to low Pi, we hypothesized that mutants in these genes would be sensitive to Pi-deprivation. Using primary root growth rate as a proxy for sensitivity, we found that rgf2-1 (Salk_145834 – null mutant with T-DNA insertion in first intron) grew significantly slower than wild type after 1, 2, 3, and 4 d on low Pi media (P < 0.001; Fig. 3a,b). Comparisons between rgf2-1 and rgf1-1 (Salk_132484 – null mutant with T-DNA insertion in promoter), rgf3-1 (Salk_053439 – null mutant with T-DNA insertion in first intron), or with the triple mutant, rgf1rgf2rgf3, showed significant slowing of root growth in rgf2-1 over time, although no difference in root length was observed at the time of transfer (Fig. 3a; Table S3). Furthermore, exogenous addition of RGF1 peptide did not rescue the rgf2-1 growth phenotype (Fig. 3b). Taken together, these data indicate that RGF2 promotes longitudinal root growth under conditions of Pi-deprivation, whereas RGF1 and RGF3 do not. Therefore, RGF1, RGF2, and RGF3 do not act redundantly to modulate root length under conditions of Pi-deprivation. Interestingly, no phenotype was observed for the triple mutant suggesting that these gene products may interact epistatically to control root length under these conditions.

Fig. 3.

Arabidopsis thaliana root lengths measured over time reveal that (a) rgf2 is hypersensitive to low phosphate (Pi). (b) Unlike rgf1 mutants, exogenous addition of 10 nM RGF1 peptide does not return rgf2 mutants to wild type growth rates. Error bars represent ± SD. #, P < 0.001; *, P < 0.05. DPT, d post transfer.

RGF1 modulates circumferential cell number

Because RGFs are induced in response to low Pi within the cortex and epidermis, we hypothesized that these genes may regulate circumferential cell number in response to low Pi. Compared with wild type, rgf1-1 exhibits hypersensitivity to low Pi with respect to circumferential cortex and epidermis cell quantities (Figs 4a, S3). Conversely, rgf2-1 mutants showed fewer circumferential divisions compared with wild type in response to low Pi. These results indicate that different roles exist for each peptide in modulating circumferential meristem cell number (Figs 4B, S3, and see later Fig. 6). rgf3-1 did not show significant differences in cell number for cortex, epidermis, or endodermis layers when compared with wild-type roots after three and four days (Figs 4c, S3). Additionally, triple mutants exhibited a cortex cell number ratio of 1.2 (P < 0.01) after 3 d on low Pi media. This is similar to the rgf1-1 phenotype, but aberrant cortex cell ratios in the triple mutant do not persist after 4 d as occurs with the single mutant. This suggests that the relationship between RGF1, RGF2, and RGF3 is additive and not epistatic or redundant.

Fig. 4.

Ratios of mutant to wild type circumferential cell numbers reveals distinct sensitivities to phosphate (Pi)-deprivation in Arabidopsis thaliana. (a) rgf1 mutants exhibit an increase in cell numbers by 3 d post transfer (DPT), whereas (b) rgf2 mutants show a decrease. (c) rgf3 mutants show a slight increase in epidermis cell number by 2 DPT, but show wild type-like sensitivity over time. (d) Triple mutants for the RGFs look similar to rgf1 mutants at 3 DPT, but are like wild type after 4 d suggesting an additivity in phenotype. #, P < 0.001; **, P < 0.01; *, P < 0.05.

Fig. 6.

Schematic model demonstrating roles for RGF1 and RGF2 in Arabidopsis thaliana in low phosphate (Pi) conditions. In response to low Pi, RGF2 promotes vertical root growth and radial divisions in epidermis, cortex, and endodermis (yellow), while RGF1 represses radial divisions in these tissues.

To further distinguish between RGF1 and RGF2, we attempted to rescue the rgf2-1 mutant phenotype with the addition of exogenous RGF1 peptide (Matsuzaki et al. 2010). After 3 and 4 d on low Pi media supplemented with 10 nM RGF1 peptide, rgf1-1 mutants had similar cell numbers to wild type on low Pi media (Figs 5a, S3). However, rgf2-1 mutants did not return to wild-type cell numbers in the presence of RGF1 peptide (Figs 5b, S3). Taken together, these data indicate that exogenous RGF1 peptide cannot rescue rgf2-1 mutants, further supporting disparate roles for these peptides in root development under conditions of Pi-deprivation. Interestingly, triple mutant cell numbers do not return to wild type either upon exposure to RGF1 peptide. This further supports a nonepistatic relationship among these genes.

Fig. 5.

Exogenous RGF1 peptide does not return rgf2 mutants to wild type-like sensitivity in Arabidopsis thaliana. (a) rgf1 mutants show a return to wild type cell numbers in conditions of low phosphate (Pi). (b) rgf2 shows significantly fewer cells compared with wild type after 4 d on low Pi. (c) rgf3 shows no significant change in cell numbers. (d) Triple mutants do not return to wild type cell numbers. #, P < 0.001; **, P < 0.01; * P < 0.05.

Discussion

Several features of root development are modulated in response to Pi-deprivation to improve nutrient acquisition potential. With a clearer understanding of genetic and cellular mechanisms that underlie these changes, we may be able to design plants that use Pi more efficiently. To understand mechanisms operating within the growing meristem, we set out to describe changes to morphology with cell-type resolution. Using reporter lines, we showed that cell number around the central root axis is increased via radial divisions of the epidermis, cortex, and endodermis. A similar increase in circumferential cortex number had been observed previously within differentiating tissues and was suggested to result in increased root hair density (Ma et al., 2001). Furthermore, the timing of the start of these divisions precedes slowing of primary root growth, QC disorganization, and expansion of QC-marker expression that occurs in response to low Pi. These changes exemplify a shift from indeterminate to determinate growth within the root, following a model that has been suggested for root response to low Pi (Sanchez-Calderon et al., 2005). Although the timing of radial divisions depends on light conditions, increases in root hair number were found to follow these divisions. These data suggest that additional cortex cell divisions, following deprivation of Pi, results in signaling between the dividing cortex cell layer and overlying epidermis tissue. Therefore, plants use an unknown mechanism to interpret the availability of Pi, then change tissue layer morphology by inducing radial divisions in the cortex and other tissues, which likely feeds signals into a mechanism that already exists for root hair specification (Kwak & Schiefelbein, 2007). If this model is correct, plants reduce the energy cost of adapting development for acquisition of nutrients in low Pi conditions by modulating existing signaling networks.

Transcriptomic studies have identified many Pi-responsive genes, however little agreement exists between these datasets. Here, we aimed to elucidate processes at cellular resolution by performing expression microarray analyses on specific root tissues following low Pi exposure with fine temporal resolution. We hypothesized that by profiling meristematic epidermis and cortex tissues specifically, we could identify genes that control radial divisions of these tissues in response to low Pi. Our analysis revealed six genes that were induced within dividing tissues around the time of meristematic radial divisions. Interestingly, three of these genes had previously been associated with Pi transport, which fits with our understanding that roots respond by upregulating processes that will increase Pi uptake and/or transport throughout the plant. Another two of the six genes identified have been linked to root hair outgrowth, which is also consistent with our model as these structures are involved in nutrient uptake. The sixth gene, RGF3, had previously been linked to root meristem growth, making this a candidate regulator. RGF3 is a secreted signaling peptide that was shown to act redundantly with RGF1 and RGF2 to modulate meristem length (Matsuzaki et al., 2010). After finding that these transcripts were coinduced with respect to time and tissue-type in our study, we hypothesized that these signaling peptides influenced radial divisions under low Pi conditions. Interestingly, individual mutant phenotypes for each of these three genes were found to differ, suggesting that these peptides have unique functions in response to low Pi. rgf2-1 was found to be hypersensitive to deprivation of Pi, while rgf1-1, rgf3-1, and rgf1rgf2rgf3 were not. Furthermore, addition of exogenous RGF1 peptide did not rescue the phenotypes observed for rgf2-1. Taken together, these data suggest that these genes act redundantly to influence root meristem length when plants are in nutrient replete conditions, but have distinct roles in meristem growth under low Pi conditions (Fig. 6). Although further study with double mutants is needed, these data may indicate that additive, not epistatic, interactions exist between these gene products under these conditions. However, there is a caveat that while these insertional mutants have been characterized as null alleles under normal growth conditions, it is possible that under low phosphate conditions they might have varying strengths.

Understanding mechanisms of plant development in response to Pi-deprivation is of paramount importance when considering food and fuel demands of a growing, and increasingly industrialized, world population. Here, we have shown how molecular mechanisms involved in root patterning and morphology are linked in unexpected ways to modulate root development in response to nutrient availability. This work not only contributes to our understanding of gene-by-environment interactions, but may provide a foundation for design of crop species with optimized phosphate-use efficiency.