MULTIPASS, a rice R2R3-type MYB transcription factor, regulates adaptive growth by integrating multiple hormonal pathways


  • Romy Schmidt,

    1. Institute of Biochemistry and Biology, University of Potsdam, Potsdam, Germany
    2. Max Planck Institute of Molecular Plant Physiology, Potsdam, Germany
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Jos H. M. Schippers,

    Corresponding author
    1. Max Planck Institute of Molecular Plant Physiology, Potsdam, Germany
    • Institute of Biochemistry and Biology, University of Potsdam, Potsdam, Germany
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Delphine Mieulet,

    1. Centre de Coopération Internationale en Recherche Agronomique pour le Développement, Unité Mixte de Recherche Genetic Improvement and Adaptation of Mediterranean and Tropical Plants, Montpellier Cedex 5, France
    Search for more papers by this author
  • Toshihiro Obata,

    1. Max Planck Institute of Molecular Plant Physiology, Potsdam, Germany
    Search for more papers by this author
  • Alisdair R. Fernie,

    1. Max Planck Institute of Molecular Plant Physiology, Potsdam, Germany
    Search for more papers by this author
  • Emmanuel Guiderdoni,

    1. Centre de Coopération Internationale en Recherche Agronomique pour le Développement, Unité Mixte de Recherche Genetic Improvement and Adaptation of Mediterranean and Tropical Plants, Montpellier Cedex 5, France
    Search for more papers by this author
  • Bernd Mueller-Roeber

    Corresponding author
    1. Max Planck Institute of Molecular Plant Physiology, Potsdam, Germany
    • Institute of Biochemistry and Biology, University of Potsdam, Potsdam, Germany
    Search for more papers by this author

For correspondence (e-mails;


Growth regulation is an important aspect of plant adaptation during environmental perturbations. Here, the role of MULTIPASS (OsMPS), an R2R3-type MYB transcription factor of rice, was explored. OsMPS is induced by salt stress and expressed in vegetative and reproductive tissues. Over-expression of OsMPS reduces growth under non-stress conditions, while knockdown plants display increased biomass. OsMPS expression is induced by abscisic acid and cytokinin, but is repressed by auxin, gibberellin and brassinolide. Growth retardation caused by OsMPS over-expression is partially restored by auxin application. Expression profiling revealed that OsMPS negatively regulates the expression of EXPANSIN (EXP) and cell-wall biosynthesis as well as phytohormone signaling genes. Furthermore, the expression of OsMPS-dependent genes is regulated by auxin, cytokinin and abscisic acid. Moreover, we show that OsMPS is a direct upstream regulator of OsEXPA4, OsEXPA8, OsEXPB2, OsEXPB3, OsEXPB6 and the endoglucanase genes OsGLU5 and OsGLU14. The multiple responses of OsMPS and its target genes to various hormones suggest an integrative function of OsMPS in the cross-talk between phytohormones and the environment to regulate adaptive growth.


Plant hormones shape plants by modulating growth in response to endogenous and environmental cues (Wolters and Jürgens, 2009). Adaptation to environmental conditions occurs largely through hormone homeostasis (Achard et al., 2006; Carabelli et al., 2007; Park et al., 2007). Major developmental growth regulators include auxin, brassinolide (BL), cytokinin (CK) and gibberellic acid (GA), whereas abscisic acid (ABA) and ethylene are often implicated in stress responses (Wolters and Jürgens, 2009). Under unfavorable conditions, plants produce stress hormones to promote survival. These hormones are essential for modulating the redistribution of resources from growth to plant protection or defense (Wang et al., 2007; Wilkinson et al., 2012).

Auxin and GA are positive plant growth regulators that may be counteracted by stress hormones such as ABA (Park et al., 2007; Yaish et al., 2010). ABA synthesis is induced by salt stress in rice (Oryza sativa) (Moons et al., 1995), with the level and duration correlating with salt tolerance. In Arabidopsis thaliana, over-expression of a GH3 enzyme gene involved in auxin inactivation results in reduced levels of free auxin and increased stress tolerance (Park et al., 2007). During abiotic stress, expression of various GH3 genes is modulated by ABA. A connection between auxin and ABA has also been implied for the rice transcription factor (TF) ABI5-like1 (ABL1), which controls stress-responsive genes through ABA-responsive elements in their promoters. The abl1 mutant shows reduced ABA responsiveness but hypersensitivity to auxin (Yang et al., 2011). Another example of cross-talk is provided by the TF OsAP2–39, which is a positive regulator of ABA biosynthesis but also induces expression of ELONGATED UPPERMOST INTERNODE (EUI), encoding a GA-deactivating enzyme (Yaish et al., 2010). ABA also affects the expression of CK biosynthesis and breakdown genes in rice roots (Tsai et al., 2012). The interactions between phytohormones and their regulation at the genetic level are still poorly understood in rice, but have become a major focus in current crop breeding strategies (Wilkinson et al., 2012).

The adaptive growth response at the genetic level is executed by stress-responsive TFs. After the initial growth cessation upon salinity stress, the development of the root system is adapted to reach out for novel water sources. Expansins (EXPs) act as positive regulators of growth, and their expression is induced in elongating tissues (Cho and Kende, 1997; Lee and Kende, 2001). Based on phylogenetic analysis, EXPs are grouped into two related classes: α- and β–EXPs (Cosgrove, 2000). Grasses such as rice contain a far greater number of β–EXP genes than dicots such as Arabidopsis (Lee et al., 2001). Inducible over-expression of OsEXPA4 stimulates plant growth by increasing cell size, while knockdown diminishes plant growth (Choi et al., 2003). OsEXPB3 was shown to be localized to the primary cell wall of the vascular system (Lee and Choi, 2005). Phytohormones are implicated in the regulation of EXP gene expression, as cis-elements specific to GA, auxin, ABA and ethylene are commonly found in the promoters of most EXP genes (Lee et al., 2001). Not surprisingly, the expression of EXP genes is down-regulated during growth under saline conditions (Walia et al., 2005). However, the down-regulation of cell wall-modifying genes is transient. During drought stress, the plant modulates root growth by activating EXPs and xyloglucan:xyloglucosyl transferases in a spatio-temporal manner (Harb et al., 2010; Sasidharan et al., 2011). Thus, modulation of cell-wall extensibility during abiotic stress is an important aspect of the adaptive growth response.

Here we studied the molecular mechanism by which the R2R3-type MYB TF encoded by MULTIPASS (OsMPS) controls adaptive growth. Expression of OsMPS is repressed by the growth-promoting hormones auxin, GA and BL, and growth defects of OsMPS over-expression plants are partially rescued by auxin application. Furthermore, we showed that OsMPS is a direct regulator of genes encoding expansins, and propose that it functions as a hub for multiple hormones to control plant growth.


OsMPS is a salt-responsive R2R3-type MYB transcription factor gene from rice

OsMPS (Os02g40530) was identified in a screen for TF genes in rice roots (cv. Nipponbare) responding to short-term salt stress. OsMPS expression is induced after 30 min and 1 h of salt stress (Figure 1a). Up-regulation of OsMPS was also observed in leaves after salt stress (Figure S1). The OsMPS locus spans three exons that form an open reading frame of 870 nucleotides (Figure 1d,e). OsMPS is classified as an R2R3-type MYB TF belonging to clade C14 (Zhang et al., 2012a). This clade includes AtMYB71, AtMYB79 and AtMYB121 from Arabidopsis and two uncharacterized rice MYB TFs (Os04g42950 and Os03g04900). Using the web tool Phytozome (Goodstein et al., 2012), homologous proteins from Sorghum bicolor, Zea mays and Setaria italica were identified and shown to share high sequence identity with OsMPS at the DNA-binding domain (Figure 1f). OsMPS contains a putative nuclear localization signal immediately downstream of the R2R3 domain, and transient expression of the OsMPS coding sequence fused at its 3′ end to the CFP-encoding gene in rice protoplasts revealed a fluorescence signal in the nucleus (Figure 1c).

Figure 1.

OsMPS genomic structure and expression upon salt stress.
(a) Expression levels of OsMPS in roots after salt stress for 10, 30, 60 or 180 min relative to mock-treated roots. Asterisks indicate a significant difference compared with mock-treated roots (Student's test,  0.05). Values are means ± SD (= 3). FC, fold change.
(b) GUS activity in roots of 2-week-old OsMPS:GUS seedlings. Left, control; right, 100 mm NaCl for 2 h.
(c) Subcellular localization of the OsMPS–CFP fusion protein in rice protoplasts. Red, chloroplast autofluorescence; cyan, CFP.
(d) Protein structure of OsMPS containing the R2R3 DNA-binding domain (H1–H3). Additional helices are colored gray.
(e) Gene structure of OsMPS. Black boxes indicate exons and numbers indicate nucleotide positions.
(f) Sequence alignment of the R2R3 DNA-binding domain of OsMPS (Os02g40530) and homologous proteins in Osativa (Os04g42950, Os03g04900), Sbicolor (Sb04g02621), Zmays (GRMZM2g027), Sitalica (Si018007m) and Athaliana (AtMYB71/At3g24310, AtMYB79/At4g13480, AtMYB121/At3g30210). Helical structures (H1–H3) within the R2R3 domain are indicated. NLS, putative nuclear localization signal.

OsMPS is expressed in vegetative and reproductive tissues

The expression profile of OsMPS was analyzed using the PlaNet Browser (Mutwil et al., 2011). OsMPS is expressed in most tissues, including shoots, roots, anthers and seeds, but not in endosperm, stigma, ovary or the embryo (Figure S2). To examine the tissue-specific expression of OsMPS in planta, a 2 kb sequence directly upstream of the transcriptional start site was cloned and used to drive GUS expression. OsMPS expression was detected in almost all tissues examined (Figure 2). In reproductive tissues, GUS activity was present in the anthers and vascular tissue of the spikelet (Figure 2a,b). Furthermore, OsMPS is expressed in the vascular tissues of the coleoptile and young leaves (Figure 2c,d). GUS activity was detected in the main root and root hairs, but was absent from the growing zone (Figure 2e–g). GUS activity spreads throughout the leaf at the onset of senescence and is wound-inducible (Figure 2h,i). In agreement with the expression data for OsMPS, the roots showed increased GUS activity after 2 h of salt stress (Figure 1b).

Figure 2.

Expression pattern of OsMPS.
GUS activity was observed in (a) anther, (b) spikelet, (c) coleoptile, (d) young leaf, (e) root, (f) root hairs, (g) root elongation zone, (h) senescing leaf, and (i) wounded leaf.

OsMPS affects plant growth

To characterize the function of OsMPS in rice, over-expression (OE) and knockdown (KD) lines were generated. For each construct, at least 15 independent lines were established. Based on the expression level of OsMPS, several non-segregating lines (T3 generation) were selected for characterization, including OE1–2, OE2–5, OE6–2 and OE6–5 for over-expression, and KD4–5, KD6–3 and KD9–2 for knockdown (Figure S3).

Over-expression of OsMPS resulted in decreased plant stature at the seedling, vegetative and heading stage (Figure 3a–c). Shoots and roots of OE1–2, OE2–5 and OE6–2 seedlings were significantly shorter than those of empty-vector (EV) seedlings at both 4 and 7 days after sowing (Figure 3d,g), whereas KD6–3 and KD9-2 seedlings had longer roots at 4 or 7 days after sowing, respectively. Upon salt stress, both KD and OE lines showed a shorter shoot and/or root length compared to EV seedlings at 4 and/or 7 days after sowing (Figure 3e,h). As OsMPS over-expression plants were severely reduced in size under control conditions, we determined the relative growth reduction upon salt stress. Interestingly, relative shoot and/or root growth of OE1–2, OE2–5 and OE6–2 seedlings was significantly less affected by salt stress than in EV seedlings (Figure 3f,i). In contrast, KD lines showed a more pronounced relative reduction in shoot and/or root length under salt stress than the EV line.

Figure 3.

OsMPS expression levels affect plant growth.
(a–c) Growth phenotype of EV, OsMPS knockdown (KD) and over-expression (OE) lines at (a) 7 days, (b) 4 weeks and (c) heading stage.
(d, e) Shoot and root lengths of EV, OsMPS KD and OE lines (d) under control conditions and (e) salt stress (100 mm) at 4 DAS.
(f) Relative shoot and root length (% of unstressed values) under salt stress (100 mm NaCl) at 4 DAS.
(g, h) Shoot and root lengths of EV, OsMPS KD and OE lines under (g) control conditions and (h) salt stress (100 mm NaCl) at 7 DAS.
(i) Relative shoot and root lengths (% of unstressed values) under salt stress (100 mm NaCl) at 7 DAS.
(j, k) Fresh weight (FW) of 4-week-old EV and OsMPS transgenic plants under (j) control and (k) salt stress (100 mm NaCl; 7 days).
(l) Relative FW (% of unstressed values) under salt stress (100 mm NaCl; 7 days).
(m, n) Dry weight (DW) of 4-week-old EV and OsMPS transgenic plants under (m) control and (n) salt stress (100 mm NaCl; 7 days).
(o) Relative DW (% of unstressed values) under salt stress (100 mm NaCl; 7 days).
Data in (d)–(i) represent means ± SE (= 32). Data in (j)–(o) represent means ± SD (= 5; six plants each). Data in (f), (i), (l) and (o) are given relative to unstressed plants of the same genotype. Asterisks in all panels indicate significant differences compared with EV (Student's t test,  0.05). EV, empty vector; DAS, days after sowing.

At the age of 4 weeks, OE plants showed a reduced root size, while KD plants showed an enlarged root structure (Figure 3b). The fresh weight (FW) and dry weight (DW) of OE6–2 shoots and roots were reduced compared with EV plants, while KD4–5 and KD6–3 showed an increased weight for the root and/or shoot (Figure 3j,m). We examined the effect of OsMPS on biomass accumulation under salt stress using 4-week-old plants. Roots and/or shoots of both OE6–2 and KD4–5 displayed a reduced weight compared to the EV control line after 1 week of salt stress (Figure 3k,n). Similarly, KD6–3 plants showed reduced root FW after salt stress, but had a higher shoot FW than the EV line. However, the relative reduction in DW of roots and shoots observed in lines KD4–5 and KD6–3 was stronger than in stressed EV plants (Figure 3l,o). In contrast, the relative root biomass of stressed OE-6–2 plants was less reduced than in EV plants.

As OsMPS affects plant stature throughout its life cycle, we determined epidermal cell sizes of the 2nd leaf of the main tiller of 6-week-old plants. Cells of KD4–5 leaves at the tip and middle region were significantly larger and longer than those of EV leaves (Figure 4). In contrast, over-expression of OsMPS resulted in decreased cell size and length.

Figure 4.

OsMPS affects the cell size and length of leaf epidermal cells.
Cell sizes were determined for the basal, middle and tip regions of the uppermost leaf of 6-week-old OsMPS knockdown (KD4–5) and over-expression plants (OE6–2). Data represent means ± SE of six plants (26 cells each). Asterisks indicate significant differences compared with EV control plants (Student's t test,  0.05). EV, empty vector.

OsMPS mis-expression causes metabolic changes related to growth

Metabolism underlies plant development during both optimal and adverse environmental conditions (Stitt et al., 2010; Ribeiro et al., 2012). To determine the effect of OsMPS expression on metabolite levels, a GC–MS-based analysis of primary metabolites in roots and leaves of 4-week-old plants was performed.

With respect to amino acids, OE6–2 roots accumulated significantly higher levels of Pro, Gly, Ile, Lys, Phe, Thr, Tyr and Val than the EV line under control conditions (Table S1). Of note, Phe and Tyr are precursors for the production of phenylpropanoids and the subsequent generation of lignin (Gray et al., 2012). Interestingly, Asp was significantly decreased in OE6–2 roots but increased in KD4–5, while the levels of γ-aminobutyric acid (GABA) were down-regulated in both lines compared to EV roots (Table S1). After 24 h of salt stress, the levels of Ala, Asp, GABA, Glu, Gln, Lys and Thr significantly increased in EV roots compared to control conditions. In the KD4–5 and OE6–2 lines, Asp, GABA, Glu, and Gln increased compared to their levels under control conditions; however, Ala, Lys and Thr did not significantly change in the KD4–5 and OE6–2 lines after salt stress. Furthermore, the levels of Gly, Ile, Pro and Val, which are higher in OE6–2 roots under control conditions, decreased significantly upon salt stress, although not below EV levels (Table S1).

Similar to roots, most amino acids accumulated under control conditions in OE6–2 leaves (Table S2). In EV leaves, only Gly was decreased upon salt stress, whereas, in KD4–5 leaves, lower levels of Ala, GABA, Gly, His, Ile, Pro and Val were observed compared to non-stress conditions. Furthermore, up-regulation of Asp, Ser and Thr was found in OE6–2 leaves.

Roots of OE6–2 plants exhibited higher levels of glucose and trehalose under control conditions, whereas roots of both KD4–5 and OE6–2 had lower levels of raffinose and galactinol. Upon salt stress, raffinose decreased in all lines tested, while galactinol was lower in EV and KD4–5 roots after salt stress as compared to their respective controls. Furthermore, leaves of OE6–2 showed higher levels of most measured sugars compared to EV (Table S2). In leaves, a significant decrease in glucose, fructose, galactose and xylose for KD4–5 and OE6–2 was found as compared to EV plants, but the levels of these metabolites in OE6–2 did not decrease below those of the EV line, whereas this did occur for KD4–5.

Interestingly, in roots and/or shoots of OE6–2 plants, a low level of the tricarboxylic acid cycle intermediates 2–oxoglutarate, malate and fumarate was observed. Tricarboxylic acid cycle intermediates are directly linked to plant development as their abundance is a potential growth signal (van der Merwe et al., 2010; Finkemeier et al., 2013). Upon salt stress, a general increase in tricarboxylic acid cycle intermediates in the roots of all lines examined was observed. In leaves, only the levels of malate in KD4–5 and EV lines decreased significantly upon salt stress compared to their controls.

OsMPS affects heading date and panicle-related traits

OsMPS is expressed in anthers and spikelets (Figure 2a,b), suggesting a role in reproductive development. OE6–2 plants showed a delay in heading and flowering by 21 days, and a reduced number of panicles per plant compared to the EV line (Figure 5a–c). Down-regulation of OsMPS caused early heading and flowering (Figure 5a,b) but KD4–5 plants developed fewer panicles per plant (Figure 5c). On the other hand, altered OsMPS expression did not affect panicle length or the number of spikelets per panicle (Figure 5d,e) but caused a reduction in the number of grains per panicle and plant (Figure 5f,g). Furthermore, grains of OE1–2, OE6–2 and OE6–5 lines were significantly reduced in length and weight compared to EV grains, while those of KD4–5 and KD9-2 showed an increase in length (Figure 5i–k). In combination with the reduced panicle number, OE6–2 plants had a grain yield of 9% and KD4–5 plants had a grain yield of 72% relative to EV plants (Figure 5h).

Figure 5.

Effect of OsMPS on panicle-related traits.
(a) Panicle emergence, (b) flowering time, (c) panicle number and (d) panicle length of EV, OsMPS knockdown (KD4–5) and over-expression (OE6–2) plants.
(e) Florets per panicle, (f) number of grains per panicle, (g) number of grains per plant and (h) grain yield per plant of KD4–5 and OE6–2 relative to EV (= 15).
(i) Grain phenotype of EV (1), KD4–5 (2), KD9-2 (3), OE1–2 (4), OE6–2 (5) and OE6–5 (6) grains. Hulled seeds left and dehulled seeds right. Scale bar = 1 mm.
(j, k) Length (j) and weight (k) of OsMPS transgenic seeds relative to EV (= 50).
Asterisks indicate significant differences compared with EV plants (Student's t test, * 0.05, ** 0.001). EV, empty vector.

OsMPS expression is regulated by plant hormones

The involvement of plant hormones in the regulation of OsMPS expression was analyzed in silico and in vivo (Figure 6). The OsMPS promoter used for GUS analysis contains three ABA-responsive elements (ACGT[G/T]), one ARF-specific auxin-response element (AuxRE; TGTCTC) and one GA-responsive element (GARE; TAACAGA) (Figure 6a). ABA induced expression of OsMPS after 2 and 6 h of treatment, whereas auxin (naphthalene-1–acetic acid, NAA) and GA3 transiently repressed OsMPS expression in roots (Figure 6b). CK (6–benzylaminopurine) induced OsMPS expression at 2 and 6 h of treatment, and epi-brassinolide (epi-BL) transiently repressed it. After 6 h of treatment with auxin, GA or epi-BL, the expression of OsMPS was not significantly different from control treatments. These observations indicate that OsMPS acts downstream of multiple hormonal pathways.

Figure 6.

OsMPS expression is modulated by various hormones.
(a) Presence of elements responsive to GA (GARE), ABA (ABRE) and auxin (AuxRE) in the OsMPS promoter.
(b) Relative expression level of OsMPS upon auxin (NAA), GA3, BR (epi-BL), CK and ABA application in wild-type roots. Values are means ± SE (= 3). Asterisks indicate significant differences compared with mock-treated roots (Student's t test,  0.05). FC, fold change.
(c) Phenotype of control and 0.1 μm NAA-treated 7-day-old seedlings.
(d) Shoot and root lengths of 7-day-old over-expression (OE) seedlings grown on 0.1 μm NAA relative to mock-treated seedlings of the same genotype. Values are means ± SE (= 16).
(e) Shoot and root lengths of 7-day-old EV, OsMPS OE and knockdown (KD) seedlings treated with 1 μm ABA relative to mock-treated seedlings of the same genotype. Values are means ± SE (= 32).
Asterisks indicate significant differences compared with mock-treated seedlings of the same genotype (d) and hormone-treated empty-vector (EV) seedlings (e) (Student's test,  0.05).

Auxin promotes growth by stimulating cell elongation (Zhang et al., 2008), although at high levels, it may inhibit growth. As OsMPS is repressed by auxin, we tested whether growth retardation of OsMPS over-expression seedlings may be rescued by auxin application. Treatment with NAA (0.1 μm; 7 days) stimulated shoot and root growth of OsMPS OE lines and partially restored plant size (Figure 6c,d). EV seedlings appeared unaffected by the minimal level of exogenous NAA used. To assess the involvement of OsMPS in the response to ABA, seedlings were grown for 7 days on MS medium containing 5 μm ABA (Figure 6e). The shoot length of KD4–5 plants was moderately reduced, whereas it was significantly increased in the case of OE6–2 plants relative to EV seedlings.

OsMPS modulates the expression of hormone and cell wall-related genes

An initial microarray analysis on roots of OE6–2, KD4–5 and EV plants was performed to identify genes regulated by OsMPS. This resulted in the identification of 247 genes showing a ≥two-fold down-regulation in OE plants and a ≥two-fold up-regulation in KD plants compared to EV. GO enrichment analysis using PLAZA (Proost et al., 2009) revealed that, among others, the biological processes ‘sexual reproduction’ and ‘cell wall organization/biogenesis’ were significantly enriched (Table S3). Furthermore, 460 genes were ≥two-fold up-regulated in OE roots but ≥two-fold repressed in KD roots. GO enrichment analysis on these genes revealed a role in metabolic processes (Table S4). Based on the functional annotation, a subset of genes was selected for quantitative RT–PCR analysis to confirm the microarray data and obtain insight into the function of OsMPS (Table S5). Auxin-related genes, including indole-3–acetic acid inducible protein (OsIAA) genes (i.e. OsIAA4, OsIAA12 and OsIAA20) and auxin response factor (ARF) genes (i.e. ARF10 and ARF19) were found to be repressed in OE6–2 but induced in KD roots (Figure 7a). Furthermore, ARF4 and ARF21 were repressed in OE6–2, and OsIAA11 and OsIAA26 were induced in KD4–5. Moreover, BR biosynthesis and signaling genes were found to be down-regulated upon over-expression of OsMPS (Figure 7a). These genes include OsBRI1, OsBAK1, OsBZR1, OsLIC, OsIBH1 and OsBSL2 (Bai et al., 2007; Wang et al., 2008; Zhang et al., 2009, 2012b), and an EXORDIUM ortholog (Os02g5200; Schröder et al., 2009). In KD lines, only the BR biosynthesis gene OsDWARF11 (Tanabe et al., 2005) showed increased expression. The expression of ACO1 and ACO3, catalyzing the oxidation of 1–aminocyclopropane-1–carboxylic acid to ethylene (Iwamoto et al., 2011), was up-regulated in roots of KD4–5, while expression of OsABA8ox2, encoding an ABA-inactivating enzyme (Zhu et al., 2009), was repressed in OsMPS over-expression roots.

Figure 7.

OsMPS-dependent genes are regulated by multiple hormones.
Relative expression levels (log2 FC) of genes involved in (a) auxin, BR, ethylene and ABA biosynthesis/signaling, (b) cell expansion, (c) cell-wall biosynthesis/modification, (d) reactive oxygen species production/detoxification, and (e) development. Heatmaps represent mean expression values (= 3) for OsMPS knockdown (KD4–5) or over-expression (OE6–2) roots and after 2 or 6 h treatment with auxin (NAA), GA, BL, CK (6–benzylaminopurine) or ABA in wild-type roots.

Cell expansion requires cell-wall relaxation, which is achieved by the action of EXPs, endoglucanases (GLUs) and yieldins (Hayashi et al., 1984; McQueen-Mason et al., 1992; Okamoto and Okamoto, 1995). Seven EXP genes were found to be differentially expressed (Figure 7b). OsEXPB3 and OsEXPB6 were strongly induced in KD4–5 roots, but repressed in OE6–2 roots. In addition, OsEXPA4, OsEXPA8 and OsEXPA22 were down-regulated in OE6–2, indicating that OsMPS negatively regulates EXP expression. Furthermore, OsEXPB2 was significantly induced in KD4–5, whereas OsEXPA14 was repressed in both the OE and KD lines. Additionally, OsGLU5 and OsGLU14, which are expressed during lateral root development (Yoshida and Komae, 2006), were induced in KD4–5 roots and/or decreased in OE6–2 (Figure 7c). Reactive oxygen species contribute to cell growth by loosening the cell-wall matrix (Passardi et al., 2004). Enzymes responsible for the production of reactive oxygen species include cell wall-located class III peroxidases (PRXs). Eight PRX genes were found to be significantly down-regulated in OE6–2, while four PRX genes were induced in KD4–5 (Figure 7d). Taken together, the expression analysis supports a role for OsMPS in controlling cell wall-related genes.

Co-expression network analysis reveals a potential role of OsMPS in the regulation of cell-wall remodeling

In order to evaluate whether the genes regulated by OsMPS have an intrinsic association with each other, an extensive co-expression analysis was undertaken using the RiceFREND and ATTED–II web servers (Obayashi et al., 2011; Sato et al., 2013). EXPB genes (OsEXPB2, OsEXPB3 and OsEXPB6) were used as guide genes for both tools. Although RiceFREND is restricted to displaying small gene networks (100 genes), the analysis demonstrated that OsEXPB genes are highly co-regulated with other cell wall-related genes (Figure S4 and Methods S1). Furthermore, expression of the selected EXPB genes was intimately connected with expression of other EXP, PRX, xyloglucan:xyloglucosyl transferase and auxin-related genes, and several ethylene biosynthesis genes (Tables S6 and S7). The obtained co-expressed genes were cross-referenced with the OsMPS microarray data. Many genes co-expressed with OsEXPB2, OsEXPB3 and OsEXPB6 were also differentially expressed in OE6–2 and/or KD4–5. These genes include the EXP genes OsEXPA4 and OsEXPA8, the cell-wall genes OsGLU14 and OsCESA3, the PRX genes OsPRX45, OsPRX46, OsPRX65, OsPRX95, OsPRX108, Os02g14170 and Os08g42030, and the auxin-related genes OsIAA12, OsIAA14 and OsIAA23. As the co-expression associations overlap with OsMPS-dependent genes identified by the microarray studies, the results support the notion that OsMPS functions as a regulator of these genes. Indeed, mining of expression data revealed that OsMPS induction during cold, drought and salt stress co-occurs with down-regulation of genes co-expressed with β–class EXP genes (Figure S4).

The effect of OsMPS on gene expression may be mimicked by auxin, CK and ABA applications

As OsMPS expression is modulated by several hormones, the response of its potential target genes towards auxin, BL, GA, CK and ABA was tested (Figure 7). Down-regulation of OsMPS affected the expression of auxin-related genes in a similar manner as auxin treatment. Five OsIAA genes, ARF10, ARF19 and to some extent two indole-3–acetic acid-amido synthetase genes involved in auxin inactivation (OsGH3.3 and OsGH3.13) were up-regulated after auxin treatment (Figure 7a). GA application also induced several OsIAA genes. Repression of auxin-related genes as in OsMPS over-expression plants was observed in the wild-type after CK treatment. Application of CK reduced the expression of OsGH3.3, OsGH3.13, OsIAA4, OsIAA11, ARF10 and ARF21. Similarly, ABA treatment caused down-regulation of OsGH3.13, OsIAA11 and OsIAA12, while OsIAA20 was induced (Figure 7a). The down-regulation of BR-related genes in OsMPS over-expression plants was only weakly reproduced by hormone treatments.

In OsMPS knockdown plants, several EXP genes were up-regulated, of which OsEXPA4, OsEXPB3 and OsEXPB6 were auxin-inducible. In contrast, OsEXPB2 and OsEXPA14 were repressed after GA treatment, while epi-BL application did not show an effect. Repression of EXP as observed by OsMPS over-expression was also observed after ABA or CK treatment. Both hormones down-regulated the expression of OsEXPA4, OsEXPA8, OsEXPB2, OsEXPB3 and OsEXPB6, while OsEXPA14 expression was exclusively repressed by CK. In contrast, OsEXPA22 was found to be induced by ABA and CK (Figure 7b). Several cell wall-related genes, including OsGLU5, OsGLU14 and a cellulase-like protein-encoding gene (CSLA1) were auxin-inducible (Figure 7c). The down-regulation of cell wall-related genes in OsMPS over-expression plants was reproduced by CK and ABA treatment. CK treatment resulted in down-regulation of OsGLU5, OsGLU14, CSLA1 and OsCESA3. ABA treatment repressed OsGLU5 and OsGLU14 and weakly down-regulated OsCESA3 (Figure 7c).

Increased expression of OsVIVIPAROUS1 (OsVP1) and CSLD5, as in OE6–2, was observed after CK or GA treatment of wild-type roots (Figure 7c,e). Down-regulation of PRX gene expression by OsMPS over-expression was also observed after treatment with CK or ABA (Figure 7d). Both hormones repressed OsPRX95, OsPRX104 and Os02g14170. In addition, OsPRX45, OsPRX77, OsPRX108 and Os06g48020 were down-regulated only by CK application. While over-expression of OsMPS repressed OsPRX46 and OsPRX52, ABA and CK induced the expression of these genes. Epi-BL treatment affected the expression of OsPRX45, OsPRX95 and OsPRX108 (Figure 7d). Thus, the effect of OsMPS on gene expression overlaps with the response to ABA, CK and auxin.

OsMPS binds to the promoter of expansin and endoglucanase genes

The Arabidopsis OsMPS homolog AtMYB71 was shown to bind a ‘TAACTG’ DNA sequence (Wang et al., 2002). Therefore, promoter sequences (1.5 kb) of genes differentially expressed in OsMPS transgenic plants were screened for the presence of the AtMYB71 binding site (Figures 8 and S5). Two α–class EXP genes, OsEXPA4 and OsEXPA8, and three β–class EXP genes, OsEXPB2, OsEXPB3 and OsEXPB6, contain the cis-element. Furthermore, several genes involved in cell-wall biosynthesis/modification, e.g. OsGLU5 and OsGLU14, possess the AtMYB71 binding site. We tested whether OsMPS interacts with the ‘TAACTG’ motif by an electrophoretic mobility shift assay (EMSA) using probes that span the binding site (Figure 8a,b). Incubation of OsMPS with OsEXPA4 promoter probes caused a band shift, which was diminished by addition of unlabeled probe (Figure 8c). Additionally, we observed that OsMPS binds to DNA probes containing the ‘TAACTG’ motif from the promoters of OsEXPB3, OsEXPB6, OsGLU5 and OsGLU14. No retention was observed when probes were incubated with OsMPS-free protein extract, indicating the specificity of the reaction.

Figure 8.

OsMPS associates with EXP and GLU promoters.
(a) Position of the OsMPS recognition motif (M; black box) in the promoters of putative direct targets. EMSA probes (blue arrows) and ChIP-qPCR target regions (gray arrows) are indicated.
(b) OsMPS consensus binding site (red outline) present in EMSA probes.
(c) EMSAs performed using probes specific to OsEXPA4, OsEXPB3, OsEXPB6, OsGLU5 and OsGLU14. Binding of OsMPS caused a band shift (black arrows). Comp., competition with unlabeled probe (100-fold excess). NC, probe only. NC*, incubation with wheatgerm extract and empty vector (lacking the OsMPS coding sequence).
(d) ChIP-qPCR results for OsMPS towards the promoters of OsEXPA4, OsEXPA8, OsEXPB2, OsEXPB3, OsEXPB6, OsGLU5 and OsGLU14. Negative controls (NC1–3) represent upstream regions of OsEXPA4, OsGLU5 and OsGLU14 promoters. Asterisks indicate significant differences compared with an IgG control ( 0.05). FC, fold change.

To determine whether OsMPS regulates its proposed target genes in vivo, a transient chromatin immunoprecipitation assay coupled to quantitative PCR (ChIP-qPCR) was performed on wild-type rice protoplasts transformed with the 35S:OsMPS-CFP construct (Figure 8d). OsMPS binds to the promoter of OsEXPA4, OsEXPA8, OsEXPB2, OsEXPB3 and OsEXP6. Furthermore, enrichment for OsGLU5 and OsGLU14 promoter fragments spanning the AtMYB71-specific binding site was observed, indicating that OsMPS directly modulates the expression of genes involved in cell-wall modulation. In contrast, no enrichment was detected for three negative controls covering upstream sequences from OsEXPA4 (−1.7 kb), OsGLU5 (−2.4 kb) and OsGLU14 (−2.1 kb) that lack the cis-element (Figure 8d).


Plant hormones allow plants to adapt to changing environments by mediating growth, development and nutrient distribution (Peleg and Blumwald, 2011). Although hormone response pathways have been elucidated in detail, only little is known with respect to molecular components that regulate cross-talk (Depuydt and Hardtke, 2011). Here, we characterized OsMPS and revealed that it may form a common hub for cross-talk between auxin and ABA/CK to control adaptive growth.

TFs play a central role in the regulation of plant responses towards adverse growth conditions (Wolters and Jürgens, 2009). R2R3-MYB proteins are well known for their role in abiotic stress tolerance and development (Park et al., 2010; El-Kereamy et al., 2012; Makkena et al., 2012; Yang et al., 2012). In the absence of stress, knockdown of OsMPS increased biomass accumulation, while OsMPS over-expression impaired growth. However, down-regulation of OsMPS negatively affected biomass accumulation under salt stress, while over-expression resulted in a less severe relative biomass reduction (Figure 3). Metabolic profiling revealed that OsMPS over-expression results in accumulation of amino acids and decreased levels of tricarboxylic acid cycle intermediates, which may indicate a lower rate of protein synthesis and metabolism, correlating with the smaller plant size observed (Stitt et al., 2010; Ribeiro et al., 2012).

OsMPS is transiently induced during the initial phase of salinity stress (Figure 1a), in agreement with the occurrence of growth cessation as the first response towards stress (Chaves et al., 2009). Similar to mis-expression of OsMYB4 (Park et al., 2010), over-expression or knockdown of OsMPS had costly trade-offs regarding panicle development and grain yield. Although their panicles had a similar number of spikelets, a significant decrease in grain number was observed for both lines (Figure 5e,f). A potential role for OsMPS in seed setting is supported by the presence of GUS activity in anthers of reporter lines (Figure 2a). Previously, it was demonstrated that root-specific over-expression of OsNAC10 improves grain yield in rice, while its ectopic expression negatively affects grain filling (Jeong et al., 2010). Therefore, it would be of interest to generate cell type- or tissue-specific OsMPS over-expression lines.

OsMPS expression is modulated by multiple hormones. Auxin, BL and GA, which are known to promote cell elongation (Depuydt and Hardtke, 2011), transiently decrease the expression of OsMPS. Consistently, OsMPS knockdown plants showed increased size and length of leaf epidermal cells (Figure 4). On the other hand, OsMPS is induced by CK and ABA (Figure 6b), and over-expression reduced leaf cell size (Figure 4). The promoter of OsMPS contains ABA-responsive, AuxRE and GARE regulatory elements, suggesting a direct link with upstream TFs controlled by these hormones. The ABA-responsive element is a known recognition site for AREB/ABF TFs including OsABF2 and ABL1, which are important positive regulators of ABA signaling and abiotic stress responses in rice (Zou et al., 2008; Yang et al., 2011).

OsMPS negatively regulates the expression of several ARF and IAA genes (Table S5 and Figure 7a). Interestingly, the growth inhibition of OsMPS over-expression lines was partially rescued by low concentrations of exogenous auxin (Figure 6c–e). In agreement, over-expression of OsIAA11, which is induced in roots of OsMPS knockdown plants, causes an increased crown root length and thickness (Zhu et al., 2012). Furthermore, the ARF1 target gene CROWN ROOTLESS1 (CRL1) is essential for crown root growth (Inukai et al., 2005). Hence, the observed effect of OsMPS on root growth may be through differential expression of auxin-related genes. CK is known to have negative effects on de novo auxin-induced root formation (Kitomi et al., 2011). Interestingly, the expression pattern of AUX/IAA genes in OsMPS over-expression plants is similar to that of wild-type seedlings treated with CK (Figure 7a). In rice, the crown rootless5 (crl5) mutant is impaired in root growth. CRL5 encodes an AP2/ERF TF that is induced by auxin and activates the expression of OsRR1, a negative regulator of CK signaling. Over-expression of either CRL5 or OsRR1 reduces CK sensitivity and increases root formation (Kitomi et al., 2011). Thus, an altered balance between CK and auxin may be responsible for the observed effect of OsMPS on root growth.

A prerequisite for cell elongation is cell-wall loosening to allow for morphological change (Depuydt and Hardtke, 2011). Auxin-induced acidification of the cell wall activates EXPs, which loosen the non-covalent bonds between polysaccharides allowing turgor-driven growth (Rayle and Cleland, 1992; Nakayama et al., 2012). Seven EXP genes were found to be repressed by OsMPS. Moreover, OsMPS binds in vivo to the promoters of OsEXPA4, OsEXPA8, OsEXPB2, OsEXPB3 and OsEXPB6 (Figure 8d). OsEXPA4 has been reported to enhance plant growth when over-expressed (Choi et al., 2003). Interestingly, all EXP genes directly regulated by OsMPS are repressed by ABA (Figure 7b) and short-term salt stress (Figure S4). We propose that down-regulation of EXP genes during the initial phase of salt stress is mediated by ABA through induction of OsMPS. In addition, OsMPS directly controls the expression of OsGLU5 and OsGLU14 (Figure 8). These GLUs hydrolyze the two major hemicelluloses in type II cell walls (Yoshida and Komae, 2006). OsGLU5 (OsCEL9A) and its paralog OsGLU14 (OsCEL9F) have been previously reported to be auxin-inducible genes that are expressed during lateral root formation (Yoshida et al., 2006). Thus, OsGLU5 and OsGLU14 are positive regulators of growth whose expression is restricted during abiotic stress.

In summary, this study shows that OsMPS is induced by salt stress and the phytohormones ABA and CK (Figure 9). In contrast, hormones promoting growth down-regulate OsMPS. OsMPS acts as a negative regulator of auxin-related genes and cell wall-remodeling genes. The expression level of OsMPS negatively correlates with root biomass accumulation. As root growth is controlled by the balance between CK and auxin, OsMPS may function as a novel mediator between these two hormonal pathways. Taken together, the results show that OsMPS is required for the adaptive growth response in rice during adverse environmental conditions.

Figure 9.

OsMPS regulates hormone-dependent growth.
In the presence of auxin, BL or GA, OsMPS is repressed, resulting in activation of EXPANSIN (EXP) genes of both α- and β–classes, and two endoglucanase (GLU) genes promoting cell-wall remodeling and growth. In contrast, ABA and salt stress inhibit growth by down-regulating EXP and GLU expression via induction of OsMPS. The hormone-dependent regulation of OsMPS expression allows growth adaptation of the root to changes in environmental conditions.

Experimental procedures

Plant material and growth conditions

Analysis was performed using non-segregating transgenic lines (T3 generation) established in the rice Nipponbare background. Soil-grown plants were cultured in phytotrons as described by Degenkolbe et al. (2009). Growth on MS medium or in hydroponic culture was performed as described by Schmidt et al. (2012). Plants were grown at 26/22°C and 75/70% relative humidity (day/night) with a day length of 12 h and a light intensity of 700 μmol m−2 sec−1.

Constructs and rice transformation

The artificial microRNA specific to OsMPS was designed as described by Warthmann et al. (2008) and cloned into the pC5300 OE vector as described previously (Schmidt et al., 2012). For over-expression, the OsMPS coding sequence was amplified by PCR and recombined into pC5300. For the OsMPS:GUS construct, a 2 kb promoter sequence upstream of the transcriptional start site was cloned into pMDC162 (Curtis and Grossniklaus, 2003) using a two-step cloning strategy involving pENTR/D–TOPO (Invitrogen, Constructs were transformed into rice calli (cv. Nipponbare) via Agrobacterium tumefaciens strain EHA105 as described by Sallaud et al. (2003). To generate 35S:OsMPS-CFP, the coding sequence of OsMPS was cloned into the pGHPGWC vector (Zhong et al., 2008). Protoplast isolation and transformation were performed as described by Zhang et al. (2011). Fluorescence imaging was performed using a confocal laser scanning microscope (SP5; Leica Microsystems, Primers used for cloning are listed in Table S8.

RNA extraction and expression analysis

RNA isolation, cDNA synthesis and expression profiling using quantitative RT–PCR were performed as described previously (Schmidt et al., 2012). Rice ACTIN (Os03g50885) was used as an internal control (Figueiredo et al., 2012). Oligonucleotide sequences used for expression profiling of OsMPS and OsMPS-dependent genes were designed using QuantPrime (Arvidsson et al., 2008) and are listed in Table S9. For microarray analysis, RNA extracted from roots of 4-week-old hydroponically grown plants was hybridized to the Affymetrix 57k Rice Genome GeneChip (Affymetrix, To identify differentially expressed genes, the log2 transformed signal ratio of each gene was calculated, and a log2 (ratio) ≥1.0 or ≤0.5 was used as the cut-off. All probe sets matching transposable elements were removed from the dataset. Probes were considered for further analysis irrespective of detection call (P: present; M: marginal; A: absent).

Salt and hormone treatments

Dehulled seeds were surface-sterilized as described by Schmidt et al. (2012) and placed on vertical square plates with MS medium containing 100 mm NaCl, 1 μm ABA or 0.1 μm NAA. For estimation of biomass reduction under salt stress (7 days, 100 mm NaCl), plants were grown hydroponically and analyzed as described previously (Schmidt et al., 2013). For expression analysis, 2-week-old hydroponically grown wild-type seedlings were treated with 10 μm NAA, 50 μm GA3, 10 μm epi-BL, 10 μm 6–benzylaminopurine or 50 μm ABA for 2 or 6 h. Roots were harvested and analyzed in triplicate.

Detection of GUS activity

OsMPS:GUS plants (T2 generation) were stained overnight in GUS solution as described by Jefferson (1987).

Cell size measurements

Measurements of leaf blade epidermal cell size and length were performed as described by Barrôco et al. (2006) using CellP software (Olympus, after imaging with an Olympus BX51 microscope (Olympus) using differential interference contrast optics.

Metabolite profiling analysis

GC–MS analysis was performed on roots and leaves of 4-week-old hydroponically grown plants. Extraction and derivatization of metabolites from tissues for GC-MS analysis were performed exactly as described by Lisec et al. (2006). Peak detection, retention time alignment and library matching were performed using TagFinder 4.0 (Luedemann et al., 2012). Metabolites were quantified based on the peak intensity of a selective mass. The amount of metabolites was analyzed as the relative metabolite level calculated by normalization of the peak intensity to that of the internal standard ribitol and the fresh weight.

Electrophoretic mobility shift assay and ChIP-qPCR

Electrophoretic mobility shift assays were performed using in vitro expressed OsMPS protein as described by Schmidt et al. (2013). To this end, the OsMPS coding sequence was cloned into the pF3A WG Flexi vector (Promega, and translated using TNT® SP6 High Yield Wheat Germ Mastermix containing 1 μl FluoroTec Green Lys (Promega). EMSAs were performed using Cy5-labeled probes (Table S10) using a LightShift® chemiluminescent assay kit (Pierce, according to the manufacturer's instructions. Separation of protein–DNA complexes was performed on a 5% native polyacrylamide gel. For detection of the Cy5 signal, a Typhoon scanner (GE Healthcare, was used.

A transient ChIP-qPCR assay with 35S:OsMPS-CFP was performed as described by Du et al. (2009) using rice protoplasts (Zhang et al., 2011), whereby, after fixation of the cells with formaldehyde, a EpiQuick kit (Epigentek, was used as described previously (Lai et al., 2012). For calculation of enrichment, the obtained CT values were normalized against the input CT, and ΔΔCT [CT(IgG) − CT(anti-GFP)] was calculated. Oligonucleotides were designed using Primer3 (Rozen and Skaletsky, 2000), and are listed in Table S11.


This work was in part supported by grants from the ERA-NET Plant Genomics program (TRIESTER, numbers 0313993A and ANR-06-ERAPG-005-01) through the Federal Ministry of Education and Research in Germany and the Agence Nationale de la Recherche in France, and the RicE Functional Genomics (REFUGE) platform, Montpellier, France, funded by the Agropolis Foundation. R.S. thanks the FAZIT Stiftung for a PhD fellowship. We thank Eugenia Maximova (Max Planck Institute of Molecular Plant Physiology, Germany) for microscopic work and Christopher Herbst (University of Potsdam, Germany) for technical assistance.