• Open Access

Over-expression of OsPIN2 leads to increased tiller numbers, angle and shorter plant height through suppression of OsLAZY1


(Tel/fax +0086 25 84396246; email ghxu@njau.edu.cn)


Crop architecture parameters such as tiller number, angle and plant height are important agronomic traits that have been considered for breeding programmes. Auxin distribution within the plant has long been recognized to alter architecture. The rice (Oryza sativa L.) genome contains 12 putative PIN genes encoding auxin efflux transporters, including four PIN1 and one PIN2 genes. Here, we report that over-expression of OsPIN2 through a transgenic approach in rice (Japonica cv. Nipponbare) led to a shorter plant height, more tillers and a larger tiller angle when compared with wild type (WT). The expression patterns of the auxin reporter DR5::GUS and quantification of auxin distribution showed that OsPIN2 over-expression increased auxin transport from the shoot to the root–shoot junction, resulting in a non-tissue-specific accumulation of more free auxin at the root–shoot junction relative to WT. Over-expression of OsPIN2 enhanced auxin transport from shoots to roots, but did not alter the polar auxin pattern in the roots. Transgenic plants were less sensitive to N-1-naphthylphthalamic acid, an auxin transport inhibitor, than WT in their root growth. OsPIN2-over-expressing plants had suppressed the expression of a gravitropism-related gene OsLazy1 in the shoots, but unaltered expression of OsPIN1b and OsTAC1, which were reported as tiller angle controllers in rice. The data suggest that OsPIN2 has a distinct auxin-dependent regulation pathway together with OsPIN1b and OsTAC1 controlling rice shoot architecture. Altering OsPIN2 expression by genetic transformation can be directly used for modifying rice architecture.


The plant hormone auxin [(indole-3-acetic acid (IAA)] is a essential regulator of many plant developmental processes, including embryogenesis, root patterning, vascular tissue differentiation, apical dominance, phototropism, gravitropism and other physiological processes (Friml et al., 2003; Mattsson et al., 2003; Kimura and Kagawa, 2006; Palme et al., 2006). Auxin is mainly synthesized in leaf primordial and young leaves and directionally transported to the root tip. At a cellular level, the directional transport of auxin results from the asymmetric distribution of different auxin membrane carriers, including the influx carrier AUX1 protein family (Bennett et al., 1996), the efflux-facilitating PIN-FORMED (PIN) family members (Friml et al., 2003; Blilou et al., 2005) and a number of the p-glycoprotein ABC transporters (Geisler and Murphy, 2006). It has been reported that PINs perform a rate-limiting role in catalysing the efflux of auxin from cells, and their asymmetric cellular localization determines the direction of cell-to-cell flow which is central to auxin-regulated growth processes (Petrásek et al., 2006).

Eight PIN genes, named from AtPIN1 to AtPIN8, have been identified in the genome of Arabidopsis. Each member displays a unique tissue-specific expression pattern and subcellular localization (Friml et al., 2003; Blilou et al., 2005; Paponov et al., 2005). The pin1 mutants exhibit pin-formed inflorescences and defective development of vascular tissue (Okada et al., 1991; Gäelweiler et al., 1998). Mutations in PIN2 display agravitropic root growth phenotypes (Chen et al., 1998; Müller et al., 1998; Shin et al., 2005). PIN2 deployment plays a pivotal role in transmitting auxin shootward through the peripheral layers of the root meristem and elongation zone (Muday and Rahman, 2008). It was proposed that rootward PIN2 and PIN1 are differentially sensitive to shootward relocation (Rahman et al., 2010). The other homologues, PIN3, PIN4 and PIN7, appear to function in tropisms, root meristem patterning and establishing embryonic polarity, respectively (Friml et al., 2002a,b, 2003; Blilou et al., 2005). The rice genome contains twelve putative PIN genes (Wang et al., 2009), including four PIN1 genes (named OsPIN1a-1d), OsPIN2, three PIN5 genes (OsPIN5a-c), OsPIN8 and three monocot-specific PIN genes (OsPIN9, OsPIN10a and OsPIN10b). Suppression of OsPIN1b expression by RNA interference (RNAi) in rice showed a phenotype exhibiting inhibition of the emergence and development of adventitious roots (Xu et al., 2005).

A unique role of auxin in rice is in the control of tiller angle (McSteen, 2010). For monocotyledonous crops, tiller angle is an important agronomic trait that has been considered for breeding programmes (Wang and Li, 2005). The regulation of tiller angle is a combination of environmental factors and hormone homeostasis (Myers et al., 1994; Li et al., 2007; Yu et al., 2007). It has been reported that the angle of rice tillers is correlated with the gravitropism of the rice leaf sheath bases (Abe et al., 1996; Hu et al., 2007). OsTAC1 was reported to play a critical role in controlling rice architecture. High expression of OsTAC1 results in wider tiller angle and more tillers, while low expression leads to erect tillers with a smaller tiller angle (Yu et al., 2007). Xu et al. (2005) reported larger tiller angles and more tillers in OsPIN1b RNAi mutant rice plants while no change of tiller angle or number was found in OsPIN1b over-expression lines. These authors demonstrated that OsPIN1b facilitated auxin transport; therefore, the RNAi mutants increased tiller angle. However, mutation of OsLazy1, a grass-specific gene playing a negative role in polar auxin transport, also increased tiller angle of rice (Li et al., 2007; Yoshihara and Iino, 2007). It is therefore still an important question whether polar auxin transport promotes or inhibits tiller angle in rice.

We used transgenic over-expression of OsPIN2 (AK101191), the only rice homologue of AtPIN2, to show that plants have a similar phenotype, with increased tiller angle, more tillers and fewer adventitious roots, like OsPIN1b suppression mutants. Furthermore, we demonstrated that in OsPIN2-over-expressing plants, free IAA distribution was changed in roots, shoots and especially at the root–shoot junction. We also detected the expression patterns of OsTAC1, OsPIN1b and OsLazy1 in the OsPIN2-over-expressing plants. The results demonstrate that OsPIN2 can be used in molecular breeding for improving rice architecture.


OsPIN2 encodes a PIN2 family member

OsPIN2 was proposed as an auxin efflux transporter and a member of the PIN family (Wang et al., 2009). The cDNA sequence of OsPIN2 (AK101191) comprises 2276 nucleotides and alignment between the cDNA, and the genomic sequence revealed that the gene has seven exons and six introns (Figure S1A). The ORF encodes a protein of 629 amino acid residues and shows 56.3% sequence identity with Arabidopsis AtPIN2 (Figure S2). Hydropathy and transmembrane (TM) motif analyses of the amino acid sequence revealed that the OsPIN2 protein contains three characteristic regions, including a hydrophobic region with five TM domains, a predominantly hydrophilic core, followed by another hydrophobic region with four TM segments (Figure S1B).

Generation of transgenic rice showing OsPIN2 over-expression and phenotypes of transformants

Ubiquitin promoter (pUbi) has been long reported as a useful strong promoter in a variety of applications in gene transfer studies and drives the gene expression most actively in rapidly dividing cells (Cornejo et al., 1993). To investigate the function of OsPIN2 in rice, we made transgenic rice plants over-expressing the full length of cDNA driven by pUbi. Southern blotting analysis revealed that they came from two transgenic lines (O1 and O2, respectively) distinct from each other and both had one copy of the transgene (Figure 1a). Real-time RT-PCR analyses showed that transcriptional expression of OsPIN2 was moderate in the root–shoot junctions and roots but very faint in the leaves of WT (the untransformed plants as control) (Figure 1b). In both the transgenic lines, OsPIN2 had the most abundant transcripts in the root–shoot junction, and similar levels in the leaves and roots (Figure 1b). The increases in OsPIN2 transcriptional expression driven by pUbi in these tissues reached by 3–15 times.

Figure 1.

 Phenotypes and molecular analyses of OsPIN2 over-expression transgenic plants (O1 and O2) and untranformed wild-type plant (WT). (a) Southern blot analysis of transgene copy number in T2 transgenic rice and WT. Genomic DNA was digested with the restriction enzyme HindIII and BamHI. The Hyg probe was used for hybridization. M, marker; P, p1390–Ubi as positive control. (b) Real-time quantitative RT-PCR analysis of endogenous OsPIN2 in the leaf, root–shoot junction and root of WT, O1 and O2. RNAs were extracted from different parts of 14-day-old seedlings. Values are shown with mean ± standard deviation (n = 3). (c–e) Tiller angle (c), plant height (e) and tiller number (d) of WT, O1 and O2 at the stage showing in I. Values are shown with mean ± standard deviation (n = 5). The tiller angle was measured between the outermost tiller in the left side and the outermost tiller in the right side (Jin et al., 2008). (f–h) The 80-day-old WT (f), T0 generation of O1 (g) and O2 (h) grown in the same field plot. (i–j) T2 generation of WT (left), O1 (middle) and O2 (right) transgenic plants was grown in the greenhouse for 80 day (i) and 120 day (j).

Over-expression of OsPIN2 significantly increased tiller angle and numbers and decreased height of the rice in comparison with WT (Figure 1i). At 80 days after germination (Figure 1c–j), O1 and O2 plants had a wider tiller angle (the angle between the outermost tillers in the left and right side) (44 ° and 48 °, respectively) than the WT (22 °) (Figure 1c). In addition, the two transgenic lines had many more tillers per plant (21 and 23, respectively) than the WT (17 on the average) (Figure 1d). In contrast, the whole plant height was very significantly decreased by OsPIN2 over-expression (Figure 1e). The transgenic plant phenotypes were inheritable from T0 (Figure 2f–h) to T2 generations (Figure 2i–j).

Figure 2.

 Schematic map of integrated T-DNA insertion site in rice genome. (a) T-DNA containing ubi-promoter in the expression vector p1390-Ubi. (b, c) Red bar shows the insertion site of O1 (b) and O2 (c) in different Oryza sativa japonica group chromosome clone.

At the ripening stage (Figure 1j), the OsPIN2-over-expressing plants showed shorter panicle length, less number of grains per panicle and lower grain weight per panicle in comparison with the control (Table 1). Even though the seed setting rate was similar between O1, O2 and WT, O1 plants had a 35% increase in the effective tiller number, leading to a 16% grain yield increase per plant relative to control plants (Table 1). O2 plants showed the same tendency but not significant increase in panicle number and grain yield (Table 1). OsPIN2 over-expression also decreased the grain length and breadth by 4%–7% and 12%–16%, respectively (Table 1). The changes in the grain size lead to 10%–12.5% decreases of 1000-grain weight compared to WT. Moreover, over-expression of the OsPIN2 significantly decreased the number of adventitious roots and the total root length by 22%–28% (Table 2).

Table 1.   Statistics of agronomic traits and yields of wild type (WT) and OsPIN2 transgenic seedlings in the pot experiment
  1. Numbers are presented as mean ± SE. The number of observations in each mean is 5 except grain length, grain breadth, 1000-grain weight and yield per plant, which was recorded from ten seedlings. Means in the same column followed by the same letter are not significantly different (P < 0.05, LSD test).

Plant height (cm)71.6 ± 1.1a57.0 ± 2.8b52.8 ± 1.9b
Total tiller number per plant37.33 ± 2.81b53.8 ± 2.48a49.33 ± 2.50a
Effective tiller number per plant31.84 ± 3.21b43.04 ± 3.53a39.47 ± 2.3a
Panicle length (cm)21.24 ± 0.54a17.73 ± 0.64b17.84 ± 0.43b
Grain weight (g/panicle)1.10 ± 0.06a0.74 ± 0.04b0.67 ± 0.06b
Seed setting rate (%)73.21% ± 2.31a75.84% ± 1.19a74.21% ± 1.49a
Grain number per panicle59.47 ± 4.01a44.93 ± 3.49b36.38 ± 1.57b
Grain length (mm)7.25 ± 0.06a6.91 ± 0.06b6.82 ± 0.07b
Grain breadth (mm)3.19 ± 0.03a2.80 ± 0.06b2.73 ± 0.04b
Length/breadth2.28 ± 0.02b2.47 ± 0.05a2.45 ± 0.03a
1000-grain weight (g)23.20 ± 0.41a20.81 ± 0.27b20.30 ± 0.09b
Yield (g/plant)15.76 ± 0.54b18.33 ± 0.63a17.30 ± 1.02ab
Table 2.   Phenotypes of wild type (WT) and OsPIN2 transgenic seedlings with NPA treatment
TreatmentGTSL (cm)RL (cm)TRL (cm)ARNLRL (cm)LRD (no./cm)
  1. GT, genotype; SL, shoot length; RL, root length; TRL, total root length; ARN, adventitious root number; LRL, lateral root length; LRD, lateral root density.

  2. Means in the same column followed by the same letter are not significantly different among the WT and the two transgenic lines with and without NPA treatment (P < 0.05, LSD test). Numbers are presented as mean ± SE. The number of observations in each mean is 5; 21-day-old seedlings were treated with 20 μm NPA at the root–shoot junctions for 7 days with five replicates.

−NPAWT19.6 ± 0.38a14.1 ± 0.40b235 ± 14.5b42.8 ± 0.58a0.06 ± 0.01b18.1 ± 0.68a
O116.5 ± 0.73b13.64 ± 0.3b184 ± 16.8c31.3 ± 2.85b0.08 ± 0.01b14.8 ± 2.00ab
O216.5 ± 0.56b13.33 ± 0.3b178 ± 14.0c30.8 ± 1.65b0.09 ± 0.02b14.5 ± 1.47ab
WT19.7 ± 0.41a16.2 ± 0.61a296 ± 15.7a33.7 ± 1.67b0.15 ± 0.01a13.9 ± 0.41b
+NPAO115.7 ± 0.24b13.2 ± 0.28b196 ± 7.4c30.8 ± 1.85b0.14 ± 0.02a13.3 ± 1.46b
O216.7 ± 0.61b13.06 ± 0.2b193 ± 17.7c29.8 ± 2.42b0.15 ± 0.01a13.2 ± 2.51b

To confirm that the changed phenotype of O1 and O2 was caused by over-expression of OsPIN2 in the transgenic plants, we detected the location of one copy T-DNA insertion in these two lines. The T-DNA containing ubi-promoter was inserted in the chromosome 3 and 10 in the genome of O1 and O2, respectively (Figure 2). There was no putative gene in the insertion site for both O1 and O2. This direct genetic evidence further supported the OsPIN2 contribution to the altered phenotype of the transgenic rice in Figure 1.

Auxin transport altered by OsPIN2 over-expression in rice

To assess whether the over-expression of OsPIN2 affected the free auxin levels, the concentration of endogenous free IAA in various organs of the transgenic and WT plants was quantified by high-performance liquid chromatography (HPLC). In the first and second leaves from the top, O1 and O2 transgenic plants contained free IAA concentrations that were 11%–35% lower than that in the control plants (Figure 3a). In the sheath of the first leaf, the transgenic plants and WT had nearly the same concentration of free IAA, while in the sheath of the second leaf, O1 and O2 plants had 38%–53% lower free IAA concentration than that in WT (Figure 3b). In contrast, in the root–shoot junction (shoot base), the two transgenic lines contained 65%–128% more free IAA than that in WT (Figure 3c).

Figure 3.

 High-performance liquid chromatography analysis of endogenous free indole-3-acetic acid concentrations in different tissues of 60-day-old wild type, O1 and O2 plants. Data are means of five replicates ± standard deviation. (a) First (left) and second (right) leaf from the top. (b) Sheath of the first (leaf) and second (right) from the top. (c) The root–shoot junctions. (d) Three different regions of root: 0–4 cm (left), 4–8 cm (middle) and 8–12 cm (right) from the root tip.

To detect the effect of OsPIN2 over-expression on auxin distribution, we separated individual roots into three segments: 0–4 cm including the root apex and elongation zones; 4–8 cm, the lateral root area; and 8–12 cm which was adjacent to the root–shoot junction. Both transgenic and WT roots showed the same patterns of free IAA concentration with an abrupt decrease from the tip to the root base (Figure 3d). However, the over-expression of OsPIN2 resulted in a several fold increase in free IAA concentration in each of the root segments (Figure 3d).

The IAA distribution was further visualized by DR5::GUS, a specific reporter that contains seven repeats of a highly active synthetic auxin response element and can reflect the in vivo auxin level (Ulmasov et al., 1997) (Figure 4). The GUS staining showed that expression of DR5::GUS in the O1 and O2 plants was strongly suppressed in the leaf blade tip in comparison with the control (Figure 4A,F,K). The GUS abundance was similar in the leaf blade and sheath of the O1, O2 and control plants (Figure 4A–O). In the root–shoot junction, DR5::GUS expression appeared concentrated in the phloem cells of control plants (Figure 4a), whereas it showed a non-tissue-specific pattern, i.e. expressing in all cell types in O1 and O2 lines (Figure 4b,c). Furthermore, IAA distribution was also altered in the roots (Figure 4d–l). In the root apex, DR5::GUS was polarized in control root apical meristem; however, the GUS expression lacked polarity from apical meristem to elongation zone in the O1 and O2 lines (Figure 4d,g,j). Furthermore, GUS staining in the lateral roots was much stronger in O1 and O2 than in control plants (Figure 4e,h,k). The DR5::GUS expression patterns matched and supported the distribution that was observed from the quantification of free IAA levels in the different organs (Figure 3).

Figure 4.

 Histochemical localization of GUS activity in wild type (A–E; a; d–f), O1 (F–J; b; g–i) and O2 (K–O; c; j–l) carrying DR5::GUS. Plants were stained for GUS activity for 2 h (d–l), 24 h (A–O; a–c). Bar = 1 mm. A, F and K, Tip areas of leaf blade. B, G and L, The first leaves from the top. C, H and M, Leaf sheaths of the first leaf from the top. D, I and N, the second leaves from the top. E, J and O, Leaf sheaths of the second leaf from the top. a, b and c, The root–shoot junctions. d, g and j, Root tips. e, h and k, A part of 4–8 cm from the root tip. f, i and l, A part of 8–12 cm from the root tip.

OsPIN2 is involved in polar auxin transport

To verify whether OsPIN2 is involved in polar auxin transport, 21-day-old seedlings of WT, O1 and O2 were treated with 20 μm NPA at the root–shoot junction. In WT, the targeted application of NPA significantly increased the length of total roots, particularly the lateral roots, but decreased the adventitious root number and lateral root density, while in the two transgenic plants, the NPA treatment did not significantly affect the total root length and number, except the length of the lateral roots which was also increased very largely (Table 2).

The different effects of NPA on the auxin transport of O1, O2 and WT could be visualized by different DR5::GUS activity in the roots (Figure 5). In WT, GUS activity was restricted to the root tip (Figure 5a, d) when NPA was applied at the root–shoot junction. However, there was little effect of the NPA treatment on the GUS activity and distribution in the roots of O1 and O2 plants (Figure 5b, c, e, f).

Figure 5.

 GUS expression in root tips of wild type (a, d), O1 (b, e) and O2 (c, f) seedlings carrying DR5::GUS with NPA treatment or not. Root tips were stained for GUS activity for 2 h. Bar = 100 μm. (a–c) root tips of 21-day-old seedlings grown in IRRI nutrition solution. (d–f) root tips of 21-day-old seedlings under the treatment of 20 μm NPA at the shoot base for 7 days.

Expression of tiller angle-related genes and OsPIN1b in OsPIN2-over-expressing plants

OsLazy1 was recognized as a negative regulator of polar auxin transport playing an important role in regulating rice shoot gravitropism, which in turn influences tiller angle (Li et al., 2007). Mutation of the Lazy1 gene in rice showed a similar phenotype to the O1 and O2 plants, having increased tiller numbers and angles. In this study, we detected that the transcription of OsLazy1 was suppressed very greatly in the leaves of both O1 and O2 plants, and in the roots of O1 plants (Figure 6a).

Figure 6.

 RT-PCR analysis of two tiller angle–related genes and OsPIN1 expression level in wild type (WT), O1 and O2. OsActin was used as a control. The amplification of OsLazy1 (a), OsTAC1 (b), OsPIN1 (b) and OsActin (c) transcripts was performed from total RNA isolated from the shoots or roots of the 60-day-old WT, O1 and O2. The RT-PCR products were resolved on TAE 1.5% agarose gel. The amount of products was compared at 35 cycles for OsLazy1, 33 cycles for OsTAC1, 30 cycles for OsPIN1 (b) and 25 cycles for OsActin. The bar graphs represent the relative expression of OsLazy1, OsTAC1 and OsPIN1 (b) normalized to the OsActin, estimated by relative quantification of the PCR products resolved by gel electrophoresis, respectively. Values are shown with mean ± standard deviation (n = 3).

It has been reported that TAC1 (tiller angle controller 1) is essential for the regulation of tiller angle, which controls plant architecture in rice (Yu et al., 2007). Mutation of OsTAC1 resulted in compact plant architecture with a tiller angle close to zero (Yu et al., 2007). However, expression of OsTAC1 both in the roots and leaves of the larger tiller angle plants was not changed (in O1 plants) and decreased (in O2 plants) in comparison with its expression in the control (Figure 6b). This result suggests that the altered architecture of rice by OsPIN2 over-expression was not occurring through a OsTAC1 regulated pathway, at least not at the OsTAC1 transcriptional level.

Among the 12 putative auxin efflux transporters in rice, OsPIN1b was the only one with a known in planta role (Xu et al., 2005). Suppression of OsPIN1b by RNAi resulted in a significant increase in tiller number and angle, while its over-expression driven by a 35S promoter did not show any significant effect on tiller numbers (Xu et al., 2005). Interestingly, no significant alteration of OsPIN1b expression occurred both in the roots and shoots of rice by over-expressing OsPIN2 (Figure 6c), demonstrating that there are different auxin-dependent regulatory pathways for PIN1b and PIN2 involved in controlling rice canopy architecture (Figure 6c).


Rice canopy architecture including the plant height, tiller number and angle, and panicle morphology are crucial traits for grain yield (Jiao et al., 2010). Auxin distribution that is mediated by AUX1, PIN and p-glycoprotein ABC transporters (Bennett et al., 1996; Friml et al., 2003; Blilou et al., 2005) has long been recognized to alter plant architecture. The general function of PINs is to transport auxin out of the cell and control auxin-regulated growth processes (Petrásek et al., 2006). In the rice genome, twelve PIN family members have been identified (Wang et al., 2009), but only one of them (OsPIN1b) has been functionally characterized (Xu et al., 2005). By over-expression of OsPIN2 through the transgenic approach in rice, we have shown that OsPIN2 plays an important role in the control of canopy architecture (Figure 1 and Table 2). The quantitative measurement of IAA concentration by HPLC and visible detection using DR5::GUS reporter activity have revealed that OsPIN2 affects free auxin distribution in various organs. Localized external NPA application at the root–shoot junction had little effects on the root growth of the OsPIN2 over-expression plants (Figure 5), which demonstrated that OsPIN2 affected the polar auxin transport directly. In addition, transcriptional analyses have demonstrated that OsPIN2 regulated OsLazy1, but not OsTAC1 and OsPIN1b in rice, suggesting that OsPIN2 has a distinct role, along with OsPIN1b and OsTAC1 in the auxin-regulated growth of the leaf canopy.

The gain of function of OsPIN2 in rice plants

It was previously reported that knockdown of OsPIN1b expression very significantly increased the number of tillers and tiller angle, while over-expression did not alter the shoot architecture of rice (Xu et al., 2005). In the rice genome, OsLazy1 has been identified as a novel grass-specific protein playing a negative role in polar auxin transport; its mutation resulted in an increase in the rice tiller angle (Abe et al., 1996; Li et al., 2007; Yoshihara and Iino, 2007). It is interesting to speculate whether enhancing OsLazy1 expression could compact rice architecture further. In addition, OsTAC1 was defined as a tiller angle control gene in rice, its expression level positively correlated with tiller angle and numbers (Yu et al., 2007).

Transgenic plants over-expressing OsPIN2 showed larger tiller angles, lower plant height, more tillers than WT (Figure 2c–e), which was similar to the OsPIN1b knockdown mutants (Xu et al., 2005). The change of phenotype caused by alteration of OsPIN2 and OsPIN1b expression supported the common observation that rice plant height was negatively correlated with tiller number (Wang and Li, 2005). Therefore, auxin might play a key role in the crosstalk between the plant height and branching, an aspect of canopy architecture which is poorly understood.

Over-expressing OsPIN2 also led to fewer adventitious roots (Table 2), which was similar to the OsPIN1b knockdown mutants (Xu et al., 2005). NPA is a well-characterized inhibitor of the polar auxin transport, which strongly inhibits auxin efflux (Luschnig, 2001). Blocking polar auxin transport at the root–shoot junction inhibits lateral root initiation (Reed et al., 1998). Zhou et al. (2003) reported that application of the NPA in rice root–shoot junction blocked the initiation and growth of adventitious and lateral roots in rice, the same as that we observed for WT in this study (Table 2). However, we found that the localized NPA treatment had little effect on the root growth in the OsPIN2 over-expression lines (Table 2). This suggests that the higher level of OsPIN2 expression had partially overcome the inhibition effect of the NPA treatment on the formation of adventitious roots and altered the sensitivities to NPA.

In Arabidopsis, it has been shown that AtPIN1 and AtPIN2, which are distinct representatives of PIN family members, can functionally replace each other in planta when expressed in the same cells and localized on the same side of the cell (Wiśniewska et al., 2006; Zhang et al., 2010). Although there are at least 12 PIN family members in the rice genome (Wang et al., 2009), our results demonstrated that the expression level of OsPIN2 correlated with tiller angle and numbers (Figure 1). While abundant expression of OsPIN1b was necessary to keep rice architecture compact (Xu et al., 2005), OsPIN2 and OsPIN1b show opposing roles in the control of rice architecture.

The tiller angle of rice control photosynthetic efficiency and planting density therefore determines the final yield (Fu and Zuo, 2007). In our pot experiment which gave no space limits to the spreading growth of O1 and O2 plants, the transgenic plants had higher yields than WT (Table 1). In our field plot trail (1.3 × 1.3 m), the grain yield of O1 plants was not significantly affected by planting density decreased from 59 (15 × 15 cm) to 46 (15 × 20 cm) plants/m2 (Table S4). In the field, neither the extreme-spreading nor the compact rice type was actually beneficial for grain production (Wang and Li, 2008). Genetic manipulation of OsPIN2 has the potential to regulate agronomical valuable traits.

Auxin distribution was altered by OsPIN2 over-expression

Indole-3-acetic acid exists in free and conjugated forms, corresponding to the bioactive and stored hormones, respectively. Local free IAA is bioactive and contributes to the regulation of plant growth and development (Ljung et al., 2002; Bajguz and Piotrowska, 2009). We measured the free IAA level to determine whether the phenotypes we observed were caused by the distribution of free IAA (active IAA). In higher plants, free IAA concentration is usually very low, often in the nanomolar range (Woodward and Bartel, 2005). In 12-day-old whole seedlings of Arabidopsis, free IAA levels were around 80 ng/g (Sun et al., 2010). Kai et al. (2007) found in 2-week-old rice seedlings that the free IAA level was around 35 ng/g and was higher in the aerial parts when compared with the roots. In this study, we examined eight parts of the 60-day-old seedling to see the distribution pattern of the free IAA. The free IAA level varied in the range of 100–4000 μg/g in the different tissues (Figure 3a–d). In the aerial parts, the level of free IAA in the OsPIN2 over-expression seedlings was much lower than in WT (Figure 3a, b), while the difference between transgenic seedlings and WT was the opposite, higher in roots and at the root–shoot junction (Figure 3c, d).

In Arabidopsis, AtPIN2 localizes to the basal side of the cortical cells and to the apical (shoot-facing) side of the epidermal and root cap cells, and it mainly regulates root basipetal auxin transport (Feraru and Friml, 2008). The reversed IAA distribution pattern in shoots, roots and at the root–shoot junction between WT and the transgenic plants may be caused by the disturbed polar subcellular localization of OsPIN2. Our results showed that the content of free IAA in either WT or OsPIN2 over-expression plants decreased from apical to base of the roots. In the tips and middle segment of the roots, the transgenic plants contained several fold higher free IAA than WT (Figure 3), confirming that in rice roots, basipetal auxin transport was also enhanced by OsPIN2 over-expression.

The rice lazy1 mutant exhibited a tiller-spreading phenotype through greatly enhanced polar auxin transport and thus altering the endogenous IAA distribution in shoots (Li et al., 2007). In the lazy1 mutant with DR5::GUS, the GUS expression had moved down to the basal part of the mutant coleoptiles, indicating that altered endogenous IAA distribution in the mutant plants contributed to the spreading growth of the tillers (Li et al., 2007). In the OsPIN2-over-expressing plants, a different GUS staining pattern could be seen in leaf tip, root–shoot junction and three different parts of the roots (Figure 4). The observations of DR5::GUS staining were in accordance with the measured free IAA levels in the root–shoot junction and the roots. At the root–shoot junction, the transgenic plants showed broader GUS staining without the cell specificity in comparison with controls (Figure 4a–c), while in the root tip, there was more obvious GUS staining in the elongation zone relative to control plants (Figure 4d,g,j).

In the shoots, except the tip areas, the GUS staining (Figure 4B–O) did not match well with the measurements of free IAA (Figure 3a, b). This disparity might be caused by one fact that the rice leaves had to be cut into pieces before GUS staining because the hard silicon-enriched walls and cuticle prevents the substrate from entering into leaf cells. The wounded edge of the leaf allowed more substrate into cells for the GUS reaction than the non-wound area (Figure 4B–O).

The GUS expression profile in O1 and O2 illustrated that enhanced basipetal polar auxin transport was the main reason for the change in endogenous IAA distribution in the OsPIN2-over-expressing plants. From the changes of both DR5 activity and free IAA concentration in the different organs, we conclude that the altered architecture of by OsPIN2 over-expression was caused by auxin redistribution, especially at the root–shoot junction.

Tiller angle was increased by auxin transport in OsPIN2-over-expressing plants through suppression of OsLazy1

Tiller angle is an important agronomic trait that contributes to the architecture of the rice canopy (Xu et al., 1998). Previous physiological studies suggested that tiller angle was controlled by gravitropism, which was regulated by auxin and its polar transport (Abe et al., 1996). An Atpin2 mutant displayed an agravitropic phenotype with the reduced root elongation (Müller et al., 1998). The Arabidopsis bushy and dwarf1 (bud1) mutants are deficient in polar auxin transport, and they showed an altered architecture with more branches and had lost apical dominance (Dai et al., 2006). Recent studies showed that the Arabidopsis UPRIGHT ROSETTE (URO) mutant displays bushy plant architecture because of a decreased basipetal IAA transport (Prusinkiewicz et al., 2009; Shimizu-Sato et al., 2009; Sun et al., 2010). Our OsPIN2-over-expressing plants illustrated that in rice, the enhanced IAA transport to the root–shoot junction also resulted in increased tiller angle and number. Therefore, both in monocot and in dicot plants, auxin transport is required for tiller gravitropism and architecture.

In rice, OsPIN1 promotes auxin transport (Yu et al., 2007), and OsLazy1 inhibits auxin transport (Li et al., 2007). Suppression of OsPIN1 expression resulted in a wider tiller angle (Yu et al., 2007), whereas mutation of OsLazy1 led to the same phenotype (Li et al., 2007). The question of whether auxin transport promotes or inhibits the spreading growth habit has not been addressed previously. Yoshihara et al. (2007) suggested that there were two signalling pathways including LAZY1-dependent and -independent gravity responses in rice shoots, and LAZY1 might interact with the PIN proteins directly or indirectly in regulating lateral translocation of auxin. When OsPIN2 was over-expressed, OsLazy1 was obviously down-regulated while OsTAC1 and OsPIN1 expression was little changed (Figure 6). The results suggest that the increased tiller angle and numbers produced by OsPIN2 over-expression were through down-regulation of OsLazy1, and OsPIN2 did not interact with OsPIN1 and OsTAC1. The predicted interaction between OsPIN2 and OsLazy1 gives us more insight into the molecular mechanism of canopy branching in rice.

In addition to PIN family genes, a number of the p-glycoprotein ABC transporters were reported to catalyse the cellular auxin efflux, such as AtABCB1 and AtABCB19 (Noh et al., 2001; Geisler et al., 2005; Nagashima et al., 2008). We did transcriptional analyses for other OsPINs and rice orthologue genes of AtABCB1 and AtABCB19 in the roots and leaves of WT, O1 and O2. However, there was little change of these gene expressions in the transgenic lines (Figures S3 and S4).

In conclusion, OsPIN2 is a functional auxin efflux transporter and is involved in LAZY1-dependent gravity responses of rice shoots. The transporter plays an important role, distinct from OsTAC1 and OsPIN1, in control of auxin-dependent canopy architecture. OsPIN2 can be directly used in the molecular breeding of rice cultivars with improved canopy architecture in the future.

Experimental procedures

Construction of vectors and rice transformation

A 1.8-kb cDNA fragment of a full-length OsPIN2 gene was amplified from Oryza sativa L. ssp. Japonica cv. Nipponbare cDNA, using the primers listed in Table S1. PrimeSTAR HS DNA Polymerase (TaKaRa Biotechnology Co., Ltd, Dalian, China) was used to amplify the OsPIN2 gene, and the two-step PCR parameters were 95 °C for 5 min followed by 98 °C for 10 s, 68 °C for 2 min (30 cycles) and 72 °C for 10 min. The BamHI and SpeI restriction sites were added in the 5′ end of forward and reverse primers, respectively, to facilitate cloning into the expression vector p1390–Ubi (Chen et al., 2004). The constructed plasmid was named as pUbi::OsPIN2 and confirmed by sequencing. The construct was then transformed using Agrobacterium tumefaciens (strain EHA105) as described by Upadhyaya et al. (2000). After 50 mg/L hygromycin (Roche, http://www.roche.com/index.htm) screening, T0 generation transgenic plants were grown for seed. The T-DNA insertion numbers were tested in T1 generation transgenic plants by real-time quantitative PCR (Bubner and Baldwin, 2004; Yang et al., 2005), and two low copy number lines (O8-4, O20-1) were selected from five positive lines and designated as O1 and O2, respectively (Table S2). Furthermore, the copy numbers in T2 generation of O1 and O2 lines were confirmed by Southern blotting (DIG High Prime DNA Labeling and Detection Starter Kit I; Roche). The fragment of the coding sequences of Hyg genes labelled with digoxigenin was used as a probe that was prepared by PCR according to the supplier’s instructions (Roche). Thermal asymmetric interlaced (tail-) PCR (Genome Walking Kit; http://www.takara.com.cn) was used to detect the T-DNA insertion site of O1 and O2 lines.

For detecting the IAA distribution patterns of WT, O1 and O2 plants, the pDR5::GUS construct was transformed into WT and the T2 generation of O1 and O2 plants using Agrobacterium tumefaciens (strain EHA105) was used for GUS staining. The pDR5::GUS construct was kindly provided by Professor Ping Wu’s group of Zhejiang University.

Plant materials, growth condition, NPA treatment and measurements of architecture parameters

The seed sterilization procedure was as described previously (Chang et al., 2009). IRRI nutrient solution contained 1.25 mm NH4NO3, 0.3 mm KH2PO4, 0.35 mm K2SO4, 1 mm CaCl2, 1 mm MgSO4, 0.5 mm Na2SiO3, 20 μm EDTA-Fe, 9 μm MnCl2, 20 μm H3BO3, 0.77 μm ZnSO4, 0.32 μm CuSO4 and 0.39 μm (NH4)6Mo7O24. The solution pH was adjusted to 5.5, and the solution was replaced every 2 days. The germinated rice seeds were planted in a hydroponic culture containing a half-strength nutrient solution in the first week and a complete nutrient solution beyond that point. Plants were grown in a greenhouse under 14 h light at 26–32 °C and 10 h dark at 18–22 °C. For determining gene expression and endogenous free IAA in roots and shoots, WT, O1 and O2 seedlings were grown for 60 days in IRRI solution; 21-day-old seedlings of WT, O1 and O2 were used to analyse the effects of the polar auxin transport inhibitor, NPA, on the plant growth. Localized application of NPA was performed through dispensing diluted agar containing 20 μm NPA directly from the pipette across the root–shoot junction. Untransformed control was used as WT in this research. Five replicates were used in all experiments.

The total root length and lateral root number on the adventitious root were measured using the scanner-based image analysis system WinRhizo (Regent Instruments, Montreal, QC, Canada). The tiller angle was measured between the outermost tiller in the left side and the outermost tiller in the right side as described by Jin et al. (2008).

Semi-quantitative RT-PCR analysis

Total RNA from the roots and leaves was isolated using TRIzol reagent (Invitrogen, http://www.invitrogen.com). The first-strand cDNA was synthesized with an oligo (dT)-18 primer and reverse transcriptase (Fermentas, Hanover, MD). RT-PCR was performed using the gene-specific primers shown in Table S3. The PCR parameters for the detection of OsActin, OsLazy1, OsTAC1, OsPINs, OsAUX1, OsABCB1 and OsABCB19 were 95 °C for 5 min, followed by 25, 35, 33, 30–40, 28, 30 and 30 cycles of 94 °C for 30 s, 55 °C for 30 s, 72 °C for 30 s respectively and 72 °C for 5 min.

Real-time quantitative PCR analysis

The quantitative PCR was performed using MyiQ Single Color Real-time PCR system (Bio-Rad, Hercules, CA) with SYBR Premix Ex Taq (Takara). Each reaction contained 1 μL the first-strand cDNA as template, in a total volume of 20 μL reaction mixture. Primer sets for OsPIN2 were: 5′-CAACACCTACTCCAGCCTC-3′ and 5′- TGGACCAGTCAAGAACCTC -3′; and for OsActin were: 5′-TTATGGTTGGGATGGGACA-3′ and 5′-AGCACGGCTTGAATAGCG-3′, respectively. The amount of template RNA and the cycle number, which provided a linear range of gene amplification, were determined for all the genes (Chen et al., 2007). The relative expression level of OsPIN2 was shown as a percentage of copies of OsPIN2 to that of OsActin.

Quantification of endogenous IAA

The first and second leaves from shoot top and their sheaths were separated at the lamina joint. Whole roots were cut into three parts: 0–4 cm, 4–8 cm and 8–12 cm from each root tip. Also, 0.5 cm root–shoot junction was removed for analysis from each tiller. Samples were weighed to give the fresh weight and put into liquid nitrogen immediately. The sample preparation and measurement of free IAA by HPLC were carried out according to the method described by Lu et al. (2008). A standard IAA sample was obtained from Sigma-Aldrich (St. Louis, MO).

GUS Staining

The same samples as analysed for IAA quantification were used for histochemical GUS staining (Ai et al., 2009). Green tissues were treated with ethanol prior to observation to remove the chlorophyll pigmentation. The stained tissues were photographed using an OLYMPUS MVX10 stereomicroscope, with a colour CCD camera (Olympus, http://www.olympus-global.com).


The research was funded by China National Basic Research Program 973 (No. 2011CB100302), the National Natural Science Foundation (No. 30971854 and 31071846), Jiangsu Natural Science Foundation (BK2010440), Qing Lan Project Priority Academic Program Development of Jiangsu Higher Education Institutions and the transgenic project (2008ZX08001-0052, 2009ZX08009-126B). Dr. Tony Miller from John Innes Centre, Norwich Research Park of UK, carefully corrected the manuscript.