The plant-specific G protein γ subunit AGG3 influences organ size and shape in Arabidopsis thaliana

Authors

  • Shengjun Li,

    1. State Key Laboratory of Plant Cell and Chromosome Engineering, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing 100101, China
    2. Graduate School, Chinese Academy of Sciences, Beijing 100039, China
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Yaju Liu,

    1. State Key Laboratory of Plant Cell and Chromosome Engineering, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing 100101, China
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Leiying Zheng,

    1. Centre of Bioenergy, Institute of Botany, Chinese Academy of Sciences, Beijing 100093, China
    Search for more papers by this author
  • Liangliang Chen,

    1. State Key Laboratory of Plant Cell and Chromosome Engineering, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing 100101, China
    2. Graduate School, Chinese Academy of Sciences, Beijing 100039, China
    Search for more papers by this author
  • Na Li,

    1. State Key Laboratory of Plant Cell and Chromosome Engineering, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing 100101, China
    Search for more papers by this author
  • Fiona Corke,

    1. Department of Cell and Developmental Biology, John Innes Centre, Norwich NR4 7UH, UK
    Search for more papers by this author
  • Yaru Lu,

    1. State Key Laboratory of Plant Cell and Chromosome Engineering, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing 100101, China
    2. College of Biological Sciences, Hebei Normal University, Shijiazhuang, China
    Search for more papers by this author
  • Xiangdong Fu,

    1. State Key Laboratory of Plant Cell and Chromosome Engineering, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing 100101, China
    Search for more papers by this author
  • Zhengge Zhu,

    1. College of Biological Sciences, Hebei Normal University, Shijiazhuang, China
    Search for more papers by this author
  • Michael W. Bevan,

    1. Department of Cell and Developmental Biology, John Innes Centre, Norwich NR4 7UH, UK
    Search for more papers by this author
  • Yunhai Li

    1. State Key Laboratory of Plant Cell and Chromosome Engineering, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing 100101, China
    Search for more papers by this author

Author for correspondence:
Yunhai Li
Tel: +86 10 64807856
Email: yhli@genetics.ac.cn

Summary

  • Control of organ size and shape by cell proliferation and cell expansion is a fundamental developmental process, but the mechanisms that set the size and shape of determinate organs are largely unknown in plants.
  • Molecular, genetic, cytological and biochemical approaches were used to characterize the roles of the Arabidopsis thaliana G protein γ subunit (AGG3) gene in organ growth.
  • Here, we describe A. thaliana AGG3, which promotes petal growth by increasing the period of cell proliferation. Both the N-terminal region and the C-terminal domains of AGG3 are necessary for the function of AGG3. By contrast, analysis of a series of AGG3 derivatives with deletions in specific domains showed that the deletion of any of these domains cannot completely abolish the function of AGG3. The GFP-AGG3 fusion protein is localized to the plasma membrane. The predicted transmembrane domain plays an important role in the plasma membrane localization of AGG3. Genetic analyses revealed that AGG3 action requires a functional G protein α subunit (GPA1) and G protein β subunit (AGB1).
  • Our findings demonstrate that AGG3, GPA1 and AGB1 act in the same genetic pathway to influence organ size and shape in A. thaliana.

Introduction

The mechanisms that control organ size are just beginning to be revealed in both animals and plants. In animals, the target of rapamycin (TOR) pathway and the Hippo pathway have been identified as two major pathways of organ size control (Arsham & Neufeld, 2006; Dong et al., 2007; Zeng & Hong, 2008). Although there are parallels between the TOR-signaling pathway in plants and animals (Deprost et al., 2007), many of the factors (e.g. Warts, Hippo, Salvador, and Yorkie) influencing organ size in animals have no obvious counterparts in plants (Wu et al., 2003; Huang et al., 2005). In addition, several plant-specific factors (e.g. BIG BROTHER, PEAPOD and DA1) of organ size control have been identified in Arabidopsis thaliana (Disch et al., 2006; White, 2006; Li et al., 2008). These studies suggest that plants possess novel mechanisms of organ size control. However, the genetic and molecular mechanisms that set the final size of determinate organs remain elusive in plants.

Plant organ growth occurs via an initial proliferative phase in which cell numbers increase, followed by dramatic cell size increases. During later stages of organ growth, increase in cell size is often associated with an increase in ploidy resulting from endoreduplication (Sugimoto-Shirasu & Roberts, 2003). Therefore, plant organ growth is driven by both cell proliferation and cell expansion which partially overlap in time; these processes are assumed to be coordinated (Horiguchi et al., 2006). Several factors that control organ growth by increasing cell proliferation have been described in plants. The Auxin-Regulated Gene involved in Organ Size (ARGOS) promotes organ growth through AINTEGUMENTA (ANT), a transcription factor that promotes cell proliferation (Krizek, 1999; Mizukami & Fischer, 2000; Hu et al., 2003). The GROWTH-REGULATING FACTORS (AtGRFs) and the GRF-INTERACTING FACTORS (AtGIFs) promote cell proliferation in leaves (Kim et al., 2003; Kim & Kende, 2004; Horiguchi et al., 2005; Lee et al., 2009). The A. thaliana cytochrome P450 KLUH/CYP78A5 protein promotes organ growth in a non-cell-autonomous manner (Anastasiou et al., 2007). By contrast, several factors that control organ growth by limiting cell proliferation have also been reported. For example, PEAPOD1 (PPD1) and PEAPOD2 (PPD2) redundantly restrict organ growth by promoting the early arrest of dispersed meristematic cell (DMC) proliferation during organ development (White, 2006). The E3 ligase BIG BROTHER (BB) restricts cell proliferation (Disch et al., 2006). The putative ubiquitin receptor DA1 functions redundantly with BB/EOD1(enhancer of da1-1) to control the final size of organs by restricting the duration of proliferative growth in A. thaliana (Li et al., 2008). The Mediator complex subunit 25 (MED25)/PHYTOCHROME AND FLOWERING TIME 1 (PFT1) has recently been reported to restrict the period of cell proliferation and cell expansion (Xu & Li, 2011), indicating that the transcriptional machinery plays an important role in organ size control. These studies suggest that modulation of the time and location of cell proliferation is one of the key mechanisms for determining organ size in plants.

Plant organ size is also affected by cell expansion. Several factors that control organ growth by influencing cell expansion have been described, such as P450 ROTUNDIFOLIA3 (ROT3), ANGUSTIFOLIA (AN), ARGOS-LIKE (ARL), BIGPETALp (BPEp), and 26S proteasome AAA-ATPase subunit RPT2a (Kim et al., 1998, 1999; Hu et al., 2006; Szecsi et al., 2006; Kurepa et al., 2009; Sonoda et al., 2009; Varaud et al., 2011). Furthermore, many experiments show that cell proliferation and cell expansion can compensate each other to achieve an optimal species-specific organ size (Potter & Xu, 2001; Horiguchi et al., 2006). For example, loss-of-function mutations in ANGUSTIFOLIA3 (AN3) and SWELLMAP (SWP) cause decreased cell numbers associated with increased cell volume in A. thaliana (Kim et al., 1998; Autran et al., 2002). Although the mechanisms that cause compensation are largely unknown, compensation implies an organ-wide mechanism that controls organ growth.

To further understand molecular mechanisms that set the final size of determinate organs, we have sought to identify genes whose loss and gain of function produce opposite effects on organ growth. Here we report that one such gene appears to be an atypical heterotrimeric G protein γ subunit (AGG3) (Chakravorty et al., 2011). G protein signaling participates in a variety of growth and developmental processes in plants and animals (Temple & Jones, 2007; Mizuno & Itoh, 2009). G protein-coupled pathways transmit a signal, via a membrane-bound receptor and heterotrimeric G proteins consisting of Gα, Gβ, and Gγ subunits, to downstream enzymes known as effectors (New & Wong, 1998). Mammals possess 17 Gα, five Gβ, and 12 Gγ genes (Jones & Assmann, 2004), while the A. thaliana genome contains one canonical G protein α subunit (GPA1), one G protein β subunit (AGB1), and at least three G protein γ (AGG) genes (Temple & Jones, 2007; Chakravorty et al., 2011). Mutations in the A. thaliana GPA1 and AGB1 genes cause defects in plant growth and development (Ullah et al., 2001; Trusov et al., 2008). Arabidopsis thaliana AGG3 has been recently shown to affect the guard cell K+ channel, morphological development and ABA responses (Chakravorty et al., 2011), but has no previously identified function in cell proliferation. Our results show that loss of AGG3 function causes small organs with fewer cells, whereas overexpression of AGG3 results in large organs with increased numbers of cells. Genetic analyses demonstrate that the role of AGG3 in organ size and shape control is dependent on functional GPA1 and AGB1. Our findings provide new insights into the role of AGG3 in setting organ size and shape.

Materials and Methods

Plant material and growth conditions

Arabidopsis thaliana (L.) Heynh. ecotype Columbia (Col-0) was the wild-type line used. All mutants, short organ 1-1 (stn1-1)agg3-1 (SALK_018024), stn1-2/agg3-2 (SAIL_164_H09), stn1-3/agg3-3 (SAIL_1209_B01), gpa1-101 (SALK_115996), and agb1-2 (SALK_061896), were in the Col-0 background. All the above mutants were backcrossed into Col-0 for three generations. T-DNA insertions were confirmed by PCR (for primers, see Supporting Information Table S1). Plants were grown in soil at 22°C under long-day conditions (16 h light : 8 h dark).

Morphological and cellular analyses

Detailed protocols for measurements of floral organs, leaves, seeds, and cells, histochemical staining for β-glucuronidase activity, and microscopy are described in Methods S1 and S2.

Kinematic analysis of petal growth was performed according to a previously described method (Disch et al., 2006; Li et al., 2008; Xu & Li, 2011). Briefly, petal primordia were dissected from the open flower (stage 14) backward to the youngest flower bud for which this was manually possible. To investigate the influence of the stn1-1 mutant on cell proliferation, a ProCYCB1;1:CDB-GUS reporter (Anastasiou et al., 2007) was introgressed into the stn1-1 mutant. The total cell number and the number of cells with GUS activity in petals were counted and expressed as a mitotic index (percentage of cells with GUS activity/number of total cells). Each value represents measurements from at least 11 petals.

Vector constructs and plant transformation

A 4378-bp genomic DNA fragment containing the entire At5g20635 coding region and the 1910-bp upstream sequence was subcloned into the pCR8/GW/TOPO TA cloning vector (Invitrogen). This genomic DNA fragment was then subcloned into the Gateway binary vector pMDC164 to generate a gCOM construct (Curtis & Grossniklaus, 2003).

The 35S:AGG3, 35S:GFP-AGG3, 35S:AGG3△192–251, 35S:AGG3△139–251 and 35S:AGG3△1–104 constructs were made using a PCR-based Gateway system. The specific primers for these constructs are shown in Table S1. PCR products were subcloned into the pCR8/GW/TOPO TA cloning vector (Invitrogen) using TOPO enzyme and sequenced. AGG3, AGG3△192–251, AGG3△139–251 and AGG3△1–104 were then subcloned into the Gateway binary vector (pMDC32) containing the 35S promoter or the pMDC43 vector containing the 35S promoter and GFP (Curtis & Grossniklaus, 2003).

A series of AGG3 derivatives with deletions in specific protein domains were generated by following the instruction manual of PfuUltra II Fusion HS DNA Polymerase (Stratagene, US). The primers used are listed in Table S1. These AGG3 derivatives were then subcloned into the binary vector pMDC32 or pMDC43 (Curtis & Grossniklaus, 2003).

The 1910-bp AGG3 promoter was subcloned into SacI and NcoI sites of the binary vector pGreen-GUS to generate the transformation plasmid ProAGG3:GUS. Brassica rapa AGG3 coding DNA sequence (CDS) was subcloned into HindIII and KpnI sites of the binary vector 35SpGreen to generate the transformation plasmid 35S:BrAGG3. The specific primers for the AGG3 promoter and the BrAGG3 gene are AGG3p-F, AGG3p-R, BrAGG31-F and BrAGG3-R, respectively (Table S1).

Detection of GFP fluorescence

GFP fluorescence in petals and leaves was detected using Zeiss LSM 710 NLO confocal microscopy. Cells were treated with 0.8 M mannitol for 1 h to cause plasmolysis.

RT-PCR and quantitative real-time RT-PCR

Total RNA was extracted from A. thaliana roots, stem, leaves, seedlings and inflorescences by using an RNAprep pure Plant Kit (TianGEN, Beijing, China). Reverse transcription (RT)-PCR was performed as described previously (Li et al., 2006). cDNA samples were standardized based on the amount of ACTIN7 transcript using the primers ACTIN7-F and ACTIN7-R (Table S1).

Quantitative real-time RT-PCR analysis was performed with a Lightcycler 480 machine (Roche) using the Lightcycler 480 SYBR Green I Master (Roche). ACTIN2 mRNA was used as an internal control, and relative amounts of mRNA were calculated using the comparative threshold cycle method. The primers used for RT-PCR and quantitative real-time RT-PCR are described in Table S1.

Protein extraction and western blot analysis

Leaves from agg3-2 and transgenic plants expressing 35S:GFP-AGG3 and 35S:GFP-AGG3△108–125 were harvested, and the soluble and membrane proteins were extracted according to the method described previously (Li et al., 2002) with minor modification. Detailed methods for protein extraction and western blot analysis are described in Methods S3.

Results

Isolation and genetic analysis of stn1 mutants

We identified short organ 1 (stn1) mutants in a screen of publicly available T-DNA mutant collections of A. thaliana. stn1-1, stn1-2 and stn1-3 mutants showed a very similar phenotype of small flowers (Fig. 1a–d; Table S2). Progeny of crosses of the three lines demonstrated these to be alleles of one locus. We observed that heterozygous stn1-1/+ plants had the wild-type phenotype and the F2 population showed a segregation ratio of three wild type to one mutant (67 : 21), indicating that stn1-1 is a single recessive mutant. stn1 mutant plants had shorter sepals, petals, stamens, carpels and fruits than wild-type plants, whereas the width of sepals, petals and carpels in stn1 mutants was not greatly altered compared with that observed in wild type (Fig. 1a–c; Table S2). The leaf blades and petioles of stn1 were slightly shorter than those of wild type (Fig. 1d; Table S2). In addition, stn1 mutants exhibited reduced plant height compared with wild type (Table S2). By contrast, the primary root length of stn1 seedlings was comparable with that of wild-type seedlings (Table S2).

Figure 1.

SHORT ORGAN 1 (STN1) increases the period of cell proliferation. (a–d) Flowers (a), petals (b), fruits (c), and sixth leaves (d) of Arabidopsis thaliana Columbia (Col-0) (left) and stn1-2 (right). (e) The size of epidermal cells in the tip, maximum width (MW), and base regions of wild-type and stn1-2 petals (> 80). The open flowers (stage 14) were used to measure the size of epidermal cells. (f) The size of palisade cells in the tip, middle, and base regions of the sixth leaves of wild type and stn1-2 (> 80). (g) The number of epidermal cells in the petal-length (PLD) and petal-width (PWD) directions (> 20), and the number of palisade cells in the leaf-length (LLD) and leaf-width (LWD) directions (> 10). (h) Growth of wild-type, stn1-1, and 35S:STN1#7 petals (mean ± SE). The largest petals of each series are from open flowers (stage 14). The smallest petals are from the youngest buds that could be manually dissected. The next flower bud was marked as the starting point (0 h). The x-axis indicates how long the bud took to develop an open flower. (i) The size of epidermal cells over time in the maximum width region of wild-type, stn1-1 and 35S:STN1#7 petals (mean ± SE). The arrow indicates the start point where epidermal cells in stn1-1 are significantly larger than those in wild type (< 0.01). (j) Mitotic index vs time in wild-type and stn1-1 petals. Values (e, f, g, and j) are given as mean ± SE relative to the respective wild-type values, set at 100%. **, < 0.01 compared with the wild type (Col-0) by Student’s t-test. Bars, (a–c) 1 mm; (d) 5 mm.

STN1 increases the duration of proliferative growth

During organogenesis, the final size of an organ is determined by both cell number and cell size. To assess their contributions to organ size in stn1, we measured the area of epidermal cells in the tip, maximum width and base positions of wild-type and stn1-2 petals. The area of epidermal cells in stn1-2 was not significantly changed compared with wild type (Figs 1e, S1a). Similarly, the area of pallisade cells in the tip, middle and base regions of the sixth leaves was comparable in wild type and stn1-2 (Figs 1f, S1b). These results imply that stn1 mutants affect cell proliferation. To determine the effect of STN1 on cell proliferation, we counted the number of epidermal cells in both the petal-length and petal-width directions. The number of epidermal cells in the petal-length direction in stn1-2 was reduced, whereas the number of epidermal cells in the petal-width direction was not significantly affected (Figs 1g, S1a). In addition, the number of pallisade cells in the leaf-length direction in stn1-2 was reduced, whereas the number of epidermal cells in the leaf-width direction was not significantly changed (Figs 1g, S1b). These analyses show that the changes in cell number account for the effects of stn1 on organ size.

To understand how STN1 affects cell proliferation, we analyzed the growth dynamics of petals in wild-type, stn1-1 and 35S:STN1 plants. Compared with wild type, stn1-1 mutant petals stopped growing earlier, producing a smaller final size (Figs 1h, S1c). By contrast, petals in 35S:STN1 plants continued to grow for a longer time, leading to a larger final size (Figs 1h, S1c). This result implies that STN1 increases the period of cell proliferation. Consistent with this, cells in stn1-1 petals entered the phase of postmitotic expansion earlier than those in the wild type, while cells in 35S:STN1 petals expanded later compared with those in wild-type petals (Figs 1i, S1d). We further measured the mitotic index using a ProCYCB1;1::CDB-GUS reporter fusion in wild-type and stn1-1 petals (Anastasiou et al., 2007). Cells in stn1-1 petals stopped proliferating earlier than those in wild-type petals (Figs 1j, S1e), indicating that STN1 increases the period of cell proliferation.

STN1 encodes a plant-specific heterotrimeric G protein γ subunit, AGG3

stn1-1, stn1-2, and stn1-3 were identified from the sites of T-DNA insertion in the first intron of the gene At5g20635 (Fig. 2a). T-DNA insertions were confirmed by PCR using T-DNA specific and flanking primers (Fig. S2). A plasmid containing wild-type At5g20635 coding DNA sequence (CDS) driven by a 35S promoter was introduced into the stn1-2 mutant. Nearly all transgenic lines exhibited complementation of stn1 phenotypes (Fig. 2b–d). In addition, transformation of stn1-2 with a genomic copy of At5g20635 restored a wild-type phenotype (Fig. 2b,e). Therefore, At5g20635 is the STN1 gene. During the preparation of this article, At5g20635 was reported to encode a plant-specific heterotrimeric G protein γ subunit (AGG3), which is involved in guard cell K+ channel regulation, morphological development and ABA responses (Chakravorty et al., 2011). stn1-1, stn1-2 and stn1-3 mutants are the recently reported agg3-1, agg3-2 and agg3-3 alleles (Chakravorty et al., 2011), respectively. Thus, we use agg3 instead of stn1 and rename the STN1 gene AGG3 in the following text.

Figure 2.

Identification of the SHORT ORGAN 1 (STN1) gene in Arabidopsis thaliana. (a) STN1 gene structure. The start codon (ATG) and the stop codon (TAA) are indicated. Closed boxes indicate the coding sequence, open boxes indicate the 5′ and 3′ untranslated regions, and lines between boxes indicate introns. T-DNA insertion sites (stn1-1, stn1-2, and stn1-3) in the STN1 gene are shown. (b) Flowers of wild type, stn1-2, cCOM#5, and gCOM#3 (from left to right). cCOM is stn1-2 transformed with the wild-type At5g20635 coding DNA sequence (CDS) driven by the 35S promoter. gCOM is stn1-2 transformed with a genomic copy of At5g20635. Bar, 1 mm. (c) Petal length (PL), petal width (PW), and petal area (PA) of wild type, stn1-2, cCOM#5, and cCOM#3. (> 20). (d) The number of epidermal cells in the petal-length (PLD) and petal-width (PWD) directions in wild type, stn1-2, cCOM#5, and cCOM#3 (> 20). (e) Petal length (PL), petal width (PW), and petal area (PA) of wild type, stn1-2, gCOM#3, and gCOM#9 (> 20). The opened flowers (stage 14) were used to measure the size of petals and the number of epidermal cells. Values (c–e) are given as mean ± SE relative to the respective wild-type values, set at 100%. **, < 0.01 compared with stn1-2 by Student’s t-test.

AGG3 was predicted to contain one predicted transmembrane domain (108–126) (http://www.ch.embnet.org/software/TMPRED_form.html), one putative tumor necrosis factor receptor (TNFR)/nerve growth factor receptor (NGFR) family cysteine-rich signature (126–162), the von Willebrand factor type C (VWFC) cysteine-rich modules (126–169; 142–192; 169–216; 192–239), and one INSULIN family signature (234–248) (http://www.expasy.ch/prosite) (Fig. S3). By performing a BLAST search in the databases, we identified 20 AGG3 homologs in 12 plant species (Fig. S4). AGG3 homologs in other plant species show a 20.1–69.1% amino acid sequence identity with AGG3. The homolog in Brassica rapa has the highest amino acid sequence identity with AGG3 (69.1%) (Fig. S3). A major quantitative trait locus (QTL) for rice GRAIN SIZE 3 (GS3) and DENSE AND ERECT PANICLE1 (DEP1)/PANICLE ERECTNESS (qPE9-1) have 29.4% and 22.5% amino acid sequence identities with AGG3, respectively (Fig. S5) (Fan et al., 2006; Huang et al., 2009; Zhou et al., 2009). However, rice GS3 and DEP1/qPE9-1 have been proposed to negatively affect cell proliferation (Fan et al., 2006; Huang et al., 2009; Takano-Kai et al., 2009; Mao et al., 2010), suggesting that A. thaliana AGG3 and its homologs in rice might have divergent functions.

Overexpression of A. thaliana AGG3 and Brassica rapa AGG3 increases organ size

To further characterize AGG3 function, in particular gain of function phenotypes, we expressed AGG3 under the control of the 35S promoter in the Col-0 wild type and isolated 78 transgenic lines. Transgenic plants had significant increases in AGG3 mRNA compared with wild-type plants (Fig. S6a). Most 35S:AGG3 transgenic plants showed larger flowers than wild type (Fig. 3a). Large petals in 35S:AGG3 transgenic plants are attributable to increased cell number in both the petal-length and petal-width directions (Fig. 3d). By contrast, the area of epidermal cells in 35S:AGG3 petals was comparable with that in wild-type petals (Fig. 3e). 35S:AGG3 transgenic plants exhibited longer fruits than wild-type plants (Fig. 3b,f), and the number of seeds per fruit was increased to 112% that of wild type (Fig. 3g). We further measured the area of agg3-1 and 35S:AGG3 seeds. Seeds from 35S:AGG3 transgenic plants were larger than wild-type seeds, whereas agg3-1 produced slightly smaller seeds than wild type (Fig. 3h), indicating that AGG3 is involved in the control of seed size.

Figure 3.

Organ size in 35S:AGG3 transgenic Arabidopsis thaliana plants. (a) Flowers of Columbia (Col-0), 35S:AGG3#6, and 35S:AGG#7 transgenic lines. 35S:AGG3 is the wild type (Col-0) transformed with the wild-type G protein γ subunit (AGG3) coding DNA sequence (CDS) driven by the 35S promoter. (b) Fruits of Col-0, 35S:AGG3#6, and 35S:AGG3#7 transgenic lines (from left to right). (c) Petal length (PL), petal width (PW), and petal area (PA) of Col-0, 35S:AGG3#6, and 35S:AGG3#7 transgenic lines (> 20). The opened flowers (stage 14) were used to measure the size of petals. (d) The number of epidermal cells in the petal-length (PLD) and petal-width (PWD) directions. (e) The area of epidermal cells in Col-0 and 35S:AGG3#7 petals. (f) Fruit length of Col-0, 35S:AGG3#6, and 35S:AGG3#7 transgenic lines (> 20). The mature fruits were used to measure the size of fruits. (g) Seed number per fruit of Col-0, agg3-1, and 35S:AGG3#7 transgenic plants (> 20). (h) Projective seed area of Col-0, agg3-1, and 35S:AGG3#7 transgenic plants (> 80). The seeds were classified into three groups (0.08–0.10, 0.10–0.11, and 0.11–0.13 mm2). Values for each group are expressed as a percentage of the total seed number analyzed. Values (c–g) are given as mean ± SE relative to the respective wild-type values, set at 100%. **, < 0.01 compared with the wild type (Col-0) by Student’s t-test. Bars, (a, b) 1 mm.

A homolog in Brassica rapa shares high amino acid sequence identity with A. thaliana AGG3 (Fig. S3), suggesting that B. rapa AGG3 and A. thaliana AGG3 might have similar functions. If so, one would expect that overexpression of B. rapa AGG3 might rescue the phenotypes of agg3 mutants. To test this possibility, we generated transgenic plants overexpressing B. rapa AGG3 (35S:BrAGG3) in the A. thaliana agg3-2 mutant and isolated 34 transgenic plants. Transgenic plants had high expression levels of BrAGG3 (Fig. S6b). Overexpression of BrAGG3 complemented the phenotypes of agg3-2 (Fig. 4a,b). The number of epidermal cells in 35S:BrAGG3 petals was restored to that in wild-type petals (Fig. 4c). In addition, we expressed BrAGG3 in the Col-0 wild type and found that plants overexpressing BrAGG3 produced larger leaves, longer fruits, and heavier seeds than wild type (Fig. 4d–f). These analyses indicate a similar conserved function between BrAGG3 and AtAGG3.

Figure 4.

Organ size in 35S:BrAGG3 transgenic Arapidopsis thaliana plants. (a) Flowers of Col-0, agg3-2, and 35S:BrAGG3;agg3-2#1. 35S:AGG3;agg3-2 is agg3-2 transformed with the BrAGG3 coding DNA sequence (CDS) driven by the 35S promoter. The small flower phenotype of agg3-2 was rescued by overexpression of BrAGG3. Bar, 1 mm. (b) Petal length (PL), petal width (PW), and petal area (PA) of Col-0, agg3-2, and 35S:AGG3;agg3-2#1 (> 20). (c) The number of epidermal cells in the petal-length (PLD) and petal-width (PWD) directions in Col-0, agg3-2, and 35S:BrAGG3;agg3-2#1 (> 20). (d–f) Leaf area, fruit length and seed weight of Col-0 and 35S:BrAGG3#3. The opened flowers (stage 14) were used to measure the size of petals (b) and the number of epidermal cells (c). Values (b–f) are given as mean ± SE relative to the respective wild-type values, set at 100%. **, < 0.01 compared with the wild type (Col-0) by Student’s t-test.

AGG3 is a temporally and spatially expressed gene

AGG3 transcripts were detected in various tissues by RT-PCR and quantitative real-time RT-PCR analysis, including roots, stems, leaves, seedlings, and inflorescences (Fig. 5a,b). To determine the spatial expression pattern of AGG3 in more detail, we generated transgenic plants containing AGG3 promoter:GUS fusions (ProAGG3:GUS). In 10-d-old seedlings, GUS activity was observed in true leaves, cotyledons and roots (Fig. 5c). However, agg3 mutants exhibited normal primary root growth (Table S2). It is possible that AGG3 might function redundantly with other factors to influence root growth. Higher GUS activity was detected in younger leaves than older ones (Fig. 5c). AGG3 was also expressed in guard cells (Fig. 5d). In flowers, GUS expression was detected in sepals, petals, stamens and carpels (Fig. 5e–h). AGG3 was highly expressed during the early stages of floral organ formation, but levels were dramatically reduced at the later stages of floral organ development (Fig. 5e-h). We further compared the GUS expression patterns of ProAGG3:GUS and ProCYCB1;1:CDB-GUS and found that AGG3 is strongly expressed in the proliferation phase of petal development (Fig. S7). Therefore, GUS histochemical staining indicates that AGG3 is a temporally and spatially regulated gene.

Figure 5.

Expression patterns of G protein γ subunit (AGG3). (a, b) RT-PCR (a) and quantitative real-time RT-PCR (b) analysis of AGG3 gene expression in Arabidopsis thaliana. Total RNA was isolated from 10-d-old seedling roots (R), stems of 40-d-old plants (St), seventh and eighth leaves of 40-d-old plants (L), 10-d-old seedlings (Se), and inflorescences of 40-d-old plants (In). (c–h) AGG3 expression activity was monitored by ProAGG3:GUS transgene expression. Six GUS-expressing lines were observed, and all showed a similar pattern, although they differed slightly in the intensity of the staining. Pictures were taken with a stereomicroscope. (c–h) Histochemical analysis of GUS activity in a 10-d-old seedling (c), guard cells (d), sepals (e), petals (f), stamens (g), and carpels (h). The arrow in (d) indicates the guard cell. Bars, (d) 20 μm; (c, e–h) 1 mm.

The role of the predicted transmembrane domain and the putative CaaX motifs in the intracellular localization of AGG3

AGG3 is predicted to contain a transmembrane domain (Fig. S3), suggesting that AGG3 might be localized in the plasma membrane. To determine the subcellular localization of AGG3, we expressed a GFP-AGG3 fusion protein under the control of the 35S promoter in the A. thaliana agg3-2 mutant. Overexpression of GFP-AGG3 complemented the phenotypes of agg3-2 (Fig. 6a,e). We also expressed a GFP-AGG3 fusion protein under the control of the 35S promoter in wild-type plants. Like 35S:AGG3 plants, transgenic lines overexpressing GFP-AGG3 showed larger flowers than wild type (Fig. 6b,f). These results demonstrate that the GFP-AGG3 fusion protein is functional. As shown in Fig. 6(c), GFP fluorescence in 35S:GFP-AGG3;agg3-2 transgenic lines was localized at the cell periphery. To determine whether GFP-AGG3 was localized in the plasma membrane or cell walls, plasmolysis was induced with a high mannitol level. GFP fluorescence was observed in the shrunken plasma membrane (Fig. 6d), indicating that AGG3 is associated with the plasma membrane. Similarly, fluorescent signals in 35S:GFP-AGG3 transgenic lines were detected as sharp and thin lines surrounding the cell periphery (Fig. S8a). We further prepared membrane and soluble protein fractions from leaves of agg3-2 and 35S:GFP-AGG3;agg3-2 transgenic plants. GFP-AGG3 fusion proteins were visualized by immunobloting with antibodies specific for GFP. Consistent with confocal imaging, GFP-AGG3 fusion proteins were detected in the membrane fraction, but not in the soluble fraction (Figs 6k, S8a).

Figure 6.

Intracellular localization of GFP-AGG3. (a) Flowers of Col-0, agg3-2 and 35S:GFP-AGG3;agg3-2#1 plants (left to right). 35S:GFP-AGG3;agg3-2 is agg3-2 transformed with GFP-AGG3 driven by the 35S promoter. The small flower phenotype of agg3-2 was rescued by overexpression of GFP-AGG3. (b) Petals of Col-0 (left) and 35S:GFP-AGG3#5 (right). Overexpression of GFP-AGG3 increased the petal size of wild-type plants. Bars, 1 mm. (c, d) Subcellular localization of GFP-AGG3 in epidermal cells of 35S:GFP-AGG3;agg3-2#1 petals (c); cells were plasmolysed with 0.8 M mannitol (d). The GFP-AGG3 was detected in the plasma membrane. (e) Petal length (PL), petal width (PW), and petal area (PA) of Col-0, agg3-2, and 35S:GFP-AGG3;agg3-2#1 plants (> 20). (f) Petal length, petal width and petal area of Col-0, 35S:GFP-AGG3#5, and 35S:GFP-AGG3#8 plants (> 20). (g–j) Subcellular localization of GFP-AGG3 (g), GFP-AGG3△108–125 (h), GFP-AGG3△130–141 (i), and GFP-AGG3△192–251 (j) in petal epidermal cells. The red arrows indicate the nuclei, and the white arrows show the vesicle-like spots. (k) Immunoblotting using GFP-specific monoclonal antibody. Leaves from agg3-2 (a3) and 35S:AGG3;agg3-2#1 transgenic plants (a3-T) were used to isolate total soluble proteins and membrane proteins. Total soluble and membrane proteins (5 μg per lane) were separated on a 10% SDS-PAGE gel. The immunoblotting (IB) was carried out with an antibody against GFP (Anti-GFP). The GFP-AGG3 proteins were detected in the membrane fraction in 35S:AGG3;agg3-2#1 transgenic plants (a3-T), but not in the soluble fraction. The opened flowers (stage 14) were used to measure the size of petals (e, f). Values (e, f) are given as mean ± SE relative to the respective wild-type values, set at 100%. **, < 0.01 compared with the wild type (Col-0) by Student’s t-test.

To further determine the role of the transmembrane domain in the subcellular localization of AGG3, AGG3△108–125 without the transmembrane domain was fused to GFP under the control of the 35S promoter. Stable transgenic A. thaliana plants were assessed by fluorescence confocal microscopy. GFP signals of GFP-AGG3△108–125 were detected at the cell periphery and also observed in nuclei and small vesicle-like spots (Figs 6h, S8b–d). To determine whether some of the GFP-AGG3△108–125 proteins were still localized in the plasma membrane, we prepared the membrane protein fraction from leaves of 35S:GFP-AGG3△108–125 transgenic plants. As shown in Fig. S9, GFP-AGG3△108–125 fusion proteins were also detected in the membrane protein fraction. A recent study reported that GFP signals of GFP-AGG3 were detected in the plasma membrane and small vesicle-like spots, but not in nuclei (Chakravorty et al., 2011). In this study, fluorescent signals of GFP-AGG3 were never observed in nuclei and small vesicle-like spots (Fig. 6g). These results indicate that the deletion of the transmembrane domain affected the subcellular localization of AGG3. Interestingly, fluorescence in GFP-AGG3△108–125-expressing lines generally appeared much weaker than in GFP-AGG3 lines (Fig. 6g,h).

In addition, the putative isoprenylation CaaX motifs have been suggested to be critical for the membrane localization of AGG3, but no experimental evidence for this has been produced to date (Fig. 7a) (Chakravorty et al., 2011). To ascertain the role of the putative CaaX motifs in the intracellular localization of AGG3, AGG3△130–141 and AGG3△192–251 were fused to GFP at the N terminus under the control of the 35S promoter. In construct GFP-AGG3△130–141, the first putative CaaX motif had been deleted, and in construct GFP-AGG3△192–251, the second putative CaaX motif was removed (Fig. 7a). Fluorescent signals in GFP-AGG3△130–141 transgenic lines were predominantly limited to the periphery of the cell, but they also showed small vesicle-like spots (Fig. 6i), indicating that the deletion of the first CaaX motif affected the subcellular localization of AGG3. By contrast, GFP signals in GFP-AGG3△192–251 transgenic lines were detected as sharp and thin lines surrounding the cell periphery (Fig. 6j), indicating that the deletion of the second putative CaaX motif did not affect the plasma membrane localization of AGG3. Interestingly, the deletion of the transmembrane domain or the CaaX motifs did not completely disrupt the plasma membrane localization of AGG3, suggesting that the transmembrane domain and the putative CaaX motifs might jointly participate in the plasma membrane localization.

Figure 7.

Functional analysis of the domains in G protein γ subunit (AGG3). (a) Schematic diagram of AGG3 and its deletion mutants. The predicted AGG3 protein contains one predicted transmembrane domain (TM), one putative tumor necrosis factor receptor (TNFR)/nerve growth factor receptor (NGFR) domain (TNFR), four overlapping von Willebrand factor type C (VWFC) modules (four blue lines), one INSULIN motif (I), and two putative CaaX motifs. The N-terminal region has a remote overall similarity to the organ size regulation (OSR) domain of rice GRAIN SIZE 3 (GS3) or the N-terminal region of DENSE AND ERECT PANICLE1 (DEP1)/PANICLE ERECTNESS (qPE9-1) (red line). A series of AGG3 derivatives with deletions in specific protein domains are indicated. (b) Petal length, fruit length, and leaf length of Arabidopsis thaliana Columbia (Col-0), agg3-2, and transgenic lines overexpressing AGG3 and its derivatives with deletions in specific protein domains. **, < 0.01 compared with agg3-2 by Student’s t-test.

Functional analysis of the domains in AGG3

We further asked whether the N-terminal region or the C-terminal domains of AGG3 are required for AGG3 function in organ size control. To address this question, we overexpressed a truncated AGG3△1–104 without the N-terminal region and a truncated AGG3△139–251 without the C-terminal domains in the agg3-2 mutant background, and generated 71 and 63 transgenic lines, respectively (Fig. 7a). None of the 35S:AGG3△1–104 and 35S:AGG3△139–251 transgenic plants could rescue the flower and leaf phenotypes of the agg3-2 mutant (Fig. 7b), indicating that both the N-terminal region and the C-terminal domains are important for AGG3 function.

To further characterize the functions of the domains in AGG3, we generated a series of AGG3 derivatives in which specific protein domains were deleted, including AGG3△35–43 and AGG3△44–52 with deletions in conserved amino acids corresponding to the putative organ size regulation (OSR) domain of rice GS3 (Fig. S5) (Mao et al., 2010), AGG3△108–125 without the transmembrane domain, AGG3△130–141 without the first CaaX motif, AGG3△126–162 without the TNFR domain and the first CaaX motif, AGG3△191–198 with a deletion in the third VWFC, AGG3△229–251 without the putative insulin motif and the second CaaX motif, and AGG3△192–251 without the putative insulin motif, the second CaaX motif and a part of the fourth putative VWFC (Fig. 7a). Complementation studies were performed by transforming constructions directly into agg3-2 homozygous plants and screening for complementation in at least 30 T1 individuals. Complementation was confirmed in subsequent generations and was assessed based on the flower and leaf phenotypes. Unexpectedly, all these constructs complemented the flower and leaf phenotypes of agg3-2 (Fig. 7b), indicating that deletion of any of these domains could not completely abolish the function of AGG3 in organ growth control. Surprisingly, AGG3△108–125 without the transmembrane domain and AGG3△130–141 without the first putative CaaX motif also complemented the phenotypes of the agg3-2 mutant (Fig. 7b). One explanation for this is that GFP-AGG3△108–125 and GFP-AGG3△130–141 are still present in the plasma membrane although the deletion of the transmembrane domain or the putative CaaX motifs affects the subcellular localization of AGG3 (Fig. 6h,i).

The role of AGG3 in organ size and shape control is dependent on a functional AGB1

Arabidopsis thaliana agg3 and agb1 mutants exhibited similar phenotypes such as short fruits, rounded leaves and shortened floral buds (Lease et al., 2001). A recent study demonstrated that AGG3 strongly interacts with AGB1 (Chakravorty et al., 2011), suggesting that AGG3 and AGB1 could act in the same signaling pathway. However, no genetic evidence for this has been obtained to date. To determine the genetic relationships between AGG3 and AGB1, we obtained the agb1-2 mutant (SALK_061896) harboring the T-DNA insertion in the AGB1 gene and generated an agb1-2 agg3-3 double mutant (Fig. S10a,b,e). The agb1-2 mutant exhibited smaller plants than wild type, partially as a result of the shorter length of the petiole (Fig. 8a,b,n) (Trusov et al., 2008). The agg3-3 leaves exhibited similar characteristics to agb1-2 leaves, although agg3-3 leaves displayed a less severe phenotype than agb1-2 leaves (Fig. 8a,b). The agb1-2 agg3-3 double mutant plants were comparable to agb1-2 single mutant plants (Fig. 8a). Similarly, the sixth leaves of agb1-2 agg3-3 were similar to those of the agb1-2 single mutant (Fig. 8b). We further investigated the size and shape of agb1-2 agg3-3 petals and fruits. The agb1-2 mutant formed short petals as a result of a reduced cell number in the petal-length direction (Fig. 8c,g,h). The size of agb1-2 agg3-3 petals was identical to that of agb1-2 petals, and the number of petal cells in agb1-2 agg3-3 was comparable to that in agb1-2 (Fig. 8c,g,h). Fruit length of agb1-2 was similar to that of agg3-3, while agb1-2 fruits were slightly wider than agg3-3 fruits (Fig 8d,i). The agb1-2 agg3-3 double mutant phenocopied the agb1-2 single mutant in fruit size and shape (Fig. 8d,i). Together, our genetic analyses demonstrate that AGG3 and AGB1 act in the same genetic pathway to affect organ size and shape.

Figure 8.

The role of G protein γ subunit (AGG3) in organ size and shape control is dependent on functional G protein α subunit (GPA1) and G protein β subunit (AGB1). (a) 35-d-old Arabidopsis thaliana Columbia (Col-0), agg3-3, gpa1-101, gpa1-101 agg3-3, agb1-2, and agb1-2 agg3-3 plants (from left to right). The flowering stems were removed. (b) The sixth leaves of Col-0, agg3-3, gpa1-101, gpa1-101 agg3-3, agb1-2, and agb1-2 agg3-3 plants (from left to right). (c, d) Petals (c) and fruits (d) of Col-0, agg3-3, agb1-2, and agb1-2 agg3-3 (from left to right). (e, f) Petals (e) and fruits (f) of Col-0, agg3-3, gpa1-101, and gpa1-101 agg3-3 (from left to right). (g) Petal length (PL) and petal width (PW) of Col-0, agg3-3, agb1-2, and agb1-2 agg3-3 (> 20). (h) The number of epidermal cells in the petal-length (PLD) and petal-width (PWD) directions in Col-0, agg3-3, agb1-2, and agb1-2 agg3-3 (> 20). (i) Fruit length (FL) and fruit width (FW) of Col-0, agg3-3, agb1-2, and agb1-2 agg3-3 (> 20). The middle region of fruits was used to measure the fruit width. (j) Petal length and petal width of Col-0, agg3-3, gpa1-101, and gpa1-101 agg3-3 (> 20). (k) The number of epidermal cells in the petal-length and petal-width directions in Col-0, agg3-3, gpa1-101, and gpa1-101 agg3-3 petals (> 20). (l) Fruit length and fruit width of Col-0, agg3-3, gpa1-101, and gpa1-101 agg3-3 (> 20). (m) The length-to-width ratio of the sixth leaves in Col-0, agg3-3, gpa1-101, gpa1-101 agg3-3, agb1-2, and agb1-2 agg3-3 plants. Error bars represent SE (> 20). (n) Petiole length of the sixth leaves of Col-0, agg3-3, gpa1-101, gpa1-101 agg3-3, agb1-2, and agb1-2 agg3-3 plants (> 20). Values (g–l, n) are given as mean ± SE relative to the respective wild-type values, set at 100%. Bars, (a, b) 1 cm; (c–f) 1 mm.

To further understand the interaction between AGB1 and AGG3, we examined the mRNA expression level of AGB1 in wild-type, agg3-2 and 35S:AGG3 transgenic plants. The AGB1 mRNA level in agg3-2 was significantly reduced compared with that in wild type, while the AGB1 mRNA level in 35S:AGG3 was not changed (Fig. S10f). We further asked whether overexpression of AGB1 could rescue the phenotypes of agg3-2. As shown in Fig. S11(a,b), overexpression of AGB1 could not complement the petal size phenotype of agg3-2. One explanation for this is that cells usually employ combinations of AGB1 and other subunits to mediate growth signals.

As plants overexpressing AGG3 produced larger floral organs than wild type, we asked whether overexpression of AGG3 could rescue the phenotypes of agb1-2. To address this question, we crossed agb1-2 with the 35S:GFP-AGG3 line that formed large floral organs and generated 35S:GFP-AGG3;agb1-2 plants. The GFP signal intensities in 35S:GFP-AGG3 and 35S:GFP-AGG3;agb1-2 plants were essentially similar (Fig. S12a,b), indicating that the agb1-2 mutation does not affect the stability of AGG3. Surprisingly, the size of 35S:GFP-AGG3;agb1-2 petals was indistinguishable from that of agb1-2 petals (Fig. S12d), indicating that the large flower phenotype of 35S:GFP-AGG3 is dependent on a functional AGB1.

The role of AGG3 in organ size and shape control is dependent on a functional GPA1

Light-grown gpa1 produced rounded leaves and shortened flowers (Ullah et al., 2001; Trusov et al., 2008), like those observed in agb1 and agg3 mutants. It is noteworthy that some effects of GPA1 deficiency are dependent on the genetic background. For example, gpa1-1 and gpa1-2 exhibited slightly longer fruits and smaller plants than the Ws wild type, while gpa1-3 and gpa1-4 formed shorter fruits and larger plants than the Col-0 wild type (Trusov et al., 2008). To determine whether AGG3 genetically interacts with GPA1, we obtained the gpa1-101 homozygous line (SALK_115996) harboring the T-DNA insertion in the GPA1 gene (Fig. S10c,d). RT-PCR analysis revealed that gpa1-101 had no detectable transcripts of the GPA1 gene (Fig. S10e). The gpa1-101 mutant formed slightly larger plants and longer petioles than wild type (Fig. 8a,b,n), like those seen in gpa1-4 (Trusov et al., 2008). The gpa1-101 agg3-3 double mutant plants were reminiscent of the gpa1-101 single mutant plants (Fig. 8a). The petiole length of gpa1-101 agg3-3 was the same as that of the gpa1-101 single mutant (Fig. 8b,n). The gpa1-101 mutant formed short but wide petals as a result of an altered cell number in both the petal-length and petal-width directions (Fig. 8e,j,k). Petal size and petal cell number in gpa1-101 agg3-3 were more like those measured in gpa1-101 (Fig. 8e,j,k). The gpa1-101 mutant fruits were slightly shorter than wild-type fruits, but longer than agg3-3 fruits. Fruit length of gpa1-101 agg3-3 was indistinguishable from that of the gpa1-101 single mutant (Fig. 8f,l). Together, these results indicate that AGG3 and GPA1 function in the same genetic pathway to control organ growth.

We then investigated the mRNA expression level of GPA1 in wild-type, agg3 and 35S:AGG3 transgenic plants. The GPA1 mRNA level in agg3 was significantly decreased compared with that in wild type, while the GPA1 mRNA level in 35S:AGG3 was not changed (Fig. S10g). We further expressed GPA1 under the 35S promoter in the agg3-2 mutant and observed that overexpression of GPA1 could not rescue the petal size phenotype of agg3-2 (Fig. S11a,c). One possible reason for this is that combinations of GPA1, AGB1 or other subunits are required to mediate organ growth.

To further confirm the genetic interaction between GPA1 and AGG3, we crossed gpa1-101 with the 35S:GFP-AGG3 line and generated 35S:GFP-AGG3;gpa1-101 plants. The GFP signal intensity in 35S:GFP-AGG3;gpa1-101 plants was similar to that in 35S:GFP-AGG3 plants (Fig. S12a,c), indicating that gpa1-101 does not affect the stability of AGG3. The size of 35S:GFP-AGG3;gpa1-101 petals was comparable to that of the gpa1-101 single mutant (Fig. S12d), indicating that the altered organ size in 35S:GFP-AGG3 is dependent on a functional GPA1.

Discussion

In this study we identified the role of AGG3 in organ growth control in A. thaliana. Loss-of-function mutants in AGG3 formed small organs, whereas plants overexpressing AGG3 produced large organs (Figs 1, 3), indicating that AGG3 positively influences organ growth. The size of organs is determined by cell proliferation and cell expansion in plants (Sugimoto-Shirasu & Roberts, 2003; Horiguchi et al., 2006). Our results show that changes in cell number are related to the effects of AGG3 on organ size and shape (Fig. 1g). Higher expression of AGG3 was detected in younger organs than in older ones (Fig. 5c,e-h), supporting the proposed role of AGG3 in promoting cell proliferation. However, expression of AGG3 was also detected in some tissues without proliferative growth, such as elongation zones in roots and vascular tissues (Fig. 5c), suggesting that AGG3 might be involved in other unrecognized biological processes. Several organ size regulators influence the transition from proliferation to expansion in A. thaliana, resulting in a shorter or longer period of proliferative growth (Krizek, 1999; Mizukami & Fischer, 2000; Hu et al., 2003; Disch et al., 2006; Anastasiou et al., 2007; Li et al., 2008). AGG3 affects organ growth by increasing the duration of proliferative growth (Fig. 1h–j), suggesting that modulation of this period is a crucial point of organ growth control.

AGG3 has been proposed to be an atypical G protein γ subunit (AGG3) involved in guard cell K+ channel regulation and ABA responses (Chakravorty et al., 2011). The agg3 mutants exhibited hyposensitivity in ABA inhibition of light-induced stomatal opening, whereas ABA-mediated promotion of stomatal closure in the agg3 mutants was similar to that in wild type (Chakravorty et al., 2011). By contrast, agg3 mutants were hypersensitive to ABA during germination (Chakravorty et al., 2011). ABA signaling has the potential to link environmental inputs, such as stress, to organ growth, and ABA is also required for normal plant growth (Cheng et al., 2002; Finkelstein et al., 2002). Interestingly, the A. thaliana da1-1 mutant with large organs as a result of increased cell proliferation exhibited ABA-insensitive growth (Li et al., 2008). Thus, it is possible that AGG3 may mediate ABA inhibition of proliferative growth during organogenesis.

AGG3 has recently been shown to strongly interact with A. thaliana G protein β subunit AGB1 (Chakravorty et al., 2011), but genetic relationships between AGG3 and AGB1 or GPA1 were not determined. In this study, we generated agb1-2 agg3-3 and gpa1-101 agg3-3 double mutants and overexpressed AGG3 in agb1-2 and gpa1-101 mutant backgrounds (Fig. 8), respectively. Our genetic analyses demonstrate that the role of AGG3 in organ size and shape control is dependent on a functional AGB1. Similarly, gpa1-101 is epistatic to agg3-3, and a functional GPA1 is required for the large organ size phenotype of 35S:GFP-AGG3 plants. These results confirm that AGG3, GPA1, and AGB1 act in the same genetic pathway to influence organ size and shape. In addition, expression levels of AGB1 and GPA1 in agg3-2 were reduced (Fig. S10f,j), further supporting their genetic interactions. However, overexpresson of GPA1 or AGB1 could not rescue the phenotypes of agg3-2 (Fig. S11). One explanation for this is that combinations of GPA1, AGB1 or other subunits might be required to mediate organ growth.

The localization of G proteins to the cytoplasmic face of the plasma membrane is critical to their signal function in animal cells (Magalhaes et al., 2011). Arabidopsis thaliana AGB1 is localized to the cytoplasmic face of the plasma membrane by interacting with G γ subunits. GFP-AGG3 was localized in the plasma membrane of epidermal cells (Fig. 6c,d), consistent with a recent report that AGG3 strongly interacts with AGB1 (Chakravorty et al., 2011). AGG3 contains a predicted transmembrane domain and two putative CaaX motifs that are potential targets for protein prenylation and S-acylation, which serve as an important secondary signal to target proteins to the plasma membrane (Zeng et al., 2007; Chakravorty et al., 2011). However, the role of these domains in plasma membrane localization was not investigated in a previous study (Chakravorty et al., 2011). GFP-AGG3△108–125 without the transmembrane domain was localized in the plasma membrane, nuclei and small vesicle-like spots (Fig. 6h), indicating that the transmembrane domain is required for the subcellular localization of AGG3. GFP signals of GFP-AGG3△130–141 without the first putative CaaX motif were also detected in the plasma membrane and small vesicle-like spots (Fig. 6i). In a previous study, small vesicle-like spots have been observed in GFP-AGG3 transgenic plants (Chakravorty et al., 2011). However, we did not observed small vesicle-like spots in GFP-AGG3 transgenic plants in this study (Fig. 6d). It is possible that small vesicle-like spots might result from GFP-AGG3 proteins that fail to properly localize to the plasma membrane. By contrast, the deletion of the second putative CaaX motif did not affect the plasma membrane localization of AGG3 (Fig. 6j). Interestingly, the deletion of any of the transmembrane domain or the putative CaaX motifs could not completely abolish the plasma membrane localization of AGG3, suggesting that the transmembrane domain and the putative CaaX motifs jointly affect the plasma membrane localization of AGG3. The predicted transmembrane domain may play a predominant role in the subcellular localization of AGG3.

Proteins that share significant similarity with AGG3 are found in plants. Previous studies reported that its rice homologs GS3 and DEP1/qPE9-1 have a predicted phoshatidylethanolamine-binding protein (PEBP) domain at their N-terminuses (Fan et al., 2006; Huang et al., 2009). Recently, the putative PEBP domain was no longer predicted by analysis of the GS3 sequence (Mao et al., 2010). This region of GS3 was renamed the OSR domain, which is both necessary and sufficient to repress seed size in rice (Mao et al., 2010). The N-terminal region of A. thaliana AGG3 has a remote overall similarity to the OSR domain of GS3 or the N-terminal region of DEP1/qPE9-1 (Fig. S5). However, the N-terminal region of AGG3 (AGG3△139–251) could not complement the phenotypes of agg3-2 (Fig. 7b), indicating that the N-terminal region is not sufficient to promote organ growth. Similarly, the C-terminal domains also could not rescue the phenotype of agg3-2 (Fig. 7b). To investigate the roles of domains in AGG3, a series of AGG3 derivatives with deletions in specific domains were used to complement the agg3-2 mutant. AGG3△35–43 and AGG3△44–52 with deletions in conserved amino acids of the putative OSR domain complemented the phenotypes of agg3-2 (Fig. 7b), indicating that these conserved amino acids are not fully required for the function of AGG3 in organ growth. The deletion of any of the predicted transmembrane domain or either of the two putative CaaX motifs did not completely disrupt the function of AGG3. It is possible that their respective deletions did not completely disrupt the plasma membrane localization of AGG3, although the deletion of the predicted transmembrane domain or the first putative CaaX motif caused ectopic localization of AGG3 (Fig. 6h,i). Similarly, all other AGG3 derivatives with deletions in C-terminus also rescued the phenotypes of agg3-2 (Fig. 7b). The C-terminal region is extremely rich in cysteine residues and is predicted to contain four overlapping VWFC domains, which may be redundantly required for the function of AGG3 in organ growth. Consistent with this hypothesis, the deletion of any of VWFC domains did not dramatically affect the function of AGG3 (Fig. 7b), while the deletion of the whole C-terminal region completely abolished the function of AGG3 in organ growth (Fig. 7b).

Rice GS3 was identified as a regulator of grain length and weight, and loss of GS3 function resulted in large seeds with more cells (Fan et al., 2006; Takano-Kai et al., 2009; Mao et al., 2010). DEP1 is a major yield QTL, and the dep1 allele enhanced meristematic activity and promoted cell proliferation (Huang et al., 2009). These studies suggest that rice GS3 and DEP1/qPE9–1 may function as negative regulators of cell proliferation. By contrast, DEP1/qPE9–1 has been proposed to influence cell expansion, rather than cell proliferation (Zhou et al., 2009). In addition, transient expression of the DEP1/qPE9–1–GFP fusion protein in rice protoplasts was reported to be localized in the membrane (Zhou et al., 2009), whereas GFP signals were detected in the nuclei in DEP1/qPE9–1–GFP transgenic rice plants (Huang et al., 2009). Unexpectedly, our data clearly show that A. thaliana AGG3 positively affects cell proliferation. In fact, loss-of-function mutants in the rice G protein α subunit showed different phenotypes from the A. thaliana gpa1 mutants (Ueguchi-Tanaka et al., 2000; Ullah et al., 2001; Trusov et al., 2008). It is plausible that the functional divergence between A. thaliana AGG3 and its rice homologs might be related to their intracellular localization.

The final size of seeds and organs is an important agronomic trait that affects seed yield and biomass. Overexpression of AGG3 substantially increased fruit length, seed number per fruit and seed size of A. thaliana wild-type plants (Fig. 3e,f,g). As a factor influencing organ growth, AGG3 (and its orthologs in crops such as soybean (Glycine max) and oilseed rape (Brassica napus)) could be used to engineer high seed yield and increased biomass in these key crops.

Acknowledgements

We thank Michael Lenhard for ProCYCB1;1:CDB-GUS seeds and the Nottingham Arabidopsis Stock Centre (NASC) and the Arabidopsis Biological Resource Center (ABRC) for stn1, gpa1 and agb1 seeds. This work was supported by grants from the National Basic Research Program of China (2009CB941503) and the National Natural Science Foundation of China (91017014; 30921003) to Y.H.L. and the National Natural Science Foundation of China (31000531) to Y.J.L., and EC Contract FP6-037704 (Agronomics) awarded to M.W.B.

Ancillary