SPOROCYTELESS modulates YUCCA expression to regulate the development of lateral organs in Arabidopsis


  • Lin-Chuan Li,

    1. National Laboratory for Protein Engineering and Plant Genetic Engineering, Peking-Yale Joint Research Center for Plant Molecular Genetics and AgroBiotechnology, College of Life Sciences, Peking University, Beijing 100871, China;
    Search for more papers by this author
    • *

      These authors contributed equally to this work.

  • Gen-Ji Qin,

    1. National Laboratory for Protein Engineering and Plant Genetic Engineering, Peking-Yale Joint Research Center for Plant Molecular Genetics and AgroBiotechnology, College of Life Sciences, Peking University, Beijing 100871, China;
    Search for more papers by this author
    • *

      These authors contributed equally to this work.

  • Tomohiko Tsuge,

    1. Institute for Chemical Research, Kyoto University, Uji, Kyoto 611-0011, Japan;
    Search for more papers by this author
  • Xian-Hui Hou,

    1. National Laboratory for Protein Engineering and Plant Genetic Engineering, Peking-Yale Joint Research Center for Plant Molecular Genetics and AgroBiotechnology, College of Life Sciences, Peking University, Beijing 100871, China;
    Search for more papers by this author
  • Mao-Yu Ding,

    1. National Laboratory for Protein Engineering and Plant Genetic Engineering, Peking-Yale Joint Research Center for Plant Molecular Genetics and AgroBiotechnology, College of Life Sciences, Peking University, Beijing 100871, China;
    Search for more papers by this author
  • Takashi Aoyama,

    1. Institute for Chemical Research, Kyoto University, Uji, Kyoto 611-0011, Japan;
    Search for more papers by this author
  • Atsuhiro Oka,

    1. Institute for Chemical Research, Kyoto University, Uji, Kyoto 611-0011, Japan;
    Search for more papers by this author
  • Zhangliang Chen,

    1. National Laboratory for Protein Engineering and Plant Genetic Engineering, Peking-Yale Joint Research Center for Plant Molecular Genetics and AgroBiotechnology, College of Life Sciences, Peking University, Beijing 100871, China;
    Search for more papers by this author
  • Hongya Gu,

    1. National Laboratory for Protein Engineering and Plant Genetic Engineering, Peking-Yale Joint Research Center for Plant Molecular Genetics and AgroBiotechnology, College of Life Sciences, Peking University, Beijing 100871, China;
    2. The National Plant Gene Research Center (Beijing), Beijing 100101, China;
    Search for more papers by this author
  • Yunde Zhao,

    1. Section of Cell and Developmental Biology, University of California at San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0116, USA
    Search for more papers by this author
  • Li-Jia Qu

    1. National Laboratory for Protein Engineering and Plant Genetic Engineering, Peking-Yale Joint Research Center for Plant Molecular Genetics and AgroBiotechnology, College of Life Sciences, Peking University, Beijing 100871, China;
    2. The National Plant Gene Research Center (Beijing), Beijing 100101, China;
    Search for more papers by this author

Author for correspondence:
Li-Jia Qu
Tel: 86 10 6275 3018
Fax: 86 10 6275 1841
Email: qulj@pku.edu.cn


  • • Auxin is essential for many aspects of plant growth and development, including the determination of lateral organ shapes.
  • • Here, the characterization of a dominant Arabidopsis thaliana mutant spl-D (SPOROCYTELESS dominant), and the roles of SPL in auxin homeostasis and plant development, are reported.
  • • The spl-D mutant displayed a severe up-curling leaf phenotype caused by increased expression of SPOROCYTELESS/NOZZLE (SPL/NZZ), a putative transcription factor gene that was previously linked to sporocyte formation. The spl-D plants also displayed pleiotropic developmental defects including fewer lateral roots, simpler venation patterns, and reduced shoot apical dominance. The leaf and floral phenotypes of spl-D and SPL over-expression lines were reminiscent of yucca (yuc) triple and quadruple mutants, suggesting that SPL may regulate auxin homeostasis. Consistent with this hypothesis, it was found that over-expression of SPL led to down-regulation of the auxin reporter DR5-GUS, and that many auxin-responsive genes were down-regulated in spl-D leaves. Interestingly, the expression of YUC2 and YUC6, two key genes in auxin biosynthesis, was significantly repressed in spl-D plants.
  • • Taken together with the genetic and phenotypic analysis of spl-D/yuc6-D double mutant, these data suggest that SPL may regulate auxin homeostasis by repressing the transcription of YUC2 and YUC6 and participate in lateral organ morphogenesis.


Auxin is involved in the regulation of almost all aspects of plant growth and development (Woodward & Bartel, 2005). Tremendous progress in the auxin field has enriched our understanding of auxin, including its biosynthesis (Cheng et al., 2006), polar transport (Petrasek et al., 2006; Wisniewska et al., 2006), and signal transduction (Dharmasiri et al., 2005; Kepinski & Leyser, 2005; Tan et al., 2007). Identification of the key components in auxin pathways and characterization of various auxin mutants have made it possible to examine the specific roles of auxin in different developmental processes (Cheng et al., 2007). Recently, it was reported that auxin played an essential role in the determination of the morphology of lateral organs, especially that of leaves and flowers (Cheng et al., 2007; Li et al., 2007). Mutants with defects in auxin signaling or homeostasis often produce curly leaves or leaves with distorted shapes. The mutant auxin-resistant 1 (axr1) of Arabidopsis (Arabidopsis thaliana) produces small and epinastic leaves, which are similar to the wild-type leaves in structure (Lincoln et al., 1990). In contrast, mutation in NON-PHOTOTROPHIC HYPOCOTYL 4/AUXIN RESPONSE FACTOR 7 (NPH4/ARF7) (Harper et al., 2000), SHORT HYPOCOTYL 2/INDOLE-3-ACETIC ACID 3 (SHY2/IAA3) (Tian & Reed, 1999), BODENLOS/INDOLE-3-ACETIC ACID 12 (BDL/IAA12) (Hamann et al., 2002) and AUXIN RESISTANT 3/INDOLE-3-ACETIC ACID 17 (AXR3/IAA17) (Leyser et al., 1996) results in hyponastic (up-curling) leaf phenotypes. We recently identified a gain-of-function mutant of IAA CARBOXYLMETHYLTRANSFERASE 1 (IAMT1), iamt1-D, that exhibited a hyponastic leaf phenotype, whereas knockdown of IAMT1 by RNA interference (RNAi) produced epinastic leaves (Qin et al., 2005). IAMT1 was shown to convert IAA into Methyl-IAA (MeIAA) in vitro (Zubieta et al., 2003; Qin et al., 2005). Although the precise role of MeIAA in planta is not clear, the correlation between leaf curvature and the expression level of IAMT1 suggests that auxin homeostasis plays an important role in the maintenance of flat leaf shape (Qin et al., 2005). This is supported by the fact that many mutants or transgenic plants over-producing auxin possess epinastic leaves, for example rooty (King et al., 1995), superroot 2 (sur2) (Delarue et al., 1998; Barlier et al., 2000) and yucca (Zhao et al., 2001).

Studies of the YUCCA (YUC) genes (YUC1, YUC2, YUC4 and YUC6) that encode the key enzymes in auxin biosynthesis provided direct evidence that auxin homeostasis determines the morphology of lateral organs (Zhao et al., 2001; Cheng et al., 2006, 2007). Although yuc single mutants display no obvious abnormalities, the double, triple, and quadruple mutants of the YUC genes produce progressively crinkled leaves, and impaired but simpler venation patterns, indicating that YUC genes play unique but overlapping roles in auxin biosynthesis to regulate plant development (Cheng et al., 2006). Although the identity and shape of the floral organs are severely affected in these yuc mutants, the severity of the defects is different among the mutants (Cheng et al., 2006). As the expression of more YUC genes was affected, fewer flowers with more severe defects were formed, indicating that auxin is involved not only in flower initiation, but also in determination of flower pattern and morphology (Cheng et al., 2007). Interestingly, the leaf and flower defects in the yuc mutants could be rescued by iaaM, a bacterial auxin biosynthesis gene, but not by application of exogenous auxin (Cheng et al., 2006). Furthermore, these YUC genes displayed different expression patterns in seedlings and flowers (Cheng et al., 2006). Because YUC genes determine the local production of auxin, transcription of these genes is likely to be strictly regulated spatiotemporally. However, the regulation of YUC gene expression is poorly understood.

Another hierarchical regulatory pathway of the curly-leaf phenotype has been intensively studied in Arabidopsis. CURLY LEAF (CLF) is a gene member of the polycomb family and is necessary to repress the expression of a floral homeotic gene AGAMOUS (AG) in leaves (Goodrich et al., 1997). A clf mutant displays a narrow, hyponastic leaf phenotype, similar to that reported when AG was constitutively expressed (Mizukami & Ma, 1992, 1997). Further analysis found that AG was mis-expressed in clf leaves and that CLF interacted with FERTILIZATION-INDEPENDENT ENDOSPERM (FIE) to regulate the transcription of AG (Goodrich et al., 1997; Katz et al., 2004). Recently, the AG protein was shown to be able to bind to the CArG-box-like sequence in the 3′ untranslated region of SPOROCYTELESS (SPL), which directly activated the expression of SPL, indicating that SPL is a direct downstream target of AG (Ito et al., 2004). However, it was not clear whether SPL also played a role in leaf morphogenesis. Previous studies demonstrated that SPL was involved in the promotion of differentiation of the microsporocyte cells and anther-wall cells in the stamens (Schiefthaler et al., 1999; Yang et al., 1999). SPL was also involved in the proximal–distal and adaxial–abaxial pattern formation in ovules (Schiefthaler et al., 1999; Balasubramanian & Schneitz, 2000, 2002).

Here, we report the identification and characterization of a gain-of-function mutant of SPL, spl-D, that produced severely hyponastic leaves and defective flowers. We found that YUC2 and YUC6 expression was repressed in spl-D. Phenotypical analysis of the double mutant spl-D/yuc6-D indicates that SPL may regulate the morphology of lateral organs by repressing auxin production.

Materials and Methods

Plant materials and growth conditions

Arabidopsis thaliana (L.) Heynh ecotype Columbia was used as the wild type in this work. The spl-D mutant was obtained from a T-DNA insertion pool as previously described (Qin et al., 2003). Seeds were surface-sterilized in 15% NaClO solution for 15 min and washed 4–6 times with sterilized water. Sterilized seeds were plated on half-strength Murashige & Skoog (1962) (MS) medium containing 1% sucrose and incubated in darkness at 4°C for 2 d. The plates were then transferred to a culture incubator at 22°C under a 16 : 8 h light:dark photoperiod. For morphological measurement, seedlings were sowed in soil 8 d after germination and placed in a glasshouse at 22°C under long-day conditions as described above. The plants were photographed at the appropriate growth stages. For the root elongation essay on IAA, seedlings were treated following the procedure described previously (Qin et al., 2005).

Histochemistry and observation of vein pattern

To obtain spl-D mutant and 4Enhancer-SPL-2 transgenic lines expressing the DR5-GUS fusion, heterozygous plants were crossed to the DR5-GUS expression line. Plants showing a curly-leaf phenotype in F2 were screened for DR5-GUS expression in young leaves with 1 µM IAA treatment for 4 h and then subjected to GUS staining. GUS staining was performed as described previously (Qin et al., 2005). F3 seeds were harvested from those displaying GUS expression, and lines expressing GUS in all F3 plants were used for subsequent analysis. For histochemical analysis, the progeny of these plants were plated on half-strength MS medium containing DL-phosphinothricin (PPT), and only the resistant seedlings were subjected to GUS staining. To obtain spl-D mutants expressing YUC2-GUS and YUC6-GUS reporter genes, heterozygous plants were crossed to these YUC gene transcriptional reporter lines. Plants showing a curly-leaf phenotype in the F2 were screened for GUS expression in inflorescences by GUS staining as mentioned above. The subsequent screening and analysis procedures were as already described.

For vein-pattern observation, cotyledons and the first pair of true leaves from spl-D mutants and wild-type plants (> 10 plants for each sample) were subjected to analysis. Methods for sample treatment and photography were as described previously (Lin et al., 2005).

Identification of T-DNA insertion site and co-segregation analysis

The flanking sequence of the T-DNA insertion was determined by thermal asymmetric interlaced PCR (TAIL-PCR). The specific and arbitrary degenerate primers used in this study were the same as previously described (Qin et al., 2003). Three primers, P-1, P-2 and P3-1, were designed for co-segregation analysis, with P-1 and P-2 corresponding to Arabidopsis genomic sequences flanking the T-DNA insertion, while P3-1 was from the sequence of the T-DNA insert (Fig. 3a). Primer sequence information is provided in Table 3. The combination of P-1 and P-2 amplified a 455-bp fragment from the wild-type background, while the combination of P3-1 and P-1 amplified a 299-bp fragment from the mutant background. PCR reactions were carried out for 30–36 cycles (94°C for 30 s; 58°C for 30 s; 72°C for 30 s) and PCR products were separated on 1–1.5% agarose gels (Qu et al., 2003).

Figure 3.

SPOROCYTELESS (SPL) over-expression-induced morphology defects in leaves and flowers of Arabidopsis. (a–c) Morphology of wild-type (WT), 4Enhancer-SPL-2 and SPL-RNAi-5 plants. In 4Enhancer-SPL-2 plants, expression of SPL was driven by its own promoter and strengthened by four copies of the Cauliflower mosaic virus (CaMV) 35S enhancer. SPL-RNAi-5 plants were obtained by transformation of a SPL RNA-interference (RNAi) construct into spl-D heterozygous plants. (d) Expression of SPL in wild-type, 4Enhancer-SPL-2 and SPL-RNAi-5 plants as determined by northern blot with 18S RNA as an equal-loading control. (e–h) Leaves and inflorescences of wild-type and 35S-SPL transgenic plants with weakly, moderately and strongly distorted phenotypes. Upper image, leaf morphology; lower image, inflorescence morphology. Bar, 1 cm. (i) Expression of SPL in wild-type and 35S-SPL transgenic plants as determined by reverse transcription–PCR (RT-PCR) with the TUBULIN2 (TUB2) gene as an internal control.

Table 3.  Primer list
Co-segregation and cloning primers
Real-time PCR primers

Over-expression and RNAi constructs and plant transformation

For cloning of the SPL gene, the primer pairs SPL-3 and SPL-4 were designed according to the cDNA sequence from the National Center for Biotechnology Information (NCBI) (accession number NM_118867). The coding sequence of SPL was amplified from flower cDNA of wild-type Arabidopsis using pfu polymerase (Invitrogen, La Jolla, CA, USA) and cloned into the EcoRV site of pBluescript II SK(+) (Stratagene, La Jolla, CA, USA). Recombinant plasmids with the SPL gene in both sense (designated pBSPL) and antisense (designated pASPL) orientations were characterized and sequenced. The XbaI-KpnI fragment of pBSPL was then subcloned into pJim19 (Kan), giving rise to clone pJim19-SPL. To obtain the 4Enhancer-SPL construct, a 1726-bp genomic sequence, which contained part of the SPL promoter region (600 bp upstream of the start codon ATG) and SPL gene sequence, including two introns, was amplified using the primers P-2 and SPL-4. This fragment was subsequently cloned into the EcoRV site of pBluescript II SK (+) (Stratagene) and confirmed by sequencing, giving rise to pBSPLG. The 4Enhancer-SPL was constructed by ligation of the following DNA fragments: the EcoRI-KpnI fragment from pBSPLG, the HindIII-KpnI fragment from pQG110, and the HindIII-EcoRI enhancer tetrad fragment from pA4Enhancer described previously (Qin et al., 2005). SPL-RNAi was constructed by ligation of the four DNA fragments: the XbaI-KpnI fragment from pQG110, the XbaI-SalI SPL fragment from pASPL, the SalI-EcoRI 1-kb GUS fragment from pBGUS, and the EcoRI-KpnI fragment from pBSPL. Transgenic plants were generated by Agrobacterium–mediated transformation by the floral dip method. The pJim19-SPL and 4Enhancer-SPL constructs were transferred into wild-type plants, and T1 generation and progenies were plated on half-strength MS medium containing 50 mg l−1 kanamycin for transformant selection, while the SPL-RNAi construct was transferred into the spl-D heterozygous plants. For SPL-RNAi transgenic line selection, plants with kanamycin resistance were subjected to co-segregation analysis as described above to identify the spl-D homozygous background and the expected plants were used for further analysis.

Northern blot, reverse transcription–PCR (RT-PCR), real-time quantitative PCR analysis, and microarray analysis

Total RNA was isolated from rosette leaves of 2- or 4-wk-old plants using TRIzol reagent (Invitrogen). As previously described (Qu et al., 2003), 15 µg of total RNA was subjected to RNA gel blots. For microarray analysis, aerial parts of 2-wk-old spl-D homozygous and wild-type seedlings were collected and RNA samples were prepared according to the protocols of the manufacturer (Affymetrix). The hybridization and data treatment were as described preciously (Lin et al., 2005). First-strand cDNAs were generated with the Superscript II RNase H-Reverse Transcriptase Kit (Invitrogen) according to the manufacturer's instructions, and then used as the templates for RT-PCR amplification. PCR was performed at 96°C for 2 min, with different numbers of cycles at 94°C for 30 s, 58°C for 30 s and 72°C for 60 s. RT-PCR products were electrophoresed on 1.0–1.5% agarose gels and visualized by staining with ethidium bromide. Real-time quantitative PCR was carried out on a RotorGene 3000 (Corbett Research, Mortlake, Australia) according to the SYBR green detection protocol (SYBR Premix Ex Taq system, TaKaRa). The product amounts were determined in each cycle with the RotorGene software. Differences in cycles during the linear amplification phases between different genes, which were compared with the transcript of Arabidopsis TUBULIN4 (TUB2), were used to examine relative expression levels of the tested genes. The primers used to determine the expression levels of auxin homeostasis-related genes and the internal control TUB2 are listed in Table 3.


Isolation and morphological characterization of the spl-D mutant

A collection of T-DNA activation tagged Arabidopsis lines (Qin et al., 2003) was used to screen for mutants with defects in leaf morphology. Here we report the analysis of one of the mutants, spl-D. Upon selfing, the T2 progenies segregated into three distinctive phenotypes, a wild-type phenotype (Fig. 1a), an intermediate curly-leaf phenotype with normal shoot apical meristem (SAM) (Fig. 1b), and a phenotype with severely up-curling rosette leaves and loss of SAM (Fig. 1c). Genetic analysis showed that the segregation ratio for the plants with these three phenotypes was 76 : 166 : 83, consistent with the expected ratio of 1 : 2 : 1. These results indicated that the curly-leaf phenotypes were caused by a semi-dominant mutation in a single nuclear gene. To confirm this, T3 progeny from the self-pollinated wild-type-like plants and plants with the intermediate phenotype were analyzed. As expected, all the T3 progeny from wild-type-like T2 plants produced only wild-type-like phenotypes, whereas those progeny from the intermediate-phenotype plants segregated into plants with phenotypes of three severities in a ratio of 1 : 2 : 1.

Figure 1.

The phenotype of SPOROCYTELESS dominant (spl-D) mutants of Arabidopsis. (a–c) Morphology of wild-type (WT), spl-D heterozygous, and spl-D homozygous plants. Note the severely upward-curling leaves of the mutant plants and loss of the shoot apical meristem (SAM) in the homozygous plants. All plants photographed were 4 wk old. (d) Inflorescences of wild-type and spl-D heterozygous plants. Fewer buds were produced by spl-D plants than wild-type plants. In the extreme case, only two or three flowers were borne on inflorescences in the mutant. (e) Flowers of wild-type and spl-D heterozygous plants. (f) Siliques of wild-type and spl-D heterozygous plants. (g) Seeds of wild-type and spl-D heterozygous plants. (h) Seven-week-old plants of wild-type and spl-D heterozygous plants.

The spl-D mutant displayed a severely hyponastic leaf phenotype, with rod-like rather than flat leaves (Fig. 1b,c). Compared with wild-type leaves, the leaves of spl-D were narrow, small, and severely up-curling along the longitudinal axes (Supplementary Material Fig. S1). Although the severity of the phenotypes progressively increased in the later-produced leaves, all leaves of spl-D possessed a constant Gaussian curvature of zero (data not shown; Nath et al., 2003). At lower temperature, the degree of curvature of the spl-D leaves decreased, resembling the phenotype of clf (Goodrich et al., 1997). In addition to the curly-leaf phenotype, SAMs were aborted in homozygous spl-D plants after seven or eight rosette leaves had been produced (Fig. 1c). Interestingly, the tissue destined to be SAM was replaced by leaf-like tissues (Supplementary Material Fig. S2) that never bolted, suggesting that transition of the SAM from a vegetative to a reproductive stage was disrupted. The heterozygous spl-D plants produced fewer and smaller flowers (Fig. 1d,e), shorter and wrinkled siliques (Fig. 1f, Table 1), and shriveled seeds (Fig. 1g). Moreover, when grown in soil, 6-wk-old heterozygous spl-D plants were about half of the height of the wild-type plants (Table 1). With age, the heterozygous spl-D plants gradually lost the apical dominance and produced more axillary shoots from the basal part of the rosettes (Fig. 1h). The hyponastic curly-leaf phenotype and loss of apical dominance of the spl-D plants implied that auxin homeostasis may be impaired.

Table 1.  Statistical analysis of Arabidopsis SPOROCYTELESS dominant (spl-D) plants
VariableWild typespl-D/spl-D/spl-D
  • a

    Cotyledons and the fifth rosette leaves were measured at 4 wk after germination.

  • b

    Seedlings grown in darkness for 5 d.

  • c

    Two-week-old seedlings were measured.

  • d

    Leaves were treated with Formalin-Aceto-Alcohol (FAA) solution to soften them, and then expanded for measurement.

  • e

    Measured at 6 wk after germination.

Cotyledon widtha (mm)3.1 ± 0.1 (n = 10)2.5 ± 0.3 (n = 15)2.3 ± 0.5 (n = 10)
Cotyledon length (mm)5.5 ± 0.4 (n = 10)4.5 ± 0.4 (n = 15)4.2 ± 0.5 (n = 10)
Hypocotyl lengthb (mm)10.7 ± 0.7 (n = 10)8.1 ± 0.6 (n = 10)5.8 ± 0.5 (n = 10)
No. of lateral rootsc5.8 ± 0.5 (n = 20)1.1 ± 0.4 (n = 20)
Apparent leaf widtha (mm)10.4 ± 0.5 (n = 5)1.8 ± 0.2 (n = 10) 
Real leaf widthd (mm)9.7 ± 0.3 (n = 5)4.2 ± 0.32.2 ± 0.3 (n = 10)
Leaf length (mm)25.8 ± 0.9 (n = 5)14.9 ± 1.2 (n = 10)11.4 ± 1.8 (n = 10)
Plant heighte (cm)30.95 ± 2.35 (n = 6)15.1 ± 2.39 (n = 10)

Cloning and molecular characterization of spl-D mutants

In order to determine which gene was impaired in spl-D, we adopted TAIL-PCR to amplify and identify the flanking sequence of inserted T-DNA (Qin et al., 2003). We isolated a T-DNA insert at the promoter region of At4g27330 (Fig. 2a), which was previously identified as SPOROCYTELESS/NOZZLE (SPL/NZZ) and shown to be involved in sporocyte formation (Schiefthaler et al., 1999; Yang et al., 1999). Among 325 T3 plants, 76 were wild type, 166 were heterozygous, and 83 were homozygous for the T-DNA insertion (Fig. 2b). All the plants homozygous for the T-DNA insertion showed a severe up-curling leaf phenotype and later lost their SAM (data not shown). These results clearly indicated that the observed phenotypes strictly co-segregated with this T-DNA insertion.

Figure 2.

Identification of the T-DNA insertion site and affected genes in Arabidopsis SPOROCYTELESS dominant (spl-D) plants. (a) Sketch of the genomic region flanking the T-DNA insertion site in spl-D plants. The thick red arrows indicate the four 35S enhancers from pSKI015. The thin red arrows indicate the primers used for the co-segregation analysis. RB, T-DNA right border; 4 × 35Se, Cauliflower mosaic virus (CaMV) 35S enhancer tetrad; BAR, Basta resistance gene; LB, T-DNA leaf border. (b) Co-segregation analysis of the T-DNA insertion and curly-leaf phenotypes. P-1 and P-2 amplified a 455-bp fragment from the wild type, and P-1 and P3-1 amplified a 299-bp fragment from the homozygous spl-D plants. (c) Expression of SPL in wild-type, spl-D heterozygous and spl-D homozygous plants as determined by reverse transcription–PCR (RT-PCR) with the TUBULIN2 (TUB2) gene as an internal control. (d) Expression of SPL in wild-type, spl-D heterozygous and spl-D homozygous plants as determined by northern blot with 18S RNA as an equal-loading control.

Because the T-DNA carried four copies of the Cauliflower mosaic virus (CaMV) 35S enhancer, the expression levels of the genes within approx. 10 kb upstream and downstream of the insertion site were examined by RT-PCR. The results showed that only the expression of SPL (At4g27330) was elevated in the spl-D mutant (Fig. 2c), which was further confirmed by northern blots (Fig. 2d). Therefore we concluded that the phenotypes of the spl-D mutants might be caused by the elevated expression of SPL.

Confirmation by transgenic studies

To determine whether the elevated expression of SPL was responsible for the observed leaf phenotypes, we carried out three experiments. Firstly, SPL, under the control of its own promoter with four copies of the CaMV 35S enhancer, was transformed into wild-type plants. Among 45 T1 transgenic plants, 12 plants displayed an up-curling leaf phenotype, one of which was designated as 4Enhancer-SPL-2 and was subjected to further analysis (Fig. 3b). In addition to the leaf phenotypes, 4Enhancer-SPL-2 plants flowered slightly earlier and produced smaller flowers (Fig. 3b, bottom). Secondly, an RNAi construct of SPL was constructed and transformed into heterozygous spl-D plants. Out of 28 independent T1 transgenic lines obtained, 14 resembled the wild-type phenotype. Genotyping of these 14 lines showed that most of them harbored the kanamycin-resistance gene in the homozygous spl-D background (data not shown). Taking SPL-RNAi-5 as an example, no obvious difference was found in leaf and flower morphology between SPL-RNAi-5 and the wild type (Fig. 3c), although the fertility of SPL-RNAi-5 was reduced (data not shown). Northern blots showed that SPL was highly expressed in 4Enhancer-SPL-2, whereas no SPL mRNA was detected in SPL-RNAi-5 leaves (Fig. 3d). These two functional complementation experiments demonstrate that the spl-D phenotype is indeed caused by the elevated expression of SPL.

We also generated transgenic plants in which over-expression of SPL was driven by a CaMV 35S promoter. Out of 69 T1 transgenic plants, 43 plants had defects in the leaves and flowers (Fig. 3f–h). These transgenic plants are roughly grouped into three groups based on the severity of the phenotypes, that is, a phenotype with slightly wrinkled rosette leaves, an intermediate phenotype with smaller, hyponastic and partially wrinkled leaves, and a phenotype with much smaller and severely hyponastic leaves (Fig. 3f–h). RT-PCR analysis showed that the severity of the leaf and flower phenotypes in these transgenic plants correlated with the expression levels of SPL (Fig. 3i), demonstrating that over-expression of SPL resulted in defective leaf and flower development.

Auxin-related deficiency in spl-D plants

To investigate whether auxin homeostasis was impaired in spl-D, we carried out a hypocotyl elongation assay at high temperature, a condition that stimulates auxin biosynthesis and hypocotyl elongation (Gray et al., 1998). The results showed that the hypocotyls of 9-d-old wild-type seedlings grown at 30°C were significantly longer than those grown at 22°C, whereas hypocotyl elongation at high temperature was impaired in spl-D plants (Fig. 4a,b).

Figure 4.

Auxin-related phenotypes in Arabidopsis SPOROCYTELESS dominant (spl-D) plants. (a, b) Induction of hypocotyl elongation by high temperature. Wild-type (WT) and spl-D mutant seedlings were grown on half-strength Murashige and Skoog (MS) plates at 22 and 30°C for 9 d and photographed at the same magnification. (c, d) Comparison of lateral roots between wild-type and spl-D plants, which were grown on half-strength MS plates for 14 d and photographed at the same magnification. (e–h) Comparison of vein pattern between wild-type (e, g) and spl-D (f, h) plants. Cotyledons (e, f) and first true leaves (g, h) were discolored with 100% ethanol at 37°C for 4 h and then photographed. Bar, 1 mm. (i) Growth of primary roots of wild-type (squares) and spl-D (triangles) plants on vertically oriented culture plates. (j) Growth of the fifth leaves of wild-type (squares) and spl-D (diamonds) plants.

We further examined the lateral roots of spl-D. As shown in Fig. 4(c,d), 2-wk-old spl-D seedlings produced fewer lateral roots than wild-type plants (Table 1), indicating that auxin homeostasis or signaling may be impaired in spl-D (Reed et al., 1998; Xie et al., 2000; Rogg et al., 2001; Fukaki et al., 2002; Gray et al., 2003). When compared with wild type, spl-D had fewer veins in both cotyledons (Fig. 4e,f) and the first pair of rosette leaves (Fig. 4g,h). The vascular defects were similar to the venation patterns of the auxin transport-defective mutant bud1 or auxin biosynthetic yuc triple or quadruple mutants (Cheng et al., 2006; Dai et al., 2006), suggesting that spl-D phenotypes may be caused by disruption of an auxin-related process (Fukuda, 2004; Cheng et al., 2006; Scarpella et al., 2006).

To further investigate the relationship between auxin and spl-D, we examined the growth rate of aerial and underground parts of spl-D plants. The results showed that the growth rate of spl-D roots was slightly reduced (Fig. 4i), and the growth of the fifth leaves was significantly retarded in spl-D plants (Fig. 4j). Because auxin has been considered to play a role in cell proliferation and in the determination of organ size (Lincoln et al., 1990; Hu et al., 2003), our data suggest that the phenotypes of spl-D may be linked to defects in auxin pathways.

Effect on auxin homeostasis in spl-D mutants

The auxin-reporter DR5-GUS line is a marker line commonly used for approximate assessment of auxin concentration and distribution in planta (Ulmasov et al., 1997). To further characterize the effects of SPL over-expression on auxin homeostasis, we crossed the DR5-GUS line with spl-D to analyze the distribution and quantity of auxin (Mattsson et al., 2003; Koizumi et al., 2005; Cnops et al., 2006). Auxin concentration and distribution in spl-D were significantly affected by SPL over-expression. As shown in Fig. 5, in 3-d-old wild-type seedlings, GUS expression was first detected at the junction of the hypocotyl and primary root, and at the junction of the hypocotyl and cotyledons (Fig. 5a), whereas in spl-D seedlings GUS staining was faint at these two junctions (Fig. 5b). In 5-d-old seedlings, GUS expression was gradually concentrated to the marginal region of the cotyledons in the wild type (Fig. 5c), whereas no GUS staining was detected in cotyledons of spl-D seedlings (Fig. 5d). When the first true leaf emerged, GUS staining was focused at the tip of the leaf in the wild type (Fig. 5e), whereas no visible GUS staining was detected in spl-D mutants (Fig. 5f). In the root tips of 8-d-old seedlings, DR5-GUS staining was also stronger in wild-type than spl-D seedlings (Fig. 5g). In inflorescences and flowers, although the expression pattern of DR5-GUS was unchanged, that is, mainly in the anthers at certain specific developmental stages, the expression level was lowered in spl-D (Fig. 5h). In order to confirm this at the molecular level, we examined the expression of some auxin-responsive genes in spl-D mutants by quantitative RT-PCR (Table 2; Mao et al., 2005). Consistent with the weaker DR5-GUS staining phenotype in spl-D mutants, the transcription of these auxin-inducible genes was also significantly suppressed (Fig. 5i). Although the DR5-GUS assay is only a proxy for assessment of auxin concentrations and needs be treated with suitable caution, our DR5-GUS data, together with the quantitative RT-PCR results, suggest that auxin homeostasis or signaling is possibly impaired in spl-D.

Figure 5.

Altered auxin accumulation and distribution in Arabidopsis SPOROCYTELESS dominant (spl-D) plants. (a–f) Comparison of GUS activity in DR5-GUS and DR5-GUS/spl-D plants. Seedlings were photographed 3, 5 and 7 d after germination and at the same magnification. (a, c, e) DR5-GUS seedlings; (b, d, f) DR5-GUS/spl-D seedlings. (g) GUS activity in the root tips of 7-d-old DR5-GUS and DR5-GUS/spl-D seedlings. (h) GUS activity in the inflorescences of 5-wk-old DR5-GUS and DR5-GUS/spl-D plants. (i) Down-regulation of auxin-inducible genes in spl-D (gray bars) leaves (wild-type, black bars).

Table 2.  Transcription of auxin homeostasis and responsive genes were altered in Arabidopsis SPOROCYTELESS dominant (spl-D) plants as revealed by microarray analysis
Probe setLocus no.RegulationDescription
  1. RNA was extracted from aerial parts of 2-wk-old wild-type and spl-D seedlings and then subjected to microarray analysis (Affymetrix ATH1); I, Increased in spl-D; D, decreased in spl-D. Numbers in brackets show the regulation ratio (log2).

Auxin homeostasis-related genes
245244_atAt1g44350I(1.7)IAA-amino acid hydrolase 6
248192_atAt5g54140I(0.7)IAA-amino acid hydrolase homolog (ILL3)
256178_s_atAt1g51780I(0.7)Auxin conjugate hydrolase (ILL5)
248163_atAt5g54510I(1.4)Auxin-responsive GH3 protein, putative (DFL-1)
262263_atAt1g70940D(0.5)Auxin transport protein, putative (PIN3)
Auxin-responsive genes
256010_atAt1g19220I(1.5)Auxin-responsive factor (ARF19)
258399_atAt3g15540I(0.8)Early auxin-induced protein (IAA19)
259790_s_atAt1g29430D(1.8)Auxin-induced protein (SAUR62)
259783_atAt1g29510D(1.8)Auxin-induced protein (SAUR67)
257506_atAt1g29440D(1.7)Auxin-induced protein (SAUR64)
259961_atAt1g53700D(1.6)Auxin-induced protein kinase
259773_atAt1g29500D(1.4)Auxin-induced protein (SAUR65)
264605_atAt1g04550D(1.3)Putative auxin-induced protein (IAA12)
261776_atAt1g76190D(1.2)Hypothetical protein similar to putative auxin-induced protein
265117_atAt1g62500D(1.2)Protease inhibitor/seed storage/lipid transfer protein (LTP) family protein
252972_atAt4g38840D(1.1)Auxin-induced protein-like auxin-inducible SAUR gene (SAUR14)
263664_atAt1g04250D(1.1)Putative auxin-induced protein (IAA17/AXR3–1)
255403_atAt4g03400D(1)Putative GH3-like protein similar to soybean GH3 auxin-inducible protein
250286_atAt5g13320D(1)Auxin-responsive-like protein Nt-gh3 deduced protein
253255_atAt4g34760D(0.8)Auxin-responsive family protein
263656_atAt1g04240D(0.7)Auxin-responsive protein/indole acetic acid-induced protein 3 (IAA3)
249651_atAt5g37020D(0.5)Auxin response factor 8 (ARF8)
265182_atAt1g23740D(0.4)Putative auxin-induced protein
255645_atAt4g00880D(0.3)Auxin-responsive family protein
257769_atAt3g23050D(0.3)Auxin-responsive protein/indole acetic acid-induced protein 7 (IAA7)

We did not observe significant differences between spl-D and wild-type plants in the root growth inhibition assay (Supplementary Material Fig. S3a). We also did not see changes in the expression of the auxin efflux carrier PIN-FORMED gene family members between wild type and spl-D (Supplementary Material Fig. S3b), indicating that spl-D phenotypes probably are not caused by defects in auxin signaling or transport.

Repression of YUC2 and YUC6 expression by SPL

Because of the phenotypic similarities between the yuc quadruple mutants and spl-D mutants, we investigated whether auxin homeostasis genes, for example YUCCA genes, were affected in spl-D mutants. We first determined the transcription levels of YUC1, YUC2, YUC4 and YUC6 in the young leaves of spl-D mutants by quantitative RT-PCR. Interestingly, although the transcription of YUC1 and YUC4 was not significantly affected (data not shown), the mRNA level of YUC2 and YUC6 was reduced to 36% and 25%, respectively (Fig. 6a). We further examined the expression of 23 other auxin homeostasis-related genes involved in auxin biosynthesis or metabolism. AAO1 (AT5g20960), GH3.6 (At5g54510) and three genes encoding IAA amino acid hydrolase (AT1g44350, At5g54140, and At1g51780) were significantly up-regulated in spl-D mutants (Fig. 6a). Furthermore, in the leaves of the 35S-SPL transgenic plants, YUC2 expression decreased as the leaf curvature increased (data not shown). YUC2 and YUC6 were suppressed in the inflorescences of 4Enhancer-SPL-2 and 35S-SPL-middle transgenic lines (Fig. 6b). These data indicated that SPL repressed YUC2 and YUC6 expression in leaves and inflorescences.

Figure 6.

YUC2 and YUC6 repression in Arabidopsis SPOROCYTELESS dominant (spl-D) plants. (a) Transcriptional alternation of auxin homeostasis-related genes in leaves of spl-D (gray bars) plants (wild-type (WT), black bars). YUC2 and YUC6 were down-regulated significantly in spl-D plants. (b) Transcription of YUC2 (black bars) and YUC6 (gray bars) was repressed in the inflorescences of SPL-over-expression plants as determined by reverse transcription–PCR (RT-PCR) with the TUBULIN2 (TUB2) gene as an internal control. (c) Seedlings stained for YUC2-GUS expression in 3-d-old wild-type and spl-D plants. (d, e) Inflorescences stained for YUC6-GUS expression in 6-wk-old wild-type and spl-D plants.

To further determine the effect of SPL activity on the regulation of YUC2 and YUC6, we examined the expression patterns of YUC2 and YUC6 in spl-D plants. In 4-d-old wild-type seedlings, YUC2 was strongly and ubiquitously expressed in cotyledons, especially in the vascular tissues (Fig. 6c, left). Compared with the wild type, YUC2-GUS staining was much weaker in the cotyledons of spl-D seedlings, consistent with our real-time PCR results (Fig. 6c, right). As GUS staining is not detectable in seedlings of the YUC6-GUS reporter line (Cheng et al., 2006), we determined YUC6-GUS expression in the inflorescences and flowers. In wild-type flowers, staining of YUC6-GUS was detected specifically in stamens and pollen (Fig. 6d; Cheng et al., 2006). However, while the pattern alternation of YUC6-GUS staining was observed in spl-D plants, YUC6-GUS staining was also very faint in the anthers of spl-D mutants (Fig. 6e).

Furthermore, we investigated the repression effect of SPL protein on the transcription of YUC2 and YUC6 in the spl/spl mutants (Ler background). In leaves, the expression levels of YUC2 and YUC6 showed no significant difference from those of wild type (data not shown), possibly because of the extremely low expression level of SPL in leaves. In flowers of the spl/spl mutants, however, both YUC2 and YUC6 were up-regulated by c. 1.5-fold (Supplementary Material Fig. S4). The data from both SPL gain-of-function and loss-of-function mutants suggest that the SPL protein is involved in the regulation of YUC expression at the transcription level.

Partial rescue of spl-D phenotypes by yuc6-D

If the decreased expression of YUC2 and YUC6 resulted in similar defects in spl-D plants, there should be similar phenotypes between SPL over-expressor lines and the yuc2 yuc6 double mutant. We found that 35S-SPL transgenic plants displayed extremely low fertility, whereas yuc2 yuc6 double mutants were completely sterile. As shown in Fig. 7(a), stamen elongation was defective in these plants, resulting in the failure of pollination. With age, the yuc2 yuc6 double mutant and 35S-SPL transgenic plants gradually lost their apical dominance (Fig. 7b).

Figure 7.

Partial rescue of SPOROCYTELESS dominant (spl-D) phenotypes by yuc6-D in Arabidopsis. (a) Comparison of inflorescence defects in 35S-SPL-weak and yuc2 yuc6 plants. (b) Both the 35S-SPL-weak transgenic plants and the yuc2 yuc6 double mutants partially lost their apical dominance at 6 wk after germination. (c) Flower size was partially rescued in F1 plants from the yuc6-D × spl-D cross. (d) Curly-leaf phenotype was alleviated in F1 plants of yuc6-D and spl-D combination. Upper image, adaxial surface; lower image, abaxial surface. (e) Inflorescence structure was alleviated in F1 plants of the yuc6-D and spl-D combination. (f) Expression of YUC6 and SPL was determined in different genotypes by reverse transcription–PCR (RT-PCR).

To further confirm that auxin deficiency in spl-D plants was caused by the repression of YUC2 and YUC6 transcription, we crossed plants heterozygous for spl-D with yuc6-D, which was a dominant T-DNA insertion mutant of YUC6. The yuc6-D plants displayed classical auxin-accumulating phenotypes: epinastic blades with long petioles and enhanced apical dominance (Fig. 7d,e). In the F1 plants (heterozygous for each locus), the rosettes were slightly curled upward along the mid-vein and the curvature was weaker than that of the spl-D single mutant. However, the rosette leaves of the F1 plants also persisted in curling downward at the leaf margin as in the yuc6-D single mutant (Fig. 7d). In addition, the leaves of F1 plants were intermediate in size between those of the single-mutant parents (Fig. 7d). Moreover, the inflorescence architecture of F1 plants was similar to that of the yuc6-D plants, whereas the flowers were more like those of spl-D plants (Fig. 7c,e). The morphology of leaves and flowers of the F1 plants indicated that yuc6-D could partially alleviate spl-D phenotypes. We also found by RT-PCR analysis that both SPL and YUC6 were highly expressed in the F1 plants (Fig. 7f).


SPL expression level is critical for the development of lateral organs

SPL was reported to encode a nuclear protein that played a key role in microsporogenesis and megasporogenesis (Schiefthaler et al., 1999; Yang et al., 1999; Balasubramanian & Schneitz, 2000, 2002). Because the expression level of SPL in the vegetative tissues of wild-type Arabidopsis is too low to be detected by northern blot or RT-PCR, and the SPL null mutant displays no obvious abnormalities except for sterility, the role of SPL in the development of vegetative tissues has long been neglected (Schiefthaler et al., 1999; Yang et al., 1999; Fig. 2c). In this study, we found that elevated expression of SPL expression by the CaMV 35S enhancer tetrad resulted in not only severe curly-leaf phenotypes, but also growth defects in primary roots, leaf expansion and floral size. More strikingly, the SAM of the homozygous spl-D plant was completely lost. These defects in vegetative organs strongly suggest that SPL is an important regulator not only in the maintenance of normal morphology of lateral organs, but also in the transition of SAM from the vegetative to the productive stage.

Increasing evidence suggests that activation tagging is an effective method for identifying important developmental genes whose function can be regulated by their transcription level (Weigel et al., 2000; Dai et al., 2006). We found that phenotypes of the activation tagging mutant spl-D were probably a result of localized elevation of endogenous SPL expression, rather than that of constitutive (i.e. epitopic) expression. Firstly, the 4Enhancer-SPL-2 transgenic plants recapitulated the curly-leaf phenotype observed in spl-D, whereas constitutive over-expression of SPL resulted in variable abnormalities in leaf morphology. Secondly, with the increase of the SPL expression level in the 35S-SPL transgenic plants, the curly-leaf phenotype was more severe and closer to those of spl-D and 4Enhancer-SPL-2, suggesting that the expression level of SPL is critical to the maintenance of flatness of leaf shape. Thirdly, the slight phenotypic difference between spl-D and 35S-SPL-strong plants suggests that the local expression pattern of SPL is important to the implementation of its function. Therefore, the spatial and temporal pattern and strength of endogenous SPL are stringently controlled in vegetative tissues.

The curly-leaf phenotype of spl-D resembles those observed in transgenic plants in which the floral homeotic gene AGAMOUS was constitutively expressed (Mizukami & Ma, 1992, 1997; Favaro et al., 2003). This is consistent with the previous report that SPL is directly activated by AG (Ito et al., 2004). This also indicates that AG possibly affects leaf morphogenesis mainly through SPL activity. The leaf morphology of spl-D plants also resembled the phenotype of the clf-25 mutant, agreeing with a previous report that AG is ectopically expressed in clf-25 rosettes (Goodrich et al., 1997). These observations raised the possibility that CLF suppresses AG expression while AG in turn activates SPL transcription in vegetative tissues. As the carboxyl-terminal region of the SPL protein showed weak transcriptional activity (Supplementary Material, Fig. S5) and SPL has been shown to be directly activated by AG at the transcription level (Ito et al., 2004), the regulation of SPL activity at both the transcription and protein levels suggests that SPL may be an important, and even key, transcription factor regulating leaf morphology in coordination with internal genetic information and external environmental stimuli. Characterization of SPL-interaction proteins and components downstream of SPL in vegetative tissues will add new layers to our understanding of this hierarchical regulatory network for curly-leaf phenotype. The severe curly-leaf phenotype of spl-D can also be used as an effective criterion in screening the suppressor of SPL in the future.

SPL modulates auxin action during lateral organ development

It is not yet clear how SPL directs the differentiation of specific archesporial cells into microsporocytes/megasporocytes (Yang et al., 1999). In the anthers of spl, the primary sporogenous cells do not produce microsporocytes; instead, they become vacuolated to form a cluster of undifferentiated cells, suggesting that cell cycle regulation is disrupted in the spl anther (Yang et al., 1999). Auxin acts as a crucial signal regulating cell division and expansion to coordinate lateral organ morphogenesis and environmental changes (Leyser, 2002). However, whether SPL regulates anther cell division and differentiation through auxin has not yet been determined conclusively. In spl-D, we propose that the SPL function is partially implemented through modulation of auxin activities. Firstly, the growth rate of primary roots as well as the expansion of the fifth leaf was significantly slower in spl-D than in wild type. Secondly, we found that cyclinB1-GUS staining in spl-D plants was weaker and the expression of some cell cycle-related genes downstream of the auxin pathway was decreased (Supplementary Material Fig. S6). These observations suggest that SPL overexpression represses cell division. However, the spl-D mutant displayed auxin-related defects, for example defective hypocotyl elongation at higher temperatures, fewer lateral roots, a simpler vein pattern, reduced DR5-GUS staining, and decreased auxin-inducible gene expression. Moreover, the IAA concentration was significantly decreased in spl-D, whereas auxin signaling and polar auxin transport were not affected. Some auxin-homeostasis genes were also affected in spl floral tissue (H. Ma, pers. comm.). These data from gain-of-function and loss-of-function mutants provide strong support for the theory that SPL modulates auxin homeostasis in plant development.

SPL represses YUC gene expression

Recently, Cheng et al. (2006, 2007) demonstrated that YUCCA flavin monooxygenases, which catalyze a rate-limiting step in auxin biosynthesis, play a vital role in the formation of embryonic and postembryonic organs. Four of the 11 YUCCA genes, YUC1, YUC4, YUC2 and YUC6, were reported to play essential roles in auxin biosynthesis (Cheng et al., 2006). The quadruple mutant yuc1 yuc2 yuc4 yuc6 exhibited a similar curly-leaf phenotype to spl-D, consistent with the evidence that the expression level of YUC2 and YUC6 was significantly decreased in spl-D plants (Fig. 6). The conclusion that YUC2 and YUC6 are repressed by SPL over-expression is supported by the evidence that DR5-GUS staining in spl-D plants was weaker than that in the control plants, suggesting that the production of free IAA is reduced in spl-D. In addition, these auxin-related defects, that is, a defective vein pattern, reduced shoot apical dominance and defective flower morphology, are found in both spl-D and yuc2 yuc6 double mutants. Furthermore, yuc6-D partially rescued the spl-D leaf shape and inflorescence architecture phenotypes. These data suggest that YUC2 and YUC6 probably account for the auxin-related phenotypes in spl-D mutants. That the leaves of yuc2 yuc6 double mutants are flat rather than curly suggests that there are additional YUCCA genes related to auxin biosynthesis or homeostasis contributing to the spl-D phenotypes, as shown in the microarray data. In addition, Electrophoretic Mobility Shift Assays (EMSA) analysis showed that the SPL protein could not directly bind to the promoters of YUC2 and YUC6 in vitro (data not shown), suggesting that SPL may regulate the expression of YUC2 and YUC6 indirectly, or that there are other components involved in this regulation. Characterization of the downstream regulated gene(s) or the protein(s) interacting with SPL will shed light on how the expression of YUC genes is regulated and how the homeostasis of auxin is established. However, because the expression of SPL was not detectable in leaves and spl mutants exhibited no obvious defects in leaf morphology, the possibility that the leaf developmental defects found in spl-D may not entirely reflect the endogenous role of SPL cannot be completely excluded, and further studies are required to exclude this possibility.

In summary, we demonstrate in this study that SPL plays an essential role in the regulation of lateral organ morphogenesis, mainly by affecting auxin homeostasis through repressing the expression of YUC2 and YUC6. Our findings will facilitate further clarification of the role of auxin in lateral organ morphogenesis.


This work was supported by the National Natural Science Foundation of China (GN 30625002 to L-JQ and 30628012 to YZ). We thank Dr Hong-Wei Guo (Peking University, Beijing, China), Professor Kiyotaka Okada (Kyoto University, Kyoto, Japan) and Professor Hirokazu Tsukaya (University of Tokyo, Tokyo, Japan) for helpful comments on the manuscript; Dr Hong Yan (China Agriculture University, Beijing, China) and Mr Xi-Zeng Mao (Peking University, Beijing, China) for microarray data analysis; Miss Fei Gao (University of Missouri) and Miss Su Chen (University of Michigan) for help in genotyping the mutants; and Dr Masatoshi Taniguchi (Kyoto University, Kyoto, Japan) for help with the RNA blot.