Overexpression of the kiwifruit SVP3 gene affects reproductive development and suppresses anthocyanin biosynthesis in petals, but has no effect on vegetative growth, dormancy, or flowering time

Summary Overexpression of SVP3 affects kiwifruit flower and fruit development. The reduced petal pigmentation results from interference with transcription of the kiwifruit flower tissue-specific R2R3 MYB regulator.

Relatively small changes in sequence can have a major effect on STMADS gene function. In Arabidopsis thaliana, the two STMADS genes SVP and AGL24 act in an opposite manner despite their close homology. SVP plays a key role as a repressor of the transition to flowering; a single amino acid substitution causes the loss of function resulting in an early flowering phenotype (Hartmann et al., 2000;Méndez-Vigo et al., 2013). In contrast, AGL24 acts as a promoter of flowering because its inactivation results in late flowering (Yu et al., 2002). Both proteins have further roles during early stages of flower development to repress expression of class B and C floral homeotic genes and maintain floral meristem identity (Gregis et al., 2009). Protein interactions and formation of complexes with other MADS-domain proteins are essential for STMADS function (Liu et al., 2009;Gregis et al., 2013). Interaction with APETALA1 (AP1) in particular resulted in floral reversion phenotypes observed in transgenic Arabidopsis upon ectopic expression of SVP-like genes (Liu et al., 2007;Gregis et al., 2008;Fornara et al., 2008;Wu et al., 2012;Jaudal et al., 2013). For these and perhaps other reasons, the function of STMADS observed upon ectopic expression in Arabidopsis might be different from their function in the species of origin (Trevaskis et al., 2007;Jaudal et al., 2013).
In woody perennial plants, the STMADS subfamily genes have been reported to act as growth inhibitors, which correlated with maintenance of plant dormancy (Mazzitelli et al., 2007;Bielenberg et al., 2008;Diaz-Riquelme et al., 2009;Horvath et al., 2010;Li et al., 2010;Sasaki et al., 2011;Yamane et al., 2011). Deletion of the six SVP-like DORMANCY-ASSOCIATED MADS-BOX (DAM) genes in peach (Prunus persica) resulted in failure to enter dormancy under cold or short-day induction, and the expression of a subset of these genes was elevated during endodormancy (Bielenberg et al., 2008;Li et al., 2009;Yamane et al., 2011). Similarly, a negative correlation of expression with endodormancy release was observed for six tandemly arrayed DAM genes predicted to act as transcriptional repressors in Japanese apricot (Prunus mume) (Yamane et al., 2008;Sasaki et al., 2011). Ectopic expression of one of these genes in transgenic poplar resulted in premature growth cessation and terminal bud set, demonstrating the role in growth inhibition . However, neither of these genes was functionally characterized by ectopic expression in the plant species of origin.
In the kiwifruit species Actinidia chinensis and A. deliciosa, four SVP genes (SVP1-SVP4) have been identified, with potentially distinct roles in bud dormancy and flowering (Wu et al., 2012). Their expression was mostly confined to vegetative tissues, with SVP3 showing the highest relative expression. SVP1, SVP2, and SVP4 were elevated in buds over the winter dormancy period. In contrast, SVP3 accumulation in buds did not demonstrate seasonal changes, but delayed flowering in transgenic Arabidopsis and was able to rescue the Arabidopsis svp41 mutant. Ectopic expression resulted in floral reversion phenotypes, caused by the strong capacity for heterodimerization with Arabidopsis MADS box proteins responsible for flower development (Wu et al., 2012).
In the current study, functional analysis of SVP3 was performed using ectopic transgenic analysis in kiwifruit species A. deliciosa and A. eriantha, which have different chilling requirements and bloom time, and a model plant species tobacco (Nicotiana tabacum). Normal vegetative growth and floral transition, but abnormal flower development, reduced petal pigmentation, and abnormal fruit and seed development were observed. The underlying mechanism of reduced anthocyanin accumulation in petals was examined. Based on the results obtained, the biological function of SVP3 is discussed.

Gene isolation and vector construction
The A. chinensis SVP3 coding sequence was cloned under the control of the Cauliflower mosaic virus (CaMV) 35S promoter (35S:SVP3) into the pART277 binary vector (Gleave, 1992) as previously described (Wu et al., 2012). The resulting plasmid was transformed into Agrobacterium tumefaciens strain EHA105 for transformation into A. deliciosa and A. eriantha, and into GV3101 for transformation into N. tabacum. A construct with a reporter gene uidA (GUS; β-gluronidase) under the control of the CaMV 35S promoter (35S:GUS) in appropriate Agrobacterium strains was used to transform control plants (Wang et al., 2006). Actinidia eriantha SVP3 (GenBank accession no. KJ123703) was amplified from the bud cDNA using gene-specific oligonucleotide primers, 5'-ATGGCGAGAGAGAAGATCAAGA-3' (forward) and 5'-TGGTGTGACATTTCAAGTTCG-3' (reverse). The predicted amino acid sequence alignment with A. deliciosa and A. chinensis SVP3 homologues was performed using Vector NTI version 9.0. (Invitrogen) Clustal W (opening 15, extension penalty 0.3).

Plant transformation and growth conditions
The transformation procedures for A. deliciosa and A. eriantha were as previously described (Wang et al., 2006(Wang et al., , 2007. Once roots were established, transgenic plants were transferred to soil and grown in the containment glasshouse at the Plant & Food Research, Mt Albert, Auckland, New Zealand for 18 months. Flowering time and floral phenotype were assessed in the following spring season. Nicotiana tabacum 'Maryland Mammoth' and 'Samsun' transformations were carried out on young leaf discs excised from in vitro grown shoots (Horsch et al., 1985). Transgenic tobacco plants were grown in a glasshouse at 20 °C under long-day (16/8 h light/dark; for 'Samsun') or short-day conditions (8/16 h light/dark; for 'Maryland Mammoth'). The seeds from these transgenic plants were collected and germinated on half-strength Murashige and Skoog (MS) agar medium (Murashige and Skoog, 1962) supplemented with 50 μg ml -1 kanamycin. Following the segregation tests, two homozygous lines of both cultivars at T 2 generations were used for flowering time and floral phenotype assessment.

RNA extraction and expression studies
Actinidia eriantha root, stem, leaf, axillary bud, shoot apex, flower, and fruit tissue collection was carried out on vines growing in the glasshouse at the Plant & Food Research, Mt Albert, Auckland, New Zealand, during the spring and summer season of 2011-2012. Actinidia chinensis root, stem, leaf, flower, fruit, and axillary and apical bud collection was carried out on vines growing at the Plant & Food Research orchard near Kerikeri, New Zealand, during the spring and summer season of 2005-2006. Total RNA was extracted from kiwifruit tissue as previously described (Chang et al., 1993). Total RNA was isolated from tobacco flowers using the Trizol reagent (Invitrogen). A 5 μg aliquot of total RNA was treated with DNase I (Ambion) and reversetranscribed at 37 °C using the BluePrint ® Reagent kit for reverse transcription-PCR (RT-PCR; TaKaRa) according to the manufacturer's instructions. Amplification and quantification were carried out using the LightCycler ® 480 System and LightCycler ® 480 SYBR Green I Master Mix (Roche Diagnostics). A minimum of three technical replicate reactions were performed and a non-template control was included in each run. Thermal cycling conditions were 95 °C for 5 min, followed by 50 cycles of 95 °C for 10 s, 60 °C for 10 s, and 72 °C for 20 s, and a melting temperature cycle, with constant fluorescence data acquisition from 65 °C to 95 °C. The data were analysed using the Target/Reference ratio calculated with the LightCycler ® 480 software 1.5 (Roche Diagnostics). The expression was normalized to previously characterized reference genes, kiwifruit actin (Wu et al., 2012) and tobacco Ntα-Tub1 (Pattanaik et al., 2010).

Biochemical analyses
For chlorophyll quantification, petals collected at visually similar developmental stages were frozen and ground in liquid nitrogen. Ground tissues were suspended in 80% acetone with 2.5 M phosphate buffer pH 7.5 and incubated for 1 h in the dark, followed by centrifugation at 16 000 g for 10 min. The absorbance of 200 μl of the supernatant was measured spectrophotometrically at 663 nm and 646 nm (SpectraMax 384, Molecular Devices, Sunnyvale, CA, USA). The total chlorophyll (C a+b ) concentration was calculated using the equation C a+b (μg ml -1 ) =7.15A 663 +18.71A 646 (Lichtenthaler and Rinderle, 1988).
Anthocyanins and flavonoids were extracted from lyophilized samples and analysed by high-performance liquid chromatography (HPLC) and liquid chromatography-mass spectrometry (LC-MS) as described previously (McGhie et al., 2011;Dare et al., 2013). Quantification of cyanidin 3-O-galactoside was achieved by calibration at 520 nm by HPLC analysis. The peaks were identified by a comparison of retention times with authentic standards of cyanidin 3-O-galactoside (Polyphenols Laboratories, Sandnes, Norway).

Microscopic analyses
Tobacco petal epidermis was obtained by careful peeling and observed under a light microscope (Olympus Vanox AHT3, Olympus Optical Co. Ltd, Tokyo, Japan). Photographs were taken using a CoolSNAP Colour camera system (Roper Scientific Ltd, Tucson, AZ, USA).

Results
Overexpression of SVP3 has no significant effect on growth or dormancy in A. deliciosa To investigate the role of kiwifruit SVP3, transgenic A. deliciosa was generated using the SVP3 cDNA driven by the CaMV 35S promoter (Wu et al., 2012). Ten independent transgenic lines with high SVP3 transgene expression levels were generated and compared with six control lines (Fig. 1A). Monitoring during callus formation, plantlet growth in tissue culture, and upon transfer to soil revealed no obvious difference between 35S:SVP3 lines and control lines. Eight 35S:SVP3 and three control lines were further monitored for spring bud break. The number of breaking buds or developing shoots was recorded weekly and is presented as a percentage of total buds. Large variation in the time of the first visible bud break, positions of breaking buds, number of developing branches, and rate of shoot outgrowth was observed for all lines; however, the first visible bud break time of the 35S:SVP3 lines was in the range recorded for control lines and was followed by normal growth of multiple branches (Fig. 1B, C). Therefore, it was concluded that SVP3 does not have a significant effect on growth and dormancy in A. deliciosa under glasshouse conditions.

SVP3 affects reproductive development in transgenic A. eriantha
Actinidia deliciosa has a high chilling requirement (Wall et al., 2008) and late bud break, and fails to flower under glasshouse conditions. Therefore, to evaluate further the role of SVP3, the non-cultivated kiwifruit species A. eriantha was chosen for transformation, as it provided advantages including low chilling requirement and early bud break, fast maturity, and prolific flowering in glasshouse conditions (Wang et al., 2006). Actinidia eriantha SVP3 shares almost identical sequence and expression pattern with the previously studied cultivated kiwifruit species A. deliciosa and A. chinensis ( Supplementary Fig.  S1A, B available at JXB online; Wu et al., 2012). Six transgenic lines with moderate to high levels of SVP3 transgene expression and three control lines were generated ( Fig. 2A). None of the lines showed any obvious difference during callus formation, growth in tissue culture, and establishment in the soil; similar growth rate and architecture were observed ( Supplementary Fig. S1 available at JXB online), and in all lines the time of bud break and flowering time (appearance of visible floral buds) were comparable ( Table 1). The number of flowers was highly variable and not correlated with transgene expression levels, but the duration of flower development and appearance of flowers were altered significantly in transgenic lines, particularly in the highest expressing Line 1 (Table 1).
For further detailed analysis, Line 1 and Line 2 were chosen as representatives of high and intermediate SVP3 transgene expression, respectively. Flower buds in Line 1 and to a lesser extent in Line 2 were not fully enclosed with sepals (Fig. 2B); the number and order of developing floral organs appeared normal (Fig. 2C), but flowers in Line 1 remained smaller (Fig. 2D). Fruit in Line 1 was misshapen and smaller compared with control or Line 2 fruit (Fig. 2E, F) and the transgenic seed was smaller, lighter ( Fig. 2G-I), and failed to germinate.
Floral reversion phenotypes observed upon expression of STMADS genes in Arabidopsis result from interaction with Arabidopsis floral homeotic MADS box proteins, particularly AP1 (Fornara et al., 2008;Gregis et al., 2009;Wu et al., 2012). To test if phenotypes observed upon ectopic expression of SVP3 in A. eriantha arose from similar MADS box protein interactions, yeast two-hybrid analysis was performed using previously characterized kiwifruit floral MADS box genes (Varkonyi-Gasic et al., 2011). SVP3 interacted with kiwifruit AG and SEP proteins SEP1, SEP3, and SEP4 with similar strong interaction intensities; a weaker interaction was identified with FUL and FUL-like, but no interaction was detected with AP3, PI, and the negative control CEN (Fig. 2J).

SVP3 affects petal colour in transgenic A. eriantha by transcriptional repression of key genes in the anthocyanin pathway
In addition to the described phenotypes, the colour of the petals of 35S:SVP3 lines was dramatically reduced (Fig. 2D, Table 1) and the chlorophyll content was higher (Fig. 3A).
The presence and intensity of the peak corresponding to the major kiwifruit anthocyanin, cyanidin 3-O-galactoside, correlated with the visual petal phenotype in flower buds and open  (Fig. 3B, C), and further quantification confirmed differential accumulation of cyanidin 3-O-galactoside (Fig. 3D) and several minor flavonoid compounds ( Supplementary Fig.  S2 available at JXB online).
Analysis of the relative expression of the genes encoding enzymes of the anthocyanin biosynthetic pathway (Montefiori et al., 2011) revealed similar expression profiles in representative 35S:SVP3 and control lines, with some variation between flower bud and open flower samples. The exception was the transcript of F3GT1, which was absent or barely detectable in flower buds of 35S:SVP3 lines, in contrast to relatively high accumulation observed in the control flower bud (Fig. 4A). Analysis of regulatory genes of the anthocyanin biosynthetic pathway (Fraser et al., 2013) revealed high relative expression of MYB110a and low relative expression of MYB10 in flower buds of the control line. In 35S:SVP3 lines these transcripts were absent or barely detectable, and in the control line both transcripts showed a decrease in open flowers (Fig. 4B).

SVP3 does not alter flowering time but affects reproductive development and petal colour in transgenic tobacco
To investigate further whether kiwifruit SVP3 affects reproductive development and anthocyanin accumulation in a conserved manner, transgenic tobacco lines were generated using the 35S:SVP3 construct. A day-neutral tobacco 'Samsun' and a short-day variety 'Maryland Mammoth' were used. This approach provided the opportunity to establish if SVP3 affects flowering time in an annual representative of asterids, the clade of eudicots to which the perennial kiwifruit belongs. Eight independent transgenic lines of tobacco 'Samsun' and three of 'Maryland Mammoth' were generated. Two homozygous T 2 lines of each cultivar with high transgene expression and two corresponding wild-type control lines were chosen for detailed analysis. No significant differences were observed in vegetative growth and flowering time, but the duration of flower development was extended, flower fertility was diminished, and seed germination was delayed (Table 2). Spikeshaped sepals in transgenic plants failed to enclose flower buds fully, the corolla remained green, and the distal part of the petal was altered in both colour and shape (Fig. 5A). Additional phenotypes were identified, including occasional homeotic conversion of stamen to petal and change in the ovary shape and its position relative to attachment of other floral parts (Fig. 5B). Changes in total chlorophyll, anthocyanin, and flavonoid content and presence of stomata in green petal segments were also noted ( Supplementary Fig.  S3 available at JXB online). To establish if reduced pigment accumulation resulted from transcriptional repression of key regulators of anthocyanin biosynthesis, quantitative PCR (qPCR) analysis of basic helix-loop-helix (bHLH) genes, NtAn1a and NtAn1b, and the R2R3 MYB gene, NtAn2 (Pattanaik et al., 2010;Bai et al., 2011), was performed. NtAn2 was significantly and consistently down-regulated in transgenic lines, while NtAn1a and NtAn1b expression showed only minor variations (Fig. 5C).

Overexpression of SVP3 affects reproductive development but not floral transition in transgenic kiwifruit and tobacco
Ectopic expression of the STMADS gene subfamily often results in altered flowering time, severe floral abnormalities, delayed flower anthesis, and senescence (Hartmann et al., 2000;Yu et al., 2002;Lee et al., 2007;Fornara et al., 2008;He et al., 2010;Li et al., 2010), and affects duration of endodormancy in woody perennial species . Overexpression of SVP3 in A. deliciosa showed no major effect on vegetative growth and the timing of first visible bud break in glasshouse conditions, although insufficient winter chilling and large variation in the position and number of synchronously breaking buds might have affected the rate of visible shoot outgrowth. Overexpression of SVP3 in A. eriantha and tobacco showed no significant effect on plant growth, bud break, and flowering time, supporting the findings in A. deliciosa. However, flower, fruit, and seed development were severely affected. The size and shape of sepals were abnormal and they failed to enclose flower buds fully, consistent with the phenotype observed in Arabidopsis (Wu et al., 2012) and similar to other reports (Masiero et al., 2004;Liu et al., 2007;Trevaskis et al., 2007;Fornara et al., 2008). The petals contained vegetative features including increased chlorophyll content and stomata, similar to phenotypes observed upon overexpression of potato StMADS16 and sweet potato IbMADS3-1 in tobacco (Garcia-Maroto et al., 2000;Shin et al., 2011). Flower development was slower, resulting in significantly delayed anthesis and an increased proportion of sterile flowers, which failed to produce fruit (Fornara et al., 2008;Li et al., 2010). Additional phenotypes included underdeveloped and misshapen fruit, abnormal seed development, reduced seed germination rate, and extended seed germination, suggesting that SVP3 affected all stages of reproductive development. The severity of the observed phenotypes during reproductive development, combined with normal vegetative growth, wild-type bud break and flowering time, and relatively high levels of expression in vegetative tissues (Wu et al., 2012), indicate that SVP3 has a role in the regulatory network of vegetative organogenesis and early stages of floral transition as the key repressor of floral organ development. It is hypothesized that SVP3 acts to retain vegetative features during plant growth and floral transition, and has to be removed to enable subsequent stages of reproductive development. Ectopic expression phenotypes, combined with constitutive vegetative expression, suggest that there is no involvement of SVP3 in maintenance of winter dormancy and general growth inhibition. However, SVP3 may have a role in these processes as an interacting partner of other, ratelimiting proteins, and a loss-of-function phenotype would be required to understand SVP3 function fully.

SVP3 interferes with anthocyanin biosynthesis in petals via repression of an anthocyanin-related MYB transcription factor
Overexpression of MADS-box genes is often associated with floral abnormalities, including changes in petal pigmentation. Light green sepaloid petals were observed upon overexpression of Arabidopsis AGAMOUS-LIKE 20 (AGL20 or SOC-1) in Arabidopsis (Liu et al., 2007) and of the mustard AGL20 orthologue MADSA in short-day tobacco 'Maryland Mammoth' (Borner et al., 2008). Similarly, expression of STMADS genes such as potato StMADS16 and sweet potato IbMADS3-1 in tobacco resulted in development of chlorophyll-enriched petals (Garcia-Maroto et al., 2000;Shin et al., 2011). However, these reports did not investigate anthocyanin biosynthesis and accumulation in the petals of transgenic plants, and very little is known about potential MADS-box protein interference with the regulatory genes of the anthocyanin pathway. This study demonstrated that SVP3 reduced anthocyanin accumulation in petals of transgenic A. eriantha in a quantitative manner. The underlying mechanism was dramatically reduced expression of the key structural gene F3GT1 (Montefiori et al., 2011) and transcriptional repression of regulatory MYB transcription factors, in particular MYB110a, the key gene required for kiwifruit petal pigmentation (Fraser et al., 2013). A similar mechanism was observed in transgenic tobacco, where SVP3 interfered with transcription of the regulatory MYB NtAn2. At this stage, it is unclear if SVP3 acts by direct or indirect repression of the regulatory MYB transcription. Currently, there is little knowledge about upstream regulators of anthocyanin-related R2R3 MYB expression. Some MADS box genes from the AP1/SQUA class have been implicated in regulation of anthocyanin accumulation. IbMADS10 expression correlated with red pigmentation in sweet potato, and ectopic expression resulted in anthocyanin accumulation in transgenic sweet potato callus and transgenic Arabidopsis (Lalusin et al., 2006(Lalusin et al., , 2011. VmTDR4 expression was linked with colour development and anthocyanin-related gene expression in bilberry, while silencing of this gene reduced anthocyanin levels and altered expression of the regulatory R2R3 MYB transcription factor (Jaakola et al., 2010). STMADS proteins have a capacity to interact with other MADS box proteins, and interaction with AP1 affected flowering time and flower development in transgenic Arabidopsis (Fornara et al., 2008;Gregis et al., 2008;Wu et al., 2012). Therefore, a similar mechanism might be responsible for repression of MYB110a and NtAn2 transcription and reduced anthocyanin accumulation in transgenic kiwifruit and tobacco flowers. The yeast two-hybrid assays showed that SVP3 could interact with kiwifruit AP1/SQUA proteins FUL and FUL-like, and this interaction may interfere with activation of R2R3 MYB transcription. Even stronger interactions were observed with AG and SEP MADS box proteins required for normal flower and  fruit development (Varkonyi-Gasic et al., 2011). Therefore, interference with establishment of homeotic MADS box protein complexes, as previously shown in Arabidopsis (Fornara et al., 2008;Gregis et al., 2008;Wu et al., 2012), is the likely mechanism by which SVP3 interferes with various aspects of normal reproductive development. Further functional studies using transgenic kiwifruit and tobacco generated in the course of this study will aid in establishing direct targets and protein partners of SVP3, providing understanding of the pathways and plant processes regulated by SVP3. These plants also provide an exciting opportunity for deeper understanding of the regulation of anthocyaninrelated MYB expression in flowers.

Supplementary data
Supplementary data are available at JXB online. Figure S1. SVP3 in Actinidia eriantha. Figure S2. LC-MS analysis of flavonoid compounds in petals of 35:SVP3 and control A. eriantha. Figure S3. Analysis of transgenic 35S:SVP3 tobacco.