Treatments with different auxins were conducted with Vitis vinifera L. cv. Shiraz fruit (Adelaide Hills, South Australia, 35.018223 S, 138.838220 E). Berries were sprayed twice (29 December 2009, 13 January 2010) during the preveraison period with 80 mg l⁻¹ IAA (Sigma-Aldrich, St Louis, USA), 50 mg l⁻¹ NAA (Gibco BRL Life Technologies, Grand Island, USA) or 20 mg l⁻¹ BTOA (American Cyanamid Company, Princeton, USA) in 0.1% (v/v) Chemwet 1000 (Nufarm, Laverton, Australia). Control fruit were sprayed with a 0.1% (v/v) Chemwet 1000 solution. The trial was of a randomized triplicate design, the sample size per replicate and treatment was 30 bunches. Samples of 60 randomly harvested berries per replicate were taken 1, 2, and 7 days post the initial spray (dpis), followed by weekly sampling throughout development. Sampling was completed between 09.30 h and 14.30 h, berries were weighed, immediately deseeded, frozen in liquid nitrogen, and stored at –80 °C until used. Anthocyanins and total soluble solids (TSS) were measured for each of these replicates.

For the analysis of developmental changes in gene expression Vitis vinifera L. cv. Cabernet Sauvignon flowers (at the 50% cap fall stage: anthesis (0 wpf)) and berries from 2–16 weeks post-flowering (wpf) were collected at fortnightly intervals, (23 November 2004–16 March 2005) from a commercial vineyard (Clare Valley, South Australia, 33.794364 S, 138.628353 E) in the 2004/2005 season. Sampling was completed between 09.30 h and 14.30 h, berries (100–150 berries sampled at each time point) were immediately deseeded and the tissue was frozen in liquid nitrogen and stored at –80 °C until used.

For the ex planta berry induction experiment grape berries (Vitis vinifera L. cv. Shiraz) were sampled from a vineyard in the Adelaide Hills, South Australia (35.018223 E, 138.838220 S) at either 22 d (12 January 2009) or 12 d (22 January 2009) before veraison between 09.00–10.00 h and kept on ice until used.

Frozen berries were ground to a powder using an IKA A11 basic analytical mill (IKA, Staufen, Germany). For the measurement of TSS (degrees Brix) 100 mg of berry powder was thawed on ice, the tissue was pelleted by centrifugation at 18 000 g for 5 min and the supernatant was analysed with an RFM710 digital refractometer (Bellingham Stanley, Tunbridge Wells, UK). For anthocyanin determination, 300 mg of powdered sample was added to 1.5 ml of MeOH containing 1% (v/v) HCl. Anthocyanins were extracted at room temperature in the dark on a rotating mixer for 1 h. The tissue was pelleted by centrifugation at 18 000 g for 15 min and the supernatant retained. Depending on the developmental stage the supernatant was diluted up to 20-fold with MeOH, 1% (v/v) HCl. Total anthocyanins were measured spectrophotometrically by reading absorbance at 520 nm immediately following centrifugation.

GH3-related plant sequences were identified by BLASTP searches of the non-redundant protein database (P value ≤10⁻²⁰) using the Vitis vinifera L. GH3-1 protein sequence as the query. Non-redundant sequences that, based on comparison with the grapevine GH3-1 protein, were considered full-length were included in the analysis. NCBI or GenBank accession numbers for all sequences used in the phylogenetic analysis are listed in Supplementary Table S1 at JXB online. The sequences were aligned using the ClustalX (version 2.0.10) program (^{Larkin et al., 2007}) in the multiple alignment mode and the Neighbor–Joining unrooted tree was generated with PHYLIP 3.6 (^{Felsenstein, 1989}). The Invitrogen Vector NTI AlignX software (version Advanced 10) was used to determine sequence similarities and identities.

Total RNA was extracted from grape berry tissues according to ^{Davies and Robinson (1996)} and further purified as described by ^{Symons et al. (2006)}. First-strand cDNA for quantitative real-time PCR was synthesized with Superscript III enzyme (Invitrogen, Carlsbad, USA) using 1 μg of RNA and the Oligo (dT)₂₀ primer in a reaction volume of 20 μl following the manufacturer's instructions. Quantitative RT-PCR was conducted as described by ^{Symons et al. (2006)} using the following gene-specific primer pairs: GH3-1 (XM_002271216), 5′-ATCTACGAGCGCAAACAAGTCC-3′ and 5′-GTGTGAGTTGGTGCCAGTTGAG-3′; for GH3-2 (XM_002283850), 5′-CTGAGTTGTGGAACCCAGTGAC-3′ and 5′-GCGGATGTAGAAGTTGGGAAAG-3′; for GH3-3 (XM_002283193), 5′-ACATTCTTCCGGCATACGAAGT-3′ and 5′-TGAGCTGGTCGTCACCACTTAT-3′; for GH3-4 (XM_002263317), 5′-TGATGCTCCTATGCAGATCCTC-3′ and 5′-CAAGTGACTGGATTCCACAACC-3′; for GH3-5 (XM_002276205), 5′-GGAACATTTGACAAGCTCATGG-3′ and 5′-CCCTTGAATTCAGAAGCTCGAT-3′; for GH3-6 (XM_002268242), 5′-TAGGATAGTGAAGCCTGGCACA-3′ and 5′-CTCTTTGGATTTGATGCACCTG-3′ and, for normalization of cDNA levels, Actin2 (AM465189), 5′-GCACCCTTCGCACGATATGA-3′ and 5′-TGACGCAAGGCAAGGACTGA-3′. Each PCR was performed in triplicate. To calculate the copy number of the GH3 genes in each reaction, the purified gene fragments used for the standard curves were quantified using PicoGreen (AGRF, Adelaide, South Australia) and the number of molecules in each standard dilution was determined according to ^{Whelan et al. (2003)}. The specificity of the reactions was confirmed by melt curve analysis as well as separation on agarose gels and the identity of each product was verified by sequencing (AGRF, Adelaide, South Australia).

Berries were sampled from 40 bunches of 10 vines at two time points prior to veraison (22 d and 12 d), sterilized in 0.05% (v/v) Tween 20 containing one Milton antibacterial tablet l⁻¹ (Milton Australia, Laverton North, Victoria) for one hour and washed three times with sterile nanopure water. All the following procedures were carried out in a laminar flow under sterile conditions. A thin slice was cut off horizontally around the brush area of each berry to facilitate compound absorption and 20 berries were placed on Petri dishes filled with 25 ml of Gamborg's media, 0.025% (w/v) casein hydrosylate, 0.8% (w/v) agar, pH 5.7–5.8 and one or more of the following additives (final concentrations), respectively: IAA (0.5 μM), NAA (0.5 μM), BTOA (0.5 μM), 12% (w/v) sucrose.

Berries were placed on the plates with the brush facing the agar, the plates were sealed with Parafilm and kept in the dark at room temperature. After 24 h the berries were harvested, deseeded, and frozen in liquid nitrogen.

[Indole-D₅]IAA and [1,4-¹³C₂]-L-aspartic acid were from Cambridge Isotope Laboratories (Andover, USA), L-tryptophan methyl ester hydrochloride and L-aspartic acid dimethyl ester hydrochloride were purchased from Sigma-Aldrich (St Louis, USA).

IAA–amino acid conjugates were synthesized according to ^{Ilić et al. (1997)} using either unlabelled IAA or [indole-D₅]-labelled IAA as the starting substrate. The same protocol was used for the synthesis of NAA–Asp and BTOA–Asp, but since no labelled forms of NAA and BTOA were commercially available [1,4-¹³C₂]-L-aspartic acid was used to obtain labelled conjugates of these compounds. Conjugate-containing fractions were further purified by reversed-phase HPLC using an Agilent 1100 series system equipped with a Luna C18 column [250×10 mm, 5 μm (Phenomenex, Torrance, CA)], using isocratic elution [40% (v/v) MeOH, 0.2% (v/v) acetic acid (IAA–Asp, Rt=7.8 min; BTOA–Asp, Rt=13.2 min) or 60% (v/v) MeOH, 0.2% (v/v) acetic acid (IAA–Trp, Rt=8.4 min; NAA–Asp, Rt=6.2 min)] at a flow rate of 4.5 ml min⁻¹. The column eluent was monitored using a DAD detector observing at a single wavelength (280 nm).

The coding region of GH3-2 was amplified by PCR from a Cabernet Sauvignon berry cDNA template using gene-specific primers (5'-TATCATATGGCAGTTGATTCCGGTCTGTC-3', 5'-ATAGCGGCCGCGTCCGTACGTCGTCGCTCTG-3') with additional NdeI and NotI sites (in bold). Cloning, heterologous expression, and purification of GH3-2-His (C-terminal fusion) as well as TLC-based assays for IAA–amino acid conjugate formation were performed as described by ^{Böttcher et al. (2010}a) for GH3-1.

To identify suitable protein amounts and incubation times to be used for the determination of steady-state kinetic parameters for GH3-1 and GH3-2 the following standard assay conditions were used: 50 mM TRIS/HCl (pH 8.6), 3 mM MgCl₂, 3 mM ATP, 1 mM DTT, 1 mM aspartic acid (Asp) or tryptophan (Trp), 1 mM IAA, 25 °C, 50 μl volume. The reactions were initiated by adding 0–1 μg of purified GH3-1 or GH3-2 from glycerol stocks [10% (v/v) glycerol, 1 mM DTT, 5–10 mg ml⁻¹ protein, stored at –80 °C for up to one month] and stopped after 0–20 min by adding 10 μl 50% (v/v) HCl. After the addition of 250–750 pmol of labelled IAA–-Asp or IAA–Trp as internal standards the samples were extracted twice with 150 μl ethyl acetate. The extract was dried and the residue resuspended in 30 μl 60% (v/v) MeOH, 1% (v/v) acetic acid to be analysed by LC-MS. Steady-state parameters for both enzymes were determined using the standard assay but varying the concentration of one substrate at a time: (0–20 mM) Asp, (0–20 mM) Trp, (0–5 mM) ATP, (0–2 mM) auxins (IAA, NAA or BTOA). Reactions were stopped after 0, 2, 5, and 10 min, appropriate labelled standards were added and the samples were acidified to pH 1–2. Samples were extracted as described above and subjected to LC-MS analysis (below). The reactions products were quantified, initial velocities were calculated from the linear range and fit to the Michaelis–Menten equation using non-linear regression (SigmaPlot 11.0).

For the quantification of IAA and IAA–Asp in grapes, auxins were extracted from 100 mg of grape berry tissue, spiked with 500 pmol of [indole-D₅]IAA and [indole-D₅]IAA–Asp as internal standards, as described by ^{Kowalczyk and Sandberg (2001)}. After extraction and diethyl ether partitioning the aqueous phase was acidified to pH 1–2 and applied to a 50 mg Env⁺ SPE column (Isolute, Uppsala, Sweden). The column was washed with water (1 ml) and then eluted with 80% (v/v) MeOH, 1% (v/v) acetic acid (2.5 ml). The dried residue was resuspended in 40 μl 60% (v/v) MeOH, 1% (v/v) acetic acid to be analysed with an Agilent LC-MS system (1200 series HPLC coupled with a 6410 triple quad mass spectrometer). The sample (10 μl) was first separated on a Luna C18 column [75×4.6 mm, 5 μm, (Phenomenex, Torrance, CA)] held at 30 °C using the following solvent conditions: 0–8 min isocratic 60% (v/v) MeOH, linear gradient from 60% (v/v) to 95% (v/v) MeOH in 1 min, held for 5 min, from 95% (v/v) to 60% (v/v) in 1 min, held for 6 min, 0.4 ml min⁻¹. The effluent was introduced into the ESI ion source (nebulizer pressure 35 psi) with a dissolvation gas temperature of 300 °C at a flow of 8 l min⁻¹, with the capillary voltage set to 4 kV. The detection was performed by multiple reaction monitoring (MRM) in positive ion mode. The optimization of fragmentation was done with purified IAA–Asp and IAA as well as the labelled standards using Agilent MassHunter Optimizer software. With the collision energy ranging between 14–46 eV, quinolinium ions were the major fragments for all analysed compounds (m/z 130 for unlabelled IAA and IAA–Asp, m/z 134 for labelled IAA and IAA–Asp) and were used for quantitation.

For the analysis of conjugates produced by the in vitro reactions with recombinant GH3-1 and GH3-2, 10 μl of the dissolved residues were introduced into the ESI ion source as described above. After optimization of the fragmentation conditions, the fragments used for quantitation were as follows: for labelled and unlabelled IAA and IAA–Asp as described above, m/z 130 for unlabelled IAA–Trp, m/z 134 for labelled IAA–Trp, m/z 141 for NAA, labelled and unlabelled NAA–Asp, m/z 161 for BTOA, labelled and unlabelled BTOA–Asp. For the analysis of BTOA and BTOA–Asp the chromatography conditions were changed to: 0–11 min isocratic 60% (v/v) MeOH, linear gradient from 60% (v/v) to 95% (v/v) MeOH in 1 min, held for 5 min, from 95% (v/v) to 60% (v/v) in 1 min, held for 6 min, 0.4 ml min⁻¹.

All statistical analyses were performed using SPSS 15.0 (SPSS, Chicago, Illinois, USA).

The ripening-delaying effects of a variety of auxins have been described for many different fruit, but the molecular mechanisms causing this delay remain elusive.

In this study preveraison Shiraz berries were treated twice with either the natural auxin IAA, the synthetic auxin NAA, or the synthetic auxin-like compound BTOA, all of which have previously been reported to delay certain aspects of the ripening of grape berries (^{Weaver, 1962}; ^{Hale, 1968}; ^{Coombe and Hale, 1973}; ^{Davies et al., 1997}; ^{Jeong et al., 2004}; ^{Deytieux-Belleau et al., 2007}; ^{Böttcher et al., 2010}b). Following the treatments the progress of berry development was monitored at regular intervals through to harvest (Fig. 1A–C). Veraison of the Control fruit was found to be 27 dpis, when Brix (Fig. 1A) and anthocyanin (Fig. 1B) levels as well as berry weight (Fig. 1C) increased rapidly. IAA-treated berries followed a very similar trend with no significant difference in berry weight and only minor differences in Brix (27 dpis) and anthocyanin (17 and 27 dpis) levels when compared with the Control (Table 1). The NAA and BTOA treatments led to a delayed accumulation of TSS from 27 dpis through to the harvest of Control fruit at 62 dpis (23.1° Brix) with BTOA-treated fruit having significantly lower Brix levels than NAA-treated berries throughout this period (Fig. 1A; Table 1). Berries sprayed with NAA reached a similar Brix level (23.8°) to the Control 13 d after harvest of the Control fruit, BTOA-treated fruit were collected another 6 d later and by that time had a slightly higher Brix level of 24.6°.

Anthocyanin accumulation was delayed in both NAA- and BTOA-treated fruit (Fig. 1B) with both treatments leading to anthocyanin levels significantly lower than those of the Control fruit between 17–49 dpis (Table 1). This was mainly due to a particularly slow rate of accumulation between 27–36 dpis, when a rapid increase in anthocyanin levels was observed in Control and IAA-treated berries. Compared with all other treatments BTOA-treated berries had the lowest anthocyanin levels at 36 and 49 dpis. No sample of NAA- and BTOA-treated berries was taken at harvest of the Control fruit (62 dpis), but when harvested 13 d (NAA) and 19 d (BTOA) later the anthocyanin concentration measured in auxin-treated berries was 1.2-fold (NAA) and 1.4-fold (BTOA) higher than that of Control berries. Since sugar and anthocyanin accumulation seem to be linked in developing grapes (^{Böttcher et al., 2010}b) this moderate increase in anthocyanin levels might be related to the slightly higher Brix levels of the auxin-treated fruit at harvest when compared with the Control berries (Fig. 1A).

Interestingly, changes in berry weight (Fig. 1C) throughout development qualitatively differed between fruit sprayed with NAA or BTOA. Although statistically different from the Control at only one time point (36 dpis; Table 1), the growth of NAA-treated Shiraz berries followed a similar trend to that previously described for NAA-treated Shiraz berries (^{Böttcher et al., 2010}b). The rate of berry growth was delayed during the first 30–40 dpis, but then accelerated beyond the rate of Control fruit growth, leading to bigger berries at harvest (Fig. 1C) probably due to an increase in cell size (^{Böttcher et al., 2010}b). BTOA-treated berries were smaller than berries from any other treatment from 36 dpis through to harvest (Fig. 1C; Table 1), which is in accordance with previous studies examining the effect of BTOA on Shiraz berry development (^{Hale, 1968}; ^{Davies et al., 1997}). Although the smaller size of the BTOA-treated fruit was, in part, due to a longer delay in berry expansion post-veraison, it was also a result of a lower rate of berry size increase (Fig. 1C).

From the presented data it can be seen that the effect on grape berry ripening was different for each of the three auxinic compounds used in the experiment. Treatments with IAA had no effect on growth or colour accumulation of developing berries and in disagreement with previously published data (^{Deytieux-Belleau et al., 2007}) there was also no delay in sugar accumulation. On the other hand both NAA and BTOA delayed sugar and anthocyanin accumulation with BTOA being more effective, but were clearly distinguished in their effect on berry growth. It has been reported previously that the efficiency in the delay in fruit ripening depends on the compound used, for example, 2,4-dichlorophenoxyacetic acid (2,4-D) was a better inhibitor of the ripening of banana slices than IAA (^{Vendrell, 1969}) and BTOA caused a greater delay in the ripening of Doradillo grapes than the auxin efflux inhibitor 2,3,5-triiodobenzoic acid (TIBA) (^{Coombe and Hale, 1973}). As for the opposite effects of NAA and BTOA on berry growth it has been reported in several previous studies that, depending on the physiological response assayed, BTOA does not always act as would be expected from a classical auxin. When ^{Crane (1952)} treated parthenocarpic fig (Ficus carica L.) achenes with BTOA they developed a completely sclerified endocarp, indistinguishable from that of pollinated fruit. This effect was not found with any other auxin tested. In another example ^{Steward and Shantz (1956)} applied BTOA to carrot root explants and artichoke tuber explants which induced growth by cell division and not, as caused by IAA treatment, by cell enlargement. It is possible that the differential effect of BTOA on berry expansion may be due to it having very strong auxin-like properties in grape but it is also possible that BTOA is acting in a pleiotropic manner.

There are three main families of early auxin response genes (^{Abel and Theologis, 1996}), which are the Small Auxin Up RNA (SAUR) genes, the Auxin/IAA-inducible (AUX/IAA) genes, and the Gretchen Hagen (GH3) genes (^{Chapman and Estelle, 2009}) and the expression of these genes can be used to elucidate the molecular mechanisms involved in the different responses to exogenous auxins. In the scope of this study the IAA-amido synthetase encoding GH3 genes were of particular interest due to the known selectivity of these enzymes for their auxin substrates (^{Staswick et al., 2005}) and the implication of one grapevine GH3 protein (GH3-1) in the ripening process of berries (^{Böttcher et al., 2010}a). Interestingly, the expression of GH3-1 was induced in preveraison berries treated with IAA, NAA, and BTOA (^{Böttcher et al., 2010}a). This raises the question if the differences in ripening delay caused by exogenous application of IAA, NAA, and BTOA (Fig. 1) might be related to the activity of GH3 proteins.

In order to identify all putative IAA-conjugating grapevine GH3 proteins that are responsive to auxins, a BLASTP similarity search on the NCBI database was performed using the grapevine GH3-1 protein as the query. From this search, eight additional non-redundant grapevine GH3-like proteins, consisting of 578–613 amino acids, were predicted and these were designated GH3-2 to GH3-9 (Table 2).

To examine the relationship between the grapevine GH3 proteins and those from other plant species, an unrooted phylogenetic tree was constructed from the alignment of 67 full-length GH3 protein sequences (Fig. 2). GH3 proteins can be divided into three groups (I–III) according to their sequence similarity and their function (^{Staswick et al., 2002}, ²⁰⁰⁵). Group I contains Arabidopsis GH3-11 (JAR1) which specifically adenylates jasmonic acid (^{Staswick et al., 2002}). Some members of the numerous group II proteins (Fig. 2), including Vv GH3-1, have been shown to be IAA-amido synthetases conjugating mainly IAA to a variety of amino acids (^{Staswick et al., 2005}; ^{Ding et al., 2008}; ^{Böttcher et al., 2010}a). Group III proteins seem to be restricted to the genera of Arabidopsis, Brassica, and Gossypium (^{Terol et al., 2006}). The first characterization of a member of this group (At GH-12) recently showed that, by conjugation of 4-substituted benzoates, this protein might be involved in the regulation of chorismate-derived pathways (^{Okrent et al., 2009}). Two grapevine proteins (Vv GH3-7, Vv GH3-9) were present in group I, Vv GH3-1–Vv GH3-6 belonged to group II, and Vv GH3-8 clustered outside of the groups together with a Populus sequence (Pt GH3-14) and a rice protein (Os GH3-6). This rice sequence has already been described as a GH3-related sequence with no clear phylogenetic position (^{Terol et al., 2006}).

Grapevine GH3 proteins shared 31–86% identity and 51–92% similarity with the highest per cent sequence similarity found between Vv GH3-1 and Vv GH3-2. The overall identity and similarity between the grapevine sequences and other plant GH3 proteins ranged from 29–88% and 48–93%, respectively.

With GH3-1 already a potential candidate as a deactivator of exogenous auxins in grapes it was tested whether the corresponding genes of any of the five remaining group II grapevine GH3 proteins were expressed in berries (Fig. 3A) and induced by auxin treatments (Fig. 3B). All five GH3 genes displayed high transcript accumulation in Cabernet Sauvignon flowers and young berries, peaking at flowering (GH3-5), 2 wpf (GH3-6), 4 wpf (GH3-3, GH3-4) or 6 wpf (GH3-2) (Fig. 3A). During this period of berry development auxin levels are high (^{Inaba et al., 1976}; ^{Cawthon and Morris, 1982}; ^{Zhang et al., 2003}; ^{Deytieux-Belleau et al., 2007}; ^{Böttcher et al., 2010}a), therefore the activity of these GH3 proteins in the early stages of berry development might contribute to the control of auxin homoeostasis. In contrast to GH3-1, which has a second peak of expression at veraison (^{Böttcher et al., 2010}a), transcript levels of all the other five group II GH3 genes declined to be low at veraison and remained low for the rest of berry development.

When Shiraz berries at 22 d and 12 d prior to veraison were exposed to 0.5 μM IAA, NAA, or BTOA for 24 h in an ex planta induction experiment, GH3-2 was the only gene out of the five tested that showed a significant induction of expression in response to all three auxin treatments (Fig. 3B). At both stages of berry development tested, NAA and BTOA were stronger inducers than IAA, the maximum being a 20-fold induction of GH3-2 transcript levels 12 d preveraison by NAA. A similar but less pronounced response to the three auxins has also been reported for the closely related GH3-1 (^{Böttcher et al., 2010}a).

The treatments with IAA and NAA also led to a significant but marginal increase of GH3-3 expression in berries 22 d preveraison, whereas there was a decrease in transcript levels of this gene in response to all three auxins 12 d preveraison (Fig. 3B). Of the three other GH3 genes tested, two (GH3-5, GH3-6) were not induced by auxin treatments and the expression of GH3-6 was repressed in berries 12 d preveraison (Fig. 3B). The transcript accumulation of GH3-4 was up-regulated 2–3-fold by IAA, NAA, and BTOA 22 d prior to veraison, when transcript levels were very low. In berries 12 d preveraison, the expression of GH3-4 did not change in response to NAA and was significantly repressed by treatments with IAA and BTOA.

Since GH3-1 and GH3-2 were the only group II genes displaying a clear induction in expression in response to the application of IAA, NAA, and BTOA, all further experiments focused on those two genes and the corresponding proteins. Like GH3-1 (^{Böttcher et al., 2010}a), the predicted coding sequence of GH3-2 contains all three sequence motifs involved in ATP/AMP binding (¹⁰⁸SSGTSAGERK¹¹⁷ (Motif I), ³³⁷YASSE³⁴¹ (Motif II), ⁴¹³YRVGD⁴¹⁷ (Motif III)) that are characteristic of the acyl-adenylate/thioester-forming enzyme superfamily (^{Chang et al., 1997}). To get initial information about the in vitro function of GH3-2, the purified recombinant protein (Fig. 4A) was tested for its ability to conjugate IAA to 20 amino acids. The reaction mixtures were analysed by thin layer chromatography (TLC) (Fig. 4B) and products with a lower mobility than IAA were only observed in reactions containing Asp and Trp. These putative IAA conjugates had the same R_F as the corresponding standards (Fig. 4C) and were confirmed to be IAA–Asp and IAA–Trp using liquid chromatography-mass spectrometry (LC-MS) (data not shown). Due to the adenylation activity of GH3-2, the formation of the conjugates was dependent on ATP. An incubation at 70 °C resulted in a complete loss of activity of the enzyme (Fig. 4C).

The high degree of similarity between the protein sequences of GH3-2 and GH3-1 seems to translate into identical substrate specificities since IAA–Asp and IAA–Trp have also been reported as the sole reaction products of GH3-1 (^{Böttcher et al., 2010}a). To date a similarly strict requirement for just one or two amino acids has only been demonstrated for the Arabidopsis group III protein GH3-12, which, in conjunction with 4-aminobenzoate as the acyl substrate, only conjugated glutamic acid (^{Okrent et al., 2009}).

To analyse the specificities of GH3-1 and GH3-2 for their amino acid and acyl substrates, kinetic parameters for both enzymes were determined by quantifying the accumulation of conjugates using LC-MS/MS. Suitable protein amounts and reaction times for the formation of IAA–Asp and IAA–Trp were identified by adding increasing amounts of protein (0–1 μg) to standard reaction mixtures and stopping the reactions at various times (0–20 min). For both enzymes and substrates it was found that between 0.25–0.75 μg protein and for up to 15 min the reactions were linear and proportional to the amount of added protein (data not shown). For the determination of initial velocities under varied experimental conditions, all reactions were initiated by the addition of 0.5 μg of either GH3-1 or GH3-2 and stopped immediately after the addition of enzyme (0 min) as well as after 2, 5, and 10 min. The initial velocity data (nmol min⁻¹ mg⁻¹ protein) obtained for each substrate concentration series was fit to the Michaelis–Menten equation, exemplified by the formation of IAA–Asp with varying IAA concentrations (Fig. 5). Table 3 shows the kinetic parameters obtained in that way for the conjugation of IAA with Asp (IAA, ATP, and Asp varied) and with Trp (Trp varied). GH3-1 and GH3-2 showed very similar trends in their substrate affinities (K_m) and turnover rates (k_cat), having higher affinities for IAA than for ATP or the amino acids. With KmIAA10.4 μM (GH3-1), KmIAA28.4 μM (GH3-2) and corresponding turnover rates of 14.3 min⁻¹ (GH3-1) and 32.6 min⁻¹ (GH3-2) the grapevine enzymes had 4–30-fold higher affinities and catalytic efficiencies (k_cat/K_m) than have been reported for acyl substrates of rice (^{Chen et al., 2009}, ²⁰¹⁰) and Arabidopsis (^{Okrent et al., 2009}) GH3 proteins. The grapevine GH3s conjugated Asp 100-fold (GH3-1) and 80-fold (GH3-2) more efficiently than Trp, making Asp the preferred amino acid substrate. In addition the physiological concentrations of Trp in grapes are about 100-fold lower than the KmTrp values determined for GH3-1 and GH3-2, whereas Asp concentrations are about 10-fold higher than the respective KmAsp values (Table 3; C Davies, MR. Thomas, P Corena, personal communication). This is in agreement with the previous finding that IAA-Asp accumulated in considerable amounts in grapes whereas IAA–Trp could not be detected in berries of various developmental stages (^{Böttcher et al., 2010a}).

For GH3-1 and GH3-2 a change of the acyl substrate from IAA to NAA or BTOA with Asp as the amino acid substrate led to dramatic reductions in affinities, turnover rates, and resulting efficiencies (Table 4). With NAA as the auxin, substrate efficiencies were 50-fold (GH3-1) and 21-fold (GH3-2) lower when compared with IAA (Table 3) and the formation of the BTOA–Asp conjugate was either not detected (GH3-1) or was 47 000-fold less efficient than the formation of IAA–Asp (GH3-2).

It has been suggested before that the naphthalene ring of NAA might not be a perfect steric fit for the active site of GH3 proteins (^{Chen et al., 2010}) which possibly resulted in the decrease in catalytic efficiencies observed for GH3-1 and GH3-2. BTOA, with a sulphur atom in the indole ring and an oxyacetic acid side chain is structurally even less closely related to IAA than NAA which could explain why it was such a poor substrate for the grapevine GH3 proteins. Assuming that NAA and BTOA applied to preveraison berries cannot be metabolized by alternative pathways their inefficient conjugation by GH3-1 and GH3-2 (Table 4) could lead to the berries being exposed to high levels of free auxinic compounds for extended periods which, in turn, could be the reason for the observed delay in ripening (Fig. 1). The fact that BTOA is less acceptable as a GH3 substrate than NAA, which itself is a worse substrate than IAA, matches the degree of berry ripening delay reported in this study (Fig. 1; Table 1). Although the presented data suggest that NAA and BTOA are poor substrates for the analysed grapevine GH3 proteins, it has to be considered that changing the acyl substrate might influence the affinity for the amino acid substrate as has recently been demonstrated for the group III Arabidopsis GH3-12 (^{Okrent et al., 2009}).

If a delay in berry ripening after treatment with NAA and BTOA was caused by a lack of GH3-catalysed conjugation of these compounds, the opposite would have to be true for exogenous IAA which had no effect on the timing of ripening. It was therefore determined if the application of IAA to preveraison berries (Fig. 1) led to the formation of IAA–Asp. One day after berries were sprayed with IAA a 7.5-fold increase in IAA levels was measured (Fig. 6A) which was matched by a 200-fold increase in IAA–Asp levels (Fig. 6B). The discrepancy between the elevation of IAA and IAA–Asp levels is probably due to the fact that conjugation and therefore accumulation of the IAA–Asp conjugate started in the 24 h period before the first sampling. The concentration of IAA was back to Control levels 2 d after the application, whereas IAA-Asp levels decreased more slowly and were still elevated 1 week after the treatment. The reduction of the IAA–Asp levels is probably due to oxidation of the bound IAA (^{Östin et al., 1998}; ^{Staswick et al., 2005}), a process that links this conjugate to the irreversible inactivation of IAA.

In planta activities of GH3 enzymes in Arabidopsis (^{Staswick and Tiryaki, 2004}; ^{Staswick et al., 2005}; ^{Okrent et al., 2009}) and rice (^{Ding et al., 2008}) appear consistent with in vitro findings and therefore the increased levels of IAA–Asp in IAA-treated berries (Fig. 6B) could be an indication of GH3-1/GH3-2 activity. To investigate this possibility further, the gene expression of GH3-1 (Fig. 6C) and GH3-2 (Fig. 6D) was analysed in the same berry tissues. Transcript levels of GH3-1 were similar in Control fruit and IAA-treated berries with GH3-1 expression transiently increasing 3.5–4-fold 2 d after the applications (Fig. 6C). This indicates that the treatment itself induced GH3-1 expression, possibly as a response to wounding. By contrast, transcript levels of GH3-2 were increased 12-fold 1 d after the IAA treatment followed by a decrease back to Control levels 1 week after the auxin application. From these result it can be deduced that GH3-2 might be linked with the accumulation of IAA–Asp and therefore the decrease of free IAA levels in IAA-treated preveraison berries. An involvement of GH3-1 in the conjugation of applied IAA was not supported by the presented data, but it cannot be excluded that GH3-1 might have participated in the IAA–Asp formation in the time period between treatment and the first sampling after 24 h. It also has to be considered that, depending on the basal levels of GH3 proteins available, the transcriptional induction of GH3 genes might not be requisite for the initial conjugation of IAA.

In conclusion, acyl substrate preferences of an auxin-inducible IAA-amido synthetase expressed in preveraison berries might be responsible for the variations in ripening delay caused by the application of different auxinic compounds.

Studies on a variety of plants have previously shown that the effect of auxinic compounds is highly dependent on the plant species, the plant organ, the compound and its concentration, as well as the physiological process assayed (Ferro et al., 2006). It will be interesting to investigate whether these differences in auxin response can be correlated with expression patterns and specificities of GH3 proteins.

The ubiquitious distribution of GH3 genes in the plant kingdom, as well as numerous studies on their function in Arabidopsis, rice, and the moss Physcomitrella patens have established GH3-catalysed auxin conjugation as an important factor in plant growth, organ development, and response to light and stress (^{Woodward and Bartel, 2005}; ^{Ludwig-Müller, 2011}). The findings presented in this paper further support the addition of fruit ripening (^{Liu et al., 2005}; ^{Böttcher et al., 2010}a) to the list of GH3-regulated processes.