In the second step of the phenyl propanoid pathway, cinnamic acid is hydroxylated at the four position (see Fig. 2A) by HT-P450 hydroxylase, C4H (EC1.14.13.11), to produce p-coumaric acid. An apple orthologue, MdC4H1, was identified from the HortResearch apple EST database by sequence similarity 15 to the Arabidopsis gene, At-C4H (At2g30490), and a full length cDNA clone inserted into the pGreenII 62-SK vector (Fig. 1). As NADPH(+) HT-P450-reductase activity (EC 1.6.2.4) has been shown to be necessary for assaying HT-P450 activity in both bacteria 16 and yeast 9, we also identified a cDNA homologous to the Arabidopsis HT-P450 reductase (At4g30210), corresponding to Ad HT-P450 reductase, from kiwifruit and inserted it into pGreenII 62-SK.

Initial experiments assessed the level of enzyme activity from transiently expressed Md-C4H1 and the effect of Ad-P450 reductase 10 days after Agrobacterium infiltration. The extent of conversion from cinnamic acid to p-coumaric acid, measured by HPLC, was similar for both the MdC4H1 and the MdC4H1 + Ad-P450 reductase, and significantly higher than for a transiently transformed empty vector control (data not shown). In addition, assaying Agrobacterium cultures just before infiltration did not reveal endogenous C4H enzyme activity in the bacteria. This suggested that the endogenous activity in the tobacco leaf provided sufficient NADP-dependent P450 reductase capacity and that additional transient expression of the reductase was not necessary. In this regard, transient assays in plants differ from other expression systems where an exogenous reductase is required for optimal enzymatic activity. The additional expression of the viral silencing suppressor P19 did not significantly increase the enzyme activity (data not shown). While expression of P19 has been shown to enhance the expression of other transiently expressed genes in other studies 5, in the case of HT-P450 it appears that gene expression levels are not impaired by gene silencing. This indicates that either the microsomal membrane surface area available to facilitate the reaction or the availability of the infiltration substrate is limiting

Having established our ability to measure the conversion of cinnamic acid to p-coumaric acid, we further optimized the assay to determine the optimal period for substrate infiltration. The HPLC peak heights for both substrate (cinnamic acid) and product (p-coumaric acid) were plotted as a function of the time between substrate infiltration and extraction. The conversion of cinnamic acid to p-coumaric acid increased up to one hour, although subsequently the level of p-coumaric acid dropped (Fig. 2B). The concentration of cinnamic acid decreased over the whole time period tested. No p-coumaric acid was detected in the control. This decrease in the product may suggest that p-coumaric acid was modified by endogenous enzymes during that time. This is further supported by the appearance of a second product peak at an earlier retention time in the 2 h time point chromatograph.

We next investigated the time between infiltration of Agrobacterium and the substrate. Time points were chosen up to six days (with 60 min between substrate infiltration and extracting leaves) (Fig. 2C). At t = 0, there was also little conversion of cinnamic acid to p-coumaric acid, demonstrating a low endogenous C4H activity of the N. benthamiana leaf and the absence of C4H activity from the infiltrated Agrobacterium (Fig. 2C). The levels of un-reacted cinnamic acid declined as p-coumaric acid accumulated over the time period to six days (Fig. 2C). Thus it was concluded that six days' infiltration of Agrobacterium and 60 min infiltration of substrate provided a sensitive assay of HT-P450s.

We further verified the utility of the method with two other genes; Actinidia deliciosa galactose dehydrogenase, 17, and L-galactose-1-phosphate phosphatase 18. Both genes expressed high levels of activity for their respective enzymes. In the case of galactose dehydrogenase, expression levels were ~100 times higher than P19 infiltrated control leaves after 7 days after infiltration, while the phosphatase showed high levels of expression of a L-Galactose-1-P specific phosphatase compared with undetectable activity in the control (data not shown).

The promoter sequences for four chalcone synthase (CHS) genes (EC 2.3.1.74) from each of four species: Arabidopsis, apple (Malus domestica), pea (Pisum sativum) and petunia (Petunia hybrida) were each inserted into the multiple cloning site of pGreenII 0800-LUC (Fig. 3). In all cases the promoter was modified to introduce an NcoI site at the 3' end of the sequence (at the CHS initiation codon, ATG), allowing the promoter to be cloned as a transcriptional fusion with the LUC gene. Thus, TFs that bind the promoter and increase transcription could be identified by an increase in LUC activity. The promoters for Malus domestica MdCHS1 [GenBank: DQ026297] identified as a 1.3 Kb PCR fragment in the Ecl136II-linker-ligation library [genebank DQ022678], Arabidopsis CHS (TT4; AT5G13930 insert reference), petunia CHS-A [GenBank: X14591] and Pisum sativum CHS-1a [GenBank: X80007] were isolated from genomic DNA. In the same construct, a REN gene under the control of a 35S promoter provided an estimate of the extent of transient expression (Fig. 3). Activity is expressed as a ratio of LUC to REN activity.

The reporter construct, transformed into Agrobacterium, was mixed with an Agrobacterium strain carrying a 35S-MYB construct and co-infiltrated into tobacco leaves.

In the absence of a MYB TF, the LUC to REN ratio was low. This background level of activity presumably represents basal levels of MYBs present in tobacco leaves. In virtually all cases, the addition of a TF to the infiltration mixture increased the relative level of LUC activity compared with the background promoter activity in the absence of Agrobacterium containing a candidate TF (Fig. 4). The majority of TFs tested showed a similar, slightly higher than background, level of activity. We speculate that many of these TFs will have a low affinity to these promoter sequences, and thus cause a small non-specific trans-activation. We hypothesize that where the interaction between TF and promoter was more specific trans-activation, there is a significant increase in the LUC activity relative to REN (Fig. 4). For six of the promoter-TF comparisons, multiple repeats of the assay using six independently infiltrated leaves with the same Agrobacterium culture were used to generate the standard error of the data (Fig. 4B). Preliminary assays showed there were little day-to-day, plant to plant or leaf positional effects on the calculated ratio (data not shown). In addition, the concentration of Agrobacterium in the infiltration buffer did not affect the ratio significantly (data not shown). In these experiments an error of 14–17% was observed, similar to previous transient assay error estimations. We chose not to estimate the error for the remaining assay but set a threshold such that the REN value had to be above 100 relative light units (RLU) to be assured of a comparable degree of error in the remainder of the data.

The utility of the luciferase vector was demonstrated by identification of specific TFs that showed significantly higher LUC to REN ratios than background, compared with the average of other TFs. Instances where a promoter-TF interaction was identified usually resulted in at least ten times higher LUC to REN ratio than the average background compared with other MYBs, which showed a 1 to 2 fold increase in the relative LUC activity of the promoter alone. Thus we have identified from a pool of 20 transcription factors, those TFs likely to be involved in the regulation of the CHS gene (Fig. 4).

The ability of these transient assays to discriminate between strong and weak trans-activation was significantly influenced by the period of transient expression and the ratio of Agrobacterium carrying the transcription and luciferase reporter cassettes. Transient expression demonstrated clear distinctions between TFs after three to four days. However, prolonged transient expression over eight days (data not shown) reduced differences between TFs, suggesting an accumulation of TF protein to levels that are able to interact with a range of promoter sequences in a generic fashion, rather than with those target promoters that would be utilized in vivo. This promiscuous behavior is not surprising, given the sequence conservation within the R2R3 MYB DNA binding domain 19. It is therefore important to note that the non-specific trans-activation in transient assays can be avoided by optimising the period of time between inoculation and assay.

We also found that increasing the ratio of the TF-containing Agrobacterium to the promoter-LUC-REN fusion containing Agrobacterium gave a clearer difference between high and low strength trans-activation. We were unable to determine TF interactions when the promoter-LUC fusion compromised less than 10% of the total infiltrate; this was because the correspondingly low luciferase activity made ratio measurements more variable.

There were significant trans-activation similarities between most of these CHS promoters; notably MdMYB22 [GenBank: DQ074470] was the strongest TF in trans-activating LUC in three of the CHS promoters tested: Arabidopsis (Fig. 4A), Petunia (Fig. 4D) and Apple (Fig. 4B). MdMYB22 has sequence homology to the maize P gene [GenBank: M73029] and AtMYB12 [AT2G47460] which has been shown to regulate genes involved in the flavanol biosynthesis pathway 20. In addition MdMYB18 [GenBank: DQ074466] up-regulated LUC when fused to the Pea CHS-1a promoter (Fig. 4C).

For each promoter tested there was a distribution of background trans-activation, which could be distinguished from a specific promoter interaction by the distribution of trans-activation (Fig. 5A). By examining the petunia CHS-A promoter data there appear to be three groups of TFs. Group 1 showed little or no trans-activation; this we take to be the background promoter activity in these assay conditions. Group 2 showed a normal distribution of activities significantly above background. This group included MdMYB18 [GenBank: DQ074466] and represents interactions we take to be poor or non-specific trans-activation. Group 3, in this case MdMYB22 [GenBank: DQ074470], was a TF that appears to have a specific trans-activation of the promoter. None of the apple MYB TFs was able to trans-activate the Arabidopsis dihydro flavonoid reductase promoter (TT3, At5g42800) (Fig. 5B), although AtMYB-75 (AtPAP1, At1g56650), which is known to regulate the anthocyanin pathway 21 was able to trans-activate this promoter fusion; this further demonstrates that the trans-activation profile for these CHS promoters was specific.

We have adapted the transient assay system in order to measure the effects on RNA silencing of the structure of the dsRNA hairpin and the influence of silencing suppressers. We were particularly interested in determining the role that RNA processing might play in efficient dsRNA production, and whether the presence of introns within hairpin cassettes can enhance the efficiency of RNA silencing activation 22. Our RNA silencing assay system consists of the LUC and REN luciferase reporter genes in opposite orientation on a single T-DNA, each under the transcription regulation of the 35S promoter (Fig. 6A). Between the two reporter genes is a LUC hairpin sequence (hpLUC), under the control of the nopaline synthase (NOS) promoter. In the absence of a NOS-hpLUC hairpin both the LUC and REN activities accumulated (Fig. 6B).

To test the efficiency of transient silencing and to quantify the effect that an intron has on this efficiency, we built two constructs: pGreenII 0800-3.6 (+) and pGreenII 0800-3.6 (-), each including an intron in either splicable (+) or inverse (-) orientation of the hairpin (Fig. 6B). Both the NOS-hpLUC cassettes were effective in significantly reducing the level of LUC activity (Fig. 6B). This reduction in LUC activity was the result of the activation of RNA silencing because the co-infiltration of Agrobacterium expressing P19, a viral suppressor of gene silencing 5, was effective in overcoming the silencing effect of the LUC hairpin, and restoring LUC levels to those of the control (Fig. 6C).

Notably, the relative level of LUC in the hairpin construct that contained a functional, splicable intron was lower (0.0171 ± 0.0014) than from the equivalent cassette where the intron was in an inverted orientation (0.0394 ± 0.0099) such that it cannot be processed. This observation appears to be consistent with previous reports 22 which showed an increased in the percentage of transgenic plants with a gene silencing phenotype when the dsRNA hairpin constructs contained an intron. However, in these transient assays the absolute levels of LUC were very similar for both hairpin constructs, and the difference in relative luciferase level could also be achieved by elevating the levels of REN in the construct that contained the splicable intron. As the NOS-hpLUC cassette and the 35S-REN cassettes are in tandem, it is possible that the NOS promoter or the processing of the intron is influencing the transcriptional activity of the REN gene. To test this possibility we built and assayed a further two constructs, pGreenII 0800-3.1 (+) and (-) (Fig. 6B). In these constructs the NOS-hpLUC cassette directs transcription in the opposite orientation to that of pGreenII 0800-3.6, such that any cis-activation from the NOS cassette should influence 35S-LUC activity, but not 35S-REN activity. In transient assays using these constructs there was no significant difference between the hairpin with a splicable intron (0.0492 ± 0.0032) and the equivalent construct with an inverted intron (0.0504 ± 0.0029). This suggests that including an intron in the construction of a cassette that generates a dsRNA hairpin may function to influence the expression of any downstream tandem genes. In the case of the pHannibal vector, 13 this would correspond to the kanamycin resistance gene NptII, used for transformation selection. This suggests that the enhanced percentage of silencing described by Smith et al 22 may in part be due to enhanced recovery or stability of transgenic lines through improved expression of the kanamycin selection gene. We were unable to identify by RT-PCR, any RNA species that represented a hybrid RNA between the NOS promoter and either the LUC or REN gene, suggesting that the processing event may act as a transcriptional enhancer rather than effect the initiation of transcription at a distance.

RPH-138: GTGAGAGGTCCTAAGCTTATGTCCGGTTAT; RPH-139: CTTGACTGGCGAGAATTCCCACGATCTCTTT; RPH-140: ATTGACAAGGATGGATCCCTACATTCTGGA; RPH-141: CACGATCTCTTTTTCCGTCATCGTCT; RPH-146: TGGCCTTTATGAGGGAATTCCTGATTTTTC; RPH-179: TGGCGGTTTTGGTACCCCGGGTCAAC; RPH-180: CCATCACCATGGTAGTATACACCAAC; RPH-198: CACACAGTTGGGAGGAGTTGCTGTCCC; RPH-199: CTTGCGAACTTCTTCGACGGTCACCAT; RPH-212: AATTGGTACCGATATCGAGCTC; RPH-213: AGCTGAGCTCGATATCGGTACC; RPH-332: ACTCCTCGACTGTCACCATGGTTGCTTG; RPH-333: GCGGAAGGGTACCGAATTCATAGCAACTGG

The cauliflower mosaic virus (CaMV) 35S expression cassette (KpnI-BglII) of pJIT-62 26 was inserted into the multiple cloning site (MCS) (KpnI-BamHI) of pGreenII-0000 24. Flanking sequences from the LB to the 35S-promoter and from the CaMV-terminator to the RB were deleted by digestion, (StuI-KpnI and SpeI-HpaI respectively), T4 pol and re-ligation to produce pGreenII-62-1. The pUC9 MCS was converted to the pBluescript MCS by ligating two oligonucleotides (RPH-212 and RPH-213) to the EcoRI-HindIII cut vector, generating KpnI and SacI sites, then inserting a KpnI-SacI fragment from pBluescript SKII+ (Stratagene).

The HortResearch apple EST libraries were constructed in either the Lambda ZAP Express or the Lambda ZAP II vector systems (Stratagene), resulting in cDNAs cloned unidirectionally as EcoRI-XhoI fragments in pBK-CMV or pBluescript SK^-, respectively.

Genes encoding various enzymes and transcription factors were cloned into one of two plant expression vectors derived from the pART7/27 binary vector system 27. Both vectors carry the same transcriptional regulatory signals for plant gene expression, namely the CaMV35S promoter and octopine synthase terminator. The T-DNA border elements, chimeric kanamycin selectable marker and vector backbone are identical in both vectors. The two derivatives differ in that one is suitable for conventional restriction enzyme cloning of cDNAs whilst the second facilitates Gateway recombination cloning of the cDNAs.

In order to improve the compatibility of restriction sites in the multiple cloning site of pART7 with those of the pBK-CMV or pBluescript SK^- EST library vectors, the XhoI-XbaI region of pART7 was replaced with the SalI-XbaI multiple cloning site region of pBK-CMV, generating pSAK7. cDNAs from the EST libraries were cloned into pSAK7 as either EcoR1-XhoI or BamHI-XhoI fragments, placing them under the transcriptional control of the CaMV 35S promoter. The 35S-cDNA-ocs3' cassette was then cloned as a Not1 fragment into pART27, to generate the plant gene expression construct.

Where conventional cloning was more problematic, due to the lack of suitable restriction sites, a Gateway-adapted version of the pART7/27 plant transformation system was utilised. This Gateway-adapted version was produced by cloning the CaMV35S promoter, multiple cloning site and octopine synthase transcriptional terminator cassette of pART7 as a NotI fragment into pART27. Subsequently, the 1711 bp Gateway RfA cassette (Invitrogen Corp.) was cloned into the SmaI site of the multiple cloning site to generate pHEX2 (35S-attR1-Cm^R-ccdB-attR2-ocs3'). cDNAs from the EST libraries were cloned into pHEX2 using Gateway recombination technology and all Gateway reactions were performed as recommended by the manufacturer (Invitrogen Corp.). cDNAs were amplified using universal primers designed to the multiple cloning site regions of the pBK-CMV or pBluescript SK^-. The primers used for pBluescript SK- clones were 5-GGGGACAAGTTTGTACAAAAAAGCAGGCTCCCCGGGCTGCAGGAATTC-3' and 5'- GGGGACCACTTTGTACAAGAAAGCTGGGTCCGGGCCCCCCCTCGAG-3' and the primers used for pBK-CMV clones were 5'-GGGGACAAGTTTGTACAAAAAAGCAGGCTTCGACACTAGTGGATCCAAAGAATTC-3' and 5'-GGGGACCACTTTGTACAAGAAAGCTGGGTGCCGCTCTAGAAGTACTCTCGAG-3'

Amplification with these primers results in PCR products with attB ends which were recombined with the attP sites of the Gateway pDONR201 vector, creating pENTRY vectors. All pENTRY vector clones were sequence verified to ensure the fidelity of the cDNA sequence. Gateway attP × attR reactions were then carried out with the pENTRY vector and the pHEX2 Destination vector, to generate the plant gene expression construct.

Enzyme expression plasmids. The apple cDNA clone of MdC4H1 [GenBank: DQ075002] was inserted into pGreenII 62-SK as an EcoRI-XhoI fragment. The genes for L-galactose dehydrogenase [GenBank: AY176585] and L-galactose-1-P phosphatase [GenBank: AY787585] from kiwifruit were cloned as previously described 18 and transformed into Agrobacterium using standard methods.

The Renilla reporter gene pRL-null (Promega, Madison, WI) was modified to remove the mammalian 5'UTR intron by digestion with NheI-SpeI, T4 Pol, followed by re-ligation. This gene was inserted into the expression cassette p35S-2 24, and the flanking sequences deleted as above. This cassette was inserted into the HpaI site of pGreenII 0000, to produce pGreen 0800-1. A 35S-LUC-expression cassette was inserted into pGreenII 0800 to produce pGreenII 0800-35S-LUC. The flanking sequence between the CaMV terminator and the RB was deleted (SpeI-StuI, T4 pol, ligation). Finally the pBluescript MCS replaced the 35S promoter to produce pGreenII 0800-LUC. These vectors are available on request.

A 0.97 kb region of the Arabidopsis CHS promoter (At5g13930) was amplified by PCR from the Arabidopsis ecotype Columbia using the primers RPH-179 and RPH-180, then digested with KpnI and NcoI and cloned into the MCS of pGreenII 0800-LUC. The 1.04 kb pea CHS-1a promoter [GenBank: X80007] was subcloned into pGreenII 0800-LUC as an EcoRI-NcoI fragment 28. A 0.92 kb Petunia CHS-A promoter [GenBank: X14591] was amplified by PCR using primers RPH-332 and RPH-333 from a V26 genomic DNA, digested with KpnI and NcoI and cloned into pGreenII 0800-LUC. The 1.3 kb Apple CHS1 promoter [GenBank: DQ022678] was isolated from Malus domestica Royal Gala, using the Genome Walker kit (Clonetech) with gene specific primers RPH-198 and RPH-199, then cloned into pGEM T-easy (Promega) and subcloned as a SalI-NcoI fragment into pGreenII 0800-LUC. pwo-Polymerase (Roche) was used for all PCR amplifications and cloned genes were sequenced to confirm no sequence modifications were incorporated.

A short hairpin was built by asymmetric amplification of a region of the LUC gene: primers RPH-138 and RPH-146 were used to amplify a 454 bp fragment introducing a HindIII and EcoRI site near the end of the PCR product. The second PCR product corresponding to the antisense region of the hairpin was a 283 bp PCR product of amplification using primers RPH-140 and RPH-139. The first primer, RPH-140, was from the same region as RPH-138 but introduced a BamHI site near the end of the PCR product. RPH-139 was 142 bp closer to RPH-138 and RPH-140 than RPH146, and also introduced an EcoRI site near the end of the amplified product. The PCR products were digested with appropriate restriction enzymes and these two digested PCR products were used in a 3-way ligation to HindIII-BamHI cut pNOS-7 24 to produce the NOS-hpLUC cassette. This NOS-hpLUC was modified to replace the loop region with an intron by amplifying the NOS-hpLUC cassette with primer RPH-141, common to both the sense and antisense region, with pwo-polymerase. An intron sequence from a kiwifruit terpene synthase genes [GenBank: DQ026298] was also amplified with primers RPH-099 and RPH-100. The intron amplification product was treated with polynucleotide kinase (NEB) and used in a ligation with the non-phosphorylated amplification product of the NOS-hpLUC cassette. The resulting cassettes either contained an intron in +ve (RPH-099 to RPH-100 relative to the NOS promoter) or -ve (RPH-100 to RPH-099 relative to NOS promoter) configuration. Accordingly the +ve intron configuration had a splicable GT to AG intron arrangement, the -ve intron configuration did not. The nos-hpLUC cassette and intron containing derivatives were inserted into a 35S-LUC, 35S-REN cassette; pGreenII 0800 35S-LUC, as an EcoRV fragment. In this way the reporter cassettes were generated with the intron containing NOS-hpLUC cassette in both orientations. pGreenII 0800-3.6 had the NOS promoter directing transcription in the same orientation as the 35S-REN and towards the T-DNA Right border (forward). pGreenII 0800-3.1 had the NOS promoter directing transcription of the hpLUC in the same orientation as the 35S-LUC and towards the T-DNA left border (reverse) (Fig. 6A).

The pSoup helper plasmid 24 was converted to a T-DNA carrying version by inserting a BglII fragment that contained the T-DNA from pGreen0000 into the unique BamHI site of pSoup creating pSoup0000. A 35S-P19-CaMV fusion was isolated from pBIN-61-P19 5 as a SacI-BglII partial and inserted into the pSoup0000 multiple cloning site to produce pSoup-P19.