Transcription factors, sucrose, and sucrose metabolic genes interact to regulate potato phenylpropanoid metabolism

Much remains unknown about how transcription factors and sugars regulate phenylpropanoid metabolism in tuber crops like potato (Solanum tuberosum). Based on phylogeny and protein similarity to known regulators of phenylpropanoid metabolism, 15 transcription factors were selected and their expression was compared in white, yellow, red, and purple genotypes with contrasting phenolic and anthocyanin profiles. Red and purple genotypes had increased phenylalanine ammonia lyase (PAL) enzyme activity, markedly higher levels of phenylpropanoids, and elevated expression of most phenylpropanoid structural genes, including a novel anthocyanin O-methyltransferase. The transcription factors Anthocyanin1 (StAN1), basic Helix Loop Helix1 (StbHLH1), and StWD40 were more strongly expressed in red and purple potatoes. Expression of 12 other transcription factors was not associated with phenylpropanoid content, except for StMYB12B, which showed a negative relationship. Increased expression of AN1, bHLH1, and WD40 was also associated with environmentally mediated increases in tuber phenylpropanoids. Treatment of potato plantlets with sucrose induced hydroxycinnamic acids, flavonols, anthocyanins, structural genes, AN1, bHLH1, WD40, and genes encoding the sucrose-hydrolysing enzymes SUSY1, SUSY4, and INV2. Transient expression of StAN1 in tobacco leaves induced bHLH1, structural genes, SUSY1, SUSY4, and INV1, and increased phenylpropanoid amounts. StAN1 infiltration into tobacco leaves decreased sucrose and glucose concentrations. In silico promoter analysis revealed the presence of MYB and bHLH regulatory elements on sucrolytic gene promoters and sucrose-responsive elements on the AN1 promoter. These findings reveal an interesting dynamic between AN1, sucrose, and sucrose metabolic genes in modulating potato phenylpropanoids.


Introduction
Plants synthesize an array of phenylpropanoids with diverse roles including in plant growth and development, flowering, pigmentation, signal transduction, and structural integrity ( Fig. 1; Koes et al., 2005;Vogt, 2010). Phenylpropanoids are also important plant dietary constituents that possess various health-promoting properties, including against cardiovascular disease and cancers (Parr and Bolwell, 2000). White potatoes (Solanum tuberosum L.) contain modest amounts of phenylpropanoids and are the third largest source of dietary phenylpropanoids because of high consumption (Chun et al., 2005). Red-and purple-flesh potatoes contain decidedly higher amounts of phenylpropanoids (André et al., 2007a) and the increase is largely, but not solely, due to anthocyanin biosynthesis .
Numerous factors mediate expression of phenylpropanoid genes, including sugars and various transcription factors including MYBs (Dubos et al., 2010). MYBs have single or multiple imperfect repeats (R) of structurally conserved DNA-binding domains ( Fig. 2A). R2R3 genes comprise the largest of the four classes of plant MYBs and have a conserved N-terminal DNA-binding domain that is in direct contact with the DNA and a variable C-terminal domain that activates or represses its targets (Dubos et al., 2010). R2R3 MYBs are divided into 22 subgroups in Arabidopsis thaliana based on conserved motifs (Stracke et al., 2001). Various R2R3 MYBs regulate phenylpropanoid biosynthesis, some of which interact with basic helix-loop-helix (bHLH) proteins (Grotewold, 2005). A mutant maize (Zea mays) P1 MYB activates some flavonoid genes but not anthocyanins in the absence of a bHLH, whereas AtMYB12 regulates flavonol biosynthesis without a bHLH interaction (Grotewold et al., 2000;Mehrtens et al., 2005).
Unlike the flavonol branch of the pathway, anthocyanin biosynthesis is typically regulated by a complex in which MYB, bHLH, and WD40 transcription factors interact. When this complex is formed, MYB and bHLH bind to promoters with consensus nucleotide sequences like MACCWAMC and CANNTG (Sablowski et al., 1994;Zimmermann et al., 2004). The first 200 aa of bHLH proteins are required to interact with MYB transcription factors, and aa 200-400 interact with WD40 proteins (Pattanaik et al., 2010). The C-terminal ACT-like domain facilitates binding of MYB to the promoter (Grant, 2006). WD proteins have four to eight imperfect tandem repeats and interact with other proteins through the WD repeat region (Neer et al., 1994).
Sucrose modulates transcriptional and post-translational regulation of many pigment-related genes (Koch, 1996). Sucrose induces anthocyanins in Arabidopsis through induction of PAP1/MYB75 (Production of Anthocyanin Pigment 1) and fails to induce anthocyanins in the pap1 mutant (Teng et al., 2005;Solfanelli et al., 2006). Anthocyanin induction in Arabidopsis seems specific to sucrose, but in grapes (Vitis vinifera), other sugars also stimulate anthocyanin synthesis (Gollop et al., 2002).  As a staple food, potatoes are an attractive target for phytonutrient enhancement. Tuber-specific overexpression of the MYB transcription factor StMtf1 resulted in elevated amounts of phenylpropanoids (Rommens et al., 2008). Tuber anthocyanin synthesis in the periderm is controlled in part by three loci, D, P, and R. P and R were found to be structural genes, whereas D encodes an R2R3 MYB (Jung et al., 2005(Jung et al., , 2009Zhang et al., 2009a). The D locus maps to a region of chromosome 10 that harbours StAN2, which was later renamed StAN1 (Jung et al., 2009). StAN1 expression correlated with anthocyanin levels in drought-stressed potatoes (André et al., 2009).

A
The goal of this study was to determine how sugars and transcription factors modulate biosynthesis of potato phenylpropanoids, including hydroxycinnamic acids, flavonols, and anthocyanins. The involvement of sugars and 15 transcription factors was characterized and provided evidence that AN1 along with bHLH1, sucrose, and sucrose metabolic genes interact to regulate the pathway.

Plant materials
Small tubers (~25-50 g) from NY144, Challenger, ORO4198-1, AmaRosa and Magic Molly were harvested from field-grown plants in Moses Lake, WA, USA, in August 2011. Tubers were peeled and frozen in liquid nitrogen within minutes of harvest. Analysis of environmental influences on transcription factor expression was performed on potatoes grown in Texas, Florida, and four locations in Alaska (Payyavula et al., 2012). For sucrose feeding studies, internodes from 1-month-old potato (cv. Purple Majesty) plants were propagated on MS medium supplemented with 0 or 120 mM sucrose at 25 °C with 16 h light. Plantlets were collected after 5 d.

Phenylpropanoid analysis
Phenolics were extracted from 50 mg freeze-dried homogenized sample with 1.5 ml of 50% methanol, 1 mM EDTA, and 2.5% metaphosphoric acid using a validated method (Shakya and Navarre, 2006). Total phenolics were estimated by the Folin-Ciocalteu method (Singleton and Rossi, 1965). Individual phenolics were analysed on a 100 × 4.6 mm Onyx monolithic C-18 (Phenomenex) column with an Agilent 1100 HPLC system equipped with a quaternary pump, refrigerated autosampler, and column heater. Detectors were a DAD and SL ion trap with an electrospray ionization (ESI) source operated in both positive and negative ion mode, as described previously . Anthocyanins were extracted from 50 mg of dry powder using a total of 2 ml of 50% methanol containing 2.5% formic acid. Total anthocyanins were estimated by a pH differential method (Wrolstad et al., 2005). Individual anthocyanins were determined by liquid chromatography/ mass spectrometry (LC/ MS) as described previously (Payyavula et al., 2012). Retention times and MS data for the quantitated compounds are shown in Supplementary Tables S1 and S2 at JXB online.
RNA extraction, cDNA synthesis, and quantitative real-time PCR (qRT-PCR) RNA was extracted using hot cetyl trimethylammonium bromide (Chang et al., 1993), with slight modification (Payyavula et al., 2012). Samples were quantified with a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies) and quality was assayed on a 1% agarose gel. cDNA synthesis used 2 μg of total RNA, Moloney murine leukemia virus reverse transcriptase (New England BioLabs), and oligo(dT) 20VN primers at 42 °C for 2 h. Quantitative PCR was performed in 384-well plates on a LightCycler 480 (Roche) with 3.5 ng of equivalent cDNA. Relative abundance was calculated with the ΔC T method (Livak and Schmittgen, 2001) using elongation factor-1, actin, and ribosomal protein L2 for template normalization. The primers are listed in Supplementary Table S3 at JXB online. Statistical analysis was performed with standard analysis of variance using SAS 9.2 Proc GLM. The means were obtained using LSMEANS with Tukey adjustments for the degrees of freedom and P<0.05 default separation (Rourke et al., 2005). Pearson correlation coefficients were calculated using Microsoft Excel and visualized with HeatMapper Plus (Verhaak et al., 2006).

Phylogenetic tree and protein similarity
Protein sequences collected by the best BLAST match from the Potato Genome Sequencing Consortium (PGSC) database (http:// solanaceae.plantbiology.msu.edu/pgsc_download.shtml) or NCBI were aligned and a phylogenetic tree developed using MEGA4 (Tamura et al., 2007). The evolutionary history was inferred using the neighbour-joining method with 500 bootstrap replicates. The evolutionary distances were computed using the Poisson correction method with units of the number of amino acid substitutions per site. Percentage similarity was calculated with GeneDoc (Nicholas et al., 1997).

Antioxidant and enzyme assays
A ferric reducing antioxidant power (FRAP) assay was performed as reported elsewhere (Benzie and Strain, 1996). To an aliquot of 15-30 μl of phenolic extract, 1 ml of the pre-warmed (37 °C) FRAP reagent was added, incubated 5 min at 37 °C, and then centrifuged for 1 min. Absorbance at 593 nm (A 593 ) was read and antioxidant capacity was estimated as trolox equivalents. Phenylalanine ammonia lyase (PAL) catalyses the first reaction of the phenylpropanoid pathway ( Fig. 1) and activity was measured as described previously (Zucker, 1965). In a 1.5 ml tube, 700 μl of 50 mM sodium borate buffer (pH 8.8) and 200 μl of 50 mM phenylalanine were pre-warmed at 37 °C prior to adding 100 μl of extract. The reaction was continued for 1 h at 37 °C and then stopped with 200 μl of 12% trichloroacetic acid and measured at A 290 . Sugars were extracted twice at 80 °C for 15 min from 25 mg of freeze-dried sample with a total of 2 ml of 80% ethanol. Pigments were removed by re-extracting with 50 mg of activated charcoal. Sucrose and glucose were estimated using Sigma kits (SCA20 and GAHK20).
A single positive colony was cultured in 5 ml of LB medium overnight and used to inoculate 50 ml of medium. Cells were harvested and redissolved in 10 mM MgCl 2 containing 100 μM acetosyringone and adjusted to an optical density of 0.5. Cultures were then diluted (1:1) with the gene silencing suppression vector p19 TBSV of tomato bushy stunt virus (Voinnet et al., 2003) to avoid co-suppression. Samples were infiltrated into leaves of 3-week-old tobacco plants (Nicotiana tabacum cv. Samsun and Nicotiana benthamiana) maintained in a growth chamber under conditions of 15 h light. Leaves were harvested at 4 d post-infiltration.

Phylogenetic analysis and protein similarity
A total of 12 MYBs were selected for characterization based on BLAST searches conducted against the potato genome database (The Potato Genome Sequencing Consortium, 2011) using nucleotide sequences of transcription factors described previously in the literature. Arabidopsis has 198 MYB genes (Yanhui et al., 2006) of which AtPAP1 (GenBank accession no. AF325123) and AtPAP2 (MYB90, AF325124) are known to be involved in anthocyanin synthesis (Dubos et al., 2010). Three putative potato MYB genes, StMYB75A, StMYB7B, and StMYB75C, were identified based on homology to AtPAP1 and AtPAP2. StMYB12A and StMYB12B are homologous to SlMYB12 (EU419748) and AtMYB12 (NM_130314), which regulate flavonols in tomato (Solanum lycopersicum) and Arabidopsis (Mehrtens et al., 2005). StAN1 and StAN2 (AY841127 and AY841131) are MYBs implicated in regulating anthocyanin biosynthesis in tuber skin and drought-stressed potatoes (André et al., 2009;Jung et al., 2009). Numerous sequences were found in the PGSC database with homology to StAN1 and StAN2, but only the two with highest similarity were chosen. StMTF1 (EU310399) and StMTF2 (CV506186) are MYBs shown to regulate potato phenylpropanoids with varying efficacy (Rommens et al., 2008). StMYb73A and StMYb73B were collected by blasting a MYB73 sequence that was upregulated 44-fold in the purple portions of tubers compared with the white portions (Stushnoff et al., 2010). StMYB10 was homologous to NtAn2 (FJ472647) and MdMYB10 (EU518249), which regulate anthocyanin biosynthesis in tobacco floral tissue and apple (Malus domestica), respectively (Espley et al., 2007;Pattanaik et al., 2010).
StWD40 is clustered with NtTTG2, PhAN11, and VvWDR1, and is 97% similar to NtTTG2 and 94 and 88% similar to PhAN11 and VvWDR1, respectively (Supplementary Table S4C), which regulate anthocyanin synthesis in petunia and grapes (Vetten et al., 1997; Matus et al., 2010). These proteins are slightly distinct at the N-terminal end but are highly conserved in the middle and C-terminal end ( Supplementary Fig. S1C).

Basal phenylpropanoid metabolism
Five genotypes, NY144, Challenger, ORO4198-1, AmaRosa and Magic Molly with white, light yellow, dark yellow, red, and purple flesh, respectively, were selected for analysis (Fig. 3A). These genotypes were chosen because they were expected to have markedly different phenylpropanoid profiles; consequently, a comparative analysis of transcription factor expression would be informative. Prior to transcription factor analysis and to provide a context to interpret the results, phenylpropanoid profiles and structural gene expression were evaluated in each genotype. Tubers were collected from field-grown plants, peeled, and processed within minutes of harvest to avoid potential post-harvest effects. Total phenolic levels ranged from 2.9 to 12 mg g -1 (Fig. 3B) and were 3-4-fold higher in red and purple potatoes than in white. Anthocyanins were detectable only in red and purple cultivars, with higher amounts in the purple (10.4 mg g -1 ) than in the red (6.5 mg g -1 ) genotype. Carotenoids were highest (60.7 μg g -1 ) in the dark yellow genotype.
Concentrations of shikimate-derived aromatic amino acids were lowest in the light yellow genotype (Fig. 4D). Concentrations of phenylalanine (Phe), the precursor for phenylpropanoid biosynthesis, did not associate with total phenolic or 5CGA concentrations. However, the enzyme activity of PAL, which catalyses the first committed step of phenylpropanoid pathway where cinnamic acid is formed by deamination of Phe, was about 35-45-fold higher in potatoes with higher amounts of phenylpropanoids ( Supplementary  Fig. S2 at JXB online). Another major source of tuber antioxidant capacity is ascorbic acid, which varied from 1.2 to 2.0 mg g -1 among genotypes (Fig. 4D). Antioxidant capacity was measured by FRAP and was greater in red and purple potatoes ( Supplementary Fig. S2).

Phenylpropanoid structural gene expression
Relative expression of PAL was about 40-fold higher in red and purple potatoes compared with white (Fig. 5A), which is consistent with the amount of phenylpropanoids and PAL enzyme activity in these genotypes ( Fig. 4 and Supplementary Fig. S2). The primary pathway to CGA biosynthesis is thought to be through hydroxycinnamoyl-CoA quinate transferase (HQT). Although 5CGA was the most abundant phenylpropanoid, HQT transcript expression did not track with the 5CGA levels (Fig. 5A). Expression of ρ-coumarate 3-hydroxylase (C3H) and hydroxycinnamoyl transferase (HCT) expression was 2-5-fold higher in red than in other genotypes. The expression of late genes like chalcone synthase (CHS), dihydroflavonol reductase (DFR), and UDP-glucose:flavonoid 3-O-glucosyltransferase (UFGT) were strongly expressed in red and purple genotypes and correlated with anthocyanin levels (Fig. 5B). To our knowledge, no anthocyanin O-methyltransferase (AOMT) gene has been reported in potato or any Solanaceae. By similarity search with grape AOMT (Hugueney et al., 2009), three isoforms were identified and aligned ( Supplementary Fig.  S1D). AOMT1 and AOMT2 shared high similarity (90%), so only one set of primers was used for both. The expression of AOMT1/2 showed no specific trend, while that of AOMT3 was strongly expressed in red and purple potatoes (Fig. 5B).

Transcription factor analysis
Expression of the 15 potato transcription factors listed in the phylogenetic analysis was studied by qRT-PCR in the flesh of the five genotypes. Expression of MTF1, MTF2, MYB10, MYB12A, and MYB75A were below quantifiable levels. Primers amplified an appropriately sized fragment from genomic DNA eliminating sequence compatibility as an issue. The expression data for the remaining ten transcription factors is shown in Fig. 5C. AN1 was by far the most abundantly expressed of those analysed, and was more than 100-fold higher in the flesh of red and purple genotypes. To evaluate whether differences in the AN1 protein might account for the differential phenylpropanoid profiles observed among the five genotypes, full-length coding regions were cloned and sequenced. The sequence was identical among Challenger, AmaRosa, and Purple Majesty, whereas a white and yellow cultivar each had two nucleotide changes resulting in leucine-to-phenylalanine and leucine-to-valine substitutions ( Supplementary Fig. S3a at JXB online).
Sequencing of ~1500 bp of the StAN1 promoter region revealed substantial differences. The promoter region was 97% similar between the low-phenylpropanoid white and yellow genotypes, and 93% similar between the purple and red potatoes. However, only 37% similarity occurred between the white/ yellow and red/purple genotypes. Interestingly, the white/yellow genotypes had only one sucrose-responsive (SURE) element, whereas the red and purple potatoes had six ( Supplementary  Fig. S3b). Similarly, no methyl jasmonate-responsive elements were present in the promoter region examined in the white and yellow potatoes, but five were present in red and four in purple potatoes. Despite using various primers, the promoter sequence of Challenger could not be amplified.
The expression of AN2, MYB75B, and MYB75C was much lower and did not associate with total phenylpropanoid concentrations. Excluding AN1, the expression levels of MYB73A and MYB73B were higher than other transcription factors but did not associate with phenylpropanoid concentrations. MYB12B was poorly expressed in all genotypes, but was the only gene that showed a clear inverse association with phenylpropanoid amounts. MYB12B expression was 2-4fold higher in white and yellow genotypes compared with that in red and purple potatoes.
The expression of bHLH1 in tuber flesh was ~10-20-fold lower than that of AN1 and was detected only in red and purple potatoes (Fig. 5C), whereas bHLH2 was expressed in all potatoes but did not show any association with phenylpropanoid concentrations. WD40 expression was 3-5-fold higher in the red and purple potatoes.

Environmental effects on transcription factor expression
These results suggested that the differences in the phenylpropanoid concentrations among different cultivars were probably partly determined by AN1, bHLH1, and WD40, i.e. they have a role in determining the inter-genotypical variation. However, phenylpropanoid concentrations can also vary significantly within potatoes of a single genotype. Potatoes from the same cultivar grown under the same management regime can have different concentrations in different years or locations. To determine whether AN1, bHLH1, and WD40 were also associated with intra-genotypical variation, purple potatoes grown in four locations in Alaska, plus in Texas and Florida, were examined. The Alaskan potatoes had significantly higher amounts of phenylpropanoids (Payyavula et al., 2012) that correlated with higher expression of AN1, bHLH1, and WD40 ( Supplementary Fig. S4 at JXB online). AN1 expression was strongest in tuber samples from Wiseman, the northern-most site located in the Arctic Circle, and least in the southern-most locations. Similar patterns were observed for bHLH1 and WD40.

Evaluation of the role of sugars in phenylpropanoid metabolism
Sugars were measured in the five genotypes (Fig. 6). The red and purple genotypes accumulated up to 30 and 60% higher sucrose and glucose levels, respectively, which is consistent with a role for sugars in modulating tuber phenylpropanoid content (Zucker and Levy, 1959;Leggewie et al., 2003). On the other hand, there were differences among phenylpropanoids in white and yellow potatoes (Fig. 4A), which did not correlate with sugar content. For example, the dark yellow genotype had higher phenylpropanoid concentrations than the white genotype but equivalent amounts of sucrose and lower glucose.
Preliminary sugar feeding studies showed increases in phenylpropanoids in both white and purple potatoes. Subsequently, the cultivar Purple Majesty was selected for in-depth analysis, and the effect of sucrose on metabolites and gene expression was studied in plantlets cultured on 0 and 120 mM sucrose (Fig. 7). Total phenolics increased by 40% reaching 14 mg g -1 in 120 mM sucrose-treated plantlets (Fig. 7A). Sucrose induced an almost 5-fold increase in total chlorogenic acid content. Numerous other phenylpropanoids including multiple flavonols showed a strong induction by sucrose, with the most abundant flavonol, Kmp, induced almost 15-fold ( Fig. 7A-C). Likewise, the anthocyanin branch of the pathway was strongly induced, with total anthocyanins increasing by ~85% (Fig. 7A). PtCRG was the predominant anthocyanin and increased by 2.5-fold with sucrose feeding (Fig. 7E). Among the less-abundant anthocyanins, some decreased in response to sucrose, such as petunidin 3-rutinoside-5-glucoside (PtRG), which may reflect competition with the more abundant anthocyanins for common precursors. Unlike a majority of the phenylpropanoids examined, the three shikimate-derived aromatic amino acids decreased in sucrose-fed plantlets (Fig. 7D).
Expression of the structural genes PAL, CHS, F3H, DFR, UFGT, AOMT3, and transcription factors AN1 and bHLH1 was more than 3-fold higher in plantlets fed 120 mM sucrose, but WD40 and AN2 showed only a slight increase (Fig. 7F, G). Genes encoding isoforms of the sucrose-hydrolysing enzymes sucrose synthase (SUSY) and invertase (INV) were also studied. The expression of SUSY1 was several fold higher than SUSY4. Both were induced by sucrose, and SUSY4 underwent a >100-fold increase in expression after sucrose feeding (Fig. 7F). Expression of INV1 was greater than INV2, but did not increase in response to sucrose, whereas INV2 expression increased.

Transient leaf infiltration assays
The functionality of the selected potato transcription factors was tested using tobacco leaf infiltration. Leaves infiltrated with empty vector, bHLH, WD40, or combined bHLH plus WD40 did not show pigmentation (Fig. 8A). Anthocyanin accumulation was time dependent and all leaves infiltrated with any combination of AN1 showed pigmentation by 96 h post-infiltration (Fig. 8A). However, at 48 h, visible purple pigmentation was only observed on leaves co-infiltrated with AN1+bHLH ( Supplementary Fig. S5 at JXB online). This accelerated anthocyanin induction was observed across multiple independent experiments. After 72 h, leaves infiltrated with AN1 alone or AN1+WD40 constructs also accumulated purple pigmentation, but were less intense than leaves infiltrated with AN1+bHLH or AN1+bHLH1+WD40.
Cyanidin 3-rutinoside (CR) was by far the most abundant anthocyanin induced in N. tabacum ( Fig. 8E and Supplementary Fig. S6A, E at JXB online). In contrast, when N. benthamiana leaves were infiltrated with StAN1, the major anthocyanin was delphinidin 3-rutinoside (DR; Supplementary Fig. S6B, F). Purple and blue potatoes contain petunidin and malvinidin derivatives, which are methylated products of delphinidin ( Fig. 4F; Hillebrand et al., 2009). A possible reason for not forming downstream products of delphinidin in tobacco could be due to a lack of AOMT activity. To address this question and address the functionality of the potato gene, N. benthamiana leaves were infiltrated with potato AOMT3 alone or in combination with AN1. Anthocyanins were not observed in leaves infiltrated with AOMT3 alone (Supplementary Fig. S6C), but when infiltrated with AOMT3 and AN1, the amounts of DR were greatly reduced, and two major peaks appeared of m/z 625 and 639 (M+H) + that were identified as the methylated anthocyanins petunidin 3-rutinoside (PtR) and malvidin 3-rutinoside (MR) (Supplementary Fig. S6D, G, H).
From the above data, it was clear that the metabolite changes were predominantly due to AN1 infiltration, so subsequent experiments did not use combinatorial infiltrations. Changes in sucrose and glucose levels, PAL activity, and the expression of several key phenylpropanoid genes, transcription factors, and sugar metabolism genes were estimated in leaves infiltrated with empty vector, bHLH1, WD40, or AN1 (Fig. 8F-H and Supplementary Fig. S7 at JXB online). In tobacco, four PAL isoforms have been reported, of which PAL1 and PAL4, and PAL2 and PAL3 have high similarity (Reichert et al., 2009). Therefore, PAL1 and PAL4 were amplified together with only one primer set, as were PAL2 and PAL3. Expression of both PAL1/4 and PAL2/3 increased, along with the late pathway genes CHS, F3H, and DFR in AN1-infiltrated samples (Fig. 8G). PAL enzyme activity increased more than 25-fold in AN1-infiltrated samples ( Supplementary Fig. S7). Notably, the expression of NtbHLH was increased in AN1-infiltrated samples. In leaves infiltrated with StbHLH1, the expression of bHLH1 increased, probably because NtbHLH primers amplified StbHLH1, which has a high sequence similarity.
Strikingly, infiltration with AN1 induced a marked change in sugar metabolism, as seen by the stimulation of SUSY1, SUSY4, and INV1 expression (Fig. 8H) and the sizeable decrease in sucrose and glucose concentrations (Fig. 8F).

Promoter elements
The regulatory elements in the promoters of selected potato genes were predicted in the 1500 bp sequence upstream of ATG using the PLACE database (Higo et al., 1999). We were not able to retrieve promoter sequences for UFGT and AOMT3. The interaction of transcription factors and sugars was of particular interest; therefore, MYB, bHLH, and SURE elements were examined. The occurrence of the three MYB recognition sites, MYBCORE (CNGTTR; Solanol et al., 1995), MYBPLANT (MACCWAMC; Sablowski et al., 1994), and MYBPZM (CCWACC; Grotewold et al., 1994); one bHLH recognition site (G-box) MYCCONSENSUS (CANNTG; Blackwell and Weintraub, 1990;Hartmann et al., 2005); and one SURE (AATAGAAAA; Grierson et al., 1994) for different gene promoters is shown in Supplementary Table S5 at JXB online. MYBCORE was present in all the promoters except INV1 and was over-represented in WD40 and SUSY1. MYCCONSENSUS was the most abundantly represented element and was present multiple times (two to nine) in all the promoters except HQT. SURE was present only in AN1, WD40, SUSY1, HCT, and HQT (Supplementary Table S5).

Discussion
Relatively few studies have focused on the regulation of phenylpropanoid metabolism in a tuber crop. To identify candidate transcription factors that regulate the phenylpropanoid pathway, metabolite and gene expression profiles were determined in tubers of five genotypes with decidedly different phenylpropanoid profiles (Fig. 3B). Field-grown tubers were used so results would relate to the crop and avoid the potentially non-representative results noted previously in studies using greenhouse-grown potatoes (Chawla et al., 2012;Navarre et al., 2013). Peeled samples were used to more clearly differentiate the samples by ensuring that all samples represented distinct flesh colours and not a mix such as red skin and white flesh. Moreover, phenylpropanoid metabolism differs between tuber skin and flesh, and less is known about the flesh (Jung et al., 2005) than periderm, in which genetic analysis has revealed three major loci controlling pigmentation, D, R, and P (De .

AN1, bHLH1, and WD40 regulate tuber anthocyanins and other phenylpropanoids
Red and purple potatoes had higher expression of phenylpropanoid structural genes and contained higher amounts of phenylpropanoids, not just anthocyanins but colourless compounds such as 5CGA (Figs 3B and 4A, B). Compared with concentrations reported previously in transgenic tubers overexpressing the R2R3 MYB StMTF1 (Rommens et al., 2008), the wild-type red and purple potatoes in this study contained higher levels of phenylpropanoids, including almost 4-fold and 2-fold higher amounts of 5CGA and Kmp, respectively. Analysis of the potential involvement of 15 transcription factors was measured by qRT-PCR. AN1 was the most abundantly expressed transcription factor and was >100-fold higher in red and purple potatoes than in white. Besides AN1, only two of the other examined transcription factors, bHLH1 and WD40, showed a positive correlation with phenylpropanoid concentrations ( Supplementary Fig. S8 at JXB online). Over 30 phenylpropanoid-related compounds were measured, of which most were only minor components of total soluble tuber phenylpropanoids in red and purple potatoes (Fig. 4). Notably 5CGA and a single anthocyanin accounted for 80-90% of the total. Amounts of CGA and total phenylpropanoids strongly correlated with AN1 expression in genotypes with different phenylpropanoid profiles, and in response to environmental signals, sucrose feeding, and AN1 infiltration. These results suggested that AN1 can mediate marked changes in phenylpropanoids outside the anthocyanin pathway, as chlorogenic acid is synthesized by a different branch of the pathway from anthocyanins. Consistent with this finding is the fact that 5CGA is found in much higher amounts in red and purple potatoes than in white or yellow (André et al., 2007b;Navarre et al., 2011).
To our knowledge, this is the first report of a potato WD40 whose expression correlates with phenolics and anthocyanins ( Supplementary Fig. S8). WD40 was not able to induce phenylpropanoid expression on its own in infiltration studies (Fig. 8A), consistent with previous results where overexpression of WD40 from Medicago truncatula (MtWD40) in hairy roots failed to induce flavonols, proanthocyanidins, or anthocyanins (Pang et al., 2009). In contrast, overexpression of MtWD40 was able to complement the lack of red-pigmented phenotype of a mutant line NF0977, which suggests that WD40 is required but not sufficient for anthocyanin pathway activation (Pang et al., 2009). However, ectopic expression of a grape WD40 induced anthocyanins in Arabidopsis leaves (Matus et al., 2010). These results illustrate the complex interactions among the transcription factors. bHLH1 but not bHLH2 showed a strong association with phenylpropanoid expression. A quantitative trait locus study linked a potato bHLH to anthocyanin synthesis and reported that it was expressed in all the coloured genotypes and in 21 of 53 white or yellow genotypes, suggesting that it is required but not sufficient for anthocyanin synthesis (Zhang et al., 2009b). While we were unable to determine if these bHLHs are the same because sequence information was not available for the previous bHLH, both are localized on chromosome 9. The expression patterns of AN1, bHLH1, and WD40 suggest that they are determinants of the amounts of phenylpropanoids that a given genotype will contain. AN1, bHLH1, and WD40 were also implicated in the intra-genotypical variation that occurs in potatoes in response to environmental variations, such as when the same purple genotype was grown in Alaska, Texas, and Florida ( Supplementary Fig. S4). The higher expression in the Alaskan-grown potatoes might be due to the lower temperatures or longer day length, and potentially reflects greater stress.
Unlike a previous microarray study that reported an ~40fold increase in MYB73 in the purple flesh of a sectored potato that had both white and purple regions (Stushnoff et al., 2010), elevated expression of MYB73 was not observed in the red or purple genotypes relative to the white or yellow potatoes (Fig. 5C). MYB12, which regulates the flavonol branch of the pathway in Arabidopsis (Mehrtens et al., 2005), appeared to be negatively associated with tuber phenylpropanoid content. The minimal repression domain (TLLLFR) present at the C terminus of AtMYBL2, a negative regulator of anthocyanin biosynthesis in Arabidopsis (Matsui et al., 2008), was not found in StMYB12B or in other MYB transcription factors in this study. Nor was the ERF-associated amphiphilic repression (EAR) motif, associated with repression of anthocyanin gene transcription, present in any in these potato MYBs (Ohta et al., 2001;Lin-Wang et al., 2011). Expression of other transcription factors varied among the genotypes but did not seem to be associated with phenylpropanoid content. Among the potato MYBs examined, StMTF1 has the highest percentage similarity with AtPAP1 (Supplementary Table S4); however, the StMTF1 transcript was not present at detectable levels in any of the genotypes examined, suggesting that the native gene does not have a major role in tuber phenylpropanoid metabolism. A previous study showing a stimulatory effect of StMTF1 (Rommens et al., 2008) may be due to its being expressed under a heterologous promoter. Potentially StMTF1 under its own promoter is active in other tissues or conditions. Importantly, the stimulatory effects of AN1 on phenylpropanoid metabolism were not limited to anthocyanins because other compounds were also upregulated, notably 5CGA, the predominant soluble phenylpropanoid in potatoes (Fig. 8B). The increase in expression of CHS and F3H (Fig. 8G) along with a 30-fold increase in Rut (Fig. 8D) suggested that StAN1 upregulates 3´-hydroxylated flavonones. Collectively, the relationship between StAN1 expression and high pelargonidin and petunidin concentrations in coloured potatoes and the induction of cyanidin or delphinidin derivatives in StAN1infiltrated N. tabacum or N. benthamiana leaves suggest that StAN1 regulates flavanone 3-hydroxylase (F3H), flavonoid 3-hydroxylase (F3´H), and flavonoid 3´,5´-hydroxylase (F3´5´H).
Differences were seen in aromatic amino acid pools among the samples. No obvious relationship was observed between these amino acids and phenylpropanoids in the field-grown tubers, whereas sucrose feeding lowered aromatic amino acid concentrations but increased phenylpropanoids, AN1, bHLH1, and WD40 expression. However, AN1 infiltration decreased sucrose and glucose but increased the amounts of aromatic amino acids. The increase was particularly clear with tyrosine and not quite so dramatic with phenylalanine and tryptophan. Thus, sucrose feeding and AN1 infiltration both increased phenylpropanoids, whereas sucrose treatments decreased aromatic amino acid amounts but AN1 increased them. Understanding the basis for these differences will require further study.

AOMT
To our knowledge, a gene for AOMT has not been reported previously in potatoes or other solanaceous plants. AOMT3 but not AOMT1/2 showed a strong correlation with anthocyanins ( Supplementary Fig. S8) and altered anthocyanin profiles when infiltrated into leaves. AOMT3 probably plays a key role in determining the type of anthocyanins that accumulate in potatoes, and its identification offers another potential target for efforts to manipulate tuber anthocyanin composition. The failure of N. benthamiana leaves to synthesize methylated anthocyanins when infiltrated with StAN1 suggests that either AN1 does not regulate AOMT or that tobacco leaves lack a functional AOMT. BLAST searches for an AOMT3 homologue in tobacco did not identify any candidates.

Sucrose induces AN1 and phenylpropanoid biosynthesis
In addition to providing carbon for phenylpropanoid metabolism, sugars also regulate anthocyanin biosynthesis (Teng et al., 2005). During grape berry development, the increase in sugars modulates expression of anthocyanin biosynthetic genes (Boss et al., 1996). High-phenylpropanoid red and purple potatoes had substantially higher amounts of sucrose and glucose than white or yellow tubers (Fig. 6). Similar relationships were seen in a tuber developmental study in which sugar and phenylpropanoid concentrations were correlated (Navarre et al., 2013). To gain direct evidence, potato plantlets were cultured on medium supplemented with 0 or 120 mM sucrose, resulting in large increases in phenylpropanoids (Fig. 7A). Sucrose stimulated significant increases in the expression of AN1, bHLH1, and WD40, demonstrating that the stimulatory effect of sucrose on potato phenylpropanoid metabolism is at least partly modulated through these transcription factors. Supporting this finding is the higher expression of AN1, bHLH1, and WD40 in the field-grown potatoes that had highest sucrose and phenylpropanoid concentrations.
The presence of SURE elements in the promoter of AN1 ( Supplementary Fig. S3b) is consistent with its regulation by sucrose. In addition to greater amounts of sucrose, purple and red potatoes contained six SURE elements in the AN1 promoter, whereas white and yellow potatoes had a single SURE element. A mechanism for the 5-fold increase in CGA (Fig. 7A) in response to sucrose is suggested by the MYB, bHLH, and SURE elements present in the promoter of HQT. SURE elements were not observed in the bHLH1 promoter, but bHLH1 expression was upregulated by sucrose treatment. A possible explanation is that the potato AN1 induced expression of tobacco bHLH in transient assays (Fig. 8G), which is consistent with a previous report of tobacco transcription factors (Bai et al., 2011). AN1-infiltrated leaves showed a 24 h delay in anthocyanin formation compared with AN1+bHLH1 co-infiltrated leaves ( Supplementary Fig.  S5), and the delay may reflect the time needed for AN1 to recruit bHLH. The promoters of potato PAL, CHS, F3H, and DFR did not have SURE elements, but had MYB and bHLH regulatory elements (Supplementary Table S5), suggesting that the increased expression after sucrose feeding was due to AN1 and bHLH1. Anthocyanin levels in potato correlated with the expression of ANS and UFGT (Hu et al., 2011;Keifenheim et al., 2006) and higher expression of these genes was observed in this study in the high-phenylpropanoid potatoes and sucrose-treated potato plantlets (Figs 5B and 7F).
Interestingly, sucrose treatments of potato plantlets induced large increases in AN1 and bHLH1 expression, and to a lesser extent that of WD40 (Fig. 7G), whereas infiltration of potato AN1 into tobacco leaves increased expression of SUSY and INV genes (Fig. 8H). This suggests the possibility of a regulatory loop in which sucrose increases AN1 expression but AN1 decreases sucrose concentrations by inducing sucrolytic enzymes that liberate hexoses that are channelled to the phenylpropanoid pathway (Fig. 9). In contrast, bHLH1 and WD40 infiltration alone did not increase expression of SUSY or INV or mobilize sucrose. The extent that sucrose-AN1 interactions modulate tuber phenylpropanoid metabolism is an interesting question that awaits future research.
In addition to their important in planta roles, phenylpropanoids are desirable in the diet because of their health-promoting properties, which include antioxidant, anti-inflammatory, hypotensive, and chemopreventative effects (Manach et al., 2004;Prior, 2003;Kaspar et al., 2011;Vinson et al., 2012). High-phenylpropanoid potatoes would be especially valuable because of the high consumption of this staple food. The potential of potatoes to provide dietary phenylpropanoids, including anthocyanins, is significant. For example, the purple potatoes in this study contained anthocyanin amounts that approach those of high-anthocyanin transgenic tomatoes , and exceed amounts in many fruits Fig. 9. Proposed model for the interaction of AN1, sucrose, and sucrose metabolic genes to modulate phenylpropanoid metabolism. Sucrose stimulates AN1 expression, which with bHLH1 and WD40 regulates phenylpropanoid expression, but also sucrolytic genes that can provide carbon for induced phenylpropanoid metabolism and decrease sucrose. The red line indicates the cleavage of sucrose. Solid green lines indicate targets stimulated by AN1. The dashed green arrow shows AN1 expression stimulated by sucrose concentrations, which can decrease in response to AN1 expression. and vegetables (see Table 2 in Wu et al., 2006). Collectively, these data suggest that interactions among sucrose, sucrolytic enzymes, and AN1 modulate the pathway, and that AN1 is a key regulator of the most abundant tuber phenylpropanoids. Increased understanding of tuber phenylpropanoid metabolism will facilitate efforts to develop potatoes with optimal types and concentrations of phenylpropanoids.