Anthocyanins are a major group of pigments involved in a large range of plant functions. They are also key to the consumer appeal of many ornamental, fruit and vegetable products, and are of increasing interest to researchers for their possible dietary benefits to human health (Gould et al., 2008; Martin et al., 2011; Davies and Espley, 2013). Their biosynthesis, as part of the larger phenylpropanoid pathway, is well characterized at both the biochemical and molecular level (Grotewold, 2006). In addition to DNA sequences being available for most of the biosynthetic enzymes, sequences have also been identified from several species for proteins that regulate the biosynthetic gene transcription (Grotewold, 2006; Feller et al., 2011; Hichri et al., 2011; Davies et al., 2012). Of particular note are R2R3-MYB and basic helix-loop-helix (bHLH) type transcription factors (TFs) that form a complex with a WD-Repeat protein (the “MBW” complex) to activate the biosynthetic genes. This co-action of R2R3-MYB and bHLH proteins in regulating anthocyanin biosynthesis has been consistently found to occur in a range of monocot and eudicot species. Based on mutant analyses, in eudicots, the anthocyanin-related TFs are principally involved in activating a subset of the anthocyanin biosynthetic genes, the late biosynthetic genes (LBGs), with other R2R3MYBs being the principal regulators of the early biosynthetic genes (EBGs; Grotewold, 2006; Feller et al., 2011).

The availability of anthocyanin-related TF genes has allowed their use to modify the amount and spatial distribution of anthocyanins produced in transgenic plants and cell lines of several species (Schwinn and Davies, 2006; Hichri et al., 2011; Dixon et al., 2013). Indeed, TF genes have increasing become the focus of biotechnology approaches to altering plant metabolism, as they can coordinately regulate several biosynthetic genes, overcoming the need to identify a rate-limiting step or introduce multiple transgenes for biosynthetic enzymes. Despite the successes with anthocyanin TF transgenes, there are also cases where no phenotypic change has been induced in the transgenic plants (Bradley et al., 1999), or where the pigmented phenotype is only observed under stress conditions of high-light or cold (Goldsbrough et al., 1996; Ray et al., 2003; Albert et al., 2009; Rowan et al., 2009). This suggests environmental induced changes in the action of endogenous regulatory partners of the TF transgene product are required for effective activation of the anthocyanin pathway. Moreover, some flavonoid related R2R3-MYB transgenes have induced pleiotropic changes in the transgenic plants. For example, tomato plants over-expressing the grapevine VvMYB5a and VvMYB5b had not only modified flavonoid production but also dwarfism and changes in chlorophyll and β-carotene production (Mahjoub et al., 2009).

Petunia is an established model species for understanding the regulation of anthocyanin production and its modification in transgenic plants. Petunia has specific vegetative and floral anthocyanin pigmentation patterns that are determined by the expression patterns of individual R2R3-MYB and bHLH activators (Quattrocchio et al., 1998, 1999; Albert et al., 2011, 2014) and R3-MYB and R2R3-MYBs with a repressive action on anthocyanin biosynthesis (Albert et al., 2014). Over-expression of the maize bHLH Leaf color (35S:LC) in the petunia line Mitchell (MP, which contains a mutation in the main petal R2R3-MYB activator, an2) enhanced light-induced vegetative and floral pigmentation phenotypes (Albert et al., 2011). In particular high anthocyanin levels, producing dark colored leaves, were generated in 35S:LC MP plants grown under high-light conditions in growth chambers. However, over-expression of two endogenous R2R3-MYB activators, DEEP PURPLE (DPL) and PURPLE HAZE (PHZ), generated MP lines that accumulated high levels of anthocyanins in their leaves under standard greenhouse conditions, without the high-light requirement of 35S:LC plants. Albert et al. (2014) showed that the expression of the LC transgene did not change with light conditions, but transcripts for DPL and PHZ increased in abundance while those for the R2R3-MYB repressor MYB27 decreased upon high-light treatment. These TF transgenic lines have proven very useful in elucidating the interactions of the MBW components in the regulation of anthocyanin biosynthesis (Albert et al., 2014).

In Antirrhinum majus (antirrhinum), the Rosea1 (ROS1), Rosea2, and Venosa genes encode R2R3-MYB TFs that regulate anthocyanin pigmentation of the flowers, in conjunction with the bHLH factors DELILA and MUTABILIS (Goodrich et al., 1992; Schwinn et al., 2006; Shang et al., 2011). ROS1 is required for wild-type levels of pigmentation in the flowers and vegetative parts, while Rosea2 and Venosa determine specific pigmentation patterns, for example venation (Schwinn et al., 2006; Shang et al., 2011).

In the study reported here the effect of introducing a transgene for production of ROS1 into the heterologous ornamental species petunia and lisianthus (Eustoma grandiflorum) has been examined. These species belong to two Asterid orders, the Solanales (petunia) and Gentianales (lisianthus), that contain many economically important ornamental species and that are generally placed next to each other in phylogenetic trees. While many studies have been conducted on the effects of anthocyanin-related TF transgenes in Solanales species, studies are lacking for any Gentianales species. The plants studied include R2R3-MYB/bHLH double transgenic petunias, generated by crossing the 35S:ROS1 lines with the 35S:LC lines characterized by Albert et al. (2009, 2011, 2014). The influence of the environment on the phenotypes of the 35S:LC, 35S:ROS1, and 35S:LC/35S:ROS1 petunia lines was assessed by comparison of plants in an outdoor field trial to those in standard greenhouse conditions. As flavonoids have multiple roles in plants, including influences on auxin transport (Peer et al., 2011; Cheynier et al., 2013), changes in the pathway activity may alter general plant performance, as observed in tomato (Mahjoub et al., 2009). Moreover, anthocyanin-based vegetative pigmentation has been postulated to help to improve the general stress tolerance of leaves (Gould et al., 2008), and petunia plants with increased anthocyanin levels in leaves showed increased light saturation and light compensation points without reducing the maximal photosynthetic assimilation rate (Albert et al., 2009). Therefore, the transgenics were assessed not only for the changes in anthocyanin production, but also for seed germination rate, plant stature, and plant survival under field conditions.

Introduction of the 35S:ROS1 transgene into either lisianthus or MP resulted in an increase in the anthocyanin production of the transgenic plants, demonstrating that expression of heterologous anthocyanin-related R2R3-MYBs can affect spatial and temporal pigmentation patterns in these species. However, as observed with flavonoid-related R2R3-MYB or bHLH transgenes in other species, the phenotypes resulting varied markedly depending on plant species, organ, developmental age, or environmental conditions. In lisianthus, the 35S:ROS1 transgene induced additional pigmentation in the sepals and petals, but had no apparent affect on anthocyanin biosynthesis in callus or plantlets in culture, mature petals or vegetative parts. Both the spatial and temporal accumulation of anthocyanins was altered in sepals and petals. In petals, precocious pigmentation was induced. This novel pigmentation was confined to the lower half of the petals, while in non-transgenics the first pigmentation to appear is at the petal tips. The fact that most lisianthus cultivars have developmentally late petal pigmentation, with the buds being colorless, is considered a significant disadvantage for their use as cut flowers and ornamentals, and breeding programs have previously tried to introduce an early coloration trait to a wider variety of lisianthus germplasm (Oren-Shamir et al., 1999). The additional pigmentation in sepals was also predominately at the sepal base. A basal pattern of anthocyanin accumulation is common in the floral organs of many species, and in at least some cases is under the control of R2R3-MYB genes (Albert et al., 2011; Cooley et al., 2011). The lack of change in mature petal or vegetative phenotypes was not due to the expression pattern of the 35S:ROS1 transgene, as transcript was detected in a range of flower stages and in leaves (Figure 2).

The novel phenotypes observed with the 35S:ROS1 transgene are unlikely to be due to the transformation event itself, as opposed to the action of the ROS1 gene product. Mitchell petunia is a commonly used line for plant transformation studies, and the changes in phenotype observed here have not been observed when using a wide range of other transformation vectors (e.g., Deroles and Gardner, 1988; Davies et al., 2003; Lewis et al., 2006). The same applies for lisianthus, as phenotypes similar to those found here have not been observed when transgenics of the same cultivars have been generated with either empty vector controls or transformation vectors for various flavonoid biosynthetic genes (Ledger et al., 1997; Schwinn et al., 1997; Deroles et al., 1998; Nielsen et al., 2002).

No information is available on the endogenous regulators of the anthocyanin pathway in lisianthus, but the ability of ROS1 to induce anthocyanin biosynthetic gene expression suggests that, like other species studied to date, R2R3-MYB and bHLH factors control anthocyanin pigmentation in this species. The strong induction of the LBG ANS, and the association of this induction with enhanced pigmentation, supports the previous results from gene expression analysis that anthocyanin pigmentation in lisianthus flowers is normally controlled through the regulation of the LBGs (Davies et al., 1993; Oren-Shamir et al., 1999; Noda et al., 2004). The increased transcript levels for the EBG CHS in the 35S:ROS1 transgenics shows that CHS is also responsive, as a direct or indirect target, to R2R3-MYB factors. The EBGs are commonly principally activated by R2R3-MYB factors that do not require a MBW complex (Feller et al., 2011; Cheynier et al., 2013), however, it is possible they can also be regulated by the anthocyanin-related MBW complex. Based on the color of the precociously pigmented flowers of ROS1-lisianthus, cyanidin-derived anthocyanins accumulated at this stage, not the delphinidin-derived anthocyanins found in the mature petals. This indicates that the genes encoding the flavonoid 3′5′ hydroxylase (F3′5′H) were not up-regulated by the 35S:ROS1 transgene, and matches the pigment types seen in line 54 transgenics containing an antisense flavonol synthase transgene that redirects flavonoid biosynthesis from flavonols to anthocyanins (Nielsen et al., 2002). Variability in the activation of F3′5′H genes by R2R3-MYB/bHLH transgenes has been found previously, with ROS1/DELILA but not LC/C1 activating the F3′5′H of tomato (Butelli et al., 2008). Although not effective in lisianthus or tomato, 35S:LC was able to greatly increased transcript abundance for F3′5′H in MP transgenics (Bradley et al., 1998).

The R2R3-MYBs DPL and PHZ have previously been show to mediate light-activation of anthocyanin biosynthesis in flowers and vegetative tissues of petunia (Albert et al., 2009, 2011). Over-expression of DPL or PHZ in MP resulted in strong anthocyanin pigmentation throughout leaves and stems, without the requirement for high-light levels (Albert et al., 2014). However, this was not the case in the 35S:ROS1 MP. The main novel phenotypes in 35S:ROS1 MP were the surface pigmentation of some callus clumps and an increase in pigmentation of the outer layer(s) of stem tissue. Given that these are the areas exposed to light, it would be interesting to test whether high-light treatment of the transgenics would result in stronger phenotypes. In the case of GMYB10-containing tobacco lines, a significant pigmentation increase only occurred when plants were under high-light conditions (Elomaa et al., 2003). The role of light in regulating anthocyanin production in lisianthus is not known. However, the precocious anthocyanin biosynthesis induced in flowers of 35S:ROS1 lisianthus was much weaker in regions of the petal that were covered by the sepals (Figure 1). This indicates that light may promote anthocyanin production in the petals, and that ROS1 over-expression was not able to overcome this effect. Collectively, the limited and very specific changes in pigmentation observed in 35S:ROS1 lisianthus suggest that anthocyanin biosynthesis in this species is under tight temporal and spatial control, and that even in the transgenics the regulation of the pathway remains primarily under the control of endogenous factors rather than being control by ROS1. For both MP and lisianthus, the specificity of pigmentation observed is unlikely to be due to specific expression patterns of the transgene (Figure 2).

In MP, ROS1 induced pigmentation through the regulation of both EBGs (CHS) and the LBGs (ANS), with a much more marked change seen for ANS transcript levels compared to CHS (Figure 5). This is in line with the function of ROS1 in antirrhinum, in which it is required for LBG activation and the pigmentation of stem and petal tissue (Schwinn et al., 2006). Although both CHS and ANS are expressed at significant levels in the petals of the MP transgenics (Figure 5) the acyanic petal (due to an2) and anther (due to an4) phenotypes were not complemented. It was surprising ROS1 was unable to complement the an2 and an4 mutations in MP, since these two loci encode homologous anthocyanin-related R2R3-MYB factors (Quattrocchio et al., 1998, 1999; Schwinn et al., 2006). However, 35S:AN2 expression in MP (W115) also failed to fully complement the an2 petal phenotype, producing a stem pigmentation phenotype similar to that of 35S:ROS1 but only a weak pigmentation phenotype in petals (Quattrocchio et al., 1999). Neither was full petal pigmentation restored in the DPL and PHZ transgenics (Albert et al., 2011). When the grapevine R2R3-MYBs VvMYBA1, VvMYB5a, and VvMYB5b were over-expressed in MP, VvMYB5b transgenics had some increase in petal anthocyanin production, but only VvMYBA1 showed significant complementation of the an2 phenotype (Cavallini et al., 2014). In contrast, both the 35S:AN2 and 35S:VvMYBA1 fully complement an2 in a different genetic background than MP (W115 × W59; Quattrocchio et al., 1999; Cavallini et al., 2014). Ectopic expression of AN2, DPL or PHZ also complemented the phenotype of the an4 mutation in MP and produced pigmented anthers (Quattrocchio et al., 1998; Albert et al., 2011). Neither over-expression of ROS1 or LC alone resulted in pigmented anthers, but the 35S:ROS1 × 35S:LC plants did have pigmented anthers, suggesting that alone, ROS1 and LC may not be able to function effectively with all the endogenous regulatory partners but that together they can activate additional biosynthetic gene promoters. Similarly, transient expression of 35S:ROS1 into 35S:LC MP leaf pieces showed induction of vegetative anthocyanin production (Albert et al., 2009). However, both ROS1 and LC must be able to interact with some endogenous MBW partners, as they do generate some increased pigmentation and analysis of mutant lines of many species has shown the requirement of the full MBW complex for activity.

The results show the large variations produced by both the genetic background of the host and the specific TF transgene used. Often transgenes for single TFs can enhance pigmentation in tissues already producing some pigment, while combinations of TF transgenes may result in de novo pigmentation. There is unlikely to be a single explanation to account for the variant patterns produced by the different TF transgenes in different species. It may be related to the availability of adequate precursors, the characteristics of the specific TF introduced, the presence or absence of interacting endogenous TFs, or characteristics of the endogenous TFs. The differences seen for the TF could relate to both varying promoter recognition of specific biosynthetic genes (e.g., F3′5′H, Butelli et al., 2008) and differences in hierarchical regulation of the TFs in different species (Albert et al., 2014). Interestingly, both LC (bHLH) and ROS1 (MYB) can induce pigmentation in leaves of MP (Figures 4 and 6), and both in the same sub-epidermal cells, suggesting that simple models based on a specific MYB and/or bHLH being limiting are unlikely to apply. The more recent findings of the involvement of repressor TFs in the MBW complex may help in explaining such results (Yuan et al., 2013; Albert et al., 2014).

The transgenic petunia plant lines generally showed no difference in performance in the greenhouse or field, as measured by seed germination, plant survival or plant stature. The exception was one backcross line of 35S:LC to MP, which had slightly reduced plant survival in the field. However, as this was not found for any other 35S:LC lines, including inbreed lines, this is thought to be due to issues with that specific cross or field variation effects. In general, however, no pleiotropic affects such as those found with 35S:VvMYB5 (Mahjoub et al., 2009) were observed from the 35S:LC or 35S:ROS1 transgenes. There was also no observed gain in plant performance, as measured by the general character of plant stature. It has previously been shown that the 35S:LC MP plants have increased light saturation and light compensation points (Albert et al., 2009), but this did not have a significant effect under the summer field conditions tested. Replication of the trial at different locations would be desirable to assess whether the lack of difference in overall plant performance between non-transgenic and transgenic lines was consistent under different climatic conditions. The clear phenotypic difference between the plants transferred to the field and those retained in the greenhouse was the induction of anthocyanin pigmentation in the foliage within 4 days of planting. The main environmental differences in the field compared to the greenhouse were higher light (including UV) levels and cooler night time temperatures (as the greenhouse was heated to maintain a minimum temperature of 15∘C). Albert et al. (2009) showed that exposing 35S:LC MP plants to light levels above those normally found in a greenhouse could strongly promote anthocyanin production in the foliage, so that is likely the main environmental factor responsible for the anthocyanin induction in the field. However, temperatures below 10∘C have also been shown to promote anthocyanin biosynthesis in some species (Gould et al., 2008), so it is possible that the more variable field temperatures could also contribute to increasing pigmentation.

Overall, the results support the potential application of anthocyanin-related TF transgenes for generating novel phenotypes in ornamental crop species. However, the results also highlight the requirement to identify specific TFs or recipient genetic backgrounds that will enable the generation of strong pigment phenotypes, and the need to conduct both greenhouse and field tests of the new plant germplasm.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.