•Floral scent is a complex trait of biological and applied significance. To evaluate whether scent production originating from diverse metabolic pathways (e.g. phenylpropanoids and isoprenoids) can be affected by transcriptional regulators, Arabidopsis PRODUCTION OF ANTHOCYANIN PIGMENT1 (PAP1) transcription factor was introduced into Rosa hybrida.
•Color and scent profiles of PAP1-transgenic and control (β-glucuronidase-expressing) rose flowers and the expression of key genes involved in the production of secondary metabolites were analyzed. To evaluate the significance of the scent modification, olfactory trials were conducted with both humans and honeybees.
•In addition to increased levels of phenylpropanoid-derived color and scent compounds when compared with control flowers, PAP1-transgenic rose lines also emitted up to 6.5 times higher levels of terpenoid scent compounds. Olfactory assay revealed that bees and humans could discriminate between the floral scents of PAP1-transgenic and control flowers.
•The increase in volatile production in PAP1 transgenes was not caused solely by transcriptional activation of their respective biosynthetic genes, but probably also resulted from enhanced metabolic flux in both the phenylpropanoid and isoprenoid pathways. The mechanism(s) governing the interactions in these metabolic pathways that are responsible for the production of specialized metabolites remains to be elucidated.
Roses are among the world’s most important ornamentals, including 200 species and > 18 000 cultivars used as cut flowers, garden and pot plants (Gudin, 2000). Modern rose varieties possess varied color and scent profiles deriving from their diverse progenitors (Schulz, 2003; Verhoeven & Brandenburg, 2003). Flavonoids, carotenoids or both determine the color of rose flowers, which is dependent on the variety: yellow and orange hues are commonly contributed by carotenoids; other shades of red, pink, violet and mauve are generally contributed by flavonoids (Schulz, 2003). The main flavonoids contributing to rose petal color are the anthocyanins cyanidin 3,5-diglucoside, pelargonidin 3,5-glucoside and paeonidin 3,5-diglucoside (Ogata et al., 2005). Co-pigments, such as flavonol glycosides, also greatly influence the flower’s final color (Jay et al., 2003). The molecular basis for the transcriptional regulation of flavonoid biosynthesis is well established, and has been shown to operate through complexes of transcription factors. These complexes include basic helix–loop–helix (bHLH) transcription factors which interact with R2R3 MYB proteins, and WD40 proteins required for the activity of the regulatory complex (Koes et al., 2005; Hichri et al., 2011). PRODUCTION OF ANTHOCYANIN PIGMENT1 (PAP1) is a Myb transcription factor from Arabidopsis thaliana which has been shown previously to exert broad activation of the phenylpropanoid pathway (Borevitz et al., 2000; Mathews et al., 2003; Matousek et al., 2006; Ben Zvi et al., 2008a; Li et al., 2010).
Analyses of floral scent in numerous rose varieties have led to the identification of over 400 volatile compounds that derive from terpenoid, phenylpropanoid/benzenoid and fatty acid pathways (Guterman et al., 2002; Scalliet et al., 2008). On the basis of their chemical structure, these volatile compounds can be classified into five groups: hydrocarbons, alcohols, esters, aromatic ethers and others (including aldehydes, aliphatic chains, rose oxides and norisoprenes) (Verhoeven & Brandenburg, 2003). In nature, the unique combination of scent molecules emitted by flowers is detected by the olfactory receptors of insects, enabling them to find and visit their flower(s) of choice (Shalit et al., 2004; Smith et al., 2006). Moreover, the diurnal pattern of floral volatile emission from wild species of the genus Rosa is synchronized with the active hours of its bee pollinator (Hendel-Rahmanim et al., 2007).
The characterization of the pattern of scent production in rose has revealed the existence of numerous levels of complexity in the regulation of floral volatile production (Hendel-Rahmanim et al., 2007). Attempts to characterize rose’s genetic machinery of scent production have led to the identification and characterization of several genes coding for enzymes responsible for the production of floral scent compounds, among them GERMACRENE D SYNTHASE (GDS), CAROTENOID CLEAVAGE DIOXYGENASE1 (CCD1), ORCINOL O-METHYLTRANSFERASE1 (OOMT1), GERANIOL/CITRONELLOL ACETYL TRANSFERASE1 (AAT1) and PHENYLACETALDEHYDE SYNTHASE (PAAS), which catalyze the production of germacrene D, β-ionone, orcinol methyl ether, geranyl acetate and phenyl acetaldehyde, respectively (Guterman et al., 2002; Lavid et al., 2002; Shalit et al., 2003; Farhi et al., 2009; Huang et al., 2009). In addition to structural scent-related genes, several Myb transcription factors putatively involved in the regulation of floral scent production have been identified in roses (Guterman et al., 2002; Yan et al., 2010).
High commercial value is a major driving force for the breeding of modern rose cultivars (Chandler & Lu, 2005). Genetic engineering approaches have been implemented in rose for the alteration of traits towards, for example, resistance to pathogens, improved morphological traits and color modification (Chandler & Lu, 2005; Tanaka et al., 2005; Chandler & Tanaka, 2007; Katsumoto et al., 2007). Although fragrance is still among the traits most identified with rose species, it is almost absent in many modern cut flower varieties (Chandler & Lu, 2005; Chandler & Tanaka, 2007). Only c. 20% of all rose species known today are classified as ‘fragrant’ and 30% are not fragrant at all (Schulz, 2003). Recently, progress has been made towards establishing a marker-assisted breeding platform for scented rose varieties, and several loci associated with specific scent patterns in roses have been mapped (Spiller et al., 2010).
Rosa hybrida‘Pariser Charme’ has dark red flowers as a result of anthocyanin accumulation (Yokoi, 1974), which produce several volatile compounds deriving from diverse biosynthetic origins (Schulz, 2003). The availability of an embryogenic callus-based genetic transformation system (Dohm et al., 2001) developed for this cultivar enabled us to introduce PAP1 into ‘Pariser Charme’ rose plants to investigate the extent of the effect of PAP1 on the production of floral volatiles from diverse biochemical origins. Remarkably, the expression of PAP1 in flowers of ‘Pariser Charme’ roses led to increased levels of volatile compounds originating from various biosynthetic origins, for example, phenylpropanoids, monoterpenes, sesquiterpenes and norisoprenes; this increase was not solely dependent on the transcriptional activation of the respective structural genes. The change in the volatile profile of PAP1-transgenic rose was olfactorily significant, and distinguishable by both humans and honeybees.
Materials and Methods
Construction of chimeric genes, plant transformation and regeneration
PAP1 (GenBank accession no. AF325123) coding for the Myb transcription factor was cloned from Arabidopsis‘Colombia’ DNA and inserted into the binary vector pCGN1559, as described in Ben Zvi et al. (2008a). Agrobacterium tumefaciens strain AGLO (Lazo et al., 1991), carrying the binary plasmid pCGN1559 or pCGN7001 (containing the uidA gene coding for β-glucuronidase (GUS) (Comai et al., 1990)), was used for the stable transformation of rose to generate PAP1-transgenic or control, GUS-transgenic plants, respectively. Transformation of embryogenic callus initiated from Rosa hybrida L. cv Pariser Charme (Dohm et al., 2001), selection and regeneration of transgenic plants were conducted as described by Dohm et al. (2001). Callus was cultured in Murashige and Skoog basal medium (Sigma) supplemented with 0.01 mg l−1 indole-3-butyric acid (IBA), 2 mg l−1 6-benzylaminopurine (BAP) and 0.1 mg l−1 gibberellic acid (GA3) for shoot induction, and 0.01 mg l−1 IBA, 0.5 mg l−1 BAP and 0.2 mg l−1 GA3 for shoot multiplication and elongation. Shoot rooting was induced by subculture on half-strength Murashige and Skoog medium supplemented with 0.1 mg l−1 IBA. The regenerated plantlets were planted into a 1 : 1 mixture of standard potting soil and perlite for glasshouse adaptation. Plants were then transferred to 18-cm-diameter containers with potting soil (Shaham, Giva’t Ada, Israel).
Plant growth and plant material
Plants were grown in a glasshouse under 25°C : 22°C day : night temperatures and a natural photoperiod for 2 yr in Rehovot, Israel. Control GUS-expressing plants were intermixed with the PAP1-transgenic lines. Flowers were analyzed at developmental stage 4 (when the sepals are beginning to retract, but the petal whorl is still closed) and stage 7 (the outer petal whorl is open, but the inner petal whorl is still closed and the reproductive organs are not yet visible).
Histochemical assay of GUS expression
GUS activity was analyzed histochemically as described previously (Ben Zvi et al., 2008b). Tissue samples were incubated for a few hours to overnight at 37°C in a 0.1% (w/v) X-Gluc (5-bromo-4-chloro-3-indolyl β-d-glucuronic acid sodium salt; Biosynth Inc., Staad, Switzerland) solution containing 0.1 M sodium phosphate buffer (pH 7.0), 10 mM EDTA and 0.1% (w/v) Triton X-100. When necessary, green tissues were bleached, after staining, by immersion in 50% ethanol for a few hours, followed by several washes with 70% ethanol.
To determine the anthocyanin content, three tissue replicates (each 100 mg fresh weight (FW)) from leaves and stage 4 flowers of control and PAP1-transgenic lines 6, 11, 12 and 13 were extracted in 1 ml of methanol containing 1% (v/v) HCl. Following overnight incubation in the dark at − 4°C with shaking at 150 rpm, the extract was centrifuged at 10 500 g for 10 min. The anthocyanin content in the supernatant was determined on the basis of the absorbance at 530 nm (Ben Zvi et al., 2008a).
RNA extraction and real-time quantitative PCR analysis
Gene expression analyses were conducted on PAP1-transgenic lines 6, 11, 12 and 13 and GUS-transgenic control plants. Petals at stage 4 were collected at 21:00 h, when the genes involved in scent production are optimally expressed (Guterman et al., 2002). Flower and leaf tissues (100 mg) were extracted as described previously (Guterman et al., 2002), and total RNA was treated with RNase-free DNase (Fermentas, Glen Burnie, MD, USA). First-strand cDNA was synthesized using 1 μg of total RNA, oligo d(T) primer and Reverse Transcriptase ImProm-II™ (Promega). Control samples were generated without the addition of reverse transcriptase to the reaction. PCR was performed for 40 cycles (94°C for 15 min and then cycling at 94°C for 10 s, 60°C for 30 s and 72°C for 20 s). To confirm that the analyzed samples were not contaminated with DNA, real-time PCR using ACTIN primers was also conducted with samples generated without reverse transcriptase. The primers (forward and reverse, respectively) used for PHENYLALANINE AMMONIA LYASE (PAL; GenBank accession no. BQ105227) were 5′-GCTAGGGCTGCATACGAGAG-3′ and 5′ GTTCCAACCACTGAGGCAAT-3′; for CHALCONE SYNTHASE (CHSAB038246), 5′-GCGACACCCATCTCGATAGT-3′ and 5′-TCGTTGAGGCTCTTCTCGAT-3′; for CHALCONE ISOMERASE (CHI; AY040321), 5′-GCCAGCAATACTCGGAGAAG-3′ and 5′-AAGCCAATCGTCAATGATCC-3′; for FLAVANONE 3-HYDROXYLASE (F3H; BQ105585), 5′-GCAACGAAATTCCGATCATT-3′ and 5′-GAACTCTCTGGCGAGACTGG-3′; for DIHYDROFLAVONOL-4-REDUCTASE (DFR; D85102), 5′-GCAACGAAATTCCGATCATT-3′ and 5′-GAACTCTCTGGCGAGACTGG-3′; for ANTHOCYANIN SYNTHASE (ANS; BI977949), 5′-GCTCGTCAACAAGGAGAAGG-3′ and 5′-GGTAGAGGCGAGAGCTTCCT-3′; for GERMACRENE D SYNTHASE (GDS; BQ105086), 5′-CTCCGCCAGTTCAAACAAGT-3′ and 5′-TTGTAACCATGCTGCCTCAG-3′; for CAROTENOID CLEAVAGE DIOXYGENASE1 (CCD1; EU327776), 5′-CGAAAATTGAGGTTGGAGGA-3′ and 5′-GCATGGAACCCATATGGAAC-3′; for EUGENOL SYNTHASE (EGS), 5′-CTTTAATGGCGGCGATGAT-3′ and 5′-AACCCGGCCAAGTCTAAAGT-3′; for PHENYLACETALDEHYDE SYNTHASE (PAAS; DQ192639), 5′-CCAGAATTTCGGCATTTCAT-3′ and 5′-TCAAAAACTCCGGATTCGTC-3′; for GERANIOL/CITRONELLOL ACETYL TRANSFERASE1 (AAT1; BQ106456), 5′-CCCACAACGTATTTCCCAGT-3′ and 5′-ATGCCTGCTTCGAAATCATC-3′; for ORCINOL O-METHYLTRANSFERASE1 (OOMT1; AF502433), 5′-CAATCCATCCAACCAAATCC-3′ and 5′ ATCGTTTTGGAACCAAGTGC-3′; for ACTIN (AB239789), 5′-CAATGTGCCCGCTATGTATG-3′ and 5′-CTGTAAGGTCACGTCCAGCA-3′. cDNA amounts were standardized to the expression of two reference genes: ACTIN (AB239789) and α-TUBULIN (AF394915); as these gave similar results, only those standardized to ACTIN are shown. Real-time quantitative PCR was performed in the presence of SYBR Green I dye (ABgene, Rockford, IL, USA) in a Corbett Research Rotor-Gene 6000 cycler, and data analysis was performed using Rotor-Gene 6000 series software 1.7 (QIAGEN Inc., Valencia, CA, USA).
Collection and extraction of volatile compounds
Volatiles were collected from stage 7 flowers (Guterman et al., 2006; Hendel-Rahmanim et al., 2007). For headspace analyses, individual detached rose flowers were enclosed in a 500-ml glass container with appropriate openings. Emitted volatiles were collected for 24 h using an adsorbent trap consisting of a glass tube containing 100 mg Porapak Type Q polymer (80/100 mesh; Alltech, Fresno, CA, USA) and 100 mg 20/40 mesh activated charcoal (Supelco, Bellefonte, PA, USA), held in place with plugs of silanized glass wool (Guterman et al., 2006). Trapped volatiles were eluted using 1.5 ml of hexane, and 2 μg of isobutylbenzene was added to each sample as an internal standard. To determine the pool sizes of volatile compounds in the petals, flowers collected at 21:00 h were weighed, ground in liquid nitrogen and extracted in 10 ml of hexane containing 2 μg of isobutylbenzene as the internal standard. Following overnight incubation with shaking at 150 rpm, the extract was centrifuged at 10 500 g for 10 min and the supernatant was filtered through a 25-ml syringe with a 0.2-μm sterile nylon filter. Samples were evaporated using liquid nitrogen to a final volume of 200 μl before injection into a gas chromatography-mass spectrometry (GC-MS) instrument.
GC-MS analyses of volatile compounds
GC-MS analyses (1 μl sample) of the extracted volatile compounds were performed in an instrument that included a Pal autosampler (CTC Analytic, Zwingen, Switzerland), a TRACE GC 2000 equipped with an Rtx-5SIL MS (Restek; inside diameter, 0.25 μm; 30 m × 0.25 mm; Bellefonte, PA, USA) fused-silica capillary column and a TRACE DSQ quadrupole mass spectrometer (ThermoFinnigan; Hemel Hempstead, Hertfordshire, UK). Helium was used as the carrier gas at a flow rate of 0.9 ml min−1. The injection temperature was set to 250°C (splitless mode), the interface to 280°C and the ion source adjusted to 250°C. The analysis was performed under the following temperature program: 2 min of isothermal heating at 40°C, followed by a 5°C min−1 oven temperature ramp to 250°C and a final increase from 250 to 270°C at 80°C min−1. The transfer line temperature was 250°C. The system was equilibrated for 1 min at 70°C before injection of the next sample. Mass spectra were recorded at three scans per second, with a scanning range of 40–450 mass-to-charge ratio and an electron energy of 70 eV. Compounds were tentatively identified (> 95% match) by comparison with the National Institute of Standards and Technology/Environmental Protection Agency/National Institutes of Health (NIST/EPA/NIH) Mass Spectral Library (Data Version: NIST 05, Software Version 2.0 d) using the XCALIBUR v1.3 program (ThermoFinnigan) library. Further identification of the compounds was based on a comparison of mass spectra and retention times with those of authentic standards (Sigma) analyzed under similar conditions.
Two-alternative forced choice (2AFC) scent (orthonasal route) trials were conducted by untrained panelists differing in sex and age. Each floral sample contained two open flowers at stage 7. Each trial was conducted on three different occasions in a room with a constant temperature of 24 ± 2°C and panelists were blindfolded during the trials. Panelists (n =24 per trial) were first presented with either a PAP1-transgenic or control transgenic floral sample, and then with the reciprocal sample. Panelists were then asked to indicate the sample with the stronger scent for scent strength trials, or the sample with the more appealing scent for the preference trials. A chi-squared test was conducted to examine the effect of sample presentation order on the tests. A binomial distribution was conducted on the proportion of expected responses according to the null hypothesis of the PAP1-transgenic floral sample being more strongly scented/preferred over the control sample. P <0.05 was considered to be significant.
Proboscis extension response (PER) of honeybees to rose flower odors
The response of honeybees (Apis mellifera) to odor from PAP1-transgenic line 11 vs control flowers was evaluated using 72 naive foragers: the honeybee colony was kept inside a netted enclosure (6 × 12 m2) so that the tested foragers were naive to floral odors. Each bee was placed in a sectioned hollow plastic tube and conditioned for PER according to standard procedures (Shalit et al., 2004) as follows: the bee was fed 2 μl of a 50% (w/v) sucrose solution and allowed to acclimate for 1.5 h before the start of the trial. Odors were sampled from three control or PAP1-transgenic flowers enclosed in 500-ml glass vials containing one air inlet connected to a pump and an outlet connected to a 1-ml glass syringe. A constant computer-controlled stream of air was pushed into the vial and the air stream exiting the vial from the syringe flowed over the bee’s antennae and was immediately vented into an exhaust vent behind the bee (Shalit et al., 2004). Bees were subjected to two trials consisting of either GUS-transgenic control or PAP1-transgenic odor samples. The order of the odor sample presentation was: control/control, PAP1-transgenic/PAP1-transgenic, control/PAP1-transgenic or PAP1-transgenic/control (18 bees for each sample presentation order). Bees were tested sequentially with an intertrial interval of 8 min. During a trial, an odor was delivered for a period of 7 s. We recorded whether the bee extended its proboscis during the first 4 s, before delivery of an unconditioned stimulus (0.4 μl of a 50% sucrose solution) with the last 3 s of odor delivery.
The bees’ differential responses were defined as ‘0’ or ‘1’ for the same or different response, respectively, in the two trials. The binomial distribution of the bees’ differential responses to odors from the two trials was determined, and the differential response to either the same (control/control or PAP1-transgenic/PAP1-transgenic) or different (control/PAP1-transgenic or PAP1-transgenic/control) odors was analyzed by chi-squared test. In addition, the differential responses to the same (control/control or PAP1-transgenic/PAP1-transgenic) odors and the effect of the order of the odor sample presentation (control/PAP1-transgenic or PAP1-transgenic/control) were also analyzed using chi-squared test. P <0.05 was considered to be significant.
All statistical calculations were performed using JMP IN 5 (JMP IN Software, SAS Institute Inc., Cary, NC, USA).
PAP1-transgenic rose plants accumulate increased anthocyanin levels
Transgenic rose plants cv Pariser Charme expressing cauliflower mosaic virus (CaMV) 35S-driven PAP1 or uidA coding for the GUS reporter gene (Fig. 1) were generated following agroinoculation of embryogenic callus (Dohm et al., 2001). The transgenic nature of the PAP1 lines was confirmed using reverse transcription PCR analyses: PAP1 expression was detected in independent lines 6, 11, 12 and 13, but not in transgenic control plants (Fig. 1d). PAP1-transgenic lines developed and matured in a similar manner to their control counterparts, but demonstrated increased pigmentation throughout plant development, from the early plantlet stages in tissue culture to mature plants in the glasshouse (Fig. 1a,b). PAP1-transgenic lines accumulated six- to ninefold higher levels of anthocyanin per gram of leaf tissue than the control transgenic plants (Fig. 1c).
PAP1 has previously been reported to exert broad transcriptional activation of the phenylpropanoid pathway (Tohge et al., 2005). Analysis of the transcriptional pattern of the anthocyanin biosynthetic pathway in PAP1-transgenic rose lines revealed a significant increase in transcript levels of both CHS and ANS when compared with control plants (Fig. 2a,b). Transcript levels of CHI, F3H and DFR were similar in both PAP1-transgenic and control plants (Fig. 2c–e). RNA levels of PAL, catalyzing the first committed step in the phenylpropanoid pathway, were also comparable in PAP1-transgenic and control plants (Fig. 2f).
Flowers of PAP1-transgenic rose plants produce higher levels of volatile compounds from phenylpropanoid and isoprenoid biosynthetic pathways
PAP1-transgenic roses flowered normally in the glasshouse and accumulated 2.5–5-fold higher levels of anthocyanins than control GUS-transgenic flowers (Fig. 1c). We next examined whether the effect of PAP1 extended to the production of volatile secondary metabolites in independent PAP1-transgenic lines. Qualitative and quantitative characterization of volatile compounds accumulating in and emitted from flowers of PAP1-transgenic rose lines 6, 11 and 12 was based on GC-MS analyses conducted on floral tissue extracts and floral headspace samples of detached flowers. The volatile profile obtained from detached flowers has been shown to correspond to that obtained from flowers growing intact on plants (Helsper et al., 1998). The volatile composition of control and PAP1-transgenic flowers consisted of compounds originating from the phenylpropanoid, isoprenoid and fatty acid biosynthetic pathways (Table 1). The level of the phenylpropanoid compound eugenol accumulated in flowers of PAP1-transgenic lines was up to 20-fold higher than that in flowers of control plants. The latter accumulated very low levels of eugenol, representing c. 0.4% of the total internal pool of volatile compounds, whereas, in PAP1 flowers, eugenol levels reached 5–9% of the total pool of volatiles (Table 1). Levels of all other phenylpropanoid compounds in the extract and headspace, including that of the major accumulated compound benzyl alcohol, were similar in PAP1-transgenic and control flowers. Levels of the fatty acid volatile derivatives cis-3-hexenyl acetate and hexanal were also similar in PAP1-transgenic and control flowers (Table 1). Interestingly, an increase in the levels of emitted volatiles of isoprenoid origin was revealed in flowers from all PAP1-transgenic lines when compared with control flowers. Specifically, levels of the major emitted terpenoid volatile compound, the sesquiterpene germacrene D, were up to 8.5-fold higher in PAP1-transgenic flowers. Emission of the norisoprenoid compound β-ionone was also dramatically increased (up to sixfold) in PAP1-transgenic flowers. By contrast, the internal pool levels of detected terpenoid compounds were similar in PAP1-transgenic and control flowers. Overall, the sum of emitted volatile compounds was up to approximately fourfold higher in flowers of PAP1-transgenic lines when compared with control flowers, whereas the level of all accumulated volatile compounds was not significantly different between the two (Table 1).
Table 1. Volatile compounds accumulating in and emitted from PRODUCTION OF ANTHOCYANIN PIGMENT1 (PAP1)- and β-glucuronidase (GUS)-transgenic (control) Rosa hybrida cv Pariser Charme flowers
The levels of emission ((e), μg per flower per 24 h) and internal pools ((p), μg per flower) of volatile compounds produced by flowers of PAP1-transgenic lines 6, 11 and 12 (Pap-6, Pap-11, Pap-12, respectively) when compared with GUS-transgenic control flowers (Gus) were analyzed quantitatively by gas chromatography-mass spectrometry (GC-MS). Flowers at stage 7 (outer petal whorl is open, but inner petal whorl is still closed and reproductive organs are not yet visible) were sampled. Values represent the means of three repeats; standard error is indicated in parentheses. Levels of emitted/accumulated volatiles were compared between PAP1- and GUS-transgenic flowers by Student’s t-test following one-way ANOVA: significant difference from control flowers *, P <0.05. The identification of compounds was confirmed by comparison of mass spectra and retention times with those of authentic compounds analyzed under similar conditions. The volatile compounds presented in the table represent 98% of the total detected volatiles.
0.01 (± 0.002)p
0.22 (± 0.13)p*
0.13 (± 0.06)p*
0.22 (± 0.11)p*
0.56 (± 0.08)e
0.38 (± 0.07)e
0.39 (± 0.19)e
0.49 (± 0.15)e
0.037 (± 0.001)p
0.036 (± 0.001)p
0.036 (± 0.001)p
0.047 (± 0.014)p
1.3 (± 0.23)p
1.44 (± 0.43)p
1.08 (± 0.24)p
0.89 (± 0.18)p
0.48 (± 0.12)e
2.66 (± 0.41)e*
1.49 (± 0.72)e*
4.1 (± 0.92)e*
0.3 (± 0.11)p
0.45 (± 0.16)p
0.37 (± 0.14)p
0.3 (± 0.05)p
0.12 (± 0.01)e
0.64 (± 0.09)e*
0.47 (± 0.15)e*
1 (± 0.4)e*
0.06 (± 0.01)e
0.3 (± 0.06)e*
0.29 (± 0.11)e*
0.36 (± 0.08)e*
0.05 (± 0.0009)p
0.05 (± 0.001)p
0.07 (± 0.01)p
0.08 (± 0.01)p
0.05 (± 0.02)e
0.18 (± 0.1)e
0.39 (± 0.22)e*
0.05 (± 0.01)e
0.07 (± 0.02)p
0.12 (± 0.04)p
0.13 (± 0.03)p
0.15 (± 0.03)p*
0.22 (± 0.04)e
0.51 (± 0.04)e*
0.5 (± 0.14)e*
0.6 (± 0.04)e*
0.01 (± 0.0008)p
0.01 (± 0.002)p
0.01 (± 0.002)p
0.01 (± 0.003)p
Fatty acid derivatives
0.08 (± 0.003)p
0.09 (± 0.019)p
0.1 (± 0.031)p
0.07 (± 0.007)p
0.51 (± 0.02)p
0.62 (± 0.02)p
0.53 (± 0.06)p
0.52 (± 0.03)p
Sum of emitted volatiles
1.49 (± 0.07)
4.66 (± 0.46)*
3.53 (± 0.93)*
6.59 (± 0.96)*
Sum of volatile internal pools
2.39 (± 0.32)
3.01 (± 0.69)
2.38 (± 0.4)
2.32 (± 0.54)
Increased levels of volatiles in PAP1-transgenic lines are not caused solely by transcriptional activation of their respective biosynthetic genes
Floral scent production presents highly ordered patterns and is regulated at several levels. The transcriptional regulation of structural genes represents one of the main mechanisms controlling volatile production (Verdonk et al., 2005; van Schie et al., 2006; Colquhoun et al., 2010, 2011; Spitzer-Rimon et al., 2010). To examine whether increased levels of volatile compounds produced in PAP1-transgenic rose flowers stem from the transcriptional up-regulation of their respective scent biosynthetic genes, quantitative real-time PCR analyses were performed on flowers just before anthesis (stage 4) when the machinery for scent production is already active (Dudareva & Pichersky, 2006). Transcript levels of GDS were significantly higher in all PAP1-transgenic lines (Fig. 3a), in agreement with the emission levels of germacrene D in independent PAP1-transgenic lines: the highest levels of both transcript and emission were detected in PAP1 line 12, followed by line 6 and line 11 (Fig. 3a, Table 1). However, levels of CCD1 and EGS transcript were not correlated with increased levels of their respective volatile products, β-ionone and eugenol, among flowers of PAP1-transgenic lines: transcript levels of EGS did not differ significantly between PAP1-transgenic and control flowers, whereas a c. 10-fold reduction in CCD1 transcript level was observed in PAP1-transgenic flowers when compared with controls (Table 1, Fig. 3b,c).
To further characterize the relationship between the levels of volatile compounds and the transcripts of their respective structural genes, we analyzed the expression of several additional rose scent biosynthetic genes. Transcript levels of OOMT1 (Lavid et al., 2002), AAT1 (Shalit et al., 2003) and PAAS (Farhi et al., 2009) were increased in flowers of PAP1-transgenic lines when compared with controls (Fig. 3d–f). However, their respective volatile products were below detection levels in both the headspace and extracts of PAP1-transgenic and control flowers. Phenylethyl alcohol, which has been shown to derive from phenylacetaldehyde (Yang et al., 2009), was also not detected in any of the examined flowers.
Humans and bees distinguish olfactory features of PAP1-transgenic rose flowers
Floral scent has aesthetic value for humans and is of key importance to the plant in attracting pollinators (Smith et al., 2006; Raguso, 2008). To examine whether the change in volatile profile of PAP1-transgenic flowers is detectable to human scent perception, floral scent strength and preference were evaluated in PAP1-transgenic vs control flowers using 2AFC tests (Fig. 4). Irrespective of the order of sample presentation, PAP1-transgenic floral samples (from all three independent lines) were given a significantly higher score for scent strength than control flowers: 81%, 79% and 93% of the panelists indicated that PAP1 floral samples from transgenic lines 6, 11 and 12, respectively, were more strongly scented than floral samples from control plants (Fig. 4, top row, P <0.05). In a separate set of trials designed to test scent preference, significant preference for PAP1-transgenic floral samples was revealed: 79%, 69% and 90% of the panelists preferred the scent of floral samples from PAP1-transgenic lines 6, 11 and 12, respectively, to that of the control floral samples (Fig. 4, bottom row, P <0.05).
To examine whether PAP1-transgenic and control rose flower odors could be distinguished by honeybees – the native pollinators of some wild rose species (Shalit et al., 2004) – we monitored the bees’ PER to PAP1-transgenic line 11 (which showed the smallest increase in volatile emission levels) vs control flowers. Bees were subjected to two trials consisting of the presentation of control and PAP1-transgenic odor samples in one of the following orders: control/control, PAP1-transgenic/PAP1-transgenic, control/PAP1-transgenic or PAP1-transgenic/control. The bees’ differential responses were defined as ‘0’ or ‘1’ for same or different responses, respectively, during the two trials. A significantly higher differential PER (P <0.0001) was revealed when bees were subjected to odors from different (control/PAP1-transgenic or PAP1-transgenic/control) floral samples when compared with bees subjected to odors from the same (control/control or PAP1-transgenic/PAP1-transgenic) floral samples (Fig. 5a). Differential responses of bees subjected to odors from the combinations of the same floral samples were very low and similar to each other, indicating high consistency of the bees’ responses to both floral odors (Fig. 5b). The bees’ ability to discriminate between PAP1-transgenic and control flower odors was further reflected by the significance of the effect of the order of odor presentation (control/PAP1-transgenic vs PAP1-transgenic/control) on the differential PER: 0.38 vs 0.77 (P =0.01) for control/PAP1-transgenic vs PAP1-transgenic/control sample order presentation (Fig. 5c).
PAP1 has been reported previously to exert transcriptional activation of the phenylpropanoid pathway, with overexpression of this gene in several plant systems –Arabidopsis, tomato and Petunia– leading to increased accumulation of anthocyanins (Borevitz et al., 2000; Mathews et al., 2003; Matousek et al., 2006; Ben Zvi et al., 2008a; Li et al., 2010; Shi & Xie, 2011). Flowers of the latter also produce increased levels of volatile phenylpropanoid compounds. The expression of PAP1 in the rose ‘Pariser Charme’ enabled us to explore its effect on volatile production in flowers that produce an array of volatile compounds originating from diverse biochemical origins – sesquiterpenes, monoterpenes, norisoprenoids, fatty acid derivatives and phenylpropanoid volatile compounds. This is in contrast with petunia cv Blue Spark which produces exclusively phenylpropanoid volatile compounds (Ben Zvi et al., 2008a).
Flowers of mature PAP1-transgenic roses, similar to PAP1-transgenic petunia (Matousek et al., 2006), produced higher levels of anthocyanins than flowers from control plants. Analysis of the volatile profile revealed increased levels of volatile compounds from various biosynthetic origins in flowers of PAP1-transgenic rose plants when compared with controls. Specifically, PAP1-transgenic flowers accumulated higher levels of the phenylpropanoid compound eugenol and emitted higher levels of isoprenoid compounds. Increased emission of germacrene D in PAP1-transgenic flowers could be correlated with an increased transcript level of GDS, suggesting that the enhanced emission of germacrene D was caused by transcriptional activation of its structural gene. As the level of the internal pool of germacrene D was similar in flowers of PAP1-transgenic and control plants, increased emission levels of germacrene D probably resulted from release of the surplus produced as a result of the increased generation of this compound in PAP1-transgenic flowers. Similarly, analyses of germacrene D production in Rosa hybrida‘Fragrant Cloud’ revealed that its internal pool is constant, and does not reflect the temporal diurnal changes in germacrene D in the headspace of the flower (Hendel-Rahmanim et al., 2007). Analyses of β-ionone levels revealed that, similar to germacrene D, although the internal pool was not affected by PAP1, the emitted levels were up to sixfold higher in PAP1-transgenic vs control flowers. The native emission pattern of β-ionone from flowers of rose, as well as petunia, has been shown to correlate strongly with the transcript levels of CCD1, encoding the enzyme that cleaves β-carotene to yield β-ionone (Simkin et al., 2004; Huang et al., 2009). However, in PAP1-transgenic roses producing increased levels of β-ionone, the transcript level of CCD1 was strongly down-regulated. These differences between metabolite and transcript levels in the transgenic background cannot be explained by the time of sampling as, at a different time point (13:00 h), CCD1 transcript levels were still similar in PAP1-transgenic and control flowers (not shown). However, the discrepancy between metabolite and transcript levels might be ascribable to differences in flux in the isoprenoid pathway, similar to the previously described effects of flux in the phenylpropanoid pathway on benzoid volatile production (Zuker et al., 2002; Ben Zvi et al., 2008a; Colon et al., 2010). Interestingly, as, in contrast with its product eugenol, the transcript levels of EGS were not affected by PAP1, the increase in this metabolite may also be explained by PAP1-enhanced flux in the phenylpropanoid pathway (Ben Zvi et al., 2008a).
The complexity of floral scent was further evidenced by the observation that the enhancement of PAAS, AAT1 and OOMT1 transcript levels in PAP1 transgenes did not yield increased levels of their respective candidate volatile products (phenylacetaldehyde, geranyl/citronellyl acetates and orcinol methyl ether). The lack of increase in the levels of these volatiles might be caused by secondary modification of the metabolite, for example, the glycosylation of phenylacetaldehyde/phenylethyl alcohol described in rose flowers (Hayashi et al., 2004). However, the application of a β-glucosidase mixture to PAP1-transgenic rose flowers did not enhance volatile levels (data not shown). Similarly, in Petunia, application of glycosidase did not affect volatile levels in PAP1-expressing flowers (Ben Zvi et al., 2008a). Hence, the lack of substrate availability may be a more plausible explanation for the inability of PAP1 to enhance the production of the aforementioned metabolites. Indeed, the lack of available substrate has been shown previously to be a limiting factor in volatile ester production in various plant systems, including rose flowers (Shalit et al., 2003; Beekwilder et al., 2004; Guterman et al., 2006).
The changes in the volatile profile of PAP1-transgenic rose vs control flowers detected by GC-MS analyses were also clearly detectable by the olfactory analyses, revealing a strong preference for the three analyzed PAP1-transgenic lines (as reported by 69–90% of panelists) over control flowers. This is one of a few examples in which floral scent is up-regulated through genetic transformation to levels that are detectable by untrained panelists (Zuker et al., 2002; Lucker et al., 2004; Davidovich-Rikanati et al., 2007). Interestingly, there was a direct correlation between the levels of volatiles detected by GC-MS in the different lines, that is, from highest to lowest in lines 12, 6 and 11, and the olfactory scores given to these lines by the panelists.
Previous reports (Wright et al., 2002, 2005) have demonstrated that bees can discriminate among flowers that vary in scent intensity and/or relative concentrations of compounds. Olfactory trials employing bees of PAP1 vs control rose flowers revealed, for the first time, that the transgenic approach to floral scent manipulation might yield changes in floral odor that are sufficiently strong to be detectable by pollinators: bees conditioned to the odor of either control or PAP1-transgenic flowers were able to distinguish between the two. The bees’ ability to discriminate between floral odors of PAP1-transgenic and control plants was significantly higher when they were first presented with, and conditioned to, the odor of PAP1-transgenic flowers, further validating the notion that the change in odor level/composition is sufficient for discrimination by bees (Wright et al., 2005; Smith et al., 2006; Reinhard et al., 2010).
The effect of PAP1 on volatiles from both the phenylpropanoid and isoprenoid pathways could be envisioned, although, to date, investigations in planta have shown an exclusive association of PAP1 with phenylpropanoid metabolism (Tohge et al., 2005). For example, a comparison of PAP1-expressing Arabidopsis cells with their wild-type counterparts revealed that the high levels of anthocyanin accumulating in the former lead to the activation of an array of genes from different metabolic pathways, not only phenylpropanoids (Shi & Xie, 2011). Furthermore, careful inspection of the transcriptome of PAP1-overexpressing Arabidopsis plants (Dare et al., 2008) using an annotated Arabidopsis gene dataset (Lange & Ghassemian, 2003) revealed the transcriptional up-regulation of several genes assigned to isoprenoid metabolism. Among these were the functionally characterized genes coding for mevalonate diphosphate decarboxylase (MPDC1; At2g38700) (Cordier et al., 1999) and 2-C-methyl-d-erythritol 4-phosphate cytidyltransferase (MCT; At2g02500) (Rohdich et al., 2000). The enzymes MPDC1 and MCT are involved in very early stages of isoprenoid production via mevalonate (MVA) and 2-C-methyl-d-erythritol 4-phosphate (MEP) pathways, respectively. In addition, PAP1-transgenic Arabidopsis seedlings accumulated increased transcript levels of terpene synthase homolog (At3g32030) (Lange & Ghassemian, 2003; Dare et al., 2008). Interaction between different secondary metabolic pathways, that is, phenylpropanoid and isoprenoid, was also revealed in light signal transduction tomato mutants (Enfissi et al., 2010), as well as in Impomoea flowers (Majetic et al., 2010) and rice (Kim et al., 2010). The mechanism(s) underlying these inter- and intra-relations in the metabolic pathways responsible for the production of specialized metabolites remains to be deciphered.
This work was funded by Israel Science Foundation grant nos. 269/09 and 432/10 and BARD grant no. US-4322-10. A.V. is an incumbent of the Wolfson Chair in Floriculture.