Keywords

11.1 Introduction

The genus Nicotiana (Solanaceae) consists of 76 species, which have flowers with tubular corollas that display diverse colors, including white, pink, magenta, purple, red, green, yellow, and white with ultraviolet light reflectance (Goodspeed 1954; McCarthy et al. 2015). Approximately half of Nicotiana species are allopolyploids that arose from both whole-genome duplication and interspecific hybridization (Chase et al. 2003; Knapp et al. 2004; Clarkson et al. 2004, 2010). These hybrid species may be predicted to display intermediate phenotypes between their diploid progenitors; however, Nicotiana allotetraploids can show transgressive floral colors (which fall outside the range of the diploid progenitors), and related polyploid accessions from the same origin can display floral color variation (McCarthy et al. 2015). We are interested in the genetic basis of these floral color differences.

Nicotiana tabacum L originated approximately 0.6 million years ago from maternal N. sylvestris and paternal N. tomentosiformis progenitors (Clarkson et al. 2004, 2010, 2017). First-generation allotetraploid lines that share the same parentage as natural N. tabacum were created synthetically by K. Y. Lim, Queen Mary University of London (London, UK). Floral color in accessions of natural N. tabacum ranges from light pink to magenta, and independent synthetic lines also display floral color differences, although the range is limited from light pink to dark pink (McCarthy et al. 2015, 2017). To determine the gene expression differences underlying this variation in floral color, we focused on two natural N. tabacum accessions (N. tabacum 095-55, magenta flowers, termed 095-55, and N. tabacum “Chulumani”, light-pink flowers, termed Chulumani), two first-generation synthetic lines (QM24, light-pink flowers, and QM25, dark-pink flowers), and their diploid progenitors, N. sylvestris (white flowers) and N. tomentosiformis (dark-pink flowers) (Fig. 11.1).

Fig. 11.1
figure 1

Flavonoid biosynthetic pathway and Nicotiana flower colors and pigments. a Flavonoid biosynthetic pathway with structural enzymes in large font next to arrows; reproduced from McCarthy et al. (2017). CHS, chalcone synthase; CHI, chalcone isomerase; F3H, flavanone 3-hydroxylase; F3′H, flavonoid 3′-hydroxylase; F3′5′H, flavonoid 3′,5′-hydroxylase; FLS, flavonol synthase; DFR, dihydroflavonol 4-reductase; ANS, anthocyanidin synthase. Shared substrates appear in small font (DHK, dihydrokaempferol; DHQ, dihydroquercetin; DHM, dihydromyricetin), and resulting flavonol (black text) and anthocyanin (colored text) pigments. Bar graphs represent the pigment concentration (mg/g fresh tissue) of kaempferol, quercetin, pelargonidin, and cyanidin in the Nicotiana accessions examined in this chapter. b Floral limb photographs of the Nicotiana accessions used. Nicotiana sylvestris (white) and N. tomentosiformis (dark pink) are the maternal and paternal progenitors, respectively, of N. tabacum. Natural tobacco accessions: N. tabacum 095-55 (dark pink) and N. tabacum “Chulumani” (light pink). Synthetic lines QM24 (light pink) and QM25 (dark pink). c Floral bud photographs of N. tabacum 095-55 at 60, 85, and 95% of anthesis length, the developmental time points used in this study. 095-55, Nicotiana tabacum 095-55; Chulu, N. tabacum “Chulumani”; sylv, N. sylvestris; tomf, N. tomentosiformis. QM24 and QM25 are first-generation synthetic lines

Full size image

In many plant species, anthocyanin pigments are responsible for the red, pink, purple, and blue floral colors. Equally important for pollination are flavonols, which are not visible to humans, but absorb ultraviolet light and thereby provide visual cues to birds and bees. In a previous study, we quantified anthocyanidins (anthocyanins without attached sugars) and flavonols in N. sylvestris, N. tomentosiformis, 095-55, Chulumani, QM24, and QM25 to determine the biochemical basis of floral color differences. The predominant anthocyanidin in all samples was cyanidin, which provides a pink color. The amount of cyanidin was consistent with flower color: low levels in the light-pink flowers of Chulumani and QM24, none in the white flowers of N. sylvestris, and high levels in the dark-pink flowers of N. tomentosiformis and QM25 and the magenta flowers of 095-55 (McCarthy et al. 2017; Fig. 11.1). In addition to cyanidin, the natural N. tabacum accessions contain a small amount of transgressive pelargonidin, which is not present in either progenitor or in the synthetic lines. Floral tissue in all accessions contains the flavonols kaempferol and quercetin. Kaempferol levels are low in N. sylvestris, 095-55, and QM25, high in N. tomentosiformis and Chulumani, and intermediate in QM24. Quercetin levels are low in N. sylvestris, high in N. tomentosiformis, and intermediate in natural and synthetic N. tabacum accessions (McCarthy et al. 2017; Fig. 11.1).

Anthocyanin and flavonol pigments are produced by the flavonoid biosynthetic pathway, a branched pathway that yields multiple pigment types (Grotewold 2006; Fig. 11.1). The enzymes chalcone synthase (CHS), chalcone isomerase (CHI), and flavanone 3-hydroxylase (F3H) act early in the pathway and are necessary for the production of both flavonols and anthocyanins. The enzymes flavonoid 3′-hydroxylase (F3′H) and flavonoid 3′,5′-hydroxylase (F3′5′H) shift the pathway along different branches toward different pigments. Flavonol synthase (FLS) produces flavonols from the intermediate substrates dihydrokaempferol, dihydroquercetin, and dihydromyricetin, whereas dihydroflavonol 4-reductase (DFR) and anthocyanin synthase (ANS) convert these substrates into anthocyanidins, which can then be converted to anthocyanins (Grotewold 2006). The transcription factors anthocyanin 1 (AN1, basic helix-loop-helix), anthocyanin 2 (AN2, R2R3 MYB), and WD40 act together as a complex to induce anthocyanin production (Grotewold 2006).

Many evolutionary changes in floral color have been attributed to mutations in genes of the flavonoid biosynthetic pathway. Floral color shifts from blue to red are the result of inactivation, down-regulation, or deletion of F3′5′H or F3′H genes, resulting in production of red pelargonidin pigmentation instead of blue cyanidin or delphinidin (Zufall and Rausher 2004; Streisfeld and Rausher 2009; Des Marais and Rausher 2010; Smith and Rausher 2011; Wessinger and Rausher 2015). Several studies of the genetic basis of shifts from pigmented to white flowers have identified the cause as inactivation of the R2R3 MYB gene (orthologous to AN2 in Nicotiana) that regulates the pathway (Quattrocchio et al. 1999; Schwinn et al. 2006; Hoballah et al. 2007). Some studies, however, have found that partial reduction or complete abolition of anthocyanin pigment is driven by repression of the pathway by an R3 MYB transcription factor (Yuan et al. 2013; Gates et al. 2017). Our questions focus on the genetic basis of the variety of floral colors observed in N. tabacum accessions and are as follows: (1) Do the expression levels of genes involved in anthocyanin synthesis (DFR, ANS, AN1, and AN2) predict cyanidin concentration, i.e., is the expression of these genes up-regulated in dark-pink flowers? (2) Are light-pink flowers lighter than dark-pink flowers because DFR and ANS are activated later in development, resulting in lower cyanidin accumulation? (3) In pairwise comparisons of dark-pink and light-pink flowers, are the same genes involved in the flavonoid biosynthetic pathway differentially expressed?

11.2 Materials and Methods

11.2.1 Plant Growth and Material

We grew plant material in greenhouses with natural sunlight and temperatures between 10 and 30 °C. We used the following plant material: N. sylvestris A04750326 (Radboud University, Nijmegen, The Netherlands), N. tomentosiformis BRNO 4103 (acquired from A. Kovařík, Brno, Czech Republic), N. tabacum 095-55 (IPK Gatersleben, Gatersleben, Germany), N. tabacum “Chulumani” (collected in the field in Bolivia by S. Knapp), and two first-generation synthetic lines, QM24 and QM25 (created by K. Y. Lim at Queen Mary University of London by crossing 4× autotetraploid N. sylvestris and 4× autotetraploid N. tomentosiformis). Because we used multiple accessions of N. tabacum, we will refer to all six plant lines as accessions.

11.2.2 Developmental Series, RNA Extraction, and cDNA Synthesis

We examined floral development in each accession to identify stages for use in transcriptome analyses by timing of the appearance of floral pigment. We measured corolla tube length, including the floral limb (the portion of the flower that opens at anthesis), at anthesis in 5–10 flowers per accession by dissecting the corolla tube, pinning it flat, photographing it, and measuring length using ImageJ (version 1.51 k; Rasband 1997). We used the mean tube length at anthesis to calculate the percentage of anthesis length at which pigment appeared for each accession. At 60% of anthesis length, none of the accessions displayed floral pigmentation, but different accessions began to show pigmentation by 85% or 95% of anthesis length. We therefore selected 60, 85, and 95% of anthesis corolla length as stages for transcriptome analysis. Floral buds were collected using the corolla length values ±1 mm. We dissected the corolla tissue, cut the stamens from the corolla, and placed the tissue in liquid nitrogen. We collected three biological replicates from different plants, where available, for each accession at each developmental time point. Each independent synthetic accession consisted of a single first-generation individual, so three replicate flowers were collected from each plant for these lines. Because of the scarcity of material, we collected only two biological replicates for N. tomentosiformis at 85% of anthesis length and two flowers for the synthetic line QM24 at 95% of anthesis length.

We extracted RNA using the RNeasy Mini Plant Kit (Qiagen, Hilden, Germany), DNase-treated using the Turbo-DNase Kit (Ambion, Thermo Fisher Scientific, Waltham, MA, USA), and synthesized cDNA from 1 µg of RNA using the SuperScript III Reverse Transcriptase Kit (Invitrogen, Carlsbad, CA, USA) with the provided oligo-dT primer.

11.2.3 Digital Droplet PCR (ddPCR)

To quantify FLS and DFR transcript accumulation, we performed ddPCR on a QX200 ddPCR system (Bio-Rad Laboratories, Hercules, CA, USA). We used QX200 ddPCR EvaGreen SuperMix and 2 pmol of each primer with the following program: 95 °C for 5 min, followed by 40 cycles of 95 °C for 30 s and 60 °C for 60 s, followed by 4 °C for 5 min and 90 °C for 5 min, using a ramp rate of 2 °C per second. We normalized transcript accumulation using the geometric mean of three reference genes: elongation factor 1 alpha (EF1α), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and ribosomal protein gene L25. Table 11.1 lists the primer sequences. We analyzed the ddPCR data using QuantaSoft Analysis Pro, version 1.0 (Bio-Rad Laboratories).

Table 11.1 Primer sequences used for digital droplet PCR
Full size table

11.2.4 Illumina Library Preparation and Sequencing

To prepare cDNA libraries for 60, 85, and 95% of anthesis length samples from all six accessions, strand-specific libraries were constructed from mRNA as previously described (Zhong et al. 2011) and sequenced on a HiSeq 2500 system (Illumina, San Diego, CA) with 1 × 85 bp reads. Sequencing data from a smaller, preliminary experiment was also included to create our reference assembly (see below in the “Transcriptome analyses” section for details), using corolla tissue at anthesis from three biological replicates each of N. tabacum 095-55 and N. tabacum 51789 (this accession was not used in the larger analysis). Libraries were prepared with the NEBNext Ultra Directional RNA Library Prep Kit for Illumina (New England Biolabs, Ipswich, MA, USA) according to the manufacturer’s protocols. Library quality was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA), and libraries were sequenced on an Illumina NextSeq 500 platform at the University of California, Riverside Institute for Integrated Genome Biology Genomics Core Facility with 1 × 150 bp reads.

11.2.5 Transcriptome Analyses

We processed raw sequencing reads on the University of California, Riverside Biocluster by trimming adaptors using Cutadapt (version 1.15; Martin 2011), followed by filtering with a quality score of 30 and a minimum length of 60 bp using Sickle (version 1.33; Joshi and Fass 2011). We initially performed analyses of cleaned reads using a genome-guided approach in TopHat (version 2.1.1), Cufflinks (version 2.2.1; Trapnell et al. 2012), and cummerbund (version 2.20; Goff et al. 2013) with the available genomes of N. sylvestris (Sierro et al. 2013), N. tabacum (Edwards et al. 2017), and N. tomentosiformis (Sierro et al. 2013). Although N. sylvestris and N. tomentosiformis are the progenitors of the allotetraploid N. tabacum, the percentage of reads mapped varied widely (20–70%) both within and among accessions and depending on which genomes were used as the reference. We therefore adopted a de novo approach using the Trinity pipeline (version 2.4; Grabherr et al. 2011; Haas et al. 2013).

We generated a single reference assembly using all sequencing reads (60, 85, and 95% of anthesis length from the four N. tabacum accessions, N. sylvestris, and N. tomentosiformis and at anthesis from N. tabacum 095-55 and N. tabacum 51789), and normalized them in silico to 50× to improve the efficiency of the assembly by decreasing the number of reads included (Haas et al. 2013). We mapped sequencing reads back to the reference assembly using RNA-Seq by Expectation Maximization implemented with Bowtie2 (version 2.2.9; Langmead and Salzberg 2012). We annotated the reference assembly using the Transdecoder (version 3.0) plugin in Trinity by performing BLASTx and BLASTp searches against the Swiss-Prot (The UniProt Consortium 2017) and Pfam (Finn et al. 2016) databases. Because of short read lengths, it was difficult to assemble full gene transcripts; therefore, any given gene locus may have several corresponding contigs in the de novo assembly. Additionally, likely due to short read lengths and nonuniformity of polymorphisms between the N. sylvestris and N. tomentosiformis genomes, there were difficulties isolating homeologous sequences in the de novo assembly; therefore, some contigs correspond to N. sylvestris copies, some contigs correspond to N. tomentosiformis copies, and some correspond to both homeologs.

We performed differential expression analyses with the bioconductor package limma/voom (version 3.6; Ritchie et al. 2015). We ran two sets of comparisons, across accessions within developmental time points and across developmental time points within accessions. To ensure robust differences, we classified differentially expressed contigs as those with p-value < 0.05, false discovery rate < 0.05, and log2 fold change > 2 (absolute value).

11.2.6 Identifying Flavonoid Biosynthesis Transcripts in the de novo Assembly

To identify contigs in the de novo assembly that correspond to genes in the flavonoid biosynthetic pathway, we performed BLAST searches using previously identified flavonoid biosynthetic genes from Iochroma and Petunia species (both in family Solanaceae) to find putative orthologs from N. sylvestris and N. tomentosiformis in GenBank. A single hit was found for each gene in the flavonoid biosynthetic pathway for N. sylvestris and N. tomentosiformis, apart from CHS for which two genes were found. We then executed custom BLAST searches in Geneious (version 10, Auckland, New Zealand) with the detected N. sylvestris and N. tomentosiformis orthologs against the de novo reference assembly to identify the contigs of genes likely involved in flavonoid biosynthesis in Nicotiana. We extracted these flavonoid biosynthesis contigs from read count data for each transcriptome and from the differential expression analyses for use in additional investigations.

11.2.7 Comparative Quantitative Expression of Flavonoid Biosynthetic Genes

To determine whether increase in DFR and ANS expression accounts for the higher cyanidin concentration observed in dark-pink-flowered species, we generated strip charts in R (version 3.4.2) using the ggplot2 package (version 2.2.1; Wickham 2009). We plotted the log2 fold change values of differentially expressed flavonoid biosynthetic genes for pairwise comparisons between accessions within each developmental time point. To establish whether the magnitude and direction of expression differences in these transcripts reflected those observed in the transcriptome overall, we overlaid these strip charts onto violin plots (which display the distribution of the data in a manner similar to that of a histogram) of the differentially expressed genes (DEGs) from the whole transcriptome using ggplot2. As our findings suggested that some floral color differences involved more complex dynamics than simply increased levels of DFR and ANS, we also examined relative expression dynamics across the pathway to identify patterns that involved more than a single gene. To do this, we summed the transcript levels for each gene (e.g., CHS) and produced heatmaps in R using the mixOmics package (version 6.3.1; Cao et al. 2016). To further investigate whether competition for the same substrates may affect pigment concentrations, we calculated the FLS:DFR ratio using both transcriptome and ddPCR data and compared it to floral phenotype and total anthocyanidin concentration.

11.2.8 Comparative Quantitative Expression of Flavonoid Biosynthetic Genes in Light-Pink and Dark-Pink Flowers

To determine whether accessions with the same floral color display similar relative transcript levels of flavonoid biosynthetic genes, we plotted Venn diagrams using the R packages Venn diagram (version 1.6.18; Chen and Boutros 2011) and gplots (version 3.0.1; Warnes et al. 2009). We compared DEGs between N. tabacum accessions and each of their progenitors to identify those shared by light-pink or dark-pink flowers that differ from progenitor expression levels. We also examined overlap in pairwise comparisons between N. tabacum accessions with light-pink and dark-pink flowers (095-55 versus Chulumani, 095-55 versus QM24, and Chulumani versus QM25).

11.2.9 Comparative Quantitative Expression of Flavonoid Biosynthetic Genes Across Development

To establish whether differences in developmental timing of DFR and ANS activation underlie changes in floral color, we examined patterns of expression of flavonoid biosynthetic genes across development within each accession by generating strip charts overlaid on violin plots and heatmaps, as described in the Sect. 11.2.7.

11.3 Results

11.3.1 Patterns of Differential Gene Expression Among Accessions and Across Development

To elucidate the genetic basis of floral color variation among white, light-pink, and dark-pink Nicotiana flowers, we used transcriptome analyses and ddPCR to quantify the expression of flavonoid biosynthetic pathway genes from two natural and two synthetic accessions of N. tabacum and their diploid progenitors. Because of difficulty in mapping the transcriptomes to the publicly available Nicotiana genomes, we created a de novo assembly for sequence analysis. Sequencing coverage was low for one biological replicate of QM24 at 60% of anthesis length; therefore, we excluded this replicate from further analyses. The number of reads mapped from each replicate ranged from 3.66 to 9.42 million (Table 11.2).

Table 11.2 Number of reads mapped to the de novo reference assembly
Full size table

In pairwise comparisons of accession transcriptomes at a given developmental time point, we found the highest number of DEGs between progenitor species and the smallest number between synthetic lines (Table 11.3). There was also a higher number of DEGs between N. tabacum accessions and their progenitors than between N. tabacum accessions (Table 11.3). In comparisons across developmental time points within each accession, the highest number of DEGs was found between 60 and 95% of anthesis length, followed by 60 and 85% and between 85 and 95% (Table 11.4). Not surprisingly, far more genes were differentially expressed between accessions than across development within an accession.

Table 11.3 Differentially expressed contigs across the transcriptomes of Nicotiana accessions
Full size table
Table 11.4 Differentially expressed contigs across developmental time points for each Nicotiana accession
Full size table

11.3.2 Higher Levels of Anthocyanin-Specific Biosynthetic Transcripts Are Correlated with Pink Versus White Phenotype Comparisons, but Variation Between Pink Phenotypes Is More Complex

To determine whether relatively higher expression levels of DFR and ANS in dark-pink flowers might underlie differences in floral color, we compared DEGs involved in the flavonoid biosynthetic pathway between pairs of accessions at each developmental time point (Fig. 11.2). In comparisons between white (N. sylvestris) and pink (all other accessions) flowers, most of the differentially expressed flavonoid biosynthetic transcripts were higher in pink flowers. These included genes that produce anthocyanins (DFR, ANS, AN1, and AN2) as we hypothesized, but also other genes in the pathway (CHS, CHI, F3H, F3′H, and FLS). In comparisons between light-pink and dark-pink flowers, some of the DEGs were up-regulated in lighter flowers, whereas others were up-regulated in darker flowers. This contrasted with the findings for comparisons involving N. sylvestris (Fig. 11.2), suggesting that the production of dark- versus light-pink flowers is not simply the result of higher expression of anthocyanin biosynthetic genes, but that more complex expression dynamics in the pathway are involved.

Fig. 11.2
figure 2

Differential gene expression across Nicotiana accessions at a 60%, b 85%, and c 95% of anthesis length. The violin plots represent transcriptome-wide differentially expressed contigs (log2 fold change > 2 (absolute value)). The points represent differentially expressed contigs from the flavonoid biosynthetic pathway, colored for each gene. Dashed lines at 2 and −2 represent the cutoff for differentially expressed contigs. Pairwise comparisons are shown along the x-axis, and the accessions are color-coded according to flower color: gray = white; light pink = light pink; magenta = dark pink. 095-55, Nicotiana tabacum 095-55; Chulu, N. tabacum “Chulumani”; sylv, N. sylvestris; tomf, N. tomentosiformis. QM24 and QM25 are first-generation synthetic lines. Differentially expressed genes that are up-regulated in the first accession of the pair are in the positive (upper) section of each graph and those that are up-regulated in the second accession of the pair are in the negative (lower) section of each graph. CHS, chalcone synthase; CHI, chalcone isomerase; F3H, flavanone 3-hydroxylase; F3′H, flavonoid 3′-hydroxylase; F3′5′H, flavonoid 3′,5′-hydroxylase; FLS, flavonol synthase; DFR, dihydroflavonol 4-reductase; ANS, anthocyanidin synthase; AN1, anthocyanin 1; AN2, anthocyanin 2

Full size image

To determine whether relative expression levels across the pathway as a whole are consistent with a role in determining different color phenotypes, we generated heatmaps for the summed contig counts for each flavonoid biosynthetic gene across accessions. CHS expression levels were higher than expression levels in the other genes, so to facilitate interpretation of expression patterns among the genes present at lower levels, we also generated heatmaps without CHS (Fig. 11.3). Some patterns were consistent with white/light-pink (N. sylvestris, Chulumani, and QM24) versus dark-pink (N. tomentosiformis, 095-55, and QM25) floral color categories. Early in development (60% of anthesis length), DFR levels were lower in white/light-pink flowers. At 60 and 85% of anthesis length, F3H levels were higher in white/light-pink flowers. We also saw, however, patterns that did not correspond to floral color categories. The low DFR and F3H levels were not maintained in light-pink flowers at later developmental stages. In addition, at 60 and 85% of anthesis length, FLS expression was higher in N. tomentosiformis, Chulumani, and both synthetic lines than in N. sylvestris and 095-55, corresponding to lower concentration of flavonols in the white N. sylvestris and dark-pink 095-55 flowers. At 95% of anthesis length, FLS expression was similar across accessions. Expression of the regulatory genes AN1 (at all developmental time points) and AN2 (at 60 and 85% of anthesis length) was higher in N. tomentosiformis, 095-55, and Chulumani than in the other accessions (Fig. 11.3).

Fig. 11.3
figure 3

Heatmaps of summed transcript levels for flavonoid biosynthetic genes in Nicotiana accessions. Transcript levels were compared at 60% (a, b), 85% (c, d), and 95% (e, f) of anthesis length, with (a, c, e) and without (b, d, f) CHS, which encodes chalcone synthase. CHI, chalcone isomerase; F3H, flavanone 3-hydroxylase; F3′H, flavonoid 3′-hydroxylase; F3′5′H, flavonoid 3′,5′-hydroxylase; FLS, flavonol synthase; DFR, dihydroflavonol 4-reductase; ANS, anthocyanidin synthase; AN1, anthocyanin 1; AN2, anthocyanin 2. Accessions: 095-55, Nicotiana tabacum 095-55; Chulu, N. tabacum “Chulumani”; sylv, N. sylvestris; tomf, N. tomentosiformis. QM24 and QM25 are first-generation synthetic lines

Full size image

Because FLS and DFR encode proteins that function in different branches of the pathway and because their expression levels vary across accessions at 60% of anthesis length, we determined whether differences in the expression of these genes correlated with differences in floral color. Consistent with floral color differences, DFR levels at 60% of anthesis length were significantly higher in dark-pink flowers than in light-pink or white flowers (with the exception of 095-55 versus Chulumani). However, DFR levels were similar in white versus light-pink, light-pink versus light-pink, and dark-pink versus dark-pink comparisons (Fig. 11.2a). In contrast, FLS expression differences at 60% of anthesis length were not consistent with differences between flower colors; for example, in light-pink versus dark-pink comparisons, FLS expression was significantly higher in QM24 versus 095-55 but similar in Chulumani versus QM25 (Fig. 11.2a). Because FLS and DFR act on the same substrates and may compete for them, we calculated FLS:DFR expression ratios to establish whether this ratio was consistent with a role in determining floral color. In both the transcriptome and ddPCR data, the FLS:DFR ratio was high (>35 fold change) at 60% of anthesis length in light-pink flowers (Chulumani and QM24), whereas other accessions (with the exception of N. sylvestris in the ddPCR data) at 60% and all accessions at 85 and 95% of anthesis length had much lower FLS:DFR ratios (Fig. 11.4a, b). To determine whether the high ratios in light-pink (Chulumani and QM24) and white (N. sylvestris) flowers were consistent with lower levels of anthocyanidins, suggesting diversion of the pathway to flavonols by FLS, we plotted total anthocyanidin concentration against FLS:DFR ratio at 60% of anthesis length (Fig. 11.4c). Although the relationship was not linear, the accessions with high FLS:DFR ratios tended to have lower anthocyanidin concentration, suggesting that a high FLS:DFR ratio early in floral development may play a role in preventing the accumulation of anthocyanidins.

Fig. 11.4
figure 4

FLS:DFR expression ratio from transcriptome (a) and digital droplet PCR (ddPCR) (b) data at 60%, 85%, and 95% of anthesis length in Nicotiana accessions. FLS encodes flavonol synthase and DFR encodes dihydroflavonol 4-reductase. Black diamonds and lines represent the mean and standard deviation, respectively. c Total anthocyanidin concentration versus transcriptome FLS:DFR ratio with points representing flower color. 095-55, Nicotiana tabacum 095-55; Chulu, N. tabacum “Chulumani”; sylv, N. sylvestris; tomf, N. tomentosiformis. QM24 and QM25 are first-generation synthetic lines

Full size image

11.3.3 Different Pathway Modifications Underlie Shifts Between Light-Pink and Dark-Pink Flowers

To determine whether expression differences of specific genes in the flavonoid biosynthetic pathway at specific developmental time points underlie the differences between light-pink and dark-pink flowers, we used pairwise DEG comparisons between accessions. We generated Venn diagrams of all four N. tabacum accessions versus each of their diploid progenitors within developmental time points (Fig. 11.5a). We then extracted the identity of the contigs from the sectors of the Venn diagram that were shared between light-pink and between dark-pink accessions. Dark-pink accessions (095-55 and QM25) had higher DFR and ANS expression levels than N. sylvestris (Fig. 11.5b), which was expected because of the presence of cyanidin in the dark-pink flowers (Fig. 11.1a). Similar differences were expected for light-pink versus white comparisons, perhaps to a lesser magnitude, but this was not observed. Light-pink accessions (Chulumani and QM24) had higher FLS expression levels than N. sylvestris, but dark-pink accessions (095-55 and QM25) did not (Fig. 11.5b). The higher FLS expression levels exclusively in light-pink flowers were not expected. This suggests that in the development of light-pink flowers, the flux of the flavonoid biosynthetic pathway may be shunted toward the production of flavonols, whereas in dark-pink flowers it may be directed toward anthocyanins. In the comparisons of the N. tabacum accessions to N. tomentosiformis, light-pink (but not dark-pink) accessions had lower F3′H and DFR expression levels, whereas dark-pink (but not light-pink) accessions had higher ANS expression levels (Fig. 11.5b). These findings were consistent with lower cyanidin production in light-pink flowers but the higher ANS expression in dark-pink N. tabacum accessions does not seem to result in higher cyanidin production compared to N. tomentosiformis, perhaps because there is no accompanying increase in DFR expression levels.

Fig. 11.5
figure 5

Shared differentially expressed genes from the flavonoid biosynthetic pathway in comparisons between Nicotiana accessions. a Example Venn diagram of differentially expressed genes from pairwise comparisons of each N. tabacum accession with one diploid progenitor; analyses were performed for comparisons with both N. sylvestris and N. tomentosiformis, but only an example diagram is shown here. Light-pink and dark-pink sections highlight genes that are differentially expressed exclusively in light-pink accessions versus progenitor or dark-pink accessions versus progenitor, respectively. b Summary of the Venn diagram data for N. tabacum accessions versus progenitor at 60% of anthesis length. The genes in the table are those found in the light-pink and dark-pink sectors of (a) for comparisons between the N. tabacum accessions and N. sylvestris or N. tomentosiformis. These represent differentially expressed genes exclusive to light-pink or dark-pink accessions versus progenitor comparisons; pink text indicates higher expression in the N. tabacum accession, and black text indicates higher expression in the progenitor. (c, d) Venn diagrams of comparisons between light-pink and dark-pink N. tabacum accessions at 60% (c) and 85% (d) of anthesis length. Light-pink text represents genes that are up-regulated in light flowers; dark-pink text represents genes that are up-regulated in dark flowers; black text represents genes for which the expression levels do not correlate with flower color. Blue, purple, and orange labels and circles represent separate light-pink versus dark-pink comparisons. CHS, chalcone synthase; F3′H, flavonoid 3′-hydroxylase; F3′5′H, flavonoid 3′,5′-hydroxylase; FLS, flavonol synthase; DFR, dihydroflavonol 4-reductase; ANS, anthocyanidin synthase; AN1, anthocyanin 1; AN2, anthocyanin 2. Accessions: 095-55, Nicotiana tabacum 095-55; Chulu, N. tabacum “Chulumani”; sylv, N. sylvestris; tomf, N. tomentosiformis. QM24 and QM25 are first-generation synthetic lines

Full size image

To ascertain whether the same genes were differentially expressed in all comparisons of light-pink and dark-pink flowers, we generated Venn diagrams to compare differential expression in pairwise comparisons between light-pink and dark-pink N. tabacum accessions (095-55 versus Chulumani, 095-55 versus QM24, and Chulumani versus QM25). No DEGs were shared among all three comparison sets (Fig. 11.5c, d), suggesting that similar colors were achieved in different ways. In comparisons between 095-55 and both light-pink accessions at 60 and 85% of anthesis length, however, FLS expression levels were higher in the light-pink flowers. In contrast, light-pink flowers had lower DFR levels in 095-55 versus QM24 and QM25 versus Chulumani comparisons at 60% of anthesis length. Either high FLS or low DFR expression levels can result in a high FLS:DFR ratio, suggesting that the factor driving differences between light-pink and dark-pink flowers may be the ratio of FLS to DFR expression. No other consistent differences were detected between light-pink and dark-pink flowers, but lower levels of the regulatory genes AN1 (at 60 and 85% of anthesis length) and AN2 (at 60% of anthesis length) were observed in QM24 than in dark-pink 095-55, although not in other light-pink versus dark-pink comparisons. In addition, CHS was found at higher levels in the light-pink flowers of Chulumani versus 095-55 at 60 and 85% of anthesis length, but not in any other comparisons. This variation in results suggests that the mechanism behind the differences between light-pink and dark-pink floral color may be a combination of several factors.

11.3.4 DFR Up-Regulation Occurs Later in Development in Light-Pink Flowers

Gene expression comparisons between light-pink and dark-pink flowers found lower DFR expression in light-pink flowers at 60% of anthesis length; we therefore assessed whether light-pink floral color may be the result of activation of DFR later in development. DFR expression was up-regulated at 85% of anthesis length compared to 60% of anthesis length and remained at high levels at 95% in N. tomentosiformis, Chulumani, and both synthetic lines (Fig. 11.6a–f). In heatmaps showing relative transcript levels across development within each accession, very low levels of DFR expression were observed in N. sylvestris, Chulumani, and QM24 at 60% of anthesis length (Fig. 11.6g–l). To compare expression across development in all accessions, we generated strip plots of DFR expression levels from the six transcriptomes (Fig. 11.7). N. sylvestris displayed low levels of DFR throughout development, which is consistent with its lack of anthocyanidin pigmentation. DFR levels in N. tomentosiformis, 095-55, and QM25 were significantly higher than those of N. sylvestris at 60% of anthesis length (Fig. 11.2a), indicating that DFR had already been activated in these accessions at this time point. In contrast, light-pink-flowered accessions (Chulumani and QM24) at 60% of anthesis length had DFR expression levels similar to those of N. sylvestris, suggesting that activation was delayed in light-pink flowers (after 60% but before 85% of anthesis length). This delay in DFR activation may underlie the increased FLS:DFR ratio observed in these accessions, which may play a role in the decrease of anthocyanidin accumulation.

Fig. 11.6
figure 6

Gene expression dynamics across developmental time points in Nicotiana accessions. Differentially expressed genes between different time points in a N. sylvestris (sylv), b N. tomentosiformis (tomf), c N. tabacum 095-55 (095-55), d Chulumani (Chulu), e the synthetic line QM24, and f the synthetic line QM25. The violin plots represent transcriptome-wide differentially expressed genes (log2 fold change > 2 (absolute value)). The points represent differentially expressed flavonoid biosynthesis genes, colored for each gene. Dashed lines at 2 and −2 represent the cutoff for differentially expressed genes. The positive (upper) portion of the graph represents genes that are up-regulated at the earlier developmental time point; the negative (lower) portion represents genes up-regulated at the later developmental time point. Heatmaps of summed transcript levels for flavonoid biosynthetic genes across development for g N. sylvestris, h N. tomentosiformis, i 095-55, j Chulumani, k QM24, and l QM25 without CHS, which encodes chalcone synthase. CHI, chalcone isomerase; F3H, flavanone 3-hydroxylase; F3′H, flavonoid 3′-hydroxylase; F3′5′H, flavonoid 3′,5′-hydroxylase; FLS, flavonol synthase; DFR, dihydroflavonol 4-reductase; ANS, anthocyanidin synthase; AN1, anthocyanin 1; AN2, anthocyanin 2

Full size image
Fig. 11.7
figure 7

DFR expression levels for each Nicotiana accession at 60, 85, and 95% of anthesis length. Black diamonds and lines represent the mean and standard deviation, respectively. 095-55, Nicotiana tabacum 095-55; Chulu, N. tabacum “Chulumani”; sylv, N. sylvestris; tomf, N. tomentosiformis. QM24 and QM25 are first-generation synthetic lines

Full size image

11.4 Discussion

11.4.1 Higher Expression Levels of Anthocyanin-Specific Genes Differentiate White from Pink Flowers

In comparisons between white-flowered N. sylvestris and each pink accession, expression of some or all anthocyanin-specific genes (DFR, ANS, AN1, and AN2) and of other genes in the pathway was higher in pink flowers. These results suggest that the difference between white and pink flowers arises from changes in pathway regulation and/or from expression of the genes that produce anthocyanins, consistent with reports that overexpression of either AN1 or AN2 increases anthocyanin concentration (Pattanaik et al. 2010; Bai et al. 2011). However, no AN2 ortholog was detected in the N. sylvestris genome. Thus, it is not surprising that AN2 expression was higher in all pink-flowered accessions than in N. sylvestris at all developmental time points examined. AN2 activates DFR and ANS expression in Nicotiana (Pattanaik et al. 2010); therefore, the absence of the AN2 gene from N. sylvestris is consistent with the lower levels of DFR and ANS observed in this species. Silencing or inactivation of AN2 orthologs has been shown to yield a white-flowered phenotype in N. tabacum (Pattanaik et al. 2010), Petunia (Quattrocchio et al. 1999; Hoballah et al. 2007), and Antirrhinum (Schwinn et al. 2006). Taken together, these results suggest that lack of AN2 expression due to loss of the gene from the genome may be responsible for the white floral phenotype in N. sylvestris.

11.4.2 Competition Between FLS and DFR May Underlie the Light-Pink Floral Phenotype

Variation in pink floral phenotypes appears to be driven by complex dynamics across the flavonoid biosynthetic pathway, consistent with differences in the degree of pigmentation rather than presence or absence. In comparisons between light-pink and dark-pink flowers, the latter exhibit increased expression of DFR, ANS, AN1, and/or AN2, whereas light-pink flowers have higher levels of CHS and/or FLS. These contrasting patterns suggest that the flux of the pathway may be directed toward producing anthocyanins in dark-pink flowers, but shunted toward flavonols in light-pink flowers. Light-pink accessions displayed high FLS:DFR ratios at 60% of anthesis length, whereas dark-pink accessions did not. In pairwise comparisons between light-pink and dark-pink accessions, this high FLS:DFR ratio in light-pink flowers was attributable to lower expression levels of DFR, higher expression levels of FLS, or both. Across development, FLS expression levels were high in N. tomentosiformis, Chulumani, and both synthetic lines at 60% of anthesis, but these accessions display both light- and dark-pink phenotypes, suggesting that high FLS expression alone is not sufficient to produce a light-pink floral phenotype. DFR expression levels at 60% of anthesis length in the light-pink accessions Chulumani and QM24 were similar to those observed for N. sylvestris throughout development, suggesting that DFR activation is developmentally delayed until 85% of anthesis length in these accessions. In contrast, N. tomentosiformis, 095-55, and QM25 had higher levels of DFR expression than N. sylvestris at 60% of anthesis length, indicating that DFR activation had already occurred by 60% of anthesis length in these accessions. These results suggest that both high expression levels of FLS and low expression levels of DFR early in development, most likely due to delayed activation of DFR, may be required to generate the light-pink floral phenotype; future studies should test this hypothesis.

FLS and DFR are in competition for the same substrates, the dihydroflavonols dihydrokaempferol, dihydroquercetin, and dihydromyricetin, to produce flavonols and anthocyanins, respectively. This competition, however, involves more than only enzyme affinity for substrates. Both FLS and DFR have been shown to interact with CHS in vivo, and FLS may interfere with the DFR and CHS interaction (Crosby et al. 2011). In addition, a study in grape (Vitis vinifera) reported that flavonols can bind to DFR and inhibit its activity (Trabelsi et al. 2008). Several studies have shown that up-regulation or silencing of FLS can alter the visible floral color phenotype, despite its regulation of the production of flavonols, which are not visible to the human eye. Overexpression of FLS yields white or light-pink floral phenotypes in Mimulus (Yuan et al. 2016), Petunia (Sheehan et al. 2016), and N. tabacum (Luo et al. 2015). Similarly, silencing FLS increases anthocyanin content in Arabidopsis (Stracke et al. 2009), pink-flowered Petunia and N. tabacum (Holton et al. 1993), and white-flowered Petunia, although, in this study, anthocyanin pigmentation was only observed in floral buds and was not maintained at anthesis (Davies et al. 2003). This evidence supports our conclusion that differences between light- and dark-pink flower phenotypes may be attributable to competition between FLS and DFR, although several questions remain. Does FLS have a higher affinity for the shared substrates than DFR? If so, why does FLS not always outcompete DFR? How high does the FLS:DFR ratio need to be to tip the balance toward producing flavonols at the expense of anthocyanins? Further studies are required to address these questions.