Anthocyanins, a class of flavonoid secondary metabolite compounds (Liu et al., 2018), are responsible for providing orange to blue colours in plant tissues, and their biosynthetic and regulatory mechanisms have been widely characterized (Zhao and Tao, 2015). However, there is still debate on the mechanism of anthocyanin transport from the ER to the vacuole (Poustka et al., 2007; Saito et al., 2013). Strong evidence for the involvement of transport proteins (Goodman et al., 2004; Francisco et al., 2013), with a special role of GST enzymes, has been reported in several plant species (Alfenito et al.; Larsen et al., 2003; Kitamura et al., 2004). In the present study, we demonstrated that the Bract1 gene functions as an anthocyanin transporter in poinsettia and that a highly mutable repeat in its coding region leads to frequent deletions and therefore to a colour-deficient phenotype.
Bract1 is a functional GST gene related to anthocyanin transport in poinsettia
GSTs are a large and diverse group of enzymes with multifunctional roles, especially in the detoxification of xenobiotics as well as in responses to biotic and abiotic stresses (Agrawal et al., 2002; Dixon et al., 2010). The classification of GSTs is based on sequence conservation, genomic organization, and physiochemical properties, among other features (Edwards et al., 2000; Lallement et al., 2014; Islam et al., 2019). Based on our previous study (Vilperte et al., 2019), we identified 95 GST genes in poinsettia and phylogenetically classified them into nine different classes based on their similarity with known Arabidopsis GSTs (Fig. 4). To date, 14 GST classes have been identified in plants: tau (U), phi (F), lambda (L), DHAR, theta (T), zeta (Z), EF1Bγ, TCHQD, microsomal prostaglandin E-synthase type 2 (mPGES-2), GHR, metaxin, Ure2p, hemerythrin (H) and iota (I) (reviewed by (Lallement et al., 2014)).
A large number of GSTs have been identified in plant species, such as 49 in Capsella rubella (He et al., 2016), 55 in Arabidopsis (Dixon and Edwards, 2010), 61 in Citrus (Licciardello et al., 2014) and 139 in L. chinensis (Hu et al., 2016). Bract1 clusters with high bootstrap support with anthocyanin-related GSTs from other species (e.g., AtTT19, PhAN9 and VvGST4), with all of these GSTs belonging to the phi class. Known anthocyanin-related GSTs belong almost exclusively to the phi class, except for Bronze-2 from maize, which belongs to the tau class (Marrs et al., 1995). Further support for Bract1 being a member of the phi-type plant GST genes is provided by the presence of two introns as a characteristic of this group of genes, such as AN9 from petunia and TT19 from Arabidopsis (Alfenito et al.; Mueller et al., 2000).
Complementation studies using Arabidopsis tt19 mutants have been widely applied as proof of concept for the function of GSTs as anthocyanin transporters (Alfenito et al.; Mueller et al., 2000; Kitamura et al., 2012; Hu et al., 2016; Pérez-Díaz et al., 2016; Jiang et al., 2019; Kou et al., 2019; Liu et al., 2019). Due to the high amino acid conservation of GST enzymes involved in flavonoid accumulation among species (Zhao, 2015), they can complement each other’s anthocyanin-deficient mutants (Alfenito et al.; Larsen et al., 2003). However, similar to our observation for Bract1, not all of these genes complemented both the shoot and seed phenotypes (Luo et al., 2018; Jiang et al., 2019). A direct complementation of poinsettia white mutants with the functional Bract1 would ultimately prove its function in bract colouration. However, neither Agrobacterium-mediated infiltration nor biolistic particle delivery system (a.k.a. gene gun) were successful for transient expression studies (data not shown). Stable transformation in poinsettia have been done using electrophoresis-based methods (Vik et al., 2001; Clarke et al., 2006), but no stable transgenic poinsettia was obtained. Successful stable transformation via Agrobacterium-mediated infiltration has been previously achieved, but the process is time-consuming (Clarke et al., 2008). Attempts to perform stable transformation of poinsettia with Bract1 alleles will bridge the current knowledge gap but are out of the scope of the present study.
A loss-of-function mutation in Bract1 is the cause of the “white paradox” in poinsettia
Based on our results, we hypothesize that deletion of one unit of the repeat in the Bract1 gene is responsible for most of the white genotypes in poinsettia. This hypothesis is strongly supported by the evidence that the tt19/35S::Bract1_mut mutant was not able to complement the anthocyanin phenotype in the Arabidopsis tt19 mutant, unlike the tt19/35S::Bract1 mutant. Mutations in GSTs leading to colourless phenotypes have been previously reported. A mutation in the fl3 gene in carnation leads to a light pink phenotype, but a brighter phenotype is observed upon complementation by petunia AN9 and maize Bz2 (Larsen et al., 2003). In peach, four alleles of a GST gene (Riant) were identified, with two of them containing frameshift mutations and unable to complement the Arabidopsis tt19 phenotype. Varieties containing copies of the mutated alleles in a homozygous state showed flowers with white variegated phenotypes (Cheng et al., 2015). Last, a single-nucleotide polymorphism (SNP) in the strawberry RAP gene, leading to a premature stop codon, results in a mutant with green petioles and leaves. The non-functional rap gene was not able to complement Arabidopsis tt19, while wild-type RAP was successful (Luo et al., 2018).
In our analyses, all six independently generated white mutants of red varieties displayed the same deletion of a 4 bp repeat in Bract1, whereas the original varieties all contained a fully functional copy of the gene. In addition, co-segregation of the deletion with the white phenotype was observed in a segregated population of 190 progeny. Furthermore, a novel mutation leading to a homozygous recessive allele of Bract1 among 184 samples obtained from irradiated cuttings of the heterozygous line SK183 led to a white phenotype (Table 3). Altogether, the results of this study present strong evidence that the four-base deletion in Bract1 is the cause of the red-to-white shift in the poinsettia varieties analysed here. However, as anthocyanin biosynthesis involves several steps, other regulatory and structural genes might give rise to white mutants as well, as has been shown in numerous other examples (Koseki et al., 2005; Morita et al., 2012; Ben-Simhon et al., 2015; Luo et al., 2016). We did not detect these genes in our current plant material perhaps due to the much higher mutation rate of the Bract1 gene than of less mutable genes.
Table 3
List of poinsettia varieties used in the present study.
Variety ID | Variety name | Bract colour | Observation |
1 | Christmas Feelings | Red | |
2 | Christmas Glory | Red | |
3 | Joy | Red | |
4 | Bravo | Red | |
5 | Titan | Red | |
6 | SK130 | Red | |
7 | Christmas Feelings Pearl | White | Mutation from Chr. Feelings |
8 | Christmas Glory White | White | Mutation from Chr. Glory |
9 | Joy White | White | Mutation from Joy |
10 | Bravo White | White | Mutation from Bravo |
11 | Titan White | White | Mutation from Titan |
12 | SK130 White | White | Mutation from SK130 |
13 | Vintage | Red | |
14 | Christmas Aurora | Red | |
15 | Happy Day | Red | |
16 | Tabalunga | Red | |
17 | Christmas Day | Red | |
18 | Christmas Eve | Red | |
19 | Noel | Red | |
20 | Valentino | Red | |
21 | Prestige Red | Red | |
22 | Christmas Cracker | Red | |
Bract1 contains a short highly mutable four-base repeat
Upon X-ray treatment, red poinsettia plants produce progeny bearing white phenotypes with high frequencies, often based on only 10 irradiated cuttings (von Tubeuf, Selecta One, pers. comm., Selecta One). This phenomenon is associated with a deletion in a short repeat in the Bract1 gene of white mutants in a homozygous state. The mutations in all six independent mutant pairs that we detected are exactly identical, which indicates that the X-ray treatment did not directly cause the mutation but rather led to changes indirectly by stimulating the DNA repair mechanisms via replication errors, by increasing recombination or by the other mechanisms discussed for mutations in repeat sequences (Pearson et al., 2005). The possible involvement of replication-based errors is supported by our observation that upon amplification of the repeat via standard PCR from cloned Bract1 wild-type or mutant allele, a low level of variants carrying four-base indels can always be detected (data not shown).
Radiation is frequently used as a tool for mutagenic breeding in poinsettia. In contrast to ethyl-methanesulphonate (EMS)-based chemical mutagenesis, which produces point mutations with high frequency (Greene et al., 2003), ionizing radiation (e.g., X-rays and γ-rays) induces DNA oxidative damage, such as double-strand breaks (DSBs), base substitutions, deletions and chromosomal alterations, at a lower frequency, frequently resulting in loss of gene function (Morita et al., 2009; Kazama et al., 2011; Jo and Kim, 2019). SSRs are among the most variable types of repetitive sequences in the genome (Ellegren, 2004). Studies have shown that SSR instability increases with plant development (Golubov et al., 2010) and abiotic stress (Yao and Kovalchuk, 2011). This might be another explanation for the frequent observation of repeat changes in the Bract1 gene after X-ray irradiation, although the small number of repeats (i.e., three) of four base pairs each does not fit the most widely applied criteria used to define SSRs, which usually focus on sequences with a larger number of repeats.
However, little information about the genetics and dynamics is available for short repeats. A majority of studies compared historical events for mostly shorter SSRs (2 and 3 bp repeats with larger repeat numbers) in present-day populations or the dynamic repeats responsible for human diseases (mostly trinucleotide repeats), which usually display effects beyond those of large numbers of repeats (> 30 repeats (Pearson et al., 2005))
Our observation that a large number of mutation events could be observed in the side shoots of ten irradiated plants indicates an unusually high mutation rate, which is in contrast to the few reports in which exact mutation rates have been reported for vegetatively propagated crops (Schum and Preil, 1998). In one example, the woody ornamental plant Tibouchina urvelliana was irradiated three independent times with a 45 Gy dose, resulting in 0.06% dwarf mutants each time (Schum and Preil, 1998). However, several authors reported that the radiosensitivity of vegetative tissues varies greatly among species and tissues (Esnault et al., 2010), so exact comparative estimations of mutation frequencies have a very limited accuracy among species and conditions. However, experiments with transgenic Arabidopsis lines harbouring constructs designed to analyse restoration of GUS open reading frames by either recombination or by restoring in-frame translation by mutations in SSRs demonstrate the occurrence of easily detectable numbers of somatic mutation events (Golubov et al., 2010; Yao and Kovalchuk, 2011). Together with the careful selection of side shoots after X-ray irradiation of poinsettia, this finding may explain the high rate of recessive mutations detected here.
In this study, we showed that the poinsettia Bract1 gene is an active GST gene involved in the expression of anthocyanins in poinsettia bracts. Furthermore, a 4 bp deletion in a short repeat within the coding region of Bract1 is the most likely cause of many mutations that lead to a white bract colour. This mutation occurs with an unusually high frequency and is presumably an indirect effect of X-ray mutagenesis. Future analyses using mutagenesis in transgenic Arabidopsis lines harbouring Bract1 might help elucidate the causes of the high instability of this repeat. Moreover, this result might also serve as a reference for the study of other repeat-containing structural genes as potential mutational hot spots in plant genomes.