2015 Volume 84 Issue 3 Pages 261-268
Two types of floral morph, called thrum and pin, occur in heterostylous species. The thrum and pin flowers have shorter and longer styles, respectively. Heterostyly in Linum has been studied since Darwin’s era. The floral morph, self-incompatibility, and related phenotypes have been well characterized using a natural population, but genetic analysis using a segregated population has not been reported, and the mode of inheritance of heterostyly in Linum remains to be investigated. We prepared a segregated population by crossing thrum and pin flowers of Linum grandiflorum Desf. and investigated style and stamen lengths. On the basis of the style to stamen length ratio, the population could be divided into thrum and pin clearly at a ratio of 1:1. Style length of pins was 1.6 times longer than that of thrums. To investigate the factor regulating the difference in style length, we further measured the style cell length of thrums and pins. The style cell length of pins was longer than that of thrums, whose rate was comparable to the rate of the style length ratio between thrum and pin flowers. These findings indicate that the floral morph of heterostyly in Linum is controlled by a single diallelic locus, and that a difference in the cell expansion rate caused thrum and pin morphs in the style, such as in a typical heterostylous species. PCR genotyping showed that TSS1, an S candidate gene reported previously, cosegregated completely with the thrum phenotype, indicating strong linkage between the S locus and TSS1. Furthermore, three flower colors (red, pink, and white) were observed at a 1:2:1 ratio, and no white thrum flowers or red pin flowers were found in this population. These flower color phenotypes could also be controlled by a diallelic locus, whose two alleles (R and r) would be incompletely dominant, and therefore, may be linked to the S locus.
Self-incompatibility (SI) is a genetic system to reject “self” pollen and receive “non-self” pollen. This system is expected to prevent inbreeding and promote outcrossing, which is necessary to maintain genetic diversity in the population (de Nettancourt, 2001; Franklin-Tong, 2008). On the basis of the floral morph polymorphism, SI systems are classified into homostylous SI and heterostylous SI (heterostyly). Heterostyly in Primula and Fagopyrum, which are model species to study heterostyly, is controlled by a single diallelic locus, called the “S locus” (Barrett, 1992; McCubbin, 2008). Two types of flower, called thrum and pin, occur in heterostylous species. The style recognizes pollen from the same morph flowers as “self” and rejects it, and then fertilization is successful between different floral morphs. In most heterostylous species, thrum is heterozygous (S/s) and has a short-style flower with long stamens. Pin is homozygous recessive (s/s) and has a long style. The S locus controls not only the SI phenotype but also the floral morph; thus, it is expected that the same gene(s) controls both SI and floral morph phenotypes, or that the two phenotypes are controlled by different genes linked completely at the S locus. In Primula, genetic studies demonstrated that the S locus of Primula comprises at least four genes, S, G, P, and A, which determine the SI phenotype, the style length, the pollen size, and the anther height, respectively (Barrett, 1992). The tight linkage and convergent evolution of SI and floral characteristics in heterostylous plants have been attracting the interest of geneticists and evolutionary biologists since the age of Charles Darwin (Darwin, 1877). Many attempts to isolate the S gene have been reported in Primula, Fagopyrum, and Turnera, but the S gene has never been identified (Labonne and Shore, 2011; Li et al., 2011; Manfield et al., 2005; Yasui et al., 2012). We have also studied heterostyly in Linum in order to isolate the S gene and reported a good S gene candidate (Ushijima et al., 2012).
The genus Linum consists of more than 180 species, with members distributed throughout the temperate and subtropical regions of the world (McDill et al., 2009; Muir and Westcott, 2012). Some species have been domesticated for centuries. L. usitatissimum L. and L. perenne L. are used as sources of linen and linseed oil. Some of these species are garden ornamentals because of their brightly colored and attractive flowers, such as the red-flowered L. grandiflorum Desf., blue-flowered L. perenne, and yellow-flowered L. flavum L. Furthermore, Linum is historically important for studying heterostyly. Darwin characterized the relationship between the floral morph and fertilization in L. grandiflorum and L. perenne (Darwin, 1863, 1877). We previously isolated an S gene candidate called the thrum style-specific (TSS1) gene from L. grandiflorum (Ushijima et al., 2012). TSS1 is expressed in the thrum style only and is found in thrum genomes. Half of all thrum pollen grains possess the TSS1 gene. These findings indicate that TSS1 is linked to the S locus and thrum is heterozygous (S/s), but no genetic evidence exists because of the lack of a segregated population. Furthermore, genetic control of heterostyly in Linum has never been demonstrated and is unclear, although the 1:1 morph style frequency ratio in a natural population (Wolfe, 2001) indicates single diallelic locus regulation, as shown in typical heterostyly species. All circumstantial evidence suggested that the genetic control and mode of inheritance of heterostyly in Linum are similar to those in other heterostylous species. However, the dominance of pin was also observed in Hypericum and Armeria (Baker, 1966; Ornduff, 1979), in which pin is heterozygous (S/s) and thrum is homozygous recessive (s/s). Considering this exception, genetic analysis using a segregated population is required to characterize the genetic control of heterostyly in Linum.
It is essential to elucidate genetic control and the mode of inheritance of heterostyly in Linum in order to attempt to isolate the S gene. Unlike in other species, stamens show no or small length differences between thrums and pins (Barrett, 1992), and intermediate style-length plants have been observed (Wolfe, 2001), which makes it difficult to determine flower type. A precise definition of Linum floral morphs will be required to conduct precise phenotyping for genetic study. In this study, we prepared a segregated population and characterized the thrum and pin floral morph structural features, the mode of inheritance of the floral morph in Linum, and the linkage between the S locus and the TSS1 gene. Characterization of the population showed that the floral morph of heterostyly in L. grandiflorum was controlled by a single diallelic locus, and that TSS1 was linked to the S locus. Furthermore, petal pigmentation polymorphism, which also seemed to be controlled by a single diallelic locus, was observed in this population.
Two L. grandiflorum cultivars, ‘Scarlet flax’ (red flowers) and ‘Bright Eye’ (white flowers), were cultivated in an experimental field at Yamagata University. Two pink-flowered progeny with different style lengths were obtained from a cross of these cultivars. The pink-flowered pin (plant no. A) and the pink-flowered thrum (plant no. B) were crossed reciprocally, and a segregated population, consisting of 51 plants designated as the AB14 F1 population, was obtained.
Measurement of floral organ length and petal color scoringThree to five buds/flowers were collected from all AB14 F1 population segregates at two stages: one stage was 1 day before flowering (day −1) and the other was the day of flowering (day 0). Style and stamen lengths were measured with digital vernier calipers, and petal color was scored.
Measurement of style cell lengthWe measured cell lengths of the thrum and pin styles according to Chen et al. (2007), with slight modifications. Pistils were excised fresh, fixed in formaldehyde:acetic acid:80% ethanol (FAE = 1:1:8) for 1 day, and stored at 4°C until use. The fixed pistils were transferred to freshly prepared 75% ethanol and washed three times within 24 h. The washed pistils were rehydrated in 1 M Tris-HCl buffer (pH 9.0) for 1 day and stained with Calcofluor white solution (Sigma-Aldrich Japan, Tokyo, Japan), which was diluted to 1:10 with 1 M Tris-HCl buffer (pH 9.0) at room temperature. After staining, the pistils were transferred to water in a laboratory dish, observed, and photographed with a fluorescence stereoscope (MVX10; Olympus, Tokyo, Japan). Epidermal cell length was measured using ImageJ software (National Institutes of Health, Bethesda, MD, USA).
Isolation of DNA and polymerase chain reaction (PCR) genotypingFloral organs and leaves were collected from AB14 F1 population plants and stored at −80°C until use. Genomic DNA was isolated from anthers with NucleoSpin TriPrep (Macherey-Nagel, Duren, Germany) or from leaves with NucleoSpin Plant II (Macherey-Nagel). The TSS1 fragment was amplified from genomic DNA with the LgSSH48qtF- (5'-CAT CAT CGC CAA CCA AAG ATC-3') and LgSSH48qtR-specific primer set (5'-TGG CAG CCA ATG AAC CAG A-3'; Ushijima et al., 2012). Actin was amplified as an internal control with the LgACTqtF1 (5'-GAT CTG GCA TCA CAC GTT CTA C-3') and LgACTqtR1 primer set (5'-TGG GTC ATC TTC TCA CGA TTG-3'). PCR was performed with ExTaq (Takara Bio, Otsu, Japan) using a program of 30 cycles at 94°C for 10 s, 60°C for 30 s, initial denaturing at 94°C for 2.5 min, and final extension at 60°C for 1 min. PCR products were separated on a 2.0% agarose gel and stained with SYBR Green I (Takara Bio).
Style and stamen lengths of segregates in the AB14 F1 population were measured at two developmental stages (days −1 and 0) to define structural differences between thrums and pins (Table 1). The distribution histograms are shown in Figure 1. Two peaks were found in the style length histograms. Style length of the short-style population was 2.5–4.5 mm on day −1, and that of the long-style population was 5.0–8.5 mm (Fig. 1A). This biased distribution was almost completely maintained after flowering (day 0; Fig. 1D). In contrast, only one peak was detected in the stamen length histograms (Fig. 1B, E). This distribution pattern indicated that stamen length did not differ between thrums and pins, unlike style length. In most heterostylous species of Primula, Fagopyrum, and Turnera, morphological polymorphism is observed not only in style length but also in stamen length (McCubbin, 2008). However, stamen structure is monomorphic in Linum (Barrett, 1992; Darwin, 1877), and our result supports this finding.
AB14 F1 segregating population used in this study.
Distribution histograms for the style to stamen length ratio. Style and stamen lengths were measured 1 day before flowering (day −1; A–C) and the day of flowering (day 0; D–F). Three to five biological replicates were used for the plots.
Four plants showed style lengths that were intermediate between the two distribution peaks on day 0, ranging from 6.0 to 7.5 mm. As no significant polymorphism was detected in stamen length (Fig. 1B, E), it was difficult to distinguish intermediate plants as thrum or pin based only on style length. We then calculated the style to stamen length ratio (Fig. 1C, F). The distribution histogram for the style to stamen length ratio was divided clearly into two peaks on day 0, and the boundary was 0.5–0.6. The same distribution pattern was observed for the style to stamen length ratio on day −1. Therefore, we defined the shorter style-length plants (ratio < 0.5) as thrum, based on the style to stamen length ratio on day 0, and the longer ones (ratio > 0.6) as pin.
The average floral organ lengths and ratios are summarized in Figure 2 and Table 2. The Student’s t-test result supported the significant difference found in style lengths between thrums and pins on both day −1 and day 0 (Fig. 2). In contrast, no significant differences in stamen length were found, as shown in the distribution histograms. Twenty-six and 25 AB14 F1 population plants were scored as thrum and pin, respectively, and the thrum to pin ratio was 1:1. This segregation indicated that the floral morph of heterostyly in Linum was also controlled by a single diallelic locus, as in other typical heterostylous species of Primula and Fagopyrum (Barrett, 1992; McCubbin, 2008). The thrum is heterozygous (S/s) and the pin is homozygous recessive (s/s) in other species. Therefore, the Linum thrums were also expected to be heterozygous and possess a dominant S allele. However, our genetic study with the AB14 F1 population could not determine whether the thrum was heterozygous (S/s) because homozygous dominant (S/S) was essential, and the S/S homozygote could not be generated due to the robust SI of Linum.
Average style and stamen lengths. Gray bars denote the average length of thrum organs and white bars denote that of pins. Asterisk (*) indicates a highly significant difference from thrum (P < 0.01).
Average length and ratio of floral organs.
In this study, we investigated and characterized the floral morph phenotype but not the SI phenotype. Considering the tight linkage of the S locus genes (Barrett, 1992; McCubbin, 2008), it is unlikely but cannot be ruled out that recombination occurs between genes controlling SI and floral morph phenotypes, if two phenotypes are regulated by different genes at the S locus in Linum. Investigation of the SI phenotype may be required in future study in order to isolate the genes controlling SI.
Cell elongation rate causes the style length differenceCell lengths of the top, middle, and bottom style portions were measured in order to investigate further the differences between the thrum and pin styles (Fig. 3). The boxplot of style cell length is shown in Figure 4. Cell lengths of the pins in all three portions were longer than those of thrums. The ratios (pin to thrum) of the top, middle, and bottom portions were approximately 1.5, 1.4, and 1.3, respectively (Table 3). Average style length of pins was 1.5 times longer than that of thrums (Table 2), which was comparable to the cell length ratios. The comparable pin to thrum ratios between the organ and cell lengths indicated that the differences in thrum and pin style lengths were caused by a difference in the cell expansion rate, but not that in cell division. The longer cells of the pin style have also been observed in Primula (Heslop-Harrison et al., 1981; Manfield et al., 2005), indicating that regulation of cell expansion is a key step to determine style length in Primula and Linum. Although the primary regulating factor is unclear and the factors may differ between the two species, cell wall modifying enzymes may function downstream of the style length determination pathway and be shared between species.
Comparison of style cell length. Style cells were stained with Calcofluor white (A, B). Cell lengths of the top (C, D), middle (E, F), and bottom (G, H) portions were photographed with a fluorescence stereoscope and measured using ImageJ software. The length distribution is shown in Figure 4, and the average lengths are summarized in Table 3.
Boxplot of style cell length. Cell length distributions are shown by the boxplots. Gray boxes denote the style cell length of the thrum and white boxes denote that for the pin.
Average length of stylar cell (μm).
We isolated TSS1 previously, but the association between the S locus and TSS1 has not been demonstrated genetically because of the lack of a segregated population (Ushijima et al., 2012). We used 25 thrums and 25 pins from the AB14 F1 population generated in this study and performed PCR with TSS1-specific primers. As shown in Figure 5 and Table 1, the TSS1 fragment was amplified from all thrums, but no such fragment was found in pins. The thrum-specific amplification of TSS1 suggests that TSS1 is cosegregated completely with the thrum phenotype in this population, and that TSS1 is linked to the S locus. This result also suggests that the thrum is S/s heterozygous. The 1:1 segregation of the two floral morphs indicates that one is heterozygous (S/s) and the other is homozygous recessive (s/s). Thrums and pins share at least one recessive s allele, but an s allele-specific fragment was not detected as a floral morph-specific band by genotyping in the population used in this study. Therefore, the floral morph-specific fragment, such as TSS1, would be derived from the dominant S haplotype, indicating that the thrum is heterozygous, and the pin is homozygous recessive.
PCR genotyping for the thrum style-specific gene (TSS1). TSS1 genotypes of parents (A and B) and the AB14 F1 population segregates were determined by PCR. The actin gene was used as an internal control. PCR fragments were electrophoresed on a 2.0% agarose gel and stained with SYBR Green I. The A and B phenotypes are for pin and thrum, respectively. P and T denote pin and thrum, respectively. The electrophoresis pattern of the representative segregates is shown in this figure. All genotyping results are listed in Table 1.
The three flower colors of red, white, and pink were found in the AB14 F1 population (Fig. 6A). Although the base area of the petal (the center of the flower) was red-pigmented in all flowers, the color of other areas was polymorphic. The factor(s) regulating this petal pigmentation could function in the style (Fig. 6B). Red pigmentation was observed in the white flowers at the bottom of the style, similar to the other flowers, but the top and middle portions of the style were white. In contrast, the entire style from red and pink flowers was red-pigmented.
Petal and style color differences in the AB14 F1 population. Three flower colors were observed in this population (A). Similar pigmentation was found for the style color (B).
The AB14 F1 population used in this study was the F1 hybrid from a cross of a pink-flowered thrum and a pink-flowered pin, whose parents had red thrum flowers and white pin flowers. There were 11, 30, and 10 red, pink, and white flowers, respectively (Table 4). A chi-square test indicated that the petal color segregation ratio was 1:2:1, showing that petal color was also controlled by a single diallelic locus, and that the two alleles (designed R and r) seemed to be typically incompletely dominant. In this case, red and white flowers are homozygous (R/R and r/r, respectively), and pink flower is heterozygous (R/r). Joshi et al. (1961) also reported petal color segregation from a cross of red and white flowers in L. grandiflorum and observed the partial dominant petal pigmentation between two flowers, as in this study. In contrast, Lyakh (2013) generated and characterized the segregated populations from crosses between crimson (red) and pink flowers. This author assumed not only a crimson flower allele but also a pink one, and demonstrated that the crimson allele was completely dominant over the pink one. Although the level of genetic diversity among L. grandiflorum species was not clear, the differences of the genetic background might have led to the conflicting results. It is suggested that a locus other than the R locus defined in this study is related to the petal pigmentation in Linum.
Flower color frequencies in AB14 F1 population.
Incompletely dominant petal pigmentation has been reported and well characterized in snapdragon (Niv locus) and morning glory (A locus) (Bollmann et al., 1991; Coen and Carpentaer, 1988; Johzuka-Hisatomi et al., 2010). Both the Niv and the A loci encode chalcone synthase (CHS). Petals homozygous for the wild-type allele (Niv+/Niv+ or A/A) synthesize anthocyanin and express the red pigmentation phenotype because a sufficient amount of the CHS transcript is transcribed in the cell. The mutated alleles (niv-525 or a), in which the transportable element is inserted, result in no anthocyanin synthesis and white flowers. The heterozygote for wild-type and mutant alleles (Niv+/niv-525 or A/a) may synthesize anthocyanin, but less anthocyanin is synthesized than that synthesized by homozygotes for the wild-type allele. These plants show a pigmentation phenotype intermediate between those of homozygotes. For example, Ipomea segregates from a heterozygote (A/a) are consistently wild-type A/A with red flowers, A/a heterozygous plants have pale red flowers, and a/a mutants have white flowers at a 1:2:1 ratio. This intermediate pigmentation is caused by a decrease in the level of the CHS transcript, whose expression level in A/a petals is about half of that in A/A petals (Bollmann et al., 1991; Coen and Carpentaer, 1988; Johzuka-Hisatomi et al., 2010). An anthocyanidin triglycoside has been isolated from L. grandiflorum red petals as the main anthocyanin, and the pigment structure is delphinidin 3-O-3-xylosylrutinoside 78 (Toki et al., 1995). The base part of the white flower petal was red, indicating that the anthocyanin synthesis pathway functioned in white flowers. Expression of the structural genes encoding enzymes involved in the anthocyanin synthesis pathway, such as CHS, DFR, and F3H, is regulated by specific transcription factors, such as MYB (Nishijima et al., 2013; Rahim et al., 2014; Vimolmangkang et al., 2013). These results suggest that the genes regulating the expression of the structural genes, such as transcription factors, are located at the R locus in L. grandiflorum. Alternatively, some structural genes are members of a multigene family and the regulation of gene expression has diverged. Eight CHS genes of pea were differentially expressed during development, in response to various environmental cues, elicitor induction, and at organs (Harker et al., 1990). GUS promoter assay of Brassica CHS genes showed that CHS1 is expressed in petal strongly, but the expression of CHS3 is found in the leaf vein only (Zhou et al., 2013). These facts suggest another possibility that the mutation in the structural gene, as in the case of snapdragon and morning glory, caused defective anthocyanin synthesis and white petal if the R allele only functions at the tip of the petal.
Pink flowers were found in both thrums and pins. In contrast, red flowers were found in only thrums but not in pins, and white flowers were found in pins. This bias indicates that the petal pigmentation phenotype cosegregates with the heterostylous phenotype, and that the locus regulating petal pigmentation is closely linked to the S locus. If the genotypes of the original parents are SR/sR (red-flowered thrum) and sr/sr (white-flowered pin), respectively, all hybrids are expected to be pink-flowered (SR/sr or sR/sr) because they would all be R/r heterozygotes. Segregates from a cross between a pink-flowered thrum and a pin are expected to be SR/sR plants with short styles and red flowers, SR/sr plants with short styles and pink flowers, sR/sr plants with long styles and pink flowers, and sr/sr with long styles and white flowers at a 1:1:1:1 ratio. In contrast, no white flowers with short styles (Sr/sr) and no red flowers with long styles (sR/sR) should be obtained from this population unless recombination between the S and R loci occurs. The segregation ratio (11:15:15:10) observed in this study fits this assumption (Table 4), and the R locus could be linked to the S locus. Alternatively, the R gene is not identical to the S gene because pink-flowered thrums and pins share a dominant R allele. Although the gene regulating petal pigmentation is not the S gene, petal pigmentation would be an obvious marker to help in the precise phenotyping of flower type.
In conclusion, we generated a segregated population of L. grandiflorum and defined the precise structural features of thrums and pins. Our definitions were based on the style to stamen length ratio, making it possible to divide the segregates into two clear morphs. Genetic study demonstrated that the floral morph of heterostyly in Linum was controlled by a single diallelic locus, and that TSS1 was linked to the S locus. In addition to the thrum-specific expression pattern and other features reported previously (Ushijima et al., 2012), linkage between TSS1 and the S locus supports that TSS1 is a good candidate for the stylous S gene. The SI phenotype of the style and pollen is controlled by different genes in homomorphic SI species, and complete linkage between floral morph and SI phenotype in heterostylous species indicates that two or more genes are located at the S locus and regulate the heterostylous phenotypes (Franklin-Tong, 2008). We have isolated one candidate S gene for the style, but no candidate for the pollen S gene was found. The segregated population generated in this study, whose segregates were phenotyped precisely, will be useful for investigating the pollen S gene and other candidate genes.
