Introduction

Transition to flowering is controlled by a combination of endogenous developmental signals and environmental cues, such as day length, light intensity and temperature. Some annual and biennial plants require exposure to cold temperatures to gain flowering competence in a process-termed vernalization. A central gene in the vernalization pathway in Arabidopsis is the MADS-box transcription factor FLOWERING LOCUS C (FLC) that acts as a repressor of flowering through repression of flowering promoters such as FLOWERING LOCUS T (FT) and SUPPRESSOR OF OVEREXPRESSION OF CONSTANS1 (SOC1) (Whittaker and Dean 2017). FLC is part of a small gene family that also includes several FLC-LIKE genes termed MADS AFFECTING FLOWERING 1–5 (MAF1–5) which, similar to FLC, act as flowering repressors (Scortecci et al. 2001; Ratcliffe et al. 2003). Major regulators of FLC include FRIGIDA and genes in the autonomous pathway that activate and repress FLC, respectively, through RNA-mediated chromatin-modification mechanisms, resulting in increased levels of FLC chromatin methylation (recently reviewed by Whittaker and Dean 2017).

During vernalization, FLC undergoes several epigenetic histone modifications, among them deposition of the repressive histone mark H3K27me3. H3K27me3 deposition is mediated by a protein complex, Polycomb Repressive Complex 2 (PRC2), associated with several PLANT HOMEODOMAIN (PHD) finger proteins from the VERNALIZATION INSENSITIVE 3 (VIN3) family (Sung and Amasino 2004; Kim and Sung 2013). The VIN3 family members in Arabidopsis encode DNA-binding proteins and include, in addition to VIN3, four VIN3-LIKE genes termed VERNALIZATION INSENSITIVE 3-LIKE 1–4 (VIL1–VIL4) (Sung et al. 2007). The VIL genes contain three domains: a PHD domain associated with chromatin modification, a VIN3-interacting domain (VID) and a fibronectin type III (FNIII) domain involved in protein–protein interactions (Sung et al. 2006; Fu et al. 2007). All members of the VIN3 family act as repressors of different sets of FLC family members in Arabidopsis in a target- and timing-specific manner during vernalization (Kim and Sung 2013). In Arabidopsis, VIL1 is functional in both the vernalization and photoperiod pathways. In the vernalization pathway, VIL1 and VIN3 regulate FLC and the FLC-clade member FLM, while in the photoperiod pathway VIL1 regulates FLM independent of VIN3 (Sung et al. 2006).

VIL genes exist in diverse flowering plants, both monocots and dicots (An et al. 2015). Three VIL genes have been identified in wheat, which requires vernalization for flowering (Fu et al. 2007). However, their involvement in controlling vernalization-mediated flowering is unknown. In the related species Brachypodium distachyon, mutation in BdVIL4 results in late flowering under cold temperatures and enhanced branching (An et al. 2015). Rice, which does not require vernalization for flowering, contains four VIL genes (Fu et al. 2007). Among them, OsVIL2 has been shown to epigenetically repress the flowering repressor OsLFL1, a putative B3 DNA-binding domain-containing transcription factor (Yang et al. 2013). Furthermore, mutation in OsVIL2 causes late flowering and additional shoot alterations, such as changes in leaf angle via altered cell division in the lamina joint, tiller number and leaf morphology, as well as sterility (Zhao et al. 2010). This indicates that in addition to flowering control, OsVIL2 regulates broad aspects of shoot architecture. Similar to OsVIL2, OsVIL3 epigenetically represses the flowering repressor OsLF, a bHLH transcription factor (Wang et al. 2013). Therefore, whereas VIL genes repress flowering in Arabidopsis and rice by a common chromatin-mediated modification mechanism, the specific target genes and pathways in which they operate diverge among species.

To dissect the genetic control of transition to flowering in pepper (Capsicum spp.), an important vegetable crop that is insensitive to vernalization and photoperiod for flowering induction, we screened an ethylmethane sulfonate (EMS)-mutagenized population for alterations in this trait. This screen identified several flowering-promoter genes, including CaFT-LIKE [the putative pepper ortholog of SINGLE FLOWER TRUSS (SFT) and FT, the tomato and Arabidopsis florigen genes, respectively; unpublished results], CaBLIND (CaBL) and CaJOINTLESS (CaJ) (Jeifetz et al. 2011; Cohen et al. 2012). CaBLIND acts as a regulator of auxiliary shoot branching and has a secondary role as a flowering promoter. CaJOINTLESS acts as a flowering promoter in the primary and sympodial shoots in addition to its effect on suppression of inflorescence development. Flowering-suppressor genes included FASCICULATE and CaAP2 (Elitzur et al. 2009; Borovsky et al. 2015). FASCICULATE is the ortholog of tomato SELF-PRUNING (SP) that acts as antagonistic to SFT in controlling flowering and shoot architecture, while CaAP2 is a homolog of a clade of six APETALA2 transcription factors that regulate transition to flowering in Arabidopsis. Furthermore, two genes, CaS and CaANANTHA, which are required for flower formation, were isolated (Lippman et al. 2008; Cohen et al. 2014). While FLC-LIKE and VIL genes are present in solanaceous genomes, their function in regulation of transition to flowering and shoot architecture has not been reported yet. In a search for additional pepper-flowering mutants, we identified a late-flowering mutant that exhibits strong pleiotropic effects on shoot architecture and was found to be disrupted at CaVIL1, the pepper ortholog of Arabidopsis VIL1. In the present study, we provide evidence that CaVIL1 is a flowering promoter and characterize its relationship with other pepper flowering genes.

Materials and methods

Plant material and mapping

The mutant E-2698 was isolated from an EMS-mutagenized population using Capsicum annuum cv. Maor as the wild-type (WT) parent (Paran et al. 2007). E-2698 was further backcrossed twice to ‘Maor’ and fixed by self-pollination. To map the mutation, we constructed an F2 population of 100 individuals from a cross of E-2698 × Capsicum frutescens BG 2816. For initial chromosomal assignment, we employed bulked segregant analysis (Michelmore et al. 1991) by constructing mutant and WT bulks, each containing 15 plants. The two bulks were screened with known COSII markers representing the 12 pepper chromosomes (https://solgenomics.net). After assignment of the mutation to chromosome 5, a larger F2 set of 400 individuals was evaluated for the mutant phenotype and 78 individuals were identified as mutants and used for fine mapping with additional markers from chromosome 5.

Generation of double mutants

Since E-2698 is male sterile, the double mutants E-2698 E-172 and E-2698 E-252 were generated through crosses between the heterozygous E-2698/+ and the corresponding mutants. E-172 is mutated in CaFT-LIKE, while E-252 is mutated in CaJOINTLESS (Cohen et al. 2012). The F2 population of each cross was screened for double mutants by PCR markers (Table S1).

Measurement of flowering time and leaf area

Flowering time was measured as the number of leaves on the primary stem until first flower. Fully opened leaves were scanned using an HP Scanjet 4890 PhotoScanner at 300 dpi. All scanned images were analyzed using Tomato Analyzer software version 3.0 for the leaf area measurement (Rodríguez et al. 2010).

Microscopy

For scanning electron microscopy (SEM) analysis, leaf disks were fixed in 70% ethanol, and dehydrated through an ethanol series (70, 90, 95 and 100%, 1 h each). Critical point drying was performed as previously described (Alvarez et al. 1992). SEM images were taken in a Hitachi S-3500 N instrument. For histological analysis, the tissues were fixed in FAA (1.85% formaldehyde, 5% glacial acetic acid, 63% ethanol, v/v), and dehydrated through an ethanol series (50, 70, 90, 95 and 100%, 1 h each). The ethanol was slowly replaced with Histoclear (K-CLEAR) and then with paraffin. The paraffin-saturated tissues were embedded in paraffin, sectioned in a microtome (Leica RM2245), and stained with Safranin O/Fast green. The stained sections were observed in a Leica IM1000 microscope, and digital images were taken with a CCD camera (DC2000, Leica, Germany).

RNA extraction and quantitative real-time PCR

Shoot apical meristems (SAMs) (approximately 3-mm-long tissue from the shoot apex) were collected for RNA extraction at the floral transition stages (4- and 6-expanded leaves stages in ‘Maor’ and E-2698, respectively). Three SAMs of the same genotype were used for one extraction. Total RNA was extracted from the tissues using the GeneElute™ Mammalian Total RNA Extraction Miniprep Kit (Sigma). Genomic DNA was removed by DNaseI (Sigma) treatment. Total RNA (500 ng) served for first-strand cDNA synthesis using Verso cDNA Synthesis Kit (Thermo Fisher Scientific, Massachusetts, USA). Quantitative real-time (qRT)-PCR was carried out in a total reaction volume of 15 µL using SYBR Premix Ex-Taq II (Takara) in a Rotor-Gene 6000™ thermal cycler (Corbett Research, Mortlake, Australia). At least three biological and two technical repeats were used for each sample. The primers used for the qRT-PCR are listed in Table S1. Results were analyzed using Rotor-Gene 6000 software 1.7 (Corbett). The relative expression levels of the genes were normalized against CaUbiquitin (DQ975458.1).

Phylogenetic analysis

The amino acid sequences encoded by the tomato and Arabidopsis genes were obtained from the public databases Sol Genomics Network (SGN) (https://solgenomics.net/) and The Arabidopsis Information Resource (TAIR) (https://www.arabidopsis.org/Blast/), respectively. Amino acid sequence alignments were performed with Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/) using the default settings. Evolutionary history was inferred using the maximum likelihood method, and the phylogenetic tree was prepared by bootstrap analysis with 1000 replicates using MEGA6 software (Tamura et al. 2013).

Results

E-2698 is a pepper EMS mutant with delayed transition to flowering

Phenotypic screening of the pepper EMS-mutagenized population for alterations in flowering time identified the late-flowering mutant E-2698. Compared to ‘Maor’ which flowers after 10.5 ± 0.2 leaves on the primary stem, E-2698 flowered after 19.2 ± 0.4 leaves (Fig. 1a, b). The sympodial units of the mutant may contain another 1 or 2 leaves as found for other late-flowering pepper mutants (Cohen et al. 2012). Analysis of the segregation pattern of the mutant phenotype in the M2 generation indicated monogenic recessive inheritance of the mutation.

Fig. 1
figure 1

Flowering time in wild-type cv. Maor and the late-flowering mutant E-2698. a Pictures of ‘Maor’ and E-2698, b flowering time measured as the number of leaves to first flower on the primary stem. ‘Maor’ and E-2698 flower after 10.6 ± 0.2 and 19.2 ± 0.4 leaves, respectively. Significant difference between ‘Maor’ and E-2698 was determined by Student’s t-test

Full size image

Mutation in E-2698 results in multiple pleiotropic effects on shoot and root structure

In addition to delayed flowering, E-2698 exhibited multiple developmental variations in shoot structure. The total plant height of the mutant was ~ 1.8 times greater than that of ‘Maor,’ whereas the primary stem length was 2.5-fold longer in the mutant. There was no significant difference in plant growth after bifurcation of the primary stem. The difference in length of the primary stem resulted from a greater number of leaves before the first flower, but also from ~ twofold longer internodes in the mutant. The diameter of the primary stem was also larger in the mutant compared to the WT (Table 1).

Table 1 Morphological parameters of ‘Maor’ and E-2698 shoot structure
Full size table

Leaves of the mutant were more than double the size of the WT leaves (Table 1), and they were markedly wrinkled due to a distorted venation pattern with twisted petioles (Fig. 2a, b). The mutant developed more lateral root branches than the WT, resulting in a bushier root system in the fully grown plants (Fig. 2c). E-2698 is male sterile, and the stamens dry up before the flower opens. The flowers were larger in the mutant, with an irregular number of floral organs and occasionally, one leaf-like sepal.

Fig. 2
figure 2

Leaf, root and lateral branching variations in E-2698 compared to ‘Maor.’ a The leaves of E-2698 are enlarged and wrinkled with a twisted petiole and distorted venation, b enlarged view of the mid-vein region on the adaxial side of the leaf of ‘Maor’ and E-2698, c roots of E-2698 have a bushy appearance due to a larger number of lateral roots compared to ‘Maor,’ d lateral branches developed in E-2698 and are indicated with white arrows, e the angle of lateral branching from the primary stem is significantly reduced in E-2698, f relative expression of CaBL in the shoot apical meristem is significantly higher in E-2698. Significant differences between ‘Maor’ and E-2698 were determined by Student’s t-test. The scale bars = 5 cm (a, c) and 2000 µm (b)

Full size image

In ‘Maor,’ lateral shoot branching is limited and axillary buds are markedly activated only after removal of the shoot apical meristem (SAM). In contrast, E-2698 developed several lateral shoots even without SAM removal, indicating weaker apical dominance in the mutant (Fig. 2d). Moreover, the axillary branches of E-2698 were restricted to the upper nodes of the mutant, i.e., the nodes above 8–10 leaves. There was no difference in the axillary branching pattern in the mutant compared to ‘Maor’ after removal of the apex. Furthermore, the angle of the axillary branch forming from the primary stem in E-2698 was almost half (17° ± 1.2) that of ‘Maor’ (36° ± 2.8) (Fig. 2e). To test the possible involvement of CaBL, a positive regulator of axillary branch development in pepper (Jeifetz et al. 2011), in the mutant branching phenotype, we checked its relative expression in the shoot apical meristem. The qRT-PCR analysis indicated 3.4-fold upregulation of CaBL in E-2698 compared to ‘Maor’ in accordance with the increased axillary branching in the mutant (Fig. 2f).

The E-2698 plant had a reduced vascular system in all of its parts. The stem and root sections showed very weak development of vascular bundles. Comparison of transverse sections of the primary stem from the topmost region showed an irregular stem shape and reduced lignification in the mutant (Fig. 3a). Similarly, the radial arrangement of the vascular bundles was highly distorted in the mutant root (Fig. 3b). A smaller size and staggered arrangement of the vascular bundles were observed in the petiole and leaf sections as well. Instead of a single mid-vein, E-2698 had multiple distorted mid-veins in the leaves. Although the petiole was flat in the mutant compared to ‘Maor,’ the mid-vein region was much thicker in the mutant, while the leaf lamina was thinner (Fig. 3c, d). Cell size was smaller in the mutant compared to the WT in all of the sections analyzed. Palisade mesophyll cells of the leaf were shorter and highly distorted in the mutant (Fig. 3e). SEM images show the smaller size and irregular shape of the epidermal cells of the mutant leaf compared to the WT (Fig. 3f, i). Similarly, stomatal size was smaller in the mutant (Fig. 3g). The number of cells and stomata per unit area was higher in the mutant (Fig. 3h, j).

Fig. 3
figure 3

Defective vascular development and cell development in E-2698. a Transverse sections of the primary stem from the topmost position showing defective vascular development and stem shape in E-2698, b transverse sections of the root showing irregular and defective vascular development in E-2698, c transverse sections of the leaf petiole showing flattened petiole and reduced vasculature in E-2698, d transverse sections of leaf lamina (LL) along with the mid-vein (MV) showing thin leaf lamina and multiple mid-veins in E-2698, e enlarged view of transverse sections of leaf lamina showing deformed palisade mesophyll (PM) cell arrangement and spongy mesophyll (SM) cells in E-2698, f SEM images of abaxial leaf surface showing smaller epidermal cells and stomata in E-2698, g, h stomatal size is reduced and stomatal number per unit area is higher in E-2698, i, j epidermal cell area on the abaxial leaf surface is reduced and number of cells per unit area is higher in E-2698. Sections in a, b, c, d and e are stained with Safranin O/Fast green. Scale bars = 200 µm (a), 1000 µm (b, c, d) and 50 µm (e, f). Asterisks indicate significant differences between ‘Maor’ and E-2698 by Student’s t-test (***P < 0.001, **P < 0.01, *P < 0.05)

Full size image

E-2698 is mutated in the pepper homolog of VIL1

To map the disrupted gene in E-2698, we constructed an F2 population from a cross of E-2698 x C. frutescens BG 2816. After phenotyping the population for flowering time (number of leaves to first flower on the primary stem), we constructed two bulks of mutant and WT phenotypes which were screened with COSII markers scattered throughout the genome. Initial genotyping located the mutation on chromosome 5 near COSII marker C2_At1g27385 (https://solgenomics.net). For fine mapping, we increased the population size and enriched the region with additional flanking markers using a panel of 78 mutants selected from the F2 population. By screening this panel, C2_At1g27385 (92,745,176–92,746,602 bp on chromosome 5, the Zunla genome v2.0, Qin et al. 2014) was found as the most closely linked marker, 1 cM away from the gene coding for the late-flowering phenotype. We searched this region for flowering-related genes based on the annotated genome sequence (https://solgenomics.net) and found complete linkage between the late-flowering phenotype and the gene Capana05g001237 (90,503,704–90,506,723 bp) on chromosome 5 coding for a VIN3-like 1 protein designated as CaVIL1. Sequence comparison of Capana05g001237 in ‘Maor’ and E-2698 revealed an early stop codon in the mutant (Fig. S1).

In addition to E-2698, we identified two other delayed-flowering mutants from the same EMS-mutagenized population—E-42 and E-2316—with pleiotropic phenotypes similar to that of E-2698. Sequence analysis of the open reading frames of the three mutants showed that the three mutations are allelic. The mutations in E-2698, E-2316 and E-42 were 1675C > T, 1639C > T and 1453C > T base substitutions, leading to truncation of the protein after 559 amino acids, 547 amino acids, and 485 amino acids, respectively, and elimination of the VID (Fig. S1).

A phylogenetic tree was constructed using the amino acid sequences of CaVIL1 homologs from pepper, tomato and Arabidopsis. In addition to CaVIL1, the small gene family in pepper contains CaVIL2 and CaVIN3. The tree showed greater similarity of CaVIL1 to VIL1 homologs from tomato and Arabidopsis than to other members of the CaVIL1 family in pepper (Fig. 4).

Fig. 4
figure 4

Phylogenetic tree of VIN3 family members in pepper, tomato and Arabidopsis constructed for the amino acid sequences of CaVIL1 homologs using MEGA6. The tree is drawn to scale, with branch lengths measured in number of substitutions per site

Full size image

CaVIL1 is expressed throughout plant development

To understand the importance of CaVIL1 in the development of pepper plant organs, we determined its expression in various tissues of ‘Maor’ by qRT-PCR. The gene was expressed in all of the examined tissues, suggesting a potential role in diverse plant developmental processes in accordance with the observed multiple pleiotropic phenotypic effects. The significant increase in expression of CaVIL1 at the SAM at the 4-expanded-leaf stage that corresponds to flower initiation in ‘Maor’ compared to the vegetative stage (2 expanded leaves) was in accordance with its role as a flowering promoter. The gene was expressed at a moderately high level in the sympodial and axillary meristems. The high expression in the flower bud suggests its role in flower development and floral organ formation (Fig. S2).

Genetic interaction of CaVIL1 with other flowering regulators

To elucidate the relationship of CaVIL1 with the flowering promoters CaFT and CaJ, E-2698 was crossed with Caft (late-flowering EMS mutant E-172 that is homologous to tomato sft, unpublished results) and Caj (E-252, Cohen et al. 2012). F2 populations from both crosses were constructed, genotyped for the mutations and analyzed for flowering time. The Cavil1 Caft and Cavil1 Caj double mutants flowered after 38 ± 0.38 and 55 ± 4.25 leaves, respectively (Fig. 5a–d). The effect of both double mutants was greater than the estimated sum of the effects of the individual mutants assuming complete additivity. The effect of each mutant was calculated by subtracting the mean of the wild-type phenotype from the mean of the mutant. This indicates a synergistic positive interaction of CaVIL1 with the two flowering-promoter genes. The shoot structure of both double mutants resembled that of Cavil1 (Fig. 5a, c), indicating epistasis of the Cavil1 shoot phenotype. The expression of CaFT was significantly lower in the expanded leaf of E-2698 compared to ‘Maor’ (Fig. 6a), indicating positive regulation of CaFT by CaVIL1. In contrast, the expression of CaJ in the SAM at flowering transition was significantly higher in E-2698 compared to ‘Maor’ (Fig. 6b). We also determined that expression of the floral repressor CaAP2 (Borovsky et al. 2015) in the SAM was upregulated in E-2698, indicating that CaVIL1 represses CaAP2 (Fig. 6c).

Fig. 5
figure 5

Flowering time of double mutants Cavil1 Caft and Cavil1 Caj. a, c Images of F2 progeny corresponding to single and double mutant genotypes, b, d flowering time (number of leaves to first flower on the primary stem) of F2 progeny corresponding to single and double mutant genotypes. Differences among means were determined by Tukey–Kramer range test at P < 0.05 and are indicated by different lowercase letters. Asterisks above the double mutant columns indicate significant differences (P < 0.0001) between the double mutants and the expected flowering time of the double mutants assuming complete additivity of the single mutant effects as determined by Student’s t-test. The effect of each mutant was calculated by subtracting the mean of the wild-type phenotype from the mean of the mutant. The number of F2 individuals in each genotypic class is indicated in parenthesis in b and d

Full size image
Fig. 6
figure 6

Interaction of CaVIL1 with CaFT, CaJ and CaAP2. Relative expression of a CaFT, b CaJ and c CaAP2 in E-2698 compared to ‘Maor,’ suggesting that CaVIL1 positively regulates CaFT and negatively regulates CaJ and CaAP2. Expression of CaFT was determined in expanded leaves at the stage of 6 leaves, while expression of CaJ and CaAP2 was determined at the SAM at flowering transition. Significant differences between ‘Maor’ and E-2698 were determined by Student’s t-test

Full size image

CaVIL1 represses the expression of FLC-LIKE genes in pepper

In Arabidopsis, VIN3 and VIL1 proteins interact to repress FLC through chromatin modification, enhancing vernalization-mediated flowering (Sung et al. 2007). Although FLC has not been found in the pepper genome, we identified a cluster of seven FLC-LIKE genes on chromosome 12, similar to the cluster of FLC-clade genes on chromosome 5 of Arabidopsis (Ratcliffe et al. 2001). The relatedness among pepper and Arabidopsis FLC-clade genes is presented in a phylogenetic tree that showed a clear grouping among pepper and Arabidopsis proteins (Fig. 7a). Arabidopsis MAF1–5 proteins grouped together with FLC in a subgroup of MIKC MADS-box proteins. Additional representatives from the other four subgroups of the MADS-box proteins (Parenicova et al. 2003) were used as outgroup and were separated from the FLC-clade members. Characteristic of MADS-box genes, Arabidopsis FLC and MAF has a MADS-box motif involved in DNA binding and a K-box motif involved in protein–protein interactions (Li et al. 2016). In pepper, the MADS-box domain is predicted in three genes in the FLC-LIKE clade, whereas the K-box domain is predicted in four genes within the clade (Fig. S3). Only Capana12g000913 contains the two domains, which indicates that most FLC-LIKE genes are likely pseudogenes. To determine whether the Cavil1 mutation affects the expression of the FLC-LIKE genes, we checked their expression in the SAM at flowering transition by qRT-PCR; four genes (Capana12g000929, Capana12g000901, Capana12g000894 and Capana12g000930) were significantly upregulated in E-2698 (Fig. 7b). The expression of Capana12g000913 and Capana12g000902 was undetectable, and that of Capana12g000908 did not differ from ‘Maor.’

Fig. 7
figure 7

Phylogenetic tree of FLC-LIKE family members and relationship between FLC-LIKE genes and CaVIL1. a Phylogenetic tree constructed for the amino acid sequences of FLC and FLC-LIKE genes in Arabidopsis (MAF1–5) and FLC-LIKE genes in pepper using MEGA6. Except for the FLC-clade members that belong to the MIKC subgroup of MADS-box genes in Arabidopsis, representative members from the other four subgroups (Parenicova et al. 2003) are included as an outgroup in the tree, b CaVIL1 represses the expression of FLC-LIKE genes. Expression was determined at the SAM at flowering transition in E-2698 compared to ‘Maor’ by qRT-PCR. Significant differences between ‘Maor’ and E-2698 were determined by Student’s t-test

Full size image

Discussion

Flowering-time regulation by CaVIL1 and its interaction with FLC-LIKE genes

This is the first report in the Solanaceae on the function of a VIN3 gene member in controlling flowering time in pepper. VIN3 family members were first characterized in Arabidopsis, for which loss of function resulted in loss of flowering response to vernalization and a late-flowering phenotype (Sung et al. 2007). Pepper CaVIL1 is part of a small gene family containing two VIL genes and a homolog of Arabidopsis AtVIN3 (Fig. 4). The interaction of AtVIL1 with AtVIN3 for FLC repression during vernalization was reported by Sung et al. (2007). They found that chromatin modification of FLC does not occur in either vin3 or vil1 mutants. Similar to Arabidopsis, CaVIL1 contains a VIN3-interacting domain as well as a PHD domain involved in chromatin modification (Fig. S1). However, because pepper flowering is not influenced by vernalization, we postulate that CaVIL1 interacts with CaVIN3 for chromatin modification of unknown targets in a vernalization-independent manner, as has been demonstrated for rice (Yang et al. 2013). Furthermore, unlike VIL1 that also acts in the photoperiod pathway in regulating flowering in Arabidopsis, pepper flowering is not affected by day length, and therefore, CaVIL1 is likely recruited by autonomous developmental pathways that control transition to flowering in this species.

In Arabidopsis, the vernalization response is mediated by repression of FLC. FLC genes have not been identified in tomato or pepper genomes. However, seven FLC-LIKE genes were clustered on pepper chromosome 12, most of them likely pseudogenes. The significantly higher expression of four FLC-LIKE genes in the Cavil1 mutant compared to ‘Maor’ indicates that CaVIL1 may regulate flowering time through repression of these genes. Transcription repression of these pseudogenes by CaVIL1 indicates their possible function in regulating pepper flowering. There is evidence that pseudogenes can interact with functional gene members and regulate their expression (Tutar 2012; Xiao et al. 2016). Further studies are needed to validate our hypothesis of the involvement of FLC-LIKE genes in flowering regulation of pepper. These might include the generation of knock-out lines of expressed pseudogene members, checking for interactions among CaVIL1, CaVIN3 and FLC-LIKE, and chromatin modification of FLC-LIKE proteins.

CaVIL1 interacts with other flowering regulators in pepper

To determine the relationship between CaVIL1 and the previously identified pepper flowering promoters CaFT and CaJ, we determined their expression in the Cavil1 background. We furthered constructed double mutants of CaVIL1 with CaFT and CaJ and tested their shoot architecture and flowering time. The extremely late flowering time of both double mutants indicated synergistic interactions of CaVIL1 with both genes (Fig. 5b, d). The expression of CaFT was significantly inhibited in Cavil1, indicating that CaVIL1 activates the expression of the putative pepper florigen. In contrast, the expression of CaJ was significantly higher in Cavil1, indicating that CaVIL1 suppresses its expression. As CaJ promotes flowering in pepper (Cohen et al. 2012), this result was unexpected and may indicate the involvement of additional unknown genes that interact with CaVIL1 and CaJ. Furthermore, our expression studies showed upregulation of the flowering repressor CaAP2 in the Cavil1 mutant, indicating repression of this gene by CaVIL1. Our results show that CaVIL1 is a major regulator of flowering and interacts with other flowering promoters and repressors in pepper, only some of whose identities are known. The exact pathway by which CaVIL1 operates to control transition to flowering in pepper, which has a sympodial shoot structure and does not respond to vernalization and photoperiod is partly known. To better decipher this pathway, additional studies, such as transcriptome analysis and chromatin immunoprecipitation sequencing (ChIP-Seq) of Cavil1, are required.

Pleiotropic effects of the Cavil1 mutation on shoot architecture

The E-2698 mutant is defective in many aspects of plant architectural development. Accordingly, CaVIL1 was expressed in all of the examined plant parts, suggesting its role in overall plant development. The PHD domain in CaVIL1 is expected to recognize histone marks of target genes, allowing interaction with chromatin-remodeling complexes for gene regulation. PHD proteins are involved in diverse developmental processes in plants, including flowering time, root development, embryonic meristem initiation, seed germination and pollen maturation (reviewed by Mouriz et al. 2015). In Arabidopsis, FLC members are known as targets of VIN3 and VIL proteins (Kim and Sung 2013). VIN3 homologs in rice, which does not respond to vernalization, are also involved in flowering regulation via histone modification (Yang et al. 2013). In pepper, the specific targets of CaVIL1 are not yet known; their identification will require the application of ChIP-Seq as indicated above.

CaVIL1 controls lateral branching development

The E-2698 mutant developed robust lateral branches mostly in the upper part of the shoot. This weak apical dominance was associated with high expression of CaBL in the mutant (Fig. 2). Since knock-out of CaBL leads to suppression of axillary bud development (Jeifetz et al. 2011), it is likely that increased expression of the gene will lead to release of axillary branching, as has been demonstrated in Arabidopsis (Muller et al. 2006). It has recently been shown that sugar demand at the apex, and not auxin, is the initial regulator of apical dominance, and that axillary bud release is correlated to sugar relocation to the axillary buds (Mason et al. 2014). We observed very little vascular development in the upper portion of the primary stem of E-2698. This might restrict sugar transport to the apical meristem, which is required for its active growth. This restricted transport may lead to the accumulation of sugars in the axillary buds resulting in their activation. Comparatively well-developed vasculature in the lower portion of the mutant stem may actively transport the sucrose synthesized from the lower leaves, and thus, there is no accumulation of sugar in the lower nodes and the axillary buds in that region remain inactive.

The decrease in the angle of the lateral branch in E-2698 could be attributed to alterations in auxin signaling and the magnitude of the antigravitropic offset component (Roychoudhry et al. 2013). A rice mutant leaf inclination2 (lc2) is mutated in a VIN3-LIKE homolog and is characterized by multiple shoot alterations, including a change in leaf inclination angle as a result of unequal cell division in the collar region (Zhao et al. 2010). Similar to E-2698 and the two additional allelic Cavil1 mutants, the lc2 phenotype is associated with loss of the VID domain, indicating its importance to the function of VIL genes. The expression of LC2 in rice was induced by application of several hormones—gibberellin, auxin and brassinosteroids. It is likely that unequal cell division is the reason for the leaf wrinkling and/or the decrease in the lateral branch angle in E-2698. Comparative transcriptome analysis and hormonal profiling of shoot tissues will provide a better understanding of hormonal involvement in the formation of the complex phenotypes attributed to CaVIL1.

Author contribution statement

IP designed the research and wrote the manuscript. VM, YB, IK and HZ conducted the experiments and analyzed the data. All authors read and approved the manuscript.