The ethylene receptors CpETR1A and CpETR2B cooperate in the control of sex determination in Cucurbita pepo

Abstract

High-throughput screening of an ethyl methanesulfonate-generated mutant collection of Cucurbita pepo using the ethylene triple-response test resulted in the identification of two semi-dominant ethylene-insensitive mutants: etr1a and etr2b. Both mutations altered sex determination mechanisms, promoting conversion of female into bisexual or hermaphrodite flowers, and monoecy into andromonoecy, thereby delaying the transition to female flowering and reducing the number of pistillate flowers per plant. The mutations also altered the growth rate and maturity of petals and carpels in pistillate flowers, lengthening the time required for flowers to reach anthesis, as well as stimulating the growth rate of ovaries and the parthenocarpic development of fruits. Whole-genome sequencing allowed identification of the causal mutation of the phenotypes as two missense mutations in the coding region of CpETR1A and CpETR2B, each one corresponding to one of the duplicates of ethylene receptor genes highly homologous to Arabidopsis ETR1 and ETR2. The phenotypes of homozygous and heterozygous single- and double-mutant plants indicated that the two ethylene receptors cooperate in the control of the ethylene response. The level of ethylene insensitivity, which was determined by the strength of each mutant allele and the dose of wild-type and mutant etr1a and etr2b alleles, correlated with the degree of phenotypic changes in the mutants.

Ethylene receptor mutants reveal the involvement of ethylene perception in the control of sex expression, sex determination, and fruit set in Cucurbita pepo.

Introduction

Monoecious and dioecious plant species produce unisexual flowers (male or female) either in the same plant (monoecy) or in separate plants (dioecy). They are believed to be derived from a hermaphrodite ancestor, by different mechanisms, which result in the suppression of either stamen or carpel primordia development during the formation of female or male flowers, respectively (Dellaporta and Calderon-Urrea, 1993; Pannell, 2017). The genetic control mechanisms underlying sex determination in plants are diverse, ranging from heteromorphic sex chromosomes, as occurs in the dioecious species Silene latifolia and Rumex acetosa, to a number of non-linked genes, as occurs in the monoecious species of the family Cucurbitaceae (Jamilena et al., 2008; Pannell, 2017)

Cultivars of Cucurbitaceae species, including Cucumis sativus (cucumber), Cucumis melo (melon), Citrullus lanatus (watermelon), and the species of the genus Cucurbita (pumpkins and many squashes), are mainly monoecious, although certain cultivars of C. sativus, C. melo, and C. lanatus are andromonoecious, producing male and hermaphrodite flowers on the same plant (Malepszy and Niemirowicz-Szczytt, 1991; Perl-Treves, 2004; Boualem et al., 2009; Manzano et al., 2016). Both monoecious and andromonoecious plants go through two flowering phases of development: an initial phase in which the plant produces only male flowers, and a second phase in which the plant alternates the production of pistillate and male flowers (Perl-Treves 2004; Manzano et al., 2013, 2014; Zhang et al, 2017). The transition to pistillate flowering, and the number of pistillate flowers per plant, vary within the different cultivars of each species. In Cucumis, natural genetic variation includes other sexual phenotypes, such as gynoecious (a plant producing only female flowers) and androecious (a plant producing only male flowers). Neither the andromonoecious phenotype nor the gynoecious and androecious phenotypes have been observed in the genus Cucurbita, although some cultivars show a partially andromonoecious phenotype characterized by the occurrence of male and bisexual flowers, that is, pistillate flowers with partially developed stamens and no pollen (Martínez et al., 2014).

Sex determination mechanisms in cucurbit species are controlled by ethylene. Under treatments that reduce ethylene biosynthesis or perception, the monoecious plants are converted into partially or completely andromonoecious ones, demonstrating that ethylene participates in the sex identity of female flowers, and that an ethylene threshold is required to arrest stamen development in the female flower (Manzano et al., 2011; Zhang et al., 2017). Besides the control of individual floral buds, ethylene also participates in the control of sex expression within the plant. A reduction in ethylene biosynthesis or perception delays the transition to pistillate flowering and reduces the number of pistillate flowers per plant in C. sativus, C. melo, and C. pepo, but has the opposite effect in C. lanatus (Manzano et al., 2011, 2014). By contrast, treatment with ethylene or ethylene-releasing agents induces the production of male flowers in C. lanatus but promotes the production of pistillate flowers in the other species (Rudich et al., 1969; Byers et al., 1972; Den Nijs and Visser, 1980).

The genes and mutations responsible for cucurbit sex phenotypes are currently being sought. The arrest of stamen development in the female flowers of the different species requires the functioning of the ethylene biosynthesis orthologs CmACS7, CsACS2, CpACS27A, and CitACS4, which are specifically expressed in the female flowers of C. melo, C. sativus, C. pepo, and C. lanatus, respectively. Loss-of-function mutations in these genes led to andromonoecy in C. melo, C. sativus, and C. lanatus (Boualem et al., 2008, 2009; Ji et al., 2016; Manzano et al., 2016), but to only partial andromonoecy in C. pepo (Martínez et al., 2014). The androecious and gynoecious phenotypes also resulted from two independent mutations: in C. melo, androecy resulted from a mutation in the CmACS11 gene (Boualem et al., 2015), while gynoecy was produced by a mutation in the CmWIP1 gene (Martin et al., 2009). CmACS11 represses the expression of CmWIP1 to permit the coexistence of male and female flowers in monoecious species (Boualem et al., 2015).

Having performed extensive screening of an ethyl methanesulfonate (EMS)-generated mutant collection of C. pepo in the search for ethylene-insensitive mutants (García et al., 2018) in order to gain further insights into the genetic network regulating sex determination in cucurbits, in this paper we present a molecular and functional characterization of two semi-dominant mutations that affect two ethylene receptors of C. pepo: CpETR1A and CpETR2B. These mutations confer ethylene insensitivity on the plant, resulting in the conversion of female flowers to bisexual or hermaphrodite ones; that is, monoecy to andromonoecy. The mutations also alter the development and maturity of different floral organs in pistillate flowers, including ovaries and fruit.

Materials and methods

Plant material

The ethylene-insensitive mutants analyzed in this study were selected from a high-throughput screening of a C. pepo mutant collection by the triple-response assay (García et al., 2018), consisting of shortening and thickening of hypocotyls and roots in seedlings germinated in the dark with an external input of ethylene (Bleecker et al., 1988). The etr1a and etr2b mutants analyzed in this paper correspond to the ein2 and ein3 mutants isolated by García et al. (2018).

Before phenotyping, etr mutant plants from each family were crossed for two generations with the background genotype MUC16, and the resulting BC₂ generation was selfed to obtain the BC₂S₁ generation. Given that homozygous etr1a and etr2b mutants were female sterile, they were always derived from selfed progenies of BC heterozygous plants. Sterility also prevented us from obtaining double homozygous mutants. The heterozygous double mutants (wt/etr1a wt/etr2b) were obtained by crossing heterozygous wt/etr1a as female and homozygous etr2b/etr2b as male, and genotyping the offspring for the causal mutations. The double heterozygous plants were also female sterile, preventing us from obtaining the double homozygous etr1a/etr1a etr2b/etr2b.

Phenotyping for sex expression and sex determination traits

First, BC₁S₁ or BC₂S₁ plants from etr1a and etr2b mutant families were classified according to their level of triple response to ethylene in wild-type (WT), intermediate (wt/etr), and ethylene-insensitive mutants (etr/etr), and then transplanted to a greenhouse and grown to maturity under local greenhouse conditions without climate control, and under standard crop management, in Almería, Spain.

For the ethylene response assay, seeds were germinated for 2 days in the absence of ethylene and then placed in a growth chamber containing 50 ppm ethylene in darkness for 5 days. The identified mutants showed more elongated hypocotyls and roots than WT, resembling seedlings grown in air. The ethylene sensitivity of each mutant genotype was estimated by using three replicates with at least 20 seedlings of the same genotype. Ethylene sensitivity was assessed as the percentage of reduction in hypocotyl length relative to air-grown seedlings: [(H₀–H₁/H₀)×100], where H₀ corresponds to the hypocotyl length of air-grown seedlings, and H₁ to the hypocotyl length of ethylene-treated seedlings. The final ethylene sensitivity was relativized to the WT seedling response, considering that the WT has 100% ethylene sensitivity, and ethylene insensitivity was calculated as (100–ethylene sensitivity).

The sex phenotype of each plant was determined according to the sex of the flowers in the first 40 nodes of each plant. A minimum of 30 WT, 30 wt/etr, and 30 etr plants were phenotyped for each mutant family. The sex expression of each genotype was assessed by determining the node at which plants transitioned to pistillate flowering, and the number of male or pistillate flower nodes. The sex phenotype of each individual pistillate flower was assessed by the so-called andromonoecy index (AI) (Martínez et al., 2014; Manzano et al., 2016). Pistillate flowers were separated into three phenotypic classes that were given a score from 1 to 3 according to the degree of their stamen development: female (AI=1), showing no stamen development; bisexual (AI=2), showing partial development of stamens and no pollen; and hermaphrodite (AI=3), showing complete development of stamens and pollen. The average AI of each plant and genotype was then assessed from the resulting AI score of at least five individual pistillate flowers from each plant, using a minimum of 30 plants for each genotype.

The growth rates of ovaries and petals in both pistillate and male flowers in each of the WT and mutant plants were assessed by measuring the length of ovaries and petals every 3 days in at least 12 flowers of each genotype, starting with flowers ~2 mm in length. The anthesis time was estimated as the number of days taken for a 2 mm pistillate or male floral bud to reach anthesis. The effect of the etr1a and etr2b mutations on the vegetative vigor of each plant was assessed by determining the average plant height, the total number of nodes, and the average internode length in the main shoot of WT and mutant plants at 60 days after transplantation.

Identification of etr1a and etr2b mutations by whole-genome sequencing analysis

To identify the causal mutations of the etr1a and etr2b phenotypes, WT and mutant plants, which were both derived from BC₂S₁ segregating populations, were subjected to whole-genome sequencing (WGS). In total, 120 BC₂S₁ seedlings from each mutant family were subjected to the ethylene triple-response assay, and the plants exhibiting the ethylene-insensitive mutant phenotype were separated from the WT. The phenotype of those seedlings was verified in the adult plants, since WT plants were monoecious while homozygous mutant plants were andromonoecious or partially andromonoecious.

The DNA of each plant was isolated by using the Gene JET Genomic DNA Purification Kit (Thermo Fisher), and the DNA from 20 plants showing the same phenotype in each mutant family were pooled into different bulks. Two DNA bulks were generated for each mutant family: one WT bulk and one mutant bulk. The four DNA bulks were used to prepare paired-end multiplex libraries by using the KAPA Library Preparation kit (Kapa Biosystems) and sequenced in 126 base-read pairs on an Illumina HiSeq2000 instrument, following the manufacturer's protocol. Base calling and quality control were done using the Illumina RTA sequence analysis pipeline, according to the manufacturer's instructions. The average fold-effective coverage of all the samples was between 15.5 and 17.2, and the percentage of genomic bases with a fold coverage higher than 5 was between 78.4 and 83.3% (see Supplementary Table S1 at JXB online). Raw reads were mapped on to the C. pepo genome v3.2 by using the GEM mapper (Marco-Sola et al., 2012) in order to generate four bam files. Indels were realigned with GATK, and duplicates were marked with Picard v1.110. In each sample we counted all the ‘good-quality bases', selected according to the following filter parameters: base quality ≥17, mapping quality ≥20, read depth ≥8. Variants were called using GATK's HaplotypeCaller in gVCF mode (McKenna et al., 2010), and then the GenotypeGVCF tool was used to perform joint genotyping on all the samples together. Only the single nucleotide polymorphisms (SNPs) were selected by using GATK's SelectVariants tool. Sites with multiple nucleotide polymorphisms were filtered out.

With regard to the segregation analysis of each mutant phenotype, the genotype of WT plants should be considered as 0/0, as the reference genome. The mutant bulks were completed by using DNA from the most ethylene-insensitive plants. However, given the semi-dominant nature of the etr1a and etr2b mutations, we expected some plants to be heterozygous for the mutations (0/1), although most of them were expected to be homozygous (1/1), with a mutant allelic frequency (AF) close to 1. Therefore, we filtered our data according to the following parameters: genotype quality ≥20, read depth ≥4 (this specific base was covered with at least four reads), AF ≥0.8 in the two mutant bulks, and AF ≤0.2 in the WT bulks.

Validation of the identified mutations by high-throughput genotyping of individual segregating plants

To confirm that the identified mutations were the causal mutations of the etr1a and etr2b phenotypes, more than 200 individual BC₂S₁ segregating populations were genotyped by using real-time PCR with TaqMan probes. The multiplex PCRs were done using the Bioline SensiFAST™ Probe No-ROX Kit, a set of forward and reverse primers amplifying the polymorphic sequence, and two allele-specific probes descriptive of the SNP of interest (C–T). The WT probe was labeled with FAM dye, while the mutant probe was labeled with HEX reporter dye. The BHQ1 quencher molecule was used in both probes (Supplementary Table S2). After genotyping, plants were phenotyped for ethylene triple response in etiolated seedlings and for sex expression in adult plants, enabling us to see whether the mutant alleles co-segregated with the etr1a or the etr2b mutant phenotype.

Assessment of relative gene expression by quantitative RT–PCR

Gene expression analysis was performed by using quantitative reverse transcription (qRT)–PCR on three biological replicates, each one resulting from an independent extraction of total RNA from samples that were pooled from at least three plants with the same genotype.

RNA was isolated according to the protocol of the GeneJET Plant RNA Purification Kit (Thermo Fisher). cDNA was then synthesized by using the RevertAid RT Reverse Transcription Kit (Thermo Fisher). The expression levels of genes were evaluated through qRT–PCR by using the Rotor-Gene Q thermocycler (Qiagen) and SYBR® Green Master Mix (BioRad). Supplementary Table S3 shows the primers used for qRT–PCR reactions for each analyzed gene.

Bioinformatics and statistical analyses

Alignments were performed using the BLAST alignment tools at NCBI (http://www.blast.ncbi.nlm.nih.gov/) and Clustal Omega at EMBL-EBI (https://www.ebi.ac.uk/Tools/msa/clustalo/).

The phylogenetic relations of ETR1, ETR2, and ERS1-like ethylene receptors were studied using MEGA7 software (Kumar et al., 2016), which allowed the alignment of proteins and the construction of phylogenetic trees using the Maximum Likelihood method based on the Poisson correction model (Zuckerkandl and Pauling, 1965), with 2000 bootstrap replicates. The protein sequences (Supplementary Table S4) were obtained using The Arabidopsis Information Resource (https://www.arabidopsis.org/) and the Cucurbit Genomics Database (http://cucurbitgenomics.org/).

Multiple data comparisons were obtained by analysis of variance with the significance level P<0.05, and each two averages were compared using Fisher's least significant difference method.

Results

etr1a and etr2b are two semi-dominant ethylene-insensitive mutations affecting sex determination

etr1a and etr2b are two independent, ethylene-insensitive mutant families that were found after screening a mutant collection of C. pepo for ethylene triple response (García et al., 2018). To ensure accurate phenotyping, mutant plants were backcrossed with the background genotype MUC16 for two generations, and then selfed. The resulting BC₁S₁ or BC₂S₁ generations segregated for three ethylene-response phenotypes in etiolated seedlings: ethylene-sensitive plants (WT), intermediate plants (wt/etr), and ethylene-insensitive plants (etr) (Fig. 1). The segregation ratio of the three ethylene triple-response phenotypes in BC₁S₁ and BC₂S₁ generations indicated that the two ethylene-insensitive mutations were semi-dominant, and that the intermediate phenotype corresponded to heterozygous plants (wt/etr1a or wt/etr2b) (Supplementary Table S5) (García et al., 2018).

Fig. 1.

Ethylene triple-response phenotypes of WT and heterozygous and homozygous single and double mutants for etr1a and etr2b. When grown in air, both the WT and the mutants showed the same growth. When exposed to ethylene, the WT responded with drastic reductions in the length of the hypocotyl and roots, while the mutants had a minor but differential response. The average level of ethylene sensitivity (ES) was assessed as the percentage of reduction in hypocotyl length relative to air-grown seedlings, and assuming that WT is 100% ethylene sensitive. The percentage ethylene insensitivity was then calculated as (100–ES). Assessments were performed in three replicates with at least 20 seedlings for each genotype.

WT plants responded to ethylene with reduced length of the roots and the hypocotyl, and also increased hypocotyl thickness, in comparison to those grown in air. In contrast, the length of the hypocotyl and roots of ethylene-treated homozygous etr1a and heterozygous etr2b seedlings was reduced to a lesser extent in relation to those grown in air (Fig. 1); this indicated that the level of insensitivity of the two mutants was not total, as they possessed a partially ethylene-insensitive phenotype. Assuming that WT plants are completely ethylene sensitive (0% ethylene insensitivity), we estimated the percentage of ethylene insensitivity in homozygous and heterozygous seedlings of the two mutants. The homozygous etr1a seedlings showed a more severe ethylene-insensitive phenotype than the homozygous etr2b seedlings (85.5% ethylene insensitivity for etr1a/etr1a versus 63.7% in etr2b/etr2b). Heterozygous mutants displayed an intermediate percentage of ethylene insensitivity (51.1% in wt/etr1a and 43.5% in wt/etr2b) (Fig. 1).

Given that ethylene is the main regulator of sex determination in cucurbits, we assessed whether the etr1a and etr2b mutations altered the sexual phenotype of the plants. Staminate and pistillate flowers in the first 40 nodes of WT, heterozygous, and homozygous mutant plants were assessed (Supplementary Fig. S1). All plants showed two sexual phases of development. In the first phase, plants produce only male flowers. The second phase starts after the pistillate flowering transition and is characterized by the production of male and pistillate flowers alternately (Fig. 2A). The duration of these two phases of sexual development was affected by the etr mutations. Homozygous etr1a and etr2b mutants showed a significant delay in pistillate flowering transition, as well as a reduction in the number of pistillate flowers per plant (Fig. 2B, C). Heterozygous etr1a and etr2b mutants showed an intermediate phenotype for both traits (Fig. 2B, C), confirming the semi-dominant nature of the two mutations.

Fig. 2.

Effect of etr1a and etr2b on sex expression. (A) Schematic representation of the distribution of male and female flowers in WT and etr1a or etr2b (etr) mutant plants. The male and female phases of development, and the node at which plants start to produce pistillate flowers (pistillate flowering transition), are indicated. Blue = male flower; red = female flower; yellow = bisexual or hermaphrodite flower. (B) Comparison of pistillate flowering transition in WT plants and single and double mutants. (C) Comparison of the number of pistillate flowers per plant in WT and single and double mutant plants. Error bars represent SE. Different letters in (B) and (C) indicate statistically significant differences (P<0.05) between samples.

The most evident phenotypic alteration in the etr1a and etr2b mutants was the conversion of monoecy into partial or complete andromonoecy. WT plants produced exclusively female flowers after the pistillate flowering transition, indicating a complete arrest of stamen development in pistillate flowers (Fig. 3A). In the homozygous etr1a mutants, almost all female flowers were transformed into hermaphrodite flowers, showing fully developed stamens with fertile pollen (Fig. 3B). In etr2b mutants, a high percentage of female flowers were transformed into bisexual (pistillate flowers with immature stamens) or hermaphrodite flowers; however, a small number of female flowers remained (Fig. 3B). In heterozygous plants bearing both mutations, only a very small proportion of the female flowers were converted into bisexual flowers (Fig. 3B), indicating that with regard to the sex of the flower, etr1a and etr2b mutations are recessive, and therefore not semi-dominant as shown for ethylene triple response and sex expression. Given that the degree of stamen development in pistillate flowers was variable, plants were classified according to the AI, ranging from AI=1 (monoecious) to AI=3 (andromonoecious) (Fig. 3A). All ethylene-sensitive WT plants showed an average AI of 1. Homozygous plants containing either etr1a or etr2b mutations displayed an average AI of 2.8 and 1.8, respectively, indicating that the etr1a mutation rendered plants almost completely andromonoecious, while the etr2b mutation rendered them partially andromonoecious. The average AI for heterozygous plants was 1.1 for both etr1a and etr2b types, very similar to that of WT plants (Fig. 3C).

Fig. 3.

Effect of etr1a and etr2b mutations on sex determination in C. pepo. (A) Phenotypes of pistillate flowers in WT and mutant plants: female [andromonoecy index (AI)=1], bisexual (AI=2), and hermaphrodite (AI=3). Female flowers develop no stamen, while bisexual and hermaphrodite flowers develop immature stamens and mature stamens with pollen, respectively. (B) Percentage of male, female, bisexual, and hermaphrodite flowers in WT plants and etr1a and etr2b single and double mutants. (C) AI of WT plants and etr1a and etr2b single and double mutants. The average AI of each genotype was assessed in at least five individual pistillate flowers from each plant, with each genotype containing a minimum of 30 plants. AI varies between 1 (complete monoecy) and 3 (complete andromonoecy). Error bars represent SE. Different letters indicate statistically significant differences (P<0.05) between samples.

etr1a and etr2b alter petal and ovary/fruit development and affect plant vigor

Table 1 and Fig. 4 show the effects of the etr1a and etr2b mutations on petal and ovary/fruit development. In the hermaphrodite flowers of etr1a, the petal growth rate was reduced and resembled petal development in male flowers. Petal maturity and subsequent anthesis of the flower were delayed (Fig. 4B). Anthesis time, which is the period of time taken for a 2 mm floral bud to reach anthesis and to open, was longer in male WT flowers (average 22 days) than in female WT flowers (average 14 days) (Table 1). Hermaphrodite etr1a flowers took an average of 22 days to reach anthesis (range 15–40 days). Under the greenhouse conditions used, several hermaphrodite flowers did not reach anthesis; the petals remained green and closed for more than 40 days (Fig. 4B). The development to maturity of petals in bisexual flowers of etr2b was also delayed in comparison with WT female flowers. However, the delay was less pronounced than in etr1a (Fig. 4B, Table 1). No alterations in petal development or anthesis time were observed in etr1a and etr2b male flowers, and no change was found in petal development or anthesis time of female flowers of etr1a or etr2b heterozygous plants (Fig. 4B; Table 1).

Table 1.

Anthesis time of pistillate and male flowers in etr1a and etr2b mutant families

Family	Flower	Anthesis time (days)
		wt/wt	wt/etr	etr/etr
etr1a	Pistillate	14.3±1.2 b	14.8±1 b	22.7±1.4 a
	Male	22.4±0.7 a	23.1±1.8 a	22.7±1.8 a
etr2b	Pistillate	13.8±1.3 b	14.3±1.1 b	20±4.2 a
	Male	22.4±0.5 a	21.6±1.1 a	22±1.7 a

Different letters within the same row indicate significant differences between means (P<0.05).

Fig. 4.

Effect of etr1a and etr2b mutations on the growth rate of petals and ovaries of male and pistillate flowers. (A) Images of pistillate flowers in WT and either etr1a or etr2b mutants (etr) at 7, 14, 20, or 24 days after the flower’s ovary reached 4 mm in length. The WT pistillate flower reached anthesis at ~15 days, and in the absence of pollination, the floral organs abscised and the fruit aborted 2–3 days after anthesis. In the two etr mutants, the anthesis time was markedly delayed, but the fruit grew normally in the absence of pollination (parthenocarpic fruits). (B) Comparison of the growth rate of WT and mutant petals. Flowers were labeled when their ovaries were 4 mm long, and then measured every 3 days for 25 days. P = pistillate flowers, M = male flowers. Yellow circles indicate the time at which more than 80% of the flowers reached anthesis. (C) Comparison of the growth rate of WT and mutant ovaries/fruits over a period of 25 days. Error bars represent SE.

Significant differences in ovary size were detected between WT and etr1a and etr2b pistillate flowers (Fig. 4). While WT ovaries reached ~8 cm at anthesis and then aborted, the flowers of etr1a and etr2b reached 16–40 cm (or even more) at anthesis (Fig. 4C). These can be considered as parthenocarpic fruits because the flowers remained closed, rendering them unavailable for pollination. During the first 16 days, the growth rates of WT and mutant ovaries were similar. After 16 days, WT ovaries aborted, while those of homozygous etr1a and etr2b flowers maintained their growth up to anthesis. Since the anthesis time of etr1a and etr2b pistillate flowers was achieved much later than that of WT plants, the size of the mutant ovary at anthesis was much larger (Fig. 4C). The ovary growth rate of heterozygous wt/etr1a flowers was intermediate between that of wt/wt and homozygous etr1a/etr1a (Fig. 4C), while heterozygous wt/etr2b ovaries displayed the same growth rate as WT plants. Pollination was attempted in homozygous mutant flowers that reached anthesis, but none of the fruits in the homozygous etr1a and etr2b mutant plants were able to set seeds under the given conditions. Since the pollen of these mutants is fertile, these results indicate female sterility associated with the etr1a and etr2b mutations. As no pure seed could be obtained from etr1a or etr2b homozygous mutants, they were maintained by selfing heterozygous mutant plants.

Vegetative development was enhanced in mutant adult etr1a/etr1a and etr2b/etr2b plants. Table 2 shows the differences in plant height, number of nodes, and internode length of plants grown under the same conditions. The two mutations were associated with increased plant growth rate and height compared with WT plants. These differences in height and growth rate were mainly due to an increase in the internode length, which was approximately twice as high in mutants as in WT plants. The number of nodes developed by WT and mutant plants was, however, very similar (Table 2).

Table 2.

Effects of the etr1a and etr2b mutations on plant vegetative development

	Plant height (cm)	Node number	Internode length (mm)
WT	85.43 b	47.2 b	13.5 b
etr1a/etr1a	112.5 a	50.7 a	24.1 a
WT	93.9 b	47.1 a	14.8 b
etr2b/etr2b	113.3 a	48.5 a	28.8 a

Different letters indicate significant differences between wild-type (WT) and homozygous mutants (P<0.05).

Phenotype of etr1a and etr2b double mutants

The interaction between the etr1a and etr2b mutations was studied in double-heterozygous plants for both mutations. As mutant bisexual and hermaphrodite flowers were sterile, heterozygous wt/etr1a plants were pollinated with pollen from homozygous etr2b/etr2b plants. The progeny segregated as 1:1 for single-heterozygous wt/wt wt/etr2b and double-heterozygous wt/etr1a wt/etr2b plants. However, double-heterozygous plants were also female sterile, which made it impossible to generate double-homozygous plants bearing the two mutations. The reciprocal cross produced the same results.

Double-heterozygous mutants showed a more severe ethylene-insensitive phenotype (61.3% insensitivity) than single-heterozygous mutants (51.1% insensitivity for wt/etr1a and 43.5% for wt/etr2b), and their phenotype resembled the triple response of etr2b/etr2b (Fig. 1). The maleness effect of the etr1a and etr2b mutations was also enhanced in the double-heterozygous wt/etr1a wt/etr2b plants compared with the phenotype of the single-heterozygous mutants. The female flowering transition was delayed up to an average of 17.9 nodes, and pistillate flower production was also significantly reduced to about 4.3 flowers resulting from the first 40 nodes of the plant, again resembling the phenotype exhibited by the homozygous single mutants (Fig. 2B, C).

In the double-heterozygous wt/etr1a wt/etr2b plants, most female flowers were converted into bisexual or hermaphrodite flowers (average AI=2.3), contrasting with single-heterozygous plants, which produced nearly 100% female flowers (average AI=1.1) (Fig. 3B, C). Thus, the combination of the two etr mutations, albeit in heterozygous conditions, had a similar effect on the flower sexual phenotype as the homozygous single mutations (Fig. 3B, C). These data therefore indicate that the two mutations have an additive effect on the sex phenotype of pistillate flowers, and that the effect is dependent upon the number of mutant alleles for the two loci.

Identification of the etr1a and etr2b mutations

The causal mutations of the etr1a and etr2b phenotypes were identified by WGS. One DNA WT bulk and one DNA mutant bulk were made for each mutant family, each bulk consisting of a pool of DNA from 20 plants from the same BC₂S₁ segregating population. With regard to the mutant bulk, only the ethylene-insensitive plants that showed the most strongly andromonoecious phenotype were selected, in order to assure that they were homozygous for the mutations. The four DNA bulks were subjected to WGS and the results were mapped against the C. pepo reference genome version 3.2. The identified SNPs were filtered by using different criteria. Common variants between samples of the two families were discarded, as they were considered likely to correspond to spontaneous nucleotide polymorphisms in the MUC16 genetic background (García et al., 2018). SNPs corresponding to canonical EMS mutations (C>T and G>A) were selected, and filtered for their quality and depth, as well as for their mutant allele frequency (AF), in WT and mutant DNA samples. We expected the WT bulks to present the genotype 0/0 (AF=0) and the mutant bulks to present the genotype 1/1 (AF=1).

Assuming minor contamination of the bulks with some heterozygous plants, we discarded SNPs with AF<0.8 in the mutant samples and AF>0.2 in the WT samples. After filtering, two EMS candidate mutations were selected for etr1a and one candidate mutation was selected for etr2b (Supplementary Table S1). The sequences surrounding the candidate mutations (±500 bp) were then used in BLAST searches against the DNA and protein databases at NCBI, which detected an EMS canonical C>T transition in both etr1a and etr2b families, located in the coding region of the ethylene receptor genes CpETR1A and CpETR2B, respectively (Table 3). To verify that these were the causal mutations of the etr1a and etr2b phenotypes, more than 200 plants segregating for either etr1a or etr2b were genotyped for the WT and mutant alleles of CpETR1A and CpETR2B. The results demonstrated a perfect co-segregation of the etr1a and etr2b sex phenotypes with mutations in CpETR1A and CpETR2B, respectively (Table 4). The other candidate mutation for etr1a segregated independently of the mutant phenotype.

Table 3.

Genotype, depth (DP), and allele frequency (AF) of causal mutations of the etr1a and etr2b phenotypes

	DNA bulk	Causal mutations
		LG07: 6 891 436 (C>T)			LG03:11 824 350 (C>T)
		Genotype	DP	AF	Genotype	DP	AF
etr1a	WT	0/0	20	0.15	0/0	19	0
	Mutant	1/1	20	1	0/0	9	0
etr2b	WT	0/0	16	0	0/0	10	0
	Mutant	0/0	8	0	1/1	14	1

Table 4.

Co-segregation analysis of the CpETR1A and CpETR2B mutations with the ethylene-insensitive phenotypes in BC₂S₁ populations segregating for etr1a, etr2b, and double mutants

Segregating population	CpETR1A or CpETR2B genotypes	Triple response to ethylene			Sexual phenotype		Total
		Sensitive	Intermediate	Insensitive	Monoecious	Andromonoecious
etr1a	0/0	80	–	–	80		226
	0/1	–	61		61
	1/1	–		85		85
etr2b	0/0	85	–	–	85		234
	0/1	–	62		62
	1/1	–		87		87
Double mutants	0/0; 0/1	–	22		22		47
	0/1; 0/1	–		25		25

With regard to the ethylene triple response, the plants that were homozygous for the WT alleles were sensitive to ethylene, and those that were homozygous for the mutant alleles of CpETR1A or CpETR2B were ethylene insensitive; heterozygous plants showed intermediate triple-response phenotypes (Table 4). Moreover, plants that were homozygous for any of the identified mutations were all andromonoecious, while those that were either homozygous for the WT allele or heterozygous were all monoecious (Table 4). This finding demonstrates that mutations identified in the two ethylene receptor genes CpETR1A and CpETR2B co-segregated with the ethylene-insensitive phenotype and andromonoecy of the etr1a and etr2b mutants.

Gene structure of CpETR1 and CpETR2

The de novo assembly of the C. pepo genome, published recently, revealed a whole-genome duplication, which occurred just before the speciation event that created the genus Cucurbita (Sun et al., 2017; Montero-Pau et al., 2018). Accordingly, we found that the genomes of the Cucurbita species contain two paralogs for each of the ethylene receptors ETR1, ERS1, and ETR2. No ERS2- or EIN4-like receptors were found in the genomes of these species. The C. pepo genome has two ETR1 duplicates (CpETR1A and CpETR1B), which showed more than 90% homology and mapped on chromosomes 7 and 11, and two ETR2 duplicates (CpETR2A and CpETR2B), which showed more than 89% homology and mapped on chromosome 8. The duplicates maintained the same molecular structure: six exons and five introns for the two CpETR1 paralogs, and three exons and two introns for the two CpETR2 paralogs (Fig. 5A, B).

Fig. 5.

Molecular structure of CpETR1A and CpETR2B, and phylogenetic analysis of ETR1- and ETR2-like ethylene receptors. (A) CpETR1A; (B) CpETR2B. Within each gene, black boxes indicate untranslated regions and grey boxes correspond to exons. Numbers indicate the size of the fragment (bp). Missense mutations in the gene, and amino acid substitutions in the protein, are indicated in red. EMS mutations C>T are numbered with respect to the coding sequence. The three transmembrane subdomains in the ethylene-binding domain, as well as the GAF, histidine-kinase, and receiver domains, are indicated for each ethylene receptor. (C) Phylogenetic relationships between the ethylene receptors of six cucurbit species (Cucurbita pepo, Cucurbita moschata, Cucurbita maxima, Cucumis melo, Cucumis sativus, and Citrullus lanatus) and those of Arabidopsis. The six ethylene receptors of C. pepo are positioned in three clusters corresponding to ETR1, ERS1, and ETR2 of Arabidopsis. In the three Cucurbita species, ETR1 and ETR2 are duplicated in relation to Arabidopsis.

The etr1a mutation was located at nucleotide position 284 of the first exon of the CpETR1A gene, and the etr2b mutation was located at nucleotide position 1018 of the first exon of the CpETR2B gene (Fig. 5A, B). The deduced ETR1A and ETR2B receptors had the same domains as those of the Arabidopsis homologs: an ethylene-binding domain with three transmembrane segments; one GAF domain; one histidine-kinase domain; and a response regulator receiver domain. The two mutations resulted in an amino acid substitution in each protein: A95V in the ethylene-binding domain of CpETR1A, and E340K in the coiled-coil domain between the GAF and histidine-kinase domains of CpETR2B (Fig. 5A, B). All ETR1- and ETR2-like proteins in the NCBI database were found to contain the WT amino acid at these two particular positions, indicating that the amino acids affected by the two mutations are highly conserved in very different plant species (Supplementary Figs S2 and S3).

To assess the genetic relationships between ETR1- and ETR2-like ethylene receptors in C. pepo, a phylogenetic tree was inferred from the ETR1, ERS1, and ETR2 protein sequences of different cucurbit species—Cucurbita maxima, Cucurbita moschata, C. melo, C. sativus, and C. lanatus—together with the five ethylene receptors of Arabidopsis (Fig. 5C, Supplementary Table S4). In the other cucurbits, only one single gene was found for each ethylene receptor. In agreement with the allotetraploid origin of the genus Cucurbita (Sun et al., 2017), only one of the paralogs in subgenome B (CpETR1B, CpERS1B, and CpETR2B in C. pepo) clustered with their orthologs in the rest of the cucurbit species (Fig. 5C).

Effects of etr1a and etr2b on ethylene receptor gene expression

The expression patterns of the mutated CpETR1A and CpETR2B and their duplicated paralogs (CpETR1B and CpETR2A9) were studied in leaves, roots, shoots, and shoot apices of WT plants (Fig. 6). The four genes were expressed in all the analyzed tissues, suggesting that both duplicates maintain their expression in these tissues.

Fig. 6.

Expression patterns of CpETR1 and CpETR2 ethylene receptor genes in different plant organs of C. pepo WT plants. The relative level of each transcript was quantified by qRT–PCR in three independent replicates of each tissue. The shoot apex consists not solely of the apical meristem but also of small leaves and floral buds. Different letters indicate statistically significant differences (P<0.05) between samples.

We also investigated whether the etr1a or etr2b mutations alter the patterns of expression of the four ETR genes. Gene expression was compared in WT and mutant female floral buds at very early stages of development, that is, when stamen arrest takes place in the WT female flower (stage T0), and at 1 day before anthesis of WT flowers (stage T5); in the latter case, we separated the flower into the ovary and a tissue comprising the petals, style, and stigma (Fig. 7). The etr1a mutation inhibited the expression of CpETR1A and CpETR2A in the tissue that comprised the petals, style, and stigma of T5 flowers, and also reduced the transcription of CpETR2B in T0 flowers (Fig. 7). By contrast, the etr2b mutation inhibited the expression of CpETR1B and CpETR2A in T0 floral buds, and the expression of CpETR1A in the ovary of T5 flowers.

Fig. 7.

Relative expression of CpETR1 and CpETR2 ethylene receptor genes in female flowers of C. pepo. The relative level of each transcript was quantified by qRT–PCR in three independent replicates of each tissue. T0 corresponds to complete female flowers 2 mm in length; T5 corresponds to pre-anthesis-stage female flowers, separated into the ovary and a tissue comprising the petals, style, and stigma (PSS). The comparison of gene expression was performed between homozygous WT and homozygous mutant flowers derived from plants in the same segregating population. Different letters indicate statistically significant differences (P<0.05) between samples.

Discussion

etr1a and etr2b are two missense mutations in ethylene receptors leading to semi-dominant ethylene insensitivity

Ethylene is perceived by a family of two-component histidine kinase receptors that repress the ethylene signaling cascade in the absence of ethylene but become inactivated upon ethylene binding (Hua and Meyerowitz, 1998). Mutations in ethylene receptor genes fall into two main categories: (i) dominant gain-of-function mutations, conferring ethylene insensitivity, and (ii) recessive loss-of-function mutations that have little effect as single mutations but show a constitutive ethylene response in combination, for example, in double, triple, and quadruple mutants of Arabidopsis. Both etr1a and etr2b of C. pepo correspond to the first type of mutation, since they are semi-dominant and result in plant ethylene insensitivity (Fig. 1).

In Arabidopsis dominant mutants, the ethylene-insensitive phenotypes are caused by single amino acid substitutions in the transmembrane ethylene-binding domain of any of the five ethylene receptors described in this species (Bleecker et al., 1988; Chang et al., 1993; Guzmán and Ecker, 1990; Hua et al., 1995, 1998; Wang et al., 2006). The etr1a mutation described here is also a missense mutation (A95V), situated in the third transmembrane domain of the N-terminal ethylene-binding site of CpETR1A (Fig. 5), which causes a strong reduction in ethylene sensitivity in etiolated seedlings (Fig. 1). The mutation is contained in a conserved segment of the protein, close to the T94M mutation of Arabidopsis ETR1, which is known to disrupt the ability of the receptor to bind ethylene and to strongly affect ethylene sensitivity (Wang et al., 2006; Resnick et al., 2008). The etr2b mutation, however, is a missense mutation (E340K) within the coiled-coil domain between the GAF and histidine-kinase domains of CpETR2B (Fig. 5). Given that this domain does not participate in ethylene binding, it is likely that the ethylene insensitivity of etr2b is caused by a lack of transduction of the ethylene signal, as has been suggested for other dominant ethylene-insensitive mutations (Hall et al., 1999). Thus, etr1a may disrupt the ethylene-binding site, and etr2b may alter ethylene signal transduction, but both mutations should convert the CpETR1A or CpETR2B receptors to a constitutive signaling-on state that represses the ethylene response (Fig. 8). The differing levels of ethylene insensitivity shown by single etr1a and etr2b mutants of C. pepo could indicate that CpETR1A (subfamily I) has a more prominent role in ethylene perception than CpETR2B (subfamily II). In Arabidopsis, both etr1-1 and etr1-4 mutants are insensitive to ethylene, but etr1-2, etr2-1, and ein4-3 maintain a reduced response to ethylene (Hall et al., 1999).

Fig. 8.

Model of sex determination in cucurbit species, integrating the function of the ethylene receptors ETR1 and ETR2 of C. pepo with other sex-determining genes identified in C. melo and C. sativus (ACS11, ACO2, WIP1, and ACS2/7). The ethylene biosynthesis enzymes ACS11 and ACS2/7 have different spatiotemporal expression patterns, causing the arrest of carpels or stamens required for a flower to develop as male or female, respectively (left panel). The two biosynthetic pathways are connected by the transcription factor WIP1, which represses the transcription of ACS2/7, and is negatively regulated by the ethylene-producing enzymes ACS11and ACO2 (Boualem et al., 2015; Che and Zhang, 2019). The ethylene receptors ETR1 and ETR2 should perceive and transmit signaling of the ethylene synthesized by both the ACS11 and ACS2/7 pathways. The inhibition of the ACS2/7 ethylene response releases the arrest of stamens in female flowers, resulting in the production of bisexual or hermaphrodite flowers (andromonoecy). On the other hand, the inhibition of the ACS11 ethylene response can induce WIP1, which enhances the arrest of carpels and so leads to the formation of male flowers. Red and dotted lines in the right panel indicate increased or decreased effects, respectively, produced by the etr1 and etr2 mutations.

A whole-genome duplication occurred just before speciation of the genus Cucurbita, which explains why only the species of the genus Cucurbita have two subgenomes (A and B) (Sun et al., 2017; Montero-Pau et al., 2018). In C. pepo, we found that the paralogs of CpETR1 (CpETR1A and CpETR1B) and CpETR2 (CpETR2A and CpETR2B) in each subgenome are both expressed and could be functional (Figs 6 and 7). This tetraploid origin of C. pepo could account for the semi-dominant nature of etr1a and etr2b. In Arabidopsis, ethylene-insensitive mutants show the same phenotype in both homozygous and heterozygous conditions; however, an extra WT allele of the same gene in triploid plants (mut/wt/wt) for either etr1-1, etr1-2, or ein4-3 reduced ethylene insensitivity compared with homozygous or heterozygous diploid mutants (mut/mut or wt/mut). This suggests that the dominant insensitivity of these mutant alleles is mediated by interaction between WT and mutated ethylene receptor isoforms (Hall et al., 1999). Our findings also indicate that the CpETR1A and CpETR2B ethylene receptors exert their action cooperatively rather than independently. Different members of the Arabidopsis ethylene receptors form homomeric and heteromeric complexes that may facilitate receptor signal output (Binder and Bleecker, 2003; Xie et al., 2006; Grefen et al., 2008; Gao et al., 2008; Gao and Schaller, 2009), and mutant receptors can repress the ethylene response only when coupled with a WT receptor, which implies that each tissue can have mixed receptor complexes with different receptor signal output strengths (Li et al., 2009).

The cooperation between WT and mutant alleles of CpETR1A and CpETR2B is also suggested by the phenotypes of single and double mutants, where homozygous single mutants for etr1a or etr2b (85% and 63.7% ethylene insensitivity, respectively) show similar ethylene insensitivity to the heterozygous double mutant wt/etr1 wt/etr2b (61.3%) but stronger insensitivity than that of the heterozygous single mutants wt/etr1a and wt/etr2b (53.5% and 43.5%, respectively). The suppression of the ethylene response in C. pepo is therefore dependent on the dosage of WT and mutant alleles for either CpETR1A or CpETR2B (Figs 8 and 9).

Fig. 9.

Cooperation of the CpETR1A and CpETR2B ethylene receptors in the regulation of the ethylene response and ethylene-associated traits in C. pepo flower and fruit development. In the absence of ethylene, CpETR1A and CpETR2B repress the ethylene response. In the presence of ethylene, the repression of the ethylene response is determined by the dosage of WT and mutant alleles, and also by the strength of each mutant allele, with etr1a having a stronger effect than etr2b. The resulting ethylene sensitivity regulates the development of pistillate floral organs and sex determination. As the number of mutant alleles increases, the response to ethylene is reduced. This increases the degree of andromonoecy, delays the pistillate (female) floral transition, and decreases the number of pistillate flowers per plant. It also delays the maturity of pistillate flowers, favoring growth of the ovary and parthenocarpic development of the fruit. FF, female flowers.

Ethylene insensitivity in the two etr mutants was not, however, associated with the expression level of the genes. The etr1a and etr2b mutations scarcely affected the expression of ethylene receptor genes, although some of them were down-regulated in female flowers, probably owing to a mechanism that aids the recovery of ethylene sensitivity that may be lost as a result of the mutations. Functional compensation between ethylene receptor gene families has been found to occur between NR and LeETR4 of tomato (Tieman et al., 2000) and in certain loss-of-function ethylene receptor mutants of Arabidopsis (Zhao et al., 2002; Chen et al., 2007; Harkey et al., 2018), but not in rice (Wuriyanghan et al., 2009). Nevertheless, given that some of the ETR genes were not down-regulated in all etr1a or etr2b tissues, it is also likely that reduction in gene expression reflects a differential regulation of the analyzed ethylene receptor genes.

Mutations in CpETR1A and CpETR2B alter sex determination and expression

Sex determination in individual floral buds of monoecious and dioecious species is controlled by diverse mechanisms that suppress the development of either stamen or carpel primordia in floral buds that will result in female or male flowers, respectively (Fig. 8; Kater et al., 2001; Bai et al., 2004). In the Cucurbitaceae, the main sex regulator is ethylene, but the genetic network controlling sex determination in these species is still poorly understood. With the exception of WIP1, all major genes controlling sex determination encode key enzymes involved in ethylene biosynthesis (ACS and ACO; Fig. 8). The gene ACS2/ACS7 controls monoecy, and loss-of-function mutations lead to plants with male and hermaphrodite flowers (andromonoecy) (Boualem et al., 2008; Boualem et al., 2009; Martínez et al., 2014; Ji et al., 2016; Manzano et al., 2016). Mutations in this gene do not affect sexual expression, that is, the ratio between male and female flowers in the plant (Martínez et al., 2014; Manzano et al., 2016). On the other hand, the genes ACS11 and ACO2 are required for the female flower development pathway, and mutations in these genes result in plants with only male flowers (androecy) (Boualem et al., 2015; Chen et al., 2016). Finally, the transcription factor WIP1 is required for male flower development and is negatively regulated by ACS11, and its dysfunction results in plants with only female flowers (gynoecy) (Boualem et al., 2015; Hu et al., 2017).

So far, no ethylene receptor has been positioned in the genetic network controlling sex determination in this group of species (Fig. 8), although there was some evidence indicating their participation. Thus, the transcription level of ETR2 and ERS1 is higher in gynoecious than monoecious apical shoots of C. sativus (Yamasaki et al., 2000), and and the down-regulation of CsETR1 in the stamens of female cucumber flowers appears to be required for the arrest of stamen development (Wang et al., 2010). Moreover, transgenic C. melo plants overexpressing the Arabidopsis ethylene-insensitive allele etr1-1 are altered in sex determination and sex expression (Little et al., 2007; Switzenberg et al., 2015). Given that etr1a and etr2b not only disrupt female flower development (converting monoecy into andromonoecy) but also significantly increase the number of male flowers in the plant, it is likely that ETR1 and ETR2 integrate the two ethylene biosynthesis pathways that result in the determination of male and female flowers, perceiving and signaling the ethylene produced by ACS2/7 as well as that produced by ACS11 and ACO2 (Fig. 8).

The mechanisms triggering the sex of each specific floral meristem must be regulated by the level of ethylene sensitivity conferred by ethylene receptors (Fig. 9). In fact, the degree of conversion of female flowers into bisexual and hermaphrodite flowers, the delay in pistillate flowering transition, and the increase in the number of male flowers per plant are all correlated with the level of ethylene sensitivity in etr1a and etr2b single and double mutants. The final level of ethylene sensitivity in the tissue will be affected by the strength of each mutation, but also by the number of WT and mutant ethylene receptors in the tissue (Hall et al.,1999), and by the cooperation between them for repressing ethylene signaling (Fig. 9).

The ethylene-insensitive mutations also alter the growth rate of petals and carpels, making mutant bisexual and hermaphrodite flowers reach anthesis later than WT female flowers. This delay in anthesis time could be associated with the development of stamens, since the anthesis time of the mutant hermaphrodite flower was similar to that of the male flower. Moreover, the mutant ovary continues to grow as long as the petals remain green and do not reach anthesis (Fig. 4). These data suggest that fruit set in C. pepo is not triggered by pollination, but is a default developmental program, which has a checking point at anthesis. If the flower is pollinated, development of the ovary continues and the fruit sets. In the absence of pollination and fertilization, growth of the ovary is aborted. Fruit set is known to be regulated positively by hormones such as auxins and gibberellins, and negatively by ethylene (Martínez et al., 2013; Shinozaki et al., 2018; Shnaider et al., 2018). Therefore, the reduction of ethylene sensitivity in the etr1a and etr2b mutants could trigger parthenocarpy by delaying flower anthesis and consequently the checking point for fruit set.

Supplementary data

Supplementary data are available at JXB online.

Table S1. Sequence and coverage statistics.

Table S2. Primers and TaqMan probes used for genotyping etr1a and etr2b mutations.

Table S3. Primers for qRT–PCR analysis.

Table S4. Proteins used to perform the phylogenetic analysis.

Table S5. Segregation of ethylene-sensitive, -intermediate and -insensitive plants in the offspring of backcrossed (BC₁ and BC₂) and selfed (BC₁S₁ and BC₂S₁) generations.

Fig. S1. Distribution of staminate and pistillate flowers in the 40 first nodes of the plant.

Fig. S2. Alignment of the CpETR1A amino acid sequence with homologous sequences from diverse species.

Fig. S3. Alignment of the CpETR2B amino acid sequence with homologous sequences from diverse species.

Conflict of interest

The authors state that no conflict of interest exists regarding this publication.