Horticulture Research, 2022, 9: uhac131 https://doi.org/10.1093/hr/uhac131 Article Modulating climacteric intensity in melon through QTL stacking Miguel Santo Domingo1 , Lorena Areco1 , Carlos Mayobre1 , Laura Valverde1 , Ana Montserrat Martín-Hernández1 , 2 , Marta Pujol1 , 2 , * and Jordi Garcia-Mas1 , 2 , * Downloaded from https://academic.oup.com/hr/article/doi/10.1093/hr/uhac131/6601557 by guest on 05 May 2023 1 Centrefor Research in Agricultural Genomics (CRAG) CSIC-IRTA-UAB-UB, Edifici CRAG, Campus UAB, Bellaterra, 08193 Barcelona, Spain 2 Institut de Recerca i Tecnologia Agoralimentàries (IRTA), Edifici CRAG, Campus UAB, Bellaterra, 08193 Barcelona, Spain *Corresponding authors: E-mail: jordi.garcia@irta.cat, marta.pujol@irta.cat Abstract Fruit ripening is one of the main processes affecting fruit quality and shelf life. In melon there are both climacteric and non-climacteric genotypes, making it a suitable species to study fruit ripening. In the current study, in order to fine tune ripening, we have pyramided three climacteric QTLs in the non-climacteric genotype “Piel de Sapo”: ETHQB3.5, ETHQV6.3 and ETHQV8.1. The results showed that the three QTLs interact epistatically, affecting ethylene production and ripening-related traits such as aroma profile. Each individual QTL has a specific role in the ethylene production profile. ETHQB3.5 accelerates the ethylene peak, ETHQV6.3 advances the ethylene production and ETHQV8.1 enhances the effect of the other two QTLs. Regarding aroma, the three QTLs independently activated the production of esters changing the aroma profile of the fruits, with no significant effects in fruit firmness, soluble solid content and fruit size. Understanding the interaction and the effect of different ripening QTLs offers a powerful knowledge for candidate gene identification as well as for melon breeding programs, where fruit ripening is one of the main objectives. Introduction mation, and other traits are independent, as carotenoid synthesis, sugar accumulation or, partially, flesh soften- Fruit development is one of the most important and energy-consuming processes during plant development. ing [6]. Also, the availability of different mapping popu- Ripening is the last step of this process, producing several lations, such as Recombinant Inbred Lines (RILs) [7, 8] or biochemical and physiological changes in the fruit and Introgression Lines (ILs) [9, 10], together with sequenced making it attractive for animal and human consumption. genome [11], makes melon a suitable model for studying Moreover, ripening has a significant role in plant breeding both climacteric and non-climacteric fruit ripening. as it has a major effect in fruit shelf life and fruit quality. The availability of IL populations makes possible to During this process, fruits undergo several changes, such mendelize and study QTLs minimizing the effect of the as softening due to cell wall degradation or accumulation genetic background. These populations have been used of pigments [1–3]. These changes happen in both climac- as genetic resources for several crops, as tomato [12], teric and non-climacteric fruits. The difference between strawberry [13] or peach [14]. In melon, they have been both types is the presence of a peak in ethylene produc- widely used to study different traits, including ripening tion coupled to an increase in the respiration rate in the [9, 10, 15, 16]. In previous studies, two IL populations were climacteric ones, or its absence in the non-climacteric [4]. used to dissect climacteric ripening in a non-climacteric The classical model organism to study fruit ripening has background. In an IL population with non-climacteric generally been tomato, due to the availability of several genetic background of Piel de Sapo (PS) and the non- mutants with delayed or failed climacteric ripening [3]; climacteric Songwan Charmi (SC) as donor parent, two but, as tomato is a climacteric fruit, there is much less QTLs were found triggering a climacteric response [17, information about non-climacteric ripening. 18]. Located in chromosome 3 and 6, they were named Melon (Cucumis melo L.) has emerged as an interesting ETHQB3.5 and ETHQV6.3, respectively. Besides, in an IL model to study fruit ripening, as it has both climacteric population with the same recurrent parent PS as genetic and non-climacteric varieties within the same species background, and Védrantais (Ved, climacteric) as donor [5]. In melon fruits, some ripening traits are dependent parent, another major QTL was found, located in chro- on ethylene, as aroma production or abscission layer for- mosome 8 and named ETHQV8.1 [19]. Received: 23 March 2022; Accepted: 29 May 2022; Published: 3 June 2022; Corrected and Typeset: 1 August 2022 © The Author(s) 2022. Published by Oxford University Press on behalf of Nanjing Agricultural University. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 2 | Horticulture Research, 2022, 9: uhac131 The effect of these QTLs has been previously studied. When we studied the aroma profile, the three lines ETHQB3.5 was found to provoke the production of a peak behaved differently. In flesh tissue, Ved accumulated of ethylene, compared to PS and SC [17, 18]. ETHQB3.5 the biggest amount of total volatile organic compounds is located within the interval 26 000 631 – 28 759 416 bp (VOCs) followed by SC, and PS was the one accumulating in chromosome 3 (melon genome v3.6.1) [20], but the the least amount of VOCs, being the differences signif- causal gene is still unknown. ETHQV6.3, encoded by icant (p < 0.05) among the three genotypes (Table S2A). a NAC transcription factor (MELO3C016540.2), is also In rind tissue, both PS and SC accumulated low amount capable of producing a climacteric response in PS, and of VOCs, and Ved significantly (p < 0.05) accumulated increase and advance the ethylene production when both more VOCs than the other two genotypes (Table S2B). ETHQB3.5 and ETHQV6.3 alleles from SC are present in Looking at the number of different compounds produced PS [18]. In a climacteric background, MELO3C016540.2 by each line, Ved was also producing the most diverse mutants showed a delayed production of ethylene [21]. aroma profile in both tissues, with around 60 different Downloaded from https://academic.oup.com/hr/article/doi/10.1093/hr/uhac131/6601557 by guest on 05 May 2023 ETHQV8.1, located in the interval 9 603 217 – 9 757 373 bp VOCs, followed by SC with 50 and PS with less than in chromosome 8 (melon genome v3.6.1), has been 30 different VOCs in flesh and less than 20 different also studied in both climacteric and non-climacteric compounds in rind tissue (Figure 1F, I) (Table S2). PS had genetic backgrounds. In a non-climacteric background, a non-climacteric aroma profile, being aldehydes the introgressing the climacteric allele produces a cli- major component in both rind and flesh tissues (92% and macteric response, and in a climacteric background, 55% of total amount of volatile compounds, respectively) introgressing the non-climacteric allele produces a delay with a lesser accumulation of esters (Figure 1G, H, J, K) and a decrease in ethylene production [19]. However, (Table S3). Ved had a major component of esters in we do not know the relationship between these three both rind and flesh (75.5% and 86%, respectively) and QTLs when they are combined in the same line and a minor component of aldehydes, although the total the consequences for melon ripening process and fruit production of aldehydes was similar to PS in flesh and quality. significantly bigger in rind (p < 0.05) (Figure 1G, H, J, K) The aim of this work is to study the effect of three (Table S3). In the case of SC, it had a different VOCs QTLs, ETHQB3.5, ETHQV6.3 and ETHQV8.1, in the non- profile depending on the tissue. Regarding flesh tissue, climacteric background of PS and their possible interac- although the total amount of produced VOCs is half the tions, to further modulate the climacteric response and one found in Ved, it had a similar profile with a major associated traits happening during melon fruit ripening. component of esters (78%) (Figure 1G, H) (Table S3A, B). When focusing into the rind tissue, the production of VOCs was also reduced compared to Ved, and the Results profile also changed, with an intermediate behavior Fruit ripening behavior and fruit quality of the between Ved and PS with a mean of 40% of VOCs being parental lines esters and 40% aldehydes (Figure 1J, K) (Table S3C, D). The three parental lines used in this study had differ- Regarding other minor compounds, SC was the only ent fruit ripening behavior (Figure 1). Piel de Sapo (PS) line producing alkanes and furans in the rind, while is a Spanish variety, belonging to melo subspecies and PS produced furans in the flesh and phenols in rind inodorous group. It produced a low amount of ethylene, (Table S3). insufficient to trigger the climacteric response. We con- In a deeper analysis, when looking into the compounds sidered PS as a classical non-climacteric cultivar [19]. that are most related to melon aroma such as (E, Z)-2,6- Songwan Charmi (SC) is a Korean accession belonging Nonadienal, Nonanal, (Z)-6-Nonenal, Ethyl acetate, Ethyl to the agrestis subspecies, considered non-climacteric, 2-methylbutanoate and Ethyl butanoate (Table S4A) [22], but presenting some phenotypes associated with the although SC is more similar to Ved than to PS regarding climacteric response, as the production of aroma. It also aroma profile, it produces more melon and cucumber- produced ethylene during ripening, although in much like aromas, while Ved produces more floral and fruity less quantity if compared to climacteric types. Védrantais ones (Table S4B, C). Otherwise, PS has a general low pro- (Ved) is a French variety, from the melo subspecies as duction of these VOCs, both in flesh and rind tissues PS, and the cantaloupensis group. It had a typical cli- (Table S4B, C). macteric fruit ripening behavior, with a sharp ethylene peak and noticeable related climacteric traits as abscis- Pyramiding ILs with 2 and 3 QTLs sion layer formation at 35 days after pollination (DAP) We developed a collection of ILs in the PS background (Figure 1A, B). with homozygous introgressions covering one, two or Regarding fruit quality phenotypes, PS produced sig- three QTLs related with climacteric ripening, coming nificantly bigger fruits compared to SC and Ved. Also, from two different parental lines (SC for ETHQB3.5 SC had the smallest fruits of the three parental lines and ETHQV6.3, and Ved for ETHQV8.1) (Figure 2) in (Figure 1C). We did not observe significant differences two generations. The ILs developed in previous studies among the three parental lines in soluble solid content were renamed consistently with the new developed ILs nor firmness of the f lesh (Figure 1D, E) (Table S1). (Table 1). Domingo et al. | 3 Downloaded from https://academic.oup.com/hr/article/doi/10.1093/hr/uhac131/6601557 by guest on 05 May 2023 Figure 1. Phenotypic evaluation of the three parental lines PS, SC and Ved used in the study, showing: A) ethylene production (ETH), B) earliness of abscission layer formation (EALF), C) fruit weight (FW), D) soluble solid content (SSC), E) f lesh firmness (FIR), F) diversity of VOCs in f lesh, G) percentage of esters over the total amount of VOCs in f lesh, H) percentage of aldehydes over the total amount of VOCs in f lesh, I) diversity of VOCs in rind, J) percentage of esters over the total amount of VOCs in rind, K) percentage of aldehydes over the total amount of VOCs in rind. Letters represent significant difference between groups (p-value <0.05). Fruit ripening behavior of the developed ILs Regarding the climacteric symptoms earliness of The main goal of this study was to dissect the ripening aroma production (EARO), earliness of abscission layer behavior of the developed ILs to understand the role of formation (EALF) and harvest date (HAR), a Principal each QTL and their interactions. Component Analysis showed that only one component 4 | Horticulture Research, 2022, 9: uhac131 Downloaded from https://academic.oup.com/hr/article/doi/10.1093/hr/uhac131/6601557 by guest on 05 May 2023 Figure 2. External and internal visualization of the fruits of the lines used in this study, including the parental lines. IL = Introgression Line, SC = Songwan Charmi, PS = Piel de Sapo, Ved = Védrantais. Table 1. Melon lines used in this work. SC = Songwan Charmi, VED = Védrantais Melon linea name QTL (allele-code) Previous line name Reference 1QSC3 SC-ETHQB3.5 8 M35 Moreno et al. 2008 [17] Vegas et al. 2013 [18] 1QSC6 SC- ETHQV6.3 8 M40 Vegas et al. 2013 [18] 1QVED8 VED-ETHQV8.1 VED8.2 Pereira et al. 2020 [19] 2QSC36 SC-ETHQB3.5 SC- ETHQV6.3 8 M31 Vegas et al. 2013 [18] 2QSCVED38 SC-ETHQB3.5 VED-ETHQV8.1 - This work 2QSCVED68 SC- ETHQV6.3 VED-ETHQV8.1 - This work 3QSCVED368 SC-ETHQB3.5 SC- ETHQV6.3 - This work VED-ETHQV8.1 a The genetic background of all the lines is Piel de Sapo (PS) explained 98% of the variation (Figure S2). EALF is the 0.005). While in PS the abscission level is low, when trait which covariates more with this component, so, introgressing one of the QTLs coming from SC most we analyzed this trait as representative of the three of the melons presented complete abscission from the main climacteric symptoms. After model selection, plant. On the other hand, when introgressing the Ved we selected the multiple linear model with the single allele from ETHQV8.1 the level of abscission increased effects and the three double interactions as the best but not as much as for SC alleles in chromosome 3 or fitting model (R2 = 0.86). Taking as reference PS, with 6 (Figure S4). the non-climacteric alleles for the three QTLs, the The ethylene production of each of the ILs developed linear model showed a significant predicted advance in is presented in Figure 4A. With the single introgression the symptoms for ETHQB3.5, ETHQV6.3 and ETHQV8.1 of any of the QTLs, PS turned climacteric, but in dif- (Figure 3) (Table 2). Also, all the double interactions ferent degrees. While ETHQB3.5 and ETHQV6.3 provoked were predicted as significant (Figure 3) (Table 2). As seen an important peak of ethylene production around 36– in the model, the three double interaction effects are 37 DAP, ETHQV8.1 only triggered a flat peak of ethy- significantly “less than additive”, noticed in the different lene production around 43–45 DAP. In the double ILs, we signs of the three main effects compared to the three observed a diverse degree of climacteric, from an early interaction effects. The model predicts a minor advance one in 2QSC36 to a late one in 2QSCVED38. Also, we can in EALF when introgressing more than one QTL than the observe that when pyramiding more than one QTL, an expected by the addition of the main effects of these earlier climacteric response and a higher production of QTLs (Table 2) (Figure S3). ethylene was induced. With the triple IL 3QSCVED368 Regarding the semi-quantitative ripening-related level we observed the sum of both effects, earlier and higher of abscission (ABS), it shows a statistically significant production, being the line with the highest climacteric interaction between ETHQB3.5 and ETHQV6.3 (p-value = behavior. While Ved had its ethylene production peak at Domingo et al. | 5 Downloaded from https://academic.oup.com/hr/article/doi/10.1093/hr/uhac131/6601557 by guest on 05 May 2023 Figure 3. A) Representation of the phenotype Earliness of abscission layer formation (EALF) in the ILs. Significance based on t-test p-value <0.05. B), C), and D) Interaction plots showing the two-by-two interactions of the three studied QTLs. Significance is calculated through a multiple linear model. Table 2. Multiple linear model used to analyze the earliness of abscission layer formation (EALF) in the ILs. R2 = 0.86 Coefficients: Effect Estimate Std. Error t value Pr(>|t|) Significance PS Intercept 53.16 0.97 54.715 < 2e-16 ∗∗∗ ETHQB3.5-SC Main −17.29 1.20 −14.397 < 2e-16 ∗∗∗ ETHQV6.3-SC Main −16.78 1.18 −14.17 < 2e-16 ∗∗∗ ETHQV8.1-VED Main −8.73 1.20 −7.272 6.19E-10 ∗∗∗ ETHQV6.3-SC:ETHQV8.1-VED Interaction 3.16 1.35 2.343 0.022226 ∗ ETHQB3.5-SC:ETHQV8.1-VED Interaction 5.17 1.35 3.835 0.000289 ∗∗∗ ETHQB3.5-SC:ETHQV6.3-SC Interaction 8.59 1.34 6.415 1.95E-08 ∗∗∗ 36 DAP (Figure 1), 3QSCVED368 had it at 28 DAP, one week Aroma production and fruit quality of the earlier. developed ILs When we analyzed ethylene production related traits, Another important trait for fruit quality in melon is the we saw that for earliness of ethylene production and the volatile compounds composition, as their combination peak of this production, a significant effect of ETHQV6.3 shapes the aroma profile of the fruit. The introgression was detected, and the interaction between the two of the climacteric ripening QTLs produced a significant QTLs ETHQV8.1 and ETHQV6.3 and the triple one was increase in the total amount of volatiles production, also significant (Figure 4B, Table S5A, B). Concerning the compared to PS, in both fruit flesh and rind (Figure 5) SLOPE, which is related with the sharpness of the peak (Table S2A, B) Analyzing the different compounds, we explaining how quick the ethylene production increases, can observe that, although there was variability within the model predicted a significant effect of ETHQB3.5 each line, the most important effect was due to the narrowing the peak, and also an interaction between accumulation of esters (mainly acetic acid esters) in ETHQV6.3 and ETHQV8.1 when they were together fruit flesh and rind, and the accumulation of aldehy- (Figure 4C, Table S5C). For ethylene production related des (mainly benzaldehyde) in the rind (Figure S5). In PS, traits, our data suggest that ETHQV6.3 has a significant there was little production of esters in both rind and effect for the earliness of the peak, and ETHQB3.5 for its flesh, but when we introgressed the QTLs, the fruits sharpness. On the other hand, although ETHQV8.1 does significantly accumulated a bigger quantity of esters in not have a significant effect on its own, it enhanced the flesh, even reaching the levels of esters production in effect of the other two QTLs. Ved (Table S3A) (Figure 5A). There was also an increase 6 | Horticulture Research, 2022, 9: uhac131 Downloaded from https://academic.oup.com/hr/article/doi/10.1093/hr/uhac131/6601557 by guest on 05 May 2023 Figure 4. A) Ethylene production (ETH) of the ILs during fruit ripening, and in dashed lines, the two donor parents Ved and SC. B) earliness of ethylene production (DAPE) and C) slope of the log10 transformed data in the exponential part of the peak (SLOPE). Significance fixed at p-value <0.05. of ester production in rind, although generally, the levels of alcohols when introgressing ETHQV8.1, approaching reached were not higher than Ved (Table S3C) (Figure 5B). the levels of SC and Ved in flesh and rind (Table S3A, C). In both rind and flesh, the accumulation of esters became Also, when introgressing the QTLs, the fruits produced the main component of the aroma profile in the intro- more terpenes, especially the lines with ETHQB3.5, but gression lines (more than 75% and 50% of total volatiles not reaching the levels of Ved (Table S3A, C). in flesh and rind, respectively), similar to the climacteric Moreover, we studied the association of all the VOCs cultivar Ved and significantly higher than in PS in all with individual QTLs, considering the large variability the pyramided lines (Table S3B, D) (Figure 5C, D). Regard- within genotypes. Using linear models, we selected the ing the aldehydes (the second group in importance in additive model to see the individual association of each melon fruit aroma profile), the total amount was very single QTL with the compounds, and then selected the variable. In flesh, there was not a significant increase in models which explained more than 40% of the variance any of the lines, even in the other parents Ved and SC (adjusted R2 > 0.4), indicating a major genetic control (Table S3A) (Figure 5E). In rind tissue, although we did of the trait by the QTLs individually. In the case of observe a significant increase in some lines, the behav- rind, two compounds met this condition: eucalyptol and ior was very variable without a clear trend (Table S3C, benzeneacetaldehyde (Table S6). Both were associated Figure 5F). However, looking at the percentage of aldehy- significantly with ETHQB3.5, and highly associated des we did observe a significant trend. While in PS the with ETHQV6.3 in the case of benzenacetaldehyde, and aldehydes were the main contributor to the aroma profile with ETHQV8.1 in the case of eucalyptol (Table S6A, B). in both rind and flesh (more than 90% and 50% respec- When focusing on the flesh, four compounds were tively), in the introgression lines the percentage was sig- associated with the QTLs. Eucalyptol appeared also nificantly lower in both tissues probably due to the high associated with ETHQV8.1 and ETHQB3.5 in flesh, and percentage of esters (Table S3B, D) (Figure 5G, H). If we phenylethyl alcohol with ETHQV8.1 and ETHQV6.3 look at the other analyzed compounds, we can observe (Table S6C, D). Moreover, another compound, 1-octen- that when we introgressed the climacteric QTLs, the lines 3-ol was highly associated with ETHQV8.1 (Table S6F). lost the capacity of accumulating furans in fruit flesh All of these compounds increased their production with (Table S3A). We can also observe a bigger accumulation the donor parent allele. Interestingly, the last compound Domingo et al. | 7 Downloaded from https://academic.oup.com/hr/article/doi/10.1093/hr/uhac131/6601557 by guest on 05 May 2023 Figure 5. Production of esters and aldehydes, showing both total relative production respect to the internal standard (3-hexanone) and percentage over the total VOCs. A), C), E) and G) in f lesh tissue, and B), D), F) and H) in rind tissue. highly associated with one QTL was propanoic acid, 2- of the ILs (Table S2A). Regarding the melon-like aroma methyl-, ethyl ester, one of the compounds described compounds nonanal and 6-nonenal, we detected a signif- as main contributor of melon aroma (Table S4A). We icant increase of nonanal in flesh in the line 3QSCVED368 detected a highly significant association (p < 0.01) of (p < 0.05), and non-significant increase in other intro- this compound with ETHQV8.1, increasing its production gression lines, such as 1QSC6 and 2QSC36, when com- when introgressing the donor allele (Table S6E). pared to PS (Table S2A). The major effect was observed in Regarding the analysis of the main contributors to the floral and fruity aroma compounds. By introgressing the melon aroma (Table S4A), we can observe differ- the QTLs, we could mostly recover the phenotype of Ved ent behaviors. Although SC produced the cucumber-like in both rind and flesh for those compounds. Individu- aroma compound 2–6 nonadienal in rind, none of the ally, there was an increase in the production of ethyl ILs recovered this phenotype, while we observed non- acetate, and butanoic acid ethyl ester in 1QVED8, pro- significant increase of its production in f lesh in most ducing bigger amounts than Ved, but we did not observe 8 | Horticulture Research, 2022, 9: uhac131 this trend in other lines carrying ETHQV8.1. In these climacteric symptoms. ETHQV8.1 was the weakest compounds, the lesser effect was observed when intro- QTL, provoking a lower and delayed ethylene peak gressing ETHQB3.5, being 1QSC3, 2QSC36, 2QSCVED38 (Figure 4A). ETHQB3.5 caused a fast production of the and 3QSCVED368 similar to PS (Table S2A, B). For the ethylene peak (Figure 4C), and ETHQV6.3 significantly other compounds, although there was an increase in advanced both the ethylene production and the peak the ILs compared to PS, we did not observe significant intensity (Figure 4B). When observing the climacteric changes or trends (Table S2). symptoms, the analysis showed a significant effect For the fruit quality traits, there was no significant when introgressing ETHQV6.3, ETHQB3.5 or ETHQV8.1 effect of either of the QTLs, and we did not observe (Table 2), being the effect bigger for the first two QTLs. significant differences among the lines compared to PS This bigger effect of the QTLs coming from SC was for fruit weight, sugar content or firmness (Figure S6). previously observed [18, 19] and ETQHQV8.1 has been However, we detected a trend of smaller fruits for any of reported to be unstable, provoking a mild climacteric Downloaded from https://academic.oup.com/hr/article/doi/10.1093/hr/uhac131/6601557 by guest on 05 May 2023 the introgressed QTLs (Figure S6A). behavior depending on the environmental conditions [19]. When combining two of the QTLs in the same line, we observed a statistical interaction in the three Discussion combinations, bringing out a potential interaction among In this study, we present the pyramiding of three QTLs the causal genes. The general trend of these QTLs is a involved in climacteric ripening coming from two differ- “less than additive” effect, being this model significant ent populations sharing the non-climacteric recurrent for the three double interactions, predicting a lighter parental line PS. While ETHQB3.5 and ETHQV6.1 come effect when introgressing two QTLs that the expected from an IL population with the non-climacteric donor if it was additive (Table 2) (Figure S3). As melons are parent Songwan Charmi PI 161375 [9, 23], ETHQV8.1 reported to start ripening at around 30 DAP, coinciding comes from an IL population with the climacteric Ved with the starting of sucrose accumulation [35], maybe as the donor parent [19]. This pyramiding strategy has the mathematically estimated additive value is too early, been used in different crops and with several traits, during the fruit growth stage. Melons carrying two or such as improving yield under stress conditions or blast three of the QTLs should ripe before 20 DAP when resistance in rice [24, 25], fungi resistance in pea [26] or the fruit is still under development and probably it is insect resistance in soybean [27]. Similar strategies, but not ready for ripening. Thus, linear models seem an with a single donor parent, have been used to improve appropriate tool for this type of experiment, as they can yield in different crops, as cucumber [28] or tomato detect statistical interactions among variables, in this [29]. In the case of melon, there have been studies of case QTLs, and have been used previously for this kind of pyramided QTLs to improve resistance to gummy stem studies [26, 29]. After model selection, the most suitable blight [30] and to advance ripening [18]. In all of these model was the one with the three double interactions studies, the use of ILs with combinations of different that explains 88.55% of the variance, evidencing also QTLs has proved as a powerful tool to uncover their that there is no need to include the triple interaction effect in diverse phenotypes, as well as their interactions. to predict the ripening behavior of the population. This However, pyramiding genes involved in fruit quality significance of all the interactions may be indicating or ripening has not been yet fully implemented in that the phenotype is due to the interactions between cultivated crops [31]. With the genomic tools available the different alleles of those QTLs. in melon, such as the reference genome and more than When we looked individually to each of the QTLs, one thousand accessions resequenced [11, 32], molecular we observed a clear effect of them in the ethylene markers can be easily obtained and used for shortening production profile. ETHQV6.3 advanced the ethylene the process of pyramiding QTLs by MAS. production, as well as the ethylene peak. This result matches previous studies showing that TILLING mutants QTLs interact to modulate the climacteric of MELO3C016540.2 in a climacteric background delay ripening behavior in melon fruits the production of ethylene but not the amount or the To date, most of the work about climacteric ripening shape of the peak [21]. In the case of ETHQB3.5 and has been done in tomato fruits. Several mutants with ETHQV8.1, although the causal genes are not yet known, delayed ripening have been used to characterize genes our results could help in their identification. Our results controlling climacteric response [3, 33, 34]. However, suggest that the role of ETHQB3.5 is to narrow the inducing early climacteric ripening in a non-climacteric ethylene peak, decreasing the days from the start of the background, as we present in this work, has not been ethylene production to the maximum production. In a addressed. recent study, two candidate genes have been suggested in Although the effects of the three QTLs here described the ETHQB3.5 interval: MELO3C011432.2, a WRKY family have been studied separately, this is the first time that transcription factor, and MELO3C011365.2, related with they are combined and analyzed in the same experiment. signal transduction [36]. ETHQV8.1, although it has a The three QTLs produce a climacteric response in PS weaker effect compared to the other two QTLs, has an independently, with production of ethylene and related important role in enhancing the response in combination Domingo et al. | 9 with both of them. Three candidate genes have been to be expressed later than 30 DAP in melon [38], later than suggested in the region of this QTL: MELO3C024516.2, the harvest date of 3QSCVED368, probably leading to a MELO3C024518.2 and MELO3C024520.2, encoding a minor accumulation of esters in this line compared to the demethylase, a negative regulator of ethylene signal other ILs. We also detected changes in the composition transduction and an ethylene responsive transcription of the aroma in the pyramided lines. The introgression factor, respectively [19]. All three of them may play a of ETHQV8.1 from Ved enhances the production of fruity role in fruit ripening, and our work can contribute to and floral aromas, and the introgression of ETHQB3.5 understand the specific function of ETHQV8.1. does not produce a strong effect in the main contributors Our results suggested an important role of the genetic to melon aroma (Table S4). Therefore, even if ETHQV8.1 background, since ETHQB3.5 and ETHQV6.3, both carry- has not the highest impact in ethylene production, it ing SC alleles, behave differently when introgressed in PS. seems to affect key VOCs implicated in melon aroma. This effect reveals the complex architecture of ripening Our results suggest a potential use of different QTLs to Downloaded from https://academic.oup.com/hr/article/doi/10.1093/hr/uhac131/6601557 by guest on 05 May 2023 in melon, as it has been reported before in other crops shape the aroma profile in melon. Although we observed [31]. Moreover, when introgressing two (2QSCVED68) or a high variation regarding VOCs accumulation in both three (3QSCVED368) QTLs in PS, we can obtain a line that rind and flesh tissues, some compounds seemed to be is even more climacteric than Ved, which is known to be a associated with a particular QTL, making it possible to typical climacteric cantaloupe type. This finding suggests fine-tune the melon aroma by pyramiding different QTLs. that there are some key genes governing the climacteric Moreover, we demonstrated that is possible to introduce response, and combining their natural variation, we can aroma in an inodorous cultivar by introgressing any of convert a non-climacteric variety with long shelf life the three QTLs, suggesting that the control of aroma into an early maturation and aromatic melon line. production depends mainly on ethylene production. By combining these three QTLs we can obtain mel- ons with different degrees of climacteric response. As Minor effect in fruit quality caused by the ethylene production is one of the main traits affecting introgressions fruit shelf life, here we have implemented a tool to con- Our results showed that there were not significant differ- trol its production with only three major QTLs that can ences between PS and the pyramided lines related to fruit potentially be fine-tuned using other minor QTLs found quality traits as weight, soluble solid content and firm- in previous studies [19]. This approach could be used to ness. However, there was a trend showing a decrease in develop new melon cultivars with different degrees of fruit weight in all the ILs. A relationship between ripening shelf life in both directions. We could develop aromatic and fruit size has been reported in tomato [31] and melon climacteric varieties from a non-climacteric variety as [39]. However, the ILs here reported contain large regions shown in this study; and from a climacteric cultivar as with many genes, making it possible that the effect on Ved, pyramiding ILs with delayed ripening [10, 16], we fruit size may be the result of other genes present inside could develop long shelf life climacteric varieties. the introgression. In the case of fruit firmness, although the softening of the fruit is partially regulated by ethylene The aroma profile can be modified in an inodorus [6], we did not detect any significant effect. melon The results of this study demonstrate the possibility Piel de Sapo (PS), the recurrent parent of the ILs, belongs of pyramiding different QTLs in the same genetic back- to the inodorous group, the term meaning “no aroma” in ground to shape and fine-tune the ripening process in Latin. As a non-climacteric variety, PS is characterized by melon fruits. With only three QTLs, we can obtain a a mild or even undetectable aroma profile. In this study, line with an extreme climacteric behavior in the non- we introgressed different QTLs in the PS background climacteric cultivar Piel de Sapo background. In addition, triggering a climacteric response during ripening, and all we can modify the aroma profile of the ripe fruit using of them induced accumulation of ester compounds and these QTLs. The combination of these three major QTLs a meaningful change in the aroma profile of the fruit. represents a powerful tool to modify and fine-tune cli- Regarding the aroma profile, we cannot differentiate the macteric fruit ripening in melon breeding programs. pyramided lines when smelling them, as all of them are climacteric and produce a similar aroma. The analysis of melon fruits by GC–MS showed that for the climac- Materials and methods teric lines, the main contributor to the aroma profile Plant material are esters [22, 37], while the non-climacteric PS does not The starting lines derive from two different IL popula- produce esters, being aldehydes the main contributors tions, both constructed in the genetic background of the to the aroma profile. The line 3QSCVED368, the earliest Piel de Sapo (PS) variety (inodorous group). Three of the climacteric line, had a lower production of esters than initial lines belong to a previously characterized IL popu- the other lines, suggesting that this effect may be due lation [9, 18] with the Korean accession Songwan Charmi to its quick ripening and fruit abscission, preventing the (SC, PI 161375) as the donor parent. These three lines are accumulation of esters. Alcohol acyl-transferases, the 1QSC3 (previously known as 8 M35), 1QSC6 (previously enzymes driving esters biosynthesis, have been reported named 8 M40) and 2QSC36 (known as 8 M31), carrying 10 | Horticulture Research, 2022, 9: uhac131 Table 3. Traits evaluated in the experiment Category Trait (units) Code Fruit weight (g) FW Fruit quality Soluble solid content (◦ Brix) SSC Flesh firmness (kg cm−2 ) FIR Earliness of aroma production (DAP∗ ) EARO Aroma production ARO Fruit ripening Earliness of abscission layer formation (DAP∗ ) EALF Level of abscission ABS Harvest date (DAP∗ ) HAR Ethylene production (μL kg−1 h−1 ) ETH Earliness of ethylene production (DAP∗ ) DAPE Downloaded from https://academic.oup.com/hr/article/doi/10.1093/hr/uhac131/6601557 by guest on 05 May 2023 Ethylene production Earliness of the ethylene peak (DAP∗ ) DAPP Slope of the peak SLOPE a DAP Days after pollination the QTLs ETHQB3.5, ETHQV6.3 and both, respectively. The flanking markers (ETHQB3.5 and ETHQV8.1). ETHQV6.3 second IL population used in this study has the variety was genotyped using a marker designed into the causal Védrantais (Ved, cantalupensis) as the donor parent, and is gene, MELO3C016540.2 [21]. still under construction. The line 1QVED8 was previously described as VED8.2, carrying the QTL ETHQV8.1 [19]. Growing conditions For a more understandable and consistent naming, the Seedlings were germinated and grown in an indoor initial lines were renamed. All lines are described in greenhouse (CRAG, Barcelona) under controlled con- Table 1. ditions for three weeks. Growing of selected plants was performed in a greenhouse (Caldes de Montbui, Breeding scheme Barcelona) in randomized blocks. In flowering time, The pyramided ILs with double and triple introgressions pollinations were manually performed until one fruit were developed by crosses and Marker Assisted Selection per plant was obtained. For the development of the (Figure S1). To construct the double ILs 2QSCVED38 and new lines, two generations were obtained in the same 2QSCVED68, two single ILs were first crossed and then year during 2018 and 2019, and the phenotyping was self-pollinated. 192 F2 plants of each cross were geno- performed during summer 2020. The harvest point was typed with f lanking markers selecting two individual determined based on the presence of abscission layer, as plants containing the two introgressions in homozygo- this is a trait highly associated with ethylene production. sis, which were self-pollinated to obtain the progeny Harvest date was fixed at: a) abscission date when the for phenotyping. For the triple IL, the already devel- fruit abscised; b) five days after the appearance of the oped double introgression line 2QSC36 was crossed with abscission layer; c) 56 days after pollination (DAP) when IL 1QVED8, and then self-pollinated. 384 F2 individuals fruits were non-climacteric and did not form abscission were genotyped and two plants carrying the three QTLs layer, considering them fully ripe at that point. For each in homozygosis were selected, which were self-pollinated genotype, ten plants were grown and depending on the to obtain the progeny for phenotyping. With this scheme, fruit set of each genotype more than seven fruits were 7 lines with all the possible homozygous combinations of evaluated. the three QTLs were developed in two generations, and another generation was needed to multiply the seeds. Phenotyping All traits related to fruit quality and ripening behavior DNA extraction and genotyping are listed in Table 3. Traits were classified according to DNA was extracted from young leaves with two different three different categories: fruit quality (fruit weight (FW), protocols depending on the use of the DNA: for quick soluble solid content (SSC) and firmness (FIR)); fruit genotyping and short-term storage, alkaline-lysis extrac- ripening-related (aroma production (ARO), earliness of tion was used [40], and for genotyping and long-term aroma production (EARO), earliness of abscission layer storage, a CTAB protocol [41] with some modifications formation (EALF), level of abscission (ABS) and harvest was used [8]. date (HAR)); and ethylene-related (maximum ethylene The genotyping was performed using two SNP geno- production (ETH), earliness of ethylene production typing systems: KASPar SNP Genotyping System (KBio- (DAPE) defining the day when ethylene was detected, sciences, Herts, UK), and PACE2.0 SNP Genotyping earliness of the ethylene peak (DAPP) and the slope of System (3CR Bioscience, Essex, UK). Primers were the peak measuring the log10 transformed data in the designed following each genotyping system instructions exponential part of the peak (SLOPE)). ARO and ABS (Table S7). Two of the QTLs were genotyped using were inspected daily from 20 days after pollination (DAP) Domingo et al. | 11 until their appearance. ABS was recorded according to PCAs were performed with the R package “factoextra”. the following scale: 0 (absence of abscission layer); 1 The multiple linear models were performed with the (partial abscission layer formation); 2 (formation of an function “lm”, using as factor the correspondent allele of almost complete abscission layer and presence of a scar); each QTL. Model selection was performed under Akaike and 3 (complete abscission layer formation and fruit Information Criterion. ANOVA and pairwise t-test were abscission). performed using R package “rstats” with Holm correction. Fruit f lesh firmness was recorded at harvest in differ- In general, significance was fixed at p-value <0.05. ent regions of the fruit (proximal, median and distal) with a penetrometer (Fruit Test™, Wagner Instruments), and the mean value was calculated. Flesh juice was used to Acknowledgements measure total soluble solids content (Brix index) with a This work was supported by grants AGL2015–64625-C2– refractometer (Atago™). Fruit weight was measured at 1-R, RTI2018-097665-B-C2, SEV-2015-0533 and CEX2019- Downloaded from https://academic.oup.com/hr/article/doi/10.1093/hr/uhac131/6601557 by guest on 05 May 2023 harvest. 000902-S funded by Ministerio de Ciencia e Inno- vación MCIN/AEI/10.13039/501100011033 and FEDER, Ethylene production and by the CERCA Programme/Generalitat de Catalunya Gas chromatography – mass spectrometry (GC–MS) was and 2017 SGR 1319 grant from the Generalitat de used to measure ethylene production of melon fruits in Catalunya to JGM. M.S.D. and L.V. were supported by planta as previously described [42]. Brief ly, melon fruits grants BES-2017-079956 and PRE2018-086627 funded by were covered with a sealed plastic bag, and then filled MCIN/AEI/10.13039/501100011033 and by “ESF Investing with atmospheric air. After 1 h, gas sample was extracted in your future”. CM was supported by 2019 FI_B 00124 fel- using a syringe and introduced in a vial for GC–MS lowship from the Secretaria d’Universitats i Recerca del analysis. Departament d’Empresa i Coneixement de la Generalitat Ethylene production was monitored from 25 DAP until de Catalunya and the co-funding of the European Social fruit harvest. For the ILs and Ved, ethylene was measured Fund (ESF “ESF is investing in your future”) from the every other day while it was undetectable and every day European Union. Special thanks to Fuensanta García for when first detected. For the non-climacteric lines (PS and assistance in field and lab operation and Martí Bernardo SC), we measured the atmosphere of the chamber every for assistance in data analyses. three days. In total, three to four fruits were analyzed per line. Data was expressed under the four ethylene-related Author contributions traits: ETH, DAPE, DAPP and SLOPE (Table 3). J.G.-M. and M.P. conceived and designed the research. M.S.D., LA, LV and CM performed the experiments. AMM- Aroma profiling H contributed in the management of the plant material. The aroma profiling of the rind and the f lesh tissue of M.S.D. analyzed the data and wrote the original draft. melon fruits was analyzed at harvest with GC–MS as M.P. and J.G.-M. reviewed and edited the manuscript. All previously described [22]. In summary, 2 g of grinded authors read and approved the manuscript. frozen tissue were added to a vial, with 7 ml of saturated NaCl solution and 3-hexanone as internal standard. Until gas chromatograph (GC) mass spectrometer (MS) anal- Data availability statement ysis was performed, tubes were stored up to one week The authors declare that the data supporting the study at 4◦ C in the dark. Vials were analyzed with GC–MS findings are presented in the article and additional using Solid Phase Micro Extraction (SPME) in a 7890A GC supporting files are available from the corresponding coupled to a 5975C MS and a GC PAL 80 autosampler authors (J.G.-M. and M.P.) upon request. (Agilent Technologies® , Santa Clara, CA). Volatile organic compounds (VOCs) were identified by comparing their mass spectra with both the NIST 11 library and their Conflict of interest Kovats retention index. Comparison to the 3-hexanone The authors declare that they have no conflict of interest. internal standard peak area was used to estimate the relative content of each compound. Supplementary data Statistical analyses Supplementary data is available at Horticulture Research Statistical analyses and graphical representations were online. performed using the software R v3.5.3 with the RStudio v1.0.143 interface [43]. Data should not contain any missing data in order to References compute the analyses. Therefore, the harvest day value 1. Seymour GB, Manning K, Eriksson EM et al. Genetic identifica- was imputed to non-climacteric fruits for EARO and EAFL tion and genomic organization of factors affecting fruit texture. when these traits were absent. J Exp Bot. 2002;53:2065–71. 12 | Horticulture Research, 2022, 9: uhac131 2. Prasanna V, Prabha TN, Tharanathan RN. Fruit ripening 22. Mayobre C, Pereira L, Eltahiri A et al. Genetic dissection of aroma phenomena-an overview. Crit Rev Food Sci Nutr. 2007;47: biosynthesis in melon and its relationship with climacteric 1–19. ripening. Food Chem. 2021;353:129484. 3. Giovannoni JJ. Fruit ripening mutants yield insights into ripening 23. Argyris JM, Pujol M, Martín-Hernández AM et al. Combined use control. Curr Opin Plant Biol. 2007;10:283–9. of genetic and genomics resources to understand virus resis- 4. Giovannoni JJ. Genetic regulation of fruit development and tance and fruit quality traits in melon. Physiol Plant. 2015;155: ripening. Plant Cell. 2004;16:S170–80. 4–11. 5. Ezura H, Owino WO. Melon, an alternative model plant for 24. Sandhu N, Dixit S, Swamy BPM et al. Marker assisted breeding to elucidating fruit ripening. Plant Sci. 2008;175:121–9. develop multiple stress tolerant varieties for flood and drought 6. Pech JC, Bouzayen M, Latché A. Climacteric fruit ripening: prone areas. Rice. 2019;12:8. ethylene-dependent and independent regulation of ripening 25. Guan H, Hou X, Jiang Y et al. Feature of blast resistant near- pathways in melon fruit. Plant Sci. 2008;175:114–20. isogenic lines using an elite maintainer line II-32B by marker- 7. Périn C, Hagen L, de Conto V et al. A reference map of Cucumis assisted selection. J Plant Pathol. 2019;101:491–501. Downloaded from https://academic.oup.com/hr/article/doi/10.1093/hr/uhac131/6601557 by guest on 05 May 2023 melo based on two recombinant inbred line populations. Theor 26. Lavaud C, Baviere M, le Roy G et al. Single and multiple resis- Appl Genet. 2002;104:1017–34. tance QTL delay symptom appearance and slow down root 8. Pereira L, Ruggieri V, Pérez S et al. QTL mapping of melon fruit colonization by Aphanomyces euteiches in pea near isogenic quality traits using a high-density GBS-based genetic map. BMC lines. BMC Plant Biol. 2016;16:166. Plant Biol. 2018;18:324. 27. Ortega MA, All JN, Boerma HR et al. Pyramids of QTLs 9. Eduardo I, Arús P, Monforte AJ. Development of a genomic enhance host–plant resistance and Bt-mediated resistance to library of near isogenic lines (NILs) in melon (Cucumis melo L.) leaf-chewing insects in soybean. Theor Appl Genet. 2016;129: from the exotic accession PI161375. Theor Appl Genet. 2005;112: 703–15. 139–48. 28. Robbins MD, Casler MD, Staub JE. Pyramiding QTL for multiple 10. Pereira L, Santo Domingo M, Argyris J et al. A novel introgression lateral branching in cucumber using inbred backcross lines. Mol line collection to unravel the genetics of climacteric ripening Breeding. 2008;22:131–9. and fruit quality in melon. Sci Rep. 2021;11:11364. 29. Gur A, Zamir D. Mendelizing all components of a pyramid of 11. Garcia-Mas J, Benjak A, Sanseverino W et al. The genome of three yield QTL in tomato. Front Plant Sci. 2015;6:1096. melon (Cucumis melo L.). Proc Natl Acad Sci U S A. 2012;109: 30. Zhang N, Xu BH, Bi YF et al. Development of a muskmelon 11872–7. cultivar with improved resistance to gummy stem blight and 12. Bernacchi D, Beck-Bunn T, Emmatty D et al. Advanced backcross desired agronomic traits using gene pyramiding. Czech J Genet QTL analysis of tomato. II. Evaluation of near-isogenic lines car- Plant Breed. 2017;53:23–9. rying single-donor introgressions for desirable wild QTL-alleles 31. Wang R, Lammers M, Tikunov Y et al. The rin, nor and Cnr derived from Lycopersicon hirsutum and L. pimpinellifolium. spontaneous mutations inhibit tomato fruit ripening in additive Theor Appl Genet. 1998;97:170–80. and epistatic manners. Plant Sci. 2020;294:110436. 13. Urrutia M, Rambla JL, Alexiou KG et al. Genetic analysis of 32. Zhao G, Lian Q, Zhang Z et al. A comprehensive genome vari- the wild strawberry (Fragaria vesca) volatile composition. Plant ation map of melon identifies multiple domestication events Physiol Biochem. 2017;121:99–117. and loci influencing agronomic traits. Nat Genet. 2019;51: 14. Serra O, Donoso JM, Picañol R et al. Marker-assisted introgression 1607–15. (MAI) of almond genes into the peach background: a fast method 33. Vrebalov J, Ruezinsky D, Padmanabhan V et al. A MADS-box gene to mine and integrate novel variation from exotic sources in long necessary for fruit ripening at the tomato ripening-inhibitor intergeneration species. Tree Genet Genomes. 2016;12:96. (Rin) locus. Science. 2002;296:343–6. 15. Castro G, Perpiñá G, Monforte AJ et al. New melon introgression 34. Manning K, TÖr M, Poole M et al. A naturally occurring epi- lines in a Piel de Sapo genetic background with desirable agro- genetic mutation in a gene encoding an SBP-box transcrip- nomical traits from dudaim melons. Euphytica. 2019;215:169. tion factor inhibits tomato fruit ripening. Nat Genet. 2006;38: 16. Perpiñá G, Cebolla-Cornejo J, Esteras C et al. ‘MAK-10’: a long 948–52. shelf-life Charentais breeding line developed by introgression 35. Saladié M, Cañizares J, Phillips MA et al. Comparative tran- of a genomic region from Makuwa melon. HortScience. 2017;52: scriptional profiling analysis of developing melon (Cucumis melo 1633–8. L.) fruit from climacteric and non-climacteric varieties. BMC 17. Moreno E, Obando JM, Dos-Santos N et al. Candidate genes and Genomics. 2015;16:440. QTLs for fruit ripening and softening in melon. Theor Appl Genet. 36. Oren E, Tzuri G, Dafna A et al. QTL mapping and genomic analy- 2008;116:589–602. ses of earliness and fruit ripening traits in a melon recombinant 18. Vegas J, Garcia-Mas J, Monforte AJ. Interaction between QTLs inbred lines population supported by de novo assembly of their induces an advance in ethylene biosynthesis during melon fruit parental genomes. Hortic Res. 2022;9:uhab081. ripening. Theor Appl Genet. 2013;126:1531–44. 37. Obando-Ulloa JM, Moreno E, García-Mas J et al. Climacteric or 19. Pereira L, Santo Domingo M, Ruggieri V et al. Genetic dis- non-climacteric behavior in melon fruit: 1. Aroma volatiles. section of climacteric fruit ripening in a melon popula- Postharvest Biol Technol. 2008;49:27–37. tion segregating for ripening behavior. Hortic Res. 2020;7: 38. Yahyaoui FEL, Wongs-Aree C, Latché A et al. Molecular and 187–18. biochemical characteristics of a gene encoding an alcohol acyl- 20. Pereira L. Genetic dissection of fruit quality and ripening traits transferase involved in the generation of aroma volatile esters in melon. PhD thesis, Universitat Autònoma de Barcelona, during melon ripening. Eur J Biochem. 2002;269:2359–66. Barcelona, 2018. 39. Chatzopoulou F, Sanmartin M, Mellidou I et al. Silencing of 21. Ríos P, Argyris J, Vegas J et al. ETHQV6.3 is involved in melon ascorbate oxidase results in reduced growth, altered ascorbic climacteric fruit ripening and is encoded by a NAC domain acid levels and ripening pattern in melon fruit. Plant Physiol transcription factor. Plant J. 2017;91:671–83. Biochem. 2020;156:291–303. Domingo et al. | 13 40. Lu J, Hou J, Ouyang Y et al. A direct PCR–based SNP marker– tification of ethylene in attached fruit headspace at 1 p.p.b. assisted selection system (D-MAS) for different crops. Mol Breed- by gas chromatography–mass spectrometry. Plant J. 2017;91: ing. 2020;40:9. 172–83. 41. Doyle J. DNA Protocols for Plants. In: Hewitt GM, Johnston AWB, 43. R Core Team (2020). R: A Language and Environment for Statistical Young JPW, eds. Molecular Techniques in Taxonomy. Springer Berlin Computing. R Foundation for Statistical Computing, Vienna, Aus- Heidelberg: Berlin, Heidelberg, 1991,283–93. tria. https://www.gbif.org/tool/81287/r-a-language-and-enviro 42. Pereira L, Pujol M, Garcia-Mas J et al. Non-invasive quan- nment-for-statistical-computing. Downloaded from https://academic.oup.com/hr/article/doi/10.1093/hr/uhac131/6601557 by guest on 05 May 2023