Contrasting effects of ethylene and auxin on tomato fruit color
The hormonal treatments induced significant color changes within 96 hours (Figure 2). Treatment with ACC accelerated significantly the transition from green to orange/red compared to controls. On the contrary, treatment with IAA induced a significant delay in the transition from green to orange/red compared to controls. After 96 h, IAA-treated fruits began to turn orange and then never became red (data not shown).
In fruits treated with a combination of ACC and IAA, color evolution was slower than in controls, but faster than the fruits treated by IAA alone, indicating that IAA treatment is epistatic over ACC treatment. In the presence of the auxin antagonist PCIB, fruits turned red faster than control ones and the color change kinetics were very similar to those treated with ACC (Figure 2A). These results confirmed previous studies showing that IAA slows down ripening of tomato fruits [2,6], and that ACC accelerates it [2].
Effects of hormonal treatments on carotenoid, chlorophyll and ABA accumulation
To further investigate the influence of hormonal treatments on fruit pigment composition, fruit extracts were analyzed. At 96 hours, the main carotenoids in control fruits were lutein and β-carotene (Figure 3). Large amounts of chlorophylls a and b were observed, together with trace amounts of lycopene, violaxanthin, neoxanthin, luteoxanthin, ζ-, δ- and α-carotene. The upstream compounds phytoene and phytofluene were not detectable. This composition is typical of a ripening stage between the “Breaker” and “Orange” stages of ripening [41].
The ACC and PCIB treatments induced large changes in carotenoid composition at 96 hours (Figure 3). Lycopene was greatly induced, becoming a major pigment, together with β-carotene which was also induced and lutein which was unaffected. The upstream compounds phytoene, phytofluene and ζ-carotene and the downstream compounds δ- and α-carotene were also induced, while the β-xanthophylls, neoxanthin and violaxanthin were reduced.
The IAA treatment reduced significantly lycopene accumulation compared to controls while it did not affect α-, β- or δ-carotene accumulation. It also led to higher levels of neoxanthin, violaxanthin and chlorophyll a than in the controls (Figure 3).
The 9-cis forms of neoxanthin and violaxanthin are the precursors of abscisic acid (ABA) [23,24], a phytohormone known to control ripening of many fruits, including tomato, in which it triggers ethylene biosynthesis and thus accelerates ripening [25]. ABA levels were decreased by the ACC and PCIB treatments and increased by the IAA treatment (Figure 4), mimicking the evolution of neoxanthin/violaxanthin, thus suggesting that the accumulation of these compounds might be directly correlated. This observation is consistent with the idea that in the tomato fruit, levels of neoxanthin and violaxanthin are rate-limiting for ABA accumulation [26]. Finally, the ACC and PCIB treatments led to an increased degradation of chlorophyll b (Figure 3).
Our results detail the auxin effects on carotenoid accumulation, thus completing preliminary observations that were not detailing this aspect [6]. Our results also detail carotenoid changes induced by ACC, following previous studies showing that ethylene treatments accelerated chlorophyll degradation, the appearance of orange color [10,27] and the accumulation of lycopene [28]. It is noticeable that PCIB, which acts as an auxin antagonist, induced the same effects as ACC.
Effects of hormonal treatments on gene expression
In order to investigate if the above hormone-induced phenotypes were controlled at least partially at the gene expression level, we determined the levels of all transcripts involved in carotenoid biosynthesis by quantitative Real Time PCR (qPCR) at two different times after the hormonal treatments (Figure 5).
As observed in Figure 5A, IAA treatment resulted in lower transcript levels for most of the genes upstream of lycopene (Psy1, Psy3, Pds, Ziso and Crtiso). With the exception of Psy3 which has been reported to be mainly expressed in roots, all these genes are rate-limiting for lycopene accumulation [15]. Thus, these changes in transcript levels match well the slower color change and the decreased accumulation of lycopene after treatment with IAA (Figures 2 and 3). Regarding the downstream part of the pathway (Figure 5B), the transcript levels of β-Lcy1 and Crtr-β1 genes were induced by IAA treatment, concomitant with the higher amounts of violaxanthin and neoxanthin, while Aba4 showed a biphasic response (induction at 24 h and repression at 96 h) and Nced1 a repression at 96 h. Together, these observations indicate that the ABA increase after IAA treatment is a fast response, probably due to an increase in the synthesis of its precursors violaxanthin and neoxanthin, mediated by an activation of the β-Lcy1, Crtr-β1 and Aba4 genes. The repression of Aba4 and Nced1 at 96 h may be due to a negative feedback regulation exerted by the increased ABA levels on these genes. ABA is known to increase in tomatoes prior to the ethylene peak [25].
ACC treatment led to higher levels of Psy1 and Psy2 transcripts, and also, to a lesser extent, of the Ziso, Pds, Zds and Crtiso ones (Figure 5A). All these genes encode rate-limiting steps for lycopene biosynthesis [15] and thus the observed changes in gene expression are in agreement with the faster color change and accelerated lycopene accumulation (Figures 2 and 3). Moreover, ACC treatment decreased β-Lcy1 transcript levels (Figure 5B) with unexpected increase of α-, β- and δ- carotenes, indicating that the β-Lcy1 repression was possibly offset by the unaltered levels of the other cyclase transcripts. ACC also repressed Crtr-β2 expression that was not offset by the unaltered Crtr-β1 levels, reducing the further conversion of carotene compounds into β-xanthophylls. This was confirmed by the reduced neoxanthin and ABA levels after ACC treatment (Figures 3 and 4), in spite of an induction of Aba4. It is also worth noticing that IAA and ACC affected the expression of two different hydroxylase paralogs, Crtr-βi being stimulated by IAA and Crtr-β2 being inhibited by ACC, respectively. Overall, these data explain the faster accumulation of lycopene and β-carotene, and also the lower accumulation of β-xanthophylls and ABA in ACC treated fruits than in controls.
Similar changes in transcript levels occurred in PCIB-treated fruits (Figure 5), which showed an additional repression of β-Lcy2 and an induction of Zep, as well as a very similar carotenoid profile (Figure 3) to the ACC-treated samples. There was no significant effect of any treatment on Ggps expression (Figure 5A and Additional file 1: Figure S1).
The combined IAA + ACC treatment resulted in a visual and carotenoid phenotype intermediate between those of each treatment alone and more similar to that of IAA alone, with the exception of violaxanthin, neoxanthin and ABA induction, which was less pronounced than in IAA alone (Figures 2, 3 and 4). At the transcriptional level, IAA + ACC was less inhibitory of upstream transcripts than IAA alone. Although the significance of these observations awaits clarification, it confirms the antagonistic effects of the two hormones at the biochemical and transcriptional levels.
Chlorophyll degradation in Citrus fruits is an active process mediated by chlorophyllase (Chlase) [29]. In tomato, chlorophyll degradation was affected by hormonal treatments, with IAA treatment retarding chlorophyll a degradation, both alone and in combination with ACC treatment, while chlorophyll b degradation was accelerated by both ACC and PCIB treatments (Figure 3). We measured the levels of the three Chlase transcripts identified in the tomato genome. Repression of all three transcripts was obvious 96 h after the IAA treatment (Figure 6). This correlates well with the higher levels of chlorophyll a and to a lesser extent of chlorophyll b, in both treatments with IAA (Figure 3). However, the marked decrease of chlorophyll b in the ACC and PCIB treatments does not correlate with increased Chlase transcript accumulation (Figure 6). This suggests that, in contrast to Citrus [29], tomato Chlase gene expression is not under ethylene control and that, as observed in Citrus [30], posttranscriptional mechanisms may also regulate Chlase activity in tomato.
Effects of hormonal treatments on the Rin transcript and on transcripts of the carotenoid/ABA pathway
Several genes in the carotenoid pathway are regulated by the Rin transcription factor [16,17]: Psy1, Ziso and Crtiso display direct positive regulation, Zds indirect positive regulation, and ε-Lcy and β-Lcy2 indirect negative regulation. Analyses carried out by qPCR (Figure 7A) showed that the transcript levels of Rin were stimulated by ACC and inhibited by IAA, even if the sole significant difference was noticed for ACC 96 h. The qPCR profiles of Rin (Figure 7A) and Psy1 (Figure 5A) seem to match quite well. Indeed, in keeping with the findings of Fujisawa et al. [17], high positive correlations (ρ > 0.60, and in some cases ρ > 0.80) were observed between transcript levels of Rin and Psy1 at both 24 h and 96 h, Ziso and Crtiso at 96 h, and ZDS at 24 h (Figure 7B).
In contrast, ε-Lcy did not show high correlations with Rin neither at 24 h nor at 96 h, while β-Lcy2 showed strong positive correlations at both time points. This contrasts with the findings of Fujisawa et al. [17] and suggests that lycopene cyclase transcripts are subject to additional layers of regulation. Strong positive correlations with Rin were identified for Pds and Zep at 24 h and for ABA4 at 96 h. The latter two genes mediate the biosynthesis of the ABA precursors, violaxanthin and neoxanthin (Figure 1), and thus their positive correlations with Rin may be indicative of the fact that Rin activates two hormonal cascades: one acting through ethylene [16], and one acting through ABA. Finally, Ggps4 showed a negative correlation with Rin levels at 96 h. This gene is unrelated to fruit carotenoid biosynthesis and may control the biosynthesis of other isoprenoid compounds (Falcone et al., unpublished).
Effects of hormonal treatments on fruit ethylene production
Ethylene is assumed to be a “master switch” controlling tomato fruit ripening. Therefore, it is interesting to verify if the hormonal treatments described above alter ethylene production. We measured ethylene production in hormone-treated fruits at various times after treatments (Figure 8). As expected, ACC treatment accelerated the appearance of the climacteric ethylene peak by about 2 days whereas IAA treatment repressed the ethylene production, and this repression was only partially reversed by combined IAA + ACC treatments. PCIB treatment had little effect up to 100 hours after treatment, while it slightly decreased ethylene production around 200 hours. So it seems that PCIB enhancement of carotenoid accumulation in comparison to controls (Figure 2) is not mediated by a variation in ethylene production. The IAA decrease of carotenoid accumulation in comparison to controls could be partially mediated by the repression of ethylene production.
Factorial and network analyses show associations between hormonal treatments and carotenoid levels
Factorial analyses are used to determine and describe the dependencies within sets of variables. In this study the treatments, and many observed variations, in this study the transcript levels (Figure 9A) or the carotenoid levels (Figure 9B). These factorial correspondence analyses clearly show strong positive correlation between the effects of ACC and PCIB, and their negative correlation to the effects of IAA treatment, whatever the regulatory level measured: transcript accumulation or carotenoid accumulation. It is noticeable that, at the transcript level, the IAA + ACC treatment is strongly correlated with the ACC and PCIB ones (Figure 9A), while at the carotenoid composition level - which matches the fruit phenotype more closely - it is correlated with the IAA treatment (Figure 9B). This may be due to the fact that changes in transcript accumulation occur ahead of those in metabolite accumulation, or to the fact that some of the latter changes are due to post-transcriptional events, or to both.
The transcripts correlating with the ripening delay associated to IAA treatment are lycopene cyclases (ε and β-Lcy) and, to a lesser extent, carotene hydroxylases (Crtr-β) (Figure 9A). These results confirm previous studies [18-21]. All transcripts mediating lycopene biosynthesis in tomato fruits: Psy1, Pds, Ziso, Zds, and Crtiso [15] correlate well with the accelerated ripening induced by ACC or PCIB. Also Psy3, which is much less expressed and is non-essential for lycopene biosynthesis, shows a position opposed to IAA treatment (Figure 9A) as it was strongly repressed by IAA at 96 h (Figure 5A). Same case for the position of Chlase transcripts in Figure 9A which is mainly due to the strong inhibition by IAA, rather than to a stimulation by ACC. Regarding carotenoids, the accumulation of upstream intermediates and lycopene and, to a lesser extent, of α-, β-, and δ-carotene is correlated directly with ACC and PCIB treatments. Inversely IAA and IAA + ACC treatments correlate well with chlorophylls and xanthophylls, (especially violaxanthin and neoxanthin) and their product ABA (Figure 9B). This is consistent with the fact that ripening is associated with the accumulation of cyclic carotenes and with the decrease of chlorophylls and xanthophylls.
We also applied correlation network analysis based on transcript-metabolite data integration (Figure 10). The time spent after treatments increased the strength in the network [31], and at 96 h the network shows four nodes with strong correlation values (|ρ| > 0.60) (Additional file 2: Table S2): ABA, its metabolic precursors violaxanthin and neoxanthin and Nxd, a gene essential for neoxanthin biosynthesis [33]. All four nodes exhibited a prevalence of negative correlations with the other ripening-specific variables in the network.