• Open Access

Metabolic engineering of β-carotene in orange fruit increases its in vivo antioxidant properties



Orange is a major crop and an important source of health-promoting bioactive compounds. Increasing the levels of specific antioxidants in orange fruit through metabolic engineering could strengthen the fruit's health benefits. In this work, we have afforded enhancing the β-carotene content of orange fruit through blocking by RNA interference the expression of an endogenous β-carotene hydroxylase gene (Csβ-CHX) that is involved in the conversion of β-carotene into xanthophylls. Additionally, we have simultaneously overexpressed a key regulator gene of flowering transition, the FLOWERING LOCUS T from sweet orange (CsFT), in the transgenic juvenile plants, which allowed us to obtain fruit in an extremely short period of time. Silencing the Csβ-CHX gene resulted in oranges with a deep yellow (‘golden’) phenotype and significant increases (up to 36-fold) in β-carotene content in the pulp. The capacity of β-carotene-enriched oranges for protection against oxidative stress in vivo was assessed using Caenorhabditis elegans as experimental animal model. Golden oranges induced a 20% higher antioxidant effect than the isogenic control. This is the first example of the successful metabolic engineering of the β-carotene content (or the content of any other phytonutrient) in oranges and demonstrates the potential of genetic engineering for the nutritional enhancement of fruit tree crops.


Plants are a source of many phytonutrients, including nutrients (such as vitamins) and phytochemicals to which a beneficial physiological function has been directly or indirectly attributed (such as folates, carotenoids, flavonoids, isothiocyanates, glucosinolates, polyphenols and glutathione). Phytonutrients are essential for human nutrition and contribute to the promotion of good health (Beecher, 1999; Block et al., 1992; Lampe, 1999; Nagura et al., 2009). The bioactivity of phytonutrients has been, to a certain extent, associated with their antioxidant properties, such as the capacity to scavenge free radicals, which are involved in the onset and development of many chronic degenerative diseases (e.g. low-density lipoprotein oxidation and atheroma plaque development, DNA oxidation and cancer, oxidation and ageing) (Dröge, 2002). Moreover, evidence suggests that the beneficial effects of these bioactive molecules on human health at a nutritional and/or pharmacological level are higher when the phytonutrients are ingested regularly and in specific amounts as part of the diet rather than as dietary supplements (Asplund, 2002; Cooper, 2004; Guarnieri et al., 2007). Unfortunately, due to either low access to fruits and vegetables or consumer ignorance about the types and quantities of the right foods required for benefit, optimal levels of these substances are not always reached in the diet. The biofortification of crops can foster significant progress at a fundamental level by facilitating the elucidation of the relationship between diet and health and at an applied level by improving diets and reducing the risk of chronic diseases [reviewed by Martin et al. (2011)]. In this context, genetic engineering has emerged as a powerful tool to introduce favourable changes in the metabolic pathways of plants to improve the quantity and bioavailability of phytonutrients, particularly antioxidants (Martin, 2012; Shukla and Mattoo, 2009).

Citrus is the most extensively produced and economically important fruit tree crop in the world, and among the 10.9 million tons (valued at $9.3 billion) of citrus products traded in 2009, sweet orange (Citrus sinensis L. Osbeck) accounted for approximately 60% of citrus production (FAO statistics, http://faostat.fao.org/default.aspx). Oranges contain an array of potent antioxidants, including carotenoids, vitamin C and certain phytochemicals (i.e. flavonoids and phenolics), with potential health-promoting properties (Franke et al., 2005; Guarnieri et al., 2007; Miyagi et al., 2000; So et al., 1996). Carotenoids are the main pigments responsible for the colour of the peel and pulp of citrus fruits and greatly contribute to the fruit's nutritional and antioxidant value. Although citrus fruits are a rich and complex source of carotenoids, the fruit of most orange varieties predominantly accumulates β,β-xanthophylls, which may represent more than 90% of the total carotenoids, with 9-Z-violaxanthin being the main carotenoid in the pulp of mature fruits (Alquézar et al., 2008b; Kato et al., 2004). However, the levels of other nutritionally important carotenoids (such as β-carotene) are considered suboptimal in these varieties. In addition to being the most potent dietary precursor of vitamin A, a large body of epidemiological and laboratory (in vitro, animal and cell culture) studies suggest that β-carotene offers protection against certain age-related degenerative diseases, such as various cancers (predominantly of the aerodigestive tract) (Bertram and Bortkiewicz, 1995; Chew et al., 1999; IARC, 1998; Mathews-Roth, 1982; van Poppel, 1996), type 2 diabetes (Abahusain et al., 1999; Montonen et al., 2005) and coronary heart disease (Gey et al., 1993; Shaish et al., 1995). These health-promoting effects are independent of pro-vitamin A activity and have most frequently been linked to the high antioxidant activity of β-carotene, which is one of the most efficient carotenoid singlet oxygen quenchers (Cantrell et al., 2003).

Recent advances in the identification and isolation of the genes responsible for carotenogenesis in citrus fruits (Alquézar et al., 2008b; Kato et al., 2004) and the development of genetic transformation procedures for this crop type (Peña et al., 2008) enable the production of increased β-carotene levels in orange fruits via metabolic engineering of carotenoid biosynthesis. Specifically, in this work, we sought to block by RNA interference (RNAi) in transgenic sweet orange plants the expression of an endogenous β-carotene hydroxylase gene (Csβ-CHX) involved in the conversion of β-carotene into xanthophylls (Figure S1). However, improving the nutritional quality of citrus fruits can be time-consuming, laborious and expensive because the plants' long juvenile phase delays regular fruit production for years; most citrus types need 5–15 years to begin flowering and fruiting (Peña et al., 2008). Alternatively, early flowering has been achieved in transgenic trees, including citrus plants, by constitutively overexpressing flower meristem identity genes (Bohlenius et al., 2006; Endo et al., 2005; Peña et al., 2001; Weigel and Nilsson, 1995). Then, the fructification of β-CHX-transgenic plants has been accelerated by simultaneously overexpressing the FLOWERING LOCUS T gene from sweet orange (CsFT), a key regulator of the flowering transition.

Analysing the antioxidant ability of a fortified food would be the first step in studying a food's effectiveness in exerting a specific health benefit (i.e. improved overall health or a lower risk of disease) (Shukla and Mattoo, 2009). However, the in vitro antioxidant capacity, which is often used as a claim, can be irrelevant to in vivo antioxidant effects because critical factors such as the bioavailability, metabolism, tissue distribution, dose/response and toxicity of food bioactive compounds affect the true health benefits of engineered crops (Espín et al., 2007). Thus, preclinical animal studies have become an essential first step in testing the in vivo functionality of genetically engineered food. Here, we present a strategy to induce early fruit production and increase the β-carotene level in the pulp of sweet oranges by metabolic engineering. We also confirm the increased capacity of the enriched orange juice to enhance in vivo protection against oxidative stress by approximately 20% in an animal model.


To increase β-carotene levels and simultaneously induce early fruit production in sweet orange plants, we constructed a binary vector, named HRP, containing both an intron-spliced hairpin (ihp) β-CHX RNAi cassette and an FT overexpression cassette (Figure 1a). We used this vector to transform seedlings via Agrobacterium tumefaciens-mediated T-DNA transfer. The binary plasmid pROK2-CsFT, which was previously generated in our laboratory (for details see Data S1), served as the vector system for transforming control plants in this work (Figure 1a).

Figure 1.

Induction of early flowering in transgenic sweet orange (cv. Pineapple) plants containing a sweet orange FLOWERING LOCUS T (CsFT) overexpression cassette. (a) Schematic diagram of the T-DNA region of the HRP and control (CV) vectors used for plant transformation. LB and RB, left and right T-DNA borders, respectively; 35Sp, CaMV 35S promoter; CHXi-AS and CHXi-S, antisense- and sense-oriented sequences, respectively, designed to silence the expression of the Csβ-CHX gene; PDK intron, pyruvate dehydrogenase kinase intron; OCSt, terminator region of octopine synthase gene; CsFT, FLOWERING LOCUS T from sweet orange; NPTII, neomycin phosphotransferase II selectable marker gene conferring kanamycin resistance; NOSt and NOSp, nopaline synthase terminator and promoter sequences, respectively. The transcription orientation for each cassette is indicated by black triangles. (b) Four selected HRP lines and one CV line, all carrying the CsFT transgene, exhibiting an early-flowering and fruiting phenotype compared with the WT control. All of the plants were obtained from seedling plant material, and the photograph was taken 1 year after grafting in the greenhouse. (c) Representative fruits from CV plants at the full-coloured stage after 18 months of cultivation in the greenhouse.

The FT overexpression system induces an extremely early fruiting phenotype and two fruiting cycles per year in sweet orange plants

Kanamycin-resistant regenerants obtained after performing transformation experiments with either the HRP vector or the pROK2-CsFT control vector (CV) were screened by PCR using primers specific to the CsFT transgene. Putatively transformed (PCR-positive) plants, designated as HRP and CV lines, respectively, were grafted onto vigorous non-GM citrus rootstocks in a greenhouse in April 2008 and subjected to phenotypic observation for three consecutive years. Whereas the wild-type (WT) seedlings remained in the nonreproductive vegetative growth phase for the entire study period, the FT lines (either HRP or CV) flowered for the first time in June 2009 or, at the latest, in June 2010. The FT lines not only produced fruits in an extremely short period of time (approximately 1 year after being grafted in the greenhouse), but also had two effective fruiting cycles per year instead of one. None of the FT transformants exhibited morphological features typical of juvenility (i.e. vigorous growth and thorniness). Indeed, they remained stunted, showing smaller leaves than WT seedlings (Figure 1b). Early flowering and fruiting confirmed the effective integration and expression of the FT cassette in all of the HRP and CV transformants. Moreover, the FT transgenic fruits developed normally (Figure 1c).

Transgenic HRP lines harbouring intact copies of the β-CHX RNAi cassette produce fruits with a golden coloration

The presence and integrity of the β-CHX RNAi cassette, as well as the number of transgene DNA loci integrations, were assayed in the HRP lines by Southern blot analysis (Data S1 and Figure S2). The HRP lines that did not flower the first year after grafting (2009) or did not set enough number of mature fruit were excluded from the analysis. DNA restriction with either NotI or ClaI followed by Southern blot hybridization with a 35S promoter-specific probe revealed that plant lines HRP6, HRP11 and HRP12 contained one or two nontruncated copies of the β-CHX RNAi cassette (Figure S2b). This result was confirmed by digestion with either NotI or KpnI, followed by hybridization with a CHXi-specific probe (Figure S2c). Then, based on their low loci number and whole-transgene integrity (Figure S2b,c), plant lines HRP6, HRP11 and HRP12 were selected for propagation and investigated in detail in successive seasons.

During the first stages of development (immature green, mature green and breaker), the fruits from the selected HRP6, HRP11 and HRP12 lines were visually indistinguishable from fruits transformed with the CV. However, at the full-coloured stage, the HRP fruits developed an orange yellow colour, whereas the CV fruits exhibited the bright orange coloration typical of sweet oranges. The difference in the coloration of the fully mature fruits from the HRP lines was easily distinguishable in the flavedo (outer coloured part of the peel), pulp (internal juice vesicles) and juice (Figure 2a). There were no substantial colour differences between the three HRP lines, and this golden phenotype remained stable in all three fruiting seasons and after propagation by grafting onto different rootstocks. Significant differences in the external coloration (< 0.01) of the HRP and CV fruits were confirmed by measuring the flavedo colour index (CI) at maturity (Figure 2b). To ensure complete fruit maturity at the time of measurement, all fruits were sampled when fully coloured, and the internal maturity index (MI) was assayed. Because no significant differences in the MI were detected in the HRP and CV fruits (Figure 2b), the differences in coloration cannot be attributed to an incomplete maturation of the HRP fruits. Rather, changes in the carotenoid content and/or profile are the most likely explanation.

Figure 2.

Golden phenotype of fruits from transgenic sweet orange (cv. Pineapple) plants carrying an ihp β-CHX RNAi cassette (HRP). (a) Phenotype of fruits from the selected HRP6, HRP11 and HRP12 lines, which display a golden colour at the full-coloured stage. Representative HRP (upper row) and CV (lower row) sweet orange plants (left), whole and cross-sectioned fruits (middle) and juice (right). All scale bars, 5 cm. (b) Flavedo colour index (CI) and internal maturity index (MI) of fruits from the three HRP and CV transgenic lines at the full-coloured stage. The data represent mean values ± SEM and are derived from at least three fruits from two independent plants per line analysed in three different fruiting seasons. A statistical analysis of differences between average HRP fruits and average CV fruits was conducted using Student's t-test, and the significance of the differences is indicated (**< 0.01). L, a, b, Hunter colour values; SSC, solid soluble content; TA, titratable acidity.

Silencing of Csβ-CHX results in increased β-carotene in the pulp

The golden phenotype of fruits from the transgenic lines containing the RNAi construct suggests that β-CHX gene expression is suppressed in the HRP fruits. We therefore examined endogenous Csβ-CHX mRNA abundance in fully mature fruits from the transgenic lines HRP6, HRP11 and HRP12 in comparison with the mature fruits of the CV lines by quantitative reverse transcription PCR (qRT-PCR). While the target carotenoid biosynthetic gene was readily expressed in the flavedo and pulp of the CV plants, Csβ-CHX transcript accumulation was highly reduced (up to 39-fold) in both tissues of the HRP lines (Figure 3a).

Figure 3.

Quantification of Csβ-CHX transcript levels and HPLC analysis of pulp carotenoids in transgenic fruits. (a) qRT-PCR analysis of Csβ-CHX expression in the flavedo (black bars) and pulp (grey bars) of full-coloured fruits from the CV and HRP6, HRP11 and HRP12 lines. The Csβ-CHX transcript accumulation, normalized to the CsACT levels, is expressed relative to accumulation in the flavedo of CV fruits and was analysed in at least six independent technical replicates (using two different 96-well plates). The data for the CV represent the average of data for two independent CV lines. (b) Representative HPLC chromatograms of the carotenoid profiles of orange pulp from HRP11 and CV transgenic plants, with the retention time shown on the x-axis and the intensity shown on the y-axis. E-Vio, all E-violaxanthin; 9Z-Vio, 9-Z-violaxanthin; Lut, lutein; Zea, zeaxanthin; Ant, antheraxanthin; Phy, phytoene; α-Cry, α-cryptoxanthin; Phytf, phytofluene; β-Crypt, β-cryptoxanthin; ζ-Car, ζ-carotene; α-Car, α-carotene; β-Car, β-carotene.

To estimate carotenoid accumulation in the edible part of the transgenic fruits, we conducted comparative profiling of the carotenoid content and composition of pulp samples collected at the full-coloured stage by high-performance liquid chromatography (HPLC). The results revealed that consistent with the silencing of the Csβ-CHX gene, the levels of β-carotene were significantly increased (up to 36-fold) in fruits from the three HRP lines compared with the CV fruits (Figure 3b; Table 1). In the best performing HRP line, the pulp accumulated 114.0 ng β-carotene/g fresh weight (FW) on average, while this carotenoid was barely detectable in the CV lines (Table 1). A similar trend was observed for α-carotene, whose content increased (up to 45-fold) in the HRP lines, but reached lower absolute amounts compared with β-carotene (47.2 ng/g of FW in the best performing line). By contrast, a slight reduction in xanthophyll content was detected in the pulp of the HRP lines compared with the CV lines, particularly for β,β-branch xanthophylls (which comprise β-cryptoxanthin, zeaxanthin, antheraxanthin, violaxanthin and neoxanthin) (Figure S1). Despite this decrease, the β,β-xanthophylls remained the major carotenoids in the pulp of the HRP lines (mainly 9-Z-violaxanthin), accounting for approximately 80% of the total carotenoids, whereas in the CV lines, the β,β-xanthophylls represented more than 90% of the total carotenoids. Consistent with the reduction in xanthophyll content (the major carotenoid compounds in pulp), a reduction in total carotenoids was observed in the HRP lines (Table 1). The levels of the colourless linear carotenes at the early steps of the biosynthetic pathway (phytoene, phytofluene and ζ- carotene) (Figure S1) were not significantly different from the levels achieved in the CV pulp (Table 1). In summary, the distribution of carotenoid species indicated that the β-CHX RNAi construct promoted the silencing of the Csβ-CHX gene in the sweet orange pulp, resulting in the accumulation of significant amounts of β-carotene and α-carotene accompanied by a mild general decrease in the downstream products of β-hydroxylation (xanthophylls).

Table 1. Comparative analysis of carotenoid content and composition in pulp samples from transgenic lines
 Total carotenoidsLineal carotenesβ-caroteneα-caroteneβ,β-xanthophyllsε,β-xanthophylls
  1. The values are the means ± SEM of at least three independent measurements and are given in ng/g of FW. For each line, the percentage of each carotenoid or group of carotenoids was calculated on the total carotenoid content. The total content of carotenoids was assessed as the sum of the content of individual pigments. The fold variation with respect to the CV is reported for each carotenoid or group of carotenoids and for each HRP line. The asterisks indicate the significance of the fold variation according to Student's t-test (*< 0.05; **< 0.01; ***< 0.001).

CV12060.4 ± 1412.197.5 ± 23.33.1 ± 1.91.0 ± 0.711015.6 ± 1327.0943.1 ± 113.6
HRP66509.2 ± 944.0135.6 ± 64.598.3 ± 7.547.2 ± 3.85363.3 ± 991.5864.7 ± 76.8
Fold variation−1.91.431.4***45.2***−2.1−1.1
HRP114187.3 ± 493.638.8 ± 15.2114.0 ± 2.127.1 ± 7.53511.8 ± 453.4495.7 ± 73.6
Fold variation−2.9*−2.536.4***26.0***−3.1*−1.9
HRP121885.5 ± 646.033.6 ± 17.860.4 ± 12.711.3 ± 7.61505.9 ± 611.8274.3 ± 47.5
Fold variation−6.4**−2.919.3***10.8*−7.3**−3.4*

Establishment of a C. elegans system to evaluate the in vivo antioxidant effect of orange juice

Caenorhabditis elegans has been widely used as a model to study the in vivo antioxidant capacity of different pure compounds and certain plant extracts (Artal-Sanz et al., 2006; Martorell et al., 2011, 2012; van Raamsdonk and Hekimi, 2010), but not the juice from fruits or vegetables. Therefore, it was necessary to optimize certain essential aspects concerning the experimental method prior to performing the bioassays with the orange fruit. First, based on preliminary dose–response experiments performed with commercial pasteurized orange juice (data not shown), a range of 1–2% (by vol.) was chosen as the optimal range for supplementation of the nematode growth medium (NGM) to test antioxidant effect. At higher doses, antioxidant effect was also observed, but reaching levels close to saturation. Subsequently, when including the food matrix under study (not sterilized pulp powder samples obtained from orange fruits) at a concentration of 2% in the NGM, microbial contamination was detected in the supplemented medium during the course of experiments. Therefore, it was necessary to establish a sample sterilization system prior to the supplementation of the NMG with pulp (for details, see Data S1 and Figure S3).

Another important task was to confirm the intake of the orange pulp extracts by the nematodes during the optimized culture protocol. Therefore, intake confirmation experiments were performed by feeding worm populations with WT pulp extracts and using the NGM without supplementation as a control (see 'Experimental procedures'). As the only purpose of this experiment was confirming the intake of pulp (and not the biological response of nematodes), we decided to use a high dose of supplementation (20%), which greatly facilitated the detection of carotenoids in the worm extracts. The two conditions yielded worm pellets of differing colour; the nematodes fed with the pulp extract were slightly orange (Figure 4a). Afterwards, HPLC analysis of the lysed worm pellets indicated the presence of violaxanthin in the nematodes fed with WT pulp extracts, whereas no carotenoid was detected in the control sample (worms fed with NGM) (Figure 4b,c), thus confirming the intake and bioassimilation of pulp extract by C. elegans.

Figure 4.

Bioassimilation of pulp extracts by C. elegans. (a) Comparison of worm pellets (not disrupted) obtained after feeding with wild-type pulp extracts (20%) (NGM + PULP) or a standard diet (NGM), showing significant increases in the levels of coloured carotenoids in worms fed with the citrus pulp. (b) HPLC analysis of carotenoids in disrupted worms previously fed with NGM + PULP or NGM, with the retention time shown on the x-axis and the intensity shown on the y-axis. (c) Total violaxanthin levels as measured by HPLC analysis of worms fed with NGM + PULP or NGM and normalized to the total protein content in the worm samples. E-vio, E-violaxanthin; 9-Z-vio, 9-Z-violaxanthin; ND, not detected.

β-carotene-enriched (HRP) orange juice exerts a much higher antioxidant effect than control (CV) juice in C. elegans

To investigate whether the levels of β-carotene achieved were sufficient to offer antioxidant properties in a dietary context, we tested diets supplemented with orange pulp in C. elegans. The two samples compared in the bioassay (pulp extracts from the HRP or CV fruits) were processed according to the method described in Data S1. After processing and measuring the content of carotenoids and vitamin C (Data S1; Table S1), the samples were added to the NGM agar plates at a final concentration of either 1% or 2% (by vol.). In our trials, we also included two additional feeding conditions that served as internal experimental controls: NGM without supplementation (negative control) and NGM supplemented with vitamin C (a well-known antioxidant compound) at 0.1 μg/mL. As shown in Figure 5, both doses of supplementation (1% and 2%) demonstrated the positive effect of the citrus pulp extracts (and, in particular, the HRP pulp extract) on resistance against oxidative stress in C. elegans. However, 2% was chosen as the optimal dose for supplementation because better protection was observed (Figure 5b). The animals fed with CV pulp extract had a survival rate of 52%, which was significantly (< 0.01) higher than the rate obtained in the negative control condition (34%) and similar to the rate obtained for the worms fed with vitamin C (46.13%). More interestingly, the animals fed with the HRP pulp extract had a survival rate of 71.67%, which represented a significant (< 0.01) increase of approximately 20% compared with the survival rate observed in the CV pulp extract-fed worms (Figure 5b). These results demonstrate that the worms fed with the β-carotene-enriched orange pulp were more resistant to the oxidative stressor hydrogen peroxide than worms fed with the control orange pulp.

Figure 5.

Antioxidant activity of orange pulp extracts and pure β-carotene in C. elegans. (a and b) Survival of C. elegans treated with 2 mm H2O2 on NGM agar plates, with β-carotene-enriched (HRP) or control (CV) pulp extract supplementation at 1% (a) or 2% (b). The HRP pulp extract tested in the bioassays was obtained from mixtures of pulp from the HRP6, HRP11 and HRP12 lines. The standard diet (NGM) and a diet supplemented with vitamin C (0.1 μg/mL) were used as negative and positive control treatments, respectively. (c) Survival of C. elegans treated with 2 mm H2O2 on NGM agar plates, with or without β-carotene supplementation. Doses of supplementation with pure β-carotene used were 3 μg/mL (βCAR 3) or 30 μg/mL (βCAR 30). The trials were performed in triplicate (100 worms scored per condition). In each experiment, the mean values ± SEM are presented, and treatments labelled with different letters are significantly different at < 0.01 using Fisher's protected LSD test.

Finally, to study whether the observed antioxidant effect of HRP pulp was related to the β-carotene, we performed oxidative stress response assays in C. elegans with exogenous pure β-carotene at a dose equivalent to the amount of β-carotene present in the HRP pulp extract (3 μg/mL). We also included in the bioassays a 10-fold higher dose of β-carotene supplementation (30 μg/mL) and the standard nematode diet as control. Results showed a significantly higher (< 0.01) antioxidant effect of β-carotene at a dose of 3 μg/mL compared with the NGM control condition; at a 10-fold higher dose, greater effect was observed (Figure 5c). These results confirmed the antioxidant capacity of β-carotene. However, the fact that the survival rate obtained by feeding worms with the HRP pulp extract (71.67% worm survival, Figure 5b) was higher than that achieved with pure β-carotene at 3 μg/mL (52% worm survival, Figure 5c) suggests that this particular background food matrix (orange pulp) enhances the antioxidant effect of increased β-carotene.


The possibility that dietary intervention via nutrition-enriched food may significantly decrease the incidence of certain chronic degenerative diseases, in conjunction with increased public awareness of the nutritional benefits of antioxidants for human health, has catalysed scientific efforts to increase many bioactive constituents in fruits and vegetables (Davies, 2007; Hossain and Onyango, 2004; Shukla and Mattoo, 2009). Although conventional breeding is one means of achieving this goal (Mayer et al., 2008; Nestel et al., 2006), the genetic diversity available within the sexually compatible species of any given crop limits the extent of improvement. In this regard, genetic engineering has become a refined tool to increase the antioxidant and nutrient capacity of economically important crops, not only to achieve levels favourable for highly nutritional diets, but also to enable in-depth studies on the relationships between diet, genetics and metabolism (Christou and Twyman, 2004). Moreover, a fast-track system, as the one used here based on the ectopic overexpression of CsFT in juvenile plants, greatly facilitates addressing metabolic engineering strategies aimed at improving fruit quality in plant species requiring many years to begin to flower and set fruits. To our knowledge, this is the first report in which this fast-track system has been successfully used for that purpose, opening the possibility for rapid characterization of fruit quality traits achieved by transgenic approaches in citrus and other fruit tree crops.

The important contribution of carotenoids to the nutritional value and healthy properties of certain fruits and vegetables has led to attempts to induce or increase carotenoid levels in foods, particularly β-carotene, through metabolic engineering of carotenoid biosynthesis (e.g. tomato, maize, rice, potato and canola seeds; reviewed in Botella-Pavía and Rodríguez-Concepción, 2006; DellaPenna and Pogson, 2006). Sweet orange fruit is an excellent candidate for the transgenic enhancement of β-carotene content. Increased levels of β-carotene (a lipophilic antioxidant) would complement vitamin C (a hydrophilic antioxidant highly abundant in oranges) because it is generally thought that foods rich in both soluble and membrane-associated antioxidants offer the best protection against disease (Yeum et al., 2004). Additionally, other factors, such as the low complexity of the food matrix, would potentially enhance the absorption and bioavailability of the increased β-carotene in oranges (de Pee et al., 1998). In this work, we have shown that RNAi-mediated silencing of Csβ-CHX, which regulates an important step in orange carotenogenesis, induces the accumulation of high levels of β-carotene in oranges (up to 36.4-fold with respect to control fruits). This increase was accompanied by a general, mild decrease in the accumulation of downstream xanthophylls. This result is consistent with the findings of other studies, in which increases in one carotenoid were found to occur at the expense of others (Fraser et al., 2002), suggesting feedback inhibition or rate-limiting steps within the carotenoid biosynthetic pathway and/or possible saturation of the carotenoid storage capacities within citrus fruits (Lu et al., 2006). The unexpectedly slight decrease in xanthophyll concentration in β-carotene-enriched oranges may be explained by the presence of a second putative β-CHX in the sweet orange genome (http://citrus.hzau.edu.cn/orange/, http://www.phytozome.net/) that would not be silenced by the RNAi strategy used here because its transcripts do not show enough sequence homology with Csβ-CHX RNA targets. Therefore, activity of the second putative β-CHX could likely counterbalance, at least in part, the very low Csβ-CHX transcript levels found in golden orange fruits.

Various vegetable and fruit crops have been transformed with the objective of enhancing the concentration of health-promoting phytonutrients, with special attention to antioxidants (Davies, 2007; Newell-McGloughlin, 2008; Shukla and Mattoo, 2009). Although most studies have successfully increased the amount of the target metabolite(s), only a few studies have also evaluated the antioxidant capacity of the enriched foods (Butelli et al., 2008; Rein et al., 2006). In this study, we have developed a straightforward experimental system that permits the characterization of the biological activity of transgenic citrus fruit in vivo in an inexpensive manner using C. elegans as a model organism for the functional analysis of orange juice.

We have engineered an increased level of β-carotene in sweet oranges, and C. elegans studies indicate that this level is sufficient to impart a substantial protective effect against oxidative damage when orange juice is included as part of the regular diet. The biofortified oranges exerted a higher protective effect than control oranges despite its slightly lower content of xanthophylls [oxygenated carotenoids also described as dietary antioxidants (Haegele et al., 2000)], which supports the strong antioxidant effect of dietary β-carotene reported in previous studies (Jialal et al., 1991; Meydani et al., 1994; Nakagawa et al., 1996). The effect of dietary β-carotene against oxidative stress achieved in this work (70% worm survival after hydrogen peroxide treatment) is similar to the effect reported for cocoa polyphenols (Martorell et al., 2011) and Tonalin (a conjugated linoleic acid commercial mixture) (Martorell et al., 2012). The mechanism of the antioxidant activity of β-carotene is related to this compound's hydrophobic character and ability to quench singlet oxygen and deactivate free radicals (Burton and Ingold, 1984; Rice-Evans et al., 1997). There is evidence indicating that the efficacy of β-carotene (as well as the efficacy of other phytonutrients) is heavily influenced by nutritional context (Hadad and Levy, 2012; Palozza and Krinsky, 1992; Shaish et al., 1995). Consistent with this finding, we observed that the level of resistance to oxidative stress achieved with exogenous pure β-carotene was lower than the levels of protection reached with the biofortified oranges, although both feeding conditions supplied diets with equivalent concentrations of β-carotene. This result indicates that the orange juice nutritional context has a substantial influence on the impact of dietary β-carotene, either through synergistic interactions with other constituents of the food matrix or through effects on bioavailability.

Experimental procedures

Generation of citrus transformants

A binary vector (HRP) was constructed that contained both an ihp β-CHX RNAi cassette and an FT overexpression cassette. The details of the construction are provided in the Data S1. After the HRP construct was confirmed by restriction mapping and DNA sequence analysis, the plasmid vector was transferred to A. tumefaciens strain EHA105 by electroporation and used to transform sweet orange plants (cv. Pineapple). The binary plasmid pROK2-CsFT, which contains a CsFT overexpression cassette, was used as the vector system for transforming control plants (Data S1). In both cases, the transformation of epicotyl explants from citrus seedlings was performed as previously described (Peña et al., 2001). The regenerated shoots obtained after kanamycin selection were screened by PCR using primers specific for the CsFT chimeric cassette. To avoid nonspecific amplification of the endogenous FT gene(s), the primers 35Sfinal-F (5′-CACAATCCCACTATCCTTCG-3′) and FTCs2 (5′-GGGATTGATCATCGTCTGAC-3′), which amplified the region encompassing the end of the 35S promoter and the entire CsFT transgene, were used to screen the transformants. The PCR programme used was 95 °C for 5 min, 30 cycles of 95 °C for 30 s, 58 °C for 30 s and 72 °C for 1 min, followed by 72 °C for 10 min. All PCR-positive shoots were shoot-tip-grafted onto Troyer citrange (C. sinensis L. Osb. × Poncirus trifoliata L. Raf.) seedlings growing in vitro (Peña et al., 2008). Three to 5 weeks after shoot-tip grafting, the plantlets were grafted again in a greenhouse onto 5-month-old Carrizo citrange seedlings. The putatively transformed Pineapple sweet orange plants were maintained in greenhouses for 3–4 years and grafted onto different citrus rootstocks for further analyses.

Plant material, colour index and internal maturity index

Fully ripened fruits of the transgenic HRP and CV lines were harvested during three consecutive seasons. The colour index (CI) of each fruit was measured with a Minolta colorimeter (model CR-200; Minolta Co. Ltd., Osaka, Japan) by taking three measurements in the equatorial zone of each fruit. The mean values of the lightness (L), red-green (a) and yellow-blue (b) Hunter parameters were calculated for each fruit and presented as previously described (CI = 1000a/Lb) (Jiménez-Cuesta et al., 1981).

After colour measurement, part of the flavedo and pulp tissues was separated with a scalpel, frozen in liquid nitrogen, ground to a fine powder and stored at −80 °C until analysis. Juice was extracted from the remaining pulp of each fruit and immediately analysed. The titratable acidity (TA) of the juice was determined by titration with 0.1 N NaOH solution using phenolphthalein as an indicator and expressed as grams of citric acid per 100 mL of juice. The soluble solids content (SSC) was determined by measuring the refractive index of the juice (Atago Digital Refractometer PR-101 model 0–45%; Atago Co., Ltd., Tokyo, Japan), and the data were expressed as ºBrix. The maturity index (MI) was estimated for each fruit from the SSC/TA ratio.


RNA extractions were performed from flavedo and pulp samples using the RNeasy Mini Kit (Qiagen, Hilden, Germany). The total RNA preparations were treated with recombinant DNase I (RNase-free DNase set; Qiagen) for complete genomic DNA removal, and the resultant RNA was accurately quantified in triplicate using a NanoDrop®ND-1000 (NanoDrop products, Wilmington, DE) spectrophotometer. Gene expression analysis was performed by a two-step real-time qRT-PCR method. First-strand cDNA was synthesized from 2 μg of each DNase-treated RNA in 20 μL using oligo(dT)18 and a SuperScript™ II Reverse Transcriptase kit (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. After synthesis, the cDNA was subjected to a 20-fold dilution with RNase-free water (Sigma-Aldrich, St. Louis, MO). Subsequent qPCRs were performed with a LightCycler®480 Instrument (Roche Applied Sciences, Indianapolis, IN), and fluorescence was analysed using LightCycler®480 software. The primer pair and reaction conditions used for Csβ-CHX target gene amplification were obtained from Alquézar et al. (2009). Normalization was performed using the expression levels of the ACTIN gene from C. sinensis (CsACT) (Romero et al., 2012). Fluorescence intensity data were acquired during the 72 °C extension step, and the specificity of the reactions was verified by analysing the postamplification dissociation curves. Melting curve analysis confirmed the presence of a single PCR product from all samples with no primer dimers. The relative expression of the target gene (Csβ-CHX) normalized to the expression of the housekeeping gene (CsACT) was calculated following the mathematical model described by Pfaffl (2001). cDNA from the flavedo of the CV lines at the full-coloured stage was used as a calibrator sample, and the rest of the values were expressed relative to this sample's value. The PCR efficiency values for Csβ-CHX and CsACT were approximately equal and were calculated by generating respective standard curves using cDNA serial dilutions. The values reported are the mean ± SEM of at least two independent assays. Each assay included (in triplicate) a standard curve, a no-template control and 1 μL of each test and calibrator cDNA.

Carotenoid extraction and analysis

The extraction of carotenoids from the pulp of the sweet oranges followed a previously described protocol (Alquézar et al., 2008b). The extracts were dried by rotary evaporation and stored under a nitrogen atmosphere at −20 °C until HPLC analysis. Carotenoid extracts were prepared for HPLC analysis by dissolution in 30 μL of chloroform/MeOH/acetone (5:3:2 by vol.), and a 25-μL aliquot was immediately injected. The HPLC analysis method was described previously (Alquézar et al., 2008b). Carotenoids were identified by their retention time, absorption and fine spectra (Britton, 1998). The carotenoid peaks were integrated at their individual maximum wavelengths, and the peaks' content was calculated using calibration curves of β-carotene (Sigma) for α- and β-carotene; β-cryptoxanthin (Extrasynthese, Lyon, France) for α- and β-cryptoxanthin; zeaxanthin (Sigma) for zeaxanthin; and antheraxanthin and lutein (Sigma) for lutein, violaxanthin and neoxanthin isomers. Phytoene and phytofluene standards for quantification were obtained from flavedo extracts of Pinalate sweet oranges, which accumulate large amounts of these compounds (Rodrigo et al., 2003), and were then purified by TLC (Pascual et al., 1993). Quantification was performed using Empower chromatography software (Waters Corp., Milford, MA). Carotenoids were measured for a minimum of three different fruits from two different plants per line and three consecutive fruiting seasons.

Worm feeding studies

Strains and maintenance conditions

The C. elegans strain used in this study (WT Bristol N2) and Escherichia coli OP50 were obtained from the Caenorhabditis Genetics Center at the University of Minnesota. The worms were maintained at 20 °C on NGM (3 g/L NaCl, 2.5 g/L peptone, 5 g/L cholesterol, 1 m MgSO4 and 1 m KPO4, pH 6.0) agar plates on a lawn of E. coli OP50.

Intake confirmation experiments

A sample processing system was established suitable for including citrus pulp in the nematode growth medium (NGM) (for details see Data S1). The intake confirmation experiments were performed with synchronized populations of the C. elegans WT strain Bristol N2. The nematodes were cultured on NGM or NGM supplemented with WT pulp extract (20%), and eggs were recovered in 50 plates per condition. Embryos were incubated at 20 °C until reaching the young adult stage (3 days old). The worms were then recovered with M9 buffer and washed three times to eliminate the E. coli OP50 present in the media. An additional 2-h incubation in M9 buffer was performed to facilitate the removal of gut microbiota from the nematodes. Once the supernatant was discarded, the worm pellets (containing approximately 12 500 worms per condition) were recovered in Eppendorf tubes and disrupted by sonication (three pulses, 10 W, 20 s/pulse). The evacuated and washed worms were rotated gently for 3 min at 4 °C to form a loose pellet, and the supernatant was carefully removed with a pipette. The worm pellets were ground to a powder with a micropestle (Eppendorf, Hamburg, Germany) and liquid N2. Acetone was then added to the powder (0.5 mL), followed by vortex-stirring for 1 min and centrifugation for 2 min at 15871 g (4 °C). Upon centrifugation, the acetone extracts were recovered, and the pellet was re-extracted with acetone. The colourless pellets were stored at −20 °C until subsequent measurements of protein content. Pooled acetone extracts were dried under a nitrogen atmosphere at 30 °C until reducing their volume to approximately 0.5 mL, after which the extracts were sequentially washed with two nonpolar organic solvents (ether and chloroform, 0.5 mL each) to remove any traces of water and impurities. The extracts were dried with nitrogen and stored under a nitrogen atmosphere at −20 °C until performing HPLC analysis of the carotenoids, following the protocol described above. Finally, the protein content of the stored worm pellets was measured essentially as previously described (Lamitina et al., 2004), and the content value was used to normalize the carotenoid levels. Briefly, the pellets resulting from the carotenoid extraction were treated with 0.5 mL 10% perchloric acid (PCA) to precipitate proteins. After centrifugation (9391 g, 30 min, 4 °C), the acidic supernatant was removed, and the PCA-precipitated pellets were solubilized with 0.1 N NaOH (200 μL). The protein concentration of these solutions was quantified according to Bradford (1976) using Protein Assay Dye Reagent (Bio-Rad, Hercules, CA) and bovine serum albumin as a standard.

Hydrogen peroxide-induced oxidative stress assays

To measure C. elegans survival rates after exposure to oxidative stress, we employed synchronized eggs hatched in NGM on agar plates containing the E. coli OP50 strain and in the presence of either the β-carotene-enriched (HRP) or the control (CV) pulp extracts at a final concentration of either 1% or 2% (by vol.). The HRP pulp extract was obtained from mixtures of pulp from the HRP6, HRP11 and HRP12 lines. The trials were performed with fruits from two different fruiting seasons. Ascorbic acid (0.1 μg/ mL, Sigma-Aldrich) was used as antioxidant positive control. After 7 days of growth at 20 °C, the worms were transferred to NGM plates containing 2 mm H2O2 and incubated for 5 h. The animals were then washed, and their viability was measured. Worms were considered dead when they no longer responded to prodding. The experiments were performed in triplicate. The proportion surviving after treatment with H2O2 was used to estimate the antioxidant capacity resulting from each feeding condition. The data on the survival rates were subjected to anova using Statgraphics, version 5.1 software (Manugistics, Inc., Rockville, MA), and Fisher's protected LSD test (< 0.01) was used to separate the means.


We are grateful to Alejandra Salvador, Cristina Rojas and Sawsen Sdiri (IVIA) for HPLC facilities. This work was supported by grants AGL2009-08052 and AGL2012-33429 and cofinanced by the Fondo Europeo de Desarrollo Regional (FEDER), the Ministry of Science and Innovation of Spain (MICINN), Prometeo/2008/121 from the Generalitat Valenciana and the Citrus Research and Development Foundation from Florida (CsFT project). Elsa Pons was the recipient of the MICINN Ph.D. Fellowship BES-2004-5270.