Jatropha curcas plants cultivated in Xishuangbanna, Yunnan Province, China, were used in this study as described previously (Pan and Xu, 2011). The Arabidopsis thaliana ecotype Columbia (Col-0) and the transgenic lines were grown in plant growth chambers at 22 ± 2°C under long-day (LD, 16 h light/8 h dark) or short-day (SD, 8 h light/16 h dark) conditions. Phenotype analysis was performed on homozygous (T3) Arabidopsis plants and heterozygous (T0) Jatropha plants.

Total RNA was extracted from the leaves of flowering Jatropha plants using the protocol described by Ding et al. (2008). First-strand cDNA was synthesized using M-MLV reverse transcriptase from TAKARA (Dalian, China) according to the manufacturer’s instructions. A full-length JcGA2ox6 CDS was amplified by PCR using the primers XA579 and XA580 (Supplementary Table 1), which introduced BamHI and SalI restriction sites at the ends of the JcGA2ox6 CDS fragment, respectively. The PCR products were subsequently cloned into the pGEM-T vector (Promega Corporation, Madison, WI, United States) and sequenced. All primers used in this research are listed in Supplementary Table 1.

The JcGA2ox6 amino acid sequence was deduced according to the coding sequence (GenBank accession No. KDP28294). Related sequences were identified through a BLAST search¹. To determine the amino acid identities, the alignment results were subjected to pairwise comparisons using DNAMAN 6.0. A phylogenetic tree was constructed based on the protein sequences with MEGA 5.0.² A neighbor-joining phylogenetic tree was generated with MEGA 5.0 using the Poisson model, with gamma-distributed rates and 1,000 bootstrap replicates.

To construct the plant overexpression vector 35S:JcGA2ox6, the JcGA2ox6 sequence was excised from the pGEM-T vector (Promega, Madison, WI, United States) using the restriction enzymes BamHI and SalI. Then, JcGA2ox6 was cloned into the pOCA30 vector containing the CaMV35S promoter and the 35S enhancer (Chen and Chen, 2002). The JcUEP promoter (Tao et al., 2015) was obtained by PCR from the Jatropha genomic DNA using the primers XB348 and XB349 (Supplementary Table 1), which introduced HindIII and SacI restriction sites, respectively. The PCR products were cloned into pGEM-T and sequenced. To construct the JcUEP:JcGA2ox6 plasmid, the CaMV35S promoter of the 35S:JcGA2ox6 vector was replaced with the JcUEP promoter using the restriction enzymes HindIII and SacI.

Transformation of Arabidopsis with the Agrobacterium strain EHA105 carrying the 35S:JcGA2ox6 construct was performed using the floral dip method (Clough and Bent, 1998). Transformation of Jatropha with the Agrobacterium strain EHA105 carrying the 35S:JcGA2ox6 and JcUEP:JcGA2ox6 constructs was performed according to the protocol described by Pan et al. (2010) and Fu et al. (2015).

Total RNA was extracted from frozen Jatropha tissues as described by Ding et al. (2008). Arabidopsis total RNA was extracted from frozen tissues using TRIzol reagent (Transgene, China). First-strand cDNA was synthesized using the PrimeScript^® RT Reagent Kit with gDNA Eraser (TAKARA, Dalian, China). qRT-PCR was performed using SYBR^® Premix Ex Taq^TM II (TAKARA) on a Roche 480 Real-Time PCR Detection System (Roche Diagnostics). qRT-PCR was performed using three independent biological replicates and three technical replicates for each sample. Data were analyzed using the 2^-ΔΔCT method as described by Livak and Schmittgen (2001). The transcript levels of specific genes were normalized using Jatropha JcActin1 or Arabidopsis AtActin2. The primers used for qRT-PCR are listed in Supplementary Table 1.

Three 1-cm² leaf segments were removed from mature leaves of wild-type (WT) and transgenic Jatropha using a hole punch. The total chlorophyll, chlorophyll a and chlorophyll b contents were measured following the protocol described by Arnon (1949). Each measurement was repeated three times.

The WT and T1 transgenic Jatropha (lines L43 and L27) were grown in soil for 6 weeks in a growth chamber at 28°C under a 16 h light/8 h dark photoperiod (Figure 8A). Their young stems (from shoot tip to node 2) (Figure 8B) were collected for GA quantification. The GA contents were determined by the Wuhan Greensword Creation Technology Company, and the analysis was performed as described previously (Chen et al., 2012). Three independent biological replicates and three technical replicates were measured for each sample.

A combined reverse transcriptase-polymerase chain reaction (RT-PCR) strategy was used to isolate JcGA2ox6 cDNA from Jatropha. The JcGA2ox6 coding sequence (CDS) (GenBank accession no. KDP28294) comprises 1002 bp and encodes a 333-amino-acid protein. A multiple alignment was performed using the JcGA2ox6 sequence and the sequences of GA2ox homologs from other species (Supplementary Figure 1A). JcGA2ox6 showed 84, 74, 71, and 57% sequence identity with Ricinus communis GA2ox5 (RcGA2ox5), Populus trichocarpa GA2ox2 (PtGA2ox2), Vitis vinifera GA2ox5 (VvGA2ox5) and Arabidopsis thaliana GA2ox6 (AtGA2ox6), respectively.

To better characterize the JcGA2ox genes within the large GA2ox family, a phylogenetic analysis was performed with amino acid sequences of the GA2oxs in some plant species (Supplementary Figure 1B). The analysis showed that JcGA2ox2, JcGA2ox4, and JcGA2ox6 have the highest identity with RcGA2ox2, RcGA2ox4, and RcGA2ox5 protein from Ricinus communis, respectively, and that they clustered with the C₁₉-GA2ox group of Arabidopsis. This result suggests that all of them encode this type of enzyme. In contrast, JcGA2ox7 and JcGA2ox8 clustered with the C₂₀-GA2oxs group of Arabidopsis.

To investigate the expression pattern of JcGA2ox6 in Jatropha, we performed a qRT-PCR analysis with the total RNAs extracted from various tissues of adult plants, including the roots, stems, young and mature leaves, shoot tips, inflorescences, male and female flowers, fruits at 10 days after pollination (DAP), pericarps at 30 DAP and seeds at 30 DAP. The expression profile showed that JcGA2ox6 was almost constitutively expressed in adult Jatropha (Figure 1). The primary expression levels were detected in the roots, stems, flowers and pericarps at 30 DAP, with the highest level found in male flowers. Very low expression levels were detected in the mature leaves, shoot tips, inflorescences, fruits and seeds at 30 DAP, especially the young leaves.

To determine the roles of JcGA2ox6 in plant growth and development, 35S:JcGA2ox6 was transformed into Arabidopsis for preliminary analysis. WT Arabidopsis under the same growth conditions was used as a control. Transgenic plants were confirmed by qRT-PCR analysis of JcGA2ox6 expression using rosette leaves. More than twenty independent T1 transgenic lines were generated with the 35S:JcGA2ox6 construct. Transgenic plants showed high JcGA2ox6 expression levels (Figure 2E). Under LD conditions, most transgenic lines showed a dwarf phenotype with late flowering.

We selected two independent homozygous lines (L4 and L1) in the T3 generation to examine the phenotypes. Compared with the WT plants, Arabidopsis overexpressing JcGA2ox6 were 12.1–23 cm shorter (Figure 2C and Table 1) and produced smaller rosette and cauline leaves (Figures 2A,D) under LD conditions. Furthermore, the transgenic lines bolted later under both LD and SD conditions, but there was no significant difference in rosette leaf number (Figures 2A,B and Tables 1, 2). The phenotypes of the transgenic Arabidopsis were similar to that of the GA-deficient mutant ga1–3 (Koornneef and Vanderveen, 1980), which suggested that overexpression of JcGA2ox6 reduced the endogenous GA levels. Furthermore, flower development was also affected, as smaller sepals, petals, stamens and pistils were noted (Figures 3A,B). However, the transgenic plants were fertile, producing shorter siliques and smaller seeds (Figures 3C–E).

To further test whether JcGA2ox6 behaves accordingly in Jatropha, the transgenic Jatropha overexpressing JcGA2ox6 was generated. Transgenic shoots overexpressing JcGA2ox6 under the control of the CaMV35S promoter showed abnormal phenotypes of severely crimpled leaves and no obvious stems (Supplementary Figure 2). Moreover, these shoots were unable to generate roots. Therefore, we grafted these shoots onto WT rootstocks, but they hardly grew. These observations prompted us to replace the CaMV35S promoter with a weaker, JcUEP promoter, which is an alternative to the CaMV35S promoter for driving constitutive overexpression of transgenes in Jatropha (Tao et al., 2015). We successfully generated more than 30 transgenic Jatropha lines carrying the JcUEP:JcGA2ox6 transgene.

When all plants grew in pots for 2 months, the JcUEP:JcGA2ox6 transgenic plants exhibited a dwarf phenotype with small dark-green leaves (Figure 4A). The average height of the transgenic Jatropha plants was decreased by 74% compared with that of the WT plants (Figure 4B). Since the leaves of transgenic plants turned dark green, the mature leaves (Figure 4C) were collected to measure the chlorophyll content. The average chlorophyll a, chlorophyll b and total chlorophyll contents in the transgenic plants were 1.9-, 1.3-, and 1.6-fold higher than those of the WT plants, respectively (Figure 4D). This result is consistent with studies in for example transgenic potato (Jackson and Prat, 1996; Martínez-García et al., 2001) and soybean (Suo et al., 2012), which showed that the chlorophyll concentration was inversely correlated with GA₁ or GA₁ and GA₄ contents. In Solanum species, overexpression of a gibberellin 2-oxidase gene from Phaseolus coccineus L. produced darker green leaves with higher concentrations of chlorophylls (Dijkstra et al., 2008). After 5 months of plant growth in the field, the transgenic plants were still dwarfed (Figures 5A–C). In addition, shoot branching was not affected by this transgene. The numbers of branches in WT and transgenic Jatropha were almost the same. We examined the phenotypes of two independent transgenic Jatropha lines L43 and L27, which exhibited intermediate and high expression levels of JcGA2ox6, respectively (Figure 5D). Accordingly, the heights (from the soil to the highest part of the main stem) of L43 and L27 lines were decreased by 37 and 60%, respectively, compared with that of the WT plant (Figure 5E). The retarded vegetative growth was similar to that of transgenic Arabidopsis (Figure 2C and Table 1).

When WT and transgenic plants were grown in the field for approximately 3 months, inflorescence buds emerged in both plants (Figure 5F), suggesting that the flowering time of Jatropha was not affected by overexpression of JcGA2ox6, unlike the late-flowering noted in transgenic Arabidopsis. The inflorescence numbers in the transgenic lines, contrary to our expectation, were lower than those in the WT plants (Figure 6D). Additionally, each inflorescence produced fewer male and female flowers (Figure 6E). Moreover, the inflorescences size was reduced in the transgenic Jatropha lines (Figure 6A) because of the shorter inflorescence stalks (Supplementary Figure 3A) and smaller male and female flowers (Figures 6B,C and Supplementary Figures 3B–E). In addition, similar to transgenic Arabidopsis (Figure 3B), the floral organs remained intact in transgenic Jatropha (Figures 6F,G) and plants were fertile.

Subsequently, fruit and seed development was affected as well by this transgene. Compared with WT plants, transgenic Jatropha lines L43 and L27 exhibited smaller fruits (Figure 7A), as the fruit lengths were reduced by 13.04 and 31.59%, respectively (Figure 7B), while the widths were almost the same (Figure 7C). Consequently, the seeds from the L43 and L27 lines were also small, with 12.90 and 22.04% shorter lengths, respectively, compared with the seed length in WT plants (Figures 7D–G). The results indicated that overexpressing JcGA2ox6 in transgenic Jatropha inhibited the elongation of fruits and seeds other than the widths. It is known that GA induces cell elongation (Kende and Zeevaart, 1997). We supposed a reduced GA content caused by overexpression of JcGA2ox6 would repress cell elongation in fruits and seeds. Consistent with this, in liliaceous Tricyrtis sp., most transgenic plants overexpressing TfGA2ox2 showed no significant differences in cell widths, but the lengths significantly decreased compared with those of control plants (Otani et al., 2013). Furthermore, we analyzed the seed weights and seed oil contents in the L43 and L27 lines. The oil contents were measured using a mini-spec mq-one Seed Analyzer (Bruker Optik, Ettlingen, Germany) (Pan and Xu, 2011). The results showed that the average weights of 10 seeds and the seed oil contents in transgenic lines were significantly decreased (Figures 7H,I).

To determine whether the endogenous GA contents in transgenic Jatropha (L43 and L27) were affected by JcGA2ox6, the non-13-hydroxylated GAs (GA₁₂, GA₉, GA₇, GA₄ and GA₃₄) and the 13-hydroxylated GAs (GA₅₃, GA₂₀, GA₃, GA₁ and GA₈) (Figure 8D) were determined in 6-week-old T1 and WT seedlings (Figure 8A). The results (Figure 8C) showed that the levels of bioactive GA₄ were significantly decreased in transgenic lines and were correlated with the degree of dwarfism (Figure 8A). However, the levels of GA₃₄, the deactivated product of GA₄, did not differ significantly between the WT and transgenic lines. Contrary to GA₄, the bioactive GA₃ levels were increased in transgenic lines. Therefore, we supposed that the reduced GA₄ levels caused negative feedback control of GA biosynthetic gene expression, resulting in an increase in GA₃ levels. Consequently, elevated expression levels of JcGA20ox1, JcGA20ox3, JcGA3ox1, JcGA3ox2 and JcGA3ox3 were detected in transgenic lines (Supplementary Figure 4). This effect of negative feedback regulation was previously confirmed in other plant species (Cowling et al., 1998; Suo et al., 2012). Another two bioactive GAs, GA₁ and GA₇, were not detected in either the WT or transgenic lines. However, GA₈, the deactivated product of GA₁ (Yamaguchi, 2008), accumulated in both lines, and there was a significant increase in its levels in the transgenic lines, implying that elevated expression levels of GA biosynthetic genes caused higher GA₁ levels than WT plants. This result indicated the GA₁ in Jatropha seedlings was completely deactivated by JcGA2ox. Taken together, it suggests that the dwarf phenotype of transgenic Jatropha resulted from a decrease in the endogenous bioactive GA₄. In addition, the transgenic lines displayed decreased levels of GA₂₀, GA₁₂ and GA₉, the immediate precursors of GA₁ and GA₄ (Figure 8D), while GA₅₃ did not change significantly.