Identification, Functional Study, and Promoter Analysis of HbMFT1, a Homolog of MFT from Rubber Tree (Hevea brasiliensis)

A homolog of MOTHER OF FT AND TFL1 (MFT) was isolated from Hevea brasiliensis and its biological function was investigated. Protein multiple sequence alignment and phylogenetic analysis revealed that HbMFT1 conserved critical amino acid residues to distinguish MFT, FLOWERING LOCUS T (FT) and TERMINAL FLOWER1 (TFL1)-like proteins and showed a closer genetic relationship to the MFT-like group. The accumulation of HbMFT1 was generally detected in various tissues except pericarps, with the highest expression in embryos and relatively higher expression in roots and stems of seedlings, flowering inflorescences, and male and female flowers. HbMFT1 putative promoter analysis showed that tissue-specific, environmental change responsive and hormone-signaling responsive elements were generally present. HbMFT1 was strongly induced under a short-day condition at 28 °C, with the highest expression after the onset of a day. Overexpression of HbMFT1 inhibited seed germination, seedling growth, and flowering in transgenic Arabidopsis. The qRT-PCR further confirmed that APETALA1 (AP1) and FRUITFULL (FUL) were drastically down-regulated in 35S::HbMFT1 plants. A histochemical β-glucuronidase (GUS) assay showed that HbMFT1::GUS activity was mainly detected in stamens and mature seeds coinciding with its original expression and notably induced in rosette leaves and seedlings of transgenic Arabidopsis by exogenous abscisic acid (ABA) due to the presence of ABA cis-elements in HbMFT1 promoter. These results suggested that HbMFT1 was mainly involved in maintenance of seed maturation and stamen development, but negatively controlled germination, growth and development of seedlings and flowering. In addition, the HbMFT1 promoter can be utilized in controlling transgene expression in stamens and seeds of rubber tree or other plant species.


Introduction
Rubber tree, a member of Euphorbiaceae family [1], is a monoecious species with male and female flowers on the same inflorescence and is known as an economically important crop in that it can produce natural rubber which is widely used in various aspects, such as in the rubber industry, medical health and items used in daily life because of its particular features of strong flexibility, good insulation, and plasticity as well asits waterproof quality. For higher yield many of the detected in the leaf transcriptome. However, the HbMFT2 cDNA fragment was also not successfully cloned from sampled tissues other than leaves even though several pairs of specific primers were designed, suggesting that the function of HbMFT2 may be degenerated during evolution. A 2267 bp 5 1 flanking region upstream of HbMFT1 coding region was cloned using a pair of specific primers and a 151 bp 3 1 untranslated region (UTR) was obtained by 3 1 rapid amplification of cDNA ends (RACE). Through sequence analysis, HbMFT1, which encoded 175 aa, conserved the characteristic genomic organization of the PEBP gene family, including four exons and three introns ( Figure 1A). Protein multiple sequence alignment analysis revealed that HbMFT1 possessed the characteristics of MFT. At position His88/Tyr85 in the TFL1/FT-like protein, HbMFT1 contained other Trp88-like MFT homologs, suggesting that Trp88 is highly conserved for function in MFT homologs among different species because His88/Tyr85 conferred TFL1/FT function in other plants [27,28]. HbMFT1 protein was more similar to JcMFT2, VvMFT, JcMFT1, AtMFT (showed 88.29%, 72.57%, 58.29% and 58.29% sequence identity, respectively) than to FT and TFL1 (showed 44.38% and 46.93% identity, respectively). Like other members of the PEBP family, MFT-like proteins contained intact D-P-D-x-P and G-x-H-R motifs ( Figure 1B), both of which were in favour of the combination among the ligand-binding sites [7]. Phylogenetic analysis showed that HbMFT1 was most closely related to JcMFT2, GaMFT1, GbMFT1 and VvMFT, and was clustered into MFT-like groups ( Figure 2). Interestingly, the TFL1-like proteins were divided into two distinct subgroups, in which BFT, MdCEN, ZCN3 and ZmTFL1 showed more similarity to FT-like proteins. However, all of the TFL1-like proteins in the phylogenetic tree contained two conservative amino acid residues such as TFL1 at His88 and Asp144 with an exception for BFT, which contained Tyr 85 and Glu141. It has been speculated that the presence of a charged hydrogen bond between Asp144 and His88 in TFL1 determined TFL1 activity [19]. In Arabidopsis, BFT has been confirmed to function redundantly with TFL1 when overexpressed in Arabidopsis because of the presence of a hydrogen bond between Tyr 85 and Glu141 like TFL1 [19]. Therefore, MdCEN, ZCN3 and ZmTFL1 also function similarly to TFL1 rather than FT, and fall into a TFL1-like subgroup together with BFT.

HbMFT1 Expression Analysis in Rubber Tree
The tissue-specific expression analysis was conducted by qRT-PCR and the result showed that HbMFT1 could be detected in various tissues except pericarp, with the highest expression in embryo. HbMFT1 also showed stronger transcript accumulation in roots and stems of three-month old seedlings, and flowering inflorescences (namely I5), male and female flowers of mature rubber trees ( Figure 3A). It was worthwhile to note that HbMFT1 was gradually increased along with development of leaves and inflorescences ( Figure 3A). We also found that its expressions in roots, stems and leaves were progressively decreased from the three-month-old seedlings to the ten-year-old rubber trees ( Figure 3B-D). However, the expression in shoot apices was increased from the three-month-old seedlings to the two-year-old trees, but was decreased sharply in the ten-year-old rubber trees, significantly lower than that of the three-month-old seedlings ( Figure 3E).

Cloning and cis-Elements Analysis of the HbMFT1 Promoter
Based on genome database (data unpublished), a 2267 bp putative promoter fragment upstream of HbMFT1 was obtained from cultivated rubber tree 7-33-97 and putative cis-acting elements were analyzed with the PLACE database (http://www.dna.affrc.go.jp/htdocs/PLACE/) [29]. Various putative plant regulatory elements in the HbMFT1 promoter are shown in Table S1. There were some tissue-specific elements such as GTGANTG10 and POLLEN1LELAT52 for pollen expression, TAAAGSTKST1 for guard cell expression, EBOXBNNAPA, MYCATERD1, MYCATRD22 and 2SSEEDPROTBANAPA for storage-protein expression, CANBNNAPA and PROXBBNNAPA for embryo-and endosperm-specific expression, RHERPATEXPA7 for root hair-specific expression, RAV1AAT and RAV1BAT for rosette leaf and root expression, NODCON2GM and NODCON1GM for nodule expression, DOFCOREZM and AACACOREOSGLUB1 for endosperm expression, GATABOX, SEF4MOTIFGM7S, DPBFCOREDCDC3 and SEF3MOTIFGM for seed or embryo expression, SITEIIATCYTC for anther-and meristem-specific expression, and TGTCACACMCUCUMISIN for fruit expression. In addition, there were other putative regulatory elements in response to environmental cues and hormone signals, such as temperature responsive elements, water stress responsive elements, light-responsive elements, disease responsive elements, circadian-regulated elements and the elements in response to hormones including salicylic acid, gibberellin acid (GA), jasmonic acid (JA), abscisic acid (ABA) and auxin. Therefore, it is tempting to speculate that the activity of the HbMFT1 promoter is limited in specific tissues because of the presence of tissue-specific cis-acting elements and may be regulated by various environmental stresses and hormone signals. (B) Protein multiple alignment between deduced amino acid sequence of HbMFT1 in rubber tree and phosphatidyl ethanolamine-binding protein (PEBP) family of other species. Sequence alignment was carried out using DNAMAN 6.0 software (http://www.lynnon.com/). Three triangles refer to the intron positions. I, D-P-D-x-P motif. II, G-x-H-R motif. III, the region is essential for FT/TFL1-like activity in exon IV. An asterisk indicates amino acids that are related to antagonistic functions between TFL1 and FT protein. Different colors refer to the different homology levels of aligned amino acid residues among MFT homologs. Darkblue represents 100% identity. Hotpink represents more than 75% identity. Turquoise represents more than 50% identity. The aforementioned proteins and their accession numbers: Arabidopsis   of HbMFT1 in roots, stems, mature leaves and shoot apices of rubber trees at different ages (three months, two years and ten years). RS, SS, MS and AS, roots, stems, mature leaves and shoot apices of the three-month-old seedlings; RT, ST, MT and AT, roots, stems, mature leaves and shoot apices of the two-year-old trees; RR, SR, MR and AR, roots, stems, mature leaves and shoot apices of the ten-year-old trees (adult trees). HbRH2b, HbRH8 and HbYLS8 were used as reference genes for qRT-PCR analysis. Values were means˘SE from three independent biological replicates.

HbMFT1 Promoter Activity Analysis under Different Photoperiods and Temperatures
Given that many cis-acting elements responsive to light are present in the HbMFT1 promoter, we carried out the expression analysis under long-day and short-day conditions, respectively, in mature leaves, in which the expression was higher than that of the other development-stage leaves ( Figure 3A). The qRT-PCR results revealed that the HbMFT1 transcript was mainly induced under the short-day conditions, came to a peak after the onset of day, and then progressively decreased till the fourth hour. Small oscillations were subsequently generated and lasted for the remaining four hours of the light period and the previous ten hours of darkness ( Figure 4A). However, extremely weak expression of HbMFT1 was observed under the whole long-day condition. Similarly to the expression in short-day conditions, HbMFT1 also showed slight oscillations after entering darkness, lasting six hours ( Figure 4A). In addition, the abundance of the HbMFT1 transcript was modulated by temperature. As shown in Figure 4B, as the temperature decreased from 36 to 28˝C, HbMFT1 transcript accumulation was gradually increased, and then progressively reduced as the temperature continued to decrease to 20˝C, at which minimal expression level was detected. However, small fluctuations occurred at 16 and 8˝C. This result indicated that 28˝C is the best inducing temperature for HbMFT1 mRNA accumulation. Results were from three independent biological replicates.

Characterization of HbMFT1::GUS (β-Glucuronidase) Fusion in Transgenic Arabidopsis
A binary vector containing HbMFT1::GUS fusion fragment was constructed and introduced into Arabidopsis ( Figure 5A). Two independent lines were obtained in the T1 generation, of which the roots, stems with caulines and axillary meristems, rosette leaves, flowering inflorescences and mature siliques were used for histochemical GUS staining analysis 30 days after germination. The result showed that both two lines exhibited similar GUS expression profiles, and that GUS activity was strongly observed in roots ( Figure 5B), the base of axillary meristems ( Figure 5D), stamens of flowers ( Figure 5E) and mature seeds located in mature siliques ( Figure 5F) but weakly detected in the tips and sides of cauline and rosette leaves ( Figure 5C,D), sepals and petals ( Figure 5E). However, no GUS activity was detected in stems ( Figure 5D). We further examined HbMFT1::GUS activity at 12 h after germination (HAG), and 1, 3, 5 and 7 days after germination (DAG) in T2 generation. Strong GUS activity appeared in whole seedlings and seed coats, with the highest expression in radicles from 12 HAG to 1 DAG ( Figure 5I,J), after which the GUS activity was decreased gradually in cotyledons ( Figure 5K-M). At three DAG, root meristems were elongated and the strongest GUS activity was still located in radicles also in root tips whereas faint activity was detected in the newly elongated regions of the roots ( Figure 5K). From five DAG onward, GUS activity was mainly restricted to hypocotyls and the root regions where root hairs grew but absent in root tips ( Figure 5L,M). At seven DAG, seedlings began generating true leaves, from which GUS activity was absent but slightly increased in main roots rather than axillary roots and root tips ( Figure 5M). These results indicated that the HbMFT1 promoter activity was mainly limited to roots, the base of axillary meristems, mature seeds, hypocotyls of post-germinated seedlings and stamens of flowers in transgenic plants, not completely consistent with the expression pattern in rubber tree, in which HbMFT1 showed predominant expression in seeds and male flowers. 2.6. Activity Analysis of GUS Fused with HbMFT1 Promoter in Response to ABA Treatment MFT homologs are regulated by ABA in some species [22,30] and five ABA cis-elements were found in HbMFT1 promoter in this study (Table S1). Therefore, we supposed that HbMFT1 promoter also responded to ABA. In order to confirm our hypothesis, we treated seven-day-old seedlings and rosette leaves of two transgenic lines, respectively, with 10 µM ABA [22] for 3, 6, 12 and 24 h, respectively. The treatment analysis demonstrated that the HbMFT1::GUS activity of the two lines was notably induced at 24th h and that the two lines had more strong blue stains distributed in cotyledons and roots of seedlings and the sides of rosette leaves than the control ( Figure 6A,C). Subsequently, ABA treatment at 10, 50, 100, 200 and 300 µM for 24 h, was used. The results indicated that all concentrations of ABA could induce GUS activity at 24th h, and that 200 and 300 µM of ABA induced the highest GUS activity in seedlings and rosette leaves ( Figure 6B,D). However, the transgenic seedlings and rosette leaves without ABA treatment retained the background GUS activity ( Figure 6B,D). Therefore, combining the fact of ABA cis-elements present in the HbMFT1 promoter and the result from the effect of ABA on activity of HbMFT1::GUS fusion in transgenic plants, it was reasonable to confirm that ABA acts as an activity inducer of the HbMFT1 promoter.

Overexpression of HbMFT1 Inhibited Seed Germination and Seedling Growth in Transgenic Arabidopsis Plants
HbMFT1 had the highest expression in embryos, and we speculated that it mainly functions to regulate seed development. To support this hypothesis, we compared the germination rate between the 35S::HbMFT1 transgenic and the wild-type (wt) Arabidopsis plants. More than 50 independent 35S::HbMFT1 transgenic Arabidopsis lines were obtained in the first generation. Through southern blot, we obtained five independent homozygous lines in the third generation ( Figure 7A). The qRT-PCR result showed that the expression of HbMFT1 was detected in line 35S::HbMFT1-10, 11 and 19, especially 35S::HbMFT1-11, in which the expression was the highest and remarkably higher than that of 35S::HbMFT1-10 and 19 ( Figure 7B). Therefore, we chose 35S::HbMFT1-10 and 11 for further function analysis. A notable observation was that during germination, 35S::HbMFT1-11 exhibited notably delayed germination as compared to 35S::HbMFT1-10 and wt plants ( Figure 7C). After vernalization for three days, wt and 35S::HbMFT1-10 Arabidopsis showed a respective germination rate of 48.34% and 60.55% at 36 h after germination (HAG), whereas 35S::HbMFT1-11 showed a germination rate of only 6.89% ( Figure 7C,D). Up to 48 HAG, the germination rates for wt and 35S::HbMFT1-10 were 92.06% and 89.21%, respectively, but still very low for 35S::HbMFT1-11 (38.87%). Finally, wt and 35S::HbMFT1-10 Arabidopsis completely germinated at 60 HAG, and 35S::HbMFT1-11 at 72 HAG. Intriguingly, at 12 HAG, both wt and 35S::HbMFT1-10 yielded distinctly bushy and vigorous root hairs, which were barely observed in 35S::HbMFT1/-11 ( Figure 8D). In order to further confirm the reliability of this experiment, we chose some other lines from 50 lines of T2 generation to repeat the study of seed germination ( Figure S1). The results showed that lines 35S::HbMFT1-20, 38 and 41 exhibit a delay in germination similar to that of the 35S::HbMFT1-11 line with high expression of HbMFT1, although they were not reflected in the southern blot and expression analysis of HbMFT1. Moreover, the root growth of 35S::HbMFT1-11 was also severely inhibited, whereas 35S::HbMFT1-10 exhibited a slight suppression but was not significantly influenced as compared to wt ( Figure 8A,B). At four days after germination (DAG), the roots of wt, 35S::HbMFT1-10 and 11 grew 4.8, 4.1 and 2.5 mm long, respectively, and then increased to 41.2, 33.9 and 20.3 mm long, respectively, by nine DAG (Figure 8C). For aerial parts, 35S::HbMFT1-11 grew more slowly, producing significantly less rosette leaves than wt and 35S::HbMFT1-10 ( Figure 8D). These results indicate that HbMFT1 negatively controls seed germination, growth and development.

Overexpression of HbMFT1 Delayed Flowering Time in 35S::HbMFT1 Transgenic Arabidopsis Plants
In order to determine whether expression of HbMFT1 could affect plant architecture, we analyzed the phenotype of 35S::HbMFT1 transgenic plants. Through morphological observation, overexpression of HbMFT1 did not cause any obvious morphological change in Arabidopsis under long-day conditions but delayed floral initiation. As shown in Figure 9A, 35S::HbMFT1-11 flowered at 37.24˘0.84 DAG, while 35S::HbMFT1-10 and wt flowered at 31.09˘0.83 and 30.32˘1.25 DAG, respectively. In addition, 35S::HbMFT1-11 produced 4.76˘0.60 cauline leaves and 14.94˘1.15 rosette leaves, significantly higher than those of wt (3.05˘0.40 cauline leaves and 11.63˘0.96 rosette leaves). As expected, 35S::HbMFT1-10 produced similar numbers of cauline and rosette leaves to wt due to its faint accumulation of HbMFT1 ( Figure 9B). Further analysis indicated that the delayed flowering time was related with higher expression levels of HbMFT1 and down-regulation of SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1), LEAFY (LFY), APETALA1 (AP1) and FRUITFULL (FUL) in transgenic plants, especially the expression levels of AP1 and FUL, which were notably reduced ( Figure 9C-F), even though the expression of AtFT in 35S::HbMFT1-11 was relatively higher than that of wt and 35S::HbMFT1-10 ( Figure 9G).

Discussion
Unlike FT-/TFL-like genes determined to be "florigen" and "antiflorigen", respectively, as a member of PEBP family, MFT-like genes were not intensively studied in most plant species, and their functions are hence less documented. In the present study, the expression of HbMFT1 in embryos ( Figure 3A) was in agreement with the high HbMFT1::GUS fusion activity in mature seeds of transgenic Arabidopsis ( Figure 5E) and the seed-specific cis-elements found in HbMFT1 promoter (Table S1), confirming that the primary activities of HbMFT1 in seed or embryo were similar to those of MFT homologs in Arabidopsi [22], wheat [25], Jatropha curcas [30], Picea abies [23], Zea mays [31], Citrus unshiu [32], rice [33] and tomato [34]. However, in orchid, abundance of the DnMFT transcript was strongly detected in auxiliary buds and leaves [35]. Overexpression of HbMFT1 gave rise to significant delay in seed germination in transgenic Arabidopsis relative to wild-type plants, as described for wheat (Triticum aestivum), in which Ta-MFT was induced by low temperature in mature seeds and further confirmed to inhibit germination as expressed in immature embryos driven by the maize (Zea mays) ubiquitin promoter [25]. Therefore, transgenic results in the study revealed that HbMFT1 plays a critical role in maintenance of seed development and maturation. However, AtMFT in Arabidopsis directly inhibited ABI5 through a negative feedback mechanism in response to ABA, thereby attenuating the effect of ABA on the suppression of germination [22], functioning analogously to HbMFT1 and Ta-MFT.
In addition, the expression level of HbMFT1 was progressively increased as inflorescence developed, suggesting that HbMFT1 may also be involved in the development of reproductive organs. In transgenic Arabidopsis plants ectopically expressing HbMFT1::GUS fusion, the stamens exhibited strong HbMFT1::GUS activity, which was weakly detected in sepals and petals and was consistent with pollen-specific cis-element existing in the promoter (Table S1), indicating that the HbMFT1 promoter conferred GUS activity preferential expression in stamen and may function to regulate development of the stamen. This might explain why the expression level of HbMFT1 was almost two-fold more abundant in the male flowers than the female flowers in rubber tree ( Figure 3A). MFT homologs were also speculated to be associated with development of reproductive organs in other plant species. Four MFT-like genes in Physcomitrella patens exhibited high expression in gametangia, and PpMFT2 and PpMFT4 showed strong expression in the sporophyte, suggesting their association with development of reproductive organs [26]. A high expression of PaMFT1 in Picea abies was also detected in pollen [23].
In transgenic Arabidopsis, high expression levels of HbMFT1 may affect development in two aspects. One was that post-embryo growth and development of transgenic Arabidopsis was dramatically repressed, with lower germination, shorter roots and fewer rosette leaves compared to wt before floral transition (Figures 7 and 8). The other one was that flowering was delayed, with more rosette and cauline leaves and down-regulation of SOC1, LFY, AP1 and FUL (Figure 9), each of which is the down-stream gene of FT and assigns a floral fate to meristem [11,[36][37][38][39][40][41], especially the transcript accumulations of AP1 and FUL that were reduced significantly, whereas the expression of LFY similar to SOC1 was not affected notably due to the fact that SOC1 is the direct upstream transcriptional activator of LFY [42,43]. AP1, known as a key floral meristem identity gene, is required to specify the identity of floral meristem in Arabidopsis [44]. Furthermore, FUL contributed to advancing floral initiation in addition to silique development [45]. Therefore, in the present study, HbMFT1 may negatively control the activity of AP1 and FUL through a kind of signalling pathway, thereby delaying formation of flowers. HbMFT1 as a development and flowering inhibitor acted similarly to MFT homologs reported in other plant species, such as orchid (Dendrobium nobileLindl), white and sitka spruce (P. glauca and P. sitchensis, respectively), of which each MFT homolog resulted in relative delay in flowering as heterologously expressed in Arabidopsis [35,46]. However, in grapevine, VvMFT transcript was mainly detected in shoot and the expression pattern was related with determination of inflorescence meristem as a flowering promoter [47], even though it was more closely related to HbMFT1 in genetic relationship. In Arabidopsis, Yoo et al. [21] found that over-expression of MFT in wild-type Arabidopsis could slightly accelerate flowering, but no remarkably delayed flowering phenotype was found in mft-1 mutant, suggesting its partial functional redundancy with FT. Overall, these results suggested that MFT homologs may exert different functions among plant species and their differences have yet to be investigated.
ABRE elements, to date, have been widely studied and confirmed to be involved in controlling the MFT promoter activity during seed development. In Arabidopsis, ABA-INSENSITIVE3 (ABI3) and ABI5 were considered to be two important ABA signaling components, both of which could directly bind to the MFT promoter through recognizing ABRE element to regulate germination of seeds in the ABA-signaling pathway [22]. ABI5 acted as a direct transcriptional promoter of MFT, which in turn inhibited the expression of ABI5, yielding a negative feedback loop, whereas ABI3 acted as a direct transcriptional repressor of MFT. ABI3 was identified to be an upstream promoter of ABI5 [48]. During seed development of Jatropha and Picea abies, similar expression patterns between ABI3 and MFT homologs were observed [23,30], suggesting that ABI3 indirectly promotes MFT transcriptional level by anegative feedback loop, so that MFT maintains seed maturation. Additionally, it has been reported that seed-or embryo-specific expression regulation depends on an ABA-responsive complex [32]. Therefore, in the present study, the highest expression of HbMFT1 in seeds of rubber tree suggest that it may regulate seed development depending on the ABA-signalling pathway in that the HbMFT1 promoter contains several ABRE elements (Table S1) and was confirmed to increase GUS activity in transgenic Arabidopsis with HbMFT1::GUS fusion when exposed to exogenous ABA stress ( Figure 6).
HbMFT1 was mainly induced in short-day conditions and showed small oscillations in darkness ( Figure 4A), suggesting that expression of HbMFT1 is regulated through both photoperiod and the circadian clock. Photoperiodic expression of the PEBP family has been reported in other species. Four homologs of MFT in Physcomitrella patens showed a peak expression after approximately 2 h after the onset of light in long days [26]. In addition, AtBFT [19], AtFT [19] and DnFT [35] were induced under long-day conditions. On the contrary, Hd3a [34] and ATC [20] were induced under SD conditions and showed a rhythmic expression.
In conclusion, we characterized HbMFT1 gene as a multifunctional regulator in rubber tree based on tissue-specific expression, temporal and spatial expression, putative plant regulatory elements analysis and ectopical expression analysis in Arabidopsis plants. Transgenic Arabidopsis plants ectopically over-expressing HbMFT1 substantially retarded seed germination, growth and development and flowering. Abundant HbMFT1::GUS activities in stamens and mature seeds of transgenic Arabidopsis plants were consistent with its expression in rubber tree and the putative seed-and pollen-specific cis-elements existing in the promoter. In addition, exogenous ABA could dramatically promote GUS activity in transgenic plants transformed with HbMFT1::GUS fusion because of the presence of ABRE elements in HbMFT1 promoter. Our study suggests that HbMFT1 may mainly retain the state of seeds maturation and regulate development of stamens in rubber tree. Moreover, seed-and pollen-preferential expressions indicate that the HbMFT1 promoter is ideal to regulate target gene expression in seeds or stamens of rubber tree or other plant species.

Plant Materials and Growth Conditions
Rubber trees of Hevea brasiliensis clone CATAS 7-33-97, including 3-month-old seedlings, 2-year-old, and 10-year-old rubber trees, were grown at the experimental plantation of the Rubber Research Institute, Chinese Academy of Tropical Agricultural Sciences (RRI-CATAS), Danzhou, Hainan, China. In order to study the tissue-specific expression of target gene, embryos and roots, stems, shoot apices and four developmental-stage leaves (bronze, color-change, pale-green and mature leaves) of 3-month-old seedlings, five developmental-stage inflorescences (I1: <0.5 cm; I2: =2 cm; I3: =4 cm; I4: =8 cm; I5: >8 cm and flowering), open male flowers, open female flowers and pericarps of 10-year-old rubber trees were collected. In addition, roots, stems, mature leaves and shoot apices of rubber trees at different ages were collected separately to study the spatial and temporal expression pattern of a target gene. In order to determine whether the target gene was influenced by photoperiod, 40 three-month-old seedlings with similar morphological characteristics were divided into two groups, with 20 seedlings each group. One group was planted under long-day condition (16-h light/8-h dark) at 28˝C, whereas the other group was grown under short-day condition (8-h light/16-h dark) at 28˝C. After one month, mature leaves were collected at the initiation of light (8 a.m.) and continued every 2 h during the whole day. We also assessed the effect of temperature on expression of the target gene. Under short-day condition (8-h light/16-h dark), 40 three-month-old seedlings with similar morphological characteristics were equally divided into two groups. The first group was transferred into a growth chamber, in which the culture temperatures were set at 28, 32 and 36˝C, respectively, whereas the second group was transferred into another growth chamber, in which culture temperatures were set at 28, 24, 20, 16 and 8˝C, respectively. Each temperature lasted for 3 days, and then was adjusted to next temperature. To avoid the interference of photoperiod and biological clock on expression, we harvested the mature leaves at the same time (10 a.m.).
Wild-type Arabidopsis thaliana seeds (Columbia) were vernalized for 3 days at 4˝C, and then grown on sterilized vermiculite containing 1/2ˆMS medium for 8 days before transplanted to soil. Seedlings were planted in a growth chamber under long-day conditions (16-h light/8-h dark) at 22˝C.

DNA and RNA Extraction
Genomic DNA from leaves of rubber tree was extracted according to Risterucci et al. [49]. The total RNA of both rubber tree and Arabidopsis was extracted according to the method of Tang et al. [50]. The total RNA extracted was then treated using Dnase I (Thermo Scientific, Waltham, Massachusetts, USA) to avoid contamination of genomic DNA. Agarose gel electrophoresis was used to assess the integrality of total RNA or DNA. The purity and concentration of total RNA, which was used for reverse transcription, was measured by NanDrop 2000 spectrophotometer (Thermo Fisher, Waltham, MA, USA) at wavelength of 230, 260 and 280 nm.

Isolation of Full-Length and Putative Promoter of MFT-Like Gene
JcMFT and AtMFT from Jatropha curcas and Arabidopsis were used as queries, and the local BLAST program was carried out with leaf transcriptome, which was finished by our research group (data unpublished) to search putative MFT-like genes. MFT-like genes were amplified by a pair of specific primers, HbMFT1(ORF)-F and HbMFT1(ORF)-R. In order to obtain the full-length cDNA of the MFT-like gene, 3 1 rapid amplification of cDNA ends (RACE) was conducted. Three adaptor primers named QT, Q0 and Q1 and two gene-specific primers, GSP1 and GSP2, were designed for 3 1 RACE. QT was used as primer for reverse transcription. Q0 and GSP1 were used for the first PCR program while Q1 and GSP2 were used for second PCR program. Amplified products were cloned into pMD19-T cloning vector (TaKaRa) and sequenced. A 2267 bp 5 1 flanking region utilized as putative promoter was amplified based on the result of the local BLAST program using genome database (unpublished data) and the genomic open reading frame sequence (ORF) of the MFT-like gene. The primer sequences used for 3' UTR, ORF and promoter amplification are shown in Table S2.

Bioinformatic Analysis
Genomic organization of the target gene was analyzed by aligning the cDNA with its corresponding genomic DNA using the online web server Clustal Omega (http://www.ebi.ac.uk/ Tools/msa/clustalo/) [51]. Multiple sequence alignment between target proteins and the PEBP family of other species was carried out using DNAMAN software 6.0 version (Lynnon Biosoft, San Ramon, CA, USA). For evolutionary analysis, we used MEGA (Molecular Evolutionary Genetics Analysis) software package version 5.0 (www. megasoftware.net) [52], in which N-J method with ClustalW software was used and N-J tree was produced from the results of 1000 bootstrap replicates [52].

Expression Analysis of Related Gense in Rubber Tree and Arabidopsis Transgenic Lines
We carried out quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) to assess the expression of target genes. First-strand cDNA was synthesized from three micrograms of total RNA in 20 µL reaction mixtures according to the manufacturer's instructions (RevertAid™ First Strand cDNA Synthesis Kit, Fermentas, Waltham, MA, USA). The qRT-PCR was conducted in a reaction volume of 10 µL, including 30 ng cDNA per sample, 1ˆSYBR ® PremixExTaq TM (TAKARA Biothnology Corporation, Dalian, China) and 0.2 µM each primer, and performed in 384-well plates with the CFX384 system (Bio-Rad Laboratories, Hercules, CA, USA). Relative expression analysis of each gene was calculated by Pfaffl method [53]. At2g28390 and At3G01150 from Arabidopsis [19,54,55] and HbRH2b, HbRH8 and HbYLS8 from rubber tree were used as reference genes in qRT-PCR [56]. Related primer sequences and reaction programs used for qRT-PCR are described in Table S3.

Construction of Binary Vector for Target Gene and Promoter-GUS Fusion and Transformation
Target genes were amplified based on a pair of specific primers (shown in Table S2), in which the EcoRI and XbaI enzyme site sequences were added in the 5 1 regions of forward and reverse primers, respectively. The products of amplification were digested with EcoRI and XbaI and cloned into the EcoRI-XbaI sites of pXCS vector which harbours a bar gene conferring resistance to herbicide to replace the multiple clone site (MCS). In addition, in order to construct the vector of promoter::GUS fusion, the above-mentioned 2267 bp HbMFT1 promoter fragment was subcloned into XbaI-NcoI site of pCAMBIA 3301. Both of these two types of constructions were introduced into Agrobacterium tumefaciens GV3101 by electroporation. Eventually, the resultant Agrobacterium strains were used for transformation of Arabidopsis wild-type plants according to the floral-dip method [57]. For selection of transformants, seeds vernalized for 3 days were plated on vermiculite supplemented with 1/2ˆMS medium solution for 4 days, and then sprayed with herbicide Basta at 50 mg/L every 3 days. After 10 days, bar-resistant transformants were selected and transplanted to soil in the growth chamber. Positive transformants were identified by Southern Blot method [58].

Phenotype Analysis of Transgenic Arabidopsis
In order to determine whether over-expression of HbMFT1 affects seed germination, the seeds of transgenic lines and wild-type (wt) Arabidopsis were vernalized for 3 days at 4˝C in 1/2ˆMS mediums, and then the seed germination was observed and counted using a stereo microscope (LEICA, Germany) per 12 h till the third day (namely 72 h). As for the measurement of root lengths, we took photos of the roots of the wt and transgenic lines, for at least 20 plants each, every day from the fourth to the ninth day after germination with 1 cm as a scale bar. Finally, the actual length of root for each line was calculated using Image-Pro Plus (IPP) software 6.0 (IPP 6.0-Media Cybernetics, Bethesda, MD, USA) based on scale.

ABA Treatment
Mature Arabidopsis seeds of transgenic plants were sterilized with solution containing 75% (v/v) ethanol and 0.05% Triton X-100 for 5 min, and the solution was then removed. These seeds were washed again using 95% ethanol for 5 s and transferred on a sterilized filter paper. When ethanol was completely volatilized, these seeds were sown in the plates with 1/2ˆMS medium. After vernalization at 4˝C for 3 days, these plates were transferred into a plant growth chamber at 22˘2˝C under a 16/8 h (light/dark) condition. At 7 days after germination, seedlings were divided into 10 groups (each group contained at least 6 seedlings), of which the first 5 groups were grown in 1/2ˆMS medium supplemented with 10 µM ABA for 3, 6, 12 and 24 h, respectively. The rest of the 5 groups of seedlings were grown in 1/2ˆMS medium supplemented with different concentrations of ABA (10, 50, 100, 200 and 300 µM) for 24 h. 1/2ˆMS Finally, all of seedlings sampled were stained using GUS solution. With regard to the treatment of rosette leaves with ABA, two independent single plants of each line were randomly selected as samples. Rosette leaves from a single seedling growing at 25 days after germination were treated with ABA at different concentrations and time points, with the same procedure as that for seedlings treatment with ABA. Seedlings and rosette leaves with no ABA treatment were used as controls.