The relationship between latex metabolism gene expression with rubber yield and related traits in Hevea brasiliensis

Background Expression patterns of many laticifer-specific gens are closely correlative with rubber yield of Hevea brasiliensis (para rubber tree). To unveil the mechanisms underlying the rubber yield, transcript levels of nine major latex metabolism-related genes, i.e., HMG-CoA synthase (HMGS), HMG-CoA reductase (HMGR), diphosphomevalonate decarboxylase (PMD), farnesyl diphosphate synthase (FPS), cis-prenyltransferase (CPT), rubber elongation factor (REF), small rubber particle protein (SRPP), dihydroxyacid dehydratase (DHAD) and actin depolymerizing factor (ADF), were dertermined, and the relationship between rubber yield with their expression levels was analysed. Results Except HbHMGR1, HbPMD and HbDHAD, most of these genes were predominantly expressed in latex, and bark tapping markedly elevated the transcript abundance of the analyzed genes, with the 7th tapping producing the greatest expression levels. Both ethephon (ETH) and methyl jasmonate (MeJA) stimulation greatly induced the expression levels of the examined genes, at least at one time point, except HbDHAD, which was unresponsive to MeJA. The genes’ expression levels, as well as the rubber yields and two yield characteristics differed significantly among the different genotypes examined. Additionally, the latex and dry rubber yields increased gradually but the dry rubber content did not. Rubber yields and/or yield characteristics were significantly positively correlated with HbCPT, HbFPS, HbHMGS, HbHMGR1 and HbDHAD expression levels, negatively correlated with that of HbREF, but not significantly correlated with HbPMD, HbSRPP and HbADF expression levels. In addition, during rubber production, significantly positive correlations existed between the expression level of HbPMD and the levels of HbREF and HbHMGR1, between HbSRPP and the levels of HbHMGS and HbHMGR1, and between HbADF and HbFPS. Conclusions The up-regulation of these genes might be related to the latex production of rubber trees under the stress of bark tapping and latex metabolism. The various correlations among the genes implied that there are differences in their synergic interactions. Thus, these nine genes might be related to rubber yield and yield-related traits in H. brasiliensis, and this work increases our understanding of their complex functions and how they are expressed in both high-and medium-yield rubber tree varieties and low-yield wild rubber tree germplasm. Electronic supplementary material The online version of this article (10.1186/s12864-018-5242-4) contains supplementary material, which is available to authorized users.


Background
Natural rubber, composed mainly of cis-1,4-polyisoprene, is an important raw material of high economic value from which hundreds of industrial and medical products are manufactured [1]. Currently, more than 2500 species of latex-producing plants have been reported [2,3]. In addition, both guayule (Parthenium argentatum) and dandelion (Taraxacum sp.) are alternative sources of natural rubber [4]. The perennial tropical cash crop, para rubber tree (Hevea brasiliensis), belonging to the Euphorbiaceae family, which is widely planted in Southeast Asia, is still prevalent because of its greater productivity than the sum total of all other latex-producing plants, including guayule and dandelion, and because of the outstanding physical properties of its rubber products [5]. Hevea latex is the main worldwide resource of commercial natural rubber, accounting for up to 90% of the rubber traded in global markets [2,6,7].
Natural rubber belonging to the isoprenoid family of plant natural products is biosynthesized using pyruvate-derived acetyl-CoA as a substrate through the cytosolic mevalonate (MVA) pathway [8]. Many enzymes and proteins found in latex are implicated in this process. The biosynthetic pathway of rubber can be generally divided into three stages. At the initial and later stages, isopentenyl diphosphate (IPP), which is the basic carbon skeleton for universal isoprenoid biosynthesis, and its allylic isomer dimethylallyl diphosphate, which is the starter molecule for subsequent additions of IPP, are synthesized through the MVA pathway from their initial donor, pyruvate metabolite acetyl coenzyme A (acetyl-CoA). The biosynthesis of IPP in the MVA pathway requires a series of six enzymatic reactions. Initially, three molecules of acetyl-CoA are converted to MVA consecutively through acetoacetyl-CoA and 3-hydrox y-3-methylglutaryl (HMG)-CoA. MVA is then sequentially phosphorylated and ATP-dependently decarboxylated to ultimately form IPP [9][10][11]. In this second step of the MVA pathway involving the synthesis of IPP, acetoacetyl-CoA and acetyl-CoA are aldol-condensed to HMG-CoA by HMG-CoA synthase (HMGS, EC 2.3.3.10), and HMG-CoA is further converted to MVA by HMG-CoA reductase (HMGR, EC 1.1.1.34). In the last step of the conversion of acetyl-CoA to IPP, the latter is produced from diphosphomevalonate by diphosphomevalonate decarboxylase (PMD, EC 4.1. 1.33). At the middle stage, IPP is sequentially condensed with dimethylallyl diphosphate to produce geranyl diphosphate (GPP), with GPP to form farnesyl diphosphate (FPP), and with FPP to produce geranylgeranyl diphosphate [12,13]. Trans-FPP is synthesized from GPP by the enzyme FPP synthase (FPS, EC 2.5.1.1) [14]. cis-Prenyltransferase (CPT, EC 2.5.1.20) catalyzes the sequential cis-1,4-condensation of IPP with cis-FPP and its growing chains [8,15,16]. At the later stages, in which the allylic primer for the cis-1,4-polymerization of isoprene units from IPP is comprised of GDP, FDP and GGDP, rubber hydrocarbons of cis-1,4-polyisoprene are synthesized mainly by adding approximately 15,000 IPP molecules to an FPP molecule in a cis-1,4-configuration [8,17]. Rubber elongation factor (REF) is required for different prenyltransferases from broad sources to add a number of cis-IPP molecules to rubber chains [18]. In both large and small rubber particles, REF and small rubber particle protein (SRPP) form a dense proteolipidic monomembrane with the lipid monolayer that contributes to the colloidal stability of latex [19].
In addition to the well-documented substrate acetyl-CoA in the MVA pathway, IPP is also synthesized from pyruvate and glyceraldehyde-3-phosphate through the recently discovered methyl-erythritol 4-phosphate (MEP) pathway in the chloroplasts of plants [20]. Furthermore, other sources, such as leucine amino acid metabolism, provide the intermediate 3-hydroxy-3-methylglutaryl coenzyme A for IPP backbone synthesis [21]. Dihydroxyacid dehydratase (DHAD, EC 4.2.1.9) catalyzes a key step in the biosynthetic pathway producing the branched-chain amino acids isoleucine, valine and leucine that exists in plants [22]. This enzyme is stress labile because of its Fe-S cluster [23].
Laticifers in the phloem (inner bark) tissue of rubber trees are highly specialized cells for rubber production and comprise an effective defense system against environmental stress [24,25]. Currently, latex is commercially removed from laticifer cells by regularly slicing the mature stem bark of rubber trees until the wounded laticifer is blocked, causing the latex drainage to cease [7,26,27]. Actin depolymerizing factor (ADF), an important actin-binding protein, plays a role in the plugging process of laticifers [28].
Accordingly, the transcriptional levels of genes encoding these enzymes or proteins in latex are closely related to latex metabolism, including rubber biosynthesis, in Hevea trees. The sequences of latex metabolism-related genes have been reported, and expression analyses have been conducted in the latex of a few H. brasiliensis varieties [24,[29][30][31][32][33]. However, the simultaneous characterization of these genes in diverse genotypes of rubber trees has not been performed to date. To investigate the unitary molecular events concerned with latex metabolism, we performed expression analyses on nine latex metabolis m-related genes in various tissues of newly tapped young 'CATAS 73397' rubber trees and in latex from trees under various abiotic stresses, including mechanical wounding (tapping) and ethephon (ETH) and methyl jasmonate (MeJA) treatments. In addition, a comparative expression analysis of the aforementioned genes was conducted using different Hevea clones, including high-and medium-yield varieties and low-yield germplasm. A variation analysis for rubber yield and its related traits in high-and medium-yield varieties was conducted, as was a correlation analysis of gene expression levels in high-and medium-yield varieties with rubber yields and yield-related traits. Our results produced valuable insights into the functions of individual genes in latex metabolism.

Planting materials
This study involved 20 H. brasiliensis clones from seven cultivars already recommended for cultivation in China  (RRIM 600, PR 107, TSF 523, TSF  , in an area of approximately 0.12 ha. Each tested wild germplasm consisted of four trees, and the row and plant spacing were 7 m and 3 m, respectively. The trees were opened for tapping in the eighth year after cultivating to a height of 120 cm above the highest point of the bud union. The tapping of all trees for the collection of latex was done every 3 d using a half spiral system (1/2S D/3) without ETH stimulation on panel BO-1. In the event of bad weather (such as rainy days), tapping was delayed until the following day. Tender leaves, male and female flowers, stem bark and latex used for the qRT-PCR experiments were collected from seven-year-old untapped clonal rubber trees of 'CATAS 73397'. ETH and MeJA treatments before opening the rubber tree for tapping and fresh latex collection and preservation were conducted as previously described [25,34].

Total RNA extraction and cDNA synthesis
In the experiment, each sample included three independent biological replicates, and each biological replicate comprised at least 5 individual trees from which latex was collected and pooled as described previously [35]. Briefly, fresh latex was transferred to a vacuum bottle containing liquid nitrogen 30 s after tapping, and then immediately used or stored frozen at − 80°C. Total RNA was extracted from the fresh latex using the improved sodium dodecyl sulfonate method [36,37], and further treated with RNase-free DNase I (TaKaRa) to eliminate the residual genomic DNA. To check for DNA contamination, the RNA sample was subsequently used as a template for 18S gene amplification. The cDNA was synthesized from 2 μg of total RNA of latex samples from 20 types of rubber tree clones using a RevertAid™ First-Strand cDNA Synthesis Kit ( # K1621; Fermentas, Lithuania) containing an optimized mix of oligo(dT) and random primers following the supplier's instructions. Each cDNA sample was diluted 10-fold with sterilized ddH 2 O, and 1 μl of the 1:10 dilution was employed as template for qRT-PCR.

Gene expression analysis by RT-qPCR
Based on the corresponding rubber tree EST sequences deposited in NCBI database with the accession numbers listed in Additional file 1: Table S1, specific primer pairs having a Tm range of 56-61°C and flanking amplicon sizes of 100-200 bp were designed for nine Hevea genes (HbHMGS, HbHMGR1, HbPMD, HbFPS, HbCPT, HbREF, HbSRPP, HbADF, and HbDHAD) related to latex metabolism using the Beacon Designer version 8.13 software (Premier Biosoft International, USA) (Additional file 1: Table S1). Real-time PCR was carried out in a 20-μL reaction system independently using 20 serial 10-fold dilutions of single-stranded latex transcripts as templates with primer pairs (0.5 pmol) for each of the nine genes from all of the compared clones. The 18 s RNA gene was selected as an internal control. We performed the quantitative gene expression analysis on a LightCycler 2.0 instrument (Roche, Basel, Switzerland). The optimum PCR procedure conditions were denaturation at 95°C for 3 min, 39 cycles with denaturation at 94°C for 10 s, annealing at 58°C for 15 s and extension at 72°C for 30 s, followed by a last extension at 72°C for 10 min. All qRT-PCR reactions were conducted in three biological replicates each with technical triplicates. Based on the relative quantification method described by Pfaffl [38], the transcript levels were calculated by normalization relative to the transcript abundance of the 18S reference gene on a LightCycler 4.05.

Rubber yield estimation
Every 10 d throughout the year, latex volumes from each experimental unit were determined by applying the methods of individual collection and measurement. Latex dry rubber contents (DRCs) were measured by collecting latex from no less than five individual plants per experimental unit, and the dry matter productivity of the rubber trees was calculated in gram per tree per tapping.

Data analysis
To determine significant differences, data from the gene expression analysis and rubber production determination were subjected to Duncan's new multiple comparison test and analyzed using the SAS system 8.0 (SAS Institute Inc., Cary, NC, USA). The mRNA expression levels of nine genes at the three tapping times examined and all of the tested clones, as well as yield and yield-related traits among the tested cultivars, were compared, and pairwise correlations were computed between all tapping times and all tested clones' gene expression data or between all of the tested cultivars' gene expression data and corresponding available yield and yield-related trait data.

Results
Expression profiles of latex metabolic genes in different tissues of young mature Hevea 'CATAS 73397' trees To explore the expression levels of genes related to latex metabolism in newly tapped young mature rubber trees of the clonal cultivar CATAS 73397, total RNA from leaves, female flowers, male flowers, bark, and latex were analyzed quantitatively by RT-PCR using gene-specific primer pairs. The average expression levels of the nine genes examined in latex tissue were significantly greater than those in male flowers and bark (P < 0.05), and the average expression levels of all of the genes in leaves were significant greater than those in bark (P < 0.05), but no significant differences were observed among female flowers, male flowers and bark. HbHMGS, HbFPS, HbCPT, HbREF, HbSRPP and HbADF were strongly expressed in latex tissue ( Fig. 1), with values of 20.  46-fold (HbPMD) and 1.30-, 3.22-, 6.53-,and 25.48-fold (HbDHAD) those in female flowers, male flowers, bark and latex respectively (P < 0.01). Furthermore, the HbHMGR1, HbPMD and HbDHAD mRNA quantities were greatest in the leaf organs ( Fig. 1 HbDHAD, respectively. These results imply that the majority of these genes are specifically/preferentially expressed in laticifer cells at the first tapping. As rubber synthetic genes, they are expressed where rubber is synthesized. Fig. 1 Relative expression profiles of different genes in diverse organs and tissues from H. brasiliensis 'CATAS 73397' young mature trees. Data are means of three biological replications, and error bars represent standard errors of the means. Shortest significant range (SSR) multiple comparison tests were employed to perform one-way analyses of variance at the 0.01 and 0.05 levels. The different capital and lowercase letters above the bars indicate that the differences in expression values among different tissues for each gene are extremely significant (P < 0.01) and significant (P < 0.05), respectively. The different capital and lowercase letters in cell grids indicate that the differences in expression values of different genes within the same tissue are extremely significant (P < 0.01) and significant (P < 0.05), respectively Dynamic expression levels and correlations among genes related to latex metabolism in latex during the incipient periods of tapping Nine genes implicated in latex metabolism, HbHMGS, HbHMGR1, HbPMD, HbFPS, HbCPT, HbREF, HbSRPP, HbADF and HbDHAD, were all expressed at the 1st, 4th and 7th tappings after opening 'CATAS 73397' rubber trees (Fig. 2). The mRNA levels of the nine genes in the 4th and 7th tappings were greater than in the 1st tapping (P < 0.01). The expression levels of the nine genes were the greatest at the 7th tapping, with extremely significant values of 12.77 and 1.88 times those at the 1st and 4th tappings, respectively. Thus, we believe that the nine genes were transcribed at low levels in latex before the opening of rubber tree for tapping and a significant increase in their transcript levels was induced after mechanical stimulations. The greatest expression level at the 1st tapping were that of HbDHAD, which was 7.96, 4.57, 4.28, 3.18, 1.96, 1.40, 1.38 and 1.34 times those of HbHMGS, HbPMD, HbSRPP, HbCPT, HbREF, HbFPS, HbHMGR1 and HbADF, respectively, which were extremely significant differences (P < 0.01). The HbPMD expression level was greatest at the 4th tapping, and it was significant or extremely significant at 4.63, 2.64, 1.88, 1.34, 1.29, 1.23, 1.18 and 1.18 times (P < 0.05 or P < 0.01) higher than those of HbSRPP, HbDHAD, HbREF, HbFPS, HbCPT, HbADF, HbHMGS and HbHMGR1, respectively. However, at the 7th tapping, differences in mRNA expression levels in the latex were not significant among these genes, after the exclusion of the weakly expressed HbSRPP. Similarly, the latex output per tree per tapping in newly-opened young mature rubber plantings exhibited an increasing tendency as the tapping number increased. Among the three tappings, the greatest latex yield per tree occurred at the 7th tapping, and its 48.33 mL/t was 19.33 and 2.12 times greater than the yields of the 1st and 4th tappings, respectively. A prominent positive correlation occurred between the expression levels of the nine genes in the latex tissues and the tapping time (R = 0.893, P < 0.01) and the latex yield (R = 0.892, P < 0.01).
As shown in Table 1, in the 1st, 4th and 7th tappings, an extremely significant correlation was observed between the mRNA expressions of HbPMD and HbHMGR1 in latex tissues, with a correlation coefficient of 1.000 (P < 0.01). The dynamic expression of HbCPT was significantly positively correlated with in the expression levels of both HbHMGS and HbFPS, with correlation coefficients of 0.997 and 0.998, respectively (P < 0.05). There was an extremely significant positive correlation between the expression level of HbSRPP and those of HbHMGS and HbCPT, with correlation coefficients of 0.999 and 1.000, respectively (P < 0.01). The expression changes of HbADF mRNA had significantly positive correlations with those of HbHMGS, HbFPS, HbCPT and HbSRPP, with correlation coefficients of 0.997, 0.998, 1.000 and 1.000, respectively (P < 0.05 or P < 0.01).

Exogenous phytohormone-induced expression in latex of latex metabolic genes
In comparison to the control 0 h group (tapped 0 h after treatment), the average expression levels of the nine investigated genes in the 50 mg/L ETH-treated plants in Fig. 2 Variations in the mRNA levels of different genes in latex at different tapping times from young newly tapped 'CATAS 73397' rubber trees. Different capital superscript letters above the bars denote significant differences among the three tapping times for each gene at the 1% level. Different capital and lowercase letters in cell grids denote significant differences among different genes at the same tapping time at the 1 and 5% levels, respectively the 8-and 48-h groups were high, but the average expression levels in the 2-and 24-h groups were low. Statistically, significant differences were found in the mean expression levels only between the 48-h groups and the control, but not between the 2-24-h groups and the control. In the latex, the average relative expression level of the nine genes in the 48-h group was significant at 2.09, 2.55, and 2.47 times (P < 0.05) greater than in the 0-, 2-, and 24-h groups, respectively, but there were no significant differences among the other groups (P > 0.05). The expressions levels of HbPMD, HbFPS, HbADF and HbDHAD were greatest in the 48-h group (Fig. 3a), with highly significant levels of 4.38, 11.75, 4.14, and 3. The mean expression levels of the nine tested genes in the plants treated with 30 mg/L MeJA for 2-48 h were high in comparison to the control group, while no statistically significant differences were observed in the mean expression levels between the 2-and 48-h groups and the control. In the latex, the mean relative expression level of the nine genes in the 8-h group was a significant 3.64and 2.24-fold (P < 0.05 or P < 0.01) greater than the mean levels in the 0-and 48-h groups, respectively, but there were no obvious differences between the 8-h group and the 2-and 24-h groups (P > 0.05). The expression levels of HbHMGS, HbFPS, HbCPT, HbREF and HbSRPP were the greatest in the 8-h group (Fig. 3b), with extremely significant levels of 5.58-, 2.25-, 1.28-and 3.44-fold (P < 0.01; HbHMGS), 9.08-, 2.50-, 1.35-and 3.83-fold (P < 0.05 or P < 0.01; HbFPS) and 2.85-, 1.48-, 1.44-and 1.76-fold (P < 0.01; HbCPT) those in the 0-, 2-, 24-, and 48-h groups, respectively. The expressions level of HbREF in the 8-h group was a significant 1.79-and 1.36-fold (P < 0.05 or P < 0.01) greater than its levels in the 0and 24-h groups, respectively. The expression level of HbSRPP in the 8-h group was significantly greater, at 2.24-, 1.44-and 1.27-fold (P < 0.05 or P < 0.01) than in the 0-, 2-, and 24-h groups. HbHMGR1, HbPMD and HbADF were strongly expressed in the 24-h group, with values 1.34-, 1.75-, 1.28-and 1.31-fold (P < 0.01; HbHMGR1) and 6.34-, 1.49-, 1.17-and 3.85-fold (P < 0.01; HbADF) those of the

Expression profiles of latex metabolic genes in different clones
To examine the expression patterns of natural latex metabolic enzyme/protein genes, we performed qRT-PCR expression analyses of these genes using highand medium-yielding cultivated varieties and wild germplasm with low yields. The qRT-PCR analyses of gene expression were carried out in ten varieties and ten wild germplasm for each gene. There were highly significant differences in the expression levels of the genes in latex among the 20 clones ( HbFPS was expressed at an extremely significant 1.88-fold (P < 0.01) greater level than HbCPT. HbFPS and HbDHAD had significant 1.81-and 1.64-fold (P < 0.05) greater expression levels than HbREF, and HbHMGR1 was expressed at a significant 1.68-fold (P < 0.05) greater level than HbCPT.  (Table 2). In addition, the average expression levels of different genes in clone CATAS 87662 were significantly greater than in clone AC/F/7 38/63 (P < 0.05). Nevertheless, no obvious differences were found among the other clones (P > 0.05). The expression values of the same gene in different rubber tree clones were different. The change in the amplitude of the expression value of a given gene in the 20 clones ranged from 4.17-6016.20-fold, with the greatest variation in the HbADF gene's expression level, while HbHMGS had the least variation. The coefficient of variation for the average expression values of nine genes was 26.07%, which suggests that major differences in the average gene expression levels exists among different clones.

Comparison of latex yields per tree per tapping from different cultivars in different months
Because the latex yields per tree per tapping from the majority of the wild germplasm were only enough to investigate gene expression, an assessment of their yields was not possible. A monthly dynamic analysis of latex yields per tree per tapping from ten cultivars revealed that the major differences existed among the latex yields of the cultivars in different months (Additional file 2: Table S2). Latex yields per tree per tapping in October and November were significantly greater than those in May, June and July (P < 0.05). The latex yield per tree per tapping in September was significantly greater than in July (P < 0.05). However, no remarkable differences were observed among the other months (P > 0.05). The latex yields per tree per tapping from ten cultivars decreased in the following months: October > November > September > August > June > May > July. Latex yields per tree per tapping from the same cultivar in different months underwent a certain change, the amplitude of which was 1.41-4.98-fold, with the change in 'CATAS 73397' being the largest, and the greatest yield being 4.98-fold the lowest value of 21.05 mL latex per tree per tapping. The least amount of change occurred in 'CATAS 78426' , and the lowest latex yield per tree per tapping decreased to 49.67 mL compared with the greatest yield. Thus, high and moderate harvest-time latex-yielding cultivars are sensitive to the month of harvest. That is, the latex yield per tree per tapping of different yield-level cultivars fluctuated greatly in different months.
There were obvious differences among the latex yields per tree per tapping from different cultivars in the same month (Additional file 2: Table S2). The greatest latex yields per tree per tapping among genotypes in May and June were all 'TSF 628' , those in August and October came from 'CATAS 879' , and those in July, September and November came from 'CATAS 87662' , while all of the lowest latex yields per tree per tapping in May, June, July, August and September came from CATAS 73397, and those in October and November were from 'PR 107' and 'TSF 523' , respectively. There was a significant difference, at the 1% level, between the two cultivars in the same month. The variation coefficients for the monthly mean latex yields per tree per tapping and latex yields per tree per year for the ten rubber tree cultivars were

Comparison of DRCs in different cultivars and different months
The dynamics of the monthly DRCs of the ten cultivars revealed that there was a greater deviation in DRC among all of the cultivars in different months (Additional file 3: Table S3). The DRC in May was very significantly greater than those in October and November (P < 0.01), and significantly greater than that in August (P < 0.05). DRCs in June, July and September were all very significantly greater than that in November (P < 0.01). The DRC in July was much greater than that in October, which was significant greater than that in November (P < 0.05). The DRCs of 10 cultivars decreased gradually in the following order: May > July > September > June > August > October > November. A certain variation appeared among the DRCs of the same cultivars in different months, and the range of variation was 1.15-1.76 times, with the maximum variation occurring in 'CATAS 72059' , and the greatest content was 1.76 times of the lowest content, at 18.92%. The minimum variation was present in 'PR 107' , and the lowest content decreased 5.08% compared with the largest value. Thus, the alternate month had a great effect on high, moderate and low DRC-containing cultivars. There was an obvious monthly fluctuation in the DRCs of various cultivars. DRCs in the same month showed marked differences among different cultivars (Additional file 3: Table S3). The DRCs of 'CATAS 87662' in May, June, September and November were the greatest among the different genotypes, while the DRCs of 'PR 107' were greatest in July, August and October. The DRCs of 'TSF 523' in July, August and October were all the lowest; those of 'TSF 192' were the lowest in June and September, and 'CATAS 879' and 'CATAS 72059' had the lowest DRC values in May and November, respectively. There was a very significant difference between the two cultivars in same month (P < 0.01). The variation coefficient for the monthly mean DRC value of the ten rubber tree cultivars was 10.89%, which indicated that the differences in monthly mean DRCs was not negligible among the various genotypes.

Differential analyses of dry rubber yields per tree per tapping of different cultivars and different months
The monthly changes in the dry rubber yields per tree per tapping of 10 cultivars indicated that their dry rubber yields of all had obvious differences during the different months (Additional file 4: Table S4). The dry rubber yield per tree per tapping in September was extremely significantly greater than the yields in May, June and July (P < 0.01), and significantly greater than the yield in August (P < 0.05). The dry rubber yield per tree per tapping in October was very significantly greater than the yields in June and July (P < 0.01) and significantly greater than the yield in May (P < 0.05). The dry rubber yields per tree per tapping in August and November were significantly greater than those in June and July (P < 0.05). The dry rubber yields per tree per tapping of ten cultivars decreased as follows: October > November > September > August > May > June > July. Although differences in the monthly tappings' average dry rubber yield per tree was mainly determined by the latex-producing ability of the Hevea trees, this latex-producing ability can be affected by the leaf phenology of the rubber tree. Additionally, daily mean photosynthetically active radiation and total precipitation within a month, as well as temperature and wind levels, during tapping are important factors that influence the monthly tappings' average dry rubber yield per tree. Dry rubber yields per tree per tapping of the same cultivar showed a certain variance, of 1.33-6.70-fold, among different months, with the greatest variance being that of 'CATAS 73397′. The greatest dry rubber yield per tree per tapping was 6.70-fold that of the lowest yield, which was 3.89 g. The minimal variance occurred in 'CATAS 78426′, in which the lowest dry rubber yield per tree per tapping decreased by 11.65 g compared with the greatest value. Thus, high and moderate harvest-time dry rubber yielding cultivars are susceptible to monthly changes. The dry rubber yield per tree per tapping of different yield-level cultivars showed a large fluctuation in various months.
The dry rubber yields per tree per tapping in the same month displayed marked differences among different cultivars (Additional file 4: Table S4). The dry rubber yields per tree per tapping of 'CATAS 87662' in May, July, August, September and November were the greatest among the various genotypes, while the greatest yields during June and October were from 'CATAS 78426' and 'CATAS 879' , respectively. The lowest dry rubber yields per tree per tapping in May, June, August and September were from 'CATAS 73397' , in July from 'CATAS 72059' , and in October and November from 'TSF 523'. There were significant differences between cultivars in same month (P < 0.01). The variation coefficient of monthly mean dry rubber yields per tree per tapping of the ten rubber tree cultivars was 45.03%, suggesting that the monthly mean dry rubber yields per tree per tapping has a greater variation among different genotypes.

Differences in dry rubber yields per plant per month from different cultivars
The dynamic change analysis for dry rubber yields per tree per month from 10 cultivars indicated that there were notable differences among different months for all of the cultivars (Additional file 5: Table S5). The dry rubber yield per tree per month in September was extremely significantly greater than the yields in June, July and August (P < 0.01) and significantly greater than those in May and November (P < 0.05). The dry rubber yield per tree per month in October was very significantly greater than those in June and July (P < 0.01) and significantly greater than that in August (P < 0.05). The dry rubber yields per tree per month from the 10 cultivars decreased as follows: October > September > November > May > August > July > June. The dry rubber yields per tree per month of the same cultivar in different months had definite variations of 1.33-9.88-fold. The variation in 'CATAS 73397' was the greatest, and the greatest yield was 9.88-fold the lowest value at 27.25 g dry rubber per tree per month. The lowest variation occurred in 'TSF 628' , and the lowest dry rubber yield per tree per month decreased by 108.31 g in comparison with the greatest yield. Thus, high and moderate monthly dry rubber-yielding cultivars are particularly sensitive to the change in month. The dry rubber yields per tree per month of different yield-level cultivars fluctuated with the changing months.
There were apparent differences in monthly dry rubber yields per tree in the same month among different cultivars (Additional file 5: Table S5). 'CATAS 87662' had the greatest dry rubber yield per tree per month in May, July, September and November, while 'CATAS 78426' , 'TSF 628' and 'CATAS 879' had the greatest yields in June, August and October, respectively. 'CATAS 73397' had the lowest dry rubber yield per tree per month for May, June, August and September, while 'CATAS 72059' , 'PR 107' and 'TSF 523' had the lowest yields in July, October and November, respectively. There was only a 1% significant difference between cultivars in the same month. The variation coefficients for the monthly average dry rubber yields per tree and the annual dry rubber yields per tree of ten rubber tree cultivars were both 47.12%, which demonstrated that there were conspicuous differences in both yield values among the various genotypes.

Correlation of latex metabolic gene expression levels in Hevea brasiliensis with dry rubber yield and its related traits
Correlation analyses of the relative expression levels of nine genes encoding key regulatory enzymes or proteins in natural latex metabolism in the latex tissues of the mature tapped rubber trees of ten cultivars with the latex yields per tree per tapping, monthly DRC, dry rubber yields per tree per tapping and dry rubber yields per tree per month for the 7 months from May to November 2013, as well as with the annual latex yields per tree and annual dry rubber yields per tree, were performed ( Table 3) HbCPT mRNA expression level in latex was positively correlated with the monthly dry rubber yields per tree in September and November, with correlation coefficients of 0.641 and 0.705, respectively (R 0.05 = 0.632; R 0.01 = 0.765). However, the HbREF mRNA expression level in latex was negatively correlated with the monthly dry rubber yields per tree in October, with a correlation coefficient of − 0.675. The HbDHAD mRNA expression in latex was significantly and highly significantly positively correlated with the monthly dry rubber yields per tree from May to November, except for June, and both the mean monthly and annual dry rubber yields per tree, with correlation coefficients were 0.633, 0.638, 0.667, 0.719, 0.638, 0.875, 0.738 and 0.738, respectively. However, the expression levels of HbPMD, HbSRPP and HbADF in latex were not markedly correlated with dry rubber yield or its related traits, indicating that mevalonate diphosphate decarboxylase and SRPP could not be rate limiting enzymes in cis-polyisoprene biosynthesis in rubber tree. Thus, HbHMGS, HbHMGR1, HbFPS, HbCPT, HbREF and HbDHAD may play important roles in the rubber synthesis process. Additionally, the latex and rubber yields are influenced by not only the expression of rubber biosynthesis-related genes in latex but also by the expression of latex drainage-related genes in latex. The HbREF mRNA expression level in the high-yield variety CATAS 87662, which has a high DRC, was lower than in Table 3 Correlation coefficients between the expression of latex metabolism genes in H. brasiliensis cultivars with the dry rubber yield and its related traits variety CATAS 523. This may indicate that the lower expression of HbREF in latex is favorable to rubber formation when its DRC is higher, and vice versa, because REF not only plays important roles in rubber biosynthesis but may also be involved in latex coagulation [19,39].

Mutual correlations among latex metabolic gene expression levels in H. brasiliensis
To determine whether there is a synergic interaction among the different enzyme/protein genes related to the synthesis and metabolism of rubber latex that aids in the co-regulation of rubber synthesis and accumulation, the correlations among their expression levels were analyzed. As shown in Table 4

Variation in the expression of latex metabolic genes in the CATAS 73397 clone under typical stimulation conditions
Rubber latex of H. brasiliensis is essentially the cytoplasm of laticifers (Chow et al., 2011); therefore, latex metabolism and its regenerative capacity are important factors determining rubber yield. In rubber production, the application of an appropriate level of ETH can greatly improve the latex yield of rubber trees. Bark tapping-related damage and exogenous ETH and MeJA stimulation result in the expressional changes of many genes in the laticiferous cells of rubber trees, and they especially induce the up-regulated expression of latex metabolism-related genes that are involved in latex production and drainage. To increase the overall understanding of the functions of latex production-and drainage-related genes, the transcriptional levels of HbHMGS, HbHMGR1, HbPMD, HbFPS, HbCPT, HbREF, HbSRPP, HbADF and HbDHAD in the leaves, male and female flowers, trunk bark and latex from the mature virgin rubber clonal trees of CATAS 73397 were determined by qRT-PCR. The expression levels of latex metabolic genes fluctuated considerably among different tissues from newly tapped young clonal trees of CATAS 73397. Among these, HbHMGS, HbFPS, HbCPT, HbREF, HbSRPP and HbADF were maximally expressed in latex, while HbPMD and HbDHAD were minimally expressed in latex and HbHMGR1 was moderately expressed in latex. Furthermore, SRPP was expressed in all tissues rather than being restricted to laticifers in a manner similar to that of dandelion SRPP [40] but different from that of HbREF.
Using qRT-PCR, we also determined the transcript levels of nine genes in latex from newly tapped plants under regular bark tappings and plants treated with exogenous ETH and MeJA. The expression levels of the nine genes (encoding critical enzymes or proteins associated with natural latex metabolism) in latex from young newly tapped trees were all inducible at least at one time point, with the exception of HbDHAD expression in plants treated with exogenous MeJA, and the expression levels varied based on the stimuli, the induction duration and the gene type, revealing that the up-regulation of the nine genes might be related to the latex production and drainage of rubber trees and beneficial to the systematic defense in laticifers against plant injury.
The expression levels of the nine genes were distinctly elevated in latex tissues in the trunks of the clonal rubber trees of 'CATAS 73397' without ETH treatment after the occurrence of mechanical wounding (excluding the first tapping), which suggests that the induced expression of these genes may be a form of self-protection in the tree body. The expression levels of the nine genes had positive correlations with the number of tappings (P < 0.01) and latex yield (P < 0.01), which was in agreement with the observation that, to some extent, the strength of the response to mechanical damage was positively correlated with the degree of damage. The expression levels of these genes were not only up-regulated after the 7th tapping, but also their expression values, with the exception of HbSRPP, were similar.
The expression levels of HbPMD, HbFPS, HbADF and HbDHAD were strongly induced after a treatment with 50-mg/L ETH for 48 h, which is the later stage of the treatment. A 50-mg/L ETH treatment induced the maximal expression level of HbHMGR1, as well as HbHMGS, HbCPT, HbREF and HbSRPP, at 2 and 8 h, respectively, which represented the early-middle stages of treatment. However, the strong induction of HbHMGS, HbFPS, HbCPT, HbREF and HbSRPP, as well as HbHMGR1, HbPMD and HbADF, after treatment with 30-mg/L MeJA appeared at 8 and 24 h, respectively, which represented the middle-later stage of the treatment. Although the high expression levels of HbHMGS, HbCPT, HbREF and HbSRPP induced by ETH and MeJA both occurred at 8 h after treatment, the sensitivity and speed of the defense responses of these genes to exogenous MeJA are closer to each other than to those of externally applied ETH. Moreover, there were significant positive correlations between the expression levels of both HbPMD and HbFPS with the duration of the ETH treatment (P < 0.05). Our results support the natural rubber production standard that regular bark tapping should not be allowed to begin before ETH applications to the tapping panel have been sustained for at least 24 h.
Together with tissue-specific expression levels, these results suggest that, except HbDHAD, which was absolutely unresponsive to externally applied MeJA stimuli, the expression levels of the examined genes were all increased after regular bark tapping or exogenous ETH and MeJA treatments, regardless of whether their expression levels in the latex of the mature virgin rubber trees were higher, lower or moderate compared with other tissues that were analyzed. Our data corroborate that the mRNA expression levels of HbREF and HbSRPP are highly expressed in the laticifer latex in H. brasiliensis and can be stimulated by mechanical wounding (tapping) and ETH treatments (Berthelot et al., 2014a).
Dynamic changes in the rubber yield and yield-related traits during different tapping months Dynamic trends of yield per tree and the relative characteristics of rubber from May to November after tapping were determined for 10 rubber tree cultivars that showed extremely significant differences in dry rubber yields and the yield-related characteristics. There were significant monthly changes in the rubber yield and its related characteristics. Furthermore, the correlation analysis revealed that the number of tapping months was highly positively correlated with the latex yield and dry rubber yield of individual plants per tapping, with correlation coefficients of 0.929 and 0.889 (R 0.05 = 0.755; R 0.01 = 0.875), respectively, and extremely negatively correlated with the monthly dry rubber content, having a correlation coefficient of − 0.886, but not with the monthly dry rubber yield of single plant, which had a correlation coefficient of 0.732. The reason for the lack of correlation may be that the tapping times of some cultivars in different months varied from those of other cultivars. Therefore, on the whole, the rubber accumulated gradually with an increase in tapping months, which was consistent with the local phenomenon of rush-harvesting rubber for 3 months, from approximately September to November in Hainan.

Monthly changes in rubber yield and its related traits among genotypes of H. brasiliensis
The monthly mean latex yields per tree per tapping of the 10 varieties ranged from 61.74 to 162.80 mL, with a variation coefficient of 12.46-52.59%. The monthly mean dry rubber contents of the ten varieties fluctuated from 26.61 to 36.87%, with a variation coefficient of 5.95-18.81%. The monthly mean dry rubber yields per tree per tapping of the ten varieties varied from 14.27 to 60.81 g, with a variation coefficient of 10.28-68.72%. The monthly average dry rubber yields per tree of the ten varieties ranged from 112.02 to 507.75 g, with a variation coefficient of 12.88-74.87%. Thus, there were evident differences in the monthly changes in the rubber yield and its related traits during various harvest periods among the ten different rubber tree varieties.
Differential gene expression profiles in the latex metabolic pathway and their relevance to dry rubber yield and yield-related traits 198' and ' AC/AB/15 54/980' , and the expression levels of HbREF were a significant 17.88-, 4.11-, 1.91-and 4.01-fold (P < 0.01) lower than those of HbCPT in the '6-62' , 'RO/PB/ 1 2/78' , 'MT/IT/14 30/18' and ' AC/F/7 38/63' clones, respectively. The different results of the two studies are presumably due to the testing of different HbREF family members and different experimental techniques for detecting HbCPT and HbREF expression levels. Aoki et al. [15] used the rubber elongation factor (GenBank accession number X56535) and cDNA library sequencing, whereas a new rubber elongation factor gene HbREF (GenBank accession number MF361124) cloned in our laboratory and real-time qRT-PCR were used in the present study. Previously, REF mRNA transcript levels in high-yielding clones were a significant three-to four-fold higher than in low-yield clones, and the REF gene expression pattern had a positive correlation with latex (rubber) yield [48]. However, in this study, REF mRNA transcripts were not relatively more abundant in high-yielding clones than in low-yield clones, and the HbREF expression levels in 20 clones with contrasting yields showed no correlations with latex yield, except for latex yields per tree per tapping in November. The different results of the two studies may be due to the selection of different experimental materials with different growth years and different tapping frequencies for assessing HbREF expression levels of different HbREF family members. Priya et al. [48] used 18-year-old regularly tapped trees that were subjected to tapping every 2 d with a GenBank KX179469 sequence, while young tapped trees were subjected to tapping every 3 d with a GenBank MF361124 sequence in the present study. Similarly, no correlation was found between the SRPP gene expression pattern and latex (rubber) yield. HbREF and HbSRPP expression levels in the tested clones with contrasting yields may not have correlate with rubber accumulation of rubber because these families have many members that are involved in rubber synthesis and linked on rubber particles [19]. Nonetheless, the induction of both REF and SRPP by regular tapping after the opening of young mature rubber trees of clone CATAS 73397 correlated with increased natural rubber synthesis, and REF and SRPP gene expression levels were also induced by both ETH and MeJA treatments.
During the latex flow of mature rubber trees, actin was intercepted at the ends of the cut laticifers and participated in forming proteinaceous networks having microfilament skeletons. The laticifer blocking and wound healing of tapped H. brasiliensis can be realized by the proteinaceous network. Finally, they cease the latex flow of rubber trees [49,50]. Actin depolymerizing factor (ADF) exists widely in eukaryotes and is an important actin-binding protein. Because the equilibrium processes of the actin cytoskeleton are critical for its function, ADF may play a role in the plugging process of the laticifer by modulating actin filament polymerization and depolymerization to regulate the dynamic assembly of the actin cytoskeleton [28]. qRT-PCR analysis in this study showed that not only is HbADF predominantly expressed in latex compared with the other eight genes, but also the expression level of this gene varied considerably in the different clones, suggesting that it is involved in modulating the latex drainage of rubber trees.
Isoprenoids are formed by the polymerization of so-called "isoprene activating molecule" IPP. IPP is synthesized from acetyl CoA through acetoacetyl-CoA and mevalonate [25,51] and is also produced by leucine deamination and transformation to form 3-hydroxy-3-methylglutaryl coenzyme A [21]. In plant cells, pyruvate is not only the principal precursor of alanine, valine and leucine, but it is also the intermediate molecular of IPP and polyisoprene synthesis. Furthermore, dihydroxy-acid dehydratase is the third most common enzyme of the biosynthetic pathway forming branched-chain amino acids, including isoleucine, valine and leucine. In this research, the expression level of HbDHAD in latex was significantly positively correlated to different degrees and extents with the latex yield per tree per tapping, average monthly latex yield per tree per tapping, monthly DRC, monthly dry rubber yield per tree and mean monthly and annual dry rubber yields per tree. Thus, the gene might be essential for rubber biosynthesis in H. brasiliensis.

Conclusions
Several clones were used to evaluate the production potential of rubber trees with different origins and genetic backgrounds. We selected the mature virgin rubber clonal trees of CATAS 73397 treated with regular bark tapping, exogenous ETH and MeJA, as well as untreated trees, and investigated the expression levels of nine genes involved in latex metabolism in these clones. Several laticifer-specific genes were involved in the MVA pathway of rubber synthesis or latex metabolism, such as IPP synthesis. In this study, we demonstrated that HbHMGS, HbFPS, HbCPT, HbREF, HbSRPP and HbADF were strongly expressed in latex, while HbHMGR1, HbPMD and HbDHAD were strongly expressed in leaves. Moreover, the nine genes' expression levels in latex during the 1st, 4th and 7th tappings were positively correlated with the times of the tappings and the latex yield in 'CATAS 73397′. Both bark tapping and ETH and MeJA stimulation could significantly increase the transcript levels of HMGS, HMGR, PMD, FPS, CPT, REF, SRPP, DHAD and ADF, which showed their greatest expression levels during the last stage of incipient tapping after tree opening. Under ETH and MeJA treatments, the greatest gene expression levels occurred in the later and early-middle stages of treatment and in the early-middle stages of treatment, respectively. A few significant variations in gene expression existed among different genotypes of H. brasiliensis. Although these are initial studies on the correlations of latex metabolism-related gene expression levels with rubber yield and yield-related traits, the study advances our knowledge of these genes and how they are expressed in both high-and medium-yield rubber tree varieties and low-yield wild rubber tree germplasm.