Characterization of cis-polyisoprene produced in Periploca sepium, a novel promising alternative source of natural rubber

Introduction

Natural rubber produced by plants, known as polyisoprene, consists of isoprene units (C₅H₈)_n linked together in a 1,4 cis-configuration^1,2,3,4, which is the most widely used isoprenoid polymer. Natural rubber is considered one of the four most important and scarce industrial raw materials, along with steel, petroleum and coal, and its products have penetrated all aspects of industry and public life. Unlike most other biopolymers, natural rubber cannot be replaced by synthetic alternatives in many of its most significant applications due to its unique properties, which include resilience, elasticity, malleability, abrasion, impact resistance, and efficient heat dispersion^4,5,6,7,8. In 2023, the Association of Natural Rubber Producing Countries (ANRPC, http://www.anrpc.org) reported that the global production of natural rubber was 14.89 million tons, but its consumption reached over 15.12 million tons, and demand is still steadily increasing. Although more than 2000 species of higher plants produce latex consisting of cis-polyisoprene^4,6,9,10,11, Hevea brasiliensis Muell. Arg (rubber tree) is currently the only commercial source of natural rubber because of its high rubber yield and the excellent physical properties of rubber products^4,11,12. However, H. brasiliensis has strict climatic requirements, limiting its cultivation to specific tropical regions and preventing its expansion to nontropical countries^6,7,10. Over 90% of natural rubber is produced in Southeast Asia, particularly in Malaysia, Thailand and Indonesia^6,13. Consequently, the problem of insufficient capacity for natural rubber production has become increasingly urgent and has attracted worldwide attention.

In an effort to develop a novel alternative plant for commercial production of natural rubber in nontropical regions, no less than 8 botanical families, 300 genera and 2500 species that produce natural rubber in latex have been identified⁶. Only few species in addition to H. brasiliensis are known to produce rubber with relatively high molecular weights, i.e. Parthenium argentatum Gray and Taraxacum kok-saghyz Rodin, both of which are considered promising alternative rubber sources; other plants either have not yet been studied in sufficient detail or produce rubber of inferior quality and low-molecular-weight, which are not adequate for the rubber industry^{6,7,11,14,15,16}. However, P. argentatum and T. kok-saghyz have several disadvantages that become major barriers to their viable commercialization: the former does not tolerate the low winter temperatures in temperate-frigid regions and has been reported to trigger allelopathic influences on crops and livestock or cause severe human health problems^17,18, and the latter, which usually grows in cool and cold regions, mainly produces rubber in the roots, and unearthing roots is extremely labour intensive and may cause environmental problems^{6,15,19,20,21}.

In the present research, we selected 3 kinds of shrubs or herbs that produce latex, Periploca sepium Bunge, Apocynum venetum L. and Cynanchum chinense R.Br., for polyisoprene extraction and identification; these species are widely distributed in the arid and semiarid areas of Northwest, Northeast and North China. The molecular weight distributions of the extractives were analysed via gel permeation chromatography (GPC). Only P. sepium, which has a higher molecular weight distribution of polyisoprene, was considered promising as an alternative natural rubber source and was selected for further characterization. The molecular structure of its polyisoprene was determined by nuclear magnetic resonance (NMR), and the anatomical structures involved in the accumulation or biosynthesis of natural rubber were observed via Nile Red staining; in addition, three rubber particle-associated proteins, namely, cis-prenyltransferase (CPT), small rubber particle protein (SRPP) and rubber elongation factor (REF), were evaluated by immunostaining via spectral confocal laser scanning microscopy (SCLSM). The interactions between the three proteins were assayed via bimolecular fluorescence complementation (BiFC), and the comprehensive influences of the genes encoding the three proteins on rubber biosynthesis were evaluated in transgenic P. sepium via CRISPR-Cas9 genome editing. This research will promote the development of P. sepium as an alternative source for the production of natural rubber.

Results

Extractive identification from three kinds of plants that produce latex

Three kinds of plants (shrubs) that produce latex, P. sepium, A. venetum and C. chinense were selected for the rubber component polyisoprene extraction. The total contents of polyisoprene from the three plants were different by 1.12%, 0.51% and 0.21%, respectively (p < 0.01, Fig. 1a). The extracted polyisoprenes were subjected to GPC analysis for molecular weight distribution measurement. All of the polyisoprenes extracted from the three plants presented typical trimodal distribution curves. However, those of A. venetum and C. chinense were mainly centred at approximately 10²–10³, whereas that of P. sepium showed two peaks at approximately 10² and 10³, and a distinct peak at approximately 10⁵ (Fig. 1b). These results revealed that the polyisoprene extracted from P. sepium has a broader molecular weight distribution than those extracted from the other two plants do, especially in the 10⁴-10⁶ region.

**Fig. 1: Total contents and molecular weight distributions of the polyisoprenes extracted from 3 plants.**

According to the analysis of rubber in H. brasiliensis, polyisoprene usually has a bimodal distribution from 10⁴-10⁶²², and the low-molecular-weight distribution less than 10³ is considered polyprenols or dolichols (linear polyisoprene containing 9-23 isoprene units)^23,24,25. It can be concluded that the extractives from A. venetum and C. chinense with low molecular weights or small molecular masses are mainly polyprenols, and only those with a high molecular weight distribution greater than 10⁵ from P. sepium are suggested to be long-chain polyisoprene. The high molecular mass accounting for the peak area was 50.33%, and its number average molecular weight (Mn), weight average molecular weight (Mw) and molecular weight distribution index (Mw/Mn) were 4.24 × 10⁴, 3.89 × 10⁵ and 9.17, respectively (Fig. 1c, P. sepium peak 3).

Structural characterization of polyisoprene extracted from P. sepium

The polyisoprene extracted from P. sepium was analysed by ¹H-NMR and ¹³C-NMR spectroscopy. Four major signals appeared at 2.04, 5.12, 2.05, and 1.68 ppm, which were assigned to the methylene (-CH₂-), methyne (-CH=), methylene (-CH₂-), and methyl (-CH₃) protons (H1, H3, H4, and H5) in the ¹H-NMR spectra, while five major signals appeared at 39.35, 135.18, 125.00, 26.37, and 23.40 ppm, which were assigned to the methylene (-CH₂-), olefinic (-Ċ=), methyne (=CH-), methylene (-CH₂-), and methyl (-CH₃) carbons (C1, C2, C3, C4, and C5) in the ¹³C-NMR spectra, respectively (Fig. 2a). These signals were very similar to those of the cis-1,4-polyisoprene standard from H. brasiliensis, whereas they were different from those of the pure trans-1,4-polyisoprene standard from Eucommia ulmoides, particularly the characteristic signals at 1.98 and 16.01 ppm ascribed to the methyl (CH₃-) proton (H1) and methyl (-CH₃) carbon (C5) in the trans-isoprene units. This finding coincided with previous reports^24,26,27 and clearly indicated that the central isoprene units of the polyisoprene extracted from P. sepium are linked to each other head-to-tail in the cis-1,4-configuration.

Fig. 2: ¹H NMR and ¹³C NMR spectra of the polyisoprene extracted from *P. sepium.*

Some minor signals were observed at 1.57, 1.58, 1.59, 1.60 ppm in the ¹H-NMR spectra and at 19.74, 15.95, 16.00, 16.13 ppm in the ¹³C-NMR spectra (Fig. 2b), which were assigned to the methyl (-CH₃) protons or carbons (H5 or C5) in the ω-terminal (ω(trans)), trans-1 (ω-trans–trans), trans-2 (trans–trans–trans), and trans-3 (trans–trans–cis) isoprene units; these signals were distinctly different from the signals at 1.68 or 23.40 ppm in corresponding proton or carbon (H5 or C5) of the central isoprene units in the ¹H-NMR or ¹³C-NMR spectra described above that were assigned to cis-configuration (Fig. 2a). These signals indicate that its ω-terminal isoprene units are followed by 2-3 isoprene units in the trans-1,4-configuration^26,28,29.

The signals at 4.57 and 62.28 ppm in the ¹H-NMR and ¹³C-NMR spectra were assigned to the methylene (-CH₂-) proton or carbon (H4 or C4) in the α-terminal isoprene unit, and the signals at 3.64 and 45.36 ppm in the ¹H-NMR and ¹³C-NMR spectra were assigned to the trimethyl (-(CH₃)₃) structure (Fig. 2a). These signals suggest that its α-terminal isoprene unit may consist of monophosphate and diphosphate (cis–α(cis)-(OPO₃)_1-2) linked with phospholipids (N⁺(CH₃)₃)³⁰. The results demonstrated that the main chain of the polyisoprene extracted from P. sepium is cis-1,4-polyisoprene composed of an ω-terminal trans-isoprene unit, 2-3 trans-isoprene units, cis-isoprene units, and an α-terminal cis-isoprene unit linked with phospholipids (Fig. 2a, Proposed structure of the main chain polyisoprene extracted from P. sepium).

In addition, four minor signals appeared at 4.73, 2.03, 1.26, and 1.60 ppm, which were assigned to the methylene (CH₂=), methyne (-ĊH-), methylene (-CH₂-), and methyl (-CH₃) protons (H1, H3, H4, and H5) in the ¹H-NMR spectra, and five minor signals appeared at 109.33, 150.52, 42.81, 29.34, and 19.74 ppm, which were assigned to the methylene (CH₂=), olefinic (=Ċ-), methyne (-ĊH-), methylene (-CH₂-), and methyl (-CH₃) carbons (C1, C2, C3, C4, and C5) in the ¹³C-NMR spectra, respectively (Fig. 2c). These signals corresponded to the 3,4-isoprene unit, indicating that the polyisoprene extracted from P. sepium contains an abnormal 3,4-isoprene side group at approximately 3.6%, which was calculated from the signal area ratio between the methylene (CH₂=) proton (H1) in 3,4-isoprene and the methyne (=CH-) proton (H3) in cis-1,4-isoprene^31,32.

Anatomical observation of the structures involved in rubber accumulation in P. sepium

The relationships between the anatomical structures involved in the accumulation or biosynthesis of natural rubber and the particle-associated protein distributions in the stems of P. sepium were observed via SCLSM. The images in the cross-section under SCLSM bright field showed that P. sepium stem consisted distinct layers of cortex, phloem, cambium, xylem and pith (Fig. 3a–i). Some small circular holes similar to the laticifer structure were mainly distributed in the secondary phloem adjacent to the cambium and the pith layers, and the laticifer cells were surrounded by irregular parenchyma cells in the secondary phloem (Fig. 3a-ii) but were usually surrounded by 5-6 regular parenchyma cells in the pith (Fig. 3a-iii).

Fig. 3: Natural rubber accumulation, rubber particle-associated protein distributions and interactions in *P. sepium.*

Since the cryosections were stained with Nile Red and Calcofluor White stain, according to the reference regions of interest (ROIs) selection and measurement, the cell wall, cis-polyisoprene and lipid body could be separated by their specific fluorescence wavelengths at maximum intensities of 445 nm, 555 nm and 575 nm, which were unmixed in blue, yellow and magenta, respectively (Supplementary Fig. 1). In the cross-section, fluorescence images (yellow) revealed that the cis-polyisoprenes appeared mainly in the secondary phloem and the pith (Fig. 3b-i, ii, iii): most accumulated in the laticifer cells, and some dispersed in the surrounding cells. In the tangential section, fluorescence images (yellow) clearly showed that cis-polyisoprenes presented dispersed or nonarticulated linear shapes (Fig. 3b-iv, v, vi).

Immunostaining using primary antibodies against the three abundant rubber particle proteins PsCPT, PsSRPP and PsREF, which are considered playing important roles in natural rubber biosynthesis, were detected via SCLSM. The specificity of these antibodies against proteins extracted from P. sepium had been confirmed by Western blotting (Supplementary Fig. 2). The molecular weight of PsCPT (37.9 kDa) almost agreed with expectation, whereas those of PsSRPP and PsREF were higher than the calculated 26.6 kDa and 22.9 kDa, suggesting that they are post-translationally modified proteins³³. Fluorescence images (magenta) of the cross-sections revealed that all of the three proteins appeared in the secondary phloem and the pith (Fig. 3c–i, iv, vii), almost entirely overlapping with the cis-polyisoprenes present. In the magnified images, the fluorescence distributions in the secondary phloem revealed that PsCPT and PsSRPP were more abundant than PsREF was (Fig. 3c-iii, vi, ix), especially PsSRPP, which even extended to some radial rays in the secondary xylem (Fig. 3c-vi), whereas all proteins were present sporadically in the pith (Fig. 3c-ii, v, viii).

BiFC assay of protein interactions between PsCPT, PsSRPP and PsREF

To further assay protein-protein interactions in living cells, PsCPT, PsSRPP and PsREF were cloned and constructed into the BiFC vectors pSPYNE and pSPYCE with a split YFP system, and were coinfiltrated into tobacco leaves via Agrobacterium-infiltrated transient expression, respectively. Upon coexpression of pSPYCE-PsCPT with pSPYNE-PsSRPP, pSPYCE-PsREF with pSPYNE-PsSRPP, or pSPYCE-PsCPT with pSPYNE-PsREF, strong YFP fluorescence signals were detected around the plasma membranes in the epidermal cell layers of the leaves (Fig. 3d), indicating that the protein-protein interactions occur between PsCPT with PsSRPP, PsSRPP with PsREF, and PsCPT with PsREF.

Influence evaluation of PsCPT, PsSRPP and PsREF on rubber biosynthesis in P. sepium

The comprehensive influences of the three genes encoding PsCPT, PsSRPP and PsREF on natural rubber biosynthesis were evaluated via CRISPR-Cas9 genome editing. Binary CRISPR-Cas9 vectors with three different sgRNA target sites for each gene (Fig. 4a) were introduced into the stem segments of P. sepium via Agrobacterium-mediated transformation. After infection, cocultivation, differentiation and selection, several plantlets are regenerated. PCR analysis confirmed that some plantlets produced the predicted CRISPR associated protein 9 (Cas9), synthetic green-fluorescent protein with S65T mutation (sGFP) and neomycin phosphotransferase (NPT) II DNA fragments, indicating that the transgenes were present in these transgenic plants.

Bidirectional sequencing of the genomic DNA isolated from the PCR+ plantlets indicated that a considerable portion of the plantlets exhibited mutations within or near the sgRNA target sites of the three genes, with most deletions ranging from 1 to 37 nt, but rare substitutions or insertions (Supplementary Figs. 3–5).

The total contents of the cis-polyisoprenes extracted from the transgenic plants and their molecular weight distributions were determined via GPC analysis. Three representative mutagenesis plants of each PsCPT, PsSRPP and PsREF gene, namely, transgenic PsCPT-06, PsCPT-41, and PsCPT-63; PsSRPP-31, PsSRPP-34, and PsSRPP-36; and PsREF-17, PsREF-18, and PsREF-59, showed reduced total cis-polyisoprene contents to different degrees compared with that of the wild-type (nontransgenic negative, WT) control (p < 0.01), with the exception of PsCPT-63 (Fig. 4b). PsCPT-41 had a 37-nt deletion in the third sgRNA target site, resulting in the loss of 12 amino acid (aa) residues, along with 169 aa and termination codon alterations downstream, and the high-molecular-weight peak at approximately 10⁴-10⁶ was almost completely missing; in contrast, PsCPT-06, which had 2 nt substitutions in the second sgRNA target site and consequently only 2 aa alterations, exhibited the high-molecular-weight peak that was shifted forwards and smaller, and its area decreased from 46.35% in the WT to 5.96%; another PsCPT-63, which had 2 nt substitutions within and near the first sgRNA target site and only 1 aa alteration, exhibited the high-molecular-weight peak that was shifted forwards and its area decreased to 14.55%. Similarly, in PsSRPP-31 or PsSRPP-36, which had 2- or 1-nt deletions in the first sgRNA target site, resulting in 159 or 154 aa and termination codon alterations downstream, respectively, the high-molecular-weight peak was completely absent, whereas PsSRPP-34, which had 7- and 2-nt deletions in the first and second sgRNA target sites, and 1-nt substitution near the second sgRNA target site, resulting in a loss of 3 aa and 19 aa alterations downstream, exhibited the high-molecular-weight peak that was shifted forwards and significantly reduced in area to 5.97%; PsREF-17, which had 2-, 5-, 27-nt deletions in the three sgRNA target sites, resulting in a loss of 11 aa and 130 aa and termination codon alterations downstream, exhibited no high-molecular-weight peak; PsREF-18, which had 2- and 1-nt deletions in the first and third sgRNA target sites and a 1-nt insertion in the second sgRNA target site, resulting in 140 aa and termination codon alterations downstream; and PsREF-59, which had 5- and 1-nt deletions in the first and third sgRNA target sites and a 1-nt substitution in the third sgRNA target site, resulting in a loss of 2 aa and 145 aa alterations downstream, both exhibited a forwards-shifted high-molecular-weight peak with a significantly reduced area, i.e. 5.81% or 7.43% (Fig. 4c, d). These results indicated that mutagenesis of these three genes, PsCPT, PsSRPP and PsREF, can down-regulate the biosynthesis of cis-polyisoprenes in P. sepium.

Discussion

In recent decades, the demand for natural rubber has continued to increase, and further production of natural rubber from H. brasiliensis has been limited to simply planting more acreage and cultivating technical improvements, because of this species’ strict growing conditions, susceptible to various infectious diseases, small genetic diversity, etc^6,10,21,34. A shortage of natural rubber is expected in the near future. Although synthetic polyisoprene rubber has been successfully obtained using petroleum derivatives as starting materials under stereospecific catalysts and its application scopes continue to expand, unexpectedly, despite all the efforts, some properties (such as superior mechanical properties) of natural rubber are difficult to match even when both polymers have the same stereoregularity. This is because natural rubber is considered a naturally occurring nanocomposite. Its nanostructure and non-rubber components like proteins and lipids, even if the differences are very small, also influence rubber properties. That is to say, synthetic rubber alternatives are still unable to completely replace natural rubber in many significant applications^4,5,6,7,8. Consequently, it is necessary to search for other species for natural rubber production to meet its ever-increasing demands^{6,7,10,11,15,16}. In this study, we present a new alternative source, P. sepium, which can produce latex in its leaves, branches, stems, roots and contains the rubber component cis-polyisoprene at approximately 1.12% (w/w, the residue weight obtained from chloroform extraction/sample fresh weight), whose molecular weight distribution is broader than those of two other plants, A. venetum and C. chinense, especially in the 10⁴–10⁶ high-molecular-weight region, according to GPC analysis. Its high molecular mass accounting for the peak area was 50.33%, and its Mw was 3.89 × 10⁵, which was slightly less than that of H. brasiliensis (Mw 5 × 10⁵-1.31 × 10⁶) and approaching or greater than those of other alternative species, P. argentatum (Mw 4.76 × 10⁵–5.77 × 10⁵), T. kok-saghyz (Mw 4 × 10³–3 × 10⁵) and Asclepias speciosa (Mw 5.2 × 10⁴)^{6,7,10,14,15,35}. Since an Mw of 10⁵ or more is appropriate for commercial applications¹⁴, P. sepium is regarded as having potential for development as an alternative source for natural rubber production.

The main chain of the polyisoprene extracted from P. sepium is cis-1,4-polyisoprene. It is composed of an ω-terminal trans-isoprene unit, 2-3 trans-isoprene units, cis-isoprene units, and an α-terminal cis-isoprene unit linked with phospholipids, as determined via ¹H-NMR and ¹³C-NMR spectroscopy analysis. The molecular structure is in good accordance with that of H. brasiliensis or other cis-polyisoprene-producing plants, such as Solidago altissima^26,29,36,37. It can be presumed that in the initiation phase, the primers (initiators) of cis-polyisoprene polymerization in P. sepium comprise 3 types of allylic diphosphates, namely geranyl diphosphate (GPP, 2 isoprene units), farnesyl diphosphate (FPP, 3 isoprene units) and geranylgeranyl diphosphate (GGPP, 4 isoprene units). All of these are in the trans-1,4-configuration and are condensed by isopentenyl diphosphate (IPP, 1 isoprene unit) and its isomer dimethylallyl diphosphate (DMAPP) through the trans-isoprenyl diphosphate synthases of GPP, FPP, GGPP. This is the reason why there are 2-3 trans-isoprene units on the ω-terminal. The chain is subsequently extended by adding thousands of IPP to the primers in the cis-1,4-configuration. This addition (condensation) is carried out through the cis-isoprenyl diphosphate synthase, which is also referred to as cis-prenyltransferase (CPT, EC 2.5.1.20) in the cis-polyisoprene producing plants. In the final phase, the α-terminal forms and its monophosphate or diphosphate linked with phospholipids by hydrogen bound, when the size of the rubber droplet reaches a point where it does not allow further polymerization^3,4,7,28,31. In addition, the polyisoprene extracted from P. sepium also contains a 3,4-isoprene abnormal side group at approximately 3.6%. This percentage appears to be higher than that from H. brasiliensis, which is usually lower than 2%^28,31,32. An abundance of side groups may increase the rubber viscosity and stability to oxygen and heat³⁸.

The cis-polyisoprenes in P. sepium stem cryosections were observed via Nile Red histochemical staining under SCLSM. Nile Red is an uncharged lipophilic dye with intense fluorescence, and cis– or trans-polyisoprenes can be successfully distinguished from lipid bodies because their fluorescence wavelengths differ by 20–30 nm^33,39,40. The cis-polyisoprenes mainly appeared in the secondary phloem adjacent to the cambium and pith layers, and most of them accumulated in the laticifer cells in dispersed or nonarticulated linear shapes, with some particles dispersed in the surrounding cells. These observations suggested that cis-polyisoprene biosynthesis in P. sepium may first start in immature laticiferous cells as granules and then accumulate and fuse in the inner space along with mature laticiferous cells to laticifers. Laticifers in high-latex production trees, such as H. brasiliensis, are often continuous, concentric layered articulated laticifers in the secondary phloem, whereas nonarticulated laticifers are distributed at the outermost laticifer layer in the bark. However, trees with low latex production, such as P. argentatum, lack latex-producing cells⁴¹ but instead produce resin in resin canals, which are predominantly present in parenchyma tissue, and in the pith, and many are not continuous³³. P. sepium is likely in between, developing laticiferous cells or laticifers in the secondary phloem and pith but in a nonarticulated structure.

The fluorescence images of immunostaining under SCLSM revealed that the cis-polyisoprenes presented almost entirely overlapping with three proteins involved in rubber synthesis, PsCPT, PsSRPP and PsREF, and all of them appeared in the secondary phloem and the pith. PsCPT and PsSRPP were more abundant than PsREF in the secondary phloem, especially PsSRPP, which even extended to some radial rays in the secondary xylem, whereas all proteins were sporadically present in the pith. These results demonstrated that the biosynthesis of cis-polyisoprenes (natural rubber) mainly takes place in the laticiferous cells within the secondary phloem and accumulates in the laticifers both in the secondary phloem and pith as an end product. The rays may correlate with supplying the necessary network for cis-polyisoprene transport^39,42. It has been reported that SRPPs generally have much higher rubber biosynthesis activity than Refs. ^39,43,44,45. According to the BiFC assay, protein-protein interactions occur between PsCPT and PsSRPP, between PsSRPP and PsREF, and between PsCPT and PsREF. This result is analogous to reports in H. brasiliensis^46,47.

To further evaluate the comprehensive influences of the three genes encoding PsCPT, PsSRPP and PsREF on natural rubber biosynthesis, CRISPR-Cas9 genome editing was performed. Compared with that of the wild-type control, the total contents of the cis-polyisoprenes in the transgenic plants with deletion, substitution or insertion mutations were reduced to different degrees, and exhibited the high-molecular-weight peak at approximately 10⁴–10⁶ shifted forwards and smaller or completely absent. Our results demonstrated that PsCPT, PsSRPP and PsREF are the key genes that apparently play important roles in natural rubber biosynthesis and editing of these genes can significantly influence natural rubber content and molecular weight in P. sepium. As mentioned above, in chain extension during natural rubber biosynthesis, CPT has been identified from rubber-producing plants as a particle-bound rubber transferase responsible for the cis-1,4-polymerization of isoprene units from IPP onto the three kinds of allylic diphosphate primers, GPP, FPP and GGPP^4,48. SRPP and REF are the two most abundant rubber particle proteins and are necessary for CPTs to incorporate IPP isoprene units into rubber molecules^4,43,45,49. SRPP is speculated to generate mainly linear molecules with both high- and low-molecular-weight rubber chains, which are expected to be composed of active chain ends for chain elongation. On the other hand, REFs mostly compose branched molecules, which are expected to be derived from low-molecular-weight rubber chains via the aggregation of phospholipids on the surface of the rubber particle and play a termination role in the rubber elongation reaction⁵⁰. Several previous reports have proved that CPTs, REFs and SRPPs are very crucial for rubber production, and down-regulation of these genes by RNA interference (RNAi) silencing or CRISPR-Cas9 knockout in T. brevicorniculatum, T. kok-saghyz, Lactuca sativa, often showed a significant reduction in latex content and natural rubber biosynthesis activity, but their molecular weight was not always changed^{51,52,53,54,55,56}.

P. sepium is a perennial shrub belonging to the family Asclepiadaceae and is widely distributed in temperate Asia, southern Europe, and tropical Africa^57,58. P. sepium is excellent because of its high resistance to cold, drought, saline alkalis, and insects, and it can be planted in arid and semiarid areas^59,60. Its suitable planting density is as high as 90,000/ha, and 135,000 kg/ha/year of branches and leaves can be harvested, according to our field experiments. Although the content of cis-polyisoprene is currently low, P. sepium is outstanding in its wide acclimatization, fast growth, and easy receptibility to genetic transformation⁶¹ and may thus become an excellent experimental model plant for natural rubber biosynthesis research and an alternative source for natural rubber production. By identifying and overexpressing key genes that regulate polyisoprene biosynthesis, for example, the genes encoding CPT, SRPP, REF and their transcription factors or specific promoters⁶² for polyisoprene chain elongation, the genes encoding GPP, FPP, GGPP synthases for the condensation of allylic diphosphate initiators, or even the genes for the precursor IPP supply in the mevalonic acid (MVA) and the 2-C-methyl-d-erythritol 4-phosphate (MEP) pathways, thereby increasing rubber content (yield) and quality (high-molecular-weight), transgenic P. sepium may become a promising rubber-producing plant that can be grown from tropical to temperate regions.

Methods

Polyisoprene extraction

To identify a plant that has potential for development as an alternative source for natural rubber production, three kinds of plants that produce latex, P. sepium, A. venetum and C. chinense, were selected. The rubber component polyisoprene was extracted from branches with leaves via Soxhlet extraction and solvent precipitation. Approximately 10 g of the branches with leaves from each plant were homogenized into powder in liquid nitrogen, and 1 g of powder from each sample was accurately weighed and pretreated with 30 mL of 10% sodium hydroxide in a boiling water bath for 3 h for degradation of the cuticle or fibres in the cell wall. The sludge was rinsed several times with deionized water until the effluent became neutral (usually five times) and then oven-dried at 40 °C before being transferred to a 250 mL Soxhlet apparatus. The sludge was treated with 100 mL of ethanol for 6 h to separate alcohol-soluble small molecule materials such as chlorophyll or pigments in advance. After oven drying at 40 °C, 100 mL of chloroform (CHCl₃; Meridian Medical Technologies, USA) was added to extract the polyisoprene in an 80 °C water bath for 6 h. The solvent was then collected and concentrated under vacuum in a rotator evaporator. The polyisoprene content of each sample was determined (% w/w) by weighing the residue obtained from chloroform extraction, which was repeated in triplicate⁶³.

GPC analysis

The molecular weight distribution of the polyisoprene extracted from the above 3 plants was determined via GPC (HLC-GPC8320, Tosoh, Japan) using two PL gel MIXED-B (10 μm, 7.5 × 300 mm) columns. The polyisoprene samples at 2 mg/mL in chloroform were prepared by filtering through a 0.45 μm microporous membrane. GPC was carried out at a column temperature of 40 °C, using chloroform as the eluent at a flow rate of 0.8 mL/min for 30 min. Twelve kinds of polystyrene (Mw 1.0 × 10³–1.2 × 10⁶, TSKgel Standard Polystyrene Oligomer Kit, Tosoh, Japan) at 1 mg/mL in chloroform were used as the calibration standard. A differential refractive detector (RI) was used to record the signal, and Mn, Mw, Mw/Mn of each sample were automatically calculated via the Eco-SEC Workstation provided by the GPC instrument.

NMR spectroscopy analysis

The polyisoprene extracted from P. sepium, a cis-1,4-polyisoprene standard from H. brasiliensis (average Mw 38000; Sigma-Aldrich, USA), and a pure trans-1,4-polyisoprene standard from E. ulmoides (a gift from Northwest A&F University, China), were dissolved in deuterated chloroform (CDCl₃, 99.8% Chloroform-d + 0.05% v/v tetramethylsilane; Cambridge Isotope Laboratories, USA) at a concentration of 50 mg/mL. ¹H-NMR and ¹³C-NMR measurements were performed with a liquid NMR spectrometer (Avance NEO 700; Bruker, Germany) at 700 MHz and 298 K (25 °C) for 1 min and 4 h (corresponding to 16 scans and 7000 scans) respectively. Chemical shifts were reported with tetramethylsilane (TMS, 0.00 ppm) as an internal standard. The molecular structure was characterized by the chemical shift of the absorption peak in the NMR spectrum data via MestReNova software (Mestrelab Research SL, Spain).

Stem cryosection preparation

P. sepium stems approximately 10 mm in diameter were cut into 5-10 cm pieces and immediately fixed in 4% (w/v) freshly prepared paraformaldehyde (Macklin, China) in phosphate-buffered saline (PBS). The fixed stem was cut into 8 mm long pieces, rapidly embedded in Tissue-Tek O.C.T. compound (Sakura Finetechnical, Japan) and snap-frozen with liquid nitrogen. 60 μm-thick cryosections, including cross-sections and tangential sections, were prepared via a cryostat (CM-1950, Leica Microsystems, Germany) at −20 °C.

Histochemical staining

For histochemical staining, cryosections were mounted on glass slides and gently washed with PBS to remove the O.C.T. compound and were stained with 50 μg/mL Nile Red (Sigma-Aldrich, USA) in 50% ethanol for approximately 30 s to fluorescent stain the rubber components, such as cis-polyisoprenes and lipid bodies, followed by counterstaining with Calcofluor White stain (Sigma-Aldrich, USA) for approximately 30 min to stain the cell wall. After three washes in PBS, the sections were mounted in Vectashield mounting medium (Vector Laboratories, USA) and sealed with nail varnish as previously reported^39,40.

Primary antibody production

The three genes that encoded CPT, SRPP and REF were isolated from P. sepium according to genome and transcriptome sequencings, namely, PsCPT (AB082516), PsSRPP (PQ130390) and PsREF (PQ130391). The corresponding antigenic peptides MEKRSDQTSILENLC, CTDPADQPQMTEQEK and CMAASMATMKKESDG were synthesized and conjugated to keyhole limpet haemocyanin (KLH). A total of 400 μg of the synthetic peptides were used to immune New Zealand rabbits, which were boosted with 200 μg four times at 2-week intervals according to the company’s standard protocol (Beijing Protein Innovation, China). Blood serum was collected from the carotid artery of the rabbit, and the titres of the antisera were over 1:25600, as determined via indirect enzyme-linked immunosorbent assay (ELISA). The specificity of each primary antibody against the proteins extracted from P. sepium was confirmed by Western blotting, and Alexa Fluor 555-labelled Donkey Anti-Rabbit IgG (Abcam, UK) was used as the secondary antibody.

Immunostaining

The immunostaining was modified from a previously reported method^33,39. The cryosections were washed with PBS three times. The proteins in the sections were fixed with 4% paraformaldehyde in PBS for 30 min at room temperature, washed with PBS three times, blocked with 10% horse serum (Beijing Solarbio Science & Technology, China) in PBS for 30 min and then washed three times. The sections were subsequently reacted with 50 μL of each rabbit primary antibody diluted 1:100 in PBS, containing 10% horse serum and 0.1% Triton X-100 (Beijing Solarbio Science & Technology, China), and incubated overnight at 4 °C, protected from light. After being washed with PBS three times, the sections were incubated with 50 μL of the secondary antibody, Alexa Fluor 555-labelled Donkey Anti-Rabbit IgG diluted 1:200 for 2 h at room temperature. The sections were washed with PBS three times and further stained with Calcofluor White stain and mounted in mounting medium as described above.

SCLSM analysis

SCLSM analysis of the stained sections was performed at 25 °C via a spectral confocal laser scanning microscope system equipped with HC PL APO 10×, 20×, 40× lenses and Las X 4.4.0 software (Stellaris 5 DMi8, Leica Microsystems, Germany). The fluorescence was excited via the 405 nm line of a blue solid laser, and the 520 nm and 555 nm lines of a pulsed white laser, in all cases with laser power set to 2.0-10.0%. The emission spectra in the ranges of 410-560 nm, 525–675 nm, and 560–610 nm with 5 nm bandwidths were recorded for the detection of the cell wall stained with Calcofluor White stain, cis-polyisoprene stained with Nile Red and the secondary antibody stained with Alexa Fluor 555, respectively. Fresh latex of P. sepium tissue and the cis-1,4-polyisoprene standard from H. brasiliensis stained with Nile Red were used as reference samples. The ROIs fluorescence spectra from each of the 30 locations in the cell walls, cis-polyisoprenes and lipid bodies were measured and averaged from three successive scans to improve the signal-to-noise ratio. Image processing, including spectral unmixing, was performed via Las X 4.4.0 software.

BiFC assay

The full-length coding sequence (CDS) without the termination codon of each gene, PsCPT, PsSRPP and PsREF, was synthesized and constructed into binary vectors pSPYNE with N-terminal (1-155 amino acids + termination codon) of yellow fluorescent protein (YFP) and pSPYCE with C-terminal (156-239 amino acids + termination codon) of YFP for the BiFC assay⁶⁴. The resultant vectors, pSPYNE-PsSRPP, pSPYNE-PsREF, pSPYCE-PsCPT, and pSPYCE-PsREF, were introduced into Agrobacterium tumefaciens strain EHA105 respectively, and were used for transient expression in tobacco according to the protocol described. The Agrobacterium harbouring pSPYCE-PsCPT and pSPYNE-PsSRPP, pSPYCE-PsCPT and pSPYNE-PsREF, pSPYCE-PsREF and pSPYNE-PsSRPP were suspended in infiltration buffer containing 10 mM 4-morpholineethanesulfonic acid (MES), 10 mM MgCl₂, and 200 mM acetosyringone, pH 5.7, to an optical density (OD₆₀₀) of 0.8 and then coinfiltrated into the abaxial side of Nicotiana benthamiana leaves. 48 h after infiltration, epidermal cell layers of the leaves were assayed for YFP fluorescence at an excitation wavelength of 514 nm.

CRISPR-Cas9 vector construction

Three different sgRNA (small guide RNA) target sites (N20) followed by NGG (the protospacer-adjacent motif, PAM) of each gene, PsCPT, PsSRPP and PsREF, were predicted and selected via CRISPR direct (http://crispr.dbcls.jp) (Supplementary Table 1). Each sgRNA target site was separated by a 77-bp-long pre-tRNA^Gly sequence⁶⁵. The three tandemly arrayed tRNA-sgRNA architectures of each gene were synthesized and constructed into a binary CRISPR-Cas9 vector, referred to as pPsCPT-3×sgRNA-Cas9, pPsSRPP-3×sgRNA-Cas9, and pPsREF-3×sgRNA-Cas9 (Fig. 4a). The resultant vectors were introduced into A. tumefaciens strain EHA105.

Gene transformation

Agrobacterium-mediated transformation and transgenic plant analysis were performed according to our previous reports⁶¹. Briefly, the stem segments of P. sepium were inoculated with A. tumefaciens strain EHA105 harbouring the three vectors. Through callus selection, adventitious shoot and root induction, the regenerated kanamycin-resistant plantlets were confirmed via PCR analysis to investigate the presence of the transgenes, using Cas9, sGFP and NPT II gene primers (Supplementary Table 2).

Transgenic plant analysis

Genomic DNA was isolated from the PCR-positive (PCR+) plantlets to screen for mutants by bidirectional sequencing spanning all the sgRNA target sites, using primer pairs (Supplementary Table 3) for each target gene, PsCPT, PsSRPP and PsREF. The plantlets containing mutations in the target genes were transplanted to a greenhouse for further growth for one year. Polyisoprene extraction and GPC analysis of the mutagenesis plants were performed as described above⁶³.

Statistics and reproducibility

In polyisoprene extractions from the three kinds of plants, P. sepium, A. venetum and C. chinense, and from the transgenic plants via CRISPR-Cas9 genome editing of PsCPT, PsSRPP and PsREF, all were repeated in triplicate (n = 3). Their one-way analysis of variance (ANOVA) and least significant difference (LSD) was calculated via Statistica software (StatSoft GmbH, Germany). The Dot-plots were generated using GraphPad Prism software (GraphPad Software, USA) and the statistical details were described in the figure legends.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.