Engineering Triterpene and Methylated Triterpene Production in Plants Provides Biochemical and Physiological Insights into Terpene Metabolism1[OPEN]

Targeting nonnative triterpene biosynthesis within the cell sheds light on cofactors and substrates and uncovers a new homeostatic mechanism. Linear, branch-chained triterpenes, including squalene (C30), botryococcene (C30), and their methylated derivatives (C31–C37), generated by the green alga Botryococcus braunii race B have received significant attention because of their utility as chemical and biofuel feedstocks. However, the slow growth habit of B. braunii makes it impractical as a production system. In this study, we evaluated the potential of generating high levels of botryococcene in tobacco (Nicotiana tabacum) plants by diverting carbon flux from the cytosolic mevalonate pathway or the plastidic methylerythritol phosphate pathway by the targeted overexpression of an avian farnesyl diphosphate synthase along with two versions of botryococcene synthases. Up to 544 µg g−1 fresh weight of botryococcene was achieved when this metabolism was directed to the chloroplasts, which is approximately 90 times greater than that accumulating in plants engineered for cytosolic production. To test if methylated triterpenes could be produced in tobacco, we also engineered triterpene methyltransferases (TMTs) from B. braunii into wild-type plants and transgenic lines selected for high-level triterpene accumulation. Up to 91% of the total triterpene contents could be converted to methylated forms (C31 and C32) by cotargeting the TMTs and triterpene biosynthesis to the chloroplasts, whereas only 4% to 14% of total triterpenes were methylated when this metabolism was directed to the cytoplasm. When the TMTs were overexpressed in the cytoplasm of wild-type plants, up to 72% of the total squalene was methylated, and total triterpene (C30+C31+C32) content was elevated 7-fold. Altogether, these results point to innate mechanisms controlling metabolite fluxes, including a homeostatic role for squalene.

Terpenes and terpenoids represent a distinct class of natural products (Buckingham, 2003) that are derived from two universal five-carbon precursors: isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP). In eukaryotic fungi and animals, IPP and DMAPP are synthesized via the mevalonate (MVA) pathway, whereas in prokaryotes, they are synthesized via the methylerythritol phosphate (MEP) pathway. In higher plants, the pathways are present in separate compartments and are believed to operate independently. The MVA pathway in the cytoplasm is predominantly responsible for sesquiterpene (C15), triterpene (C30), and polyprenol (greater than C45) biosynthesis and associated with the endoplasmic reticulum (ER) system. The MEP pathway resides in plastids and is dedicated to monoterpenes (C10), diterpenes (C20), carotenoids (C40), and long-chain phytol biosynthesis. All these compounds are usually produced by plants for a variety of physiological (i.e. hormones, aliphatic membrane anchors, and maintaining membrane structure) and ecological (i.e. defense compounds and insect/animal attractants) roles (Kempinski et al., 2015). Terpenes are also important for various industrial applications, ranging from flavors and fragrances (Schwab et al., 2008) to medicines (Dewick, 2009;Niehaus et al., 2011;Shelar, 2011).
The utility of terpenes as chemical and biofuel feedstocks has also received considerable attention recently. Isoprenoid-derived biofuels include farnesane (Renninger and McPhee, 2008;Rude and Schirmer, 2009), bisabolene (Peralta-Yahya et al., 2011), pinene dimers (Harvey et al., 2010), isopentenal (Withers et al., 2007), and botryococcene (Moldowan and Seifert, 1980;Hillen et al., 1982;Glikson et al., 1989;Mastalerz and Hower, 1996). The richness of branches within these hydrocarbon scaffolds correlate with their high-energy content, which enables them to serve as suitable alternatives to crude petroleum (Peralta-Yahya and Keasling, 2010). Indeed, some of them are already major contributors to current-day petroleum-based fuels. One of the best examples of this is the triterpene oil accumulating in the green alga Botryococcus braunii race B, which is considered a major progenitor to oil and coal shale deposits (Moldowan and Seifert, 1980). This alga has been well studied, and the major constituents of its prodigious hydrocarbon oil are a group of triterpenes including squalene (C30), organism-specific botryococcene (C30), methylated squalene (C31-C34), and methylated botryococcene (C31-C37; Metzger et al., 1988;Huang and Poulter, 1989;Okada et al., 1995), which can be readily converted into all classes of combustible fuels under hydrocracking conditions (Hillen et al., 1982).
The unique biosynthetic mechanism for the triterpenes in B. braunii was recently described by Niehaus et al. (2011), and a series of novel squalene synthaselike genes were identified (Fig. 1). In short, squalene synthase-like enzyme, SSL-1, performs a head-to-head condensation of two farnesyl diphosphate (FPP) molecules into presqualene diphosphate, followed by a reductive rearrangement to yield squalene (C30) by the enzyme SSL-2, or is converted by SSL-3 to form botryococcene through a different reductive rearrangement (Niehaus et al., 2011). Methylated derivatives are the dominant triterpene species generated by B. braunii race B (Metzger, 1985;Metzger et al., 1988), and these derivatives are known to yield higher quality fuels due to their high energy content and the hydrocracking products derived by virtue of having more hydrocarbon branches. Triterpene methyltransferases (TMTs) that can methylate squalene and botryococcene have been successfully characterized by Niehaus et al. (2012). TRITERPENE METHYLTRANSFERASE1 (TMT-1) and TMT-2 prefer squalene C30 as their substrate for the production of monomethylated (C31) or dimethylated (C32) squalene, while TMT-3 prefers botryococcene as its substrate for the biosynthesis of monomethylated (C31) or dimethylated (C32) botryococcene (Fig. 1). These TMTs are believed to be insoluble enzymes; they exhibit large hydrophobic areas, and their activities were only observed in vitro using yeast microsomal preparations (no activity was observed when expressed in bacteria; Niehaus et al., 2012).
Although plants and microbes are the natural sources for useful terpenes, most of them are produced in very small amounts and often as complex mixtures. In contrast, B. braunii produces large quantities of triterpenes, but its slow growth makes it undesirable as a viable production platform (Niehaus et al., 2011).  (Niehaus et al., 2011;Niehaus et al., 2012). SSL-1 catalyzes the condensation of two farnesyl diphosphate (FPP) molecules to presqualene diphosphate (PSPP), which is converted to either squalene or botryococcene by SSL-2 or SSL-3, respectively. Squalene can also be synthesized directly from the condensation of two FPP molecules catalyzed by squalene synthase (SQS). TMT-1 and TMT-2 transfer the methyl donor group from S-adenosylmethionine (SAM) to squalene to form monomethylated and dimethylated squalene, whereas TMT-3 acts on botryococcene to form monomethylated and dimethylated botryococcene (Niehaus et al., 2012).
Nevertheless, metabolic engineering and synthetic biology offer many strategies to manipulate terpene metabolism in various biological systems to achieve high-value terpene production with high yield and high fidelity for particular practical applications (Nielsen and Keasling, 2011). Many successes have been achieved in engineering valuable terpenes in heterotrophic microbes, such as Escherichia coli (Nishiyama et al., 2002;Martin et al., 2003;Ajikumar et al., 2010) and Saccharomyces cerevisiae (Ro et al., 2006;Takahashi et al., 2007;Westfall et al., 2012;Zhuang and Chappell, 2015). The strategies developed in these efforts usually take advantage of specific microbe strains whose innate biosynthetic machinery is genetically modified to accumulate certain prenyldiphosphate precursors (e.g. IPP or FPP), which can be utilized by other introduced terpene synthase (s) for the production of the desired terpene(s). For example, greater than 900 mg L 21 bisabolene was produced when bisabolene synthase genes from plants were introduced into FPP-overproducing E. coli or S. cerevisiae strains (Peralta-Yahya et al., 2011). High levels of farnesane production for diesel fuels were also achieved by reductive hydrogenation of its precursor farnesene, which was generated from a genetically engineered yeast (e.g. Saccharomyces cerevisiae) strain using plant farnesene synthases (Renninger and McPhee, 2008;Ubersax and Platt, 2010). However, terpene production using microbial platforms is still dependent on exogenous feedstocks (i.e. sugars) and elaborate production facilities, both of which add significantly to their production costs.
Compared with microbial systems, engineering terpene production in plant systems seems like an attractive target as well. This is because plants can take advantage of photosynthesis by using atmospheric CO 2 as their carbon resource instead of relying on exogenous carbon feedstocks. Moreover, crop plants such as tobacco (Nicotiana tabacum) can generate a large amount of green tissues efficiently when grown for biomass production (Schillberg et al., 2003;Andrianov et al., 2010), making them a robust, sustainable, and scalable platform for large-scale terpene production. Nonetheless, compared with microbial platforms, there are only a few examples of elevating terpene production in bioengineered plants. This is due partly to higher plants being complex multicellular organisms, in which terpene metabolism generally utilizes more complex innate machinery that can be compartmentalized intracellularly and to cell/tissue specificities (Lange and Ahkami, 2013;Kempinski et al., 2015). Significant efforts have been made to overcome these obstacles to improve the production of valuable terpenes in plants, including monoterpenes (Lücker et al., 2004;Ohara et al., 2010;Lange et al., 2011), sesquiterpenes (Aharoni et al., 2003Kappers et al., 2005;Wu et al., 2006;Davidovich-Rikanati et al., 2008), diterpenes (Besumbes et al., 2004;Anterola et al., 2009), and triterpenes (Inagaki et al., 2011;Wu et al., 2012). Among these, engineering terpene metabolism into a subcellular organelle, where the engineered enzymes/ pathways can utilize unlimited/unregulated precursors as substrates, appears most successful. For example, Wu et al. (2006Wu et al. ( , 2012 expressed an avian farnesyl diphosphate synthase (FPS) with foreign sesquiterpene/triterpene synthases targeted to the plastid to divert the IPP/DMAPP pool from the plastidic MEP pathway to synthesize high levels of the novel sesquiterpenes patchoulol and amorpha-4,11diene up to 30 mg g 21 fresh weight and the triterpene squalene up to 1,000 mg g 21 fresh weight. This strategy appears to be particularly robust because it avoids possible endogenous regulation of sesquiterpene and triterpene biosynthesis, which occurs normally in the cytoplasm, and relies upon more plastic precursor pools of IPP/DMAPP inherent in the plastid, which are primarily derived from the local CO 2 fixation (Wright et al., 2014).
The goal of this study was to evaluate the prospects for engineering advanced features of triterpene metabolism from B. braunii into tobacco and, thus, to probe the innate intricacies of isoprenoid metabolism in plants. In order to achieve this, we first introduced the key steps of botryococcene biosynthesis into specific subcellular compartments of tobacco cells under the direction of constitutive or trichome-specific promoters. The transgenic lines expressing the enzymes in the chloroplast were found to accumulate the highest levels of botryococcene. Triterpene methyltransferases were next introduced into the same intracellular compartments of selected high-triterpene-accumulating lines. A high yield of methylated triterpenes was also achieved in transgenic lines when the TMTs were targeted to the chloroplast. Through careful comparison of the levels of triterpenes and the methylated triterpene products in the various transgenic lines, we have also gained a deeper insight into the subcellular distribution of the triterpene products in these transgenic lines as well as a better understanding of methylation metabolism for specialized metabolites in particular compartments. These findings all contribute to our understanding of the regulatory elements that control carbon flux through the innate terpene biosynthetic pathways operating in plants.

Targeting Botryococcene Biosynthesis to the Cytoplasm Versus Chloroplasts
An earlier study demonstrated that plastid-targeted engineering of a foreign squalene synthase (SQS) and an FPS can successfully divert carbon flux from the MEP pathway to accumulate a high level of squalene in transgenic tobacco (Wu et al., 2012). That study revealed that the availability of IPP/DMAPP precursors was adequate but that strong regulatory mechanisms were absent in the chloroplast for novel squalene (C30) production to occur. This, in turn, led us to utilize this strategy to engineer botryococcene (C30) biosynthesis into tobacco plants. However, botryococcene biosynthesis requires two squalene synthase-like enzymes, SSL-1 and SSL-3, to catalyze successive reactions to make the botryococcene product (Niehaus et al., 2011). This is in contrast to squalene biosynthesis, which requires only a single enzyme, squalene synthase ( Fig. 1; Niehaus et al., 2011). We chose to overexpress two chimeric versions of botryococcene synthase. One is SSL1-3 (Fig. 2B), which is a fusion of the SSL-1 and SSL-3 enzymes by a peptide linker, which exhibited a 2-fold greater accumulation of botryococcene when expressed in yeast in comparison with simple coexpression of the two enzymes separately (Niehaus et al., 2011). The second design is referred to as SSL1-3M (Fig.  2B), in which the SSL1-3 chimeric enzyme has 71 amino acids of the C terminus of B. braunii squalene synthase (M) appended to its C terminus. This construct thus contains a membrane-spanning domain that was hypothesized to improve botryococcene productivity in engineered yeast by integrating the enzyme into the ER membrane in order to promote proximity between enzymes for substrates (Niehaus et al., 2011). The overall gene constructs thus consist of either botryococcene synthase SSL1-3 or SSL1-3M directed by a cassava mosaic promoter (Pcv; Verdaguer et al., 1996) and the avian FPS (Tarshis et al., 1994) driven by the cauliflower mosaic virus 35S promoter (Pca; Benfey et al., 1990). An N-terminal, plastid-targeting signal sequence (tp) from the Rubisco small subunit gene of Arabidopsis (Arabidopsis thaliana; Lee et al., 2006) was also inserted onto the chimeric SSL1-3 constructs to target these enzymes to the chloroplast compartment, whereas constructs without the signal sequence would target the encoded proteins to the cytoplasmic compartment. The respective gene constructs ( Fig. 2B; Table I) were introduced into tobacco accession KY 1068 by standard Agrobacterium tumefaciens transformation methodology. Thirty or more T0 independent transgenic lines were generated, and the leaf materials from different transgenic plants were extracted and analyzed by gas chromatography-mass spectrometry (GC-MS) and gas chromatography-flame ionization detection (GC-FID). When evaluated by gas chromatography, a unique molecule was detected in the extraction from some of the transgenic plants (Supplemental Fig. S1E) that was not evident in any of the wild-type plants (Supplemental Fig. S1A). This unique chemical peak had identical retention time and mass spectrum (Supplemental Fig. S2A) to a botryococcene standard Figure 2. Triterpene contents of independent T0 transgenic lines transformed for novel botryococcene synthesis. A, Schematic outline of the MVA and MEP pathways operating in the cytoplasm and chloroplast compartments, respectively, and conceptual strategies to divert carbon flux from these two pathways for the biosynthesis of botryococcene by heterologous expression of avian FPS and a chimeric botryococcene synthase (SSL1-3) targeted to membranes or not with a C-terminal membrane-spanning domain (M). GPP, geranyl diphosphate; GGPP, geranylgeranyl diphosphate B, Tobacco accession KY 1068 was transformed with the indicated gene constructs, consisting of the chimeric botryococcene synthase SSL-1 fused to SSL-3 via a linker peptide with (SSL1-3M) or without (SSL1-3) a membrane-associating domain (Niehaus et al., 2011), coupled with the avian FPS gene. Expression of both engineered genes was under the direction of strong, constitutive viral promoters (Pcv and Pca, respectively; Benfey et al., 1990;Verdaguer et al., 1996). A plastid-targeting signal sequence (tp) was fused to the 59 end of the respective genes where indicated. Thus, the constructs with tp will target the enzymes to the chloroplasts, and those without tp will express the enzyme in the cytoplasm. C, Antibiotic-selected T0 lines propagated in the greenhouse were assessed from botryococcene accumulation at a relatively young age (1 month old; light green) and mature age (6 months old; green) by GC-MS. Three representative, independent elite transgenic lines from each engineered construct were chosen to illustrate their capacity for botryococcene production. FW, Fresh weight. (Niehaus et al., 2011) and was confirmed as botryococcene by 1 H-NMR and 13 C-NMR analyses (Supplemental Figs. S3 and S4).
We observed that transgenic lines engineered with the construct that directed botryococcene synthase (SSL1-3) along with FPS to the chloroplasts (tpSSL1-3+tpFPS;  Table I), which is about a 70-to 90fold increase over the level of botryococcene (6.3 mg g 21 fresh weight [maximum] and 3.5 mg g 21 fresh weight [average]) accumulated in the lines with the same enzymes targeted to the cytoplasm (SSL1-3+FPS; Fig. 2B). These results indicate that the chimeric SSL1-3 enzyme efficiently uses FPP as a substrate derived from the universal C5 precursors present in the chloroplast compartment, supported by the accompanying engineered FPS. In contrast, the failure to enhance botryococcene yield by cytosolic engineering might be due to the cytosolic FPP pool being low and highly regulated, even if the avian FPS is used to override potential regulatory mechanisms in the cytoplasm (Wu et al., 2006(Wu et al., , 2012. The overall production of botryococcene by plastid targeting of this metabolism, and its increase over that achieved by cytosolic engineering, coincided well with what was found earlier for engineering squalene biosynthesis by Wu et al. (2012).
A similar contrast was also found in the comparison of production by engineering the membrane-tethered version of botryococcene synthase in the chloroplasts (tpSSL1-3M) with that directing the same metabolism to the cytoplasm (SSL1-3M). A relatively high amount of botryococcene accumulation was achieved by plastidic engineering, with a maximum level of 222 mg g 21 fresh weight and an average level of 131 mg g 21 fresh weight, which is about a 10-to 20-fold increase over that for cytosolic engineering, with a maximum of 16.4 mg g 21 fresh weight and an average of 5.8 mg g 21 fresh weight (Table I; Fig. 2). As with the soluble form of SSL1-3, the low production by cytosolic engineering of SSL1-3M suggests a limited flux of carbon and/or restrictive regulation occurring in the cytoplasm but absent in the chloroplast. The membrane targeting of SSL1-3M to internal membranes like the ER may enhance access to more readily available substrates in the cytoplasm, which might account for why the cytosolic engineered lines of SSL1-3M accumulated a higher level of botryococcene than was achieved by lines engineered for cytosolic, functionally soluble SSL1-3. In contrast, plastid-targeted SSL1-3M yielded only half the level of botryococcene as the plastid, soluble SSL1-3 form. Why this differential response of the soluble and membrane forms of the SSL1-3 enzyme in the chloroplast and cytoplasm exists is unknown, but certainly, the stroma compartment is physically distinct from the cytoplasm, as are the thylakoid membranes versus the ER membranes.

Development-Dependent and Tissue-Specific Accumulation
As expected, botryococcene accumulation demonstrated a significant developmental dependence. The level of botryococcene accumulating in mature leaves was 2-to 4-fold higher than in young leaves (Table I; Fig. 2). This makes sense biochemically, because the engineered botryococcene synthase and FPS were expressed constitutively and, thus, had more time to biosynthesize and accumulate botryococcene over the developmental time course of leaf maturation. In addition, there is no known mechanism in plants or any other organisms for the catabolism of botryococcene. Hence, accumulation should primarily reflect biosynthesis. Table I. Chemical assessment of T0 transgenic lines for their botryococcene contents Wild-type tobacco (accession 1068) was transformed with each indicated construct, consisting of a chimeric botryococcene synthase gene (SSL1-3 or SSL1-3M) and the avian FPS gene inserted downstream of constitutive promoters (cassava vein mosaic viral promoter [Pcv] and cauliflower mosaic viral promoter [Pca], respectively) or enhanced trichome-specific promoters (two 35S enhancers fused to cembratrienol synthase and hydroxylase promoters, respectively; Ennajdaoui et al., 2010;Wang et al., 2002a). More than 20 independent lines for each indicated construct were generated and grown under greenhouse conditions. The first fully expanded leaf from each plant was sampled for botryococcene content after 1 month (young) and 6 months (mature). Botryococcene levels were analyzed by gas chromatography-flame ionization detection, and the average (ave), as well as the minimum (min) and maximum (max), are reported. The number of transgenic lines under the direction of constitutive promoters having crinkle leaf and dwarf phenotypes (Fig. 8), and those under the direction of the trichome-specific promoters scored as chlorotic and dwarf (Fig. 4), are noted. To determine if the various transgenic plants accumulated triterpenes in other tissues besides leaves, the triterpene chemical profiles across various tissues were determined for select transgenic lines (Fig. 3). The squalene accumulating lines selected for comparison were homozygous T2 generation plants, while the lines selected for botryococcene accumulation were T1 heterozygous lines that had been selected for transgene presence under antibiotic pressure. The squaleneaccumulating lines were previously described by Wu et al. (2012) and designated as tpSS+tpFPS #5. The botryococcene lines were tpSSL1-3M+tpFPS #31 and tpSSL1-3+tpFPS #10. Triterpene content (either squalene or botryococcene) was found in all tissues examined, but the levels varied dramatically. Leaf accumulation was greatest and up to 64-fold more than that found in roots. Low amounts of triterpenes, never exceeding 25 mg g 21 , were also observed in other tissues, like veins and stems.

Trichome-Specific Expression of Botryococcene Metabolism
Trichomes are specialized organs located on the surface of the aerial parts of plant species, which can be the site of abundant specialized metabolite biosynthesis, accumulation, and secretion. In tobacco, for instance, up to 15% of the leaf dry weight has been attributed to the secretion of leaf exudate from trichomes (Wagner et al., 2004). Such a large contribution to leaf biomass relative to the actual volume of the glandular trichomes makes trichome engineering an attractive target for metabolite bioengineering (Ennajdaoui et al., 2010). Hence, in an effort to direct botryococcene biosynthesis to secretory trichomes, the trichome-specific promoters of the cembratrienol synthase and cembratrienol hydroxylase genes (Wang et al., 2002a;Ennajdaoui et al., 2010) were used to direct botryococcene synthase and FPS expression, respectively, to the secretory trichomes of tobacco. To strengthen the overall trichome-specific expression, the 35S double enhancer element was also appended to the 59 end of each of the trichome promoters (Wu et al., 2012). Four constructs harboring SSL1-3 or SSL1-3M with FPS, plus or minus chloroplast-targeting N-terminal sequences (tp), were thus introduced into the Nicotiana tabacum accession 1068, an accession documented to have high trichome density (Nielsen, 1982). Almost 30 independent transgenic lines were generated for each construct, the resulting transgenic lines were propagated in greenhouse facilities, and differently sized leaves were analyzed for their botryococcene content (Table I).
Targeting botryococcene metabolism to the chloroplasts of trichomes resulted in only a modest accumulation of botryococcene in young and maturing leaf tissue (less than 30 mg g 21 fresh weight on average), which was 10-to 20-fold greater than that accumulated in the lines with the enzymes directed to the cytoplasm of trichome cells (Table I). However, more unexpectedly, plants with putative enhanced, trichome-specific expression of SSL1-3 and FPS targeted to the chloroplast showed a strong chlorotic, white, mottling, and dwarf phenotype (Fig. 4), which certainly contributed to difficulties in propagating these materials. This phenotype was also much more severe than anything observed with the trichome-specific expression of  squalene biosynthesis (Wu et al., 2012). These adverse phenotypes were surprising, because we had hoped to engineer botryococcene metabolism away from tissues important for normal growth and development. Instead, we speculate that the trichome-specific promoters in combination with the double 35S enhancers may not behave in the anticipated manner (i.e. are not trichome specific) and exhibit ectopic expression (higher than that of the other constitutive promoters used here) in tissues that are more sensitive to this type of metabolism (e.g. meristems), resulting in the deleterious phenotypes described.

Engineering Triterpene Methyltransferases into Particular Subcellular Compartments of Tobacco Plants
The success in engineering squalene and botryococcene C30 production in transgenic tobacco led us to take advantage of these high-triterpene-accumulating lines for possible triterpene methylation. Our working hypothesis was that if we introduced triterpene methyltransferases into these lines, the accumulating triterpene (C30) could be converted to their methylated forms (C31 and C32) if the methyltransferases were targeted to where the triterpenes were synthesized and accumulated, assuming the methyl donor substrate SAM was available in sufficient quantities for the methyltransferase activity. We also wished to evaluate the substrate specificity of the TMTs for squalene and botryococcene as was done in yeast by Niehaus et al. (2012). Due to the hydrophobic regions present in the TMTs (which may function as transmembrane domains), it was equally important to evaluate if these enzymes could be expressed and function in the chloroplast and cytoplasm compartments.
To address these questions, all three of the TMT genes were individually constructed with a strong constitutive promoter (cauliflower mosaic virus 35S promoter [Pca]), plus or minus a plastid targeting signal sequence (tp), and engineered separately into squaleneand botryococcene-accumulating lines ( Fig. 5; Table II). Many independent transformants for each of the parental lines engineered previously for squalene biosynthesis and accumulation in the plastid compartment (a T2 homozygous line, line #5, tpSQS+tpFPS) or botryococcene accumulation in the plastid compartment (T1 generation heterozygous for tpSSL1-3+tpFPS, line #10, or T1 generation heterozygous for tpSSL1-3M+tpFPS, line #31) were subsequently transformed with each of the methyltransferases targeted to the cytoplasm (TMT-1, TMT-2, and TMT-3) or plastid compartment (tpTMT-1, tpTMT-2, and tpTMT-3), and the resulting transformants were screened for their triterpene chemical content and composition. More specifically, each of the transgenic lines was evaluated for triterpene (C30), monomethylated (C31), and dimethylated (C32) triterpene contents by GC-MS (Supplemental Figs. S1 and S2). Because of the large number of transgenic lines generated and evaluated, Table II summarizes the data for all the various lines, and Figure 5 illustrates the design strategies and provides analysis for three example lines for each TMT construct.
When the transgenic TMT-1 enzyme was targeted to the chloroplasts of the squalene-accumulating line, the maximum level of squalene methylation was 91%, with the average across all the lines being 65% (Table II). Of the methylated forms, approximately two-thirds was in the dimethylated form (Table II; Fig. 5). When TMT-2 was plastid targeted in the same parental background line, the maximum level of squalene methylation was 82% of the total, with an average of 51% for the 36 lines accumulating methylated squalenes (Table II). In contrast to the TMT-1 lines, only about one-third of the methylated squalene was in the dimethylated form in the TMT-2 plastid-targeted lines (Table II; Fig. 5).
When TMT-1 or TMT-2 was targeted to the cytoplasm rather than the chloroplasts, only 4% of the total squalene, on average, was methylated by either of the methyltransferases. However, 100% of the methylated squalene was in the dimethylated form (Table II; Fig. 5). The small amount of squalene available to the methyltransferases under these conditions could arise from cytosolic squalene synthesized by the native machinery or that synthesized by mistargeted engineered squalene synthase not properly transported into the chloroplast.
In contrast, only 5% (average) of the total squalene was methylated when TMT-3 was targeted to the chloroplast of the high-squalene-producing line, suggesting that TMT-3 exhibited weak catalytic activity toward squalene (Supplemental Table S1), which was also observed when substrate specificity for TMT-3 was investigated in yeast (Niehaus et al., 2012). Targeted expression of TMT-3 to the cytoplasm in the same line did not result in any methylation products, also corroborating the inability of TMT-3 to utilize the limited amounts of squalene found in this compartment (Supplemental Table S1).
When the TMT-3 enzyme was introduced into the botryococcene-accumulating lines, we observed a large proportion of methylated botryococcene only when TMT-3 was targeted to the chloroplasts (Table II; Fig. 5). Using parental line tpSSL1-3M+tpFPS, 87% of the botryococcene was maximally methylated, with 54% of the total botryococcene being methylated on average across all 18 lines evaluated. In comparison, maximally, 66% of the botryococcene was methylated in parental line tpSSL1-3+tpFPS, but more typically, 35% (average) was methylated (Table II). The apparent improved efficiency of botryococcene methylation in the tpSSL1-3M line versus the tpSSL1-3 line, however, might be more of a reflection on the total botryococcene levels rather than the efficiency of the methylation reaction itself. The line engineered with tpSSL1-3M accumulates about Table II. Chemical assessment of T0 transgenic lines targeting TMT activity to the chloroplast or the cytoplasm of high-squalene-accumulating (tpSQS) or botryococcene-accumulating (tpSSL1-3 and tpSSL1-3M) lines for their methylated triterpene contents More than 30 independent lines were generated for each transformation construct, consisting of one of the three TMT genes targeting triterpene methyltransferase activity to the chloroplast (with tp) or the cytoplasm (without tp) of the indicated parental lines. The squalene-accumulating line (tpSQS+tpFPS #5) or the botryococcene-accumulating lines (tpSSL1-3+tpFPS #10 and tpSSL1-3M+tpFPS #31) were transformed with the indicated TMT construct and evaluated after 4 months by GC-FID/GC-MS for methylated triterpenes. The number of transgenic lines accumulating methylated squalene or botryococcene was scored, along with the average (ave) percentage of nonmethylated (C30), monomethylated (C31), or dimethylated (C32) triterpenes relative to total triterpene (C30+C31+C32) content. The percentage of methylated triterpene (C31+C32) to total triterpene is denoted as conversion, and the number of plants with less than 10%, between 10% and 50%, and more than 50% conversion for each construct was counted accordingly. The highest amount of conversion as well as the average conversions are noted. The average total triterpene content for each line is also noted, as is the average triterpene content for three or four plants for each of the parental (control) lines.
half as much botryococcene as tpSSL1-3, and assuming similar expression levels of TMT-3 in both parental lines and comparable amounts of SAM availability, the greater percentage of methylated botryococcene in tpSSL1-3M may simply reflect the smaller pool of botryococcene available for secondary modifications. Little to none of the chloroplast-synthesized botryococcene was methylated when either TMT-1 or TMT-2 was targeted to the chloroplast (Supplemental Table  S1), further demonstrating the striking substrate preference of TMT-3 for botryococcenes and of TMT-1 and TMT-2 for squalene. By comparison, only a small proportion of methylated botryococcene was formed when TMT-3 was targeted to the cytoplasm. For parental line tpSSL1-3M+tpFPS, 14% maximal and on average only 6% of total botryococcene was methylated, whereas 10% maximal and on average only 3% of the botryococcene produced in parental line tpSSL1-3+tpFPS was methylated (Table II). As suggested above for squalene, the low level of methylated botryococcene produced by cytosolic TMT-3 could arise from either botryococcene (C30) produced by mistargeted SSL1-3(M), generating triterpene substrate in the cytoplasm, or leakage of botryococcene from the chloroplast-synthesized pool.
The observation of methylated squalene in the lines with squalene biosynthesis targeted to the chloroplast, yet the methyltransferases expressed in the cytoplasm, raised a question about what pool of squalene was being methylated. Was it the squalene synthesized in the chloroplast by the engineered squalene synthase, or could it be a reflection of the squalene synthesized by the native biosynthetic machinery operating in association with the ER? Because differentiating between pools of native versus engineered squalene in the tpSQS+tpFPS line has proven to be technically difficult, an alternative approach was sought. Gene constructs directing expression of the TMT-1 and TMT-2 enzymes targeted to the chloroplast and cytoplasm were introduced into control, wild-type tobacco, and the resulting transgenic lines were screened for methylated squalene. Interestingly, a significant proportion of methylated squalene (average 41% of total squalene) was observed when TMT-1 was targeted to the chloroplasts (Table III), where there is no evidence for squalene biosynthesis or accumulation (Aharoni et al., 2003). In this case, one plausible explanation is that some mistargeting of the engineered TMT-1 results in the methylation of cytoplasmic biosynthesized squalene. In contrast, when TMT-1 and TMT-2 expression were targeted to the cytoplasm of wild-type plants, a high proportion of methylated squalene (average 72% and 67% of total squalene, respectively) was found (Table III; Fig. 6). Even more surprising, the level of total squalene (C30+C31+C32) in transgenic lines expressing TMT-1 was elevated to a maximum of 55 mg g 21 fresh weight and an average of 36 mg g 21 fresh weight (Table  III). This was about 4-to 7-fold greater than the level of endogenous squalene (C30) accumulating in wild-type plants (Table III; Fig. 6). Equally important, the level of nonmethylated squalene remained relatively constant in all these lines at 6 to 10 mg g 21 fresh weight, with all the additional triterpene accumulating as monomethylated and dimethylated squalene.

Developmental Triterpene Accumulation and Methylation
To explore the possible influence of development processes on the methylation status of squalene and botryococcene, three independent lines for each expression vector combination (squalene biosynthesis targeted to the plastid compartment with plastidtargeted TMT-1 or TMT-2, and botryococcene biosynthesis plus TMT-3 directed to the plastid compartment) were grown in the greenhouse for approximately 3 months, then leaves at four developmental positions on the plants were profiled for their triterpene levels and Table III. Chemical assessment of T0 transgenic lines targeting TMT activity to the chloroplast or the cytoplasm of wild-type plants for their methylated squalene contents Independent lines were generated for each of the indicated constructs consisting of either TMT-1 or TMT-2 targeted to the chloroplast (with tp) or the cytoplasm (without tp) of wild-type plants. T0 antibiotic-selected transgenic plants were propagated in the greenhouse for up to 5 months before their triterpene and methylated triterpene contents was determined by GC-MS. The number of transgenic plants accumulating methylated squalene were scored, and their average (ave) percentage of nonmethylated (C30), monomethylated (C31), or dimethylated (C32) squalene was determined relative to the total squalene (C30+C31+C32). The percentage of methylated squalene (C31+C32) to total triterpene is denoted as conversion, and the number of plants with less than 10%, between 10% and 50%, and more than 50% conversion for each construct was counted accordingly. The highest amount of conversion as well as the average conversions are noted. The average total squalene content for each line is also noted, as is the average squalene content for three of the wild-type control plants.  Figure 7 represent the average determinations for leaves from three independent transformants for each engineered combination. As expected, the total triterpene levels showed a successive increase with leaf maturation. The more mature leaves generally had more total triterpene accumulation (Fig. 7, left axis). However, the average ratio of methylated (C31+C32) to total triterpene (C30+C31+C32) at the various leaf positions remained essentially the same, from 55% to 75%, for all the transgenic lines (Fig. 7B, right axis).

Phenotypes of Triterpene-Accumulating Plants
Over 75% of the botryococcene-accumulating lines directing this metabolism to the chloroplast exhibited moderate to distinguishing phenotypes, including dwarfing, chlorosis, and mottling (Table I; Fig. 8, A-C). Figure 6. Methylated triterpene contents in independent transgenic lines targeting TMT activity to the chloroplast or cytoplasm of wildtype plants. A, Conceptual strategies for how endogenous squalene biosynthesized by wild-type plants in the cytoplasm might be methylated by methyltransferases (TMT-1 and TMT-2) directed to the chloroplast or cytoplasm. B and C, Gene constructs harboring TMT genes targeting the triterpene methyltransferase to the chloroplast (with tp) or the cytoplasm (without tp) were transformed into wild type plants (B), and the antibiotic-selected T0 lines propagated in the greenhouse (C) were assessed for their triterpene contents (4 months old) by GC-MS. D, The levels of squalene (C30; green) and methylated squalenes (C31, blue; C32, orange) accumulating in three independent lines for each construct were determined in the first fully mature leaf of each line by GC-MS. FW, Fresh weight. Figure 7. Accumulation of total triterpene content and the fraction of triterpene converted to methylated triterpenes over a time course of leaf development. Transgenic lines directing squalene (tpSQS+tpFPS) biosynthesis and squalene methylation by TMT-1 or TMT-2 to the chloroplast compartment, or directing botryococcene (tpSSL1-3M+tpFPS) biosynthesis and botryococcene methylation by TMT-3 to the chloroplast compartment, were grown under greenhouse conditions for 4 months. Leaves at the noted positions relative to the apex of the plant were collected, and their triterpene and methylated triterpene contents were determined by GC-MS. Absolute levels of total triterpenes (C30, C31, and C32) are reported in the histogram (blue), while the fraction of triterpene converted to methylated forms are denoted by the scatterplot (red). FW, Fresh weight.
These phenotypes were not observed in any transgenic lines in which the engineered metabolism was targeted to the cytoplasm (Table I) and was obviously different from what was observed in any of squaleneaccumulating plants (Fig. 8E, left;Wu et al., 2012). The visual observations thus indicated that botryococcene accumulation had some undefined effects on chloroplast development, plant morphology, and growth, while squalene accumulation did not. Moreover, there were not any noticeable differences in phenotypes between triterpene-accumulating plants and their respective methylated triterpene counterpart plants (Fig. 8, D and E); hence, the methylation of botryococcenes was not able to restore the wild-type phenotype.

DISCUSSION
This work successfully transplanted several steps of triterpene metabolism occurring in the alga B. braunii race B into tobacco plants, leading to a high-level accumulation of botryococcene and methylated triterpenes. The most robust triterpene accumulation relied upon the strategy of diverting the five-carbon precursors IPP and DMAPP from the MEP pathway operating in the chloroplast to triterpene (C30) biosynthesis by the coexpression of an FPS plus a triterpene synthase. The accumulating triterpenes (C30) in the transgenic plants could be further modified by targeting yet another enzyme activity, TMT, to the chloroplasts of these transgenic plants. Hence, engineering novel expression of the enzymes FPS, triterpene synthases, and TMTs created a new metabolic channel redirecting carbon flux from the MEP pathway to the biosynthesis and accumulation of unique and unusual triterpenes (Figs. 1, 2A, and 5A).
The strategy was successful in taking advantage of engineering terpene metabolism in the plant chloroplasts. First, chloroplasts offer an abundance of carbon passing through the MEP pathway, and diverting an intermediate and carbon flux from this pathway does not adversely impact the biosynthetic needs in the chloroplasts for large amounts of carotenoids and chlorophylls. The second, equally important, observation is that the chloroplast provides an ideal environment for heterologous terpene production, perhaps due to the endogenous regulation of the MEP pathway in plastids, allowing for a new carbon sink in the form of the introduced triterpenes, as compared with the MVA pathway operating in the cytosol preventing the high IPP/DMAPP flux into these compounds (Kempinski et al., 2015).
This approach has now been demonstrated to be applicable for the metabolic engineering of various types of terpene compounds, including monoterpenes, sesquiterpenes, and triterpenes in tobacco plants. However, we also note that the accumulation level of each type of terpene differed substantially between the respective terpene targets. Plants engineered for triterpene production accumulated 200 to 1,000 mg g 21 fresh weight of triterpene, whereas sesquiterpene production has not exceeded 30 mg g 21 fresh weight and monoterpene accumulation is maximally around 1 mg g 21 fresh weight (Wu et al., 2006;Kempinski et al., 2015). Such stark differences strongly suggest that the limitation in specific terpene class accumulation lies with Figure 8. Phenotypes of transgenic lines expressing botryococcene and squalene biosynthetic genes. Some transgenic lines targeting botryococcene biosynthesis to the chloroplast (tpSSL1-3+tpFPS), and hence accumulating high levels of botryococcene, are dwarfed (A, left plant; right, wild-type plant) and exhibit chlorotic, mottled, and wrinkled leaf morphologies (B and C). High methylated botryococene accumulating lines (D, right plant) generated by engineering tpTMT-3 engineered into a line (tpSSL1-3+tpFPS) accumulating only non-methylated botryococcene (D, left plant) did not result in a wild type phenotype (E, right plant). In contrast, plants targeting squalene biosynthesis to the chloroplast (tpSQS+tpFPS) and accumulating high levels of squalene (E, left plant) did not exhibit such phenotypes, nor when triterpene methyltransferases (TMT-1 or TMT-2) were also targeted to the chloroplast of squalene accumulating line (E, middle plant) compared to the wild type (E, right plant). the engineered terpene synthase or that certain introduced terpene compounds may have differing effects on physiological homeostasis and growth. Consistent with this notion, we found that overexpression and chloroplast targeting of the soluble form of SSL1-3 with FPS yielded similar levels of botryococcene accumulation to that for squalene, which was achieved by plastid-targeted engineering of a yeast, soluble squalene synthase along with the avian FPS. However, two times more botryococcene was observed with SSL1-3 than the SSL1-3M enzyme form with FPS targeted to the chloroplast. This suggests that the chimeric enzyme SSL1-3 functions as well as the single yeast squalene synthase enzyme targeted to the chloroplast and exhibits a higher catalytic capacity than SSL1-3M could.
TMTs are functionally insoluble enzymes that exhibit an unexpectedly high catalytic activity for the methylation reaction when engineered into both the chloroplast and cytoplasm compartments of the appropriate transgenic plant lines. Up to 91% of the C30 triterpenes accumulating in the high-yielding lines was subsequently transformed to monomethylated or dimethylated triterpene when one of the three TMT genes targeted methyltransferase activity to the chloroplast. The methylation ratio of 51% to 91% by TMTs directed to the plastid compartment versus 3% to 14% by TMTs targeted to the cytosol provides additional evidence that the distribution of triterpene C30 in the hightriterpene-accumulating transgenic lines remained in the chloroplast. This was not unexpected, because the C30 triterpenes are supposedly synthesized in the chloroplast and methylation in the cytosol would require some mechanism, either active or passive, to export the novel triterpene out of the chloroplasts to the cytoplasm.
Therefore, in order to account for the small but significant methylation of triterpenes occurring in the cytoplasm, at least four possible routes remain plausible. First, the methylated squalene produced by targeting TMT-1 and TMT-2 to the cytoplasm in the wild-type plants proves that natively synthesized squalene can be methylated by TMTs. Second, the small amount of methylated botryococcene generated in plants in which TMT-3 was directed to the cytoplasm while high botryococcene biosynthesis was directed to the chloroplasts [tpSSL1-3(M)+tpFPS] could arise from a low level of botryococcene (C30) biosynthesized by mistargeted SSL1-3(M) ( Table II). This notion implies that TMTs can methylate cytosolic triterpene (C30) produced by mistargeted triterpene synthase as well as by the native triterpene machinery. Third, expressing the tpTMT-1 in wild-type plants also resulted in methylated products, which must be derived from cytosolic endogenous squalene catalyzed by mistargeted TMT-1. This evidence, not surprisingly, suggests that our chloroplast-targeting strategy is not 100% effective and supports our contention that mistargeted TMTs are also able to methylate cytosol-localized triterpenes. Fourth, the cytosolic engineered TMTs may have a way to access the plastid-localized squalene. The recent discovery that plastid envelope-localized substrates can be accessed by enzymes targeted to the ER membrane through a continuity of ER and chloroplast (Mehrshahi et al., 2013) offers one possible explanation. Of course, the methylation status of triterpenes could come about by some combination of routes, which might also be variable upon plant development and growth habit.
An issue raised during the initial phases of this work was whether there would be sufficient SAM to support the formation of the methylated triterpenes. This concern arose because of an appreciation for how important SAM is to the methylation of macromolecules as well as very diverse small molecules (Bouvier et al., 2006;Sauter et al., 2013) and its known biosynthesis in the cytoplasm (Ravanel et al., 1998(Ravanel et al., , 2004. Fortunately, concern for SAM availability seemed unfounded regardless of whether the methylation reactions were targeted to the chloroplasts or the cytoplasm. Although there was reduced TMT efficiency in the cytoplasm, this is most likely due to the reduced amount of triterpene available for methylation and, thus, reflects TMT efficiency and not SAM availability. Equally interesting was the observation that plants engineered for botryococcene accumulation tended to exhibit distinct phenotypic outcomes like dwarfism, chlorosis, and mottling, while plants accumulating high levels of squalene did not show any of these adverse effects. Why this might be so is currently unknown. However, if one could discern how the plants were able to accumulate high levels of squalene without any negative impact on growth performance, then one might be able to use this information in the engineering of advanced accumulation mechanisms for terpenes like botryococcene. One suggestion worth examining is how botryococcenes versus squalene might differentially interdigitate into membranes and disrupt normal biochemical functions. Hence, engineering alternative means for sequestering these molecules could alleviate physiological consequences and improve overall accumulation (e.g. engineering in lipid droplet-forming proteins).
Finally, squalene biosynthesis is known to be a key committed step in sterol biosynthesis, and squalene has been suspected of serving a regulatory role (Wu et al., 2012). The results presented here, where introducing cytosolic forms of the TMT enzymes elevated overall squalene and methylated squalene levels 4-to 7-fold higher than normal, directly address this issue. By diverting squalene to its methylated forms, some innate mechanism had to be evoked to allow for additional squalene production to occur. This, we suggest, is providing an important glimpse into the regulatory complexity of squalene biosynthesis, which is crucial for the homoeostatic control of sterol biosynthesis in the plants.

Expression Vector Construction and Plant Transformation
The design of gene constructs and assembly for engineering botryococcene biosynthesis were based on the work described previously by Wu et al. (2006Wu et al. ( , 2012 using standard molecular methodologies. Gene constructs consisted of a peptide fusion of SSL-1 (GenBank accession no. HQ585058.1) and SSL-3 (GenBank accession no. HQ585060.1) connected by a triplet repeat peptide linker of Gly-Gly-Ser-Gly, with or without appending the C-terminal end (71 amino acids) of the Botryococcus braunii squalene synthase (GenBank accession no. AF205791.1) onto the C terminus of SSL-3 and the FPS gene (P08836; Tarshis et al., 1994). The chimeric SSL1-3 genes and FPS gene were inserted downstream of the strong constitutive promoters Pcv (Verdaguer et al., 1996) and Pca (Benfey et al., 1990), respectively. For trichome-specific expression of triterpene biosynthesis, the trichome-specific promoters Pcbt (Ennajdaoui et al., 2010) and Pcyp16 (Wang et al., 2002a) were fused to 59 end of botryococcene synthase genes and the FPS gene, respectively. The duplicated cauliflower mosaic virus 35S enhancer elements (Benfey et al., 1990) were fused to the 59 end of each trichome promoter. A chloroplast-targeting signal sequence encoding for the first 58 amino acids of the Arabidopsis (Arabidopsis thaliana) Rubisco small subunit gene (NM23202; Lee et al., 2006) was fused in frame with the 59 end of the respective terpene synthase genes. The gene cassettes were assembled together in a helper vector described by Wu et al. (2012) by standard molecular biology methods, and the various DNA segments were verified by DNA sequencing. The gene cassettes were then introduced into pBDON, a modified Ti plasmid vector harboring a hygromycin resistance gene by DNA recombination (Wu et al., 2006).
The triterpene methyltransferase genes TMT-1 (JN828962.1), TMT-2 (JN828963.1), and TMT-3 (JN828964.1) were inserted directly into the plant transformation vector pKYLx71 (Schardl et al., 1987) harboring a 35S viral promoter and a kanamycin resistance gene. In order to target TMT genes to the chloroplast, the chloroplast-targeting signal sequence noted above was then inserted in frame with the 59 termini of the respective TMT genes.
The engineered Ti plasmid vectors were introduced into Agrobacterium tumefaciens GV3850 by electroporation, and the resulting A. tumefaciens lines were used to genetically engineer tobacco (Nicotiana tabacum) T1 accession 1068 (Nielsen, 1982), or transgenic line tpSQS+tpFPS #5 (T2 homozygous generation) with a high level of squalene, as described previously by Wu et al. (2012), or the high-botryococcene-accumulating transgenic lines (tpSSL1-3+tpFPS-10 or tpSSL1-3M+tpFPS-31; T1 heterozygous generation) generated in this study. Leaf explants were transformed with the respective gene constructs, and the resulting calli were selected on tissue culture medium with hygromycin (50 mg L 21 ) for engineering botryococcene biosynthesis and with both hygromycin (50 mg L 21 ) and kanamycin (250 mg L 21 ) for engineering methylated triterpene biosynthesis. The culture medium (1 L) contained 4.2 g of Murashige and Skoog salts (Phytotechnology Laboratories), 0.112 g of B5 vitamins (Phytotechnology Laboratories), 30 g of sucrose, 9 g of agar, 1 mg of indole-3-acetic acid, and 2.5 mg of benzylaminopurine (Sigma). The selected calli were grown under sterile tissue culture conditions to regenerate plantlets. The selected T0 plantlets were then propagated in the greenhouse and assessed for triterpene content by GC-MS or GC-FID analysis.

Plant Propagation and Segregation Selection
All the T0 plantlets after hygromycin or kanamycin selection were grown in common commercial vermiculite/soil blends in a greenhouse and fertilized weekly with water-soluble fertilizer (20-20-20 for nitrogen, phosphorus, and potassium). Insect control was performed as needed. The T0 plants were allowed to flower in the greenhouse, and the T1 seed was collected for subsequent cycles of propagation. Segregation of the hygromycin and kanamycin resistance trait in the T1 seed lines was also evaluated by germinating sterilized seeds on 50 mg L 21 hygromycin and 250 mg L 21 kanamycin in T-tissue culture medium (4.2 g of Murashige and Skoog salts, 0.112 g of B5 vitamins, 30 g of Suc, and 9 g of agar in 1 L of medium).

Triterpene (Squalene, Botryococcene, Methylated Squalene, and Methylated Botryococcene) Determinations
Fifty to 150 mg of transgenic leaf material was collected from the uppermost, fully expanded leaves of tobacco plants grown in a greenhouse. The other plant tissues (roots, stems, and veins) were collected from plants grown in tissue culture for chemical analysis. The terpene content for each sample was determined by the methods described previously by Wu et al. (2012). Each plant sample was ground in liquid nitrogen, then extracted with 2 to 4 mL of a hexane:ethyl acetate (85:15, v/v) mixture containing 200 ng of a-cedrene as an external standard for quantification and calculations of recovery. The extracts were concentrated to 500 mL under a nitrogen stream without drying the sample. The concentrated extracts were then partially purified by passing through a silica column (500 mg, prepared in a glass wool-plugged glass pipette) and further eluted with 1 mL of the hexane solvent. After concentration of the combined eluate under a stream of nitrogen, aliquots were injected onto a GC-MS device equipped with an HP5-MS capillary column (30 m 3 0.32 mm, 0.25-mm phase thickness) with a temperature program of 70°C for 1 min followed by a 4°C min 21 gradient to 250°C. Mass spectra were recorded at 70 eV, scanning from 35 to 500 atomic mass units, and experimental samples were compared with standards that were used previously in earlier studies (Niehaus et al., 2011(Niehaus et al., , 2012Wu et al., 2012) for verification.
The structure of purified botryococcene from tobacco was determined by 1 H-NMR and 13 C-NMR spectral analyses, which were also described in an earlier study (Wu et al., 2012). Botryococcene was extracted from leaf material of a transgenic line (tpSSL1-3+tpFPS #10) targeting the chimeric botryococcene synthase SSL1-3 and FPS to the plastid compartment under the direction of the constitutive promoters.
One hundred grams of leaf material was ground in liquid nitrogen and then extracted with 1.2 L of hexane:ethyl acetate (85:15, v/v); the extract was concentrated to 5 mL and fractionated on a silica column with 5-mL aliquots of hexane as the eluting solvent. Fractions were monitored by GC-MS for the desired triterpene compound. Enriched fractions were pooled and concentrated under nitrogen, and the entire sample was processed by silica HPLC using hexane as the eluting solvent (Niehaus et al., 2012;Wu et al., 2012). Alternatively, the crude extract was resuspended in hexane and fractionated via silica gel chromatography, with a final purification step provided by HPLC. Recovery of a 6-mg purified botryococcene sample with a 50% yield was obtained. 1 H-NMR and 13 C-NMR spectra were recorded on a 400-MHz Varian J-NMR spectrometer at 300 K, and chemical shifts were referenced relative to solvent peaks, namely dH 7.24 and dC 77 for CDCl 3 .

Supplemental Data
The following supplemental materials are available.
Supplemental Figure S1. Triterpene and methylated triterpene contents determined in leaf extracts from elite transgenic plants by GC-MS.
Supplemental Table S1. Screen of T0 transgenic lines targeting select TMT activities to the chloroplast or the cytoplasm of high squalene (tpSQS+tpFPS) and botryococcene (tpSSL1-3+tpFPS) accumulating lines for their methylated triterpene content.