Building a biofactory: Constructing glandular trichomes in Cannabis sativa

Introduction

Like other plants, Cannabis sativa L. (cannabis) has evolved many ways to defend themselves from herbivores, including the production of specialized metabolites [1]. The bracts and calyces of mature cannabis pistillate flowers have multiple glandular trichome types, such as stalked or sessile ‘capitate’ glandular trichomes, or the small bulbous glandular trichomes [2]. The most abundant and important for metabolite production are stalked glandular trichomes, which are composed of a disc of secretory cells topped with a storage cavity full of resin in the subcuticular apoplast [3]. Cannabis has somewhat unusual stalked glandular trichomes (Figure 1(a)): they have features of both capitate glandular trichomes such as secretory cells borne on a long stalk (as seen in the Solanaceae, e.g. Ref. [4]), as well as stereotypical peltate glandular trichome features, such as a disc of secretory cells topped with the subcuticular storage cavity (typical of the Lamiaceae [5]).

1. Download : Download high-res image (1MB)
2. Download : Download full-size image

Figure 1. Schematic model of cannabis stalked glandular trichome development. (a) A mature glandular trichome showing the extracellular storage cavity and secretory cells on the large multicellular stalk. (b) The developmental trajectory of stalked glandular trichome, from the protodermal initial (left) through a sessile-like intermediate with small metabolite droplets to the mature and over-mature trichomes (right). (c) The presecretory disc cells, interconnected by multiple cytoplasmic bridges, are differentiated for metabolite production. The cells are dominated by abundant ER and unusual nonphotosynthetic plastids with crystalline lattice-like cores. Formation of the storage cavity will occur through delamination of the cell wall. (d) Subcellular localization of the cannabinoid biosynthetic pathway with GPP production in plastids, prenyltransferases associated with plastid, and abundant ER contact sites linking plastid to PM. Putative ABC transporters could export CBGA into the wall, where THCAS catalyzes the final step of THCA biosynthesis. Metabolites interact with specialized cell wall components like arabinogalactans to form small metabolite droplets that coalesce into a large central droplet in the mature trichome. Abbreviations: ABC = adenosine triphosphate–binding cassette; CBGA = cannabigerolic acid; GPP = geranyl diphosphate; ER = endoplasmic reticulum; PM = plasma membrane;THCA = tetrahydrocannabinolic acid; THCAS = tetrahydrocannabinolic acid synthase.

Female plants grown commercially for medicinal and recreational cannabis production form highly branched inflorescences, with densely packed repeating floral modules of a reduced leaf, bract, and perigonal bract/calyx surrounding the style and ovary [6,7]. Glandular trichome density on bracts increases from weeks 3–8 of the flowering period accompanied by increased head and stalk size, as documented by light and cryo-scanning electron microscopy [7,8]. This floral architecture and high density of stalked glandular trichomes at maturity make this plant a metabolic powerhouse.

Cannabis glandular trichomes have often been referred to as ‘cell factories’ [9,10] due to their intense production of cannabis resin, a high-value mixture of specialized metabolites including cannabinoids and terpenes. Our understanding of the cannabis trichome morphology was shaped by the groundbreaking work of Mahlberg et al. (reviewed in Ref. [2]). Since then, there have been advances in the molecular basis of cannabinoid biosynthesis (reviewed by Gülck and Møller [11]), the diverse terpenoid biochemistry of cannabis (reviewed by Lange and Srividya [12]), and identification of the ‘skunky’ volatile sulfur compounds [13]. In this review, we adopt the biofactory analogy to frame new cell biology findings in the context of recent studies of cannabis biochemistry, transcriptomics, and proteomics.

Site identification – formation and differentiation of epidermal outgrowths

The first step in cannabis glandular trichome development is the differentiation of a protodermal cell to trigger a series of cell divisions resulting in an epidermal outgrowth (Figure 1(b)) [2]. The exact regulatory mechanisms that trigger the developmental trajectory of protodermal cells in cannabis is unknown; however, mechanisms involved in glandular trichome initiation in other species, like Solanum lycopersicum and Artemisia annua, may provide useful comparisons [14]. Multiple lines of evidence implicate transcription factors (TFs) of the HD-ZIP subfamily IV [15] and the R2R3-MYB family, especially members of the MIXTA subfamily [16,17], in initiation of glandular trichomes in these systems. Genome-wide analysis identified nine HD-ZIP IV TFs in cannabis, and expression data implicated CsHDG5 as a potential initiator of trichome development [18]. This is further supported by the similarity of CsHDG5 to AaHD1 [18], the HD-ZIP IV TF required for trichome initiation in A. annua that acts downstream of the AaMIXTA-AaHD8 complex [17,19]. In cannabis, 99 R2R3-MYB proteins have been identified, with several showing preferential expression in glandular trichomes [20]. However, only one of the cannabis R2R3-MYB TFs has been studied for a potential role in trichome initiation: CsMYB106/CsMIXTA1 (designated as a MIXTA-like TF based on phylogenetic analysis) [21]. In this study, transient expression of CsMIXTA1 in Nicotiana tabacum resulted in increased glandular trichome density and size, suggesting that CsMIXTA1 could act as a positive regulator of trichome development [21]. The missing element in our understanding of cannabis trichome initiation is our lack of information on how individual cannabis TFs interact in functional complexes to trigger the diverse cell fates.

Following glandular trichome initiation, protodermal cells differentiate into secretory, abscission, or stalk cells (Figure 1(a)). The secretory disc cells form an interconnected syncytium, and these are the cells responsible for cannabinoid and terpene biosynthesis, transport, and storage (Figure 1(c)) [22]. The trichome heads are easily dislodged at the top of the stalk, where the stalk forms a narrow constriction with a cluster composed of two to four specialized abscission cells [8,22,23]. These abscission cells contain plastids with unusual thylakoid membranes, and they form an interface with the secretory disc but are symplastically isolated [22]. Supporting the secretory and abscission cells are the stalk epidermal and hypodermal cells, which are continuous with the underlying mesophyll cell layer [3,22].

Two-photon microscopy of the intrinsic fluorescence in cannabis flowers revealed that while a subpopulation of the developing glandular trichomes terminally differentiates into canonical peltate sessile trichomes, the majority of developing glandular trichomes differentiate into stalked trichomes with a peltate-like head [24]. As they develop, they gain a higher ratio of monoterpenes to sesquiterpenes and accumulate more cannabinoids in a large lipid droplet in the storage cavity (Figure 1(b)) [24].

Assembly line – a supercell equipped with specialized organelles for production

Cannabis glandular trichomes, like Mentha (mint) species’ trichomes, are non-photosynthetic and rely on import of photosynthates from the surrounding tissues to support specialized metabolism. In mint, raffinose oligosaccharides have been proposed as the carbon source imported into the trichome heads [25]. Cannabis proteomics studies detected raffinose biosynthetic enzymes in whole flowers, while the enzymes for raffinose breakdown were found in trichome heads [23]. Flux-balance analysis, using transcriptomic and metabolomic data sets, models the flow of carbon through primary metabolism into secondary metabolism in a variety of glandular trichomes [26]. Although cannabis was not among the species modeled [26], cannabis proteomic data experimentally support a mint-like model with reducing power flowing directly into secondary metabolism [23].

Cannabinoid biosynthesis in cannabis glandular trichomes requires both olivetolic acid (OA), from the OA pathway and geranyl diphosphate (GPP) [27]. While both the plastidial 2C-methyl-D-erythritol-4-phosphate (MEP) and the cytoplasmic mevalonate (MVA) pathways can contribute to the production of the C₅ isoprenyl precursors, GPP is predominantly derived from the MEP pathway, e.g. during monoterpene biosynthesis [28]. In cannabis trichome heads, MEP pathway genes have higher expression than MVA genes [23,28,29], suggesting that the GPP used to make cannabinoids is more likely to be derived from the MEP pathway. High-resolution electron microscopy demonstrates that the cannabis secretory disc nonphotosynthetic plastids, which are producing all this GPP, contain an unusual crystalline lattice-like core of membrane tubules, superficially resembling prolamellar bodies of etioplasts (Figure 1(d)) [2,22]. Unlike prolamellar bodies, the crystalline cores in these plastids persist in the presence of light. Prolamellar bodies’ membrane tubules are shaped by polymers of protochlorophyllide oxidoreductases, the light-sensitive enzymes that activate chlorophyll production [30]. The paracrystalline membrane structures in cannabis trichomes suggested the hypothesis that GPP synthase (GPPS) could form these cores; however, immunogold transmission electron microscopy (TEM) localization showed that there was no enrichment of GPPS in the paracrystalline cores over the surrounding stroma [22]. Thus, the specific composition of these unusual intraplastidial membranes remains unknown.

Unlike the plastidial MEP pathway, OA is synthesized in the cytoplasm [31] when hexanoyl-CoA is extended and cyclized by a tetraketide synthase (TKS) working together with OA cyclase [32]. Aromatic prenyltransferases (aPTs) catalyze the prenylation of the OA, thus forming the cannabinoid intermediate cannabigerolic acid (CBGA). Two highly expressed aPTs, CsaPT1 and CsaPT4, are the top candidates for CBGA synthase in cannabis trichomes [33]. CsaPT1 and CsaPT4 have slightly different substrate preferences [34] and gene expression patterns in the floral tissues [35]. Furthermore, plastid localization signal sequences and localization in heterologous systems suggest these aPTs are plastid-localized (Figure 1(d)) [33]. We speculate that the cytosolic production of OA and the plastid localization of aPTs points to the formation of CBGA at the plastid outer envelope, and this would position CBGA to partition into nearby membranes such as the endoplasmic reticulum.

While the cannabinoid precursor CBGA is made inside the cell, there are several lines of evidence that indicate that the final step of tetrahydrocannabinolic acid (THCA) biosynthesis occurs outside secretory cells, in the storage cavity. Heterologous expression of THCA synthase tagged with green fluorescent protein (GFP) in N. tabacum shows GFP fluorescence in trichome heads [36]. Proteomics analysis of cannabis resin further narrows the localization of THCAS and cannabidiolic acid synthase (CBDAS) to the storage cavity of cannabis glandular trichomes [37]. Immunogold TEM, using an antibody specific to THCAS, localized THCAS on the surface cell walls of the secretory disc cells underlying the storage cavity but not to the anticlinal or basal cell walls (Figure 1(d)) [22].

One implication of the extracellular localization of these cannabinoid synthases is that the common precursor CBGA is the cannabinoid exported from the secretory disc to the storage cavity, where it can encounter a variety of cannabinoid synthases. Correlation analysis of transcriptomics and metabolomics across nine diverse strains of cannabis demonstrated a robust correlation between CBDA synthase gene expression and CBDA metabolite levels, with some high-CBDA plants expressing nonfunctional THCA synthase transcripts, and high-THCA plants expressing almost identical THCA synthase alleles [28]. In strains that accumulate both CBDA and THCA, functional versions of both CBDAS and THCAS are expressed [38]. A study of 104 genomes concluded that during cannabis domestication, there was loss of function of either THCAS or CBDAS genes, in hemp (CBDA-rich) or drug-type cannabis cultivars, respectively [39]. In addition to the allelic variation among genomes, three putative regulators of cannabinoid synthase transcriptional activity were identified as CsAP2L1, CsWRKY1, and CsMYB1 [40]. Regulation of synthase gene expression is developmentally necessary but is not the rate-limiting factor of cannabinoid production [41]. Instead, transcriptomic data suggest that regulation of genes involved in CBGA production (GPPS, PT, TKS, OAC) are the key drivers in overall cannabinoid content [41]. Thus, the flux of carbon into CBGA production could control the quantity, while which functional cannabinoid synthases are expressed would control cannabinoid composition.

Distribution – transportation of specialized metabolites

The lipidic components of cannabis resin must be trafficked from the plastids, across the plasma membrane, and through the cell wall for deposition in the storage cavity. Early hypotheses suggested that intracellular transport of metabolites occurred through the formation of intracellular inclusions [2]. However, ultrarapid cryofixation and freeze-substitution provide a clear demonstration that these inclusions were artifacts of the chemical fixation process [42] and instead revealed abundant endoplasmic reticulum (ER)–plastid and ER–plasma membrane (PM) contact sites (Figure 1(d)) [22]. In plant, yeast, and mammalian cells, numerous membrane contact sites between the ER and organelles can form a network that functions in lipid transfer [43]. The presence of a similar ER network in cannabis secretory cells suggests the hypothesis that specialized metabolites could move through membrane contact sites.

The mechanisms of cannabinoid and terpene transport across the PM are currently unknown. Adenosine triphosphate–binding cassette (ABC) transporters have been functionally characterized as exporters of specialized metabolites in a wide variety of plant species [44]. In Petunia hybrida, a half transporter, PhABCG1, has been characterized to facilitate transport of volatile organic compounds [45]. Two other full transporters of the subfamily G identified in N. tabacum and A. annua, NtPDR1 and AaPDR3, were also found to function in diterpene and sesquiterpene transport, respectively [46,47]. Transcriptomic and proteomic studies have detected high expression of ABC transporters in cannabis glandular trichomes suggesting potential functions in cannabinoid and terpene export (Figure 1(d)) [23,24]. A recent study utilizing phylogenetic and co-expression analysis with cannabinoid pathway genes points to one ABCB and six ABCG transporters as interesting candidates [48], however, functional characterization has yet to be performed on any candidate transporters.

Warehouse assembly – storage cavity construction by cell wall remodeling

Once exported from secretory cells, cannabinoids and terpenes are stored in an inflated apoplastic compartment on the top of the glandular trichome. As cannabinoids have been demonstrated to exert a cytotoxic effect on their own tissues [36], through a mechanism involving mitochondrial disruption [49], sequestration of these compounds in an extracellular space minimizes cannabinoid autotoxicity. Formation of the extracellular cavity occurs by delamination and separation of disc cell walls, resulting in a wall layer that remains contiguous with the plasma membrane (surface cell wall) and a second that remains contiguous with the cuticle (subcuticular layer) (Figure 1(d)) [2,50]. Cell wall directed antibody labeling demonstrated that the formation of the storage cavity in cannabis trichomes is accompanied by extensive cell wall remodeling, including decreased xyloglucan content [50]. Delamination of the subcuticular layer from the surface wall was marked by altered homogalacturonans, and as metabolites accumulated in the storage cavity, arabinan and arabinogalactan epitopes were detected surrounding the small droplets [50]. These specialized cell wall components disappeared in the mature trichome, where a single large monoterpene-rich droplet coalesced, filling the distended cavity [50].

The formation of the storage cavity is associated with cuticle thickening as metabolites accumulate [2,50]. Two non–mutually exclusive hypotheses can explain the cuticular thickening: first, that biosynthesis and deposition of cuticular lipids from the glandular cells continues throughout development; and second, that the observed thickening is the result of cannabis metabolites permeating the cuticle. One study examining cannabis glandular trichome phenotypes during maturation speculated that import of cuticle precursors causes the color change of glandular heads from clear to milky white [51]. Correlated light and cryo-scanning electron microscopy microscopy showed that once trichomes have reached maturity, turning an amber color, resin droplets are secreted on the surface of the glandular head, and the cuticle collapses inward (Figure 1(b)) [8]. Thickening of the cuticle during glandular trichome development suggests a potential function in containment of resin and/or modulation of volatile emissions, as seen in petunia flowers [52].

Conclusions

The findings presented in this review provide an updated model of the physical, chemical, and genetic mechanisms that allow cannabis glandular trichomes to operate as biofactories for specialized metabolite synthesis. With this, novel hypotheses can be generated regarding the unusual cellular features seen at the nanoscale, such as paracrystalline bodies in plastids, extensive membrane contact sites, cell wall remodeling, and extracellular cannabinoid synthesis. Recent progress in the use of viral-induced gene silencing [53] and transient transformation of cannabis flowers [54] offer exciting new tools to test these hypotheses. Meanwhile, as the molecular toolkit for cannabis continues to be developed, new studies using transcriptomics, genomics, and metabolomics have been essential for advancing our knowledge on cannabis glandular trichome biology.

Funding

This work was supported by the Natural Sciences and Engineering Research Council of Canada Discovery [grant number RGPIN-2019-04592].