Trichomes form a large group of plant structures that are uni- or multicellular epidermal appendages, classified by their origin, form, function, and secretion (Werker, 2000). These structures are responsible for synthesis and storage of cannabinoids and terpenes in C. sativa L., accumulating in resin heads (Hudson, 1963). They protect plants from light stress (Lydon et al., 1987), high heat (Levin, 1973; Lapinjoki et al., 1991), and herbivore pressure (Pillemer and Tingey, 1976; Alahakoon et al., 2016). Other mechanisms, including water absorption through dew collection, salt secretion, and alluring function, are reported (Werker, 2000).

All aerial parts of the cannabis plant are covered with trichomes, and can be classified as either “glandular” or “non-glandular” (Dayanandan and Kaufman, 1976). Glandular trichomes contain more bioactive/psychoactive compounds than non-glandular trichomes (Raman et al., 2017; Livingston et al., 2020). Glandular trichomes are found on all anatomical plant parts except the hypocotyl and cotyledon, and non-glandular trichomes are found on stems, leaves, petioles, stipules, bract, and tepals (Raman et al., 2017).

Glandular trichome classification relates to morphological traits and composition of the chemical substance secreted (Werker, 2000). Three types of glandular trichomes in the cannabis plant are described and size-differentiated: capitate-stalked, capitate-sessile, and bulbous trichomes (Dayanandan and Kaufman, 1976; Hammond and Mahlberg, 1977). Capitate-stalked trichomes are found exclusively on flowering regions, whereas capitate-sessile and bulbous trichomes are found everywhere except the hypocotyl and cotyledon (Raman et al., 2017). In C. sativa L., high THC-containing strains had a bigger resin head on their glandular trichomes than in low-THC industrial hemp (Small and Naraine, 2016). Capitate-stalked glandular trichomes have more secretory disc cells than other plants and secrete specialized metabolites in the subcuticular oil storage cavity, instead of through pores formed in the cuticle (Tissier, 2012; Huchelmann et al., 2017). Excretory cells secrete a resin in a subcuticular cavity (Small and Naraine, 2016). This resin contains high concentrations of the economically important cannabinoids, with psychoactive and medicinal properties (Dayanandan and Kaufman, 1976; Small and Naraine, 2016). Optimal cannabinoid and terpene biosynthesis in glandular trichomes is of paramount importance to bud quality (El-Alfy et al., 2010; Friedman and Devinsky, 2015).

Cannabinoids, also called meroterpenes or terpenophenols, are PSM synthesized by members of the Cannabaceae family, and several other plant species, including Echinacea purpurea, Echinacea angustifolia, Acmella oleracea, Helichrysum umbraculigerum, and Radula marginata (Bauer et al., 2008). More than 20% of isolated cannabis PSMs are cannabinoids (Chandra et al., 2017). The two major cannabinoids, Δ⁹-THC and cannabidiol (CBD), are used to classify cannabis (Bruci et al., 2012; Piluzza et al., 2013; Hilderbrand, 2018), and differentiation between marijuana and hemp is often based on Δ⁹-THC content from cannabis biomass. Cannabis extract with a Δ⁹-THC percentage greater than 0.3% is classified as a medical marijuana product, whereas C. sativa L. with a Δ⁹-THC content of less than 0.3% is cultivated as hemp (Hilderbrand, 2018). Three C. sativa L. chemotypes have further been distinguished and classified, determined by the relative proportions of Δ⁹-THC and CBD: drug-type (Δ⁹-THC is the predominant cannabinoid, known as marijuana), intermediate-type (both Δ⁹-THC and CBD are predominant), and fiber-type (CBD is the predominant cannabinoid, known as hemp) (Bruci et al., 2012; Piluzza et al., 2013). This differentiation based on cannabinoid content or cannabis cultivars is inadequate, particularly for the medical industry, since it does not reflect or match the therapeutic and medical properties (Russo, 2019). The term “chemovar,” which considers the specific ratios of cannabinoids, flavonoids, and terpenes, will likely be a better tool in the development of cannabis-assisted medicine (Baron, 2018).

Figure 1 shows the cannabinoid biosynthesis pathway and precursor formation. Primary biosynthesis steps are impacted by UV radiation and blue light (Dolzhenko et al., 2010; Booth et al., 2017; Jin et al., 2019; Nazari and Zarinkamar, 2020). Cannabinoid biosynthesis starts as isopentenyl diphosphate (IPP), formed from glyceraldehyde 3-phosphate (G3P), and pyruvate in plastids (Mcgarvey and Croteau, 1995). Formation of IPP in plastids is ensured by 1-deoxy d-xylulose-5-phosphate synthase (DXS), part of the methylerythritol phosphate (MEP) pathway (Lichtenthaler, 1999). The 5-carbon isoprenoid then is linked with isopentenyl diphosphate (IPP) and dimethylallyl pyrophosphate (DMAPP) through isopentenyl-diphosphate delta-isomerase (IPPi). These are condensed into geranyl diphosphate (GPP, C₁₀) via GPP synthase (GPPS) (Ruzicka, 1953; Hunter, 2007). GPP also acts as a precursor for monoterpene biosynthesis. The enzymes DXS, IPPi, and GPPS are upregulated by UV radiation and blue light in peppermint (Mentha x piperita) and water mint (Mentha aquatica) (Dolzhenko et al., 2010; Nazari and Zarinkamar, 2020).

Olivetolic acid (OA) sets cannabinoid and monoterpene biosynthesis apart. It is produced through a type III polyketide synthase, leading to formation of the cannabinoid precursor, cannabigerolic acid (CBGA) (Gagne et al., 2012). The first step of OA formation is ensured by a unique tetraketide synthase (TKS) and olivetolic acid cyclase (OAC) (Luo et al., 2019). This step uses 1 hexanoyl-CoA and 3 malonyl-CoA to form OA via a tetraketide (TK) intermediate. OA is then prenylated by geranylpyrophosphate: olivetolate geranyltransferase 4 (PT4) to form the central precursor molecule CBGA, which can then be further modified into constituents such as Δ⁹-THC, CBD, and cannabichromene (CBC) (Flores-Sanchez and Verpoorte, 2008b; Luo et al., 2019). CBGA is converted to cannabidiolic acid (CBDA), tetrahydrocannabinolic acid (THCA), and cannabichromentic acid (CBCA) by cannabidiolic acid synthase (CBDAS), tetrahydrocannabinolic acid synthase (THCAS), and cannabichromentic acid synthase (CBCAS), respectively. During these steps, cannabinoids are naturally converted from their acid forms during storage or heating (decarboxylation) as non-enzymatic catalyzed reactions (Veress et al., 1990). THCA and CBDA reactions are oxygen-dependent and produce hydrogen peroxide, as opposed to the CBCA reaction, which is oxygen-independent and can be inhibited by hydrogen peroxide (Sirikantaramas et al., 2004; Taura et al., 2007; Degenhardt et al., 2017).

CBCA is most actively synthesized in young cannabis seedlings and can be found in both drug-type and fiber-type cannabis plants, yet its concentration is relatively low compared to other cannabinoids (Kushima et al., 1980; Chandra et al., 2017). CBCA is converted to CBC (Gaoni and Mechoulam, 1966), cannabicyclolic (CBL), and cannabicyclolic acid (CBLA) through irradiation or decarboxylation (Shoyama et al., 1972). CBDA is the precursor of CBD, and THCA is the acidic precursor of Δ⁹-THC. THCA can be converted to Δ⁸-THC, cannabinol (CBN), and cannabinolic acid (CBNA) (Mechoulam and Gaoni, 1965; Elsohly and Slade, 2005).

Terpenes are a large class of organic molecules responsible for flower aroma; they include β-caryophyllene, limonene, and linalool, which are present in 50 to 70% of all studied plants (Knudsen et al., 2006; Booth et al., 2017). For monoterpene biosynthesis, GPPS condenses one unit of IPP and DMAPP to form GPP, and GPP is converted into monoterpene form via mono-terpene synthase (TPS). Sesquiterpene biosynthesis requires two units of IPP to be added to a DMAPP unit. This sequential modification of DMAPP is ensured by farnesyl diphosphate synthase (FPPS) (Kulkarni et al., 2013). FPP is converted into sesquiterpenes via sesqui-TPS (Booth et al., 2017). Involvement of other enzymes such as cytochrome P450s leads to more complex terpenes (diterpenes, C₂₀) (Grof, 2018; Booth and Bohlmann, 2019).

Independent of the inflorescence stage, major monoterpenes found in indoor-grown C. sativa L. “Finola” are α-pinene, β-pinene, β-ocimene, limonene, myrcene, and terpinolene (Booth and Bohlmann, 2019). Major sesquiterpenes expressed in trichomes are α-humulene, β-caryophyllene, bergamotene, and farnesene. As inflorescence matures, monoterpene accumulation increases relative to sesquiterpenes (Figure 1) (Booth et al., 2017). Although more than 120 terpenes have been identified in C. sativa L., many (including corresponding TPS genes) require further characterization (Aizpurua-Olaizola et al., 2016; Booth and Bohlmann, 2019). Since robust analytical standards are lacking, reported terpene profiles in C. sativa L. may contain some unknown terpene compounds, especially sesquiterpenes. A recent study reported more than 30 different TPS genes in the “Purple Kush” genome, and only 9 of 30 have been characterized (Günnewich et al., 2007; Booth et al., 2017). Elucidation of the underlying mechanisms surrounding terpene biosynthesis in cannabis plants may lead to further exploration and different medical applications for this PSM group (Aliferis and Bernard-Perron, 2020).

Terpenoids (a modified class of terpenes with different functional groups) are by far the most diverse group, with at least 80,000 different compounds (Christianson, 2017; Zhou and Pichersky, 2020). In recent years, cannabis terpenoids have slowly gained interest (Arena et al., 2016; Booth et al., 2017; Mudge et al., 2019). Studies have reported that terpenoids are powerful metabolites that have an interactive effect (or an “entourage effect”) with cannabinoid receptors (Gertsch et al., 2008). However, terpene composition in cannabis resin is dependent upon genetic, environmental, and developmental factors, and highly variable terpene profiles additionally exist between individual plants (Fischedick et al., 2010; Hazekamp and Fischedick, 2012; Booth et al., 2017). Terpene diversity in cannabis resin is responsible for scent and flavor qualities of cannabis flowers (Booth et al., 2017).

Members of the phenol family, flavonoids, form an important PSM group that aids in the plant's responses to sunlight and UV radiation (Downey et al., 2006; Warner et al., 2021). More than 20 flavonoid types in C. sativa L. have been identified, such as quercetin and kaempferol (Brenneisen, 2007). Others, such as cannflavins A, B, and C, are uniquely found in cannabis (Barrett et al., 1985, 1986; Radwan et al., 2008). Cannabis-specific flavonoids show promising therapeutic effects because of their anti-inflammatory activities (Barrett et al., 1985, 1986).

Cannabis-specific flavonoid biosynthesis is not well-established. Figure 2 shows the proposed biosynthetic pathway(s) for cannflavin A and B in C. sativa L. (Flores-Sanchez and Verpoorte, 2008b; Rea et al., 2019). The general pathway for cannflavin biosynthesis begins with p-coumaroyl-CoA derived from phenylalanine, phenylalanine ammonia-lyase (PAL), cinnamate 4-hydroxylase (C4H), and 4-Coumarate:CoA ligase (4CL). p-coumaroyl is covered to luteolin and cannflavin A and B via regiospecific methylation and prenylation reactions (Rea et al., 2019). Alternate routes for cannflavin A/B biosynthesis, beginning with feruloyl-CoA or caffeoyl-CoA with 3 malonyl-CoA, are also proposed (Flores-Sanchez and Verpoorte, 2008b). Although it has not been reported in C.sativa L., upregulated chalcone synthase (CHS) gene expression is observed in several plant species under abiotic stress such as UV radiation, as well as biotic stressors such as bacterial or fungal infection (Lipphardt et al., 1988; Dao et al., 2011).

Unlike cannabinoids and terpenes, flavonoid spatial and temporal distribution in cannabis plants does not follow the same pattern (Aizpurua-Olaizola et al., 2016). Rather, higher flavonoid content is reported in C. sativa L. leaves than other plant tissues (Flores-Sanchez and Verpoorte, 2008a; Jin et al., 2020). Apart from this, flavonoid concentration seems to decrease with plant tissue age (both leaves and inflorescence), in which higher flavonoid content is observed in young cannabis plants (Flores-Sanchez and Verpoorte, 2008a; Drinić et al., 2018). Low flavonoid content in cannabis oil and seeds is reported (Frassinetti et al., 2018; Moccia et al., 2019; Siano et al., 2019), while flavonoids are absent in glandular trichomes (Flores-Sanchez and Verpoorte, 2008b).

Recent studies show that flavonoid accumulation in inflorescence is variety-dependent and could be an indicator of the susceptibility of the variety to oxidative stress (Pavlovic et al., 2019; Giupponi et al., 2020). Pavlovic et al. (2019) reported the hemp variety “Futura 75” had higher cannabispiran concentration than “Finola.” This variety-dependent response is displayed elsewhere, where the hemp variety “Carmagnola Cs” has up to 25% more total phenol content (TPC) than other varieties, such as “Kompolti” (Izzo et al., 2020). Harnessing the radical scavenger activity and screening ability of flavonoids against UV radiation is a promising means of increasing flavonoid production in medical varieties (Agati and Tattini, 2010). Although it is out of the scope in this review, it is still worth to mention that the differences in the flavonoid quantifying methodologies, such as solvents used, matrix to solvent ratio, and characterization methods may result in flavonoid concentration discrepancies (Drinić et al., 2018; Frassinetti et al., 2018; Pellati et al., 2018).

Different wavelength ranges in UV radiation result in varying cannabinoid accumulation (Lydon et al., 1987; Magagnini et al., 2018). It has been nearly four decades since the first study suggesting that UV-B (280–315 nm) radiation affects cannabinoid accumulation in cannabis plants (Lydon et al., 1987). UV-B radiation did not impact cannabinoid content in both drug- and fiber-type cannabis plants, with the exception of Δ⁹-THC in bud tissues of drug-type cannabis plants. When the daily dosage of UV-B radiation increased from 0 to 13.4 kJ m⁻², the Δ⁹-THC content increased from 25 to 32% (Lydon et al., 1987), suggesting that Δ⁹-THC was a UV-B photo-protectant (Pate, 1994). It was further noted that UV-B radiation increases trichome numbers. Altitude may be equally important. Increased solar UV radiation results in higher CBDA, terpene, and cannaflavin content in the hemp variety “Kompolti” (Giupponi et al., 2020). Notably, UV radiation sources used in both studies had relatively broad spectra, compared to electrical UV radiation sources, such UV-discharge lamps and light-emitting diodes (LEDs). It is unknown if there is was an interactive effect between UV-A (315–380 nm) and UV-B radiation, as a high percentage of UV-A radiation was present in both the UV-B and control light treatments (Mirecki and Teramura, 1984; Lydon et al., 1987; Giupponi et al., 2020). A subsequent study examined the impact of UV-A radiation on cannabinoid accumulation, and reported increased cannabinoid levels other than Δ⁹-THC (Magagnini et al., 2018). Low percentages of UV-A radiation (2%) from full-spectrum LED arrays induced an increase of several cannabinoids, including CBD, CBG, Δ⁹-THC, and tetrahydrocannabivarin (THCV), compared to a high pressure sodium (HPS) lamp that contained 1% of UV-A radiation (Magagnini et al., 2018). Clearly, more studies are required to clarify the impact of UV radiation on cannabis PSM accumulation.

The impact of visible light on cannabis PSM accumulation has been investigated with different lighting configurations and different wavelengths (Hawley, 2018; Magagnini et al., 2018; Namdar et al., 2019) (Table 1). A high percentage of blue light cause increased cannabinoid content in cannabis inflorescence (drug-type cannabis, high amount of THC) (Hawley, 2018; Namdar et al., 2019; Danziger and Bernstein, 2021). Hawley (2018) examined the impact of subcanopy lighting with two different light spectra, 440 + 660 nm (blue + red, BR) and 440 + 530 + 660 nm (blue+ green + red, BGR), on cannabinoid and terpene accumulation. Increased Δ⁹-THC content and high CBGA levels were observed under both subcanopy BR and BGR lighting. Subcanopy BGR lighting had a higher impact on terpene accumulation than BR lighting, on both upper and lower canopies (Hawley, 2018). Increased CBGA content under LED lighting was similarly reported (Namdar et al., 2019; Danziger and Bernstein, 2021). During the flowering stage, light treatment with rich-blue light from overhead blue-red LED fixtures increased CBGA content and the CBGA: THCA ratio (Namdar et al., 2019).

Conflicting results on the interactive effects between blue light and cannabinoid content, however, were reported recently in fiber-type cannabis (hemp) (Wei et al., 2021; Westmoreland et al., 2021). Westmoreland et al. (2021) investigated the impact of light spectra on fiber-type cannabis and reported that neither CBD nor THC accumulation was impacted by spectral quality. The authors reported that this was likely caused by high light level (900 μmol·m⁻²·s⁻¹) used as the saturation state of photoreceptors was reached, resulting in low sensitivity of cannabinoid accumulation to spectral quality (Westmoreland et al., 2021). Wei et al. (2021) also reported that no significant correlation between blue light fraction and cannabinoid yield was found in fiber-type cannabis; however, note that in this study the light levels used was between 28 and 540 μmol·m⁻²·s⁻¹. As such, it is unknown that whether such variation on the interactive effect between spectral quality and cannabinoid accumulation is caused by light levels or cannabis chemotypes. Apart from blue light, supplemental green light induced cannabis PSM accumulation, including Δ⁹-THC and terpenes (limonene, linalool, and myrcene) (Hawley, 2018). No physiological theories explain how supplemental green light induces cannabis PSM accumulation. Clearly, both spectral properties and cannabis chemotype used highly impact cannabinoid accumulation, and further investigation on the links between spectral properties, cannabis chemotype, and photoreceptor is required to clarify the spectral effects.