Evolution of the Polyphenol and Terpene Content, Antioxidant Activity and Plant Morphology of Eight Different Fiber-Type Cultivars of Cannabis Sativa L. Cultivated at Three Sowing Densities

2.2. Cannabinoid Acids Content of the Studied Cultivars

Cannabinoids are biosynthesised in plant tissues in the acid form. Cannabinoic acids such as cannabidiolic acid (CBDA) and cannabigerolic acid (CBGA) are the dominant cannabinoids constituents of fibre type Cannabis sativa [16]. Tetrahydrocannabinolic acid (THCA), CBDA and CBGA were the only forms detected in our samples. Each hemp sample was frozen on the same day within a few hours after the harvest, preventing the conversion of cannabinoid acids into their neutral forms via a decarboxylation reaction, which can be triggered by heat and light during the drying process [17]. As previously described [16], the fresh inflorescences of the present study were shown to contain moderate levels of cannabinoid acids CBDA and low levels of THCA, with the exception of cultivars Santhica 27 and 70, which were characterised by significantly lower levels of THCA and CBDA, but significantly higher level of their precursor CBGA in comparison with all other cultivar studied (t-Student test, Table 1).

All eight cultivars in the present study comply with European and Swiss regulations by containing less than 0.2% THC in its THCA form, the main psychoactive constituent of Cannabis sativa. They also comply with their commercial description, stating for all of them a content <0.2% THC, and 0% THC for Santhica 27 and 70.

Five different chemotypes of Cannabis are described in the literature, based on the THCA/CBDA ratio [18]. Finola, Felina 32, Fibror 79, Futura 75, KC Virtus (male and female) and Fedora 17 had a low THCA/CBDA ratio (<1), classifying them as chemotype III, typical for fibre-type plants. Santhica 27 and Santhica 70 were the only two cultivars where no THCA was detected and in which the CBDA levels were significantly lower than in other cultivars (<0.2 mg/g). Moreover, CBGA was detected in significantly higher concentrations compared to the other cultivars. This high CBGA content with undetectable THCA classifies Santhica 27 and 70 as chemotype IV.

2.3. Total Phenolic Content and Antioxidant Activity

The total phenolic content (TPC) and the antioxidant activity of the hemp samples collected between July and September were measured using spectrophotometric methods (Table 2).

The TPC ranges from 22.05 mg GAE/g dw (Fedora 17, density 150, harvested on the 22nd of July at early flowering stage), to 4.72 mg GAE/g dw (Santhica 70, density 300, harvested on the 10th of September at the end of flowering), and are in line with previous studies on industrial hemp species [19,20].

The TPC decreased significantly during the flower’s development for all cultivars studied, by a range of 25.5% to 76.3% between July and September, depending on cultivars. The highest TPC was found in the inflorescences harvested at early flowering stage in July for all cultivars and all sowing densities. This phenomenon has already been described for other fruits, such as peaches [21] or medlar fruits [22], but to our knowledge, never for hemp inflorescences. As described by Pourcel et al., polyphenolic compounds, such as flavonoids, can be oxidised within the plant due to an increased activity of different polyphenol oxidases and peroxidases during normal seed and plant development [23]. This oxidation process offers a physico-chemical protection of the plant and seeds tissues against different stresses [23].

The DPPH radical scavenging activity of all the extracts was also evaluated and decreases as well over the maturation period. The antioxidant activity, as determined by DPPH assays, had a significant correlation with the total phenolic content calculated for all flowering stages, sowing densities and cultivars (R² = 0.96, Table S2). The extract of Fedora 17, harvested in July (early flowering; concentration 3.2 µg/mL), was able to scavenge 43% of DPPH radical and was therefore the most effective antioxidative extract. In comparison, the extract of Santhica 70, harvested in September (end of flowering), which contained 79% less total polyphenols, scavenged only 8% of the DPPH radical at the same concentration.

When comparing the different data obtained, Fedora 17 (density 150), Felina 32 (density 300), harvested in July and KC Virtus (density 30) harvested at the end of August (all in early flowering stage), were the richest in total polyphenols, showing the highest antioxidant activity, whereas Finola (density 150), Futura 75 (density 150) and Santhica 70 were the poorest. It should be noted that for the KC Virtus variety grown at a density of 150 plants per m², the TPC values and antioxidant activities remain stable throughout the early flowering period, before dropping during full flowering.

2.4. Flavonoids Composition

HPLC-UV-MS/MS analysis revealed the presence of several flavonoids within the inflorescences of the different cultivars. In total, fifteen different compounds were identified based on retention time, absorbance spectra, exact mass and mass fragmentation patterns (Table 3). When possible, the identification was confirmed using the commercially available reference standard.

The identification of the glycosylated flavonoids was mainly realised based on MS and MS/MS data acquired in the positive ion mode. Structure elucidation of flavonoids glycosides using tandem mass spectrometry has been extensively described [24,25,26]. Important structural information can be obtained on the carbohydrate sequence, the aglycone part and on the C– or O– linkage between the aglycone and the glycoside. Fragmentation spectra of each compound are accessible in the supplementary material (Figures S1–S9).

However, for compounds 3 and 4, the observation of the MS/MS fragmentation patterns recorded using high collision energy (70 eV) did not allow a distinction to be made between kaempferol or luteolin moieties. The absorbance spectra of the two compounds were hence recorded and compared (Figures S10 and S11). Shifts can be observed in the maximum absorbance wavelengths between kaempferol (max. at 265 nm and 365 nm) and luteolin (max. at 252 nm, 267 nm and 349 nm) (Table S3 and Figures S10 and S11). Compounds 3 and 4 showed similar absorbance spectra to the one of luteolin (max. absorbance at 255 nm, 267 nm and 350 nm), and were therefore identified as luteolin-C-glucoside (3) and luteolin-C-hexoside-O-rhamnoside (4).

The flavonoid content varied from 7.37 mg/g of dry weight (Felina 32, density 300, harvested on the 22nd of July, early flowering stage) to 0.06 mg/g (Santhica 70, density 300, harvested on the 10th of September, end of flowering stage) (Tables S3–S12). As already observed for the total phenolic content and the antioxidant activity (Table 2), the content in flavonoids decreased with flower maturation, and was higher in the inflorescences harvested in the early flowering stage, for all cultivars.

Regarding the composition, even if every cultivar differed in the concentration of the flavonoid contents detected, the female inflorescences contained mainly flavone derivatives, apigenin, luteolin and chrysoeriol glycosides being the most abundant compounds, regardless of the cultivar. Luteolin-7-O-glucuronide (8) was the most abundant flavonoid in all cultivars, followed by apigenin-7-O-glucuronide (9), vitexin-2’’-O-glucoside (6) and vitexin-2’’-O-rhamnoside (7), which concurs with the findings of Vanhoenacker et al. [27]

Male inflorescences of the two dioicous cultivars of the study (KC Virtus and Finola) contained overall less flavonoids than female inflorescences, as well as two unique flavonol compounds, quercetin-O-sophoroside (1) and kaempferol-O-sophoroside (5) (Tables S6–S8 and S13), found for the first time by Ross et al. in the pollen of Cannabis sativa [28].

The data obtained on the total phenolic content, the antioxidant activity and the concentration of the different flavonoids within the hemp extracts were statistically analysed using the Principal Component Analysis (PCA) method. The first outcome of the analysis is given by the biplot in Figure 2 (Figure 2a represents the direction of each variable that went into the PCA; Figure 2b is the biplot of the first two principal components of PCA). The total seventeen principle components explain 100% of the total variance. The two first principal components, PC1 and PC2, represent 68% of the variance. In PC1 the corresponding loading factors were positive for all flavonoid glycosides, and negative for the aglycone parts (Figure 2a). For greater ease of comprehension, the data were coloured by their phenological stage.

Hemp samples of all cultivars harvested in early flowering stage (circled in blue), full flowering (circled in violet) and end of flowering stages (circled in green) are clearly separated along the PC1 axis. Early flowering samples showed higher content in different flavonoids, in total phenolic content and antioxidant activity than inflorescences collected at full flowering or end of flowering stages. Inflorescences at full flowering and end of flowering group together, indicating a similar content in terms of flavonoids, total phenolic content and antioxidant activity (Figure 2; Table 2; Figures S4–S13).

To analyse the effect of the sowing density on the flavonoid content of the different cultivar, a second PCA was conducted, this time by keeping only the data from the harvest date showing the highest total phenolic and flavonoids content, and antioxidant activity, for each cultivar. The two first components account for 60.2% of the variance. The data presented in Figure 3 were coloured according to cultivars.

The result of the second PCA (Figure 3) shows that the effect of the sowing density differs depending on the cultivars.

For the cultivar Felina 32, it can be seen in Figure 3b (light green colour) that the two densities are apart in the biplot graph, indicating a difference in chemical composition between the two planting densities. Looking in detail, it can be seen that cultivar Felina 32 grown at the density of 300 seeds/m² contains significantly more flavonoids than the inflorescences from density 150 (Table S5; p < 0.01; +40%). For the cultivars KC Virtus, Fibror 79 and Futura 75, no differences can be seen between the densities 150 and 300 seeds/m² harvested at the same phenological stage, as shown in the blue circle in Figure 3b. Looking in detail at the flavonoid composition, the TPC and the antioxidant activity, no significant differences are seen. However, the cultivar KC Virtus sown at density 30 seeds/m² and harvested at the early flowering stage (end of August 2019) grouped with the two samples of densities 150 and 300 harvested at early flowering stage in July (Figure 3b; green circle). Looking in detail, this sample and the one harvested in full flowering stage in September contained significantly more total polyphenol and flavonoids than the inflorescences harvested at the same period and sown at a higher sowing density, even though they were at the same phenological stage (Table 2; Tables S6–S8). The plants sown at 30 seeds/m² also showed differences in terms of plant morphologies, with bigger inflorescences (Figure 1). The same observation was made for the cultivar Fibror 79, where the inflorescences of the plants grown at density 30 seeds/m² contained significantly more total polyphenols and flavonoids on the 27th of August (end of early flowering stage) than inflorescences of the densities 150 and 300, yet at the same flowering stage (Table 2; Table S9). Even though no trend has emerged, however, it appears that for some cultivars a low planting density would have a positive impact on the concentration of polyphenols within the flowers.

Male plants of the dioicous cultivars KC Virtus form a cluster, as can be seen in Figure 3b (circled in black). This is explained by the fact that male inflorescences of KC Virtus contain two unique flavonoids, kaempferol-O-sophoroside and quercetin-O-sophoroside [28], as well as 2 aglycones, apigenin and luteolin (arrows on Figure 3a), but also contain overall fewer amounts of flavonoids and total phenolic compounds, having a smaller antioxidant activity than female inflorescences (Table 2; Tables S6–S8). The male inflorescences of Finola cultivar contain less kaempferol-O-sophoroside and quercetin-O-sophoroside, and do not contain any aglycones (Table S13), and therefore are not clustered with KC Virtus male inflorescences.

As the antioxidant activity of a plant extract depends on the content of various class of phytocomponents, a correlation analysis was performed (Table S2). This analysis revealed that several specific flavonoid compounds of hemp inflorescences were positively correlated to the antioxidant activity: luteolin-7-O-glucuronide and luteolin-C-hexoside-O-rhamnoside having the highest degree of correlation (0.81, p < 0.05), followed by isoorientin-2’’-O-glucoside (r = 0.76, p < 0.05), luteolin-C-glucoside (0.72, p < 0.05) and chrysoeriol-O-glucuronide isomer 1 (0.71, p < 0.05). Those compounds are therefore regarded as the most potent antioxidative compounds among all flavonoids found in hemp inflorescences. On the other hand, the three aglycones—luteolin, apigenin and chrysoeriol—found in few samples were negatively correlated with the antioxidant activity.

The overall analysis reveals that cultivars Felina 32, grown at 300 seeds/m², KC Virtus, grown at 150 and 300 seeds/m², Fibror 79, grown at 150 or 300 seeds/m² and KC Virtus, grown at 30 seeds/m² (all harvested at the early flowering stage), contain the highest amount of flavonoids, of total polyphenols as well as have the highest antioxidant activity. As the consumption of dietary flavonoids, and among them of flavonols and flavones, have been shown to have protective effects against ovarian and colorectal cancers, and to prevent coronary heart disease [29], those three cultivars are therefore presumed to be the most potent in terms of health benefit.

4.2. Plant Material and Agricultural Conditions

Eight industrial hemp cultivars of Cannabis sativa L. (Finola, Fedora 17, KC Virtus, Santhica 27, Santhica 70, Fibror 79, Felina 32, Futura 75) were cultivated in the garden of the Grüental campus of the Zurich University of Applied Sciences (ZHAW) in 2019 (Wädenswil, Switzerland, latitude 47°13′02.3″ N, longitude 8°40′57.5″ E). The experiment lasted from 7th of May 2019 (sowing) until 25th of September 2019 (uprooting). The main characteristics of the cultivars are reported in Table 5.

The cultivars of the study are in compliance with the EC regulation (No 809/2014 of 17 July 2014) and are therefore authorised for cultivation in Switzerland.

The soil is a deep sandy clay soil. The mineral nitrogen content was measured in the soil at various depths before fertilisation. The layer between 0 and 30 cm deep contained 22 kg N–NO3/ha, the layer between 30 and 60 cm contained 8 kg N–NO₃/ha and the layer between 60 and 90 cm contained 2 kg N–NO3/ha. Fertilisation using Plantaktiv Typ K (16% N, 6% P₂O₅, 26% K₂O, 3.4% MgO; Hauert, Grossaffoltern, Switzerland) was done on 07.06.2019 and 26.06.2019. After solubilisation in water, the fertiliser was distributed on the field with an irrigation sprinkler (approx. 20 kg N/ha 4 weeks after sowing, approx. 40 kg N/ha 7 weeks after sowing). No pesticides were used, as well as no irrigation, except the water brought during the fertilisation. Some climatic parameters of the study site are given in Table 6.

Hemp seeds were sown by hand at a sowing depth of 4 cm and at different sowing densities (30/150/300 seeds/m², depending on the species, as indicated in Table 1) on the 7th May 2019, in 7 rows with 1.05 m width, and 22 cm within rows. The lengths of the plots differed according to the sowing density: 20 linear metres were sown for the sowing density of 30 seeds/m², 12 linear metres for the sowing density of 150 seeds/m² and 8 linear metres for the sowing density of 300 seeds/m². The borders of the strip were planted with the cultivar KC Virtus, in order to prevent an edge effect. These border plants were not harvested.

4.3. Harvests

Plants were harvested at four different periods, from early flowering to end of flowering in July 2019 (11 weeks after sowing), August 2019 (14 and 16 weeks after sowing) and September 2019 (18 weeks after planting). The phenological stage of each cultivar for each harvest date is indicated in Table 7 [45].

For the 300 and 150 plants/m² densities, 1 m² was harvested at each time point. For the 30 plants/m² density, between 10 and 20 stems were harvested. Inflorescences, leaves and stems were separated. Half of the harvest was dried in a heating cabinet at 45 °C for 4 days, acclimatised at room temperature before the different part were weighed to determine the proportions of inflorescences, leaves and stems described in Figure 1. For the second half of the harvest, only the fresh inflorescences were kept and frozen at −20 °C immediately after the harvest, then crushed a couple of days after with liquid nitrogen and ground using an Ika A11 basic analytical mill. The ground samples were kept at −80 °C until further analysis (total phenolic content, antioxidant activity, cannabinoid acids, flavonoids and terpenes analysis).

4.4. Hemp Extraction for TPC, Antioxidant Activity and HPLC-MS/MS Analysis

In total, 56 hemp samples were analysed. A total of 30 g of each frozen sample was freeze-dried. Then, 0.4 g of dried powder was weighed in a 15 mL centrifuge tube and extracted with 5 mL of a 70:30 methanol:water mixture. The tube was sonicated in an ultrasound bath for 10 min, then centrifuged at 4000 rpm for 2 min. Extractions were made in duplicate.

The supernatant was filtered through 0.2 µm polytetrafluoroethylene (PTFE) filters (Chromafil extra, Machery-Nagel AG, Oensingen, Switzerland) into HPLC 2 mL amber vials and analysed directly through HPLC-MS/MS.

The remaining supernatant was used to determine total phenolic content and antioxidant activity.

4.5. Total Phenolic Content

The total phenolic content (TPC) of the extracts were determined by the Folin–Ciocalteu reagent method [46]. To obtain a working solution, 2 M Folin–Ciocalteu reagent was diluted in a ratio of 1:3 with water. Hemp extract was diluted 50 times so that finally 1 mL of the diluted extract was mixed with 1 mL of Folin–Ciocalteu reagent, 2 mL of a 20% anhydrous sodium carbonate solution, 2 mL of water, mixed well and left for reaction in the dark for 2 h. The absorbance of the reaction mixture was measured at 750 nm using a spectrophotometer (Genesys™ 10S, Thermo Fisher Scientific, Reinach, Switzerland). All samples were carried out in triplicate for each sample (mean ± SD). Gallic acid was used as a calibration standard and results were expressed as milligrams of gallic acid equivalent per gram of dry hemp (mg GAE/g). The calibration curve of gallic acid was in the calibration range of 0.005–0.05 mg/mL with a linear regression line of y = 20.351x − 0.0059 with R² = 0.9962.

4.6. Antioxidant Activity

The antioxidant activity of the extracts was determined using DPPH assay following the protocol described by Pedan et al. [47]. The DPPH working solution was prepared by dissolving 24 mg of DPPH in 550 mL methanol and left to react in the dark for 24 h at room temperature to obtain an absorptance of 1.1 ± 0.2 units at 515 nm. The extract was then diluted 25 times in water, and 150 µL of the diluted extract was allowed to react with 2850 µL of the DPPH working solution. The absorbance at 515 nm was noted after an incubation time of 30 min using a spectrophotometer (Genesys™ 10S, Thermo Fisher Scientific, Reinach, Switzerland). The inhibition percentage for scavenging DPPH radical was calculated according to the equation:

where ABS_c is the absorbance of the blank solution at 515 nm, and ABS_s is the absorbance of the sample after 30 min of incubation with DPPH at 515 nm.

4.7. HPLC-MS/MS Analysis of Flavonoids

LC-MS/MS analyses were conducted on a system consisting of an Agilent 1290 Infinity II chromatographic system coupled to an Agilent 6530 Q-TOF mass spectrophotometer. Separation of analytes was performed using an Agilent Poroshell 120 EC-C18 (2.1 × 150 mm, 2.7 µm) column preceded by a guard column (Agilent EC-18, 2.1 × 5 mm, 2.7 µm). The flow rate was set to 0.12 mL/min, and the column temperature set at 50 °C. The two elution mobile phases were made up of water +0.1% trifluoroacetic acid (TFA) (mobile phase A) and 95:5 acetonitrile: water +0.1% TFA (mobile phase B). Gradient elution was as follows: 0–1 min, 10% B; 3 min, 12% B; 4 min, 15% B; 18 min, 16% B; 20 min, 19% B; 21 min, 50% B; 22 min, 60% B, 24–33 min, 65% B; 34–38 min, 100% B. Re-equilibration time was 11 min. Injection volume was 1 µL. All samples were analysed in duplicate.

The MS analyses were performed using Agilent 6530 Q-TOF instrument in positive ionisation mode (ESI), in the spectral range of 100–1700 Da. Nitrogen served as the nebuliser and collision gas. The MS parameters were as follows: gas temperature, 320 °C; drying gas, 10 L/min; nebuliser, 50 psi; sheath gas temperature, 350 °C; sheath gas flow, 12 L/min; capillary voltage, 3500 V. For MS/MS analysis, four different collision energy were applied: 15/30/45/70 eV using Agilent auto MS/MS mode.

For the quantification, the reference standards were used when possible. When not, the respective aglycone was used as reference compound. Calibration curves were built using the peak area at the corresponding wavelength for each reference under the chromatographic conditions described above, and calibration solutions were analysed in duplicate. The equations of the obtained linear curves are available in Table 8.

4.8. Plant Extraction for GC-MS and GC-FID Analysis of Terpenes

In order to analyse the terpene profiles, an extraction approach was developed which enables a targeted extraction of the hemp mono- and sesquiterpenes directly out of the plant material, resulting in extracts with relatively low amounts of the co-extracted lipid fraction. For this purpose, 2 g of frozen hemp samples were extracted with 20 mL of acetonitrile, homogenised using a Polytron homogeniser at 1000 U/min for 1 min and further extracted under magnetic stirring for 30 min. After the extraction, the solids were allowed to sediment. Subsequently, 7 mL of the supernatant were removed and spiked with 70 µL of injection standard stock solution (amylacetate and methylpelargonat at 1 g/L). In the next step, 3 mL ultra-pure water were added to spiked supernatant and the preparation was vortexed for 30 s with 0.4 g of silica gel C18, in order to bind the co-extracted long chain fatty acids and other nonpolar compounds. After the sedimentation of the silica gel C18, 7 mL of the supernatant were removed and diluted with 20 mL ultrapure water, after which 4 g of sodium chloride were added and dissolved by shaking. Furthermore, 5 mL of hexane were added, and a liquid/liquid extraction was performed by shaking vigorously for 30 s. After phase separation, approximately 1 mL of organic phase was filled into a GC vial. The extracts obtained were used for both GC-MS screening and the quantitation by means of GC-FID.

4.9. GC-MS Analysis of Terpenes

For compound identification, mass spectra were generated by TSQ 8000 Evo mass spectrometer coupled to a Thermo Trace GC equipped with a TriPlus RSH Autosampler (Thermo Scientific). The samples (1 µL) were injected on column. The separation of the volatiles was performed on a DB-FFAP capillary column (30 m × 0.25 mm, 0.25 µm film thickness; Agilent) connected with a fused silica pre-column capillary (5 m × 0.53 mm; Handel Müller GmbH) using helium as the carrier gas at a constant flow rate of 2.5 mL/min, by application of the following temperature program: 40 °C for 6 min, then raised to 240 °C at a rate of 7 °C/min and held at 240 °C for 25 min. The injection port was set at 50 °C, and the transfer line to 240 °C with an increase of 14.5 °C/s.

The identification was done in the electron ionization (EI) mode (70 eV). The spectra for identification were recorded in the full scan mode from 35–250 amu with a scan time of 11 ms. The ion source temperature was 220 °C. NIST mass spectral library software (version 2.2; Standard Reference Data Program of the National Institute of Standards and Technology, distributed by Agilent Technologies) was used to assist in compound identification. Most of the compounds identified by the MS-Spectra were verified by commercially available standards (see supporting information S14).

4.10. GC-FID Quantification of Terpenes

For the quantitation of the analytes identified by GC-MS, the experiments were performed using a GC system (Thermo Trace GC Ultra) equipped with a Flame Ionisation Detector (FID). Separation of the compounds was conducted on a column DB-1701 30 m × 0.25 mm (0.25 µm film thickness; Agilent Technologies, Basel, Switzerland), using helium as the carrier gas. The injection volume was 1 µL, and the flow rate 2.5 mL/min. The injector temperature was 200 °C with a split ratio of 1:10 and the FID temperature was 250 °C. The oven temperature was programmed as follows: the column was held initially at 40 °C for 5 min, then increased to 190 °C at 7 °C/min, then increased to 280 °C at 50 °C/min. For the GC-FID quantitation, standards and blanks were prepared. For blank, 7 mL of acetonitrile +70 µL of stock solution of injection standard at 1 g/L, then prepare as described above.

For calibration, three different concentrations of each analyte were prepared by diluting 30, 70 and 140 µL of the standard stock solution (1 g/L) in 7 mL of acetonitrile and 70 µL of stock solution of injection standard (1 g/L), then prepared as described above. The determined calculations showed a calibration range from 0.1 to 92 mg/kg and a coefficient of determination (R²) ≥ 0.999.

The authors would like to thank Katrin Jedrys and Sandra Panarese for their valuable work throughout this project.