Selective Extraction of Cannabinoid Compounds from Cannabis Seed Using Pressurized Hot Water Extraction

Phytochemicals of Cannabis sativa mainly for the use in the different industries are that of delta-9-tetrahydrocannabinol (THC) and cannabidiol (CBD). Pressurized hot water extraction (PHWE) is seen as an efficient, fast, green extraction technique for the removal of polar and semi-polar compounds from plant materials. The PHWE technique was applied to extract cannabinoid compounds from Cannabis sativa seed. Response surface methodology was used to investigate the influence of extraction time (5–60 min), extraction temperature (50–200 °C) and collector vessel temperature (25–200 °C) on the recovery of delta-9-tetrahydrocannabinol (THC), cannabinol (CBN), cannabidiol (CBD), cannabichromene (CBG) and cannabigerol (CBC) from Cannabis sativa seed by PHWE. The identification and semi quantification of cannabinoid compounds were determined using GCXGC-TOFMS. The results obtained from different extractions show that the amount of THC and CBN was drastically decreasing in the liquid extract when the temperature rose from 140 to 160 °C in the extraction cell and the collector′s vessel. The optimal conditions to extract more CBD, CBC, and CBG than THC and CBN were set at 150 °C, 160 °C and 45 min as extraction temperature, the temperature at collector vessel, and the extraction time, respectively. At this condition, the predicted and experimental ratio of THCt (THC + CBN)/CBDt (CBD + CBC+ CBG) was found to be 0.17 and 0.18, respectively. Therefore, PHWE can be seen as an alternative to the classic extraction approach as the efficiency is higher and it is environmentally friendly.


Introduction
Cannabis L Sativa is a plant belonging to the family of Cannabaccae and grows to around 2-5 m in height [1]. The plant is considered dioecious since it contains both male and female parts [2]. It exhibits several psychoactive and medicinal activities. The use of Cannabis has been recorded throughout history in many industries, such as food, cosmetics, paper, clothing, and pharmaceutical [3,4]. The phytochemicals within the plant have gained interest overtime for its medicinal purposes, mainly for its use against cancer-inducing effects [4,5].
Up today, there are above 500 compounds within the Cannabis L Sativa plant which are known, the most interesting is a set of compounds known as cannabinoids, there are about 70 different cannabinoids which are unique to the Cannabis L Sativa plant [5,6]. Among these, the most psychoactive compound being delta-9-tetrahydrocannabinol (THC), the metabolic breakdown of the compound, which is cannabinol (CBN), is still psychoactive, but less than that of THC. The second most abundant compounds are that of cannabidiol (CBD), cannabichromene (CBG) and cannabigerol (CBC). These

Chemicals and Reagents
Chemicals used in this study were of analytical grade. Methanol was used as a solvent and purchased from Sigma-Aldrich (Johannesburg, South Africa).

Plant Material
One species of cannabis was subjected to extraction and analysis. Seed samples of Cannabis L Sativa were collected from farmers in Gauteng province in August 2019. Prior to processing, seeds were dried at 50 • C for 4 h and the water content was found to be 5.5-7.9% weight. The dried samples were crushed to powder and kept in a sealed container until extraction.

Response Surface Methodology
The influence of extraction temperature (80-200 • C), the temperature at collector vessel  • C) and extraction time (15-60 min) on the selective recovery of cannabinoid compounds were investigated using response surface methodology (RSM). Full factorial design, comprising of 30 experiment run with three center point replicates, was created in MODDE Pro (Sartorius Stedim Biotech, Malmö, Sweden), to assess the recovery of THC, CBN, CBD, CBG, and CBC from the Cannabis sativa seeds. During the optimization process, the pressure and flow rate were kept constant at 105 bar and 0.5 mL min −1 , respectively, according to our previous work [24]. The partial least square regression was applied to evaluate the fitting of the model and response surface. The adequacy of the models was evaluated by the R 2 and Q 2 values (where R 2 shows the model fit and Q 2 shows an estimate of the future prediction precision). The F-test was used to assess the significance of the coefficients of regression. The modeling was done with a quadratic model like linear, squared, and interaction terms.

Pressurized Hot Water Extraction
Pressurized hot water extraction instrument was constructed in the laboratory and equipped with an extraction cell, gas chromatography (GC) oven for controlling the extraction temperature and time, and a collector vessel ( Figure 1). This was in accordance with the method described by [24] with some modifications. For each run, a mass of 5 g of Cannabis sativa seed powder was filled into the PHWE extraction cell. Before each extraction, the cell loaded with the powder was pre-heated at a specific extraction temperature for 10 min. The dynamic extractions were done under various extraction conditions as specified by the CCD matrix ( Table 1). The approach used in the pressurized hot water extraction for cannabinoid compounds is to extraction.    The responses were expressed in percentage relative peak area, which is the peak area of each compound over the total peak area in the chromatogram; these percentages are generated by the instrument. During the extraction process, the collector s vessel was heated at different temperatures in order to investigate the effect of these temperature variations on the removal of cannabinoid compounds in the obtained liquid extract.

GCXGC-TOFMS Method
A LECO 2D GCXGC chromatography with a time-of-flight mass spectrometer (LECO, Johannesburg, South Africa) was used for the identification and analysis of cannabinoid compounds from Cannabis sativa seed. This experimental GCXGC-QTOF-MS/MS method developed by Marrotti et al. [29] was used with some modification. The separation was carried out in a 30 m × 0.25 mm × 0.25 µm BPX5 column (SGE Analytical, Johannesburg, South Africa). The injection was carried out in spit mode (ratio 20:1). The carrier gas was helium (99.999% purity) and was used as at a flow rate of 1 mL min −1 . The oven temperature was set at 100 • C (for 2 min) and increased to 280 • C with rate of 15 • C min −1 . Temperatures applied were 300 • C for injector, 250 • C for transfer line, 250 • C for ion source, and 150 • C for quadrupole. Data were acquired in the full scan mode with mass ranging from 45-600 amu. The total ion chromatograms (TIC) were integrated. Raw data (m/z) generated by UHPSFC were processed using the ChromaTOF software version 4.5.1. (LECO Corp, St. Joseph, MI, USA).

Quality Assurance
Deionized water was used throughout the study. Glassware was properly cleaned, and all the reagents were of analytical grade. All extractions were done in triplicates. All measurements were inter-day repeatability measurements taken over a week. In addition, the optimum extraction condition was partially validated for repeatability and reproducibility using relative peak areas.

Model Fitting and Predictive Efficiency
The model fitting and predictive efficiency between the experimental and predicted data at optimum extraction conditions were investigated by using the absolute average deviation (AAD), root mean square error (RMSE), mean absolute error (MAE), standard error of prediction (SEP), model predictive error (MPE), and chi-square statistic (χ 2 ), and correlation coefficients (R 2 ). The equations used to calculate these factors are described in Table 2. Table 2. Equations of error functions.

Identification of Cannabinoid Compounds
Cannabinoid compounds were identified using GC-TOF/MS. The chromatogram and fragmentation patterns are presented in Figures 2-4. Five cannabinoid compounds such as THC, CBN, CBD, CBC, and CBG were identified from the plant extracts. Five cannabinoid compounds have been identified in the cannabis extracts. Figure 2 shows a chromatogram obtained for a cannabis extract.

Identification of Cannabinoid Compounds
Cannabinoid compounds were identified using GC-TOF/MS. The chromatogram and fragmentation patterns are presented in Figures 2-4. Five cannabinoid compounds such as THC, CBN, CBD, CBC, and CBG were identified from the plant extracts. Five cannabinoid compounds have been identified in the cannabis extracts. Figure 2 shows a chromatogram obtained for a cannabis extract.
The mass spectrum fragmentation obtained for THC, CBN, CBD, CBC, and CBG were compared with the fragmentation provided by the Leco database library.  The mass spectrum fragmentation obtained for THC, CBN, CBD, CBC, and CBG were compared with the fragmentation provided by the Leco database library.
The fragment partens were compared to the Leco database (NIST library). The mass spectrum fragmentation obtained for THC, CBN, CBD, CBC, and CBG were compared with the fragmentation provided by the Leco database library. The obtained fragmentation pattern is presented in Figures 3 and 4. The similarity for all the fragmentation patterns for the targeted compounds with the library databe ranged from 80-98%.

Experimental Design: Response Surface Methodology
Collector vessel temperature, extraction time and temperature were chosen for optimization since they are the main parameters that can influence the selectivity extraction of cannabinoid compounds from Cannabis sativa seed, and thereby, also the recovery efficiency. The extracted amounts of cannabinoid compounds expressed in relative peak area (%) at different experiments of the design are summarized in Table 1. The model was created and calculated by fitting using partial least square regression. The fitted model showed a total described variance range from 90%-95% (R 2 = 0.96-0.99) and a cross-validated predictability ranged from 95-99% (Q 2 = 0.95 -0.99), where R 2 shows the model fit and Q 2 shows an estimation of the future prediction and precision [24]. The linearity of the predicted vs. observed values plot ( Figure 5) highlighted the validity of the model and its capability to predict the best condition of the extraction within the range of the design. The coefficients plot ( Figure 6) reveals that collector vessel temperature, extraction temperature and time have a significant negative influence on the extracted amount of Tetrahydrocannabinol (THC) and Cannabinol (CBN) with a p-value of 0.003, whereas extraction time has shown a significant positive influence on the amount extraction of cannabidiol (CBD), cannabichromene (CBG) and cannabigerol (CBC).

Experimental Design: Response Surface Methodology
Collector vessel temperature, extraction time and temperature were chosen for optimization since they are the main parameters that can influence the selectivity extraction of cannabinoid compounds from Cannabis sativa seed, and thereby, also the recovery efficiency. The extracted amounts of cannabinoid compounds expressed in relative peak area (%) at different experiments of the design are summarized in A C B   Figure 6. Coefficient plot.
The response surface described by a counter-plot in Figure 7 shows the direct and interaction impact of extraction time and temperature on the extracted amount of cannabinoid compounds from cannabis seeds. Increasing the extraction temperature and collector vessel temperature from 140 to 160 • C for 60 min decreased the extracted amount of THC and CBN, while the amount of CBD, CBC and CBG are raised in the extract. This is probably due to the vaporization of the THC and CBN portion at these temperatures, which ranged in temperature in liquid extract and increased the solubility of CBD, CBC and CBG in the liquid extract [20,29]. The increase of temperature from 140 to 160 • C in the extraction cell increased the solubility of the cannabinoids compounds. The collected extract was heated in the extraction cell at the temperature range from 140 to 160 • C. At this point, a portion of THC and CBN was evaporated from the extract, resulting in the final product having lower levels of THC and CBN and high amounts of CBD, CBC and CBG. Since the PHWE was equipped with the trapping system, which contained methanol, the evaporated THC and CBN were captured in this solvent. This resulted in the increase of levels of CBN and THC in the trapping solution ( Figure 8). However, a further increase in temperature above 160 • C removed both psychoactive and non-psychoactive compounds from the extract.

Model Fitting and Predictive Efficiency Analysis
The response surface methodology approach was investigated for its predictive performance and estimation capabilities. The statistical factors including AAD (%), RMSE, MAE, SEP (%), MPE (%), and chi-square (χ 2 ) have been assessed by equations described in Table 2. The obtained results of the statistical analysis are shown in Table 3.
The results showed that RSM has a higher ability to investigate the interaction of the main parameters involving the extraction of cannabinoid compounds from cannabis seed by PHWE. Partial least square regression analysis was carried out between the response (THC, CBD, CBC, CBG, and CBN) values predicted by RSM models with their corresponding experimental value. Also, The RSM model predictions are much closer to the line of perfect prediction as presented in Figure 1. Also, this higher predictive accuracy of the RSM can be attributed to its ability to approximate the linearity for a small data set. All statistical error parameters are very low, proving good fitting of the model.

Universal Extraction Condition of an Extract rich in CBD, CBC, and CBG
It is clear that the optimal pressurised hot water extraction condition would be the one for an extract with lower THC and CBN, and high CBD, CBC and CBG. For systematic optimization, a desirability function was used to find the setting of extraction temperature, collector vessel temperature and extraction time where a high amount of CBD, CBC and CBG can be extracted. Accordingly, the best condition was found at 150 • C, 160 • C and 45 min for extraction temperature, collector vessel temperature and extraction time, respectively. The amounts of cannabinoid compounds expected to be extracted under these conditions using the predicted model are shown in Table 4. The universal extraction conditions were applied to the Cannabis sativa seed to determine the exact extracted amounts. It was observed that the RSM predicted values for the extraction of cannabinoid compounds were within the standard deviation of the experimental values. Also, the repeatability and reproducibility ranged from 89.45 ± 2.87 to 92.61 ± 4.5% and 89.70 ± 5.60 to 92.56 ± 2.31%, respectively. The ratio between the THCt (THC + CBN) and CBDt (CBD + CBC+ CBG) was found to be 0.18, which means that the final extract at the optimal condition has more CBD, CBC and CBG than THC and CBN.

Conclusions
The present study was conducted in order to investigate the influence of extraction temperature, extraction time and collector vessel temperature on the pressurized hot water extraction of cannabinoid compounds from Cannabis sativa seed. Response surface methodology was performed as an optimization tool. The evaluation of different error statistical parameters such as AAD, RMSE, MAE, MPE, χ 2 , and R 2 have shown that RSM is an excellent statistical tool in terms of prediction and estimation capabilities. The optimization of the PHWE process by RSM predicted that the ratio between THCt (THC + CBN) portion and CBDt (CBD + CBC+ CBG) portion was 0.17 under optimum conditions of 150 • C, 160 • C and 45 min for extraction temperature, collector vessel temperature and extraction time, respectively, which means that the final extract at the optimal condition has higher amounts of CBD, CBC and CBG than THC and CBN. To conclude, the optimization of PHWE using the RSM approach provides an effective guideline for an extraction process that yields a formulated cannabinoid nutraceutical. The PHWE technique has produced Cannabis sativa extract with high amounts of non-psychoactive compounds and lower levels of THC and CBN.