Biological Activity of Cannabis sativa L. Extracts Critically Depends on Solvent Polarity and Decarboxylation

Minor cannabinoid and non-cannabinoid molecules have been proposed to significantly contribute to the pharmacological profile of cannabis extracts. Phytoplant Research has developed highly productive cannabis cultivars with defined chemotypes, as well as proprietary methods for the extraction and purification of cannabinoids. Here, we investigate the effect of solvent selection and decarboxylation on the composition and pharmacological activity of cannabis extracts. A library of forty cannabis extracts was generated from ten different cannabis cultivars registered by Phytoplant Research at the EU Community Plant Variety Office. Plant material was extracted using two different solvents, ethanol and hexane, and crude extracts were subsequently decarboxylated or not. Cannabinoid content in the resulting extracts was quantified, and biological activity was screened in vitro at three molecular targets involved in hypoxia and inflammation (NF-κB, HIF-1α and STAT3). Changes in transcriptional activation were strongly associated to solvent selection and decarboxylation. Two decarboxylated extracts prepared with hexane were the most potent at inhibiting NF-κB transcription, while HIF-1α activation was preferentially inhibited by ethanolic extracts, and decarboxylated extracts were generally more potent at inhibiting STAT3 induction. Our results indicate that solvent selection and proper decarboxylation represent key aspects of the standardized production of cannabis extracts with reproducible pharmacological activity.


Introduction
Cannabis sativa L. is an herbal medicine which contains at least 554 identified compounds including 150 phytocannabinoids, terpenophenolic active molecules which are nearly exclusive to this plant, and other phyto-molecular families such as terpenoids, flavonoids and oxylipins [1]. The main two cannabinoids responsible for the pharmacological properties of cannabis extracts are ∆9-tetrahydrocannabinol (THC), the principal active component in drug-type cannabis, and cannabidiol (CBD), the major cannabinoid molecule found in fiber-type cannabis or hemp [2]. These compounds are formed in the cannabis plant as their 2-carboxylic precursors, ∆9-tetrahydrocannabinolic acid (THCA) and cannabidiolic acid (CBDA), whom rapidly decarboxylate to render the neutral forms upon heating [3]. Pharmacological interest has been attributed to THCA and CBDA, as well as to other less-abundant, non-psychoactive cannabinoids such as cannabigerol (CBG), the molecular precursor of THC and CBD, and ∆9-tetrahydrocannabivarin (THCV) and cannabidivarin (CBDV), their propylic scope to focus on non-classical cannabinoid targets that could mediate cellular responses elicited by cannabis extracts. A plant extract library (PHYRESCEL) was prepared by systematically processing 10   PHYRESCEL contained 40 different entries that were further screened for their pharmacological activity towards NF-κB, HIF-1α and STAT3, three molecular targets known to mediate cellular responses to inflammation and hypoxia.  Table 1).

Reagents for Plant Material Extraction, Purification of Cannabinoids and Analysis of Cannabinoid Content
n-hexane and absolute ethanol (Scharlab, Barcelona, Spain) were used as solvents for the extraction of plant material. Petroleum ether bp 40-60 • C (Scharlab, Barcelona, Spain) or n-hexane were used for the purification of cannabinoids by crystallization (CBD, CBG and CBGA), and purified water were used in addition to the n-hexane and absolute ethanol in the purification of cannabinoids by liquid-liquid chromatography (THC, THCA, CBDA, CBDV and CBGV) used in this study. LC-MS-grade ammonium formate from Sigma-Aldrich (San Luis, MO, USA) and acetonitrile, formic acid and methanol from Scharlab (Barcelona, Spain) were used for preparation of chromatographic mobile phases and working solutions. Standards (purity > 98%) of cannabinoids THC, THCA, THCV, CBG, CBGA, CBDV, CBD, CBDA, cannabidivarinic acid (CBDVA), cannabichromene (CBC), ∆8-tetrahydrocannabinol (∆8-THC), cannabinol (CBN) and cannabicyclol (CBL) were used to make the corresponding calibration curves. All standards were from Cerilliant (Round Rock, TX, USA), commercialized as a solution in methanol (1 mg·mL −1 ).

Extraction Procedure
100 g of plant material were extracted, at room temperature, three times in organic solvent (n-hexane or ethanol) under constant stirring at 190 rpm in a mini OVAN OS10E S orbital shaker (Ovan, Barcelona, Spain). Extracts were then filtered under vacuum (Laboxact KNF, Trenton, NJ, USA) through a glass plate porous filter (Pobel, Madrid, Spain) with a pore size of 16-40 µm. Extracts were then evaporated to dryness in a RS-3000V rotaevaporator from Selecta (Barcelona, Spain) at 40 and 50 • C for hexane and ethanol, respectively. This process yielded a total of 20 extracts from Phytoplant varieties, 10 extracted with hexane, and 10 with ethanol. Dry extracts were then divided into two glass vials and one was placed into a "Conterm" oven (Selecta, Barcelona, Spain) and incubated at 150 • C for 60 min to induce decarboxylation of acid cannabinoids to their neutral forms. In this way, a total of 40 extracts were obtained: 20 decarboxylated (DEC) and 20 non-decarboxylated (NDC) (see Figure 1).

Analysis of Cannabinoid Content
For LC-DAD analysis, extracts (40 mg) were diluted with 5 mL of methanol, placed in an ultrasound bath for 25 min, and centrifuged for 5 min at 3000 rpm. Then, the supernatant containing cannabinoids was collected and further diluted 1:200 prior to analysis. An 1260 Agilent liquid chromatography (Santa Clara, CA, USA) set-up consisting of a binary pump, a vacuum degasser, a column oven, an autosampler and a diode array detector (DAD) equipped with a 150 mm length × 2.1 mm internal diameter, 2.1 µm pore size Raptor ARC-18 column with a Raptor ARC-18 EXP guard column cartridge 2.7 µm, both from Restek (Bellefonte, PA, USA), was used for the quantification of cannabinoids. The device was controlled by OpenLAB CBS software (Agilent, Santa Clara, CA, USA). Full spectra were recorded in the range of 200-400 nm. The analysis was performed in isocratic mode using water with 5 mM ammonium formate and 0.1% formic acid and acetonitrile with 0.1% formic acid as mobile phases. Flow rate was set at 0.375 mL·min −1 , the injection volume at 3 µL and the injector needle was washed between injections with methanol. Chromatographic peaks were recorded at 225 nm. All determinations were carried out at 35 • C. All samples were analyzed in duplicate. The concentration of cannabinoids was calculated by peak area versus a calibration curve prepared with a commercial standard. The results for each cannabinoid were calculated as relative weight (%) in the extract.

Sample Preparation for Biological Activity
Stock solutions of extracts and purified cannabinoids were prepared in DMSO at a concentration of 100 mg·mL −1 and stored at −80 • C until used. Experimental solutions were freshly prepared on the day of the experiment by diluting stock solutions in DMSO. Final concentrations tested were 10, 25 and 50 µg·mL −1 in the case of plant extracts and 5, 12.5 and 25 µg·mL −1 for purified cannabinoids (THC, THCA, CBD, CBDA, CBDV, CBG, CBGA). Dosage range was determined to allow for direct comparison between the biological activity of purified cannabinoids and plant extracts, anticipating a 50% w/w concentration of the main cannabinoid in the extract.

Cell Lines
Anti-NF-kB activity was determined using the Raw-KBF-Luc cell line, stably transfected with the plasmid KBF-Luc, which contains three copies of the NF-kB binding site from the major histocompatibility complex promoter, fused to a fragment of the simian virus 40 (SV40) promoter driving the firefly luciferase reporter gene. The increase in NF-κB levels activates the pCRE-Luc system, inducing the expression of the luciferase reporter gene. The determination of STAT-3 activity was performed using the Raw-STAT3-Luc cell line, which is stably transfected with the plasmid 4xM67 pTATA TK-Luc that contains STAT3 binding sites and whose activation induces the expression of the firefly luciferase reporter gene. The determination of HIF-1α activity was performed using the cell line NIH-3T3-Epo-Luc. This cell line is stably transfected with the plasmid Epo-Luc, which contains three tandem copies of the hypoxia response element (HRE) of the erythropoietin (EPO) promoter and whose activation induces the expression of the firefly reporter gene. The determination of cytotoxicity was performed using the mouse brain neuroblastoma cell line N2a obtained from ATCC (Manassas, VA, USA).

Determination of Biological Activity
100 µL of a cell suspension were seeded in 96-well plates (500,000 cells·mL −1 ) and incubated for 24 h. Cells were then treated for 30 min either with cannabis extracts or purified cannabinoids and then stimulated with either LPS (1 µg·mL −1 , in the case of NF-κB and STAT-3) [21] or deferoxamine (150 µM, in the case of HIF-1α) for 6 h [22]. Deferoxamine, a potent hypoximimetic compound acting as an iron chelating agent, or LPS were used alone as positive control and cells without stimulus were used as negative control. Cells were then washed twice with PBS 1X and lysed in 100 µL of lysis buffer (25 mM Tris-phosphate pH 7.8, 8 mM MgCl2, 1 mM DTT, 1% Triton X-100 and 7% glycerol) for 15 min at room temperature under orbital shaking. Luciferase activity was determined with a TriStar LB 942 plate multi-reader (Berthold Technologies, GmbH & Co. KG, Bad Wildbad, Germany) following the instructions of the Luciferase Assay system from Promega. Resulting RLU (Relative Light Units) were Separations 2020, 7, 56 7 of 16 normalized to positive control (deferoxamine or LPS) in order to calculate the percentage of activity left. Percentage of inhibition was calculated as: where T are the RLU of cells treated with cannabis extracts, C are the RLU of cells challenged with LPS of deferoxamine, and B are the RLU of un-stimulated control cells.

Determination of Cytotoxicity
100 µL of a cell suspension were seeded in 96-well plates (40,000 cells·mL −1 ) and incubated for 24 h. Cells were then treated for 24 h either with cannabis extracts or purified cannabinoids and then the media removed and 50 µL of MTT was added (1.25 mg·mL −1 ) and incubated for 4 h. Then the supernatant was removed and 100 µL of DMSO were added and the absorbance recorded at 550 nm with a TriStar LB 941 plate multi-reader (Berthold Technologies, GmbH & Co. KG, Bad Wildbad, Germany). Resulting absorbances were normalized to control (untreated cells) and blank (DMSO) to calculate the percentage of cell viability according to the following formula: where T is the absorbance of the cannabis extract treated cells, C is the absorbance of the untreated cells and B is the absorbance of the blank.

Statistical Analysis
Cannabinoid concentration is reported as percentage (w/w). Biological activity is expressed as the mean ± standard deviation of two or three independent assays. Dose-response inhibition of NF-κB activity was analyzed by a one-way ANOVA followed by a Dunnett's multiple comparisons test with significance set at p < 0.05, using GraphPad Prism (Version 8.1.1) (San Diego, CA, USA).

Results
Extraction yield was found to be cultivar-dependent, ranging from the less-producing variety, Octavia (main cannabinoid, CBG), to the largest producer, Moniek (main cannabinoid, THC). Although not significant differences were found, hexane extraction yields were slightly lower compared to ethanol: 4.9% vs. 6.8% for Octavia and 28.4% vs. 28% for Moniek (Table 1). This may be indicative of a richer phytocomplex with more polar molecules like flavonoids and polyphenols in the case of ethanol whom, contrary to hexane, is nor cancerous nor toxic for the environment. Accordingly, average weight loss during decarboxylation ranged from 0.37% and 6.37% for hexane extracts and between 8.65% and 16.67% in the case of ethanol, likely due to the amount of water extracted by ethanol from the plant material. All extracts pass the residual solvents analysis accordance of European Pharmacopoeia 9.0. (2.4.24, Identification and control of Residual Solvents using Methodology system A) for Class I solvents the HEX cannabis extracts and for Class III solvents the ETH cannabis extracts. Of note, the salient carbon dioxide molecule generated during decarboxylation represents approximately 12% of the molecular weight of THCA, CBDA and CBGA [3]. Decarboxylation did not produce significant changes in the color or physical appearance of extracts.
No significant differences were appreciated between solvents in the average total content of cannabinoids among extracts, which was 71.1 ± 15.7 % for hexane and 68.0 ± 10.0% for ethanol. In the case of non-decarboxylated extracts the average of total content in hexane and ethanol extracts were 74.30 ± 12.15 % and 68.1 ± 7.9% respectively (Supplementary Table S1). Figures 2 and 3 depict the relative concentrations of major and minor cannabinoids, a total of 13 is reported. Because of their low quantity and similar relative abundance across extracts, CBL, ∆8-THC, CBN, CBC and THCV were not tested in the in vitro assays. case of non-decarboxylated extracts the average of total content in hexane and ethanol extracts were 74.30 ± 12.15 % and 68.1 ± 7.9% respectively (Supplementary Table S1). Figures 2 and 3 depict the relative concentrations of major and minor cannabinoids, a total of 13 is reported. Because of their low quantity and similar relative abundance across extracts, CBL, Δ8-THC, CBN, CBC and THCV were not tested in the in vitro assays.   Figure 4 displays the degree of transcriptional inhibition (as % of activation) at the most effective, non-toxic dose for each cannabis extract in PHYRESCEL at the three targets investigated, NF-κB (left panel), HIF-1α (middle panel) and STAT3 (right panel), following incubation with LPS or deferoxamine. First, our results indicate that decarboxylated (DEC) extracts, rich in neutral cannabinoids such as THC, CBD and CBG, were consistently more toxic for the cell lines used in the assay, which share some proliferative features of cancerous cells, than their non-decarboxylated (NDC) counterparts irrespectively of the solvent used. Accordingly, most DEC extracts (>95%) were toxic at a dose of 50 μg·mL −1 , with less than 20% of NDC extracts exhibiting cytotoxicity at the same dose ( Figure 4). This observation agrees with the literature, where the cytotoxic effect of neutral case of non-decarboxylated extracts the average of total content in hexane and ethanol extracts were 74.30 ± 12.15 % and 68.1 ± 7.9% respectively (Supplementary Table S1). Figures 2 and 3 depict the relative concentrations of major and minor cannabinoids, a total of 13 is reported. Because of their low quantity and similar relative abundance across extracts, CBL, Δ8-THC, CBN, CBC and THCV were not tested in the in vitro assays.   Figure 4 displays the degree of transcriptional inhibition (as % of activation) at the most effective, non-toxic dose for each cannabis extract in PHYRESCEL at the three targets investigated, NF-κB (left panel), HIF-1α (middle panel) and STAT3 (right panel), following incubation with LPS or deferoxamine. First, our results indicate that decarboxylated (DEC) extracts, rich in neutral cannabinoids such as THC, CBD and CBG, were consistently more toxic for the cell lines used in the assay, which share some proliferative features of cancerous cells, than their non-decarboxylated (NDC) counterparts irrespectively of the solvent used. Accordingly, most DEC extracts (>95%) were toxic at a dose of 50 μg·mL −1 , with less than 20% of NDC extracts exhibiting cytotoxicity at the same dose ( Figure 4). This observation agrees with the literature, where the cytotoxic effect of neutral  Figure 4 displays the degree of transcriptional inhibition (as % of activation) at the most effective, non-toxic dose for each cannabis extract in PHYRESCEL at the three targets investigated, NF-κB (left panel), HIF-1α (middle panel) and STAT3 (right panel), following incubation with LPS or deferoxamine. First, our results indicate that decarboxylated (DEC) extracts, rich in neutral cannabinoids such as THC, CBD and CBG, were consistently more toxic for the cell lines used in the assay, which share some proliferative features of cancerous cells, than their non-decarboxylated (NDC) counterparts irrespectively of the solvent used. Accordingly, most DEC extracts (>95%) were toxic at a dose of 50 µg·mL −1 , with less than 20% of NDC extracts exhibiting cytotoxicity at the same dose ( Figure 4). This observation agrees with the literature, where the cytotoxic effect of neutral cannabinoids on immortalized cell lines has been thoroughly noted as a limitation of cell-based assays in the pharmacological characterization of both cannabinoids [23,24] and cannabis extracts [25]. Therefore, we were not able to obtain complete dose-response information regarding the potency of most entries in PHYRESCEL, particularly in the case of either DEC_HEX or DEC_ETH groups ( Figure 4). However, despite this toxicity issue, we were able to draw relevant information regarding the differential effects of solvent selection and cannabinoid decarboxylation on the biological activity of cannabis extracts. Although this not always correlated with cannabinoid content, the overall results suggest that DEC extracts displayed higher potency, with a majority of non-toxic ID50 values in the range of 10-25 µg·mL −1 , while NDC extracts were generally most effective at 50 µg·mL −1 with a few exceptions (Figure 4). cannabinoids on immortalized cell lines has been thoroughly noted as a limitation of cell-based assays in the pharmacological characterization of both cannabinoids [23,24] and cannabis extracts [25]. Therefore, we were not able to obtain complete dose-response information regarding the potency of most entries in PHYRESCEL, particularly in the case of either DEC_HEX or DEC_ETH groups ( Figure  4). However, despite this toxicity issue, we were able to draw relevant information regarding the differential effects of solvent selection and cannabinoid decarboxylation on the biological activity of cannabis extracts. Although this not always correlated with cannabinoid content, the overall results suggest that DEC extracts displayed higher potency, with a majority of non-toxic ID50 values in the range of 10-25 μg·mL −1 , while NDC extracts were generally most effective at 50 μg·mL −1 with a few exceptions (Figure 4). The molecular mechanism by which cannabinoids modulate NF-κB signaling remains controversial. For example, CBD but not THC has been shown to reduce the activity of the NF-κB pathway in BV-2 microglial cells [26]. However, synthetic cannabinoid dexanabinol, non-antioxidant CB1 receptor agonist WIN 55,212-2 and phytocannabinoid CBN, which weakly binds to CB1 receptors, inhibited NF-κB transcriptional activity and apoptosis in a neuron-like cell line, which suggests that neither the antioxidant properties of cannabinoids nor binding to cannabinoid receptors are responsible for this effect [27]. Our results show how most extracts did not block more than 50% the transcriptional activation of NF-κB in macrophages induced by LPS ( Figure 4). Accordingly, THC and CBD displayed cellular toxicity at higher doses and were inactive towards NF-κB at the dose of 10 μg·mL −1 ( Table 2). Discrepancies with cited literature may be due to differences in the cell lines used as well as in the experimental design [25]. Nevertheless, we were able to identify four cannabis extracts capable of inhibiting NF-κB activation more than 40% at non-toxic doses ( Table 2). The molecular mechanism by which cannabinoids modulate NF-κB signaling remains controversial. For example, CBD but not THC has been shown to reduce the activity of the NF-κB pathway in BV-2 microglial cells [26]. However, synthetic cannabinoid dexanabinol, non-antioxidant CB1 receptor agonist WIN 55,212-2 and phytocannabinoid CBN, which weakly binds to CB1 receptors, inhibited NF-κB transcriptional activity and apoptosis in a neuron-like cell line, which suggests that neither the antioxidant properties of cannabinoids nor binding to cannabinoid receptors are responsible for this effect [27]. Our results show how most extracts did not block more than 50% the transcriptional activation of NF-κB in macrophages induced by LPS ( Figure 4). Accordingly, THC and CBD displayed cellular toxicity at higher doses and were inactive towards NF-κB at the dose of 10 µg·mL −1 ( Table 2). Discrepancies with cited literature may be due to differences in the cell lines used as well as in the experimental design [25]. Nevertheless, we were able to identify four cannabis extracts capable of inhibiting NF-κB activation more than 40% at non-toxic doses ( Table 2). Contrary to their toxic potential, the inhibitory activity of these extracts did not seem to correlate with their cannabinoid content. To further characterize this finding, we conducted a refined dose-response experiment with extracts of Magda (ETH_NDC, Figure 5A), Beatriz (ETH_DEC, Figure 5B), Pilar (HEX_DEC, Figure 5C) and Sara (HEX_DEC, Figure 5D) and were able to confirm their potency and toxic range, with all of them inhibiting NF-κB activation in a dose-dependent manner before reaching toxic concentrations. Such effect that was not found when the same plant varieties were subjected to different processing conditions or with pure cannabinoids. Therefore, it is plausible to speculate that these extracts may contain additional molecules capable of blocking NF-κB transcriptional activation, either acting alone or as a part of a phytocomplex. For example, flavonoid Cannflavin A and a phenanthrenequinone derivative, Denbinobin, present in the cannabis variety "Carma", elicit synergistic effects in combination with CBD and CBG to inhibit NF-κB activation [28]. Denbinobin alone also inhibits NF-κB [29].
Separations 2020, 7, x FOR PEER REVIEW 9 of 15 Contrary to their toxic potential, the inhibitory activity of these extracts did not seem to correlate with their cannabinoid content. To further characterize this finding, we conducted a refined doseresponse experiment with extracts of Magda (ETH_NDC, Figure 5A), Beatriz (ETH_DEC, Figure 5B), Pilar (HEX_DEC, Figure 5C) and Sara (HEX_DEC, Figure 5D) and were able to confirm their potency and toxic range, with all of them inhibiting NF-κB activation in a dose-dependent manner before reaching toxic concentrations. Such effect that was not found when the same plant varieties were subjected to different processing conditions or with pure cannabinoids. Therefore, it is plausible to speculate that these extracts may contain additional molecules capable of blocking NF-κB transcriptional activation, either acting alone or as a part of a phytocomplex. For example, flavonoid Cannflavin A and a phenanthrenequinone derivative, Denbinobin, present in the cannabis variety "Carma", elicit synergistic effects in combination with CBD and CBG to inhibit NF-κB activation [28]. Denbinobin alone also inhibits NF-κB [29]. Hypoxia-inducible factor-1α (HIF-1α) has been identified as a key regulator of the hypoxic response with pleiotropic effects, such as promotion of invasion, angiogenesis, switch to glycolytic metabolism, and up-regulation of cell survival-related molecules [30]. The HIF-1α signaling pathway has been proposed to mediate the neuroprotective role for CB2 receptors in microglia [31]. A single study has reported the ability of CBD to dose-dependently reduce the expression of HIF-1α in glioma cells [32]. Results from our biological screening suggest that among phytocannabinoids, CBDV, CBD and CBG were able to reduce the activation of HIF-1α induced by deferoxamine, with IC50s under 12.5 μg·mL −1 for CBD and CBDV, and 25 μg·mL −1 for CBG (Table 3). Hypoxia-inducible factor-1α (HIF-1α) has been identified as a key regulator of the hypoxic response with pleiotropic effects, such as promotion of invasion, angiogenesis, switch to glycolytic metabolism, and up-regulation of cell survival-related molecules [30]. The HIF-1α signaling pathway has been proposed to mediate the neuroprotective role for CB2 receptors in microglia [31]. A single study has reported the ability of CBD to dose-dependently reduce the expression of HIF-1α in glioma cells [32]. Results from our biological screening suggest that among phytocannabinoids, CBDV, CBD and CBG were able to reduce the activation of HIF-1α induced by deferoxamine, with IC50s under 12.5 µg· mL −1 for CBD and CBDV, and 25 µg·mL −1 for CBG (Table 3). Additional studies would be required to confirm and better characterize this finding. As for the extracts, ETH extracts were clearly more potent than HEX, with a total of nine ETH extracts found active at inhibiting HIF-1α ( Figure 4, Table 3). Among these, five NDC extracts presented similar potencies, with IC50s in the 25-50 µg·mL −1 range, irrespective of their major cannabinoid component (either CBDA, THCA or both). However, this observation did not correlate with the reported potency for acid cannabinoids in this assay, which were not active at the highest concentration tested (Table 3). This may be suggestive of the presence of non-cannabinoid active molecules, selectively extracted by the polar solvent ethanol and not by the non-polar solvent hexane, occurring in five different cannabis varieties (Moniek, Magda, Mati, Pilar and Theresa). Several flavonoids have been shown to actively inhibit HIF-1a transcription [33]. Flavonoids are more polar than cannabinoids and, therefore, more likely to be extracted by ethanol than hexane, thus providing a plausible explanation as to why ETH extracts from the same cannabis cultivar are active, but not their HEX counterparts. On the other hand, the degree of inhibition displayed by the remaining four ETH_DEC extracts seems to correlate with their cannabinoid content, dominated by the presence of CBD, CBDV and CBG. However, the reason why their HEX_DEC counterparts were either not active or toxic remains unclear. Theresa and Mati varieties produced ETH_NDC and ETH_DEC extracts active at inhibiting HIF-1a with the ETH_DEC IC50 of <25 µg·mL −1 , more active than the ETH_NDC with a IC50 between 25-50 µg·mL −1 Further studies are currently undergoing to help clarifying this finding by employing more powerful analytical and statistical techniques, such as quantitative time-of-flight (QTOF) detection and principal component analysis (PCA), similarly to what recently reported by other authors [25].
Signal transducer and activator of transcription 3 (STAT3), a member of the STAT families, is involved in regulating essential cellular functions in cancer, such as proliferation, differentiation, survival, invasion, angiogenesis, and metastasis [34]. STAT3 activation can be effectively modulated by several key receptors of the endocannabinoid system such as CB1 [35,36], CB2 [37] and G protein-coupled receptor 55 (GPR55), which has been postulated as a novel cannabinoid receptor [38]. Further, THC induces trophoblast dysfunction in pregnant mice by suppressing STAT3 signaling [39]. Indeed, our results show how THC nearly abolished STAT3 activation at the lower dose tested (5 µg·mL −1 ). All other cannabinoids tested were also able to suppress the activation of STAT3 induced by LPS except for CBGA ( Figure 6). Accordingly, all PHYRESCEL entries resulted active at, at least, one non-toxic dose, with DEC extracts being about five-times more potent than NDC extracts ( Figure 4). However, major differences between solvents were not observed, probably due to the efficacy of the cannabinoid components present in the extracts. While remarkable, more directed studies will be required to confirm and better characterize this inhibitory effect displayed by cannabinoids over LPS-induced STAT3 activation.

Conclusions
Here we disclose the preparation and preliminary characterization of PHYRESCEL, a botanical drug library containing 40 cannabis extracts with varying cannabinoid content. By screening the pharmacological activity of PHYRESCEL in vitro, we were able to pin-point several cannabis extracts that differentially inhibited the transcriptional activation of NF-κB, HIF-1α and STAT3, three molecular targets known to mediate cellular responses to inflammation and hypoxia. The efficacy of isolated minor cannabinoids at inhibiting these transcription factors is also reported. Our results support the notion that cannabis extracts possess differential pharmacological properties compared to isolated cannabinoid molecules. The composition and pharmacological activity of different cannabis extracts obtain from a single cultivar differed largely depending on the organic solvent used and the subsequent decarboxylation process. Although the translation value of these findings is unclear, we expect these functional coordinates to aid in the identification of novel, natural modulators of cellular responses to pro-inflammatory and hypoxic stimuli present in certain varieties of Cannabis sativa L. developed by Phytoplant Research.
Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, Table S1: Cannabinoid composition of all extracts determined by HPLC-DAD. Concentrations are expressed as mean ± standard deviation.

Conclusions
Here we disclose the preparation and preliminary characterization of PHYRESCEL, a botanical drug library containing 40 cannabis extracts with varying cannabinoid content. By screening the pharmacological activity of PHYRESCEL in vitro, we were able to pin-point several cannabis extracts that differentially inhibited the transcriptional activation of NF-κB, HIF-1α and STAT3, three molecular targets known to mediate cellular responses to inflammation and hypoxia. The efficacy of isolated minor cannabinoids at inhibiting these transcription factors is also reported. Our results support the notion that cannabis extracts possess differential pharmacological properties compared to isolated cannabinoid molecules. The composition and pharmacological activity of different cannabis extracts obtain from a single cultivar differed largely depending on the organic solvent used and the subsequent decarboxylation process. Although the translation value of these findings is unclear, we expect these functional coordinates to aid in the identification of novel, natural modulators of cellular responses to pro-inflammatory and hypoxic stimuli present in certain varieties of Cannabis sativa L. developed by Phytoplant Research.