Use of 13C-qNMR Spectroscopy for the Analysis of Non-Psychoactive Cannabinoids in Fibre-Type Cannabis sativa L. (Hemp)

Cannabis sativa L. is a dioecious plant belonging to the Cannabaceae family. The discovery of the presence of many biologically-active metabolites (cannabinoids) in fibre-type Cannabis (hemp) has recently given rise to the valorisation of this variety. In this context, the present study was aimed at the multi-component analysis and determination of the main non-psychoactive cannabinoids (cannabidiol, cannabidiolic acid, cannabigerol and cannabigerolic acid) in female inflorescences of different hemp varieties by means of 13C quantitative nuclear magnetic resonance spectroscopy (qNMR). The method proposed here for the first time for the determination of cannabinoids provided reliable results in a competitive time with respect to the more consolidated HPLC technique. In fact, it gave sufficiently precise and sensitive results, with LOQ values lower than 750 μg/mL, which is easily achievable with concentrated extracts, without affecting the quality of 13C-qNMR spectra. In conclusion, this method can be considered as a promising and appropriate tool for the comprehensive chemical analysis of bioactive cannabinoids in hemp and other derived products in order to ensure their quality, efficacy and safety.


Introduction
Cannabis sativa L. was one of the first cultivated plants, and it was further adapted by humans to a variety of needs: technical (textiles, construction and paper industries), nutritional (food), medicinal and therapeutic, since it contains over 150 bioactive phytocannabinoids [1]. It is an annual dioecious plant belonging to the Cannabaceae family and it growths almost in every developed country of the world [2]. Taxonomic studies have raised a critical issue, owing to the huge variability of specimens within the same genus but, in the scientific community, a monotypic classification is currently preferred [3]. According to this agreement, one single species (C. sativa) has been supposed, and several chemotypes based on the specific cannabinoid profile have been identified [4]. The two principal phenotypes of interest are the drug-type and the fibre-type Cannabis, also known as "hemp". The first is rich in psychoactive ∆ 9 -tetrahydrocannabinol (∆ 9 -THC), and it is used for medicinal or in complex mixtures with a reduced time of analysis and a lower solvent usage, compared with the HPLC technique [28,29].
In the light of all the above, the aim of this study was the development of a reliable 13 C-qNMR method for the comprehensive characterisation and determination of the main non-psychoactive cannabinoids (CBD, CBDA, CBG, CBGA) in eight different hemp varieties. The obtained results were compared with those provided by a previously developed method based on high-performance liquid chromatography (HPLC) [7,8,30]. The 13 C-qNMR method developed in this study for the first time produced satisfactory results in a competitive time with respect to the most widespread and consolidated technique for cannabinoid analysis (HPLC), thus demonstrating to be suitable for a multi-component analysis of hemp extracts. This technique could also be useful to gain a holistic view of the composition and chemical profile of different hemp varieties, and it can be applied for the quality control of both the plant material and derivatives.

NMR Spectroscopic Data of Extracts from Hemp Inflorescences
Typical 1 H, 13  in complex mixtures with a reduced time of analysis and a lower solvent usage, compared with the HPLC technique [28,29].
In the light of all the above, the aim of this study was the development of a reliable 13 C-qNMR method for the comprehensive characterisation and determination of the main non-psychoactive cannabinoids (CBD, CBDA, CBG, CBGA) in eight different hemp varieties. The obtained results were compared with those provided by a previously developed method based on high-performance liquid chromatography (HPLC) [7,8,30]. The 13 C-qNMR method developed in this study for the first time produced satisfactory results in a competitive time with respect to the most widespread and consolidated technique for cannabinoid analysis (HPLC), thus demonstrating to be suitable for a multi-component analysis of hemp extracts. This technique could also be useful to gain a holistic view of the composition and chemical profile of different hemp varieties, and it can be applied for the quality control of both the plant material and derivatives.      The spectra of samples and standard compounds were compared, and the results led to the assignments for CBD, CBG, CBDA and CBGA shown in Figure 4. The chemical shifts in deuterated chloroform for 1 H and 13 C are shown in Table 1-2, expressed as ppm related to tetramethylsilane (TMS).   The spectra of samples and standard compounds were compared, and the results led to the assignments for CBD, CBG, CBDA and CBGA shown in Figure 4. The chemical shifts in deuterated chloroform for 1 H and 13 C are shown in Tables 1 and 2, expressed as ppm related to tetramethylsilane (TMS).  The spectra of samples and standard compounds were compared, and the results led to the assignments for CBD, CBG, CBDA and CBGA shown in Figure 4. The chemical shifts in deuterated chloroform for 1 H and 13 C are shown in Table 1-2, expressed as ppm related to tetramethylsilane (TMS).      * Signals with T 1 < 2 s used for quantification. ** Signal with T 1 > 2 s used for quantification. T 1 was measured, when possible, by using standard compounds.

NMR Spectroscopic Data of Extracts from Hemp Inflorescences
As shown in Figure 1, the protonic spectrum presents a huge number of overlapped peaks, thus requiring an extensive use of deconvolution for signal integration, which can introduce some accuracy issues. To overcome this problem, in this work, 13 C-qNMR was evaluated for the first time for the determination of cannabinoids in hemp extracts [26]. As shown in Figure 2 and Table 2, carbon signals are better resolved and easier to deal with for quantitative purpose. In this context, in these experimental conditions evident signal doubling, due to conformational equilibria, and line broadening did not occur. In particular, all the signals used for CBD and CBG quantification showed a measured average half width of 1.25 Hz (maximum 1.5 Hz). CBDA and CBGA signals were slightly broader; in this case, the average half width was 3.3 Hz (maximum 4 Hz). This line broadening effect did not invalidate or affect the quantification of these compounds, considering that the differences between qNMR and HPLC data are limited and not significant (Table 3).  On the other hand, it is well known that 13 C-qNMR presents some complexities, which often makes the use of this technique quite problematic. In particular, the major issues related to 13 C-qNMR are: 1.
low sensitivity, due to the lower 13 C isotopic abundance (1%) in comparison to 1 H; 2.
carbons typically have long T 1 relaxation time; in this way, signals could not reach their full intensity; 3. the 13 C signal intensity is affected by the Nuclear Overhauser Enhancement (NOE) effect, that depends on 13 C type (from primary to quaternary).
In order to reduce the impact of these issues, the following measures were applied. With regards to the sensitivity problem, it was possible to acquire spectra of concentrated hemp extracts without disregarding the spectrum quality in terms of resolution, intensity and resonance shifting, due to the matrix effect. To overcome the issue related to 13 C long relaxation time, T 1 for each carbon of standard compounds was tentatively measured. Some of the nuclei did not provide results in an acceptable time, due to the molecule structure or to the low amount of standard available (CBDA and CBGA). For this reason, only signals with T 1 < 2 s were initially selected for quantification (marked with * in Table 2). Furthermore, some characteristic signals with T1 > 2 s were also tentatively used for the two acidic compounds (marked with ** in Table 2). Finally, a pulse delay (D 1 ) equal to 10 s (5*T 1 ) was used. This parameter allowed us to obtain the full relaxation in an acceptable time, thus making possible to perform the analysis in a competitive time (47 min) with respect to the most common HPLC technique. The last question was solved by lowering to zero the proton decoupling power.

13 C-qNMR Method Validation
The Concentration Conversion Factor (CCF), obtained from the analysis of standards and used by the software to calculate the concentration of compounds in samples, was measured in the range of expected concentrations, by diluting a CBD standard solution. The CCF value was constant (0.71605 ± 0.00004) in all the considered concentration ranges, thus assuring a good linearity (r 2 = 0.9998). The method proposed here provided also a good intra-and inter-day precision, with percentage relative standard deviation (%RSD) values of 0.8 and 1.5%, respectively. With regards to the limit of detection (LOD), in these experimental conditions the lowest concentrations which presented clearly distinguishable signals from the spectral background were 0.6, 0.5, 0.6 and 0.5 mM for CBD, CBDA, CBG and CBGA, respectively, corresponding to LOD values of 188, 180, 190 and 180 µg/mL in the NMR tube. The lowest quantification limits (LOQ), by applying the below mentioned S/N restriction, corresponded to 2.11, 2.15, 2.10 and 2.18 mM for CBD, CBDA, CBG and CBGA, respectively, with relative LOQ values of 660, 746, 664 and 746 µg/mL in the NMR tube.

Quantitative Analysis of Hemp Inflorescences
Quantitative results of hemp inflorescences obtained by 13 C-qNMR are shown in Table 3, compared to those obtained by HPLC [7,8,30]. Data are expressed as mean (mg/g) ± standard deviation (SD), dry weight. These values correspond to the overall means of three replicates for all the considered signals for each compound. The results provided by 13 C-qNMR and HPLC are comparable, without meaningful differences. It should also be considered that the signals tentatively used for CBDA and CBGA with T 1 > 2 s (** in Table 2) provided very similar quantitative results as well as those with the lower relaxing times, thus demonstrating the method's ruggedness. In particular, 13 C-qNMR seems to slightly underestimate the CBD and CBG content with respect to HPLC, since this situation occurs for every cultivar. The mean recovered amounts of CBD and CBDA were significantly higher respect those of CBG and CBGA; this finding is supported by other studies, in which the measured concentrations for these cannabinoids were undetectable or very low [8,30]. In the majority of samples (Antal, Carmagnola, China, Fibrante, Futura and Santhica), the acidic compounds were more concentrated with respect to their neutral counterparts, demonstrating that the extraction method is able to limit the decarboxylation process. Codimono was the cultivar which provided the highest content of CBD, while Futura was the one with the highest amount of CBDA. With regards to CBG, only one hemp variety (Santhica) showed a concentration over the LOQ value. CBGA contents were determined instead for Antal, Carmagnola, Futura and Santhica, with the latter two varieties being the richest ones.
The 13 C-qNMR method here proposed for the quantification of non-psychoactive cannabinoids in hemp provided reliable results in comparison to the most commonly and consolidated HPLC technique. The validation confirmed the ruggedness of this qNMR approach, which was found to be precise and sufficiently sensitive. In conclusion, this 13 C-qNMR method is suitable and advantageous for a multi-component analysis of cannabinoids in hemp extracts. This technique could be useful to gain a holistic view of the composition and chemical profile of different hemp varieties, and it could be applied in the quality control of both the plant material and its derivatives.

Hemp Plant Material
Eight samples of fibre-type female inflorescences were analysed in this study, including Antal, Carma, Carmagnola, China, Codimono, Fibrante, Futura and Santhica. These samples (about 100-500 g of dry material each), belonging to different breeding lines, were cultivated under the same agronomic conditions and they were kindly provided by Dr. Gianpaolo Grassi of the research centre CREA-CIN (Rovigo, Italy). Each sample was certified for a content of ∆ 9 -THC below 0.2% (w/w). All the samples considered in this study were approved for commercial use by the European Union [31].

Sample Preparation
Hemp inflorescences were manually separated from twigs and seeds. After this procedure, the samples were stored at +4.0 • C until analysis. Cannabinoids were extracted from hemp inflorescences by means of dynamic maceration, as previously described in detail [7,8]. A weighed amount of each sample (1 g) was extracted with 20 mL of absolute EtOH at room temperature for 15 min, under magnetic stirring. The solution was then paper filtered and the residue was extracted with the same procedure twice more with 20 and 10 mL of same solvent, respectively. The filtrates were then combined and adjusted to 50 mL with EtOH in a volumetric flask. The extraction procedure was repeated twice for each sample.
To prepare NMR tubes, the extracts were evaporated under vacuum at 30 • C and the residue was dissolved in 1 mL of CDCl 3 (0.03% TMS); finally, 600 µL of this solution were transferred to a WILMAD ® NMR tube, 5 mm, Ultra-Imperial grade, 7 in. L, 528-PP (Sigma-Aldrich, Milan, Italy).

NMR Spectroscopy and Spectra Acquisition Procedures
To characterize samples 1 H-NMR, 13 C-NMR, two-dimensional 1 H-13 C Heteronuclear Multiple-Bond Correlation (HMBC) and 1 H-13 C Heteronuclear Single Quantum Coherence (HSQC) spectra were acquired with a Bruker FT-NMR Avance III HD 600 MHz spectrometer (Bruker Biospin GmbH Rheinstetten, Karlsruhe, Germany). All the experiments were performed at 300 K and non-spinning. The assignments for the major compounds were carried out by using standard compounds and by comparison of 1 H-NMR and 13 C-NMR spectra with literature data. After introducing the sample into the probe, at least 5 min were required to achieve a thermal equilibrium. Afterwards, the magnetic field was locked, the probe head was tuned and matched and, finally, the sample was shimmed. To assure the highest reproducibility, all these procedures were automatically executed. The correct 90 • pulse was calibrated for protonic and bi-dimensional spectra with the Bruker AU program "pulsecal" and the receiver gain was set.

13 C-qNMR Procedure
Quantitative NMR was performed by using the Concentration Conversion Factor (CCF) method, implemented in Mnova®12.0 (Mestrelab Research, S.L., Santiago de Compostela, Spain) software. A CBD standard solution (1 mg/mL) was used as the external reference compound. The CCF provided by this analysis was verified and corrected by using a commercial standard of 40% dioxane in benzene-d 6 (Bruker Biospin GmbH Rheinstetten, Karlsruhe, Germany). This solution is normally used for 13 C sensitivity test; in our case, the signal at 66.69 ppm (corresponding to the four equivalent carbons) was used for the quantification. In order to determine the correct delay time (D 1 ), T 1 was measured for both dioxane and the standard compounds, when possible. In the case of acidic compounds, the T 1 of the signals related to the presence of carboxylic groups was directly estimated in the sample richest in CBDA and CBGA. For T 1 determination, the Bruker Sequence "T 1 IR" was employed. The following parameters were applied: a list of 8 increasing delay times (from 0.2 to 20 s); delay time, 20 s; number of scans, 32; total acquisition time, 2 h. A D 1 time equal to 10 s (5*T 1 ) was set to ensure the nuclei complete relaxation during acquisition. Moreover, only signals presenting a signal to noise ratio (S/N) ≥ 50 were used for quantification. An exponential window function, with a line broadening (lb) equal to 0.3, was selected for spectra transformation. Each spectrum was calibrated according to TMS signal and then an automatic zero order phase and base line correction were applied, before peak integration. The quantification tool of Mnova requires a multiplet analysis for the integration; therefore, after the initial spectra processing, a manual multiplet analysis was performed and the peak area of signals belonging to the target compounds were compared to the area of those generated by the CBD standard solution, whose spectrum was acquired under the same conditions. The results were then corrected with the correction factor obtained from the dioxane standard solution. The resonances suitable for quantification are marked with * in Table 1. In general, more than one signal should be integrated for each compound. In the cases of CBD and CBDA, eight carbons having the required characteristics were chosen and they were used to calculate the average concentration of related compounds. For CBG and CBGA, six resonances were selected. In addition, both CBDA and CBGA showed three well resolved peaks, in the region ranging from 160 to 180 ppm, related to the presence of the carboxylic group on the aromatic ring, which were also used for quantification and they are marked with ** in Table 1. All the acquisitions were performed in triplicate.

HPLC Analysis
The HPLC analysis was intended to verify 13 C-qNMR results and it was carried out by using the previously validated method [7,8,30].

Method Validation
Since that the validation procedure for the extraction and HPLC analysis of hemp extracts has already been discussed [7,8,30], the attention was focused on the 13 C-qNMR method in the present work. Linearity was investigated by analyzing both standard solutions at different concentrations (from 2 to 100 mM) and a sample prepared with different dilution factors. Linearity was evaluated by calculating the CCF of the selected signals for quantitative analysis in the aforementioned samples and verifying its constancy inside the concentration range. Precision was assessed by analyzing the same extract five times for three different days. The LOD for the target compounds (CBD, CBDA, CBG and CBGA) was determined by gradually diluting a sample containing measurable amounts of the four molecules, until a S/N ratio equal to 3 was reached. Since only signals with a S/N ratio > 50 were considered to be suitable for quantification, the LOQ value was calculated by considering the area of a signal with a S/N equal to 49.