Talanta 228 (2021) 122249

Contents lists available at ScienceDirect

Talanta
journal homepage: www.elsevier.com/locate/talanta

In-depth cannabis fatty acid profiling by ultra-high performance liquid

chromatography coupled to high resolution mass spectrometry
Susy Piovesana a, Sara Elsa Aita a, Giuseppe Cannazza b, c, Anna Laura Capriotti a, *,
Chiara Cavaliere a, Andrea Cerrato a, Paolo Guarnaccia d, Carmela Maria Montone a,
Aldo Laganà a, c
a
Department of Chemistry, Sapienza University of Rome, Piazzale Aldo Moro 5, 00185, Rome, Italy
b
Department of Life Sciences, University of Modena and Reggio Emilia, Via Giuseppe Campi 287, 41125, Modena, Italy
c
CNR NANOTEC, Campus Ecotekne, University of Salento, Via Monteroni, 73100, Lecce, Italy
d
Dipartimento di Agricoltura, Alimentazione e Ambiente, University of Catania, 95123, Catania CT, Italy

A R T I C L E I N F O A B S T R A C T

Keywords: Industrial hemp (Cannabis sativa L.) represents an important plant, used for a variety of uses including phar­
Fatty acids maceutical and nutraceutical purposes. As such, a detailed characterization of the composition of this plant could
Hemp inflorescence help future research to further exploit the beneficial effects of hemp compounds on the human health. Among the
Liquid chromatography
many compounds of hemp, fatty acids represent an interesting class of minor components, which has been
High resolution mass spectrometry
Software-assisted analysis
overlooked so far. In this work, an untargeted approach based on liquid-chromatography coupled to a high-
Quantitative analysis resolution mass spectrometry and a dedicated structure-based workflow for raw data interpretation was
employed for the characterization of fatty acids from hemp inflorescences. A simple method, without any
chemical derivatization, was developed for extraction and characterization of fatty acids leading to the tentative
identification of 39 fatty acid species in the five hemp samples. A quantitative analysis on the untargeted data
was initially performed, using peak areas as surrogate of analyte abundance for relative quantitation. Five fatty
acids resulted the most abundant in all hemp samples, with ca. 90% of the total peak area. For these compounds a
targeted quantitative method was validated, indicating that the most abundant ones were linolenic acid
(1.39–7.95 mg g-1) and linoleic acid (1.04–7.87 mg g-1), followed by palmitic acid (3.74–6.08 mg g-1), oleic acid
(0.91–4.73 mg g-1) and stearic acid (0.64–2.25 mg g-1).

1. Introductionintroduction characterized in detail by our research group [6,7] due to their biolog­
ical importance in plant tissues [8–10]. Other minor components, such
Industrial hemp (Cannabis sativa L.) is a versatile herbaceous crop, as fatty acids (FAs), amino acids, proteins, sugars, vitamins, and pig­
which is widely used for a variety of applications, from the production of ments, are also present in hemp inflorescence and are also worth to be
fabrics to nutraceutical and pharmaceutical purposes [1]. To date, over characterized for their potential benefit on human health, due to both
500 compounds have been isolated from hemp inflorescence [2,3] and synergic and antagonist interactions [2,3].
more than 150 of them are phytocannabinoid compounds [4]. Recently, In this context, FAs, which are classified according to the number of
121 phytocannabinoids were simultaneously identified by ultra-high double bonds in their carbon chain into saturated fatty acids (SFAs),
performance liquid chromatography (UHPLC) coupled with monounsaturated fatty acids (MUFAs), and polyunsaturated fatty acids
high-resolution mass spectrometry (HRMS) and a dedicated (PUFAs), are particularly interesting for human health [11]. FAs can
structure-based workflow for MS data analysis [5]. Another major prevent different chronic inflammatory processes and diseases, such as
chemical class in hemp is that of terpenes, which are responsible for the cancer, hypertension, cardiovascular diseases, rheumatoid arthritis, and
odor and flavor of the different cannabis strains and possess potent neurodegenerative processes [12]. To date, while several papers have
anti-cancer, anxiolytic, and immunostimulant properties [3]. Some less investigated the composition of FAs in hemp seed oil [13–16] and
abundant compounds include polyphenols phospholipids, which were highlighted their capabilities in reducing the risk of cardiovascular

* Corresponding author. Department of Chemistry, Università di Roma, “La Sapienza” Piazzale Aldo Moro 5 00185, Rome, Italy.
E-mail address: [email protected] (A.L. Capriotti).

https://doi.org/10.1016/j.talanta.2021.122249
Received 19 January 2021; Received in revised form 19 February 2021; Accepted 22 February 2021
Available online 26 February 2021
0039-9140/© 2021 Published by Elsevier B.V.
S. Piovesana et al. Talanta 228 (2021) 122249

disease [17] and in the prevention of inflammation [18], only a few 2.2. Preparation of stock solutions, working standard solutions, and
papers were focused on the comprehensive characterization of FAs in calibration mixtures
hemp inflorescences. ElSohly and his co-workers reported in a review
article the identification of 23 FAs [2], recently 18 terpenoid alcohols Stock (10 mg mL− 1) and working solutions (1 mg mL− 1) in methanol
and FAs were identified by comprehensive two-dimensional gas chro­ were prepared for each compound and stored at − 20 ◦ C. For the five
matography (GC × GC) coupled to both low- and high-resolution most abundant FAs, the quantitative analysis was done. The calibration
time-of-flight mass spectrometry and 27 lipids, including 16 FAs and mixture was prepared from the individual working solutions at a con­
derivatives were identified by GC–TOF/MS analysis [19,20]. GC-MS is, centration of 0.025 mg mL− 1 for oleic acid and palmitic acid and 0.060
in fact, a routine analytical platform in FA profiling, and it has been mg mL− 1 for linoleic, linolenic, and stearic acid. Mixtures were stored at
considered the gold standard in FA analysis, especially for its capability − 20 ◦ C. Eight-point calibration curves were constructed in the range of
to obtain trans and cis isomer separation of unsaturated FAs. However, 0.25–47.00 μg mL− 1.
GC-MS hyphenated technique is extremely laborious since it requires
chemical derivatization due to the non-volatile nature and low ther­
mostability of these compounds [21,22]. In the last years, the use of 2.3. Hemp inflorescence extraction
liquid chromatography (LC) coupled to MS in the analysis of FAs has
increased [23–25]. LC is an appealing alternative to GC and it is suitable The analyzed industrial hemps were three monoecious varieties,
for both polar and nonpolar, nonvolatile or thermounstable compounds. namely Futura 75, Felina 32 and Fedora 17, and two dioecious varieties,
However, there are drawbacks in this approach as well, in the relatively namely Kompolti and Finola. Felina, Futura and Kompolti were culti­
limited ionization efficiency for direct detection of FAs, which are vated in Salerno (Consorzio Canapa del Sud, Campania), Fedora was
detected in the negative ionization mode. Other drawbacks include that cultivated in Licodia Eubea (Azienda Agricola Tinnirello, Sicily) and
the best chromatographic shapes are obtained with acid modifiers in the Finola was cultivated by Azienda Agricola Giovanna Brogna (Cassano
mobile phase, but at the same time they suppress analyte ionization allo Jonio, Calabria).
[26]. Some chemical derivatization methods were described to improve Inflorescences were gathered at a full-flowering stage in the middle
the LC separation and MS detection in couplings, and allowed to reach of May 2019. The inflorescences were mashed, freeze-dried by a Heto
resolutions similar to the one of the GC technique but with a most PowerDry LL1500 (Thermo Fisher), finely ground in a mortar, and
effective sensitivity [27,28]. Nevertheless, also in this case, laborious stored at − 20 ◦ C until use. FAs were extracted following a previously
extraction and/or pre-separation procedures were required before optimized method for polar lipids in hemp inflorescences [7].
derivatization to avoid hydrolysis of esterified FAs from other lipid Briefly, 1 g of each sample was extracted three times using 20 mL of
classes. An approach to avoid derivatization and to improve the frag­ MeOH/CHCl3 (1:1, v/v), under vortex agitation for 30 min; the super­
mentation behavior of unsaturated FAs in electrospray ionization natants were isolated by centrifugation for 20 min at 3000×g, pooled
(ESI)-tandem MS (MS/MS) uses the post-column addition of metal cat­ and evaporated to dryness by an IKA RV 8 rotary evaporator (IKA-Werke
ions, like lithium, for double bond position determination [29]. As far as GmbH & Co. KG, Staufen, Germany). The residue was dissolved in 6 mL
LC stationary phases are concerned, free FAs are separated by reversed of MeOH/H2O/CHCl3 (80:15:5, v/v), with 5 mmol L− 1 H3PO4. All ex­
phase, especially C18 [30]. Recently, Kamphorst et al. have demon­ periments were carried out in duplicate.
strated that the use of the less hydrophobic C8 stationary phase assured
the elution of FAs up to 36 carbon atoms [24], while Hellmuth et al.
developed a simple, rapid, and low-cost analytical platform based on 2.4. Ultra-high performance liquid chromatography-HRMS analysis
diphenyl stationary phase chromatography coupled to triple quadrupole
MS/MS detection for the identification of FA species [31]. The UHPLC Vanquish binary pump H (Thermo Fisher Scientific,
In this work, detailed profiling of the FA content of 5 different in­ Bremen, Germany), equipped with a thermostated autosampler and a
dustrial hemp varieties, grown in different geographic regions, was thermostated column compartment, was used for the analysis of FAs in
carried out to extend the knowledge on hemp inflorescence composition. hemp inflorescence samples. Twenty μL of each sample was injected on a
In particular, for the first time a simple and rapid untargeted UHPLC- C18 Kinetex EVO chromatographic column (100 × 2.1 mm, 1.7 μm
HRMS/MS method without chemical derivatization was developed. particle size; Phenomenex, Torrance, CA, USA). The column was oper­
Identification of FAs was performed by Compound Discoverer software ated at 300 μL min− 1 and 40 ◦ C. Mobile phases were water (A) and
using a predefined workflow for food analysis. Finally, the method was MeOH/i-PrOH 80:20 (v/v) (B) both containing 0.05% acetic acid and 5
validated for the quantitative analysis of the most abundant FAs tenta­ mmol L− 1 ammonium acetate. The gradient was as follows: 40% B for 2
tively identified by the untargeted analysis. min; 40–99% B in 18 min; 99% B for 15 min; the column was finally re-
equilibrated at 40% B for 9 min. The chromatographic system was
2. Materials and methods coupled to a Q Exactive Q Exactive™ Hybrid Quadrupole-Orbitrap mass
spectrometer (Thermo Fisher Scientific, Bremen, Germany) via a heated
2.1. Chemicals and reagents ESI (HESI) source. The HESI source was operated in the negative polarity
ionization mode with the following parameters: spray voltage 2500 V;
HPLC-grade chloroform, methanol (MeOH), and water used for capillary temperature 320 ◦ C; auxiliary gas was set at 15 arbitrary units
sample pre-treatment were supplied by VWR International (Milan, (a.u.); auxiliary gas heater temperature 400 ◦ C; sheath gas 35 a.u.; S-lens
Italy). Ultra-pure water and isopropanol (i-PrOH) of LC-MS grade were RF level was 100%. HRMS analysis was performed in the range m/z
purchased from Thermo Fisher Scientific (Waltham, MA, USA); LC-MS 200–2000 with a resolution (full width at half maximum, FWHM, m/z
grade methanol was provided by Romil Pure Chemistry (Pozzuoli, NA, 200) of 70,000. Automatic gain control (AGC) target value was 5e5, with
Italy). Ammonium acetate, acetic acid, and FA standard compounds a max ion injection time set of 200 ms. TOP 5 data-dependent acquisi­
were purchased from Merck (St. Louis, MO, USA). Standard compounds tion (DDA) was used, with 35,000 resolution (FWHM m/z 200) for MS/
and sample manipulation were always performed using glass equip­ MS analysis. Higher-energy collisional ionization was performed at 40%
ment, such as Pasteur pipettes, glass syringes and glass tubes, to avoid normalized collision energy, using an isolation window width of 2 m/z
lipid adsorption to plastics and tubes were covered with aluminum foil and AGC target value of 5e5. Dynamic exclusion was set to 6 s. Raw data
to prevent oxidation. files were acquired by Xcalibur software (version 3.1, Thermo Fisher
Scientific). Three technical replicates were performed for each experi­
mental replicate. Runs were performed in the same day.

2
S. Piovesana et al. Talanta 228 (2021) 122249

2.5. Data analysis and fatty acid identification mass analyzers allow resolution values up to 500,000 FWHM at the
expense of the acquisition rate, and consequently, of the data points
Compound identifications were obtained according to level 2a con­ collected for each peak. Therefore, in the analysis of complex samples, a
fidence level in metabolomics analysis [32]. FA identification was car­ compromise has to be found between these two parameters, since the
ried out by Compound Discoverer™ 3.1 (Thermo Fisher Scientific, higher the mass resolution is set, the higher the discrimination obtain­
Waltham, MA, USA) using the Food Research workflow template with able between apparently identical m/z is. Nonetheless, many authors
few modifications. After spectra selection and alignment and compound showed that operating at the highest resolution setting, does not provide
detections, adducts were grouped and blank signals were removed. The a significatively larger number of identified features or a reduced mass
Fill Gaps tool was enabled, to grant a better evaluation of peak areas. error [33,34].
Spectra matching was performed against FooDB and Lipid Maps data­ In this way, a compromise could be achieved between the number of
bases with a mass tolerance of 5 ppm. The Apply Spectral Distance tool, detected features and more than three data points per peak, which is
which provides a ranking for compound annotation based on isotopic necessary for quantitative analysis. A large amount of information can
pattern comparison, was enabled as well as the Apply mzLogic tool, the be retrieved and used for qualitative and quantitative analyses. The first
latter used to rank annotations for unknown compounds based on goal of the work was a qualitative untargeted characterization of the
MS/MS data. hemp inflorescence samples. At this level, a structural assignment is
performed without preselection of the analytes. From this perspective, a
total of 39 FAs was tentatively identified in the 5 different hemp in­
2.6. Method validation for the five most abundant fatty acids
florescences, comprising 21 SFAs, 10 MUFAs, and 8 PUFAs; all 39 FAs
were in common in the 5 analyzed hemp varieties (Table 1). Details on
The analytical method validation concept was based on the FDA
adduct, molecular weight, Δ error in ppm, major product ions, and peak
guidance document for bioanalytical method validation guidance for
area intensity of identified FAs are reported in Supplementary Material
industry (https://www.fda.gov/regulatory-information/search-fda-g
Table S1.
uidance-documents/bioanalytical-method-validation-guidance-indust
Our simple and rapid method, without any sample pretreatment and
ry). Analyte stability was initially checked as freeze and thaw stability
any chemical derivatization, but with a simple widely used extraction
test, short-term temperature stability test, long-term stability test, stock
with a solvent mixture (MeOH/CH3Cl) was able to identify 39 FAs, the
solution stability test, and post-preparative stability test.
highest number ever reported in hemp inflorescence. A comparison with
Validation parameters included selectivity, precision, recovery, and
a recent work showed the superiority of our approach based on LC-MS/
calibration curve of the analytes in spiked samples. Compound identi­
MS analysis for the comprehensive characterization of FAs. The inte­
fication was accepted if the retention time, accurate mass, and MS/MS
grated analytical platform developed by Delgado-Povedano et al. based
fragmentation of the candidate compound in the sample matched the
on LC-MS platform was not able to identify FAs while the GC-MS
ones of the reference standard. The non-spiked hemp pool extract was
approach led to the identification of a total of 27 lipids including only
used to assess the absence of interferents at the retention times of the
16 FAs [20]. There are several possible explanations for the lack of FA
analytes. Accuracy, precision and recovery were determined by spiking
identifications by LC-MS/MS in the paper by Delgado-Povedano et al.
a pool of hemp material before extraction at 3 different concentration
probably due to the solvent used for the extraction of the hemp matrix,
levels (C1 0.025 μg mL− 1; C2 0.25 μg mL− 1; C3 1.25 μg mL− 1). Recoveries
chromatographic conditions and the choice of the resuspension mixture
R (%) for individual analytes were calculated according to:
for the extracted analytes before the LC injection.
R% =
Aset1 − A0
x 100 The qualitative untargeted characterization was complemented with
Aset2 a relative quantitative analysis for comparison of the common FAs
among the 5 hemp varieties. Strategies have been described to obtain
where Aset1 is the area of the sample fortified before extraction, A0 is the quantitative data from untargeted studies, and the use of peak areas is
area of the sample without any spiking (endogenous amount), Aset2 is the most suitable to compare results within a single study, under the
the area of spiked extracts. Precision was assessed as intra-assay preci­ same experimental conditions and for samples acquired very close in
sion (repeatability) and inter-assay precision (intermediate precision) time. Under these conditions changes of the signal of one specific
and the values were expressed as relative standard deviation (RSD). compound can be associated with the change of the concentration of this
Signal suppression or enhancement due to matrix effect was evaluated as compound from sample to sample. Therefore, relative peak areas can be
follows: used to follow trends [35]. This approach was used in this work for
Aset2 − A0 comparison of the relative abundances of FAs. Areas could be calculated
ME% = x 100 using Compound Discoverer after compound tentative identification.
Aset3
The most abundant FAs were PUFAs, which comprised 50.8–62.0% of
where Aset3 is the area of standard in pure solvent. the total FA content, and were followed by SFAs, which accounted for
27.8–39.5%, and MUFAs, which constituted 10.7–21.8% (Fig. 1).
3. Results and discussion Under our experimental conditions, α-linolenic and γ-linolenic acids
could not be distinguished. The retention time of the two acids was very
3.1. Fatty acid profiling in hemp inflorescences close, as such peaks were partially juxtaposed. The more abundant of the
two, based on the composition of hemp oil, is α-linolenic acid, which is
FAs in hemp inflorescence samples were extracted and injected onto ca. 16% of FA composition in hemp seeds, with γ-linolenic also a major
a UHPLC column without any preliminary chemical derivatization or component but significantly less abundant (ca. 3%) [36]. α-Linolenic
purification. This simple sample preparation method is advantageous acid was expected to be the most abundant in the analyzed hemp sam­
over the commonly employed derivatization GC-MS methods. Less ples as well.
sample manipulation is needed when derivatization is avoided, while
the use of UHPLC coupled to HRMS in DDA provided a sensitive 3.1.1. Saturated fatty acids
detection and high confidence identification. In this work, the untar­ Palmitic acid was the predominant SFA in all the examined hemp
geted characterization of FAs in hemp inflorescences was done at high inflorescences. Fedora 17 and Finola contained the largest amount
resolution for the full scan acquisition, as recently suggested by Najdekr (around 23%), while Felina 32 had the smallest percentage (14.7%) of
et al. [33], at 70,000. In fact, this resolution value is necessary to this acid. Stearic acid, ranging from 5.5% to 7.9%, was the second most
retrieve the highest amount of information. The most advanced Orbitrap abundant SFA in all varieties. Arachidic acid, a plant metabolite, was

3
S. Piovesana et al. Talanta 228 (2021) 122249

Table 1
Tentatively identified fatty acids (FAs) in the five industrial hemp accessions. For each compound, the related retention time (RT), formula, molecular weight, and area
% are provided. Percentages derived from ratios of individual peak areas and total peak areas of the identified FAs is also shown.
Name RT Formula Monoisotopic %Area_Fedora 17 %Area_Felina 32 %Area_Futura 75 %Area_Finola %Area_Kompolti
Mass

Myristic acid (14:0) 3.9 C14H28O2 228.2086 1.6% 0.3% 0.5% 0.3% 1.1%
Pentadecanoic acid (15:0) 5.0 C15H30O2 242.2244 0.2% 0.1% 0.2% 0.1% 0.1%
Hexadecadienoic acid (16:2) 3.5 C16H28O2 252.2087 <0.1% <0.1% <0.1% <0.1% <0.1%
Palmitelaidic acid (16:1) 4.6 C16H30O2 254.2243 0.6% 0.4% 0.5% 0.3% 0.4%
Palmitic acid (16:0) 6.1 C16H32O2 256.2396 22.6% 14.7% 17.6% 22.8% 15.3%
Margaric acid (17:0) 7.3 C17H34O2 270.2556 1.1% 0.5% 0.5% 0.6% 0.5%
Stearidonic acid isomer 1 (18:4) 3.2 C18H28O2 276.2087 <0.1% 0.1% 0.2% <0.1% 0.1%
Stearidonic acid isomer 2 (18:4) 3.4 C18H28O2 276.2088 0.1% 0.1% 0.3% <0.1% 0.1%
Linolenic acid (18:3) 4.3 C18H30O2 278.2240 30.8% 14.2% 35.7% 21.8% 12.6%
Linoleic acid (18:2) 5.3 C18H32O2 280.2397 18.7% 36.1% 25.3% 18.9% 34.9%
Oleic acid (18:1) 6.6 C18H34O2 282.2553 9.8% 19.3% 8.8% 15.0% 19.7%
Stearic acid (18:0) 8.6 C18H36O2 284.2711 6.3% 7.0% 5.4% 6.6% 7.9%
Nonadecadienoic acid (19:2) 6.0 C19H34O2 294.2557 0.1% 0.1% 0.1% 0.1% 0.1%
Nonadecenoic acid (19:1) 7.5 C19H36O2 296.2712 0.2% 0.4% 0.5% 1.7% 0.4%
Nonadecanoic acid (19:0) 9.6 C19H38O2 298.2869 0.2% 0.1% 0.1% 0.2% 0.1%
Eicosatrienoic acid (20:3) 6.3 C20H34O2 306.2557 0.1% 0.1% 0.1% 0.1% 0.1%
Eicosadienoic acid (20:2) 7.6 C20H36O2 308.2713 0.1% 0.2% 0.2% 0.2% 0.2%
Gadoleic acid (20:1) 9.2 C20H38O2 310.2869 0.4% 1.0% 0.6% 1.5% 1.1%
Arachidic acid (20:0) 11.4 C20H40O2 312.3025 2.9% 2.2% 1.4% 2.8% 2.5%
Heneicosenoic acid (21:1) 10.1 C21H40O2 324.3025 <0.1% <0.1% <0.1% <0.1% <0.1%
Heneicosanoic acid (21:0) 12.9 C21H42O2 326.3181 0.3% 0.3% 0.2% 0.6% 0.3%
Brassidic acid (22:1) 11.9 C22H42O2 338.3182 0.1% 0.2% 0.2% 0.5% 0.2%
Behenic acid (22:0) 14.2 C22H44O2 340.3336 2.0% 1.5% 0.9% 3.3% 1.4%
Tricosenoic acid isomer 1 (23:1) 13.4 C23H44O2 352.3337 <0.1% <0.1% <0.1% <0.1% <0.1%
Tricosenoic acid isomer 2 (23:1) 12.9 C23H44O2 352.3338 <0.1% <0.1% <0.1% <0.1% <0.1%
Tricosanoic acid (23:0) 15.6 C23H46O2 354.3493 0.4% 0.4% 0.3% 0.6% 0.4%
Nervonic acid (24:1) 14.7 C24H46O2 366.3492 <0.1% 0.1% <0.1% 0.3% 0.1%
Lignoceric acid (24:0) 16.8 C24H48O2 368.3647 1.1% 0.5% 0.3% 1.2% 0.5%
Pentacosenoic acid (25:1) 16.0 C25H48O2 380.3652 <0.1% <0.01% <0.1% <0.1% <0.1%
Pentacosanoic acid (25:0) 18.0 C25H50O2 382.3806 0.1% 0.1% <0.1% 0.1% 0.1%
Cerotic acid (26:0) 19.1 C26H52O2 396.3961 0.1% 0.1% <0.1% 0.3% 0.1%
Heptacosanoic acid (27:0) 19.8 C27H54O2 410.4118 <0.1% <0.1% <0.1% <0.1% <0.1%
Montanic acid (28:0) 21.1 C28H56O2 424.4277 <0.1% <0.1% <0.1% <0.1% <0.1%
Nonacosanoic acid (29:0) 22.0 C29H58O2 438.4431 <0.1% <0.1% <0.1% <0.1% <0.1%
Melissic acid (30:0) 22.8 C30H60O2 452.4588 <0.1% <0.1% <0.1% <0.1% <0.1%
Hentriacontanoic acid (31:0) 22.8 C31H62O2 466.4744 <0.1% <0.1% <0.1% <0.1% <0.1%
Pentatriacontanoic acid (35:0) 25.6 C35H70O2 522.5369 <0.1% <0.1% <0.1% <0.1% <0.1%
Octatriacontanoic acid (38:0) 27.2 C38H76O2 564.5834 <0.1% <0.1% <0.1% <0.1% <0.1%
Tetracontanoic acid (40:0) 28.5 C40H80O2 592.6148 <0.1% <0.1% <0.1% <0.1% <0.1%

Fig. 1. Percentages of tentatively identified FAs in the five industrial hemp accessions according to their degree of unsaturation and their chain length.

also present at a percentage ranging from 1.4% to 2.9%. SFAs with odd 3.1.2. Unsaturated fatty acids (MUFAs and PUFAs)
carbon numbers were also identified at small levels, i.e. pentadecanoic The most abundant MUFA was oleic acid with a low percentage in
acid (C15:0), margaric acid (17:0), and nonadecanoic acid (C19:0). Futura variety (8.8%) and a higher percentage in Kompolti (19.7%). The
quantity of other MUFAs was generally below 0.1%. Regarding PUFAs,
the proportion of PUFAs found in the Cannabis Sativa L. species ranged

4
S. Piovesana et al. Talanta 228 (2021) 122249

from 41.1% in Finola to 62.0% in Futura 75 (Fig. 1). Among the detected multiple reaction monitoring used on traditional triple quadrupole
8 PUFAs, the predominant ones were linoleic and linolenic acids with a instrumentation for targeted analyses. DDA was preferred over PRM as it
percentage ranging from 18.7% to 36.1% and from 12.6% to 35.7%, is much more widespread and requires less method development and
respectively. optimization of parameters. Moreover, the selected analytes are abun­
These data agreed with the profile described for other plants, where dant in the investigated matrix, therefore the sensitivity boost of PRM is
the major unsaturated FAs were three C18 species, namely, oleic, lino­ not required [39]. Finally, scan time with high resolution can become
leic, and linolenic acids, which play regulatory roles in plant defense and very long in PRM compared to DDA, a fact which can reduce the number
are also important economic traits of oil crops. These two C18 species of data points per peak and, in turn, a precise quantitative analysis.
are dietary essential FAs since human beings are unable to bio-
synthesize them. Moreover, oleic, linoleic, and linolenic acids consti­ 3.2.1. Selectivity, precision, and recovery
tute raw materials for the production of biofuels, cosmetics, detergents, Standard compounds were used for comparison of retention times
and pharmaceuticals [37]. and MS/MS spectra. The non-spiked hemp pool extract was used to
Wide variations in the FA composition were shown among all hemp assess the absence of interferents at the same retention times of the
inflorescence species. Regarding the relative abundance of each species, analytes. However, endogenous amounts of the analytes were detected.
for Felina 32 e Kompolti varieties, the order of abundance of the Initially, recoveries were calculated for 12 representative fatty acids,
tentatively identified FAs was 18:2 > 18:1 > 16:0 > 18:3 > 18:0, with subtracting the endogenous quantity (C0). The accuracy of the proposed
linoleic acid presenting around 35% followed by oleic acid at 20% and a method was expressed as average recovery at three spiking levels (C1,
low percentage of linolenic acid (12–14%). Conversely, in Fedora 17 and C2, C3) with related relative standard deviation (Table 2 for the five most
Futura 75 species, the most abundant FA was linolenic acid with a abundant analytes, data on all 12 FAs are reported in Table S2). Assay
percentage of 31% and 36%, respectively. As the aforementioned FAs accuracy ranging from 80% to 109% was achieved.
were the most abundant in all industrial hemp samples, absolute The reproducibility of the method was determined by the intra- and
quantitation and method validation were carried out as described in inter-day precision detection of hemp extracts added with known
section 3.2. quantities of FAs (at intermediate value of fortification C2). Intraday
Felina 32 and Kompolti varieties were very similar in their FA profile accuracies were determined from five parallel experiments within one
with about 30% of SFAs, 20% MUFAs, and 50% PUFAs. Fedora 17 day, and intraday accuracies were determined from three parallel ex­
species presented 40% of SFAs and just 10% of MUFAs. Finola and periments of the sample solution for three consecutive days. The results
Futura 75 presented the lowest and highest abundance of PUFAs with 62 showed that the intra- and inter-day RSDs were lower than 8% and 5%,
and 41%, respectively. respectively (Table 2), indicating that the method was reproducible.
The similarity in Felina 32 and Kompolti FA composition was also
confirmed by the chain length of the identified FAs; about 17% of FAs 3.2.2. Calibration curve and matrix effect
possessed a chain length shorter than 18, with palmitic acid being the Validation of the analytical method ideally requires calibration
most abundant one, 75% consisted of C18 chain FAs, and 7% possessed curves to be built in the same matrix, especially using ESI MS, because
chains longer than 18 carbon atoms. Futura 75 had also a similar dis­ components in the sample can lead to matrix effects such as ionization
tribution, even if the percentage of FAs with chains shorter than 18 is enhancement or suppression. In this work, the analytical validation was
quite higher. Fedora 17 and Finola presented lower concentrations of carried out on a hemp extract pool, and the background subtraction
C18 FAs at about 65%. Fedora 17 presented the highest abundance of method was employed for building calibration curves. Absolute area
FAs with a chain shorter than C18 (about 26%), while Finola presented values from the calibration curves were used to evaluate the matrix ef­
the highest concentration of long and very-long-chain FAs (13.5%). The fect. This was calculated to evaluate the effect of signal enhancement or
similar profile of Felina 32, Futura 75, and Kompolti inflorescence could suppression due to matrix components, without use of deuterated pure
be ascribed to their geographical area of growth rather than to their standards for each analyte. Matrix effect was calculated for all hemp
genotype. Inflorescences are, in fact, largely influenced by the pedocli­ varieties and was found acceptable, within 10% variation (95–110%).
matic conditions of growth. Further studies with larger sample sets Good linearity was achieved for all analytes in the tested concen­
could be carried out to determine the actual role of the individual tration range, and the calibration curve equations were linear with the
variability of each variety rather than their geographical area of growth. correlation coefficient (R2) ≥ 0.9853 (Table 2).
It is worth mentioning that very-long-chain FAs (up to C40) were The results indicated that this method was sensitive to the determi­
identified in all industrial hemp accessions, even at low concentrations. nation of FAs in different hemp varieties, meaning that it could be
Untargeted LC-HRMS analysis, which allows derivatization-free identi­ employed even in biological applications where less sample is available,
fication of FAs, is, therefore, suitable for the characterization of a very and where the identified analytes are present at lower concentrations.
broad range of alkyl chains.
3.3. Absolute quantitation of FAs in the five hemp inflorescence
3.2. Method validation accessions

The untargeted analysis is a powerful method for the qualitative FAs were quantified in the 5 different hemp varieties. Three bio­
characterization of a complex mixture, but one drawback is that quan­ logical replicates were performed for each hemp variety. Values are
titative analysis is not as precise as in a targeted experiment. As such, a shown in Table 3 with the RSD. The observed peaks were assigned to FA
targeted method was validated for quantification of the abundant spe­ species by comparison of the experimental retention time and accurate
cies in the analyzed samples. Following the untargeted analysis, 5 FAs mass to the ones of known standards.
were the most abundant species, i.e. palmitic acid (16:0), stearic acid
(18:0), oleic acid (18:1), linoleic acid (18:2) and linolenic acid (18:3). 4. Conclusions
Absolute quantification is necessary for more reliable quantitative
characterization and to communicate the significance of the results in a The manuscript describes an untargeted characterization of FAs in
clear manner, which can be compared from one laboratory to another industrial hemp inflorescences by UHPLC coupled to HRMS. The
[35]. In this context, no official guidelines are reported for targeted developed method, compared to that based on GC-MS, the gold standard
quantitative analysis by HRMS, though several have been described in for FA analysis, does not require a laborious clean-up procedure or
the literature [38]. Method validation was performed in DDA even if derivatization before characterization and identification. Moreover, the
parallel-reaction monitoring (PRM) is the Orbitrap equivalent to use of UHPLC coupled to HRMS in DDA provided a sensitive detection

5
S. Piovesana et al. Talanta 228 (2021) 122249

Table 2
Extraction recoveries (R%) in matrix of 12 fatty acids at three different concentrations (C1: 0.025 μg mL− 1; C2: 0.25 μg mL− 1; C3 1.25 μg mL− 1).
Adduct Monoisotopic (R ± RSD)% Precision (C2) (%) Calibration range (μg mL− 1) R2
Mass

C1 C2 C3 Intra-day Inter-day

Palmitic acid (16:0) [M − H]- 256.2396 (80 ± 3)% (100 ± 1)% (100 ± 1)% 7% 4% 0.25–12.50 0.9909
Linolenic acid (18:3) [M − H]- 278.2240 (91 ± 2)% (99 ± 1)% (99 ± 1)% 8% 5% 0.95–47.00 0.9987
Linoleic acid (18:2) [M − H]- 280.2397 (98 ± 1)% (100 ± 1)% (106 ± 1)% 8% 5% 0.95–47.00 0.9853
Oleic acid (18:1) [M − H]- 282.2553 (99 ± 1)% (99 ± 1)% (99 ± 1)% 7% 5% 0.60–30.00 0.9909
Stearic acid (18:0) [M − H]- 284.2711 (80 ± 2)% (97 ± 3)% (93 ± 4)% 5% 4% 0.25–12.50 0.9947

Cannazza: Conceptualization, Supervision. A. L. Capriotti: Conceptual­

Table 3
ization, Writing – original draft, Writing – review & editing, Project
Concentration of each analyte in the different hemp accessions (mean of n = 3
administration, Supervision. C. Cavaliere: Supervision, Writing – review
biological replicates and n = 2 technical replicates for each hemp variety).
& editing. A. Cerrato: Data curation, Visualization. P. Guarnaccia: Re­
Analyte Concentration (mg g− 1) ± RSD sources. C. M. Montone: Methodology, Data curation, Visualization. A.
Fedora 17 Felina 32 Futura 75 Finola Kompolti Laganà; : Conceptualization, Funding acquisition, Supervision.

References
Palmitic acid 3.74 ± 4.75 ± 6.08 ± 3.74 ± 5.35 ±
(16:0) 0.02 0.01 0.01 0.02 0.01
[1] H.P.V. Rupasinghe, A. Davis, S.K. Kumar, B. Murray, V.D. Zheljazkov, Industrial
Linolenic acid 2.54 ± 2.16 ± 7.95 ± 1.39 ± 2.02 ±
hemp (Cannabis sativa subsp. sativa) as an emerging source for value-added
(18:3) 0.02 0.04 0.01 0.02 0.06
functional food ingredients and nutraceuticals, Molecules 25 (2020) 4078, https://
Linoleic acid 1.04 ± 7.44 ± 5.27 ± 1.04 ± 7.87 ± doi.org/10.3390/molecules25184078.
(18:2) 0.03 0.02 0.02 0.03 0.02 [2] M.A. ElSohly, D. Slade, Chemical constituents of marijuana: the complex mixture of
Oleic acid 0.91 ± 4.73 ± 2.09 ± 1.60 ± 5.27 ± natural cannabinoids, Life Sci. 78 (2005) 539–548, https://doi.org/10.1016/j.
(18:1) 0.01 0.01 0.01 0.02 0.03 lfs.2005.09.011.
Stearic acid 0.64 ± 1.76 ± 1.39 ± 0.68 ± 2.25 ± [3] C.M. Andre, J.F. Hausman, G. Guerriero, Cannabis sativa: the plant of the thousand
(18:0) 0.01 0.01 0.01 0.01 0.01 and one molecules, Front. Plant Sci. 7 (2016) 1–17, https://doi.org/10.3389/
fpls.2016.00019.
[4] L.O. Hanuš, S.M. Meyer, E. Muñoz, O. Taglialatela-Scafati, G. Appendino,
Phytocannabinoids: a unified critical inventory, Nat. Prod. Rep. 33 (2016)
and high confidence identification.
1357–1392, https://doi.org/10.1039/C6NP00074F.
The method allowed to tentatively identify 39 FA species in the five [5] C.M. Montone, A. Cerrato, B. Botta, G. Cannazza, A.L. Capriotti, C. Cavaliere,
hemp samples. A preliminary quantitative analysis, based on areas for C. Citti, F. Ghirga, S. Piovesana, A. Laganà, Improved identification of
each compound, was then used to assess the composition of each hemp. phytocannabinoids using a dedicated structure-based workflow, Talanta 219
(2020) 121310, https://doi.org/10.1016/j.talanta.2020.121310.
The five most abundant species were palmitic, linolenic, linoleic, oleic [6] A. Cerrato, G. Cannazza, A.L. Capriotti, C. Citti, G. La Barbera, A. Laganà, C.
and stearic acids, for which an absolute quantitative method was M. Montone, S. Piovesana, C. Cavaliere, A new software-assisted analytical
developed and validated. The study contributes to a detailed charac­ workflow based on high-resolution mass spectrometry for the systematic study of
phenolic compounds in complex matrices, Talanta 209 (2020) 120573, https://doi.
terization of minor components of industrial hemp, and is potentially org/10.1016/j.talanta.2019.120573.
suitable for other studies aimed at characterizing FAs qualitatively and [7] M. Antonelli, B. Benedetti, G. Cannazza, A. Cerrato, C. Citti, C.M. Montone,
quantitatively for hemp breeding or applications. S. Piovesana, A. Laganà, New insights in hemp chemical composition: a
comprehensive polar lipidome characterization by combining solid phase
enrichment, high-resolution mass spectrometry, and cheminformatics, Anal.
Declaration of competing interest Bioanal. Chem. 412 (2020) 413–423, https://doi.org/10.1007/s00216-019-02247-
6.
[8] I.O. Minatel, C.V. Borges, M.I. Ferreira, H.A.G. Gomez, C.-Y.O. Chen, G.P.P. Lima,
The authors declare that they have no known competing financial Phenolic compounds: functional properties, impact of processing and
interests or personal relationships that could have appeared to influence bioavailability. Phenolic Compd. - Biol. Act., InTech, 2017, https://doi.org/
the work reported in this paper. 10.5772/66368.
[9] M. Soto-Hernndez, M. Palma-Tenango, M. del R. Garcia-Mateos (Eds.), Phenolic
Compounds - Biological Activity, InTech, 2017, https://doi.org/10.5772/63693.
Acknowledgements [10] A.K. Cowan, Phospholipids as plant growth regulators, Plant Growth Regul. 48
(2006) 97–109, https://doi.org/10.1007/s10725-005-5481-7.
[11] P.C. Calder, Functional roles of fatty acids and their effects on human health,
Funding for this research was provided by Sapienza University of J. Parenter. Enteral Nutr. 39 (2015) 18S–32S, https://doi.org/10.1177/
Rome with the Scientific Research grants, protocol number 0148607115595980.
RM120172B8825C01, and protocol number AR11916B55532026. [12] C. Cantele, M. Bertolino, F. Bakro, M. Giordano, M. Jędryczka, V. Cardenia,
Antioxidant effects of hemp (Cannabis sativa L.) inflorescence extract in stripped
Additional funding was obtained within the UNIHEMP research
linseed oil, Antioxidants 9 (2020) 1131, https://doi.org/10.3390/antiox9111131.
project “Use of iNdustrIal Hemp biomass for Energy and new bio­ [13] C. Leizer, D. Ribnicky, A. Poulev, S. Dushenkov, I. Raskin, The composition of
cheMicals Production” (ARS01_00668) funded by Fondo Europeo di hemp seed oil and its potential as an important source of nutrition,
J. Nutraceuticals, Funct. Med. Foods 2 (2000) 35–53, https://doi.org/10.1300/
Sviluppo Regionale (FESR) (within the PON R&I 2017-2020—Axis
J133v02n04_04.
2—Action II—OS 1.b), grant decree UNIHEMP prot. n. 2016 of July 27, [14] M. Kiralan, V. Gül, S. Metin Kara, Fatty acid composition of hempseed oils from
2018; CUP B76C18000520005). different locations in Turkey, Spanish J. Agric. Res. 8 (2010) 385, https://doi.org/
10.5424/sjar/2010082-1220.
[15] M. Petrović, Ž. Debeljak, N. Kezić, P. Džidara, Relationship between cannabinoids
Appendix A. Supplementary data content and composition of fatty acids in hempseed oils, Food Chem. 170 (2015)
218–225, https://doi.org/10.1016/j.foodchem.2014.08.039.
Supplementary data to this article can be found online at https://doi. [16] J. Deferne, D.W. Pate, Hemp seed oil: a source of valuable essential fatty acids,
J. Int. Hemp Assoc. 3 (1) (1996) 4–7.
org/10.1016/j.talanta.2021.122249. [17] J.A. Fleming, P.M. Kris-Etherton, The evidence for α-Linolenic acid and
cardiovascular disease benefits: comparisons with eicosapentaenoic acid and
Credit author contribution statement docosahexaenoic acid, Adv. Nutr. 5 (2014) 863S–876S, https://doi.org/10.3945/
an.114.005850.

S. Piovesana: Writing – review & editing. S. E. Aita: Investigation. G.

6
S. Piovesana et al. Talanta 228 (2021) 122249

[18] S. Jin, M.-Y. Lee, The ameliorative effect of hemp seed hexane extracts on the [30] I. Losito, L. Facchini, A. Valentini, T.R.I. Cataldi, F. Palmisano, Fatty acidomics:
Propionibacterium acnes-induced inflammation and lipogenesis in sebocytes, PloS evaluation of the effects of thermal treatments on commercial mussels through an
One 13 (2018), e0202933, https://doi.org/10.1371/journal.pone.0202933. extended characterization of their free fatty acids by liquid chromatography –
[19] F.A. Franchina, L.M. Dubois, J.-F. Focant, In-Depth cannabis multiclass metabolite fourier transform mass spectrometry, Food Chem. 255 (2018) 309–322, https://
profiling using sorptive extraction and multidimensional gas chromatography with doi.org/10.1016/j.foodchem.2018.02.073.
low- and high-resolution mass spectrometry, Anal. Chem. 92 (2020) 10512–10520, [31] C. Hellmuth, M. Weber, B. Koletzko, W. Peissner, Nonesterified fatty acid
https://doi.org/10.1021/acs.analchem.0c01301. determination for functional lipidomics: comprehensive ultrahigh performance
[20] M.M. Delgado-Povedano, C. Sánchez-Carnerero Callado, F. Priego-Capote, liquid chromatography–tandem mass spectrometry quantitation, qualification, and
C. Ferreiro-Vera, Untargeted characterization of extracts from Cannabis sativa L. parameter prediction, Anal. Chem. 84 (2012) 1483–1490, https://doi.org/
cultivars by gas and liquid chromatography coupled to mass spectrometry in high 10.1021/ac202602u.
resolution mode, Talanta 208 (2020) 120384, https://doi.org/10.1016/j. [32] E.L. Schymanski, J. Jeon, R. Gulde, K. Fenner, M. Ruff, H.P. Singer, J. Hollender,
talanta.2019.120384. Identifying small molecules via high resolution mass spectrometry: communicating
[21] J.M. Halket, V.G. Zaikin, Derivatization in mass spectrometry—1, Silylation, Eur. J. confidence, Environ. Sci. Technol. 48 (2014) 2097–2098, https://doi.org/
Mass Spectrom. 9 (2003) 1–21, https://doi.org/10.1255/ejms.527. 10.1021/es5002105.
[22] W. Schliemann, C. Ammer, D. Strack, Metabolite profiling of mycorrhizal roots of [33] L. Najdekr, D. Friedecký, R. Tautenhahn, T. Pluskal, J. Wang, Y. Huang, T. Adam,
Medicago truncatula, Phytochemistry 69 (2008) 112–146, https://doi.org/ Influence of mass resolving power in orbital ion-trap mass spectrometry-based
10.1016/j.phytochem.2007.06.032. metabolomics, Anal. Chem. 88 (2016) 11429–11435, https://doi.org/10.1021/acs.
[23] I. Brondz, Development of fatty acid analysis by high-performance liquid analchem.6b02319.
chromatography, gas chromatography, and related techniques, Anal. Chim. Acta [34] G. La Barbera, A.L. Capriotti, C. Cavaliere, S. Piovesana, R. Samperi, R. Zenezini
465 (2002) 1–37, https://doi.org/10.1016/S0003-2670(01)01467-2. Chiozzi, A. Laganà, Comprehensive polyphenol profiling of a strawberry extract
[24] J.J. Kamphorst, J. Fan, W. Lu, E. White, J.D. Rabinowitz, Liquid (Fragaria × ananassa) by ultra-high-performance liquid chromatography coupled
chromatography–high resolution mass spectrometry analysis of fatty acid with high-resolution mass spectrometry, Anal. Bioanal. Chem. 409 (2017)
metabolism, Anal. Chem. 83 (2011) 9114–9122, https://doi.org/10.1021/ 2127–2142, https://doi.org/10.1007/s00216-016-0159-8.
ac202220b. [35] A. Kruve, Strategies for drawing quantitative conclusions from nontargeted liquid
[25] S.-H. Chen, Y.-J. Chuang, Analysis of fatty acids by column liquid chromatography, chromatography–high-resolution mass spectrometry analysis, Anal. Chem. 92
Anal. Chim. Acta 465 (2002) 145–155, https://doi.org/10.1016/S0003-2670(02) (2020) 4691–4699, https://doi.org/10.1021/acs.analchem.9b03481.
00095-8. [36] J.I. Alonso-Esteban, M.J. González-Fernández, D. Fabrikov, E. Torija-Isasa, M. de
[26] L. Xiang, L. Zhu, Y. Huang, Z. Cai, Application of derivatization in fatty acids and C. Sánchez-Mata, J.L. Guil-Guerrero, Hemp ( Cannabis sativa L.) varieties: fatty
fatty acyls detection: mass spectrometry-based targeted lipidomics, Small Methods acid profiles and upgrading of γ-Linolenic acid–containing hemp seed oils, Eur. J.
4 (2020) 2000160, https://doi.org/10.1002/smtd.202000160. Lipid Sci. Technol. 122 (2020) 1900445, https://doi.org/10.1002/ejlt.201900445.
[27] X. Li, A.A. Franke, Improved LC− MS method for the determination of fatty acids in [37] M. He, C.-X. Qin, X. Wang, N.-Z. Ding, Plant unsaturated fatty acids: biosynthesis
red blood cells by LC− Orbitrap MS, Anal. Chem. 83 (2011) 3192–3198, https:// and regulation, Front. Plant Sci. 11 (2020), https://doi.org/10.3389/
doi.org/10.1021/ac103093w. fpls.2020.00390.
[28] W.-C. Yang, J. Adamec, F.E. Regnier, Enhancement of the LC/MS analysis of fatty [38] A. Kaufmann, High-resolution mass spectrometry for bioanalytical applications: is
acids through derivatization and stable isotope coding, Anal. Chem. 79 (2007) this the new gold standard? J. Mass Spectrom. 55 (2020) e4533, https://doi.org/
5150–5157, https://doi.org/10.1021/ac070311t. 10.1002/jms.4533.
[29] F.-F. Hsu, J. Turk, Distinction among isomeric unsaturated fatty acids as lithiated [39] A. Kaufmann, Analytical performance of the various acquisition modes in Orbitrap
adducts by electrospray ionization mass spectrometry using low energy MS and MS/MS, J. Mass Spectrom. 53 (2018) 725–738, https://doi.org/10.1002/
collisionally activated dissociation on a triple stage quadrupole instrument, J. Am. jms.4195.
Soc. Mass Spectrom. 10 (1999) 600–612, https://doi.org/10.1016/S1044-0305
(99)00041-0.