**Hempseed Lignanamides Rich-Fraction: Chemical Investigation and Cytotoxicity towards U-87 Glioblastoma Cells**

**Ersilia Nigro 1,2,**† **, Giuseppina Crescente 1,**† **, Marialuisa Formato 1,**† **, Maria Tommasina Pecoraro <sup>1</sup> , Marta Mallardo 1,2, Simona Piccolella <sup>1</sup> , Aurora Daniele 1,2 and Severina Pacifico 1,\***


Academic Editor: Stefano Dall'Acqua

Received: 18 February 2020; Accepted: 25 February 2020; Published: 26 February 2020 **Abstract:** The weak but noteworthy presence of (poly)phenols in hemp seeds has been long

overshadowed by the essential polyunsaturated fatty acids and digestible proteins, considered responsible for their high nutritional benefits. Instead, lignanamides and their biosynthetic precursors, phenylamides, seem to display interesting and diverse biological activities only partially clarified in the last decades. Herein, negative mode HR-MS/MS techniques were applied to the chemical investigation of a (poly)phenol-rich fraction, obtained from hemp seeds after extraction/fractionation steps. This extract contained phenylpropanoid amides and their random oxidative coupling derivatives, lignanamides, which were the most abundant compounds and showed a high chemical diversity, deeply unraveled through high resolution tandem mass spectrometry (HR-MS/MS) tools. The effect of different doses of the lignanamides-rich extract (LnHS) on U-87 glioblastoma cell line and non-tumorigenic human fibroblasts was evaluated. Thus, cell proliferation, genomic DNA damage, colony forming and wound repair capabilities were assessed, as well as LnHS outcome on the expression levels of pro-inflammatory cytokines. LnHS significantly inhibited U-87 cancer cell proliferation, but not that of fibroblasts, and was able to reduce U-87 cell migration, inducing further DNA damage. No modification in cytokines' expression level was found. Data acquired suggested that LnHS acted in U-87 cells by inducing the apoptosis machinery and suppressing the autophagic cell death.

**Keywords:** *Cannabis sativa* L.; phenylamides; lignanamides; hemp seeds; high resolution tandem mass spectrometry; U-87 glioblastoma cells; cytotoxicity

### **1. Introduction**

Plant foods, thanks to the functionality of their bioactive secondary metabolites, are considered to be both safe and able to promote good health [1], explaining some important targeted effects in humans, and preventing affluence diseases, such as cardiovascular diseases and cancers [2]. Bioactive compounds are also in ancient crops, whose actual recovery further renewed the phytochemical research into the discovery of beneficial substances, which could be present, after food processing, in the daily meals or be exceptionally lost in produced waste and by-products [3]. This is the case of lignans, chemically characterized by a phenylpropanoid core, which are reported to exert numerous biological effects in mammals, including antitumor and antioxidant activities [4]. They act as phytoestrogens and are converted by intestinal microflora into mammalian lignans or enterolignan compounds [5].

Lignans are abundant in dietary sources like whole-grain cereals and legumes. The preventive benefits of some edible seeds, and the increasing intake of no-conventional foods as chia, quinoa, flax, canola and pumpkin seeds could be ascribed to their richness in these compounds [6]. For instance, flaxseed (*Linum usitatissimum* L.) is reported to contain about 75–800 times more lignans than cereal grains, legumes, fruits and vegetables, together with antioxidant flavonols and hydroxycinnamic acids [7], and lariciresinol was one of the main constituents of pumpkin seeds (*Cucurbita pepo* L.) [8].

Hemp seeds (non-drug type of*Cannabis sativa*L.) also contain, beyond proteins, and polyunsaturated fatty acids, bioactive lignan derivatives, known as lignanamides [9–12]. These latter could be found in hemp fruits together with their biosynthetic precursors, namely phenylamides, whose presence is functional in another seed, such as oat seed, which produces avenanthramides (AVAs) with important anti-inflammatory and antiproliferative effects [13]. Until a few years ago, phenylamides and lignanamides seemed to constitute a small group of natural products, whose distribution in plant kingdom was thought to be limited to plants of the Cannabaceae and Solanaceae family. Indeed, lignanamides are also reported from *Mitrephora thorelii* (Annonaceae) and *Corydalis saxicola* (Papaveraceae), and, recently, five pairs of enantiomeric lignanamides were obtained from *Solanum nigrum*, and melongenamides A–D were isolated from the roots of *Solanum melongena* L. [14–16]. Furthermore, new lignanamides and neolignanamides were isolated from *Lycium chinense* [17,18], highlighting that the diversity of these compounds is so far to be really known. Moreover, (±)-sativamides A and B, two pairs of nor-lignanamide enantiomers featuring a unique benzo-angular triquinane skeleton, were isolated from the fruits of *Cannabis sativa*, and were observed to be able to reduce endoplasmic reticulum (ER) stress-induced cytotoxicity in neuroblastoma cells [19].

The ability of lignanamides to display interesting and diverse biological activities, including feeding deterrent activity, insecticidal effects, anti-inflammatory and neuroprotective activity [20,21], addresses the research in new analytical challenges for their ready exploitation from hempseed meal. In this context, starting on the great and renewed interest in hemp seeds and their by-products, as source of essential nutrients, in sufficient amount and ratio to satisfy the dietary human demand, commercial hempseeds underwent ultrasound assisted extraction first with *n*-hexane, and after an oil-like mixture recovery, the obtained by-product was investigated for its polyphenol content. A fraction rich in lignanamides (LnHS) was achieved and chemically profiled through HR-MS/MS tools operating in negative ion detection. Furthermore, taking into account recent literature, which highlights the protective effects of pure lignanamides on central nervous system cell lines [15,21], LnHS was investigated for its anti-cancer properties versus U-87 malignant glioblastoma (GMB) neuroepithelial cells. Indeed, U-87 cells are known to move very fast and to aggregate as clusters, showing rapid migration, and highest invasion ability [22]. As malignant gliomas are the most common primary brain tumors, among which GBM is the most malignant and highly aggressive [23], the aim of this study was to understand the effects of low doses of constituted LnHS fraction on U-87 viability cell line as well-established model of malignant glioblastoma cell line in terms of apoptosis and autophagic cell death. In addition, the effects of LnHS on U-87 colony formation efficacy and cell migration was analyzed. Finally, the ability of LnHS of inducing oxidative and inflammation processes was investigated through evaluation of the expression of Sirt1 and Sirt2, as well as of some pro-inflammatory cytokines. All tests were performed in comparison to non-tumorigenic human fibroblasts (thereafter indicated as HF).

#### **2. Results and Discussion**

The interest in the polyphenols contained in hemp seed was for a long time obscured by its high content in essential polyunsaturated fatty acids, mainly linoleic and α-linolenic acids, though different flavonoid glycosides were identified in cold-pressing hemp seed oil [24,25]. Indeed, after oil recovery, hemp seed meal yet represents an important food polyphenol source. Several phenylamides and lignanamides were previously isolated and structurally identified mainly by means of nuclear magnetic resonance (NMR) spectroscopic tools [20,26]. Herein, in order to get insight into the chemical composition of commercial hempseeds, ultrasound assisted maceration was applied on seeds first ground with a knife mill and crushed into liquid N<sup>2</sup> to better preserve the integrity of the fruit's constituents.

The cryo-crushed matrix, made as a friable powder, preliminarily underwent solid-liquid extraction process using *n*-hexane, obtaining an oil-like extract and the defatted-matrix. This latter was then ultrasound-assisted macerated using ethanol (Figure S1). The alcoholic fraction was further fractionated (please see Section 3) obtaining, among the others, a fraction, hereafter referred to as LnHS, able to strongly modify the morphology of SH-SY5Y cells (Figure S2). Thus, in the consciousness that extraction/fractionation steps are fundamental into defining the chemistry of a phytochemical extract, in order to get insight into this fraction chemical composition, negative HR-MS and HR-MS/MS spectra, as well as spectra by ultraviolet diode array detection (UV-DAD), were acquired. Phenylamides and lignanamides were found to be the main compounds (Tables 1 and 2), whereas flavonol glycosides were the minor constituents (Table 3). Thus, the fraction underwent an extensive cytotoxic screening on U-87 cells, exhibiting a promising behavior in fighting migration and invasion features of these glioblastoma cells.

#### *2.1. HR-MS Analysis: Phenylamides*

Compounds **1**, **2**, **5**, **8**, **11**, **17** and **33** were recognized as phenylamides, conjugates of aliphatic polyamines or arylmonoamines and hydroxycinnamic acids, suggested as defensive plant-specific molecules. Compound **1** was tentatively identified as *N*-caffeoyloctopamine (Figure 1), previously isolated from hempseed cakes in a screening aimed to identify novel arginase inhibitors [27], and further identified among hempseed constituents with potential anti-neuroinflammatory activity [25].

**Figure 1.** (**A**) and (**B**) TOF-MS and TOF-MS/MS spectra for compound **1**; (**C**) proposed fragmentation pathway of the [M − H]<sup>−</sup> ion; (**D**) UV-DAD spectrum. In **C** panel, the theoretical *m*/*z* value is reported below each structure.

The [M − H]<sup>−</sup> ion at *m*/*z* 314.1039 underwent dehydration (likely through octopamine moiety) providing the TOF-MS<sup>2</sup> fragment ion at *m*/*z* 296.0910, or rearranged to give, through 45 Da neutral loss, the fragment ion at *m*/*z* 269.0818.








**Table 2.** *Cont.*




This could be due to (NH<sup>3</sup> + CO) neutral loss eliminated in one step (as HCONH2), or, more likely, eliminated in rapid sequential steps [28]. The N-CO α-cleavage, a characteristic fragmentation observed as a common pattern in all natural and synthetic amides [29], could drive the genesis of the ion at *m*/*z* 161.0244, whereas ions at *m*/*z* 135.0455 (base peak), 134.0373, and 133.0302, and 132.0217 were attributable to dihydroxycinnamoyl residue (Table 1). UV-DAD spectra of compound **1** showed absorption bands at 324, 296, and 216 nm, plus a shoulder at 242 nm. This latter, together with the broad band at 324 nm was attributable to aromatic moieties π→π\* transitions, whereas the n→π\* and π→π\* electronic transitions referred to the amidic group were responsible for the absorption at 296 and 216 nm. Compound **2** showed the [M − H]<sup>−</sup> ion at *m*/*z* 298.1086 according to the molecular formula C17H17NO4. TOF-MS<sup>2</sup> spectrum suggested the occurrence of *N*-*p*-coumaroyloctopamine, before reported as an inducible phenolic amide in potato tuber tissue [30]. The deprotonated molecular ion gave rise to the ion [M − H − H2O]<sup>−</sup> at *m*/*z* 280.0981, which in turn provided the ions at *m*/*z* 134.0607 and 119.0502. The ion at *m*/*z* 145.0293 confirmed coumaroyl moiety, as well as the fragment ion at *m*/*z* 160.0408, which could be from the cleavage at the N-Cα bond and the following 1,4 nucleophilic addition on the β'-carbon of the phenylpropanoid side chain. The presence of the coumaroyl moiety was further revealed through the UV-DAD spectrum of the compound, which is was similar to that previously reported for this hydroxycinnamoyl amide (HAA) [26] and it was consistent with the loss of catechol group as highlighted by the blue shift of the absorption band detectable at 324 nm in compound **1** (Figure S3).

A constitutional isomer of the previous compound was *N*-caffeoyltyramine, which was found to inhibit macrophage-mediated inflammatory responses through the suppression of the production of NO and pro-inflammatory cytokines [31]. This compound was tentatively identified under peak **5**. In this case, the [M <sup>−</sup> H]<sup>−</sup> ion at *<sup>m</sup>*/*<sup>z</sup>* 298.1090 dissociated providing the TOF-MS<sup>2</sup> fragment ion at *m*/*z* 135.0459 as base peak, likely corresponding to 2-hydroxy-4-vinylphenolate. Following the cleavage at the N-Cα bond, the fragment ion at *m*/*z* 178.0520 was formed, whereas the ion at *m*/*z* 190.0518 could be from Cα-Cβ bond breakdown [32]. The CH2CO loss likely consisted in the fragment ion at *m*/*z* 256.0974 (Figure 2). In order to corroborate this latter hypothesis, hydrogen–deuterium (H/D) exchange reaction was carried out on pure *N*-caffeoyltyramine. The TOF-MS<sup>2</sup> spectra of the 3*d*-derived underwent rearrangement to give the ion at *m*/*z* 180.0619, whereas the loss of an ethen-1-one-2-d provided the fragment ion at *m*/*z* 258.1098. The electronic absorption spectrum of compound **5** (Figure 2C), similarly to that of compound **1**, with which shared the common caffeoyl component, evidenced the bathochromic and hyperchromic effect of -OH functional auxochromic groups of the double absorption band, which was at 317 and 294 nm, whereas the other band was at 237 nm. The compound was a constituent of other inestimable food sources such as hot pepper (*Capsicum annuum*), a spice used worldwide [33], Goji berry, the fruit of *Lycium barbarum* [34] and seeds of *Annona crassiflora* Mart., a fruitful tree native to the Brazilian Cerrado biome [35].

Compounds **8** and **17**, which showed the [M − H]<sup>−</sup> ion at *m*/*z* 312.1243 and 312.1241, respectively, were tentatively identified as *N*-feruloyltyramine geometrical isomers (Figure S4). The deprotonated molecular ion underwent in both TOF-MS<sup>2</sup> spectra methyl radical loss to achieve the fragment ion at *m*/*z* 297.1015(4), which gave rise to fragment ions at *m*/*z* 190.0510(3) and 178.0513, or more favorably led to ion at *m*/*z* 148.0530(5) (base peak) through CO-Cα' bond cleavage. The CH<sup>3</sup> • loss generated the radical ion at *m*/*z* 134.0373(6), as well as the ion at *m*/*z* 135.0452(6).

The [M − H]<sup>−</sup> ion at *m*/*z* 282.1140 for compound **11** was in accordance with the molecular formula C17H17NO4. UV-DAD and TOF-MS/MS spectra were according to *N*-*p*-coumaroyltyramine, an antioxidant HAA compound with inhibiting effect on acetylcholinesterase, cell proliferation, platelet aggregation [36]. In particular, in TOF-MS/MS spectrum, beyond the fragment ion at *m*/*z* 162.0558, which was consistent with coumaramide, the abundance of 4-vinylphenolate further confirmed the acylic moiety identity (Figure S5). Finally, compound **33** was tentatively identified as tri-*p*-coumaroylspermidine (Figure S6). This latter, until now never reported among hemp seed constituents, was firstly reported to reduce the mycelial growth of the oat leaf stripe pathogen

*Pyrenophora avenae* and also the infective ability of powdery mildew (*Blumeria graminis*f. sp. hordei) [37], and recently described as constituent of *Salvia* and *Lavandula* species [38].

**Figure 2.** TOF-MS/MS spectra of compound 5 (**A**) and of its 3d-derived (**B**). UV-DAD spectrum of the compound (**C**). Proposed fragmentation pathway of the [M − H]<sup>−</sup> ion for compound **5** (**D**) and the 3*d*-derived (**E**); theoretical *m*/*z* value is reported below each structure.

#### *2.2. HR-MS Analysis: Lignanamides*

The great part of the other identified compounds belongs to the lignanamide class (Table 2), which in *Cannabis sativa* fruit appears highly variable. In fact, it was reported to be constituted by aryl(dihydro)naphthalene-type, benzodioxanes-type and β-arylether-type compounds, as well as nor-lignanamides with a peculiar benzo-angular triquinane skeleton [10]. The arylnaphthalene lignanamide, cannabisin A, and three phenyldihydronaphthalene lignanamides were among the first isolated compounds in hemp fruits, and were herein tentatively identified thanks to their abundance, which made them easily isolable, and common fragmentation pattern features. Furthermore, UV-DAD spectra were also acquired and compared to those of pure standard compounds. In particular, cannabisin A, an arylnaphthalene lignanamide isolated so far from fruits of *Cannabis sativa* [39], was putatively identified in compound **16** (Figure S7). The [M − H]<sup>−</sup> ion at *m*/*z* 593.1941 (error ppm 2.0), in accordance with the molecular formula C34H30N2O8, gave the fragment ion at *m*/*z* 456.1110, which in turn, thanks to CO neutral loss, provided the ion at *m*/*z* 428.1151. The loss of 163.06 Da, likely corresponding to isocyanic acid (HNCO) + *p*-hydroxystyrene, also could directly occur from the deprotonated molecular ion supplying the fragment ion at *m*/*z* 430.1313. As previously reported, it appeared as HNCO elimination competes with backbone cleavage [40]. To support the hypothesis of *p*-hydroxystyrene residue loss, the ion at *m*/*z* 336.0520 was observed to be formed, which in turn, after HNCO loss, gave the ion at *m*/*z* 293.0457. This latter could be also obtained from the base peak due to 163.06 Da loss. UV-DAD spectrum of peak **16** was fully comparable to that previously reported for the pure reference compound [26]. In fact, UV-DAD spectrum exhibited a characteristic strong maximum at 256 nm and only small peaks between 280 and 350 nm.

The [M − H]<sup>−</sup> ion of compounds under peaks **15**, **22**, **28**, **32**, **36** and **38** was in line with the C34H32N2O<sup>8</sup> molecular formula and calculated exact mass equal to *m*/*z* 595.2086. TOF-MS/MS spectra of **15** and **22** were almost super-imposable, and their UV-DAD spectra mostly resembled the previously reported for cannabisin B electronic absorptions [26], with maximums at 218, 254, 282, 310 and 342 nm (Figure 3 and Figure S8). In particular, TOF-MS/MS spectra showed the ions at *m*/*z* 269.08 as base peak, whereas the ions at *m*/*z* 485.17 (− catechol), 432.14 (likely due to the loss of HNCO + *p*-hydroxystyrene), and 322.10, which could be favorably formed when the moiety weighing 163.06 Da

was lost together with a catechol unit, were detected with their relative high abundance. Cannabisin B, among lignanamides from hempseed, was one of the first to be investigated for its biological behavior, and it was found that it showed a marked antiproliferative action on HepG2 cell [11].

**Figure 3.** TOF-MS/MS spectra of compounds **15** and **22** (**A** and **C**, respectively). UV-DAD spectra of the compounds (**B** and **D**). In grey panel, the structure of cannabisin B is reported, without emphasizing stereochemical features.

Compound **28** was likely 3,30 -didemethyl-grossamide. The [M − H]<sup>−</sup> ion dissociated, as observed in the TOF-MS/MS experiment, providing the ion *m*/*z* 458.1262, attributable to a tyramine moiety direct neutral loss, as well as the more abundant ion at *m*/*z* 432.1468. This latter could further loss a tyramine unit providing the ion at *m*/*z* 295.0616 or also, more favorably supplied the base peak at *m*/*z* 269.0828, which was in accordance with a 2-hydroxy-4-(7-hydroxy-5-vinyl-2,3-dihydrobenzofuran-2-yl)phenolate (Figure S9). The occurrence of a phenylcoumaran lignandiamide with 2,3-dihydrobenzofuran nucleus was supposed, and it could be derived through the 8-50 coupling of two caffeoyl alcohols [41]. The comparison of the UV-DAD spectrum with that reported in literature allowed us to putatively further identify the compound [26]. Finally, the [M − H]<sup>−</sup> ion of compound under peak **32**, which fragmented into the base peak at *m*/*z* 298.1083, and minor fragment ions at *m*/*z* 178.0503 and 135.0444, could be attributable to 3,30 -demethyl-heliotropamide, an oxopyrrolidine-3-carboxamide, previously isolated from hemp fruits [20], whereas the benzodioxanes-type lignandiamides, cannabisin M and cannabisin Q, were tentatively identified from TOF-MS/MS spectra of compounds **36** and **38** (Figure S10). Charge-driven collision-induced dissociation could favor the formation of a phenoxide anion, which retro-cleaved leading the fragment ion at *m*/*z* 298.1088 (base peak).

Phenyl(dihydro)naphtalene-type lignanamides could be also the compounds under peaks **4**, **6** and **13**. The most polar compounds **4** and **6** showed the deprotonated molecular ions at *m*/*z* 609.1886 and 609.1887, respectively and were distinguishable with the other extract's compounds for their content in octopamine, beyond tyramine, as polyamine moiety. In particular, both the [M − H]<sup>−</sup> ions underwent water neutral loss (likely from the octopamine residue) providing a fragment ion at *m*/*z* 591.18 and presented the base peak ion at *m*/*z* 456.11 (Figure S11). This latter resembled that observed in compound **16**, in accordance with an arylnaphthalene lignandiamide core. Furthermore, TOF-MS/MS spectrum of compound **4** also displayed the fragment ion at *m*/*z* 472.1058, which could be due to the loss of the tyramine moiety, and further underwent water loss to yield the ion at *m*/*z* 454.0944 (Figure S11). The neutral loss of octopamine supplied for compound **6** directly the ion at *m*/*z* 456.1102, whose chemical feature likely resembled that of previously identified compound cannabisin I [27]. This latter was tentatively identified through TOF-MS and TOF-MS/MS spectra of compound **21**. Instead, compound **13** was likely a hydroxy derivative of *N*-caffeoyltyramine phenyldihydronaphtalene dimer. Its deprotonated molecular ion at *m*/*z* 613.2192 gave rise, through H2O neutral loss, to the base peak at *m*/*z* 595.2080, which in turn could lose 110.03 Da to provide the ion at *m*/*z* 485.1700, or could yield through 163.06 Da loss the ion at *m*/*z* 432.1446 (Figure S12). Both the ions at

*m*/*z* 595.2080 and 432.1446 could undergo phenyldihydronaphthalene moiety cleavage providing the ions at *m*/*z* 475.1864 and 312.1227, respectively. All the other detected fragment ions could be from 110, 163 or 120 Da losses. Furthermore, a dimer of *N*-caffeoyloctopamine and *N*-caffeoyltyramine was hypothesized to be compound **26**, whose deprotonated molecular ion at *m*/*z* 611.2047 gave rise the abundant ions at *m*/*z* 314.1044 and 298.1088, attributable to caffeoyloctopamine, and caffeoyltyramine, respectively (Figure S13).

H2O and CH2O losses were detectable in TOF-MS/MS spectra of peaks related to compounds **19** and **20**, which showed the deprotonated molecular ions at *m*/*z* 508.1968 and 508.1990, respectively, likely corresponding *erythro* and *threo*-diastereoisomers of cannabisin H [42]. The fragment ions at *m*/*z* 312.1244 and 312.1248, could be formed following the cleavage of β-aryl ether moiety, whereas the further methyl radical loss yielded the ions at *m*/*z* 297.1015 and 297.1009. Diagnostic ions at *m*/*z* 195.0653(69), likely α-hydroxyconiferyl alcohols, and the deformylation products at *m*/*z* 165.0557(60), appeared to support our hypothesis (Figure 4). The *erythro* diastereomer, together with grossamide K, was isolated for the first time as a phenolic constituent of the bark of the kenaf (*Hibiscus cannabinu*s var. *Salvador*) [42].

**Figure 4.** TOF-MS/MS spectra of compounds (**A**) **19** and (**B**) **20**, tentatively identified as cannabisin H isomers. The proposed fragmentation pathway of their [M − H]<sup>−</sup> ion was reported (**C**); the theoretical *m*/*z* value is reported below each structure.

The deprotonated molecular ions of compounds **23–25** and **35** were in accordance with the molecular formula C35H34N2O<sup>8</sup> and the occurrence of lignandiamides in which the hydroxycinnamoyl moieties were represented by caffeoyl and coniferyl alcohols, whereas tyramine constituted the amine part. In TOF-MS/MS spectra of compounds **23–25**, the base peak ion was at *m*/*z* 446.16, allowing us to confirm that the concurrent loss of isocyanic acid and *p*-hydroxystyrene could be advantageously observed as informative of tyramine presence. Moreover, other diagnostic fragment ions could be observed and differentiate the major bonding types encountered in hemp fruit lignandiamides. The aryldihydronaphtalene-type core likely characterized both compounds **23** and **24** which were distinguishable through fragment ions at *m*/*z* 499.1891 and 485.1757, respectively, which were in accordance with their relative catechol or guaiacol loss. This finding was in line monolignol

cross-coupling with a caffeoyl-end-unit for compound **23** and a guaiacol-end unit for compound **24**. Based on previous observation, TOF-MS/MS spectrum of compound **35**, which appeared to fragment via a pathway similar to that of compound **28**, allowed us to hypothesize demethylgrossamide occurrence. In fact, also in this case, the [M − H]<sup>−</sup> ion underwent tyramine neutral loss with the genesis of the ion at *m*/*z* 472.1428, whereas the loss of 163.06 Da gave the most abundant ion at *m*/*z* 446.1633, which in turn lost a methyl radical, providing the radical anion at *m*/*z* 431.1391. Furthermore, as already observed in compound **28**, the loss of two 163.06 Da units could occur giving the ion at *m*/*z* 283.0979. This latter, which represented the base peak, also furnished the radical ion at *m*/*z* 268.0743 through methyl radical loss (Figure 5). UV-DAD spectra were in accordance with tentatively assigned lignandiamide skeleton, and represented a useful tool to unravel the lignan nucleus of compound **25**, which was supposed to belong to aryldihydronaphtalene class [26].

**Figure 5.** TOF-MS/MS spectra of compounds (**A**) **23**, (**B**) **24**, (**C**) **25**, and (**D**) **35**.

Based on previous MS observations, and UV-DAD spectra, compounds **27**, and **29**, whose pseudomolecular ions were in accordance with the C36H36N2O<sup>8</sup> molecular formula, and showing the base peak at *m*/*z* 460.1799(1803), and its demethylated radical ion at *m*/*z* 445.1569(1), were suggested to be aryldihydronaphtalene-type lignanamide isomers, whereas compound **37** was tentatively identified as cannabisin F, and compound **39** was putatively assigned as grossamide (Figure S14). This latter compound, whose MS/MS fragmentation pattern in positive ion mode was previously reported [43], was found to exert anti-neuroinflammatory action, being able to inhibit the secretion of pro-inflammatory mediators (e.g., IL-6, and TNF-α), reducing LPS-mediated IL-6 and TNF-α mRNA levels [44]. Furthermore, neuroprotection by cannabisin F was ascertained against LPS-induced inflammatory response and oxidative stress in BV2 microglia cells [21].

Grossamide K, a phenylcoumaran-type lignanamide with previously reported antimelanogenic activity [45], was tentatively identified based on TOF-MS and TOF-MS/MS spectra related to peak **30**. The deprotonated molecular ion at *m*/*z* 490.1875 provided the fragment ions at *m*/*z* 472.1769 and 460.1769, following H2O and formaldehyde neutral losses. Charge-driven CH2O loss could be initiated when phenoxide ion abstracted the proton from aliphatic OH function. Both the ions underwent methyl radical loss to achieve ions at *m*/*z* 457.1541 and 445.1529, respectively, which, in turn, gave diradical anions at *m*/*z* 442.1300 (base peak) and 430.1300. This latter could provide the fragment ion at *m*/*z* 297.1125 through CO-Cα cleavage, or the ion at *m*/*z* 338.1027 by dehydrogenation and hydroxystirene loss. Methyl radical and CO losses were further detected (Figure S15).

Finally, compounds **31** and **34** were tentatively assigned as cannabisin E isomers based on their deprotonated molecular ion at *m*/*z* 641.2516, which provided the ion at *m*/*z* 489.2053(64) through 4-hydroxy-3-methoxy benzaldehyde loss. The detection of the ion at *m*/*z* 151.0404(6) as base peak, attributable to 4-formyl-2-methoxyphenolate, seemed to confirm the hypothesis (Figure S16).

#### *2.3. HRMS Analysis: Flavonol Glycosides*

Compounds **3**, **7**, **9**, **10**, **12**, **14** and **18** were putatively flavonol glycosides (Figures S17–S20, Table 3), whose presence was previously identified as bioactive components of hemp seeds, and its cold-pressed oil derived product [24]. In particular, collision-induced dissociation of the [M − H]<sup>−</sup> ion at *m*/*z* 433.0777 for compound **3**, gave radical aglycone ion [aglycone − H]<sup>−</sup> at *m*/*z* 300.0273 and its aglycone ion at *m*/*z* 301.0362, which were consistent with quercetin and the loss of a pentose moiety (−132.0415 Da). The presence of the pentose moiety was further suggested for compounds **9** and **10**, whose aglycone moiety was kaempferol, as suggested by the [aglycone <sup>−</sup> H]– and aglycone radical ion at *<sup>m</sup>*/*<sup>z</sup>* 285.0406(08) and 284.0328(29), as well as the ions at *m*/*z* 255.0299(303) and 227.0352(1), which could come from the [aglycone − H]•− ion. The loss of deoxyhexose moiety was shared by compounds **7**, **12** and **14**. Quercetin was identified as aglycone residue in compound **7**, whereas compounds **12** and **14** were kaempferol deoxyhexoside isomers. Finally, the detected loss of acetic acid moiety in the TOF-MS/MS spectrum of compound **18** as well as the occurrence of [aglycone–H]– , [aglycone − H]•− and [aglycone − 2H]•− at *m*/*z* 301.0361; 300.0278 and 299.0194, were in accordance with quercetin-7-O-acetyldeoxyhexose [46,47]. Quantifying flavonoid content as peak area percentage, it was found that the compounds constituted only the 2.8%, whereas lignanamides were the most representative compounds with the 79.0%. Compounds deriving from the coupling of two phenylamides, both preserving the amine moiety (e.g., metabolites showing the [M − H]<sup>−</sup> at *m*/*z* 593, 595, 609, 623, 641), were the most abundant and they were present in comparable amount (Figure S21).

### *2.4. LnHS Inhibits Cell Survival of Glioblastoma U-87 Cell Line but not of HF*

As some of compounds in LnHF were reported to exert neuroprotective and anti-neuroinflammatory effects [20,21], or also to be able to induce dramatic morphological changes and autophagic cell death [11], the effects of the treatment of LnHS hempseed mixture on primary glioblastoma cell lines was investigated. Indeed, a prior analysis of the effects of LnHS was carried out on undifferentiated SH-SY5Y cells. It was evidenced that LnHS at dose levels higher than 25 µg/mL, induced clear cell morphological changes as cells shrunk and loss cell adhesion (Figure S2). Thus, in order to better understand and define a preliminary LnHS-cell death inducing mechanism, the effects on cytotoxicity was further assessed on U-87 glioblastoma cells, which are characterized by a rapid migration, and a highest invasion capacity [48]. Human Fibroblast (HF) cells were used in order to evaluate LnHS effects on non-tumorigenic cells. MTT assay was performed to assess the effects of LnHS on cell viability, at seven dose levels (0.5, 1.25, 2.5, 5, 10, 25, 50 µg/mL), and at three exposure times (24 h, 48 h and 72 h). Data acquired showed that LnHS exerted toxic effects on U-87 cells. Indeed, U-87 cell viability was strongly compromised at the highest LnHS tested dose, at each exposure time considered (Figure 6). The dose level 25 µg/mL significantly affects the U-87 viability after 48 and 72 h of incubation, *p* < 0.05 (Figure 6A). LnHS did not show relevant toxic effects on HF cells, and only the highest dose (50 µg/mL) appeared to induce, after the longer 72 h exposure time, cell viability decrease (Figure 6B). In addition, the quantification of lactate dehydrogenase levels, considered an indicator of cell damage for necrotic cell death [49], evidenced that LnHS 25 µg/mL and 50 µg/mL dose levels exerted a time-dependent percentage increase of LDH release (Figure 6C,D). Furthermore, the colony formation assay [50] showed that LnHS-treated U-87 cells did not retain the capacity to produce colonies (Figure 6E), when LnHS dose exceeded only 5 µg/mL.

**Figure 6.** Cell viability of U-87 and human HF cell (**A** and **B**, respectively) was assessed by MTT assay after 24, 48 and 72 h of exposure. Data from LDH release assay at 24, 48 and 72 h exposure times were in panels **C** (U-87 cells) and **D** (HF cells). Values are the mean ± SE of two independent experiments performed in triplicate. \**p* < 0.05 vs. untreated cells. (**E**) Representative images from colony forming efficiency of U-87 cells grown in presence of LnHS or vehicle control for ten days; the experiment was performed in duplicate.

#### *2.5. LnHS Induced Genomic DNA Damage in U-87 Cell Line but not in HF*

In order to verify if the observed cytotoxicity was the result of LnHS ability to induce a genotoxic damage, comet assay was performed on U-87 and HF cells, after 24 h of exposure to the hempseed mixture at 0.5, 2.5, 5, 25 and 50 µg/mL concentration levels (Figure 7). U-87 cells showed DNA damage when treated with LnHS 25 and 50 µg/mL (Figure 7A), whereas no genotoxic effects was detectable in human fibroblast cells.

**Figure 7.** Representative images of U-87 and HF cells treated with different LnHS doses and subjected to the comet assay (panels **A** and **B** respectively). ctrl: untreated cells; green arrows indicate comets with a tail. Experiments were performed in duplicate.

#### *2.6. LnHS Inhibited Cell Migration of U-87 Cell Line*

As invasiveness is one of the pathophysiological features of human malignant gliomas [51], the effects of LnHS on the U-87 cells migration, in comparison to HF, were tested through wound healing assay. Both the cell types were subjected to scraped wounds: U-87 cells were treated with five

doses (0.5, 2.5, 5, 25 and 50 µg/mL), whereas HF cells with three LnHS doses (5, 25 and 50 µg/mL). Although LnHS significantly inhibited the migration of both cell types at the highest dose, U-87 appeared more sensitive to LnHS (Figure 8). In fact, already at a dose of 5 µg/mL, the hempseed fraction inhibited the migration of the U-87 cells; the maximum effect was evident at 50 µg/mL, where the cell toxicity is also evident.

**Figure 8.** U-87 cells underwent a scraped wound and were then treated with different LnHS doses (0.5, 2.5, 5, 25 and 50 µg/mL). Cells were photographed immediately following the scratch (0 h), after 12, 24, 36, 48, 72 and 96 h. Untreated cells (ctrl), were used as a control. Representative figures are shown from one of two independent experiments.
