Polar components of hemp seeds were extracted through a very sustainable procedure under different pH conditions. The extraction yield from hemp seeds was 30.27 ± 0.03% at pH 5, 35.13 ± 0.11% at pH 7 and 25.28 ± 0.23% at pH 10. For defatted seeds, the yield of the extraction carried out at pH 7 was 34.46 ± 0.15%.

All of the extracts (Exs) were revealed to be complex mixtures on analysis by ¹H-NMR spectroscopy (Figure 1).

The signals in the region between 3–5 ppm in the ¹H-NMR spectra suggest the presence of carbohydrates, whereas those between 0.5–3 ppm could be attributable to aliphatic carbon chains. The signals in the range of 6–8 ppm can be ascribed to the presence of peptides or amino acids. The NMR spectra suggest that aliphatic proton signals and the signals attributable to amino acids were more abundant in Ex7d and Ex4.5 compared to the other two extracts. By contrast, in Ex7 and Ex10, the carbinolic proton signals were more pronounced. Finally, Ex7d and Ex4.5 spectra showed signals between 6 and 8 ppm that are attributable to aromatic rings, whereas the same signals in Ex7 and Ex10 were present at a very low intensity.

All of the extracts were subjected to chemical derivatization to convert all of the compounds into volatile derivatives. They were then analysed by GC-MS. The results of chemical analyses (Figure 2 and Figure 3, Table 1 and Table 2) agreed with the results of NMR analysis. Indeed, all of the extracts were shown to contain a mixture of fatty acids, amino acids, and monosaccharides. Ex7d and Ex4.5 are composed mainly of amino acids and saturated fatty acids, while Ex7 and Ex10 are composed mainly of glycerol, the sugars glucose and galactose and unsaturated fatty acids. The compounds were identified by comparing retention times in both the TIC and the mass spectrum with those of authentic standards, which were also derivatised with the same procedure. As an example, Figure S1 shows the chromatograms and the mass spectra of both Ex4.5 and galactose and the galacturonic acid standard used for its identification. All of the components of each extract were identified in the same way. The absence of cannabidiol and tetrahydrocannabinol peaks from these chromatograms could be due to the high molecular weight of these compounds after derivatization [59]. In fact, even when the final temperature of the GC program was 300 °C, the library did not detect any of these components. On the other hand, the comparison of our results with those obtained using other aqueous extracts from hemp seeds suggested no other differences [59].

To evaluate the effect of extracts on bacterial viability, the dried Exs, diluted to the desired final concentration in BHI medium, were analysed to investigate their antimicrobial activity against S. epidermidis RP62A.

To this end, an appropriate dilution (0.125 OD600nm) of S. epidermidis RP62A bacterial culture in the exponential phase was used. No antimicrobial activity against S. epidermidis RP62A strains was observed for any tested Exs (maximum concentration tested 500 µg mL⁻¹) (Figure S2).

The antibiofilm effect of the extracts at a concentration of 1 mg mL⁻¹ was examined against two staphylococcal strains: S. epidermidis O-47 [60,61] and S. epidermidis RP62A [62]. The S. epidermidis O-47 strain is methicillin-susceptible, and it was isolated from patients with orthopaedic implant infections [63]. It is a naturally occurring agr mutant [61] and produces only very low levels of delta-toxin and other phenol-soluble modulins (PSMs), the transcription of which is agr-controlled [61,64]. As previously reported [65], the agr-negative genotype of S. epidermidis O-47 is responsible for increased expression of the surface protein AtlE, a bifunctional adhesin/autolysin abundant in the cell wall of S. epidermidis, and this overexpression enhances biofilm formation on polymer surfaces. S. epidermidis RP62A (ATCC 35984) is a methicillin-resistant strain [66] isolated from an outbreak of intravascular catheter-associated sepsis [67].

As shown in Figure 4, almost all of the extracts were able to inhibit biofilm formation by both staphylococcal strains. The only exception was Ex10, which was less active. It is interesting to note that Ex7d, Ex7 and Ex4.5 had similar activity profiles against the two target strains; they are all more effective on the S. epidermidis O-47 biofilm than on the S. epidermidis RP62A biofilm. The two S. epidermidis strains differ in several aspects [60]. One of the reported differences between S. epidermidis RP62A and S. epidermidis O-47 is the amount of AtlE present in the cell envelope [61]. Future studies will clarify whether this characteristic is involved in the difference in the extracts’ efficacy. In any case, the phenotypic and genotypic heterogeneity of S. epidermidis strains [68,69,70] could limit the clinical applicability of extracts. Therefore, future studies will be dedicated to testing the extracts against different S. epidermidis strains.

The observation that the extracts have similar activity profiles against the two target strains could suggest that in the three extracts, the molecule/s involved in antibiofilm activity is/are the same and that the slight difference in activity is related to the concentration of the compound/s. It is difficult to ascribe this behaviour to specific key component/s, as the antibiofilm effect of such complex mixtures can depend on interactions (enhancing and/or inhibiting the bioactivity) between the components.

Because all of the extracts were active at a concentration of 1 mg mL⁻¹, their anti-biofilm effect against staphylococcal strains was evaluated at different concentrations within the 0.5–0.01 mg mL⁻¹ range. As reported in Figure 5, the effect was concentration-dependent, and the activity of the extracts decreased as their concentrations decreased. Ex7d showed the highest activity against both target strains, even at the lowest tested concentration (0.01 mg mL⁻¹).

In order to characterize the compositions of the extracts that exhibited antibiofilm activity (Ex7d, Ex7 and Ex4.5), the protein content was evaluated by BCA assay and SDS-PAGE followed by Comassie-blue staining. No proteins were present in any of the tested extracts, as shown in Figure S3. The absence of proteins in the tested extracts agrees with findings in the literature, in which it is reported that degreasing or defatting is a fundamental step in extracting proteins from plant seeds, as the presence of oils in general causes the formation of lipid-protein complexes, reducing the extraction yield [71]. Nonetheless, no proteins were found even in Ex7b, which was obtained from defatted hemp seeds, because the extraction pH (pH 7) was not optimal for dissolving hydrophilic proteins [71].

The biocompatibility of the extracts that exhibited antibiofilm activity (Ex7d, Ex7 and Ex4.5) was evaluated by using immortalised human keratinocytes to assess the potential applicability of these extracts to humans. HaCaT cells were incubated with increasing concentrations of each extract (i.e., from 0.1 to 100 µg mL⁻¹), and their viability was evaluated at 24 h and 48 h. Figure 6A,B shows that none of the analysed extracts significantly affect cell viability at the investigated concentrations at 24 and 48 h, respectively. Only Ex4.5 slightly reduced cell viability after 24 h at a concentration of 100 µg mL⁻¹ and after 48 h at a concentration of 50 µg mL⁻¹. The absence of toxic effects makes the extracts suitable for the intended application.

According to an activity-guided approach, Ex7d was further fractionated by HPLC on a gel filtration column because it exhibited the highest antibiofilm activity against the two S. epidermidis strains (Figure S4). This strategy aims to evaluate the antibiofilm effect of the separated fractions and thus to potentially identify the component/s responsible for the observed activity. The obtained fractions (A–E) were tested against both S. epidermidis strains at a concentration of 0.1 mg mL⁻¹ (Figure 7).

Fraction B proved to be the most effective in countering biofilm formation, especially against the O-47 strain, the biofilm production of which was reduced by 90%. The other fractions induced a reduction in biofilm occurrence in both strains that ranged between 20% and 60%. Fraction B was therefore subjected to NMR analysis. The 2D-NMR spectra suggest that this fraction is still a mixture of different components.

The 2D-NMR results shown in Figure 8 suggest that the purified fraction responsible for the antibiofilm activity is mainly composed of fatty-acid chains, glycerol, and alditols. Alditols were revealed here for the first time, likely because their concentration in the unpurified extract was not high enough for them to have been detected before. In particular, the presence of the -CH₂OH proton signals around δ 3.6–3.8 ppm in correlation with -CH₂OH carbon signals at δ 62 ppm and -CHOH indicate the presence of glycerol. Moreover, the additional -CHOH signals in the range 3.8–4.5 ppm for ¹H nucleus and in the 70–80 ppm region for ¹³C signals suggest the presence of inositol. Finally, the numerous -CH₂- signals may indicate the presence of methylene groups of fatty acids.

Until now, researchers have primarily investigated organic extracts from hemp seeds for antibiofilm activity. A bioactive ethanol extract from hemp seeds was analysed by mass spectrometry, and free fatty acids and phospho-glycerol derivatives, as well as flavanols and amino acids, were found [72]. According to these results, it can be hypothesised that the antibiofilm activity is mainly attributable to glycerolipids, although it is not possible to exclude the possibility of synergy with the other components.