Alkaloid profiling of Italian alpine herbs using high resolution mass spectrometry (Orbitrap-MS)

Abstract Alpine pastures have not yet been extensively studied with regard to the presence of alkaloids in herbaceous plants. In this work the alkaloid profiles were characterised from a selection of 62 herbs collected from alpine pastures in north-eastern Italy. High-performance liquid chromatography coupled to a hybrid quadrupole-orbitrap mass spectrometer was used to evaluate the presence of 41 different alkaloids and quantify them using a targeted approach. Provisionally, 118 alkaloids were identified, including both free and glycosylated forms, making use of a homemade database and a suspect screening approach. rucifoline, gramine, heliotrine, lycopsamine, seneciphylline, and veratramine were quantified with concentrations ranging from 6 to about 100 µg kg−1 in 6 plants. Herbaceous plants belonging to the most well-represented plant families (Poaceae, 9 species; Asteraceae, 7; Lamiaceae, 6) showed distinct and characteristic alkaloid profiles and were correctly reclassified with an average accuracy of 85% (Partial Least Squares – Discriminant Analysis). Graphical Abstract


Introduction
The alps are characterised by a remarkable geomorphological and ecological variability and therefore associated with the presence of a considerably high number of different plant species. In terms of flora, this mountain range involves 4,500 vascular plant species, representing 39% of European flora (Morschel 2004). The composition of alpine herbage is not frequently addressed in the literature (Peiretti et al. 2015;Lefebvre et al. 2016) and, to the best of our knowledge, no systematic and extensive studies have been dedicated to alkaloids, even though they belong to some of the most important secondary metabolites in plants. Alkaloids are basic nitrogen-containing organic compounds deriving from amino acids or arising from amination of another type of substrate, which may be a terpene or a steroid. Alkaloids are used by plants as protective agents against predator attack (Dobler 2001) and therefore playing an important role in the interaction of plants with their environment (Fester 2010). Some alkaloids from plant extracts used in traditional medicines are responsible for the beneficial health effects (Applequist and Moerman 2011;Liao et al. 2016;Zubair et al. 2016), while on the other hand some alkaloids could be a potential threat for the human body. Pyrrolizidine (Pyz) alkaloids for example, are hepatotoxic, mutagenic and carcinogenic (Paßreiter 1992;Hartmann 2007) and show a high occurrence in Ranunculaceae, Berberidaceae, Menispermaceae, Piperaceae, Cactaceae, Papaveraceae, Gentianaceae and Solanaceae (Yang et al. 2009). On the other hand, Pyz is almost absent in families such as Fagaceae, Betulacae, Casuarinaceae and Juglandaceae, suggesting that alkaloids can be used as a marker of the botanical family (Li and Willaman 1968). Nevertheless, the presence in the literature of studies about the glycosylated alkaloids in plants (Larcher and Nardin 2019), with exception of steroidal alkaloids, which are well known to be abundant in plants of the genera Solanum and Veratrum (Aniszewski 2007).
This study aimed to define the alkaloid profiles of a variety of herbs characterising natural pastures in the eastern Italian Alps. In the present work a selection of 62 herbs was used in order to carry out preliminary assessments on the distribution of these metabolites in alpine plants, which in reality involves a very large variety of different species. Despite addressing only a part of the existing alpine plant species, this approach was considered suitable for the evaluation of possible recurring presences of metabolites within botanical families. Up until now the studies addressing the alkaloid composition of alpine plants are often considering only a single alkaloid in a specific plant family (Chandra and Purohit 1980), or the most typical alkaloids in one specific plant (Busqu e et al. 2010). In the present work it was aimed to obtain a general alkaloid profiling in which, for the first time, a large number of alpine plants was included. Therefore, this study considerably increases the knowledge on alkaloids in the field of alpine plant phytochemistry. Moreover, we focused our attention on edible species, being particularly relevant for the investigation of the possible dangers for the grazing animals that eat these plants during the summer season.
The alkaloid profiles were obtained by liquid chromatography, using an online SPE clean-up system for plant extracts, combined with high-resolution mass spectrometry for rapid and selective detection of different alkaloids.

Results and discussion
First, a quantifiable evaluation of the content of the most known and toxic alkaloids in alpine herbs was made, considering the ones of which standards were available on the market. Subsequently, the investigation on alkaloid profiles was extended by carrying out a suspect screening analysis.out. As in this case pure standards were not available, the research was focused on alkaloids for which the retention times were established by analysing several plants (see Supplementary Material, Suspect screening study of alkaloids) rich in those compounds. Moreover, some MS spectra fragmentations from literature or online libraries have been used for confirmation and a more robust identification (Table S6). The results obtained from these two studies were subsequently used for statistical analysis.

Alkaloid profiling of individual alpine herb species: targeted study
Some studies highlight the possible danger of alkaloids for grazing animals that are commonly involved in poisoning episodes (Cortinovis and Caloni 2015;Matsuura and Fett-Neto 2015). Considering that the analysed plants were harvested in the summer period, the knowledge of their alkaloid content could be interesting for evaluating possible toxicological problems for grazing animals that include these plants in their feed. A study based upon the cattle intake of different Senecio species with high dehydroPyzs content, determined that the lethal doses were for S. jacobaea and S. longilobus about 870 g day À 1 whereas for S. riddellii 273 g day À 1 (Molyneux et al. 2011). Considering that the Pyz content found in the alpine plants resulted 10000 times lower than these Senecio species (0.3 to 6% dry plant) and that a cow usually eats about 60 kg of herbs and a sheep 4 kg of herbs, it is very improbable that the Pyz content found in Capsella bursa-pastoris, Veronica chamaedrys, Potentilla crantzii, and Hypericum maculatum, can be harmful. Moreover, the LD50 reported by WHO Food Additives (World Health Organization 2020) for lycopsamine, seneciphylline, and heliotrine (1500 mg kg-1 bw, 77 mg kg-1 bw, and 510 mg kg-1 bw, respectively for rats) show that the concentrations that we found in these specific plants should not be of any concern. Finally, it should be taken into account that the cattle feed on a variegated pool of herbs, which lowers the estimated Pyz intake considerably.
One indole (Ind) alkaloid, gramine, was detected in Stachys alopecurus with a concentration of 118 mg kg À1 . Gramine is the simplest indole alkaloid with antibacterial properties towards the growth of Pseudomonas (Sepulveda and Corcuera 1990). Considering this antimicrobial activity, different studies were conducted to evaluate the effect on digestibility of plants. Coulman et al. (1977) reported that gramine inhibited the in vitro digestibility of orchard grass but found only little effect on the in vitro digestibility of cellulose or plant material. Besides, herbs with high gramine content are not suitable for grazing because of the bitter taste that reduces their palatability (Popp et al. 2016). The LD50 of gramine is 62.9 mg kg À1 bw in rats (Erspamer 1954).
Finally, veratramine, one of the steroidal (Str) alkaloids, was detected in Phleum rhaeticum with a concentration of 10 mg kg À1 . This alkaloid occurs in several Veratrum species (Liliaceae) and is known to antagonise the Na þ channel gating mechanism (Pelletier 1998). Together with other alkaloids of the same family it has been evaluated for its ability to inhibit the growth, proliferation, and migration of the human metastatic prostate cancer cell line PC-3 (Khanfar and El Sayed 2013). nlike veratramine, only veratrum alkaloids with an intact furan ring were teratogenic for cattle (Keeler 1978) and the LD50 for this compound was calculated to be 15.9 mg kg-1 bw in rats (Wang et al. 2008).
According to these results, the presence of potential toxic alkaloids turned out to be infrequent and with very low concentrations.

Alkaloid profiling of individual alpine herb species: suspect screening study
With regard to the suspect screening approach (Table S6), up to 60 different alkaloids were tentatively identified in the 62 alpine plant samples. In particular, piperidines (Pprs) and Inds were detected in approximately 50% of samples, terpenoids (Trns) and pyridine (Pyr) in about 40%, tropanes (Trps) in 32%, isoquinolines (Iqns), Pyzs and Strs in about 20%, pyrrolidines (Pyls) in 13%, and quinolines (Qnls) and acridones (Acds) in about 2% of samples. Considering the individual plants, Capsella bursa pastoris, Achillea millefolium, Poa alpine and Carex sempervirens showed the highest number of different alkaloids (9, 9, 11, and 12 respectively). The suspect screening alkaloid profiles of the herbs are reported in Table S2.
Among the most common alkaloids, 3-acetyltropine was found in 18 plants, and gelsemine, norharmane and actinidine in 13 plants, Valerianine@12.9 in 11 plants, Harmane@15.5 and Valerianine@8.9 in 10 plants. Trps, such as 3-acetyltropine mainly occur in the family Solanaceae and have mydriatic and anaesthetic properties (Barton et al. 1999). Inds such as gelsemine, norharmane and harmane have a bicyclic structure consisting of a benzene ring fused to a five-membered pyrrole ring. The difference between Inds is the presence of carbonyl, methoxyl and hydroxyl groups at different positions, which support different bioactivities among which anti-inflammatory, antinociceptive, anti-tumoural, antioxidant and antimicrobial (Rosales et al. 2020). In particular, gelsemine had analgesic, anti-anxiety, and anaesthetic effects (Liu et al. 2019), while aromatic b-carbolines norharman and harman have been associated with a number of human diseases including Parkinson's disease, tremor, addiction and cancer (Pfau and Skog 2004). Among Pyrs, actinidine and valerianine are the principal valerian alkaloids which are used to treat sleeping disorders, restlessness, and anxiety (Pato cka and Jakl 2010).
Considering the very low signal of the peaks, it was often not possible to verify the total correspondence of the isotopic pattern of the molecule and the typical fragments. The identification was therefore based on the accurate mass (mass tolerance <5 ppm) and the RT (þ/-0.5 min). Moreover, in the screening study it was not possible to quantify the content, as the standards were not available. Besides, the detected signals were in all cases very low. Nevertheless, it is important to specify that in this study many of the detected alkaloids are not reported in the literature in relation to toxicity problems, even though they are often used for their curative properties.

Discrimination of botanical families
This study made it possible to describe the complex compositional alkaloid profiles of a large selection of alpine herbs, specifically characterising the three most representative botanical families (Poaceae, Asteraceae and Lamiaceae). Although the alkaloids seem typical and different from herb to herb, evaluating a large number of samples allowed us to find some significant and interesting similarities between botanical families.
Significant differences (Tukey's HSD test) characterised the individual alkaloid content in the main herb families. Considering families represented by at least 3 sampled plants, the following alkaloid variability is considerably important. Caryophyllaceae were significantly richer in harmol compared to Apiaceae and Rosaceae and in norharmane compared to Lamiaceae. Furthermore, Plantaginaceae plants were richer in 3acetyltropine than Asteraceae.
A PLS-DA model was used in order to assess whether plants belonging to the same taxonomic group were characterised by a particular alkaloid profile, and conversely whether a specific alkaloid profile could allow prediction of the corresponding botanical family. To assure a correct representation for the botanical family, the model was based on the alkaloid profile of the three most abundant families -Poaceae, Asteraceae and Lamiaceae -respectively comprising nine, seven and six individual herb species (X-matrix). Three variables (Y-matrix) were defined for the families and 16 alkaloids selected among those who were present in at least three herb samples were used for obtaining a two-factor PLS-DA discriminant model. The relationship between the two sets of variables is represented in Figure S2.
The families are located towards the ring of the plot, well separated in three different quadrants, indicating that the alkaloid profile of herbs may be a viable tool for grouping them according to the botanical family. Table S3 shows the full cross-validation results of the PLS-DA model together with the evaluation of prediction performance.
The average accuracy of the discriminant multiclass model was 85%. The Asteraceae and Lamiaceae discriminant models achieved the accuracy percentage of 0.82 and 0.91 respectively, misclassifying three and two species of each selected family (sensitivity¼ 0.57 and 0.67, respectively). The false negative herbs were Crepis aurea, Hieracium pilosum and Achillea clavennae for Asteraceae and Betonica alopecurus and Prunella grandiflora for Lamiaceae. Furthermore, the Asteraceae and Poaceae models were affected by two and three false positive family identification of individual plants (specificity ¼ 0.93 and 0.69, respectively). Indeed, Betonica alopecurus and Prunella grandiflora were erroneously validated as an Asteraceae species while Crepis aurea, Hieracium pilosum and Achillea clavennae were recognised as Poaceae. Figure S3 shows the graphs of the VIPs, with the two valerianine isomers and dihydropiperlonguminine being the most powerful alkaloids for discriminating Asteraceae, whereas harmane and harmol for Poaceae and piperine and valerine for Lamiaceae were found most important.

Conclusions
This study made it possible to describe the complex compositional alkaloid profiles of a selection of alpine herbs, specifically characterising three of our most representative botanical families. Although the alkaloids seem typical and different from herb to herb, evaluating many samples made it possible to find some significant and interesting similarities between botanical families. The present work resulted in new insides in the botanical field, considering that the alkaloid profile of alpine plants is rarely described in literature. The quantification of alkaloids in specific species did not reveal a potentially dangerous content for the animal consumption. Although the alkaloids were not quantifiable, the screening profile is considered a promising method to be used in a wide field of applications, for example for the evaluation of alkaloid transfer from the animal to the milk.

Disclosure statement
No potential conflict of interest was reported by the authors.