Plants and plant-derived products may contain naturally occurring toxins that can be harmful to humans. Pyrrolizidine alkaloids (PAs) are a class of natural toxins that have drawn increased amounts of attention [1]. PAs are secondary metabolites of plants produced by a defense mechanism against insects. To date, more than 600 PAs and their N-oxides (PANOs) have been identified from nearly 6000 plant species [2]. PAs are widely present in various plant-derived foods, such as spices, honey, and herbal teas, and can potentially pose a risk to human health via dietary intake [3]. Among these PAs, 1,2-unsaturated PAs have been proven to be carcinogenic and hepatotoxic to humans [3]. After metabolic (oxidative) activation of PAs into dehydropyrrolizidine (DHP) esters, adducts with DNA are formed and considered the major cause of the carcinogenic effects of PAs [4]. Incidents of liver damage caused by the consumption of PAs in foods have been reported [5] and are considered to be one of the major causes of hepatic venous occlusion disease (HVOD), which can lead to cirrhosis and liver failure. Long-term exposure to these pollutants has been associated with genotoxic and carcinogenic effects [1].

Due to the widespread presence of PAs/PANOs in different types of food, their presence in food should be recognized as a food safety issue. An assessment of PA/PANO levels during food processing can provide a realistic picture of PA exposure. For example, Casado et al. (2023) summarized the effects of heat treatment, fermentation, infusion preparation, milling, washing, and soaking on PAs during food processing [6]. Additionally, the most relevant analytical procedures for their determination in different food products were included from 2010 to 2020. The development of sensitive analytical methods can lead to a better understanding of PA occurrence [7]. However, it is important to note that the lack of sufficient toxicological data on PAs hinders their risk assessment. In addition, due to insufficient toxicity data on PAs, the toxic levels of PAs vary widely, and species vary widely in their sensitivity to PA exposure. Moreover, the risk assessment standards and legislation for PAs vary among different countries. Within the framework of current risk assessments, individual PAs are considered a group of equipotent substances with carcinogenic effects [8]. The large number of PAs/PANOs occurring in plants makes it impossible to generate comprehensive in vivo data on toxicity. Merz and Schrenk (2016) reported interim relative potency (REP) factors for the toxic and genotoxic potency of 1,2-unsaturated PAs based on limited cytotoxicity data [9].

2. Chemical Structure and Toxicity of PAs

PAs are mostly formed by a pyrrolizidine ring and an esterified organic acid, with the pyrrolizidine ring referred to as the necine and the acid part as the necic acid (Figure 1A). Based on the presence or absence of unsaturated double bonds at positions C-1 and C-2 of the necine structure, PAs are divided into saturated and unsaturated types. Saturated PAs with a saturated necine base, such as the platynecine (PLA) type, are known to be non-toxic. The unsaturated PAs are further divided into retronecine (RET), heliotridine (HEL), and otonecine (OTO) types (Figure 1B). OTO-type PAs include clivorine and senkirkine, while RET-type PAs include retrorsine and senecionine. HEL-type PAs include heliotrine and lasiocarpine (Figure 1C). Additionally, nitrogen atoms on the necine moiety can be oxidized to form N-oxides, which coexist with PAs in most plants [4]. However, otonecine-type PAs are not able to form a corresponding PA N-oxide due to the methylated nitrogen in the necine base core structure. Saturated PAs generally exhibit low or no toxicity, while 1,2-unsaturated PAs are of great concern due to their hepatotoxic, carcinogenic, and genotoxic properties [8]. After being absorbed in the small intestine, PAs are transferred to the liver, where they are metabolized by cytochrome P450 enzymes (CYP450) to form active primary metabolites called dehydropyrizidine alkaloids (DHPAs), which are then subsequently hydrolyzed to form dihydropyran derivatives (DHPs) (Figure 1D). DHPAs and DHPs have a strong electrophilicity and can quickly interact with macromolecules in cells, including DNA and proteins, forming pyrrole–DNA adducts, pyrrole–protein adducts, protein–DNA cross-links and protein–protein cross-links [12]. On the other hand, PAs can be hydrolyzed by nonspecific esterase enzymes into necine and necic acid. These necines and necic acids are non-toxic and can bind to polar molecules and be excreted in the urine (Figure 1D). Furthermore, recent studies have shown that PANOs can also cause hepatotoxicity, but the hepatotoxicity in humans is much lower than that of their corresponding PAs [13]. The metabolites of PANOs are generally non-toxic and are excreted in the urine, but excessive amounts of these metabolites can be transformed into toxic epoxides, damaging cellular functions [14]. Studies have shown that in rodents, PAs are primarily metabolically activated by the CYP3A and CYP2B subfamilies. CYP3A4 is involved in the metabolic activation of PAs in humans [15]. The metabolic activation of OTO-type PAs by CYP3A4 is greater than that of RET-type and HEL-type PAs. This results in the formation of more DNA or protein adducts and, therefore, a higher level of toxicity than RET- and HEL-type PAs [16].

3. Toxic Effects of PAs

3.1. Acute Toxicity

Since the early 19th century, it has been observed that livestock that consume plants belonging to the genera Heliotropium, Senecio, or Crotalaria experience slow emaciation and weakness, and has autopsies have revealed hepatocyte necrosis ^[17][18][63,64]. Acute poisoning by PAs can significantly affect the liver, leading to acute veno-occlusive disease characterized by hepatomegaly, hemorrhage, ascites, and even death in severe cases [1]. It has been reported that a 6-month-old female infant was diagnosed with hepatic veno-occlusive disease (HVOD) after the ingestion of PAs at approximately 0.8 to 1.7 mg/kg body weight (b.w.) per day for 2 weeks [1]. Similarly, a 2-month-old male infant ingesting 3 mg/kg (b.w.) PAs per day died after approximately 4 days ^[19][65]. Based on epidemiological data, the EFSA Panel on Contaminants in the Food Chain estimated that the daily intake of PAs ranging from 1 to 3 mg/kg (b.w.) per day for 4 to 14 days can cause acute toxicity [8]. It was shown that 7R-configured macrocyclic diesters of PAs, including retrorsine, seneciphylline, and senecionine, constitute the most potent group causing acute toxicity ^[20][21][22][66,67,68]. The 7S-synthesized Pas heliotrine and lasiocarpine show acute toxic effects similar to macrocyclic diesters ^[20][66]. Compared with 7S-heliotrine, 7S-lasiocarpine, and 7R-configured macrocyclic diesters of PA, 7R-echimidine, 7S-heliotrine, 7R-indicine, and 7R-intermedine had lower acute toxic effects ^[23][69]. Furthermore, the acute toxicity of PANOs generally appears to be lower than that of their parent PAs [13].

3.2. Cytotoxicity

The cytotoxicity induced by Pas is also structurally dependent. Li et al. (2013) evaluated the cytotoxicity of four PAs, namely, seneciphylline, senecionine, retrorsine, and riddelliine, on HepG2 cells using MTT and bromodeoxyuridine (BrdU) incorporation assays. MTT results showed that the IC₂₀ value of senecionine was 0.66 mM, which was 2.4, 1.9, and 2.1 times greater than those of retrorsine, senecionine, and riddelliine, respectively. Moreover, the BrdU assay showed similar results ^[24][70]. Reuel A Field et al. (2015) assessed the effects of 11 PAs on cell morphology, mitochondrial function, and lactate dehydrogenase (LDH) activity in CRL-2118 chicken hepatoma cells. MTT and LDH assays revealed that lasiocarpine had the greatest cytotoxicity, followed by riddelliine, heliotrine, seneciphylline, and senecionine. The cytotoxic effects of these PAs are characterized by significant cell swelling and vacuoles. On the other hand, the cytotoxicities of riddelliine N-oxide, senecionine N-oxide, and heliotrine N-oxide were lower than those of monocrotaline, intermedine, and lycopsamine ^[25][71]. In general, the cytotoxicity caused by macrocyclic diesters (RET and HEL types) with cyclic and acyclic ester structures, such as lasiocarpine, seneciphylline, and riddelliine, was greater than that caused by monoesters (heliotrine, lycopsamine, and intermedine) and PANOs. Current studies on the cellular toxicity mechanisms of PAs have focused mainly on oxidative stress and apoptosis, with oxidative stress being the primary cause of cytotoxicity. Several studies revealed that DHPAs and DHP not only bind to DNA and proteins to form adducts but also bind to glutathione (GSH) to form adducts. When GSH is depleted and not supplied in a timely manner, it leads to oxidative stress and consequent cytotoxicity ^[26][72]. Previous studies have shown that exposure of rat hepatocytes to adonifoline, monocrotaline, and clivorine significantly reduces intracellular GSH levels, and the activities of glutathione peroxidase (GSH-Px), glutathione reductase (GR), and glutathione S-transferase (GST) also significantly decrease ^[26][27][72,73]. In addition, apoptosis is another crucial factor contributing to cytotoxicity. Several studies have demonstrated that PAs can cause hepatotoxicity by activating apoptosis ^[28][74]. Clivorine treatment of L-02 cells resulted in a significant increase in intracellular caspase-3 enzyme activity and increased expression levels of cleaved PARP and cleaved caspase-3 proteins ^[29][30][75,76]. Furthermore, the induction of apoptosis and subsequent cytotoxicity can also be attributed to the downregulation of the antiapoptotic factor Bcl-xl and the upregulation of Fas expression ^[29][31][75,77]. Table 1 summarizes the IC₅₀ and IC₂₀ values for cytotoxicity caused by different types of PAs.

a IC₂₀ b IC₅₀; IC20 values refer to the calculated 20% inhibitory concentrations on cell viability; IC50 values refer to half-maximal inhibitory concentrations on cell viability.

3.3. Genotoxicity and Carcinogenicity

Animal experiments have shown that 1,2-unsaturated PAs can cause hepatocarcinoma in rodents ^[33][34][79,80]. The main reason is that DHPAs and DHPs generated by 1,2-unsaturated PAs in the liver bind to DNA to form adducts, resulting in abnormal biological processes and genotoxicity, which are also considered to be the main reasons for the carcinogenic effects of PAs ^[4][35][4,81]. In recent years, a series of in vitro assays have demonstrated that most 1,2-unsaturated PAs are genotoxic ^[35][81]. Williams et al. (1980) developed a quantitative detection method using radioactive precursors to detect DNA damage, and the results showed that lasiocarpine and riddelliine can induce nonprogrammed DNA synthesis in primary rat hepatocytes ^[36][82]. Monocrotaline, riddelliine, senecionine, and seneciphylline can induce DNA repair and HGPRT gene mutation in rat hepatocytes ^[37][83]. The number of DNA adducts induced by PAs with different structures results in differences in genotoxic and carcinogenic potentials. Xia et al. (2013) quantitatively analyzed the DNA adducts of nine PAs, including lasiocarpine, riddelliine, retrorsine, retronecine, heliotrine, clivorine, monocrotaline, senkirkine, and lycopsamine, in rat livers. The results showed that the levels of DNA adducts formed by macrocyclic diesters (retrorsine, riddelliine, and monocrotaline) and acyclic diesters (lasiocarpine) were much greater than those formed by monoester (glycosamine), HEL-type (heliotrine), and OTO-type clivorine (clivorine and senkirkine) ^[38][84]. Louisse et al. (2019) assessed the genotoxicity of 37 PAs in HepaRG cells using a γH2AX assay, and the results showed that the genotoxicity of cyclic diester and macrocyclic diester PAs was greater than that of monoester PAs and PANOs ^[39][85]. In general, the genotoxicity of PAs is characterized by higher levels of acyclic and macrocyclic diesters than monesters, and RET- and HEL-type PAs generally exhibit greater toxicity than OTO-type PAs.