Analysis of pyrrolizidine alkaloids in Crotalaria species by HPLC-MS/MS in order to evaluate related food health risks

Abstract

Pyrrolizidine alkaloids (PAs) are one of the most significant groups of plant toxins in the world and are an important cause of poisoning in livestock, resulting in significant financial and production losses each year (Kellerman et al. 1996). Pyrrolizidine alkaloids may also enter the human food chain as contaminants of grains, via animal products such as milk, eggs and honey or may be consumed as constituents of herbal medicines (ANZFA 2001). Not all PAs are toxic. Pyrrolizidine alkaloids affecting human health are the esters of 1,2-unsaturated hydroxymethyl dehydropyrrolizidines (DHP). Before it can be converted to DHP, PAs need to have certain essential features, which include an unsaturated 3-pyrrole ring, one or two hydroxyl groups attached to the ring, one or two ester groups and a branched acid moiety (Mattocks 1986). These compounds can be metabolized in the liver to nucleophillic pyrroles which cause damage to hepatocytes (Winter and Segall 1989). Although the involvement of PAs in the development of hepatic veno-occlusive disease is well established (Bras et al. 1961), there is still uncertainty concerning the consequences of long-term, low-dose exposure in humans. Exposure to PAs through the use of herbal remedies may also be a contributing factor to the high rates of liver cancer and cirrhosis seen in Africa (Steenkamp et al. 2000). Crotalaria spp. are known to contain toxic PAs and various incidences of human poisoning through contaminated grains have been recorded in the scientific literature (IPCS 1989). Legislation controlling the allowable levels of toxic seeds in grains in South Africa is generally much stricter than in many other grain producing countries. The Soybean and Sunflower Forum recently commissioned a study (Eloff et al. 2003) to review published and unpublished information on toxic seed that could affect human health in South Africa and to make recommendations accordingly. Crotalaria sphaerocarpa is one of the problem plants discussed in the review and is apparently the only species which regularly contaminate grain in certain areas in South Africa. There is uncertainty at present about the number of these seeds that should be allowed in grains and the threat that this may pose to human health. Based on the review a provisional recommended level of 10 seeds of C. sphaerocarpa per 10 kg of grain was proposed as an approximated safe level in the report. As emphasized by the authors (Eloff et al. 2003), this absolute level is based on assumptions that must still be tested. As a follow-up on the report, a sensitive LC-MS/MS method for the determination of toxic PAs in plants was developed in this study. The characteristic fragments produced by 1,2-unsaturated necine bases under specific MS/MS conditions were used to discriminate between the toxic and non-toxic PAs. The concentration of these PAs were then determined using multi-reaction-mode experiments. Quantitative results were calculated against a retrorsine calibration curve and expressed as µg retrorsine equivalents per gram plant material. Various extraction methods described in the literature were investigated. A final liquid-liquid extraction method was used to extract unsaturated PAs from small amounts (about one gram) of milled plant samples. Recoveries from spiked lucerne samples were 98% for retrorsine and 105% for monocrotaline. To determine the applicability of the LC-MS/MS method the unsaturated PA content of C. laburnifolia and C. dura were investigated. Crotalaria laburnifolia, which is regarded as non-toxic, contained low concentrations (< 20 µg.g-1) of unsaturated PAs. Crotalaria dura, on the other hand, is known to be toxic to livestock and the concentration of unsaturated PAs was significantly higher (585 µg.g-1). The toxic PA content of Senecio inaequidens was also determined after an incident of livestock poisoning. The plant material contained very high concentrations of retrorsine (11.5 mg.g-1) and senecionine (0.5 mg.g-1) which were also present in the rumen content collected post-motally. These results confirmed the suspected toxicity of S. inaequidens. The LC-MS/MS method was also used to follow variations in unsaturated PA content in C. sphaerocarpa plants during the growing season. Pyrrolizidine alkaloids were present in the roots of the growing plants as N-oxides and also found in the mature aerial parts, where it was present mainly as the basic alkaloids. The method was used to determine the concentration of unsaturated PAs, in various C. sphaerocarpa seeds from different locations, in order to calculate the allowable level of C sphaerocarpa seed in maize. Of all the seed samples analyzed, the highest unsaturated PA concentration found was 150 µg.g-1. The allowable level of seed was calculated using this result and was found to be 656 seeds per 10 kg maize, based on the Australian and New Zealand Food Authority level of 0.1 µg.kg-1.day-1. If these results are confirmed with systematic statistical samples of C. sphaerocarpa seed from different grain production areas, the allowable level could be increased substantially. This may have an economic benefit to grain producers.