The strength of phytotoxicity of ethanol extracts on test plants varied greatly with the plant parts (Table 1). Leaf and fruit extracts exhibited the strongest biological activity on all test species compared with other extracts, completely inhibiting seed germination of all test species at 0.05 g/mL. Stem and root extract exhibited moderate to weak phytotoxic activity on four test plants, while litter extract showed the least effect. Given the fact that fruit has a much higher biomass than leaf, it was chosen for further investigation.

Among all the organic extracts of X. italicum fruit, the chloroform extract consistently exhibited the most potent effect on four test species. Therefore, it was selected for further chromatographic separation. Amaranth seemed to be the most sensitive test species; petroleum ether and n-butanol extract significantly affected its seedling growth, meanwhile, their effects on the other three test species was much weaker (Table 2).

The structure of the purified active compound was identified as xanthinosin by comparing its spectral data with published literature [20,21,22] (Figure 2). We obtained 820 mg of xanthinosin from 8 kg of dried X. italicum fruits (0.01% yield). Xanthinosin has been previously isolated from X. italicum, X. macrocarpum and X. strumarium, and it was confirmed to possess antileishmanial and antitumor activity [22,23,24]; however, no phytotoxic activity has ever been reported for it. Due to the fact that xanthinosin also exists in X. strumarium, a native Xanthium plant, our current study thus cannot provide any evidence to support the “novel weapons hypothesis”, although it is possible that X. italicum does produce unknown novel allelochemicals [25].

The genus Xanthium is the richest source of xanthanolides [26]. Besides xanthinosin, dozens of other xanthanolide lactones have been isolated from Xanthium plants, and many of them have been proven to possess a variety of biological activities including antimicrobial, anti-ulcerogenic, anti-inflammatory, antitumor and plant growth regulatory activity [23,27,28,29,30]. So far, the majority of reported chemical constituents of X. italicum are sesquiterpene lactones. X. italicum contains many other xanthanolide sesquiterpenes besides xanthinosin. Kovacs et al. isolated four xanthanolides, i.e., xanthatin, 4-epixanthanol, 4-epi-isoxanthanol, and 2-hydroxyxanthinosin, from leaves of X. italicum with antitumor activity [29]. Marco et al. identified five xanthanolides from aerial parts of X. italicum including xanthinosin and 4-epixanthanol [22]. Tsankova obtained six xanthanolides from leaves of X. italicum, mainly xanthinin and xanthatin [30]. It is noteworthy to mention that xanthinin was reported to possess plant growth regulatory effects [31]. Another xanthanolide, 8-epi-xanthatin, which doesn’t exist in X. italicum, also exhibited plant growth regulatory activity [32]. In case both xanthinosin and xanthinin (or other xanthanolides, too) can be released to the environment and persist in soil at effective concentration for a certain period of time to affect other plants’ growth, it is necessary to investigate whether these co-occurring sesquiterpenoids could result in synergistic effects [33]. Although xanthinosin (possibly xanthinin and 8-epi-xanthatin, too) has the potential to be exploited as an environmentally friendly herbicide, still, there is a long way to go before it can be successfully commercialized. There are some issues that limit the utilization of naturally occurring substances directly as herbicides including their low yields, weak inhibitory activity, high toxicity, instability, short persistence in soil, etc.; on the other hand, there are successful cases of using natural products as new lead compounds to synthesize commercial herbicides. Examples include mesotrione, a synthesized analogue of leptospermone which is produced by the roots of the “bottle brush” plant Callistemon citrinus [34], and cinmethylin, a derivative of 1,4-cineole which is a natural phytotoxin found in the essential oils of a number of plants [35].

Xanthinosin didn’t affect seedling growth of any test plants when the concentration was low (32 µM). From 160 µM, significant inhibitory effects were observed on most test plants, especially lettuce, whose root elongation was inhibited by 58%. When treated with 800 µM xanthinosin solution, two dicot plants, i.e., amaranth and lettuce, seemed to be more sensitive than two monocot species, wheat and ryegrass. Root growth of amaranth and lettuce was inhibited by 78% and 89%, respectively, while the numbers were 69% lower for wheat, and 66% for ryegrass (Figure 3). When the concentration reached 4 mM, seed germination of all test species was mostly completely inhibited. Shoot growth of all species responded to xanthinosin in a similar way, but to a lesser extent (Figure 4).

IR spectra were measured on a Perkin-Elmer 783 spectrometer with KBr disks. UV spectra were taken on a Perkin Elmer Lambda 25 UV/VIS spectrometer. ¹H- and ¹³C-NMR spectra were recorded on a Varian Unity Inova instrument (at 400/100 MHz). EIMS were measured on a Micromass Platform EI-200 GC/MS instrument at 70 eV by direct insertion probe.

Plants and fresh litter of X. italicum were collected randomly from a roadside monoculture in Changji city, Xinjiang province in September, 2011 (N43°54′40.2″, E87°17′7.6″). Plants were identified by Professor Wenfang Wang from Linyi University. The plant materials were air dried in our laboratory at room temperature for two weeks.

Two dicot plants, amaranth and lettuce, and two monocots, wheat and ryegrass, were chosen as test plants for this assay. Seeds were surface sterilized with 0.5% HgCl₂ before use. Air dried plants of X. italicum were separated into stems, leaves, fruits, roots and litter. Five g of each part were ground into powder and soaked in 80% ethanol (100 mL) for 24 h to afford 0.05 g/mL extract. Three mL of ethanol extract of each plant part were then added to Petri dishes (9 cm diameter) lined with Whatman No. 3 filter paper. After complete evaporation of ethanol, distilled water (3 mL) was added to each Petri dish followed by addition of 10 seeds. Petri dishes were sealed with Parafilm to prevent water loss and stored in the dark at 25 °C. Seedlings were measured after 4 days of cultivation for amaranth, lettuce and wheat, and 5 days for ryegrass. Three replicates were made for all phytotoxic bioassays.

Fruits of X. italicum were selected for further isolation because of their strong phytotoxic activity and high biomass. Two hundred g of dried fruits of X. italicum were first extracted in ethanol for a week and then subjected to solvent-solvent partition using the following organic solvents sequentially: petroleum ether, chloroform, ethyl acetate and n-butanol, to afford corresponding extracts. The extracts along with the remaining water phase were filtered and evaporated to dryness in vacuo, weighed, and redissolved in 95% ethanol (for organic extracts) or distilled water (for the water phase) to give 1 mg/mL solutions. They were then tested for their phytotoxic effects at 1 mg/mL concentration using the same method for plant part extracts. Again, amaranth, lettuce, wheat and ryegrass served as test plants.

Eight kg of dried fruits were ground into powder and exhaustively extracted with 80% ethanol at room temperature for three weeks (3 × 40 L). The ethanol extract was concentrated under reduced pressure to yield a dark brown residue which was subsequently subjected to solvent-solvent partition into petroleum ether, chloroform, ethyl acetate, and n-butanol subfractions. Results from the bioassays indicated that the chloroform extract possessed the strongest inhibitory effect. Thus 10 g chloroform extract were fractionated on silica gel eluted with a step gradient elution (CHCl₃, CHCl₃-MeOH at 98:2, 96:4, 9:1, 8:2, 7:3, 6:4, 1:1, MeOH). Twenty-four major fractions were collected based on TLC profiles; they were all subsequently assayed for phytotoxic activity on lettuce at 1 mg/mL. Among them, fraction 5 and 6 completely inhibited seed germination of lettuce, and they were chosen for further study. Fractions 9–10, 16–17 also exhibited moderate biological effects. However, considering the low yield of these fractions, we gave up on them. Fraction 5 was then further purified by Sephadex LH-20 to afford an oily liquid that was later identified as xanthinosin; fraction 6, which contains lots of impurities, was subjected to silica gel column chromatography, after which we lost track of the active ingredient and thus failed to identify it in this fraction.

Phytotoxic activity of purified xanthinosin obtained from the above purification procedure was assessed by conducting bioassays with similar procedures at 8, 40, 200, 1,000 µg/mL concentrations against the same four test plants.

The significance of effects of extracts of X. italicum as well as xanthinosin on seedling growth of test species was first examined by ANOVA (p < 0.05) and then analyzed using Fisher’s LSD test at p < 0.05 level.