Occurrence of carcinogenic illudane glycosides in drinking water wells


 Background Ptaquiloside (PTA), caudatoside (CAU) and ptesculentoside (PTE) are carcinogenic illudane glycosides found in bracken ferns (Pteridium spp.) world-wide. The environmentally mobile PTA entails both acute and chronic toxicity and comparable risk might be associated with the structurally similar CAU and PTE. It is of great concern if these compounds are present in drinking water wells in bracken dominated regions, since they might pose a threat to human health. This study investigates the presence of PTA, CAU, PTE, and their corresponding hydrolysis products pterosins B (PtB), A (PtA) and G (PtG) in water wells in Denmark, Sweden and Spain. In total, 77 water samples from deep groundwater wells (40 – 100 m) and shallow water wells (8 – 40 m) were collected and preserved in the field, pre-concentrated in the laboratory and analysed by liquid chromatography-mass spectrometry (LC-MS). Results Deep groundwater wells contained neither illudane glycosides nor their pterosins. However, seven private shallow wells contained at least one of the illudane glycosides and/or pterosins at concentrations up to 0.27 µg L-1 (PTA), 0.75 µg L-1 (CAU), 0.05 µg L-1 (PtB), 0.03 µg L-1 (PtA) and 0.28 µg L-1 (PtG). Conclusions Detected concentrations of illudane glycosides in some of investigated wells exceeded the suggested maximum tolerable concentrations of PTA, although they were used for drinking water purpose. Contaminated wells were characterized by shallow depth, lower pH and electrical conductivity compared to deep groundwater wells where no illudane glycosides or pterosins were found.

United Kingdom [29]. In contrast to PTA, occurrence of CAU and PTE in the aqueous environment has been much less investigated. A recent study reported the first occurrence of CAU and PTE in surface waters up to 0.07 and 5.3 µg L -1 , respectively [30]. This finding suggests that assessments of water toxicity based on PTA only were underestimated. CAU and PTE are slightly more polar compounds and as stable as PTA in surface waters [30]. The reactive cyclopropyl group is marked with a circle [20] [32].
Recent attempt of modelling PTA fate in plant-soil matrix indicates that intense precipitation events during cold periods and with a fully developed canopy induce high probability of significant PTA leaching [33]. Furthermore, predicted high concentration of PTA in groundwater is also attributed to macropore transport [33]. Macropores have proven to be important for fast leaching of organic contaminants such as pesticides, bypassing biologically active layers, and thus resulting in low extent of biodegradation [34]. Leaching of pesticides is pronounced in fractured clayey till soils compared to sandy soils due to rapid well-connected macropores [35]. Hence, transport of organic contaminants from upper soil layers into deep groundwater is frequently observed for a variety of pesticides and herbicides [35]- [37]. It has never previously been investigated whether illudane glycosides are mobile and stable enough, to reach deep aquifers used for drinking water supply.
In Denmark, groundwater-based drinking water utilities mainly apply deep groundwater abstraction wells, and subsequently water is treated at waterworks before distribution to the consumers. In contrast to this, in some areas the consumers rely on private water wells, typically more shallow and the raw water might not be treated prior to consumption. While deep groundwater wells usually abstract the water from confined aquifers, shallow water wells typically abstract water from unconfined aquifers and may be affected by seepage water. Several studies suggest that shallow wells are more prone to contamination by organic contaminants [38] [39]. Pesticides are found in higher concentration in Danish shallow water wells as compared to deep groundwater wells [39]. Therefore, shallow water wells might also be more vulnerable to contamination by illudane glucosides.
It is currently unknown whether the bracken illudane glycosides PTA, CAU and PTE pose a threat to groundwater-based drinking water supplies. Thus, it is important to: 1) Investigate if the illudane glycosides are found in wells used for drinking water production, and subsequently 2) Identify the characteristics of vulnerable wells. To do so, we studied the presence of PTA, CAU and PTE and their corresponding pterosins PtB, PtA and PtG in a variety of water wells in bracken dominated areas in Denmark, Sweden and Spain, as function of well depth and water chemistry (pH and electric conductivity).

Solvents, chemicals and resins
Acids, bases and buffers (glacial acetic acid, formic and hydrochloric acids, sodium hydroxide, ammonium acetate) were all of analytical grade from Sigma-Aldrich (Germany). LC-MS grade acetonitrile was obtained from Merck Millipore (LiChrosolv hypergrade for LC-MS, Germany) and LC-MS grade methanol was purchased from Honeywell (LC-MS Chromasolv, Germany). MilliQ water (electrical resistivity 18.2 MΩcm, TOC less than 2 µg/L) was produced by Sartorius Ultrapure water system (Sartorius Stedim Biotech GmbH, Germany). Polyamide for column chromatography was from Fluka Analytical, Sigma-Aldrich Co (Germany). Oasis MAX (20 cc, 60 mg Sorbent, 30 mm particle size) was purchased from Waters (Milford, USA).

Preparation of analytical standards
No certified reference materials of PTA, CAU and PTE exist and therefore analytical standards were produced in-house. They were prepared from bracken plant material by the method described by Kisielius et al. [40].

Analytical procedure
Prior to LC-MS analyses, all water samples were pre-concentrated by solid-phase extraction (SPE) by a factor 250. The SPE method used in this study was optimized for PTA and PtB and validated for various groundwater samples [41]. A SPE protocol with more details is provided in Figure S2 of the supplementary material (SM). Together with each set of SPE samples, positive (for PTA and PtB spiked at concentration of 0.5 µg L -1 ) and negative controls (DI water) were assessed at the same time. In order to determine the efficiency of the SPE method for CAU, PTE, PtA and PtG, 50 mL of MilliQ water was spiked to a concentration of 0.5 µg L -1 for each compound. Following the SPE protocol, samples were processed in separate SPE cartridges and analysed by LC-MS (n = 2). The recovery results for each compound are presented in Table 1. The samples were analysed using Agilent 1260 Infinity HPLC System equipped with Agilent 6130 Single Quadrupole mass spectrometer by the method described by Kisielius et al. [40].  [40]. All samples with reported presence of illudane glycosides or pterosins were confirmed with spiking test and accurate retention times of the analytes. Furthermore, at least one of the confirmation ions was detected. More details on spiking test are provided in Figure S1. The concentrations of the compounds are reported as average concentrations measured in duplicates. In case the compound was detected in only one of the sample pairs the concentration was reported as a trace.

Identity of illudane glycosides and pterosins in bracken
Bracken plant material was collected from 11 locations during the water sampling campaign in order to quantify the content of illudane glycosides. From each location, 40 cm tip of a fully matured frond was collected in duplicates. Plant materials were dried in an ordinary plant press and grinded to powder prior to extraction according to Kisielius et al. [40]. Aqueous extracts were passed through a filter vial with 0.2 μm pore size membrane (Syringeless filter device Mini-Uniprep, GE Healthcare Life Sciences UK) prior to LC-MS analyses by which the contents of PTA, CAU and PTE and the pterosins were quantified using in-house produced standards [40].

Origin of water samples
Water samples from 77 wells in Denmark, Sweden and Spain were collected for analyses of PTA, CAU, PTE and the corresponding pterosins between June and October 2019. Water chemistry (pH and EC), well depth and more details on each investigated location is provided in Table S1. The majority of investigated wells belong to HOFOR utility (44 locations), the largest drinking water utility in Denmark, that is providing drinking water to approximately one million customers in the Greater Copenhagen area. The remaining investigated locations were selected based on presence of dense bracken ferns in their vicinity, diverse depth of water wells, geographical area and water type. In order to identify water wells with these features, an advertising campaign was performed through different public media prior to monitoring. Raw water (with no treatment) was abstracted from single wells. In case of HOFOR wells, mixed groundwater was collected (raw water mixed from several single wells). Only one location (number 9) does not represent a water well, but is seepage water collected from a creek (3 -4 m deep) just after a rain event. Wells 68 and 69 were sampled on two occasions, summer and fall 2019.
Most of the monitored wells serve for drinking water purpose. The 44 groundwater wells operated by HOFOR utility included in the study were deep groundwater wells (40 -100 meters). The remaining 33 wells were less than 40 meters deep. In the following text, HOFOR wells are classified as deep wells (44) and the others as shallow wells (33).

Water sample collection
Water samples were collected and preserved in the field according to the method of Skrbic et al. [41]. The preservation protocol was originally developed for PTA and PtB, but was further validated for CAU and PTE and their pterosins in this study. 50 mL groundwater was collected in amber glass bottles, buffered in the field by adding between 0.05 and 1 mL of 0.5 M ammonium acetate adjusted to pH 5 with glacial acid for stabilizing the PTA against hydrolysis. Samples were transported to the lab without ice for transportation up to 2 hours (otherwise, placed on ice and transported within 4 hours) and stored at 4 °C. SPE was performed in the lab within 72 hours after sampling. Final samples were stored in vials at -18 °C until LC-MS analyses. Filtration with sterile syringe through 0.45 μm pore size filters was performed only on two occasions, when very shallow/murky water was collected (locations 9 and 62). The samples were collected in duplicates, except for the HOFOR samples that were collected as single samples. The field blanks (DI water) were collected and treated as field samples during transport, preparation and analysis. The samples from Spain were packed with dry ice and delivered to Denmark by express mail (within 2 days).

Identity of illudane glycosides and pterosins in bracken
The most commonly found illudane glycoside in bracken plant material was PTA, which was detected at all eleven sampled locations, while at eight locations CAU could also be detected ( Table 2). The highest PTA concentration was measured at location 12 (3.23 mg g -1 dry weight) and the highest CAU content was detected at location 66 (0.42 mg g -1 dry weight). Neither PTE nor its degradation product PtG was detected in any of the analysed plant samples. Table 2. Measured content of illudane glycosides and corresponding pterosins in bracken ferns. The "-" represents no detection (< LOD) and "± " represents standard deviation (n = 2). Average coefficient of variation equals 33%, 37%, 21% and 44% for PTA, PtB, CAU and PtA, respectively. Location numbers correspond to the map in Figure 2.

Leaching of illudane glycosides after a rain event
The presence of illudane glycosides and the corresponding pterosins was determined in seepage water collected just after a rain event in a bracken dominated region (location 9). At this location, both PTA (1.27 µg L -1 ) and CAU (0.28 µg L -1 ) and their hydrolysis products were detected (Table 3).

Water samples
We sampled 77 water wells in Denmark (64), Sweden (5) and Spain (8) representing different geological settings. Neither bracken illudane glycosides (PTA, CAU and PTE) nor their hydrolysis products (PtB, PtA and PtG) were detected in 70 of the investigated wells (91%). However, in seven wells (9%) at least one of the compounds was detected (Figure 2). The deep groundwater wells did not contain illudane glycoside or pterosins. All seven wells in which at least one of the compounds was detected were shallow wells ( Figure 3). In terms of geography, one of the shallow wells with presence of illudane glycosides and/or pterosins was located in Denmark, four in Sweden and two in Spain (Figure 2). Apart from well 62, all these wells serve for drinking water purpose for humans and/or livestock (Table 3). Three shallow wells contained CAU in concentrations up to 0.75 µg L -1 (Table 3). However, PtA, the degradation product of CAU, occurred in six out of the seven wells, indicating even more frequent presence of CAU. PTA was present in three wells up to 0.27 µg L -1 , while two wells contained PtB. PTE was not detected in any of the wells, while its degradation product PtG was detected in three of them. Table 3. Water wells with presence of illudane glycosides and/or corresponding pterosins. Empty cells represent no detection (< LOD) and "± " represents standard deviation (n = 2). Average coefficient of variation equals 8%, 24%, 6%, 15% and 42% for PTA, PtB, CAU, PtA and PtG, respectively. Only location 9 does not represent a water well. In wells 62, 68 and 70 we detected both CAU and PtA. Similarly, PTA and PtB were detected in well 77. Detection of glycosides and corresponding pterosin in the same well validates the presence of glycosides. Well 62 was very shallow and not used for drinking purpose, but contained CAU and PtA at concentration up to 0.03 µg L -1 . Presence of PTA has previously been reported for this location [41]. Wells 72 and 77 showed presence of PtA, PtB and trace levels of PTA. The depth of these wells is unknown.
To include seasonal variations, we sampled well 68 and 69 both during summer and fall. In the summer, location 68 contained both CAU and PtA, while only PtA was found in the fall. At location 69, PtB and PtA were detected in the summer, but again only PtA appeared in the fall.

Characteristics of the water wells in relation to presence of illudane glycosides
To investigate determining factors for the presence of illudane glycosides in water wells, we analysed three characteristics of the sampled wells: depth, electrical conductivity (EC) and pH. First, for the subset of wells with known depths (59 out of 77 wells), the shallow wells with presence of illudane glycosides and/or pterosins were on average 19 meters deep (based on only three wells with known depths). Shallow wells without detected compounds were on average 30 meters deep, while deep wells were 40+ meters.
Secondly, EC in the deep wells were 940 ± 80 µS cm -1 , while conductivity in the shallow wells were 516 ± 79 µS cm -1 , showing different water chemistry in the deep versus the shallow wells (p-value << 0.001) (Figure 4). Additionally, shallow wells containing illudane glycosides and/or pterosins had lower average EC of 465 ± 181 µS cm -1 when compared to the other shallow wells which had an average EC of 530 ± 89 µS cm -1 but the difference was not statistically significant.
Finally, the wells with presence of illudane glycosides and/or pterosins had lower average pH (6.99 ± 0.22) when compared to the other shallow and deep wells (pH 7.16 ± 0.15 and 7.19 ± 0.05, respectively) ( Figure 4). However, this difference was also not significant.

Presence of illudane glycosides in drinking water wells and bracken
In this study, we detected illudane glycosides and their corresponding hydrolysis products in drinking water wells for the first time. Interestingly, CAU occurred more frequently than PTA, even though PTA is considered the most abundant carcinogen in bracken [43] [44]. Based on estimated LogP values and order of chromatographic elution, CAU is observed to be more polar than PTA (Figure 1) [40][31], and is thus more prone to be mobile in the environment, and potentially end up in water recipients. In a recent study, concentrations of CAU and PTE in surface water were also found to be similar or higher than PTA [30]. Additionally, the same study showed that content of CAU in bracken was higher in Sweden than in Denmark [30], which could explain higher CAU occurrence in wells in Sweden in this study.
In the present study, we did not detect PTE and PtG in any of the collected bracken plant samples, and in accordance with this none of the water samples contained PTE. In Northern Europe PTE has only been detected rarely in bracken and in very low concentrations compared to PTA and CAU [30]. Nevertheless, we detected PTE's degradation product PtG in three wells, indicating prior presence of PTE (at locations where no PTE or PtG were found in the bracken plant material). The content of illudane glycosides in bracken is decreasing from June to November [27][30], thus their hydrolysis products pterosins might still be present in water sampled in autumn while the glycoside (mother compound) is not present anymore, or the content is below LOD. The content of PTE in bracken populations is highly variable with random distribution [30]. Hence, bracken plant material collected for this study might not be the representative for PTE occurrence in the study area.
Maximum measured illudane glycoside concentrations in drinking water wells in this study were 0.27 µg L -1 for PTA and 0.75 µg L -1 for CAU. The maximum estimated tolerable concentration of PTA in drinking water is 0.002 µg L -1 (based on a model with one cancer incidence per million) [45], and the PTA analogue CAU, is presumably equally toxic [20]. The tolerable concentration were thus exceeded with more than a factor of 100 in these specific drinking water samples, indicating a health threat to humans or animals who are supplied with this drinking water.

Leaching of illudane glycosides after a rain event -pulse effects
Both PTA (1.27 µg L -1 ) and CAU (0.28 µg L -1 ) as well as their hydrolysis products were detected in seepage water coming from clayey sand soil in a small creek (location 9). The sampling was performed just after a rain event during summer when concentration of illudane glycosides in bracken is expected to be the highest [27] [30]. Even though the water sample at this location was not collected from a water well, this demonstrates potential of bracken toxins to leach quickly from plant material through the top soil layers.
A rainfall event had occurred prior to sampling from a drinking water well in Sweden (well 66), when PTA was detected (0.27 µg L -1 ). Similar situation was observed for locations 68 and 69 that were sampled on two occasions. During summer, CAU, PtA and PtG were detected after a rain event, while only PtA was found after a rain event in the autumn (location 68). Precipitation related pulses of higher PTA concentrations have already been observed for surface water and shallow groundwater [27][29] [41]. The presence of fissured bedrock in the sediment profiles could facilitate fast transport of CAU to deeper layers and explain CAU presence in a 40 m deep well (location 66 and 68). The conditions leading to an extreme toxin wash out event (pulse effect) are when intense precipitation takes place over the fully developed bracken canopy with maximum PTA content, moist soils and continous macropores/fracture systems stretching from the soil surface to groundwater [33]. Such an event is expected to be more likely for shallow wells due to better macropore connectivity over shorter than long distances.
This study was neither designed to cover temporal/spatial presence of illudane glycosides nor precipitation related pulses. Furthermore, by sampling the raw groundwater mixed from different wells in this study, a potential contamination could be diluted to below detection limit by water from non-contaminated wells.

Characteristics of the water wells in relation to presence of illudane glycosides
Water wells with presence of illudane glycosides and/or pterosins in this study shared several features. They were mostly privately owned, shallow and with low pH and EC. Electrical conductivity is typically higher in deep groundwater [46] as confirmed by this study. Opposed to that, in the shallow wells including wells with presence of illudane glycosides, we observed significantly lower EC that indicates a likely connection with rain and surface water (i.e. presence of seepage water). In addition, water in shallow wells (0 -40 meters) that contain illudane glycosides and/or pterosins had slightly lower average pH in comparison to deep groundwater wells (40 -100 meters). Lower pH (pH 4.5 -6.5) is known to have stabilizing effect on PTA hydrolysis [24]. Furthermore, a very short distance to high bracken biomass (< 10 m) and rain event prior to sampling supported illudane glycoside occurrence.
The majority of wells with presence of illudane glycosides (6 out of 7) serve for drinking water supply. In Denmark there are 55,000 similar private single-abstraction drinking water wells [47] and it is estimated that that 1 in 10 citizens of the EU receives drinking water from small and local systems, including private wells [48]. These systems are more vulnerable to illludane glycoside contamination and might not have any sort of water treatment.
Illudane glycosides and their hydrolysis products were not detected in deep groundwater wells investigated in this study (40 -100 m). This complies with the fact that PTA's half life in alkaline water is short, ranging from seconds to hours at pH > 7 [24]. Thus, in 80 days the PTA concentration will be reduced by 99.9% at pH 7 (or much shorter at higher pH) [24], indicating PTA is not sufficiently stable to reach deep groundwater aquifers as travel times from soil surface to deep aquifers take 25 to 100 years [46] when macropore transport does not intervene. In contrast, shallow wells investigated in this study showed presence of illudane glycosides and/or pterosins. Shallow wells are easily affected by younger seepage water that is more vulnerable to illudane glycosides contamination. The presence of cracks or macropores in soil profiles of shallow wells could support fast transport of illudane glycosides in relation to pulses [35] [49]. Similar observations were noticed when leaching of pesticides was investigated in clayey tills with up to 5-6 meters deep macropores [50]. In that study, leaching of pesticides into shallow groundwater was noticed and the transport occurred in macropores and was driven by precipitation events [50]. Hence, leaching of illudane glycosides to shallow drinking water wells is possible, it is site specific and probably occurs in pulses.

Conclusions
In this study, we detected at least one illudane glycoside or pterosin from bracken fern in seven of the 77 studied wells. Thereby we demonstrated that leaching of illudane glycosides to drinking water wells is possible and could pose a threat to human health. Deep groundwater wells did not contain bracken illudane glycosides, probably due to the long travelling time to these aquifers. However, we detected bracken illudane glycosides in shallow wells, which could be due to intrusion of seepage water. A dense bracken biomass aboveground is a risk factor, especially in relation to rain events.
Six out of seven wells with presence of illudane glycoside were used for drinking water supply. The concentrations in these drinking water wells were 0.27 -0.75 µg L -1 , which violate the maximum estimated tolerable PTA concentration in drinking water by approximately 100 -300 fold. In this study, CAU was more frequently detected in drinking water wells than PTA. Hence risk assessments are likely to underestimate water toxicity if they are based on PTA only and should also include CAU and PTE. In particular, these results are of great importance for consumers supplied by shallow wells in bracken dominated regions.   Solid-phase extraction (SPE) protocol Figure S2. SPE protocol [41]. Figure S3. Soil map of Denmark according to the Danish Soil Classification [42].