1. Introduction Cyanobacteria are ubiquitous prokaryotic photoautotrophs, formally known as blue-green algae, attracting increased scientific and public attention, due in part to the suite of secondary metabolites they produce. Many of these low molecular weight compounds are toxic, causing adverse effects on humans, animals, plants and Protoctista [1,2,3]. Cyanotoxins can affect a variety of biological pathways, including hepato-, neuro-, cyto-, dermato-, geno-toxicity and carcinogenicity [4,5], with the neurotoxins including both anatoxins and saxitoxins. Saxitoxins (STX) are alkaloid neurotoxins, blocking voltage-gated sodium channels, causing nausea, leading to respiratory arrest and death in severe cases (Paralytic shellfish poisoning) [6]. Known STX producing cyanobacteria genera are Aphanizomenon, Dolichospermum (Anabaena), Lyngbya, Cylindrospermopsis and Planktothrix. Anatoxin-a (ATX), 2-acetyl-9-azabicyclo[4.2.1]non-2-ene [7], is a low molecular weight bicyclic secondary amine, neurotoxin that irreversibly binds to nicotinic acetylcholine receptors, causing loss of coordination, muscular fasciculations, convulsions and death by respiratory paralysis [1,8]. Known ATX producing cyanobacteria genera include Dolichospermum (Anabaena), Aphanizomenon, Phormidium and Microcoleus [9,10,11]. In addition to ATX itself, other analogues have been reported such as homoanatoxin-a (HTX), dihydroanatoxin-a (dhATX) and epoxyanatoxin-a (epoxy-ATX) [12,13,14,15,16,17]. (Figure 1).In freshwaters, neurotoxins are prevalent and by extension, have attracted a high level of interest from the scientific community and water resource managers. Additionally, data suggests that human activities, namely global warming and eutrophication are expected to increase the number of cyanobacteria [18,19], therefore these toxins represent a serious threat to human, animal, and ecosystem health [3].In addition to human health impacts through ingestion of drinking water and recreational exposure, neurotoxic blooms of cyanobacteria have caused animal poisonings globally for many years [1,3,18,20,21,22,23]. Animal deaths, due to the anatoxin-producing Anabaena were recorded in North America in the 1920s and 1950s [24]; the 1952 bloom at Lake Storm, Iowa, caused the deaths of several thousand birds and mammals. In Canada during the 1970s there were a series of cattle deaths attributed to ingestion of cyanobacterial scums. Dolichospermum flos-aquae (formerly Anabaena flos-aquae) was cultured and administered orally to calves to determine a minimum lethal dose of 360 to 480 mL of concentrated culture in a 60 kg animal. After 30 min under veterinary supervision, the calves suffered muscle fasciculations, paralysis and respiratory collapse and required intubation after 4 h, and they were maintained for 28 h before the experiment was terminated. The causative compound ATX was identified and isolated and found to have a LD50 of 50 µg kg−1 by MBA following intraperitoneal (i.p.) injection [25]. Other fatal animal poisonings linked to anatoxins have been reported in Canada, including cattle and waterfowl [26,27]. Dogs also show susceptibility to anatoxins poisoning. It is thought that the musty taste and odours of cyanobacteria attract the attention of dogs scavenging habits, and they like to swim allowing cyanobacteria to stick to fur followed by ingestion or through grooming [28]. An investigation into dog and other animal deaths near a lake in South Dakota, reported a bloom dominated by Dolichospermum flos-aquae, and identified ATX-a as the causative compound by HPLC. [25]. Benthic cyanobacteria were also accountable for two dog deaths from a river in France, with ATX identified from Phormidium favosum, by liquid chromatography coupled to tandem mass spectrometry [22]. Anatoxins in Finnish lakes have also been linked to reports of animal poisonings including both cattle and dogs [29,30], as well as reports of ATX and dihydroanatoxin-a (dhATX) being linked to dog fatalities in Germany [15]. Outside Europe, in the US, between 2007 and 2011, as part of the Harmful Algal Bloom-related Illness Surveillance System (HABISS), 67 cases of canine cyanotoxicosis events in freshwaters were reported, 58% were fatal and of those 32% were ascribed to ATX [21]. Multiple reports also exist of anatoxin-producing cyanobacteria in New Zealand water bodies, resulting in dog deaths [17,31,32].Closer to the UK, ATX has been reported in several freshwater lakes in Ireland, with links to fatal canine poisonings in multiple regions [33,34,35]. In the UK itself, five cases of cyanotoxicosis in dogs were reported after drinking from a Loch in Scotland; four cases were fatal by neurotoxicosis within 30 min. The causative cyanobacterium was discovered to be benthic Oscillatoria. ATX was identified and extracted from shoreline samples, stomach contents and cultured Oscillatoria, using liquid and gas chromatography [36,37]. Routine monitoring is not conducted in inland UK water bodies, so hazard assessment involves primarily the responsive microscopy testing of water samples following reports of cyanobacterial presence from public or water-body owners, with action taken to restrict water access when cell densities exceed threshold limits [38]. Other schemes exist including the citizen science reporting of potentially toxic cyanobacterial blooms throughout the country, which can help publicise the presence of potentially toxic blooms [39]. Nevertheless, without a systematic risk management system in place, the cyanobacterial blooms occurring on a regular basis during the warmer months of the year, are likely to continue to result in exposure incidents affecting animal as well as potentially human health. Indeed, multiple reports of dog poisonings following cyanobacterial-exposure are found every year in the UK media and news reports, e.g., [40,41,42,43], which has resulted in a range of advisory services published on-line to warn owners of the risks from blue-green algae in some freshwater locations [44,45,46].

In June 2022, an incident was reported involving the suspected poisoning of a dog at a freshwater reservoir in SW England. The dog had been one of seven dogs in a visiting group, walking along the side of the water body. The affected dog was observed to lick or nibble upon some stranded material including a dead fish at the shoreline. Symptoms of paralysis were observed within minutes of exposure, with death of the dog occurring after 45 min. Consequently, an investigation was conducted to establish the cause of death with a view to determining if cyanobacterial neurotoxins presented a risk to animal and human health. Specifically, analysis was conducted to establish the potential presence of either anatoxins and/or saxitoxins in postmortem samples obtained from the deceased animal.

3. DiscussionIncidents involving the poisonings of dogs following exposure to anatoxins whilst not routinely confirmed, have been reported worldwide in recent years. Whilst many reports exist in the media, few of these have been scientifically linked to cyanobacterial sources and verified through examination of clinical or post-mortem samples. Nevertheless, dog deaths have on occasions been unambiguously linked to the presence of anatoxins, including ATX and associated analogues (dhATX, HTX) in water and cyanobacterial sources such as benthic mats [15,50]. Anatoxins are also known to be distributed globally, throughout a wide range of climatic environments. In Europe, 25% of German waterbodies were found to contain ATX, with toxin presence also confirmed in Finland, Ireland and France, with the highest toxin prevalence occurring in benthic cyanobacterial mats, as opposed to within surface waters [16,22,33,34,51,52].In this study, a 32 kg male dog exhibited signs of paralysis within 10 min of suspected oral exposure to a dead fish and/or other nearby biota situated on the shore of a freshwater reservoir in SW England in June 2022. Death occurred within 45 min of exposure, despite attempts at CPR. The dog was one of seven present in the location, all of whom had played along the shoreline, with the affected dog being the only one observed to have encountered the dead fish. The other dogs exhibited no signs of sickness or distress. Anatoxins are known to be rapidly absorbed and distributed throughout the body after exposure. Administration of lethal doses of ATX in animal studies results in death within minutes due to muscular paralysis and asphyxiation [53,54], which fits well with the observations in this case [16]. Consequently, the symptomology and poisoning showed evidence for ingestion of anatoxins, with the source of the toxins likely to be the fish and/or other surrounding biota, as opposed to direct exposure to toxins in the water.Correlating concentrations of toxin in post-mortem samples to toxicity exposure levels is very difficult, given the lack of data regarding metabolism, including absorption and excretion rates as well as potential biotransformation reactions. In terms of lethal doses of anatoxins, the LD50 for oral ingestion has been suggested as >5 mg/kg body weight [16], and between 1–10 mg/kg body weight [53,54] which for a 32kg dog equates to >32–320 mg of ATX. The toxin concentrations quantified in this study using HILIC-MS/MS were approximately 0.6 mg/mL and 1.0 mg/g of ATX and 6.0 mg/mL and 21 mg/g in urine and stomach contents, respectively (Table 3). Noting the sample amounts received were 2.16 mL urine and 0.083 g stomach contents, representing a small proportion of the total samples originally taken, these concentrations equate to approximately 1.3 mg (urine) and 0.08 mg (stomach) of ATX, with 13 mg (urine) and 1.75 mg (stomach) of dhATX. With dhATX shown to be more than four times more toxic than ATX [17], the combination of both toxins equates to a total of approximately 60 mg of ATX equivalents. Given that other ATX analogues were also detected, albeit with unknown toxicity equivalents, and the toxins would have likely distributed to other tissues following absorption, and that the sample sizes received at Cefas were only sub-samples of larger sample volumes used for general extensive toxicological screening including organic contaminants, there is strong evidence that the toxins were ingested at levels within the 32–320 mg range of LD50 limits for ATX. Such doses fit with multiple literature reports where high concentrations of ATX have been quantified in many different species of cyanobacteria, with ATX concentrations reaching 13 mg/g cyanobacteria in laboratory cultures and environmental samples from Finland and Canada [10,55], as well as other examples quantified in environmental samples from France, Ireland, New Zealand, Germany, Kenya and Iran [16]. A summary of ATX concentrations quantified to date in dog stomach samples is summarised in Table 4, alongside detection of other ATX analogues, and concentrations in related environmental samples. As shown, concentrations were reported in approximately half of the samples analysed, with values varying hugely. ATX concentrations reach as high as 36 mg/g stomach contents in samples from New Zealand, over 30,000 times higher than levels quantified in this study. Conversely, the stomach contents of dead dogs from Germany contained as low as 0.025 µg/g ATX, significantly lower than the amounts determined here.Clinical samples from the deceased animal included urine, blood, a blood clot of unknown origin and stomach contents. These were extracted and analysed using two different mass spectrometric methods; HILIC-MS/MS for targeted quantitation and LC-HRMS for accurate mass confirmation and quantitation. Reverse-phase LC-MS/MS, used for analysis of lipophilic cyanotoxins, was also modified for detection of hydrophilic toxins, but retention times of these target analytes were too low, so quantitation was not performed. SRMs generated by HILIC-MS/MS for hydrophilic cyanotoxins showed no detectable peaks for CYN or any saxitoxins, but did show the clear presence of ATX, most notably in the urine and stomach content extracts. Chromatographic separation between ATX and Phe was achieved with the refined HILIC gradient, removing the potential for false detection or over-estimation of the former, given the shared SRM transitions associated with both compounds due to the isobaric nature of both precursor and product ions [62] and previous false reports of human fatality published following LC-MS detection of Phe [63]. SRMs for other known ATX analogues were also generated, with none detected with the exception of clear SRM peaks for dhATX in the urine and stomach samples. Further confirmation of ATX and dhATX was obtained through the LC-HRMS analysis of the same extracts, with evidence generated for XIC detection of both ATX and dhATX in all four samples, with the urine and stomach extracts showing concentrations within the calibration range of the method. Specific confirmation was generated through both accurate mass detection of expected base peak ions and with the confirmation of product ions matching those utilised for targeted HILIC-MS/MS detection and those reported previously in the literature. The LC-HRMS approach also allowed differentiation between the isobaric ATXs and Phe, through acquisition of readily distinguishable [M+H]+ XICs at 166.1154 and 166.0790, respectively [17,63]. Notably, the results demonstrated the high prevalence of dhATX in the clinical samples, with the urine and stomach contents containing ten times the concentrations in comparison to ATX (Table 3). These findings agree with those reported recently in Germany following canine fatalities linked to consumption of benthic cyanobacterial mats [15] as well as prior reports suggesting dhATX is the congener most likely to cause dog deaths [31]. DhATX is known to present intracellularly within cyanobacteria, evidencing natural production [17,64], but is also considered as the major degradation product of ATX. The tentative identification of the epoxy and keto ATX degradation products further confirms the presence of ATX in these samples, and highlights the requirement for additional analytical standards for these compounds to aid in confirmation and quantitation.To date, the maximum concentration of ATX dissolved in lake waters has been 444 µg/L, in samples taken from a lake in Ireland [33]. Whilst it is not inconceivable for concentrations to exceed this level, given the absence of routine monitoring for these toxins with only occasional research surveys conducted, it seems highly unlikely that toxin levels were high enough in the water to result in poisoning through drinking lake water alone. As shown in previous studies where extremely high concentrations of toxins were confirmed in benthic cyanobacterial mats [15,32,65,66], it seems that in this study the exposure occurred through contact with such material, including potentially cyanobacterial debris present on the surface of the dead fish. The absence of any illness in the other dogs at the lake that had only contact with the reservoir water, certainly seems to support the hypothesis that the affected animal was exposed to much higher doses of toxins, noting that cyanobacterial mats and scums can contain a thousand to a million-fold increase in cyanobacterial cell numbers in comparison to normal cyanobacterial cell populations [67]. Indeed, it was noted at the time of the incident that contact with some kind of material at the edge of the water was seen, so potentially this may have been ingested. Unfortunately, water samples were not taken at the time of the incident by the authorities, and on returning to the scene to collect the fish, the remains had been removed. Given the incident was not notified to Cefas until a month later, water samples were not taken given that the samples would not have been representative of the water present at the time of the poisoning, particularly as rapid decay of ATXs has been reported in biological and environmental matrices [63,67,68]. Similarly, no collection of algae was made after the incident to help support the hypothesis that benthic cyanobacteria may have been a cause. However, from the evidence collated here, there is clear confirmation of anatoxin poisoning in the dog as a consequence of exposure to the neurotoxins at the lakeside.

Consequently, the chemical analysis conducted using two independent mass spectrometric methods provided excellent evidence confirming the hypothesis of ATX and dhATX-related poisoning in the affected dog, with the likely source being material present on or within the dead fish at the water’s edge, as opposed to toxins dissolved in the lake water.