*2.3. Phycology*

Intact colonies of *Microcystis* were observed in the canine vomitus sample, confirming exposure to cyanobacteria (Figure 2). Other cyanobacteria were not observed. While a water sample was not submitted in conjunction with this exposure event, reports from the Florida Department of Environmental Protection (FDEP) support the vomitus phycological observations match the dominant genera (*Microcystis*) present in the St. Lucie River HAB at the time of exposure [42].

**Figure 2.** A *Microcystis* colony observed in the canine vomitus sample acquired within 6 h of exposure. The micrographs are at 400× with brightfield (top), phase-contrast (middle) and epi-fluorescence (bottom). The scale bar represents 25 μm.

#### *2.4. Adda Microcystin*/*Nodularin (MC*/*NOD) Levels*

Results from the three microcystin/nodularin (MC/NOD) analytical tests on all specimens are reported in Table 3. Enzyme-linked immunosorbent assay (ELISA) values (representing freely extractable Adda MCs/NODs) were, in general, supported by the MMPB (2-methyl-3-methoxy -4-phenylbutyric acid) data (representing total Adda MCs/NODs). This indicates that the Adda ELISA is able to react to conjugated forms of microcystin, as those excreted in urine. However, total Adda MCs (measured as MMPB) were higher in organ specimens, supporting that the MMPB method accounts for some fraction of protein bound MCs. The only non-metabolized specimen, the vomitus, had total Adda

MCs (MMPB) measured at 46,000 <sup>±</sup> 8000 ng g−1, confirming a high dose of exposure. The vomitus sample was significantly diluted with meal items (>30 grams submitted), so the dose was in excess of 1,380,000 ng total MCs. Analysis of the vomit using ELISA resulted in a lower level of MCs (25,000 <sup>±</sup> 1800 ng g−1), which may be due to MC losses to the matrix (high protein dietary items), partial MC degradation or due to differences in analytical technique used. Targeted LC-MS/MS of 19 MC variants and NOD-R confirmed the presence of 7 MC variants (Figure 3) and the absence of NOD-R. The dominant variant present was MC-LR (14,000 <sup>±</sup> 100 ng g<sup>−</sup>1), followed by [Dha7]MC-LR (170 <sup>±</sup> <sup>21</sup> ng g−1), MC-HilR (140 <sup>±</sup> 28 ng g−1), [Asp3]MC-LR (82 <sup>±</sup> 0 ng g−1), MC-LY (23 <sup>±</sup> 16 ng g−1), MC-LW (18 <sup>±</sup> 3 ng g<sup>−</sup>1), and MC-LF (14 <sup>±</sup> 2 ng g<sup>−</sup>1). Other variants were detected in a non-targeted MS scan, but due to a lack of standards available to verify identities or levels, they are not reported here (work ongoing). Targeted MCs in the vomit accounted for 56% of the ELISA measurement and 30% of the total Adda MCs by MMPB. The results of the targeted MC analysis were key to the decision to target MC-LR in the remaining collected specimens.

**Table 3.** Results of the Adda MC/NOD analyses conducted on all matrices. Data is reported in parts per billion, with organ and hair samples reported by weight (ng g<sup>−</sup>1) and liquid samples by volume (ng mL<sup>−</sup>1). Data is reported <sup>±</sup> the standard deviation for samples with duplicate extractions. MMPB analysis represents total Adda MCs, ELISA represents freely extracted Adda MCs and MC-LR is LC-MS/MS analysis of the freely extracted variant. PE = Post exposure. UE = Unexposed individual.


<sup>1</sup> above the limit of detection, but below the limit of quantification. 2. false positive data (refer to text for explanation).

MC-LR was detected in all the tested specimens from exposed animals with exception of the hair (C-GR #1) and one blood sample (C-GR #2). The negative control specimens were all below detection for targeted MC-LR. The remaining MC-LR concentrations (metabolized specimens) only accounted for 0.5%–2.8% of ELISA data and 0.3–2.3% MMPB data.

The MMPB data was integral to results interpretation, as the data is representative of total Adda MCs (and nodularins, when present), regardless of form (free, bound, partially degraded). Since the approach to quantification was pre-oxidation spiking (standard addition), confidence in MMPB data was higher than that of ELISA. The spike returns in Table 3 are shown only for reference and are used to illustrate how the method is impacted by various complex matrices (as compared to MC-LR oxidized in water). A low level of detection (sub-ppb) was achieved for urine and blood samples. The strength of this approach was illustrated when 2.6 ng mL−<sup>1</sup> total MCs was detected in the urine sample of C-Pom over 3 weeks after suspected exposure. While this level could not be confirmed using alternate techniques due to higher test MDLs, the MMPB urinalysis of another exposure case (C-GR#2) supported the continued excretion of Adda >60 days post exposure (Figure 4). In addition to renal elimination, it was determined that hair was a potential route of elimination. MMPB chromatograms of a dog hair sample collected 72 days post exposure with overlaid pre-oxidation spikes of MC-LR can be viewed in Figure 5, with a negative control sample. A total MCs of 180 ng g−<sup>1</sup> (dry weight) was determined to be present in the C-GR#1 hair specimen.

**Figure 3.** The LC-MS/MS chromatograms of MC variants confirmed present in the C-GR#2 vomit sample with a sum of 14,000 ng g−<sup>1</sup> MCs. The MC-LR scale is on the right due to high levels detected in comparison to the other variants. MC-LR > [Dha7]MC-LR > MC-HilR > [DAsp3]MC-LR > MC-LY > MC-LW > MC-LF. Transitions monitored are reported in Table S1.

**Figure 4.** Data derived from the log of total Adda MCs (by MMPB) of the urine collected from one of the surviving dogs (C-GR#2) plotted against days post exposure. Urine was collected within 1 day from initial exposure event, and the animal continued to excrete MC metabolites >60 days post exposure. The MDL for total MCs in urine was determined to be 0.2 ng mL<sup>−</sup>1.

**Figure 5.** MMPB LC-MS/MS chromatograms (*m*/*z* 207→131) showing sample peaks (blue) overlaid with their paired pre-oxidation MC-LR spikes (red) at 200 ng g−<sup>1</sup> for A (exposed dog C-GR#1) and 100 ng-g−<sup>1</sup> for B (unexposed dog C-CBR). Total MCs detected in the C-GR#1 hair at 72 days post exposure was determined to be 180 ng g<sup>−</sup>1, illustrating hair as a potential route of MC elimination.

The testing of the organs of the deceased canine (C-SP) revealed that the kidney had higher total Adda MCs (MMPB), free MCs (ELISA) and MC-LR when compared to the liver. In similar fashion, the urine was higher than the bile, with the least MCs measured in heart blood. Regardless of method used to test the specimens, the levels of MCs were as follows: urine > bile > kidney > liver > blood. Figure 6 illustrates the MCs detected using the 3-techniques in the specimens collected from the deceased dog (C-SP).

**Figure 6.** Results showing the different specimens collected from the dog (C-SP) that succumbed to intoxication 2 days post exposure and reported as ppb (ng mL−<sup>1</sup> or ng g<sup>−</sup>1). As illustrated, the urine contained the highest amounts of MCs (all methods), followed with the bile, kidney, liver and finally heart blood. All specimens were collected post-mortem.

It should be noted that the method of quantification for each technique is different, with interpretations of spike returns and final data essential to understanding results. The Adda ELISA method employs an external curve (certified reference material of MC-LR) for quantification, with the only assessable quality controls being replicate extraction data (reported as the average with standard deviations in Table 3) and spike returns. Both high (>130%) and low (<70%) spike returns were obtained from ELISA data. Exaggerated spike returns observed with ELISA may indicate overestimates of natively reported MCs, as supported by MMPB data in the bile sample. In contrast, low spike returns (e.g., hair and liver specimens), indicate matrix inhibition. The same extracts (and

MC-LR spikes) were analyzed by both Adda ELISA and LC-MS/MS and can be directly compared. The MC-LR spikes analyzed by LC-MS/MS returned 57% to 111%, while the same spikes were 0% to 211% when analyzed with the ELISA, further supporting that the ELISA resulted in both exaggerated and inhibited responses.

Further complicating ELISA data interpretation was the positive (≥15 ng g−<sup>1</sup> MCs/NODs) response elicited from a negative control liver sample. The negative control specimen was collected from an animal without a recent exposure to waterways or cyanobacteria. Furthermore, MMPB analysis confirmed that Adda MCs/NODs were not present over 4 ng g−1, which is more sensitive than the ELISA MDL of 15 ng g<sup>−</sup>1. Sample extract clean-up (SPE), dilution (100-fold) and hexane washing were not sufficient in mitigating false positive data for the negative control liver sample. In comparison, prior to hexane washing, the positive liver and kidney samples were 260 and 1105 ng g<sup>−</sup>1, respectively. Post hexane washing, the levels were similar at 258 and 1080 ng g−1, with little change to spike returns (193→201%, 106→111%). Hexane washing did lower the negative control liver assay responses (analyzed with 100-fold dilution) from 1.78 to 0.52 ng mL−<sup>1</sup> (correlating to 178 to 52 ng g<sup>−</sup>1), and increased the spike return from 12% to 33%. However, this was not sufficient in preventing the false positive assay result.

#### **3. Discussion**

Harmful algal blooms (HABs) can be hazardous to humans, animals and the environment [43,44]. Reports of bloom formation with toxin production are becoming more frequent, likely due to anthropogenic influence leading to nutrient accumulation in waterways and changing global environmental conditions [45–47]. The 2018 *Microcystis* bloom in the St. Lucie River is a fitting example of the sublethal and lethal toxic effects that may be observed in dogs exposed to HABs.

The clinical signs and pathology of blue-green algal toxicity can mimic a handful of toxicoses, including, but not limited to, xylitol, sago palm, rodenticide, *Amanita*, and ricin. Rapid detection of the inciting toxin can be problematic in clinical settings. Samples of tissues may be difficult or dangerous to obtain and few laboratories provide cyanotoxin testing for animal tissues. In this work, the source water was not available for testing, but *Microcystis* and MCs were determined present in the source water during the event by the Florida Department of Environmental Protection [42]. The vomit from one of the six dogs did have measurable MCs, with MC-LR found to be 97% of the total targeted MCs. This observation is similar to other work analyzing MCs in the St. Lucie River [48], where MC-LR was 98% of the total targeted variants by LC-MS/MS. Specimens tested in past dog exposure events that succumbed to MC intoxication included the source water, feces [4], liver [5], and vomitus [5,6]. In this work, clinicopathological observations, post-mortem pathology, and MC testing were all used to confirm that the exposed dog morbidity was a result of MC intoxication and findings correlate with previous reports [4–6,49]. This work illustrates the most comprehensive multidisciplinary reporting on microcystin induced animal morbidity and mortality following a HAB exposure event. Confirmation of cyanobacteria intoxication is rare in a clinical setting due to a multitude of reasons. Improved antemortem methods for testing is essential for diagnosis and timely treatment recommendations, especially in the face of a multi-animal exposure event.

For the first time, urine was used to diagnose an ongoing MC intoxication event. The importance of this finding cannot be overstated, as urine is a specimen that can be easily, quickly and non-invasively collected by veterinarians faced with a suspected toxicosis, even if several days or weeks have passed. The depuration of MCs, even months after exposure, is likely due to the accumulation of MCs in tissues and bound to proteins. The MCs are then slowly released over time, with the Adda being conserved. The achieved sensitivity and broad specificity of the MMPB test, especially for urine, was integral in confirming exposure and metabolism. Figure 7 outlines the recommended oxidation and extraction approach for screening urine samples in the event of a suspected exposure case. This approach could be applied to other sentinel species, or even following human exposure events.

*Toxins* **2019**, *11*, 456

**Figure 7.** Schematic showing the recommended approach to the oxidation, extraction and clean-up of urine for the purposes of total Adda MCs by the MMPB method. In order to properly quantitate, it is essential that pre-oxidation spiking of MC-LR be conducted on duplicate sample.

With regard to more long term, chronic exposure, it has been demonstrated that hair might be a candidate for assessing exposure. This is the first use of MC testing on mammalian hair following an intoxication event and positive results were observed using the MMPB technique. This preliminary exercise shows promise with regard to potential avenues of testing in both acute and chronic mammalian exposure cases. However, confirmatory testing beyond the MMPB analysis did not detect free MC-LR in the hair above the established method detection limits. This area of research should be expanded in order to determine the applicability of this specimen for monitoring exposed populations of mammalian species.

Pharmacokinetic studies of orally ingested MCs in canines have not been conducted to date, making interpretation of measured toxin levels challenging. The amount of toxin needed to cause the observed gross and microscopic lesions has not been standardized in canine patients and extrapolation from lab animal models may not be representative. In experimental settings, the median lethal dose (LD50) for MCs is typically derived from an intraperitoneal (i.p.) route, which has been reported as low as 36 μg per kg body weight for mice exposed to MC-LR [50]. The oral LD50 of MC-LR has been reported to be as high as 5,000 μg per kg body weight for mice, and even higher for rats, supporting interspecies variability [10]. The likely presence of other protein phosphate inhibitors concomitant with MC-LR, such as other MC variants and peptides (e.g., anabaenopeptins), can also complicate interpretations of toxin data in reference to toxicity [51,52]. Since other MC variants were detected in this event, and MMPB data indicated even more variants were present but not accounted for in targeted analyses, correlating MC-LR levels from previous toxicity studies would provide little benefit. The inter-/intra-species variability combined with complicated real-world exposure, results in a high level of uncertainty with regard to MC dose and observed toxicity.

In this study, the dose for each animal was unknown, but it can be inferred from MMPB analysis of the urine samples from a surviving individual (32,000 ng mL<sup>−</sup>1) at 1-day post exposure and from the deceased animal (41,000 ng mL<sup>−</sup>1) at 2-days post exposure, that doses were similarly high and caused the observed hepatoxicity. Additionally, the single collected vomit sample provided evidence that >46 μg kg−<sup>1</sup> total Adda MCs was initially ingested, but it is impossible to determine what level entered metabolism. Post metabolism, several potential MC metabolites were observed in the urine employing an MS scan, but require more sophisticated analyses, such as high-resolution mass spectrometry and deconjugation experiments, to elucidate the profile. Work to identify these metabolites is ongoing.

Although the oral exposure route has been confirmed in at least one of these animals, the potential for compounded exposure through inhalation cannot be ruled out. Inhalation of MC-LR by mice has a reported LD50 of 43 μg kg−<sup>1</sup> [53], similar to that reported for the i.p. route. The higher toxicity observed via inhalation coupled with evidence that MCs can become aerosolized during cyanobacteria blooms [54,55] may present additional health risks to those living in close proximity to an ongoing bloom. Research has shown that low dose, subchronic exposure can result in accumulation of MCs in the mammalian liver [56]. Since dogs are considered a proposed sentinel species and they intimately share the human environment, canine exposure events such as this support the importance of a pro-active approach with regard to HABs in relation to human and animal health.

Collection of appropriate specimen type and selection of optimal analytical test are of high importance for accurate diagnosis. It was illustrated that the Adda ELISA provided false positive data and exaggerated assay responses, likely due to non-specific binding to kit antibodies. This highlights the need to confirm any ELISA data, especially for the analysis of matrices more complicated than water, for which the assay was intended. Matrix effects and exaggerated ELISA responses have been observed in other work [36,40] and require, at minimum, a secondary method of confirmation when a positive assay response is observed. In this work, both LC-MS/MS of targeted MCs and total Adda MCs by MMPB were conducted in addition to the MC Adda ELISA. However, due to method specificity, targeting MC variants, such as the MC-LR targeted in this work, under-represents total MCs, both due to MC structural variability and metabolic transformations. Therefore, if only one method is to be used in initial screening of samples in a suspected exposure case, it is recommended that the MMPB method be used. The MMPB approach provided a value representative of total (e.g., bound, free, conjugated) Adda MCs for the evaluated samples with a low detection limit.

Although clinicopathological data and epidemiology supported involvement of six dogs in the described morbidity/mortality event, only four dogs were confirmed to have been exposed to MCs using analytical techniques. Due to monetary constraints and lack of published sampling/testing protocols for antemortem testing of cyanotoxin in canines, confirmation could not be made in the other three cases. In the present study, a viable test using free catch urine collection has been established. Furthermore, a set protocol for directed toxin testing is described and can hopefully be employed in similar events to quickly and accurately diagnose MC intoxication. The development of an antemortem assay using non-invasive collection techniques is sure to have a significant impact on the diagnosis, treatment, and exposure of animals to HABs.

### **4. Materials and Methods**
