Subacute Cardiovascular Toxicity of the Marine Phycotoxin Azaspiracid-1 in Rats

Abstract

Azaspiracids (AZAs) are marine toxins produced by Azadinium spinosum that get accumulated in filter feeding shellfish through the food-web. The first intoxication was described in The Netherlands in 1990, and since then several episodes have been reported worldwide. Azaspiracid-1, AZA-2, and AZA-3 presence in shellfish is regulated by food safety authorities of several countries to protect human health. Azaspiracids have been related to widespread organ damage, tumorogenic properties and acute heart rhythm alterations in vivo but the mechanism of action remains unknown. Azaspiracid toxicity kinetics in vivo and in vitro suggests accumulative effects. We studied subacute cardiotoxicity in vivo after repeated exposure to AZA-1 by evaluation of the ECG, arterial blood pressure, plasmatic heart damage biomarkers, and myocardium structure and ultrastructure. Our results showed that four administrations of AZA-1 along 15 days caused functional signs of heart failure and structural heart alterations in rats at doses ranging from 1 to 55 µg/kg. Azaspiracid-1 altered arterial blood pressure, tissue inhibitors of metalloproteinase-1 plasma levels, heart collagen deposition, and ultrastructure of the myocardium. Overall, these data indicate that repeated exposure to low amounts of AZA-1 causes cardiotoxicity, at doses that do not induce signs of other organic system toxicity. Remarkably, human exposure to AZAs considering current regulatory limits of these toxins may be dangerously close to clearly cardiotoxic doses in rats. These findings should be considered when human risk is estimated particularly in high cardiovascular risk subpopulations.

Azaspiracids (AZAs) are marine algal toxins produced by the dinoflagellate Azadinium spinosum (Tillmann et al., 2009) and accumulated in filter feeding shellfish. The first human intoxication episode was in The Netherlands after ingestion of contaminated mussels from Killary Harbour, Ireland (MCMahon, 1996; Satake et al., 1998). Azaspiracid poisoning (AZP) is an acute gastrointestinal syndrome (James et al., 2004) and to date, at least 6 human intoxication episodes have occurred (Furey et al., 2010). Currently, AZAs are considered worldwide distributed natural contaminants, and food safety authorities have regulated the presence of AZA-1, AZA-2, and AZA-3 in seafood to protect consumers’ health (European Comission Regulation, 2004; Furey et al., 2010).

Widespread organ damage, tumorogenic properties, and heart arrhythmogenicity have been described in vivo for AZAs (Aasen et al., 2010; Ferreiro et al., 2014a; Ito et al., 2000, 2002, 2006); however, the molecular target remains unknown. Azaspiracids in vitro studies reveal several effects on cell biology, among them cytotoxicity, cytoskeleton damage, apoptosis activation, calcium modulation, blockage of human ether-a-go-go related gene (hERG) channels and hERG trafficking alterations (Alfonso et al., 2005; Ferreiro et al., 2014b; Twiner et al., 2012; Vilariño et al., 2006, 2007). In addition, AZA-1 has been detected in mice hearts after oral administration (Aasen et al., 2010) and in vitro experiments indicate that at least some AZA-1 effects at the cellular level are irreversible (Vilariño, et al., 2007). Collectively, these data suggest a possible accumulative AZA-related cardiotoxicity that should be explored.

European Medicines Agency recommendations (EMA, 2005) and specialized publications (Guth, 2007; Stummann et al., 2009) guide the strategy to evaluate potential heart toxicity. Screening for effects on hERG function by patch clamp is the in vitro method of choice to explore cardiotoxicity. In vivo studies are also required and include functional and structural heart alterations. Heart rate (HR), arterial blood pressure (ABP) and the ECG are commonly monitored to evaluate functional cardiovascular effects (Stummann, et al., 2009). An ECG reflects the heart electrical activity, and it can reveal cardiomyocyte disorders or heart structural injuries that involve anomalous generation or propagation of the cardiac impulse (Farraj et al., 2011). Arterial blood pressure measurement is also critical to evaluate the hemodynamic response of the cardiovascular system.

Structural heart injury has been historically evaluated by histopathological analysis, but quantification of cardiotoxicity biomarkers is currently recommended to complete cardiotoxicity studies (Kettenhofen and Bohlen, 2008). Cardiac troponins (cTns), owed to their high specificity and stable release, are considered the “gold standard” biomarkers to detect cardiomyocyte necrosis (O'Brien, 2008). Brain natriuretic peptides (BNPs) are early and highly sensitive biomarkers to assess acute or chronic cardiac impairment. Tissue inhibitors of metalloproteinases (TIMPs), together with matrix metalloproteinases, are important regulators of extracellular matrix (ECM) remodeling after heart injury (Bowers et al., 2010). Among TIMPs, TIMP-1 is a well-characterized plasma biomarker of heart remodeling processes (Vanhoutte and Heymans, 2010; Zile et al., 2011).

The aim of this work was to study subacute in vivo cardiotoxicity of AZA-1 (Figure 1) in a rat model by evaluating cardiovascular function parameters, cardiac biomarkers, and heart structure/ultrastructure.

FIG. 1.

Structures of azaspiracid-1 to azaspiracid-6.

MATERIALS AND METHODS

Reagents. Azaspiracid-1 (purity ≥ 99%) (Figure 1) was a quality controlled standard supplied by Laboratorio CIFGA S.A. (Lugo, Spain). Dimethyl sulfoxide (DMSO) was from Sigma-Aldrich Co. LLC. (St. Louis, MO). General health profile (GHP) chemistry panel was obtained from IDEXX Laboratories (Barcelona, Spain). Milliplex Map KITs were from Millipore (Billerica, MA). All chemicals were reagent grade quality.

Animals and in vivo experimental design. In vivo studies were performed with Sprague Dawley female rats aged 8–16 weeks (180–260 g). The rats were housed in a temperature- and humidity-controlled room (21°C±2°C, 50%±5% relative humidity) and maintained on a 12 h/12 h light/dark cycle. They were caged in pairs with free access to food and water. After the first toxin/vehicle administration, the rats were caged individually although they were allowed to see and smell their congeners.

AZA-1 or vehicle (DMSO) was administered ip on days 1, 5, 9, and 13, and the data were collected on day 15. Intraperitoneal administration was selected over IV injection due to a better adequacy for subacute studies and over the oral route for a better estimation and lower variability of the amount of toxin available for systemic exposure, owed to the lack of information about AZA-1 oral bioavailability. Doses of 1, 10, and 55 µg/kg AZA-1 were injected to 6, 4, and 6 rats, respectively. Each treated rat received the same dose in all 4 injections. For ip injection, the solvent of the AZA-1 stock solution (methanol) was evaporated and the toxin was reconstituted with DMSO. Saline solution (Grifols Engineering, Barcelona, Spain) was added subsequently to provide a final concentration of 10% DMSO (each rat received 50 µl of DMSO per 200 g of body weight) and 0.4, 4, or 22 μg/ml AZA-1 (for the 1, 10, and 55 µg/kg doses, respectively, final injection volume was 500 µl/200 g in all rats). Eight vehicle control rats were injected with the same concentration of DMSO in saline solution. The experiments could not be performed simultaneously for adequate ECG monitoring and therefore vehicle controls were spread along the experimental period to ensure uniformity of all conditions; for this reason, the number of controls exceeds the number of rats for each dosing group.

The rats were evaluated for signs of toxicity twice a day. Two days after the last toxin administration, on day 15, the rats were anesthetized with isoflurane (FI_ISO 1.5%–2%), and the cardiovascular function was monitored using ECG and ABP measurements. The ECG was monitored for 1 h. Then ABP was measured by placing a catheter in the right carotid artery. Afterward, another catheter was placed in the jugular vein for blood sample collection. Two blood samples of 400 and 500 μl were collected in EDTA tubes for the detection of biomarkers and hematological analysis, respectively. A third blood sample of 500 μl was collected in a heparin tube for biochemical analysis. All animals were euthanized by exsanguination at the end of the experiment, except those that died during the 15-day treatment period or those euthanized prematurely to prevent suffering. The number of animals per dosing group was reduced to the minimum necessary to obtain significant results of cardiovascular toxicity. All animal procedures were conducted according to the principles approved by the Institutional Animal Care Committee of Universidad de Santiago de Compostela.

Electrocardiography. Electrocardiography was recorded with a MAC* 800 Resting ECG System (GE corporate, Madrid, Spain) using lead II. The electrodes were placed as in Ferreiro et al. (2014a). An ECG recording of every rat was obtained for 15 min, under anesthesia, 2-10 days before starting toxin administration (basal). Any abnormality at this point prompted the exclusion of the animal from the study. ECG was also registered on day 15 of the experiment for 1 h. The following ECG parameters were analyzed: HR, QT interval, PR interval, and T wave length. The corrected QT interval (QTc) calculation was done as in Ferreiro et al. 2014 (Ferreiro et al., 2014a).

Arterialblood pressure measurement. Arterial blood pressure was measured using a direct method with a Diascope 2 S&W Medico Teknik instrument (Albertslund, Denmark). Three independent measurements of ABP were done for each rat. Data are reported as systolic ABP (SAP), diastolic ABP (DAP), and mean ABP (MAP).

Cardiac andinflammatory biomarkers. Cardiac troponin I (cTnI), cardiac troponin T (cTnT), BNP, TIMP-1, CXCL5/LPS-induced chemokine (LIX), interleukin-6 (IL-6), keratinocyte chemoattractant/human growth-regulated oncogene (KC-GRO), and tumor necrosis factor-α (TNF-α) were measured in 1 plasma sample collected at the end of each experiment using a commercial assay based on the Luminex XMap technology. Blood samples were centrifuged immediately after collection to separate the plasma fraction and stored at −80°C until their analysis. A rat cardiovascular disease (CVD) panel 1 Milliplex Map KIT was used for cTnI, cTnT, BNP, and TIMP-1 quantifications, and immunology/immune response rat cytokine/chemokine Milliplex Map KIT for LIX, IL-6, KC-GRO, and TNF-α quantifications in plasma (100 µL) following the instructions provided by the manufacturer. All samples were assayed in duplicate.

Biochemistryanalysis. Biochemistry parameters were analyzed in 1 plasma sample collected at the end of each experiment. Plasma fraction was separated by centrifugation and immediately analyzed with the IDEXX VetTest Chemistry Analyzer. A prepacked GHP panel was used to test 12 parameters: albumin (ALB), alkaline phosphatase (ALKP), alanine aminotransferase (ALT), blood urea nitrogen (BUN), calcium (Ca), cholesterol (CHOL), creatine kinase (CK), creatinine (CREA), globulin (GLOB), glucose (GLU), phosphorus (PHOS), and total protein (TP).

Hematologicalanalysis. A blood sample collected at the end of every experiment was immediately analyzed in the IDEXX ProCyte Dx Haematology Analyser for the hematological parameters: haematocrit (HCT), red blood cells (RBC), reticulocytes, and white blood cells (WBC) (neutrophils, lymphocytes, monocytes, eosinophils, and basophils). Because this analyzer system was not validated for rats, WBC counts were also performed in stained blood smears under a microscope (×100 magnification) to verify the hematology analyzer results. In each blood smear, 20 microscope fields were randomly chosen and 2 smears were examined per rat. Lymphocyte and neutrophil percentages were confirmed by optical microscopy.

Histologicaldamage evaluation. Samples of heart, lung, thymus, brain, spleen, pancreas, liver, kidney, stomach, and small and large intestine were collected for histological evaluation immediately after euthanasia. Macroscopic organ examination was done during the extraction process, and the samples were then prepared for light microscopy (LM) and transmission electron microscopy (TEM) observation.

For LM, all organ samples were fixed by immersion in buffered 10% formalin and Bouin for 24 h at 4°C. After that, the tissues were processed for hematoxylin and eosin (H&E) staining and examined under the light microscope as in Ferreiro et al. (2014a). Heart collagen content was evaluated with Sirius Red staining. Heart specimens were stained with 0.1% Sirius red in saturated aqueous solution of picric acid for 1 h. Then, they were rinsed in acidified water (0.5% acetic acid in distilled water), dehydrated in 100% ethanol, and cleared twice in xylene (Junqueira et al., 1979). The percentage of collagen in Sirius Red stained heart sections was measured with ImageJ 1.43 software. Three nonconsecutive analogous fields were randomly chosen in each heart section to obtain high-resolution images (×40). The surface corresponding to collagen staining was quantified in 3 images from 3 sections for each rat, obtaining 9 measurements per rat. Collagen content was reported as percentage of collagen area versus total area of tissue (Hadi et al., 2010).

For TEM, all organ samples (1 mm³) were fixed by immersion in TEM fixative solution (2.5% glutaraldehyde in 0.1 M cacodylate trihydrate buffer) for 4 h at 4 °C. Then the samples were rinsed with 0.1 M cacodylate trihydrate buffer. Postfixation by immersion in 1% OsO₄ in 0.1 M cacodylate trihydrate buffer was performed for 60 min. Finally, after a second rinse fixed tissues were dehydrated in graded ethanol solutions, including 1 bath with 70% ethanol and 0.5% uranyl acetate, rinsed in propylene oxide, and embedded in Epon 812 (Momentive Specialty Chemicals Inc., Houston, TX). A Leica Ultracut UCT ultramicrotome from Leica Microsystems GmbH (Wetzlar, Germany) was used to obtain ultrathin sections of tissue samples, and they were counterstained with uranyl acetate and lead citrate. Ultrastructural analysis of 1 mm² samples was performed with a JEOL JEM-1011 Transmission Electron Microscope (Jeol Ltd, Tokyo, Japan). Five fields were randomly chosen in every heart sample and evaluated for integrity of cell structures at several magnification levels.

Dataanalysis. Data were plotted as mean ± SEM. Statistical significance was determined using t test for unpaired data. ANOVA was used for multiple comparisons. P  <  .05 was considered for significance.

RESULTS

For cardiotoxicity evaluation, AZA-1 was administered ip at doses of 1, 10, and 55 µg/kg to 6, 4, and 6 rats, respectively. The same dose was repeated every 4 days to complete a total of 4 administrations during the experiment, which was ended on day 15 after the first toxin injection. A concentration of 55 µg/kg was selected as the first test dose because it caused heart function alterations in acute intravenous studies (Ferreiro et al., 2014a) and it is one-fourth to two-third of the reported acute ip LD₅₀ in mice (Marine Institute, NDP 2007-13). Dosing levels were subsequently reduced to 10 µg/kg, with moderate signs of toxicity, and to 1 µg/kg to test lower exposures. The exact same procedure was performed in 8 control rats in the absence of toxin. The following results were obtained from these experiments.

Symptoms

The rats were evaluated twice a day for toxicity signs during the total administration period. The symptoms observed are reported in Table 1. In the first 24 h after toxin injection, the symptoms were more evident, later the rats progressively recovered. The symptoms were greater in intensity for high dosing levels. The most remarkable sign was the progressive development of ascites along the 15 days, which was clearly evident in the rats dosed with 10 and 55 µg/kg at the end of the experiment. At necropsy, the fluid in the abdominal cavity was clear and volumes of approximately 16 ml were recovered for every animal.

TABLE 1.

Symptoms Displayed During AZA-1 Treatment

Symptoms	Control (8 Rats)	1 µg/kg (6 Rats)	10 µg/kg (4 Rats)	55 µg/kg (6 Rats)
Apathy	—	6, +	4, ++	5, +++
Reduced appetite	—	2, ++	4, ++	5, +++
				1, ++
Reduced level of spontaneous activity	—	2, ++	4, +	5, +++
				1, +
Squint-eyes	—	6, +	4, ++	5, +++
				1, ++
Piloerection	8, +	6, ++	4, ++	5, +++
				1, ++
Decreased grooming	—	2, +	4, ++	5, +++
				1, ++
Accumulation of porphyrin secretions (eye, nares)	—	2, +	4, ++	5, +++
				1, ++
Ascites (day 15)	—	—	4, +++	1, +++
Paralysis of limbs	—	—	—	2, +++
Death	—	—	—	3 (day 3)
Euthanized	—	—	—	2 (day 6)
Body weight increment (%)
Day 4 versus day 1	2±2	3±1	−5±2	3, −11±2
Day 15 versus day 1	11±2	8±2	13±5	1, 22
(mean ± SEM)

The number of rats that displayed the symptom and the intensity are indicated. Intensity: — not present; + low; ++ moderate; +++ severe.

Furthermore, all rats were weighted before every injection and at the end of the toxin treatment. The percentage of body weight gain on days 4 and 15 is reported in Table 1. On day 4, the rats that received AZA-1 showed a weight loss; however, on day 15, their weight gain was similar to controls.

AZA-1 Effects on Rat ECG

The evaluation of heart functionality after repeated administrations of AZA-1 was done using ECG recordings obtained before and when the treatment was completed for every rat. All ECG parameters were measured at a recording speed of 50 mm/s (Figure 2A). No alterations of HR (Figure 2B), QTc interval, T-wave, and PR-interval durations (Figs. 2C–E) were observed after 4 administrations of 1 μg/kg or 10 μg/kg AZA-1. The rat treated with 55 μg/kg AZA-1 that survived the complete treatment did not evidence any ECG alteration (data not shown).

FIG. 2.

Azaspiracid (AZA)-1 subacute effects on rat ECG and arterial blood pressure. ECG parameters were analyzed before and after 4 ip administrations of 1 or 10 μg/kg of AZA-1 or carrier (n = 6, 4, and 8, respectively). Control rats followed the same experimental protocol in the absence of toxin. A, Representative ECG recording at 50 mm/s. ECG landmarks and measurements of QT interval, R-R interval, T-wave duration, and PR interval are indicated. B, Heart rate, (C) QTc interval, (D) T-wave, and (E) PR interval measured in AZA-1–treated and control rats before and after administrations. F, Representative ECG fragments of (a) ventricular extrasystoles (VES) episodes recorded in a rat that received 4 injections of 1 μg/kg AZA-1 and (b) inverted QRS complex recorded for a rat that received 2 injections of 55 μg/kg AZA-1. G, Systolic ABP (SAP), diastolic ABP (DAP), and mean ABP (MAP) in rats that received 4 administrations of 1, 10, or 55 μg/kg of AZA-1 or carrier (n = 6, 4, 1, and 4, respectively, mean ± SEM. *Statistically different from control).

Cardiac rhythm alterations were also evaluated. Ventricular extrasystoles (VES), 26 in 1 h, were observed in 1 of 6 rats injected with 1 μg/kg AZA-1 (Figure 2Fa). ECG was also recorded for two rats that had to be euthanized after a second administration of 55 µg/kg AZA-1. One of them presented bradycardia and inverted QRS complex (Figure 2Fb).

AZA-1 Effects on ABP

The effect of repeated administrations of AZA-1 on ABP was evaluated by direct measurements on day 15. Arterial blood pressure was measured in the rats treated with 1, 10, and 55 µg/kg AZA-1 and in 4 controls. The results showed statistically significant reductions in SAP, DAP, and MAP of AZA-1–treated rats (Figure 2G). The 2 rats that had to be euthanized after the second administration of 55 µg/kg AZA-1 presented a severe hypotension state (MAP = 36  ± 1.2 mm Hg). In fact, the rat showing inverted QT complex entered apnea after the ABP measurement and died.

Cardiac and Inflammatory Biomarkers

The levels of cTnI, cTnT, BNP, TIMP-1, LIX, IL-6, KC-GRO, and TNF-α were evaluated in plasma of rats injected with 1 (n = 6), 10 (n = 4), and 55 (n = 1) µg/kg AZA-1 or carrier (DMSO, n = 8). The samples were analyzed with a CVD kit and an immunology/immune response (rat cytokine/chemokine) kit. The CVD kit results indicate that cTnI, cTnT, and BNP plasma levels were not statistically different in AZA-1–treated rats and controls (Figs. 3A–C). On the contrary, TIMP-1 plasma levels were significantly increased (6.8 times above controls) in rats treated with 10 µg/kg AZA-1 (Figure 3D). An increase of plasmatic TIMP-1 was also observed in the rat that completed 4 administrations of 55 µg/kg AZA-1 (4.6 times higher than control levels; Figure 3D). One of the rats that received only 2 doses of 55 µg/kg AZA-1 had remarkably high levels of TIMP-1 (50 times higher than control levels, data not shown), the other rat died during anesthesia before blood samples could be collected. The immunology/immune response kit results showed no significant increase of LIX, IL-6, KC-GRO, or TNF-α in the plasma of toxin-treated rats versus controls (Fig. 3E–H).

FIG. 3.

Cardiac and inflammatory biomarkers in plasma of rats treated with azaspiracid (AZA)-1. Biomarkers were quantified in blood samples collected from every rat after 4 ip administrations of 1, 10, or 55 μg/kg of AZA-1 or carrier (n = 6, 4, 1, and 8, respectively). A, Cardiac troponin I (cTnI), (B) cardiac troponin T (cTnT), (C) brain natriuretic peptides (BNP), (D) tissue inhibitors of metalloproteinase (TIMP)-1, (E) CXCL5/LPS-induced chemokine (LIX), (F) interleukin (IL)-6, (G) keratinocyte chemoattractant/human growth-regulated oncogene (KC-GRO), and (H) tumor necrosis factor-α (TNF-α) concentrations in plasma of rats exposed to AZA-1 (mean ± SEM, *statistically different from control).

Biochemical and Hematological Parameters

The biochemical parameters ALB, ALKP, ALT, BUN, Ca, CHOL, CK, CREA, GLOB, GLU, PHOS, and TP were quantified as described in methods to obtain an overall functional evaluation of various organic systems. The results showed only a statistically significant increase in ALT and ALKP plasma levels of rats treated with 1 and 10 μg/kg AZA-1, respectively, when compared with controls (Supplementary Table 1); however, in both cases, the levels were within physiological values. The rat that received 4 administrations of 55 µg/kg AZA-1 had increased levels of CK at the end of the experiment (Supplementary Table 1). One of the rats prematurely euthanized after 2 injections of 55 µg/kg showed an increase of BUN, CREA, ALT, CHOL, and CK (data not shown). The other rat died after ABP was measured and blood samples could not be collected. The hematological analysis showed that 1 or 10 μg/kg AZA-1 did not cause any significant alterations of HCT and counts of lymphocytes and neutrophils (Supplementary Table 2). Finally, the 55 μg/kg-treated rat that survived 4 administrations did not show hematological alterations (Supplementary Table 2), but 1 rat prematurely euthanized after 2 administrations evidenced a marked increase of neutrophil percentage (33.5%) and a reduction of lymphocyte percentage (44.6%).

Heart Tissue Structure and Ultrastructure

After euthanasia, internal organs were examined for macroscopic alterations and collected to perform histopathological analysis. No rat presented evidence of organ damage by macroscopic observation, except for one rat dosed with 55 µg/kg AZA-1 that was prematurely euthanized, which liver appeared pale.

Histopathological damage of heart, lung, thymus, brain, spleen, pancreas, liver, kidney, stomach, and small and large intestine tissues was evaluated by LM. Only alterations in the heart were observed that could be unequivocally related to the presence of toxin. Slight intracellular vacuolization and an apparent increased separation between adjacent cardiomyocytes were observed in H&E-stained heart sections of rats treated with 1, 10, and 55 µg/kg AZA-1. Sirius red staining of control and AZA-1–treated samples (Figure 4A, a and b, respectively) revealed an increase of collagen in hearts of AZA-1–treated rats (Figure 4B). Ultrastructural effects of repeated AZA-1 administrations were assessed by TEM in heart and liver, which was included considering the macroscopic alteration in one rat. Electron microscopy images revealed notorious changes in cardiomyocyte ultrastructure. The cardiomyocytes of control rats (Figure 4C) showed regular arrangement and morphology of the myofibrils, the typical dark-light band pattern of the sarcomeres and abundant spherical or elongated mitochondria, which contained packed, regular cristae and were well-arranged in rows between myofibrils. In rats treated with 1 µg/kg AZA-1, slight myofibril disorganization and irregular mitochondria arrangement with sporadic size and shape alterations and vacuolation of their cristae were detected (Figure 4D). The cardiomyocytes of 10 and 55 µg/kg–treated rats showed disorganization, fragmentation and loss of the striated pattern of their myofibrils. Furthermore, the mitochondria were disarranged, irregular in size and shape, and displaying cristae vacuolation (Figs. 4E and F). The damage increased in intensity and extension with the dose. In the hearts of rats treated with 55 µg/kg, it was difficult to find unaffected cardiomyocytes in these randomly selected sections, in rats exposed to 10 µg/kg, a few unaltered or less damaged cells could be found, and in 1 µg/kg–treated rats, damaged cardiomyocytes were less abundant. Except for the 55 µg/kg dose in which some cell structures were difficult to identify, other cell structures were evaluated and they were not altered (Supplementary Fig. 1). Only the rat that evidenced macroscopic pale liver showed ultrastructural alterations in the liver, mainly autophagosomes and mitochondrial damage (Supplementary Fig. 2).

FIG. 4.

Azaspiracid (AZA)-1 effects on heart collagen content and on heart muscle ultrastructure. Collagen content and ultrastructural damage was evaluated in heart samples of rats exposed to 1, 10, and 55 μg/kg of AZA-1 and control rats (n = 6, 4, 1, and 3, respectively). A, Representative images of heart sections from (a) control and (b) 1 μg/kg AZA-1–treated rat, respectively, stained with Sirius red, arrows indicate collagen. B, Collagen content expressed as percentage of image surface corresponding to collagen as judged by Sirius red staining (mean ± SEM, *statistically different from control). C, Representative electron micrograph of the myocardium of control rats showing regular arrangement of myofibrils, typical dark-light band pattern of the sarcomeres and spherical or elongated mitochondria, with visible packed cristae. D, Representative electron micrograph of cardiomyocytes of rats that received 4× 1 μg/kg AZA-1 showing slight myofibril disorganization and irregular arrangement of mitochondria. E, Representative electron micrograph of cardiomyocytes of rats that received 4× 10 μg/kg AZA-1 showing notorious myofibril disorganization, loss of mitochondria alignment between myofibrils and evident mitochondria vacuolation and disruption of cristae integrity. F, Representative electron micrograph of cardiomyocytes of rats that received 4× 55 μg/kg AZA-1 showing disorganization, fragmentation, and loss of striated pattern of myofibrils and irregular mitochondria, with vacuolation and disruption of cristae (scale bar: 2000 nm). Collagen (c), intercellular space (is), mitochondria (mi), myofibrils (m), nucleus (N), sarcomeres (s), and t-tubules (t-T).

DISCUSSION

Several recently published studies indicate AZAs potential cardiotoxicity. In vitro effects on hERG channels, including blockade (Ferreiro, et al., 2014a; Twiner et al., 2012) and trafficking alterations (Ferreiro et al., 2014b), and acute in vivo arrhythmogenicity of AZA-2 clearly support heart toxicity (Ferreiro et al., 2014a). Actually, other polyether compounds of marine origin such as yessotoxins, okadaic acid, palytoxins, brevetoxins, or ciguatoxins among others have shown cardiotoxic potential. Palytoxins, brevetoxins, and ciguatoxins cause heart dysfunction probably by alteration of ion fluxes through the plasma membrane (Frelin et al., 1990; Marquais and Sauviat, 1999; Templeton et al., 1989). Yessotoxins and okadaic acid induce ultrastructural alterations in cardiomyocytes (Aune et al., 2002; Terao et al., 1990; Tubaro et al., 2004) although until now no alterations of cardiovascular function have been related to these toxins (Ferreiro et al., 2015).

Our results demonstrate that repeated administrations of AZA-1 cause alterations of the cardiovascular function and myocardium structural and ultrastructural damage. Azaspiracid-1–induced heart injury was evidenced by elevated TIMP-1 plasma levels, collagen content increase, and ultrastructural cardiomyocyte damage. These observations are probably related. Cardiomyocyte injury triggers remodeling mechanisms, which involve collagen deposition at the insult site (Bowers et al., 2010). Tissue inhibitors of metalloproteinases are key regulators of ECM, and their upregulation has been related to heart injury and remodeling processes (Bowers et al., 2010; Vanhoutte and Heymans, 2010). The observation of ultrastructural damage of cardiomyocytes, even at nonlethal doses, reveals clear chronic cardiotoxic potential of AZA-1. This cardiomyocyte damage would induce remodeling-related increase of collagen in the heart, consistent with the simultaneous elevation of plasmatic TIMP-1 levels in rats treated with 10 µg/kg AZA-1. The absence of an increase of other heart damage biomarkers, BNP, cTnI, and cTnT, is probably due to the experimental design and their kinetics. Cardiac troponins kinetics (peak at 2–6 h; 6 h half-life for cTnI in rats) (O'Brien, 2008) suggest that sampling times might not have been appropriate to detect alterations of these biomarkers considering the interval between AZA-1 administrations and sample collection. On the contrary, the increase of TIMP-1 later after injury fits better the study sampling protocol (Peterson et al., 2000). Blood samples were not collected at earlier times to minimize stress, which might alter the results.

Interestingly, AZA-1–induced liver toxicity, previously described in mice (Ito, et al., 2000, 2002) was observed only in 1 rat that received 2 doses of 55 µg/kg and had to be euthanized. No liver alterations were detected in rats that received the 2 lower doses. These findings might reflect different species sensitivity to AZA-induced hepatotoxicity and confirm the appearance of cardiotoxicity at AZA doses that do not trigger hepatotoxicity.

Heart function implications of the previously discussed alterations were evidenced by ascites development. An elevated content of collagen increases the stiffness of the myocardium yielding a less compliant heart and diastolic dysfunction (Bowers et al., 2006, 2010). Ultimately, excessive remodeling leads to fibrosis and heart failure (HF) (Bowers, et al., 2010). ECG recordings did not evidence any functional abnormalities at the lower doses. The appearance of VES only in 1 rat that received 1 µg/kg is not enough evidence to suggest a correlation between VES and subacute AZA toxicity. On the other hand, although speculative, we are tempted to suggest that the inverted QRS complex and bradycardia in 1 rat treated with 55 µg/kg, might be due to AZA toxicity. Actually, QRS complex changes have been associated with bundle branch block and myocardial fibrosis, which have been related to heart function deterioration (Farraj et al., 2011; Zannad et al., 2007). This rat showed marked and generalized myocardium injury, which would be consistent with these ECG alterations. The results do not show a dose-effect relationship for ECG data; however, alterations of ECG often do not occur concurrently with HF (Kemp and Conte, 2012). Although echocardiography would have been more adequate to detect HF-related dysfunction, overall the data obtained during this study—ultrastructural damage of cardiomyocytes, ascites development in the absence of liver damage, plasma TIMP elevation, and increased collagen content in the heart—demonstrate cardiotoxic activity.

Hypotension is also a sign of cardiovascular imbalance, which could be related to HF (Kemp and Conte, 2012; Sabino et al., 2013). However, the fact that hypotension appears in rats dosed with 1 µg/kg in the absence of ascites and elevated TIMP-1 suggests that AZA-1 might have a direct functional effect on the vasculature. Arterial blood pressure could also be decreased by an inflammatory response, but several immune response markers, some of them previously related with HF (Van Kimmenade and Januzzi, 2012), were not altered in AZA-1–treated rats. Blood urea nitrogen and CREA, nephrotoxicity biomarkers (Tonomura et al., 2010), were raised in the rat with the most severe signs of toxicity, probably secondary to hemodynamics alteration. Moreover, this rat showed neutrophilia and lymphopenia, a blood dyscrasia described for other toxic compounds (Winnie and Uetrecht, 2013).

Azaspiracid structure-activity relationship is not well known, probably because its biological target has not been identified yet. However, AZA-2 and AZA-3 (Figure 1) have also in vivo toxicity orally and by ip injection (Marine Institute, NDP 2007-13), and therefore it is possible that they have cardiotoxic potential. Intraperitoneal toxicity of AZA-6 (Figure 1) is similar to AZA-2 (Marine Institute, NDP 2007-13). Azaspiracid-4 and AZA-5 (Figure 1), on the contrary, displayed lower in vivo toxicity (Ofuji et al., 2001), and consequently, lower heart toxicity should be expected. From in vivo and in vitro studies, it seems that the entire AZA-1 molecule, with its 2 ABCD and FGHI ring complexes, and its particular stereochemistry as well as the carboxyl group are necessary for toxic activity (Ito 2008; Ito et al., 2006; Vilariño et al., 2006, 2008). Additional results obtained by in vitro structure-activity studies with different analogs do not seem to correlate well with in vivo toxicity (Marine Institute, NDP 2007-13).

Finally, these new cardiotoxicity data should be considered when evaluating human health safety. Actually the 10 µg/kg ip dose used in this study, when corrected for administration route and inter- and intraspecies variation is close to the estimated lowest observable adverse effect level (LOAEL) in humans that was used for regulatory purposes. Toxicological data in mice indicate that an ip dose would be equivalent to a 5–10 times higher dose if administered po (ip LD₅₀ of 74 µg/kg; po LD₅₀ of 443–774 µg/kg) (Aune et al., 2012; Dewi et al., 2014; Marine Institute, NDP 2007-13). The factor accounting for variability within species should be at least 3, whereas the minimum factor for inter-species variation should not be lower than 10 (EFSA, 2008). Applying these corrections the equivalent dose for humans (oral exposure) would be about 1.6–3.3 µg/kg for a toxin ip dose of 10 µg/kg (that corresponds to a 50–100 µg/kg oral dose, divided by 3×10), which is in the range of the value suggested by EFSA CONTAM Panel of a LOAEL for AZA of 1.9 µg/kg. Current regulations (160 µg of AZA-1 equivalents/kg of shellfish meat) have been established based on a deterministic estimation of a 0.32% probability of exceeding 1 µg/kg for a 60 kg adult and a 400 g portion (EFSA, 2008). However, we have estimated the equivalent dose using 10 µg/kg, but we also observed ultrastructural heart damage in rats at levels of exposure 10 times lower. In addition, the intraspecies variation factor should probably be increased to account for highly sensitive individuals, such as high cardiovascular risk populations.

The main limitation of this toxicological study is that it was performed using ip administration of AZA-1, and therefore it is difficult to estimate the toxic potency by the oral route. Azaspiracid-1 is known to be absorbed and reach systemic blood flow after oral administration (Aasen et al., 2010) and undoubtedly heart damage would be possible in these conditions. However, bioavailability by this route is unknown. Oral bioavailability, and therefore toxicity, in rodents may vary greatly depending on the administration procedure (gavage or feeding) and other factors (Munday, 2006; Munday et al., 2004). A better estimation of potential human toxicity could be achieved if oral bioavailability of AZAs was known in humans.

In summary, the marine phycotoxin AZA-1 induces cardiotoxicity after repeated administrations in rats, at doses that did not cause toxic signs in other organic systems. Azaspiracid-1 cardiotoxic effects are evidenced by structural and ultrastructural heart damage and functional cardiovascular alterations. These cardiotoxicity data should be considered when evaluating the safety of current regulatory limits, mainly for populations with a pre-existing risk of CVD.

SUPPLEMENTARY DATA

Supplementary data are available online at http://toxsci.oxfordjournals.org/.

FUNDING

The research leading to these results has received funding from the following FEDER cofunded-grants. From CDTI and Technological Funds, supported by Ministerio de Economía y Competitividad, AGL2012-40185-CO2-01 and Consellería de Cultura, Educación e Ordenación Universitaria, GRC2013-016, and through Axencia Galega de Innovación, Spain, ITC-20133020 SINTOX. From CDTI under ISIP Programme, Spain, IDI-20130304 APTAFOOD. From the European Union’s Seventh Framework Programme managed by REA – Research Executive Agency (FP7/2007-2013) under grant agreement Nos. 265409 µAQUA, 315285 CIGUATOOLS and 312184 PHARMASEA.