O. vulgaris of both sexes (males: N = 32; females: N = 18; body weight, mean ± SEM: 574 ± 25 g) were caught in the Bay of Naples (Italy) by local fishermen and transported to Arcavacata (Cosenza, Italy) according to the best-practice for long-duration transportation (Byrne et al., 2004; review in Fiorito et al., 2015).

Octopuses were housed individually in opaque tanks (40 × 50 × 80 cm) with circulating sea water (18–22°C) and maintained according to the best practice (e.g., Agnisola et al., 1996) for up to 3 days prior to killing. Animals were randomly assigned to the various experimental groups.

Age of octopuses cannot be estimated in live individuals, but assessed post-mortem, and does not appear to correlate with body weight (Canali et al., 2011). Therefore, information on age of the octopuses was not available at the time the animals were assigned to experimental groups.

Research studies involving “live cephalopods” within the EU are covered by Directive 2010/63/EU (European Parliament and Council of the European Union, 2010) and its subsequent transposition into legislation of Member States (see also: Smith et al., 2013; Fiorito et al., 2015).

The studies reported here were performed before Italy transposed the Directive (March 2014). In addition, these experiments required killing the animal for the sole purpose of removing tissue. Therefore, they fall outside the scope of the Directive 2010/63/EU, provided that an approved method of killing is used (see below). Ethical review of the experiments at institutional level was undertaken at the study site (Cosenza, Italy).

This study adhered to the ethos of the Directive which specifies that killing must be performed by an “adequately educated and trained person” using an approved method (European Parliament and Council of the European Union, 2010). Although cephalopods are included in the Directive, no specific recommendations of methods for killing are listed in Annexe IV. In view of this we have applied the general principles outlined in Annexe IV (1a) to cephalopods (Andrews et al., 2013; Fiorito et al., 2015) that, when killed the animal should be unconscious, should remain unconscious until death ensues and is confirmed.

In all experiments included here the animals fulfilled the criteria for general anesthesia (as mentioned above, and see: Andrews and Tansey, 1981; Andrews et al., 2013; Gleadall, 2013; Fiorito et al., 2015) when killed.

Depending upon the protocol, the animals were exposed to one of the following treatments:

MgCl₂: Immersion for 20 min in 2 L of 3.5% magnesium chloride dissolved in sea water (sw) at room temperature (18–21°C). At this time the animals are immobile, pale, lack a righting reflex, are unresponsive to handling, and ventilation is suppressed or absent (see review of criteria in Fiorito et al., 2015).

After treatments the brain was destroyed to complete killing, as described in Directive 2010/63/EU Annexe IV (2b) (see recommendations in: Andrews et al., 2013; Fiorito et al., 2015).

Immediately after brain destruction, the systemic heart was isolated according to Foti et al. (1985) and Houlihan et al. (1987). The heart dissection was performed at 4°C and took ~ 15 min.

In O. vulgaris, blood enters the heart through two auricles and leaves the ventricle through three arteries: the dorsal (cephalic) aorta and the abdominal and gonadial arteries. A rich network of coronary veins is present on the surface of the heart which drain the blood directly from the ventricular lumen during ventricular systole (Agnisola et al., 1990). Both the auricles and the dorsal aorta were cannulated while the gonadial and abdominal arteries were ligatured at their base. The outflow from the coronary veins passed out into the chamber and was collected by an overflow system. The fusiform ganglion was removed.

The heart was connected to the perfusion apparatus where the two auricles received perfusion fluid at the same controlled input pressure. The perfusion solution (PS) contained filtered sea water containing 2.78 mM anhydrous glucose, and was gassed with 99.5% oxygen and 0.5% carbon dioxide. The pH of the perfusion medium was 8.0 (as in: Houlihan et al., 1987; Agnisola et al., 1989; Agnisola and Houlihan, 1994) and is close to the pH of O. vulgaris haemolymph (pH = 7.8 ± 0.1; D'Aniello et al., 1986).

The basal perfusion conditions were chosen to reproduce in vivo resting hemodynamic parameters as previously reported (Agnisola et al., 1989; Agnisola and Houlihan, 1994). The preload (input pressure) was adjusted to obtain a stroke volume (SV) of 0.7–0.8 ml/g of ventricular weight, a value that is approximately physiological for resting animals of a similar size to those used here (Wells et al., 1987). The diastolic output pressure was always set at 2 kPa above the preload which corresponds to the diastolic aortic pressure in resting animals (Wells and Smith, 1987). Experiments were performed at room temperature (18–21°C). The heart beats spontaneously and does not require pacing at this temperature. Under these conditions, aortic pressure, heart rate, aortic, and coronary flow were measured for 15–20 min to assess their stability; if parameters had not stabilized within the range established in previous studies the experiment was not continued. About 15% of hearts failed to meet the criteria for stabilization, but there was no obvious relationship between body weight, sex, and anesthetic protocol and failure to stabilize.

Preload and afterload (in kPa) were defined as the mean input and output pressures, respectively. The pressure measurements were referred to the level of perfusate in the perfusion chamber and corrected for cannulae resistance. Pressures and heart rate (HR, beats/min) were measured through two MP-20D pressure transducers (Micron Instruments, Simi Valley, CA) connected to a PowerLab data acquisition system and analyzed using Chart software (AD-Instruments, Ugo Basile, Comerio, Italy).

Other cardiac parameters were measured as follows:

Cardiac output (Q, ml/min/g) and Coronary output (CorO, ml/min/g) were derived from the dorsal aorta outflow and coronary vein outflow, respectively, collected over 1 min, and weighed. Values were corrected for fluid density and normalized per gram ventricle wet weight.

The separation between aortic and coronary output is necessary because during contraction, a proportion of the total cardiac output is collected in the coronary veins passing through the walls of the heart (Foti et al., 1985; Houlihan et al., 1987; Agnisola and Houlihan, 1994). Thus, contraction of the isolated ventricle resulted in ejecting flow through the cannulated aorta (i.e., aortic output) and through the walls of the heart, into the cut coronary veins (i.e., coronary output). In figures Q, SV, and SW are related to aortic output.

The following chemicals were used: Ethanol (99%, APPLICHEM, CAS Number 64-17-5); Glucose (SIGMA, CAS Number: 50-99-7); Magnesium chloride hexahydrate (SIGMA ALDRICH, CAS Number: 7791-18-6).

The osmolality of the sea water, perfusion medium and the anesthetic formulations used for the experiments was measured using an autocal osmometer (Roebling) according to recommendations for the use of osmometry methods for biological samples (Sweeney and Beuchat, 1993). For each solution four samples were measured (at 20°C) in triplicate.

Results are expressed as mean ± SEM of percentage changes obtained from individual experiments unless otherwise stated. Statistical analysis was performed on raw data (not %) following Zar (1999). All data were tested for normality. We utilized repeated-measures ANOVA followed by Bonferroni's Multiple Comparison test whenever appropriate. Multivariate Analysis of Variance (MANOVA) was utilized to test effects of different treatments for Frank-Starling curves. Mixed Model ANOVA was utilized as method for testing effects of repeated-measures when the sample numbers were not matched for all treatments/parameters. Differences were considered statistically significant at p < 0.05. For all statistical analyses we used SPSS (rel. 14.0, SPSS Inc—Chicago, 2005).

Osmolality values of the solutions used for in vivo (utilized to anesthetize the animals) and in vitro (by perfusion) studies are reported in Table 1. The osmolality of sea water was 1125.25 ± 1.39 mOsm/Kg (n = 4) and the osmolality of the perfusion solution (PS: Sea water+glucose) was 1139.67 ± 1.17 mOsm/Kg (n = 4). In our conditions, all solutions had higher osmolality values than sea water and PS, with the exception of MgCl₂ (1:1) where the osmolality was about 10% lower than either PS or sea water. It is interesting to note that haemolymph in Mediterranean O. vulgaris has an osmolality of ~1300 mOsm/Kg or higher (D'Aniello et al., 1986; Wells and Wells, 1989).

The baseline hemodynamic parameters in the spontaneously beating isolated hearts from animals with the five different pre-isolation treatments are summarized in Table 2. Irrespective of the different anesthetic treatments (including hypothermia), at baseline conditions the perfused hearts were able to produce physiological values of stroke volume and to work at hemodynamic loads comparable with those reported by other authors for in vitro preparations (Agnisola et al., 1989; Agnisola and Houlihan, 1994). No differences were detected in baseline parameters of hearts with different treatments, with exception of afterload pressure and CorO (Table 2). Although the afterload values were in the physiological range (see as reference: Wells et al., 1987; Wells and Smith, 1987) differences emerged only when comparing MgCl₂ vs. hypothermia and MgCl₂ vs. Mix 45′ (Table 2). In addition, the coronary outputs were only significantly different when comparing hypothermia vs. MgCl₂ (1:1) or Mix 20′ (Table 2).

Frank-Starling curves obtained from cardiac preparations of animals treated with MgCl₂, MgCl₂(1:1), Mix 20′, Mix 45′, and hypothermia showed that, with increasing preload, a significant raise in Q, SV, and SW occurred (Figure 1). The preload increases did not produce a significant change in the heart rate in any of the treatments, thus the increases in cardiac output were due solely to an increase in stroke volume.

The Frank-Starling curves for the individual treatments are plotted in Figure 1. Hearts isolated from animals under either hypothermia or MgCl₂in sea water endured a higher number of pre-load increments (6 increments) before reaching the plateau of the Frank-Starling curve (i.e., the optimal preload at which point Q remains constant) followed by Mix (20′) and MgCl₂ (1:1) (5 increments), and Mix 45′ (4 increments).

The individual Frank-Starling curves (Figure 1) for each pre-isolation treatment revealed that over a similar pressure range the curves for MgCl₂(1:1) and Mix 45′ were flatter than those for all other pre-isolation treatments. This is confirmed by the maximum values reached for all cardiac functions parameters (see legend in Figure 1). At the maximal input pressure used (until there was no further increase in Q), the largest response in terms of Q, SV, and SW was obtained with hypothermia followed by treatment with Mix 20′ and MgCl₂. MgCl₂(1:1), and Mix 45′ showed the worst response revealed by an impaired ability to respond to preload increases (Figure 1).

Repeated-measures ANOVA confirmed this view, revealing no significant differences between any of curves for each treatment for any of the parameters considered (SV, SW, HR, and Q, see Figure 1), but also highlight a significant treatment^*preload effect for all parameters considered (p < 0.001), with the exception of HR (p = 0.999), thus confirming that the hearts all responded to increases in preload. Inspection of the Frank-Starling curves (Figure 1) revealed that hearts isolated under hypothermia, Mix 20′ and MgCl₂ were all capable of increasing stroke volume by ~50% or more, whereas this was not the case for the MgCl₂ (1:1) and Mix 45′ hearts. Bonferroni's Multiple Comparison Tests after Multivariate Analysis of Variance (MANOVA) confirmed this view. Significant differences in the preload-stroke volume relationship were observed only for hypothermia vs. Mix 45' (preload values 0.6, 0.8, 0.9 kPa, p < 0.05) and for hypothermia vs. MgCl₂(1:1) only at the two highest preloads (p < 0.05). A similar view emerged considering stroke work where significant differences resulted for hypothermia vs. Mix 45′ (p < 0.05) and MgCl₂ vs. Mix 45′ (p < 0.05, only at the last preload value). No other significant differences emerged when considering other parameters.

Figure 1 also shows that the Frank-Starling curves for hypothermia, Mix 20′ and MgCl₂ are parallel to each other; in addition, the responses of the hearts isolated under MgCl₂(1:1) and Mix 45′ are similar to each other. In the latter the preload-stroke volume relationship was shallow in comparison to that in hearts isolated under hypothermia, Mix 20′ or MgCl₂. These differences in slope are also reflected in calculations of the change in stroke volume per kPa change in input pressure (δ kPa) using the values from the individual Frank-Starling curves; for hypothermia (n = 6): 2.01 ± 0.21 ml/g/kPa; for MgCl₂ (n = 5): 1.4 ± 0.10 ml/g/kPa; for MgCl₂ (1:1; n = 5): 1.01 ± 0.16 ml/g/kPa; for Mix 20′ (n = 5): 1.52 ± 0.14 ml/g/kPa; for Mix 45′ (n = 6): 1.08 ± 0.18 ml/g/kPa.

Perfusion with increasing concentrations of different anesthetic solutions was carried out on systemic hearts removed under hypothermia and the effects on HR, SV, and Q are plotted in Figure 2. All anesthetics, except ethanol, induced bradycardia in a concentration-related manner. In particular, MgCl₂ and Mix significantly reduced HR, Q, and SV (p < 0.01 after Mixed Model ANOVA, except HR with EtOH) at the higher concentrations tested (Figure 2).

MgCl₂ (1:1) appears to be a more potent cardio-suppressive agent than either MgCl₂ or Mix because it induced a significant reduction of cardiac parameters starting at a lower MgCl₂ concentration. All anesthetic solutions were able to produce cardiac arrest at the higher concentrations used (Figure 3) with the incidence of arrest lowest with MgCl₂ and ethanol alone. In all cases the heart could be restarted in <10 min, when re-perfused with PS alone (data not shown).

To investigate if the MgCl₂ formulations used for anesthesia in vivo could have a direct effect on cardiac function, the systemic hearts were removed from animals killed under hypothermia and perfused with increasing concentrations of anesthetic solutions. The results show that, in contrast to ethanol alone, all the anesthetic formulations containing MgCl₂ (sw, 1:1 and Mix) produced a concentration-related, acute onset bradycardia which at the higher concentrations arrested some hearts (Figure 3).

The bradycardia induced by the various MgCl₂ formulation in vitro is consistent with reports that the heart rate is “very low” in O. vulgaris anesthetized with 3.5% MgCl₂ in sea water (Fiorito et al., 2015) and “slow” in the same species anesthetized with MgCl₂+EtOH (Grimaldi et al., 2013). Ethanol alone (from 0.125 to 1.0%) did not induce a significant bradycardia indicating that chronotropic effects of the MgCl₂+EtOH mixture (i.e., MgCl₂ 1.12 % + EtOH 1%) are not due to the ethanol. It should also be noted that the highest concentration of MgCl₂ tested in vitro (2%) and which caused arrest in 50% of hearts was lower than the 3.5% commonly used for anesthesia in vivo. This may indicate that in vivo additional mechanisms such as ganglionic or central nervous system reflexes operate to protect the myocardium.

In addition to the bradycardia, a significant reduction in stroke volume was observed with the effects being more marked with MgCl₂ (1:1) and Mix compared to the same concentration (1%) of MgCl₂ in sea water (Figure 2). Again, comparison of the effects of Mix (1.12% + 1%) and ethanol alone (1%) shows that whilst there is a concentration related bradycardia with Mix (Figure 2), this is not the case with ethanol alone, although both Mix and ethanol can produce cardiac arrest (Figure 3).

It should be also noted that the highest concentration of ethanol we studied in vitro (1%) is lower than that commonly used (2–3%) to anesthetize cephalopods when used as the sole agent (review in Fiorito et al., 2015).

This study has established that three formulations of magnesium chloride used to anesthetize cephalopods affect the cardiac function and in the case of MgCl₂ (1:1) and Mix 45′ we observed persistent effects. Investigation of mechanisms responsible was beyond the scope of this study but, for completeness, the most likely mechanisms are outlined below to provide pointers for future investigation.

Formulations of magnesium chloride utilized to anesthetise cephalopods cause bradycardia in the isolated heart. This is consistent with the descriptive reports of heart rate in anesthetized octopuses (e.g., Young, 1971a; see also: Andrews et al., 2013; Fiorito et al., 2015). A marked fall in heart rate together with the effects of magnesium chloride formulations on stroke volume will reduce cardiac output leading to a decrease in brain perfusion. The fall in haemolymph flow to the brain together with the marked reduction of ventilation is likely to lead to ischaemia of the brain (and other tissues) that could contribute to the anesthetic state induced by the various magnesium chloride formulations. Direct effects of the magnesium chloride formulations on the heart are likely to be compounded by the suppression of ventilation and muscular activity of the arms caused by anesthesia which contribute to venous return (Wells, 1983; King et al., 2005). In addition, acute hypoxia is itself associated with bradycardia and reduced aortic flow in cephalopods, as reported, for example, for Nautilus pompilius (Boutilier et al., 2000) and O. vulgaris (Wells, 1983). Furthermore, in vivo the anesthetic formulations may also have effects on the brain vasomotor lobe or peripherally on the cardiac ganglia.

Despite the effects of magnesium chloride formulations on ventilation and the heart, following anesthesia on returning an O. vulgaris to fresh sea water all externally assessed parameters return within ~30 min (review in Fiorito et al., 2015). Recovery from anesthesia is usually uneventful with octopus feeding within an hour (for review see Fiorito et al., 2015). However, marked suppression of cardiac function is an undesirable property of an anesthetic, particularly as in the case of the 3.5% MgCl₂ (1:1 sw:dw) and prolonged exposure to MgCl₂ and ethanol mixture (Mix 45′) where it is associated with residual deleterious cardiac effects.

In conclusion, this study has demonstrated for the first time the acute direct effects on cardiac function of three formulations of magnesium chloride used as anesthetics in cephalopods following Messenger et al. (1985). A direct effect of Mg⁺⁺ on cardiac calcium fluxes can account for the marked bradycardia observed in animals anesthetized with MgCl₂. Evidence was obtained for a residual effect of MgCl₂ on cardiac function that could compromise recovery, but this was apparent only with either prolonged exposure (45 min) or using the sw:dw formulation.

Based upon the overall assessment of the acute (bradycardia/arrest and decreased stroke volume: Protocol II) and protracted (reduced Frank-Starling curves: Protocol I) effects, 3.5% MgCl₂ in sea water and Mix formulations had the least deleterious combination of effects provided exposure time is minimized (within 30 min or less). These formulations may be suitable for procedures where relatively short duration (~20 min) anesthesia is required and for studies using non-invasive approaches (e.g., Grimaldi et al., 2007; Margheri et al., 2011).

In addition, these results pave the way for mechanistically oriented studies aimed to re-examine the most common anesthetic practices used with cephalopods in relation to the application of Directive 2010/63/EU.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.