Critical temperatures in the cephalopod Sepia officinalisinvestigated using in vivo³¹P NMR spectroscopy

An oxygen limitation of thermal tolerance was proposed to be a unifying principle in ectothermic water-breathing animals(Pörtner, 2001; Pörtner, 2002a) and has been supported by the studies on several invertebrates [Arenicola marina (Sommer et al.,1997), Maja squinado(Frederich and Pörtner,2000), Sipunculus nudus(Zielinski and Pörtner,1996), Littorina littorea(Sokolova and Pörtner,2003), Laternula elliptica(Peck et al., 2002) and the fish species Rutilus rutilus(Cocking, 1959), Pachycara brachycephalum (Mark et al.,2002) and Gadus morhua(Sartoris et al., 2003; Lannig et al., 2004)]. According to this hypothesis, ectothermic animals subjected to acute temperature change first experience a loss in aerobic scope and, second,oxygen deficiency beyond low and high critical temperature thresholds(T_c), ultimately leading to a transition to an anaerobic mode of energy production and time-limited survival. It was further proposed that oxygen limitation mechanisms first set in at high levels of organismal complexity, namely the integrated function of the major convection systems of the oxygen delivery apparatus.

Accordingly, previous studies have identified both ventilatory and circulatory (where present) systems in invertebrates to be limiting for oxygen transport during acute temperature change(Zielinski and Pörtner,1996; Frederich and Pörtner, 2000), while circulatory insufficiency was suggested to be the first limiting process in fish species acutely subjected to high and low temperature extremes (Heath and Hughes, 1973; Mark et al.,2002; Lannig et al.,2004).

The present study, working with the cuttlefish Sepia officinalis,will focus on an animal model that, although being an invertebrate, is characterised by several vertebrate-like features. Namely, sophisticated behaviours, a closed, high-pressure blood convection system with low blood volume and a highly efficient counter-current gill gas exchange system similar to that of fish (Wells and Wells,1982; Wells and Wells,1991). These features make the cuttlefish an ideal candidate to test the proposed universal character of thermal tolerance limitation mechanisms (Pörtner,2002a).

An obvious first step is the evaluation of whether or not cuttlefish will display oxygen-limited thermal tolerance upon acute exposure to low and high temperature extremes, as evidenced by transition to an anaerobic mode of energy production at rest. Advent of mitochondrial anaerobiosis in highly aerobic liver tissue determined high T_c in fish(Pachycara brachycephalum and Zoarces viviparus), while the onset of anaerobic metabolism in white muscle only occurred shortly before death (van Dijk et al., 1999). Choice of tissues therefore seems to be crucial when investigating T_cs. Continuously working, aerobic organs with a permanently high oxygen demand are the first to suffer from oxygen deficiency and functional failure. Accordingly, we decided to monitor cephalopod mantle muscle energy status.

Cephalopod mantle tissue is permanently active and is involved in ventilatory work but also in jet-propelled locomotion: as a result, the mantle muscle evolved as a complex organ that contains thin outer layers of aerobic circular muscles, which aid during slow swimming contractions, and thick layers of anaerobic muscle fibres in the central part of the mantle muscle organ, which produce the high-pressure amplitude contractions during fast swimming (Bone et al., 1994a; Bone et al., 1994b; Bartol, 2001). The inner and outer layers of circular fibres possess high densities of mitochondria(Bone et al., 1981; Mommsen et al., 1981), leading to enhanced baseline oxygen demand. Approximately 30% of the mantle volume consists of radial muscle fibres that have a key role in ventilation(Milligan et al., 1997). By contracting, these radial muscle fibres decrease mantle muscle organ diameter to aid in refilling the mantle cavity with fresh seawater during each ventilation cycle. Funnel collar flap movements in combination with a passive relaxation of radial fibres aid during exhalation of the respiratory water. Contractions of circular muscle fibres are not involved in exhalation(Bone et al., 1994a).

Due to their permanent workload, muscle tissues active in ventilation are probably a good indicator tissue for the determination of T_cs. Other organs/tissues that are less vital during short-term stresses (i.e. digestive system, reproductive system) may even be temporarily shut down while essential fuels (nutrients, oxygen) are reallocated towards the organs that are maintained in continuous operation. Examples for such selective energy allocation towards certain organs and metabolic depression within other organs/tissues during acute environmental stressors are manifold in the animal kingdom [e.g. the mammalian diving response (Hochachka, 2000)associated with cellular metabolic depression (e.g. Buck et al., 1993)]. If,during thermal stress, anaerobic metabolism is needed to fuel ventilatory muscle contractions, obviously time-limited survival has set in and critical temperatures are being reached.

Using in vivo³¹P NMR spectroscopy, we had a technique available to continuously monitor concentrations of mantle muscle intracellular high-energy phosphate compounds and intracellular pH(pH_i) in unrestrained animals. Net utilization of the phosphagen(phospho-l-arginine, PLA) to fuel muscle contractions is a sign of the start of anaerobiosis in molluscs. We could simultaneously monitor in vivo performance of the very muscle fibres observed with the NMR setup by recording mantle cavity pressure oscillations. Such pressure oscillations in the mantle cavity are a consequence of rhythmic action of the mantle muscle organ's ventilatory and locomotory muscles(Bone et al., 1994a; Bone et al., 1994b).

The complexity of mantle muscle structure and the complexity of various ventilatory and locomotory functions may be confounding factors in the analysis. It was thus necessary to first learn more about the various mantle cavity pressure patterns present and their influence on muscle tissue energy status under control conditions, to later use the acquired knowledge to distinguish between (putative) effects of high-pressure mantle contractions related to spontaneous exercise and those of ventilatory pressure cycles on tissue energy status at thermal extremes.

We hypothesized that at both high and low temperatures, mantle muscle metabolism would need to switch to anaerobic metabolism during resting conditions to sustain ventilatory activity. This report therefore concentrates on muscle energy status and the effects of ventilatory activity on mantle metabolism. It will, at the same time, address the possibly interfering effects of spontaneous, exercise-related high-pressure circular muscle contractions on muscle energy status. In a companion study (F.M., C.B. and H.-O.P., manuscript submitted for publication), we analyse how the pressure patterns generated during resting ventilation relate to oxygen extraction from the ventilatory stream, metabolic rate and the costs of ventilatory movements.

European cuttlefish (Sepia officinalis L.) used in the present study were grown from egg clusters trawled in the Bay of Seine (France) in May 2002. The animals were raised in a closed re-circulated aquaculture system(20m³ total volume, protein skimmers, nitrification filters,UV-disinfection units) at the Alfred–Wegener–Institute on a diet of mysids (Neomysis integer) and brown shrimp (Crangon crangon) under a constant dark:light cycle (12 h:12 h) and constant temperature regime (15±0.1°C). Water quality parameters were monitored three times per week. Concentrations of ammonia and nitrite were kept below 0.2 mg l^–1, and nitrate concentrations below 80 mg l^–1. Salinity was maintained between 32 and 35‰, and water pH between 8.0 and 8.2. All animals were raised in the same 3 m³ volume tank. Five animals (104.2±7.4 g wet mass, mean± s.d.) were used for experimentation.

Experimental animals were starved for 24 h and then transferred to the experimental set-up. Surgery was conducted on the first day, followed by an overnight acclimatization period within the experimental chamber (control 1). In vivo³¹P NMR spectra showed that anaesthesia during surgery resulted in a transient accumulation of inorganic phosphate(P_i), which could be fully reversed within 4–6 h of recovery under control conditions. On the second day, animals were cooled from control temperature (15°C) to a lower critical temperature, then warmed to and kept at control temperature overnight (control 2), after which they were finally warmed until an upper critical temperature was reached on the third day. Temperature was changed in a stepwise procedure at an average rate of 1 deg. h^–1. Specifically, a 3°C temperature change was accomplished during the first hour of a 3 h period, while temperature was kept constant for in vivo³¹P NMR spectroscopy and mantle cavity pressure measurements during the two subsequent hours. Assay temperatures were 14°C/11°C/8°C on the second day, and 17°C/20°C/23°C/26°C on the third experimental day. Temperatures were changed further at a rate of 1 deg. h^–1 if critical temperatures were not reached within the outlined temperature window. As the accumulation of P_i due to phosphagen utilization indicates limited energy production by aerobic metabolism, the appearance of significant P_i peaks in in vivo³¹P NMR spectra and their persistence over an extended time period (>60 min) defined critical temperatures.

To implant a catheter for ventilatory monitoring, animals were anaesthetized with a 0.4 mol l^–1 MgCl₂ solution that was mixed 1:1 with seawater (Messenger, 1985) at 15°C for 3–3.5 min, then placed (ventral side up) on a wet leather cloth to prevent skin injuries. During surgery, animals were perfused with aerated seawater (with 0.04 mol l^–1 MgCl₂) through the funnel aperture. A PE cannula, required to record postbranchial pressure, was connected to a 23-gauge hypodermic needle, led through the entire mantle cavity and then fed through the posterior ventro-lateral section of the mantle muscle. Cannulae(Portex PE tubing, i.d. 0.58 mm, o.d. 0.96 mm, flared at the opening) were held in place by two 4 mm-diameter plastic washers on the inside and outside,embracing the mantle muscle in a sandwich-like fashion. PE tubes were connected to MLT-0699 disposable pressure transducers, signals were amplified with a ML-110 bridge amplifier and were further fed into a PowerLab/8SP data acquisition system (AD Instruments GmbH, Spechbach, Germany). Pressure transducers were calibrated daily.

Following surgery, animals were placed in a Perspex perfusion chamber analogous to the one used by Mark et al.(Mark et al., 2002) for eelpout (Fig. 1A). Plastic sliders within the chamber could be adjusted to the animals' dimensions and used to restrict the space available for roaming activity. The chamber was connected to a closed recirculation seawater system and placed within the magnet as described by Bock et al. (Bock et al., 2002). Water quality was maintained with a protein skimmer(Aqua Care, Herten, Germany) and a nitrification filter (Eheim Professional 2;Eheim, Deizisau, Germany). Water quality was monitored daily, and parameters were kept within the limits indicated above.

In vivo³¹P NMR spectroscopy experiments were conducted in a 47/40 Bruker Biospec DBX system with a 40 cm horizontal wide bore and actively shielded gradient coils (50 mT m^–1). A 5 cm ¹H/³¹P/¹³C surface coil was used for excitation and signal reception. The coil was placed directly under the animal chamber in such a manner as to gain maximum signal from the posterior mantle muscle section (Fig. 1B). A calculated 80–90% of sensitive coil volume was filled by mantle muscle tissue and ∼10–20% by tissues from the organ sac and coelomic fluid. As Sepia officinalis mantle muscle tissue is characterized by high phosphagen (PLA) concentration [∼34 μmol g^–1 wet mass(Storey and Storey, 1979)], it is quite likely that the in vivo³¹P NMR spectra almost exclusively represent the adenylate pool and pH_i of the mantle musculature.

Concentrations of Arg and Oct were estimated using published values(Storey and Storey, 1979) and assuming that a decrease in 1 μmol g^–1 wet mass [PLA]results in a concomitant 0.67 μmol g^–1 wet mass increase in [Arg] and a 0.33 μmol g^–1 wet mass increase in [Oct](Storey and Storey, 1979) (as witnessed during moderate and severe hypoxia and during exercise). Intracellular free [Mg²⁺] was estimated from ³¹P NMR spectra as described by Doumen and Ellington(Doumen and Ellington,1992).

Pressure oscillations in the cephalopod mantle cavity are generated to create both a ventilatory water stream past the gills (mainly by concerted action of the collar flap muscles of the funnel apparatus and radial mantle muscles) and a jet stream to elicit swimming and escape movements (mainly by action of circular and radial mantle muscles). While the former is associated with low-pressure amplitudes, the latter can cause amplitudes of up to 25 kPa(Wells and Wells, 1991). According to Bone et al., slow swimming in cuttlefish starts with pressure amplitudes of 0.1–1 kPa, while maximum ventilation pressure amplitudes(at rest) are 0.15 kPa (Bone et al., 1994). For an analysis of the interfering influence of spontaneous activity on temperature-dependent muscle energetics,we set a pressure threshold of 0.2 kPa as the starting point for non-ventilation-related mantle pressure generation. All such pressure oscillations will be termed `swimming jets' (SJ) hereafter. For all five animals, frequencies of SJs of >0.2 kPa amplitude were analysed at all temperatures. In addition, mean mantle pressure amplitudes of these SJs(MMPA_SJ) were recorded (and grouped into 12 amplitude classes, i.e. 0.2–1 kPa, 1–2 kPa, [...], 11–12 kPa).

Mantle contractions associated with high-amplitude pressure pulses were present at all temperatures. The major challenge in the present study was thus to distinguish between the effects of spontaneous exercise and the effects of routine ventilatory mantle muscle activity (related to ventilation only) on muscle metabolic status. An extensive base of control ³¹P NMR spectra at 15°C was available to investigate activity patterns and patterns of metabolite change in the mantle organ.

As a second step, an attempt was made to correlate metabolic changes observed within the mantle muscle with non-ventilatory muscle contractions (of pressure amplitudes >0.2 kPa). It is known from previous work that such mantle muscle contractions are fuelled in part by phosphagen breakdown, as aerobic metabolism cannot provide sufficient amounts of ATP to match demand at very high ATP fluxes (e.g. Pörtner et al., 1993; Finke et al.,1996). For this, in vivo³¹P NMR spectra were scanned for relative increases in [P_i] (which is equivalent to a net phosphagen breakdown). A total of 30 intervals from all five animals was randomly selected, and all associated mantle contractions within each interval were analysed. To find a (putative) causal relationship between spontaneous exercise and metabolite changes in the mantle organ, it was necessary to identify those suitable time intervals and pressure amplitudes that actually have an effect on muscle phosphagen stores. In an exercise study on squid(Illex illecebrosus), it could be demonstrated that mantle phosphagen levels returned close to control levels within 10 min after fatiguing exercise, where PLA had decreased by –22.5 μmol g^–1wet mass from an initial concentration of >30 μmol g^–1wet mass (Pörtner et al.,1993). Thus, in our case, a major spontaneous exercise event,occurring 20 min prior to ³¹P NMR spectrum acquisition, might not be reflected in the latter due to a putative rapid recovery phase of PLA stores.

Consequently, intervals (21 min 20 s between the acquisition of two ³¹P NMR spectra plus the acquisition time of the second spectrum;thus, a total of 25 min) were divided into 11 two-minute segments and one 3 min segment (s1–s12; see Fig. 1c). Both the frequency and amplitude of SJs greater than 0.2 kPa were determined for each segment. Pressure amplitudes were grouped into classes with the following class means: 0.6 kPa (=0.2–1 kPa), 1.5 kPa(=1–2 kPa) [...] up to 11.5 kPa (=11–12 kPa) and a jet index (JI,in kPa segment^–1) obtained for each segment by adding the products of class amplitude means and jet numbers within the respective amplitude classes. For example, five jets between 2 and 3 kPa and three jets between 5 and 6 kPa within one segment yield a JI of(5×2.5)+(3×5.5)=29 kPa. 12 variables were created by adding JIs from segments within intervals `a' to `l' (see Fig. 1C). Furthermore, each of these variables was split up by varying the pressure threshold used for JI calculation (only jets >0.2 kPa or >1 kPa or >2 kPa, [>...], or>11 kPa used for calculations). Thus, for our example above, a JI calculated from jets >3 kPa would result in 3×5.5=16.5 kPa. This variation in duration of the interval and in the pressure amplitudes taken for JI calculation created a total of 12×12=144 different variables that could be tested in an iterative linear regression analysis to explain a maximum of the variability observed in [P_i] changes as obtained by successive in vivo³¹P NMR spectra. In a similar fashion,maximum jet density (JD) of Sjs >0.2, >1 [>...], >11 kPa within segments of intervals a to l (Fig. 1C) was calculated (again, 144 possible variables) as a second factor that might influence [P_i] changes in mantle muscle tissue. For example, JD was 15 for jets >1 kPa if we considered interval d (see Fig. 1C), and the density of jets >1 kPa (in jets 2 min^–1) in segments s9, s10, s11 and s12 were 9, 12, 15 and 3.

Simple linear, exponential and sigmoidal regression analyses were performed using SigmaPlot 8.0 (SSPS Inc., Point Richmond, CA, USA). Multiple linear regression analysis was also performed using Statistica (Statsoft, Tulsa, OK,USA), as were all other statistics. Comparisons between values grouped according to temperature or animal were conducted using one-factorial analysis of variance (ANOVA) and subsequent post-hoc testing with Student–Newman–Keuls. t-tests were used to compare SJ frequencies obtained during the day with those obtained at night.

Typical ventilation pressure amplitudes in all experimental animals were lower than 0.1 kPa, typically occurring for ≥57 min of each control hour. The remaining time was filled with spontaneous high-pressure mantle contractions of >0.2 kPa. Occurrence of such SJs was observed in all animals, with pressure amplitudes distributed as shown in Fig. 2A. It appeared that roughly 73% of all non-routine ventilation pressure cycles of >0.2 kPa were characterized by an amplitude lower than 2 kPa. Only 0.2% of all pressure cycles of >0.2 kPa showed pressure amplitudes of 11–12 kPa.

The mean SJ pressure amplitude of all SJs >0.2 kPa was found to be 1.7 kPa. Frequencies of 108 (181) pressure cycles >0.2 kPa h^–1during daily (nightly) control measurements were recorded; these did not appear to be evenly distributed over time but rather were found to occur in groups of 3–50 SJs. A significant difference in high-amplitude pressure cycle frequency between night-time and daytime control measurements was also evident, with more jets observed at night (t=3.5, d.f.=4, P<0.03, N=5).

Concentrations of in vivo³¹P NMR visible metabolites and pH_i were found to be as variable as mantle pressure recordings. Fig. 3 gives an example: data are shown from 14 subsequent control ³¹P NMR spectra taken from animal 4 between 23:22 h and 04:47 h during the second control phase of the experiment (at constant 15°C). While muscle ATP concentrations remained constant over the whole period, [PLA] decreased from 33.4 to 23.8 μmol g^–1 wet mass (Δ[PLA]=–29%) between 50 and 100 min. This was mirrored by a concomitant increase in [P_i] from 0.8 to 11.6 μmol g^–1 wet mass. Recovery of the phosphagen pool in combination with a decrease in [P_i] started at t=100 min and lasted for 150–200 min. pH_i values followed a similar pattern when compared with [PLA] but were delayed by ∼50 min. During the first 25 min of phosphagen utilization, pH_i increased from 7.43 to 7.49. Thereafter, pH_i values decreased continuously to a minimum value of 7.32 at t=150 min. A value close to control was reached again after 325 min. The observed P_i accumulation of 11.6 μmol g^–1 wet mass was the highest observed in any of the five animals under control conditions (Table 1).

Patterns of metabolite concentration changes and correlated pH_ichanges appeared to be comparable between all animals. To elaborate these patterns, changes in metabolite concentration and pH_i were analysed from in vivo³¹P NMR spectra over time(Fig. 4). As would be expected from Fig. 3, Δ[PLA]changed linearly with Δ[P_i], with a negative slope close to one (Fig. 4B), while ATP concentrations remained stable at ∼7–9 μmol g^–1wet mass (ANOVA, F_7,272=1.83, P<0.07; Fig. 4D). pH_ichanged linearly with increasing Δ[P_i] between less than–3 and 3 μmol g^–1 wet mass(Fig. 4A) but deviated significantly from this relationship once [Pi] accumulated above 3 μmol g^–1 wet mass. pH_i fell back to 7.44 (control pH was 7.45; see Table 1) at those higher inorganic phosphate accumulations. From 282 intervals analysed for Fig. 4, only nine intervals(3%) were characterized by such a high increase in [P_i]. Correspondingly, intracellular Δ[H⁺] changed linearly in the respective range of Δ[P_i] values(Fig. 4C). Increases in[P_i] of >3 μmol g^–1 wet mass did not result in a further proton buffering but rather led to a net increase in[H⁺] by 3 nmol l^–1.

As high spontaneous SJs could be related to the accumulation of inorganic phosphate observed in mantle muscle, true control values for [ATP], [PLA] and pH_i were calculated only from spectra with [P_i]<1.5μmol g^–1 wet mass, as this was the maximum [P_i]found during prolonged routine ventilation sequences[Table 1; only sporadic(<10) jets of pressure amplitudes of <1 kPa for at least 30 min]. Mean pH_i values were comparable between animals, except for animal 3,which showed a significantly higher mean muscle pH_i. Variability in[H⁺_i] was low under control conditions, with a relative standard deviation (CV) of about ±15% at a mean concentration of 36.1 nmol l^–1. ATP concentrations were comparable between animals 2–5, while animal 1 had a significantly lower muscle [ATP] than animals 2, 4 and 5. Although there were significant differences found in [PLA] between animals, it has to be considered that all mean concentrations were found within a range of –1.6 to +2.8% of the mean value of 33.6 μmol g^–1 wet mass. The ratio of [PLA] over [ATP] proved to be relatively stable between animals (4.0–4.6), with ratios being comparable from animals 2–5 and only animal 1 being characterised by a significantly higher ratio (Table 1).

Therefore, grouping animals into pre-P_i accumulation (group A,means of in vivo³¹P NMR spectra 60 min prior to P_i accumulation) and P_i accumulation [group B, means of in vivo³¹P NMR spectra during P_i accumulation(60 min duration), with the start of accumulation defined as at least two successive spectra with a [P_i] of >1.5 μmol g^–1 wet mass] enabled us to improve the resolution of metabolic patterns at both ends of the temperature window(Table 3). In the cold, phase B mean temperature was 7°C. From phase A to B, [PLA] had decreased to 30.9μmol g^–1 wet mass while [ATP] remained constant. pH_i was also comparable between phases A and B. At the warm end of the temperature spectrum, the picture was similar. At a mean temperature of 26.8°C, [PLA] decreased to 30.5 μmol g^–1 wet mass,while [ATP] and pHi did not change from phase A to B. Looking at the last ³¹P NMR spectra taken at each temperature (B_extreme in Table 3) illustrates that at high temperatures, pH_i is significantly decreased compared with phase A. A trend towards decreased pH_i values is also visible at the low-temperature B_extreme, although it is not (yet) significant. Free-energy changes of ATP hydrolysis |dG/dξ |decreased by 3.3 kJ mol^–1 at the low temperature B_extreme and by 5.1 kJ mol^–1 at the high B_extreme. Mean values did not drop below 50 kJ mol^–1, if one assumes that phosphagen utilization is distributed evenly among all muscle fibre types [radial and circular fibres(r+c) in Table 3] present in the sensitive volume of the ³¹P NMR coil.

In a second step, we looked at [P_i] variability during 25 min intervals with no high-pressure jets of >0.2 kPa present. At both low and high extreme temperatures (phase B), a mean (± s.d.) accumulation of[P_i] could be found (high temperature,Δ[P_i]=1.15±0.3 μmol g^–1 wet mass, N=8 intervals; low temperature,Δ[P_i]=0.53±0.43 μmol g^–1 wet mass, N=9 intervals), while during (randomly) chosen control intervals with no high-pressure cycles present, no inorganic phosphate increases could be found at all (Δ[P_i]=–1.65±1.9 μmol g^–1 wet mass, N=8 intervals). Rather, negativeΔ[P_i] values dominated under control conditions, as periods without any SJs at all were found predominantly during recovery times from(spontaneous) exercise.

In summary, we have found significant increases in [P_i] (=phosphagen use) in the mantle muscle organ at both high and low temperature extremes in all five animals investigated. At a control temperature of 15°C, [P_i] variability could almost completely (89%) be explained by the occurrence of high-pressure SJs. At high extreme temperatures, SJs could not be related to the observed increases in[P_i]. Inorganic phosphate accumulation was also observed during intervals without any SJs present, and thus was likely to be caused by elevated levels of routine ventilation alone. At low extreme temperatures,apparently both processes (routine ventilation energy demands and spontaneous exercise energy demands) contributed to [P_i] accumulation.

Using non-invasive in vivo³¹P NMR spectroscopy, we were able to continuously monitor key components of the S. officinalis mantle muscle energy system in unrestrained animals subjected to acute temperature changes. Our aim was to identify potential threshold temperatures, below or above which aerobic metabolism would insufficiently provide the energy required for muscle maintenance, routine activity(ventilatory contraction of radial fibres) and facultative exercise(high-pressure contractions performed by aerobic and anaerobic circular fibres). A number of invasive studies on cephalopod mantle organ tissue has provided the necessary information to interpret the observed changes in NMR visible metabolites ([PLA], [P_i], [ATP]) and pH_i,following the advent of anaerobic metabolism, both in mitochondria and the cytosol (Storey and Storey,1979; Gaede, 1980; Finke et al., 1996; Pörtner et al., 1991; Pörtner et al., 1993; Pörtner et al., 1996; Zielinski, 1999; Zielinski et al., 2000).

During control conditions, we had the unique possibility to study the effects of facultative mantle muscle exercise on the animals' intracellular energy status. We could show that animals displayed a higher activity (as evidenced by the occurrence of SJs with an amplitude of >0.2 kPa) during 1–3% of the total time. Similar levels of activity were recently found for a tropical cuttlefish in its natural habitat [S. apama was found to be active (as estimated from elevated mantle pressures) for approximately 3% of the day (Aitken et al.,2005)].

Control mantle muscle organ adenylate levels obtained in our study, as witnessed by the ratio of [PLA] over [ATP], are comparable with other studies that analysed metabolite concentrations invasively. Our ratio of 4.2 (0.4 s.d.) compares well with a value of 3.9 for S. officinalis(Storey and Storey, 1979), 4.4 for the squid Lolliguncula brevis(Finke et al., 1996) or 3.5 for the squid Loligo pealei(Pörtner et al., 1993). We were able to clearly demonstrate that PLA stores were being utilized during high-pressure SJs and could establish a quantitative relationship between exercise levels and decreases in [PLA] or increases in [P_i],respectively. The best relationship obtained suggests that SJs with a MMPA of>2 kPa cannot be fuelled entirely by means of aerobic energy production,similar to findings in the squid L. brevis(Finke et al., 1996). This fits the picture that slow-swimming (`cruising') mantle pressure amplitudes in S. officinalis do not seem to exceed 2 kPa (fig. 1b in Wells and Wells, 1991). Bone et al. could demonstrate that during slow swimming at pressure amplitudes between 0.1 and 1.0 kPa, (aerobic) circular fibres become involved in the pressure generating process (Bone et al.,1994a). Thus, it is quite likely that slow swimming at <2 kPa can be fuelled entirely by metabolism of the thin aerobic muscle layers of the mantle periphery. More extreme exercise with pressure amplitudes of >2 kPa progressively involves central anaerobic fibres(Bone et al., 1994b) that exploit their phosphagen reserves in order to propel the animal at higher speeds. Gradual involvement of central (mainly) anaerobic fibres (rather than an all-or-nothing transition) to support aerobic fibres at an increasing workload has been recently demonstrated for swimming squid (L. brevis) (Bartol,2001).

The fact that we found a linear relationship between jet indices constructed from mantle pressure amplitudes multiplied with jetting frequencies and phosphagen use is not surprising, as pressure production should be a direct function of circular mantle muscle fibre force generation. Muscle fibre force has been found to depend on the number of crossbridges in the force generating state per cross-sectional area (e.g. Wannenburg et al., 1997);thus, mantle cavity pressure in cephalopods should be directly proportional to ATP flux rates in working mantle muscle. This corresponds to the results of Webber and O'Dor, who found correlated changes in mantle pressure integral and whole-animal rate of oxygen consumption(Ṁ_O2) during various levels of exercise in a squid (I. iIllecebrosus)(Webber and O'Dor, 1986).

The inclusion of a second variable into the regression model (jet density)significantly enhanced the fraction of explainable variation in inorganic phosphate concentration changes. This implies that high amplitude jets of>5 kPa pose a higher threat to cellular phosphagen reserves when they occur in quick succession, rather than distributed over a longer time interval. This could be due to oxygen depletion or aerobic fibre fatigue at high jet density,resulting in pressure generation exclusively by anaerobic fibres during high jet density, high-pressure time intervals.

It should be emphasized that PLA stores were never used extensively under control conditions (15°C) and during facultative, spontaneous exercise. Mean maximum decreases in [PLA] (= increases in [Pi]; see Table 1) under control conditions amounted to 6.66 μmol g^–1 wet mass, which corresponds to roughly 20% of phosphagen reserves, although one animal(replicate 4, Fig. 3) depleted 35% of its phosphagen reserves on one occasion.

Fig. 3 demonstrated that during initial phosphagen transphosphorylation and P_i accumulation,pH_i can be buffered and remains unchanged. pH_i decreased only during prolonged phosphagen utilization, probably due to glycolysis and concomitant octopine formation (Storey and Storey, 1979; Pörtner,1987; Pörtner,2002b). Work on in vitro preparations of scallop(Argopecten irradians) contracting phasic adductor muscles supports this conclusion (Chih and Ellington,1985). While, during initial exercise (40 muscle contractions),proton consumption by phosphagen utilization exceeded proton production by octopine formation, resulting in a net alkalosis of about 0.09 pH units(Δ[H+]=–16 nmol l^–1), further exercise(40–200 contractions) led to progressively declining pH_ivalues due to glycolytic proton production outmatching proton consumption by the phosphagen. Fig. 4 provides a more quantitative picture and shows clearly that anaerobic metabolism is seldom employed to the degree that net cellular acidification occurs. Only 3%of all (randomly chosen) intervals analysed for Fig. 4 showed a [P_i]accumulation of >3 μmol g^–1 wet mass. Such a degree of phosphagen utilization goes along with the onset of muscle acidosis,suggesting that glycolytic proton production outmatches proton buffering by phosphagen use.

Apparently, cuttlefish avoid intracellular acidification by terminating exercise in most cases as soon as glycolytic proton production equals phosphagen proton buffering capacity. Upon removal of inorganic phosphate during recovery, a glycolytic proton surplus, which cannot be buffered,results in a slight decrease in pH_i(Fig. 4A,B). Potentially adverse effects of decreased pH_i values on muscle function(reviewed by Fitts, 1994) are thus shifted into recovery phases. Absolute changes in intracellular proton activities are low (in the nmole range; Fig. 4B), a feature also observed by Chih and Ellington (Chih and Ellington, 1985). A recent in vivo³¹P NMR spectroscopy study on forced scallop exercise(Bailey et al., 2003) confirmed the metabolic patterns obtained in the older in vitro study on stimulated adductor muscle preparations.

Fig. 5 depicts the changes in mantle organ metabolite levels with temperature. Despite the dramatic changes in ventilatory power output over the entire temperature range, [ATP]is strictly conserved (Fig. 5A), a phenomenon commonly encountered in studies on muscles of marine ectothermic animals subjected to acute temperature change (i.e. Mark et al., 2002; Sartoris et al., 2003; Zielinski, 1999) and generally referred to as the `stability paradox'(Hochachka and Somero, 2002).[PLA] was also constant between 11 and 23°C.

We could not find any evidence for an alphastat pattern of pH_iregulation (Reeves, 1972; see Burton, 2002 for a review) in the investigated temperature range. Typically, changes of around –0.018 pH units deg.^–1 are expected to ensure constant levels of imidazol and protein ionization. For fish species, such a pattern could be demonstrated in white muscle (Borger et al., 1998; Van Dijk et al.,1997; Van Dijk et al.,1999; Bock et al.,2001). As for molluscs, an alphastat pattern of pH_iregulation was absent in the stenothermal marine bivalve Limopsis marionensis (Pörtner et al.,1999). Despite confounding effects of mantle muscle exercise at all temperatures, we found a decrease in pH_i with temperature by about –0.006 pH units deg.^–1 over the full temperature range examined. Omitting pH_i values at 8 and 26°C, as phosphagen utilization was observed to start at these temperatures, gives an even lower rate of change of –0.004 pH units deg.^–1. Between 11 and 17°C [the typical natural temperature range of this population of cuttlefish in the English Channel(Boucaud-Camou and Boismery,1991)], pH_i values are nearly identical. The absolute temperature-dependent changes in pH_i are <0.05 units between 11 and 23°C, which is lower than the range of change observed during facultative exercise at control temperature. Future studies should address the time dependence of pH_i regulation in response to temperature. Our study focused on short-term temperature effects and may not have allowed mantle pH_i to fully reach new steady-state values after each thermal challenge. As it stands, the question of temperature-dependent pH_i regulation in cephalopods must remain open.

Patterns of metabolite changes observed at extreme temperatures exactly mirrored those during exercise under control conditions. The start of phosphagen utilization was observed in all five animals at both temperature extremes during phase B. Mean temperatures during this phase were 26.8°C and 7°C.

The analysis of mechanisms at the high end of the temperature spectrum proved to be easier than at the low end. As no relationship between the few SJs (see Fig. 2B,C) and the use of the phosphagen could be established and, also, since [P_i]increases could be found during periods of ventilation at rest, obviously aerobic metabolic limitation had set in independent of effects of spontaneous SJ exercise at warm temperatures. Still, the results are difficult to interpret as mantle muscle is a complex organ that consists of different muscle fibre types with different functions(Bone et al., 1981; Bone et al., 1994a; Bone et al., 1994b; Bartol, 2001). Radial muscle fibres aid in refilling the mantle cavity during ventilation by contracting and thus enlarging mantle cavity volume. Bone et al. were the first to demonstrate (Bone et al.,1994a) that expiration in the cuttlefish under control conditions(18°C) and at rest (mantle pressure amplitudes of 0.05–0.15 kPa) is not brought about by contraction of the outer, aerobic layers of circular fibres but rather by the movements of the collar flaps (muscular funnel appendages) (see Tompsett,1939) that expel water rhythmically from the mantle cavity. Maximum resting ventilation MMPA recorded in our experimental animals were lower than 0.15 kPa (F.M., C.B. and H.-O.P., submitted), thus we assume that during our entire experimental series, radial fibres had been the only constantly working myofilaments within the sensitive volume of our ³¹P NMR coil (Fig. 1B). Our companion study revealed that ventilation pressures stagnate at temperatures beyond 26°C. Fig. 6A shows ventilation pressure amplitudes (at rest) at temperatures close to the upper T_c for two experimental animals (the ones with the highest and the lowest T_cs), while Fig. 6B gives correlated increases in [P_i] (circles). It is quite evident from this figure that declining ventilation pressures and phosphagen use are tightly coupled(all other experimental animals showed similar patterns). We thus conclude that an energetic limitation of radial muscle fibres is responsible for the observed increases in [P_i] and the correlated decreases in ventilation pressures once phosphagen usage starts. Radial fibres have a mitochondrial content as low as that of central `anaerobic' circular fibres(Bone et al., 1981; Mommsen et al., 1981) and thus may be especially sensitive to enduring ventilation exercise at high intensities. Considering that radial fibres constitute about 30% of total mantle volume in the cuttlefish (Milligan et al., 1997), and assuming that solely radial fibres deplete their phosphagen stores as ventilation pressures increase while, on the other hand, circular fibre energy status remains constant, it is possible to estimate metabolite changes for the radial fibre compartment. Based on such considerations, radial fibre |dG/dζ | for animals 1 and 5 would drop severely as phosphagen use proceeds(Fig. 6B). Following such a rationale, metabolite changes were recalculated for phase B (denoted r in Table 3) for all five animals. Thus, mean [PLA] reserves would decrease by 75% from 33 to 8.5 μmol g^–1 wet mass, mirrored by an increase in [P_i] in the same range and a 25% reduction in [ATP]. pH_i would be significantly decreased to 7.19 and |dG/dζ | would drop to below 44 kJ mol^–1 in the radial fibre compartment of the mantle organ. These calculations correspond to similar|dG/dζ | values for mantle muscle of three species of squid following fatiguing exercise, which ranged from 42 to 47 kJ mol^–1 (Pörtner et al., 1996), while values for two species of exercise-fatigued eelpout (P. brachycephalum, Z. viviparous)(Hardewig et al., 1998) white muscle ranged from 46.6 to 48 kJ mol^–1. Possibly, reductions in the free energy of ATP hydrolysis as calculated for cuttlefish radial muscle could contribute to muscle fibre fatigue in that vital muscle due to functional impairments of ATPase functions. Kammermeier et al. found a drop in contractile performance of perfused rat hearts (38°C) below a|dG/dζ | of 48 kJ mol^–1(Kammermeier et al., 1982). They calculated threshold values of |dG/dζ |required for proper function of the various ATPases engaged in muscular work to range from 45 to 53 kJ mol^–1(Kammermeier, 1987; Kammermeier, 1993). Only recently, Jansen et al. found maintenance of [Na⁺_i]homeostasis prevented by |dG/dζ | values below 50 kJ mol^–1 due to a limitation of the sodium pump(Na⁺/K⁺-ATPase) in perfused rat hearts(Jansen et al., 2003). Less information on critical |dG/dζ | values for muscle function is available for ectothermic animals: Combs and Ellington calculated an energy requirement of 41 kJ mol^–1 for blue mussel(Mytilus edulis) sarcolemmal Ca²⁺–ATPase(Combs and Ellington, 1995). A minimum |dG/dζ | value of 46 kJ mol^–1 for the sodium pump of crayfish abdominal muscle was calculated by the same authors (Combs and Ellington, 1997), although they found changes in[Na⁺_i] homeostasis well above 50 kJ mol^–1 already. They speculated that global|dG/dζ | might not reflect the|dG/dζ | close to the sodium pump.

Also, both elevated intracellular proton and inorganic phosphate concentrations have frequently been connected with muscular fatigue independent of |dG/dζ |(Allen and Westerblad, 2001; Fitts 1994). Increased[P_i] is thought to reduce muscle force by reversing the force-generating P_i release step by mass action(Hibberd et al., 1985). According to Debold et al. (rat muscle fibres), the [P_i] dependency of muscle fatigue is temperature related, with a more pronounced effect of[P_i] at low temperatures(Debold et al., 2004). For ectothermic (marine) animals such studies have, to our knowledge, not been undertaken.

Judging from the presented results and literature data, it appears that progressive phosphagen usage is indeed causative of the observed stagnation in ventilatory pressure generation at high temperature extremes, although the exact mechanisms still need to be elucidated.

At low temperatures, we could demonstrate that SJs contributed significantly to increases in [P_i]. 50% of variability inΔ[P_i] could be attributed to facultative exercise of circular anaerobic fibres, while the other half remained unexplained. As neither SJ frequency nor the fraction of MMP_SJ in MMP_tot increased at lower temperatures (similar frequencies/pressures were observed between 8 and 11°C; Fig. 2B,C) and no P_i accumulation occurred at 11°C(Fig. 5B), a major loss in aerobic scope for spontaneous activity had evidently occurred towards low temperature extremes, resulting in the net use of phosphagen stores for spontaneous activity. The other 50% of variability that could not be explained by the linear regression is likely due to a limitation in oxygen flux towards radial muscles, thus limiting their active role in ventilation. It seems that,at low temperatures, all muscle fibre types might suffer from phosphagen breakdown. Accordingly, we calculated |dG/dζ |values assuming that mantle phosphagen depletion took place more homogeneously. In B_extreme, |dG/dζ |had dropped from 55 to 51 kJ mol^–1. This does not exclude the existence of heterogeneity or even of putative intracellular|dG/dζ | gradients. Hubley et al. calculated distinct intracellular concentration gradients of the vertebrate phosphagen,creatine phosphate (PCr), and free energy of ATP hydrolysis in working fish white muscle, depending on the maximum distance from the nearest mitochondrion(Hubley et al., 1997). Maximum intracellular |dG/dζ | gradients of 7 kJ mol^–1 were calculated. Interestingly, the study suggests that in fish white muscle such |dG/dζ | gradients will be less pronounced in animals acclimated to high temperature (25°C) and acutely exposed to low temperature (5°C). This is mainly due to differing temperature relationships of metabolic rates and intracellular diffusion coefficients for ATP and PCr (i.e. Q₁₀ for D_PCr was only found to be 1.28, while, typically, Q₁₀ values for metabolism are>2). If this holds true for cuttlefish muscle as well, we can expect drops in |dG/dζ | to have less severe impact on intracellular ATPase function, and thus muscle function, during oxygen limitation developing at acute exposure to low temperatures.

However, although oxygen limitation of thermal tolerance in the cold likely progresses at a slower pace, phosphagen usage during periods without spontaneous activity is a certain sign of a cellular energy limitation,leading to time-limited survival of the organism.

Van Dijk et al. found no changes in fish (P. brachycephalum, Z. viviparus) white muscle energy status at high critical temperatures (|dG/dζ | did not drop below 60 kJ mol^–1), while in liver, an accumulation of succinate indicated a progressive switch to mitochondrial anaerobiosis(Van Dijk et al., 1999). Furthermore, two recent in vivo³¹P NMR studies on P. brachycephalum (Mark et al.,2002) and Gadus morhua(Sartoris et al., 2003) white muscle confirmed that only moribund animals displayed a drop in pH and|dG/dζ | in this tissue at temperatures beyond T_c. Sartoris et al. noted that even immediate cooling could not reverse detrimental changes in tissue energy status(Sartoris et al., 2003).

In our experiments, cuttlefish survived exposure to the high T_c, when rapidly cooled to control temperatures (F.M. and C.B., personal observation), suggesting that the observed net use of the phosphagen reflects an energetic limitation of actively working muscle sections rather than an energetic limitation of all muscle, including the resting bulk of circular and anaerobic (`white') muscle fibres. This emphasizes that active tissues are the first to be affected by temperature-induced oxygen deficiency. Effects in white or passive muscle set in at the end of a progressive oxygen limitation cascade and thus only occur once severe damage to other organs, especially those involved in driving blood circulation, has already set in, leading to ventilatory and circulatory failure as in fish (Van Dijk et al.,1999; Mark et al.,2002). Accordingly, progressive radial fibre fatigue of the cephalopod mantle characterizes an early limitation in response to temperature. It will eventually affect all other tissues, as oxygen uptake and distribution will likely be negatively affected by decreasing ventilatory pressures (Fig. 6A,B). However,whether radial muscle fatigue observed in our study results from limited ventilatory muscle power output per se or is caused by a progressively limited oxygen supply through the blood circulation or by a combination of both processes remains to be established.

This study was carried out in support of the project `The cellular basis of standard and active metabolic rate in the free-swimming cephalopod, Sepia officinalis' (NERC Grant PFZA/004). The authors wish to thank Rolf Wittig and Timo Hirse for their excellent technical support, Raymond and Marie-Paule Chichery (Université de Caen) for providing cuttlefish eggs in 2002 and 2003 and all student helpers that were engaged in raising the animals in our laboratory.