Hormonal control of luminescence from lantern shark (Etmopterus spinax) photophores

Lantern sharks (Etmopteridae) are deep-water Chondrichthyes, which received their name from their ability to produce a visible light through a chemical reaction. Due to the relative rarity of most of these sharks and the limited accessibility of their environment, the function of this bioluminescence has never been experimentally tested. Recent works strongly support the implication of this phenomenon in camouflage by counter-illumination as well as in intraspecific behaviours such as mating and schooling(Claes and Mallefet, 2008; Claes and Mallefet, 2009).

The velvet belly lantern shark (Etmopterus spinax) emits a blue luminescence from thousands of tiny (ca. 150 μm) photophores organised into nine different luminous zones forming a complex luminous pattern (Claes and Mallefet,2008; Claes and Mallefet,2009). These epidermic organs are relatively simple in structure,lacking reflectors or specialised filters(Hubbs et al., 1967). The light from the photocytes (i.e. the photogenic cells) passes through the lens cell(s) before being emitted to the outside. A pigmented iris-like structure(ILS) is present between the photocytes and lens cells, and a pigmented sheath protects the underlying tissues; this sheath is penetrated by large sinuses bringing blood inside the photophore's lumen(Ohshima, 1911).

To be efficiently used by the shark in its natural environment, this light emission must be precisely controlled. Counter-illumination, for example, is only efficient if the light produced has the same characteristics (intensity,wavelength and angular distribution) as found in the luminous background(Clarke, 1963; Denton et al., 1972; Denton et al., 1985; Harper and Case, 1999). Moreover, the shark should be able to switch off its luminescence when not needed.

Control mechanisms of shark luminescence are poorly understood and rarely investigated (Case and Strause,1978). The slow onset of luminescence(Ohshima, 1911) and the absence of photophore innervation (Johann,1899) in some investigated species, led Harvey to hypothesise a hormonal control of luminescence in sharks(Harvey, 1952). This control could, for example, target the chromatophores of the ILS whose action has already been suggested to allow a regulation of the amount of light emitted outside (Ohshima, 1911; Iwai, 1960). This could explain why Herring failed to induce luminescence in the luminescent cookie-cutter shark, Isistius brasiliensis, using acetylcholine as well as adrenaline (Herring and Morin,1978), which is currently considered the main neurotransmitter triggering luminescence in Osteichthyes(Baguet, 1975; Baguet and Marechal, 1978; Baguet and Christophe, 1983)(J.M., unpublished data).

Assuming the hypothesis of a hormonal control of E. spinax's luminescence through the ILS, three different hormones, implied in the regulation of the physiological colour change in Elasmobranchs (sharks and rays), are good candidates: prolactin (PRL), α-melanocyte-stimulating hormone (α-MSH) (which stimulate melanosome dispersion and therefore induce skin darkening after visual stimulation) and melatonin (MT) [which is thought to regulate (through changes in α-MSH release) skin colouration resulting from a non-visual perception of light level](Visconti et al., 1999; Gelsleichter, 2004).

In this work, the control mechanism of luminescence in the velvet belly lantern shark was pharmacologically investigated. Using isolated ventral skin patches of E. spinax, we performed a screening of test substances for neurotransmitters and hormones to identify those eliciting luminescence(extrinsic control) and then explored the intrinsic control pathways of these drugs.

Fifty-seven adult velvet belly lantern sharks, Etmopterus spinaxLinnaeus [30–52 cm total length (TL)] were captured during four field collections (February 2008, April 2008, December 2008 and February 2009) by long lines lowered in a deep area (depth≥200 m) of the Raunefjord, Norway(Table 1). Living sharks were brought to the High Technological Center of Bergen (`Hoyteknologisenteret i Bergen', HiB) or to the Espeland Marine Station and were housed in 1 m×1 m×1 m tanks placed in dark, cold (6°C) rooms. Following the rules of HiB for experimental fish care, all sharks were killed by a blow on their heads before experimentation took place.

The photophore preparations used in this work were skin patches of E. spinax specimens dissected out from their ventral luminous zone using a metal cap driller (diameter=0.55 cm); this preparation will be called `ventral skin patches' (Fig. 1A–C). After dissection, these ventral skin patches were placed in a shark saline [292 mmol l^–1 NaCl, 3.2 mmol l^–1 KCl, 5 mmol l^–1 CaCl₂, 0.6 MgSO₄, 1.6 mmol l^–1 Na₂SO₄,300 mmol l^–1 urea, 150 mmol l^–1trimethylamine N-oxide, 10 mmol l^–1 glucose, 6 mmol l^–1 NaHCO₃; total osmolarity: 1.080 mosmol; pH 7.7(Bernal et al., 2005)]following the method of Claes and Mallefet(Claes and Mallefet, 2008). The number of photophores present in a ventral skin patch of each animal was counted under binocular microscope (Leitz Diaplan, Oberkochen, Germany).

For screening of neural control, ventral skin patches were transferred to small Perspex chambers filled with 200 μl of the shark saline; the light-emitting surface areas of the patches were orientated to the photo-detector of a luminometer (Berthold FB12; Pforzheim, Germany) calibrated using a standard 470 nm light source (Beta light; Saunders Technology, Hayes,UK). Light emissions were recorded for 10 min using Berthold single kinetic mode (Sirius protocol manager v1.4). Data were collected every 0.2 s on a laptop computer to build original curves.

The usual slow kinetics of hormonal drugs permitted to conduct several long-lasting (about one hour) experiments simultaneously. For this purpose,preparations were placed in 100 μl saline-filled holes of one multiplate luminometer (Berthold MPL2/Orion; Pforzheim, Germany) calibrated with the same standard light source. According to the number of experiments, Berthold Simplicity software parameters were adjusted in order to obtain at least one point every minute. The luminescent responses were characterised using different parameters (Fig. 1D):the maximum intensity of light emission [L_max, in megaquanta per second (Mq s^–1)], the total quantity of light emitted during the experiment [L_tot, in teraquanta per hour (Tq h^–1)], and the time to reach L_max from the stimulation time [TL_max,in seconds (s)]. Light parameters were standardised by skin surface area (in cm^–2).

As a first step to trigger light emission, test substances(Table 2) were injected on ventral skin patches to reach the desired working dilution(10^–4 and 10^–6 for neural and hormonal testing, respectively; Table 2). In addition, potassium chloride (KCl, 200 mmol l^–1) was also injected in order to induce luminescence by nerve depolarisation. Ventral skin patches were injected with shark saline as control. For each drug, we compared the L_tot values obtained for treated and control skin patches from different adult sharks. After this screening, only drugs having significantly produced more (or less)light than the control were considered as good candidates for a more detailed analysis. Pharmacological analysis consisted in (i) dose–light response curves to determine the sensitivity of the tissue, (ii) use of antagonists(Table 2) to signal the presence of a specific receptor in the tissue investigated, and (iii) use of second messenger analogues and enzyme activators and inhibitors(Table 2) to determine the intrinsic control of the photophores in E. spinax. When inhibitors or antagonists were used, the ventral skin patches were first immersed for 20 min[except for the janus kinase 2 (JAK2) inhibitor, which necessitated a longer pre-treatment (Sandberg et al.,2005)] in a saline containing the inhibitor (or antagonist) at the desired final concentration and then received an injection of saline containing both the inhibitor (or antagonist) at the desired final concentration and the drug at a concentration that allows the final concentration to be the best working dilution. Due to large individual variability in the amplitude of light responses in E. spinax,dose–light response curves were given in relative units, and effect of antagonists, inhibitors and activators in percentage of control, i.e. the luminescence induced by the tested drug injected alone. All the experiments were conducted at room temperature (18°C).

All analyses [Student's t-tests and linear regressions] were performed with the software SAS/STAT (STAT Institute Inc., 1990, Cary, NC,USA). Two regression slopes were only considered to be significant if there was no overlapping between their 95% slope intervals. Each value was expressed with its standard error (means ± s.e.m.), and `N' equals the number of ventral skin patches used for a specific treatment (which actually corresponds to the number of individual sharks used).

The effect of classical neurotransmitters, KCl and hormonal drugs on ventral skin patches was investigated in adult specimens of E. spinax. Saline injection was used as control. Classical neurotransmitters and KCl evoked a minor light emission whose L_tot was, on average, not significantly (P>0.05) different from the L_tot obtained with a saline injection(Table 3). Injection of MT and PRL always induced light responses, which were significantly(P<0.05) higher than those of their control(Table 3). These light emissions typically lasted at least 30 min. Ventral skin patches pre-treated with α-MSH, however, produced significantly (P<0.05) less light than the saline control (Table 3).

Ventral skin patches stimulated by MT and PRL produced a glow quickly after injection but the time course was very different according to the hormone considered (Fig. 2A). The two hormones showed a highly significant (P<0.001) but different relationship between L_tot and L_max(Fig. 2B). The dose–light response curve of both hormones showed a similar pattern. The L_max progressively increased and attained its highest value at 10^–6 mol l^–1 to stay more or less constant at higher concentrations while the TL_maxdecreased when concentration increased(Fig. 3A,B). L_max in response to PRL and MT was also found to vary according to the time of year. In April, both hormonal drugs showed higher L_max than during the winter months and especially in February (Fig. 4A). However,the TL_max stayed constant over the year and corresponded to 40.68±2.95 min (N=44) for MT and 18.36±0.73 min(N=49) for PRL (Fig. 4B).

An additive effect for L_max and L_tot was obtained with a simultaneous injection of 10^–6 mol l^–1 MT and 10^–6mol l^–1 PRL (Table 4). The luminescence resulting from this injection, however,showed a TL_max equal to the TL_max of PRL and significantly (P<0.01) smaller than the TL_max of MT (Table 4).

To estimate the inhibitory effect of α-MSH, dose–light response curves were conducted with a simultaneous injection of either PRL or MT at their most effective stimulating concentration (10^–6 mol l^–1) and α-MSH at different concentrations. The patterns of the dose–light response curves are given in Fig. 5A,B. Depending on the experiment, the control was either PRL or MT injected alone at 10^–6 mol l^–1. At 10^–8 mol l^–1, no significant effect of α-MSH on the time course of luminescence was detected. At 10^–7 mol l^–1, however, α-MSH very significantly decreased the TL_max of MT (P<0.01) and the L_max of PRL (P<0.001) whereas all light parameters of both stimulating hormones were decreased by a 10^–6 mol l^–1 injection of α-MSH(Fig. 5A,B). While the TL_max and the L_max of MT-induced light response as well as the L_max of PRL-induced light response were significantly reduced (P<0.001) by α-MSH, this hormone had a lesser effect (P<0.05) on the TL_max of PRL-induced luminescence (Fig. 5A,B).

The treatment of ventral skin patches with 10^–4 mol l^–1 MT1/MT2 receptor antagonist luzindole and 2 10^–4 mol l^–1 MT2-specific receptor antagonist 4P-PDOT did not induce luminescence per se.

The MT (10^–6 mol l^–1)-induced luminescence was highly significantly (P<0.001) decreased by 2 10^–4 mol l^–1 and 2 10^–5 mol l^–1 4P-PDOT as well as by 2 10^–4 mol l^–1 luzindole (Fig. 6A,B). 4P-PDOT had, however, no effect on MT-induced luminescence(10^–6 mol l^–1) at lower concentrations(Fig. 6A).

The treatment of ventral skin patches with 10^–4 mol l^–1 dibutyryl-cAMP (db-cAMP), a membrane-permeable analogue of adenosine 3′,5′-cyclic monophosphate (cAMP),10^–4 mol l^–1 forskolin (FSK) and 10^–5 mol l^–1 of the two adenylyl cyclase inhibitors (SQ22,536 and MDL-12,330A) did not induce luminescence per se.

Even though 10^–4 mol l^–1 FSK significantly (P<0.001) decreased the MT-induced luminescence(10^–6 mol l^–1) from ventral skin patches,this luminescence was not affected by 10^–4 mol l^–1 db-cAMP, 10^–5 mol l^–1SQ22,536 and 10^–5 mol l^–1 MDL-12,330A(Fig. 7A).

The JAK2 inhibitor did not induce luminescence from ventral skin patches per se but significantly (P<0.001) decreased the PRL-induced luminescence (10^–6 mol l^–1) from ventral skin patches pre-treated during 5 h(Fig. 7B).

This work represents the first detailed study of control mechanisms of photophore luminescence in a cartilaginous fish, the velvet belly lantern shark (E. spinax), at both the extrinsic and intrinsic levels. It provides evidence for the first time of a hormonal control of luminescence in a fish. It also highlights the diversity of fish luminescence physiology and confirms that this capability evolved independently multiple times(Hastings, 1983; Herring, 1987). Finally, by working on isolated ventral skin patches, it offers a new valid technique to investigate luminescence control mechanisms of sharks whose photophores are extremely tiny (ca. 150 μm) compared with those of bony fishes.

The different classical neurotransmitters used in this study failed to induce luminescence from isolated photophores of E. spinax. This and the absence of a reaction to KCl strongly suggest that, contrary to luminescent bony fishes, shark's photophores are not under direct nervous control. This is in agreement with the observation of Johann(Johann, 1899), who was not able to find any innervation in the photophores of the velvet belly lantern shark.

MT and PRL, however, show a dose-dependent stimulatory effect on light emission from photophores of isolated ventral skin patches of E. spinax. The comparison of the dose–light response curves show similar sensitivity of the photophores to both hormones between 10^–5 mol l^–1 and 10^–8 mol l^–1. The time course of luminescence triggered by the two hormones is, however, different but reproducible: while response to MT injection is characterised by a slow increase of light intensity (which can last several hours; J.M.C., unpublished data), PRL induced a relatively quicker response, which lasts one hour at maximum. This, with the additive effect of a simultaneous injection of MT and PRL, suggests that these two hormones use a different pathway to induce light emission in E. spinax's photophores. In addition to these hormonal stimulating effects,an inhibitory effect of the α-MSH on photophore luminescence was also shown. These combined results strongly suggest that the light emission of lantern shark is controlled by different hormones. This would explain the slow onset of luminescence observed in this species(Ohshima, 1911). Assuming luminescence to be hormonally controlled, the hormones certainly use the large blood sinuses that go through the pigmented sheath of shark's photophores to act on a specific target inside these luminous organs.

In vertebrates, the effects of MT are mediated through high and low affinity receptors. While low affinity receptors (MT3) correspond to a cytosolic enzyme implicated in detoxification processes (the `quinone reductase-2'), high affinity receptors are transmembrane receptors(Vanecek, 1998; Barrett et al., 2003; Boutin et al., 2005). In fishes, three different high affinity receptors have been found: MT1 (=Mel1a),MT2 (=Mel1b) and Mel1c (Barrett et al.,2003). These high affinity MT receptors can be coupled with several intrinsic pathways, including adenylyl cyclase inhibition (which induces a decrease in intracellular cAMP) or phospholipase C (PLC) activation,which finally modulate intracellular Ca²⁺, or even viaguanylyl cyclase inhibition (Vanecek,1998; Barrett et al.,2003).

In this work, MT1/MT2 receptor antagonist luzindole (10^–4mol l^–1) and MT2-specific receptor antagonist 4P-PDOT (2 10^–4 mol l^–1) both inhibit the action of MT,with 4P-PDOT showing a dose-dependent (from 2 10^–7 mol l^–1 to 2 10^–4 mol l^–1)inhibitory effect on MT-induced luminescence. These results are strongly in favour of the presence of melatoninergic MT2 receptors in the ventral photophores of E. spinax. Moreover, the strong inhibitory effect of FSK (10^–4 mol l^–1), which activates the adenylyl cyclase and increases intracellular cAMP concentration, on MT-induced luminescence indicates that the shark MT2 receptor is negatively coupled to the cAMP pathway. The absence of effects of adenylyl cyclase inhibitors SQ22,536 (10^–5 mol l^–1) and MDL-12,330A(10^–5 mol l^–1) on light emission triggered by MT are coherent if we consider that the intensity of light emitted after MT injection correspond to the maximum physiological response of the tissue. The contradictory absence of effects with the cAMP analogue dB-cAMP(10^–4 mol l^–1) might reveal the presence of a high phosphodiesterase activity inside the photophore that prevents cAMP to act. The lack of effect of db-cAMP has already been observed in other animal preparations and could also be explained by solubility or permeability problems of this substance in preparations(Vanderlinden et al.,2004).

Due to its numerous actions in many different biological mechanisms, PRL is considered as the most versatile hormone in vertebrates(Bole-Feysot et al., 1998; Manson, 2002). Effects of PRL are mediated by the PRL receptor (PRLR), a receptor of the class 1 cytokine receptor superfamily (Bole-Feysot et al.,1998). The PRLR is known to be present in several bony fish species but, to our knowledge, has never been reported in a cartilaginous species (Sandra et al., 1995; Tse et al., 2000; Le Rouzic et al., 2001; Santos et al., 2001; Higashimoto et al., 2001). The binding of PRL to the PRLR induces a homodimerisation of this receptor and activates a tyrosine kinase called JAK2. When activated, JAK2 phosphorylates different proteins and initiates different downstream cascades, including JAK/Stat (signal transducers and activators of transcription) and MAPK(mitogen-activated protein kinase) pathways(Bole-Feysot et al., 1998; Freeman et al., 2000). The strong inhibition of PRL luminescence observed after pre-treatment with the inhibitor of JAK2 suggests strongly that the luminescence triggered by PRL is mediated by the action of JAK2, which is also in favour of the presence of PRLR in the shark photogenic organs.

The survival of a luminescent organism depends on its capability to control its light emission. In multicellular animals, this control occurs at two levels: (i) the intrinsic control level, which focuses on intracellular events and on the photogenic reaction, and (ii) the extrinsic control level whose action is extracellular. Luminescent bony fishes, especially, show a high diversity of extrinsic control mechanisms depending on the nature of their photogenic organs. While luminescence of fishes harbouring symbiotic bacteria in their photophores can be controlled either mechanically via dark shutters and chromatophores or physiologically by controlling the blood supply to these organs, luminescence of fishes with photophores containing endogenous systems are neurally controlled and often show accessory structures, including optical filters and reflectors (Case and Strause, 1978; Herring and Morin, 1978; Denton et al.,1985).

In bony fishes (Osteichthyes), luminescence is triggered by adrenaline (and sometimes noradrenaline) while serotonin (5-HT) is known to inhibit light emission in the midshipman fish Porichthys notatus(Table 5). In addition, a modulatory effect of nitric oxide (NO) on adrenaline-stimulated luminescence has been detected in the hatchetfish Argyropelecus hemigymnus(Table 5). Direct neural control of luminescence generally allows precise and rapid on/off light switching, which is useful in predator avoidance and in intraspecific communication.

In the velvet belly lantern shark, photophore luminescence is controlled by hormones, which are also involved in the skin pigmentation control of Elasmobranchs via the action of melanophores(Visconti et al., 1999; Gelsleichter, 2004). This strongly suggests that, in this shark, light emission from photocytes is, at least partly, controlled via the action of the ILS (formed by pigmented cells) as it has been sometimes suggested in different lantern shark species (Ohshima, 1911; Iwai, 1960). This hormonal control system of E. spinax's luminescence explains the slow kinetic of light emission, especially in the case of MT luminescence. Far from being unsuited to this shark's ecology, this slow luminescence kinetic could be very convenient for vertical migrations during which the ambient light intensity may increase/decrease gradually. Lantern sharks like other deep-sea shark species possess a `pineal window' and it has already been suggested that visual information collected by the pineal gland help these fishes to accomplish their vertical migrations(Clark and Kristof, 1990). Assuming that the visual information from the pineal gland is mediated by a release of MT in the blood (Underwood,1989; Arendt, 1997)and because MT triggers light emission in a dose-dependent manner in E. spinax, this hormone could make a direct connection between ambient light and luminescence intensity in this shark. This could allow E. spinaxto match the intensity of downwelling light with its luminescence, which is particularly convenient for camouflage by counter-illumination, a function already suggested for this shark's luminescence(Claes and Mallefet, 2008; Claes and Mallefet, 2009). Despite the lack of a rapid fine tuning of luminescence by neural control and the absence of ocular photophore, which are used as a luminescence reference standard by counter-illuminating midwater bony fishes(Nicol, 1962; Herring, 1977; Denton et al., 1985; Warrant and Locket, 2002), the hormonal long-lasting luminescence associated to pineal gland might therefore represent a new way to counter-illuminate efficiently. Nevertheless more experiments are needed in order to properly document counter-illumination in E. spinax. In Elasmobranchs, PRL and α-MSH are though to be released and induce skin darkening after a visual stimulation(Visconti et al., 1999; Gelsleichter, 2004). Similarly, the action of these hormones on shark luminescence may be dependent on visual perception and therefore to be involved in behaviours necessitating quicker on/off light switching than responses to MT can support, such as in schooling, mating or predator escape.

In this study we show that PRL and MT trigger light emission in E. spinax. In addition we report a possible seasonal variation in the response to these hormones, with a higher response in April than in December and February. A similar seasonal variation in luminescence intensity has been suggested for the krill Meganyctiphanes norvegica and is thought to be linked to higher ambient light levels in the summer period(Krönström et al.,2007). Seasonal variation in MT-induced luminescence is not surprising because plasma MT content vary with day-length and the season. Using this molecule as a biological internal clock, fishes can therefore anticipate changes in their environment to which they have to synchronise their biological rhythms (Falcón et al., 2007). Because MT is known to affect PRL release in some fish species (Falcón et al.,2003) (Falcón et al.,2007), seasonal variation in PRL-induced luminescence may be directly dependent of seasonal variation in plasma MT content.

LIST OF ABBREVIATIONS

 
• α-MSH

  α-melanocyte-stimulating hormone

 
• cAMP

  cyclic adenosine monophosphate

 
• db-cAMP

  dibutyryl cyclic adenosine monophosphate

 
• FSK

  forskolin

 
• ILS

  iris-like structure

 
• JAK2

  janus kinase 2

 
• L

  light emission

 
• L_max

  maximum intensity of light emission

 
• L_tot

  total light emission

 
• MT

  melatonin

 
• MT1

  melatonin receptor 1

 
• MT2

  melatonin receptor 2

 
• MT3

  melatonin receptor 3

 
• PRL

  prolactin

 
• PRLR

  prolactin receptor

 
• TL

  total length

 
• TL_max

  time between stimulation and maximum intensity light emission