Spectral Diversity and Regulation of Coral Fluorescence in a Mesophotic Reef Habitat in the Red Sea

The phenomenon of coral fluorescence in mesophotic reefs, although well described for shallow waters, remains largely unstudied. We found that representatives of many scleractinian species are brightly fluorescent at depths of 50–60 m at the Interuniversity Institute for Marine Sciences (IUI) reef in Eilat, Israel. Some of these fluorescent species have distribution maxima at mesophotic depths (40–100 m). Several individuals from these depths displayed yellow or orange-red fluorescence, the latter being essentially absent in corals from the shallowest parts of this reef. We demonstrate experimentally that in some cases the production of fluorescent pigments is independent of the exposure to light; while in others, the fluorescence signature is altered or lost when the animals are kept in darkness. Furthermore, we show that green-to-red photoconversion of fluorescent pigments mediated by short-wavelength light can occur also at depths where ultraviolet wavelengths are absent from the underwater light field. Intraspecific colour polymorphisms regarding the colour of the tissue fluorescence, common among shallow water corals, were also observed for mesophotic species. Our results suggest that fluorescent pigments in mesophotic reefs fulfil a distinct biological function and offer promising application potential for coral-reef monitoring and biomedical imaging.


Introduction
The obligatory symbiotic association with photosynthetic algae of the genus Symbiodinium (zooxanthellae) renders the major builders of shallow, warm-water reef scleractinian corals, Up to now, the role of fluorescence in corals in low-light habitats at depths >40m has remained largely unstudied due to the difficulties in accessing these habitats using standard SCUBA diving. Recent advantages in technical diving approaches [34,35] and fluorescence detection techniques [36][37][38] facilitate the exploration of coral fluorescence in mesophotic corals. Here, we provide first insights in the prevalence, spectral properties and regulation of coral fluorescence in a mesophotic coral community in the Red Sea in Eilat, Israel.
Representative specimens were then collected and transferred to the IUI laboratory in separate Ziploc bags and stored in black plastic containers for further analyses. Samples of S. pistillata were collected from a depth range of 1-65 m. We sampled horizontal branches with equal exposure to down-welling light on the upper part and bottom reflectance on the lower part of the branches. The fragments were kept in the IUI flow-through aquarium system inside a dark room until they were photographed and fluorescence spectra were recorded.

Characterisation of the underwater light field
Downwelling spectral irradiance measurements were conducted using a profiling reflectance radiometer PRR800 (Biospherical Instruments Inc., San Diego, USA) with 19 channel detection (300-900nm) and an integrated Photosynthetically Active Radiation (PAR) sensor. The instrument was deployed at midday (11:00-13:00) on the 29 th October 2014 from a boat, using the free fall technique in order to avoid shading or reflectance by the vessel and to keep the light sensor in a vertical position. The instrument was lowered from the boat while its distance from sea bed was kept constant (~1 m) comparing the boat sonar data to the depth sensor data of the instrument. Data were analysed using PROFILER software (Biospherical Instruments Inc., San Diego, USA). Depth distribution and fluorescence of prevalent coral species Fluorescence in shallow water corals (depths between 0.5 and 3 m) was surveyed at night using a blue-light torch and yellow-barrier filter goggles (Nightsea). The reefs in front of the IUI were assessed over a 500 m transect.
Additionally, coral fluorescence was explored during technical SCUBA diving along one belt transect of 50 m 2 at depths of 2, 5, 10, 15, 40, 50 and 65 m using the underwater fluorescence imaging system for excitation and fluorescence detection. Four localities were explored (Dekel Beach, Oil Jetty, Nature Reserve and IUI) in the Gulf of Eilat/Aqaba during 2010 to 2015. All localities had a fringing reef structure and mesophotic submerged bunks down to a maximal depth of 72 m. Coral identification was achieved by (1) using keys provided by Veron [39,40], Scheer & Pillai [41], Wallace [42] and Hoeksema [43], (2) comparison with the coral collection at the Tel-Aviv University Steinhardt Museum of Natural History and National Research Center, and (3) evaluation by expert coral taxonomists.

Coral culture
Five to ten fragments from each colony of the studied E. paradivisa colonies were kept in "light" and "dark" conditions in the flow-through aquarium system at the IUI for one year. Light levels experienced by the "light"-treated fragments were determined with a Li-Cor LI-193SA 4 pi probe attached to a Li-1400 data logger (Li-Cor) at noon on several days over the different seasons. The upper parts of the corals were exposed to maximum fluxes of 10-20 μmol photons m -2 s -1 . The "dark"-treated fragments were kept in complete darkness. At the end of the experiment, the fluorescence of the corals from the different treatments was documented by spectroscopic measurements and photography (see below).
Replicate fragments of a red colour morph of Echinophyllia sp. were cultured in the experimental mesocosm of the Coral Reef Laboratory at the University of Southampton [44]. To test the effect of short wavelengths on the coral fluorescence, the five replicate corals were exposed to light with different spectral compositions but with identical photon fluxes of~20 μmol photons m -2 s -1 for six months. One group was kept under full spectrum white light (containing violet wavelengths) provided by a metal halide lamp fitted with an Aqualine 10000 burner (13000 K; AquaMedic). The other group was incubated under cyan light (~480nm, lacking violet wavelengths <430 nm) emitted from Aqua Ray Marine Blue LEDs (Tropical Marine Centre).

Fluorescence photography in the laboratory
Coral fluorescence was photographed through a yellow long-pass filter (Nightsea) upon excitation of the tissue fluorescence with a~450-nm light source (Nightsea). Standardised conditions (excitation light intensity and exposure times) were established to produce images that allow a comparison of tissue fluorescence intensity.

Fluorescence spectroscopy
Fluorescence was excited with blue light torches (440-460 nm) and a yellow long-pass filter (Nightsea) to block out the excitation light.
Fluorescence emission spectra were measured with a USB2000 spectrometer and analysed with the SpectraSuite-Spectrometer Operating Software (Ocean Optics Inc.). The background was subtracted and the spectra were smoothed by a running median smoother in SigmaPlot 12 (Systat Software Inc.).
Fluorescence emission spectra of the aquarium-cultured specimens of Echinophyllia sp. were recorded with a fibre optic probe coupled to a Cary Eclipse fluorescence spectrometer (Varian). A 0.5 cm spacer attached to the probe was used to measure the fluorescence of comparable areas and enable a quantitative comparison of tissue fluorescence between the experimental samples from the different light treatments.

Photoconversion under the fluorescence microscope
Following a six-month exposure to cyan light lacking violet wavelengths, the tissue fluorescence of Echinophyllia sp. was imaged with an MZ10 Olympus fluorescence microscope equipped with a long-pass GFP filter set (AHF). Green-to-red photoconversion of the fluorescent pigments was induced by local exposure to~380 nm light for 10 minutes using a UV filter set (AHF). The photoconversion was subsequently imaged using the above mentioned GFP filter set.

Results
Green fluorescence in shallow water corals was found in most of the colonies surveyed in the reefs in front of the IUI. Notably, essentially no orange-red coral fluorescence was observed with the exception of two orange-red fluorescent individuals, one Dipsastraea sp. coral and one corallimorpharian (Discosoma sp.), both of which were growing in crevices underneath massive corals.
Visual inspection of corals within the belt transects in depths between 40-65 m revealed that green and orange-red fluorescence excited by the natural, blue-dominated environmental light was widespread and often bright enough to be detected by the naked eye or by a standard underwater camera. The fluorescence imaging system proved helpful in improving the contrast between the fluorescence signal and the background (Fig 2).

Spectral properties
We collected corals from mesophotic depths with representative and visually significant fluorescence in order to gain an overview of the range of the fluorescent pigments present at these depths. After transfer to the laboratory, the fluorescence emission of the living corals was immediately recorded (Figs 3-5, Table 1). Green fluorescence with maxima in the range of 512-522 nm could be found in essentially all examined species. Yellow fluorescence with emission maxima between 522 and 558 nm was detected in more than a third of the species. Emission maxima in the orange-red range from 578 to 584 were recorded in six species. Most of the fluorescence spectra were dominated by a single, narrow peak characteristic of GFP-like proteins [18]. The spectroscopic data of some individuals of E. paradivisa and G. somaliensis showed variable contributions of green and yellow fluorescent pigments with overlapping emission spectra.
In E. aspera, the typical signature of green-to-red photoconvertible proteins was detected [17,32,48]. The emission spectrum shows a local maximum at~514 nm which can be attributed to the green, unconverted chromophore, and a second peak at~578 nm with the characteristically pronounced vibronic sideband at~630 nm, emitted by the red, photoconverted chromophore (Fig 3).
All species showed detectable chlorophyll fluorescence with an emission maximum around 680 nm. Representative chlorophyll emission spectra in the corals G. somaliensis and E. paradivisa are shown in Figs 4 and 5, respectively.

Fluorescence patterns and colour polymorphisms
Many species collected from depths >50 m were characterised by a uniform distribution of fluorescence over the upper colony surface: e.g. O. egyptensis, A. ocellata, G. minor, P. granulosa and E. aspera (Fig 3), while G. somaliensis was represented by three distinct colour morphs which could be easily distinguished by their green, yellow and orange-red fluorescence (Fig 4). The colour morphs also showed differences in the tissue-specific accumulation of the pigments. Whereas in the green and yellow morphs the fluorescence was concentrated in the polyps, the fluorescent pigments in the red morph were distributed over the whole colony. In E. paradivisa too, three colour morphs (green, yellow and red) could be distinguished based on the spectral signature and the localisation of tissue fluorescence (Fig 5). In the green and yellow morphs, fluorescence was concentrated in the tips of the tentacles. In the red morph, in contrast, green light emission was evenly distributed over the length of the tentacles and absent from the tentacle tips, which showed instead a dark red fluorescence emitted by the chlorophyll of the zooxanthellae.
Deep-water individuals of the branching species S. pistillata and A. squarrosa showed a flattened branch morphology. In both cases, the green fluorescence was restricted to the upper part of the branches (Fig 6). Light-dependent accumulation of fluorescent pigments We incubated replicate nubbins of the different colour morphs of E. paradivisa for 12 months under low-intensity full spectrum daylight and in complete darkness. The photosynthetic  pigments of the zooxanthellae were lost completely from the specimens in the dark treatment as evident from the bleached appearance of the corals and lack of the chlorophyll emission peak at~680 nm (Fig 5). The yellow fluorescence too (emission maximum at~545 nm) vanished completely. However, under these conditions, the intensity of the green tentacle fluorescence remained essentially unaltered.
Accumulation of fluorescent pigments was restricted to the upper side of outer branch parts in S. pistillata and A. squarrosa collected from 60 m depth (Fig 6). Long term experimental exposure of Echinophyllia sp. to low intensity of~480 nm-light, lacking wavelengths <430 nm resulted in a greening of the otherwise orange-red fluorescent corals (Fig 7). Short-term local   [45] e [46] f [47] doi:10.1371/journal.pone.0128697.t001  exposure with light around~380 nm resulted in a green-to-red photoconversion of tissue fluorescence.

Spectral properties
Our assessment of coral fluorescence in Eilat reefs in shallow water (depths between 0.5 and 3 m) encountered essentially no yellow (rarely found also in other reef regions [16]) or red fluorescent specimens. In contrast, green, yellow and red fluorescent pigments were found frequently in corals from mesophotic depths 50m. Fluorescence at depths >40m is readily visible by eye and can be documented photographically even without optical filter systems owing to the fact that at these depths longer wavelengths are depleted relative to the fluorescence-exciting shorter wavelengths (Fig 1). Thereby, the contrast of the emitted light in the yellow-green and orange-red spectral region is enhanced. The specific signature of the emission spectrum of E. aspera with emission maxima at 514 nm and 578 nm identified its dominant pigment as a photoconvertible protein of the class of green-to-red photoconvertible proteins [32,48]. The chromophore of this specific type of GFPlike protein requires exposure to light in the spectral range between 360 to 430 nm (maximal efficiency at 390 nm) in order to undergo the irreversible photochemical reaction that changes the chromophore from a green(514 nm) to a red (578 nm) emitting form [48]. A GFP-like protein with an amino acid sequence with high similarity to the green-to-red photoconvertible EosFP has been isolated from Echinophyllia echinata [16].
Several previously studied fluorescent proteins show a reversible increase of their fluorescence intensity and small shifts in the position of their emission maxima with increasing pressure in the range of 0.1-500 MPa [49,50]. The fluorescence intensity of the red fluorescence protein dsRed, for example, was maximally increased by~8% at a pressure of 200 MPa as compared to normal atmospheric levels [49]. Since the coral pigments in mesophotic depths are exposed to increased pressure of only~0.5-1.1 MPa, it is unlikely that their spectral properties in situ deviate substantially from those determined in our measurements at atmospheric pressure Our analysis of 16 species of corals (Table 1) from mesophotic depths did not detect a significant contribution of cyan fluorescence with emission maxima in the range of 470-500 nm to the overall fluorescence of these animals. In contrast, cyan fluorescent proteins are frequently encountered in shallow water corals and can dominate the quality of light emission of several species [16,19]. A pronounced cyan fluorescent maximum was detected in the strain of Indo-Pacific Echinophyllia sp. that was cultured for the long-term laboratory experiments and in colour morphs of M. cavernosa [30,32], suggesting that the absence of CFP might not be a uniform feature of corals with a preference for less light-exposed habitats.

Fluorescence patterns and colour polymorphism
Shallow-water corals display a broad range of fluorescence patterns that result from the tissuespecific expression of GFP-like proteins [51]. Many of the corals from depths >50 m in this work showed a rather homogenous distribution of light emission over the whole colony. Exceptions were found among individuals of E. paradivisa, which showed a variable localisation of the fluorescent pigments in the tentacles; and among representatives of G. somaliensis, in which the fluorescence was either distributed evenly over the colony or restricted to the polyps. The latter two species also showed a pronounced colour polymorphism related to variations in the expression of the dominant fluorescent pigment, a phenomenon that is common among anthozoans [30,32,[52][53][54]. As demonstrated for the coral Acropora millepora, these colour polymorphisms can be the result of the pigment-encoding genes being present in multiple copies [55]. Depending on how many genes are active, the corals will become more or less colourful. In shallow waters, where the GFP-like proteins can fulfil a photoprotective function, this genetic framework can enable corals to invest either in expensive high-level pigmentation, offering benefits under light stress, or to rely on low tissue pigment concentrations and use the conserved resources for other purposes, a preferable strategy in low light-exposed environments.

Regulation of pigment production
Previous studies showed that the expression of GFP-like proteins in shallow water corals is often upregulated at the transcriptional level by the light intensity, specifically by blue light [19]. At low-light intensities, the tissue fluorescence decreases, usually to almost undetectable levels [19,24].
In contrast, corals such as L. hemprichii or M. cavernosa, which frequently occupy less lightexposed habitats showed a constitutive expression of their fluorescent pigments [31,32].
We incubated Echinophyllia sp. and E. paradivisa under reduced light intensity to examine whether the constitutive expression of fluorescent pigments was a common phenomenon in corals from low-light environments.
Long-term experimental exposure of Echinophyllia sp. to low intensity of cyan light (~480 nm), which lacked wavelengths <430 nm resulted in a colour change from orange-red to green. This greening could be reversed by a brief, local exposure to short wavelength light (~380 nm). A comparable response was previously observed for the red fluorescent protein mcavRFP in the tissue of M. cavernosa, suggesting that the exposure to short-wavelength light is essential for these corals to develop their dominant red tissue fluorescence [31]. Since the production of the pigment in its native green state is sustained at low light levels [31], the relevant genes are most likely constitutively expressed and light is only required for the post-translational modification of the pigments. The E. aspera specimens analysed in the present study showed bright orange-red fluorescence in situ at a depth of >50 m, indicating that even at this depth, the underwater light field contains sufficient amounts of photons in the short-wavelength range to induce efficient photoconversion.
Variable responses were observed after prolonged incubation of the different colour morphs of E. paradivisa in darkness. All the morphs completely lost the chlorophyll fluorescence of the zooxanthellae. Accordingly, the prevailing red emission of the red morph was lost. In contrast, in all the morphs, the green fluorescence with an emission maximum at~515 nm persisted essentially unaltered in corals cultured in the dark. This indicates that the relevant genes are constitutively expressed.
Interestingly, the 545 nm-peak of the yellow fluorescent pigment (most pronounced in the yellow morph) was not detected in the spectra measured after dark incubation. This may indicate that this pigment is either a yellow fluorescent protein with a light-dependent expression or that it represents a derivative of the green fluorescent emitter that is formed in a light-dependent reaction. Further studies will elucidate the putative mechanism responsible for this observation.
The restricted accumulation of green fluorescent pigments in the upper-side of S. pistillata and A. squarrosa branches collected from 60 m depth suggests an influence of the irradiating light on the regulation of pigment production. This localised accumulation in the most lightexposed branch parts has been frequently observed for light-regulated GFP-like proteins in shallow water corals, in particular among acroporids [24,29].

Biological function of fluorescence at mesophotic depths
Recent studies have demonstrated the ability of both FPs and CPs to fulfil a photoprotective function for the zooxanthellae in some shallow-water corals by screening photosyntheticallyactive radiation (PAR) [24,29]. In line with their function, these photoprotective pigments are tightly controlled by the light environment. The need for photoprotection can be ruled out for corals from mesophotic reefs and may explain the absence of dominant cyan fluorescence at greater depths, since the absorption properties of these pigments appear to be well suited for screening of algal chlorophyll [19]. Nevertheless, species such as S. pistillata with a broad depth distribution ranging from very shallow water to mesophotic depths (Table 1), may still reveal the light-mediated regulation mechanisms associated with the photoprotective function that fluorescence fulfils in the shallow water population.
We have found that a number of brightly fluorescent species, including E. aspera and E. paradivisa, have their maximal abundance at depths deeper than 30 m. The production of individual pigment molecules might be considered comparably cheap [31], since only a single gene is required to produce the functional FP and their turnover is slow, presumably due to the stable beta-can fold. However, the high tissue concentrations of up to 7-14% of the total soluble protein [31,32] can only be reached by the simultaneous expression of multiple copies of the same gene [24]. Hence, maintaining high tissue fluorescence is energetically costly, and it is therefore very likely that the fluorescent pigments in mesophotic corals have a particular biological function that differs from photoprotection. This assumption is supported by the lightindependent, constitutive expression of their FP genes, which is atypical for the photoprotective pigments from shallow water corals [31,32]. Furthermore, in contrast to shallow water corals, yellow and orange-red fluorescence were more abundant in the species with a preference for greater depths (Table 1). Also Leptoseris spp. from Hawaii showed a shift of the dominant FPs from shorter (cyan) in shallow waters to longer (green) wavelengths in mesophotic depths [15]. The trend that fluorescence in mesophotic corals tends to be red-shifted compared to the shallow water representatives further points to a distinct biological function in corals from deeper habitats. Alternative functions of FPs, which have been discussed, include modulation of the activity of regulatory photosensors analogous to phytochromes and cryptochromes of higher plants [56], links to visual ecology of the reef fishes [57,58] and PAR enhancement [12]. Future experimental studies are required to confirm the function of fluorescence in mesophotic corals.

Coral reef monitoring
Coral fluorescence can be detected with a high signal-to-background ratio and has been suggested as an indicator of coral health and live coral cover in reef-monitoring approaches [19,20,[59][60][61]. Indeed, the ability to detect coral recruits in situ is increased by screening for their fluorescence [62,63]. The use of advanced wide field-of-view fluorescence imaging shows the potential of coral fluorescence to be used in semi-automated image analysis for habitat mapping [37]. Since coral fluorescence in mesophotic reefs appears to be widespread, bright and spectrally diverse, fluorescence imaging represents a promising approach for the monitoring of mesophotic reefs.

Bio-prospecting
Fluorescent proteins of the GFP-family have revolutionised biomedical research, as in vivo labels of proteins or markers of gene expression [64,65]. The range of differently coloured FPs isolated from marine invertebrates has facilitated multicolour imaging of biological processes in live cells, while photoconvertible proteins have enabled advanced tracking of dynamic processes in living cells and organisms, such as protein movement [66,67]. Furthermore, they are also essential tools for super-resolution microscopy [68]. Red fluorescent proteins are particularly desirable for imaging application, since their detection is facilitated by the better penetration of cells and tissues by long-wavelength light and reduced cellular autofluorescence in the red emission range [69]. The high number of long-wave emitting FPs that we have encountered in the present study suggests that mesophotic reefs may hold novel marker proteins useful for biomedical research applications.