Pushing the limits of photoreception in twilight conditions: The rod-like cone retina of the deep-sea pearlsides

Abstract

Most vertebrates have a duplex retina comprising two photoreceptor types, rods for dim-light (scotopic) vision and cones for bright-light (photopic) and color vision. However, deep-sea fishes are only active in dim-light conditions; hence, most species have lost their cones in favor of a simplex retina composed exclusively of rods. Although the pearlsides, Maurolicus spp., have such a pure rod retina, their behavior is at odds with this simplex visual system. Contrary to other deep-sea fishes, pearlsides are mostly active during dusk and dawn close to the surface, where light levels are intermediate (twilight or mesopic) and require the use of both rod and cone photoreceptors. This study elucidates this paradox by demonstrating that the pearlside retina does not have rod photoreceptors only; instead, it is composed almost exclusively of transmuted cone photoreceptors. These transmuted cells combine the morphological characteristics of a rod photoreceptor with a cone opsin and a cone phototransduction cascade to form a unique photoreceptor type, a rod-like cone, specifically tuned to the light conditions of the pearlsides' habitat (blue-shifted light at mesopic intensities). Combining properties of both rods and cones into a single cell type, instead of using two photoreceptor types that do not function at their full potential under mesopic conditions, is likely to be the most efficient and economical solution to optimize visual performance. These results challenge the standing paradigm of the function and evolution of the vertebrate duplex retina and emphasize the need for a more comprehensive evaluation of visual systems in general.

Fig. 1. Vertebrate opsin gene phylogeny and pearlside opsin gene expression.

(A) The pearlside retinal transcriptomes (n = 5 per species) contained three opsin genes: one rod opsin (rh1) and two rhodopsin-like 2 (rh2) cone opsins. Black spheres indicate Bayesian posterior probabilities >0.8. Asterisk indicates that this opsin gene class was not present in the pearlside transcriptome. lws, long-wavelength sensitive; sws1 and sws2, short-wavelength sensitive 1 and 2; va, vertebrate ancient opsin (outgroup). A detailed phylogeny and GenBank accession numbers are shown in fig. S1. Pearlside-specific accession numbers are given in table S4. (B) The per-species mean of the proportional opsin gene expression shows the almost exclusive use of cone opsins in pearlside vision.

Fig. 2. Vertebrate phylogenies of phototransduction cascade genes and phototransduction cascade gene expression.

(A and B) Vertebrate Gα transducin (A) and vertebrate arrestin (B) gene phylogenies. Black spheres indicate Bayesian posterior probabilities >0.8. Detailed phylogenies and GenBank accession numbers are shown in figs. S3 and S4. Pearlside-specific accession numbers are given in table S4. gnat2, G protein subunit α transducin 2; gnat1, G protein subunit α transducin 1; “taste,” G protein subunit α transducin 3; arrb2, arrestin β2; arr3, arrestin 3; sag, s-antigen arrestin [saga is present in the outer segment, and sagb is present in the synapses (16)]; arrb1, arrestin β1. (C and D) The per-species mean of the proportional transducin (C) and arrestin (D) gene expression shows the almost exclusive use of cone transducin (gnat2) and cone arrestin (arr3) in pearlside vision.

Fig. 3. Absorbance spectra of photopigments expressed in two representative Maurolicus spp.

(A) Experimentally determined absorbance spectra of M. muelleri and human (Homo sapiens; control) rod opsin photopigments, reconstituted with 11-cis-retinal. For all pigments, representative dark (filled circles) and light-bleached (open circles) spectra are shown, with difference spectra (open diamonds) that have been fitted with a Govardovskii rhodopsin/vitamin A₁ template (63) (black line) in the inset to determine the λ_max. (B) Predicted spectral sensitivities of rh1 and the two rh2 opsins found in the two pearlside species, M. muelleri (black) and M. mucronatus (gray).

Fig. 4. Opsin R-ISH and the distribution of the two photoreceptor types in the retina of M. muelleri.

(A and C) Retinal cryosections showing the expression of rh1 (A) and rh2 (C) opsin genes in cryosections. Arrowheads highlight labeled cells. Note that both genes are expressed in rod-like photoreceptor cells. (B) Distribution of rh1 photoreceptors labeled with anti-rhodopsin antibodies. Each dot represents one labeled photoreceptor. Black arrows indicate the orientation of the retina. N, nasal; V, ventral. (D) Topographic map of rh2 photoreceptor densities (cells × 10⁴ mm⁻²). Percentages indicate the proportion of each cell type. Scale bars, 50 μm (A and C) and 1 mm (B and D).

Fig. 5. Morphology of the two photoreceptor types in M. muelleri.

(A) Schematic of the rod-like cone (yellow; left) and rod (blue; right) drawn from the 3D reconstruction using 3View. OS, outer segment; IS, inner segment; SE, synaptic ending; Di, discs; Mt, mitochondria; ILM, inner limiting membrane; Nc, nucleus. Note the displaced nucleus and synaptic ending in the rod. (B) Immunofluorescence labeling of transverse retinal cryosections. Rod outer segments were labeled with anti-rhodopsin antibodies (red), inner segments with NeuN (white), cell nuclei with 4′,6-diamidino-2-phenylindole (DAPI) (blue), and synaptic connections with synapsin (green). Note that NeuN does not usually stain photoreceptor inner segments, but in M. muelleri, the inner segments of the rods were strongly labeled compared to the rod-like cones. PRL, photoreceptor layer; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; c, rod-like cone; r, rod; cse, rod-like cone synaptic ending; rce, rod synaptic ending. Scale bar, 5 μm. (C to F) TEM of transverse retinal sections showing the two photoreceptor types (C), the polysynaptic ending of the rod-like cone (D), the oligosynaptic ending of the rod (E), and the sealed discs and incisures of the outer segments (F). The white arrowheads in (D) and (E) show the synaptic ribbons, and the black arrowheads in (F) show the incisures present in the rod-like cone. Scale bars, 2 μm (C) and 1 μm (D to F).

Fig. 6. Ambient light environment and pearlside visual capabilities.

(A) Light levels associated with different photoreceptor functionalities. M. muelleri is only active during mesopic and low-level photopic light intensities (39). R, rod; C, cone. Scotopic vision is defined by the use of rods. Mesopic vision is defined by the use of both rods and cones limited by cone threshold and rod saturation. Photopic vision is defined by the use of cones and ends when light intensities start to be damaging (75). Environmental light sources (from left to right) are as follows: starlight, full moon, civil twilight, sunset/sunrise, and sunlight (76). Figure partially redrawn from Hood and Finkelstein (75). (B) Spectral sensitivity curves of the pearlside M. muelleri rh2 (a, black line; predicted λ_max = 441 nm; fig. S7), the deep-sea myctophid Symbolophorus evermanni rh1 [b, dark gray line; λ_max = 476 (23)], and the nocturnal squirrelfish Neoniphon sammara rh1 [c, light gray line; λ_max = 502 nm (77)] along with the relative downwelling vector irradiance spectra (courtesy of S. Johnsen) of their respective light environments: twilight (−6.5° solar elevation) at the surface (d, black dashed line), downwelling light at 500 m (e, dark gray dashed line), and moonlight (full moon at 70° elevation above horizon) at the surface (f, light gray dashed line). Note how the spectral sensitivity of each species is tuned to the light spectra of their respective habitat.