Bat eyes have ultraviolet-sensitive cone photoreceptors

Abstract

Mammalian retinae have rod photoreceptors for night vision and cone photoreceptors for daylight and colour vision. For colour discrimination, most mammals possess two cone populations with two visual pigments (opsins) that have absorption maxima at short wavelengths (blue or ultraviolet light) and long wavelengths (green or red light). Microchiropteran bats, which use echolocation to navigate and forage in complete darkness, have long been considered to have pure rod retinae. Here we use opsin immunohistochemistry to show that two phyllostomid microbats, Glossophaga soricina and Carollia perspicillata, possess a significant population of cones and express two cone opsins, a shortwave-sensitive (S) opsin and a longwave-sensitive (L) opsin. A substantial population of cones expresses S opsin exclusively, whereas the other cones mostly coexpress L and S opsin. S opsin gene analysis suggests ultraviolet (UV, wavelengths <400 nm) sensitivity, and corneal electroretinogram recordings reveal an elevated sensitivity to UV light which is mediated by an S cone visual pigment. Therefore bats have retained the ancestral UV tuning of the S cone pigment. We conclude that bats have the prerequisite for daylight vision, dichromatic colour vision, and UV vision. For bats, the UV-sensitive cones may be advantageous for visual orientation at twilight, predator avoidance, and detection of UV-reflecting flowers for those that feed on nectar.

Figure 1. Rod and cone photoreceptors in the retina of C. perspicillata and G. soricina.

(A–C) Vertical sections of C. perspicillata retina immunostained for rod opsin (A), long-wave-sensitive (L) opsin (B) and short-wave-sensitive (S) opsin (C). Commonly, the antibodies labelled only the photoreceptor outer segments, but the S opsin antibodies also weakly labelled the somata and axons. (D–F) Double immunofluorescence labelling for the cone opsins in a flat-mounted retina of G. soricina. Examples of cones expressing both opsins are indicated by arrows, cones expressing S opsin only by circles. Cone outer segments containing roughly equal amounts of both opsins appear whitish in the merge. All micrographs shown at same magnification. ONL, outer nuclear layer; OPL, outer plexiform layer.

Figure 2. Combination of in situ hybridization and immunohistochemistry in a vertical section of C. perspicillata retina.

(A) Short-wave-sensitive (S) cone opsin transcript in a cone photoreceptor soma and inner segment. (B) Immunolabelling of S opsin in the photoreceptor outer segment. (C) Merging the two labels demonstrates that the transcript and the protein are in the same cell. OS, layer of photoreceptor outer segments; IS, layer of photoreceptor inner segements; ONL, outer nuclear layer.

Figure 3. ERG responses of C. perspicillata and G. soricina at mesopic conditions.

(A) Sample ERG responses from C. perspicillata to 550 nm light stimuli of increasing intensity (stimulus indicated on abscissa; duration 200 ms, stimulus intensities indicated near the traces, multiplied by 10¹¹ quanta•s⁻¹•cm⁻²). Each trace shows the average of 30–60 responses and is shifted vertically for clarity. (B) Intensity-response curves for 500 nm test flashes of increasing intensity in C. perspicillata (filled squares) and G. soricina (open squares). The peak response in G. soricina occurs at an approximately 10-fold lower intensity than in C. perspicillata. (C) Light adaptation in C. perspicillata was tested with 551 nm background illuminations of different intensities [0.18 • 10¹¹ quanta•s⁻¹•cm⁻² (circles), 0.79 • 10¹¹ quanta•s⁻¹•cm⁻² (triangles), 3.6 • 10¹¹ quanta•s⁻¹•cm⁻² (diamonds)]. 500 nm test flashes of increasing intensity were presented. With increasing background illumination, the response to a given flash intensity decreases. Squares represent the situation with no adapting light (same curve as in B). Data points in (B) and (C) show mean±s.e.m.; n = 6 for C. perspicillata and n = 3 for G. soricina. (D) Cone contributions to the ERG were determined using spectral stimuli to obtain the action spectra S(λ) for C. perspicillata (filled black squares) and G. soricina (open black squares). Sensitivities were measured at 13 wavelengths (λ) ranging from 365 nm to 682 nm. Flash sensitivity at each wavelength was determined from the intensity required to reach a b-wave criterion response of 15 µV and normalized to 0 at 365 nm. (E) For C. perspicillata, action spectra were also measured during chromatic adaptation to background lights of 551 nm (filled green squares; 3.6 • 10¹¹ quanta•s⁻¹•cm⁻²) or 656 nm (filled red squares; 28.1 • 10¹³ quanta•s⁻¹•cm⁻²) to assess the λ_max of the UV-sensitive pigment. The bleaching effect of the green background was stronger than that of the red background at intermediate wavelengths (450–550 nm), demonstrating rod-specific bleaching in this part of the spectrum. At the long- and the short-wave ends of the spectrum the effect of the green background was reduced in comparison to the effect of the red background. This indicates contribution of a UV and a long-wave cone photopigment. Test flashes in the chromatic adaptation measurements were 365 nm, 452 nm, 500 nm, 520 nm, 551 nm, 604 nm, and 649 nm. Data points in (D) and (E) show mean±s.e.m.; n = 9 for C. perspicillata and n = 8 for G. soricina. Absolute sensitivity at 365 nm was 5.84•10⁻¹⁰ 1(/quanta•s⁻¹•cm⁻²) for C. perspicillata and 1.99•10⁻¹⁰ 1/(quanta•s⁻¹•cm⁻²) for G. soricina.

Figure 4. Spectral transmittance of cornea and lens of C. perspicillata from 250 to 750 nm.

Mean values for four corneas and two lenses of two adult individuals are shown. Both cornea and lens showed high transmittance in the UV-range (310–380 nm) of the spectrum. Transmittance of the cornea was <10% below 280 nm but rose sharply to more than 80% transmittance at 300 nm and up to 100% towards 750 nm. The lens showed a sharp rise from <10% transmittance at 300 nm to 50% at 310 nm, and then a continuous increase to 100% transmittance at 750 nm.

Figure 5. Comparison of action spectra for G. soricina obtained by different methods.

Our ERG measurements (left ordinate, squares) were performed at mesopic light conditions, whereas behaviour data (right ordinate, circles) were collected under scotopic light conditions . Absolute sensitivity is plotted against wavelength. For better comparison, the two sensitivity curves are shifted vertically to overlap at 520 nm; the behavioural response is actually about 1 log unit more sensitive (see discussion). The most noticeable difference is a higher UV sensitivity in the mesopic ERG curve. Sensitivities are given in 1/(quanta•s⁻¹•cm⁻²).