A unique visual pigment expressed in green, red and deep-red receptors in the eye of the small white butterfly, Pieris rapae crucivora

Animal eyes need to have different classes of spectral photoreceptors for colour vision. The honeybee (Apis mellifera), the classical example of an insect species with colour vision, has compound eyes with short (UV)-,middle (blue)- and long (green)-wavelength photoreceptors, which provide the physiological basis of a trichromatic colour vision system. Many butterflies employ so-called red receptors, with spectral sensitivities peaking near 600 nm or even at longer wavelengths (Arikawa et al., 1987; Bernard,1979; Matic,1983). These long-wavelength photoreceptors are most likely used for discriminating flower colours upon searching for food(Kelber and Pfaff, 1999; Kinoshita et al., 1999).

The sensitivity wavelength range of a photoreceptor cell is principally determined by the absorption spectrum of its visual pigment. A visual pigment molecule consists of an opsin protein with an 11-cis retinal chromophore. Absorption of light by the visual pigment molecule converts the 11-cis retinal to all-trans retinal, which then triggers the signal transduction cascade, resulting in a change of the membrane potential of the photoreceptor cell. The wavelength range where light effectively isomerizes the chromophore depends on the interaction of certain amino acids of the opsin with the chromophore; i.e. the opsin's amino acids together with the chromophore determine the spectral sensitivity of a photoreceptor. A distinct spectral sensitivity correlates with a unique amino acid sequence, as has been established for honeybees(Mardulyn and Cameron, 1999; Townson et al., 1998). This also holds for the Japanese yellow swallowtail butterfly, Papilio xuthus, where green and red receptors express different mRNAs encoding different visual pigment opsins (Kitamoto et al., 1998).

Optical factors often play a modulatory role in the photoreceptor spectral sensitivity. For example, the spectral sensitivity of the red receptors of Papilio xuthus peaks at 600 nm, but the spectrum is considerably narrower than predicted for a visual pigment with peak absorbance at 600 nm. The red receptors of Papilio are located in the proximal tier of those ommatidia where the rhabdom is surrounded by red pigmentation, which acts as a red transmittant spectral filter. While the absorption spectrum of the visual pigment peaks at 575 nm, the filter shifts the spectral sensitivity such that it peaks at 600 nm (Arikawa et al., 1999b).

We here report an extreme case of spectral filtering in the eye of the small white butterfly, Pieris rapae crucivora. Pieris has three types of long wavelength photoreceptors, peaking at 560 nm (green), 620 nm (red),and 640 nm (deep-red), accordingly called L560, L620 and L640 receptors,respectively (Qiu and Arikawa, 2003a,b). The Pieris eye consists of three distinct types of ommatidia, which are characterized by the perirhabdomeral pigmentation: a pale-red pigment in type I and III ommatidia and a deep-red pigment in type II ommatidia(Qiu et al., 2002). In all ommatidial types, two of the four distal photoreceptors, R3 and R4, are L560 receptors. The proximal photoreceptors of type I and type III ommatidia are L620 receptors, whereas the proximal photoreceptors of type II ommatidia are L640 receptors (see Table 1).

These findings pose several questions. First, how are the spectral sensitivities of the long-wavelength photoreceptors of Pierisdetermined, and do they have different visual pigments? What is the function of the perirhabdomeral pigmentation? At the proximal end of the rhabdom, a tapetal mirror exists that reflects light back into the rhabdom. Do the mirrors affect the spectral sensitivity of photoreceptors? In order to answer such questions, we first performed a molecular cloning study to identify multiple different mRNAs encoding opsins for long-wavelength absorbing visual pigments. What we found was a single opsin of a long-wavelength absorbing visual pigment opsin, which we term PrL (Pieris rapae long-wavelength visual pigment). Furthermore, we carried out histological in situhybridization to localize the photoreceptors that express PrL mRNA. Quite unexpectedly, the mRNA encoding PrL was detected in all the L560, L620 and L640 receptors in the Pieris eye. The three long-wavelength-sensitive Pieris photoreceptors can be explained with a unique visual pigment when assuming that the two types of perirhabdomeral pigments act as long-wavelength transmittant, spectral filters.

We used spring-form individuals of Pieris rapae crucivoraBoisduval. The butterflies were taken from a laboratory stock culture derived from eggs laid by females captured in the field around the campus of Yokohama City University, Yokohama, Japan. The hatched larvae were reared on fresh kale leaves at 19°C under a light regime of 8 h:16 h light:dark. The pupae were stored at 4°C for at least 3 months and then allowed to emerge at 25°C. The adults were used within 4 days after emergence.

The method of molecular cloning of Pieris opsins was as described previously (Kitamoto et al.,1998). Briefly, retinal mRNA was extracted using the QuickPrep Micro mRNA purification kit (Amersham Pharmacia Biotech Inc., Piscataway, NJ,USA) from eyes rapidly frozen in liquid nitrogen. For amplifying the cDNAs of long-wavelength absorbing visual pigments by RT-PCR, we designed two sets of degenerate primers based on amino acid sequences conserved in long-wavelength absorbing visual pigments of the swallowtail butterflies Papilio xuthus (Kitamoto et al.,1998) and Papilio glaucus(Briscoe, 1998) and the hawkmoth Manduca sexta (Chase et al., 1997); for the sequence of the primers, see the legend of Fig. 1. RT-PCR using all of these primers identified a single DNA fragment with an opsin-like sequence. To obtain the full-length cDNA, we carried out 3′ and 5′-RACE.

To compare the amino acid sequence deduced from the cloned cDNA with opsins of other insects so far identified, the sequences were aligned using an alignment program (CLUSTAL W 1.6), and then a phylogenetic analysis was performed by the neighbour joining method (PHYLIP 3.572), with octopus opsin as an outgroup.

The compound eyes of Pieris were fixed in 4% paraformaldehyde in 0.1 mol l^–1 sodium phosphate buffer (pH 7.2) for 0.5–2 h at 25°C. After dehydration with an ethanol series, we embedded the eyes in paraplast. The paraplastembedded eyes were sectioned at ∼8 μm thickness with a rotary microtome.

Probes for in situ hybridization were designed to hybridize to∼400 bases of the mRNA at the non-coding region downstream of the C-terminal. The corresponding cDNA region was first subcloned into pGEM-3zf(+)vector, and then digoxigenin (DIG)-labelled cRNA was generated using the DIG-RNA labelling kit (Roche, Mannheim, Germany).

For labelling, sections were first de-paraffinized and treated with hybridization solution [300 mmol l^–1 NaCl, 2.5 mmol l^–1 EDTA, 200 mmol l^–1 Tris-HCl (pH 8.0),50% formamide, 10% dextran sulphate, 1 mg ml^–1 yeast tRNA,1× Denhardt's medium], containing 0.5 μg ml^–1 of the cRNA probe, at 45°C overnight. After a brief rinse, the sections were incubated in 50% formamide in 2× SSC (saline sodium citrate buffer) at 55°C for 2 h and then treated with RNase (10 μg ml^–1)at 37°C for 1 h. The probes were further visualized by anti-DIG immunocytochemistry.

Four clusters of pigment surround the rhabdoms distally in all ommatidia of the fronto-ventral eye of Pieris. The pigment clusters thus act as absorption filters for the R5–8 proximal photoreceptors. The pigment is pale-red in type I and III ommatidia and deep-red in type II ommatidia. The R5–8 photoreceptors in the different ommatidial types have spectral sensitivities depending on the wavelength, λ, peaking in the red and deep-red, respectively (Qiu and Arikawa,2003b). The central hypothesis of the present paper is that the two types of distal screening pigment create the two types of R5–8 spectral sensitivities by selective spectral filtering.

The distal absorbance, A_d, as given by equation 5,consists of three terms: (1) the absorbance of the dioptrics; (2) the absorbance of the various visual (and possibly other absorbing) pigments within the rhabdom and (3) the absorbance of the red screening pigments near the rhabdom boundary. The first term is probably, in very good approximation, a constant, i.e. independent of wavelength (Stavenga, 2004). The second term is described by A_dv=–log(T_dv), where the transmittance of the distal visual pigment, T_dv, in the green to red wavelength range is given by an expression similar to equation 2: T_vd=exp(–f_3,4ηκ_maxαL_d),where f_3,4=f₃+f₄is the sum of the rhabdom volume fractions of photoreceptors R3 and R4, and L_d is the length of the distal rhabdom tier. Anatomical data show that f_3,4=0.3 and L_d=250μm (Qiu et al., 2002).

With four degenerate primers (Fig. 1), we performed RT-PCR with the mRNA extracted from the Pieris eye. In any combination of primers we could amplify cDNA fragments, which eventually appeared to originate from a single mRNA encoding a visual pigment opsin. After performing 3′ and 5′-RACE, we found that the cDNA has a single open reading frame of 1146 bp, encoding 382 amino acids (Fig. 1). Thephylogenetic relationship of the novel opsin with other insect opsins indicates that it falls into the group of long-wavelength absorbing visual pigments(Fig. 2). We therefore termed the opsin PrL (Pieris rapae long-wavelength absorbing visual pigment).

Fig. 3 shows the results of histological in situ hybridization, together with a diagrammatical sketch of the ommatidium (Fig. 3A) and unstained histological sections of the Pierisretina through the distal (Fig. 3B) and the proximal (Fig. 3C) tier of the retina. In the distal tier, the probe for PrL mRNA hybridized the R3 and R4 photoreceptors in all ommatidia(Fig. 3D). In the proximal tier, the same probe labelled the R5–8 photoreceptors in all ommatidia(Fig. 3E). We could not identify any obvious labelling in the R9 basal photoreceptors.

The spectral sensitivity of the distal R3,4 photoreceptors, determined by intracellular recordings (Qiu and Arikawa,2003a), well approximates the absorption spectrum of a visual pigment peaking at 563 nm (Fig. 4A). We therefore may conclude that PrL, expressed in R3,4, is an R563 visual pigment. This conclusion immediately implies that all R5–8 photoreceptors have an R563, although their spectral sensitivities severely deviate from an R563 absorption spectrum(Fig. 4B). The R5–8 photoreceptors form two classes, L620 and L640, that correlate with the colour of the pigment clusters in the ommatidium of the photoreceptor, being pale-red(PR; in type I and III ommatidia) or deep-red (DR; in type II ommatidia),respectively.

The two photoreceptor classes can be explained with a simple model, based on the assumption that the pale-red and deep-red pigments act as spectral filters, which are positioned fully distally, i.e. in front of the proximal rhabdom, formed by the rhabdomeres of photoreceptors R5–8. The absorbance spectra of the screening pigments can be estimated with the procedure outlined in the Materials and methods, as visualized in Fig. 4B,C. The first step is the calculation of the log absorptance of the proximal photoreceptors,log(1–T_p), where T_p is the transmittance of the proximal tier of the rhabdom. Red pigments are assumed to be transparent at long wavelengths, and hence the log sensitivity curves should match the log absorptance spectrum at long wavelengths. This indeed occurs for both L620 and L640 receptors(Fig. 4B). The difference between the log absorptance and log sensitivity curves then yields the absorbance spectra for the retinal material situated distally(Fig. 4C).

We identified a single long-wavelength absorbing opsin (L-opsin) mRNA in the eye of Pieris rapae crucivora. This is quite parsimonious compared with Papilio xuthus, which has three different L-opsins,PxL1–3 (Kitamoto et al.,1998), and therefore we used various sets of degenerate primers to check extensively whether other long-wavelength opsin mRNAs exist in the Pieris eye. So far, we have been unable to trace more than the unique long-wavelength opsin, PrL, in Pieris. The complementary observation that all R3–8 photoreceptors, without exception, express the unique PrL mRNA strongly indicates that Pieris relies on a single opsin for the different long-wavelength photoreceptors. We note here that our PrL labelling has so far failed to clearly label R9 photoreceptors. If R9 photoreceptors do not have PrL, they must express some other long-wavelength absorbing visual pigment, because electrophysiological recordings indicate that R9s are red sensitive (Shimohigashi and Tominaga,1991). Of course, we cannot exclude the possibility that the, as yet unknown, visual pigment of R9 is expressed in the other red receptors,R5–8, together with PrL: coexpression of multiple visual pigments has been found in many photoreceptors in the eye of Papilio xuthus(Arikawa, 2003). Furthermore,it should be mentioned that the actual absorbance spectrum of PrL in the different classes of photoreceptors could, in principle, be different due to post-translational modification or specific interactions with unique cellular components, but there is no clear evidence for such phenomena.

The absorbance spectra of the screening pigments in the pale-red (PR)ommatidia (type I and III) and deep-red (DR) ommatidia (type II) in the distal retina of the Pieris eye were estimated by subtraction of the experimental log spectral sensitivities from the calculated log absorptance spectrum of the proximal rhabdom tier. The resulting spectra show a substantial absorption of approximately 1 and 2 log units at wavelengths of<600 nm, but moreover clearly show a difference in spectral cut-off for the two types of screening pigments. Whereas the PR absorbance is low above∼620 nm, the DR absorbance is only minor above 660 nm. This is in full agreement with measurements of the eye shine, which showed that eye reflectance sharply drops at wavelengths below 620 and 660 nm for the two ommatidial types that can be distinguished with optical methods(Qiu et al., 2002). It should be noted, however, that the derived absorbance spectra of Fig. 4C are not exclusively due to the screening pigments. The visual pigments in the distal rhabdom tier also function as spectral filters. The absorbance, given by spectrum A_dv in Fig. 4B (see Materials and methods), is maximally ∼0.17, so the effect of the distal visual pigment is minor. The absorbance spectra include the possible spectral effects of the dioptric apparatus. Although the facet lens and crystalline cone will be fully transparent, the channelling of light into the rhabdom will not be spectrally flat. Presumably, the spectral modulations are very minor (see the case of fly eyes: Stavenga, 2003a,b). Furthermore, as explained in the Materials and methods, the light reflected by the tapetal mirror slightly modulates the absorptance spectrum of the proximal photoreceptors. The maximal contribution from the tapetum is given by log(1+T_p), but it is probably much smaller. At any rate,the absorbance spectra of Fig. 4C will not essentially change when the contributions of distal visual pigment filtering and tapetal mirror are fully accounted for. A detailed analysis cannot be given here because first the absorption characteristics of the short-wavelength visual pigments and then the existing short-wavelength spectral filters have to be analyzed in more detail(Arikawa et al., 1999a).

Many lepidopterans appear to have a long-wavelength absorbing visual pigment in six of the eight main photoreceptors. The long-wavelength visual pigments characterized photochemically in the nymphalids Vanessa cardui (Briscoe et al.,2003) and Polygonia c-album(Vanhoutte, 2003) absorb maximally at ∼530 nm, and the similarly dominant visual pigment of the moth Manduca sexta peaks at 520 nm(White et al., 2003). This can be compared with the honeybee, where six of the eight main photoreceptors have a 530 nm visual pigment (M. Kurasawa, M. Giurta and K. Arikawa, manuscript in preparation). (Note also that, in flies, six large photoreceptors have the same visual pigment, peaking at 490 nm.) In Apis, Vanessa and Manduca, the two additional photoreceptors are UV and/or blue receptors, which are distributed heterogeneously in the ommatidial lattice. The same long-wavelength visual pigment appears to exist in all long-wavelength receptors, however, indicating that these animals have a homogeneous retina, when considering the long-wavelength range. This agrees with optical observations on common nymphalids, which show a homogeneous eye shine (Stavenga, 2002a).

The heterogeneous eye shine observable in many butterflies, including Pieris, demonstrates that many butterflies have diversified their long-wavelength receptors. As a first example, the dorsal parts of the eyes of the satyrine Bicyclus anynana have ommatidia, which all have the same green–orange eye shine. Ventrally, red-reflecting ommatidia intersperse the yellow–orange shining ones. Spectral measurements strongly suggest that in the red-reflecting ommatidia red spectral filters occur, presumably shifting the spectral sensitivity of sets of long-wavelength receptors(Stavenga, 2002b). Pieris applies one of two types of red spectral filters in the ommatidia in the fronto-ventral eye area. Papilio xuthus also uses two types of screening pigment, i.e. either a red or a yellow pigment, but moreover combines these pigments with different types of long-wavelength-sensitive visual pigments. Papilio xuthus exercises further extravagance by expressing two different long-wavelength visual pigments in certain photoreceptors. Presumably, these differences increase the potential for colour vision. At least, Papilio xuthus can boast extreme colour discrimination capacities(Kinoshita and Arikawa, 2000; Kinoshita et al., 1999).

Eye shine studies suggest that the diversification of spectral properties strongly depends on species and eye region(Stavenga, 2002a). For example, whereas the ommatidia in the main part of the eye of Pieris rapae reflect red or deep-red light, the ommatidia in the dorsal part of the eye reflect in the yellow wavelength range, in agreement with the absence of red screening pigment dorsally (Ribi,1979). The red spectral filters in the fronto-ventral area create extremely long-wavelength shifted photoreceptors, presumably to improve the capacity to discriminate food plants(Kelber, 1999), which are more often seen with the ventral than the dorsal eye area.

The work was financially supported by Grants-in-Aid for scientific research from the Mitsubishi Foundation Grant for Natural Science #33-05, Grant-in-Aid for Scientific Research from JSPS 14204080, and Special grant for promotion of science from YCU to K.A., and by the EOARD to D.G.S.