Regulation of dopa decarboxylase expression during colour pattern ...

2303

Development 125, 2303-2313 (1998) Printed in Great Britain © The Company of Biologists Limited 1998 DEV5191

Regulation of dopa decarboxylase expression during colour pattern formation in wild-type and melanic tiger swallowtail butterflies P. Bernhardt Koch1, David N. Keys2, Thomas Rocheleau3, Katherine Aronstein3, Michael Blackburn3, Sean B. Carroll2 and Richard H. ffrench-Constant3,* 1Department of General Zoology, University of Ulm, Albert-Einstein-Allee 11, D-89081 Ulm, Germany 2Howard Hughes Medical Institute and Laboratory of Molecular Biology, Bock Laboratory, University of

Wisconsin-Madison, Madison, WI 53706, USA 3Department of Entomology, 237 Russell Laboratories, 1630 Linden Drive, University of Wisconsin-Madison, Madison, WI 53706, USA *Author for correspondence (e-mail: [email protected])

Accepted 3 April; published on WWW 19 May 1998

SUMMARY The eastern tiger swallowtail butterfly Papilio glaucus shows a striking example of Batesian mimicry. In this species, females are either wild type (yellow and black) or melanic (where most of the yellow colour is replaced by black). In order to understand how these different colour patterns are regulated, we examined the temporal order of wing pigment synthesis via precursor incorporation studies, enzyme assays, and in situ hybridisation to mRNA encoding a key enzyme, dopa decarboxylase. We show that dopa decarboxylase provides dopamine to both of the two major colour pigments, papiliochrome (yellow) and melanin (black). Interestingly, however, dopa decarboxylase activity is spatially and temporally regulated, being utilised early in presumptive yellow tissues and later in black. Further, in melanic females, both dopa decarboxylase activity and early papiliochrome synthesis

INTRODUCTION Colour pattern formation is biologically important because the overall pattern of an organism interacts directly with natural selection through such processes as cryptic colouration and mimicry. The breadth of colour pattern diversity displayed in butterflies (Nijhout, 1991) reflects the complex interplay between natural selection and colour phenotypes, making butterfly wing patterns an excellent system in which to study the evolution of the underlying developmental mechanisms. One of the most highly studied examples of such an ecologically significant change in pattern formation involves the repeated occurrence of melanism in Lepidoptera (Kettlewell, 1973; Nijhout, 1991). Melanism, which refers to a mutant phenotype expressing black melanin in an atypical pattern, has primarily been examined in two ecological situations. Firstly, ‘industrial’ melanism, where melanism is thought to provide cryptic colouration within the context of industrially darkened backgrounds. Secondly, Batesian mimicry, where melanics of one species mimic another that is

are suppressed in the central forewing and this normally yellow area is later melanised. These results show that the regulation of enzyme synthesis observed in the yellow/black pattern of a single wing, is similar to that involved in melanism. We infer that dopa decarboxylase activity must be regulated in concert with downstream enzymes of either the melanin and/or the papiliochrome specific pathways, forming part of a developmental switch between yellow or black. This modification of multiple enzyme activities in concert is consistent with a model of melanisation involving coordinate regulation of the underlying synthetic pathways by a single Y-linked (female) factor. Key words: Colour formation, Papilio glaucus, Batesian mimicry, Butterfly, Sex-linkage, Dopa decarboxylase, Papiliochrome, Melanin, β-Alanyl-transferase

avoided by predators (Kettlewell, 1973). This process can involve either an overall darkening of wing colour, or patternspecific changes of lightly coloured areas to black. A striking example of proposed Batesian mimicry is the sexlinked melanism displayed by the eastern tiger swallowtail Papilio glaucus of North America (Family Papilionidae) (Clarke and Sheppard, 1959, 1962; Clarke and Clarke, 1983; Scriber et al., 1996). Females of this species can be either ‘wild type’ (predominantly yellow with black ‘tiger’ stripes) or ‘melanic’ where much of the wild-type yellow pattern is replaced with black (Fig. 1A). These melanic females (as well as a number of other North American butterfly species; Scott, 1986) are thought to mimic the pipevine swallowtail, Battus philenor, which flies throughout much of the same range as P. glaucus and is distasteful to predators such as birds (Brower, 1958). In Lepidoptera females are heterogametic XY (also referred to as WZ) and males homogametic XX. Previous genetic analysis in P. glaucus therefore suggested that melanism is controlled by a single locus on the Y chromosome that is either absent from the X or present but suppressed

2304 P. B. Koch and others (Nijhout, 1991). Similarly, Scriber described a single major sex-linked gene (the Y-linked black or b gene) and a modifying gene (the X-linked suppressor, s) which suppresses the melanic phenotype in hybrid crosses to the closely related P. canadensis (Scriber et al., 1996). Colour pattern in butterfly wings is formed by the colour and arrangement of individual scales on the wing surface (Nijhout, 1991). Wings develop in the larva from embryonic cells which form wing buds or anlagen, and by the end of pupation the fully everted wing disc has adopted its final shape, tracheal pattern and total pre-inflation area. During the pupal stage, scale cells differentiate from the wing disc epidermis and it is these cells that synthesise the colour pigments shortly before adult emergence. The pigments that underlie these colours belong to a range of different chemical classes: pteridines forming white, yellow or red; ommatins forming red or red-brown and melanins forming grey or black (Nijhout, 1980; Koch, 1992; Koch and Kaufmann, 1995). Interestingly, some pigments are only found in certain groups of Lepidoptera, for example the yellow papiliochromes discussed here are exclusive to the papilionids (Umebachi, 1985). We were interested in testing whether there is a developmentally ordered succession of pigment synthesis in butterfly wing pattern formation and if it is conserved between different butterfly families. If so, given a highly conserved order of pigment synthesis, how is such a complex temporally and spatially controlled developmental pathway altered by apparently single major genes to give melanic traits? To address the former issue, we examine the developmental order of pigment synthesis in four different butterfly species. To address the latter issue, we used the wild-type and melanic female P. glaucus as a model system. We predicted from previous work that melanin (Koch, 1994; Koch and Kaufmann, 1995) and papiliochrome (Ishizaki and Umebachi, 1988; Ishizaki and Umebachi, 1990) synthesis would utilise a common enzymatic pathway through the use of dopa Fig. 1. Sex-linked melanism in P. glaucus and developmental staging of pigment synthesis. (A) Dorsal wing surface of adult wild-type (left) and melanic (right) female P. glaucus. Note the preservation of the distal row of yellow bands on the wing margins of the melanic females. (B) Developmental staging of colour pattern formation in wild-type P. glaucus on both dorsal and ventral wing surfaces. (C) Succession of dorsal pigmentation for both wild-type and melanic females. Developmental stages VII-0 were defined with reference to visible pigmentation and time to adult eclosion. Stage VII, wings are soft and uncompacted with green colouration due to hemolymph. Stage VI, wings compact and become pinkish in colour. However, no scale pigmentation is yet visible. Stage V, pigmentation starts with white and red/yellow also become visible. Stage IV, red and yellow pigmentation becomes intense but there is no black apparent. Stage III, black pigmentation starts on the dorsal, but not ventral, side of the wings. Stage II, the wings are fully blackened dorsally. Stage I, pigmentation is complete. Stage 0, adult eclosion. Note, wings in B and C are actual size and stages VII and O are not shown.

decarboxylase (DDC), which catalyses the conversion of dopa to dopamine. This dopamine would then be used as a substrate in either the papiliochrome- or melanin-specific elements of the synthetic pathway (Fig. 3A). However, it is not clear at which

Wing colour pattern formation in butterflies 2305 stage, or stages, in the synthetic pathway the developmental mechanisms that determine yellow versus black control the switch between these pigments. Previous work has shown that Ddc is transcriptionally regulated in spatially specific patterns during pattern development in the wings of Precis coenia (Bergander et al., 1996), a butterfly from a different family, the Nymphalidae, which synthesises both dopamine- and nondopamine-derived colour pigments. Here, we take several approaches to elucidate the regulatory changes controlling the switch between papiliochrome and melanin synthetic pathways in P. glaucus. We describe the temporal development of wing colour patterning both in the papilionid P. glaucus and also in three nymphalid species. Irrespective of spatial pattern, the different pigments are synthesised in a common order, always finishing with black melanin. This ordered appearance of pigments appears to be a universal feature of butterfly wing pattern formation, suggesting that temporal as well as spatial regulation of pigment synthesis may play a key role in patterning. Using precursor uptake, enzyme activity assays, and in situ hybridisation, we describe several components of pigment biosynthesis. We show that the regulation of Ddc, as well as the uptake of pathway-specific precursors, correlates with the timing of the pigment synthesis cascade in different areas of the wing. Despite being common to both pathways, differential

spatial and temporal regulation of DDC activity is associated with the developmental switch between yellow and black pattern elements. Further, the temporally ordered nature of pigment synthesis allows us to show that in the mimetic melanic females there is no early precursor incorporation, DDC activity or papiliochrome deposition, in the proximal (normally yellow) areas of the wing. The replacement of yellow with black in melanic individuals therefore appears to involve a switch which utilises the same developmental control of branch points in the underlying biosynthetic pathways as those used to determine yellow versus black elements in an individual normal wing. We propose that melanism is associated with a single Y chromosome (female) linked factor involved in suppressing early papiliochrome synthesis and also in promoting later atypical melanisation of the same formerly yellow area. Further, as dopamine is required in both biochemical pathways, we speculate that this factor must also regulate other downstream enzymes in the synthetic pathway. MATERIALS AND METHODS Insects and staging For the staging of wing colour development in P. glaucus we obtained pupae of progeny of field-collected wild-type (yellow and black) or

Fig. 2. The temporal order of pigment synthesis in wings of other butterfly species. Comparison of the developmental succession (from top to bottom) of pigment formation on the dorsal and ventral wing surfaces of three different species of nymphalid butterfly: (A) the map butterfly, Araschnia levana L., (B) the peacock, Inachis io L. and (C) the buckeye, Precis coenia Butler. Note the conservation of the succession of pigment synthesis in the three different species with the early deposition of the white pteridines, followed by the red ommatins and finally the black of melanisation. This mimics the succession of yellow/red then black synthesis in the Papilio (Fig. 1) but involves different chemical classes of pigments (see text). Scale bar, 1 cm.

2306 P. B. Koch and others melanic (largely black) females. As melanic females almost invariably produce melanic female offspring (Clarke and Sheppard, 1959, 1962; Scriber et al., 1996), we could reliably predict that upon wing development such females would show the black melanic phenotype. Pupae were from the second generation of P. glaucus that undergo an obligate overwintering pupal diapause which lasts a minimum of 100 days (M. Scriber, personal communication). Pupae were therefore stored at 4°C for >100 days and then taken out in batches and transferred to a growth chamber at 26°C. On the same day the pupae were transferred to 26°C, they were each injected with 20hydroxyecdysone (0.5 µg dissolved in 5 µl of Grace’s insect tissue culture medium) to trigger adult development. To compare the development of colour formation in a range of butterfly species we performed similar staging experiments in three nymphalids (Nymphalidae): the map butterfly, Araschnia levana L., the buckeye, Precis (Junonia) coenia Butler, and the peacock, Inachis io L. Methods for staging of nymphalids such as P. coenia have been described elsewhere (Koch and Nijhout, 1990). Following injection, the developmental status of each P. glaucus pupa was estimated by manually testing the hardness of the cuticle over the developing wings. This area of cuticle softens noticeably during the last three days before eclosion, the time during which wing pigmentation is initiated and the wings become readily dissectable upon removal of the overlying pupal cuticle. After dissection, wings destined for visual staging or mRNA extraction were transferred into Grace’s medium and held on ice for up to 2 hours prior to staging or freezing. Wings were staged visually according to the colour and extent of different pigments (see Results). For mRNA extraction, batches of 40 wings (each from 10 staged pupae) were transferred from the Grace’s medium, dried by touching to a paper tissue and then flash frozen in liquid nitrogen and stored at −80°C. For in vitro precursor incubation studies, wing pairs from individual pupae were stored separately. For in situ hybridisation wings were fixed immediately after dissection, as described below. Precursor incorporation In order to study the incorporation of specific radiolabelled precursors into P. glaucus pigments we used the following radiolabels: L-[3-14C]tryptophan (DuPont/NEN, 54.2 Ci/mol), β-[114C]alanine (DuPont/NEN, specific activity 53.0 Ci/mol) and L-[U14C]tyrosine (Amersham, 473 Ci/mol). For in vivo incorporation, radiolabells were injected (1.0 µCi β-alanine, 0.5 µCi tyrosine or 0.2 µCi tryptophan in 10 µl of Grace’s medium with a HamiltonBonaduz 10 µl syringe) directly into the hemocoel through an intersegmental membrane on the lateral side of the pupal abdomen. The time to adult emergence was recorded individually for each pupa. Eclosing adult butterflies were left to harden their wings for 24 hours and were then frozen at −20°C prior to wing scale removal for autoradiographic analysis (Koch, 1991). For in vitro experiments individual radiolabelled precursors were added to Grace’s medium (Sigma). As commercial Grace’s medium lacks sufficient concentrations of pigment synthesis precursors for proper levels of colour formation, we supplemented the medium with unlabelled L-kynurenine, β-alanine and L-tyrosine at final concentrations of 0.5, 2.5 and 1.8 mmol/liter respectively. Individual pairs of fore- and hindwings were then incubated for 5 hours (in 2 ml of modified Grace’s medium in a 12-well tissue culture plate) in pulse-chase experiments (with 0.2 µCi/ml tryptophan, 1 µCi/ml βalanine or 0.5 µCi/ml tyrosine) as described elsewhere (Koch and Kaufmann, 1995). The left and right wings of the same individual butterflies were incubated with different precursors in order to compare their incorporation pattern at exactly the same stages of development. DDC enzyme activity DDC enzyme activity in wing tissues was measured radiochemically

in an assay described elsewhere (McCaman et al., 1972). For tissue samples of different wing patterns (see Fig. 5B, C for source locations of dissected tissue samples) we punched out 1.2 mm diameter discs (20-26 per assay) of tissue from the developing wing using a hollow steel rod or ‘punch’ (Koch and Kaufmann, 1995). It is important to note that samples derived from the marginal black wing band always also contain the distal yellow marginal spots, which are present in both the wild-type and melanic morphs. Such mixture of tissues is inevitable given the fine scale of the microdissection.

Ddc cloning and mRNA expression To examine the underlying basis of differential DDC enzyme activity we analysed both the levels and the spatial distribution of the Ddc encoded message. We used Ddc cDNA probes in both a developmental northern analysis and also for in situ hybridisation of RNA probes to whole mounted butterfly wings. Initially, we used a Ddc cDNA probe from a moth, the tobacco hornworm Manduca sexta, as a heterologous probe for fluorescent in situ hybridisation. Following the cloning of a P. glaucus cDNA fragment, we switched to using this homologous probe for developmental northern analysis and 33P-labelled in situ hybridisation. The M. sexta cDNA probe (a 3.1 kb insert in the Bluescript cloning vector, Stratagene) was a kind gift from Dr K. Hiruma from the laboratory of Dr L. M. Riddiford, University of Washington, Seattle. We also cloned a short section of the P. glaucus Ddc cDNA by use of degenerate primers in the polymerase chain reaction (GenBank accession number AF036963). We used a nested primer design with two forward (F1 and F2) and two reverse (RC1 and RC2) primers. PCR amplification conditions were: 75°C for 2 minutes, 94°C for 2 minutes (94°C for 1 minute, 45°C for 1 minute and 72°C for 1 minute; for 34 cycles), and then 75°C for 5 minutes. PCR products were electrophoresed on 1% ethidium bromide stained agarose gels to check for products of the predicted size (F2-RC1, 233 bp and F1-RC2, 245 bp). Products were then cloned into the TA cloning vector (Invitrogen) and sequenced on both strands using dye terminator chemistry on an ABI-373A fluorescent sequencer (Perkin-Elmer).The sequences of the PCR primers were: F1: 5′-CATGTGGACGCKGCCTATGC-3′, F2: 5′-TCTTTCAATTTCAAYCCACA-3′, RC1: 5′-TGRAGATTYTCKACRCCRTACAGCC-3′, RC2: 5′-GGRATTTGCCAGTGACGRTAGTC-3′. For northern analysis of the relative abundance of Ddc mRNA in different wing developmental stages, total RNA was extracted from sets of 40 wings for each stage. Wings were homogenised in lysis buffer and poly(A)+ mRNA isolated with an OligotexTM mRNA extraction kit (Quiagen). 5 µg of poly(A)+ mRNA was loaded per lane on a 1% agarose-formaldehyde gel, electrophoresed overnight and then capillary blotted onto a Hybond N membrane (Amersham). Developmental northerns were probed both with the M. sexta and P. glaucus radiolabelled ([32P]dCTP) cDNA probes. For in situ hybridisation to whole mounted butterfly wings, P. glaucus wings were dissected and then fixed immediately in 4% paraformaldehyde in phosphate-buffered saline (PBS). Wings were treated as described elsewhere (Carroll et al., 1994), except that for the fluorescent in situ hybridisation, wings were also sonicated for 5 minutes (in addition to the normal proteinase K digestion) in order to improve penetration of the secondary probe. During the early stages of wing development (VI to IV) fluorescent immunological detection of a digoxigenin-labelled M. sexta RNA probe, corresponding to expression in presumptive yellow areas, was possible using a MRC1024 confocal microscope (BioRad) and a Cy5-conjugated mouse anti-digoxin IgG monoclonal antibody. However, at later stages in development the Cy5 fluorescent signal was quenched by the increasing level of pigmentation found in the developing wing. Ddc expression in situ, at this and in later stages of development, was therefore analysed with the use of radiolabelled (33P) P. glaucus Ddc riboprobes and autoradiography.

Wing colour pattern formation in butterflies 2307 RESULTS

Temporal and spatial patterns of pigment precursor incorporation In order to analyse the biosynthetic pathways underlying wing colour pigment synthesis in P. glaucus, and to test how these are altered in melanism, we followed the formation of yellow and black pigments using radiolabelled precursors predicted to be incorporated into specific chemical classes of pigments. We examined the temporal and spatial incorporation of the three main pigment precursors tyrosine, β-alanine and tryptophan both in vivo (after their injection into developing pupae) and in vitro (by adding precursors as supplements to wings isolated in tissue culture medium). Based on the current state of knowledge of papiliochrome synthesis in Papilio xuthus (Ishizaki and Umebachi, 1990; Umebachi, 1993) and melanin formation in Precis coenia (Koch, 1994; Koch and Kaufmann, 1995) (Fig. 3A), we predicted that both tryptophan and β-alanine would be incorporated into the yellow or orange papiliochromes. Further, that tyrosine would be the major precursor for the

Stages of pigment formation in P. glaucus and other butterflies We were interested in determining if the staged synthesis of butterfly wing colour pigments is a universal phenomenon among different butterfly species. We therefore analysed the different stages of colour formation in the papilionid P. glaucus and compared it to that in the nymphalid species Araschnia levana, Inachis io and Precis coenia. In P. glaucus pupal development lasts 10-12 days following injection of 20-hydroxyecdysone (mean = 11.1 days, s.e. ±0.1, n=126) and pigmentation started 2 days before adult emergence. In wild-type and melanic swallowtails (Fig. 1B,C), pigmentation is first visible in those scales that will form the white/blue iridescent scales in the prospective pattern. Secondly, red (starting on the ventral wing surface) and yellow pigmentation appears. Finally, black pigmentation appears on the dorsal surface of the wing and this melanisation continues on both sides of the wing surface until eclosion. Developmental stages (VII to 0, where VII is the earliest stage and 0 is the latest) were defined with respect to visible pigmentation and adult emergence and are described in Fig. 1B. In contrast, in melanic females (Fig. 1C), the yellow pattern elements do not appear in the central (proximal) parts of the wings as they do in wildtype females, despite the normal appearance of the red spots (distal hindwing) and terminal row of yellow bands (distal fore- and hindwing) before melanisation. This observation provides the first evidence that presumptive papiliochrome (yellow) pigment formation is absent from the central wing and that this area is later melanised black. We contrasted the temporal succession of pigment synthesis in the papilionid P. glaucus to that in three different nymphalid species (Fig. 2). Again, as in P. glaucus, colour pattern in the three nymphalids develops in a clear succession of stages corresponding to periods of different pigment synthesis and deposition. Initially white colouration appears, then red and then black. Note, however, that the red pigments in the nymphalids (ommatins) belong to a different chemical class than the reds and yellows of the papilionids (papiliochromes). One substantial difference can be observed in the nymphalids in that there is an additional final stage in which grey or grey-brown Fig. 3. The yellow and black pigment biosynthetic pathways. (A) A diagram of the pigments are deposited. These latter pigments proposed biosynthetic pathways underlying the formation of yellow and black or chemical processes may therefore not be pigments in swallowtail butterflies (see text for relevant literature citations). Only present in papilionids. two synthetic enzymes (circled) are indicated. (B-F) Autoradiographs showing in The observation that butterfly pigment vivo incorporation of radiolabelled precursors into the wing scales of P. glaucus synthesis, in a range of species from different during pigment formation. Precursors used were β-alanine, injected 2 days prior to families, follows a similar ordered cascade of emergence into both wild-type (B) and melanic females (C) and tryptophan (D) or pigment production is important because it tyrosine, injected either 3 days (E) or 1 day (F) before adult emergence. Note how suggests that differences in colour pattern within β-alanine (B) and tryptophan (D) are incorporated into the yellow areas of adult (e.g. melanism) and between species may be wings and that β-alanine is also incorporated into the distal row of yellow spots at programmed by similar developmental changes the edge of the wing of melanic females (C), whereas tyrosine is incorporated only into yellow areas when injected early (E), and later into black areas (F). involving temporal regulation.

2308 P. B. Koch and others

Fig. 4. Pattern of incorporation of pigment precursors at different stages in wing colour development. Autoradiographs showing in vitro incorporation of radiolabelled precursors into the wing pigments of P. glaucus after a 5 hour pulse incorporation. (A) Pigmentation stages corresponding to the initiation of precursor incubation are shown above for reference (see Fig. 1B). (B) Incorporation of tyrosine (left wing pairs at top) and β-alanine (right wing pairs from the same wild-type females at bottom). (C) Incorporation of tryptophan into wild-type female wings. (D) Incorporation of tyrosine and β-alanine into wings of melanic females. Note again, as in vivo (Fig. 3), how β-alanine is incorporated into yellow and red wing areas in both the wild-type and melanic females, whereas in vitro tryptophan is only incorporated late (stage I, panel C) into black areas of the developing wing (see text). Note also that tyrosine is widely incorporated into the abnormally melanised area of the melanic females (D). Wings are actual size. Two data points (stage I in B and stage IV in D) are missing due to shortage of staged wing material.

melanin synthesis pathway and would also be incorporated into papiliochrome, i.e. that it would be incorporated into both black and yellow pigments. This latter expectation arises from the fact that dopamine is complexed with β-alanine (by N-βalanyl-transferase) to give N-β-alanyl dopamine, a necessary component for papiliochrome synthesis (Umebachi, 1985).

In order to test the hypothesis that both tryptophan and βalanine are incorporated into the yellow (or red) papiliochromes, we examined autoradiographs of adult wings following pupal injection of precursors (Fig. 3B-F) and the temporal pattern of precursor incorporation of wings at different stages of development in vitro (Fig. 4). As predicted,

Wing colour pattern formation in butterflies 2309 β-alanine was incorporated only in the yellow areas of wildtype and melanic females both in vivo (Fig. 3B,C) and in vitro (Fig. 4B,D). In the case of the melanic females the yellow areas remaining are specifically the terminal row of yellow bands at the distal edge of both wings (Figs 3C, 4D). Tryptophan was also incorporated in vivo into yellow areas (Fig. 3D). Unexpectedly, in vitro, at later stages in pigment synthesis tryptophan was not incorporated into yellow but showed significant incorporation into black (Fig. 4C), suggesting it has the potential to contribute to both chemical pigments. Tyrosine was indeed incorporated, in vivo and in vitro, into both yellow and black areas as we expected. However, there was a distinct temporal difference. Tyrosine was first incorporated into yellow areas early in development (Figs 3E, 4B,D) and later incorporated into the black pattern (Figs 3F, 4B,D). This suggests that the switch between the black and yellow synthesis pathways is under tightly regulated developmental control. The patterns of precursor incorporation into the various colour pattern elements differ between the wild-type females and the black melanic females. In the central presumptive black areas of the melanic females there is no β-alanine or tyrosine incorporation during papiliochrome synthesis. Tyrosine is only incorporated late into melanin (Fig. 4B,D). This failure to incorporate β-alanine suggests that papiliochrome synthesis is absent from this area of the wing and is consistent with the observation that the yellow pattern element is missing from the central wings of the melanic females. Thus, the yellow versus black elements of the wild-type pattern are controlled by the differential timing of pigment synthesis and the black area of the melanic females appears to be regulated by the absence or suppression of yellow pigment synthesis in the central part of the wing. Spatial and temporal distribution of DDC mRNA and enzyme activity The observation that wing pattern pigment synthesis is strictly regulated both in time and space suggests early developmental control. Specifically, in P. glaucus, the patterns observed suggest that DDC may selectively produce dopamine in different tissues at different times (i.e. early in yellow and late in black) and that this enzyme may therefore play a key role in regulating the decision of a specific pattern element to be yellow or black.

In order to determine the potential role of the DDC enzyme in the differential timing of pigment synthesis, we examined the distribution of DDC enzyme activity in the whole wing and in the yellow and black areas of both wild-type and melanic

Fig. 5. DDC enzyme activity is differentially regulated in melanic P. glaucus. (A) Histogram of DDC enzyme activity in whole wings from wild-type and melanic females, superimposed on a line graph of the relative abundance of Ddc mRNA in the different wild-type wing stages (relative to maximal abundance at stage IV which is given a value of 1.0). (B) Comparison of the DDC activity in the yellow and black areas of wildtype (yellow) females. (C) Comparison of the DDC activity in tissue dissected from the central forewings of wild-type (yellow) or melanic (black) females. Note how the early DDC enzyme activity is lower in the central forewing of melanic females corresponding to the absence of papiliochrome synthesis in this specific area of the wing. Insets show the relevant wing colours and tissue samples analysed. Error bars are standard errors of the mean and numbers correspond to the number of samples analysed.

2310 P. B. Koch and others females. DDC enzyme activity in whole wings is low at stages VII-VI in both wild-type and melanic females. Later, in stage IV, activity was 2.4 times higher in whole wings from wildtype yellow females than in those from melanic black individuals. Subsequently, DDC activities were essentially similar (Fig. 5A). We also examined the time course of DDC activity within yellow or black areas from wild-type females. Activity increases first in prospective yellow areas and then later in black areas (Fig. 5B). This shows that DDC enzyme activity is regulated in a spatial and temporal fashion with initial activity peaking in yellow areas of presumptive papiliochrome synthesis and later in the presumptive black areas corresponding to melanin synthesis. We note that DDC enzyme activity increased later in the wild-type black pattern, despite the fact that the dissected tissue includes the terminal row of yellow bands found at the edge of the swallowtail wing. This spatial and temporal regulation of DDC enzyme activity was confirmed by a comparison of the central yellow areas of wild-type females with the same area from the black melanic females (the region in which yellow pigment synthesis appears to be lacking in the mutant). In this case, DDC activity in wild-type tissues increased earlier, and peaked higher, than that in the melanic tissues (Fig. 5C). This is consistent with the absence of papiliochrome synthesis in the central forewing of the melanic butterflies; i.e. DDC is not active to provide dopamine for papiliochrome synthesis. To determine if these differences in enzyme activity in differently coloured wing areas are controlled by differential spatial and temporal transcription of the Ddc gene, we examined the temporal and spatial distribution of Ddc message by northern analysis and in situ hybridisation. Developmental northern analysis of mRNA from whole wings shows that the 2.6 kb P. glaucus Ddc message is transcribed at very low levels before visible wing pigmentation (stage VII). During wing colour synthesis, the relative abundance of Ddc mRNA increases rapidly through stages VI-IV, peaks at stage IV, and then declines slightly through the remainder of wing colour development (Fig. 5A). Analysis of Ddc mRNA distribution by fluorescent in situ hybridisation shows that message abundance is tightly regulated both spatially and temporally. Initially, Ddc message appears during stages VI-IV and is found in the presumptive yellow areas of the wing. Ddc mRNA is abundant in the presumptive yellow areas of the central forewing (Fig. 6C) and the marginal yellow bands (Fig. 6B,D) of the wild-type pattern, where it is readily contrasted with much lower levels in the surrounding black pattern. Higher magnification reveals the clear-cut boundary in Ddc expression between individual presumptive yellow scales, which show Ddc message at this stage, and the adjacent presumptive black scales that do not (Fig. 6B). To examine Ddc transcription in later stages we utilized 33Plabelled RNA probes in order to avoid problems associated with fluorescent signal quenching due to high pigment concentrations in older wing tissue. Autoradiography of wholemounted wings confirms the early accumulation of Ddc message in the presumptive yellow pattern (Fig. 6E) but also shows that transcription is subsequently turned off in areas destined to be yellow and turned on in those that will be black (Fig. 6F).

DISCUSSION The order of colour pigment synthesis is conserved between different butterfly families The four butterfly species examined here, from two different families the Papilionidae and the Nymphalidae, all exhibit a similar developmentally ordered succession of pigment synthesis (compare Fig. 1 with Fig. 2). In the nymphalids, following the appearance of the white pteridines (Koch, 1992), the tryptophan-derived red ommatins (Koch, 1991, 1993) are synthesised before the start of melanisation. In the papilionid P. glaucus the yellow and red papiliochromes, which are tightly linked to the tryptophan pathway, are again synthesised before melanisation. Thus, despite the fact that the final coloured pigments have different chemical compositions in the different butterfly families, the tryptophan-derived pigments are always deposited before those derived from tyrosine being melanin. Different butterfly species therefore appear to display an evolutionary conserved succession of pigment synthesis, despite divergence in the details of the synthetic pathways involved in depositing the final coloured pigments. Differential expression of Ddc is involved in P. glaucus colour pattern formation P. glaucus presents a valuable model for the study of butterfly wing pattern formation both because of the uniqueness of the papiliochrome pigments that form the yellow colour of the wing and body (Umebachi, 1985) and also because of the existence of melanic ‘black’ females which represent a striking example of Batesian mimicry (Nijhout, 1991). Given the likely complexity of a system in which different pigments are synthesised at different times in different places, there are several possible mechanisms whereby colour differentiation could occur. Specifically, pigment production could involve differential uptake of precursors into individual scale cells, differential spatial and temporal expression of specific enzymes involved in pigment synthesis, or a combination of both (Koch, 1992). Here, via a combination of precursor incorporation studies, enzyme activity assays, and in situ hybridisation analysis of Ddc mRNA abundance, we show that the differential spatial and temporal expression of enzymes underlying pigment synthesis controls the yellow/black colour decision taken by each individual wing scale cell. We also demonstrate for the first time, that DDC supplies dopamine early in wing development for papiliochrome synthesis in presumptive yellow areas and later for melanisation of the black ‘tiger’ stripes. This clear inversion of the spatial pattern of Ddc message accumulation shows that Ddc gene expression is dynamically regulated and contributes dopamine both to early papiliochrome synthesis and later to the melanisation of the black stripes and borders of the swallowtail wing. However, the correlation of DDC activity with the differential timing of yellow/black patterning also implies that other elements of the pigment synthesis pathway must also be regulated to dictate the production of one pigment or another. Furthermore, we show that early in wing colour development, the high levels of DDC enzyme activity normally present in the central yellow wing areas of wild-type females are much reduced in the same pattern elements of melanic

Wing colour pattern formation in butterflies 2311 Fig. 6. Ddc gene expression is temporally and spatially regulated in P. glaucus wing colour development. Diagram of developing wing (A) shows the relative location of panels showing in situ analysis of Ddc mRNA (B-D). (B-D) Fluorescent detection of Ddc message in developing wings of wild-type (yellow and black) females. (B) Detail of presumptive yellow marginal band showing expression of DDC in individual scales (stage IV). Note the high Ddc signal strength in the presumptive yellow scales and lack of signal in those destined to be black. (C) Yellow-black boundary in forewing. (D) Presumptive yellow terminal band on forewing (stage IV). (E-F) In situ hybridisation of 33P-labelled antisense Ddc probes. Note early (stage IV) Ddc mRNA accumulation in the presumptive yellow pattern, including the yellow marginal bands (E), and later (stage II) accumulation in the presumptive black pattern (F), with a corresponding decrease in signal in the yellow marginal bands.

females. The presence or absence of DDC activity is therefore correlated with the decision of tissues to be either yellow or black. Thus, in melanics, the early abnormal absence of DDC activity is co-ordinately regulated with the subsequent atypical melanisation of the same formerly yellow area. Control of Ddc mRNA abundance is therefore a key component of the regulatory process underlying yellow/black colour formation. In this respect, while we are uncertain how Ddc transcription and/or message accumulation may be regulated in such a complex spatial and temporal pattern, we note that Ddc is regulated by ecdysteroids during melanin synthesis in both the butterfly P. coenia (Koch, 1995) and also the moth, M. sexta (Hiruma and Riddiford, 1993; Hiruma et al., 1995). It is therefore possible that the timing of DDC production in

WILD TYPE FORM EARLY LATE

MELANIC FORM EARLY LATE

PIGMENT COLOR AND DEPOSITION PATTERN

DDC activity tyrosine uptake β-alanine uptake tryptophan uptake

DDC activity tyrosine uptake tryptophan uptake

DDC activity tyrosine uptake β-alanine uptake tryptophan uptake

DDC activity tyrosine uptake tryptophan uptake

BIOSYNTHESIS ACTIVITY PATTERN

Fig. 7. Diagram illustrating a model of the developmental basis of melanism in P. glaucus. In wild type, early papiliochrome synthesis forms the yellow pattern and, later, the surrounding areas are melanised black. In the melanic females, this early papiliochrome synthesis is absent from the central presumptive yellow areas of the wings and is later replaced by atypical black melanin. Note that Ddc mRNA abundance correlates precisely with the two different patterns (Fig. 6) as DDC first supplies dopamine to the yellow papiliochrome and later to black melanin. In melanic females DCC enzyme activity in the central forewing is lower in the absence of papiliochrome synthesis. As DDC supplies dopamine to both the yellow and black pigment specific parts of the synthesis pathway (Fig. 3A), the model infers that DDC activity must be regulated in concert with other pigment specific elements of the pathway. Such a model, where a single Y linked factor coordinately regulates a number of synthetic components, helps explain how a complex developmental trait could be controlled by a single key genetic determinant.

2312 P. B. Koch and others different parts of the wing is coordinated by differing levels of ecdysteroids during different stages of wing development. Some of our observations also suggest that selective precursor uptake contributes to pattern- and stage-specific pigment deposition. For example, we note that DDC enzyme activity remains high in yellow areas of the wing towards adult eclosion, even though Ddc mRNA synthesis has ceased in the area at this time and radiolabelled tyrosine is not incorporated. This observation would be consistent with tyrosine not being available in yellow scales for papiliochrome synthesis at the time when tyrosine is being incorporated into black melanin. Two additional points on pattern-specific incorporation and scale cell precursor uptake are of note which both relate to tryptophan. Firstly, tryptophan was incorporated into the black areas of isolated wings in a poorly understood, but previously documented mechanism (Koch, 1991). Secondly in contrast to nymphalid butterflies, where the scale cells can directly metabolise tryptophan (Koch, 1991), in P. glaucus tryptophan was incorporated in vivo but not in vitro (Figs 3D, 4C) showing that the scale cells are not capable of metabolising tryptophan to kynurenine. This is consistent with previous observations from P. xuthus (Ishizaki and Umebachi, 1984, 1988; Ishizaki and Umebachi, 1990) which predict that kynurenine, as well as β-alanine and tyrosine-glucoside, are precursors present in the swallowtail hemolymph (Ishizaki and Umebachi, 1984) which are later taken up by the scale cells for papiliochrome synthesis. A model for the genetic basis of melanism in P. glaucus In order to combine these observations on the differential expression of DDC activity into a working model of the genetic basis of melanism in P. glaucus, we must first consider previous classifications of melanism in Lepidoptera. Previous work has stressed three alternative models: (1) darkening of the background colour, (2) broadening of dark pattern elements or (3) darkening of light coloured pattern elements (Nijhout, 1991). In P. glaucus, prior analysis of the black melanic females emphasised a type 1 mechanism, i.e. ‘darkening of the normal background colour (yellow)’ (Nijhout, 1991). This conclusion is re-inforced by the observation that the black tiger stripes of the swallowtail remain visible underneath the melanisation of the central wing areas (Nijhout, 1991) and also because the degree of melanism can show variable penetrance (Ritland, 1986). This description therefore suggests that black melanin may be superimposed over the yellow papiliochrome. In this study, we were therefore interested in determining if the developmental pathways resulting in black are superimposed on those forming yellow, or whether they actually replace them. Our data on the spatial and temporal regulation of Ddc expression during pigment synthesis suggests that black is actually replacing the yellow in the central proximal pattern element. These observations help explain how different pigments can be laid down at different locations by the same enzyme. However, they cannot fully explain the molecular basis of melanism, as DDC expression alone does not appear sufficient to drive pigment synthesis towards black. A model for the molecular basis of melanism must therefore explain how it is that dopamine is produced both in presumptive yellow and black tissues and how, following the decision of a scale cell to

be either yellow or black, dopamine is then driven either into the yellow papiliochrome pathway or into black melanin. In theory, there are a number of alternative mechanisms by which such specific colour choices could occur. Firstly, specific cells could show unique expression of colour-specific synthetic enzymes. For example, N-β-alanyl-transferase, the enzyme that shunts dopamine into the papiliochrome pathway (Fig. 3A), may be expressed exclusively in presumptive yellow cells. Similarly, the enzyme catalysing conversion of dopamine to dopamine quinone, the first melanin-specific step in the pathway, could be expressed exclusively in areas destined to be black. Secondly and alternatively, the complete complement of enzymes required to make one of the pigments might be expressed in both tissue types, making each competent to produce that pigment. Pattern-specific expression of enzymes for the alternative pigment would then shunt all available substrate into a secondary pigment-specific pathway only in the areas in which they are produced. Thirdly, all the enzymes involved in papiliochrome synthesis could be exclusively expressed early in both tissue types, while those involved in melanisation would be expressed in both tissues late. In this last hypothesis, each separate colour pathway would respond to the spatially and temporally differentiated pool of dopamine provided by differential expression of Ddc. Although the current data do not allow us to distinguish between these possibilities, it is significant that all of the alternatives require coordinated regulation of additional steps in pigment biosynthesis alongside DDC activity. Further, the observation that the switch that determines yellow versus black colour is always both complete and precise (intermediate scale colours are not seen) suggests that a single developmental system must be coordinately regulating all the synthetic enzymes involved in the colour decision. Thus, we propose that scale cell colour determination involves developmental changes leading to the regulation of multiple synthetic enzymes in concert. Our observations of adult P. glaucus pattern phenotypes, the biochemical data and the spatial and temporal control of Ddc expression (summarised in Fig. 7) suggest a model in which regulation of DDC activity plays a key, but not exclusive, role in yellow/black patterning. Although in the context of the present study it is premature to directly correlate previously described genetic loci with those examined here, we note that the hypothesis of a single sex linked factor (b) and an associated modifier (s) (Scriber et al., 1996), is consistent with our theory that there is a single genetic factor coordinately regulating a range of different synthetic enzymes. Although there are a number of different genes in the melanin synthesis pathway whose deregulation could lead to atypical melanisation of wing patterns, we suggest that the early lack of underlying yellow pigment (suppression of papiliochrome synthesis) is a critical observation which is subsequently correlated with the later atypical melanisation of the presumptive yellow pattern. One of the coordinately regulated enzymes may therefore be N-βalanyl-transferase which shunts dopamine out of the melanin pathway and into papiliochrome synthesis (Hopkins and Kramer, 1992, Umebachi, 1993). In this context, we also note that this gene product is a probable homolog of that encoded by Drosophila ebony (Caizzi et al., 1987), mutants of which show black body colour. Early suppression of N-β-alanyl-

Wing colour pattern formation in butterflies 2313 transferase, in concert with reduced DDC activity, could therefore lead to failure of the initiation of the papiliochrome specific synthesis pathway. It will therefore be interesting to investigate which genes in addition to Ddc are regulated, and to identify the earliest regulated components of pigment synthesis. We thank John True, David Lewis and Ron Galant for critical reading of the manuscript, all members of the laboratory of S. B. C. for their help and support, and Philip Holzbauer for provision of pupae.

REFERENCES Bergander, S., Koch, P. B., Spelz, S. and Fuchs, U. (1996). Pattern specific melanin synthesis in butterfly wings is controlled by ecdysteroid dependent expression of dopadecarboxylase. Verh. Dtsch. Zool. Ges., 89, 324. Brower, J. V. Z. (1958). Experimental studies of mimicry in some North American butterflies II. Battus philenor and P. glaucus. Evolution 12, 123136. Caizzi, R., Ritossa, F., Ryseck, R.-P., Richter, S. and Hovemann, B. (1987). Characterization of the ebony locus in Drosophila melanogaster. Mol. Gen. Genet. 206, 66-70. Carroll, S. B., Keys, D. N., Paddock, S. W., Panganiban, G. E. F., Selegue, J. E. and Williams, J. A. (1994). Pattern formation and eyespot determination in butterfly wings. Science 265, 109-114. Clarke, C. A. and Clarke, F. M. M. (1983). Abnormalities of wing pattern in the Eastern Tiger Swallowtail butterfly, Papilio glaucus. System. Entomol. 8, 25-28. Clarke, C. A. and Sheppard, P. M. (1959). The genetics of some mimetic forms of Papilio glaucus. Linn. J. Genet. 56, 236-260. Clarke, C. A. and Sheppard, P. M. (1962). The genetics of the mimetic butterfly Papilio glaucus. Ecology 43, 159-161. Hiruma, K., Carter, M. S. and Riddiford, L. M. (1995). Characterization of the dopa decarboxylase gene of Manduca sexta and its suppression by 20hydroxyecdysone. Dev. Biol. 169, 195-209. Hiruma, K. and Riddiford, L. M. (1993). Molecular mechanisms of cuticular melanization in the tobacco hornworm, Manduca sexta (L.) (Lepidoptera: Sphingidae). Int. J. Insect Morphol. Embryol. 22, 103-117. Hopkins, T. L. and Kramer, K. J. (1992). Insect cuticle sclerotization. Annu. Rev. Entomol. 37, 273-302. Ishizaki, Y. and Umebachi, Y. (1984). Synthesis, accumulation and excretion of kynurenine during the pupal and adult stages of Papilio xuthus. Insect Biochem. 14, 563-567. Ishizaki, Y. and Umebachi, Y. (1988). Level changes of β-alanine, dopamine,

and N-β-alanyldopamine during the pupal stage of Papilio xuthus (Lepidoptera: Papilionidae). Comp. Biochem. Physiol. 90C, 83-87. Ishizaki, Y. and Umebachi, Y. (1990). Further studies on dopamine and Nacetyl-dopamine during the pupal stage of Papilio xuthus (Lepidoptera: Papilionidae). Comp. Biochem. Physiol. 97B, 563-567. Kettlewell, B. (1973) Evolution of Melanism. Oxford: Clarendon Press. Koch, P. B. (1991). Precursors of pattern specific ommatin in red wing scales of the polyphenic butterfly Araschnia levana L.: haemolymph tryptophan and 3-hydroxykynurenine. Insect Biochem. 21, 785-794. Koch, P. B. (1992). Seasonal polyphenism in butterflies: a hormonally controlled phenomenon of pattern formation. Zool. Jb. Physiol. 96, 227240. Koch, P. B. (1993). Production of [14C]-labeled 3-hydroxy-L-kynurenine in a butterfly, Heliconius charitonia L. (Heliconidae), and precursor studies in butterfly wing ommatins. Pigment Cell Res. 6, 85-90. Koch, P. B. (1994). Wings of the butterfly Precis coenia synthesize dopamine melanin by selective activity of dopadecarboxylase. Naturwissenschaften 81, 36-38. Koch, P. B. (1995). Colour pattern specific melanin synthesis is controlled by ecdysteroids via dopa decarboxylase in wings of Precis coenia (Lepidoptera: Nymphalidae). Eur. J. Entomol. 92, 161-167. Koch, P. B. and Kaufmann, N. (1995). Pattern specific melanin synthesis and DOPA decarboxylase activity in a butterfly wing of Precis coenia Hubner. Insect. Biochem. Molec. Biol. 25, 73-82. Koch, P. B. and Nijhout, H. F. (1990). Color pattern specific proteins in black scales in developing wings of Precis coenia Hubner (Nymphalidae, Lepidoptera). Roux’s Arch. Dev. Biol. 199, 289-294. McCaman, M. W., McCaman, R. E. and Lees, G. J. (1972). Liquid cation exchange - a basis for sensitive radiometric assays for aromatic amino acid decarboxylases. Anal. Biochem. 45, 242-252. Nijhout, H. F. (1980). Ontogeny of the color pattern on the wings of Precis coenia (Lepidoptera: Nymphalidae). Dev. Biol. 80, 275-288. Nijhout, H. F. (1991). The Development and Evolution of Butterfly Wing Patterns. (ed. V. A. Funk and P. F. Cannell), p. 297. Smithsonian Series in Comparative Evolutionary Biology, Washington and London: Smithsonian Institution Press. Ritland, D. B. (1986). The effect of temperature on the expression of the dark morph phenotype in female Papilio glaucus (Papilionidae). J. Res. Lepidopt. 25, 179-187. Scott, J. A. (1986) The Butterflies of North America. p. 583. Stanford: Stanford University Press. Scriber, J. M., Hagen, R. H. and Lederhouse, R. C. (1996). Genetics of mimicry in the tiger swallowtail butterflies, Papilio glaucus and P. canadensis (Lepidoptera: Papilionidae). Evolution 50, 222-236. Umebachi, Y. (1985). Papiliochrome, a new pigment group of butterfly. Zool. Sci. 2, 163-174. Umebachi, Y. (1993). The third way of dopamine. Trends Comparat. Biochem. Physiol. 1, 709-720.