Redox alters yellow dragonflies into redRyo Futahashia,1, Ryoji Kuritab, Hiroaki Manoc, and Takema Fukatsua

aBioproduction Research Institute and bBiomedical Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba 305-8566, Japan; and cDivision of Evolutionary Biology, National Institute for Basic Biology, Okazaki 444-8585, Japan

Edited by David L. Denlinger, Ohio State University, Columbus, OH, and approved June 6, 2012 (received for review April 26, 2012)

Body color change associated with sexual maturation—so-callednuptial coloration—is commonly found in diverse vertebratesand invertebrates, and plays important roles for their reproduc-tive success. In some dragonflies, whereas females and youngmales are yellowish in color, aged males turn vivid red upon sex-ual maturation. Themale-specific coloration plays pivotal roles in,for example, mating and territoriality, but molecular basis of thesex-related transition in body coloration of the dragonflies hasbeen poorly understood. Here we demonstrate that yellow/redcolor changes in the dragonflies are regulated by redox states ofepidermal ommochrome pigments. Ratios of reduced-form pig-ments to oxidized-form pigments were significantly higher inred mature males than yellow females and immature males.The ommochrome pigments extracted from the dragonflieschanged color according to redox conditions in vitro: from redto yellow in the presence of oxidant and from yellow to red inthe presence of reductant. By injecting the reductant solution in-to live insects, the yellow-to-red color changewas experimentallyreproduced in vivo in immature males and mature females. Dis-continuous yellow/red mosaicism was observed in body colorationof gynandromorphic dragonflies, suggesting a cell-autonomousregulation over the redox states of the ommochrome pigments.Our finding extends the mechanical repertoire of pigment-basedbody color change in animals, and highlights an impressively sim-ple molecular mechanism that regulates an ecologically importantcolor trait.

red dragonflies | nuptial color change | redox-dependent color change |Sympetrum | Crocothemis

Nuptial color change, the body color transition associatedwith sexual maturation, is commonly found in diverse ani-

mals such as mammals, birds, reptiles, amphibians, fish, andinsects (1–4). The distinct color patterns reflecting their re-productive mode must have been shaped by sexual selection, andgenerally play important roles for their reproductive success.Among insects, many dragonflies display remarkable nuptialcoloration (5). The sex-specific color patterns are generally im-portant for partner recognition before mating (5, 6). Territorialmales visually recognize vividly colored invader males and ag-gressively attack them, whereas they permit invasion of dull-colored females as potential mates (7). However, the molecularmechanisms of the sex-related transition in body coloration havebeen poorly understood.In many red dragonflies of the genera Crocothemis and Sym-

petrum, young males are yellow in color and turn red upon sexualmaturation, whereas females remain yellowish throughout theirlifetime (Fig. 1). Here we report that the yellow-to-red colorchange in the dragonflies is regulated by redox states of epidermalommochrome pigments, which extends the mechanical repertoireof pigment-based body color change in animals and highlightsan impressively simple molecular mechanism that regulates theecologically important external trait.

Results and DiscussionIdentification of Red-Yellow Pigments of Dragonflies. We attemptedto extract the red-yellow pigments from abdominal epidermis ofthe red dragonflies Crocothemis servilia, Sympetrum darwinianum,and Sympetrum frequens. The pigments were efficiently recovered

by 0.5% hydrochloric acid in methanol, which has been often usedas solvent for ommochrome pigments (8, 9). By using HPLC andMS, we analyzed the epidermal pigments extracted from sexuallymature males and females of C. servilia, S. darwinianum, andS. frequens, and identified two ommochrome pigments, xan-thommatin and decarboxylated xanthommatin, irrespective of sexand species of the dragonflies (Fig. 2 and Fig. S1). Cinnabar red-colored males of S. darwinianum and S. frequens, and yellow-colored females of all three species, contained decarboxylatedxanthommatin as the major pigment (Fig. 2 B–F). Meanwhile,crimson red-colored males of C. servilia contained two majorpigments, xanthommatin in addition to decarboxylated xan-thommatin (Fig. 2A).

Redox-Dependent Color Change of Ommochrome Pigments ofDragonflies. Ommochrome pigments generally change their colorunder oxidative/reductive conditions (Fig. S2A) (8). We confirmedredox-dependent color changes of the pigments extracted from thedragonflies in vitro: the red pigment extracts from sexually maturemales turned yellow by an addition of oxidant (NaNO2; Fig. 3A),whereas the yellow pigment extracts from immature males turnedred by an addition of reductant (ascorbic acid; Fig. 3B). Chemicallysynthesized decarboxylated xanthommatin and xanthommatin ex-hibited similar redox-dependent color changes: red under the re-ductive condition and yellow under the oxidative condition (Fig. 3C and D). The dull red color of reduced decarboxylated xan-thommatin agrees with the cinnabar-red color of mature malesof S. darwinianum and S. frequens, whereas the vivid red color ofreduced xanthommatin accounts for the crimson-red color of ma-ture males of C. servilia (Figs. 2 and 3 C and D).

Redox-Dependent Color Change of Dragonflies in Vivo. By injectingascorbic acid solution into the abdomen of the dragonflies, weconfirmed a similar color change in vivo: not only yellow im-mature males but also yellow mature females changed their colorto red within several hours after the reductant injection (Fig. 3 Eand F). On the contrary, by injecting an oxidant solution into theabdomen of mature males, a red-to-yellow color change wasobserved but only very subtly and locally, and the insects wereseverely damaged and usually died soon (Fig. S2B).

Direct Measurement of Redox Status of Ommochrome Pigments inDragonflies. Redox conditions of a soluble substance can bemeasured by using an electrochemical method with a rotatingworking electrode by applying lower and higher voltages than theredox potential of the soluble substance (Fig. S2C). By using thistechnique (10), we electrochemically evaluated the relativeabundance of the oxidized and reduced forms of the ommo-chromes in the pigment extracts from the dragonflies. In all three

Author contributions: R.F. and T.F. designed research; R.F., R.K., and H.M. performedresearch; R.F. analyzed data; and R.F. and T.F. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1To whom correspondence should be addressed. E-mail: ryo-futahashi@aist.go.jp.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1207114109/-/DCSupplemental.

12626–12631 | PNAS | July 31, 2012 | vol. 109 | no. 31 www.pnas.org/cgi/doi/10.1073/pnas.1207114109

Dow

nloa

ded

by g

uest

on

Apr

il 13

, 202

0

species, the levels of the red reduced form were significantlyhigher in the mature males than in the immature males andfemales: the reduced form ratios were from 90% to 100% inmature males, whereas the values were approximately 55% to75% in mature females and also in immature males and females(Fig. 3 G–I).

Analysis of Water-Soluble Reductants. We examined reduction ac-tivities in water extracts of the male and female dragonflies. Al-though ommochrome pigments are poorly water-soluble, mostommochromes are assumed to be associated with specific bindingproteins in vivo (11), and some of the complexes are water-solubleto some extent (12). Accordingly, water extracts of mature maledragonflies were more reddish in color than those of femaledragonflies (Fig. 4 A and B), and exhibited stronger reductionactivities than those of female dragonflies (Fig. 4C). Addition ofascorbate oxidase did not affect the reduction activities in thewater extracts of the dragonflies (Fig. 4 D and E), indicating thatthe reduction activities are not attributable to ascorbic acid in-herently present in the dragonflies. HPLC analysis of the waterextracts from the male dragonflies revealed that, of five peakfractions collected (Fig. 4F), the reduction activities were re-covered in fraction 4, representing decarboxylated xanthomma-tin, and fraction 5, corresponding to xanthommatin (Fig. 4G).

Possible Mechanisms Underlying Sex-Specific Redox Change ofOmmochrome Pigments. Conceptually, male-specific accumula-tion of natural reductants such as ascorbic acidmay account for themale-specific presence of the reduced ommochrome pigments (13).However, our experimental data refuted this possibility: we couldnot detect any natural reductants other than the pigments them-selves in the water extracts from the male dragonflies (Fig. 4). Theeyes of the fruit fly Drosophila melanogaster contain a high amountof reduced form of xanthommatin, in which a significant activity ofxanthommatin reductase was detected (14), although reductasegenes responsible for the activity have not yet been characterized.In the wings of Heliconius butterflies, ommochrome-binding pro-teins have been hypothesized to stabilize the redox status of thepigments (15). It is conceivable, although speculative, that male-specific enrichment of reductases and/or pigment-binding proteins

Fig. 1. Stage- and sex-specific adult color change in dragonflies. (A) C.servilia immature male (Left), mature male (Center), and mature female(Right). (B) Adult males and females of three red dragonflies C. servilia, S.darwinianum, and S. frequens.

Fig. 2. Identification of ommochrome pigments from dragonflies. (A–F) Chromatograms of ommochrome pigments from males and females of three reddragonflies. Blue lines denote the acetonitrile gradient. (G and H) Electrospray ionization mass spectra of synthesized pigments. (G) Decarboxylated xan-thommatin. (H) Xanthommatin.

Futahashi et al. PNAS | July 31, 2012 | vol. 109 | no. 31 | 12627

EVOLU

TION

Dow

nloa

ded

by g

uest

on

Apr

il 13

, 202

0

may be involved in the redox-dependent color differences in themale and female dragonflies.

Insight from Gynandromorphic Dragonflies. In the abdomen of thedragonflies, the red/yellow pigments are localized within epidermalcells as pigment granules (8, 16) (Fig. 5 A and B). Gynandromor-phic dragonflies have been recorded occasionally (17, 18), and theyconsistently exhibit discontinuous yellow/red mosaicism in theirbody coloration (Fig. 5 C and D). These observations suggest thepossibility that a cell-autonomous mechanism is involved in theregulation of the dragonfly body color, which may entail, for ex-ample, a sex-specific cellular response to hormonal signals.

Conclusion and Perspective. In conclusion, the sex-specific matu-ration color change in dragonflies is regulated by redox states ofthe ommochrome pigments. Pigment-based color changes areknown from diverse animals, which have been attributed tovarious mechanisms such as de novo pigment synthesis, pigmentdegradation, change in pigment localization, and accumulationof food pigments (19). Our finding may extend the mechanicalrepertoire of the pigment-based body color change, which maybe operating not only in dragonflies but also in butterflies andother organisms (15). In Japan, appearance of red dragonflieshas been regarded as a symbol of seasonal change, giving a

natural feeling of summer/autumn transition adopted by nu-merous versicles, arts, songs, and paintings (5, 20, 21) (Fig. S3).This study highlights a simple molecular mechanism underlyingthe ecologically important, as well as aesthetically impressive,biological phenomenon.

Materials and MethodsExperimental Animals and Chemical Treatment. C. servilia, S. darwinianum,and S. frequens were collected in Tsukuba, Ibaraki Prefecture, Japan.S. darwinianum and S. frequens start to emerge from June to July, andmature males become reddish at the beginning of September and disappearin December. C. servilia emerges from May to October, and adult males turnred several days after emergence. For pigment extraction, abdominal epi-dermis was dissected from a single adult insect and extracted with 1 mL of0.5% hydrochloric acid in methanol. In acidic methanol, the redox states ofthe ommochrome pigments were stable, and the color was maintained atleast for 6 mo when stored at −30 °C. The pigments were reduced or oxi-dized by adding a few drops of 1% ascorbic acid (Sigma) or 1% NaNO2

(Wako), respectively. For the redox-dependent color change analyses in vivo,ascorbic acid (Sigma) and NaNO2 was dissolved in distilled water at 0.1 mg/mL. The reductant/oxidant solution (5 µL each) was microinjected into eachlive insect abdomen (Fig. 3 E and F and Fig. S2B, arrows) using a glass cap-illary. For histological observation, adult abdominal epidermis was fixedovernight in freshly prepared 4% (wt/vol) paraformaldehyde in PBS solutionat 4 °C. After dehydration by acetone, the tissue was embedded in paraffin,

Fig. 3. Redox-dependent color change of the ommochrome pigments. (A and B) Pigment extract from a mature male and an immature male of C. servilia. (Cand D) Reduced and oxidized forms of synthetic decarboxylated xanthommatin and xanthommatin. (E and F) Reductant-induced yellow/red color changein vivo. Arrows indicate the injection sites. (E) An immature male of S. darwinianum. (F) A mature female of C. servilia. (G–I) Reduced form ratios of theextracted ommochrome pigments: (G) C. servilia, (H) S. darwinianum, and (I) S. frequens. Statistically significant differences are observed between barsmarked “a” and “b” as shown (Games–Howell test). Means and 95% confidence intervals are shown.

12628 | www.pnas.org/cgi/doi/10.1073/pnas.1207114109 Futahashi et al.

Dow

nloa

ded

by g

uest

on

Apr

il 13

, 202

0

Fig. 4. Analysis of water-soluble reductants in red dragonflies. (A) Abdominal epidermis of a male and a female of C. servilia used in the analysis. (B)Water extracts of a male and a female of C. servilia. (C) Linear sweep voltammograms of the water extracts from a male and a female of C. servilia. (D)Quantification of reduction activities using standard ascorbic acid solutions (0, 2, 4, 6, and 8 nmol/µL) as positive control. Lower: Samples supplementedwith ascorbate oxidase. (E ) Quantification of reduction activities in water extracts of a male and a female of C. servilia. Abdominal epidermis was dissectedfrom a single adult insect and extracted in 0.1 mL, 0.2 mL (1/2), and 1 mL (1/10) of distilled water. Lower: Samples supplemented with ascorbate oxidase. (F )Chromatograms of water extracts of a male and a female of C. servilia. Fraction numbers are shown beneath the chromatogram. (G) Quantification ofreduction activities in the five fractions of the water extract from a male of C. servilia. Fractions 4 and 5 correspond to decarboxylated xanthommatin andxanthommatin, respectively.

Futahashi et al. PNAS | July 31, 2012 | vol. 109 | no. 31 | 12629

EVOLU

TION

Dow

nloa

ded

by g

uest

on

Apr

il 13

, 202

0

processed into 5-μm serial tissue sections, mounted on glass slides, and ob-served under a light microscope without staining.

Liquid Chromatography and MS. Abdominal epidermis was dissected froma single adult insect, and pigments were extracted with 1 mL of 0.5%hydrochloric acid in methanol. After centrifugation, the supernatant wassubjected to reversed-phase HPLC directly. HPLC analysis was conducted byusing a Waters Alliance 2695 HPLC system fitted with a Gemini-NX C18column (150 × 4.6 mm, 3 μm particle size; Phenomenex). The absorbancedata were collected with a Waters 2487 dual-wavelength absorbance de-tector. The elution gradient was based on a ternary solvent system [time in

min [% solvent A/% solvent B/% solvent C]: 0 min (80/10/10); 5 min (30/60/10); 6 min (80/10/10); and 20 min (80/10/10)]. Solvent A consisted of 0.1%trifluoroacetic acid (TFA) in water, solvent B consisted of 0.1% TFA inacetonitrile, and solvent C consisted of 10 mM ascorbic acid in 0.1% TFAwater. For HPLC analysis of water extracts, ascorbic acid was not used inmobile phase [0 min (90/10/0); 5 min (40/60/0); 6 min (90/10/0); and 20 min(90/10/0)]. The column temperature was set at 30 °C, and the flow rate wasconstant at 1 mL/min. The peak fractions were monitored based on 450 nm(for pigment extract) or 250 nm (for water extract) absorbance and col-lected manually. The collected pigment fractions were subjected toMALDI-TOF MS. The mass spectra were recorded on a Reflex III system(Bruker Daltonik) by using α-cyano-4-hydroxycinnamic acid and 2,5-dihy-droxybenzoic acid as matrices. Standard samples of xanthommatin anddecarboxylated xanthommatin were chemically synthesized as describedpreviously (22). Standard samples were verified by mass measurementsusing MALDI-TOF MS and electrospray ionization MS (23). Electrosprayionization mass spectra were recorded with an LCQ Finnigan spectrometer(LCQ Duo; Thermo Finnigan).

Measurement of Oxidation/Reduction Current. Abdominal epidermis was dis-sected from a single adult insect and pigments were extracted with 1 mL of0.5% hydrochloric acid in methanol or 1 mL of distilled water. After cen-trifugation, the supernatant was collected and 1 mL of distilled water wasadded just before measurement (within 1 min). The diluted sample waselectrochemically analyzed (10) as follows. A three-electrode electrochemicalcell with a glassy carbon disk (3 mm in diameter) as a working electrode, aplatinum wire as a counter electrode, and Ag-AgCl (3 M NaCl) as a referenceelectrode were used. Before electroanalysis, the glassy carbon electrode waspolished in Al2O3/water slurry, and then voltammetry was conducted inblank buffer (0.25% hydrochloric acid in 50% methanol) between −0.2 Vand 0.6 V until a stable voltammetric response was obtained. The workingelectrode was rotated at 1,500 rpm by a rotating electrode system (modelRRDE-3A; ALS), and a linear sweep voltammogram was obtained by scanningof electrode potential from −0.2 V to 0.6 V at a scan rate of 5 mV/s by usinga potentiostat (model 900; CH Instruments). The reduction current and ox-idation current were calculated at 0.1 V and 0.45 V by subtracting reduction/oxidation currents of blank buffer from those of the sample, respectively.

Measurement of Reduction Activity. Reduction activity was measured by usingan Ascorbic Acid Assay Kit II (Bio-Vision) according to the manufacturer’sinstructions. Abdominal epidermis was dissected from a single adult insectand extracted in 0.1 mL, 0.2 mL (Fig. 4E, “1/2”), and 1 mL (Fig. 4E, “1/10”) ofdistilled water. Each extract was divided into two samples, and backgroundvalue was calculated by adding ascorbate oxidase to one sample. Standardascorbic acid solutions (0, 2, 4, 6, and 8 nmol/µL) were used as positivecontrol. Fe+3 is reduced to Fe+2 by any antioxidants present, and the re-sultant Fe+2 was chelated with a colorimetric probe, thereby exhibiting anintense blue color, which can be monitored by measuring the change inabsorption at 593 nm.

ACKNOWLEDGMENTS. We thank M. Umemura, Y. Makino, H. Sawada, andY. Oba for technical support; N. Ishizawa and I. Kawashima for photos ofgynandromorphic dragonflies; K. Okubo and S. Okubo for photos ofdragonfly arts and products; and M. Moriyama and Y. Matsuura for insectsamples. This work was supported by Japan Society for the Promotion ofScienceGrant-in-Aid for Scientific Research Grant 23780058 (to R.F.) andMinistry of Education, Culture, Sports, Science and Technology Grant-in-Aidfor Scientific Research Grant 22128007 (to T.F.). This study was carried outunder the National Institute for Basic Biology Cooperative Research Program(project no. 11-373).

1. Leonard J, Cordoba-Aquilar A (2010) The Evolution of Primary Sexual Characters inAnimals (Oxford Univ Press, London).

2. van der Sluijs I (2008) Divergent Mating Preferences and Nuptial Coloration in SiblingSpecies of Cichlid Fish (Leiden Univ Press, Leiden, The Netherlands).

3. Olsson M (1994) Nuptial coloration in the sand lizard, Lacerta agilis - an intra-sexuallyselected cue to fighting ability. Anim Behav 48:607–613.

4. Hill GE, McGraw KJ (2006) Bird Coloration, Volume 2: Function and Evolution (HarvardUniv Press, Cambridge, MA).

5. Corbet PS (1999) Dragonflies: Behavior and Ecology of Odonata (Cornell Univ Press,New York).

6. Cordoba-Aquilar A (2008) Dragonflies & Damselflies: Model Organisms for Ecologicaland Evolutionary Research (Oxford Univ Press, London).

7. Frantsevich LI, Mokrushov PA (1984) Visual stimuli releasing attack of a territorialmale in Sympetrum (Anisoptera: Libellulidae). Odonatologica 13:335–350.

8. Linzen B (1974) The tryptophan to ommochrome pathway in insects. Adv InsectPhysiol 10:117–246.

9. Riou M, Christidès JP (2010) Cryptic color change in a crab spider (Misumena vatia):Identification and quantification of precursors and ommochrome pigments by HPLC.J Chem Ecol 36:412–423.

10. Bard AJ, Faulkner LR (1980) Electrochemical Methods: Fundamentals and Applications(Wiley, New York).

11. Sawada H, Nakagoshi M, Mase K, Yamamoto T (2000) Occurrence of ommo-chrome-containing pigment granules in the central nervous system of thesilkworm, Bombyx mori. Comp Biochem Physiol B Biochem Mol Biol 125:421–428.

12. Maki N, Kajiura Z, Nakagaki M, Takei R (1995) Purification and identification of serumred pigment–protein complex from the rb mutant of the silkworm, Bombyx mori.J Seric Sci Jpn 64:46–55.

Fig. 5. Localization of the ommochrome pigments in dragonflies. (A andB) Intracellular localization of the ommochrome pigments in the abdomi-nal epidermis. (A) Tissue section of a mature male of S. darwinianum. (B)Tissue section of a mature female of S. darwinianum. (C ) Gynandromorphof S. frequens showing the left side with male coloration and the right sidewith female coloration (photo courtesy of Naoya Ishizawa). (D) Gynandro-morph of S. parvulum showing patches of red and yellow coloration (photocourtesy of Itsuro Kawashima). Insets: Magnified images of abdomens.

12630 | www.pnas.org/cgi/doi/10.1073/pnas.1207114109 Futahashi et al.

Dow

nloa

ded

by g

uest

on

Apr

il 13

, 202

0

13. Jimenez A, et al. (1998) Role of the ascorbate-glutathione cycle of mitochondria andperoxisomes in the senescence of pea leaves. Plant Physiol 118:1327–1335.

14. Santoro P, Parisi G (1986) A new enzyme from Drosophila melanogaster – in vitroconversion of xanthommatin into its dihydro form by means of xanthommatin re-ductase. J Exp Zool 239:169–173.

15. Gilbert LE, Forrest HS, Schultz TD, Harvey DJ (1988) Correlations of ultrastructure andpigmentation suggest how genes control development of wing scales of Heliconiusbutterflies. J Res Lepid 26:141–160.

16. Prum RO, Cole JA, Torres RH (2004) Blue integumentary structural colours in drag-onflies (Odonata) are not produced by incoherent Tyndall scattering. J Exp Biol 207:3999–4009.

17. Ishizawa N (1990) A gynandromorphic specimen of Sympetrum frequens Selys. Tombo33:37–39.

18. Torralba-Burrial A, Ocharan FJ (2009) Two gynandromorphs of Sympetrum striolatum(Charpentier, 1840) (Odonata: Libellulidae). Entomol Sci 12:182–187.

19. Stevens M, Merilaita S (2011) Animal Camouflage: Mechanisms and Function (Cam-bridge Univ Press, London).

20. Inoue K, Tani K (2010) All About Red Dragonflies (Tombow, Osaka).21. Ueda T (2004) How Do the Japanese See Dragonflies? (Kyoto Univ Press, Kyoto).22. Butenandt A, Schiedt U, Biekert E (1954) Uber ommochrome. III. Synthese des xan-

thommatins. Liebigs Ann Chem 588:106–116.23. Iwahashi H, Ishii T (1997) Detection of the oxidative products of 3-hydroxykynurenine

using high-performance liquid chromatography–electrochemical detection–ultravio-let absorption detection–electron spin resonance spectrometry and high-perfor-mance liquid chromatography–electrochemical detection–ultraviolet absorptiondetection–mass spectrometry. J Chromatogr A 773:23–31.

Futahashi et al. PNAS | July 31, 2012 | vol. 109 | no. 31 | 12631

EVOLU

TION

Dow

nloa

ded

by g

uest

on

Apr

il 13

, 202

0