RESEARCH ARTICLES

Characterization of the Long-Wavelength Opsin from Mecoptera andSiphonaptera: Does a Flea See?

Sean D. Taylor,* Katharina Dittmar de la Cruz,* Megan L. Porter,� and Michael F. Whiting**Brigham Young University, Department of Integrative Biology; and �Brigham Young University, Department of Microbiology andMolecular Biology

Mecoptera and Siphonaptera represent two insect orders that have largely been overlooked in the study of insect vision.Recent phylogenetic evidence demonstrates that Mecoptera (scorpionflies) is paraphyletic, with the order Siphonaptera(fleas) nesting as sister to the family Boreidae (snow fleas), showing an evolutionary trend towards reduction in gross eyemorphology within fleas. We provide the first molecular characterization of long-wavelength opsins from these threelineages (opsin gene from fleas [FL-Opsin], the Boreidae [B-Opsin], and a mecopteran family [M-Opsin]) and assessthe effects of loss of visual acuity on the structure and function of the opsin gene. Phylogenetic analysis implies a phys-iological sensitivity in the red-green spectrum for these opsins. Analysis of intron splice sites reveals a high degree ofsimilarity between FL-Opsin and B-Opsin as well as conserved splice sites across insect blue-green and long-wavelengthopsins. Calculated rates of evolution and tests for destabilizing selection indicate that FL-Opsin, B-Opsin, andM-Opsin areevolving at similar rates with no radical selective pressures, implying conservative evolution and functional constraintacross all three lineages.

Introduction

Insects represent a group of organisms in which visualperception and the associated structural components arehighly developed (Briscoe and Chittka 2001). The classInsecta is one of the largest and most diverse classes oforganisms on the planet, accounting for over 70% ofall known species (Resh and Carde 2003). Insects haveadapted to nearly every ecological niche available, fromarctic to tropic and mountaintop to river bottom. However,despite this wide diversity of habitats and lifestyles, findinglinks in the adaptation of visual systems to visual ecologyhas proven to be difficult (Briscoe and Chittka 2001). It isimportant, therefore, to continue to characterize the visualsystem for a diversity of insect groups in order to gain abetter understanding of the selective pressures and con-straints that guide visual evolution in insects.

The photoreceptive cells in the insect compound eyecontain light-sensitive visual pigments, which consist of anopsin apoprotein, bound by a protonated Schiff’s base link-age to a photosensitive chromophore molecule, usually11-cis-retinal. The visual cascade is initiated when the chro-mophore absorbs a photon of light, causing it to isomerizeto the trans form. A corresponding conformational changein the opsin protein, a member of the G protein–coupledreceptors, initiates a transducin-mediated response, whichultimately sends a neural signal to the brain. For a reviewof opsin physiology, see Nathans (1987). A single individ-ual may possess several opsin variants, each sensitive todifferent wavelengths of light, thus conferring a specificrange of spectral sensitivity to that organism (Kochendoerferet al. 1999). Typically, insects possess trichromatic vision,conferred by at least three opsin classes: UV-, blue-, andgreen-sensitive variants. All insects studied have greenreceptors (;530 nm), most express UV (;330 nm) and

blue (;440 nm), and only a few have been shown to con-tain red (.565 nm) (Briscoe and Chittka 2001). Molecularcharacterizations and spectral tuning studies have been per-formed on opsin proteins found in many insect groups,including bees (Bellingham et al. 1997), moths and butter-flies (Kitamoto et al. 1998; Briscoe 2000), mantids (Townerand Gartner 1994), fireflies (Cronin et al. 2000), and fruitflies (Neufeld, Carthew, and Rubin 1991; Carulli et al.1994).

Mecoptera and Siphonaptera represent two insectgroups that have largely been overlooked in the study ofvision. Mecoptera is a relatively small, holometabolousinsect order with approximately 600 described species(Penny and Byers 1979). For purposes of this study, twofamilies are of particular interest. Panorpidae (true scor-pionflies) is the most speciose family of mecopterans. Theyhave a scavenging lifestyle, feeding primarily on dead, soft-bodied arthropods and often robbing spiders’ webs of theencased insects (Byers and Thornhill 1983). They havewell-developed eyes and are highly visually oriented, witha maximum spectral sensitivity peak at 490–520 nm and aminor peak at 360 nm (Burkhardt and De LaMotte 1972).Boreidae (snow fleas) are wingless, alpine insects that liveand feed exclusively on mosses and are most often collectedon winter snow (Penny 1977). The eyes of boreids aresmaller and appear slightly less well developed than thoseof panorpids. We have observed that boreids will oftenescape collection by jumping when they are too closelyapproached, indicating that vision plays an important rolein sensing the environment. Little is known, however, ofboreid visual ecology.

Siphonaptera (fleas) is a highly specialized holometab-olous insect order with approximately 2,380 described spe-cies (R. E. Lewis and J. H. Lewis 1985). Fleas are strictvertebrate ectoparasites and are of tremendous economicimportance as vectors of pathogens, transmitting diseasessuch as plague, murine typhus, and tularemia (Dunnetand Mardon 1991). Little is known about flea vision,due mainly to the radical divergence in flea eye structure

Key words: opsin, evolution, Mecoptera, Siphonaptera, insect vision.

E-mail: kd98@email.byu.edu.

Mol. Biol. Evol. 22(5):1165–1174. 2005doi:10.1093/molbev/msi110Advance Access publication February 9, 2005

Published by Oxford University Press 2005.

Dow

nloaded from https://academ

ic.oup.com/m

be/article/22/5/1165/1066918 by guest on 21 July 2022

from the typical insect bauplan (Crum, Knapp, and Wite1974). Fleas show a transformation of the multifacetedeyes and ommatidia of most insects, replaced instead withheavily sclerotized, atypical ocelli, or ‘‘eyespots,’’ or insome cases, a complete absence of any eye at all (Dunnetand Mardon 1991). The species used in our analysis exhibita diversity of eyespot morphologies, ranging from largeeyespots in Pulex irritans, to small eyespots in Ctenoph-thalmus agyrtes and Jellisonia sp., to no discernable eye-spot in Uropsylla tasmanica. Despite the drastic reductionin eye structure, Crum, Knapp, and Wite (1974) demonstra-ted that cat fleas (Ctenophalides felis) and oriental rat fleas(Xenopsylla cheopis) have peak light sensitivities at 330and 530 nm, respectively, although the light elicited differ-ent responses in the two species. Osbrink and Rust (1985)further demonstrated that visual cues seem to be among theprimary factors influencing host selection in cat fleas. How-ever, this eyespot probably cannot form an acute image butmerely acts as a light sensor (Osbrink and Rust 1985; Rustand Dryden 1997; Land 2003).

Whiting (2002) recently demonstrated that Mecopterais paraphyletic, with Siphonaptera nesting within Mecop-tera as sister group to Boreidae, and the obscure familyNannochoristidae placed as sister to Boreidae1 Siphonap-tera (fig. 1). This phylogenetic hypothesis provides an evo-lutionary framework in which to evaluate shifts in visualecology and search for possible correlates in morphologicaland molecular changes associated with vision in theseinsects. Given that nannochoristids are visual mecopterans,optimization of vision on this phylogeny implies that therewas a reduction of visual acuity in the flea lineage. We com-pare opsin structure and function in the ectoparasitic fleas,which have a single-lens light sensor and only general light-

dark sensitivity, with that of their boreid sister group, whichare detritus feeders and visually oriented insects. These inturn are compared with their highly visual panorpid rela-tives, which have a fully developed, image-forming eye.Because nannochoristids are rare insects, we have notincluded them in this study.

This study focuses on the effects of the shift to para-sitism in fleas, the associated loss of visual acuity, and themolecular processes underlying the evolution of the opsinprotein in these insects. Given the demonstrated photosen-sitivity of the fleas and scorpionflies to green light, we iso-lated the long-wavelength (LW) opsin gene from fleas (FL-Opsin), their close sister group, the Boreidae (B-Opsin),and the Panorpidae, a highly visual mecopteran family(M-Opsin). We provide the first molecular characterizationof insect opsins from these three groups. In addition, wecompare the gene structure and evolutionary history ofthese three opsin groups in order to assess the effects of lossof visual acuity on the structure and function of the opsingene. We specifically test for relaxation of functional con-straint on the FL-Opsin protein due to the gross reduction ineye structure and visual sensitivity.

Materials and MethodsDNA Analysis: Polymerase Chain Reaction andSequencing

LW opsin sequences were obtained from four flea spe-cies representing four families, six panorpid species repre-senting the single diverse genus Panorpa, and one boreidspecies. For a detailed list of all taxa included in the anal-ysis, see Appendix 1 (Supplementary Material online).

Genomic DNA was extracted from specimens usingthe Qiagen DNeasy extraction kit using manufacturer’sguidelines. Because of the low yield of DNA expected fromfleas, the DNA was eluted in only 100 ll of solution.Genomic DNA vouchers and specimen vouchers are depos-ited in the Insect Genomics Collection, M. L. BeanMuseum, Brigham Young University.

Polymerase chain reaction (PCR) primers were devel-oped from conserved regions of arthropod LW opsins and,once sufficient sequences were obtained, redesigned for usein Mecoptera and Siphonaptera. The amplified region spansthe first six of the transmembrane helices and;60% of theseventh. Primer sequences are given in table 1. PCR ampli-fication used the following three-step cycling protocol withAmplitaq Gold polymerase (ABI, Foster City, Calif.): 35cycles of denaturation at 94�C for 1 min, annealing at55�C for 1 min, and extension at 72�C for 3 min, followed

FIG. 1.—Condensed phylogeny of Mecoptera and Siphonaptera pre-sented by Whiting (2002) and eye morphology of Panorpidae, Boreidae,and Siphonaptera.

Table 1Primers Used to Amplify LW-Opsin

Primer Sequence (5# / 3#)

LF1 (fw) CAYTGGTAYCARTWYCCICCIATGBarney (fw) GAYMGITAYAAYGTIATIGTIAARGGRh6 5.1F (fw) GGMTGGAAYMGRTATGTWCCTGARGMike (fw) ATGMGIGAICARGCIAARAARATGRh6 5.1R (rv) CYTCAGGWACATAYCKRTTCCAKCCGoogle (rv) CATYTTYTTIGCYTGITCICKCATScylla (rv) TTRTAIACIGCRTTIGCYTTIGCRAA

1166 Taylor et al.

Dow

nloaded from https://academ

ic.oup.com/m

be/article/22/5/1165/1066918 by guest on 21 July 2022

by an additional 7 min of elongation at 72�C. PCR productswere visualized by gel electrophoresis and purified usingthe Millipore Montage purification system. Initial amplifi-cations were attempted on all taxa using LF1 and Scyllaprimers in order to obtain the entire target region in a singleamplification. If unsuccessful, smaller overlapping seg-ments were amplified with internal primers.

Because opsin variants arose through a series of geneduplication events (Deininger, Fuhrman, and Hegemann2000), the possibility exists that paralogous opsin variantsmay coamplify. In order to separate potential paralogouscopies, the PCR products were cloned using the pCR2.1-TOPO cloning kit (Invitrogen) following the manufac-turer’s protocol. Ten clones from each PCR product werepicked and then reamplified and sequenced using the pro-videdM13 vector primers (Invitrogen, Carlsbad, Calif.) andBigDye Terminator chemistry (ABI). Sequence data werecollected using the ABI Prism 3730 capillary autosequencerin the BYU DNA Sequencing Center.

The vector ends were trimmed from all cloned prod-ucts and the resulting sequence blasted in GenBank toinsure that an opsin product was indeed amplified. Clonesand overlapping PCR regions from the same taxa were ini-tially included in the phylogenetic analysis as separate ter-minals to verify that the sequences did not representparalogous gene copies. If they were shown to grouptogether, the sequences were then combined, with any dis-crepancies coded as ambiguous or missing data, andincluded as a single terminal in the final analysis.

Phylogenetic Analysis

Sequences from crustaceans and the LW, blue-green(BG), blue (BL), and UV insect opsin variants were down-loaded from GenBank and used to determine the phyloge-netic placement of FL-Opsin, M-Opsin, and B-Opsin (seeAppendix 1). The sequence of bovine rhodopsin, as theonly opsin to have its tertiary structure characterized,was included as an out-group. Initial alignments were per-formed manually in Sequencher 4.2 (GeneCodes 2003).Introns were identified through comparison to GenBankinsect cDNA sequences (fig. 2). The resulting codingsequences were translated using MacClade (W. P. Maddisonand D. R. Maddison 2003) and the amino acid sequencesaligned via ClustalX (Thompson et al. 1997). The align-ment was further refined by performing a homology align-ment of the Jellisonia sp. amino acid sequence (the mostcomplete opsin isolated in this study) to high-resolutionbovine opsin crystal structure (Palczewski et al. 2000) inthe Swiss-Pdb Viewer and submitted to the SwissModelServer for homology modeling (Guex and Peitsh 1997).This allowed us to optimize our overall nucleotide align-ment based on both amino acid sequence homology andstructural homology.

Phylogenetic analyses were performed using thenucleotide sequences after intron removal. The phylogenywas reconstructed using Bayesian methods coupled withMarkov chain Monte Carlo (BMCMC) techniques asimplemented in MrBayes v3.0b4, which allows for mixed

FIG. 2.—Comparison of FL-Opsin, M-Opsin, and B-Opsin intron splice sites to other insect opsins, based on Briscoe (1999). Numbers above hatchmarks indicate position of splice site relative toDrosophila melanogasterRh1 amino acid sequence. Numbers in parentheses indicate observed size (bp) ofintrons (shown only for LW opsins). Regions in black were not amplified in this analysis. Intron positions and lengths determined from alignment of ournovel opsin sequences with the following: Drosophila melanogaster Rh1 (X65877), Papilio glaucus Rh3 (AF098283), Bee LW1 (Bombus impatiens:AY485302, Bombus terrestris: AY485301, Drosophila afflicta: AY485303, Drosophila rinconis: AY485304, Osmia rufa: AY572828), Bee LW2 (B.impatiens: AY485306, B. terrestris: AY485305, D. afflicta: AY485308, D. rinconis: AY485307, O. rufa: AY572829, Apis mellifera: XM_397398.1),Anopheles LW (GPRop5, GPRop6, GPRop7; see Appendix 1, available as Supplementary Material online).

Flea and Mecoptera Opsin 1167

Dow

nloaded from https://academ

ic.oup.com/m

be/article/22/5/1165/1066918 by guest on 21 July 2022

model analyses (Huelsenbeck and Ronquist 2001; Ronquistand Huelsenbeck 2003). Because different codon positionshave different structural constraints, the data set was parti-tioned into first, second, and third codon positions. Modelsfor each partition were selected using the procedure imple-mented in Modeltest v.3.6 (Posada and Crandall 1998).Mixed models were used with unlinked parameters betweenpartitions, treating model parameters as unknown variableswith uniform default priors. Three independent BMCMCanalyses were run, each consisting of 10 independentMarkov chains started from a random tree and run for3 3 106 generations, with every 1,000th generationsampled. To confirm that each separate Bayesian analysisconverged and mixed well, the fluctuating value of the like-lihood and all phylogenetic parameters were examinedgraphically. To check for congruence among independentruns, the mean likelihood score and parameter values andthe posterior probabilities (pP) for individual clades werecompared. All sample points prior to reaching stationaryvalues were discarded as burn-in, and the remaining treesfrom each independent analysis were combined to calculatethe maximum a posteriori (MAP) tree (Huelsenbeck andImennov 2002; Huelsenbeck et al. 2002).

Opsin Evolution

The ratio of nonsynonymous versus synonymouschanges (dN/dS) for the insect LW portion of the tree wasestimated with maximum likelihood using the codon-basedsubstitution model of the CODEML software package inPAML v3.14 (Yang 1998). Several different site-specificmodels in which selective pressure varies among differentsites but the site-specific pattern is identical across all lin-eages were implemented (Yang et al. 2000): model M0(null model with no variation among sites), M1 (‘‘neutral’’model, with two categories of site with fixed dN/dS ratios of0 and 1), M2 (‘‘selection’’ model with three categories ofsite, two with fixed dN/dS ratios of 0 and 1, and a third esti-mated dN/dS ratio), M3 (‘‘discrete’’ model with three cate-gories of site, where the dN/dS ratio is free to vary for eachsite), M7 (‘‘beta’’ model—10 categories of site, with 10 dN/dS ratios in the range 0–1 taken from a discrete approxima-tion of the beta distribution), and M8 (‘‘beta plus omega’’model—10 categories of site from a beta distribution as inmodel M7 plus an additional category of site with a dN/dSratio that is free to vary from 0 to greater than 1). PAMLestimates the dN/dS ratios that are free to vary under thesemodels, as well as the proportion of sites with each ratio. Allmodels were run twice with starting omega values of lessthan and greater than 1 as suggested by the PAML manualto test for entrapment in local optima (Yang 1997). Like-lihood ratio tests, to determine whether more complex mod-els provided a significantly better fit to the data than moresimple models, were performed by comparing the likeli-hood ratio test statistic (�2[ln L1� ln L2]) to critical valuesof the chi-square distribution with the appropriate degreesof freedom (Yang 1998).

Although dN/dS ratios are useful for detecting selection,they do not indicate how the identified selection affects theoverall structure and functionof theprotein.Amino acid sub-stitutions can have a wide range of effects on a protein

depending on the difference in physicochemical propertiesand location in theprotein structure; inorder to evaluate theseeffects we used the insect LWportion of our Bayesian tree inTreeSAAP v3.0 (Woolley et al. 2003) to test for evidence ofselection among 31 amino acid properties in the three line-ages of interest. TreeSAAP allows us to identify these prop-erty changes and classify them into categories on a gradientfrom conservative to radical change. Based on data set–specific nucleotide substitution patterns, a neutral modelof expected change for each category is generated. This nullmodel is then compared to the observed numbers of aminoacid replacements in the data set, and a z-score is calculatedfor each category. Categories where the observed number ofamino acid replacements is significantly different than neu-tral expectations are considered as potentially being affectedby selectivepressures. In this study,we specifieda scaleof20categories of change. Greater than expected numbers ofreplacements (relative to the neutral model) in categories1–3 indicate significant stabilizing selection, whereas thesame situation in categories 18–20 indicate significant desta-bilizing selection (table 3). Bymapping the identified aminoacid replacementsbackonto theprotein structure,weare ableto make inferences about the influence of potential selectivepressures onprotein function and on the evolutionary historyof particular lineages (Woolley et al. 2003).

Results and DiscussionOpsin Gene Isolation and Phylogenetic Analysis

PCRamplificationof theLWopsin gene for theMecop-tera and Siphonaptera yielded;1,030- and;1,200-bp frag-ments, respectively.Wewere unable to obtain;460bp fromthe 5# end ofPanorpa gracilis andBoreus coloradensis. Thetranslated sequencesused in the analysis rangedbetween203and 278 amino acids. Interestingly, Siphonaptera have aunique four–amino acid–long expansion segment in trans-membrane helix 6, at amino acid positions 119–222 in ouralignment (fig. 3).

The MAP tree (ln L 5 �24,065.002 and pP 5 0.002)used in PAML and TreeSAAP analyses is shown in figure 4.Although not entirely congruent with the species phylogenypresented by Whiting (2002), there are nonetheless severalsignificant features about this topology. Briscoe (2000)demonstrated that opsins will cluster together primarilyby physiological similarity, i.e., similar spectral sensitiv-ities, and secondarily by species relationships. Our datastrongly support these results, with opsins of physiologicalsimilarity clustering together across the topology. FL-Opsins and M-Opsins are well supported as monophyleticgroups within the insect LW clade, suggesting sensitivity inthe red-green spectrum. We note, however, that althoughsupport for the ordinal grouping of the FL-Opsins, M-Opsins, and B-Opsin sequences is strong, the opsin genegenerally provides little or no support for the resolutionof these ordinal relationships to each other.

Furthermore, a recent study of the Anopheles gambiaegenome revealed the presence of a higher number of opsingenes than any other characterized insect,with at least half ofthe identified copies being duplicates of LW-sensitive genes(Hill et al. 2002). Phylogenetic analysis of this LW genecomplement indicated at least one early gene duplication

1168 Taylor et al.

Dow

nloaded from https://academ

ic.oup.com/m

be/article/22/5/1165/1066918 by guest on 21 July 2022

within insects before the emergenceof the ordersOrthoptera,Mantodea, Hymenoptera, Lepidoptera, and Diptera(Spaethe and Briscoe 2004). In our analysis, the B-Opsin,FL-opsin, and M-opsin sequences do not cluster with eithertheA. gambiae or the beeLW2genes (fig. 4). Analysis of ouralignment using themethods described by Spaethe and Bris-coe (2004) produced a tree similar to their results, with thenewly characterized B-Opsin, FL-Opsin, and M-Opsinsequences grouping within the insect LW1 clade (resultsnot shown). Although the weak support in the LW cladeof our MAP tree makes inferences about opsin gene dupli-cation and evolution difficult, the phylogenetic relationshipsof the opsin gene copies in A. gambiae and Hymenopteraindicate that there may be many more yet unidentified geneduplication events in the insect LW clade.

Opsin Gene Structure

Figure 2 shows comparative intron splice site positionsof FL-Opsin, B-Opsin, and M-Opsin relative to other insectLW opsins. Within the regions amplified, the putativeSiphonaptera opsin gene contains five introns, two Panor-pidae, and four Boreidae. Interestingly, the B-opsin and FL-opsins (putative LW1) and bee LW2 share a unique intronat position 156, although they are not a monophyletic clade

in our phylogeny. Manual mapping of this character ontothe MAP tree (fig. 4) indicates a single gain of this intronsite, with secondary loss in the Bee LW1 clade.

Briscoe (1999) reported three splice sites conservedacross insect BG and LW opsin groups (sites 190, 239,and 332, fig. 2). Our data support this observation for bothsites 190 and 239, although secondary intron deletions areobserved in the bee LW1 (position 239) and Anopheles andPanorpidae (position 190) lineages. Site 332 is outside theregion amplified in this study and cannot be further eval-uated. Additionally, our data indicate a unique splice siteat 119 that, with the exception of bee LW1 and Anopheles,appears to be characteristic of LW opsins, and we predict itwill also be present in boreids.

Of the remaining intron insertions, sites 79 in Papilioand 108 in bee LW1 appear to be autapomorphies (fig. 2).Several sites (277, Papilio and bee LW1 and LW2; 296,Boreidae and Siphonaptera) appear in only a few taxa andmay indicate their presence before the divergence of thoselineages. All of the Anopheles LW opsin gene copies inves-tigated (GPRop5, GPRop6, GPRop7) have lost all intronsplice sites except 239, implying that patterns of introninsertion-deletion can change considerably within lineages.

Even though two groups may share a particular intronsplice site, the length of the insert was observed to vary. For

FIG. 3.—Two-dimensional representation of the homology model of FL-Opsin showing preservation of the seven-helix structure. Numbering beginswith the first amino acid included in this analysis. Black circles represent amino acids not obtained. Boxed residues show an expansion region found onlyin FL-Opsin. Circles with a cross in them show key conserved residues described by Briscoe (1999). The results of the TreeSAAP analysis are mappedonto the model. Blue-colored circles show sites of conservative amino acid (AA) replacement, and red-colored circles show sites of radical AA replace-ment found along the branches leading to FL-Opsin, M-Opsin, or B-Opsin clades. Circles shaded in light gray indicate sites strictly conserved across thethree lineages. White circles represent AA residue variation among the three lineages.

Flea and Mecoptera Opsin 1169

Dow

nloaded from https://academ

ic.oup.com/m

be/article/22/5/1165/1066918 by guest on 21 July 2022

example, at position 296 the boreid has a long intron (537bp) and the fleas have relatively short introns (58–84 bp).Intron lengths vary even among taxa within the same group,e.g., sequence lengths for position 119 in Jellisonia sp., C.agyrtes, and P. irritans are 99, 119, and 66 bp, respectively.These results indicate that the splice sites appear to be morephylogenetically conserved than the actual makeup of theintron.

Removal of introns reveals a preserved open readingframe in the boreid-, panorpid-, and flea-lineages. Homol-ogy modeling of the inferred amino acid sequence of theJellisonia sp. FL-Opsin suggests that the seven transmem-brane helices characteristic of G protein–coupled receptors

and all known insect opsins are preserved (fig. 3). Addition-ally, Briscoe (1999) described seven amino acid motifs thatare necessary for a functional opsin protein and that are con-served across all known insect opsins. Four of these motifsare located within the region amplified in this study, all fourof which are conserved in FL-Opsin, B-Opsin, andM-Opsin (fig. 3). These residues include (1) Leu36 and(2) Asn41 found in all G protein–coupled receptors (num-bered according to the first residue obtained in FL-Opsin);(3) the G protein–coupled receptor motif, E/D R, in the thirdtransmembrane region involved in transducin activation(Asp102 and Arg103); and (4) a disulfide bridge connectingtransmembrane region 3 with the second cytostolic loop at

FIG. 4.—Maximum a posteriori estimate of phylogeny (MAP) tree used in PAML and TreeSAAP analyses, ln L 5 �24,065.002 and pP 5 0.002.Posterior probabilities .0.95 are indicated by a gray circle on the corresponding branch. A detailed list of all taxa names and their respective accessionnumbers can be found in Appendix 1. The shared insertion at site 156 between Boreidae, Siphonaptera, and bee LW2 is mapped onto the respectivebranches, indicating a gain (red rectangle) and a loss (white rectangle).

1170 Taylor et al.

Dow

nloaded from https://academ

ic.oup.com/m

be/article/22/5/1165/1066918 by guest on 21 July 2022

Cys78/155. As expected, many of these key residues occur inregions for which the amino acid sequence is conservedacross all insect LW opsins.

Opsin Evolution

The results of the PAML analysis are summarized intable 2. For the majority of the comparisons made, we failedto reject the null hypothesis that models that allow for pos-itive selection fit the data significantly better than modelsthat do not. However, model M3 showed a significantly bet-ter score (P5 0.001) than the neutral model M1, indicatingthat selective pressure is present. Although model M3 fit thedata better, no substitutions were found in categories wherex . 1, rejecting the hypothesis of diversifying selection.This result is confirmed by examining the individual ratesof evolution found inM-Opsin, B-Opsin, and FL-Opsin andcomparing them to the other lineages in the topology. If, forinstance, in the flea lineage there was either strong selectionaway from the ancestral protein or a loss of functional con-straint of the opsin protein due to the reduction in eye struc-ture, we expect to see a greater number of nucleotidesubstitutions as well as a larger proportion of nonsynony-mous to synonymous substitutions (Yokoyama et al. 1995).The dN/dS ratios on the branch leading to the FL-Opsinclade are not significantly different (P 5 0.05) from thoseleading to the B-Opsin and M-Opsin clades (see table 2).These ratios are also nearly identical to the average ratesof evolution across the entire topology (P5 0.05). In addi-tion, all ratios are observed to be less than 1. As stated ear-lier, this indicates that any selective pressure on opsinevolution is purifying across all the insects observed suchthat the structural and functional properties of the proteinare being conserved (Yang et al. 2000).

Analysis in TreeSAAP also confirmed that a low rateof radical diversifying selection (destabilizing selection)was occurring with respect to amino acid properties. Only9.6% of all sites showed destabilizing selection among 7 ofthe 31 properties tested. On the other hand, 21.4% of allsites are under strongly stabilizing selection for 22 of 31amino acid properties tested, and 60% of the amino acidresidues in FL-Opsin, B-Opsin, and M-Opsin were identi-cal. Although several of the properties showing destabiliz-ing selection (e.g., turn tendencies, coil tendencies, or

power to be at the N-terminal) seem to be related to struc-tural aspects of the protein, none are properties previouslyidentified to affect spectral tuning (i.e., polarity). Addition-ally, all of the conserved residues as described by Briscoe(1999) also show conservation in our data set. These resultsare shown in figure 3 mapped onto the two-dimensionalprojection of FL-Opsin. See table 3 for a complete list ofall properties tested. As a whole, these data not only implyfunctional constraint but also similar functional propertiesacross FL-Opsin, B-Opsin, and M-Opsin.

Sensitivity to Topology

Because of the weak support of the ordinal relation-ships found in our analysis, the PAML and TreeSAAP anal-yses were also calculated after constraining the data to theWhiting (2002) species tree. The results in all cases wereidentical, suggesting that the data are relatively insensitive

Table 2Results from PAML Analysis of FL-Opsin, B-Opsin, andM-Opsin Lineages

Differentfrom

Neutrala

dN/dSb

Model Code ln LFL-Opsin B-Opsin M-Opsin

M0 (one-ratio) �3,175.547 — 0.0424 0.0424 0.0424M1 (neutral) �3,137.736 — 0.0651 0.0649 0.0649M2 (selection) �3,137.736 N 0.0651 0.0649 0.0649M3 (discrete) �3,092.321 Y 0.0437 0.0437 0.0437M7 (beta) �3,099.321 — 0.0610 0.0611 0.0610M8 (beta&w) �3,099.321 N 0.0610 0.0611 0.0610

a Results from likelihood ratio test (P 5 0.01).b Ratios calculated from branches leading to FL-Opsin, B-Opsin, and M-Opsin

clades, respectively, and compared to the average ratio across the tree for each model

analyzed. In each model, rates are not significantly different (P 5 0.05) from the

average rate across the tree.

Table 3Conservative (Stabilizing) and Radical (Destabilizing)Amino Acid Properties Under Selection Identified from 31Tested Amino Acid Properties in TreeSAAP Analysisa

Amino Acid Property

ConservativeSelectionCategories

RadicalSelectionCategories

1 2 3 18 19 20

Alpha-helica tendencies XAverage number of surroundingresidues

X

Beta-structure tendencies XBulkiness X XBuriedness X XChromatographic index X X XCoil tendencies XComposition XCompressibility XEquilibrium constant XHelical contact area XHydropathy X X XIsoelectric point XLong-range non-bondedenergy

X

Mean root square mean fluctuationdisplacement

Molecular volumeMolecular weightNormalized consensushydrophobicity

X

Partial specific volume XPolar requirement X XPolarity X XPower to be at the C-terminal X XPower to be at the middle ofalpha helix

Power to be at N-terminal XRefractive index XShort/-medium range non-bondedenergy

X

Solvent accessible reduction ratio XSurrounding hydrophobicity XThermodynamic transferhydrophobicity

X

Total non-bonded energy X XTurn tendencies X

a Twenty-two properties show selection at least once in the first three categories

(conservative selection). Only seven properties show selection in the last three cat-

egories (radical selection).

Flea and Mecoptera Opsin 1171

Dow

nloaded from https://academ

ic.oup.com/m

be/article/22/5/1165/1066918 by guest on 21 July 2022

to organization of the ordinal relationships, as predicted byYang et al. (2000). We observe that despite the overlyingreduction in eye morphology and loss of visual sensitivityacross these three lineages, there remains strong conserva-tive, purifying selection on the opsin protein.

Functional Constraint of FL-Opsin, B-Opsin,and M-Opsin

Despite the considerable variation in lifestyle, habitat,and visual structure, the LW opsin gene appears to beremarkably well preserved across panorpids, boreids, andfleas. This is not entirely surprising, given that all three lin-eages demonstrate sensitivity and responsiveness to visualcues at some level. Perhaps with gross reduction in flea eyemorphology and the incapability to perceive images, wemight have expected to see some corresponding changein the mechanism of photoreception itself. However, as dis-cussed above, not only does FL-Opsin appear to haveretained functional constraint but also no evidence is foundfor unusual selective pressure on the amino acid sequenceand structural properties. On the contrary, we observed ahigh degree of conservative selection, preserving aminoacid properties and structure similar to other functionalopsins. These results provide evidence that although vastdifferences in perceptual abilities exist between Panorpidae,Boreidae, and Siphonaptera, e.g., ability to form images,the underlying physiology involved in the mechanismsof photoreception appears to be preserved.

In addition, the preservation of opsin structure inSiphonaptera, combined with the apparent reduction ofmacroscopic eye structure may also indicate the presenceof an extraretinal photoreceptor system. Indeed, prelimi-nary transmission electron studies on flea eyespots indicateno preservation of typical microscopic adult insect eyestructures (e.g., rhabdom, crystalline cone, corneal lens)as well as a heavily sclerotized layer of chitin coveringthe flea ‘‘eyes’’ (Dittmar de la Cruz, unpublished data). Thisstructural degradation implies that photoreception may nolonger occur within the eye. Extraretinal, nonvisual photo-receptors are common among insects and have been foundat sites located in the central nervous system (Briscoe andNagy 1999; Shimizu et al. 2001), the posterior margin of thecompound eye (Yasuyama and Meinertzhagen 1999), andthe peripheral abdominal segments (Arikawa 1997; see G.Fleissner and G. Fleissner 2003 for a review of nonvisualphotoreceptors). A number of LW opsin copies have beenimplicated as extraretinal (Briscoe and Nagy 1999; Spaetheand Briscoe 2004), and Spaethe and Briscoe (2004) spec-ulate that these extraretinal opsins evolved early in inver-tebrate evolution. Interestingly, similar to the peripheralphotoreceptors found on the last abdominal segments(genitalia) of Papilio xuthus (Arikawa et al. 1980), fleashave a sensillial plate (or pygidium) on the last abdominalsegment, which is packed with receptor-like structures.Unfortunately, the function of these ‘‘receptors’’ is still asource of speculation.

Furthermore, extraretinal photoreceptors appear to beinvolved in the photic entrainment of the circadian clock(Shimizu et al. 2001; Malpel, Klarsfeld, and Rouyer2002). Koehler, Leppla, and Patterson (1989) demonstrated

the presence of circadian rhythms in fleas, which could bemediated by these types of nonvisual receptors.

Other studies have also shown examples of organismsthat do not have functional eye structures, yet seem to retainfunctional opsin proteins. R. Yokoyama and S. Yokoyama(1990), for example, presented a case where putative func-tional opsin genes were isolated from Astyanax fasciatus,a blind cave fish. This phenomenon was confirmed byCrandall and Hillis (1997), who demonstrated that somespecies of cave crayfish still possess a functional opsin withno significant differences from opsins isolated from surfacespecies. As there is no light available, there would be littlereason to maintain a traditional photoreceptor. They con-cluded that opsin must be involved in other pathwaysbesides vision, although this remains highly speculative.

Conclusions

(1) We have isolated and characterized the first LW opsingenes from two orders: Mecoptera and Siphonaptera.

(2) Phylogenetic analysis implies a physiological sensitiv-ity in the red-green spectrum for these opsins, consistentwith previously identified spectral sensitivities.

(3) Analysis of intron splice sites reveals the presence oftwo introns (190 and 239) conserved across the BGand LW groups. In addition, one intron (119) seemsto be unique to LW opsins.

(4) Flea opsin sequences showed a unique, four–aminoacid–long expansion segment.

(5) Calculated rates of evolution indicate that FL-Opsin,B-Opsin, and M-Opsin are evolving at similar rateswith no radical selective pressures, implying conserva-tive evolution and functional constraint.

The amino acid composition of FL-Opsin, B-Opsin,and M-Opsin is remarkably well preserved. Over 60% oftheaminoacid residuesare identical across the three lineages,with an additional 22% that are under stabilizing, conserva-tive selection.Among theseareseveralkeyaminoacidmotifsthatarenecessaryforproperopsinfunction.Panorpidae,Bor-eidae, and Siphonaptera exhibit vast differences in lifestyleand ecology that are reflected in the adaptations of the visualsystems of each group. These lines of evidence indicate thatdespite the reduction ineyestructureand lossofvisualacuity,fleashave retained a remarkable similarity inLWopsin struc-ture and amino acid properties.Although it is highly unlikelythat fleas perceive visual images, our data support the possi-bility offleasbeing sensitive toLWlight.That theunderlyingmechanismsofphotoreceptioncouldbepreservedacrossvis-ual systems of such immensely different qualities, speaksmuch of the flexibility, ease of adaptation, and potential ofour biological world.

Supplementary Material

Supplementary data are available atMolecular Biologyand Evolution online (www.mbe.oupjournals.org).

Acknowledgments

We gratefully acknowledge K. Miller, S. Cameron, andG. Svenson for their support and comments on this work.We

1172 Taylor et al.

Dow

nloaded from https://academ

ic.oup.com/m

be/article/22/5/1165/1066918 by guest on 21 July 2022

arealsograteful toD.McClellanforhis invaluable insightsandhelp with the TreeSAAP analysis, as well as to M. Perez-LosadaforhishelpwithPAML.ThankstoJ.Pricefortheinitialdesign of opsin PCR primers, to M. Hastriter for his indefat-igable knowledge of fleas and the use of his collection, and toR.Trowbridge for his photography. This researchwas fundedby NSF grants DEB-9983195 (M.F.W.) and DEB-0120718(M.F.W.), theKarl-Enigk-Foundation for Experimental Para-sitology (Hannover, Germany), and the BrighamYoungUni-versity Office of Research and Creative Activities.

Literature Cited

Arikawa, K., E. Eguchi, A. Yoshida, and K. Aoki. 1980. Multipleextraocular photoreceptive areas on genitalia of butterfly,Papilio xuthus. Nature 288:700–702.

Arikawa, K. 1997. Hindsight by genitalia: photo-guided copula-tion in butterflies. J. Comp. Physiol. A. 180:295–299.

Bellingham, J., S. E. Wilkie, A. G. Morris, J. K. Bowmaker, andD. M. Hunt. 1997. Characterization of the ultraviolet-sensitiveopsin gene in the honey bee, Apis mellifera. Eur. J. Biochem.243:775–781.

Briscoe, A. D. 1999. Intron splice sites of Papilio glaucus PglRh3corroborate insect opsin phylogeny. Gene 230:101–109.

———. 2000. Six opsins from the butterfly Papilio glaucus:molecular phylogenetic evidence for paralogous origins ofred-sensitive visual pigments in insects. J. Mol. Evol.51:110–121.

Briscoe, A. D., and L. Chittka. 2001. The evolution of color visionin insects. Annu. Rev. Entomol. 46:471–510.

Briscoe, A. D., and L. Nagy. 1999. Spatial expression of opsinsin the retina and brain of the tiger swallotail Papilio glaucus.Am. Zool. 39:254B.

Burkhardt, D., and I. De LaMotte. 1972. Electrophysiologicalstudies on the eyes of Diptera, Mecoptera, and Hymenoptera.Pp. 137–145 in R. Wehner, ed. Information processing inthe visual systems of arthropods. Springer-Verlag, Berlin,Germany.

Byers, G. W., and R. Thornhill. 1983. Biology of the Mecoptera.Annu. Rev. Entomol. 28:203–228.

Carulli, J. P., D.-M. Chen, W. S. Stark, and D. L. Hartl. 1994.Phylogeny and physiology of Drosophila opsins. J. Mol. Evol.38:250–262.

Crandall, K. A., and D. M. Hillis. 1997. Rhodopsin evolution inthe dark. Nature (Lond.) 387:667–668.

Cronin, T.W., M. Jarvilehto, M.Weckstrom, and A. B. Lall. 2000.Tuning of photoreceptor spectral sensitivity in fireflies(Coleoptera: Lampyridae). J. Comp. Physiol. 186:1–12.

Crum, G. E., F. W. Knapp, and G. M.Wite. 1974. Response of thecat flea, Ctenocephalides felis (Bouche), and the oriental ratflea, Xenopsylla cheopis (Rothschild), to electromagnetic radi-ation in the 300–700 nanometer range. J. Med. Entomol.11:88–94.

Deininger, W., M. Fuhrman, and P. Hegemann. 2000. Opsin evo-lution: out of thewild greenyonder?TrendsGenet.16:158–159.

Dunnet,G.M.,andD.K.Mardon.1991.Siphonaptera.Pp.705–716in CSIRO, ed. The insects of Australia: a textbook for studentsand research workers. CSIRO and Cornell University Press,Melbourne, Australia.

GeneCodes. 2003. Sequencher. v4.2. Ann Arbor, Mich.Guex, N., andM. C. Peitsh. 1997. SWISS-MODEL and the Swiss-

Pdb viewer: an environment for comparative protein modeling.Electrophoresis 18:2714–2723.

Fleissner G., and G. Fleissner. 2003. Nonvisual photoreceptorswith emphasis on their putative role as receptors of naturalZeitgeber stimuli. Chronobiology 20(4):593–616.

Hill, C. A., A. N. Fox, R. J. Pitts, L. B. Kent, P. L. Tan, M. A.Chrystal, A. Cravchik, F. H. Collings, H. M. Robertson, andL. F. Zwiebel. 2002. G protein-coupled receptors in Anophelesgambiae. Science 298:176–178.

Huelsenbeck, J. P., and N. S. Imennov. 2002. Geographic origin ofhuman mitochondrial DNA: accommodating phylogenticuncertainty and model comparison. Syst. Biol. 51:155–165.

Huelsenbeck, J. P., B. Larget, R. E. Miller, and F. Ronquist. 2002.Potential applications and pitfalls of Bayesian inference phy-logeny. Syst. Biol. 51:673–688.

Huelsenbeck, J. P., and F. Ronquist. 2001. MrBayes: Bayesianinference of phylogeny. Bioinformatics 17:754–755.

Kitamoto, J., K. Sakamoto, K. Ozaki, Y. Mishina, and K. Arikawa.1998. Two visual pigments in a single photoreceptor cell: iden-tification and histological localization of three mRNAs encod-ing visual pigment opsins in the retina of the butterfly Papilipxuthus. J. Exp. Biol. 201:1255–1261.

Kochendoerfer, G. G., S. W. Lin, T. P. Sakmar, and R. A. Mathies.1999. How color visual pigments are tuned. Trends Biochem.Sci. 8:300–305.

Koehler, P. G., N. C. Leppla, and R. S. Patterson. 1989. Circadianrhythm of cat flea (Siphonaptera: Pulicidae) locomotion unaf-fected by ultrasound. J. Econ. Entomol. 82:516–518.

Land, M. F. 2003. Eyes and vision. Pp. 393–406 inV. H. Resh andR. T. Carde, eds. Encyclopedia of insects. Academic Press,New York.

Lewis, R. E., and J. H. Lewis. 1985. Notes on the geographicaldistribution and host preferences in the order Siphonaptera.J. Med. Entomol. 22:134–152.

Maddison, W. P., and D. R. Maddison. 2003. MacClade: analysisof phylogeny and character evolution. Sinauer Associates,Sunderland, Mass.

Malpel, S., A. Klarsfeld, and F. Rouyer. 2002. Larval optic nerveand adult extra-retinal photoreceptors sequentially associatewith clock neurons during Drosophila brain development.Development 129:1443–1453.

Nathans, J. 1987. Molecular biology of visual pigments. Annu.Rev. Neurosci. 10:163–194.

Neufeld, T. P., R. W. Carthew, and G. M. Rubin. 1991. Evolutionof gene position: chromosomal arrangement and sequencecomparison of the Drosophila melanogaster and Drosophilavirilis sina and Rh4 genes. Proc. Natl. Acad. Sci. USA88:10203–10207.

Osbrink, W. L., and M. K. Rust. 1985. Cat flea (Siphonaptera:Pulicidae): factors influencing host-finding behavior in the lab-oratory. Ann. Entomol. Soc. Am. 78:29–34.

Palczewski, K., T. Kumasaka, T. Hori et al. (12 co-authors). 2000.Crystal structure of rhodopsin: a G protein-coupled receptor.Science 289:739–745.

Penny, N. D. 1977. A systematic study of the family Boreidae(Mecoptera). Kans. Univ. Sci. Bull. 51:141–217.

Penny, N. D., and G. W. Byers. 1979. A check-list of the Mecop-tera of the world. Acta Amazon. 9:365–388.

Posada, D., and K. A. Crandall. 1998. Modeltest: testing the modelof DNA substitution. Bioinformatics 14:817–818.

Resh, V. H., and R. T. Carde. 2003. Encyclopedia of insects. Aca-demic Press, New York.

Ronquist, F., and J. P. Huelsenbeck. 2003. MrBayes 3: Bayesianphylogenetic inference under mixed models. Bioinformatics19:1572–1574.

Rust, M. K., and M. W. Dryden. 1997. The biology, ecology, andmanagement of the cat flea. Annu. Rev. Entomol. 42:451–473.

Shimizu, I., Y. Yamakawa, Y. Shimazaki, and T. Iwasa. 2001.Molecular cloning of Bombyx cerebral ospin (Boceropsin)

Flea and Mecoptera Opsin 1173

Dow

nloaded from https://academ

ic.oup.com/m

be/article/22/5/1165/1066918 by guest on 21 July 2022

and cellular localization of its expression in the silkworm brain.Biochem. Biophys. Res. Commun. 287:27–34.

Spaethe, J., and A. D. Briscoe. 2004. Early duplication and func-tional diversification of the opsin gene family in insects. Mol.Biol. Evol. 21:1583–1594.

Thompson, J. D., T. J. Gibson, F. Plewniak, F. Jeanmougin, andD. G. Higgins. 1997. The Clustal X windows interface: flexiblestrategies for multiple sequence alignment aided by qualityanalysis tools. Nucleic Acids Res. 24:4876–4882.

Towner, P., and W. Gartner. 1994. The primary structure of man-tid opsin. Gene 143:227–231.

Whiting, M. F. 2002. Mecoptera is paraphyletic: multiple genesand phylogeny of Mecoptera and Siphonaptera. Zool. Scr.31:93–104.

Woolley, S., J. Johnson, M. J. Smith, K. A. Crandall, and D. A.McClellan. 2003. TreeSAAP: selection on amino acid proper-ties using phylogenetic trees. Bioinformatics 19:671–672.

Yang, Z. 1997. PAML: a program package for phylogeneticanalysis by maximum likelihood. Comput. Appl. BioSci.13:555–556.

Yang, Z., R. Nielsen, N. Goldman, and A.-M. K. Pedersen. 2000.Codon-substitution model for heterogeneous selection pres-sure at amino acid sites. Genetics 155:431–449.

Yasuyama, K., and I. Meinertzhagen. 1999. Extraretinal photore-ceptors at the compound eye’s posterior margin in Drosophilamelanogaster. J. Comp. Neurol. 412:193–202.

Yokoyama, R., and S. Yokoyama. 1990. Convergent evolutionof the red- and green-like visual pigment genes in fish,Astyanax facsiatus, and human. Proc. Natl. Acad. Sci. USA87:9315–9318.

Yokoyama, S., A. Meany, H. Wilkens, and R. Yokoyama. 1995.Initial mutational steps toward loss of opsin gene function incavefish. Mol. Biol. Evol. 12:527–532.

Claudia Schmidt-Dannert, Associate Editor

Accepted January 17, 2005

1174 Taylor et al.

Dow

nloaded from https://academ

ic.oup.com/m

be/article/22/5/1165/1066918 by guest on 21 July 2022