MBE Advance Access originally published online on October 19, 2006
Molecular Biology and Evolution 2007 24(1):253-268; doi:10.1093/molbev/msl152
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Research Articles |
Molecular Characterization of Crustacean Visual Pigments and the Evolution of Pancrustacean Opsins
* Department of Microbiology and Molecular Biology, Brigham Young University
Department of Biological Sciences, University of Maryland Baltimore County
Department of Integrative Biology, Brigham Young University
E-mail: porter{at}umbc.edu.
 |
Abstract |
|---|
 TOP Abstract IntroductionMethodsResultsDiscussionSummaryAcknowledgementsReferences |
|---|
Investigations of opsin evolution outside of vertebrate systems have long been focused on insect visual pigments, whereas other groups have received little attention. Furthermore, few studies have explicitly investigated the selective influences across all the currently characterized arthropod opsins. In this study, we contribute to the knowledge of crustacean opsins by sequencing 1 opsin gene each from 6 previously uncharacterized crustacean species (Euphausia superba, Homarus gammarus, Archaeomysis grebnitzkii, Holmesimysis costata, Mysis diluviana, and Neomysis americana). Visual pigment spectral absorbances were measured using microspectrophotometry for species not previously characterized (A. grebnitzkii = 496 nm, H. costata = 512 nm, M. diluviana = 501 nm, and N. americana = 520 nm). These novel crustacean opsin sequences were included in a phylogenetic analysis with previously characterized arthropod opsin sequences to determine the evolutionary placement relative to the well-established insect spectral clades (long-/middle-/short-wavelength sensitive). Phylogenetic analyses indicate these novel crustacean opsins form a monophyletic clade with previously characterized crayfish opsin sequences and form a sister group to insect middle-/long-wavelength–sensitive opsins. The reconstructed opsin phylogeny and the corresponding spectral data for each sequence were used to investigate selective influences within arthropod, and mainly "pancrustacean," opsin evolution using standard dN/dS ratio methods and more sensitive techniques investigating the amino acid property changes resulting from nonsynonymous replacements in a historical (i.e., phylogenetic) context. Although the conservative dN/dS methods did not detect any selection, 4 amino acid properties (coil tendencies, compressibility, power to be at the middle of an
-helix, and refractive index) were found to be influenced by destabilizing positive selection. Ten amino acid sites relating to these properties were found to face the binding pocket, within 4 Å of the chromophore and thus have the potential to affect spectral tuning.
Key Words: TreeSAAP • selection • opsin evolution • Crustacea • visual pigment absorbance
|
Introduction |
|---|
|
TOPAbstractIntroductionMethodsResultsDiscussionSummaryAcknowledgementsReferences |
|---|
Visual pigment research has long been of interest to a number of biological disciplines, including sensory ecologists, visual physiologists, biochemists, and molecular evolutionists. Composed of a chromophore bound to an integral membrane protein (opsin), visual pigments are phenotypically characterized by the wavelength of maximal absorption (
max). As most visual pigments contain the same chromophore, variation in opsin sequence is responsible for most of the observed variation in max. The ability to study the genetic mechanisms behind phenotypic variation has made opsin a model evolutionary system. In vertebrates, these studies often take the form of site-directed mutagenesis studies, where the effect of changes at a single amino acid residue on max can be quantified (Nathans 1990a
; Neitz et al. 1991; Asenjo et al. 1994; Yokoyama and Radlwimmer 1999; Wilkie et al. 2000; Yokoyama et al. 2000; Yokoyama and Radlwimmer 2001; Cowing et al. 2002; Yokoyama and Tada 2003). This work suggests that spectral tuning involves the distribution of charged or polar residues within the chromophore-binding pocket (Nathans 1990a; Neitz et al. 1991; Chan et al. 1992). Due to the difficulty of expressing visual pigments, most studies of "nonvertebrates" (i.e., insects, cephalopods, crayfish, and Limulus) have used comparative methods and homology modeling to identify sites potentially important in spectral tuning (Chang et al. 1995; Chase et al. 1997; Crandall and Cronin 1997; Briscoe 2001, 2002), whereas a brave few have investigated spectral tuning using Drosophila-based heterologous expression systems (Britt et al. 1993; Salcedo et al. 1999, 2003). These studies have not only uncovered some similarities among spectral tuning sites in vertebrates and insects (Briscoe 2001; Salcedo et al. 2003), but also illustrated that there are a number of unique residues affecting spectral tuning in insects.
Given that the visual pigment genes diversified into the observed spectral clades sometime after the separation of the protostomes and deuterostomes (Pichaud et al. 1999; Arendt et al. 2004), some differences in functionality are expected. In fact, several significant differences in the mechanism of photoactivation have been documented, most notably differences related to the Schiff base. In vertebrates, a negatively charged counterion (E113) stabilizes the positive charge of the Schiff base (Sakmar et al. 1989; Nathans 1990b; Zhukovsky et al. 1992). However, in insects and cephalopods, this same site is a tyrosine (or in UV-sensitive opsins, a phenylalanine). Studies have shown that this residue does not function as a counterion, and the exact method of photoactivation is not yet completely understood (Nakagawa et al. 1999; Salcedo et al. 2003).
Although a plethora of studies have looked at structure/function relationships in vertebrate opsins from an evolutionary perspective, few studies have investigated similar issues in the arthropod visual pigment diversification. Consequently, much less is known about the mode and tempo of opsin evolution in these animals and most of the evolutionary work related to visual pigments has focused on insect systems. Insects have at least 4 main spectral classes: long-wavelength–sensitive (LWS), middle-wavelength–sensitive (MWS), and 2 short-wavelength–sensitive (SWS) groups (UV and blue) (O'Tousa et al. 1985; Montell et al. 1987; Zuker et al. 1987; Feiler et al. 1988; Carulli and Hartl 1992; Feiler et al. 1992; Carulli et al. 1994; Towner and Gartner 1994; Smith et al. 1997; Salcedo et al. 1999; Briscoe 2001, 2002; Briscoe and Chittka 2001; Taylor et al. 2005). Although there have been a large number of physiological studies of spectral sensitivities in other major arthropod groups, particularly within crustaceans, how these spectral variants relate to the defined insect opsin clades is unknown. The only other nonvertebrate taxa in which opsin sequences have been explicitly investigated are horseshoe crabs (Smith et al. 1993), molluscs (Brown PK and Brown PS 1958; Hubbard and St George 1958; Hall et al. 1991; Morris et al. 1993), and crustaceans (Sakamoto et al. 1996; Crandall and Cronin 1997; Crandall and Hillis 1997; Oakley and Huber 2004).
Here we investigate opsin evolution in insect, crustacean, and mollusc systems and add to the knowledge of arthropod opsins by expanding the characterized opsin sequences from within the Crustacea. Although a number of insect groups have been the focus of evolutionary studies (most notably Drosophila and Lepidoptera; Carulli et al. 1994; Briscoe 2000, 2001, 2002) and representatives from known insect spectral variants are often used to place phylogenetically novel opsins and to form hypotheses about uncharacterized spectral sensitivities, few studies have focused on explicitly investigating the selective forces acting upon this diversity of opsins. We have identified opsin sequences from 6 previously uncharacterized crustacean species and determined the max in those species without previous estimates. These data complement previously characterized insect opsins, expanding the data set available to investigate selective influences from insects to "pancrustacea." Furthermore, within these data are many opsin sequences with corresponding characterization of the visual pigment spectral sensitivities. By combing the molecular and physiological data sets and analyzing them with established (PAML; Yang 1997) and newly described (TreeSAAP; Woolley et al. 2003) methods for detecting selection, we test the hypothesis that visual pigment spectral tuning throughout animal diversity is related to the distribution of charged or polar residues relative to the chromophore-binding pocket (Nathans 1990a; Neitz et al. 1991; Chan et al. 1992; Salcedo et al. 2003).
|
Methods |
|---|
|
TOPAbstractIntroductionMethodsResultsDiscussionSummaryAcknowledgementsReferences |
|---|
Taxon Sampling and Outgroup Choice
Opsin sequence data were collected from 6 crustacean species: Archaeomysis grebnitzkii, Holmesimysis costata, Mysis diluviana, Neomysis americana (Mysida); Homarus gammarus (Decapoda); and Euphausia superba (Euphausiacea). In addition, 3 sequences from the stomatopod crustacean Neogonodactylus oerstedii that were characterized for a Ph.D. thesis (Brown 1996) were included in our analyses. To these data, we added any opsin sequence data for nonvertebrate species that represented at least half of the transmembrane-spanning portion of the gene and had corresponding
max values previously reported in the literature (table 1). Sequences chosen for analyses were limited to those with spectral characterization in order to evaluate any sites identified to be under selection in terms of spectral tuning of the visual pigments. Outgroups were chosen from the vertebrate opsin clade, based on the hypothesis that duplication of opsin genes occurred independently in the lines of descent leading to protostomes and deuterostomes (Pichaud et al. 1999). Bovine rhodopsin was included as the only G protein–coupled receptor (GPCR) to have been crystallized (Palczewski et al. 2000). The inclusion of this sequence in our alignment allowed for sites identified by selection detection methods to be mapped to the bovine protein structure (see below). Two representatives of pineal opsin and an "orphan" human opsin paralog (GPR52) were included as outgroups, representing basal lineages to the vertebrate visual pigment clade (Bellingham et al. 2003; Fredriksson et al. 2003). Finally, human melatonin receptor 1A, a GPCR that has been phylogenetically placed close to the human visual pigment clade (Fredriksson et al. 2003), was chosen as a distant outgroup.
|
Microspectrophotometry
Of the 6 crustacean species molecularly characterized, 2 have been previously characterized with respect to spectral sensitivities: H. gammarus and E. superba. In order to link the analyses of selection at the molecular level to function (i.e., spectral tuning), spectral sensitivities of the 4 remaining species were characterized using microspectrophotometry (MSP). For spectral analyses, live specimens were shipped overnight in dark conditions to the University of Maryland Baltimore County. In most cases, animals were used within a week of collection. All organisms were dark adapted at least overnight, but more commonly for several days, before use. Eyes were removed under dim red light, mounted in tissue medium, and flash frozen. Frozen eye samples were sectioned immediately using a cryostat microtome to produce
14-µm thick sections. Individual sections were mounted on cover slips and scanned under dim red light on a microscope for usable rhabdom structures. Suitable sections were mounted in Ringer's buffer solution between coverslips sealed with a ring of silicone grease. The equipment and general procedure used for MSP have been described by Cronin et al. (1996). Briefly, a linearly polarized scanning beam was placed within a single rhabdom. Scans were made from 400 to 700 nm, with measurements taken at 1-nm steps. Each dark-adapted rhabdom was scanned twice to check for stability. If the 2 scans were identical, the second was saved as the direct absorption spectrum of the dark-adapted photoreceptor. The rhabdom was then exposed to 2 minutes of bright white light, followed by a second absorption scan. During bright-light treatments, the field diaphragm of the substage illuminator was closed down to produce a spot of 10 µm diameter at the level of the rhabdom, minimizing local heating of the preparation. For each photoreceptor, the rhodopsin absorption spectrum was taken as the difference between the initial, dark-adapted spectrum and the final, photobleached spectrum. The wavelength of maximum absorption (max) was estimated for each difference spectrum using a least squares procedure (see Cronin et al. 1994). We compared all photobleach difference spectra with standard rhodopsin and porphyropsin templates derived by Stavenga et al. (1993) and subsequently averaged together those that closely resembled either template. The average spectrum was then fitted to the corresponding template again to determine a max value that best represents the spectra of the measured visual pigments. Results from 6 to 10 rhabdoms, representing 2 or more individuals, were obtained for each species.
DNA Extraction, Polymerase Chain Reaction, Cloning, and Sequencing
All specimens were stored in 70–95% ethanol and kept at 4 °C until extracted. Genomic DNA was extracted using DNeasy kits (Qiagen) following the manufacturer's instructions. Polymerase chain reaction (PCR, Saiki et al. 1988) products for the opsin gene were amplified using a semi-nested degenerate PCR strategy. An initial PCR using primers LWF1a: 5'-TGG TAY CAR TWY CCI CCI ATG AA-3' and OPSRD: 5'-CCR TAN ACR ATN GGR TTR TA-3' (Chang et al. 1996) with standard conditions (2.5 µl 10x Taq buffer, 4 µl 10 mM deoxynucleoside triphosphates, 2.5 µl each of two 10 mM primers, 1.25 U Eppendorf HotMaster Taq, 12.5 µl double distilled water) was run on a Perkin-Elmer 9700 machine for 35 cycles of 95 °C for 30 s, 48 °C for 45 s, and 70 °C for 1:15 min, followed by chain extension at 72 °C for 15 min. The first round PCR reactions were then diluted 1:10 with sterile water and used as template for a second round PCR of another 35 cycles using primers LWF1a and Scylla: 5'-TTR TAI ACI GCR TTI GCY TTI GCR AA-3' (Taylor et al. 2005). Second round PCR products were visualized by agarose (1.2%) gel electrophoresis. Visible opsin DNA bands were excised and cleaned from the agarose gel using a GeneClean II kit (Bio 101). Purified opsin PCR products were cloned using the TOPO TA Cloning Kit (Invitrogen, San Diego, CA), following manufacturer's instructions. For each species examined, 3–10 clones were lysed in 50 µl buffer (10 mM Tris-HCl; 0.1 mM ethylenediaminetetraacetic acid, pH 8.0) for 10 min at 96 °C and inserts were PCR amplified from lysed cells using plasmid-specific primer pairs M13(-20) (5'-GTAAAACGACGGCCAGT-3') and M13(-24) (5'-AACAGCTATGACCATG-3') and the following PCR conditions: denaturation at 94 °C for 1 min, primer annealing at 55 °C for 1 min, and chain extension at 72 °C for 3 min, for 30 cycles. Cloned PCR products were purified using the Millipore Montage purification system and sequenced in both directions on an ABI Prism 3730 capillary autosequencer using the ABI big-dye Ready-Reaction kit and following the standard cycle sequencing protocol but using 1/16th of the suggested reaction volume.
Phylogenetic Analyses
Opsin sequences were either generated in the lab (see above section) or were downloaded from GenBank (http://www.ncbi.nlm.nih.gov; see table 1). All of the species included in the analyses were from species where both the opsin sequence and the max have been previously characterized or were measured in this study. Only sequences that spanned more than half of the opsin transmembrane domains were used to reduce spurious arrangements caused by short sequences and to increase the power of tests of selection by decreasing missing data (see below section). MacClade (Maddison DR and Maddison WP 2001) was used to create an initial nucleotide sequence alignment, which was then converted to amino acid sequences for alignment. Amino acid sequences were aligned in ClustalX (Thompson et al. 1997) using a Blosum log2 weight matrix generated using MatrixGen (http://matrixgen.sourceforge.net/) with aligned class a GPCRs from the GPCR Web site (http://www.gpcr.org). The initial nucleotide sequence alignment was then adjusted to match the resulting amino acid alignment and fine-tuned based on structural information (Chang et al. 1995; Palczewski et al. 2000). The final nucleotide alignment is available online (http://inbio.byu.edu/faculty/kac/crandall_lab/pubs.html).
Phylogenetic trees were reconstructed from both the nucleotide and amino acid alignments. The nucleotide phylogeny was reconstructed using Bayesian methods coupled with Markov chain Monte Carlo (BMCMC) inference as implemented in MrBayes v3.04b (Ronquist and Huelsenbeck 2003). Because different codon positions have different functional constraints, the data set was partitioned into first-, second-, and third-codon positions for mixed-model analyses. Model selection for each partition followed the procedure outlined by Posada and Buckley (2004) for Akaike Information Criterion (AIC) as implemented in Modeltest v3.6 (Posada and Crandall 1998). Four independent BMCMC analyses were run with each consisting of 4 chains. Each Markov chain was started from a random tree and run for 6.0 x 106 cycles, sampling every 1,000th generation. Model parameters were treated as unknown variables with uniform default priors and were estimated as part of the analysis. To confirm that our Bayesian analyses converged and mixed well, we monitored the fluctuating value of likelihood and all phylogenetic parameters graphically and compared means and variances of all likelihood parameters and likelihood scores from independent runs using the program Tracer v1.2 (Rambaut and Drummond 2003). All sample points prior to reaching stationary were discarded as burn-in. The posterior probabilities (pP) for individual clades obtained from separate analyses were compared for congruence and then combined and summarized on a majority-rule consensus tree (Huelsenbeck and Imennov 2002; Huelsenbeck et al. 2002). A phylogeny based on the amino acid alignment was constructed in the program PHYML (Guindon and Gascuel 2003), which allows for the fast estimation of large data sets within a maximum likelihood (ML) framework. The best-fit model for the amino acid alignment was determined using ProtTest v1.2.6 (Abascal et al. 2005), which uses the phylogenetic analyses library in conjunction with PHYML to compute the likelihood for each of 64 candidate models of protein evolution. The fit of each of these models to our data set was then determined using a second-order AICc framework with sample size equal to the total number of characters (i.e., alignment length). Branch support values were estimated from 100 PHYML bootstrap replicates as bootstrap proportions (BP).
Investigating Selective Influences
The influence of selective forces on the evolution of visual pigments was investigated using a suite of methodologies. First, selection was investigated using nonsynonymous to synonymous substitution rate ratios (dN/dS) as calculated in a likelihood framework in the CODEML module of PAML v3.14 (Yang 1997). Selection was determined by using likelihood ratio tests to evaluate nested site-specific models with and without incorporating selection (Yang et al. 2000). For site-specific models, models M1a (nearly neutral) versus M2a (positive selection) and M7 (beta) versus M8 (beta&
) were tested as suggested in the PAML v3.14 manual. For site classes where > 1, Bayes empirical Bayes calculations of posterior probabilities are implemented in models M2a and M8 to identify the particular sites under positive selection (Yang et al. 2005). All models were run twice with starting omega values of less than and greater than 1 to test for entrapment in local optima (Yang et al. 2000). Likelihood ratio tests, to determine whether particular models provided a significantly better fit to the data, were performed by comparing the likelihood ratio test statistic (–2[lnL1 – lnL2]) with critical values of the Chi square distribution with the appropriate degrees of freedom (Yang 1997).
A common measure of selective pressure in protein-coding genes is dN/dS = , where > 1, =1, or <1 indicate positive selection, neutral evolution, and purifying selection, respectively. Many advances have been made in the estimation of this ratio that increase its power of detecting selection, including models that are either lineage or site specific (Nielsen and Yang 1998; Yang 1998; Yang and Nielsen 1998; Yang et al. 2000; Anisimova et al. 2002; Forsberg and Christiansen 2003; Bielawski and Yang 2004). Site-specific models will be used to investigate the presence of positive selection across the entire pancrustacean opsin phylogeny. However, if adaptive evolution occurs at only a few time points and affects only a few amino acids, site-specific models may still lack power in detecting positive selection (McClellan et al. 2005). For protein-coding genes such as opsin where strong structural and functional constraints can lead to a large proportion of invariable residues (31–99% similarity in arthropod opsin sequences), dN/dS methods are still likely to be very conservative. Furthermore, beyond detecting the presence of positive selection at either specific sites or lineages within a phylogeny, dN/dS methods do not provide information on the type of positive selection detected (directional or nondirectional, stabilizing or destabilizing), have very little power to detect purifying selection, and offer little insight into how the identified selection affects the overall structure and function of the protein.
Recent methods have taken the investigation of selection in protein-coding genes further by addressing several of these issues. A different approach for detecting selection in amino acid sequences is to look at the magnitudes of property change of nonsynonymous residues across a phylogeny. Amino acid substitutions have a wide range of effects on a protein depending on the difference in physicochemical properties and location in the protein structure. This approach provides further resolution to differentiating between types of selective pressures with the ability to detect positive and negative and stabilizing and destabilizing selection and offers insights into the structural and functional consequences of the identified residues under selection (McClellan et al. 2005). We used the program TreeSAAP v3.2 (Woolley et al. 2003) to test for selection on amino acid properties within our opsin data set. For each property examined, a range of possible 1-step changes as governed by the structure of the genetic code was determined and divided into 8 magnitude categories of equal range, with lower categories indicating more conservative changes and higher categories denoting more radical changes. In order to construct an expected distribution of amino acid property change, each of the 9-nt changes in every codon of every DNA sequence within the data set was evaluated, with each nonsynonymous change assigned to 1 of the magnitude categories for each property independently. These property changes were then summed across the data set, constructing a set of relative frequencies of change for each of the 8 magnitude categories to establish the null hypothesis under the assumption of neutral conditions (McClellan and McCracken 2001). If distributions of observed changes fail to fit the expected distributions based on goodness-of-fit scores and z-scores, the null hypothesis of neutrality is rejected. Because we are interested in the evolution of spectral variation, we targeted sites identified to be under positive destabilizing selection, defined as selection for radical amino acid changes resulting in structural or functional shifts in local regions of the protein (McClellan et al. 2005). In terms of TreeSAAP analyses, positive destabilizing selection is defined a priori as properties with significantly greater amino acid replacements than neutral expectations for magnitude categories 7 and 8 (e.g., the 2 most radical property change categories). Using this criterion, identified sites were mapped onto the protein structure to investigate the structural/functional impact of the selection. Within TreeSAAP, 31 amino acid properties are evaluated across a phylogeny using either the entire data set or a sliding window analysis. For our purposes, properties and magnitudes showing significantly more observed than expected numbers of changes at the P 
0.001 level were first identified with an overall analysis of our data. The identified properties were then subjected to a sliding window analysis, investigating varying window sizes (15, 20, 30, and 40 codons in width) to determine the range that maximizes the signal. The results of the sliding window analyses were used to identify regions in the protein that differ significantly from a nearly neutral model at a significance level of P = 0.001. Finally, we identified the particular amino acid residues within each of these regions that contained positive destabilizing selection for each property. Using a high-resolution (2.6 Å) bovine rhodopsin template (1L9H.pdb) from the Protein Data Bank (http://www.rcsb.org/pdb/; Berman et al. 2000) with the program Swiss-PdbViewer v.3.7 (http://www.expasy.org/spdbv; Guex and Peitsch 1997), we mapped the identified residues to the opsin protein structure using our alignment as a template. Using features of this program, residues were evaluated for their proximity to the chromophore-binding pocket using a conservative distance of 4 Å to infer a residue's potential interaction with the chromophore. Previous studies of functionally important amino acids in class A GPCRs demonstrated that there are 2 classes of residues: those that mediate conformational change and receptor activation of G proteins, which are expected to be similar across receptors, and those that mediate ligand binding, which are expected to be highly specific due to the structural diversity of ligands (Madabushi et al. 2004). By limiting our selection of identified amino acids to those with potential to interact with the chromophore, we are looking at those residues most likely to be important to opsin-specific functioning rather than of general importance to GPCR function. Throughout this paper, domains are labeled as indicated in figure 1. All references to amino acid residues are given using bovine rhodopsin numbering to make inferences directly comparable to previous studies.
|
|
Results |
|---|
|
TOPAbstractIntroductionMethodsResultsDiscussionSummaryAcknowledgementsReferences |
|---|
MSP and Molecular Characterization of Novel Crustacean Opsin Sequences
The mysid visual pigment
max ranged from 496 to 520 nm (fig. 2). Most rhabdom scans fit the Stavenga et al. (1993) rhodopsin template, but several scans from M. diluviana indicated the presence of a porphyropsin pigment (utilizing the A2 chromophore). Here, we present only the rhodopsin data; the porphyropsin data are presented elsewhere (Jokela-Määttä et al. 2005). Besides the M. diluviana porphyropsin, there was no evidence for the presence of additional spectral classes of visual pigments in any of the species investigated.
|
The isolated crustacean opsin sequences were 284 amino acids in length and spanned from 9 amino acids before the start of transmembrane
-helix (TM) I to the middle of TMVII (AA site 299). Sequences were submitted to GenBank under the accession numbers: DQ852573–DQ852598 (newly characterized crustacean species) and DQ646869–DQ646871 (N. oerstedii). Similar to genes previously sequenced from crayfish and cephalopods (Morris et al. 1993; Crandall and Cronin 1997; Crandall and Hillis 1997), none of the isolated crustacean sequences contained any introns. Within each species, the clone sequence variability was less than 1%. This variability is less than expected from Taq error (2.3 x 10–6 to 2.1 x 10–4 errors per basepair per cycle; Bracho et al. 1998) given our PCR methodologies (2 rounds of initial PCR, cloning, and PCR confirmation of cloned products), and therefore clones were considered to be identical and were merged into a consensus sequence for each species for further analyses. When aligned with other opsin sequences, the new crustacean sequences exhibited the characteristic indel region of arthropods in cytoplasmic loop (CL) 3 of 14 amino acids.
Phylogenetic Analyses of Pancrustacean Opsins
For each codon position, the best-fit models all corresponded to general-time reversible model (GTR) + I + G (AICw: position 1 = 0.56; position 2 = 0.92; position 3 = 0.90). For the amino acid data, the best-fit model was WAG + G + F (AICcw = 0.77, = 1.1) (Whelan and Goldman 2001). Phylogenetic analyses using these models with either nucleotide or amino acid data produced similar trees, with only minor differences in the relationships of tip taxa within well-supported clades. Both sets of analyses placed the novel crustacean opsin sequences with previously characterized crustacean sequences in a monophyletic clade sister to the insect MWS/LWS clade (fig. 3). Within each insect spectral clade, species cluster roughly by higher-order taxonomy (i.e., Lepidoptera, Hymenoptera, and Diptera). In contrast, although the crustacean clade is strongly supported (BP = 98; pP = 1.00), the currently sampled taxa exhibit very little taxonomic clustering with generally low branch support values. For example, neither the decapod species (crayfish lineage + H. gammarus) nor the 4 mysid species form monophyletic groups. The 3 genes sequenced from the stomatopod N. oerstedii do cluster, although branch support is low for the placement of N. oerstedii Rh2 (BP < 50, pP = 0.89). Interestingly, the N. oerstedii opsins do not cluster by spectral sensitivity, with strong support for the Rh1 (the corresponding visual pigment has max = 489 nm; Cronin and Marshall 1989) + Rh3 (max = 522 nm; Cronin and Marshall 1989) clade (BP = 100, pP = 1.00). Furthermore, the opsin located in the N. oerstedii peripheral ommatidia (Rh2) that is most similar to other crustacean compound eyes contains the most divergent opsin sequence (40.9–48.3% amino acid difference) relative to other crustacean LWS opsins.
|
The general topology of this tree demonstrates the presence of a monophyletic arthropod LWS clade containing representatives from the Hexapoda, Crustacea, and Chelicerata (BP = 100; pP = 1.00). Curiously, within the Diptera, there has been a gene duplication event leading to the formation of a distinct MWS clade, representatives of which have not yet been found in other insects (Spaethe and Briscoe 2004
). A similar situation exists within the Crustacea, with the only sequenced MWS opsin (Hemigrapsus sanguineus; Sakamoto et al. 1996) falling outside of the main arthropod LWS clade, indicating an earlier gene duplication event; however, sequence data from other crustaceans are necessary to confirm that this gene is not a copy unique to H. sanguineus. Sister to the arthropod LWS/MWS is a monophyletic insect SWS clade (BP = 100; pP = 1.00), containing well-supported lineages representing both blue (BP = 96, pP = 1.00) and UV (BP = 99, pP = 1.00) spectral variants. Finally, sister to the arthropod visual pigment clade is the cephalopod lineage (BP = 100, pP = 0.99).
Selective Influences in Pancrustacean Opsins
Analyses of dN/dS ratios using PAML did not detect any sites under positive selection, and none of the models incorporating selection parameters contained site classes with > 1 or were significantly better than neutral models based on likelihood ratio tests. In contrast, TreeSAAP analyses identified 4 amino acid properties to be under positive destabilizing selection in our opsin data set (P < 0.001): coil tendencies, compressibility, power to be at the middle of the -helix, and refractive index. Evaluation of various sliding window sizes indicated that a window of 20 amino acids provided the best signal-to-noise ratio. Sliding window analyses using a 20 amino acid window size identified that selection on these properties occurred primarily in the transmembrane domains (fig. 4); however, the distribution of sites under selection within each domain differed between properties. For example, the majority of sites across the phylogeny exhibiting selection for refractive index were found in TMII (30.6%), TMIII (13.9%), CL3 (22.2%), and extracellular loop (EL) 2 (19.4%), whereas those found for coil tendencies were in CL2 (32.3%), TMV (14.9%), TMIII (11.5%), and EL2 (10.3%).
|
Using a high-resolution model of bovine rhodopsin (1LH9.pdb), the sites identified to be in regions of the protein under destabilizing selection were mapped to opsin (fig. 5). For compressibility, sites were concentrated on the extracellular end of TMI and in CL2 and scattered throughout TMIII, TMIV, and TMVI. Interestingly, every single amino acid change under selection identified for compressibility was a change to an alanine. Outside of the transmembrane domains, coil tendencies and power to be at the middle of the
-helix also identified sites within CL2 and coil tendencies, power to be at the middle of the -helix, and refractive index sites within EL2. Using features of the program Swiss-PdbViewer v.3.7 (http://www.expasy.org/spdbv; Guex and Peitsch 1997), selected residues were evaluated for their proximity to the chromophore-binding pocket using bovine rhodopsin as a reference sequence. Using a conservative distance of 4 Å to infer a residue's potential interaction with the chromophore, 10 sites were isolated from those identified by TreeSAAP analyses: 113 (compressibility), 117 (compressibility), 118 (compressibility), 121 (coil tendencies), 122 (compressibility), 186 (power to be at the middle of the -helix), 187 (power to be at the middle of the -helix, refractive index), 189 (coil tendencies, power to be at the middle of the -helix), 207 (coil tendencies, power to be at the middle of the -helix), and 265 (compressibility). These residues are clustered in 2 areas of the protein in TMIII (113–122) and EL2 (186–207). In addition to these sites, TreeSAAP analyses identified several residues—90 (compressibility, power to be at the middle of the -helix), 123 (coil tendencies, compressibility), 164 (compressibility), and 274 (compressibility, power to be at the middle of the -helix)—which are within 10 Å of the chromophore and have been identified in other studies of spectral tuning (Wilkie et al. 2000; Briscoe 2002; Salcedo et al. 2003).
|
|
Discussion |
|---|
|
TOPAbstractIntroductionMethodsResultsDiscussionSummaryAcknowledgementsReferences |
|---|
Pancrustacean Opsin Evolution
Almost all investigated insect opsins fall within 3 spectral clades—long-wavelength, UV, and blue sensitive. These genetic clades fit well with physiological characterizations of spectral sensitivity in insects, supporting the hypothesis that the ancestral visual system was trichromatic (Chittka 1996
, 1997). However, although the ancestral state may have been trichromatic, the current diversification of opsin genes appears to be more complex. For example, the complement of 6 visual pigments described from Drosophila represent 3 gene duplications that most likely occurred only in the Dipteran lineage (2 duplications leading to the MWS Rh1 and Rh2 and 1 duplicating a UV gene leading to Rh3 and Rh4). More recent studies have identified similar gene complement expansion in insect LWS opsins within specific lineages (Briscoe 1998, 2000; Hill et al. 2002) and early within insect diversification (Spaethe and Briscoe 2004). Although the expression patterns and spectral sensitivities of most of these additional LWS opsin genes have not been investigated, recent studies suggest that at least some of these copies may have an extraocular expression and potentially be involved in circadian regulation systems (Shimizu et al. 2001; Briscoe and White 2005). Similarly, investigations of ostracod opsins have documented differential expression in the median versus the compound eyes as well as the presence of at least 8 expressed SWS loci, indicating many recent gene duplications (Oakley and Huber 2004). Given the demonstrated high copy number of expressed LWS genes in insects and SWS genes in ostracods, a similar pattern of gene expansion is likely to be documented in the crustacean LWS clade once more sequence data are obtained. The low taxonomic clustering observed in the crustacean clade also suggests the presence of unidentified gene duplication events occurring at least before the divergence of the Mysida and Decapoda. Interestingly, however, the crustaceans form 2 main clades that are well supported by BMCMC analyses (pP > 0.98). With the exception of N. oerstedii Rh1 (489 nm), these 2 clades are roughly divided by spectral sensitivities, with 1 shorter-wavelength clade (496–501 nm) composed of M. diluviana, A. grebnitzkii, and E. superba opsins and 1 longer-wavelength clade (512–533 nm) with sequences from the remaining species. A particularly interesting crustacean group to conduct further investigations of opsin evolution will be the stomatopod crustaceans, here represented by the species N. oerstedii, which contain up to 16 different visual pigments that span the ultraviolet to visible spectrum of light (Cronin and Marshall 1989, 2004; Cronin et al. 2000). Characterization of additional LWS opsin sequences will be required to elucidate the validity of the observed patterns of evolution in crustaceans. An unusual feature of arthropod opsin evolution is the large indel present in the CL3 domain. It exists in all the opsins investigated in this study, with the most conserved stretch taking the form of R(E/D)QAKKM(N/G). Because the function of the CL3 domain has been linked to G protein docking and signal transfer, and a relatively large proportion of its sites were identified to be under destabilizing positive selection for compressibility (16.1%) and refractive index (22.2%) (fig. 4), this conserved region should be of interest to future comparisons of opsin functionality between "invertebrates" and vertebrates.
Selective Forces in Pancrustacean Opsins
The ability to study the genetic mechanisms behind phenotypic variation has made opsin a model evolutionary system for studying the selective influences leading to genetic adaptation. The combination of genetic, physiologic, and biochemical studies have shown a direct correlation between the environment and visual pigment spectral sensitivity (Lythgoe 1972, 1980; Crescitelli et al. 1985; Partridge et al. 1988, 1989; Partridge 1989; Douglas et al. 1998) and demonstrated the effects of single amino acid changes on tuning this spectral sensitivity (Nathans 1990a; Neitz et al. 1991; Asenjo et al. 1994; Yokoyama and Radlwimmer 1999; Wilkie et al. 2000; Yokoyama et al. 2000; Yokoyama and Radlwimmer 2001; Cowing et al. 2002; Yokoyama and Tada 2003), making opsins one of the clearest, and best-studied, protein groups for the quantification of adaptation at the molecular level. Because the observed variation in visual pigment spectral absorbance is the result of adaptive selection on the opsin gene, this gene family is therefore an excellent system for testing methods of detecting how selection acts on protein sequences. In our analyses, estimated dN/dS ratios did not detect any evidence of selection; however, these results only confirm that the opsin gene is a generally conservative protein-coding gene that requires the use of alternative criteria to reveal molecular adaptations (McClellan et al. 2005). Using a method that evaluates only nonsynonymous replacements with respect to the relative change in a suite of amino acid properties, 4 properties were identified to be under positive destabilizing selection across the phylogeny: coil tendencies, compressibility, power to be at the middle of the -helix, and refractive index. Interestingly, none of the identified properties measure aspects of polarity or charge, 2 of the properties hypothesized to be important in residues affecting vertebrate spectral tuning (Nathans 1987, 1990b; Neitz et al. 1991; Chan et al. 1992). Instead, the identified properties mostly relate to structural aspects. Thus, although polarity and charge may be important in fine-tuning spectral absorbance, in a broad-scale evolutionary context, structural aspects of residue replacement seems to have been more important to overall opsin function. In particular, compressibility seems to be a significant property impacting opsin evolution, with 186 nonsynonymous replacements identified from across the phylogeny, with all replacements being to the residue alanine. Alanine is 1 of 7 amino acids identified by Gromiha and Ponnuswamy (1993) that determine overall protein compressibility, and, with the exception of proline, has the smallest calculated compressibility (–25.5 x 10–15 m3 mol–1 Pa–1). Previous studies have indicated that alanine replacements can stabilize -helices and have implicated alanine content in helical thermal stability (Argos et al. 1979; Lyu et al. 1990; O'Neil and DeGrado 1990; Padmanabhan et al. 1990; Zhang et al. 1991; Ptitsyn 1992; although see Pinker et al. 1993 for a different opinion). The large proportion of nonsynonymous replacements in TMI (10.2%), TMIII (17.7%), TMIV (18.3%), TMV (3.8%), and TMVI (13.4%) indicates that -helical stability may play an important role in pancrustacean opsin evolution. The exclusive use of alanine over proline at these sites is most likely due to the fact that proline residues introduce bends into -helices (Riek et al. 2001).
Most amino acid sites corresponding to the 4 identified properties occur in transmembrane domains (fig. 4). Based on site-directed mutagenesis studies in vertebrates, transmembrane domains III, VI, and VII are known to have significant interaction with the chromophore, and interactions among these 3 helices are involved in restraining the structure of GPCRs in the inactive, nonsignaling state (Sakmar et al. 1989; Zhukovsky and Oprian 1989; Nathans 1990a, 1990b; Nakayama and Khorana 1991; Chan et al. 1992; Filipek et al. 2003). Additionally, the cytoplasmic ends of helices III, VI, and VII move during GPCR activation to create a binding crevice for the G protein (Filipek et al. 2003). Unfortunately, the selective influences in TMVII could not be evaluated in this study because many of the available opsin sequences are too short. However, the fact that 3 of the properties (coil tendencies, compressibility, and power to be at the middle of the -helix) had significant amounts of detected selection in TMIII indicates that this helix may play a crucial role in the functional diversification of arthropod opsins.
Using a cutoff distance of 4 Å, 10 sites from the TreeSAAP analyses were identified that potentially affect the chromophore-binding pocket. This criterion is very conservative given that residues at distances of up to 10 Å have been identified as regulating chromophore wavelength absorption in other studies of spectral tuning (Wilkie et al. 2000; Briscoe 2002; Salcedo et al. 2003). In the bovine rhodopsin structure, 18 residues are within 4 Å of the chromophore; we detected destabilizing selection at 55.6% of those residues. Comparisons of these sites with studies using evolutionary trace (ET) analyses to identify transmembrane residues of importance to class A GPCRs in general and to opsins specifically indicate that at least 2 of our identified residues (113 and 207) correspond to previously identified residues specific to opsin function, whereas an additional 3 residues (118, 121, and 265) are of general importance to GPCR functioning (Madabushi et al. 2004). Our analyses identified an additional 2 transmembrane residues (117 and 122) to be under destabilizing selection, which were not found using the ET methods, as well as 3 residues located in EL2, which were not included in the ET analyses. This comparison illustrates that TreeSAAP methods are capable of detecting selection at residues of importance to general GPCR- and opsin-specific functioning, while also finding novel sites not predicted using other methods. Further TreeSAAP analyses–comparing sites found in opsins versus broader sets of class A GPCRs are needed to tease apart the residues that are most important to opsins.
The 10 residues identified to be under destabilizing selection and within 4 Å of the chromophore are clustered in the extracellular end of TMIII and in EL2. The EL2 folds back into the cavity formed by the membrane-embedded domains, forming part of the chromophore-binding pocket and acting as a "plug" preventing solvent access to the Schiff base (Yan et al. 2003). Recent studies have shown this loop is important for the thermal stability of the dark state of rhodopsin (Janz et al. 2003), and in particular, E181 in bovine rhodopsin affects the stability and spectral absorption of metarhodopsin II (Yan et al. 2003), leading to the hypothesis that E181 is a counterion for metarhodopsin I (Teller et al. 2003; Yan et al. 2003). Interestingly, although the rhodopsin counterion E113 is not conserved outside of vertebrates, the E181 residue is conserved, implying that there may be more similarities in metarhodopsin dynamics between nonvertebrates and vertebrates than in rhodopsin spectral tuning. All of the sites identified in this region (186, 187, and 189) demonstrated positive destabilizing selection for power to be at the middle of the -helix, suggesting that similar forces are acting upon adaptation of residues in the EL2 "plug."
The sites clustered at the extracellular end of TMIII (113, 117, 118, 121, and 122) were demonstrated to be under selection for compressibility. These sites are all on turns of the -helix that face the chromophore-binding pocket, and several of these sites have been identified as affecting spectral tuning in vertebrates (Nathans 1990a, 1990b; Yokoyama et al. 1999; Yokoyama and Tada 2000; Shi et al. 2001; Yokoyama and Tada 2003). Intriguingly, site 113 (the vertebrate counterion site) was identified to be under positive destabilizing selection in this data set. This site is occupied by either a tyrosine in visible-absorbing or a phenylalanine in UV-absorbing arthropod pigments. However, previous studies have shown this site is not used as a counterion (Nakagawa et al. 1999) and that the observed amino acid polymorphism is not responsible for the difference in absorption spectra between UV and visible pigments (Salcedo et al. 2003). However, the identification of the site in this study implies that the observed polymorphism serves unidentified function in visible versus UV opsins.
In addition to the sites in TMIII and EL2, several residues (90, 123, and 274) identified in our analyses to be under positive destabilizing selection for more than 1 property are within 10 Å of the chromophore and have been identified in other studies of spectral tuning (Wilkie et al. 2000; Briscoe 2001, 2002; Salcedo et al. 2003). The identification of these sites confirms that amino acids affecting spectral tuning are selected for impacts on structural aspects of the helices (in this study, compressibility, power to be at the middle of the -helix, and coil tendencies). These amino acid properties in particular may affect the internal packing of the chromophore-binding site and thereby affect spectral tuning of the chromophore and signal transduction by the opsin protein.
|
Summary |
|---|
|
TOPAbstractIntroductionMethodsResultsDiscussionSummaryAcknowledgementsReferences |
|---|
In this study, we contribute to the knowledge of "pancrustacean" opsin evolution by identifying amino acid sites within the chromophore-binding pocket under selection. Techniques investigating amino acid property changes detected 4 amino acid properties (coil tendencies, compressibility, power to be at the middle of an
-helix, and refractive index) to be historically influenced by destabilizing positive selection. Ten amino acid sites relating to these properties were found to be facing the binding pocket, within 4 Å of the chromophore, with potential to affect spectral tuning. Comparisons with previous studies investigating opsin functional residues indicate that some of these sites are specific to opsin functioning, some are general to GPCR functioning, and some are novel residues not previously identified. This research demonstrates the ability of TreeSAAP to identify amino acid residues of functional importance in an evolutionary context, providing a method for targeting sites for further studies, particularly in proteins that have not yet been structurally determined.
|
Acknowledgements |
|---|
|
TOPAbstractIntroductionMethodsResultsDiscussionSummaryAcknowledgementsReferences |
|---|
Many thanks to T. Frank and T. Targett and members of their laboratories for help in acquiring specimens. We would like to thank T. Oakley and an anonymous reviewer for insightful comments on how to improve this manuscript, and A. Brown for permission to use the N. oerstedii sequences. This study was supported by grants from the National Science Foundation (DEB-0206537 to M.L.P. and K.A.C. and IBN-0235820 to T.W.C.).
|
Footnotes |
|---|
Adriana Briscoe, Associate Editor
|
References |
|---|
|
TOPAbstractIntroductionMethodsResultsDiscussionSummaryAcknowledgementsReferences |
|---|
Abascal F, Zardoya R, Posada D. (2005) ProtTest: selection of best-fit models of protein evolution. Bioinformatics 21:2104–2105.
Anisimova M, Bielawski JP, Yang Z. (2002) Accuracy and power of Bayes prediction of amino acid sites under positive selection. Mol Biol Evol 19:950–958.
Arendt D, Tessmar-Raible K, Snyman H, Dorresteijn AW, Wittbrodt J. (2004) Ciliary photoreceptors with a vertebrate-type opsin in an invertebrate brain. Science 306:869–871.
Argos P, Rossmann MG, Grau UM, Zuber H, Frank G, Tratschin JD. (1979) Thermal stability and protein structure. Biochemistry 18:5698–5703.[CrossRef][Medline]
Arikawa K, Inokuma K, Eguchi E. (1987) Pentachromatic visual system in a butterfly. Naturwissenschaften 74:297–298.[CrossRef][ISI]
Arikawa K, Scholten DGW, Kinoshita M, Stavenga DG. (1999) Tuning of photoreceptor spectral sensitivities by red and yellow pigments in the butterfly Papilio xuthus. Zool Sci 16:17–24.[Medline]
Asenjo AB, Rim J, Oprian DD. (1994) Molecular determinants of human red/green color discrimination. Neuron 12:1131–1138.[CrossRef][ISI][Medline]
Bellingham J, Wells DJ, Foster RG. (2003) In silico characterisation and chromosomal localisation of human Rrh (peropsin)—implications for opsin evolution. BMC Genomics 4:3.[CrossRef][Medline]
Berman HM, Westbrook J, Feng Z, Gilliland G, Bhat TN, Weissig H, Shindyalov IN, Bourne PE. (2000) The protein data bank. Nucleic Acids Res 28:235–242.
Bielawski JP and Yang Z. (2004) A maximum likelihood method for detecting functional divergence at individual codon sites, with application to gene family evolution. J Mol Evol 59:121–132.[ISI][Medline]
Bracho MA, Moya A, Barrio E. (1998) Contributions of Taq polymerase-induced errors to the estimation of RNA virus diversity. J Gen Virol 79:2921–2928.[Abstract]
Briscoe AD. (1998) Molecular diversity of visual