Molecular Evolution of Human Visual Pigment Genes’

Shozo Yokoyama and Ruth Yokoyama Department of Ecology, Ethology, and Evolution, University of Illinois at Urbana-Champaign

By comparing the published DNA sequences for (a) the genes encoding the human visual color pigments (red, green, and blue) with (b) the genes encoding human, bovine, and Drosophila rhodopsins, a phylogenetic tree for the mammalian pigment genes has been constructed. This evolutionary tree shows that the common ancestor of the visual color pigment genes diverged first from that of the rhodopsin genes; then the common ancestor of the red and green pigment genes and the ancestor of the blue pigment gene diverged, and finally the red and green pigment genes diverged from each other much more recently. Nucleotide substitutions in the rhodopsin genes are best explained by the neutral theory of molecular evolution. However, important functional adaptations seem to have occurred twice during the evolution of the color pigment genes in humans: first, to the common ancestor of the three color pigment genes after its divergence from the ancestor of the rhodopsin gene and, second, to the ancestor of the red pigment gene after its divergence from that of the green pigment gene.

Introduction

In human eyes, two kinds of photoreceptor cells, cones and rods, exist. Cones function in bright light and are responsible for color vision, whereas rods function in dim light and do not perceive color. The photosensitive molecule, or visual pigment, in rods is rhodopsin, which consists of a protein, opsin, that is covalently linked to 1 I-cis retinal via a protonal Schiff ‘s base linkage between the e-amino group of a lysine and the aldehyde of retinal ( Wald 1968). Color vision is mediated by three different photoreceptors, i.e., green, red, and blue visual pigments, each containing 1 I-cis retinal. Vision begins when a photon is absorbed by a visual pigment and isomerizes retinal from the 11-cis retinal to the all-trans retinal configuration. A late step in visual excitation in dim light is the hyperpolarization of the plasma membrane (Nathans 1987).

All pigment absorption spectra have about the same characteristic bell shape, and each pigment is specialized by its wavelength of maximal absorption. The three pigments that mediate human color vision have absorption maxima at -420 nm (the blue-sensitive pigment ) , - 530 nm (the green-sensitive pigment), and - 560 nm (the red-sensitive pigment), and the rhodopsin absorbs maximally at -495 nm (Nathans 1987). The genes encoding the rhodopsin in bovine (Nathans and Hogness 1983) and in human (Nathans and Hogness 1984) have been sequenced, and there is 93% identity between the two deduced amino acid sequences. Recently, the genes encoding the blue, green, and red pigments in human have also been sequenced (Nathans et

1. Key words: rhodopsin, color pigment genes, molecular evolution.

Address for correspondence and reprints: Dr. Shozo Yokoyama, Department of Ecology, Ethology, and Evolution, University of Illinois at Urbana-Champaign, Shelford Vivarium, 606 East Healey Street, Champaign, Illinois 6 1820.

Mol. Biol. Evol. 6(2): 186-197. 1989. 0 1989 by The University of Chicago. All rights reserved. 0737-4038/89/0602-0006$02.00

186

Evolution of Color Vision 187

al. 1986b). The color pigments have -40% amino acid identity with the human rhodopsin (Nathans and Hogness 1984), and the red and green pigments show 96% identity to each other but only 43% identity with the blue pigment (Nathans et al. 19863), On the basis of these comparisons, Nathans et al. (1986b) suggested that short-wavelength (blue), long-wavelength (red and green), and rod pigments all di- verged from a common ancestor about the same time >500 million years ago (Mya).

In the present paper, we construct a phylogenetic tree of the color pigment genes and rhodopsin genes in human and of the rhodopsin gene in bovine. Then, we evaluate the lengths of all branches of the evolutionary tree constructed. The analyses provide information on the sequence of the gene duplications, the evolutionary rates of nu- cleotide and amino acid substitutions, and possible selective pressure for acquiring a new function at one of the duplicate loci. Specific amino acid changes during the evolution of the visual color pigment genes can also be identified at many sites. These changes provide information on the cause of the different spectral absorptions of the pigments.

DNA Sequence Data of Visual Pigment Genes

The DNA sequences used in the analyses were the blue (Bu), green (Gu), and red (Rn) pigment genes from human (Nathans et al. 1986a) and the rhodopsin genes from human (Rhu) (Nathans and Hogness 1984), bovine (RUB) (Nathans and Hogness 1983), and Drosophila melanogaster (Rhb) (O’Tousa et al. 1985; Zuker et al. 1985). The Rhb gene is useful as the outgroup not only in constructing the rooted evolutionary tree but also in estimating all of the branch lengths. Two Rhb genes, which show the identical nucleotide sequence, were isolated and characterized by two different groups using the same genomic library (Maniatis et al. 1978). The homologous coding regions of these visual pigment genes with different lengths are compared. Both the red and green pigments consist of 364 amino acids, whereas the blue pigment consists of 348 residues. The rhodopsins of human and bovine are 348 residues long, and that of Drosophila contains 373 residues. To compare these DNA sequences, we have followed the alignment of Nathans et al. (19863) for the mammalian visual pigment genes and that of Zuker et al. (1985) for the rhodopsin genes from bovine and Drosophila.

The Phylogenetic Tree of the Mammalian Visual Pigment Genes

The proportion of different nucleotides for each pair of nucleotide sequences (p) and the corresponding numbers of nucleotide substitutions per nucleotide site (d) are shown for the first, second, and third codon positions separately (table 1). The values of d are obtained from the relationship d = -(3/s)ln[l - (4/3)p], which assumes equal substitution rates among the four types of nucleotides (Jukes and Cantor 1969 ) . The d values obtained by other methods (Kimura 198 1; Takahata and Kimura 198 1; Gojobori et al. 1982) did not differ much from these.

In constructing the phylogenetic tree of the different genes, the neighbor-joining (NJ) method (Saitou and Nei 1987) was used with four sets of data, i.e., the distance matrices from the first, second, third, and all codon positions. These different data sets consistently gave the same evolutionary tree. Figure 1 shows the tree, with branch lengths as estimated by considering all codon positions. A phylogenetic tree of the different genes has also been constructed by applying the unweighted pairwise group with arithmetic mean (UPGMA; e.g., see Nei 1987) to all codon positions. The tree topology obtained was the same as that in figure 1.

Two characteristics of the tree topology are worth mentioning. First, the color

188 Yokoyama and Yokoyama

Table 1 p and d for the Three Codon Positions

GENE AND CODON

POSITION RH GH

GENES COMPARED

BH RhH MB fiD

RH: First ........ Second. ..... Third . . , , . , .

GH: First. ....... 0.03 + 0.009 Second. ..... 0.02 + 0.007 Third. ...... 0.02 + 0.008

BH: First ........ 0.66 + 0.064 Second.. .... 0.50 f 0.05 1 Third ....... 1.02 + 0.100

RhH : First. ....... 0.76 + 0.074 Second.. .... 0.49 + 0.050 Third. ...... 0.93 + 0.009

RhB: First. ....... 0.77 + 0.076 Second. ..... 0.48 f 0.050 Third ....... 0.86 f 0.085

RhD: First ........ 1.25 f 0.137 Second ...... 1.09 kO.112 Third ....... 1.20 f 0.130

101364 151/345 1651345 1661345 2141352 61364 1261345 1251345 1231345 2031352 91364 1921345 1841345 1761345 2 1 l/352

0.64 f 0.063 1581345 1581345 2201347 0.49 f 0.050 1231345 1221345 2081347 0.99 f 0.101 1931345 1821345 2 lo/347

0.73 + 0.071 0.71 + 0.069 201348 2071348 0.48 + 0.049 0.48 I!I 0.049 9/348 1821348 0.96 f 0.097 1.03 + 0.105 791348 2031348

0.75 f 0.074 0.7 1 + 0.068 0.06 Z?I 0.014 2051348 0.47 + 0.048 0.47 Z!I 0.048 0.03 f 0.093 1801348 0.88 f 0.088 0.91 Z!I 0.089 0.27 f 0.032 1951348

1.26 AZ 0.140 1.40 + 0.167 1.18 + 0.127 1.15 + 0.123 1.09 f 0.111 1.20 + 0.131 0.90 + 0.087 0.87 + 0.085 1.16 f 0.124 1.23 Z!I 0.131 1.13 f 0.123 1.03 + 0.105

1491345 1611345 1631345 2151352 1241345 1221345 120/345 2021352 1901345 1871345 1791345 2081352

NOTE.-values above the diagonal are p values; those below the diagonal are d values.

pigment genes and the rhodopsin genes have evolved independently from each other since the divergence of their common ancestral genes. The divergence apparently occurred much earlier than the divergence of the two species human and bovine. Second, among the color pigment genes, the common ancestor of the RH and GH diverged from the ancestor of the BH first, and then the Rn and Gn diverged from each other much more recently.

The tree topology in figure 1 generally supports Nathans et al.‘s ( 19863) proposal that short-wavelength, long-wavelength, and rod pigment genes all diverged from a common ancestor about the same time and that the green and red pigment genes diverged much more recently. However, there is one important difference. The present analysis shows that the common ancestor of the color pigment genes and that the rhodopsin genes diverged first and that the three different color pigment genes evolved after that. The distinction between the evolutionary tree proposed by Nathans et al. ( 1986b) and that of figure 1 becomes critical in studying the biological implication of evolutionary rates for different branches, as will be seen in the next section.

Branch Lengths for the Evolutionary Tree of the Visual Pigment Genes

The Rho becomes necessary as the outgroup in evaluating the lengths of the branches A-B and A-D in figure 1. The branch lengths are estimated by using the four sets of distance data which distinguish different codon positions, as we did before.

A I

0.31

C 0.02 [--

RH

0.10

0.37

- 0.01

GH

0.07

r

RhH

0.20

I D

1

0.04

RhB

FIG. I.-Phylogenetic tree constructed for the visual pigment genes by using nucleotide substitutions at all codon positions. The rhodopsin gene of Drasophilu is used as the outgroup in constructing the rooted

190 Yokoyama and Yokoyama

Table 2 Branch Lengths for the Evolutionary Tree of the Visual Pigment Genes, Estimated by Using the Distances from the First, Second, Third, and All Codon Positions

COD~N POSITION

BRANCH First Second Third All

A-B ........ B-C ........ C-R,, ....... C-GH ....... B-B,., ....... A-D ........ D-RhH. ..... D-RhB ......

0.12 f 0.038 0.12 + 0.034 0.03 f 0.038 0.10 zk 0.021 0.29 + 0.034 0.22 & 0.028 0.46 f 0.049 0.31 + 0.021 0.02 f 0.008 0.02 f 0.007 0.01 + 0.006 0.02 f 0.004 0.01 z?I 0.004 0.00 & 0.002 0.01 + 0.006 0.01 + 0.002 0.35 f 0.038 0.27 Z!I 0.033 0.53 f 0.052 0.37 f 0.023 0.25 f 0.033 0.10 f 0.018 0.27 f 0.041 0.20 f 0.017 0.03 * 0.010 0.02 f 0.008 0.18 + 0.025 0.07 * 0.009 0.03 f 0.010 0.01 f 0.005 0.09 + 0.017 0.04 +- 0.006

In figure 1, the branch lengths estimated by using the nucleotide substitutions at all codon positions are shown.

Branch lengths and associated SEs were computed separately by distinguishing the first, second, third, and all codon positions (table 2). If we consider the lengths estimated by using all codon positions, total lengths for the branches A-Rhn and A- RhB are 0.27 and 0.24, respectively, whereas the lengths for the branches A-RH, A- Gn, and A-Bu are 0.43,0.42, and 0.47, respectively. All six sets of comparison show that the color pigment genes are evolving significantly faster than the rhodopsin genes. When a series of averages is considered, the total branch length for the color pigment genes from the branch point A is 0.44 (= [(0.02 + 0.01)/2 + 0.31 + 0.37]/2 + O.lO), whereas that for the rhodopsin genes is 0.25, showing that the former is evolving about 1.7 times faster than the latter.

It is also possible to compare branch lengths for the color pigment genes and for rhodopsin genes. When all codon positions are considered, the total lengths of the branches B-Rn, B-Gn, and B-Bn are 0.33, 0.32, and 0.37, respectively, which are not significantly different from each other. The lengths of C-Rn and C-Gn are 0.02 and 0.0 1, respectively, which are significantly different (P < 0.05). Similarly, the lengths of the branches D-Rhn and D-RhB are 0.07 and 0.04, respectively, which are also significantly different (P < 0.05). Thus, when all nucleotide substitutions are considered, evolutionary rates differ significantly between the RH and Gu and between the RhH and RhB after the divergence from their common ancestors. These differential rates of evolution, however, should not be related automatically to the acquisition of new functions. As shown below, biological implications of the evolutionary rates for the color pigment genes and rhodopsin genes are very different.

In six of the eight branches, the length is longest when the distance of the third codon position is used, is shorter when that of the first codon position is used, and is shortest when the second codon position is used. This is expected, because nucleotide substitutions at the second positions tend to produce more drastic changes in the physicochemical properties of amino acids than do those at the first positions and because a majority of mutational changes at the third position do not cause amino acid changes ( Kimura 1983 ) .

Two exceptions to this rule are the branches A-B and C-Rn, where the branch length for the third codon position is the shortest (table 2). Note that these branches

Evolution of Color Vision 19 1

Table 3 Branch Lengths for the Evolutionary Tree of the Visual Pigment Genes, Estimated by Using the Distances from Synonymous and Nonsynonymous Substitutions

SUBSTITUTION WEIGHTED

BRANCH Synonymous Nonsynonymous AVERAGE

A-B ........ B-C ........ C-R,., ....... C-GH ....... B-BH ....... A-D ........ D-Rh,, ...... D-RhB ......

-0.05 & 0.062 0.10 + 0.022 0.09 0.79 I?I 0.097 0.23 f 0.020 0.36 0.01 f 0.008 0.02 Ik 0.004 0.01 0.03 + 0.009 0.00 + 0.002 0.01 0.96 zk 0.1 I’2 0.29 It 0.022 0.44 0.35 f 0.064 0.18 f 0.017 0.22 0.33 + 0.045 0.02 f 0.005 0.09 0.11 & 0.023 0.01 f 0.004 0.03

are the points when the common ancestor of the color pigment genes diverged from that of the rhodopsin genes and when the ancestor of the red pigment gene diverged from that of the green pigment gene. Thus, there is a unique opportunity for positive selection of new functions, i.e., absorption of new wavelengths, along these two branches. One way to examine such a possibility is to compare the rates of synonymous and nonsynonymous substitutions directly.

The numbers of synonymous and nonsynonymous substitutions were estimated by using the methods of Miyata and Yasunaga ( 1980) and Nei and Gojobori ( 1986). The two methods gave very similar results, and therefore the distances obtained by the former method were again applied to the NJ method. Results are shown in table 3, together with the weighted average. It should be noted that the weighted averages are generally very close to the branch lengths estimated by using all codon positions (see table 2, fig. 1). As may be expected on the basis of the results in table 2, in six of the branches the length estimated by synonymous substitution is always significantly longer than that estimated by nonsynonymous substitution. However, in the branch A-B, nonsynonymous substitution produces a significantly longer branch length than does synonymous substitution. In the branch C-Ru, the length estimated by nonsy- nonymous substitution is longer than that estimated by synonymous substitution, but the difference is not statistically significant.

It should be noted that the negative value obtained for synonymous substitution in the branch A-B seems due to sampling error rather than to an error in the tree topology. This is because the topology in figure 2 is supported consistently by different data sets, as already noted. Recently, sequence homology has been demonstrated be- tween the visual pigment proteins, G-protein coupled receptors, and the human mas oncogene (e.g., see Nathans 1987). The G-protein coupled receptors includes the adrenergic and muscarinic cholinergic receptors. The phylogenetic analysis of these genes supports the tree topology of figure 1 (Yokoyama et al., submitted).

When the distance matrix for the number of synonymous substitutions is US&, the total lengths for branches A-Ru, A-Gu, A-B H, A-Rb, and A-R~B are 0.75 f 0.116,0.77 f 0.119,0.91 + 0.128,0.68 & 0.078, and 0.46 f 0.068, respectively. In the comparison of the total branch lengths between the color pigment genes and rhodopsin genes, only those branch lengths including RhB are significantly different and the color pigment genes are evolving faster than the bovine rhodopsin gene. HOW-

Evolution of Color Vision 193

These results suggest that the modes of evolution for the rhodopsin genes and color pigment genes differ. During the evolutionary processes of the two rhodopsin genes, the number of synonymous substitutions is significantly larger than that of nonsynonymous substitutions. This is consistent with the neutral theory of molecular evolution (Kimura 1983 ), which maintains that nonsynonymous substitution is more functionally constrained than synonymous substitutions. A similar functional con- straint can also be seen in the branches B-Bn, B-C, and C-GH. Thus, a large portion of the evolutionary process of the human color pigment genes may also be explained by the neutral theory. However, two branches A-B and C-Rn show a larger number of nonsynonymous substitutions than of synonymous substitutions. This may imply that positive selection for absorption at a new wavelength occurred twice in the past.

These conclusions have been obtained by using the Rhn gene as the outgroup. The same conclusion can be reached when G-protein coupled receptor genes are used as the outgroup. For example, consider porcine muscarinic receptor genes (Kubo et al. 1986; Kobilka et al. 1987) and P-adrenergic receptor genes in hamster (Dixon et al. 1986), turkey (Yarden et al. 1986), and human (Frielle et al. 1986), rather than Rhn . Then, the number of synonymous and nonsynonymous substitutions for branch A-B is 0.07 + 0.0 18 and 0.12 f 0.0 14, respectively. The respective numbers for branch D-Rhu are 0.00 + 0.001 and 0.02 f 0.005. The difference at each branch is statistically significant (P < 0.05 for both). In all other branches, the number of synonymous substitutions is larger than that of nonsynonymous substitutions.

Discussion

Although the existence and number of color pigment genes seem to vary among different species, the rhodopsin gene exists over a wide range of species. For example, when a bovine rhodopsin complementary DNA probe has been used, the homologous DNA segments have been detected in vertebrate and invertebrate organisms, in a unicellular alga, and even in an archaebacterium (Martin et al. 1986). Since archae- bacteria and unicellular algae seem to have already coexisted - 1.5 billion years ago (e.g., see Schopf et al. 1983)) the rhodopsin gene must have originated in the very early stage of the evolution of life.

Assuming that the amino acid sequence divergence between human and bovine rhodopsins is 1 Om9/ site/year, Nathans et al. ( 1986b) estimated the point of the di- vergence of short-wavelength, long-wavelength, and rod pigments to be >500 Mya. Using the branch lengths estimated in the present analyses, we evaluated approximate divergence times of the points A, B, and C in figure 1. For that purpose, the lengths estimated by the numbers of nonsynonymous substitutions were used, where the lengths of the branches D-Rhn (0.02 + 0.005) and D-RhB (0.01 + 0.004) do not differ significantly from each other. The time of mammalian radiation is considered to be 60-80 Mya (e.g., see Nei 1987). If we take point D in figure 1 as being 75 Mya, then the rates of nonsynonymous substitution per site per year for Rhn and RhB are 0.32 X 10 -9 and 0.16 X 10 -9, respectively, giving an average value of 0.24 X 10 -9. The total branch length for the rhodopsin gene is 0.18 + (0.02 + 0.0 1) / 2 = 0.20 (see table 3). Thus, if we assume that the rate of nonsynonymous substitution along A-RhH or A-RhB is 0.24 X 10s9, the divergence time of point A in figure 1 is 0.20/( 0.24 x 10-9) = 833 Mya. Similarly, when a series of averages is considered, the total branch length, from branch point A, for the color pigment genes is [ (0.02 + 0.00)/2

194 Yokoyama and Yokoyama

+ 0.23 + 0.29]/2 + 0.10 = 0.36 (see table 3), which is 1.8 times larger than that for the rhodopsin genes. Then, the divergence time at point B in figure 1 may be estimated in three ways. The total branch lengths for the branches B-Ru, B-Gu, and B-Bu are 0.25,0.23, and 0.29, respectively (see table 3). The corresponding estimates for point B are 579 Mya [=0.25/(0.24 X 10-g)/l.8], 532 Mya, and 671 Mya, ranging from 530 to 670 Mya. Similarly, branch point C is estimated to be 9.3 Mya if branch C- GH is used, whereas it is 34.7 Mya if branch C-RH is used.

These estimates show that the ancestor of the human color pigment genes diverged from that of the rhodopsin genes -800 Mya. At this point in time, a new kind of photoreceptor cell, i.e., cones, may have been created. This functional adaptation may be reflected in the higher rate of nonsynonymous substitution at branch A-B in figure 1. Then, -200-300 million years later, the color pigment gene again duplicated and the development of short- and long-wavelength visual pigments must have begun. The divergence between the red and green pigment genes is much more recent, i.e., 9-35 Mya. It has been shown (Bowmaker et al. 1983; Jacobs 1983) that old-world monkeys and humans appear to have the same green and red cone pigments, whereas new-world monkeys have only a single long-wavelength visual pigment. The divergence time between new- and old-world monkeys is considered to be -50 Mya, whereas that between old-world monkeys and humans is -30 Mya ( Romeo-Herrera et al. 1973). Thus, it is most likely that the red and green pigment genes diverged just before the separation of old-world monkeys and humans, i.e., -30 Mya.

The human visual pigment genes have been assigned to different chromosomes, i.e., the rhodopsin gene to chromosome 3, the blue pigment gene to chromosome 7, and both red and green pigment genes to the X chromosome (Nathans et al. 1986a). The existence of the visual pigment genes on different chromosomes shows a dynamic aspect of genome rearrangement during evolution. We have shown that the color pigment genes evolved faster than the rhodopsin genes and that they evolved with different rates. One interesting aspect of this variation is that branch B-Gu in figure 1 has the shortest length. As noted before, rhodopsin maximally absorbs the wavelength 495 nm, whereas the blue, green, and red pigments maximally absorb the wavelengths 420 nm, 530 nm, and 560 nm, respectively. Thus, the slowly evolving green pigment absorbs the wavelength closest to that absorbed by the rhodopsin and may be closest to the common ancestor of the three color pigment genes.

Since possible historical events in the evolution of the visual pigment genes have here been considered, it is of interest to locate specific amino acid changes along the different branches leading to the present human color pigment genes. The detection of many such amino acid changes is possible by deducing amino acids from the DNA sequences. For example, the 37th codons of the red and green pigment genes ( AGA) and of the rhodopsin genes (CGC) code for arginine, whereas the corresponding codon (GTG) of the blue pigment gene codes for valine. On the basis of the topology of figure 1, then, the amino acid substitution must have occurred from arginine to valine along branch B-Bn. Similarly, for both the red and green pigment genes, the 44th codon (AAT) codes for asparagine, whereas the others (CAG) code for glutamine, showing that the amino acid substitution occurred from glutamine to asparagine along branch B-C. On the basis of this logic, 8 1 directed amino acid changes can be identified at the specific branches in figure 1. Among these, 36 replacements are located on branch B-Bn ,4 1 on branch B-C, 4 on branch C-Rn, and 0 on branch C-Gu.

Figure 2 is the modified version of the schematic diagram for bovine rhodopsin, a diagram devised by Hargrave et al. ( 1983; see also Nathans et al. 19863), and con-

Evolution of Color Vision 195

taining the cytoplasmic face, seven transmembrane segments, and the luminal face. In figure 2, both the directed amino acid substitutions and the 91 conserved amino acids among the three human color pigment genes and the two mammalian rhodopsin genes are shown. All of the visual pigments have a lysine at the 3 12th residue in the seventh transmembrane segment in figure 2, a residue that is the site of covalent attachment to 114s retinal (Bownds 1967; Wang et al. 1980; Mullen and Akhtar 198 1), showing one of the most important sites. However, that residue is not perfectly conserved at the nucleotide level. Vertebrate rhodopsin is known to interact with several peripheral membrane and soluble proteins, which include G protein (trans- ducin), opsin kinase, phosphoprotein phosphatase, and a 48kilodalton protein that interacts with phosphorylated opsin (Nathans 1987). Binding of some of these proteins and the cytoplasmic loops is suspected, and the strong conservation of the cytoplasmic loops is consistent with the functional constraint. Six consecutive amino acids in the second transmembrane segment are replaced in branch B-Bn and, furthermore, all four amino acid replacements in branch C-Ru are restricted to the sixth and seventh transmembrane segments. These substitutions may have contributed to the develop ment of blue and red wavelength absorptions.

Nathans et al. (1986b) observed that the net intramembrane charge of the blue pigment was + 1, that of rhodopsin was 0, and that of the green and red pigments was - 1. They suggest that the net intramembrane charges of the three color pigments are important, at least in part, for spectral absorption differences. On the basis of the directed amino acid changes detected, the change in the net charge associated with the development of the three color pigments can be determined directly. Generally, lysine and arginine are positively ( + ) charged, asparatic acid and glutamic acid are negatively ( - ) charged, and other amino acids are all neutral (0). Detectable charge changes are also shown in figure 2, where the changes (--0) and (O++) are denoted as + 1, whereas the changes (++O) and (O+-) are denoted as - 1. On the basis of the directed amino acid changes, only one charge change in the intramembrane region can be detected, which shows the change from negative to neutral, i.e., + 1, along branch B-Bn . This agrees with the assertion of Nathans et al. ( 1986b).

When the net charges of the visual pigments are considered, a somewhat different result emerges. The net changes along branches B-Bn , B-C, and C-Ru (and C-Gn ) are -3, +4, and 0, respectively. Thus, the increase in positive net charge is associated with the absorption of the longer wavelength, whereas the increase in negative net charge is associated with the absorption of the shorter wavelength. On the basis of figure 13 in Nathans et al. ( 19863)) the net charges for the blue, rhodopsin, and red (or green) pigments for the entire protein are + 16, + 16, and +22, respectively. This also shows that the increase in the net charge toward the positive direction associates with the absorption of the longer wavelength. The net charge’s exact effect on the spectral absorption differences, however, remains to be characterized. Elucidation of three-dimensional structures of visual pigments will be one of the important steps in understanding functional implications of visual pigment sequences.

Acknowledgments

Comments by Drs. W. M. Fitch, B. G. Hall, J. Nathans, and M. Nei were greatly appreciated. This work was supported by National Institutes of Health grant GM- 28672 and National Science Foundation grants BBS 87-14603 and DMB870004N.

196 Yokoyama and Yokoyama

LITERATURE CITED

BOWMAKER, J. K., J. D. MOLLON, and G. JACOBS. 1983. Microspectrophotometric results for old and new world primates. Pp. 57-68 in J. D. MOLLON and L. T. SHARPE, eds. Colour vision. Academic Press, New York.

BOWNDS, D. 1967. Site of attachment of retinal in rhJaopsin. Nature 216: 1178- 118 1. DIXON, R. A. F., B. K. KOBILKA, D. J. STRADER, J. L. BONOVIC, H. G. D~HLMAN, T. FRIELLE,

M. A. BOLANOWSKI, C. D. BENNETT, E. RANDS, R. E. DIEHL, R. A. MUMFORD, E. E. SLATER, I. S. SIGAL, M. G. CARON, R. J. LEFKOWITZ, and C. STRADER. 1986. Cloning of the gene and cDNA for mammalian /3-adrenergic receptor and homology with rhodopsin. Nature 321:75-79.

FRIELLE, T., S. COLLINS, K. W. DANIEL, M. G. CARON, R. J. LEFKOWITZ, and B. K. KOBILKA. 1987. Cloning of the cDNA for the human pl-adrenergic receptor. Proc. Natl. Acad. Sci. USA 84:7920-7924.

GOJOBORI, T., K. ISHII, and M. NEI. 1982. Estimation of average number of nucleotide sub- stitutions when the rate of substitution varies with nucleotide. J. Mol. Evol. l&414-423.

HARGRAVE, P. A., J. H. MCDOWELL, D. R. CURTIS, J. K. WANG, E. JUSZCZAK, S. L. FONG, J. K. MOHANNA RAO, and P. ARGOS. 1983. The structure of bovine rhodopsin. Biophys. Structural Mech. 9:235-244.

JACOBS, G. H. 1983. Within-species variation in visual capacity among squirrel monkeys (Siumiri sciureus): sensitivity differences. Vision Res. 23:239-248.

JUKES, T. H., and C. R. CANTOR. 1969. Evolution of protein molecules. Pp. 2 l-l 32 in H. N. MUNRO, ed. Mammalian protein metabolism. Vol. 3 Academic Press, New York.

KIMURA, M. 198 1. Estimation of evolutionary distances between homologous nucleotide se- quences. Proc. Natl. Acad. Sci. USA 78:454-458.

- 1983. The neutral theory of molecular evolution. Cambridge University Press, Cam- . bridge.

KOBILKA, B. K., H. MATSUI, T. S. KOBILKA, T. L. YANG-FENG, U. FRANCKE, M. G. CARON, R. J. LEFKOWITZ, and J. W. REGAN. 1987. Cloning, sequencing, and expression of the gene coding for the human platelet uz-adrenergic receptor. Science 238:650-656.

KUBO, T., K. FUKUDA, A. MIKAMI, A. MAEDA, H. TAKAHASHI, M. MISHINA, T. HAGA, K. HAGA, A. ICHIYAMA, K. KANGAWA, M. KOJIMA, H. MATSUO, T. HIROSE, and S. NUMA. 1986. Cloning, sequencing and expression of complementary DNA encoding the muscarinic acetylcholine receptor. Nature 323:4 1 l-4 16.

MANIATIS, T., R. C. HARDISON, E. LACY, J. LAUER, C. O’CONNELL, 0. QUON, G. K. SIM, and A. EFSTRATIADIS . 1978. The isolation of structural genes from libraries of eukaryotic DNA. Cell 15:687-70 1.

MARTIN, R. L., C. WOOD, W. BAEHR, and M. L. APPLEBURY. 1986. Visual pigment homologies revealed by DNA hybridization. Science 232: 1266- 1269.

MIYATA, T., and T. YASUNAGA . 1980. Molecular evolution of mRNA: a method for estimating evolutionary rates of synonymous and amino acid substitutions from homologous nucleotide sequences and its application. J. Mol. Evol. 16:23-26.

MULLEN, E., and M. AKHTAR. 198 1. Topological and active site studies on bovine rhodopsin. FEBS Lett. 132:26 l-264.

NATHANS, J. 1987. Molecular biology of visual pigments. Annu. Rev. Neurosci. 10:163-194. NATHANS, J., and D. S. HOGNESS. 1983. Isolation, sequence analysis, and intron-exon arrange-

ment of the gene encoding bovine rhodopsin. Cell 34:807-8 14. - 1984. Isolation and nucleotide sequence of the gene encoding human rhodopsin. Proc.

Nat11 Acad. Sci. USA 81:485 l-4855. NATHANS, J., T. P. PIANTANIDA, R. EDDY, T. B. SHOWS, and D. S. HOGNESS. 1986~. Molecular

genetics of inherited variation in human color vision. Science 232:203-210. NATHANS, J., D. THOMAS, and D. S. HOGNESS. 19868. Molecular genetics of human color

vision: the genes encoding blue, green, and red pigments. Science 232: 193-202.

Evolution of Color Vision 197

NEI, M. 1987. Molecular evolutionary genetics. Columbia University Press, New York. NEI, M., and T. GOJOBORI . 1986. Simple methods for estimating the numbers of synonymous

and nonsynonymous nucleotide substitutions. Mol. Biol. Evol. 3:4 18-426. O’TOUSA, J., W. BAEHR, R. L. MARTIN, J. HIRSCH, W. L. PAK, and M. L. APPLEBURY. 1985.

The Drosophila melanogaster nina E gene encodes an opsin. Cell 40:839-850. ROMEO-HERRERA, A. E., H. LEHMANN, K. A. JOYSEY, and A. E. FRIDAY. 1973. Molecular

evolution of myoglobin and the fossil record: a phylogenetic synthesis. Nature 246:389-395. SAITOU, N., and M. NEI . 1987. The neighbor-joining method: a new method for reconstructing

phylogenetic trees. Mol. Biol. Evol. 4:406-425. SCHOPF, J. W., J. M. HAYES, and M. R. WALTER. 1983. Archean microfossils: new evidence

of ancient microbes. Pp. 2 14-239 in J. W. SCHOPF, ed. Earth’s earliest biosphere: its origin and evolution. Princeton University Press, Princeton, N.J.

TAKAHATA, N., and M. KIMURA. 198 1. A model of evolutionary base substitutions and its application with special reference to rapid change of pseudogenes. Genetics 98:641-657.

WALD, G. 1968. The molecular basis of visual excitation. Nature 219:800-807. WANG, J. K., J. H. MCDOWELL, and P. HARGRAVE. 1980. Site of attachment of 1 1-cis retinal

in bovine rhodopsin. Biochemistry 19:5 11 l-5 117. YARDEN, Y., H. RODRIGUEZ, S. K.-F. WONG, D. R. BRANDT, D. C. MAY, J. BURNIER, R. N.

HARKINS, E. Y. CHEN, N. RAMACHANDRAN, A. ULLRICH, and E. M. Ross. 1986. The avian B-adrenergic receptor: primary structure and membrane topology. Proc. Natl. Acad. Sci. USA 83:6795-6799.

YOKOYAMA, S., K. E. ISENBERG, and A. E. WRIGHT. Adaptive evolution of G protein coupled receptor genes. J. Mol. Evol. (submitted).

ZUKER, C., A. F. COWMAN, and G. M. RUBIN. 1985. Isolation and structure of a rhodopsin gene from D. melanogaster. Cell 40:85 l-858.

BARRY G. HALL, reviewing editor

Received June 16, 1988; revision received October 24, 1988