Ultraviolet (UV) and violet vision in vertebrates is mediated by UV and violet visual pigments that absorb light maximally (λmax) at ∼360 and 390–440 nm, respectively. So far, a total of 11 amino acid sites only in transmembrane (TM) helices I–III are known to be involved in the functional differentiation of these short wavelength-sensitive type 1 (SWS1) pigments. Here, we have constructed chimeric pigments between the violet pigment of African clawed frog (Xenopus laevis) and its ancestral UV pigment. The results show that not only are the absorption spectra of these pigments modulated strongly by amino acids in TM I–VII, but also, for unknown reasons, the overall effect of amino acid changes in TM IV–VII on the λmax-shift is abolished. The spectral tuning of the contemporary frog pigment is explained by amino acid replacements F86M, V91I, T93P, V109A, E113D, L116V, and S118T, in which V91I and V109A are previously unknown, increasing the total number of critical amino acid sites that are involved in the spectral tuning of SWS1 pigments in vertebrates to 13.
BY the early 1930s, it was already known that the behavior of many insects was affected strongly by their ultraviolet (UV) vision. It took an additional 60 years to fully appreciate UV vision in vertebrate species (Jacobs 1992), but we now know that many fish, amphibian, reptilian, avian, and mammalian species use UV vision for foraging, mate choice, and communication (Burkhardt 1982, 1989; Harosi 1985; Fleishman et al. 1993; Viitala et al. 1995; Bennett et al. 1996; Hunt et al. 1997; Sheldon et al. 1999). In vertebrates, UV vision is mediated by visual pigments that absorb light maximally (λmax) at ∼360 nm and violet (or blue) vision is mediated by violet (or blue) pigments with λmax-values of 390–440 nm. These pigments belong to an evolutionarily distinct short wavelength-sensitive type 1 (SWS1) pigment group (Yokoyama and Yokoyama 1996; Yokoyama 2000a; Ebrey and Koutalos 2001). Compared with those with violet vision, organisms with UV vision have an advantage of recognizing certain UV-reflecting objects much more quickly, but they lack precision in viewing their surroundings and are also subjected to a higher chance of developing retinal damages caused by UV light. Thus, whether or not organisms have UV vision must depend strongly on species-specific behavioral and physiological requirements (Shi and Yokoyama 2003).
Visual pigments consist of a transmembrane (TM) protein, opsin, and the chromophore, 11-cis-retinal. The protonated Schiff-based chromophore in solution absorbs light at 440 nm (Kito et al. 1968). By interacting with different types of opsins directly or indirectly, however, the identical chromophore in various pigments can have a wide range of λmax-values between 360 and 560 nm, which is referred to as the spectral tuning of visual pigments (Kochendoerfer et al. 1999).
Over the past several years, significant progress has been made in clarifying the molecular basis of spectral tuning in the SWS1 pigments. The molecular analyses of avian and mammalian pigments show that the differentiations of UV and violet pigments have occurred by the accumulations of various amino acid replacements at a total of 11 sites, 46, 49, 52, 86, 90, 93, 97, 113, 114, 116, and 118 in TM helices I–III, where amino acid site numbers are standardized by those of the bovine rhodopsin (Wilkie et al. 2000; Yokoyama and Shi 2000; Yokoyama et al. 2000; Babu et al. 2001; Shi et al. 2001; Cowing et al. 2002; Fasick et al. 2002; Shi and Yokoyama 2003; Parry et al. 2004; Yokoyama and Tada 2005; Yokoyama et al. 2005b). Furthermore, engineered ancestral pigments of a wide range of vertebrates show that most vertebrate ancestors, including the common ancestor, had UV pigments. Most contemporary UV pigments have inherited their UV sensitivities from the vertebrate ancestor by retaining most of these 11 critical amino acids, whereas violet pigments acquired their new functions by modifying several critical amino acids. The avian SWS1 pigments are an exception, where the ancestral pigment abandoned UV sensitivity by replacing four critical amino acid changes, but some descendants regained it by a single amino acid change (Shi and Yokoyama 2003).
The spectral tuning of SWS1 pigments is characterized by strong synergistic interactions among the critical amino acids. The interactions are often so strong that λmax-shifts can be detected only when at least two critical amino acids are mutated (Shi et al. 2001). Some single amino acid changes at sites 86 and 90 can shift the λmax-value of SWS1 pigments (Wilkie et al. 2000; Yokoyama et al. 2000; Cowing et al. 2002; Fasick et al. 2002; Shi and Yokoyama 2003; Parry et al. 2004; Yokoyama and Tada 2005; Yokoyama et al. 2005b), but they also interact with other amino acids (see discussion). From these analyses, it is now widely accepted that the absorption spectra of SWS1 pigments are modulated by amino acids in TM helices I–III. This “assumption” is based on a rather limited number of species and its validity has not been examined.
To test the generality of the assumption, it is necessary to extend the analyses to species that are outside of the avian and mammalian lineages. At present, African clawed frog (Xenopus laevis) is the only species that has been shown to have violet pigments, using an in vitro assay (Starace and Knox 1998). Furthermore, the ancestral UV pigment between the frog and salamander pigments has been engineered, making evolutionary analyses of SWS1 pigment in the amphibian lineage feasible (Shi and Yokoyama 2003). To better understand the molecular basis of spectral tuning of SWS1 pigments, therefore, we have constructed chimeric pigments consisting of all combinations of TM helices I, II, III, and IV–VII between the frog and its ancestral pigments. The analyses reveal totally unexpected results; that is, not only has the frog pigment acquired the contemporary absorption spectrum by accumulating amino acid replacements in TM helices I–IV but also the overall effect of individual and all interaction terms that include TM helices IV–VII on the λmax-shift is virtually zero.
MATERIALS AND METHODS
Construction of chimeric SWS1 pigments and site-directed mutagenesis:
Shi and Yokoyama (2003) engineered SWS1 pigments of major vertebrate ancestors by introducing a total of ∼70 amino acid changes into several extant pigments, one of which is the ancestral amphibian pigment between African clawed frog (X. laevis) and salamander (Ambystoma trigrinum) (ancestral pigment c in Shi and Yokoyama 2003). The opsin cDNAs of the frog and ancestral amphibian pigment have been subcloned into pBluescript vector KS(+). The N and C termini, whose boundaries are identified by NdeI and MfeI sites, respectively (Figure 1), have been replaced by the corresponding segments of the SWS1 pigment of American chameleon (Anolis carolinensis). These modifications do not change the λmax-values of visual pigments and keep the backgrounds of all wild-type and chimeric pigments identical (Shi and Yokoyama 2003). All mutant pigments with single and multiple amino acid changes were generated by using the QuickChange site-directed mutagenesis kit (Stratagene, La Jolla, CA), and the primers were designed using the protocol recommended by the manufacturer. Various chimeric pigments were constructed by recombining their cDNA fragments of restriction enzyme recognition sites, NdeI, HindIII, BsmBI, SphI, and MfeI (Figure 1). NdeI and MfeI sites, which had been introduced to replace their N and C termini by the corresponding segments of the chameleon pigment, can introduce different amino acids from those of the ancestral and frog pigments and the “correct” amino acids have been reintroduced by the mutagenesis method. The amino acid sequences of all mutant pigments were confirmed by a Sequitherm Excel II long-read kit (Epicentre Technologies, Madison, WI) with dye-labeled M13 forward and reverse primers. Reactions were run on a LI-COR (Lincoln, NE) 4200LD automated DNA sequencer. The confirmed mutants were subcloned into expression vector pMT5.
Expression and spectral analyses of pigments (in vitro assay):
The opsin cDNAs of full length were subcloned into the EcoRI and SalI restriction sites of the expression vector pMT5 and these plasmids were expressed in COS1 cells by transient transfection (e.g., Yokoyama 2000b). In short, the pigments were regenerated by incubating the opsins with 11-cis-retinal (Storm Eye Institute, Medical University of South Carolina) and purified using immobilized 1D4 (Industry Liaison Office, University of British Columbia) in buffer W1 [50 mm N-(2-hydroxyethyl) piperazine-N′-2-ethanesulfonic acid (HEPES) (pH 6.6), 140 mm NaCl, 3mm MgCl2, 20% (w/v) glycerol, and 0.1% dodecyl maltoside]. UV visible spectra were recorded at 20° using a Hitachi U-3000 dual beam spectrophotometer. Visual pigments were bleached for 3 min using both a 366-nm UV light illuminator and a 60-W standard light bulb equipped with a Kodak Wratten no. 3 filter at a distance of 20 cm. Data were analyzed using Sigmaplot software (Jandel Scientific, San Rafael, CA).
Functional differentiation of the frog pigment from the ancestral pigment:
The ancestral and frog pigments have λmax-values of 359 nm (Figure 2A) and 423 nm (Figure 2L), respectively. To identify TM helices that are involved in the differentiation of the frog pigment, we have constructed 10 chimeric pigments, f(67)a, a(67)f(99)a, a(99)f(152)a, f(99)a, f(67)a(99)f(152)a, a(67)f(152)a, f(152)a, a(67)f(99)a(152)f, f(67)a(152)f, and a(152)f, where amino acids 31–66, 67–98, 99–151, 31–98, 31–66, and 99–151; 67–151, 31–151, 67–98, and 152–311; 31–66 and 152–311; and 152–311 of the ancestral pigment were replaced by the corresponding segments of the frog pigment, respectively, where the amino acid sites are standardized by those of the bovine rhodopsin (Figure 2, B–K, and Table 1). The chimeric pigments show that if we replace the TM helices I–III of the ancestral pigment by the corresponding segments of the frog pigment, then its λmax-value becomes 422 nm (Figure 2H and Table 1). On the other hand, if we replace the TM helices I–III of the frog pigment by the corresponding segments of the ancestral pigment, then its λmax-value becomes 360 nm (Figure 2K and Table 1). Thus, it seems that the frog pigment has acquired its absorption spectrum by accumulating amino acid replacements only in TM helices I–III, which is consistent with the observation that amino acid replacements in that region are responsible for the spectral sensitivity of mammalian and avian SWS1 pigments (Shi et al. 2001).
To evaluate the effects of TM helices I, II, and III on the λmax-shift, let us first consider the evolution of the frog pigment from the ancestral UV pigment. Thus, let θI, θII, and θIII be the magnitudes of the λmax-shift caused by the replacements of TM helices I, II, and III of the ancestral pigment by the corresponding segments of the frog pigment, respectively, and Z be the absorption spectrum of the ancestral pigment. Furthermore, let θI×II, θI×III, θII×III, and θI×II×III be the synergistic effects between TM helices I and II, between TM helices I and III, between TM helices II and III, and among TM helices I, II, and III on the λmax-shift, respectively. Then, using the λmax-values in Figure 2, A–H (see also Table 1),
Solving these equations, Z = 359 nm and θI, θII, θIII, θI×II, θI×III, θII×III, and θI×II×III are 0, 24, 51, 6, 7, −13, and −12 nm, respectively (Table 2). Thus, TM helices II and III have major individual effects in spectral tuning of the frog pigment. Although it does not cause any λmax-shift individually, TM helix I not only increases the λmax-value by 6 and 7 nm by interacting with TM helices II and III, respectively, but also decreases the λmax-value by 12 nm with TM helices II and III. It should be noted, however, that the overall effect of TM helix I on the λmax-shift (θI + θI×II + θI×III + θI×II×III) is only 1 nm and negligible. This overall effect gives an impression that the spectral tuning of the frog pigment is determined exclusively by amino acid changes in TM helices II and III.
Reverse changes from the frog pigment to the ancestral pigment:
We have seen that chimeric pigments a(67)f(99)a(152)f, f(67)a(152)f, and a(152)f have λmax-values of 404, 361, and 360 nm, respectively (Figure 2, I–K). We have constructed four additional chimeric pigments, a(99)f, f(99)a(152)f, f(67)a(99)f, and a(67)f, where amino acids at sites 31–98, 99–151, 67–98, and 31–66 of the frog pigment were replaced by the corresponding segments of ancestral pigment, respectively. Their λmax-values range from 404 to 418 nm (Table 1). The comparison of the λmax-values of f(67)a(152)f and a(67)f pigments again suggests that the functional differentiation of the frog pigments has been achieved by amino acid replacements mainly in TM helices II and III (Table 1). When we consider all eight pigments with TM helices IV–VII of the frog pigment, Z = 360 nm and θI, θII, θIII, θI×II, θI×III, θII×III, and θI×II×III are 1, 44, 44, 3, −1, −30, and 2 nm, respectively (Table 2). Thus, with the exceptions of Z and θI, the two sets of θ-values estimated by considering the TM helices IV–VII of ancestral and frog pigments differ significantly. These differences come from interactions between TM helices I–III and IV–VII. We can see this by simply comparing the λmax-values of a(67)f(99)a and a(67)f(99)a(152)f in Table 1, where the 21-nm difference occurs due to different types of TM helices IV–VII.
The role of TM helices IV–VII in the spectral tuning of SWS1 pigments:
To evaluate the individual and joint effects of TM helices I, II, III, and IV–VII on the λmax-shift more generally, let us define additional parameters θIV–VII, θI×IV–VII, θII×IV–VII, θIII×IV–VII, θI×II×IV–VII, θI×III×IV–VII, θII×III×IV–VII, and θI×II×III×IV–VII as the individual and synergistic effects of TM helices IV–VII and the other TM helices on the λmax-shift. Then, using all 16 λmax-values in Table 1, Z and all 15 θ-values can be evaluated (Table 2). As we can see in Table 2, TM helices IV–VII had significant roles during the functional differentiation of the frog pigment.
As noted earlier, when the ancestral and frog backgrounds are considered separately, θII-values differ by 20 nm, which can now be explained formally by the simultaneous replacements of TM helices II and IV–VII, θII×IV–VII. Similarly, the difference between the θIII-values with the two different backgrounds can be explained by the interaction between TM helices III and IV–VII (Table 2). The amazing thing about these interactions is the fact that despite their large individual synergistic effects, the sum of all θ-values that involve TM helices IV–VII, i.e., θIV–VII + θI×IV–VII + θII×IV–VII + θIII×IV–VII + θI×II×IV–VII + θI×III×IV–VII + θII×III×IV–VII + θI×II×III×IV–VII, is virtually zero. The exact reason for this fine spectral tuning is not known, but it appears that the frog pigment has achieved its λmax-value by going through some rigid “structural constraints” (see discussion).
Critical amino acid replacements in TM helices I–III of the contemporary frog violet pigment:
When the amino acids in TM helices I, II, and III of the ancestral and frog pigments are compared, they differ at eight, five, and nine sites, respectively (Figure 1). Among these, F49L in TM helix I; T93P in TM helix II; and E113D, L116V, and S118T in TM helix III of UV pigments are known to be involved in the spectral tuning of SWS1 pigments either individually or jointly (Babu et al. 2001; Shi et al. 2001; Shi and Yokoyama 2003). When T93P, L116V, and S118T are introduced into the ancestral pigment, they do not cause any significant λmax-shift individually (Figure 3). Babu et al. (2001) have found that the reverse mutation, D113E, in the frog pigment decreases the λmax-value by 12 nm. Much to our surprise, however, when E113D is introduced into the ancestral pigment, the mutant pigment also decreases the λmax by 4 nm (Figure 3), a direction that is the opposite of that predicted from the previous observation.
To identify all amino acid changes in TM helices I–III that are involved in the spectral tuning of the contemporary frog pigment, we have introduced a series of amino acid changes into the TM helices I–III of the ancestral pigment. The mutagenesis results show that the λmax-shifts caused by F86M, V91I, and T93P in TM helix II and V109A, E113D, L116V, and S118T in TM helix III agree well with the corresponding values caused by a(67)f(99)a and a(99)f(152)a pigments (Figure 3 and Table 3). Amino acid changes F49L/F86M/V91I/T93P in TM helices I–II and F49L/V109A/E113D/L116V/S118T in TM helices I and III also generate λmax-values of 385 and 411 nm, respectively, which are similar to those of f(99)a and f(67)a(99)f(152)a pigments (Figure 3 and Table 3); however, the former values are 4–6 nm lower than the latter values (Table 3). Without F49L, these mutant pigments decrease λmax-values by 3–5 nm (Figure 3), showing that F49L had some minor roles during the evolution of the frog pigment.
When seven amino acid changes F86M/V91I/T93P/V109A/E113D/L116V/S118T are introduced into the ancestral pigment, the mutant has a λmax-value of 421 nm (Figure 4A and Table 3). When we add F49L into this mutant pigment, it increases the λmax-value only by 1 nm (Table 3). On the other hand, when the reverse changes M86F/I91V/P93T/A109V/D113E/V116L/T118S are introduced into the frog pigment, a λmax-value of 355 nm is achieved (Figure 4B). Thus, the violet sensitivity of the contemporary frog pigment can be explained fully by the amino acid replacements at the seven sites.
Previously, molecular analyses of avian and mammalian SWS1 pigments showed that the functional differentiation of the violet pigments had occurred by amino acid replacements at 11 sites in TM helices I–III, particularly by those in TM helices I–II. Our molecular analyses of the frog violet pigment add four new pieces of information: (1) sites 91 and 109 are also involved in the spectral tuning; (2) amino acid replacements in TM helix III can also contribute significantly to the functional differentiation of SWS1 pigments; (3) interactions of critical amino acids within and between TM I–III and TM IV–VII had a significant role in the functional differentiation of the frog pigment; and (4) the overall effect of amino acid changes in TM IV–VII on the λmax-shift is essentially zero.
The involvement of TM helices IV–VII in the spectral tuning of SWS1 pigments is not unique to the frog pigment. As noted earlier, the mouse UV pigment and human blue pigment have λmax-values of 359 and 414 nm, respectively. When TM helices I–III of the mouse pigment are replaced by those of the human pigment, the chimeric pigment achieves a λmax of 414 nm, whereas the human pigment with the reverse changes has a λmax of 360 nm. Again, the functional differentiation of the two pigments seems to have occurred by amino acid replacements only in TM helices I–III (Shi et al. 2001). Unfortunately, when TM helix I of the mouse pigment is replaced by the corresponding segment of the human pigment, 11-cis-retinal does not bind to the opsin and the λmax-value of the chimeric pigment cannot be measured. For the human and mouse pigments, therefore, we cannot evaluate all individual and synergistic effects of different TM helices on the λmax-shift. However, when TM helices II–III of the mouse pigment are replaced by the corresponding segments of the human pigment, the chimeric pigment has a λmax-value of 405 nm (Shi et al. 2001), whereas when TM helices II–VII of the mouse pigment are replaced by those of the human pigment, the mutant pigment has a λmax-value of 413 nm (Y. Shi and S. Yokoyama, unpublished data). Since the two pigments differ only in their TM helices IV–VII, the different λmax-values are generated by the interaction between TM helices II–III and TM helices IV–VII. This comparison clearly shows that the synergistic interactions between amino acids in the two regions have occurred during the functional differentiation of the human and mouse pigments as well. Since the λmax-values of the mouse and human pigments can be exchanged by interchanging their TM helices I–III, the overall effect of amino acid changes in TM helices IV–VII of the mouse pigment on the λmax-shift is also zero.
As indicated earlier, some amino acid changes at sites 86 and 90 in TM helix II can cause significant λmax-shifts in SWS1 pigments. They are Y86F in the bovine pigment (λmax-shift = −71 nm) (Fasick et al. 2002) and wallaby (λmax-shift = −59 nm) (Yokoyama and Tada 2005), Y86F in the goldfish pigment (λmax-shift = −60 nm) (Cowing et al. 2002), V86F in guinea pig (λmax-shift = −53 nm) (Parry et al. 2004), F86S in the common ancestral pigment of birds and reptiles (λmax-shift = 17 nm) (pigment e in Shi and Yokoyama 2003), S86F in the elephant pigment (λmax-shift = −52 nm) (Yokoyama et al. 2005b), S90C in the pigeon and chicken pigments (λmax-shift = −34 and −46 nm) (Yokoyama et al. 2000), S90C in the mouse pigment (λmax-shift = −2 nm) (Shi et al. 2001), and C90S in the budgerigar and zebra finch pigments (λmax-shift = 35–38 nm) (Wilkie et al. 2000; Yokoyama et al. 2000). Hence, the same amino acid changes in different SWS1 pigments generate different λmax-values. Although the cause of such variation is not known, it is likely that the variation is caused not only by interactions with other critical amino acids in TM I–III but also by those between TM helices I–III and IV–VII. To test such possibilities, construction of various chimeric pigments between two comparable pigments will be necessary, as demonstrated by our analyses of the frog pigment.
Comparisons of the mammalian and amphibian pigments demonstrate that the functional differentiation of violet pigments from the ancestral UV pigment can be understood fully only by studying both individual and synergistic effects of critical amino acids in all seven TM helices. Critical amino acids in TM helices IV–VII are unknown and remain to be identified. Once such sites are identified, it is possible to explore the molecular mechanisms that underlie the evolution of SWS1 pigments. At the chemical level, it is highly likely that UV pigments have unprotonated Schiff base-linked chromophores and violet pigments have protonated Schiff base-linked chromophores (Babu et al. 2001; Kusnetzow et al. 2001; Shi et al. 2001; Dukkipati et al. 2002; Fasick et al. 2002). The identification of all critical amino acid sites will also be an important step in clarifying their roles in the protonation of UV pigments or deprotonation of violet pigments.
As represented by Doi et al. (1990), followed by more than 50 articles entitled “Structure and Function of Rhodopsins” by Khorana and his colleagues, the properties of key amino acids common to various visual pigments have been characterized. These analyses have improved our understanding of how visual pigments work. Compared with these biochemical analyses, our analyses intend to uncover how visual pigments have managed to change their structure-function relationships during evolution. Certainly, every amino acid change modifies the tertiary structure of proteins, but not all structural changes lead to a change of protein function (e.g., Perutz 1983). We have seen that the overall effect of all critical amino acid changes in TM helices IV–VII on the λmax-shift is negligible. This observation strongly suggests that critical amino acid changes have not been incorporated into the frog or human SWS1 pigments at random; rather, the observed amino acid replacements might have occurred because certain structural requirements are met. This also may mean that the molecular changes can take some limited number of possible evolutionary paths. To explore the mechanisms of such evolutionary processes, it will be useful to perform tertiary structural analyses and quantum chemical computations on the basis of various chimeric and other mutant pigments with known λmax-values (Yokoyama et al. 2005a). The concept of structural constraints that emerges from such analyses of protein evolution will offer a new feature of structure and function of visual pigments.
Comments by Ruth Yokoyama, Naoyuki Takahata, and two anonymous reviewers were greatly appreciated. This work was supported by a grant from the National Institutes of Health and by a setup fund from Emory University.
↵1 Present address: Department of Medicine Endocrinology, University of Oklahoma Health Science Center, 941 Stanton L. Young Blvd., Oklahoma City, OK 73104.
Communicating editor: N. Takahata
- Received May 18, 2005.
- Accepted July 21, 2005.
- Copyright © 2005 by the Genetics Society of America