4. Discussion

↓91

4.1. Nocturnal vs. diurnal 

4.1.1. Characterisation of the echidna rhodopsin

The rhodopsin of the short-beaked echidna was successfully expressed in vitro and was found to be functional, as indicated by the dark and light absorption spectra (Fig. 17B, chapter 3.1.4.). With a λmax at 496.5 nm, it absorbs light in a more blue-shifted range than that of bovine. The rhodopsin of its sister taxon, the platypus, has its absorption peak at 498 nm (Davies et al. 2007).

Though the bleaching with HCl acid did not take place as quickly in the echidna as in the bovine, the formation of the protonated Schiff base was complete within the first 5 minutes, which also indicates that this pigment is functional (Fig. 18B, chapter 3.1.5.). The molar extinction coefficient was determined to be 34 921 M-1 cm-1, which is much lower than that predicted for bovine (Tab. 11, chapter 3.1.5.) (Wald and Brown 1953, Shichi et al. 1969, Daemen et al. 1970, Hong and Hubbell 1972, Oprian et al. 1987). The molar extinction coefficient is a measure of how strongly a protein absorbs light at a given wavelength. Since one photoexcited rhodopsin molecule activates hundreds of copies of transducin (Sagoo and Lagnado 1997, Menon et al. 2001), one would assume that for vision at low light levels, the rhodopsin would be adapted to absorb a single photon very strongly and trigger the activation of as many transducin molecules as possible. Thus, one would expect the rhodopsin of a nocturnal animal to be better adapted to scotopic vision, displaying a high molar extinction coefficient.

↓92

In the hydroxylamine assay, the echidna rhodopsin, as well as its two mutants, reacted to hydroxylamine more than bovine and the ancestral rhodopsins (Fig. 19B, chapter 3.1.6.). Since this assay has commonly been used to characterise rod and cone opsins, this result suggests that the expressed echidna rhodopsin is cone-like. However, cone opsins react to hydroxylamine much stronger (Imai et al. 1995, Das et al. 2004).

The determination of the meta II decay rate, which is the active state of rhodopsin in which the GDP- for GTP-exchange on the G-protein transducin is catalyzed, thereby activating it and eventually generating an electrical response in the photoreceptor cell, provides an interesting result (Tab. 13, chapter 3.1.7.). With a mean value of 7.92 min-1, the echidna rhodopsin has a much lower t1/2 than bovine (13.38 min-1).

It has been suggested that having a longer signaling state increases the sensitivity of the photoreceptor cells (Imai et al. 1997, Kuwayama et al. 2002, Shichida and Matsuyama 2009). Thus, a higher meta II time constant would be advantageous for scotopic vision (Sugawara et al. 2010). Cones, which are less photosensitive than rods, show considerably faster meta II decay rates than rods and less activation of the visual transduction cascade (Wald et al. 1955, Wilden et al. 1986, Langlois et al. 1996). Hence, with a t1/2 half that of bovine, the echidna rhodopsin displays another cone-like characteristic.

↓93

It has also been shown that rhodopsins sometimes display cone-like characteristics and cones sometimes behave rod-like (Crescitelli 1980, Crescitelli 1988, Kawamura and Yokoyama 1998, Yokoyama and Blow 2001). In the gecko, which has a pure rod retina, the green Rh2 cone pigment shows rod-like biochemical characteristics (Crescitelli 1980, Crescitelli 1988). In the anole, the SWS2 cone pigment displays a rod-like insensitivity to hydroxylamine; a result which is still open to interpretation (Kawamura and Yokoyama 1998). On the other hand, the anole rhodopsin is sensitive to hydroxylamine, and thus cone-like, probably as an adaptation to a pure cone retina (Kawamura and Yokoyama 1998). Yokoyama and Blow (2001) suggested that substituting a Glycine (G) for a Methionine (M) at site 89 is likely to serve as determinant of rod and cone properties. However, the echidna has a Glycine (G) at this site, like all other taxa included in this study (Tab. 4, chapter 2.2.2.). Imai et al. (1997) suggested that substitutions at site 122 are associated with rod and cone pigments, as E122Q and E122I bovine rhodopsin mutants showed sensitivity to hydroxylamine. However, both the gecko and the echidna rhodopsin have a Glutamate (E) at this site (Tab. 4, chapter 2.2.2.).

Furthermore, rhodopsins can be expressed in cones, and cone opsins in rods (Kawamura and Yokoyama 1998). In the tiger salamander, the Rh2 rods and SWS2 cones both contain the same SWS2 opsin, but use different transducin types (Ma et al. 2001).

The echidna has long been thought to possess a pure rod retina (Walls 1942, O’Day 1952); a finding which was refuted by the identification of twin cones present in the retina (Young and Pettigrew 1991). Thus, the cone-like meta II decay rate and the hydroxylamine sensitivity of the echidna rhodopsin might reflect an adaptation to rhodopsin being expressed in twin cones as well and thus, show cone-like characteristics as an adapatation to expression in cones, as is the case in the anole (Kawamura and Yokoyama 1998). Investigating the biochemical properties of the cone pigments of the echidna would be an interesting study possibly providing more clearity.

↓94

However, the results derived from this study as well as others (Crescitelli 1980, Crescitelli 1988, Kawamura and Yokoyama 1998, Yokoyama and Blow 2001) point out how variable visual pigments, even from the same opsin class, are in their biochemical and functional properties and that changes are not necessarily a result of ecological constraint, as previously assumed.

The echidna rhodopsin displays many reptilian characteristics in its eye, such as morphologically similar bipolar cells, a cartilaginous sclera, and a flattened lense (Bolk et al. 1934, Young and Pettigrew 1991). Interestingy, in the rhodopsin amino acid sequence, there is an observable trend that monotremes more frequently share the same residue with reptiles and other non-mammalian vertebrates than with Theria (see chapter 3.3.2. and 3.3.3.). At rather conservative residues, only nine amino acids are shared with other mammalian taxa, whereas twelve amino acids are shared with non-mammals (Tab. 4, chapter 2.2.2.). Most interesting is an insertion of five amino acids at the end of the amino acid sequence in monotremes and all non-mammalian taxa, which is known to interact with rhodopsin kinase, which is a downstream effector of rhodopsin and, thus, a crucial component in the visual signaling cascade (Nathans and Hogness 1983).

A mosaic of derived and plesiomorphic characters in monotremes, as present in the rhodopsin amino acid sequence, has also been reported from anatomic, genomic, physiological, and developmental studies (Bolk et al. 1934, Gresser and Noback 1935, Griffiths 1989, Young and Pettigrew 1991, Warren et al. 2008, Werneburg and Sánchez-Villagra 2010). On the one hand, these findings strengthen the yet controversial Theria hypothesis that monotremes are the most basal mammals (Janke et al. 2002, Rowe et al. 2008). On the other hand, the odd mosaic pattern in the echidna amino acid sequence might be responsible for the cone-like and yet contradictory results derived from the functional and biochemical assays.

4.1.2. Characterisation of the two echidna mutants

↓95

As seen in Figure 17C-D (chapter 3.1.4.), the expression of the two echidna mutants T158A and F169A was also successful and both pigments are functional. With 494.5 nm and 495.5 nm for T158A and F169A, respectively, the determined λmax are close to the one determined for the echidna rhodopsin, which is a bit blue-shifted from where bovine has its absorption peak.

For the acid bleach, the protonated Schiff base had formed in mutant F169A as fast as in bovine, whereas in mutant T158A it took a bit longer. This result nevertheless indicates that both expressed pigments are functional (Fig. 18C-D, chapter 3.1.5.).

The molar extinction coefficients vary: with a value of 31 411 M-1 cm-1, mutant T158A has a ε similar to the one determined for echidna, whereas F169A (40 254 M-1 cm-1) has a ε similar to bovine rhodopsin (Tab. 11, chapter 3.1.5). The molar extinction coefficient is a measure of how strongly the rhodopsin absorbs light at λmax. Thus, this result suggests that site 169 affects the strength of photon absorption in the echidna. Borhan et al. (2000) figured that site 169, which is not a conserved residue in the GPCR family of proteins, is cross-linked to the all-trans chromophore in intermediates lumirhodopsin, meta I, and meta II (Fig. 13, chapter 2.1.8.). Furthermore, this site is likely to be involved in transducin activation (Borhan et al. 2000). Only two of seven rhodopsins expressed in this study, i.e. echidna and T158A mutant, display a low ε, and, interestingly, these two have a Phenylalanine (F) instead of an Alanine (A) at site 169, suggesting that a F, as opposed to an A, decreases the strength of photon absorption as measured by the molar extinction coefficient. However, the benefit of decreasing the strength of photon absorption in the nocturnal echidna remains to be elucidated. For future research, it would be interesting to determine the ε of the platypus rhodopsin, as it has a unique Leucine (L) at this site (Tab. 4, chapter 2.2.2.).

↓96

Together with the echidna rhodopsin, the two mutants show sensitivity to hydroxylamine (Fig. 19C-D, chapter 3.1.6.). The strong increase in relative difference absorbance at 363 nm, which is where the retinal oxime absorbs, indicates that the hydroxylamine entered the chromophore binding pocket as it does in cones (Kawamura and Yokoyama 1998). However, the assay was only performed twice for the two mutants, due to technical reasons, and should be reproduced for reliability. Still, the present finding suggests that the echidna is sensitive to hydroxylamine and that substitutions at site 158 and 169 are not involved in regulating this.

4.1.3. Inferring life habits from absorption maxima of living taxa

It has long been hypothesised that the range of absorption maxima in rhodopsin corresponds with life habits in vertebrates (Chang et al. 2002a, Chang 2003, Yokoyama et al. 2008, Zhao et al. 2009b). In particular, a red-shifted absorption range (> 500 nm) is said to be advantageous for vision at low-light levels, whereas a blue-shifted absorption range (< 500 nm) is said to be an adaptation to a deep-water habitat (Muntz 1976, Yokoyama et al. 2008). Yokoyama et al. (2008) classified rhodopsins into four classes based on their absorption maxima and light environments: deep-sea (≈ 480-485 nm), intermediate (≈ 490-495 nm), surface (≈ 500-507 nm), and terrestrial red-shifted (≈ 525 nm). Chang et al. (2003) pointed out that birds tend to have longer wavelength-absorbing rhodopsins. In addition, a number of studies focus on spectral tuning sites, accepting the assumption that the absorption range allows for inferences of life habits (Kochendörfer et al. 1999, Altun et al. 2008, Zhao et al. 2009b). Sugawara et al. (2010) hypothesized that substitutions at sites 83 and 292 are responsible for a blue-shift in λmax values indicating adaptation to a deep-water habitat.

However, this assumption has never been verified nor statistically tested. Thus, the aim was to statistically test if there is a correlation between wavelength absorption and lifestyle, and if a potential correlation is linked to phylogeny. Therefore, absorption maxima of 42 tetrapod taxa were collected from the literature and the life habits of the taxa were classified into three groups, i.e. 1 for diurnal, 2 for nocturnal, and 3 for aquatic or semi-aquatic (Tab. 29).

↓97

Table 29. 42 tetrapod taxa used in a Kruskal-Wallis test.

Taxon

λmax

Lifestyle

Reference

Alligator mississippiensis

499

3

Lythgoe 1972, Smith et al. 1995

Ambystoma tigrinum

502

2

Makino et al. 1999

Anas platyrhynchos

505

1

Bowmaker et al. 1997

Anolis carolinensis

491

1

Kawamura and Yokoyama 1998

Bos taurus

500

1

Nathans and Hogness 1983

Bufo bufo

502

2

Ala-Laurila et al. 2002, Fyhrquist et al. 1998

Bufo marinus

503

1

Ala-Laurila et al. 2002, Fyhrquist et al. 1998

Caluromys philander

504

2

Hunt et al. 2003

Carassius auratus

492

3

Chang et al. 2002a

Columba livia

504

1

Bowmaker et al. 1997, Yokoyama et al. 2008

Coturnix japonica 

505

1

Bowmaker et al. 1997

Felis felis

500

2

Bridges 1970

Gallus gallus

504

1

Bowmaker et al. 1997, Yokoyama et al. 2008

Globicephala melas

488

3

Fasick and Robinson 2000

Harbour seal

501

3

Fasick and Robinson 2000

Homo sapiens

495

1

Chang et al. 2002a

Leiothrix lutea

500

1

Bowmaker et al. 1997

Macaca fascicularis

491

1

Baylor et al. 1984, Schnapf et al. 1988, Nickels et al. 1995

Melopsittacus undulatus

509

1

Bowmaker et al. 1997

Mesoplodon bidens

484

3

Fasick and Robinson 2000

Mirounga angustirostris

483

3

Southall et al. 2002

Mus musculus

498

2

Lythgoe 1972, Baehr et al. 1988

Ornithorhynchus anatinus

498

3

Davies et al. 2007

Taxon

λmax

Lifestyle

Reference

Oryctolagus cuniculus

502

2

Chang et al. 2002a

Petromyzon marinus

500

3

Zhang and Yokoyama 1997

Phoca groenlandicus

498

3

Fasick and Robinson 2000

Physeter macrocephalus

483

3

Southall et al. 2002

Polychrus marmoratus

497

1

Loew et al. 2002

Puffinus puffinus

505

1

Bowmaker et al. 1997

Python regius

494

2

Sillman et al. 1999

Raja erinacea

500

3

Chang et al. 2002a

Rana pipiens

502

2

Chang et al. 2002a

Rana temporaria

502

2

Koskelainen et al. 2000

Rattus norvegicus

500

2

Chang et al. 2002a

Sminthopsis crassicaudata

512

2

Hunt et al. 2003.

Spheniscus humboldti

504

3

Bowmaker et al. 1997

Strix aluco

503

2

Bowmaker et al. 1997

Tachyglossus aculeatus

497

2

Taeniopygia guttata

504

1

Bowmaker et al. 1997, Yokoyama et al. 2008

Trichechus manatus

502

3

Fasick and Robinson 2000

Xenopeltis unicolor

499

2

Davies et al. 2009

Xenopus laevis

502

3

Koskelainen et al. 2000

Lifestyle: 1 corresponds to diurnal, 2 to nocturnal, and 3 to aquatic life habits.

Analysis of variance (ANOVA) is the most commonly used technique for comparing the means of groups of measurement data. This kind of test is used when one deals with a nominal variable, which classifies observations into categories, and a measurement variable. The Kruskal-Wallis test is the non-parametric version of a one-way ANOVA and compares the medians of three or more samples (Fowler et al. 1995).

With a p-value = 0.5053, there are no significant differences in the medians between the samples. Interestingly, this indicates that inferring a lifestyle based on an animal’s rhodopsin absorption maximum is not statistically founded. Though it has been shown that amino acid substitutions at particular sites cause shifts in wavelength absorption, this study shows that ecological inferences based on λmax are not justified (Janz and Farrens 2001).

↓98

Since the Kruskal-Wallis test revealed that there is no correlation between wavelength absorption and lifestyle based on our data, a second test for correlation which also considers phylogeny (e.g. Independent Contrast Analysis) was redundant.

However, it should also be pointed out that one weak point of this analysis might be that published absorption maxima were determined inconsistently by differing methods, i.e. either after expression in COS-1 or HEK293 cells, or rhodopsins were purified from ROS, or they were determined using microspectrophotometry (MSP). Others determined the λmax based on a difference spectrum (dark spectrum - light spectrum) after in vitro expression. However, the effect on the consistency of the measurement based on the method of data acquisition has never been elucidated either. It would be useful to test if the various methods of λmax determination produce significantly different results.

4.1.4. Conclusions

The expressed echidna and its two mutant rhodopsins are functional pigments as indicated by the dark and light absorption spectra. Acid treatment also showed that the pigments are functional. Hydroxylamine assays and meta II decay rates by fluorescence spectroscopy indicate some cone-like properties of the three rhodopsins, which might have resulted from expression of rhodopsin in twin cones present in the echidna retina. Furthermore, though the role of the molar extinction coefficient in dim-light vision is not yet elucidated in detail, a substitution at site 169 has been found to be involved in decreasing the strength of photon absorption in the echidna. Paradoxically, a low ε appears disadvantegeous for scotopic vision.

↓99

The echidna rhodopsin seems to have achieved cone-like characteristics during its evolution, picturing its rhodopsin to be as enigmatic as the animal itself. Furthermore, it was shown that the protein-coding sequence of the rhodopsin of monotremes shares more amino acids with reptiles and amphibians than with other mammals. This mosaic pattern might be responsible for the yet contradictory results from the biochemical and functional assays.

A statistical test rejected any relationship between absorption maxima and life habits at different light levels. Thus, any habitat categorisation based on λmax, as is commonly done, is deficient.

The results interestingly show that, in contrast to prior assumptions, variation in the biochemical and functional properties of visual pigments seem unlikely to be due to ecological constraints, but rather result from interactions of the various proteins involved in the visual signaling cascade.

↓100

4.2. The ancestral rhodopsins

4.2.1. Characterisation of the three ancestral rhodopsins

The three inferred and successfully in vitro expressed ancestral pigments bound to 11-cis retinal to form functional pigments, as indicated by dark and light spectra (Fig. 17E-G, chapter 3.1.4.). The treatment with HCl acid showed that all pigments denatured within the first five minutes, which also indicates that all are functional pigments (Fig. 18E-G, chapter 3.1.5.). The functionality of these pigments is important, as the amino acid sequences were inferred using Maximum likelihood estimates; if they had not shown any functionality, the inference would have borne errors and the models chosen would have to be changed to better fit the data.

The expressed ancestral pigments have absorption peaks at 500 nm, 501 nm, and 500.5 nm for Amniota, Mammalia, and Theria, respectively, which is within the close range of bovine rhodopsin (Oprian et al. 1987, Stavenga et a. 1993).

↓101

The determined molar extinction coefficients are all higher than the one predicted for bovine (Tab. 11, chapter 3.1.5.), which indicates that more transducin molecules can be activated by the active state of rhodopsin. It has been shown that substituting a F for an A at site 169 decreases the strength of photon absorption in the echidna, as measured by the ε (see chapter 4.1.2.). Like bovine and all other placentals, the three ancestral pigments share an A at site 169. The determination of a high ε in all ancestral pigments suggests that, in addition to site 169, another site is likely to be involved in regulating the strength of photon absorption. Furthermore, accepting the common belief that a high ε is advantageous for vision at low light levels, the fact that the amniote rhodopsin has a ε similar to Mammalia and Theria indicates that the amniote ancestor had a rhodopsin maintaining a high degree of photon absorption as well, functioning well at low light levels.

The hydroxylamine assays showed that like bovine, neither of the three ancestral pigments reacted to hydroxylamine, also indicating rod-like pigments (Fig. 19E-G, chapter 3.1.6.).

In Figure 21D-F (chapter 3.2.3.), the 3D structure of the three inferred ancestral sequences were predicted based on their secondary structure. Though all three bear a few substitutions which differ from bovine, there is no change in conformation seen in the predicted 3D structure.

↓102

4.2.2. The meta II decay rate

The meta II decay assay by fluorescence spectroscopy, which measures the time constant for the active state of rhodopsin that is crucial for the visual signaling cascade, produced a very interesting result. Here, the t1/2 of the amniote rhodopsin is as high as that of bovine (Tab. 12, chapter 3.1.7.). They have t1/2 mean values of 16.05 min-1 (Amniota) and 17.01 min-1 (bovine) (Fig. 23), which are within the published range of bovine (Oprian et al. 1987, Stavenga et al. 1993). The Mammalia and Theria pigments, however, show a slower meta II decay rate (Fig. 23). The mammalian rhodopsin displays a mean t1/2 of 24.17 min-1 and the therian rhodopsin

one of 28.17 min-1 (Fig. 23). However, the last assay run did not provide a very confident R2 value in Theria; disregarding this one, the t1/2 is even higher, with a mean value of 32.67 min-1 (Fig. 23).

↓103

Figure 23. Phylogeny showing meta II decay rates derived from this study. 

It has been hypothesised that there is a correlation between the lifetime of meta II and the amplitude of rod response, indicating that larger amounts increase the signal arriving at the brain, as more transducin molecules can be activated (Shichida and Matsuyama 2009, Sugawara et al. 2010). Hence, a low meta II decay rate would be advantageous for scotopic vision (Sugawara et al. 2010). If this assumption was true, the results derived from this study would indicate that the mammalian rhodopsin experienced a change in function leading to better vision at low-light levels compared to the amniote ancestor, and that this functional change was preserved in the therian rhodopsin as well. In contrast, Sakurai et al. (2007) suggested that differences in the amplitude of the photoresponse are more likely due to intrinsic properties such as temperature dependence, rather than interactions between rhodopsin and other proteins from the visual signaling cascade.

Various meta II decay rates with varying assay conditions, such as temperature, have been examined, but most of these lack bovine as positive control. Hence, to ensure reliability, only two studies presenting t1/2 of chicken, human, and salamander, including bovine as positive control, were considered here (Okada et al. 1994, Imai et al. 2005). Human rhodopsin has a t1/2 similar to bovine, whereas chicken and salamander rhodopsin also display a t1/2 half that of bovine (Imai et al. 2005). However, values listed by Imai et al. (2005) cannot be tracked back in the literature. Thus, only the chicken meta II decay rate by Okada et al. (1994) can be used for comparative interpretations: with a value of 4.42 min-1 (bovine: 9.93 min-1), chicken rhodopsin displays a t1/2 half that of bovine, as is the case in the echidna. With chicken being a crepuscular animal displaying a rapid meta II decay rate like that of the nocturnal echidna, the t1/2 of meta II seems unlikely to allow for inferences on activity patterns as suggested previously (Shichida and Matsuyama 2009, Sugawara et al. 2010). The low t1/2 in bovine (9.93 min-1) might be due to experiments being performed at T=15ºC (Okada et al. 1994), whereas in this study, temperature was set to 25ºC. In addition, Imai et al. (2005) pointed out that meta II data obtained from spectroscopic assays using in vitro synthesised pigments differs from data acquired by membrane preparations. Okada et al. (1994) prepared chicken rhodopsin from ROS, in contrast to in vitro expression used in this study. For future research, assays with chicken should be replicated under the same conditions used in this study.

↓104

Alternatively, the rapid meta II decay rates in chicken and echidna, which has a „reptilian-like“ amino acid sequence, could display a phylogenetic pattern. However, the amniote ancestor has a t1/2 similar to bovine. But, this finding could be a result of the ancestral sequences being inferred using Maximum likelihood estimates, and, thus, being hypothetical. If indeed chicken and echidna display rapid meta II decay rates due to phylogeny, then the high t1/2 values for Mammalia and Theria would indeed indicate an adaptation to dim-light vision, as suggested previously (Shichida and Matsuyama 2009, Sugawara et al. 2010). However, the results rather emphasise that inferring ecological traits based on the investigation of single steps within the visual signaling cascade is problematic.

Sugawara et al. (2010) also found evidence to suggest that substitutions at site 83, among others, were resposible for a blue shift in the absorption spectrum of cichlid fishes, the result of adaptating to the blue-green photic environment in deep water. In nocturnal bats, this substitution was found to cause accelerated meta II formation rates, possibly as an adapatation to dim-light vision (Sugawara et al. 2010). Interestingly, the nocturnal echidna, whose rhodopsin has its λmax at 496.5 nm, which is slightly blue-shifted from bovine rhodopsin, also beares an Asparagine (N), which is said to cause a blue-shift and an accelerated meta II formation rate in fish and bats, in contrast to an Aspartic acid (D) in all others. Though being crepuscular, the chicken rhodopsin, which has a D at site 83, displays a rapid meta II decay rates. However, it has been suggested that, in contrast to meta II formation rates, meta II decay rates are not affected by substitutions at site 83 (Sugawara et al. 2010). Thus, analysing meta II formation rates might be helpful in future research.

Furthermore, Sugawara et al. (2010) discussed that residues 140 to 150 and 226 to 247 are involved in association with transducin, which is the crucial component affected by the meta II state (Weitz and Nathans 1993, Imai et al. 2005). There is one site within these regions where amino acids of the ancestral sequences differ from bovine, i.e. site 228 (Fig. 20, chapter 3.2.1.). Here, Amniota differs from bovine, Mammalia, and Theria in substituting a Cysteine (C) for a Phenylalanine (F) (Fig. 20, chapter 3.2.1.). Since bovine shares the same amino acid with Mammalia and Theria, this residue is unlikely to have influenced the detected high t1/2 value. Since all other residues are conserved, the suggested regions are unlikely to be involved in accelerating the meta II decay rate in Mammalia and Theria.

↓105

Another amino acid known to cause differences in meta II decay rates between chicken green opsin and rhodopsin is site 189 (Kuwayama et al. 2002). However, Amniota has a different amino acid at this site, i.e. Valine (V), than bovine, Mammalia, and Theria, which share an Isoleucine (I). Thus, this site is unlikely to affect the meta II decay rate in Mammalia and Theria (Fig. 20, chapter 3.2.1.). In addition, the replacement of Isoleucine by a Valine at this site caused no changes in the meta II decay rate, as indicated by site-directed mutagenesis (Kuwayama et al. 2002).

In conclusion, the inconsistency of the results derived from this as well as other studies emphasizes the high variability in the functional properties of visual pigments and demonstrates that single assays do not provide an adequate picture of the highly complex and interconnected visual system; not to mention their problematic use to infer the activity patterns of entire organisms.

4.2.3. Weak points of Maximum likelihood Inferences 

Though ancestral sequence reconstruction provides knowledge of ancient organismal biology where the fossil record reaches its limits, Maximum likelihood estimates also have their limits (Chang 2002a).

↓106

For example, it is possible that the ancestral reconstructions, which were inferred using likelihood methods, might not reflect the actual ancient gene sequence (Smith et al. 2010). However, they can be used as a good starting point for experimental tests (Chang 2003,

Ugalde et al. 2004). Here, ancestral sequences with the highest likelihood were chosen for in vitro expression (Tab. 8, chapter 3.1.2.). Future directions already involve in vitro expression of additional sequences, which were randomly sampled from the Bayesian distribution, as was done by Gaucher et al. (2010).

Furthermore, codon usage bias describes the phenomenon that the frequencey of occurence of codons in a protein-coding DNA sequence varies among species. It has been found to be present in rhodopsin (Chang and Campbell 2000). For example, reptile and amphibian rhodopsins tend to have more A’s and fewer G/C’s than all other sequences (Chang and Campbell 2000). Thus far, the Maximum likelihood approach used in this study does not account for this bias, which is a weak point of the approach. Unfortunately, at this time there is no better approach.

↓107

Also, ancestral reconstruction is sensitive to model choice (Chang 2003). However, since the inferred and expressed ancestral pigments were functional as indicated by dark-light spectra and acid bleach, it seems likely that the models used fit the data well.

4.2.4. Conclusions

It has been suggested that two unique properties were acquired by the rhodopsin from its cone ancestors for mediating scotopic vision: stability and a high amplification ability for phototransduction (Sakurai et al. 2007). A high amplitude of the single-photon response is likely to be achieved by a long lifetime of meta II, as more transducin molecules can be activated (Imai et al. 2005, Imai et al. 2007, Sugawara et al. 2010).

Meta II decay rates have been found to accelerate from node Amniota to Mammalia, suggesting that the mammalian rhodopsin experienced changes in order to adapt to dim-light vision. In Theria, this high meta II t1/2 is preserved. In contrast, a rapid meta II decay rate has been measured for the nocturnal echidna in this study and has been reported for the crepuscular chicken (Okada et al. 1994). Thus, the meta II decay data is inconsistent with activity patterns in echidna and chicken, and rather suggests that the visual system is too complex and interconnected, involving many proteins, to allow for ecological interpretations based on single biochemical and functional reactions.

↓108

Though the dark and light spectra indicated that all three ancestral pigments are functional, it must be emphasized that ancestral sequence reconstruction has its limitations, such as its hypothetical character as well as the non-consideration of a codon usage bias.

4.3. Positive selection on non-synonymous substitutions along the Therian branch

4.3.1. Therian diversity during the Late Jurassic 

The discovery of about 200 additional and exceptionally well preserved Mesozoic mammal fossils in the last 25 years have shaken the view of early mammals being only generalized forms (Luo 2007). Recently, it has been discovered that it is uncommon for any Mesozoic mammalian group to experience little or much delayed diversification (Luo 2007). Instead, early mammalian evolution is characterised by many short lineages in successive clusters (Luo 2007); although, this former view is still valid for the earliest forms such as Eozostrodon and Megazostrodon as well as for members of the Mesozoic Jehol Biota ecosystem (Luo 2007). However, there is now strong evidence for ecological specializations in many other early mammalian clades (Fig. 24) (Luo 2007). Though not very abundant in the Mesozoic, early mammals were highly diverse: modern lifestyles such as semi-aquatic, swimming, ambulatory, scansorial, climbing, fossorial, volant, and others had already evolved convergently in different taxa and clades during the Triassic and Jurassic (Fig. 24) (Luo 2007). Also a predatory carnivorous diet had evolved multiple times in unrelated mammalian groups during the Jurassic and Cretaceous, indicating an early evolution of food divergence (Luo 2007).

There is now evidence that there were six major diversification events in mammalian evolution, three of which occurred along the mammalian branch. As indicated by a grey dot in Figure 24, a first ecological diversification in early mammalian taxa took place during the Late Triassic and Early Jurassic (Luo 2007). It was followed by another remarkable diversification in ecological specializations in docodonts during the Middle Jurassic (see grey dot in Fig. 24) (Luo 2007). In the Late Jurassic, a third diversification followed within theriiform groups and taxa (see grey dot in Fig. 24) (Luo 2007).

↓109

Importantly, these three major diversifications happened along the branch leading from the node Mammalia to the node Theria, which is where the selective constraint analyses detected significant evidence for positive selection acting on the rhodopsin (Fig. 22, chapter 3.4.4.). Assuming that the earliest mammalian forms had indeed been nocturnal, it seems likely that the rhodopsin had undergone major changes in response to these new habitats at different light levels, in particular a semi-aquatic/swimming or fossorial/digging lifestyle; adaptations which are likely to be detected by selective constraint analyses. 

Figure 24. Phylogeny of Mesozoic and extant mammalian groups (after Luo 2007). 

Grey dots indicate starting points of ecological diversification events.

Ecological specialisations in early mammals include a semi-aquatic, swimming, ambulatory, scansorial, climbing, fossorial, and volant lifestyle. In detail, a swimming lifestyle first evolved in docodonts such as Haldanodon and Castorocauda (Fig. 24). Haramiyidans as well as early theriiform taxa, such as Fruitafossor and Repenomamus, were burrowing (Fig. 24). Volaticotherium was a gliding form (Fig. 24). Henkelotherium was arboreal and Vincelestes scansorial (Fig. 24). Early mammalian forms such as Sinocodon, Morganucodon, and others, as well as the theriiform taxon Yanoconodon were ground-dwelling (Fig. 24). 

4.3.2. The tetrapod opsin complement

↓110

The ancestral complement of visual pigments in tetrapods comprises four cone opsins for colour vision and one rhodopsin for vision at night and/or dim-light. As seen in Figure 25, this ancestral opsin set is reduced in all tetrapod clades. No green-sensitive opsin Rh2 has been found in any amphibian, but since it is found in reptiles and fish, it must have been present in the ancestor of amphibians and amniotes (Fig. 25) (Bowmaker 2008). All mammals have lost Rh2 (Fig. 25) (Hunt et al. 2009).

Davies et al. (2007) found exon 5 of the SWS1 gene in platypus, but Wakefield et al. (2008) found it neither in the platypus nor in the echidna and, thus, SWS1 is not functional in any living monotreme (Fig. 25). Zhao et al. (2009b) hypothesised that an ecological switch to a low-light habitat coincided with the loss or absence of functionality of the SWS1 opsin in marine mammals. All terrestrial mammals that have lost SWS1 are nocturnal (Peichl 2005, Carvalho et al. 2006, Jacobs 2009). One might infer that the early monotreme activity pattern had been nocturnal, as has been suggested by Crompton et al. (1978).

Theria, on the other hand, have lost SWS2, which absorbs blue light at around 410-490 nm (Cowing et al. 2008, Hunt et al. 2009). One might hypothesise that the strong positive selection both at branch level and acting on sites along the Therian branch, might be related to the fact that Theria had lost Rh2 and SWS2, and that their ancestor was only able to absorb UV (SWS1), red (LWS), and dark light (Rh1), as opposed to an amniote ancestor with an opsin set of Rh1, SWS1, SWS2, LWS, and Rh2 (Fig. 25). The loss of a visual pigment possibly puts another opsin, here rhodopsin, under selective constraint, in order to take over functional aspects; a selective constraint which is likely to be detected by the selective constraint analyses used in this study.

↓111

Figure 25. Visual pigment loss in tetrapods.

4.3.3. Selective constraint on synonymous substitutions in the mammalian
rhodopsin

Selection for particular codons, i.e. codon usage bias, has long been thought to be free of selection, suggesting an unbiased codon usage, as these substitutions do not lead to adaptive changes in the protein (Kimura 1968). However, this assumption has been challenged, and selection for synonymous sites has been found to be present in plants, bacteria, and invertebrates in order to increase translation efficiency/accuracy (Ikemura 1985, Wright et al. 2004, Cutter and Charlesworth 2006). In mammals, codon usage bias due to selective constraint was found to enhance mRNA stability and tRNA translation efficiency/accuracy, to maintain efficient splice control, and to ensure proper protein folding (Ikemura 1985, Parmley et al. 2006, Shabalina et al. 2006, Drummond and Wilke 2008). Furthermore, it has been suggested that genes with a high level of expression are likely to experience selective constraint on synonymous substitutions (Sharp et al. 1995). Rhodopsin is a highly expressed gene and mammalian rhodopsin has been found to have undergone a strong codon usage bias (Pugh and Lamb 1993, Chang and Campbell 2000).

↓112

A collaborative study using the same data set has shown that rhodopsin experienced selective constraint acting on synonymous substitutions in rhodopsin along the branch leading to Mammalia (Du 2010, unpublished MSc thesis). A strong codon usage bias towards G/C nucleotides at the 3rd position of four-fold codons was observed (Fig. 26) (Du 2010, unpublished MSc thesis). The LRTs of estimated data show significance (p < 0.001) (Du 2010, unpublished MSc thesis).

Figure 26. Distribution of G/C-ending codons in mammalian rhodospin gene.

Synonymous codons with highest fitness are highlighted by red codon ending. 

A preference for G/C-ending codons over A/T-ending codons has been found to increase mRNA stability and tRNA translation efficiency in mammals, suggesting an increase in rhodopsin molecules (Ikemura 1985, Shabalina et al. 2006, Drummond and Wilke 2008). Though in the majority of mammals, the retina is dominated by rods, nocturnal animals have been found to possess even more rod photoreceptors in their retina (Szél et al. 1996, Peichl 2005). An increase in rhodopsin molecules in the retina of the vertebrate eye is said to have resulted when adaptating to vision at night and/or low light leves (Kaskan et al. 2005, Peichl 2005).

↓113

This suggests that the mammalian rhodopsin had experienced changes in synonymous sites that led to an increased expression of molecules in the retina, which would have been supportive for adaptating to a nocturnal habitat. In addition, the study showed that there are mechanisms regulating adaptation to dim-light vision other than selection on non-synonymous sites causing adaptive changes.

4.3.4. Conclusions

Selective constraint can act either on synonymous or on non-synonymous substitutions. However, the effect on the protein is different. Positive selection acting on non-synonymous substitutions changes the amino acid sequence, which might affect the functionality or biochemical properties of a protein in order to adapt to external changes, whereas selective constraint on synonymous substitutions does not change the subsequent amino acid but instead increases mRNA stability and tRNA translation efficiency/accuracy (Ikemura 1985, Yang 2002, Shabalina et al. 2006, Drummond and Wilke 2008). Interestingly, selective constraint analyses investigateing both types of selection have shown that the mammalian rhodopsin had experienced important changes in both synonymous and non-synonymous substitutions: selective constraint acting on synonymous substitution sites along the branch leading to Mammalia was detected, and positive selection on non-synonymous substitutions was found within mammals, along the branch leading to Theria. These results suggest that early mammals have increased their number of rhodopsin molecules in order to adapt to a nocturnal habitat. Subsequently, their rhodopsin underwent functional and biochemical changes when taxa began exploring new habitats at different light levels, as indicated by the fossil record.

Furthermore, with only SWS1, LWS, and Rh1 opsins left in the retina, Theria have a very reduced opsin set as opposed to an amniote ancestor with an opsin set of Rh1, SWS1, SWS2, LWS, and Rh2. In order to compensate, the loss of an opsin is likely to put adaptive constraint onto another opsin, which possibly causes adaptive changes which are likely to be detected by selective constraint analyses.

4.4. Summary and future prospects

↓114

This thesis represents an integrative approach that combines paleontology and molecular biology in order to address an interesting question in evolutionary history: were the first mammals nocturnal?

1) The in vitro expression of the rhodopsin of the nocturnal echidna, together with two mutants T158A and F169A, was successful. All pigments are functional with λmax slightly blue-shifted from that of bovine. Results of the meta II decay assay, which measures the t1/2 of the active state of rhodopsin that is a crucial step in the visual signaling cascade, revealed a cone-like characteristic in the echidna rhodopsin, namely, a low t1/2. This finding stands in sharp contrast to prior assumptions that a high t1/2 is advantageous for scotopic vision. Hydroxylamine assays also describe these three pigments as cone-like, possibly a result of being expressed in cones as well. Further assays of the two mutants revealed that site 169 is involved in decreasing the strength of photon absorption in the echidna rhodopsin; another contradictory finding as a high strength of photon absorption is believed to be advantageous for vision at low light levels. The echidna rhodopsin is as enigmatic as the echidna itself.

2) Ancestral sequences for the nodes Amniota, Mammalia, and Theria were inferred using Maximum likelihood estimates and their in vitro expression was successful. All pigments were found to be functional and rod-like, with λmax within the range of bovine. Mammalia and Theria rhodopsin display high meta II half life times; a finding thought to be linked with adaptation to vision at low-light levels.

↓115

However, with regards to inconsistency in the available data, it must be emphasized that the visual signaling cascade is a complex and interconnected system involving numerous proteins. Therefore, inferences based on single biochemical and functional assays are problematic and do not allow for ecological interpretation.

3) Selective constraint analyses on non-synonymous substitutions were carried out. Interestingly, positive selection on non-synonymous sites, which is known to be adaptive, was found along the therian branch. This finding corresponds with recent paleontological data of three major events of ecological diversification along this branch. Changes involved in adapting to a new habitat at different light levels are likely to be detected by selective constraint analyses. Furthermore, selective constraint analyses on synonymous substitutions have revealed that the rhodopsin experienced non-adaptive changes, which nevertheless increase mRNA stability and/or tRNA translation efficiency/accuracy along the mammalian branch. This suggests a scenario in which rhodopsin molecules increased in number somewhere along the branch leading to crown mammals, in order to adapt to a low-light environment, followed by adaptive changes in the rhodopsin due to constraints resulting from ecological diversification or the loss of several cone opsins.

To date, the fossil record does not provide much information concerning nocturnality in early mammals, as preservation of soft-tissue is lacking (Ruben 1995). However, recently, it has been found that eyeball morpholgy is associated with the activity pattern of an animal (Walls 1942, Hall 2008a, Hall 2008b, Schmitz 2009). With scleral plates and other eyeball parameters being well preserved, it is now possible to infer activity patterns of extinct organisms, such as birds (Schmitz 2009). Scleral plates are not found in mammals, not even in the earliest forms or non-mammalian therapsids, but they are present in basal therapsids such as biarmosuchians, dinocephalians, anomodonts, and theriodontids (Fig. 2, chapter 1.1.3.) (Romer 1956, Sidor and Welman 2003, Sidor et al. 2004). Though ocular parameters needed for inferring activity patterns vary in birds and primates (Hall 2008), inferring visual capacities in therapsids based on eyeball dimensions could provide further insight as to whether early mammals had indeed been nocturnal.

↓116

Furthermore, from the molecular perspective, a switch from nocturnality to diurnality, or vice versa, has been observed in double-knockout mice lacking the inner-retinal photopigment melanopsin (OPN4) and RPE65, a key protein involved in retinal chromophore recycling (Doyle et al. 2008). Investigating these proteins by means of molecular evolution would be an intriguing direction for future research.

In addition, in order to visualize how the visual system works in a broader sense and to elucidate differences in the rhodopsin of a nocturnal and a diurnal animal, characterising the rhodopsin of a nocturnal placental sister taxon of bovine as second positive control is the next natural step for future research, and further assays, such as retinal regeneration, meta II formation rate, or transducin activation are needed (Chang et al. 2002a, Chang 2003, Janz and Farrens 2004, Sakurai et al. 2007, Sugawara et al. 2010).

In conclusion, this thesis contributes to knowledge about the origin and evolution of mammals in that three ancestral pigments inferred for the nodes Amniota, Mammalia, and Theria by Maximum likelihood estimates were successfully expressed in vitro, and were found to be functional and rod-like. The determination of meta II half life times tentatively indicate functional adaptation to vision at low light levels in the mammalian and therian rhodopsin. Furthermore, selective constraint analyses describe a scenario in which early mammals had increased the number of rhodopsin molecules in the retina, which was followed by adaptive changes in the amino acid sequence along the therian branch that were likely the result of exploring various novel habitats. Therefore, Crompton et al.’s hypothesis that early mammals had been nocturnal is supported by the results derived from this study.

↓117

In the coming years, the continued collaboration of paleontology and molecular biology could prove fruitful for addressing macroevolutionary questions and for peering deep into the past.


© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 4.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML-Version erstellt am:
28.09.2011