Constanze Bickelmann: Visual pigment evolution and the paleobiology of early mammals
[Titelseite] Kapitel: 1. \| 2. \| 3. \| 4. [Acknowledgements] [Literature cited] [Publications] [Conference presentations] [Erklärung] Zitierabschnitt: [Inhaltsverzeichnis]

3. Results

↓55

3.1. In the molecular lab

3.1.1. The echidna rhodopsin sequence

The sequencing of the echidna rhodopsin gene sequence was successful. Table 7 shows the genomic DNA (gDNA) and complementary (cDNA) sequence, and Figure 15 shows the protein-coding amino acid sequence with amino acids differing from bovine rhodopsin highlighted in green.

Table 7. Genomic DNA (gDNA) sequence, and complementary DNA (cDNA) sequence of the rhodopsin of the short-beaked echidna, Tachyglossus aculeatus.

	1
gDNA	ATGAATGGGA	CGGAGGGCCA	GGACTTTTAC	ATCCCCATGT	CCAATAAGAC	GGGGATTGTC
cDNA	ATGAATGGGA	CGGAGGGCCA	GGACTTTTAC	ATCCCCATGT	CCAATAAGAC	GGGGATTGTC

gDNA	AGGAGTCCCT	TTGAGTATCC	CCAGTATTAC	CTGGCAGAGC	CATGGCAGTA	CTCGGTCCTC
cDNA	AGGAGTCCCT	TTGAGTATCC	CCAGTATTAC	CTGGCAGAGC	CATGGCAGTA	CTCGGTCCTC

gDNA	GCTGCGTATA	TGTTCATGCT	CATCATGCTG	GGGTTCCCCA	TCAACTTCCT	CACGCTGTAC
cDNA	GCTGCGTATA	TGTTCATGCT	CATCATGCTG	GGGTTCCCCA	TCAACTTCCT	CACGCTGTAC

gDNA	GTCACCATCC	AGCACAAGAA	ACTCCGCACC	CCTCTCAACT	ACATCCTCCT	GAACCTGGCA
cDNA	GTCACCATCC	AGCACAAGAA	ACTCCGCACC	CCTCTCAACT	ACATCCTCCT	GAACCTGGCA

gDNA	TTTGCCAACC	ACTTCATGGT	GTTGGGTGGT	TTCACCACAA	CCCTGTATAC	TTCCCTGCAT
cDNA	TTTGCCAACC	ACTTCATGGT	GTTGGGTGGT	TTCACCACAA	CCCTGTATAC	TTCCCTGCAT
						*360*
gDNA	GGCTACTTTG	TTTTTGGACC	TACGGGCTGC	AACATCGAAG	GCTTCTTTGC	CACACTGGGA
cDNA	GGCTACTTTG	TTTTTGGACC	TACGGGCTGC	AACATCGAAG	GCTTCTTTGC	CACACTGGGA

gDNA	GGTAAGTTTC	CTCCAGGAGT	CCCCCTAGGA	GACGCTCTCC	TGGGCTATGA	CTTTTTTCCT
cDNA	----------	----------	----------	----------	----------	----------


gDNA	CCTGAAGGGA	GAGGAAAGAT	GTCAGCACCT	CCTCCCCACC	TGGGTAGGCC	GCCTTGCCGG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	CGGAAGTCAT	TTTCGAGCTA	ATACCGAGAA	GAGGCTGCTT	TGGCTAATAC	TGGGGACCGA
cDNA	---------	---------	---------	---------	---------	---------

gDNA	GGTCACAGCA	GATCGGGTCA	GTCACTCCAG	AGTCTCTGTC	CCACTCAGCC	CTGGCCCTTT
cDNA	----------	----------	----------	----------	----------	----------

gDNA	CTCTTGGAAT	TCTGAGTCTT	TTGGAAGGAG	AGTGCGGGCC	CCGAGATGAG	GACTGTTAAT
cDNA	----------	----------	----------	----------	----------	----------

gDNA	CGTTAACAGA	GAATGGCAGA	GACCAGCCTG	AGGCCTCCGA	GCAGGAGGTC	TTGTGGGATC
cDNA	----------	----------	----------	----------	----------	----------

gDNA	TGAGGGCAGG	GAGGACAGAA	ATATGGCACT	GGGGCGGAGA	GGGAGGCAGG	TCACCTTCTG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	TTTGGCACCC	AAGTCTCTGG	TAAGGAGTAT	GGGGTTCAGG	GAAGCCATCA	GGGAGCACAC
cDNA	----------	----------	----------	----------	----------	----------
gDNA	AGAGGCTTGG	AGTCTGACCC	CATTCTGCCA	CAAGCTTCCC	TTAAATGAGT	TCCTCGACCT
cDNA	----------	----------	----------	----------	----------	----------

gDNA	CTCTGCCCTT	CAGTTTGTCC	ACTGAGACTG	GGGTGGGGAG	AGAGACCCAG	GGGAGCAGAC
cDNA	----------	----------	----------	----------	----------	----------

gDNA	ACCTCAAAAC	ATGAAGTTCC	ATTATCAATC	CTAAAACCGC	CCTGAGAGTC	TAAATCAGGG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	GAGATTGGGA	GAGGTTGCCC	TTTTGTTCTG	GACCTGTAGC	TTCCCCAAGG	ATATCGCTAT
cDNA	----------	----------	----------	----------	----------	----------

gDNA	CTGGGGCAGG	AACCTATGGC	TCTTGCCTCA	GCTCAACCTC	CTGCTCCTGC	AGCCAGAGTG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	GGAGCCTGGC	ATGGGACAGG	GACGGTGTCT	GATCTGATGA	GCTGGTATCT	ACCCCCGCAC
cDNA	----------	----------	----------	----------	----------	----------

gDNA	CTTAGCCCAG	TGCTTTGGCA	CACATGAGCA	CTAAATAGAT	ACCCTAACTA	GCTTTGTGTC
cDNA	----------	----------	----------	----------	----------	----------
	*1266*
gDNA	TTGCAGGTGA	GATTGCGCTC	TGGTCTCTGG	TGGTGTTGGC	TATCGAGCGG	TATATCGTGG
cDNA	-----GGTGA	GATTGCGCTC	TGGTCTCTGG	TGGTGTTGGC	TATCGAGCGG	TATATCGTGG

gDNA	TCTGCAAGCC	TATGAGCAAC	TTCCGGTTTG	GGGAGAACCA	TGCCATCATG	GGTGTGACTT
cDNA	TCTGCAAGCC	TATGAGCAAC	TTCCGGTTTG	GGGAGAACCA	TGCCATCATG	GGTGTGACTT
						*1434*
gDNA	TCACTTGGAT	CATGGCCCTG	GCCTGTGCCT	TCCCCCCACT	CGTTGGCTGG	TCCAGGTACA
cDNA	TCACTTGGAT	CATGGCCCTG	GCCTGTGCCT	TCCCCCCACT	CGTTGGCTGG	TCCA------

gDNA	GGAGCTGCCT	GAAACCTGCT	CAGTAGCCCA	AGGGAAAGCC	CTGAAATGCC	AGGAGGAGGA
cDNA	----------	----------	----------	----------	----------	----------

gDNA	ACTCAGAGGG	GTTGGGATGG	GAGGGCATCC	TCAACTGTGC	CAGTGACGAA	GCTAGGTCTG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	CCAGGGTACC	TGCTCCCCTT	CTTCAACTTG	GCTTTTCCCT	AATCCTTAGC	TAACCTGGGG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	TTTCAAGTCA	AGCATCTTGA	ACAGAGCTAC	CCAAATCCTC	TGATGCAGCG	CTCCCATTGA
cDNA	----------	----------	----------	----------	----------	----------

gDNA	TATTGACCAT	GAGTTCTCCG	AGCCCATGGA	GATGGGGAGA	GATCACGTCT	CTGGAATTGG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	TGTTTGACAG	TGGGGAAATG	GCAGCTGTGG	AGGTGGTGTG	AGTTGGGAGT	GTCATTTGTT
cDNA	----------	----------	----------	----------	----------	----------

gDNA	TTAAAGAGAA	CAACCATAAT	AAAAATGACA	TTTGTTAAGC	GCTCTTTCTG	TGCCAAGCAC
cDNA	----------	----------	----------	----------	----------	----------

gDNA	TGTACTAAGC	GCTGGGGTAG	GTACAGGATA	ATCAGGTTAG	GCACAGTCCC	TGTCCCACCT
cDNA	----------	----------	----------	----------	----------	----------

gDNA	GGGATGAAGA	GTCTAAGTGG	AGGGGACTAT	TCATCCATAA	AGGTGTTTAG	TCCTGCTGAG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	GTGCAAAGAA	GTTCAGTGAC	TTGCTTAAAG	TCACACAGCA	GGCAGGTGGC	AGCTCTGGGA
cDNA	----------	----------	----------	----------	----------	----------

gDNA	TTAGAACCCA	GGTCCTCTGA	CTTCTAGTCT	GGTGCTCTCT	CCACTAAGCC	ACACTGCTTC
cDNA	----------	----------	----------	----------	----------	----------

gDNA	TCCCAGCTCT	AAAGGGTGAT	TAGAGAATCC	TTGGGCCAGA	GGAATCTCCC	TCAGCAGATT
cDNA	----------	----------	----------	----------	----------	----------

gDNA	GTCTCCACTT	CAGCCTCCAG	CAAAGCTATC	CCAGCCTCAG	CAGGCACCAA	CATGCCTGAC
cDNA	----------	----------	----------	----------	----------	----------
gDNA	CAACTGTCAA	GAAGATTCTA	CACCCTCTCC	CGGGGATCTG	TCATAGCTAA	GGAATACCAG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	ATCTCTTCTG	CAGTCGAAGC	CCATGCCTTG	ATCAAAAGCT	GTTCCCCTTC	CTCCTTACAG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	AAAGTCTAAA	CCCATCATAT	AATCTTTAGG	TTGAATGCCT	CCAATATGCC	CTCTTTGCCA
cDNA	----------	----------	----------	----------	----------	----------

gDNA	ATCTCCTCAC	ACATCTACCT	AGGGGGGCTG	CTAAATGGTA	ATGCGGTCAA	TCTGTCTGCA
cDNA	----------	----------	----------	----------	----------	----------
	*2461*
gDNA	GATATATCCC	CGAGGGTATG	CAGTGTTCGT	GTGGGATTGA	CTACTACACT	CTCAAACCTG
cDNA	GATATATCCC	CGAGGGTATG	CAGTGTTCGT	GTGGGATTGA	CTACTACACT	CTCAAACCTG

gDNA	AGGTCAACAA	TGAGTCCTTT	GTCATCTACA	TGTTTGTGGT	TCACTTCACC	ATCCCAATGA
cDNA	AGGTCAACAA	TGAGTCCTTT	GTCATCTACA	TGTTTGTGGT	TCACTTCACC	ATCCCAATGA
					*2628*
gDNA	CAATCATTTT	CTTCTGCTAC	GGCCGCCTGG	TCTTCACTGT	CAAAGAGGTG	AGCAAACCGT
cDNA	CAATCATTTT	CTTCTGCTAC	GGCCGCCTGG	TCTTCACTGT	CAAAGAG G--	----------

gDNA	CTCACGTGCA	TCTACCTGGG	GAGATTGGTT	CTGGTGTTCT	CTGCTGGCCT	AGCCCCTTTC
cDNA	----------	----------	----------	----------	----------	----------
						*2760*
gDNA	CTCAACTGCT	CCCCTCACGA	TTTCCTGCCT	GACCATCCCT	CTCTGCCCCC	CATTTTAGGC
cDNA	----------	----------	----------	----------	----------	---------C

gDNA	TGCAGCCCAG	CAGCAGGAGT	CCGCCACCAC	GCAGAAAGCT	GAGAAGGAAG	TCACCCGCAT
cDNA	TGCAGCCCAG	CAGCAGGAGT	CCGCCACCAC	GCAGAAAGCT	GAGAAGGAAG	TCACCCGCAT

gDNA	GGTGATCATC	ATGGTCATTG	CTTTCCTGAT	CTGCTGGGTG	CCCTACGCCA	GTGTGGCATT
cDNA	GGTGATCATC	ATGGTCATTG	CTTTCCTGAT	CTGCTGGGTG	CCCTACGCCA	GTGTGGCATT

gDNA	CTACATCTTC	ACACACCAGG	GATCAAACTT	CGGCCCCATC	TTCATGACTG	CCCCGGCTTT
cDNA	CTACATCTTC	ACACACCAGG	GATCAAACTT	CGGCCCCATC	TTCATGACTG	CCCCGGCTTT
						*2998*
gDNA	CTTTGCCAAG	AGTTCTGCGA	TCTACAACCC	AGTCATCTAC	ATTATGATGA	ACAAGCAGGT
cDNA	CTTTGCCAAG	AGTTCTGCGA	TCTACAACCC	AGTCATCTAC	ATTATGATGA	ACAAGCAG--

gDNA	AACCGAGAGC	GTGTCTGGTT	TGTCCTTACA	TATAAGTTAA	GGTGCGGCAA	GAGCCCCCAG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	CAGGCCGGGG	GGCGGGGGGG	AGGCAGGCAG	ATTCAATCAG	TCAATGGCAT	TTATCTAGTT
cDNA	----------	----------	----------	----------	----------	----------

gDNA	CTTGCTTATG	GTGGGCAGAG	TACTGGCCTG	AGCGTGTGGG	AAAATCCAAT	ACAATGGGGC
cDNA	----------	----------	----------	----------	----------	----------

gDNA	AGGTAGATGT	GATCCCTGCC	CCCAAGGAGC	TTACAGTCTA	GAGGGTCTAA	GTGGGTAGGG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	CAGGACAAGA	GTCTCGGAAG	GGCCCAGCCA	ATCGGCATGA	GGTAACAGGG	CCCCAAAAGT
cDNA	----------	----------	----------	----------	----------	----------

gDNA	TGGGAGACAG	GGGTTCTGGT	CTCCGTCCCT	CTTCCAGCTT	TGGTCCCCTC	TGACCTCCGG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	TAAACTTCTC	TATCCATACC	TCAGGGTGAC	AGTACTTGCC	TTCTCCCTTC	ACCTCTCAAG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	GATGAAGTAG	GGCAGAGTGA	AAGGGAACCC	AGATGAAGCC	AAATTCTCCG	GAGGGAGGTG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	CTCGCTCTGC	CAAGGTTGAA	GTCTGTTCCG	TTGACATCCT	CATGGGCTTC	TGTGGGCCTG
cDNA	----------	----------	----------	----------	----------	----------

gDNA	CAAAAATTGG	GTGGAAGACC	CCCCAAGTAC	CCTGCTGCAC	TGGTGCCAGA	ACTCAAGCTG
cDNA	----------	----------	----------	----------	----------	----------
gDNA	TCTGCTACCT	CCCCCTCCTC	ATTGTGCCAT	TGTTAGCATC	CTGCTGGGGA	TGGGGTGGGC
cDNA	----------	----------	----------	----------	----------	----------

gDNA	CTGGCGTGCC	TGAGCTTGGC	TATCAGCCTG	ATCTAGAAAG	GGGCTGACTG	TTGATTGTGG
cDNA	----------	----------	----------	----------	----------	----------

					*3762*
gDNA	TCTCCTTGTC	CTGGTTTCCA	ACCTAATGCT	TCCTCCCCCA	GTTCCGGAAC	TGCATGCTCA
cDNA	----------	----------	----------	----------	-TTCCGGAAC	TGCATGCTCA

gDNA	CCACCATCTG	CTGCGGCAAG	AACCCGCTGG	GCGATGATGA	GGCTTCGGCC	ACAGCTTCCA
cDNA	CCACCATCTG	CTGCGGCAAG	AACCCGCTGG	GCGATGATGA	GGCTTCGGCC	ACAGCTTCCA

					*3887*
gDNA	AGACCGAGCA	GTCTTCCGTG	TCCACCAGCC	AGGTTTCTCC	AGCATAG
cDNA	AGACCGAGCA	GTCTTCCGTG	TCCACCAGCC	AGGTTTCTCC	AGCATAG

Exons are highlighted in red.

↓56

Figure 15. Secondary structure of the echidna rhodopsin (modified after Sakmar et al. 2002).

Amino acids differing from bovine rhodopsin are highlighted in green.

3.1.2. Three ancestral sequences

Three inferred ancestral sequence for the nodes Amniota, Mammalia, and Theria were inferred by the M3 model in PAML (Tab. 8).

↓57

Table 8. Most likely hypothetical ancestral nucleotide sequences for the nodes Amniota, Mammalia, and Theria, inferred by maximum likelihood estimates.


	0	1	2	3	4	5
Amniota	MNGTEGPNFY	VPMSNKTGVV	RSPFEYPQYY	LAEPWQYSAL	AAYMFMLILL	GFPINFLTLY
Mammalia	MNGTEGPNFY	VPMSNKTGVV	RSPFEYPQYY	LAEPWQYSVL	AAYMFMLIVL	GFPINFLTLY
Theria	MNGTEGPNFY	VPFSNKTGVV	RSPFEYPQYY	LAEPWQFSVL	AAYMFMLIVL	GFPINFLTLY

					1	1
	6	7	8	9	0	1
Amniota	VTIQHKKLRT	PLNYILLNLA	VADLFMVLGG	FTTTMYTSMN	GYFVFGPTGC	NIEGFFATLG
Mammalia	VTIQHKKLRT	PLNYILLNLA	VADLFMVFGG	FTTTLYTSLH	GYFVFGPTGC	NIEGFFATLG
Theria	VTIQHKKLRT	PLNYILLNLA	VADLFMVFGG	FTTTLYTSLH	GYFVFGPTGC	NLEGFFATLG

	1	1	1	1	1	1
	2	3	4	5	6	7
Amniota	GEIALWSLVV	LAIERYVVVC	KPMSNFRFGE	NHAIMGVAFT	WIMALACAAP	PLFGWSRYIP
Mammalia	GEIALWSLVV	LAIERYVVVC	KPMSNFRFGE	NHAIMGVAFT	WIMALACAAP	PLVGWSRYIP
Theria	GEIALWSLVV	LAIERYIVVC	KPMSNFRFGE	NHAIMGVAFT	WIMALACAAP	PLVGWSRYIP

	1	1	2	2	2	2
	8	9	0	1	2	3
Amniota	EGMQCSCGVD	YYTLKPEVNN	ESFVIYMFVV	HFTIPLTIIF	FCYGRLVCTV	KEAAAQQQES
Mammalia	EGMQCSCGID	YYTLKPEVNN	ESFVIYMFVV	HFTIPMTIIF	FCYGRLVFTV	KEAAAQQQES
Theria	EGMQCSCGID	YYTLKPEVNN	ESFVIYMFVV	HFTIPMIVIF	FCYGQLVFTV	KEAAAQQQES

	2	2	2	2	2	2
	4	5	6	7	8	9
Amniota	ATTQKAEKEV	TRMVIIMVIS	FLICWVPYAS	VAFYIFTNQG	SDFGPIFMTV	PAFFAKSSAI
Mammalia	ATTQKAEKEV	TRMVIIMVIA	FLICWVPYAS	VAFYIFTHQG	SNFGPIFMTV	PAFFAKSSAI
Theria	ATTQKAEKEV	TRMVIIMVIA	FLICWVPYAS	VAFYIFTHQG	SNFGPIFMTL	PAFFAKSSAI

	3	3	3	3	3	3
	0	1	2	3	4	5
Amniota	YNPVIYIVMN	KQFRNCMITT	LCCGKNPLGD	DETSAAAGTT	KTETSSVSTS	QVSPA
Mammali a	YNPVIYIMMN	KQFRNCMLTT	LCCGKNPLGD	DEASATAGTS	KTETSSVSTS	QVSPA
Theria	YNPVIYIMMN	KQFRNCMLTT	LCCGKNPLGD	DEASATAGTS	KTETSQVATS	QVSPA

3.1.3. Western blot

In order to confirm that the correct proteins had been expressed in HEK293 cells, a SDS-PAGE analysis transferred onto a nitrocellulose membran was performed on harvested samples (Fig. 16).

Fig. 16A shows the bovine rhodopsin used as control, as well as the echidna protein and the two mutants. Fig. 16B shows the bovine control and the three ancestral pigments. The bovine sample was diluted 1:2 due to its high expression yield.

↓58

Bovine rhodopsin has a molecular weight of around 30 kDa (Frank and Rodbard 1975, Reeves et al. 1996); the corresponding band is seen in Fig. 16. All other samples display two distinct bands at around 36 and 40 kDa (Fig. 16). The echidna rhodopsin and the two mutants are 363 amino acids long, which is 5 amino acids longer than bovine rhodopsin, due to a non-therian insertion from AA 358 to 363 in the former (Tab. 4, chapter 2.2.2.). The ancestral pigments also carry this insertion and are 367 amino acids long (Tab. 8, chapter 3.1.2.). This is responsible for the greater size of expressed visual pigments other than bovine. Bovine rhodopsin monomers are seen in an additional band at around 32 kDa (Fig. 16). In addition, echidna rhodopsin and the two mutants show two additional faint bands at around 50 kDa (Fig. 16A). However, the presence of multiple bands is most likely due to proteins undergoing different post-translational modifications, which can differ in different cell types (Reeves et al. 1996, Wong 2006). Proteins are often synthesized with an extra short peptide in the N-terminal end in order to keep the protein in a nonfunctional form until it is activated into the more mature form, or to guide the protein through various compartments in the cell (Wong 2006). Thus, a subsequent treatment with N-glycosidase F would help to remove all N-linked glycosidation, but unfortunately this was not possible due to technical reasons.

The molecular weight (MW) of each expressed protein was determined with an online tool for calculating the MW based on an input protein sequence (Tab. 9). The results indicate that the upper band in each lane in the western blot, which is at around 40 kDa, is the correct one.

(http://www.expasy.ch/tools/pi_tool.html).

↓59

Table 9. Molecular weight estimates based on protein sequences

Rhodopsin	MW
Bovine	38.54
Echidna	39.96
Mutant T158A	39.93
Mutant F169A	39.88
Amniota	39.60
Mammalia	39.67
Theria	39.69

Interestingly, the echidna rhodopsins and the two mutants show only faint bands after being exposed for 3 minutes, whereas bovine and the ancestral pigments show a very strong band, after being exposed for only 1 second. This indicates that the ancestral pigments were expressed much better, which could be due to the fact that their gene sequences had been optimized for expression in mammalian cells.

Figure 16. Western blot analysis of expressed rhodopsin pigments.

(A) From left to right: Bovine, Echidna, mutant T158A, and mutant F169A rhodopsin. (B) From left to right: Bovine, Amniota, Mammalia, and Theria rhodopsin.

3.1.4. Dark and light spectra

↓60

Figure 17 shows dark aborption spectra of all visual pigments expressed in this study. For an accurate determination of λ_max, absorption spectra were curve fitted following Govardovskii’s method (Govardovskii et al. 2000). Ideally, for a reliable determination of λ_max, the curve fitting should be performed at least three times on rhodopsin data from different expressions. However, this was not possible due to technical reasons.

Nonetheless, the following absorption peaks were determined and are shown in Table 10. With a determined λ_max at 500 nm, the bovine rhodopsin expressed in this study shows an absorption peak that falls within the published range (Oprian et al. 1987, Stavenga et al. 1993).

↓61

Table 10. Absorption peaks of all rhodopsins expressed in this study.

Rhodopsin	λ_max in nm
Bovine	500
Echidna	496.5
Mutant T158A	494.5
Rhodopsin	λ_max in nm
Mutant F169A	495.5
Amniota	500
Mammalia	501
Theria	500.5

Absorption spectra were curve fitted following Govardovskii’s method (Govardoskii et al. 2000).

After the dark absorption spectra were taken, pigments were bleached with light for 60s (Fig. 17). A light-bleached opsin shows a characteristic absorption curve with a peak at 380 nm, due to the unquenching of tryptophan after irradiation and subsequent deprotonation of the Schiff base (Farrens and Khorana 1995, Schädel et al. 2003, Salom et al. 2006). This shift in λ_max indicates that each expressed pigment is indeed functional (Fig. 17).

Figure 17. Dark (in red) and light (in black) absorption spectra of expressed and purified rhodopsins, i.e.

(A) bovine, (B) echidna, (C) mutant T158A, (D) mutant F169A, (E) Amniota, (F) Mammalia, and (G) Theria rhodopsin. λ_max of expressed rhodopsins: Bovine: 500 nm, Echidna: 496.5 nm, mutant T158A: 494.5 nm, mutant F169A: 495.5 nm, Amniota: 500 nm, Mammalia: 501 nm, and Theria: 500.5 nm.

↓62

The ratio of UV to visible absorbance (A₂₈₀/A_max) was also determined using the dark absorption spectra data. It is the amount of protein in a sample over the amount of absorbing protein in the sample, i.e. the expression yield. For expression in COS-1 cells, a ratio of around 3 was observed (Oprian et al. 1987), as opposed to a ratio of 1.6-1.7 when prepared from rod outer segments (ROS) (Hong et al. 1982). Sakamoto and Khorana (1995) prepared bovine rhodopsin from ROS and reported a ratio of 1.7-1.8. ROS prepared bovine rhodopsin displayed a ratio of around 2 (Radding and Wald 1956). A ratio below 1.6 is considered to indicate a purity close to 100% (Ernst et al. 2007).

All rhodopsins expressed in this study, including bovine, showed a A₂₈₀/A_max ratio in the same range per expression. Ratios between 2.3 to 3.7 were observed.

3.1.5. Acid bleach

Acid bleaches were performed on echidna and its mutants as well as on the three ancestral rhodopsin pigments, including bovine as positive control (Fig. 18). A shift from λ_max to 440 nm at 20°C indicates the break-off of the chromophore from the opsin, relinquishing a protonated Schiff base 11-cis retinal free in solution (Kito et al. 1968); hence, a functional rod pigment.

↓63

Figure 18 shows the difference absorbance over time of all acid treated pigments. The white circles indicate difference absorbance at 440 nm and are expected to increase and then stabilize once the chromophore and the opsin are indeed detached. The black circles indicate difference absorbance at λ_max of each rhodopsin and are expected to decline.

In Figures 18B-D there is an initial drop in difference absorbance at 440 nm, which can be explained by bubbles that formed when adding the HCl and which disturbed the reading of the spectrophotometer.

However, the echidna rhodopsin and the two mutants did not react to the acid as quickly as bovine, which occured immediately right after the addition (Figs. 18A, B). Still, within 10 minutes the protonated Schiff base (PSB) had formed. The two mutants reacted to HCl similar to echidna (Figs. 18C, D). For all ancestral rhodopsins, the acidification was complete within 5 minutes; the therian rhodopsin reacted as quickly as the bovine one (Figs. 18E-G).

↓64

Figure 18. Acid bleaches of (A) bovine, (B) echidna, (C) mutant T158A, (D) mutant F169A, (E) Amniota, (F) Mammalia, and (G) Theria rhodopsin.

White circles indicate absorption at 440 nm; black ones indicate absorption at λ_max.

In addition, the molar extinction coefficient of a visual pigment can be estimated based on acid treatment data (Radding and Wald 1956, Starace and Knox 1998). It is a measure of how strongly a chemical absorbs light at a given wavelength.

There are various extinction coefficients published for bovine rhodopsin (estimated λ_max = 498 - 500 nm), ranging from 40 600 to 43 000 M^-1 cm^-1 (Wald and Brown 1953, Shichi et al. 1969, Daemen et al. 1970, Hong and Hubbell 1972, Oprian et al. 1987). All estimated extinction coefficients are shown in Table 11.

↓65

Table 11. Molar extinction coefficients determined for all proteins expressed in this study.

Rhodopsin	ε in M^-1 cm^-1
Bovine	40 622
Echidna	34 921
Mutant T158A	31 411
Mutant F169A	40 254
Amniota	49 169
Mammalia	46 961
Theria	45 460

3.1.6. Hydroxylamine sensitivity

All pigments, including bovine rhodopsin as positive control, were treated with 1 M hydroxylamine (NH₂OH) for 2 hrs (Fig. 19). Hydroxylamine assays are used to distinguish between rod and cone opsins, with cone opsins reacting quickly to the compound and forming a retinal oxime, which absorbs light at around 363 nm, and rod opsins not shifting their absorption peak for an extended period of time (Wald et al. 1955, Fager and Fager 1981, Okano et al. 1989, Wang et al. 1992, Starace and Knox 1998). Bovine rhodopsin is known to stay stable in the presence of hydroxylamine for at least 12 hrs (Kawamura and Yokoyama 1998).

In this study, the bovine rhodopsin positive control reacted little to hydroxylamine for the 2 hours during which the measurements were taken, though the dots are very scattered, which is due to the spectrophotometer (Fig. 19A). Also, the degree of increase in difference absorbance at λ_{363 nm}is not very high. An incipient rise is normal, as long as the curve evens out after several minutes. The observed drop in difference absorbance is due to the presence of bubbles or a change in properties of the solution, as was also the case in the acid bleach (Fig. 19A).

↓66

Interestingly, the echidna rhodopsin and the two mutants reacted to hydroxylamine more than bovine, as indicated by an increase in difference absorbance of more than 0.005 (Figs. 19B-D). However, cone opsins react to hydroxylamine much stronger (Kawamura and Yokoyama 1998, Starace and Knox 1998). Also, since there were only two runs performed for each mutant, a third run should be performed for a more reliable result.

Figures 19E-F show, though the data points are also somewhat scattered, that the amniote and mammalian rhodopsins react to hydroxylamine just as little as the bovine one. For the Theria rhodopsin, one of the three curves rises slightly, but the other two do not show a strong increase in absorption (Fig. 19G).

The determination of t_1/2 of hydroxylamine treated pigmnents was not possible, because the data points are too scatterered and R² values are not reliable.

↓67

In conclusion, there is some indication that echidna and the two mutants are not as stable in the presence of hydroxylamine as bovine rhodopsin, which indicates cone-like characteristics. All ancestral pigments, however, are as insensitive to hydroxylamine as bovine rhodopsin.

Figure 19. Hydroxylamine assays performed on (A) bovine, (B) echidna, (C) mutant T158A, (D) mutant F169A, (E) Amniota, (F) Mammalia, and (G) Theria rhodopsins.

Circles indicate different runs.

3.1.7. Meta II decay by fluorescence spectroscopy

Meta II is the active state of rhodopsin and a key intermediate in the visual signaling cascade where the crucial transducin activation takes place (Fig. 13, chapter 2.1.8.) (Weitz and Nathans 1993, Imai et al. 2005, Sugawara et al. 2010). Here, the opsin and the chromophore are still bound but the Schiff base is deprotonated, unquenching tryptophan, and has its λ_max at 380 nm (Farrens and Khorana 1995, Sakmar et al. 2002, Heck et al. 2003, Salom et al. 2006). The rhodopsin meta II state is induced by light bleach and finished with the addition of fresh 11-cis retinal, which binds to rhodopsin molecules.

↓68

The results of the meta II decay rate assays performed in this study are given in tables 12 and 13. Bovine meta II decay rates are more or less within the expected range of 15 min^-1 (Tab. 12, 13) (Janz and Farrens 2001, Reeves et al. 1996). The amniote rhodopsin displays a t_1/2 similar to bovine (Tab. 12). Most striking are the results for the mammalian ancestor, where t_1/2is much higher than those of bovine and amniote (Tab. 12). Also, the therian rhodopsin displays a high t_1/2, similarto the mammalian one (Tab. 12). On the other hand, the echidna displays a much lower t_1/2 than bovine (Tab. 13). Due to technical reasons, meta II decay rates were not determined for the two mutants.

Table 12. Meta II decay results and their coefficient of determination (R²) of ancestral pigments and bovine rhodopsin as positive control.

Expressed rod pigment	t_1/2 in min^-1	R²
Bovine	16.46 ⁽¹⁾	0.9985
	17.24 ⁽²⁾	0.9990
	21.39 ⁽³⁾	0.9932
	12.95 ⁽⁴⁾	0.9921
Amniota	16.74 ⁽¹⁾	0.9989
	16.54 ⁽²⁾	0.9992
	17.07 ⁽³⁾	0.9976
	13.85 ⁽⁴⁾	0.9924
Mammalia	21.33 ⁽¹⁾	0.9986
	22.36 ⁽²⁾	0.9994
	22.43 ⁽³⁾	0.9938
	30.54 ⁽⁴⁾	0.9989
Theria	33.98 ⁽¹⁾	0.9988
	25.30 ⁽²⁾	0.9987
	38.72 ⁽³⁾	0.9977
	14.69 ⁽⁴⁾	0.9120

Hyphenated numbers in brackets indicate number of expression and assay run.

↓69

Table 13. Meta II decay results and coefficients of determination (R²) of echidna rhodopsin and bovine as positive control.

Expressed rod pigment	t_1/2 in min^-1	R²
Bovine	12.2 ⁽⁵⁾	0.9979
	13.8 ⁽⁶⁾	0.9979
	13.0 ⁽⁷⁾	0.9977
	14.1 ⁽⁸⁾	0.999
	13.8 ⁽⁹⁾	0.9987
Echidna	10.3 ⁽⁵⁾	0.9957
	9.9 ⁽⁶⁾	0.9968
	6.6 ⁽⁷⁾	0.9943
	6.1 ⁽⁸⁾	0.9974
	6.7 ⁽⁹⁾	0.9972

Hyphenated numbers in brackets indicate number of expression and assay run.

3.2. The ancestral sequences and their structure

3.2.1. Interesting sites

Site-directed mutagenesis is often used in vision research in order to identify key sites being responsible for causing dramatic changes within the visual pigment (Imai et al. 1997, Carvalho et al. 2006).

For the three inferred ancestral proteins, there are 10 residues at which Amniota and Mammalia differ from the Therian sequence (Fig. 20). Amniota differs from Mammalia and Theria at 28 sites (Fig. 20).

↓70

According to Hildebrand et al. (2009), residues 37, 39, and 290 are located within the hole where the chromophore enters the binding pocket and might be involved in holding it. Site 95 is not believed to be involved in shifting λ_max (Yokoyama et al. 2008). Residue 112 may be of interest as it is next to 113, which was found to be a negatively charged counterion that stabilizes the positively charged PBS (Hildebrand et al. 2009, Shichida and Matsuyama 2009). Substitutions at site 189 cause differences in the molecular properties of rods and cones (Imai et al. 2007, Lamb et al. 2007). Mutants with substitutions at this site were found to fold incorrectly (Doi et al. 1990). A site-directed mutagenesis study by Chang et al. (2002a) showed that site 218 does not have any effect on spectral tuning or transducin activation. According to Wakefield et al. (2008), site 308 causes spectral tuning in human and platypus.

Interestingly, all three ancestral sequences have the insertion of five amino acids between position 349 and 353, which is lost in all living Theria, but retained in living monotremes and living non-mammalian tetrapods (Tab. 4, chapter 2.2.2.). Its presence in the hypothetical

Theria sequence reflects the arithmetic of the Maximum Likelihood approach and indicates that it became lost independently in marsupials and placentals.

↓71

Figure 20. Amino acid alignment of the three inferred ancestral rhodopsins.

Blue bars indicate residues where Amniota and Mammalia differ from Theria. Pink bars indicate residues where Amniota differs from Mammalia and Theria, and where Bovine is different from Mammalia and Theria. Yellow bars indicate residues where Amniota differs from Mammalia and Theria, and where Bovine shares the same residue with Mammalia and Theria. The red boxes indictate BEB sites inferred by PAML (Tab. 22, chapter 3.4.3.).

Future directions for research already involve creating and expressing mutants at some of these interesting sites, allowing the determination of if and which ones are responsible for differences in the biochemistry and functionality of the ancestral pigments. Such a study could potentially elucidate which changes these sites experienced while the organism was adapting to a new environment.

3.2.2. Rhodopsin 3D structure

↓72

Rhodopsin is a well studied G protein-coupled receptor. It is now possible to examine its 3D structure with the help of molecular visualization programs, such as PyMOL (www.pymol.org). This method helps to locate sites that might influence the biochemical and functional properties of the rhodopsin of various taxa. In addition, it is possible to infer the 3D structure of hypothetical rhodopsins based on their protein-coding sequence, in order to see if differing amino acids have any effect on the 3D structure of the protein.

Figure 21. Rhodopsin 3D structure of all pigments from this study.

(A) shows the echidna rhodopsin with amino acids differing from bovine rhodopsin highlighted in gray. Red marks indicate the substitutions of mutants (B) T158A and (C) F169A. (D-F) Ancestral pigments, i.e. (D) Amniota, (E) Mammalia, and (F) Theria.

3.3. Comparing protein-coding rhodopsin sequences from living taxa

Taking a closer look at the 27 tetrapod rhodopsin amino acid sequences, several interesting substitutions were identified, i.e. substitutions unique to a taxon, a monophyletic group, or individual clades (Tab. 4, chapter 2.2.2.).

↓73

3.3.1. Substitutions unique to a taxon

The lungfish bears the highest number of unique substitutions of all taxa studied, which is nine in total. This is followed by dunnart (seven substitutions), and toad, snake, anole, and bovine (five substitutions each).

The echidna has two unique substitutions at site 158 and 169, which were also chosen for site-directed mutagenesis (see chapter 2.1.3).

↓74

Interestingly, rhodopsin sequences from eutherian (placental) taxa, especially Euarchontoglires (i.e. Glires and Primates), do not exhibit that many unique substitutions compared to the rest of tetrapods. Furthermore, sequences of the manatee, dog, guinea pig, and human do not display a single unique substitution.

3.3.2. Substitutions unique to monophyletic groups

Reptiles, including birds, have a couple of very interesting unique features in their rhodopsin sequences: together with the lungfish, they have lost a residue at site 337; and at site 133, except for the alligator, they share a Valine instead of the Isoleucine present in all other taxa.

Amino acids shared by most mammalian sequences are at residues 95, 99, 100, 107, 216, 228, 308, 318, and 333.

↓75

Monotremes carry two unique substitutions, i.e. at residues 39 and 344. In general, they share more amino acids with reptilian and other non-mammalian vertebrates than with Theria, such as at residues 13, 83, 88, 112, 225, 346, and 348. In addition, monotreme sequences have an insertion of five amino acids between position 349 and 353, which is lost in Theria, but retained in lungfish, coelacanth, amphibian, and reptilian sequences (Hunt et al. 2003). These residues are known to interact with rhodopsin kinase (Nathans and Hogness 1983).

Marsupial rhodopsins have a Glutamic acid at residue 26, whereas other tetrapod taxa have a Tyrosine, except for unique substitutions in bovine and polar bear.

Placental sequences differ from all others, except for alligator and lungfish, only at site 63.

↓76

At site 333, Afrotheria have a unique substitution: a Glycine.

3.3.3. Similar substitutions in different clades

At site 338, lungfish and coelacanth are the only taxa that bear an amino acid at all; it is lost in all tetrapods. Lungfish and coelacanth sequences share an Aspartic acid with squamates at site 33, while all others have a Glutamic acid. At site 286, lungfish and coelacanth share a Valine with reptilian sequences, except for the chicken, which has an Isoleucine like mammals and amphibians. At residue 39, lungfish and coelacanth and reptiles share an Alanine, monotremes have a Valine, marsupials a Cysteine, and placentals a Methionine, except for the guinea pig.

At residue 290, amphibians share a Valine, reptiles and artiodactyls an Isoleucine, and marsupials, afrotherians, and carnivors a Leucine. All others are not consistent.

↓77

Amphibian and archosaur sequences share an Asparagine at site 277, all others have a Histidine.

Monotremes share a Valine with amphibians at site 81. At residue 88, they share a Leucine with amphibians and reptiles, except for salamander and chicken.

At residue 63, monotreme and marsupial rhodopsins share a residue with reptilian ones rather than placentals. And an Isoleucine at site 137 distinguishes monotremes and marsupials from placentals.

↓78

At site 37, therian rhodopsin sequences share a Phenylalanine with that of most reptiles.

Afrotheria have a few substitutions that they share with other groups, for example, at site 328 they have a Phenylalanine in common with lungfish and coelacanth and amphibians. At residue 331, they share a Glutamic acid with lungfish and coelacanth and some reptiles.

3.4. Selective constraint acting on the rhodopsin visual pigment

3.4.1. Introduction

In order to test the hypothesis that early mammals had indeed been nocturnal, selective pressure acting on the visual pigment resposible for vision at night, the rhodopsin, was assessed by using a maximum likelihood approach that estimates ω, which is the ratio of non-synonymous substitutions to synonymous substitutions. To determine the type and degree of selective constraint, branches of interest were selected as foreground branches with their own estimated ω, one that is different from the background branches, which have a combined ω estimated for all branches. When these two groups are compared, if one has a higher ω, then either the one with the higher ratio has experienced relaxed purifying selection, or the one with the lower value has undergone stronger purifying selection than the other. Positive selection is indicated if ω is significantly greater than 1.

↓79

Here, the amniote, reptilian, mammalian, monotreme, therian, marsupial, and placental branches were of interest and marked separately. Then, comparison of alternative and null models as well as significant LRTs tell us if there was positive or relaxed purifying selection acting on the rhodopsin.

3.4.2. Branch models

A comparison of MB2a and MB1n using siginficant LRTs determines whether the foreground branch is significantly different from the background d_N/d_Sratio. If the ω ratio of the foreground branch in MB2a is estimated to be greater than 1, this indicates either relaxed purifying or positive selection. Comparing MB2a and MB2n tests whether the branch of interest has a d_N/d_S ratio that is significantly different from 1, if supported by significant LRTs. If ω₁ is estimated to be greater than 1, positive selection is indicated.

↓80

The first branch of interest is the amniote one. With Amniota marked as a foreground branch, we find a value of 999 in model MB2a (Tab. 14). In PAML 4, the number 999 is the upper bound set for ω, meaning the actual value is not known, it might even represent infinity (Yang 2007). The LRT of the comparison of MB2a and MB1n does not show significance. Hence, the foreground value 999 is not significantly different from the background value 0.0532 and, thus, there is no indication for any positive selection along this branch (Tab. 14). Testing whether this value is significantly different from 1, does not show statistical significance using the LRT comparing models MB2a and MB2n (Tab. 14). However, because the foreground ratio is much larger than the background branch, this indicates slightly relaxed selective constraint.

Table 14. Branch model estimates for the branch Amniota. np is number of parameters, LnL is log likelihood of the model.

Model	ω₀	ω₁	np	LnL	p-value
Alternative model MB2a	0.0532	999	54	-10646.2

First null model MB1n	0.05432	0.05432	53	-10649.5	MB2a vs MB1n
					0.0703

Model	ω₀	ω₁	np	LnL	p-value
Second null model MB2n	0.0533	1	53	-10646.3	MB2a vs MB2n
					0.8069

In the reptilian branch, the foreground ratio in null model MB2n is 999 compared to a background ratio of 0.0537 (Tab. 15). However, neither LRTs of comparing models MB2a and MB1n nor models MB2a and MB2n provide statistical support (Tab. 15). As for Amniota, this also suggests slightly relaxed purifying selection.

↓81

Table 15. Branch model estimates for the branch Reptilia. np is number of parameters, LnL is log likelihood of the model.

Model	ω₀	ω₁	np	LnL	p-value
Alternative model MB2a	0.0537	999	54	-10647.5

First null model MB1n	0.05432	0.05432	53	-10649.5	MB2a vs MB1n
					0.1549

Second null model MB2n	0.0537	1	53	-10647.6	MB2a vs MB2n
					0.7218

For Mammalia, the alternative model MB2a, with foreground and background ratios estimated separately, estimates a foreground ratio of 0.0794 and a background ratio of 0.0538 (Tab. 16). The LRT comparing MB2a and MB1n is not significant and indicates that this value is not significantly different from the background ratio (Tab. 16). However, the LRT comparing MB2a and MB2n is statistically significant (Tab. 16). The foreground ratio is significantly different from 1, and since it is close to the background ratio, this is an indication of purifying selection similar to the background branches.

Table 16. Branch model estimates for the branch Mammalia. * indicates statistical significance. np is number of parameters, LnL is log likelihood of the model.

Model	ω₀	ω₁	np	LnL	p-value
Alternative model MB2a	0.0538	0.0794	54	-10649.0

First null model MB1n	0.05432	0.05432	53	-10649.5	MB2a vs MB1n
					0.4965

Second null model MB2n	0.0522	1	53	-10656.9	MB2a vs MB2n
					0.0049*

↓82

In monotremes, the estimated foreground ratio is less than 1, more precisely 0.0209 compared to a background ratio of 0.056 (Tab. 17). Both model comparisons that are different from the background and also different from 1, are found to be statistically significant by the LRTs (Tab. 17). Hence, stronger purifying selection than the background branches was detected in the monotreme branch.

Table 17. Branch model estimates for the branch Monotremata. * indicates statistical significance. np is number of parameters, LnL is log likelihood of the model.

Model	ω₀	ω₁	np	LnL	p-value
Alternative model MB2a	0.056	0.0209	54	-10645.6

First null model MB1n	0.05432	0.05432	53	-10649.5	MB2a vs MB1n
					0.0490*

Second null model MB2n	0.0514	1	53	-10681.5	MB2a vs MB2n
					0.000000002*

In the therian branch, a foreground ratio of 8.7588 was estimated in the null model MB2a (Tab. 18). The LRT comparing MB2a and MB1n indicates that this foreground ratio is significantly different from the background ratio 0.0528 (Tab. 18). But testing whether the elevated ω is significantly different from 1 by comparing MB2a and MB2n, we do not find statistical support by the LRT (Tab. 18). However, since ω is still greater than the background ratio, this indicates relaxed purifying or weak positive selection compared to the background.

↓83

Table 18. Branch model estimates for the branch Theria.

Model	ω₀	ω₁	np	LnL	p-value
Alternative model MB2a	0.0528	8.7588	54	-10644.3

First null model MB1n	0.05432	0.05432	53	-10649.5	MB2a vs MB1n
					0.0224*

Second null model MB2n	0.0529	1	53	-10644.3	MB2a vs MB2n
					0.8332

* indicates statistical significance. np is number of parameters, LnL is log likelihood of the model.

For Marsupialia, a foreground ratio of 0.0186 and a background ratio of 0.00553 was estimated (Tab. 19). The comparison of models MB2a and MB1n using the LRT does not find statistical support, but there is support when comparing MB2a and MB2n (Tab. 19). Since ω₁ is close to the background ω and significantly smaller than 1, this indicates that purifying selection, similar to that of the background, was acting along this branch.

Table 19. Branch model estimates for the branch Marsupialia.

Model	ω₀	ω₁	np	LnL	p-value
Alternative model MB2a	0.0553	0.0186	54	-10647.5

First null model MB1n	0.05432	0.05432	53	-10649.5	MB2a vs MB1n
					0.1548

Second null model MB2n	0.0532	1	53	-10659.5	MB2a vs MB2n
					0.0005*

* indicates statistical significance. np is number of parameters, LnL is log likelihood of the model.

↓84

The last branch of interest is Placentalia. The foreground ratio is 0.0044, compared to a background ratio of 0.0526 (Tab. 20). Both LRTs provide statistical significance, indicating that ω₁ is not only significantly different from ω₀but also from 1. Because ω₁ is approaching 0 this is evidence for purifiying selection (Tab. 20). Since the estimated foreground ratio is also much smaller than the background ratio, this indicates even stronger purifying selection along this branch compared to the background branches (Tab. 20).

Table 20. Branch model estimates for the branch Placentalia.

Model	ω₀	ω₁	np	LnL	p-value
Alternative model MB2a	0.0526	0.0044	54	-10633.0

First null model MB1n	0.05432	0.05432	53	-10649.5	MB2a vs MB1n
					0.00005*

Second null model MB2n	0.0520	1	53	-10692.33	MB2a vs MB2n
					< 0.0000000001*

* indicates statistical significance. np is number of parameters, LnL is log likelihood of the model.

3.4.3. Branch-site models

However, positive selection acts on sites. If there are a lot of sites positively selected along a branch of interest, this signal will be detected by branch models. But if there are only a few sites experiencing positive selection, their signal might be overruled by the other negatively selected sites along that branch. In order to test whether positive selection is acting only on a few sites, branch-site models that detect single positively selected sites, were applied as well.

↓85

In branch-site models, the comparison of alternative model MA and first null model M1a, tests whether there are sites with a ω greater than 1. It is a test for either positive selection or relaxed purifying selection (Yang 2007). Comparing the alternative model MA and the second null model MA1 tests whether sites with an elevated ω ratio are indeed significantly greater than 1. This tests for positive selection only and is called the branch-site test of positive selection (Yang 2007).

For the amniote branch, model MA detects positively selected sites which is indicated by the estimated ω_2a+b value 10.643 (Tab. 21). However, the LRT comparing models MA and M1a does not provide statistical support, neither does the LRT comparing models MA and MA1 (Tab. 21). Thus, there is no indication for positive selection, nor for relaxed purifying selection. However, the BEB analysis did identify six positively selected sites, but all show low posterior probabilities < 95% (Tab. 22).

↓86

Table 21. Branch-site model estimates for the branch Amniota. np is number of parameters, df is degrees of freedom in Likelihood Ratio Test, LnL is log likelihood of the model.

Model	ω	np	df	LnL	p-value
Alternative model MA	ω₀ = 0.04521	56		-10567.6
	ω₁ = 1
	ω_2a = 10.643
	ω_2b = 10.643

First null model M1a	ω₀ = 0.04565	54	2	-10567.7	MA vs M1a
	ω₁ = 1				0.40568

Second null model MA1	ω₀ = 0.0453	55	1	-10569.4	MA vs MA1
	ω₁ = 1				0.67395
	ω_2a = 1
	ω_2b = 1

Table 22. Positively selected sites estimated by BEB analysis in branch-site model MA (Yang et al. 2005), with posterior probabilities, for branches Amniota, Reptilia, Monotremata, Theria, Marsupialia, and Placentalia.

Branch of interest marked as foreground branch	Positively selected site	Posterior probability	Mutation
Amniota	46	0.535	F
	93	0.568	V
	328	0.903	F
	335	0.773	S
	344 (342)	0.899	A
	349 (347)	0.896	S

Reptilia	290	0.611	A
	336	0.900	A

Monotremata	344 (342)	0.992**	Q

Theria	13	0.997**	M
	37	0.967*	Y
	49	0.575	L
	162	0.534	I
	218	0.653	V
	225	0.993**	R
	290	0.921	A
	345 (343)	1.000**	S
	346 (344)	1.000**	S
	348 (346)	0.972**	S

Marsupialia	26	0.987*	Y
	39	0.979*	A

Placentalia	39	0.546	A

Numbers in brackets refer to numbering in bovine rhodopsin. The programm PAML prints out an * if the posterior probability is > 95%, and ** if the probability is > 99% (Yang 2007).

In Reptilia, again, the alternative model MA identifies positively selected sites, which is displayed by the elevated ω of 12.236 in site class ω_2a+b (Tab. 23). But neither comparing models MA and M1a nor models MA and MA1 show statistical support by the LRTs (Tab. 23). So again, there is no evidence for relaxed purifying or positive selection along this branch. Nevertheless, the BEB analysis estimated two positively selected sites, though with low posterior probabilities < 95% (Tab. 22).

↓87

Table 23. Branch-site model estimates for the branch Reptilia. np is number of parameters, df is degrees of freedom in Likelihood Ratio Test, LnL is log likelihood of the model.

Model	ω	np	df	LnL	p-value
Alternative model MA	ω₀ = 0.04525	56		-10568.6
	ω₁ = 1
	ω_2a = 12.236
	ω_2b = 12.236

First null model M1a	ω₀ = 0.04565	54	2	-10567.7	MA vs M1a
	ω₁ = 1				0.68249

Second null model MA1	ω₀ = 0.04527	55	1	-10568.6	MA vs MA1
	ω₁ = 1				0.8792
	ω_2a = 1
	ω_2b = 1

In Mammalia, the estimated ω_2a+b value is 1 (Tab. 24). This indicates that there are no sites under positive selection in the foreground branch. Also, statistical support for testing for positive selection or relaxed purifying selection is not given by the LRTs (Tab. 24). The results suggest that the mammalian branch has a similar selective constraint to the background branch.

Table 24. Branch-site model estimates for the branch Mammalia. np is number of parameters, df is degrees of freedom in Likelihood Ratio Test, LnL is log likelihood of the model.

Model	ω	np	df	LnL	p-value
Alternative model MA	ω₀ = 0.04507	56		-10568.9
	ω₁ = 1
	ω_2a = 1
	ω_2b = 1

First null model M1a	ω₀ = 0.04565	54	2	-10567.7	MA vs M1a
	ω₁ = 1				0.77565

Second null model MA1	ω₀ = 0.04527	55	1	-10568.9	MA vs MA1
	ω₁ = 1				1
	ω_2a = 1
	ω_2b = 1

↓88

Also in the monotreme branch, the MA model identified positively selected sites as indicated by the value 50.166 in site class ω_2a+b (Tab. 25). Neither model comparison provides statistical significance using the LRTs. The monotreme branch has a selective constraint similar to the background branch (Tab. 25). However, the BEB analysis estimated one positively selected site, but since the LRT comparing models MA and MA1 was not significant, this predicted site is not statistically significant either (Tab. 22).

Table 25. Branch-site model estimates for the branch Monotremata. np is number of parameters, df is degrees of freedom in Likelihood Ratio Test, LnL is log likelihood of the model.

Model	ω	np	df	LnL	p-value
Alternative model MA	ω₀ = 0.04537	56		-10566.1
	ω₁ = 1
	ω_2a = 50.166
	ω_2b = 50.166

First null model M1a	ω₀ = 0.04565	54	2	-10567.7	MA vs M1a
	ω₁ = 1				0.19004

Second null model MA1	ω₀ = 0.04541	55	1	-10568.5	MA vs MA1
	ω₁ = 1				0.11969
	ω_2a = 1
	ω_2b = 1

For Theria, the MA model estimates a high ω_2a+b ratio of 999 (Tab. 26). This is a signal for positively selected sites. This time, both the comparison of models MA and M1a as well as the one of models MA and MA1 are statistically significant, which is indicated by the LRTs (Tab. 26). This is a clear signal of positive selection acting on sites along this branch. In a second step, the BEB analysis estimated ten BEB sites in total, but only six have posterior probabilites >95% and are, thus, reliable (Tab. 22).

↓89

Table 26. Branch-site model estimates for the branch Theria.

Model	ω	np	df	LnL	p-value
Alternative model MA	ω₀ = 0.04443	56		-10549.8
	ω₁ = 1
	ω_2a = 999
	ω_2b =999

First null model M1a	ω₀ = 0.04565	54	2	-10569.4	MA vs M1a
	ω₁ = 1				0.000055*
Second null model MA1	ω₀ = 0.04443	55	1	-10559.0	MA vs MA1
	ω₁ = 1				0.00241*
	ω_2a = 1
	ω_2b = 1

* indicates statistical significance. np is number of parameters, df is degrees of freedom in Likelihood Ratio Test, LnL is log likelihood of the model.

A signal of positively selected sites was also detected along the marsupial branch, as indicated by the ω_2a+b value 509.91 in the alternative model MA (Tab. 27). Statistical support is given by the LRT of comparing MA and MA1, which means that the sites which are greater than 1 are significantly greater than 1, an indication for positive selection (Tab. 27). Two predicted BEB sites with confident posterior probabilities < 95% are shown in Table 22.

Table 27. Branch-site model estimates for the branch Marsupialia.

Model	ω	np	df	LnL	p-value
Alternative model MA	ω₀ = 0.04553	56		-10563.3
	ω₁ = 1
	ω_2a = 509.91
	ω_2b = 509.91

First null model M1a	ω₀ = 0.04565	54	2	-10567.7	MA vs M1a
	ω₁ = 1				0.04848*

Second null model MA1	ω₀ = 0.04546	55	1	-10568.0	MA vs MA1
	ω₁ = 1				0.03139*
	ω_2a = 1
	ω_2b = 1

np is number of parameters, df is degrees of freedom in Likelihood Ratio Test, LnL is log likelihood of the model.

↓90

In the placental branch, the ω_2a+b ratio was estimated to equal 1, indicating the presence of no positively selected sites along this branch (Tab. 28). However, neither model comparison is statistically supported by the LRTs (Tab. 28). Thus, no evidence for relaxed purifying or positive selection is found. For placentals, the BEB analysis estimated one positively selected site with a low posterior probability < 95% (Tab. 22).

Table 28. Branch-site model estimates for the branch Placentalia.

Model	ω	np	df	LnL	p-value
Alternative model MA	ω₀ = 0.04565	56		-10569.4
	ω₁ = 1
	ω_2a = 1
	ω_2b = 1

First null model M1a	ω₀ = 0.04565	54	2	-10567.7	MA vs M1a
	ω₁ = 1				1
Model	ω	np	df	LnL	p-value
Second null model MA1	ω₀ = 0.04565	55	1	-10569.4	MA vs MA1
	ω₁ = 1				1
	ω_2a = 1
	ω_2b = 1

np is number of parameters, df is degrees of freedom in Likelihood Ratio Test, LnL is log likelihood of the model.

3.4.4. Summary

In conclusion, the branch-site analyses found evidence for positive selection acting only on the rhodopsin along the branches Theria and Marsupialia (Fig. 22). All other branches experienced slightly relaxed purifying selection (Amniota and Reptilia), purifying selection similar to background branches (Mammalia), or even stronger purifying selection compared to the background branch (Monotremata and Placentalia) (Fig. 22).

↓91

Figure 22. Summary figure showing selective constraints acting on rhodopsin along branches and on sites.

© 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 4.0	Zertifizierter Dokumentenserver der Humboldt-Universität zu Berlin	HTML-Version erstellt am: 28.09.2011

	Figure 15. Secondary structure of the echidna rhodopsin (modified after Sakmar et al. 2002).

	Amino acids differing from bovine rhodopsin are highlighted in green.

	Figure 16. Western blot analysis of expressed rhodopsin pigments.

	(A) From left to right: Bovine, Echidna, mutant T158A, and mutant F169A rhodopsin. (B) From left to right: Bovine, Amniota, Mammalia, and Theria rhodopsin.

	Figure 17. Dark (in red) and light (in black) absorption spectra of expressed and purified rhodopsins, i.e.

	(A) bovine, (B) echidna, (C) mutant T158A, (D) mutant F169A, (E) Amniota, (F) Mammalia, and (G) Theria rhodopsin. λ_max of expressed rhodopsins: Bovine: 500 nm, Echidna: 496.5 nm, mutant T158A: 494.5 nm, mutant F169A: 495.5 nm, Amniota: 500 nm, Mammalia: 501 nm, and Theria: 500.5 nm.

	Figure 18. Acid bleaches of (A) bovine, (B) echidna, (C) mutant T158A, (D) mutant F169A, (E) Amniota, (F) Mammalia, and (G) Theria rhodopsin.

	White circles indicate absorption at 440 nm; black ones indicate absorption at λ_max.

	Figure 19. Hydroxylamine assays performed on (A) bovine, (B) echidna, (C) mutant T158A, (D) mutant F169A, (E) Amniota, (F) Mammalia, and (G) Theria rhodopsins.

	Circles indicate different runs.

	Figure 20. Amino acid alignment of the three inferred ancestral rhodopsins.

	Blue bars indicate residues where Amniota and Mammalia differ from Theria. Pink bars indicate residues where Amniota differs from Mammalia and Theria, and where Bovine is different from Mammalia and Theria. Yellow bars indicate residues where Amniota differs from Mammalia and Theria, and where Bovine shares the same residue with Mammalia and Theria. The red boxes indictate BEB sites inferred by PAML (Tab. 22, chapter 3.4.3.).

	Figure 21. Rhodopsin 3D structure of all pigments from this study.

	(A) shows the echidna rhodopsin with amino acids differing from bovine rhodopsin highlighted in gray. Red marks indicate the substitutions of mutants (B) T158A and (C) F169A. (D-F) Ancestral pigments, i.e. (D) Amniota, (E) Mammalia, and (F) Theria.