Phylogenetic analysis shows extensive nAChR duplications in the time range of 1R and 2R
The phylogenetic analysis of the nAChR genes from a broad selection of vertebrate species listed in Methods, rooted with the human 5HTR3A and 5HTR3B receptors, is shown in Fig. 1. Also the GABA-A family has been used as outgroup and gave the same topology (not shown). The tree shows that among the members of the nAChR family, the subfamily comprised of the CHRNA9 and CHRNA10 genes forms the earliest branch, in agreement with previous analyses [9, 10] but in contrast to a recent study [13]. This clade has two amphioxus sequences as closest relatives showing that the duplication of the CHRNA9/CHRNA10 ancestor coincides with the time range of 1R and 2R, as confirmed by our paralogon analysis (see below). The CHRNA9 and CHRNA10 are present in all species included in this analysis, except that CHRNA10 has not been identified in Australian ghostshark. Furthermore, 3R duplicates are retained in all teleost species (Additional file 1). However, the genes in this subfamily have evolved at quite different rates, for instance the CHRNA10 gene in human, mouse and opossum has evolved faster than in the other vertebrates, see Additional file 1. The next branch point leads to the clade forming the CHRNA7/CHRNA8/CHRNA11 gene subfamily, also closely related to a group of invertebrate chordates as well as protostomes, including both fruitfly and C. elegans (Fig. 1). Notably, there seems to have been a local expansion in C. elegans, resulting in nine genes (Additional file 1). The CHRNA7 gene is present in all vertebrates investigated. The zebrafish and medaka genomes have retained a 3R duplicate of CHRNA7. The CHRNA8 gene has not been identified in mammals, but it is present in chicken, lizard, frog, cartilaginous fish and ray-finned fish. A 3R duplicate of the CHRNA8 gene is present in stickleback, medaka and fugu, but not in zebrafish (Additional file 1). Interestingly, during the analysis an additional subfamily member that has not been previously described was identified, here referred to as CHRNA11. It is present in lizard, coelacanth, cartilaginous fish and ray-finned fish and 3R duplicates are present in medaka, stickleback and fugu (Additional file 1). The 3R duplicates of CHRNA11 differ considerable from each other in their evolutionary rates. Also CHRNA11 in Australian ghostshark has diverged considerably.
The ancestor of the remainder of the nAChR family is also closely related to two groups of invertebrate sequences, where there have been extensive local duplications in C. elegans and fruitfly (Additional file 1). The ancestral deuterostome gene duplicated further giving rise to one ancestral gene for the four vertebrate subfamilies consisting of the CHRNB1/CHRNB1.2, CHRND, CHRNE/CHRNG and CHRNB2/CHRNB4/CHRNB5 genes, respectively (Fig. 1). The CHRNB2/CHRNB4/CHRNB5 clade has as sister group an amphioxus clade containing no less than nine genes, indicating an amphioxus-specific expansion. Subsequently, the olfactores clade underwent a duplication leading to the ancestors of the vertebrate clades comprised by the CHRNB1/CHRNB1.2 and CHRND/CHRNE/CHRNG genes on the one hand, and the CHRNB2/CHRNB4/CHRNB5 genes on the other. Each of these two clades has a sister group in tunicates (Fig. 1). Possibly, the duplication took place already in the chordate ancestor, whereupon one of the two clades was lost in amphioxus. Finally, additional duplications took place in the vertebrates. The ancestor of the CHRNB2/CHRNB4/CHRNB5 subfamily was triplicated at a time point that coincides with 1R and 2R, supported by a group of tunicate sequences present basally to the subfamily (Fig. 1), as confirmed by our paralogon analysis (see below). The CHRNB2 gene has not been identified in spotted gar or Australian ghostshark and the gene has not retained any 3R paralogs in teleosts (Additional file 1). The CHRNB4 gene has not been identified in lizard nor coelacanth, but it is present in turtle. This gene has retained no 3R duplicates in teleosts. The CHRNB5 gene has a different phylogenetic distribution, as it is found in Australian ghostshark, spotted gar and teleosts, and in the teleost lineage it has retained 3R duplicates in zebrafish, medaka and fugu.
The CHRNB1 gene encodes a subunit of the NMJ receptor but it has not been found in chicken, lizard, frog or Australian ghostshark. As this subunit is an obligate member of the mammalian NMJ heteropentamer, the genomes of additional species genomes were screened in silico and CHRNB1 was found to be present in other reptiles, such as turtle, python and American alligator. The CHRNB1 gene seems to have no surviving duplicates from tetraploidization events. In spotted gar and teleosts, on the other hand, a local duplication of CHRNB1 in the ancestor of ray-finned fishes gave rise to the CHRNB1.2 gene (Fig. 1 and Additional file 1), located in close proximity to the CHRNB1 gene on the same chromosome (see Fig. 7). Alternatively but less parsimoniously, the duplication could have occurred before the actinopterygian–sarcopterygian split, in which case it would have been followed by a loss of CHRNB1.2 in the sarcopterygian lineage. The common ancestor of the NMJ subunit genes CHRND, CHRNE and CHRNG first underwent a local duplication, giving rise to the CHRND gene and the ancestor of CHRNE and CHRNG, then the 1R and 2R duplications resulted in the CHRNE and CHRNG genes (Fig. 1). The CHRND has not been found in stickleback, but is present in the rest of the species investigated. The CHRNE gene has not been found in chicken, frog or Australian ghostshark but it is present in python, turtle and American alligator. Both the CHRNB1 and CHRNE genes are present in lizard and CHRNE in frog, possibly indicating that the genes exist also in the chicken genome but have not yet been sequenced. The CHRNG was also not identified in the Australian ghostshark, nor in medaka. Close inspection of the phylogenetic tree (see Additional file 1) shows quite varying evolutionary rates among tetrapod CHRNE sequences. Neither of the CHRNB1, CHRNB1.2, CHRND, CHRNG or CHRNE genes have retained additional teleost duplicates. Also, no close relatives of these genes were found in amphioxus but both of the tunicate species have a gene most closely related to the ancestor of these NMJ genes, and it has evolved very rapidly in the tunicate lineage (Fig. 1 and Additional file 1).
An ancestral CHRNA-like gene generated four copies, each one becoming the ancestor of one of the following subfamilies: CHRNA1 (a single member), CHRNA5/CHRNB3, CHRNA3/CHRNA6 and CHRNA2/CHRNA4. The most basal lineage of these four is the one formed by the CHRNA1 gene. Despite its function as a subunit present in the NMJ receptors, our phylogenetic analyses show that the CHRNA1 gene clusters together with the other α-subunit genes rather than the other four NMJ subunit genes (Fig. 1). The CHRNA1 gene is present in all species investigated, and in contrast to the rest of the NMJ genes, it has retained 3R duplicates in stickleback, medaka and fugu. A local duplicate of CHRNA1 is also present in frog, as previously described [19]. Two tunicates are found basal to CHRNA1 and in common with the CHRNB2/CHRNB4/CHRNB5 subfamily, the CHRNA1-like sequences in amphioxus contain a species-specific expansion with four genes present (Fig. 1 and Additional file 1).
The remaining three CHRNA-like genes were all duplicated during what seems to be the time period spanning 1R and 2R. The subfamily containing the CHRNA5 and CHRNB3 genes has one basal tunicate relative. The CHRNA5 gene has not been found in lizard, but it is present in turtle. The CHRNB3 is present in all species included in this analysis. The CHRNB3 gene has retained the 3R duplicate in teleosts, in contrast to CHRNA5 which retained no duplicate in any of the teleosts. The CHRNA3 and CHRNA6 genes form a separate subfamily, with a group of tunicate sequences as closest relatives. The CHRNA3 gene has not been identified in lizard, but is present in turtle. The CHRNA6 gene has not been found in coelacanth. There are 3R duplicates for CHRNA6 in medaka, stickleback and fugu, but none for CHRNA3. In the CHRNA2/CHRNA4 gene subfamily, finally, the CHRNA2 and CHRNA4 genes have not been identified in Australian ghostshark. A CHRNA4 3R duplicate is present in zebrafish. The CHRNA2 gene has retained a duplicate both in stickleback and zebrafish. There are two close relatives in amphioxus to the CHRNA2/CHRNA4 subfamily.
Taken together, the phylogenetic analyses show that the nAChR family can be divided into 10 subfamilies, each of which had one ancestral gene before the origin of the vertebrates and the two vertebrate tetraploidizations. Eight of these 10 ancestral genes seem to have orthologs in either tunicates or amphioxus or both. The ancestor of the clade consisting of CHRNB1/CHRNB.2, CHRND, and CHRNE-CHRNG appears to have triplicated after the tunicate lineage branched off, but before 1R, and subsequently 1R/2R generated the two additional duplicates shown in Fig. 7 (resulting in the CHRNB1/CHRNB1.2 pair and the CHRNE/CHRNG pair). The analyses show that the 10 ancestral subfamily genes expanded in 2R, resulting in the 19 subunit genes present today in vertebrates. Furthermore, the timing of nAChR gene duplications resulting in the additional genes present in the teleosts coincides with the teleost specific tetraploidization 3R, except for CHRNB1.2 that arose as a local duplicate of CHRNB1 basally in the ray-finned fish lineage and the local CHRNA1 duplicate in frog. Information about the nAChR sequences included in the analysis is provided in Additional file 2 and the multiple sequence alignment file is provided in Additional file 3. Additional alignment algorithms were tested for their possible advantage on the dataset as well as for control (these algorithms were: CLUSTAL and PRANK, data not shown), however MUSCLE was found to be most optimal for this dataset.
Exon-intron organization differs among nAChR subunit genes
In addition to the sequence-based phylogenetic analysis, an intron-based phylogenetic analysis was carried out, i.e., a phylogenetic tree was created based on intron insertions into the vertebrate nAChR genes, similar to a previous analysis [9]. Figure 2 shows the exon-intron organization of the 19 vertebrate nAChR genes, with numbered introns, together with important protein features, namely positions of N-linked glycosylation sites, cysteine-pairs and single cysteines. This analysis displays all cysteine sites, although the ones present in the TM regions are not expected to have any particular role. The most parsimonious explanation for order of intron insertions into the nAChR genes is that the first three introns (intron 1–3, Fig. 2) were present in the common ancestor of all nAChR vertebrate genes, since these introns are shared between all genes. After CHRNA9/CHRNA10 had branched off from the ancestor of the rest of the genes, an intron (intron 5, Fig. 2) was inserted dividing exon 3 into two. Subsequently, after the ancestor of the CHRNA7/CHRNA8/CHRNA11 subfamily had branched off, the ancestor of the remainder of the genes received an intron (intron 15) before the region encoding TM4 (Fig. 2). Apart from these six introns (intron 1–5 and 15, Fig. 2) that are shared between several nAChR subfamilies, 14 additional introns have been gained specifically to subsets of subfamilies.
The exon-intron organization analysis divides the nAChR family into four groups. First comes the CHRNA9/CHRNA10 subfamily, which as already mentioned has three introns (intron 1–3, Fig. 2) in common with all nAChR genes. In addition it has a fourth intron (intron 4, Fig. 2) which was gained independently in the ancestors of CHRNA9/CHRNA10 and CHRNA7/CHRNA8/CHRNA11, respectively since it is present in all these genes. Alternatively, this intron was lost in the ancestor of all the other nAChR genes. The CHRNA9/CHRNA10 group of genes contains the lowest number of introns, four in total (intron 1–4, Fig. 2). When comparing N-linked glycosylation sites and cysteines some differences between the CHRNA9 and CHRNA10 genes are observed, for instance there is one glycosylation site located just after the Cys-loop encoded by CHRNA9 but not CHRNA10 gene, whereas the glycosylation site encoded by the second exon is present only in some of the CHRNA9 orthologs.
The second group based on exon-intron organization is formed by the CHRNA7/CHRNA8/CHRNA11 subfamily which just as CHRNA9/CHRNA10 has a quite distinct exon-intron organization. They contain nine introns (intron 1–9 in Fig. 2), four of which are unique to this subfamily (intron 6–9, Fig. 2). This is the only subfamily of genes where there is no glycosylation site present in any of the receptor subtypes close to the Cys-loop (Fig. 2). It seems that these genes (especially the CHRNA7 and CHRNA8) have gained a higher number of cysteines in their last exon, which includes a part of the ICD as well as TM4.
The third group is formed by the NMJ subunit genes. The CHRNB1/CHRNB1.2 genes share an identical exon-intron organization and they have two intron positions in common with CHRND/CHRNE/CHRNG (intron 11 and 13, Fig. 2), presumably inserted in their common ancestral gene. In addition, CHRNB1/CHRNB1.2 share one intron (intron 12, Fig. 2) with CHRNA1. Further, the ancestral CHRNB1/CHRNB1.2 gene received two additional introns (intron 10 and 14, Fig. 2). The positions encoding N-linked glycosylation sites and cysteines in the CHRNB1 and CHRNB1.2 sequences differ between the genes. In the second exon there is a glycosylation site encoded in CHRNB1.2 which is not present in CHRNB1. Also, the CHRNB1.2 gene codes for an extra cysteine. The genes in the CHRND/CHRNE/CHRNG clade have four unique introns (intron 16–19, Fig. 2), which were most likely gained in the CHRND/CHRNE/CHRNG ancestor. They all encode only one glycosylation site which is in close proximity to the Cys-loop. Additional differences are found for cysteine positions (Fig. 2). The remaining NMJ gene, CHRNA1, has a slightly different exon organization, containing fewer introns than the rest of the NMJ genes. As already mentioned, it shares one intron with CHRNB1/CHRNB1.2 (intron 12, Fig. 2) and another one with CHRNB1/CHRNB1.2 and CHRND/CHRNE/CHRNG (intron 13, Fig. 2). Intron 12 and intron 17 (Fig. 2) have the same splice phase although the intron position differs by one codon in CHRND/CHRNE/CHRNG relative to CHRNB1/CHRNB1.2 and CHRNA1. Therefore the sequences of the CHRND/CHRNE/CHRNG genes were analyzed in detail to see if consensus splice donor-acceptor sites are present adjacently in the sequence, which could indicate that intron 12 and 17 are indeed the same intron insertion event that has subsequently undergone a one-codon shift by mutations. However, no obvious such possibility could be found, unless there have later been multiple substitutions that have eradicated any similarity, which would be a less parsimonious explanation. Hence, intron 17 in the CHRND/CHRNE/CHRNG clade could not be concluded to be the same as intron 12 in the CHRNB1/CHRNB1.2 and the CHRNA1 sequences. These similarities and dissimilarities in exon-intron organization results in two possible scenarios. Either the CHRNA1 gene shared a common ancestor with the rest of the NMJ genes and therefore shares with these introns 12 and 13, whereupon it branched off and received intron number 20. Or, the common ancestor of CHRNB1/CHRNB1.2 and CHRNA1 may have received an intron at the same position independently and the CHRND/CHRNE/CHRNG ancestor may have received an intron one codon away, a scenario consistent with the clustering of CHRNA1 with the neuronal α-genes in the sequence-based tree. What further differentiates the CHRNA1 gene from the rest of the NMJ genes is the cysteine pair located in its sixth exon, which is characteristic for the α-subunits.
The fourth “intron-position clade” is the one that contains most of the gene family members, namely the CHRNA2-CHRNA6/CHRNB2–5 genes. They all share an organization that includes a very large exon 5, which distinguishes them from the rest of the nAChR genes. However, the length of the fifth exon as well as some features regarding locations of glycosylation sites, cysteines and the cysteine pair differ between the subunits. For instance, despite its position in the tree the CHRNB3 gene lacks the cysteine-pair, just as the CHRNB2/CHRNB5/CHRNB4 genes (Figs. 1 and 2). Also, not all of the CHRNA5 orthologs contain the cysteine pair.
As described in the methods section the exon-intron comparison is based on the human and spotted gar sequences, and on one occasion the zebrafish (for CHRNA11). However, although the intron positions are quite well conserved in the rest of the species, there are some specific events associated with a few of the genes. For instance, some intron positions differ in opossum (in the CHRNG and CHRNE genes). However, it may be difficult to conclude whether these positions are true or whether they are results of artefacts in the genome assembly. When it comes to the teleosts, many of them have gained extra introns into the region encoding the ICD, the most variable part of the genes. Medaka, stickleback, fugu and zebrafish have all gained introns in the CHRNA5 and CHRNA9 genes and medaka, stickleback and fugu have gained introns in the CHRNA1, CHRNA6, CHRNA10, CHRNB5 and CHRNB3 genes. Further, stickleback has gained introns specifically in CHRNA2, CHRNA4, CHRNB4 and CHRNG and zebrafish in the CHRNA3 gene. The Australian ghostshark CHRNB5 gene has also gained one intron. Finally, some introns seem to have been lost in medaka, stickleback and fugu for the CHRNA8 gene and zebrafish lacks the first intron in the CHRNA10 gene (Data not shown; available from the authors upon request).
Synteny and paralogon analysis confirms expansion of the nAChR family following the vertebrate tetraploidizations
To test the phylogenetic results that indicate the existence of 10 ancestral (in the vertebrate predecessor) nAChR genes that expanded to 19 genes in the first vertebrate ancestor after 1R and 2R, the neighboring chromosomal regions of all nAChR genes were investigated to check for chromosome or block duplications consistent with 1R and 2R. Such related chromosome regions are said to belong to the same paralogon [20], i.e., a set of related chromosomal regions sharing members from the same gene families as a result of chromosome duplication. The 1R and 2R events together resulted in paralogons with four members (double tetraploidization) and the ensuing 3R event in teleosts gave paralogons with up to eight members. Investigation of the chromosomal positions for the nAChR genes showed that the members of each subfamily shared neighboring families, i.e., the neighboring genes also belong to subfamilies that have members on the same chromosomes as the nAChR subfamily, implying that they arose by block (chromosome) duplication. As the phylogenetic analyses (Fig. 1, Additional file 1) showed that the duplications occurred at the origin of the vertebrates, they all probably duplicated as a result of the 1R and 2R which took place in that time.
The CHRNA5/CHRNA3/CHRNB4 gene cluster is present on human chromosome 15, where the CHRNA7 gene is also found. The phylogenetic analyses had indicated that the CHRNB2, CHRNB5 and CHRNB4 genes share a common ancestor, which had triplicated in 1R and 2R, or rather quadrupled after which one of the members was lost. The same scenario was indicated for the common ancestor of the CHRNA7, CHRNA8 and CHRNA11 genes. Following exclusion criteria stated in methods section, 11 gene families found in regions surrounding the nAChR genes just mentioned were analyzed and included, namely: AQP, ANP32E, DENND4, LINGO, APH1, ARHGEF, MTMR, MEGF, SV2, MYO1 and CELF (Fig. 3). Based on the results of the phylogenetic analyses [21], the chromosomal positions for each member of these families were used to deduce a likely evolutionary scheme for human, chicken and spotted gar (Fig. 3). The similarity in the gene repertoires between the four chromosomal regions and species argues strongly that they arose as a result of quadruplication in 1R and 2R. Following a tetraploidization event, some duplicates may be lost, which is in fact a quite common scenario. Losses of genes, or genes that have not yet been identified, are indicated with crosses. Although the crosses are displayed on chromosomes, we cannot know the location of the gene at the time of the loss. However, four of the neighboring families (the ANP32, LINGO, ARHGEF and CELF) have retained all four copies, and they define four member chromosomes of this paralogon in the chicken genome. Although the rest of the neighboring gene families investigated have lost one and sometimes two members, those that are still present are located on the same four chromosomes as the gene families with full quartets, thus supporting the 1R and 2R events generating a paralogon. Three of the paralogon members in human are intact chromosomes and likewise for the spotted gar. As seen in the figure, the second chromosomal line in the human genome actually includes four different chromosomes (chromosome 9, 8, 5, 18), and the spotted gar orthologs are split into two different chromosomes (LG2 and LG4). This can be explained by translocations that have occurred after 1R and 2R, as the corresponding gene family members are located on one chromosome only in chicken (chromosome Z). Taken together, the chromosomal results from these species lead to the conclusion that these gene families all derive from an ancestral chromosome that was quadrupled in 1R and 2R and this paralogon has in a previous study been referred to as paralogon A [15]. In the present study we refer to it as paralogon 1 in relation to the nAChR gene family evolution.
The CHRNA2 and CHRNA4 genes are located in separate regions compared to the other nAChR genes, on human chromosome 8 and 20, respectively. These regions and their neighboring gene families have been analyzed in depth in previous studies from our lab, where analyses showed that these regions belong to a paralogon that resulted from a 2R event [22, 23], namely paralogon B according to the classification provided in [15], which is also displayed in Fig. 4 as paralogon 2 in this study. Two neighboring gene families with complete quartets are included in Fig. 4.
Next the CHRNA9 and CHRNA10 genome regions were analyzed. Following exclusion criteria stated in methods section, nine gene families were analyzed and included, namely FAT, TENM, ATP8A, FRY, PDS5, USP, STARD, MTMR and SLC7A. As in Fig. 3, following the phylogenetic analyses the chromosomal position of each gene family member was noted for human, chicken and spotted gar (Fig. 5). Three of the gene families in this region have retained all four copies in chicken and spotted gar (TENM, STARD and SLC7A) whereas in the human genome the STARD and SLC7A families are triplets. The other neighboring families have lost one or two members in all three species. As not so many gene families with three or four members were found in these genomic regions, also families with two members were included in the analysis. However, despite the smaller data set and some losses, this analysis supports gene family expansions through the 1R and 2R events. As in paralogon 1, there has been a translocation in human where the first member of the paralogon has gene families located on chromosome 4 or 8, and the second paralogon member is comprised by chromosomes 11 and 13, but in chicken and spotted gar each paralogon member is located on a single chromosome. According to the classification provided in [15] this set of chromosomes corresponds to paralogon C, here referred to as paralogon 3 (Fig. 5).
The CHRNA1 gene is a single gene and has no 1R and 2R paralogs remaining today. It is located in close proximity to the HOXD cluster in human, chicken and spotted gar, belonging to paralogon E according to the classification in [15] and here referred to as paralogon 4 (Fig. 6). This paralogon has been analyzed in great detail previously [24,25,26].
The remaining NMJ genes are located on human chromosomes 2 (CHRNB1 and CHRNE) and 17 (CHRND and CHRNG). As already stated, duplicates seem not to have been retained for these genes after 1R and 2R, except for the duplication resulting in the CHRNG and CHRNE genes. Following analyses of the chromosomal regions surrounding these genes, 16 gene families were included namely: ACAP, DLG, RNF, ARRB (for phylogenetic analysis see previous study [27]), PER, HDLBP, KCNAB, TNIK, GNB (for phylogenetic analysis and chromosomal positions see previous study [28]), PIK3C, PLOD, LRCH, PCOLCE, STAG, GIGYF and GPC. After phylogenetic analyses, the chromosomal regions are displayed in Fig. 7. As this is a region that has been subject to many translocations, the analyses become more complicated and in order to gain as much information as possible also families with only two members were included in the analyses. These genomic regions seem to be a mixture of paralogons B and F, according to the classification in Nakatani et al. [15] which further indicates the complexity of the regions, here referred to as paralogon 5 (Fig. 7). However, despite these complications our analyses find nothing that would argue against gene family expansions through the 1R and 2R events. This figure also shows that spotted gar LG2 has a local gene duplication of the CHRNB1 gene resulting in CHRNB1.2 in the ancestor of ray-finned fishes as discussed above.
The nAChR family expansion after the teleost specific tetraploidization
The phylogenetic analyses of the nAChR genes (Fig. 1, Additional file 1) indicate that in total 11 of the 19 nAChR genes in the vertebrate ancestor after 2R have retained duplicates in at least one of the teleosts (CHRNA9, CHRNA10, CHRNA7, CHRNA8, CHRNA11, CHRNB5, CHRNA1, CHRNA2, CHRNA4, CHRNA6 and CHRNB3). From the phylogenetic tree these duplications could be interpreted as having occurred at the time of the teleost tetraploidization 3R. The obvious exception is CHRNB1.2 which is a local duplicate. However, in order to verify that these duplicates are results of 3R a similar analysis as for the two initial vertebrate tetraploidizations was carried out for the paralogon with the largest number of nAChR genes, namely paralogon 1. Two neighboring gene families from the analysis of chromosomal regions in relation to 1R and 2R, with members present in zebrafish, medaka, stickleback and fugu, were chosen for an in depth analysis in teleosts; the CELF and SV2 gene families (Fig. 8). To be noted here is that the CELF4 and the SV2C duplicates are located on separate chromosomes in all teleost species included. This was also noted in the 1R and 2R synteny analysis, where these genes are located on separate chromosomes in human and spotted gar, whereas in chicken they are located in the same chromosomal region (Fig. 3) which shows that despite translocations the chromosomal regions belong to the same paralogon. In order to complement the analysis, the NOCT and PROM gene families were added. As this analysis shows that the duplicates analyzed are located in chromosomal regions harboring the same repertoire of gene families, we conclude that together with the phylogenetic results, the most parsimonious interpretation is that the duplicate nAChR genes in the teleosts are a result of 3R.
These results for the nAChR genes are summarized in Fig. 9. Our analyses show that the nAChR family consists of 10 subfamilies as defined by the ancestral repertoire of genes before 1R. The ancestral genes of these subfamilies were duplicated during the vertebrate tetraploidizations and following losses resulted in 19 ancestral vertebrate genes. Subsequently, mammals lost 3 additional genes resulting in 16 subunit genes present in humans today. Chicken seems to have lost two of these and two more, resulting in 15. The spotted gar has retained 18 of the ancestral 19 (lost only CHRNB2) and has gained one additional copy, the local CHRNB1.2 duplicate (Fig. 8a). Following the teleost tetraploidizations, 20 genes (including the CHRNB1.2 local duplicate) present in the teleost predecessor expanded to 31 nAChR genes present in the teleost ancestor (Fig. 8b). A few losses subsequently resulted in 27 nAChR genes present in zebrafish today, 28 in medaka, 27 in stickleback and 28 in fugu (Additional file 4).