Fig 1.
Summary of the ML trees obtained when analyzing tunicate TNFSF genes.
For each sequence, the species name and accession number are indicated. When several sequences derived from the same contig, an additional number indicates where the TNFSF sequence starts. The topology shown corresponds to the optimal tree after manual correction of the alignment (see main text for details). The box displays the bootstrap values for seven key branches. From left to right, the values in the box correspond to: 1) the original CLUSTALX alignment; 2) the manually-corrected alignment; and, 3) this same corrected alignment once the highly divergent Oikopleura sequence is eliminated. Six highly-supported orthogroups are defined according to the topology of the tree and the patterns of presence/absence of genes in tunicate species. The final, manually-corrected alignment is detailed in the S1 File (see Supporting information).
Fig 2.
Number of genes per orthogroup in the model tunicates analyzed.
The orthogroup to which the highly divergent Oikopleura sequence belongs is unclear. It is here assigned to TNFSF-T4, according to the results shown in Fig 1, but subsequent analyses did not confirm that association (see main text).
Fig 3.
Synteny data for the region containing TNFSF-T1 genes in different tunicate species.
The names and chromosome positions (in parentheses) refer to the most likely human orthologs of the detected tunicate genes. Color codes indicate the degree of proximity respect to TNFSF genes in the human genome. Brown boxes indicate close proximity/adjacency to a TNFSF gene. Dark blue and dark green colors indicate that the corresponding human genes are located in regions that harbor TNFSF genes in our species, but are not adjacent to them. Light blue and light green colors indicate that a gene is found in a chromosome where TNFSF genes are present, but outside the regions that contain TNFSF genes. The blue shades refer to genes on human chromosomes 1, 6, 9 and 19, which contain regions derived from a particular ancestral chromosome. Green boxes indicate genes on human chromosomes 3, 13, 17 and X, which also have regions with a common origin (see details in the main text). N. a.: sequence data not available. Species for which no flanking genes could be identified are not included.
Fig 4.
Synteny data for the regions around TNFSF-T2 and TNFSF-T3, which are tandemly arranged in many species.
Color codes and conventions as in Fig 3.
Fig 5.
Synteny data for the regions which include TNFSF-T4 (top) or TNFSF-T5 (bottom).
Fig 6.
Local synteny data for the regions where tunicate TNFSF-T6 genes are located.
Fig 7.
Optimal ML tree obtained when the tunicate sequences are analyzed together with selected vertebrate orthogroups.
This tree was obtained after manually refining the alignment provided by the NWNSI routine in MAFFT, which was found to be the one that generated the best trees among the 12 alternative alignment programs (see Methods). Using the WAG + F + R6 model and a perturbation strength equal to 0.5, the ML value of the uncorrected tree was -lnL = 76249.295. ML analyses based on the corrected alignment, which had the same number of columns, were similarly performed. Using the same substitution model but with perturbation strength = 0.2, a slightly better tree was found (-lnL = 76210.959), which is the one shown here. The final alignment can be found in the supporting information of the paper (S2 File). Vertebrate genes fall into four distinct groups, each one derived from an ancestral gene, TNFSF-V11 to -V22. Color coding: 1) blue: tunicate genes; 2) black: sequences of gnathostome vertebrates; and, 3) red: cyclostome vertebrate genes.
Fig 8.
Most parsimonious model explaining the distribution of TNFSF genes in living tunicates.
Colors are used to differentiate genes of each of the six orthogroups. Triangles indicate duplications and rectangles, gene losses. The topology of the tree is based on the known relationships among the model species analyzed [51–53]. Three genes are deduced to exist in the ancestor of all tunicates, which all duplicated to become six after the divergence of appendicularians. TNFSF-T4 genes appeared relatively recently, within the stolidobranchian lineage. It was not possible to determine which genes were lost in appendicularians, given that the only sequence found in Oikopleura could not be reliably assigned to any orthogroup. Species names highlighted in yellow indicate colonial tunicates.
Fig 9.
The 4G hypothesis explains all the data available in vertebrates and tunicates.
Top panel: early evolution of the TNFSF genes and origin of the chordate ancestral genes and main tunicate orthogroups. A single gene became duplicated (D) in tandem. Later, a duplication plus transposition of that tandem (D/T) gave rise to the four ancestral chordate genes (TNFSF-C11 to -C22). TNFSF-C11 gave rise to tunicate TNFSF-T1 genes, TNF-C21 to TNFSF-T5 and -T6 and TNFSF-C22 to TNFSF-T2 and TNFSF-T3. One of those two last genes was much later duplicated again, in stolidobranchians, giving rise to TNFSF-T4 (an event not depicted in this figure). TNFSF-C12 was lost in the tunicate lineage. Bottom panel: evolution of the TNF superfamily in vertebrates after their divergence from tunicates. Descendants of the four ancestral chordate genes are found in living vertebrates. The ancestor of all vertebrates suffered a whole genome duplication (WGD1), so eight TNFSF genes were present in its genome. Much later, gnathostomes suffered a second whole-genome duplication (WGD2) with again duplicated the number of TNFSF genes (not shown in this figure). All the details of the subsequent evolution of the TNF superfamily in vertebrates can be found in [21].
Fig 10.
The very different expression patterns of TNFSF-T1a (left) and TNFSF-T1b (right) along Ciona development.
The bottom panel indicates the regions that correspond to each tissue type in this two-dimensional representation [46].
Fig 11.
Overlapping developmental expression patterns of Ciona TNFSF-T2, TNFSF-T3a and TNFSF-T3b genes, which are also similar to those found for TNFSF-T1b (see Fig 10).
Fig 12.
The expression patterns of the related genes TNFSF-T5 and TNFSF-T6 are qualitatively different.
While TNFSF-T5 has again a pattern quite similar to those of TNFSF-T1, -T2 and both -T3 genes (Figs 10, 11), TNFSF-T6 has a singular, very broad expression pattern.
Fig 13.
Expression levels of Ciona TNFSF genes in adult samples (data obtained from [47]).
Top panel: raw expression levels, measured in RPKMs (reads per kilobase of transcript per million mapped reads). Bottom panel: Spearman’s correlation values. In yellow, significant correlations (p < 0.01; FDR-corrected).