The evolution of the 9aaTAD domain in Sp2 proteins: inactivation with valines and intron reservoirs

Abstract

The universal nine-amino-acid transactivation domains (9aaTADs) have been identified in numerous transcription activators. Here, we identified the conserved 9aaTAD motif in all nine members of the specificity protein (SP) family. Previously, the Sp1 transcription factor has been defined as a glutamine-rich activator. We showed by amino acid substitutions that the glutamine residues are completely dispensable for 9aaTAD function and are not conserved in the SP family. We described the origin and evolutionary history of 9aaTADs. The 9aaTADs of the ancestral Sp2 gene became inactivated in early chordates. We next discovered that an accumulation of valines in 9aaTADs inactivated their transactivation function and enabled their strict conservation during evolution. Subsequently, in chordates, Sp2 has duplicated and created new paralogs, Sp1, Sp3, and Sp4 (the SP1-4 clade). During chordate evolution, the dormancy of the Sp2 activation domain lasted over 100 million years. The dormant but still intact ancestral Sp2 activation domains allowed diversification of the SP1-4 clade into activators and repressors. By valine substitution in the 9aaTADs, Sp1 and Sp3 regained their original activator function found in ancestral lower metazoan sea sponges. Therefore, the vertebrate SP1-4 clade could include both repressors and activators. Furthermore, we identified secondary 9aaTADs in Sp2 introns present from fish to primates, including humans. In the gibbon genome, introns containing 9aaTADs were used as exons, which turned the Sp2 gene into an activator. Similarly, we identified introns containing 9aaTADs used conditionally as exons in the (SP family-unrelated) transcription factor SREBP1, suggesting that the intron-9aaTAD reservoir is a general phenomenon.

Keywords: CBP; E2A; Gal4; Gcn4; KIX; KLF; MED15; MLL; Met4; TAF9; WT1; p53.

Fig. 1

The activation domain subfamilies. a The regions with identified 9aaTADs in Gal4 (in red), Sp1 (in purple), their mutants, and NF1C/CTF (in sand brown) were tested in a reporter assay with a hybrid LexA DNA-binding domain for the capacity to activate transcription. The mutants of Gal4 and Sp1 were designed to substitute aspartates (in red) for glutamines (in purple) and vice versa. The sequences used in hybrid constructs are shown. b The NF1/CTF activator, orthologs and paralogs are shown to demonstrate 9aaTAD motif conservation in their kin (adjacent conserved cysteines are in black). At the bottom of the figure, the entire region of the previously reported proline-rich activation domain of NF1/CTF is shown; the activation domain 9aaTAD is marked (prolines in sand brown). The 9aaTADs are colored for quick orientation. The LexA-Gal4 hybrid constructs assayed in the L40 strain for transactivation activity are shown

Fig. 2

The SP family. a The 9aaTADs identified in the SP family are shown for humans, early deuterostomes/hemichordates/Saccoglossus kowalevskii (acorn worms), protostomes/arthropods/crustacea/Daphnia pulex (water fleas), cnidarians/Acropora digitifera (corals), and poriferans/Oscarella carmela (demosponges). The designation Sp69 reflected uncertainty of origin (a single member of the SP6–9 clade in those animals), whereas the designation Sp59 in Oscarella carmela pointed to a common ancestor of the latter Sp5 and SP6–9 clade. The peptides with the capacity to activate transcription are in red and those without are in black (fivefold induction below the threshold). The activation domains with similarity are referred to by ~ marks. A link of the Sp5 genes and Sp6–9 genes is marked in gray boxes and, for the Sp2 genes, in yellow boxes. The most intriguing variations are in purple. The dash symbol indicates amino acid deletion in human Sp8. The 9aaTADs are colored for quick orientation and sequences used in hybrid constructs are colored and bold emphasized. Valines in the 9aaTAD are in black and underlined. The DNA-binding domain signature (DBD) is shown and was used to distinguish SP clades and their ancestral genes. b The 9aaTADs identified in the SP family are shown for human (h in gray), for chicken (Gallus gallus, g in green; except Sp7, which is lost in the chicken genome and we replaced it by the bird Pseudopodoces humilis, x in purpura, gene Sp7:XP005533328), for lobe-finned bony fish Latimeria chalumnae (coelacanth, l in red), for ray-finned bony fish Danio rerio (zebrafish, d in black) and cartilaginous fish Callorhinchus milii (elephant shark, e in sand brown). Selected activation domains were tested in a reporter assay with a hybrid LexA DNA-binding domain for the capacity to activate transcription. The peptide designations with capacity to activate transcription are with red ID and those without are in black. The most intriguing variations in Sp2 are in purple. The activation domains 9aaTAD are colored for fast orientation. The valines in activation domains are shown in black. Dash symbol means amino acids deletion. Duplications of the Sp2 gene (rending in Sp1, Sp3 and Sp4 paralogs) are in boxes

Fig. 3

SP2 clade evolution. The 9aaTADs identified by similarity (BLAST search in metazoans) in the SP2 clade from early-branched metazoans to humans are shown. Some of them were tested in a reporter assay with a hybrid LexA DNA-binding domain for the capacity to activate transcription. The peptide IDs with the capacity to activate transcription are in red, and those without are in black (fivefold induction above threshold). The most intriguing variations are in purple and black. The paralogs originating from Sp2 gene duplication in elephant sharks are in gray, and the Sp2 paralogs of cartilaginous fish and hemichordates, which show high sequence similarity, are in gray boxes. The 9aaTADs are colored for quick orientation and sequences used in hybrid constructs are colored and bold emphasized. Valines in the 9aaTADs are in black. Valine-rich (V1 in position p1, V2 in position p6, V3 in position p8) activation domains (ADs) with tryptophan (in position p3) are in hemichordates and cartilaginous fish. A link of the first Sp1–4 genes with the Sp2 orthologs and paralogs (in gray) is shown (similarities of hemichordates and cartilaginous fish Sp2 9aaTADs are underlined). Diversified branches are indicated

Fig. 4

Schema of SP1–4 clade evolution. a The evolution of Sp2 could be followed from sea sponges, where we identified the functional 9aaTAD and where the Sp2 protein possesses an activation function. Surprisingly, we found the Sp2 activation domain to be inactivated by valine accumulation in acorn worms, which belong to hemichordates, and we could follow the conservation through all vertebrates, including humans. The Sp2 gene gave rise to the SP2 clade with three other orthologs, Sp1, Sp3 and Sp4. By valine substitution in the activation domains, Sp1 and Sp3 became activators again as ancestral 9aaTADs in sponges. b Phylogenetic trees were generated for the SP1–4 clade, including the hemichordate Saccoglossus kowalevskii (single gene, SkoSp2, from the SP1–4 clade, in magenta), cartilaginous fish Callorhinchus milii (elephant shark; after Sp2 gene duplication, four paralogs CmiSp1–4 from the SP1–4 clade) and Rhincodon typus (whale shark, RtySp2 is the only paralog available from an incomplete sequencing result), resulting in the SP2 clade (in yellow) including SkoSp2, CmiSp2 and RtySp2 and a newly diversified Sp2-paralogs clade; c for the SP family of the hemichordate Saccoglossus kowalevskii (genes SkoSp2, 5 and 9) and poriferan Oscarella carmela (genes OsaSp2 and OsaSp5), resulting in an SP2 clade, SP5/SP59 clade (in black) and SP69 clade (in red); and D) for the SP6–9 subfamily of the hemichordate Saccoglossus kowalevskii (solitary gene, SkoSp69, in red) and cartilaginous fish Callorhinchus milii (after duplication, four paralogs, CmiSp6-9, in orange). The simple phylogeny tool was used to generate phylogenetic trees online at http://www.ebi.ac.uk from ClustalW2 alignments on EMBL-EBI server. The DNA-binding domain signature (DBD) is shown and was used to distinguish SP clades and their ancestral genes

Fig. 5

Reformation of the activation domain in the gibbon Sp2 gene. The Sp2 activation domain encoded by exon 4 conserved in placental mammals (primates, cats, elephants and mice are shown in gray) was lost in hyraxes and gibbons. Their new exons originated from their ancestral introns (in sand brown, exon junction in red, exon 3 in purple, exon 5 in green). This repair took place on the border of exon 3 and exon 4, exactly as in the Takifugu and Tetraodon Sp2b genes and crane Sp2 gene. In the gibbon’s case, the intron provided a new 9aaTAD, and their Sp2 genes were rendered activator function. The missing gibbon ancestral intron was substituted in our figure by the human intron, which is close to that of all primates, and should substitute well for the gibbon ancestral intron. Slashes denote frameshifts in the intron sequence. The 9aaTADs are colored for quick orientation and sequences used in hybrid constructs are colored and bold emphasized. The exon junctions are marked by a red hashtag. *UCSC genomic annotation for the Procavia capensis (rock hyrax) Sp2 gene in UCSC scaffold_25:102095-130179. We did not identify any obvious 9aaTAD motif in the obtained hyrax exon

Fig. 6

Activation domain backup in introns of the unrelated transcription factor SREBP. The exon 1 and exon 2 junction to exon 3 of the SREBP1 gene encodes two distinct 9aaTADs (exon 1 in red, exon 2 in sand brown, exon 3 in purple). The first activation domain, 9aaTAD-I, is encoded by exon 1 (used in isoforms 1, 2 and 5 but not in isoforms 3, 4 and 6), and the second, 9aaTAD-II, is encoded by exon 2 (used in isoforms 3, 4 and 6 but not in isoforms 1, 2 and 5). The alternative splicing always resulted in the presence of a solitary 9aaTAD in all isoforms. The peptides with the capacity to activate transcription are shown in the chart. The 9aaTAD motif conservation in the SREBP family and the similarity with p53 9aaTADs are shown in the lower panel. The 9aaTADs are colored for quick orientation. The exon borders are marked by colored hashtags

Fig. 7

Alignment of activation sequences of S. purpuratus SP and KLF paralogs revealed activation domains in the KLF family. a The Strongylocentrotuspurpuratus (sea urchin) SpuSp69 sequence was aligned to SpuKLF2/3/7/15 sequences (green box for homology region, blue box for 9aaTADs and black box for DNA-binding domain). Human and other metazoan SP proteins have homology regions. The 9aaTADs in the human KLF proteins were identified by sequence similarity to the activation domain in SpuKLFs (KEGG blast with scoring matrix blosum62 and the entire human SP sequences were used initially followed by secondary BLAST search with the entire Spu Sp69 sequence; ClustalW2 on EMBL-EBI server for a set of identified homologous genes was used for alignments). The homology regions (HR) and DNA-binding domain signatures (DBD) are shown and were used to align SP / KLF sequences and to identify the 9aaTADs. b According to sequence similarity, the KLF and WT1 9aaTADs were clustered into Sp1-like and Oaf1-like activation domain subgroups (10.1016/j.ygeno.2007.02.003). *Homo sapiens Hsa, Saccoglossuskowalevskii (acorn worm) Sko, Capitellateleta (annelid worm) Cte, Daphniapulex (water flea) Dpu, Acroporadigitifera (acroporid coral) Adi, Oscarellacarmela (slime sponge) Oca, Crassostreagigas (pacific oyster) Cgi), Latimeriachalumnae (coelacanth) Lch, Branchiostomafloridae (lancelet) Bfo. The valines and twin isoleucines (in HsaKLF6 only) are in black, phenylalanines in positions of valines (in HsaKLF3) are in red and tryptophans are in green. Hidden amino acids are marked with points and numbers. Initial methionines are underlined. The regions with similarity of active and deactivated KLFs are in brown boxes. The 9aaTADs are colored for quick orientation and sequences used in hybrid constructs are colored and bold emphasized. c The KLF genes from Sarcophilusharrisii (Tasmanian devil) Shr, Latimeriachalumnae (coelacanth) Lcm, Callorhinchusmilii (elephant shark) Cmk, Branchiostomafloridae (Florida lancelet) Bfo, Saccoglossuskowalevskii (acorn worm) Sko and Amphimedon queenslandica are listed in Suppl. Figure S9. The designation numbers with brackets reflect uncertainty of origin. Sequences in blue fulfill the 9aaTAD pattern, but sequences in gray and marked with an asterisk do not. The first 9aaTAD position p1 is the most variable and is therefore not shown in the similarity overview. Valines and isoleucines are in black, and tryptophans are in green. *Repressor function for the KLF12 activation domain was only deduced from their reported repressor function and the sequence similarity to KLF3 and KLF6

Fig. 8

Valine accumulation in yeast Met4 activation domain. The accumulation of valines did not inactivate yeast Met4 activation domain. The 9aaTADs are colored for quick orientation and sequences used in hybrid constructs are colored and bold emphasized in the table