Comprehensive identification of SWI/SNF complex subunits underpins deep eukaryotic ancestry and reveals new plant components

Abstract

Over millions of years, eukaryotes evolved from unicellular to multicellular organisms with increasingly complex genomes and sophisticated gene expression networks. Consequently, chromatin regulators evolved to support this increased complexity. The ATP-dependent chromatin remodelers of the SWI/SNF family are multiprotein complexes that modulate nucleosome positioning and appear under different configurations, which perform distinct functions. While the composition, architecture, and activity of these subclasses are well understood in a limited number of fungal and animal model organisms, the lack of comprehensive information in other eukaryotic organisms precludes the identification of a reliable evolutionary model of SWI/SNF complexes. Here, we performed a systematic analysis using 36 species from animal, fungal, and plant lineages to assess the conservation of known SWI/SNF subunits across eukaryotes. We identified evolutionary relationships that allowed us to propose the composition of a hypothetical ancestral SWI/SNF complex in the last eukaryotic common ancestor. This last common ancestor appears to have undergone several rounds of lineage-specific subunit gains and losses, shaping the current conformation of the known subclasses in animals and fungi. In addition, our results unravel a plant SWI/SNF complex, reminiscent of the animal BAF subclass, which incorporates a set of plant-specific subunits of still unknown function.

Fig. 1. Occurrence of SWI/SNF subunits in different eukaryotes.

Blue-filled circles indicate at least a positive hit for the search of a given component (rows) in specific species (columns). Light blue circles indicate the presence of a plausible or distant ortholog hit. Names for each subunit derive from consensus names (see Table 1). The presence of a given homolog was determined as described in Materials and Methods. Ancestral prediction indicates the most likely ancestor that contained a given subunit based on the presence in distinct lineages and species. *protozoan species are entitled together even though they do not form a natural group. The lower cladogram indicates phylogenetic relationships between species. LECA Last Eukaryotic Common Ancestor, LOCA Last Opisthokontan Common Ancestor, LMCA Last Metazoan Common Ancestor, H. sapiens Homo sapiens, M. musculus Mus musculus, D. rerio Danio (Brachydanio) rerio, S. purpuratus Strongylocentrotus purpuratus, D. melanogaster Drosophila melanogaster, C. elegans Caenorhabditis elegans, A. queenslandica Amphimedon queenslandica, S. pombe Schizosaccharomyces pombe, S. cerevisiae Saccharomyces cerevisiae, A. bisporus Agaricus bisporus, U. maydis Ustilago maydis, R. irregularis Rhizophagus irregularis, R. globosum Rhizoclosmatium globosum, D. discoideum Dictyostelium discoideum, T. brucei Trypanosoma brucei, A. thaliana Arabidopsis thaliana, S. lycopersicum Solanum lycopersicum, O. sativa Oryza sativa, A. trichopoda Amborella trichopoda, P taeda Pinus taeda, S. cucullata Salvinia cucullata, S. moellendorffii Selaginella moellendorffii, P. patens Physcomitrella (Physcomitrium) patens, S. fallax Sphagnum fallax, M. polymorpha Marchantia polymorpha, A. agrestis Anthoceros agrestis, M. endlicherianum Mesotaenium endlicherianum, C. braunii Chara braunii, K. nitens Klebsormidium nitens, C. atmophyticus Chlorokybus atmophyticus, M. viride Mesostigma viride, C. reindhardtii Chlamydomonas reindhardtii, C. subellipsoidea Coccomyxa subellipsoidea, O. tauri Ostreococcus tauri, C. paradoxa Cyanophora paradoxa, C. merolae Cyanidioschizon merolae.

Fig. 2. SMARCF subunits are found in all eukaryotes.

a Domain architecture of known SMARCF subunits in H. sapiens, S. cerevisiae, C. subellipsoidea, and A. thaliana. Scale bar, 100 amino acids. b Graphical summary of the evolution of SMARCF domain architectures as predicted from Supplementary Fig. 1. The scale bar indicates the primary sequence length. Phylogram represents the suggested relationship between SMARCF subunits. Arrow indicates the duplication of a single SMARCF into SMARCF1 and SMARCF2 in a fungal and animal common ancestor. Domains are predicted based on Pfam and InterProScan hits, and depicted as colored boxes as indicated in the figure. ARM fold represents a series of ARM-fold hits (IPR016024 and IPR00025) and BAF250_C (PF12031/IPR033388); ARID, AT-rich interaction domain (PF01388/IPR001606); RFX DBD, RFX DNA-binding domain (PF02257/IPR003150); C2H2 Zf, zinc finger, C2H2 type (PF00096). H.s. Homo sapiens, S.c. Saccharomyces cerevisiae, R. irregularis Rhizophagus irregularis, C.s. Coccomyxa subellipsoidea, A. castellanii Acanthamoeba castellanii*, A.t. Arabidopsis thaliana, C. braunii Chara braunii. *A. castellanii (Amoebozoa), is a close relative of D. discoideum with a bona fide SMARCF subunit.

Fig. 3. SMARCG subunits are ancestral PHD-containing proteins.

a Occurrence of SMARCG subunits in different eukaryotes. Blue-filled circles indicate at least a positive hit for the search of a specific SMARCG subtype (rows) in specific species (columns). Each subtype is based on known architectures (PHF10, DPF1-3, and Swp82/Rsc7) or the newly described plant architecture (TPF). Enlarged circles represent subunits with mixed architectures as represented in (c) or lack of duplication into Swp82 and Rsc7 subunits. b T-Coffee derived multiple sequence alignment of PHD-regions of SMARCG proteins from several eukaryotes. The numbers above columns are referred to residue position in AT3G52100 (TPF1). PHDs structure and relevant residues are indicated below the alignment. Tryptophan residues suggesting binding to methylated H3K4 are marked in red. c Summary of the evolution of SMARCG architecture showing the primary structure and domain composition of representative species. Dikaryan SMARCG structure is represented by Saccharomyces cerevisiae Swp82 protein due to common domain architecture in Swp82 and Rsc7 proteins. RiRsc7 and RgRsc7, R. irregularis, and R. globosum SMARCG subunits. Domains are predicted based on Pfam and InterProScan hits, and depicted as colored boxes as indicated in the legend. PHD plant homeodomain (several Pfam hits, clan zf-FYVE-PHD CL0390/IPR019787), SAY supporter of activation of yellow (predicted from bibliography/absent in Pfam/InterProScan), Req requiem/DPF N-terminal domain (PF14051/IPR025750), CRC chromatin remodeling complex Rsc7/Swp82 subunit (PF08624/IPR013933), Tudor-like is derived from several Pfam hits all belonging to the clan Tudor (CL0049), or predicted from multiple sequence alignment. The species list and databases used can be found in Supplementary Table 2. The scale bar indicates the primary sequence length.

Fig. 4. Proposed evolution of the SWI/SNF family in eukaryotes.

Model of the evolutionary history of the SWI/SNF complexes from a predicted ancestral complex (a-BAF) composed of an ATPase core module (subunits A, 2xN, L, and J) and a core scaffold module of subunits 2xC, D, and I. Early branching of the a-BAF into an ancestral non-canonical BAF (a-ncBAF) by the acquisition of the K subunit, and an ancestral canonical BAF (a-cBAF) by the addition of B, F, and G subunits most likely occurred in the LECA prior to the split between Archeaplastids and the lineage that gave rise to the opisthokonts. The presence of the K and I signature subunits suggests that ncBAF is present in Metazoa, Archaeplastida, and Fungi (probably lost in derived lineages as Saccharomycetes where K and I are absent). Incorporation of the E subunit into the a-cBAF gave rise to a predicted opisthokontan-type of cBAF that further diverged into the ancestral opisthokontan BAF and PBAF (a-oBAF and a-oPBAF) by the incorporation of the signature H subunit in the PBAF line, and the loss of the L and I in the a-oPBAF and a-oBAF lines, respectively. The duplication and subfunctionalization of the F subunit into the F1 and F2 types of F subunits (for a-oBAF and a-oPBAF, respectively) is also predicted to have occurred during oBAF-oPBAF emergence in the LOCA, before fungi and animals diverged. The Archaeplastid lineage retained a single cBAF that early lost the subunits I and L, and subsequently lost the ARID domain of the F subunit during Streptophyta emergence, originating the LFR type of the F subunit. The acquisition of the M subunit and the loss of the J subunit represent lineage-specific events occurring after the metazoan and fungal split. A putative split of the ancient H subunit into the fungal Rsc1/2/4 subunits could have occurred after the divergence of the a-oPBAF into RSC complexes. The presence of each predicted type of complex in extant lineages is indicated, with lineage-specific subunit composition specified for S. cerevisiae and A. thaliana, representing SWI/SNF complexes´ flexibility to acquire and lose ancillary subunits. Timeline and indicated events are an estimate. A, SMARCA; B, SMARCB; C, SMARCC; D, SMARCD; E, SMARCE; F, SMARCF (including ARID1, ARID2, and LFR); G, SMARCG (including Swp82/Rsc7, Req/DPFs, and TPFs); H, SMARCH; I, SMARCI; J, SMARCJ; K, SMARCK; L, SMARCL; N, SMARCN (including Actin and Actin-related proteins); 58, Rsc58; 3/30, Rsc3/30; 7, Lbd7; 6, Snf6; 11, Snf11; 102, Rtt102; 1, Htl1; 2, SHH2; P1, PSA1; P2, PSA2; 5, BRD5; O, OPF.