Sterol and genomic analyses validate the sponge biomarker hypothesis

Abstract

Molecular fossils (or biomarkers) are key to unraveling the deep history of eukaryotes, especially in the absence of traditional fossils. In this regard, the sterane 24-isopropylcholestane has been proposed as a molecular fossil for sponges, and could represent the oldest evidence for animal life. The sterane is found in rocks ∼650-540 million y old, and its sterol precursor (24-isopropylcholesterol, or 24-ipc) is synthesized today by certain sea sponges. However, 24-ipc is also produced in trace amounts by distantly related pelagophyte algae, whereas only a few close relatives of sponges have been assayed for sterols. In this study, we analyzed the sterol and gene repertoires of four taxa (Salpingoeca rosetta, Capsaspora owczarzaki, Sphaeroforma arctica, and Creolimax fragrantissima), which collectively represent the major living animal outgroups. We discovered that all four taxa lack C30 sterols, including 24-ipc. By building phylogenetic trees for key enzymes in 24-ipc biosynthesis, we identified a candidate gene (carbon-24/28 sterol methyltransferase, or SMT) responsible for 24-ipc production. Our results suggest that pelagophytes and sponges independently evolved C30 sterol biosynthesis through clade-specific SMT duplications. Using a molecular clock approach, we demonstrate that the relevant sponge SMT duplication event overlapped with the appearance of 24-isopropylcholestanes in the Neoproterozoic, but that the algal SMT duplication event occurred later in the Phanerozoic. Subsequently, pelagophyte algae and their relatives are an unlikely alternative to sponges as a source of Neoproterozoic 24-isopropylcholestanes, consistent with growing evidence that sponges evolved long before the Cambrian explosion ∼542 million y ago.

Keywords: Amorphea; Porifera; sponges; steranes; sterols.

Fig. 1.

(A) Structure of 24-ipc, illustrating the canonical carbon numbering system for sterols. The isopropyl group on carbon 24 is highlighted in red. (B) Distribution of sterol synthesis genes across eukaryotes. This phylogeny represents a consensus tree based on previous phylogenetic studies (6, 20, 35). In some cases, multiple genes have been combined into one category based on shared enzymatic function. SMT gene copy numbers and sterol alkylation are provided at the bottom. (C) Sterol side chains from taxa analyzed in this study. See Tables S1 and S2 for relevant data and references.

Fig. S1.

Sterol nuclei and side chain structures identified in S. arctica, C. fragrantissima, C. owczarzaki, and S. rosetta. The key shows compound numbers referenced from Table S1 as well as the associated sterol nuclei and substituent R groups.

Fig. S2.

Results from Osc (OSc/Erg7) protein alignment and tree building. (A) Alignment of OSc/Erg7 peptide 453 and surrounding amino acids; this peptide is considered diagnostic of lanosterol (valine-453) versus cycloartenol (isoleucine-453) synthesis (14). (B) Best maximum-likelihood tree inferred from full OSc/Erg7 alignment (also provided in Dataset S1). Although there is uncertainty about the placement of apusozoan and amoebozoan genes, all other bikonts and amorphean sequences form monophyletic groups. This strongly suggests that lanosterol synthesis in higher plants resulted from convergent evolution following gene duplication. Genes for lanosterol (Las1) and cycloartenol (Cas1) synthesis in the plant A. thaliana are noted with blue and red arrows, respectively, to illustrate this point.

Fig. 2.

Protein modeling of SMT genes. (A and B) Predicted structure of S. cerevisiae Erg6 (A) and A. thaliana SMT2 (B); α-helices are colored red, β-sheets yellow, and predicted methyltransferase sites blue. The seven central β-sheets are numbered in A. (C and D) I. fasciculata’s SMT superimposed on Erg6 (C) and SMT2 (D). The proteins were aligned in the Swiss-PdbViewer, using the isoleucine active site and its two surrounding peptides as a guide. Coloration signifies the root-mean-square distance (RMSD) between the superimposed structures.

Fig. S3.

Maximum-likelihood trees for eukaryote SMTs. Alignments and newick trees are provided in Dataset S1. Nodes in A with bootstrap values > 90 are marked with red circles. Note that in topology A (A), the sponge gene PfiSMT2 clades with an SMT from the apusozoan Thecamonas trahens, whereas in topology B (B), the gene clades with an SMT from the bikont O. tauri. Both scenarios feature long branches and poor bootstrap support values, which suggests that PfiSMT2 suffers from long-branch attraction caused by high sequence divergence and missing data (see the alignment in Dataset S1). We also note that the basally branching amoebozoan SMTs (“Amoebozoa SMT2?” in A) might not represent genuine SMT homologs, as they lack conserved C-terminal sterol methyltransferase domains and have an unusual N-terminal SRPBCC ligand-binding domain (National Center for Biotechnology Information accession no. cl14643).

Fig. 3.

(A) Two competing hypotheses for the origin of the second demosponge SMT. Topology A is the prefered methodology we used to discern orthologs from paralogs, but topology B is also consistent with our data. (B) Molecular clock of SMT divergence times, based on topology A with a long root prior (see Table 1 and Dataset S1 for all time-calibrated trees). Gene duplication events are labeled with stars. The gene duplication events relevant to sponge and pelagophyte C₃₀ sterol synthesis are noted with arrows. The age estimate of Neoproterozoic 24-isopropylcholestane is provided in the red box. White circles indicate calibration points, which are described in Tables S3.

Fig. S4.

A comparison of the RaxML tree shown in Fig. S3B (presented as a cladogram on the right and the species tree error-corrected NOTUNG tree on the left). Yellow circles in the RaxML tree indicate nodes with high (>90%) bootstrap support values. Red squares in the NOTUNG tree indicate putative gene duplication events; gray branches indicate putative gene loss events. Pink lines note taxa that have been rearranged.