Sea Urchin Polyketide Synthase SpPks1 Produces the Naphthalene Precursor to Echinoderm Pigments

Abstract

Nearly every animal species on Earth contains a unique polyketide synthase (PKS) encoded in its genome, yet no animal-clade PKS has been biochemically characterized, and even the chemical products of these ubiquitous enzymes are known in only a few cases. The earliest animal genome-encoded PKS gene to be identified was SpPks1 from sea urchins. Previous genetic knockdown experiments implicated SpPks1 in synthesis of the sea urchin pigment, echinochrome. Here, we express and purify SpPks1, performing biochemical experiments to demonstrate the sea urchin protein is responsible for the synthesis of 2-acetyl-1,3,6,8-tetrahydroxynaphthalene (ATHN). Since ATHN is a plausible precursor of echinochromes, this result defines the biosynthetic pathway to the ubiquitous echinoderm pigments and rewrites the previous hypothesis for echinochrome biosynthesis. Truncation experiments showed that, unlike other type I iterative PKSs so far characterized, SpPks1 produces the naphthalene core using solely ketoacylsynthase (KS), acyltransferase, and acyl carrier protein domains, delineating a unique class of animal nonreducing aromatic PKSs (aPKSs). A series of amino acids in the KS domain define the family and are likely crucial in cyclization activity. Phylogenetic analyses indicate that SpPks1 and its homologs are widespread in echinoderms and their closest relatives, the acorn worms, reinforcing their fundamental importance to echinoderm biology. While the animal microbiome is known to produce aromatic polyketides, this work provides biochemical evidence that animals themselves also harbor ancient, convergent, dedicated pathways to carbocyclic aromatic polyketides. More fundamentally, biochemical analysis of SpPks1 begins to define the vast and unexplored biosynthetic space of the ubiquitous animal PKS family.

Graphical Abstract

Introduction

The discovery of a polyketide synthase (PKS) in sea urchins nearly 20 years ago came as a great surprise.¹ While microbes elaborate many polyketides,^2,3 including useful pharmaceuticals, the animals were thought to be biosynthetically impoverished by comparison. Since that initial discovery, PKSs have been found in many animal genomes, including most species with well-assembled sequence available in GenBank that we have investigated with the exception of the placental mammals. A few exceptions include hybrid polyketide-peptide products from nematodes and polyene pigments from birds, which have been investigated via chemical and genetic methods.^4,5 More recently, an unexpected new group of fatty acid synthase- (FAS)-like PKSs in animals was uncovered, leading to the synthesis of the sacoglossan polypropionates.⁶

However, no protein from the main animal PKSs clade has been biochemically characterized.⁶ This is a major limitation because animal PKSs lead to largely unknown chemistry. The lack of biochemical knowledge concerning animal PKS function or iterative PKS “programming” makes it impossible to predict the products or to engineer animal PKSs. For example, fish such as zebrafish express a PKS that is essential to otolith (ear) formation, and yet the chemical product is unknown, leading to a glaring gap in knowledge in fish biology.⁷ Another example of this limitation can be found in the echinoderms, where PKSs such as SpPk1 are clearly implicated in pigment biosynthesis, yet the actual chemical products and thus the biosynthetic pathways leading to these important and ubiquitous compounds remain unknown.^8,9 The functions of the unique animal PKSs clearly differ from related PKSs of similar catalytic domain organization found in fungi or bacteria, and yet the molecular basis of their activities remains a mystery. Solving this problem would enable us to finally understand the vast and unknown polyketide diversity in the potentially ~10 million animal species on Earth.

Echinoderms are common marine animals with five-fold symmetry, including the sea stars, sea urchins, sea cucumbers, brittle stars, and crinoids. Chemically diverse, aromatic hydrocarbon pigments are found in all five sea urchin groups (Figure 1).¹⁰ Although the biological roles of the pigments are not known with certainty, anthraquinone derivatives have been detected in fossil crinoids dating to >200 million years of age, indicating that they are likely essential to echinoderm biology.¹¹ Various biological activities have been attributed to echinoderm naphthoquinones,^12–14 leading to the use of echinochrome A (1) (Histochrome^™) in Russia as an antioxidant drug.

Figure 1.

Echinoderm compounds and their biosynthesis. (A) Sea urchin polyketides found in S. purpuratus and elsewhere, with pigment colors shown. (B) Examples of other echinoderm polyketide structural families. (C) Domain architecture of SpPks1 in comparison to other iterative PKSs and FAS enzymes.

Echinochromes, spinochromes, and related naphthoquinones are the best studied echinoderm pigments. They are found in sea urchins such as Strongylocentrotus purpuratus, which is deeply red-purple as a result.¹⁵ Feeding studies with radiolabeled precursors indicated that 1 is a polyketide derived from acetate.¹⁶ The S. purpuratus gene SpPks1 was identified as the first animal genome-encoded PKS, and its knockdown led to loss of coloration, strongly implicating the gene in PKS biosynthesis.¹ CRISPR-Cas9 knockouts resulted in albino sea urchins, further validating the early work.^9,17 Knockdown of putative flavin monooxygenases SpFmo also led to albino sea urchin larvae.¹ Specific knockout of SpFmo subtypes led to color changes in the urchins.¹⁸ Moreover, SpPks1 orthologs are found in other echinoderms.¹⁹ Taken together, these data suggested that SpPks1 and its related genes in echinoderms were likely responsible for polyketide pigment production. However, biochemical evidence of function is lacking. The chemical product of SpPks1, and thus the biosynthetic steps in the echinochrome pathway, remain unknown.

Naphthoquinones and other aromatic polyketides are widely represented products of microbial metabolism, yet any biosynthesis through an SpPks1-like enzyme would require completely novel biochemistry. Aromatic carbocyclic polyketides are made by several PKS mechanisms in bacteria, fungi, and plants.^20–25 SpPks1 differs from these classes because it is a canonical type I iterative PKS (consisting of ketosynthase (KS), acyltransferase (AT), dehydratase (DH), C-methyltransferase (cMT), enoylreductase (ER), ketoreductase (KR), and acyl carrier protein (ACP) domains) that lacks a thioesterase (TE) domain, which is typically involved in release of the polyketide chain in tandem with cyclization.^6,26 In contrast to SpPks1, proteins with these domains act to produce reduced polyketides or fatty acids.

Here, we aimed to solve the longstanding mystery of echinochrome biosynthesis by taking a biochemical approach. As a result, we describe a new class of animal carbocyclic aromatic PKSs (aPKSs), opening the field to further studies of the widespread class of animal PKSs.

Results and Discussion

Enzyme expression and optimization.

SpPks1 (GenBank accession: XM_788471) was expressed as a tagged construct in Saccharomyces cerevisiae BJ5464 harboring the npgA phosphopantetheinyl transferase (PPTase) gene (Figure S2).²⁷ Yeast was chosen because it is an excellent tool for the expression of a wide variety of fungal PKSs with similar domain architectures to the urchin PKS (although much different in sequence and, as described below, function).²⁸ Moreover, the same strain was previously successfully used by our group in the first biochemical analysis of any animal PKS, the FAS-like PKS from sacoglossan mollusks.⁶ In turn, while the native yeast PPTase is sometimes able to partially pantetheinylate PKSs, NpgA is well known to effectively convert many PKSs to the active holoprotein.²⁹ In fact, in our hands the limiting factor is often the difficulty of lysing the yeast cell without damaging the large PKS proteins, leading to our application here of a cryo-microfluidizer for efficient lysis. When we expressed and purified SpPks1, and then performed sequencing of the protein, the ACP domain was modified to the holo form, as we did not observe the unmodified fragment in any condition (Figure S3).

Incubation of the purified SpPks1 with malonyl-CoA led to two new compounds observed by UPLC-MS in negative mode at m/z 247.0241 and third compound at m/z 233.0463 (Figures 2 and S4). These compounds were not affected by the presence or absence of cofactors NADPH or SAM in the reaction mixture, potentially indicating that redox and methylating domains were inactive in SpPks1. The products were unstable, making scaleup challenging. These initial data led us to suggest that the three new compounds comprised nonreduced polyketides.

Figure 2.

SpPks1 is an ATHN (6) synthase. UPLC-MS Extracted ion chromatograms are shown. Enzyme assay conditions: (i) Boiled SpPks1; (ii) Active SpPks1; (iii) synthetic standard containing compounds 6-8; (iv) co-elution of [¹³C₃]-malonyl-CoA enzyme assay product and synthetic standard. Colors of pigments are shown. The major enzymatic product of SpPks1 is 6. Compounds 7 and 8 are present at ~5% of the amount of 6, based upon the chromatogram peak height, and represent oxidative degradation products of 6.

SpPks1 was purified using several different methods. Ni-NTA resin provided impure SpPks1 bound to chaperones, which were partly removed using size-exclusion chromatography (SEC) and/or anion exchange chromatography. SEC also confirmed that like other PKSs, SpPks1 elutes as a dimer. Enzyme fractions and purified products were monitored by enzymatic activity and SDS-PAGE. The best method ultimately included Ni-NTA, SEC, and a final step incorporating anti-FLAG, which provided rigorously pure protein with minimal contaminants (Figure S2).

Using the purified enzyme, variables were modified including using different buffers between pH 6.5–9.0, revealing sodium phosphate pH 8.0 as the optimum condition (Figure S11). NaCl and other salts were used, leading to an optimum of 100 mM NaCl. The enzyme was functional at many different temperatures, although it slowed down greatly at 16 °C. Crucially, the enzyme reaction products were not stable at higher temperatures. Thus, we selected a temperature at which the enzyme was at its highest rate without undue degradation of products. The best condition was determined to be 100 mM sodium phosphate, 100 mM NaCl, 1 mM TCEP at 23 °C.

SpPks1 is a 2-acetyl-1,3,6,8-tetrahydroxynaphthalene (ATHN, 6) synthase.

The m/z for the third enzymatic reaction product closely matched the formula C₁₂H₁₀O₅ for ATHN (6), an unstable compound that is readily converted into acetylflaviolin isomers (7) and (8), which in turn have formulas C₁₂H₈O₆, matching the masses of the two major compounds in the mass spectra.²⁶ This behavior matches previous descriptions of 7 and 8 being slightly more stable than 6, although they are also further oxidized into multiple uncharacterized compounds.^22,26 Thus, these data suggested that SpPks1 produces ATHN (6) in vitro. Often, only a single acetylflaviolin has been observed.^26,30 The ratio of 7 to 8 was altered over time due to different rates of spontaneous synthesis and degradation (Figure S10).

Compound 6 and its oxidation products are notoriously difficult to isolate from natural sources or enzymatic reactions, and indeed in previous work using microbially derived 6, a synthetic standard was required.²⁶ Before performing this work, we sought further evidence supporting the proposed molecular formula, performing reactions in the presence of [¹³C₃]-malonyl-CoA and SpPks1. The resulting products exhibited increased masses (m/z = 259.0634 and 245.0839), each differing from the unlabeled products by 12 Da (Figure S5). This confirmed the proposed molecular formula and revealed that all 12 carbons in the enzymatic products originated in 6 units of malonate (Figures 2 and S5).

Thus, these data led to the hypothesis that SpPks1 produces ATHN (6). To provide definitive proof of structures, we generated synthetic standards of 6-8 following a previously described method.²⁶ Synthetic 7-acetyl-6,8-bis-benzyloxynaphthalene-1,3-diol is stable, but when deprotected it produces unstable 6-8 in situ.²⁶ We thus deprotected the diol and used it immediately in UPLC-MS in comparison to SpPks1 reaction products. Further, we validated the identity of product 6 by obtaining ¹H, HSQC, and HMBC NMR spectra for the unpurified reaction product (Table S2 and Figure S28). The synthetic compounds exhibited exactly the same retention times, mass spectra, and MS/MS fragmentation as the enzyme reaction products (Figures 2, S6 and S7). Moreover, the [¹³C₃]-malonyl-CoA reaction product was added to the synthetic standard, which is not enriched in ¹³C. Only single peaks were observed in the UPLC-MS chromatogram for compounds 6-8. However, a double set of peaks were observed in the mass spectra, corresponding to the [U-¹²C] synthetic and [U-¹³C] enzymatic products. Taken together, this evidence definitively demonstrated that SpPks1 produces 6.

Further, we wondered whether 6 was indeed the native reaction product in vivo, or whether it might be an artifact of production in vitro. To test this, we analyzed the organic extract of S. cerevisiae expressing SpPks1 in comparison to control strains. Peaks for 6-8 were readily identified. To confirm the identities of these compounds, S. cerevisiae expressing SpPks1 was further fed with 2-[¹³C]-acetate, leading to the isolation of products greatly enriched in ¹³C as found in the enzymatic labeling reaction. No other new products were observed in S. cerevisiae expressing SpPks1 (Figures 3 and S25).

Figure 3.

ATHN is synthesized in yeast cells. S. cerevisiae expressing SpPks1 was incubated in the presence of: (ii) no extra malonate or acetate; (iii) 10 mM malonate; or (iv) 10 mM malonate and 10 mM 2-[¹³C]-acetate. S. cerevisiae-pxw55 as control shown in (i). A: UPLC-MS Extracted ion chromatograms, B: Mass spectra of extracted ion chromatograms. These results show that S. cerevisiae expressing SpPks1 makes 6 in vivo.

Remarkably, 6 is also the primary product of fungal nonreducing PKS (NRPKS) WdPks1,³⁰ even though NRPKSs have completely different domain architectures and biosynthetic logic in comparison with urchin SpPks1.

Acetyl-CoA is the preferred starter unit.

Initial reactions with SpPks1 revealed that malonyl-CoA was sufficient to synthesize 6, and addition of acetyl-CoA did not improve yield. This led us to propose that malonyl-CoA is the starter unit for SpPks1. To test this hypothesis, we used combinations of ¹³C-labeled acetyl- and unlabeled malonyl-CoA in enzyme reaction mixtures. In the presence of ¹³C-acetyl-CoA and malonyl-CoA, compounds 6-8 were observed at 1 Da greater than when only unlabeled substrates were used, indicating the incorporation of a single acetate group (Figure S16). This indicates a preference for acetyl-CoA as the starter unit.

Malonyl-CoA was also found to initiate synthesis. Even in the presence of ¹³C-acetyl-CoA, a set of small peaks corresponding to unlabeled 6-8 was observed, indicating that some malonyl-CoA was also being used as the starter unit, but that acetyl-CoA was preferred. Purification of SpPks1 using several different FPLC methods, as well as treatment of SpPks1 with dilute detergents, did not eliminate the ability of the enzyme to initiate with malonyl-CoA, showing that it was unlikely to result from an impurity in the enzyme reaction mixture (Figure S17). Moreover, the malonyl-CoA was measured by MS before and after the enzymatic reaction, and also the purchased material was investigated by ¹H NMR (Figure S27). This analysis revealed an insignificantly small amount of acetyl-CoA in malonyl-CoA preparations, which would not explain the incorporation of malonate as a starter unit.

Because of the instability of 6-8, kinetic analysis was performed by labeling the resulting CoA with iodoacetamide, measuring the area under the curve for the resulting CoA derivative using LC-MS in positive mode. Each run was normalized using an internal standard and compared to a standard curve of the desired product. SpPks1 concentrations of 1.9, 3.8 and 7.6 μM (based upon monomer concentration) were used to confirm enzyme linearity in the reaction conditions (Figure S22). The SpPks1 concentration of 3.8 μM was selected for kinetic analysis (Figure S23). SpPks1 displayed k_cat = 0.34 min⁻¹ and K_m=0.94 μM for malonyl-CoA. (Figure 4). This indicates an estimated yield of 28.3 ± 0.6 μM of 6 per hour of reaction time. Kinetic analysis in the presence of 1 mM acetyl-CoA revealed that the enzyme reaction rate was virtually unchanged in the presence of acetyl-CoA (k_cat = 0.34 min⁻¹; K_m=0.89 μM for malonyl-CoA). Thus, starter selection is not the rate-limiting step of the enzyme. While the SpPks1 prefers acetyl-CoA when available, this is not reflected in the overall rate of catalysis to produce 6, and SpPks1 is equally efficient even when acetyl-CoA is not present.

Figure 4.

Michaelis-Menten kinetics for SpPks1. SpPks1 reaction with different concentrations of malonyl-CoA. (A) In presence of 1 mM acetyl-CoA, (B) without acetyl-CoA. The standard deviation is shown for each value.

Since echinochrome has a reduced sidechain in comparison to 6, we anticipated that ethylmalonyl-CoA might be a competent substrate (Figure S18). However, experiments using potential starter units such as ethylmalonyl- and butyryl-CoA, did not afford new products (Figure S20). Collectively, these results suggest that malonyl-CoA and acetyl-CoA are the natural starter units of SpPks1, although acetyl-CoA is preferred (Figures S16 and S19).

SpPks1 cyclization mechanism.

An alignment of SpPks1 and related PKSs revealed that many homologs in echinoderms lack a crucial tyrosine residue that is required for activity of the KR domain (Figures 5 and S25).^31,32 This implied that the KR was inactive, leading us to hypothesize that only the KS, AT, and ACPdomains are necessary to catalyze synthesis of 6. Alternatively, potentially another domain such as the inactive KR might be involved in templating cyclization of the nascent polyketide chain. To test this hypothesis, we constructed a gene encoding the KS-AT didomain, which was expressed in E. coli. A second, ACP-only construct was expressed in E. coli BAP1, which contains a chromosomal copy of the PPTase Sfp.³³ We designed these proteins based upon previous successful approaches used to dissdomains are necessary to catalyze synthesis of 6. Alternatively, potentially another domain such as the inactive KR might be involved in templating cyclization of the nascent polyketide chain. To test this hypothesis, we cons tructed a gene encoding the KS-AT didomain, which was expressed in E. coli. A second, ACP-only construct was expressed in E. coli BAP1, which contains a chromosomal copy of the PPTase Sfp.³³ We designed these proteins based upon previous successful approaches used to dissect fungal type I PKSs, which include the post-AT linkers found in bacterial PKS reconstitution.^25,34 ATHN (6) and acetylflaviolins (7, 8) were synthesized in the presence of both KS-AT and ACP, but not with KS-AT or ACP alone (Figures 5 and S14). This result confirmed that only the three domains are necessary and sufficient for full catalytic activity. In comparison to the wild-type enzyme that was simultaneously run as a control, the excised domains produced only about ~1% the amount of 6, indicating a significant loss of efficiency. A caveat with this experiment is that the truncated protein constructs precipitate when concentrated, so that the efficiency of the process was not optimized.

Figure 5.

Overexpression of minimal PKS. (A) An alignment of KR domains, showing that essential residue Y2233 is absent in SpPks1. (B) SpPks1 protein constructs used in this study. (C) Extracted ion chromatograms (m/z 233.0455) showing production of ATHN (6) by the minimal PKS.

Usually, the KS domain controls the length of a polyketide chain.^35–37 However, in previously characterized type I and type II aromatic PKSs, other domains beyond KS, AT, and ACP are required to promote the fidelity of cyclization.^25,38–40 When those additional domains are absent, polyketides of the correct chain length are synthesized, but they are aberrantly folded. For example, ATHN (6)-producing fungal nonreducing PKSs (NRPKSs) require a product template (PT) domain to catalyze the first ring formation, while the pyrone 9 is produced when the TE domain is excised (Figure 6).^20,25,26,41 This trend is found throughout type I and II PKSs, where folding and controlled cyclization requires additional domains.

Figure 6.

Comparison of aromatic polyketide assembly by various type I PKSs, highlighting unique features of SpPks1.

In experiments with the SpPks1 minimal domains (KS-AT and ACP), the yield of 6-8 decreased in comparison to the wild-type enzyme, but 6-8 remained the only detectable products. To confirm that pyrone 9 was not an additional side-product, we used a synthetic standard of 9 synthesized for a previous project.²⁶ Pyrone 9 was not observed either in the wild-type SpPks1 assay or in the KS-AT/ACP domain experiments (Figures S9 and S14). These results showed that SpPks1 is unique in comparison to known aromatic PKSs from bacteria and fungi. In particular, no other PKS cyclizes the substrate with high fidelity using only KS-AT-ACP, without the need for additional domains. This reactivity also explains why there is no need for a TE in SpPks1, since the last cyclization step would simultaneously offload the product from the enzyme.⁴²

We hypothesized that cyclization is catalyzed by the KS domain, since that is the site where the growing polyketone would be sequestered. However, localizing cyclase activity is a notoriously difficult problem.²⁵ To gain insight into potential molecular factors that might favor cyclization, we modeled the SpPks1 KS structure using trRosetta, templated against 7S6D (TM-score = 0.947).⁴³ We found KS-SpPks1 shared RMSD 0.997 with LovB (7CPX) from fungi, RMSD 0.847 with DEBS5 (2HG4) from actinobacteria, and RMSD 0.749 with CurL (4Mz0) from cyanobacteria, indicating KS-SpPks1 is structurally similar to KS domains from fungal highly reducing PKSs (HRPKSs) (Figure S15). Alignment of KS domains revealed that echinoderm PKS1 enzymes have a conserved set of residues GM(L)MD neighboring the active site cysteine and histidines (Figure S15). These residues are homologous to those in the LovB structure that bind the pantetheine arm of the ACP.⁴⁴ Potentially, this motif, GM(L)MD, might change the position of binding, favoring cyclization. A second conserved motif, AHSS(V), lines the back of the GM(L)MD motif and is not in direct contact with the substrate or pantetheine. In turn, it is possible that both the ACP and KS participate in controlling the high-fidelity cyclization catalyzed by SpPks1.

Biosynthetic hypothesis.

Based upon ¹⁴C isotope feeding experiments, the prevailing biosynthetic proposal for echinochromes and spinochromes is that they are made via condensation of 5 acetate groups into tetrahydroxynaphthalene (THN, 10), which is then oxidized.^15,16 Subsequently, for compounds in which an ethyl or acetyl sidechain is present, this moiety isadded. Here, we revise that proposal and show that the full 6-acetate carbon skeleton is synthesized by the SpPks1 enzyme. Subsequent redox steps, possibly catalyzed in part by SpFmo and/or other enzymes, explain the formation of spinochromes and echinochromes (Figures 7 and S21). The removal of the acetyl side chain and replacement with oxygen might be catalyzed by an oxygenase, as is known for the Baeyer-Villigerases from fungal aromatic polyketide metabolism.⁴⁵ Alternatively, a hydrolytic mechanism is used in fungal naphthalene biosynthesis, although in that case the side chain is replaced by hydrogen, whereas in sea urchin compounds it is replaced by hydroxyl.⁴⁶ The reduced side chain of 1 cannot be fully explained by ourdata, but we considered two major possibilities. First, reduction of the ketone on the acetyl group, followed by a reverse Michael reaction, would transiently yield a quinone methide. Second, previous chemical studies suggest that the reduced side chain may be a result of oxidative dimerization.⁴⁷ The inability of SpPks1 to incorporate ethylmalonyl-CoA or butyryl CoA suggests that the sidechain is probably reduced post-PKS.

Figure 7.

Previous biogenetic hypothesis for echinochrome and spinochromes compared with new findings supported by this study. Blue: previous proposal. Green: Revised proposal defined in this study. Black: Both proposals. Note that this figure only includes two out of the many products isolated from echinoderms that may also play a role as intermediates. A more detailed proposal is presented in Figure S21.

SpPks1 represents an ancient animal-genome encoded pathway that is widely distributed in echinoderms.

The S. purpuratus genome encodes two distinct PKS proteins, SpPks1 and SpPks2, which are about ~30% sequence identical. In contrast to the ATHN synthase SpPks1, SpPks2 (GenBank accession: NP_001239013), consists of domains KS-AT-DH-ER-KR-ACP-TE. SpPks2 is not involved in echinochrome synthesis, and instead it may be important in calcium accumulation⁴⁸ or in the immune response,⁴⁹ although the chemical products of SpPks2 are unknown. An early phylogenetic analysis of animal PKS proteins grouped SpPks1 and SpPks2 together with the other animal PKSs.⁵⁰ Here, we assembled many further echinoderm PKSs from sequence read archive (SRA) data in GenBank and included these as well as many new annotated echinoderm PKSs in a phylogenetic tree (Figure S26). Based upon these new sequences, we found that the echinoderm PKS clade is shared with their nearest neighbors, the acorn worms, at the base of the animal PKS tree. The clade splits into two branches represented by SpPks1 and SpPks2, with representatives found in acorn worms and all echinoderm groups. These results suggest that the common ancestor of modern echinoderms and acorn worms contained both SpPks1 and SpPks2 homologs (at least 500 MYA). The maintenance of these PKS genes since ancient times suggests that the genes are crucial to the echinoderm survival. The SpPks1 phylogenetic tree was congruent with echinoderm phylogeny, indicating vertical inheritance.

Assemblies of 712 echinoderm and 12 acorn worm SRAs led to the identification of 372 SpPks1 homologs (Table S2), 163 SpPks2 homologs, 160 cytoplasmic fatty acid synthase (FAS) homologs, 77 mitochondrial FAS homologs, and 5 other PKS sequences (of which 4 were found in acorn worms). Thus, SpPks1 and SpPks2 homologs appear to be virtually universal in echinoderms and acorn worms.

Animal FAS and PKS sequences are difficult to assemble for technical reasons,⁶ so their absence from a transcriptome is not meaningful, and the >500 SpPks homologs present in existing SRAs underestimates the true number. For example, here using BLAST searching we found SpPks1 homologs in the SRAs from 340 individual samples. However, BLAST did not detect any SpPks1 homologs in SRAs from many samples. We used the read mapping method (see SI Methods) to detect unassembled, individual reads encoding SpPks1 homologs in the SRAs from 292 additional samples. Thus, the number of PKS homologs found is certainly an underestimate.

Out of 81 sequence-distinct SpPks1 homologs that were long enough to include the KR, 71 contained KR-inactivating mutations (Figure S24). Thus, most of these proteins likely synthesize aromatic polyketides, potentially explaining the many diverse aromatic polyketides found throughout Phylum Echinodermata (Figure 8). The remaining 10 proteins may still have nonfunctional KR domains via other mutations, but experimental work is required to determine their products. By contrast, in the SpPks2 clade, the KR domains appear to be active based upon sequence analysis. In addition, 73 unique SpPks1 homologs were long enough to include both the KS GM(L)MD region and to overlap the KR Tyr mutation. Of these, 66 had the inactivating KR mutation.

Figure 8.

Phylogeny of SpPks1 homologs (KS domains) in echinoderms and acorn worms. Nodes were supported by SH-aLRT % and ultrafast bootstrap %. Colors of circles indicate the taxonomic class of the echinoderm. This tree shows that SpPks1 homologs clade according to their phylogenetic origin; the phylogenetic tree parallels the known phylogenetic tree of echinoderms and acorn worms, supporting an ancient origin of SpPks1 prior to the divergence of these groups at least 500 MYA.

Conclusion

In comparison to the large numbers of biosynthetic genes encoded in animal genomes, the resulting enzymes and specialized metabolites are almost completely uncharacterized and unknown. For the most part, the biosynthetic capacity so far characterized in animals comes from the symbiotic microbiome, and not from the animals themselves. Countless new classes of natural product enzymes from animal genomes await characterization and discovery.

Here, we provide the in vitro characterization of the sea urchin aPKS, SpPks1, demonstrating the enzyme’s chemical product and defining the early biosynthetic pathway to echinoderm pigments. aPKSs have not been previously characterized, yet phylogenetic analysis suggests that SpPks1 represents a family of aromatic PKSs present in echinoderms that is likely important to their biology. Therefore, here we define this as the founding member of the aPKS group of Type I PKSs.

Moreover, while several animal PKSs have been identified through genetic methods, this is the second animal PKS to be characterized biochemically (and the first in the main animal PKS clade), out of what is a widespread family that is found throughout the animal kingdom. Thus far, genetic and chemical methods have revealed animal-derived PKSs that produce linear, acetate-derived polyenes that are important as bird pigments,⁵ as well as polyketides that are part of nematode lipopeptides.^4,51 Genetic and biochemical methods have firmly established two further PKS products: carbocyclic aromatic polyketides by the urchin SpPks1 described here, and methylated pyrones synthesized by the sacoglossan mollusk enzyme, EcPKS1.^1,6,9

A current major limitation in the field is that so little is known about the biochemistry of this new class that it is impossible to predict the functions of these novel enzymes that are found in virtually all animals. Thus, it is important to characterize as many representative enzymes as possible. However, expressing animal PKSs has been a major goal of the field since they were identified in 2003, but they have proven very challenging, leading us to develop several methods to circumvent these problems (see SI Methods).

Collectively, this and previous studies of animal biosynthesis reveal the unexpectedly rich secondary metabolism encoded in animal genomes, including many complex motifs more familiarly associated with microbial biosynthesis. As seen in the example of SpPks1, even when the enzymes produce compounds found in microbes, they sometimes do so through unexpected and novel mechanisms. It is time to look beyond microbes to the animals themselves as sources of chemical and biochemical novelty.

ABBREVIATIONS

ACP

acyl carrier protein

AT

acyltransferase

ATHN

2-acetyl-1,3,6,8-tetrahydroxynaphthalene

cMT

C-methyltransferase

DH

dehydratase

ER

enoylreductase

FAS

fatty acid synthase

KR

ketoreductase

KS

ketosynthase

MYA

million years ago

PKS

polyketide synthase

NRPKS

nonreducing PKS

PT

product template domain

SRA

sequence read archive

TE

thioesterase