bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1 biosynthesis of complex polyketides in a photosynthetic partnership

2 Joshua P. Torres1, Zhenjian Lin1, Jaclyn M. Winter1, Patrick J. Krug2, and Eric W. Schmidt1*

3 1Department of Medicinal Chemistry, University of Utah, Salt Lake City, UT 84112 USA 4 2Department of Biological Sciences, California State University, Los Angeles, CA 90032 USA 5 *Correspondence to: [email protected] 6

7 ABSTRACT

8 are rich sources of complex polyketides, including pharmaceuticals, cosmetics, and

9 other products. Most polyketides are associated with microbial or plant metabolism1. For this

10 reason, symbiotic bacteria or dietary organisms are often the true producers of compounds found

11 in animals2,3. Although increasing evidence suggests that animals themselves make some

12 compounds4-7, the origin of most polyketides in animals remains unknown. This problem makes

13 it difficult to supply useful animal compounds as drugs and severely constrains our

14 understanding of chemical diversity and the scope of biosynthesis in nature. Here, we

15 demonstrate that animals produce microbe-like complex polyketides. We report a previously

16 undocumented but widespread branch of fatty acid synthase- (FAS)-like proteins that have been

17 retooled by evolution to synthesize complex products. One FAS-like protein uses only

18 methylmalonyl-CoA as a substrate, otherwise unknown in animal lipid metabolism, and is

19 involved in an intricate partnership between a and captured . The enzyme’s

20 complex, methylated polyketide product results from a metabolic interplay between algal

21 chloroplasts and animal host cells, and also likely facilitates the survival of both symbiotic

22 partners, acting as a photoprotectant for plastids and an antioxidant for the slug8-12. Thus, we find bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

23 that animals can unexpectedly synthesize a large and medically useful class of structurally

24 complex polyketides previously ascribed solely to microbes, and can use them to promote

25 symbiotic organelle maintenance. Because this represents an otherwise uncharacterized branch

26 of polyketide and fatty acid metabolism, we anticipate a large diversity of animal polyketide

27 products and enzymes awaiting discovery.

28

29 MAIN

30 Complex, branched polyketides, are canonically ascribed to microbial metabolism, yet they are

31 commonly found in animals13,14. The sacoglossan polypropionate pyrones are among the most

32 structurally complex polyketides isolated from animals (Fig. 1). Sacoglossans consume algae,

33 and some maintain the stolen algal chloroplasts (kleptoplasts) for weeks or in exceptional

34 cases, for several months15. Feeding studies showed that fixed carbon obtained from de novo

35 is efficiently incorporated into polypropionates via chloroplasts16. Labeled

36 propionate is also incorporated17-20, suggesting that the mollusk compounds are made via the

37 methylmalonate pathway, which has historically been associated with bacterial metabolism24.

38 The polypropionates react via oxidative and photocyclization pathways that are likely non-

39 enzymatic to form the mature natural products8-10,16. It is thought that these reactions protect the

40 slugs from damage associated with photosynthesis, and that they may thus be necessary for a

41 photosynthetic lifestyle8-12. Thus, the polypropionates are known to originate in the mollusks and

42 to be central to mollusk biology, yet their biochemical sources were unknown.

43 bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

44 We took an unbiased approach to define the molecular source of polypropionates. To test the

45 bacterial origin hypothesis, we sequenced the metagenomes of polypropionate-containing

46 sacoglossans Plakobranchus cf. ocellatus “aff. sp. 1”21 and diomedea. We were unable to

47 identify good candidate polypropionate synthases from the bacterial metagenomes, suggesting a

48 possible origin in the animals or chloroplasts. We took advantage of recently released genomes

49 and transcriptomes of three sacoglossans11,22-24, Elysia chlorotica, E. timida, and E. cornigera, to

50 test this hypothesis. E. chlorotica contains polypropionates ent-9,10-deoxytridachione and

51 elysione25, while E. timida contains ent-9,10-deoxytridachione and several relatives (Fig. 1,

52 Extended Data Table 1)26. Both species obtain fixed carbon from chloroplasts that they maintain

53 for one (E. timida) or several (E. chlorotica) months. In contrast, no polypropionates have been

54 reported from E. cornigera, which only retains chloroplasts for a few days27.

55 Fig. 1: Biogenesis of sacoglossan polypropionates.

56 bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

57 a Previous feeding studies demonstrated that kleptoplasts in the sacoglossans fix carbon that is

58 used to make polypropionates. Propionate is also a precursor, presumably via methylmalonate.

59 A further series of hypothetical steps, including synthesis by a PKS, methylation, and

60 photochemical and oxidation reactions would lead to b the known natural polypropionates of

61 diverse structures. These are found in c sacoglossan species investigated in this work, including

62 E. chlorotica and E. timida, but not E. cornigera.

63

64 Based upon recent reports of animals that make polyene fatty acids5,6, we suspected that the

65 animals might encode modified fatty acid synthase (FAS) or PKS enzymes. Examining the

66 available E. chlorotica transcriptomes, we found a number of fragmented transcripts encoding

67 multiple type I FAS enzymes. Reassembly of those transcripts from raw read data revealed four

68 different FAS-like enzymes (Extended Data Table 2). Two of these were the cytoplasmic and

69 mitochondrial FAS enzymes of primary metabolism. The remaining two (which we named

70 EcPKS1 and EcPKS2) formed their own distinct group in a global PKS/FAS tree that included

71 bacterial, fungal, and animal PKS variants (Fig. 2A, Extended Data Fig. 1). EcPKS1, EcPKS2,

72 and cytoplasmic FAS were encoded in the animal genomes, and not in chloroplasts (Fig. 2C,

73 Extended Data Fig. 2). EcPKS1 homologs were only found in E. chlorotica and E. timida, and

74 not in E. cornigera, while the other genes were found in all three. This observation was

75 consistent with EcPKS1 being the putative polypropionate synthase.

76

77 Like other type I FAS and PKS enzymes, EcPKS1 and EcPKS2 contain multiple domains

78 responsible for product formation (Fig. 2B). Enoylreductase (ER) domains reduce double bonds bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

79 in the nascent polyketide chain. Because EcPKS1 and EcPKS2 lack functional ER domains, theyey

80 likely both synthesize polyenes (Fig. 2D)5,28. The acyltransferase (AT) domain is responsible forr

81 substrate selection, and therefore would be responsible for choosing malonyl-CoA (straight-

82 chain lipids) or methylmalonyl-CoA (methyl-branched lipids)29. In bacteria, amino acid residuess

83 responsible for AT selectivity are well understood30. Since no sequenced animal FASs/PKSs

84 prefer methylmalonyl-CoA, there is no model for AT selectivity in animals. Gratifyingly,

85 EcPKS1, EcPKS2, and several relatives, contained a conserved GHSXGE sequence motif found

86 in bacterial ATs that is known to be important in substrate selection. In bacteria, when X is

87 methionine, the substrate is methylmalonyl-CoA. In EcPKS1, a GHSMGE sequence was

88 observed, whereas that of EcPKS2 was GHSLGE (Fig. 2E). This implied that EcPKS1 might usese

89 methylmalonyl-CoA, while EcPKS2 should use malonyl-CoA. Thus, EcPKS1 had all of the

90 sequence and expression features consistent with the polypropionate synthase.

91 Fig. 2: Sacoglossan PKSs, a new class of FAS-like PKS enyzmes.

92 bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

93 a Sacoglossan PKSs represent a group separate from the known FAS and PKS proteins from

94 bacteria, fungi and animals. Numbers indicate consensus support in percent. Outgroup DpsA is

95 a type II PKS sequence from bacteria, while the remainder are type I enzymes. b The domain

96 architecture of the sacoglossan PKSs, EcPKS1 and EcPKS2 includes ketosynthase (KS),

97 acyltransferase (AT), methyltransferase (MT), dehydratase (DH), enoylreductase (ER),

98 ketoreductase (KR) and acyl carrier protein (ACP). This domain architecture is identical to the

99 sacoglossan FAS, EcFAS except for the absence of a C-terminal thioesterase domain. c EcPKS1 is

100 encoded in the genome of E. chlorotica. d The acyltransferase (AT) domain sequence indicates

101 loading preference for methymalonate while the e enoylreductase (ER) domain lacks key

102 NADPH binding site residues making this function inactive.

103

104 EcPKS1 gene was synthesized and expressed solubly in Saccharomyces cerevisiae BJ5464

105 harboring the npgA phosphopantetheinyl transferase gene (Extended Data Fig. 3)31. Incubation of

106 the purified enzyme with methylmalonyl-CoA and NADPH led to a series of new peaks in the

107 chromatographic trace (Fig. 3A, Extended Data Fig. 4). The major peak had a UV spectrum and

108 mass consistent with a triene pyrone that is three carbons smaller than the major natural products

109 of E. chlorotica (Fig. 3B). A second, minor tetraene pyrone product had mass and UV features

110 consistent with the putative precursor of tridiachiones and related natural products17.

111

112 Like other unmethylated, polyunsaturated pyrones, the enzymatic reaction products were

113 challenging to isolate and characterize because of chemical instability. To gain further evidence

114 for the presence of the pyrone structure, we hydrogenated and derivatized the enzymatic reaction bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

115 mixture, and then analyzed the products by gas chromatography-mass spectrometry (GCMS).

116 This method was chosen because of its predictive value: hydrogenation of the triene should

117 produce four separable stereoisomers, with predictable fragmentation serving to localize methyl

118 groups. Hydrogenation gave four isomers with nearly identical fragmentation patterns around

119 some of the methyl groups, supporting the structural assignment (Extended Data Fig. 5).

120

121 Although some animal FAS/PKS enzymes accept methylmalonyl-CoA in vitro, malonyl-CoA is

122 preferred32-34. We sought to determine whether EcPKS1 specifically incorporates

123 methylmalonate, or whether it also uses malonyl-CoA like other enzymes. We incubated

124 EcPKS1 with malonyl-, methylmalonyl-, acetyl-, and propionyl-CoA under various conditions

125 (Fig. 3D, Extended Data Fig. 6). These experiments demonstrate that EcPKS1 uses only

126 methylmalonyl-CoA, and not other substrates. Further, our data suggest that other CoA esters

127 compete with methylmalonyl-CoA but are not incorporated into any products, indicating

128 stringent selectivity for methylmalonate. S-adenosylmethionine (SAM) is the source of methyl

129 groups in many branched polyketides28. We co-incubated SAM with malonyl-CoA and

130 methylmalonyl-CoA in the presence of EcPKS1, showing that SAM is not involved in the

131 biosynthesis (Extended Data Fig. 6). Thus, EcPKS1 is the first characterized methylmalonate-

132 specific FAS/PKS from animals.

133

134 These data revealed that EcPKS1 is a mollusk polypropionate synthase. In combination with

135 previous experimental evidence, a biosynthetic pathway from CO2 to complex polyketides can

136 be defined. Symbiotic chloroplasts fix CO2, which is then converted into methylmalonyl-CoA.

137 The animal enzyme EcPKS1 synthesizes the tetraene pyrone, which would be converted to bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

138 tridachiones and related natural products by methylation, photorearrangement, and oxidation.

139 These later events may be important in photosynthesis and are strongly supported by previous

140 synthetic chemistry experiments in which the γ-methylated tetraene pyrone is used as starting

141 material to synthesize the natural products (Extended Data Fig. 7)35-37.

142

143 The close connection between chloroplasts and polyene pyrones extends beyond biosynthesis. It

144 has been widely observed that polyene pyrones are found in sacoglossans that engage in long-

145 term photosynthesis, while those lacking polyene pyrones also derive little or no nutrition from

146 photosynthesis (Extended Data Table 1). We found that, in E. diomedea, polyene pyrones are not

147 found in developing embryos, which lack chloroplasts (Extended Data Fig. 8). Moreover, when

148 examining transcriptomes of E. chlorotica at various growth stages, we found that EcPKS1 is

149 highly expressed as the animals metamorphose from the planktonic larval to the benthic juvenile

150 slug stage, just before chloroplasts are acquired (Extended Data Fig. 9). The expression

151 continues well into the stage when the animal maintains stable chloroplasts. In E. timida,

152 expression of the EcPKS1 ortholog EtPKS1 is downregulated after prolonged starvation, when

153 sequestered chloroplasts have degraded.

154

155 To examine the association of EcPKS1 with photosynthetic ability further, we reinvestigated the

156 P. cf. ocellatus and E. diomedea metagenomes with the hypothesis that EcPKS1 homologs would

157 be found in each, as these species are also capable of long-term retention. Indeed,

158 homologs of EcPKS1, EcPKS2, and cytoplasmic FAS were discovered (Extended Data Fig. 10).

159 These were mollusk-encoded, as they came from intron-bearing contigs similar to those

160 encoding their relatives in E. chlorotica. Our data thus strongly support previous suggestions that bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

161 pyrones are critical for maintaining long-term photosynthetic activity in sacoglossans by servingg

162 antioxidant and photoprotective roles.

163 Fig: 3 EcPKS1 is a methylmalonate-specific animal FAS/PKS that synthesizes the tridachione

164 precursor.

165

166 a Incubation of EcPKS1 with methylmalonyl-CoA and NADPH resulted in the synthesis of

167 polyene 1 and tridachione precursor 2. b UV spectra of the synthesized products showing

168 features of polyene propionates consistent with the structural assignment. c Incubation with

169 other acyl-CoA substrates (acetyl-CoA: a-CoA, malonyl-CoA: m-CoA, propionyl-CoA: p-CoA,

170 methylmalonyl-CoA: mM-CoA ) with NADPH did not yield new products. d Using propionyl-CoA bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

171 as the loading molecule inhibited the reaction, as observed in decreased yields. e Using

172 different loading and extender units other than methylmalonate did not produce any products.

173 Graphs shown in a and c-e are HPLC-DAD experiments, where the x-axis gives the elution time

174 in minutes and the y-axis is absorbance units at λ = 325 nm. The plots shown in b are UV

175 spectra with the y-axis showing absorbance and the x-axis showing wavelength in nanometers.

176

177 Polypropionates in marine mollusks are abundant and structurally diverse13. Aside from those

178 described in this study, no other sequenced mollusks are known to contain polypropionates.

179 Examining non-propionate mollusks in GenBank, we found that octopus, sea hares, clams, and

180 oysters contain many FAS-type PKSs that have not been previously noted. However, none of

181 these enzymes falls within the sacoglossan polypropionate group. So far two sequenced

182 transcriptomes encode the methylmalonate-specific motif, from the hemichordate Saccoglossus

183 kowalevskii and the beetle Leptinotarsa decemlineata. These data imply that the ability to

184 synthesize complex polyketides is a ubiquitous feature among mollusks, and it may therefore

185 play an important and unrecognized biological role in the animals.

186

187 Using these newly identified sequences, we generated a taxonomically broader phylogenetic tree

188 based upon the ketosynthase domains of FAS/PKS genes (Extended Data Fig. 10). Many of the

189 mollusk PKSs, as well as several from other phyla, fall in a single large group that includes

190 animal FAS. A second large group contains all of the characterized type I PKS genes from

191 bacteria, animals, and fungi. EcPKS1 is the only characterized PKS within the large FAS-like

192 PKS cluster, representing a new biochemical frontier for PKS research. bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

193

194 A remaining mystery involves the in vitro product selectivity observed in our study. The major

195 enzymatic product incorporated six units of methylmalonate, while the minor product

196 incorporated seven. While many mollusk polypropionates contain five or seven equivalents of

197 methylmalonate, none of the reported compounds contains six. We are therefore still missing a

198 chain length-determining factor. EcPKS1 lacks a thioesterase (TE) domain, which is often

199 important for chain length determination38. We attempted to use base to improve offloading of

200 the product39, but were unsuccessful, indicating that an as-yet unknown factor governs chain

201 length (Extended Data Fig. 6). We speculate that this factor is likely to be the unknown

202 methyltransferase involved in the next step of the biosynthetic pathway.

203

204 In summary, EcPKS1 represents the first example of complex polyketide biosynthesis in animal

205 metabolism. We use biochemical evidence to demonstrate that structurally elaborate polyketides,

206 a family previously associated with microbial metabolism, are animal products as well. The

207 biosynthesis is part of an elaborate molecular dance in which kleptoplasts from algae fix CO2.

208 Fixed carbon is transformed into methylmalonyl-CoA and modified by the mollusk EcPKS1

209 enzyme to synthesize UV- and oxidation-blocking pyrones that protect the mollusk and its

210 chloroplasts from photosynthetic damage.

211

212 Methods

213 Genomic DNA extraction, sequencing and assembly bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

214 Single live specimens of Plakobranchus cf. ocellatus “aff. sp. 1” sensu (collected by SCUBA in

215 the Solomon Islands, June 22, 2006; 9° 1.02’S 160° 13.06’ E) and Elysia diomedea (purchased

216 from www.Liveaquaria.com, Rhinelander, WI, USA) were used in this work. Tissue sections

217 each approximately 0.5 cm3 were frozen using liquid nitrogen and ground using a sterile mortar

218 and pestle. Ground tissue was resuspended in lysis buffer (2 mL; 100 mM NaCl, 10 mM EDTA,

219 1.0 % SDS, 50 mM Tris-HCl at pH 7.5) followed by addition of proteinase K (25 µL; 20

220 mg/mL) and incubated in a water bath at 55 °C until lysis was complete (clear solution). The

221 solution was removed from the water bath and RNAse (2 µL; 20 μg/mL) was added. After 15

222 minutes of incubation at 37 °C, saturated KCl (200 µL) was added to the solution. The mixture

223 was incubated in an ice bath for 10 minutes and then centrifugated at 3,739 x g at 4 °C.

224 Phenol:chloroform extraction of the supernatant40 led to purified DNA. Sequencing was

225 performed at the Huntsman Cancer Institute’s High Throughput Genomics Center, University of

226 Utah. Sacoglossan genomes were sequenced using Illumina HiSeq 2000 sequencer with 350 bp

227 inserts and 125 bp paired-end runs. Raw reads were merged using BBMerge41. Non-merged

228 reads were filtered and trimmed using Sickle with the parameters (pe sanger -q 30 –l 40)42. The

229 trimmed and merged FASTQ files were assembled using metaSPAdes43 with standard

230 parameters in the Center for High Performance Computing at the University of Utah.

231

232 Identification of polypropionate PKS genes in published sacoglossan transcriptomes

233 Sacoglossan transcriptome reads were acquired in previous work by other groups 11,23 from

234 19 specimens of Elysia chlorotica, Elysia timida, and Elysia cornigera (Extended Data Table

235 3). We downloaded the reads from the NCBI Sequence Read Archive (SRA) and assembled

236 them using IDBA-UD44. The assembled transcriptomes were queried by Basic Local bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

237 Alignment Search Tool (BLAST) using the Homo sapiens FASN gene (NCBI Accession:

238 NP_0045095.4) involved in human fatty acid metabolism. Putative full-length, assembled

239 FASN-related transcripts were translated using ExPASy translation tool. FASN and related

240 sacoglossan transcripts consist of multiple domains, which were annotated using

241 Conserved Domain Database (NCBI)45 and Pfam 32.0 (EMBL)46. Geneious 10.2.2 software

242 package was used to perform alignments and tree building. Sequence alignments were

243 made using Multiple Sequence Comparison by Log Expectation47, trimmed and realigned.

244 Phlyogenetic trees were constructed using the Unweighted Pair Group Method with the

245 Jukes-Cantor model48.

246

247 Construction of EcPKS1 expression plasmid

248 The plasmid was constructed by homologous recombination in yeast. The predicted EcPKS1

249 gene was synthesized in three fragments containing 70-bp overlapping regions (Genewiz). The

250 fragments were designed so that their ends overlapped with the promoter and terminator sites of

251 yeast expression plasmid39. The expression vector was linearized using NdeI and PdeII (New

252 England Biolabs) following the manufacturer’s protocol, and the EcPKS1 fragments and

253 linearized vector were cotransformed into Saccharomyces cerevisiae BJ5464 using the S.C.

254 EasyComp Transformation Kit (Sigma). Transformants were plated onto synthetic complete

255 media and selected by transformation to uracil prototrophy. Yeast colonies were combined, and

256 plasmids were isolated using the QIAprep Spin Miniprep Kit (QIAGEN). Plasmids were

257 amplified by transformation into chemically competent Escherichia coli DH5α (New England

258 Biolabs), which was plated onto LB agar (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl, agar

259 20 g/L) with ampicillin (50 μg/mL) and incubated for 24 hours at 37 oC. Individual colonies were bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

260 picked and inoculated in to 5 mL LB broth with ampicillin (50 μg/mL) and incubated at 30 °C

261 with overnight shaking. Plasmids were obtained using the QIAPREP Spin Miniprep Kit

262 (QIAGEN) and screened by digestion with BamHI (New England Biolabs) followed by agarose

263 gel electrophoresis. Plasmid pJTPKS1 containing the expected insert was verified by Sanger

264 sequencing.

265

266 Overexpression and purification of EcPKS1

267 Plasmid pJTPKS1 was transformed into Saccharomyces cerevisiae BJ5464-NpgA (MATα ura3-

268 52 trp1 leu2-Δ1 his3Δ200 pep::HIS3 prb1d1.6R can1 GAL), which was selected on uracil-

269 deficient agar (1.39 g/L Yeast Synthetic Drop-out Media Supplements without uracil (Sigma-

270 Aldrich), 6.7 g/L Yeast Nitrogen Base (Sigma-Aldrich), 40 mL/L 50% glucose solution, agar 20

271 g/L) and incubated at 30 °C for 48 hours. A single colony of BJ5464-NpgA-EcPKS1 was used to

272 inoculate uracil-deficient broth (5 mL; 1.39 g/L Yeast Synthetic Drop-out Media Supplements

273 without uracil (Sigma-Aldrich), 6.7 g/L Yeast Nitrogen Base (Sigma-Aldrich), 40 mL/L 50%

274 glucose solution) and incubated at 30 oC with shaking at 150 rpm. After 24 hours, 1 mL of the

275 culture was used to inoculate 6 x 1 L yeast peptone dextrose broth (10 g/L yeast extract, 20 g/L

276 peptone, 20 g/L glucose). The culture was grown at 30 oC under constant shaking at 180 rpm.

277 After 72 hours, the cells were pelleted at 3,739 x g for 20 minutes at 4 oC. The cell pellets were

278 resuspended in lysis buffer (50 mM NaH2PO4, 150 mM NaCl, 10 mM imidazole, pH 8.0) and

279 sonicated on ice at 1-minute intervals until homogenous. The resulting mixture was centrifugated

280 at 28,928 x g for 40 minutes at 4 oC to separate the supernatant from the cell debris. The

281 supernatant was filtered using a 0.45-micron PVDF syringe filter (Millex-HV, Sigma) before

282 adding Ni-NTA and incubating the mixture at 4 oC for 12 hours. Soluble EcPKS1 in the lysis bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

283 buffer was loaded onto Ni-NTA resin in a gravity column. The resin was washed with 25 mL

284 each of 20 mM and 50 mM imidazole in 50 mM Tric-HCl buffer (500 mM NaCl, pH 8.0). The

285 protein was eluted using 3 x 5 mL 250 mM imidazole in 50 mM Tris-HCl buffer pH 8.0. 15 mL

286 of eluted EcPKS1 was concentrated to 200 μL, buffer exchanged (15 mL of 50 mM Tris-HCl, 2

287 mM EDTA, 5 mM DTT, pH 8.0) and further concentrated to 200 μL final volume using Amicon

288 Ultra 100 MWCO centrifugal filters (EMD Millipore). Protein concentration was calculated to

289 be 6.2 mg/L using the Bradford assay with bovine serum albumin (New England Biolabs) as a

290 standard.

291

292 in vitro characterization of EcPKS1

293 Optimization of the enzymatic reaction was done using 100-µL enzyme reaction mixtures

294 consisting of 10 μM EcPKS1, 2 mM NADPH, 1 mM DTT and 2 mM methylmalonyl-CoA

295 buffered using 100 mM sodium phosphate (over a range of pH 6.5-8.0) and incubated at room

296 temperature under dark conditions. After 16 hours of incubation, reaction products were

297 extracted twice with ethyl acetate (200 μL). The organic layer was dried using a speed vacuum

298 concentrator and reconstituted with methanol and analyzed by HPLC using a Hitachi Primade

299 HPLC system equipped with a PDA detector and autoinjector. Aliquots (20 μL) of the sample

300 were injected onto a reversed-phase C18 analytical column (Luna C18 4.6 x 100mm,

301 Phenomenex) with a gradient starting at 20% mobile phase B (H2O + 0.1 % TFA: MeCN) for 5

302 minutes and increased until 100 % mobile phase B for 20 minutes at a rate of 0.7 mL/min. The

303 optimum working pH for EcPKS1 was determined to be 7.0 (Extended Data Fig. 6). HRMS

304 analysis was performed at the University of Utah’s Mass Spectrometry and Proteomics Core

305 using an Agilent 1290 equipped with an Agilent 1290 FlexCube and Agilent 6350 Accurate bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

306 Mass Q-TOF dual ESI mass spectrometer. Sample (10 μL) was injected onto a UPHLC C18

307 column (2.1 x 150 mm Eclipse Poroshell, 1.9 μ ID) set at 30 °C. Elution was done using solvents

308 containing 10 mM ammonium formate (AF) and 0.1% formic acid (FA). Mobile phase A

309 contained water with AF and FA, while mobile phase B contained 95% acetonitrile (aqueous)

310 with AF and FA. The gradient started at 10% mobile phase B for 1 minute, increasing to 100% B

311 over 10 minutes. The flow rate was 0.5 mL/min. Electrospray ionization was done in positive

312 mode using nitrogen as sheath gas at 350 °C flowing at 12 L/min. Capillary and nozzle voltages

313 were set to 3500 V and 1000 V, respectively.

314

315 EcPKS1 product offloading assay

316 Six 100-µL reaction mixtures containing 2 mM methylmalonyl-CoA, 1 mM DTT, 2 mM

317 NADPH and 10 μM EcPKS1 in 100 mM phosphate buffered at pH 7.0 were incubated in the

318 dark at room temperature for 16 hours. After incubation, 1 M NaOH (20 µL) was added to three

319 of the reaction mixtures, which were then incubated at 65 °C for 10 minutes. Reaction products

320 were extracted into ethyl acetate extraction analyzed by HPLC. For comparison, three control

321 reaction mixtures were made using the same method, but not subjected to NaOH treatment

322 (Extended Data Fig. 6).

323

324 Methytransferase domain activity assay

325 Reaction mixtures (100-µL) containing 2 mM malonyl-CoA, 1 mM DTT, 2 mM NADPH, 5 mM

326 S-adenosyl-L-methionine (SAM), 1mM MgCl2 and 10 μM EcPKS1 in 100 mM phosphate buffer

327 (pH 7.0) were incubated in the dark at room temperature. Two parallel enzymatic reaction bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

328 mixtures containing additional 2 mM methlymalonly-CoA and one without SAM and MgCl2 but

329 with 2 mM methylmalonyl-CoA were carried out for comparison. After 16 hours of incubation,

330 reaction products were extracted with two volumes of ethyl acetate method and analyzed using

331 HPLC (Extended Data Fig. 6).

332

333 Acyl substrate specificity assays

334 Acyl-CoA substrates malonyl-Co A, acetyl-CoA, propionyl-CoA, and methylmalonyl-CoA were

335 purchased from CoALA Biosciences. Enzyme reaction mixtures (100 μL) consisted of enzyme

336 EcPKS1 (10 μM), sodium phosphate pH 7.0 (100 mM), NADPH (2 mM), DTT (1 mM) and

337 acyl-CoA substrates (2 mM). When multiple acyl-CoA substrates were used in a single reaction,

338 they were used at 2 mM each. Reaction mixtures were incubated at room temperature for 16

339 hours and protected from light to avoid photochemical rearrangements (Extended Data Fig. 6).

340

341 Chemical characterization of enzyme reaction products

342 To generate EcPKS1 products for chemical characterization, fifty enzyme reaction mixtures

343 (100-µL) containing 2 mM methylmalonyl-CoA, 1 mM DTT, 2 mM NADPH and 10 µM

344 EcPKS1 in 100 mM phosphate buffer (pH 7.0) were incubated at room temperature for 48 hours

345 with addition of 200 mM NADPH (2 µL) to replenish NADPH after the 24th hour. Enzyme

346 reaction mixtures were pooled and loaded into C18 resin packed in cotton-plugged borosilicate

347 glass Pasteur pipets. The resin was washed twice with water (1 mL) followed by 20%

348 acetonitrile:water (1 mL) before elution with acetonitrile (3 mL). The eluant was dried using a

349 speed vacuum concentrator and reconstituted in methanol 400 µL. Pd/C (0.5 mg) was added to bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

350 the solution and stirred for 16 hours under a H2 atmosphere. The mixture was then filtered

351 through Celite to remove Pd/C and dried a using speed vacuum concentrator. GC-MS/MS

352 analysis was performed at the University of Utah’s Metabolomics Core.

353

354 GC-MS analysis used an Agilent 7200 GC-QTOF and an Agilent 7693A automatic liquid

355 sampler. Dried samples were suspended in dry pyridine (100 µL; EMD Millipore) containing O-

356 methoxylamine hydrochloride (40 mg/mL; MP Bio) and incubated for one hour at 37 °C in a

357 sand bath. An aliquot (50 µL) of the solution was added to autosampler vials. N-Methyl-N-

358 trimethylsilyltrifluoracetamide containing 1% chlorotrimethylsilane (60 µL; Thermo) was added

359 automatically via the autosampler and incubated for 30 minutes at 37 °C. After incubation,

360 samples were vortexed and the prepared sample (1 µL) was injected into the gas chromatograph

361 inlet in the split mode with the inlet temperature held at 250 °C. A 1:1 split ratio was used for

362 analysis. The gas chromatograph had an initial temperature of 60°C for one minute followed by a

363 10 °C/min ramp to 325 °C and a hold time of 10 minutes. A 30-meter Agilent Zorbax DB-5MS

364 with 10 m Duraguard capillary column was employed for chromatographic separation. Helium

365 was used as the carrier gas at a rate of 1 mL/min.

366

+ 367 Polyene (1): UV (MeCN) λmax 279, 318 nm. HRESIMS m/z 289.1814 [M+H] (calcd for

368 C18H24O3, 289.1798). GCMS/MS of 1c: Parent mass m/z 366.2590 (calcd for C21H38O3Si,

369 366.2548), Fragments: m/z 73.0475 (calcd for C3H9OSi, 73.0474), m/z 155.0873 (calcd for

370 C11H23, 155.1800), m/z 211.1141 (calcd for C20H35O3Si, 211.0790), m/z 253.1239 (calcd for

371 C13H21O3Si, 253.1260), m/z 351.2322 (calcd for C20H35O3Si, 351.2355). bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

372

+ 373 Tridachione precursor (2): UV (MeCN) λmax 260, 355 nm. HRESIMS m/z 329.2120 [M+H]

374 (calcd for C21H28O3 , 329.2111).

375

376 Chemical analysis of E. diomedea

377 Chemical extracts were prepared by homogenizing egg ribbons and tissue sections

378 (approximately 10 cm3) of E. diomedea with acetone (5 mL). The homogenate was centrifuged at

379 3,739 x g for 20 minutes at 4 oC and the supernatant recovered by aspiration. The acetone extract

380 was dried using a rotovap and resuspended in methanol. Extracts were analyzed using a HPLC

381 using a Hitachi Primade HPLC system equipped with a PDA detector and autoinjector. Aliquots

382 (20 µL) were loaded onto a C18 analytical column (Luna C18 4.6 x 100mm, Phenomenex) with

383 a gradient starting at 30% mobile phase B (H2O + 0.1 % TFA: MeCN) and 70% mobile phase A

384 (H2O) for 5 minutes and increased until 100 % mobile phase B for 25 minutes at a rate of 1.0

385 mL/min. Purification of the major polypropionate in E. diomedea tissue extract was performed

386 using a semi-preparative HPLC column (Luna 5u C18 250 x 10 mm) at a flowrate of 2.5

387 mL/min. 1H NMR data were obtained using a Varian 500 (1H 500 MHz) NMR spectrometer

388 equipped with a 3mm Nalorac MDBG probe, utilizing residual CDCl3 signal for referencing.

389 Natural products, including tridachione, were identified by comparing data to those reported in

390 the literature. A similar procedure was used to isolate and identify products from P. cf. ocellatus.

391

392 Discovery of EcFAS, EcPKS1, and EcPKS2 analogs in P. cf. ocellatus and E. diomedea bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

393 Gene prediction was done using AUGUSTUS49. Full-length EcPKS1, EcPKS2, and EcFAS genes

394 were used as queries to search for PKS gene-containing contigs by tblastn search (E-value = 1.0

395 x e-10) in P. cf. ocellatus and E. diomedea genomic assembly. Contig hits were then submitted to

396 AUGUSTUS to predict genes based on parameters built using the reported E. chlorotica genome

397 assembly and protein sequences50. The resulting predicted gene fragments were queried against

398 EcPKS1, EcPKS2 and EcFAS using blastp search (E-value = 1.0 x e-10) and were ranked

399 according to bitscore and E-value to determine the best hit. The best hits for EcPKS1, EcPKS2

400 and EcFAS were assigned to each query gene then aligned and assembled to create the predicted

401 PKS and FAS orthologs in P. cf. ocellatus aff. sp. 1 and E. diomedea (PoPKS1, PoPKS2, PoFAS

402 for P. cf. ocellatus aff. sp. 1 and EdPKS1, EdPKS2 and EdFAS for E. diomedea).

403

404 EcPKS1 gene expression analysis

405 Raw reads were trimmed by Sickle with parameters (pe sanger –q 30 –l 40). Trimmed reads were

406 pooled by species, and de novo assembly for each species was done (EC and ET) using

407 maSPAdes with default parameters. Trimmed raw reads of each specimen were multimapped to

408 the reference transcriptome assembly using Salmon51 with parameters (salmon index –t –i index

409 –k 31; salmon quant --index –valudateMappings –libType A –dumpEq –r). The multitude of

410 contigs produced by assembly were hierarchically clustered, and cluster count summarization

411 was performed using Corset 1.0452 according to shared read information with parameters (-f true

412 –g –n – salmon+eg_classes). Transcript abundance of clusters (EcPKS1 gene) was estimated

413 using EdgeR53 by normalized counts per million.

414 bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

415 Data and materials availability

416 Plasmids are available from the corresponding author with a standard MTA. All sequences

417 described in this study have been deposited in GenBank (See Extended Data Table 3 for

418 accession numbers).

419

420 References

421 1 Newman, D. J. & Cragg, G. M. Natural products as sources of new drugs from 1981 to

422 2014. J. Nat. Prod. 79, 629-661 (2016).

423 2 Morita, M. & Schmidt, E. W. Parallel lives of symbionts and hosts: chemical

424 mutualism in marine animals. Nat. Prod. Rep. 35, 357-378 (2018).

425 3 Zan, J. A.-O. et al. A microbial factory for defensive kahalalides in a tripartite marine

426 symbiosis. Science epub (2019).

427 4 Castoe, T. A., Stephens, T., Noonan, B. P. & Calestani, C. A novel group of type I

428 polyketide synthases (PKS) in animals and the complex phylogenomics of PKSs.

429 Gene 392, 47-58 (2007).

430 5 Cooke, T. F. et al. Genetic mapping and biochemical basis of yellow feather

431 pigmentation in budgerigars. Cell 171, 427-439 (2017).

432 6 Sabatini, M. et al. Biochemical characterization of the minimal domains of an

433 iterative eukaryotic polyketide synthase. FEBS J. 285, 4494-4511 (2018).

434 7 Torres, J. P. & Schmidt, E. W. The biosynthetic diversity of the animal world. J. Biol.

435 Chem. In press (2019).

436 8 Zuidema, D. R. & Jones, P. B. Triplet photosensitization in cyercene A and related

437 pyrones. J. Photochem. Photobiol. B 83, 137-145 (2006). bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

438 9 Zuidema, D. R., Miller, A.K., Trauner, D. & Jones, P. B. Photosensitized conversion of

439 9,10-deoxytridachione to photodeoxytridachione. Org. Lett. 7, 4959-4962 (2005).

440 10 Zuidema, D. R. & Jones, P. B. Photochemical relationships in sacoglossan

441 polypropionates J. Nat. Prod. 68, 481-486 (2005).

442 11 de Vries, J. et al. Comparison of sister species identifies factors underpinning plastid

443 compatibility in green sea slugs. Proc. Royal Soc. B 282, 20142519 (2015).

444 12 Powell, K. J. et al. Photochemical activity of membrane-localized polyketide derived

445 marine natural products. Tetrahedron 74, 1191-1198 (2018).

446 13 Davies-Coleman, M. T. & Garson, M. J. Marine polypropionates. Nat. Prod. Rep. 15,

447 477-493 (1998).

448 14 Morgan, E. D. Biosynthesis in Insects. (The Royal Society of Chemistry, 2004).

449 15 Trench, R. K. Chloroplasts as functional endosymbionts in the mollusc Tridachia

450 crispata (Bërgh), (Opistobranchia, ). Nature 222, 1071-1072 (1969).

451 16 Ireland, C. & Scheuer, P. J. Photosynthetic marine mollusks: In vivo 14C incorporation

452 into metabolites of the sacoglossan Placobranchus ocellatus. Science 205, 922-923

453 (1979).

454 17 Cutignano, A., Cimino, G., Villani, G. & Fontana, A. Shaping the polypropionate

455 biosynthesis in the solar-powered mollusc Elysia viridis. ChemBioChem 10, 315-322

456 (2009).

457 18 Di Marzo, V. et al. Cyercenes, novel pyrones from the ascoglossan mollusc Cyerce

458 cristallina. Tissue distribution, biosynthesis and possible involvement in defense

459 and regenerative processes. Experientia 47, 1221-1227 (1991). bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

460 19 Manker, D. C., Garson, M. J. & Faulkner, D. J. De novo biosynthesis of polypriopionate

461 metabolites in the marine pulmonate Siphonaria denticulata. Chem. Commun. 1061-

462 1062 (1988).

463 20 Cutignano, A., Villani, G. & Fontana, A. One metabolite, two pathways: convergence

464 of polypropionate biosynthesis in fungi and marine molluscs. Org. Lett. 14, 992-995,

465 (2012).

466 21 Krug, P. J. et al. Integrative species delimitation in photosynthetic sea slugs reveals

467 twenty candidate species in three nominal taxa studied for drug discovery, plastid

468 symbiosis or biological control. Mol. Phylogenet. Evol. 69, 1101-1119 (2013).

469 22 Rauch, C., Christa, G., de Vries, J., Woehle, C. & Gould, S. B. Mitochondrial genome

470 assemblies of Elysia timida and Elysia cornigera and the response of mitochondrion-

471 associated metabolism during starvation. Genome Biol. Evol. 9, 1873-1879, (2017).

472 23 Cai, H. et al. A draft genome assembly of the solar-powered sea slug Elysia chlorotica.

473 Sci. Data 6, 190022 (2019).

474 24 Chan, C. X. et al. Active host response to algal symbionts in the sea slug Elysia

475 chlorotica. Mol. Biol. Evol. 35, 1706-1711 (2018).

476 25 Dawe, R. D. & Wright, J. L. C. The major polypropionate metabolites from the

477 sacoglossan mollusc, Elysia chlorotica. Tetrahedron Lett. 27, 2559-2562 (1986).

478 26 Gavagnin, M., Spinella, A., Castelluccio, F. & Cimino, G. Polypropionates from the

479 Mediterranean mollusk Elysia timida. J. Nat. Prod. 57, 298-304 (1994).

480 27 Rumpho, M. E., Pelletreau, K. N., Moustafa, A. & Bhattacharya, D. The making of a

481 photosynthetic animal. J. Exp. Biol. 214, 303, (2011). bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

482 28 Kennedy, J. et al. Modulation of polyketide synthase activity by accessory proteins

483 during lovastatin biosynthesis. Science 284, 1368-1372 (1999).

484 29 Keatinge-Clay, A. T. The uncommon enzymology of cis-acyltransferase assembly

485 lines. Chem. Rev. 117, 5334-5366 (2017).

486 30 Haydock, S. F. et al. Divergent sequence motifs correlated with the substrate

487 specificity of (methyl)malonyl-CoA:acyl carrier protein transacylase domains in

488 modular polyketide synthases. FEBS Lett. 374, 246-248 (1995).

489 31 Lee, K. K., Da Silva, N. A. & Kealey, J. T. Determination of the extent of

490 phosphopantetheinylation of polyketide synthases expressed in Escherichia coli and

491 Saccharomyces cerevisiae Anal. Biochem. 394, 75-80 (2009).

492 32 Juarez, P., Chase, J. & Blomquist, G. J. A microsomal fatty acid synthetase from the

493 integument of Blattella germanica synthesizes methyl-branched fatty acids,

494 precursors to hydrocarbon and contact sex pheromone. Arch. Biochem. Biophys. 293,

495 333-341 (1992).

496 33 Gu, P., Welch, W. H. & Blomquist, G. J. Methyl-branched fatty acid biosynthesis in the

497 German cockroach, Blatella germanica: Kinetic studies comparing a microsomal and

498 soluble fatty acid synthetase. Insect Biochem. Mol. Biol. 23, 263-271 (1993).

499 34 Gu, P., Welch, W. H., Guo, L., Schegg, K. M. & Blomquist, G. J. Characterization of a

500 novel microsomal fatty acid synthetase (FAS) compared to a cytosolic FAS in the

501 housefly, Musca domestica. Comp. Biochem. Physiol. B 118, 447-456 (1997).

502 35 Eade, S. J. et al. Biomimetic synthesis of pyrone-derived natural products: Exploring

503 chemical pathways from a unique polyketide precursor. J. Org. Chem. 73, 4830-4839

504 (2008). bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

505 36 Moses, J. E., Adlington, R. M., Rodriguez, R., Eade, S. J. & Baldwin, J. E. Biomimetic

506 synthesis of (±)-9,10-deoxytridachione. Chem. Common. 1687-1689 (2005).

507 37 Sharma, P., Lygo, B., Lewis, W. & Moses, J. E. Biomimetic synthesis and structural

508 reassignment of the tridachiahydropyrones. J. Am. Chem. Soc. 131, 5966-5972

509 (2009).

510 38 Gokhale, R. S., Hunziker, D., Cane, D. E. & Khosla, C. Mechanism and specificity of the

511 terminal thioesterase domain from the erythromycin polyketide synthase. Chem.

512 Biol. 6, 117-125 (1999).

513 39 Ma, S. M. et al. Complete reconstitution of a highly reducing iterative polyketide

514 synthase. Science 326, 589-592 (2009).

515 40 Green, M. R. & Sambrook, J. Isolation of high-molecular weight DNA using organic

516 solvents. Cold Spring Harb. Protoc. pdb prot093450 (2017).

517 41 Bushnell, B., Rood, J. & Singer, E. BBMerge – Accurate paired shotgun read merging

518 via overlap PLoS One 12, e0185056 (2017).

519 42 Joshi, F. J. Sickle: A sliding-window, adaptive, quality-based trimming tool for FastQ

520 files (Version 1.33) (2011).

521 43 Bankevich, A. et al. SPAdes: A new genome assembly algorithm and its applications

522 to single-cell sequencing. J. Comput. Biol. 19, 455-477 (2012).

523 44 Peng, Y., Leung, S. M., Yiu, F. Y. L. & Chin, F. Y. IDBA-UD: a de novo assembler for

524 single-cell and metagenomic sequencing data with highly uneven depth

525 Bioinformatics 28, 1420-1428 (2012).

526 45 Marchler-Bauer, A. et al. CDD/SPARCLE: functional classification of proteins via

527 subfamily domain architectures Nucleic Acids Res. 45, D200-D203 (2017). bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

528 46 El-Gebali, S. et al. The Pfam protein families database in 2019. Nucleic Acids Res. 47,

529 D427-D432 (2019).

530 47 Edgar, R.C. MUSCLE: multiple sequence alignment with high accuracy and high

531 throughput. Nucleic Acids Res. 32, 1792-1797 (2004).

532 48 Sokal, R. R. & Michener, C. D. A Statistical Method for Evaluating Systematic

533 Relationships, University of Kansas (1958).

534 49 Hoff, K. J. & Stanke, M. Predicting genes in single genomes with AUGUSTUS. Curr.

535 Protoc. Bioinformatics 65, e57 (2019).

536 50 Cai, H. et al. A draft genome assembly of the solar-powered sea slug Elysia chlorotica

537 Sci. Data 6, 190022 (2019).

538 51 Patro, R. et al. Salmon provides fast and bias-aware quantification of transcript

539 expression. Nat. Methods 14, 417-419 (2017).

540 52 Davidson, N. M. & Oshlack, A. Corset: enabling differential gene expression analysis

541 for de novo assembled transcriptomes. Genome Biol. 15, 410 (2014).

542 53 Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor package for

543 differential expression analysis of digital gene expression data. Bioinformatics 26,

544 139-140 (2010).

545 54 Ireland, C. & Faulkner, J. The metabolites of the marine molluscs Tridachiella

546 diomedea and Tridachia crispata. Tetrahedron 37, 233-240 (1981).

547 55 Díaz-Marrero, A. R. et al. Validating an endoperoxide as a key intermediate in the

548 biosynthesis of elysiapyrones. Org. Lett. 10, 3057-3060 (2008).

549 56 Fu, X., Hong, S. P. & Schmitz, F. J. New polypropionate pyrones from the Philippine

550 sacoglossan mollusc Placobranchus ocellatus. Tetrahedron 56, 8989-8993 (2000). bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

551 57 Ksebati, M .B. & Schmitz, F. J. Tridachiapyrones: propionate-derived metabolites

552 from the sacoglossan mollusk Tridachia crispata J. Org. Chem. 50, 5637-5642

553 (1985).

554 58 Carbone, M. et al. First chemical study of the sacoglossan Elysia patagonica:

555 Isolation of a γ-pyrone propionate hydroperoxide. Biochem. Systematics Ecol. 49,

556 172-175 (2013).

557 59 Pierce, S. K. & Curtis, N. E. in International Review of Cell and Molecular Biology Vol.

558 293 (ed K. W. Jeon) 123-148 (Academic Press, 2012).

559 60 Christa, G. et al. Phylogenetic evidence for multiple independent origins of functional

560 in Sacoglossa (, ). Org. Divers. Evol. 15, 23-

561 36 (2015).

562 61 Pierce, S. K., Curtis, N. E. & Middlebrooks, M. L. Sacoglossan sea slugs make routine

563 use of photosynthesis by a variety of species-specific adaptations. Invert. Biol. 134,

564 103-115 (2015).

565 62 Viteri, C. Estudios experimentales de comportamiento en la babosa marina

566 fotosintética Elysia diomedea (Opisthobranchia: Sacoglossa) ante distintas

567 condiciones de luz y variedad de dieta B.S. thesis, Pontificia Universidad Católica del

568 Ecuador, (2012).

569 63 Schmitt, V. et al. Chloroplast incorporation and long-term photosynthetic

570 performance through the life cycle in laboratory cultures of Elysia timida

571 (Sacoglossa, Heterobranchia). (Sacoglossa, Heterobranchia). Front. Zool. 11, 5

572 (2014). bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

573 64 Laetz, E. M., Rühr, P. T., Bartolomaeus, T., Preisfeld, A. & Wägele, H. Examining the

574 retention of functional kleptoplasts and digestive activity in sacoglossan sea slugs.

575 Org. Divers. Evol. 17, 87-99 (2017).

576 65 Evertsen, J. & Johnsen, G. In vivo and in vitro differences in chloroplast functionality

577 in the two north Atlantic sacoglossans (Gastropoda, Opisthobranchia) Placida

578 dendritica and Elysia viridis. Mar. Biol. 156, 847-859 (2009).

579 66 Cartaxana, P., Trampe, E., Kühl, M. & Cruz, S. Kleptoplast photosynthesis is

580 nutritionally relevant in the sea slug Elysia viridis. Sci. Rep. 7, 7714 (2017).

581 67 Hinde, R. & Smith, D. C. The role of photosynthesis in the nutrition of the mollusc

582 Elysia viridis. Biol. J. Linnean Soc. 7, 161-171 (1975).

583 68 Evertsen, J., Burghardt, I., Johnsen, G. & Wägele, H. Retention of functional

584 chloroplasts in some sacoglossans from the Indo-Pacific and Mediterranean. Mar.

585 Biol. 151, 2159-2166 (2007).

586 69 Maeda, T. et al. Algivore or phototroph? Plakobranchus ocellatus (Gastropoda)

587 continuously acquires kleptoplasts and nutrition from multiple algal species in

588 nature. PLoS One 7, e42024 (2012).

589

590 Acknowledgements

591 We thank James Cox and John Alan Maschek from the University of Utah Metabolomics Core

592 Facility, which is supported by 1 S10 OD016232-01, 1 S10 OD021505-01 and 1 U54

593 DK110858-01. Yi Tang kindly provided S. cerevisiae BJ5464-NpgA. Paul Jones from Wake

594 Forest University and John Moses from LaTrobe University provided helpful discussion and

595 advice concerning the chemistry of mollusk pyrones. We are grateful to the people and nation of bioRxiv preprint doi: https://doi.org/10.1101/764225; this version posted September 10, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

596 Solomon Islands for permission to perform the study on P. ocellatus following Nagoya Protocol

597 standards. J.P.T., Z.L., and E.W.S. are supported by NIH R35GM122521 and U01TW008163.

598

599 Author information

600 Affiliations

601 Department of Medicinal Chemistry, University of Utah, Salt Lake City, UT 84112 USA

602 Joshua P. Torres, Zhenjian Lin, Jaclyn M. Winter & Eric W. Schmidt

603 Department of Biological Sciences, California State University, Los Angeles, CA 90032 USA

604 Patrick J. Krug

605 Contributions

606 J.P.T. and E.W.S. conceived of the project. J.P.T. performed experiments and bioinformatics to

607 discover EcPKS1. Z.L. performed bioinformatics. J.M.W. provided assistance and guidance

608 concerning yeast expression and biochemical characterization experiments. P.J.K. helped to

609 interpret the results and helped with mollusk . E.W.S. and J.P.T. wrote the manuscript

610 with input from all authors.

611

612 Corresponding author

613 Correspondence to Eric W. Schmidt ([email protected]).

614 Competing interests

615 The authors declare no competing interests.