Evolutionarily conserved aspects of animal nutrient uptake and transport in sea anemone vitellogenesis

SUMMARY The emergence of systemic nutrient transport was a key challenge during animal evolution, yet it is poorly understood. Circulatory systems distribute nutrients in many bilaterians (e.g., vertebrates and arthropods) but are absent in non-bilaterians (e.g., cnidarians and sponges), where nutrient absorption and transport remain little explored at molecular and cellular levels. Vitellogenesis, the accumulation of egg yolk, necessitates high nutrient inﬂux into oocytes and is present throughout animal phyla and therefore represents a well-suited paradigm to study nutrient transport evolution. With that aim, we investigated dietary nutrient transport to the oocytes in the cnidarian Nematostella vectensis (Anthozoa). Using a combination of ﬂuorescent bead labeling and marker gene expression, we found that phagocytosis, micropinocytosis, and intracellular digestion of food components occur within the gonad epithelium. Pulse-chase experiments further show that labelled fatty acids rapidly translocate from the gonad epithelium through the extracellular matrix (ECM) into oocytes. Expression of conserved lipid transport proteins vitellogenin ( vtg ) and apolipoprotein-B ( apoB ) and colocalization of labeled fatty acids with a ﬂuorescently tagged ApoB protein further support the lipid-shut-tling role of the gonad epithelium. Complementary oocyte expression of very low-density lipoprotein receptor ( vldlr ) orthologs, which mediate endocytosis of bilaterian ApoB- and Vtg-lipoproteins, supports that this evolutionarily conserved ligand/receptor pair underlies lipid transport during sea vitellogenesis. we lipid- and lipid our a that an lipid the and provided a basis for the of ‘‘half’’). the LG+I+G4) likelihood trees bootstrapping probabilities MrBayes v3.2.7 LG+I+G8 (Vtg) or WAG+I+G4 (LDLR, LRP1) two parallel runs, a temperature of 0.2 (LDLR, LRP1) or 0.05 (Vtg), eight 1 (LDLR, LPR1) or 2 (Vtg) swaps each swapping a frequency of every 20.000 (LDLR), 80.000 (Vtg) or 100.000 (LPR1) generations. consensus tree a burn-in of Maximum-likeli-hood (v/v) Triton X-100 in 1xPBS pH 7.4), the tissue was digested with 2.5 m g/ml Proteinase K in PTx for 5 minutes at room temperature (RT). Enzymatic digestion is stopped by two washes in 2mg/ml glycine/PTx. Samples were then washed in 1% (v/v) trietha-nolamine (TEA) in PTx, followed by a 1%TEA/ 3 m g/ml acetic anhydride and a 1%TEA/6 m g/ml acetic anhydride washing steps. After two PTx washes, tissue pieces were post-ﬁxed for 2 min in 0.2% (v/v) glutaraldehyde/PTx and 1 hour in 3.7% formaldehyde (FA, v/v)/ PTx. After washes in PTx and hybridization buffer (HB: 50% (v/v) 8M urea, 5x SSC pH 4.5, 0.3% (v/v) Triton X-100, 1% (w/v) SDS, 100 m g/ml heparin, 5 mg/ml Torula yeast RNA), we performed an overnight blocking step at 60 (cid:2) C in blocking buffer (5% (w/v) dextran sulfate (MW > 500,000, Sigma-Aldrich), 3% (w/v) Blocking Reagent (Roche) in HB). On the next day, digoxigenin (DIG)-labelled probes (see Gene cloning and RNA probe synthesis) were diluted to 0.75ng/ m l in blocking buffer and denatured for 10 minutes at 80 (cid:2) C before addition to the samples. Hybridization was conducted at 60 (cid:2) C over two days. Unbound probe was removed via a series of 60 (cid:2) C washes in HB/2X SSCT (0.03M sodium citrate, 0.3M NaCl, 0.3% (v/v) Triton X-100) solutions [75/25, 50/50, 25/75, 0/100 (v/v)] and in 0.1X SSCT. Additional washes in 0.1X SSCT/PTx were performed at RT [66/33, 33/66, 0/100 (v/v)], followed by consecutive washes in PBTx (0.1% (w/v) BSA, 0.3% (v/v) Triton X-100 in 1X PBS). Samples were then blocked for 1 hour in blocking solution (1% (w/v) Blocking Reagent (Roche)/maleic acid buffer (150mM maleic acid, 100mM NaCl, pH=7.5)) and incubated overnight with 1:2000 anti-DIG alkaline phosphatase antibody (Roche)/blocking solution (preabsorbed for 1h). On the next day, unbound antibody was removed by 10 increasingly long washes (2-30 minutes over a period of 2h30) with PBTx, after which samples were washed with alkaline phosphatase buffer (AP: 100mM NaCl, 50mM MgCl 2 , 100mM Tris pH9.5, 0.1% (v/v) Tween-20). The staining reaction (4.5 m l/ml NBT, 3.5 m l/ml BCIP in AP buffer) was performed at RT for 5 minutes (strongly expressed genes) to 8 hours (weakly expressed genes). The reaction was stopped by washing in PTw (0.1% (v/v) Tween-20 in 1X PBS), and unspeciﬁc staining removed by a 2 minutes 100% ethanol wash. Staining quality was further improved by storing the samples in 87% glycerol at 4 (cid:2) C for several days before further processing. followed by a 30 minutes permeabilization step at 4 (cid:2) C in PTx. The click reaction was performed using the reagents of a Click-iT EdU Cell Proliferation Kit for Imaging (Invitrogen) following the manufacturer’s protocol. The staining reaction was conducted for 30 minutes at room temperature. The tissue was then washed in 1x PBS three times over a period of 2 hours to remove unbound ﬂuorophore and stained with DAPI nuclear staining before cryosection.


In brief
In the sea anemone Nematostella vectensis, the vitellogenin-/apoBexpressing gonad epithelium exhibits high endocytosis and fatty acid translocation levels during vitellogenesis. Complementary expression of vldl receptor genes in oocytes supports that ECM-based lipoprotein transport is evolutionarily conserved between cnidarians and bilaterians.

INTRODUCTION
The emergence of circulatory systems ensured a constant nutrient supply to all tissues during the evolution of large and complex body plans in many bilaterians. 1,2 Non-bilaterians, in contrast, exhibit much simpler body plans that lack circulatory systems and consist of mostly two cell layers without intermediate tissues. It is therefore widely assumed that systemic nutrient distribution by circulatory systems has emerged in bilaterian lineages. [1][2][3] As dietary nutrient uptake and transport are poorly studied on a cellular or molecular level in any non-bilaterian animal, the evolutionary origin of animal nutrient transport systems remains speculative. A prevailing hypothesis proposes that fluid and nutrient transport through an increasingly porous extracellular matrix (ECM), as reflected by blood vascular system development in many bilaterians, represents the ancestral condition from which bilaterian circulatory systems have further specialized. [1][2][3] Here, we test this hypothesis by investigating whether ECM-based nutrient uptake and transport mechanisms are conserved between cnidarian and bilaterian vitellogenesis.
Yolk production necessitates large amounts of nutrients, especially lipids and proteins, and is one of the most widely studied nutrient transport processes in bilaterians. 4 Vitellogenin (Vtg) and ApolipoproteinB (ApoB; Apolipophorins in insects), two proteins of the large lipid transfer protein (LLTP) superfamily, play conserved roles during bilaterian lipid transport. 5,6 Vtg is an egg yolk precursor protein that is primarily produced in female extra-ovarian tissues (e.g., vertebrate liver, insect fat body) and mediates lipid transport specifically into oocytes. 4,[7][8][9] ApoB, in contrast, shuttles lipids systemically between the gut epithelium, vitellogenic organs, and other peripheral tissues. [10][11][12] In many bilaterians, Vtg and ApoB lipoproteins move through hemolymph or blood vascular systems 4 and are endocytosed into oocytes or other target tissues (e.g., muscles) by conserved orthologs of the very low-density lipoprotein (VLDL) receptor family (including arthropod Vtg or Apolipophorin receptors). [13][14][15] Although Vtg or ApoB orthologs are almost ubiquitously present among animal phyla, [16][17][18] their role in systemic lipoprotein transport is poorly studied in many animals, especially non-bilaterians.
The sea anemone Nematostella vectensis exhibits a typical cnidarian body plan consisting of only two epithelia: the inner gastrodermis and outer epidermis. Food uptake occurs through the mouth into the blind-ended gastric cavity. As gastrodermal cilia create internal fluid flows, the gastric cavity is commonly thought to constitute the main nutrient distribution system in this phylum. 19 Gastrodermal folds (mesenteries) reach into the gastric cavity and likely represent an ancestral feature of cnidarians. 20,21 They are subdivided into functionally distinct regions such as the gonad 22-24 ( Figure 1A). Oocytes or spermaries locate to mesenteries within the oral half of the Nematostella body column ( Figure 1A, ''gonad region''). Here, they are embedded in the ECM, the mesoglea, between two layers of somatic gonad epithelia 24,25 ( Figure 1A). Their mesogleal localization, isolated , and trophic tract (B 00 -E 00 ) after 30 min (B-B 00 ), 2 h (C-C 00 ), and 24 h (D-D 00 ) of incubation. A similar distribution of particle uptake is seen when incubating with 20 nm FS (E-E 00 ). Distal end of mesenteries oriented to the top (B-E 00 ). All data figures represent single confocal plane images of adult Nematostella mesentery cross sections. Arrowheads in (B 0 ) and (B 00 ): intracellular beads as estimated by their relative location to cortical F-actin enrichment. Magenta, phalloidin (F-actin); blue, DAPI. Scale bars, 25 mm (B-E) and 10 mm (E 0 and E 00 ). CGT, cnidoglandular tract; CT, ciliated tract; ep, epidermis; gastr, gastrodermis; IT, intermediate tract; mes, mesentery; meso, mesoglea; ooc, oocyte; retr. m, retractor muscle; SGE, somatic gonad epithelium; SF, septal filament; tent, tentacle; TT, trophic tract. See also Figure S1. from direct food access, is common among anthozoans (sea anemones, corals, sea pens), scyphozoans (''true'' jellies), cubozoans (cube jellies), and hydrozoan medusae 19,26 and thus likely ancestral among cnidarians. This raises the question of whether nutrients reach the cnidarian oocytes via the ECM using molecules and mechanisms that are evolutionarily conserved between cnidarians and bilaterians.
Here, we have investigated how nutrients, especially lipids, are transported from the gastric cavity to the oocyte in Nematostella to better understand the evolution of animal dietary lipid transport systems. The comparison with well-studied bilaterians, the phylogenetic sister group to cnidarians, has allowed the reconstruction of key cellular and molecular aspects of the lipid transport system present in the last common ancestors of cnidarians and bilaterians.

RESULTS
High endocytic activities are restricted to specialized regions of the Nematostella mesentery The body regions absorbing food particles or nutrients are only vaguely described in sea anemones [27][28][29] and are currently unknown in Nematostella. We therefore used 1 mm and 20 nm bovine serum albumin (BSA)-coated fluorescent microspheres (FS, FluoSpheres) to probe for phagocytosis or micropinocytosis, respectively, in adult Nematostella polyps [30][31][32] (Figures 1B-1E 00 and S1). We found that 20 nm beads, but not the 1 mm beads, were enriched in the body wall epidermis after 24 h of incubation (Figures S1A and S1B). This result suggests that micropinocytosis of nano-sized particles from the surrounding water is the principal absorptive mechanism of the outer epithelium. Both bead sizes were present in the gastrodermis of the body wall and were strongly enriched in tentacles (Figures S1C and S1D) and in three different parts of the mesenteries: the intermediate tract  Figures 1A-1E 00 ). Notably, we have not observed any bead uptake into trophonema cells (n = 11 trophonemata with 20 nm FS, n = 14 trophonemata with 1 mm FS; asterisks in Figures S1E and S1F). Phagocytosis of 1 mm FS becomes apparent after 30 min of incubation ( Figures 1B-1B 00 ) and leads to intracellular accumulation after 2 and 24 h ( Figures 1C-1D 00 ). A similar uptake distribution is observed using fluorescently labeled, heat-killed E. coli cells, confirming the physiological relevance of BSA-coated bead experiments (Figures S1G-S1G 00 ).
Interested to explore whether endocytosed microspheres become mobilized, we used a 4-h incubation pulse of 20 nm beads and found similar distributions after 20-h-, 7-day-, and 14-day-long chase periods (Figures S1H-S1K 00 ), indicating that no transport occurs between body regions. Biweekly incubations over 2 weeks led to an intracellular accumulation of 20 nm beads in the apical half of SGE cells (Figures S1L and S1M). Notably, a subset of SGE cells, which due to their abundance in the SGE are likely mucus cells, 33 shows no nanobead uptake (Figures S1L and S1M). We found altogether that three mesenterial regions, including the SGE, show increased phagoand micropinocytosis activities in Nematostella.
Phagocytic cell types are well studied on the molecular level in the context of immunity (e.g., macrophages) but barely in the context of nutrition. 34 We therefore investigated the spatial gene expression profiles of endocytic pathway markers and transcription factors in the nutritive phagocytic cells of the Nematostella mesenteries (Figures 2 and S2). Genes were selected for analysis based on either their informative value as conserved marker genes for cellular processes (e.g., phagocytosis, lysosomal digestion), their co-occurrence in putative endocytic ''metacells'' of a whole-polyp, single-cell RNA sequencing dataset 35 (Table S1; Data S1), or as a spatial reference to juvenile nutrient storage regions. 36 We found that lipopolysaccharide binding protein/bactericidal permeability-increasing protein (lbp/bpi; Figure S2A) and C-type lectin mannose receptor (mannR; Figure S2B) gene orthologs of Nematostella are specifically expressed in the endocytic TT (both genes), IT (lbp/bpi only), and SGE (mannR only) (Figures 2A-2B 00 and 2J). Their bilaterian orthologs recognize bacterial lipopolysaccharide and glycans as well as soluble macromolecules and large particulate matter. MannR is additionally known to induce phagocytosis and clathrin-mediated endocytosis. [37][38][39] Their expression in Nematostella thus supports a dual role for endocytic regions in both immunity and nutrition. Next, we checked the expression of Nematostella orthologs of cdc42 and rhoA genes, which play key roles in phagocytic cup formation, and of low-density lipoprotein receptor-like (ldlr-like), clathrin light chain (clatLC), and clathrin heavy chain (clatHC) genes, all encoding for proteins involved in clathrin-mediated endocytosis. Expression of all these genes overlaps with bead uptake in IT, TT, and SGE regions (Figures 2C-2G 00 and 2J), supporting phagocytosis and clathrinmediated endocytosis as their main endocytic mechanisms. Additionally, the predominant expression of gene orthologs encoding for glycosidases (a-glucosidase, a-mannosidase), proteases (cathepsin), and cholesterol transport/recycling (npc2) proteins, in all or a subset of the three endocytic mesenterial regions (Figures S2C-S2K), supports increased intracellular digestion levels. In juvenile Nematostella, the median, lipid-storing part of the mesentery was previously reported to co-express a combination of foxC, six4/5, and nkx3/bagpipe transcription factor genes, revealing striking similarities to lateral mesoderm derivatives of bilaterians, such as somatic gonad or nutrient storage tissues (e.g., insect fat body). 36,40 How this juvenile region relates to adult endocytic structures was yet to be determined. While nkx3/ bagpipe expression levels were below detection limits, foxC and six4/5 genes were expressed in all endocytic regions of the adult mesentery ( Figures 2H-2J), indicating similarities to the lipid-uptaking region of juvenile mesenteries.
In order to test how the expression profile of Nematostella endocytic cells compares with putative trophic phagocytes or lipidstoring cells of other animals, we searched previously published single-cell transcriptomes of the stony coral Stylophora pistillata, the planarian Schmidtea mediterranea, and the sponge Amphimedon queenslandica for metacells/clusters that share at least 2 of our candidate genes (Table S1; Data S1). We found the highest overlap with metacells/clusters identified as Stylophora gastrodermis or ''alga-hosting'' cells, Schmidtea cathepsin+ parenchymal, intestinal, or neuronal cells, or as Amphimedon pinacocytes or archeocytes. Notably, all identified metacells/ clusters (except Schmidtea neural cells) refer to cell types previously characterized as phagocytosic and/or lipid-storing. In addition, the presence of six4/5 or foxC orthologs in a subset of identified metacells/clusters throughout all datasets supports a potential conservation of six4/5 and foxC in endocytic and/or lipid-storing cell types across animals. Our characterization of cellular particle uptake modalities and gene expression profiles has revealed that distinct mesenterial regions are specialized in food particle endocytosis. We next tested whether these regions provide dietary nutrients to the oocytes by studying the cells and molecules Dietary fatty acids reach the oocyte by trans-epithelial transport Spatial detection of neutral lipid storage using Oil Red O (ORO) staining confirmed that lipids are a major yolk component in Nematostella ( Figure S3A). It also showed high levels of neutral lipids largely overlapping with endocytic regions (SGE, IT, and TT) in both male and female adults ( Figures S3A-S3C). In contrast, lipid levels appeared low in non-endocytic regions such as the ciliated tract (CT) or cnidoglandular tract (CGT), suggesting that only specialized endocytic cells and gametes possess lipid transport or storage abilities. We explored the dynamics of fatty acid uptake and transport from the gastric cavity toward the oocyte by developing a whole-body pulse-chase assay using a ''clickable,'' alkyne-modified oleic acid 41 (alkyne-OA). Previous studies used alkyne-OA to reliably mimic oleic acid and study its metabolism in yeast, Drosophila, and mammalian cells. 41 As oleic acid is abundantly found in anthozoan storage lipids, it is well-suited to track dietary fatty acids in Nematostella. 42 In oocytes, alkyne-OA strongly accumulated in vesicles after 20 h and 7 days of chase ( Figures 3D and 3E), confirming major dietary fatty acid movement from the gastric cavity into ECM-based oocytes. We increased the temporal control and resolution of alkyne-OA delivery by directly injecting alkyne-OA/BSA solution into the gastric cavity ( Figures 3F-3K and S3I-S3N). This optimized delivery method yielded similar results to feeding enriched brine shrimps, which relied on individual rates of extracellular digestion (compare Figures S3J and S3K Figures S3E and S3F). We found that 30 min after delivery, alkyne-OA localized to large apical vesicles in SGE cells ( Figure 3G). Within 1 h of incubation, smaller vesicles appeared in median and basal regions of SGE cells, suggesting rapid intracellular apical-to-basal translocation of alkyne-OA (Figure 3H, white arrowheads). Between 1 and 2 h after delivery, alkyne-OA vesicles started accumulating intracellularly at the  (Figures 3I and 3K, red arrowheads; Figure S3N). A 24-h-long incubation revealed the uptake of alkyne-OA in trophonema cells of the female SGE (24 out of 24 trophonemata; Figures 3J and 3K). These observations are in full agreement with the trans-epithelial transport of alkyne-OA from the gastric cavity through the SGE into oocytes via the surrounding ECM. Our inability to detect alkyne-OA in the ECM could indicate a rapid clearance by high levels of endocytosis into oocytes, which is strongly supported by ultrastructural observations. 25 Notably, the absence of endocytic activity in trophonema cells and of alkyne-OA in the adjacent oocyte region questions their previously proposed role as main nutrient-shuttling cells in Nematostella and other sea anemones. 22,24, 44 We were next interested to investigate how lipid transport through the SGE and ECM could be mediated on the molecular level.
To do so, we studied the expression of the single Nematostella orthologs of vtg and apoB apolipoprotein genes ( Figure S4A) and three Nematostella vldl receptor (vldlr) paralogs conserved among sea anemones and corals (Figures S4B and S4C). We confirm previous reports from Nematostella and other anthozoans that the SGE expresses high levels of vtg transcripts ( Figures 4A-4A 00 ). [45][46][47][48] In contrast, no vtg transcript was detected in oocytes despite Vtg representing ±60% of the protein content of the mature egg in Nematostella. 49 As cytoplasmic connections between oocytes and the gonad epithelium are absent, 25,50 this indicates that Vtg protein accumulation in anthozoan oocytes is largely dependent on endocytosis. vldlr-A1, -A2, and -B genes ( Figures S4B and S4C) are all expressed in growing oocytes of different sizes ( Figures 4B-4E), supporting a role in receptormediated apolipoprotein uptake during vitellogenesis. vldlr-A1 and vldlr-A2 are additionally expressed in uncharacterized cells of the IT and reticulate tract of the septal filament (SF) ( Figures 4B 00 and 4C 00 ). Altogether, our data strongly suggest that Vtg mediates lipid transport from SGE cells into growing oocytes by VLDLR-mediated endocytosis, as widely found among bilaterians. 4,51 Although ApoB-encoding genes are found in genomes of bilaterians, cnidarians, and placozoans ( Figure S4A), their expression profiles and role in lipid transport have not been characterized in any non-bilaterian animal. In Nematostella, apoB expression colocalizes with vtg expression in the SGE but further extends to the IT and TT, thus fully overlapping with endocytic regions (Figures 5A-5C). This broader expression suggests a role for ApoB in lipid transport beyond vitellogenesis, as typically found in bilaterians. 52 To further explore this possibility, we aimed to study the localization of ApoB proteins and their potential colocalization with alkyne-OA. We therefore generated a transgenic reporter line using CRISPR-Cas9-mediated knockin technology, resulting in a genomically encoded, C-terminal ApoB-PSmOrange fusion protein ( Figure S5). As expected, ApoB-PSmOrange protein detected by immunofluorescence localizes to endocytic regions with high apoB transcript levels (compare Figures 5A-5C with 5D-5I). Large ApoB-PSmOrange-containing vesicles are distributed throughout the cells of the IT (Figures 5D and 5E) and male SGE ( Figure 5H). In cells of the TT and female SGE, similarly sized vesicles accumulate apically (Figures 5F, 5G, and 5I). Growing oocytes were devoid of the ApoB-PSmOrange signal ( Figure 5G), as supported by low levels of ApoB protein detected by mass spectrometry in spawned eggs. 49 Notably, however, ApoB-PSmOrange protein was detected in spermaries, with higher intensities in peripheral, immature cells compared with central, mature spermatozoa ( Figure 5H). Here, ApoB-PSmOrange and alkyne-OA localize to adjacent but clearly distinct vesicles ( Figures 6E-6E 00 ). While we found broad overlap between tissues with high ApoB-PSmOrange protein and alkyne-OA uptake levels, we detected only partial intracellular colocalization in the male and female SGE after a 2-h alkyne-OA pulse ( Figures 6A-6B 00 ). Interested to investigate a potential role for ApoB in systemic lipid transport, we compared the distribution of ApoB-PSmOrange and alkyne-OA between a 2-h pulse ( Figures 6A-6B 00 ) and a 4-h pulse, followed by a 7-day chase ( Figures 6C-6E 00 ). Strikingly, we found ApoB-PSmOrange/alkyne-OA colocalization in mesenchymallike mesogleal cells of the trophic region only in the pulse-chase experiment ( Figures 6C-6D 00 ). This result shows that oleic acid became mobilized in endocytic epithelial cells, as indicated before by intracellular apical-to-basal movements within TT cells ( Figures S3E-S3H), and translocated through the ECM into mesogleal cells. The most likely explanation, to be tested in future studies, is that ApoB-mediated lipid transport from epithelial to mesogleal cells enables subsequent systemic lipid distribution in sea anemones.

DISCUSSION
In many bilaterian animals, dietary nutrients travel from the gut epithelium to oocytes via ECM-based circulatory systems or, more rarely, through coelomic cavity fluids. 4,53,54 Due to the lack of data from non-bilaterian animals, the emergence of systemic nutrient transport and its underlying cellular and molecular features during animal evolution has remained speculative. [1][2][3] We have, therefore, retraced the cellular path of dietary nutrients, more specifically lipids, from the gastric cavity to the oocytes during vitellogenesis in the sea anemone Nematostella vectensis. We have shown that although the entire gastrodermal epithelium is in direct contact with food particles and nutrients, it does not show a uniform distribution of phagocytosis, micropinocytosis, and lysosomal activities. Instead, our analyses of bead uptake assays and marker gene expression show that these activities are enriched in the tentacle gastrodermis as well as the IT, SGE, and TT mesenterial regions. In corals and symbiotic sea anemones (e.g., Exaiptasia), carbon-fixing Symbiodinium sp. dinoflagellates locate mainly to the tentacle gastrodermis and the IT and SGE regions. 23,55 These parallels confirm the importance of these regions in nutrient acquisition in sea anemones and corals and raise questions about the symbionts' ability to permanently populate these highly endocytic regions in some anthozoan species but not in others.
As our molecular knowledge of animal phagocytosis relies almost entirely on the study of innate immunity cell types (e.g., macrophages), we provide one of the first molecular characterizations of animal trophic phagocytes. 34 Notably, the set of genes found expressed in nutritive endocytic regions in Nematostella, composed of immunity-related pattern recognition proteins (LPS/BPI, C-type lectins), phagocytic cup and clathrinmediated endocytosis components, as well as lysosomal enzymes and transporters, is largely shared with immune cells in vertebrates and flies. [56][57][58][59] This raises the question of whether Nematostella trophic phagocytes are bi-functional and carry out nutritive as well as immunity-related functions, e.g., in controlling the gastrodermal surface microbiome composition or preventing symbiont settlement. Currently, a reconstruction of trophic and immune phagocytic cell-type evolution remains difficult, as only little comparative molecular data are available from bilaterian trophic phagocytes. 34,60 Nevertheless, our comparison with available single-cell transcriptomes reveals expression profile ll OPEN ACCESS similarities, mainly with trophic phagocytic cells such as coral alga-hosting and gastrodermal cells, planarian phagocytic parenchymal and intestinal cells, 34,61 and sponge pinacocytes and archeocytes. 40 The shared expression of foxC and six4/5 transcription factors in endocytic/lipid-storing cells of Nematostella, corals, planarians, and sponges is reminiscent of their coexpression in lateral mesoderm regions of bilaterians, whose typical derivatives include blood immune cells, nutrient storage tissues (e.g., insect fat body), or somatic gonad tissue. 62 More in-depth analysis of a broader set of species is needed to reveal the evolutionary relationship of endocytic/lipid-storing cell types across animal phyla.
We identified the SGE as highly relevant among mesenterial endocytic regions for dietary nutrient transport during vitellogenesis. Indeed, our data show that the SGE plays an important role in dietary particle and lipid uptake and as a Vtg-producing tissue. Our alkyne-oleic acid tracking experiments strongly suggest that dietary fatty acids are rapidly absorbed and transported through the SGE into the ECM before being internalized via endocytosis by oocytes, as corroborated by ultrastructural observations. 25 As Vtg transcripts are highly expressed in the SGE and the corresponding protein is highly abundant in the mature egg, 49 we propose that it constitutes the key lipid transporter from the SGE through the ECM into oocytes during vitellogenesis in Nematostella. We found three Nematostella vldl receptor paralogs expressed during oocyte growth, of which VLDLR-B was previously identified in the proteome of mature eggs. 49 These results support the idea that Vtg endocytosis is mediated by VLDL receptors in Nematostella. Together, our data suggest that a Vtg ligand/VLDL receptor pair has been evolutionarily conserved during vitellogenic lipid transport since the last common bilaterian-cnidarian ancestor.
ApoB protein, involved in systemic lipid transport in bilaterians, appears to have no direct role during vitellogenesis in Nematostella. Although abundantly found in the SGE, it was detected at low levels in oocytes via mass spectrometry 49 and below immunofluorescence detection limits. Instead, we found ApoB-positive vesicles in spermatogonia or immature spermatozoa, suggesting a role in lipid transport or metabolism during spermatogenesis. Loss of ApoB function has been linked to male infertility so far only in mice 63,64 and further investigations are necessary to corroborate an evolutionarily conserved function of ApoB during animal spermatogenesis. In addition to the SGE, ApoB-PSmOrange protein levels were high in all other endocytic tissues of Nematostella. Partial colocalization with apical alkyne-OA suggests that ApoB might function in intracellular fatty acid transport. Strikingly, we also found ApoB protein broadly colocalizing with pulse-chased alkyne-OA in single mesogleal cells. Based on location and lipid content, these cells are highly reminiscent of motile amoebocytes that were previously identified in several anthozoans. [65][66][67] Our observation therefore supports previous assumptions that amoebocytes function in systemic nutrient transport via the mesoglea in anthozoans. [65][66][67] A prevalent hypothesis has stated that bilaterian circulatory systems evolved from a simple ECM-based fluid and nutrient transport system 2 but has so far lacked experimental support. Here, we show first evidence that Vtg/ApoB-and VLDLR-mediated lipid transport through the ECM space is not only conserved in bilaterians but also found in a sea anemone. We suggest, therefore, that the local, ECM-based transport of yolk precursors predates the cnidarian-bilaterian split. Furthermore, we propose that this simple nutrient transport system formed a framework for the evolution of intricate ECM-based circulatory systems, ensuring a systemic nutrient supply in increasingly large and complex bilaterians.

Materials availability
Plasmids and genetically modified Nematostella vectensis generated for this study are available upon completion of a Material Transfer Agreement.
Data and code availability cDNA fragments amplified and cloned for generating in situ hybridization probes have been deposited in Genbank and are publicly available as of the date of publication. Accession numbers are listed in Data S2. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS
Nematostella vectensis culture Nematostella vectensis polyps are derived from the original culture established by C. Hand and K. Uhlinger. 81 Adult animals (> 6 months) were maintained in the dark at 18 C in 16ppm diluted sea water (Nematostella medium, NM) and fed 5 days per week with brine shrimp nauplii. Spawning was induced approximately every three weeks by a 12-hours shift in temperature (from 18 C to 25 C) and exposure to light, as described previously. 82 For injection, fertilized egg packages were incubated in a 3% (w/v) cysteine/NM solution to remove the egg jelly. Injected animals were raised at 25 C for 2 to 3 months, and subsequently transferred to 18 C for regular induction. The use of female or male specimen is specified in the figures and figure legends throughout the manuscript wherever relevant.
For all particle uptake assays, animals were initially relaxed in 0.1M MgCl 2 /NM for 20 minutes and were kept in this medium for incubation times below 4 hours. For longer incubations, animals were kept in MgCl 2 /NM for the first 4 hours and then transferred to fresh NM. The diluted Fluospheres or E. coli solution was injected inside the body cavity through the mouth using a thinned glass Pasteur pipet. The animals were kept in the dark at 18 C during incubation.
At the end of the assays, animals kept in NM were transferred back to 0.1M MgCl 2 /NM solution and left to relax for 20 minutes. The body cavity of all animals was flushed and inflated with this medium through the mouth to ensure full extension of the mesenteries. The polyps were then transferred to a 3.7% (v/v) Formaldehyde (Merck)/NM solution and opened longitudinally along the body column to ensure penetrance of the fixative and to allow non-internalized particles to be washed off. Polyps were fixed overnight at 4 C in the dark, after which the mesenteries were dissected in fixative and cut to 3-5mm long pieces. The tissue pieces were washed in 1x PBS/0.1% (v/v) Tween-20, counterstained for F-actin with Phalloidin (1:17, Alexa-488 or Alexa-647; Thermo Fisher) and DAPI nuclear staining (1:1000) before being processed for cryosectioning. Sectioning was performed at À25 C in a Leica Cryostat CM1850 and the 12mm sections were mounted on Thermo Scientific SuperFrost Plus adhesion slides (Thermo Fisher). After drying, the sections were post-fixed for 10 min in 3.7% (v/v) FA/1x PBS, rinsed 2x in 1xPBS, and mounted in 87% glycerol (non-fluorescent samples) or ProLong Gold antifade reagent (Molecular Probes, fluorescent samples).
Oil Red O staining Animals were fixed, dissected and mesentery pieces sectioned with a cryotome and post-fixed as described above. Oil Red O (ORO) stock solution was prepared by diluting 5mg/ml ORO powder in isopropanol, shaking at room temperature for 4 hours. The stock solution was further diluted to 60% (v/v) in milliQ water, left shaking at room temperature for another 2 hours and filtered using a 0.22mm syringe filter to produce a working solution. 94,95 Glass slides with mesentery sections were incubated for 15 minutes with the ORO working solution, then washed several times in 1xPBS.
Transmitted light and confocal imaging Images of in situ hybridization-and ORO-stained tissue were taken on a Nikon Eclipse E800 using a 20x air, 40x air or 60x oil-immersion objective. Images of fluorescent assays were taken either on a Leica SP5 confocal microscope using standard PMT detectors and a 20x or 63x oil-immersion lens, on a Leica SP8 confocal microscope using HyD detectors and 40x water-immersion or 100x oilimmersion objectives, or on an Olympus FLUOVIEW FV3000 confocal microscope using standard PMT detectors and a 60x silicon oil immersion lens. Transmitted light images were corrected for levels and colour balance and cropped using Photoshop CC. Fluorescent stacks were processed, cropped and the levels corrected using Fiji. 80