Bacterial precursors and unsaturated long-chain fatty acids are 2 biomarkers of North-Atlantic demosponges

21 Sponges produce distinct fatty acids (FAs) that (potentially) can be used as 22 chemotaxonomic and ecological biomarkers to study endosymbiont-host interactions and the 23 functional ecology of sponges. Here, we present FA profiles of five common habitat-building 24 deep-sea sponges (class Demospongiae, order Tetractinellida), which are classified as high 25 microbial abundance (HMA) species. Geodia hentscheli, G. parva, G. atlantica, G. barretti, 26 and Stelletta rhaphidiophora were collected from boreal and Arctic sponge grounds in the 27 North-Atlantic Ocean. Bacterial FAs dominated in all five species and particularly isomeric 28 mixtures of mid-chain branched FAs (MBFAs, 8- and 9-Me-C 16:0 and 10 and 11-Me-C 18:0 ) 29 were found in high abundance (together ≥ 20% of total FAs) aside more common bacterial 30 markers. In addition, the sponges produced long-chain linear, mid- and a(i)- branched 31 unsaturated FAs (LCFAs) with a chain length of 24‒28 C atoms and had predominantly the 32 typical Δ 5,9 unsaturation, although also Δ 9,19 and (yet undescribed) Δ 11,21 unsaturations were 33 identified. G. parva and S. rhaphidiophora each produced distinct LCFAs, while G. atlantica, 34 G. barretti , and G. hentscheli produced similar LCFAs, but in different ratios. The different bacterial precursors varied in carbon isotopic composition (δ 13 C), with MBFAs being more 36 enriched compared to other bacterial (linear and a ( i )-branched) FAs. We propose biosynthetic pathways for different LCFAs from their bacterial precursors, that are consistent with small isotopic differences found in LCFAs. Indeed, FA profiles of deep-sea sponges can serve as chemotaxonomic markers and support the conception that sponges acquire building blocks


Introduction
In addition, sponge FA composition may have taxonomic value, at least on a higher 66 classification level (e.g. class level), since Demospongiae, Hexactinellida ('glass' sponges), 67 Calcarea, and Homoscleromorpha have distinct FA profiles (17). However, the 68 chemotaxonomic value on a lower classification level is disputable, since composition may 69 alter with environmental conditions (18). The FA composition of sponges, especially 70 combined with (natural abundance) stable isotope analysis, has been shown a valuable tool to 71 infer dietary information on sponges, such as feeding on coral mucus (19), phytoplankton 72 (20) and methane-fixing endosymbionts (21). 73 The North-Atlantic Ocean is home to extensive sponge grounds, that are widespread 74 along the continental shelves, seamounts, and on the abyssal plains (22,23). Geodiidae and 75 other sponge species of order Tetractinellida (class Demospongiae) are major constituents of 76 these sponge grounds, representing >99 % of sponge ground benthic biomass (23-25). 77 Geodiidae spp. are high microbial abundance (HMA) sponges that harbor rich, diverse and 78 specific microbial communities (bacteria and archaea) involved in several biogeochemical 79 processes, as observed in G. barretti (26). This is reflected in the FA composition of G. 80 barretti that is dominated by bacterial FAs (12), including the distinct MBFAs that represent 81 28% of total FAs (12). However, the FA profiles of other Geodiidae are not described in the 82 literature yet. 83 In this study we analyzed the FA profiles of five common deep-sea Tetractinellids, 84 from different assemblages distinguished by temperature in the North Atlantic: the Arctic 85 sponge ground assemblages accommodate G. parva, G. hentscheli, and Stelletta spp. (e.g. S. 86 with the C isotope (δ 13 C) signatures of LCFAs and bacterial precursors. The high abundance 91 of endosymbiont markers that are precursors of LCFAs, indicate that these deep-sea sponges 92 use their endosymbionts as metabolic source. 6 114 115 Lipid extraction and FAME preparation 116 Approximately 100 mg of sponge powder of each individual sponge was used per 117 extraction. Sponge lipids were extracted with a modified Bligh and Dyer protocol (30), which 118 was developed at NIOZ Yerseke (31-33). We adjusted this protocol by replacing chloroform 119 with dichloromethane (DCM), because of lower toxicity. The whole protocol is available 120 online: dx.doi.org/10.17504/protocols.io.bhnpj5dn. In short, sponge tissue samples were 121 extracted in a solvent mixture (15 mL methanol, 7.5 mL DCM and 6 mL phosphate (P)-122 buffer (pH 7-8)) on a roller table for at least 3 hours. Layer separation was achieved by 123 adding 7.5 mL DCM and 7.5 mL P-buffer. The DCM layer was collected, and the remaining 124 solution was washed a second time with DCM. The combined DCM fraction was evaporated 125 to obtain the total lipid extract (TLE), which was subsequently weighed. An aliquot of the 126 TLE was separated into different polarity classes over an activated silica column. The TLE 127 was first eluted with 7 mL DCM (neutral lipids), followed by 7 mL acetone (glycolipids) and 128 15 mL methanol (phospholipids). The phospholipid (PL) fraction, which was used for further 129 analysis, was converted into fatty acid methyl esters (FAMEs) using alkaline methylation 130 (using sodium methoxide in methanol with known δ 13 C). Alkaline methylation is 131 recommended for complex lipid mixtures (34). After methylation, FAMEs were collected in 132 hexane and concentrated to ~100 µL hexane for gas chromatography (GC) analysis. 133 For this study, two individual sponge samples per species were selected for detailed 134 analysis. Aliquots of the FAME samples were used for double bond identification using 135 dimethyl disulfide (DMDS) derivatization (35). Samples reacted overnight at 40°C in 50 µL 136 hexane, 50 µL DMDS and 10 µL 60 mg/mL I 2 . The reaction was stopped by adding 200 µL 137 hexane and 200 µL Na 2 S 2 O 3 . The hexane layer was collected, and the aqueous phase was 138 washed twice with hexane. The combined hexane fraction was dried, subsequently eluted . CC-BY 4.0 International license made available under a (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 The copyright holder for this preprint this version posted October 9, 2020. ; https://doi.org/10.1101/2020.10.09.332833 doi: bioRxiv preprint over a small Na 2 SO 4 column using in DCM: methanol (9:1) to remove any water and re-140 dissolved in hexane in a GC-vial for GC-analysis. Another aliquot of FAME sample was used 141 for methyl-branching identification using catalytic hydrogenation with Adams catalyst (PtO 2 ) 142 and hydrogen. Each FAME sample, dissolved in ~3 mL ethyl acetate with 10-30 mg PtO 2 and 143 a drop of acetic acid, was bubbled with hydrogen gas for at least 1 h, after which the reaction 144 vial was closed, and stirred overnight at room temperature. Subsequently, each sample was 145 purified over a small column consisting of MgSO 4 (bottom) and Na 2 CO 3 (top) using DCM 146 and analyzed after re-dissolving it in ethyl acetate.

167
The lipid yield of G. barretti, G. hentscheli, G parva, and S. rhaphidiophora was 168 similar, around 2-3 % of dry weight (DW). Only G. atlantica had a lower lipid yield, about 169 1.6 % of DW. The FA profiles of PL resembled those of TLE (Table S1). However, 170 identification was more difficult using TLEs, because LCFAs co-eluted with sterols, hence 171 PL chromatography was used for identification and composition analysis. FAs are presented in both ω and Δ (IUPAC) annotation to avoid unambiguity and in a 180 hybrid form, which is typical of sponge LCFA annotation (17,38) (Table 1). Unsaturation is 181 described as C x:y , where x is the number of C atoms and y is the number of double bonds, 182 which is followed by Δ and all double bond positions from the carboxylic acid end in Δ 183 notation, and the position of the first double bond from the methyl (terminal) end in ω 184 notation (Table 1, Fig. 1). Methyl branching according to IUPAC notation is described as y- 185 Me-C x , where y is the position of the branching from the carboxylic acid end and x is the 186 number of C atoms at the backbone, excluding the branching (Fig. 1). The ω notation follows . CC-BY 4.0 International license made available under a (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 The copyright holder for this preprint this version posted October 9, 2020. ; https://doi.org/10.1101/2020.10.09.332833 doi: bioRxiv preprint the terminology of IUPAC for MBFAs, but deviates for terminally branched FAs. The 188 penultimate (ω2) and pen-penultimate methyl branching (ω3) are described with ω notation 189 as iso (i-C x ) and anteiso (a-C x ) where x is the total number of C atoms, including the 190 branching (Table 1

315
Stable C isotope values (δ 13 C) 316 Stable C isotope values (δ 13 C) of dominant FAs ranged between -18 ‰ (95 percentile) 317 and -26 ‰ (5 percentile) and showed similar patterns across all demosponges (Fig. 4, Table   318 2). The δ 13 C values of the dominant MBFAs, Me-C 16:0 , Me-C 18:0 , and also Me-C 18:1 ω12 (and . CC-BY 4.0 International license made available under a (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 The copyright holder for this preprint this version posted October 9, 2020. ; https://doi.org/10.1101/2020.10.09.332833 doi: bioRxiv preprint ω14) were enriched in 13 C compared to other bacterial fatty acids, (a(i)-C 15:0 , C 16:1 ω7, 320 C 18:1 ω7) (Fig. 4, Table 2). The most depleted FA was i-C 17:1 ω7 (-25.7 ± 1.3 ‰ δ 13 C). The 321 different LCFA isomers were analyzed as one, because isomers co-eluted or were at least not 322 well separated on GC (Fig. 2). However, we could assign separate isotope values for (a)i-323 C 27:2 and Me-C 26:2 (Fig. 4). The LCFAs showed less isotopic variation compared to bacterial 324 FAs, but still ranged between -25 and -19 ‰ (5-95 percentile) (Fig. 4, Table 2). Me-C   Fig. 4: δ 13 C composition of precursors and dominant LCFAs in analyzed demosponges. 334 Sponge species were pooled together, and the median is indicated in the box plot as black . CC-BY 4.0 International license made available under a (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 The copyright holder for this preprint this version posted October 9, 2020. showed that Poribacteria are a prominent phylum in G. barretti (46,47), but are rare or even . CC-BY 4.0 International license made available under a (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 The copyright holder for this preprint this version posted October 9, 2020. ; https://doi.org/10.1101/2020.10.09.332833 doi: bioRxiv preprint absent in G. hentscheli (48), which shows a dominance of Acidobacteria, Chloroflexi, and 360 Proteobacteria, phyla that are also abundant in G. barretti (46,47). This suggests that either 361 the MBFAs belong to one of the above-mentioned phyla, or that the MBFAs are shared 362 among microbial phyla, as their chemotaxonomic resolution is lower compared to genomic 363 analysis. In the environment, MBFAs are primarily found in nitrogen and sulfur reducers 364 (chemoheterotrophs) and oxidizers (chemoautotrophs) that are mostly members of the (large) 365 proteobacteria family (49-52). Nitrogen and sulfur reduction and oxidation processes are 366 conducted in deep-sea sponges such as G. barretti (26,53,54), and oxidation processes are 367 coupled to CO 2 fixation, although associated CO 2 fixation is likely to contribute < 10 % of 368 the carbon demand of deep-sea sponges (55). The poribacteria in sponges were also 369 characterized as mixotrophic bacteria, able to fix CO 2 using the ancient Wood-Ljungdahl 370 (reversed acetyl-CoA) pathway (56). The isotopic enrichment in MBFAs (Fig. 4, Table 2), 371 agrees with earlier observations for G. barretti (57), and might thus be linked to nitrogen and 372 sulfur transforming processes and potentially CO 2 fixation. It will be interesting to perform 373 an isotope-tracer study (55) with 13 C-CO 2 to assess CO 2 incorporation in the abundant 374 MBFAs, perhaps combined with nitrification (or sulfur oxidation) inhibitors, similar to 375 Veuger et al. (58). 376 The most depleted FA (i-C 17:1 ω7, Fig. 4, Table 2 (Fig. 4, Table 2). Such values can be the result of isotopic averages 382 from different pathways, since they are more general bacterial markers, or they might . CC-BY 4.0 International license made available under a (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 The copyright holder for this preprint this version posted October 9, 2020. ; https://doi.org/10.1101/2020.10.09.332833 doi: bioRxiv preprint represent general heterotrophy on organic matter with δ 13 C value from -24 to -22 ‰ in the 384 western Arctic (60). 385 The low contribution of FAs with a chain length of C 20 to C 24 typical of 386 phytoplankton and zooplankton (e.g. C 20:5 ω3 and C 22:6 ω3) indicates that sinking zoo-and . CC-BY 4.0 International license made available under a (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 The copyright holder for this preprint this version posted October 9, 2020. ; https://doi.org/10.1101/2020.10.09.332833 doi: bioRxiv preprint and/or i-C 27:2 Δ 5,9 , present in all species analyzed (Fig. 5, Table 1), are common LCFAs of 409 demosponges, (e.g. (17,38,65), for an overview see (8)). We found (mid-)Me-branched Δ 5,9 410 LCFAs in all species, except G. parva (Fig. 5, Table 1) (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 The copyright holder for this preprint this version posted October 9, 2020. ; https://doi.org/10.1101/2020.10.09.332833 doi: bioRxiv preprint analysis provides an additional method to identify each species. Furthermore, the distinct 434 LCFAs could be useful as trophic markers to study the ecological role of deep-sea sponges in 435 the environment. No geographical differences in LCFA composition of Arctic Tetractinellids 436 were found (Table S1) suggesting that the environment has a limited influence on the LCFA 437 composition, which is a prerequisite for using LCFA as chemotaxonomic markers. 2 C atoms at the carboxylic acid end and desaturate at Δ 5 and Δ 9 (visualized in (10)), 444 revealing C 16:0 as precursor for the common C 26:2 Δ 5,9 , while C 16:1 ω7 was identified as 445 precursors for C 26:2 Δ 9,19 (Fig. 5). There is no evidence for branching to be introduced by 446 sponges, so i-and a-C 15:0 were identified as precursors of i-and a-C 27:2 Δ 5,9 (Fig. 5) (38), 447 while Me-C 16:0 has been identified as precursor for Me-C 26:2 Δ 5,9 and Me-C 28:2 Δ 5,9 (Fig. 5) 448 (13,16,70). Finally, we hypothesize that i-C 17:1 ω7 is the precursor for i-C 25:1 ω7 and i-449 C 27:2 Δ 9,19 found in G. atlantica, G. barretti, and G. hentscheli (Fig. 5). 450 Application to the present study showed that most LCFAs could be linked to 451 precursors via established pathways, with hypothetical intermediates since hardly any were 452 found in detectable abundance (Fig. 5). The C isotopic differences in bacterial precursors 453 were (partially) reflected in C isotopic composition of LCFAs (Fig. 4, Table 2), although the 454 differences were not as prominent in LCFAs compared to their precursors. One explanation is 455 that a mixture of C sources is used by the host to elongate precursors to LCFAs, while also 456 methodological aspects might contribute. A (much) longer analytical column might help 457 improving separation of LCFAs.
. CC-BY 4.0 International license made available under a (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 The copyright holder for this preprint this version posted October 9, 2020. ; https://doi.org/10.1101/2020.10.09.332833 doi: bioRxiv preprint  (Fig. 5). However, the typical Δ 5,9 unsaturation is more convenient to 469 show with Δ annotation, as an ω notation would alter with varying C chain length (Fig. 5). 470 Ambiguity arises in ω notation of methyl-branching, because a(i) notation is used for 471 terminally branched FAs and describes total C atoms (including the methyl group(s)), while 472 Me notation is used for MBFAs and describes the C number of the backbone (excluding 473 methyl group (s)) and counts the position of the branching from the carboxylic acid end (and 474 not the terminal (ω) end, Fig. 1). This might lead to confusion about the total C number, 475 which is needed to correct measured isotope values for the extra methyl group, and about the 476 ω position of unsaturation (start counting from the end of the backbone, excluding the 477 methyl-group) and the conversion from ω to Δ notation (Fig. 1). We added this discussion to 478 create awareness and would like to recommend including a description of the notation in the 479 methods and presenting both nomenclature when a mixture of notation styles is used. (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 The copyright holder for this preprint this version posted October 9, 2020.  . CC-BY 4.0 International license made available under a (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 The copyright holder for this preprint this version posted October 9, 2020.  . CC-BY 4.0 International license made available under a (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 The copyright holder for this preprint this version posted October 9, 2020. (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 The copyright holder for this preprint this version posted October 9, 2020. ; https://doi.org/10.1101/2020.10.09.332833 doi: bioRxiv preprint  . CC-BY 4.0 International license made available under a (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 The copyright holder for this preprint this version posted October 9, 2020. ; https://doi.org/10.1101/2020.10.09.332833 doi: bioRxiv preprint profiles of DIC, DOC, and POC. Biogeosciences. 2012;9(3):1217-24.
. CC-BY 4.0 International license made available under a (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 The copyright holder for this preprint this version posted October 9, 2020. ; https://doi.org/10.1101/2020.10.09.332833 doi: bioRxiv preprint al. A new cytotoxic fatty acid (5Z,9Z)-22-methyl-5,9-tetracosadienoic acid and the . CC-BY 4.0 International license made available under a (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 The copyright holder for this preprint this version posted October 9, 2020. . CC-BY 4.0 International license made available under a (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 The copyright holder for this preprint this version posted October 9, 2020. ; https://doi.org/10.1101/2020.10.09.332833 doi: bioRxiv preprint