In Situ Capture and Real-Time Enrichment of Marine Chemical Diversity

Analyzing the chemical composition of seawater to understand its influence on ecosystem functions is a long-lasting challenge due to the inherent complexity and dynamic nature of marine environments. Describing the intricate chemistry of seawater requires optimal in situ sampling. Here is presented a novel underwater hand-held solid-phase extraction device, I-SMEL (In Situ Marine moleculELogger), which aims to concentrate diluted molecules from large volumes of seawater in a delimited zone targeting keystone benthic species. Marine benthic holobionts, such as sponges, can impact the chemical composition of their surroundings possibly through the production and release of their specialized metabolites, hence termed exometabolites (EMs). I-SMEL was deployed in a sponge-dominated Mediterranean ecosystem at a 15 m depth. Untargeted MS-based metabolomics was performed on enriched EM extracts and showed (1) the chemical diversity of enriched seawater metabolites and (2) reproducible recovery and enrichment of specialized sponge EMs such as aerothionin, demethylfurospongin-4, and longamide B methyl ester. These EMs constitute the chemical identity of each targeted species: Aplysina cavernicola, Spongia officinalis, and Agelas oroides, respectively. I-SMEL concentrated sponge EMs from 10 L of water in a 10 min sampling time. The present proof of concept with I-SMEL opens new research perspectives in marine chemical ecology and sets the stage for further sustainable efforts in natural product chemistry.

(a) Schematic view of the main items within I-SMEL: identification of the instrument items: 1 electronic controller of the pump, 2 peristaltic pump, 3 valves, 4 SPE supports, 5 pump outlet where a flexible container can be adapted, 6 the battery, (b) zoom in the upper part of I-SMEL containing the peristaltic pump, its controller and the battery in stainless steel enclosures, (c) closer overview on the entire instrument divided in two parts: the upper one with all the electronics and the pump outlet, and the lower one, where the SPE supports are tightly screwed and turned toward the bottom, thus toward the organisms to sample, (c-f) view inside the chamber, where the SPE supports have to be fixed one by one, (e) upside view of I-SMEL with the battery, the waterproof magnetic push button activating the pump and the three valves that can be maintained closed or opened allowing the seawater to go through a selected SPE disk.

Notes
The instrument is made of six primary components as shown here and in figure 1 of the manuscript.The dimensions of the capture chamber can be adjusted according to the research objectives or the target organism.Similarly, the peristaltic pump's specificity (e.g.flow rate) can be tailored to match the intended purpose.The sizes of the two stainless steel enclosures that house respectively the battery and electronic controller are also contingent upon the desired level of autonomy.These two enclosures, along with the electronic controller, were constructed in-house.  .For all those reasons, we chose DVB as a polymeric phase.

S1.2. Description of the SPE supports
The disks were preferred compared to the cartridges as they offer a larger surface of exchange, allowing faster flow rate to be applied, compatible with shorter sampling time.Also, in general, SPE disks are more appropriate when the water to be filtered is charged with different types of particulates.The disks were chosen instead of resins as they are, for now, easier to use and to change underwater and more importantly, compatible with a standardized SPE elution/extraction in the laboratory using automated SPE instruments.
As I-SMEL encloses three distinct SPE supports, it would be possible to perform in situ enrichment of exometabolites using different, yet complementary, SPE matrices (e.g.HLB and DVB).3b.

S4. Complement to Figure 3: Marine Chemodiversity through the analysis of FBMN
A total of 764 FT was attributed to different sample groups.Out of them, 119 features were found to be detected in two or three sponge EMs, and did not cluster with any crude extracts or ACS samples.These features were not included in the representation of feature distribution as they were not specific to a given sponge community.See Excel spreadsheet 103 exomet_fig3R1vf_MN on ZENODO at https://doi.org/10.5281/zenodo.7820941.

S6. MS data processing and annotated MS² spectra of reproducibly detected EMs
All MS 2 data were acquired with a Bruker Impact II (ESI-Q-ToF) using positive electrospray ionization (ESI+) as follows: nebulizer gas N 2 at 3.5 bar; dry gas at 12 L.min -1 , capillary temperature at 200°C and voltage at 4500 V. MS/MS acquisition mode was set with a scan rate of 8 Hz (full scan from 50-1200 m/z), and a mixed collision energy (CE) at 20-40 eV (50% time at each collision energy, stepping mode).

UHPLC-MS data processing
Following their calibration, the acquired MS data were converted to the open format *.mzXML using MSConvert (Proteowizard) 7 and further processed on MZmine 3.2.8 8                The proposed structure for M367T14.3 is in agreement with previously reported data. 12The corresponding cluster index is between brackets (see also table below).

Figure S1. 2 .S1. 3 .Figure S1. 3 .
Figure S1.2.Illustration of how the SPE disks (47 mm diameter) are installed inside the SPE supports, which are made of 4 different parts to assemble.Two PTFE cases and two stainless steel grids of different mesh.The grid on top of the SPE disk has a 2 mm mesh which retains the coarsest particles (4).The preconditioned SPE disk is placed inside the holder prior to the dive.All four parts enclosing the disk are maintained tightly together by outside screws(5, 6).

S2.Figure S2. 1 .
Figure S2.1.Sampling protocol used in this study for EXP1 to EXP3.For EXP1 and 2: one replicate experiment corresponds to the filtration of a total of 2 L of seawater repeated above either 5 different community (EXP1) or 5 different sponge specimens (EXP2).During the sampling all three valves are opened.Therefore, for one replicate a total of 10 L of seawater is passed through 3 SPE disks simultaneously.The three disks are then eluted individually then pooled to provide an average extract of marine chemical seascape (n = 1, EXP1) or an average extract of cumulated sponge EMs (n = 1, EXP2).Both EXP1 and 2 were replicated three times, resulting in 3 replicate extracts for each EXP.➢ 1 replicate experiment = one average extract from 3 SPE disks simultaneously from 5 different substrates.In situ sampling duration 5 x 2 min = sampling volume 10 L (5 x 2 L).

For EXP3 :
one replicate experiment corresponds to the filtration of 10 L of seawater above one sponge individual, on three SPE disks simultaneously.After individual elution of these disks, the solvents are combined to obtain one single extract enriched in individual sponge EMs.EXP3 was replicated three times to obtain 3 individual EM extracts/sponge species.➢ 1 replicate experiment = one average extract from 3 SPE disks simultaneously from 1 single sponge individual.In situ sampling duration 1 x 10 min = sampling volume 10 L.

Figure S2. 2 .
Figure S2.2.depicts three SPE disks from in situ capture ready to be eluted.Each elution provides a 30 mL extract/disk.These extracts are then pooled and dried to constitute one replicate sample, which is diluted in 1.5 mL MeOH MS-grade prior to LC-MS analysis.

S7. 4 .Figure S7. 4 .
Figure S7.4.Stacked 1 H NMR spectra of all EM extracts from EXP2.The figure shows that each averaged sponge EM extract showed unique 1 H NMR fingerprints.Characteristic NMR signals corresponding to sponge specialized metabolites were observed for A. cavernicola and S. officinalis.For both sponges, several metabolites with a common core structure were identified in all EM extracts.Consequently, these metabolites collectively contributed to detectable 1 H signals in the NMR spectrum.For A. oroides, the only detection of longamide B methyl ester as reproducible EM poses a challenge for its detection by NMR.
Peristaltic dosing pump: ERDEMIL TM coupled to a Maxon TM microreductor motor Battery: 12 V, 250 mA rechargeable NiMH battery Controller: Pic16F electronic card Connecting tubing: flexible PTFE tubing 8/11 mm diameter.I-SMEL chamber is delimited by plexiglass.All the other parts are made of PTFE.
5mong the different polymeric solid phases, DVB is the most used to enrich water-soluble natural products(Berlinck etal.2021) 1 because it is able to retain structurally diverse dissolved metabolites of wide polarity range.As such, DVB solid phase (e.g.Bond-Elut-PPL) is used to capture Dissolved Organic Matter (DOM) from filtered seawater (Dittmar et al. 2008, Petras et al. 2017) 2,3 .DVB resins have also been previously deployed in device aiming at passively sampling microalgae toxins in situ (SPATT Solid Phase Adsorption Toxin Tracking [Roué et al. 2018])4.More recently, DVB resin (HP-20) lead to the isolation of metabolite directly enriched in the ecosystem(Bogdanov et al. 2023)5.Finally, previous work focusing on collecting EMs released by sponge in aquarium used DVB polymeric SPE(Vlachou et al. 2018) S2.3.Notes on the rationale behind the choice of SPE DVB disksDVB polymeric matrix:

Table S3 .1. Mass of eluted EM extracts from EXP1 and 2
* We had an experimental mishap with the third replicate and did not measure the extraction yields.EXP: type of Experiment; CAPT: in situ capture replicate, n = 3, R: technical Replicate of DVB disk.Each capture was performed on the three DVB disks simultaneously.ACS = Average Chemical Seascape, SO = Spongia officinalis, AC = Aplysina cavernicola, AO = Agelas oroides.

Table S4 .2. Distribution a of features (FT) or nodes in the MN between identified groups
bValues used in figure

Table S4 .3. Evaluation of molecular diversity within the classified features Number of features per sample type
A total of 744 features were annotated using SIRIUS, 488 of them had a Natural Product (NP) pathway probability score > 0.8 and 309 of them (41%) were specific to each group as represented here.

Table S6 .1. MS data of identified or annotated exometabolites
found in Aplysina cavernicola crude extracts and reproducibly detected as exometabolites

Table S6 .2. MS data of putatively identified exometabolites
found in Spongia officinalis crude extract and reproducibly detected as exometabolite The Cluster index identifies nodes in the GNPS molecular network, the most intense peak of the isotopic pattern was used for feature codification.bMajor MS² fragments and their intensities were selected using the GNPS Metabolomics USI tool.cConfidence level of metabolite identification according toSchymanski et al. 2014.

Table S6 .3. MS data of putatively identified exometabolites
The Cluster index identifies nodes in the GNPS molecular network, the most intense peak of the isotopic pattern was used for feature codification.b The monoisotopic peak was selected for molecular formula determination.c Major MS² fragments and their intensities were selected using the GNPS Metabolomics USI tool.d Confidence level of metabolite identification according to Schymanski et al. 2014. 9NMR acquisition parameters were set up as follows for all EM extracts: found in Agelas oroides crude extract and reproducibly detected as exometabolite a