Deposit feeding of a foraminifera from an Arctic methane seep site 1 and possible association with a methanotroph revealed by 2 transmission electron microscopy 3

LPG, Laboratoire de Planétologie et de Géodynamique, Univ. Angers, Université de Nantes, CNRS, LPG, SFR 6 QUASAV, Angers, 49000, France 7 CAGE, Centre for Arctic Gas Hydrate, Environment and Climate, UiT, The Arctic University of Norway, Tromsø, 8 9010, Norway 9 ZMT, Leibniz Centre for Tropical Marine Research, Bremen, 28359, Germany 10 Woods Hole Oceanographic Institution, Geology & Geophysics Department, Woods Hole, 02543, MA, USA 11 Cell and Plant Physiology Laboratory, CNRS, CEA, INRAE, IRIG, Université Grenoble Alpes, Grenoble, 38054 12 France 13 Department of Arctic and Marine Biology, UiT, The Arctic University of Norway, Tromsø, 9037, Norway 14 Université de Nantes, CNRS, Institut des Matériaux Jean Rouxel, IMN, Nantes, 44000 France 15

Abstract. Several foraminifera are deposit feeders that consume organic detritus (dead particulate 20 organic material along with entrained bacteria). However, the role of such foraminifera in the 21 benthic food-web remains understudied. As foraminifera may associate with methanotrophic 22 bacteria, which are 13 C-depleted, feeding on them has been suggested to cause negative δ 13 C values 23 in the foraminiferal cytoplasm and/or calcite. To test whether the foraminiferal diet includes 24 methanotrophs, we performed a short-term (1 d) feeding experiment with Nonionellina 25 labradorica from an active Arctic methane-emission site (Storfjordrenna, Barents Sea) using the 26 marine methanotroph Methyloprofundus sedimenti, and analyzed N. labradorica cytology via 27 Transmission Electron microscopy (TEM). We hypothesized that M. sedimenti would be visible, 28 as evidenced by their ultrastructure, in degradation vacuoles after this feeding experiment. 29 Sediment grains (mostly clay) occurred inside one or several degradation vacuoles in all 30 foraminifers. In 24% of the specimens from the feeding experiment degradation vacuoles also 31 contained bacteria, although none could be confirmed to be the offered M. sedimenti. Observations 32 of the area adjacent to the aperture after 20 h incubation revealed three putative methanotrophs, 33 close to clay particles. These methanotrophs were identified based on internal characteristics such 34 as a type I stacked intracytoplasmic membranes (ICM), storage granules (SG) and gram-negative 35 cell walls (GNCW). Furthermore, N. labradorica specimens were examined for specific 36 adaptations to this active Arctic methane-emission site; we noted the absence of bacterial 37 endobionts in all specimens examined but confirmed the presence of kleptoplasts, which were 38 often partially degraded. Based on these observations, we suggest that M. sedimenti can be 39 consumed by N. labradorica via untargeted grazing in seeps and that N. labradorica can be 40 generally classified as a deposit feeder at this Arctic site. These results suggest that if 41 methanothrophs are available to the foraminifera in their habitat, their non-selective uptake could 42 make a substantial contribution to altering δ 13 Ctest values. This in turn may impact metazoans 43 grazing on benthic foraminifera by altering their δ 13 C signature. 44 45 46 benthic foraminiferafeeding experimentgrazing -marine methanotrophs -Arctic methane 47 seeps-transmission electron microscopyultrastructurekleptoplasts-protistmolecular 48 identification 49

Introduction 50
In methane seep sites, the upward migration of methane affects the pore-water chemistry of near-51 surface sediments, where benthic foraminifera inhabiting the sediment interface have been shown 52 to live (e.g. Dessandier et al., 2019). Extremely light isotopic signals of δ 13 C have been measured 53 in seep-associated foraminiferal calcite tests (Wefer et al., 1994;Rathburn et al., 2003;Hill et al., 54 2004b;Panieri et al., 2014). One explanation of low δ 13 C signals in foraminifera could be due to 55 the ingestion of 13 C-depleted methanotrophs (Mccorkle et al., 1990;Wefer et al., 1994;Rathburn 56 et al., 2003;Panieri, 2006). Recently, specimens of the foraminifer Melonis barleeanus 57 (Williamson, 1858) collected from an active methane seep site were closely associated with 58 putative methanotrophs at their apertural region (Bernhard and Panieri, 2018). 59 The observation by Bernhard and Panieri (2018) brought to light the need to examine feeding 60 habits of foraminifera living on or around methane seeps. The species M. barleeanus could feed 61 on aerobic methane-oxidizing bacteria (methanothrophs), which are abundant in the water column 62 around methane seeps (Tavormina et al., 2010). Methanotrophs produce the biomarker diplopterol, 63 which has an extremely light δ 13 C signature (− 60 ‰) and makes methanotrophs isotopically very 64 light themselves (Hinrichs et al., 2003 hose was used to sample the upper most surface layer (0-1 cm). The wet sediment was collected 141 in petri dishes and wet sieved to a size range of 250-500 μm, which served as source of living 142 (cytoplasm containing) foraminifera. The species N. labradorica, which was the visibilly 143 abundant, was subsequently used for feeding experiments described in detail below.

Geochemistry 148
For geochemical analysis a push corer (PUC2) was used (referred to as geochemistry core) to 149 obtain measurements on δ 13 CDIC and sulfate, as blade corer (BLC18) did not allow those 150 measurements. PUC2 was taken in close vicinity to BLC18, ~5m apart (Fig 1). Pore-water samples 151 were taken from PUC2 using rhizons that were inserted through pre-drilled holes in the core tube 152 at intervals of 1 cm. Acid washed 20-ml syringes were attached to the rhizons for pore water 153 collection. Depending on the amount of pore water collected, the samples were split for δ 13 CDIC 154 and sulfate measurements. To the samples 10 µL of saturated HgCl2 (aq) was added to stop 155 microbial activity, and stored in cold conditions (5°C). δ 13 CDIC was determined using a 156 ThermoScientific Gasbench II coupled to a ThermoScientific MAT 253 IRMS at the Stable Isotope 157 Laboratory (SIL) at CAGE, UiT. Anhydrous phosphoric acid was added to small glass vials 158 (volume 4.5 mL), that were closed and flushed with helium 5.0 gas before the pore water sub-159 sample was measured. A pore-water sub-sample (volume 0.5 mL) was then added through the 160 septa with a syringe, followed by equilibration for 24 h at 24°C to liberate the CO2 gas. Three solid 161 calcite standards with a range of +2 to -49 ‰ were used for normalization to δ 13 C -VPDB. 162 Correction of measured δ 13 C by -0.1 ‰, was done to account for fractionation between (g) and 163 (aq) in sample vials. Instrument precision for δ 13 C on a MAT253 IRMS was 1σ +/-0.1 ‰. Sulfate 164 was measured with a Metrohm ion chromatography instrument equipped with column Metrosep 165 A sup 4, and eluted with 1.8 mmol/L Na2CO3 + 1.7mmol/L NaHCO3 at the University of Bergen. (NMS) and sterile filtered sea water using 125-mL Wheaton ® serum bottles with butyl septa and 171 aluminum crimp caps (Teknolab ® ). Methane was injected to give a headspace of 20% methane in 172 air, and the bottles were incubated without shaking at 15°C in darkness. Purity of the cultures and 173 cell integrity was verified by microscopy and by absence of growth on agar plates with a general 174 medium for heterotrophic bacteria (tryptone, yeast extract, glucose and agar). 175 Transmission Electron Microscopy was performed on culture aliquots to allow morphological 176 comparison to previously published work (Tavormina et al., 2015). Methyloprofundus sedimenti 177 strain PKF-14 cells have a gram-negative cell wall, coccoid to slightly elongated shape and characteristic stacked intracytoplasmic membrane (ISM) and storage granules (SG) (Fig 2c). 179 Additionally,16S rRNA gene sequencing was performed (data not shown) to confirm it to be 180 similar to the published Methyloprofundus sedimenti (Tavormina et al., 2015). 181

Experimental setup 182
On the ship, Nonionellina labradorica ( Fig. 2) specimens showing a dark greenish brown 183 cytoplasm were picked using sable artist brushes under a stereomicroscope immediately after wet 184 sieving the sediment using natural seawater delivered from the ship pump. Living specimens had 185 a partly inorganic covering surrounding the test, which was gently removed using fine artist 186 brushes. Another Nonionellidae, Nonionella iridea, was similarly embedded with a cyst / covering 187 in sediment 188 Our specimens were subsequently rinsed twice in filtered artificial seawater to remove any 189 sediment before placing them into the experimental petri dishes. Care was taken that those were 190 minimally exposed to light during preparation of the experiment, as kleptoplasts are known to be Five foraminifera, which served as initial/field specimens (Table 1) incubation duration influenced response of the foraminifera to the methanotroph. One petri dish 206 containing five foraminifera, which were un-fed and fixed at 20 h, served as a negative "control". 207 After the end of the respective incubation times, each foraminifer was picked with a sterilized fine 208 artist brush, which was cleaned in 70% ethanol between each specimen, and placed individually 209 into a fixative solution (4% glutaraldehyde and 2% paraformaldehyde dissolved in ASW). 210

Ultrastructure of methanotroph culture used in the feeding experiment 266
Metyloprofundus sedimenti is characterized by a typical type I intracellular stacked membrane 267 (ISM), storage granules (SG) and typical gram-negative cell wall (GNCW) (Fig. 2). These features 268 were used to identify M. sediment. 269 270

General ultrastructure 272
All 17 specimens were considered living at the time of observation (Fig. 3), as the mitochondria 273 had characteristic double membranes and occasionally visible cristae (Nomaki et al., 2016). 274 Cytoplasm exhibited several vacuoles and kleptoplasts concentrated in the youngest chambers 275 (Fig. 3a) and, in some specimens, the nucleus with nucleoli was visible (Fig. 3b). Kleptoplasts 276 were numerous throughout the cytoplasm and occurred in the form of a single chloroplast (Fig.  277 3a-b), or as double chloroplasts (Fig. S2). Not all kleptoplasts were intact, some showed peripheral 278 degradation of the membranes indicated by an increasing number of white areas between pyrenoid, 279 lamella and thylakoids (Fig. S2). Peroxisomes in N. labradorica occurred mostly as pairs (Fig. 3c) 280 or small clusters of 3-4 spherical organelles (Fig. S1a-b). The mitochondria occurred often in small 281 clusters of two to five throughout the cytoplasm and were oval, round or kidney-shaped in cross 282 section (Fig. 3e-f). Sometimes, but not always, peroxisomes were associated with endoplasmic 283 reticulum ( Fig. S1c) but could also occur alone. Golgi apparatus (Fig 3d) had intact membranes, 284 often occurring near mitochondria. 285

Ultrastructure of aperture-associated bacteria 287 288
In total three putative methanotrophs were identified in the vicinity of two foraminifer specimens 289 (sample E39, Fig. 4; E37, Fig. 5). Those were identified next to reticulopodial remains in the cross-290 section (Fig. 4b). As an aid for identification of M. sedimenti we used the characteristics shown in 291 the literature (Tavormina et al. 2015) and a our own TEM observation obtained from M. sedimenti 292 culture (Fig. 2c). As noted, Methyloprofundus sedimenti is characterized by a typical type I 293 intracellular stacked membrane (ISM), storage granules (SG) and typical gram-negative cell wall 294 (GNCW) (Fig. 2). On specimen E39 from the 20 h treatment, we found the methanotroph 295 exhibiting the clearest internal structure, having both typical type I stacked intracytoplasmic 296 membranes (ICM+SG) and a second putative methanotroph showing SG+GNCW (Fig. 4). 297 Specimen E36, from the 20 h treatment, hosted another putative methanotroph showing three large 298 SG (Fig. 5). Storage granules occur throught this putative methanotroph (Fig. 5c).     The sediment grains are easy to recognize in the TEM image as angular grains spiking out of the 310 vacuoles, next to organic debris, which can have many different shapes. Each specimen had at 311 least one degradation vacuole with sediment particles present (Table 1). If a sediment particle was 312 visible, the vacuole was defined as a degradation vacuole (dv), and if it was not then it was defined 313 as a standard vacuole (v) (Fig. 6). Sediment particles are likely the remains of clay grains from the 314 seafloor, and hence show that the vacuole must contain cell foreign objects, around which 315 degradation processes have started. Next to sediment particles, 4 out of 17 specimens examined 316 (23%) had a few bacteria of various sizes inside their degradation vacuoles (Fig 6 b-c). 317 : 2 µm, b,c: 0.5 µm.

Foraminiferal genetics 319
Six of 13 specimens analyzed for genetics were positively amplified and sequenced (Fig. S3). The  . It has also been suggested that foraminifera may 363 sometimes be transported into seeps and can also occur at tze SMTZ, but they likely not live in 364 those sediment layers permanently (Bernhard and Bowser, 1999).  (Pascal et al., 2008) and is a 379 logical consequence from detritus feeding. Certain foraminifera have been shown to selectively 380 ingest algae/bacteria according to strain (Lee et al., 1966;Lee and Muller, 1973). From laboratory 381 cultures we know that several foraminifera cultures require bacteria to reproduce, as antibiotics 382 inhibited reproduction (Muller and Lee, 1969). Future studies will need to employ additionally 383 molecular tools to additionally determine the food contents inside the cytoplasm (e.g. (Salonen et 384 al., 2019). A recent study by used metabarcoding to assess the contribution of bacterial OTUs 385 associated with intertidal foraminifera, and revealed that Ammonia sp. T6 can predateon metazoan 386 taxa, whereas Elphidium sp. S5 and Haynesina sp. S16 are more likely to ingest diatoma 387