DNA-stable isotope probing (DNA-SIP) identifies marine sponge-associated bacteria actively utilizing dissolved organic matter (DOM).

Sponges possess exceptionally diverse associated microbial communities and play a major role in (re)cycling of dissolved organic matter (DOM) in marine ecosystems. Linking sponge-associated community structure with DOM utilization is essential to understand host-microbe interactions in the uptake, processing, and exchange of resources. We coupled, for the first time, DNA-stable isotope probing (DNA-SIP) with 16S rRNA amplicon sequencing in a sponge holobiont to identify which symbiotic bacterial taxa are metabolically active in DOM uptake. Parallel incubation experiments with the sponge Plakortis angulospiculatus were amended with equimolar quantities of unlabelled (12 C) and labelled (13 C) DOM. Seven bacterial amplicon sequence variants (ASVs), belonging to the phyla PAUC34f, Proteobacteria, Poribacteria, Nitrospirae, and Chloroflexi, were identified as the first active consumers of DOM. Our results support the predictions that PAUC34f, Poribacteria, and Chloroflexi are capable of organic matter degradation through heterotrophic carbon metabolism, while Nitrospirae may have a potential mixotrophic metabolism. We present a new analytical application of DNA-SIP to detect substrate incorporation into a marine holobiont with a complex associated bacterial community and provide new experimental evidence that links the identity of diverse sponge-associated bacteria to the consumption of DOM.

Introduction

Sponges are among the oldest extant metazoans and a unique model system to study early metazoan–microbe interactions because of their exceptionally diverse associated microbial communities (Thomas et al., 2016; Moitinho‐Silva et al., 2017a). Sponge holobionts, consisting of the sponge host and its associated microbiota, are important aquatic ecosystem drivers because of their efficient filtering system, their interactions with surrounding organisms and within the holobiont itself in terms of nutrient cycling and providing habitat (Webster and Taylor, 2012; Webster and Thomas, 2016; de Goeij et al., 2017; Pita et al., 2018). In the oceans, primary producers, such as phytoplankton, macroalgae, and dinoflagellates in corals release the product of their photosynthetic activity in the form of dissolved organic matter (DOM), which is the main pool of organic matter in seawater and an important food source for sponges (Tanaka et al., 2011; de Goeij et al., 2013; Thornton, 2014; Hansell and Carlson, 2015; Rix et al., 2017; Bart et al., 2020). In fact, dissolved organic carbon (DOC) – a component of the DOM pool – can contribute to 60%–97% of the daily carbon intake of sponges (Yahel et al., 2003; de Goeij et al., 2008; Mueller et al., 2014; Hoer et al., 2018; McMurray et al., 2018). Sponges consume DOM and release particulate organic matter as detritus, which is readily available to higher trophic levels (de Goeij et al., 2013) or stored as biomass, which is hypothesized to be predated upon by spongivores (McMurray et al., 2018). This so‐called sponge loop is of high ecological relevance in tropical coral reefs where sponges are estimated to cycle DOC in the same order of magnitude as the primary production rates of the entire ecosystems (de Goeij et al., 2013). In the deep‐sea, sponges are also very abundant, forming large sponge‐dominated ecosystems called sponge grounds (Maldonado et al., 2017), or as part of cold‐water coral reefs (Cathalot et al., 2015). The processing of DOM by deep‐sea sponges has been experimentally confirmed (Rix et al., 2016; Leys et al., 2018; Maier et al., 2020; Bart et al., 2020; Bart et al., 2021), but the ecological relevance is not yet established.

Organic and inorganic nutrient uptake (from here on collectively called ‘nutrients’) is believed to occur through a symbiotic relationship within the sponge holobiont with reciprocal translocation of resources: the sponge microbiota can benefit by a supply of nutrients, such as ammonia, released by the host, while the host can profit from the internal nutrient supply performed by the diverse microbial metabolism (Taylor et al., 2007; Hentschel et al., 2012; Webster and Taylor, 2012; Zhang et al., 2019). Although the proposed symbiotic functions are extensive and mostly described for photoautotrophic nutrient cycling within the sponge holobiont (Wilkinson, 1983; Steindler et al., 2002; Erwin and Thacker, 2008; Weisz et al., 2010; Freeman and Thacker, 2011; Fiore et al., 2013), examples of heterotrophic interaction (e.g., the uptake and cycling of organic matter) between sponge host cells and microbial symbionts are scarce. Recent nanoscale secondary ion mass spectrometry (nanoSIMS) studies have visualized and quantified DOM assimilation by the sponge host cells and associated symbionts (Achlatis et al., 2019; Rix et al., 2020) along with the transfer of carbon and nitrogen within the holobiont (Hudspith et al., 2021). Both sponge holobionts with high and low abundances of microbial symbionts were found to assimilate DOM, with different relative contributions of host and symbionts. Whereas microbial symbionts extensively assimilated DOM, the filter cells (choanocytes) of the sponge host also readily take up DOM (de Goeij et al., 2009; Achlatis et al., 2019; Rix et al., 2020) and even translocation of the assimilated DOM from the host cells to the symbionts was found (Hudspith et al., 2021). However, the identity of the microbial groups actively involved in DOM uptake and processing within the sponge holobiont remains unresolved.

In recent years, environmental microbiology has taken a step further in linking metabolic activity with the phylogenetic identity of uncultivated microorganisms through the development of stable isotope‐based techniques coupled with next generation sequencing. The principle of stable isotope probing (SIP) is to amend an experimental incubation with a substrate labelled with a heavy stable isotope (e.g., 13C or 15N) and track the labelled compounds into cellular biomass components, like DNA or RNA (Radajewski et al., 2000; Dumont and Murrell, 2005; Whiteley et al., 2007; Neufeld et al., 2007a). A long ultracentrifugation step (36–65 h) creates a density gradient along which the cellular components labelled by the heavy isotope can be isolated into multiple fractions and sequenced, thus linking the identity of the organisms to the uptake of specific substrates or to specific metabolic functions (Radajewski et al., 2000; Neufeld et al., 2007a). In the marine environment, SIP has been used to identify bacterioplankton taxa incorporating readily bioavailable DOM as opposed to the refractory component of DOM, which generally comprises 75%–90% of the DOM pool (Nelson and Carlson, 2012; Bryson et al., 2017; Liu et al., 2020). SIP is therefore a promising approach to track direct incorporation of DOM into the biomass of responding microbial populations and to understand DOM metabolism in the sponge microbiota. Here, we conducted for the first time a DNA‐SIP experiment in the sponge Plakortis angulospiculatus (Porifera, Homoscleromorpha) to investigate which of the associated bacterial taxa can actively incorporate DOM. The species P. angulospiculatus was chosen because it is widespread across the Caribbean and contains a high abundance of associated microbes (Hudspith et al., 2021). Seven bacterial amplicon sequence variants (ASVs) belonging to the phyla PAUC34f, Proteobacteria, Poribacteria, Nitrospirae, and Chloroflexi were identified as active DOM processors in P. angulospiculatus, therefore providing new experimental evidence of their ecological roles in the sponge holobiont.

Experimental procedures

Collection and maintenance of sponges

Individuals of the sponge Plakortis angulospiculatus were collected by SCUBA diving from between 5 and 30 m water depth on the fringing reef close to Piscadera Bay on Curaçao (12°12' N, 68°96' W). All experimental work was conducted at the CARMABI Research Station between May and August 2018. Collection and experiments were performed under the research permit (#2012/48584) issued by the Curaçaoan Ministry of Health, Environment and Nature (GMN) to the CARMABI Foundation. After collection, sponge individuals were trimmed to sizes of 8–10 cm3 with at least two functioning oscula (i.e., outflow opening; active pumping tested with fluorescent dye) and cleared from epibionts. The sponges were immediately transported without air exposure to the CARMABI aquarium facilities and kept in 100 l flow‐through aquaria, with a flow rate of 3 l min−1, supplied with seawater pumped directly from the reef at 10 m water depth. All individuals were allowed to recover from tissue damage caused by the initial trimming and to acclimatize in the aquaria up to 4 weeks prior to incubation experiments (Alexander et al., 2015). Only visually healthy individuals (actively pumping, with open oscula) were used in the experiments.

SIP incubations

Two different treatments were amended: 12C‐unlabelled DOM and 13C‐labelled DOM. Per treatment, three sponge replicates and one seawater control without sponge were incubated. The added tracer‐DOM was extracted from 1 g of unlabelled (12C) or 1 g of labelled (13C) lyophilized algal cells (>98% atom purity, ULM‐2177 and CLM‐2065 Agmenellum quadruplicatum, Cambridge Isotope Laboratories). Briefly, the algal cells were resuspended in Milli‐Q water, sonicated for 15–20 min and then centrifuged for 10 min at 8000g. Subsequently, the supernatant was first filtered over a 0.7‐μm GF/F grade filter and then over a 0.2‐μm polycarbonate membrane filter to remove possible algal cell remnants from the DOM solution. This process was repeated with the pellet left after the centrifugation step, in order to extract most of the DOM. The filtered DOM was freeze dried and analysed by elemental analyser–isotope ratio mass spectrometry (EA‐IRMS) to determine the C and N content and isotopic composition. The freeze‐dried DOM was then solubilized in ultrapure water (18.2 MΩ‐cm type I, Elga Purelab Classic UV) at CARMABI Research Station prior to the incubations. Experimental incubations were carried out in 3 l incubation chambers for 6 h. The sponges were transferred, without air exposure, to air‐tight, stirred, incubation chambers (de Goeij et al., 2013), which were filled with 0.7‐μm filtered seawater (FSW). The resuspended DOM was added at the beginning of the incubation using a sterile syringe (at a final concentration of 120 μM DOC). One FSW control (without a sponge) was also incubated per treatment under the same conditions to exclude possible feeding of the sponge on enriched seawater bacterioplankton. All incubations were conducted in the dark and dissolved oxygen concentrations were monitored continuously with an optical probe (OXY‐4; PreSens, Regensburg, Germany). Incubation chambers were placed in a flow‐through aquarium (water pumped up from 10 m water depth from the nearby reef) to ensure near in situ temperatures. At the end of each incubation, the sponges were rinsed in non‐labelled 0.7‐μm FSW and dipped in ultrapure water to remove salts before sampling. Each sponge was cut in half, one half was kept for 13C bulk tissue (i.e. holobiont tissue) stable isotope analysis, and the other half was immediately snap‐frozen and stored at −80°C until DNA extraction. The seawater from each control incubations (3 l) was collected and filtered over a Sterivex filter (GP 0.22 μm) using a peristaltic pump. The Sterivex filters were stored at −80°C until DNA extraction.

Stable isotope analysis

Sponge tissue samples for bulk stable isotope analysis were weighted before (wet weight) and after freeze‐drying (dry weight). Dry tissue was milled with a mortar and pestle. For each sample, a subset of thoroughly mixed tissue powder was decalcified by acidification with a few drops of fuming HCl until effervescence ceased and freeze‐dried again. Duplicate (10–30 mg) acidified and non‐acidified samples were transferred to tin cups for bulk ∂13C isotope analysis. The carbon content and the 13C/12C isotopic composition was measured simultaneously with a high‐temperature combustion element analyser (Vario Isotope cube, Elementar GmbH, Langenselbold, Germany) coupled with an isotope ratio mass spectrometer (BioVision, Elementar, Stockport, UK). A two‐point calibration curve was used to correct the isotopic data. The 13C values were measured against caffeine (IAEA‐600, ∂13C = −27.77‰) and UL‐d‐glucose (IAEA‐309B, ∂13C = 535.3‰), used as low and high internal stable‐isotope standards, respectively. The 13C stable‐isotope data were calculated as previously described (Rix et al., 2020; Bart et al., 2020; for details see Supporting Information). Isotope data from the incubations are corrected for: the background ∂13C values (measured in the sponges incubated with unlabelled DOM), the sponge dry weight, the incubation time, and the enrichment of the DOM source used as substrate. Isotope‐tracer‐DOM assimilation rates are expressed as μmol Ctrac er mmol Csponge −1 h−1.

DNA extractions

DNA was extracted from the sponge tissue samples using the DNeasy® Blood & Tissue Kit (Qiagen). The tissue was first homogenized in 180 μl Buffer ATL and 20 μl proteinase K using a small sterile pestle that fits in a 1.5 ml microcentrifuge tube. Afterwards, all tissue samples were incubated overnight at 56°C. After the incubation, 200 μl of Buffer AL were added to each sample, followed by 200 μl of ethanol [100% (v/v)] and mixed thoroughly by vortexing. Before adding the homogenate onto the DNeasy mini spin column, all samples were centrifuged at 6000g for 2 min to precipitate unlysed material. The supernatant was then transferred onto the spin column and the protocol was further followed as described by the manufacturer. DNA from the Sterivex filters was also extracted using DNeasy® Blood & Tissue Kit (Qiagen). The volumes of Buffer ATL and proteinase K were doubled as deviation from the original manufactory protocol and pipetted into the filter cartridge after which the whole filter was incubated in an oven at 55°C with a sample rocker. After the incubation, 400 μl of Buffer AL were pipetted into the filter cartridge through the luer‐lock side and the filter was incubated for another 20 min at 70°C to deactivate the proteinase K. The entire volume was extracted from the cartridge using a sterile 3 ml syringe by taking it up into the syringe and transferred to a 2 ml microcentrifuge tube. Then, 400 μl of 100% (v/v) ethanol was added and after mixing by vortexing the entire sample was loaded onto the spin column. The protocol was further followed as described by the manufacturer. The concentration and purity of the extracted DNA were checked with a NanoDrop 1000™ spectrophotometer (Thermo Fischer Scientific, MA, USA).

Density gradient centrifugation and fractionation

For DNA separation, caesium chloride (CsCl) gradients were prepared as described by Neufeld et al. (2007a). Briefly, around 2 μg of DNA was mixed with 4.8 ml of 7.163 M CsCl solution and the appropriate gradient buffer (GB) (0.1 M Tris, 0.1 M KCl and 1 mM EDTA) volume to obtain a final density of 1.725 g ml−1. The solution was transferred to a sterile 4.9 OptiSeal polypropylene centrifuge tube (Beckham Coulter, CA, USA) and centrifuged in an Optima L‐90 K ultracentrifuge (Beckham Coulter) fitted with a VTi 90 rotor (Beckham Coulter), for 40 h at 177 000g and 20°C, with the vacuum on, maximum acceleration and deceleration without brake. After centrifugation the gradients were recovered dropwise from the bottom of the tube by injecting ultrapure water with 0.2% (v/v) of loading dye on top of the tube using a LEGATO® 100 syringe pump (KD Scientific, MA, USA) with a speed of 0.5 ml min−1. A total of 15 fractions of 330 μl each were collected per sample. The refractive index (nD‐TC) of each gradient fraction was measured using an AR200 digital refractometer (Reichert, NY, USA) and converted to buoyant density based on a standard curve of nD‐TC versus GB‐CsCl solution. The DNA from all fractions was precipitated with 1 μl of glycogen (20 μg) and two volumes of polyethylene glycol (PEG) 6000 solution. Pelleted DNA was washed with 70% (v/v) ethanol and resuspended in nuclease‐free water. The DNA concentration of each fraction was measured with Qubit® dsDNA HS Assay Kit and Qubit® 2.0 Fluorometer (Invitrogen, CA, USA). DNA was stored at −20°C until further analysis. For each experimental treatment, a DNA density profile was created (Fig. 1). Equal density ranges were selected for the sponge and the seawater control profiles, based on the observed differences in the amount of DNA between labelled and unlabelled treatments. DNA fractions were grouped in the following density ranges: super heavy (fractions 2, 3, 4), heavy (fractions 5, 6, 7), medium (fractions 8, 9, 10, 11) and light (fractions 12, 13, 14), and the fractions within each range were pooled volumetrically (5 μl of each fraction). Because fraction 1 (the heaviest fraction) and fraction 15 (the lightest fraction) contain mostly CsCl or sterile water they were excluded for further analysis.

Fig 1

Distribution of the DNA in the CsCl gradients of (A) sponge DNA (n = 3) and (B) seawater bacterioplankton DNA (n = 1). The line charts show the percentage of total gradient DNA contained in each fraction of 12C treatment (blue line) and 13C treatment (red line) (data shown as mean % ± SD). Density ranges constituting each pooled fraction are separated by the dotted grey arrow and indicated as Super Heavy, Heavy, Medium, and Light.

16S amplicon sequencing and data analyses

Pooled fractions of the CsCl gradient were submitted for 16S amplicon sequencing. The V3‐V4 region of the 16S rRNA gene was amplified using the primer pair 341f/806r (dual‐barcoding approach; Kozich et al., 2013; primer sequences: 5'‐CCTACGGGAGGCAGCAG‐3′ and 5'‐GGACTACHVGGGTWTCTAAT‐3'). The PCR conditions were as follows: initial denaturation step for 30 s at 98°C, 30 cycles of denaturation for 9 s at 98°C, annealing for 30 s at 55°C and extension for 30 s at 72°C, followed by a final extension step of 10 min at 72°C. The product quality and quantity were checked using gel electrophoresis. PCR products were normalized (SequalPrep normalization plate kit; Thermo Fisher Scientific, Waltham, USA) and pooled. Amplicon libraries were sequenced using the MiSeq v3 chemistry sequencing kit (2 × 300 bp) on an Illumina MiSeq platform (MiSeq FGx; Illumina, San Diego, USA). Raw sequences were quality filtered and trimmed based on quality scores. Amplicon sequence variants (ASVs) were computed with the DADA2 algorithm within QIIME2 (version 2018.11). To train the error model, 1 million reads were used. Chloroplasts and mitochondrial sequences were removed from further analyses. Phylogenetic ASV trees were generated with the FastTree2 plugin. A primer‐specific trained Naive Bayes taxonomic classifier was used to classify representative ASVs based on the Silva 132 99% OTUs 16S database. Weighted UniFrac distances were calculated and sample separation in ordination space visualized by non‐metric multidimensional scaling (nMDS) in RStudio Version 1.2. Permutational multivariate analyses of variance (PERMANOVAs) were performed with 999 permutations to determine whether bacterial communities were statistically significantly different between treatments. Furthermore, significant enrichment between treatments and density fractions was determined and ranked using the linear discriminant analysis effect size (LEfSe) algorithm, with a P value of 0.05 and LDA threshold of 2 (Segata et al., 2011). Microbiome Analyst software (Chong et al., 2020; www.microbiomeanalyst.ca) was used to perform PERMANOVAs and LEfSe analyses. Among the significantly enriched taxa (P < 0.05, LEfSe) in the sponge‐associated bacterial community, substrate incorporation was validated based on two criteria: (i) the increasing relative abundance of a taxon in the 13C treatment occurs in at least two consecutive density fractions, with highest abundance in the heaviest of the two fractions, and ii) the relative abundance of the taxon in the 13C treatment is greater than its relative abundance in the 12C treatment in both density fractions. Because only one seawater sample was collected per treatment, the seawater bacterial community comparisons were only made between treatments, using the density fractions as technical replicates, and thus allowing only conclusion on taxa enrichment between treatments, but not between density fractions. Taxa detected only in one of the two treatment were removed from the enrichment analysis in order to exclude possible artefacts of the methodology itself. The sequence data have been deposited in the NCBI database under BioProject ID PRJNA698441.

Results

Sponge tissue was enriched with tracer‐DOM in the 13C treatment as compared to the 12C treatment (210.97 ± 36.62 and − 18.57 ± 0.47 ∂13C ‰ mean ± SD, respectively), showing a mean carbon assimilation rate of 0.46 ± 0.07 μmol Ctracer mmol Csponge −1 h−1 (mean ± SD).

The density of all fractions ranged between 1.68 and 1.77 g ml−1, with a linear trend decreasing from the bottom to the top, indicating proper gradient formation (Supporting Information Fig. S1). DNA extracted from the sponge samples did not show a clear distinction in distribution when looking at the light fractions; however, in the medium and heavy fractions (see magnification panel in Fig. 1A), an increase in the DNA concentration was visible in the 13C treatment as compared to the 12C treatment (Fig. 1A). The distinction in DNA distribution was evident in the DNA extracted from the seawater controls, where the highest peak of DNA abundance shifted from the light fraction of the 12C treatment to the medium fraction of the 13C treatment (Fig. 1B).

16S amplicon sequencing

After filtering and quality control, 908 906 bacterial sequences were obtained from 32 samples, resulting in an average frequency of 28 403 sequence reads per sample. We identified 2102 bacterial amplicon single nucleotide variants (ASVs) affiliated to 23 bacterial phyla. Non‐metric multidimensional scaling (nMDS) based on weighted UniFrac distances clearly separated the sponge‐associated bacterial community and the seawater bacterioplankton community (Supporting Information Fig. S2). An nMDS ordination plot of the sponge‐associated bacterial community alone showed a slight separation between the density fractions along the horizontal axis with heavier ones (darker shaded ‘SH’ and ‘H’) clustering to the left and lighter ones (lighter shaded ‘M’ and ‘L’) to the centre‐right (Fig. 2). Overall, the three replicates of each fractions of the 13C treatment were more dispersed compared with the more closely clustered replicates of the 12C treatment.

Fig 2

Sponge‐associated bacterial community composition visualized by a non‐metric multidimensional scaling plot on weighted UniFrac distances at the ASV level. Each marker is one bacterial community, with symbols and colours indicating the combination of treatment (12C circles in shades of blue and 13C triangles in shades of red) and density fractions (SH: Super Heavy, H: Heavy, M: Medium, and L: Light).

Taxonomic assignment revealed that in the sponge‐associated bacterial community the predominant phyla were Chloroflexi (35.9%), Proteobacteria (22.7%), Acidobacteria (14.2%), Actinobacteria (10.2%), Gemmatimonadetes (4.9%), Nitrospirae (3.8%), Spirochaetes (1.8%), Entotheonellaeota (1.7%), and PAUC34f, Poribacteria, Nitrospinae, and Dadabacteria, each accounting for 0.9% (Fig. 3A). The seawater bacterioplankton community, instead, consisted mostly of Proteobacteria accounting for 94.2% on average (Fig. 3B).

Fig 3

Relative abundance of bacterial phyla among the pooled density fraction of the 12C and 13C treatments in (A) the sponge‐associated bacterial community and (B) the seawater control. Density fractions are indicated as SH: Super Heavy, H: Heavy, M: Medium, and L: Light.

There was a significant difference in the sponge‐associated bacterial community among the density fractions of the 13C‐labelled and 12C‐unlabelled DOM treatments (PERMANOVA, F‐value = 1.647; R 2 = 0.42; P = 0.048). A total of 10 sponge‐associated bacterial ASVs resulted significantly enriched (P < 0.05, LEfSe), of which seven in the 13C treatment and three in the 12C treatment (Fig. 4 and Table 1). Sponge‐associated bacterial ASVs significantly enriched in the 13C treatment super heavy fraction were a PAUC34f bacterium and a Proteobacteria of the genus Endozoicomonas (Fig. 4A and B) and in the heavy fraction a Poribacterial ASV, a Nitrospira genus and three ASVs belonging to the Chloroflexi SAR202 clade (Fig. 4C–G). The three sponge‐associated ASVs significantly enriched in the 12C super heavy fraction were an Entotheonellaeota bacterium, a PAUC26f Acidobacterium and an uncultured Chloroflexi bacterium (Fig. 4H–J). LEfSe analysis showed that four of these sponge‐associated ASVs were significantly enriched at all taxonomical levels (from phylum to ASVs), namely, the Poribacteria, PAUC34f, and Nitrospirae taxa in the 13C treatment and the Entotheonellaeota ASV in the 12C treatment (Table 1).

Fig 4

Relative abundance profiles of significantly enriched sponge‐associated ASVs across treatments and density fractions (n = 3). A–G: Enrichment in the 13C treatment. H–J: Enrichment in the 12C treatment. Blue dots represent unlabelled fractions (12C amendments). Red triangles represent labelled fractions (13C amendments). Density fractions are indicated as SH: Super Heavy, H: Heavy, M: Medium, and L: Light. Asterisks (*) indicate in which density fraction the significant enrichment (LefSE, P < 0.05) was found for the sponge‐associated taxa.

Table 1

Bacterial taxa that were significantly enriched in either the 13C or 12C treatment in the sponge‐associated (n = 3) or in the seawater bacterial communities (n = 1, four technical replicates).

Taxa						Activity	LefSE (P < 0.05)
Phylum	Class	Order	Family	Genus	ASV	Treatment	Sponge	Seawater
Sponge‐associated
PAUC34f	_	_	_	_	c21208a04e6186513386c2c482345edc	13C SH	*
Proteobacteria	Gammaproteobacteria	Oceanospirillales	Endozoicomonadaceae	Endozoicomonas	a2478cc476af4056defe87a849618323	13C SH	**(A)
Poribacteria	_	_	_	_	64b6cef452f5e253205a0543df547dab	13C H	*
Nitrospirae	Nitrospira	Nitrospirales	Nitrospiraceae	Nitrospira	658a5a79e887116ab355ee0a52e3dd0f	13C H	*
Chloroflexi	Dehalococcoidia	SAR202clade	Uncultured Chloroflexibacterium	Uncultured Chloroflexibacterium	60bdd0e2fee3b5ff24c4bf975df23c58	13C H	*(F,G,A)
			Uncultured sponge symbiont PAWS52f	Uncultured sponge symbiont PAWS52f	aac86c50ed061422950976fe61a5e949	13C H	*(F,G,A)
			_	_	a2478cc476af4056defe87a849618323	13C H	*(A)
Entotheonellaeota	Entotheonellia	Entotheonellales	Entotheonellaceae	_	c4682f3866a514d67f9fb8f721f208f5	12C SH	*
Acidobacteria	Acidobacteriia	Solibacterales	Solibacteraceae(Subgroup3)	PAUC26f	12f08292a37324271780e278b29d091a	12C SH	*(C,O,F,G,A)
Chloroflexi	TK30	uncultured bacterium	Uncultured bacterium	Uncultured bacterium	52ced10f6a64091cc72afd91c25f3827	12C SH	*(C,O,F,G,A)
Seawater
Bacteroidetes	Bacteroidia	Flavobacteriales	Flavobacteriaceae	Tenacibaculum	f21f749f2b1805136d34383f8c97ece4 eff75c879b746cdcedf6afb6760e97ae 24ca9fa104b19f14339d1a7311813089	13C		*
		Cytophagales	Cyclobacteriaceae	Marinoscillum	7275d37e008a936d8d5bc763e6bb4bc4	13C		*
				Roseivirga	38d7233bd6197cf2336e07a3048db6cb	13C		*
			Hymenobacteraceae	Pontibacter	e97fb331fff682a80370cee733bd1d78	13C		*
Proteobacteria	Alphaproteobacteria	Caulobacterales	Hyphomonadaceae	Ponticaulis	a813833b5bf150261ffc393184d30d44 40d8840a3be2a4fe8dde5eda0476ad31	12C		*
				uncultured	6de8b46c22914de5a20a9ee515922538	12C		*
		Sphingomonadales	Sphingomonadaceae	_	561ccfc9d063356c8373a600171733ec	12C		*
		Rhodobacterales	Rhodobacteraceae	Shimia	206c94637323740ba5098d13f68ca613	12C		*
				Thalassobius	6dcfa855474bb1059df1cb23148cb7a7	12C		*
				Tropicibacter	9f2a8d79acd772807cef2413a7eec54f	12C		*
				_	01a15b2901a87b10a9c86180308bed9f 23213ccd4beac18b2bc709906fd88799 f9b669c4bc34d832d53e35560fa82bb3 07d7253127dd990ba7624c7583eae864 d465efd32b6588a04333c4ecf8cb4b12 a3db2151875b48c2902e982d93392359 ce1f4d53c75557d9f75dea813b22abf1	12C		*
				Nautella	ac400ce27d3fc3110fc2470116a8ed5b	13C		*
				Pelagibaca	9f90f4404e5d2136a0a3e6c7b338728d	13C		*
				Ruegeria	64eb1f3c8c959ed10dcf396ed914c6f0	13C		*
				Thalassobius	9bd0341d1fa5ed1bc0b72992262bc0ce 1238794ef18ee8cb8a3d859c49baff1b	13C		*
				_	1ce2fdedd72ccc2531126fdddfc0a17e 5e812dd01970e02d58df8803208c4ccf 29fbbbc6e3ca28a22617dd162afcfa20	13C		*
		Rhodospirillales	Thalassospiraceae	Thalassospira	0bb177fe74aef34be459e2e44c6db8aa	13C		*
	Gammaproteobacteria	Vibrionales	Vibrionaceae	Vibrio	0d03338e4efc93449bc6fd3ff698df17	13C		*(A)
		Vibrionales	Vibrionaceae	Vibrio	d236227c6c2e37e2a2e29320e7511b4a 88f99b7a436d728ba7dba26a1e76e946	12C		*(A)
		Pseudomonadales	Moraxellaceae	Acinetobacter	ddcd85a2958ed67c1d62118ca454acec	12C		*(A)
			Pseudomonadaceae	Pseudomonas	70bc0124178ab46e0f1ebc57a31d814e	12C		*(A)
		Cellvibrionales	Cellvibrionaceae	uncultured	822420f29aeac1825f5e8da048d2a70b	12C		*(A)
		Alteromonadales	Marinobacteraceae	Marinobacter	91ec927d4b3dd369fccbf441b0797054	12C		*
			Pseudoalteromonadaceae	Pseudoalteromonas	4706b0f83ab177d31fead600620e5d4a	12C		*
			Alteromonadaceae	Alteromonas	31ea5a141edff498d612057a56d03ee8	13C	*(O,F,G,A)	*
					7d2144b913b1bbe6cb18a3c96aa6a8ef	13C		*
				_	a599a94b6f7d8247f7adb108b6f536aa a69949ac60af0f6e6e35ca9f7f8d95d8	13C		*
			Colwelliaceae	Thalassotalea	46f9eaa760044af4acffa4a79c577b82	13C		*
		Oceanospirillales	Litoricolaceae	Litoricola	62425045241579aa2274efe55060a69b 73504068893298f99a4da53ea1bdef49	13C		*
			Nitrincolaceae	Neptuniibacter	3b307f47731270021d0f8334d4185d0e	13C		*
			Saccharospirillaceae	Reinekea	ce54be95d7dcd5bc326bf8c67da61668	13C		*
				Oleibacter	02ed0c05bcf4178b7d5bffa548811bc2	12C		*
Firmicutes	Bacilli	Bacillales	FamilyXII	Exiguobacterium	f2a5b9165468b028bc5ad45d02ef0a67	12C		*

“*” represents significant enrichment at all listed taxonomical levels; *() represents significant enrichment only at the taxonomical level included in parenthesis where C = Class, O = Order, F = Family, G = Genus and A = ASV; Density fractions are indicated as SH: Super Heavy, H: Heavy, M: Medium, and L: Light.

A significant change was also detected between the 13C and 12C treatment in the seawater bacterioplankton community composition (PERMANOVA, F‐value = 32.129; R 2 = 0.41; P = 0.028). Specifically, 48 ASVs were found to be significantly enriched (P < 0.05, LEfSe) in one of the two treatment, and 42 of them at all taxonomical levels (Table 1). The Bacteroidetes phylum was enriched in the 13C treatment (9.8%) compared to 12C treatment (0.4%) (Supporting Information Fig. S3a). Within this phylum, the enriched genera were Tenacibaculum, Marinoscillum, Roseivirga, and Pontibacter. The Proteobacteria phylum was overall enriched in the 12C treatment (98.9%) compared to 13C treatment (89.6%), however when looking at the class level a division was clear between the Alphaproteobacteria, which abundance increased in the 12C treatment (27.35% vs 54.4%, 13C vs 12C treatment, respectively; Supporting Information Fig. S3b), and the Gammaproteobacteria, which abundance instead increased in the 13C treatment (46.1% vs 34.5%; Supporting Information Fig. S3c). Ten gammaproteobacterial ASVs were enriched in the seawater 13C treatment and belonged to the genera Vibrio, Alteromonas, Thalassotalea, Litoricola, Neptuniibacter, and Reinekea. One of the Alteromonas ASVs that resulted enriched in the seawater bacterial community was also enriched in the sponge‐associated bacterial community from the order to the ASV level in the 13C treatment. The other eight Gammaproteobacterial ASVs were enriched in the seawater 12C treatment and belonged to the genera Oleibacter, Vibrio, Acinetobacter, Pseudomonas, Marinobacter, Pseudoalteromonas, and an uncultured Cellvibrionaceae. Among the Alphaproteobacteria, nine ASVs were enriched in the labelled treatment, including the genera Thalassospira, Nautella, Pelagibaca, Ruegeria, and Thalassobius, while 15 other alphaproteobacterial ASVs were enriched in the 12C treatment, encompassing the genera Ponticaulis, Shimia, Thalassobius, Tropicibacter, a Sphingomonadaceae bacterium, and seven other Rhodobacteraceae. Exiguobacterium of the phylum Firmicutes was also enriched in the seawater 12C treatment (Supporting Information Fig. S3d).

Discussion

DNA‐SIP has been widely used in environmental microbiology, but there are only few studies that applied this technique in vivo to understand the role of microbes that live in symbiotic association with a host (Shao et al., 2014; Alonso‐Pernas et al., 2017). To the best of our knowledge, this is the first study applying DNA‐SIP to a marine holobiont. Here we linked the identity of sponge‐associated bacterial taxa to their ability of incorporating DOM – a complex mixture of low‐ and high‐molecular weight biomolecules, including proteins, lipids, and carbohydrates (Nebbioso and Piccolo, 2012) –, which is the largest source of organic matter in the oceans (Hansell and Carlson, 2015) and is largely inaccessible to other heterotrophic marine metazoans.

Stable isotope analysis, density gradient centrifugation and fractionation

The challenge of DNA‐SIP experiments is employing sufficient substrate (e.g., DOM) concentration and incubation time, but staying as close as possible to the in situ ambient conditions (Dumont and Murrell, 2005; Neufeld et al., 2007b). Too short incubation times may cause insufficient above‐background enrichment for the successful isolation of 13C‐labelled DNA (Teng‐xiang et al., 2016), whereas too long incubation times could cause enrichment bias due to cross‐feeding by secondary consumers on the metabolic by‐products released by the initial consumers (Neufeld et al., 2007b). Experimental conditions become more unpredictable in holobionts with complex microbial communities compared to pure bacterial cultures.

The treatment conditions used in our experiment were suitable to detect isotopic enrichment in the labelled samples compared to the unlabelled ones. Furthermore, our approach resulted in isotopic enrichment and DOM incorporation rates similar to those observed in previous studies where diatom or cyanobacterial DOM was used as a labelled substrate for estimating DOM uptake and release by sponges and their associated microbes (de Goeij et al., 2013; Achlatis et al., 2019; Rix et al., 2020; Bart et al., 2020; Hudspith et al., 2021; Campana et al., 2021). The suitability of our treatment conditions was also proven by the clear shift in DNA distribution visible in the seawater bacterioplankton, indeed the highest peak of DNA abundance shifted from the light fraction in the 12C treatment to the medium fraction in the 13C treatment after a 6 h incubation time (Fig. 1B). This shift indicates an increase in the density of the DNA caused by the incorporation of the heavier (13C) isotope by the active microbial groups, as observed previously in a DNA‐SIP study on DOM uptake by Sargasso Sea bacterioplankton (Liu et al., 2020).

Unlike the bacterioplankton, the DNA distribution profile of the sponge bacterial symbionts did not show a clear and general shift towards heavier fractions in the 13C treatment as compared to the 12C treatment. Only the medium to heavier fractions became slightly heavier in the 13C treatment (Fig. 1A). This could be explained by the relative low abundances of the ASVs that resulted significantly enriched in the sponge‐associated bacterial community. In fact, the total difference in relative abundance between the labelled and unlabelled treatments of the significantly enriched ASVs was around 4%. Furthermore, in these samples, the extracted DNA contained both the DNA of the sponge host and its associated symbionts and we hypothesize that the host DNA was not enriched in 13C as much as the DNA of the associated symbionts. For instance, in another sponge with high abundance of associated microbes, the percent contribution of host sponge cells to bulk uptake of algal tracer DOM was only 35%, with the remaining 65% assigned to the symbiont microbes (Rix et al., 2020). Most likely, the density shift of the bacterial DNA may have been masked by the presence of unlabelled host DNA. One option to avoid this effect could be to perform a priori cell separation of the sponge tissue before DNA extraction (Fieseler et al., 2006; Thomas et al., 2010; Freeman et al., 2013; Astudillo‐García et al., 2018; Engelberts et al., 2020). A couple of studies have applied cell separation techniques with results reaching up to 95%–99% pure microbial fractions (Freeman et al., 2013; Rix et al., 2020) but do not report the number and composition of host and symbiont cells lost in the numerous washing steps during the process. Therefore, using separated microbial cell fractions for microbial community composition studies may not be representative of the overall microbial diversity because bacterial cells lost during the washing steps could result in a biased relative abundance of some bacterial taxa or in the complete disappearance of less abundant taxa (Fieseler et al., 2006). Cell separation can be nonetheless useful in (meta)genomic studies where sufficient coverage of given microbial lineages is required (Fieseler et al., 2006; Bayer et al., 2018; Mori et al., 2018; Schorn et al., 2019; Moeller et al., 2019). Another option to reduce host DNA contamination could be the use of methylation‐based approaches, such as the NEBNext Microbiome DNA Enrichment Kit and the Molzym MolYsis Basic kit, to enrich microbial DNA after extraction (Thoendel et al., 2016). However, similarly to the cell separation method, there are some considerations to be made regarding this approaches. The NEBNext microbiome DNA enrichment kit has been used on one sponge species for the preparation of metagenomic libraries (Burgsdorf et al., 2019), but its application has not been yet optimized for the use on other sponge species. Furthermore, it is unknown whether this techniques may or may not introduce bias in relative abundance analysis, for example, due to the loss of some bacterial taxa.

Bacterial taxa actively incorporating ‐labelled DOM

To date, no clear or consistent shift in the sponge microbial community composition was found in relation to rates of DOM uptake through the sponge holobiont (Gantt et al., 2019). Our study demonstrates that, at least, seven sponge‐associated bacterial ASVs, belonging to the phyla PAUC34f, Proteobacteria, Poribacteria, Nitrospirae, and Chloroflexi, are active incorporators of 13C‐labelled DOM. The relative abundance of these ASVs varied significantly among different density fractions. Higher relative abundance of a microbial taxon in the heavier fractions of the 13C‐labelled treatment can indicate higher metabolic activity in incorporating a certain isotope‐tracer substrate, i.e. DOM in our case. Therefore, the pattern observed in our results suggest that in the sponge P. angulospiculatus a PAUC34f bacterium and an uncultured Proteobacteria of the genus Endozoicomonas could be the first DOM consumers followed by a Poribacteria, a Nitrospira and three Chloroflexi bacteria. We cannot, however, exclude that other microbial taxa present in P. angulospiculatus were also able to process DOM to a certain extent. In fact, a NanoSIMS study on the uptake of 13C and 15N‐labelled DOM by the same sponge species indicated that after a 3‐h  pulse, about 50% of the microbial symbiont cells were enriched in 13C (Hudspith et al., 2021). Our results rather provide direct evidence that the seven mentioned bacterial taxa had significant metabolic activity in the consumption and incorporation of the labelled carbon (13C) source into their DNA within 6 h of incubation time. If other bacteria were able to take up the stable‐isotope‐enriched dissolved organic carbon, but not yet incorporate it into their DNA, the isotopic signal could be detected by NanoSIMS but missed with the DNA‐SIP method in the observed time frame. Longer DNA‐SIP incubations may allow to identify sponge‐associated incorporating taxa with slower division time, but cofounding results could occur. Hudspith et al. (2021) found that there is translocation of DOM processed by sponge cells – predominantly the ‘choanocytes’ or sponge filter cells – to the microbial symbionts 48 h after a 3‐h pulse with 13C/15N‐labelled DOM. It would be challenging to distinguish direct microbial processing of DOM versus translocated host metabolites. The bacterial taxa enriched in the sponge‐associated bacterial communities differed from those enriched in the seawater control; therefore, we can exclude an enrichment bias caused by sponge feeding on the seawater bacterioplankton. The only sign of potential seawater bacteria ingestion by the sponge was the enrichment of the Alteromonas ASV in the 13C‐labelled treatment of both sponge and seawater bacterial communities (Table 1). Therefore, we did not consider the Alteromonas ASV as an indication of endosymbiotic processing of DOM.

To further relate the uptake of certain organic compounds from the DOM pool to the enriched bacterial taxa in the 13C treatment, we can assess their reconstructed metabolic pathways as heterotrophic carbon consumers. The metabolic reconstruction of two metagenome‐assembled genomes (MAGs) of PAUC34f, also known as sponge‐associated unclassified lineage (SAUL), suggests that the members of this phylum possess genes involved in the tricarboxylic acid cycle (TCA), glycolysis, pentose phosphate pathway, Wood–Ljungdahl pathway, and oxidative phosphorylation (Astudillo‐García et al., 2018). Moreover, genes encoding glycoside hydrolases (GH), glycoside transferases (GT), polysaccharide lyases (PL) and carbohydrate esterases (CE) were also present, suggesting the ability of PAUC34f to degrade glycolipids, glycopeptides and glycoproteins, compounds typically found within the sponge mesohyl and as dissolved organic matter in seawater (Blunt et al., 2011; Astudillo‐García et al., 2018). Endozoicomonas bacteria of the phylum Proteobacteria are globally distributed endosymbionts and their genomes have been found enriched in genes for carbon sugar transport, indicating a potential role in the cycling of carbohydrates to their host (Neave et al., 2017). The Poribacteria phylum, originally discovered and described in marine sponges (Fieseler et al., 2004), has a well‐described genomic repertoire (Siegl et al., 2011; Giles et al., 2013; Kamke et al., 2014; Slaby et al., 2017; Podell et al., 2018). The predicted functions of the central carbohydrate metabolism of Poribacteria include complete pathways for glycolysis, oxidative phosphorylation, the TCA cycle, branches of the pentose phosphate pathway, and carbon fixation via the Wood–Ljungdahl pathway (Podell et al., 2018). Furthermore, polysaccharides and other complex carbohydrate degradation pathways can be carried out by polysaccharide lyases, and glycoside hydrolases (GH); the genes encoding these enzymes are abundantly present in the genomes of Poribacteria (Kamke et al., 2013; Podell et al., 2018; Robbins et al., 2021). In analogy with the PAUC34f and the Poribacteria, the metabolism of the Chloroflexi SAR202 clade includes glycolysis, TCA cycle, pentose phosphate pathway and the respiratory chain as energy‐producing pathways, additionally, autotrophic carbon fixation via the reductive citrate acid cycle and the Wood–Ljungdahl pathway were also identified (Bayer et al., 2018). A relevant metabolic specialization in SAR202 is the potential to degrade recalcitrant DOM, due to a large repertoire of oxidative enzymes that may help in the oxidation of recalcitrant alicyclic ring structures to more labile carboxylic acid (Landry et al., 2017; Bayer et al., 2018). The reconstructed genome of a member of the sponge‐associated genus Nitrospira proposes an autotrophic carbon metabolism via the reductive TCA (rTCA) cycle (Moitinho‐Silva et al., 2017b). Nonetheless, there is some evidence of a possible mixotrophic lifestyle, indeed some Nitrospira from marine ecosystems or activated sludge can use simple organic substrates, such as pyruvate, acetate, formate and glycerol for carbon assimilation, but pure heterotrophic growth has not been observed yet (Lucker et al., 2010; Koch et al., 2015; Pachiadaki et al., 2017).

The Bacteroidetes phylum was the most active consumer of DOM in the (0.7‐μm filtered) seawater bacterioplankton. The Proteobacteria phylum was overall enriched in the 12C treatment, but when looking at lower taxonomic level we can see differential activities between members of the Alphaproteobacteria and Gammaproteobacteria, with the latter being the most active in the 13C treatment. It is well recognized that Bacteroidetes (especially Cytophagales and Flavobacteriales) and Gammaproteobacteria are able to mineralize organic aggregates, such as cellulose and chitin, which are compounds that are often found in the high‐molecular mass (HMW) fraction of DOM (Kirchman, 2002; Edwards et al., 2010). Our bacterioplankton observation, however, serves here mainly as control to distinguish sponge bacterial symbionts from seawater bacteria and indicate possible bacterioplankton feeding as opposed to endosymbiont processing of DOM. The limited replication and the filtration over a 0.7‐μm filter (that likely changed the ambient bacterioplankton community composition) of our seawater control prevents solid ecological conclusions from the bacterioplankton treatment. Previous studies have assessed the uptake of different types of DOM by bacterioplankton community of the Sargasso Sea through DNA‐SIP, showing that a wide variety of taxa (among which several groups of the Proteobacteria, Flavobacteria, Actinobacteria and Verrucomicrobia) are capable of incorporating DOM, and that the quality and type of DOM influences the response of these taxa in incorporating such DOM (Nelson and Carlson, 2012; Liu et al., 2020). Likewise, DNA‐SIP can become a powerful tool to understand which sponge‐associated microbial groups mediate the uptake of DOM released by different primary producers present on coral reefs, such as macroalgae, and corals.

Most of the DOM incorporating taxa identified in P. angulospiculatus are characteristic taxa of high microbial abundance (HMA) sponges. In HMA sponges, the bacteria can constitute up to 40% of their biomass and have a distinctive community composition, while sponges with low‐microbial abundance of associated microbes, i.e., low microbial abundance (LMA) sponges, generally possess a microbial community similar to that of the seawater in both composition and abundance (Taylor et al., 2007; Moitinho‐Silva et al., 2017b; Gantt et al., 2019). DOM‐uptake strategies within the sponge holobiont have been shown to vary between LMA and HMA sponges: while the LMA sponge Dysidea avara relied for more than 95% on DOM uptake by host choanocyte cells, in the HMA sponge Aplysina aerophoba the microbial symbionts accounted for the majority (65%) of DOM uptake (Rix et al., 2020). Since the microbial taxa enriched in P. angulospiculatus are indicator taxa of HMA sponges, our results suggests that in HMA sponges DOM incorporation may depend on specific microbial taxa rather than on host cell uptake. However, to validate the hypothesis that sponges with different quantities of associated microbes have evolved different strategies to exploit DOM in the ocean (Rix et al., 2020), a similar DNA‐SIP study in an LMA sponge would be necessary.

Conclusions

This manuscript, to best of our knowledge, represents the first application of a DNA‐SIP experiment in combination with 16S amplicon sequencing in a marine holobiont. We identified sponge‐associated bacterial taxa that are metabolically active in the uptake and processing of DOM. Seven ASVs, belonging to the phyla PAUC34f, Proteobacteria, Poribacteria, Nitrospirae and Chloroflexi were significant incorporators of DOM in P. angulospiculatus. PAUC34f, Poribacteria, and Chloroflexi have similar metabolic capabilities and could be directly involved in the degradation of labile and recalcitrant organic matter through heterotrophic carbon metabolism, while Nitrospirae may have an indirect role through carbon fixation of respired organic carbon or a potential mixotrophic metabolism. The DNA‐SIP technology identified the first DOM‐processing taxa within the holobiont and could be used in the future to trace the uptake of different types of DOM sources or other non‐DOM substances in sponge–microbe interactions.

Author contributions

S.C. and J.M.G. designed the experiments. S.C. performed the experiments, analysed the samples and wrote the manuscript. S.C. and K.B. analysed the data. All authors reviewed and contributed to the manuscript writing.

AppendixS1: supporting information

Click here for additional data file.