Contribution of mono and polysaccharides to heterotrophic N2 fixation at the eastern Mediterranean coastline

N2 fixation should be a critical process in the nitrogen-poor surface water of the eastern Mediterranean Sea. Despite favorable conditions, diazotroph abundance and N2 fixation rates remains low for reasons yet explained. The main goal of this study was to investigate the limiting nutrients for diazotrophy in this oligotrophic environment. Hence, we conducted dedicated bottle-microcosms with eastern Mediterranean Sea water that were supplemented with mono and polysaccharides as well as inorganic nitrogen and phosphorous. Our results indicate that the diazotrophic community expressing nifH was primarily represented by heterotrophic Proteobacteria. N2 fixation and heterotrophic bacterial activity increased up-to tenfold following two days of dark incubations, once seawater was supplemented with organic carbon substrate in the form of glucose (monosaccharides) or gum-xanthan (polysaccharide surrogate). Furthermore, our results point that carbon-rich polysaccharides, such as transparent exopolymer particles, enhance heterotrophic N2 fixation, by forming microenvironments of intense metabolic activity, high carbon: nitrogen ratio, and possibly low O2 levels. The conclusions of this study indicate that diazotrophs in the eastern Mediterranean coast are primarily limited by organic carbon substrates, as possibly in many other marine regions.

In this study, we focused on the role of monosaccharides (i.e. glucose) and polysaccharides (i.e. transparent exopolymer particles) as limiting factors for bacterial metabolism and diazotrophic activity in the oligotrophic EMS. To do so, we conducted nutrient enrichment bioassays using surface EMS water during summertime when oligotrophy is at its peak. Our results indicate that the active diazotrophs community comprised mostly heterotrophic Proteobacterial diazotrophs that were primarily limited by organic carbon substrates and not phosphorus as previously hypothesized. The conclusions of this research shed new light on the role of heterotrophic diazotrophy and their limiting factors in the EMS; one of the most oligotrophic marine environments in the world.

Materials and Methods
Water sample collection. Surface water were collected from the EMS (32° 49′ 34.86 N, 34° 57′ 23.11 E, Haifa, Israel) by pumping seawater from a shallow (~5 m) station during the summer months; September 2013, June and August 2014. Seawater was sampled for inorganic nutrients, chlorophyll a, heterotrophic bacterial abundance, primary and bacterial production, respiration, transparent exopolymer particles (TEP), β -glucosidase activity, N 2 fixation and sequencing of the nifH gene (genomic DNA [gDNA] and complementary DNA [cDNA]).
Polysaccharide-addition experiment. Gum-xanthan (GX) enriched microcosms (in biological triplicates) were conducted during August 2014 to elucidate the role of TEP as a labile carbon source for heterotrophic diazotrophy. Initially, surface EMS water was filtered through a 5-μ m polycarbonate membrane filter (PALL Corp.) in order to retain most of the large aggregates. The pre-filtered water (4.5L) was either supplemented with gum-xanthan (300 μ g L −1 ) or left unamended (controls) for two days under heterotrophic-favored conditions (i.e. dark and DCMU) as described above. GX was dissolved in deionized water (100 ml) and extensively homogenized (Thomas Scientific Model D1000) for 15 minutes as described by Rahav et al. 29 .
Inorganic nutrients. Triplicate water samples were collected in 15-mL acid-washed plastic scintillation vials and were immediately frozen (− 20 °C). Nutrient concentrations were measured using a segmented flow Technicon Auto-Analyzer II (AA-II) system 30 . The limits of detection (twice the standard deviation of the blank) were 0.08 μ M for nitrate + nitrite and 0.008 μ M for phosphate.
Chlorophyll a extraction (chl.a). Seawater samples (300 mL) were passed through a Whatman GF/F filter, extracted overnight in cold acetone (90%) under dark conditions and determined by the non-acidification method 31 using a Turner Designs (Trilogy) fluorometer with 436 nm excitation and 680 nm emission filters.
Primary production (PP). Photosynthetic carbon fixation rates were estimated using the 14  Bacterial production (BP). Rates were estimated using the 3 H-leucine (Amersham, specific activity: 160 Ci mmol −1 ) incorporation method 33 . A conversion factor of 1.5 kg C mol −1 per mole leucine incorporated was used, assuming an isotopic dilution of 2.0 34 . See supplementary information for further details.
Heterotrophic bacterial abundance. Samples (1.8 mL) for bacterial abundance were collected, fixed with 50% glutaraldehyde (0.15% final conc. v/v., Sigma G7651), incubated in room temperature for 10 min, and frozen in liquid nitrogen until analyses. Prior to counting, samples were fast thawed at 37 °C and stained with 0.5 nM SYTO9 (Applied Biosystems) in the dark for 15 min 35 . Sub-samples (100 μ L) were analyzed with an Attune acoustic focusing flow cytometer (Applied Biosystems) equipped with a syringe-based fluidic system at 408 and 488 nm wavelengths at a flow rate of 25 μ l min −1 . Beads (nominal size 0.93 μ m) (Polysciences) were used as a size standard.
Dissolved oxygen (DO) and bacterial respiration rate (BR). Oxygen was measured with the Winkler titration method modified by Carpenter 36 . Water was immediately fixed with MnSO 4 and KI + NaOH and sealed without headspace in 300 mL Winkler bottles (Wheaton). Once oxygen was residue, H 2 SO 4 was added to the samples and titrated with Na 2 S 2 O 3 using a Metrohm 785 DMP titrino auto-burette and double platinum electrode (end-point titration precision, ± 1 μ mol L −1 ).
Bacterial respiration (BR) rates were determined according to the following equation: Where DO (T0) is the initial dissolved oxygen concentration and DO (T48) is the dissolved oxygen concentration after two days of dark incubation. We assumed that BR accounted for 90% of the entire community respiration as both chl.a and primary production decreased in the same proportion under dark (with DCMU) conditions. Bacterial carbon demand (BCD). BCD was defined as the sum of carbon assimilation measured by bacterial production plus carbon oxidation through bacterial respiration. Oxygen respiration was converted into carbon consumption assuming a respiratory quotient (RQ) of 1 37 . For isotope ratio mass spectrometry, a standard curve to determine N mass was generated for each sample run. Based on natural abundance, N mass on the filters, incubation times, and precision of the mass spectrometer, our calculated detection limit for 15 N uptake was 0.02 nmol N L −1 d −1 .
Extraction and sequencing of the nifH gene. Samples (1L) were filtered through 0.2 μ m Supor filters (PALL Corp.) and placed in a sterile DNase/Rnase Free Whirl-Pak bag. The samples were than snap frozen in liquid nitrogen and stored at − 80 °C. DNA was extracted using the phenol-chloroform method according to Man-Aharonovich et al. 21 . RNA extraction was carried with a mirVana RNA isolation kit (Ambion). Genomic DNA (gDNA) contamination was removed using the DNase I digestion Turbo DNA-free kit (Ambion) and removal was verified by PCR (16S universal primers 519F-1492R) prior to reverse transcription. Complementary DNA (cDNA) synthesis was accomplished using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) following manufacturer's instructions. Nitrogenase Fe protein transcripts (nifH) were amplified using a nested PCR strategy 43  Sequencing analysis. Merged Illumina reads were quality filtered and analyzed with the Quantitative Insights Into Microbial Ecology (QIIME) pipeline 44 . The remaining reads were binned into operational taxonomic units (OTUs), defined at 97% similarity, using the UCLUST algorithm 45 . Taxonomy was assigned with BLAST and a database of nifH sequences from Heller et al. 46 . Phylogenetic trees were generated with FastTree in QIIME 47 , and visualized with the Interactive Tree of Life (IToL) and Topiary Explorer v1.0 packages.
Many of the OTUs were unidentifiable, and therefore nifH sequences from representative phototrophs and heterotrophs were added to the trees as outgroups, using Anabaena sphaerica for rooting (Table S1). OTUs that clustered with either a phototrophic or a heterotrophic outgroup were considered phototrophs or heterotrophs, respectively. Further details on quality filtering criteria and sequencing data analyses are indicated in the supplementary information.
Statistical analyses. Changes in bacterial production, primary production, bacterial carbon demand, β -glucosidase and N 2 fixation rates in the different treatments were evaluated using a one-way analysis of variance (ANOVA), followed by a Fisher LSD multiple comparison post hoc test with a confidence of 95% (α = 0.05). A Pearson linear correlation was carried out between N 2 fixation and bacterial cell specific activity and TEP (α = 0.05). All statistic tests were done using the XLSTAT software.

Results and Discussion
Physicochemical and biological characteristics of the study site. The coastal water was characterized by typical EMS summer conditions, with warm, saline and well-oxidized conditions (Table 1). Average dissolved inorganic nitrogen and phosphorus were in the lower end of oligotrophic environments 30 , resulting in basal chl.a concentrations (phytoplankton proxy) and low primary production rates (Table 1). Heterotrophic bacterial abundance and production rates remained unchanged in all sampling dates (Table 1) and were similar to previously reported summertime concentrations and rates in the EMS 18,48 .
The diazotrophic community, based on nifH DNA analyses, was represented by both heterotrophic bacteria and autotrophic cyanobacteria (Fig. 1). The retrieved sequences converged into ten distinct groups ascribed to clusters I and III 49 . We acknowledge that since only 1-L of water was sampled, species with low abundances such as Trichodesmium may have been overlooked. Yet, based on previous studies in the EMS, bloom-forming N 2 fixers such as Trichodesmium were rarely detected, even after the filtration of ~20-L of seawater 21,22 . Although the EMS is presumably an ideal environment for cyanobacterial diazotrophs (sunlit, warm and nitrogen-poor, Table 1), studies spanning over a decade have determined that most of the diazotrophs species cluster within the Proteobacterial clades, mainly composed of alpha, beta and gamma-proteobacteria 21,23 . Our results indicate that the diazotrophic diversity was highly compatible with long-term summer observations of the coastal and open EMS [21][22][23] .
Concurrent with the absence of bloom forming cyanobacterial diazotrophs, measured N 2 fixation rates were overall low (Table 1) and similar to those measured in the open [24][25][26]50 and coastal 18 EMS water. Currently, the reasons for the absence of large bloom forming diazotrophs or the low N 2 fixation rates in this nitrogen-poor system are unknown.
Phosphorus was the most studied limiting nutrient in the EMS due to the high N:P ratio in its deep water suggesting that diazotrophs are hindered by P availability. Therefore, only P, P + Fe or dust were examined to date as a diazotrophy-limiting nutrients in the surface of the EMS 26,28 . Yet, the addition of P yielded inconclusive responses with insignificant changes in N 2 fixation rates across the Levantine Basin during July 2009 50 and a 3-fold increase   Table S1). during June-July 2008 28 . Concurrently, dust additions were found to elevated N 2 fixation via the release of P and Fe 28 and/or by supplying airborne diazotrophs associate with the dust particles 51 . A recent study across the Mediterranean Sea determined that N 2 fixation measured under dark conditions equaled the rates retrieved under ambient light 50 . These results suggest that heterotrophic diazotrophs play a significant role in supplying new bioavailable nitrogen to the Mediterranean Sea 6,12 . Heterotrophic diazotrophs, unlike phototrophs (e.g. Trichodesmium and Crocosphaera), cannot harvest light through photosynthesis to maintain the energetic needs of the nitrogenase complex. Instead, these organisms scavenge and mineralize organic substrates. Assuming heterotrophic diazotrophs are important contributors to N 2 fixation in the EMS 8,50 , we postulated that the prime factor that limits diazotrophy is organic carbon.
Diazotrophic feedback to the addition of organic carbon. Our results indicate that N 2 was primarily fixed by heterotrophic-bacterial diazotrophs under ambient light ( Fig. 2A-C) and dark (Fig. 2D-F) conditions once organic carbon was not limiting. Incubating EMS water under ambient light conditions with the different nutrient additions did not affect phytoplankton concentrations (ranging from 0.27 to 0.43 μ g chl.a L −1 ); apart than NP addition which triggered elevated phytoplankton biomass (1.60 μ g chl.a L −1 ). These results are in agreement with previous studies from the EMS indicating that phytoplankton (not-necessarily diazotrophs) are co-limited by N and P 52 .
Similarly, no significant changes were found in heterotrophic bacterial abundance across the different microcosms (6 × 10 8 to 8 × 10 8 cells L −1 , one-way ANOVA and a Fisher LSD means comparison test, P > 0.05). However, following glucose additions (G, GN, GP and GNP), primary production significantly decreased ( Fig. 2A), while bacterial production drastically increased (Fig. 2B). These results suggest that glucose (as an organic carbon surrogate), rather than phosphorus 52 , is the prime limiting nutrient for heterotrophic bacterial activity at our study site (EMS). It is likely that once organic carbon is present and bacterial growth is prompt, both N and P become the limiting factors for bacterial activity. Further, we suggest that phytoplankton and heterotrophic bacteria are competing for the available nutrients as previously found in open sea EMS 48,52 .
At the end of the light incubations, N 2 fixation rates were overall low (< 0.5 nmol N L −1 d −1 ) and similar to values across the EMS 24,50 . However, when G was added (G, GN, GP, GNP), N 2 fixation rates were enhanced by 2 to 7 fold compared with the control microcosms (Fig. 2C). The compatibility between the increased bacterial biomass and activity with the elevated N 2 fixation rates indicate that diazotrophic heterotrophs are the prime N 2 -fixing organisms in our system once organic carbon is available. We surmise that the overall low N 2 fixation rates, even under ambient light conditions that hypothetically should favor cyanobacterial diazotrophs, result from a competition with phytoplankton for any available N or P that are essential for bacterial metabolism. In the experiment where microcosms were incubated under dark conditions, phytoplankton concentrations were low regardless of the supplements combinations and ranged between 0.10 to 0.23 μ g chl.a L −1 . Correspondingly, primary production rates were significantly lower than under the light condition, with negligible changes between the different microcosms, apart than NP (Fig. 2D). Similarly to the microcosm incubations conducted under ambient light, heterotrophic bacterial abundance remained unchanged, regardless of the different nutrient additions (6 × 10 8 to 10 × 10 8 cells L −1 , one-way ANOVA and a Fisher LSD means comparison test, P > 0.05). However, bacterial production rates increased once glucose was added, and peaked (4.7 ± 0.2 μ g C L −1 d −1 ) in the microcosms containing the GNP addition (Fig. 2E). N 2 fixation rates, similarly to the bacterial production, were significantly higher following glucose additions and drastically increased (12 fold) with GNP supplements (Fig. 2F). Since phototrophic activity was suppressed by the long-dark incubations, it is highly likely that only heterotrophs and facultative microorganisms could utilize the available nutrients and proliferate. Hence, the high N 2 fixation rates measured in the dark microcosms-especially following glucose addition (up to 3.84 nmol N L −1 d −1 ), were mostly attributed to heterotrophic diazotrophs.
Furthermore, based on nifH gene expression, 56% of the active diazotrophic communities in the control bottle microcosms were heterotrophs following two days of dark conditions (Fig. 3). These clusters were represented mostly by alpha-Proteobacteria phylotypes as well as by marine Stromatolites; a complex community of benthic heterotrophic Protobacteria and phototrophic heterocyst-forming cyanobacteria 53 . In the glucose supplemented microcosms (G, GN, GP and GNP) the relative abundance of these heterotrophic nifH expression bacteria increased up to 85% highlighting their importance to N 2 fixation (Fig. 3).
The biochemical and phylogenetic results indicate that during the summer period at the EMS coastline, active diazotrophs are mostly heterotrophs and often limited by organic carbon. Furthermore, our results point on a strong dependence between the availability of organic carbon in the form of glucose, bacterial metabolism and N 2 fixation by heterotrophic diazotrophs. Once monosaccharides such as glucose are available, assimilated carbon can be directed to meet the energetic demands of the nitrogenase complex and prompt N 2 fixation 1 . The above results are in agreement with reports from other aquatic regions such as the South Pacific Ocean 12-14 and from oxygen minimum zones 15,16 .
Glucose metabolism by bacterial heterotrophs may cause a rapid uptake of inorganic nitrogen and phosphorus, thus leading to nutrient stress. Therefore, when glucose is available in addition to phosphorous and/or nitrogen, the activity of bacteria, including of heterotrophic diazotrophs, intensifies, resulting in high N 2 fixation rates (Fig. 2F). This trend was also observed following addition of a mixture of amino-acids (typically composed by organic carbon and nitrogen) to surface water in the Solomon and Bismarck seas 13 , the aphotic water in the eastern Tropical south Pacific Ocean 15 and the northern Red Sea 29 .
The dependence on organic carbon and nutrient availability was further highlighted by the positive linear correlation between heterotrophic cell-specific activity (bacterial production per cell) and N 2 fixation (Fig. 4A). This relationship points that supplementing the EMS water with glucose enhanced heterotrophic bacterial cell specific activity, which was tunneled into the energetically expensive N 2 fixation process 1 . Taken together the above results indicate on a complex regulation of heterotrophic or possibly mixotrophic diazotrophs activity by varying nutrients and combinations, but primarily due to organic carbon availability.  Table S1, similar to the analysis performed in Fig. 1 Another indication of the importance of organic carbon to diazotrophy is the positive and significant relationship between N 2 fixation and carbon-rich polysaccharides in the form of TEP measured in the different microcosms (Fig. 4B). TEP are acidic polysaccharide hydrogels, intensely sticky, three-dimensional-supramolecular networks, ranging in size from ~0.4 to > 200 μ m, and ubiquitous in marine 54 and freshwater 38 environments. TEP may serve as a surface and substrate for planktonic organisms such as heterotrophic bacteria and cyanobacteria, thus forming microenvironments of intense microbial activity with high carbon to nitrogen content 40,54 . A recent study has suggested that these carbon-rich hydrogels may also act as favorable "hotspots" for diazotrophy in oligotrophic conditions 8 .
Impact of carbon-rich polysaccharides on heterotrophic N 2 fixation. Carbohydrate-microgels such as TEP were found to promote heterotrophic bacterial activity and N 2 fixation in dedicated polysaccharide-enriched microcosms (Fig. 5). EMS water was first pre-filtered (5 μ m) before they were distributed into the different microcosms, thus removing 60% of the phytoplankton biomass and reducing TEP by approximately 80% (Table 2). However, the pre-filtration step did not change the bacterial abundance ( Table 2). The filtrate was then supplemented with gum xanthan (GX); a commercially available polysaccharide comprised mostly of mannose, glucuronic acid and glucose 54 . The added GX was ~50% of the TEP concentration usually found in the EMS surface 17 , thus represents a moderate yet realistic scenario.
The addition of GX as a pure TEP surrogate triggered a 10-fold increase in N 2 fixation relative to the control microcosms following two days of dark incubations (Fig. 5A). Concurrent with the elevated N 2 fixation, a substantial increase in the abundance of heterotrophs that expressed the nifH genes was found following GX addition (2.5 fold compare to the control, Fig. 5B). Similarly to the previous experiment (Fig. 3), the majority of these nifH expressing diazotrophs were associated with Proteobacterial phylotypes.
We suggest that the elevated N 2 fixation rates were mediated by biodegradation and metabolism of polysaccharides such as GX by heterotrophic diazotrophs. β -glucosidase is one of various ectoenzymes that are secreted by bacteria to hydrolyze polysaccharides (such as TEP) into bio-available molecules 54,55 . Indeed, β -glucosidase activity increased by 45% once GX was supplemented (Fig. 5C). At the same time, bacterial production and respiration were enhanced by 80% and 50% respectively (Table 2), corresponding to a 50% increase in heterotrophic bacterial carbon demand (BCD) when compared to the control microcosms (Fig. 5D). The BCD reflects the total carbon biomass that is required to sustain the metabolic demands of heterotrophic bacteria 56 . Our results indicate that TEP hydrolysis of GX by β -glucosidase accounted for 75 μ g C L −1 d −1 , which was approximately 15% of the total BCD (Table 2, Fig. 5). Yet, the direct contribution of polysaccharide hydrolysis to heterotrophic diazotrophs carbon demand is currently unclear.
Based on our results, we propose three, not mutually exclusive pathways, in which TEP can support heterotrophic diazotrophy: (i) TEP are often found as bio-aggregates that are heavily colonized by bacterial communities 38,57 . These tight bacteria-hydrogel associations may enhance polysaccharide hydrolysis by increasing the efficiency of ectoenzymes such as β -glucosidase 58 ; thereby support the high energetic requirements of N 2 fixation 1 . (ii) TEP comprises high C:N ratio 40,54 . Therefore, as carbon availability enhances heterotrophic bacterial activity, nitrogen becomes the limiting factor and therefore diazotrophy is promoted 26 . (iii) It was estimated that diazotrophs (such as Crocosphaera watsonii) direct approximately 60% of the energetic costs of N 2 fixation to remove intracellular oxygen since nitrogenase is an oxygen-sensitive protein 1 . Large aggregates (> 1 mm) may result in oxygen gradients that decrease from the surface to the anaerobic center 59 . The anoxic conditions within large aggregates, such as TEP, may benefit heterotrophic diazotrophs by shielding the nitrogenase complex from Microcosms were conducted under dark + DCMU incubations for 48 h. Each nutrient addition type has a different color-code. Note that the GNP treatments were not included in the linear correlation. A Pearson linear correlation was carried out between N 2 fixation and bacterial cell specific activity and TEP (P < 0.05). extracellular oxygen, thus reducing the overall energetic requirements and promoting N 2 fixation. Similarly, Pseudomonas stutzeri strain BAL361 isolate from the Baltic Sea surface water formed aggregates (1-4 mm), which controlled O 2 diffusion into the cells and facilitate N 2 fixation under an oxygenated environment 60 . We postulate that bacterial diazotrophs associated with bioavailable hydrogels are not restricted solely to TEP, but would be found with other types of bioaggregates in the aquatic environment such as marine snow, fecal pellets and detritus matter.  Chl.a μ g L −1 0.14 0.03 ± 0.01 0.02 ± 0.02 0.8 ± 1.0 Primary production μ g C L −1 d −1 0.89 0.08 ± 0.02 0.04 ± 0.01 0.5 ± 0.2

Conclusions
The oligotrophic EMS is presumably an ideal environment for diazotrophy, and yet, for over a decade, only low N 2 fixation rates have been measured throughout with no reports of basin-wide diazotrophs blooms. The results of this study indicate that N 2 was fixed primarily by heterotrophic bacteria. Furthermore, we highlight, for the first time, that the prime limiting factor for heterotrophic bacterial diazotrophs, in the EMS is organic carbon (Figs 2 and 5) rather than phosphorus as previously suggested. In fact, we show that only when organic carbon is provided in the form of monosaccharides (glucose) or polysaccharides (e.g. TEP), which results in high C:N ratio, bacterial metabolism and biological N 2 fixation are intensified. Such carbon-rich inputs are often occurring along the Israeli coast as well as other coastal environments along the Mediterranean Sea from various anthropogenic sources. These coastal water masses, along with its subsequent nutrients and possibly carbon-fueled diazotrophs, routinely intrudes into the open EMS and therefore may pose an important ecological role on both the coastal and open EMS water. A better understanding of the link between heterotrophic bacterial-diazotrophs and organic carbon supplements such as carbon-rich TEP hydrogels might prove critical, as climate change predicts an expansion of oligotrophy in many marine and coastal environments. Under these circumstances, microenvironments such as TEP might benefit with diazotrophs, and thereby hold important ecological significance not only to carbon sequestration but also to the nitrogen cycle.