Introduction

Seagrasses are one of the most productive ecosystems, placed in continental seashores of all continents but Antarctica [1], providing key ecological services. As relevant coastal primary producers, seagrasses fix atmospheric CO2, acting as carbon sinks and O2 releasers. They also sustain secondary production, trap particles, stabilize sediments, are key actors of biogeochemical processes (e.g., nitrogen cycles), and form the habitat and nursery for several micro- and macro-organisms [2]. Their associated microbiome takes part in the major ocean’s biogeochemical cycles (e.g., nutrient cycling of nitrogen (N), phosphorus (P), and/or iron (Fe)) and microbial food webs [3,4,5], essential for the livelihood of seagrasses. Within the bacterial consortia, primary producers (e.g., Prochlorococcus and Synechococcus) play a critical role through their CO2-fixing capacity [6], together with other bacteria in sulfide detoxification (which is highly toxic for plant growth [7]), and biological N2-fixation (BNF) which compensates denitrification, converting dinitrogen gas (N2) to inorganic N (i.e., ammonia, NH3) [8]. The role of environmental factors (e.g., nutrient availability) modulating bacterial community composition, diversity, functioning, and, thus, the health and productivity of the seagrasses, remains to be investigated.

The implication of the microbial communities associated with seagrasses, which can be found as epiphytic or endophytic, has been profusely illustrated on the Mediterranean endemic seagrass Posidonia oceanica (L.) Delile [9,10,11]. N2-fixers or diazotrophs can supply the entire N-demand of the plant, having a key role in nutrient cycling [12], considering the Mediterranean Sea (MS) oligotrophy. The MS is a semi-enclosed sea chronically limited by P [13], where P and N concentrations do not obey the Redfield ratio (i.e., 23–28:1) [14, 15]. Besides, Mediterranean waters are subjected to Saharan atmospheric dust deposition, which controls Fe concentrations in the water column [16]. On the other hand, the Humboldt Current System (HCS), which extends from the west coast of South America (Southern Chile) up to Ecuador and the Galapagos Islands, is one of the most productive marine ecosystems as a result of the transport of sub-Antarctic-enriched nutrient waters and often poor in Fe [17,18,19]. There, the endemic Heterozostera chilensis J. Kou seagrass thrives settled after long-distance dispersion of its ancestor, i.e., Heterozostera nigricaulis, from temperate waters of Australia across the Pacific Ocean [20, 21]. Noteworthy, it is intriguing how H. chilensis (former H. nigricaulis) was capable to settle in the cold waters of the HCS, concretely in Puerto Aldea, Bahia Chascos, and Isla Damas [21], where historically seaweeds have proliferated, but no seagrasses [22]. Among the many intrinsic characteristics that may have allowed H. chilensis to survive, the role of its associated microbiota in its maintenance and sustenance remains unclear.

Elucidating the nutrient regimes of N, P, and Fe of these microbial communities is of especial relevance as they limit microbial growth and activity [17], usually controlled by upwelling/downwelling processes or by atmospheric dust deposition [23]. The average ratio of N/P/Fe in microbial communities obeys to Redfield ratio, 16 (N):1 (P):0.0075 (Fe), and hence N and P are the major elements needed for microbial biomass, constituting around 7% and 1% of cell biomass, respectively [24, 25]. In P-limited waters, bacteria are capable of hydrolyzing P-esters contained in dissolved organic phosphorus (DOP) releasing dissolved inorganic phosphorus (DIP), through the so-called alkaline phosphatases (APases). Bacterial APases can be classified into three main families depending on the associated co-factor (i.e., Mg2+, Zn2+, Fe3+, and/or Ca2+): PhoA, PhoX, and PhoD [26]. Thus, metals, as Fe, are essential for enzyme activity as structural components. Nonetheless, Fe homeostasis has to be tightly regulated since it can promote oxidative stress by the generation of reactive oxygen species (ROS) through Fenton and Haber-Weiss reactions [27].

Multifactorial studies investigating the interactive effects of N, P, and Fe in marine microorganisms are very scarce [28,29,30,31]. Low dissolved inorganic N (DIN, e.g., NO3−), inorganic P (PO43−), and Fe concentrations can impair growth, enhancing oxidative stress, morphological changes, and apoptosis processes, as well as limiting N2-fixation rates [31]. In oligotrophic waters (e.g., North Atlantic Ocean or the MS), Fe can limit P-mechanisms as the alkaline phosphatase activity (APA) [32], and APA can be subjected to DIN concentration [30]. In the global ocean, microbial communities are faced with a wide range of nutrient regimes (e.g., from oligotrophy to eutrophic) and abiotic scenarios, selecting the best-adapted species worldwide [33]. Analyzing the biogeographic patterns of cosmopolitan species is of special relevance to better address the success of these species in adapting and surviving to a wide range of scenarios, an issue poorly studied with bacteria associated with seagrasses.

Gammaproteobacteria is one of the most distributed marine heterotrophic bacteria in the oceans [34]. Among them, the Halomonadaceae family (which include Modicisalibacter, Halotalea, Zymobacter, Carnimonas, Cobetia, Kushneria, and Salinicola) is found in almost any saline environment, being moderate halophilic bacteria and exopolysaccharides producers [35, 36]. As a result of the re-classification of the Halomonas marina, Arahal et al. (2002) described the genus Cobetia as a member of the Halomonadaceae [20], which comprises five species (i.e., C. amphilecti, C. crustatorum, C. lioralis, C. marina, and C. pacifica) [37,38,39]. Cobetia spp. are isolated as hydrocarbon-degrading, biosurfactants-producing, which display a nutritional versatility and different metabolic profiles (e.g., sugar utilization and assimilation), being sources of psychrophile enzymes. Besides, some strains have a key role in improving the water quality [40,41,42,43]. Although the ecophysiology and adaptation responses of Cobetia spp. remains poorly explored, the genus Cobetia is well distributed throughout the world oceans having been isolated from different marine ecological niches, pointing to their high adaptability.

Our genomic analysis proved that we isolated two strains that belong to the same species, inside the genus Cobetia, placed in geographically unrelated endemic seagrasses: P. oceanica from the MS and H. chilensis from the HCS. Based on our previous studies in which we described the sensitivity of the diazotrophic bacteria to different nutrient regimes and emerging pollutants [30, 31, 44], the aim of this study, considering the contrasting features of the MS and HCS at nutrient level, was to evaluate the physiological responses of the Cobetia spp. isolated, using batch cultures. For this purpose, we conducted long-term (i.e., 12 days (d)) and short-term experiments (i.e., 4 d/96 h) towards different nutrient regimes (i.e., N, P, and Fe) and pH levels, measuring their physiological and biochemical responses.

Materials and Methods

Isolation, Sequencing, Identification, and Genomic Features of UIB 001 and 4B UA Strains

Isolation of the Strains

The strains UIB 001 and 4B UA were isolated from the roots of P. oceanica and leaves of H. chilensis, respectively. Posidonia oceanica roots were collected from Cala Nova (Mallorca) in Balearic Islands, Spain (39° 33′ 02.3″ N 2° 36′ 02.1″ E), while H. chilensis leaves were collected from the seashore of Puerto Aldea (Coquimbo), Chile (30° 17′ 37.83″ S–71° 36′ 24.55″ O). Posidonia oceanica roots were triturated in Tris-EDTA 1 mM pH 7.5, whereas H. chilensis leaves were rasped to gather the epiphytic community, which was maintained in sterile seawater. In both cases, the microbial consortia were cultured in modified artificial seawater (ASW) medium (L−1: 25.0 g NaCl, 1.0 g MgCl2·6H2O, 5.0 g MgSO4·7H2O, 0.7 g KCl, 0.15 g CaCl2·2H2O, 0.1 g KBr, 0.04 g SrCl2·6H2O, and 0.025 g H3BO3), with 1 ml L−1 trace metal [L−1: 2.86 g H3BO3, 1.81 g MnCl2·4 H2O, 0.22 g ZnSO4·7H2O, 0.39 g NaMoO4·2H2O, 0.079 g CuSO4·5H2O, and 0.049 g Co(NO3)2·6H2O], glucose/citrate (final concentration 0.1% (v/v)), and with agarose (1.5% (p/v)). Cultures had N2 as sole N source (with low P and Fe concentrations) and were incubated at 20 °C. The resulting colonies were subcultured in marine agar (MA) to obtain pure cultures.

Genome Sequencing

Total genomic DNA of the strains UIB 001 and 4B UA were extracted using the optimized version of Salvà Serra et al. (2018) [45]. Illumina whole-genome sequencing was performed (Eurofins Scientific, Luxemburg). Briefly, the DNA libraries were prepared following an optimized protocol and standard Illumina adapter sequences. Paired-end sequences were determined using the Illumina HiSeq platform (read mode, 2 × 150 bp). Additionally, DNAs were sequenced with a MinION Mk101B sequencer (Oxford Nanopore Technologies, Oxford, UK) for the generation of long-read sequences. The DNA library was prepared, using the rapid barcoding sequencing kit vR9 (SKQ-RBK004) and loaded into a FLO-MIN106 vR9.4 flow cell. The sequencing process was performed for 48 h (h) on MinKNOWN software v1.4.2 (Oxford Nanopore Technologies). The raw reads obtained were base called using Guppy software v3.5.1 (Oxford Nanopore Technologies). Quality of Illumina and nanopore reads were determined using fastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and NanoPlot software (https://github.com/wdecoster/NanoPlot), respectively. A hybrid assembly was done using Unicycler v0.4.7 [46] combining the Illumina and nanopore reads. The qualities of the genome sequences obtained were assessed using the Quality Assessment Tool for Genome Assemblies (QUAST) v4.5 [47]. For submission to GenBank, the whole-genome sequences were annotated with the Prokaryotic Genome Annotation Pipeline (PGAP) [48, 49].

Strain Identification

Complete 16s rRNA gene sequences were extracted from the obtained genomes and analyzed using EzBioCloud to determine the closely related species [50]. Afterward, the genome sequences or the strains UIB 001 and 4B UA were compared with the genomes of the closely related species, which were obtained from NCBI. Comparative 16S rRNA gene sequence analyses were done using the Kimura two-parameter model to calculate evolutionary distances. Cluster analyses and phylogenetic trees were built by Neighbor-Joining, using the MEGA7 software [51]. Bootstrap values were determined for 1000 replications. Average nucleotide identities by BLAST (ANIb) were determined using the JSpeciesWS tool [52]. The digital DNA–DNA hybridization (dDDH) similarity values were determined with the Genome-to-Genome Distance Calculator (GGDC) [53]. Only results of the recommended formula 2 (sum of all identities found in high-scoring segment pairs (HSPs) divided by the overall HSP length) in the GGDC analyses were considered.

Genomic Difference Detection Through Pan-Genome Analysis and Single Nucleotide Polymorphisms (SNPs)

Pan-genome analysis was performed following an already described procedure [54]. Briefly, homologous determinations were performed using the software Get_Homologues [55], based on two different algorithms: Cluster of Orthologous Genes Triangle (COGT) and Orthologous Markov Cluster (OMCL). The total number of clusters conforming the pan-genome was determined with the consensus of COGT and OMCL. Exclusive proteins of each strain, extracted from the pan-genome determination, were assigned to a functional category using the eggNOG-mapper v2 online tool [56]. Single nucleotide polymorphism (SNPs) analyses were done using Snippy (https://github.com/tseemann/snippy), using the default parameters and the genome of strain UIB 001 as a reference.

Experimental Design

Prior to the experiments, temperature (4–42 °C) and NaCl tolerance (0–20% (p/v)) were tested. Besides, biochemical tests were carried out in the Spanish Type Culture Collection (CECT) using API 20NE and API ZYM test kits (Biomerieux), following the manufacturer’s instructions and adding Marine Cations Supplement in a 1/10 ratio, and incubated at 30 °C. To evaluate their ability to cope with different nutrient regimes and changes in cell morphology, bacterial cells were grown in ASW with agarose (1.5% (p/v)), containing optimal nutrients (+N, +P, +Fe, and +carbon source (C)), N2 as sole N source (−N, +P, +Fe, and +C), without P and Fe (+N, −P, −Fe, and +C), without N, P, and Fe (−N, −P, −Fe, and +C), with N, P, Fe, and not C (+N, +P, +Fe, and −C), and without N, P, Fe, and C (−N, −P, −Fe, and −C).

Cobetia sp. isolates were cultured and acclimatized in marine broth (MB) at 20 °C and 150 r.p.m. The subcultures for seeding were centrifuged and washed with their corresponding medium without N, P, and Fe. All the experiments were performed in triplicate in sterile 50 mL Falcon tubes and/or 2 mL microplates in modified ASW culture media (without agarose), in batch cultures. We inoculated 7 × 105 cell mL−1 in each treatment. Samples were manipulated in a glass-clean hood to avoid Fe contamination, and cultures were maintained over 4–12 d, at 20 °C and 120 r.p.m.

Different inorganic/organic N (i.e., NH3 from NH4Cl and/or C4H11NO3 [Tris]), inorganic P (i.e., PO43− from K2HPO4), and Fe (from ferric citrate [C6H5FeO7]) concentrations were tested at different pH (6–8), as shown in the Supplementary Table S1, based in previous works [30, 31]. Briefly, for N, four levels of inorganic (NH3) and/or organic N (Tris) were tested: Non-N [0 mM], 0.8 mM NH3, 1 mM Tris, and 1 mM Tris +0.8 mM NH3; for PO43−, two levels were tested: [Low PO43−, 0.005 μM] and [High PO43−, 50 μM], as well as for Fe, [Low Fe, 1 nM] and [High Fe, 1 μM]. All these concentrations were combined in three different experimental designs: (I) long-term studies (12 days) performing growth curves (where growth rates (μ) and duplication times (Tg) were measured) at pH 8 and (II) and (III) short-term studies (96 h), testing N, and PO43− concentration effects at different pH levels, respectively. During the experiment, subsamples were taken to measure growth, alkaline phosphatase activity (APA), and reactive oxygen species (ROS). Besides, (IV) we measured APA under different levels of PO43− (0.005–100 μM) and Fe (1 nm and 1 μM) at pH 8 after 96 h (Supplementary Table S1).

Flow Cytometry and Growth Responses

Flow cytometry was used to assess changes in cell abundances. For long-term studies, subsamples of the culture (1.5 mL) were daily sampled, fixed with 36% (p/v) of formalin and froze at −20 °C until reading at the end of the experiments. For the rest of the experiments, measurements were performed without freezing nor fixing. The cytometer used was the BD FACSJazzTM, previously calibrated with the SPHEROTM Ultra Rainbow Fluorescent Particles, 3.08 μm. The cells were separated by adjusting the voltages, according to size and complexity: forward scatter (FCS) versus side scatter (SSC), respectively. Cell counts were adjusted to count for 10 s or 1000,000 events. To measure the growth rates (μ) and duplication times (Tg), we followed the equations used in Fernández-Juárez et al. (2020) [31]. To evaluate the effect of the low nutrient concentration (i.e., serial, simultaneous, or independent) of the growth response to P and Fe (with N replete), the log ratio effect-size criteria based on the mean treatment and control response was assessed as in Fernández-Juárez et al. (2020) [31].

Microscopical Analysis

Microscopical images were taken with the Leica TCS SPE confocal microscope, Leica Microsystems. The samples were placed on a clean slide with DAPI at a final concentration of 0.1 mg mL−1. The samples were kept in the dark until reading, taking the images with the ×100 objective combining the brief-field I (BF) channel. Images were processed with ImageJ software, and cell volume was calculated following Hillebrand et al. (1999) [57].

Alkaline Phosphatase Activity

Alkaline phosphatase activity (APA) was evaluated using a fluorometric assay and following the MUF-P hydrolysis [30]. At the end of the experiment, APA was tested at low levels of PO43− (i.e., 0.005 μM) with low or high levels of Fe (1 nm or 1 μM) combined with different pH levels (pH 6–8). Moreover, a battery of concentrations of PO43− (0.005, 0.05, 0.5, 5, 50, and 100 μM) combined with different levels of Fe (1 nm and 1 μM) at pH 8 was performed (Supplementary Table S1). An endpoint assay was conducted, using 5 μM MUF-P. After 1 h of incubation in darkness at room temperature, APA was measured in a microtiter plate that contained buffer borate pH 10 (3:1 of sample: buffer). With a Cary Eclipse spectrofluorometer (FL0902M009, Agilent Technologies), MUF production (fmol MUF cell−1 h−1) was measured at 359 nm (excitation) and 449 nm (emission), using a calibration standard curve with commercial MUF (Sigma-Aldrich). Results were normalized by cell number.

Structural Analysis of the Alkaline Phosphatase

Sequences of PhoD of the UIB 001 and 4B UA (locus_tag: HA399_02660 and H2O77_02640) in FASTA format were sent to the I-Tasser server for protein 3D-structure prediction [58], with their domains previously checked in Pfam 32.0 [59]. The predicted structures were sent to POSA for a structural alignment between them [60]. The description of the Ca2+ and Fe3+ coordination positions of these proteins was based on the descriptions of the catalytic center of the PhoD from C. amphilecti KMM 296 (WP_043333989) [61] and the PhoD crystal structure of Bacillus subtilis (PDB: 2YEQ). Residues from PhoD from UIB 001 and 4B UA strains against the PhoD of C. amphilecti KMM 296 were mapped through alignments with Uniprot Clustal Omega [62]. The predicted structures and the corresponding structural alignments were visualized with Pymol [63].

Reactive Oxygen Species Production

ROS detection was measured with the molecular probe 2′,7′-dichlorofluorescein diacetate (DCFH-DA, Sigma) [31]. Briefly, bacterial samples were placed in a 96-well microplate (Thermo Scientific) spiked with DCFH-DA at a final concentration of 15 μg mL−1. The green fluorescence product, 2′,7′-dichlorofluorescein (DCF), generated after its oxidation by ROS was measured at 25 °C in the Eclipse spectrofluorometer (FL0902M009, Agilent Technologies) for 1 h with an excitation of 480 nm and emission of 530 nm. The results were expressed as the slope of the linear regression obtained and normalized by cell (arbitrary units [AU] cell−1). DCFH-DA was added in ASW without cells as blanks under the same conditions.

Statistical Analysis

A parametric univariate analysis of variance (ANOVA) factor and post hoc (Bonferroni) was used to study the effect of N, P, and Fe concentrations in the UIB 001 and 4B UA strains. The statistical analyses were performed using the SPSS software v21 (IBM Corp year 2012).

Results

Characterizing Cobetia Sp. UIB 001 and 4B UA

Genomic Analyses

Based on 16S rRNA, ANIb, and GGDC analyses, we revealed that both strains from the endemic Mediterranean seagrass P. oceanica and the endemic HCS seagrass H. chilensis, respectively, corresponded to the same species and belonging to the Cobetia genus (Table 1, Fig. 1, Supplementary Tables S2, S3 and S4). The 16s rRNA analysis of the UIB 001 and 4B UA strains showed a similarity between 99.5 and 99.9% with the type strains C. amphilecti KMM 1516T, C. litoralis KMM 3880T, Cobetia marina JCM 21022T, and C. pacifica KMM 3879T (Supplementary Table S2). Besides, the 16S rRNA phylogenetic tree showed a clade formed by these species, supported with high bootstrap values (Fig. 1). Based on these results, the UIB 001 and 4B UA strains seem to be more closely related to C. amphilecti and C. litoralis (Supplementary Table S2). In agreement with the 16s rRNA analysis, ANIb and DDH analyses also confirmed that strains UIB 001 and 4B UA belong to the same species, showing a similarity over 96% (ANIb) and 70% (dDDH) between them (Table 1 and Supplementary Tables S3 and S4). Comparisons with genomes of the Cobetia genus present in GenBank were carried out to provide deeper insights into their taxonomic affiliation (Supplementary Table S3 and S4). There is an important lack of genome sequences of type strains inside the Cobetia genus, being C. marina the unique species for which the type strain has been genome sequenced (JCM 21022T, accession number CP017114). The genome of C. amphilecti KMM 296, which is not the type strain, showed the highest score by ANIb and dDDH (over 96 and 70%, respectively), indicating that, presumably, our strains might be representatives of C. amphilecti (Supplementary Tables S3 and S4). However, due to the lack of type strain genomic information, it was not possible to perform an accurate final taxonomic classification at the species level. Overall, with our results, we can affirm that the strains isolated belong to the Cobetia genus, and they are not C. marina. Thus, the strains were named Cobetia sp. UIB 001 and 4B UA.

Table 1 Genomic features of Cobetia sp. UIB 001 and 4B UA
Full size table
Fig. 1
figure 1

Complete 16s rDNA phylogenetic tree of Cobetia spp. and other closely related groups, including the strains Cobetia sp. UIB 001 and 4B UA (*)

Full size image

Genomic Features

Sequencing and assembly procedures generated two complete and closed genomes. The genome sequence of the strain UIB 001 was composed of one chromosome (4,177,647 bp) and one putative plasmid (10,679 bp), while the genome of strain 4B UA was composed of one chromosome (4,319,205 bp) and two putative plasmids (4702 and 2015 bp). The main genomic features of each genome are included in Table 1 and Fig. 2A and B. The complete genome sequences were deposited at DDBJ/EMBL/GenBank under the accession numbers CP058244 to CP058245 (Cobetia sp. UIB 001) and CP059843 to CP059845 (Cobetia sp. 4B UA).

Fig. 2
figure 2

Genome BLAST atlas of the chromosomes for the strains A UIB 001 and B 4B UA, showing the overall similarity of the chromosome organization between them. Position of CDS+ and CDS−, as well as GC content, is indicated

Full size image

Genomic Differences

The pan-genome analysis determined that both strains shared 3087 homologous proteins (Fig. 3A). Functional analysis of the entire set of proteins of each strain displayed that both strains shared almost the same distribution of gene functions (Supplementary Fig. S1). Additionally, 197 proteins were exclusively present in the strain UIB 001, whereas the strain 4B UA had 278 specific proteins (Fig. 3A). It is worth mentioning that the functional category with the most representatives was “Function unknown” (Fig. 3B). Results from the SNPs analysis showed up to 66,929 SNPs between both strains. Among the “classical” proteins implicated in the adaptation and survival to low N, P, and Fe availability, we detected up to 1045 SNPs, 470 SNPs for proteins related to N-metabolism, 277 SNPs for proteins related to P-metabolism, and 298 SNPs for proteins related to Fe-metabolism (Fig. 3C).

Fig. 3
figure 3

Genomic differences between Cobetia sp. UIB 001 and 4B UA. A Pan-genomic analysis between Cobetia sp. UIB 001 and 4B UA, showing the number of shared proteins between the strains, as well as the number of exclusive proteins of each strain. B Functional categorization by COG analyses of the exclusive proteins of UIB 001 and 4B UA strains. C Number of SNPs detected in genes implicated in N, P, and Fe metabolism. The bar chart represents the number of SNPs, whereas, lines represent the number of annotated genes implicated in N, P, and Fe metabolism

Full size image

Phenotypic Features

From the phenotypic point of view, these rod-shaped bacteria (1–1.5 μm length and 0.7–0.8 μm width) were able to grow in nutrient-depleted media, i.e., in N-, P-, and Fe-depleted solid media (with 1.5% (p/v) agarose), as long as carbon (C) source was present (Table 2). In agreement with the species more related to them (C. amphilecti KMM 1516T and C. litoralis KMM 3880T), both isolated had a tolerance over 15% NaCl (yet growth was not detected at 20% NaCl) and up to 42 °C (Table 2). Overall, biochemical tests showed that both Cobetia sp., UIB 001 and 4B UA, had the same metabolic profiles and similar to the most related type strains (Table 2).

Table 2 Differential phenotypic characteristics of the Cobetia sp. UIB 001 and 4B UA and the most related Cobetia sp. species, Cobetia amphilecti KMM 1516T and Cobetia litoralis KMM 3880T, whose results were obtained from Romanenko et al. (2016)
Full size table

Long- and Short-Term Effects of Varying Nutrient Concentration (N, P, and Fe)

Effect on Morphology

Nitrogen, P, and Fe triggered differential responses on the morphology of the UIB 001 and 4B UA strains (Fig. 4A–I). With N2 as the sole N source, rod-shaped cells became more circular (≃1.2 μm length and 0.97 μm width) than under optimal nutrient conditions (≃1.16 μm length and 0.76 μm width) (Fig. 4C, D, and I). Under P/Fe or total nutrient (N, P, and Fe) depletion, cells became larger and wider (≃1.68 μm length and 0.95 μm width) than under optimal concentrations (≃1.16 μm length and 0.76 μm width) (Fig. 4E–I). Overall, under low nutrient concentrations, both strains’ volume (μm3) significantly increased (Fig. 4I).

Fig. 4
figure 4

Microscopic analysis of Cobetia sp. UIB 001 and 4B UA under different N, P, and Fe nutrient regimes. A Optimal condition, +N + P + Fe + C, for the UIB 001 strain. B Optimal condition, +N + P + Fe + C, for the 4B UA strain. C N2 as the sole N source, −N + P + Fe + C, for the UIB 001 strain. D N2 as the sole N source, −N + P + Fe + C, for the 4B UA strain. E P- and Fe-depleted condition, +N−P−Fe + C, for the UIB 001 strain. F P- and Fe-depleted condition, +N−P−Fe + C, for the 4B UA strain. G Nutrient-depleted condition, −N−P−Fe + C, for the UIB 001 strain. H Nutrient depleted condition, −N−P−Fe + C, for the 4B UA strain. I Length, width, and volume changes derived from changes in nutrient regimes for Cobetia sp. UIB 001 and 4B UA. For A–H, microscopical images were taken at ×100 with 4.73 of zoom

Full size image

Changes of N Concentrations

Long-Term Effect of N

Long-term studies testing N2 as the sole N source showed differential effects between UIB 001 and 4B UA (ANOVA, p < 0.05, Fig. 5A and B). N-depletion for UIB 001 reduced cell growth by 3-fold (μ, 0.18 d−1 and Tg, 3.87 d) compared to optimal nutrient conditions (μ, 0.51 d−1 and Tg, 1.37 d) (Bonferroni test, p < 0.05, Fig. 5B). Nevertheless, UIB 001 strain was capable to grow under this condition, showing its diazotrophic features. Although we did not find any annotated nif genes, UIB 001 possesses a cbb3-type cytochrome oxidase (locus_Tag: HA399_01480) found in N2-fixing organisms and had active N2-fixation rates (Supplementary Fig. S2A). Besides, both strains were capable of growth in Burk’s N-free medium, recommended for detecting N2-fixing organisms [64] (Supplementary Fig. S2B).

Fig. 5
figure 5

Nitrogen (N) concentration effect for Cobetia sp. UIB 001 and 4B UA. A–B Long-term effect. A Growth curve under the optimal condition and N2 as the sole N source and B growth rate (μ, d−1, as a bar chart) and duplication time (Tg, d, as a line) under the optimal condition and N2 as the sole N source. C–D Short-term (cell abundance, cell mL−1) after 96 h under different N sources (inorganic, NH3, or organic, Tris) at C pH 8 and D at pH 6 and 7. Values are the mean ± SE between the replicates. Letters (for UIB 001) and numbers (for 4B UA) indicate significant differences according to the treatment inside each strain, and asterisks (*) indicate significant differences between each strain inside each treatment, using a post hoc test (Bonferroni test, p < 0.05) after ANOVA over the whole dataset

Full size image

Short-Term Effects of N along with pH

Different N sources (i.e., inorganic, NH3, and/or organic, Tris) triggered different effects on UIB 001 and 4B UA at pH 8 after 96 h (ANOVA, p < 0.05, Fig. 5C). For both, increasing N sources had a positive impact, reaching the highest growths at 0.8 mM of NH3 and 1 mM of Tris for UIB 001 and 4B UA, respectively (Fig. 5C). However, combining inorganic and organic N sources (i.e., 1 mM Tris + 0.8 mM NH3) impaired the growth of both strains (Bonferroni test, p < 0.05, Fig. 5C). At low N levels (i.e., 0.08 mM NH3), UIB 001 responded better than 4B UA. However, organic N (i.e., 1 mM Tris) was more toxic to UIB 001 than 4B UA (Bonferroni test, p < 0.05, Fig. 5C).

Lowering pH to 6 or 7 affected growth to a higher degree than the N source itself (ANOVA, p < 0.05, Fig. 5D). Indeed, the significant differences found previously were banished when pH was dropped, without any differences between strains (ANOVA, p > 0.05, Fig. 5D), except for UIB 001 at pH 7, in which 0.8 mM of NH3 had a positive impact on cell abundance (Bonferroni test, p < 0.05, Fig. 5D).

Changes of P and Fe Concentrations

Long-Term Effect of P and Fe

Not surprisingly, higher growth rates (μ) were reached under optimal PO43− and Fe concentrations than at low PO43−, decreasing growth ≃ 2-fold for both strains, and more pronouncedly at low Fe levels (ANOVA, p < 0.05, Fig. 6A and B). Noteworthy, higher μ and lower Tg were reached by UIB 001, compared to 4B UA, independently of the nutrient levels (Bonferroni test, p < 0.05, Fig. 6B).

Fig. 6
figure 6

Phosphorus (P) and Fe concentration effect for Cobetia sp. UIB 001 and 4B UA. A–B Long-term effect studies. A Growth curve under optimal nutrient condition and low PO43− and/or Fe concentrations (0.005 μM and/or 1 nM, respectively) and B growth rate (μ, d−1, as a bar chart) and duplication time (Tg, d, as a line) curve under optimal nutrient condition and low PO43− and/or Fe concentrations. C Short-term studies (cell abundance, cell mL−1) after 96 h under different PO43− and Fe concentrations at pH 6, 7, and 8. In A-C values are the mean ± SE between the replicates. Letters (for UIB 001) and numbers (for 4B UA) indicate significant differences according to the treatment inside each strain, and asterisks (*) indicate significant differences between each strain inside each treatment, using a post hoc test (Bonferroni test, p < 0.05) after ANOVA over the whole dataset. PO43− is represented as Pi. D Log P and Fe responses. Each data point represents each replicate (red: UIB 001; blue: 4B UA) in which X-axis is the P log response and Y-axis is the Fe log response. Dashed red (for Cobetia sp. UIB 001) and blue lines (Cobetia sp. 4B UA) represent the critical threshold values (data points outside the critical values are significantly different (at p = 0.05) from P0Fe0)

Full size image

Short-Term Effects of P and Fe Along with pH

Short-term studies (during 96 h) showed that for both strains, PO43−/Fe concentrations and pH levels (6, 7, and 8) variations significantly affected cell abundances (ANOVA, p < 0.05, Fig. 6C). Overall, low concentrations of PO43− and Fe combined with different pH levels (6–8) did not significantly differ between strains (ANOVA, p > 0.05, Fig. 6C), albeit a higher tolerance to lower pH was observed for the UIB 001 strain. However, at pH 8 and low PO43− and Fe levels, UIB 001 grew as it did at high PO43− and Fe levels (Bonferroni test, p > 0.05, Fig. 6C), as opposed to 4B UA (Bonferroni test, p < 0.05, Fig. 6C). Decreasing pH levels had a dramatic effect on cell abundances (i.e., pH 6 and 7) (Bonferroni test, p < 0.05, Fig. 6C), showing that pH 8 was the optimum for UIB 001 and 4B UA strains independently of PO43− and Fe levels (Fig. 6C).

Effect of Low Concentrations of P and Fe

The responses to low PO43− and Fe availability (under long term or short term) for both strains revealed an independent effect of P and Fe (P response >0, Fe response >0 and P + Fe response >0, p < 0.05, Fig. 6D). The threshold for P and Fe was slightly lower for UIB 001 than 4B UA, displaying a better response to low concentrations of P and Fe.

Effects of P on APA and Structure of PhoD

The alkaline phosphatase activity (APA) was extremely influenced by pH at low PO43− levels, i.e., 0.005 μM (ANOVA, p < 0.05, Fig. 7A). Alkaline pH (i.e., pH 8) enhanced APA an average of 40-fold in both strains compared with the acidify conditions (Bonferroni test, p < 0.05, Fig. 7A). Different Fe levels (1 nM–1 μM) did not cause any significant differences in APA at [Low PO43−], i.e., 0.005 μM (ANOVA, p < 0.05, Fig. 7A). Increasing levels of PO43− (up to 100 μM) induced different inhibition responses on each strain (ANOVA, p < 0.05, Fig. 7B), detecting positive feedback with Fe, especially for the Cobetia sp. UIB 001 (Fig. 7B). For the UIB 001, at pH 8 and high levels of Fe (1 μM), APA did not respond to higher P levels (up to 50 μM) and required the highest PO43− level (100 μM) to inhibit its activity (Bonferroni test, p < 0.05, Fig. 7B), unlike for 4B UA in which APA at 50 μM was significantly inhibited (Bonferroni test, p < 0.05, Fig. 7B).

Fig. 7
figure 7

P-acquisition mechanisms for Cobetia sp. UIB 001 and 4B UA. A Alkaline phosphatase activity (APA) under low levels of PO43− (0.005 μM) and differential concentration of Fe (1 nm and 1 μM). B APA under different levels of PO43− (0.005–100 μM). Solid bars are low Fe level conditions, whereas the stripped bars are under high levels conditions. Values are the mean ± SE between the replicates. Letters (for UIB 001) and numbers (for 4B UA) indicate significant differences according to the treatment inside each strain using a post hoc test (Bonferroni test, p < 0.05) after ANOVA over the whole dataset. PO43− is represented as Pi. C–D Protein structural analysis base on the description of the PhoD of C. amphilecti KMM 296. C Structural analysis of PhoD from UIB 001 (red) and 4B UA (blue) strains. D Catalytic center for PhoD: UIB 001 strain, in red (aa letters) and blue (aa structure), and 4B UA, in black (aa letters) and yellow (aa structure)

Full size image

Structural analysis revealed the potential protein structure and the catalytic center for both PhoD of UIB 001 and 4B UA (Fig. 7 C and D). PhoD sequences shared an identity of 88%, while at the structural level, they shared an identity of 76%. Both PhoD had the same catalytic center as predicted for the PhoD of C. amphilecti KMM 296 (WP_043333989): ASP (241/238), TYR (244/241), ASP (306/303), HIS (308/305), ASN (312/309), ASN (313/310), ARG (375/372), ASP (472/469), and HIS (474/471) (Fig. 7D).

Effects of N, P, and Fe on ROS Production Along with pH

Different pH levels had a significant effect on ROS production for both strains (ANOVA, p < 0.05, Fig. 8A). Alkaline pH (i.e., pH 8) reduced an average of 15-fold ROS compared with the acidify conditions (Bonferroni test, p < 0.05, Fig. 8A). At pH 8, ROS production was affected by the different sources of inorganic and organic N (ANOVA, p < 0.05, Fig. 8B). However, no significant differences were achieved between strains (ANOVA, p > 0.05, Fig. 8B), decreasing ROS production with an increase of N (ANOVA, p < 0.05, Fig. 8B). Indeed, different PO43− and Fe levels did not have any effect on ROS production for both strains (ANOVA, p > 0.05, Fig. 8C), except for 4B UA at high concentrations of PO43− and Fe, with ROS production decreasing significantly compared to [Low PO43−-High PO43−] condition (Bonferroni test, p < 0.05, Fig. 8C).

Fig. 8
figure 8

Reactive oxygen production (ROS) production for Cobetia sp. UIB 001 and 4B UA. A At different pH levels (6–8). B At different concentrations of N source. C At different combinations of PO43− and Fe. Values are the mean ± SE between the replicates. Letters (for UIB 001) and numbers (for 4B UA) indicate significant differences according to the treatment inside each strain using a post hoc test (Bonferroni test, p < 0.05) after ANOVA over the whole dataset. PO43− is represented as Pi

Full size image

Discussion

For decades, the study of microbial biodiversity has sought to determine why microbes live where they live, embracing the tenet that “Everything is everywhere, but the environment selects” [33]. Here, we isolate and identify two strains of the same species (i.e., as endophytic and epiphytic bacteria, Cobetia sp. UIB 001 and 4B UA) which thrive on two different seagrasses in contrasting environments: Posidonia oceanica found in the Mediterranean Sea (MS) and the Heterozostera chilensis (formerly H. nigricaulis) found in the Humboldt current system (HCS). As noticed before, the MS is an oligotrophic semi-enclosed sea, where P. oceanica is widely extended, while the HCS is a nutrient-rich cold current where H. chilensis thrives in just a few locations, settled relatively recently [21]. Our results suggest that regardless of the contrasting environments considered, the same endophytic and epiphytic bacteria can perform analogous functions and adapt similarly to stressful scenarios.

After culturing the whole epiphytic community of both seagrasses, the gamma-proteobacteria Cobetia sp. (UIB 001 and 4B UA) arose above all in both seagrasses. Although we made an effort to taxonomic classify these isolates, it was only possible to determine that both strains are the same Cobetia species, presumably C. amphilecti, and they do not belong to the Cobetia marina species. Due to the lack of type strain genome sequences of the genus Cobetia, precise taxonomic classifications cannot be performed from the genomic point of view. From a genomic view, high G + C % contents were detected in both strains (i.e., 62–63%), which has previously been correlated to N availability [65]. This means that the ecological niche of Cobetia sp. UIB 001 and 4B UA are rich in N, probably by the high rates of N2-fixation detected in P. oceanica (i.e., higher than the rest of tropical seagrasses) [9] and the high N availability in the HCS [19]. Even though Balabanova et al. (2016ab) did not show physiological data, it revealed the potential genomic adaptation to the changing nutrient regimes of C. amphilecti, reporting different metabolic and biochemical profiles between differential strains placed in different marine niches [66, 67]. Yet, this work reveals that the Cobetia sp. isolates respond differently to seawater conditions. Thus, we hypothesize that these different traits might be the consequence of a different genomic evolution subjected to environmental pressures that led to the development of specific proteins and the significant number of SNPs that were found.

According to classical classifications [68, 69], both Cobetia sp. strains (UIB 001 and 4B UA), might belong to the group of oligotrophic bacteria, as they were capable of growing with minimal content of organic matter and nutrients (i.e., 1–15 mg of C mL−1 and N2 as the sole N source). Indeed, these strains may be considered as extreme oligotrophic bacteria, as they were able to grow with N2 as the sole N source, and P- and Fe-depleted conditions, being just dependent on the carbon (C) source (e.g., citrate or glucose), as metabolic tests revealed. Oligotrophic bacteria usually have small sizes (≤ 1 μm) to optimize the low availability of resources available by increasing their S/V ratio. However, microbes can expand their size to increase the cell surface area to accommodate enough uptake sites to meet uptake demands [70], as observed in both Cobetia sp. Through this plasticity to respond to the changing environment, bacteria play a key role in marine food webs and nutrient recycling (e.g., accumulation, export, remineralization, and transformation of nutrients), being a reservoir for nutrients as C, N, and P [71, 72], as Cobetia sp. does for the sustenance and maintenance of both seagrasses.

The ability to grow with N2 as the sole N source and the N2-fixation activity found in the UIB 001, remarks its potential role as N2-fixer, as previously seen in other Cobetia spp., isolated as hydrocarbon-utilizing microbiota [73, 74]. Although non-putative N2-fixation genes annotated were found in their genomes (i.e., UIB 001 and 4B UA), genes typically involved in N2-fixing bacteria as ADP-ribosyl (dinitrogen reductase) glycohydrolase (draG), which products can regulate the nitrogenase complex (i.e., formed by NifH, NifK, and NifD) [75], and cbb3-type cytochrome c oxidase, required for symbiotic N2-fixation [76], were found in both bacteria. This observation may suggest that the UIB 001 might contain alternative genes for N2-fixation. Noteworthy, the utilization of trace amounts of atmospheric ammonia through high-affinity ammonium transporters (AmtB) might be also a potential source of N for extreme oligotrophic bacteria [77], which explains why isolated 4B UA was capable of growth in (semi)solid media with N2 as sole N source.

In vast oligotrophic areas of the ocean, N2-fixation is limited by P [78], while, in eutrophic coastal areas, P can trigger harmful algal blooms [79]. In our experiments, only the 4B UA strain isolated from the eutrophic HCS was P limited, as opposed to the oligotrophic MS isolate (UIB 001) that was relatively unaffected. Contrariwise, APA of the UIB 001 strain was prolonged with a higher concentration of PO43− (without enhancing ROS production, unlike the 4B UA), suggesting a higher P-demand to achieve homeostasis. Previously, it was observed that under low concentrations of P, bacteria associated with P. oceanica (e.g., Halothece sp. species) were able to release dissolved inorganic phosphorus (DIP) from dissolved organic phosphorus (DOP) through the alkaline phosphatases (APases), which are expressed and activated through the PhoB-PhoR system [30]. Alkaline phosphatase D (PhoD) varied between the two strains at the sequence and structural levels, 88% and 76%, respectively, pointing to the structural differences as the most relevant in the differential APA, due to the lack of differences at the catalytic center.

Both APases are Fe-dependent (such as antioxidant enzymes or the nitrogenase complex [27, 80]) and are therefore impacted by the extended iron limitation throughout the ocean, including the MS and the HCS [17]. For both strains, APA was enhanced by increasing Fe levels since we predict in our PhoD model the coordination with one Fe3+ atom. However, high Fe levels did not induce ROS production at the optimal pH 8. Ferric iron, Fe3+, is insoluble at neutral pH, but at a lower pH, its solubility increases. At acidified pH, organic ligands have less affinity for Fe, reducing Fe absorption and thus affecting cell growth [81]. Moreover, ROS production was enhanced under acidic pH (i.e., pH 6), possibly by lower intracellular Fe content for antioxidant enzymes (e.g., superoxide dismutase, catalase, glutathione reductase, or glutathione peroxidase).

The increasing carbon dioxide (CO2) concentrations emitted to the atmosphere are irremissibly increasing ocean acidification (OA), posing serious risks to marine life [82]. However, lower pH levels may ameliorate some of the concomitant impacts of OA, i.e., reducing the impacts of un-ionized NH3 on the activity of some cytosolic enzymes [83]. Both Cobetia sp. were significantly affected by high N sources at pH 8, but not at pH 6 or 7, not accompanied by ROS increments, suggesting that antioxidant enzymes were not affected by un-ionized NH3. Moreover, it has been observed that OA may play a significant role in P eutrophication [84]. Previous works pointed to the sensitivity of the APA of Cobetia spp. exposed to acidification [61, 85, 86].

Our multifactorial design showed that the same species growing in different ocean basins (i.e., MS or HCS) displays different physiological and biochemical responses to different nutrient regimes (N, P, and/or Fe). Not surprisingly, the Mediterranean strain (exposed to oligotrophic water of the MS) coped with low N and P concentrations more efficiently. Nonetheless, the appearance of oligotrophic bacteria in nutrient replete waters (i.e., 4B UA) may point to the origin of this species, as its host, H. chilensis, reached the Chilean coast after long-distance dispersal from its native East Australian Current [21], characterized by warm waters and oligotrophy. Moreover, this species settled ~30°S, where a drop in primary production has been associated with the poleward migration of the South Pacific Anticyclone [87], pointing to substrate availability as the more important factor controlling bacterial activity and abundance in the HCS [88].

Conclusions

In summary, the present study brings new clues to better understand how biotic (host) and abiotic (i.e., nutrient levels and pH) interactions affect the plasticity and phenotypic acclimation of epiphytic bacteria thriving in contrasting environments, pointing to the intriguing relationships between epiphytes and seagrasses on the nutrients cycling around the global ocean.