Abstract
Phosphorus (P) limits primary production in regions of the surface ocean, and many plankton species exhibit specific physiological responses to P-deficiency. The metabolic response of Micromonas pusilla, an ecologically relevant marine photoautotroph, to P-deficiency was investigated using environmental metabolomics and comparative genomics. The concentrations of some intracellular metabolites were elevated in the P-deficient cells (e.g., xanthine) and genes involved in the associated metabolic pathways shared a significant motif in the non-coding regions of the gene. The presence of the conserved motif suggests that these genes may be co-regulated, and the motif may constitute a regulatory element for binding a transcription factor (i.e., Psr1). A putative phosphate starvation response gene (psr1-like) was identified in M. pusilla with homology to well characterized psr1/phr1 genes in freshwater algae and plants. This gene appears to be present and expressed in other marine algal taxa, such as the abundant haptophyte Emiliania huxleyi, in chronically phosphorus-limited field sites. Results from the present study have implications for our understanding of taxon-specific roles in mediating P cycling in the ocean.
Introduction
Phosphorus (P) is found in lipid membranes, genetic material, and energy storage compounds, making it a critical element for life. Because most marine microorganisms preferentially take up P as inorganic phosphate (PO43−), concentrations of dissolved PO43− are low in most of the surface ocean (<1 μM; ref. 1) thereby limiting phytoplankton productivity (2).
Marine microbes respond to chronic P-deficiency by investing resources in organic and inorganic P uptake (3–7), remodeling cellular metabolism and structures (8–13), and storing P (14,15). Different mechanisms for enacting these responses have evolved in widely distributed and often sympatric microbial taxa. For example, within the diatom, Phaeodactylum tricornutum, proteomics indicated broadly depressed metabolic activity as a response to P-starvation, with down-regulated energy metabolism, amino acid metabolism, nucleic acid metabolism, and photosynthesis, and up-regulated protein degradation, lipid accumulation, and photorespiration (16). In contrast, the prymnesiophytes Prymnesium parvum and Emiliania huxleyi, and the dinoflagellates Prorocentrum donghaiense and Amphidinium carterae, maintained energy-generating processes (i.e., photosynthesis) and carbon fixation in response to P-deficiency (17–20). Interestingly, P. donghaiense increased nitrate assimilation under P-deficiency (20), whereas E. huxleyi reduced nitrate uptake with a tight coupling between P and nitrogen (N) pools (19). These varied strategies all minimize non-critical P utilization and maximize P uptake, and thus play important roles in structuring phytoplankton assemblages and P biogeochemical cycling in the oceans (11, 19, 21–23).
Picoeukaryotes, including the abundant and cosmopolitan groups Micromonas and Ostreococcus spp., are members of the “green” lineage (Chlorophyta) and are important marine primary producers (24–26). Recent culture experiments with Micromonas pusilla indicated that this group is likely to be successful under conditions of increased acidification and low nutrient (P) availability, yet the mechanisms behind their success are elusive (27). Similar to other phytoplankton (4), M. pusilla may remodel lipids to increase non-P containing lipids (28) and up-regulate genes for P-transporters and polyphosphate accumulation (29). However, the broader physiological response of M. pusilla and other prasinophytes to P-deficiency has not been fully explored.
The green algae, Chlorophyta, share a more recent common ancestor with land plants than other marine algal groups such as diatoms and haptophytes (30,31). Consequently, their physiological response to P-deficiency may share more traits with land plants than with other algal taxa. Indeed, P-deficiency is common in lakes and terrestrial systems (32). The physiological response of model chlorophytes such as the mustard plant, Arabidopsis thaliana, and the freshwater green alga, Chlamydomonas reinhardtii, to P-deficiency led to the identification of a phosphate starvation response gene described as psr1 in C. reinhardtii (33) and as phr1 in plants (34). This gene encodes a transcription factor (TF), a protein that binds to specific DNA regions to activate or repress transcription of one or more genes. The specific region of DNA to which TFs bind is typically characterized by a short repetitive nucleotide sequence, or motif, found up- or downstream of a given gene or within introns (35,36). In A. thaliana, regulatory motifs for the Phr1 protein were more abundant in known P-responsive genes than in the rest of the genome (37), linking Phr1 to the regulation of P-responsive genes. Phr1/Psr1 TFs have not been described, to our knowledge, in marine algae.
Here, we investigate the impact of P-deficiency on the physiology of the ecologically relevant picoeukaryote, Micromonas pusilla CCMP1545. Many metabolites are produced by metabolic pathways under genetic regulatory control; thus, our combined metabolomics and genomics approach provides mechanistic insight into the physiological response of the organism to P-deficiency (e.g., 38, 39) and its underlying genetic regulation. We used a targeted metabolomics approach to analyze the suite of molecules produced by M. pusilla within the cells and in the external medium. In combination, we employed a comparative genetics approach to identify 1) a psr1-like gene in M. pusilla and other abundant marine phytoplankton species and 2) a potential regulatory element that may be the DNA-binding site of the Psr1-like protein in putatively P-responsive genes. We found evidence for the expression of psr1-like genes in M. pusilla and other marine algae under P-deficient conditions in cultures and in the field (i.e., Tara Oceans metatranscriptomes). These results have implications for better understanding and predicting the metabolic response of diverse phytoplankton groups to P-deficiency.
Materials and Methods
Culture of Micromonas pusilla CCMP 1545
All glassware was acid-cleaned and combusted at 450°C for at least 4 h. We grew an axenic culture of Micromonas pusilla CCMP1545 from the National Center for Marine Algae and Microbiota (Boothbay, ME, USA) in L1-Si media (https://ncma.bigelow.org/algal-recipes) with 0.2-μm filtered (Omnipore filters; Millipore, MA, USA) autoclaved seawater from Vineyard Sound (MA, USA). We maintained the cultures at 22°C under a 12:12 light:dark regime (84 μmol m−2 s−1). After at least three generations, we split the culture into two parallel cultures of L1-Si media amended with (a) 36 μM phosphate (P-replete) and (b) 0.4 μM phosphate (P-deficient). We completed three transfers of these parallel cultures prior to this experiment (Fig S1 a,b). For each media type, we inoculated 9 flasks of media with exponential-phase cells to achieve 320 ml and ~300,000 cells in each flask (33 ml P-replete; 37 ml P-deficient), with three cell-free control flasks. We grew cultures for two weeks (Fig S2a, b, c) and removed samples approximately 1 hr into the light cycle at the time of experimental setup (T0; day 0), in rapid growth phase (T1; day 4 P-deficient; day 6 P-replete), and in stationary phase (T2; day 5 P-deficient; day 13 P-replete).
At each sampling point, we removed 1 ml for cell counts and chlorophyll fluorescence, 30 ml for total organic carbon (TOC; Methods S1), and 20 ml of filtrate (see below) for nutrients. We monitored cell abundance daily by flow cytometry (Guava, Millipore) and assessed photosystem II efficiency by measuring the variable and maximum chlorophyll fluorescence (Fv/Fm) using fluorescence induction and relaxation (FIRe; Satlantic LP, Halifax, Cananda). We used chlorophyll a (692 nm) and side scatter of M. pusilla cultures to optimize settings for flow cytometry analysis. We assessed potential bacterial contamination by viewing DAPI-stained cells of each time point (Methods S1).
Metabolite extraction and instrument methods
We processed cultures for intracellular and extracellular metabolite extraction as described previously (ref. 40; Methods S1). For intracellular metabolites, we collected cells via filtration on 0.2 μm PTFE filters (Omnipore, Millipore) and stored them at −80°C. Metabolites were extracted in 500 μl of cold 40:40:20 acetonitrile:methanol:water with 0.1 M formic acid. Extracts were reduced to near dryness under vacuum centrifugation and reconstituted in 500 μl of 95:5 MilliQ water: acetonitrile with deuterated biotin (final concentration 0.05 μg ml−1) added as a high-performance liquid chromatography (HPLC) injection standard; 100 μl of the extract was used for targeted metabolomics analysis. We extracted extracellular metabolites using solid phase extraction (SPE) with 1 g/6 cc PPL cartridges (BondElut, Agilent, Santa Clara, CA, USA) as described previously (41) and modified by Longnecker (42) (Methods S1). Targeted metabolomics analysis was performed as described in Kido Soule et al. (43). All metabolomics data are available in the MetaboLights database (http://www.ebi.ac.uk/metabolights/) with accession number MTBLS295.
R statistical software (v3.0.2; R Core Team) was used for statistical analyses. We used a Welch’s two sample t-test to compare the cell-normalized concentrations of specific metabolites (log transformed) between P-deficient and P-replete treatments.
Characterization of the phosphate starvation response (psr1) gene and DNA-binding motifs
We used the psr1 gene sequence from C. reinhardtii (ref. 33, NCBI XM_001700501.1) to query the genome of M. pusilla using the BLAST option through Integrated Microbial Genomes (IMG, blastx, e-value: 1e−5). Conserved domains within putative sequences were characterized using NCBI conserved domain search (CD-search). We then used the putative M. pusilla psr1-derived amino acid sequence to query several databases using BLAST (tblastn, e-value: 1e−5), including: a re-assembly of the Marine Microbial Eukaryote Transcriptional Sequencing Project (MMETSP) (44), NCBI non-redundant protein sequences (nr), the NCBI metagenome proteins (env nr), the Tara Oceans eukaryotic unigenes (45), and several eukaryotic algal genomes from IMG (Methods S1). Visualization of the Tara Oceans queries were performed using the basemap library in Python and custom Python scripts (Methods S1). Inspection of sequences were performed with multiple sequences alignment with C. reinhardtii psr1-derived amino acid sequence. Sequence alignment and phylogenetic analysis was performed in MEGA 5.2.1 (46) using MUSCLE (47) with default settings in MEGA (Methods S1).
We searched for a regulatory binding element, or motif, for the Psr1-like protein within a now-archived gene model of M. pusilla CCMP1545 (FrozenGeneCatalog_20080206 (ver 1), Table S1), then again within the updated gene model from Bachy et al. (29) (Table 1). Specifically, we analyzed gene sequences of interest for a regulatory motif using the motif-based sequence analysis tools Multiple Em for Motif Elicitation (MEME Suite) 4.10.1 (48). The genes of interest (Table 1) were selected based on metabolites that were elevated or depressed in concentration in the P-deficient cultures although not necessarily statistically different between treatments (i.e., malate, several amino acids, vitamins, Fig 2) and gene expression information from Bachy et al. (29). These included genes related to central carbon metabolism, lipid metabolism, vitamin metabolism (pantothenic acid, folic acid), and nucleotide metabolism (aspartic acid). We also included genes within the carbohydrate and lipid metabolic pathways based on published work (49, 50) as well as the gene for proline oxidase (POX), a key enzyme up-regulated in E. huxleyi in response to N- and P-deficiency (19, 51). After initial motif discovery, we optimized our putative motif by varying discovery parameters such as motif width and number within genes of interest and within genes that were not expected to be regulated in response to P-deficiency, including asparagine synthetase (JGI gene ID: 9681548) and acetyltransferase like/FAD linked oxidase (9687568)). We attempted to identify motifs using the motif comparison tool (Tomtom; ref. 52) against the A. thaliana database (Protein Binding Microarray (PBM) motifs, (36)) and significant matches were used to guide the optimization of motif discovery settings and inclusive sequences.
Using the selected sequences and optimized parameters, we searched for motifs that are overrepresented in the query sequences (“positive”) relative to a set of background sequences (“negative”), the latter we defined using the genome scaffolds. These positive and negative sequences were used to create a position specific prior (PSP) distribution file (48). Discriminative motif discovery used the PSP file and the set of M. pusilla gene sequences of interest (parameters: minw = 5, maxw = 10, nmotifs = 3, mod = oops (one occurrence per sequence)). Each gene sequence consisted of concatenated 500 nucleotides upstream and downstream of the gene as well as untranslated regions (UTRs) and intronic regions, thereby excluding the coding regions.
Two recent publications provided transcriptomes for testing our hypotheses regarding expected DE genes in P-deficient and P-replete conditions. Whitney & Lomas (53) published transcriptomes of M. commoda (formerly M. pusilla) RCC299, while Bachy et al. (29) published transcriptomes of M. pusilla CCMP1545. We utilized these data to test our hypotheses based on results of our metabolomics analyses with M. pusilla CCMP1545 and in silico genomic analysis. We performed the same protocol for motif discovery and identification on gene sequences from M. commoda RCC299 (gene model: Micromonas pusilla NOUM17/FrozenGeneCatalog_20090404) and CCMP1545 (gene model: Micromonas pusilla CCMP1545/MBARI_models(ver_1)).
Results
Metabolic response of Micromonas pusilla CCMP 1545 to phosphorus deficiency
M. pusilla grew in P-replete (0.45 day−1) and P-deficient (0.03 day−1) media (Fig S2). Total organic carbon concentrations mirrored growth curves, increasing substantially in the P-replete cultures, and increasing slightly in the P-deficient cultures (Fig S2). Inorganic nutrient concentrations and photochemical efficiency generally decreased over time in both treatments (Results S2).
Metabolites were generally less abundant under P-deficiency in the intracellular (Fig 1) and extracellular (Fig S3) fractions, although several intracellular metabolites were most abundant during rapid growth (T1) under P-deficiency, including nucleobases, amino acids, carbohydrates, and vitamins (Fig 1). Only a few intracellular metabolites showed significantly different concentrations between treatments: (a) the nucleobase xanthine, which was significantly elevated under P-deficiency (t-test, p = 0.01), and (b) the amino acid derivatives, N-acetylglutamate and tryptamine, which were significantly higher in P-replete samples (t-test, p = 0.01, p = 0.02, respectively) (Fig 1). We observed significant decreases in the intracellular ratio of purine nucleosides to their nucleobases under P-deficiency at T1 (Fig 2). Intracellular malate, succinate, and citrate exhibited divergent responses, although their abundances were not significantly different between treatments (Fig 1). Malate was more abundant in the P-deficient cells, while succinate concentrations were similar between treatments and replicates and citrate was detected in only one of three replicates in each treatment (Fig 1). We noted similarly varied responses to P-deficiency in the purine nucleosides xanthosine and inosine, with invariant xanthosine abundances but higher average concentrations and variability of inosine (Fig 1).
Metabolite concentrations are, in part, determined by genetic regulation of the enzymes producing or degrading them and we hypothesized that the varied metabolite responses could provide insight into the regulation of these genes. The description of a phosphate starvation response (psr1) gene containing a myb DNA-binding domain, led us to investigate the presence of psr1, and a myb-like motif in genes that could be regulated by the psr1 gene product in M. pusilla and other marine algae.
Characterization of the phosphate starvation response (psr1) gene in marine algae
We found four statistically-significant sequences (Table 2) in response to querying the now-archived M. pusilla CCMP1545 genome with the psr1 gene from C. reinhardtii (strain cc125, NCBI accession AF174532). Only one of these sequences (JGI ID 61323) had the two characteristic myb domains of the psr1 gene in C. reinhardtii (myb-like DNA-binding domain, SHAQKYF class (TIGR01557) and myb predicted coiled-coil type transfactor, LHEQLE motif (pfam PF14379)) (Fig 3a). The glutamine-rich regions, characteristic of TF activation domains, and putative metal binding domain originally described for psr1 in C. reinhardtii(33) were not detected in the putative M. pusilla psr1 gene, thus we refer to this as a psr1-like gene. Conserved domain analysis (54) revealed two other domains in the putative psr1-like-derived amino acid sequence from M. pusilla: 1) PLN03162 super family (NCBI cl26028), a provisional golden-2 like transcription factor, and 2) a DUF390 super family domain (NCBI cl25642), comprising proteins of unknown function.
We identified putative psr1-like genes or transcripts in field datasets (Fig 3b, 4, 5, Table S2), representing diverse taxa such as Dinophyta (e.g., the symbiotic clade C Symbiodinium sp.), and Haptophyta (e.g., the coccolithophore E. huxleyi). These representative sequences did not necessarily cluster by taxonomic origin (e.g., cluster III, Fig 4). Transcripts from MMETSP and Tara Oceans were generally short sequences and often contained one of the two characteristic domains for psr1-derived amino acid sequences (Results S2). Field-derived transcripts of the psr1-like gene were geographically dispersed in the Tara Oceans dataset with the highest relative expression in surface samples and in the North Atlantic Ocean and Mediterranean Sea (Fig 5a). The majority of transcripts from the Tara Oceans dataset occurred in samples with low phosphate concentrations (≤ 0.5 μM) but there was no significant correlation between psr1 expression and phosphate concentration (Fig 5b). We observed a significant positive relationship between the presence of Micromonas psr1-like genes and the relative expression of one of the TCA cycle genes in Micromonas, fumarase, which were up-regulated in both culture-based studies of M. pusilla (29) and M. commoda (53) under P-deficiency (Fig 5c, Discussion S3). We never found genes with homology to the M. pusilla putative psr1-like gene, in genomes or transcriptomes within diatoms, the Heterokonta superphylum or the cryptophytes.
Putative regulatory element in Micromonas pusilla genes of interest
MEME analysis yielded a significant motif in M. pusilla genes involved in central carbon metabolism, lipid metabolism, and nucleotide metabolism (Fig S4). The conserved motif is similar in nucleotide sequence to several TF-binding sites in A. thaliana (Table S3, Fig S5; 36), including myb-like TF-binding motifs. Several iterations of this analysis using genes not tied to the metabolites of interest yielded either no significant motif or motif sequences lacking similarity to myb-like domains. We observed a similar conserved motif in DE transcripts in M. pusilla CCMP1545 under P-deficiency (29) (Table 1, Fig S4). While the actual nucleotide sequence recognized by a TF must be experimentally determined (36), the presence of a significant motif in each model and only in the expected genes, indicates that this motif is likely biologically significant in M. pusilla. Specifically, this motif, with similarity to myb-like domains in A. thaliana, may represent a regulatory element where the Psr1-like protein binds.
Comparison of metabolomics-based predictions of P-responsive genes in M. pusilla CCMP1545 to transcriptomics analysis
We queried the DE transcripts in M. pusilla CCMP1545 (29) and compared these to our gene predictions based on our metabolomics data. As Bachy and colleagues (29) used a more recent gene model than our original analysis, we confirmed the identity of the psr1-like gene in the updated gene model (Fig S5). We analyzed a similar set of genes in search of a regulatory element where the Psr1-like TF might bind (Fig S4), including DE genes in the P-deficient transcriptome relative to the replete (Table S1 from 29). Under P-deficiency, the psr1-like gene (JGI gene ID 360) was highly up-regulated (log-2 fold change = 4.3) and nearly all of the predicted P-responsive genes were DE (Table 1). In contrast to the M. commoda P-deficient transcriptome, M. pusilla exhibited many DE genes in the tricarboxylic acid (TCA) cycle (Results S2) but not pyruvate carboxylase (gene ID 6984). Additionally, between M. pusilla and M. commoda, different genes involved in nucleotide and lipid metabolism were DE and contained a conserved motif with similarity to myb-like regulatory element in A. thaliana (Fig 6). Lastly, the POX gene observed in M. commoda and in the archived gene model for M. pusilla, is not present, or at least not similarly annotated in the updated gene model. Instead, a copper amine oxidase (gene ID 5863) contained a significant conserved motif (Table 1, Figure S4) and is DE under P-deficiency.
Comparison to Micromonas commoda RCC299 gene expression under P-deficiency
We queried the DE genes in M. commoda RCC299 (26) exposed to P-deficient conditions (53) for the psr1-like gene as well as for genes that we expected to be regulated by the Psr1-like TF based on our M. pusilla observations and the literature (e.g., 19, 55, 56). In M. commoda, the psr1-like gene (JGI gene ID: 60184) was significantly up-regulated in the P-deficient treatment (Table S4), and we found a significant conserved motif in 18 DE genes in M. commoda. The significant motif identified in M. commoda differed in sequence from that discovered in M. pusilla and occurred in a different set of genes. However, putative binding motifs for Psr1 are present in the same pathways in the two Micromonas species and are similar to other DNA-binding motifs in A. thaliana (Fig S6 Results S2; ref. 36). The psr1-like genes in RCC299 and CCMP1545 exhibited relatively low homology to each other but were equally similar to the psr1-derived amino acid sequence in C. reinhardtii (Table S5).
Discussion
Metabolomics and comparative genomic analysis of the response of M. pusilla CCMP1545 to P-deficiency has led to three conclusions: 1) there is a shift in intracellular metabolite composition, 2) a psr1-like gene is expressed in M. pusilla and other marine phytoplankton, and 3) the genes regulated by the Psr1-like protein appear to differ amongst algal species.
Micromonas pusilla exhibits a metabolic shift in response to P-deficiency
Most intracellular metabolites did not differ significantly in concentration between treatments. Because these metabolites represent many central metabolic pathways, it appears that M. pusilla does not respond to P-deficiency through a global decrease in metabolic activity, as was observed for a diatom (16). Interestingly, we found examples of functionally-related metabolites that behaved differently. For example, glutamate, proline, and methionine were variable but higher on average under P-deficiency, while phenylalanine and tryptophan were not. In the TCA cycle, malate was variable but higher in concentration on average under P-deficiency, while citrate was lower and succinate was relatively unchanged. The opposing responses of these TCA intermediates was surprising as others (e.g., 49, 57) have shown that transcripts or proteins involved in the TCA cycle tend to exhibit a coordinated response to nutrient limitation. Additional non-uniform metabolite responses occurred, with vitamins and glucosamine-6-phosphate higher on average under P-deficiency, but other P-containing compounds (e.g. glycerol-3-phosphate) and sugars were depleted. Previous work has suggested that organisms placed under nutrient stress are capable of diverting carbon flux between pathways (reviewed in 39). Changes in carbon flux through the cell could result in coordinated abundance shifts in seemingly unrelated metabolites, as observed in this study. If the shunting of carbon is a coordinated response within M. pusilla, then we would expect genes involved in the synthesis and/or catabolism of these compounds to be co-regulated, and potentially mediated by a specific transcription factor.
Presence and expression of psr1 in marine algae
We detected a putative psr1-like gene in the genome of M. pusilla and in the genomes and transcriptomes of major algal lineages (e.g., prasinophytes, dinoflagellates). A gene annotated as psr1 is present in Ostreococcus tauri(58), although to our knowledge the potential role of this gene in the metabolism of O. tauri or other marine algae has not been discussed. The M. pusilla psr1-like gene contains two myb domains similar to those in psr1 from C. reinhardtii (33), and comparable to those in phr1 in A. thaliana (34). No psr1-like gene was identified in queried diatom transcriptomes or genomes, suggesting that a phosphate starvation response gene akin to psr1 is present in diverse but phylogenetically constrained phytoplankton groups.
We found significant similarity between the M. pusilla psr1-like gene and transcript sequences from the MMETSP and Tara Oceans datasets, indicating that this gene is expressed by marine algae in situ. High-identity hits to sequences in the GOS and Tara Oceans datasets underscore this gene’s prevalence in the oceans, particularly in regions characterized by chronically low phosphorus concentrations. Although some MMETSP and Tara Oceans sequences contained only one (of two) characteristic Psr1 domains, these short sequences did not include enough of the C-terminal end to capture the myb coiled-coil domain. Thus, further investigation is required to confirm whether the identified transcripts are from psr1-like genes. Several GOS sequences, previously described as protein of unknown function, had homology to both characteristic Psr1 domains, suggesting that they should be re-annotated as psr1-like genes. The GOS sequences are likely derived from prasinophytes (i.e., Micromonas spp., Ostreococcus spp.) and haptophytes (e.g., E. huxleyi), highlighting the need to elucidate the role of psr1 in these organisms.
Potential role of Psr1 in the marine algal response to P-deficiency
We observed a conserved and enriched motif that may function as a regulatory element across genes involved in nucleotide biosynthesis, the TCA cycle and glycolysis, carbon fixation, fatty acid metabolism, and phosphate transport or salvage. The conserved motifs discovered in M. pusilla and M. commoda had unique sequences and were present in a slightly different set of genes (Fig 6). The motif detected in M. pusilla is similar in nucleotide sequence to a myb DNA-binding domain in A. thaliana. This similarity is potentially significant, because the psr1/phr1 genes in C. reinhardtii and A. thaliana contain a myb DNA-binding domain. Thus, we hypothesize that the conserved motif detected in the Micromonas species represents binding sites for the Psr1-like transcription factors (TF).
We observed the conserved motif only in DE genes under P-deficiency by each species (29, 53), suggesting that these genes could be co-regulated. In bacteria, genes within the well-characterized Pho-regulon, for example, all contain specific sequences (the PHO box) where the transcription factor binds to activate or repress the gene (59). Different bacterial species and strains contain distinct PHO box sequences. In M. commoda, the discovered motif was not significantly similar to myb-like DNA-binding motifs in A. thaliana and was not similar to the motif discovered in M. pusilla. This was surprising at first, as we expected that the myb domain of the Psr1-like proteins in the two species would interact with similar binding regions in the genome. However, psr1-like derived amino acid sequences from M. pusilla and M. commoda were only 47% similar. TF protein sequences with up to 79% similarity have been shown to have distinct DNA-binding motif profiles (36), so these two proteins could reasonably have distinct DNA-binding motifs. Moreover, as the conserved motif was present in different sets of genes in each Micromonas species, it is likely that the Psr1-like protein regulates different genes in different taxa and may serve as a niche defining feature between these taxa. The observed variability in the psr1-like gene and the associated binding motifs may be an adaptation to distinct environments as M. pusilla was collected in the temperate English Channel, while M. commoda was isolated from the tropical Atlantic (26).
The transcriptomes of M. commoda (53) and M. pusilla (29) provided a means to test our hypotheses regarding the up- or down-regulation of certain genes in response to P-deficiency. The psr1-like gene in M. commoda and M. pusilla was one of the highest DE genes in each experiment (29, 53), suggesting that this potential TF plays a critical role in the response of Micromonas to P-deficiency. Based on previous studies, we expected genes involved in triacylglycerol (TAG) production to increase under P-deficiency, but found divergent responses between the two species. Two TAG production genes were up-regulated in M. commoda, including the penultimate step in TAG production (putative phosphatidic acid phosphatase), which contained a putative regulatory element where the Psr1-like TF could bind. By contrast, neither of the two TAG metabolism genes that were DE in M. commoda were DE in M. pusilla; instead two different genes involved in fatty acid biosynthesis and metabolism were DE in M. pusilla. Additionally, a starch-binding protein was highly up-regulated in the M. pusilla transcriptome under P-deficiency (gene ID 9633, ref. 29). These observations suggest that M. pusilla may invest resources in starch storage while M. commoda may invest in lipid storage, a testable hypothesis for future work.
Variable and non-uniform dynamics of several TCA cycle metabolites between treatments suggested that this pathway may be involved in the metabolic response to P-deficiency. Indeed, nearly all of the genes involved in the end of glycolysis and in the TCA cycle were up-regulated in M. pusilla and contained the conserved motif that may function as a regulatory element where the Psr1-like TF could bind. We propose that carbon flow through the TCA cycle may increase as a means to fuel TAG or starch production and to pull potentially damaging energy away from the photosystems (60, 61). Further work is needed to test this hypothesis and to explore the physiological implications for Micromonas spp. in situ. In contrast, only a few genes involved in the TCA cycle were up-regulated in M. commoda, including genes involved in malate and oxaloacetate (OAA) metabolism and a gene for pyruvate carboxylase (PC) which converts pyruvate to oxaloacetate. The expression of these TCA-cycle genes may reflect an increase in the malate-aspartate shuttle, a redox reaction that drives the production of NAD+ from NADH in the mitochondrial membrane. In M. commoda, it is possible that PC-based conversion of pyruvate to OAA combined with an increase in the malate-aspartate shuttle function as a stress response that allows continued glycolysis and ATP production via oxidative phosphorylation (62, 63). In this scenario, carbon flow into the TCA cycle would be limited and the genes controlling malate, aspartate, oxaloacetate and pyruvate would be co-regulated. This hypothesis is supported by the observation of a conserved motif that may function as a regulatory element in the relevant genes and by preliminary gene expression analysis of one of the TCA cycle genes that was up-regulated in both M. pusilla and M. commoda transcriptomes under P-deficiency, the fumarase gene. If fumarase expression is regulated by the Psr1-like protein, expression of these two genes (fumarase and psr1) should be correlated in field populations. We observed support for this hypothesis within the Tara Oceans dataset, although we observe weaker relationships between Psr1-like gene expression and other genes that were up-regulated in either M. pusilla or M. commoda only (Discussion S3).
We observed shifts for several purine nucleosides between treatments for M. pusilla and detected the conserved motif in the genes for a nucleoside phosphatase and aspartate transcarbamylase. These results suggest increased nucleotide salvage through purine nucleotides and nucleosides, a mechanism described in M. commoda (53) and in other phytoplankton (13, 64). Interestingly, while a 5’-nucleotidase was observed to be DE in M. commoda (53) under P-deficiency, it was not DE in M. pusilla (29). In the M. pusilla transcriptome (29), a nucleotide phosphatase was up-regulated and may function in a similar capacity in nucleotide salvage.
Another point of contrast between M. pusilla and M. commoda is the presence and expression of a proline oxidase (POX) gene. The POX gene was highlighted by Rokitta et al. (19,51) in nutrient limitation experiments (including P-deficiency) with the haptophyte E. huxleyi where this enzyme may have a role in stabilizing the mitochondrial membrane and in detecting cellular nitrogen levels. The POX gene was up-regulated in M. commoda but not present in M. pusilla. By contrast, a copper amine oxidase was up-regulated in M. pusilla. The POX gene in M. commoda and the copper amine oxidase gene in M. pusilla contained the conserved motif that may function as a regulatory element for the Psr1-like TF. The role of these genes in nutrient limitation should be targets for further investigations (Discussion S3).
The potential impact of Psr1-regulated genes on the ecological roles of Micromonas spp. is unknown, but our results suggest that the Psr1-like protein coordinates a metabolic shift in these organisms under P-deficiency, altering the intracellular flow of carbon and other elements. More comprehensive examination of these metabolic responses, which likely vary to some extent among these organisms, will be paramount to improving models of trophic carbon flow. More experiments are needed to characterize the structure and role of the psr1-like gene in Micromonas spp. and other phytoplankton, including: (1) confirming the presence of the psr1-like gene with genetic experiments, (2) determining whether the Psr1-like protein is more abundant under P-deficiency, (3) identifying the taxon-specific genes affected by the Psr1-like protein, (4) verifying the interaction of the Psr1-like protein with the hypothesized binding sites, and (5) comparing the genetic and metabolic responses to P-deficiency between organisms containing the psr1-like and those that do not. Exploring the underlying biology of the psr1 gene will facilitate mechanistic understanding of the complex metabolic response of these organisms to P-limitation and will enhance our ecological and biogeochemical predictions.
Acknowledgements
The authors would like to thank Krista Longnecker for performing the TOC analysis and Matthew Johnson and Elizabeth Harvey for assistance with flow cytometry and FIRe analysis. This research was funded by the Gordon and Betty Moore Foundation through Grant GBMF3304 to E. Kujawinski.
Footnotes
Competing interest statement: The authors declare that they have no conflict of interest