https://orcid.org/0000-0002-8330-4071
Figures
Abstract
In marine ecosystems, microbial communities often interact using specialised metabolites, which play a central role in shaping the dynamics of the ecological networks and maintaining the balance of the ecosystem. With metabolomics and transcriptomics analyses, this study explores the interactions between two marine microalgae, Skeletonema marinoi and Prymnesium parvum, grown in mono-cultures and non-contact co-cultures. As a growth indicator, the photosynthetic potential, measured via fluorescence, suggested chemical interaction between S. marinoi and P. parvum. Using Liquid Chromatography-Mass Spectrometry (LC-MS) data, we identified 346 and 521 differentially produced features in the endo- and exometabolome of S. marinoi and P. parvum, respectively. Despite limited tandem mass spectrometry data (MS2) for these features, we structurally annotated 14 compounds, most of which were previously under-studied specialised metabolites. Differential gene expression analysis was then performed on the transcriptomes of the microalgae, which uncovered differentially expressed genes involved in energy metabolism and cellular repair for both species. These metabolic changes depict the adaptation of both species in the co-culture. However, further data acquisition and investigation will be necessary to confirm the type of interaction and the underlying mechanisms.
Citation: Zulfiqar M, Abel A-S, Barth E, Syhapanha K, Poulin RX, Dehiwalage SNCW, et al. (2025) Analysis of metabolomics and transcriptomics data to assess interactions in microalgal co-culture of Skeletonema marinoi and Prymnesium parvum. PLoS One 20(7): e0329115. https://doi.org/10.1371/journal.pone.0329115
Editor: Barathan Balaji Prasath, Gujarat Institute of Desert Ecology, INDIA
Received: January 30, 2025; Accepted: July 10, 2025; Published: July 28, 2025
Copyright: © 2025 Zulfiqar et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: MZ, GP, and CS funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany's Excellence Strategy - EXC 2051 - Project-ID 390713860. (https://www.dfg.de/en/research-funding/funding-initiative/excellence-strategy; https://www.microverse-cluster.de). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: No authors have competing interests.
1. Introduction
Marine ecosystems cover almost 70% of the earth’s surface, represent 95% of the known biosphere, and are a hub of undiscovered chemodiversity [1,2]. Marine microorganisms thrive in hostile environmental conditions and are constantly adapting through the exchange of genetic material and signalling metabolites [3–5]. The diverse metabolites released from the interactions between microbes and their environment enhance stress tolerance and their survival capabilities [6,7]. These marine metabolites have potential applications in pharmaceuticals, biofuels, and agriculture [8]. Despite these benefits, research on microbial interactions is challenging due to the complexity of microbial communities, which requires inventive experimental and in silico techniques that combine existing and novel methodologies to understand the metabolic exchange involved in metabolic communication [9].
Exploration of natural products (NPs) in microbial communities is conducted in various ways. Bioassay-guided fractionation is one of the classic techniques used in drug discovery to isolate and characterise bioactive NPs [10]. Since the emergence of omics tools, such as genomics, transcriptomics, and metabolomics, meta-analyses of large-scale datasets have been extensively used to determine the taxonomic and metabolic profiles of marine microorganisms, enabling elucidation of biomolecules and their potential roles within the microbial communities [11]. Combined transcriptomics and metabolomics analysis can accelerate NP discovery by estimating the differential gene expression and differentially produced metabolites within the community [12,13]. However, deriving meaningful insights from such data analysis is challenging. Automating omics data analysis is essential for easier comprehension of biological processes and overcoming the computational resource limitations of managing and analysing omics data [14].
Co-cultures of two microorganisms could serve as a baseline to study complex interactions among microbes via omics techniques. For this project, we cultivated two unicellular marine microalgae, Skeletonema marinoi and Prymnesium parvum, in a mono- and co-culture experimental setup [15,16]. Skeletonema marinoi, a centric diatom, is known for its beneficial role in the marine food chain and biogeochemical cycling of carbon, nitrogen and silica [17,18]. Prymnesium parvum, also known as “golden algae”, is a marine and freshwater haptophyte that releases toxins called prymnesins and is involved in causing harmful algal blooms (HABs) [19,20]. Given the contrasting observed roles of the two ubiquitous microalgae, we designed a co-culture experiment based on the possibility of metabolic interactions among them. We implemented a workflow to analyse co-culture data from the endo- and exometabolome of two microorganisms. Additionally, transcriptomics data were acquired under the same conditions. Statistical analyses, including Principal Component Analysis (PCA) and Partial Least Square (PLS), were performed to identify differentially produced metabolic features and differentially expressed genes. This research is a step forward in unravelling the complexities of marine microbial interactions using omics approaches.
2. Materials and methods
2.1. Co-culture preparation
Non-contact co-culture chambers, consisting of two glass flasks separated by a 0.22 µm permeable PVDF membrane (Durapore®, Merk, USA) were acid-washed and autoclaved prior to assembly under sterile conditions. The co-culture experiment was conducted using 13 co-culture chambers (4x treatment/control (2) + 1 media blank), each with a total capacity of 600 mL [15,16]. We cultivated two microalgal species in this co-culture experiment, Skeletonema marinoi (RCC75, Roscoff Culture Collection, Roscoff, France) and Prymnesium parvum (UTEX2797, UTEX Culture Collection of Algae, Austin, TX, USA). Pre-cultures were grown for 1–2 weeks in Artificial Sea Water (ASW) media, to reach exponential growth, then normalized to the same initial chlorophyll a fluorescence (ex: 430 nm, em: 665 nm) (Varioskan Flash, Thermo Fisher Scientific,Vantaa, Finland) using ASW media. This allowed for a starting inoculation ratio of 1:1 based on chlorophyll a fluorescence.
For each co-culture chamber, 250 ± 10 mL of ASW media was added to each half chamber, followed by 50 mL of microalgal pre-cultures. Monoculture controls of both S. marinoi (NCBI-Taxonomy ID: 267567) and P. parvum (NCBI-Taxonomy ID: 97485) were established by inoculating both half-chambers with the same species. Co-cultures were established by inoculating one half-chamber with S. marinoi and the other half-chamber with P. parvum. The media blank was established by inoculating each half-chamber with ASW (no microalgae).
All pre-cultures and co-cultivation chambers were cultivated in temperature-controlled cultivation chambers at 18 °C ± 1.5 on a 14/10 h diurnal day/night cycle with a photon flux of 47−63 µmol m-2 s-1. The co-cultures were shaken and randomly repositioned daily to promote adequate diffusion of metabolites.
2.2. Sample preparation
For sampling, 1 mL syringes were used with long syringe needles daily to extract diatom cultures from the randomly located chambers. A 96-well plate was used for chlorophyll a fluorescence measurement (200 µL), and a 1.5–2 mL Eppendorf tube for sample preservation (800 µL) using 1% Lugol’s preservation solution. Sampling was conducted every other day starting from inoculation (day 0), until day 8 to monitor the growth. For sampling, a 1 mL aliquot was removed from each half-chamber under sterile conditions and chlorophyll a fluorescence measurements were taken as a proxy for cell growth. After final sampling on day 8, cultures were prepared for metabolomic and transcriptomic analysis. For transcriptomic analysis, 2 x 50 mL aliquots of each half-chamber were passed through a 2.7 µm GF/D glass microfiber filter (Whatman, Maidstone, UK) to collect microalgal cells. Filters were stored frozen at −80 °C prior to submission for transcriptomic analysis. An additional 50 mL of each half-chamber was filtered through a 2.7 µm GF/D glass microfiber filter for metabolomic analysis. Dried filters were stored frozen at −80 °C prior to endometabolomic extraction. Filtrate was extracted on 200 mg HLB SPE columns (Oasis®, Waters, Eschborn, Germany) following manufacturer’s instructions. The eluent was evaporated under vacuum to dryness, argon sparged, and stored frozen at −80 °C until analysis. Filters were extracted for endometabolites following the protocol by Vidoudez and Pohnert (2012) [21]. Briefly, filters were thawed, and 5 mL of fresh extraction mix (methanol:ethanol:chloroform, 1:3:1) was added to the filter. Samples were placed in ultrasonic bath for 10 min then centrifuged for 15 min at 15k rpm at 4 °C. Supernatant was transferred to glass vials and evaporated to dryness under vacuum. Dried extracts were argon sparged and stored frozen at −80 °C until analysis.
2.3. Metabolomics
2.3.1. Data acquisition.
Ultra High-Performance Liquid Chromatography (UHPLC) coupled with High-Resolution Mass Spectrometry (HR-MS) was carried out with Thermo (Bremen, Germany) UltiMate HPG-3400 RS binary pump, WPS-3000 autosampler, and THERMO QExactive plus orbitrap Mass Spectrometer coupled to a heated electrospray source (HESI). The separation was run on a Thermo 100 mm, C18 RP (100 × 2.1 mm; 2.6 µm) column with a gradient of 100% A (Water+2% acetonitrile + 0.1% formic acid) to 100% B1 (acetonitrile) in 8 min and keeping 100% B1 for 3 min. All biological replicates were profiled via electrospray ionisation (ESI) in both positive and negative modes with a resolution of 70,000 and a mass range of 80–1200 m/z. A pooled quality control sample, made from 20 µL of each normalised (to cell count) biological replicate, was profiled after every four biological replicates for retention time shift corrections via the Compound Discoverer software (Thermo Scientific, version 3.2). The media blank sample were also profiled via LC-MS to remove all metabolites originating from media components. The quality control samples were used to generate a master inclusion list of metabolites of interest derived from phytoplankton metabolism. This list included 1467 features and 758 features for the exometabolomic and endometabolomic samples, respectively. The master inclusion list was separated into smaller, 100-feature lists for ease of data collection. Each 100-feature list included features spanning the full retention time span of the analysis window. The quality control samples were then profiled with each inclusion list via data-dependent acquisition mode (DDA) to generate fragmentation patterns for each parent ion included in the master inclusion list. The DDA analysis was also conducted in both positive and negative modes, using a three-stepped normalised collision energy of 15, 30, and 45. Fig 1 illustrates the metabolomics data acquired from this experimental setup.
(A) The control groups were S. marinoi mono-culture and P. parvum mono-culture, while the treatment groups were S. marinoi in the presence of P. parvum and P. parvum in the presence of S. marinoi. (B) From the extracts of these groups, we acquired the endo- and exometabolome using LC-MS2, and C) RNA sequencing data.
2.3.2. Data preprocessing and statistical analysis.
The MS1 processing workflow was developed with the LC-MS data generated using the mono-culture and co-culture samples. An overview of the workflow is given in Fig 2. The LC-MS spectrometry data RAW files were converted to mzML open format processed by the GNPS “Conversion Drag and Drop” file converter [22] (accessible at https://gnps-quickstart.ucsd.edu/conversion). Preprocessing and analysis were done in R, based on the “Identifying ESsenTIal Molecular variables in Terrestrial Ecology” (iESTIMATE) repository [23] (accessible at https://github.com/ipbhalle/iESTIMATE). The processing was performed with the packages XCMS (version 3.18.0), Spectra (version 1.6.0), and MSnbase (version 2.22.0) [24]. In the following processing steps, the parameters were optimised for four different conditions: endometabolome in positive and negative modes and exometabolome in positive and negative modes. The parameters for the peak picking algorithm were optimised using the R package Isotopologue Parameter Optimization (IPO) (version 1.22.2) [25].
The first section demonstrates the MS1 preprocessing workflow, which starts with the.mzML files. These files go through the parameter optimisation and subsequent steps, such as peak detection, grouping, retention time correction, and re-grouping of peaks. The preprocessed data is then converted to a feature list which are subjected to further statistical analysis and are used to link the MS1 features to their corresponding MS2 fragmentation spectra. Separately, the MS2 Fragmentation spectra undergo Metabolome Annotation Workflow (MAW), which annotates chemical structures to the metabolic features. The final candidate list is searched in the suspect list from each origin organism to find previously identified metabolites in the new experimental setup dataset.
The first function in the workflow is the data_preparation function, which generates a descriptive table containing metadata and phenotypic data, called the phenodata table. The data were subset to an instrument-specific retention time range of 0–700 seconds and MS level = 1. Chromatographic peak picking was performed by the peak_detection function using the centWave algorithm [26] from the XCMS package (findChromPeaks). The following parameters were used for the function grouping_1, which used groupChromPeaks for grouping of peaks: minFraction = 0.7, bw = 2.5 (standard deviation of smoothing kernel). One resulting peak group was termed as a feature, representing a metabolite in the samples. Retention time correction was performed in the rt_correction function by adjustRtime with the peakGroups method. The following parameters were used: minFraction = 0.7, smooth = loess, span = 0.5, and family = gaussian. With the function grouping_2 a second grouping was performed with the same parameters as before. A feature table was created in the featureValues function. Intensities of the feature table were transformed logarithmically, and missing values were imputed to 0 in the feature_transformation function. The lower end of the normal distribution of the feature data was used as an intensity cutoff for the creation of a presence/absence table, with a 5% cutoff in the binary_list_creation function. The workflow with documentation is available at https://github.com/zmahnoor14/MAW/tree/main/co-culture [27]. The data is forwarded 1) for statistical analysis and 2) to link MS1 precursor masses to their fragmentation MS2 spectra.
For univariate statistical analysis, the chlorophyll a fluorescence was measured through 0–8 days of cell growth in mono-culture and co-culture samples. Two-way ANOVA with replicates was performed to verify if there was a significant change in the chlorophyll a fluorescence between mono-culture and co-culture conditions for each species separately. As a posthoc test, we also performed Welch t-test (two-tailed t-test with a sample of unequal variance). For the metabolomics data, diversity analysis on feature level was performed using the presence/absence matrix, measuring the metabolite richness by calculating the Shannon diversity index (H’) for each sample origin. Multivariate statistical analysis was performed on the feature tables from the MS1 workflow to elucidate differentially abundant features. Principal component analysis (PCA) was performed using the prcomp function [28]. Partial least square (PLS) analysis was performed using the f.select_features_pls function from the iESTIMATE. The parameters for model construction were principal components = 5, tune length = 10, and quantile threshold = 0.95.
2.3.3. Suspect list preparation and metabolome annotation.
For metabolite annotation, we generated, based on literature and public databases search, two separate lists of metabolites known to be produced by both microalgae, termed suspect lists. In a previous study, the suspect list for Skeletonema spp was published (https://doi.org/10.5281/zenodo.10143554; [29]). For the present research, we assembled a suspect list for Prymnesium parvum as well, using different databases: compound, reaction, pathway, and enzyme databases such as Kyoto Encyclopedia of Genes and Genomes (KEGG) [30–32], Comprehensive Marine Natural Products Database (CMNPD) [33], Chemical Entities of Biological Interest (ChEBI) [34], PubChem [35,36], Universal Protein Resource (UniProt) [37], BRaunschweig ENzyme DAtabase (BRENDA) [38], MetaboLights [39,40], MetaCyc [41,42], and different publications on P. parvum metabolites [43–48]. Suspect list was curated with RDKit [49], PubChemPy [50], and pybatchclassyfire [51]. The protocol for curation is given in the methodology section of [29].
The chemical structure annotation was performed using the Metabolome Annotation Workflow (MAW) [52]. All four conditions (endo_pos, endo_neg, exo_pos, and exo_neg) were combined into four.mzML input files for MAW. To perform dereplication, MAW-R (spectral database dereplication module) used GNPS [22], HMDB 5.0 [53,54], and MassBank [55] (https://doi.org/10.5281/zenodo.7519270) [56], and MAW-SIRIUS (compound database dereplication module with SIRIUS5 [57], using COCONUT database [58]. Different candidates from MAW-R and MAW-SIRIUS were analysed to select top candidates using MAW-Py. Chemical classes were assigned using ClassyFire [59] and CANOPUS [60]. Finally, the putative candidate list obtained from MAW was screened for common compounds within the suspect lists of both species. The workflow was executed using Docker containers with R and Python scripts (https://github.com/zmahnoor14/MAW) [61].
2.3.4. Linking annotations to experimental conditions.
Following the MS1 preprocessing, statistical analysis, and MS2 annotation workflow, the MS1 features were linked to MS2 fragmentation spectra using inclusion lists to identify the chemical structures assigned to each differentially produced metabolite. Note that here the MS1 features are termed as metabolites. To accomplish this, a function named linking_with_inc_list was implemented, which searched for MS2 precursor masses [m/z] in the inclusion list within the specified [m/z] minimum and maximum range of the MS1 feature list. If the condition was met, the retention time range from the inclusion list was used to search for the retention time median value in the MS1 feature list. The corresponding feature ID was allocated to the MS2 precursor mass when the second condition was also fulfilled. In cases where retention time median values did not precisely align within the rt min-max range, this information was noted as a comment alongside the corresponding feature ID. After establishing the link between the MS1 feature ID and the MS2 precursor mass, the precursor masses were searched in the results obtained with MAW, allowing for the assignment of feature IDs to chemical structure annotations. This was accomplished using a function called link_ann_inc, which linked the inclusion list feature ID to each precursor mass in the MAW results, applying the same conditions for precursor mass and retention time ranges. Subsequently, the MAW results were provided with the corresponding feature IDs and conditions leading to a structure assignment to differentially produced metabolites (https://github.com/zmahnoor14/MAW/blob/main/co-culture/linking.R). The complete metabolomics workflow is illustrated in Fig 2.
2.4. Transcriptomics
2.4.1. Data acquisition.
The data acquisition was performed by the commercial company GENEWIZ, from Azenta life Sciences (Leipzig Germany). Total RNA was extracted using the Qiagen RNeasy Plant Mini kit and Plant RNA isolation aid kit following the manufacturer’s instructions (Qiagen, Hilden, Germany). RNA samples were quantified using Qubit 4.0 Fluorometer (Life Technologies, Carlsbad, CA, USA), and RNA integrity was checked with RNA Kit on Agilent Tapestation (Agilent Technologies, Palo Alto, CA, USA). The RNA sequencing library was prepared using the NEBNext Ultra II RNA Library Prep Kit for Illumina using the manufacturer’s instructions (New England Biolabs, Ipswich, MA, USA). Briefly, mRNAs were initially enriched with Oligod(T) beads. Enriched mRNAs were fragmented for 15 minutes at 94 °C. First-strand and second-strand cDNA were subsequently synthesized. cDNA fragments were end-repaired and adenylated at 3’ends, and universal adapters were ligated to cDNA fragments, followed by index addition and library enrichment by PCR with limited cycles. The sequencing library was validated on the Agilent TapeStation (Agilent Technologies, Palo Alto, CA, USA) and quantified by using Qubit 2.0 Fluorometer (ThermoFisher Scientific, Waltham, MA, USA) as well as by quantitative PCR (KAPA Biosystems, Wilmington, MA, USA). The sequencing libraries were multiplexed and clustered onto a flow cell. After clustering, the flow cell was loaded onto the Illumina HiSeq 4000 according to the manufacturer’s instructions. The samples were sequenced using a 2x150bp Paired-End (PE) configuration. The HiSeq Control Software (HCS) conducted image analysis and base calling. Raw sequence data (.bcl files) generated from Illumina HiSeq was converted into Fastq files and de-multiplexed using Illumina bcl2fastq 2.20 software. One mismatch was allowed for index sequence identification. Fig 1 illustrates the transcriptomics data acquired from this experimental setup.
2.4.2. RNA seq quality control and transcriptome assembly.
Raw sequencing reads were assessed for quality using FastQC (version 0.11.9) (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Adaptor trimming, quality filtering, and read preprocessing were performed using fastp (version 0.23.2) [62]. In detail, 5’ and 3’ bases with a Phred quality score below 30 were cut, and reads with more than 1 N base, an average quality score below 30, or a length of fewer than 15 bases were removed.
Before assembly, the quality filtered reads were classified using Kraken2 (version 2.1.2) [63] using its standard index and including the genome assemblies of the diatoms Thalassiosira pseudonana (GCF_000149405.2, [64]) and Skeletonema costatum (GCA_018806925.1, [65]). All reads classified as viral, prokaryotic, and human were removed from the datasets.
Trinity (version 2.13.2) [66] was used to assemble the transcriptome of the remaining read data, keeping only contigs with a minimal length of 200 nt. After assembly, Kraken2 was applied again to filter all assembled contigs not classified as originating from the clade Stramenopiles. BUSCO (version 5.4.5) [67] was applied to evaluate the assembly quality using the transcriptome mode and the stramenopiles_odb10 evaluation dataset.
2.4.3. Transcriptome annotation and differential gene expression.
Using the tool Augustus (version 3.5.0) [68], the putative open reading frames (ORFs) encoded by the assembled transcripts were annotated with the species option applied for Skeletonema costatum and the option to search on both strands. Using InterProScan (version 5.59_91.0) [69] and based on the databases Pfam [70], TIGRFAM [71], PANTHER [72], ProSiteProfiles [73], and FunFam [74], the predicted protein sequences were further classified with their potential function, functional domains, as well as, associated GO terms and KEGG pathways. Processed reads were aligned to the assembled transcriptome using Hisat2 (version 2.2.1) [75] with standard parameters. Read counting was performed using featureCounts (version 2.0.1) [76], with the annotation described as above as a reference. Differential gene expression analysis was performed using DESeq2 (version 1.38.3) [77] statistical significance defined as FDR adjusted p-value < 0.05. The source code is available at https://github.com/Bioinformatics-Core-Facility-Jena/SE20220705_97 [78].
3. Results
3.1. Chlorophyll a fluorescence measurement as a growth indicator
In this co-culture experiment, Skeletonema marinoi and Prymnesium parvum were cultivated in a co-culture chamber as 1) control groups represented by mono-cultures of each microalga individually, and 2) treatment groups, represented by co-culture of both species grown together, but separated from each other by a permeable membrane barrier. The chlorophyll a fluorescence was measured for eight days for all groups. One of the potential mechanisms of actions of allelopathy, the process of chemical exudates from one phytoplankton to affect a co-occurring competitor, is via inactivation or modulation of photosynthetic efficiency [79]. However, we are working under the assumption of a linear correlation between the growth of microbial cells and the increase in chlorophyll a concentration. The chlorophyll a fluorescence over the eight days of culturing the microbial cells in mono-culture and co-culture for both species are shown in Fig 3, derived from S1 Table. Fig 3a shows a lower chlorophyll a fluorescence in S. marinoi when grown as a co-culture in the presence of P. parvum, compared to mono-culture controls. Fig 3b shows a steady increase in the chlorophyll a fluorescence of P. parvum, indicating no difference in the co-culture P. parvum compared with mono-culture controls, except for at day 8. Growth of all cultures throughout the experiment indicates that they were not nutrient-limited.
Both plots show the number of days on the x-axis and the chlorophyll a fluorescence measurement in relative fluorescence units (RFU) on the y-axis. All eight replicates for monocultures and all 4 replicates for co-culture were used to generate the fluorescence plot of the cultures over eight days (see S1 Table). The error bars represent the standard deviation among the mean of the replicates for the particular day. a) The plot for the chlorophyll a measurement of S. marinoi shows lower chlorophyll a fluorescence of the co-culture (purple) as compared to the mono-culture (green) as a general trend. b) The plot for the chlorophyll a measurement of P. parvum shows no distinction between the co-culture (purple) and the mono-culture (green) groups. However, there is an increase in the chlorophyll a concentration compared to the control after day 6 in the co-culture.
Based on the chlorophyll a measurements plot, it can be suggested that P. parvum negatively influences the growth of S. marinoi, while S. marinoi doesn’t influence the growth of P. parvum for the first six days of growth as a co-culture. We performed a two-way ANOVA test with replicates (Table 1 and S2 Table) to confirm this hypothesis. ANOVA was performed with 4 replicates for co-culture and 4 replicates for mono-culture for each species separately. The analysis showed that for S. marinoi, both the culture conditions (mono- or co-culture) and date of sampling (0,2,4,6,8) explained a significant amount of variation in chlorophyll a fluorescence. For P. parvum, only date of sampling (0,2,4,6,8) explained a significant amount of variation in chlorophyll a fluorescence. To compare the individual days after significant p-value for fluorescence measurements of S. marinoi cultures in ANOVA, we performed a t-test to verify which days had significant differences in the chlorophyll a fluorescence. Despite non-significant findings in the ANOVA analysis of the P. parvum growth curve, a notable visual difference in the curves on day 8 led us to conduct a t-test on the chlorophyll a fluorescence in P. parvum cultures. Table 1 shows the two-way ANOVA test p-values for Sample as source of variation, and the t-test p-values for individual days for S. marinoi and P. parvum. The t-test performed on S. marinoi fluorescence data shows significant differences (p-value =< 0.05) for days 0, 4, 6, and 8. While for P. parvum, the p-value only dropped below 0.05 at the concentrations measured for day 8, which confirms the chlorophyll a measurement plots in Fig 3.
3.2. Metabolomics analysis of endo- and exometabolome
3.2.1. Suspect lists from two marine microalgae.
For this study, a suspect list of metabolites known to be produced by P. parvum was generated using publicly available databases and literature search. The suspect list for P. parvum consists of 222 primary and secondary metabolites, together with information on formula, molecular mass, SMILES, InChI, IUPAC names, source, and ChemONT classification [59], which classifies chemical compounds based on their chemical structures into a hierarchical system. The suspect list and a sunburst plot can be accessed on Zenodo with https://doi.org/10.5281/zenodo.7864506 [80]. We also used the published suspect list of S. marinoi to obtain the list of known metabolites from this diatom [29,81], which consisted of 893 compounds from both S. marinoi and S. costatum https://doi.org/10.5281/zenodo.7519270 [56].
The secondary metabolites detected from the Skeletonema spp. suspect list found in the endo- and exometabolome were: 7-mercaptoheptanoic acid, a medium-chain fatty acid, hexadeca-6,9,12-trienoic acid, which is a fatty acyl [82], and octatrienal which is a polyunsaturated aldehyde known to be produced by marine diatoms [83], and 8,11,14,17-eicosatetraenoic acid, all of which emphasise the diversity of fatty acids of S. marinoi. Toluene, a compound commonly associated with oceanic phytoplankton contributing to the remote marine atmosphere, was detected in the S. marinoi endometabolome [84]. Finally, uteroverdine, which has a structure resembling biliverdin, was also found in the S. marinoi suspect list for the exometabolome. Lumichrome, being a blue-fluorescing compound formed by riboflavin photolysis, was present in S. marinoi’s exometabolome. Both S. marinoi and P. parvum shared the presence of fucoxanthin in their endometabolome, hinting at chlorophyll a production as part of the photosystems. Also shared between the two species were monopalmitin and monomyristin, glycerolipids previously identified in microalgae [85]. P. parvum’s exometabolome showcased viburnitol, a cyclitol [86], which can have various roles, such as signal transduction, cell wall formation, and osmoregulation, and can act as antioxidants [87].
3.2.2. Statistical evaluation of differentially produced metabolites from MS1 data.
The MS1 spectra were processed to investigate observed growth effects using metabolomics preprocessing pipeline from XCMS. Overall, 24,862 features were detected for the endometabolome and 34,671 features for exometabolome from the whole dataset. After preprocessing was performed, various statistical tests were applied to the endo- and exometabolome of the co-cultures and mono-cultures. To assess the chemodiversity, Shannon diversity (H’) index was used, where higher values indicate greater diversity. In both cases, the co-culture had a higher H’ index than the mono-cultures for the same species, depicting higher chemodiversity (S1 Fig). For exometabolome, P. parvum showed higher chemodiversity of spectral features, with H’ = 10.03 for mono-culture and H’ = 10.07 for co-culture, than S. marinoi, with H’ = 9.89 for mono-culture and H’ = 9.98 for co-culture. For endometabolome, S. marinoi showed higher diversity, with H’ = 9.59 for mono-culture and H’ = 9.60 for co-culture, as compared to P. parvum, which shows H’ = 9.45 for mono-culture and H’ = 9.50 for co-culture.
Partial Least Square (PLS) analysis was performed to assess which spectral features caused this chemical diversity and are differentially produced among different conditions of S. marinoi and P. parvum (S2 Fig). The PLS analysis of the endometabolome of S. marinoi resulted in the identification of 22 differentially produced metabolites (i.e., differentially produced features), where 7 were also present in mono-culture conditions, while 15 were only abundant in co-culture conditions. For P. parvum, all 30 differentially produced metabolites were significantly more abundant in the co-culture condition. The exometabolome of S. marinoi resulted in 324 differentially produced metabolites, of which 304 were more abundant in co-culture conditions. Similar results were obtained for the exometabolome of P. parvum, where among the 491 features, 418 metabolites were more abundant in the co-culture condition.
3.2.3. Metabolic exchange in endo- and exometabolome.
The exometabolome is shared among the two species within the extracellular space and cannot be easily traced back to the species releasing these metabolites. Both Shannon diversity index and PLS plots indicate a significant change in the exometabolome in co-culture conditions as compared to the mono-culture conditions. To verify that these effects are not mere a combined effect of the mono-culture exometabolomes of the two species in a shared co-culture environment, rather an actual change in the co-culture exometabolome, a comparative PCA analysis of the exometabolome was performed. The first PCA plot shown in S3 Fig depicts all exometabolome conditions (PC1 = 24.9%, and PC2 = 7.1%). Each condition forms a separate cluster on the PCA plot. The second PCA plot shown in S3 Fig is obtained by calculating the sum of half of the mono-culture conditions (8 mono-culture replicates for each species/ 2) to be comparable with the 4 co-culture replicates per species (PC1 = 16.7%, and PC2 = 11.6%). The combined mono-culture conditions of exometabolome clustered together, while the co-culture conditions formed separate clusters for P. parvum and S. marinoi. The comparison PCA plot verified that the changes in the exometabolome are not a combined effect of the individual species’ mono-culture exometabolome; instead, the observed difference is due to the actual adaptation or metabolic exchange occurring in the co-culture condition.
To analyse the metabolic exchange among the species in co-culture, we searched for shared metabolites in the endometabolome of the two microalgae in co-culture separately, as the exometabolome in the co-culture conditions is mixed and cannot be assigned to one particular species.
We searched for all 30 differentially abundant metabolites found in the co-culture endometabolome of P. parvum within S. marinoi conditions to identify shared differential metabolites. Among these 30 metabolites, only 7 were exclusive to P. parvum co-culture condition, suggesting that 7 of the 30 metabolites are only produced by P. parvum in the presence of S. marinoi. 10 metabolites were found in the exometabolome of mono-culture S. marinoi, suggesting that these metabolites were produced and released by S. marinoi in the extracellular environment (in both mono- and co-cultures) and when produced in co-culture, were taken up by P. parvum. 6 metabolites were found in both mono- and co-culture endometabolomes of S. marinoi, which P. parvum also takes up. The remaining 7 metabolites were shared in all the above-mentioned conditions.
A similar search was performed for the shared metabolites in the co-culture endometabolome of S. marinoi. In this case, out of 15 differentially abundant metabolites produced by S. marinoi cells (endometabolome) in co-culture conditions, 7 metabolites were found to be exclusively from S. marinoi co-culture source in general. 8 of these 15 metabolites were found in exometabolome of mono-culture P. parvum, out of which 3 were also present in both endometabolome conditions of mono- and co-culture of P. parvum. This suggests that 3 metabolites are produced by P. parvum cells regardless of S. marinoi presence, and in total, 8 of the released metabolites produced and released by P. parvum in the extracellular space as mono-culture (and also produced and released in co-culture conditions) are taken up by S. marinoi. The shared metabolites, along with the number of differentially produced metabolites and the annotated metabolites, are depicted in Fig 4.
The first section refers to the mono-culture of both microalgae in endo- and exometabolome. The metabolites are depicted in each section with a coloured outline and no colour filling. The second section refers to the experimental condition of co-culture for both microalgae in endo- and exometabolome, with coloured filling representation. With our dataset, we could only annotate the metabolite structures in co-culture conditions. The third section refers to the shared metabolites which were detected in the co-culture endometabolome of both microalgae. We do not consider the shared metabolites in extracellular space, which is represented by a dotted outline.
3.2.4. Structural annotation and classification of MS2 data.
The MS2 fragmentation spectra were acquired in data-dependent acquisition (DDA) mode for the precursor masses [m/z] in the inclusion list of selected high-intensity peaks from MS1 spectra. MAW [52] was used to annotate chemical structures from spectral databases and the COCONUT database [58] to the MS2 spectra. The number of features annotated within each MSI level is given in Table 2. No internal standards were used in this study (hence, no MSI level 1 structure identifications were assigned with MAW).
With the DDA mode, only a few differentially produced metabolites were fragmented in the MS2 acquisition. Among the significant features of the endometabolome, the data acquired by the negative mode revealed no differentially produced metabolites, while the positive mode acquired only one in P. parvum co-culture condition. 10 differentially produced metabolites were annotated from exo_pos, and 3 by exo_neg samples (Table 2). The annotation results for differentially produced metabolites from MAW, represented in Table 3, unveiled the occurrence of secondary metabolites from various chemical classes.
3.3. Transcriptomics analysis to assess differentially expressed genes across mono-culture and co-culture
Using transcriptomics data, we investigated the dynamics of gene expression in mono-cultures and co-cultures. The transcriptome assemblies for S. marinoi and P. parvum were performed with the Trinity tool [66]. Within the assembly for S. marinoi, kraken2 [63] predicted 25% of the reads to be from humans, 24% from bacteria, 38% with no hits, and ~1% from viruses and other sequences. About 12% of assembled transcripts were assigned directly to S. costatum or the group of Stramenopiles. After the quality check, Busco analysis [67] was performed to measure the quality of the transcriptome: non-restrictive transcriptome, which had both transcripts that belonged to S. marinoi and unclassified transcripts, and restrictive transcriptome, which had transcripts that only belonged to S. marinoi. Since the restrictive transcriptome had a high score of 96%, we went further with the restrictive transcriptome assembly. For P. parvum, the measured contamination led to bacterial (32%) and human (17%) contamination within the sequenced samples and from virus and other sequences (~0.5%). Approximately 3% of assembled transcripts were assigned to S. costatum (more precisely to Stramenopiles), which are most likely P. parvum transcripts closely related to the Stramenopiles group, which was the closest one to P. parvum. Most P. parvum transcripts (47%) were within the unclassified group. So, the non-restrictive transcriptome, with a score of 80.7%, was used for further analysis.
Using the tool Augustus [68], 46598 of the assembled transcripts were annotated as potential Open Reading Frames (ORFs) for S. marinoi, and 123266 were annotated for P. parvum. From within these transcripts, Augutus classified 32478 ORFs for S. marinoi and 59328 for P. parvum with functional domains. For S. marinoi, Augustus predicted that the assembled potential transcripts could encode for more than one ORF. For many of these transcripts, the predicted coding sequences (CDS) are located on opposite strands, i.e., one CDS was predicted in the sense direction of the assembled transcript and the other CDS in the anti-sense direction. To better estimate the number of unique transcripts, MMSeqs2 [90] was used. It assembled the transcripts based on the proteins they encoded, resulting in 10812 transcripts for S. marinoi and 75575 for P. parvum (Table 4).
The transcriptome assembly with ORFs predictions was then followed by the differential gene expression analysis to identify the differentially expressed ORFs (Fig 5). The differentially expressed ORFs with an FDR-adjusted p-value less than 0.05 and sorted genes based on the log fold change values resulted in 664 differentially expressed upregulated and 514 downregulated ORFs for S. marinoi and 755 differentially expressed upregulated and 247 downregulated ORFs for P. parvum. Table 5 lists the top 10 differentially expressed upregulated ORFs from S. marinoi and P. parvum with their functional annotation when identified. The transcriptome sequences are available with the BioProject ID PRJNA1004186, while the differential ORF protein sequences are available on Zenodo with the https://doi.org/10.5281/zenodo.10397384 [91].
This scatter plot displays the relationship between the log fold change and the adjusted p-value of each gene (represented by the dots). The tentative names are also displayed for the top 30 differentially expressed ORFs. The horizontal line marks the adjusted p-value threshold, whereas the two vertical lines indicate possible log fold change thresholds (1 and −1).
The top 10 differentially expressed upregulated ORFs in both species that could not be annotated with a function or domain were then investigated for an attempt of manual functional annotation, using TMHMM [92], p–BLAST [93], and InterProScan [69] for hints to identify the category of their possible function (Table 5). Among the ORFs of unknown function in S. marinoi, the ORF number g12608 could be predicted to encode a transmembrane protein, exhibiting 78% homology with a hypothetical protein from the haptophyte phytoplankton Chrysochromulina tobinii. In contrast, the ORF g255 is likely to encode an intracellular peptide with no detectable homology. In total, 229 ORFs in the differential expression analysis in S. marinoi had completely unknown functions and were only homologs with genes of unknown function in public databases. For P. parvum, four ORFs among the top 10 differentially expressed upregulated ones were with unknown functions. Among the four ORFs, ORF number g87732 was predicted to encode an intracellular protein, showing no significant similarity to any known or hypothetical proteins, while ORF number g90799 was predicted to encode a transmembrane protein with otherwise no known similarity to anything else in public databases. Furthermore, the subsequent ORFs numbers g68158 and g114069 also appeared to be intracellular. The response of P. parvum (highest fold change is 25.1531) is way higher than S. marinoi (highest fold change is 6.0440). These findings highlight the considerable number of genes with yet-to-be-characterised functions in both S. marinoi and P. parvum and an increased level of differential expression in the genome of P. parvum when grown as co-culture with S. marinoi.
4. Discussion
4.1. Influence of co-culturing on microalgae
Microbial communities are complex adaptive biosystems and are highly dynamic, with varying biodiversity and chemodiversity depending on the exogenous stimuli [94]. However, to understand the complexities of such systems, the controlled cultivation of two microorganisms can be considered as a basis for studying highly complex communities. Both microalgae selected for this research are cosmopolitan, found in marine ecosystems, and hold profound impacts on the aquatic biosphere [95,96]. The first microalgae in this experiment is the diatom S. marinoi, which is a valuable source of different natural products such as polyunsaturated fatty acids [83], serves as a food source for other marine organisms and has been demonstrated to convert CO2 in cement waste gas into renewable energy [97], depicting ecological and economic applications. The second microalgae, Prymnesium parvum, is considered a toxic species, which causes a decrease in biodiversity during harmful algal blooms (HABs) while growing at an exponential rate [98], indicating a toxic or growth-inhibiting behaviour on other members of the microbial community. It remains unexplored whether the growth of P. parvum benefits from diatoms like S. marinoi and other microalgae.
Our investigations into the co-culture of S. marinoi and P. parvum revealed distinct changes in chlorophyll a fluorescence, which is used as a proxy for cell growth in this experiment. Specifically, S. marinoi demonstrated a significant reduction in chlorophyll a fluorescence when co-cultured with P. parvum as compared to when grown as a mono-culture, whereas P. parvum in co-cultures demonstrated no significant difference overall, except the last day of fluorescence measurement in the cultures (as demonstrated in Fig 3 and Table 1). These observations could be due to several reasons. P. parvum is also known to be a mixotroph; hence, it relies on both photosynthesis and other organisms for food sources [99]. The cell lytic activity of the metabolites released from P. parvum could cause S. marinoi cellular depletion and the transfer of nutrients from S. marinoi. Supervised statistical analysis was performed to elucidate the possible metabolic exchange via metabolome and transcriptome.
4.2. Exploring responses to mutual presence of microalgae through metabolome and transcriptome
To investigate the metabolic interactions observed in co-culture of the two microalgae, suggested by a shift in the chlorophyll a concentrations, we employed the Partial Least Squares (PLS) analysis on endo- and exometabolome of all conditions from both microalgae. This supervised statistical regression model linked the intensity of differentially produced metabolic features to their respective occurrences in the co-culture and mono-culture conditions (S2 Fig). The 15 co-culture-specific features for the endometabolome of S. marinoi speculate the presence of metabolites involved in the defence activation of S. marinoi in the presence of P. parvum. For P. parvum endometabolome 30 differentially produced metabolites were revealed, suggesting the production of P. parvum allelochemicals, with adverse effects on S. marinoi. Alternatively, these features could have also originated from S. marinoi, indicating nutrient uptake from P. parvum.
To further investigate the potential metabolic exchange and uptake between the endometabolomes of S. marinoi and P. parvum, we analysed whether the differentially produced metabolites are shared among the endometabolomes (Fig 4). First, the differentially abundant metabolites in the endometabolome of co-culture P. parvum were cross-referenced with the exometabolome of the mono-culture of S. marinoi, and both the mono- and co-culture endometabolome of S. marinoi. The 23 common metabolites between endometabolome (co-culture) P. parvum and different conditions of S. marinoi metabolome suggest that P. parvum consumes these S. marinoi metabolites. Furthermore, the differentially abundant metabolites in endometabolome of co-culture S. marinoi were cross-referenced with exometabolome of the mono-culture of P. parvum, and both the mono- and co-culture endometabolome of P. parvum. In this scenario, atleast 8 metabolites in S. marinoi endometabolome in co-culture condition were from P. parvum; however, these metabolites can be termed as common metabolites between two species, as they were from the set of 15 metabolites found in the co-culture condition of S. marinoi. Either these could be the exudates exchanged between the two species, or the production of these metabolites is induced by the presence of other species. However, these speculations need further experimental confirmation.
For exometabolome (S2 Fig), we observed an increase in differentially produced metabolites as compared to endometabolome, which could be ascribed to the experimental design, given the potential for metabolic exchange across the permeable membrane. It is plausible that these features, originating from either microalga, indicate a predominant metabolic uptake direction favouring P. parvum. An additional factor to consider is the non-axenic nature of the mono-cultures and co-cultures. Some of these significant metabolites might be sourced from bacteria existing symbiotically within the phycosphere of these microalgal cells [100]. While the co-culture’s exometabolome cannot be easily attributed to a specific microalga, its distinct profile suggests metabolic interaction between S. marinoi and P. parvum rather than a mere aggregation of their individual metabolomes (S3 Fig). Even with these exometabolome features, we are missing non-polar metabolites, as the data was acquired in the Reverse Phase (RP).
Furthermore, a comprehensive transcriptome assembly coupled with protein annotations was performed for differential gene expression analysis. The resulting volcano plots (Fig 5) depict a change in responses to mutual presence by S. marinoi and P. parvum in co-culture samples; the differential gene expression analysis revealed a total of 664 differentially expressed transcripts for S. marinoi and 755 for P. parvum. The increased number of transcripts for P. parvum could indicate the genetic exchange from S. marinoi cell lysis towards the P. parvum side of the co-culture chamber through the permeable membrane. This has not been confirmed with the current methodologies applied to this study. Moreover, within the top 10 differentially expressed transcripts, P. parvum had a higher fold change for transcripts of unknown function than the log fold change of the top 10 differentially expressed transcripts in S. marinoi with unknown function (Table 5). These results suggest that both species undergo significant transcriptional remodelling, potentially reflecting adaptive mechanisms or competitive behaviours to thrive in the presence of the other, especially in the case of P. parvum.
4.3. Metabolome and transcriptome annotation analysis and technical aspects
The differential metabolite profiling from the co-culture revealed intriguing dynamics that suggest the underlying responses and adaptations of these microorganisms in a co-culture setting; however, these metabolites were not studied in the context of marine microalgae before. In P. parvum co-culture (endometabolome), stilbenes, a phenolic allelochemical compound class, was annotated [101], known to have roles in defence mechanisms and regulate critical signalling pathways such as JAK-STAT, MAPK, and NF-κB pathways by diminishing the transcription of inflammatory factors and thus preserving a homeostatic environment [102]. Meanwhile, the exometabolome of P. parvum predominantly showed fatty acyls, macrolides [103], Saccharolipids [104], and Isoflavonoids, involved in protection against oxidative stress and immunomodulation [105]. These detected metabolites show a survival response from P. parvum; however, they could also belong to S. marinoi as exometabolites are exchanged through the permeable membrane. S. marinoi, in contrast, demonstrated the presence of lactones, phenols, and prenol lipids. Certain other known compounds, such as the fungal pigment compound boletocrocin E [106], tumonoic acid A (exhibiting anti-inflammatory properties) [107], commonly found in bacteria and marine cyanobacterium Blennothrix cantharidosmum [108], and pogopyrone B, a benzopyran, sourced from the plant Pogostemon heynianus, were detected [109]. These metabolites are less studied for their functions and could be due to misannotation, because of the limited chemical structures associated with marine microalgae in the databases [29]. However, their putative presence suggests that as a co-culture, S. marinoi and P. parvum produce a competitive environment. We also used suspect lists from both S. marinoi and P. parvum for all MS2 features to identify secondary metabolites specific to either species or both, none of which were differentially produced (Table 3). Using the suspect list, the origin of specific metabolites could be identified, most of which belonged to S. marinoi, which is more extensively studied along with its closely related species S. costatum; however, P. parvum is mostly studied for its toxins and hence had a much smaller suspect list of known compounds and less secondary metabolite annotations from the suspect list.
MS2 for only 14 differential metabolites (Table 3) is likely due to the data acquisition from the inclusion lists generated using Compound Discoverer software, which focused on the most abundant MS1 features in the DDA mode. Since all of the co-culture analysis was performed using open source software and R packages after both MS1 and subsequent MS2 data acquisition, no prior processing or statistical analysis was performed using Compound Discoverer to pinpoint the most relevant differential features before MS2 data acquisition, and there’s a high likelihood that several significant features were overlooked for MS2 fragmentation because they were differentially abundant among different conditions and not across all samples [110]. A future follow-up using open source code would be to first acquire MS1, statistically analyse the differentially abundant metabolites, enrich those features (as they might be lost due to low intensity throughout the samples) and acquire MS2 for those metabolites.
The transcriptional landscape of S. marinoi and P. parvum in the co-culture reflected both primary and secondary metabolic adaptations, as observed from the top 30 differentially expressed upregulated ORFs (only 10 differentially expressed ORFs mentioned in Table 5; the rest can be found in Zenodo with the https://doi.org/10.5281/zenodo.10397384 [91]) for S. marinoi and P. parvum. In S. marinoi, the presence of genes associated with core cellular processes like DNA repair, cell signalling, apoptosis, and immune response, based on the identification of a protease related to ATP-dependent degradation of ubiquitinated proteins [111], provided the metabolic readjustments that might be necessary for survival in a shared environment. Fatty acid synthesis through the condensation reaction and enzymes associated with oxidative deamination suggest potential lipid-based defence strategies or energy storage responses as a buffer against potential resource competition in the presence of P. parvum [112]. The genes involved in photosynthesis, such as photosystem II d2 protein, signify a possible photosynthetic activity as seen in chlorophyll a levels or adaptation to changes in light quality or intensity that may arise from the growth dynamics of the co-culture [113]. In P. parvum, the identification of genes, such as ADP-ribosylation/crystallin j1, highlights possible cellular regulatory and repair mechanisms. Similarly, the enoyl-CoA hydratase/isomerase and the dihydrodipicolinate synthetase families, both central to fundamental metabolic pathways, might indicate energy utilisation. Various transporters like carboxylic acid and copper transport protein suggest an exchange of metabolites and the metabolic content in the extracellular space. Moreover, the presence of genes associated with type IV pilus inner membrane component and the division/cell wall cluster transcriptional repressor could indicate a heightened focus on biofilm formation [114].
For the transcriptome assembly of both species, we encountered a substantial number of potentially assembled transcripts and annotated ORFs mentioned in Table 4. Most likely, many of the assembled transcripts have duplicates (sequence similarity). This could result from paralogs, small assembly mistakes, and alternative splicing events. To analyse the differentials ORFs, we used the total number of annotated ORFs so as not to lose any important differential ORFs with no functional annotation. Transcriptomics data analysis provides some insights into the co-culture dynamic; however, there are many missing annotations specific to secondary metabolism, specifically for P. parvum. When analysed using BLAST-P, several differentially expressed upregulated ORFs with unknown functions in this co-culture experiment showed minimal homology to hypothetical proteins. Without well-defined gene sets for these species, identifying unique genes and pathways specific to the microalgae can be challenging due to limited existing data to reference. Specifically for P. parvum, most of the top differentially expressed upregulated ORFs suggested primary metabolism-relevant proteins, and others with unknown functions and low structural homology to known proteins could have more significant insights into the secondary metabolism of the co-culture of these two microalgae.
5. Conclusion
Chemical interactions regulate many marine life processes, such as balancing the ecosystem and the type of interaction among the microbial key players, such as predation or defence [115]. Natural products which cause marine chemical interaction affect not only the endometabolome of the species but also the exometabolome in the environment. This study describes the co-culture setup between two microalgae, Skeletonema marinoi and Prymnesium parvum. The cell growth data suggested a negative growth impact on S. marinoi when grown in the presence of P. parvum. This hypothesis was then tested with statistical analysis, metabolome, and transcriptome annotations, leading to three conclusions: (1) there is an effective exchange of metabolites among the two microalgae, (2) S. marinoi is most affected by this co-culture setup, and (3) there is a significant increase in the number of differentially expressed ORFs for both S. marinoi and P. parvum, which suggests adaptation. This work signifies that much remains undiscovered in our understanding of the marine biosphere, and the microalgae metabolic interactions should be further explored.
Supporting information
S1 Table. Chlorophyll a fluorescence measurement in relative fluorescence units (RFU) for 8 days (d0-d8) in S. marinoi mono-culture (Sm control), S. marinoi co-culture (Sm co-culture), P. parvum mono-culture (Pp mono-culture), and P. parvum co-culture (Pp co-culture).
https://doi.org/10.1371/journal.pone.0329115.s001
(DOCX)
S2 Table. ANOVA analysis of variance for the species.
The sample describes the overall difference in the data when comparing mono-culture to co-culture data. Days describe the difference in the data between the days of sample collection. The values determining a significant difference for the corresponding groups are marked in green. SS = sum of squares to represent the squared deviation from the mean, df = degrees of freedom, MS = mean square which is calculated by dividing (SS) by (df), representing the average amount of variation within each variation source, F = F-statistics to determine whether significant differences exist between groups and within groups, P-value = probability of the null hypothesis being accepted or rejected, F crit = critical F-value, the value from an F-distribution table that corresponds to the significance level of 0.05. Another indicator of significant variation among the samples for S. marinoi is the calculated F-value (F) being higher than the critical F-value (F crit). In contrast, the opposite holds for P. parvum.
https://doi.org/10.1371/journal.pone.0329115.s002
(DOCX)
S1 Fig. Shannon Diversity index H’ for exometabolome and endometabolome samples.
Co-culture samples are coloured purple and mono-culture samples are coloured green. a) exometabolome feature diversity with the left two samples showing P. parvum and the right samples S. marinoi. b) endometabolome feature diversity with P. parvum on the left and S. marinoi on the right. Mono = mono-culture, Co = co-culture, P.p = P. parvum, S.m = S. marinoi.
https://doi.org/10.1371/journal.pone.0329115.s003
(TIF)
S2 Fig. Partial Least Square Analysis of endo and exometabolome of S. marinoi and P. parvum.
The dendrogram clusters the conditions and features based on their intensities, with red indicating high intensity, white indicating no change, and blue representing low intensity or absence. a) For endometabolome, 22 differentially abundant features were selected for S. marinoi, out of which only seven were abundant in mono-culture. b) Conversely, all 30 differentially abundant features detected for P. parvum endometabolome were abundant in co-culture conditions. For exometabolome, a higher number of differentially abundant features were detected for both species, with more abundant features found in co-culture of S. marinoi and P. parvum. c) In total, the S. marinoi differentially abundant features were 324, out of which 304 features were abundant in co-culture. d) For P. parvum, the total number was 491, out of which 418 features were abundant in co-culture.
https://doi.org/10.1371/journal.pone.0329115.s004
(TIF)
S3 Fig. Principal component analysis (PCA) for the observed effects on the exometabolome of co-cultures.
a) Shows co-culture (orange for S. marinoi and purple for P. parvum) and mono-culture (yellow for S. marinoi and pink for P. parvum) samples with principal component (PC) 2, explaining 7.1% of the variance, plotted against PC2, explaining 24.9% of the variance. b) Shows combined mono-culture for the S. marinoi and P. parvum exometabolome in green, while the S. marinoi co-culture is shown in orange, and P. parvum co-culture is shown in purple, where PC 2 describes a variance of 11.6%, and PC1 describes a variance of 16.7%. The PCA in b) clearly distinguishes between the combined mono-cultures and individual co-cultures, indicating metabolic transformation occurring in the co-culture of both species.
https://doi.org/10.1371/journal.pone.0329115.s005
(TIF)
Acknowledgments
Authors would like to acknowledge Nico Ueberschaar (ORCID: 0000-0002-4192-490X) for providing the standardised MS1 data acquisition protocol.
The RAW and mzML metabolomics data are submitted and under curation on the MetaboLights [39,40] repository with the study number MTBLS8313 (www.ebi.ac.uk/metabolights/MTBLS8313). The mzML files are also available on Zenodo with the following DOIs: (LC-MS1 https://doi.org/10.5281/zenodo.10143127; LC-MS2 https://doi.org/10.5281/zenodo.10143233 [116,117]). All the metabolomics result files are available at Zenodo with https://doi.org/10.5281/zenodo.10143554 [118]. The currently available versions of the GNPS, HMDB, and MassBank are also archived on Zenodo with the https://doi.org/10.5281/zenodo.7519270 [56]. The suspect list for Skeletonema spp. is available on Zenodo with the https://doi.org/10.5281/zenodo.5772755 [81] and for Prymnesium parvum with the https://doi.org/10.5281/zenodo.7864506 [80]. The script for the suspect list curation is available at https://github.com/zmahnoor14/MAW-Diatom/blob/main/suspectlist_curation.py [119], applied to both suspect lists. The code and scripts for MS1 processing, linking, and statistical analysis are available at https://github.com/zmahnoor14/MAW/tree/main/co-culture [27], while the MS2 processing and annotation were carried out using MAW v1.5.0 available on https://github.com/zmahnoor14/MAW [61]. The transcriptomics sequencing data is available on the BioProject PRJNA1006530 [120]. The fasta sequences for the proteins encoded by the predicted ORFs are also archived on Zenodo with the https://doi.org/10.5281/zenodo.10397384 [91]. Code for transcriptome analysis is available on the GitHub repository: https://github.com/Bioinformatics-Core-Facility-Jena/SE20220705_97 [78].
References
- 1. Costello MJ, Cheung A, De Hauwere N. Surface area and the seabed area, volume, depth, slope, and topographic variation for the world’s seas, oceans, and countries. Environ Sci Technol. 2010;44(23):8821–8. pmid:21033734
- 2.
Barre SL. Marine biodiversity and chemodiversity — the treasure troves of the future. In: Biodiversity - The dynamic balance of the planet. IntechOpen; 2014.
- 3. Rotter A, Barbier M, Bertoni F, Bones AM, Cancela ML, Carlsson J, et al. The Essentials of Marine Biotechnology. Front Mar Sci. 2021;8.
- 4. Young JN, Schmidt K. It’s what’s inside that matters: physiological adaptations of high-latitude marine microalgae to environmental change. New Phytol. 2020;227(5):1307–18. pmid:32391569
- 5. Dorrell RG, Kuo A, Füssy Z, Richardson EH, Salamov A, Zarevski N, et al. Convergent evolution and horizontal gene transfer in Arctic Ocean microalgae. Life Sci Alliance. 2022;6(3):e202201833. pmid:36522135
- 6.
Hunter-Cevera J, Karl D, Buckley M. Marine Microbial Diversity: The Key to Earth’s Habitability. American Society for Microbiology; 2005.
- 7. Mathur V, Ulanova D. Microbial Metabolites Beneficial to Plant Hosts Across Ecosystems. Microb Ecol. 2023;86(1):25–48. pmid:35867138
- 8. Barbosa AJM, Roque ACA. Free Marine Natural Products Databases for Biotechnology and Bioengineering. Biotechnol J. 2019;14(11):e1800607. pmid:31297982
- 9. Song H-S, Cannon W, Beliaev A, Konopka A. Mathematical Modeling of Microbial Community Dynamics: A Methodological Review. Processes. 2014;2(4):711–52.
- 10. Koehn FE, Carter GT. The evolving role of natural products in drug discovery. Nat Rev Drug Discov. 2005;4(3):206–20. pmid:15729362
- 11. Aguiar-Pulido V, Huang W, Suarez-Ulloa V, Cickovski T, Mathee K, Narasimhan G. Metagenomics, Metatranscriptomics, and Metabolomics Approaches for Microbiome Analysis. Evol Bioinform Online. 2016;12(Suppl 1):5–16. pmid:27199545
- 12. Luo Y, Cobb RE, Zhao H. Recent advances in natural product discovery. Curr Opin Biotechnol. 2014;30:230–7. pmid:25260043
- 13. Mishra A, Medhi K, Malaviya P, Thakur IS. Omics approaches for microalgal applications: Prospects and challenges. Bioresour Technol. 2019;291:121890. pmid:31378447
- 14. Donati S, Mattanovich M, Hjort P, Jacobsen SAB, Blomquist SD, Mangaard D, et al. An automated workflow for multi-omics screening of microbial model organisms. NPJ Syst Biol Appl. 2023;9(1):14. pmid:37208327
- 15. Paul C, Barofsky A, Vidoudez C, Pohnert G. Diatom exudates influence metabolism and cell growth of co-cultured diatom species. Mar Ecol Prog Ser. 2009;389:61–70.
- 16. Paul C, Mausz MA, Pohnert G. A co-culturing/metabolomics approach to investigate chemically mediated interactions of planktonic organisms reveals influence of bacteria on diatom metabolism. Metabolomics. 2012;9(2):349–59.
- 17. Johansson ON, Pinder MIM, Ohlsson F, Egardt J, Töpel M, Clarke AK. Friends With Benefits: Exploring the Phycosphere of the Marine Diatom Skeletonema marinoi. Front Microbiol. 2019;10:1828. pmid:31447821
- 18. de Vargas C, Audic S, Henry N, Decelle J, Mahé F, Logares R, et al. Ocean plankton. Eukaryotic plankton diversity in the sunlit ocean. Science. 2015;348(6237):1261605. pmid:25999516
- 19. Binzer SB, Svenssen DK, Daugbjerg N, Alves-de-Souza C, Pinto E, Hansen PJ, et al. A-, B- and C-type prymnesins are clade specific compounds and chemotaxonomic markers in Prymnesium parvum. Harmful Algae. 2019;81:10–7. pmid:30638493
- 20. Johnsen TM, Eikrem W, Olseng CD, Tollefsen KE, Bjerknes V. Prymnesium parvum: The Norwegian experience. JAWRA J Am Water Resour Assoc. 2010;46(1):6–13.
- 21. Vidoudez C, Pohnert G. Comparative metabolomics of the diatom Skeletonema marinoi in different growth phases. Metabolomics. 2011;8(4):654–69.
- 22. Wang M, Carver JJ, Phelan VV, Sanchez LM, Garg N, Peng Y, et al. Sharing and community curation of mass spectrometry data with Global Natural Products Social Molecular Networking. Nat Biotechnol. 2016;34(8):828–37. pmid:27504778
- 23.
Peters K. iESTIMATE computational analysis framework for Eco-Metabolomics data. IPB Halle; 2022 [cited 2023 Jun 12] In: GitHub [Internet]. Available from: https://github.com/ipb-halle/iESTIMATE
- 24. Gatto L, Gibb S, Rainer J. MSnbase, Efficient and Elegant R-Based Processing and Visualization of Raw Mass Spectrometry Data. J Proteome Res. 2021;20(1):1063–9. pmid:32902283
- 25. Libiseller G, Dvorzak M, Kleb U, Gander E, Eisenberg T, Madeo F, et al. IPO: a tool for automated optimization of XCMS parameters. BMC Bioinformatics. 2015;16:118. pmid:25888443
- 26. Tautenhahn R, Böttcher C, Neumann S. Highly sensitive feature detection for high resolution LC/MS. BMC Bioinformatics. 2008;9:504. pmid:19040729
- 27. Abel AS, Zulfiqar M. MAW/co-culture at main [cited 2023 Dec 23]. In: GitHub [Internet]. Available from: https://github.com/zmahnoor14/MAW/tree/main/co-culture
- 28.
R Core Team. R: A language and environment for statistical computing. [Internet]. Vienna, Austria: R Foundation for Statistical Computing; 2022. Available from: https://www.R-project.org/
- 29. Zulfiqar M, Stettin D, Schmidt S, Nikitashina V, Pohnert G, Steinbeck C, et al. Untargeted metabolomics to expand the chemical space of the marine diatom Skeletonema marinoi. Front Microbiol. 2023;14:1295994. pmid:38116530
- 30. Kanehisa M. Toward understanding the origin and evolution of cellular organisms. Protein Sci. 2019;28(11):1947–51. pmid:31441146
- 31. Kanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 2000;28(1):27–30. pmid:10592173
- 32. Kanehisa M, Furumichi M, Sato Y, Kawashima M, Ishiguro-Watanabe M. KEGG for taxonomy-based analysis of pathways and genomes. Nucleic Acids Res. 2023;51(D1):D587–92. pmid:36300620
- 33. Lyu C, Chen T, Qiang B, Liu N, Wang H, Zhang L, et al. CMNPD: a comprehensive marine natural products database towards facilitating drug discovery from the ocean. Nucleic Acids Res. 2021;49(D1):D509–15. pmid:32986829
- 34. Hastings J, Owen G, Dekker A, Ennis M, Kale N, Muthukrishnan V, et al. ChEBI in 2016: Improved services and an expanding collection of metabolites. Nucleic Acids Res. 2016;44(D1):D1214-9. pmid:26467479
- 35. Kim S, Thiessen PA, Bolton EE, Chen J, Fu G, Gindulyte A, et al. PubChem Substance and Compound databases. Nucleic Acids Res. 2016;44(D1):D1202-13. pmid:26400175
- 36. Kim S, Chen J, Cheng T, Gindulyte A, He J, He S, et al. PubChem 2023 update. Nucleic Acids Res. 2023;51(D1):D1373–80. pmid:36305812
- 37. UniProt Consortium. UniProt: the universal protein knowledgebase in 2021. Nucleic Acids Res. 2021;49(D1):D480–9. pmid:33237286
- 38. Chang A, Jeske L, Ulbrich S, Hofmann J, Koblitz J, Schomburg I, et al. BRENDA, the ELIXIR core data resource in 2021: new developments and updates. Nucleic Acids Res. 2021;49(D1):D498–508. pmid:33211880
- 39. Kale NS, Haug K, Conesa P, Jayseelan K, Moreno P, Rocca-Serra P, et al. MetaboLights: An Open-Access Database Repository for Metabolomics Data. Curr Protoc Bioinformatics. 2016;53:14.13.1-14.13.18. pmid:27010336
- 40. Haug K, Cochrane K, Nainala VC, Williams M, Chang J, Jayaseelan KV, et al. MetaboLights: a resource evolving in response to the needs of its scientific community. Nucleic Acids Res. 2020;48(D1):D440–4. pmid:31691833
- 41. Caspi R, Altman T, Billington R, Dreher K, Foerster H, Fulcher CA, et al. The MetaCyc database of metabolic pathways and enzymes and the BioCyc collection of Pathway/Genome Databases. Nucleic Acids Res. 2014;42(Database issue):D459-71. pmid:24225315
- 42. Caspi R, Billington R, Keseler IM, Kothari A, Krummenacker M, Midford PE, et al. The MetaCyc database of metabolic pathways and enzymes - a 2019 update. Nucleic Acids Res. 2020;48(D1):D445–53. pmid:31586394
- 43. Rasmussen SA, Meier S, Andersen NG, Blossom HE, Duus JØ, Nielsen KF, et al. Chemodiversity of Ladder-Frame Prymnesin Polyethers in Prymnesium parvum. J Nat Prod. 2016;79(9):2250–6. pmid:27550620
- 44. Henrikson JC, Gharfeh MS, Easton AC, Easton JD, Glenn KL, Shadfan M, et al. Reassessing the ichthyotoxin profile of cultured Prymnesium parvum (golden algae) and comparing it to samples collected from recent freshwater bloom and fish kill events in North America. Toxicon. 2010;55(7):1396–404. pmid:20184911
- 45. Rahat M, Reich K. The B12 vitamins and methionine in the metabolism of Prymnesium parvum. J Gen Microbiol. 1963;31:203–9. pmid:13990463
- 46. Bertin MJ, Zimba PV, Beauchesne KR, Huncik KM, Moeller PDR. Identification of toxic fatty acid amides isolated from the harmful alga Prymnesium parvum carter. Harmful Algae. 2012;20:111–6.
- 47. Liu Z, Koid AE, Terrado R, Campbell V, Caron DA, Heidelberg KB. Changes in gene expression of Prymnesium parvum induced by nitrogen and phosphorus limitation. Front Microbiol. 2015;6:631. pmid:26157435
- 48. Taylor RB, Hill BN, Langan LM, Chambliss CK, Brooks BW. Sunlight concurrently reduces Prymnesium parvum elicited acute toxicity to fish and prymnesins. Chemosphere. 2021;263:127927. pmid:32814137
- 49. Landrum G. RDKit: Open-Source Cheminformatics Software. 2016 [Internet]. Available from: http://www.rdkit.org/, https://github.com/rdkit/rdkit
- 50. Swain M. Introduction — PubChemPy 1.0.4 documentation [Internet]. [cited 2022 Nov 18]. Available from: https://pubchempy.readthedocs.io/en/latest/guide/introduction.html#pubchempy-license
- 51. Allard PM. pybatchclassyfire. 2020 [cited 2022 Sep 22]. In GitLab: [Internet]. Available from: https://gitlab.unige.ch/Pierre-Marie.Allard/pybatchclassyfire
- 52. Zulfiqar M, Gadelha L, Steinbeck C, Sorokina M, Peters K. MAW: the reproducible Metabolome Annotation Workflow for untargeted tandem mass spectrometry. J Cheminform. 2023;15(1):32. pmid:36871033
- 53.
Wishart DS. Human Metabolome Database: Downloads [Internet]. HMDB Downloads; 2021 [cited 2022 Sep 22]. Available from: https://hmdb.ca/downloads
- 54. Wishart DS, Guo A, Oler E, Wang F, Anjum A, Peters H, et al. HMDB 5.0: the Human Metabolome Database for 2022. Nucleic Acids Res. 2022;50(D1):D622–31. pmid:34986597
- 55. Horai H, Arita M, Kanaya S, Nihei Y, Ikeda T, Suwa K, et al. MassBank: a public repository for sharing mass spectral data for life sciences. J Mass Spectrom. 2010;45(7):703–14. pmid:20623627
- 56. Zulfiqar M. Spectral Databases (GNPS, HMDB, MassBank). 2023 [cited 2023 Jun 29]. In: Zenodo [Internet]. Available from: https://zenodo.org/record/7519270
- 57. Böcker S, Letzel MC, Lipták Z, Pervukhin A. SIRIUS: decomposing isotope patterns for metabolite identification. Bioinformatics. 2009;25(2):218–24. pmid:19015140
- 58. Sorokina M, Merseburger P, Rajan K, Yirik MA, Steinbeck C. COCONUT online: Collection of Open Natural Products database. J Cheminform. 2021;13(1):2. pmid:33423696
- 59. Djoumbou Feunang Y, Eisner R, Knox C, Chepelev L, Hastings J, Owen G, et al. ClassyFire: automated chemical classification with a comprehensive, computable taxonomy. J Cheminform. 2016;8:61. pmid:27867422
- 60. Dührkop K, Nothias L-F, Fleischauer M, Reher R, Ludwig M, Hoffmann MA, et al. Systematic classification of unknown metabolites using high-resolution fragmentation mass spectra. Nat Biotechnol. 2021;39(4):462–71. pmid:33230292
- 61. Zulfiqar M. zmahnoor14/MAW. 2022 [cited 2022 Sep 2]. In: GitHub [Internet]. (Metabolome Annotation Workflow). Available from: https://github.com/zmahnoor14/MAW
- 62. Chen S, Zhou Y, Chen Y, Gu J. fastp: an ultra-fast all-in-one FASTQ preprocessor. Bioinformatics. 2018;34(17):i884–90. pmid:30423086
- 63. Wood DE, Lu J, Langmead B. Improved metagenomic analysis with Kraken 2. Genome Biol. 2019;20(1):257. pmid:31779668
- 64. Armbrust EV, Berges JA, Bowler C, Green BR, Martinez D, Putnam NH, et al. The genome of the diatom Thalassiosira pseudonana: ecology, evolution, and metabolism. Science. 2004;306(5693):79–86. pmid:15459382
- 65. Sorokina M, Barth E, Zulfiqar M, Kwantes M, Pohnert G, Steinbeck C. Draft genome assembly and sequencing dataset of the marine diatom Skeletonema cf. costatum RCC75. Data Brief. 2022;41:107931. pmid:35242913
- 66. Grabherr MG, Haas BJ, Yassour M, Levin JZ, Thompson DA, Amit I, et al. Full-length transcriptome assembly from RNA-Seq data without a reference genome. Nat Biotechnol. 2011;29(7):644–52. pmid:21572440
- 67. Seppey M, Manni M, Zdobnov EM. BUSCO: Assessing Genome Assembly and Annotation Completeness. In: Kollmar M, editor. Gene Prediction: Methods and Protocols. New York, NY: Springer; 2019. p. 227–45. (Methods in Molecular Biology).
- 68. Keller O, Kollmar M, Stanke M, Waack S. A novel hybrid gene prediction method employing protein multiple sequence alignments. Bioinformatics. 2011;27(6):757–63. pmid:21216780
- 69. Jones P, Binns D, Chang H-Y, Fraser M, Li W, McAnulla C, et al. InterProScan 5: genome-scale protein function classification. Bioinformatics. 2014;30(9):1236–40. pmid:24451626
- 70. Mistry J, Chuguransky S, Williams L, Qureshi M, Salazar GA, Sonnhammer ELL, et al. Pfam: The protein families database in 2021. Nucleic Acids Res. 2021;49(D1):D412–9. pmid:33125078
- 71. Haft DH, Selengut JD, White O. The TIGRFAMs database of protein families. Nucleic Acids Res. 2003;31(1):371–3. pmid:12520025
- 72. Mi H, Thomas P. PANTHER Pathway: An Ontology-Based Pathway Database Coupled with Data Analysis Tools. In: Nikolsky Y, Bryant J, editors. Protein Networks and Pathway Analysis. Totowa, NJ: Humana Press; 2009. p. 123–40. (Methods in Molecular Biology).
- 73. Sigrist CJA, de Castro E, Cerutti L, Cuche BA, Hulo N, Bridge A, et al. New and continuing developments at PROSITE. Nucleic Acids Res. 2013;41(Database issue):D344-7. pmid:23161676
- 74. Scheibenreif L, Littmann M, Orengo C, Rost B. FunFam protein families improve residue level molecular function prediction. BMC Bioinformatics. 2019;20(1):400.
- 75. Kim D, Paggi JM, Park C, Bennett C, Salzberg SL. Graph-based genome alignment and genotyping with HISAT2 and HISAT-genotype. Nat Biotechnol. 2019;37(8):907–15. pmid:31375807
- 76. Liao Y, Smyth GK, Shi W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics. 2013;30(7):923–30.
- 77. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550.
- 78. Jena BCF. Bioinformatics-Core-Facility-Jena/SE20220705_97. 2023 [cited 2023 Dec 23]. In: GitHub: [Internet]. Available from: https://github.com/Bioinformatics-Core-Facility-Jena/SE20220705_97
- 79. Tan K, Huang Z, Ji R, Qiu Y, Wang Z, Liu J. A review of allelopathy on microalgae. Microbiology (Reading). 2019;165(6):587–92. pmid:30688632
- 80. Zulfiqar M, Costa S. Suspect List of Prymnesium parvum. 2023 [cited 2023 Dec 23]. In: Zenodo: [Internet]. Available from: https://zenodo.org/records/7864506
- 81. Zulfiqar M, Schmidt S. Skeletonema species Suspect List. 2021 [cited 2023 Sep 3]. In: Zenodo: [Internet]. Available from: https://zenodo.org/record/5772755
- 82. d’Ippolito G, Romano G, Iadicicco O, Miralto A, Ianora A, Cimino G, et al. New birth-control aldehydes from the marine diatom Skeletonema costatum: characterization and biogenesis. Tetrahedron Lett. 2002;43(35):6133–6.
- 83. Ribalet F, Wichard T, Pohnert G, Ianora A, Miralto A, Casotti R. Age and nutrient limitation enhance polyunsaturated aldehyde production in marine diatoms. Phytochemistry. 2007;68(15):2059–67. pmid:17575990
- 84. Rocco M, Dunne E, Peltola M, Barr N, Williams J, Colomb A, et al. Oceanic phytoplankton are a potentially important source of benzenoids to the remote marine atmosphere. Commun Earth Environ. 2021;2(1):1–8.
- 85. Breuer G, Evers WAC, de Vree JH, Kleinegris DMM, Martens DE, Wijffels RH, et al. Analysis of fatty acid content and composition in microalgae. J Vis Exp. 2013;(80):50628. pmid:24121679
- 86. Raina J-B, Lambert BS, Parks DH, Rinke C, Siboni N, Bramucci A, et al. Chemotaxis shapes the microscale organization of the ocean’s microbiome. Nature. 2022;605(7908):132–8. pmid:35444277
- 87. Antonowski T, Osowski A, Lahuta L, Górecki R, Rynkiewicz A, Wojtkiewicz J. Health-Promoting Properties of Selected Cyclitols for Metabolic Syndrome and Diabetes. Nutrients. 2019;11(10):2314. pmid:31574903
- 88. Steinbeck C, Han Y, Kuhn S, Horlacher O, Luttmann E, Willighagen E. The Chemistry Development Kit (CDK): an open-source Java library for Chemo- and Bioinformatics. J Chem Inf Comput Sci. 2003;43(2):493–500. pmid:12653513
- 89. Willighagen EL, Mayfield JW, Alvarsson J, Berg A, Carlsson L, Jeliazkova N, et al. The Chemistry Development Kit (CDK) v2.0: atom typing, depiction, molecular formulas, and substructure searching. J Cheminform. 2017;9(1):33. pmid:29086040
- 90. Steinegger M, Söding J. MMseqs2 enables sensitive protein sequence searching for the analysis of massive data sets. Nat Biotechnol. 2017;35(11):1026–8. pmid:29035372
- 91. Barth E, Zulfiqar M. Transcriptome Analysis from a co-culture of Skeletonema marinoi and Prymnesium parvum. 2023 [cited 2023 Dec 23]. In: Zenodo [Internet]. Available from: https://zenodo.org/records/10397384
- 92. Hallgren J, Tsirigos KD, Pedersen MD, Armenteros JJA, Marcatili P, Nielsen H, et al. DeepTMHMM predicts alpha and beta transmembrane proteins using deep neural networks. bioRxiv [Preprint]. 2022 [cited 2023 Feb 5]:2022.04.08.487609. Available from: https://www.biorxiv.org/content/10.1101/2022.04.08.487609v1
- 93. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol. 1990;215(3):403–10. pmid:2231712
- 94. Konopka A, Lindemann S, Fredrickson J. Dynamics in microbial communities: unraveling mechanisms to identify principles. ISME J. 2015;9(7):1488–95. pmid:25526370
- 95. Kaartvedt S, Johnsen TM, Aksnes DL, Lie U, Svendsen H. Occurrence of the Toxic PhytoflagellatePrymnesium parvumand Associated Fish Mortality in a Norwegian Fjord System. Can J Fish Aquat Sci. 1991;48(12):2316–23.
- 96. Wang H, Chen F, Mi T, Liu Q, Yu Z, Zhen Y. Responses of Marine Diatom Skeletonema marinoi to Nutrient Deficiency: Programmed Cell Death. Appl Environ Microbiol. 2020;86(3):e02460-19. pmid:31757826
- 97. Olofsson M, Lindehoff E, Frick B, Svensson F, Legrand C. Baltic Sea microalgae transform cement flue gas into valuable biomass. Algal Research. 2015;11:227–33.
- 98. Wagstaff BA, Pratscher J, Rivera PPL, Hems ES, Brooks E, Rejzek M, et al. Assessing the Toxicity and Mitigating the Impact of Harmful Prymnesium Blooms in Eutrophic Waters of the Norfolk Broads. Environ Sci Technol. 2021;55(24):16538–51. pmid:34882392
- 99. Carpenter KJ, Bose M, Polerecky L, Lie AAY, Heidelberg KB, Caron DA. Single-Cell View of Carbon and Nitrogen Acquisition in the Mixotrophic Alga Prymnesium parvum (Haptophyta) Inferred From Stable Isotope Tracers and NanoSIMS. Front Mar Sci. 2018;5.
- 100. Mugnai S, Derossi N, Hendlin Y. Algae communication, conspecific and interspecific: the concepts of phycosphere and algal-bacteria consortia in a photobioreactor (PBR). Plant Signal Behavior. 2023;18(1):2148371.
- 101. Ampofo J, Abbey L. Microalgae: Bioactive Composition, Health Benefits, Safety and Prospects as Potential High-Value Ingredients for the Functional Food Industry. Foods. 2022;11(12):1744. pmid:35741941
- 102. Al-Khayri JM, Mascarenhas R, Harish HM, Gowda Y, Lakshmaiah VV, Nagella P, et al. Stilbenes, a Versatile Class of Natural Metabolites for Inflammation-An Overview. Molecules. 2023;28(9):3786. pmid:37175197
- 103. Kanoh S, Rubin BK. Mechanisms of action and clinical application of macrolides as immunomodulatory medications. Clin Microbiol Rev. 2010;23(3):590–615. pmid:20610825
- 104. Park J, Choi J, Kim D-D, Lee S, Lee B, Lee Y, et al. Bioactive Lipids and Their Derivatives in Biomedical Applications. Biomol Ther (Seoul). 2021;29(5):465–82. pmid:34462378
- 105. Miadoková E. Isoflavonoids - an overview of their biological activities and potential health benefits. Interdiscip Toxicol. 2009;2(4):211–8. pmid:21217857
- 106. Kahner L, Dasenbrock J, Spiteller P, Steglich W, Marumoto R, Spiteller M. Polyene pigments from fruit-bodies of boletus laetissimus and B. rufo-aureus (basidiomycetes). Phytochemistry. 1998;49(6):1693–7. pmid:11711083
- 107. Engene N, Choi H, Esquenazi E, Byrum T, Villa FA, Cao Z, et al. Phylogeny-guided isolation of ethyl tumonoate A from the marine cyanobacterium cf. Oscillatoria margaritifera. J Nat Prod. 2011;74(8):1737–43. pmid:21751786
- 108. Harrigan GG, Luesch H, Yoshida WY, Moore RE, Nagle DG, Biggs J, et al. Tumonoic acids, novel metabolites from a cyanobacterial assemblage of Lyngbya majuscula and Schizothrix calcicola. J Nat Prod. 1999;62(3):464–7. pmid:10096859
- 109. Hohmann C, Schneider K, Bruntner C, Brown R, Jones AL, Goodfellow M, et al. Albidopyrone, a new α-pyrone-containing metabolite from marine-derived Streptomyces sp. NTK 227. J Antibiot. 2009;62(2):75–9.
- 110. Zuo Z, Cao L, Nothia L-F, Mohimani H. MS2Planner: improved fragmentation spectra coverage in untargeted mass spectrometry by iterative optimized data acquisition. Bioinformatics. 2021;37(Suppl_1):i231–6. pmid:34252948
- 111. Brinkmann K, Schell M, Hoppe T, Kashkar H. Regulation of the DNA damage response by ubiquitin conjugation. Front Genet. 2015;6.
- 112. Jüttner F. Liberation of 5,8,11,14,17‐eicosapentaenoic acid and other polyunsaturated fatty acids from lipids as a grazer defense reaction in epilithic diatom biofilms. J Phycol. 2001;37(5):744–55.
- 113. Wu H, Roy S, Alami M, Green BR, Campbell DA. Photosystem II photoinactivation, repair, and protection in marine centric diatoms. Plant Physiol. 2012;160(1):464–76. pmid:22829321
- 114. de Carvalho CCCR. Marine Biofilms: A Successful Microbial Strategy With Economic Implications. Front Mar Sci. 2018;5.
- 115. Lauritano C, Galasso C. Microbial Interactions between Marine Microalgae and Fungi: From Chemical Ecology to Biotechnological Possible Applications. Mar Drugs. 2023;21(5):310. pmid:37233504
- 116. Zulfiqar M, Syhapanha K. Liquid Chromatography Mass Spectrometry Data (LCMS-1) from Microalgal Co-culture of Skeletonema marinoi and Prymnesium parvum. 2024 [cited 2023 Dec 23]. In: Zenodo: [Internet]. Available from: https://zenodo.org/records/10143127
- 117. Zulfiqar M, Syhapanha K. Tandem Mass Spectrometry Data (LCMS-2) from Microalgal Co-culture of Skeletonema marinoi and Prymnesium parvum. 2024 [cited 2023 Dec 23]. In: Zenodo: [Internet]. Available from: https://zenodo.org/records/10143233
- 118. Zulfiqar M. Analysis of Metabolomics Data to Assess Interactions in Microalgal Co-culture of Skeletonema marinoi and Prymnesium parvum. 2023 [cited 2023 Dec 23]. In: Zenodo: [Internet]. Available from: https://zenodo.org/records/10143554
- 119. Zulfiqar M. MAW-Diatom/suspectlist_curation.py at main. [cited 2023 Dec 23]. In: Zenodo: [Internet]. GitHub. Available from: https://github.com/zmahnoor14/MAW-Diatom/blob/main/suspectlist_curation.py
- 120. Zulfiqar M. ID 1006530 - BioProject - NCBI [Internet]. [cited 2023 Dec 23]. Available from: https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1006530