Halomonas meridiana Slthf1 (DSM 15724; Gram negative bacterium) was obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ). It should be noted this strain has been synonymized with H. aquamarina based on phylogenomic classifications (Dobson and Franzmann, 1996), but H. meridiana remains validly published as a heterotypic synonym according to the International Code of Nomenclature of Prokaryotes (ICNP). Due to constantly evolving taxonomy, we refer to H. meridiana Slthf1 as “Slthf1” in the proceeding text. Slthf1 was isolated in a deep-sea hydrothermal environment. Such environments could have supported prebiotic chemistry on Earth (Martin et al., 2008; Sojo et al., 2016) and may occur within icy moon oceans (Vance et al., 2007; Hsu et al., 2015; Russell et al., 2017). Typical phenotypic characteristics of the hydrothermal-vent habitat of origin are exhibited by Slthf1 (Kaye and Baross, 2004; Kaye et al., 2004; Takahashi et al., 2020). This includes not only adaptability to high salinity (growth in up to 22% (w/v) NaCl) and high alkalinity (tolerance up to pH 12), but also, genomic adaptations to the cold (possessing three cold shock protein genes, growth at −1 °C) and high-pressure deep-sea environment (growth at 550 bar). This combination of polyextremophilic adaptations makes Slthf1 a superior model organism compared to halophilic archaea for studying potential life in cold, saline-alkaline environments similar to those presented in icy moon subsurface oceans. Additionally, Slthf1 has no known specialised adaptations to ammonium. This was an intentional choice. Ammonium content in the oceans of icy moons is such that adaptation to ammonium may not be required for survival. The intention of this study was to assess survival in ammonium, not to study an already established ammonium adaptation. The complete genome sequence for this organism is also available [DDBJ, accession no. AP022821] (Takahashi et al., 2020). Aerobic culture of Slthf1 was performed in glass conical Erlenmeyer flasks in an orbital benchtop shaking incubator set to rotate at 150 RPM, 28 °C. Slthf1 was cultivated in a yeast media consisting of 1 g/100 mL Bacto™ yeast extract (Becton, Dickinson and Company), 0.2 M NaCl (Thermo Fisher Scientific, CAS Number: 7647-14-5) and distilled water.

Solutions of ammonium and sulfate salts were prepared to 0.1 M, 0.25 M, 0.5 M, 0.75 M and 1 M from 2 M stock solutions diluted into yeast media. Ammonium salts included: (NH₄)₂SO₄; Fisher Scientific, CAS Number: 7783-20-2, ammonium nitrate (NH₄NO₃; Scientific Laboratory Supplies, CAS Number: 6484-52-2) and ammonium chloride (NH₄Cl; Honeywell Research Chemicals, CAS Number: 12125-02-9). In addition to (NH₄)₂SO₄, sulfate salts included sodium sulfate (Na₂SO₄; Sigma Aldrich CAS number: 7757-82-6), and potassium sulfate (K₂SO₄; Acros Organics, CAS number: 7778-80-5). Owing to limited solubility of K₂SO₄, molarities beyond 0.5 M were not tested. (NH4)₂SO₄ was also prepared in yeast media at concentrations of 0.05 M, 0.125 M and 0.375 M to achieve equivalent NH4+ concentrations of 0.1 M, 0.25 M and 0.75 M, respectively, in accordance with the stoichiometry of NH4+ ions in other ammonium brines. The pH of solutions was determined with a Jenway 3510 benchtop pH meter. All solutions were between pH 5.5 and pH 6.5. Solution pH remained unmodified in order to preserve a high NH4+/NH₃ ratio at acidic pH. Solutions of matching pH were created by addition of HCl or NaOH into yeast media. Solutions were matched to within ± 0.01 pH units. All solutions were filter-sterilized through a 0.22-micron pore before use.

The growth kinetics of Slthf1 cultivated in ammonium and sulfate salts was determined by recorded optical density (OD) measurements at 600 nm (OD₆₀₀). Overnight Slthf1 culture was inoculated to OD₆₀₀ = 0.05 into the selected brines. Controls were prepared by Slthf1 inoculation into unamended yeast media. Negative controls had no inoculation. Samples were seeded into a 96-well plate with a low evaporation lid and measurements taken with a BMG SPECTROstar Nano Microplate Reader over 48 h (h) at 28 °C. For cell viability assays, (NH₄)₂SO₄ solutions were prepared to 1 M in yeast media. Brines were inoculated with overnight culture of Slthf1 to OD₆₀₀ = 0.05 in a 96-well plate. Cultures were incubated in a tabletop shaker at 28 °C for 72 h. Cell viability was examined using colony forming units (CFU) on yeast media agar and incubated at 28 °C for 3 days prior to enumeration. To prevent condensation, 96-well plate lids were treated with a solution of Triton X-100 (0.05%) in 20% ethanol in all growth experiments.

Growth curves were analyzed to determine growth rate and final cell density at 600 nm. Growth rate, μ, was calculated as per Equation 1, where N₀ is the OD₆₀₀ at the beginning of a selected time interval (t₀) in the exponential growth phase; N is the OD₆₀₀ at the end of a selected time interval (t) in the exponential growth phase. t and t₀ were recorded in minutes.

Final cell concentration was indicated by the final OD₆₀₀ reached after 48 h. Measurement of OD₆₀₀ vs. cell viability confirms that an increase in OD₆₀₀ reflects increased viability and proliferation of cells (Supplementary Figure S1).

Water activities were measured in the laboratory with a Rotronic HP23-AW water activity meter (Rotoronic AG, Bassersdorf, Switzerland). Solutions were prepared and measured after a time interval of 1.5 h to allow equilibration.

Slthf1 was cultivated overnight and inoculated to OD₆₀₀ = 0.05 into 0.5 M (NH₄)₂SO₄ or unamended yeast media (control) within a 24-well plate. Growth at 28 °C was assessed by OD₆₀₀ readings every 30 min using a BMG SPECTROstar Nano Microplate Reader. Slfth1 was harvested at OD₆₀₀ = 0.5 following 28 h growth. An aliquot of each sample was placed into a microcentrifuge tube and briefly incubated on ice. Samples were retained for transmission electron microscopy (TEM) as described in the TEM preparation section. The remaining samples were quenched by rapid cooling in a dry ice-ethanol bath (70% v/v ethanol). Samples were vigorously mixed to prevent freezing. Any spent medium was discarded by centrifugation at 1,000 × g for 10 min at 4 °C followed by removal of supernatant. Metabolites were extracted by application of ice-cold chloroform/methanol/water (1:3:1). During metabolite extraction, cell lysis was encouraged by sonication of samples in water for 5 min at 37 kHz in an ultrasonication bath (Elmasonic S 60 H) maintained at 4 °C with ice. Extraction mixtures were shaken at 1,200 RPM for 1 h at 4 °C and centrifuged at 13,000 × g for 3 min at 4 °C. The metabolite-rich supernatant was harvested into sterile microcentrifuge tubes and maintained at −80 °C until analysis. A quality control sample was created by pooling equal volumes of metabolites from all samples, which was also maintained at −80 °C until analysis.

Global metabolomic profiling was conducted using liquid chromatography (LC) coupled with ion mobility (IM) quadrupole time-of-flight (qTOF) mass spectrometry (MS). The system consisted of an Agilent 1290 Infinity II series ultra-high-performance liquid chromatography (UHPLC) setup interfaced with an Agilent 6560 IM-qTOF mass spectrometer equipped with a Dual Agilent Jet Stream Electron Ionization source. Chromatographic separation was achieved using an InfinityLab Poroshell 120 HILIC-Z UHPLC column (2.1 mm × 50 mm, 2.7 μm) coupled to an InfinityLab Poroshell 120 HILIC-Z guard column (3.0 mm × 2.7 μm), both sourced from Agilent Technologies (689775–924 and 823750–948, respectively). A gradient elution was performed over 3.5 min, utilizing an organic solvent (acetonitrile) in combination with an aqueous buffer, either low-pH (10 mM ammonium formate, pH 3) for positive ionization or high-pH (10 mM ammonium acetate, pH 9) for negative ionization. Data were collected using MassHunter Data Acquisition 10.0 software, with 1 μL of each sample injected at a flow rate of 800 μL/min. A pooled quality control (QC) sample, comprising equal volumes of all experimental samples, was injected five times at the beginning of the experiment to equilibrate the column and after every subsequent set of five test samples to monitor system stability during data acquisition. Mass spectrometry data were acquired over a m/z range of 50–1,700, with a scan rate of 0.8 scans per second. The metabolomic analysis was performed at the EdinOmics research facility (RRID: SCR_021838) at the University of Edinburgh.

Analysis of the raw data files was performed by the Agilent MassHunter software suite. Specifically, ion multiplexed and calibration files underwent demultiplexing with the PNNL PreProcessor v2020.03.23, utilizing default settings for tasks such as demultiplexing, moving average smoothing, saturation correction, and spike removal. For recalibration, accurate mass and drift time adjustments were made using AgtTofReprocessUi and IM-MS Browser 10.0, respectively. Molecular features were extracted using Mass Profiler 10.0, with parameters set for retention time tolerance (±0.3 min), drift time tolerance (±1.5%), and accurate mass tolerance (± 5 ppm + 2 mDa). Feature annotation was carried out by matching accurate mass and collision cross-section (CCS) values to the McLean CCS Compendium PCDL library (Nichols et al., 2018). Statistical analyses was performed via the MetaboAnalyst 6.0 online platform (Pang et al., 2024), with data log-transformed and Pareto-scaled before analysis. Annotated molecular features were used to generate principal component analysis (PCA), volcano analysis, unpaired t-test and box plots. For pathway analysis, compound names were first converted to ID labels according to the human metabolome database (HMDB). Compound HMDB ID with relative intensities were submitted to the MetaboAnalyst 6.0 online platform pathway analysis tool. Data was log-transformed, auto-scaled and examined against the H. meridiana SCSIO 43005 KEGG pathway library using global test and relative betweenness centrality methods. Altered pathways with a p-value < 0.05 and FDR < 0.05 were considered significant. Significantly altered metabolites in the unpaired t-test that were also identified as altered in the pathway analysis are depicted in box and whisker plots. The plots were retrieved following t-test analysis on the MetaboAnalyst 6.0 online platform. Normalized values are presented. The raw data associated with this study is available in the Supplementary Data Sheet. This study focuses on metabolomic changes in (NH₄)₂SO₄, but the broader metabolomic profiling also included samples cultivated in NH₃ and NaOH. For the purpose of this study, only (NH₄)₂SO₄ and control samples were included. The metabolomics of NH₃ and NaOH exposed samples were addressed in a separate analysis (Hopton et al., 2025).

Cultures of Slthf1 cultivated in 0.5 M (NH₄)₂SO₄ were harvested during metabolomics sampling, prior to extraction. Cells were pelleted by centrifugation at 5,000 × g and supernatant removed. The pellet was washed and resuspended in phosphate-buffered saline (PBS). Following centrifugation at 5,000 × g and supernatant removal, the cell pellets were fixed in 3% glutaraldehyde prepared in 0.1 M sodium cacodylate buffer (pH 7.3) for 2 h, followed by three 10 min washes in 0.1 M sodium cacodylate. Post-fixation was carried out using 1% osmium tetroxide in 0.1 M sodium cacodylate for 45 min, followed by a series of three 10 min washes in 0.1 M sodium cacodylate. The samples were dehydrated sequentially in ethanol solutions at 50%, 70%, 90%, and 100% for 15 min each. This was repeated in triplicate and followed by two 10 min washes in propylene oxide. The samples were embedded in TAAB 812 resin. Sections of 1 μm thickness were prepared using a Leica Ultracut ultramicrotome, stained with Toluidine Blue, and examined under a light microscope to identify regions of interest. Ultrathin sections (60 nm thick) were cut from these selected regions, stained with uranyl acetate and lead citrate, and observed using a JEOL JEM-1400 Plus TEM. Representative images were acquired with a GATAN OneView camera at 4K resolution and subsequently processed using ImageJ software (version 57).

Normality of data was assessed with the Shapiro-Wilk test. For comparison of two groups, equal variance was assessed with an F-test. Groups of equal variances were analyzed by unpaired two-tailed t-test. Groups of unequal variances were assessed by unpaired two-tailed t-test with Welch's correction. For analysis of three or more groups, equal variance was assessed by the Brown-Forsythe test. Samples of equal variance were analyzed by analysis of variance (ANOVA) followed by Tukey's post-hoc test. For samples where variance was not equal, Welch's ANOVA test with Tamhane's T2 post-hoc test was applied. For datasets with non-normal distribution, means were compared using the Kruskal–Wallis test with Dunn's multiple comparisons test. Statistical tests are specified in figure legends. Results where p < 0.05 were considered significant. All data was compiled from at least three biological replicates (n = 3–5). Data is presented as the mean ± standard deviation (SD). All figures and statistical analyses were produced using GraphPad Prism version 8.0.2 (GraphPad Software Inc.).

Growth of Slthf1 over 48 h in increasing concentrations of (NH₄)₂SO₄ was investigated to assess concentration thresholds of growth in (NH₄)₂SO_4. Concentrations of 0.1 M, 0.25 M, 0.5 M, 0.75 M and 1 M were utilized, with unamended yeast media, 0 M (NH₄)₂SO₄, as a control. The resulting growth curves are depicted in Figure 1A. Growth progressively declined with increasing (NH₄)₂SO₄, with minimal cell density observed at 1 M (NH₄)₂SO_4. Cell viability assay confirmed that Slthf1 remained viable at 1 M (NH₄)₂SO₄ after 72 h incubation (Figure 1B). Growth rate and final cell density at 48 h are shown in Figures 1C, D, respectively. Overall, growth of Slthf1 was limited by increasing concentrations of (NH₄)₂SO_4. Growth rate was non-significant from control at concentrations of 0.1 M (p = 0.599). Successive reduction in growth rate compared to control was observed in 0.25 M (p < 0.05), 0.5 M (p < 0.01), 0.75 M (p < 0.01) and 1 M (p < 0.01) brines. However, reduction in growth rate does not affect final cell density when grown up to 0.5 M (NH₄)₂SO₄; at 48 h, there was no significant difference between the OD₆₀₀ in control solutions compared to 0.1 M (p = 0.766), 0.25 M (p = 0.825) and 0.5 M (p = 0.815). Cell density was lower compared to control in 0.75 M (p < 0.01) and 1 M (NH₄)₂SO₄ (p < 0.01). Cell density remained above OD₆₀₀ = 2.00 when cultivated in the control, 0.1 M, 0.25 M and 0.5 M (NH₄)₂SO₄. Cell density was below OD₆₀₀ = 1.00 in 0.75 M (NH₄)₂SO₄ (OD₆₀₀ = 0.847 ± 0.448). The average cell density of Slthf1 after 48 h incubation in 1 M (NH₄)₂SO₄ was 0.2 OD₆₀₀ ± 0.029.

The established survival limits of Slthf1 in increasing concentrations of (NH₄)₂SO₄ could be due to altered water availability, salinity, osmotic pressure, ion induced toxicity or pH changes. To investigate these possibilities, Slthf1 was cultivated in ammonium (NH₄Cl, NH₄NO₃) and sulfate salts (Na₂SO₄, K₂SO₄) at concentrations of 0.1 M (Figure 2A), 0.5 M (Figure 2B) and 1 M (Figure 2C), in addition to (NH₄)₂SO₄. Growth was assessed by OD₆₀₀ after 48 h incubation. For comparison against ammonium salts, (NH₄)₂SO₄ was prepared to concentrations of 0.05 M, 0.25 M and 0.5 M to ensure ionic levels of NH4+ were equivalent to NH₄Cl and NH₄NO₃ at 0.1 M, 0.5 M, 1 M, respectively. Salinity, osmolarity and ionic strength of each brine is displayed in Table 1. Slthf1 was also grown in unamended yeast media pH-matched to (NH₄)₂SO₄ brines using NaOH or HCl. Slthf1 reached an OD₆₀₀ > 2 in all brines at 0.1 M (Figure 2A). The OD₆₀₀ at 48 h of Slthf1 in 0.05 M (NH₄)₂SO₄ (0.1 M NH4+) was found to be not significantly different in 0.1 M NH₄Cl (p = 0.380) and the pH-matched solution at pH 6.38 (p > 0.999) (Figure 2A). There was a higher OD₆₀₀ in 0.1 M NH₄NO₃ compared to 0.05 M (NH₄)₂SO₄ (p < 0.01). When matching the molar concentration of SO42- ion, the OD₆₀₀ at 48 h was lower in 0.1 M (NH₄)₂SO₄ compared to 0.1 M Na₂SO₄ (p < 0.01) and 0.1 M K₂SO₄ (p < 0.05). Growth in the pH-matched solution at pH 6.38 was not significantly different from growth in 0.1 M (NH₄)₂SO₄. At 0.5 M, there was no significant difference between the OD₆₀₀ at 48 h for any of the tested brines compared to growth in 0.25 M (0.5 M NH4+) and 0.5 M (NH₄)₂SO₄ (Figure 2B) (p-values in Supplementary Table S1), despite differential salinity, osmolarity and ionic strengths between certain brines.

Alterations to cell density became evident when brine concentrations reached 1 M (Figure 2C). Slthf1 grew in 1 M NH₄Cl (p = 0.883) and a pH-matched solution at pH 5.8 (p = 0.933) with a non-significant change to OD₆₀₀ at 48 h compared to 0.5 M (NH₄)₂SO₄ (1 M NH4+). Cell density was maintained above OD₆₀₀ = 1.8 in these solutions. OD₆₀₀ in 1 M NH₄NO₃ was lower compared to (NH₄)₂SO₄ (p < 0.05), with a final OD₆₀₀ < 0.05 indicating severely limited growth. Thus, the molarity of NH4+ ion alone does not determine growth outcomes. Growth in 1 M (NH₄)₂SO₄ was significantly lowered compared to growth in 1 M Na₂SO₄ (p < 0.01) and a pH-matched solution at pH 5.7 (p < 0.01), despite the fact that Na₂SO₄ displayed higher salinity, and equal osmolarity and ionic strength compared to (NH₄)₂SO₄ (Table 1). This confirms molarity of SO42- alone does not determine growth outcomes. The difference between the OD₆₀₀ at 48 h in 1 M Na₂SO₄ and the pH-matched solution at 5.7 was found to be non-significant (p = 0.108). Water availability was assessed by water activity measurements of the brines at each concentration −0.1 M (Figure 2D), 0.5 M (Figure 2E) and 1 M (Figure 2F). The water activity of all brines was found to be above 0.9 a_w (Figures 2D–F). There was a non-significant difference between the a_w of ammonium salts and the a_w of sulfate salts (p-values in Supplementary Table S2). The results of these tests suggest that neither toxicity by individual ions, osmotic stress, ionic strength, salinity nor pH were contributing factors that limit growth at higher concentrations of (NH₄)₂SO₄.

Metabolites can indicate stress (Avci, 2024; Sharma et al., 2025), and can additionally be utilized as biomarkers in the search for life (Fairén et al., 2020; Weber et al., 2023). To examine the stress response and adaptations in (NH₄)₂SO₄, comparative untargeted metabolomics was performed in Slthf1 cultivated under two conditions: 0.5 M (NH₄)₂SO₄ and unamended yeast media (0 M (NH₄)₂SO₄, hereafter denoted “control”). The annotated molecular features with relative intensities underwent multivariate and univariate statistical analysis, the results of which are shown in a PCA scores plot (Figure 3A), volcano plot (Figure 3B), and unpaired t-test (Supplementary Table S3). The PCA scores plot shows clear separation of the 0.5 M (NH₄)₂SO₄ group from control group with no overlap. The metabolites attributed to this group differentiation were identified using univariate volcano analysis, using a fold change greater than 2 and a p-value < 0.05 (adjusted using FDR correction). Volcano analysis revealed significant elevation of 17 molecular features (p < 0.05, FDR corrected), and significant reduction of 24 molecular features (p < 0.05, FDR corrected) in 0.5 M (NH₄)₂SO₄ cultivated samples compared to control samples. The complete volcano analysis dataset for this comparison is shown in Supplementary Table S4. The altered metabolites included amino acids and derivatives; there was an enrichment of aspartate (FC = 7.24, p < 0.05) and D-allo-isoleucine (FC = 8.42, p < 0.0001), and a reduction to the levels of serine (FC = 1.2 × 10⁻⁸, p < 0.0001), glutamine (FC=1.15 × 10⁻⁸, p < 0.0001) and N-acetyl-L-aspartate (FC = 0.159, p < 0.05) in 0.5 M (NH₄)₂SO₄ cultivated samples compared to control samples.

Lipids were also found to be significantly altered in the volcano analysis. Figure 4A depicts lipid alterations as Log2(FC) from control sample. The levels of unsaturated phosphatidylcholine (PC) 36:05 (FC = 1.6 × 10⁹, p < 0.0001), PC (18:1/18:1) (del9-trans) (FC = 4.83, p < 0.01), PC [16:1(9Z)/16:1(9Z)] (FC = 11.35, p < 0.05) and phosphatidylethanolamine (PE) 37:01 (FC = 3.16 × 10⁷, p < 0.001) and PE (O-34:03) (FC = 46.15, p < 0.05) were higher in (NH₄)₂SO₄ cultivated Slthf1. There were lower levels of unsaturated 40-carbon phosphatidylserine (FC = 3.16 × 10⁻⁷, p < 0.001), and a small but significant elevation in the levels of saturated stearic acid (FC = 2.43, p < 0.01) in Slthf1 cultivated in (NH₄)₂SO₄ compared to the control. These alterations suggest cell wall modulation; morphological changes in Slthf1 cultivated in control and 0.5 M (NH₄)₂SO₄ solutions are shown in Figures 4B, C, respectively. Cells in both conditions exhibited irregular, undulating outer membrane morphology with an enlarged periplasm between inner and outer membrane. Cytoplasm showed an abundance of ribosomes and nucleoids in both conditions. PHA-like granules were also apparent in both conditions but significantly greater in number in the control condition. This possibly suggested nitrogen limitation in the growth media that was satisfied by addition of NH4+ in cells cultivated in 0.5 M (NH₄)₂SO₄. Cells without membranes, indicating cell lysis events, were evident with greater occurrence in the (NH₄)₂SO₄ cultivated cells. There was electron-dense material observed between cells cultivated in (NH₄)₂SO₄ that may indicate microbial interactions with (NH₄)₂SO_4.

To identify pathways altered upon 0.5 M (NH₄)₂SO₄ exposure, a pathway enrichment analysis of the annotated metabolites was conducted using the MetaboAnalyst 6.0 platform. Significantly altered metabolites in the unpaired t-test (Supplementary Table S3) that were also identified as altered in the pathway analysis are depicted in box and whisker plots in the following sections. Pathway analysis identified 12 significantly (p < 0.05) altered pathways. Figure 5A depicts the altered pathways. Associated significance values are in Supplementary Table S5. The altered pathways correspond to sphingolipid, nitrogen, purine, glyoxylate and dicarboxylic, amino acid, folate, pyruvate, butanoate metabolism and the citric acid cycle. The levels of serine were found to be lower in 0.5 M (NH₄)₂SO₄ cultivated cells relative to control samples (p < 0.0001). This resulted in the pathway of sphingolipid metabolism to appear significantly altered. However, based on the complete genome of Slthf1, we do not believe this organism to be capable of sphingolipid metabolism. The next most significantly altered pathway corresponded to nitrogen metabolism, attributed to the significant reduction to glutamine (p < 0.0001) in (NH₄)₂SO₄ samples compared to control samples. Glutamate was not identified as significantly altered. The reduced levels of glutamine could suggest, in NH4+ surplus, nitrogen assimilation shifted from utilizing glutamine. The proposed alternative pathway for nitrogen metabolism is presented in Figure 5B.

Numerous metabolites involved in purine metabolism were identified as altered (22/70) following 0.5 M (NH₄)₂SO₄ cultivation (Figure 5A). Figure 6 depicts boxplots of the metabolites within purine metabolism that were found to be significantly altered in both the pathway and volcano analysis—glutamine (p < 0.0001), deoxyguanosine (p < 0.001), guanosine (p < 0.0001), hypoxanthine (p < 0.0001) and inosine (p < 0.0001). The KEGG pathway for purine metabolism attributes these molecules to purine biosynthesis (glutamine) (Figure 6A), guanine metabolism (deoxyguanosine, guanosine) (Figure 6B) and adenine ribonucleotide degradation (hypoxanthine, inosine) (Figure 6C). Pathway analysis also identified 3′,5′-cyclic-GMP as significantly elevated (p < 0.05) and adenine as significantly reduced (p < 0.05) in 0.5 M (NH₄)₂SO₄ cultivated samples compared to control, further suggesting reduced turnover of 3′,5′-cyclic-GMP for production of guanine and reduced adenine biosynthesis.

Pathway analysis indicated four amino acid pathways altered in response to 0.5 M (NH₄)₂SO₄ cultivation (Figure 5A). Box plots of the altered metabolites identified in the pathway analysis related to amino acid metabolism that were also found to be altered in the volcano analysis are depicted in Figure 7. These correspond to the metabolism of glycine, serine and threonine (p < 0.001) (Figure 7A), D-amino acids (p < 0.01) (Figure 7B) and alanine, aspartate and glutamate (p < 0.001) (Figure 7C). All comparisons made below reference metabolites altered in samples cultivated in 0.5 M (NH₄)₂SO₄ compared to control samples. Five metabolites out of 33 total metabolites were found to be altered in glycine, serine and threonine metabolism. Of these, two were found to be significantly reduced: serine (p < 0.0001) and dimethylglycine (p < 0.0001). Reduction of serine may account for the reduction to PS. Four metabolites related to D-amino acid biosynthesis were found to be altered. These included significantly lower levels of serine (p < 0.0001). It is notable that D-allo-isoleucine was also found to be significantly elevated in the volcano analysis (p < 0.001). In alanine, aspartate and glutamate metabolism, 11/28 metabolites were found to be altered. These included significantly higher levels of aspartate (p < 0.01), fumarate (p < 0.0001), and significantly lower levels of N-acetyl-L-aspartate (p < 0.01), glutamate (p < 0.05), glutamine (p < 0.0001) and succinate (p < 0.05). Not depicted in Figure 7, the significant reduction to the levels of glutamate (p < 0.05) and glutamine (p < 0.0001), and significant elevation to the levels of aspartate (p < 0.01) and fumarate (p < 0.0001), were also relevant to arginine metabolism.

The citric acid cycle was found to be significantly altered (p < 0.05) in the pathway analysis (Figure 5A). Pathway analysis revealed significant reduction to succinate (p < 0.05) and elevation to fumarate (p < 0.0001) in the 0.5 M (NH₄)₂SO₄ cultivated sample compared to control. It is notable that elevated levels of O-succinyl-homoserine (FC = 1.44 × 10⁹, p < 0.0001), a succinate precursor, and reduced levels of 4-Guanidinobutanoate (FC = 0.431, p < 0.05), a product of arginine degradation which is subsequently converted to succinate, were also identified in the volcano analysis. Figure 8A depicts box plots of significantly altered metabolites in the citric acid cycle identified in both the pathway and volcano analysis. Pyruvate metabolism was also found to be significantly altered (p < 0.01) with significant alteration to fumarate (p < 0.0001), and non-significant alteration of (S)-lactate (p = 0.269) and pyruvate (p = 0.116). Reduced levels of glutamate (p < 0.05), succinate (p < 0.05), and acetoacetate (p = 0.119) in the 0.5 M (NH₄)₂SO₄ cultivated cells compared to control also caused butanoate metabolism to be found as significantly altered (p < 0.01). The most significantly altered pathway was that of glyoxylate and dicarboxylate metabolism (p < 0.001). Three metabolites of this pathway were found to be significantly reduced in the 0.5 M (NH₄)₂SO₄ cultivated cells compared to control: glutamate (p < 0.05), glutamine (p < 0.0001) and serine (p < 0.0001). Figure 8B depicts box plots of significantly altered metabolites in the glyoxylate cycle identified in both the pathway and volcano analysis. Notably, we additionally found the methylaspartate cycle intermediate N-methylaspartate to be reduced in the volcano analysis (FC = 0.108, p < 0.05) in (NH₄)₂SO₄ cultivated cells (Figure 3B).