**Intracellular and Extracellular Metabolites from the Cyanobacterium** *Chlorogloeopsis fritschii,* **PCC 6912,**

#### **Bethan Kultschar 1,\*, Ed Dudley 2, Steve Wilson <sup>3</sup> and Carole A. Llewellyn 1,\***

**During 48 Hours of UV-B Exposure**


Received: 12 March 2019; Accepted: 13 April 2019; Published: 16 April 2019

**Abstract:** Cyanobacteria have many defence strategies to overcome harmful ultraviolet (UV) stress including the production of secondary metabolites. Metabolomics can be used to investigate this altered metabolism via targeted and untargeted techniques. In this study we assessed the changes in the intra- and extracellular low molecular weight metabolite levels of *Chlorogloeopsis fritschii* (*C. fritschii*) during 48 h of photosynthetically active radiation (PAR) supplemented with UV-B (15 μmol m−<sup>2</sup> s−<sup>1</sup> of PAR plus 3 μmol m−<sup>2</sup> s−<sup>1</sup> of UV-B) and intracellular levels during 48 h of PAR only (15 μmol m−<sup>2</sup> s<sup>−</sup>1) with sampling points at 0, 2, 6, 12, 24 and 48 h. Gas chromatography–mass spectrometry (GC–MS) was used as a metabolite profiling tool to investigate the global changes in metabolite levels. The UV-B time series experiment showed an overall significant reduction in intracellular metabolites involved with carbon and nitrogen metabolism such as the amino acids tyrosine and phenylalanine which have a role in secondary metabolite production. Significant accumulation of proline was observed with a potential role in stress mitigation as seen in other photosynthetic organisms. 12 commonly identified metabolites were measured in both UV-B exposed (PAR + UV-B) and PAR only experiments with differences in significance observed. Extracellular metabolites (PAR + UV-B) showed accumulation of sugars as seen in other cyanobacterial species as a stress response to UV-B. In conclusion, a snapshot of the metabolome of *C. fritschii* was measured. Little work has been undertaken on *C. fritschii*, a novel candidate for use in industrial biotechnology, with, to our knowledge, no previous literature on combined intra- and extracellular analysis during a UV-B treatment time-series. This study is important to build on experimental data already available for cyanobacteria and other photosynthetic organisms exposed to UV-B.

**Keywords:** cyanobacteria; *C. fritschii*; UV-B; PAR; time-series; intracellular; extracellular; metabolites; GC–MS

#### **1. Introduction**

Cyanobacteria are gram-negative bacteria with the ability to photosynthesise, assimilating CO2 into a variety of biochemical compounds through different metabolic pathways [1]. Cyanobacteria can thrive in a wide variety of extreme habitats such as high ultraviolet radiation (UVR) due to their adaptive capabilities such as the production of secondary metabolites [1]. Metabolomics can be used to determine changes at the metabolite level during varying environmental stimuli and is a useful tool in cyanobacterial research [2]. The metabolome provides information closely reflecting the interaction between an organism and its environment. Some metabolites produced by cyanobacteria under stress

conditions are unique and are of increasing interest from a biotechnological perspective as sustainable sources of ingredients in a variety of industries [3,4].

The effect of UVR on cyanobacteria has been widely researched including the interaction with biomolecules, production of reactive oxygen species (ROS) which cause oxidative stress, impaired growth, partial inhibition of photosynthesis and decreased enzyme activity [5–7]. UVR also has a role as an activator of secondary metabolite production such as mycosporine-like amino acids (MAAs) [8] and other protective secondary metabolites [9]. Many studies have been conducted to identify these targeted intracellular metabolites during UV-B and UV-A exposure in *Lyngbya* sp. CU2555 [10], *Nostoc commune* [11], *Anabaena variabilis* PCC 7937 [12], *Calothrix* sp. [13] and *Chlorogloeopsis fritschii* (*C. fritschii)*, PCC 6912, [14] to name a few. Other studies conducted have sought to evaluate changes at the protein level [15,16], targeted and untargeted metabolomic analysis using different intensities of UV-B [17] and combined metabolomic and proteomic analysis during UV-A exposure [18].

Cyanobacteria convert CO2 into reduced carbon which forms the backbone of metabolites and are central to life. Like many other microorganisms, cyanobacteria release these carbon-based primary and secondary metabolites into their surrounding area which drives carbon cycling within microbial communities [19,20]. These released metabolites are by-products of metabolism within cells and make up a small proportion of the dissolved organic matter (DOM) pool within freshwater and marine ecosystems [19]. Consisting of a variety of chemical compositions such as; polysaccharides, proteins, lipids, organic compounds or inorganic molecules, they are released for communication, structural organisation, and defence against biotic and abiotic factors [20–22]. The uptake and release of metabolites change with varying environments; examples include the release of exopolysaccharides during high light and UVR [11,23].

Monitoring industrially relevant metabolites released by microorganisms into their surroundings is a widely used technique in the fermentation industry. It can be used in bioprocess monitoring, fermentation biomarker identification, for monitoring metabolite levels in fermentation processes and microbial contamination [24,25].

Combining intracellular and extracellular analysis is useful in the study of cyanobacteria providing a more holistic picture of metabolite production during growth and its response to different environmental conditions [25,26].

Little work has been undertaken on monitoring both intracellular and extracellular metabolites in cyanobacteria especially *C. fritschii,* a potential candidate for use in industrial biotechnology due to its scalability [27] and tolerance to different growth conditions [28–30]. In this study, we observe the changes in metabolites produced by *C. fritschii* during 48 h of UV-B exposure as detected by untargeted gas chromatography-mass spectrometry (GC–MS). We were able to identify metabolites with altered levels comparing UV-B treatment (PAR + UV-B) to cultures irradiated with PAR only.

#### **2. Results**

#### *2.1. Intracellular and Extracellular Analysis of C. fritschii during UV-B Stress Response*

The metabolite profiles of *C. fritschii* cultures (*n* = 3) were investigated during 48 h of UV-B exposure. At each time point (0, 2, 6, 12, 24 and 48 h) intracellular and extracellular metabolites were analysed by untargeted GC–MS to evaluate the global changes in metabolite production during UV-B stress.

A total of 300 and 412 peaks were detected from the intracellular and extracellular time-series data respectively (Tables S1 and S2). Using a match factor of 60% or above, 135 and 218 peaks were putatively identified within the intracellular and extracellular data respectively (Tables S1 and S2). The identified chemical structures belonged to a variety of classes such as; acids, alcohols, amino acids, aromatics, fatty acids, heterocycles and sugars.

A Principle Component Analysis (PCA) model was used as an unsupervised multivariate statistical tool to plot and visualise the variance between UV-B exposed samples over time. A total variance of 37.2% for intracellular (Figure 1A, PC1 = 17%, PC2 = 11.1%, PC3 = 9.1%) and 36.4% for extracellular (Figure 1B, PC1 = 18.9%, PC2 = 9.2%, PC3 = 8.3%) was observed.

**Figure 1.** Principle component analysis (PCA) of (**A**) intracellular and (**B**) extracellular gas chromatography-mass spectrometry (GC–MS) data of UV-B exposed (PAR + UV-B) *Chlorogloeopsis fritschii* (*C. fritschii*) cultures showing PC1 vs PC2 only. Each ring represents distribution of biological replicates. S1 = replicate 1, S2 = replicate 2, S3 = replicate 3.

The results of the PCA for intracellular samples (Figure 1A) showed good separation over time between the control (0 h) and 6, 12, 24 and 48 h of UV-B. Less variation was observed between 0 and 2 h of UV-B with clustering seen between 12, 24 and 48 h of UV-B. This result was consistent with the two sample T-test results comparing control (0 h) with each time point where the number of significant features increases with length of UV-B exposure (Table S1). After a one-way analysis of variance (ANOVA) with repeated measures, 112 statistically significant peaks were observed with *p* ≤ 0.05 (Figure S1A), 10 of which remained significant after Bonferroni correction.

From the extracellular data PCA (Figure 1B) a similar pattern was observed with increasing variance with increasing length of UV-B exposure. Statistically significant changes between control (0 h) and each time point, measured using a two-sample T-test also showed increasing significance (*p* ≤ 0.05) with increasing length of UV-B up to 24 h (Table S2). A one-way ANOVA with repeated measures calculated 114 statistically significant peaks with *p* ≤ 0.05 (Figure S1B).

#### 2.1.1. Intracellular Metabolites

28 metabolites (13 represented for simplicity, Figure 2), selected as being involved in the central carbon and nitrogen metabolism within cyanobacteria, were identified within the intracellular GC–MS results Table S3). Many changes in metabolite levels were observed comparing between time points. Glucose, pyruvate and lactate all decreased in abundance after UV-B exposure with significant reduction after 2 h (pyruvate *p* ≤ 0.05, 0 vs. 2 h), 6 h (glucose *p* ≤ 0.05, 2 vs. 6 h) and 12 h (lactate *p* ≤ 0.05, 0 vs. 12 h). Lactate was present during the whole time course whereas glucose and pyruvate were below detection limit after 6 h (*p* ≤ 0.05) and 24 h (*p* ≤ 0.001) respectively.

6 proteinogenic amino acids were detected; serine (ser), glycine (gly), glutamate (glu), proline (pro), tyrosine (tyr) and phenylalanine (phe). All detected amino acids decreased after 6 or 12 h of exposure with the exception of pro. A decrease in tyr, phe and gly was seen after 6 h (tyr *p* ≤ 0.05, 0 vs. 6 h; phe *p* ≤ 0.01, 2 vs. 6 h; gly *p* ≤ 0.05, 2 vs. 6 h) of UV-B followed by no detection at 6, 12, 24, and 48 h (tyr *p* ≤ 0.05; phe *p* ≤ 0.01; gly *p* ≤ 0.05). Ser and glu decreased significantly after 12 h of treatment (ser *p* ≤ 0.05, 0 vs. 12 h; glu *p* ≤ 0.05, 0 vs. 12 h), ser was below detection limit between 12 and 48 h (*p* ≤ 0.05) whereas glu was detected throughout the time series. Proline showed no significant decrease after UV-B exposure with a significant increase observed after 24 h (*p* ≤ 0.05, 6 vs. 24 h).

**Figure 2.** Schematic representation of a generalised reduced carbon metabolism in the cyanobacterium *C. fritschii* showing glycolysis, the citric acid (TCA) cycle, amino acid and fatty acid biosynthesis. Primary metabolites identified in intracellular samples using GC–MS are highlighted in blue with each insert presenting mean values of normalised abundance (normalised to internal standard and dry weight) ± standard error of each metabolite during supplemented UV-B exposure PAR + UV-B). Statistical significance between control (0 h) and UV-B exposure (2, 6, 12, 24 and 48 h) and between each treatment time point was measured using a two-sample T-test with equal variance; \* = 0.05 ≥ *p* ≥ 0.01, \*\* = 0.01 ≥ *p* ≥ 0.001 and \*\*\* = *p* ≤ 0.001.

The fatty acids stearic acid, palmitic acid and mystiric acid all decreased significantly after 2 h of treatment (*p* ≤ 0.05) and their abundance remained lowered throughout the time series (stearic acid *p* ≤ 0.01, 0 vs. 48 h; palmitic acid *p* ≤ 0.05, 0 vs. 48 h; mystiric acid *p* ≤ 0.01, 0 vs. 48 h).

#### Carotenoid and MAA Analysis

Carotenoid concentration (Figure 3A) and MAA content (Figure 3B) were analysed by UV-visible spectroscopy and high performance liquid chromatography (HPLC) respectively. Total carotenoid concentration decreased after 2 h (*p* ≤ 0.05); with a steady significant increase up to 48 h with a final concentration of 2.59 μg/mg dry weight (*p* ≤ 0.05).

**Figure 3.** Carotenoid and mycosporine-like amino acid (MAA) analysis of *C. fritschii* extracts during UV-B exposure. **(A)** Total carotenoid concentration as measured by UV-visible spectroscopy and **(B)** shinorine and mycosporine-glycine (m-gly) content measured by high-performance liquid chromatography (HPLC) analysis. All values are the mean of three biological replicates (normalised to dry weight) ± standard error. Statistical significance was measured using a two-sample T-test with equal variance; \* = 0.05 ≥ *p* ≥ 0.01, \*\* = 0.01 ≥ *p* ≥ 0.001 and \*\*\* = *p* ≤ 0.001.

As described above, UV-B also induces the production of the photoprotective compounds, MAAs. The two forms found in *C. fritschii* are mycosporine-glycine (m-gly) and shinorine [29], both were detected during this experiment with peaks identified using their retention time and absorption maxima (λmax) values. As expected an increase in shinorine (retention time ~4.9 min, λmax = 334 nm) was observed with increasing length of UV-B exposure. No significance was observed with m-gly (retention time ~10.8 min, λmax = 310 nm) content during this experimental time series.

#### 2.1.2. Extracellular Metabolites

29 biologically relevant dissolved metabolites (Table S3) were detected within the extracellular data set (13 represented for simplicity, Figure 4). Citrate, a component of BG-11 medium [31] and involved in the citric acid (TCA) cycle, was consistently detected throughout the time series (ANOVA, *p* ≤ 0.05) along with succinate. Other TCA substrates; malate and fumarate were also detected at 0 h with decreasing abundance after 2 h of UV-B (malate, *p* ≤ 0.001). Other metabolites detected at 0 h which decreased after UV-B exposure were leucine (*p* ≤ 0.01, 0 vs. 6 h), putrescine (*p* ≤ 0.05, 0 vs. 2 h) octanoic acid (*p* ≤ 0.001, 0 vs. 24 h) mystiric acid (*p* ≤ 0.01, 2 vs. 24 h) and fructose (*p* ≤ 0.01, 0 vs. 2 h). Accumulation of the sugars galactose, xylose, lyxose and arabinose was seen after 6 h (galactose *p* ≤ 0.01; arabinose *p* ≤ 0.01) 12 h (arabinose *p* ≤ 0.05), 24 h (arabinose *p* ≤ 0.05; xylose *p* ≤ 0.05; lyxose *p* ≤ 0.01) and 48 h (galactose *p* ≤ 0.05; arabinose *p* ≤ 0.001; lyxose *p* ≤ 0.01) of UV-B exposure. Trehalose was also identified throughout the time series with no significant changes.

**Figure 4.** Time-series exometabolomics data of *C. fritschii* over 48 h of UV-B exposure showing primary metabolites found in extracellular samples only. Statistical significance was measured using a two-sample T-test comparing control (0 h) and UV-B exposure (2, 6, 12, 24 and 48 h) and between each treatment time point, \* = 0.05 ≥ *p* ≥ 0.01, \*\* = 0.01 ≥ *p* ≥ 0.001 and \*\*\* = *p* ≤ 0.001.

#### *2.2. Intracellular Analysis of C. fritschii (PAR Only)*

#### Intracellular Metabolites

A time-series analysis of PAR only (without UV-B supplementation) over 48 h (Table S4) revealed 35 key primary intracellular metabolites (Table S3). 12 were commonly identified between PAR only conditions and during UV-B supplementation (nine represented for simplicity in pathway schematic, Figure 5).

In general, comparing both supplement UV-B and PAR only, the nine common metabolites (Figure 6) showed differences in log2 fold change (FC). The metabolites detected during supplemented UV-B showed negative log2(FC) values which corresponds to reduced metabolite abundances compared to 0 h. Positive log2 (FC)) was generally observed for metabolites detected during PAR only indicating increased abundances compared to 0 h. The main exception is 5-oxoproline that reduced in both UV-B + PAR and PAR only experiments.

**Figure 5.** Schematic representation of a generalised reduced carbon metabolism in the cyanobacterium *C. fritschii* showing glycolysis, TCA cycle, amino acid and fatty acid biosynthesis. Primary metabolites identified in intracellular samples using GC–MS are highlighted in orange with each insert presenting mean values of normalised abundance (normalised to internal standard and dry weight) ± standard error of common metabolites found during PAR only conditions and supplemented UV-B (Figure 2). Statistical significance between 0 h and each time point as well as between time points were measured using a two-sample T-test with equal variance; \* = 0.05 ≥ *p* ≥ 0.01, \*\* = 0.01 ≥ *p* ≥ 0.001 and \*\*\* = *p* ≤ 0.001.

A significant accumulation of glu (*p* ≤ 0.01) was observed after 12 h of PAR only whereas a significant decrease was seen after 12 h of UV-B exposure. Tyr accumulation (*p* ≤ 0.001) was also seen after 2 and 12 h of PAR only conditions with a significant decrease during UV-B treatment. Palmitic and stearic acid remained relatively stable over time with a significant reduction during UV-B exposure observed. Gly and ser abundance remained consistent with significant decreases after 6 and 12 h respectively during UV-B exposure.

**Figure 6.** The changes in common metabolites identified in both UV-B + PAR and PAR only experiments comparing 0 h with 2, 24 and 48 h. Data are presented as Log2 Fold Change (FC) ± standard error.

#### **3. Discussion**

#### *3.1. Intracellular Metabolite Changes and Pathway Analysis*

UV-B exposure is known to reduce growth, photosynthesis and nitrogen fixation in cells to divert energy from key primary pathways to adaptive mechanisms such as; the production of secondary metabolites, MAAs; antioxidant production and DNA/protein repair [32]. A reduction in average dry weight was measured over 24 h of UV-B exposure (Figure S2) with the recovery of the initial biomass concentration and further growth measured between 24 and 48 h. These results showed no significance (*p* ≥ 0.05) across the time series suggesting acclimation of cells to UV-B where damage to photosynthetic systems are counterbalanced by repair and mitigation strategies. A reduction in carotenoid concentration (Figure 3A) was observed after 2 h (*p* ≤ 0.05) due to damage to photosystems caused by UV-B. Accumulation of total carotenoids could be indicative of antioxidant activity as a response to ROS production [6,33].

A reduction in glucose, pyruvate and lactate (Figure 2) could indicate a reduction in CO2 fixation via photosynthesis and further biochemical processes. This could be due to the reduced production of ATP and NADPH2 from photosynthesis [34].

The 13 selected intracellular metabolites (Figure 2) were reduced across the UV-B time series indicating the reduction of cellular processes. A decrease in phe and tyr could be due to their role as precursors to many secondary metabolites such as aromatic nitrogen-containing alkaloids [13]. M-gly and shinorine are produced via a combination of the shikimate/pentose phosphate pathway which

also involves the addition of gly and ser to form the final MAAs [9,35]. The reduction of these amino acids coincides with an increase in MAA levels (Figure 3B).

5-oxoproline and glu are involved in glutathione metabolism. 5-oxoproline reduction (*p* ≤ 0.01) could be due to its interconversion into glutamate which is further converted into the antioxidant glutathione [36]. Glu is also produced from the assimilation of nitrogen during nitrogen fixation which is reduced during UV-B exposure [34].

Pro has been studied in many UV-B experiments involving different photosynthetic organisms and its accumulation is thought to have a role in stress response by providing additional defence as a ROS scavenger and molecular chaperone [37–39]. Accumulation of proline has been observed in *Nostoc punctiforme* during 24 h of UV-A stress [18], in the model plant organism *Arabidopsis* after 24 h of UV-B treatment [40], and also in *C. fritschii* after 24 h of UV-B exposure within this study.

Overall less significant differences were observed during PAR only conditions (Figure 5) compared to PAR supplemented with UV-B (Figure 2).

#### *3.2. Extracellular Metabolite Changes*

The movement of metabolites and substrates between cells and their surrounding environment (or vice versa) can occur via passive and active uptake and efflux systems. Reactions can also occur on cell surface membranes and as transformations of media components [41]. Identification of extracellular metabolite uptake and release from cyanobacteria is, therefore, a complex process due to the high turnover rates of intracellular processes [25]. Extracellular metabolites can be released during stress and as by-products of intracellular reactions [19]. Sugars such as galactose, arabinose, lyxose and xylose are actively released during UV-B stress [5] as seen in this experiment (Figure 4).

7 metabolites from the identified biologically relevant pool were found in both intra- and extracellular metabolite samples (Figure S3). The fatty acid mystiric acid shows a similar pattern of reduced abundance with increasing length of UV-B in both samples (intracellular *p* ≤ 0.05; extracellular *p* ≤ 0.05). Ethanolamine, involved in glycerophospholipid metabolism, and 2-oxobutanoate, involved in amino acid biosynthesis show opposite patterns with intracellular levels decreasing (*p* ≤ 0.05) and extracellular levels increasing (ethanolamine, *p* ≤ 0.01, 2-oxobutanoate, *p* ≤ 0.001) (S1: Figure 3). Further detailed analysis using a combination of -omic techniques and the application of isotopic labelling such as 13C flux balance analysis would be required to better understand the uptake and release of extracellular metabolites and their possible use within cyanobacterial metabolite production [42].

#### **4. Conclusions**

In summary, an untargeted GC–MS workflow was used to evaluate intra- and extracellular metabolites under supplemented UV-B exposure (PAR + UV-B). Most significantly we found a reduction of intracellular metabolites such as the amino acids, tyr, phe, ser, gly and glu and the accumulation of pro, which to our knowledge has not been previously reported in *C. fritschii.* Compared to PAR only, intracellular metabolites showed less significant changes with amino acids tyr, phe and glu accumulation observed.

Although a time series analysis was conducted, this only represents a minuscule proportion of the true changes within the metabolome. This study is important to build on experimental data already available for cyanobacteria and other photosynthetic organisms exposed to UV-B. To understand the changes in primary metabolites and metabolic process with increasing length of UVR exposure to help further understand secondary metabolite production and adaptation of cyanobacteria to UV stress. Further studies are needed to understand and verify these processes within cyanobacteria to aid in the understanding of UV stress adaptation at the metabolite level.

#### **5. Materials and Methods**

#### *5.1. Organism and Growth Conditions*

The cyanobacterium *C. fritschii,* PCC 6912, was obtained from the Pasteur culture collection (PCC) and grown in autoclaved deionised water with filtered BG-11 growth medium (Sigma Aldrich). The strain was maintained in 50 mL BG-11 at a temperature of 27 ± 2 ◦C under continuous PAR illuminated at 15 μmol m−<sup>2</sup> s−<sup>1</sup> (measured using a PAR light sensor, Enviromonitors, West Sussex, UK). Experimental cultures were pre-grown in 300 mL BG-11 media under the same conditions with constant shaking at 80 rpm.

#### *5.2. Experimental Setup*

#### 5.2.1. Supplemented UV-B Experiment (PAR + UV-B)

After 6 days of pre-growth, triplicate experimental *C. fritschii* cultures were transferred into three Quartz Erlenmeyer flasks (H.Baumbach & CO.LTD, Suffolk, UK) at an optical density at 750 nm (OD750nm) of approx. 0.14 to allow even UV-B exposure. The cultures were exposed to a total of 48 h of UV-B radiation using a UVB broadband (290–315 nm, centered at 310 nm) fluorescent tube (Philips TL 20W/12 RS SLV/25, Proflamps, Eindhoven, The Netherlands) emitting 3 μmol m−<sup>2</sup> s−<sup>1</sup> of UV-B radiation (measured using a UVR light sensor, Enviromonitors, West Sussex, UK). The experiment was carried out under continuous PAR at 15 μmol m−<sup>2</sup> s−<sup>1</sup> and shaking at 100 rpm for even UVB exposure of cells. For time course analysis samples were collected at no UV-B (0 h), 2, 6, 12, 24 and 48 h for dry weight, pigment, MAA, and GC–MS analysis.

#### 5.2.2. PAR Only Experiment (PAR only)

*C. fritschii* cultures were pre-grown for 6 days prior to experimental analysis. After 6 days of pre-growth, triplicate experimental *C. fritschii* cultures (OD750nm of approx. 0.13) were grown at a temperature of 27 <sup>±</sup> 2 ◦C under continuous PAR illuminated at 15 <sup>μ</sup>mol m−<sup>2</sup> s−<sup>1</sup> (measured using a PAR light sensor, Enviromonitors, West Sussex, UK) and continuous shaking at 100 rpm. For time course analysis samples were collected at 0, 2, 6, 12, 24 and 48 h for dry weight and GC–MS analysis.

#### *5.3. Sample Harvest and Growth Analysis*

Forty mL volumes of UV-B exposed (*n*=3) and PAR only cultures (*n*=3) were harvested at each time point by centrifugation at 4400 rpm for 20 min to produce a pellet and supernatant. The supernatants (40 mL) were collected and freeze-dried (Edwards, super modulyo) for 72 h. The remaining pellets were transferred into pre-weighed Eppendorf's and freeze-dried for 24 h (Scanvac, CoolSafeTM, LaboGeneTM, Vassingerød, Denmark) for dry weight measurements. Both pellets (PAR + UV-B and PAR only) and dried supernatant (PAR + UV-B only) were stored at −20 ◦C until analysis. OD was monitored using absorbance at 750 nm using a UV-visible spectrophotometer (Shimadzu, UV-2550, Kyoto, Japan).

#### *5.4. GC–MS Analysis*

#### 5.4.1. Sample Preparation

Polar and non-polar metabolites were extracted from UV-B exposed and PAR only dried cell pellets for GC–MS analysis. Briefly, approx. 0.5–1 mg (UV-B exposed) or 1.5–3 mg (PAR only) of dried biomass was re-suspended in 1 mL methanol:chloroform:water (2:2:1) and sonicated using a sonicator probe (Fisher Scientific, FB50) using 6 cycles of 20 s pulses at 40 Hz at 0 ◦C. After centrifugation (5 min at 12,000 rpm), 100 μL of each solvent layer (both methanolic and chloroform layers) were aliquoted into new Eppendorf's and evaporate to dryness using a rotary vacuum concentrator (Eppendorf concentrator 5301).

Dried supernatant (PAR + UV-B) was re-suspended in 1 mL of methanol and centrifuged (4000 rpm, 5 min). Two hundred μL was aliquoted into new Eppendorf's followed by evaporation to dryness and derivatisation as below.

#### 5.4.2. Sample Derivatisation

To each 200 μL dried sample, 30 μL of methoxyamine hydrochloride (23 mg) in pyridine (1.5 mL) was added and samples were heated at 70 ◦C for 45 min. Once cooled to room temperature, 50 μL of MSTFA+TMCS (Thermo ScientificTM, product no: TS-48915) was added and samples heated for an additional 90 min at 40 ◦C. Once cooled to room temperature, 10 μL of tetracosane dissolved in hexane (2 mg/mL) was added as an internal standard. Derivatised samples were transferred into auto-sample vials ready for analysis.

#### 5.4.3. GC–MS Analysis

Derivatised sample (1 μL) was loaded onto an Agilent HP-5MS capillary column (30 m × 0.25 mm <sup>×</sup> 0.25 um) in splitless mode at 250 ◦C. The GC was operated at a constant flow of 1 mL min−<sup>1</sup> helium. The temperature program started at 60 ◦C for 1 min, followed by temperature ramping at 10 ◦C min−<sup>1</sup> temp of to a final 180 ◦C, this was followed by a second temperature ramping at 4 ◦C min−<sup>1</sup> to a final temp of 300 ◦C and held constant at 300 ◦C for 15 min. Data acquisition included a mass range of 50 to 650 and resulted in .D data files for analysis.

Chromatograms were deconvoluted using AMDIS (Automated Mass Spectral Deconvolution and Identification System) followed by alignments using the online portal SpectConnect, http: //spectconnect.mit.edu/ [43], before identifying peaks using Golm metabolome database, www.gmd. mpimp-golm.mpg.de/, and the NIST 05 (National Institute of Standards and Technology) library [44]. MetaboAnalyst, www.metaboanalyst.ca/, was used for statistical analysis [45,46].

#### 5.4.4. GC–MS Data Processing

GC–MS data sets need deconvolution of co-eluting compounds, the freely available software AMDIS was used to process the chromatograms (.D) and produce .ELU files for alignment and conservative component identification using SpectConnect [43,47]. Settings for AMDIS were as followed [18]; Resolution = medium, sensitivity = medium, shape requirement = medium and component width was 10. The resulting .ELU files were uploaded to SpectConnect to produce matrices for further analysis and processing in Excel 2010 (Microsoft, USA). The integrated signal (IS) matrix generated was used for relative quantification of peaks. Triplicate missing data points (within time points) were assumed to be lower than the detection limit and replaced with half of the minimum integrated signal within each data set. Data normalisation was carried out using the peak area of the internal standard tetracosane and dry weight of each sample (for intracellular data only; approx. 0.5–1 mg of UV-B exposed cells and 1.5–3 mg of PAR only cells). All duplicate retention times were removed before further analysis.

#### 5.4.5. Identification

Identification of peaks was carried out in AMDIS by analysing each chromatogram using the Golm databases as a target library, followed by searching the NIST 05 library with a match factor of 60% or above. Reports were exported in .xls format from AMDIS with the first hit only included for further processing using Excel 2010 (Microsoft, Redmond, WA, USA). A true hit was considered when two or more biological replicates (within the same time-point) contained the peak. If none of the time points contained 'true hits,' the peaks were removed before further analysis. Metabolites reported belonged to level 2 (putatively annotated compounds) and level 4 (unknown compounds) identifications in accordance with the Metabolomics Standards Initiative [48].

#### 5.4.6. Statistical Analysis

MetaboAnalyst was used to statistically analyse the IS peak lists (in .csv format) of the time series data using the time-series/two-factor module. Multivariate analysis was carried out using each column as a different time point and each row representing a metabolite (data type = peak intensity table; study design = time-series only; data format = samples in columns) [46]. Missing data points were uploaded as blanks and replaced with half of the minimum integrated signal within each data set. Peaks were normalised to total sum of peaks, log-transformed and mean centered prior to statistical analysis. PCA, a one-way repeated ANOVA (*p* ≤ 0.05) and hierarchical heat map clustering was used to evaluate the data. A two-sample T-test with equal variance was also used as a univariate statistical tool to evaluate data comparing 0 h with each treatment time point (2, 6, 12, 24 and 48 h) as well as between each time points in Excel.

#### *5.5. MAA analysis*

#### 5.5.1. Sample Preparation

0.5–1 mg of UV-B exposed dried biomass was re-suspended in 100% HPLC grade methanol (1 mL) and left in the dark at 4 ◦C overnight (24 h). After centrifugation (5 min at 12000 rpm), the supernatant was removed and evaporated to dryness using a rotary vacuum concentrator. The dried extract was re-dissolved in 600 μL of deionised water and transferred to autosample vials for HPLC analysis [49].

#### 5.5.2. HPLC Analysis

HPLC analysis was performed using an Agilent 1100 system equipped with a binary pump (G1312A), an autosampler injector (ALS, G1313A), thermostatted column compartment (G1316A) and diode array detector (DAD, G1315A) connected via an interface module to a computer running ChemStation software. The stationary phase was an AlltimaTM AltechTM C18, 4.6 <sup>×</sup> 150 mm, 5 <sup>μ</sup>m column heated to 35 ◦C. The mobile phases consisted of; Eluent A: Water (0.01% TFA, *v*/*v*) and Eluent B: 70% methanol (0.054% TFA, *v*/*v*) with a gradient of; 99% A for 10 min, to 80% A over 5 min, to 1% A over 5 min, held for 3 min and increased to 99% A over 2 min. The samples were injected at a volume of 100 μL and MAA's were monitored at wavelengths of; 310, 320 and 330 nm, absorption spectra between 200–400 nm were stored in each detected peak.

#### *5.6. Pigment Analysis*

#### Sample Preparation

To approx. 0.5–1 mg of dried biomass, 100% HPLC grade methanol (1 mL) was added and vortexed to re-suspend. Samples were sonicated under low light conditions using a sonicator probe for 6 cycles of 20 s pulses at 40 Hz at 0 ◦C. After centrifugation (5 min at 12,000 rpm), the supernatant was removed and absorbance spectra measured using a UV-visible spectrophotometer between 400–800 nm with 100% methanol as a blank. Carotenoid concentration was calculated using the equations as described in [50,51].

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2218-1989/9/4/74/s1, Figure S1: Hierarchical heatmap visualisation of the significant (A) intracellular and (B) extracellular peak intensities (*p* ≤ 0.05) during UV-B exposure using a one-way repeated measure ANOVA in MetaboAnalyst. Data is arranged in triplicate with increasing length of UV-B exposure from left to right (0–48 h). S1 = replicate 1, S2 = replicate 2, S3 = replicate 3; Figure S2: Dry weight measurements of *C. fritschii* cultures during 48 h of supplemented UV-B exposure. All values are the mean of three biological replicates ± standard error; Figure S3: Time-series metabolomics data of *C. fritschii* during 48 h of UV-B exposure showing primary metabolites found in both intra- and extracellular during GC–MS analysis. (A) Metabolites showing statistically significant (*p* ≤ 0.05) changes over time in intra- and extracellular data; (B) metabolites showing statistically significant (*p* ≤ 0.05) changes over time in intracellular samples only. Statistical significance was measured using a two-sample T-test comparing control (0 h) to each treatment time point (2, 6, 12, 24 and 48 h) and between treatment time points, \* = 0.05 ≥ *p* ≥ 0.01, \*\* = 0.01≥ *p* ≥ 0.001 and \*\*\* = *p* ≤ 0.001; Table S1: Intracellular compounds detected during UV-B exposure

using GC–MS including; Retention time (RT), name, class, model (*m*/*z*), normalised mean abundance (*n* = 3) ± standard error (SE). T-test and ANOVA results showing statistical significant compounds, \* = 0.05 ≥ *p* ≥ 0.01, \*\* = 0.01 ≥ *p* ≥ 0.001, \*\*\* = *p* ≤ 0.001; Table S2: Extracellular compounds detected during UV-B exposure using GC–MS including; Retention time (RT), name, class, model (*m*/*z*), normalised mean abundance (*n* = 3) ± standard error (SE). T-test and ANOVA results showing statistical significant compounds, \* = 0.05 ≥ *p* ≥0.01, \*\* = 0.01 ≥ *p* ≥ 0.001, \*\*\* = *p* ≤ 0.001; Table S3: Biologically relevant metabolites detected during PAR + UV-B (intra- and extracellular metabolites) and PAR only (intracellular metabolites), including possible biosynthetic pathways; Table S4: Intracellular compounds detected during PAR only exposure using GC–MS including; Retention time (RT), name, class, model (*m*/*z*), normalised mean abundance (*n* = 3) ± standard error (SE). T-test and ANOVA results showing statistical significant compounds, \* = 0.05 ≥ *p* ≥ 0.01, \*\* = 0.01 ≥ *p* ≥ 0.001, \*\*\* = *p* ≤ 0.001.

**Author Contributions:** Authors in this study contributed in the following areas: B.K. was responsible for the design of the experiment, execution of experimental work and data analysis. E.D. supported experimental and data acquisition. B.K. wrote the manuscript with input from C.A.L., E.D. and S.W. Supervision by C.A.L., E.D. and S.W.

**Funding:** This research was funded by the Biotechnology and Biological Sciences Research Council (BBSRC iCASE studentship), UK, grant number BB/N503630/1.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **E**ff**ects of Copper and pH on the Growth and Physiology of** *Desmodesmus* **sp. AARLG074**

#### **Nattaphorn Buayam 1,2, Matthew P. Davey 3,\*, Alison G. Smith <sup>3</sup> and Chayakorn Pumas 2,\***


Received: 3 April 2019; Accepted: 27 April 2019; Published: 30 April 2019

**Abstract:** Copper (Cu) is a heavy metal that is widely used in industry and as such wastewater from mining or industrial operations can contain high levels of Cu. Some aquatic algal species can tolerate and bioaccumulate Cu and so could play a key role in bioremediating and recovering Cu from polluted waterways. One such species is the green alga *Desmodesmus* sp. AARLG074. The aim of this study was to determine how *Desmodesmus* is able to tolerate large alterations in its external Cu and pH environment. Specifically, we set out to measure the variations in the Cu removal efficiency, growth, ultrastructure, and cellular metabolite content in the algal cells that are associated with Cu exposure and acidity. The results showed that *Desmodesmus* could remove up to 80% of the copper presented in Jaworski's medium after 30 min exposure. There was a decrease in the ability of Cu removal at pH 4 compared to pH 6 indicating both pH and Cu concentration affected the efficiency of Cu removal. Furthermore, Cu had an adverse effect on algal growth and caused ultrastructural changes. Metabolite fingerprinting (FT-IR and GC-MS) revealed that the polysaccharide and amino acid content were the main metabolites affected under acid and Cu exposure. Fructose, lactose and sorbose contents significantly decreased under both acidic and Cu conditions, whilst glycerol and melezitose contents significantly increased at pH 4. The pathway analysis showed that pH had the highest impact score on alanine, aspartate and glutamate metabolism whereas Cu had the highest impact on arginine and proline metabolism. Notably both Cu and pH had impact on glutathione and galactose metabolism.

**Keywords:** algae; copper; FT-IR; metabolite fingerprinting; pathway analysis; TEM

#### **1. Introduction**

Copper (Cu) is widely used in industry [1] and as a result there is a very high demand of raw Cu extracted from mines. This demand is projected to increase 2.8 to 3.5 fold by 2050 [2]. Consequently, Cu containing waste (in particular wastewater) is generated daily from mining and industrial operations. The concentration of Cu in such mine drainage and water systems can vary substantially [3]. Additionally, the pH of the water systems is lowered (pH 3–5) resulting in acid mine drainage (AMD) [4,5]. A range of studies [3,6,7] have determined the level of Cu directly from industrial and acid mine run-off is around 10 to 80 times higher (up to 160 mg/L) than control level, indicating an urgent need for remediation. For example, the run-off from an abandoned Cu mine in Norway has a Cu concentration of 13.9 mg/L (pH 2.9), whereas the nearby receiving lake has just 0.04 mg Cu/L (pH 6.6) [3].

High levels of Cu exposure can be toxic to most living organisms [8–11] as heavy metals are non-biodegradable and environmentally persistent, which may be deposited on surfaces and then absorbed into the tissues of organisms [12]. Although the toxicity of Cu is recognized, the permissible amount of Cu in effluents slightly differs around the world. For example, the Ministry of Industry of Thailand has announced that the maximum permissible limit of Cu in industrial effluent is 2.0 mg Cu/L, which is the same as the permissible limit of Cu ions in drinking water permitted by the World Health Organization (WHO). The United States Environmental Protection Agency (USEPA) regulations state that Cu in industrial effluents must not exceed 1.3 mg Cu/L [1].

Despite the toxicity, some micro-algal species are able to grow in waterways that contain high concentrations of Cu, alongside the associated altered pH. This is in part due to Cu being an essential trace element that is required for normal algal growth [13,14] and also due to some algae (both macroand micro-algae) being highly efficient in bioaccumulating heavy metals [15]. As such, bioaccumulation of the Cu from these polluted waterways is a potential approach for heavy metal bioremediation. Although the efficiency of heavy metal removal by algae and its effect on algal growth has been well characterized for some species [8,10,16–19] the biochemical mechanisms that are associated with algae tolerating high concentrations of Cu tolerance are not clearly defined or understood. In addition, *Desmodesmus* (Chlorophyta, Chlorophyceae) is the genera that frequently found distributed in the freshwater resources over the north and north-eastern of Thailand [20].

The aim of this study was to determine how microalgae are able to tolerate large alterations in their external Cu and pH environment. To this end we studied a green microalga *Desmodesmus* sp. (Chlorophyta, Chlorophyceae) AARLG074, a species isolated from and commonly found in a natural water reservoir in the north of Thailand. Specifically, we set out to measure the variations in the Cu removal efficiency, growth, ultrastructure, and cellular metabolite content in algal cells that are associated with Cu exposure and acidity. The Cu concentrations used in this study were set to the permittable Cu concentration in wastewater of Thailand (2 mg Cu/L) and Cu concentrations in the acid mine drainage system and industrial effluents (14–164 mg Cu/L) [3,6]. The pH values were also set according to the pH of AMD [4,5].

#### **2. Results**

#### *2.1. Copper Removal E*ffi*ciency*

The Cu removal efficiency of the cultures was assessed by growing *Desmodesmus* sp. AARLG074 in JM medium supplemented with 0, 2, 20, and 50 mg Cu/L at pH4 and pH6 for 168 h (7 days). The Cu removal efficiency was highest at pH6 where up to 83% of Cu in the media was absorbed within 30 min after inoculation (Table 1). The highest Cu absorption (93.5%) was measured 168 h after exposure to Cu in medium supplemented with 2 mg Cu/L, pH 6. Even in medium supplemented with 20 mg Cu/L the absorption after 24 h exposure at pH 6 was 92.8%.

In contrast, when algae were grown at pH 4 only 16.5% of Cu in the medium was removed at a Cu concentration of 20 mg Cu/L. The Cu absorption fluctuated over seven days with approximately 4–40% absorption at pH 4 and 64–94% at pH 6. Nonetheless, statistically significant lower Cu removal efficiency values were observed in all time points at pH 4, compared to pH 6 (Table 1).


**Table 1.** Percent removal of copper from growth media containing *Desmodesmus* sp. AARLG074 over 168 h (7 days) growth at pH4 and pH6. Each value represents the mean ± SD (*n* = 3) with different superscript letter in the same time point indicating statistically significant differences (Two-way ANOVA, Tukey HSD, *p* < 0.05) using R version 3.4.3.

*2.2. Copper and Acid Exposure A*ff*ected the Cell Density, Pigment Content and Ultrastructure in Desmodesmus sp. AARLG074*

Cell and colony density—The cell density was calculated based on the number of colonies within four categories, based on number of cells in each colony (single, duplet, triplet, quadruplet) (Figure 1). The cell density of algae grown under control conditions (no Cu, pH 6) gradually increased and reached the highest cell density after 120 h cultivation (Figure 1A). However, the cell density of Cu-treated algae cultures at pH 6 was indicated a longer stationary phase (up to 72 h) with the highest cell densities occurring after 120 h cultivation. The cell density of algae grown in non-Cu supplemented JM at pH 4, which reflects that effects of acidic stress (pH 4) on algal growth, was significantly less than the cell densities of algae grown pH 6 (Figure 1A) (Supplementary Materials Table S1). As measured in cultures grown at pH 6, the cell density of cultures grown in combined Cu and pH 4 media also had a statistically-significantly negative effect on the cell density of *Desmodesmus* sp. AARLG074, when compared to no Cu media controls (Figure 1A).

**Figure 1.** The effects of copper and pH on cell density and the percentage of number of cells per colony of *Desmodesmus* sp. AARLG074. (**A**) The mean cell density of *Desmodesmus* sp. AARLG074 grown at pH 4 (red line) and pH 6 (blue line) and copper concentrations (0, 2, 20, 50 mg/L) over 168 h. (**B**) The mean percentage of single (blue), duplet (pink), triplet (purple) and quadruplet (green) cells per colony of *Desmodesmus* sp. AARLG074 growing at pH 4 and pH 6 and copper concentration over 168 h. Values are mean ± SD (*n* = 3).

The majority of colonies over the testing period were duplet (Figure 1B). The percentage of duplet colonies increased over 16 h cultivation under both control (0 mg Cu/L) and low Cu (2 mg Cu/L) conditions at pH 4 or pH 6. Additionally, the percentage of triplet colonies (three-cells-colony) significantly increased when exposed to high copper (50 mg Cu/L) for 168 h. The percentage of quadruplet colonies (four-cells-colony) also increased after 24 h cultivation, but this effect was negated when Cu was added to the media (Figure 1B and Supplementary Materials Tables S2–S5).

Pigments—The pigment content (chlorophyll *a*, chlorophyll *b* and total carotenoids) of the cultures significantly decreased in the algal cultures containing Cu (Figure 2). This effect was more severe at pH 4 compared to pH 6 (Figure 2A and Supplementary Materials Tables S7–S9). The amount of pigments in the control (0 mg Cu/L) and low Cu (2 mg Cu/L) supplemented cultures gradually increased over time. This was not observed under high Cu supplemented conditions (20 and 50 mg Cu/L), where pigment concentrations were statistically lower than cultures grown at 0 and 2 mg Cu/L.

**Figure 2.** Pigment content (chlorophyll *a* (pink), chlorophyll *b* (green) and total carotenoids (blue)) of *Desmodesmus* sp. AARLG074 grown at pH 4 and pH 6 with varied copper concentrations (0, 2, 20, 50 mg/L) over 168 h. Data expressed as (**A**) pigments per volume of culture (ug per ml) and (**B**) per algal cell (ng per cell). Values are mean ± SD (*n* = 3).

This lower pigment content in the culture would partly be due to the lower cell count per ml of culture, but also a lower pigment concentration per cell as there was a similar response to Cu and pH when the pigment values were expressed as amount of pigments per cell (Figure 2B and Supplementary Materials Tables S7–S9). The amount of pigments in un-supplemented Cu cultures grown in either pH 4 or 6 gradually increased over time (Figure 2B). Also, under low Cu supplemented conditions (2 mg Cu/L) the pigment content per cell increased until 16 h and remained stable until the end of experiment, in both pH 4 and 6. However, the amount of chlorophyll *a* in cultures grown at 50 mg Cu/L was significantly lower (Supplementary Materials Table S9). The amount of chlorophyll *a* in pH 4 grown cultures was lower than pH 6 but not statistically different.

Ultrastructure—The ultrastructure of *Desmodesmus* sp. AARLG074 was altered after 24 h of Cu exposure as observed using TEM (Figure 3). Control algal cells that were grown in non-copper supplemented JM media at pH 4 had few starch granules with little or no periplasmic space (PM) (Figure 3A). However, control algal cells grown in non-copper supplemented JM media at pH 6 had many starch granules (S), again with little or no periplasmic space (PM) (Figure 3B).

In Cu supplemented media, the largest change in ultrastructure was observed in *Desmodesmus* sp. AARLG074 exposed to 50 mg Cu/L in both pH 4 (Figure 3C) and 6 (Figure 3D). The body of the cell had retracted from the cell wall resulting in an increased periplasmic space (PM) indicating that plasmolysis had occurred (Figure 3C,D). In addition, fewer starch grains (S) and membrane whorls (MW) were observed in the 50 mg Cu/L-treated cells at pH 6 (Figure 3D). The thylakoidal space was also more apparent under the high Cu and pH 4 (Figure 3C) conditions, compared to those grown under the high Cu and pH 6 conditions (Figure 3D).

**Figure 3.** Electron micrographs of *Desmodesmus* sp. AARLG074 after 24 h growth in JM media at pH 4 and pH 6 at various Cu concentrations. The ultracellular structure of *Desmodesmus* sp. AARLG074 cultivated in the 0 mg Cu/L media at pH 4 (**A**) and pH 6 (**B**). Numerous starch granules (S) were observed. The ultrastructure of *Desmodesmus* sp. AARLG074 that grew in 50 mg Cu/L supplemented media at pH 4 (**C**) and pH 6 (**D**). Abbreviations: P, pyrenoids; PM, periplasmic space; S, starch granule; TH, thylakoids; V, vacuole; VD, vacuole deposit.

#### *2.3. The E*ff*ects of Copper and pH on Metabolite Composition of Desmodesmus sp. AARLG074*

Fourier transform-infrared spectrometry (FT-IR) was used to study the effects of different Cu concentrations and pH conditions on the metabolite composition of *Desmodesmus* sp. AARLG074 over 168 h of cultivation. The results, based on the principal component analysis (PCA) (Figure 4), showed that the metabolite fingerprints of *Desmodesmus* sp. AARLG074 cultivated in high Cu (20 and 50 mg Cu/L) at pH 6 were different from the metabolite fingerprints of algae that were cultivated in any other conditions. The specific wavenumber regions that were strongly associated with high copper concentration treated *Desmodesmus* sp. AARLG074 at pH 6 only was in the range of 835–1090 cm−<sup>1</sup> (Figure 5B), which was associated with wavenumbers relating to polysaccharides.

**Figure 4.** Score scatter plot from Principal Component Analysis (PCA) plot of FT-IR scans of *Desmodesmus* sp. AARLG074 cultured in JM under various copper concentrations (0, 2, 20 and 50 mg Cu/L) at pH 4 and pH 6. Each dot represents one biological replicate. The dot color represents the cultivation pH condition-pH 4 (green) and pH 6 (blue) and the labelled number beside dots represent the copper concentration as mg/L in the JM. The results shown here were analyzed using SIMCA-P (Version 15.0.2).

**Figure 5.** Score contribution plot values from PCA (PC1 and PC2 loading) from Fourier transform-infrared spectrometry (FT-IR) scans of *Desmodesmus* sp. AARLG074. FT-IR wavenumber score values are negative if they contribute towards PCA loadings associated with (**A**) growing in media at pH 4 (green) and positive for pH 6 (blue) and (**B**) negative if they contribute towards PCA loadings associated with no copper (green) and positive for copper supplemented media (red).

The detailed metabolite profiling by GC-MS did not reveal any major clustering of samples based on the pH and Cu concentration growth conditions in any of the principal components (Figure 6). Even though the PCA of the GC-MS data did not cluster by treatment, the score contribution plots of 40 key metabolites between absent Cu and Cu-supplemented treatment and between pH 4 and pH6 (Figure 7) were calculated to indicate the main metabolite differences between culture conditions. The score contribution of the full list of metabolites associated with the different Cu or pH are shown in Supplementary Materials Figures S1 and S2, respectively. The key metabolites based on

the score contribution plots of both pH and Cu were polysaccharides, which corresponded to FT-IR results above (Figure 7). The other metabolites listed were lipids, amino and organic acids and vitamins. In pH 4 grown cells, the dominant metabolites were melezitose, galactose, hexadecane and mannose whereas in cells grown at pH 6 the dominant metabolites were phosphoric acid and cystathionine (Figure 7). The dominant metabolites in algae grown under high Cu supplemented media were sorbose, octacosane, hexadecanol, cystathionine and xylofuranose. The key metabolites under control conditions (zero Cu) were threose, lactose, galactose, maltose, ribose mannitol and lipids (e.g., decane and digalactosylglycerol). The statistical analysis of the detected metabolites showed that there were five statistically different metabolites among the treatments which were fructose, glycerol, lactose, melezitose and sorbose (Supplementary Materials Figure S3). Four of five statistically different metabolites among the treatments belonged to polysaccharides which supported the FT-IR results.

**Figure 6.** Metabolic profiling of the *Desmodesmus* sp. AARLG074 cultivated under different copper and pH conditions. The score scatter plot from PCA plot of GC-MS results of *Desmodesmus* sp. AARLG074 that were cultured in JM under various copper concentration (0, 2, 20 and 50 mg/L) at pH 4 and pH 6. Each dot represents one biological replicate. The colour of dot represents the cultivation pH condition-pH 4 (green) and pH 6 (blue) and the labelled number beside dots represent the copper concentration as mg/L in the JM.

The impact of Cu and pH on the metabolic pathways (based on the number of metabolites per pathway being identified and changing in intensity using the GC-MS data) were analysed using the MetaboAnalyst Pathway-Analysis function [21]. The results which had *P* values less than 0.05 are shown in Table 2. The pathway analysis of different pH condition revealed that pH had impacted on multiple amino acid, sugar and lipid metabolic pathways. The greatest impact of pH was shown on alanine, aspartate and glutamate metabolism. Additionally, Cu exposure had also affected various protein metabolic pathways with the greatest impact on arginine and proline metabolism. Moreover, the result indicated that both pH and Cu had an effect on glutathione metabolism and galactose metabolism. Although the impact levels were low there were altered metabolic pathways related to photosynthesis- pigment synthesis and carbon fixation in photosynthesis. The results showed that seven amino acid or protein synthesis pathways had been affected by pH. However, only four amino acid synthesis pathways (arginine and proline metabolism, lysine biosynthesis, beta-alanine metabolism and butanoate metabolism) had been affected by Cu (Table 2).

**Figure 7.** The score contribution plot values (top 40) that contribute towards GC-MS PCA loading plots of *Desmodesmus* sp. AARLG074 under different treatments (**A**) pH and (**B**) copper. The score contribution plot values (top 40) were ranked in order of importance and are negative if they contribute towards PCA loading plot for the pH 4 (green) and positive if they contribute towards the pH 6 (blue) and positive for copper stress (red) and negative if they contribute towards control (without copper) (green). The full list of the metabolites is presented in supplementary (Figures S1 and S2).



#### **3. Discussion**

#### *3.1. Tolerance of High Cu Exposure and Acidity in Desmodesmus sp. AARLG074—Growth and Structural Alterations Associated with Cu Removal*

The aim of this study was to determine how *Desmodesmus* are able to physiologically tolerate large alterations in their external Cu and pH environment. *Desmodesmus* was indeed able to grow, although less so, in media supplemented with Cu. The Cu in the growth media was removed in the algae cultures with up to 83% of the copper present in the medium within 30 min after exposure, after which the absorption stabilizing over 168 h. We did not measure the Cu content in the cells but previous studies have shown that this biphasic absorption process may be due to a rapid non-metabolic dependent adsorption followed by a slow metabolic dependent uptake process [9,15]. The early absorption (<30 min) could be due to adsorption of Cu to the outer cell components (e.g., polysaccharides, mucilage and cell walls). The slower fluctuation in Cu removal efficiency (Table 1) would then be uptake into the living algal cell. In addition, the reduction in Cu removal efficiency at 20 mg Cu/L, pH 6 after 168 h exposure might be due to cell death and lysis, releasing the Cu back to the medium, or the living algae by export Cu out to the environment [15]. Moreover, the results also indicate that the pH is a critical factor that influences copper removal. The maximum removal, 93%, was when the algae were grown in JM media containing 2 mg Cu/L at pH 6. The efficiency of Cu absorption at pH 4 was significantly lower than pH 6, which was similar to several previous studies showing that heavy metal biosorption is a pH dependent process [22–26].

Despite the ability to tolerate Cu in the growth media, the exposure to Cu had an adverse effect on growth and cellular pigment content. Such effects have been previously observed and reported in other marine and freshwater algal species [8,10,16–19]. Although overall cell growth was reduced in cultures containing Cu we also found that the percentage of cell type was changed. There was a higher proportion of triplet cells in cultures that were cultivated in high Cu media at both pH 4 and pH 6 (Figure 1B). We can speculate that this might be one of the mechanisms that *Desmodesmus* sp. AARLG074 use for acclimating to live in a copper containing environment as multiple cell colonies have reduced cell surface area being exposed to the surrounding copper. Furthermore, the pigment analysis showed that chlorophyll *a*, *b* and total carotenoids content in algal cultures (Figure 2A) or per cell (Figure 2B), was lowered when cultivated in Cu supplemented media, this effect was much more severe in pH 4, a phenomena similar to previous studies [18,27–29]. Several studies have shown that the pigment level was reduced when algae were exposed to heavy metals [18,27–29], but none of them explain the reason of this phenomena. According to our TEM, ultrastructural changes of the algal cells occurred when exposed to Cu and acidity (Figure 3) notably at the high Cu concentration (50 mg Cu/L). One of the most drastic changes was the increase of intra-thylakoid space and disorganization of membrane which was similar to previous studies [19,24,29–32]. The alteration of the chloroplast ultrastructure was correlated with a lower chlorophyll content value (Figure 2). In addition, increasing the thylakoid space might be due to an excess of Cu accumulated inside due to Cu transporters locate on the pyrenoid and thylakoid membranes [33–35].

Furthermore, the Cu treated algae had fewer starch granules compared to the control, which might be due to an imbalance of the energy generated and used under Cu exposure as a consequence of chloroplast disorganization and fewer pigments [19,36]. Studies on another species of algae exposed to Cu, Desmidium swartzii, also had electron-dense inclusions in chloroplast and starch grains, of which Cu were detected in those compartments [37]. The presence of membrane whorls (MW) has previously been observed in heavy metal treated algae, where metals were detected inside the membrane whorl and as such has been proposed as a possible metal detoxification mechanism [38]

#### *3.2. Metabolic Alterations Associated with Exposure to Cu and Acidity in Desmodesmus sp. AARLG074*

The FT-IR results showed that different Cu and pH conditions affected a similar class of metabolites within the cells (Figures 4 and 5). The wavenumbers that differentiated treatments were 835–1090 cm−<sup>1</sup> (Figure 5A,B) which coincide with polysaccharide metabolites [39,40]. However, the GC-MS metabolite profiling showed that most of the key metabolites differed under the different pH or Cu conditions were sugars and amino acids (Figure 7). The PCA plots of both FT-IR metabolites fingerprinting and GC-MS metabolites profiling did indicate that there were relatively small differences in the metabolome of *Desmodesmus* sp. AARLG074 under various pH and copper treatments. This corresponds to a previous report on the effect of Cu on the transcriptome and metabolome of *Ectocarpus siliculosus* (Ochrophyta, Phaeophyceae) [41].

Five metabolites (glycerol, melezitose, lactose, fructose and sorbose) were highlighted as statistically different between treatments, which four of five are sugars (Figure S3) indicating some agreement between the FT-IR and GC-MS findings. From this list, the pH might be one of the crucial factors that affected the level of glycerol as there was considerably more glycerol in cells grown under pH 4 whereas there was no change under pH 6. The high level of glycerol at pH 4 might be explained for several reasons, largely osmoregulatory and membrane composition. Accumulation of intracellular glycerol in algae function as an osmoregulator which lowers internal osmotic potential and so prevent excessive water loss during high Cu exposure [42]. Additionally, released glycerol may also serve as a sink for products of photosynthesis [42]. Glycerol is also part of plasma membrane component and high glycerol induce membrane stiffening [43,44] and changes in membrane composition will affect the membrane permeability. The plasma membrane of algae living in acidic environments is fairly impermeable for protons, requiring relatively little energy for active pumping against a passive proton influx [45]. Thus, the higher level of glycerol detected in pH 4 might be one of the mechanisms that this green microalgae *Desmodesmus* sp. AARLG074 used to cope with the acidic and/or Cu stress. Moreover, melezitose (syn: melicitose), is an allelopathic chemical in marine planktonic microalgae [46]. This chemical is also found to be upregulated under osmotic stress in higher plants [47,48]. However, to our knowledge there is no previous report on an association between melezitose and Cu or pH in algae.

Beyond the top five metabolites mentioned above, one of metabolites that was highlighted in the score contribution plots as being important in cells grown in both pH 6 and Cu conditions was cystathionine (Figure 7). The cystathionine is an important intermediate compound in the biosynthesis of cysteine which is linked to glutathione metabolism [49]. The addition of glutamate to cysteine leads to form glutamyl cysteine, which the addition of glycine to this glutamyl cysteine leads to form glutathione [49]. This corresponded to our pathway analysis that showed Cu had impacted on glutathione metabolism (Table 2). Importantly, glutathione have been reported as an important intracellular ligands involved in metal sequestration and detoxification in algae [50].

Although the top five metabolites that were statistically different between the treatments were sugars the pathways that had the highest impact score for different pH and Cu were alanine, aspartate and glutamate metabolism (impact = 0.54) and arginine and proline metabolism (impact = 0.27) (Table 2). The pH treatment had the biggest impact on alanine, aspartate and glutamate metabolism, whereas Cu had the biggest impact on arginine and proline metabolism. The impact scores are a combination of the number of metabolites detected in the pathway and the fold change between treatments of those metabolites. This implies that although the metabolites may not statistically different between treatments, the impact score may still be on when taking all the metabolites in that pathway into consideration. Care should be taken when interpreting the in silico pathway analysis scores as the pathways were not measured directly, only the individual metabolites that could be in one or more metabolic pathway. However, such pathway analyses do provide us with an initial snapshot of which pathways were the most likely to be changing, which provides useful information for further targeted studies. There are similarities with the metabolites and pathways highlighted in the analysis in other organisms, especially bacteria [51–55]. Glutamate has been reported to be involved in acid tolerance in *Streptococcus mutans* [52], in both acid and Cu tolerance of *Escherichia coli* [53] and the accumulation of potassium glutamate in *E*. *coli* has been shown to occur immediately after osmotic shock to provide temporary protection to the cells [53,55]. In addition, aspartic acid and glutamic acid are also reported to be involved in acid tolerance in *Acetobacter pasteurianus* [51]. Proline is well

known as playing an important role in plants and microorganism osmoprotection, metal chelation and general antioxidative defense molecule and signaling molecule [52,56,57]. Enhancement and accumulation of proline in heavy metals and acid exposed algae and bacteria has also been previously reported [48,57–59].

To conclude, we have shown that cultures of are able to tolerate and grow on media in both high Cu and acidic conditions (pH 4) and indeed removes over 80% Cu from the media within 30 min. The Cu concentration did affect Cu removal percentages. We showed that at the highest Cu removal efficiency and cell density, approximately 0.002 ng Cu could be removed per cell. This information shows that *Desmodesmus* sp. AARLG074 has the potential to industrially recover Cu from the algae that would otherwise be lost. However, despite its tolerance, the higher concentrations of Cu did negatively affect its growth rate. Additionally, we have shown that exposure and tolerance to Cu and acid conditions was associated with changes in its morphology, ultrastructure and overall metabolite composition of the cells, with polysaccharide and amino acid biochemistry changing the most.

#### **4. Materials and Methods**

#### *4.1. Organism and Culture Conditions*

The unicellular green microalga *Desmosdesmus* sp. AARLG074 was obtained from the Applied Algal Research Laboratory (AARL), Department of Biology, Faculty of Science, Chiang Mai University. This strain was isolated from the northern Thailand natural water reservoir [60]. The algal feed stock was maintained in Jaworski's medium (JM), pH 7.0 under continuous white light emitting diode (LED) illumination (75 μmol photon/m2/s) at 25 ◦C and shaken by hand twice daily. The optical density 650 nm and cell density was measured until it reached the log phase after which the cultures were used as stock for further experiments. To study the effects of pH and Cu on Cu removal efficiency, algal growth, ultrastructure and metabolites the algae were inoculated into 1l flasks containing 800 mL Jaworski's medium (JM) supplemented with 0, 2, 20 and 50 mg Cu/L by using CuSO4·5H2O as a Cu source. The pH of medium was adjusted to pH 4 and pH 6 using hydrochloric acid (HCl). Each treatment was conducted in triplicate biological samples (*n* = 3), in total 24 flasks were used. Atomic absorption spectroscopy (AAS) (see below method) confirmed the initial copper concentration in the treatment of 0, 2, 20, and 50 mg Cu/L was 0, 1.97 ± 0.11, 20.59 ± 0.55, and 50.66 ± 1.03 mg/L respectively.

#### *4.2. Copper Absorption E*ffi*ciency*

To determine the copper absorption efficiency, 10 mL of culture from three biological replicate samples (*n* = 3) was collected using an autopipette at various time points from 0 to 168 h (0, 0.5, 8, 16, 24, 72, 120 and 168 h) and then centrifuged at 3000 rpm for 5 min. The Cu concentration in the supernatant was measured using flame atomic absorption spectroscopy (AAS) as described in Mota et al. (2015) [61] and the removal efficiency (%) was calculated using the following equation:

$$\text{Removal efficiency} \left(\% \right) = \left(100 \text{\%} \right) \times \left(\text{Ci} - \text{Cf} \right) \tag{1}$$

where: Ci is the initial copper concentration (mg/L), Cf is the final (residual) copper concentration (mg/L).

#### *4.3. The E*ff*ects of Copper and pH on Cell Density, Pigments and Ultrastructure*

The three biological replicates of algae that were cultured in different conditions were collected at the same time point as described above. *Desmodesmus* sp. are colony forming. As 1, 2 and 4 cell colonies were observed (3 cell colonies were rarely found) the cell density (cells per ml of culture) was counted into four categories based on the number of cells per colony—single, duplet, triplet and quadruplet using a hematocytometer. The percentage of each type of colony found in the cultures was calculated to investigate the effect of pH and Cu on the change of percentage of each colony presented in algal community over time. Chlorophylls and carotenoids were extracted using 90% methanol as modified method from Saijo (1975) [62] and Lichtenthaler and Buschmann (2001) [63]. In addition, the amount of pigment per cell was calculated based on the amount of pigment in culture (μg/mL) and cell density (ng/cell). Transmission electron microscopy (TEM) was used to investigate the ultrastructural change of *Desmodesmus* sp. AARLG074 after 24 h exposed to Cu and acidic stress [38].

#### *4.4. Metabolite Analysis*

Ten ml of samples were collected at 0, 0.5, 8, 16, 24, 72, 120 and 168 h into the treatment phase and centrifuged at 3000 rpm at 4 ◦C for 5 min. The supernatant was discarded and the pellets were freeze-dried (freezone labconco, USA) and transferred to the Department of Plant Sciences, University of Cambridge, UK. Metabolite fingerprints of freeze-dried samples were obtained using a Perkin-Elmer Spectrum Two FT-IR, within the wavenumbers of 600–4500 cm−1. The spectra were normalized against air.

The metabolite profiling of freeze-dried samples was further investigated in more detail using GC-MS by extracting soluble polar and non-polar metabolites in methanol-chloroform-water method as described in Davey et al. (2005) [64]. The compounds within the polar methanol-water phase were derivatized by N-Methyl-N-trimethylsilyl-trifluoroacetamide (MSTFA) and Trimethylsilyl (TMS) as described by Dunn et al. (2011) [65]. Then, 1 μL (splitless) of the derivatized extracts were separated and profiled by GC-MS (GC-MS (Thermo Scientific Trace 1310 GC with 211 ISQ LT MS, Xcaliber v2.2) with a ZB-5MSi column (30 m, 0.25 mm ID, 0.25 μm film 212 thickness, Phenomenex, UK). GC-MS spectra were aligned to an internal standard (phenyl-β-d-glucopyranoside hydrate 98%) and processed using Thermo TraceFinder (v3.1) using genesis peak search method to aid identification based on molecular mass as previously described [64,66]. Pathway analysis of the identified metabolites were done both on SIMCA-P program (v15.0.2 Umetrics, Sweden) and MetaboAnalyst open source software (version 4.0, pathway analysis tool) using the *Arabidopsis thaliana* metabolic pathway library [21], www.metaboanalyst.ca). The results of pathway analysis that have *p* value ≤ 0.05 only were shown in the Table 2. R-script for the metaboanalyst software can be downloaded at https://github.com/xia-lab/MetaboAnalystR [21].

#### *4.5. Statistical Analyses*

To indicate the effect of initial Cu and pH on copper absorption efficiency, cell density and pigments of *Desmodesmus* sp. AARLG074 a two-way ANOVA (after a Lavene's normality distribution) with Tukey HSD test (*p* ≤ 0.05) was performed using car, multcompView and lsmeans library on R version 3.4.3. as previously described [67]. Multivariate analyses to test whether the effect of initial Cu and pH on the algae could be discriminated based on their identified and unidentified metabolites (from FT-IR fingerprints or GC-MS profiling datasets) were performed using Principal Component Analysis (PCA) [68] on Pareto Scaled data (FT-IR absorbance values or GC-MS identified peak area units) within the SIMCA-P v14.1 PCA analysis pipeline (Umetrics, Sweden) to produce standard score scatter plots and ranked score contribution plots of how each variable (FT-IR wavenumber or GC-MS metabolite) contributed to clustering within the PCA score scatter plot.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2218-1989/9/5/84/s1, Table S1: The effects of copper and pH over 7 days on cell density (cells/mL) of *Desmodesmus* sp. AARLG074, Table S2: The effects of copper and pH over 7 days on percentage of single colony (1-cell-colony) of *Desmodesmus* sp. AARLG074, Table S3: The effects of copper and pH over 7 days on percentage of duplet colony (two-cells-colony) of *Desmodesmus* sp. AARLG074, Table S4: The effects of copper and pH over 7 days on percentage of triplet colony (three-cells-colony) of *Desmodesmus* sp. AARLG074, Table S5: The effects of copper and pH over 7 days on percentage of quadruplet colony (four-cells-colony) of *Desmodesmus* sp. AARLG074, Table S6: The effects of copper and pH over 7 days on chlorophyll *a* (μg/mL) of *Desmodesmus* sp. AARLG074, Table S7: The effects of copper and pH over 7 days on chlorophyll *b* (μg/mL) of *Desmodesmus* sp. AARLG074, Table S8: The effects of copper and pH over 7 days on total carotenoids (μg/m) of *Desmodesmus* sp. AARLG074, Table S9: The effects of copper and pH over 7 days on chlorophyll *a* (10−<sup>3</sup> ng/cell) of *Desmodesmus* sp. AARLG074, Table S10: The effects of copper and pH over 7 days on chlorophyll *b* (10−<sup>3</sup> ng/cell) of *Desmodesmus* sp. AARLG074, Table S11: The effects of copper and pH over 7 days on total carotenoids (10−<sup>3</sup> ng/cell) of *Desmodesmus* sp. AARLG074, Figure S1: Score contribution plot values showing the full list which contribute towards GC-MS PCA loading plots of

*Desmodesmus* sp. AARLG074 under different pH condition, Figure S2: Score contribution plot values showing the full list which contribute towards GC-MS PCA loading plots of *Desmodesmus* sp. AARLG074 under different copper condition, Figure S3: Statistical analysis of GC-MS metabolite profiling of Desmodesmus sp. AARLG074 under copper and acidic stress.

**Author Contributions:** All authors have contributed to the conceptualization, investigation, data analysis and writing, reviewing and editing of this work.

**Funding:** We would like to thank the Development and Promotion of Science and Technology Talents Project (DPST Scholar), which is jointly supported by the Ministry of Science and Technology, the Ministry of Education, and the Institute for the Promotion of Teaching Science and Technology (IPST) who provided financial support to conduct this research in Thailand and aboard.

**Acknowledgments:** We truly appreciated the kindly help, support and advice of Jeeraporn Pekkoh. Also thank you all supportive and cheerful lab members of the Applied Algal Research Laboratory (AARL), Department of Biology, Faculty of Science, Chiang Mai University, Thailand and lab members of 220 Plant Metabolism Group, Department of Plants Sciences, University of Cambridge, United Kingdom. This research work was partially supported by Chiang Mai University.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
