Effects of reused water on the growth of E. gracilis
Since the UFM can cut off substances with a molecular weight ≥ 50 kDa, viruses, bacteria, macromolecular proteins, polysaccharides, and other substances can be filtered out [10, 12], so only unconsumed ions and DOM remain in the reused water. This study found that the DW of algae cells gradually decreased in the reused water with successive cycles. The DW of algae with each cycle of cultivation decreased by 13.1% (UFM-R1, p < 0.05), 28.6% (UFM-R2, p < 0.05), and 79.2% (UFM-R3, p < 0.01) compared to the control group on the last day of cultivation (Fig. 1). By the third cycle of cultivation, the growth of algae cells had been severely inhibited. This suggests that the increase in the presence of growth inhibitors with successive cycles of water reuse reduces algal growth beyond a tolerable range. This phenomenon is similar to the growth inhibition observed for other microalgae such as S. acuminatus [10], Chlorella. SDEC-18 [18], and N. oceanica [12]. We postulate that accumulated ions and algae cells secretions of DOM in the reused water are the main factors that inhibit the growth of E. gracilis.
The effect of accumulated ions on the growth of E. gracilis
Microalgae can selectively absorb some types of ions from inorganic nutrients and assimilate them into their own organic matter. But some ions such as Cl− and Na+, cannot be absorbed by microalgae. These residual ions, which accumulate in reused water, can destroy the balance of osmolality of the microalgae, thereby inhibiting their growth [8, 9]. This study found that when algae cells were cultivated in more than eightfold the concentration of PEM medium (NH4Cl as the nitrogen source) compared to the control, the relative cell density was lower, algal cytochromes were almost absent, and the algae cells were elongated (Fig. 2a, b). When the concentration of the culture medium was within fivefold of the control medium, the DW of algal cells was about 2.7 g/L, which was not significantly different from the control group (p > 0.05). However, when the concentrations of the medium were increased to above fivefold that of the control medium, the DW gradually decreased. In fact, when the concentration of the medium was increased by eight-, nine-, and tenfold, the DW of cells decreased by 81.3% (p < 0.01), 85.0% (p < 0.01), and 92.5% (p < 0.01), respectively (Fig. 2c). In addition, the Cl− concentration was increased as the medium concentration increases, the maximal concentration was reached 11,996.4 mg/L in the tenfold medium (Fig. 2d), suggesting that accumulated Cl− in the medium may be a key growth inhibitor of E. gracilis.
To prove the above hypothesis, we used urea instead of NH4Cl as the nitrogen source (equal nitrogen content) and found that as the concentration of urea increased, the DW of algae cells increased significantly. When the concentration of the culture medium was four to tenfold, the biomass of algal cells was stably maintained at about 2.7 g/L, and there was no significant difference among them (p > 0.05) (Fig. 2c). In addition, when urea was used as a nitrogen source, the cells appeared to be fuller. With an increase in the concentration of the culture medium, the relative cells density increased and the relative content of chlorophyll gradually increased as well (Fig. 2e, f). At the same time, the salinity and osmolality in the medium were much lower than those of the medium using NH4Cl as the nitrogen source. For example, the salinities at tenfold medium concentration were 29.4 psu versus 9.1 psu for NH4Cl versus urea, respectively. The osmolalities under these conditions were 727.0 mosm versus 167.0 mosm for NH4Cl versus urea, respectively. The salinity value for NH4Cl was 3.2-fold greater than urea (Fig. 2g, p < 0.01), and the osmolality for NH4Cl was 4.3-fold greater than urea (Fig. 2h, p < 0.01). These results show that the growth of algae cells was not inhibited with a fivefold increase in salinity (< 15.6 psu) and osmolality (< 361.1 mosm) of the medium, and we can confirm that during the UFM-R3 culture cycle, the growth of E. gracilis was not hindered by the accumulated ions in the reused water. This phenomenon has also been confirmed by the cultivation of S. acuminatus in reused water [10, 19]. Our work also showed that the growth of E. gracilis had a certain tolerance range to ions. If this tolerance range was exceeded, the growth of E. gracilis was inhibited. In addition, it was determined that the traditional medium PEM with NH4Cl as the nitrogen source was not suitable for the continuous recycling of cultivation water or for batch-fed cultivation of E. gracilis (such as heterotrophic batch-fed fermentation). When urea is used, it serves as an ideal nitrogen source because it reduces the salinity and osmolality in the culture medium.
Identification of growth inhibitors in E. gracilis secretions
The growth of microalgae is not affected by certain osmotic pressures for reused water, so growth inhibitors may exist in the DOM secreted by microalgae. However, some DOM can promote the growth of microalgae while some have an inhibitory effect on microalgae growth [3], so further study of these DOM characteristics is required. This study also found that E. gracilis continuously secreted DOM during the culture process. By the time UFM-R3 was reached, the DOM concentration was 189.21 mg/L, while the control group contained only 54.92 mg/L DOM, a 3.4-fold difference (p < 0.01) (Fig. 3a). This indicates that, at elevated concentrations, DOM may have an inhibitory effect on the growth of E. gracilis.
3D-FEEM fluorescence spectroscopy is fast and has excellent selectivity and sensitivity for fluorescent substances [20]. Therefore, in this study, this technique was used to identify the types of DOM secreted by algae cells. Chen et al. [21] used 3D-FEEM fluorescence spectroscopy to identify the following substances in the DOM present in cultivation water: aromatic proteins (AP), fulvic acid-like substances (FA), soluble microbial by product-like material (SMBM), and HA. We used those assignments to determine which types of DOM were present in our cultivation water samples (they are labeled with roman numerals in the spectrum in Fig. 3b. See caption b). It can be seen from Fig. 3b that the abundance of organic compounds with fluorescent signals in the reused water from high to low as: HA, SMBM, FA, and AP. Our spectra showed that HA was the potential main type of DOM present in E. gracilis secretions.
In order to further identify the growth inhibitors, we divided the DOM into six major fractions using fractional distillation (Fig. 3c). The percentages from high to low were: HiN (32%), HoA (27%), HoN (25%), HiB (7%), HoB (6%), and HiA (3%). From this result, we know that the DOM is mainly composed of HiN, HoA, and HoN, suggesting that HA, a potential inhibitor of E. gracilis, is composed of these organic acids. We also know that the slope of peaks in a UV spectrum (at 254 nm, given in AU/cm) for organic matter represents the content of organic functional groups that contribute to fluorescence, such as C=C bonds, C=O bonds, and aromatic rings. The importance of ultraviolet spectra for detecting pollutants in the water treatment process was described by Altmann et al. [22]. In this study, we tested the UV254 of 80-fold-concentrated DOM and found that the fluorescence intensity from high to low was: HoN (0.58 AU/cm), HoA (0.53 AU/cm), HiA (0.50 AU/cm), HiB (0.44 AU/cm), HoB (0.30 AU/cm), and HiN (0.24 AU/cm) (Fig. 3d). The inhibition of growth, IG%, of E. gracilis for each of these organic substances were: 28.8%, 25.0%, 24.6%, 19.2%, 12.1%, and 5.0% (Fig. 3e), respectively. It suggests that all of these DOM fractions can inhibit the growth of E. gracilis, especially HoN, HoA, and HiA. In addition, it is obvious that the UV254 absorption value is linearly related to the IG% for E. gracilis based on the graph in Fig. 3f for which R2 = 0.9. The degree of inhibition was positively correlated with the content of luminescent functional groups in the DOM. Based on these results, we confirmed that all fractions of DOM with C=O bonds, C=C bonds, and aromatic rings have an inhibitory effect on the growth of E. gracilis. In other words, inhibiting the growth of E. gracilis mainly depended on the concentration of different fractions.
According to the above results, DOM mainly includes HA, which was mainly composed of three organic compounds: HiN, HoA, and HoN (Fig. 3b). However, when using U254 signal to characterize these organics, in addition to HoA and HoN with relatively high signal intensity, HiA, HiB, and HoB also had an inhibitory effect on the growth of E. gracilis, suggesting that inhibitors other than HA may also be present in the recycled culture media. These growth inhibitory factors may be derived from SMBM, FA, and AP (Fig. 3b). These hydrophilic/hydrophobic fractions also have inhibitory effects on the growth of microalgae, such as FA has been proven to have an inhibitory effect on Scenedesmus species [13], indicating that this fraction, as well as HA with its highly fluorescent signals, may be potential inhibitors. However, both the concentration and the UV254 signal intensity of HiA, HiB, and HoB were lower than HoA and HoN derived from HA. In addition, although the concentration of HiN derived from HA was relatively higher, it obviously reduced the inhibitory effect on the growth of E. gracilis. Therefore, this study finally confirmed that the main growth inhibition of E. gracilis was HA, and the hydrophobic HoA and HoN organics fractions with higher content and higher UV254 signal intensity played a key inhibitory role. Lu et al. [10] only fractionated HoN-containing fatty acids and showed that they have an inhibitory effect on the growth of S. acuminatus. In addition, Zhang et al. [11] showed that all of the fractions could inhibit the growth of Scenedesmus sp. LX1, especially, HiB, HoB, and HiA. However, HoN and HoA showed the strongest inhibition of E. gracilis. This suggests that different microalgae may have different tolerances to different classifications of DOM. This scientific problem requires further research.
The influence of E. gracilis secretions on its physiology and biochemistry
The Fv/Fm ratio reflects the ability of microalgae to dissipate, absorb, and transmit light energy during photosynthesis. It is a useful parameter that indicates physiological state and growth rate, and is also an internal probe of the relationship between microalgae and their environment [13, 23, 24]. The Fv/Fm ratio for algae cells was only 0.12 in water containing HA, while that of the control group was 0.64, which is 5.3-fold difference (Fig. 4a, p < 0.01). From this result, it is obvious that HA significantly reduces the algae cell’s photosynthetic efficiency. Similarly, studies on S. acuminatus [13] and Arthrospira platensis [23] also showed comparable Fv/Fm reductions when cultivated in reused water, which means that HA has a negative effect on the photosynthetic system of these microalgae too. Thus, this impact has a certain universality.
The paramylon content and TFA content of E. gracilis in the experimental group containing DOM were 7.1% and 12.2%, respectively, while the control group was 21.2% and 35.2%. Both of these values were significantly lower than the control group, which showed a decrease of 66.5% (Fig. 4b, p < 0.01), and 65.3% (Fig. 4c, p < 0.01), respectively. These results were confirmed for the TFA of Scenedesmus sp. LX1 [11]. These results indicate that the HA secreted by E. gracilis may interfere with its own photosynthesis, and that this leads to inhibition of the synthesis of organic matter in the algae cells. The mechanism behind this process is worthy of our in-depth study in the future.
Study on the mechanism of E. gracilis growth inhibition by its own secretions
When UHPLC–QTOF-MS was used to detect metabolites in E. gracilis cells and cultivation media, the range of metabolites detected in negative ion mode was greater than that in positive ion mode, so this study only analyzes metabolites that were observed in negative ion mode to describe the mechanism whereby algae cells secrete DOM. With this analysis, we observed 4130 metabolites (Additional file 2). These metabolites were analyzed by PCA and OPLS-DA, and we can see clear separation between intracellular (IEG) and extracellular (EEG) metabolites (Additional file 1: Fig. S2, S3), indicating that there were significant differences in the metabolites in these two groups. When the OPLS-DA permutation test was performed on the data, the categorical variable Y was randomly changed 1000 times (Additional file 1: Fig. S4) and the original model R2Y was equal to 1, indicating that the established model conforms to the real situation for the sample data. The original model had a Q2 value equal to 0.997, which is very close to 1. This means that if a new sample were added to the model, it would fall within the existing distribution of data points. In general, the original model is robust and can explain the difference between the two sets of samples well. No overfitting was required to fit our data to it.
This study used VIP > 1 and a P-value < 0.05 to screen metabolites, and 108 different metabolites were obtained (see Additional file 2). According to the heat map cluster analysis, the relative concentration of 69 and 39 metabolites in the EEG and IEG were up-regulated, respectively (Additional file 1: Fig. S5). After these metabolites were annotated by the KEGG database, important metabolic pathways were screened according to their position and role in the relevant metabolic pathways (Additional file 2). According to the bubble chart, there are nine main metabolic pathways that are relevant: valine, leucine, and isoleucine biosynthesis; linoleic acid metabolism; arginine biosynthesis; the TCA cycle; pyruvate metabolism; purine metabolism; tyrosine metabolism; pyrimidine metabolism; and phenylalanine metabolism (Fig. 5). Among these, the first two are the key metabolic pathways. Some of the metabolites in these metabolic pathways were highly expressed inside the cell, and some were highly expressed outside the cell, and the latter group of metabolites may be secreted from the cell into the cultivation water. Three pathways—linoleic acid metabolism, the TCA cycle, and valine, leucine, and isoleucine biosynthesis—involve C=O and C=C bonds, while purine and pyrimidine metabolism contribute aromatic rings and C=O bonds. These metabolites accumulate in the medium and gradually become HA, which contains various functional groups (Fig. 6).
Palmitic acid was one of the metabolites secreted by S. acuminatus to inhibit growth [10]. Similarly, this study found that the palmitic acid in linoleic acid metabolism was higher in concentration in the culture medium. Therefore, it was further proved that palmitic acid was also one of the key factors that inhibit the growth of E. gracilis. 2-Isopropylmalate is an intermediate product of valine, leucine, and isoleucine biosynthesis. Excessive secretion of this intermediate product from algal cells into the cultivation medium may also inhibit the growth of E. gracilis. However, exactly how this intermediate metabolite inhibits the growth of E. gracilis is a question that requires in-depth research in the future.
HA containing multiple functional groups can complex iron ions that are essential for photosynthesis in microalgae. However, Sun et al. [25] found that the underlying mechanism of the inhibitory effect for cyanobacteria was not to reduce the bioavailability of iron, but to inhibit the oxidative damage of cells mediated by peroxidase-mediated. More and more evidence shows that HA could directly interact with certain large plants and algae through their different functional groups, thereby interfering with photosynthesis and growth. Due to their low molecular weight (< 50 kDa), these substances can easily pass through cell membranes. When these quinone-containing metabolites enter the chloroplast, they interfere with the electron transport processes of photosynthesis [26, 27]. In fact, the toxic effects of quinones on the growth and photosynthesis of Scenedesmus strains have been confirmed [28]. In addition, we have previously found that there was no significant difference between the experimental group and the control group under heterotrophic conditions containing HA (data not disclosed) and the Fv/Fm ratio was significantly reduced (Fig. 4a), which means that these inhibitors may primarily attack the photosynthetic system of the E. gracilis chloroplast. However, no metabolites related to quinones were found in the different metabolites screened in this study (Additional file 1: Fig. S5), indicating that the photosynthetic machinery of E. gracilis was not affected by quinones. Moreover, it is possible that different functional groups (e.g. C=C and C=O bonds, aromatic rings) interfere with the electron transport processes of photosynthesis. How these functional groups in the compounds secreted by different metabolic pathways interfere with the photosynthetic system of microalgae requires further in-depth study.
Removal of growth inhibitors
Markiewicz et al. [29] have confirmed that DOM in sewage is adsorbed effectively by AC. The fluorescence spectrum after AC treatment showed that the fluorescence signal was very weak (Fig. 7a), indicating that almost all of the HA that can fluoresce had been removed. In addition, the growth curve for the experimental group was almost the same as that of the control group (Fig. 7d). By the time of the last day of culture, the DWs of algal cells were 2.4 g/L (experimental group) and 2.4 g/L (control group), with no significant difference (p > 0.05). This indicates that AC is effective at completely adsorbing and removing substances that inhibit the growth of E. gracilis. Although the reused water of cultivated N. oceanica [12] and S. acuminatus [13] showed a relatively significant effect from AC treatment, the biomass obtained was slightly lower than the control group, indicating that some growth inhibitors could not be removed. However, AC effectively adsorbs growth inhibitors secreted by E. gracilis in this study. We would like to develop recyclable AC technology, such as biological AC, to increase the utilization rate so that it can be more convenient for large-scale reuse of water resources to cultivate E. gracilis.
AOPs have been widely used in the field of wastewater treatment. Oxidizers create a large number of free radicals under ultraviolet catalysis, such as hydroxyl radicals. These free radicals have strong oxidizing properties and can oxidize organic acids with unsaturated bonds [15]. According to the 3D-FEEM spectra of the reused water after AOP treatment, the fluorescence signal of the UV/H2O2/O3 group was weaker (Fig. 7b), followed by UV/H2O2 (Fig. 7c), indicating that the oxidation efficiency was higher with the participation of O3. In addition, the biomasses of the UV/H2O2/O3 group and the UV/H2O2 group on the last day were 2.89 g/L and 2.59 g/L, respectively, with the UV/H2O2/O3 experimental group significantly higher than the control group (p < 0.05). These results indicate that the advanced oxidation method not only eliminates the growth inhibitors, but may also oxidize these inhibitors into small organic molecules that could be absorbed by algae cells, thereby increasing their biomass. Our results show that the growth inhibitors were mainly HAs with luminescent functional groups (C=O and C=C bonds, aromatic rings). O3 and UV/H2O2 have been shown to work well for the treatment of Scenedesmus sp. LX1 [16] and S. acuminatus GT-2 [17] reused water, respectively. This study combines these two methods and shows that both methods together are more effective at removing growth inhibitors than either O3 or UV/H2O2 alone. Therefore, we believe that UV/H2O2/O3 is an ideal and efficient method for the removal of inhibitors of E. gracilis.
According to our previous research, the free radicals in the reused water after treatment with AOPs could also inhibit the growth of microalgae. Therefore, we need to optimize the AOP treatment process in the future by optimizing the treatment time, the concentration of the oxidizing agent, and the development of indicators for online detection of the concentration of free radicals in reused water (for example, the vitamin C reducing agent neutralization method). Use of AOPs is more conducive to the wide application of water reuse for algae cultivation. In addition, through these treatments, again it is clear that HA secreted by E. gracilis is a main growth inhibitor.
Based on the above results, we have proposed a cyclic culture model for E. gracilis (Fig. 8). The conceptual model is optimal when urea replaces NH4Cl as a nitrogen source and the reused water is filtered through an UFM and then treated with UV254/H2O2/O3. This model improves the availability of reused water, reduces the cost of cultivation, and increases the biomass of microalga E. gracilis.