Synergistic algicidal effect and mechanism of two diketopiperazines produced by Chryseobacterium sp. strain GLY-1106 on the harmful bloom-forming Microcystis aeruginosa

A potent algicidal bacterium isolated from Lake Taihu, Chryseobacterium sp. strain GLY-1106, produces two algicidal compounds: 1106-A (cyclo(4-OH-Pro-Leu)) and 1106-B (cyclo(Pro-Leu)). Both diketopiperazines showed strong algicidal activities against Microcystis aeruginosa, the dominant bloom-forming cyanobacterium in Lake Taihu. Interestingly, these two algicidal compounds functioned synergistically. Compared with individual treatment, combined treatment with cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu) significantly enhanced algicidal activity, accelerated the increase in intracellular reactive oxygen species (ROS) levels in M. aeruginosa, and further decreased the activities of antioxidases, effective quantum yield and maximal electron transport rate of M. aeruginosa. The results also showed that the algicidal characteristics of cyclo(4-OH-Pro-Leu) are distinct from those of cyclo(Pro-Leu). Cyclo(4-OH-Pro-Leu) mainly interrupted the flux of electron transport in the cyanobacterial photosynthetic system, whereas cyclo(Pro-Leu) mainly inhibited the activity of cyanobacterial intracellular antioxidases. A possible algicidal mechanism for the synergism between cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu) is proposed, which is in accordance with their distinct algicidal characteristics in individual and combined treatment. These findings suggest that synergism between algicidal compounds might be used as an effective strategy for the future control of Microcystis blooms.


Results
Isolation and identification of algicidal bacteria. A total of 212 isolates were obtained from surface water samples collected from Meiliang Bay of Lake Taihu in October 2012. Among these strains, approximately 8% ( Bacterial growth occurred under both aerobic and anaerobic conditions, but did not take place at temperatures of ≥ 37 °C. The Gram-negative isolate hydrolyzed casein and was phosphatase-, catalase-and oxidase-positive, but did not show arginine-, ornithine-and lysine-decarboxylase or phenylalanine deaminase activity. Acid was formed from D-glucose, but no gas was formed. The strain did not produce hydrogen sulfide from triple-sugar iron agar. When the 16S rRNA gene sequence was compared with the sequence available in the GenBank database, the strain GLY-1106 most closely resembled the type strain Chryseobacterium piscium LMG 23089 T (99% identity, GenBank accession number NR_042410). Most of the physiological properties of strain GLY-1106 were similar to those of C. piscium (see supplementary Table S1 online). However, some properties were different: in contrast to C. piscium, the isolate GLY-1106 utilized maltose and D-glucose rather than acetic acid, and it also tested negative for phenylalanine deaminase activity and did not grow in the presence of 5% NaCl. By combining the physiological and biochemical phenotypic characteristics and the results of phylogenetic analysis (see supplementary Algicidal mode of Chryseobacterium sp. strain GLY-1106. As shown in Supplementary Fig. S3, the algicidal activity of the strain GLY-1106 culture (A = 98.9%, t = 6 days) was slightly higher than that of its cell-free filtrate (A = 90.7%, t = 6 days), and much higher (P < 0.01) than that of its re-suspended washed cells (A = 37.5%, t = 6 days). No significant (P > 0.05) difference was observed between the algicidal activities of cell-free filtrate and heat-treated cell-free filtrate. These results indicated that strain GLY-1106 exhibited algicidal activity mainly by producing heat-stable extracellular algicidal compounds to attack cyanobacterial cells indirectly. Algicidal range of Chryseobacterium sp. strain GLY-1106. The algicidal activity of strain GLY-1106 was evaluated according to its efficiency in inhibiting on the dominant bloom-forming cyanobacterial species in Lake Taihu and other cyanobacterial and algal species (Table 1). Similar to most of the cyanobacterial species tested, especially those isolated from Meiliang Bay in Lake Taihu, strain GLY-1106 exhibited strong algicidal effects. After 6 days of co-culture, strain GLY-1106 showed the highest algicidal activity (A = 98.8%) against M. aeruginosa 9110 among all the cyanobacterial species tested. In addition, the strain GLY-1106 also exhibited an algicidal effect on Chlamydomonas sp. BS3 (A = 95.2%, t = 6 days), a eukaryotic algal strain from Lake Taihu.
Extraction and purification of the algicidal compounds. The ethyl acetate extract of the cell-free culture filtrate of strain GLY-1106 showed strong algicidal activity against M. aeruginosa 9110 on cyanobacterial lawns. Following silica gel chromatography of the extract, one fraction which exhibited algicidal activity was obtained from the effluent of the chromatographic column. When the fraction had been collected and applied to semi-preparative high performance liquid chromatography (HPLC), two primary fractions in the effluent from HPLC exhibited algicidal activity on cyanobacterial lawns, i.  23 .
Dynamics of the cellular density of strain GLY-1106, concentration of algicidal compounds, and cyanobacterial biomass during the algicidal process. As shown in Fig. 2, the cellular density of strain GLY-1106 was significantly positively related to the concentration of algicidal compounds (cyclo(4-OH-Pro-Leu): r = 0.975, P < 0.01; cyclo(Pro-Leu): r = 0.994, P < 0.01), and significantly  Table 1. Algicidal effect of strain GLY-1106 against various cyanobacterial and algal strains. * Isolated from Meiliang Bay of Lake Taihu. The algicidal activities were obtained after 6 days of coculture. Data are the mean ± SD from three independent assays. negatively related to the cyanobacterial biomass (r = − 0.948, P < 0.01) in general during the algicidal process of strain GLY-1106 against M. aeruginosa 9110. After 12 h of co-cultivation, when the cellular density of strain GLY-1106 increased to 8.3 × 10 6 CFU ml -1 in the co-culture of M. aeruginosa 9110 and strain GLY-1106, the algicidal compounds were still undetectable and the algicidal activities were not observed. After 24 h of co-cultivation, when the cellular density strain GLY-1106 increased to 5.1 × 10 7 CFU ml -1 , only the algicidal compounds cyclo(Pro-Leu) was detected, at a concentration of approximately 0.19 μ g ml -1 , leading to weak algicidal activity. When the cellular density of strain GLY-1106 increased to 1.2 × 10 8 CFU ml -1 on the 2nd day, the concentrations of cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu) reached 0.51 and 0.77 μ g ml -1 , respectively, resulting in a sharp drop in the cellular density of M. aeruginosa 9110 to 43.8% of the control (without inoculation of strain GLY-1106). After 6 days of co-cultivation, M. aeruginosa 9110 almost disappeared, the cellular density of strain GLY-1106 increased to 7.36 × 10 8 CFU ml -1 and the concentration of cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu) rised to 5.03 and 6.92 μ g ml -1 , respectively.

Cyclo(4-OH-Pro-Leu) and cyclo(Pro
Cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu) synergistically enhanced the damage to cyanobacterial physiology. To further investigate the underlying mechanism of the synergistic algicidal effects of cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu) on M. aeruginosa 9110, the physiological responses of the cyanobacterium to the individual and combined treatments were determined, including the changes in reactive oxygen species (ROS) level, malondialdehyde (MDA) content, antioxidase activity, effective quantum yield (Φ e ), and maximal electron transport rate (rETR max ). As shown in Fig. 5a, the intracellular ROS levels in M. aeruginosa 9110 after individual treatment with cyclo(4-OH-Pro-Leu) at 0.4 μ g ml -1 and 0.8 μ g ml -1 reached a maximum after 12 h and 9 h of exposure, respectively, and these two peak ROS levels were 1.28 and 1.36 times higher, respectively, than those of the control at the same time points. The ROS levels after individual treatment with cyclo(Pro-Leu) at 0.4 μ g ml -1 and 0.8 μ g ml -1 showed a slowly increase with time. After 48 h of exposure, the ROS levels in M. aeruginosa 9110 after individual treatment with cyclo(Pro-Leu) (0.4 μ g ml -1 ) and cyclo(Pro-Leu) (0.8 μ g ml -1 ) were 1.21 and 1.32 times higher than that of the control, respectively. The intracellular ROS level in M. aeruginosa 9110 with the combined treatment of cyclo(4-OH-Pro-Leu) (0.4 μ g ml -1 ) plus cyclo(Pro-Leu) (0.4 μ g ml -1 ) exhibited a greater and continuous tendency to increase compared with that after the individual treatments, and the increased ROS levels after the combined treatment were significantly (P < 0.01) higher than those after the individual treatments at each time point during the incubation process. After 48 h of exposure, the increased ROS level following the combined treatment was 2.26 times higher than the sum of the individual treatments with cyclo(4-OH-Pro-Leu) (0.4 μ g ml -1 ) and cyclo(Pro-Leu) (0.4 μ g ml -1 ).
As an indicator of lipid peroxidation 24 , the MDA content in M. aeruginosa 9110 (Fig. 5b) showed a continuous increase in all the treatments. During the algicidal process, the increased MDA content with the combined treatment (0.4 μ g ml -1 cyclo(4-OH-Pro-Leu) plus 0.4 μ g ml -1 cyclo(Pro-Leu)) was significantly (P < 0.01) higher than those in the individual treatments. After 48 h of exposure, the increased MDA content after the combined treatment was 1.75 times higher than the sum of the individual treatments with cyclo(4-OH-Pro-Leu) (0.4 μ g ml -1 ) and cyclo(Pro-Leu) (0.4 μ g ml -1 ).
As shown in Fig. 6, the response of the antioxidant enzyme system to the algicidal compounds cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu) was obviously different during the algicidal process. The activities of superoxide dismutase (SOD) (Fig. 6a) following the individual treatment with cyclo(4-OH-Pro-Leu) showed an activating effect compared with the control, and generally exhibited a tendency first to increase then to descend. The SOD activities after the individual treatment with cyclo(4-OH-Pro-Leu) (0.4 μ g ml -1 ) and cyclo(4-OH-Pro-Leu) (0.8 μ g ml -1 ) reached the highest levels after 15 and 12 h of exposure, respectively. However, the activities of SOD (Fig. 6a) following the individual treatment with cyclo(Pro-Leu) showed an inhibitory effect compared with the control and generally exhibited a tendency to decline throughout the algicidal process. The activities of SOD with the combined treatment (0.4 μ g ml -1 cyclo(4-OH-Pro-Leu) plus 0.4 μ g ml -1 cyclo(Pro-Leu)) showed a much greater reducing trends with time. In particular, the activities of SOD after the combined treatment decreased dramatically after 6 h of exposure, and were significantly lower than those following the individual treatments. After 48 h of exposure, the activities of SOD with the combined treatment were only 21.3% of those in the control. During the algicidal process, the activities of catalase (CAT) (Fig. 6b) and peroxidase (POD) (Fig. 6c) varied and resembled those of SOD (Fig. 6a). Following individual treatment with cyclo(4-OH-Pro-Leu), the activities of SOD (cyclo(4-OH-Pro-Leu) (0.4 μ g ml -1 ): r = 0.722, P < 0.01; cyclo(4-OH-Pro-Leu) (0.8 μ g ml -1 ): r = 0.611, P < 0.05) and POD (cyclo(4-OH-Pro-Leu) (0.4 μ g ml -1 ): r = 0.556, P < 0.05; cyclo(4-OH-Pro-Leu) (0.8 μ g ml -1 ): r = 0.558, P < 0.05) were significantly and positively correlated with ROS levels. However, the activities of these three antioxidases were significantly and negatively (r = -1.000, P < 0.01, in all cases) correlated with ROS levels following individual treatment with cyclo(Pro-Leu) and the combined treatment.
As shown in Fig. 7a,b, the Φ e and rETR max following all treatments showed a tendency to decline during the algicidal process. The Φ e and rETR max following individual treatment with cyclo(4-OH-Pro-Leu) (0.8 μ g ml -1 ) or cyclo(4-OH-Pro-Leu) (0.4 μ g ml -1 ) decreased gradually with time, and were 60.9% and 39.7%, or 75.6% and 49.93% of the control after 48 h of exposure, respectively. The Φ e and rETR max following individual treatment with cyclo(Pro-Leu) (0.8 μ g ml -1 ) or cyclo(Pro-Leu) (0.4 μ g ml -1 ) began to decrease after 12 h, and were 89.3% and 80.5%, or 94.6% and 88.2% of the control after 48 h of exposure, respectively. The Φ e and rETR max after the combined treatment (0.4 μ g ml -1 cyclo(4-OH-Pro-Leu) plus 0.4 μ g ml -1 cyclo(Pro-Leu)) decreased at a higher rate, and were lower than those following individual treatments at each time point during the algicidal process. In particular, after 6 h, with the highest and continuous increase in ROS level in the combined treatment (Fig. 5a), these two photosynthetic parameters decreased markedly compared with those following individual treatments and the control. After 48 h of exposure, the Φ e was 34.1% of the control, while rETR max was almost 0% with the combined treatment. During the algicidal process, the decrease in rETR max was greater than that of Φ e with the combined treatment and individual treatment with cyclo(4-OH-Pro-Leu), but not individual treatment with cyclo(Pro-Leu). In addition, both Φ e and rETR max were significantly (r = −1.000, P < 0.01) and negatively correlated with the ROS level following the combined treatment.

Discussion
Cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu) are diketopiperazines, which are the simple peptide derivatives. Diketopiperazines constitute a large class of small molecules synthesized by a broad range of microorganisms which show several useful biological properties 25 , and act as antibacterial compounds 26 , antitumor and immunosuppressive agents 27 , and so on. The biosynthesis of diketopiperazines is thought to be catalyzed either by nonribosomal peptide synthetases or by cyclodipeptide synthases 28 . Cyclo(Pro-Gly), produced by Shewanella sp. strain Lzh-2 14 and Stenotrophomonas sp. strain F6 17 , and cyclo(Pro-Val), produced by Bacillus sp. strain Lzh-5 16 , showed algicidal activity against M. aeruginosa. Compared with cyclo(Pro-Gly) and cyclo(Pro-Val), cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu) showed the similar or stronger algicidal activities against M. aeruginosa. To the best of our knowledge, the cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu) isolated from Chryseobacterium spp. in this study are newly discovered algicidal substances.
Chemicals with similar structures often show distinct biological properties. Xiao and colleagues 6 isolated a pair of chiral flavonolignans from barley straw, and found that, after short-term exposure, one of the chiral flavonolignans caused significant damage to cyanobacterial cell membranes, but had no influence on the intracellular ROS level, whereas the other chiral flavonolignan induced an increase in intracellular ROS level, but no damage to the cell membrane. In the current study, the difference in the chemical composition between cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu) is only one hydroxyl group, but their biological properties are vastly different. Synergism may be a common characteristic of some co-occurring secondary metabolites and a powerful driving force in their evolution 29 . Synergism occurring naturally has been observed in bacteriocins, which are peptides of ribosomal origin produced by lactic acid bacteria 30 . The co-produced secondary metabolites of some Actinomycetes act synergistically against either a target microorganism or competitors for nutrients 29,31 . In the present study, the combination of cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu) significantly enhanced the algicidal activity against the bloom-forming M. aeruginosa. To the best of our knowledge, this is the first report of the synergistic algicidal action of diketopiperazine algicidal substances.
ROS are involved in damage to living organisms under environmental stress 32 . Excessive ROS may cause irreversible oxidative damage to intracellular components, finally leading to cell death 33 . The reaction centers of the photosystem are the major generation sites of ROS in cells 34 . During the photosynthesis process in cyanobacteria, chlorophyll a in photosystem II initially captures photons with sufficient energy (λ < 680 nm), and generates electrons; after this, the electrons are transferred via cytochrom bf to photosystem I and are used to produce the NADPH (dihydronicotinamide adenine dinucleotide phosphate) 1 . The photoproducton of ROS can be enhanced by decreased capacity of the electron transport flux 34 . On the other hand, superoxide dismutase (SOD), catalase (CAT) and peroxidase (POD) are three important antioxidases that can scavenge the excessive ROS and protect cells from damage caused by oxygen free radicals 20 . During the algicidal process in the present study, individual treatment with cyclo(4-OH-Pro-Leu) caused a much greater decrease in photosynthetic quantum yield and electron transport rate of M. aeruginosa than individual treatment with cyclo(Pro-Leu), and the degree of reduction in the electron transport rate was larger than that in the photosynthetic quantum yield, indicating that cyclo(4-OH-Pro-Leu) may induce excessive production of ROS in M. aeruginosa mainly by interrupting the electron transport flux. In addition, the antioxidase system was activated by individual treatment with cyclo(4-OH-Pro-Leu) during the algicidal process, and the activities of SOD, CAT and POD were much higher than those following the individual treatment with cyclo(Pro-Leu). In contrast to cyclo(4-OH-Pro-Leu), individual treatment with cyclo(Pro-Leu) caused a significant decrease in antioxidase activities during the algicidal process, but showed a weak or even negligible influence on photosynthetic quantum yield and electron transport rate, which suggests that cyclo(Pro-Leu) may induce excessive levels of ROS in M. aeruginosa mainly by inhibiting the activities of antioxidases. It can be seen that the algicidal characteristics of cyclo(4-OH-Pro-Leu) are distinct from those of cyclo(Pro-Leu). Compared with individual treatment with cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu), the combined treatment not only significantly aggravated the antioxidases activities, but also interrupted electron transport flux in the photosystem of M. aeruginosa, resulting in a synergistic algicidal effect. With regard to the marked decrease in the activities of antioxidases in the combined treatment after 6 h of exposure, the oxidative damages induced by the continuous increase of ROS level could play a key role in this phenomenon. Accordingly, the mechanism underlying the synergistic algicidal effect of cyclo(4-OH-Pro-Leu) (mainly inhibiting photosynthesis) and cyclo(Pro-Leu) (mainly inhibiting antioxidase activity) may be as follows: both the photosynthetic capacity and antioxidase activities in cyanobacterial cells are inhibited by the two substances simultaneously, accelerating the disruption of redox homeostasis and the increase in intracellular ROS level and lipid peroxidation; in addition, the higher level of ROS may further accelerate the decline in antioxidase activities 35 and photosynthetic capacity 36,37 , resulting in a greater increase in ROS and more oxidative damage in cyanobacterial cells.
Based on the results of this study, it could be seen that the synergism really exists between some algicidal compounds produced by algicidal bacterium, although there was no such phenomenon found in the previous studies on the algicidal compounds derived from algicidal bacteria. Therefore, determining whether or not algicidal synergism exists among more known algicidal compounds derived from algicidal bacteria and exploiting the collaborative mechanism would be of great significance in the future. In addition, synergism between algicidal compounds might be used as an effective strategy for the future control of Microcystis blooms. ) of standard cyclo(4-OH-Pro-Leu) or cyclo(Pro-Leu) was added to a 9.9-ml culture of M. aeruginosa 9110 to yield a final concentration of 0.1, 0.2, 0.4, 0.6, 0.8, 1.0, 1.5, 2, 2.5, 3, 4, 5, 10, 20, 30, 40, 50, or 100 μ g ml -1 , respectively. In addition, control (without addition of algicidal compounds) was prepared using an equal amount of sterile distilled water instead of the standard algicidal compound solution. After 24 h of incubation under cyanobacterial growth conditions, the corresponding survival rate (R s ) of M. aeruginosa 9110 was examined. The EC 50 values were calculated from the relevant dose response curves by probit analysis 39 . Synergistic effect of cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu) against M. aeruginosa 9110. To examine whether there was a synergistic effect between cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu), combined treatment and corresponding individual treatments with these two algicidal compounds were conducted with an initial cyanobacterial density of 1 × 10 7 cells ml −1 . Briefly, in the combined treatment, 0.5-ml aliquots of stock solutions (0.08 mg ml −1 ) of standard cyclo(4-OH-Pro-Leu) and cyclo(Pro-Leu) were added to a 99-ml culture of M. aeruginosa 9110 (99-ml culture of M. aeruginosa 9110, 0.5 ml of cyclo(4-OH-Pro-Leu), and 0.5 ml of cyclo(Pro-Leu); total, 100 ml), respectively. In the individual treatments, an aliquot (1 ml) of each stock solution (0.04 or 0.08 mg ml -1 ) of standard cyclo(4-OH-Pro-Leu) or cyclo(Pro-Leu) was added to a 99-ml culture of M. aeruginosa 9110 to yield a final concentration of 0.4 or 0.8 μ g ml -1 , respectively. In addition, 1-ml sterile distilled water was added to a 99-ml culture of M. aeruginosa 9110 as the control (without addition of algicidal compounds). The cyanobacterial cultures in the treatment groups and control were incubated for 48 h under cyanobacterial growth conditions. During the incubation period, aliquots from both treatment groups and the control were sampled to determine algicidal activities and other relevant physiological indices at the following time points: 0, 3rd, 6th, 9th, 12th, 15th, 18th, 24th, and 48th hour.

Flow cytometry-based analyses on the intracellular ROS level. Flow cytometric measurements
were carried out using a FACSAria II flow cytometer (BD Biosciences, USA) equipped with Diva software for data acquisition. FlowJo software (Tree Star, USA) was used for data analysis. For each sample of cyanobacterial cells, 20,000 events were collected by flow cytometer. Cyanobacterial intracellular ROS formation was determined at the single-cell level following the method described by Wang and colleagues 40 , and the membrane-permeable dye 2′ ,7′ -dichlorodihydroflurescein diacetate (No. D6883, Sigma) was used as a probe. When the probe molecules enter cells, they may be transformed into 2′ ,7′ -dichlorodihydrofluorescein (H 2 DCF) by intracellular esterase. Once the intracellular ROS generated, H 2 DCF would be converted into highly fluorescent 2′ ,7′ -dichlorofluorescein (DCF). Therefore, we determined the fluorescence intensity (FI) of DCF by flow cytometry to indicate the extent of ROS generation. Approximately 10 6 cells in the treatment groups and control were harvested by centrifugation at 3,000 × g for 10 min, washed twice with sterile phosphate-buffered saline (PBS) solution (50 mM, pH 7.0) and re-suspended in 1 ml PBS (50 mM, pH 7.0). The cyanobacterial suspensions containing 10 μ M H 2 DCF-DA were incubated in the dark at 25 °C for 60 min and then washed twice with sterile fresh PBS solution. The stained cells were then analyzed by flow cytometry. Changes in ROS levels as compared with the control were evaluated using the following formula 40 Determination of antioxidase activity and lipid peroxidation. Determination of photosynthetic performance. The photosynthetic performance of photosystem II was determined using a PHYTO-PAM phytoplankton analyzer (Walz, Germany) following the method described by Ou and colleagues 42 . A 5-ml sample of the suspension was analyzed immediately after harvesting. When a sample was acclimated to the light in its environment, effective quantum yield (Φ e ) can be calculated as follows: where F s and F m' are the corresponding light-acclimated steady-state and maximum fluorescence, respectively, and △ F is the difference between F m' and F s . The photosynthetic parameter (Φ e ) is an approximation of the fraction of absorbed energy used for photochemistry in the total energy at a specific time and is therefore dimensionless 43 . The other photosynthetic parameter (rETR max ), which indicates the maximum photosynthetic capacity with the unit of μ mol m -2 s -1 44 , was calculated following the method described by Ralph and Gademann 45 . In practice, these parameters can be obtained directly from the PHYTO-PAM analyzer.
Data analysis. The data in this study were obtained from three replicates and are presented as means± standard deviation. One-way analysis of variance (ANOVA) was carried out using SAS 9.1.3 (SAS Institute Inc., Cary, NC, USA). Comparisons between the means were conducted using Duncan's Multiple Range Test. The correlation analysis and probit analysis were conducted using SPSS v 20.0 (IBM Corp., Armonk, NY, USA).