Strains of cyanobacteria belonging to the genus, Microcystis, were cultured in the Department of Biological Sciences at Florida International University (Miami, FL). Strains included those isolated from temperate freshwater systems of the Great Lakes, MC 81-11 (Bay of Quinte, Lake Ontario) and MC 95-11 (Lake Erie), and from sub-tropical South Florida, MC 36-1 (Doctors Lake, Hog Point, FL, USA), as well as a strain (299) of M. aeruginosa obtained from the reference collection of the University of Toronto Culture Collection of algae and cyanobacteria (Table 1). All isolates (except M. aeruginosa 299) were taxonomically identified (to genus) by microscopic observations using classical morphological criteria given in Komarek and Anagnostidis [20,21]. The cultures were grown in 3L flasks in BG11 medium under cool-white fluorescent light (intensity 30 µEm⁻²s⁻¹) with aeration as previously described [10,11,22]. The cells were harvested after three to four weeks by centrifugation and freeze-dried.

LPS were extracted from collected cyanobacterial biomass using the standard “hot phenol-water extraction” method as modified by Bernadova et al. [23]. Briefly, freeze-dried biomass from cultures were re-suspended in pyrogen-free water, added to equal volume of 90% phenol (in water) and extracted (with stirring) at 68 °C for 20 min. The aqueous layer was removed by centrifugation (2500 g for 30 min), and the remaining phenol extracted a second time with water. The pooled aqueous fractions were dialyzed (3500 molecular weight cut-off cellulose) against distilled water, and lyophilized. The lyophilized sample was re-dissolved in 0.1 M Tris buffer with 25 µg·mL⁻¹ RnaseA (Sigma-Aldrich, St. Louis, MO, USA), and incubated at 37 °C for 16 h. An equal volume of phenol was added to the resuspended sample, and after incubation (at room temperature for 4 min), the aqueous layer was collected by centrifugation, dialyzed a second time and subsequently lyophilized to produce the LPS in powder form. In addition to the LPS extracted from cyanobacterial biomass, a standard of LPS from Escherichia coli (0128:B12), a heterotrophic bacterium, was purchased from Sigma-Aldrich (St. Louis, MO, USA).

“Endotoxin activity” is a commonly employed measure of LPS, particularly in the investigation of heterotrophic bacteria, and was used here to identify possible correlation between this activity and other biological activities. Specifically, we utilized the standard Limulus Amoebocyte Lysate (LAL) Endosafe Endochrome-K Assay that gives a quantitative detection of endotoxins by kinetic-chromogenic methods. Analyses were performed by Charles River Laboratories, Inc. (Wilmington, MA). Endotoxin activity was expressed as endotoxin unit per mg of LPS (EU/mg). In addition to endotoxin analysis, an enzyme-linked immunosorbent assay (ELISA) specific to the widespread microcystins (MCYSTs), and related nodularins (Abraxis Kits, Warminster, PA), was performed to test for MCYST content in the prepared LPS, and eliminate any possible contribution of this cyanobacterial toxin.

Zebrafish embryos of the L3 line (obtained from the University of Miami Rosenstiel School of Marine and Atmospheric Science-RSMAS) were used for all experiments presented here. Maintenance and breeding of the zebrafish were carried out by standard methods, and specifically as described by Berry et al. [10,11]. After successful breeding, the eggs were collected, rinsed and placed in a Petri dish with E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl₂, and 0.33 mM MgSO₄; [24]). Unfertilized and poor quality (i.e., clearly moribund) embryos were removed from the dish and the remaining embryos were subsequently used within 3 hpf (4- to 64-cell stage) for exposures.

For evaluation of detoxifying enzymes, embryos were exposed to 250 μg/L of LPS from E. coli and Microcystis isolates in E3 medium based on a method modified from Cazenave et al. [25] for 24 h. Specifically, 200 embryos per replicate were placed in polystyrene culture dishes, containing LPS (diluted to 250 µg/L) in 30 mL E3 medium; for negative controls, embryos were placed in dishes containing 30 mL of E3 medium only. All exposures were done in triplicate (600 embryos total). The embryos were then “flash frozen” in liquid nitrogen and stored at −80 °C until enzyme extraction. The enzyme preparations were carried out according to the methods of Wiegand et al. [26] by which the samples were mechanically homogenized using phosphate buffer at a pH of 7.4. The cell debris was removed, and the supernatant was collected for enzyme studies. To normalize enzyme activity, total protein content was determined by the “Bradford Assay” [27] using bovine serum albumin as a standard.

Enzyme (GST, GPx, GR, SOD and CAT) activities, as well as total glutathione and relative reduced/oxidized glutathione (GSH/GSSG) were evaluated using commercially available kits (Sigma-Aldrich, St. Louis, MO; Cayman Chemical Company, Ann Arbor, MI), as per manufacturer’s instructions and standard techniques, and expressed relative to total protein. Spectrophotometric measurements were made using a Biotek Synergy HT multi-well plate reader. Total glutathione was determined based on spectrophotometric measure (412 nm) of the production of the 5-thio-2-nitrobenzoic acid (TNB) product; samples were deproteinated using metaphosphoric acid and triethanolamine prior to assays. Oxidized glutathione (GSSG) was specifically determined by derivatizing reduced GSH using 2-vinylpyridine (prior to assays, but following deproteination). GST activity (µmol/mL/min) was measured colorimetrically, specifically based on the conjugation of GSH to 1-chloro-2, 4-dinitrobenzene (CDNB), and subsequent measurement of the product at 340 nm. GPx was measured based on the decrease in NAPDH (340 nm) that is consumed in the recycling of oxidized GSSG to GSH, following reduction of hydroperoxide. GR, on the other hand, was determined, based on the conjugation of GSH to 5,5′-dithiobis(2-nitrobenzoic acid) following recycling of GSSG to the reduced form by GR. Both GPx and GR activity of embryo preparations were compared to standards of purified enzymes. CAT and SOD activity were both determined colorimetrically using kits from Cayman Chemical Company; CAT assays specifically measured the production of formaldehyde (from methanol) using the chromagen, 4-amino-3-hydrazino-5-mercapto-1,2,4-triazole (540 nm), whereas SOD activity was measured based on the dismutation of superoxide radicals, generated by xanthine oxidase, using tetrazolium salts (440 nm).

For the enzymes activities assays, a one-way ANOVA followed by F-test was carried out to determine if LPS treatments were statistically different from the control (p ≤ 0.05).

Cyanobacterial LPS were isolated from laboratory cultures of geographically diverse Microcystis strains, including temperate (MC 81-11 and MC 95-11) and sub-tropical (MC 36-1) representatives. LPS were found to account for up to 4.1% of the dried biomass weight (Table 1). In general agreement with prior studies of heterotrophic bacterial [28] and cyanobacterial [29] LPS, considerable variation in extraction yield (0.1–4.1%) between Microcystis strains was observed (Table 1). In fact, LPS from an additional sub-tropical Microcystis isolate were prepared, but yielded too little for subsequent experimental studies (data not shown).

Endotoxin activities of LPS preparations were measured, using the standard LAL assay, and found to be highly variable (Table 1). Specifically, endotoxin activity (per unit weight, i.e., EU/mg) associated with LPS preparations varied over nearly three orders of magnitude. This variability in endotoxin activity agrees with previous studies [23,30,31] in which the LPS from several, diverse cyanobacterial species were studied, and the endotoxin activities, based on the Limulus assay, likewise, were seen to vary considerably (i.e., 2–3 orders of magnitude).

Indeed, compared to heterotrophic bacterial LPS, little is known about the structural and chemical composition of cyanobacterial LPS, in general, and specifically how chemical composition/structure differences may relate to endotoxin activity. Raziuddin et al. [32] compared LPS from two strains of Microcystis aeruginosa to Salmonella arbortus equi, and found that cyanobacterial LPS had considerably less endotoxin activity than the heterotrophic bacteria (based the LAL assay). More generally, cyanobacterial LPS have been found to be less active in all of the models typically associated with endotoxin activity [8,30].

Only one recent study [33] has provided a complete structural elucidation of a cyanobacterial LPS. Specifically, Snyder et al. [33] characterized the structure of LPS from a marine cyanobacterial genus, Synechococcus. In this study, cyanobacterial LPS (compared to typical heterotrophic bacterial LPS) was found to represent a simplified, and possibly “primordial,” structure. In particular, Synechococcus LPS lacked the usual heptose and Kdo-replaced, instead, by 4-linked glucose-in the oligosaccharide core, and contained a lipid A moiety uniquely characterized by odd-chain hydroxylated fatty acids and lack of phosphates. Similarly, Lindsay et al. [29] recently evaluated LPS preparations from several laboratory strains of cyanobacteria, including Microcystis, and did not detect Kdo in any of the samples evaluated. Unlike preparations from Microcystis, LPS from Synechococcus showed no LAL activity. Taken together with existing toxicological information, these findings strongly suggest that chemical differences relative to heterotrophic bacterial LPS- and particularly, perhaps, the lipid A structure (associated with TLR4-mediated effects; see Introduction) or oligosaccharide core-may underlie the relatively low endotoxin-like activity of cyanobacterial LPS.

In addition to LPS, Microcystis is a well-known producer of the heptapeptide toxin, microcystin. Microcystins have been linked to modulation of each of the detoxification pathways investigated in the present study [34,35,36,37]. Evaluation of LPS preparation using ELISA, specific to the characteristic Adda moiety of microcystins (and related nodularins), suggested that these toxins were not present (within the limit of quantitation, i.e., ≥0.15 ppb) in any of the samples evaluated, and thus not associated with observed effects on detoxification pathways (discussed below).

Relative to heterotrophic bacterial LPS, there have been very limited studies on the toxicology of cyanobacterial LPS. In one very recent study, specifically using LPS isolated from one of the same cyanobacterial strains (i.e., M.aeruginosa 299) investigated here, Mayer et al. [8] showed that cyanobacterial LPS modulates expression of several proinflammatory markers in the rat brain microglia model. However, effects were typically observed at two to three orders of magnitude higher concentrations for cyanobacterial, compared to heterotrophic bacterial, LPS. Similarly, Keleti and Sykora [30] previously showed that cyanobacterial LPS from Oscillatoria brevis and Anabaena cylindrical are about ten times less toxic, in adrenalectomized mice, than heterotrophic bacterial (i.e., Salmonella) LPS.

In addition to these studies, several recent studies (including the present) have specifically investigated the effects of cyanobacterial LPS in various teleost fish models [15,16,37,38]. Unlike mammalian models, teleost fish have been found to either lack, or contain an evolutionarily divergent form of, TLR4 [19] as the recognized mediator of endotoxin, e.g., immunomodulatory, activity [39]. Moreover, this evolutionary divergence of TLR4 has been clearly linked to the resistance of teleost fish to the immunomodulatory effects of heterotrophic bacterial LPS [17,18]. For example, Sepulcre et al. [18] showed that, although heterotrophic bacterial LPS stimulated immunomodulatory effects in zebrafish (which express an evolutionarily divergent TLR4 ortholog, whereas more evolutionarily advanced fish do not), these effects were only observed at significantly (approximately 10³-fold) higher doses of LPS, and occurred via TLR4-independent pathways.

In the present study, we evaluated the effects of LPS from Microcystis in relation to several detoxification enzymes/pathways in the zebrafish embryo model. No difference in mortality, or apparent impairment of development, was observed in zebrafish embryos exposed to any of the LPS treatments. However, a significant modulation of several detoxifying pathways was measured.

GST activity decreased after 24 h exposure of zebrafish embryos to both heterotrophic bacterial and cyanobacterial LPS (at 250 µg/L). Compared to the control group (0.15 nkatal/mg protein), GST activity was significantly decreased in embryos that were exposed to E. coli (0.076 nkatal/mg protein), M. aeruginosa 299 (0.078 nkatal/mg protein), MC 36-1 (0.11 nkatal/mg protein) and MC 81-11 (0.056 nkatal/mg protein) LPS. Embryos exposed to MC 95-11 LPS showed a decrease (0.12 nkatal/mg protein), although it was not statistically significant (Figure 1). Inhibition of GST (compared to “No LPS” controls) varied considerably between LPS preparations, ranging 61.9% inhibition for the most active (MC 81-11) to approximately 20–25% for the least active (20.6% and 25.5%, respectively for MC 95-11 and MC 36-1).

Inhibition of GST is consistent with prior studies in zebrafish and other teleost fish models [15,37,38]. Best et al. [15] similarly demonstrated GST inhibition by cyaonbacterial LPS in the zebrafish embryo model. More recently, Wang et al. [37] found apparent down-regulation of GST mRNA in response to LPS in the Nile tilapia (Oreochromis niloticus).

In addition to GST, we also investigated two additional enzymes, GPx and GR, which reciprocally utilize glutathione, and specifically recycle the peptide between its reduced and oxidized forms. While no effects of LPS on GR were observed, GPx moderately increased in zebrafish embryos exposed to all LPS evaluated (Figure 2). GPx showed a significant increase from the controls (0.07 nkatal/mg protein) in embryos exposed to E. coli (0.13 nkatal/mg protein), and MC 95-11 (0.13 nkatal/mg protein). Though increases in the embryos exposed to M. aeruginosa 299 (0.12 nkatal/mg protein), MC 36-1 (0.11 nkatal/mg protein), and MC 81-11 (0.12 nkatal/mg protein) were observed, they were not statistically significant (Figure 2). To the authors’ knowledge, this is the first report of the effect of LPS on GPx in the zebrafish embryo model. However, this is consistent with prior studies of tilapia by Wang et al. [37] who, likewise, showed a moderate up-regulation of GPx transcription by LPS.

In order to evaluate the role of oxidative stress, as well as the possible effects of LPS on glutathione levels, in relation to the modulation of GST and GPx, we measured both total glutathione, as well as relative ratios of reduced (GSH) and oxidized (GSSG) forms. Heterotrophic bacterial LPS is associated with oxidative stress in multiple model systems [40]. The relative concentration of GSH/GSSG, and specifically a decrease in this ratio—due to the relative depletion of GSH, and subsequent accumulation of the oxidized GSSG—has been widely employed as an indicator of oxidative stress in a range of organisms [41], including zebrafish [42]. In agreement with this, we observed a slight, but not statistically significant, decrease in GSH/GSSG ratio (Figure 3) in embryos exposed to E. coli LPS (Figure 3). The slightness of the effect would be consistent with prior studies [19] which indicate teleost fish, and specifically zebrafish, are generally resistant to TLR4-mediated effects (e.g., immunomodulatory, oxidative stress) of heterotrophic bacteria LPS.

In contrast, we actually observed an increase, although not statistically significant, in the ratio of reduced/oxidized glutathione in embryos exposed to cyanobacterial LPS (Figure 3), suggesting an alternative mechanism, and not oxidiative stress, for concomitant modulation of GST and GPx (see below). In further support of this, no change in either SOD or CAT activity was observed (data not shown) in embryos exposed to LPS. Unlike GST and GPx that primarily act via interaction with ROS-generating xenobiotics or products of oxidative damage (e.g., lipid peroxides), both SOD and CAT directly detoxify ROSs (i.e., superoxide and hydrogen peroxide, respectively), and are thus generally associated most directly with oxidative stress.

Rather than oxidative stress, cyanobacterial LPS increased both total glutathione and reduced GSH (Figure 4) levels in exposed embryos. Embryos exposed to all cyanobacterial LPS treatments showed a significant increase total glutathione pool (i.e., both GSH and GSSG; Figure 4A), relative to the untreated (i.e., “No LPS”) controls (7.28 µM). In contrast, there was no significant increase-and, in fact, a slight decrease (7.22 µM)-in total glutathione for embryos exposed to E. coli LPS (Figure 4A). Exclusive measurements of reduced and oxidized glutathione determined that GSSG content showed no significant difference compared to the control, but a significant increase in GSH content for embryos exposed to LPS from M. aeruginosa 299 (5.86 µM, p < 0.05), MC 81-11 (6.33 µM, p < 0.01), and MC 95-11 (6.21 µM, p < 0.01) (Figure 4B). GSH levels also increased, relative to controls (5.42 µM) in embryos exposed to LPS from MC 36-1 (5.68 µM), as well as E. coli (5.70 µM), but increases were not statistically significant.

These results, taken together, support an alternative mechanism (i.e., not oxidative stress), and perhaps a possible direct interaction between LPS and glutathione-based detoxification pathways. The increase in total glutathione pool (Figure 4A) clearly suggests that cyanobacterial LPS either directly or indirectly elevate levels of the peptide. In support of a direct interaction between LPS and GST expression, and consequent activity, Wang et al. [37] cloned the putative soluble GST-α gene from tilapia—the transcription of which was shown to increase with LPS exposure-and identified a putative “LPS response element.” Comparison to the sequence of a GST-α (BC056725) cloned from zebrafish, in fact, identifies a similar sequence, and further supports the possible involvement of transcription level regulation of GST by LPS in this model. Likewise, the same study showed a concomitant increase in GPx transcription (in the trout model) with LPS exposure, in agreement with moderate effects on GPx activity observed here.