Effects of organic solvents on the high performance liquid chromatographic analysis of the cyanobacterial toxin cylindrospermopsin and its recovery from environmental eutrophic waters by solid phase extraction

Abstract

The effect of organic solvents on the high performance liquid chromatography (HPLC) analysis of cylindrospermopsin using photodiode array detection was examined since organic solvents are commonly used to extract this toxin from cyanobacteria and in the mobile phase compositions used in HPLC. Increasing concentrations of methanol resulted in an increase in the UV absorbance of purified cylindrospermopsin according to spectrometry, but to a marked decrease during HPLC analysis when the concentration of this solvent was greater than 50% methanol, or when acetonitrile concentrations exceeded 30% (v/v). Precipitation of cylindrospermopsin at these high concentrations of organic solvents was not observed. Solid phase extraction methods were developed to recover the toxin from spent extracellular growth medium after laboratory culture of Cylindrospermopsis raciborskii strain CR3 as an aid to toxin purification and from spiked environmental water samples. Using C18 and polygraphite carbon cartridges in series, 100% recoveries of cylindrospermopsin were achieved for lake waters spiked at 1 µg l⁻¹.

Cylindrospermopsin, Cyanobacterial toxin, Analysis, Solid phase extraction

1 Introduction

Cylindrospermopsin is a cyanobacterial guanidine alkaloid hepatotoxin [1] originally isolated from cultures of Cylindrospermopsis raciborskii[2]. The toxin has since been isolated from cultures of Aphanizomenon ovalisporum[3,4], Umezakia natans[5] and Raphidiopsis curvata[6]. In comparison to the more widely known hepatotoxic microcystins of cyanobacteria, which include over 70 variants [7], only two further naturally occurring analogues of cylindrospermopsin have been identified, namely deoxycylindrospermopsin [8] and 7-epicylindrospermopsin [9]. Of these, deoxycylindrospermopsin has been shown to be approximately one tenth as toxic as cylindrospermopsin [8] and 7-epicylindrospermopsin appears to have a similar toxicity to cylindrospermopsin [9]. Studies such as these have indicated that the uracil moiety of cylindrospermopsin may be required to express toxicity [10].

Several methods are used to detect and analyse cylindrospermopsin, including mouse bioassay [11], and high performance liquid chromatography (HPLC) employing either photodiode array detection (HPLC-PDA) [5] or mass spectrometry (HPLC-MS/MS) [12]. Influences of organic solvent concentration on the quantification of cylindrospermopsin by HPLC-PDA are identified in the present communication.

The recovery of cylindrospermopsin from solution has been performed using resins [5] and graphitised carbon solid phase extraction (SPE) cartridges have recently been applied to recover the toxin from the cells and growth medium of cultures of C. raciborskii[13]. In this study, we further optimised the extraction of cylindrospermopsin from spent C. raciborskii medium by combining C18 and graphitised carbon solid phases. This procedure was also shown to be effective to extract the toxin from spiked environmental waters. These procedures are recommended to extract and purify cylindrospermopsin for analytical and research purposes, since the toxin is not available commercially.

2 Materials and methods

2.1 Purification and HPLC-PDA analysis of cylindrospermopsin

C. raciborskii strain CR3 [14] was grown in 10-l batch cultures in BG11 medium minus nitrate [15]. The cells were harvested during late growth phase and removed from the culture medium by tangential flow filtration using a Millipore Pellicon 2 tangential filter (Millipore, Watford, UK). The filtrate was retained and stored for SPE and toxin purification. Cylindrospermopsin was purified using Waters equipment including a 510 pumping system, 471 WISP, 484 UV detector and a Waters fraction collector employing a Phenomenex Luna semiprep column (C18, 5 µm, 150×10 mm i.d.). Analytical detection of cylindrospermopsin was performed using a Waters 2690 separations module and a 996 photodiode array detector. Separation was achieved using a method modified from that of Harada et al. [5] employing a Cosmosil C18 column (Phenomenex; 5 µm, 150×4.6 mm i.d.) with a linear gradient of 1–12% (v/v) methanol/water over 24 min at 40°C. Chromatograms were monitored at 262 nm with spectra from 200 to 300 nm.

2.2 Effect of solvents on the UV absorbance and HPLC-PDA analysis of cylindrospermopsin

Aliquots of purified cylindrospermopsin (10 µg) were dispensed into glass vials and dried down under N₂ gas at 50°C. These were then resuspended to 10 µg ml⁻¹ with solutions of increasing methanol concentration in 10% increments, from 100% (v/v) MilliQ water up to 100% (v/v) methanol (HPLC grade; Rathburn, Walkerburn, UK). The absorbance of these solutions was measured in quartz cuvettes by a UV spectrophotometer at 262 nm (Pharmacia LKB Ultrospec II, Uppsala, Sweden), blanked against the appropriate concentration of methanol. These same solutions were then analysed by HPLC-PDA as described above. After analysis, the samples were dried down and resuspended in methanol/water to the same concentration in the reverse solvent direction and analysed again by HPLC-PDA. Further 10-µg aliquots were prepared and dried down. These were resuspended as for the methanol solutions, except that methanol was replaced by HPLC grade acetonitrile (Rathburn). These solutions were analysed by HPLC-PDA, dried down, resuspended in the reverse solvent order to the same concentration and again analysed. Finally, 10-µg aliquots were dissolved to a concentration of 10 µg ml⁻¹ in either 100% (v/v) MilliQ water, 50% (v/v) aqueous methanol or 100% (v/v) methanol in glass vials. The contents of the vials were then removed and placed in fresh glass vials and 100 µl of water were added to the original vials. After vortexing, the contents were removed and all solutions were analysed by HPLC-PDA (n=3).

2.3 Effect of tap and eutrophic environmental waters on the HPLC analysis of cylindrospermopsin

Dundee municipal tap water and environmental water from eutrophic lakes (Lochs Rescobie and Balgavies, and Monikie Island Pond, Angus, UK) were filtered through GF/C filters (Whatman, Maidstone, UK) and spiked with purified cylindrospermopsin at concentrations of 10, 5, 2, 1 and 0.5 µg ml⁻¹. All solutions were analysed by HPLC-PDA (n=3).

2.4 SPE procedure for the recovery of cylindrospermopsin from culture medium and environmental waters

For the recovery of cylindrospermopsin from spent culture medium and spiked water, stacked SPE cartridges were used with C18-packed material (Jones Chromatography, Mid Glamorgan, UK) placed in series before Hypersep Hypercarb SPE cartridges (Thermo Hypersil-Keystone, Runcorn, Cheshire, UK). For all samples, the cartridges were primed with 100% methanol and washed with 100% water at flow rates specified by the manufacturer. The samples were applied after priming at a flow rate of 1 ml min⁻¹ and after application, were eluted from the solid phases. For cylindrospermopsin isolated from cell-free spent culture medium, the toxin was extracted by the addition of 10 ml of 100% (v/v) methanol or 10% incremental steps of methanol in water from 0 to 100% (v/v). For the elution of extracted cylindrospermopsin from spiked environmental waters, 100% (v/v) methanol plus 0.1% (v/v) trifluoroacetic acid was used. Once extracted, the solvents were dried down and resuspended in MilliQ water before HPLC-PDA analysis.

3 Results

The absorbance at 262 nm of cylindrospermopsin prepared in methanol increased to a maximum, with increasing methanol concentration up to 50% v/v (Fig. 1A), above which absorbance remained approximately constant. Polynomial regression analysis (second and fourth order) showed good correlation with R² values of at least 0.94. HPLC-PDA analysis of these same solutions showed the opposite trend with a decrease in peak height, with the subsequent indicated cylindrospermopsin concentration reaching a minimum at 50% (v/v) methanol which continued to 100% (v/v) methanol (Fig. 1B). However, after these solutions were dried down and resuspended in the reverse solvent order, the opposite trend was apparent, indicating that the higher the methanol concentration, the lower the HPLC response (Fig. 1B). The effect of methanol on analysis by HPLC was compared to the effect of acetonitrile (Fig. 1C). Again, the same response occurred although concentrations of 30% (v/v) acetonitrile were required to obtain the same response as with 50% methanol. Finally, to ascertain whether the observations with organic solvents were due to precipitation of cylindrospermopsin within the glass vial, new vials containing dried cylindrospermopsin were resuspended in water, 50% (v/v) methanol and 100% (v/v) methanol (Fig. 1D). Once resuspended, the solutions were removed and water was added to all vials to dissolve potentially precipitated cylindrospermopsin. With the initial re-dissolving, a decrease in the absorbance by HPLC was observed, although when the vials were rinsed with water and absorbances determined, the observed toxin concentrations for all three re-dissolution treatments were the same, indicating that precipitation of cylindrospermopsin from solution had not taken place.

1

Effects of solvents on the analysis of cylindrospermopsin. Cylindrospermopsin resuspended in methanol from 0 to 100% (v/v) methanol was analysed by a spectrophotometer (A) at 262 nm with second order (—) and fourth order (—–) polynomial regression. The cylindrospermopsin solutions were also analysed by HPLC-PDA, resuspended in methanolic (B), and acetonitrile-containing (C) solutions, resuspended from 0 to 100% (v/v) solvent (█), or dried and again resuspended from 100% to 0% (v/v) organic solvent (•). D: To rule out the possibility of precipitation in organic solvent, cylindrospermopsin was redissolved in 0, 50 and 100% (v/v) methanol (█), supernatant removed and the MilliQ water added to the vials (•). All samples were analysed by HPLC-PDA (n=3). Vertical error bars represent standard deviation.

For subsequent investigations into the analysis of cylindrospermopsin in eutrophic waterbodies, filtered lake water, with tap water for comparison, was spiked with the purified toxin to final concentrations of between 0.5 and 10 µg ml⁻¹ (Table 1). The HPLC response for tap water and untreated Loch Rescobie water was good, with recovered cylindrospermopsin concentrations similar to the spike concentration. The recovered toxin concentrations for spiked waters from Monikie were lower than the spike concentrations, although the recovery was at least 86%, indicating that environmental waters did not interfere greatly with the HPLC analysis of cylindrospermopsin.

1

Mean recoveries of cylindrospermopsin from tap and environmental eutrophic waters spiked with toxin at concentrations between 0.5 and 10 µg ml⁻¹ as analysed by HPLC (n=3)

Spike concentration (µg ml−1)	Tap water	Monikie Island Pond	Loch Rescobie
10	10.14±0.07	8.98±0.08	10.26±0.04
5	5.02±0.04	4.68±0.05	5.21±0.00
2	1.95±0.00	1.80±0.01	2.07±0.00
1	0.96±0.02	0.87±0.01	1.05±0.02
0.5	0.47±0.02	0.43±0.00	0.55±0.00

Values are means±S.D.

HPLC analysis of actively growing cultures of C. raciborskii strain CR3 has indicated that at least 60% of the total detectable cylindrospermopsin is present in the extracellular growth medium (data not shown). Methods were developed to recover this abundant source of toxin. The combination of SPE using C18, followed by Hypercarb cartridges, facilitated the isolation of cylindrospermopsin, and HPLC-PDA analysis indicated that a large proportion of the impurities were removed by the C18 packing. Application of 4 mg of cylindrospermopsin to this combined setup resulted in association of cylindrospermopsin with both C18 and Hypercarb (Fig. 2A) with cylindrospermopsin quantitative ratios of 1:80 respectively, according to HPLC-PDA analysis. Elution of the Hypercarb cartridges with volumes of methanol increasing by 10% increments resulted in the release of cylindrospermopsin between 0 and 50% (v/v) with a maximal concentration eluted at 30% (v/v) methanol. Due to the high concentrations of cylindrospermopsin present in the spent medium, the use of modifiers for solid phase elution was not necessary, as other compounds eluted from the cartridges did not interfere with the HPLC analysis of cylindrospermopsin. However, the use of this system for the recovery of cylindrospermopsin from environmental eutrophic waters required method development to permit the recovery of low concentrations of the toxin. In order to elute these low concentrations, and allow sample concentration, 100% methanol supplemented with 0.1% (v/v) TFA was found to be the most suitable. When cylindrospermopsin isolated from environmental water was eluted from the Hypercarb cartridge, 8 ml of elution solvent were required (Fig. 2B). In some fractions, interfering compounds prevented the detection of cylindrospermopsin by HPLC-PDA, although this was only a problem for the second millilitre of solvent applied (Fig. 2B). When the cylindrospermopsin was eluted as one pool using 100% methanol/0.1% TFA, good recoveries were found (Table 2). In all cases, full recovery was achieved for environmental waterbody samples spiked to 1.14 µg l⁻¹.

2

SPE of cylindrospermopsin from spent extracellular growth medium after batch culture of C. raciborskii to late growth phase (A) and spiked environmental eutrophic waters (B) using combined C18, followed in series by polygraphite carbon cartridges. A: Extracellular growth medium containing cylindrospermopsin (grey bar) was passed through the SPE system and eluted from C18 (white bar) and polygraphite carbon (hatched bars) before analysis by HPLC-PDA. B: Environmental water samples from Lochs Balgavies (white bar) and Rescobie (light grey bar) and Monikie Island Pond (dark grey bar) spiked with cylindrospermopsin at 1 µg l⁻¹ were eluted with 1 ml incremental volumes of 100% (v/v) methanol+0.1% (v/v) TFA and analysed by HPLC-PDA. Asterisks represent undetected cylindrospermopsin due to the presence of interfering compounds.

2

Recovery of cylindrospermopsin from three raw waters obtained by SPE (n=2) and analysis by HPLC-PDA

Source	Spike concentration (µg l−1)	Recovered concentration (µg l−1)	Spike recovery (%)
Loch Balgavies	1.14	1.17	103
Loch Rescobie	1.14	1.22	107
Monikie Island Pond	1.14	1.28	112

4 Discussion

In comparison to the microcystins, methods for the recovery and analysis of cylindrospermopsins of cyanobacteria have received little attention. The early method of Harada et al. [5] was a rapid screening procedure involving four steps: extraction, cleanup, separation and determination. The determination was performed by HPLC-PDA and cleanup involved the use of HP resin, after which the cylindrospermopsin was eluted and applied to a C18 cartridge to remove lipophilic compounds. However, the use of resins and C18 separately is more time-consuming and more complicated than in the present study and the unit costs are more expensive. Since the Harada et al. study [5], other methods of SPE have been used to isolate and concentrate this toxin. In particular, Norris et al. [13] investigated a wide range of commercially available SPE sorbents over a wide range of pH values. Of these, graphitised carbon was found to recover cylindrospermopsin from spent growth medium when analysed by HPLC-MS/MS. However, as the graphitised carbon cartridge was used on its own, there was a greater potential for breakthrough with small applied volumes, compared to a combined system having a C18 cartridge to remove lipophilic compounds [5] and other components which may interfere with the HPLC-PDA analysis of cylindrospermopsin.

The ability to recover and concentrate cylindrospermopsin from environmental waters can help to accurately monitor toxin concentrations when sensitive analytical methods such as HPLC-MS/MS, which do not require sample concentration, are not available. The use of C18 followed by Hypercarb cartridges in series permits other organic substances to be removed in one step, allowing the HPLC detection of cylindrospermopsin after the removal of a large number of other organic compounds that may be present in environmental waters. As reported by Norris et al. [13], breakthrough of cylindrospermopsin from graphite cartridges was found to occur after 4–5 l of medium had been passed through the cartridges. However, by preceding the graphite cartridges with C18 cartridges, we found that the volume of spent growth medium through the cartridges could be greatly increased. At least 10 l of spent medium could be passed through without breakthrough (data not shown), although the maximum volume would be expected to depend on the concentration of total organic compounds within the sample for extraction. Organic products in the spent medium might have the capacity to displace cylindrospermopsin from the cartridge. If so, a two-stage system might have the advantage of preventing these compounds from interfering with cylindrospermopsin for potential binding sites within the carbon cartridge, therefore permitting a larger volume of medium to be passed through the system and a greater amount of cylindrospermopsin to be retained.

In order to determine the concentration of cylindrospermopsin in environmental waters, a variety of solvents may be employed in chromatographic methods such as HPLC-PDA. When cylindrospermopsin was redissolved in a range of solvent concentrations, a loss of the HPLC-PDA absorbance signal occurred at concentrations greater than 50% methanol and 30% acetonitrile. Organic solvents have been found to influence the detection and quantification of microcystin-LR [16]. Methanol is widely used to extract microcystins, 70–75% methanol being optimal [17,18]. Moreover, the UV absorbance of microcystin-LR is maximal in 70% (v/v) methanol when determined by HPLC-PDA and UV spectrophotometry. For cylindrospermopsin, absorbance in the UV spectrophotometer at high methanol concentrations has higher absorbance at 262 nm, unlike the response by HPLC detection. As precipitation of cylindrospermopsin was not observed with high solvent concentrations, the loss of the HPLC signal indicates that the HPLC procedure is involved in the reduced response visualised with cylindrospermopsin. The use of C18 analytical columns with cylindrospermopsin dissolved in organic solvents at high concentrations may result in dissociation or association of cylindrospermopsin molecules due to their planarity resulting in the reduced responses visualised. Furthermore, retention time differences which could be attributed to poor retention with increasing organic solvent concentrations were not observed.

The concentration of cylindrospermopsin used for environmental raw water spiking and achieved by analysis in this study (∼1 µg l⁻¹) is in the range advocated as a potential guideline value for the protection of human health [19] and well below the limits of detection for HPLC-PDA (400 µg l⁻¹; 25-µl injection, no sample pre-concentration). Further assessment of this SPE system together with HPLC-PDA is necessary to assess the lower limits of detection and linearity for cylindrospermopsin analysis in diverse raw and treated waters that may contain a wide range of total organic carbon concentrations, which could interfere with this system for environmental use.

Acknowledgements

We thank Dr. G.K. Eaglesham for a purified cylindrospermopsin standard and Prof. R.A. Herbert for the kind loan of the tangential flow equipment. We also thank the European Union for financial assistance (CYANOTOX project ENV4-CT-98).