Extraction of Cyanobacterial Endotoxin

John Papageorgiou, Thomas A. Linke, Con Kapralos, Brenton C. Nicholson,Dennis A. Steffensen

CRC for Water Quality and Treatment, Australian Water Quality Centre, Private Mail Bag 3,Salisbury, South Australia, 5108, Australia

Received 16 April 2003; revised 6 November 2003; accepted 7 November 2003

ABSTRACT: To simplify our efforts in acquiring toxicological information on endotoxins produced by cya-nobacteria, a method development study was undertaken to identify relatively hazard-free and efficientprocedures for their extraction. One article sourced and two novel methods were evaluated for their ability toextract lipopolysaccharides (LPSs) or endotoxins from cyanobacteria. The Limulus polyphemus amoebocytelysate (LAL) assay was employed to compare the performance of a novel method utilizing a 1-butanol–water(HBW) solvent system to that of Westphal’s (1965) phenol–water system (HPW) for the extraction of endotoxinfrom various cyanobacteria. The traditional HPW method extracted from 3- to 12-fold more endotoxin from sixdifferent cyanobacterial blooms and culture materials than did the novel HBW method. In direct contrast, thenovel HBW method extracted ninefold more endotoxin from a non–microcystin producing Microcystis aerugi-nosa culture as compared to the HPW method. A solvent system utilizing N,N�-dimethylformamide–water(HDW) was compared to both the HPW and HBW methods for the extraction of endotoxin from naturalsamples of Anabaena circinalis, Microcystis flos-aquae, and a 1:1 mixture of Microcystis aeruginosa/Micro-cystis flos-aquae. The LAL activities of these extracts showed that the novel HDW method extracted two- andthreefold more endotoxin from the Anabaena sample that did the HBW and HPW methods, respectively. TheHDW method also extracted approximately 1.5-fold more endotoxin from the Microcystis flos-aquae sampleas compared to both the HBW and HPW methods. On the other hand, the HBW method extracted 2- and14-fold more endotoxin from the Microcystis flos-aquae/Microcystis aeruginosa mixture than did the HPW andHDW methods, respectively. Results of this study demonstrate that significant disparities exist between thephysicochemical properties of the cell wall constituents not only of different cyanobacterial species but also ofdifferent strains of the same cyanobacterial species, as showing by the varying effectiveness of the solventsystems investigated. Therefore, a sole method cannot be regarded as universal and superior for the extractionof endotoxins from cyanobacteria. Nevertheless, the ability of the novel HBW and HDW methods to utilizeeasily handled organic solvents that are less hazardous than phenol render them attractive alternatives to thestandard HPW method. © 2004 Wiley Periodicals, Inc. Environ Toxicol 19: 82–87, 2004.

Keywords: lipopolysaccharide (LPS); endotoxin; extraction; cyanobacteria

INTRODUCTION

Lipopolysaccharides (LPSs), or endotoxins, as they arecommonly referred to, are major components of the cellwall in most gram-negative bacteria, marine pseudomonads,and cyanobacteria (Buttke et al., 1975). LPSs usually con-sist of a polysaccharide backbone comprising an O-specificoligosaccharide chain and a relatively small oligosaccharide(core region), which, in turn, is covalently bonded to a lipid

Correspondence to: John Papageorgiou; e-mail: john.papageorgiou@sawater.com.au

Contract grant sponsor: American Water Works Association ResearchFoundation.

Contract grant sponsor: Cooperative Research Centre for Water Qualityand Treatment.

Published online in Wiley InterScience (www.interscience.wiley.com).DOI 10.1002/tox.10152

© 2004 Wiley Periodicals, Inc.

82

unit known as lipid A (Sykora et al., 1980). It is wellestablished that LPSs of gram-negative organisms (e.g.,Escherichia coli) are responsible for the severe pathophys-iological effects (fever, trauma, multiorgan failure, and sep-tic shock) observed following infection of a host (Galanos etal., 1985; Holst et al., 1996). Consequently, much work hasbeen undertaken on the elucidation of the structure andbiological significance (with respect to toxicity) of LPSsderived from gram-negative bacteria. In direct contrast,limited information exists on the toxicological propertiesand effects of LPSs produced by cyanobacteria that are ofincreasing concern to water and health authorities through-out the world.

The first step involved in the process of determining thetoxicological properties of LPSs requires that they be sep-arated and isolated from other cell-wall constituents. Therehas been significant work done to develop methods for theisolation and purification of LPSs from bacteria (predomi-nantly enterobacteria). To date, 18 LPS extraction methodsare available, and no single method is universally applicable(Ridley et al., 2000). Nevertheless, Westphal’s (1965) hotphenol–water (HPW) method and Galanos’s (1969) hotphenol–chloroform–petroleum ether (HPCP) method arecommonly used for the extraction of LPSs from gram-negative bacteria. In direct contrast, no articles were foundthat address the extraction of cyanobacterial LPSs usingsolvent systems other than those of the HPW method and, toa lesser extent, the HPCP method. The potential healthhazard from phenol, which makes significant precautionsnecessary when handling it on a regular basis, prompted usto identify alternative, less hazardous methods for the ex-traction of cyanobacterial endotoxins.

We decided to initially probe the performance of a sub-stitute hot phenol–water method utilizing 1-butanol as theorganic solvent component. 1-Butanol was chosen as thephenol substitute because it has a similar polarity to phenol,as shown by its solubility in water, yet, in direct contrast, itis a liquid that is nontoxic and easily handled over a widetemperature range (0–100°C). In addition, 1-butanol can beremoved from an extraction mixture by simple rotary evap-oration under low vacuum, which is not possible with phe-nol. In this study, hot (65°C) butanol–water (HBW) wascompared to HPW for the extraction of endotoxins (LPSs)from two cultured Microcystis aeruginosa strains (MANand MAP), one cultured Anabaena circinalis strain (AC),two natural Phormidium strains (PTL and PPR), and anatural sample of Microcystis botrys (MB). The Limuluspolyphemus amoebocyte lysate (LAL) assay (Bang et al.,1956; Solum et al., 1970, 1973; Young et al., 1972) wasused to determine the endotoxicity of the extracts obtainedand hence to compare the performance of the extractionmethods.

Another organic solvent commonly employed to dissolverecalcitrant organic compounds in synthetic and analyticalorganic applications is N,N�-dimethylformamide (DMF).DMF is a dipolar aprotic volatile liquid; however, unlike

1-butanol, DMF is soluble in both water and relativelynonpolar solvents at room temperature. Despite DMF beingregarded as moderately toxic but less hazardous than phe-nol, we decided to compare the performance of a DMF–water (HDW) solvent system to that of HBW and HPW forthe extraction of cyanobacterial endotoxin. Natural samplesof Anabaena circinalis (ACN) and Microcystis flos-aquae(MF) and a natural 1:1 mixture of Microcystis aeruginosa/Microcystis flos-aquae (MAMF) were chosen as the repre-sentative cyanobacteria in this study. MAMF was extractedby all three methods in duplicate or triplicate to determinethe coefficient of variation (CV) of each method.

MATERIALS AND METHODS

Equipment

A hot-plate stirrer (RCT basic, IKA, Staufen, Germany)equipped with a precise temperature controller (ETS-D 4fuzzy, IKA, Germany) was used in all extractions. Semipu-rification of the extracts was carried out using Pellicon5–100 K and �100 K molecular-weight cutoff ultrafiltrationcross-flow membranes (Millipore Corporation, Bedford,Massachusetts, USA) in conjunction with a peristalticpump. Gas chromatographic–mass spectrometric (GC-MS)analysis of fatty acid methyl esters was performed using aHewlett Packard (HP) system consisting of a 5890 GC oven,5MS GC column, and a 5971 MS detector.

Chemicals and Chromatography Materials

Kieselgel 60 F254 aluminium–supported thin layer chroma-tography (TLC) plates (Merck, Darmstadt, Germany) wereused in the detection of carbohydrates.

Organic solvents and reagents were of analytical gradeand supplied by Aldrich Chemical Company (St. Louis,Missouri, USA). Milli-Q-water (endotoxin content: � 0.06EU/mL, Millipore Corporation, USA) was used in the prep-aration of aqueous buffers and in experiments requiringwater.

Cyanobacteria

Natural Samples

Samples of a benthic Phormidum sp. were collected fromTorrens Lake, Adelaide, Australia, in January 1999 (PTL)and from Paskeville Reservoir, Yorke Peninsula, SouthAustralia, in January 2000 (PPR). A 1:1 mixture of Micro-cystis aeruginosa/Microcystis flos-aquae was collectedfrom Murray Bridge, South Australia, in January 2002(MAMF), and a sample of Microcystis botrys was collectedfrom Lake Elphinstone, Queensland, Australia, in Septem-ber 2002 (MB). A sample of Anabaena circinalis wascollected from Birdwood Yabbie Farm, Adelaide Hills,

EXTRACTION OF CYANOBACTERIAL ENDOTOXIN 83

Australia, in December 2000 (ACN). A sample of Micro-cystis flos-aquae was collected from the Bolivar TreatmentLagoons, Adelaide, South Australia, in November 2002(MF). All samples were freeze-dried and stored at �20°Cuntil extracted.

Cultured Samples

Australian Water Quality Centre (AWQC) nonaxenic unial-gal cultures of Microcystis aeruginosa strain 338 (MAN),Microcystis aeruginosa strain 309 1 CA (MAP), andAnabaena circinalis 118 AR (AC) were subcultured up to avolume of 10 L using a 1:10 dilution factor. Subculturingtook place approximately every 14 days. Cultures weregrown in ASM media (Gorham et al., 1964) at an irradianceof 25 �mol m�2 s�1 (PAR 400–700 �m) and at a temper-ature of 25°C. Cyanobacterial cells were isolated by cen-trifugation at 10 000 � g for 30 min, freeze-dried, and thenstored at �20°C prior to extraction.

Extraction of Cyanobacterial Endotoxin

Freeze-dried cyanobacteria were used in all extractions.Representative HPW, HBW, and HDW methods for theextraction of cyanobacterial endotoxin on a 1-g sample aredescribed below. Modified versions of these methods(HPW-2, HBW-2, and HDW-2) adopted in the second partof the study for the extraction of endotoxins from naturalsamples of Anabaena circinalis (ACN) and Microcystisflos-aquae (MF) are also described below. Coefficients ofvariation (CVs) of the HPW, HBW and HDW methods werecalculated from the yields of the replicate MAMF freeze-dried extracts obtained.

Phenol/Water (HPW)

To a suspension of freeze-dried cyanobacteria (1 g) in water(20 mL) at 65°C was added a mixture of hot (65°C) phenol(18 mL) and water (2 mL). The suspension was left to stirat 65°C for 30 min. The mixture was then cooled to about5°C and centrifuged for 30 min at 4000 � g. The aqueoussupernatant was separated and the phenol layer re-extractedwith water (20 mL) followed by centrifugation as describedabove. The combined aqueous supernatant was ultrafilteredusing two ultrafiltration cross-flow membranes (5–100 Kand �100 K molecular-weight cutoff cartridges) in con-junction with a peristaltic pump giving two fractions (5–100K and �100 K). Semipure fractions were freeze-dried,which afforded two powders in all cases.

1-Butanol/Water (HBW)

To a suspension of cyanobacteria (1 g) in water (20 mL)was added a mixture of 1-butanol (18 mL) and water (2mL). The suspension was left to stir at 60°C for 30 min. Themixture was cooled to room temperature and centrifuged at

4000 � g for 30 min. The water layer (bottom clear layer)was isolated, and the remaining interphase and butanollayers, together with suspended particles, were diluted withwater (20 mL) and recentrifuged. The combined water ex-tract was ultrafiltered and processed as described for HPWto give two powders in all cases.

N,N�-Dimethylformamide/Water (HDW)

To a suspension of cyanobacteria (1 g) in water (20 mL),was added a mixture of N,N�-dimethylformamide, or DMF(18 mL), and water (2 mL). The suspension was left to stirat 60°C for 30 min. The mixture was cooled to roomtemperature and centrifuged at 4000 � g for 30 min. Thesupernatant was evaporated to dryness (rotary evaporator),and the remaining residue was redissolved in Milli-Q-water(300 mL). The solution was then ultrafiltered and processedas described for HPW to give two powders in all cases.

Phenol/Water (HPW-2), 1-Butanol/Water(HBW-2), and N,N�-Dimethylformamide/Water(HDW-2)

Freeze-dried MF and ACN (2 g each) were extracted induplicate as described for HPW, HBW, and HDW, respec-tively, except that the combined aqueous supernatants(HPW-2 and HBW-2) or aqueous solutions (HDW-2) of theextracts were processed as follows: Aqueous supernatantsand solutions of extracts were ultrafiltered using one ultra-filtration cross-flow membrane (�100 K molecular-weightcutoff cartridges), and what was retained was diluted to afinal volume of 200 mL with Milli-Q-water and stored at4°C until further use.

Analytical Methods

Gravimetric Analysis

All cyanobacteria samples used and extracts obtained werefreeze-dried prior to being weighed.

Qualitative Sugar Analysis

An aliquot (�10 �L) of an aqueous solution of each extractwas spotted on a TLC plate and dried at 70°C for 1 min. TheTLC plate was totally immersed in a solution of 10:1acetone and 15% sulfuric acid for 2 s, removed, dried, andthen further heated to char the spots. Formation of a brownspot confirmed the presence of carbohydrate material.

Qualitative Fatty Acid Analysis

A sample (�5 mg) of each extract was hydrolysed in 6 MHCl (2 mL) for 6 h at 100°C. The reaction mixture wascooled to room temperature and extracted with petroleum

84 PAPAGEORGIOU ET AL.

ether (40°C–60°C fraction). A solution of diazomethane*(0.5 mL) in t-butyl methyl ether (TBME) was added to thepetroleum ether extract (1 mL), and the resulting mixturewas left to stand for 30 min. Silica was then added to thereaction mixture, and the mixture was left to stand foranother 10 min to remove excess diazomethane. An aliquotof each solution was then analysed by GC-MS.

Fatty acid methyl esters were separated (see equipmentand materials for details of system and column) using thefollowing temperature program: 50°C for 1 min, then ramp-ing to 150°C at 15°C/min, then ramping to 300°C at 25°C/min and holding at this temperature for 10 min. Fatty acidmethyl esters were detected by MS in electron impact (EI)mode, where the MS interface temperature was 230°C andthe source temperature was 150°C.

Endotoxicity Assay

Assays were performed according to the following proce-dure: Weighed samples (1-10 mg) of freeze-dried semi-purified fractions were dissolved in water (1 mL) and sub-jected to Limulus polyphemus amoebocyte lysate (LAL)assay (Bang et al., 1956; Solum et al., 1970, 1973; Young etal., 1972). In the case where semipurified extracts weredirectly generated in solution form (HPW-2, HBW-2, andHDW-2 methods), aliquots (1 mL) of the extracts weresubjected to the LAL assay without further manipulation.Each assay was performed in duplicate using a BioWhit-taker QCL-1000 Quantitative Chromogenic LAL kit (Cat.No. 50-647U, BioWhittaker, MD, USA) following the mi-croplate method provided with the kit, which is describedbelow.

Endotoxin activity was measured against a 4-log stan-dard curve from 0.1 to 1.0 endotoxin units per milliliter(EU/mL) of E. coli. standard LPS (prepared from E. coli.

055:B5, referenced against the United States Pharmacopeia(USP) standard endotoxin). Assays were performed at 37°Cin 96-well sterile microtiter plates. To each well were added50 �L of water and either the standard or sample, followedby 50 �L of the LAL solution (enzyme). The mixtures wereheld at 37°C for 10 min, and then 100 �L of the substratesolution (chromogen) was added to each well, and theresulting content was held at 37°C for an additional 6 min.Then 100 �L of a stop reagent (stops the action of theactivated LAL enzyme and therefore the release of thechromogen) was added to the mixtures. The absorbance ofeach microplate well was read at 405–410 nm (distilledwater was used to zero the spectrophotometer prior toreading the absorbance values). Endotoxicity was given inendotoxin units per milliliter of water (EU/mL), which wasconverted to endotoxin unis per milligram (EU/mg) offreeze-dried semipurified extract for solid samples.

RESULTS AND DISCUSSION

Evaluation of Solvent Systems for Extractionof Cyanobacterial Endotoxin

The first part of this study compared the performance of thenovel HBW and known HPW methods for the extraction ofendotoxins from six cyanobacterial species/strains. Table Ishows the amounts and yields of the freeze-dried semipu-rified HPW and HBW cyanobacterial extracts obtained.Between 150 and 1000 mg of material was extracted in eachexperiment, depending on availability.

The data in Table I indicate that the HBW methodextracted approximately 1- to 3-fold more material in total(sum of 5–100 K and �100 K yields) from all the cya-nobacteria samples examined as compared to the HPWmethod. Table II shows the LAL activities of the HBW andHPW semipurified cyanobacterial extracts in endotoxinunits per milligram of freeze-dried extract (EU/mg), to-gether with total endotoxin activity (EU) in the �100 Kmolecular weight extracts.

The data in Table II indicate high endotoxin content inboth HPW and HBW � 100 K semipurified molecularweight fractions. Of the HPW and HBW 5–100 K fractions,

*Preparation of diazomethane: A solution (8 mL) of N-methyl-N-nitroso-p-toluenesulfonamide (p-TSN) in anhydrous TBME (20% w/w ofp-TSN in TBME) was slowly added (dropwise) to a methanolic potassiumhydroxide solution (20 mL, 10% w/w of KOH in MeOH) at 40°C undernitrogen. The reaction mixture was stirred until an opaque yellow colorformed, which indicated that diazomethane was generated. Nitrogen wasthen bubbled through the reaction mixture in order to transfer diazometh-ane via a tube (immersed in TBME) into a glass bottle containing TBME(100 mL). The solution was sealed and stored in the freezer until use.

TABLE I. Amounts and yields of freeze-dried semipurified cyanobacterial extracts

Cyanobacteria

Amount (mg) and Yield (% w/w)*

HBW (5–100 K) HPW (5–100 K) HBW (100 K�) HPW (�100 K)

MAN 12.6 (3.19) 7.4 (1.59) 47.2 (11.98) 28.6 (6.16)MAP 20.2 (4.52) 22.4 (6.55) 39.7 (8.88) 17.4 (5.09)AC 12.2 (7.79) 9.0 (5.79) 18.7 (11.93) 16.8 (10.80)MB 23.3 (4.76) 5.5 (1.19) 33.0 (6.74) 10.5 (2.27)PPR 10.8 (1.29) 12.2 (1.09) 20.9 (2.49) 14.7 (1.28)PTL 10.1 (1.02) 9.3 (0.88) 10.7 (1.03) 9.6 (0.91)

* Values in brackets are percentage yields (weight of freeze-dried extract/weight of freeze-dried cyanobacteria � 100) of extracts obtained.

EXTRACTION OF CYANOBACTERIAL ENDOTOXIN 85

only those from AC exhibited relatively high endotoxicity.Qualitative carbohydrate analysis of the extracts showedthat all the �100 K fractions were relatively rich in carbo-hydrate material. Chemical hydrolysis of the �100 K ex-tracts liberated both sugars and fatty acids, the latter de-tected by GC-MS of their methyl esters, hence confirmingthe presence of LPSs, which most likely are responsible forthe observed endotoxicity. It was envisaged that these ex-tracts were contaminated with negligible amounts of gram-negative bacterial-derived LPSs because none of their typ-ical fatty acid profiles (dominant in 3-hydroxylated C-10,C-12, and C-14 acids) were detected (Rietschel et al., 1984).Only palmitic and/or stearic acid dominated these extracts,which are also main components of LPSs from other cya-nobacteria (Weckesser et al., 1979). That only the HBW andHPW 5-100 K extracts from AC were significantly endo-toxic, as compared to the analogous extracts obtained fromthe other cyanobacteria samples, implies that LPS from ACexists in both high- (aggregate) and low-molecular-weightforms, which is a common property of gram-negative bac-terial LPSs (Mueller-Loennius et al., 1998).

Total endotoxin activity values indicate that the novelHBW method extracted approximately ninefold more endo-toxin from the MAN strain than the HPW method. Incontrast, the HPW method extracted 3.3-, 3.3-, 12.4-, 3.7-,and 9.9-fold more endotoxin from MAP, AC, MB, PPR andPTL, respectively. The significant variation in performancebetween the two methods more than likely reflects thevariation in physicochemical properties and LPS content of

the cell walls of different cyanobacteria strains and species,which is also apparent in gram-negative bacteria (Ridley etal., 2000).

The second part of this study compared the performancesof the HPW, HBW, and HDW methods for the extraction ofendotoxins from three cyanobacterial bloom materials.

A preliminary assessment of the reproducibility of themethods was included. Table III shows the yields of repli-cate HPW, HBW, and HDW extracts from MAMF, includ-ing the coefficient of variation (CV) of each method.

The CV of each method indicates that the HBW methodis the most reproducible, followed by the known HPWmethod and then the HDW method. It is expected that thereproducibility of all three methods would significantlyimprove over a larger number of replicate extractions car-ried out on a near-identical scale with respect to each other.

Table IV shows the endotoxicity of HPW, HBW, andHDW � 100 K extracts from ACN and MF, in endotoxinunits per milliliter of aqueous extract (EU/mL), and ofMAMF, in endotoxin units per milligram of freeze-driedextract (EU/mg). As described in the experimental section,modified versions of the three extraction methods (HPW-2,HBW-2, and HDW-2) were employed to reduce the turn-around time for LAL analysis of the ACN and MF extracts.

As observed for the extracts in Table II, all �100 Kmolecular weight endotoxin–rich fractions comprised bothcarbohydrate and fatty acid (dominant in palmitic and/orstearic acid) material, suggesting that LPSs were responsi-ble for the observed endotoxicity. The data showed that the

TABLE II. Endotoxicity of freeze-dried semipurified extracts from various cyanobacteria, as measured by LAL

Cyanobacterium

LAL Activity (EU/mg) Total LAL Activity (EU)

HBW(5–100 K)

HPW(5–100 K)

HBW(�100 K)

HPW(�100 K)

HBW(�100 K)

HPW(�100 K)

MAN 23.20 nd 1.09 � 106 2.01 � 105 5.15 � 107 5.75 � 106

MAP 5.40 33.2 7.26 � 103 5.43 � 104 2.88 � 105 9.45 � 105

AC 1.40 � 104 1.97 � 104 6.83 � 104 2.48 � 105 1.28 � 106 4.17 � 106

MB 845 613 7.17 � 104 2.81 � 106 2.37 � 106 2.95 � 107

PPR 41.9 nd 4.55 � 104 2.39 � 105 9.51 � 105 3.51 � 106

PTL nd 173 1.05 � 105 1.16 � 106 1.12 � 106 1.11 � 107

nd: Not done.

TABLE III. Yields of replicate HPW, HBW, and HDW semipurified extracts obtained from MAMF together with theCV of each extraction method

ReplicateExtraction

Yield of Freeze-Dried Extracts (% w/w)*

HBW5–100 K

HBW�100 K

HPW5–100 K

HPW�100 K

HDW5–100 K

HDW�100 K

1 nd 7.89 0.88 5.28 3.0 26.62 2.78 6.86 0.82 3.44 1.78 15.13 2.24 5.56 1.42 5.22 nd ndMean (%) 2.51 6.77 1.04 4.65 2.39 20.85CV (%) 15.1 17.3 31.7 22.6 35.6 39

* Percentage yield of extracts calculated in weight of freeze-dried extract/weight of freeze-dried cyanobacteria � 100.

86 PAPAGEORGIOU ET AL.

novel HDW method extracted 2.1- and 2.9-fold more endo-toxin from the ACN sample compared with the novel HBWmethod and the standard HPW method, respectively. Sim-ilarly, the HDW method extracted 1.3- and 1.5-fold moreendotoxin from the MF sample compared with the HBWand HPW methods, respectively. Interestingly, the HBWmethod produced an extract from MAMF that was 2.3- and14.1-fold more endotoxic than those obtained from theHPW and HDW methods. Once again, the significant vari-ation observed in the performance of all three methods overthe range of cyanobacteria studied highlights the vast dif-ferences in the physicochemical properties of the cell wallsof different cyanobacteria strains and species and the easewith which LPSs are extracted from this material withvarious solvents.

In conclusion, the results from these investigations dem-onstrate that the performance of the novel HDW methodand, to a lesser degree, the novel HBW method was com-parable to that of the established hot phenol–water (HPW)procedure. However, the significant variation observed inthe performances of all three methods with particular cya-nobacteria indicates that neither of the methods investigatedcan be regarded as universal for the extraction of cyanobac-terial LPSs. Nevertheless, that DMF and 1-butanol are eas-ier to handle and far less hazardous than phenol make boththe HBW and HDW methods especially attractive for theextraction of endotoxins from multiple samples of cya-nobacteria. It is envisaged that application of both the HDWand HBW methods will expedite our efforts to investigatethe endotoxicological effects of cyanobacterial LPSs.

The authors sincerely thank Dr. Andrew Humpage, PeterBaker, and Dr. Peter Hobson, from the Australian Water QualityCentre (AWQC), Adelaide, South Australia, for providing cya-nobacteria material, and Karen Simpson, also from the AWQC, forperforming GC-MS analyses. Kerry Munroe, from the Institute ofMedical and Veterinary Science, Adelaide, South Australia, is alsoacknowledged for performing LAL analyses.

REFERENCES

Bang FB. 1956. A bacterial disease of Limulus polyphemus. BullJohns Hopkins Hosp 98:325.

Buttke TM, Ingram LO. 1975. Comparison of lipopolysaccharidesfrom Agmenellum quadruplicatum to Escherichia coli and Sal-monella typhimurium by using thin-layer chromatography. JBacteriol 124:1566–1573.

Galanos C, Luderitz O, Westphal O. 1969. A new method for theextraction of R-lipopolysaccharides. Eur J Biochem 9:245–249.

Galanos C, Luderitz O, Rietschel Eth, Westphal O, Brade H, BradeL, Freudenberg M, Schade U, Imoto M, Yoshimura H, Ku-sumoto S, Shiba T. 1985. Synthetic and natural Escherichia colifree lipid A express identical endotoxic activities. Eur J Bio-chem 148:1–5.

Gorham PR, McLachlan J, Hammer UT, Kim WK. 1964. Isolationand culture of toxic strains of Anabaena flos-aquae (Lyngb.). deBreb Verh internat Verein Limnol 15:796–804.

Holst O, Ulmer AJ, Brade H, Flad HD, Rietschel Eth. 1996.Minireview: biochemistry and cell biology of bacterial endotox-ins. FEMS Immunol Med Microbiol 16:83–104.

Muller-Loennies S, Zahringer U, Seydel U, Kusumoto S, UlmerAJ, Rietschel ET. 1998. What we know and don’t know aboutthe chemical and physical structure of lipopolysaccharide inrelation to biological activity. Prog Clin Biol Res 397:51–72.

Ridley BL, Jeyaretnam BS, Carlson RW, 2000. The type and yieldof lipopolysaccharide from symbiotically deficient Rhizobiumlipopolysaccharide mutants vary depending on the extractionmethod. Glycobiol 10:1013–1023.

Rietschel Eth, Wollenweber H, Russa R, Brade H, Zahringer U.1984. Concepts of the chemical structure of lipid A. Rev InfectDis 6:432–438.

Solum NO. 1970. Some characteristics of the clottable protein ofLimulus polyphemus blood cells. Thromb Diath Haemorrh 23:170.

Solum NO. 1973. The coagulagen of Limulus polyphemus hemo-cytes. A comparison of the clotted and non-clotted forms of themolecule. Thromb Res 2:55.

Sykora JL, Keleti G, Roche R, Volk DR, Kay GP, Burgess RA,Shapiro MA. 1980. Endotoxins, algae and Limulus Amoebocytelysate test in drinking water. Water Res 14:829–839.

Weckesser J, Drews G. 1979. Lipopolysaccharides of photosyn-thetic prokaryotes. Annu Rev Micrbiol 33:215–239.

Westphal O, Jann K. 1965. Bacterial lipopolysaccharides: Extrac-tion with phenol–water and further applications of the proce-dure. Methods in Carbohydrate Chemistry 5:83–91.

Young NS, Levin J, Prendergast RA. 1972. An invertebrate coag-ulation system activated by endotoxin: evidence for enzymaticmechanism. J Clin Invest 51:1790.

TABLE IV. Endotoxicity of >100 K HPW, HBW, andHDW extracts from natural cyanobacterial bloomsamples

Cyanobacteria

LAL Activity (EU/mL)

HBW HPW HDW

ACN 1.15 � 103 841 2.43 � 103

MF 7.58 � 103 6.28 � 103 9.53 � 103

MAMF 3.89 � 105* 1.69 � 105* 2.76 � 104*

* Endotoxicity values are in EU/mg.

EXTRACTION OF CYANOBACTERIAL ENDOTOXIN 87