Cyanobacterial heterocyst glycolipids in cultures and environmental samples: Diversity and biomarker potential

Cyanobacterial heterocyst glycolipids in cultures and environmental samples: Diversity

and biomarker potential

Lars Wormer,a,* Samuel Cires,b David Velazquez,b Antonio Quesada,b and Kai-Uwe Hinrichs a

a Organic Geochemistry Group, Zentrum fur Marine Umweltwissenschaften (MARUM), University of Bremen, Bremen, GermanybDepartamento de Biologıa, C. Darwin, 2, Universidad Autonoma de Madrid, Madrid, Spain

Abstract

The envelope of cyanobacterial heterocysts contains characteristic glycolipids termed heterocyst glycolipids(HGs). Six cultures of heterocyst-forming cyanobacteria, as well as samples from 2 polar ecosystems and 12Spanish freshwaters, were analyzed for their HG content. We tentatively identified four novel HGs in thecyanobacterial strains, including derivatives with a nonhexose sugar moiety. We observed unexpectedly diverseHG distributions in cultures, with very singular fingerprinting for Aphanizomenon aphanizomenoides andAphanizomenon ovalisporum, thus supporting the taxonomic robustness of HGs. The analysis of HGs inenvironmental samples revealed even larger diversity and increased abundance of certain HGs that are minorcomponents in cultured strains. Interpretation of HG patterns could be applied to support taxonomiccharacterization of the community of heterocystous cyanobacteria and to detect changes in the cyanobacterialconsortia mediating N2 fixation. In addition to taxonomy, environmental factors seem to influence thebiosynthesis of HGs. For example, in our studied environmental samples, temperature appears to influence therelative distribution of keto-hexacosanol and hexacosanediols, as previously proposed for cultures. Our resultsextend the knowledge of HG distributions in cyanobacteria and the environment and illustrate the potential ofHGs as biomarkers for applications in both ecological and paleoenvironmental studies.

Cyanobacteria played an important role in the history oflife on Earth. The first potential evidence is 3.5-Gyr-oldfossils from Apex Chert (Australia) with a morphology thatresembles cyanobacteria (Schopf 2000). About a billionyears later, cyanobacteria were responsible for oxygenationof the atmosphere during the Great Oxidation Event(Summons et al. 1999). Due to their broad physiologicalcapabilities these organisms have been able to colonize anenormous variety of environments. Many of these envi-ronments are considered extreme (e.g., polar regions, high-salinity environments, hot deserts, or hydrothermal sourc-es; Whitton and Potts 2000). Still, most cyanobacteriathrive in less extreme environments, as either benthic orpelagic communities. Under certain nutrient conditions,pelagic cyanobacteria may grow massively and form so-called cyanobacterial ‘blooms.’

Some cyanobacterial genera have the capacity toassimilate atmospheric nitrogen (N2). This nitrogen fixa-tion plays a key role in the global N supply and is commonin environments such as eutrophic lakes, sediments ofoligotrophic systems, microbial mats, oceans, or rice fields(Howarth et al. 1988; Karl et al. 1997; Quesada et al. 1998).Nitrogen fixation is catalyzed by the nitrogenase enzymecomplex, which is extremely sensitive to oxygen. Cyano-bacteria developed various strategies to limit the presenceof oxygen during nitrogenase activity, including temporal(Stal and Krumbein 1987) or spatial separation of N2

fixation and O2 production, or even both (Berman-Franket al. 2003). Two cyanobacterial orders, namely Nostocalesand Stigonematales, achieve spatial separation by celldifferentiation into heterocysts that lack photosynthetic

activity while offering an anoxic environment in which N2

is reduced and incorporated as NHz4 .

Heterocysts differ morphologically from vegetative cellsbecause of their unique cell envelope with two additionallayers surrounding the outer cell membrane. The firstpolysaccharide layer is mainly thought to offer mechanicalsupport and protect the second glycolipid or ‘laminated’layer, which is responsible for limiting oxygen diffusioninto the cell (Nicolaisen et al. 2009). This laminated layeris formed by unique glycolipids (i.e., so-called ‘heterocystglycolipids’ [HG]; Nichols and Wood 1968). The maincomponents of this compound class were structurallycharacterized by Bryce et al. (1972) and Lambein andWolk (1973) in Anabaena cylindrica. Soriente et al. (1993)reported that these lipids also occur in akinetes ofCyanospira rippkae, after observing them in an akinetesuspension obtained after centrifugation of an old culture.This report raises the possibility that HGs may not beexclusive to heterocysts but may also occur in akinetes inheterocystous cyanobacteria.

Studies of different cyanobacterial species have expand-ed the range of known HGs. Based on chain length,number, and type of functional moieties (diol, triol, keto-ol,and keto-diol), 10 different compounds have been de-scribed in Nostocales and Stigonematales (Gambacortaet al. 1998; Fig. 1). The length of the n-alkyl chains rangesfrom 26 to 32 C-atoms, and appears to be limited to evencarbon-atom numbers. Functional groups (hydroxyl and/orcarbonyl moieties) are found at the C-3, v-1 and v-3positions, with the most common position of the carbonylgroup at C-3. Different types of glycosidically boundhexoses have been described, including a-glucosides, a-galactosides, b-glucosides, and a-mannosides (Gambacortaet al. 1996, 1998, 1999).* Corresponding author: [email protected]

Limnol. Oceanogr., 57(6), 2012, 1775–1788

E 2012, by the Association for the Sciences of Limnology and Oceanography, Inc.doi:10.4319/lo.2012.57.06.1775

1775

Analysis of HGs focused mainly on cyanobacterialisolates and has allowed a systematic description of themost common HG classes (Bauersachs et al. 2009a). Still,the number of species tested is low compared to the largediversity of heterocystous cyanobacteria, and further workis needed to provide evidence for taxonomic heterogeneityconcerning HG synthesis. First biomarker applications inmicrobial and sedimentary samples were recently reported:HGs were detected in Mediterranean sapropels S1 andS5 of Pleistocene Age, in which they signified N2 fixationin the photic zone during times of sediment formation(Bauersachs et al. 2010). In Early to Middle Eocenesediments from the central Arctic Ocean, HGs suggesteda symbiotic relationship between N2 fixing cyanobacteriaand Azolla species (Bauersachs et al. 2010). Microbial matsfrom the North Sea (Bauersachs et al. 2011) revealed arelatively simple HG profile that consisted of hexacosane-diol and octacosane-triol. The presence of these twocompounds was consistent with the species compositionof the mats, which demonstrated the taxonomic specificityof HGs.

Due to the HGs diagnostic link to N2 fixation bycyanobacteria, this compound group has not only a hightaxonomic value but also strong process-related biomarkerpotential. Given the demonstrated preservation potential ofHGs in the geologic record (Bauersachs et al. 2010), thesecompounds qualify as unique and powerful tools to trackthe activity of N2-fixing cyanobacteria in ancient environ-ments. Existing records of past cyanobacterial N2-fixation

are based on indirect evidence, such as d15N values ofchlorophyll (Sachs and Repeta 1999) or bulk sediment(Haug et al. 1998), while putative cyanobacterial lipidbiomarkers such as 2-methyl hopanoids (Summons et al.1999) provide no distinct signature for N2-fixation.

In this study, we further explored the diversity andextended the range of known HGs in cyanobacterial strainsand in various environments in order to more fully de-velop the biomarker potential. We devised a dedicatedUltra High Performance Liquid Chromatography MassSpectrometry (UHPLC-MS) method, which we applied tothe analysis of HGs in six diazotrophic benthic and pelagiccyanobacterial strains, including three strains of the genusAphanizomenon, of which (to our knowledge) only twoother strains have been characterized (Bauersachs et al.2009a). In a second step, we examined HG distribution ina wide range of cyanobacteria-dominated environmentalsamples. This includes pelagic cyanobacterial communitiesand a complete survey of HG distribution in bloomsamples from 12 freshwater reservoirs in Spain. Addition-ally, microbial mats from both polar regions werecharacterized based on their HG composition. Our studyfurthers the understanding of HG diversity in theenvironment and its correlation with N2 fixation undercertain environmental conditions, while establishing taxo-nomic relationships with specific cyanobacterial species orgenera. Consequently, this study expands the foundationfor future applications of HGs as biomarkers of N2 fixationin both ecological and paleoenvironmental settings; this

Fig. 1. Generic HG structure and structural variations described so far (simplified afterGambacorta et al. 1998). The position of the carbonyl group may vary.

1776 Wormer et al.

includes improved analytical assays for their analysis incomplex matrices.

Methods

Cyanobacterial cultures—All strains were grown inmedia without sources of combined nitrogen. The fourpelagic strains of the Nostocaceae family obtained fromthe Universidad Autonoma de Madrid Culture Collection(UAM) and benthic Nostoc sp. MA4 were grown in BG110

medium (Rippka et al. 1979) and the strain of theRivulariaceae family in a modification of CHU No. 10medium (Chu 1942). Additionally, Aphanizomenon ovali-sporum UAM 290 was also cultivated with sodium nitratein BG11 medium in order to evaluate changes in HGcomposition under nondiazotrophic conditions (see Table 1for details on culture conditions). Cultures were harvestedbefore reaching stationary phase, freeze-dried, and storedfrozen (220uC) until extraction.

Environmental samples—Two microbial mat sampleswere obtained at Byers Peninsula (Livingston Island, SouthShetland Islands, Antarctica; Table 2). The peninsula,60.6 km2 in area, is largely free of ice and snow coverduring almost the entire austral summer season. Thepurple-greenish microbial mat studied here grew on theplateau of the peninsula, in a semipermanent shallow pond(South Western pond [SW]) and was 3–4 mm in thickness.

Detailed characterization of these microbial mats can beobtained from Velazquez et al. (2011). Samples werecollected on 08 December 2006, just after snow and icecover melting (SW I), and on 20 January 2007 (SW II).Additionally, a cyanobacterial biofilm was collected atEllesmere Island (Nunavut, Canada) from the W4 pond.Ellesmere Island is the northern part of the ArcticArchipelago and its 900-m-thick ice fields are a trueremnant of the last continental glaciers that covered muchof North America during the last Ice Age. Thus, the area isessentially a polar desert, with thermal oases that are warmand moist enough to support life. Temperature measure-ments were obtained by automatic temperature dataloggers(Tidbit. Onset) placed on the community throughout thewhole summer season. Samples for microscopic observa-tions and taxonomic determinations were taken with 15-mm or 22-mm inner-diameter metal core samplers. Stonesand coarse sediments were removed, and the samples werekept frozen at 220uC for microscopic determination.Qualitative determinations were carried out by using bothbright-field and epifluorescence microscopy at the fieldcamp immediately after collection. Morphological classifi-cation followed Komarek and Anagnostidis (1989, 1999,2005) and Broady and Kibblewhite (1991).

Samples from cyanobacterial blooms were obtainedfrom 12 freshwater reservoirs across Spain (Table 2).Water-column profiles of environmental parameters (pH,conductivity, dissolved oxygen, and temperature) were

Table 1. Description of cyanobacterial cultures analyzed in the present study.

Order Family Life strategy Growth medium T (uC)Irradiance (mmolphotons m22 s21)

Cylindrospermopsis raciborskiiUAM 520 Nostocales Nostocaceae pelagic BG110

28 20

Aphanizomenon aphanizomenoidesUAM 523 Nostocales Nostocaceae pelagic BG110

28 20

Aphanizomenon gracile UAM 521 Nostocales Nostocaceae pelagic BG110 28 20Aphanizomenon ovalisporum

UAM 290 Nostocales Nostocaceae pelagic BG110 and BG1128 60

Nostoc sp. MA4 Nostocales Nostocaceae benthic BG110 18 11Calothrix sp. MU27 Nostocales Rivulariaceae benthic CHU 10 18 11

Table 2. Description of environmental samples analyzed in the present study.

Sample type Location Coordinates Temperature (uC)

South Western Pond (SW) microbial mat Antarctica, South Shetland Islands 62u379060S 61u429140W 2.3, 4.9W4 pond biofilm Canada, Ellesmere Island 83u009080N 75u029200W 4.5Alcantara pelagic Spain, Tajo Watershed 39u429480N 6u289370W 28.1Brovales pelagic Spain, Guadiana Watershed 38u219190N 6u419590W 29.6Cachamuinas pelagic Spain, Norte Watershed 42u209080N 7u489030W 21.3Cuerda del Pozo pelagic Spain, Duero Watershed 41u519120N 2u449350W 20.9Cueva Foradada pelagic Spain, Ebro Watershed 40u589100N 0u419380W 20.5La Barca pelagic Spain, Norte Watershed 43u189560N 6u189270W 22.6Navalcan pelagic Spain, Tajo Watershed 40u029050N 5u069480W 24.3Nogales pelagic Spain, Guadiana Watershed 38u339300N 6u449030W 25.4Oliana pelagic Spain, Ebro Watershed 42u069280N 1u189080E 20.2, 21.2Peares pelagic Spain, Norte Watershed 42u289170N 7u439220W 21.0Rosarito pelagic Spain, Tajo Watershed 40u069130N 5u189240W 27.0Valuengo pelagic Spain, Guadiana Watershed 38u189160N 6u409010W 28.3

Cyanobacterial heterocyst glycolipids 1777

measured using an YSI 6920 multiprobe, at 20-cm depthintervals. Samples stem from the depth of maximumchlorophyll a concentration, which was established withthe same device; sampling was done with a 5-liter plexiglasswatersampler equipped with a magnetic closing device(Uwitec). Samples were maintained at 4uC, transported tothe lab, and processed immediately. Samples were filteredthrough Whatmann glass-fiber (GF/F) filters (nominalporosity: 0.7 mm), which subsequently were stored at220uC.

Identification and quantification of cyanobacterialgenera in the freshwater samples were achieved in twoways. Fresh samples were left undisturbed overnight inorder to allow buoyant cyanobacteria to concentrate atthe surface to be collected and identified. In some casesphytoplankton samples were additionally fixed with acidLugol’s solution and counted, after settling, under aninverted microscope following Utermohl’s method (Uter-mohl 1958). Cyanobacterial taxa were identified accord-ing to Geitler (1932) and Komarek and Anagnostidis (1989,1999, 2005) and quantified as cell concentrations (cellsmL21) by counting cells under 4003 magnification.Average cell biovolume was calculated for each species byassimilating cells to regular geometric bodies and measur-ing dimensions of $ 100 cells per species, thus obtainingbiovolume of each species per volume of water (mm3 m23).

Sample extraction—Samples were extracted following amodified Bligh and Dyer method (Bligh and Dyer 1959;Sturt et al. 2004). Cell material or GF/F filters wereultrasonically extracted together with 3 g of combustedsea sand in a solvent mixture (v : v) of methanol(MeOH), dichloromethane (DCM), and phosphate buffer(0.05 mol L21; 2 : 1 : 0.8). After centrifugation, the super-natant was collected and the residue extracted two moretimes. Supernatants were pooled and DCM and water wereadded in order to allow optimal phase separation. Aftertransferring the organic layer to a glass vial, the aqueousphase was extracted three more times with DCM. Pooledorganic layers were then washed three times with water andevaporated under N2 flow. The sample was then dissolvedin 0.5 mL DCM : MeOH (9 : 1) and prepared for analysis.

UHPLC-MS analysis—HG detection and identificationwas carried out on a Bruker maXis ultra-high-resolutionquadrupole time-of-flight mass spectrometer (QToF-MS)with electrospray ionization coupled to a Dionex Ultimate3000 UHPLC. Chromatographic separation was achievedwith a Waters Amide column (150 3 2.1 mm, 1.8-mmparticle size) used in hydrophilic interaction liquid chro-matography mode. Solvent A consisted of acetonitrile andisopropanol (3 : 1) and solvent B of methanol and water(1 : 1); in both cases 0.065% of formic acid and ammoniumhydroxide were added. A gradient run was performedstarting with 0% B for 3 min, and then slowly rising to 6%B during 11 min, followed by 2 min washing with 60% B;subsequent column re-equilibration for 6 min completedthe chromatographic run. Flow rate was 0.4 mL min21.LC-MS grade acetonitrile and methanol were purchasedfrom Merck Chemicals. Isopropanol, formic acid 98%, and

ammonium hydroxide solution . 25% were obtainedfrom Sigma Aldrich. Isopropanol and deionized waterwere additionally distilled in the lab. The method wasvalidated by comparing eight samples with the onedescribed by Bauersachs et al. (2009b). Our protocol issuitable for HG analysis because all HGs observed by themethod utilizing a diol column (Bauersachs et al. 2009b)were detected, while peak areas were consistently higher,which resulted in a more sensitive analytical assay.

Detection was performed in positive ionization modewhile scanning a mass-to-charge (m/z) range from 300 to1000. MS2 scans were obtained in data-dependent mode;for each MS full scan, up to three MS2 experiments,targeting the most abundant ions, were performed. Activeexclusion limits the times a given ion is selected forfragmentation (three times every 0.5 min) and thus allowedto also obtain MS2 data for less abundant ions. Detectionof HGs was achieved by monitoring exact masses ofpossible parent ions (present as either H+ or NHz

4 adducts)in combination with characteristic losses of the sugarmoiety and subsequent losses of up to four H2O molecules,as described by Bauersachs et al. (2009b). No authentic HGstandards are available; therefore, reported concentrations,obtained by comparison of responses of parent ions relativeto known amounts of an internal standard (1,2-dihenar-achidoyl-sn-glycero-3-phosphocholine, Avanti Lipids),should be viewed as semiquantitative. No detailed struc-tural characterization of the glycosidic and alkyl moietieswas carried out; thus, the glycosidic moieties (‘headgroups’)are generically referred to as hexose, while hydroxyl andcarbonyl groups are assumed at the commonly reportedpositions described above (C-3, v-1, and v-3). For brevity,hexose-HGs will subsequently be referred to only by thelength of the alkyl chain and the number and type offunctional moieties, omitting the sugar headgroup (e.g.,keto-triacontanediol refers to hexose keto-triacontanediol).

Data analysis software—The comparison of the lipidextracts from A. ovalisporum UAM 290 grown underdiazotrophic and nondiazotrophic conditions was per-formed with the Bruker MetaboliteDetect software (version2.0, 2008). MetaboliteDetect allows sophisticated com-parison of two full-scan data sets. Instead of simplysubtracting chromatograms, the applied eXpose algorithmaccounts for tolerances in retention time and mass position.In this study, we generated an accurate delta chromato-gram that shows masses detected more strongly in thediazotrophic culture than in the nondiazotrophic one.Major peaks in this chromatogram were identified based onthe exact mass of both MS and MS2 spectra.

Hierarchical clustering of culture and environmentalsamples based on HG distribution was achieved with theStatistics package for Excel XLSTAT (version 2011.4.03).The Pearson correlation coefficient was used for construct-ing similarity dendrograms.

HG26-index—Based on three heterocystous cyanobacte-rial strains grown at two temperatures, Bauersachs et al.(2009a) established the HG26-index (i.e., a ratio defined asC26 keto-ol / (C26 diol + C26 keto-ol)) and suggested to

1778 Wormer et al.

negatively correlate with temperature. We applied thisindex to the environmental samples in order to study itsrelationship to water temperature.

Results

Novel HG compounds—The analysis of diazotrophicallygrown Aphanizomenon aphanizomenoides UAM 523 re-vealed the presence of two novel HG types, tentativelyidentified as hexose-3,29-triacontanediol and hexose-3-keto-29-triacontanol (Fig. 2). Previously only triacontane-triols and keto-triacontanediols have been described. Thefragmentation pattern of these two compounds is typicalfor HGs, namely an initial loss of the hexose headgroup(162.053 Da) and subsequent losses of H2O (18.011 Da).Figure 2 provides an example of the structural relation-ships that can be confidently assigned on the basis ofobserved accurate mass of parent ion and fragments andmasses calculated for the proposed molecular structures.Additionally, retention time supported assignment. Withthe chromatographic separation employed, HG retentionrelates to the number and type of their functional moieties.The retention times of the compounds tentatively identifiedas hexose-3,29-triacontanediol and hexose-3-keto-29-tria-contanol are almost identical to those observed for

corresponding hexacosane and octacosane derivatives andthus support our assignments.

Two additional novel compound types were confidentlyassigned as HGs by comparison of A. ovalisporum UAM 290grown under either diazotrophic or nondiazotrophic condi-tions. Figure 3 shows a delta chromatogram obtained bysubtracting nondiazotrophic growth from diazotrophicgrowth. The illustrated peaks are thus characteristic fordiazotrophic A. ovalisporum. Peaks IV and V are consistentwith hexose hexacosanediol (m/z 5 577.467), while peak Iseems to be the core hexacosanediol resulting from the lossof the sugar headgroup (m/z 5 415.415). The peaks II andIII could not be assigned to any known HG structure. Exactmasses (m/z 5 561.472; 547.457) suggest these compoundscould be hexacosanediols, with a substitution of the hexoseheadgroup by deoxyhexose (or methylpentose) and apentose, respectively (Fig. 4). The fragmentation patternobtained for both compounds is consistent with thishypothesis. Fragments are identical to those from hexosehexacosanediol, and are explained by the sequential loss ofthe sugar headgroup and three successive losses of H2O. Theonly difference between hexose hexacosanediol and thesetwo novel compounds is the magnitude of the initial loss,which is equivalent to hexose (162.053 Da), deoxyhexose(146.058 Da), and pentose (134.042 Da), respectively.

Fig. 2. (a, b) MS and (c, d) MS2 spectra, and suggested structure and fragmentation pattern for compounds extracted from A.aphanizomenoides UAM 523 and tentatively identified as (a, c) hexose-3,29-triacontanediol and (b, d) hexose-3-keto-29-triacontanol.Difference between measured and calculated m/z is given in brackets.

Cyanobacterial heterocyst glycolipids 1779

Taxonomic distribution of HGs—The novel hexose-3,29-triacontanediol and hexose-3-keto-29-triacontanol are bothexclusive to the culture of A. aphanizomenoides. However,the highly characteristic HG distribution of A. aphanizo-menoides is not only expressed in the presence of these twonovel HGs. In addition, octacosanediol, keto-octacosanol,triacontanetriol, and keto-triacontanediol derivatives dom-inate, while hexacosane HGs are virtually absent. Exceptfor A. aphanizomenoides, the three pelagic and the benthicrepresentative of Nostocaceae show a relatively simple HGdistribution (Fig. 5) with hexacosanediol clearly dominat-ing. Keto-hexacosanol is important in Nostoc sp. MA4 andAphanizomenon gracile, while A. ovalisporum is character-ized by the presence of tentatively identified pentose anddeoxyhexose-hexacosanediol. Octacosane derivatives areonly present in small quantities in these cultures. In the caseof Cylindrospermopsis raciborskii UAM 520, dominance ofhexacosane is accompanied by substantial amounts ofoctacosanediol and octacosanetriol. In the case of Calo-thrix sp. MU27, octacosanetriol and keto-octacosanediolare dominant.

HG profiling and cyanobacterial occurrence in environ-mental samples—Hexacosane-based HG derivatives aredominant in most environmental samples (82.4%; Fig. 5),with two polar samples displaying particularly strikingdominance: the W4 pond biofilm (Arctic) and the Januarysample from SW pond (Antarctic). The December samplefrom the SW pond shows a more complex distribution. Atthe beginning of the summer season after ice has retreated,HGs are composed of hexacosane (38%), octacosane(38%), and triacontane (24%) derivatives. At the secondsampling, about 2 months after ice retreat, the sample isclearly hexacosane-dominated (93%) and absolute HGabundance is significantly decreased; combined abun-dance of all hexacosane derivatives (peak area per g ofdry weight) is 5.6-fold (of octacosane derivatives 87-foldand of triacontane derivatives even 316-fold [data not

shown]). In the W4 pond sample, Nostoc sp. was the onlyheterocyst-forming cyanobacterial species detected. Thecyanobacterial community in both samples from SWpond was composed of Nostoc sp. and Tolypothrix sp.(Microchaetaceae). Unfortunately no quantitative taxo-nomic data concerning their relative abundance areavailable.

In the pelagic samples analyzed (Fig. 5), hexacosanederivatives are dominant in 93% of the samples, but only intwo samples do they account for . 80% of the HG pool.Octacosane derivatives are abundant in all samples andtriacontane derivatives are abundant in six of the analyzedsamples. Octacosane derivatives account for 74% in theBrovales Reservoir.

Figure 6 shows relative abundance of cyanobacterialspecies in the analyzed samples. Nostocaceae are abundantin all samples and are dominant in 10 out of 14 samples;among Nostocaceae, Anabaena are dominant in sixsamples, Aphanizomenon in five, Anabaenopsis in two, andCylindrospermopsis in one. On a species level, the commu-nities are rather simple and are mainly dominated by onlyone or two species.

Environmental factors influencing HG distribution—Tem-perature has been suggested as a factor influencing HGpatterns and, based on laboratory experiments, Bauersachset al. (2009a) established the HG26-index. In the presentstudy, highest values (0.31) for the index are observed in thepolar samples; freshwater samples from the colder reser-voirs in Northern Spain show values between 0.12 and 0.26;and finally, samples from the warmer reservoirs in southernSpain present lowest values, ranging between 0.01 and 0.04.When plotting HG26 against temperature (Fig. 7), datacould be fitted to a decreasing sigmoidal curve. Relativelyconstant HG26 values could thus be expected at lowtemperatures (, 15uC) and at very high temperatures (.25uC). Between these extreme ranges a linear decrease canbe established.

Fig. 3. Delta chromatogram showing peaks exclusive to diazotrophic growth of A. ovalisporum UAM 290 and corresponding MSspectra. Peak I is related to core 3,25-hexacosanediol, peaks IV and V are assigned as hexose-3,25-hexacosanediol. Further discussion ofpeaks II and III, tentatively identified as pentose-3,25-hexacosanediol and deoxyhexose-3,25-hexacosanediol, is shown in Fig. 4.

1780 Wormer et al.

Discussion

Taxonomic distribution of HGs—The new data of theanalyzed cultures confirm differences in HG alkyl chainlength between Nostocaceae and Rivulariaceae (Table 3).Abundance of hexacosane derivatives in the Nostocaceaefamily is in accordance with Gambacorta et al. (1999) andBauersachs et al. (2009a) and is observed for four of thestrains tested in the present study. The dominance ofoctacosanetriol and keto-octacosanediol in Calothrix sp.MU27 is supportive of previous results (Gambacorta et al.1998; Bauersachs et al. 2009a) and confirms the charac-teristic HG profile of the genus Calothrix. Previously,octacosanetriol HGs had only been found in traceconcentrations in members of the Nostocaceae family(Gambacorta et al. 1996; Bauersachs et al. 2009a). Thepresent study though, in which we observed significantpresence of octacosanetriol in C. raciborskii and, to a lesserdegree, A. gracile, clearly shows that this compound class isnot exclusive to Rivulariaceae. Keto-octacosanediols,which have not been described to be abundant inNostocaceae, may serve as more robust chemotaxonomicmarkers for the Calothrix genus. However, based on thepresent data and the possible effect of environmental

factors on HG synthesis, specific HGs should not beconsidered exclusive and thus characteristic for a certaingenus or species. Instead, when trying to characterizeheterocystous cyanobacteria based on their HG signature,we suggest a focus on HG patterns rather than on singleHGs.

The HG distribution of A. aphanizomenoides clearlydiffers from both the cultures analyzed in this study andmost strains of the genus Aphanizomenon or the familyNostocaceae (see Table 3 for an overview of HG distribu-tion in cultured strains). In addition to the absence ofhexacosane derivatives and the dominance of octacosanederivatives, the high abundance of triacontane compoundsin A. aphanizomenoides is noteworthy. Such long-chainedHGs had been related to Scytonemataceae, Microchaeta-ceae, and Stigonematales (Gambacorta et al. 1998) andwere not detected in Nostocaceae and Rivulariaceae byBauersachs et al. (2009a). The greatest similarity to the HGpattern of A. aphanizomenoides is probably found inAnabaena sphaerica and Anabaena sp. strain WSAF(Gambacorta et al. 1996), and in members of the familyMicrochaetaceae (Gambacorta et al. 1998).

The atypical HG distribution in A. aphanizomenoidesrelative to other strains of the genera Aphanizomenon,

Fig. 4. (a, b) MS and (c, d) MS2 spectra and suggested structure and fragmentation pattern for compounds extracted from A.ovalisporum UAM 290 and tentatively identified as (a, c) deoxyhexose-3,25-hexacosanediol and (b, d) pentose-3,25-hexacosanediol(compounds II and III in Fig. 3). Difference between measured and calculated m/z is given in brackets.

Cyanobacterial heterocyst glycolipids 1781

Anabaena (in which A. aphanizomenoides was previouslyincluded), or Nostoc supports recent findings by Zapome-lova et al. (2009). These authors applied polyphasiccharacterization to strains of Anabaena reniformis and A.aphanizomenoides and observed that these clustered sepa-rately from other pelagic Anabaena and Aphanizomenonstrains. Also, secondary metabolite profiles of the testedstrains were considerably different from reference strains.Similarly Stuken et al. (2009) observed A. aphanizomenoidesto cluster separately based on the analysis of partialsequences of three genes. Likewise, HG compositiondifferentiates A. aphanizomenoides from other Aphanizo-menon and Anabaena species. This example clearly demon-strates the taxonomic resolution of HG fingerprinting.

HG profiling and cyanobacterial occurrence in environ-mental samples—The simple hexacosanediol-dominatedHG profile from the W4 pond is consistent with Nostocsp. being the only member of the community. Severalspecies of this genus were characterized by hexacosanedioland keto-hexacosanol (Table 3). The absence of keto-hexacosanol in the W4 sample might be explained bytemperature-dependent synthesis of this compound (Bauer-

sachs et al. 2009a). In the two mat samples from SW pond,we observed a change of HG composition and decrease inconcentration, probably signaling a change of both thecommunity mediating N2 fixation and the importance ofthis process in this ecosystem.

The HGs of a Tolypothrix tenuis strain consist ofoctacosane and triacontane derivatives, while the genusNostoc is generally dominated by hexacosane derivatives(Gambacorta et al. 1998; Table 3). Considering thepresence of both Nostoc sp. and Tolypothrix sp. in theSW pond, we suggest that after retreat of ice, the activity ofN2 fixation started and probably involved both species, asindicated by the presence of hexacosane, octacosane, andtriacontane derivatives. At the start of the season, nutrientsare extremely scarce and thus N2 fixation may be needed tobuild up community development. Such fixation at lowtemperatures has been previously demonstrated in incuba-tion experiments with this microbial mat even at 0uC(Velazquez et al. 2011). At the end of the season, fixationwas strongly reduced and carried out exclusively byhexacosane-dominated Nostoc sp. This reduction in N2

fixation would be in accordance with field observationsthat indicated a 10-fold increase in dissolved inorganic

Fig. 5. Lipid distribution in (I) cyanobacterial cultures, (II) polar mats and biofilm, and (III) bloom samples from Spanishfreshwaters. Relative abundance (%) is shown. Dotted line indicates contribution of hexacosane derivatives.

1782 Wormer et al.

nitrogen in the pond water at the second sampling date (D.Velazquez, unpubl.). The available data do not provideinsight into the causes for the inferred temporal successionsof heterocystous cyanobacteria but stress the potential ofHG analysis for quantitative and qualitative profiling ofmodern cyanobacterial ecosystems and N cycling.

In the pelagic samples, HG distribution is far morediverse than in cultured strains, in the polar benthicsamples described above, and in the marine microbialmat samples studied by Bauersachs et al. (2011). In asimilarity dendrogram based on relative abundance of HGs(Fig. 8), eight groups can be established by setting thetruncation level to 0.9. The major factor responsible for thisclustering is the taxonomic composition of the cyanobac-terial community (Fig. 6). Most of the samples (17 of 23)are closely related and included in groups A to C. Samplesin these clusters are dominated by Nostocaceae, which arepresumed to be hexacosane-dominated and include, forexample, C. raciborskii and the sample from RosaritoReservoir dominated by C. raciborskii in cluster A orbenthic Nostoc sp. MA4 and Nostoc-dominated samplesfrom ponds W4 and SW II in cluster B. Cluster C iscomposed of samples dominated either by the genusAnabaena or by Aphanizomenon sp. Even though HGdiversity is high and some unexpected HGs occur inthese environmental samples, the observed HG patternsconsistently reflect the HG patterns expected for the mostabundant heterocystous cyanobacteria.

Fig. 6. Relative abundance (%) of cyanobacterial species in the analyzed pelagic samples.

Fig. 7. HG26-index [C26keto-ol / (C26 diol + C26 keto-ol)] inenvironmental samples vs. temperature and fitted three-parametersigmoidal curve (R2 5 0.83, p , 0.01).

Cyanobacterial heterocyst glycolipids 1783

The five remaining clusters account for only six samples.Cluster D is composed of A. aphanizomenoides UAM 523and the sample from Brovales Reservoir. Both shareoctacosane and triacontane dominance and almost com-plete absence of hexacosane HGs. Considering that thecyanobacterial bloom in Brovales was dominated by A.aphanizomenoides, this HG signature is consistent withtaxonomy. Similarly, this taxonomic issue would alsoexplain the lone clustering of the sample from NogalesReservoir, where A. aphanizomenoides is abundant but notdominant. Surprisingly, the Navalcan Reservoir is groupedin cluster A, although it is dominated by A. aphanizome-noides. This discrepancy may be caused by a misidentifi-cation due to the morphological similarity to A. gracile orA. ovalisporum. Indeed the sample from Navalcan isclustered together with the only A. gracile–dominatedsample (Oliana II). In this sense, the extreme difficulty ofmorphological characterization of some species of thegenus Aphanizomenon could make the use of HG profilingfor taxonomic purposes attractive. For example, in the case

of the Alcantara sample, paralytic shellfish toxin (PST) hadbeen previously detected, but the producing organismcould not be identified (Wormer et al. 2011) because itshared some morphological characteristics (trichome widthand shape and position of akinetes) with both A. gracileand A. aphanizomenoides. Based on HG data, we nowsuggest that the PST-producing organism in the Alcantarasample is not A. aphanizomenoides but rather A. gracile.This explanation is consistent with PST production byother A. gracile strains (Pereira et al. 2004; Ballot et al.2010).

The remaining three clusters are each composed of onlyone sample. In the case of Calothrix sp. MU27 and theDecember sample from SW pond, clustering can beexplained, respectively, by the distinct HG profiles ofRivulariaceae and Microchaetaceae. The last cluster iscomposed of A. ovalisporum UAM 290, whose HGdistribution is characterized by the presence of thetentatively identified pentose and deoxyhexose hexacosa-nediol derivatives. If the distinct HG composition observed

Table 3. HGs detected in a selection of Nostocales strains, based on (a) Soriente et al. (1993), (b) Gambacorta et al. (1996),(c) Gambacorta et al. (1998), (d) Bauersachs et al. (2009a), and the present study (bold). Data are adapted to a relative abundance scale(+++ dominant, ++ major presence, + minor presence or traces, — not detected).

Strain FamilyHexose

hexacosane-diolHexose

keto-hexacosanolHexose

octacosane-diolHexose

octacosane-triol

Anabaena cylindrica 10C b Nostocaceae +++ ++ — +Anabaena cylindrica ATCC29414 b Nostocaceae +++ ++ — +Anabaena cylindrica CCY9921 d Nostocaceae +++ ++ — —Anabaena sphaerica b Nostocaceae — — +++ —Anabaena sp. 7120 b Nostocaceae +++ ++ — —Anabaena sp. WSAF b Nostocaceae — — +++ —Anabaena sp. CCY0017 d Nostocaceae +++ ++ ++ —Anabaena sp. CCY9402 d Nostocaceae — — +++ —Anabaena sp. CCY9613 d Nostocaceae ++ ++ — —Anabaena sp. CCY9614 d Nostocaceae +++ ++ — —Anabaena sp. CCY9910 d Nostocaceae +++ ++ ++ —Anabaena sp. CCY9922 d Nostocaceae +++ ++ — —Anabaenopsis sp. CCY0520 d Nostocaceae +++ ++ ++ —Aphanizomenon aphanizomenoides UAM 523 Nostocaceae + — +++ +Aphanizomenon gracile UAM 521 Nostocaceae +++ ++ + +Aphanizomenon ovalisporum UAM 290 Nostocaceae +++ + + —Aphanizomenon sp. CCY0368 d Nostocaceae +++ ++ ++ —Aphanizomenon sp. CCY9905 d Nostocaceae +++ ++ ++ +Cyanospira rippkae a Nostocaceae — — ++ —Cylindrospermopsis raciborskii UAM 520 Nostocaceae +++ + ++ ++Nodularia chucula CCY0103 d Nostocaceae +++ ++ — —Nodularia sp. CCY9414 d Nostocaceae +++ ++ — —Nodularia sp. CCY9416 d Nostocaceae +++ ++ — —Nostoc lickia b Nostocaceae +++ ++ — —Nostoc sp. CCY0012 d Nostocaceae +++ ++ — —Nostoc sp. CCY9926 d Nostocaceae +++ ++ — —Nostoc sp. MA4 Nostocaceae +++ ++ + +Calothrix desertica c Rivulariaceae — — — ++Calothrix sp. CCY0018 d Rivulariaceae — — — +++Calothrix sp. CCY0202 d Rivulariaceae + + — +++Calothrix sp. CCY0327 d Rivulariaceae — — — +++Calothrix sp. CCY9923 d Rivulariaceae — — ++ +++Calothrix sp. MU27 Rivulariaceae — — + +++Microchaete sp. c Microchaetaceae — — ++ —Tolypothrix tenuis c Microchaetaceae — — +++ —Scytonema hofmanni c Scytonemataceae — — — —

1784 Wormer et al.

in A. ovalisporum UAM 290 is common to other membersof this species, then the novel HG derivatives could becomeuseful taxonomic markers, especially when considering itsmorphological similarity to A. gracile and A. aphanizome-noides, its potential toxicity (Banker et al. 1997), and thesuggested invasion of mid-latitude ecosystems by thisspecies (Mehnert et al. 2010).

Environmental factors influencing HG distribution—Although taxonomic relationships may explain the domi-nance of certain HGs, the high diversity of HG distribu-tions in the reservoir samples does not correspond well withthe relatively simple taxonomic composition of heterocys-tous cyanobacteria. Even in blooms clearly dominated bya single taxon, like the ones in Peares, Oliana, CuevaForadada, or Cachamuinas reservoirs, HG diversity ismuch greater than in samples from cyanobacterial cultures.The dominance of hexacosane derivatives is generallystrongly reduced in the reservoir samples, and compoundsthat are rare in Nostocaceae cultures are more abundant.

This phenomenon suggests that HG fingerprinting ismore sensitive to taxonomic diversity and/or the existenceof subpopulations than are classical morphological ap-proaches. In addition, unconstrained environmental factorsmay also influence HG biosynthesis. Under the controlledconditions in which isolated strains are grown, the needfor different HG types may not be as high as in naturalenvironments, thus resulting in more simple HG profiles. Inthis sense, Kangatharalingam et al. (1992), for example,demonstrated that heterocyst glycolipid-layer thickness wasdependent on exogenous oxygen tension. Whether suchthickening could imply changes in the HG compositionremains unknown. The potential of certain conditions toalter an organism’s HG profile exists, given that distinctsignaling cascades are regulating the synthesis of differentHGs involved in heterocyst formation (Nicolaisen et al.2009). Efforts toward the investigation of the effect ofenvironmental factors on HG fingerprinting are importantfor the further development of this compound class asbiomarkers.

Hexoseketo-octacosanol

Hexose keto-octacosane-diol

Hexosetriacontane-diol

Hexosetriacontane-triol

Hexose keto-triacontanol

Hexose keto-triacontane-diol

Pentosehexacosane-diol

Deoxyhexosehexacosane-diol

— + — — — — — —— + — — — — — —— — — — — — — —++ — — ++ — ++ — —— — — — — — — —++ — — ++ — ++ — —++ — — — — — — —++ — — — — — — —— — — — — — — —— — — — — — — —++ — — — — — — —— — — — — — — —— — — — — — — —++ + + ++ + ++ — —— — — — — — — —+ — — — — — ++ ++— — — — — — — —++ + — — — — — —++ — — — — — — —+ + — — — — + +— — — — — — — —— — — — — — — —— — — — — — — —— — — — — — — —— — — — — — — —— — — — — — — —+ — — — — — — —— +++ — — — — — —— ++ — — — — — —— ++ — — — — — —— ++ — — — — — —++ ++ — — — — — —— ++ — + — — — —

+++ — — — — — — —++ — — ++ — ++ — —— — — +++ — ++ — —

Table 3. Extended.

Cyanobacterial heterocyst glycolipids 1785

One of the factors altering HG synthesis is temperature.In laboratory experiments, Bauersachs et al. (2009a)compared HG composition in cyanobacterial strains grownat two different temperatures and observed decreasedamounts of keto-hexacosanol relative to hexacosanediol athigher temperature. The authors suggested that thisadaptation is linked to optimized gas diffusion because, athigher temperatures, oxygen removal via respiration is moreeffective and would thus allow increased gas diffusion intothe heterocyst. The established HG26-index (Bauersachs etal. 2009a) is proposed to negatively correlate with temper-ature. The HG26-index utility is here demonstrated over awide range of temperatures in environmental samples. Theindex is sensitive to major temperature changes andconsiderable variations are observed in our sample set.However, the ability to resolve temperature differencesabove 25uC or below 15uC is poor. The physiologicaladaptation underlying the HG26-index may no longer beeffective under these temperature regimes and, therefore, isreplaced by different adaptation strategies. Interestingly, asimilar sigmoidal relationship with temperature was ob-served by Rattray et al. (2010) for the distribution of selectedladderane lipids that form part of the anammoxosome ofcertain Planktomycetes; significant changes in ladderanelipid composition were observed between 12uC and 20uC butonly small variations were observed outside this range.

Further studies are needed to define the temperaturerange in which the HG26-index is useful and to assess otherfactors influencing relative abundance of keto-hexacosanoland other HGs. In the pelagic strains tested in the present

study, for example, abundance of keto-hexacosanol relativeto hexacosanediol varies strongly even when they are grownat the same temperature. Such variations could be attributedto a constitutive taxonomic issue, but could also be related tochanging HG composition during growth of the analyzedcultures. HG fingerprinting over complete growth curves ofbenthic and pelagic diazotrophic cyanobacteria could shedsome light on these and other relevant issues. In paleoenvi-ronments, the HG26-index may differentiate HG productionduring cold and warm seasons and may thus provide insightsinto the dynamics of past nitrogen cycling.

Our study of heterocyst-forming cyanobacterial strainsand of diverse environmental samples has revealed a highdiversity of these compounds. The diverse distributions mayprovide valuable taxonomic information and/or may offerinsights into temporary and spatial successions of N2-fixingcyanobacteria. HG analysis in recent sediments at appropri-ate temporal resolution may provide information on the lifecycle and colonization strategies in an ecosystem. Applica-tions in paleoenvironments will potentially add importantinsights into the dynamics and modes of past nutrient cycling.

AcknowledgmentsThis work was funded by the Deutsche Forschungsge-

meinschaft (DFG) through grant Inst 144/300-1 (LC-QToFsystem) and the DFG-Research Center and Excellence Cluster‘The Ocean in the Earth System’ and by the European ResearchCouncil via project ‘Deep subsurface Archaea: carbon cycle, lifestrategies and role in sedimentary ecosystems (DARCLIFE).’

We thank Pilar Mateo and Esther Berrendero (UniversidadAutonoma de Madrid) for providing us with biomass of Nostoc

Fig. 8. Hierarchical clustering of environmental samples and cyanobacterial cultures basedon HG fingerprinting.

1786 Wormer et al.

sp. MA4 and Calothrix sp. MU27, and three anonymousreviewers for helpful comments and corrections that significantlyimproved the quality of the manuscript.

References

BALLOT, A., J. FASTNER, AND C. WIEDNER. 2010. Paralytic shellfishpoisoning toxin-producing cyanobacterium Aphanizomenongracile in Northeast Germany. Appl. Environ. Microb. 76:1173–1180, doi:10.1128/AEM.02285-09

BANKER, R., S. CARMELI, O. HADAS, B. TELTSCH, R. PORAT, AND

A. SUKENIK. 1997. Identification of cylindrospermopsin inAphanizomenon ovalisporum (Cyanophyceae) isolated fromLake Kinneret, Israel. J. Phycol. 33: 613–616, doi:10.1111/j.0022-3646.1997.00613.x

BAUERSACHS, T., J. COMPAORE, E. C. HOPMANS, L. J. STAL, S.SCHOUTEN, AND J. S. SINNINGHE DAMSTE. 2009a. Distributionof heterocyst glycolipids in cyanobacteria. Phytochemistry 70:2034–2039, doi:10.1016/j.phytochem.2009.08.014

———, ———, I. SEVERIN, E. C. HOPMANS, S. SCHOUTEN, L. J.STAL, AND J. S. SINNINGHE DAMSTE. 2011. Diazotrophicmicrobial community of coastal microbial mats of thesouthern North Sea. Geobiology 9: 349–359, doi:10.1111/j.1472-4669.2011.00280.x

———, E. C. HOPMANS, J. COMPAORE, L. J. STAL, S. SCHOUTEN,AND J. S. SINNINGHE DAMSTE. 2009b. Rapid analysis of long-chain glycolipids in heterocystous cyanobacteria using high-performance liquid chromatography coupled to electrosprayionization tandem mass spectrometry. Rapid Commun. Mass.Spectrom. 23: 1387–1394.

———, E. N. SPEELMAN, E. C. HOPMANS, G.-J. REICHART, S.SCHOUTEN, AND J. S. SINNINGHE DAMSTE. 2010. Fossilizedglycolipids reveal past oceanic N2 fixation by heterocystouscyanobacteria. Proc. Natl. Acad. Sci. U. S. A. 107:19190–19194, doi:10.1073/pnas.1007526107

BERMAN-FRANK, I., P. LUNDGREN, AND P. FALKOWSKI. 2003.Nitrogen fixation and photosynthetic oxygen evolution incyanobacteria. Res. Microbiol. 154: 157–164, doi:10.1016/S0923-2508(03)00029-9

BLIGH, E. G., AND W. J. DYER. 1959. A rapid method of total lipidextraction and purification. Can. J. Biochem. Phys. 37: 911–917, doi:10.1139/o59-099

BROADY, P. A., AND A. L. KIBBLEWHITE. 1991. Morphologicalcharacterization of Oscillatoriales (Cyanobacteria) from RossIsland and Southern Victoria Land, Antarctica. Antarct. Sci.3: 35–45, doi:10.1017/S095410209100007X

BRYCE, T. A., D. WELTI, A. E. WALSBY, AND B. W. NICHOLS. 1972.Monohexoside derivatives of long-chain polyhydroxy alcohols—novel class of glycolipid specific to heterocystous algae. Phy-tochemistry 11: 295–302, doi:10.1016/S0031-9422(00)90006-2

CHU, S. P. 1942. The influence of the mineral composition ofthe medium on the growth of planktonic algae. Part 1.Methods and culture media. J. Ecol. 30: 284–325, doi:10.2307/2256574

GAMBACORTA, A., E. PAGNOTTA, I. ROMANO, G. SODANO, AND

A. TRINCONE. 1998. Heterocyst glycolipids from nitrogen-fixing cyanobacteria other than Nostocaceae. Phytochemistry48: 801–805, doi:10.1016/S0031-9422(97)00954-0

———, I. ROMANO, A. TRINCONE, A. SORIENTE, M. GIORDANO,AND G. SODANO. 1996. Heterocyst glycolipids from fivenitrogen-fixing cyanobacteria. Gazz. Chim. Ital. 126:653–656.

———, A. TRINCONE, A. SORIENTE, AND G. SODANO. 1999.Chemistry of glycolipids from the heterocysts of nitrogen-fixing cyanobacteria. Curr. Top. Phytochem. 2: 145–150.

GEITLER, L. 1932. Cyanophyceae, p. 1–1196. In R. Kolkwitz [ed.],Dr. L. Rabenhorst’s Kryptogamen Flora. Vierzehnter Band:Die Algen. Koeltz Scientific. p. 1–1196.

HAUG, G. H., T. F. PEDERSEN, D. M. SIGMAN, S. E. CALVERT, B.NIELSEN, AND L. PEDERSEN. 1998. Glacial/interglacial variationin nitrogen fixation in the Cariaco Basin during the last580 kyr. Paleoceanogr 13: 427–432, doi:10.1029/98PA01976

HOWARTH, R. W., R. MARINO, J. LANE, AND J. J. COLE. 1988.Nitrogen-fixation in fresh-water, estuarine, and marineecosystems. 1. Rates and importance. Limnol. Oceanogr. 33:669–687, doi:10.4319/lo.1988.33.4_part_2.0669

KANGATHARALINGAM, N., J. C. PRISCU, AND H. W. PAERL. 1992.Heterocyst envelope thickness, heterocyst frequency andnitrogenase activity in Anabaena flos-aquae—influence ofexogenous oxygen-tension. J. Gen. Microbiol. 138: 2673–2678, doi:10.1099/00221287-138-12-2673

KARL, D., R. LETELIER, L. TUPAS, J. DORE, J. CHRISTIAN, AND D.HEBEL. 1997. The role of nitrogen fixation in biogeochemicalcycling in the subtropical North Pacific Ocean. Nature 388:533–538, doi:10.1038/41474

KOMAREK, J., AND K. ANAGNOSTIDIS. 1989. Modern approach tothe classification-system of Cyanophytes 4 - Nostocales. Arch.Hydrobiol. 82: 247–345.

———, AND ———. 1999. Cyanoprokaryota 1. Teil: Chroococ-cales, p. 1–548. In H. Ettl, G. Gartner, H. Heynig, and D.Mollenhauer [eds.], Susswasserflora von Mitteleuropa. Gus-tav Fischer Verlag. [Freshwater Flora from Central Europe.]p. 1–548.

———, AND ———. 2005. Cyanoprokaryota 2. Oscillatoriales,p. 1–760. In B. Budel, G. Gartner, L. Krienitz, and M.Schagerl [eds.], Susswasserflora von Mitteleuropa. Elsevier.[Freshwater Flora from Central Europe.]

LAMBEIN, F., AND C. P. WOLK. 1973. Structural studies onglycolipids from envelope of heterocyst of Anabaena cylin-drica. Biochemistry 12: 791–798, doi:10.1021/bi00729a002

MEHNERT, G., F. LEUNERT, S. CIRES, K. D. JOHNK, J. RUCKER, B.NIXDORF, AND C. WIEDNER. 2010. Competitiveness of invasiveand native cyanobacteria from temperate freshwaters undervarious light and temperature conditions. J. Plankton Res. 32:1009–1021, doi:10.1093/plankt/fbq033

NICHOLS, B. W., AND B. J. B. WOOD. 1968. New glycolipid specificto nitrogen-fixing blue-green algae. Nature 217: 767–768,doi:10.1038/217767a0

NICOLAISEN, K., A. HAHN, AND E. SCHLEIFF. 2009. The cell wall inheterocyst formation by Anabaena sp. PCC 7120. J. BasicMicrobiol. 49: 5–24, doi:10.1002/jobm.200800300

PEREIRA, P., R. H. LI, W. W. CARMICHAEL, E. DIAS, AND S.FRANCA. 2004. Taxonomy and production of paralyticshellfish toxins by the freshwater cyanobacterium Aphanizo-menon gracile LMECYA40. Eur. J. Phycol. 39: 361–368,doi:10.1080/09670260410001714723

QUESADA, A., M. NIEVA, F. LEGANES, A. UCHA, M. MARTıN, C.PROSPERI, AND E. FERNANDEZ-VALIENTE. 1998. Acclimationof cyanobacterial communities in rice fields and responseof nitrogenase activity to light regime. Microb. Ecol. 35:147–155, doi:10.1007/s002489900069

RATTRAY, J. E., AND OTHERS. 2010. Impact of temperature onladderane lipid distribution in Anammox bacteria. Appl.Environ. Microbiol. 76: 1596–1603, doi:10.1128/AEM.01796-09

RIPPKA, R., J. DERUELLES, J. WATERBURY, M. HERDMAN, AND R.STANIER. 1979. Generic assignments, strain histories andproperties of pure cultures of cyanobacteria. J. Gen. Micro-biol. 111: 1–61, doi:10.1099/00221287-111-1-1

SACHS, J. P., AND D. J. REPETA. 1999. Oligotrophy and nitrogenfixation during Eastern Mediterranean sapropel events.Science 286: 2485–2488, doi:10.1126/science.286.5449.2485

Cyanobacterial heterocyst glycolipids 1787

SCHOPF, J. W. 2000. The fossil record: Tracing the roots ofcyanobacterial lineage, p. 13–35. In B. Whitton and M. Potts[eds.], The ecology of cyanobacteria: Their diversity in timeand space. Kluwer Academic. p. 13–35.

SORIENTE, A., A. GAMBACORTA, A. TRINCONE, C. SILI, M.VINCENZINI, AND G. SODANO. 1993. Heterocyst glycolipids ofthe cyanobacterium Cyanospira rippkae. Phytochemistry 33:393–396, doi:10.1016/0031-9422(93)85526-W

STAL, L. J., AND W. E. KRUMBEIN. 1987. Temporal separation ofnitrogen-fixation and photosynthesis in the filamentous,nonheterocystous cyanobacterium Oscillatoria sp. Arch.Microbiol. 149: 76–80, doi:10.1007/BF00423140

STUKEN, A., R. J. CAMPBELL, A. QUESADA, A. SUKENIK, P. K.DADHEECH, AND C. WIEDNER. 2009. Genetic and morphologiccharacterization of four putative cylindrospermopsin produc-ing species of the cyanobacterial genera Anabaena andAphanizomenon. J. Plankton Res. 31: 465–480, doi:10.1093/plankt/fbp011

STURT, H. F., R. E. SUMMONS, K. SMITH, M. ELVERT, AND K.-U.HINRICHS. 2004. Intact polar membrane lipids in prokaryotesand sediments deciphered by high-performance liquid chro-matography/electrospray ionization multistage mass spec-trometry—new biomarkers for biogeochemistry and microbi-al ecology. Rapid Commun. Mass. Spectrom. 18: 617–628.

SUMMONS, R. E., L. J. JAHNKE, J. M. HOPE, AND G. A. LOGAN.1999. 2-Methylhopanoids as biomarkers for cyanobacterialoxygenic photosynthesis. Nature 400: 554–557, doi:10.1038/23005

UTERMOHL, H. 1958. Zur Vervollkommung der quantitativenPhytoplankton-Methodik. Mitt. Int. Ver. Limnol. 9: 1–38.[Completing methods for phytoplankton quantification.]

VELAZQUEZ, D., C. ROCHERA, A. CAMACHO, AND A. QUESADA.2011. Temperature effects on carbon and nitrogen metabolismin some Maritime Antarctic freshwater phototrophic com-munities. Polar Biol. 34: 1045–1055, doi:10.1007/s00300-011-0964-7

WHITTON, B., AND M. POTTS. 2000. The ecology of cyanobacteria:Their diversity in time and space. Kluwer Academic.

WORMER, L., S. CIRES, R. AGHA, M. VERDUGO, C. DE HOYOS, AND

A. QUESADA. 2011. First detection of cyanobacterial PSP(paralytic shellfish poisoning) toxins in Spanish freshwaters.Toxicon 57: 918–921, doi:10.1016/j.toxicon.2011.02.022

ZAPOMELOVA, E., J. JEZBEROVA, P. HROUZEK, D. HISEM, K.REHAKOVA, AND J. KOMARKOVA. 2009. Polyphasic character-ization of three strains of Anabaena reniformis and Aphani-zomenon aphanizomenoides (Cyanobacteria) and their reclas-sification to Sphaerospermum gen. nov (incl. Anabaenakisseleviana). J. Phycol. 45: 1363–1373, doi:10.1111/j.1529-8817.2009.00758.x

Associate editor: John Albert Raven

Received: 13 April 2012Accepted: 10 August 2012Amended: 14 August 2012

1788 Wormer et al.