Several allelochemicals isolated from cyanobacteria or higher plants specifically inhibit photosynthesis of target organisms (Einhellig et al., 1993; Gonzalez et al., 1997; Smith and Doan, 1999). This results in a lower primary production and might consequently cause slower growth of competing photoautotrophs. To prove effects on primary production, the radiocarbon assay (fixation of radioactive labeled ¹⁴C) has been frequently used, e.g. with extracts from Zostera marina (Harrison and Durance, 1985) and Chara globularis (Wium-Andersen et al., 1983). More detailed studies involved separate measurements of the activity of photosystems I and II (PSI and PSII). Sorgoleone, a p-benzoquinone from sorghum (Sorghum bicolor), inhibits the electron transport between Q_A and Q_B (Gonzalez et al., 1997). Many benthic filamentous cyanobacteria produce specific inhibitors of PSII (see Smith and Doan, 1999), like cyanobacterin from Scytonema hofmannii or fischerellin A from Fischerella muscicola and Fischerella ambigua. The former inhibits the photosynthetic electron transport on the acceptor side of PSII (Gleason and Paulson, 1984; Gleason and Case, 1986), whereas the latter interrupts the electron transport at four different sites (Srivastava et al., 1998) and interferes additionally with membrane integrity. These allelopathically active compounds act differently from synthetic herbicides, an apparently common trait of natural herbicides (Duke et al., 2000). Such compounds reveal a great potential for new target sites not yet exploited by commercial herbicides.

Myriophyllum spicatum (also named milfoil hereafter) is a highly competitive submerged macrophyte with strong allelopathic potential (Planas et al., 1981; Agami and Waisel, 1985; Gross et al., 1996). The major active allelochemical has been identified as β-1,2,3-tri-O-galloyl-4,6-(S)-hexahydroxydiphenoyl-d-Glc (tellimagrandin II; Gross et al., 1996). This and other hydrolyzable polyphenols can account for up to 10% of the dry weight of M. spicatum. Part of the inhibitory potential of M. spicatum is attributable to the complexation with and inactivation of cyanobacterial and algal extracellular enzymes, such as alkaline phosphatase (Gross et al., 1996; Gross, 1999). However, this mechanism does not fully explain the strong allelopathic activity. Allelochemicals from M. spicatum interfered with photosynthetic carbon uptake of two common phytoplanktonic diatoms as shown with the radiocarbon method (Gross and Sütfeld, 1994).

Therefore, we investigated the impact of allelopathically active compounds from M. spicatum on photosynthesis of target organisms, especially cyanobacteria. Our objective was to identify the active compound and to characterize its target site and mode of action. We achieved this by measuring PSI and PSII activity, and by application of thermoluminescence (TL) and electron paramagnetic resonance (EPR) spectroscopy.

Table I Content of tellimagrandin II in the SPE fractions as used in the Clark electrode measurements

Figure 1 Inhibition of photosynthetic O2 evolution of Anabaena sp. PCC 7120 by addition of different fractions of the solid phase extraction (SPE) of M. spicatum extract (1, water; 2, 25% [v/v] MeOH; 3, 50% [v/v] (more ...)

Because tellimagrandin II was not available in sufficient amount for dose-response measurements, we used tannic acid, a closely related commercially available polyphenol. Tannic acid inhibited oxygen evolution of intact Synechocystis sp. PCC 6803 cells. Concentrations of 5, 50, and 100 μg mL⁻¹ reduced photosynthetic oxygen evolution to 82%, 57%, and 36% of the control (335 ± 35 μmol O₂ mg⁻¹ Chl h⁻¹, mean ± se, n = 6 per treatment). Significant differences to the controls were found above concentrations of 50 μg mL⁻¹ (Tukey's HSD, P < 0.05).

To exclude a putative uncoupler effect of polyphenols in the extract, PSI activity was measured in the absence of NH₄Cl. Under these conditions, no change in the rate of O₂-uptake was seen in the presence of the lipophilic extract fractions (data not shown).

The TL curves obtained with thylakoid membranes are shown in Figure 2. In the control, the B-band had a maximum emission temperature at about 36°C. A small shoulder was seen at about 55°C (C-band), which was assigned to a Tyr_D⁺Q_A⁻ recombination (Demeter et al., 1993; Johnson et al., 1994), where Tyr_D is a Tyr residue (in PSII) that can be photooxidized. In the presence of the lipophilic fraction, the TL emission was shifted to a higher temperature of about 41°C. It has been described recently that the maximum emission temperature of the B-band is upshifted in the presence of 20 mm formate by approximately 8°C (Bock et al., 2001). Formate competes with HCO₃⁻, one of the physiologically occurring ligands of the non-heme iron between the primary and the secondary quinone acceptors, Q_A and Q_B, and so impairs electron transfer (Diner et al., 1991).

Figure 2 TL signals of thylakoid membranes with no addition (black line) and in the presence of the lipophilic SPE fraction of M. spicatum (dashed line). The amount of extract used in the experiment was equivalent to 190 μg TAE mL−1.

In addition, TL intensities were measured as a function of the number of excitation flashes in the presence and absence of the lipophilic fraction. In the absence of an inhibitor, the B-band oscillates with a period of four representing an active electron transport chain. The period of four oscillation is caused by the turnover of the Mn cluster during the water splitting process. For the production of one molecule oxygen from two water molecules, four light-induced charge separations in the reaction center are required and four positive charges are accumulated. In the presence of the lipophilic fraction, containing approximately 190 μg mL⁻¹ TAE, the oscillation was strongly damped indicating interference with the electron transfer on either the donor or the acceptor side of PSII (data not shown). The same effect was observed using 5 μg mL⁻¹ tannic acid.

Results of TL measurements with PSII-enriched membrane are shown in Figure 3 and Table II. As can be seen for the control, the maximum emission temperature is positioned at a lower temperature (17°C) than observed with thylakoid membranes (36°C). This effect is because of a partial loss of Q_B during the isolation procedure; the TL-band reflects a mixture of the Q-band (S₂Q_A⁻ recombination) and the B-band (S_2,3Q_B⁻ recombination). The addition of 20 mm formate shifted the emission temperature to 22°C. The same effect was obtained by the addition of 3 μm tellimagrandin II. Higher formate concentrations (100 mm) led to a larger upshift (ΔT = 12°C). When adding the lipophilic fraction containing about 6 μm tellimagrandin II, a shift of the maximum emission temperature by 12°C was observed similar to the one in the presence of 100 mm formate. Tannic and gallic acid, as well as (+)-catechin also caused a comparable, dose-dependent shift of the maximum temperature of the B-band of PSII-enriched membrane fragments. The emission temperature of the Q-band measured in the presence of the commercial herbicide DCMU, which binds to the Q_B-binding site, was not changed by the addition of tellimagrandin II or formate. Therefore, we conclude that the redox properties of Q_A were not altered by tellimagrandin II and that the site of action is localized after Q_A in the electron transfer chain.

Figure 3 TL signals of PSII-enriched membrane fragments with no addition (black line) and in the presence of 3 μm tellimagrandin II as pure substance (dashed line) or 20 mm formate (dotted line), respectively.

Table II Shift of the B-band of PSII-enriched membrane particles in thermoluminescence after addition of formate, M. spicatum extracts, tellimagrandin II, or pure phenolic compounds

The donor side of PSII is functional in the presence of M. spicatum lipophilic extract fraction (Table III). The S₂ g = 2 multiline EPR signal (Dismukes and Siderer, 1981) is observed unchanged. This indicated that PSII is capable of charge separation, without any damage to the Mn-cluster. In addition, the functionality of the acceptor side was investigated by measuring the iron-semiquinone (Fe²⁺Q_A⁻) EPR signal. Depending on the ligand to the non-heme iron, this signal appears in two forms, the g = 1.9, when HCO₃⁻, the physiological ligand, is present and the g = 1.82 at low pH (Zimmermann and Rutherford, 1986) or when formate is bound (Vermaas and Rutherford, 1984). Both forms were present in the control and also after addition of the lipophilic fraction. Other molecules have also been shown to bind to the non-heme iron, such as cyanide (Koulougliotis et al., 1993) or carboxylate anions (Deligiannakis et al., 1994; Petrouleas et al., 1994). In those cases, alternative signals from the Fe²⁺Q_A⁻ complex were reported (g = 1.98, alternate forms of g = 1.82, respectively). Upon addition of 120 mm of cyanide or 100 mm of formate, the g = 1.98 and g = 1.82 forms, respectively, remained unchanged. This lack of competition with other compounds that bind to the non-heme iron indicates that the effect of the active substance on PSII is not attributable to direct binding to the non-heme iron. Addition of 5 mm K₃[Fe(CN)₆] leads to an oxidation of the non-heme iron in the control. This oxidation was not possible when the lipophilic fraction was present (Table III; Fig. 4). This may indicate that the redox midpoint potential of the iron in PSII is shifted to a more positive value by tellimagrandin II or other phenolic compounds. Fluctuations of the midpoint redox potential of the non-heme iron and their effect on electron transfer from Q_A to Q_B are discussed in Deligiannakis et al. (1994).

Table III EPR signals of PSII membrane particles in the absence and presence of the lipophilic fraction of M. spicatum extract

Figure 4 Difference EPR spectra of the g = 8 and g = 5.6 EPR signal of the oxidized non-heme iron in control (A) and in the presence of lipophilic fraction of M. spicatum (B). EPR conditions: microwave frequency, 9.41 GHz; modulation amplitude, 25 Gauss; microwave (more ...)

We showed that lipophilic extracts of M. spicatum inhibit photosynthetic electron transport of cyanobacteria and other photoautotrophs. Polyphenols, allelopathically active secondary metabolites in this submerged aquatic angiosperm, were obviously responsible for a reduction in photosynthetic oxygen evolution of various axenic cultures of cyanobacteria as well as unialgal epiphytes isolated from Lake Constance.

SPE fractions containing tellimagrandin II, the major allelopathically active compound (Gross et al., 1996), caused the strongest inhibition of photosynthesis, indicating that this compound is, at least to a great part, responsible for the observed effect (Table I; Fig. 1). During the experiments, tellimagrandin II was not available in sufficient amounts to perform extended dose-response curves, because the substance is not commercially available. Tannic acid (synonym for penta- to decagalloyl-Glc) exhibited a dose-dependent inhibition of photosynthesis in Synechocystis sp. PCC 6803. Although we cannot fully exclude other, yet unknown, secondary metabolites to exert additional inhibitory activity toward photosynthesis, our results with purified extracts and commercially available polyphenols indicate that gallotannins are primarily responsible for the observed inhibition of photosynthesis.

We showed that the lipophilic fraction inhibits photosystem II activity but not PSI. The oxygen consumption of PSI in our assay decreased slightly in the presence of extract, but the difference was never statistically significant. With M. spicatum allelochemicals apparently affecting PSII, we could use TL to refine determination of the target site. TL is a very sensitive and highly specific method suitable to gain information about details of the mode of action of inhibitors within PSII (Vass and Inoue, 1992). The lipophilic extract from M. spicatum caused a shift of the B-band in thylakoid membranes and PSII particles similar to shifts observed with formate applied to purified PSII particles (Figs. 2 and 3). A comparable shift of the B-band was also induced by low concentrations (3 μm) of purified tellimagrandin II. The changes in TL were strictly different from that of commercial phenolic herbicides that bind to the Q_B-binding site, thereby blocking forward electron transport, resulting in the loss of the B-band and the formation of the Q-band (Vass and Demeter, 1982).

Formate, which causes a similar shift in TL as tellimagrandin II or lipophilic extracts from M. spicatum, binds to the non heme iron between Q_A and Q_B as shown by EPR spectroscopy (Vermaas and Rutherford, 1984). However, measurements of the Fe²⁺Q_A⁻ signal by EPR in the presence of tellimagrandin II did not reveal a direct interaction of this compound with the non-heme iron (Table II). This is not surprising, because tellimagrandin II is a much larger molecule (M_r 938) than formate (M_r 45), and it is very unlikely that this allelochemical directly binds to the non-heme iron. However, an effect of tellimagrandin II on the redox characteristics of the non-heme iron was shown. In the presence of extract, the oxidation of the non-heme iron by high K₃[Fe(CN)₆] concentrations was no longer possible, which indicates a rise in the midpoint potential of the non-heme iron (see Deligiannakis et al., 1994). This in turn can interfere with the electron transport between Q_A and Q_B and thus explains both the observed decrease in oxygen evolution and the shift of the maximum temperature in the TL measurements. Because ferricyanide interfered with the lipophilic extract as verified by HPLC analysis (data not shown), it was not used in Clark electrode measurements. In the EPR measurements, however, we used it in large surplus to guarantee that it was available in sufficient amount for the desired oxidation.

Most of our studies were performed with isolated spinach thylakoids or cyanobacteria from culture collections. The inhibition of photosynthetic oxygen evolution of an epiphytic cyanobacterium and a chlorophyte from Lake Constance indicates that extracts of M. spicatum act in the same way on naturally occurring primary producers. Further evidence for an ecologically important interference with photosynthesis of epiphytes and phytoplankton was recently demonstrated by others. Allelopathically active exudates from M. spicatum inhibited various cyanobacteria and algae (Körner and Nicklisch, 2002). This study showed, using PAM fluorometry, that allelochemicals from M. spicatum interfere with PSII of competing primary producers. Preliminary results in our laboratory indicate that lipophilic exudates of M. spicatum cause a shift of the B-band in TL comparable with the effect of extract or purified tellimagrandin II. Released polyphenols presumably interfere with photosystem II of algae and cyanobacteria. The release of various polyphenolic compounds by M. spicatum, among them tellimagrandin II, ellagic acid and catechin, was shown in several studies (Gross and Sütfeld, 1994; Gross et al., 1996; Nakai et al., 1999). Exudates from M. spicatum were shown to be growth inhibitory only when a constant exposure to freshly released allelochemicals was provided (Nakai et al., 1999).

In the present study we found that in particular, tellimagrandin II is responsible for the inhibition of PSII activity; a synergistic effect with other polyphenols is likely. At a first glance, a 40% inhibition of PSII activity may be considered weak. Full inhibition of PS by any allelopathically active compound might, however, favor the emergence of resistance in target species (Lopez-Rodas et al., 2001). We suggest that even a moderate reduction of photosynthetic activity may impair normal growth of co-occurring epiphytes and phytoplankton. Our measurements of oxygen evolution with intact cyanobacteria and a chlorophyte reveal that the allelochemicals can pass through cell membranes and reach their target sites. This suggests that such an inhibition is likely to occur in situ.

Anabaena sp. PCC 7120 and Anabaena variabilis sp. ATCC 29413, strain P9, were generally used as axenic indicator organism for photosynthetic inhibition caused by allelopathically active substances from M. spicatum. In addition, we used axenic Synechocystis sp. PCC 6803 and two unialgal epiphytes isolated from submersed macrophytes of Lake Constance (Synechococcus sp. Myr9801 and Chlamydomonas sp. Elo99–5B). Cultures were maintained in cyanobacteria medium (Jüttner et al., 1983) and grown in 300-mL Erlenmeyer flasks at 24°C under permanent light (100 μmol photons m⁻² s⁻¹) on a rotary shaker (100 rpm). Chl of cyanobacteria was extracted with methanol and the concentration was determined with an extinction coefficient of ε₆₆₆ = 6.58 × 10⁴ m⁻¹ cm⁻¹ (Ogawa and Vernon, 1971). The Chl content of Chlamydomonas sp. was determined according to Marker et al. (1980).

To distinguish effects on PSI or PSII, assays were conducted with thylakoid membranes or PSII-enriched membrane fragments from spinach (Spinacia oleracea). The preparation of thylakoid membranes was carried out according to Jensen and Bassham (1966), modified after Laasch (1987). PSII-enriched membrane fragments were prepared from spinach as described by Berthold et al. (1981) with modifications by Johnson et al. (1994). The Chl concentration was determined according to Arnon (1949).

This crude extract was either directly used in measurements of photosynthesis or was fractionated with SPE. Methanol was evaporated from the crude extract, and the remaining water phase was passed over a preconditioned C18-SPE-cartridge (Bond Elut, 500 mg of sorbens, Varian Medical Systems, Palo Alto, CA). Two different SPE fractionations were performed. The first method separated only between the aqueous eluent (named hydrophilic fraction thereafter) and the methanolic eluate containing the compounds adsorbed on the cartridge (named lipophilic fraction). The second procedure used a stepwise elution of the adsorbed compounds. The aqueous eluent was used as one fraction in subsequent assays. Adsorbed lipophilic compounds were then stepwise eluted with 25%, 50%, 75%, and 100% (v/v) methanol (water:methanol mixtures [v/v]). The different fractions were evaporated to dryness, resuspended in 50% (v/v) methanol (1 mL per 125 mg dry weight), and stored at −20°C until use.

Pure phenolic compounds related to polyphenols present in M. spicatum, gallic acid, tannic acid, or (+)-catechin (Gross et al., 1996; Nakai et al., 2000), were used in Clark-electrode and TL measurements, all dissolved in 50% (v/v) ethanol. Because tannic acid refers to several similar compounds (penta- to decagalloyl-Glc), no exact M_r can be provided for this substance. According to the Folin-Ciocalteau assay, our M. spicatum extracts contained between 160 and 170 μg TAE mg⁻¹ dry weight. In control experiments, we used the pure phenolic compounds in concentrations comparable with the TAE concentrations detected in M. spicatum extracts.

TL was measured with an apparatus as described by Krieger et al. (1998). Samples were incubated in the dark at 20°C for 120 s and cooled down to −5°C. At this temperature, a single-turnover flash was given, and the sample was then heated to 60°C with a rate of 0.4°C per s. For measuring the oscillation of TL intensity, saturating single-turnover flashes (1–6; frequency, 1 Hz) were given in subsequent assays. The area under the curve corresponds to the TL intensity and was determined using the curve fitting procedure according to Ducruet and Miranda (1992).

EPR spectroscopy was performed using a spectrometer (ER-300, Bruker, Rheinstetten/Karlsruhe, Germany), equipped with a cryostat (ESR 9, Oxford Biomedical Research, Oxon, UK), a frequency counter (5350B, Hewlett-Packard, Palo Alto, CA), and a Bruker 035 M NMR gaussmeter. Illumination at 200 K was performed in an ethanol/solid CO₂ bath in an unsilvered dewar flask, using a 300-W projector lamp. Typical Chl concentrations for the PSII-enriched membranes were 3 to 4 mg Chl mL⁻¹. All control samples were treated with the same amount of methanol. Although the concentrations of methanol used in this study did not interfere with PSII activity, they modify the spectroscopic properties of oxidation states of the Mn-cluster (e.g. see Messinger et al., 1997; Schansker et al., 2002).

We gratefully acknowledge technical assistance from Katharina Kienzler and Claudia Feldbaum, statistical guidance from Daniel Baumgärtner, and linguistic improvements from Raymond Newman. The manuscript further benefited from helpful discussions with Daniela Erhard, Peter Böger, and two anonymous peer reviewers. Christine Postius isolated strain Synechococcus sp. Myr9801 and Enikö Ivanyi isolated strain Chlamydomonas sp. Elo99-5B.