Iron limitation in the marine cyanobacterium Trichodesmium reveals new insights into regulation of photosynthesis and nitrogen fixation


  • Hendrik Küpper,

    1. Universität Konstanz, Mathematisch-Naturwissenschaftliche Sektion, Fachbereich Biologie, D-78457 Konstanz, Germany;
    2. University of South Bohemia, Faculty of Biological Sciences and Institute of Physical Biology, Branišovská 31, CZ-370 05 České Budejovice, Czech Republic;
    Search for more papers by this author
  • Ivan Šetlík,

    1. University of South Bohemia, Faculty of Biological Sciences and Institute of Physical Biology, Branišovská 31, CZ-370 05 České Budejovice, Czech Republic;
    2. Academy of Sciences of the Czech Republic, Institute of Microbiology, Department of Autotrophic Microorganisms, Opatovický mlýn, CZ-37981 Třeboň, Czech Republic;
    Search for more papers by this author
  • Sven Seibert,

    1. Universität Konstanz, Mathematisch-Naturwissenschaftliche Sektion, Fachbereich Biologie, D-78457 Konstanz, Germany;
    Search for more papers by this author
  • Ondrej Prášil,

    1. University of South Bohemia, Faculty of Biological Sciences and Institute of Physical Biology, Branišovská 31, CZ-370 05 České Budejovice, Czech Republic;
    2. Academy of Sciences of the Czech Republic, Institute of Microbiology, Department of Autotrophic Microorganisms, Opatovický mlýn, CZ-37981 Třeboň, Czech Republic;
    Search for more papers by this author
  • Eva Šetlikova,

    1. University of South Bohemia, Faculty of Biological Sciences and Institute of Physical Biology, Branišovská 31, CZ-370 05 České Budejovice, Czech Republic;
    2. Academy of Sciences of the Czech Republic, Institute of Microbiology, Department of Autotrophic Microorganisms, Opatovický mlýn, CZ-37981 Třeboň, Czech Republic;
    Search for more papers by this author
  • Martina Strittmatter,

    1. Universität Konstanz, Mathematisch-Naturwissenschaftliche Sektion, Fachbereich Biologie, D-78457 Konstanz, Germany;
    Search for more papers by this author
  • Orly Levitan,

    1. Bar Ilan University, Mina and Everard Goodman Faculty of Life Sciences, Ramat Gan, 52900, Israel
    Search for more papers by this author
  • Jens Lohscheider,

    1. Universität Konstanz, Mathematisch-Naturwissenschaftliche Sektion, Fachbereich Biologie, D-78457 Konstanz, Germany;
    Search for more papers by this author
  • Iwona Adamska,

    1. Universität Konstanz, Mathematisch-Naturwissenschaftliche Sektion, Fachbereich Biologie, D-78457 Konstanz, Germany;
    Search for more papers by this author
  • Ilana Berman-Frank

    1. Bar Ilan University, Mina and Everard Goodman Faculty of Life Sciences, Ramat Gan, 52900, Israel
    Search for more papers by this author

Author for correspondence:
Hendrik Küpper
Tel:+49 7531 884112
Fax:+49 7531 884533


  • • As iron (Fe) deficiency is a main limiting factor of ocean productivity, its effects were investigated on interactions between photosynthesis and nitrogen fixation in the marine nonheterocystous diazotrophic cyanobacterium Trichodesmium IMS101.
  • • Biophysical methods such as fluorescence kinetic microscopy, fast repetition rate (FRR) fluorimetry, and in vivo and in vitro spectroscopy of pigment composition were used, and nitrogenase activity and the abundance of key proteins were measured.
  • • Fe limitation caused a fast down-regulation of nitrogenase activity and protein levels. By contrast, the abundance of Fe-requiring photosystem I (PSI) components remained constant. Total levels of phycobiliproteins remained unchanged according to single-cell in vivo spectra. However, the regular 16-kDa phycoerythrin band decreased and finally disappeared 16–20 d after initiation of Fe limitation, concomitant with the accumulation of a 20-kDa protein cross-reacting with the phycoerythrin antibody. Concurrently, nitrogenase expression and activity increased. Fe limitation dampened the daily cycle of photosystem II (PSII) activity characteristic of diazotrophic Trichodesmium cells. Further, it increased the number and prolonged the time period of occurrence of cells with elevated basic fluorescence (F0). Additionally, it increased the effective cross-section of PSII, probably as a result of enhanced coupling of phycobilisomes to PSII, and led to up-regulation of the Fe stress protein IsiA.
  • • Trichodesmium survives short-term Fe limitation by selectively down-regulating nitrogen fixation while maintaining but re-arranging the photosynthetic apparatus.



fluorescence kinetic microscopy/microscope (for two-dimensional (imaging) measurements of fluorescence kinetics)


fast repetition rate fluorimeter


basic fluorescence yield of a dark-adapted sample, fluorescence in nonactinic measuring light


maximum fluorescence yield of a dark-adapted sample

inline image

maximum fluorescence yield of a sample during exposure to actinic light, i.e. diminished by nonphotochemical quenching

inline image

maximum fluorescence yield of a fully light-adapted sample at the end of the actinic light period of the measurement, diminished by nonphotochemical quenching

inline image

fluorescence yield under actinic irradiance immediately before the measurement of inline image


variable fluorescence yield

Fv = Fm − F0

i.e. response to a supersaturating flash in the dark-adapted state of photosystem II (PSII)


maximal efficiency of dark-adapted PSII (In this study, activity of PSII was measured using the variable fluorescence (Fv; see above) without the usual normalization to Fm (e.g. Maxwell & Johnson, 2000) because the latter value would be influenced by the fluorescence emitted by the uncoupled antenna that leads to the elevated F0; see Results and Küpper et al., 2004 for details)


photochemical quenching, measured as the difference between inline image and inline image, i.e. the response to a supersaturating flash during actinic light exposure (this was done without normalization, for the same reason as described for Fv vs Fv/Fm above)


nonphotochemical quenching, measured as the difference between the dark-adapted Fm and the light-adapted inline image, i.e. inline image (without the usual normalization to inline image for the same reason as described for Fv vs Fv/Fm above)


light-emitting diode


phosphate-buffered saline (buffer used for extracting phycobiliproteins)


Biological fixation of atmospheric nitrogen is performed by certain cyanobacteria when bioavailable forms of nitrogen (nitrate and ammonia) are in short supply. The nitrogen-fixing enzyme, nitrogenase, is irreversibly inactivated when exposed to molecular oxygen (reviewed by Postgate, 1998). Therefore, nitrogen-fixing (diazotrophic) cyanobacteria must prevent nitrogenase from being damaged by oxygenic photosynthesis (reviewed by Gallon, 1992, 2001; Bergman et al., 1997; Berman-Frank et al., 2003). Most diazotrophic cyanobacteria achieve this by separating photosynthesis and nitrogen fixation either spatially, by differentiating highly specialized cells called heterocysts, or temporally, by fixing nitrogen at night (usually found in unicellular diazotrophic cyanobacteria). By contrast, filamentous nonheterocystous marine cyanobacteria of the genus Trichodesmium execute both processes during the light period without irreversible differentiation of specialized cells. Trichodesmium is abundant and forms widespread blooms, thousands of kilometres wide, over the subtropical and tropical oceans, and contributes a larger fraction to the total marine nitrogen fixation than any other organism (Capone et al., 1997, 2005; Westberry & Siegel, 2006).

In Trichodesmium, active photosynthetic components (such as photosystem I (PSI) and PSII complexes, Rubisco and carboxysomes) are found in all cells, even those containing nitrogenase (Janson et al., 1994; Berman-Frank et al., 2001a). During nitrogen fixation, nitrogenase is expressed in a fraction of the cells that are often arranged consecutively along the trichome (Lin et al., 1998; Berman-Frank et al., 2001a). Protection against oxygen in Trichodesmium involves a complex interaction between reversible spatial and temporal segregation of photosynthesis, respiration and nitrogen fixation (Kana, 1993; Carpenter & Roenneberg, 1995; Chen et al., 1999; Berman-Frank et al., 2001a; Küpper et al., 2004). Earlier research demonstrated light- and PSII-dependent (3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU)-sensitive) oxygen consumption by the Mehler reaction (Kana, 1993; Berman-Frank et al., 2001a; Milligan et al., 2007) as well as an unusually high dark respiration rate (Kana, 1993; Carpenter & Roenneberg, 1995) suggesting that oxygen-consuming mechanisms protect nitrogenase. Measurements of chlorophyll (Chl) fluorescence kinetics yield comprehensive information about the regulation of photosynthesis in cyanobacteria (reviewed e.g. by Campbell et al., 1998).

Investigations of Chl fluorescence kinetics in cyanobacteria have led to the discovery and better understanding of daily activity cycles in Cyanothece sp. (Meunier et al., 1997, 1998; review by Sherman et al., 1998), Synechococcus sp. (Behrenfeld & Kolber, 1999) and Plectonema boryanum (Misra & Mahajan, 2000). The fluorescence kinetics of cyanobacteria differ in several ways from those of green plants and algae, mainly because of the presence of phycobilisomes in cyanobacteria instead of the light harvesting complex (LHC) II in Chlorophyta (Campbell et al., 1998). In particular, ‘state transitions’ between associations of the phycobilisomes with PSII (‘state I’) and associations with PSI (‘state II’) profoundly change many parameters of the fluorescence kinetics. In two previous studies, we utilized Chl fluorescence kinetic microscopy (FKM; Küpper et al., 2000) to resolve the spatial and temporal patterns of photosynthetic activity in Trichodesmium in relation to nitrogen fixation (Berman-Frank et al., 2001a; Küpper et al., 2004). Thus, we showed that Trichodesmium trichomes have a homogeneous high activity of PSII during most of the day, and show a reversible partial differentiation of cells for the period of nitrogen fixation. This partial differentiation involves a decline in oxygen production by enhancing the Mehler reaction correlated to a reversible change in PSII activity (Berman-Frank et al., 2001a; Küpper et al., 2004, Milligan et al., 2007). Chl fluorescence kinetic microscopy revealed that, during the period of high nitrogen fixation, some cells had a much higher basic Chl fluorescence yield (F0) than all the cells outside the diazotrophic period; these cells have been termed ‘bright zones/cells’ (Berman-Frank et al., 2001a). Rapid reversible switches between fluorescence levels were observed, which indicated that the elevated F0 of the bright cells originates from reversible uncoupling of PSII antenna proteins from the PSII reaction centre (Küpper et al., 2004). Two physiologically distinct types of bright cells were observed (Küpper et al., 2004). Bright I cells, with approximately double F0 compared with the normal F0 in the nondiazotrophic state, had a high PSII activity and were correlated with nitrogen fixation. Type II bright cells, by contrast, had more than three times the normal F0, exhibited hardly any PSII activity measurable by variable fluorescence and were not related to nitrogen fixation, but to stress. In addition to the two high-fluorescence states, cells were observed to reversibly enter a low-fluorescence state.

Biological nitrogen fixation is also controlled by the bioavailable iron (Fe), which is limiting in many regions of the world's oceans (reviewed e.g. by Morel & Price, 2003). The bioavailable Fe concentration in the oceans is still a matter of debate, as a consequence of the fact that not only free Fe (at the surface down to < 0.1 nM; Morel & Price, 2003) but also the much more abundant bound and colloidal Fe (high nM range) is often bioavailable, which was also recently investigated with Trichodesmium (Wang & Dei, 2003). Amongst the organisms most susceptible to Fe limitation are the photosynthetic diazotrophs, as a result of their high requirement for Fe in nitrogenase (19 Fe per nitrogenase), in addition to the Fe-containing proteins of the photosynthetic units. Thus, diazotrophs have higher intracellular Fe quotas than nondiazotrophic phytoplankton (Raven et al., 1999; Kustka et al., 2003; Tuit et al., 2004). The availability of Fe influences N2 fixation in cyanobacteria by its direct effect on Fe-rich protein synthesis of nitrogenase, and by effects on photosynthesis, growth, and global productivity (Paerl et al., 1987, Rueter et al., 1990; Falkowski 1997; Berman-Frank et al., 2001b; Fu & Bell, 2003). Thus we expect that the availability of Fe would further influence the regulation of photosynthesis for nitrogen fixation in Trichodesmium.

In the present study, we used newly available techniques to extend our understanding of the response of Trichodesmium to Fe limitation. Biophysical measurements were combined with assays of nitrogenase activity, Chl and carotenoid composition, western blot analysis of protein expression, and measurements of carbon, nitrogen, and phosphate content. These investigations were performed in Trichodesmium cultures grown in chemostats either at the normal or a continuously reduced Fe concentration.

Materials and Methods

Culture media and culture conditions

Cultures of Trichodesmium IMS101 were grown in YBCHK medium, with the following composition: 420 mM NaCl, 10 mM KCl, 20 mM MgCl2, 10 mM CaCl2, 25 mM MgSO4, 2.5 mM NaHCO3, 464 µM H3BO3, 780 µM KBr, 50 µM KH2PO4, 68 µM NaF, 25 µM LiCl, 2 µMRbCl, 1 µM FeNa-EDTA, 450 nM NaIO3, 80 nM Na2MoO4, 20 nM MnCl2, 7 nM ZnSO4, 7 nM NiSO4, 2.5 nM CoCl2 and 1 nM CuSO4, dissolved in redistilled water. The 1 : 1 ratio of Fe3+ and ethylenediaminetetraacetic acid (EDTA) in our experiments may have caused some formation of (invisible) Fe hydroxide precipitates, but this affected neither the controls (as judged by their healthy physiology and growth) nor the cultures with Fe limitation (as shown by the decrease and final cessation of growth under these conditions). Pre-cultivation was in batch cultures in glass tubes of 3 cm inner diameter and 250 ml volume. Altogether, we carried out three experiments with step-down type Fe removal in batch cultures and two experiments with gradual Fe removal in chemostat cultures. As the chemostat cultures with gradual Fe removal are closer to the natural situation, all data presented here are taken from these experiments. The step-down batch culture experiments yielded very similar results, however, in terms of filament fragmentation, in vivo single-cell and acetone extract absorption spectra, and maintenance of the photosynthetic apparatus.

In the step-down experiments, Fe was removed by filtering Trichodesmium filaments onto GF/F filters (Whatman;, washing them with Fe-free medium, and re-suspending them in Fe-free medium in the glass culture tubes. Because in such step-down experiments the cells died within 5–6 d, to simulate more natural conditions in the later experiments described in the Results we used chemostat cultures and a slow decrease of Fe concentration instead. From 3 wk before the experiments and throughout the experiments, the cultures were grown in 4-l flat glass chemostats at a flow rate of 0.5 to 1 l d−1 which maintained the Fe-replete cultures at an optical density (OD750) of 0.05. At the start of the experiments, the medium flowing into the low-Fe chemostats was exchanged for medium containing no Fe-EDTA, so that the Fe in the chemostat was gradually diluted out. The decrease of the Fe concentration was stopped after 2 wk by adding 50 nM Fe-EDTA into the medium pumped into the low-Fe chemostats. The chemostat experiments were performed twice, each replicate with one Fe-replete and one Fe-limited culture. The preliminary step-down batch culture experiments were carried out three times, again each replicate with one Fe-replete and one Fe-limited culture.

Both the tubes and the chemostats were aerated with air. The cultures were maintained with a 12 : 12 h light:dark cycle (light on from 09:00 to 21:00 h local time) and 27 : 25°C day:night temperature. The photon flux density during the light period followed a sinusoidal cycle simulating natural conditions, with a peak intensity of c. 300 µmol m−2 s−1 supplied by OSRAM® Dulux L 55W/12-950 (Osram, München, Germany; fluorescent tubes.

Sampling times

Samples for measurements of nitrogen fixation, pigment quantification, FKM analyses and fast repetition rate fluorimeter (FRRf) measurements were taken five times per day to resolve the changes in the daily activity pattern of Trichodesmium: at 08:15 h (just before the onset of the light period), 11:00 h (at the beginning of pigment synthesis in the morning), 14:15 h (in the middle of the light period and at the peak of nitrogen fixation), 17:00 h (during the decline of nitrogen fixation), and 21:15 h (directly after the end of the light period). Samples requiring very large amounts of culture were taken only three times per day (at 08:15, 14:15 and 21:15 h). This applied to the samples for western blots and carbon:nitrogen:phosphorus (C:N:P) analyses.

Fluorescence kinetic measurements at the single-cell level

Photosynthetic performance was analysed at the single-cell level using the updated and extended version (Küpper et al., 2007a) of a fluorescence kinetic microscope (FKM) originally described in Küpper et al. (2000), produced by Photon Systems Instruments (Brno, Czech Republic, The most important new features of the version of the FKM described in Küpper et al. (2007) are the ability to excite and detect fluorescence kinetics at various wavelengths accessible by computer-controlled filter wheels, and the recording of spectrally resolved, in addition to spatially resolved, (imaging) fluorescence and absorption kinetics as a consequence of the addition of a high-sensitivity fibre-optic spectrometer.

Preparation of samples for FKM measurements

Samples were prepared as previously described (Küpper et al., 2004) with a few modifications. Embedding of the living Trichodesmium filaments was carried out with 0.75% SeaKem Gold agarose (Cambrex BioScience Rockland, Inc., Rockland, ME, USA;, which yields a gel strength of > 1000 g cm−2 at this concentration. This high gel strength helped to reduce the movement of Trichodesmium filaments during the measurement. Temperature-controlled (27°C) air-saturated medium was pumped through the measuring chamber at a rate of 25 ml min−1. For each time-point of the daily activity cycle, fresh samples were prepared from the cultures in the chemostats.

FKM measurement parameters

All FKM measurements lasted 300 s. At the third second, a 600-ms pulse of saturating light (c. 4000 µmol m−2 s−1) was given for measurement of the maximum fluorescence yield of a dark-adapted sample (Fm). This was followed by 90 s of darkness, after which F0 was measured for 5 s (measuring light irradiance 5 µmol m−2 s−1, tested to be nonactinic for Trichodesmium). Then, 100 s of actinic light (c. 1000 µmol m−2 s−1) was applied to analyse Kautsky induction, and finally 100 s of measurement with no actinic light was used to measure dark relaxation. For the analysis of photochemical and nonphotochemical quenching, further 600-ms saturating pulses were applied during the actinic light exposure and in the relaxation period (4 pulses and 3 pulses, respectively). The excitation light in the range of 410–510 nm almost excited Chl and the short-wavelength form of phycoerythrin (phycourobilin =‘CU phycoerythrin’) as determined from the optical properties of Trichodesmium (Subramaniam et al., 1999). The measurements were performed with an automatic subtraction of background signals and a maximum time resolution of 40 ms. A lower time resolution was applied for the slower kinetics. Each image of the resulting fluorescence kinetic records had a resolution of 512 × 512 pixels at 4096 grey values (12 bits). The FKM was also used as a regular epifluorescence and bright-field microscope to observe the cultures and measure the length of the filaments. The fluorescence kinetic measurements were analysed as described in detail in Küpper et al. (2000, 2004, 2007a) using the FluorCam6 software from Photon Systems Instruments.

Integrative (population-level) biophysical measurements

Fast repetition rate measurements of Chl fluorescence kinetics  Fast repetition rate fluorometry (FRRf) via the ‘FIRe’ system (Satlantic Instruments, Halifax, Nova Scotia) was used to assess the effective cross-section of the PSII-associated antenna according to Strasser et al. (1999), the reoxidation rates of the quinone A in the PSII reaction centre (QA-), and photosystem connectivity (Kolber et al., 1998). Samples were measured without pre-concentration. Fluorescence excitation was via blue (Chl a and phycoruobilin exciting) light-emitting diodes (LEDs; excitation at 450 ± 30 nm), and emission was detected using a > 678-nm long pass filter combination.

Low-temperature spectroscopy  77-K fluorescence emission spectra were recorded using a laboratory-built portable spectrofluorometer. The sample was prepared by filtering Trichodesmium cells onto a GF/F filter (Whatman), mounting this into a filter holder, freezing it in liquid nitrogen, and immersing this sample in a custom-designed dewar at liquid nitrogen temperature. Spectra were recorded using the Avantes USB-2000 fibre optic spectrometer and the SpectraWin software (Avantes Corp., Eerbeek, the Netherlands). Excitation was performed using a white LED with filters transmitting between 450 and 550 nm, that is, exciting Chl, phycourobilin and phycoerythrin. The fluorescence signal was recorded using a 590-nm long pass filter.

Measurement of nitrogen fixation

Nitrogenase activity was assayed by the acetylene reduction method. Samples of 10 ml volume were taken from all three cultures at five time-points throughout the daily cycle. Samples were incubated in gas-tight glass vials, in which 10 ml of air (20% of total air volume) was exchanged for 10 ml of acetylene, for 1 h under the same light and temperature conditions as used for the cultures. Afterwards, 5 ml of the gas phase in the vials was taken out and analysed by gas chromatography on an FID-GC gas chromatograph (310 GC; SRI Instruments, Torrence, CA, USA). Control injections and analyses were performed with vessels containing pure YBCHK medium without cells in order to check for possible acetylene turnover not connected to Trichodesmium and for possible ethylene contamination in the air and/or acetylene.

Analysis of protein, Chl and carotenoid composition

FKM measurement of single-cell absorption spectra In vivo single-cell absorption spectra were recorded in the FKM using the white LED for transmittant light as the light source, as described in detail in Küpper et al. (2007a).

Analysis of Chl, carotenoids and phycobilisomes in extracts   Chlorophyll and carotenoids were extracted with 100% acetone after filtrating 80 ml of culture on 25-mm GF/F glass fibre filters (Whatman), freezing the filters in liquid nitrogen and then lyophilizing them. After 1 d of extraction at 4°C in the dark, absorbance spectra from 350 to 750 nm were recorded at 0.2-nm intervals at an optical bandwidth (slit) of 1 nm. From these spectra, carotenoids and Chl were analysed using the Gauss peak spectra method according to Küpper et al. (2000, 2007b).

Phycobiliproteins were extracted after completion of the acetone extraction described above, which along with the pigments removed the membranes from the cells. This pretreatment made the phycobilisomes easily accessible for extraction by incubation with phosphate-buffered saline (PBS: 40 g l−1 NaCl, 1 g l−1 KCl, 7.2 g l−1 Na2HPO4 and 1.2 g l−1 KH2PO4) for 1 d at 4°C in the dark. Spectra of these extracts were recorded in the same way as those of the acetone extracts.

Western blot analysis of protein expression  Samples were harvested by filtration on GF/F filters (Whatman), frozen in liquid nitrogen, and stored at −20°C until analysis. Total proteins were extracted from collected samples by grinding the frozen cells in liquid nitrogen and heating the cell suspension in preheated (60°C) extraction buffer composed of 125 mM Tris/HCl, pH 6.8, 4% (w/v) SDS, 200 µM phenylmethylsulfonyl fluoride (PMSF) (dissolved in 100% ethanol) and 100 mM dithiothreitol (DTT) for 10 min at 100°C after short vortexing. Cell debris was collected by centrifugation twice for 15 min at 25 000 g at room temperature. Proteins from the collected supernatant were precipitated by the addition of 4 volumes of 100% cold acetone during the incubation at −20°C for approx. 2 h. The proteins were collected by centrifugation for 15 min at 25 000 g at 4°C and the pellets were washed with 70% (v/v) ethanol. The protein pellets were dried and re-suspended in 1 : 100 diluted extraction buffer. The amount of protein was determined using the RC DC Protein Assay kit (BioRad, Hercules, CA, USA; with an optional precipitation step to improve accuracy as recommended in the manufacturer's manuals. Absorption was measured at 750 nm.

For western blotting, routinely 6 µg of total protein extract was loaded onto 10 or 15% Laemmli gels (Laemmli, 1970) before their transfer to Hybond-P membranes (Amersham Biosciences, Uppsala, Sweden) using a semi-dry blotting system (Biometra, Göttigen, Germany) as described in Towbin et al. (1979). Primary antibodies against D1 protein from PSII (anti-PsbA), nitrogenase (anti-NifH) and PsaC from PSI (anti-PsaC) were purchased from AgriSera (Vännäs, Sweden), and that against B-phycoerythrin (anti-B-PE) from Rockland Immunochemicals (Gilbertsville, PA, USA), and used at the dilutions recommended by suppliers. Blots were developed using horseradish peroxidase-coupled immunoglobulin G (IgG; Sigma-Aldrich, Hamburg, Germany) followed by chemiluminescence detection (ECL; Amersham Bioscience, Uppsala, Sweden) or using alkaline phosphatase-coupled IgG (Sigma-Aldrich) followed by Nitro Blue Tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) colorimetric reaction (Roche Diagnostics GmbH, Basel, Switzerland).

For quantification of signals in western blots, membranes and X-ray films were scanned and signals of interest were quantified by densitometry scanning. Although the chemistry behind any detection on western blots is not a linear function of protein abundance, the resulting numbers clearly do provide an indication of trends of increase/decrease.

Analysis of C:N:P ratios

Samples were harvested by filtration on pre-combusted (450°C for 4 h) and acid-washed GF/F filters (Whatman). After filtration the samples were lyophilized, and analysed for C:N and for P. The C:N analysis was performed on an elemental soil analyser using a thermoconductivity detector (NC 2110; ThermoQuest, Basingstoke, UK). P analysis was performed using a modified version of standard protocols for particulate P (Protocol from the Hawaii Institute of Marine Biology, Analytical Services laboratory at the University of Hawaii). The method relies on the release of organically bound P compounds as orthophosphate, by high-temperature and pressure combustion. The released orthophosphate reacted with a mixed reagent containing sulphuric acid, molybdic acid and trivalent antimony to form phosphomolybdic acid. The solution was reduced to a blue molybdenum complex by the ascorbic acid in the mixture, which was then measured spectrophotometrically (880 nm).


Growth parameters: cell number and filament length, nitrogen fixation, and C:N:P ratios

Cultures were initiated at an OD750∼ 0.01 as in our earlier studies, and we then let the cultures adjust their growth rate to the dilution rate of the chemostats, which was c. 0.18 d−1. For the Fe-replete culture, this equilibrium was reached at an OD750 of 0.5–0.6 (Fig. 1b). The filaments of the Fe-replete culture exhibited the phenotype characteristic of healthy Trichodesmium, that is, they were up to several millimetres long (Fig. 1a1). After 1 wk of Fe limitation, growth gradually slowed down and finally stopped (Fig. 1b), and the filaments fragmented and shortened (Fig. 1a2). Under transmitted light microscopy, no other morphological differences were observed between the Fe-limited and Fe-replete cultures (Fig. 1a3,1a4).

Figure 1.

Morphology and growth of the iron (Fe)-replete and the Fe-limited cultures. (a) Morphology of Trichodesmium filaments in the Fe-replete and Fe-limited cultures as observed using brightfield microscopy in the fluorescence kinetic microscope (FKM). Left, ×5 objective; right, ×40; top, Fe-replete; bottom, Fe-limited culture 18 d after the induction of Fe limitation, at c. 50 nM Fe(III)-ethylenediaminetetraacetic acid (EDTA). (b) Cell density, measured as optical density (OD750), of the Fe-limited cultures expressed as a percentage of the value for the Fe-replete cultures. Squares, control; circles, Fe-limited. The points represent the average and standard error of two independent replicate experiments for both conditions.

The Fe-replete culture with 1 µM total Fe in the medium showed high rates of nitrogen fixation. The maximum rate per volume of culture was reached in the late afternoon (Fig. 2a). Normalization to Chl (to compensate for the growth of the culture) shifted the maximum nitrogenase activity to the middle of the light period (Fig. 2b) as a result of Chl synthesis in the light period. Fe limitation very quickly inhibited nitrogen fixation. Nine days after starting to dilute out the Fe in the chemostat, when the Fe concentration declined to c. 200 nM (the total Fe content of the medium calculated from the initial concentration and the dilution rate), the nitrogen fixation per volume of culture was reduced to c. 25% of that of the Fe-replete culture (Fig. 2a) or 50% when normalized to Chl (Fig. 2b). Three days later, when Fe in the medium was reduced to ∼100 nM, nitrogen fixation of the Fe-limited culture declined to ∼10% of that of the Fe-replete culture, and became almost undetectable 16 d after induction of Fe limitation. Although the biomass of the Fe-limited culture continued declining, 20 d after induction of Fe limitation nitrogenase activity (normalized to Chl; Fig. 2b) or OD750 (not shown) increased again, almost reaching rates measured in the Fe-replete culture. Nevertheless, the Fe-limited cultures died 3 wk after induction of Fe limitation (Fig. 1b).

Figure 2.

Nitrogen fixation in iron (Fe)-replete and Fe-limited cultures of Trichodesmium. The Fe-replete cultures showed the typical daily trends; the data points shown represent the average and standard error of four measurements of two independent experiments. (a) Nitrogen fixation measured as nmol ethylene h−1 ml−1 culture. (b) Nitrogen fixation measured as nmol ethylene h−1µmol−1 chlorophyll (Chl).

Elemental analysis of carbon, nitrogen and phosphorus reflected the decreased nitrogen fixation with increasing C:N and a decrease in the N:P ratio (Fig. 3). While the N:P ratio declined continuously until 12 d after induction of Fe limitation, the C:N ratio remained steady at c. 130% of the control value (C:N ∼10.2; Fig. 3b) after 9 d of Fe limitation. This was caused by a slower decrease of the C:P ratio compared with the N:P ratio at the beginning of the experiment. From 12 d after induction of Fe limitation, the two parameters (i.e. N:P and C:P) decreased at about the same rate. At the end of the experiment, N:P was c. 28% of the control (N:P ∼9) and C:P was c. 39% of the control (C:P ∼93; Fig. 3).

Figure 3.

Effect of iron (Fe) limitation on carbon:nitrogen:phosphorus (C:N:P) ratios of Trichodesmium. (a) C:N:P ratios averaged over the whole experiment for the two Fe-replete cultures and between 9 and 20 d after the start of Fe limitation for the two Fe-limited cultures. Thus, the means and standard errors shown are from 28 replicate samples for the Fe-replete (cross-hatched bars) and 34 samples for the Fe-limited (hatched bars) cultures. (b) Development of C:N:P ratios during progressive Fe limitation. For the day 0 level, again all values for the Fe-replete cultures are averages (i.e. n = 28); for all further time-points all samples taken on that day were averaged (n = 1–5). Squares, C:N; circles, N:P; triangles, C:P.

Changes of protein and pigment composition

FKM measurements of single-cell absorption spectra   Mea-surement of absorption spectra of individual living cells in the FKM allows direct monitoring of cellular pigment changes without any artefact-prone sample preparation. Fe limitation led to a slight increase in the total content of Chl (peaks at 680 and 435 nm minus the reference point at 750 nm) and carotenoids (shoulder at 465 nm) per cell (Fig. 4a). Phycoerythrin content (peaks at 542 and 568 nm minus the valley at 593 nm) remained constant. Phycourobilin content per cell, measured from the peak amplitude at 495 nm to the local minimum before it, was higher in the Fe-limited cultures compared with the Fe-replete cultures (Fig. 4a).

Figure 4.

Effect of iron (Fe) limitation on pigment composition. (a) Single-cell in vivo absorption spectroscopy of Trichodesmium filaments. Each graph represents the average of all single-cell spectra measured at all five time-points of the daily activity cycle, and the average of the two replicate experiments. For the absorption spectrum of the Fe-replete cells, additionally all days of the measurement were averaged. Background was corrected only by setting optical density (OD750) to zero. No normalization was performed. (b) Analysis of chlorophyll (Chl) and carotenoid composition in acetone extracts. Averages and standard errors of all samples measured between 12 and 20 d after the start of Fe limitation are shown; altogether, six samples from two independent experiments for each data point are shown. Values for ‘chlorophylls’ are the sum of Chl a and all Chl a degradation products (chlorophyllid a, pheophytin a, and oxidation products). (c) Changes in phycobilisome composition of Trichodesmium under Fe limitation analysed using absorption spectra of extracts in phosphate-buffered saline. Each spectrum shown represents the average of five or 10 samples taken at five different time-points of the daily cycle. As the single-cell in vivo spectra, where artefacts of extraction can be excluded, showed constant phycoerythrin per cell (see a), the spectra shown here are normalized to the phycoerythrin peak of the Fe-replete cultures.

Analysis of chlorophyll, carotenoids and phycobiliproteins in extracts   Partial separation of pigments into an acetone extract for Chl plus carotenoids and a PBS extract for phycobiliproteins, combined with narrower absorption peaks in solution, allowed a quantitative analysis of Chl and carotenoids. Moreover, the PBS extract enabled better assessment of relative changes between phycobiliproteins as a result of the removal of overlapping absorption of Chl and carotenoids.

The acetone extracts revealed a slow decrease of the Chl:carotenoid ratio until 1 d before the end of the experiments, and a steep decrease of this ratio on the last day of the experiment (Fig. 4b). The most dramatic change induced by Fe limitation, however, was the appearance of high concentrations of keto-carotenoids (mainly echinenone), which were only a very minor component in Fe-replete cultures (Fig. 4b).

As no direct normalization per cell is possible for extracts, we used the observation from the in vivo spectra (see above) showing constant phycoerythrin content throughout the experiment, and normalized the PBS extract spectra to this peak. Spectra of PBS extracts differed from the single-cell in vivo spectra mainly in two ways. First, while in the in vivo spectra phycourobilin increased only slightly (Fig. 4a), in the PBS extracts a peak at c. 504 nm, close to the normal absorption of phycourobilin (497 nm), dominated the spectra 18 and 20 d after the start of Fe limitation (Fig. 4c). Additionally, the PBS extract spectra showed that phycocyanin (peak at 620 nm) decreased under Fe limitation (Fig. 4c). Phycocyanin could not be analysed in the single-cell absorption spectra because of overlapping of the phycocyanin peak with the side peak of the red absorbance band (Qy band) of Chl a.

Western blot analysis of protein expression  In the Fe-replete cultures dinitrogenase reductase NifH was consistently abundant throughout the 19 d of sampling, with the highest expression level assayed in the early afternoon (Fig. 5a). Strongly reduced levels of NifH were found in samples collected in the mornings and evenings, which is consistent with the diurnal rhythms reported for this protein previously (Church et al., 2005). By contrast, Fe limitation led to a very rapid decline of NifH levels, corresponding to declining nitrogenase activity (compare Figs 2 and 5). An approximately 60% lower NifH level was measured as early as the 9th day in Fe-limited culture collected in the early afternoon as compared with the Fe-replete controls (Fig. 5a,b). Only traces of this protein were detected in cultures exposed to prolonged Fe limitation. Yet, in the Fe-limited cultures, both NifH expression and in vivo nitrogenase activity partially recovered at the end of the experiments. Western blots with antibodies against subunit C of PSI (PsaC; Fig. 5a) and the D1 protein of PSII (not shown), both containing Fe as a cofactor, showed no measurable change in the amount of these proteins under Fe-limitation stress (Fig. 5a,b). The band of PsaC is at c. 47 kDa, indicating that in Trichodesmium this protein forms covalently linked trimers, which have also been found in many other organisms. In accordance with the in vivo absorption spectra (Fig. 4a), phycoerythrin expression remained constant until c. 12 d after the start of Fe limitation. On the following 8 d of the experiments, however, the amount of regular (16 kDa) phycoerythrin decreased and an additional, larger (20 kDa) band reacting with the phycoerythrin antibody appeared (Fig. 5a,b). On the last day of the experiments, only the 20-kDa band was present in immunoblots with phycoerythrin antibody (Fig. 5a,b), while the single-cell in vivo spectra (Fig. 4a) still showed the normal phycoerythrin absorption peak, and a phycourobilin-like peak was found in the PBS extract spectra (Fig. 4c).

Figure 5.

Western blot analysis of protein expression under iron (Fe) limitation stress in Trichodesmium. (a) Western blot analysis using antibodies directed against dinitrogenase reductase (NifH), subunit C of photosystem I (PSI) (PsaC) and phycoerythrin (PE). Equal protein was loaded per lane (6 µg total protein). Samples were taken daily just before the onset of the light period (M, morning), in the middle of the light period (A, afternoon), and directly after the end of the light period (E, evening). The days are separated by vertical lines on the blot. The evening sample of the 8th day was lost. The asterisk marks the 20-kDa protein band that accumulated during Fe limitation and reacted with phycoerythrin antibody. (b) Quantification of western blot signals shown in (a). The maximal value of control and Fe-limited samples was set as 100%. Although the chemistry behind any detection on western blots is not a linear function of protein abundance, the results of background-corrected densitometric scanning shown here clearly do provide an indication of trends of increase/decrease.

Biophysics of photosynthesis

FKM measurements  The imaging FKM measurements were used to analyse the daily cycle of photosynthetic activity of the cells in terms of basic fluorescence yield (F0), maximal dark-adapted variable fluorescence, photochemical quenching in actinic light, and nonphotochemical quenching. As in Trichodesmium F0 can change independently of variable fluorescence, changes in F0 can lead to artefactual changes in activity parameters that are normalized to F0 or another parameter that contains F0 (e.g. Fm, the maximum fluorescence yield of a sample during exposure to actinic light (inline image) and fluorescence under actinic irradiance immediately before the measurement of inline image (inline image)) (Küpper et al., 2004). Thus, as in our previous work (Küpper et al., 2004) and as done also by Behrenfeld & Kolber (1999), we did not normalize to Fm or inline image for calculating fluorescence quenching parameters. As the values of our nonnormalized parameters are recorded under constant conditions (irradiance, measurement geometry and detector), variations that render them unreliable in certain other experimental conditions (e.g. measurements with leaf clips, flexible fibre optics, etc.) do not apply in our case. For the FKM measurements, these data even have an inherent normalization that is not influenced by the changes in F0, because the values as shown in our figures are per cell.

Basic fluorescence yield (F0)   Because of slightly narrower emission filters (short-wavelength cut-off at c. 660 nm in this study, compared with 650 nm previously), in the current study allophycocyanin (max. at c. 670 nm) and phycocyanin (max. at c. 650 nm) fluorescence contributed less to the signal measured in the imaging kinetics compared with our earlier studies (Berman-Frank et al., 2001a; Küpper et al., 2004). As it is probably the phycobilisomes that lead to changes in F0 (Küpper et al., 2004), their reduced contribution to the measured F0 signal made the groups with elevated F0 less distinctly separated here compared with our earlier studies (Berman-Frank et al., 2001a; Küpper et al., 2004). In addition, the numerical values of F0 and all other nonnormalized parameters are higher by a factor of c. 40 in the current study because of the different measuring camera used. Based on the histogram of all measured cells (Fig. 6), in the current study the levels of F0 defined by Küpper et al. (2004) were found to be the following (relative fluorescence yields): ‘low F0/quenching’: < 310; ‘normal’: 310–490 (= classical state II); ‘bright I’ (= classical state I): 490–825; ‘bright II/very bright’: > 825.

Figure 6.

Results of fluorescence kinetic microscope (FKM) measurements of Trichodesmium. (a) Histogram of the basic fluorescence yield (F0) of all cells measured in the Fe-replete cultures showing the separation into groups of different F0, for comparison with our previous study (Küpper et al., 2004). (b–e) Daily pattern of relative frequencies the values of fluorescence kinetic parameters. The numbers at the colour scales show the relative frequencies represented by each colour. The numbers labelling the y-axis of the panels indicate the value of the parameter. The label ‘_i’ after a parameter indicates that it was measured after acclimation to actinic irradiance. Left column of colour maps, Fe-replete cultures; right column, Fe-limited (cells measured between 12 and 20 d after the start of Fe limitation). (b) F0; (c) variable fluorescence (Fv); (d) photochemical quenching after acclimation to actinic irradiance (Fqp_i); (e) nonphotochemical quenching after acclimation to actinic irradiance (Fqnp_i).

The daily pattern of F0 of the Fe-replete culture differed slightly from the pattern observed before (Berman-Frank et al., 2001a; Küpper et al., 2004). Groups of bright I and quenching cells were found not only around noon, but also at other times of the day (Fig. 6b, left panel). This is probably related to the rather broad maximum of nitrogen fixation in the current experiments (Fig. 2).

Fe limitation stress caused three major differences in the daily cycle of F0 (Fig. 6b, right panel). First, the appearance of bright I cells (F0 of c. 490 to 825) was rare compared with the Fe-replete culture and they were almost absent during the time of the day when nitrogen fixation occurs. This concurs with the results of the nitrogenase activity measurements, which showed almost no nitrogen fixation from 2 wk of Fe limitation onwards. Secondly, bright II cells appeared throughout the daily cycle except in the early dark period, and not only in the evening as observed in the Fe-replete cultures. On several occasions, sudden lysis of such cells was observed. Finally, in the Fe-limited culture we detected only a few low-fluorescence cells in the middle of the light period.

Dark-adapted maximal variable fluorescence (Fv = Fm − F0)   The Fv of the Fe-replete culture (Fig. 6c, left panel) showed the typical pattern known for Trichodesmium (Berman-Frank et al., 2001a; Küpper et al., 2007a), with high values in the nondiazotrophic period and low values, for most of the cells, in the diazotrophic period. The group of cells with high Fv values in the diazotrophic period turned out to be bright I cells. In the Fe-limited culture, this pattern was lost almost completely (Fig. 6c, right panel). Practically all cells showed intermediate Fv values throughout the day, and only very few cells with the characteristics of bright I or quenching cells were found in the light period.

Photochemical quenching under actinic light inline image  Photochemical quenching (Fig. 6d) under actinic light showed slight changes in the daily activity cycle compared with F0 and Fv. Only a few cells with decreased or increased Fqp were found in the light period, and a similar pattern was found in the Fe-limited culture.

Nonphotochemical quenching inline image  Compared with Fqp, much larger fluctuations in the daily activity cycle, and as a result in Fe limitation, were found in nonphotochemical changes of fluorescence yield (Fig. 6e). In the Fe-replete culture, strong Fqnp was found in the dark period, which steeply declined towards the middle of the light period. This daily cycle was almost completely lost in the Fe-limited cultures.

77-K fluorescence emission spectroscopy  The emission spectra showed drastic changes induced by Fe limitation (Fig. 7). In the Fe-replete cultures, the highest peak obtained was for Chl (PSII) emission at c. 695 nm, with a smaller contribution seen from phycocyanin at 620–650 nm and a side peak that is usually attributed to PSI at 750 nm. Under low Fe, we observed a clear shift of the main emission peak from 695 to 684 nm and a dramatic increase of phycocyanin bands at 645 and 655 nm.

Figure 7.

77-K fluorescence emission spectra of Trichodesmium. Average spectra found under iron (Fe)-replete (solid line; whole experiment) and Fe-limited (dotted line; 15–20 d after start of Fe limitation) conditions are shown. The peak at 695 nm in the Fe-replete cultures is photosystem II (PSII) to which the spectra were normalized, while the strong 685-nm peak in the Fe-limited cultures is the PSI-associated chlorophyll (Chl) protein IsiA.

Fast repetition rate Chl fluorescence kinetic measurements   FRRf measurements yielded information about the functional cross-section of the PSII light-harvesting antenna (Fig. 8). In the Fe-replete cultures, FRRf measurements showed an increase in the functional antenna cross-section during the light period and a larger functional cross-section of the PSII antenna in the Fe-limited compared with the Fe-replete cultures; in other words, a decreased ratio of reaction centre to antenna complexes connected to PSII. Additionally, F0 and Fv values obtained by FRRf were used as a comparison to the single-cell spatially resolved measurements obtained with the FKM. The average F0 and Fv of all cells measured in the FKM (discussed in detail above) correlated accurately with the results from integral measurements using the FRR fluorimeter (compare Figs 6 and 8).

Figure 8.

Effect of iron (Fe) limitation on daily patterns in photosynthetic parameters of Trichodesmium measured by fast repetition rate fluorescence. (a, c) Fe-replete cultures; (b, d) Fe-limited cultures; (a, b) parameters of fluorescence yield (basic fluorescence yield (F0), squares; maximum fluorescence yield (Fm), circles; variable fluorescence yield (Fv), triangles); (c, d) the functional absorption cross-section of photosystem II (PSII) (σPSII, squares).


This study has revealed a strategy employed by Trichodesmium to survive phases of Fe limitation. It involves selectively down-regulating nitrogenase while maintaining and modifying PSI, PSII and phycobilisomes via the coupling of photosynthetic components, expression of IsiA and expression of an alternative isoform of phycoerythrin.

Most studies of Fe limitation in Trichodesmium cultures to date, including preliminary experiments of our own (see Materials and Methods) and a newly published study by Shi et al. (2007), have applied a sudden step-down type removal of Fe, leading to rather quick (usually within 5–6 d) death of the cells. These experiments do not necessarily reflect the physiological behaviour in nature, where a gradual decrease of the available Fe by dilution or Fe uptake into cells would be more typical. Therefore, in this study we investigated the effects of Fe limitation on Trichodesmium in long-term (3-wk) experiments combining steady-state growth in chemostat cultures and a gradual removal of the Fe.

Fe limitation led to a dramatic decrease in filament length and a decline in the growth rate of the cultures during both step-down and gradual decrease of Fe (Fig. 1). This may be caused simply by the starvation of the cells, rendering the filaments more fragile. However, it may form part of the survival strategy of Trichodesmium, as it increases the surface/volume (S/V) quotient and thus may enhance the uptake of Fe and other nutrients. Increasing the S/V quotient for improved nutrient uptake is a well-known strategy to cope with nutrient limitation, and was recently found also in the related cyanobacterium Oscillatoria, where the quantitative increase of the S/V quotient by filament fragmentation was calculated (Kruskopf & Du Plessis, 2006). Filament fragmentation as an emergency survival strategy could be the evolutionary adaptive value of the programmed cell death (PCD) that was recently found in Trichodesmium under Fe and phosphorus limitation and high irradiance (Berman-Frank et al., 2004).

More surprising were the differential effects of Fe limitation on the processes of nitrogen fixation and photosynthesis. Both processes require large amounts of Fe. The nitrogenase complex contains 50 Fe atoms: 2 × 7 for the Fe7MoS9 cluster and 2*8 for the P cluster in the α2β2-heterotetramer of the FeMo-protein (= dinitrogenase) and 1*4 for the Fe4S4 cluster in the γ2-homodimer of the Fe-protein (= dinitrogenase reductase), the latter usually being present in a 5 : 1 ratio to the FeMo-protein (Kustka et al., 2003). For the photosynthetic electron transport chain, 23 Fe atoms are needed: 12 in PSI, three in PSII, one in cytochrome c and six in the Cytb6/f complex (see e.g. Raven et al., 1999; Kustka et al., 2003). However, only nitrogenase was strongly and rapidly down-regulated under Fe limitation. This down-regulation applied both to the activity and protein level (Figs 2, 5) and to the mRNA abundance (Shi et al., 2007). Photochemical activity, by contrast, remained high, as shown by both photochemical yield of dark-acclimated PSII and photochemical quenching in actinic light (Figs 5, 8). Moreover, no changes were found in the protein levels of the Fe-requiring components of PSI or PSII (Fig. 5). Additionally, no down-regulation of the nitrogen-requiring phycobilisomes was observed even after prolonged Fe limitation that ultimately led to the death of the cultures (Figs 4, 5), which contrasts with results for other cyanobacteria (e.g. Sandström et al., 2003).

As a result of the maintained photosynthetic activity and down-regulated nitrogen fixation, the C:N ratio of the cells increased and the N:P ratio decreased more rapidly than the C:P ratio decreased, consistent with earlier findings (Berman-Frank et al., 2001b). A slow but steady decrease of the C:P ratio was also observed (Fig. 3), indicating a reduced efficiency of photosynthetic carbon assimilation despite the maintenance of all components of the photosynthetic light reactions. This may be related to Fe-stress-induced changes in photosynthesis that were shown by the biophysical measurements, that is, loss of most of the diurnal cycle of activity (Figs 6, 8) and coupling of much of the antenna to PSII, leading to the observed increase in antenna cross-section (Fig. 8).

This strategy of selectively down-regulating nitrogenase and nitrogen fixation while maintaining photosynthetic capacity under Fe deficiency could allow Trichodesmium to survive short-term Fe limitation stress. Trichodesmium inhabits oligotrophic subtropical and tropical areas characterized by periodic pulses of Fe supply via atmospheric deposition or upwelling events and reduced Fe availability between pulse events. Such fluctuations of Fe supply were described for example by Gao et al. (2001) and their relation to Trichodesmium blooms by Westberry & Siegel (2006). During such periods of Fe limitation, Trichodesmium may benefit from reducing the energetically and Fe-expensive processes of nitrogen fixation and biomass growth in order to save the energy-delivering photosynthetic apparatus. The degradation of nitrogenase would yield an emergency supply of Fe and nitrogen for the remaining metabolism. In this early phase of Fe starvation, the kinetic properties of photosynthesis shifted to accommodate the different metabolic requirements. The distribution of cells in different activity states (characterized by F0, Fv, Fqp and Fqnp as described by Küpper et al., 2004) throughout the daily cycle resembled the nondiazotrophic state that is only found in Fe-replete cells in the early morning before the diazotrophic period (Fig. 6).

As a defence strategy against the inhibition of metabolism by Fe limitation, most cyanobacteria express proteins involved in Fe uptake and intracellular Fe distribution. In contrast to other photosynthetic organisms, however, they also express the Chl-binding Fe-stress-induced protein A (IsiA) in response to Fe starvation (reviewed e.g. by Michel & Pistorius, 2004). IsiA forms large highly ordered aggregates either around PSI or not connected to any photosystem, and seems to have a dual function: light harvesting and dissipation of excess excitons (Yeremenko et al., 2004). IsiA expression was also found in the current study, both from the strong up-regulation of echinenone (Fig. 4b), which is a cofactor of IsiA (Ihalainen et al., 2005; UniProtKB/Swiss-Prot entry Q55274), and from the appearance of the IsiA-characteristic emission peak at 684 nm that dominated the 77-K fluorescence emission spectra of Fe-limited cells (Fig. 7). The latter feature has long been known to be a characteristic feature of IsiA-containing photosystems as they specifically occur under Fe-limitation stress in cyanobacteria (e.g. Öquist 1971, 1974a,b; review by Michel & Pistorius, 2004). The very high intensity of the 685-nm IsiA fluorescence peak combined with its almost complete overlap with the emission band of PSII reaction centres at 695 nm clearly makes the estimation of PSII:PSI ratios based on 77-K spectra impossible. As this problem was not generally known, except to those people working directly on IsiA, it caused artefactually high PSII:PSI ratios in past Fe-limitation studies (e.g. Raven et al., 1999; Berman-Frank et al., 2001b).

When nitrogen starvation of the cells became too severe to continue the strategy described above, the cells up-regulated nitrogenase as a last resort (Figs 2, 5). As photosynthesis was still maintained, with constant levels of PSI as measured by PsaC, the Fe required for nitrogenase assembly must have come from the degradation of other Fe-containing proteins. According to our results, this ultimate attempt to acclimate to Fe deficiency involved an as yet unknown type of restructuring of the phycobilisomes. Although absorption per cell remained almost unchanged, with only a slight shift from normal phycoerythrin to the phycoerythrin derivative phycourobilin, the analysis of protein expression revealed the appearance of a 20-kDa protein band reacting with the phycoerythrin antibody, which did not react with any other band at the end of the experiments. Therefore, it appears likely that this band represented another phycoerythrin isoform that progressively replaced the regular 16-kDa isoform after the re-induction of nitrogenase at the end of our experiments. Judging from the decreasing intensity of the bands on western blots (Fig. 5b) compared with the constant absorption revealed by single-cell spectra, it seems probable that this putative isoform of phycoerythrin contains more chromophores per amount of protein backbone, which may help the cells synthesize nitrogenase despite the lack of previous nitrogen fixation. The features of this novel phycoerythrin isoform will be a subject of future research.


We are very grateful to Martin Trtílek (Photon Systems Instruments, Brno, Czech Republic) for continuous support and updates of the fluorescence kinetic microscope and the FluorCam software, and to Miloslav Šimek (Academy of Sciences of the Czech Republic, Biology Centre, Institute of Soil Biology) for gas chromatography analysis of the nitrogenase activity samples. This research was supported by grants from the Deutsche Forschungsgemeinschaft (KU 1495/4-1 to HK and El 179/4-4 to IA) and a collaborative linkage grant from NATO (EST.MD.CLG 981009) to IA, IBF, HK and OP. Grant MSMT 1P05ME824 from the Academy of Sciences of the Czech Republic provided further support to OP, IS and ES, and OL was further supported by an Israeli Ministry of Science Fellowship for Women in Science. The portable emission spectrofluorometer was developed in the laboratory of OP with support via grant 1QS500200570 from the Academy of Sciences of the Czech Republic. The research in the laboratory of OP is supported by the institutional research concepts AV AV0Z50200510 and MSM6007665808. We are grateful for a generous donation of various research instruments from the Degussa AG (formerly Hüls AG, Marl) to HK.