Search

One of the main problems with classifier training for imbalanced data is defining the correct learning criterion. On the one hand, we want the minority class to be correctly recognized, and on the other hand, we do not want to make too many mistakes in the majority class. Commonly used metrics focus either on the predictive quality of the distinguished class or propose an aggregation of simple metrics. The aggregate metrics, such as Gmean or AUC, are primarily ambiguous, i.e., they do not indicate the specific values of errors made on the minority or majority class. Additionally, improper use of aggregate metrics results in solutions selected with their help that may favor the majority class. The authors realize that a solution to this problem is using overall risk. However, this requires knowledge of the costs associated with errors made between classes, which is often unavailable. Hence, this paper will propose the semoos algorithm - an approach based on multi-objective optimization that optimizes criteria related to the prediction quality of both minority and majority classes. semoos returns a pool of non-dominated solutions from which the user can choose the model that best suits him. Automatic solution selection formulas with a so-called Pareto front have also been proposed to compare state-of-the-art methods. The proposed approach will train a svm classifier ensemble dedicated to the imbalanced data classification task. The experimental evaluations carried out on a large number of benchmark datasets confirm its usefulness.

Colloidal quantum dots (QDs) are nanoparticles that are able to photoreduce redox proteins by electron transfer (ET). QDs are also able to transfer energy by resonance energy transfer (RET). Here, we address the question of the competition between these two routes of QDs' excitation quenching, using cadmium telluride QDs and cytochrome c (CytC) or its metal-substituted derivatives. We used both oxidized and reduced versions of native CytC, as well as fluorescent, nonreducible Zn(II)CytC, Sn(II)CytC, and metal-free porphyrin CytC. We found that all of the CytC versions quench QD fluorescence, although the interaction may be described differently in terms of static and dynamic quenching. QDs may be quenchers of fluorescent CytC derivatives, with significant differences in effectiveness depending on QD size. SnCytC and porphyrin CytC increased the rate of Fe(III)CytC photoreduction, and Fe(II)CytC slightly decreased the rate and ZnCytC presence significantly decreased the rate and final level of reduced FeCytC. These might be partially explained by the tendency to form a stable complex between protein and QDs, which promoted RET and collisional quenching. Our findings show that there is a net preference for photoinduced ET over other ways of energy transfer, at least partially, due to a lack of donors, regenerating a hole at QDs and leading to irreversibility of ET events. There may also be a common part of pathways leading to photoinduced ET and RET. The nature of synergistic action observed in some cases allows the hypothesis that RET may be an additional way to power up the ET.

Plastoglobules (PGs) might be characterised as microdomains of the thylakoid membrane that serve as a platform to recruit proteins and metabolites in their spatial proximity in order to facilitate metabolic channelling or signal transduction. This study provides new insight into changes in PGs isolated from two plant species with different responses to chilling stress, namely chilling-tolerant pea (Pisum sativum) and chilling-sensitive bean (Phaseolus coccineus). Using multiple analytical methods, such as high-performance liquid chromatography and visualisation techniques including transmission electron microscopy and atomic force microscopy, we determined changes in PGs’ biochemical and biophysical characteristics as a function of chilling stress. Some of the observed alterations occurred in both studied plant species, such as increased particle size and plastoquinone-9 content, while others were more typical of a particular type of response to chilling stress. Additionally, PGs of first green leaves were examined to highlight differences at this stage of development. Observed changes appear to be a dynamic response to the demands of photosynthetic membranes under stress conditions.

In this study, we investigated an experimental and Monte-Carlo computational characterization of self-assembled antennae built using CdTe colloidal quantum dots (QDs). These clusters provide efficient excitation of phycocyanine (PC) or phycobilisomes (PBSs). PBSs are light-harvesting complexes (LHCs) of cyanobacteria, made of several PC units, organized in disks and rods. Each PC contains three separate cofactors. Therefore, we analyzed variations in multi-donor and multi-acceptor systems. The self-assembled QD clusters were formed mostly by electrostatic interactions, possibly due to the introduction of a positive charge on an originally negatively charged nanoparticle surface. Our results suggest that PC may accept energy from multiple nanoparticles localized at a distance significantly longer than the Forster radius. The excitation transfers between particular nanoparticles with possible delocalization. The maximal energy transfer efficiency was obtained for the PC/PBS : QD ratio from 1 to 20 depending on the QD size. This cannot be fully explained using computational simulations; hence, we discussed the hypothesis and explained the observations. Our self-assembled systems may be considered for possible applications in artificial light-harvesting systems because absorption spectra of QDs are different from the absorption characteristics of PC/PBS. In addition, huge clusters of QDs may effectively increase the optical cross-section of so-created nanohybrids.

While quantum dots (QDs) are useful as fluorescent labels, their application in biosciences is limited due to the stability and hydrophobicity of their surface. In this study, we tested two types of proteins for use as a cover for spherical QDs, composed of cadmium selenide. Pumilio homology domain (Puf), which is mostly α-helical, and leucine-rich repeat (LRR) domain, which is rich in β-sheets, were selected to determine if there is a preference for one of these secondary structure types for nanoparticle covers. The protein sequences were optimized to improve their interaction with the surface of QDs. The solubilization of the apoproteins and their assembly with nanoparticles required the application of a detergent, which was removed in subsequent steps. Finally, only the Puf-based cover was successful enough as a QD hydrophilic cover. We showed that a single polypeptide dimer of Puf, PufPuf, can form a cover. We characterized the size and fluorescent properties of the obtained QD:protein assemblies. We showed that the secondary structure of the Puf proteins was not destroyed upon contact with the QDs. We demonstrated that these assemblies do not promote the formation of reactive oxygen species during illumination of the nanoparticles. The data represent advances in the effort to obtain a stable biocompatible cover for QDs.

Nonlinear optical properties of colloidal CdSe quantum dots conjugated with proteins were investigated in a wide spectral range with the two-photon excited emission technique using a tunable femtosecond laser system in the range from 700 to 1000 nm. The most significant value of the two-photon absorption cross section σ2 for 2.9 nm size hydrophilic CdSe quantum dot was equal to 4 505 GM, while the two-photon excitation action cross section σ2·QY was found to be 0.101 GM at 900 nm. The nonlinear absorption properties are present as appropriate cross sections normalized per molecular weight which allows comparison of the nonlinear properties of the studied quantum dots with various nanosystems or organic dyes. The properties of these proteins covering hydrophilic colloidal quantum dots can be potentially useful for nonlinear bioimaging.

The imbalanced data classification remains a vital problem. The key is to find such methods that classify both the minority and majority class correctly. The paper presents the classifier ensemble for classifying binary, non-stationary and imbalanced data streams where the Hellinger Distance is used to prune the ensemble. The paper includes an experimental evaluation of the method based on the conducted experiments. The first one checks the impact of the base classifier type on the quality of the classification. In the second experiment, the Hellinger Distance Weighted Ensemble ( HDWE ) method is compared to selected state-of-the-art methods using a statistical test with two base classifiers. The method was profoundly tested based on many imbalanced data streams and obtained results proved the HDWE method's usefulness.

Many different decision problems require taking a compromise between the various goals we want to achieve into account. A specific group of features often decides the state of a given object. An example of such a task is the feature selection that allows increasing the decision’s quality while minimizing the cost of features or the total budget. The work’s main purpose is to compare feature selection methods such as the classical approach, the one-objective optimization, and the multi-objective optimization. The article proposes a feature selection algorithm using the Genetic Algorithm with various criteria, i.e., the cost and accuracy. In this way, the optimal Pareto points for the nonlinear problem of multi-criteria optimization were obtained. These points constitute a compromise between two conflicting objectives. By carrying out various experiments on various base classifiers, it has been shown that the proposed approach can be used in the task of optimizing difficult data.

The bionanohybrids are the junctions of at least two objects of different origin: abiotic and biotic. The abiotic part is a nanoparticle (often a fluorescent quantum dot), the biotical one may be a protein (especially fluorescent one or redox-active one), nucleic acid, carbohydrate as well as a simple organic molecule. When such a junction undergoes illumination, the energy transfer between the partners is possible. The nanoparticles, depending on their characteristics, may be donors, acceptors or mediators of the energy transfer. In most cases, the mechanism of the transfer is the Förster resonance energy transfer (FRET) or the electron transfer (ET). Here, we reviewed the newest achievements in the field with special attention paid to those bionanohybrids which allow FRET or ET. Such nanohybrids are important not only for exploration of the mechanism of the partner interaction but mainly for working out nanobiodevices for biosensing and nanotools for modern therapies.

Direct interaction of ferredoxin:NADP+ oxidoreductase (FNR) with thylakoid membranes was postulated as a part of the cyclic electron flow mechanism. In vitro binding of FNR to digalactosyldiacylglycerol and monogalactosyldiacylglycerol membranes was also shown. In this paper we deal with the latter interaction in more detail describing the effect for two FNR forms of Synechocystis PCC 6803. The so-called short FNR (sFNR) is homologous to FNR from higher plant chloroplasts. The long FNR (lFNR) form contains an additional domain, responsible for the interaction with phycobilisomes. We compare the binding of both sFNR and lFNR forms to native and non-native lipids. We also include factors which could modulate this process: pH change, temperature change, presence of ferredoxin, NADP+ and NADPH and heavy metals. For the lFNR, we also include phycobilisomes as a modulating factor. The membrane binding is generally faster at lower pH. The sFNR was binding faster than lFNR. Ferredoxin isoforms with higher midpoint potential, as well as NADPH and NADP+, weakened the binding. Charged lipids and high phosphate promoted the binding. Heavy metal ions decreased the rate of membrane binding only when FNR was preincubated with them before injection beneath the monolayer. FNR binding was limited to surface lipid groups and did not influence hydrophobic chain packing. Taken together, FNR interaction with lipids appears to be non-specific, with an electrostatic component. This suggests that the direct FNR interaction with lipids is most likely not a factor in directing electron transfer, but should be taken into account during in vitro studies.

Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) is one of the best studied enzymes. It is crucial for photosynthesis, and thus for all of biosphere’s productivity. There are four isoforms of this enzyme, differing by amino acid sequence composition and quaternary structure. However, there is still a group of organisms, dinoflagellates, single-cell eukaryotes, that are confirmed to possess Rubisco, but no successful purification of the enzyme of such origin, and hence a generation of a crystal structure was reported to date. Here, we are using in silico tools to generate the possible structure of Rubisco from a dinoflagellate representative, Symbiodinium sp. We selected two templates: Rubisco from Rhodospirillum rubrum and Rhodopseudomonas palustris. Both enzymes are the so-called form II Rubiscos, but the first is exclusively a homodimer, while the second one forms homo-hexamers. Obtained models show no differences in amino acids crucial for Rubisco activity. The variation was found at two closely located inserts in the C-terminal domain, of which one extends a helix and the other forms a loop. These inserts most probably do not play a direct role in the enzyme’s activity, but may be responsible for interaction with an unknown protein partner, possibly a regulator or a chaperone. Analysis of the possible oligomerization interface indicated that Symbiodinium sp. Rubisco most likely forms a trimer of homodimers, not just a homodimer. This hypothesis was empowered by calculation of binding energies. Additionally, we found that the protein of study is significantly richer in cysteine residues, which may be the cause for its activity loss shortly after cell lysis. Furthermore, we evaluated the influence of the loop insert, identified exclusively in the Symbiodinium sp. protein, on the functionality of the recombinantly expressed R. rubrum Rubisco. All these findings shed new light onto dinoflagellate Rubisco and may help in future obtainment of a native, active enzyme.

Here, we explore a mechanism of quantum dots related photoreduction of two redox-active proteins, cytochrome c and ferredoxin, by detailed analysis of fluorescence decay and reconstruction of time-resolved emission spectra (TRES). We used two types of cadmium telluride quantum dots, with diameters of 2.6 nm and 3.9 nm and maximum emissions at 550 nm and at 650 nm, respectively, which are known to be able to reduce proteins with different efficiencies. First, we observed that for a pure quantum dots solution, the fluorescence decay can be well fitted by three components. The average fluorescence lifetimes, as well as separate time constants, depend on the nanocrystal diameter. In the presence of proteins, fluorescence decay is faster and cytochrome c has a greater impact than ferredoxin. The TRES experiment showed that a fraction of the medium τ decay component is dominant in a pure quantum dot solution, with the maximum corresponding to the steady-state spectrum. The addition of ferredoxin does not change this pattern, while the presence of cytochrome c strongly promotes the shortest τ. Additionally, potassium iodide titration experiments were used to verify the origin of individual decay components. We propose that reduction occurs by electron transfer from both conductive band and surface trap states.

The GLABRA (GL1) gene, belonging to the transcription factor-encoding myb gene family, is responsible for trichome formation in Arabidopsis thaliana (L.) Heynh. The leaves and stems of glabra1 mutant plants are devoid of trichomes. Having an easily observable phenotype, the gl1 mutation was one of the first markers established for genetic mapping of Arabidopsis thaliana. Since then, the GL1 gene has been assigned roles in other processes, also related to leaf structure. In this study we present some previously undescribed effects of the gl1 mutation on dark-induced senescence. This process was induced by covering selected mature leaves of Columbia wild-type and gl1 Arabidopsis with black paper for 4 days, while the plants remained growing in a normal photoperiod. While no visible differences in the external symptoms of senescence could be observed in the darkened leaves, the expression of senescence-associated genes was significantly lower in gl1 plants as compared to the wild type. The darkening of leaves led to a decrease in photosynthetic activity and the expression of photosynthesis-associated genes, in comparison to the control leaves. This effect was much less pronounced in gl1 than in the wild type plants. Therefore, gl1 plants seem to be less susceptible to dark-induced aging, suggesting a possible role for the GL1 gene in controlling the onset and progress of senescence. This result is also of practical importance, since gl1 is the genetic background of many other mutants. It may therefore be advisable to revise some of the results obtained with such mutants in light of findings presented here.

In this study, we tested the possibility of creating complexes of two proteins by fusing them with heterodimerizing helices. We used the fluorescent proteins GFP and mCHERRY expressed with a His-tag as our model system. We added heterodimer-forming sequences at the C- or N- termini of the proteins, opposite to the His-tag position. Heterodimerization was tested for both helices at the C-terminus or at the N- terminus and C-terminus. We observed complex formation with a nanomolar dissociation constant in both cases that was higher by one order of magnitude than the Kds measured for helices alone. The binding of two C-terminal helices was accompanied by an increased enthalpy change. The binding between helices could be stabilized by introducing an additional turn of the helix with cysteine, which was capable of forming disulphide bridges. Covalently linked proteins were obtained using this strategy and observed using fluorescence cross-correlation spectroscopy. Finally, we demonstrated the formation of complexes of protein dimers and quantum dots.

Senescence is the final stage of plant development, affecting individual organs or the whole organism, and it can be induced by several environmental factors, including shading or darkness. Although inevitable, senescence is a complex and tightly regulated process, ensuring optimal remobilization of nutrients and cellular components from senescing organs. Photoreceptors such as phytochromes and cryptochromes are known to participate in the process of senescence, but the involvement of phototropins has not been studied to date. We investigated the role of these blue light photoreceptors in the senescence of individually darkened Arabidopsis thaliana leaves. We compared several physiological and molecular senescence markers in darkened leaves of wild-type plants and phototropin mutants (phot1, phot2, and phot1phot2). In general, all the symptoms of senescence (lower photochemical activity of photosystem II, photosynthetic pigment degradation, down-regulation of photosynthetic genes, and up-regulation of senescence-associated genes) were less pronounced in phot1phot2, as compared to the wild type, and some also in one of the single mutants, indicating delayed senescence. This points to different mechanisms of phototropin operation in the regulation of senescence-associated processes, either with both photoreceptors acting redundantly, or only one of them, phot1, playing a dominant role.

Colloidal CdTe quantum dots (QD) were able to reduce both heme and iron–sulfur cluster containing proteins. Reduction was depended on quantum dots size. CdTe nanocrystals emitting light with maximum at 550 nm (QD-550, r ∼ 1.5 nm) reduced both ferredoxin (Fd) and cytochrome c (Cyt c). The process was followed by UV/Vis absorption spectroscopy and steady state fluorescence spectroscopy. For CdTe emitting longer wavelength (QD-610 and QD-670) reduction of Fd was less efficient. QD emitting light with maximum at 750 nm (QD-750, r ∼ 3.3 nm) reduced only Cyt c. Reduction of proteins by QDs was photo-dependent and did not demand oxygen presence. As shown by gel filtration and fluorescence correlation spectroscopy, Fd formed complexes with QD while Cyt c did not bound steadily. The stable complex was not necessary for photoreduction, although might influence its kinetics. Time-resolved fluorescence studies showed than electron transfer rate depends on QD size and also is higher for QD–Cyt c when compared to QD–Fd. The mechanism of process was additionally explored by detailed analysis of QD–protein complex formation and by measurements of cyclic voltamperometry and zeta potential. These results open new possibilities in controlling of natural redox processes.

Direct interaction of ferredoxin:NADP

Oxide nanoparticles doped with rare-earth ions are tested for applications as nanomarkers in biology and medicine, e.g. for early cancer diagnostics. In this work we describe properties of such markers obtained by a microwave hydrothermal technology. Nanomarkers based on oxide materials (ZnO) are grown and tested for medical and biological applications.

In the present studies, we have found a fragment of amino acid sequence, called TFT motif, both in light-dependent protochlorophyllide oxidoreductase (LPOR) and in the L subunit of dark-operative (light-independent) protochlorophyllide oxidoreductases (DPOR). Amino acid residues of this motif shared similar physicochemical properties in both types of the enzymes. In the present paper, physicochemical properties of amino acid residues of this common motif, its spatial arrangement and a possible physiological role are being discussed. This is the first report when similarity between LPOR and DPOR, phylogenetically unrelated, but functionally redundant enzymes, is described.

Here, the transition of colloidal CdSe quantum dots (QDs) from hydrophobic to hydrophilic environments after coating the surface with long peptide chains of membrane scaffold proteins (MSP) is reported. The intermediate step included the solubilization of QDs with detergents, where n-octyl glucoside was the most promising ligand. Furthermore, size analysis by fluorescence correlation spectroscopy, gel filtration and atomic force microscopy suggested that the obtained QD-MSP conjugates were primarily discoidal and were likely formed from single QDs tightly encircled by helix belts. In addition, Fourier-transformed infrared spectroscopy analysis confirmed the preservation of the secondary structure of most proteins during conjugate formation, with no signs of denaturation. The obtained QD-MSP conjugates were optimal in terms of stability in water environments, suggesting that it is possible to obtain QDs with single peptide coats and providing the first guidelines for future research in this direction.

The paper explores the eld of de novo protein design, as a source of material for e ective hybrid nanostructures. Main design approaches, namely the intuitional and the computational strategy, are brie y overviewed. The achievements in the eld are illustrated with several examples, starting from historical heme binding maquettes to novel non-natural enzymes. Separate paragraph covers the problem of designing peptides, which may act as anchor between biological and non-biological parts of nanostructures. The advantages of de novo designed proteins and still existing problems of the eld are discussed.

Here, we compare two approaches of protein design. A computational approach was used in the design of the coiled-coil iron-sulfur protein, CCIS, as a four helix bundle binding an iron-sulfur cluster within its hydrophobic core. An empirical approach was used for designing the redox-chain maquette, RCM as a four-helix bundle assembling iron-sulfur clusters within loops and one heme in the middle of its hydrophobic core. We demonstrate that both ways of design yielded the desired proteins in terms of secondary structure and cofactors assembly. Both approaches, however, still have much to improve in predicting conformational changes in the presence of bound cofactors, controlling oligomerization tendency and stabilizing the bound iron-sulfur clusters in the reduced state. Lessons from both ways of design and future directions of development are discussed. This article is part of a Special Issue entitled: Photosynthesis Research for Sustainability: from Natural to Artificial.

Photosynthetic organisms utilize interacting pairs of chlorophylls and bacteriochlorophylls as excitation energy donors and acceptors in light harvesting complexes, as photosensitizers of charge separation in reaction centers, and maybe as photoprotective quenching centers that dissipate excess excitation energy under high light intensities. To better understand how the pigment's local environment and spatial organization within the protein tune its ground- and excited-state properties to perform different functions, we prepared and characterized the simplest possible system of interacting bacteriochlorophylls within a protein scaffold. Using HP7, a high-affinity heme-binding protein of the HP class of de novo designed four-helix bundles, we incorporated 13(2)-OH-zinc-bacteriochlorophyllide-a (ZnBChlide), a water-soluble bacteriochlorophyll derivative, into specific binding sites within the four-helix bundle protein core. We capitalized on the rich and informative optical spectrum of ZnBChlide to rigorously characterize its complexes with HP7 and two variants, in which a single heme-binding site is eliminated by replacing histidine residues at positions 7 or 42 by phenylalanine. Surprisingly, we found the ZnBChlide binding capacity of HP7 and its variants to be higher than for heme: up to three ZnBChlide pigments bind per HP7, or two per each single histidine variant. The formation of dimers within HP7 results in dramatic quenching of ZnBChlide fluorescence, reducing its quantum yield by about 80%, and the singlet excited-state lifetime by 2 orders of magnitudes compared to the monomer. Thus, HP7 and its variants are the first examples of a simple protein environment that can isolate a self-quenching pair of photosynthetic pigments in pure form. Unlike its complicated natural analogues, this system can be constructed from the ground up, starting with the simplest functional element, increasing the complexity as needed.

The ferredoxin:NADP+ oxidoreductase (FNR) catalyses the ferredoxin-dependent reduction of NADP+ to NADPH in linear photosynthetic electron transport. The enzyme also transfers electrons from reduced ferredoxin (Fd) or NADPH to the cytochrome b(6)f complex in cyclic electron transport. In vitro, the enzyme catalyses the NADPH-dependent reduction of various substrates, including ferredoxin, the analogue of its redox centre - ferricyanide, and the analogue of quinones, which is dibromothymoquinone. This paper presents results on the cadmium-induced inhibition of FNR. The K(i) value calculated for research condition was 1.72 mM. FNR molecule can bind a large number of cadmium ions, as shown by the application of cadmium-selective electrode, but just one ion remains bound after dialysis. The effect of cadmium binding is significant disturbance in the electron transfer process from flavin adenine dinucleotide (FAD) to dibromothymoqinone, but less interference with the reduction of ferricyanide. However, it caused a strong inhibition of Fd reduction, indicating that Cd-induced changes in the FNR structure disrupt Fd binding. Additionally, the protonation of the thiol groups is shown to be of great importance in the inhibition process. A mechanism for cadmium-caused inhibition is proposed and discussed with respect to the in vitro and in vivo situation.

Comparative study was performed to assess the content and proportions of photosynthetic pigments and the violaxanthin cycle (VXC) activity in winter-green and summer-green leaves of bugleweed (Ajuga reptans L.) plants grown in shaded (photosynthetically active radiation, PAR 150 μmol/(m2 s)) and sunny (PAR 1200 μmol/(m2 s)) habitats in the Botanic Garden of Jagiellonian University (Krakow, Poland). In overwintered and newly formed leaves of shade plants, the content of green and yellow pigments was two times higher than in leaves of sun plants. The shade plants were distinguished by accumulation of β-carotene, while lutein was predominant in leaves of sun plants. Under the action of strong light (2000 μmol/(m2s)), the level of violaxanthin deepoxidation in winter-green leaves of shade and sun plants increased five- to sixfold, whereas it changed insignificantly in summer-green leaves of shade plants. It is concluded that, in a shadetolerant species A. reptans, the photosynthetic apparatus of winter-green leaves in sun and shade plants and of summer-green leaves in sun plants is protected against excess insolation by high activity of VXC. The carotenoids of summer-green leaves in shade plants are supposed to function mainly as light-harvesting pigments.

Ferredoxin:NADP(+) oxidoreductase is an enzyme associated with the stromal side of the thylakoid membrane in the chloroplast. It is involved in photosynthetic linear electron transport to produce NADPH and is supposed to play a role in cyclic electron transfer, generating a transmembrane pH gradient allowing ATP production, if photosystem II is non-functional or no NADP(+) is available for reduction. Different FNR isoforms have been described in non-photosynthetic tissues, where the enzyme catalyses the NADPH-dependent reduction of ferredoxin (Fd), necessary for some biosynthetic pathways. Here, we report the isolation and purification of two FNR isoproteins from wheat leaves, called FNR-A and FNR-B. These forms of the enzyme were identified as products of two different genes, as confirmed by mass spectrometry. The molecular masses of FNR-A and FNR-B were 34.3 kDa and 35.5 kDa, respectively. The isoelectric point of both FNR-A and FNR-B was about 5, but FNR-B appeared more acidic (of about 0.2 pH unit) than FNR-A. Both isoenzymes were able to catalyse a NADPH-dependent reduction of dibromothymoquinone and the mixture of isoforms catalysed reduction of cytochrome c in the presence of Fd. For the first time, the pH- and ionic strength dependent oligomerization of FNRs is observed. No other protein was necessary for complex formation. The putative role of the two FNR isoforms in photosynthesis is discussed based on current knowledge of electron transport in chloroplasts.