Isothermal titration calorimetry of membrane protein interactions: FNR and the cytochrome b₆f complex

Abstract

Ferredoxin-NADP⁺ reductase (FNR) was previously inferred to bind to the cytochrome b₆f complex in the electron transport chain of oxygenic photosynthesis. In the present study, this inference has been examined through analysis of the thermodynamics of the interaction between FNR and the b₆f complex. Isothermal titration calorimetry (ITC) was used to characterize the physical interaction of FNR with b₆f complex derived from two plant sources (Spinacia oleracea and Zea maize). ITC did not detect a significant interaction of FNR with the b₆f complex in detergent solution nor with the complex reconstituted in liposomes. A previous inference of a small amplitude but defined FNR-b₆f interaction is explained by FNR interaction with micelles of the undecyl β-D maltoside (UDM) detergent micelles used to purify b₆f. Circular dichroism, employed to analyze the effect of detergent on the FNR structure, did not reveal significant changes in secondary or tertiary structures of FNR domains in the presence of UDM detergent. However, thermodynamic analysis implied a significant decrease in an interaction between the N-terminal FAD-binding and C-terminal NADP⁺-binding domains of FNR caused by detergent. The enthalpy, ΔH^o, and the entropy, ΔS^o, associated with FNR unfolding decreased four-fold in the presence of 1 mM UDM at pH 6.5. In addition to the conclusion regarding the absence of a binding interaction of significant amplitude between FNR and the b₆f complex, these studies provide a precedent for consideration of significant background protein-detergent interactions in ITC analyses involving integral membrane proteins.

Significance

This study concerns structure properties and membrane-protein interactions of the cytochrome b₆f complex for which a 2.5 Å crystal structure has been obtained. Regarding electron transfer between FNR and b₆f complex, a question is whether FNR, an electron donor to the cytochrome b₆f complex, interacts with the complex to an extent that both proteins should be considered part of a “super-complex”. This inference had been based on a tendency for co-purification and electron exchange. Here, isothermal titration calorimetry was used to examine the existence of a significant interaction between FNR and the b₆f complex. Such an interaction was not found. Co-purification of FNR with the b₆f complex is associated with a significant affinity of FNR for undecyl β-D maltoside detergent.

Introduction

The cytochrome b₆f complex (Fig. 1 A), for which a structure has been obtained at 2.5-Å resolution by x-ray diffraction analysis (1), has a central role in the function and regulation of electron transport and energy transduction in oxygenic photosynthesis (2). The b₆f complex can accept electrons from the peripheral electron transfer protein, FNR, ferredoxin-NADP⁺ reductase (3), which also functions as an electron acceptor of ferredoxin in the pathway of NADP⁺ reduction. This FNR-b₆f connection can participate in a “cyclic” electron transport pathway (4,5), providing an additional source of membrane energization for ATP synthesis.

Figure 1.

Crystal structures of cytochrome b₆f complex (A) and FNR (B). (A) Ribbon diagram of dimeric cytochrome b₆f complex previously determined at 2.5-Å resolution (PDB: 4OGQ). Each monomer contains the following subunits: cytochrome b dimer (yellow) binding two transmembrane hemes; subunit IV (light blue), cytochrome f (pale green), the iron-sulfur protein (brown) and, on the periphery of each monomer, four hydrophobic single transmembrane helix polypeptides (Pet G,K,L,M) colored in gray, cyan, magenta and orange, respectively. Hemes and one chlorophyll a per monomer are shown as red and green sticks, and iron-sulfur clusters as brown and yellow spheres. (B) Ribbon diagram of FNR (PDB: 1FND; 1.7-Å resolution). The N-terminal domain containing the FAD group is dark yellow, and the C-terminal domain light blue. FAD (orange) is located in the interface between N- and C-domains. Trp296 and Trp309 (green), whose interaction generates a split-CD signal, are in the hydrophobic cavity. Phospho-AMP (blue) represents the position of the NADP⁺-binding site in the C-domain.

The crystal structure of FNR (Fig. 1 B) has been determined to a resolution of 1.7 Å (6) and in complex with ferredoxin at 2.6-Å resolution (Fig. S1; (7)). The ability of FNR to bind to the thylakoid membrane has been documented (8, 9, 10, 11). Analysis of the FNR crystal structure suggested its buried hydrophobic cavity as a possible membrane attachment site (6), whereas the recruiting of FNR to thylakoids involves binding to the polyproline type II helix found in the Tic62 and Trol proteins (12,13). The presence of FNR at significant, but substoichiometric levels, in purified highly active plant (spinach) cytochrome b₆f complex suggested the possibility that FNR is a structural component of the b₆f complex (14). This inference had also been made previously on the basis of co-purification (15,16).

To examine the interaction between the b₆f complex and FNR more quantitatively, the present study considers the thermodynamics of their interaction to determine whether FNR interacts with the b₆f complex as part of a defined integral co-complex or, like plastocyanin, functions in electron exchange but not as an integral component of the b₆f complex (17). In the present study, isothermal titration calorimetry (ITC) was used to determine the thermodynamic parameters of FNR interaction with b₆f complex derived from two plant sources (spinach, Zea maize). ITC measurements conducted in the presence and absence of detergent, including b₆f complex reconstituted in liposomes, did not detect a significant affinity between FNR and b₆f complex. The measured parameters documenting low affinity were attributed to an interaction of FNR with detergent, or with the liposome membrane surface. A preliminary summary of ITC analysis of the interaction between FNR and the b₆f complex has been presented (18) in which the thermodynamic information derived from the ITC analysis had not yet been determined, and which is now considered obsolete.

Materials and Methods

Purification of dimeric cytochrome b₆f complex from spinach and maize

Dimeric cytochrome b₆f complex was isolated from spinach leaves and purified according to a protocol described previously (19). After extraction of the b₆f complex from sodium bromide-treated thylakoids with a combination of two detergents, 30 mM octyl glucoside (OG) and 0.1% sodium cholate, at a chlorophyll concentration of 2 mg/mL, insoluble material was removed by centrifugation at 250,000 x g for 60 min. Ammonium sulfate was added to 37.5% (w/v) of saturation. After removal of precipitate, the supernatant was loaded on a propyl-agarose column preequilibrated with 20 mM N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES) (pH 7.6), 50 mM sodium chloride, 2 mM ethylenediamine tetraacetic acid, 2 mM benzamidine, 2 mM ε-amino-caproic acid, ammonium sulfate (37.5% of saturation), and 1 mM undecyl β-D maltoside (UDM) detergent. Extensive washing of the column with this solution removes green contaminants and exchanges the detergent from OG to UDM. The cytochrome b₆f complex is eluted with the same buffer also containing ammonium sulfate at 20% saturation. Collected fractions were concentrated and desalted using size-exclusion chromatography (SEC) on a Superdex 200 column (GE Life Science). This step also removes “green contaminants,” which are eluted before the dimeric b₆f complex, monomeric b₆f, and FNR when it was present. Dimeric and monomeric b₆f, and FNR eluted at 10.9, 12.3, and 15.2 mL, respectively (Superdex 200, 10 x 300 column; Fig. S2). Further details of the purification are described in Baniulis et al. (19). The concentration of b₆f complex was determined using a molar extinction coefficient of 25 mM⁻¹cm⁻¹ at 554 nm applied to the redox difference spectrum of cytochrome f (20). Electron transfer activity of the cytochrome b₆f complex was tested using the reduction of plastocyanin by decyl-plastoquinol (19) dependent on the cytochrome b₆f complex. Purified dimeric b₆f complex, which did not contain FNR, was used in the ITC studies.

Reconstitution of the b₆f complex in liposomes

The cytochrome complex was reconstituted in liposomes consisting of the phospholipids DOPC/POPE (1:1 mol/mol). Lipid aliquots dissolved in chloroform were mixed and placed in vacuum overnight to remove solvent. Liposomes were mixed with the cytochrome complex in the presence of 2% OG at a ratio of 2600 mol/mol, phospholipid to cytochrome dimer. After a 20-min incubation, the detergent was removed by dialysis. The dimer concentration in proteoliposomes was determined from the concentration of cytochrome f measured by the amplitude at 554 nm in redox difference (sodium ascorbate minus ferricyanide) spectra.

Purification of recombinant spinach FNR

Recombinant spinach FNR with an N-terminal His-6 tag and TEV protease recognition site was expressed in E. coli BL21(DE3) cells using the vector p28a(+). After lysozyme treatment, cells were broken by sonication. The supernatant after ultracentrifugation was applied to an Ni-NTA column. Protein eluted from the column with 0.5 M imidazole was dialyzed and digested with TEV protease to remove the His-tag. SEC on a Superdex 200 column was used for final purification. This variant of FNR carries the mutation V269F, which significantly increases the thermal stability of FNR, which was essential for crystallization (6). The FNR concentration containing the flavin adenine dinucleotide (FAD) prosthetic group was determined spectrophotometrically using a molar extinction coefficient of 10.7 mM⁻¹ cm⁻¹ at 457 nm for FAD (7).

Isothermal titration calorimetry

For detection of the interaction of FNR with the cytochrome b₆f complex, a Low Volume Nano ITC Calorimeter (TA Instruments, Germany) containing a total cell volume of 0.3 mL (active volume, 178 μL) was used. The burette was equipped with a 50-μL syringe and syringe needle that also worked as a stirring paddle. Titrations were carried out in the temperature range 5–25°C, with a stirring rate of 300 rpm. 20 injections of 2.45 μL of ligand were performed with a 300-s interval between injections. For ITC measurements, buffers with a small and moderate enthalpy of protonation (ΔH^P, kJ/mol) were used: sodium acetate (0.49), sodium phosphate (5.2); morpholino-ethanesulfonic acid (MES) (15.5), HEPES (21.1) (21,22). To obtain valid thermodynamic parameters, both components, protein and ligand, were dissolved in the same buffer solution (21).

Far and near-UV circular dichroism spectral analysis

Circular dichroism (CD) spectra in the far-UV range (185–260 nm) were measured in quartz cuvettes (Hellma) with optical path lengths of 0.02, 0.1, and 0.2 mm using a “Chirascan” spectropolarimeter (Applied Photophysics, Leatherhead, U.K.) equipped with a Peltier thermal control unit. The far-UV CD spectral amplitudes expressed in units of mean molar residue ellipticity were analyzed using the DichroWeb package of Selcon3, Contin, and CDSSrt programs (23).

Near-UV CD spectra (250–320 nm; step, 0.1 nm) were measured in a quartz cuvette with a 1-cm optical pathlength. To reduce spectral noise, 15 spectra were averaged for each determination.

Thermodynamic analysis of the stability of FNR domains using thermal “melting” of FNR measured by the far-UV CD signal at 222 nm

CD analysis of the thermal unfolding of the FNR domains was performed at 222 nm in a stirred quartz cuvette with an optical pathlength of 1 cm and a heating rate of 0.5°C/min in the temperature range 10–90°C, with a 0.4°C step. Simultaneous measurement of the CD signal and absorbance allows control of the reversibility of the protein unfolding, which is important for thermodynamic analysis. The buffer solution contained 10 mM sodium acetate, MES, or HEPES titrated to the necessary pH with 50 mM sodium phosphate and 50 mM sodium chloride.

Enthalpy (ΔH^o) and temperature midpoint (T_m) parameters of FNR unfolding were derived from the slope of the thermal melting curves (folded-to-unfolded transition) as described (24,25). Considering the temperature-dependent equilibrium FNR(folded) ↔ FNR(unfolded) the folded fraction, f, is determined experimentally through the amplitude of the CD signal at 222 nm:

$$f\equiv \frac{\left[F\right]}{\left[U\right]+\left[F\right]}=\frac{\theta \left(T\right)-{\theta }_U}{{\theta }_F-{\theta }_U}$$ (Equation 1)

where [F] and [U] are concentrations of folded and unfolded FNR, respectively, Θ(T) is the measured ellipticity at any temperature T, Θ_F, is the ellipticity of fully folded protein, and Θ _U, the ellipticity of the unfolded protein. To fit the changes of CD at 222 nm as a function, f, of temperature, T^oK, the following expression for f, derived from the van 't Hoff equation, was used:

$$f=\frac{1}{1+\exp \left[-\frac{\Delta H^0}{R}\left(\frac{1}{T_m}-\frac{1}{T}\right)\right]}$$ (Equation 2)

where T_m is the midpoint temperature at which the protein is 50% unfolded, ΔH⁰ is the enthalpy of the transition U→F, from the unfolded to the folded states, and R is the gas constant. Combining Equations (1) and (2) directly relates the measured CD signal, θ(T), to the parameters T_m and ΔH⁰:

$$\theta \left(T\right)=\frac{{\theta }_F-{\theta }_U}{1+\exp \left[-\frac{\Delta H^0}{R}\left(\frac{1}{T_m}-\frac{1}{T}\right)\right]}+{\theta }_U$$ (Equation 3)

Parameters T_m and ΔH⁰ for the U→F transition were obtained by fitting the experimentally measured CD signals, θ (T), with a function for the model (3). As the Θ_F and Θ_U are generally unknown, these parameters were also adjusted by the fitting program to obtain the best match between the experimental data and the model. See Supporting Material for details of the calculation.

The measured “melting curves” exhibit a background with a relatively small slope. It is proposed to originate from noncooperative melting of weakly bound segments of FNR such as the helix and strand termini, which have an incomplete hydrogen bond structure (23). It was modeled as an additional transition component with a relatively small ΔH⁰. The inclusion of this additional broad transition to account for the background slope in fitting of the data was critical for proper determination of ΔH⁰ of the sharp highly cooperative transitions associated with two FNR domains (Fig. 7). See Supporting Material (Figs. S5–S7, Equations S1–S6) for more detail.

Figure 7.

Effect of UDM detergent on thermal “melting” of FNR domains. The temperature dependence of the CD signal at 222 nm was used to measure unfolding of FNR domains in the absence and presence of UDM in the temperature range 10–90°C. The measured melting curves in the absence of UDM at pH 5.0 (triangles) and pH 6.5 (circles), as well as in the presence of 1 mM UDM at pH 6.5 (squares), were fit with biphasic (pH 5.0) and monophasic (pH 6.5) sigmoidal functions to apply thermodynamic analysis of the stability of FNR domains as described in Materials and Methods.

The entropy change of the U → F transition was calculated by considering the thermodynamic relationship at equilibrium (Gibbs energy ΔG = 0 at T_m):

$$\Delta H=T_m\Delta S.$$ (Equation 4)

Results

ITC of the interaction of FNR with dimeric cytochrome b₆f complex in detergent solution

ITC is a sensitive method for studies of protein-ligand interactions. The characteristic thermodynamic parameters of an interaction, affinity (Kd), standard enthalpy (ΔH^o), entropy (ΔS^o), and stoichiometry, can be determined in a single experiment (26,27).

Purified active cytochrome b₆f complex is dimeric (1,28). ITC measurements were carried out in the same buffer solutions containing 50 mM sodium chloride and 50 mM sodium phosphate in the ITC cell and syringe, respectively, which contained the b₆f complex and FNR, at pH 6.5 or 7.5 with 1 mM UDM detergent. Titrations of spinach FNR with the cytochrome b₆f complex purified from spinach were conducted in a temperature range from 5 to 25°C (Fig. 2 A and B). These measurements showed insignificant heat production, comparable to that measured in a control titration of water to water (not shown). In addition, the titration did not show the sigmoidal shape of a titration curve that is associated with saturation of binding sites (26). This implies an insignificant contribution of an enthalpic interaction of FNR with the b₆f complex. However, the data shown do not rule out the possibility of purely entropic binding (ΔH⁰ = 0) of FNR with the b₆f complex. Similar results showing the absence of an enthalpic interaction of FNR with the b₆f complex were obtained at pH 5.5 and 7.5 (data not shown).

Figure 2.

ITC measurement of FNR binding to the b₆f complex; temperature dependence. (A) Titration at pH 6.5 at 10°C (triangles) or at 20°C (circles) in buffer containing 10 mM MES, 50 mM NaCl, 1 mM UDM. The cell contains 16 μM b₆f dimer and the syringe, 160 μM FNR. (B) Titration at pH 7.5 at 5°C (triangles), 15°C (squares), or 25°C (circles) in buffer containing 5 mM HEPES, 50 mM sodium phosphate, 1 mM UDM. The cell contains 9.6 μM b₆f dimer; the syringe contains 107 μM FNR.

Higher plant FNR has several isoforms including a thylakoid membrane-binding isoform (11). ITC titration of the Z. maize membrane-binding form of FNR (FNR2 (29)) with dimeric b₆f complex purified from maize also did not reveal an interaction of FNR2 with the b₆f complex (Fig. S3 A and B).

Water-soluble FNR carries several hydrophobic sites [8,9] that are involved in FNR binding to thylakoid membranes (12,28). The lack of interaction of FNR with the b₆f complex in UDM buffer solution could be explained by masking of the hydrophobic sites on FNR by the detergent. Therefore, titrations were performed using FNR in buffer without UDM, whereas b₆f complex was studied in buffer containing UDM. This titration resulted in the moderate heat release with indication of saturation of binding sites (Fig. S4 A). However, the same titration profile and heat release were obtained for FNR titration into UDM buffer without b₆f (Fig. S4 B). These results imply that most or all heat production in these titrations originates from the interaction of FNR with UDM detergent.

ITC measurement of FNR interaction with UDM at sub-critical micelle concentration (0.2 mM) did not detect any heat release or uptake (not shown). Therefore, thermodynamic parameters of the interaction of spinach FNR with UDM were determined with the assumption that FNR interacts with UDM micelles (Fig. 3). The concentration of the UDM micelles, 14 μM, was calculated using a detergent aggregation number of 71 molecules per micelle (Anatrace). The thermodynamic parameters for the interaction of spinach FNR with UDM micelles in the absence of the b₆f complex are as follows: K_d, 1.3 ± 0.4 x 10⁻⁵ M⁻¹; n, 2 ± 1; ΔH^o, −25 ± 2 kcal/mol; ΔS^o, −62 ± 2 cal/mol^xK. In the presence of the complex, they were as follows: K_d, 2.7 ± 0.5 x 10⁻⁵ M⁻¹; n, 4 ± 2; ΔH^o, −28 ± 3 kcal/mol; ΔS^o, −22 ± 4 cal/mol^xoK (Table 1), where n is the number of binding sites.

Figure 3.

ITC analysis of FNR interaction with UDM micelles. Cell contains 1 mM UDM (approximately 14 μM UDM micelles), 10 mM MES (pH 6.5), 50 mM sodium chloride. Syringe contains 160 μM FNR in the same buffer without UDM. The thermodynamic parameters for the interaction of spinach FNR with UDM micelles are as follows: K_d = 1.4 x 10⁻⁵ M⁻¹; n = 3; ΔH^o = −25 kcal/mol; ΔS^o = −62 cal/mol ^oK. First data point with diluted reagent was excluded from analysis.

Table 1.

Interaction of FNR with UDM micelles; thermodynamic parameters

Parameters	b6f absent	With b6f
Kd, M−1	(1.3 ± 0.4) x 10−5	(2.7 ± 0.5) x 10−5
n, binding sites	2 ± 1	4 ± 2
ΔH0, kcal/mol	−25 ± 2	−28 ± 3
ΔS0, cal/mol K	−62 ± 2	−22 ± 4

Syringe: FITC is a sensitive NR in buffer solution without UDM detergent. Cell: buffer solution containing 1 mM UDM in the presence and absence of the b₆f complex.

Interaction of FNR with the cytochrome b₆f complex reconstituted into liposomes

To avoid possible detergent interference with the FNR-b₆f interaction, the cytochrome b₆f complex was also reconstituted into liposomes consisting of the phospholipids DOPC/POPE (1:1 mol/mol), 2600 mol phospholipid per cytochrome dimer. The ITC titration of FNR to proteoliposomes with 5 μM b6f dimer resulted in a significant heat release for each of the 20 injections of FNR (Fig. 4, circles). However, despite an FNR concentration 24 times that of the cytochrome complex, the isotherm did not show a sigmoidal decline associated with saturation of binding sites by FNR.

Figure 4.

Significant heat release upon titration of proteoliposomes with FNR is caused by FNR interaction with the liposome surface. Syringe: 0.12 mM FNR; cell: proteoliposomes containing 5 μM dimeric cytochrome complex and 20 mM lipid (circles) or liposomes (triangles) with 20 mM lipid, DOPC/DOPE, 1:1 mol/mol. Buffer: 10 mM MES, 0.1 M NaPi (pH 6.5).

A similar heat release and an absence of saturation of binding sites were seen in an FNR titration with liposomes that did not contain the b₆f complex (Fig. 4, triangles), implying that the FNR interaction with b₆f proteoliposomes is a consequence of FNR binding directly to the lipid surface of the liposomes. The absence of saturation of binding sites on the lipid surface of liposomes (total lipid content of liposomes, 20 mM) is attributed to insufficient amount of the titrant (0.12 mM FNR).

Can the V269F mutation in FNR prevent interaction with the b₆f complex?

It has been suggested (6) that the hydrophobic cavity in the C-domain is the site of FNR binding with thylakoid membranes, and that the V269F mutation in this cavity, which was introduced to increase thermal stability of the C-domain, might prevent this binding because the bulky hydrophobic side chain of Phe decreases the accessible hydrophobic area of the cavity. 1 mM UDM was not sufficient to saturate all UDM-binding sites of FNR with Val269. Assuming that the hydrophobic cavity might also be the site of FNR interaction with the b₆f complex, ITC measurement of the interaction of FNR with the b₆f complex was done in buffer containing 2 mM UDM. The titration of wild-type FNR containing the b₆f complex, both in 2 mM UDM, did not show an interaction between the b₆f complex and wild-type FNR (Fig. 5).

Figure 5.

ITC measurement of interaction of wild-type FNR (Val269) with the b₆f dimer. The syringe contained 0.15 mM FNR, and the cell 5 μM b₆f dimer. Buffer: 10 mM MES, 50 mM NaPi (pH 6.5); 2 mM UDM.

Does UDM detergent alter the secondary and tertiary structure of FNR and thereby prevent binding to the b₆f complex?

The absence of a significant binding interaction of FNR with the cytochrome complex could be explained by changes in the conformation of native FNR caused by interaction with the detergent. Therefore, circular dichroism analysis was used to compare secondary and tertiary structures of FNR in the absence and presence of the UDM detergent.

Far-UV CD spectra of spinach FNR were similar in the absence and presence of 1 mM UDM (Fig. 6 A). The content of ordered secondary structure was estimated using the DichroWeb program for CD spectra analysis (23). Results of the analysis of α-helical and β-sheet structure content show that UDM does not alter the secondary structure of FNR (Table 2). The α-helix content determined by the far-UV CD analysis is in agreement with the fraction of amino acid residues in helices, which is found in the crystal structure of FNR (PDB: 1FND (6)).

Figure 6.

Far- and near-UV CD analysis of the effect of UDM on the secondary and tertiary structure of FNR. (A) CD spectra measured in the absence (circles) and presence of UDM (1 mM, triangles) overlap showing no dependence on UDM. (B) Near-UV CD spectra from which tertiary structure changes are derived. CD spectra were measured in a quartz cuvette with 1-cm optical path in the absence (circles) and presence of 1 mM UDM (triangles), or after FNR treatment with 1% SDS (squares). Buffer: 10 mM MES, 0.1 M sodium phosphate (pH 6.5).

Table 2.

Ordered secondary structure content of FNR

Structure, %	Helix	Sheet	Remainder
No UDM	28 ± 3	21 ± 2	51 ± 5
+1 mM UDM	29 ± 4	20 ± 2	51 ± 4

α-Helix and β-sheet content were obtained from the far-UV CD spectra, using the DichroWeb site (23). Results of analysis from the Selcon3, Contin, CDStr, and K2D programs were averaged.

The ordered secondary structure content of folded proteins is a relatively conservative parameter for detection of small changes in protein conformation. Another approach for detection of changes in protein folding is the CD signal in the near-UV region (Fig. 6 B) arising from Trp and Tyr groups, which is induced by a constrained environment of the FNR hydrophobic core (30). Spinach FNR contains 6 Trp and 12 Tyr, which are mostly (5 Trp and 8 Tyr) located in the C-terminal domain. The near-UV CD spectrum of FNR is dominated by a prominent bisignate CD signal (“split-CD” signal) with an amplitude crossover point at approximately 280 nm, and with positive and negative peaks at 272 and 287 nm, respectively (Fig. 6 B, circles). The search for Trp residues in the crystal structure of FNR (6) suggests that the split-CD signal could originate from excitonic dipole-dipole coupling (31) between the neighboring Trp296 and Trp309 located in the hydrophobic cavity of the C-domain. The distance between the respective CE2 atoms is 5.4 Å (Fig. 1 B). This cavity is distant from the interdomain interface, the site of electron transfer (6). The split-CD signal was eliminated by addition of 1% sodium-dodecyl-sulfate (SDS), which causes FNR unfolding (Fig. 6 B, squares), implying that this signal is induced by the tertiary structure folding of FNR. UDM detergent did not affect this CD signal (Fig. 6 B, triangles), implying that UDM does not change the tertiary structure folding of FNR, at least not of its C-terminal domain.

Effect of detergent (UDM) on the interdomain interaction of FNR

The question of whether UDM can impair interdomain interaction in FNR, which would affect an interaction with the b6f complex, was considered.

Spinach FNR has two distinct structural domains of approximately the same size, approximately 150 amino acids (32), as shown in Fig. 1 B. The N-terminal domain (blue) contains the FAD redox group (orange) and several long loops. An ordered secondary structure is represented mostly by an antiparallel β-sheet and a relatively small content of α-helical structure (6). The C-terminal domain (dark yellow) contains the NADP⁺ binding site (blue). Six α-helices with five parallel β-strands form a central β-sheet core (Fig. 1 B (6,29)).

To investigate the effect of UDM on interdomain interaction in FNR, thermal melting functions were measured and modeled using van 't Hoff equation with one or two sharp transition components depending on pH (Fig. 7), as described in Materials and Methods ((Equation 1), (Equation 2), (Equation 3), (Equation 4)) and in Supporting Material (Figs. S5–S7, Equations S1–S6). A slightly sloped background spanning the temperature range of measurements, ascribed to noncooperative melting of the FNR secondary structures, which are not involved in the tertiary structure interactions, was modeled as an additional transition. Although ΔH⁰ and T_m parameters of that broad underlying feature could not be reliably determined, its overall shape was reproduced well (Fig. S7). The inclusion of that underlying component into fitting models to account for sloped background was found to have a major effect on determination of ΔH⁰ values of the sharp transition(s) of interest (Figs. S5–S7).

The thermal melting function of FNR at pH 5.0 can be fit best by two distinct transitions, at 53.2 ± 1.6°C and 71.5 ± 3.4°C (Table 3), derived from the thermodynamic analysis of the melting curves (Fig. 7, inverted triangles), which are associated with unfolding of the N- and C-terminal domains, respectively, inferred from the relative domain complexity. At pH 6.5, both domains melt synchronously in a relatively narrow temperature range, Tm = 64.3 ± 0.9°C (Fig. 7, circles). This suggests a significant increase in thermal stability of the N-terminal and destabilization of the C-terminal domains. Such changes in the stability of FNR domains are inferred to be caused by a strong interaction between N- and C-terminal domains involving hydrophobic sites and an electrostatic interaction at pH 6.5. The recombinant FNR used in this study has a pI of 6.4, and the pI of the N- and C-terminal domains is 7.8 and 5.6, respectively.

Table 3.

Thermodynamic parameters of FNR unfolding with and in the presence and absence of 1 mM UDM at pH 6.5

Parameter	UDM absent	+1 mM UDM
pH 6.5
ΔHo, kcal/mol	−173 ± 16	−41 ± 10
ΔSo, cal/mol∗K	−514 ± 47	−120 ± 30
Tm, oC	64.3 ± 0.9	67.2 ± 3.2
pH 5.0
ΔHo1, kcal/mol	−94 ± 8	−190 ± 6
ΔHo2	−80 ± 30	−120 ± 2
ΔSo1, cal/mol∗K	−273 ± 25	−560 ± 16
ΔSo2	−245 ± 95	−374 ± 5
Tm1, oC	71.5 ± 3.4	69.4 ± 0.4
Tm2	53.2 ± 1.6	47.3 ± 1.0

Parameters are obtained from the thermal melting analysis (Fig. 7). The error margins were determined as root mean-square values obtained in fits to two to three separate measurements in each condition.

In the presence of 1 mM UDM at pH 6.5 the main melting transition is still monophasic. However, the slope of the melting transition of FNR is significantly decreased and the melting transition is less well defined (Fig. 7, squares).

The enthalpy of FNR unfolding at pH 6.5 in the absence and presence of 1 mM UDM derived from three independent experiments was −173 ± 16 and 41 ± 10 kcal/mol, respectively (Table 3). The T_m values derived from these fits were 64.3 ± 0.9 and 67.2 ± 3.2°C. Such a four-fold decrease in the enthalpy and the entropy of FNR unfolding in the presence of UDM could be caused by UDM micelle insertion in the interface between the N- and C-terminal domains, affecting the strength of interdomain interaction in FNR. This inference is confirmed by the thermodynamic analysis of the melting of FNR at pH 5.0 where N- and C-terminal domains of FNR melt independently (Fig. 7, triangles). In the presence of 1 mM UDM the enthalpy and entropy change, associated with unfolding of both domains, increased and the T_m values decreased (Table 3).

Discussion

Concerning the association of FNR with the cytochrome b₆f complex

The presence of FNR in the purified cytochrome b6f complex from higher plants (spinach) was previously documented and discussed in the context of 1) a component of the complex or 2) an artifact of purification (15,32). FNR was consistently observed in SDS-PAGE analysis of the b6f complex at a substoichiometric level relative to cytochrome f (14). This was confirmed by electrospray ionization mass spectroscopy and Western blot analysis (14). In retrospect, the study in Zhang et al. (14) documented conditions under which FNR co-purified with b6f complex, but it is acknowledged that this was not a sufficient proof that FNR is an integral component of the b6f complex.

The involvement of FNR in electron transfer function of the cytochrome b6f complex was supported by a study of Szymanska et al. (33), which utilized cytochrome b₆f complex isolated according to Hurt and Hauska (34), based on a method of purification of the ATP-synthase complex (35). The authors reconstituted cytochrome b₆f complex into liposomes and found that these proteoliposomes have NADPH-plastoquinone oxidoreductase activity. However, this activity was not observed when thylakoid membranes were treated with the chaotropic salt, sodium bromide, before detergent extraction. This procedure was utilized to remove Rubisco and other peripheral membrane proteins including FNR (33), consistent with the inference that this activity was supported by an FNR contaminant. More relevant evidence for FNR interaction with the cytochrome b₆f complex is the 60% inhibition of NADPH-PQ oxidoreductase reaction by NQNO, an inhibitor of the cytochrome b₆f complex (36,37). However, inhibition by NQNO was shown only with the digitonin extract of the b₆f complex, a purification protocol that perhaps did not provide sufficient purity of the b₆f complex (33,38,39) to allow the inference that the b₆f complex is the target for NQNO.

ITC parameters associated with the interaction of FNR with UDM detergent micelles

ITC was employed in the present study to detect and characterize the thermodynamic parameters of interaction between purified b6f complex and FNR. However, ITC measurements in UDM detergent solution did not detect any significant heat release or consumption associated with the proposed interaction in the pH range 5.5–7.5 and at temperatures from 5 to 25°C (Fig. 2 A and B). It is inferred that enthalpic interaction between FNR and b₆f complex is insignificant and their binding, if any, is purely entropic. Measurements at pH < 5.5 could not be conducted as they were affected by significant monomerization of the b6f complex.

The ITC study showed that FNR has a significant affinity for UDM detergent micelles characterized by the following parameters: K_d, 1.3 ± 0.3 x 10⁻⁵ M⁻¹; n, 2 ± 1; ΔH^o, −25 ± 2 kcal/mol; ΔS^o, −62 ± 2 cal/mol^xK (Fig. 3). FNR was also shown to have a significant affinity for lipid membranes (Fig. 4). Such an affinity of FNR for detergent micelles, as well as for lipid membranes detected by ITC (Fig. 4) in this study, is consistent with the ability of FNR to bind nonspecifically with thylakoids, in contrast with its specific interaction through binding with Tic62 via the polyproline type II helix (12,13).

Effect of UDM detergent on FNR structure

CD analysis did not detect changes in the secondary structures of FNR caused by the presence of detergent (Fig. 6 A; Table 2). The analysis in the near-UV region did not detect the loss of CD signals from aromatic side chains (Fig. 6 B). However, information derived from these spectra is limited by the presence of a strong split-CD signal arising from the excitonic dipole-dipole coupling between Trp296 and Trp309, which are located in the hydrophobic cavity of the C-domain away from the interface between FNR domains (Fig. 1 B). This cavity was previously proposed to form an attachment site of FNR to the thylakoid membrane (6). UDM detergent does not affect the split-CD signal, implying that protein folding in this region of the C-domain and possibly of the entire C-domain does not change in the presence of 1 mM UDM.

The N- and C-terminal domains of FNR have pI values of 5.6 and 7.8, respectively. At pH 5.0, the thermal melting of the domains is biphasic (Fig. 7, green inverted triangles), implying that the FNR domains unfold independently. At pH 6.5 the thermal melting of FNR is monophasic (Fig. 7, blue circles). Both domains melt simultaneously in the narrow temperature range with T_m = 64.3°C, forming a single cooperative unit (40). This implies a strong interaction between domains that results in significant stabilization of the N-domain and a decrease in stability of the C-domain.

Thermal melting of FNR in the presence of the UDM detergent is still monophasic. However, the cooperativity of the melting decreased significantly (Fig. 7, triangles), whereas the position of the melting transition (T_m = 67.2 ± 3.2°C) did not change significantly (Table 3).

The comparative thermodynamic analysis of the slope of thermal melting curves at pH 6.5 revealed a four-fold decrease in ΔH^o and ΔS^o of FNR unfolding in the presence of 1 mM UDM (Table 3). The interface between FNR domains is primarily hydrophobic (6). It is proposed that the effect of detergent can be explained by the insertion of UDM micelles into the interface region between FNR domains, which could significantly decrease the electrostatic interaction between N-terminal (pI, 7.8) and C-terminal (pI, 5.6) domains. This inference is consistent with the thermal melting of FNR at pH 5.0 (Fig. 7 triangles) where the FNR domains unfold independently.

The significant decrease in the enthalpy and entropy at pH 6.5 in the presence of UDM could be explained by the decrease of the size of the cooperative unit. At pH 6.5 the cooperative unit could include both domains of FNR (Fig. 1 B). In the presence of UDM detergent the cooperativity of FNR melting is significantly decreased, possibly as a result of UDM micelle(s) insertion in the interdomain space.

The four-fold decrease in the entropy at pH 6.5 in the presence of UDM could imply a small increase in FNR domain ordering and thus correlates with the increase of T_m. It is proposed that the decrease in interdomain interaction caused by UDM restores the C-domain structure disorganized by interaction with the N-domain in the absence of detergent.

ITC measurement of interactions between FNR and photosynthetic electron transfer proteins

A recent study (41) of the FNR interaction with the PSI-LHCI complex can serve as a positive control for the study presented here based on ITC measurements of the interaction of FNR with the cytochrome b6f complex in detergent solution. Titration of the PSI-LHCI complex with FNR, both in 50 mM HEPES, 0.05% n-dodecyl-β-D maltoside detergent, was characterized by Kd = 0.8 μM, and ΔG = −8.3 kcal/mol (41). A similar ITC study of the interaction of ferredoxin with the PSI-LHCI complex in detergent solution (42) detected a weak interaction, close to the detection limit of the ITC method (raw heat rate [RHR] changes <0.1 μcal/s). ITC measurement of the interaction of FNR with Fd in the absence of detergent (43) documented an affinity of ∼1 μM under conditions of low ionic strength (50 mM tricine, pH 8.0) in which interprotein electrostatic interactions can contribute significantly. Under similar conditions, but in the presence of detergent (1 mM UDM), ITC measurements describing the interaction of FNR and b₆f complex did not detect a significant interaction.

Conclusion

In the present study, ITC was used to examine the existence of a significant physical interaction between FNR and the b₆f complex of oxygenic photosynthesis. A significant interaction between FNR and the b₆f complex was not found. It is inferred that binding of the b₆f complex and FNR, if any, is purely entropic. The frequently observed co-purification of FNR with the b₆f complex is associated with a significant affinity of both FNR and the b₆f complex for UDM detergent.

Author contributions

S.D.Z. designed and performed research, analyzed data, and wrote the manuscript. S.S. contributed thermodynamic analysis. Y.M. purified, characterized biochemical properties of recombinant and wild-type FNR. G.K. was responsible for background knowledge of the FNR system. W.A.C. conceived the project, framed questions, analyzed data, and wrote the manuscript.

Editor: Heiko Heerklotz.