A photosynthetic–respiratory electron transport chain chimera based on photosystem I and cytochrome c oxidase on graphene oxide

Significance

Multifunctional biocatalytic cascade systems arising from nanostructured material surfaces have provided valuable platforms for investigations of enzymatic synergism, recovery, and reusability, proving to be important in basic research and applied industrial processes. Here, we report development of a photoswitchable bienzymatic system consisting of colocalized His-tagged derivatives of photosystem I (PSI) core-complex trimers isolated from Synechococcus sp. PCC 7002 and cytochrome c oxidase (CcO) from Rhodobacter sphaeroides. After their attachment to the surface of carboxylated graphene oxide (GO) nanosheets functionalized with Ni-nitrilotriacetic groups, the resulting visible light-driven O₂ consumption relied upon the GO electroactive support together with tunable redox species. For this photoswitchable galvanic biohybrid nanosystem (GBN), assembled by means of coordination chemistry, we interpret these results as arising from PSI described as a simplified quantum system.

Abstract

The self-assembly of enzyme proteins on 2D nanomaterials has enabled the construction and functional control of viable biochemical pathways. However, enzymatic cascades, which combine essential components of the photosynthetic and respiratory electron transport chains in tandem, have thus-far remained elusive. Herein, we have investigated a galvanic biohybrid nanosystem coupling photosystem I and cytochrome c oxidase on the surface of graphene oxide nanosheets in colloidal suspension. The oriented immobilization of the enzymes was facilitated by Ni-coordination sites tethered to the carbon basal plane, with negligible parasitic O₂ consumption. Transient absorption and electrochemical measurements provided evidence of electron transfer between donors and acceptors, leading to light-induced O₂ consumption of up to 70 out of 120 O₂ molecules/s/CcO unit. Graphene oxide behaves as an electronic reservoir and as an electroactive support, enabling electron transport, in concert with cytochrome c, as well as small-molecule redox mediators and reductants. This study provides a state-of-the-art approach for the exploration of photoelectron transfer in membrane-free suspensions of nanosurface-anchored photosynthetic–respiratory enzymatic chains.

Individual enzymes and multienzyme systems stabilized on biopolymers and nanostructured materials have been used as catalysts in lab-scale chemical reactions, applied research, and industrial processes: multifunctional biocatalytic cascade systems can be designed by multienzyme colocalization, fostering enzymatic synergism, recovery, and reusability (1). Notably, graphene derivatives have emerged as versatile enzymatic supports both as nanosheet suspensions (2) and following adsorption onto solid surfaces (3).

As a consequence, a considerable number of biocatalytic cascade systems based on graphene derivatives have recently been developed (2–13) (Table 1). These include colocalized glucose oxidase (GOx) and horseradish peroxidase (HRP), as bridged by Janus DNA and adsorbed onto graphene oxide (GO) within a metal organic framework (4). Since GOx catalyzes the conversion of D-glucose to H₂O₂ and gluconic acid, while HRP reduces H₂O₂ to water, inactivation of GOx is thereby prevented, while its activity becomes enhanced. This has provided a precise, robust system for glucose determination, which holds promise in personal diabetes detection. GOx and HRP have also been immobilized on the surface of GO through a combination of covalent and noncovalent interactions involving the use of 1-pyrenebutanoic acid derivatives, along with 2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid)(ABTS)-encapsulated liposomes (5). This has served to facilitate a one-step colorimetric assay for glucose detection. A related system has also emerged in which GOx and HRP were covalently immobilized on nanocomposite sheets of GO-chitosan, as linked in a 2-D graphene framework through avidin–biotin interaction (6). The confined bienzyme complex exhibited synergistic catalytic efficiency in glucose biosensing and could be regulated by altering interlayer spacing within the 2D scaffold.

Table 1.

Examples of 2D-nanomaterial-based enzyme cascade systems established with graphene oxide derivatives

Colocalized enzyme cascade	Graphene derivative	Physical configuration	Application	Comments	References
Glucose oxidase (GOx)-Horse-radish peroxidase (HRP)	Graphene oxide (GO)	Absorbed on solid surface	Precise glucose detection	GOx inactivation by H2O2 prevented	(4)
GOx-HRP	GO linked via 1-pyrene-butanoic acid derivatives	Covalently and non-covalently bound on GO surface (liposome encapsulated)	Precise glucose detection	Single-step glucose assay	(5)
GOx-HRP	GO-chitosan nano- composite sheets	Avidin–biotin linked in 2D graphene framework	Precise glucose detection	Synergistic catalytic efficiency	(6)
GOx-Cellulase	Covalently GO bound	Nanosheet suspension	Carboxymethyl cellulose conversion to gluconic acid	One-pot conversion	(7)
GOx-Glucoamylase	Reduced GO	Noncovalently bound in nanosheet suspension	Production of gluconic acid from starch	One-pot conversion	(8)
GOx-Myeloperoxidase (MPO)	Triazole functionalized GO nanosheets	Covalently bound	D-glucono-1,5-lactone and H2O2 released, MPO transforms H2O2 to NaOCl, GO derivatives degraded to smaller nanosheets	Antitumor agent Doxorubicin released from nanosheet surface by laser irradiation	(3)
Acetylcholine esterase- Choline oxidase	Size-tunable GO	GO immobilized	Colorimetric assay for organophosphorus pesticide detection	High sensitivity and stability	(9)
Cytochrome P450 bienzyme (CYP1A2- CYP3A4)	Reduced GO	Nanocomposite with Au nano- particles + chitosan	Cascade drug metabolism pathway (conversion of clopidogrel to clopidogrel carboxylic acid	Electrochemically driven transformation of antiplatelet medication	(10)
Pyranose-2-oxidase-Cytochrome P450 peroxygenase (CYP-OleTJE)	Colocalized on GO	Nanosheet suspension	Cascade for α-olefin biosynthesis, oxidase generates H2O2 cosubstrate for P450 peroxygenase conversion of fatty acids into α-olefins	α-olefin yields improved to 143%, liquid–liquid extraction augmented yields, allows enzyme reuse	(11)
Arginase-Urease	Reduced GO field effect transistor	Enzymes embedded in polyethylene-imine	L-arginine biosensor, arginase decomposition to L-ornithine and urea, urease catalyzes urea breakdown	Overall pH rise triggers signal-transduction	(12)
Photosystem I-Cytochrome c oxidase (Photoswitchable bienzymatic system)	GO nanosheets	Nanosheet suspension of His-tagged, Ni-coordinated enzymes	Photosynthetic and respiratory electron transport chains artificially linked in tandem	Catalyzed light-induced electron transport coupled to O2 consumption, with GO acting as electron reservoir and electroactive site, enabling electron transport	This study

Table lists examples of colocalized enzyme systems immobilized on graphene derivatives, existing as nanosheet suspensions or solid surfaces together with additional structural components.

In further investigations of 2D-nanomaterial-based enzyme cascade systems, cellulase together with GOx were covalently bound to GO, enabling the conversion of carboxymethyl cellulose to gluconic acid (7), while GOx and glucoamylase noncovalently bound to reduced GO has been used in the efficient production of gluconic acid from starch (8). A synergistic enzyme cascade comprised of GOx and myeloperoxidase (MPO) was covalently assembled onto graphene nanosheets functionalized with triazole (3). Under these circumstances, GOx converts glucose to D-glucono-1,5-lactone and H₂O₂ in the presence of molecular O₂, while MPO transforms H₂O₂ to sodium hypochlorite. NaOCl causes the graphene derivatives to biodegrade to smaller nanosheets, while doxorubicin loaded on the surface is released by laser-irradiation, serving as an antitumor agent. Overall, this enzymatic nanoplatform mimics white blood cell activity and also exhibits neutrophil-like activity. In a further example of a biocatalytic cascade system, acetylcholinesterase and choline oxidase were immobilized onto GO, permitting the detection of organophosphorus pesticides (9).

In other related efforts, the cytochrome P450 (CYP) bienzymatic complex was assembled on gold nanoparticles/chitosan/reduced-graphene-oxide and driven electrochemically (10). The isoenzyme CYP1A2 generates 2-oxo-clopidogrel from the substrate clopidogrel, and then the isoenzyme CYP3A4 converts 2-oxo-clopidogrel to clopidogrel carboxylic acid, thus facilitating in-vitro exploration of a multienzyme complex metabolic pathway. In addition, the synthesis of α-olefins from fatty acids was accomplished by a cascade reaction carried out by pyranose-2-oxidase and a cytochrome P450 peroxygenase from Jeotgalicoccus sp. 8456 (OleT_JE), a member of the CYP152 enzyme family (11). The first enzyme produces ketoglucose and the H₂O₂ cosubstrate from D-glucose in the presence of O₂, while the second enzyme converts the fatty acids to α-olefins. The yield of the final product was 143% higher than in the case of the free enzymes.

A reduced graphene oxide field-effect transistor was employed in the detection of L-arginine as part of a biosensor fabricated together with an enzymatic cascade of arginase and urease embedded in polyethyleneimine (12). The stereospecificity and high selectivity toward this amino acid were based upon the arginase-catalyzed decomposition of L-arginine to L-ornithine and urea, while urease catalyzes the subsequent breakdown of urea to ammonium cations, hydrocarbonate, and hydroxide anions. An overall pH increase provided a trigger for signal transduction. In an important example of the development of a novel bioconjugates-driven colorimetric immunoassay, visible light irradiation of GO was employed in the photoswitching of HRP enzymatic activity, thereby obviating the need for peroxides and acidic stop solutions as additives (13). Typical substrates such as 3, 3′, 5, 5′-tetramethylbenzidine and ABTS are oxidized due to the in-situ formation of superoxide anions ($O_2^{\bullet -}$) and photogenerated holes ($h^{+}$). The same bioconjugate can act as an efficient signal reporter for bioassays, including in the detection of α-fetoprotein serum levels, which is useful in maternal screening for chromosomal abnormalities such as neural defects and Down’s syndrome (14).

Both photosynthetic and respiratory enzymes have been separately investigated on solid polymeric surfaces and on GO derivatives in suspension (2, 15). In the latter case, the native trimeric and nonnative monomeric photosystem I (PSI) core complexes (CCs) isolated from Thermosynechococcus elongatus were embedded in a conductive polymer film comprised of osmium 2,2′-bipyridine chloride and polyvinylimidazole (15). The polymeric film was the central component of a photoelectrochemical cell, including methyl viologen (MV) as a final acceptor. The trimers were shown to display saturation at higher light intensities as opposed to the monomers and to exhibit energy sharing among the subunits. In the former case (2), nanoconjugates based on carboxylated graphene oxide (CGO) functionalized with nickel nitrilotriacetic groups (CGO-NiNTA) coordinating His-tagged cytochrome c oxidase (CcO) from Rhodobacter sphaeroides have been fabricated. CcO stabilized on the CGO-based support was capable of an O₂ reduction reaction induced by cytochrome c (Cyt c) in solution, with a turnover rate as high as 240 O₂ molecules per s per CcO unit.

To the best of our knowledge, no attempts have been carried out to combine PSI and CcO and to overcome the challenges of matching these redox enzymes in a functional electronic cascade. In this respect, it is useful to recall the main characteristics of PSI and CcO. PSI is a multiprotein enzyme and is typically found in cyanobacterial membranes in trimeric form (16). PSI indirectly accepts electrons generated within PSII and increases their reduction potential by translocation through the thylakoid membrane. The 2.5-Å resolution crystal structure of PSI isolated from the thermophilic cyanobacterium Synechococcus elongatus has shown that it is comprised of twelve protein subunits (PsaA–PsaF, PsaI–PsaM, and PsaX) and 128 cofactors (17). The latter consists of 96 chlorophyll (Chl) a molecules, two phylloquinones, three Fe₄S₄ clusters, 22 β-carotene molecules, and in addition 4 embedded lipids and a presumed Ca²⁺ cation. The PsaA-PsaB polypeptide heterodimer houses the Chl a dimer (Chl_1A and Chl_1B) forming the primary P₇₀₀ electron donor at the base of a branched electron transfer chain. The A branch contains the A₀ primary acceptor formed by a dimer of accessory Chl_2A and secondary Chl_3A which transfers electrons to the A₁ phylloquinone molecule (Q_KA), subsequently conveyed onto the single Fe₄S₄ cluster F_X (18). The extrinsic PsaC protein (19, 20) hosts the additional Fe₄S₄ clusters, F_A and F_B, which function as the respective sequential terminal electron acceptors. In this article, we define a PSI core complex (PSI CC) as a PSI monomer including nearly all the structural elements listed above. PSI from cyanobacterial sources tends to form trimers of the unit structure (designated as PSI CCs) (19).

The R. sphaeroides CcO (20–22) represents a well-described aa₃-type cytochrome c oxidase, belonging to the A-family mitochondrial-like cytochrome c oxidases, which serves as an important CcO model system. CcO is not to be confused with the alternative R. sphaeroides cbb3-type oxidase, which acts as a high affinity c-type cytochrome oxidase, that functions under low O₂ tension, with both respiratory and regulatory roles (23). The R. sphaeroides CcO is comprised of four subunits, CoxI (CtaD), CoxII (CtaC), CoxIII (CtaE), and CoxIV (SU I, II, III, IV, respectively), together with 1 heme a, 1 heme a₃, 3 Cu atoms (including 2 Cu_A and 1 Cu_B which forms a Fe-Cu binuclear center with heme a₃), 1 Mg and 1 Ca atom (22). In our previous study (2), a His₆-tag was inserted at the C-terminus of CoxI through genetic modification of the ctaD structural gene (24) capable encoding the 63,147 Da protein as deduced from the full-length gene sequence (25), altered to 64,077 Da with addition of the hexa-His tag. The His₆-tagged CoxI version was designed for linkage of the CcO complex by immobilization on CGO-NiNTA nanosheet suspensions (2).

In this study, we assembled a photoswitchable bienzymatic system colocating PSI CC trimers isolated from Synechococcus sp. PCC 7002 and CcO isolated from R. sphaeroides on the surface of GO nanosheets. We designate this photosynthetic/respiratory bienzymatic cascade immobilized on 2D nanostructures in suspension as a Galvanic biohybrid nanosystem (GBN). Visible light-irradiation of PSI was used to trigger the reduction of CcO and initiate its O₂ consumption activity. Joliot-type spectroscopy (JTS) was employed for kinetic investigation of the immobilized PSI as a function of light-irradiance and concentration of the sacrificial electron donor, while a Clark electrode served to independently probe the O₂ reduction reaction of CcO. The behavior of PSI was simulated based on a kinetic model representing PSI as a simplified but sufficiently sophisticated quantum system.

The main motivation for our study was to explore a cell-free, membrane-free nanosystem that contains redox enzymes representing the photosynthetic and respiratory electron transport chains as kinetic models of photoelectron transport. The improved understanding of photoswitchable nanosystems in suspension will serve to shed light on redox cascades with potential artificial photosynthesis, bioelectronic and biotechnology applications. Our approach enables bulk kinetic investigations of the GBN and takes advantage of the enormous surface provided by an ensemble of GO-based biocircuits. This feature will greatly benefit potential use cases as well, including fuel generation.

Results

Photosynthetic/Respiratory Bioconjugates and Galvanic Biohybrid Nanosystem.

We have assembled a chimeric photosynthetic/respiratory electron transport chain based on PSI CCs (trimers), CcO (monomer), and a GO derivative (GO-NiNTA) in suspension (Fig. 1A and SI Appendix, Figs. S1 and S2). The chemical structure of a fragment of GO-NiNTA is shown in SI Appendix, Fig. S3. The size of GO-NiNTA is determined by that of the precursor GO which averages at about 3 μm in its main dimension (26). PSI and CcO enzymes are present on the surface at a molar ratio of ~3:1 on a trimer-to-monomer basis. PSI serves as the primary site of a charge separation event which entails the oxidation of $P_{700}$ to $P_{700}^{+•}$ (Fig. 1 B and C). PSI can be understood as a two- or three-level energy system. The re-reduction of $P_{700}^{+}$ to $P_{700}$ is accomplished by ascorbate (Asc) transferring electrons to dichlorophenolindophenol (DCPIP). DCPIP is a highly specific chemical donor (27), capable of serving as a redox mediator by docking at the luminal side of PSI (28). The high reactivity and accessibility of DCPIP serves to accelerate P700⁺ reduction, showing kinetic similarities with plastocyanin, the natural P700⁺ electron donor in thylakoids (28).

Fig. 1.

Representation of PSI-(GO-NiNTA)-CcO as a galvanic biohybrid nanosystem (GBN). (A) The structures of CcO, Cyt c, and trimeric PSI components are shown along with their related chemical/biochemical pathways as accomplished with appropriate redox mediators. Ox, oxidized; Red, reduced. This represents a composite image of the protein structures obtained from Pymol and is approximately drawn to scale in top–down perspective. The same image is shown in SI Appendix, Fig. S2 from different points of view. The depiction of Cyt c floating in solution shows this protein approaching from above its binding site with CcO on SU II; four associations of this type are necessary to reduce O₂ to 2 H₂O molecules. The orange-colored PSI protein is PsaL, which is involved in the trimerization process. The single yellow CcO α-helix depicts SU IV. SU IV is a labile component not essential for activity which is absent in our CcO, but is included for the sake of completeness. (B) PSI-CcO shown as a two-level and (C) as a three-level energy system. See text for further details.

Asc is added in excess (under flooding conditions) as typical in kinetic chemistry and biochemistry studies. This choice is dictated in order to obtain a simplified pseudo-first-order behavior. Asc alone does not transfer electrons to PSI in the absence of horse-heart Cyt c. We cannot exclude that Asc may donate electrons to CcO directly although the energy difference is greater than the difference between Cyt c and CcO (vide infra). Therefore, the latter electron-transfer should be favored. MV acts as the final electron acceptor on the stromal side of PSI. In our GBN, at the same time, horse-heart Cyt c acts as a redox protein donor highly specific for CcO, while also providing electrons to PSI. Cyt c₆ is more specific to cyanobacterial PSI (15, 29), but our goal was to maximize O₂ consumption in the GBN. Reduced CcO carries out the O₂ reduction reaction to H₂O.

GO-NiNTA is obtained through functionalizing the surface and the edges of the carbon basal plane with nitrilotriacetic groups coordinating Nickel cations (NiNTA) (2, 30). NTA is a tridentate ligand often used in immobilized metal affinity chromatography applications for the immobilization of recombinant proteins labeled with a terminal His-tag (31, 32). The mechanism of protein immobilization is well known and entails the replacement of three water molecules weakly coordinated to Ni²⁺ by His⋅⋅⋅Ni²⁺ interactions (33). The equilibrium dissociation constant is 10 μM. GO-NiNTA is not simply an anchoring support for His-tagged enzymes, but also acts as an electroactive platform enabling electron transport among the associated enzymes. Chemically modified GO can accept electrons or donate electrons depending on the position of its Fermi level with respect to surrounding redox species (2). Therefore, GO can mediate electron transport to and from redox molecules with which it comes into contact (vide infra).

While the Nickel-coordination sites are likely to direct the initial attachment of His-tagged PSI CCs and CcO, nonspecific binding ultimately prevails. Indeed, after immobilization on GO-NiNTA, treatment with imidazole (Im) does not remove the His-tagged enzymes either together or individually. This was shown by flow cytometry measurements performed on the GO-NiNTA-PSI and GO-NiNTA-CcO bioconjugates, as well as on the GBN PSI-(GO-NiNTA)-CcO (SI Appendix, Figs. S4–S6). Statistical analysis employing the dependent t test of paired samples (before and after Im treatment) indicates that there is no significant difference between the mean of sample differences of both the side scattering integral (SS INT) and the fluorescence integral (FL1 INT) with respect to the forward scattering integral (FS INT). This result is estimated to be true with 95% confidence. Such SS INT, FL1 INT, and FS INT analyses provide meaningful information about the internal complexity of the samples, their fluorescence, and their size, respectively.

The maps of FS time of flight (FS TOF) vs. FS INT (SI Appendix, Figs. S4–S6) indicate that there is no change in aggregation level before and after Im treatment. Singlet events are detected below 300 units of FS TOF and correspond to individual bioconjugates and GBN particles. Above 300 units of FS TOF doublets and multiplet events signify aggregates of the bioconjugates and GBN particles.

Joliot-Type Spectroscopic Analysis.

The GBN represented in Fig. 1A and its subcomponents can be investigated by means of JTS (Fig. 2) to directly measure PSI CCs (29), either free in suspension or anchored to the surface of GO-NiNTA at pH of 6.5 (29) in MES buffer, in order to assess PSI photosynthetic activity. JTS entails the use of probing light at 700 to 740 nm filtered at 705 ± 3 nm, green actinic light at 522 nm, and orange actinic light at 639 nm. Changes in the difference absorption spectrum with respect to the probing light between the sample and the reference can be measured with time. Moderate irradiation at 522 nm (50 μE/m²s) causes the primary charge separation within PSI to approach a steady state, while intense orange light (3,000 μE/m²s) is capable of saturating the photooxidation process, driving complete oxidation of $P_{700}$ to $P_{700}^{+\bullet }$. When both actinic lights are turned off, the chemical donors present in solution can re-reduce PSI to its original state in the dark phase. The changes can be understood in terms of appropriate kinetic equations. The oxidation of $P_{700}$ to $P_{700}^{+\bullet }$ can be initially described in terms of a two-level system, as shown by Eq. 1:

$$P_{\mathrm{700}}\begin{array}{c}\stackrel{k_1\left[h\nu \right]}{⟶}\hfill \\ \stackrel{k_{-1}\left[D\right]}{⟵}\end{array}{P}_{\mathrm{700}}^{+\bullet }+e^{-},$$ [1]

Fig. 2.

Absorbance changes of PSI CCs, GO-NiNTA-PSI, and CcO-(GO-NiNTA)-PSI in suspension as determined by JTS. (A) Effect of DCPIP concentrations on PSI-CCs. (B) Effects of light irradiance on PSI CCs. (C) Effects of DCPIP concentrations on GO-NiNTA-PSI bioconjugate. (D) Effect of DCPIP concentrations on CcO-(GO-NiNTA)-PSI GBN. Regions I and II represent the light and dark phases, respectively.

where the rate constant of the forward reaction is $k_1[hv]$ and is proportional to light irradiance; the reverse reaction has a rate constant equal to $k_{-1}[D]$ that is proportional to the concentration of reduced donor D (e.g., DCPIP). In our model we assume that chemical re-reduction in the dark phase (i.e., $k_{-1}[D]$) occurs at the same rate as in the light phase, when primary charge separation is taking place within PSI. We are not aware that other authors have explicitly adopted such assumption. The mass conservation principle implies that:

$${\left[P_{700}\right]}_{\mathit{tot}}=\left[P_{700}^{+•}\right]+\left[P_{700}\right].$$ [2]

The reaction rate can be expressed in terms of the consumption of $P_{700}$ or production of $P_{700}^{+}$, as shown in Eq. 3:

$$-\frac{d[P_{700}]}{\mathit{dt}}=\frac{d[P_{700}^{+•}]}{\mathit{dt}}=k_1[h\nu ][P_{700}]-k_{-1}[D][P_{700}^{+•}].$$ [3]

Setting the time derivatives and, consequently the right-hand side equal to 0 provides the steady-state condition:

$$\frac{{\left[P_{700}^{+•}\right]}_{s.s.}}{{\left[P_{700}\right]}_{s.s.}}=\frac{k_1\left[h\nu \right]}{k_{-1}\left[D\right]}.$$ [4]

When the light is off, the light term goes to zero, and differential Eq. 3 can be easily solved to predict the change of $P_{700}^{+•}$ in the dark phase:

$$\left|{\left[P_{700}^{+•}\right]}_t\right|=\left|{\left[P_{700}^{+•}\right]}_0\right|·e^{-k_{-1}\left[D\right]·t}.$$ [5]

If Eq. 3 is solved in the general case, by invoking Eq. 2, the general solution in the light phase is found to be

$$\left|{\left[P_{700}^{+•}\right]}_t\right|=\left|\frac{{\left[P_{700}\right]}_{\mathit{tot}}{k}_1\left[h\nu \right]}{k_1\left[h\nu \right]+k_{-1}\left[D\right]}\right|\left(1-{e^{-}}^{\left\{k_1\left[h\nu \right]+k_{-1}\left[D\right]\right\}·t}\right).$$ [6]

Eq. 6 has the form of a Poisson distribution (16). The simple model for PSI illustrated above was used as a starting point in the analysis of JTS data collected for PSI CCs in suspension. In Fig. 2A, the kinetic profiles of PSI CCs in suspension are shown as a function of increasing DCPIP concentration. Region I from time t = 0 corresponds to the “light phase.”

The comparison of Eq. 6 with Eq. 7 indicates that besides a Poisson distribution term, there must be an additional exponential term, as in the function $g(t)$:

$$g(t)=y_1+A_0·\left[1-\exp \left(-\frac{t}{{\tau }_1}\right)\right]+A_2·\exp \left(-\frac{t}{{\tau }_2}\right).$$ [7]

Eqs. 7 and 11 (see Materials and Methods section) are equivalent under the following conditions:

$$y_0=y_1+A_0{A}_1=-A_0.$$ [8]

The additional term $A_2·\exp \left(-\frac{t}{{\tau }_2}\right)$ is not contemplated in our model elucidated above and requires a separate justification.

PSI can be better explained as a three-level system in terms of the energy levels $P_{700}$, $P_{700}^{\ast }$, and $P_{700}^{+•}$ (Fig. 1C). The oxidation of $P_{700}$ to $P_{700}^{+•}$ via $P_{700}^{\ast }$ corresponds to the time constants ${\tau }_{h\nu }$. The reverse reaction from $P_{700}^{+•}$ to $P_{700}$ corresponds to the time constant ${\tau }_D$. Based on Eq. 6 in comparison with Eq. 8, it can be deduced that the ${\tau }_{h\nu }$ and ${\tau }_D$ are related to ${\tau }_1$ by (Eq. 10):

$$\frac{1}{{\tau }_1}=\frac{1}{{\tau }_{h\nu }}+\frac{1}{{\tau }_D}.$$ [9]

The additional exponential term can be connected to resonant nonradiative exciton transfer (Förster mechanism) (34) between PSI CCs in suspension or immobilized on the surface of GO-NiNTA. The time constant of this transfer is ${\tau }_2$, where $\frac{1}{{\tau }_2}=k_2\left[A^{+}\right]$. $A^{+}$ corresponds to oxidized donors ($D^{+}$) or $P_{700}^{+•}$ of neighboring PSI CCs. ${\tau }_2$ implicitly depends on light due to the light-dependence of $\left[A^{+}\right]$.

The effect of light irradiance on PSI CCs is shown in Fig. 2B. It can be seen that the steady state is achieved much more rapidly with increasing green light intensity. When PSI is immobilized on GO-NiNTA yielding GO-NiNTA-PSI it is possible to show that PSI remains active albeit with lower activity, as shown in Fig. 2C. A higher concentration of DCPIP is necessary to see appreciable differences in the JTS kinetic profiles: concentrations of 100 to 180 μM and 200 to 280 μM were used in the light and dark phases, respectively. PSI CCs are comprised of trimers His-tagged on the luminal side (Fig. 1A). Access to the acceptor-side pocket is likely to be hindered if coordination to the Ni ions occurs as expected. A preferential orientation may explain the need for a higher concentration, but this explanation is weakened by the fact that nonspecific binding of PSI also takes place as confirmed by flow cytometry. Indeed, nonspecific binding is the dominant interaction in our system. Labeling on the stromal side would have been preferable if metal-coordination was the prevailing interaction. Since nonspecific binding takes over, this requirement proved to be not as crucial because PSI can take many different possible orientations. Alternatively, or concomitantly, it is possible that the Fermi level of GO-NiNTA is low with respect to pristine GO due to the Ni²⁺ positive charges introduced on the surface.

The coassembly of PSI and CcO on GO-NiNTA to give PSI-(GO-NiNTA)-CcO was investigated in Fig. 2D. The appearance of the JTS kinetic profiles is greatly affected by the presence of CcO as opposed to GO-NiNTA-PSI. This evidence indicates that CcO, as expected, behaves as an electron acceptor with respect to PSI and consumes O₂ in the process. The need to use higher concentrations for some of the samples is necessary to detect sufficient differences in activity. We did measure all samples in the entire concentration range, but meaningful activity was observed only for the traces we have shown. In the case of PSI-(GO-NiNTA)-CcO we observe differences in activity in a wider range. DCPIP is capable of reaching PSI even at low concentration, but the re-reduction is slower. This can certainly be due to steric hindrance to some extent. A higher concentration will be able to access PSI more easily but could also provide electrons to CcO indirectly through GO.

The JTS data can be analyzed in terms of four time constants, namely, ${\tau }_1$, ${\tau }_2$, ${\tau }_{h\nu }$, and ${\tau }_D$. The changes of the prefactors $A_1$ and $A_2$ can be described in terms of fractional percentage ${\chi }_j:$

$${\chi }_j=\frac{A_j}{A_1+A_2},$$ [10]

with $j=1,2$. Alternatively, $A_1$ and $A_2$ (positive in value) can be normalized with respect to the saturation level $A_{\mathit{sat}}$ (negative in value) caused by the orange light: $-\frac{A_j}{A_{\mathit{sat}}}$. In Fig. 3, the values of the fitting parameters ${\chi }_j$, $-\frac{A_j}{A_{\mathit{sat}}}$, and the logarithm of the time constants ${\tau }_i$ are plotted with respect to the DCPIP added concentration and light irradiance in the case of PSI CCs. Based on Fig. 3 A and B, the contribution of term 1 increases, while that of term 2 decreases with DCPIP concentration, as shown in the range of 10 to 140 μM. The change of the logarithms of ${\tau }_1$, ${\tau }_2$, ${\tau }_{h\nu }$ and ${\tau }_D$, with the same range of DCPIP concentration (Fig. 3C), shows that there are two regions. In region I, the time constants exhibit different slopes and converge to a common slope in region II. This trend suggests that the steady state is achieved much faster at higher concentrations of donors. When considering the light effect, the importance of term 1 decreases to zero, while term 2 increases to 100% (Fig. 3 D and E). It can be seen that ${\tau }_2$ is most dramatically affected by increasing light irradiance (Fig. 3F). As indicated above, ${\tau }_2$ depends on light because $\left[A^{+}\right]$ also does. Increased light irradiance increases Förster resonance between neighboring PSI CCs.

Fig. 3.

Analysis of fitting parameters of absorbance changes of PSI CCs in suspension determined by JTS. (A–C) Effect of DCPIP added concentration and (D–F) Effect of light irradiance.

A similar analysis of the fitting parameter as a function of the DCPIP concentration can be conducted for the samples GO-NiNTA-PSI and PSI-(GO-NiNTA)-CcO (assembled in situ), as shown in Fig. 4. The presence of a higher added concentration of reduced DCPIP is necessary for overcoming the competition of the electroactive support with PSI for accepting the incoming electrons. As evinced from Fig. 4A, term 1 slightly increases while term 2 slightly decreases with increasing donor concentration when plotting terms 1 and 2. Fig. 4B indicates that after normalization with respect to the saturation level, both terms decrease with donor concentration with a similar slope. We deduce that therefore, GO plays the role of a redox mediator for both types of events. This behavior is different from that observed with PSI CCs in suspension, where term 1 clearly increases, while term 2 decreases with DCPIP concentration (Fig. 3 A and B). These observations suggest that the presence of the support serves to couple term 1 events and those associated with term 2. Furthermore, term 1 is systematically more important than term 2 in comparison with PSI CCs. This indicates that Förster resonance becomes less relevant when PSI is immobilized onto GO. The time constants (Fig. 4C) decrease with DCPIP concentration but are systematically slightly higher than the profiles of the same quantities for PSI CCs (Fig. 3C). In particular, for GO-NiNTA-PSI, ${\tau }_2$ is more affected by donor concentration than other time constants due to its dependence on $\left[A^{+}\right]$.

Fig. 4.

Analysis of fitting parameters for absorbance changes of GO-NiNTA-PSI and CcO-(GO-NiNTA)-PSI in suspension as determined by JTS. (A–C) Effect of DCPIP added concentration on GO-NiNTA-PSI and (D–F) Effect of DCPIP added concentration on CcO-(GO-NiNTA)-PSI.

In the case of PSI-(GO-NiNTA)-CcO, the fractional percentages of terms 1 and 2 are roughly constant with DCPIP concentration, with a dominant term 1 (Fig. 4D). The values of $-\frac{A_i}{A_{\mathit{sat}}}$ decrease with donor concentration for both terms 1 and 2, as observed for GO-NiNTA-PSI (Fig. 4E). Again, it is surmised that a coupling occurs between the events underlying term 1 and term 2 as mediated by the electroactive support. The DCPIP concentration interval that was explored was 20 and 260 μM. The PSI activity in PSI-(GO-NiNTA)-CcO is systematically lower than that of GO-NiNTA-PSI (Fig. 2 D and C, respectively). At the same time, the contribution of term 1 is systematically more important than that of term 2, even more than observed with GO-NiNTA-PSI, and contrary to PSI CCs (compare Fig. 4 D and E with Figs. 3 A and B and 4 A and B). We deduce that PSI is capable of transferring electrons to CcO in the electronic cascade we have established. Presumably, the reduction of CcO is accompanied by O₂ consumption (vide infra). This is also reflected by the profiles of the time constants of PSI-(GO-NiNTA)-CcO and GO-NiNTA-PSI (Fig. 4 C and F, respectively). The values of the logarithms of ${\tau }_1$, ${\tau }_2$,${\tau }_{h\nu }$ and ${\tau }_D$ with DCPIP concentration are systematically somewhat higher for PSI-(GO-NiNTA)-CcO than for GO-NiNTA-PSI. This observation indicates that CcO is an effective electron acceptor for PSI despite the competing MV. Overall, these results demonstrate that we have succeeded in assembling PSI-(GO-NiNTA)-CcO as a functional GBN. We are not aware whether the biphasic to monophasic behavior that we measure for free PSI (Fig. 3 A and B) has been previously documented. This phenomenon needs to be compared with the biphasic behavior of immobilized PSI, which appears at all concentrations (Fig. 4 A, B, D, and E). We believe that the second exponential term (Fig. 2) represents aggregation and therefore Förster energy transfer even in the case of PSI alone.

Electrochemical Oxygen-Consumption Measurements.

Independent measurements of O₂ consumption using a Clark electrode were conducted as a function of pH on PSI CCs, GO-NiNTA-PSI, and PSI-(GO-NiNTA)-CcO in the presence of the sacrificial donors dithiothreitol (DTT) and/or Cyt c with MV as terminal acceptor. Absolute O₂ level may vary with electrode response after calibration and can be due to differences in the colloidal suspensions. Another source of change may be related to alterations in Zeta-potential resulting from photoexcitation. Also, the ionic strength of the medium will change for different concentrations of DTT even for the same sample. All these factors, including electrodic drift can modify the absolute level of the traces but not as much as their slope changes. Furthermore, a slow equilibration with atmospheric oxygen inevitably occurs when the light is off. In Fig. 5, the results for pH 7.4 are shown, reflecting the optimum physiological conditions for CcO (2). A more complete set of data on initial rates of O₂ consumption, after addition of the sacrificial donors and with light on and off conditions, is given in SI Appendix, Tables S1–S3, for the respective pH values of 7.4, 8.0, and 9.0. In the dark, minimal O₂ consumption is observed, which can be attributed to the presence of Ni coordination sites. The initial rate of O₂ consumption, when only DTT is present under light conditions, increases with increasing DTT concentration (Fig. 5A). Even without DTT, PSI can reduce O₂ to some extent by what is known as the Mehler reaction (35). Once Cyt c is added, O₂ consumption is systematically higher and nearly constant for all levels of DTT addition (Fig. 5B). This is due to the fact that Cyt c reduced by DTT appears to have a higher specificity as a redox donor to PSI as opposed to reduced DTT. The bioconjugate GO-NiNTA-PSI is capable of consuming O₂, albeit less than for free PSI CCs (Fig. 5C). The inclusion of CcO yields the complete PSI-(GO-NiNTA)-CcO nanosystem. This GBN can consume O₂ without DTT, just by virtue of Cyt c reduced by PSI (Fig. 5C). However, when DTT is added, a deeper level of oxygen O₂ consumption can be achieved. It should also be noted that significantly more oxygen O₂ is consumed when CcO operates in series with PSI (Fig. 5D). It can be calculated that the oxygen O₂ consumption was as high as 117 O₂ molecules per second per CcO unit in the GBN (pH = 7.4). The baseline oxygen O₂ consumption in the dark is 45 O₂ molecules per second per CcO unit. This is equivalent to a bioconjugate GO-NiNTA-CcO hindered by the presence of inactive PSI CCs. Therefore, 62% of CcO activity in the light is triggered by the photoelectronic events within PSI (72/117). These rates should be compared with caution to what was previously observed in the bioconjugate CGO-NiNTA-CcO, namely, 240 O₂ molecules per second per CcO unit (2). It should also be noted that no oxygen O₂ consumption occurs in free CcO in suspension under similar conditions (SI Appendix, Fig. S7) (2).

Fig. 5.

Effect of donor and acceptor additions on time course of O₂ consumption by suspension of PSI CCs, GO-NiNTA-PSI, and CcO-(GO-NiNTA)-PSI. (A) PSI CCs with DTT; (B) PSI CCs with oxidized Cyt c and DTT; (C) GO-NiNTA-PSI with oxidized Cyt c and DTT; (D) CcO-(GO-NiNTA)-PSI with oxidized Cyt c and DTT. Down and up arrows indicate when white light was turned on and off, respectively; asterisks denote DTT addition. A given arrow applies to different traces showing a change of slope in a similar vertical position when there is no room for individual arrows. Additions of DTT at time t before the beginning of the experiment, if applicable, are indicated as * (t <0 s).

We explored the initial rates of O₂ consumption of the GBN and its subcomponents in the dark following the addition of DTT, after turning on white light and immediately after switching it off, as a function of pH (Fig. 6 and SI Appendix, Fig. S8). It can be seen that PSI CCs under illumination consume O₂ at the highest rate at pH = 9.0. (Fig. 6 A–C). When PSI is immobilized onto GO-NiNTA, the O₂ consumption under light is lower than before illumination when DTT is present (green trace) (Fig. 6 D–F). The best performance in the light of PSI-(GO-NiNTA)-CcO occurs at pH = 7.4 (Fig. 6G). This is expected considering that the physiological pH of CcO is close to this value of acidity (2). On the contrary, at pH = 9.0, the highest O₂ consumption takes place with DTT before the light is switched on (Fig. 6I). This is due to the fact that the concentration of deprotonated thiols of DTT, with their reduction activity, becomes much more significant as pH approaches ${\mathit{pK}}_{a,1}=9.2$ and ${\mathit{pK}}_{a,2}=10.1$. The comparison of Panel C and Panel I indicates that pH 9.0 did not remove the Fe-S clusters from PSI with a DTT concentration below 2 mM. We note that Asc was not used in our oxygen consumption measurements as was the case in our previous study on CGO-NiNTA-CcO (2).

Fig. 6.

Initial O₂ consumption rates of PSI CCs, GO-NiNTA-PSI, and CcO-(GO-NiNTA)-PSI as a function of pH plotted vs. added volume of 0.1 M DTT in presence of Cyt c. (A–C) PSI CCs; (D–F) GO-NiNTA-PSI; (G–I) CcO-(GO-NiNTA)-PSI. The rates at pH 7.4 were measured from the profiles shown in Fig. 5 B–D. Green symbols, measurements before light turned on; orange and black symbols, measurements made under light on and light off conditions, respectively. See SI Appendix, Fig. S8 for comparison.

Discussion

The assembly of the PSI-(GO-NiNTA)-CcO GBN requires the immobilization of the membrane enzymes PSI trimers and CcO on the surface of GO-NiNTA nanosheets in suspension. Both PSI and CcO are His-tagged, and their affinity for Ni²⁺ coordination sites directs initial immobilization. Ultimately, the nonspecific binding to hydrophobic patches on the carbon basal plane locks both proteins into place. The attachment of these enzymes is likely to occur with different orientations on both sides of the 2D nanosurfaces. Regardless of the manner in which enzymatic anchoring occurs, electron transport among the enzymes, the electroactive support, and the redox shuttles are determined by their intrinsic thermodynamic properties and by the electromagnetic potential landscape established on the self-assembled nanostructures.

CcO alone does not consume oxygen (SI Appendix, Fig. S7). We have previously shown that the presence of CGO-NiNTA is necessary for oxygen consumption in CGO-NiNTA-CcO (2). Zhu et al. incontrovertibly demonstrated that CGO-NiNTA is an electroactive support. Furthermore, basic physics principles support our previous and present findings about the conducting nature of the GO-based support. GO is known to present $\pi$ conjugated regions on the carbon basal plane (26). A moderate horizontal electron/hole conductivity, which is enabled by a percolation mechanism, does exist in GO (36). More importantly, a one-atom thick GO nanosheet is known to conduct vertically as well. PSI and CcO are present on both sides of GO-NiNTA and are in intimate contact with the support either through the NiNTA ligands or by nonspecific binding with GO. In a sense these proteins are wired to the GO-NiNTA support. Therefore, a combination of horizontal and vertical electron transfer will allow electrons to virtually reach any position on the carbon basal plane and on the surface of the proteins. Vertical electron transfer between PSI and CcO is likely to be faster and most efficient. Indeed, PSI and CcO are redox proteins which transfer electrons internally and present regions with different electromagnetic potentials on their surfaces. Externally available electrons on the carbon basal plane will be able to spontaneously move toward more positive midpoint reduction potential values (E_h) and toward positive electromagnetic potentials in general. Stabilization of CcO by protein–protein interaction cannot be excluded, but the possibility of electron transfer is undeniable whenever contact is established between the proteins and the intrinsically conductive support. Too-tight protein–protein stabilization would also cause steric hindrance, which does not appear to be the case here. While we do not know the details of the spatial distribution of the proteins on the GO surface, we do know that complete binding occurred between GO-NiNTA and the proteins.

The relationship between the energy levels of the redox mediators required for operation of the PSI-(GO-NiNTA)-CcO is elucidated by the energy diagram presented in Fig. 7. Asc or DTT (the latter was used in the O₂ measurements) reduce DCPIP⁺ to DCPIP with subsequent protonation. An incoming photon causes a primary charge separation within PSI, generating the radical cation $P_{700}^{+•}$ from $P_{700}$ via $P_{700}^{\ast }$. An electron of higher reduction potential (−450 mV) is released on the stromal side as DCPIP re-reduces PSI on the luminal side. Asc or DTT can also reduce Cyt c, which in turn reduces heme a₃ and the Cu_B site of CcO (20). Consequent reduction of O₂ to water occurs in the presence of H⁺, serving as the O₂-reducing chemical protons, as opposed to the pumped protons, which establish an electrochemical proton gradient. Both PSI and CcO in their reduced states can transfer electrons to GO-NiNTA, changing the Fermi level of this nanomaterial and thereby adapting to the inflow and outflow of electrons as provided by each complex.

Fig. 7.

Energy diagram showing midpoint reduction potential and vacuum energy levels of the galvanic biohybrid nanosystem PSI-(GO-NiNTA)-CcO. Energy levels are shown for Asc, Cyt c, DCPIP⁺, and MV; DTT can also serve as sacrificial donor. The color-gradient used for GO-NiNTA signifies that the Fermi-level position can vary.

In contrast to the artificial electron donors used in this study, Bruce et al. reported a JT LED-driven analysis of P700⁺ reduction kinetics using the natural electron donor, Cyt c₆, together with the trimeric PSI complex from T. elongatus (29). Western blotting established a physiologically relevant intracellular ratio of ~1.7 Cyt c₆/PSI. The JTS LED pump–probe system facilitated the use of lower PSI levels, avoiding the much higher concentrations required for previous laser-based, transient absorption analyses of PSI electron transfer kinetics (37, 38). As a result, light-source dependent kinetic alterations were facilitated by the JTS system, in which the intensity and duration of LED flashes could be varied (29). Accordingly, these differences were found to be protein-concentration dependent and could be attributed to multiple turnovers arising during the Cyt c₆-induced $P_{700}^{+•}$ reduction, occurring in the dark phases of the extended illumination regimes. The kinetics of $P_{700}^{+•}$ reduction was characterized by a two-phase recovery with a fast-phase half-time of 6.7 ms and a slow-phase of 62.3 ms. It is worth noting that when our PSI-CC was subjected to a time-resolved optical specroscopy, the photooxidation kinetics could also be fitted to a pair of time constants with half-times of 75 ms and 3 s, respectively (SI Appendix, Fig. S9).

It is also noteworthy that a variety of GBN-based systems are available, which possess substantial practical and environmental importance (39–44). Among these are electrobiotechnological platforms based on Shewanella oneidensis, a metal-reducing facultatively anaerobic bacterium that forms conductive nanowire filaments, donating electrons which enhance microbial pollutant removal (41). Recently, an S. oneidensis- generated GBN has been developed (45) in which the endogenous oxidative stress-resistant extracellular polymer, acting as an electron shuttling medium, was modified to surround nano zero-valent iron, capable of transferring electrons to an outer membrane-bound cytochrome c. The subsequent extracellular electron transfer serves to facilitate a variety of metal-coupled anaerobic pollutant remediation processes (46).

Our study has successfully demonstrated the viability of the assembled GBN, including both photosynthetic and respiratory enzymes in an extracellular and membrane-free nanoscale arrangement. JTS and electrochemical measurements provide complementary information about the kinetics of O₂ consumption, supporting our conclusions. A kinetic model based on the understanding of PSI as a three-level quantum system provides an appropriate framework for data interpretation. This approach has allowed us to single out the role of Förster resonance (38) among the photosynthetic core complexes. Further investigations should be directed toward imaging macromolecular-level details of the GBN. Additional kinetic studies could also be performed to test the effects of specific enzyme inhibitors. The proposed electronic cascade constitutes an example of a multienzyme redox electron transport chain and can be extended to other enzymes with suitable reduction potentials. In the future, various applications of practical significance in industrial photoswitchable enzymatic processes can be envisioned.

Materials and Methods

Experimental Design.

PSI CC trimers with a His₆-tag on the C-terminus of the PsaJ protein (stromal side) were isolated from cultures of Synechococcus sp. PCC 7002 ΔPsaJ/PsaJ-His variant using a standard protocol (19). The structural integrity of the PSI CCs was confirmed by UV-visible spectroscopy (SI Appendix, Fig. S10), while the integrity of in vitro photosynthetic activity was assessed by examining diagnostic kinetic parameters measured by time-resolved optical spectroscopy (SI Appendix, Fig. S9) (37). CcO, His₆-tagged on the C-terminus of subunit I, was isolated from R. sphaeroides following a previous report (2). GO was synthesized using a modified Hummers’ method (26) and functionalized with nickel nitriloacetic coordination sites, resulting in GO-NiNTA as previously described (2, 30). The His-tagged PSI and CcO enzyme proteins were immobilized onto the surface of GO-NiNTA, as described below. The resulting GBN was characterized by independently probing the oxidation and re-reduction of PSI by JTS and the O₂ consumption activity of CcO using an electrochemical method. Kinetic models of increasing complexity were proposed to interpret the JTS results and evaluate the manner in which PSI is directly responsible for triggering the functional response of CcO, producing an electron transport chain on the surface of GO.

Generation of the His₆-Tagged PsaJ Variant.

PsaJ is a small (4.4 kDa), hydrophobic subunit that spans the membrane once, with its C-terminus located at the luminal side. The His₆-tagged PsaJ variant was constructed by first deleting the native copy of the psaJ gene from the chromosome of Synechococcus sp. PCC 7002 using an erythromycin (Em) resistance cassette. A sequence of linear DNA containing the flanking regions (approximately 500 bp) before and after the psaJ gene, as well as the Em resistance cassette in between, was transformed with wild-type Synechococcus sp. PCC 7002. The Em resistance cassette replaces the native psaJ gene through homologous recombination of the flanking regions. Then, a C-terminally his-tagged PsaJ sequence was introduced on the pAQ1-Ex-cpc plasmid, which contains gentamicin resistance. This pAQ1-Ex-cpc plasmid has homology with the native pAQ1 plasmid that is present in Synechococcus sp. PCC 7002 at high copy numbers. Genes may be cloned in this plasmid under the control of the phycobilisome promoter (cpc), and this fragment, along with the promoter and gentamicin cassette, are flanked by regions of DNA that promote homologous recombination into a neutral site sequence in the resident pAQ1. This allows for expression of His-tagged psaJ directly from the pAQ1 plasmid in a background devoid of wild-type PsaJ.

Maintenance and Growth of the His₆-Tagged PsaJ Variant.

The His₆-tagged PsaJ variant was kept on A+ plates supplemented with vitamin B12, appropriate antibiotics, and sodium thiosulfate (an effective mold inhibitor). The culture was restreaked at least every 14 d and maintained at 30C in normal light conditions (50 to 100 μmol photons m⁻² s⁻¹). The variant was grown on A+ medium supplemented with B12 vitamin and appropriate antibiotics, 3% CO₂ (air balanced) bubbling, normal light conditions (100 μmol photons m⁻² s⁻¹) at 37 °C. A trimeric PSI band was selected after two sucrose gradient ultracentrifugations, as previously noted (19).

Self-Assembly of Enzyme Bioconjugates and Galvanic Biohybrid Nanosystem.

The initial concentrations of PSI CCs and CcO were measured by standard methods (2, 19). PSI was quantified by measuring the visible absorbance at 663 nm (19). The concentration of the purified CcO was determined by measuring the visible absorbance of the fully reduced enzyme at 606 and 640 nm (2). A GO-NiNTA suspension was prepared in Tris buffer (e.g., pH 7.4) by diluting 130 μL of 25% m/m GO-NiNTA aqueous suspension with 870 μL of Tris buffer solution. GO-NiNTA-CcO bioconjugates were self-assembled by mixing 100 μL of CcO sample (0.300 mg/mL, 129.9 kDa) (2) with 130 μL of the GO-NiNTA Tris suspension for 2 h. GO-NiNTA-PSI bioconjugates were formed by mixing 150 μL PSI (0.179 mg/mL, 360 kDa/monomer) with 40 μL of GO-NiNTA Tris suspension. It should be noted that our CcO preparations consisted of a three-subunit complex from which CoxIV is absent, since they were derived from a plasmid capable of encoding only the ctaCDE structural genes with the appropriately attached His-tag (2). The GBN was prepared by mixing 150 μL of PSI with 40 μL of GO-NiNTA-CcO bioconjugate suspension for 2 h. All suspensions were washed with Tris buffer three times with following resuspension to a final volume of 1 mL. Analysis by UV-visible spectroscopy of the washes indicated complete immobilization of both CcO and PSI CCs onto GO-NiNTA. The pH was adjusted to 7.4, 8.0, and 9.0.

Joliot-Type Spectroscopy.

JTS measurements were conducted using a JTS-10 instrument (BioLogic USA) to probe the photoelectronic behavior of PSI in PSI CCs, GO-NiNTA-PSI, and PSI-(GO-NiNTA)-CcO in MES buffer suspensions at pH 6.5. Ficoll (20% v/v) was used to stabilize the bioconjugates and the GBN. Measurements were conducted using a polychromatic probing light (700 to 740 nm) filtered at $705\pm 3$ nm and by controlling the exposure of sample and reference to green (522 nm) and orange (639 nm) actinic LED lights. The irradiance of the green light was set at 50 $\mathrm{\mu }\mathrm{E}/\left({\mathrm{m}}^2\mathrm{s}\right)$ while that of the orange light was chosen among discrete values comprised between 50 and 4,700 $\mathrm{\mu }\mathrm{E}/\left({\mathrm{m}}^2\mathrm{s}\right)$. The JTS measurements were repeated at least three times per sample. Region I (light phase) is fitted by using a double exponential function:

$$f(t)=y_0+A_1·\exp \left(-\frac{t}{{\tau }_1}\right)+A_2·\exp \left(-\frac{t}{{\tau }_2}\right).$$ [11]

Region II (dark phase) instead requires a simple exponential function. Data were interpreted by considering 2-level and 3-level system models of PSI.

Oxygen Consumption.

A volume of 150 μL of PSI, GO-NiNTA-PSI, or PSI-(GO-NiNTA)-CcO suspension was diluted to 1 mL in Tris buffer and inserted into a Clark-type electrode cell (Hansatech). The temperature was maintained at 37 ± 0.1 °C. Five μL of a 0.1 M MV and 10 μL of 1.0 mM horse-heart Cyt c were added to each suspension. After establishing equilibrium, the reaction was left to proceed for 5 min. Samples were subsequently exposed to light (Schott KL 2500 LCD) at 3,000 K for 2 min. When PSI was completely reduced, 5,10, 15, and 20 μL of 0.1 M DTT was added following the same procedure, with MV and Cyt c added at the beginning. Every sample was measured by adding the reductant DTT using a syringe through a rubber silicon septum connected to the Clark electrode. Measurements were repeated at least three times per sample.

Flow Cytometry Measurements.

Flow cytometry measurements were performed by means of a model TN Gallios Flow Cytometer, equipped with 10 colors and 3 lasers (red at 638 nm, blue at 488 nm, and violet at 405 nm). Data processing required the Kaluza software version 1.5a (Beckman Coulter) in combination with the Igor Pro 8 software. Setting parameters and experimental conditions were chosen in agreement with a previous study (2). Each flow cytometry map was based on 100,000 events per sample.

Statistical Analysis.

Curve fitting of the JTS data was performed with linear, polynomial, and nonlinear regression using the software Igor Pro. The Gaussian variance formula was used to determine the error propagation from experimental fitting parameters to calculated parameters (47). For O₂ consumption measurements, a range of variability error of 7% was assumed for O₂ consumption rates on data acquired on the same day. Flow cytometry data were statistically analyzed based on a dependent t test of paired samples.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Study data are included in the article and/or SI Appendix.