Highly oriented photosynthetic reaction centers generate a proton gradient in synthetic protocells

Reconstitution of RC in Giant Unilamellar Vesicle Membrane by Means of the Droplet Transfer Method.

Giant unilamellar vesicles (GUVs) (47) were prepared using the droplet transfer method (48) (Fig. 1) because we envisaged that this method could be suitable for reconstituting transmembrane proteins with a high degree of physiological orientation. Purified RC from R. sphaeroides was first obtained by a well-established procedure requiring the detergent lauryldimethylamine N-oxide (LDAO) to extract the protein from the photosynthetic membrane and to solubilize it in aqueous solutions (49). A homogeneous micellar solution was obtained with fully photoactive RCs surrounded by a toroid of LDAO molecules that shield the LM core from the aqueous environment (50). To prepare RC@GUVs, the RC micelle solution was emulsified in mineral oil containing a mixture of phosphatidylcholine and phosphatidylglycerol (POPC:POPG, 9:1). This emulsion was then layered on the aqueous solution, generating a biphasic system, and RC@GUVs were obtained after centrifugation (Fig. 1C).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Preparation of GUVs by the droplet transfer method (48). (A) Water in oil (w/o) droplets, prepared by the emulsification of an aqueous solution (I-solution) in a lipid-rich oil phase, are transferred to an aqueous solution (O- solution) by centrifugation. (B) For preparing RC@GUVs, a detergent-stabilized RC solution (RC-micelles) is emulsified in oil, giving the w/o droplets. Owing to asymmetric RC-micelle structure a preferential “physiological” RC orientation is expected, namely, with the H subunit (in orange) facing toward the aqueous core of the droplets (the cytoplasm-like GUV lumen), and the photoactive dimer (SI Appendix, Fig. S2 A and B) facing the GUV exteriors (in white). (C) RC@GUVs (POPC:POPG, 9:1) as imaged by confocal microscopy. Red-fluorescent AE-RC was reconstituted in calcein-containing GUVs. (C1) Green fluorescence channel (calcein). (C2) Red fluorescence channel (AE-RC). (C3) Bright field. (C4) Overlay of the C1, C2, and C3 channels.

Considering the RC reconstruction mechanism in vesicle membrane, it is reasonable to assume that micelles when dispersed in w/o will deliver their protein cargo at the droplet w/o interface, mainly driven by hydrophobic interactions. Moreover, because RCs present asymmetric distribution of hydrophilic and hydrophobic regions, protein-containing micelles will have a preferential orientation while approaching, and interacting with, the lipid monolayer of the w/o droplet, because the large hydrophilic H subunit prefers the aqueous phase (SI Appendix, Fig. S2 A and B). It is expected that the chemical vectoriality of both RC-micelles and lipid monolayer will favor only one of the possible protein orientations in the w/o droplets before and during their transfer to the aqueous phase, so that a population of highly oriented RCs in the GUVs membrane should be obtained.

RC@GUVs prepared in such a way have an average diameter of 20 ± 10 μm (statistical analysis performed on a population of 150 GUVs; SI Appendix, Fig. S3) and are morphologically stable for at least 2 d when stored in the dark at room temperature. Quantitative image analysis shows calcein does not leak out from GUVs after 2 d from the preparation (SI Appendix, Fig. S4), proving also that traces of detergent, present as a consequence of the RC encapsulation, do not significantly affect the membrane stability.

The concentration of lipids and photoactive RCs, collected in a 100-μL volume of the thus-prepared GUVs suspension, was determined spectroscopically (SI Appendix, section S2g), resulting 440 μM and 0.2 μM, respectively, and hence a protein/lipid molar ratio of 1/2,200 was reached. RC@GUVs are characterized by a quite high RC density (∼1,200 RC molecules per μm²), corresponding to roughly one-third of the RC average density in the intracytoplasmic membranes of photosynthetic bacteria (51, 52). The collected GUVs were washed twice before further use, to remove any external fragments of RCs.

Fluorescently labeled RCs were used to monitor the spatial distribution of the protein in GUVs. As a fluorophore we selected a suitable fluorescent dye belonging to the aryleneethynylenes class, because these molecules emit (53) light efficiently and can be easily functionalized to be covalently conjugated to biomolecules (SI Appendix, section S2a). In this work, we used the 7-AE fluorophore (AE) (SI Appendix, Fig. S6) (54), which absorbs light at 445 nm and emits it in the red region at 602 nm (SI Appendix, Fig. S7). The AE is covalently linked through an amide bond to the protein lysine residues by exploiting the succinimidyl N-hydroxysuccinimidyl ester derivative AE-NHS as an activated compound toward the reaction with amine groups in the lysines (SI Appendix, Fig. S8) (54). The AE-RC conjugate can be easily visualized by confocal microscopy, allowing its localization in AE-RC@GUVs. Fig. 2C shows images obtained by confocal laser scanning microscopy where vesicles display a uniform red fluorescent ring overlapping with the vesicle membrane, demonstrating a homogeneous incorporation of RC in the lipid bilayer of all GUVs.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Charge recombination of RCs reconstituted in giant vesicles after a saturating light flash. The points represent the experimental data, and the lines are the biexponential best fit curves. Data refer to charge recombination in the absence (blue points) and in the presence (dark green points) of excess of reducing agent (cyt²⁺). In a control experiment (red points), a full recovery of RC photoactivity is measured after the addition of an electron acceptor, the dQ, and the exhaustion of cyt²⁺. Note that values in the y axis represent the absolute values of ΔA₈₆₅. (Inset) Theoretical charge recombination curves in the absence (blue) and in the presence (green) of cyt²⁺, corresponding to different RC orientation (100, 50, and 0% of physiological orientation). The histogram represents the initial amplitude of the curves ΔA₈₆₅(0). The green bar marked with the asterisk refers to the experimental trace reported in the main plot.

RC Is Active and Highly Oriented.

The photoactivity of reconstituted RC can be assessed by inducing the formation of an electron-hole couple by a short light flash and monitoring the time of the charge recombination reaction by following the absorbance at 865 nm (the detailed mechanism is reported in SI Appendix, section S3b). Fig. 2 shows the time decay of the charge-separated state induced by light flash: blue dots are the recovery of the dimer signal from the excited state after a saturating flash in RCs reconstituted in giant vesicles. From the initial absorbance, ΔA₈₆₅(0) = 2.21 ± 0.03 milliabsorbance (mAU), the actual RC amount in the RC@GUVs preparation can be determined, which corresponds to ∼10% of the protein initially loaded in the w/o droplets. The biexponential fitting of the recorded trace (blue line) reveals that the fast charge-recombination from D⁺Q_A^-• (A_f) accounts for about 71% of the overall signal, whereas the slow recombination from D⁺Q_AQ_B^-• (A_s) contributes in a minor way (29%), showing that under these experimental conditions the Q_B-site is only occupied partially (Table 1, first row). The orientation of the RC population in GUV membrane can be assessed by using the water-soluble cytochrome c₂ (30), the physiological electron donor to the photooxidized dimer. In fact, both the reduced (cyt²⁺) and the oxidized (cyt³⁺) forms of the cytochrome are unable to cross the membrane. Therefore, the reduced cyt²⁺ added in excess externally to preformed RC@GUVs reacts only with the oxidized dimers exposed to the outer solution. The electron donation from the reduced cytochrome to the oxidized dimer, D⁺ + cyt²⁺→ D + cyt³⁺, occurs very fast on the microsecond time scale, preventing the charge recombination reaction. The dimers reduced by the cyt²⁺ will not contribute to the absorbance recorded at 865 nm. On average, if the RCs reconstituted in the GUVs dispose across the lipid bilayer in random orientation, only half of the dimers face toward the bulk solution. Under this condition, a saturating flash of light will generate the full population of D⁺, but the signal will appear halved because the dimers oriented toward the bulk are rereduced on a very fast time scale by cyt²⁺. The other extreme possibilities, that is, fully oriented RCs with the dimer facing the GUVs core, or fully oriented with the dimer facing the external aqueous solution, will give the full signal ΔA₈₆₅(0) in the presence of cyt²⁺ or the complete absence of signal, respectively (Fig. 2, Inset). The actual ratio of the D⁺ absorbance change in the presence [ΔA₈₆₅(0)_cyt] and in the absence [ΔA₈₆₅(0)] of cyt²⁺ gives the fraction of RCs oriented in the bilayer with the dimer exposed to the outer solution. Hence, fully oriented RCs will have the ratio ΔA₈₆₅(0)_cyt/ΔA₈₆₅(0) value equal to 0 when all RCs are oriented with the dimer outward. The ratio assumes a value of 1 when all RCs are oriented with the dimer facing the GUV water core. All other intermediate possibilities will have a ratio value ranging from 0 to 1.

Fig. 2 (green points) shows the recovery of D in RC@GUVs in the presence of externally added cyt²⁺. A small signal ΔA₈₆₅(0)_cyt 0.21 ± 0.01 mAU is recorded, accounting for 9.5 ± 0.6% of the ΔA₈₆₅(0) value recorded in the absence of cyt²⁺ (Table 1, second row). This clearly indicates that the vast majority of photoactive proteins in RC@GUVs prepared by the droplet transfer method, 90 ± 1%, are uniformly oriented and expose the dimer to the outer aqueous phase. Notably, this result also demonstrates that the large majority of vesicles prepared by the droplet transfer method are unilamellar, as reported elsewhere (55). In fact, if RC were embedded in any internal lipid structure, as in the internal membranes of multilamellar vesicles, it would not react with cyt²⁺ and therefore it would count as oppositely oriented.

As a further experimental test to check RC functionality, a suitable amount of decylubiquinone (dQ) was then added to oxidize all cyt²⁺ molecules, as a result of the RC photocycle (SI Appendix, Fig. S2C). In fact, dQ is a ubiquinone analogous that binds to the RC Q_B-site and accepts electrons as well (56). When added to RC@GUVs suspension, it is expected that dQ will insert into the lipid membrane and diffuse and bind to RC Q_B-site. RC@GUVs were illuminated with repeated light pulses until the exhaustion of cyt²⁺, which is converted to cyt³⁺, whereas dQ is reduced to decylhydroubiquinone dQH₂. Thus, having removed all of the exogenous electron donors, the charge recombination signal reappeared. As shown in Fig. 2 (red points) and Table 1 (third row), the measured ΔA₀ value, in the presence of dQ (1.94 ± 0.03 mAU), is close to the original 2.21 ± 0.03 mAU value, demonstrating unequivocally the biochemical activity and the high orientation of RCs in GUVs. As expected, the slow pathway for charge recombination (k_s = 1.86 ± 0.06 s⁻¹) now becomes more relevant (55%), due to the presence of dQ in the Q_B-site.

RC Converts Light Energy into a pH Gradient Across the GUV Membrane.

The spontaneously achieved high orientation of RCs in the bilayer of the GUVs having roughly 90% of the dimer facing the aqueous bulk and, consequently, ∼90% of the Q_B-site facing the vesicle lumen, can be exploited to efficiently build a light-driven pH gradient across the GUV membrane. Under continuous actinic illumination, and thanks to the electron-hole couple formation, the electrons will flow from the external donor (cyt²⁺) to the acceptor (dQ in the Q_B-site) that will uptake protons from the vesicle aqueous core to form the quinol dQH₂. Ultimately, this compound accumulates in the bilayer. The net result of the photocycle is an intravesicle alkalinization that can be revealed using the pH-sensitive probe pyranine. Pyranine-containing RC@GUVs, prepared with low buffer capacity, were hence added with an excess of dQ, a small amount of cyt²⁺, and an excess of ferrocyanide acting as secondary electron donor. Under continuous irradiation, the pathway shown in Fig. 3 is established. The net stoichiometry of the main process is the oxidation of two ferrocyanides to ferricyanide, and the reduction of dQ to dQH₂,2Fe(CN)64-+dQ+2H+→2hν2Fe(CN)63-+dQH2,removing two protons per dQ molecule from the vesicle lumen.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Scheme of RC@GUVs function under red-light illumination. (A) RC is reconstituted in a highly oriented manner (90%) in the membrane of GUVs, whose average diameter is 20 μm. The asterisk marks a nonphysiologically oriented RC. (B) Detail of the photochemical mechanism generating the pH gradient. hv, light energy.

Continuous red-light irradiation of pyranine-containing RC@GUVs (SI Appendix, Fig. S9) generated an increase of pyranine fluorescence over the whole vesicle population, shown in Fig. 4A for two different RC concentrations in the final suspension: 10 and 20 nM, respectively. As can be seen, by doubling the RC concentration in the preparation this amplifies the pH-increase rate by a factor of 2.11 ± 0.02.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Generation of a pH gradient by RC@GUVs. (A) Bulk fluorescence measurements of pyranine-containing RC@GUVs, which have been suspended in a fluorescence cuvette and illuminated from the top (SI Appendix, Fig. S9). Blue and red points refer to RC@GUVs with final RC concentration of 10 nM and 20 nM, respectively. Black lines represent the best-fit straight line, whose slopes are (2.64 ± 0.03) × 10⁻⁴ a.u.⋅min⁻¹ and (5.57 ± 0.03) × 10⁻⁴ a.u.⋅min⁻¹, respectively, for the blue and red datasets. (B) Confocal images of three pyranine-containing RC@GUVs illuminated with red light. (C) Quantitative image analysis reveals the increase of intravesicle pH in time (fluorescence values converted by means of a calibration; SI Appendix, section S2h). The best-fit slope is 0.061 ± 0.004 pH units per min. (D) Comparison between the experimentally observed pH increase in the aqueous core of giant vesicles: circles with error bars (as in C) and the theoretical outcomes (colored bands). a.u., arbitrary units.

The incipient proton gradient across the membrane of individual RC@GUVs was visualized by directly illuminating the vesicles in a microscopy slide well and imaging them with confocal microscopy. Fig. 4B reports a series of fluorescence micrographs referring to pyranine-containing RC@GUVs at increasing irradiation time. Pyranine fluorescence increases over time as expected and the fluorescence intensity obtained by image analysis was converted to pH units via a calibration curve (SI Appendix, Fig. S10). The internal pH linearly increases in time, as shown in Fig. 4C, with a slope of 0.061 pH unit per min, equivalent to one pH unit in 16.4 min. The average rate of pH increase was converted in the rate of translocated proton per RC by a physicochemical model that takes into account the GUV size, the RC density, and chemical composition of the vesicle lumen. According to some simplifying assumptions (detailed in SI Appendix, sections S5 c and e) the observed pH increase corresponds to a calculated RC turnover rate of about 1.0 ± 0.1 protons per min per protein, equivalent to 2.5 × 10⁶ protons per min per GUV. This value is our best estimate of RC function in GUVs in current experimental conditions and corresponds to about 10% of the maximal RC turnover rate calculated from the photon flux density delivered to the microscope well (SI Appendix, section S5e). Moreover, it contributes for a proton motive force of ca. 3.6 mV⋅min⁻¹ (ΔpH min⁻¹ × 59 mV). To test the robustness of the RC@GUV, the same sample was irradiated in a fluorimetric cuvette for 30 min immediately after the preparation and later on 24 h (stored in the dark at room temperature) by showing a comparable increase in the fluorescence of the encapsulated pyranine (SI Appendix, Fig. S11). These experiments show that GUVs retain the encapsulated pyranine and, at the same time, that the RC activity is largely (ca. 80%) maintained (SI Appendix, section S3c).

Moreover, based on the developed kinetic model (SI Appendix, section S5b), a statistical estimation of the pH change over time in the entire GUV population was obtained taking into account the vesicle polydispersity in size and in RC content. Because the GUV size distribution is experimentally known (SI Appendix, Fig. S13), by assuming a random distribution for the RC surface concentration it is possible to derive the bivariate density function P_Ves(D, C_RC) that estimates the probability P_Ves(D, C_RC) dD dC_RC to find a GUV with diameter in the [D, D + dD[ range and RC concentration in the [C_RC, C_RC + dC_RC[ interval (SI Appendix, Fig. S14). According to this model, the calculated displacements of the pH time course, weighted by the density probability P_Ves(D, C_RC) for the whole vesicle population, are reported as a green band (1–60 μm) in Fig. 4D. The shown large diversity in GUV performances depends much more on the vesicles size dispersion than on the random distribution of the RC proteins in the lipid membrane. In fact, the red band, which refers to vesicles with a restricted size range (15–30 μm), exhibits a more uniform behavior (Fig. 4B and SI Appendix, Fig. S16) that is much closer to those of the GUVs monitored experimentally. The comparison with the experimental data are good enough to validate the theoretical approach, although a statistical analysis on a larger vesicle population would be necessary. Because the number of RCs per GUV scales with the vesicle surface, whereas the variations of the proton concentration scales with the GUV volume, the model predicts small RC@GUVs generate a pH gradient faster than large ones (SI Appendix, Fig. S15). It is also possible to estimate theoretically the behavior of the smaller RC@GUVs with diameters in the <12.5-μm range that represent the 27% of the entire population and remove intravesicle protons from two to four times faster than the average (SI Appendix, Fig. S15), resulting in a theoretical pH increase rate up to 0.106 pH units per min. This suggests that RC@GUV with optimized size in the range between 10–15 μm would perform more efficiently and uniformly than those shown in this first report. Microfluidics fabrication could be used to produce almost completely monodispersed vesicle samples.