Control of protein activity by photoinduced spin polarized charge reorganization

Significance The role of well-placed charges within proteins in mediating biological functions, from protein-protein association to enzyme kinetics, is well documented. Here, we go beyond this static picture and show that charge motions can exert significant effects on protein function. Injecting charge from a photosensitizer, we demonstrate a threefold decrease in enzymatic activity and a twofold increase of antibody-antigen binding. These effects depend on the specific position of the photosensitizer on the protein. Our results point to charge reorganization as a form of allostery that complements known allosteric mechanisms such as conformational changes and dynamics.


Introduction
Biomolecules within the living cell are subject to extensive electrical fields, particularly next to membranes(1). Indeed, a role for bioelectricity has been well established at the organismal level (2). While the importance of electrostatics in protein functions such as protein-protein association and enzymatic activity has been well documented (3), very little is known on how biomolecules respond to external electric fields, or in other words, what may be the potential contribution of polarizability to protein function. Multiple protein activities involve electrostatic effects (3). For example, it is recognized that the association kinetics of proteins can be accelerated by charged residues positioned close to the interaction sites on their surfaces (4). Recent work on enzyme catalysis has given rise to a picture of pre-organized charges at catalytic sites, directly influencing substrate molecules and lowering enzymatic reaction barriers in this manner (3,5,6). These mechanisms for charge influence on protein function invoke essentially fixed charge distributions, and do not take into account the potential role of charge regulation and reorganization due to external electric fields (7). Yet, it is important to appreciate that any interaction between two proteins, as well as between a protein and other molecular species, involves the formation of an effective electric field that results from the difference in electrochemical potentials of the two interacting bodies.  (8,9) and experimental work from our labs (10,11) have indeed hinted at a role for charge reorganization as an allosteric signal in proteins. Here we decisively establish this role by studying the effect of phototriggered charge injection on both protein-protein association kinetics and enzyme kinetics.
We find a rich spectrum of responses that depends on the position of the photoexcited group as well as on the spin polarization of the rearranging charges. The spin dependence is likely associated with the chiral induced spin selectivity effect (12).

Modulating protein-protein association
We site-specifically labeled phosphoglycerate kinase (PGK), a 415-residue protein, with the photosensitizer ([Ru(2,2′-bipyridine)2(5-iodoacetamido 1,10-phenanthroline)] 2+ (Ru) (13). In particular, we created the mutant C97S/Q9C, in which the native cysteine at position 97 was changed to a serine, and a cysteine residue was inserted at position 9 ( Fig. 1). Ru can inject either an electron or a hole into the protein, potentially modulating the charge distribution (i.e., the electric polarization) within the protein. We first studied the binding of an anti-His antibody to a polyhistidine tag at the C terminus of PGK (Fig 2A). The Ru-PGK construct was attached to a gold surface to facilitate uniform illumination and readout of antibody-antigen interaction. The antibody molecules were labeled with the dye Alexa 647, which allowed counting individual events of protein-protein association at the surface at different times, following the addition of the antibody to the solution. The experiment was performed either under illumination with a linearly polarized (LP) 470 nm laser or in the dark (Fig 2B-E). The kinetics traces in Fig. 2F demonstrate that under illumination association was significantly enhanced at early times. In particular, at 2 s, illumination increased the association rate by a factor of 2.25±0.05. At longer times the difference between the two sets decreased, reaching a similar value at 8 s, due to saturation of the binding of antibody molecules to the surface. In the rest of the article, we will therefore report only rate differences at 2 s. The experiment was repeated with PGK molecules that were not labeled with Ru, and no effect of illumination was observed (Fig. 2G). We further repeated the same experiment on glass to rule out any potential contribution of the gold surface, and the results were similar (Fig. 2H).
As it is known that electron transport through a protein may be spin selective, due to the chirality of the protein and its secondary structure (12,14,15), we asked whether illumination with circularly polarized light can modulate the observed effect. The experiment on the gold surface was therefore repeated with either right or left circularly polarized light. Circularly polarized light is likely to generate excitations with one spin state (16), so that the injected charge into the protein (either positive or negative) would be spin polarized. Remarkably, the enhancement of the association kinetics was observed only with left circularly polarized (LCP) light, and not with right circularly polarized (RCP) light (Fig. 2I). These results indicate, within the experimental uncertainty, that the whole photoinduced effect an outcome of essentially a single spin polarization, suggesting in turn that the charge reorganization within the protein is spin selective. Only a minor illumination effect was observed when Ru was moved to residue 290. In G-J molecules were counted 2 s following the initiation of the reaction. At least 9 regions were counted in each sample. Experiments were repeated three times (see Supporting Table 1 for all values). Error bars represents standard errors of mean.
To test the position dependence of the charge reorganization effect on association kinetics, the Ru complex was moved to residue 290, using the mutant C97S/S290C (Fig. 1). At this position, the photosensitizer is much further away from the His-tag at the C terminus compared to the previous position; the distance from residue 290 to the C terminus, residue 415, is 55 Å, based on the crystal structure 3PGK, while from residue 9 it is only ~10 Å. Repeating the same experiment, it was found that illumination (either LP, LCP or RCP) had only a minor effect on the association reaction ( Fig.   2J), pointing to a significant position dependence of the effect.

Controlling enzymatic activity
We then turned to measure the effect of photosensitization on the catalytic reaction of PGK. The enzyme catalyzes the transfer of a phosphate group from ATP to 3-phosphoglycerate (3-PG), producing ADP and 1,3-bisphosphoglycerate (1,3-BPG) ( Figure 3A). To observe a robust reaction on a surface, the His-tag at the C terminus of PGK was used to attach protein molecules to a supported lipid bilayer formed on a glass substrate (Fig. 3A). The turnover of surface-bound enzyme molecules was measured using a coupled assay, and the kinetics were gauged through a change in NADH absorbance (17). Based on the slopes of the kinetics curves in Fig. 3B-E (Supporting Table 2), and assuming a surface density of PGK molecules of ~5⋅10 11 /cm 2 (somewhat lower than expected for a close-packed layer of the protein), we calculated a turnover rate of ~200 s -1 for Q9C PGK and S290C PGK in the dark. This turnover rate is quite close to the value measured in solution with C97S PGK (226.9±7.3 s -1 ).
Remarkably, with Ru at position 290, the enzymatic rate decreased under illumination by a factor of 3.3±0.2 (Fig. 3B). As above, this rate reduction was induced by either LP or LCP illumination, but not under RCP illumination. The effect could be observed in a single experiment: when light was turned off, the slope of the kinetics curve increased ( Figure 3C). In the absence of Ru, no effect of illumination was observed ( Fig. 3C and Supporting Fig. 1A). When the Ru complex was moved to position 9, an illumination effect was still observed, but it was significantly reduced to a factor of only 1.8±0.1 ( Fig 3D); as above, the slope increased when light was turned off, and no illumination effect was observed in the absence of Ru ( Fig. 3E and Supporting Fig. 1B).

Discussion
When a protein interacts with a charged molecule/protein, charge rearrangement occurs within the protein, which may affect the interaction between the protein and the other species. The extent of charge rearrangement depends on the polarizability of the protein, and therefore polarizability may affect both interaction between proteins and enzymatic activity. Upon excitation of a photosensitizer, charge can be injected into the protein, hence affecting its polarizability, thereby modulating the effect discussed above. Charge injection can involve either an electron or a hole, and might potentially be only partial, leading in either way to an effect on the charge distribution within the protein. However, since the protein is chiral, any charge injection would be spin dependent due to the chiral induced spin selective (CISS) effect, as shown by Naaman and coworkers in multiple studies (12). Exciting the dye with circularly polarized light causes one spin to be preferentially excited. Due to the CISS effect, one specific spin can be injected more A. Schematic of the experimental setup to study the effect of illumination on enzymatic kinetics of Ru-modified PGK. PGK molecules were attached to a lipid layer supported on a glass surface through their His-tags. The enzymatic reaction of PGK is depicted in the cartoon. Enzymatic activity was measured at 25 °C using a coupled assay (see Methods) and the absorbance of NADH at 340 nm was monitored. B. A strong reduction in enzyme kinetics was observed upon either LP or LCP illumination of PGK modified with Ru at position 290, but not under RCP illumination, as compared to no light (NL). C. The slope of reaction progression changed when the initial LP illumination was stopped after 5 min (full symbols). In the absence of Ru, no effect of light was observed (empty symbols). D. The effect of light was smaller when Ru was at position 9. As in B, the effect was observed under LP or LCP illumination, but not under RCP illumination. E. As in C, but with Ru at position 9. Only the linear regions of the activity curves were fitted. Experiments were repeated three times, and this figure shows only one set. For values obtained from all experimental sets, see Supporting Table 2. efficiently into the protein. Therefore, one circular polarization is more effective than the other.
Excitation with the 'correct' circular polarization would lead to charge injection into the protein and to a charge-separated species that would typically have a much longer lifetime than the usual excitation lifetime of the molecule. On the other hand, excitation of the 'wrong' circular polarization would not lead to charge injection, and the excited state would relax quickly, either radiatively or non-radiatively.
Our results indeed indicate a significant effect of charge injection from the photosensitizer Ru into the protein both on association with an antibody and on its enzymatic reaction. Notably, the effects we measure depend on the polarization of light and in particular, respond to only one circular polarization. Our findings strongly support the notion that charge reorganization is involved, as it has been established that the motion of charge through a chiral potential is spin selective and is affected by the protein secondary structure (15). This spin dependence is due to the chiral-induced spin selectivity effect (12). Importantly, it has been shown previously that spin polarization enables long range charge transfer through chiral systems (18).
Specifically for our systems, we can only speculate on the exact effect of charge injection and in which direction charge is transferred. In the case of the antibody-protein interaction, since the antibody is directed to the His-tag, we observe the C-terminal region of the protein (where the Histag is connected) and find that it is in general more negative. Clearly, the protein-protein association reaction would benefit from this region being even more negative, meaning that an electron would likely be injected from the photosensitizer. In the case of the enzymatic reaction, charge reorganization may affect substrate binding by making the active site more negatively charged and changing its interaction with the negatively charged substrate molecules. Additionally, charge reorganization may affect the catalytic mechanism itself. We cannot be more specific about this aspect at this moment of time. In any case, since charge reorganization is found to be sensitive to circularly polarized light, it is likely that α-helical structures of the protein are involved, as αhelices have been implicated as good spin filters (15).
The photoinduced charge injection effect we observe here depends on the distance from the active site involved, rather than on the sequence separation. Thus, for protein-protein association at the C terminus of PGK, Ru at position 9 had a strong effect, while Ru at position 290 had no effect. A similar picture arose also for the enzymatic activity of PGK, though now Ru at position 290 (close to the ATP binding site) showed double the effect of Ru at position 9.
The findings here, combined with previous studies (10,11), point to a new role of charge reorganization, or of polarizability, in modulating protein activities. Surprisingly, not much is known about the involvement of polarizability in protein function, though the development of polarizable force fields for molecular dynamics simulations of biomolecules in recent years may change this situation (19). The role of charged protein residues in enzymatic catalysis has been discussed extensively by Warshel and coworkers (3), who emphasized the contribution of charges that are pre-organized to reduce the free energy of the transition state. Recent work from the Boxer lab has experimentally demonstrated that charges at the active site of the enzyme ketosteroid isomerase exert an electric field that contributes significantly to the catalytic effect (20). However, these charges are considered to be static. We suggest instead that the electric field at the active site of an enzyme may be modulated through the binding of charged groups at distant sites or by the presence of bioelectric fields.
Indeed, our current results indicate that this is the case. The excitation of the Ru moiety likely leads to a propagation of a polarization signal through the protein, reaching and affecting the active site.
A significant effect is demonstrated here on both the binding of an antibody to the His-tag of PGK and, most remarkably, on its enzymatic activity. The effect on the activity of PGK might be due either to modulation of the binding of substrates or to an effect on the catalytic step itself-this remains to be determined. In any case, these findings point to a so-far unappreciated role of electric fields in the regulation of biological activity at the molecular level. Within the cellular environment, electric fields abound particularly near membranes, and it is possible that membrane proteins and also proteins that interact with membranes are susceptible to control mediated by charge reorganization. This discovery also suggests a novel method for generating photo-controlled enzymes and sensors, based on photoexcitation of an attached group. Currently, all proposed methods to photo-control bioactivity have relied on various conformational changes induced by photoexcitation (21,22). Photo-controlling bioactivity through charge injection might be easier to implement. Future work will allow us to optimize the location of the photosensitizer and enhance the effect of light on activity even further and will teach us more about pathways of charge rearrangement in relation to protein function. For that purpose, we plan to identify biological systems that might be particularly susceptible to this type of activity regulation in proteins.

Methods
Protein expression and purification. Yeast phosphoglycerate kinase (PGK) DNA was cloned into a pET28b vector, fused to a C-terminal 6xHis tag. For site-specific labeling of PGK, the natural cysteine (C97) was replaced by a serine. A single cysteine residue was introduced using sitedirected mutagenesis, resulting in either a Q9C or a S290C PGK mutant.
Single-cysteine PGK plasmids were transformed into E. Coli BL21 pLysS (DE3) cells (Invitrogen), which were grown in LB media at 37 °C up to an optical density of 0.8-1. Rinsing with PBS removed the unattached protein molecules.
Labeling of anti-His tag antibody. In order to study the antigen-antibody reaction kinetics by observing the fluorescence of attached antibody molecules, anti-His tag antibody molecules were tagged with the dye Alexa Fluor® 647 NHS Ester (Succinimidyl Ester, ThermoFisher SCIENTIFIC, Catalog number: A20006) using the same procedure as followed in our previous paper. (11) In brief, unlabeled antibody molecules in PBS buffer were reacted with the NHS ester of the dye in a 1:1.5 ratio in presence of 0.1 M sodium bicarbonate buffer for 1 h at room temperature in the dark. Micro Bio-Spin columns with Bio-Gel P-30 (Bio-Rad) were used to remove the unlabeled dye molecules. We verified that the labeled protein did not show any optical activity at the wavelength of absorption of the Ru group using circular dichroism spectroscopy (Supporting Fig. 2).

Interaction between His-tagged PGK and anti-His antibodies with and without illumination
To study the antibody-antigen reaction kinetics, His-PGK-Ru modified gold surfaces were To test the potential contribution of the gold surface on the antigen-antibody reaction, the above experiment was repeated using a glass surface coated with His-PGK-Ru with and without illumination.

Microscopy measurements & data analysis.
Fluorescence imaging of the samples following reaction with antibody molecules was carried out following the same procedure used in our previous work.(11) A home-built total internal reflection fluorescence microscope (TIRFM) was used for the imaging. In each experiment, 10 different TIRFM movies were recorded on 10 different regions of 101 X 101 pixels (6.73μm X 6.73 μm). On each region, 100 ms frames were recorded until all molecules in the designated area were photo-bleached. TIRFM movies were analyzed using custom-written Matlab (MathWorks) routines. Individual spots, corresponding to individual antigen-antibody complexes, were identified in the first frame of a movie using a combination of thresholding and center-of-mass (CM) analysis as described previously (23). The intensity of the center of mass of each individual spot as a function of time was plotted, and change-point analysis was performed on to identify photobleaching steps and hence the number of emitters in each spot.
Some examples are shown in Figure S3 of ref (11). introduced at the appropriate angle just before the sample chamber. We verified that at the sample light was circularly polarized to within ~10% by rotating a polarizer and measuring the power. The laser intensity (~5 mW/cm 2 ) at the sample was kept constant for linear as well as circular polarization by tuning the laser power at the source. While the relatively low laser intensity implies a low efficiency of excitation, potentially long charge recombination times are likely to lead to a significant fraction of charge-separated protein molecules.

Supporting Information
Supporting 3. All slopes are given in units of absorbance change per minute. Note that slopes vary between different samples due to differences in the surface densities of the proteins; however, the ratios are similar within experimental error.