Nature is based on chiral molecules, namely molecules that appear in two forms, enantiomers, that are mirror images of each other. Interestingly, chiral biomolecules, like proteins and sugars appear in Nature mainly as one enantiomer. The origin of “homo chirality” in Nature, was – and is – discussed very intensively in the literature1. However, the focus of this work is related to a more fundamental question, i.e., why did Nature preserve chirality so persistently over the many millions years of evolution? In other words, does chirality per se, independent on the specific handedness, provide properties that serve an important role in Life? The ability of biological molecules to interact selectively with each other is at the heart of all biological processes and the basis of many pharmaceutical concepts. Two important properties – related to chirality – characterize interactions in nature, i.e., very strong enantioselectivity and the relatively fast rates of very complex processes, e.g., the rate of protein folding2-4 and the repair of damage in DNA5.
It has been suggested that an interaction term that involves the electrons’ spin can improve the enantioselectivity in reactions of chiral molecules, due to the symmetry constraints resulting from the dispersion-induced charge reorganization, which is accompanied by transient spin polarization (see Figure 1A)6. However, such a spin-related interaction term has never been measured before. Herein we use atomic force spectroscopy, and measure directly the enantioselective interaction between oligopeptides of different handedness. Simple modelling of the spin related enantiospecific interaction energies shows that the spin constraint imposes directionality in the interaction.
The enantioselectivity measured here does not correspond to any of the established biorecognition mechanisms related to structural properties, like the “lock and key” model7, induced fit and allosteric interactions8, and neither can it be explained through differential, enantiomer-specific, long-range electrostatic interactions9-13. Though existing computational methods account for enantioselectivity9-11, none is able to reproduce the high selectivity observed in the present experiments14-16. This suggests that the enantioselective spin interaction is not captured properly by the theories currently employed to describe (bio)molecules. While additional research will be needed to delineate the exact scope and pervasiveness of this effect, it can be expected to alleviate some of the notable discrepancies between observed vs. computed enantiospecificity, interaction energies, and reaction rates. Furthermore, the directionality emerging from our model may help to explain the high efficiency of many complex bio-processes, such as protein folding and enzymatic reactions, since it reduces significantly the phase space the systems have to explore.
When considering bio-related chemical processes, the distribution of charges in the reacting species is of major importance. Obviously, these distributions are a direct result of the spatial positions of the charged electrons and nuclei in the molecular systems. However, next to charge, electrons also have another property, their spin, which is their angular momentum and can have two orientation values. In organic molecules, the spin is typically not coupled significantly to the molecular frame and therefore the orientation of the spin relative to this frame is not defined. Consequently, in such a case, the electron’s spin direction does not affect the interaction between molecules, i.e., the exchange term of the dispersion interaction is spin-independent17. For chiral molecules however, this is not the case. In the last two decades it was established that when electrons are displaced in a chiral system, the rate of this displacement depends on their spin. This property was termed the Chiral Induced Spin Selectivity (CISS)18.
In chiral systems, the spin that results from the charge displacement is strongly coupled to the molecular frame, so that spins of one type are displaced faster than the other, depending on the handedness of the molecule and the direction of motion. Charge displacement occurs whenever two chiral molecules approach each other, resulting in the formation of induced electric dipoles (see Figure 1A). Consequently, the emergence of these dipoles implies that at each electric pole, there is at least a fraction of an unpaired electron, i.e., spin polarization is intrinsically associated with the electric dipole formation in chiral molecules. The concept of charge polarization accompanied by spin polarization was verified in experiments in which the interaction of chiral molecules with ferromagnetic substrates was probed19,20. Hence, when two chiral molecules interact, a spin-dependent interaction term emerges21, which is dependent upon the relative handedness of the molecules.
Here we present experiments in which the force between chiral oligomer attached to the tip of an atomic force spectrometer (AFM) and a monolayer made from oligomers that possess either the same or opposite handedness, or oligomers that are not chiral, are monitored. The energies associate with the enantiospecific interactions are larger by more than a factor five compared to the thermal energies at room temperature. This part is followed by model calculations that show the role of the spin exchange interaction and its effect on the interaction energies and their angle-dependent distribution
Previous work20 has shown that exchange interactions can be probed using modified atomic force spectroscopy (AFS). AFS is widely used to examine biological interactions and functions22 and is used for the study of binding and unbinding of proteins23,24. This study utilizes the above method to probe spin exchange interactions between helical (hence chiral) peptides and demonstrate the relation between enantiomer-selectivity and spin. A chemically modified standard gold coated AFM cantilever was used. The functionalized cantilever is used to generate a distance dependent force curve based on short-ranged spin exchange interaction. The gold AFM tip was functionalized with polyethylene glycol (PEG), 60 nm long, bound to a helical peptide, L-AHPA, where AHPA is Alpha Helix Polyalanine (AHPA) [HS-PEG-NH-AAAAKAAAAKAAAAKAAAAKAAAAKAAAAKAAAAK-COOH] (see Figure 1C, 1D). The AFM-tip was functionalized in such a way that the peptide's carboxyl group is facing the substrate (see methods section for details). The PEG acts as a spacer to reduce non-specific interactions and the whole system is immersed in ethanol to eliminate capillary forces. The measured samples consist of self-assembled monolayers of the same helical polypeptide adsorbed on a gold substrate. The adsorption process ensures a peptide alignment such that the carboxylic group is facing up, so that there are no covalent bonds possible between the oligopeptide peptide on the AFM tip and the one adsorbed on the substrate. Over a thousand curves of force vs. distance were measured, which were subsequently examined manually. Following previous studies20,25,26, only curves that showed a clear pulling event (see in supplementary Figure S1), were further analyzed as single molecule rapture events. A worm-like chain (WLC) model was then fitted on the specific interaction's pulling event and the pulling force was retrieved. The mean pulling force (MPF) was calculated by averaging over the selected pulling forces.
The force between the same and different enantiomers is presented in Figure 2. Right or left-handed helical oligopeptides (L-AHPA or D-AHPA respectively) were measured when adsorbed on the gold substrate. As control experiments, we measured a monolayer of a non-chiral - 12-mercaptododecanoic acid with a comparable length and an equivalent carboxylic head group facing up. The mean pulling force of the molecules is shown in Figure 2A. A force difference of between the interaction of the homochiral pair of oligopeptides (L-AHPA monolayer – L-AHPA) and the interaction of the heterochiral pair (D-AHPA monolayer – L-AHPA adsorbed AFM tip) is obtained. The interaction's energy is retrieved by integrating the force distance curve over the pulling event. An average energy difference of about is measured (Supplementary Figure S2).
A sum of two Gaussians is fitted to the force distributions presented in Figure 2B to highlight the concept of multiple forces that are acting between the molecules (quality of fit is shown in Figure S4). This is done to differentiate between the long-range interactions and the short-range spin exchange interactions20. We attribute the first Gaussian, at low forces, to non-specific interactions that are the same for all samples (i.e., Coulomb and dispersive forces). The second, at higher forces, is attributed to the spin-dependent exchange interaction and is different for homo and hetero chiral interactions, i.e., this force is stronger between same enantiomers (L-L) than opposite enantiomers (L-D). As a control experiment, we checked non chiral molecules as well. The MPF between the L-AHPA adsorbed on the tip and non-chiral control was lower than the force for both enantiomers. This effect will be discussed below. It is interesting to note that the MPF of the plain gold sample is lower than the MPF of the chiral L-L interaction. At first sight, this may be surprising since the interaction between a gold substrate and a carboxylic group has a covalent character, but it can be explained by the gold being coated with organic contamination27,28.
To verify that the difference in the binding of the two enantiomers results from the spin effect, a sample of gold substrate with adsorbed L-AHPA monolayer was tested (Figure 3 A1). Due to the sulfur-gold bond, the sample becomes paramagnetic29-32. By adding an external, out-of-plane, magnetic field, the spin injection into the chiral monolayer can be controlled. The same functionalized tip, as described above, was used and a constant applied magnetic field perpendicular to the surface was applied during the measurement (see Figure 3A1). The results are presented in Figure 3 A2 and Figure 3 A3. The MPF difference between up and down magnetizations is (Figure 3A2). These results support the notion that the interaction's strength is spin dependent. A clear difference is also seen in the force distributions histograms (Figure 3 A3).
The effect of the substrate magnetization on the force measured in the case of chiral molecules can also be a result of more efficient charge penetration from the substrate into the chiral molecule, when the injected charge has the preferred spin for the given handedness33. This charge transport from the substrate increases the spin density at the interaction between the molecule attached to the tip and the adsorbed molecule. To evaluate this effect, we investigated the interaction of a tip coated with achiral molecule with oligopeptide adsorbed on magnetized substrates (Supplementary Figure S5). In this case the difference in the force measured for the two directions of magnetization is within the noise range of the system. Thus, the main difference in force under opposite magnetic field is a result of spin-dependent exchange interactions.
We have shown so far that spin is affecting the bio-recognition pulling force. To relate the results to exchange interactions only, and to differentiate them from mechanical and structural related forces, we probed the interaction range. The spin-exchange interaction is characterized by overlap of wavefunctions and is therefore short-ranged. In a previous study, the decay length of the spin-dependent interaction was determined to be about 0.7 nm34. To probe the range of the effect observed in the present study, a layer of a non-chiral amino-acid (glycine), 0.4 nm long, was added to the respective helical peptide monolayers (L/D-AHPA), which are 2 nm high. The adsorption of the glycine was done following a protocol reported in the literature35,36, and the experimental layout and of the monolayer with glycine is presented in Figure 3 B1. The samples were measured with the same functionalized AFM cantilever as before. The results are presented in Figure 3 B2 and 3 B3. The force histograms in Figure 3 B3 are fitted with a sum of two Gaussians to differentiate between the long-range interactions and the short-range spin exchange interactions. A difference in the MPF of was measured suggesting the effect is still apparent but weaker then without the a-chiral separation. An additional verification of the range of the force was performed with samples having a thicker bi-layer of glycine (0.8 nm) and the results are presented in the inset of Figure 3 B3. In the case of a sample in which the chiral molecules are separated by 0.8 nm, the difference in the force between the enantiomers has completely disappeared, suggesting a length dependent of the force as expected in the case of spin-related interactions34. These results also suggest that the penetration of the helical peptide into the monolayer is also short-ranged, otherwise the structural differences would become apparent in the interaction.
To obtain an insight into the observed spin-dependent exchange interaction and its relation to enantio-selective interaction between biomolecules, we developed a “toy model” that – despite its simplicity – captures the essential physics in a phenomenological manner. Since – as mentioned above – none of the currently available electronic structure programs enable a proper description of the transient spin polarization, we opted to mimic this phenomenon through inclusion of so-called “spin polarization-sources” (vide infra) within a Valence Bond (VB) framework37. In our toy model, the two chiral helices are represented as 2 two-site systems, i.e., two H2-like molecules. For the resulting four-site system, 6 (covalent) VB determinants can be defined, corresponding to the different distributions of 2 α- and 2 β-electrons in 4 orbitals (see Supplementary Figure S7). In the dissociation limit, i.e., in the limit of infinite separation between the two molecules, the spins will not interact with each other. To phenomenologically enforce (dispersion-induced) spin polarization, we add a “spin polarization-source” to each of the molecules (Supplementary Figure S8A). These spin polarization-sources are single sites with a predefined/fixed spin which induce spin polarization in the adjacent molecules by increasing/decreasing the energy of the respective VB determinants. As can be seen from Supplementary Figure S8B and S8C, placing the two sources with an opposing spin leads to spin polarization of the same sign in both molecules; placing them with a parallel spin leads to spin polarization of the opposite sign in both molecules. Note that the distance between the source and the actual molecule is an arbitrary parameter which sets the extent of induced spin-polarization; in the calculations performed, a distance of 0.1 nm was selected, which results in net spins of approximately 0.33e on each of the individual sites of the molecules in the dissociation limit.
As expected, in the dissociation limit, the spin polarized wavefunctions obtained for the two spin-polarization-source alignments are degenerate in energy. When the spacing between the two molecules is reduced however, this degeneracy is broken: at a distance of 2-3 Å between the two molecules, the spin interaction starts to become significant.
Next to the (spin-dependent) exchange term, a Lennard-Jones potential with parameters σ=0.16 nm and ε=3.0 kcal/mol was added to the model to collectively account for all the spin-independent interaction terms. The resulting potential energy profile is shown in Figure 4A for the case of collinear approach of the two spin polarized species. The energy one samples with the AFM measurements corresponds to the difference between the bottom of the well in each potential (namely for the two spin configurations, parallel and antiparallel spins). It should be clear that the toy model provides qualitative result that are consistent with the experimental measurements. Figure 4B shows the angle and distance dependence of the energy difference for the two spin configurations. This plot reveals a clear preference for a collinear configuration due to the spin exchange interaction.
The preference for a narrow range of approach angles emerging from our model can be expected to be relevant to many complex biological processes, since it may reduce the phase space that a system has to explore before reacting, thus enhancing reaction rates. As such this relatively subtle, yet probably ubiquitous, effect may be responsible for the remarkable speed of bio-recognition effects in protein folding and in searching DNA by enzymes.
To summarize, three experiments using chiral atomic force microscopy were presented. The first, directly measured the interaction force difference between same and opposite enantiomers. The second and third experiments show interactions between helical peptides and present two main characteristics of the exchange interaction, its spin dependency and its short range ( . The third experiment also suggests that the peptide adsorbed on the tip of the AFM does not penetrate more than into the monolayer. Our findings suggest that spin-dependent exchange interaction may play a pivotal role in biorecognition processes. The energy scale for L-AHPA L-AHPA interactions is on the order of . When the chiral molecules interact, symmetry constrains that arise from the chirality create a different spin distribution for homochiral and heterochiral interaction. A toy model for the enantiomer-specific interaction shows that the interaction is controlled by the exchange interaction of electron spin pairs and exhibits a significant radial dependence. The experimental results and the model calculations suggest an additional interaction term, not taken into account so far, that may explain enhanced enantiospecificity and rates for various important processes occurring in Biology. The new interaction term introduces short range force that is especially relevant to biological systems, where the interacting systems are typically in crowded environment.