CHAPTER OUTLINE
2.2 Enthalpic and Entropic Contributions to Ligand/Drug Binding
2.3 The Strength of Non-Covalent Interactions
2.4 Desolvation and the Hydrophobic Effect
Box 2.1—Ionic interactions in a proton channel from influenza virus
2.7 C–H Bonds as Hydrogen Bond Donors
2.8 Aryl Rings as Hydrogen Bond Acceptors
Box 2.2—A π-hydrogen bond in glutathione S-transferase
Box 2.3—Hydrophobic and cation-π interactions in the binding of neurotransmitters and drugs
Box 2.4—C–F bonds as hydrogen bond acceptors and in orthogonal interactions with carbonyl groups
2.13 Case Study—Inhibitors of Factor Xa as Anticoagulants
2.1 Introduction
In the first chapter we described the nature of the covalent chemical bonds in biological molecules and drug substances. In this chapter we will discuss the various non-covalent interactions that are important in biological molecules and in their interaction with drug molecules. Although non-covalent interactions are typically orders of magnitude weaker than covalent bonds, this does not imply that they are of less importance. As we will see, these “weak” interactions are what give proteins their particular three-dimensional shape and function, enable the copying of genetic information in DNA, and govern the interactions of drug molecules with their biological targets.
We can glean the importance of non-covalent interactions by looking at the structures of the amino acids that form proteins and the nucleotide bases present in DNA and RNA. Nature creates a huge diversity of structural and functional proteins using the same 20 amino acid building blocks (Table 2.1). Individual amino acids are distinguished by the chemical nature of their side chains. They can be roughly grouped into categories as being hydrophobic (leucine, valine, etc.), aromatic (phenylalanine, tryptophan), hydrophilic/uncharged (serine, glutamine), and hydrophilic/charged (lysine, arginine, aspartate, glutamate). This diversity is no accident—nature has selected for amino acids that are capable of forming a wide range of non-covalent interactions.
Table 2.1 Structures of the Amino Acids Present in Protein Structure.
While proteins evolved primarily to serve structural and functional roles, the role of the DNA molecule is to store information. Here too nature has employed non-covalent interactions (hydrogen bonds and aryl-aryl stacking) that are well suited to the task at hand. The highly complementary yet reversible nature of A:T and G:C base pairing ensures high fidelity in the storage and copying of genetic information (Figures 2.1 and 2.2).
Figure 2.1 The familiar double helix structure of DNA is formed by two complementary strands of DNA held together by non-covalent interactions that include hydrogen bonds and aryl-aryl stacking interactions. The dimensions and location of minor and major groove are shown. A, adenine; C, cytosine; G, guanine, P, phosphate; S, sugar [deoxyribose]; T, thymine. (Reproduced, with permission, from Murray RK, Bender D, Botham KM, Kennelly PJ, Rodwell VW, Weil PA. Harper’s Illustrated Biochemistry. 29th ed. New York: McGraw-Hill Education; 2012.)
Figure 2.2 Base pairing between complementary adenine-thymine and cytosine-guanine involves hydrogen bonding (dashed lines). (Reproduced, with permission, from Murray RK, Bender D, Botham KM, Kennelly PJ, Rodwell VW, Weil PA. Harper’s Illustrated Biochemistry. 29th ed. New York: McGraw-Hill Education; 2012.)
For a drug molecule to be safe and effective it must bind to its biological target with high affinity and fidelity. Most drugs bind via non-covalent interactions and even those that bind covalently must first “recognize” their intended target via non-covalent interactions. The structures of drugs are therefore imbued with many of the same chemical features found in their biological targets, including hydrophobic regions, hydrophilic regions, hydrogen bond donors/acceptors, aromatic rings, and charged atoms. Moreover, since most drugs are synthetic substances, they can be designed to exploit other non-covalent interactions that are less common, or even absent, in biological macromolecules. In this chapter we will discuss a wide range of non-covalent interactions, with an emphasis on their importance in the binding of drugs to their biological targets.
2.2 Enthalpic and Entropic Contributions to Ligand/Drug Binding
Any favorable non-covalent binding interaction is associated with a negative free energy of binding (ΔG). This is true of two interacting proteins, two interacting small molecules, or the interaction of a small molecule (drug) with its biological target. The free energy of binding is in turn dependent on changes in enthalpy (ΔH) and entropy (ΔS) according to the familiar equation ΔG = ΔH – TΔS. What this equation reveals is that binding events can be enthalpically and/or entropically driven. Enthalpy-driven interactions tend to be those that require precise positioning of the interacting partners, as is the case, for example, with hydrogen bonds (Section 2.6) and halogen bonds (Section 2.11). Entropically driven interactions often involve the displacement of water molecules from a protein surface into bulk solution, thus increasing the overall disorder (entropy) of the system.
Another interesting consequence of the relation ΔG = ΔH – TΔS is that weak binding interactions (small ΔG) can result from large but counteracting ΔH and ΔS values. In fact, drug binding often results from a balancing of entropic and enthalpic factors that are in opposition. This counterbalancing is sometimes referred to as entropy-enthalpy compensation and while it is not a rule of intermolecular interactions it is quite common. To see why this might be so, consider that an exothermically favorable interaction such as a well-positioned hydrogen bond will necessarily require precise spatial orientation of the interacting molecules. While these constraints on motion and orientation make for an exothermic hydrogen bond (favorable ΔH), they also reduce entropy. Conversely, consider the interaction of an aliphatic side chain in a drug with a hydrophobic surface on a protein. The enthalpy of binding in this case may be small or even endothermic, but in the event several water molecules are expelled from the hydrophobic surface into bulk solution and a significant gain in entropy is the result.
As we review various types of non-covalent interactions in the sections below, it is important to remember that these interactions do not happen in a vacuum (at least not in living organisms). We will see in Section 2.4 that the enthalpy and entropy of water molecules are often crucial factors in the overall free energy of drug binding.
2.3 The Strength of Non-Covalent Interactions
The strength of a non-covalent interaction will depend on various factors, among which is the distance between the interacting groups. Generally, the strength of interaction will increase as the two groups approach one another in space, reaching a maximum attraction at some specific distance r, and becoming less attractive and eventually repulsive as the groups are forced still closer together (Figure 2.3). The specific relationship between distance and attraction is not the same for all types of non-covalent interactions. The strength of ionic interactions, for example, is inversely proportional to the distance between charges (1/r2) whereas the strength of van der Waals interactions and hydrogen bonds are proportional to 1/r6. Ionic interactions therefore can be attractive over a significant distance whereas hydrogen bonds are attractive only over a very narrow range of distances. The strength of non-covalent interactions is also inversely related to the “dielectric” of the medium (i.e., solvent) in which the interaction occurs. A high-dielectric medium such as water will favorably surround (or “solvate”) ionic species, whether positively or negatively charged. This will tend to weaken the electrostatic attraction of the charges and indeed, such ionic interactions (Section 2.5) tend to be weak when they occur on the surface of a protein, near water. The same interaction will be much stronger should it occur in the interior of a protein, an environment more akin to a low-dielectic organic solvent.
Figure 2.3 Potential energy diagram illustrating how the strength of van der Waals interactions vary as a function of the distance r in Ångstroms. (Reproduced, with permission, from Murray RK, Bender D, Botham KM, Kennelly PJ, Rodwell VW, Weil PA. Harper’s Illustrated Biochemistry. 29th ed. New York: McGraw-Hill Education; 2012.)
The positioning of interacting groups in space is also important, and as with distance, the requirements vary for different types of non-covalent interactions. Ionic interactions, for example, can be approximated as an interaction of two point charges in space. The relative orientation of such point charges is not important, only the distance between them matters. We might say that such interactions are nondirectional in nature. If the interacting groups are not point charges but polarized bonds (dipoles), then we might expect the relative positioning to be very important (and at least as important as distance). This is the case for hydrogen bonds, where a polarized carbonyl bond interacts with a polarized N–H or O–H bond. We can therefore say that hydrogen bonds are directional interactions. Between the examples of two interacting point charges and two interacting dipoles lie many other interactions with geometric requirements that fall between these two extremes.
2.4 Desolvation and the Hydrophobic Effect
The tendency of hydrophobic solutes to associate in an aqueous solution is a consequence of both favorable van der Waals interactions between the hydrophobic surfaces and more importantly, the exclusion of water molecules from those hydrophobic surfaces. This latter effect is known as the hydrophobic effect and is a consequence of some special properties of the water molecule. Water is the smallest molecule capable of both donating and accepting a hydrogen bond. This allows for a complex yet dynamic network of hydrogen bonding interactions between water molecules in solution (Figure 2.4). This arrangement is optimal from both an enthalpic and entropic perspective because every water molecule is free to move through bulk solution while continually breaking and reforming enthalpically favorable hydrogen bonds along its travels. This ideal is disrupted when a more hydrophobic solute (e.g., a drug or protein) is introduced to the solution. In response, water molecules will tend to form a highly organized “shell” around the solute so as to minimize its interaction with bulk solution. When more than one hydrophobic solute is present, it becomes energetically favorable for them to associate since fewer water molecules are required to surround the aggregate than the individual solutes. The net result of association is then the release of water molecules from the organized solvation shell and back into bulk solution, where their entropy is much greater. This is the classical description of the hydrophobic effect—an entropy-driven process that tends to minimize the amount of solvated hydrophobic surface area.
Figure 2.4 Image illustrating hydrogen bonding interactions in water molecules in solution. The central water molecule is shown donating two hydrogen bonds and accepting one hydrogen bond. Reproduced under terms of the Creative Commons Attribution-Share Alike 3.0 Unported license, commons.wikimedia.org. Copyright Thomas Splettstoesser, www.scistyle.com.
The classical, entropy-driven hydrophobic effect holds for “idealized” solutes—what we might imagine as tiny, molecule-sized drops of oil in water. Of course proteins and drugs are not spherical drops of oil and, in fact, the shape and chemical nature of a protein surface affects the entropy and enthalpy of the water molecules surrounding it. In the case of narrow crevices in proteins or flat aromatic surfaces in drugs, the hydrophobic effect can even be enthalpy-driven. This so-called “nonclassical hydrophobic effect” occurs when water molecules surrounding a surface are unable to form good hydrogen bonding contact with other water molecules. Upon release from such surfaces, the water molecules are able to form exothermically favorable hydrogen bonding geometries and distances in bulk solvent. If the distinction between the classical and nonclassical hydrophobic effects is still not clear, do not fret. The most important thing to remember is that the entropy and enthalpy of water molecules surrounding a solute (drug or protein) will have a significant effect on the association of those solutes. When it comes to the chemistry of life, water is never an unconcerned spectator.
The hydrophobic effect is hugely important in both protein folding and the binding of drugs to their biological targets. In a folded protein, peptide sequences composed of hydrophobic amino acids (Val, Leu, etc.) will naturally tend to form hydrophobic contacts so that water is released to bulk solution. An example from protein structure is the formation of α-helices that further assemble into bundles of α-helices (Figure 2.5). The formation of such structures involves both polar and hydrophobic interactions. Hence the backbone amide bonds in the α-helix form specific, directional hydrogen bonds that stabilize the helical structure. Assembly of multiple helices is often driven by the desolvation of hydrophobic side chains on the interacting helices and is thus an example of the hydrophobic effect at work.
Figure 2.5 Association of two α-helices in a “leucine zipper.” The association of the two helices is driven by the burial of hydrophobic surfaces on one side of each α-helix.
The binding of a drug to the active site of an enzyme or receptor usually involves the desolvation of hydrophobic patches on both drug and enzyme/receptor. The magnitude of the effect can be estimated by calculating the total hydrophobic surface area that is “buried” (made inaccessible to water) upon drug binding. In fact, buried hydrophobic surface area has been found to be the best single predictor of drug binding affinity across diverse drug-target interactions. Of course drug binding sites are not uniformly hydrophobic and different binding sites vary widely in their hydrophobic and hydrophilic character. The hydrophobic effect will tend to be most important when hydrophobic and/or narrow and poorly solvated surfaces are desolvated upon drug binding.
2.5 Ionic Interactions
Ionic interactions are probably the easiest type of non-covalent interaction to understand. We know that opposite charges attract one another while like charges are repulsive. Some other notable characteristics of these interactions are that their strength decreases gradually with distance and that they are nondirectional interactions. This means that ionic interactions can have attractive or repulsive effects over considerable distances and that the interacting groups do not need to be precisely positioned to exert their effects. The strength of an ionic interaction will also depend on the dielectric constant of the environment surrounding the interacting groups. For example, imagine a protein with lysine and aspartate residues in proximity on the surface of a protein. Both groups will be highly “solvated,” meaning they will be surrounded by water molecules. The positively charged amine on lysine will interact favorably with the lone pair electrons on water molecules while the negatively charged carboxylate on aspartate will interact with the more electropositive hydrogen atoms of the surrounding water molecules. These interactions with solvent will weaken the strength of the ionic interaction. If on the other hand the Lys and Asp residues are found in the hydrophobic interior of a protein, the ionic interaction will be very strong indeed. Such interactions are called salt bridges in the language of protein structure and usually contribute significantly to the stability of a particular protein conformation (or “fold”).
The hydrophobic nature of protein interiors can also affect the ionization state of amino side chains. We usually think of aspartate and glutamate side chains as anionic groups and this is indeed the predominant ionization state for these residues in water at neutral pH. The basic side chains of lysine and arginine in contrast are positively charged in aqueous solution at neutral pH. Ionization of these groups is much less favorable however in low-dielectric organic solvents or in the hydrophobic interior of proteins. The result is that acidic and basic side chains are more likely to be found in their neutral (non-ionized) state in the protein interior than when exposed to water on the surface of a protein. Another factor that can influence ionization state is the relative proximity of two or more ionizable groups. For example, when two anionic carboxylate side chains are in close proximity they will tend to repel one another. A consequence of this repulsion can be that the acidity one of the residues is reduced such that only one of the two side chains is in its ionized form. Another example of proximity effects on the acidity of amino acid side chains is described in Box 2.1.
