Standard Article

You have free access to this content

The Principles of and Reasons for Using Water as a Solvent for Green Chemistry

Part 5. Reactions in Water

  1. Ronald Breslow

Published Online: 15 MAR 2010

DOI: 10.1002/9783527628698.hgc047

Handbook of Green Chemistry

Handbook of Green Chemistry

How to Cite

Breslow, R. 2010. The Principles of and Reasons for Using Water as a Solvent for Green Chemistry. Handbook of Green Chemistry. 5:1:1–29.

Author Information

  1. Columbia University, Department of Chemistry, New York, NY, USA

Publication History

  1. Published Online: 15 MAR 2010

Chemistry Terms

Choose one or more boxes to highlight terms.

Abstract

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

Water is an environmentally benign solvent. It also has special properties useful in chemical and biochemical reactions: the hydrophobic effect can lead to altered selectivities and rate accelerations. Although these are commonly seen in water solutions – which require water solubility of the reactants – we have reported one example in which such effects were seen in the water suspension of a poorly soluble reactant. Here we describe examples, chiefly from our own work, in which the hydrophobic effect was used in substrate binding, in aromatic chlorination, in acylations by esters, in mimics of metalloenzymes and of ribonuclease, and amino acid syntheses and reactions, heterocyclic carbene chemistry, aldol condensations, Diels–Alder reactions, carbonyl reductions, and selective oxidations. The hydrophobic effect was also used to determine the geometries of transition states for some important reactions. With these attractive special advantages, including the green chemistry advantages, water should become a more usual solvent in chemical manufacturing.

1.1 Introduction

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

Chemical reactions used to manufacture important compounds such as medicinals are essentially always carried out in solution, and this is also true of the research work that is used to invent the new compounds and to develop appropriate ways to manufacture them. In the past, continuing into the present, the solvents used are normally volatile organic compounds (VOCs), and these pose an environmental problem. Their vapors can contribute to the greenhouse effect that causes global warming, and in some cases the solvent vapors can catalyze the destruction of the ozone layer that protects the Earth and its living inhabitants from short-wavelength ultraviolet solar radiation. The vapors may also be toxic to humans, plants, or animals, or they may cause diseases.

The liquids themselves can be a problem. If they are released into the earth, rivers or the ocean, they can cause direct environmental damage, while also slowly releasing their vapors. In principle, the solvents can be completely captured and purified for reuse during manufacturing, but it is difficult to prevent some loss to the environment. Hence there is interest in using environmentally benign liquids as the solvents in chemical reactions.

One possibility is supercritical carbon dioxide, which is a liquid under pressure and which has attractive solvent properties. However, unless it is completely contained and reused, it will release gaseous carbon dioxide, a greenhouse gas. Thus interest has increasingly turned to water as the solvent for chemical reactions.

Water is the solvent in which biochemical reactions are performed in Nature, and it is environmentally benign. However, it is a good solvent only for organic chemicals that have polar groups, such as alcohols and carboxylic acids. This may not be an insuperable problem. Over 20 years ago we reported that the special selectivities seen in water solution (see below) were also seen in some water suspensions, where one soluble component reacted with one that was poorly soluble (1, 2). We pointed out that such suspensions in water could well be generally more practical ways to use water in manufacturing (2). Recently, Sharpless and co-workers described a remarkable acceleration of a reaction in such a suspension, which they called reactions ON water (3, 4). The large reported rate effect was seen in only one particular case, but even without a large acceleration the selectivities that we describe below could perhaps make suspensions in water a practical way for the environmentally benign properties of water to be generally useful even with insoluble reaction components.

One industry that has switched from VOCs to water is the paint industry. We are all familiar with the water-based paints that no longer emit strong solvent odors, and these have been widely adopted for painting automobiles, for instance. It is essentially impossible to capture all the solvent vapors that are released when a vehicle is spray painted, but when the solvent is water there is no problem.

Water is not simply an environmentally benign solvent; it has special properties that are essentially unique, related to what is called the “hydrophobic effect.” This is the tendency for hydrocarbons or molecules with hydrocarbon components to avoid contact with water, and to associate instead with other hydrocarbon species in water. This is what makes aqueous soap solutions dissolve grease, and it is the driving force in biology for the associations that produce cell membranes, and that cause nucleic acids to form the famous double helix. It drives the folding of proteins into their shapes in enzymes and antibodies, and it also promotes the binding of biological substrates into enzymes and antibodies (5).

As described below, the hydrophobic effect has now been used to mimic biological chemistry and to provide remarkable selectivities in the field called biomimetic chemistry. It has even been used to permit the discovery of the geometries of the transition states for some interesting reactions, information that is otherwise inaccessible. The remainder of this chapter describes examples of the use of the unique property of water to achieve not just solubility but also selectivity, but the examples will be mainly chosen from our own work. Hence it is important to refer to a number of sources in which other authors have also described their use of water and the hydrophobic effect in chemical studies.

Some of the work of our group has been presented as chapters in the books Structure and Reactivity in Aqueous Solution (6), Green Chemistry (7), and most recently Organic Reactions in Water (8). In addition, in various review articles our work has been placed in context with that of other groups (2, 5, 9-20). The remainder of this chapter describes the various contexts in which we have seen the special properties of water as a solvent.

1.2 Binding of Two Species Together Driven by the Hydrophobic Effect in Water

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

Cyclodextrins are molecules composed of glucose units linked in rings, the most common being α-cyclodextrin (six glucose units), β-cyclodextrin (seven glucose units) and γ-cyclodextrin (eight glucose units) (Scheme 1.1). The three exposed hydroxyl groups on each glucose unit make then water soluble, but they have an internal cavity that is less polar, and that will bind hydrocarbons such as aromatic rings using the hydrophobic effect in water. In later sections it is described how such cyclodextrin–substrate complexes can catalyze reactions, imitating enzymes. Here the cases where binding alone was studied are described.

original image

Scheme 1.1. The three cyclodextrins used – α, β, and γ-cyclodextrin – and two ways in which they are symbolized.

In one example, we saw that some dipeptides would selectively bind into simple β-cyclodextrin in water (21), and that the large steroid lithocholic acid bound strongly (22), as did cocaine (23). When we linked two β-cyclodextrins together, we achieved even better binding of cholesterol (24), and such cyclodextrin dimers also showed strong and selective hydrophobic binding of compounds with two phenyl groups (25), of peptides with two hydrophobic amino acid components (26), and of oligopeptides whose binding promoted the formation of a helix (27).

We also tied two β-cyclodextrins with two links, which made a hinge that could let the two cyclodextrins close around a substrate, and also another geometry in which they were prevented from cooperating (28). As hoped, the dimer with the correct geometry was a very strong binder of hydrophobic substrates, since the double link had frozen out the incorrect geometries. Interestingly, in one study we saw that such strong double binding was reflected in a better enthalpy, rather than entropy (29). Our early work with cyclodextrin dimers has been reviewed (30). We also examined a dimer of a cyclophane, another species with an internal cavity that binds hydrophobic groups (31). The findings were similar to those with the cyclodextrin dimers. In addition, we examined some trimers of cyclodextrins, but did not see as much cooperativity as one might expect (32).

We synthesized some cyclodextrin dimers with photocleavable links as potential carriers of anticancer photodynamic sensitizers (33, 34). We also saw that some cyclodextrin dimers could bind to proteins and prevent their aggregation (35). Furthermore, we saw that some of our cyclodextrin dimers and trimers could bind to amyloid protein and prevent the aggregation that causes Alzheimer's disease (36). Some other studies with cyclodextrin dimers will be presented in Section 1.5 on mimics of metalloenzymes.

1.3 Aromatic Chlorination

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

In our earliest work using cyclodextrins to bind substrates, we examined the chlorination of anisole by hypochlorous acid in water with and without added α-cyclodextrin (37, 38). We saw that the anisole in solution was chlorinated in both the ortho and para positions, but in the complex with α-cyclodextrin only the p-chloroanisole was formed. The kinetic studies showed that the chlorination involved the prior attachment of chlorine to a hydroxyl group of the cyclodextrin, and then its transfer to the bound anisole (Scheme 1.2).

original image

Scheme 1.2. α-Cyclodextrin catalyzed the selective chlorination of anisole in water by an intra-complex transfer of a chlorine atom.

We also examined other substrates, whose behavior reflected this same mechanism (38). In this case, the cyclodextrin is acting as a mimic of the enzyme chlorinase, except that interestingly the enzyme mimic was more selective than was the enzyme itself. In a later study, we established which hydroxyl group was the chlorine transfer agent, and showed that a cyclodextrin polymer could perform the selective chlorination in a flow reactor (39).

1.4 Acylation of Cyclodextrins by a Bound Ester

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

Komiyama and Bender examined the reaction of m-nitrophenyl acetate with cyclodextrins, and saw that they transferred the acetyl group to a hydroxyl of the cyclodextrin, with a modest 250-fold rate enhancement over the hydrolysis rate in water under the same conditions (40). Our modeling of this process indicated that the starting material could occupy the cyclodextrin cavity, but that the tetrahedral intermediate for acetyl transfer would have its nitrophenyl group largely pulled from the cavity. This picture was confirmed by a study of the effect of high pressure on the reaction rate, which indicated that the volume of the transition state was larger than that of the starting complex, as such a geometric change would cause (41). Such a loss of binding would be energetically unfavorable for the reaction, accounting for the very modest rate of the acetyl transfer process. We therefore created a series of substrates that could avoid this problem.

Molecular models indicated that compound 1, based on a ferrocene core, would be able to acylate a cyclodextrin hydroxyl while still retaining most of the binding of the ferrocene unit in the cyclodextrin cavity. We synthesized 1, and saw that indeed it acylated β-cyclodextrin with a 51 000-fold rate acceleration (42). However, high-pressure studies (41) indicated that although indeed the transition state for the reaction retained most of the binding into the cyclodextrin cavity, it was not yet the ideal substrate. By modifying the cyclodextrin itself – adding a floor to the cavity – and adjusting the substrate further, we achieved a rate acceleration of ca 106-fold (43).

With an even better substrate geometry, we achieved a rate acceleration of 108-fold, and the reaction was also enantioselective (cyclodextrin is composed of chiral glucose units), with a 20 : 1 preference for one substrate enantiomer over the other (44). The optimizations and their explanations were described in a full paper (45), and theoretical calculations on the geometric factors involved were described in another publication (46).

The very high rates of the best substrates reflected a rigid geometry that favored the first step of acylation – addition of the cyclodextrin hydroxyl to the ester carbonyl to form a intermediate – but in the next step, departure of the p-nitrophenoxide ion to form the product acylated cyclodextrin, this rigidity was undesirable. Some flexibility was needed in the substrate to permit the rotation involved in this second step. When we incorporated such flexibility, both steps were well catalyzed even with an ordinary ester, where the second step could be rate limiting (47). Thus, these studies on cyclodextrin acylation by bound substrates indicated the enormous rate accelerations that can be achieved using the hydrophobic effect to promote catalyst– substrate binding in a well-designed geometry.

Such work accomplishes two goals. It indicates that incorporating the factors we believe play a role in enzymatic catalysis does indeed lead to very good catalysis, approaching the rates of the best enzymes. This helps confirm our ideas about how enzymes are able to function so effectively. At the same time, these studies strengthen ordinary chemistry. They show how to make effective catalysts with good rates and selectivities, adopting the principles but not the details of enzymatic reactions.

1.5 Mimics of Metalloenzymes Using the Hydrophobic Effect in Water

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

For hydrolytic enzymes, the formation of an acyl-enzyme intermediate is only a first step; for catalysis, the intermediate must hydrolyze to regenerate the catalyst and liberate the product carboxylate ion. In many hydrolytic enzymes, including the most effective ones, substrate binding involves both the hydrophobic effect induced by water and some binding to the metal ion itself, which is held in the enzyme by typical coordinating groups. In our first study of mimics for such enzymes, we constructed an artificial enzyme 3 comprised of an α-cyclodextrin ring for hydrophobic binding and an attached pyridinecarboxylate to bind a Ni(II) ion (48). The nickel also bound a nucleophilic oxime group.

original image

We found that p-nitrophenyl acetate was hydrolyzed in a two-step process, after it was hydrophobically bound into the cyclodextrin. First the nucleophilic oxime removed the acetyl group, in a nickel-catalyzed reaction, and then the nickel ion catalyzed the hydrolysis of this intermediate, regenerating the catalyst. The geometry permitted this process, not the direct acylation of cyclodextrin as in the systems in the previous section. However, the rate acceleration was modest, reflecting the many degrees of flexible freedom in the catalyst.

We constructed some metal ligands mirroring those in metalloenzymes such as carbonic anhydrase, and studied their ability to bind zinc(II), the metal ion in carbonic anhydrase and in carboxypeptidase, and other metal ions (49). In a study of the hydrolysis of a phosphate triester, we saw evidence that a bound Zn(II) acted as a bifunctional catalyst, delivering a hydroxide ion to the phosphorus while coordinating to the phosphate oxygen atom to stabilize the phosphorane intermediate in hydrolysis (50). We have seen such a process in many enzyme mimics that also use hydrophobic binding of substrates, as discussed below.

Carboxypeptidase uses metal ion catalysis in the hydrolysis of an amide group, a peptide bond. In a relevant study we used Co(III) to lock the amide oxygen to a metal ion [cobalt(III) is substitution inert], and saw hydrolysis of the amide with the assistance of phenol groups of the catalyst (51). Apparently the phenol group and some others that we examined play a role in the second step of amide hydrolysis, fragmentation of the tetrahedral intermediate. We also attached α-cyclodextrin to a macrocyclic zinc ligand that held the metal so strongly that it could exist as the zinc hydroxide without losing the zinc (52). The compound bound phosphate esters into the cyclodextrin using the hydrophobic effect in water, and then used the bifunctional zinc hydroxide mechanism to hydrolyze the substrate. In related work, we catalyzed the cyclization/cleavage of a model for conversion of RNA to its cyclic phosphate using the well-bound zinc macrocycle with an attached thiol or imidazole second catalytic group (53).

We constructed cyclodextrin dimers with a catalytic metal ion bound to the linking group. In the first example, esters that could hydrophobically bind into both cyclodextrins, stretching along the linking, were hydrolyzed by bound copper(II) hydroxide using the bifunctional nucleophilic bound hydroxide plus electrophilic metal ion mechanism, in one case achieving a 220 000-fold rate acceleration over uncatalyzed hydrolysis in water (54). We also saw that such a cyclodextrin dimer could bind a bis-p-nitrophosphate anion to an La(III) ion coordinated to the linking group and then achieve catalytic cleavage of the phosphate ester with added hydrogen peroxide (55). In a full paper describing such cyclodextrin dimer catalysts for ester hydrolysis, we saw as much as a 107-fold rate acceleration (56). Using cyclodextrin dimers with bound metal ions, we saw a 103-fold acceleration of the hydrolysis of a bound benzyl ester, less reactive than some of the p-nitrophenyl esters used in earlier studies (57).

In another approach, we constructed a β-cyclodextrin that had both a metal ion binder and an imidazole general base catalyst attached to the cyclodextrin, and examined the hydrolysis of a tert-butylcatechol cyclic phosphate 3 that hydrophobically bound to the cyclodextrin (58). The hydrolysis was accelerated ca 103-fold. We describe other studies of such a hydrolysis in the next section.

1.6 Mimics of the Enzyme Ribonuclease

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

Ribonucleic acid (RNA) is cleaved by the enzyme ribonuclease in an overall two-step process (Scheme 1.3). In the first step there is a cyclization/fragmentation in which the hydroxyl group on C-2 of the ribose attacks the phosphate group of the RNA chain and produces a cyclic phosphate 4 while breaking the chain at that point. In the second step, this cyclic phosphate is hydrolyzed to release the C-2 hydroxyl again while opening the cyclic phosphate ring. The enzyme can catalyze both of these rather different steps.

original image

Scheme 1.3. The enzyme ribonuclease cleaves RNA by a cyclization, then a hydrolysis of the cyclic phosphate. It is shown for uridyluridine, a dinucleotide component of RNA.

The major catalytic groups in bovine ribonuclease A are two imidazole rings of the amino acid histidine, although an ammonium ion of lysine also plays a role. At the optimum pH for the enzyme, one imidazole is protonated and serves as a general acid catalyst, whereas the other imidazole is unprotonated and acts as a general base. We decided to produce a mimic of this enzyme by using hydrophobic binding of a substrate into the cyclodextrin cavity in water, in which the cyclodextrin also had two imidazole rings replacing two hydroxyls of the cyclodextrin.

In our first study (Scheme 1.4), we attached the imidazoles on opposite sides of the cyclodextrin cavity and examined the ability of this catalyst to hydrolyze compound 3, a cyclic phosphate as a rough mimic of the cyclic phosphate that is hydrolytically cleaved in the second step of the enzymatic process (59). We saw that there was a pH optimum for this hydrolysis that was essentially identical with that of the enzyme itself, indicating that both the general base and the general acid versions of the imidazoles were cooperating in the hydrolysis process. The substrate was selectively cleaved to 5, leaving the phosphate group meta to the tert-butyl group. By moving the imidazoles out slightly, we could reverse the selectivity, now leaving the phosphate group para to the tert-butyl group (60).

original image

Scheme 1.4. β-Cyclodextrin with two attached imidazole rings catalyzes the hydrolysis of a bound cyclic phosphate ester in water with specificity, and with geometric and isotopic evidence that indicates a process involving a phosphorane intermediate.

A method called proton inventory had been applied to ribonuclease (61). By observing the rate with different ratios of H2O and D2O, it is possible to deduce whether one or two protons are moving in the rate-determining step, and it was concluded that two protons were moving. This means that as the general base imidazole is removing the proton from the C-2 hydroxyl group the general acid imidazolium ion is transferring its proton to the substrate in a simultaneous bifunctional process. We applied this test to our bisimidazolecyclodextrin enzyme mimic, and saw the same result, and with almost the same data as had been seen with the enzyme (62).

We also varied the structure of the bisimidazolecyclodextrin catalyst. We were able to synthesize isomers with the two imidazoles on neighboring glucose units, which we called the A,B isomer, and also an isomer with imidazoles on the A,C units and on the A,D units (63). If the cleavage mechanism had involved direct attack on the phosphate group while a proton was being placed on the leaving oxygen, the A,D isomer should have been the best. However, we saw that the most active isomer was A,B, with the acid and base groups on neighboring glucose units. This absolutely requires a mechanism in which the hydrolysis proceeds through an intermediate with five oxygens on phosphorus, a phosphorane, which later fragments to the final product. As we shall describe, we saw evidence for the same mechanism with a different model system.

We examined the hydrolysis of a simple dinucleotide, uridyluridine, in water solution with imidazole buffer. Since this process does not involve the hydrophobic special effects of water, it will not be described in detail and rather the relevant references are listed (64-73). The evidence points to a phosphorane intermediate for this simple buffer-catalyzed process, and we suggested that the enzyme may well be using the same mechanism, rather than a direct cleavage. There is not general agreement on this idea for the enzyme. We have published an account of both the cyclodextrin studies and the buffer studies in ribonuclease mimics (74), and an account of the result of variation in the geometries of the bisimidazolecyclodextrins and the substrate, which made it clear how important it is to have a relatively tight fit of the substrate in the hydrophobic cavity of the cyclodextrin (75).

1.7 Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

We have constructed a number of such mimics. In general, they use the hydrophobic effect in water to bind the substrates for the reactions, with the pyridoxal or pyridoxamine unit coenzyme mimics covalently attached to a cyclodextrin or a hydrophobic polymer. In a few cases we have also used the hydrophobic effect to bind reversibly a coenzyme mimic itself. Some of the resulting rate effects are truly enormous.

In our first study, we covalently linked a pyridoxamine unit to β-cyclodextrin in compound 6 and examined its ability to convert α-keto acids to amino acids in water (Scheme 1.5) (76). This directly mimics the process used by enzymes to synthesize most amino acids. We compared the conversion of phenylpyruvic acid 7 to phenylalanine 8 and of indolepyruvic acid 9 to tryptophan 10, both of which can exhibit hydrophobic binding into the cyclodextrin cavity, with the conversion of pyruvic acid 11 to alanine 12, in which there was no hydrophobic binding of the small methyl group.

original image

Scheme 1.5. Three different versions of a pyridoxamine attached to a cyclodextrin convert keto acids to amino acids in water with a preference for those hydrophobic substrates that can bind into the cyclodextrin cavity, imitating in part a biological process by which the amino acids are formed.

The two aromatic compounds had essentially the same rate as did simple pyruvic acid when pyridoxamine was used without the attached cyclodextrin, but hydrophobic binding of keto acids 7 and 9 led to about a 100-fold preference over alanine with the enzyme mimic 6. With 6, the tryptophan was formed with a 33% enantiomeric excess (ee) of the l-isomer, induced by the chirality of the cyclodextrin unit, and phenylalanine was formed with a 67% ee (77). We saw similar results with compound 13, in which the pyridoxamine was attached on one of the primary carbons of β-cyclodextrin and the other primary carbons were converted to methyl groups, forming a deeper hydrophobic pocket (78). Even higher selectivities were seen in some of our later work using the hydrophobic effect in amino acid synthesis.

In 6, the pyridoxamine is attached on the primary side of the cyclodextrin, but we also examined a compound 14 in which it was attached to the secondary side (77). All the acylations of cyclodextrin described earlier were directed to the secondary side. We found that with 14 there was also a preference for the formation of phenylalanine and tryptophan, rather than alanine, reflecting the hydrophobic binding of the two aromatic ring substrates in water. However, the preference was less than with the original primary-side linked compound 6, and the chiral inductions were also smaller. Apparently in this class of compounds, the primary side of the cyclodextrin is the better place for attachment of the pyridoxamine unit. A full paper summarized these results (79). We also prepared a couple of transaminase mimics with two links between the pyridoxamine and the cyclodextrin (80). The much better geometric control that this afforded led to very high selectivities among substrates with different geometries themselves.

Although the cyclodextrins are conveniently available compounds for incorporating hydrophobic binding in water into enzyme mimics, they are not unique. We also used some novel synthetic macrocycles that could carry a pyridoxamine unit and bind hydrophobic substrates into their cavity in water solution (81). We saw that compound 15 converted phenylpyruvic acid to phenylalanine 15 times more rapidly than did simple pyridoxamine, again reflecting acceleration by hydrophobic binding of the substrate. However, the effect was not as large as with 6, carrying a pyridoxamine attached to the primary side of β-cyclodextrin. Also, there was of course no chiral induction with the achiral synthetic macrocycle.

In a successful attempt at chiral induction, we synthesized compound 16, which has no hydrophobic binding group but has a chain carrying a basic unit that can deliver the new proton of the amino acid with geometric control (82). We saw that as much as a 94 : 6 ratio of d- to l-tryptophan was seen with the enantiomer of 16 that we used. The studies on amino acid synthesis up to this point were summarized in a full paper (83).

original image

Subsequently, we combined these ideas in enzyme mimics that had a pyridoxamine with a chirally mounted rigid proton transfer group, that was also attached to a cyclodextrin for hydrophobic binding (84). As much as a 96% ee was obtained in the best case. We also showed that we could reverse the optical preferences in some cyclodextrin–pyridoxamine enzyme mimics as we moved an attached basic group into different positions (85).

The principle biochemical function of pyridoxamine phosphate as a coenzyme is to convert keto acids to amino acids, as above, but the result of this process is also to convert the pyridoxamine unit to an aldehyde, pyridoxal phosphate. This aldehyde has biochemical functions in addition to reversing the transaminations process, so we synthesized compound 17 – with a pyridoxal unit attached to a primary carbon of β-cyclodextrin – to examine such reactions in water using hydrophobic binding (86). In Nature, tryptophan is synthesized by an enzyme with a pyridoxal phosphate coenzyme in which the pyridoxal forms an imine with the amino acid serine, and catalyzes its dehydration to a reactive olefin that then adds to the beta position of indole. The product is then hydrolyzed to tryptophan, with regeneration of the pyridoxal phosphate (Scheme 1.6).

original image

Scheme 1.6. A cyclodextrin carrying a pyridoxal unit catalyzes the conversion of indole to tryptophan, aided by hydrophobic binding, in imitation of a biochemical process.

Metzler et al. had tried to perform this reaction with simple pyridoxal, and had obtained tryptophan in a 1% yield (87). We obtained a somewhat better yield substituting β-chloroalanine for the serine, since HCl elimination was more facile than dehydration. With compound 17 carrying a hydrophobic binding group the process was three to five times better, but still tryptophan was produced in only a few percent yield. Furthermore, the chiral induction was not as large as in our previous transamination procedures for tryptophan production. The new system shows that two substrates can be coupled if one is hydrophobically bound into the cyclodextrin cavity and the other is covalently linked to the pyridoxal unit, but side reactions still predominate.

As the previous example indicates, there is a competition between reactions catalyzed by pyridoxal groups – in which the pyridoxal is recovered again at the end, so there can be catalytic turnover – and reactions with amino acids that reverse the amination process and produce α-keto acids and pyridoxamine. The latter processes do not permit turnover catalysis by the pyridoxal. Thus in part we have strived to control this dichotomy. For example, pyridoxal units can catalyze the racemization of amino acids, but the process is in competition with transamination to form pyridoxamines. We saw that an added rigid base group could do the proton transfers needed for racemization and increase the selectivity for this process (88). An even more extensive study with rigid bases was published later, again showing how racemization could be increased relative to transaminations (89). In a full paper, we described how the attachments of rigid bases could be used to control both the reactions performed and the stereochemistry achieved (90).

Pyridoxal phosphate is also the coenzyme for the coupling of glycine to acetaldehyde to form threonine (Scheme 1.7). We synthesized a pyridoxal with a loop across its face, and examined this as catalyst for the glycine–acetaldehyde reaction (91). The catalysis was successful, and an interesting reversal of optical selectivity occurred when the pH of the medium was changed.

original image

Scheme 1.7. Glycine condenses with acetaldehyde to form threonine.

Cyclodextrins and related synthetic macrocycles have been used in many artificial enzyme systems, as described above, but Klotz's group pioneered a different approach. They reasoned that real enzymes are long-chain macromolecules, so they examined the use of synthetic polymers to mimic them. The particular polymers on which they concentrated are polyamines, polyethylenimine, in which the nitrogens are separated by two ethano groups and there are both linear and branched structures (92). They and Kirby and co-workers (93) had used such a polymer to perform hydrolysis reactions, but we adopted it as the basis for mimics of synthetic enzymes, the transaminases. We have pursued artificial enzymes based on various polyamines of different sizes and produced in different ways to achieve some spectacular rate accelerations, and good chiral selectivities.

In our first approach, we used a commercial polyethylenimine with a molecular weight of about 60 000 that is fairly polydisperse and has extensive branching, with about 25% of tertiary nitrogens (94). We attached a pyridoxamine to it covalently, and then added some lauryl chains to nitrogens to produce a hydrophobic interior. The polyamine is ca 50% protonated at pH 8, so it has both general base amino groups and general acid ammonium groups, and can use these to catalyze transaminations. With this artificial enzyme 18, we saw the conversion of pyruvic acid 11 to alanine 12 in water with an 8300-fold rate acceleration over the same reaction with simple pyridoxamine, and a 10 000-fold acceleration relative to unbuffered pyridoxamine (the polymer needs no buffer, since the amino and ammonium groups provide the equivalent of buffer catalysis).

original image

On further study, we saw that this polymer–pyridoxamine compound converted indolepyruvic acid 9 to tryptophan 10 in water with a rate acceleration of 240 000-fold relative to pyridoxamine itself, reflecting the hydrophobic binding of the substrate into the polymer (95). The large effect reflected the hydrophobic character of the lauryl chains in the polymer, and with simple methyl groups instead the rate acceleration was much smaller.

We also examined other polyamines, including polyallylamine (94), and polyethylenimines of various sizes (96). Interestingly, even fairly small polyamines were almost as effective as the large one, probably indicating that the smaller ones clustered to form an effectively large structure.

Dendrimers are an interesting class of polymers that branch from a central point. We examined a group of dendrimers 19 that branched from a pyridoxamine unit, by adding two methyl acrylate ester units to each amino group, then reacting each ester group with ethylenediamine, then repeating the process again and again. We saw that the largest ones were similar in catalytic activity to the polyethylenimines that did not carry attached hydrophobic lauryl groups (97). In later work, we compared the two classes (98), and saw that attachment of chiral amino acids to the exterior of the dendrimers induced chiral amino acids (99). In this latter study, we also showed, by calculation, that the pyridoxamine that was the formal center of the dendrimer could actually lie on the surface, near the surface chiral groups.

original image

In the reactions in this section described so far, pyridoxamine units were reacting with keto acids to form amino acids, but the other products were pyridoxal units so the pyridoxamines were not regenerated. Hence we did not refer to such processes as“catalyses.” In Nature, the pyridoxals are converted back to pyridoxamines by reversing the amination process with a sacrificial amino acid, but neither we nor others had been able to achieve significant catalytic turnovers no matter which amino acids we tried to sacrifice. We finally solved this problem with an unusual class of amino acids, those such as α-methylphenylglycine 20 that could not undergo simple transaminations, but that instead performed a process we called transaminative decarboxylation (100).

In this process (Scheme 1.8), the pyridoxal forms an imine with the novel amino acid, and then promotes decarboxylation to form a delocalized anion. This anion protonates next to the pyridine ring, where most of the negative charge is centered, and in water the resultant compound hydrolyzes to generate pyridoxamine and a ketone, which in the case shown is acetophenone. We saw that this process could occur in the same medium in which transaminations by pyridoxamine was being performed, and simultaneously. Thus we obtained as many as 100 turnovers by the pyridoxamine unit, true catalysis.

original image

Scheme 1.8. Decarboxylative transaminations that convert pyridoxal derivatives to pyridoxamines, permitting turnover catalysis of amino acid synthesis.

In the same paper, we reported another important advance: the pyridoxamine was not attached to the polymer, but was reversibly bound in water when the pyridoxamine carried a hydrophobic side chain (100). The reversible system was somewhat faster than the covalently attached one, presumably because now both the substrate and the pyridoxamine bound into the same hydrophobic region in the polymer. This is similar to the real enzymatic case, in which pyridoxal phosphate is reversibly bound to the enzyme, not covalently attached. In a full paper, we examined the hydrophobic and electronic factors involved in decarboxylative transaminations reactions (101). It might be mentioned that we used a related decarboxylative transamination process in a mimic of a possible prebiotic reaction that could form natural amino acids with enantioexcesses, as part of an explanation of the origin of chirality on Earth (102).

In Nature, the transamination of keto acids forms amino acids as single enantiomers. As mentioned above (82), we had been able to perform transaminations with a chirally mounted basic group and achieve some enantioselectivity. We now turned to the production of polyamines with themselves carrying chirally attached side chains to see how well these could induce chiral selectivity in the polyamine-catalyzed reactions.

In our first study, we reported that chiral polyamines could be synthesized by the borane reduction of polypeptides, and that these induced some chiral selectivity in transamination processes (103). In a different approach, we synthesized chiral polyamines by the polymerization of oxazolines, and saw that these too induced some chiral preferences when they catalyzed transaminations (104). Further work towards the goal of high enantioselectivity in such catalyses is currently under way. A summary of the situation up to the near present was published (105), as was a general review of transamination mimics (20).

1.8 Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

Many years ago, we had shown that the coenzyme thiamine pyrophosphate acted by losing the C-2 proton of the thiazolium ring to form a compound that could be described as a thiazolium zwitterion 21 with a carbene resonance form 22 (106). Such zwitterion/carbene hybrid species were also formed with imidazolium ions (23 and 24), and they are now important ligands in metal ion catalysts, often referred to as “stabilized carbenes” (which we had described in an early paper (107)).

original image

The biochemical decarboxylations that thiamine pyrophosphate performs can also be performed in model systems by simple cyanide ion, not a practical choice for biology. They are electronically related to the benzoin condensation, for which both cyanide ion and thiazolium salts can perform as catalysts. In contrast to pyridoxamine, thiamine pyrophosphate is a true catalyst that does not need regeneration. Thus we synthesized artificial enzymes linking thiazolium ions to cyclodextrins, to use again hydrophobic binding of substrates in water.

In our first study, we attached a thiazolium salt to β-cyclodextrin in compound 25, and saw that it bound tert-butylbenzaldehyde to the hydrophobic cavity in water and promoted the ionization of the benzaldehyde proton, an important step in the benzoin condensation (108). However, the cavity of β-cyclodextrin is too small to accommodate a second benzaldehyde species, so the benzoin condensation was not catalyzed.

γ-Cyclodextrin – with a ring of eight glucose units – is able to bind two benzaldehydes side-by-side, so we prepared a related compound with a thiazolium ion linked to γ-cyclodextrin (109). This was a very effective catalyst. The flexible thiazolium ion link added a thiazolium ion to one benzaldehyde, and the deprotonated aldehyde group then added to the second benzaldehyde in the cavity. The thiazolium ion addition was then reversed and benzoin was liberated from the cavity (Scheme 1.9). Benzoin itself is linear and rigid, so the two phenyls can no longer occupy the same cavity. Thus this “enzyme” showed no inhibition by the product, and the transition state for the reaction did not resemble the geometry of the product. We are currently studying such enzyme mimics in which thiazolium, imidazolium, and triazolium ions cooperate with the polyamines described in the previous section (110).

original image

Scheme 1.9. A thiazolium salt coupled to β-cyclodextrin binds one benzaldehyde in the cavity, in water, and catalyzes deuterium exchange, but with the γ-cyclodextrin derivative two benzaldehydes are bound into the cavity and converted to benzoin.

1.9 Enolizations and Aldol Condensations

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

In the section on mimics of the enzyme ribonuclease, we described the use of cyclodextrins carrying two imidazole groups, and the at first surprising results that the cooperative base/acid hydrolysis of a substrate bound into the cyclodextrin cavity in water by the hydrophobic effect was preferentially performed when the two imidazoles were linked to adjacent glucose residues, what we called the A,B isomer. The availability of the group of A,B and A,C, and A,D isomers let us examine the geometric preference for enolization of a bound ketone with bifunctional catalysis by the two imidazoles. Thus we studied deuterium exchange into p-tert-butylacetophenone 27 by the set of bisimidazolecyclodextrins, and found that the preferred catalyst in this case was the A,D isomer (110). The pH versus rate profile showed that this was again bifunctional catalysis, with one imidazole acting as a base to remove the proton while the other, as the imidazolium ion, was hydrogen bonded to the carbonyl oxygen and delivered its proton to form the enol during deprotonation of the methyl group.

Models show that all three isomers could reach the carbonyl oxygen and the methyl hydrogen, but the direction of approach was different. The preferred isomer had a preferred direction of attack on the methyl hydrogen, pushing the electrons towards the carbonyl group (Scheme 1.10). The defined geometries of the three isomeric bisimidazolecyclodextrins could be a useful tool in discovering such details of mechanism.

original image

Scheme 1.10. A bisimidazolecyclodextrin catalyzes the enolization of a hydrophobic ketone in water. The most effective catalyst had the imidazoles far apart, indicating the geometry used for proton abstraction, the first evidence for this detail.

We used the same bisimidazolecyclodextrin isomers to perform an internal aldol condensation of a keto aldehyde. Again there was a preference for the A,D isomer, and the product of the aldol condensation reflected the substrate geometry in the cyclodextrin cavity. When we used the bisimidazolecyclodextrin to catalyze an aldol condensation of a dialdehyde substrate that could hydrophobically bind into the cyclodextrin cavity, an otherwise random aldol reaction was turned into a selective one, again because of the geometry of the bound substrate in the cavity (111). We also catalyzed the condensation of a hydrophobically bound group of various benzaldehydes with ketones that formed enamines with amino groups linked to the cyclodextrin (112). The catalyst captured the benzaldehydes into the cavity by hydrophobic binding, then captured the ketones with the catalyst amino group, then held them together and linked them in the product.

1.10 Hydrophobic Acceleration of Diels–Alder Reactions

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

In the benzoin condensation described earlier, we were able to bind two benzaldehydes into a γ-cyclodextrin ring, and then couple them with a thiazolium ion attached to the cyclodextrin. It had earlier occurred to us that the hydrophobic binding of two substrates in water into a cyclodextrin cavity could accelerate their reaction without the use of other catalytic groups, so we examined a group of Diels–Alder reactions (113). Indeed, we found that the Diels–Alder reaction of cyclopentadiene with butanone or acrylonitrile in water was accelerated by added β-cyclodextrin, where the two substrates could both fit.

However, we needed to determine their reaction rates in water without the cyclodextrin, as a comparison, and we discovered a remarkable fact: these Diels–Alder reactions were much faster in water than in other solvents. With β-cyclodextrin they were faster still, but with α-cyclodextrin they were slowed relative to water. Both substrates cannot fit into the smaller cavity, so the cyclodextrin is hidden from the dienophiles by binding into α-cyclodextrin.

We concluded that in the reactions using water alone, the reaction between diene and dienophiles was being promoted by the hydrophobic effect, and this was confirmed by the changes in rate produced with added substances that were known to increase, or decrease, the hydrophobic effect. We then examined other Diels–Alder reactions in water, including those of 9-hydroxymethylanthracene (114). Again, we saw large accelerations with this solvent, and in other cases we also saw a significant preference for the production of the endo product.

In that paper (114), we also reported that we saw such increased preference for endo addition even when the Diels–Alder components were present as suspensions, not solutions, in water. We pointed out that this indicated that the special hydrophobic effects of water can be seen even with rather insoluble reactants, and that this use of suspensions might be a more general and practical way to apply the water effect. We also made this point in our review describing the use of the hydrophobic effect in organic chemistry (2).

We showed that the preference for endo addition was a simple result of the hydrophobic effect, by using prohydrophobic and antihydrophobic additives (1). We saw that the effects were much smaller when “water-like” solvents were used instead of water (115), and explored the range of antihydrophobic effects further (116). As is described below, we were eventually able to use the effects of antihydrophobic additives in water to determine the geometries of a number of important transition states, including those of Diels–Alder reactions.

1.11 Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

Since the bringing together of a diene and a dienophiles in water was promoted by the hydrophobic effect, it seemed likely that this effect would also promote the reaction of a hydrophobic reagent with a hydrophobic substrate. As one example, we examined the competition in reduction of a methyl ketone with a phenyl or naphthyl ketone in water by borohydride anion, and by phenylborohydride anion and pentafluorophenylborohydride anion (Scheme 1.11) (117). The data indicated that the intrinsically less-reactive aryl ketones were selectively reduced by the arylborohydrides in water, but not in methanol. Thus in water the association of the aryl groups of reagent and substrate was hydrophobically promoted.

original image

Scheme 1.11. In water, hydrophobic borohydrides and amine-boranes selectively reduce those carbonyls accessible in hydrophobic complexes.

A striking example was seen in the reversal of selectivity in reducing a steroid diketone 28 in water (118). Simple borohydride anion selectively reduced the intrinsically more reactive cyclopentanone carbonyl, but pentafluorophenylborohydride anion preferred to reduce the other one in a transition state with the reagent perfluorophenyl ring hydrophobically packed on the steroid phenyl ring.

The synthesis of arylborohydrides is not trivial, and it turns out that there is a simpler way to link a borohydride to an aryl ring – coordinate borane with a pyridine nitrogen to form an amine-borane with a positive nitrogen and a negative boron. We showed that this kind of species 29 could also perform selective reductions in water by promoting the hydrophobic association of the aryl rings of substrates and reagents (119).

1.12 Selectivities in Water Induced by the Hydrophobic Effect – Oxidations

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

The enzymes called cytochrome P450s oxidize many substrates biochemically by binding them next to a heme group, and converting the heme to an iron–oxo species that transfers the oxygen atom from iron to substrate. One such classic biochemical reaction is the epoxidation of olefinic groups in substrates such as squalene. We set out to mimic such enzymes by attaching two or more binding groups to metalloporphyrins and salen (120). In this first work, we used 8-hydroxyquinoline binding groups to link to substrates with two pyridine or bipyridyl groups through bridging Cu(II) ions. We saw that indeed we could achieve substrate selectivity for those substrates that could use such copper bridging to the catalysts. In this work, the solvent was acetonitrile.

We then performed similar studies in water solution using attached cyclodextrins to doubly bind the hydrophobic substrates to metalloporphyrins, and again saw substrate selectivity (121). After this, we showed that such metalloporphyrins with attached cyclodextrin groups (e.g. 30) could selectively hydroxylate a saturated steroid 31 on a well-defined position, attaching a hydroxyl to the C-6 methylene group (122). When these systems were further examined, we saw that the epoxidation of olefinic substrates occurred with as many as 650 catalytic turnovers. However, the hydroxylation of the saturated carbon in the steroid substrate was slower, and the catalyst was oxidatively destroyed after only three to five catalytic turnovers (123).

The porphyrin ring was being oxidized, and work by others had suggested that this could be suppressed if the phenyl rings used to link the cyclodextrin to the porphyrin were replaced with perfluorophenyl groups. Therefore, we synthesized catalyst 32, and used it to catalyze the hydroxylation of substrate 31 in water solution (124). We saw that now the steroid was selectively hydroxylated at C-6 with 187 catalytic turnovers. Furthermore, the perfluoro derivative 32 was particularly easy to synthesize, attaching the cyclodextrin groups by selective displacement of para-fluorines on pentafluorophenyl groups.

original image

We then saw that we could move the selective hydroxylation position in the steroid to C-9 by using three reversible hydrophobic binding groups to attach a steroid to the catalyst in water (125). The product is a useful precursor to important steroids such as corticoids. Full descriptions of the extension of this work to other substrates and to other catalyst geometries were also published (126, 127). We also examined a catalyst carrying synthetic cyclic hydrophobic binding groups instead of cyclodextrins, and saw that these catalysts were chemically slightly more stable, but less selective than the cyclodextrin analogs (128). A full paper summarized all the results from such mimics of cytochrome P450 (129).

In these oxidations, we used iodosobenzene to supply oxygen atoms to the catalyst, but in Nature the oxidant is molecular oxygen itself. Also, in Nature the iron atom of cytochrome P450 has a thiolate ligand from cysteine, in addition to the four nitrogens of the porphyrin. Hence we added a thiolate from thiophenol, either by covalent attachment to the catalyst or by hydrophobic binding using the two cyclodextrins on the catalyst that were not involved in substrate binding (130). We now found that the cheap and convenient oxidant hydrogen peroxide could be used instead of iodosobenzene for the steroid oxidations.

We also returned to our previous use of metal ion bridging as a way to coordinate substrates to the metalloporphyrins. We saw that a new catalyst with Cu(II) bridges between substrate and catalyst gave higher turnovers and selectivities than we had obtained with the cyclodextrin-based catalysts (131).

In this work, we used the hydrophobic effect in water to bind substrates to catalysts, but we also examined using the hydrophobic effect to bind a reagent to a substrate, as we had done with carbonyl reductions. Houk had shown that the transition-state geometry for epoxidations of styrene-type olefins would not permit hydrophobic packing of a reagent such as perbenzoic acid onto the phenyl of the substrate (133), and we confirmed that this did not occur (134). However, he had calculated that epoxidation of styrene by phenyldioxirane would permit such phenyl–phenyl stacking in water. The same geometry permits epoxidation with hydrophobic stacking by aryl-bearing oxaziridinium ions, and by competition experiments we showed that this was indeed the case (132). We achieved good selectivities for the epoxidations of aryl-substituted olefins in water, but not in methanol, indicating that hydrophobic effects were involved.

In recently published work, we have examined the correlation of enantioselectivity in epoxidations by dioxiranes with the calculated transition-state energies in water. The preferred geometries again involve hydrophobic packing of reagent to substrate (135).

1.13 Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

Throughout this chapter, we have invoked ways in which the hydrophobic effect produces rate accelerations by lowering the energy of the transition state relative to that of the reactants. For example, in the previous section we saw that the predicted difference in transition-state geometry for epoxidation by peracids and by dioxiranes could be confirmed by our experiments, and explained why the dioxirane oxidations showed packing of a hydrophobic reagent to a hydrophobic substrate in water, whereas that did not occur with the peracids.

Hydrophobic effects can be modulated by added solutes in water. Simple salts such as lithium chloride increase the hydrophobic effect by contracting the water volume. However, antihydrophobic salts such as guanidinium chloride decrease the hydrophobic effect in water. We showed that this reflected an increased solvation of the hydrophobic surfaces by the guanidinium ion (133), rather than some effect on water structure, as had commonly been proposed. As we had described for the Diels–Alder reaction, the antihydrophobic effect is even greater if both the cation and the anion of the salts are able to solvate hydrophobic surfaces (116).

We realized that we could determine whether the transition states for reactions in water had some hydrophobic packing of non-polar components in the reactants by comparing the effects of the antihydrophobic additives on the starting materials – increasing their water solubility (134) – and on the transition states – determined from the effects of the additives on the rates of the reactions. As Figure 1.1 shows, an antihydrophobic additive will lower the energy of the reactants and will lower the energy of the transition state by the same amount if it has the same amount of solvent-exposed hydrophobic surface. However, the rate will be slowed if the transition state has hidden some hydrophobic surface. From a comparison of the solubility effect on reactants and the rate effect, we could say how much of the reactant hydrophobic surface was hidden in the transition state.

original image

Figure 1.1. Reactions in water follow the solid line energy curve, and with an added antihydrophobic material the energies are lowered in proportion to the amount of accessible hydrophobic surface in the starting material, the transition state, and the product. From solubility and rate data we can deduce how much hydrophobic surface is exposed in the transition state, the first evidence about the geometry of this elusive species.

This simple idea furnished the first way to determine the geometries of transition states by experiment, rather than by theory. In order to put it into practice, we abandoned salts, as bringing too many effects from ionic strength, and investigated antihydrophobic solvents. Our first studies were done by examining solubilities and rates in water, and then with a few percent of added ethanol. We checked the method by determining the geometries of transition states where the results were predictable. For example, we examined the effects in the Diels–Alder dimerization of cyclopentadiene and in the Diels–Alder addition of a maleimide to an anthracene (135). The results were completely consistent with simple ideas and with a quantum mechanical calculation of the transition-state geometry for the dimerization of cyclopentadiene.

We then applied our method to the cyanide-catalyzed benzoin condensation of benzaldehyde, and saw that it predicted a transition state with only partial overlap of the phenyls in the transition state (Scheme 1.12) (136). This was consistent with a transition state in which the nucleophilic mandelonitrile anion approached the electrophilic benzaldehyde from the rear of the carbonyl group. This new result was completely consistent with modern chemical ideas. In the same paper, we also described the reactions of nucleophiles such as thiophenoxide ion and N-methylaniline with the p-carboxylate of benzyl chloride (the carboxylate is there to achieve water solubility, as our method requires.) We saw that the two phenyl groups overlapped in this displacement by N-methylaniline, but not with the thiophenoxide ion. We pursued such displacements further in subsequent papers.

original image

Scheme 1.12. From this technique, we deduce that the hydrophobic surfaces of benzaldehydes are partially hidden from solvent water in the transition state for the benzoin reaction.

Our most striking result was that the displacement on p-carboxybenzyl chloride by phenoxide ion showed no evidence of hydrophobic overlap in the transition state, with no significant decrease in the rate with added ethanol (137). This indicated that the phenoxide ion is attacking with its n-electrons rather than with the electrons that are part of the π-system, so the two phenyl rings are far apart, whereas with the aniline nucleophile there is no such choice and the phenyls overlap in the transition state (Scheme 1.13). We also confirmed the lack of slowing with nucleophilic thiophenoxide ion, and proposed that this nucleophile used a single electron transfer (SET) mechanism, for which we offered some evidence. Both ideas were further pursued, and the results to date were described in a full paper (138).

original image

Scheme 1.13. From this technique, we deduce that the transition state for displacement by an aniline on a benzylic chloride in water involves substantial phenyl–phenyl overlap. However, when phenoxide ion is the nucleophile the n-electrons are used, and there is no overlap of the phenyl groups.

We saw further evidence for the SET mechanism with thiophenoxide ions (139). It was a dominant mechanism with iodide ion as the leaving group, as expected from the ease of one-electron cleavage of the carbon–iodine bond, and the SET mechanism showed remarkably low rate effects from steric hindrance of the nucleophile, as expected of the transition state for electron transfer if it could occur over longer distances than would be needed for an SN2 reaction. In a subsequent full paper, we showed that quantum mechanics predicted the geometry of phenoxide ion attack, using the n-electrons as we had concluded (140). However, we did raise the concern that with such ionic reactions it was critical to determine how important the decreased polarity of water with added ethanol could be, since that could also slow the rates. We found a way to deal with this concern in later work.

A powerful tool came from using dimethyl sulfoxide (DMSO) instead of ethanol as an additive to water (141). We saw that DMSO was even better than ethanol, on a molar basis, in solvating hydrophobic surfaces and thus increasing the solubilities of hydrophobic materials in water. However, DMSO is rather polar, and had a much smaller effect than ethanol on the dielectric constant of the water-based solvent. Using this, we were able to show that the slowing of displacement reactions indeed reflected the modulation of hydrophobic effects, not simply changes in solvent polarity.

In the same paper, we introduced a critical piece of evidence, the solvent effects on two different reactions of the same substrates. 2,6-Dimethylphenoxide ion can be alkylated by p-carboxybenzyl chloride in two different spots, on the oxygen and on the para-position. Others had studied this and saw, as we did, that only in water was there alkylation of the benzene ring, and that simple phenoxide ion showed only oxygen alkylation. They had proposed that the alkylation of the benzene ring reflected hindrance of the oxygen by the methyl groups at positions 2 and 6. Our results showed that something completely different was involved.

The alkylation of oxygen showed no slowing by ethanol or DMSO cosolvents, consistent with the idea that again there is no phenyl–phenyl overlap in the transition state for phenoxide oxygen alkylation. However, the alkylation of the phenyl ring showed significant slowing with the added cosolvents, indicating that such phenyl alkylations have transition states with phenyl overlaps, as indeed they must. The methyl groups do not slow oxygen alkylation; they add an extra hydrophobic group to promote the hydrophobic packing needed for carbon alkylation.

In a full paper, we described further results in detail (142). The most striking new result was that moving the two methyl groups to the 3- and 5-positions also caused C-alkylation that was not seen without the methyl groups. They were now next to the carbon being alkylated, away from the phenoxide ion, and this completely disproved the idea that the methyls were causing C-alkylation by shielding the oxygens. Some alkylation now occurred at C-2, not just at C-4 of the phenoxide ion, and even a single methyl group in the para-position of phenoxide ion was enough, by adding hydrophobic surface, to promote alkylation at C-2 while without that methyl group no C-alkylation occurred. Thus here the hydrophobic effect can take over a simple O-alkylation of phenoxide ion and divert a significant fraction to the phenyl ring. The use of our method to determine the geometries of transition states was described in a review (143).

1.14 Conclusion

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References

As this chapter describes, there are many reasons to use water as a solvent for chemical reactions. In manufacturing, it is both a “green” solvent and an inexpensive one. The finding that some of the effects seen in water solution can be seen also in water suspensions suggests that even insoluble compounds might be better used in water. However, the major real reason to pursue water as a solvent is that the hydrophobic effect leads to such remarkable new chemistry not otherwise achievable. We saw at least some evidence for this in a water suspension, not just a solution. Process chemists are urged to examine whether they can achieve similar special chemical results in reactions of interest while at the same time solving the environmental problems that so many other solvents produce.

References

  1. Top of page
  2. Introduction
  3. Binding of Two Species Together Driven by the Hydrophobic Effect in Water
  4. Aromatic Chlorination
  5. Acylation of Cyclodextrins by a Bound Ester
  6. Mimics of Metalloenzymes Using the Hydrophobic Effect in Water
  7. Mimics of the Enzyme Ribonuclease
  8. Mimics of Enzymes that Use Pyridoxamine Phosphate and Pyridoxal Phosphate as Coenzymes
  9. Artificial Enzymes Carrying Mimics of Thiamine Pyrophosphate
  10. Enolizations and Aldol Condensations
  11. Hydrophobic Acceleration of Diels–Alder Reactions
  12. Selectivities in Water Induced by the Hydrophobic Effect – Carbonyl Reductions
  13. Selectivities in Water Induced by the Hydrophobic Effect – Oxidations
  14. Using Hydrophobic Effects in Water to Determine the Geometries of Transition States for Some Important Reactions
  15. Conclusion
  16. References