This review provides the reader a concise overview of the majority of prodrug approaches with the emphasis on the modern approaches to prodrug design. The chemical approach catalyzed by metabolic enzymes which is considered as widely used among all other approaches to minimize the undesirable drug physicochemical properties is discussed. Part of this review will shed light on the use of molecular orbital methods such as DFT, semiempirical and ab initio for the design of novel prodrugs. This novel prodrug approach implies prodrug design based on enzyme models that were utilized for mimicking enzyme catalysis. The computational approach exploited for the prodrug design involves molecular orbital and molecular mechanics (DFT, ab initio, and MM2) calculations and correlations between experimental and calculated values of intramolecular processes that were experimentally studied to assign the factors determining the reaction rates in certain processes for better understanding on how enzymes might exert their extraordinary catalysis.
A drug is defined as a substance, which is used in the diagnosis, cure, relief, treatment, or prevention of disease, or intended to affect the structure or function of the body. The development of any potential drug starts with the study of the biochemistry behind a disease for which pharmaceutical intervention is seen.1
Drug discovery is a lengthy interdisciplinary endeavor. It is a consecutive process that starts with target and lead discovery, followed by lead optimization and preclinical in vitro and in vivo studies to evaluate whether a compound satisfies a number of preset criteria to start clinical development. The number of years it takes to introduce a drug to the pharmaceutical market is over 10 years with a cost of more than $1 billion dollars [1, 2].
Modifying the absorption, distribution, metabolism, and elimination (ADME) properties of an active drug requires a complete understanding of the physicochemical and biological behavior of the drug candidate [3-6]. This includes comprehensive evaluation of drug-likeness involving prediction of ADME properties. These predictions can be attempted at several levels: in vitro–in vivo using data obtained from tissue or recombinant material from human and preclinical species, and in silico or computational predictions projecting in vitro or in vivo data, involving the evaluation of various ADME properties, using computational approaches such as quantitative structure activity relationship (QSAR) or molecular modeling [7-11].
Studies have indicated that poor pharmacokinetics and toxicity are the most important causes of high attrition rates in the drug development process, and it has been widely accepted that these areas should be considered as early as possible in drug discovery to improve the efficiency and cost-effectiveness of the industry. Resolving the pharmacokinetic and toxicological properties of drug candidates remains a key challenge for drug developers .
Thus, the aim is to design drugs that have an efficient permeability to be absorbed into the blood circulation (absorption), to reach their target efficiently (distribution), to be quite stable to survive the physiological journey (metabolism), and to be eliminated in a satisfactory time (elimination). In other words, designing a drug with optimum pharmacokinetics properties can be achieved by implementing one or more of the following strategies:
Drug absorption is determined by the drug hydrophilic hydrophobic balance (HLB) value, which depends upon polarity and ionization. Very polar or strongly ionized drugs, having a relatively high HLB values, cannot efficiently cross the cell membranes of the gastrointestinal (GI) barrier. Hence, they are given by the intravenous (I.V.) route, but their disadvantage is being rapidly eliminated. Non-polar drugs, on the other hand, having a relatively low HLB values, are poorly soluble in aqueous media and hence are poorly absorbed through membranes. If they are given by injection, most probably, they will be retained in fat tissues [13-21].
Generally, the polarity and/or ionization of drug can be altered by changing its substituents, and these changes are classified under the so-called quantitative structure–activity relationships (QSAR). The following are examples for such changes:  variation of alkyl or acyl substituents and polar functional groups to vary polarity,  variation of N-alkyl substituents to vary pKa; acidic drugs with low pKa and basic drugs with high pKa values tend to be ionized and are poorly absorbed through membrane tissues,  variation of aromatic substituents to vary pKa: The pKa of aromatic amine or carboxylic acid can be varied by adding electron donating or electron withdrawing groups to the ring. The position of the substituent is important too if the substituent interacts with the ring through resonance and  bioisosteres for polar groups; carboxylic acid is a highly polar group which can be ionized and hence decreases the absorption of any drug containing it. To overcome this problem, blocking the free carboxyl group by making the corresponding ester prodrug or replacing it with a bioisostere group, which has similar physiochemical properties and has advantage over carboxylic acid in regard to its pKa, such as 5-substituted tetrazoles, is essential; 5-substituted tetrazole ring contains acidic proton such as carboxylic acid and is ionized at pH 7.4. On the other hand, most of the alkyl and aryl carboxylic groups have a pKa in the range of 2–5 [13-28].
There are different strategies that can be utilized to improve drug metabolism: (i) steric shields: Some functional groups are more susceptible to chemical and enzymatic degradation than others. For example, esters and amides are much more affordable to hydrolysis than others such as carbamates and oximes. Adding steric shields to these drugs increases their stability. Steric shields were designed to hinder the approach of a nucleophile or a nucleophilic center on an enzyme to the susceptible group. These usually involve the addition of a bulky alkyl group such as t-butyl close to the functional group. (ii) Electronic effects of bioisosteres: This approach is used to protect a labile functional group by electronic stabilization. For example, replacing the methyl group of an ester with an amine group gives a urethane functional group, which is more stable than the parent ester. The amine group has the same size and valance as the methyl group; however, it has no steric effect, but it has totally different electronic properties, because it can donate electrons via its inductive effect into the carbonyl group resulting in reducing the electrophilicity of the carbonyl carbon and hence stabilizing it from hydrolysis.
Carbachol (1 in Figure 1), a cholinergic agonist, and cefoxitin (2 in Figure 1), a cephalosporin, are stabilized in this way. (iii) Stereoelectronic modification: Steric hindrance and electronic stabilization have been used together to stabilize labile groups. For example, procaine, an ester drug, is quickly hydrolyzed, but changing the ester to the less reactive amide group reduces hydrolysis such as in the cases of procainamide (3 in Figure 1) and lidocaine (4 in Figure 1). (iv) Metabolic Blockers: Some drugs are metabolized by introducing polar functional groups at particular positions in their skeleton. For example, megestrol acetate (5 in Figure 1), an oral contraceptive, is oxidized at position 6 to give hydroxyl group at this position; however, replacing the hydrogen at position 6 with a methyl group blocks its metabolism, and consequently it results in prolonging its duration of action. (v) Removal of susceptible metabolic groups: Certain chemical moieties are particularly susceptible to metabolic enzymes. For example, a methyl group on aromatic rings is often oxidized to carboxylic acid, which then results in a rapid elimination of the drug from the body. Other common metabolic reactions include aliphatic and aromatic C-hydroxylation, O and S-dealkylations, N- and S-oxidations, and deamination. (vi) Group Shifts: Removing or replacing a metabolically vulnerable group is feasible if the group concerned is not involved in important binding interactions within the active site of the receptor or enzyme. If the group is important, then different strategy either masking the vulnerable group using a prodrug or shifting the vulnerable group within the molecule skeleton is undertaken. Salbutamol was developed in 1969 from its analog neurotransmitter, norepinephrine, using this tactic. Norepinephrine is metabolized by methylation of one of its phenolic groups by catechol O-methyl transferase. The other phenolic group is important for receptor-binding interaction. Removing the hydroxyl or replacing it with a methyl group prevents metabolism but also prevents hydrogen bonding interaction with the binding site. While moving the vulnerable hydroxyl group out from the ring by one carbon unit as in salbutamol makes, this compound unrecognizable by the metabolic enzyme, but not to the receptor-binding site (prolonged action) and (vii) ring variation; some ring systems are often found to be susceptible to metabolism, and so varying the ring can often improve metabolic stability. For example, replacement of the imidazole ring, which is susceptible to metabolism in tioconazole (6 in Figure 1) with 1,2,4-triazole ring, gives fluconazole (7 in Figure 1) with improved stability.
Making drug less resistance to drug metabolism: drug that is extremely stable to metabolism and is very slowly eliminated can cause problems in a similar manner to that susceptible to metabolism, thus resulting in an increase in toxicity and adverse effects. Therefore, designing drugs with decreased chemical and metabolic stability can sometimes be beneficial. Methods for applying such strategy are (i) introducing groups that are susceptible to metabolism is a good way of shorting the lifetime of a drug. For example, methyl group was introduced to some drugs to shorten their lifetime because methyl can metabolically undergo oxidation to polar alcohol as well as to a carboxylic acid. (ii) A self-destruct drug is one that is chemically stable under one set of conditions but becomes unstable and spontaneously cleaves under another set of conditions. The advantage of a self-destruct drug is that inactivation does not depend on the activity of metabolic enzymes, which could vary from patient to patient. For example, atracurium, a neuromuscular blocking agent, is stable at acidic pH but self-destructs when it is exposed to the slightly alkaline conditions of the blood (pH 7.4). Thus, the drug has a short duration of action, allowing anesthetists to control its blood concentration levels during surgery by providing it as a continuous intravenous drip [22-28].
It is often found that a drug fails clinical trials because of its toxic adverse effects.
This may be due to toxic metabolites, in which case the drug should be made more resistant to metabolism. It is known that functional groups such as aromatic nitro groups, aromatic amines, bromoarenes, hydrazines, hydroxylamines, or polyhalogenated groups are generally metabolized to toxic metabolites.
Side-effects might be reduced or eliminated by varying harmless substituents. For example, addition of fluorine group to UK 47265, antifungal agent, gives the less toxic fluconazole [29-34].
Prodrugs Catalyzed by Metabolic Enzymes
The principle of targeting drugs can be traced back to Paul Ehrlich who developed antimicrobial drugs that were selectively toxic for microbial cells over human cells. Today, targeting tumor cells is considered one of the most important issues that under concern among the health community. A major goal in cancer chemotherapy is to target drugs efficiently against tumor cells rather than against normal cells. One method for achieving this is to design drugs which make use of specific molecular transport systems. The idea is to attach the active drug to an important building block molecule that is needed in large amounts by the rapidly divided tumor cells. This could be an amino acid or a nucleic acid base such as uracil mustard. In the cases where the drug is intended to target against infection of gastrointestinal tract (GIT), it must be prevented from being absorbed into the blood supply. This can easily be done using a fully ionized drug which is incapable of crossing cell membrane barriers. For example, highly ionized sulfonamides are used against GIT infections because they are incapable of crossing the gut wall. It is often possible to target drugs such that they act peripherally and not in the central nervous system (CNS). By increasing the polarity of drugs, they are less likely to cross the blood–brain barrier, and thus, they are less likely to have CNS adverse effects [35-46].
The most efficient approach for overcoming the negative pharmacokinetics characteristics of a drug is the prodrug approach. This approach may be utilized in the cases where the use of the parent drug faces problems associated with solubility, absorption and distribution, site specificity, instability, toxicity, poor patient compliance, or formulation problems [47-52]. A metabolic enzyme is usually involved in converting the prodrugs to their active forms. Not all prodrugs are activated by metabolic enzymes. For example, photodynamic therapy involves the use of an external light source to activate prodrugs. When designing a prodrug, it is important to ensure that the prodrug is effectively converted to the active drug once it has been absorbed in blood supply. It is also important to ensure that any groups that are cleaved from the prodrug molecule are non-toxic .
The prodrug approach is a very versatile strategy to increase the utility of biologically active compounds, because one can optimize any of the ADME properties of potential drug candidates. In most cases, prodrugs contain a promoiety (linker) that is removed by an enzymatic or chemical reaction, while other prodrugs release their active drugs after molecular modification such as an oxidation or reduction reaction. The prodrug candidate can also be prepared as a double prodrug, where the second linker is attached to the first promoiety linked to the parent drug molecule. These linkers are usually different and are cleaved by different mechanisms. In some cases, two biologically active drugs can be linked together in a single molecule called a codrug. In a codrug, each drug acts as a linker for the other [8, 9]. The prodrug approach has been used to overcome various undesirable drug properties and to optimize clinical drug application. Recent advances in molecular biology provide direct availability of enzymes and carrier proteins, including their molecular and functional characteristics.
There are two major prodrug design approaches that are considered as widely used among all other approaches to minimize or eliminate the undesirable drug physicochemical properties while maintaining the desirable pharmacological activity. The first approach is the targeted drug design approach by which prodrugs can be designed to target specific enzymes or carriers by considering enzyme–substrate specificity or carrier–substrate specificity to overcome various undesirable drug properties. This type of ‘targeted-prodrug’ design requires considerable knowledge of particular enzymes or carriers, including their molecular and functional characteristics [35-46].
The second approach to be discussed in this review is subdivided to two major approaches: (i) Chemical approach by which the drug is linked to promoiety which upon exposure to physiological environment undergoes enzymatic catalyzed degradation to the parent drug and inactive linker. In this approach, the interconversion rate is dependent on the enzyme catalysis. This approach involves carrier-linked prodrugs and contains a group that can be easily removed enzymatically, such as an ester or labile amide, to provide the parent drug. Ideally, the group removed is pharmacologically inactive and non-toxic, while the linkage between the drug and promoiety must be labile for in vivo efficient activation. Carrier-linked prodrugs can be further subdivided into (i) bipartite, which is composed of one carrier group attached to the drug, (ii) tripartite, which is a carrier group that is attached via linker to drug, and (iii) mutual prodrugs consisting of two drugs linked together, and  bioprecursors are chemical entities that are metabolized into new compounds that may be active or further are metabolized to active metabolites, such as amine to aldehyde to carboxylic acid [48-51]; and (ii) intramolecular chemical approach designed based on calculations using molecular orbital (MO) and molecular mechanics (MM) methods and correlations between experimental and calculated values. In this prodrug approach, no enzyme is involved in the intraconversion chemical reaction of a prodrug to its parent drug. The interconversion of the prodrug is solely controlled by the rate-limiting step of the intramolecular reaction.
The prodrug design can be utilized in the followings cases: (i) enhancing active drug solubility in a physiological environment/s and consequently its bioavailability since dissolution of the drug molecule from the dosage form may be the rate-limiting step to absorption . It has been documented that more than 30% of drug discovery compounds have poor aqueous solubility . Prodrugs are an alternative way to increase the aqueous solubility of the parent drug molecule by increasing dissolution rate through attachment to ionizable or polar groups, such as phosphates, sugar, or amino acids moieties [51, 54]. These prodrugs can be used for increasing oral bioavailability and in parenteral or injectable drug delivery. (ii) Upon increasing permeability and hence absorption, membrane permeability has a significant effect on drug effectiveness . In oral drug delivery, the most common absorption routes are un-facilitated and largely non-specific, passive transport mechanisms. The lipophilicity of poorly permeable drugs can be enhanced by linking to lipophilic groups. In such cases, the prodrug strategy can be an extremely valuable option and crucially needed. Improvements in lipophilicity have been the most widely researched and successful field of prodrug research. It has been achieved by masking polar ionized or non-ionized functional groups to increase either oral or topical absorption . (iii) Modification of the distribution profile: Before the drug reaches its physiological target and exert the desired effect, it has to bypass several pharmaceutical and pharmacokinetic barriers. Today, one of the most promising site-selective drug delivery strategies is the prodrug approach, which utilizes target cell- or tissue-specific endogenous enzymes and transporters.
The suitability of a number of functional groups such as carboxylic, hydroxyl, amine, phosphate, phosphonate, and carbonyl groups for undergoing different chemical modifications facilitates their utilization in prodrug design [1, 9] In the past few decades, a variety of prodrugs based on the chemical approach have been designed, synthesized, and tested. Among those are the following.
carboxylic acid, hydroxyl, phosphate, and thiol groups can easily undergo hydrolysis via the enzymatic catalysis of esterases and phosphatases that are present in many places in the body including liver, blood, and other tissues, or via oxidative cleavage catalyzed by cytochrome P450 enzymes (CYP) [51, 55, 56].
Carboxyl esterases, acetylcholinesterases, butyrylcholinesterases, paraoxonases, arylesterases and biphenyl hydrolase-like protein (BPHL) are examples of enzymes that are responsible for the hydrolytic bioactivation of ester prodrugs . For example, biphenyl hydrolase-like protein (BPHL) is known to catalyze the hydrolysis of prodrugs such as valacyclovir (8 in Figure 2) and valganciclovir (9 in Figure 2), as well as a number of other amino acid esters of nucleoside analogs including valyl-AZT, prodrugs of floxuridine (5-fluoro-20-deoxyuridine or FUdR) (10 in Figure 2) and gemcitabine (11 in Figure 2) . Ester prodrugs are commonly used to enhance lipophilicity, thus increasing membrane permeation through masking the charge of polar functional groups and by handling the alkyl chain length and configuration . For example, acyclovir aliphatic ester prodrugs were prepared by an esterification of the hydroxyl group with lipophilic acid anhydride or acyl chloride ; hence, an enhanced lipophilicity can be achieved. Utilizing the lipophilic ester approach, some acyclovir prodrugs were synthesized and have shown an enhanced nasal and skin absorption . It has been shown that an increase in the length of an alkyl chain results in a relatively ease cleavage of the ester bond. Therefore, it might be concluded that improved binding to the hydrophobic pocket of carboxylesterase can be accomplished by increasing the length of the ester alkyl chain, while branching the alkyl chain might result in reduced hydrolysis due to a steric hindrance . Further, eighteen amino acid esters of acyclovir were synthesized as potential prodrugs intended for oral administration, and their hydrolytic reaction was shown to be catalyzed by biphenyl hydrolase . Acyclic nucleosides such as adefovir and tenofovir are monophosphorylated and do not rely on viral nucleoside kinases for initial activation; hence, they have low oral bioavailability due to the ionized phosphonate group [60, 61]. To overcome this problem, a bis(pivaloyloxymethyl) ester prodrug of adefovir (adefovir dipivoxil) and ether lipid ester prodrugs of cidofovir were explored for improving intestinal permeability . Other examples for ester prodrugs that were designed and synthesized for different purposes are thioester of erythromycin, palmitate ester of clindamycin, a number of angiotensin-converting enzyme (ACE) inhibitors which are  presently marketed as ester prodrugs, including enalapril (12 in Figure 2), ramipril, benazepril, and fosinopril, and all of them are intended for the treatment for hypertension  and ibuprofen guaiacol ester that was reported to have fewer GI side-effects with similar anti-inflammatory/antipyretic action to its parent drug when is given in equimolar doses . As mentioned before, two active drugs can be joined together such that each one behaves as a carrier moiety for the other, a strategy known as mutual prodrugs . The followings are some examples of mutual prodrugs, based on ester linkage, that were produced to overcome several shortcomings associated with therapeutic drugs used in clinical practice: Benorylate (13 in Figure 2) is a mutual prodrug of aspirin and paracetamol, coupled through an ester linkage, which is postulated to have reduced gastric irritancy with synergistic analgesic effect . Moreover, mutual prodrugs of ibuprofen with paracetamol and salicylamide have been reported to have better lipophilicity and diminished gastric toxicity than the parent drug. Another example is naproxen-propyphenazone which was synthesized to prevent GI irritation and bleeding . An alternative strategy to avoid GI side-effects is by conjugation of a nitric oxide (NO) releasing moiety to the parent NSAID drug. It has been reported that NO plays a gastro protective role along with prostaglandins . Some NO-releasing organic nitrate esters of aspirin, diclofenac, naproxen, ketoprofen, flurbiprofen, and ibuprofen have been reported to give the corresponding active parent drugs with lower gastro toxicity [66, 67]. Reduce gastro toxicity could be achieved also by linking NSAID drug with histamine H2 antagonist such as in the case of flurbiprofen-histamine H2 antagonist conjugates . The mutual prodrug approach was also applied to other therapeutic groups. For instance, sultamicillin (14 in Figure 2) in which the irreversible β-lactamase inhibitor sulbactam has been linked via an ester linkage with ampicillin has shown a synergistic effect , and upon oral administration, sultamicillin is completely hydrolyzed to equimolar proportions of sulbactam and ampicillin, thereby acting as an efficient mutual prodrug .
This approach can be exploited to enhance the stability of drugs, provide targeted drug delivery, and change lipophilicity of drugs such as acids and acid chlorides . Drugs that have carboxylic acid or amine group can be converted into amide prodrugs.Generally, they are used to a limited extent due to high in vivo stability. However, prodrugs using facile intramolecular cyclization reactions have been exploited to overcome this obstacle .
Similar to mutual ester prodrugs, there are some mutual prodrugs where the two active drugs are linked together by an amide linkage, such as atorvastatin and amlodipine which upon in vivo amide hydrolysis provide the corresponding active parent drugs. Amide prodrugs can be converted back to the parent drugs either by nonspecific amidases or by specific enzymatic activation such as renal γ-glutamyl transpeptidase. Dopamine double prodrug γ-glutamyl-l-dopa (gludopa) (15 in Figure 2) undergoes specific activation by renal γ-glutamyl transpeptidase where it achieves relatively fivefold increase in dopamine level compared with L-dopa prodrug. However, as gludopa has low oral bioavailability, dopamine [N-(N-acetyl-L-methionyl)-O,O-bis(ethoxycarbonyl)dopamine), a pseudopeptide prodrug of dopamine, was developed and has shown improved oral absorption; hence, it is given orally and is used in the treatment for renal and cardiovascular diseases. Basically, dopamine prodrugs are developed due to dopamine inactivation by COMT and MAO when administered by the oral route [71, 72].
A respected number of amine conjugates with amino acids through amide linkage have been considered for providing active drugs with remarkable enhancement in solubility such as dapsone (16 in Figure 2) .
Other examples of amide based prodrugs are allopurinol N-acyl derivatives which were found to be more lipophilic than allopurinol itself .
Carbonates and Carbamates Prodrugs
Generally, carbonates and carbamates are more stable than esters but less stable than amides . Carbamates and carbonates have no specific enzymes for their hydrolysis reactions; however, they are degraded by esterases to give the corresponding active parent drugs [75, 76]. Co-carboxymethylphenyl ester of amphetamine is an example of carbamates prodrug that can be hydrolyzed by esterase to yield amphetamine . Carbamates prodrugs are regarded as double prodrugs (pro-prodrug) because they are enzymatically activated at first which is followed by spontaneous cleavage of the resulting carbamic acid . An example for such prodrugs is fluorenylmethoxycarbonyl]-3 derivatives of insulin and exenatide  that undergo slow interconversion via carbamate bond breakdown, thus providing glucose controlling agents in an adequate rate which consequently results in lowering the risk of hypoglycemia .
Another example of carbamates prodrug is the one obtained by linking phosphorylated steroid, an estradiol, to normustard, an alkylating agent, through a carbamate linkage, which yields estramustine prodrug. The latter is used in the treatment for prostate cancer. The steroid portion has an antiandrogenic action and acts to concentrate the prodrug in the prostate gland where prodrug hydrolysis takes place and normustard action can then be exerted . Carbamates prodrugs can also be used to increase the solubility of active drugs such as cephalosporins . In addition, carbamates prodrugs have been exploited in targeted therapy such as ADEPT. In this case, the carbamate group is susceptible to the action of tyrosinase enzyme present in melanomas. This approach is usually utilized in cancer targeted therapy . The list of carbamates prodrugs is long; among other examples is the non-sedating antihistamine loratadine (17 in Figure 2), an ethylcarbamate, that undergoes in vivo interconversion to its active form, desloratidine, through the action of CYP450 enzymes , and capecitabine, an anticancer agent, that undergoes a multistep activation, to finally yield 5-fluorouracil in the liver. Capecitabine is less toxic than 5-fluorouracil, more selective, and widely used in clinical practice [81-83].
These prodrugs serve to increase the permeability of the corresponding active drugs, and they are converted back to their parent drugs by microsomal cytochrome P450 enzymes (CYP450) . Dopaminergic prodrug 6-(N,N-Di-n-propylamino)-3,4,5,6,7,8-hexahydro-2H-naphthalen-1-one is an example for such class .
N-Mannich Bases, Enaminones, and Schiff Bases (Imines)
N-Mannich base formation is another approach, which can be utilized to enhance drug's solubility. N-Mannich bases are prepared by Mannich reaction that involves reacting of NH-acidic compound, an aldehyde and an amine in ethanol . Rolitetracycline (18 in Figure 3) is the Mannich derivative of tetracycline, and it is the only one available for intravenous administration . N-Mannich bases of dipyrone, metamizole (19 in Figure 3), the methane sulfonic acid of the analgesic 4-(methylamino) antipyrine is water soluble and suitable for parenteral route, and when given orally, it is hydrolyzed in the stomach to give the parent active drug . Despite the success of N-Mannich base prodrugs to improve bioavailability of active drugs, there still some stability formulation problems arise from poor in vitro stability of some of the prodrugs . In addition, the in vivo formation of formaldehyde upon enzymatic breakdown  of these prodrugs is considered a limitation of this prodrug approach. Enamines  (α,β-unsaturated amines) are unstable at low pH, which results in their limitation for use in oral administration . Nonetheless, an ampicillin prodrug based on enamines was prepared for rectal use, and it exhibits an increased absorption compared with its active parent drug . Enaminones are enamines of β-dicarbonyl compounds that undergo ketoenolimine-enamine tautomeric equilibrium, which may offer stability to these compounds . Enaminones are generally more lipophilic than their parent drugs; hence, they have an improved oral absorption. Typically, enaminones have a relatively high chemical stability; therefore, their use as potential prodrugs is being somewhat limited. It is expected that enaminones derived from ketoesters and lactone may be subjected to enzymatic degradation; hence, a better conversion rate to the active drug can be obtained .
Phosphate and Phosphonate Prodrugs
Phosphorylation offers increased aqueous solubility to the parent drugs. A traditional example of phosphate prodrugs is prednisolone sodium phosphate, a water-soluble prodrug of prednisolone, its water solubility exceeds that of its active form, prednisolone, by 30 times , it is often used as an immunosuppressant, and it is formulated as a liquid dosage form . Another common phosphate prodrug is fosamprenavir (20 in Figure 3). Similar to prednisolone, the phosphate promoiety in fosamprenavir is linked to a free hydroxyl group, and the prodrug is 10-fold more water soluble than amprenavir. An enhanced patient compliance is achieved when using this antiviral prodrug; instead of administering the drug 8 times daily, dosage regimen is reduced into two times per day . In the gut and via the action of alkaline phosphatases, phosphate prodrugs are cleaved back to their corresponding active drugs and then absorbed into the systemic circulation . Another application of this approach is fosphenytoin (21 in Figure 3), a prodrug of the anticonvulsant agent phenytoin. Fosphenytoin has an enhanced solubility over its corresponding drug .
Colonic bacteria can be exploited in prodrug approach as a means of prodrug activation through the action of azo-reductases; this approach is applied specially in targeted drug strategy . Sulfasalazine (22 in Figure 3), used in the treatment for ulcerative colitis , is a prodrug of 5-aminosalicylic acid and sulfapyridine. Upon reaching the colon, sulfasalazine undergoes azo bond cleavage to release the active parent drug . Osalazine (23 in Figure 3), a dimer of 5-aminosalicylic acid, balsalazide, and ipsalazide in which 5-aminosalicylic acid moiety is conjugated to 4-aminobenzoyl-β-alanine and 4-aminobenzoylglycine, respectively , are other examples of prodrugs that are activated by azo-reductases. A prodrug by which 5-aminosalicylic acid is linked to L-aspartic acid is another example for such class that has shown a desirable colon-specific delivery and a 50% release of 5-aminosalicylic acid from an administered dose . Usually this approach is limited to aromatic amines, because azo compounds of aliphatic amines exhibit significant instability .
Poly Ethylene Glycol (PEG) Conjugates
PEG can be linked to drugs either to increase drug solubility or to prolong drug plasma half-life ; an ester, carbamate, carbonates, or amide spacer can be used to link the drug to PEG. Upon enzymatic breakdown of the spacer, the resultant ester or carbamate drug can be liberated by 1,4- or 1,6-benzyl elimination . Daunorubicin conjugated to PEG is an example of this kind of prodrugs. In this prodrug system, PEG is conjugated to the phenol group of the open lactone via a spacer. Controlling the rate of the free drug release can be accomplished by manipulation of the substituents on the aromatic ring .
The prodrug chemical approach involving enzyme catalysis is perhaps the most unpredicted approach, because there are many intrinsic and extrinsic factors that can affect the bioconversion mechanisms. For example, the activity of many prodrug-activating enzymes may be changed due to genetic polymorphisms, age-related physiological changes, or drug interactions, leading to adverse pharmacokinetic, pharmacodynamics, and clinical effects. In addition, there are wide interspecies variations in both the expression and function of most of the enzyme systems activating prodrugs which could lead to serious challenges in the preclinical optimization phase [3-6].
Prodrugs Based on Intramolecular Processes (Enzyme Models)
The novel prodrug approach to be discussed in this section implies prodrug design based on enzyme models (mimicking enzyme catalysis) that have been advocated to understand how enzymes work. The tool used in the design is a computational approach consisting of calculations using a variety of different molecular orbital and molecular mechanics methods and correlations between experimental and calculated rate values (activation energies) for some intramolecular processes that were utilized to understand the mechanism by which enzymes might exert their high catalysis. In this approach, no enzyme is needed for the catalysis of the intraconversion of a prodrug to its active parent drug. The release rate of the prodrug to the active drug is solely determined by the factors affecting the rate-limiting step of the intraconversion process. Knowledge gained from the mechanisms of the previously studied enzyme models was used in the design.
It is worth noting that the use of this approach might eliminate all disadvantages that are concerned with prodrug interconversion by enzymes approach. As mentioned in the introduction, the bioconversion of prodrugs has many disadvantages related to many intrinsic and extrinsic factors that can affect the process. For instance, the activity of many prodrug-activating enzymes may be varied due to genetic polymorphisms, age-related physiological changes, or drug interactions, leading to variation in clinical effects. In addition, there are wide interspecies variations in both the expression and function of the major enzyme systems activating prodrugs, and these can pose some obstacles in the preclinical optimization phase.
Intramolecular Processes (Enzyme Models) Used for the Design of Potential Prodrugs
Studies of enzyme mechanisms by Bruice and Benkovic, Jencks, Menger, Kirby, Walsh, and Bender, over the past five decades, have had a tremendous contribution to better understanding the mode and scope by which enzymes catalyze biochemical transformations [98-101].
Nowadays, the scientific community has reached a consensus that the catalysis by enzymes is based on the combined effects of the catalysis by functional groups and the ability to reroute intermolecular reactions through alternative pathways by which substrates can bind to preorganized active sites.
The rates for most of enzymatic reactions exceed 1010–1018 -fold the nonenzymatic bimolecular counterparts. For example, reactions catalyzed by the enzyme cyclophilin are accelerated by 105, and those by orotidine monophosphate decarboxylase are enhanced by 1017 .
In the last 50 years, as mentioned earlier, scholarly studies have been carried out by Bruice , Cohen , Menger , Kirby , and others  to design chemical models that have the capability to reach rates comparable to that with enzyme-catalyzed reactions. Frequently cited examples of such models are those based on rate acceleration driven by covalently enforced proximity. The most quoted example is Bruice et al.'s. intramolecular ring-closing reaction of dicarboxylic semi-esters to anhydrides . Studying this model, Bruice et al. has shown that a relative rate of anhydride formation can reach 5 × 107 upon the intramolecular ring-closing reaction of a dicarboxylic semi-ester when compared to a similar counterpart's intermolecular reaction.
Other examples of rate acceleration based on proximity orientation include  systems that obey the principles of Koshland's ‘orbital steering’ theory  that signifies the importance of the ground state angle of attack value of the hydroxyl in hydroxycarboxylic acids on the intramolecular lactonization reaction rate;  the ‘spatiotemporal hypothesis’ advocated by Menger, which implies that a type of a reaction, in proton transfer processes, whether intermolecular or intramolecular, is significantly determined by the distance between the two reactive centers involved in the hydroxycarboxylic acids lactonization reaction ;  the stereopopulation control proposed by Cohen to explain the relatively high enhancement rates in the acid-catalyzed lactonization reactions of hydroxyhydrocinnamic acids containing two methyl groups on the β position of their carboxylic groups  and Kirby's proton transfer models on the acid-catalyzed hydrolysis of acetals and maleamic acid amides which demonstrate the importance of hydrogen bonding formation in the products and transition states leading to them.
In the past 15 years, some prodrugs based on hydroxyhydrocinnamic acids have been introduced. For example, Borchardt et al. reported the use of the 3-(2′-acetoxy-4′, 6′-dimethyl dimethyl)-phenyl-3, 3-dimethylpropionamide derivative (pro–prodrug) that is capable of releasing the biologically active amine (drug) upon acetate hydrolysis by enzyme triggering. Another successful example of the pharmaceutical applications for a stereopopulation control model is the prodrug Taxol which enhances the drug water solubility and hence affords it to be administered to the human body via intravenous (I.V.) injection. Taxol is the brand name for paclitaxel, a natural diterpene, approved in the USA for use as anticancer agent .
Calculation Methods Used in the Prodrugs Design
In the past six decades, the use of computational chemistry for calculating molecular properties of ground and transition states has been a progressive task of organic, bioorganic, and medicinal chemists alike. Computational chemistry uses principles of computer science to assist in solving chemical problems. It uses the theoretical chemistry results, incorporated into efficient computer programs, to calculate the structures and physical and chemical properties of molecules.
Reaction rates and equilibrium energy-based calculations for biological systems that have pharmaceutical and biomedicinal interests are a very important challenge to the health community. Nowadays, quantum mechanics (QM), such as ab initio, semi-empirical, and density functional theory (DFT), and molecular mechanics (MM) are increasingly being used and broadly accepted as reliable tools for providing structure-energy calculations for an accurate prediction of potential drugs and prodrugs alike .
These methods cover both static and dynamic situations. In all cases, the computer time and other resources (such as memory and disk space) increase rapidly with the size of the system being studied. Ab initio methods typically are feasible only for small systems. Ab initio methods are based entirely on theory from first principles. The term ab initio was first used in quantum chemistry by Robert Parr and coworkers, including David Craig in a semiempirical study on the excited states of benzene. The ab initio molecular orbital methods (quantum mechanics) such as HF, G1, G2, G2MP2, MP2, and MP3 are based on rigorous use of the Schrodinger equation with a number of approximations. Ab initio electronic structure methods have the advantage that they can be made to converge to the exact solution, when all approximations are sufficiently small in magnitude and when the finite set of basic functions tends toward the limit of a complete set. The convergence, however, is usually not monotonic, and sometimes the smallest calculation gives the best result for some properties. The disadvantage of ab initio methods is their computational cost. They often take enormous amounts of computer time, memory, and disk space [110-112].
Other less accurate methods are called empirical or semi-empirical because they employ experimental results, often from acceptable models of atoms or related molecules, to approximate some elements of the underlying theory. Among these methods, the semi-empirical quantum chemistry methods are based on the Hartree–Fock formalism, but make many approximations and obtain some parameters from empirical data. They are very important in computational chemistry for treating large molecules where the full Hartree–Fock method without the approximations is too expensive. Semi-empirical calculations are much faster than their ab initio counterparts. Their results, however, can be very wrong if the molecule being computed is not similar enough to the molecules in the database used to parameterize the method. Among the most used semiempirical methods are MINDO, MNDO, MINDO/3, AM1, PM3, and SAM1. The semi-empirical methods have afforded vast information for practical application [113-116]. Calculations of molecules exceeding 60 atoms can be made using such methods.
Another commonly used quantum mechanical modeling method in physics and chemistry to investigate the electronic structure (principally the ground state) of many-body systems, in particular atoms, molecules, and the condensed phases, is the density functional theory (DFT). With this theory, the properties of many-electron systems can be determined using functionals, that is, functions of another function, which in this case is the spatially dependent electron density. Hence, the name density functional theory comes from the use of functionals of the electron density. DFT is among the most popular and versatile methods available in condensed-matter physics, computational physics, and computational chemistry. The DFT method is used to calculate structures and energies for medium-sized systems (30–60 atoms) of biological and pharmaceutical interest and is not restricted to the second row of the periodic table .
Despite recent improvements, there are still difficulties in using density functional theory to properly describe intermolecular interactions, especially van der Waals forces (dispersion), charge transfer excitations, transition states, global potential energy surfaces, and some other strongly correlated systems. Its incomplete treatment of dispersion can adversely affect the accuracy of DFT in the treatment of systems which are dominated by dispersion. The development of new DFT methods designed to overcome this problem, by alterations to the functional or by the inclusion of additive terms, is a current research topic.
On the other hand, molecular mechanics is a mathematical approach used for the computation of structures, energy, dipole moment, and other physical properties. It is widely used in calculating many diverse biological and chemical systems such as proteins, large crystal structures, and relatively large solvated systems. However, this method is limited by the determination of parameters such as the large number of unique torsion angles present in structurally diverse molecules .
Ab initio is an important tool to investigate functional mechanisms of biological macromolecules based on their 3D and electronic structures. The system size, which ab initio calculations can handle, is relatively small despite the large sizes of biomacromolecules surrounding solvent water molecules. Accordingly, isolated models of areas of proteins such as active sites have been studied in ab initio calculations. However, the disregarded proteins and solvent surrounding the catalytic centers have also been shown to contribute to the regulation of electronic structures and geometries of the regions of interest.
To overcome these discrepancies, quantum mechanics/molecular mechanics (QM/MM) calculations are utilized, in which the system is divided into QM and MM regions where QM regions correspond to active sites to be investigated and are described quantum mechanically. MM regions correspond to the remainder of the system and are described molecular mechanically. The pioneer work of the QM/MM method was accomplished by Warshel and Levitt [119, 127], and since then, there has been much progress on the development of a QM/MM algorithm and applications to biological systems [120, 121].
Similarly to that utilized for drug discovery, modern computational methods based on QM and MM methods could be exploited for the design of innovative prodrugs for drugs containing different functional groups such as hydroxyl, phenol, or amine. For example, mechanisms of intramolecular processes for a respected number of enzyme models that have been previously studied by others to understand enzyme catalysis have been recently computed by us and used for the design of some novel prodrug linkers [122-140]. Using DFT, molecular mechanics, and ab initio methods, numerous enzyme models were explored for assigning the factors governed the reaction rate in such models. Among the enzyme models that have been studied are (i) proton transfer between two oxygens and proton transfer between nitrogen and oxygen in Kirby's acetals [141-148], (ii) intramolecular acid-catalyzed hydrolysis in N-alkylmaleamic acid derivatives [141-148], (iii) proton transfer between two oxygens in rigid systems as investigated by Menger [149-152], (iv) acid-catalyzed lactonization of hydroxy-acids as researched by Cohen [104, 153, 154] and Menger [149-152], and (v) SN2-based cyclization as studied by Brown , Bruice [156, 157], and Mandolini . Our recent studies on intramolecularity have demonstrated that there is a necessity to further explore the reaction mechanisms for the above-mentioned processes for determining the factors affecting the reaction rate. Unraveling the reaction mechanism would allow for better design of an efficient chemical device to be utilized as a prodrug linker that can be covalently linked to a drug which can chemically, but not enzymatically, be cleaved to release the active drug in a programmable manner. For example, studying the mechanism for a proton transfer in Kirby's acetals has led to a design and synthesis of novel prodrugs of aza-nucleosides for the treatment for myelodysplastic syndromes , atovaquone prodrugs for the treatment for malaria , less bitter paracetamol prodrugs to be administered to children and elderly as antipyretic and pain killer , and prodrugs of phenylephrine as decongestant . In these examples, the prodrug moiety was linked to the hydroxyl group of the active drug such that the drug-linker moiety (prodrug) has the potential to interconvert when exposed into physiological environments such as stomach, intestine, and/or blood circulation, with rates that are solely dependent on the structural features of the pharmacologically inactive promoiety (Kirby's enzyme model). Other different linkers such as Kirby's maleamic acid amide enzyme model was also explored for the design of a number of prodrugs such as tranexamic acid for bleeding conditions, acyclovir as antiviral drug for the treatment for herpes simplex , atenolol for treating hypertension with enhanced stability and bioavailability  and statins for lowering cholesterol levels in the blood . In addition, prodrugs for masking the bitter taste of antibacterial drugs such as cefuroxime were also designed and synthesized [166-171]. The role of the linkers in the antibacterial prodrugs such as cefuroxime prodrugs was to block the free amine, which is responsible for the drug bitterness, and to enable the release of the drug in a controlled manner. Menger's Kemp acid enzyme model was utilized for the design of dopamine prodrugs for the treatment for Parkinson's disease as well . Prodrugs for dimethyl fumarate for the treatment psoriasis was also designed, synthesized and studied .
Computationally Designed Prodrugs Based on Intramolecular Amide Hydrolysis of Kirby's N-Alkylmaleamic Acids
Kirby et al. studied the efficiency of intramolecular catalysis of amide hydrolysis by the carboxyl group of a number of substituted N-methylmaleamic acids 24–30 (Figure 4) and found that the reaction is remarkably sensitive to the pattern of substitution on the carbon–carbon double bond. In addition, the study revealed that the hydrolysis rates for the dialkyl-N-methylmaleamic acids range over more than ten powers of ten, and the ‘effective concentration’ of the carboxyl group of the most reactive amide, dimethyl-N-n-propylmaleamic acid, is > 1010m. This acid amide was found to be converted into the more stable dimethyl maleic anhydride with a half-life of <1-second at 39 °C below pH 3 . Furthermore, Kirby's study demonstrated that the amide bond cleavage is due to intramolecular nucleophilic catalysis by the adjacent carboxylic acid group, and the dissociation of the tetrahedral intermediate is the rate-limiting step . Later on, Kluger and Chin researched the intramolecular hydrolysis mechanism of a series of N-alkylmaleamic acids derived from aliphatic amines, having a wide range of basicity . Their study revealed that the identity of the rate-limiting step is a function of both the basicity of the leaving group and the acidity of the solution.
To utilize N-alkylmaleamic acids, 24–30, as prodrug linkers for tranexamic acid, atenolol, acyclovir, cefuroxime, and other drugs, having poor bioavailability or/and undesirable (bitter) taste, we have unraveled the mechanism for their acid-catalyzed hydrolysis using DFT and molecular mechanics methods. Our DFT calculation results were found to be in accordance with the reports by Kirby et al.  and Kluger and Chin .
Tranexamic Acid Prodrugs Based on Kirby's N-Alkylmaleamic Acids
It is not often that a simple old generic product makes medical news. Yet this is just the case for tranexamic acid. This small molecule that is a synthetic lysine amino acid derivative has been originally developed to prevent and reduce excessive hemorrhage in hemophilia patients and reduce the need for replacement therapy during and following tooth extraction. Yet the use of tranexamic acid has been expanding beyond the small number of hemophilia patients. Perhaps the most exciting new development about tranexamic acid has been the recent publication of the results of CRASH-2, a randomized controlled trial undertaken in 274 hospitals in 40 countries with 20211 adult trauma patients. Tranexamic acid was demonstrated to safely reduce the risk of death in bleeding trauma patients.
Tranexamic acid might also have a role in bleeding conditions apart from traumatic injury. Postpartum hemorrhage is a leading cause of maternal mortality, accounting for about 100 000 maternal deaths every year. Although preliminary evidence suggests that this drug reduces postpartum bleeding, a large trial is being undertaken to assess the effect of tranexamic acid on the risk of death and hysterectomy in women with postpartum hemorrhage. Furthermore, the similarities of tissue injury after trauma and surgery create a novel model for antifibrinolytic therapy with tranexamic acid. Recently, a new oral formulation of tranexamic acid was shown to be safe and effective for treatment for heavy menstrual bleeding2 [175-180]. One of the main disadvantages of tranexamic acid is its pharmacokinetic profile. After an intravenous dose of 1 g, the plasma concentration–time curve shows a terminal elimination half-life of about 2 h. The initial volume of distribution is about 9–12 L. More than 95% of the dose is excreted in the urine as the unchanged drug via glomerular filtration. The plasma protein binding of tranexamic acid is about 3% at therapeutic plasma levels and seems to be fully accounted for by its binding to plasminogen. Tranexamic acid does not bind to serum albumin. As a result of this pharmacokinetic profile, tranexamic acid in CRASH-2 study needed to be administered using a loading dose of 1 g by intravenous infusion over 10 min followed by 1 g infused over 8 h. Although an 8-hr IV infusion may be an easy option in a hospital setting, such option may not be available in under-developed countries or at sites of accidents and battlefields. Similarly, the oral administration of tranexamic acid results in a 45% oral bioavailability. The total oral dose recommended in women with heavy menstrual bleeding was two 650-mg tablets three times daily for 5 days. Accumulation following multiple dosing was minimal2 [175-180].
Improvement in tranexamic acid pharmacokinetic properties may reduce the administration frequency via a variety of administration routes. This can be achieved by exploiting a carrier-linked prodrug strategy2 [175-180].
Continuing our study on how to utilize enzyme models as potential linkers for drugs containing amine, hydroxyl, or phenol group [122-140], we have investigated the proton transfer reactions in the acid-catalyzed hydrolysis of N-alkyl maleamic acids 24–30 [122, 142] (Kirby's enzyme model, Figure 4) reported by Kirby et al. and based on the calculation results of this system, we have proposed four tranexamic acid prodrugs, tranexamic prodrugs ProD 1- ProD 4 (Figure 5).
As shown in Figure 5, tranexamic acid prodrugs, ProD 1-Prod 4, consist of a carboxylic group (hydrophilic moiety) and a lipophilic moiety (the rest of the prodrug), where the combination of both moieties secures a relatively moderate (adequate) HLB. It is worth noting that our proposal is to exploit tranexamic acid prodrugs ProD 1-ProD 4 for oral use via enteric coated tablets. At this physiological environment, the tranexamic acid prodrugs will exist as a mixture of the acidic and ionic forms where the equilibrium constant for the exchange between the two forms is dependent on the pKa of a given prodrug.
The DFT kinetic and thermodynamic properties for tranexamic acid prodrugs ProD 1- ProD 4 (Figure 5) were calculated by DFT methods. Using the calculated DFT enthalpy and entropy values for the entities involved in the acid-catalyzed hydrolysis of tranexamic acid prodrugs ProD 1- ProD 4, the barriers (∆G‡) for all steps described in Figure 6 were calculated. The calculated values for ∆Gf‡, activation energy for the tetrahedral intermediate formation, and ∆Gd‡, activation energy of the tetrahedral intermediate dissociation, demonstrate that while the rate-limiting step for all prodrugs as calculated in the gas phase is the tetrahedral intermediate formation, the scenario is the opposite when the calculations were made in water. The water DFT calculations indicate that the rate-limiting step in the acid-catalyzed hydrolysis of tranexamic acid ProD 4 is the tetrahedral intermediate formation, while that for ProD 1- ProD 3 is the tetrahedral intermediate collapse. To evaluate the factors determining the acid-catalyzed hydrolysis rate in tranexamic acid prodrug comparison of their calculated DFT properties with previously calculated properties for the acid-catalyzed hydrolysis of 24–7 (Figure 4), atenolol prodrugs ProD 1- ProD 2 (Figure 7), acyclovir prodrugs ProD 1- ProD 4 (Figure 7), and cefuroxime ProD 1- ProD 4 (Figure 7) were made. The calculation results reveal that the rate-limiting step (higher barrier) in the gas phase for all systems studied is the tetrahedral intermediate formation. On the other hand, the picture is quite different when the calculations were carried out in dielectric constant of 78.39 (water medium). While for systems 24–30 and atenolol prodrugs ProD 1-ProD 2, the rate-limiting step was the dissociation of the tetrahedral intermediate in the reactions of cefuroxime prodrugs ProD 1- ProD 4 and acyclovir prodrugs ProD 1- ProD 4, the rate-limiting step was the formation of the tetrahedral intermediate.
For assigning the factor determining the rate-limiting step, MM2 strain energy values for the reactants (GM) and intermediates (INT2) in 24–30 and tranexamic acid prodrugs ProD 1-ProD 4 were calculated and correlated with the calculated DFT activation energy values, (∆Gd‡). Good correlation was obtained between the Es (INT) (strain energy of intermediate) in 24–30 and tranexamic acid ProD 1- ProD 3 and the activation energies for the tetrahedral intermediate breakdown (∆Gd‡). In addition, strong correlation was found between log krel (relative rate) and Es (INT) in 24–30. The calculations demonstrate that the reaction rate for systems 24–30 and tranexamic acid ProD 1-ProD 3 is dependent on the tetrahedral intermediate breakdown and its value is largely affected by the strain energy of the tetrahedral intermediate formed. Systems with less-strained intermediates such as 25 and tranexamic acid ProD 3 undergo hydrolysis with higher rates than those having more strained intermediates such as 27 and tranexamic acid ProD 1. This might be attributed to the fact that the transition state structures in these systems resemble that of the corresponding intermediates.
Calculation of the t1/2 Values for the Cleavage Reactions of Tranexamic Acid Prodrugs ProD 1-ProD 4
The effective molarity parameter is considered an excellent tool to define the efficiency of an intramolecular process. Generally accepted that the measure for intramolecular efficiency is the effective molarity (EM), which is defined as a ratio of the intramolecular rate (kintra) and its corresponding intermolecular (kinter) where both processes are driven by identical mechanisms. The major factors affecting the EM are ring size, solvent, and reaction type. Values in the order of 109-1013 M have been measured for the EM in intramolecular processes occurring through nucleophilic addition. Whereas for proton transfer processes, EM values of less than 10 M were reported  until recently where values of 1010 were reported by Kirby on the hydrolysis of some enzyme models [141-149]. Using equation 1 obtained from the correlation of log EMcalc (calculated effective molarity) vs. log EMexp (experimental effective molarity) and the t1/2 value for process 24 (t1/2 = 1 second) , the t1/2 values for tranexamic acid ProD 1 – ProD 4 were calculated. The predicted t1/2 at pH 2 for ProD 1 - ProD 4 is 556, 253 h, 70 seconds, and 1.7 h, respectively.
In Vitro Kinetics Studies
The kinetics for the acid-catalyzed hydrolysis was carried out in aqueous buffer in the same manner as that done by Kirby on his enzyme models 24–30. This is in order to explore whether the prodrug hydrolyzes in aqueous medium and to what extent or not, suggesting the fate of the prodrug in the system. Acid-catalyzed hydrolysis kinetics of the synthesized tranexamic acid ProD 1 was studied in four different aqueous media: 1 N HCl, buffer pH 2, buffer pH 5 and buffer pH 7.4. Under the experimental conditions, the target compounds hydrolyzed to release the parent drug as evident by HPLC analysis. At constant pH and temperature, the reaction displayed strict first-order kinetics as the kobs was fairly constant and a straight plot was obtained on plotting log concentration of residual prodrug verves time. Half-lives (t1/2) for tranexamic acid prodrug ProD 1 in 1N HCl, pH 2 and pH 5 were calculated from the linear regression equation correlating the log concentration of the residual prodrug versus time, and their values were 0.9, 23.9, and 270 h, respectively. The kinetic data in 1N HCl, pH 2 and pH 5 were selected to examine the interconversion of the tranexamic acid prodrug in pH as of stomach, because the mean fasting stomach pH of adult is approximately 1–2 and increases up to 5 following ingestion of food. In addition, buffer pH 5 mimics the beginning small intestine pathway. Finally, pH 7.4 was selected to examine the interconversion of the tested prodrug in blood circulation system. Acid-catalyzed hydrolysis of the tranexamic acid ProD 1 was found to be higher in 1N HCl than at both pH 2 and 5. At 1N HCl, the prodrug was hydrolyzed to release the parent drug in less than 1 h. On the other hand, at pH 7.4, the prodrug was entirely stable and no release of the parent drug was observed. As the pKa of tranexamic acid ProD1 is in the range of 3–4, it is expected that at pH 5, the anionic form of the prodrug will be dominant and the percentage of the free acidic form that undergoes the acid-catalyzed hydrolysis will be relatively low. At 1N HCl and pH 2, most of the prodrug will exist as the free acid form, and at pH 7.4, most of the prodrug will be in the anionic form, thus the difference in rates at the different pH buffers.
The t1/2 experimental value at pH 5 was 270 h, and at pH 7.4, no interconversion was observed. The lack of the reaction at the latter pH might be due to the fact that at this pH, tranexamic acid ProD 1 exists solely in the ionized form (pKa about 4). As mentioned before, the free acid form is a mandatory requirement for the reaction to proceed.
On the other hand, tranexamic acid ProD 4 has a higher pKa than tranexamic acid ProD 1 (about 6 versus 4). Therefore, it is expected that the interconversion rate of tranexamic acid ProD 4 to its parent drug, tranexamic acid, at all pHs studied will be higher (log EM for tranexamic acid ProD 4 is 14.33 versus 9.53 for tranexamic acid ProD 1).
Acyclovir Prodrugs Based on Kirby's N-Alkylmaleamic Acid Enzyme Model
Acyclovir is a synthetic acyclic purine nucleoside analog that is the first agent to be registered for the treatment for and prevention of viral infections caused by herpes simplex (HSV), varicella zoster (chicken pox), and herpes zoster (shingles) [170, 182]. Acyclovir water solubility is very poor and has an oral bioavailability of < 20%; hence, administration is necessary when high doses are needed . Orally acyclovir is mostly used as 200-mg tablets, five times daily. In addition, 6 months to a year administration of acyclovir is required in immune-competent patient with relapsing herpes simplex infection .
The oral administration therapy that currently available is associated with a number of drawbacks such as highly variable absorption and low bioavailability (10–20%). The main problem with the therapeutic effectiveness of acyclovir is its absorption that is highly variable and dose dependent, thus reducing the bioavailability to 10–20%. In commercially available dosage forms of acyclovir, the amount of drug absorbed is very low due to short residence time of the dosage forms at the absorption site. In humans, acyclovir showed poor and variable oral bioavailability (10–20%), probably due to the relatively low lipophilicity of the drug. Thus, the rate-limiting factor in acyclovir absorption is its membrane penetration .
Several approaches have been investigated to improve the oral bioavailability of acyclovir: (i) Luengo and coworkers have used different preparations of acyclovir with β-cyclodextrin to increase its solubility and hence its bioavailability; however, no significant effect of β-cyclodextrin on the oral drug bioavailability was observed . (ii) Encapsulation of acyclovir in lipophilic vesicular structure to enhance the oral absorption and prolong the existence of the drug in the systemic circulation . (iii) Yadav and coworkers have used acyclovir-loaded mucoadhesive microspheres for increasing the retention time and hence the bioavailability of acyclovir , and (iv) the search for an effective prodrug that would provide acyclovir with higher bioavailability led to the synthesis of a number of aliphatic and amino acid esters of acyclovir [188, 189].
Improvement in acyclovir pharmacokinetic properties may increase the absorption of acyclovir via a variety of dosing routes. This can be achieved by utilizing a carrier-linked prodrug strategy which could be implemented by covalently linking acyclovir to a linker to provide a drug-host system which upon exposure to physiological environment, such as stomach or intestine, can penetrate the membrane tissues and release the active drug, acyclovir, in a programmable manner.
To expand our approach for utilizing intramolecularity to design potential linkers for amine drugs, we have studied the mechanism and driving forces determining the rate of the acid-catalyzed hydrolysis in some of Kirby's acid amides (prodrugs linkers) [122, 142]. This work was carried out with the hope that such linkers might have a potential to be good carriers to the antiviral agent, acyclovir.
Based on the DFT calculation results on the acid-catalyzed hydrolysis of maleamic acid amides 24–30 (Figure 4), four acyclovir prodrugs were proposed (Figure 7). The acyclovir prodrugs, ProD 1 – ProD 4 are composed of the amide acid linker having a carboxylic acid group (hydrophilic moiety) and the rest of the prodrug molecule (a lipophilic moiety); the combination of both groups secures a prodrug moiety, having a potential to be with a high permeability (adequate HLB). The plan was to prepare acyclovir ProD 1- ProD 4 as sodium or potassium carboxylate salts due to their stability in neutral aqueous medium. It is worth noting that N-alkylmaleamic acids such as 24–30 undergo hydrolysis in acidic aqueous medium, whereas they are relatively stable at pH 7.4.
Based on a linear correlation between the calculated and experimental effective molarities (EM), the study on the systems reported herein could provide a good basis for designing prodrug systems that are less hydrophilic than their parent drugs and can be used, in different dosage forms, to release the parent drug in a controlled manner. For example, based on the calculated log EM values, the predicted t1/2 (a time needed for 50% of the reactant to be hydrolyzed to products) for acyclovir prodrugs, ProD 1–4, were 29.2 h, 6097 days, 4.6 min, and 8.34 h, respectively. Hence, the rate by which acyclovir prodrug releases acyclovir can be determined according to the structural features of the linker (Kirby's acid amide moiety).
Computationally Designed Prodrugs Based on Bruice's Enzyme Model
In this section, we summarize the results of our study on design and synthesis of novel prodrugs  via linking the active drug with a di-carboxylic semi-ester linker (Bruice's enzyme model) to produce a system that having better physicochemical properties (adequate HLB value) than its active parent drug and is able to release the latter in a chemically driven controlled manner. Bruice et al. studied the hydrolysis of di-carboxylic semi-esters 31–35 shown in Figure 8 and found that the relative rate (krel) for 5 > 4 > 3 > 2 > 1. They attributed the acceleration in rate to proximity orientation. Using the observation that alkyl substitution on succinic acid influences rotamer distributions, the ratio between the reactive gauche and the unreactive anti-conformations, they proposed that gem-dialkyl substitution increased the probability of the resultant rotamer adopting the more reactive conformation. Therefore, for cyclization to occur, the two reacting centers must be in the gauche conformation. In the unsubstituted reactant, the reactive centers are almost completely in the anticonformation to minimize steric interactions [156, 157].
For exploiting systems 31–35 (Figure 8) as prodrug linkers for drugs containing hydroxyl or phenol group, we have recently unraveled the mechanism for their cyclization using DFT and molecular mechanics calculation methods . In accordance with the results by Bruice and Pandit [156, 157], we have found that the cyclization reaction proceeds by one mechanism, by which the rate-limiting step is the tetrahedral intermediate collapse and not its formation. However, contrarily to the conclusion by Bruice's et al., we have found that the acceleration in rate is due to steric effects rather than to proximity orientation stemming from the ‘rotamer effect’ .
Atovaquone (ATQ) Prodrugs Based on Bruice's Enzyme Model
Malaria is a global public health problem, affecting 300 million clinical cases annually, and causes about 2 million deaths per year [191-193]. This protozoan disease is caused by 5 parasites species of the genus Plasmodium that affect humans (P. falciparum, P. vivax, P. ovale, P. malariae, and P knowlesi) . The only one among these parasites that can cause life-threatening complications is P. falciparum , which is dominated in Africa and to which most drug-resistant cases are attributed. Malaria can exist in a mild form that most commonly associated with flulike symptoms: fever, vomiting, and general malaise. While in the severe form caused by P. falciparum, nervous, respiratory, and renal complications frequently coexist due to serious organ failure3 .
Several medications, alone or in combination such as chloroquine, antifolates, artemisinins and others, show effectiveness and were considered as being the corner stone in malaria treatment. However, drug or multidrug resistance to these agents has been escalated and constitutes a major challenge in malaria treatment . Accordingly, the need for new anti-malarial drugs is now widely recognized, particularly those that are structurally different from existing antimalarial drugs and possess a novel mechanism of action .
Atovaquone (ATQ) (for the structure see Figure 9), a hydroxynaphthoquinone, is relatively new treatment option, active against Plasmodium spp . It has a novel mechanism of action, acts by inhibition of the electron transport system at the level of cytochrome bc1 complex . In malaria parasites, the mitochondria act as a sink for the electrons generated from dihydrofolate dehydrogenase, an essential enzyme for pyrimidine biosynthesis; inhibition of electron transport by ATQ leads to dihydrofolate dehydrogenase inhibition resulting in reduced pyrimidine biosynthesis and thus parasite replication inhibition .
It is well established that ATQ has an excellent safety profile and long half-life; besides, ATQ can be administered via oral route. However, ATQ has poor oral bioavailability (less than 10% under fasted condition) and variable oral absorption  due to its poor solubility that results from its lipophilic structure. Consequently, low and variable plasma and intracellular levels of the drug which is an important determinant of therapeutic outcome are expected . Moreover, ATQ is an expensive medication . Therefore, to achieve therapeutic success and to meet the medical needs in malaria treatment, ATQ solubility improvement strategies should be addressed. The prodrug approach has the potential to be the most successful among other approaches to overcome this shortcoming.
The study herein was conducted to design ATQ prodrugs through linking ATQ to a di-carboxylic semi-ester linker (Bruice's enzyme model) to produce a system that is more hydrophilic than its parent drug and is able to release ATQ in a chemically driven controlled manner. This will result in introducing novel ATQ prodrugs with improved bioavailability and better clinical profile.
Based on our calculation results that enabled us to unravel the cyclization reaction mechanism and to assign the factors determining the reaction rate, we have designed five ATQ prodrugs with a potential to have better water solubility than ATQ and to release the active parent drug in a sustained controlled manner (Figure 9). The relative rate, log krel (effective molarity), for processes ATQ ProD 1- ProD 5 The experimental relative rates for the intramolecular cyclization of 31–35 (Figure 8) were obtained from the division of the intramolecular rate and the corresponding intermolecular reaction [156, 157]. For obtaining the relative rates (effective molarity, EM) for processes ATQ ProD 1– ProD 5, we assume that their corresponding intermolecular process is similar to that for systems 1–5.
As an excellent correlation was obtained between the activation free energy values (∆G‡) for 31–35 and ATQ ProD 1– ProD 5 and the difference in the strain energy values of the reactants and intermediates, ΔEs (INT-GM), the calculated values of ΔEs(INT-GM) for ATQ ProD 1–ProD 5 were used to calculate their corresponding relative rates (log krel) which were 6.96, 6.47, 3.78, 6.50, and -12.82, respectively. These values demonstrate that ATQ ProD 1 and ATQ ProD 4 are the most efficient processes among all systems studied, and the least efficient are ATQ ProD 3 and ATQ ProD 5. Using the experimental t1/2 (the time needed for the conversion of 50% of the reactants to products) value for the cyclization reaction of di-carboxylic semi-ester 31 and the calculated log krel values for prodrugs ATQ ProD 1-ProD 5, we have calculated the t1/2 values for the conversion of ATQ ProD 1- ProD 5 to their parent drug. The calculated t1/2 values were found to be ATQ ProD 3, 22.44 h, ATQ ProD1, ATQ ProD2, and ATQ ProD 4, few seconds and ATQ ProD 5 few years. Therefore, the interconversion rates of atovaquone prodrugs to atovaquone can be programmed according to the nature of the prodrug linker.
Paracetamol Prodrugs Based on Bruice's Enzyme Model
The palatability of active drugs possesses significant obstacle in developing a patient convenient dosage form. Organoleptic properties, such as taste, are an important factor when selecting a drug from the generic products available in the market that have the same active ingredient. It is a key issue for doctors and pharmacists administering the drugs and particularly for the pediatric and geriatric populations .
Organic and inorganic molecules dissolve in saliva and bind to taste receptors on the tongue giving a bitter, sweet, salty, sour, or umami sensation. Bitter taste is sensed by the receptors on the posterior part of the tongue. The sensation is a result of signal transduction from taste receptors located in areas known as taste buds. The taste buds contain very sensitive nerve endings, which are responsible for the production and transmission of electrical impulses via cranial nerves VII, IX, and X to certain areas in the brain that are devoted to the perception of taste . Bitter taste receptors are believed to have evolved for organism protection against the ingestion of poisonous food products. Molecules with bitter taste [204-208] are very diverse in their chemical structure and physicochemical properties [209, 210]. In humans, bitter taste perception is mediated by 25 G-protein-coupled receptors of the hTAS2R gene family .
Drugs such as macrolide antibiotics, non-steroidal anti-inflammatory, and penicillin derivatives have a pronounced bitter taste . Masking the taste of water-soluble bitter drugs, especially those given in high doses, is difficult to achieve using sweeteners alone. As a consequence, several approaches have been studied and have resulted in the development of more efficient techniques for masking the bitter taste of molecules. All of the developed techniques are based on the physical modification of the formulation containing the bitter tastant.
Although these approaches have helped to improve the taste of some drugs formulations, the problem of the bitter taste of drugs in pediatric and geriatric formulations still creates a serious challenge to the health community. Thus, different strategies should be developed to overcome this serious problem.
Bitter tastant molecules interact with taste receptors on the tongue to give bitter sensation. Altering the ability of the drug to interact with its bitter taste receptors could reduce or eliminate its bitterness. This could be achieved by an appropriate modification of the structure and the size of a bitter compound.
Paracetamol is an odorless, bitter crystalline compound widely used as pain killer and antipyretic. Paracetamol was found in the urine of patients who had taken phenacetin, and later on it was demonstrated that paracetamol was a urinary metabolite of acetanilide (Figure 10) .
Phenacetin, on the other hand, lacks or has a very slight bitter taste . Careful inspection of the structures of paracetamol and phenacetin indicates that the only difference in the structural features in both is the nature of the group on the para position of the benzene ring. While in the case of paracetamol, the group is hydroxy, in phenacetin it is ethoxy. Another related example is acetanilide that has a chemical structure similar to that of paracetamol and phenacetin, but it lacks the group at the para position of the benzene ring. Acetanilide has a burning taste and lacks the bitter taste characteristic for paracetamol . The combined facts described above suggest that the presence of hydroxy group on the para position is the major contributor for the bitter taste of paracetamol. Hence, it is expected that blocking the hydroxy group in paracetamol with a suitable linker could inhibit the interaction of paracetamol with its bitter taste receptor/s and, hence masking its bitterness.
It seems reasonable to assume that the phenolic hydroxyl group in paracetamol is crucial for obtaining the bitter taste characteristic for paracetamol. This might be due to the ability of paracetamol to be engaged in a hydrogen bonding net with the active site of its bitter taste receptor via its phenolic hydroxyl group.
In a similar manner to the design of ATQ prodrugs, two paracetamol prodrugs were designed by linking the active drug paracetamol to Bruice's enzyme model (linker), which upon exposure to physiological environment releases the active parent drug (Figure 11).
It is worth noting that linking the paracetamol with such linkers via its phenolic hydroxyl group will hinder its bitter taste. Paracetamol ProD 1 and ProD 2 were synthesized, and their kinetics at pH 2 was studied (stomach physiological environment). The kinetics results revealed that while paracetamol ProD 1 has a t1/2 value of 1 h in pH 2, paracetamol ProD 2 showed very fast kinetics and it underwent complete interconversion to paracetamol within few minutes. The difference in the hydrolysis rates for both prodrugs is attributed to strained effects imposed in the case of paracetamol ProD 2, which upon cleavage gives maleic anhydride while in the case of paracetamol ProD 1, the byproduct is the less-strained succinic anhydride.
Studies have revealed that poor pharmacokinetics properties and toxicity-related issues are the crucial causes of high attrition rates in the drug development process. Resolving the pharmacokinetic and toxicological properties of drug candidates remains a key challenge for drug developers. The most efficient approach for overcoming these negative drug characteristics is the prodrug approach. The rationale behind the use of prodrugs is to optimize the ADME properties and to increase the selectivity of drugs for their intended target.
Prodrug design based on a computational approach consisting of calculations using molecular orbital (MO) and molecular mechanics (MM) methods and correlations between experimental and calculated values of intramolecular processes has the potential to be a very effective tool to be utilized for obtaining effective prodrugs that are capable of releasing the parent drug in a programmable fashion. In this prodrug approach, no enzyme is needed for the catalysis of the intraconversion of a prodrug to its parent drug. The interconversion of the prodrug is solely dependent on the rate-limiting step for the intramolecular reaction.
The future of prodrug design is forthcoming yet extremely challenging. Progresses must be made in better understanding the chemistry of many organic mechanisms that can be effectively exploited to push forward the development and advances of even more types of prodrugs. The understanding of the organic reactions mechanisms of intramolecular processes will be the next major milestone in this field. It is envisioned that the future of prodrug design holds the ability to produce safe and efficacious delivery of a wide range of active small molecule and biotherapeutics.
The author would like to acknowledge funding by the German Research Foundation (DFG, ME 1024/8-1) and Exo Research Organization, Potenza, Italy.
Drug.” Dictionary.com Unabridged (v 1.1) (2007), Random House, Inc., via dictionary.com. Retrieved on 20 September 2007.