1.1. Drug development and Intestinal permeability:

Most of the new emerging drug compounds are formulated as orally administered drugs due to the convenience of the oral administration route, however, the properties of some compounds could be incompatible with oral administration. In fact, pharmaceutical industry used to suffer from major financial losses because of the poor bioavailability of some new drugs after their oral administration being discovered while in the clinical development stage [1-4].

Therefore, poor drug candidates with poor biopharmaceutical properties, such as poor oral bioavailability, and aqueous solubility should be identified as soon as possible before entering the clinical development stage in which the cost of research performed for a compound is significantly high, it is even better to identify such properties before the drug synthesis to save money and time.

Over the past years, drug discovery programs were developed and they helped in the generation of a large number of lead compounds, however these compounds compared to conventional drugs have high lipophilicity, low aqueous solubility, and high molecular weight which are all unfavourable characteristics that decrease the success rates of such compounds in clinical development [5]. As a result, there has been a growing interest in the early prediction of biopharmaceutical properties by means of experimental and theoretical models.

The two main properties that influence the drug absorption from the intestinal lumen are its solubility and permeation [6-8]. Low intestinal permeability of a drug has less possibilities for its improvement when compared to poor solubility, since the drug solubility can be altered by choosing the suitable formulation. This is the reason why synthesis of compounds with structures of reasonably high permeability or even screening of compounds with poor permeability during early stages of drug development is considered as a very important and vital step.

First, it is crucial to describe the mechanisms by which drug molecules cross the intestinal barrier to reach the systemic circulation, and subsequently its site of action.

1.2. Mechanisms of permeation of compounds across intestinal membrane:

Solutes come across a number of barriers during their crossing from the intestine to the systemic circulation. There are two main routes for the transport of molecules across the intestinal membrane: 1) the transcellular route in which the intestinal membrane is penetrated by the drug molecule by the aid of channels and transporters, hence it is a carrier mediated route, 2) the paracellular route, in which the drug molecules cross the intestinal epithelium through aqueous pores in between the cells by means of a diffusion process that is not carrier mediated [9].

As described in the fluid mosaic model, the construction of the cell membrane of a double phospholipid bilayer with various lipids and embedded proteins is what gives it its unique characteristics [10, 11] (Figure (1)). As an example of these unique characteristics is, the difference in the permeability properties between the apical and basolateral sides of the intestinal membrane due to the difference in the lipid and protein compositions between the two sides. Also the cell membrane structure has a sieving effect on the diffusion of molecules with large size.

Figure 1: Intestinal cell membrane structure.

1.2.1. Passive transcellular diffusion:

Mainly this mode of transport requires molecules of reasonable lipophilicity and size as it occurs by the apical membrane penetration by the drug molecules followed by their diffusion into the cell cytoplasm, which is the rate limiting step of passive transcellular permeability [12].

Most of the drug molecules which are well absorbed across the intestinal membrane take this passive diffusion mode of transport [13-15].

1.2.2. Paracellular passive transport:

This is the mode of transport favoured by hydrophilic molecules which are incapable of penetrating the intestinal epithelial cell membrane. It takes place by the means of aqueous pores in between the cells which form a small portion of the total surface area of the intestine [16-19].

1.2.3.    Carrier-mediated transport:

1.2.3. a. Active and facilitated transport:

Nutrients and other essential compounds are extracted by the embedded proteins in the cell membrane through different carrier-mediated mechanisms. This mode of transport is only limited to a small number of drugs which are structurally resembling the original substrates of cell membrane protein transporters[20].

Specificity, saturability, and regional variability are considered to be the three main properties of carrier–mediated transport of some drugs [20].

1.2.3. b. Receptor-mediated transcytosis:

It is a subtype of transcellular transport where the drug molecule binds to a receptor found on the surface of the cell then it crosses to the other membrane surface within an endocytic vesicle formed by endocytosis. This mode of transport is not abundant and it is only limited to highly potent macromolecular drugs [21].

1.2.3. c. Efflux mechanism:

Carrier-mediated mechanisms help to enhance transcellular transport of drugs into the cell interior while on the other hand efflux mechanisms carried out by efflux proteins (e.g. P-gp) help pumping drugs in the opposite direction therefore decreasing the overall permeability of these drugs [22-24]. The efflux system main role is to avoid toxic compounds uptake or help in the excretion of such compounds across the intestinal mucosa [25].

An overall summary of transport across the intestinal membrane is shown in figure (2).

Figure 2: Mechanisms of transport across the intestinal membrane (adapted from reference [26]).

Over the years, the prediction of the biopharmaceutical properties of new drug entities (NDE) has received a growing attention where a large number of experimental (in vitro and in situ)and theoretical (statistical) models have been developed over the past three to four decades. These developed models contributed in saving money and time by helping in screening of best drug candidates and exclusion of poor candidates during drug discovery and development stage [27].

Drug intestinal permeability is one of the major biopharmaceutical properties that is worth investigating and predicting using these models. Therefore, a brief description of some of the methods used in determination and measuring of intestinal permeability is given as follows:

1.3. Methods for determination of intestinal permeability:

1.3.1.    Cell culture based models: e.g. (Caco-2 cells)

For almost forty years Caco-2 cells have been used as an in vitro model for investigation of drug absorption. Caco-2 cells originating from the isolation of human colon tumour cells (adenocarcinoma) possess some of the main and important structure and function related characteristics of the small intestine. Therefore, this model is considered to be one of the most commonly used among cell culture based models in the study of the transport of already available and newly synthesized drugs across intestinal membrane especially in the drug discovery process and preformulation studies which means less use of animals for identification of pharmaceutical compounds with the best properties [28].

It presents some advantages such as:[29]

• Human origin.
• Less use of animals in studies.
• No bioanalysis.
• Good screening model.
• Evaluation of absorption enhancing strategies, toxicity of compounds and transport mechanisms.
• Availability of techniques to improve biorelevance of model.
• And some limitations, including:[29]
• Very expensive.
• Time consuming with a long differentiation period.
• Laboratory intensive.
• Inter and intra-laboratory variability of permeability data.
• Low uptake transporters expression.

It has also been used in other applications involving [30]:

1. Evaluation of the bioactivity of plant extracts:The bioavailability of these extracts is usually unclear, as they are often composed of a complicated mixture of molecules also they are metabolised to some extent before reaching their destination cells inside the body. Since it is possible to use these extracts in the formation of new functional foods, therefore Caco-2 has proven to be a suitable method for investigation of these extracts bioactivity by co-culturing of Caco-2 with the cells where the extracts should be giving their effects.
2. Study of cells matrix interactions and wound healing in intestinal cells:A co-culture system of Caco-2 cells and myofibroblasts was found to be efficient for studying the process of intestinal epithelium wound healing and its regulation, as Caco-2 cells have the capability of producing and releasing extracellular components responsible for controlling the ability (power) and rate of intestinal epithelium cells wound healing and repair.

3-Genotoxicity of food contaminants:Human Caco-2 cells have been described in various works as popular method for studying food contaminants crossing intestinal barrier to get to systemic circulation.

Figure 3: Schematic representation of Caco-2 on a microporous (adapted from reference [31]).

1.3.2.    Membrane based models:

For over almost 4 decades, synthetic membranes have been used in studying diffusion process. Parallel artificial membrane permeation assay (PAMPA) is one of the most common membrane based models used since it was introduced by Kansy et al. in 1998 [32]. PAMPA is a method where the donor and the acceptor compartments are placed on the top of each other in a microtiter plate with a lipid infused membrane hence called ‘Sandwich’ assembly. This lipid membrane system is made of a phospholipid ‘cocktail’ supported on a filter in an organic solvent [33].

It presents some advantages such as: [29]

• Relatively low cost.
• Good predictability.
• Availability of various lipid compositions.
• High throughput.

And some of its limitations include: [29]

• The obtained value depends on pH and lipid composition.
• Membrane retention of lipophilic drugs.
• Prediction is limited only to a part of the overall absorption process.

Figure 4: Schematic representation of PAMPA model.

1.3.3.    Ex Vivo models:

In 1951, Ussing chambers set up was first developed by Ussing and Zerahn [34]. It was initially used for the studies related to ion and water transport. However, modifications were further introduced to such set up by Grass and Sweetana to include determination of drug absorption across the intestine [35].

In the Ussing set up, a tissue of an animal, usually rat, is fixed in between the two parts of a diffusion cell [36]. This model differs from ordinary diffusion cells in that both compartments of the Ussing diffusion cell are supplied with bicarbonate buffer in which an oxygen/carbon dioxide mixture is bubbled continuously to keep the excised segment viable [37].

Furthermore, when electrodes are fitted to the Ussing chambers, they become a useful model for investigating how some compounds affect the electrical characteristics of intestinal membrane physiology, as well as, to check the viability of the excised tissue [38].

Applications of Ussing chambers:

This technique is useful for various purposes including the study of transepithelial drug transport and intestinal metabolism simultaneously [39-41].

Furthermore, it appears to be particularly useful in the assessment of the effect of surface active agents or additives on tissue integrity and on the transport of compounds [42-44]. Another important application for this technique is investigation of the effect of different diseases causing changes in the intestinal membrane function with subsequent changes in permeability e.g. inflammatory bowel disease (IBD) and Crohn’s disease [45].

Among its advantages, are [29]:

• Permeability data obtained from this method correlates well with that obtained from in vivo experiments [46].
• Well oxygenation.
• No bioanalysis.

And its limitations, include [29, 47]:

• Underestimation of drug transport due to membrane retention.
• Viability and integrity of tissues used which is time dependant.
• Surfactants can only be used at low concentration especially in set ups fitted with gas lifts due to the foaming problem in chambers.
• Difficulty in obtaining the used tissues.

Figure 5: Schematic diagram of an Ussing chamber (adapted from reference [48]).

1.3.4.    In Situ intestinal perfusion models

In this approach, the small intestine of an anaesthetised rat is either chronically removed (open loop) or initially removed then may be returned to the intestine during perfusion (closed loop) by laparotomy [31].

The blood supply in this approach remains intact allowing multiple sampling therefore, studying the kinetics of the drug introduced into the intestinal segment.

Among its advantages, are [31]:

• The rat in situ model shows good correlation with in vivo human data [49].
• Avoid the exposure of the investigated drug to the stomach acidic conditions that lead to the precipitation or the breakdown of some drugs.
• First pass effect by the liver can be studied if sampling from the hepatic vein is carried out.

And some limitations, including:

• Use of anaesthesia might affect the drug intestinal absorption [50].
• Use of animal in this approach [51].

Figure 6: Schematic representation of in situ intestinal perfusion (adapted from reference [52]).

1.3.5.    Everted intestinal ring/sac

The intestinal segments used in this approach are tied from both sides (everted or not) forming sacs therefore can be used in measuring the drug transport out or into the sacs. These sacs are placed in oxygenated buffer [48].

Among its advantages are [48, 51]:

• Fast and inexpensive.
• Measures permeability in all intestinal cell types and mucus layer
• Useful method for classifying compounds with high or low permeability according to Biopharmaceutical Classification System (BCS).

Some of its limitations are [52, 53]:

• The enzymatic activity is lost within the experiment conditions.
• The viability of the intestinal tissue is lost within the experiment conditions which leads to limited sampling points.
• Absence of nervous response upon exposure to drug.

Figure 7: Schematic representation of the everted gut technique (adapted from reference [52]).

1.3.6.    In silico models for prediction of intestinal permeability through in vitro-in vivo correlation

Another method for prediction of intestinal absorption of drugs intended for oral administration is through the use of physiologically based in silico models.

In literature, a large number of publications describe many mathematical models generated for prediction of the intestinal absorption that involve the use of coefficients of permeability obtained either from in vitro models such as Caco-2 [54-59] and PAMPA [60-63] or from in situ models [64-66] in combination with some physicochemical parameters such as log P, number of hydrogen bonds, aqueous solubility…etc.

Among all of the previously mentioned physicochemical parameters, log P is considered to be one of the most important and widely investigated parameter in the field of prediction of drug pharmacokinetics such as prediction of intestinal absorption of pharmaceutical compounds.

Among the advantages of this method are [67]:

• Money and time saving method as it decreases the number of molecules synthesized and tested.
• Contribute to the decrease in the animal use.
• Reliable prediction of the pharmacokinetic and pharmacodynamics properties of pharmaceutical compounds.

Some of its limitations are [67]:

• A training in modelling and informatics is required.
• Lack of the presence of a computer programme that can completely model the biological systems complexity.

1.4. Importance of lipophilicity in medicinal chemistry and drug discovery:

Pre-formulation is considered as the first learning phase where the main physicochemical properties of a drug are determined prior to its development into a dosage form. Determination of such properties is essential for selection of the drug candidate itself and selection of the optimum delivery system to ensure its delivery to the site of action[68, 69].

Since drug lipophilicity is considered as a key descriptor that controls permeation across biological membranes[70], the evaluation or determination of the lipophilicity of a drug is important for its characterisation to ensure its potential to penetrate lipid barriers and subsequently be absorbed[71, 72].

Among the most important pre-formulation studies, is determination of drug lipophilicity which reflects the ability of a compound to dissolve in lipids or nonpolar solvents, and it is generally expressed as a partition coefficient (Log P). Partition coefficient is defined as, the ratio of the unionised drug distributed between organic and aqueous phases at equilibrium [68]. It is very useful in the prediction of various biological properties of chemicals.

In the case of ionisable compounds partitioning was found to change as a function of pH, this relationship is called the distribution coefficient (Log D) and is pH dependant[73]. Log D is defined as the ratio of the concentration of a compound in the lipid phase to the concentration of all species in the aqueous phase at given pH (organic phase is assumed to contain only unionised species). Therefore, Log P is the partition of the unionised form of the compound (in case of neutral compounds) while Log D is the net partition of ionised and unionised forms of the compound.

Log D can be estimated from Log P and pK_a [73]:

Log D acids=Log P+ Log 11+10(pH-pKa)

Log D bases=Log P+ Log 11+10(pKa-pH)

When the compound is largely unionised then, Log P is assumed to be approximately equal to Log D, then:

Log D≅Log P

A correlation has been found between the oil-water partition coefficient of simple organic compounds and their biological activity [74]. For biological purposes, long chain esters or alcohols are often selected as the organic phase for partition coefficient determination.

Since then, the octanol-water system has been widely used in most biological correlation work as n-octanol was found to be the most appropriate oil phase used for biological applications because it has similar properties to biological membranes.

The Octanol-water partition coefficient was also found to be used for a correlation of structural changes of drug chemicals with the change noticed in biological, biochemical, and toxic effects [75]. It has been determined for a diverse set of compounds creating a large dataset of octanol-water partition coefficient (K_ow). It has been widely used as a hydrophobicity parameter in pharmacological and toxicological modelling.

The methods for determination of partition coefficient are summarised in figure (8).

Figure 8: A schematic diagram illustrating the methods for determination of partition

coefficient.

1.5. Methods for determination of partition coefficient:

Being the oldest parameter in physicochemical profiling, Log P has been determined by a vast number of well-established experimental methods. These methods have been classified into two groups (direct and indirect).

1.5.1. Direct methods:

1.5.1. a. Shake Flask Method:

It is generally regarded as the most reliable method for Log P determination. The idea of this method is mainly based on an extraction procedure, where a solute is allowed to partition between a two liquid system (octanol-water) followed by determination of the concentration of that solute in each layer after equilibrium using either UV/Vis spectroscopy, fluorimetry, high performance liquid chromatography (HPLC), thin layer chromatography (TLC), gas chromatography (GC) and other detection techniques such as radiometry in the case of radioactive solutes [76, 77].

Among its advantages, are:

• Application to a wide range of solutes.
• Accurate method.

And regarding its limitations:

• Tedious and time consuming method.
• Large amount of solute is required.
• Pure solutes must be used as interference from impurities of the solute used will also partition into the liquid phases which may lead to inaccurate and erratic results [76].
• This method is not suitable for compounds of poor solubility in any of the solvent phases used as their concentration will be difficult to quantify by any of the detection techniques used.

1.5.1. b. Slow Stir Method:

It is similar to shake flask method, it only differs in the procedure of the method where a slow stirring under rigid temperature control is applied instead of vigorous shaking thus avoiding microemulsion formation[78, 79].

Advantages:

• Avoids microemulsion formation.
• Reliable for relatively all compounds.

• Does not require expensive equipment.
• Relatively faster.

Limitations:

• Strict experimental conditions should be applied as slow stirring and close temperature monitoring to avoid formation of a microemulsion.

1.5.1. c. Generator column method:

To overcome the previous limitations of the shake flask method, a generator column method was developed. In this method, the generator column is packed with a solid support coated with an organic stationary phase, when water is pumped through the column an aqueous solution is generated which is in equilibrium with the stationary phase. The concentration of the solute eluted with the aqueous phase is measured by HPLC or solvent extraction followed by GC [80-83].

Advantages:

• Avoids microemulsion formation.
• Colloidal dispersion formation can be avoided by a slow flow rate.
• Rapid equilibration by the large interfacial area.
• No loss of volatile solutes as well as no errors from adsorption as it is a continuous and closed flow system.
• Easy, inexpensive, and requires no special skill of the operator.

Limitation:

• The requirement of sophisticated expensive equipment.

1.5.1. d. Potentiometric method:

In dual phase potentiometric titration, the tested compound is titrated twice, firstly in the absence of the partitioning solvent to measure its aqueous pK_a then secondly, in the presence of a partitioning solvent (octanol) with stirring until the pH is measured. The partitioning of the unionised form of the compound in to octanol will make both titration curves different leading to a shift. Log P is calculated from a difference in Pk_a values[84, 85].

Advantages:

• Accurate and precise
• Used for ionisable compounds.

Limitation:

• Limited capacity as compounds with a pK_a out of the measurable pH range cannot be used in this method.

1.5.1. e. Counter current chromatography method (CCC):

In this method, both the mobile phase and the stationary phase are liquid where the stationary phase has no solid support. These two phases are immiscible with each other and the only physicochemical interaction that controls the retention of solutes is liquid-liquid partitioning. Also, the centrifugal field keeps both immiscible phases together. This method is considered as a direct method for determination of log P as it directly relates the distribution volume to the partition coefficient of the solute as both phases present are only liquid and there is no chemical reaction, ionisation or complexation taking place in the mobile phase or stationary phase to be considered so the distribution ratio D= K_D [86]

Advantage:

• It gives the D ratio of compounds directly and in any biphasic liquid system.

Limitation:

• The restriction over the range of the measurable D ratios where large D values need prohibitive times and mobile phase volumes to be determined.

1.5.2. Indirect methods:

1.5.2. a. In silico methods:

Since the octanol-water partition coefficient was introduced by Hansch et al. [87, 88], it has been vastly used in quantity structure activity relationship (QSAR) studies as a hydrophobicity descriptor. Lately, log P has proved to be a key descriptor for modelling and evaluation ADMET properties through a large number of developed approaches which help detect unsatisfactory pharmacokinetic properties and the toxicity of drugs at the early stages of drug discovery therefore reducing the cost of these drugs failing at later stages [89].

Since the 1970’s several computational methods have been developed for calculation or prediction of log P. These methods are classified in to two main classes:

1.5.2. a. 1. Substructure-based Method:

It is divided into two types of methods: [90]

• Atom-based Methods

The overall molecular log P is computed by this method through the additive contribution of individual atoms in the molecule.

• Fragment-based Methods

This method determines log P through the contribution of the sum of non-overlapping fragments and functional groups attached to the molecule.

This later method is better than the earlier because it includes corrections that account for electronic and steric effects.

Both types of contributions are fitted on experimentally determined log P values leading to the generation of a molecular lipophilicity map.

The main advantage of atom-based methods is the avoidance of ambiguities [91] therefore it provides good estimation results compared to the fragment-based methods which are considered to be very accurate methods [92].

Limitations:

For atomic-based Log P calculation methods [90]:

• Ambiguity in the classification system.
• Large number of atom types.
• Unrealistic value of some atom contributions.
• Perceived failure at prediction and bias towards underestimation of log P.

For fragment-based Log P calculation methods:

The inability to predict Log P for molecules with unusual functional groups as a result of lack of experimental data for molecules containing such functional groups [93, 94].

1.5.2. a. 2. Property-based (Whole molecule) method:

Log P calculation is based on physicochemical properties of the molecule under investigation such as volumes, partial charges, molecular surfaces OR different topological and electrostatic indices can be used as parameters for Log P quantification OR molecular lipophilicity potentials (MLP) [95].

Advantages:

• Substructure based methods are normally validated on a large group of data so they give more reliable and accurate results than the whole molecule approach so that is why they are more popular and more widely used [96, 97].

Limitations:

• Many of these methods are validated based on small groups of organic compounds so the feasibility of their application to a larger chemical space is not known. Despite the correlation of many physicochemical properties to log P, there is still no clear explanation for the combination of certain physicochemical properties used to compute log P [95].

1.5.2. b. Chromatographic methods

In this method, log P determination is through a simple correlation of the obtained chromatographic data (retention or mobility time) characteristic of solutes with similar compounds of known log P [98].

A calibration graph of standard reference compounds with known Log P values is plotted against their retention or migration times. Therefore, knowing the retention or migration time of the solute of interest, its Log P can be easily calculated.

Among these methods are Reversed phased high performance liquid chromatography (RP-HPLC) [98], Reversed phased thin layer chromatography (RP-TLC) [98], immobilised artificial membrane (IAM), micellar liquid chromatography (MLC), counter current chromatography (CCC), electrochemical chromatography (capillary electrophoresis (CE), micellar electrokinetic chromatography (MEKC), microemulsion electrokinetic chromatography (MEEKC) and mathematical models).

Advantages:

• Fast and ease of automation.
• Simultaneous determination of Log P for more than one solute in mixture.
• Applicable for a wide range of analytes of different lipophilicity.
• High precision, accuracy and reproducibility.

1.5.2. b. 1. Electrochemical methods:

CE has been widely used in the determination of partition coefficients where MEKC [99-101] and MEEKC [102-105] are commonly used types of CE for this purpose. These are rapid screening electrochemical methods that study and examine the transfer of charged species from one phase to another according to the type of the medium selected; one of the three previously mentioned methods is used. The introduction of micelles into CE for separation of neutral compounds according to their micelle affinity is a method called micellar electrokinetic chromatography (MEKC) which allows the calculation of the partition coefficient for the solute of interest by relating the solute partitioning within the micelle to its Log P. The use of a microemulsion instead of micelles by as a mobile phase is a method named as MEEKC where the solutes of different lipophilicity are allowed to partition into the small oil droplets in the microemulsion and the aqueous phase with different mobility allowing both the separation and the calculation of the Log P of more than one solute at the same time.

Advantages:

• These methods overcome the direct methods limitations.
• Less time consuming, swift analysis and high automation.

Limitations:

• Its limitations are related to the method development as the need for internal standard incorporation to overcome poor injection precision [106].

1.5.2. b. 2. UV spectrophotometry and spectrofluorimetry:

A Partition coefficient can also be determined by means of spectroscopic methods as spectrophotometry and spectrofluorimetry and then the obtained log P can be used as a tool for prediction of different biological activities [107, 108].

1.5.2. b. 3. RP-TLC method:

TLC is a rapid and easy tool for estimation of Log P [109-112]. This method is similar to RP-LC where the Retention factor (K) and lipophilicity parameter (R_m) of a certain compound analysed by such method are linearly plotted against log P.

Advantages:

• Samples used are of very small amounts and are not required to be pure.
• Cheap and simple.

Limitation:

• Restriction of its application to mainly small data sets of compounds of similar properties.

1.5.2. b. 4. Immobilised artificial membranes (IAMs) and Immobilised Liposome Chromatography (ILC):

These previously mentioned methods are considered fast and reliable methods to predict biological properties as drug distribution, absorption and transport across biological membranes as intestinal membranes [113-115], blood brain barriers [116, 117], and skin [118, 119] through chromatographic retention measurement [120]. Since IAMs were introduced as HPLC packaging materials by Pidgeon and Venkataram [121].

The IAMs structure is composed of synthetic phospholipid analogues linked covalently to silica propylamine particles, while in immobilised liposome chromatography (ILC), liposomes are stably immobilised in the pores of gel beads. Both methods are useful in the early profiling of drug candidates in the drug discovery process [122, 123].

The structure of these chromatographic surfaces are prepared in such a way to mimic the fluid phospholipid bilayers chemically and physically supporting drug-membrane partitioning based on lipophilicity and electrostatic interactions, thus the retention factors obtained on IAMs or liposomal columns are used for determination of the solute partition coefficient where the solute capacity factors K’_m aremeasured in liposome systems [122, 124].

In addition to their ability to predict drug membrane interaction, distribution, absorption, and transport across various biological membranes, IAMs appear to have other applications as purification of membrane proteins[125-128], immobilising enzymes [129, 130], obtaining enzyme ligand binding constants for drugs and obtaining hydrophobic parameters [131].

Limitations of ILC:

• Limited stability of liposomes [132].
• Preparation of identical columns is difficult.
• Unavoidable column to column variation because of the methods used to entrap liposomes [133].
• Laborious and very time consuming.

Therefore, IAMs appear to be a simple, rapid and reproducible method for measuring partition coefficients and better for the prediction of drug transport than ILC and other conventional expensive, time consuming and laborious methods such as Caco-2 permeability tests.

1.5.2. b. 5. Micellar Liquid Chromatography (MLC):

The use of micelles in HPLC was first introduced by Armstrong and Henry in 1980 [134], this technique is called micellar liquid chromatography (MLC) and was used to enhance retention and selectivity of various solutes that would be inseparable or poorly resolved.

Micellar liquid chromatography is a reversed phase liquid chromatographic (RP-LC) mode which uses mobile phases containing a surfactant (ionic or non-ionic) above its critical micellar concentration (CMC). The stationary phase is modified with approximately constant amounts of surfactant monomers so the presence of micelles alters the solubilising capability of the mobile phase leading to diverse interactions (hydrophobic, ionic and esteric) [135] with major implications in retention and selectivity.

The basic and very important parts of MLC are surfactants, micellar mobile phase and stationary phase.

Surfactants used in MLC:

Surfactants possess both hydrophobic and hydrophilic moieties where the hydrophobic moiety, is represented by the tail of the molecule and the hydrophilic moiety is represented by the polar head group (as shown in Figure (9)). Surfactants are classified in to different classes: anionic, cationic, and zwitter ionic or nonionic.

                              Figure 9: Structure of a Micelle (adapted from reference [136]).

Because of the dual nature of surfactants, they have the ability for self-organisation in solution. When the surfactant concentration reaches the critical micellar concentration (CMC) or more, aggregates of monomers which are called micelles are formed. The selection of the most appropriate surfactant to be used in MLC depends upon different properties, such as CMC, Krafft point, cloud point and aggregation number (AN).

– CMC:  Surfactants with a low CMC are the most appropriate type of surfactants to be used in MLC as those with a high CMC result in a viscous solution giving undesirable high system pressure and background noise in UV detectors. Sodium dodecyl sulphate (SDS), cetyltrimethyl ammonium bromide (CTAB), and Brij-35 are the most commonly used surfactants in MLC as they have low CMC values. CMC values are affected by the addition of organic modifiers to reduce retention in MLC from modification in the structure of the micelle [137].

– Krafft point: In  case of ionic surfactants, it  is the temperature at which the solubility of an ionic surfactant monomer becomes equal to the CMC [138, 139].

If the solubility is very low, then no micelles are present below the Krafft point temperature. Therefore, chromatographic work should always be carried out above this temperature to avoid surfactant precipitation.

– Cloud point: In case of non-ionic surfactants, it is the temperature above which phase separation takes place therefore chromatographic work using non-ionic surfactants should be carried out below this temperature.

Micellar Mobile Phase: The mobile phase used in MLC consists of surfactants at a concentration above its CMC, where any increase in the surfactant concentration is translated into increase in the concentration of micelles in solution while the number of the surfactant monomers in the mobile phase remains constant. Micelles provide hydrophobic and electrostatic sites (for ionic surfactants) of interaction [140].

Micelles have three sites of solubilisation:

• The core, which is hydrophobic in nature.
• The surface, which is hydrophilic in nature.
• The palisade layer which is the region between the core and the surfactant head group.

Figure 10: Structure of the palisade region of the micelle (adapted from   reference [141])

A non-polar stationary phase and a polar aqueous mobile phase are the common basic components of MLC and RPLC however in the conventional RPLC the hydro-organic mobile phase is homogenous, but in MLC the micellar mobile phase is microscopically heterogeneous as it is composed of two different media: the amphiphilic micellar aggregates (micellar pseudophase) and the aqueous-organic solvent containing surfactant monomers concentration (approximately equal to the CMC).

Organic solvents may be added to the micellar mobile phase for modification of the eluent strength [142], peak efficiency improvement and retention time reduction (via changing the micelle structure) and lowering the polarity of the aqueous solution resulting in to the so-called “Hybrid micellar mobile phase” containing micelles, surfactant monomers, molecules of organic solvent and water.

The choice of the best organic solvent used in MLC depends on the polarity of the analytes. The maximal allowable organic solvent concentration used depends on the type of organic solvent and the surfactant, where a high concentration of organic solvent leads to the disaggregation of micelles and sweeping completely the adsorbed surfactant molecules from the bonded phase thus only free surfactant molecules remain in the mobile phase [136].

Another type of surfactants is the naturally occurring biosurfactants bile salts which have a distinguished shape and unusual micellar properties compared to the conventional head and tail synthetic surfactants.

Bile Salts:

Bile salts (Figure (11)) are metabolic products of cholesterol [143, 144]. They are derived from cholic acid so, it is comprised of a rigid and slightly curved tetracyclic steroid ring based structure [143, 145, 146]. Hydrophilic groups are attached to the hydrophobic ring, these hydrophillic groups are one to three hydroxyl (OH)-groups and an acidic group. Bile salts are conjugated to either taurine or glycine amino acid. Because of their distinct structure, where the hydroxyl groups are oriented towards the concave side of the rigid steroid ring backbone so the hydrophilic part of bile salt structure is its concave side while, the hydrophobic part is represented by the convex side. It appears that as a result of the rigid structure of the steroid ring there is no complete separation between hydrophilic and hydrophobic parts in micelles [143].

Figure 11: Structure of bile salts. (A): Structural formula, (B): 3D structure, (C): Schematic representation of a dihydroxy bile salt (adapted from reference [143]).

According to the type of the bile salt, positions R₁ to R₃ can be hydroxylated. R4 is the acidic group that can be conjugated with taurine or glycine.

Based on their unique structure, there are different assumptions for explaining the micellisation process in bile salts as shown in Figure (12) where (A and B) are different primary micelles, (C) is a Disclike micelle and (D) is a hellical micelle [143]

Bile salts have both hydrophobic and hydrophilic sides; they form micelles in water by means of hydrophobic association of their hydrophobic sides. A variety of models have been proposed to describe bile salts aggregation (micellization). Among the popular models for bile salt aggregation are:

• Small’s model suggested the formation of primary aggregates through hydrophobic association between the hydrophobic parts of (2-9) monomers of bile salts followed by further aggregation of the primary aggregates via hydrogen bonding between the hydroxyl groups. Furthermore, the model proposed that the primary aggregates are globular shape while the secondary aggregates are oblate ellipsoidal in shape [147].
• Oakenfull and Fisher’s model proposed that the bile salts form dimers while in water via hydrophobic interaction. The dimers are supposed to be rod-like in structure [148].
• Kawamura et al. model supposed that the secondary aggregates are disc-shaped in structure, in which the hydrophobic sides are facing each other towards the inside while the hydrophilic sides are facing outwards towards the solvent molecules [149].
• Warren et al. model; according to this model the bile salt aggregates are formed by polar interactions between the bile salt molecules. The formed aggregates are proposed to be helical in shape which is based on the crystalline state rather than the liquid state. Tis model has been discontinued so disc-shaped hypothesis of the bile salt aggregates structure became widely recognised [150].

Figure 12: Schematic representation of different models for a bile salt micellar structure (adapted from reference [143]).

Modified Stationary Phase: The alkyl bonded C₁₈ column is the most widely used stationary phase in MLC, other columns (e.g.  C₈ and cyanopropyl) can be used [151].

Surfactants monomers incorporated in the mobile phase adsorb on the porous RPLC packing altering the various surface properties of the stationary phase, such as surface area, polarity, structure, and pore volume which leads to a major influence on chromatographic retention. The stationary phase pores are also coated by the surfactant molecules which results in decreasing their volume [152].

For most surfactants and stationary phases, the amount of the surfactant adsorbed remains constant after equilibrium between mobile and stationary phase is reached. The adsorption of a surfactant on a silica-bonded stationary phase can occur in two ways:

Hydrophobic Interaction: The hydrophobic alkyl tail of the surfactant is adsorbed on the stationary phase while the ionic head is projected outwards which gives the stationary phase some ion exchange ability with charged analytes.

Silanophilic Interaction: The ionic head group of the surfactant is adsorbed on the stationary phase giving the stationary phase more hydrophobic character.

A competition between surfactant and analyte may possibly take place. Owing to the number of interactions which are possible in MLC, separations for example electrostatic, hydrophobic, esteric, and surfactant monomers adsorb on the stationary phase leading to its modification. Therefore, the MLC system is more complex than conventional RP-HPLC with hydro-organic solvents [153].

For buffering of pH and ionic strength adjustment, ionic compounds are commonly added to the micellar mobile phases in MLC. A change in the amount of the adsorbed ionic surfactant may occur by salt addition by decreasing surfactant CMC, electrostatic repulsion and hydrophobic interactions [154].

Retention Behaviour:

The separation behaviourin MLCis explained by taking three phases into consideration which are: stationary phase, micellar pseudophase, and bulk solvent. According to the analytes differential partitioning between micelles and bulk solvent either in the mobile phase or in surfactant-coated stationary phase, separation of analytes takes place. As a result, three coefficients explain the partitioning behaviour in MLC:

P_sw        Partition coefficient between aqueous solvent and stationary phase.

P_mw       Partition coefficient between aqueous solvent and micelles.

P_ms            Partition coefficient between micelles and stationary phase.

An outline of the interactions taking place between the three phases is shown in Figure (13).

Figure 13: Summary of interactions in MLC

P_sw and P_mw have opposite effects on solute retention, since P_sw represents solute affinity with the stationary phase. As P_sw increases the retention increases whereas when P_mw increases, a decrease in retention is observed from the greater association with micelles as P_mw represents solute affinity with micelles.

Nature of interactions:

The retention behaviour of solutes in MLC depends on the interactions between the solute and the surfactant modified stationary phase and between the solute and the surfactant modified stationary phase and between the solute and micelles.

The elution of neutral analytes with non-ionic and ionic surfactants and the elution of charged analytes with non-ionic surfactants is only influenced by dipole-dipole, nonpolar and proton donor acceptor interactions[155, 156].

In addition to the previously mentioned interactions, charged analytes interact electrostatically with ionic surfactants which form charged micelles and charged surfactant layer on the stationary phase.

According to the charges of analyte and that of ionic surfactant, repulsion or attraction may occur.

In case of electrostatic repulsion, unless significant hydrophobic interaction with the modified bonded layer exists, the charged analytes cannot be retained by stationary phase and elute early at the dead time. On the other hand, in case of combined electrostatic attraction and hydrophobic interactions with the modified stationary phase, strong retention may be achieved in MLC.

Solutes are classified according to their elution behaviour into three categories, which are binding, non-binding and antibinding solutes:

• Binding solutes: solutes that bind or associate to micelles, they show decrease retention when the micelles concentration is increased.
• Non-Binding solutes: solutes that do not bind or associate to micelles, they show unaltered retention by changing the micelle concentration.
• Antibinding solutes: solutes that show increased retention with increasing the concentration of micelles, however antibinding behaviour is not very common.

Electrostatic repulsion is an important issue in antibinding behaviour, where the antibinding behaviour has never been observed between a charged solute and an oppositely charged surfactant.

Antibinding behaviour has not been observed with C₈ or C₁₈ bonded phases modified by adsorption of ionic surfactants since repulsion between solutes and the charged surfactant layer on the stationary phase tend to elute in the void volume region.

On the other hand, when using stationary phases which do not adsorb large amounts of surfactant (C₁) or cyanobonded phases where the surfactant charge is buried close to the bonded phase, antibinding behaviour is observed, this is a consequence of a compound being strongly excluded or repelled from the micelle which forces the solute on to the stationary phase where it is retained as a result of hydrophobic interactions [157].

Description of the retention behaviour in pure micellar mobile phase:

Retention behaviour of binding solutes as a function of the micellar concentration [M] (concentration of surfactant monomers forming micelles equal to total surfactant concentration minus the CMC) was explained by many proposed theoretical approaches [158].

– Armstrong & Nome partitioning model:

The model proposed by Armstrong and Nome [153] considers transitions among three environments in a micellar chromatographic system i.e. water, micelles and stationary phase.

Ve-VoVs=KФ=Pws1+v (Pmw -1)[M]

V_e: The total volume of mobile phase needed to elute a given solute from the column.

V_o: The column void volume.

V_s: The volume of the active surface of the stationary phase.

Ф: V_s/V_o phase ratio.

ν: Partial specific volume of monomers of surfactant in the micelle.

– Arunyanarat & Cline-Love model:

Arunyanarat and Cline-Love [159]assumed association equilibrium of solute in bulk aqueous solvent (A) with the stationary phase binding sites (S) and with monomers of surfactant in the micelle (M) governed by the binding constants K_AS and K_AM respectively.

K=Ф[AS]A+[AM]=ФKAS [AS]1+KAM [M]

– Foley model:

This model is based on the idea that the association between solute and micelle is a secondary equilibrium affecting the retention in the absence of micelles (K_o). Foley put forward the idea of treating the retention factor as an apparent parameter.

 

K=Ko11+KAM [M]

This model resembles the previous two models as the retention factor of free solute (K_o) coincides with P_ws in the Armstrong and Nome model and (K_AS) in the Arunyanarat and Cline-Love model whereas (K_AM) coincides with (K_AM) in the Arunyanarat and Cline-Love model.

A comparison of the pros and cons of MLC method is listed in table (1).

Table 1:  A comparison of advantage and disadvantages of micellar liquid

chromatography method

Advantages

Limitations

– It is an interesting technique for green chemistry because it uses a mobile phase containing 90 % or more water, these micellar mobile phases have low toxicity, are non-flammable and do not produce hazardous waste [136]. – The incorporation of surfactants in the mobile phase leads to altering of the interactions formed inside the column which reduces the amount of organic solvent in the mobile phase compared with that in conventional RPLC [152]. – It provides an alternative to conventional RPLC as it confers analytical procedures of greater accuracy and at a lower cost [160-163]. – It  allows direct injection of real biological samples (for example urine, plasma, serum) for analysis of untreated physiological fluids as micelles have the ability to solubilise proteins therefore no sample extraction or preparation is required prior to analysis proving to be time saving compared with other analytical methods such as HPLC and ion pairing (IP) [164]. – Analysis of various pharmaceutical compounds. – Excessive peak tailing that is observed in (IP) seen for basic drugs is reduced by the use of MLC. – It is used in the separation of hydrophilic drugs that are usually unretained in HPLC. – A novel application of MLC is separation and analysis of inorganic compounds (mostly simple ions) [165]. – It  is considered as a superior technique to ion pairing and ion exchange for separation of charged molecules and mixtures of charged and neutral species [166]. – Micelles can be considered as chemical models for biomembranes, which enable the application of MLC to hydrophobicity estimation of organic compounds [142] where partition coefficients can be calculated by plotting their capacity factors obtained from MLC against micellar concentration of surfactant used. – Micellar liquid chromatography is the same as biopartitioning micellar chromatography (BMC) but they differ in the composition of the micellar mobile phase. In BMC, a C18 stationary phase and polyoxyethylene (23) lauryl ether (Brij 35) mobile phase are used for the prediction of biological behaviour of drugs [167]. BMC is useful in obtaining many models for the prediction of various biological behaviours of different drugs for example BBB penetration[168], ocular tissue permeability [169], skin permeability[170], drug absorption [171], and mutagenicity of aromatic amines [172]. Similar to BMC, MLC has also been used in the prediction of biological behaviour such as skin permeability[173], and oral drug absorption[174].

– One of the major drawbacks of MLC systems is the reduced chromatographic efficiency compared with conventional RPLC with an aqueous organic mobile phase, this decrease in chromatographic efficiency results from an increase in the resistance of solute mass transfer from the mobile phase to the stationary phase and poor wetting of the stationary phase by the mobile phase [175, 176]. Also, the increase in the thickness of the stationary phase (by the adsorbed surfactant) has a major effect on the MLC efficiency [154, 177, 178].