Studying protein-protein and protein-polysaccharide interactions

(1)

Figure: Enzyme-inhibitor complex between trypsin (yellow) and

the Bowman-Birk inhibitor (red) from bitter gourd, copied from [5].

Wouter Knol

23-01-2018

(2)

MSc Chemistry

Analytical Sciences

Literature Thesis

Studying protein-protein

and protein-polysaccharide interactions

by

Wouter Knol

January, 2018

Supervisor:

dr. W.T. Kok

Daily Supervisor:

dr. E. Kaal

(3)

1. Introduction

Proteins and polysaccharides can be found in many types of food, and both contribute to the texture and structure of the food. These effects can be attributed to their thickening and gelling behavior and surface properties.[1]

When both biomolecules are present in foods they may interact with other molecules and form complexes. These interactions are mostly between proteins and charged (acidic) polysaccharides such as carrageenan. Interactions between non-ionic polysaccharides are generally much weaker. Since the main interactions are ionic, they depend heavily on on the pH which dictates the charge of the protein and the polysaccharide. [2] The complexes formed are crucial to the texture and the microstructure of products and are therefore interesting to study. In yoghurt, protein-polysaccharide interactions can arise from two sources, polysaccharides can be added before fermentation or they can be produced by lactic acid bacteria during fermentation in the form of EPSs (Exopolysaccharides). EPSs especially are a subject of interest in recent years due to their ability to modify yoghurt texture in a natural manner. [3] To be able to improve and predict food textures, it is necessary to

understand the interactions between the proteins and the EPSs. The study of interactions between EPSs and proteins is an analytical challenge. Another area of interest is the study of interactions between production enzymes and interacting proteins present in the production matrix. Enzymes are widely present in nature, and their high catalytic rates at mild condition make them interesting for use in a biotechnology perspective. [4] Many cases of interactions between protein inhibitors and enzymes are known. They are commonly strongly associated with dissociation constants ranging between 10-7_{and 10}-13_{M. [5] Interactions between proteins usually consist of multiple types of}

interactions. The major interaction mechanisms are steric complementarity, hydrophobic

interactions, electrostatic interactions and hydrogen bonds. Water present in interaction interfaces can stabilize the complex by forming additional hydrogen bonds and interacting with charges. [6] Many types of analytical methods have been developed for the evaluation of enzyme inhibitor interactions [4].

The aim of this review is to evaluated analytical methods which could be suitable to study

interactions between EPSs and proteins and production enzymes and inhibitors as described above. It is important to note that interactions between these molecules are preferably studied in situations mimicking native conditions, which should be considered during method development. In this thesis the most relevant methods for studying protein interactions and their advantages and disadvantages, will be discussed. Some recommendations will be made regarding method development, after reviewing the most relevant methods.

2. EPS and EPS protein interactions

EPSs are exocellular polysaccharides that are excreted by bacteria.[7] Lactic acid bacteria can excrete these EPSs during yoghurt production, affecting the texture and viscosity. The excreted EPS can form a layer around the cell called capsular EPS, or it can be more loosely attached or released into the environment. The exact function of the EPS for the cell is unsure, it likely serves a protective function against environmental factors such as toxic compounds, osmotic stress and antibiotics. Besides this ecological function EPSs also have a distinct effect on the production of fermented dairy products such as yoghurt. During the production of yoghurt the EPS contributes to a “ropy” character in the final product. The firmness of the yoghurt decreases when more EPS is present. A theory is that EPS interferes in association of casein micelles which results in a weaker protein matrix. The structure of the EPS itself is also an important factor in its contribution to texture. The chain stiffness and length have an effect of the viscosity of the yoghurt.

EPS can vary in structure from homo-polysaccharides such as dextran to more complicated hetero-polysaccharides containing more complex subunits consisting of 2 to 7 monosaccharides. Figure 1 shows an overview of hetero-exopolysaccharide structures published by Ruas-Madiedo et al. in 2002.

(5)

2

Figure 1: an overview of several hetero-polysaccharides produced by lactic acid bacteria, figure copied from [7]

Exopolysaccharides are typically long chains with high molecular weights in the range of 4 x 104_{to 6 x}

106 _{Da .[7] While interactions between EPS and whey proteins are well known and have been}

studied in the past, for example by Doublier et al. in 2000[1] and by Tolstoguzov et al. in 1991[8], very few studies have actually been conducted in native conditions. More recently (2013) attempts have been made by Gentes et al.[9] to mimic the natural environment of a protein-EPS complex in milk permeate systems [9]. The interactions between polysaccharides and proteins can be either segregative of associative.

A fully homogenous mixture of polymers in solution can only be achieved under specific conditions of polymer-protein ratio and concentration. These conditions also depend on the environmental factors such as pH and temperature. The interactions between protein and polysaccharides can decrease or improve macroscopic stability depending on these conditions. This can for instance result in

improved colloid stability or a separation of two phases. When polysaccharides and milk proteins interact, thermodynamic compatibility is observed when the biopolymer-biopolymer interaction is favored over the solvent-biopolymer interaction. When this is not the case, segregative phase separation will take place and separate regions rich in each biopolymer will form. Tuinier et al. (2000)[10] describes this phase separation in whey protein-colloids as a result of EPS. These interactions arise from a so-called depletion mechanism. These interactions are formed when the polysaccharide loses conformational energy when confined between two protein colloid particles. This leads to a depletion of polysaccharides in the region between the two protein particles. Osmotic pressure on the particles from polysaccharides in solution then leads to an attracting force between the two colloid particles.

(6)

3

This depletion theory is well developed for large spheres (colloids) and small polymers, in the case of EPS whey protein colloid interactions it is the opposite. The EPS are larger than the whey protein colloids. When dextran is mixed with whey protein BSA it leads to phase separation due to incompatibility. In contrast, the mixing of native whey proteins with EPS does not lead to phase separation, since the native whey colloids are too small to restrict the EPS confirmation enough to induce whey separation. When increasing the whey colloid size, the EPS-whey colloid mixture does show phase separation. Since in this case the colloids are then large enough to restrict the EPS confirmation enough to induce phase separation. [10] These interaction studies were however not performed under native conditions but in water. Since yoghurt is a complex system containing many compounds, such as salts acids and sugars, this research does not fully represent the mechanism in situ.

In 2008 Ayala-Hernandez et al. (2008)[11] studied EPS protein interactions in a simplified milk system using milk permeate. The interaction of EPS with latex particles was studied first, they found an increase in the diameter of the particles, which indicated that the EPS attached to the particles. This interaction was dependent on the size of the latex beads used. They found that the zeta potential (potential difference between fluid layer around particle and dispersion medium) changed very little when the EPS was added. This indicates that the EPS has a negative charge since the outer surface of the particles was also negatively charged, if the EPS had a neutral of positive charge it would result in a larger change in the zeta potential. The interaction between EPS and a whey protein stabilized emulsion at pH 3.5 was examined, the apparent diameter of the particles increased significantly with the addition of EPS. The maximum was found to be at 0.2 mg ml-1_{, this concentration of EPS}

increased the apparent diameter of the particles by roughly 140 nm. The charge went from positive to negative, which is also a clear indicator of an interaction.

Gentès et al. (2013)[9] expanded on this research and examined the interaction of four different EPSs with different structural properties, with the aim to study the interaction of structurally different EPSs with caseins and whey proteins. To realistically simulate EPS-protein interactions in yoghurt, milk permeate was used to mimic environmental conditions. This study focused more on the rheological properties of yoghurts produced by bacteria with different EPS productions with or without whey proteins present. They found that the addition of whey proteins and caseins increased rheological values, but the effect of the EPS on these values remained constant regardless of protein concentration. They found that anionic, linear and stiff EPS mainly contributed to gel firmness, probably due to electrostatic interactions with caseins. Neutral stiff EPS contributed to the viscosity because of its ability to retain water, but not to the gel firmness, possibly due to thermodynamic incompatibility.

In short, interactions between proteins and EPS have been well studied but never under native conditions. The main approach has been at a more macroscopic scale instead of looking at direct chemical interactions between molecules.

3. Protein-protein interactions

Interactions between proteins are involved in virtually every cellular process.[12] Many proteins consist of multiple subunits that interact strongly. There are many types of methods to determine affinity between proteins, for example protein affinity chromatography, affinity blotting and immunoprecipitation. This literature review will specifically focus on methods which can be used to determine binding constants. Some general remarks on protein-protein interactions will be made before an overview of methods will be given. Any interaction between molecules like proteins is governed by the binding constant, which can be given as the ratio between the rates of formation and dissociation of the complex.

(7)

4

these factors is the possibility of competing proteins. Multiple proteins might have a certain affinity for the same ligand which can influence the determined binding constant between a certain protein and ligand. This is especially the case when performing analysis in a complex sample matrix with a wide range of proteins present. Another important factor is the presence of cofactors, which can have a major influence on the complex formation. Finally, the solution conditions, such as pH or the ionic strength, play a major role in the binding constants determined. This means that determining the binding constants under native conditions is crucial to understand the binding in situ.

4. Overview of methods for determining binding interactions

A large variety of methods can be used to study the interactions between proteins and other proteins or EPSs. A quick overview of methods for studying protein binding will be given, more promising methods for the study of protein-protein or protein-EPS interactions will be elaborated on further.

Since very little studies have been performed on protein-EPS binding, methods for determining protein-drug binding constants will be discussed and their possible application to determining EPS-protein or EPS-protein-EPS-protein binding constants.

There are two approaches for studying protein-drug interactions, separative and non-separative methods. Separative methods separate the free ligand from the bound species to determine either the bound concentration or the free concentration. These methods include membrane based methods, CE and LC. The second approach depends on a change in chemical properties due to binding interactions; this mostly includes spectroscopic methods and calorimetry.

4.1 Equilibrium dialysis, ultrafiltration and ultracentrifugation

In equilibrium dialysis (ED) the drug and the protein are separated by a semipermeable membrane [13]. The drug can pass the membrane but the protein cannot, because it is larger in size. When an equilibrium is reached after a certain time the concentration of free drug can be measured in the compartment free of protein and complex. While this method is the standard for measuring protein drug interactions, it might not necessarily be suitable for studying protein-protein or protein-EPS interactions. The method depends on a large difference in molecular weight, so that only one of the compounds can pass the membrane.

There might not be a large difference in molecular weight when studying protein or protein-EPS interactions, making the method unsuited. Another disadvantage is that ED is a slow analysis method, equilibration times are typically 12-48 hours and protein adsorption to the cell wall and membrane can be significant. Furthermore, osmotic pressure can cause a volume change of up to 30%, which results in errors in determining the binding constant.

Ultrafiltration (UF) has been proposed as a faster alternative to equilibrium dialysis. UF is similar in principle to ED, in UF however there is a pressure applied which forces the solution through the membrane. This method also suffers from a specific protein binding to the membrane and the addition of pressure makes it susceptible to protein leakage across the membrane.

Another related method that can be used to determine protein drug interactions is

Ultracentrifugation, in this method the protein and drug are placed in a centrifugational field together. The protein and complex have a larger sedimentation coefficient than the drug and will thus sediment while the free drug will remain in the supernatant. This method does not suffer from membrane binding proteins like UF and equilibrium dialysis, but it does need a large difference in precipitation coefficient between drug and protein. This makes the method unsuitable for determining EPS-protein and protein-protein interactions.[13]

(8)

5

4.2 The parallel artificial membrane assay

The parallel artificial membrane assay uses two compartments separated by an artificial membrane made by coating a porous filter plate with lipids or solvents. One compartment is filled with buffer, protein and drug, while the other is filled with neat buffer, these compartments are named the donor and acceptor compartment. The concentration of free drug in the acceptor compartment is measured at different times.

The experiment is conducted twice, once with protein added and once without. Since the protein and complex cannot cross the membrane, there is a slower migration rate to the acceptor

compartment when an interacting protein is added. This difference in migration rate can be used to estimate the binding constant.

Compared to the earlier discussed methods, it offers some advantages. A major advantage is that there is no volume transfer, since the chemical membrane is not permeable to water. The method can be applied in a 96 well format for faster analysis and the membrane is not susceptible to nonspecific adsorption.

The method does however have some downsides, the liquid membrane, usually consisting of a nonpolar solvent or lipids, needs to be passable to just the drug and not the protein. This means that a careful consideration must be made about which membrane is chosen in combination with which drug. Since the membrane is nonpolar, this method is not applicable to the study of protein-protein interactions, since proteins cannot cross the membrane. EPSs will probably also be unable to cross the membrane making the method unsuited for the study of EPS- protein interactions.[13]

4.3 Spectroscopic methods

Spectroscopic methods such as UV-VIS, IR, fluorescence, NMR and circular dichroism, are based on the change in electronic and spectroscopic energy levels due to binding between ligand and protein[13]. These methods often work in solution, which means that measurements can be

performed at true equilibrium. An advantage of using spectroscopic methods is that they also give an insight in the binding mechanism next to determining the binding constant.

These methods can also give information on the three-dimensional structure of the protein and the changes in it due to binding interactions. If the UV or visible spectrum of a drug changes due to the binding interaction, this could be interpreted as an indication of the polarity of the binding site. Fluorescence can determine the location of the binding site and can be used to determine the distance between the fluorophore and the binding drug. IR measurements provide information on the secondary structure of the protein. NMR analysis can reveal the sites on the protein involved in the binding process. Circular dichroism can be used to acquire information on the three-dimensional structure of the binding site.

A combination of these methods is often used to get extensive information on the binding site. While spectroscopic methods offer a lot of information about the structure of protein and ligand and the binding sites, they do have some disadvantages. Spectroscopic methods are generally only applicable to tight binding systems and are not well suited for studying multiple equilibria. Besides that, the methods are often relatively unsensitive, which limits their application range. Another disadvantage is that highly purified samples are generally required. [13]

While these methods offer great insight on the binding sight structure and binding mechanism, they seem unsuited for initial experiment for studying protein-protein and protein-polysaccharide interactions. Since the methods require highly purified samples in high concentrations, they are unsuited for screening in a complex sample matrix. The methods might however be suited for further examination of interactions found with other methods.

(9)

6

4.4 Calorimetric methods

Isothermal titration calorimetry and differential scanning calorimetry, are the main calorimetric methods used for studying binding interactions [13]. Isothermal titration calorimetry monitors the heat uptake or release during complex formation. A ligand is added to a protein solution in a reaction cell and the binding affinity is determined based on the heat release.

This method can be used for molecules of all sizes and for molecules which are unsensitive to spectroscopic methods. Very strong and very weak binding affinities cannot be studied using this method and larger amounts of material might be required for analysis. Differential scanning calorimetry was developed to monitor protein folding and stability. The protein is heated at a controlled rate until 50% of the protein is denatured. If the experiment is repeated with a ligand present, this ligand might stabilize the protein if binding to the native form is preferred over binding to the denatured form. If this is the case, then the energy required to denature the protein will be higher. This method can thus be used to indirectly determine the binding constant. While this method can be used to determine large binding constants, it does suffer from high sample consumption and low throughput.

The main advantage of these methods, is the ability to give a thermodynamic insight on the binding mechanism. Using these methods the entropy change, the enthalpy of the binding reaction, the Gibbs free energy as well as the binding stoichiometry and the binding constants, can be determined. [13] Much like the spectroscopic methods, calorimetric methods seem unsuited for initial

experiments for determining binding constants. However, they might prove useful to further examine interactions found with other methods.

4.5 Liquid chromatography methods

Two main LC methods are used for determining the interactions between proteins and drugs, size exclusion chromatography (SEC) and high performance affinity chromatography. SEC enables the analysis of interactions in free solution. A variety of methods are available for determining binding interactions in SEC. These methods are similar to methods used in CE and will be discussed in the CE section in more detail. A major drawback to SEC is the low column efficiency and low protein recovery. Since proteins and EPSs are large biomolecules, SEC might not have the efficiency to separate complexes from EPS or protein, especially in a complex sample matrix.

In high performance affinity chromatography one of the interaction compounds, usually the protein, is immobilized on the column surface, after which the interacting compound is injected. Retention is based on the affinity with the immobilized protein. A major advantage of this method is that only a small amount of analyte is immobilized, which can then in turn be used for multiple experiments. The major drawback of this method is that the immobilized protein might not fully represent the protein under native conditions in solution. The immobilization process could affect the activity of the protein, denature it or alter its orientation. Furthermore, the carrier on which the protein is

immobilized often shows some interaction with the analyte. When dealing with strong interactions, high concentrations of organic modifiers might be required to elute the strongly interacting

compounds. These high modifier concentrations can alter the conformation of the protein and influence the interactions, which can reduce the measured binding interaction. [13]

(10)

7

5. Capillary electrophoresis

5.1 Advantages and disadvantages

Capillary electrophoresis (CE) offer some clear advantages over other discussed methods for determining binding constant in this research:

1. The method offers high efficiency and is very selective. This is especially useful in a matrix which contains multiple proteins, which would be the case when simulating a yoghurt matrix or a complex cell lysate.

2. Low sample consumption. Since EPS is only produced in low concentrations during fermentation and is relatively hard to isolate and purify, this is a major advantage of using CE for analysis. For the case of protein-enzyme interactions this might be not relevant, since plenty of sample is available. 3. Fast analysis times. If multiple environmental conditions such as pH values, salt species and ionic strengths and their effect on the binding constant have to be evaluated, short analysis times are crucial. In the case of EPS protein binding, multiple types of EPS might have to be analysed which might be very time consuming if long analysis times are present.

4. Ability to adapt buffer to mimic physiological conditions. This is crucial for determining the effect of environmental factors on the binding constant. Especially in the case of the enzyme binding to unwanted ligands, it is interesting to see if binding can be decreased by environmental factors. In the case of EPS-protein binding, native conditions are crucial to determine binding constants under realistic conditions.

While it offers some great advantages over some other previously mentioned methods, it also suffers from some drawbacks. Protein adsorption to the capillary walls is a risk and when UV detectors are used, there are generally poor detection limits. [13]

Since CE seems to be the most suitable methods for analysing protein-EPS and protein-protein interactions, the method will be elaborated in more detail than the other methods discussed.

5.2 General background

Capillary electrophoresis a method which has emerged as an alternative to many slab gel techniques. Due to effective heat dissipation of the capillary, it is possible to work at high voltages which enables high resolution analysis. The method can also be applied to study interactions between biomolecules in so-called affinity capillary electrophoresis (ACE). It should be noted that the term affinity capillary electrophoresis is used on literature in two ways, either to refer to a specific CE method for

determining binding constants or to refer to all CE methods for examining affinity interactions. To avoid confusion ACEm will be used as an abbreviation when referring to the specific method and ACE will be used an abbreviation when revering to all CE methods for determining binding interactions. ACE was first introduced in the early 1990’s and can be used to determine binding constants, rate constants and binding stoichiometry between biomolecules.[14]

In most cases ACE is used to determine the interactions between small molecules such as ions, drugs and or other analytes and larger ones such as polymers, proteins and micelles. There are multiple experimental approaches which will be discussed in more detail later in this paper.[15]

Both weak and strong binding systems can be analysed using ACE. When analysing weak or moderate binding systems, mobility based methods are often used. The binding between ligand and protein changes the electrophoretic mobility of the protein due to a change in the charge to mass ratio. In

(11)

8

these cases, the change in electrophoretic mobility of the protein as a function of the change in concentration of the interacting species in the buffer is most often used to determine the binding constant. When using mobility shifts there are some key assumptions: the equilibrium must be established instantaneous, the ligand must be distributed homogenously, one of the compounds must be in large excess and a binding stoichiometry of 1:1.

In case of other binding stoichiometry’s and multiple binding sites more complex models can be used. The speed of analysis must be compatible with the speed of the kinetics. The rate of kinetics must be fast enough to establish an equilibrium during the analysis to get both analyte and ligand to elute as one peak. Generally, the complex dissociation halftime must be less than 1% of the analysis time. If this is not the case, separate peaks or odd peak shapes might occur.

Furthermore, this also means that longer analysis times enable the analysis of binding systems with slower kinetics. The mobility measured can be seen as a weighted average of the electrophoretic mobility of the ligand and the complex and can thus be used to calculated the binding constant. [16] As mentioned earlier ACE is mostly used to determine interactions between small molecules and proteins, such as interactions between low MW drugs and proteins. It can however also be applied in the analysis of protein-protein interactions. In these cases, there are often strong interactions present, the direct integration of peak areas of the complex, protein or ligand is most commonly used to determine the binding constant. This does however require a sensitive detection method of the free and complexed species, otherwise significant errors in the determination of the binding constant are possible.[14]

5.3 Detection options

Laser induced fluorescence (LIF), in combination with dye labelled proteins, offers a more sensitive quantification method than more traditional UV detection. Kiessig et al. [17]used green fluorescent protein (GFP) in a screening method for protein-protein interactions. With this method they

investigated interactions between a cyclophilin (rDmCyp20) and the capsid protein p24. The GFP was fused to the c-terminus of the cyclophilin. This enabled LIF detection which added selectivity, thus enabling the determination of binding interactions in a complex sample. The developed method was used to successfully determine the binding constant at Kd=20 ± 1.5x 10-6 M. Furthermore, The

method was tested to screen for interacting compounds in biological extracts. Using this screening approach a tightly binding compound was found in extracts of D. melanogaster, which indicates that this method might be suitable for high-throughput screening for drug development.

More recently mass spectrometry has emerged as a detection method in CE. Vuignier et al. [18] describes its application to determining binding interactions between drugs and proteins. The use of MS offers more sensitivity that the traditional UV detection methods. The poor UV detection limits limited the binding range that could be examined using CE. MS detection enables the analysis of a larger binding range. LIF detection also greatly decreases the detection limit, but the labelling can interfere with binding interactions, a disadvantage from which MS detection does not suffer. The use of MS does however, need careful consideration of parameters like the buffer composition, rinsing steps between measurements and the MS parameters, to achieve stable MS detection. The buffer composition is one of the issues with CE-MS detection, since the buffers must be both MS compatible and they should not interfere with the binding process. Most studies which used ACE-CE-MS used Ammonium acetate or tris-acetate buffers as background electrolyte (BGE), which are standard in MS detection. A postconditioning step is usually performed to remove proteins that adsorbed to the capillary wall during the previous analysis. This is usually performed by rinsing the capillary with 0.1M NaOH. Since NaOH is not MS compatible this might cause issues, the use of 0.1% NH4OH, which is MS

compatible, yielded similar results. Furthermore, careful selection of the sheath liquid and ESI voltage are needed to achieve stable signals.[18]

(12)

9

5.4 The electroosmotic flow

Since the capillary is generally made of bare fused silica, is negatively charged at pH values of 2 and higher.[19] This negative charge induces the electro osmotic flow (EOF) when the electric field is applied. If the pH increases, more silanol groups become charged and the EOF increases. The charged silanol groups initiate the formation of a double layer at the contact point between the solution and the capillary wall. The difference in potential between both layers is called the zeta potential. The velocity of the EOF can be calculated using the Helmholtz-Smoluchowski equation.[19]

Equation 1:

𝑣𝐸𝑂𝐹=

𝜀 ∙𝜁 ∙ 𝐸 𝜂

In which ε is the permittivity, ζ the zeta potential between the two layers, E is the strength of the electric field and η the viscosity of the BGE. The EOF can easily be determined experimentally for instance by using a neutral marker or by monitoring the volume change in a receiving vial. If organic solvents are added to the BGE as modifiers, the viscosity of the BGE changes and the EOF usually decreases. The temperature also has an influence on the EOF, at a higher temperature the viscosity of the BGE decreases and the EOF increases.[19]

5.5 Capillary coatings

As mentioned earlier one of the major disadvantages of CE compared to other analysis methods, is protein adsorption to the capillary. The use of capillary coating can both reduce the degree of protein adsorption and alter the EOF. [20]

There are two main categories of capillary coatings, dynamic and permanent coatings. Dynamic coatings are based on the addition of agents to the BGE. These compounds are added to the buffer and cover the capillary wall during separations.

Permanent coatings are applied irreversibly to the surface of the capillary, by covalent bonding or by physical adsorption. Dynamic coatings are easily applicable and can provide an effective method to alter the EOF or to limit protein adsorption. One of the advantages of using a dynamic coating is that they are easy to remove. Some common forms of dynamic coatings include small ions, such as ionic surfactants, amines or neutral hydrophilic polymers. These agents are often positively charged since the capillary wall in negatively charged. The addition of these compounds can thus change the direction of the EOF or alter its strength by interacting with the capillary wall.

Some common dynamic coatings will be discussed. Amines such as spermine, agmantine and petrescine mask silanol groups which leads to increased precision and reproducibility. Quaternary ammonium derivatives, such as 1,4-Dialkyl-1,4-diazobicyclo[2,2,2]butane diiodide (M7C4M7), are very effective at decreasing protein adsorption. Their effect on the EOF can range from no effect to reversal of the EOF, depending on the exact type used. Polymers can be used to reduce the EOF, especially hydrophobic polymer which adhere to the capillary wall, since they have little affinity with the aqueous phase. Cationic polymers such as polybrene have also been used in CE. The addition of a low concentration (0.4% or less depending on type) of cationic polymer can be used to reverse the EOF. Finally, surfactants have also been used to reverse the EOF and to avoid interaction with the capillary wall. The use of surfactants is however hard to optimize because if they are used above the critical micelle concentration analytes may interact with the micelle changing their apparent

electrophoretic mobilities.

Permanent coatings are covalently bonded to the capillary or permanently adsorbed. The use of permanently adsorbed coatings has some advantages over covalently bonded coatings, they are easier to apply, easily regenerated and less dependent of the chemistry of the capillary surface. Permanently adsorbed coatings are adsorbed by interactions such as hydrogen bonding, hydrophobic interactions or electrostatic interactions. The major advantage of these coatings over dynamic coating, is that there is no need to sustain them. This means they do not have to be added to the BGE

(13)

10 during analysis.

Dusa et al. (2016)[21] reported the use of cationic polyelectrolytes as a permanently adsorbed coating, which is referred to as a semi-permanent coating in the paper. The coating had to be reapplied after 5 measurements. The poly electrolytes, poly(2-(methacryloyloxy)ethyl

trimethylammonium iodide, PMOTAI and poly(3-methyl-1-(4-vinylbenzyl)-imidazolium chloride, PIL-1 showed good stability across a large pH range (up to pH 11). Its uses were also demonstrated on the separation of betablockers, optimal separation was achieved at pH 3. [21] Gonzalez et al. (2003)[22] demonstrated the use of a physically adsorbed coating based on a

N,N-dimethylacrylamide-ethylpyrrolidine methacrylate copolymer. Since the electrical charge of the polymer could be adapted by changing the pH, the coating could thus be used to separate both basic and acidic compounds. The RSD of the migration time was determined to be lower than 3.9% (n=15, 3 days). The coated system proved to be able to separate whey proteins with good repeatability.[22] Covalently bond coatings are applied by chemical reactions to create covalent bonds with the silica wall. This is generally done by pre-treating the capillary with NaOH before fusing the desired

compound to the capillary surface by reaction with a bifunctional alkoxysilane.[20] This process is not only time consuming and difficult, but it can also be irreproducible.[19] These coatings can be used in combination with MS detection and are stable over long periods of time. Some examples of

permanent coating will be given. [20] Sola & Chiari (2012)[23] reported the use of hydrophilic polymeric coatings with reactive groups, which form covalent bonds with the silanol groups on the capillary surface. The method used a poly(N,dimethylacrylamide) based copolymer with

N-acryloyloxysuccinimide (NAS) and 3-(trimethoxysilyl)propyl-methacrylate (MAPS) as reactive groups. Leftover reactive groups were reacted with amino modified agents. The coating showed stable EOFs over a wide pH range and excellent separation of both basic and acidic proteins.[23] Other covalently bound capillary coatings include diazo based polymers which bond to the capillary under influence of UV light. [20]

6. ACE methods

Capillary zone electrophoresis offer multiple methods for studying binding constants.[24] Each with its advantages and disadvantages, the methods will briefly be discussed and then possible

applications on the study of protein-protein and protein-EPS binding will be discussed. We will speak of interactions between proteins and drugs, when describing the working principles of the methods. There are of course many more applications of ACE than just determining protein drug interactions but this provides a standard situation for explaining the methods. These methods all depend on differences in the electrophoretic mobility of the protein (µP), the drug (µD) and the formed complex

(µC) in order to separate the different species and allow quantification of the binding parameters

based on peak areas of migration time shifts.

6.1 The Hummel and Dreyer method

In the Hummel and Dreyer (HD) method a varying concentration of drug in the buffer is used, while a set amount of protein is injected.[24] The presence of the drug in de BGE creates a high background signal.[15] If a complex is formed, it will result in a positive peak corresponding to the protein and complex which migrate away from the injection plug. The bound drug in the buffer will result in a negative peak since there is less drug present, due to some of the drug being used to form the complex. The negative peak area is directly related to the amount of drug bound to the ligand. The free drug concentration can be assumed to be equal to the amount of analyte in the buffer. [24] The main advantages of this method are the ability to determine the number of binding sites, even in low affinity situations and the possibility to control the amount of free ligand independently. The use of this method does however require that some conditions are met. It is crucial that the kinetics are fast enough, as discussed earlier in the case of mobility shift based ACE methods. Furthermore, it is

(14)

11

required that the bound and free macromolecules have identical mobilities. The calculated binding constants are only correct when µC=µP. In the cases that µC>µP>µD the binding constant will be

overestimated. When µC<µP>µD the binding constant will be underestimated. The degree of this over

or underestimation is dependent on the difference between µC and µP. [25]

The reason for this over or under estimation is caused by a brief disturbance of the equilibrium condition after injection when µC≠µP. In figure 2 a simulation of an elution scheme can be seen of the

three situation discussed above.

Figure 2: Simulated elution scheme of the HD method in three different scenarios, A µC=µP>µD, B µC>µP>µD and C µC<µP>µD,

in which the Pf, Df and C axes represent the elution profiles of the free protein, the free drug and the complex respectively,

figure copied from [24].

In case A µC=µP>µD, the estimation of the binding constant is correct, the negative peak accurately

represents the amount of D that is bound. This is because free protein and formed complex move at the same speed keeping a constant equilibrium condition in which Df, Pf and C are in a constant

ration to each other. This , which result in a correct estimation of the binding consant. In case B µC>µP>µD some of the complex migrates out of the injections plug, while the equilibrium is

established leaving P behind. The P that is left behind at the rear end of the zone, will then complex with free D to maintain equilibrium. This will cause a larger negative peak area. The complex that initially moved out of the zone, will dissociate since there is no free P present and after a while an equilibrium is reached. In this case however the initial amount of extra D that was consumed, remains in the P zone causing an overestimation of D bound. In case C the opposite happens, the protein moves out of the zone causing the complex to dissociate. This causes an underestimation of D bound. [24]

De binding constant can be determined by fitting Equation 2 to a plot of the ratio or bound protein, r vs the concentration of free drug, Df. Methods of calculating binding constant will be discussed in detail under the section “Quantification”.

Equation 2:

𝑟 = ∑ 𝑛𝑖𝐾𝑖𝐷𝑓 1 + 𝐾𝑖𝐷𝑓 𝑚

𝑖=1

(15)

12

constant, r the ratio of acceptor bound and Df the concentration of free drug.

Markuszewski et al. (2001) [26] used the HD method to determine the binding constant between BSA and buspirone. The method used a Scatchard plot a plot of r/Df vs r to determine the binding

constant, more information on quantification using Scatchard plots in given in section 6.8. The method suffered from poor peak shapes due to protein adsorption to the capillary walls and slow binding kinetics. While the authors calculated a binding constant at 5.55 x 10-4_{, the results are}

questionable. Two separate positive peaks seemed to appear, which could be caused by slow kinetics combined with µC not being equal to µP. Since fast kinetics and µC=µP are both conditions of using the

HD methods, the method seems unsuited. Furthermore, the authors might have benefitted from using a nonlinear regression model by plotting r vs Df, since the data suggested that multiple binding

sites with different binding constant were present.

Rudnev et al. (2005) [27] used the HD method to determine binding constants between cisplatin and oxaliplatin to the serum transport proteins transferrin and albumin. The authors took note of the limitations of the HD method and verified before analysis that µC=µP. Since the kinetics of the

reactions were slow, the normal HD method could not be applied in its standard form. To make sure an equilibrium situation was reached, the samples were incubated with drug at different

concentrations before injection. The samples were then analysed using varying concentrations of drug in the BGE. After applying a correction factor for the drug already present in the sample, the amount of bound drug could be determined. The values for n and K were then determined, using nonlinear regression fitting equation to a plot of r vs Df.

6.2 Affinity capillary electrophoresis

Affinity capillary electrophoresis (ACEm), in ACEm one component is added to the buffer in varying concentrations, while the other is injected at a consistent volume and concentration. A drug could for instance be added to the buffer, while an interacting protein is injected. The difference in the mobility of the protein is then monitored as a function of the drug concentration in the buffer. The method is identical in experimental setup to the HD method. [24]

Since this method calculates the binding constant based on the change in electrophoretic mobility, possible factors that vary between measurements that affect it, such as the viscosity, ionic strength and capillary wall adhesion, should be eliminated or corrected for. This method is best applied, when complexation leads to a large change in electrophoretic mobility, since binding constants are based on the electrophoretic mobility change. Binding constants for 1:1 ratios can be determined by fitting of nonlinear regression functions or by linearized equations, in the section quantification this is elaborated upon further. [15] While calculating the binding constant, the amount of free D is taken to be equal to the concentration of D in the buffer. Since this method is equal in experimental setup as the HD methods the simulated elution profile (figure 2) is identical. In situation A in which µC=µP>µD it is impossible to determine the binding constant with ACEm, since there is no mobility

shift depending on the amount of complex formed. The binding constant is always over or underestimated in case B and C, since the amount of free D in the P zone is never equal to the amount of D in the BGE. [24]

Chinchilla et al. [28] demonstrated the use of ACEm in combination with a multiple injection method. This method was demonstrated by injecting a variety of vancomycin plugs with a plug of ligand between each injection at varying concentration . A schematic representation of the filled capilairy at the start of the separation can be seen in figure 3.

(16)

13

Figure 3: A schematic presentation of the capillary at the start of the separation the black zones represent the vancomycin peaks, the grey zones are ligand plugs which are varied in concentration. The resulting electropherogram shows the different vancomycin plugs, copied from [28].

At the start and at the end of the injection cycle different non-interacting standard is injected. After the injection protocol electrophoresis is performed. This analysis is performed with different

concentrations of ligand in the buffer plugs, so that the change in the electrophoretic mobility can be monitored as a function of the concentration of ligand. This method enables the determination of binding constants of species in a limited number of runs (one per ligand concentration), and gives more data points allowing more robust quantification. This method was demonstrated to determine the binding constant between vancomycin, carbonic anhydrase B and teicoplanin with their

respective ligands. The binding constant was determined using linear Scatchard plots. [28] Partial filling ACE methods

Another variation of the set of variations on the ACE method are partial filling methods. [29] These techniques are useful when there is not enough sample present for standard ACE methods. In this method, the capillary is first partially, instead of fully, filled with a ligand before the sample and a standard are introduced. The change in the electrophoretic mobility as a function of the

concentration of ligand is used to determine the binding constant. In this case the sample and ligand plug elute simultaneously at the point of detection. The electropherogram will typically show a large platform caused by the ligand plug on which the peak of the protein can be seen. Analysis of the data is then performed in a similar manner as in normal ACE.

Another partial filling method is the flow-through partial filling method. In this method, a smaller ligand plug is injected and the sample plug flows completely through the ligand plug before detection. A condition for using this method is, that the analyte elutes faster than the interacting ligand in the capilary. [29] When the ligand used causes a significant background signal in UV

detection, they also offer a good alternative to conventional ACE. Another advantage is the ability to perform the separation without introducing the ligand in to the MS when using MS detection. [30] This method was recently applied by Farcas et al. (2017) [31] to monitor interaction between thrombin and peptides as a proof of concept for the use of partial filling ACE in fragment based drug discovery. The capillary was partially filled with thrombin and a mixture of potential binding

fragments was injected. The shift of the fragments was determined compared to a run without thrombin present. The shift in electrophoretic mobility was used to determine if there was an interaction present or not. The method was performed under conditions mimicking physiological ones and proved to be a reliable screening approach.[31]

The partial filling method can also be used in combination with competitive analysis.[29] Competitive affinity capillary electrophoresis can be performed when a ligand does not significantly affect the

(17)

14

electrophoretic mobility of the protein, this might be the case when the ligand is small and has no charge. The capillary is partially filled with a charged ligand and the sample is injected with a marker. The sample and the charged ligand interact and form a complex. Then the neutral ligand is injected at varying concentrations and it flows into the domain of the complex that is established. Since the competition of the neutral compound with the charged compound does result in a change in electrophoretic mobility, this change can then be used to determine the binding constant. [29]

6.3 The Vacancy peak method

In the vacancy peak (VP) method the capillary is filled with a buffer containing both protein and drug. One of the compounds is used in a fixed concentration while the concentration of the other is varied. Instead of injecting one of the compounds, a small volume of neat buffer is injected. This results in two negative peaks in the electropherogram. One peak is caused by the vacancy of protein and complex, the other by the vacancy of drug. [24] The negative peak areas corresponding to free protein and free drug can then be used to determine the binding constant. This method uses more sample than other ACE methods since the sample must be present in a significant concentration in the buffer. The optimization of the method can be difficult since both drug and protein must contribute to the background signal. This background signal must not be too low or too high, since both cases would result in a loss in sensitivity. [15] The binding constants can be determined, as well as the binding stoichiometry. The VP method will always contain a systematic error if,the results approximate the true value the best when µC=µP in other scenarios the error will be larger. The

systematic error is caused by the dissociation of the complex in the D vacancy zone which yields a systematic error. This will cause the negative peak area to be too small, which leads to

overestimation of the amount of bound drug. In figure 4 a simulated elution profile can be seen in different scenarios.

Figure 4:Simulated elution scheme of the VP method in three different scenarios, A µC=µP>µD, B µC>µP>µD and C µC<µP>µD,

copied from [24].

In case A the amount of free D can be seen by the negative peak in the D profile. As discussed earlier the amount of free D is not correct since complex in that zone will dissociate. This generates extra free D causing the amount of free D to be to be underestimated, which causes the binding constant to be overestimated. In case B two negative peaks are visible in the elution profile of D. In this case

(18)

15

the complex migrates faster that the protein causing a zone lacking complex at the front edge of the zone. This cause P to consume free D to retain equilibrium, which causes the second negative peak. In this case the amount binding constant is overestimated. In case C the opposite happens causing underestimation of the binding constant. The binding constant can be determined using the same method as used in the HD method.[24]

6.4 Vacancy affinity capillary electrophoresis method

The vacancy affinity capillary electrophoresis method (VACE) is experimentally identical to the VP method. However, in this method the binding constant is determined based on the difference in the mobility of the negative peaks relative to the concentration of one of the compounds in the buffer. The concentration of drug is varied while the concentration of protein is kept constant or vice versa. Much like the VP method these values will always contain a systematic error. The elution profiles are identical to those shown in figure 4. When both D and P are detected, the migration shift of both compounds can be used for quantification. VACE suffers from the same complex dissociation in case A as the VP method causing overestimation of the binding constant. In case B and C the binding constants are overestimated or underestimated depending on the vacancy used to determine the binding constant. [24] Dvorak et al. (2013) [32]performed simulations comparing the ACE and VACE methods. The experiments showed that the ACE method proved to be more reliable and consistent with the used equation. They theorized that the concentration of the ligand was not the only factor that affects the electrophoretic mobility in in VACE, but that changes in buffer and analyte also affect the determined Kd.

6.5 Frontal analysis method

The frontal analysis (FA) method is quite different from the other ACE methods. The capillary is filled with neat buffer and a large sample plug is injected. The sample consists of a mix of protein and drug. The concentration of one of the compounds is varied while the other is kept constant. The working principle of the method is that if there is a sufficient difference in mobility between the drug and the complex and protein, the drug will leak out of the peak creating a plateau. The height of the plateau represents the amount of free drug in the sample. [24] An important condition for the use of this method is that the electrophoretic mobility of the drug is not equal to the electrophoretic mobility of the protein, otherwise the second plateau would not form. [15] Similar as in other methods this method yields the most accurate results when µC=µP>µD. In the case of µC>µP>D the binding values

will be too high since the amount of free D is underestimated. In the case of µC<µP>D the plateau

height is too high resulting in a determination of K which is too low. In figure 5 simulated elution profiles of the FA method in three different scenarios can be seen.

(19)

16

Figure 5: Simulated elution scheme of the FA method in three different scenarios, A µC=µP>µD, B µC>µP>µD and C µC<µP>µD,

copied from [24].

In case A the drug dissociates out of the sample plug causing a region with only P to from at the front of the plug. The amount of free D is accurate and the binding constant is determined correctly. In case B the mobility of the complex is higher migrating in to the free P zone. This causes the D zone to broaden to the front causing a lower plateau height and an overestimated binding constant. In case C the opposite happened the protein migrates out of the zone causing complex to dissociate into free D. This causes a higher platform of free D and an underestimation of the binding constant. [24] Recently Qian et al. (2017) [33] reported a method which could be used to correct for these over and underestimation of the binding constant, by applying a correction based on the electrophoretic mobility of the complex, free drug and protein. The model showed good result and was able to correct for errors in the determined binding constant caused by differences between µC and µP. [33]

Kang et al. (2016)[34] used the FA method to screen for inhibitors of protein-protein interactions. The method was developed using the protein Bcl-XL, a known interacting fluorescently labelled

peptide and a known inhibitor. The effect of the inhibitor can easily be studied by a change in the plateau height when the complex formation is inhibited and more free fluorescent peptide is present. The method used pooled inhibitor samples to increase throughput and two new inhibitors were found. By comparing the normal binding curves, to those with various amounts of inhibitor present, the inhibitor concentration needed to dissociate 50% of the complex could be determined.

(20)

17

A variation on the FA method is the frontal analysis continuous capillary electrophoresis (FACCE) method. This method was first developed by Muhoberac et al. in 1997 [35] to determine the binding constants of polyelectrolytes to proteins. This method employs continuous sampling during analysis contrary to the traditional FA method. At the start of the analysis the capillary is filled with buffer and equilibrated. After equilibration, the inlet of the capillary is placed in the sample vial and the current is applied which initiates the separation. In this method, the sampling and separation are thus not separated into two different stages. The resulting electropherogram consists of different stacking plateaus resulting from different electrophoretic mobilities of the analytes. Muhoberac et al. applied this method to study the binding between β-lactoglobulin and sodium poly(styrenesulfonate). In this situation, the polymer can bind multiple protein molecules since its high MW (465 kDa). The

electropherogram of this separation can be seen in figure 6, the first plateau is caused by the free β-lactoglobulin the second plateau is caused by the complex. The method offers multiple advantages, it enables the analysis of systems with slow kinetics since the sample is injected when its already equilibrated. Furthermore, it enables detection at lower concentrations due to the robust plateaus formed and it requires only a single

calibration curve of the plateau height of the free ligand.

6.6 Kinetic capillary electrophoresis

Krylov et al. (2005) [36] describes the theory behind kinetic capillary electrophoresis (KCE) and introduces a set of methods for the study of interactions using based on this model. Kinetic based models differ from the earlier described models in the sense that they do not assume an equilibrium between complex and free protein and drug. These methods instead determine the rate constant of association and dissociation separately, in a non-equilibrium condition and can determine

equilibrium conditions from this data.

While these methods seem very promising they do require more complicated mathematical analysis of the data and have not been widely introduced. Therefore, they will be left out of scope in this review.

Figure 6: electropherogram of the FACCE method demonstrated by Muhoberac et al., copied from [35]

(21)

18

6.7 The selection of an ACE method

Many factors must be considered when selecting an analysis method, the principles of all the relevant ACE methods have been discussed above. To assist in easy method selection some charts and tables summarizing the applicability and strengths and weaknesses of the discussed methods will be given. In figure 7 a flowchart for selecting an ACE method can be found. In Figure 8 8 the most important information about each method is mentioned.

(22)

19

(23)

20

6.8 Quantification in ACE

There are multiple methods for determining binding constant from data acquired with the ACE methods mentioned above.[37] Some of the most commonly used methods will be discussed. The binding constant itself can simply be defined as:

Equation 3:

𝐾𝑎 = [𝐶] [𝑃][𝐷𝑓]

In which [C] represents the complex concentration, [P] the concentration of protein and [Df] the

concentration of free drug. The binding ratio r can be then be defined as:

Equation 4:

𝑟 = [𝐶] [𝑃] + [𝐶]

When there are m binding possibilities, ni binding sites and the bindings do not influence each other,

then Ki can be defined as the association constant for a certain binding, Df is the concentration of the free drug and r can be taken as the average amount of bound ligand per binding molecule. Then r can be defined as: Equation 5: 𝑟 = ∑ 𝑛𝑖𝐾𝑖𝐷𝑓 1 + 𝐾𝑖𝐷𝑓 𝑚 𝑖=1

In a simple system where there is only one type of binding, where m=1 the equation can be simplified to:

Equation 6:

𝑟 = 𝑛𝐾𝐷𝑓 1 + 𝐾𝐷𝑓

In systems with two types of binding Equation 7 can be applied as derived from Equation 5.

Equation 7:

𝑟 = 𝑛1𝐾1𝐷𝑓 1 + 𝐾1𝐷𝑓

+ 𝑛2𝐾2𝐷𝑓 1 + 𝐾2𝐷𝑓

Equation 5-7 are commonly applied to the quantification of binding constants and binding

stoichiometry’s when the HD, FA and VP methods are used. These equations can be fitted to plots of r vs Df using nonlinear regression with least squares fitting. In situations with multiple equilibria these non-linearized methods are the most suited for quantification .[24] These equations can also be linearized in a so called Scatchard plot and results in a linear equation if r/Df is plotted against r. Scatchard analysis is traditionally the most common method for data processing in the study of interactions with CE. The Scatchard plot can be seen in Equation 8.

Equation 8:

𝑟 𝐷𝑓

(24)

21

Another form of the Scatchard plot, also called the Klotz equation, is [37]:

Equation 9: 1 𝑟 = 1 𝑛+ 1 𝑛𝐾𝐷𝑓

The Scatchard plot will show a straight line when a 1:1 complex is present, when there are multiple binding sites the line will deviate but several straight parts might still be present.[24] The use of nonlinear Scatchard plots is prone to errors, Kermode (1989) [38] points out many factors which can lead to nonlinear Scatchard plots, such as affinity differences between labelled and unlabelled ligands, nonspecific binding and other factors.[38] So while nonlinear Scatchard plots can be used to determine multiple binding constants, a nonlinear regression approach using equation 5 is more reliable. [24]

In ACEm, the binding constant is determined based on the change electrophoretic mobility. In that case the Scatchard plot can be written as:

Equation 10:

∆µ [𝐷𝑓]

= 𝐾 ∗ ∆µ𝑚𝑎𝑥− 𝐾 ∗ ∆µ

In this formula ∆µ is the change in mobility, [Df] the concentration of the drug in the buffer, K the

binding constant and µmax is the difference in electrophoretic mobility between protein and complex. Compared to Equation 4 it does not take in account the number of binding sites n. This means that ACE does not give information about the binding stoichiometry and can therefore generally only be applied to 1:1 binding stoichiometry systems.[37]

When determining binding constants based on VACE analysis equation 11 can be used [15].

Equation 11:

µ𝑃,𝐷 = µ𝑃,0+ (µ𝑃,𝐷,𝑚𝑎𝑥− µ𝑃,0)

𝐾[𝐷𝑓]

1 + 𝐾[𝐷𝑓]

In which µP,D is mobility of the protein in the presence of drug, µP,0 is the mobility with no drug

present and µP,D,max is the mobility of the complex.[15]

More data can be extracted from VACE measurements such as the binding stoichiometry. This can be done using a method described by Busch et al. [39].

Zierler (1989)[40] describes the correct use of nonlinear Scatchard plots. To interpret a nonlinear Scatchard plot it can be resolved into multiple straight lines. This is done by computer analysis using least squares fitting of equation 5. Zierler emphasizes that nonlinear Scatchard plot can only be resolved to find multiple binding constants by computer analysis or by a graphic method described by Rosenthal (1967)[41].

Nonlinear Scatchard plots cannot simply be interpreted by eye or by extrapolating the data points. The curve should be seen as a sum of two (or more) linear functions which reconstruct the original curve. The correct and incorrect manner of resolving a Scatchard plot into linear functions can be seen in Figure 9.

(25)

22

Figure 9: a shows the correct method to resolve a nonlinear Scatchard plot performed using a computer program, b shows an incorrect method since the dotted line reproduced using the linear lines does not reproduce the original curve, based on an image copied from Zierler (1989) [40]

Another point made by Zierler is that it is important to make sure that the experimental data points cover enough of the range to enable extrapolation to the intercept. Often there are no data points close to the intercept which can lead to misinterpretations of the data. In some cases linearity of the whole range is claimed while that might not be the case if data points closer to the intercept were taken.[40]

The methods mentioned above only apply to cases where no interaction between the binding sites is present. [42] The McGhee- von Hippel equation is a model used to describe the nonspecific binding of large ligands, such as proteins, to homogeneous one-dimensional lattices, such as certain types of EPS. This type of binding is also called multiple contact or multivalent binding. This model results in curved plots unlike traditional linear Scatchard plots. The model is based on three parameters, the binding constant K, the cooperative parameter 𝜔, which is unitless and n, the number of sites covered by a single ligand. The model can then be expressed in to following form[42]:

Equation 12: 𝑣 𝐿= 𝐾𝜔(1 − 𝑛𝑣) ∙ 1 + (2𝜔 − 𝑛 − 1)𝑣 + 𝑄 (2𝜔 − 1)(1 − 𝑛𝑣) + 𝑣 + 𝑄 ∙ ( 2𝜔(1 − 𝑛𝑣) (2𝜔 − 1)(1 − 𝑛𝑣) + 𝑣 + 𝑄) 𝑛−1

In which v is the average amount of ligand per binding site, L the free ligand activity and Q is equal to:

𝑄 = √(1 − (𝑛 + 1)𝑣)2_{+ 4𝜔𝑣(1 − 𝑛𝑣)}

While this model is more complicated than the other described models it is the only one which can be used to accurately describe the binding between proteins and large polysaccharides. Therefore it is crucial to accurately quantifying the binding between EPS and whey proteins with ACE methods. This method has been applied by Girard et al. [43] to quantify binding interactions between pectin and β-lactoglobulin.

(26)

23

6.9 Examples of analysis of binding interactions with ACE

Protein-polysaccharide interactions

ACE has been applied by Yang et al. (2009) to determine the binding constant between sulphated polysaccharide 916 and human serum albumin (HSA). The polysaccharide, consisting of a 1,4-linked b-D glucosamine with on average half a carboxymethyl group per monosaccharide and one sulphate group was added to the separation buffer. The HSA was then injected at a fixed concentration and volume. In figure 10 the resulting electropherograms can be seen.

Figure 10: The electropherograms of the ACE experiments, spectra A to F are from measurements with concentrations ranging from 0 to 2 x 10-3 M polysaccharide in the buffer. The P peak indicates the protein signal, the NM signal is a neutral marker, copied from [44].

The polysaccharide gives no signal in UV detection and is thus not visible in the electropherogram. The separation was performed in a 20 mM sodium phosphate buffer at a pH of 7.4. The applied voltage was 15 kV during the separation and the capillary was kept at a constant temperature of 25 °C. A neutral marker (benzyl alcohol) was added to correct for the change in viscosity of the buffer. The change in the mobility of the protein is otherwise caused by both the change in viscosity of the buffer and the binding with the polysaccharide. This would result in an error in the determination of the binding constant if no correction for the viscosity change is made based of the neutral marker. The neutral marker was injected before the polysaccharide to ensure that it did not affect the interactions between the protein and the polysaccharide. The concentration of protein was kept constant at 10-5_M.

To determine the binding constant a Scatchard type plot was used in the form of equation 13. The equation was modified to determine Ka based on electrophoretic mobility changes and to correct for viscosity changes with the neutral marker.

Equation 13:

1 𝑟=

1 𝐾𝑎 ∗ [𝑃] + 1

in which r is the ratio of bound protein vs total protein, [P] the molar concentration of protein and Ka the binding constant.

(27)

24

Supposing that there is a difference it the electrophoretic mobility between the complex and the protein there must be a change of the migration time if the polysaccharide is added and a complex is formed. This can be expressed by:

Equation 14:

µ𝑃𝑆 = (1 − 𝑟)µ𝑃+ 𝑟µ𝑃𝑆

In this formula µ𝑃𝑆 is the electrophoretic mobility at a certain polysaccharide concentration, µp is the

mobility of the protein with no polysaccharide present and µPSis the mobility of the complex. This

means that the mobility of the protein peak at a certain polysaccharide concentration is the average of the mobility of the complex and the protein depending on the ratio in which they are present. The equation can be rewritten as Equation 15.

Equation 15:

𝑟 = µ𝑃

𝑆_{− µ} 𝑃

µ𝑃𝑆 − µ𝑃

The change in electrophoretic mobility can thus be used to determine the ratio between bound and unbound protein. From these equations the equation which can be used to determine the binding constant can be derived in the form of equation 16.

Equation 16: 1 ∆µ_𝑃𝑆 = 1 𝐾𝑎 ∗ ∆µ𝑚𝑎𝑥 ∗ 1 [𝐴]+ 1 ∆µ𝑚𝑎𝑥 In which ∆µ𝑃𝑆= µ𝑃𝑆-µ𝑃 and ∆µ𝑚𝑎𝑥 = µ𝑆𝑃−µ𝑃

This function can be used to determine the binding constant when transformed into a linearized form. If 1/[A] is set on the X-axis and 1/∆µ𝑃𝑆 is set on the Y-axis then the binding constant Ka is equal

to the ratio between the intercept and the slope. While this formula could be used in this form it does still not take in to account the change in viscosity due to the addition of the polysaccharide. This correction was done using a neutral marker. The correction results in Equation 17.

Equation 17: 1 ∆𝑀𝑆𝑃 = 1 𝐾𝑎 ∗ ∆𝑀𝑚𝑎𝑥 ∗ 1 [𝐴]+ 1 ∆𝑀𝑚𝑎𝑥 In which ∆𝑀𝑆𝑃 = µ_𝑃𝑆 µ𝑁𝑀− µ𝑃 µ𝑁𝑀 and ∆𝑀𝑚𝑎𝑥= µ𝐴𝑃 µ𝑁𝑀− µ𝑃 µ𝑁𝑀

This equation corrects for the change in viscosity due to a change in the amount of polysaccharide. This correction is performed by using the mobility of the neutral marker at a certain polysaccharide concentration to correct the measured mobility of the protein.

The electrophoretic mobilities were calculated with a correction for the path length due to the injection protocol. By plotting _∆𝑀1

𝑆𝑃 on the Y-axis and

1

[𝐴] in the X-axis of resulted in a linear equation

with a slope of -0.00013 and an intercept of -27.196. By dividing the slope by the intercept, the Ka was calculated to be 2.1x 10-4_.[44]

Studying protein-protein and protein-polysaccharide interactions

Figure: Enzyme-inhibitor complex between trypsin (yellow) and

the Bowman-Birk inhibitor (red) from bitter gourd, copied from [5].

Wouter Knol

23-01-2018

MSc Chemistry

Analytical Sciences

Literature Thesis

Studying protein-protein

and protein-polysaccharide interactions

by

Wouter Knol

January, 2018

Supervisor:

dr. W.T. Kok

Daily Supervisor:

dr. E. Kaal

Table of contents

1. Introduction

2. EPS and EPS protein interactions

3. Protein-protein interactions

4. Overview of methods for determining binding interactions

5. Capillary electrophoresis

6. ACE methods