WO2011040876A1

WO2011040876A1 - Screening of binders on immobilized native membrane proteins

Info

Publication number: WO2011040876A1
Application number: PCT/SE2010/051052
Authority: WO
Inventors: Anders Karlsson; Roger Karlsson; Kenneth Olesen; Owe Orwar
Original assignee: Nanoxis Ab
Priority date: 2009-10-02
Filing date: 2010-09-30
Publication date: 2011-04-07
Also published as: WO2011040876A9

Abstract

The present invention relates to a method for characterizing the binding interaction of a candidate binder to a target molecule comprising - providing a microfluidic flow cell having - an inlet; - at least one supporting solid surface with at least one membraneophilic region; - a covering layer that is at least partially immobilized to the membraneophilic region, said covering layer consisting of (i) a surfactant membrane, (ii) a lipid mimicking polymer, (iii) a surfactant or emulsion system, or (iv) a liquid crystal, or a combination thereof; - the target molecule included in or bound to, connected to or associated with the covering layer; and -an outlet; said method further comprising - contacting the target molecule with the candidate binders by creating an eluent flow, comprising the candidate binders, from said inlet to said outlet; - detecting and/or identifying said candidate binders in the eluate - characterizing the binding interaction of the candidate binder to the target molecule based on the detection and/or identification of the candidate binder in the eluate. The invention also relates to a device adapted for use in said method.

Description

SCREENING OF BINDERS ON IMMOBILIZED NATIVE MEMBRANE

PROTEINS

BACKGROUND OF THE INVENTION

Screening of compound libraries against targets have been described in many aspects using various methods. Goals include identification of binders to specific targets, ranking a collection or library of candidates against a specific target according to binding strength and identification of epitopes involved in protein-protein interactions to understand protein structure and function. In many of the described techniques for screening, elaborate procedures are required to keep track of individual members of the library, and in order to categorize the individual members, labeling techniques are commonly employed, as are array techniques. Methods to construct large libraries of chemical compounds, for example through combinatorial chemistry, have made it possible to generate vast collections of ligands/binders against interesting and potential drug targets. In a related approach, the creation of these vast libraries can be done using phage display or similar technologies. In phage display, the DNA of a protein or peptide of interest is inserted into the phage gene for the phages own coat proteins. When infecting E. Coli cells, the phages will express and "display" the protein or peptide of interest on its surface.

Through this, the need for efficient and rapid high throughput screening methods for these libraries has increased dramatically.

There is an increased interest in mass spectrometry (MS) as a sensitive tool to detect, identify and resolve components of a complex library. Other detection methods such as optical detection have limitations in resolving the complexity of compound mixture due to spectral overlap of individual components. MS has been used as a pinpoint tool to characterize active library compounds in molecular recognition studies. MS based screening methods have also been used in combination with the principle of capture and release mechanisms, where ligands/binders are allowed to interact with target molecules, in some cases immobilized on solid supports (capture event), followed by a release event where the ligand-target interaction is broken. As an example, WO 97/43301 describes this approach, where the release event is done by certain displacers, chaotrope agents, pH-, salt- and temperature gradients, organic solvents, selective denaturants and detergents. MS is then used to continually follow elution of binders during the release event. During the different elution conditions, weak binders elute first followed by stronger binders. In general though, deviations from normal physiological (and relevant) conditions may induce conformational changes of one or both binding partners, thus destroying even strong interactions. Furthermore, capture and release methods often have limitations in throughput. Phage display is performed in a similar way by infusing a library of different phages across a target of interest and only those phages that display a protein that binds to the target will remain bound, while the rest are washed away. The bound phages are then eluted and the DNA of interest isolated and used to identify the interacting protein or peptide.

One approach to circumvent some of the issues described above is frontal affinity chromatography (FAC), which can be used with MS as a detection method. In frontal chromatography, the targets are immobilized on solid supports and typically packed in a column and ligands/binders are then continuously infused over the targets at certain concentrations and flow rates. Ligands/binders that have an affinity for the target will interact with the stationary phase and elute slower. Eventually though, the capacity of the column combined with binding affinity to target, will cause each ligand/binder to elute at their "breakthrough volume" for given concentrations of ligands/binders. Ligands or binders with weak interaction, intermediate interaction or strong interaction to assayed target will elute in the order; early, intermediate or late, respectively. Using MS, the identity and breakthrough volume for each assayed ligand can be determined. Breakthrough volumes can then be sorted to exclude non-binders from binders. Selected binders can be ranked relatively to ligand-target binding affinity.

In order to assay ligands, immobilization of functional target molecules to surfaces is a vital, albeit difficult step. For example, immobilization of functional target molecules enables a way to construct arrays of surfaces, each tailored with selected target(s). These arrayed surfaces, usually denoted microarrays as a collective name, can be exposed to a ligand library to study binding interactions. The use of different microarray based assays are continuously increasing and are nowadays an important tool for the sorting and controlled handling of samples for numerous screening purposes, for example complete genome micro arrays exist for gene expression studies.

In all examples above, binding interactions are strongly dependent on a correct structural conformation of both receptor and ligand. Therefore, when immobilizing the targets on surfaces, it is of utmost importance to keep the structure of the target intact. Usually some sort of modification of the target is needed in order to couple it to surfaces. Many different approaches have been made in order to make sure that the target is correctly folded with intact structure after being modified and also correctly oriented when exposed to its binding partners. Methods include the use of various types of tags that can be introduced through genetically engineered target or by chemical modification of the target molecule. One exemplary approach is to utilize biotin-streptavidin interactions.

A special class of targets is membrane proteins that are situated or associated in some manner to the lipid bilayer membrane of cells. They differ from soluble proteins by their inherent nature of being dependent on a hydrophobic environment, thus making them hard to solubilize in water-based solutions with retained structure and function. Membrane proteins are of great importance and have several vital functions, e.g. to transmit signals across membrane barriers and uphold concentration gradients. It has been estimated that as many as 2/3 of all drug targets are membrane proteins, even further highlighting the importance of membrane proteins. Immobilization of native membrane proteins presents a number of problematic issues. First of all, correct protein folding and structure must be retained and stabilized by a hydrophobic environment in order for the membrane protein to function and bind ligands correctly. Different approaches have been taken to make certain that function is retained. For example, specific detergents or mixtures of detergents have been used to extract membrane proteins and immobilize them in the presence of the detergent(s) in order to stabilize its structure. This approach is very much target dependent and often a lot of effort is done to find the right detergent(s) that extracts and stabilizes the membrane protein of interest. Other approaches include immobilization of targets in synthetic lipid membranes, where targets that have been extracted by detergents are re- inserted or reconstituted into synthetic lipid membranes. One such approach is described in WO 9942511, where receptors are immobilized onto artificial membrane supports in a column format. The artificial membrane support in that particular approach consists of a monolayer of lipid that coats beads that are packed in a column. In general though, membrane proteins that are integral to cell membranes span across the entire lipid bilayer in order to have correct folding and structure and therefore this approach is only applicable to a select number of membrane associated species. Also, lipid bilayer membranes that are directly coupled to surfaces still have trouble retaining the correct membrane protein structure due to interactions with the underlying support surface, since many integral membrane proteins have domains that protrude out from the lipid bilayer membrane. Therefore alternative ways to immobilize native membrane protein, preferably with the correct lipid composition around the membrane protein targets, is beneficial. More and more attention is given to the fact that the lipid composition of the lipid bilayer membrane may affect the membrane proteins structure and folding and hence its activity. Further, there are numerous examples of specific and tightly bound lipids to the membrane proteins; these lipids may play a crucial role in the structure function relationship.

The need for quick and easy handling and immobilization of intact and native membrane proteins for screening purposes is high. An efficient and controllable exposure of the ligands to the immobilized membrane protein targets will enable ways for sensitive frontal chromatography applications.

The present invention provides solutions to these technical issues by utilizing a microfluidic flow cell to perform frontal affinity chromatography.

SUMMARY OF THE INVENTION

In a first aspect, the present invention relates to a method for characterizing the binding interaction of a candidate binder to a target molecule comprising

- providing a micro fluidic flow cell having

- an inlet;

- at least one supporting solid surface with at least one membraneophilic region;

- a covering layer that is at least partially immobilized to the membraneophilic region, said covering layer consisting of (i) a surfactant membrane, (ii) a lipid mimicking polymer, (iii) a surfactant or emulsion system, or (iv) a liquid crystal, or a combination thereof;

- the target molecule included in or bound to, connected to or associated with the covering layer; and

-an outlet;

said method further comprising

- contacting the target molecule with the candidate binders by creating an eluent flow, comprising the candidate binders, from said inlet to said outlet;

- detecting and/or identifying said candidate binders in the eluate

- characterizing the binding interaction of the candidate binder to the target molecule based on the detection and/or identification of the candidate binder in the eluate.

In a preferred embodiment the target molecule is a protein, more preferably a membrane protein.

The present invention further relates to a flow cell design for easy and efficient exchange of fluid around immobilized native membrane proteins for frontal

chromatography applications and library screening of binders. The present invention makes use of the methods in WO 2006068619 to immobilize native membrane proteins for the purpose of using frontal affinity screening of ligand libraries towards membrane protein targets. WO 2006068619 describes ways to prepare and immobilize small proteo liposomes onto surfaces that are denoted membraneophilic. As compared to other approaches using frontal chromatography, where columns with column material that immobilizes/traps membrane material has been employed, a micro fluidic flow cell has been constructed for the same purpose, which is to select and/or rank potential binders against interesting targets (with or without the specific determination of binding affinity for individual binders).

Accordingly, in one of its method aspects, the present invention describes a device to efficiently screen libraries of ligands/binders against immobilized native membrane proteins comprising methods to: (a) prepare and provide proteoliposomes containing native membrane proteins, (b) immobilize the proteoliposomes onto surfaces inside a flow cell, (c) exchange fluids by laminar flow through the flow cell, (d) to apply a

ligand/binder library by a laminar flow through the flow cell under frontal affinity conditions, (e) to detect and identify ligands/binders using various detection systems, preferably mass spectrometry, continuously or by analyzing collected fractions from a fraction collector or similar device, (f) to select and/or rank the ligands/binders according to their break through volume, which is related to their binding strength, (g) to determine binding affinity constant(s) for ligand(s).

Preferably, the membrane preparation is prepared without the use of detergents in order to retain the functionality of the membrane proteins and the lipid composition around the membrane proteins.

In a related preferred embodiment the membrane preparation has low cytosolic protein concentration or soluble contaminants. In such a preferred embodiment the cytosolic or soluble contaminants have been removed or minimized in favor of the purity of the membrane fraction.

Furthermore, such preferred preparations are generally done by breaking the outer barrier of the cell, usually the plasma membrane of cells, a procedure called lysis, followed by centrifugation and collection of the membrane fraction.

Furthermore, in a related aspect, the membrane fraction can be washed with pure water or buffer solutions with high pH or high salt concentrations to even further reduce the soluble or cytosolic contaminants and reduce the amount of membrane associated species in favor of integral membrane proteins. Certain harsh wash protocols may however affect the membrane protein(s) activity and this must be checked for the target protein of interest. In another preferred embodiment, the membrane preparation is transformed into small proteoliposomes using various methods, preferably ultrasonication,

tip/probesonication and extrusion procedures as non- limiting examples.

In another related aspect, the resulting proteoliposomes are 10-100 000 nm in diameter, more preferably 20-500 nm in diameter.

In another preferred embodiment, the proteoliposomes are prepared using the methods described in "PLASMA MEMBRANE VESICLES AND METHODS OF MAKING AND USING SAME" and Bauer et al. 2009 (Angew Chem Int Ed Engl. 2009;48(9): 1656-9, where vesicles or blebs are budded of directly from the plasma membrane of cells.

In a related embodiment proteoliposomes may also be prepared by reconstitution procedures where a purified target is needed. In such a case, membrane proteins are extracted by detergent(s) and re-inserted into synthetic lipid bilayers or extracted natural lipids.

In another aspect, proteoliposomes may consist of natural vesicles from cells emanating from exocytosis or similar events.

In a related embodiment, virus-like particles (VLP) as those produced by

MembranePro™ FPE System, from Invitrogen, or those produced by technologies from Integral Molecular (US 7763258 B2) may also be used as proteoliposomes. In brief, the particles from the MembranePro system are created by cells that have been induced to express a viral core protein called gag. The gag protein cores are budded from the cells, more specifically from the lipid raft areas of the plasma membrane, and thus capture and display the contents in the lipid rafts. These particles therefore contain a high density of GPCRs, since these types of membrane proteins are often situated in lipid raft areas of the plasma membrane. Also, the particle preparation does not contain the same amount of contaminating material. In this aspect, contaminating material can be membranes arising from other parts than the plasma membrane, such as Golgi, ER, mitochondria etc., but also soluble proteins from the cytosol.

In another embodiment intact bacteria can also be immobilized. Preferably the flow rate should be kept low when exchanging fluid through the flow cell due to the drag forces created on the particles from the flow. Large particles will experience a larger drag force especially at high flow rates.

In a preferred embodiment, the proteoliposomes contain targets of interest, including transmembrane proteins, e.g., transmembrane alpha-helix proteins,

transmembrane beta-barrel proteins, lipid anchored membrane proteins, and peripheral membrane proteins. In related embodiments, the transmembrane proteins are selected from the group consisting of enzymes, transporters, receptors, ion channels, cell adhesion proteins, G-protein coupled receptors (GPCRs). In another related embodiment the targets can be selected from the group of lipid anchored proteins.

Preferably, the proteoliposome concentration used for the immobilization step should be 0.01-100 mg/ml, more preferably between 0,1-1,0 mg/ml, measured as the dry weight of membrane (total lipid/protein concentration).

Preferably, the amount of proteoliposome material bound to the surface should be between 1 ng and 100 μg per square centimeter, more preferably between 100 ng and 10 μg per square centimeter. In an exemplary embodiment the amount of proteoliposome material is approximately 1 μg per square centimeter.

Preferably, the ionic strength of the proteoliposomes suspension should be between 1 μΜ to 5 M, more preferably the ionic strength should be between 20-1000 mM, even more preferably the ionic strength should be between 200-500 mM. Under such preferred conditions as described above, immobilization is performed rapidly, typically within 20 minutes. Preferably though, immobilization is performed between 30 minutes to several hours, in order to make sure that the surface is saturated.

In a related aspect, blocking substrates (including but not limited to small molecules, peptides, proteins, lipids, lipid vesicles or small molecules that specifically target uncovered surface, non-limiting examples include DMSO, BSA, casein, dextrane) may be added in a second immobilization step to ensure that the surface is completely covered. Specific blocking of gold substrates has been demonstrated by PEG-thiols and PEG-functionalized alkane thiols. This is done in order to reduce the non-specific binding of ligands to exposed surfaces etc. In an exemplified embodiment, blocking of surfaces is performed by adding peptides modified with a thiol group.

In a related embodiment, in order to further reduce the exposed surface sites that may bind ligands non-specifically, a weak ligand may be added to saturate both the non- specific and specific sites. Stronger ligands will compete for the specific sites and release the weak binders.

In a preferred related embodiment the proteoliposomes are distributed over the entire exposed surfaces inside the flow cell, thus creating a homogenous distribution of target membrane proteins in the flow cell. In a preferred embodiment a wash or rinse step should follow the immobilization of proteoliposomes to the surfaces inside the flow cell. Preferably at least 2 flow cell volumes of buffer should be used to rinse the flow cell, more preferably 10-25 volumes of buffer should be used to rinse the flow cell. Preferably the wash buffer should be compatible with downstream applications, such as on-line MS analysis, and also be compatible with the surface immobilization and not detach the immobilized proteoliposomes.

In a related embodiment the buffer may consist of PBS (phosphate buffer saline), TRIS (2-Amino-2-(hydroxymethyl)-l ,3-propanediol), HEPES (4-(2- Hydroxyethyl)piperazine- 1 -ethanesulfonic acid), AMBIC (ammonium bicarbonate), AMAC (ammonium acetate), triethylammonium hydrogen carbonate or similar together with various salts including monovalent and divalent ion additives, such as NaCl, KC1, CaCi₂, MgSC>4 or similar, as non-limiting examples.

In one embodiment, the flow cell comprises at least one inlet and at least one outlet connection for creation of a flow of the liquid or the aqueous solution covering the covering layer and for the exchange of additional membrane, protein-lipid mixtures, washing solutions, staining solutions, digestive solutions and other solutions or suspensions around the covering layer. The flow cell could be connected to a detector, said detector detecting changes in concentration of substances such as binders/ligands when eluted from the device, in order to determine the breakthrough volume of said binders/ligands, when using frontal affinity chromatography.

In one embodiment, the supporting solid surface of the flow cell is planar and the membraneophilic surface region has a specific two-dimensional geometric shape. The supporting solid surface may also have a three-dimensional structure.

Preferably, flow rates of less than 100 nl up to 10 ml per minute may be used to exchange the fluid in the flow cell, more preferably 10 μΐ to 500 μΐ per minute is used to exchange the fluid inside the flow cell. Even more preferably, in a related embodiment, flow rates of 10-200 μΐ per minute are usable with on-line detection with MS. Preferably, the flow inside the flow cell is laminar.

In a preferred embodiment, the height of the channel should be between 0,01-500 μιη and the width of the channel 0.0001-1000 mm, more preferably a height 20-100 μηι and width between 1 - 100 mm, more preferably a width between 4- 100 mm. In a related preferred exemplary embodiment, a height of 30-60 μηι and a width of 25-30 mm for the channel dimensions. In the exemplary embodiment the aspect ratio is thus > 500. The flow can be described by thin liquid films, flowing with predictable laminar flow profiles through the flow cell, enabling an efficient and reproducible exchange of fluid.

Preferably, a flow of ligands/binders (as components of a compound library) may then be flowed across the immobilized native membrane proteins. In a preferred embodiment, the flow creates a homogenous ligand concentration exposure for all exposed targets, as compared to in a mesh or packed column or a gel, for example. In a preferred embodiment the flow is created using syringe pumps, liquid chromatography pumps or similar without limitations.

In one embodiment, Mass Spectrometry (MS) is used as detection method. In a preferred embodiment the flow of ligands is directly detected using an on-line coupling of the eluate from the flow cell into the MS instrument. Preferably, the mass spectrometer is an electrospray ionization mass spectrometer or a MALDI (matrix assisted laser desorption ionization) mass spectrometer In a related preferred embodiment the eluate is ionized by the method of electrospray. In such a preferred embodiment the eluate from the flow cell may be mixed with a second buffer containing some percentage of organic solvent in order to induce a good spray and ionization of ions in the eluate. In another related embodiment the MS analysis can be performed with other ionization techniques and MS instrumentations, such as MALDI etc. The sample can be spotted online on a MALDI plate or collected in fractions for subsequent spotting.

Preferably, the eluent buffer may consist of AMBIC (ammonium bicarbonate), AMAC (ammonium acetate), triethylammonium hydrogen carbonate or similar as non- limiting examples and the mixing buffer may consist of AMBIC (ammonium

bicarbonate), AMAC (ammonium acetate), triethylammonium hydrogen carbonate or similar as non-limiting examples with additives including alcohols, organic solvents and similar, such as methanol, ethanol, acetonitrile and similar solvents. The additives are preferably mixed into the mixing buffer at 0-100%, more preferably 5-50% in order to increase ionization and promote the formation of the electrospray. In a exemplary embodiment, wash buffer and ligand buffer that is flowed across the membrane protein targets is 5-20 mM ammonium actetate, pH 7,5, and the mixing buffer consist of 5 mM ammonium acetate with 20-60% methanol, pH 7,5.

In another related embodiment the binders are emanating from a phage display library, in which case the phages are flowed through the flow cell. Those phages that have epitopes that can interact with the immobilized membrane protein target will bind and those that do not have an interaction will be washed away.

In another related embodiment, the eluent may be collected by a fraction collector. The binders/ligands eluted from the device may be fractionated and analyzed separately after collecting said fractions. In such a preferred setup, ligands in buffer preferably as above, are flowed across the flow cell and volume fractions ranging from 1-1000 μΐ are collected, more preferably 10-100 μΐ fractions are collected using a fraction collector of any kind. The components of the eluate fractions are then preferably transferred to an autosampler of any kind and injected into an MS instrument (including as non-limiting examples electrospray ionization and MALDI) for analysis. The fractions may be chemically or physically treated in order to facilitate their analysis. The fractions may thus be processed and analyzed by for example Rapid Fire systems or similar and also desalted and/or focused prior analysis by various MS systems.

In a related embodiment the phages that bind to the immobilized membrane protein target can be eluted and collected for further analysis and identification of the binder.

In another related preferred embodiment, the MS instrument can be instructed to only detect certain ions, so called single ion recording (SIR) mode. It is thus possible to track and relate the SIR signal to the amount or concentration of the eluted species and determine the break-through time for all tracked species.

In another related embodiment the detection of ligands/binders in the eluted solution from the flow cell may be performed by several other different detection techniques, such as fluorescence, electrochemistry, UV-Vis detection, NMR, Infrared spectroscopy, Atomic Absorption Spectroscopy as non-limiting examples.

In a preferred embodiment, the methods above enable the determination of the relative affinities for ligands/binders of a ligand/binder library towards the immobilized native membrane protein targets in the flow cell, thus allowing a ranking of the ligands/binders. In another related preferred embodiment the methods described above enables the determination of dissociation constants for ligands/binders towards an immobilized native membrane protein target in the flow cell. In a related preferred embodiment the methods described may also be used to study effects of allosteric modulation on an orthosteric binding by detecting shifts in retention time in the presence and absence of allosteric compounds.

In a preferred embodiment, the binders referred to in this text are substances that are able to bind to and create complexes with membrane protein targets, using a variety of intermolecular forces such as ionic and hydrogen bonds as well as van der Waals forces. Strong binding reflects a high affinity for the binders to its target.

In a non-limiting exemplified embodiment, the binders can be classified as belonging to peptides, small molecules or phages. In a related embodiment the library size or the number of ligands/binders screened or flowed across the immobilized membrane proteins targets is preferably between 1 and 1 * 10¹⁰ different ligands/binders. More preferably the library size is between 1 and 1 * 10⁶ different ligands/binders. In a related aspect, the screen size or the number of different ligands/binders tracked simultaneously in a single run across one flow cell is preferably between 1 and 1 * 10⁴. Importantly, in this preferred embodiment the screen for these ligands/binders is done simultaneously across the same sample.

The invention is further specifically disclosed in the appended claims.

DESCRIPTION OF THE DRAWINGS

FIGURE 1 Figure showing the assembly of LPI Maxi FlowCell. The top and bottom plastic pieces are attached to each other by a tape constituting the actual flow chamber profile when the flow cell is assembled.

FIGURE 2 This is a time study that illustrates the flow profile in LPI™ Maxi FlowCell. A transparent Maxi FlowCell, produced for this specific purpose, was first filled with water. Green liquid was injected at a flow rate of 50 μΐ min using a syringe pump. Photos were taken every 30 seconds.

FIGURE 3 Different on-line setups for FAC-MS.

1. Ligand solution and mixing buffer solution is flowed with the flow rate Q 1 and Q2 - in this case Q1=Q2 or Q1≠Q2. The solutions are mixed using a mixing tee or similar device before entering the FAC-LPI FlowCell.

2. The ligand and mixing buffer solutions are mixed after the ligand solution has passed the FAC-LPI FlowCell. In this case Q1=Q2 or Q1≠Q2.

FIGURE 4 Off-line setup.

a) A solution of ligands is flowed through FAC-LPI FlowCell and the eluate is collected using a fraction collector. The fractions can also be processed and analyzed by for example Rapid Fire systems or similar and also desalted and/or focused prior analysis by various MS systems.

b) The fractions are then analyzed using an autosampler setup instructed to inject a fixed volume of the collected fractions into the detection system, in this example an MS instrument. FIGURE 5 An example for automated analysis and switching of ligand solutions across one or multiple FAC-LPI FlowCell(s). Ml, M2,..., M_n describes the different ligand solutions of a ligand library. The autosampler takes ligand solutions together with the HPLC pump to drive the different ligand solutions through a switching valve before injection into the FAC-LPI FlowCell(s), denoted CI, CI,..., C_n. Another pump can be used for washing steps, where the flow goes to waste. The eluate(s) are then mixed with mixing buffer before entering into the MS instrument for detection of ligands.

FIGURE 6 Demonstration of retention of a ligand to a target membrane protein. The membrane preparation contained overexpressed Ste2 in yeast.

The recorded SIR (single ion recording) traces of 3 tryptic peptides emanating from a bovine serum albumin (BSA) digestion (functioning as a mock-up peptide library) and the natural ligand, a- factor (WHWLQLKPGQPMY), which binds to the target membrane protein Ste2 in the immobilized membrane preparation inside the FAC-LPI FlowCell are shown. The SIR traces were recorded by using the setup described in FIG. 2, #2, where the ligand library is flowed through the FAC-LPI FlowCell, where the immobilized target membrane protein resides, and the eluate is mixed with a mixing or make-up buffer facilitating electrospray ionization in 1: 1 mixture before detection is done by MS. The natural ligand has a slight retention compared to the BSA peptides and reaches its breakthrough volume, where the SIR signal is equal to its infusion concentration, much later than the BSA peptides reaches their break-through volume.

FIGURE 7 Demonstration of retention to endogenous (naturally expressed) Ste2 in yeast. The recorded SIR traces are shown for 3 BSA peptides and the natural ligand, same setup as in FIG. 6. In this case the membrane preparation consisted of a preparation of yeast with natural, endogenous, expressed level of target membrane protein Ste2.

FIGURE 8 Recorded SIR traces of 3 BSA peptides and the ligand, a- factor. The membrane preparation was done on a yeast culture where the Ste2 gene had been deleted, denoted as ASte2. This example therefore represents a true control sample with the same sample as above except for the target membrane protein. The difference between the BSA peptides and the a- factor is very slight as noted by the time scale. FIGURE 9 Continued SIR traces from FIG 8. Image shows the release of ligands retained in the FlowCell. Ligands are eluted from the FAC-LPI FlowCell with buffer. The release rate of ligands can be determined and subsequently also the time needed to completely deplete ligands from the FlowCell.

FIGURE 10 Demonstration of off-line measurements using a fraction collector. Fractions of ligands are collected and transferred to an autosampler for analysis with MS. 10% methanol is added to each sample and a sample plug is injected into the MS via an injection loop. In SIR mode traces of detected target ligands can be followed, and the peak corresponding to each target peptide is integrated. Ion count readout is plotted against peak area of each individual ligand which corresponds to elution volume.

FIGURE 11 Recorded SIR traces from 2 masses from reference peptides (blue and red traces) and the ligand, a- factor (black trace). Fewer and more pure peptides were used as reference peptides compared to the case where tryptic digest of BSA was used as reference peptides. This, together with a good preparation which resulted in a high number of binders immobilized in the flow cell, gave rise to a distinct binding of the oc- factor ligand revealed as large breakthrough volume. FIGURE 12 A) 2-dimensional schematic of the flow cell. The direction of the flow is marked with a black bent arrow. The red box marks the inlet region shown in the 3- dimensional simulation in B. B) Flow profiles at the inlet to the flow cell. The speeds of the flow profiles can be read out from the color bar to the right. C) Left part. The flow in the 50 μηι high and 29 mm wide channels of the flow cell can be approximated with a 2-D parabolic flow. The left part of figure B shows the 2-D parabolic flow in the channel for a 100 μηι long arbitrary segment. The flow speeds can be read out from the color bar to the left. The left scale associated with the color bar gives the speeds for an arbitrary volume flow in the flow cell, where V is the volume flow expressed in ml min. The right scale gives the flow speeds for the upper limit volume flow 10 ml/min. Right part. Close up of 200 nm high and 300 nm wide region close to the bottom of the channel. The circle demonstrates the extensions of a 100 nm-diameter vesicle if placed in the flow. Flow speeds can be read out from the color bar to the right. The left scale associated with the color bar gives the speeds for an arbitrary volume flow in the flow cell, where V is the volume flow expressed in ml min. The right scale gives the flow speeds for the upper limit volume flow 10 ml/min.

FIGURE 13 Ranking of binding affinity of three different synthetic variants of the OC- factor (WHWLQLKPGQPMY). The different variants have one or two amino acids substituted in the amino acid sequence. Black trace: WHWLQLKAGQPMY, Brown trace: AHWLQLKPGQPMY and finally Blue trace: WHWLQLKPGQPAA. As noted it is easy to separate the strong binder from weaker ones. The first analogue WHWLQLKAGQPMY is known to be a strong binder to Ste2 and is compared with the two weaker ones, WHWLQLKPGQPAA and AHWLQLKPGQPMY.

FIGURE 14 A-B Example of screening of two known binders against membrane protein target MC4 receptor. Non-binders in the experiment was the a- factor

(WHWLQLKPGQPMY) and one of its synthetic variants (WHWLQLKAGQPMY), illustrated by black and green traces. The two binders were represented by synthetic melanocortin analog SHU9119 (red trace) and adrenocorticotropic Hormone (ACTH) 1- 16, Corticotropin (blue trace). As noted by indicating breakthrough volume (red line and arrow) ACTH 1-16 binds stronger than SHU9119 (which has a literature value for the dissociation constant, ¾ of 0,6 nM). The breakthrough volume ratio for the two different ligands give a four times stronger binding for the ACTH 1-16 ligand.

FIGURE 14A

Result from screening with flow rate 50 μΐ/minute. The signals for each of the binders and non-binders were normalized in order to visualize the results more clearly. The parameter AV(V-Vo) is indicated for each of the binders.

FIGURE 14B

After the ligands/binders had been washed away the same setup and experiment was done on the same immobilized preparation. The flow rate was in this case decreased to 25 μΐ/minute. As noted the parameter AV (V-V₀) is affected by longer retention times on the flow cell.

FIGURE 15A

Example showing retention of the peptide ligand a- factor to a membrane preparation containing Ste2 receptor. In this case the membrane preparation was purified by the lectin purification protocol described above. Also, the membrane preparation was immobilized on a HexaLane format, consisting of a straight line measuring 4 mm across and same height as in the Maxi flow cell, 50 μηι spacing. The flow rate was also lowered to 10 μΐ/minute. Although the amount of membrane material is 10 times lower than in a Maxi flow cell, still a clear retention difference between a membrane preparation with Ste2 and a membrane preparation from a knocked yeast strain ASte2 (containing no Ste2).

Blue trace is from 10 μΜ Hepes and both black and red traces are SIR m/z for the a- factor, screened at 1,5 μΜ.

FIGURE 15B

Retention of a-factor ligand to a membrane preparation of ASte2 with no Ste2 receptor. The ligand demonstrates some retention of non-specific manner. Blue trace is from 10 μΜ Hepes and both black and red traces are SIR m/z for the a-factor screened at 1 ,5 μΜ.

FIGURE 16

Example of screening of a known small molecule binder (ZM 241385) against the membrane protein target Adenosine A2a receptor. Non-binder in the experiment was Hepes, present at 10 μΜ, illustrated by light blue and orange traces. Two different runs (one control run and one sample run) were placed on top of each other to illustrate the difference between control and sample in a single graph. The blue trace, - A2a, illustrates the retention of the ZM 241385 to a control preparation consisting of HEK 293 membranes. The red trace, + A2a, illustrates the retention of the ZM 241385 to a preparation with overexpressed Adenosine A2a receptor. Both runs were performed at the same conditions and identical protocols were followed both for the membrane preparation and the FAC screen setup. The flow rate was 25 μΐ/min and the ligand screen

concentration for ZM 241385 was 400 nM. SIR recording was made for the m/z peak for the ligand and Hepes. The retention difference between control and sample was in this case ~4 minutes, corresponding to a AY of 100 μΐ.

FIGURE 17

The figure schematically shows the effect of increasing concentration of allosteric compound together with a fixed concentration of orthosteric ligand, in this specific case a positive allosteric effect is shown. ACo is the retention without allosteric modulator present and AC_1-5 illustrates a shift in the retention time when the allosteric modulator concentration is increased. A negative allosteric effect will decrease affinity of the orthosteric compound causing a decreasing retention window with increasing allosteric modulator.

FIGURE 18

The figure is a schematic of a titration experiment, which can be used to determine the dissociation constant Kd of the ligand - target interaction and the number of active sites in the flow cell, Bt. In this specific case the concentration range was chosen to titrate the interaction between the alpha factor and Ste2 target membrane protein. At low concentration of ligand the retention window between the ligand and the flow marker molecule is large (in most examples HEPES was used as flow marker molecule). When increasing ligand concentration is used, the retention window is decreased as shown in the figure.

FIGURE 19

Example of a titration of a high affinity ligand against a membrane bound target. Natural ligand, a-factor was titrated against its target, pheromone G-Protein-Coupled Receptor Ste2. Decreased retardation of the front by increased concentration of infused ligand is represented by plotting the linear relationship according to Eq.2, 1/[A]0(V-V0) versus 1/[A]0 where V0 is the retention volume of non-binder marker molecule and V is the retention volume of ligand at infused concentration [A]0. Kd and titration capacity Bt can be retrieved from abscissa and ordinate respectively where Kd is the dissociation constant for the ligand and Bt is the dynamic binding capacity of ligand.

DETAILED DESCRIPTION OF THE INVENTION

This invention describes a flow cell device which enables ways to screen libraries of ligands/binders across immobilized native membrane protein targets preferentially using the method of frontal affinity chromatography. The membrane preparations consist of small proteoliposomes prepared from cells or tissues through extrusion, sonication or reconstitution as non- limiting examples. Immobilization of the proteoliposomes is preferably performed as described in WO 2006068619 in order to create a stationary phase of native membrane protein targets in a flow cell format. This approach takes membrane preparations directly from cells or tissues without detergents, which therefore retains the structure and function of the membrane proteins and also keeps the natural lipid composition around the membrane proteins. The present invention preferably utilizes the methods of frontal affinity chromatography coupled to mass spectrometry to identify and rank members of a library of ligands/binders that can be flowed across the

immobilized native membrane protein targets through the flow cell.

The membrane proteins are immobilized in the form of small proteoliposomes on two opposing surfaces, separated by a thin spacer. The thin spacer also sets the boundaries for the flow cell channel, where the width of the channel is much greater than the spacer height, creating a high aspect ratio. The design of the microfiuidic flow cell enables a well defined laminar flow of thin liquid films of fluid across the membrane proteins. In frontal chromatography it is highly important to have control over the volumetric flow, since calculations are done on the estimates of breakthrough volumes etc.

The preparation of the stationary phase of membrane protein targets is very simple and does not require any major technical knowhow or special equipment. The flow cell is simply filled with an excess of membrane preparation in the form of small

proteoliposomes, which rapidly (minutes) adsorbs to the exposed flow cell surfaces, which are of membraneophilic character. Sample preparation

There are a plethora of sample preparation methods for different types of cellular starting material and application protocols. For the isolation of membrane proteins, a separation step to remove the soluble components is desirable. In general, protocols involve breaking up cells so that the soluble parts can be separated from the in-soluble parts - the membranes. This step is usually referred to as lysis of cells. Lysis of cells may be performed by osmotic shock, sonication, the use of specific enzymes, adding detergents or using methods/protocols that disrupt the integrity of the exterior cell membrane, in some cases referred to as the plasma membrane. After this step it is possible to separate the membrane fraction from the soluble parts. This separation is preferably done by centrifugation methods however two-phase extraction protocols and other elaborate techniques involving alcohols, detergents or other chemical mixtures kits have been developed for the purpose of enriching membrane fractions and removing soluble contaminants. Preferably though, the membrane preparation is prepared without the use of detergents to retain the functionality of the membrane proteins and the lipid composition around the membrane proteins, especially when activity or binding studies is to be performed. In a related aspect, the membranes prepared by the above methods may also be washed in order to further remove cytosolic components. Washing may be performed by adding high pH buffers, high ionic strength buffers, pure water or similar as non- limiting examples to the membranes. High pH and high ionic strength can in some cases affect ionic interactions and thus remove weakly membrane associated species from the membrane. Again centrifugation methods may be used in order to separate the membrane parts from the soluble parts. Membranes are pelleted by centrifugation forces and the supernatant contains the soluble parts, such as cytosolic contaminants. Preferably, the centrifugation is performed by exerting enough centrifugation force to pellet all membranes without pelleting large soluble protein clusters and similar. In a related preferred aspect, the centrifugation velocity may range from 10 OOOxg to 300 OOOxg depending on application protocol and membrane preparation. In a related exemplary embodiment the centrifugation velocity was performed at 48000xg for 30 minutes at 4 degrees Celsius. In a related aspect, specific parts of the cells membrane may also be collected and studied separately. In this aspect, the different cell membranes may be separated by their difference in membrane composition and hence density. A density variation between different membranes could also be created by binding of an appropriate modifier to a particular membrane. Above methods are often referred to density fractionation methods and may be done by a number of different approaches, the most common being the separation of membrane species in a density gradient. Density gradients may also be done by centrifugation methods, for example by continuous or discontinuous sucrose gradients as a non-limiting example. Other media that also create density gradients have been developed and commercialized, such as Percoll or Ficoll etcetera, often consisting of silica particles of certain sizes.

The membrane preparation is preferably transformed into small proteoliposomes using various methods, preferably ultrasonication, tip/probesonication and extrusion procedures as non-limiting examples. Sonication methods can be performed in a bath sonicator or with a probe sonicator. In both cases energy in the form of ultrasonic waves are produced that creates shearing motions in the buffer. Membrane particles and membrane sheets are then transformed into smaller membrane fragments. Due to the inherent nature of lipid membranes having a hydrophobic core, the system tries to minimize energy by closing open ends that are created by the shearing motion. Thus small proteoliposomes are created. The size of the created proteoliposomes is dependent on the energy input in the form of amplitude settings of the instrument, the pulse length and sonication time. In a related aspect, the resulting proteoliposomes are 10-100 000 nm in diameter, more preferably 20-500 nm in diameter. In another preferred embodiment, the proteoliposomes are prepared using the methods described in "PLASMA MEMBRANE VESICLES AND METHODS OF MAKING AND USING SAME, Serial No.

61/046,479" and Bauer et al. 2009 (Angew Chem Int Ed Engl. 2009;48(9): 1656-9), where vesicles or blebs are budded of directly from the plasma membrane of cells. This approach is based on the addition of so called vesiculation agents, which promotes the formation of blebs growing from the plasma membrane of cells.

In another aspect, the proteoliposomes may also be prepared by reconstitution procedures where a purified membrane protein target is needed. In such a case, membrane proteins are extracted by detergent(s) and re-inserted into synthetic lipid bilayers. In another aspect, proteoliposomes may consist of natural vesicles from cells of vesicles emanating from exocytosis or similar events.

The reason for the transformation of membrane preparations into small proteoliposomes comes from several different aspects, the first is the fact that the thin spacing only allows for a certain size range of particles, proteoliposomes and cells and similar material to be analyzed. In a preferred embodiment, the height of the channel should be between 0,01-500 μηι and the width of the channel 0,0001-1000 mm, more preferably a height 20- 100 μηι and width between 1 - 100 mm, more preferably a width between 4-100 mm. In a related preferred exemplary embodiment, a height of 30-60 μηι and a width of 25-30 mm for the channel dimensions. In the exemplary embodiment the aspect ratio is thus > 500. The flow can be described by thin liquid films, flowing with predictable laminar flow profiles through the flow cell, enabling an efficient and reproducible exchange of fluid.

The second reason being the fact that the transformation creates a homogenous distribution of targets in the membrane preparation, however certain types of

transformation may affect the orientation of target molecules. Orientation of targets in proteoliposomes is naturally an important issue, since it will affect the number of exposed targets in the flow cell. When performing tipsonication for example, in order to produce proteoliposomes, the orientation of targets is usually randomized. Mass spectrometry results show that peptides from both sides of the membrane proteins are identified, when digesting the solvent exposed parts of the membrane proteins situated in immobilized tipsonicated proteoliposomes. The exact distribution of orientation, 40:60 or 50:50 and so on may be difficult to determine, based on the fact that separate sides of a membrane protein can differ in the number of possible or theoretical cleavage sites. Orientation of targets may however be controlled to some extent as demonstrated by many different methods or samples. For example, it has been shown already back in the 1970's that inside-out or right-side-out vesicles of from red blood cells could be prepared and separated (Weiner and Lee 1972). Another example was demonstrated by Palmgren et al. where plasma membrane from plants was used to produce inside-out and right-side-out vesicles by partitioning in aqueous polymer two-phase systems (Palmgren et al. 1990). Plasma membrane vesicles from plant cells usually have 80-90% right-side-out orientation and it has been demonstrated that this can be transformed and converted into inside-out vesicles using mild detergent addition (Johansson et al. 1995). Microsomal fractions prepared directly from mild lysis of cells can also be kept intact and immobilized in the flow cell, thus creating an intact intracellular organelle stationary phase with correct orientation of target membrane proteins.

Another important example is virus- like particles (VLP) as those produced by MembranePro™ FPE System, from Invitrogen, or those created by techniques developed by Integral Molecular, may also be used as proteo liposomes. In brief, the particles from the MembranePro system are created by cells that have been induced to express a viral core protein called gag. The gag protein cores are budded from the cells, more specifically from the lipid raft areas of the plasma membrane, and thus capture and display the contents in the lipid rafts. These particles therefore contain a high density of GPCRs, since these types of membrane proteins are often situated in lipid raft areas of the plasma membrane. The particles produced most importantly contain correctly oriented membrane protein targets. Also, the particle preparation does not contain the same amount of contaminating material. In this aspect, contaminating material can be membranes arising from other parts of the plasma membrane (Golgi, ER, mitochondria etc.) but also soluble proteins from the cytosol.

And thirdly, very importantly, small particles immobilized on surfaces are much less affected by the flow inside the channel. In other words the drag force on small particles (submicron) is much less than on large particles (tens of micrometers), which also is an argument to keep your samples immobilized in small proteoliposomes.

Thus, in a preferred embodiment, the vesicles that are adhered to the surface of the flow cell are very small compared to the channel height. Preferably, the diameters of the proteoliposomes are in the range of 50-100 nm which is three orders of magnitude smaller than the height of the flow cell being 50 μιη. In figure 12, a close-up of the parabolic flow close to the wall is shown. It can be estimated that the fluid velocity 100 nm away from the wall is -0.8 % of the maximum channel flow. The shear stress in a Newtonian fluid is directly proportional to the velocity gradient. As a result, the shear force per unit area at a specific distance y from the channel centerline in a parabolic flow, is given by τ=μ(άα/άγ), where τ is the force and μ is the viscosity of the fluid. For a volume flow rate of 10 ml/min, the shear force per unit area at the top and bottom surfaces of the channel where it is 29 mm wide, is 7 kN. If taking the surface of a vesicle to be 7.5 10^A-15 m2

(corresponding to a 50 nm diameter proteoliposome) and multiplying this with the obtained force per unit area, a resulting shear force of ~50 pN is acting on the surface layer. Such forces are sufficient to cause deformations such as tubulations in large unilamellar solitary vesicles, but it is uncertain if they can cause deformations in tightly packed small vesicles as used here. However, when solutions are introduced manually into the flowcell by a pipet, a volumetric flow rate of about 2 mL/min is used. In this regime, the vesicles are exposed to forces which are well below what is expected to cause shape transformations. Importantly, in a preferred embodiment, in frontal affinity applications the applied volumetric flow rate is between 10 to 200 μΐ per minute. Even more preferably, the flow rate is 10-50 μΐ per minute. At these low flow rates, the exerted force on the small immobilized proteoliposomes is very small. Furthermore, as judged from the AFM experiments, the surface coverage is high, which means that the fluid is exposed to a rugged surface rather than a surface with solitary vesicles placed at a distance from each other.

In a related aspect, intact bacteria have also been tested and immobilized. As noted above, it is important to keep the flow rate low when exchanging fluid through the flow cell due to the drag forces created on the particles from the flow, since large particles will experience a larger drag force, especially at high flow rates. Investigation of the outer membrane proteins or the surfaceome of bacteria is critical from a vaccine development perspective, where surface exposed membrane proteins are key targets.

In a preferred working embodiment, the proteoliposomes contain targets of interest, including transmembrane proteins, e.g., transmembrane alpha-helix proteins, transmembrane beta-barrel proteins, lipid anchored membrane proteins, and peripheral membrane proteins. In related embodiments, the transmembrane proteins are selected from the group consisting of enzymes, transporters, receptors, ion channels, cell adhesion proteins, G-protein coupled receptors (GPCRs). In another related embodiment, the targets can be selected from the group of lipid anchored proteins.

In a related embodiment, membrane protein targets may also be over-expressed in host systems, thus affecting and increasing the amount of bound target on the surfaces. This will in turn affect the break through time of the ligands that are passed by the targets.

Preferably, for the immobilization of samples, the proteoliposome concentration should be 0.01-10 mg/ml, more preferably between 0.1-1.0 mg/ml, measured as the dry weight of membrane (total lipid/protein concentration). Preferably, the ionic strength of the proteoliposomes suspension should be between 1 μΜ to 5 M, more preferably the ionic strength should be between 20-1000 mM, even more preferably the ionic strength should be between 200-500 mM. Under such preferred conditions as described above, immobilization is performed rapidly, typically within 20 minutes. Preferably though, immobilization is performed between 30 minutes to several hours, in order to make sure that the surface is completely covered. In a related aspect, blocking substrates (including but not limited to small molecules, peptides, proteins, lipid, lipid vesicles, no n- limiting examples include DMSO, BSA, casein, dextrane) may be added in a second

immobilization step to ensure that the surface is completely covered and minimizes unspecific binding events and false positives when screening for binders. Specific blocking of gold substrates has been demonstrated by PEG-thiols and PEG-functionalized alkane thiols. This is done in order to reduce the non-specific binding of ligands to exposed surfaces etc.

In an exemplified embodiment, the membrane protein sample in the form of small proteoliposomes are immobilized onto gold surfaces for 1 hour and then washed with buffer followed by blocking of any exposed surfaces by adding peptides modified with a thiol group.

In a related embodiment, in order to reduce the non-specific binding even further to the exposed surface sites that may bind ligands non-specifically, a weak ligand may be added to saturate both the non-specific and specific sites. Stronger ligands will compete for the specific sites and release the weak binders. Such experiments can also be used in the sense of competition assays where a strong ligand competes and displaces a weaker ligand that binds to the same receptor/target site.

In a preferred related embodiment the proteoliposomes are distributed over the entire exposed surfaces inside the flow cell, thus creating a homogenous distribution of target membrane proteins in the flow cell. Compared to a mesh, gel or porous material used in column formats, targets are immobilized over the entire exposed surface inside the flow cell channel, thus making a homogenous surface distribution of receptor targets, increasing the accessibility for the ligands interacting with the receptors. The channel contains an open volume with no obstacles (as in a mesh, gel or porous structure) and the exposed targets will therefore experience a homogenous ligand concentration. The thin channel also enables very rapid mass transport between the mobile phase of ligands and the immobilized stationary phase of membrane protein targets.

In a preferred embodiment a wash or rinse step should follow the immobilization of proteoliposomes to the surfaces inside the flow cell. Preferably at least 2 flow cell volumes of buffer should be used to rinse the flow cell, more preferably 10-25 volumes of buffer should be used to rinse the flow cell. Preferably the wash buffer should be compatible with downstream applications, such as on-line MS analysis, and also be compatible with the surface immobilization and not detach the immobilized

proteoliposomes. In a related exemplary embodiment the buffer may consist of PBS (phosphate buffer saline), TRIS (2-Amino-2-(hydroxymethyl)-l ,3-propanediol), HEPES (4-(2-Hydroxyethyl)piperazine-l-ethanesulfonic acid), AMBIC (ammonium bicarbonate), AMAC (ammonium acetate), triethylammonium hydrogen carbonate or similar together with various salts including monovalent and divalent ion additives, such as NaCl, KC1, CaCi₂, MgSC>4 or similar, as non-limiting examples.

Preferably, flow rates of less than 100 nl up to 10 ml per minute may be used to exchange the fluid in the flow cell, more preferably 10 μΐ to 500 μΐ per minute is used to exchange the fluid inside the flow cell. Even more preferably, in a related embodiment, flow rates of 10-200 μΐ per minute are usable with on-line detection with MS.

To characterize the properties of the fluid flow in the flow cell, finite element method simulations were performed based on the Navier-Stokes equation. In figure 12 the flow profile at the inlet region of the flow cell is shown for an upper-limit volume flow rate of 10 ml min. For such high volume flows, a linear velocity of several meters per second is obtained at certain locations within the channels. However, the flow at the inlet region as well as in the rest of the channel is laminar, as shown by the simulation. The Reynolds number can be defined as Re=pud^, where p is the density of the fluid, u the characteristic velocity, d the characteristic length scale, and μ the viscosity. Not surprisingly, the highest local Reynolds number is found at the inlet region. By taking the maximum velocity shown in figure 12A, i.e. 4 m/s as the characteristic velocity, and the channel height (50 μπι) as the characteristic length scale, the Reynold's number becomes -200. Concerning the Maxi flow cell design, away from the inlet region and further out in the channel, the highest Reynolds number is found in the bent region where the channel width decreases to 3 mm, and the average cross-sectional velocity increases almost by a factor 10 compared to at the widest point of the channel. In the constricted and bent region, the Reynold's number becomes 50, which is well below the limit for turbulent flow (Re < 2300). (REF: D.J Tritton, Physical Fluid Dynamics, second edition, Oxford Science Publications, 1988)

The width of the channel is much larger than the height of the channel, and therefore we can approximate the flow in the channel as a 2-dimensional parabolic flow. For a parabolic flow the velocity u is:

where h is the height of the channel, u_av is the average velocity, and y is the distance to the centreline of the channel. In figure 12B, the velocity profile for such a 2- dimensional flow is shown. From the left scale bar, the velocities obtained in the flow cell for a volume flow of 10 ml/min can be read out, and from the right scale bar, the velocity for an arbitrary volume flow in the flow cell can be read out.

Preferably, a flow of ligands/binders (as components of a compound library) may then be flowed across the immobilized native membrane proteins. In a preferred embodiment, the flow creates a homogenous ligand concentration exposure for all exposed targets, as compared to in a mesh or packed column or a gel, for example. In a preferred embodiment the flow is created using syringe pumps or liquid chromatography pumps, as non- limiting examples. Preferably, as noted above, flow rates of 10-200 μΐ per minute are usable when using on-line MS as detection method. In a preferred embodiment the flow of ligands is directly detected using an on-line coupling of the eluate from the flow cell into the MS instrument.

In a related preferred embodiment the eluate is ionized by the method of electrospray. In such a preferred embodiment the eluent from the flow cell may be mixed with a second buffer containing some percentage of organic solvent in order to induce a good spray and ionization of ions in the eluent. In another related embodiment the MS analysis can be performed with other ionization techniques and MS instrumentations, such as MALDI etc. an The fractions can also be processed and analyzed by for example Rapid Fire systems or similar and also desalted and/or focused prior analysis by various MS systems.

Preferably, the wash buffer may consist of PBS (phosphate buffer saline), TRIS

(2-Amino-2-(hydroxymethyl)- 1 ,3-propanediol), HEPES (4-(2-Hydroxyethyl)piperazine- 1 -ethanesulfonic acid), AMBIC (ammonium bicarbonate), AMAC (ammonium acetate), triethylammonium hydrogen carbonate or similar together with various salts including monovalent and divalent ion additives, such as NaCl, KC1, CaCl₂, MgSC>4 or similar, as non- limiting examples. In a exemplary embodiment, wash buffer and ligand buffer that is flowed across the membrane protein targets is 5-20 mM ammoniumactetate, pH 7,5, and the mixing buffer consist of 5 mM ammonium acetate with 20% methanol, pH 7,5.

In another related embodiment, the eluent may be collected by a fraction collector. In such a preferred setup, ligands in buffer, preferably as above, is flowed across the flow cell and volume fractions ranging from 1-1000 μΐ is collected, more preferably 10-100 μΐ fractions is collected using a fraction collector of any kind. The components of the eluate fractions are then preferably transferred to an autosampler of any kind and injected into the MS instrument for analysis. The fractions can also be processed and analyzed by for example Rapid Fire systems or similar and also desalted and/or focused prior analysis by various MS systems.

In another related preferred embodiment, the MS instrument can be instructed to only detect certain ions, so called single ion recording (SIR) mode.

In another related embodiment the detection of ligands/binders may be performed by several different detection techniques, such as fluorescence, electrochemistry as non- limit ing examples .

In a preferred embodiment, the methods above enable the determination of the relative affinities for ligands/binders of a ligand/binder library towards the immobilized native membrane protein targets in the flow cell. In another related preferred embodiment the methods described above enables the determination of dissociation constants for ligands/binders towards an immobilized native membrane protein target in the flow cell.

In a preferred embodiment, the ligands referred to in this text are substances that are able to bind to and create complexes with membrane protein targets, using a variety of intermolecular forces such as ionic and hydrogen bonds as well as van der Waals forces. Strong binding reflects a high affinity for the ligand to bind to its target.

In a non-limiting exemplified embodiment, the ligands can be classified as belonging to peptides and small molecules.

In a related embodiment the library size or the number of ligands/binders screened or flowed across the immobilized membrane proteins targets is preferably between 1 and 1 * 10¹⁰ different ligands/binders. More preferably the library size is between 1 and 1 * 10⁶ different ligands/binders. In a related aspect, the screen size or the number of different ligands/binders tracked simultaneously in a single run across one flow cell is preferably between 1 and 1 * 10⁴.

The above methods may therefore be used as a primary screen setup, where the screen is performed using a large number of ligands/binders. In another setup, the number of ligands/binders is limited in number, preferably one at a time, and the above methods are then used in a secondary screen/assay setup where the binding characteristics between the ligand and target is investigated more thoroughly. Thus, determination of the dissociation constant Kd can be done by performing a titration experiment, where several different ligand concentrations are flowed across the same sample in sequential order with wash steps in between to remove bound ligand. The different retentions are then related to the ligand concentrations according to the relation between dissociation constant, number of binding sites, ligand concentration and the retention.

The relati nship between V-Vo and ¾ is given by equation 1.

By rearranging Eq. 1 to give a linear relationship between V and [A]o Eq. 2 (ligand infusion concentration) the slope and abscissa intercept will determine the constants B, and Kd. Titration is performed by varying [A]o within reasonable values for the binding isotherm. Since V varies with the concentration, an active site titration will give B_t through the ordinate and l/¾ as the intercept on the abscissa. The staircase model is an alternative approaches to obtain Bt and ¾ and do not require washing of flow cell between each assayed concentration of [ A]o. When using the staircase approach, increasing concentration of [A]o is continuously infused to the flow cell. The staircase approach is more rapid, however continuous infusion of ligands might introduce large errors throu h accumulation.

Using a similar approach, the screening for allosteric modulation may also be performed to study and identify allosteric compounds. An allosteric compound will bind to the same target protein, however at a different binding site than the orthosteric ligand. The binding of the allosteric compound will affect the binding of the orthosteric ligand in such a way that the binding is weakened or strengthened. In the absence of allosteric compound the ligand will have a given retention time on the flow cell and in the presence of allosteric modulators the retention time will be shorter illustrating a negative allosteric modulation and longer if it is a positive allosteric modulation. The MF-FAC on the LPI platform is very useful in this particular application since a shift in the retention time is easily detected.

Exemplary devices used in the methods of the invention are described in

International patent application WO/2006/068619, entitled "Device and Use Thereof, filed December 23, 2005. The entire content of this application is expressly incorporated herein by reference.

EXAMPLES

It should be appreciated that the invention should not be construed to be limited to the examples that are now described; rather, the invention should be construed to include any and all applications provided herein and all equivalent variations within the skill of the ordinary artisan.

TARGET: STE2 IN YEAST

Construction and induced growth of yeast strains to obtain cells with various levels of Ste2 expression

1. Construction of a strain with met25-pro motor driven GFP tagged Ste2

1.1. Preparation of genomic DNA from Saccharomyces cerevisae strain By4741

S. cerevisiae strain BY4741 (MATa; his3Al ; leu2A0; metl5A0; ura3A0) was used as template strain for the PCR based cloning of the STE2 gene. The BY4741 cells were grown overnight in rich complete medium (YPD), diluted with fresh YPD to an optical density (OD) of 0.5, and regrown to an OD of 0.7 (cells will then be in the mid- exponential growth phase). The cells harvested by centrifugation at 3000g for 5 minute and resuspended in 100 μΐ lyticase solution Yl (Quiagen RNA easy minikit). The cell- suspension was then incubated for 30 min at 30°C with gentle shaking to create protoplasts. 250 ul BL buffer (E.Z.N.A Blood DNA Kit) were then added and the protoplasts were lysed by vortexing. Genomic DNA was then prepared using the E.Z.N.A Blood DNA Kit (E.Z.N).

1.2. Cloning of Ste2

Primers GTACGGATCCAAGAATCAAAAATGTCTGATGC and

GCGTGAATTCTAAATTATTATTATCTTCAGTCC were used to amplify a DNA fragment corresponding to the full-length Ste2 receptor. PCR was run on 100 ng genomic DNA prepared from BY4741 using the PCR cycles: 98 °C 2 min , 35 cycles (98 °C for 30s, 51 °C for 45s , 72 °C for 2 min) and finally 72°C for 4 min. Pfu Ultra (Stratagene) was used as polymerase and 5% DMSO was included in the reaction. The resulting fragment was cloned into the BamHI, EcoRI site of pUG35, creating pUG35_Ste2 construct that will express a Ste2 C-terminally tagged with GFP. Expression of Ste2-GFP will be under the control of the MET25 promoter. Thus, the STE2-GFP construct is induced by depletion of methionine from the medium (see induction procedures below). 1.3. Transformation of Saccharomyces strain BY4742 with pUG35_Ste2

Transformation of BY4742 (MAT ; his3Al; leu2A0; lys2A0; ura3A0) with pUG35_Ste2 was made according to the Litium acetate method (Gietz et aL1991) creating the strain BY4742_Ste2-GFP.

2. Growth of yeast cells for vesicle preparation

2.1. Growth and harvest of the negative control (Ste2A) and the wild type with native expression of Ste2.

The wild type BY4742 (this strain is of the MATa sex that expresses STE2 in the haploid state) and the mutant carrying the gene deletion of STE2 (BY4742ste2A) was grown over night in 500 ml in synthetic medium containing all amino acids and nucleotides (YNB- complete), diluted with fresh Y B-complete medium to an OD of 0.6-1 and regrown to an OD of 1-1.8.

2.2. Growth, induction and harvest of the MET25 controlled STE2-GFP construct BY4742_Ste2 was grown to saturation in 500 ml YNB-complete -Ura, washed two times with YNB-complete-Ura-Met, diluted to an OD of 0.6-1 in YNB-complete-Ura-Met and grown for another 2.5 to 4 hours to an OD of 1-1.8. All strains were harvested by centrifugation at 3000g for 5 minutes frozen at -20C.

Isolation of Ste2 membranes

Cell pellet was suspended in 10 mM HEPES pH 7.0, 4 mM EDTA and 1 mM

phenylmethylsulphonyl fluoride (PMSF). Cells were broken by vigorously vortexing with 425-600 μηι glass beads, for homogenization of cells, equal volume of cell pellet, buffer and glass beads were used. A total of 8 rounds of vortexing were applied with 30 sec duration, solution was kept on ice-water for 30 sec in between vortexing. Unbroken cells, cell debris and glass beads was removed by centrifugation at 500g for 5 min at 4°C.

Supernatant was carefully collected and transferred to a new centrifuge tube and centrifuged at 5000g for 10 min at 4°C. Membrane fraction was collected by

centrifugation of the resulting supernatant at 48 OOOg for 30 min at 4°C. Membrane pellets were suspended in 50 mM HEPES pH 6.8, 0.15 M NaCl, 2 mM CaCl₂, 5 mM KC1, 5 mM MgCl₂, 4 mM EDTA and snap frozen in liquid nitrogen. Isolated membranes were stored at 80 °C until use. Formation of proteoliposomes

Membranes were suspended in 10 mM Tris-HCl, 300 mM NaCl pH 8. Proteoliposomes were formed by sonicating the suspension with a VibraCell™ sonicator equipped with a 2 mm ultra tip. Sonicator was set to operate with 0.5 s pulses, 0.5 s rest for 1 min.

Sonication was performed in a thick wall glass vial and at 4 °C. Proteoliposome formation was confirmed by fluorescence staining with FM1-43. Proteoliposome solution was centrifuged at 2000g for 5 min at RT (room temperature) and supernatant containing proteoliposomes were immediately immobilized into the LPI™ FlowCell at RT.

F AC setup for Ste2

The ligand syringe contained the natural ligand, a-factor mixed with peptides from trypsinated bovine serum albumin, BSA prepared in 5 mM ammonium acetate pH7.4. Concentrations of BSA peptides were determined by comparing the ion count signal with a dilution series of known standard. Natural ligand was infused at 2.9 μΜ and 7 different BSA peptides at 3-5 μΜ. Alternative to the setup described above one can assay the outlet from the flow cell in an off-line manner. The offline approach is extremely valuable if screening conditions are not compatible with direct MS detection. Fractions collected can prior to MS analysis be tailored for high detectability by means of, changing pH, desalting and exchange of buffer components. This may also be performed using a Rapid Fire setup or similar.

The setup for off-line measurement was as principle as described above but omitting the syringe containing makeup buffer. Ligands to be assayed were infused to the flow cell at 50 μΐ min^_1and the eluate was collected in discrete fractions of 50 μΐ. Collected fractions representing the profile of breakthrough volume were transferred to glass vials and placed in an auto sampler. Fractions were supplemented with methanol to a final concentration of 10% and assayed with the same settings as described below. TARGET: MC4

Membrane preparation

A human MC4 receptor membrane preparation (Chemi SCREEN™, catalog number: HTS105M) was purchased from Millipore (www.millipQre.com). In brief the preparation (Chemicon's MC4 membrane preparation) consisted of crude membrane preparations made from proprietary stable recombinant cell lines to ensure high-level of GPCR surface expression of the target. Liquid in packaging buffer: 50 mM Tris pH 7.4, 10% glycerol and 1% BSA with no preservatives. Packaging method: Membrane protein adjusted to the indicated concentration in packaging buffer (1 mg/ml), rapidly frozen, and stored at -

80°C. Prior use the membrane preparation was further washed to remove soluble proteins by first diluting the membrane preparation in wash buffer (50 mM HEPES, 5 mM MgCl₂, 1 mM CaCl₂, pH 7,4. This was done by splitting one tube containing 1 ml of membrane preparation (1 mg/ml) into two tubes, 500 μΐ in each, and adding 1,5 ml wash buffer to each tube. The tubes were then spun at 40,000 x g for one hour using a Beckman centrifuge ( ) equipped with a TLS-55 rotor. A 3 mm pellet was collected and the supernatant discarded. The pellet was re-suspended in 2 ml wash buffer and again spun at 40,000 x g for one hour. Each pellet was re-suspended in 300 μΐ wash buffer and collected. In addition 100 μΐ wash buffer was used to rinse the tubes to remove remaining membrane preparation. The samples were pooled to a total of 800 μΐ and split into 4x200 μΐ aliquots. The samples were snap frozen in liquid N₂ and stored in -80 degrees. Prior use, one aliquot was diluted with 200 μΐ 10 mM Tris, 300 mM NaCl, pH 8 and sonicated mildly on ice using a VibraCell™ sonicator equipped with a 2 mm ultra tip using 7% amplitude, 0.5 second pulses with 0.5 second rest time for 1 minute, iterated 3 times.

TARGET: Adenosine A2a

Lectin purification procedure: Use a suitable protocol to prepare membrane protein vesicles, such as proteoliposomes. Centrifuge freshly prepared proteoliposomes at 2000 x g and transfer supernatant (proteoliposomes) to a Falcon tube (15 ml), discard the pellet.

Add 1/3 volume of Lectin gel* and 1/3 of buffer, 50 mM Tris pH 7.4, 10 mM MgCl₂, 0.5 mM EDTA to 1/3 proteoliposomes volume. Incubate slurry for 30 min with gentle shaking at RT and then transfer the slurry to a gravity flow gel column.

Wash the slurry with approximately 15 bed volumes with running buffer and exchange buffer in gravity flow column by adding running buffer supplemented with 10% N- acetylglucosamine (NAG). Stop the gravity flow and incubate 10 minutes.

Elute bound material with 1.5 bed volumes of running buffer containing 10% NAG and snap freeze the eluted material. The 10% NAG will also function as cryoprotectant in liquid nitrogen in aliquots of 200-500 μΐ. Store proteoliposomes at -80°C.

Regenerate lectin resin on flow column by washing with 5 bed volumes with regeneration soltion. Incubate lectin resin for 30 min with regeneration solution. If performing multiple purification step, re-equilibrated lectin resin with running buffer. If lectin resin is to be stored, add 0.1% NaN₃ to regeneration solution as a bacteriostat.

*Binding capacity (Sigmaaldrich L1394) 1-2 mg/mL (ovomucoid). If proteliposome solution is dense, accordingly increase the lectin volume.

Lectin regeneration buffer: 0.5 M NaCl containing Mg²⁺, Mn²⁺, Ca²⁺, and Zn²⁺ (1 mM each)

LPH^M FlowCell LPI™ FlowCell consists of two flat, gold-coated, plastic pieces that are adhered together with a chamber tape. The chamber tape is shaped in a way so that a flow channel is created between the plates. Access to the channel is via two threaded holes through the top surface. The Maxi FlowCell size is 109 x 68 mm and equipped with 13 screws, evenly distributed along the tape profile, to make it withstand pressure better. The HexaLane FlowCell consist of six parallel channels of the same basic material as above, with dimensions 5 x 95 mm for each channel. Plastic material (PETG) is processed to the right size and the threaded holes are made. The surfaces are then ultra sonicated in isopropanol for 5 minutes prior to coating with chromium and then gold. The total thickness of metal is approximately 40-60 nm. The technique used for coating is PVD - Magnetron sputtering. LPI™ FlowCell was assembled in the clean room facility at Chalmers University of Technology, Gothenburg.

Frontal affinity chromatography-Mass spectrometry

Prior to Frontal affinity chromatography-Mass spectrometry FAC-MS, all liquid media such as buffers etc were filtered with 0.45 μΜ membrane filter. Immobilizations of proteo liposomes were done by injecting 380 μΐ of proteoliposome solution into the LPI™ Maxi FlowCell followed by incubation for 1-2 h at room temperature or alternatively 40 μΐ proteoliposome sample was injected into the LPI HexaLane channels) with the same incubation time. Before integrating the FlowCell with the MS setup the FlowCell were washed ~30 times with 5-20 mM Ammonium acetate pH 7.5, depending on model system. A syringe pump was fitted with two individual syringes, one ligand containing syringe and one syringe containing makeup buffer, ammonium acetate pH 7.5

supplemented with 20-60 % Methanol. Syringe holding ligand was directly fitted to the LPI FlowCell inlet by a 1/8" PEEK™ connection and the outlet was connected to a mixing tee by 1/8" PEEK™ connection. Syringe containing makeup buffer was connected to mixing tee postcolumn by a 1/8 PEEK™ connection. To prevent backflow of makeup buffer the outlet port from LPI FlowCell can if needed be equipped with a floating ball check valve. Ligands were infused over the FlowCell with a flow rate of 10-50 μΕ/ηιίη. MS analysis

A tripple-quadrupole mass spectrometer (Quattro LC, Micromass, Waters) was used to detect the peptide ligands eluting from the LPI FlowCell using electrospray ionization (ESI) and the single ion recording (SIR) mode.

For the Ste2 model system the following settings was used. Capillary voltage was held at 3,8 kV, the cone voltage was 55 V and the extrator voltage was 4 V. The source block temperature was 120 degrees Celsius and the desolvation temperature was 220 degrees Celsius. The single ion recording was performed on a select number of BSA peptides (m/z 927.69, 922.54, 820.87, 789.63, 733.62, 720.64, 689,39) and the oc-factor (m/z 842.73). Also, the same settings was used when detecting and following analogues of the a-factor. For the MC4 and A2a model systems the following settings was used. Capillary voltage was held at 3,65 kV, the cone voltage was 25 V and the extrator voltage was 3 V. The source block temperature was 120 degrees Celsius and the desolvation temperature was 250 degrees Celsius.

Data collection and data analysis

During FAC retention time of infused ligand(s) is determined by the affinity of ligand to immobilized target, equation 1. MS software was set to collect data in SIR mode and traces representing ion count of each individual peptide was continuously followed until plateau level. Data was converted and analyzed with a third degree polynomial function. Point of inflection representing the breakthrough volume was determined by the second derivative. Ligand affinity was ranked according to differences in V-Vo which in turn is inverse proportional to ¾.

Claims

1. A method for characterizing the binding interaction of a candidate binder to a target molecule comprising

- providing a microfluidic flow cell having

- an inlet;

-an outlet;

said method further comprising

- detecting and/or identifying said candidate binders in the eluate

2. The method according to claim 1, wherein characterization of the binding interaction of the candidate binder to the target molecule is done by frontal affinity chromatography and the breakthrough volume for the candidate binder is determined by detection and/or identification of the candidate binder in the eluate.

3. The method according to any preceding claim, wherein the membraneophilic region is a surface made of silicon dioxide, glass, Mica or a polymer, metal oxides and metals (e.g. aluminum oxide and gold, or any combinations thereof) or any combinations thereof.

4. The method according to any preceding claim, wherein the covering layer is constituted by a surfactant membrane, such as lipid bilayer membranes, cell membranes, lipid vesicles, proteovesicles, proteoliposomes, lipid nanotubes, lipid monolayers, cell fragments, organelles, virus-like particles and any combinations thereof; a surfactant; or emulsion system.

5. The method according to claim 4, wherein the surfactant membrane is at least partially immobilized to the membraneophilic region via lipids, sugars and/or proteins.

6. The method according to anyone of the claims 1-3, wherein the covering layer is constituted by intact cells or bacteria.

7. The method according to anyone of the claims 1-6, wherein the covering layer further comprises a blocking agent that will block the membraneophilic surfaces not covered by the covering layer, thus hindering non-specific binding to the membraneophilic surface, wherein the blocking agent can be selected from, but not limited to, small molecules such as DMSO, peptides, proteins, lipid, lipid vesicles.

8. The method according to any of the preceding claims, wherein the membraneophilic region comprises gold and wherein the blocking agent is selected from, PEG-thiols, PEG- functionalized alkane thiols and thiol linked molecules including peptides, proteins, lipid, lipid vesicles, small molecules and similar entities.

9. The method according to anyone of the claims 1-8, wherein the target molecule included in or bound to, connected to or associated with the covering layer is selected from the group consisting of peripheral and integral membrane proteins, glycosylphosphatidylinositol(GPI)- anchored proteins, phospholipids, sphingo lipids, drugs, amphiphilic agents, lipophilic agents, sterols, sugars, oligonucleotides, polymers and DNA.

10. The method according to any of the preceding claims, wherein one or several

candidate binders is an orthosteric ligand.

11. The method according to any of claims 2-10, wherein the breakthrough volume is directly related to the binding affinity of the candidate ligand for the protein, the number of active binding sites or correctly oriented target membrane proteins and the ligand concentration (equation 1).

^KJ ⁺ [^L] (Equation 1)

12. The method according to any of the preceding claims, wherein a dissociation constant (Ka) between the ligand and the target molecule is determined.

13. The method according to any of the preceding claims, wherein the covering layer including the target molecule is contacted with one or several orthosteric ligand(s) and several allosteric ligand(s).

14. The method of claim 13, wherein a shift in binding strength is detected for the orthosteric ligand by observing the change in breakthrough volume for the orthosteric ligand when said orthosteric ligand is mixed with an allosteric ligand.

15. The method of claim 13-14, wherein said orthosteric and allosteric ligands are mixed in different concentrations before contacting the covering layer.

16. A device comprising:

-at least one supporting solid surface comprising at least one membraneophilic region;

-a covering layer that is at least partially immobilized to the membraneophilic region, said covering layer consisting of (i) a surfactant membrane, (ii) a lipid mimicking polymer, (iii) a surfactant or emulsion system, (iv) a liquid crystal, or a combination thereof;

- a substance included in or bound to, connected to or associated with the covering layer; wherein said device is adapted for use in the method according to any of claims 1-15.