ACTAUNIVERSITATISUPSALIENSISUPPSALA2007

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 294

Protease Activity, Inhibition andLigand Interaction Analysis

Developments and Applications for Drug Discovery

THOMAS GOSSAS

ISSN 1651-6214ISBN 978-91-554-6866-8urn:nbn:se:uu:diva-7822

To my family

List of Papers

The thesis is based on the following papers, referred to in the text by their roman numerals: I Analysis of the pH-dependencies of the association and disso-

ciation kinetics of HIV-1 protease inhibitors. Thomas Gossas and U. Helena Danielson, J. Mol. Recognit. 2003; 16: 203-212

II Exploration of acylsulfonamides as carboxylic acid replace-ments in protease inhibitors of the hepatitis C virus full-length NS3. Robert Rönn, Yogesh A. Sabnis, Thomas Gossas, Eva Åkerblom, U. Helena Danielson, Anders Hallberg and Anja Johansson, Bioorg. Med. Chem. 2006; 14: 544-559

III Evaluation of a diverse set of potential P1 Carboxylic acid bioisosteres in hepatitis C virus NS3 protease inhibitors, Robert Rönn, Thomas Gossas, Yogesh A. Sabnis, Hanna Daoud, Eva Åkerblom, U. Helena Danielson and Anja Sandström, Bio-org. Med. Chem., Accepted.

IV Characterization of Ca2+ interactions with matrix metal-lopeptidase-12: implications for matrix metallopeptidase regulation. Thomas Gossas and U. Helena Danielson, Biochem.J. 2006; 398: 393-398

V The high affinity of hydroxamate-based MMP-12 inhibitors is due to an inherently slow dissociation rate of the enzyme-inhibitor complex. Thomas Gossas, Helena Nordström, Hans Wallberg and U. Helena Danielson, submitted.

Reprints of the papers were made with permission from the publishers

Contents

Introduction...................................................................................................11Proteases...................................................................................................11Proteases as drug targets ..........................................................................12The proteases in this study .......................................................................13

HCV NS3/NS4A protease ...................................................................14MMP-12...............................................................................................15HIV-1 protease ....................................................................................17

Protein-ligand interactions .......................................................................18Forces of interaction ............................................................................18Basic enzyme and interaction kinetics.................................................19Basic thermodynamics.........................................................................20

Aim of Present Investigation ........................................................................22

Experimental Methods ..................................................................................23Recombinant enzymes..............................................................................23

HIV-1 protease ....................................................................................23HCV NS3/NS4A protease ...................................................................23MMP-12...............................................................................................23

Interaction studies ....................................................................................24Enzyme activity assays ........................................................................24Direct binding assays...........................................................................24Substrate or no substrate? ....................................................................25

Results...........................................................................................................27Protease activity regulation ......................................................................27

MMP-12 and Calcium (Paper IV) .......................................................27HCV NS3-NS4A .................................................................................29

Enzyme-inhibitor interaction studies........................................................29The pH-dependency of HIV protease-inhibitor interactions (Paper I) 29MMP-12 inhibitor interactions (Paper V)............................................32HCV NS3 inhibitor optimization (Paper II and III).............................33

Conclusions and perspectives .......................................................................37Conclusive summary ................................................................................37Relevance for drug discovery...................................................................38

Future prospects .......................................................................................39

Sammanfattning ............................................................................................40Proteaser och läkemedelsutveckling ........................................................40Min forskning...........................................................................................41

Reglering av proteasaktivitet ...............................................................41Interaktioner mellan proteaser och hämmare.......................................41

Slutsatser ..................................................................................................42

Acknowledgements.......................................................................................43

References.....................................................................................................45

Abbreviations

HCV Hepatitis C virus HIV Human immunodeficiency virus MMP Matrix metallopeptidase SPR Surface plasmon resonance RU Resonance units IMAC Immobilized metal affinity chromatography FRET Fluorescence resonance energy transfer EDANS N- (Aminoethyl)-5-naphthylamine-1-sulfonic acid DABCYL 4-(4-dimethylaminophenylazo)benzoic acid SAR Structure activity relationship Asp Aspartic acid Glu Glutamic acid His Histidine Arg Arginine Ser Serine Nva Norvaline ACCA 1-amino-cyclopropane-carboxylic acid kon Association rate constant koff Dissociation rate constant KD Equilibrium dissociation constant Ki Inhibition constant KM Michaelis constant TIMP tissue inhibitor of metalloproteases ECM Extracellular matrix

11

Introduction

In science there has always been a desire to describe natural phenomena and to know the true nature of things, and the goal of science is often described as the quest for knowledge. A more pragmatic view is that science should be able to describe nature in order to make predictions. Empirical sciences can often find a description of a phenomenon, based on the induction of knowl-edge and the subsequent deduction of rules that are the foundation for pre-diction.

The science of enzymology deals with the structure, function and interac-tions of enzymes. This knowledge forms the basis of drug discovery. Some-times the word rational is put in front of drug discovery to indicate that the design of drugs can be derived from scientific reasoning. However, to call it rational drug discovery, it is also necessary to able to deduce from enzy-mological knowledge how a drug should be made for a specific target. By the study of protease-ligand interactions, this work contributes to the knowl-edge of enzymes and their interactions, and as such it adds a small piece in the puzzle towards the realization of rational drug discovery.

ProteasesProteases, also known as proteinases or peptidases, are enzymes that catalyse the hydrolysis of peptide bonds, or in other words they cut proteins. These enzymes occur everywhere in nature, from the smallest virus and bacteria to plants and animals, and their physiological roles are equally diverse, ranging from food digestion and haemorrhage venoms to signal cascades controlling blood clotting and apoptosis. Proteases are classified according to the chemi-cal mechanism of catalysis. There are six types of proteases known: Serine-, cysteine-, threonine-, aspartic-, glutamic- and metallo-proteases. These are further classified according to type of reaction catalyzed and sequence ho-mology. For a thorough and useful account of the classification of proteases, see the MEROPS database of peptidases [1,2].

Proteases do not just cut any proteins, but they have a well-defined sub-strate specificity. The substrate specificity is determined by the substrate binding site, which is lined with pockets in which the residues of the sub-strate peptide can bind. The residues of the substrate peptide are named P1, P2 and so on from the scissile bond toward the N-terminus. From the scissile

12

bond toward the C-terminus the residues are named P1’, P2’ and so on. The corresponding binding-pockets in the protease are called subsites and are named with an S instead of the P as in S1 or S1’ (Figure 1). This terminol-ogy, given in Schechter and Berger [3] is used throughout this thesis. The same nomenclature is used for pepdidomimetic inhibitors.

Figure 1. A substrate peptide, bound in the active site of a protease. The scissile bond is between P1 and P1´.

Proteases as drug targets Because proteases are involved in many essential processes in both humans and several pathogens, these enzymes have emerged as important drug tar-gets. The human genome encodes 500 - 600 proteases and abnormal activi-ties of many of these enzymes have been identified in numerous diseases [4]. If the disease is coupled to the excessive activity of an endogenous human protease, or proteases that are essential for the replication of a pathogen, turning off the activity, by inhibiting that protease seems like a good strat-egy. Protease inhibitors are therefore a potential cure for many infectious and endogenous diseases.

Many endogenous proteases are now recognized as part of vital signaling processes and not merely as protein degrading enzymes. As such their in-volvement in disease is complex and all aspects are not readily accessible through reductionistic methods [5]. As part of biochemical networks of sig-naling pathways, proteases may have more functions than what can be de-duced from, for example, learning the identity of their substrates. Studying the regulation of protease activity is thus a crucial part of understanding these enzymes as drug targets.

Because many proteases have well defined substrate binding-sites, sub-strate-based inhibitor design, have produced potent and selective inhibitors [6]. The strategy is to start with a substrate peptide and exchange the scissile bond with a non-cleavable analogue. Inhibitors with remarkably increased affinity are achieved when that analogue mimics the transition state of pep-tide bond hydrolysis. For example, in inhibitors of HIV-protease, hy-

13

droxyethylamine is a common amide bond isostere since it resembles the tetrahedral transition state of the enzyme-catalyzed reaction. Although tran-sition state analogy is a successful strategy to design inhibitors, other fea-tures of proteases can be utilized as well. For metalloproteases, metal chelat-ing functionalities attached to P1’-P2’-peptides have been utilized to design potent inhibitors, and the discovery of product inhibition of HCV NS3 prote-ase was the starting point for the development of highly potent inhibitors of that enzyme. Since peptides are unsuitable as drugs, for reasons of uptake and stability, the peptidic nature of the inhibitors has to be reduced. This process is greatly aided by the determination of three-dimensional structures of protease-inhibitor complexes and computational methods for modeling of protease-inhibitor interactions. The development of HIV protease inhibitors as drugs against AIDS is an example of the usefulness of structure assisted drug design [7]. Another example of protease inhibitors used successfully in the clinic is the inhibitors of angiotensin converting enzyme (ACE) for the treatment of hypertension. These examples have shown that protease inhibi-tors can be used to treat illnesses caused by both infectious agents and en-dogenous factors.

The proteases in this study Although all three enzymes in this study are proteases, they are of different classes and origin. One is a human metalloprotease while the other two are aspartic and serine proteases of viral origin. The fact that they are proteases makes it simple to study their activity, by measuring the hydrolysis of pep-tides, using the same type of methodology. Furthermore, with proteases is often associated the same type of problems and scientific questions to be solved, such as the question of substrate specificity, or, to be able to study them at all, the problem of autohydrolysis. In all other aspects they are still very different from each other and that is relevant to consider when studying them. For drug discovery it is important not to forget the biological context of the target enzymes. This is simply because the assays that are applied should make suitable model systems for in vivo situations. For example, the question may be if a fully functional catalytic domain is a relevant model system although the biologically active protein is a larger structure. Further-more, the conditions of inhibition assays should reflect the physiological conditions in order to get early and accurate predictions on drug efficacy. Such things are of relevance for studying the enzymes as drug targets and therefore will follow a short description of the enzymes of interest in this study.

14

HCV NS3/NS4A protease Hepatitis C virus non-structural protein 3 (HCV NS3) is a bifunctional pro-tein. The larger C-terminal domain is a helicase/NTPase responsible for the unwinding of RNA during viral replication. The N-terminal domain is a ser-ine protease of chymotrypsin type, with an active site consisting of the cata-lytic triad His, Asp, Ser (Figure 2). This enzyme is a validated drug target for the treatment of hepatitis C due to its essential function in polyprotein processing during viral replication [8].

Figure 2. The structure of HCV NS3 protein (PDB-file 1CU1). The larger domain at the top of the picture is the helicase domain. Below is the protease domain, with the residues of the catalytic triad in stick representation. The black strand is a truncated NS4A-peptide, which in this construct is fused to the protease domain. The struc-tural zinc, represented as a sphere, is situated in the lowermost part of the protease domain.

Although both domains retain full enzymatic activity when expressed sepa-rately, there is no evidence that separation of the two occurs in vivo. Looking at the protease domain alone, the substrate binding-site is relatively feature-less and exposed to the solvent [9]. However, the helicase domain, situated above the active site of the protease, provides a lid and could thus influence substrate and consequently also inhibitor binding [10,11]. Note also that the

15

C-terminal of the helicase domain lies in the substrate binding site of the protease. The natural substrates of the protease are the cleavage sites be-tween the non-structural proteins of the HCV polyprotein. These substrates have a consensus P1 cystein, P1’ serine or alanine and P6 glutamic or aspar-tic acid [12]. The major single determinant of substrate specificity is the S1 pocket, which fits a small residue, such as cysteine [13]. Since the substrate binding site is void of distinct binding pockets other than the S1, the overall substrate specificity is determined by many weak interactions with a long peptide, spanning a minimum of P6 to P4’ [14].

For full protease activity the small NS4A protein (54 amino acids) is needed as a co-factor. A peptide comprising a central part of NS4A is suffi-cient for maximal activity in vitro [15-17]. NS4A provides one of the -strands in the N-terminal -barrel of the protease domain (Figure 2). This brings about a conformational change that affects the position of the catalytic residues to form a functioning triad [18]. The N-terminal of NS4A is a hy-drophobic helix that is responsible for anchoring NS3 to the membrane of the endoplasmatic reticulum, where viral replication takes place [19]. NS4A is suggested not only to activate proteolysis but also to modulate the sub-strate specificity [20,21]. However, the physiological relevance of NS4A activation is not known.

MMP-12 Matrix metallopeptidases (MMPs) are zinc-dependent enzymes belonging to the family of metzincins [22]. MMPs are characterized by the conserved HEXXHXXGXXH sequence, comprising the catalytic glutamate and the metal binding histidines [1,2]. MMP-12 is mostly expressed by macrophages and is therefore also called macrophage elastase [23]. MMPs are implicated in several diseases involving general destruction of the extracellular matrix (ECM). MMP-12 is specifically a target for potential drugs against chronic obstructive pulmonary disease (COPD), rheumatoid arthritis and multiple sclerosis [24-26].

The overall structure of MMPs consists of a signal peptide for secretion, a prodomain that is proteolytically removed upon secretion, a catalytic domain responsible for peptidase activity, and a hemopexin-like domain that inter-acts with the ECM (Figure 3).

16

Figure 3. Domain organization of MMPs. Following the signal peptide (SP) is the pro-domain, which keep the enzyme inactive by interaction with the catalytic zinc ion via a cysteine thiol. The zinc ion is chelated by three histidines in the catalytic domain. The hemopexin-like domain, which in MMP-12 is autocatalytically re-moved, is linked to the catalytic domain via a hinge region.

In addition to the catalytic zinc, the catalytic domain of MMPs also binds up to three calcium ions and one zinc ion. These ions are believed to stabi-lize the structure (see Figure 7 below). The catalytic mechanism involves the activation of a water molecule for nucleophilic attack of the carbonyl carbon of the scissile bond by a glutamate residue, acting as a general base, and the histidine-coordinated zinc ion as a Lewis acid. Although several subsites affect the substrate specificity of MMPs, the S1’ pocket is a major determi-nant of substrate specificity and generally accommodates hydrophobic resi-dues and excludes charged residues. MMP-12 is an exception since it also cleaves substrates with a P1’ lysine residue, although leucine is preferred [27].

MMPs are synthesized as inactive proenzymes, and exported to the ex-tracellular matrix where they are activated. In the proMMPs, a conserved cysteine within the propeptide coordinates to the active site zinc, thereby replacing the catalytic water and obstructing the substrate binding site [28]. Activation of proMMPs in vivo is initiated either by chemicals such as nitric oxide, modifying the cysteine-switch of the pro-peptide, or by partial prote-olysis of the propeptide by other proteases [29]. Final removal of the propep-tide is accomplished by another MMP. MMP-12 is special in that also the hemopexin domain is removed, and hence the biologically active MMP-12 consists of the catalytic domain only [23]. The activated MMPs are under strict control of the tissue inhibitors of metalloproteases (TIMPs) [30]. The substrates of MMPs are the proteins that make up the connective tissue in the ECM but there are growing evidence that MMPs also act on substrates that control many biological functions. The main substrate for MMP-12 is elastine but it also hydrolyses for example plasminogen to yield angiostatin, which is an inhibitor of angiogenesis. The growing number of substrates reported, apart from ECM-components, suggests a more complex role of MMPs than mere ECM turnover [31].

17

HIV-1 protease HIV-1 protease is an aspartic acid peptidase of the retropepsin family [1,2]. Inhibitors of this enzyme are used for the treatment of AIDS, due to the es-sential function of this enzyme in polyprotein processing during viral repli-cation of HIV [32,33].

The active enzyme is a homodimer with an active site consisting of two catalytic aspartic acid residues, one from each monomer (Figure 4). These aspartates activate a water molecule for nucleophilic attack of the carbonyl carbon of the scissile amide bond. A tunnel-shaped substrate binding-site is formed between the two monomers, with two -hairpin structures called the flaps, constituting the roof of the tunnel [34]. The flaps are flexible and open in absence of substrate. Binding of substrate (or inhibitor) initiates a confor-mational change, and the flaps close over the substrate.

Figure 4. HIV-1 protease with an inhibitor (B369) bound in the active site (PDB-file 1EBY). The two catalytic aspartates are seen below the inhibitor, interacting with a hydroxyl group of the inhibitor. Above the inhibitor is the flap region, consisting of two flexible flaps, seen here in closed conformation.

There is no consensus sequence in the natural substrates, i.e. the cleavage sites in the viral polyprotein, although hydrophobic residues are generally preferred in the P1 and P1’ position and glutamate or glutamine in P2’. The substrate specificity is rather the result of all interactions spanning the resi-dues of the entire P4-P3’ sequence, and the specificity of any single subsite is affected by the amino acids present in the rest of the peptide [35]. The enzyme is also prone to autohydrolysis, since it contains internal cleav-age sites [36,37]. Thus, the enzyme is not entirely stable in the test tube.

18

Although it is not known what conditions are relevant in the immature viral particle where the protease is active, the dependence of activity on pH and ionic strength is remarkable compared to normal physiological conditions. The maximal activity in activity assays using model peptides of natural cleavage sites is reached at pH 5 and 1 M NaCl.

Protein-ligand interactions Forces of interaction Non-covalent forces are of immense importance for both the structure and interactions of biological molecules. These forces determine the folding of proteins into functional structures and mediate all interactions between molecules. Compared to the covalent bond, these forces are weak but the sum of all such forces between a protein and a ligand can lead to interactions of immense strength. As interactions are of some interest in this thesis, the different types of non-covalent forces will be briefly reviewed.

Electrostatic interactions are governed by the force between charged molecules and are described by Coulombs law. The strongest and most long range is that between two ions, but electrostatic interactions also include dipole-dipole and ion-dipole interactions. A molecule is a dipole when the electron density is unevenly distributed as a consequence of different elec-tronegativity of the elements making up the molecule. A molecule can also acquire a temporary dipole moment in the vicinity of an electrical field gen-erated by an ion or a dipole. Electrostatic forces are affected by factors such as the dielectric constant of the solvent and shielding by ions in the solvent. Hence, the strength of an ion-ion interaction on a protein surface is different from the same interaction in the solvent excluded interior of a protein [38].

Dispersion forces are weak, short-range forces that arise between atoms as their electron densities become unevenly distributed and thus create a temporary dipole. It is important to stress that these forces can be repelling as well as attractive, depending on for example the charges of two interact-ing ions or the distance between molecules in dispersion contact.

The hydrogen bond is formed when a hydrogen atom is shared by two highly electronegative elements. The bond is denoted X H Y where X and Y can be oxygen, nitrogen or fluorine. One description of the hydrogen bond is to consider it as the interaction between an acid and a base, where the XH-group is the acid and Y is the base or simply as an electrostatic at-traction between the induced positive charge of the hydrogen and the nega-tive charge of the lone electron pair of Y [39].

Hydrophobic interaction, or the hydrophobic effect, is the tendency for hydrophobic molecules to hide from polar solvents such as water. This is not an actual attractive force between molecules, but instead an effect of entropy

19

off 1 cat

on -1

k k kk k

EI I + E + S ES E + P→

and the second law of thermodynamics. The binding of a hydrophobic inhibitor to an enzyme means that water molecules ordered around the solvated inhibitor become released and thus the total entropy increases [40].

Basic enzyme and interaction kinetics The inhibition of an enzyme (E) by an inhibitor (I) is in the simplest case described by

Scheme 1

where S is a substrate that can be converted into product (P) by enzyme catalysis. The inhibition constant is defined as

offi D

on

kK = K =k

(1)

where kon is the association rate constant for the formation the enzyme inhibitor complex and koff is the dissociation rate constant. KD is the equilibrium dissociation constant for the interaction between enzyme and inhibitor, and its value, in molar units, is defined as the affinity of the interaction. Although both Ki and KD have the same theoretical definition, Kiis used when studying inhibition in an enzyme activity assay, e.g. a system like Scheme 1, while KD is used to describe the interaction between enzyme and inhibitor as measured by a direct binding assay.

The conversion of substrate into product by enzyme catalysis is described by the Michaelis-Menten equation [41]:

0 cat

M

d[P] [E ] k S = dt K + S

× × (2)

[E0] is the total enzyme concentration and kcat is the number of substrate molecules that can be turned over into product per unit time at saturating substrate concentration, and it is therefore also called the turnover number. For a reaction such as Scheme 1 but without inhibitor (I), KM is defined as

cat -1M

1

k + kK =k

(3)

A competitive inhibitor affects the value of KM but not kcat. Hence, for the complete Scheme 1, equation 2 must be modified into

0 cat

Mi

d[P] [E ] k S = dt [I]K 1 + S

K

× ×

× + (4)

20

in order to account for competitive inhibition. In order to compare substrate or enzyme specificity the specificity constant (kcat/KM) must be used and it is defined as the second order rate constant for the formation of an enzyme-substrate complex (k1) multiplied by the probability that product is formed from the ES complex.

Basic thermodynamics Interactions can also be described in terms of Gibbs free energy ( G). The energy diagram (Figure 5) illustrates the interaction between enzyme and inhibitor in terms of free energy change as described by transition state the-ory. The difference between the energy for the two separate interactants (E + I) and the transition state (EI‡) is the free energy for complex association

Gon. The corresponding free energy for complex dissociation is Goff and

on offG = G - G . (5)

Figure 5. Free energy profile for the interaction between an enzyme and an inhibitor. Gon and Goff are the change in free energy required for formation of the transition

state ([EI‡]) starting from reactants (E + I) and complex (EI) respectively. G is the change in free energy between reactants and complex and determines the affinity according to Equation 6.

As a consequence of the second law of thermodynamics, the free energy needs to decrease, i.e. the condition G < 0 must hold, for the interaction to take place spontaneously. The relationship between the equilibrium dissocia-tion constant and the free energy is

21

GRT

DK = e (6)

The free energy is linked to the kinetic rate constants by

oon

onB

k C hG = -RT ln k T

(7)

offoff

B

k hG = -RT lnk T

(8)

where h is the Planck constant, kB is the Bolzmann constant, R is the gas constant, T is absolute temperature and C is the transmission factor, which is set to 1 for the standard state of 1 M.

22

Aim of Present Investigation

The aim of the present study has been to characterize protease-ligand inter-actions in the context of drug discovery. More specifically, this included:

Exploring interactions between enzymes and inhibitors by studying the pH-dependency of HIV-protease inhibitor interactions. Development of a biosensor assay for MMP-12 and the study of in-teractions between this enzyme and inhibitors. Characterization of calcium interactions with MMP-12 for the pur-pose of a better understanding of this enzyme. To study the structure activity relationship of novel inhibitors of Hepatitis C NS3 protease.

Now that many proteases have been identified as drug targets, the next step is the discovery of compounds that inhibit their activity. Drug discovery can be done by random screening, but more appealing to the scientific mind is to use a rational approach. Rational drug discovery is based on techniques such as protein crystallography and computer modeling and is guided by the con-cept of structure activity relationships (SAR). However, the emphasis is on structure rather than activity. Activity is the biological effect of the drug and that needs to be studied in assays where molecules are in dynamic interac-tion. So, although structural information and modeling are indispensable tools in rational drug discovery, relevant information about protein-ligand interactions are crucial for SAR analysis. After all, the function is not an intrinsic property of the structure but emerges as molecules meet and inter-act. There is a constant need to investigate the methods of drug discovery in order to improve SAR and the predictions made on the basis of SAR. Be-cause interaction kinetic information can improve SAR and thus accelerate protease targeted drug discovery, it is important to study both the techniques for understanding molecular interactions and its applications.

This thesis makes a contribution towards the better understanding of pro-tease-ligand interactions, applicable to both protease inhibitor interactions and protease activity regulation, and as such it will have an impact on the field of protease inhibitor development and drug discovery in general.

23

Experimental Methods

Recombinant enzymes

HIV-1 protease DNA encoding HIV-1 protease (Q7K) was excised from a pGEM vector [42] by restriction enzyme cleavage and inserted into a pET-11a vector for expression in E. coli. Expression and purification were performed as de-scribed in Danielson et al. [43]. The procedure involved collection of inclu-sion bodies, followed by denaturation in 8 M urea and subsequent anion exchange chromatography. Refolding was performed by dialysis to decrease the urea concentration, and a final step of cation exchange chromatography concluded the purification procedure. Usually, from a 3 l culture, the yield was 2-5 mg of protease of 95 % purity, as estimated from silver stained SDS-PAGE gels. The only visible contaminations were of lower molecular weight than HIV protease, probably originating from the autohydrolytic ac-tivity of the enzyme itself.

HCV NS3/NS4A protease Expression and purification of the HCV NS3 protease-helicase/NTPase was performed as previously described [44]. The general procedure involved expression in E. coli followed by IMAC-chromatography of the cleared lys-ate. After desalting, chromatography using a poly-U sepharose column yielded 1-2 mg of pure (>98 %) protein, starting with a 6 l culture. As NS4A cofactor, a synthetic peptide corresponding to the central part of the NS4A protein was used. The amino acid sequence of the peptide, KKGSVVIV-GRIVLSGK, is the same as for strain H77, except that the fourth amino acid is S instead of C. Two lysine residues are also added in the N-terminus for increased solubility.

MMP-12 The catalytic domain of MMP-12 was obtained from an external source, where it was purified as described in Morales et al. [45]. The enzyme was pure as judged by SDS-PAGE, developed by silver staining. The only visible

24

contaminations were fragments originating from MMP-12 autohydrolysis (paper IV).

Interaction studies Enzyme activity assays Since proteases cleave their substrate into two products the principle of fluo-rescence resonance energy transfer (FRET) is conveniently used to measure protease activity, and it has been applied to study the activity of all three proteases throughout this entire work. The FRET-substrate consists of a syn-thetic peptide with an amino acid sequence corresponding to the natural cleavage site of the protease of interest. These peptides contain an excitable fluorescent group (e.g. EDANS) in one end and a quenching group (e.g. DABCYL) in the other end. Upon excitation of such a peptide, energy is transferred from the excited fluorescent group to the quenching group, by the principle of FRET, and no measurable light will be emitted. Cleavage of the peptide separates the fluorescent group and the quenching group, and the fluorescent group will emit light. Hence, formation of product can be fol-lowed in time by an increase in fluorescence. Typically, buffer and enzyme is added to wells of a 96-well plate, which is inserted into a fluorescence plate reader and thermostated, whereupon the reaction is started by adding substrate. For the determination of KM and kcat, equation 2 can be used for non-linear regression analysis of initial velocity data as a function of sub-strate concentration. By including an inhibitor in the reaction mixture, the system described in Scheme 1 can be studied and inhibition constants deter-mined by means of equation 4.

Direct binding assays In a direct binding assay, it is possible to study the interaction between two molecules directly, without for example the use of a substrate as a reporter of enzyme inhibitor interactions. In this work, SPR based biosensor technology has been used to study the interactions between proteases and inhibitors (Pa-pers I and V). An advantage of using a biosensor is that the interaction ki-netic constants can be determined, i.e. the affinity can be resolved into asso-ciation and dissociation rate constants according to equation 1. The method is based on surface plasmon resonance (SPR), which is an optical phenome-non occurring at the interface of surfaces. SPR is used to detect changes in refractive index and the signal is related to the change in mass close to the surface. Hence, the binding of an analyte to the surface is detected in reso-nance units (RU) and followed in real time. The data is presented in a sen-sorgram, which is a graph of the response as a function of time. The surface

25

consists of a carboxymethylated dextrane layer on a gold-coated glass sup-port. It is possible to immobilize molecules in the dextrane matrix by a vari-ety of chemical coupling methods. HIV-protease and MMP-12 were both immobilized by amine coupling, which means that an amide bond is formed by a reaction between primary amines on the proteins and the carboxylates on the surface. During interaction experiments a baseline response is ob-tained when only buffer flows over the surface. Injection of an analyte starts the association phase where the analyte binding to the surface is followed in time and recorded in resonance units (RU) (Figure 6). Stopping the injection starts the dissociation phase. Usually, a regeneration procedure follows, where a solution capable of removing residually bound analyte without af-fecting the binding capacity of the surface is injected. This prepares the sur-face for a new analyte injection.

Figure 6. A typical sensorgram.

Substrate or no substrate? Enzyme-inhibitor interactions can be studied using enzyme activity assays, or by using direct binding assays such as the biosensor. The difference is that the former makes use of a substrate, measuring initial rate steady state kinet-ics, yielding equilibrium constants to describe the interaction while in the latter the interaction between enzyme and inhibitor can be studied without a substrate, resulting in kinetic rate constants to describe the interaction. Fur-thermore, the chemodynamics and thermodynamics of enzyme-inhibitor interactions are more easily studied in a biosensor assay, since there is no enzyme-substrate interaction to account for. In an activity assay, catalysis sets the limit for practicable assay conditions since the signal originates from

26

hydrolysis of substrate. For the same reason, the buffer conditions of activity assays are often optimized to generate as high signal as possible. These con-ditions do not necessary mimic physiological conditions. Using a biosensor, binding experiments can be optimized for relevant physiological conditions. The efficiency of high throughput screening for the identification of binders in compound libraries can also be increased by using biosensor assays. Both methods are needed, but should be used for different purposes and ideally in combination. An activity assay must, for example be used to actually deter-mine inhibition of enzyme activity, and it can be sufficient for many SAR studies. A biosensor assay is a tool for deeper understanding of the interac-tion since it provides interaction rate information, which can improve the conclusions of SAR studies.

27

Results

Protease activity regulation MMP-12 and Calcium (Paper IV) The reason for starting to work with MMP-12 was to develop a biosensor assay in order to study the specificity of inhibitors interacting with MMPs. To be able to do that, suitable conditions had to be applied to maximize en-zyme stability and activity on the chip surface. It was known that the enzyme needed calcium for full activity. However, when studying the literature it was striking how little was known about the calcium dependence of MMPs in general and MMP-12 in particular. In fact, it seemed that the question of whether calcium has a regulatory impact on enzyme activity or if it is merely a structural requirement of this enzyme had not been asked. Hence, a charac-terization of the calcium and pH dependence of MMP-12 activity was needed (Paper IV) in order to make possible a discussion of the role of cal-cium for these enzymes.

A fluorometric assay for activity measurements was used to measure the activity as a function of [Ca2+] and pH. It was found that the activity was directly correlated to the concentration of calcium and that a model for three independent binding sites could describe the data. At pH 6.5 a two-site model described the data as well as the three-site model and at pH 4.5 a one-site model was sufficient. The binding sites turned out to have significantly different affinities for calcium (Table 1). Having determined the affinities of the binding sites at different pH-values it was also possible to estimate the pKa-values of the calcium binding sites.

Table 1. Affinities and pKa-values for the three calcium binding-sites of MMP-12.

Binding site KD at pH 7.5 (mM) pKa

High affinity 0.1 4.3 Intermediate 4.6 6.5 Low >100 >6.5

The pKa of the intermediate and low affinity binding sites were high, consid-ering that the ionizable residues of the calcium binding sites are Asp and Glu, which usually have pKa-values in proteins between 3 and 6. This may be explained by the titration of other residues than inner sphere ligands that

28

affect the binding sites. Calcium binding is not only dependent on the ligands in direct contact with the ion, but also on factors such as charge bal-ancing by distant residues and the occurrence of a hydrogen bonding net-work supporting the structure of the binding site. Nevertheless, we proposed an assignment of affinities to the binding sites, based on structural informa-tion of the inner sphere ligands since they were found to differ significantly (Figure 7). Considering the cautions addressed above, this assignment ought to be tested, for example by studying the effect of mutations of individual residues of the binding sites.

Figure 7. Structure of MMP-12 generated using the coordinates with PDB ID 1JK3 and the program WebLab Viewer. Based on the suggested assignment, the calcium ions (green) and their corresponding binding sites are numbered from high (1) to low (3) affinity and can also be seen in close-up to the left. The active site is visualized by one of the zinc ions (blue) interacting with three histidines and an inhibitor. A structural zinc ion is seen in the center of the protein.

Is there any biological relevance to the results of this study? Well, it was shown that small variations in physiological conditions would have a large effect on MMP-12 activity. One would expect that calcium binding for struc-tural reasons only, should be an interaction of high affinity, contrary to the relatively low affinities determined in this study. The results were interpreted so as that MMP-12 should be active in microenvironments of high calcium concentrations, such as during inflammation, and not elsewhere where it could have a deleterious effect. In fact, macrophages, the cells that express

29

MMP-12, are attracted to inflamed tissue by sensing the high calcium con-centration. Thus calcium binding could function as a safety mechanism that limits proteolytic activity to a certain space and time. In that sense the cal-cium requirement of the enzyme has a regulatory function by limiting MMP-12 activity to locations of high calcium concentration.

HCV NS3-NS4A In the activity assay for NS3 protease, the concentration of NS4A peptide is 25000 times that of the NS3 concentration. This indicates that the interaction is relatively weak and investigators have reported affinity values of around 5

M [44,46]. This prompted an interest in the interaction between these two proteins and raised a desire to study the full length NS4A and its interaction with NS3. In analogy with the reasoning for using full-length NS3 in inhibi-tor development, an advantage with using full-length NS4A, in for example biosensor based interaction studies would be the increased in vivo likeness of the assay. Thus, based on a published procedure [47], NS4A was PCR am-plified, using a vector that introduced a his-tag, and cloned into a MBP fu-sion protein expression vector. Sequencing of this vector showed that the DNA was correctly inserted and of the expected sequence. However, purifi-cation of this protein was unsuccessful, the main reason being low expres-sion levels, which were probably due to cell toxicity of the expressed pro-tein. Hence, alternative expression systems or strategies should be tested, but for reasons of limited time this fall outside the scope of this work.

Enzyme-inhibitor interaction studies The pH-dependency of HIV protease-inhibitor interactions (Paper I) It had been observed that inhibitors of HIV protease had lower effect in cell culture than in enzyme activity assays and that the correlation between Ki-values and ED50-values was weak [48]. One explanation could be that the physico-chemical conditions were different for the two assays. The most obvious factor was pH and hence this study focused on the effect of pH on the interaction between HIV-protease and a set of structurally diverse inhibi-tors. Seven inhibitors from five different structural classes were selected, which allowed analysis of the pH-dependencies of different structural classes as well as the contribution of different functional groups of the inhibitors.

Biosensor technology had previously been used to characterize the inter-action between inhibitors and HIV protease [49]. Using this biosensor assay, the kinetic rate constants for the interaction between HIV-protease and the inhibitors could be determined at three different pH-values. It was found that

30

the pH-dependency of the kinetic rate constants was unique for every inhibi-tor. For example, while the association rate constant was highest at pH 5.1 for both indinavir and B268, the pH-dependency of the dissociation constant for these two compounds is the opposite of each other (Figure 8). The only trends were that for all inhibitors, except indinavir, affinity was highest at acidic pH, and most inhibitors had the highest association rate at pH 5.1.

Figure 8. Interaction kinetic maps, shown as kon vs. koff for three of the studied in-hibitors, indinavir ( ), B268 ( ) and acetylpepstatin ( ). Each point is marked with the respective pH-value. The diagonals represent affinity and the pH-dependency of the interactions is visualized by the arrows.

The enzyme-inhibitor interactions were also described in terms of the change in Gibbs free energy upon association and dissociation (Figure 9). The bar graph shows the free energies, corresponding to the free energy profile de-picted in Figure 5, for acetyl-pepstatin and indinavir at different pH-values. The graph clearly shows how changing the pH has different effects on the energy for association and dissociation. Generally, G‡on had the lowest values at pH 5.1, while G‡off and affinity decreased with decreasing pH. This suggests that different forces, with different pH optima, are of impor-tance for association and dissociation. The energy difference between pH 7.4 and pH 5.1 or 4.1 for both G‡off and G‡on was 1-7 kJ/mol. That is similar to the energy lost by removing one or two hydrogen bonds.

31

Figure 9. Free energies for the interactions between HIV-1 protease and inhibitors at different pH, as indicated by the values next to each triplet of bars. G‡on = Black,

G‡off = White, G° = Grey. Note that G‡off = G‡on - G°.

Electrostatic interactions (including hydrogen bonds) that could change with pH were identified by studying crystal structures of complexes between HIV-protease and inhibitors. The structural information combined with the pH-profiles made possible the identification of crucial interactions between enzyme and functional groups of the inhibitors. For example, the results suggested that repulsion between the negatively charged carboxylates of acetyl-pepstatin and Asp130 of the enzyme at pH 7.4 was the reason for improved affinity at acidic pH as at least one of these interactants would become protonated (Figure 10).

Figure 10. The C-terminal carboxylate of acetyl-pepstatin is hydrogen bonded to the carboxylate of Asp130 at acidic pH. At basic pH the two negatively charged car-boxylates would repel each other.

32

Hence, if acetyl-pepstatin was a hit-substance to be optimized, this informa-tion could be used to replace the carboxylate with a hydrogen bond donor, perhaps an amine that would also be positively charged at neutral pH. This could potentially improve the interaction significantly. Thus, pH-profiling of enzyme inhibitor interactions can be used as a tool for optimizing inhibitor structures with respect to the electrostatic interactions between enzyme and inhibitors. The unique pH-dependency of the interaction between indinavir and the protease was assigned to the pyridine nitrogen of the inhibitor, which in contact with Arg-8 of the enzyme would be either hydrogen bonded or repelled depending on the ionization states of both groups.

The most important result from this study was the discovery that associa-tion and dissociation show different pH-dependencies, which implied that the forces of interaction are different in the two processes. This is important for understanding protein-ligand interactions in a broader sense. Further-more, this study underlines the importance of selecting relevant assay condi-tions for drug discovery, and it demonstrates the usefulness of a pH-study when optimizing new drug leads.

MMP-12 inhibitor interactions (Paper V) The starting point of this work was an interest in the specificity problem of MMP inhibitors. Several inhibitors of MMPs have been evaluated in clinical trials for various diseases. Unfortunately, all have failed due to toxicity. The reason is most likely inhibition of other MMPs. These inhibitors are de-signed to interact with the catalytic zinc via a zinc-binding group (ZBG) in the inhibitor. Thus, it is not surprising that these inhibitors are not selective since they could in fact bind to any protein with a zinc ion sufficiently ex-posed for interaction with the hydroxamate-group, e.g. all other MMPs. However, in order to address the selectivity problem, we envisioned that modern biosensor technology would be suitable, with the high information output and the opportunity to study the interaction of inhibitors and several enzymes simultaneously. With such an assay, evaluation and design of selec-tive inhibitors would be enhanced. MMP-12 was the starting point of this work.

A biosensor assay for MMP-12 was developed. Knowing the calcium and pH dependence of the enzyme activity (paper IV) was very helpful in this work. Conditions detrimental to the enzyme could be avoided during immobilization and experiments by adding calcium to all reagents and avoiding acidic conditions. Immobilized MMP-12 was in a control experi-ment found to be active. In fact, the activity of immobilized MMP-12 was higher compared to the activity measured in solution.

Suitable regeneration conditions could not be established for hydrox-amate-based inhibitors of high affinity. Regeneration strategies included salt, acid, base, detergents, metal chelators and organic solvents but were either

33

ineffective or detrimental to the binding capacity of the surface. Despite that difficulty, the assay provided association and dissociation rate information for a set of inhibitors. For inhibitors with high affinity, an experimental de-sign where a new enzyme surface was prepared for every inhibitor injection was used. For comparison, these sensorgrams could be overlaid and normal-ized to the response of a reference compound. It was found that inhibitors having a hydroxamate ZBG interacted with the enzyme with very slow dis-sociation rates. This was corroborated by the detection of time dependent inhibition in an activity assay. These experiments suggested that the interac-tion followed a one-step mechanism, such as that of Scheme 1, although a two-step mechanism, involving an additional step of for example conforma-tional change, could not be excluded.

The results provided new insights into the selectivity problem of many MMP-12 inhibitors. In analogy with irreversible inhibitors, which often suf-fer from low selectivity, the slow dissociation of the hydroxamate inhibitors of MMPs could be associated with the selectivity problem. The biosensor assay for MMP-12 should be useful when developing new selective inhibi-tors for this enzyme.

HCV NS3 inhibitor optimization (Paper II and III) A straightforward way of evaluating structure activity relationships for novel compounds is to use an activity assay and measure inhibition. The inhibition constant (Ki) can then be used to rank analogous compounds with systematic alterations of their structures. In that way, the contribution of specific func-tional groups to the interaction/inhibition can be evaluated. Such structure activity relationships can then be used in the design of more potent inhibi-tors. In this case, inhibition constants were determined for novel HCV NS3 protease inhibitors by using the assay described by Poliakov et al. [44,50].

It was previously known that the N-terminal cleavage product inhibited the protease, and peptides derived from the P1-P6 residues of the natural substrates had been demonstrated as potent inhibitors [51,52]. The optimiza-tion of these peptides led to the development of highly potent inhibitors that were tested clinical trials [8]. It has been shown that a C-terminal carboxy-late is crucial in these inhibitors [53], and that it is possible to replace it, using tetrazole or acyl sulfonamide functionalities as bioisosteres, with conserved or even improved inhibitory potency (Figure 11) [54].

34

Figure 11. Replacement of the C-terminal carboxylic acid with tetrazole or acyl sulfonamide as bioisosteres, resulted in inhibitors with improved inhibitory potency [54].

The studies focused on bioisosteric replacement of the C-terminal carboxylic acid and, since the acyl sulfonamide proved successful in hexa- and pen-tapeptides, exploration of the acyl sulfonamide functionality seemed a prom-ising strategy to discover potent inhibitors (paper II). Therefore, tri- and tetrapeptide acyl sulfonamide inhibitors were prepared and their inhibition of the enzyme was determined.

The results showed that acyl sulfonamides were better inhibitors of HCV NS3 protease than the corresponding carboxylic acid inhibitor in both tri- and tetrapeptide inhibitor scaffolds. A benzyl acyl sulfonamide was best in the norvaline-tripeptides. It was also discovered that the P1 side chain has a large impact on the interactions. For example, changing the P1-residue from norvaline (Nva) to vinyl-ACCA improved the affinity by a factor 25 for the carboxylate and 395 for the benzyl acyl sulfonamide compounds.

Combining the best P1 side chain (vinyl-ACCA) and the best acyl sul-fonamide functionality from the Nva tripeptides, the benzyl acyl sulfona-mide, did not yield a better inhibitor. This demonstrated the non-additive nature of side chain functional groups on the affinity in a protease inhibitor. In this case the vinyl-ACCA and the Nva interacts differently with the S1 subsite, ACCA more tightly, and thus the interaction with the rest of the molecule is affected. This can be explained in the same way as subsite adap-tivity, which means that what is bound in one subsite of a substrate binding site will affect what can fit into the other subsites. This is why protease in-hibitor design is not always “linear” and some randomness should be incor-porated into the design process.

Computer modeling of the interaction between the enzyme and acyl sul-fonamide inhibitors suggested that the carbonyl oxygen interacted with the oxyanion hole, while the amide nitrogen interacted with the catalytic his-tidine (Figure 12).

35

Figure 12. Sulfonamide interacting with the active site of HCV NS3 protease

Thus, it was hypothesized that the acidity of the sulfonamides was important for the interaction between inhibitors and the enzyme. Therefore, non-acidic analogues of the benzyl acyl sulfonamide were tested. These inhibitors had 10 to 20-fold higher Ki-values than the corresponding acyl sulfonamides. This demonstrated that the acidity is a crucial function in acyl sulfonamides. Since the pKa of the catalytic histidine is 6.85 it is essentially deprotonated at physiological pH [55]. At acidic pH the interaction between the deprotonated amide nitrogen of the inhibitor and the catalytic histidine should be strength-ened by the possibility to form a hydrogen bond (Figure 13). Thus, it was interesting to study the effect of pH on the inhibition.

Figure 13. The pH-dependency of the interaction between an acyl sulfonamide in-hibitor and the active site of HCV NS3 protease

A set of inhibitors was designed to have C-terminal groups of different pKa-values, and different hydrogen bonding capabilities. Inhibition constants for a selection of these inhibitors were determined at three different pH-

36

values (paper III). The results of this pH-study supported the hypothesis about the binding mode of the sulfonamides since the acidic inhibitors had around 70 times higher Ki-values at pH 8.0 than at pH 6.0. The Ki-value of an acyl cyanamide was similarly affected, suggesting it has a similar binding mode. Interestingly the inhibition constants of the two non-acidic inhibitors included in the pH-study, predicted not to be pH-dependent, had 10 times higher Ki-values at pH 8.0 than at pH 6.0. The possibility of another his-tidine residue, His528 in the helicase domain, interacting with a carbonyl oxygen in the P3 position of the inhibitors could be responsible for this pH-dependency. The pH-study revealed information about the binding mode of these inhibitors and the possibility of inhibitor interactions with the helicase domain. Furthermore, it emphasizes the importance of both assay conditions and enzyme variants (truncated or full-length) when comparing inhibition data from different laboratories.

This set of inhibitors, with diverse P1 functional groups, was also de-signed to discover other functional groups that could work as bioisosteric replacements for the carboxylic acid. Compounds with an acyl cyanamide and an acyl sulfinamide functionality were identified as promising carbox-ylic acid bioisosteres in HCV NS3 inhibitors. The acyl cyanamide and acyl sulfinamide displayed much lower potency (200 times lower EC50-value than Ki-value) in the cell based replicon assay. Although the correlation between EC50 and Ki is weak, EC50-values are generally not more than 10 to 20 times higher than the Ki-values determined by the inhibition assay. The reasons for this can be numerous, but it seems that the enzyme inhibition assay is not efficient in predicting the in vivo efficacy of these compounds. Resolving the kinetics and using dissociation rate constants instead, can give better correla-tion with cell based data [56]. This is one of the reasons for the interest in developing a biosensor assay for the HCV NS3 protease. Such an assay has been developed and will hopefully help to improve both inhibitor design and in vivo predictions in the near future (manuscript in preparation).

37

Conclusions and perspectives

Conclusive summary This investigation has led to concrete results concerning three important drug targets in terms of protease-ligand interactions. This is expected to be of particular use in the specific drug discovery projects, and for drug discov-ery in general by the increased knowledge of biomolecular interactions. The specific results are emphasized below:

Concrete results in the discovery of drugs against Hepatitis C were obtained, with the unraveling of the structure-activity relationships of acyl sulfonamide-based inhibitors and the discovery of new functional groups as bioisosteres in these inhibitors. Furthermore, the mode of binding of these functional groups in the interaction with the active site of NS3 was established by a combination of SAR, structural modeling and the investigation of the effect of pH on the interactions. The regulation of MMP-12 activity by calcium was proposed. Since this enzyme belongs to a group of important drug targets, this can have implications for understanding the involvement of MMPs in disease and also under what conditions to best optimize inhibi-tors for these enzymes. A biosensor assay for the study of interactions between MMP-12 and inhibitors was developed. It was shown that the high affinity of hydroxamate-based inhibitors was due to slow dissociation of the MMP-12-inhibitor complex. It was evident that the pH-dependency of the interaction kinetics of HIV-protease inhibitors differed in a way that could be related to their structures. This prompted the hypothesis that the forces of in-teraction have different impact on the association and dissociation phases of an interaction. The usefulness of pH as a variable to characterize protein-ligand in-teractions has been demonstrated. The results have highlighted the importance of selecting relevant targets and conditions as well as acquire relevant kinetic informa-tion in drug discovery.

38

Relevance for drug discovery The success of any given candidate drug that enters clinical trials is depend-ent upon the prediction of the in vivo efficacy of that compound by the drug discovery process. Therefore the model systems used in the process of de-signing and evaluating inhibitors need to be physiologically relevant. The model systems are based on the enzymology of the target enzyme, using information such as structure, function, substrate specificity and physiologi-cal activity of the target. One important aspect of physiological function is the regulation of enzyme activity since it may affect the conditions under which the enzyme needs to be inhibited. With that in mind, the calcium de-pendence of MMP-12 activity is definitely an issue of concern, and the pre-sent investigation has made possible a starting point to elucidate the signifi-cance and implications of the intriguing calcium dependence of MMPs. Throughout this work, pH have been demonstrated to have a major impact on the interaction between enzymes and inhibitors. This is in itself nothing new but in the context of model systems for drug discovery it demonstrates how crucial it is to use relevant conditions when evaluating inhibitors in order to obtain accurate predictions of in vivo efficacy. Furthermore, since pH was shown to have different impact on association and dissociation, it is important for drug discovery projects that optimize inhibitor structures based on kinetic rate information, and not only affinity, to be extra careful when considering the relevant pH.

Selecting the right target is of course vital in developing relevant model systems for disease. The use of full-length NS3 was based on previous indi-cations that this was important [44,57]. Since then, several investigators have abandoned the protease domain for full-length assays. This is corroborated in the present investigations by results indicating interactions between inhibi-tors and the helicase domain. Recently, new evidence for the helicase influ-ence was found in a study where mutations in the helicase domain affected protease inhibition [58]. One could also question if studying MMPs in solu-tion is the best model. In vivo, MMPs are highly associated with the ex-tracellular matrix and basement membranes. Hence, the attachment of these enzymes to a surface, as in the biosensor assay, is perhaps a better model for the study of inhibitor interactions. The observation of enhanced activity of immobilized MMP-12 in comparison to the activity in solution certainly does not contradict this theory.

Lastly, it is important to consider the parameters used to evaluate SAR. A growing amount of evidence suggests that the parameter giving the best invivo prediction of drug efficacy for many targets is the dissociation rate con-stant [56,59]. However it cannot be excluded that the association rate would be more important in some specific cases. Biosensor studies resolving affin-ity into individual rate constants have contributed immensely to the under-standing of interactions between proteases and inhibitors [60-62]. These

39

studies also support the usefulness of kinetic rate information in drug dis-covery by the detailed data generated in the HIV-protease study and the new insights into the interaction kinetics of hydroxamate inhibitors of MMPs.

To summarize this discussion, it can be concluded that in order to in-crease the rational part of rational drug discovery it is important to remember that the outcome of the drug discovery process is dependent upon the assay conditions used, the parameters that have been optimized and the target that has been used. Furthermore, the use of kinetic rate information in conjunc-tion with structural information will add an extra dimension to structure ac-tivity relationships. This is needed in order to achieve the deductive power necessary to move drug discovery in the direction towards becoming a genu-inely rational process.

Future prospects There is always a desire to extend studies and learn more. That is a natu-

ral part of being a scientist, but for reasons of limited time and resources all exciting studies will never be made. I will therefore finish by making a brief list of some thoughts for further exploration:

The MMP-12 biosensor assay should be extended for other MMPs in order to study selectivity issues. It will also be useful in the dis-covery of second-generation MMP-12 inhibitors and is in fact al-ready applied in a fragment screening approach. It would also be interesting to study the effect of mutations in the calcium binding-sites on MMP-12 activity.

To fully implement a pH-study in a drug discovery project and use the results to guide the design of novel inhibitors.

To develop a biosensor assay for NS3, using full-length NS4A linked to a membrane surface. That would make an in vitro model system which closely resemble the membrane associated NS3/NS4A in the replication complex of HCV.

40

Sammanfattning

Proteaser och läkemedelsutveckling Under 1990-talet utvecklades de första läkemedlen mot HIV och AIDS, de så kallade bromsmedicinerna. De bromsar upp sjukdomsförloppet och för-länger patienternas liv med flera år. Bromsmedicinerna hämmar enzymer som viruset behöver för att fortplanta sig och sprida sig i kroppen. Ett av enzymerna, HIV-proteas, är ett protein som klipper i andra proteiner. Det klippta proteinet delas på så vis upp i olika funktionella delar. Hämmaren fungerar så att den liknar proteinet som ska klippas sönder men kan själv inte klippas. Hämmaren fastnar då i enzymet som sätts ur funktion (Figur 14).

Figur 14. Överst ses ett proteas (saxen) klippa ett protein i två delar. Nederst i bilden har en hämmare blockerat proteaset som då inte kan klippa proteinet.

Då läkemedlen mot HIV visat sig vara så framgångsrika har forskare världen över försökt använda samma strategi för att motverka andra sjukdomar. Pro-teaser identifieras som en viktig faktor i utvecklingen av alltfler sjukdomar. Hepatit C, kronisk obstruktiv lungsjukdom (KOL), psoriasis, multipel scle-ros (MS) och vissa former av cancer är några av de sjukdomar man hoppas kunna bota genom att hämma proteaser. Därför är det viktigt att studera både hur proteaser fungerar som enzymer och hur de interagerar med hämmare.

41

Min forskning Det övergripande målet med arbetet var att studera interaktioner mellan

proteaser och ligander för att bidra till en förbättring av den del av läkeme-delsutveckingen som fokuserar på att hämma enzymer. Uttryckt i mer speci-fika termer studerades tre proteaser, HIV-proteas, Hepatit C virus (HCV) NS3 proteas och matrix metalloproteas 12 (MMP-12), med fokus på inter-aktioner med hämmare samt reglering av aktivitet.

En bra hämmare som dessutom ska kunna användas som läkemedel måste binda hårt till enzymet och helst inte binda till andra enzymer för att minime-ra risken för biverkningar. För att kunna göra sådana hämmare behövs kun-skap om målenzymets funktion och metoder för att mäta att hämmaren bin-der till och blockerar enzymet. Kunskap om interaktioner mellan proteaser och hämmare, samt hur detta bäst studeras och implementeras inom ramen för utveckling av läkemedel kan därför påskynda utvecklingen av nya läke-medel mot en rad sjukdomar.

Reglering av proteasaktivitet MMP-12 är ett enzym från människa som är viktigt för vävnadsomsättning vid exempelvis inflammation eller sårläkning. Hämmare av MMP-12 tros kunna fungera som läkemedel mot bl.a. KOL. I en av studierna visades att aktiviteten av MMP-12 var beroende av kalciumjoner och pH på ett sådant sätt att kalcium kan reglera enzymets aktivitet. Slutsatsen blev att aktiviteten av MMP-12 i vävnad kan regleras av calciumkoncentrationen.

Interaktioner mellan proteaser och hämmare En studie av pH-beroendet för interaktionen mellan hämmare och HIV-proteas genomfördes. För att erhålla så detaljerad information som möjligt användes biosensorteknik, en metod som gör det möjligt att i realtid följa hur en hämmare binder till (associerar) ett enzym och sedan hur det släpper från (dissocierar) enzymet. Resultaten visade att varje hämmare gav upphov till en interaktion med unikt pH-beroende, vilket kunde tillskrivas olika struktu-rella element i hämmarna. Dessutom visade det sig att associationen och dissociationen hade olika pH-beroende. Därför framlades hypotesen att kraf-terna som verkar attraherande och binder hämmare och enzym samman är av olika betydelse i associations- och dissociationsfas.

Även MMP-12 studerades med biosensorteknik men här låg fokus på att utröna hur en viss typ av hämmare interagerar med enzymet. Dessa så kalla-de hydroxamatbaserade hämmare binder hårt till enzymet, genom interaktion med en zinkjon som är essentiell för enzymets funktion. Det visade sig att när dessa hämmare väl har bundit till MMP-12 så sitter de kvar väldigt länge, d.v.s. dissociationen är mycket långsam.

42

För HCV NS3 proteas studerades interaktionen med hämmare indirekt genom att mäta hur de påverkar aktiviteten. Nya föreningar med potential att utvecklas till bra hämmare och sedemera läkemedel hittades. En studie över hur pH påverkade föreningarnas förmåga att hämma proteaset visade hur de interagerade med enzymets aktiva yta och gav således viktig information för fortsatt utveckling av hämmare mot HCV NS3 proteas.

SlutsatserResultaten som presenteras i avhandlingen visar att vid läkemedels-utveckling är det viktigt att komma ihåg att utfallet är beroende av vilket modellsystem som används för att efterlikna det biologiska system i vilket läkemedlet ska verka. En dröm inom enzymologi och läkemedelsutveckling är att utifrån information om ett enzymes struktur och funktion direkt kunna designa en läkemedelsmolekyl, så kallad rationell design av läkemedel. Yt-termera, kan slutsatsen dras att interaktionskinetik i samarbete med struktu-rell information ger den deduktiva förmåga som behövs för att lyckas med drömmen om rationell läkemedelsutveckling.

43

Acknowledgements

I am very grateful for the opportunity I have had to complete this thesis work, and for that I would like to thank Helena for offering me a place in her research group and for the financial support from Sven och Lilly Lawskis fond för naturvetenskaplig forskning, VINNOVA and Medivir AB.

Helena is also greatly acknowledged for being an excellent supervisor, by means of a true passion for science and teaching you have made sure that I have gradually grown as a scientist over these years.

Thanks to all my co-authors, and especially: Robert because it has been a pleasure to collaborate with you over the years and Helena N for a fruitful collaboration and for sharing with me your expertise in Biacore technology.

Thanks to all the present and past members of HD’s research group: Cynthiaand Maria for getting me started in the lab in the best possible way and for always sharing your experiences; Dan for encouraging me to join the group and sharing the path of pH; Matthis for the introductory crash course on the 2000 instrument; Anton for introducing me to HCV NS3; Göran for helpful comments on this summary and for sharing with me the hardship of working under the spell of NS3; Omar for invaluable help on complicated equations and your ability to always make people smile; Malin for leading the work with UUCC 2006 and for help on the NS4A-project; Angela for being as focused on writing as me during our time as roommates.

Bengt, Birgitta T, Micke, Gun, Gunnar, and Françoise for contributing to my scientific education and for being a source of inspiration as researchers and teachers.

Thank you, all my colleagues at the department of biochemistry and organic chemistry for friendship and professionalism. Thanks for everything from help with practical issues and scientific discussions to company around the coffee table. I would especially like to mention Ann-Christine for our time as roommates and Sandra for help on the NS4A-project, and also Lars E, Na-talia, Arna, Sanela, Inger, Birgit, Birgitta E, Ylva, Lisa B, Lilian, Usama,

44

Abeer, Per-Axel, Diana, Olof, Eric, Malena, Ann, Nisse, Lisa T and Anna-Karin.

Thanks also to the people at Medivir AB: Hans for a nice collaboration on paper V and Lotta and Susanne for support over the years.

Jag vill också passa på att tacka alla mina vänner som under arbetet med detta verk fått mig att också tänka på annat än biokemi ibland. Speciellt vill jag tacka: ”Falugänget”: Alex, Cecilia, Linus, Stina, Mattias, Anna M, Mats, Petra för allt roligt vi haft tillsammans under åren och att vi grabbar som träffades som finniga tonåringar fortfarande har barnsligt kul ihop. Magnus för du är en vän att alltid lita på och för att min lekamen inte förfal-lit helt då du lyckats släpa med mig till gymmet ibland. Familjerna Landegren-Björk och Snis-Kallvi: Vilken tur att Morgan, Felix och Alva var kompisar så att vi blev ett litet pappa-gäng under hösten. Min svärfamilj: Bror, Britt, Camilla, Ola, Selma, Helena, Gustav för er ge-nerositet och att det alltid är lika trevligt att umgås med er. Mor och Far för allt stöd och för att ni alltid uppmuntrat mig att söka mina egna vägar. Carina för att du alltid ställer upp när det behövs och Erik för vi alltid har så kul ihop. Ni är världens bästa syster och bror!

Anna för att du ville dela livet med mig och Alva för din gränslösa kärlek som kan få en att uträtta vad som helst. Ni är mitt bästa! Jag älskar er!

45

References

1. The MEROPS database of peptidases (accessed in march 2007), http://merops.sanger.ac.uk/

2. Rawlings, N.D., Morton, F.R. and Barrett, A.J. (2006) MEROPS: the peptidase database. Nucleic Acids Res. 34(DATABASE ISSUE):D270-2

3. Schechter, I. and Berger, A. (1967) On the size of the active site in proteases. I. Papain. Biochem. Biophys. Res. Commun. 27(2):157-62

4. Turk, B. (2006) Targeting proteases: successes, failures and future prospects. Nat Rev Drug Discov 5(9):785-99

5. Van Regenmortel, M.H. (2004) Reductionism and complexity in molecular biology. Scientists now have the tools to unravel biological and overcome the limitations of reductionism. EMBO Rep 5(11):1016-20

6. Dash, C., Kulkarni, A., Dunn, B. and Rao, M. (2003) Aspartic peptidase inhibitors: implications in drug development. Crit. Rev. Biochem. Mol. Biol. 38(2):89-119

7. Wlodawer, A. and Vondrasek, J. (1998) Inhibitors of HIV-1 protease: a major success of structure-assisted drug design. Annu. Rev. Biophys. Biomol. Struct. 27:249-84

8. Lamarre, D., Anderson, P.C., Bailey, M., Beaulieu, P., Bolger, G., Bonneau, P., Bos, M., Cameron, D.R., Cartier, M., Cordingley, M.G., Faucher, A.M., Goudreau, N., Kawai, S.H., Kukolj, G., Lagace, L., LaPlante, S.R., Narjes, H., Poupart, M.A., Rancourt, J., Sentjens, R.E., St George, R., Simoneau, B., Steinmann, G., Thibeault, D., Tsantrizos, Y.S., Weldon, S.M., Yong, C.L. and Llinas-Brunet, M. (2003) An NS3 protease inhibitor with antiviral effects in humans infected with hepatitis C virus. Nature 426(6963):186-9

9. Love, R.A., Parge, H.E., Wickersham, J.A., Hostomsky, Z., Habuka, N., Moomaw, E.W., Adachi, T. and Hostomska, Z. (1996) The crystal structure of hepatitis C virus NS3 proteinase reveals a trypsin-like fold and a structural zinc binding site. Cell 87(2):331-42

46

10. Yao, N., Reichert, P., Taremi, S.S., Prosise, W.W. and Weber, P.C. (1999) Molecular views of viral polyprotein processing revealed by the crystal structure of the hepatitis C virus bifunctional protease-helicase. Structure 7(11):1353-63

11. Johansson, A., Hubatsch, I., Akerblom, E., Lindeberg, G., Winiwarter, S., Danielson, U.H. and Hallberg, A. (2001) Inhibition of hepatitis C virus NS3 protease activity by product-based peptides is dependent on helicase domain. Bioorg. Med. Chem. Lett. 11(2):203-6

12. Grakoui, A., McCourt, D.W., Wychowski, C., Feinstone, S.M. and Rice, C.M. (1993) Characterization of the hepatitis C virus-encoded serine proteinase: determination of proteinase-dependent polyprotein cleavage sites. J. Virol. 67(5):2832-43

13. Pizzi, E., Tramontano, A., Tomei, L., La Monica, N., Failla, C., Sardana, M., Wood, T. and De Francesco, R. (1994) Molecular model of the specificity pocket of the hepatitis C virus protease: implications for substrate recognition. Proc. Natl. Acad. Sci. U. S. A. 91(3):888-92

14. Steinkuhler, C., Urbani, A., Tomei, L., Biasiol, G., Sardana, M., Bianchi, E., Pessi, A. and De Francesco, R. (1996) Activity of purified hepatitis C virus protease NS3 on peptide substrates. J. Virol. 70(10):6694-700

15. Butkiewicz, N.J., Wendel, M., Zhang, R., Jubin, R., Pichardo, J., Smith, E.B., Hart, A.M., Ingram, R., Durkin, J., Mui, P.W., Murray, M.G., Ramanathan, L. and Dasmahapatra, B. (1996) Enhancement of hepatitis C virus NS3 proteinase activity by association with NS4A-specific synthetic peptides: identification of sequence and critical residues of NS4A for the cofactor activity. Virology 225(2):328-38

16. Shimizu, Y., Yamaji, K., Masuho, Y., Yokota, T., Inoue, H., Sudo, K., Satoh, S. and Shimotohno, K. (1996) Identification of the sequence on NS4A required for enhanced cleavage of the NS5A/5B site by hepatitis C virus NS3 protease. J. Virol. 70(1):127-32

17. Tomei, L., Failla, C., Vitale, R.L., Bianchi, E. and De Francesco, R. (1996) A central hydrophobic domain of the hepatitis C virus NS4A protein is necessary and sufficient for the activation of the NS3 protease. J. Gen. Virol. 77(PT5):1065-70

18. Love, R.A., Parge, H.E., Wickersham, J.A., Hostomsky, Z., Habuka, N., Moomaw, E.W., Adachi, T., Margosiak, S., Dagostino, E. and Hostomska, Z. (1998) The conformation of hepatitis C virus NS3 proteinase with and without NS4A: a

47

structural basis for the activation of the enzyme by its cofactor. Clin. Diagn. Virol. 10(2-3):151-6

19. Moradpour, D., Gosert, R., Egger, D., Penin, F., Blum, H.E. and Bienz, K. (2003) Membrane association of hepatitis C virus nonstructural proteins and identification of the membrane alteration that harbors the viral replication complex. Antiviral Res. 60(2):103-9

20. Kim, J.L., Morgenstern, K.A., Lin, C., Fox, T., Dwyer, M.D., Landro, J.A., Chambers, S.P., Markland, W., Lepre, C.A., O'Malley, E.T., Harbeson, S.L., Rice, C.M., Murcko, M.A., Caron, P.R. and Thomson, J.A. (1996) Crystal structure of the hepatitis C virus NS3 protease domain complexed with a synthetic NS4A cofactor peptide. Cell 87(2):343-55

21. Landro, J.A., Raybuck, S.A., Luong, Y.P., O'Malley, E.T., Harbeson, S.L., Morgenstern, K.A., Rao, G. and Livingston, D.J. (1997) Mechanistic role of an NS4A peptide cofactor with the truncated NS3 protease of hepatitis C virus: elucidation of the NS4A stimulatory effect via kinetic analysis and inhibitor mapping. Biochemistry 36(31):9340-8

22. Bode, W., Gomis-Ruth, F.X. and Stockler, W. (1993) Astacins, serralysins, snake venom and matrix metalloproteinases exhibit identical zinc-binding environments (HEXXHXXGXXH and Met-turn) and topologies and should be grouped into a common family, the 'metzincins'. FEBS Lett. 331(1-2):134-40

23. Shapiro, S.D., Kobayashi, D.K. and Ley, T.J. (1993) Cloning and characterization of a unique elastolytic metalloproteinase produced by human alveolar macrophages. J. Biol. Chem. 268(32):23824-9

24. Snider, G.L. (2003) Understanding inflammation in chronic obstructive pulmonary disease: the process begins. Am. J. Respir. Crit. Care Med. 167(8):1045-6

25. Liu, M., Sun, H., Wang, X., Koike, T., Mishima, H., Ikeda, K., Watanabe, T., Ochiai, N. and Fan, J. (2004) Association of increased expression of macrophage elastase (matrix metalloproteinase 12) with rheumatoid arthritis. Arthritis Rheum. 50(10):3112-7

26. Rosenberg, G.A. (2002) Matrix metalloproteinases and neuroinflammation in multiple sclerosis. Neuroscientist 8(6):586-95

27. Gronski Jr, T.J., Martin, R.L., Kobayashi, D.K., Walsh, B.C., Holman, M.C., Huber, M., Van Wart, H.E. and Shapiro, S.D. (1997) Hydrolysis of a broad spectrum of extracellular matrix

48

proteins by human macrophage elastase. J. Biol. Chem. 272(18):12189-94

28. Springman, E.B., Angleton, E.L., Birkedal-Hansen, H. and Van Wart, H.E. (1990) Multiple modes of activation of latent human fibroblast collagenase: evidence for the role of a Cys73 active-site zinc complex in latency and a “cysteine switch” mechanism for activation. Proc. Natl. Acad. Sci. U. S. A. 87(1):364-8

29. Visse, R. and Nagase, H. (2003) Matrix metalloproteinases and tissue inhibitors of metalloproteinases: structure, function, and biochemistry. Circ. Res. 92(8):827-39

30. Brew, K., Dinakarpandian, D. and Nagase, H. (2000) Tissue inhibitors of metalloproteinases: evolution, structure and function. Biochim. Biophys. Acta 1477(1-2):267-83

31. Overall, C.M. and Kleifeld, O. (2006) Tumour microenvironment - opinion: validating matrix metalloproteinases as drug targets and anti-targets for cancer therapy. Nat Rev Cancer 6(3):227-39

32. Kohl, N.E., Emini, E.A., Schleif, W.A., Davis, L.J., Heimbach, J.C., Dixon, R.A., Scolnick, E.M. and Sigal, I.S. (1988) Active human immunodeficiency virus protease is required for viral infectivity. Proc. Natl. Acad. Sci. U. S. A. 85(13):4686-90

33. Piacenti, F.J. (2006) An update and review of antiretroviral therapy. Pharmacotherapy 26(8):1111-33

34. Wlodawer, A. and Erickson, J.W. (1993) Structure-based inhibitors of HIV-1 protease. Annu. Rev. Biochem. 62:543-85

35. Beck, Z.Q., Morris, G.M. and Elder, J.H. (2002) Defining HIV-1 protease substrate selectivity. Curr Drug Targets Infect Disord 2(1):37-50

36. Mildner, A.M., Rothrock, D.J., Leone, J.W., Bannow, C.A., Lull, J.M., Reardon, I.M., Sarcich, J.L., Howe, W.J., Tomich, C.S., Smith, C.W. and et, a. (1994) The HIV-1 protease as enzyme and substrate: mutagenesis of autolysis sites and generation of a stable mutant with retained kinetic properties. Biochemistry 33(32):9405-13

37. Strickler, J.E., Gorniak, J., Dayton, B., Meek, T., Moore, M., Magaard, V., Malinowski, J. and Debouck, C. (1989) Characterization and autoprocessing of precursor and mature forms of human immunodeficiency virus type 1 (HIV 1) protease purified from Escherichia coli. Proteins 6(2):139-54

38. Fersht, A. (1999) Structure and Mechanism in Protein Science: A Guide to Enzyme Catalysis and Protein Folding, W. H. Freeman and Company, New York

49

39. Atkins, P.W. (1996) The Elements of Physical Chemistry. , pp 365-378, Oxford University Press, Oxford

40. van Holde, K.E., Johnson, C.W. and Ho, S.P. (1998) Principles of Physical Biochemistry, Prentice-Hall Inc., Upper Saddle River, New Jersey

41. Michaelis, L. and Menten, M. (1913) Bichem. Z. 49:333 42. Markgren, P.-O., Lindgren, M.T., Gertow, K., Karlsson, R.,

Hämäläinen, M. and Danielson, U.H. (2001) Determination of interaction kinetic constants for HIV-1 protease inhibitors using optical biosensor technology. Anal. Biochem. 291(2):207-18

43. Danielson, H., Lindgren, M.T., Markgren, P.-O. and Nillroth, U. (1998) Investigation of an allosteric site of HIV-1 proteinase involved in inhibition by Cu2+. Adv. Exp. Med. Biol. 436:99-103

44. Poliakov, A., Hubatsch, I., Shuman, C.F., Stenberg, G. and Danielson, U.H. (2002) Expression and purification of recombinant full-length NS3 protease-helicase from a new variant of Hepatitis C virus. Protein Expr. Purif. 25(3):363-71

45. Morales, R., Perrier, S., Florent, J.M., Beltra, J., Dufour, S., De Mendez, I., Manceau, P., Tertre, A., Moreau, F., Compere, D., Dublanchet, A.C. and O'Gara, M. (2004) Crystal structures of novel non-peptidic, non-zinc chelating inhibitors bound to MMP-12. J. Mol. Biol. 341(4):1063-76

46. Bianchi, E., Urbani, A., Biasiol, G., Brunetti, M., Pessi, A., De Francesco, R. and Steinkuhler, C. (1997) Complex formation between the hepatitis C virus serine protease and a synthetic NS4A cofactor peptide. Biochemistry 36(25):7890-7

47. Back, S.H., Kim, J.E., Rho, J., Hahm, B., Lee, T.G., Kim, E.E., Cho, J.M. and Jang, S.K. (2000) Expression and purification of an active, full-length hepatitis C viral NS4A. Protein Expr. Purif. 20(2):196-206

48. Hämäläinen, M.D., Markgren, P.-O., Schaal, W., Karlén, A., Classon, B., Vrang, L., Samuelsson, B., Hallberg, A. and Danielson, U.H. (2000) Characterization of a set of HIV-1 protease inhibitors using binding kinetics data from a biosensor-based screen. J. Biomol. Screen. 5(5):353-60

49. Markgren, P.-O., Hämäläinen, M. and Danielson, U.H. (2000) Kinetic analysis of the interaction between HIV-1 protease and inhibitors using optical biosensor technology. Anal. Biochem. 279(1):71-8

50. Johansson, A., Poliakov, A., Akerblom, E., Lindeberg, G., Winiwarter, S., Samuelsson, B., Danielson, U.H. and Hallberg, A. (2002) Tetrapeptides as potent protease

50

inhibitors of Hepatitis C Virus full-length NS3 (protease-helicase/NTPase). Bioorg. Med. Chem. 10(12):3915-22

51. Steinkuhler, C., Biasiol, G., Brunetti, M., Urbani, A., Koch, U., Cortese, R., Pessi, A. and De Francesco, R. (1998) Product inhibition of the hepatitis C virus NS3 protease. Biochemistry 37(25):8899-905

52. Llinas-Brunet, M., Bailey, M., Fazal, G., Goulet, S., Halmos, T., Laplante, S., Maurice, R., Poirier, M., Poupart, M.A., Thibeault, D., Wernic, D. and Lamarre, D. (1998) Peptide-based inhibitors of the hepatitis C virus serine protease. Bioorg. Med. Chem. Lett. 8(13):1713-8

53. Ingallinella, P., Altamura, S., Bianchi, E., Taliani, M., Ingenito, R., Cortese, R., De Francesco, R., Steinkuhler, C. and Pessi, A. (1998) Potent peptide inhibitors of human hepatitis C virus NS3 protease are obtained by optimizing the cleavage products. Biochemistry 37(25):8906-14

54. Johansson, A., Poliakov, A., Akerblom, E., Wiklund, K., Lindeberg, G., Winiwarter, S., Danielson, U.H., Samuelsson, B. and Hallberg, A. (2002) Acyl sulfonamides as potent protease inhibitors of the hepatitis C virus full-Length NS3 (protease-helicase/NTPase): a comparative study of different C-terminals. Bioorg. Med. Chem. 10(12):2551-68

55. Poliakov, A., Sandström, A., Åkerblom, E. and Danielson, U.H. (2006) Mechanistic studies of electrophilic protease inhibitors of full length HCV NS3. Enzyme Inhib Med Chem Accepted

56. Shuman, C.F., Vrang, L. and Danielson, U.H. (2004) Improved structure-activity relationship analysis of HIV-1 protease inhibitors using interaction kinetic data. J. Med. Chem. 47(24):5953-61

57. Johansson, A., Hubatsch, I., Akerblom, E., Lindeberg, G., Winiwarter, S., Danielson, U.H. and Hallberg, A. (2000) Inhibition of hepatitis C virus NS3 protease activity by product-based peptides is dependent on helicase domain. Bioorg. Med. Chem. Lett. 2(5):203-6

58. Dahl, G., Sandström, A., Åkerblom, E. and Danielson, U.H. (2007 (Submitted)) Bifunctional NS3 helicase/protease from hepatitis C; Helicase residues influence the protease functionality.

59. Copeland, R.A., Pompliano, D.L. and Meek, T.D. (2006) Drug-target residence time and its implications for lead optimization. Nat Rev Drug Discov 5(9):730-9

60. Markgren, P.O., Schaal, W., Hamalainen, M., Karlen, A., Hallberg, A., Samuelsson, B. and Danielson, U.H. (2002)

51

Relationships between structure and interaction kinetics for HIV-1 protease inhibitors. J. Med. Chem. 45(25):5430-9

61. Backman, D. and Danielson, U.H. (2003) Kinetic and mechanistic analysis of the association and dissociation of inhibitors interacting with secreted aspartic acid proteases 1 and 2 from Candida albicans. Biochim. Biophys. Acta 1646(1-2):184-95

62. Shuman, C.F., Markgren,