NS3 protease

An obvious drug target in this protein is the viral serine protease domain in the N-terminal part. The enzymatic activity of this protein, which is dependent on at least 40 amino acids of NS2B, is vital for the post-translational proteolytic processing of the polyprotein precursor and is essential for viral replication and maturation of infectious dengue virions (Falgout et al 1991).

Leung et al (2001) have shown that the dengue protease domain fused to 40 amino acids of NS2B through a flexible protease resistant linker (CF40-Gly-NS3pro185) was soluble and catalytically active. A suitable substrate was identified by Li et al (2005) using a tetrapeptide library containing over 130 000 substrates. Not surprisingly this study found a preference for Arg at both P1 and P2 positions, the best substrate for all four serotypes of dengue having the sequence Bz-Nle-Lys-Arg-Arg-AMC (AMC = 7-Amido-4-methylcoumarin). The substrate profiling suggested that P1 and P2 are important both for binding and turnover, while changes in P3 mainly influence binding and P4 amino acids are essential for efficient catalysis.

With all the tools in hand we chose an assay based on dengue virus (DENV) -2 CF40-Gly-NS3pro185 with the fluorogenic substrate Bz-Nle-Lys-Arg-Arg-AMC as our high-throughput screening (HTS) assay. The result of the first HTS campaign was unfortunately not very encouraging. Despite screening over 1 000 000 compounds we were only able to identify a number of promiscuous inhibitors, which either formed aggregates or did not show useful structure—activity relationships (SAR). This outcome was not a total surprise, since serine proteases are notoriously difficult targets for HTS. Because of the importance of the target for our drug discovery effort, we are currently running a second HTS campaign with a slightly different assay format using full length DENV-2 NS3 in parallel with West Nile CF40-Gly-NS3pro185.

Traditionally most drug discovery projects for prototypical serine proteases like thrombin (Gustafsson et al 2004), elastase or HCV NS3 (Llinas-Brunet et al 1998) have started from inhibitors based on substrate peptide sequences. Such a peptido-mimetic approach can often produce very potent inhibitors in a short period of time, however turning these early leads into drugs can be extremely challenging.

The failure of our first HTS made such an approach however necessary, since it allowed us to rapidly map the active site and assess the feasibility of a project based on structure-aided design. As a first step we synthesized a number of peptides with electrophilic warheads (Fig. 2). These warheads interact with the active site serine of the enzyme to form transition state analogues which are very convenient starting points for inhibitor design.

Table 2 shows a selection of peptides that were tested on NS3 protease enzymes from dengue 2, West Nile and yellow fever virus. While the sample is rather small, the data is useful to illustrate a few general trends in SAR that we observed during our work on over 70 peptides (Z. Yin, unpublished work 2005). A non-covalent, charged warhead such as a carboxylic acid theoretically could provide binding efficiency via electrostatic interactions. A product based peptide acid was reported as a competitive inhibitor of the hepatitis C virus NS3-4A serine protease with a Ki of 0.6 |iM (Llinas-Brunet et al 1998). In our hands, carboxylic acid 1 failed to show any activity against Dengue serine protease. Simple amides are known to be relatively inert towards certain serine proteases and can serve as substrate-like enzyme inhibitors (Brady et al 1995). However tetrapeptide amide 2 showed only activity at high micromolar concentration.

Dengue Virus Protease
FIG. 2. Binding mode of transition state inhibitors 3-14. Binding pockets of side chains are labelled P1 to P4 (see text). Transition state inhibitors contain an electrophilic carbonyl group or boron atom which forms a covalent but reversible bond to the active site serine.
TABLE 2 Comparison of peptide activity on flavivirus NS3 proteasesa

K, (\iM)h

K, (\iM)h

K, (\iM)h

Inhibitors

Dengue 2

Yellow fever

West Nile

1

Bz-Nle-Lys-Arg-Arg-OH

>500

nd

nd

2

Bz-Nle-Lys-Arg-Arg-NH2

>500

nd

nd

3

Bz-Nle-Lys-Arg-Arg-H

5.8

0.36

4.10

4

Bz-Lys-Arg-Arg-H

1.30

0.34

0.75

5

Bz-Arg-Arg-H

10.00

0.74

2.15

6

Bz-Nle-Lys-Arg-Arg-Thiazole

36.60

10.50

13.20

7

Bz-Nle-Lys-Arg-Arg-CF3

0.70

0.38

0.30

8

Bz-Nle-Lys-Arg-Arg-B(OH)2

0.04

0.05

0.03

9

Bz-Lys-Arg-Arg-B(OH)2

0.20

0.04

0.04

a Preparation of the enzyme and assay conditions are described in Li et al 2005 b K,s are averages of at least two independent determinations nd, not determined a Preparation of the enzyme and assay conditions are described in Li et al 2005 b K,s are averages of at least two independent determinations nd, not determined

Surprisingly several of the standard serine protease warheads do not work very well for the examined NS3 proteases. Inhibitors incorporating an a-keto hetero-cycle moiety like 6 proved to be less active than aldehyde 3. These heterocyclic groups gave potent inhibitors with the related serine proteases elastase, chymase and thrombin (Ni & Wagman 2004). We surmise that their lack of activity may be due to some steric restriction in the active site of the enzyme, which prevented optimal binding. In our hands a-keto amides with a deprotected arginine at P1 were unstable and therefore useless as probes.

Boronic acids, trifluoroketone and aldehydes gave the best results. These pep-tides have helped us to get invaluable insight into the behaviour of the enzymes and allowed us to make rapid progress towards X-ray crystal structures of NS3 proteases. A detailed discussion of the SAR of these peptides is beyond the scope of this article, however we currently can say that the shorter peptides (e.g. Table 2, compounds 5 and 9) still maintain appreciable inhibition which bodes well for drug discovery (Z. Yin unpublished work 2005).

Table 3 shows an examination of the importance of the different binding pockets (P1—P4) in dengue 2 NS3 protease. The alanine scan suggested that P1 and P2 are very important for inhibitor binding, which is in good agreement with the substrate profiling studies. A slight surprise is that the replacement of the arginine at P2 causes a greater decrease in potency than the one at P1. The inhibitor residues at P3 only contribute a small amount of binding energy, while the norleucine at P4 can be eliminated without any affect on inhibitor potency. The latter result matches our substrate studies in which a suboptimal substitution at P4 maintained Km but displayed sevenfold decrease in kcat [substrates Bz-nKRR-ACMC (kcat = 1.39 s-1) and Bz-TKRR-ACMC (kcat = 0.20 s-1)) (Li et al 2005).

KELLER ET AL TABLE 3 Effect of alanine scan on dengue inhibition3

Inhibitors

10 11 12

Bz-Nle-Lys-Arg-Arg-H Bz-Nle-Lys-Arg-^/a-H Bz-Nle-Lys-^/a-Arg-H Bz-Nle-^/a-Arg-Arg-H Bz-^/a-Lys-Arg-Arg-H

5.8 193

a Preparation of the enzyme and assay conditions are described in Li et al 2004.

b K;s are averages of at least two independent determinations.

Moreover, they are in agreement with the recently published data that pointed out marked destabilization of the enzyme-inhibitor interactions in the presence of a small chain residue such as Ala or Ser at P4 (Chanprapaph et al 2005).

The design of inhibitors and virtual screening would be greatly facilitated by structural information. Unfortunately the currently available crystal structures (Murthy et al 2000) lack the NS2B cofactor and are therefore of uncertain value for drug discovery. As a consequence, an intensive effort was mounted to generate protein structures of different NS3 protease constructs in the presence of inhibitors, so far without conclusive results for the dengue NS3 protease. As a substitute, a homology model based on an unpublished X-ray crystal structure of West Nile NS3 protease (N. Schiering, personal communication 2005) was generated and used to rationalize the SAR. Figure 3 shows a snapshot of the homology model with the docked peptide Ac-Lys-Arg-Arg-H (J. Knox, personal communication

Our homology model suggests that the active site of dengue NS3 protease is rather flat with few distinct pockets which could be used for inhibitor binding. As discussed previously this situation makes drug discovery on this target rather challenging and is probably also the reason why the HTS did not produce a good lead. Nevertheless, because of the importance of this target for flavivirus drug discovery we will continue to pursue all avenues of lead discovery.

Was this article helpful?

0 0
Swine Influenza

Swine Influenza

SWINE INFLUENZA frightening you? CONCERNED about the health implications? Coughs and Sneezes Spread Diseases! Stop The Swine Flu from Spreading. Follow the advice to keep your family and friends safe from this virus and not become another victim. These simple cost free guidelines will help you to protect yourself from the swine flu.

Get My Free Ebook


Post a comment