|
|
 |
|
Practical application of the CADDIS process is demonstrated by an MATRIX in-house project, the development of a direct thrombin inhibitor.
Thrombin was chosen as a target because it is so well known in the research community, and any work done by MATRIX in this field can easily be compared with the efficacy of other processes such as X-ray structure analysis or 3D modelling) .
|
|
The substantial financial implications for the pharmaceutical industry of an orally available thrombin inhibitor were also a factor in the decision to focus on the thrombin inhibitor issue.
All research projects undertaken up to now (with the one exception of Xi-Melagatran) have failed, and most of these failures share more or less the same difficulties in combining a high activity, a high selectivity, a not too short elimination time and full bio-availability .
|
Figure 4: Trypsin-like proteinases in the blood
effector systems.
|
The process of thrombin inhibition is embedded in a complex network of blood coagulation. Most of the serine proteases play different roles in many control systems of the body. The risk of undesired side-effects is therefore significantly higher than usual. Fig. 5 shows a very simplified model of the blood coagulation system and its most prominent actors.
The specifications for an interesting candidate for further development have to include:
|
 High affinity with the target enzyme (high potency)
High selectivity towards the target enzyme versus related enzymes
Chemical and metabolic stability
No toxicity
Low to medium serum protein binding
High bio-availability
Half-life appropriate for oral dosing
Based on these specifications, a test program was established. All compounds were tested in all assay systems in the test program right from the beginning. The selected assay systems were:
|
|
High thrombin-inhibition, even at low compound concentrations. |
|
Low trypsin-inhibition. Trypsin is structurally very similar to thrombin and is therefore a very suitable representative for the specificity with respect to other serine proteases. The experiments for the trypsin inhibition have to be undertaken with a high (non-decreasing) concentration of the compound. |
|
The toxic impact on cells has to be investigated at constant high concentrations. The HeLa cells used are very robust and allow an ideal overview of the acute tox behaviour of the compounds. |
|
The haemolytic effect was measured. |
|
The metabolic stability was covered by a microsome test. |
|
The solubility octanol/water should be in the range log D = +1 ... +3. |
|
An additional AMES test was done after running the computer-based optimisation process. This test allows conclusions about the mutagenic potential of the compound. However, this criterion was not measured in all optimisation cycles, and was therefore not the subject of computer optimisation.
The requirement for a high bio-availability was initially left out because the currently available in-vitro test systems (Caco2) are not very expressive.
Fig. 6 outlines the computer-based optimisation process. The first graph shows the measured activities for the initial selection of molecules (0th cycle) . The inhibitory activity on thrombin, which should be as high as possible, is shown vertically. To observe differences in the more or less random initial selection very high concentrations (here: 150 ?M) of the compounds are used at this stage. This concentration is successively reduced in the next cycles. The inhibitory activity on trypsin, horizontally displayed, should be as low as possible. The high concentration of 150 µM remains unchanged throughout the project.
Two further criteria are displayed in different colours: a green dot represents a compound that is neither haemolytic nor toxic. A red dot indicates the presence of at least one of the non-desirable attributes. The three remaining criteria, serum –protein binding, solubility and metabolic stability, are also measured and optimised in each cycle, though these are not shown here.
The objective is therefore “green dots at the co-ordinates (0,1) with a minimum concentration in the thrombin-assay”.
In the second graph, the activities of the molecules proposed by the computer after the 1st modelling are shown. Noticeable improvement of the activities can be observed, though this would seem to be at the expense of an increase in the toxic and/or haemolytic effect. The data base derived from the results of the 0th cycle was too small (see left) to predict this attribute in a faithful way . The concentration in the thrombin assay was reduced to 50 µM.
The process can successfully tackle the problem of toxicity and/or haemolysis in the 2nd cycle (third graph). After another reduction of the concentration to 25 µM in the 3rd cycle, no convincing improvement of the activities is visible. On the other hand, the data are widely „spread “so that a clear differentiation of their activities is possible.
The concentration in the thrombin assay remained for this reason unchanged in the 4th cycle, and a significant improvement of the activities by simultaneous occurrence of toxicity/haemolysis can, in fact, be observed.
In the 5th cycle, the algorithm concentrates on the elimination of side-effects. It should be emphasized that the process was, of course, not instructed to search for or display the behaviour described here ; the sole task specification is at all stages the simultaneous optimisation of all criteria.
The concentration in the thrombin assay was reduced to 1 µM during the 5th and 6th cycle. Some of the compounds proposed in the 5th cycle show significant activity (at around 0.4), and major improvements in the activities can be observed in the 6th cycle. At this stage it becomes clear that two very different patterns are emerging, one with and the other without side-effects.
|
 |
Figure 5: Course of the computer-based optimisation process (CADDIS) for six cycles. In the upper line, the initial cycle and the first two optimisation cycles are shown (from left) and in the lower line the cycles 3-6 (from left). Thrombin inhibition is shown vertically. Notice that the concentration in this assay is adapted in each cycle! The concentrations in the thrombin assay are 150µM, 50µM, 25µM, 25µM, 10µM, 1µM and 1µM. The activity on trypsin is shown horizontally. In this assay the concentration is kept constant during all optimisation cycles.
|
| A number of promising compounds have now emerged from this project. The properties of the most interesting compounds in comparison with well-known compounds in this field are given in table 1. |
 |
Table 1: List of the most important properties of the compounds developed by
MATRIX in comparison with other active compounds in this field.
|
| Truly surprising - even for experts in the field - is the extremely high selectivity achieved by the process. Selectivity factors of 50,000 and above compared to other serine proteases (not all have been tested) are to date a complete novum in this area of research. |
 |
Figure 7: Elimination of several compounds developed by Matrix Goe-(6-4, 7-8, 8-1, 8-5) compared to some reference compounds (NAPAP and Melagatran) from the circulation after i.v. application of 1 mg/kg in the rat. Courtesy of Dr. J.St?rzebecher, Center for Vascular Biology and Medicine, Friedrich-Schiller-University, Jena, Erfurt.
|
| The elimination behaviour is shown in Fig. 7. As can be seen, the compounds developed by MATRIX are in a very interesting range. |
 |
| Figure 8: Elimination of some compounds developed by Matrix (6-4 and 7-8) compared to NAPAP and Melagatran by the bile after intravenous application of a dose of 1 mg/kg in the rat. Courtesy of Dr. J.St?rzebecher, Center for Vascular Biology and Medicine, Friedrich-Schiller-University, Jena, Erfurt. |
Even more exciting are the results with respect to the in-vivo metabolic stability of the compounds (Fig. 8). Most surprisingly, G?-7-8 is not eliminated by the bile – moreover 58% of the applied dosage was found after 300 min in the urine of the test animal.
Currently, final experiments are being undertaken to ensure the in-vivo oral availability of most of the compounds. X-ray structure analysis is also being conducted to uncover the nature of the binding.
|
|
