Table 5.2 shows the number of readings necessary in each case for the specified degree of accuracy Hence the optimisation experiments in the next chapter

are repeated 6 times.

5.5

Discussion

Some of the problems associated with operating a robotic system in the context of enzyme reaction optimisation have been investigated here.

Most of the commercially available robotic liquid handlers are designed to operated with liquids only or a low density solid emulsion (such as a cell suspension). However, many enzymatic systems are operated with the enzyme immobilised on a support and the support may change some of the properties of the enzyme, therefore it is necessary to find a method to dispense accurate quantities of bead slurry in the wells of a plate. The main problems identified in dispensing bead slurry by the pipetting methods investigated are:

■ The slurry which the beads are pipetted from must be homogeneous so that

equal amounts of beads are picked up each time. The Eupergit C beads tested however, were dense and so quickly settle in solution and so the vessel holding them had to be constantly agitated to maintain a homogeneous state.

■ The beads may clog the tips both when drawing the slurry up into the pipette tip and when dispensing, resulting in the slurry being effectively filtered, so that more liquid than solid is drawn or dispensed.

Until the problems of robotic bead handling are addressed, the usefulness of robotic, systems when dealing with immobilised enzymes vsdll be limited.

The accuracy of the robotic system is not one that had been discussed in this context before. Previous examinations of robotic systems have been in their use for large scale screening, where large numbers of candidates are assayed for a particular property. This differs significantly from the use of robotic systems described here and in the subsequent chapter. Instead of looking at a very large number of conditions and assaying for a simple positive or negative response, a smaller number of conditions

is permitted. However, with the quantitative analysis, the accuracy of the assay system must be much more sensitive, as an inaccurate response could skew the data and give a false impression of the system. With the screening example, only one repeat of each condition is necessary, but with the quantitative example it is necessary to determine the number of repeats which are necessary of the required degree of accuracy. Therefore, the accuracy of the system is determined in section 5.5. Figure 5.2 shows the relationship between the accuracy of a system and the number of repeats necessary. The degrees of freedom and therefore the t value is maintained constant and the standard deviation in this fictional system is varied. The different confidence intervals (L) chosen show that if either the standard deviation increases or the accuracy which is chosen (L) increases the number of repeats necessary has to be increased.

For the real system used in this example the number of repeats necessary were six. It is important to note that, while it is possible to repeat each experiment more than six times, to do so would make those extra repeats redundant. The redundant repeats would not increase the accuracy of the final result nor would they add to the information already gathered with the first six repeats. Redundancy is an important phenomenon in this context as it is tempting to do many repeats of each experiment, just because it is possible to do so. Without first investigating the accuracy of the

system it is impossible to ascertain how many repeats are really necessary.

Chapter 5: Robotics L = 0.1% 1000000 z i2

1

Standard deviation (a)

Figure 5.2:

Relationship between number of repeats (N) and standard deviation (o) for different lengths of confidence intervals (L). This graph represents the equation to calculate the number of repeats necessary for an experiment:

The degrees of freedom and therefore the value of t (Student’s t-test) remains constant.

6 Optimisation of conditions for a second enzyme

cofactor regeneration reaction

6.1 Summary

A combined, two enzyme system has been optimised using a semi-automated, high throughput screening method. The synthetic enzyme, tetralone reductase, reduces tetralone to tetralol, while a second regenerative enzyme, formate dehydrogenase, converts formate to CO2 and recycles NADH. The optimum condition of pH (6.8),

temperature (26®C) and solvent (30% octanol) were found by factorial experiments using a robotic liquid handling system to produce a semi-automated process.

6.2

Introduction

As described in the previous chapter, time to market for new pharmaceuticals is a priority. Therefore any methods which can speed the progression of process synthesis will be welcome. This chapter demonstrates the use of a semi-automated robotic system for the optimisation of the combined synthetic (tetralone reductase) and regenerative (formate dehydrogenase) reactions. To further streamline the process and to reduce the number o f experiments necessary a factorial approach was taken to the optimisation.

Factorial designs are schemes in which the effects of several factors are studied simultaneously. In the analysis of the factorial experiment the effect of each factor on the measured value is considered and also the interaction between the factors can be examined. The outcome is a model which should predict the behaviour of the measured value under any conditions of the factors in the range which they are tested.

The oxidation by formate dehydrogenase was chosen as the NADH regenerative

method. This has previously been used successfully by other workers for the

regeneration of NADH (Chenault et al, 1988). The major advantage of this enzyme is

that the product of the reaction is CO2 when the substrate formate is used. This CO2

gas will then leave the reaction liquid or can be driven o ff by gas sparging, resulting in a cleaner system for subsequent downstream processing. Despite the fact that the