Biol 591 
Introduction to Bioinformatics
Fall 2002 

Scenario 4: Metabolic modeling and the discovery of new drug targets
Our Story

You are planning strategy at a major pharmaceutical company, and you've turned your attention to Trypanosoma bruceii, the causative agent of sleeping sickness. This is a tough problem, because since T. bruceii is eukaryotic, it is immune to most of the antibiotics in our arsenal. Furthermore, the parasite is remarkably resourceful in evading the immune system. What treatments there are have side effects so serious that hospitalization is generally required only manage them (never mind the effects of the disease itself). You're looking for something better.
Trypanosoma bruceii amidst blood cells.
One cause for hope is that T. bruceii is totally dependent upon glycolysis, the breakdown of simple sugars, for the energy it needs to grow. Conceivably, a drug that selectively blocks glycolysis might slow down growth of T. bruceii with fewer side effects than current treatments. Unfortunately, no such drug currently exists, and it's difficult to know how to mount a search for one.

Your idea is to initiate the search through a computer model of glycolysis, using the model to assess which of the enzymes of the pathway might be the most effective target for pharmaceutical intervention. You hope that such a model might tell you before you lift a test tube, what level of inhibition of which enzymes is necessary to achieve what level of inhibition of growth by the parasite. If you can identify a single enzyme as the most sensitive, then the chemistry types can step in and try to devise a drug that might act on that enzyme.

So, how to model glycolysis? We want to end up with a set of equations that predicts the level of ATP (correlated with growth rate) under various circumstances, but there is no such equation. What experimental enzymology can tell is not an analytical expression for metabolite levels but rather the rates at which metabolites change, since rate constants are intrinsic properties of enzymes. You thus need a method to take experimentally determined rate constants for all the enzymes of glycolysis and use them to predict the concentration of different metabolites (particularly ATP) under different regimes. For example, you want to know how that level changes upon the addition of mythical inhibitors of each of the enzymes in the pathway.

Enzymatic rates are expressed as differntial equations, for example:

d[X]/dt = [A][B]k1 - [C]k2
i.e., the rate of increase of some metabolite X equals the sum of two processes: the production of X, proportional to the concentrations of A and B; and the consumption of X, proportional to the concentration of C.

Problem
How can we proceed from a set of such differential equations to a model that predicts ATP concentration?
How can we modify that model to permit testing the effects of hypothetical drugs?

Tools
Metabolic modeling
We'll use a homegrown program to illustrate how models are made and modified.