Professor Johnjoe McFadden and his team of scientists at the University of Surrey created the virtual TB bacillus in an effort to speed up the drug discovery process by identifying new protein targets for drugs and how some bacillus manage to fight off a drugs effect and significantly prolong the life of the microbe.
Prof. McFadden explained to DrugResearcher.com that these few 'persistent' cells keep the infection alive in a human for up to six months before succumbing to drugs while the vast majority of other microbes are killed within a few weeks. That delay can be enough to kill the patient.
However, no one knows exactly why these microbes are resistant and only last week the World Health Organisation warned that resistant forms of TB were becoming more widespread, sparking fears of a potential "untreatable epidemic".
A further problem is that although new drugs are clearly needed, drugs development is a slow process and perhaps more so with TB as the bacillus takes a long time to grow, meaning testing also takes longer.
The scientists decided to tackle both these problems through designing a tool that could perform experiments faster - within minutes instead of months - and try and find out what enzymes certain microbes use to fight of drugs.
"If we could find that out, it would revolutionise TB treatment," said Prof. McFadden. New treatments could be developed that destroy TB over a couple of months instead of a more typical six. If this were possible, millions of lives would be saved and it could even pave the way to eradicating the disease.
The first step towards building the model is to decide what to finally measure. The team settled on the flow of nutrients through the cell (such as glucose) to make biomass, as it is a good indicator of cell growth. Then, the scientists examine the entire genome of the pathogen to discover which genes are important to this process.
The function of each enzyme that is deemed important is then transformed into a set of linear equations to solve, explained Prof. McFadden. The scientists then use their expertise ("and a little guesswork," said Prof McFadden) to fill any gaps or dead-ends in the genome data.
The network of mathematical problems that is eventually created also includes some data on the structure of the microbe in order to more accurately model its growth. All that remains now is to 'train' the model using real experimental data.
To do this, the scientists pooled years of research from around the globe and supplemented it with their own experiments using TB bacillus grown in a chemostat. This type of apparatus is used to ensure variables such as nutrient concentrations, pH and cell density are all kept constant throughout - in a 'steady state' to match the computer model.
The team measured the glucose uptake both in vivo and in silico to ensure they matched - which they did. These results are then validated further by repeating gene mutations, known to stop the bacillus growing in vivo, in the model and checking the results correlate. Once Prof. McFadden had examined the full "catalogue of mutations against inactivating that gene in the model", the final validation stage can begin.
Here the enzyme targets of existing TB drugs such as isoniazid and rifampicin are blocked in the model and the growth of the bacillus predicted. Again, the results matched the in vivo data, and the scientists decided the first design of the model was complete. The results are published in the Genome Biology journal and the model itself is freely available online here for other scientists to use to their hearts content.
Prof. McFadden went on to say that he hopes scientists will use it to predict potential new drugs targets against TB and since the virtual bacillus 'grows' in nanoseconds, experiments can take anywhere between fractions of a second to a few minutes, compared to months for the real thing.
Not only is it quicker though, it also enables researchers to examine the effects of blocking combinations of more than one enzyme - in effect examining the whole pathway of a drug's effect.
All of the 848 different biochemical reactions using 726 genes in the model can interact with each other and this incredibly complex network is impossible to manipulate in traditional experiments, explained Prof. McFadden. Instead, the power of a computer is needed to make sense of the 'explosion' of interactions.
"Our model is a much more realistic way of looking into an organism's biology than simply looking at the effect of a single gene," he said.
The scientists are now in the process of building a human cell to work alongside the bacillus cell to enable scientists to examine drug targets that will kill the pathogen while leaving a human cell intact. They also plan to add more biological measures to the models, such as virulence, as well as growth.
One current limitation of the model is its lack of chemistry. It cannot predict how well a drug candidate will block a certain target. However, once this is measured - using different computational or experimental techniques, the data can then be fed into the model to predict growth - just as in the validation step.
Ultimately, they would also like to combine the biology and chemistry aspects to develop a model that can predict how well a drug will work. At the very least, scientists from around the globe can now access a wealth of information about this deadly disease and hopefully design drugs that will consign it to a place in history.