Diseases caused by fungi are a real risk for people burdened with weak immune systems, after organ transplants, or long chemotherapy treatments, for example. The fungi can do all kinds of damage, from causing pneumonia in the lungs to attacking the brain with vicious types of meningitis, or trigger life-threatening infections. Grid computing is helping with the creation of powerful new antibiotics against fungi, but with fewer side effects.
The antibiotic Amphotericin B – abbreviated as AmB – has been the drug of choice to fight fungal infections for the past 50 years. It’s brutally efficient, killing a broad spectrum of fungal agents and active against all known multi-drug resistant strains.
The catch is that AmB is toxic to the human body and it can cause organ damage in patients, especially the kidneys.
The challenge is to develop an upgraded version of AmB, with all the original’s efficiency, but fewer side effects.
Anna Neumann has been working on this problem for her PhD at the University of Technology in Gdansk, Poland. “We know how AmB works on a cellular level – it acts on a cell membrane, and forms some kind of permeable structures, most probably channels, in it,” Neumann said. The channels built by AmB allow the cell contents to leak out eventually leading to the cell’s death.
The problem is that AmB is not very discerning and attacks the human cells together with the infectious fungi cells it’s supposed to kill. There is not much difference between the way this antibiotic interacts with fungal and human cell membranes.
The keys to solving this problem are the compounds known as sterols – types of organic alcohol that are gatekeepers of cell membranes. Sterols control the physical and chemical properties of the membrane; for example, how permeable they are. The sterol in fungal cells is called ergosterol; mammals have a different type, called cholesterol.
AmB has a slight preference to attach itself to membranes containing ergosterol (hence killing the fungal cells), but this affinity is not strong and it explains why the antibiotic also attacks human cells: it sometimes can’t tell the difference between them.
Learning more about how AmB connects to the two types of cells and their sterols, how the antibiotic enters the cell membrane, and how the channels are formed, will help create safer AmB varieties. Neumann analyzed the problem with molecular dynamics simulations – computer models designed to mimic the physical movements of atoms and molecules.
Molecular dynamics models are useful for describing the behavior of atoms and molecules, and their interactions, but are also very demanding on computing power. Neumann accessed the computing resources provided by the Polish National Grid Initiative (PL-Grid) to process the molecular dynamic simulations. She used 24 computing cores for each grid job that was submitted, adding up to a total of five million CPU hours.
Neumann consumed almost 10% of the whole usage of the PL-Grid infrastructure according to Zofia Mosurska from PL-Grid. And, for a long time she was one of three PL-Grid users that held the record for highest usage every month.
According to the results, published in the Journal of the American Chemical Society,the difference in AmB’s affinity for ergosterols and cholesterols is partially due to energy levels. It’s easier, in terms of energy, for AmB to interact with the rigid and elongated molecular geometry of ergosterol than with the cholesterol. In other words, AmB needs more energy to combine with human cells than with fungal cells and it is usually the lower energy option that wins.
These conclusions, together with further analysis, will allow Neumann to propose a way to make the AmB molecule more likely to attach itself to fungal cells. Neumann said, “That would affect AmB's activity – making it more selective for fungal cells and hence less toxic.”
Since the paper was published in 2010, preliminary results look promising. Neumann has been working with a number of colleagues to propose some general ideas, which will then be verified by experimentalists. A number of simulations are currently being analyzed.
A version of this story first appeared on the EGI website.