Traversing Dimensional Boundaries – Behind the Screen to On the Shelf

---

BY DEBORAH KOMLOS

Characterizing compounds, discovering the mechanisms of molecular interactions and testing cell responses to stimuli are bound to become virtual realities — ones with very tangible results.

A handful of efforts worldwide are aiming to create a “virtual cell” — a computer simulation of a living cell that can be applied in R&D for a variety of tasks, such as assessing protein-protein and protein-drug interplay. Ultimately, having an in silico mimic of reality will save both funding and time as researchers will not need to screen as many potential drug targets, and can thereby focus in much sooner on those with most promise.

Among these global initiatives is Project CyberCell™, an endeavour being conducted by the Institute for Biomolecular Design (IBD) at the University of Alberta (Edmonton, AB), to computationally recreate the bacterium, Escherichia coli. Founded in 1998 by Michael Ellison, PhD, IBD was established with an investment of nearly $26-million from the Canada Foundation for Innovation. Intent of the initial setup, Ellison explains, was to systematically investigate the fundamental chemical rules for understanding how proteins interact, either with themselves or with ligands, such as drugs. Eventually, he says, the field of proteomics began to emerge. Thus, IBD — with expertise in the realms of protein structure and function — was nicely poised to begin massive data collection. But the key was deciding what to do with that information influx.

“When you really think about it in the big terms, what all this data will ultimately be used for is modelling a cell in silico. And once you’ve made that major shift in mindset, it really predetermines what kinds of data you want to collect, and how you’re going to go about collecting it,” says Ellison, executive director of IBD. The institute has 15 individuals involved in the project, distributed among the research areas of discovery, bioinformatics, biometrics, simulation, and structure and function.

Building Upon Basics

IBD made Project CyberCell its main scientific focus, Ellison says, with the aim of collecting data that would be needed to not only drive cell simulation, but also to validate it. The latter is important, he says, because only by testing a model can one determine how good it is. Selecting E. coli as the simulation target was not a difficult decision, Ellison adds, considering it is the simplest organism about which the most is known and it is experimentally amenable.

“I can’t reasonably imagine trying to take a cell as complicated as a mammalian cell, or even as complicated as Saccharomyces

cerevisiae, and starting there,” Ellison says. “The idea is that we can kind of cut our teeth on developing key concepts, and getting the mistakes out of the system and what we learn on E. coli could then be ultimately applied to more sophisticated organisms.”

The approach taken by Project CyberCell makes the endeavour distinctive, Ellison says. Modelling efforts that began anywhere from five to two years ago were the innovation of theoreticians, or even systems engineers, he says. As a result, those projects were created without much associated effort to collect the necessary data for model validation.

“We’ve really come into this game thinking as biochemists. For us, a cell is not a collection of pipes, flows and fluxes; it’s a collection of discrete molecules that interact and have their activities influenced by other things,” Ellison says. “As biochemists we understand really the key importance of high-quality, precise measurements that are necessary to be able to really drive and validate these simulations.”

Achieving the end result — creating a virtual E. coli model — may seem like a mammoth goal, Ellison says, but a global effort is geared toward facilitating and accelerating the ambitious undertaking.

Unified Aim, Unique Avenues

The International E. coli Alliance (IECA) was founded last summer by Ellison and the leaders of three virtual-cell teams: Igor Goryanin, PhD (E. coli Whole Cell Model, GlaxoSmithKline, U.K.), Masaru Tomita, PhD (E-Cell, Keio University, Japan), and Barry L. Wanner, PhD (E. coli Model Cell Consortium, Purdue University, Ind.)

“Rather than duplicating efforts, basically we’ve carved up key pieces of scientific territory. We’re actually working in a very complementary way,” Ellison says of the alliance. For instance, while IBD is centred on the quantitative proteomics aspect, the Japanese are taking over much of the functional genomics.

Ellison says the groups have agreed that each will pursue its own modelling, something he praises because a certain level of competition is healthy and modelling is a very idea-intensive task. “Consequently, it’s very important to get as many ideas onto the table as possible, and the only way that can really be done is if we kind of pursue what we think is best and we test it according to the rigours of scientific peer review.”

As the alliance’s other North American partner, the E. coli Model Cell Consortium led by Wanner began informally four years ago, with colleagues discussing possibilities of generating a virtual-cell model. Currently with five members on its steering committee, the consortium had its first official meeting in December 2001, and has since held ongoing gatherings involving American and international attendees, including Ellison.

With secured temporary space at Purdue, Wanner’s group aims to establish a microbial growth facility, from which the other IECA teams can be served. Not only is it critical that researchers use the same strain of E. coli, Wanner explains, but that samples are taken from the same culture. In light of recent bioterrorism fears, Wanner predicts that serving Canada may not be as difficult as shipping samples abroad. Nevertheless, he says, having a central facility dispense the materials will help establish a baseline experimental protocol among the groups, something necessary to ensure repeatability of results.

Also expected is a state-of-the-art metabolic profiling facility to be situated adjacent to the growth facility. A proposal of $1.8 million in equipment funding has been submitted for this facility, which will provide a site for high-accuracy measurement of small molecules such as intermediates in metabolism. Both new facilities will ultimately be housed at a new bioscience research centre at Purdue’s University Park, to be constructed by April 2005.

Like Ellison, Wanner sees the merit in dividing up the enormous task at hand. “We’re not going to be able to solve the whole cell model in one thick stroke. There are different subsystems,” he says. In this regard, he continues, taking advantage of each institution’s respective strengths — at Purdue, this includes excellence in analytical chemistry — will greatly propel the project.

A Computational Quest

Given the concerted effort and factoring in the anticipated advances in computing capacity, Ellison says it’s not unreasonable to predict completion of the virtual cell within about 10 years. A close relationship with IBM, which is currently building a supercomputer called Blue Gene, will enable this effort. Expected within three years and intended to be the world’s fastest supercomputer, Blue Gene will allow IBD to get a very good sense of the magnitude of the project and how long it will take to obtain something predictively useful, Ellison says. “I’m cautiously optimistic that once the Blue Gene trials are done, I think it could be a matter of three to four years beyond that before this becomes kind of a routine scientific tool used by big pharma and individual researchers alike.”

Wanner finds himself with a more moderated outlook. “Within 10 years, if we’re well funded, we hope to know all the parts and begin to assemble the pieces. To say that we’re going to have a virtual cell in 10 years I think is dreaming. We’re scientists and we’re realistic as to what we can do.”

Giving the analogy of curing cancer, Wanner points out that tremendous progress has been made over the last several decades on tackling this disease. “With better diagnostics now, there are some cures,” he says. “But it’s still a problem.”

So, within about a decade, “We’ll be a lot closer, and the only way that we’re really going to get closer is by having groups that actually work together, collaboratively,” he says, because the feat requires expertise in a range of areas — in particular, biology, chemistry, physics and mathematics.

Model Moulding

In pursuing its virtual-cell vision, Project CyberCell has necessarily required appropriate visualization software, and among IBD’s partners, Calgary, Alta.-based Computer Modelling Group Ltd. (CMG), is fulfilling this role. For over 20 years, CMG has served the oil and gas industry, and about two and a half years ago widened its simulation scope to include the biotech sector, says CMG chairman, Frank Meyer.

Admittedly, the firm is still at a “crawl stage,” Meyer says, continuously working to develop and improve its software. However, with the backing of a very mature, robust technology, CMG stands confident in helping the virtual-cell project materialize, he says.

Past biomedical modelling efforts, Meyer explains, focused on process calculations such as measuring the amount of liquid moving into or out of a tissue. CMG’s technology introduces a unique approach, he says, which models flow and bio-chem reactions in and through volumes of porous media. Thus, aside from the conventional 3-D options (measuring height, width, depth), the dimension of time is added, which then produces a 4-D model.

“Part of our software is to display on the computer screen or on a projector screen the result of visualization in 3-D,” Meyer says. “You can put on some stereoscopic glasses so you can actually have a sense of being in the cell or being in the tissue.” An example he gives is being able to virtually slice open the modelled entity and watch how temperature changes through that volume over time.

Meyer emphasizes that while CMG’s technology provides a novel research tool, it is only as good as those who use it. “You have to have people who understand it and know how to use it, and how to apply it and how to take advantage of it,” he says.

Within two years, Meyer says his firm will have developed software that is progressively more useful than the current versions. Meanwhile, everything is a work in progress, he adds, because users can take what is available now and make use of its existing capabilities.

Lakshmi Kotra, PhD likewise advocates the value of stereoscopic imaging. As director of the Molecular Design and Information Technology Center (MDIT) at the University of Toronto’s Leslie Dan Faculty of Pharmacy, Kotra finds himself extremely enthusiastic regarding the $7.3-million supercomputing and visualization centre. Opened in late January of this year, MDIT’s facility bears a stereo monitor nearly four metres wide and uses advanced molecular graphics software tools supplied by Tripos Inc. This resource will expedite drug-discovery research, Kotra says, by encouraging new perspectives on biomolecular structures and interactions.

Initiatives such as Project CyberCell will certainly expand the breadth of knowledge on cell function, Kotra says. However, there is a difference in philosophical thinking between MDIT and such virtual-cell endeavours. “If I understand four proteins in the cell today and if I can make a molecule, which is a small molecule in the lab, tomorrow, and test it on the cells the day after tomorrow, that is now reaching my goal faster. Of course after 10 years, if I have a virtual model that can tell me that, I would do that,” he says. It is facilities like MDIT that Kotra says will therefore accelerate the advancement of a drug to market as fast as possible.

“Here I think we are more closely tied to human health and outcomes tomorrow, outcomes we want to see today,” Kotra says. “There’s a lot to be done, a lot to be understood and when it comes to human health, or for that matter, any health-related problem, it’s yesterday we’d need the answers.”

Enhancing the Search

Ellison acknowledges that virtual models will never replace experimentation, but they have the potential to greatly broaden the drug-discovery pipeline by providing a means to screen candidate drugs in silico for their desired targeted effects as well as for potential side-effects on cell physiology.

The result, he says, will be cheaper science, by minimizing lab mistakes and being pointed in the right direction for conducting the right experiments. Strides will also come in conceptual gain, he says.

“You can imagine moving over entire ensembles of genes and actually being able to see how well they work in concert in the new environment, and you have the latitude to be able to sit down and optimize performance,” Ellison says.

The computer will serve as a prosthesis, he says, to help the human mind grasp an enormous level of complexity, which in E. coli consists of thousands of proteins, “bits and pieces that are running around talking to one another.”

With its simulation strategy, the IBD is “really playing the long game, the marathon, not a 100-metre run,” Ellison says. “We’re really planning for the day when the computing capacity will be sufficient to be able to deal with the model that we’re working on.”