Sifting Through Garbage

A Seattle-area Biotech Company Confronts Heart Disease

By Rok Roškar

Tuesday is garbage day in my part of Seattle. The cacophonous early morning symphony of diesel and shattered glass crawls at a snail's pace down my block and onwards as I struggle to unglue my eyelids. A fraction of almost everything consumers use on a daily basis eventually ends up in the garbage. Garbage trucks are an urban necessity, relentlessly making their rounds, hiding the byproducts of consumer culture in their voluminous metal bellies. An interested party could, in principle, dig through the contents of the garbage truck shortly after it passes my block and learn an immense amount about the lifestyles of myself and my neighbors: how lean is the meat we eat (do we eat meat?), how virgin is our olive oil, do we subscribe to leftist or rightist propaganda, which are our hobbies? The interested party could begin to understand our strengths and weaknesses.

You would be hard-pressed to find someone interested enough in my life to dig through the dumpster in order to learn about me. But, our existence provides byproducts on all scales, and garbage of particular interest can be found in the bloodstream. The byproducts in our bodies of course are not recycled magazines you never subscribed to, but may instead take the form of proteins of various shapes and sizes.

The presence of particular combinations of proteins may indicate that important changes are underway in your veins and arteries, possibly even signaling the onset of heart disease. In your blood, some proteins are picked up by the High Density Lipoprotein (HDL) molecule, which parades through the cardiovascular maze and, in the words of Erik Nilsson, "basically does a garbage run,” picking up bits of cholesterol here and proteins there. It is a garbage truck in our blood. Carefully sifting through the random scraps HDL carries along on its rounds could expose potentially fatal weaknesses of our cardiovascular system.

Nilsson is the president of Insilicos, a small biomarker discovery and diagnostics company based in Seattle. This morning, the Insilicos office space looks subdued and expressionless, much like the rest of the office-building strip of Seattle's Eastlake neighborhood where the small company resides. Nilsson, however, is anything but. His entrepreneurial enthusiasm cuts through the drone of traffic and rain outside and the humming of fluorescent lighting inside.

Nilsson and I are joined by Bryan Prazen, Insilicos chief scientist. Prazen holds a Ph.D. in analytical chemistry and is an expert in pattern recognition techniques. The idea for their enterprise hatched in part while "rafting on rickety bamboo rafts” in Thailand, where Prazen was conducting research at Chiang Mai University almost ten years ago. With the backdrop of lush greenery and shimmering waters of Thailand, they began to formulate the concept behind Insilicos.

"[Being in Thailand] gave us a chance to talk. At that point, nobody was really using classical pattern recognition applied to proteomics. It just seemed so obvious to both of us,” synthesizes Nilsson.

"Proteomics is the study of proteins at a system level,” explains Prazen. Proteins are organic compounds used and created by cells at all levels of cell function from metabolism to immune system response. A proteome is a group of proteins produced by a cell at a particular time for a particular purpose. Rather than classifying each protein individually as a separate entity, proteomics is an attempt to describe how the biological system as a whole produces protein signatures.

The tangle of proteins can be dissected and classified by mass spectrometry, which separates compounds present in a sample based on their physical characteristics. Traditional mass spectrometry methods, however, can only report on the relative amounts of species that are specified and known ahead of time. Necessarily, a large part of the signal is discarded because it contains species not previously identified. The pattern recognition approach of Insilicos is to extract information from the entire spectrum, quantifying relationships between peaks and valleys, regardless of what is known about each of them specifically. It's like sitting back and absorbing the full sound of an orchestra rather than focusing on the rhythmic growl of the tuba while ignoring the playful melodies of the flutes. They needed a testing ground for these ideas – a system small enough to be well-constrained, but "important enough to grab attention,” says Nilsson.

Nilsson and Prazen discussed their concept with Tomas Vaisar, at the time a research scientist in the laboratory of Jay Heinecke, who studies metabolism and nutrition at the University of Washington. Vaisar and Heinecke were just then investigating the proteins acquired by HDL, our trusty cardiovascular garbage truck. They had developed reliable procedures of extracting HDL from the bloodstream, but needed good analysis methods to proceed further.

HDL is dispatched from the liver and is known mainly for its function as a bulk cholesterol collector. However, Vaisar and Heinecke discovered that HDL also collects a myriad of other scraps. In particular, HDL is known to interact with cells inside atherosclerotic lesions, which are essentially fatty buildup on arterial walls. When these lesions rupture, the fatty material is released into the bloodstream and carried downstream like forest deadfall down a river during a flood. However, arteries, unlike rivers, cannot overflow their standard channels. When the fatty material encounters a constriction, blood circulation is inhibited potentially causing a collapse of the cardiovascular system, triggering a heart attack or a stroke. HDL therefore takes part directly in the progression of heart disease, carrying potential signals of the dormant beast from affected cells through the bloodstream.

Nilsson and Prazen found the perfect testing ground for their ideas. They would search for indications of heart disease by using pattern recognition methods to analyze protein content recovered from HDL of patients with and without heart disease. "[Pattern recognition] can use features that individually would not be very interesting, but together they can provide you with much more information,” explains Heinecke. Rather than trying to identify and isolate individual parts of the spectrum, their approach characterizes the features and their relation to one another. This information can then be used to contrast two groups of subjects, potentially identifying signals of disease.

However, Heinecke also warns about the downside of such a holistic approach to proteomics: identifying the source of an important signal may not always be possible. The inherent danger is that an experimental oversight would cause a systematic difference between samples that could then falsely be interpreted as a sign of disease. Nevertheless, preliminary tests on small groups of subjects reveal remarkable diagnostic success of their method.

Success with a high-profile problem like diagnosis of heart disease is a major asset to Insilicos, whose full-time roster of five appears more like a small software start-up rather than a biotech analysis provider. Nilsson himself has only dabbled in the biotech sphere since he sold his software company some ten years ago. Regardless, Insilicos has been very successful in fostering strategic relationships with important partners in academia. They have routinely secured a healthy number of research grants during the six years of operations, a testament to the strong scientific base of their activities. Just recently, their collaboration with Vaisar was awarded the Washington Technology Center grant for Research and Development. This has been a crucial step in the growth of Insilicos. "The [first phase of the] WTC grant was really the first that allowed us to focus on this problem of pattern recognition in proteomics.”

Nilsson's enthusiasm for his newly-adopted niche makes it clear that the Insilicos vision is not ending here. "I would love to do something about neurodegenerative disease, I would love to do something about cancer. But, you need to solve the problem somewhere so that you can solve it somewhere else.” Cardiovascular disease seems as important a problem as any. He admits this was partly good luck: "half of life is showing up. We picked where we wanted to show up, we showed up, and something interesting happened.” Not caring much about luck, the garbage truck keeps rolling, but the wealth of information held within may bring a healthier tomorrow.

Rok Roškar is pursuing a Ph.D. in astronomy at the University of Washington.