While my graduate work was rule-based modeling of petrochemical-related chemistries, I felt that following graduation I was more interested in applying these computational approaches to biological problems instead of petrochemistry. After a brief post-doc at Sandia National Laboratory in Livermore, I landed a position as a Biosystems Engineer at Entelos, Inc in Menlo Park, California during the peak of the dot-com bubble. Entelos was an interesting company. In retrospect, it was a pioneer in the field of quantitative systems pharmacology. As an engineer who knew how to develop mathematical models, I joined Ph.D. life scientists, who knew biology but not the math, on a truly interdisciplinary team to develop mathematical models of disease. In short, these were flight simulators for new drugs. Using the company’s own proprietary modeling platform, our team of engineers and life scientists dove into the scientific literature to find data that could be used to justify the mathematical relationships that connect a molecular target with a clinical read-out. These models were called PhysioLabs. The initial business model of the company was to develop these PhysioLabs and sell this software product to pharmaceutical companies. In short, that didn’t work.
My initial assignment was with the Asthma PhysioLab team. The Asthma PhysioLab focused on allergic asthma and simulated the how the exposure to an allergen results in a bi-phasic reduction in lung function as measured clinically by the forced expiratory volume in 1 second (FEV1). It included the intermediate steps of IgE-mediated degranulation of mast cells, histamine release, histamine and cytokine action on smooth muscle contraction, histamine-induced release of mucus in the airways, and how the combination of smooth muscle contraction, edema within the airways, and mucus release within the airways reduces the available space for airflow. This reduction in airflow then impacts the FEV1. We also modeled how existing drugs, like an anti-histamine, altered the simulated behavior. In presenting an overview of the Asthma PhysioLab to Dr. Jeffrey Drazen at the AAAAI meeting in San Diego in 2000, he said that the Asthma PhysioLab was like the best review paper ever written on atopic asthma as it visually summarized the biology but also simulated how all of the different components interact in time.
In making these models, we initially made assumptions that limited the biology that we needed to include in the model. For instance, allergic asthma is associated with a type 2 immune response. Our initial model assumed that the type of immune response was fixed, which limited the type of potential drugs that we could simulate. At the time, there was interest in targeting the type 2 cytokine Interleukin-4, which could skew the immune response away from a type 2 response. To meet this interest, we needed to expand the biological scope of the model.
Towards that aim, I focused on how dendritic cells integrate cytokine signals present in one anatomical location, like the lung, and deliver different signals when the migrate to the secondary lymphoid organs. Besides reading 100’s of papers on a particular topic, we also engaged leading scientists to give us a lay of the land, so to speak. Dr. Patrick Holt was an expert that we brought in to provide perspective on adaptive immune response in the lungs and on dendritic cells. One of the aspects of dendritic cell biology that became important was how dendritic cells integrated signals to produce the cytokine Interleukin-12. At the time, the literature surrounding Interleukin-12 was a bit of a mess.
Our understanding of biology changes with time. In part this evolution in knowledge can be driven by the realization that the tools used to measure biological components are flawed. The cancer immunology story is a good example where early experiments using animal models that supposedly were lacking a host immune response had to be reinterpreted upon realizing that these animal models were not as advertised. In the case of Interleukin-12, early studies used an antibody that measured one component (p40) of a two-component protein (p40 + p35). Later studies revealed that the measured component (p40) could be produced by dendritic cells in different forms that have different biological functions. In making the revised model, I modeled how signals present in the lung influenced how dendritic cells produce these different forms containing p40 and how these forms interacted with the Interleukin-12 receptor that is expressed on T and NK cells. This work is summarized in two modeling papers that I published after my arrival at WVU [1,2].
A bit later, the company was having challenges to grow. In an all-hands meeting, the CEO stated that we have a credibility problem when making sales calls to the pharmaceutical companies. Essentially every couple of years, pharmaceutical companies merge and there are new faces in the disease areas that are targeted by the company’s PhysioLabs. While Entelos had established their credibility with the previous scientific team at the pharma company, the new faces were skeptics. That is not surprising, as scientists are accustomed to digesting the written word and data presented to support an argument. Entelos and it’s sales team, though, were deeply embedded in a consulting culture, where panache and a yes-we-can attitude are used to support a stance instead of hard data. At the time, the company had no primary research papers that laid out it’s approach. The leaders at Entelos were resistant to publishing as they were afraid to reveal their “secret sauce.” A common refrain within Entelos was that scientists typically don’t acquire the right data – such as time-course, multiplex, and quantitative experimental designs. Though as non-participants in scientific discourse, this refrain became a bit hypocritical as without publishing what kind of data is needed, the current state of biological investigation won’t change. The CEO said then that we need to publish peer-reviewed primary publications to help close this credibility gap.
With those inspirational words, I thought what of my work could be wrapped up as an independent story to illustrate how we use scientific data. A story related to Interleukin-12 and it’s different constitutive components fit the bill and I drafted the paper. After getting feedback from some of the team leaders, the manuscript wound its’ way to the CSO, a newer arrival who’s hire was announced by much fanfare. At a one-on-one meeting to discuss the manuscript, the CSO said that the story seemed complicated and wondered if I could summarize it for him. In short I said that inferring biological activity associated with Interleukin-12 required three pieces of information. Most experimentalist measure only one or two pieces of information and thus can’t interpret their results. I think I surprised him with my succinct answer. He responded after a bit and said “Hmmm, well I’ve never really understood immunology but I think we shouldn’t publish this paper. It’s too simple of a model and that’s not really the message that we want to convey.” He then said that I was free to publish it just unaffiliated with Entelos. Yeah, me working alone at night in my garage doing science. This interaction pretty much cemented my view that it was time to move on. Ultimately the paper was accepted and I held up publishing it until I had officially started at WVU. While the contact email was still my private email, I could list my new affiliation. Upwards and onwards.
References:
- Klinke DJ 2nd, An age-structured model of dendritic cell trafficking in the lung, Am J Physiol Lung Cell Mol Physiol 2006; 291:L1038-49. PMID 17030902
- Klinke DJ 2nd, A multi-scale model of dendritic cell education and trafficking in the lung: implications for T cell polarization, Ann Biomed Eng 2007; 35:937-55. PMID 17457675
DK Design Blog 