June 1, 2017 (Vol. 37, No. 11)
Systems Biologists Are Attacking the Encrypted Messages That Would Allow Us to Predict and Change the Course of Disease
Cryptanalysis, the practice of learning how to read coded messages—and doing so without having advance knowledge of the cryptographic key—occupies not just the military services and the intelligence agencies, but the life sciences, too. Soldiers, spies, and biologists alike intercept signals, uncover hidden patterns, and analyze information systems. For biologists, however, obtaining data isn’t simply a matter of tapping phone lines or hacking servers. Instead, biologists have to “bug” living networks, the intricately connected cell-signaling pathways that usually work on our behalf, sustaining health, but sometimes betray us to the enemy—disease.
Subcellular double agents and moles may be most readily detected by biologists in the interdisciplinary field of systems biology, which tends to emphasize a holistic rather than a reductive approach to research. Using computational tools, systems biologists model how complex interactions can generate emergent properties.
Prominent systems biologists recently gathered at the Future of Health, the Institute for Systems Biology’s 16th Annual International Symposium, an event that highlighted the connection between wellness and health informatics. Discussions tackled health issues at the global level (malaria and tuberculosis) as well the individual level (disruptions of the immune system and the microbiome).
Overall, the Future of Health emphasized the importance of biological cryptanalysis. This form of cryptanalysis, like any other, is an informatically intense effort to recognize nefarious developments while they are still subject to timely preventives.
Getting to Zero Malaria
Despite major progress in the last 15 years, billions of people are still at risk of contracting malaria. According to the World Health Organization, in 2015 there were 214 million global cases of malaria and 438,000 malaria deaths (most of whom were children under five years of age). Although vector-control strategies such as bed nets can reduce the spread of malaria, drugs remain the crucial weapons for preventing it and reducing transmission. However, multidrug-resistant Plasmodium strains (the causative parasite) are now common.
When Trevor Mundel, Ph.D., president of global health at the Bill & Melinda Gates Foundation, delivered the keynote address at the Future of Health, he emphasized that systems biology strategies are needed to end malaria: “Systems biology, with its focus on advanced informatics and mathematical modeling, is one of the most promising approaches for rationally combining old and new tools. Defeating malaria demands that we understand the complex biological system that perpetuates cycles of infection and reinfection, and systems biology can produce realistic simulations that lead to smarter, customized sets of interventions in the field.”
Dr. Mundel explained that beneficial analyses could be derived from systematic analyses of various kinds. “A combination of surveillance, GPS mapping, cell phone data records, and remote sensing is helping us integrate the data coming from individual field studies by giving us insight into human mobility patterns and localized variations in climate and geography,” he detailed. “We’re even learning more about the biting behaviors of mosquitoes.”
Dr. Mundel remains optimistic that the battle against malaria will be won eventually. “The lessons we learn from applying systems biology to malaria are also strengthening the fight against other global diseases such as HIV,” he concluded. “With this powerful approach, grounded in data and empowered by technology, I am confident we can devise fresh ways to rid our world of malaria and ensure all people have the opportunity to lead healthy, productive lives.”
Tuberculosis (TB), a disease that has afflicted humankind for thousands of years, continues to cause more adult deaths than any other single infectious disease. TB is responsible for more than 10 million new cases and 2 million deaths each year. “Adding about 2 billion people who may be latently infected, a reservoir of TB exists that will last for decades,” stated Sébastien Gagneux, Ph.D., associate professor of infection biology, Swiss Tropical and Public Health Institute.
“Coupled with resistance issues, TB may soon become nontreatable by current antibiotics,” he continued. “Our goal is to better understand the diversity of the bacteria themselves and examine resulting pathogenicity in order to develop improved treatments.”
Dr. Gagneux’ group recently challenged the traditional view that little variation exists among the strains of Mycobacterium tuberculosis complex (MTBC), the causative agent of TB. “We examined genomes around the world and found that the MTBC comprises seven distinct human-adapted lineages and two lineages adapted to animals,” he reported. “Additionally, while we found that lineage 4 was present on all inhabited continents, this lineage could be further separated into 10 sublineages that differ greatly in geographic distributions. This suggests that these sublineages can be separated into specialists and generalists with respect to the number of human populations they infect.”
Using a systems biology-inspired approach, Dr. Gagneux’ group combined large-scale single-nucleotide polymorphism typing with targeted whole-genome sequencing of a global collection of 3,366 lineage 4 clinical isolates from 100 countries. According to Dr. Gagneux, the group’s findings indicated that “whereas the majority of human T-cell epitopes were conserved in all sublineages, the proportion of variable epitopes was higher in the generalist sublineages, that is, those able to thrive in a broader range of human populations.”
The conclusion of Dr. Gagneux’ studies may have an impact on future therapeutics. “It is important to globally evaluate the genetics of disease-causing pathogens, especially TB,” he insisted. “There is much to be learned by examining large collections of clinical isolates. Our studies suggest that including variable or mutant TB antigens, as opposed to those dominantly conserved, could improve the design of candidate vaccines.”
Immunity’s Black Box
Few systems approach the complexity of our own immune system, which promotes a swift and powerful host defense by coordinating a dynamic, multiscale set of hierarchically organized molecular, cellular, and organismal components. “I believe the immune system represents the original problem to be addressed by systems biology,” declared Naeha Subramanian, Ph.D., assistant professor, Institute for Systems Biology. “Immune cells respond to infection and environmental cues through a variety of intracellular and extracellular receptors.”
“Ligation of these receptors,” Dr. Subramanian continued, “leads to activation of many signaling cascades eliciting such processes as protein binding, phosphorylation, degradation, and nuclear localization that can subsequently alter gene expression.” Dr. Subramanian then related this observation to her work, which focuses on deciphering the molecular mechanisms of innate immunity: “The goal of my laboratory is to utilize a systems biology approach to examine these layers of information and derive hypotheses from emerging biological signatures.”
According to Dr. Subramanian, small alterations in gene expression resulting from infection or homeostatic dysregulation may lead to significant pathological effects associated with autoimmunity and other immune diseases. “By examining how expression changes more globally, we can detect such alterations,” she explained. “We now know, for example, that regulatory elements in the genome participate in normal immune responses.
“However, when these networks are disrupted, as by underlying mutations or polymorphisms in proteins involved in these pathways, the resulting perturbations may initiate a snowball effect that drives disease. Thus, exposure to pathogens may predispose to expression changes associated with immune disorders.”
According to Dr. Subramanian, taking a balanced, unbiased approach to data gathering and interpretation is of paramount importance. “Decades of reductionist biological studies have catalogued enormous numbers of components ranging from genes and their products to intermediate metabolites,” she noted. “In the past, we thought of one protein in one pathway leading to one disease. However, that is not true. Rather, it is more like one protein into a black box of myriad interactions.
“We need a systems biology approach to understand how these systems interact,” she concluded. “We need to know how these systems may sustain health or give rise to pathological phenotypes.”
Forecasting Disease Outcomes
Many complex chronic diseases, such as Parkinson’s, lupus, and autism, are remarkably heterogeneous across individuals. This can make treatment challenging because caregivers cannot accurately determine the course the illness will take.
“Better analytical tools are needed to help predict the patient’s trajectory of their disease,” remarked Suchi Saria, Ph.D., assistant professor, Johns Hopkins University. “Often dense sets of measurements are already measured to track an individual’s disease over time. By examining canonical progression patterns, we could better stratify and predict a patient’s trajectory and use that to target therapeutic interventions.”
After noting that systems biology is critical for unveiling hidden structures and patterns of disease progression, Dr. Saria described her laboratory’s systems biology approach: “We develop probabilistic machine-learning techniques that can draw inferences from data that represents how a complex, real-world system evolves over time. An individual’s health evolving over time is an example of such a system.
“We combine the clinical and biological knowledge we have about a disease with techniques that flexibly learn new knowledge from population-scale data. Doing so helps us improve our ability to understand and treat diseases.”
One immediate application of these prediction techniques is improving recruitment of patients for clinical trials. “Current trials in lupus and scleroderma take all comers into trials,” Dr. Saria detailed. “However, drugs often target very specific complications. For example, a drug might be aimed at mechanisms leading to lung decline. Instead of including all patients, one can increase the power of the trial by predicting and including individuals likely to experience lung decline.”
Dr. Saria added that providing timely treatments is another example of applications: “Sepsis is a whole-body inflammatory response to infection and is one of the leading causes of death in the hospital. The caregiver must interpret a diverse array of markers such as heart rate, respiratory rate, blood counts, and serum measurements. Current scoring systems are inadequate. By analyzing diverse signals using probabilistic machine-learning algorithms, we have derived severity scores that trend up to identify when a patient is at high risk. Similar techniques allow us to predict an individual’s response to therapy.”
According to Dr. Saria, the accumulation of biomedical data creates pressure for a change of thinking. “We must evaluate many more factors,” she argued. “Instead of thinking of diseases in static profiles, we need to realize diseases are better characterized as dynamic processes that evolve over time. Creating better dynamical models of disease would provide a very practical way to characterize and forecast disease progression and improve therapeutic interventions.”
The human body, no less than the planet, may be thought of in ecological terms. That is, the human body is a biological world that includes communities of symbiotic, commensal, but sometimes pathogenic microbes. These tiny organisms not only interact with each other in the human habitat, they constitute a genomic whole—a microbiome—that stands alongside our genome. In a sense, the microbiome actually towers above the human genome. The genes in the microbiome outnumber genes in the human genome by about 100 to 1.
It appears that the “good” microbes are also our partners in health because they assist in nutrient processing, vitamin production, and immunity. “Our relationship with our symbiotic bacteria is important,” noted Martin J. Blaser, M.D., professor of medicine and microbiology at New York University School of Medicine and director of the Human Microbiome Program, “especially during our early years of life, when the adult microbiome has not yet developed.”
According to Dr. Blaser, perturbing early-life microbiota can affect metabolic and immunologic development: “The quandary is that antibiotics are vital for health care. It is difficult to imagine optimal health without an umbrella of antibiotics to use when needed. Nevertheless, the biologic cost of antibiotic use is often not sufficiently taken into account when practitioners make treatment decisions.”
Dr. Blaser asserted that it is becoming increasingly clear that the well-intended use of antibiotics as medicines or prophylactics can cause long-term harm: “Observational, clinical, and epidemiologic studies focused on young children suggest that antibiotic exposure is associated with increased risk for a variety of diseases including obesity, type 1 diabetes, type 2 diabetes, inflammatory bowel diseases, allergies, and asthma. Experimental models suggest these associations are not just correlative, but rather are causal.”
Dr. Blaser suggested that the use of systems biology approaches could help unravel the maze. “Complexity of data relating to the microbiome continues to grow,” he explained. “There are lots of different organisms with lots of very complex transcriptomes and metabolomes. Factor in interactions with the host transcriptome and other systems, it become obvious that this will require very sophisticated methods to extract information from all these datasets.
“Our overall goal is to make better predictive models. Just as weather forecasters can more accurately predict the weather in one hour versus the next week, scientists also need better predictive tools for following models from before birth through adulthood. Although systems biology can help us analyze the Big Data, we still will always need good experimental models that accurately portray phenotypes.”