Systems Biology: Disciplines of Biology, Engineering and Solving Grand Challenges at the Crossroads of Biology, Computation

Systems biology is solving grand challenges by bringing together components from the Systems Biology: disciplines of biology, engineering and Solving Grand Challenges at the Crossroads of Biology, computation. The new Engineering and Computation Integrative BioSystems Institute is helping Georgia By Abby Vogel Tech researchers solve complicated problems in n March 2007, when Georgia sented that day. research projects presented human health and the Tech professors Richard Fu- “Systems biology is a rela- that day,” says Voit, the David D. environment. Ijimoto, Jeffrey Skolnick and tively new field that focuses on Flanagan Chair in the Wallace H. Eberhard Voit invited colleagues complex networks of biological Coulter Department of Biomedi- conducting systems biology re- interactions. When we held the cal Engineering at Georgia Tech search to present at a poster ses- workshop, we were really sur- and Emory University and Geor- sion, they expected a few dozen prised by the enormous inter- gia Research Alliance Eminent researchers. To their surprise, est on the Georgia Tech campus Scholar in Biological Systems. more than 100 posters were pre- and also by the broad range of Georgia Tech’s move toward

Associate professor Frank Löffler and graduate student Elizabeth Padilla-Crespo cultivate bacteria in chemostat vessels. RESEARCH HORIZONS Photo: Gary Meek systems biology research was under way. develop better pharmaceuticals, create Less than a year later, this initiative was more efficient therapies and determine CONTACTS formalized by the launch of the Integra- how to better deal with environmental David Bader tive BioSystems Institute (IBSI) with Voit at challenges that affect human health and 404.385.0004 [email protected] the helm as the founding director. quality of life. IBSI researchers aim to understand Mark Borodovsky the complex biological systems in which Developing the Necessary Tools 404.894.8432 [email protected] molecules, cells, organs, organisms and Systems biology researchers must de- ecosystems interact. Their ultimate goal is fine all of the components of the system Richard Fujimoto 404.894.5615 to solve complicated problems related to they are studying, including the regula- [email protected] human health and the environment. tory relationships between genes and Stephen Harvey “Georgia Tech created the Integrative interactions of proteins and biochemical 404.385.4498 BioSystems Institute to provide our facul- pathways. They can then use this knowl- [email protected] ty members and students with the oppor- edge to formulate conceptual and com- Mark Hay tunity to efficiently and effectively explore putational models that test and predict 404.894.8429 the intersection of the traditional science cellular functions and responses. [email protected] and engineering disciplines with the life Complete genome sequences of Melissa Kemp sciences,” says Gary Schuster, Georgia Tech various organisms are now available, 404.385.6341 [email protected] interim president and provost. and high-throughput technologies have To do this, IBSI brought together a emerged for molecular-level measure- Frank Löeffler 404.894.0279 cross-disciplinary group of researchers ments of gene expression, protein func- [email protected] from Georgia Tech’s Colleges of Science, tion and protein-protein interactions. Engineering and Computing. They These advances make it possible to mea- Eva Lee 404.894.4962 include: sure a large number of cellular compo- [email protected] • Biologists who develop strategies nents simultaneously, and to perform sys- Hang Lu for designing experiments, collect tem-level studies at the molecular level 404.894.8473 reliable data and suggest impor- with unprecedented breadth and depth. [email protected]

tant investigations that shed light The available sequenced genomes, Haesun Park on grand challenges such as cancer however, are not very meaningful until 404.385.2170 [email protected] and environmental sustainability; the genes are distinguished from less use- • Engineers who design new instru- ful information in the DNA strands. Mark Jeffrey Skolnick mentation that allows rapid, highly Borodovsky, director of Georgia Tech’s 404.407.8975 [email protected] parallel, inexpensive and accurate Center for Bioinformatics and Computa- measurements to be recorded of tional Genomics and a Regents’ Professor Patricia Sobecky 404.894.5819 molecules and their characteris- in the Wallace H. Coulter Department of [email protected] tics, localization and interactions Biomedical Engineering and the College Eberhard Voit with other components; and of Computing, has developed several al- 404.385.5057 • Computational experts who man- gorithms and computer programs that do [email protected] age, analyze and integrate the just that for genomes with different levels Howie Weiss wealth of biological data and cre- of complexity. 404-385-2134 ate accurate dynamic models of “It’s like deciphering text when you ini- [email protected]

the system’s performance. tially don’t have a clue what the sequence Joshua Weitz Summer 2008 • Vol. 25, No. 3 “These investigations will lead to the of numbers or letters means,” he says. “The 404.385.6169 [email protected] development of new understanding, new information you are looking for may be tools and new procedures that will help hidden in just five percent of the whole us all to lead longer, healthier lives,” adds sequence, but you need to find this very Schuster. important text.” Georgia Tech researchers plan to un- Since 1995 – the start of the genomic derstand and fight threatening diseases, era – Borodovsky’s line of software pro-

Photo: Gary Meek Assistant professor Hang Lu (left) and graduate student Alison Paul test microfluidic devices. grams, with the common name GeneMark, has been frequently used in laboratories worldwide to find genes. The program was used to find genes in the first com- pletely sequenced genomes of bacteria and archea. Borodovsky has long-term funding from the Nation- al Institutes of Health. Currently, he is collaborating with the U.S. Department of Energy Joint Genome Institute, the Broad Institute of the Massachusetts Institute of Technology and Harvard University, and genome cen- ters in Europe and Japan to find genes in sequences of novel complex genomes. Although genes get a lot of attention, it’s the proteins produced from the genes that perform most life functions. Unlike the relatively unchanging genome, the dynamic proteome changes every minute in response to thousands of internal and environmental signals. A protein’s chemistry and behavior are specified by the gene sequence and by the number and identities of other proteins made in the same cell at the same time and with which that protein associates and interacts. College of Computing professor David Bader and graduate student Kamesh Madduri are using a social network- ing principle called “betweenness centrality” to analyze cellular proteins interacting in complex systems. “It’s a metric in which you look at all of the proteins that are interacting and find the most important proteins – the ones where, if you remove them, you most likely disconnect the entire network,” explains Bader. The researchers have examined human protein-pro-

Photo: Gary Meek Defining the IBSI

“The key words in the name of the Integrative BioSystems Institute are ‘integrative’ and ‘systems.’ Many advances in the life sciences today require an integration of what have previ- ously been traditional disciplines, such as biology, computing and engineering. The disciplines provide strong foundations of knowledge and discovery, but the maximum benefits occur when these are integrated to study systems. Creating the

BRIEF Integrative BioSystems Institute is an organizational method of achieving this integration at Georgia Tech.”

– Don Giddens, dean of the College of Engineering, the Lawrence L. Gellerstedt Jr. Chair in Bioengineering and a Georgia Research Alliance Eminent Scholar

tein interaction datasets and discovered the human manipulation of the organisms information, the amount of data can be proteins important to the cells, including – it’s all automatic,” says Lu. Details about reduced so that the problem becomes one that has been implicated in breast the device were published in July 2008 in computationally tractable,” says Park. cancer. What is intriguing though, says the journal Nature Methods. When comparing gene data from a set Bader, is that the important proteins are The device Lu developed incorpo- of cancer patients to comparable informa- not usually those that interact with the rates complete automation, which allows tion from a set of healthy individuals, it is greatest number of other proteins. Bader her to conduct quantitative population important to determine which genes are plans to continue studying protein-pro- experiments to study the genetics of de- important out of the more than 20,000 in tein interactions with other IBSI mem- velopmental processes in complex sys- the human genome. Effective algorithms bers. tems such as the nervous system and the for doing that can potentially increase the In addition to genomics and pro- effects of small molecules on biological new knowledge about the roles of spe- teomics, another key factor in the rise of networks. cific genes. systems biology is the rapid development The advances in high-throughput The wealth of biological data and of measurement technologies. Modeling measurement techniques also require availability of computers with enormous biological systems requires highly accu- innovations from computer science. The power are beginning to enable realistic rate, high-throughput, quantitative meth- volume of data is large and the number models to be designed of living cells and ods and devices to collect data under of genes or proteins being studied at one organs that can simulate life in a quanti- many different conditions. time is often much greater than the num- tative fashion. Systems biology connects Microfluidic devices show much ber of samples. the wet and dry sciences by an iterative promise for this, and Hang Lu, an assis- College of Computing professor Hae- cycle of model-driven experimentation tant professor in Georgia Tech’s School of sun Park uses numerical linear algebra and and experiment-driven modeling. Chemical & Biomolecular Engineering, has optimization techniques to develop com- “A good mathematical model of a real engineered microfluidic devices to collect puter-based algorithms that dramatically world system should be able to quanti- multiple pieces of data simultaneously reduce the number of genes that must be tatively describe, predict and explain the from cells and microscopic organisms. studied to only those that may be respon- behavior of the system,” says Richard Fuji- “We’ve designed a lab-on-a-chip from sible for specific biomedical conditions. moto, chair of the Computational Science which we can get lots of data with very “With careful consideration of the sig- and Engineering Division in the College high spatial resolution, very high through- nificance of the biological characteristics of Computing and a member of the IBSI put and low error because we’ve taken out of the problem and without losing much executive committee.

Frontiers in Multi- Scale Systems Biology Conference

Georgia Tech’s new Integrative BioSystems Institute (IBSI) (www.ibsi.gatech. edu) will officially be presented to the public at the Frontiers in Multi- Scale Systems Biology conference Oct. 18-21, 2008, in Atlanta, Georgia. This conference will highlight topics of integrative systems biology including genomics, proteomics, metabolomics, molecular inventories and databases, modeling and simulation, high-performance computing, enabling experimental and computational technologies, and applications in cancer, neuroscience and the environment.

Photo: Gary Meek

Senior research technologist Bartosz Ilkowski stands in front of the School of Biology computing cluster.

Environmental and Health Systems Are Similar

“Modeling biological systems of human health and ecology are not very different. If ecologists draw a community interaction web showing connections among predators and prey, competitors, or host and pathogens and then remove the labels, it will look like a network of protein interactions to the human health biologists. The complexity of

BRIEF interactions and the mandate for considering the system instead of just the parts are similar for both molecular/biomedical and environmental scientists.”

– Mark Hay, the Harry and Linda Teasley Chair in Environmental Biology in the Georgia Tech School of Biology

10 Many researchers begin modeling approaches and for the development of Löeffler found microorganisms that biological systems with simple organisms, new methodologies to analyze metabo- detoxify these chemicals by biodegrad- such as bacteria, that can be easy to cul- lites and optimize the production of valu- ing them into ethylene gas, a harmless ture, manipulate genetically, maintain un- able compounds for the food industry. end-product. The genomes of some bac- der controlled conditions and examine in teria involved in this process have been the laboratory. Environmental Systems Linked to sequenced, and current efforts are aimed “Understanding small systems is a nec- Human Health at determining the expression levels of essary prerequisite before we can deal with Once researchers understand micro- thousands of genes and proteins to iden- bigger systems, such as the human body organisms in controlled laboratory sys- tify those responsible for the detoxification and its disease processes,” says Voit. “We tems, they can begin to study them in processes. need to look at microorganisms that have their more complicated natural environ- “With computational models, we can plenty of biological relevance in their own ments. The challenge is that although the better understand the cellular regulatory right, but that also allow us to explore the genomes of hundreds of microbes have networks, including how different compo- function and coordination of metabolites, been sequenced, microbial ecosystems nents in the microbe interact and how pop- enzymes and genes in organisms that are can encompass hundreds or thousands ulations of microbes interact,” says Löeffler. much less complex than humans.” of different populations, each comprising This work is sponsored by the U.S. Depart- In collaboration with Helena Santos, a large numbers of cells that often cannot be ment of Defense’s Strategic Environmental professor at the Universidade Nova de Lis- differentiated as individuals and must be Research and Development Program and boa in Portugal, Voit recently showed that considered as communities. the National Science Foundation. high-precision, dynamic experimental me- Frank Löeffler, the Carlton Wilder As- Löeffler also has funding from the U.S. tabolite data could be combined with non- sociate Professor of Environmental Engi- Department of Energy’s Environmental Re- linear systems modeling to characterize neering, is analyzes complex soil microbial mediation Sciences Program to investigate regulatory mechanisms in the bacterium ecosystems so that he can improve the effi- uranium contamination. Uranium is radio- Lactococcus lactis. These are difficult to as- ciency of environmental decontamination active and can be mobile – that is, it can sess with traditional biological analysis. processes. His work focuses on remediat- be transported with the groundwater flow With a genome of fewer than 300 genes ing groundwater contaminated by toxic and cause widespread subsurface contam- – compared to the more than 20,000 genes chlorinated solvents, such as those used in ination. in humans – this bacterium served as an at- dry cleaning and metal degreasing opera- Remediation is difficult, he says, be- tractive model for Voit’s biological systems tions. cause the actions of one set of microbes

RNA Folding

Christine Heitsch, an assistant professor in the School of Mathematics, is collaborating with School of Biology professor Steve Harvey and College of Computing professor David Bader to understand how ribonucleic acid (RNA) folds in two and three dimensions and how structural information is encoded in large RNA viral genomes. Because current prediction methods cannot reliably and efficiently treat these lengthy sequences, the researchers are developing novel combinatorial and computational approaches to the analysis, prediction and design of viral RNA secondary structures. “Once we know how things fold, we can design drugs to dock at the right places and design a sequence to fold into a specific shape,” says Bader.

11 Photo: Gary Meek

Associate professors Patricia Sobecky (left) and Martial Taillefert analyze soil samples.

can undo the work of another. Microbes tivity of phosphatases. This process, which increased drug resistance. When one bac- transform the soluble form of uranium can occur aerobically and anaerobically, terial cell acquires resistance, it can quick- into an insoluble form that won’t travel can be included in models to remediate ly transfer the resistance genes to many or spread, but certain microbes can re- contaminated areas. species. Cells commonly transfer genetic mobilize the uranium. Löeffler aims to “While investigating phosphatase material between one another via bacte- understand these microbial processes so genes, we found that another gene was be- riophages, also known as phages, which that computational models can be used ing transferred between organisms that al- are diverse and abundant viruses that ex- to predict how best to remediate a con- lowed the organisms to pump heavy metals clusively infect bacteria. taminated area. back out to help them adapt and detoxify Joshua Weitz, an assistant professor With funding from the U.S. Depart- their environment,” explains Sobecky. in the School of Biology, is modeling the ment of Energy, Patricia Sobecky, an as- Currently, Sobecky is developing and mechanisms behind the diversity that can sociate professor in the Georgia Tech applying systems approaches to better emerge from interactions between phag- School of Biology, and Martial Taillefert, an understand how these complex pro- es and bacteria. This is particularly difficult associate professor in the School of Earth cesses such as horizontal (or lateral) gene because the evolution of bacterial viruses and Atmospheric Sciences, are analyzing transfer operate. By studying the genetic is intimately linked to their hosts. aerobic processes that contribute to the material of these organisms, scientists can “I am trying to understand the strate- remediation of uranium. begin to understand how these microbes gies by which viruses interact with bacterial In articles published in the August 15, have evolved and adapted to life in ex- hosts at the cellular scale and the diversity 2007 issue of the journal Environmental treme conditions, and isolate the genes of such strategies in different environments,” Science & Technology and the December that give them their unique abilities to says Weitz, who is funded by the Defense 2007 issue of the journal Environmen- survive and proliferate. Advanced Research Projects Agency and tal Microbiology, Sobecky and Taillefert Horizontal gene transfer is also impor- the Burroughs Wellcome Fund. showed that certain bacteria could pro- tant in human disease because the pro- mote uranium immobility through the ac- cess is thought to be a significant cause of CONTINUED ON PAGE 14 12 Summer 2008 • Vol. 25, No. 3 13 ------By Eberhard Voit By Eberhard Eberhard Voit, 404.385.5057, [email protected] 404.385.5057, Voit, Eberhard The The third area Here of the is expertise goal is computational. and creation device of experimentation, aspects The technical with world living the at look to us encourages biology Systems be So endeavor. mind-altering fascinating, a is biology Systems Biological systems are complex and modeling them requires re requires them modeling and complex are systems Biological engineering. of realm the into falls expertise of area second The Voit is the The David D. Flanagan Chair in the Coulter Department of Bio of Department Coulter the in Chair Flanagan D. David The the di is and Voit Scholar Eminent Alliance Research Georgia Engineering, medical Institute. BioSystems Integrative new Tech’s Georgia of rector Contact: the construction of numerical models that reliably quantify the in the quantify reliably that models of numerical construction the a within and among the processes many components teractions “What observation, Feynman’s Richard Following system. biological enormous the at glancing and understand,” not do I create, cannot I firstwith is accomplished “creating” the in successes engineering, of actual, creation the guide later results whose models, computer systems. biological synthetic, yet of be of drivers will the the certainly success crucial computation systems biology. Motivating these drivers is a systems-oriented mindset. androlefunction the appreciate biologists Systems eyes. part different vital a as but isolation, in not – block building biological each of hierar that machines fine-tuned of intricately context larger in the bring parts, all with in concert and, ensembles larger form chically life. to molecules innate your change may irreversibly biology in systems – ware engaging world! living the of views some higher-level function that is important in a cell, organism or in organism a is cell, that important function higher-level some system? ecological areas main three The unite. to expertise of areas many in searchers high-through uses typically first The here. described are expertise of ex the measure simultaneously instance, for that, experiments put of goal ultimate The organism. an in genes of thousands of and pression inventories molecular complete establish to is researchers these component. biological each of function the characterize biol of none the high-throughput machinery, innovative Without in advances sensing, engineering ogy Brilliant would be possible. bio allow areas other many and robotics miniaturization, probing, a clarity with and rapidly be to investigated blocks building logical before. seen never resolution and COLUMN: ------Y FACULT Systems Biology—What’s All the Buzz About? the Buzz All Biology—What’s Systems

the the only community scientific a few years ago, but what mean? really it does he buzzword “systems biology” entered the entered of vocabulary biology” “systems he buzzword Systems biology complements the Systems by paradigm biology complements reductionist Reductionist biology has Reductionist with researchers provided a rapidly Integrative systems biology adds a new and exciting set of tools tools of set exciting and new a adds biology systems Integrative Understanding how cellular control works—or fails—will sug fails—will works—or control how cellular Understanding The grand challenge of systems biologists is to understand how how understand to is biologists systems of challenge grand The In contrast to cars, they search for their own food and adapt to adapt and food own their for search they cars, to In contrast On a coarse level, biological systems have quite a few similarities similarities few a quite have systems biological level, coarse a On Front and center is a clear focus on systems, which are collec which is on a and systems, center focus clear Front

one particular gene or signaling process if it did not contribute to if it did not contribute process or gene signaling one particular among the building blocks. They are intimately interwoven – after interwoven intimately are They blocks. knowledge building the among detailed without succeed biology systems could how all, in interested be anyone would why And system? a in players the of revealing how higher-level functionality emerges from interactions interactions from emerges functionality higher-level how revealing features and features The functions. results of Just this life. paradigm of reductionist aspect every affected have and amazing truly been have decades! past the over medicine in advances the at look the characterization of select components or processes. processes. or components select of characterization the their on details with along blocks, building molecular of list growing plements plements the experi more paradigm of traditional “reductionist” crisply focus on experiments in mentation which well-controlled cleaning up the environment. environment. the up cleaning com uniquely It techniques. research biological of repertoire the to tion of bacterial pathways will help harness the power of microor power the harness help will pathways of bacterial tion organics of valuable production for the ganisms biotechnological of means or and like yet for ethanol natural, bulk novel, products an academic perspective, but also for very practical reasons. reasons. practical very for also but perspective, academic an the organiza into Insights treatment. cancer for new options gest biological systems function. This is an important goal, not just from from just not goal, important an is This function. systems biological new situations and even to entirely novel environments. They They re and environments. even novel new to situations entirely the although and intelligently, spond appropriately to challenges disputed. sometimes is latter cated and intriguing. Not only do they move, but they also repro also but they do move, Not only they and intriguing. cated cell. single a of out own their on develop and duce with engineered systems, but they are incomparably more compli more incomparably are they but systems, engineered with them together and the parts seemingly come to life, with the cre the with to life, come seemingly parts the and of together them out emerges dynamic Something car. a or set television a of ation objects. of collection a cess make something happen. Typical components in the engineer the in components Typical happen. something make cess do much – don’t and transistors wheels wires, – ing world screws, put engineers let clever However, when left to own their devices. tions of components that interact with each other and in the pro in the and other each with interact that of components tions T 14 CONTINUED FROMPAGE 12 ytm booy rbe bcue the because problem biology systems multi-scale a is dynamics cholera of study The disease. the of outbreaks epidemic or endemic to lead conditions what predict cally model the transmission of cholera and doctoral scientist Hao Wang to mathemati post and Joh Richard student graduate physics Weiss, Howie professor ematics ready infected. al person a with contact require doesn’t infection – food or water contaminated ingesting by occurs transmission Cholera cholera. as such reservoirs, aquatic by ated medi diseases of dynamics the studying begin to him led interactions host-phage cases evolvednewly –strains. some in and – reproduced newly of emergence the to respond hosts and viruses both as changes population term cell and infecting phage, followed by long- researchersinfectedthe analyzefateof the The infection. of site the at begin that ics dynam host-phage of models simulation developed has Weitz bioinformaticians, and mathematicians physicists, biologists, et’ eprie n environmental in expertise Weitz’s includes that team research his With BRIEF Math of School with teamed has He year, with the possibility of extension. of possibility the with year, one for $30,000 about is award typical The Tech. units Georgia different within in way members new faculty a two least in at of program combine expertise the that research projects The research pilot disciplines. supports faculty different two by from supervised be members to by student Tech graduate new Georgia a at funding promotes members faculty program between fellowship collaborations The program. a pilot a research and established program fellowship has student (IBSI) graduate collaborative Institute BioSystems Integrative The Graduate Students ResearchJoint Proposals and ------information about how an individual’s im individual’s an how about information on depends outbreak an of emergence rm niomna dmg.Ti wr is work This damage. environmental from ecosystem’s health and its ability to recover the maximize to way a as interactions of networks complex understand to is goal The ecosystems. to nutrients and petitors ing or removing different consumers, com systems. biodiverse these sustain that interactions cific to understand the complex network of Pa South and Caribbean the in reefs coral tropical Biology,explores Environmental in Mark Hay, the and Harry Linda Teasley Chair purification. air and water and production food as such services ecosystem taining and ecosystem restoration with the goal of sus conservation to approaches novel developing are researchers Tech Georgia depend. humans which upon ecosystems natural of health the improving and ing sustain for also options management provide they health, human of knowledge a population. among spreads pathogen the how as well mune system contends with a pathogen as Hay conducts field experiments by add improve approaches systems as Just ------environmental systems that affect human affect that systems environmental Tech,Georgia At communities. connected plexity, live in and rely on diverse and inter provide. they services critical the on depends and ecosystems natural values both that population man hu growing a of face the in environment natural the of stewardship better to lead will exploration multi-scale system-wide, of type this that hopes also environmental He stressors. and species non-native of invasions change, global of effects the from recovering and for avoiding models predicting, develop to used be will ing ofHealth. tion andtheNationalInstitutes Founda Science National the by funded considered in isolation, which can often be typically is combination target and drug Each date. to development and research pharmaceutical in roles minor relatively Improving Discovery Drug cesses. pro and systems living real of complexity health are being modeled on the scale and All organisms, regardless of their com their of regardless organisms, All understand this that anticipates Hay oeig n smlto hv played have simulation and Modeling Images: Vickie Okrzesik ------Summer 2008 • Vol. 25, No. 3 15 Photo: Gary Meek Gary Photo: CONTINUED ON PAGE 18 ON PAGE CONTINUED To discover To where new drugs will bind netic code into proteins. proteins. netic code into to the ribosome, researchers must know which ribosome, the of shape structural the includes two subunits that assemble to - - - - prompting extensive effort in the design of design the in effort extensive prompting antibacterial One new agents. or improved target of antibiotics is the ribosome – the cellular workhorse that translates the ge to to treat bacterial infections, pathogenic strains have acquired antibiotic resistance, -

Assistant professor Joshua Weitz discusses cholera transmission with postdoctoral scientist Yuriy Mileyko and graduate student Richard Joh. Richard student and graduate Mileyko Yuriy scientist with postdoctoral transmission discussescholera Weitz Joshua professor Assistant

function and decreasing the symptoms of both disorders. This interdisciplinary research is research This the of interdisciplinary and function symptoms decreasing both disorders. Emory at Initiative Health Predictive Center’s Sciences Health Woodruff the by funded being University. The researchers plan The to researchers use the model and in with studies clinical biological conjunction at conducted Emory to University aimed at screen novel therapeutics dopamine altering vironmental and pharmacological factors alter how dopamine functions in healthy neuro in healthy functions dopamine how alter factors and pharmacological vironmental schizophrenia. and disease Parkinson’s like diseases neurodegenerative in and transmission gineering, postdoctoral fellow Zhen postdoctoral gineering, Qi and Gary an Miller, in professor associate Emory’s Department of and Environmental en Occupational Health, are genetic, developing a how mathematical understand better to it use to hope They network. dopamine the of model Eberhard Voit, Voit, Eberhard the David D. Flanagan Chair in the Coulter Department of Biomedical En Modeling Neurological Disorders Neurological Modeling With With the increased use of antibiotics

product. are are complex and not just the result of the contribution of a single gene or its protein tributing to the development of disease misleading because the mechanisms con Mass Spectrometry: Techniques Offer Molecular Information to Help Solve Systems Biology Problems

nderstanding biology at the systems level is difficult, espe- cially when studying complex specimens like tissue slices Uor communities of organisms in a biofilm. Scientists must be able to identify, quantify and locate the molecules present in the samples. Mass spectrometry imaging is a powerful analytical technique with the potential to unravel the molecular complexities of such biological systems. It allows researchers to visualize the spatial arrange- ment and relative abundance of specific molecules – from simple metabolites to peptides and proteins – in biological samples. At Georgia Tech, researchers from the Colleges of Sciences and Engineering have joined forces to create the Center for Bio-Imaging Mass Spectrometry (BIMS) that aims to tackle these types of challenges. “We organized this center in 2007 when we saw the enormous Professor Al Merrill, principal research scientist Cameron Sullards (left to right potential of mass spectrometry imaging tools and realized that we standing) and research scientist Yanfeng Chen display mass spectrometry results.

had a unique ensemble of people at Georgia Tech that would en- Photo: Gary Meek able us to excel in this field,” says Al Merrill, a professor in the School the large sets of complicated data collected by mass spectrometry of Biology and the Smithgall Chair in Molecular Cell Biology. systems. Mass spectrometry imaging takes advantage of the ability of biological molecules to be converted into ions that can then be sepa- Improving Current Technology rated and analyzed by a mass spectrometer. When data is collected Today, a popular technique for studying biological samples is ma- from different regions of a sample, the distribution of molecules can trix-assisted laser desorption/ionization mass spectrometry (MALDI- be used to create multidimensional images of that sample. MS). In this technique, sample preparation plays a very important As a cell biologist, Merrill sees the technique’s potential in its abil- role in image quality because it requires that a matrix compound ity to detect all of the important molecules that control cell behavior be uniformly deposited over the surface of a histological tissue slice instead of just a few. Another advantage to mass spectrometry is mounted on a special plate. the ability to test whether all of the cells are being affected in the A research team including Merrill, along with Cameron Sullards, same ways. director of Georgia Tech’s Bioanalytical Mass Spectrometry Facility, His laboratory uses mass spectrometry to profile sphingolipids, and Yanfeng Chen, a research scientist in the School of Chemistry a family of thousands of metabolites that are involved in cell-cell and Biochemistry, recently showed that the homogeneity of the communication and intracellular signaling. He also studies the types matrix could be improved. With this development, broader catego- and amounts of these metabolites that control whether cells grow ries of compounds, such as lipids, could be analyzed. or die. The researchers used an oscillating capillary nebulizer to spray “With mass spectrometry, we have not only been able to profile small droplets of matrix aerosol onto the sample surface – a process these compounds, but also to find new metabolites we think are similar to airbrushing. Using histological samples provided by Timo- important in inflammation, aging and cancer,” says Merrill. thy Cox, a professor of medicine at the University of Cambridge, the The BIMS center includes researchers like Merrill who propose researchers could profile and localize many different lipid species in biological and clinical problems that may be solved by mass spec- the samples. Specifically, they localized sphingolipids that accumu- trometry imaging. It also brings together researchers who are im- late in the brain when there is a genetic defect. This research was proving current mass spectrometry imaging technologies and de- published in the April 15 issue of the journal Analytical Chemistry and veloping innovative techniques, and researchers who are analyzing was funded by the National Institutes of Health. RESEARCH HORIZONS 16 Photo: Gary Meek

Developing New Technologies While MALDI samples must be analyzed in a vacuum, recent advances allow samples to be studied under ambient conditions. Facundo Fernandez, an assistant professor in the School of Chemis- try and Biochemistry, has been using a technique called desorption electrospray ionization (DESI). With DESI, a high-speed, charged spray containing alcohol and water is directed at a sample a few millimeters away. The solvent droplets pick up portions of the sample through interaction with the surface and then form highly charged ions that can be analyzed. Fernandez and his research team recently used DESI to analyze nearly 400 drug samples provided by public health authorities to identify counterfeit anti-malarial drugs. “We have done a lot of work using DESI to analyze pharmaceutical formulations, but we are moving into new avenues of research including looking at algae samples, as well as ovarian cancer tissue samples provided by the Ovarian Cancer Institute, which is housed at Georgia Tech and headed by School of Biology chair John Mc- Professor Thomas Orlando and graduate student Irene Anestis-Richard conduct an Donald,” says Fernandez. experiment with the single photon ionization mass spectrometer. In ovarian cancer research, little is known about how biomarkers and low-mass signaling molecules increase or decrease in abundance with treatment. Fernandez has teamed with Thomas Orlando, lected from every tissue slide matrix, perform quality control and chair of the School of Chemistry and Biochemistry, to use laser de- visualize the distribution of ions on the tissue matrix. The research- sorption single photon ionization mass spectrometry (LD/SPI-MS) to ers then use data mining methodologies – including principal com- investigate this issue. ponent analysis, independent component analysis and multivariate Because it does not use a matrix, LD/SPI-MS can detect low- analysis – to identify and compare ions of interest present in differ- mass molecules – such as sugars, amino acids, small peptides and ent locations. cytotoxic compounds – formed as result of cancer treatment. It can With the advances in software and hardware, the use of mass also achieve higher spatial resolution than is typically possible with a spectrometry in the life sciences promises to become even more desorption laser, according to Orlando. prevalent and diversified for systems biology research. “We hope LD/SPI-MS will lead to a better understanding of the molecular basis of ovarian cancer at its various stages and how treat- By Abby Vogel ment affects regulation of low-mass biomarkers in ovarian cancer cells,” says Orlando. Contacts: Facundo Fernandez, 404.385.4432 The Information Challenge [email protected] Mass spectrometry experiments produce incredible volumes Al Merrill, 404.385.2842 of data, each composed of thousands of spectra and thousands of [email protected] peaks, which makes finding molecules of interest very difficult. “We’ve focused on researching computational methods that Thomas Orlando, 404.894.4012 can clean up, visualize and look for interesting patterns in thousands [email protected] of mass spectrometry tissue images that you wouldn’t necessarily Summer 2008 • Vol. 25, No. 3 Cameron Sullards, 404.385.4249 be able to find or have time to find with the naked eye,” says May [email protected] Dongmei Wang, an assistant professor in the Wallace H. Coulter De- partment of Biomedical Engineering and a Georgia Cancer Coalition May Dongmei Wang, 404.385.2954 Distinguished Cancer Scholar. [email protected] Wang, post-doctoral fellow Mitchell Parry and graduate students Richard Moffitt and Peter Siy are providing software systems to BIMS center users. The systems acquire the thousands of ion spectra col-

17 CONTINUED FROM PAGE 14

chal Brylinski to develop a method for pre- the journal Proceedings of the National Acad- produce a functional particle at the begin- dicting where signal-triggering molecules emy of Sciences. This work was funded by the ning of the process of protein biosynthesis. called ligands will bond to target proteins. National Institutes of Health. School of Biology professor and Georgia The prediction can be made without know- In another project, Skonick and post- Research Alliance Eminent Scholar Steve ing the protein structure. doctoral scientist Roman Mezencev, re- Harvey uses computer models to investi- To complete the complex calculations search scientist Adrian Arakaki and School gate the structural shape, and in turn, func- required for the prediction, Skolnick has a of Biology chair John McDonald aim to dis- tion, of the ribosome. computing facility that contains 4,600 core cover drug targets for treating ovarian can- He developed a structural model of processors that perform calculations 24 cer. Their goal is to enhance the non-can- the large subunit of the mammalian ribo- hours a day, seven days a week. The com- cerous pathway by promoting production some by combining molecular modeling puter rates the quality of the fit to various of specific molecules whose production is techniques with cryo-electron microscopic sites on the protein, analyzing the mol- reduced by the disease. data that will allow him to screen where ecules’ abilities to either enhance or disable “It’s much easier to feed cells a molecule novel drugs will bind to the ribosome and the function of the protein, depending on that’s been inhibited because of a disease affect its function. its function in the cell. rather than designing a drug to inhibit a mol- Jeffrey Skolnick, a professor in the School Identifying the ligand-binding site is often ecule that’s being produced in excess due to of Biology, a Georgia Research Alliance Emi- the starting point for protein function deter- a disease,” explains Skolnick. nent Scholar and associate director of IBSI, mination and drug discovery. Details of the The researchers have developed a list has worked with postdoctoral scientist Mi- method were published in January 2008 in of metabolites whose production has been

Image: Eva Lee

Associate professor Eva Lee’s research is fingerprinting microvascular networks for early disease diagnosis. RESEARCH HORIZONS 18 Photo: Gary Meek reduced in diseased cells, but which may stop cancer cells from multiplying if me- tabolite production resumes. They have tested these metabolites on leukemia and ovarian cancer cell lines and results have shown a reduced growth rate of the cancer cells by 90 and 50 percent, respectively. This work was published in June 2008 in the journal Molecular Cancer.

Predicting Health Risk, Improving Disease Diagnosis and Treatment A wealth of biological and clinical data betes, premature aging, macular degener- to analyze their properties dynamically. By is available for health professionals to ex- ation and tumor metastasis. Her predictive measuring molecules downstream from amine, but existing modeling and com- models have yielded correct classification the T-cells in a high-throughput manner, putational techniques limit their ability to rates ranging from 80 to 100 percent. Lee Kemp and Lu can build mathematical handle such heterogeneous large-scale and her medical collaborators have also models to assess the quality of the T-cell data sets. This limits the ability of research- conducted blind tests, where diagnoses population before infusing the cells into ers to uncover discriminatory factors that are predicted for new patient data based the patient. That could potentially improve might assist in predicting health risk, de- on a developed rule. These tests resulted in the therapeutic outcome. tecting disease early and predicting treat- correct predictions more than 83 percent As the field of systems biology expands, ment outcome. of the time. Georgia Tech’s Integrative BioSystems Insti- Working with clinicians and experimen- These results provide a strong basis for tute is working to be on the forefront. The tal biologists, Eva Lee, an associate profes- pursuing their use as medical diagnostic, Institute will continue to encourage the in- sor in the School of Industrial and Systems monitoring and decision-making tools, Lee tegration of information about cells and or- Engineering and director of Georgia Tech’s says. This research is supported by the Na- ganisms at multiple levels and the integra- Center for Operations Research in Medicine tional Institutes of Health and the National tion of knowledge from biological science and Healthcare, has developed a general- Science Foundation. and tools from mathematics and computer purpose predictive modeling and software In addition to diagnoses, systems biology science for a better understanding of the system that determines predictive rules for a approaches can also be used to improve dis- biological systems. wide range of biological and medical prob- ease treatments. Hang Lu and Melissa Kemp, “Georgia Tech has a culture of inter- lems. These rules can assist in early disease an assistant professor in the Coulter Depart- disciplinary research and education that prediction and diagnosis, identification of ment of Biomedical Engineering and Geor- goes back several decades and enables us new target sites – genomic, cellular and mo- gia Cancer Coalition Distinguished Professor, to address issues such as systems biology lecular – for treatment and drug delivery, dis- aim to improve a new cancer therapy called in ways that other institutions can’t,” notes ease prevention and early intervention and adoptive transfer of T-cells. Don Giddens, dean of the College of En- optimal treatment design. During this process, clinicians remove gineering, the Lawrence L. Gellerstedt, Jr. For one project, Lee has developed a T-cells from a cancer patient, multiply them Chair in Bioengineering and a Georgia Re- model to predict atherosclerosis by exam- in the laboratory and then infuse them back search Alliance Eminent Scholar. ining inexpensive physiological measure- into the patient’s body to attack the cancer. ments, traditional risk factors and novel However, this type of therapy is limited by Views, opinions and/or findings contained in this biomarkers. In another, she has completed the difficulty of generating sufficient- num article are those of the scientists and should not be Summer 2008 • Vol. 25, No. 3 genomic analyses to predict the abnormal bers of active T-cells in the laboratory. construed as an official Department of the Navy methylation of certain genomic regions. “Researchers use an artificial method position or decision. The conclusions of research Methylation is a process that has been of population growth that may cause the projects described in this article that are funded by the shown to silence genes responsible for tu- cells to divide too many times and become National Institutes of Health (NIH) and the National mor suppression. unresponsive and inactive,” says Kemp. Science Foundation (NSF) are solely the responsibility Lee is also fingerprinting microvascular Kemp and Lu are using microfluidic de- of the authors and do not necessarily represent the networks to provide early diagnosis of dia- vices to stimulate and break open T-cells official views of the NIH or NSF.