)URP 'HDQ )UHHPDQ $ORRNDWWKHODWHVW6FLHQFH1HZVmurj.mit.edu/issues/MURJ29_full.pdf · 3 Letters Volume 29, Spring 2014MURJ May2015 DearMITCommunity, IamdelightedandhonoredtointroducetheSpring2015issueofthe

Introductory Letters

From Dean FreemanFrom MURJ Editors

Science News In Review

A look at the latest Science News.

Features

Science in theWorld:Fashion andVideo Games

Science outreach is important for edu-cation and for invoking enthusiasm forresearch. Dr. Tal Danino is helpinginvolve the public in biology research byusing fashion and video games.

MURJ Spotlight:Professor DavidC. Page, Directorof the WhiteheadInstitute

MURJ interviews Professor DavidC. Page, Director of the WhiteheadInstitute, Professor of Biology at MIT,and Investigator at the Howard HughesMedical Institute

Contents

3-5

6-12

14-16

17-22MURJ

Massachusetts

Instit

ute

of

Technolo

gy

Undergraduate

Research

Journal

Cover image: Modified figure from "Gathering Real World Data to PredictAdmission Statistics;" See page 28 for the full report

ReportsOn Mortgages and Refinancing

Khizar Qureshi, Cheng Su

In general, homeowners refinancein response to a decrease in inter-est rates, as their borrowing costsare lowered. However, it is worthinvestigating the effects of refinanc-ing after taking the underlying costsinto consideration. Here we developa synthetic mortgage calculatorthat sufficiently accounts for suchcosts and the implications on newmonthly payments. To confirm theaccuracy of the calculator, we simu-late the effects of refinancing over 15and 30-year periods...

Volume 29, Spring 2015 ContentsMURJ

43-47

Improving the Efficiency ofCarbon Fixation in Ralstonia eut-ropha with Carbon-concentratingMicrocompartments

Sophia Li, Jingnan Lu, ChristopherJ. Brigham3, Anthony J. Sinskey

Microorganisms such as Ralstoniaeutropha offer renewable solutions tocurrent global warming and carbonemission challenges with their abilityto use CO2 as the carbon source forvaluable products such as biodegrad-able plastic and biofuels. An impor-tant bottleneck in this process is R.eutropha’s carbon fixation efficiency,which is limited by the enzyme, ribu-lose-1,5-bisphosphate carboxylaseoxygenase (RuBisCO), since RuBisCOcan take both CO2 and O2 as sub-strates...

36-42

UROP Summaries

Simulation of BloodMicrocirculation andStudy of Sickle Cell Disease

Jean-Pierre Theron1, RaymondJansen, Wesley Harris,Francois Le Floch

Sickle cell disease is a genetic diseasethat affects the red blood cell behav-ior of a great number of people. Inan effort to help the development ofa cure, Dr Francios Le Floch pre-viously developed a sophisticatedmathematical model that simulatesthe flow and oxygen diffusion char-acteristics of capillaries. The aim ofthis project is to replicate the resultsthat Le Floch acquired as well as toexpand the calculated model...

Gathering Real World Data toPredict Admission Statistics

Noor Khouri, DimitriosPagonakis, Marta González

In this paper we design a model thatestimates the incoming flow of inter-national graduate and undergradu-ate students at the MassachusettsInstitute of Technology (MIT). Theauthors, both international studentsat the university, were motivatedto explore this question due to theirdiverse cultural backgrounds. Theanalysis uses a modified version ofthe Gravity Model used in networktheory and estimates the number ofstudents coming from each country...

28-35

24-26

MURJ 2015 CPW ResearchPoster CompetitionAn article on MURJ's second inter-disciplinary undergraduate researchsymposium.

48

3

Letters Volume 29, Spring 2014MURJ

May 2015

Dear MIT Community,

I am delighted and honored to introduce the Spring 2015 issue of theMIT Undergraduate Research Journal (MURJ) to you. This particu-lar issue marks the 15th year since MURJ was founded. Despite thepassage of time, the original vision for the publication—an interdis-ciplinary, peer-reviewed scientific journal, modeled after ScientificAmerican—still holds true today in this edition of the journal.

MURJ offers a window into the remarkable research students under-take here through the Undergraduate Research Opportunities Program(UROP). UROP was established at MIT in 1969 and is now emulatedat academic institutions around the globe. In many ways, UROPembodies what MIT’s founder, William Barton Rogers, envisioned for his then-novel, independenteducational institution: learning by doing, and bringing that knowledge to bear on the nation’s—andthe world’s—greatest challenges. Since its inception, UROP has become an integral part of an MITeducation: in any given year, 60% of undergraduates participate. And by the time they graduate, 89%of students have completed at least one UROP. These are remarkable statistics that Rogers wouldundoubtedly be pleased with!

Participating in faculty-mentored research is a powerful experience for undergraduates. However,the same can be said of the faculty experience. I have been privileged to collaborate with a number ofUROP students over the years, and it has been one of the most rewarding aspects of my work here.

MURJ also reflects the incredible breadth of research that takes place on a daily basis at the Institute,and the interdisciplinary nature of that work. Reaching across academic departments and schoolboundaries is an integral part of MIT; it’s no accident that there are over 50 interdisciplinary centers,labs, and programs here, including the new Institute for Data, Systems and Society set to launch inJuly.

I hope you enjoy this issue of MURJ as much as I have. It is a wonderful distillation of just a fraction ofthe extraordinary and unique undergraduate research projects going on in all corners of the Institute.

Sincerely,

Denny Freeman

Dean for Undergraduate EducationProfessor of Electrical EngineeringMacVicar Faculty Fellow

4

No material appearing in this publication may be reproduced withoutwritten permission of the publisher. The opinions expressed in thismagazine are those of the contributors and are not necessarily sharedby the editors. All editorial rights are reserved.

UNDERGRADUATERESEARCH JOURNALVolume 29, Spring 2015

Editors-In-ChiefElliot Akama-GarrenTatyana GubinReuben Saunders

Research EditorLakshmi Subbaraj

News EditorsRiya JagetiaLinda Jiang

Features EditorsEta AtoliaPaulomi BhattacharyaLinda Jiang

Copy EditorsMahmoud GhulmanWinter GuerraCarolyn LanzkronDaphne Superville

Web DeveloperWinter Guerra

Production AdvisorDSGraphics

Printed at Upper ValleyPrinting

May 2015

Dear MIT community,

We hope you enjoy 29th issue of the MIT UndergraduateResearch Journal, a biannual student-run publication that fea-tures groundbreaking undergraduate research from across cam-pus. MURJ is now in its 15th year of publication, and theresearch conducted and articles written by MIT undergraduatesdemonstrate, as always, phenomenal passion and talent in allscientific disciplines. The high diversity of research interestsamong MIT undergraduates is evident--in this issue, we learnabout a computational model predicting admissions statistics ofMIT international students, a method for improving the carbon-fixation abilities of a microorganism, and a comprehensive analy-sis of mortgage refinancing options.

In addition to our research articles, we also present sciencenews articles reviewing cutting-edge scientific discoveries, aswell as features articles elaborating more comprehensively onparticularly fascinating topics. In this issue, our features detailthe creativity of science education and outreach to the generalpublic, through the merging of artistic and scientific work inmediums such as fashion and video games. We also interviewand spotlight Professor David C. Page, Director of the WhiteheadInstitute, to learn his path through science and his advice to ourpeer young scientists at MIT.

During CPW 2015, MURJ hosted its second interdisciplinaryundergraduate poster session, currently the only annual all-major poster session symposium at MIT. The event was a greatsucess, and we are pleased to include an article reviewing theevent. We invite you to check out this article and the others pre-sented within on our website at murj.mit.edu, where we highlight


Volume 29, Spring 2015

the work of the MIT students published within and host an archiveof previous MURJ issues.

This journal is a collaborative effort by an extraordinary team ofdedicated students, and we would like to thank all our editorialboard and contributors for their time and effort this semester.Without the work of our hard-working staff members, MURJ wouldnot exist. We would also like to thank all of the student researcherswho graciously shared their research.

If you would like to contribute to future issues of the MITUndergraduate Research Journal, we invite you to join our team ofauthors and editors or submit your research for our Fall 2015 issue.Please contact [email protected] if you have any questions orcomments.

Best,

Elliot Akama-Garren(Co-Editor-in-Chief)

Tatyana Gubin(Co-Editor-in-Chief)

Reuben Saunders(Co-Editor-in-Chief)

5

No material appearing in this publication may be reproduced withoutwritten permission of the publisher. The opinions expressed in thismagazine are those of the contributors and are not necessarily sharedby the editors. All editorial rights are reserved.

UNDERGRADUATERESEARCH JOURNALVolume 29, Spring 2015

Research StaffJose A. EsparzaKaustav GopinathanRiya JagetiaDo Yeon ParkJennifer SwitzerKimia Ziadkhanpour

NewsContributing StaffElizabeth BergArchis R. BhandarkarChristopher GonzalezPratheek NagarajSebasthian SantiagoJennifer F. SwitzerKimia Ziadkhanpour

FeaturesContributing StaffElizabeth BergRiya JagetiaBrianna JonesJae Hyun Kim

Massachusetts Institute of Technology77 Massachusetts AvenueCambridge, MA 02139

PSB 07-02-0097



6

Science News in ReviewMURJ

Science News In Review

Long gone may be the dayswhere diabetes patients have to

puncture themselves with needlesin order to measure their bloodsugar levels. In a paper publishedin Analytical Chemistry, scientistsat the University of California, SanDiego have presented a proof-of-concept of a temporary tattoo thathas the ability to monitor glucoselevels in a noninvasive manner.

This tattoo uses iontophoretic-bio-sensing to measure glucose levelsand has been shown to be selec-tive for glucose and to be able todetect rapid rises in glucose levelsafter eating. Currently, the tattooonly works for a day, and it doesnot give a numerical read-out ofblood sugar content; the scien-tists are now working on addingthis feature while also increasingthe tattoo’s longevity and add-ing new functions such Bluetoothconnectivity, which would allowthe diabetes patients’ tattoos tosend collected data to their doc-tor. Nevertheless, this tattoo repre-sents a monumental advancement

in the development of a non-inva-sive glucose monitor.

—S. SantiagoSource: http://news.sciencemag.org/chemis-try/2015/01/temporary-tattoo-measures-blood-glucose-levels

The planets and stars that wesee around us are made up of

matter. This ordinarymatter has a counterpart – anti-matter. Particles of antimatterare identical in size and mass tothat of ordinary matter, but haveopposite other properties such ascharge and spin. A notable effectof interaction between matter andantimatter is annihilation wherethe particle and antiparticle collideand most often disappear emittingenergy.

For several decades now, cosmolo-gists and astronomers alike havebeen perplexed by the fact that ourobservable universe is abundantlymore matter than antimatter, an

imbalance known in cosmology asasymmetry. Alexander Kusenkoand his colleagues at UCLA believethat in the early universe the Higgsfield may have influenced the pro-portion. The Higgs field is associ-ated with the Higgs boson, theparticle recently discovered byphysicists at CERN working at theLHC, and is responsible for giv-ing particles and antiparticles theirmass.

In the primordial soup of theearly universe, there existed manymore particles than we see today.It is believed that antimatter andmatter existed in nearly equalproportions with a difference ofonly one in 10 billion. As the uni-verse cooled, they annihilated toleave just the matter that we seetoday. Kusenko believes that thisslight difference was the result ofa process called Higgs relaxationwherein the Higgs field changeswith the course of the universe’sevolution. Researchers at theLHC have measured this effectthrough the mass of the Higgsboson and found the field to bemore representative of the first

Tattoos: Glucometersof the 21st Century

BIOMEDICAL ENGEERING

Blood glucose measuring tattooCredit: http://news.sciencemag.org/chemistry/2015/01/temporary-tattoo-measures-blood-glucose-levels

Matter-AntimatterAsymmetry Linkto Higgs Boson

PHYSICS

All the planets and stars in the observable universe are made up of matter andnot antimatter. This imbalance is what is known as baryon asymmetry.Credit: NASA

7

Science News in Review Volume 29, Spring 2015MURJ

Abasic checklist of key factors forsuccess of a startup includes:

location, trademarking, patents,legal incorporation, and naming.

MIT affiliates Jorge Guzmanand Scott Stern tracked variousentrepreneurial ventures acrossCalifornia and reinforced somecommon perceptions as well as elu-cidated some new factors that leadto the success of a startup. Forinstance, having a shorter namelike Google or Facebook provesto be more effective than longercompany names such as CryptineNetworks.

Establishing a relationship betweenentrepreneurship success andregional location has been diffi-cult. Much of the previous attemptsfocused on quantity of ventures,whereas in the study by Guzman

and Stern they attempt to assess thequality of ventures in order to over-come this difficulty. They introducea method to measure what is calledthe entrepreneurial quality index.By accounting for naming deci-sions, type of business registration,and control of intellectual property,the researchers were able to predictthe likelihood of company growthand in turn compare the entrepre-neurial quality across cities in thestate of California.

Guzman and Stern found thatnames associated with high-tech-nology clusters are indicative ofgrowth whereas companies namedafter the founders are less likelyto grow. Registering as a corpora-tion and trademarking also sug-gest a higher probability of growthin a company. They then mappedthis quality index against locations

in California and found SiliconValley as the most dominantentrepreneurial hotspot beatingout San Francisco, San Diegoand Los Angeles. More specifi-cally, they found higher entre-preneurial quality located aroundresearch institutions, and in par-ticular Stanford, which lies atthe center of Silicon Valley. Theresearchers believe the work hasimportant implications on entre-preneurial growth dynamics,scientific research, and possiblypublic policy.

—P. NagarajSources: http://www.sciencemag.org/

cotent/347/6222/606.full.pdfhttp://news.sciencemag.org/econom-

ics/2015/02/why-facebook-and-google-succeeded

Mapping the Success of Startups

Map portraying the entrepreneurial quality in California’s Bay Area.Notable concentration on the South Bay centered on Stanford University. .Credit: GUZMAN AND STERN/SCIENCE (2015); (ILLUSTRATION) R. J. ANDREWS

ENTREPRENEURSHIP


8


few moments of the Big Bang thanwhat we expect today. Therefore asa result, Kusenko states that parti-cles and antiparticles may have hadtemporarily unequal masses due tothe early Higgs field leading to theasymmetry we have today.

—P. NagarajSource: http://www.sciencedaily.com/releases/2015/02/150225132255.htm

In a recent study that was amongthe first to explore false memories

in non-human animals, scientistswere able to ascertain that bumble-bees are prone, much like humans,to remember things incorrectly. Toconduct their research, Lars Chitikaand his colleague, Katheryn Hunt ofQueen Mary University of Londontrained bumblebees to expect areward when visiting a solid yellowartificial flower followed by one withblack and white rings or vice versa.During test trials, the bees weregiven a total of three choices amongflowers, the third of which was amixed-up version of the other two.For example, the third flower wouldhave yellow and white rings. If thistest was conducted only minutes

after their training, the bees wouldgravitate towards the one that theyhad most recently been rewardedwith – their short term memory wastherefore intact. After a period ofone to three days, however, some-thing very different happened whenthe bumblebee’s memory was putto the test: They began to select theflowers that had yellow rings, eventhough that was not a flower theyhad used in their training before.

ChitikaandHuntstatethat themerg-ing of long-term memories phenom-enon observed in these bumblebeesis what happens to humans as well.They believe that these false memo-ries are not “bugs in the system”but rather the side effects of anadaptive learning memory systemthat is parallel to human memoryformation. Further research is beingconducted and, so far, the group hasbeen able to find that extracting pat-terns and commonalities betweendifferent events in our environmentis an inherent part of our ability toadapt to our environment. Thus, theintegration of lifetime experiencescan accumulate and be incorporatedinto a comprehensive foraging deci-sion as in the case for bees and lifedecisions as in the case of humans.

—K. ZiadkhanpourSource: http://www.sciencedaily.com/releases/2015/02/150226131940.htm

Setting fires to clear farmland isan agricultural technique around

the world, but recent research sug-gests that the smoke from thesefires may lead to more violent hur-ricanes. A paper recently publishedin Geophysical Research Letterslooks at the deadly outbreak of 120tornadoes that hit the southeasternUnited States in the Spring of 2011.Researchers noted that the air inthe region contained smoke thatcould be traced back to Mexican andCentral American agricultural firesset earlier that year. Using computermodels to recreate the atmosphericconditions at that time, research-ers believe that the smoke in the aircould have bolstered the tornadoes’force. In particular, the model pro-duced faster wind speeds and largerstorm clouds than the smoke-freesimulation. These weather condi-tions- the fast wind and low, thickclouds- would not be enough tohave produced the tornadoes ontheir own, but their presence didmake these storms more intense.Perhaps the tornadoes would nothave racked up such a large deathtoll, ultimately killing 313 people, if

False Memories inBumblebees

COGNITIVE SCIENCE

Smoke Found toIncrease Tornadoes’Strength

EARTH SCIENCE

Bumblebees, like humans, can makefalse memories.Credit: http://www.sciencedaily.com/releases/2015/02/150226131940.htm

A tornado in Shoal Creek Valley, Alabama, one of the 120 to hit southeasternUnited States on April 27, 2011Credit:http://upload.wikimedia.org/wikipedia/commons/9/9c/Shoal_Creek_Valley_Alabama_Tornado_April_27%2C_2011.jpg

9


there had not been smoke in the airto boost their strength. At the veryleast, these researchers hope thattheir results will encourage meteo-rologists to factor smoke into theirfuture weather models.

—E. BergSource: http://news.sciencemag.org/earth/2015/02/smoke-distant-fires-could-create-more-deadly-tornadoes

Carbon dioxide is an importantgreenhouse gas. Atmospheric

carbon dioxide traps heat thatreflects of the Earth’s surface, pre-venting this heat from escapingthe atmosphere. Over time, thisincrease in trapped energy raisestemperatures around the world.While this has been commonknowledge for years, scientistshave only recently confirmed thistheory experimentally, finding anincreased amount of thermal radia-tion absorbed at two data collec-tion sites over the last ten years.These researchers, working at theDepartment of Energy’s LawrenceBerkeley National Laboratory, hadbeen collecting almost-daily radia-tion measurements at researchfacilities in Oklahoma and Alaskafrom 2000 to 2010. To make thesemeasurements, the researchers

used spectroscopic instrumentsthat measured both the totalamount of infrared energy trappedin the atmosphere and the radia-tion specifically emitted by carbondioxide. This data shows a 0.2 Wattper square meter increase in ther-mal energy in both locations overthis time period. This increase isconsistent with what theory wouldpredict for the amount of carbondioxide that has been emitted intothe atmosphere over this sametime period. This research is thefirst definitive and experimentalproof that global carbon dioxideemissions are warming the surfaceof the Earth.

—E. BergSource: http://www.sciencedaily.com/releases/2015/02/150225132103.htm

Today we know that the richnuances hidden in genomic

data speak volumes about peopleand their ancestors. But, in dig-ging deep back into the geneticvariations that trace to our earlyevolutionary history, could we pos-sibly unearth the origin of uniquelyhuman disorders? That is couldthe same insights from genomicdata that have previously shed light

on human evolution also speak toour understanding of mental ill-nesses like autism, schizophrenia,and bipolar disorder? A recentstudy published by Mount Sinairesearchers in Molecular Biologyand Evolution may have theanswer.

The team of researchers led by JoelDudley chose to reexamine datafrom the Psychiatric GenomicsConsortium- a compendium thatboasts genomic data from 36,989schizophrenia cases and 113,075controls. Specifically, the groupsought to explore a possible rela-tion between previously identi-fied schizophrenia-associated lociand human accelerated regions(HARs). HARs represent hotspotsof high mutation activity in thehuman genome relative to non-human species; in short, thesesmall stretches in the humangenome in tandem with nearbygenes conserved in non-humanprimates (pHARs) may very well bethe crux of what separates humansfrom all other species.

Bayesian network of genes relatedto prefrontal cortex development.Non HAR-associated schizophre-nia genes (blue), HAR-associatedschizophrenia genes (pink), othergenes (green).Credit: http://mbe.oxfordjournals.org/content/

early/2015/03/15/molbev.msv031.full.pdf+html

The DarkerHistory Hiddenin Our Genes

GENETICS

ExperimentalConfirmation thatCO2 Contributes toGlobal Warming

A photograph ofthe AtmosphericRadiationMeasurement ClimateResearch Facility inHawaii, one of twoDepartment of Energysites where thermalradiation measure-ments were recordedfor ten years.Credit: http://commons.wiki-

media.org/wiki/File:ARM_facil-

ity_Nauru.jpg


10


What Dudley and his colleagues dis-cerned was that pHARs are heavilyenriched with schizophrenia-associ-ated loci. Moreover, they discoveredthat the schizophrenia genes associ-ated with pHARs seem to be under astronger selective pressure than otherschizophrenia genes. In the short term,their analysis points to a prioritizationof genomic targets for treating schizo-phrenia; knowing that pHAR-asso-ciated schizophrenia genes are morelikely to persist in the gene pool speaksto them being lucrative targets goingforward. In the long term, similar evo-lutionary history based analyses couldbe applied for other neuropsychiatricdisorders like autism and bipolar dis-order. But for now it remains humblingthat, within the complex genetic under-pinnings of all that makes us human,lie lurking the crippling seeds of mentalillness.

—A. BhandarkarSource: http://www.sciencedaily.com/releases/2015/02/150224220013.htm

The U.S. Defense Advanced ResearchProjects Agency (DARPA) has set

out to automate cancer research by2017 with the launch of the $45 millionprogram Big Mechanism last summer.

Big Mechanism aims to develop com-puter systems that will integrate infor-mation from research papers and com-puter models of cancer mechanismsin order to frame new hypotheses forcancer development. According to theprogram’s manager Paul Cohen, itsgoal is to help scientists deal with thecomplexity of cancer in a time whereresearch has become highly special-ized. If Big Mechanism succeeds, itsmethodology could be applied to com-plicated systems in fields ranging fromclimate science to military operationsto poverty. Currently, though, the pro-gram’s focus is on untangling complexcancer pathways caused by mutationsin the Ras gene family, which accountfor about one third of human cancers.To do so, DARPA is developing systemsthat will read the literature on Ras-driven pathways, identify and extractuseful information, and convert thisinformation into formal representa-tions that can then be integrated intomodels of Ras pathways. As of January,Big Mechanism’s best-performingmachine reading system extracted40% of useful information from sixparagraphs of text and correctly deter-mined how it related to rudimentarymodels of Ras-driven cancer pathways.According to Cohen, this is an excel-lent start. He is confident that, oneway or another, Big Mechanism willpay off. “DARPA seeks revolutionarytechnology,” he said. “Sometimes thosetechnologies are turned into practice.

Sometimes, they show the world whatis possible.”

—J. SwitzerSource: http://news.sciencemag.org/math/2015/01/darpa-sets-out-automate-research

An increase in wealth may havecaused the transition from ritual-

to morality-based religions, proposesa new study from the École NormaleSupérieure (ENS) in Paris. NicolasBaumard, a psychologist at ENS,explains that it wasn’t until around500 B.C.E. that new religions focusingon morality, self-discipline, and asceti-cism appeared in Eurasia. Before thattime, religion was largely focused onrituals and short-term rewards. Usinghistorical and archaeological data,Baumard and his team have shownthat this transition— also known asthe ‘Axial Age’— coincided with a risein affluence throughout Eurasia. Aspeople became increasingly wealthy,their comfortable living allowed themto shift their focus from short termgoals—for example a good grow-ing season—to longer term consider-ations such as what would happen tothem after death. Not coincidentally,said Baumard, the new religions thatbegan to appear during the Axial Age—Stoicism, Jainism, and Buddhism—and their successors, Christianity andIslam, promoted self-discipline andshort-term sacrifice while promising apleasant afterlife for the faithful. Usinghistorical and archaeological data fromacross Eurasia during the Axial Age,Baumard and his colleagues found thatthe best predictor of the emergenceof one of these “moralizing religions”was an affluence index called “energycapture”: the number of calories worthof food, fuel, and other resources avail-able per day to each person in a givensociety. Societies that had an energycapture of more than 20,000 kilocalo-ries per day had a much higher likeli-hood of seeing the emergence of a mor-alizing religion; those below 20,000

DARPA Sets Out toAutomate CancerResearch by 2017

COMPUTATIONAL BIOLOGY

DARPA’s Big Mechanism project will use computers to investigate the mechanismsresponsible for the formation of cancerous cells, such as those depicted in the styl-ized image above.Credit: Image courtesy of dream designs at FreeDigitalPhotos.net

Morality-BasedReligions May HaveTheir Roots in Wealth

ANTHROPOLOGY

11


kilocalories almost never did. SaidBaumard, “You need to have morein order to be able to want to haveless.”

—J. SwitzerSource: http://news.sciencemag.org/biol-ogy/2014/12/wealth-may-have-driven-rise-today-s-religions

The Ebola virus is just one ofmany lethal viruses that are

currently lurking and causing analarming number of deaths inAfrica. Scientists have been activelyresearching for effective ways totreat these viruses, with the goalof potentially creating vaccinesthat could eradicate them once andfor all. One particular route thathas proven to be attractive is theuse of monoclonal antibodies totarget these viruses. This possibletherapeutic treatment was the oneexplored by two scientists at theScripps Research Institute, who,after a six-year effort, managed toidentify and uncover the structureof an antibody that binds to theMarburg virus, Ebola’s cousin. Byusing a blood sample of a Marburgfever survivor, the two scientistsstudied how the antibody attacheditself to the virus, discovering thatits attachment prevented the virusfrom binding to a receptor and even-tually its replication. Further stud-ies then showed that this antibodyalso had the ability to bind to the

Ebola virus, yet -interestingly- it wasnot present in the blood of Ebolafever survivors. These findings pro-vide a template for potential newtreatments that could be used com-bat the Marburg and Ebola fever.Furthermore, the cross reactivity ofthis antibody may allow it to bemodified in such a way that wouldpermits it use in the treatment ofother viruses that share similaritiesto the Ebola and Marburg virus,giving us a tool to fight and preventpossible future outbreaks.

—S. SantiagoSource: http://www.sciencedaily.com/releases/2015/02/150226122328.htm

Nutrient-starved bacteria canlatch onto other bacterial spe-

cies in order to share their food,recent studies have suggested.Marie-Thérèse Giudici-Orticoni ofAix-Marseille University, Franceand her colleagues have found thatwhen Clostridium acetobutylicum,which metabolizes glucose forgrowth, and Desulfovibrio vulgaris,which metabolizes lactate and sul-fate, are cultured together in a glu-cose-only medium, the D. vulgarissurvives by latching itself onto theC. acetobutylicum and sharing thelatter’s cytoplasm and proteins. In aseparate study, Christian Kost of theMax Planck Institute for ChemicalEcology in Jena, Germany and hiscolleagues performed a similarexperiment with Escherichia coliand Acinetobacter baylyi. The bacte-ria were mutated so that they couldnot produce certain essential aminoacids and then cultured in a mediumlacking the amino acid required byE. coli. Similar to the observationsof the Aix-Marseilles’ team, the E.coli was found to form nanotubesof up to 14 micrometers in orderto connect to the A. balyli. The twobacterial species then exchangedcytoplasm and amino acids so thateach received the amino acid it waslacking. According to Kost and histeam, the results suggest that these

bacteria function as interconnectedentities rather than individuals.

—J. SwitzerSource: http://news.sciencemag.org/biol-ogy/2014/12/wealth-may-have-driven-rise-today-s-religions

An entire biosphere of organismswith a chemical makeup dis-

tinct from what we see on earthmay be going undetected by scien-tists, said Victoria Orphan of theCalifornia Institute of Technologyin Pasadena at the annual meetingof the American Association for theAdvancement of Science (AAAS).When searching for alien life, scien-tists typically look for features thatindicate an earth-like planet, suchas an abundance of liquid water.However, this search precludes theexistence of life forms with an unfa-miliar biochemistry— for instance,organisms that formed in liquidhydrocarbons rather than in water.

The possible existence of this over-looked “shadow biosphere” hassome scientists stressing the need toenlarge the search for life to includeindicators more general than thosewith which we are already familiar.

Unique BiochemistryMay Be ConcealingAlien Life

ASTROBIOLOGY

Ebola: SmallMolecule, BigConsequences

MICROBIOLOGY

Starving BacteriaShare Nutrients

A stylized image of an E. coli bacte-rium, one of the bacterial species thatwas observed by scientists to sharefood with other species when nutrientdeprived.Credit: Image courtesy of Victor Habbick atFreeDigitalPhotos.net

Electron micrograph of an Ebolavirion.Credit: http://www.cdc.gov/vhf/ebola/modules/flexslider/about-ebola.jpg


12


One such “universalindicator” is a systemthat is out of equilib-rium, says CarolynPorco of the SpaceScience Institute inBoulder Colorado.Because life takes inand uses energy, italters its environment.For instance, the earthwould not have suchan oxygen-rich atmo-sphere if it weren’t forthe actions of photo-synthetic organisms.Porco has called onother scientists to comeup with a working defi-nition of life in orderto give planetary scien-tists more guidance onwhat to look for. Such

a criteria would not only be useful foralien life, however; even here on earth,there may be unique life forms goingundetected by scientists. Tucked awayin themicroscopic world are the earth’smost diverse organisms, some of whichhave already called into question ourunderstanding of life. Although virusesare currently classified as nonliving,recent discoveries of giant amoeba-infectingviruseswith complexgenomesblur the lines between living and non-living. Such unexpected discoveries,according to Orphan, suggest that weshouldn’t allow our search for life beconstrained by what we already knowis out there.

—J. SwitzerSource: http://news.sciencemag.org/biol-ogy/2015/02/shadow-biosphere-might-be-hiding-strange-life-right-under-our-noses

Life on Earth is inextricablylinked to the presence of one

molecule: water. As the “universalsolvent”, water is central in enablingthe chemical reactions necessary forbiological function. In short, life aswe know it cannot exist withoutwater.

However, recently, a team at CornellUniversity has modeled a new typeof methane-based life form thatcould exist on Saturn’s moon Titan.With surface temperatures reaching292 degrees below zero Fahrenheit,Titan is covered in oceans not ofwater, but of hydrocarbons. In thesecryogenic, organic oceans, cellscould be composed of membranesmade up of small organic nitrogencompounds--which the researchershave dubbed “azotosomes” (“azote”being the Frenchword for nitrogen)--rather than phospholipids. To thesurprise of the researchers, theseazotosomes show the same sort ofstability and flexibility in conditionsfound on Titan that phospholipidmembranes show here on Earth.Using amolecular dynamicsmethod

that screened for methane-basedcompounds able to self assembleinto membranes, the team identi-fied acrylonitrile as the most prom-ising candidate to form the buildingblocks of azotosomes. Acrylonitrileis found in Titan’s atmosphere, andshowed a strong barrier to decom-position. With a strong candidate toform cell membranes identified, the

next steps will be to study the pos-sible analogs to reproduction andmetabolism in this methane-basedlife form. The team hopes to ulti-mately send a probe to Titan itselfand sample its oceans.

—C. GonzalezSource: http://www.sciencedaily.com/releases/2015/02/150227181333.htm

Astrobiology: A NewHope

Cells composed of “azotosome" membranes may exist in Titan’s hydrocarbonoceans.Credit: http://www.nasa.gov/mission_pages/cassini/multimedia/pia11001.html`

Scientists have found even the earth’s most extremeenvironments, such as this Yellowstone hot spring, tocontain life. Life on other planets however, may proveto be more elusive.Credit: http://news.sciencemag.org/biology/2015/02/shadow-biosphere-might-be-hiding-strange-life-right-under-our-noses

Features Volume 29, Spring 2015MURJ

13

MURJ


Features

14

Volume 29, Spring 2015 FeaturesMURJ

People associate many wordswith biological science: test tubes,gloves, lab coats, chemicals,bacteria, viruses, medicine, etc. Inthis list, ‘fashion’ and ‘games’ arenowhere to be found.

Current biology research is verysophisticated and very specificto relevant scientific topics. This

allows for great research butinevitably alienates those withouthigher learning in the field. Peoplethroughout MIT and the sciencecommunity are trying to amendthese gaps via different outreachprograms, including FashionDescience and Syndemic.

Fashion Descience is a runwayshow that connects researchersand fashion designers to createlooks inspired by science. Theorganizationis focusedon“fosteringcollaborations between [the] twocreative worlds and creating a newone that provides science with anew language and gives fashion

Science in the World:Fashion and Video Games

Science outreach is important for education and for invokingenthusiasm for research. Dr. Tal Danino is helping involve the

public in biology research by using fashion and video games.BY ETA ATOLIA

2014 Fashion Descience Runway Show - MIT Media Lab.Photo Credit: Sharon Lacey

15


a new source of inspiration. It isabout bringing together distinctperspectives, ideas, and developinga new mindset that has never beenexplored before.”

In Fall 2014, Dr. Tal Danino(Post-Doc with Prof. SangeetaBhatia, MIT) and Tatiana Tejedor(Carmelo Clothing) participatedin this runway competition as ateam called ‘Quorum 54’. Theirdesign, shown to the right, usedfabric printed with fluorescencemicroscopy images of bacteria-cancer cell interactions. Inconversation with Dr. Danino, heexplained that “the material wasfurther enhanced by hand-paintingpart of the fabric with greenfluorescent protein. This treatmentmakes the bacteria glow in the darkand helps visually illustrate their‘cancer-fighting’ abilities.

The second project thatDr. Danino is working on is amicrobiology video game thatfocuses on bacterial therapiesagainst cancer. The game wasinitially developed by Alex May, agraduate student at VU UniversityAmsterdam. He envisioned it as atool to help “educate and empowerpeople while explaining some ofthe mechanisms of cancer, likemetastasis and tumor formation”.The two researchers teamedup to create a real-to-life videogame that has players (pyrogenicbacteria, chemotaxic bacteria,toxic-phage etc.) and power-ups(antibiotics, genetically engineeredplasmids, etc.) to help the gamerdefeat cancer/pathogens. Thiscollaboration of Dr. Danino (U.S.)and Alex (Netherlands) has createdsomething of an internationalplatform for science outreach. Inan interview with Alex and Dr.Danino, we were able to learn moreabout what Syndemic has to offer.

MURJ: What was yourinspiration for creatingSyndemic?

Alex: “The main inspirationfor Syndemic comes from the factthat despite the multitude of gamesthat deal with destroying bad guysor blowing things up, I have neverseen a game with pathogens/canceras the actual enemy — just zombiesand aliens, etc. If people enjoyfighting enemies, then I figurewhy not eliminate something thattruly deserves it? Games typicallyhave you fighting to save the

world and our bodies are basicallyfighting pathogens all the time,saving our lives from countlessthreats. We are constantly in a waragainst disease: people need to bemore aware of how powerful ourimmune system can be. Plus, it isempowering to be able to directlyattack diseases with 'weapons',as opposed to letting some drugtake care of it. My thinking is thatby giving people a form of agencyagainst disease, it can maybe makepeople feel a little more hopeful

Dr. Danino and Tatiana Tejedor with model wearing ‘Quorum 54’dress.Photo Credit: Marta Murcia

16


about their own bodies and whatthey may be going through.”

MURJ:What is yourvisionfor Syndemic in the future?

Alex: “Our ideal goal for thegame would be to see it used as anintroduction to educating peopleabout pathogens, the immunesystem, and biotechnology. Wewould like for it to be a 'springboard'that hooks people into doing theirown research/thinking about thesesorts of topics. And while the gamemay not be entirely realistic, itcan at least make people considerwhat is actually going on inside ourbodies and hopefully allow themto have some fun doing it. I wantpeople to ask, ‘Wow, is that whatcancer actually does to our bodies?’or ‘Is that what a virus looks like?’or ‘How does the immune systemactually protect us?’. We really likethe idea of the game itself beingcritically examined by the players,and I plan to have messages thatcompare the real world to the gameworld.”

Dr. Danino: “In addition,we would like to collaborate withelementary and middle schools touse the game to help teach youngchildren about basic biology. Itcould be a used as a visual toolto help illustrate the complextopics of disease, cancer, andimmunology—similar to how TheOregon Trail video game is used forHistory classes. The video game,ideally, would generate a lastingenthusiasm for science.”

Dr. Danino’s multitudeof science outreach projects—including Fashion Descience,Syndemic—all have one thing incommon: they use visual media(clothes, videos, pictures, etc.) todemonstrate scientific topics. Dr.Danino firmly believes that “visualmedia stimulates learning andhelpspeople more readily believe andembrace new research ideas. In

addition, it can help show intuitiveconcepts without getting boggeddown with too many details.”

Over the past few years, Dr.Danino has been part of severalscience outreach endeavors—one of his earlier projects iscoolsciencevideos.com. The websitehosts interesting science videos thatfocus on current research, biologytopics, chemistry topics, activitiesfor kids, etc. It serves as an easyway to learn advanced scientificideas.

A currently developing projectof Dr. Danino’s—in addition, toFashion Descience, Syndemic, andcoolsciencevideos.com—deals withcreating porcelain plates printedwith bacterial artwork. The projectis in collaboration with artist VikMuniz. The idea is to integrateart and science in a visuallypleasing way. The plates integrateinto everyday life and can bringpositive appreciation to bacteria/microbiology.

To learn more about Dr. TalDanino’s research please visit www.taldanino.com.

Different bacteria, plasmids, toolsthat can be used to combat cancer inSyndemic.Credit: Alex May, Tal Danino

Examples of bacterial artwork printed on porcelain plates.Credit: Tal Danino, Vik Muniz

17


Professor David C. Page, M.D.,graduated from SwarthmoreCollege with a degree in chemistryand received his M.D. fromHarvard Medical School in 1984.Subsequently, Page was appointeda Whitehead Fellow and, two yearslater, was awarded a MacArthur“Genius” Grant. He joined thefaculty of the Whitehead Instituteand MIT in 1988 and is presentlyboth a professor of biology at MITand the director of the WhiteheadInstitute. In 1992, his lab mappedthe human Y chromosome, and in1997, Page was awarded an AmoryPrize for advances in reproductivebiology by the AmericanAcademy of Arts and Sciences.In 2003, his research group wasable to sequence the human Ychromosome; the same year,Page also received the Curt SternAward from the American Societyof Human Genetics. In 2005,Page was elected to the NationalAcademy of Sciences. In 2011, Pagewon the March of Dimes Prize inDevelopmental Biology. His lab atthe Whitehead Institute currentlyfocuses on the evolutionary anddevelopmental origins of sexualdimorphism in the genome andgermline to better understand theX and Y chromosomes as well asgerm cells.

MURJ: You've had a longand illustrious career as abiologist. Going back to thebeginning, what compelledyou to study chemistry as anundergraduate?

Page: I had a lot of interestsas an undergraduate. I arrived atcollege thinking that I was goingto be an environmental lawyer.As a sophomore, I was briefly aneconomics major. Then I ended upas a chemistry major. I don’t know

exactly why. I remained interestedin all of these things, but I decidedthat this world of biologicalrevolution was unfolding and waspretty interesting. I was actuallya chemistry major and a biologyminor, so I was very much in thebiological space. For me, it wasmore of a decision about whetherI would be in the natural sciencesor if I would study something likeeconomics or political science. Iremain interested in all of thesethings. I briefly considered being a

MURJ Spotlight:Professor David C. Page

Director of the Whitehead InstituteThis issue's spotlight features Professor David C. Page, Director of

the Whitehead Institute, Professor of Biology at MIT, and Investigatorat the Howard Hughes Medical Institute

BY RIYA JAGETIA, ELIZABETH BERG, JAE HYUN KIM, BRIANNA JONES

Professor David C. PageAll photos reproduced with permission from David C. Page archives

18


religion major, just to round thingsout.

I would say, the older I get, themore I realize that I’m interestedin understanding how the worldworks, at many different levels.This is partly why I’m drawn tostudying the genetic differencesbetween males and females. It’s nothard to connect that to human life.Anybody can identify with thesesorts of questions, and so I’m justinterested in how it all works. Ifwe can deal with particular littleparts of these questions, that’s verysatisfying.

If I go back to when I was anundergrad in the mid-seventies,biology had not yet becomemolecular. It was just the beginning,the early days of that pursuit.Maybe today I’d be a biology major,but chemistry seemed to me at that

time to provide a more molecularlygrounded approach to the kinds ofthings that I was thinking about.

MURJ: Studying chemistryincollegewasonlythebeginningof a very distinguished career.Canyougive us abrief overviewofyour career?

Page: I had a bit of an oddhistory. I started doing researchwhen I was an undergrad, and Idecided to apply to both M.D. andM.D./Ph.D. programs. I got intoa bunch of M.D./Ph.D. programs,and then someone told me thatto complete an M.D./Ph.D. could

take as long as six years. I thought-- that’s too long; I’m not willing tosign up for six years. So I came tothe HST M.D. program. Four years-- fine. I had done a fair amount ofresearch as an undergrad. In theHST program, you have to do someresearch to get the M.D. I startedworking in a lab the summer aftermy first year of medical school.Looking back upon it now, I realizethat, in the summer of 1979, Ibecame the first student anywherein the world to work on whatwould become the Human GenomeProject. I never heard the phraseHuman Genome Project until sevenyears later, but I was working on itin 1979. I was there, before thebeginning. It was just an incredibleopportunity when I look back uponit. Of course, at the time, it didn’tlook like anything. I spent that firstsummer working in a lab at UMassMedical School, but I was workingtogether with a professor who wasthen at MIT, David Botstein. It wasthe prehistoric days of the HumanGenome Project.

I then went back to medicalschool doing clinical rotations,more or less thinking I’d become acardiologist. But after completingmy third year of medical school,I came back to finish up my HSTthesis. That’s when I decided totake a year’s leave of absence. Iplanned to spend six months inlab and six months working in ahospital in a third-world country.One thing led to another. It tookme longer to set up the third-worldproject than planned, and a one-year leave of absence became atwo-year leave of absence.

"It was the prehistoric days of the HumanGenome project..."

19


I ended up spending threemonths working in a hospital inLiberia, where among other things,I met my future wife, who was aCanadian medical student workingat the same hospital. So, that was avery important trip to West Africa!My medical education ended upbeing six years long by the timeI was done. I paid my own waythrough medical school, and Ididn’t get a Ph.D. But, I spentmore than two years, almost threeyears, working on what wouldbecome the Human GenomeProject. I spent three months ata hospital in Liberia, where I metmy future wife.

When I graduated frommedical school, I was the onlymember of my graduating class

who did not doan internship.Instead, six weeksafter I graduated,the WhiteheadInstitute opened,and I became thefirst fellow here.Now, realize --I didn’t have aPh.D. I hadn’tdone a post-doc,but somehow Iwas given thisopportunity torun my ownlaboratory. Thirty-one years later, I’mstill here! So that’sthe short versionof my career.

I had a lot ofdecisions to makeabout what to dowith my life. Idid go to medicalschool, I didn’t doa Ph.D., I didn’tdo an internship

or residency. I couldn’t possiblyreplay that path if I had to. I wassort of making things up as Iwent. I feel incredibly fortunateto have had a series of mentorsand opportunities along the way.I always tell people to keep theireyes open and be prepared for thatwhich they had not planned. Dothings you’re interested in, andkeep your eyes open. You mightmeet your wife in Africa; you might

stumble into the Human GenomeProject.

MURJ: One of the greatestdevelopments of our time hasbeen the sequencing of thehuman genome, which youplayed an instrumental rolein. With this development,medical care now has thepotential to become greatlypersonalized. What are yourthoughts on the ethical issuesthat may follow?

Page: Actually, I was educatedin the “ethics” sphere more than20 years ago. We were in the earlystages of preparing to sequencethe human genome, so we set upthe Whitehead Institute Task Forceon Genetics and Public Policy,bringing all sorts of people tocampus to debate issues of geneticprivacy, stem cell research, etc.What I’ve witnessed since then hasbeen a tremendous generationaltransformation. I’m of a generationthat is fairly obsessed with privacy.Your generation – as best as I cantell – has no idea of what privacyis.

MURJ: As evident throughFacebook, Twitter, and thelike.

Page: Right, right – as youmight expect, I don’t have aFacebook account. But, I think theshift is great. My generation, andeven my parents’ generation, wasperhaps too focused on keeping

"I always tell people to keep their eyes openand be prepared for that which they had not

planned...You might meet your wife in Africa; youmight stumble into the Human Genome Project."

20


information to ourselves. My senseis that those days are gone.

In light of this transformation,it’s interesting how none of theethical issues have really workedout the way we’d expectedthem to. Back in the early 90’s,we were worried that with thesequencing of the human genome,insurance companies would startdiscriminating against peoplebased on their genotypes.

MURJ: As in the movieGattaca?

Page: Yes! Contrary to ourworries, such discrimination hasn’treally happened. We realize thatgenetic determinism has its limits,as in we’re not just the resultsof our DNA sequences, but alsoof our environments, lifestyles,etc. Obviously, there are caseslike cystic fibrosis or sickle cellanemia where knowledge of theunderlying DNA sequence providesstrong predictive information, butfor most of us, it’s not nearly sosimple. As we heard PresidentObama mention in his most recentState of the Union address, the eraof Precision Medicine may addressthese individual complexities.

Going back to the question:Unlike 20 years ago, I’m not asconcerned about privacy issues.Through my own experience, I’vecome to understand that public

policy concerns tend to morphand change over time. I thinkthe questions now revolve morearound: How do we feel aboutaltering our genomes? Whichreminds me of a New York Timesarticle that I read last week on thecall for a moratorium on CRISPR-Cas 9 editing of human germ linegenomes. The first person quotedin the article, David Baltimore,was actually the founding directorof the Whitehead Institute and theperson who hired me in 1984.

Moving forward, I think we’regoing to have an endless streamof new ethical issues to consider.These issues will reinforce the needfor all of us to continue becomingmore broadly, as opposed tonarrowly, educated. For we arenot dealing with strictly scientific

questions, but rather, questionsthat require us to fall back on ourunderstanding of how the worldworks and how it should work.

MURJ: A lot of thediscussion ties back toknowledge, specificallyof educating peopleabout scientific, ethicalconsiderations. What is yourtake on science educationand/or communication forthe general public?

Page: I believe it’s the right,privilege, and responsibility ofpeople who have been so luckyas to receive an MIT education tothemselves become educators ofeveryone with whom they comein contact. It may be difficultfor students to appreciate justhow privileged they are – tobe surrounded by incredibleclassmates, taught by fantasticprofessors, all in an environmentwhere exciting things arehappening a mere 100 feet away.And to realize that they’re just atiny, tiny fraction of the Americanpopulation, of the world’spopulation, that gets to experience

"Moving forward, I think we’re going to havean endless stream of new ethical issues toconsider....questions that require us to fall

back on our understanding of how the worldworks..."

21


first-hand what the debates are allabout. So I’d like to take your questionand turn it into a call to action on ourpart to share the knowledge we’vebeen so fortunate to receive.

Because in my mind, there’s nobound to where education startsor stops. There should be constantcommunication of knowledge. Atthe time when I started training asa scientist, many of my colleaguesthought we had no particularresponsibility to concern ourselveswith ethical issues or public policydebates. Today, I doubt anybody feelsthat way. I think scientists now realizethat it’s our right and responsibility toengage with these sorts of issues – that,in fact, the future of the world dependsupon our conscientious engagement.

Back in 2013, during my veryfirst 7.016 lecture, I tried to take thestudents back to Cambridge in 1976,when the Cambridge City Council washolding hearings on whether to imposea 3-year moratorium on recombinant

DNA research. I love to point this outto students because almost 40 yearslater they look around and see 150tech companies based on recombinantDNA within a mile of our classroom.So who won that debate?

These issues are always going to bethere. They aren’t trivial issues, but it isour right, privilege and responsibilityto help guide and educate people inwhatever way we can.

MURJ: Going back to thetheme of education, what doyou like most about being aprofessor at MIT?

Page: It’s very simple: Thefantastic students. If you’re a teacher,your happiness is determined bythe quality of the students, and thestudents at MIT are great. They makebeing a professor a great pleasure, so Ifeel spoiled. I’m surrounded by greatstudents at all levels – undergraduates,grad students – and then I have greatfaculty colleagues. MIT students are

very smart, tend to be very idealistic,and very hard working. It’s thatcombination of idealism, sincerity,earnestness and intelligence. It’sperfect! I love that combination.

There is something also about thefeel of this campus, of this communitythat is very focused on excellence, butalso very down to earth. The realityis the students at MIT don’t have anyidea how good they are, and that’s awonderful characteristic.

MURJ: Many undergradslook up to MIT professors astheir role models, for a varietyof reasons. Who has been yourgreatest role model?

Page: I’ve had a lot of role models,but my strongest role models haveto be my parents. They had greataspirations for me. They did not havethe educational opportunities that Ihad. My mother completed 8th grade,and my father finished high school –but they taught me all that I needed

to know about how to be aperson.

I also had other rolemodels as well. In college,I did some research atthe NIH. I couldn’t find aplace to live, so the manI was working for, BobSimpson, offered me hisfamily’s basement. So Iactually lived with Bob’sfamily, and over the courseof a year and a half I waswith them for a total of12 months. Bob was anM.D.-Ph.D., and he wenton into basic science, andhe had a family, so hewas a walking, breathingillustration of many thingsfor me. He was also a

22


wonderful person. He was definitelyan important role model for me.

Another one of my role modelswas David Botstein, who I trainedwith during medical school. He wasfantastically intelligent, fantasticallycreative. He had high expectations ofthe people who were training inhis lab but he was very hands-off, so I had a lot of freedom. Heassembled this incredible groupof students in his laboratory.There are so many people whowere in that lab who are nowdistinguished professors allover the place. It was a greatenvironment, a great place totrain.

MURJ: Looking forward,what do you think the future ofbiology will be?

Page: Biology is the scienceof our time, and there is no end insight to that. It’s dangerous to make

predictions about the future, butthinking about how it has changedover my professional lifetime, it hasgrown enormously. It was not thescience of our time when I was inhigh school or college or even medicalschool. It has become that in the 30years since. I think the sky is the limit.

I do believe that even thoughbiology has found so many ways tobe applied, it’s going to be absolutelycritical that we maintain the places,and capabilities, and resources andsuch for people who want to do wildlycreative things in understanding howliving systems work. I think we’re goingto discover that. I think that we’re

going to have a lot more transformativediscoveries that we can’t even imagine.These are happening with increasingfrequency, the kinds of discoveries thatcompletely disrupt our understandingof how things work. There are manylevels at which we have not come toany understanding of how they work.

MURJ: What do you thinkaresomeofthebiggestcurrentproblems that biologists face?

Page: I’ll tell you about oneof the ideas I’m obsessed with,which is that 20 or 30 years fromnow, I imagine that we’re going tounderstand at a very fundamentallevel the biological differences

between human males and females.When we look back on today, wherewe have a kind of unisex humanbiology, we will see it as an odd, quaint,uninformed time. We’ll look back onthe present as a time of primitiveunderstanding.

"Biology is the science of ourtime, and there is no end insight to that...I think the sky

is the limit."

23

MURJUROP

Summaries

UROP Summaries Volume 29, Spring 2015MURJ

24

Volume 29, Spring 2015 UROP SummariesMURJ

Simulation of BloodMicrocirculation and Study ofSickle Cell DiseaseJean-Pierre Theron1, Raymond Jansen1, Wesley Harris2, Francois Le Floch31Student contributor, Class of 2015, Department of Aeronautical and Astronautical Engineering, MassachusettsInstitute of Technology (MIT), Cambridge, MA 02139, USA2Department of Aeronautical and Astronautical Engineering, Massachusetts Institute of Technology (MIT),Cambridge, MA 02139, USA3MIT Alumnus of the Department of Aeronautical and Astronautical Engineering, Massachusetts Institute ofTechnology (MIT), Cambridge, MA 02139, USA

Abstract

Sickle cell disease is a genetic disease that affects thered blood cell behavior of a great number of people. In aneffort to help the development of a cure, Dr Francios LeFloch previously developed a sophisticated mathematicalmodel that simulates the flow and oxygen diffusioncharacteristics of capillaries. The aim of this project is toreplicate the results that Le Floch acquired as well as toexpand the calculated model by calculating intermediatedata points. Through the use of Le Floch’s model apreliminary set of three data points were calculated, ofwhich the blood flow properties were consistent withthose of Le Floch. The oxygen properties, however,did not converge by the end of the project nor were allthe data points calculated. The research project willtherefore be continued.

Background

Sickle cell disease is a form of Anemia where, upondeoxygenating, the red blood cells (RBC) deform intocrescent-shaped cells. This is in part due because of thepolymerization of the hemoglobin and the subsequentformation of long stiff fibers in the cytoplasm (Fiberformation, 2004). These sickle cells then affect the flowproperties of the blood flow by, for example, blockingcapillary vessels, which can then lead to severe pain anddamage of organs.

In an attempt to help research for a cure, Dr FrancoisLe Floch, as part of his doctoral research and inconjunction with Prof Wesley Harris of the Department

of Aeronautics and Astronautics, created a mathematicalmodel that simulates the flow characteristics of sicklecell blood flow in capillary vessels. Through the use ofnumerical methods like the Immersed boundary method,the model calculates a large set of flow and oxygendiffusion equations (which include the likes of the NavierStokes equations) to get the necessary results in terms ofoxygen transfer, blood plasma dynamics, the mechanics ofred blood cells and many more physiological constraints(Le Floch, 2010, pp. 75-76). The hope is that researcherscould use this model to test treatment options since theycan now get an estimate of how the flow properties areaffected by their proposed treatment (Le Floch, 2010,pp. 160-176).

The aim of this research project was thus to modeldifferent case studies of different flow conditions. Thisincluded replicating the results of Dr Le Floch’s PhDthesis and modelling completely new scenarios.

Method

The case studies that were analyzed were groupedtogether under different overall scenarios. Thesescenarios were distinguished from each other basedon their specific flow parameters, for example differenthematocrits or fixed inlet blood flow velocity. The casestudies themselves were then distinguished based ononly one parameter, capillary diameter. The range ofcapillaries that were analyzed started at a diameter of10 µm and goes down 5 µm. In Le Floch’s research,data points were analyzed at intervals of 0.5 µm. In the

25

UROP Summaries Volume 29, Spring 2015MURJ

research project intermediate data points that fallhalf way between Le Floch’s data points (Le Floch,2010, p. 103) were also analyzed. Hence intervalsof 0.25 µm were analyzed during the researchproject.

The approach that was taken consisted offeeding predetermined flow conditions (gainedfrom research done by Le Floch) as input tothe model. The model would then run untilthe calculations converge. Convergence wasachieved when the red blood cell geometry hasn’tchanged by more than 1%, with the two points ofcomparison being at least twice the spatial periodapart (Le Floch, 2010, p. 63). The spatial periodwas indicative of the distance traveled by the RBCin the capillary and was defined as:

where L is the spatial period, VRBC is the RBC volume,Hct is the Hematocrit and R is the capillary radius.After a set of calculations had finished for a particularvessel diameter, the resulting red blood cell geometrywas used as input to the subsequent case study witha smaller capillary diameter. This process was thenrepeated for the entire range of capillary diameters.

Results

By the end of the research project, the calculationsfor all the case studies across the various scenarioshad not been completed. Preliminary results,however, did look promising as the blood flowproperty results of three case studies (10 µm, 9.5 µmand 9.0 µm capillary diameters) in a normal bloodflow scenario, where inlet blood flow velocity wasfixed, were replicated. The calculations convergedwithin the 1% margin that was set before hand andthe error between the newly acquired data and thedata that was acquired during Le Floch’s researchwas lower than 5%. The new intermediate data point,which was calculated, also fitted within the previouslyestablished trends, which validates its values.

Figures 1 to 3 provide a visual representation of howthe original data and calculated data compare. Figure1 shows how the replicated/calculated dischargehematocrit (the ratio of the volume occupied bythe RBCs to the total volume of the blood flow)compares to the original data calculated by Le Floch.Figure 2 plots RBC velocity (the velocity at which theRBC travel in the blood stream) over the capillarysizes (graphs A). Figure 3 plots apparent viscosity

Figure 1. Discharge haematocrit at steady state vs. capillary vesselradius for the original data of Le Floch and the replicated/extendeddata.

Figure 2. RBC velocity at steady state vs. capillary vessel radius forthe original data of Le Floch and the replicated/extended data.

Figure 3. Apparent viscosity at steady state vs. capillary vessel radiusfor the original data of Le Floch and the replicated/extended data.

26

Volume 29, Spring 2015 UROP SummariesMURJ

(resistance to shear deformation of a non-Newtonianfluid) of the various capillary radii

Additionally a graph of available data (Le Floch, 2010,pp. 111-112) of a scenario that studied blood flow witha lower oxygen partial pressure (PO2 = 40mmHg) wasplotted on the three figures so as to give an idea of howthe scenario that was studied compared to other similarscenarios. On Figures 2 and 3 additional available data(op. cit.) of another scenario, which used a constant inletpressure boundary condition, were plotted (graphs B) toshow how a change in boundary conditions affects thedata.

Despite the convergence of the blood flow properties,the blood oxygen properties of the three case studies hadnot converged and therefore these calculations will have tobe continued.

Future Work

As stated before, there are still a lot of case studies thatneed to be run. Furthermore the calculations for the bloodoxygen properties of the previously mentioned case studieswill also be continued to achieve convergence. The hopeis also that the additional data points that are generatedwill greatly help expand the model to make it even morerepresentative of the blood flow dynamics of sickle cellpatients. Ultimately, however, the expectation is that thisexpanded model will help medical researchers in their goalof developing a drug or technique to help those who areafflicted with sickle cell disease.

References

Fiber formation. (2004, March 05). Retrieved November03, 2014, from About Sickle Cell disease: http://www.sicklecellinfo.net/fiber_formation.htm

Le Floch, F. T. (2010). Design of a Numerical Model forSimulation of Blood Microcirculation and Study ofSickle Cell Disease (Unpublished doctoral dissertation).Massachusetts Institute of Technology, Cambridge,United States of America.

27

MURJReports

Volume 29, Spring 2015MURJReports

28

Volume 29, Spring 2015 ReportsMURJscale, while the analogous European communityconnections correspond to neighboring nations andcontiguous regions (Cerina et al., 2014).

Furthermore, Figure 1 demonstrates that Europeancountries' community networks are transnational andtend to exhibit large clustering; communities are veryclose to one another and are at most linked with theirdirect bordering countries. In other words, the probabilityof an Italian knowing someone from Denmark is a lotlower than for the case of a Californian resident knowinga person from Massachusetts. This illustrates theimportant insight that communities in the US are wellspread and well connected as compared to the tightlypacked, but less connected, European communitiesthat are dependent upon geographic proximity andcontiguity. Furthermore, although the gravity modelanalysis is independent of the nodal connectivity, thephysical borders play an important role in the networkvisualization and the resulting inferences regarding thediversity of the MIT student population.

1.2 The Gravity Model

To build a model that uses country data as inputsand returns a prediction of the number of students atMIT from each given country, we modify the traditionalgravity function that is often used in transportation

Figure 1. Normalized distance distribution for Europe(blue dots) and the US (red dots) and their best fits (Cerinaet al., 2014). The connectivity within the US follows powerlaw decay with respect to distance while Europe exhibitsan exponential decay.

Gathering Real World Data toPredict Admission StatisticsNoor Khouri1, Dimitrios Pagonakis2, Marta González31Student Contributor, Class of 2015, Department of Civil Engineering, MIT, Cambridge, MA 02139, USA

2Student Contributor, Class of 2015, Department of Civil Engineering, MIT, Cambridge, MA 02139, USA

3Human Mobility and Network Lab, Department of Civil Engineering, MIT, Cambridge, MA 02139, USA

In this paper we design a model that estimates the incoming flow of international graduate andundergraduate students at the Massachusetts Institute of Technology (MIT). The authors, bothinternational students at the university, were motivated to explore this question due to theirdiverse cultural backgrounds. The analysis uses a modified version of the Gravity Model usedin network theory and estimates the number of students coming from each country, given onlythe country's data, namely the distance from MIT, the Gross Domestic Product (PPP) per capitaand the population size. The model parameters were optimized using student data from the MITInternational Students Office (ISO) for the 2013-2014 academic year, but could be further refinedby using data from previous and later years. This model could be proposed to any InternationalStudents Office in the US as a means of diversifying the respective university's studentcommunities and to provide an aid for setting country specific admission quotas.

1 Introduction

TheMITstudentbodyconsistsof11,300undergraduateand graduate students. Among them, approximately3,100 are international students: 400 undergraduatesand 2,700 graduate students. For the 2013-14 academicyear, these students come from 110 different countries ofvarious cultures and backgrounds. Given that there are192 countries, it may appear as though the MIT studentbody is a relatively broad representation of the world, butis this the case? Are countries with financial or politicalproblems underrepresented in the MIT community?

The analysis presented in this paper addresses thesequestions by proposing a Gravity Model approach topredict the total expected number of MIT studentsfrom a given country based on parameters such as thedistance of a country to MIT and economic measures,including population size and GDP per capita. Usingnetworks as a visualization and data analysis tool, weconstruct two parallel networks: the World Network andthe MIT International Students Network, wherein eachnode represents a country that is linked to its borderingcountries. The World Network consists of 192 nodes(countries) and 300 edges (political borders), whereasthe MIT Network consists of 110 nodes and 146 edges, asubset of the World Network.

The following section describes the construction of thenetworks, including the details of each node attribute, inconjunction with the development of the model as apredictive tool.

1.1 Bordering Networks

Borders across countries tend to provide a naturalburden for community network expansions (Cerina,Chessa, Pammolli, & Riccaboni, 2014). In other words,the formation of network communities is significantlyinfluenced by country borders and the number ofneighboring countries; this property also manifestsitself in the tendency for international students at MITto form clusters with others from nearby nations. Assuch, the empirical networks we construct are visualizedby connecting each country (nodes) with its nearestneighbors.

The motivation to connect the nodes with linksrepresenting political borders is further justified by theanalysis conducted by Cerina et al (2014). Figure 1 plotsthe normalized distance distribution of the links betweencountries in Europe and between states in the US.The observed rapid decay in the European probabilitydistribution suggests that the spatially embeddednetworks that form in the US span multiple states acrossthe whole country without any characteristic geographic

29

Reports Volume 29, Spring 2015MURJscale, while the analogous European communityconnections correspond to neighboring nations andcontiguous regions (Cerina et al., 2014).

Furthermore, Figure 1 demonstrates that Europeancountries' community networks are transnational andtend to exhibit large clustering; communities are veryclose to one another and are at most linked with theirdirect bordering countries. In other words, the probabilityof an Italian knowing someone from Denmark is a lotlower than for the case of a Californian resident knowinga person from Massachusetts. This illustrates theimportant insight that communities in the US are wellspread and well connected as compared to the tightlypacked, but less connected, European communitiesthat are dependent upon geographic proximity andcontiguity. Furthermore, although the gravity modelanalysis is independent of the nodal connectivity, thephysical borders play an important role in the networkvisualization and the resulting inferences regarding thediversity of the MIT student population.

1.2 The Gravity Model

To build a model that uses country data as inputsand returns a prediction of the number of students atMIT from each given country, we modify the traditionalgravity function that is often used in transportation

applications to forecast travel demand. This gravityfunction, which is analogous to Newton's theory ofgravity, measures the attraction between two massesin terms of their population sizes and their separatingdistance. More specifically, the generic gravity definitionis given below.

Definition 1 Let locations i, j have populations Pi andPj, respectively, and let C(i,j) represent the travel costfriction factor between them. Note that C(i,j) dependson the distance ri->j between i and j. We define

to be the gravity from origin i to destination j, withfitting parameters a,b,c,dЄR .

To estimate the size of a given country's communityat MIT, it is clear that its financial state would affect thechance of its students being able to study at MIT. Hence,for our model we redefine the travel cost friction factor toaccount for both the distance and GDP (PPP) per capitaof each country relative to the US. Thus, our model ofgravity will predict the population of students at MITfrom country of origin i by the formula

Pi is the population of country i, ri->MIT is the distancefrom country i to MIT, and GDPi is the Gross DomesticProduct (PPP) per capita of country i. The parameters α,β, γ, δ are fitting parameters which shall be determinedby optimizing the model using the empirical data for thenumber of students from each international country thatis represented at MIT.

Note that this gravity model assumes that thesample of the population from which students canbe selected for admission to MIT have fulfilled theacademic requirements and have attained the necessaryqualifications. Prior to optimizing the model with theactual MIT student data, we present the sources andorganization of data used to perform this analysis.

1.2 Data

In order to construct the MIT and World networks andto develop the gravity model for predicting the number ofinternational students from a particular country whostudy at MIT, we began by collecting the necessary datafrom a variety of sources.

The international students' data for the 2013-2014academic year, categorized by country, was obtainedby contacting the International Students Office at MIT

Figure 1. Normalized distance distribution for Europe(blue dots) and the US (red dots) and their best fits (Cerinaet al., 2014). The connectivity within the US follows powerlaw decay with respect to distance while Europe exhibitsan exponential decay.

30

Volume 29, Spring 2015 ReportsMURJ

(Cheng, 2014). The country population data analyzed inthis paper was extracted from The World Bank (“Data -Population, total,” 2013). The GDP (PPP) per capita ofeach country was derived from the CIA World Factbook(“The World Factbook - GDP Per Capita (PPP),” n.d.),while the distance of each country to MIT was estimatedimplementing the Haversine formula (Sinnott, 1984).This formula outputs distance in terms of the longitudeand latitude of two points assuming that the Earth is aperfect sphere (which is a much better approximationthan Euclidean formula).

Thus, each country in both networks was assignedwith the following unique node attributes:

•Country name (Label)•Number of undergraduates at MIT•Number of graduates at MIT•Total number of students at MIT•Total population•GDP (PPP) per capita•Average longitude•Distance from MIT (using the Haversine formula)•Predicted number of students at MIT (based onresults of gravity model)As previously mentioned, the connectivity of the

networks was defined based on countries' borders. Thebordering country data was obtained from the CIA WorldFactbook and the links were created via an adjacencymatrix (Barabási, 2012, p. 31; “The World Factbook -Land Boundaries,” n.d.). In the next section we presentthe methods of analysis of the gravity model.

2 Methods and Results

Upon optimizing the gravity model using the existinginternational MIT students' data, we were able todetermine the fitting parameters necessary to predict thenumber of incoming students from a particular countryfor the upcoming academic year. This section evaluatesthe strength of the correlation between the distanceof a country to MIT, the country's population and itsGDP per capita and the number of students within theMIT international student body who originate formthe country in question. Finally, conclusions are drawnregarding the relative impacts of each parameter onthe number of students attracted to MIT from a givencountry.

2.1 Distance of a Country to MIT

Following the generic gravity model, we first exploredthe relationship between the distance of a country from

MIT and the number of students from that country whoare currently studying at MIT. This was primarily doneto gauge whether or not the distance of the country inquestion would be a good indicator for the number ofstudents it could send abroad to pursue their highereducation at MIT. Figure 2 shows the results of themodel predictions for the number of students that couldbe admitted to study at MIT plotted versus the distancefrom each country to MIT.

The predictions were derived from the node attributesof the countries comprising the MIT InternationalStudents Network for the academic year 2013-2014. Theraw data from the MIT Network is shown as individualdata points. As made evident in Figure 2, there is no clearcorrelation between the distance of a country to MITand the number of students at MIT of that origin. Bothdata sets fluctuate significantly and do not exhibit anyparticular trend. In fact, each box plot, representing themodel predictions for a logarithmic range of distances,covers a variety of values for the predicted number ofstudents at MIT. This large variability could be explainedby the ease of modern air transportation, which facilitatestraveling thousands of miles to an accessible location,such as Boston. As a result, the distance of the countryof origin to MIT is the least important factor thatcontributes to the inflow of new students; students who

Figure 2. The box plots show the predicted number ofstudents at MIT for given distances from MIT. Note thatthe upper and lower bounds of the boxplots show thequartiles, the minimum and maximum predicted values,while the red line represents the median and the whiskersextend to the extreme outlier data points. The raw data isshown as individual points (green), each representing aparticular country.

31

Reports Volume 29, Spring 2015MURJ

are geographically closer to MIT are, in fact, not morelikely to study at MIT than others who live further away,as illustrated by Figure 4. Indeed, this notion is justified bythe optimized fitting factor β, which was estimated to equal2.5-10-2, the smallest of the fitting parameters.

2.2 Population of a Country

Next, we examined the relationship between thepopulation of a country and the number of its studentsat MIT. Figure 3, which highlights the countries withthe largest populations in the world, and Figure 4, whichshows the distribution of the international students atMIT, can be compared qualitatively. Although the countrypopulations do not perfectly correlate with the size ofMIT student populations, there is clearly a relationshipbetween the two parameters.

In particular, both China and India dominate inboth the MIT and World Networks. Following on fromthe qualitative analysis, we then plotted the model-derived predicted number of students at MIT versus thepopulation size of each country. The model results and the

raw data (individual cyan data points), shown in Figure5, are closely matched and exhibit an increasing trend.Figure 5 clearly shows that there is a direct correlationbetween the population size of a country and the numberof students originating from the country in question. Thiscan be rationalized by the fact that the larger a country'spopulation is, the more likely it is to send students toMIT given the larger sample size of potential qualifiedapplicants, and thus the expected number of studentsthat can be admitted increases. It is worth noting,however, that according to the model we developed,certain countries are either over-represented (raw datapoints lie above model predictions) or under-represented(raw data points lie below model predictions). Forexample China, which has the largest populations in theMIT and World Networks, still has an over-representedpopulation at MIT according to our model. Furthermore,we found the fitting parameter of the population α, tobe 6.4*10−1, the largest of the fitting parameters in thegravity model. Hence it represents a best estimate of thenumber of students at MIT compared to the other inputparameters considered in the proposed model.

Figure 3. This visualization rep-resents the empirically constructedWorld Network wherein the coun-tries (nodes) are ranked by bothcolor and size based on the sizes oftheir respective populations. Thered nodes, representing China andIndia, have the largest populationsrelative to the other countries inthe world. Green nodes, such as theUS and Brazil represent the nexttier of large population sizes, whilethe gray nodes have comparativelysmaller populations

Figure 4. This visualization rep-resents the empirically constructedMIT Network wherein the coun-tries are ranked based on the sizesof their respective populations.China, shown in red, has the largestinternational student populationat MIT, while India, Canada andSouth Korea, shown in orange, alsoexhibit significant representations.

32


2.3 GDP (PPP) Per Capita

Given that Oct4 expression is a key property of naïveplFinally, in order to account for the financial situation ofa country and to determine its effect on students' abilitiesto study at MIT, we examined the distribution of GDPper capita across all countries in the world; the results areshown in Figure 6. According to the visualization, if GDPwere to significantly impact a student's ability to studyat MIT, countries in Europe and certain regions in Asiawould be more representative within the MIT internationalstudent body.

To test the effect of finance on the size of a country'sstudent population at MIT, we proceeded with a quantitativeanalysis in the same manner as in the previous sections.The number of students at MIT was plotted versus the GDPper capita for both the raw data and model predictions.Comparing the predicted population of our model with theactual data, we see that there is a close correlation betweena country's financial status and the number of its studentsthat receive higher education at MIT. This naturally makessense since private institutions like MIT are not easily

accessible to developing nations. The outliers of the modellocated above the predicted data represent countries of lowGDP that potentially sent students to MIT thanks to theavailability of financial aid. Note though that even thesecountries have an acceptable GDP, which presumablymeans that its students are educated enough to be awareand capable of going through the university and financialaid application process. However, students from LeastDeveloped Countries rarely have this opportunity. Theoutliers located below the model data tend to lie on thehigh side of GDP and represent countries where studentsdo not have any incentive to study at a US institutiongiven that they have access to established universities intheir countries (i.e. European countries, Singapore, Japan).Additionally, pursuing higher education in their countriescould be less of a financial burden due to reduced tuitionfees for nationals, as in UK. Finally, this could perhaps bejustified further for countries such as Qatar, which actuallyhas the highest GDP in the world that might not have aculture of sending students to study abroad. In any case, therelationship between GDP and the country's population atMIT is still notable, as evidenced by its fitting parameter γ,which was 6.0*10-1.

Figure 5. The box plots show the gravity model predictednumber of students at MIT for given country population sizes.The model predictions closely match the raw data and exhibitpositive correlation with the number of international studentsfrom a particular community at MIT. The population of a countryappears to be the best measure of the MIT student populationfrom the parameters considered in the proposed gravity model.

Figure 6. This visualization rep-resents the empirically constructedWorld Network wherein the coun-tries are ranked based on theirGDP (PPP) per capita. Larger, rednodes represent countries with ahigh GDP, whereas smaller, graynodes represent countries with alower GDP. Countries with highGDP tend to cluster in NorthAmerica, Europe, the Arabian Gulfand within certain portions of theAsian continent.

33


2.4 Gravity Model Results

The Gravity model was quite successful in accuratelypredicting the approximate number of students at MITfrom each country in the world given its demographicdata. The gravity formula can be further enhanced byadding additional parameters and existing data to improveits accuracy; suggestions for such variables are discussedin the concluding remarks. Given only 110 existing datapoints from the MIT Network, the model's estimates arequite impressive. Figure 8 shows the number of studentsfrom each country, as predicted by the model, while Figure4 illustrates the actual data for the 2013-2014 academicyear; the great contributors, such as China, India andCanada are common between the two visualizations,however, the model predicts larger student representationfrom European countries. In addition, there are severaloutliers which are either over or under-represented, forreasons the proposed gravity function does not capture,namely political and societal influences, that are perhapsmore difficult to quantify. Table 1 provides a comparison

between the raw data and model predictions for samplecountries that illustrate the model’s accuracy. Countries,such as China are over-represented according to ourmodel, while others, such as Iraq, are under-represented,most likely due to political instabilities with the US.Furthermore, Table 2 shows the variance of each of thefitting parameters that constitute the proposed gravitymodel. Evidently, population, with the smallest fittingparameter variance, has the most significant contributionin predicting the number of students within the MITinternational student community.

3 Conclusion

Through a series of qualitative analyses, which werefacilitated by the various network visualizations, inaddition to the quantitative gravity model formulation,the composition of the MIT student body has becomeclearer. Indeed, as shown by Figures 4 and 5, internationalstudents at MIT represent 110 countries of the possible 192countries comprising the World Network and thus form a

Figure 7. The box plots show the gravity model predicted number ofstudents at MIT for given country GDP (PPP) per capita values. Themodel predictions quite closely match the raw data and exhibit positivecorrelation with the number of international students from a particularcommunity at MIT. The GDP per capita of a country appears to be agood measure of the MIT student population.

Figure 8. This visualization rep-resents the predicted MIT studentpopulations, wherein the countriesare ranked based on their predictedrepresentations at MIT. The grav-ity model predicts higher represen-tations from European countries aswell as Brazil and Japan, as com-pared to the existing MIT popula-tions shown in Figure 4.

34


decently culturally diverse community. Nevertheless, theinternational student population at MIT is not an accuraterepresentation of the World Network. It is also importantto note that the most noticeable under-representation inthe MIT network are countries in the Western Africancontinent, which are also predicted to have a very lowstudent population at MIT according to the results ofour model shown in Figure 9. Furthermore, we observedthat both the population of a country and its GDP aregood predictors for the number of students at MIT, whilecountries with existing MIT students and alumni havea greater tendency of being attracted to the university.Perhaps this could be explained by MIT’s reputationamong the country's community or due to donations byalumni to the institute.

As our proposed gravity model for predicting thenumber of students that would be admitted to MIT

provides the potential for promising results, suchstudy could have many applications. For example thecollaboration between the International Students Officeat MIT and the admissions office in order to improve thediversity of its incoming international student body. Thestatistics by the model could also be used to set admissionquotas for various countries in the world.

In order to further refine the gravity model we haveused to predict the number of students from a givencountry being admitted to MIT, a number of additionalfactors would need to be introduced into the model.Firstly, it would be important to build the model basedon ISO data from several years and hence improve theaccuracy of the model. Also, since the internationalstudents data provided by the MIT ISO is categorizedaccording to the students' nationalities, we recommendthat additional student data be incorporated into the

Table 1. Note that the countries in green areunder-represented at MIT due to factors thatare not accounted in the gravity model (politi-cal, societal), while the countries in red arecountries which are over-represented at MIT.Possible reasons for this could include the factthat alumni or close relations between theInstitute and the country in question attractmore students than anticipated by the model.

Table 2. The variance table for the fittingparameters for Population (α), GDP (γ) andDistance (β). The lower the variance the tighterthe spread of the data and hence the greaterthe influence of the variable to the model.

Figure 9. This visualization repre-sents the predicted MIT student popu-lations based on the World Network,wherein the countries are rankedbased on their predicted representa-tions at MIT. The World Networkpredictions closely match the MITNetwork predictions shown in Figure8, suggesting that MIT is more likelyto attract countries that already havean established student populationat the university. Countries that aremissing in the MIT Network are pre-dicted to send few to none of its stu-dents to MIT.

35


node attributes. This is due to the fact that a student'snationality does not necessarily reflect the conditionsof the country from which the student applied. Thus,it would also be interesting to explore the effect ofadditional economic measures of development thatshed light on the strength of a country's educationalsystem, its political stability and its relations with theUS. Finally, since this model accounts for the total MITstudent population from a given country, dividing theMIT network into graduates or undergraduates couldperhaps highlight other trends that are not capturedin this simplistic version of the gravity model. Forexample, graduate students could potentially exhibitlarger concentrations of particular populations fromwhich professors at MIT have maintained collaborativerelationships with alumni originating from the countryin question.

Overall, we believe that the results presented in thispaper highlight some interesting aspects of the MITinternational students community and allow for greateravenues for future research.

ReferencesBarabási, A.-L. (2012). Network Science. Retrieved from

http://barabasilab.neu.edu/networksciencebook/Cerina, F., Chessa, A., Pammolli, F., & Riccaboni,

M. (2014). Network communities within andacross borders. Scientific Reports, 4. http://doi.org/10.1038/srep04546

Cheng, E. (2014). MIT International Student Body Data.MIT Interntional Students Office.

Data - Population, total. (2013). Retrieved fromhttp://data.worldbank.org/indicator/SP.POP.TOTL?order=wbapi_data_value_2013+wbapi_data_value+wbapi_data_value-last&sort=asc

Sinnott, R. W. (1984). Virtues of the Haversine. Sky andTelescope, 68(2), 159.

The World Factbook - GDP Per Capita (PPP).(n.d.). Retrieved from https://www.cia.gov/library/publications/the-world-factbook/rankorder/2004rank.html

The World Factbook - Land Boundaries. (n.d.). Retrievedfrom https://www.cia.gov/Library/publications/the-world-factbook/fields/2096.html

36


Improving the Efficiencyof Carbon Fixation inRalstonia eutropha withCarbon-concentratingMicrocompartmentsSophia Li1, Jingnan Lu2, Christopher J. Brigham3, Anthony J.Sinskey4,5,61Student Contributor, Class of 2015, Department of Biological Engineering, Massachusetts Institute of Technology,77 Massachusetts Avenue, Cambridge, Massachusetts 02139, USA

2Department of Chemistry, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge,Massachusetts 02139, USA

3Department of Bioengineering, University of Massachusetts Dartmouth, 285 Old Westport Road, NorthDartmouth, MA 02747, USA

4Department of Biology, 5Division of Health Sciences and Technology, 6Engineering Systems Division,Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, Massachusetts 02139, USA

Microorganisms such as Ralstonia eutropha offer renewable solutions to current global warming and carbonemission challenges with their ability to use CO2 as the carbon source for valuable products such as biodegradableplastic and biofuels. An important bottleneck in this process is R. eutropha’s carbon fixation efficiency, which islimited by the enzyme, ribulose-1,5-bisphosphate carboxylase oxygenase (RuBisCO), since RuBisCO can take bothCO2 and O2 as substrates. Cyanobacteria and some chemoautotrophs have increased the efficiency of RuBisCO byencapsulating it in carboxysomes, protein-based organelles that concentrate CO2 while blocking O2 from RuBisCO.Since these structures are not found in R. eutropha, we introduced carboxysomes into R. eutropha to improvecarbon fixation and evaluated the carboxysomes’ effect on bacterial growth and product formation. We foundthat the carbon-fixation enzymes found in carboxysomes, carbonic anhydrase and RuBisCO, had higher enzymaticactivities in strains with carboxysomes as compared to a control strain without carboxysomes. We also determinedthat carboxysomes in the R. eutropha strain H16 led to greater cell growth, but decreased product formation of thepolymer PHB when the bacteria were grown autotrophically on CO2 or formate. Thus, this study shows the utilityof carboxysomes in improving carbon fixation in nonnative microorganisms, while providing insight into potentialchallenges, such as carboxysome assembly, in increasing the yield of products that aggregate in the cytoplasm of R.eutropha.

Introduction

Climate change and the inevitable depletion of fossilfuels has led to a search for sustainable energy sources.Certain microbes, such as Ralstonia eutropha, offer asolution to these global warming and energy problems

with their ability to grow rapidly, take in CO2 as a carbonsource, and produce a wide array of useful products(Lee et al., 1999; Lu et al., 2012).

R. eutropha is a Gram-negative bacterium that isable to use renewable carbon sources such as plant oils(Budde et al., 2011), organic acids (Yang et al., 2010),

37


CO2 (Pohlmann et al., 2006), and formate (Li et al.,2012) to naturally produce polyhydroxybutyrate (PHB),a biodegradable polyester. In addition, R. eutropha hasbeen engineered to produce other valuable productssuch as isobutanol and 3-methyl-1-butanol (Lu et al.,2012), which can be used as gasoline substitutes (Atsumiet al., 2008) or as chemical feedstocks (Gogerty et al.,2010). A R. eutropha mutant strain has been shownto produce over 14 g/L of isobutanol and 3-methyl-1-butanol after 50 days of flask cultivation (Lu et al.,2012). However, increases in yield and productivityare needed to make the production of these alcoholsfeasible on an industrial scale (Lee et al., 2008).

One of the bottlenecks in R. eutropha’s abilityto produce valuable products is its carbon fixationefficiency. While R. eutropha has several carbonicanhydrases (Figure 1), a key carbon fixation enzyme,ribulose-1,5-bisphosphate carboxylase oxygenase(RuBisCO), is inefficient because it takes both CO2

and O2 as substrates (Keys, 1986). Thus, in this study,we aim to improve the efficiency of carbon fixation inR. eutropha by introducing carboxysomes, a carbon-concentrating mechanism, in the bacterium.

A carboxysome is a bacterial microcompartment thatis similar to a synthetic organelle (Yeates et al., 2008). It

consists of an icosahedral protein shell surrounding theenzymes carbonic anhydrase and RuBisCO, separatingthem from the rest of the cytoplasm (Bonacci et al., 2012).Carboxysomes are natively found in cyanobacteriaand some chemoautotrophic bacteria (Price et al.,1991), which often grow in environments with lowconcentrations of CO₂. It has been hypothesized thatthe carboxysome increases the efficiency of carbonfixation through two main mechanisms: saturatingRuBisCO with CO2 and blocking RuBisCO fromtaking O2 as a substrate for the production of wastefulbyproducts (Figure 2; Yeates et al., 2008). By increasingthe proximity of carbonic anhydrase and RuBisCO andtrapping CO2 inside the carboxysome, the specificityand efficiency of RuBisCO increases as it is saturated

with more CO2. The protein shell also shields RuBisCOfrom O2, minimizing the substrate competition betweenO2 and CO2 for RuBisCO, so less phosphoglycolate, apotentially toxic byproduct, is formed and the efficiencyof carbon fixation is increased. Through these twomechanisms (Rae et al., 2013), the carboxysome allowsthe bacteria to more efficiently fix CO₂.

Since carboxysomes have been seen to improvecarbon fixation in many bacteria due to their carbon-concentrating mechanisms, in this study we introducedcarboxysomes from the chemoautotrophic bacteriumHalothiobacillus neapolitanus into R. eutropha.We then measured the carboxysomes’ effect on the

Figure 1. Possible roles of carbonic anhydrases (red) in R.eutropha. Periplasmic Caa preferentially converts CO2 intobicarbonate. Cytosolic Can supplies CO2 to RuBisCO in theCalvin-Benson-Bassham (CBB) cycle. The roles of Cag andCan2 are less clear; Cag could play a similar role to Can whileCan2 could be involved with pH homeostasis.

RuBisCO=ribulose-1,5-bisphosphate carboxylase oxygenase

Figure 2. Carbon-concentrating mechanism of the carboxy-some. CO2 and bicarbonate (HCO3

-) enter the cell as describedin Figure 1. Carbonic anhydrase (CA) and ribulose-1,5-bispho-sphate carboxylase oxygenase (RuBisCO) are also inside theprotein shell of the carboxysome. Bicarbonate enters thecarboxysome through pores and is converted to CO2. RuBisCOthen catalyzes the carbon fixation reaction where ribulose-1,5-bisphosphate (not shown) and CO2 are converted to twomolecules of 3-phosphoglycerate (PGA). The carboxysomeshell (purple) restricts oxygen from entering and interactingwith RuBisCO to form the toxic side product, phosphoglycolate(PG). CBB cycle= Calvin-Benson-Bassham cycle.

38


enzymatic activity of the two key carbon fixationenzymes, carbonic anhydrase and RuBisCO. We alsodetermined whether the introduction of carboxysomesimproved R. eutropha’s ability to grow on carbondioxide/formate. Finally, we evaluated whether theaddition of carboxysomes led to changes in the yield ofthe biodegradable plastic, polyhydroxybutyrate, in R.eutropha.

Methods

Chemicals, bacterial strains, and plasmids

Chemicals were obtained from Sigma Aldrich unlessotherwise noted. Bacterial strains and plasmids used inthis study are included in Tables 1 and 2.Growth media and cultivation conditions

R. eutropha strains were grown in either trypticsoy broth without dextrose (TSB) (Becton Dickinson,Sparks, MD) or minimal media (Lu et al., 2012) at30°C with 10 µg mL-1 of the antibiotic gentamicin. R.eutropha strains with plasmids were grown with anadditional 300 µg mL-1 kanamycin. Carbon sourcesused in minimal media cultivation were 2% (w/v)fructose or 0.2% (v/v) formic acid.

For enzymatic assays, a single R. eutropha colonyfrom a TSB plate with 1.5% agar and 10 µg mL-1gentamicin was used to inoculate 5 mL of TSB medium

with appropriate antibiotics. After 48 hours of growth,these cultures were used to inoculate 50 mL flaskcultures of minimal media with appropriate antibiotics.Flask cultures were continuously shaken at 200 rpm ina 30°C incubator. After one day of growth, the cultureswere induced with 0.2% (w/v) of L-arabinose. Aliquotswere removed at various time points for analysis.

When R. eutropha was grown with formic acid as thesole carbon source, 1.5 mL of the TSB medium culturewas used to inoculate 150 mL of flask cultures. ThepH and the OD600 of the flask cultures was measuredperiodically. Whenever the pH of the flask cultureswas measured to be 8, formic acid was added to theflask cultures (0.1% final concentration v/v) and the pHwas adjusted back to 7. Flask cultures were inducedwith 0.5% (w/v) of L-arabinose during the exponentialgrowth phase of the bacteria.

Escherichia coli strains were grown in LB medium(Bertani, 1951) at 37°C. E. coli strains with plasmidwere grown with 50 µg mL-1 kanamycin.

Plasmid and Strain Construction

Standard molecular biology techniques as describedpreviously were used for the construction of all plasmidsand strains (Sambrook and Russell, 2001). PhusionDNA polymerase (New England Biolabs, Ipswich, MA),was used for amplification of DNA. The QIAquick gelextraction kit (Qiagen, Valencia, CA) was used to gelpurify all DNA, while the QIAprep spin miniprep kit(Qiagen, Valencia, CA) was used to extract plasmids.All restriction enzymes used were supplied by NewEngland Biolabs (Ipswich, MA).

To construct pCarboxy (Table 2), thecarboxysome operon from H. neapolitanus wasamplified through PCR. The forward primer(5’ATTACCCGGGTTGTACACGGCCGCATAATCG3’)and the reverse pr imer(5’ACTACCCGGGCTAGCTAATGCTCTGTCTCA3’)were designed with XmaI restriction sites. Afteramplification, the carboxysome operon was purifiedand digested with XmaI, forming a DNA product with afinal length of 9811 bp. The pBBad vector (Fukui et al.,2009) was also digested with XmaI. The digested vectorand carboxysome operon DNA insert were ligatedtogether and transformed into E. coli Top 10 chemicallycompetent cells (Life Technologies, Carlsbad, CA).Plasmids were then extracted and correctly constructedplasmids were found with a SalI diagnostic digest. Thecorrectly constructed plasmids were then transformedinto E. coli S17-1 (Simon et al., 1983), a donor for theconjugative transfer of mobilizable plasmids, and theplasmid was introduced intoR. eutropha via conjugationusing a standard mating procedure (Slater et al., 1998).R. eutropha with the plasmids were selected using

Table 1. Strains used in this work

Table 2. Plasmids used in this work

Table 1 continued. Strains used in this work

39

Reports Volume 29, Spring 2015MURJcolony PCR with the same primers used to amplify thecarboxysome operon.

Preparation of cell extracts for enzymatic assays

In order to prepare cell extracts of R. eutrophastrains for enzymatic assays, 10 mL of cell culture wereremoved and spun down at 4000 x g. The cell pelletswere resuspended in 2 mL of phosphate buffered saline.1 mL of resuspended cells with added 0.1 mm zirconia/silica beads (BioSpec Products, Bartlesville, OK) wereshaken five times at 5.0 m/s for 20 seconds, with 2min intervals of rest in between at 4°C by FastPrep-24(MP Biomedicals, Solon, OH). The supernatant wascollected after centrifugation for 20 min at 4°C and6500 x g for enzymatic assays. Protein concentrationswere determined by a Bradford assay with bovineserum albumin as the protein concentration standard(Bradford, 1976).

Assays of carbonic anhydrase and RuBisCO enzymaticactivities

In this study, enzyme activities reported were theaverage of triplicate assays ± standard deviation. Thecarbonic anhydrase activity assays were performed asdescribed previously (Gai et al., 2014). The RuBisCOassay is a modified spectrophotometric assay (Lan andMott, 1991). The activation buffer used as the blankfor the spectrophotometer contained 50 mM of bicineat pH 8, 20 mM of MgCl2 ·6 H2O, and 2mM DTT. Thefinal reaction mixture in the cuvette consisted of activitybuffer (5.5% v/v), 10 mM NaHCO3 ,1 mM ATP, 0.4mM NADH, 25 units of carbonic anhydrase, 5 units of3-phosphoglycerate kinase, 5 units of glyceraldehyde-3-phosphate dehydrogenase, 10% v/v cell extract and 0.2mM of ribulose-1,5-biphosphate (RBP). Everything inthis mixture except for RBP was incubated at 30°C for5 minutes before the absorbance was measured ( λ=340nm) for five minutes. After 100 seconds of measurementin the spectrophotometer, RBP was injected into thesample. NADH concentrations were calculated using anextinction coefficient of 6.22 mM-1.

Quantification of Polymer in R. eutropha

The cell dry weight (CDW) and amount ofpolyhydroxybutyrate (PHB) was measured as describedpreviously (Karr et al., 1983; York et al., 2003).

Results

Introduction of carboxysomes into R. eutropha strains

Wild type R. eutropha H16 has the ability toassimilate CO2 via the Calvin-Benson-Bassham (CBB)cycle through its possession of operons containing CBBgenes (Bowien and Kusian, 2002). However, wild-type R. eutropha does not express carboxysomes, and

carboxysome shell protein genes are not found in itsgenome (Pohlman et al., 2006). Thus, we constructedpCarboxy, an arabinose-inducible plasmid with thecarboxysome operon from H. neapolitanus (Bonacci etal., 2012) and introduced the plasmid to R. eutrophastrains H16 and Re2061 (ΔphaCAB), which lacks theability to produce PHB. The use of this plasmid allowsthe induction of gene expression of carboxysomes whenarabinose is added to the growth media. Colony PCRwas used to confirm the presence of pCarboxy in R.eutropha.

Greater carbonic anhydrase and RuBisCO enzymaticactivities in strains with carboxysomes

To investigate whether carboxysomes were beingexpressed in R. eutropha and evaluate the effect of

0

50

100

150

200

250

300

350

400

450

500

pre-‐induc1on post induc1onEnzyme Ac

*vity

(U/m

g of protein)

Re2061 Re2061/pCarboxy

A

0

10

20

30

40

50

60

70

80

90

100

pre-‐induc5on post induc5on

RubisCOen

zymeac0v

ity(U/ g of protein)

Re2061 Re2061/pCarboxy

B

Figure 3. Enzymatic activity of carbonic anhydrase (A) andRuBisCO (B) in the strain Re2061/pCarboxy, with the strainRe2061 (without pCarboxy) as the control (see Table 1). Bothstrains were cultivated in minimal media with 0.05% NH4Clwith 2% w/v fructose. All flask cultures were induced with 0.2%arabinose (final w/v) after 21 hours of growth. The no arabinosesamples were taken 1 hour before arabinose induction, and thepost-arabinose samples were taken three hours after arabinoseinduction. For the carbonic anhydrase enzymatic assay, oneEnzyme Activity Unit (U) was defined as the enzyme makesthe reaction proceed twice as fast. For the RuBisCO enzymaticassay, the units of enzymatic activity (U) were defined as mmolof NADH consumed/min. Each value represents the mean ±standard error (error bars) of n=3.

40


carboxysomes on key carbon fixation enzymes, theenzymatic activity of carbonic anhydrase and RuBisCOwere assayed in Re2061/pCarboxy, with Re2061 as thecontrol strain (Figure 3). The cultures were grown for47 hours and induced with 0.2% (w/v) arabinose at 21hours. For both carbonic anhydrase (CA) and RuBisCO,greater enzymatic activities were seen in Re2061/pCarboxy as compared to Re2061, before and afterarabinose induction (Figure 3). The highest CA andRuBisCO activities were seen after arabinose induction;3 hours after arabinose induction, Re2061/pCarboxyhad 11.4 times greater CA enzymatic activity and 2.5times greater RuBisCO enzymatic activity (Figure 3).

R. eutropha strains with carboxysomes exhibit greatercell growth on formate

To demonstrate the carboxysomes’ effect on carbonfixation in R. eutropha, we compared the growth ofH16/pCarboxy and H16/pBBad when formate was thecarbon source. Formate breaks down into carbondioxide in the media while also providing NADH,effectively providing both carbon and energy sourcesto R. eutropha. The cultures were grown for 65 hoursand induced at 20 hours of growth. H16/pBBad, whichdoes not contain the carboxysome operon, was used asthe control strain. Initially, the controls strain H16/pBBad grew to a greater optical density, but after eighthours of arabinose induction, H16/pCarboxy began

to equal and slowly overtake H16/pBBad in opticaldensity (Figure 4). After 65 hours of cultivation or 45hours after arabinose induction, the optical density ofH16/pCarboxy was 1.23 times greater than that of H16/pBBad (Figure 4).

R. eutropha wild-type strains with carboxysomesproduce less PHB under autotrophic growth

Lastly, to assess the carboxysomes’ effect on theproduction of polyhydroxybutyrate (PHB), a polymerproduced by wild-type R. eutropha, we measured theamount of PHB produced as a percentage of the celldry weight (CDW) in H16/pCarboxy with H16/pBBadas a control. Both strains were grown autotrophicallyfor 65 hours with formate as the sole carbon source asdescribed above. At both 48 and 65 hours of growth,H16/pCarboxy was found to produce less PHB thanH16/pBBad, with H16/pCarboxy producing no morethan 28% PHB and H16/pBBad producing at least 65%PHB (Figure 5). Both strains produced the most PHB at48 hours of cultivation (Figure 5).

Discussion

Carbon fixation is an important first bottleneckin R. eutropha’s ability to take in carbon dioxideas a renewable carbon source for valuable products.While R. eutropha already has the ability to growautotrophically on CO2 (Li et al., 2012), this studyaimed to improve carbon fixation in R. eutropha byintroducing carboxysomes into these microorganism,since carboxysomes have a carbon-concentratingmechanism that makes key carbon fixation enzymesmore efficient. After carboxysomes were introduced,R. eutropha strain Re2061/pCarboxy exhibitedgreater enzymatic activity of carbonic anhydrase(CA) and RuBisCO (Figure 3), providing evidencethat carboxysomes are being expressed and increaseenzymatic activities of the key carbon fixation enzymes,

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

0 10 20 30 40 50 60 70

OD 6

00

Cul1va1on Time (hr)

H16/pBBad H16/pCarboxy

Induc1on at 20 hours

Figure 4. Growth curve of H16/pCarboxy and H16/pBBadas a control on formate. Both strains were cultivated in mini-mal media with 0.05% NH4Cl with 0.2% w/v formic acid. Allflask cultures were induced with 0.5% arabinose (final w/v) at20 hours of growth. Each data point represents the mean ±standard error (error bars) of n=3.

0

20

40

60

80

100

120

48 65

%PH

B/Ce

llDryWeight

Cul:va:on :me (hr)

H16/pBBad H16/pCarboxy

Figure 5. Polyhydroxybutyrate (PHB) production in thestrain H16/pCarboxy as compared to H16/pBBad (controlstrain) on formate. Both strains were cultivated in minimalmedia with 0.05% NH4Cl with 0.2% w/v formic acid. All flaskcultures were induced with 0.5% arabinose (final w/v) at 20hours of growth. 50 mL aliquots of flask culture were takenat 48 and 65 hours of growth. Each data point represents themean amount of PHB produced as a percentage of the cell dryweight ± standard error (error bars) of n=3.

41

Reports Volume 29, Spring 2015MURJCA and RuBisCO, in R. eutropha. Carboxysomes also led togreater autotrophic growth of wild-type R. eutropha strainH16 on formate (Figure 4), implying that carboxysomesimproved R. eutropha’s ability to grow with formate asthe sole carbon source. Finally, wild-type strain H16with carboxysomes was found to produce less PHB thanH16/pBBad, the control strain, (Figure 5), implying thatan increase in cell growth on formate did not lead toa corresponding increase in polymer formation. Thedecrease in PHB production may have been caused bycompetition between carboxysomes and PHB granules forspace inside the cell. In current models of PHB biogenesis,small granules coated in protein are continually generatedin the cytoplasm before merging together (Beeby et al.,2011). Thus, the assembly of carboxysomes, which aremade up of protein shells, could have interfered with PHBgranule information. The bacteria may then had channeledmore of its carbon flux toward growth, instead of towardPHB production.

Even though PHB production was not increased,this study provides valuable insight into the potentialapplications of carboxysomes in improving carbon fixationin R. eutropha. While many studies have been donecharacterizing the mechanisms of carbon concentrationand assembly in the carboxysome, few studies have beencompleted that have demonstrated the application of thecarboxysome in heterologous hosts. Carboxysomes havebeen expressed in heterotrophic E. coli (Bonacci et al.,2012), and more recently in plant chloroplasts (Lin etal., 2014). However, carboxysomes have not been usedto improve carbon fixation and product formation inautotrophic microorganisms. In this study, carboxysomesled to higher enzymatic activities of key carbon fixationenzymes and improved autotrophic growth of R. eutrophaon formate, thus demonstrating the utility of carboxysomesin improving the efficiency of carbon fixation in anindustrially relevant microorganism, while highlightinghow carboxysome assembly can be a potential challengein increasing the yield of products that aggregate in thecytoplasm.

There are many future directions to be explored in howbacterial microcompartments like the carboxysome can beused to improve R. eutropha’s ability to produce valuableproducts. Electron microscopy can be used to illuminatehow carboxysome protein shell assembly interacts withPHB granule assembly and provide further insight in howcarboxysomes assemble in R. eutropha. In addition, theintroduction of more bicarbonate and CO2 transportersinto R. eutropha may further improve carbon fixation.Recently, overexpressing extra bicarbonate transporters ina cyanobacterial strain led to increased biomass production(Kamennaya et al., 2015), so the addition of bicarbonatetransporters in R. eutropha may enhance the carbon-concentrating mechanism of the carboxysome. R. eutrophahas also been engineered to produce branched-chainalcohols as biofuels (Lu et al., 2012), so the expression ofcarboxysomes into these alcohol-producing strains may

lead to an increase in product yield. Since these alcoholsare secreted by R. eutropha into the medium (Lu etal., 2012), it is more likely that increased carbon fluxfrom improvements in carbon fixation can be directedinto greater alcohol production without the challenge ofcompeting against intracellular granules, as with PHB.Thus, there is great potential in utilizing carboxysomes toimprove carbon fixation in R. eutropha and other bacteria,and bolster their ability to convert carbon dioxide into avariety of products.

Acknowledgments

The student contributor thanks Dr. Jingnan Lu, Dr.Christopher Brigham, and Professor Anthony Sinskey fortheir helpful discussion and mentorship; Dr. Jingnan Lufor critical review of this manuscript; Dr. Claudia Gai forassistance with enzymatic assays. This work is funded byMIT UROP Program, the Lord Foundation, MIT EnergyInitiative and Dr. Alfred Thomas Guertin ’60.

References

Atsumi, S., Hanai, T., & Liao, J. C. (2008). Non-fermentativepathways for synthesis of branched-chain higher alcoholsas biofuels. Nature, 451(7174), 86-89. Retrieved on March30, 2015 from http://www.nature.com/nature/journal/v451/n7174/full/nature06450.html

Beeby, M., Cho, M., Stubbe, J., & Jensen, G. J. (2012).Growth and localization of polyhydroxybutyrate granulesin Ralstonia eutropha. Journal of bacteriology,194(5),1092-1099. Retrieved on March 30, 2015 from http://jb.asm.org/content/194/5/1092.short

Bertani G., (1951) The mode of phage liberation by lysogenicEscherichia coli. J Bacteriol 62:293-300. Retrieved onMarch 30, 2015 from http://www.ncbi.nlm.nih.gov/pmc/articles/PMC386127/

Bonacci, W., Teng, P. K., Afonso, B., Niederholtmeyer, H.,Grob, P., Silver, P. A., & Savage, D. F. (2012). Modularityof a carbon-fixing protein organelle. Proceedings ofthe National Academy of Sciences, 109(2), 478-483.Retrieved on March 30, 2015 from http://www.pnas.org/content/109/2/478.short

Bowien B, Kusian B. (2002). Genetics and control of CO2assimilation in the chemoautotroph Ralstonia eutropha.Arch Microbiol 178:85-93. Retrieved on March 30, 2015from http://link.springer.com/article/10.1007/s00203-002-0441-3

Bradford M.M., (1976). Rapid and sensitive method forthe quantitation of microgram quantiites of proteinutilizing the principle of protein-dye binding. AnalBiochem 72:248-254. Retrieved on March 30, 2015from http://www.sciencedirect.com/science/article/pii/0003269776905273

Budde, C. F., Riedel, S. L., Hübner, F., Risch, S., Popović,M. K., Rha, C., & Sinskey, A. J. (2011). Growth andpolyhydroxybutyrate production by Ralstonia eutrophain emulsified plant oil medium. Applied microbiologyand biotechnology, 89(5), 1611-1619. Retrieved on March30, 2015 from http://link.springer.com/article/10.1007/s00253-011-3102-0

Cannon, G. C., Bradburne, C. E., Aldrich, H. C., Baker,S. H., Heinhorst, S., & Shively, J. M. (2001).Microcompartments in prokaryotes: carboxysomes

42

Volume 29, Spring 2015 ReportsMURJand related polyhedra. Applied and environmentalmicrobiology, 67(12), 5351-5361. Retrieved on March30, 2015 from http://aem.asm.org/content/67/12/5351.short

Cannon, G. C., Heinhorst, S., & Kerfeld, C. A. (2010).Carboxysomal carbonic anhydrases: Structure and rolein microbial CO2 fixation. Biochimica et BiophysicaActa (BBA)-Proteins and Proteomics, 1804(2),382-392. Retrieved on March 30, 2015 from http://www.sciencedirect .com/science/art ic le/pi i/S157096390900274X

Ducat, D. C., & Silver, P. A. (2012). Improving carbon fixationpathways. Current opinion in chemical biology, 16(3),337-344. Retrieved on March 30, 2015 fromhttp://www.sciencedirect.com/science/article/pii/S1367593112000634

Fukui T., Suzuki M., Tsuge T., Nakamura S. (2009)Microbial synthesis of poly((R)-3-hydroxybutyrate-co-3-hydroxypropionate) from unrelated carbon sourcesby engineered Cupriavidus necator. Biomacromolecules10:700–706. Retrieved on March 22, 2015 from http://pubs.acs.org/doi/pdf/10.1021/bm801391j

Gogerty, D. S., & Bobik, T. A. (2010). Formation ofisobutene from 3-hydroxy-3-methylbutyrate bydiphosphomevalonate decarboxylase. Applied andenvironmental microbiology, 76(24), 8004-8010.Retrieved on March 30, 2015 from http://aem.asm.org/content/76/24/8004.short

Kamennaya, N. A., Ahn, S., Park, H., Bartal, R., Sasaki,K. A., Holman, H. Y., & Jansson, C. (2015). Installingextra bicarbonate transporters in the cyanobacteriumSynechocystissp.PCC6803enhancesbiomassproduction.Metabolic engineering, 29, 76-85. Retrieved on March30, 2015 from http://www.sciencedirect.com/science/article/pii/S1096717615000282

Karr D.B., Waters J.K., Emerich D.W. (1983) Analysis of Poly-β-Hydroxybutyrate in Rhizobium japonicum Bacteroidsby Ion-Exclusion High-Pressure Liquid Chromatographyand UV Detection. Appl Environ Microbiol 46:1339-1344. Retrieved on March 30, 2015 from http://aem.asm.org/content/46/6/1339.short

Keys, A. J. (1986). Rubisco: its role inphotorespiration.PhilosophicalTransactionsoftheRoyalSociety of London. B, Biological Sciences,313(1162), 325-336. Retrieved on March 30, 2015 from http://rstb.royalsocietypublishing.org/content/313/1162/325.short

Lan, Y., & Mott, K. A. (1991). Determination of apparentKm values for ribulose 1, 5-bisphosphate carboxylase/oxygenase(Rubisco)activaseusingthespectrophotometricassay of Rubisco activity. Plant physiology, 95(2), 604-609. Retrieved on March 30, 2015 from http://www.plantphysiol.org/content/95/2/604.short

Lee, S. K., Chou, H., Ham, T. S., Lee, T. S., & Keasling, J.D. (2008). Metabolic engineering of microorganismsfor biofuels production: from bugs to synthetic biologyto fuels. Current opinion in biotechnology, 19(6),556-563. Retrieved on March 30, 2015 from http://www.sciencedirect .com/science/art ic le/pi i/S0958166908001420

Lee, S. Y., Choi, J. I., & Wong, H. H. (1999). Recent advancesin polyhydroxyalkanoate production by bacterialfermentation: mini-review. International Journal ofBiological Macromolecules, 25(1), 31-36. Retrieved onMarch 30, 2015 from http://www.sciencedirect.com/science/article/pii/S0141813099000124

Li, H., Opgenorth P.H., Wernick D.G., Rogers S., Wu T.Y.,Higashide W., Malati P., Huo Y.X., Liao J.C. (2012).Integrated electromicrobial conversion of CO2 to

higher alcohols. Science 335:1596. Retrieved onMarch 30, 2015 from http://www.sciencemag.org/content/335/6076/1596.short

Lin, M. T., Occhialini, A., Andralojc, P. J., Devonshire, J.,Hines, K. M., Parry, M. A., & Hanson, M. R. (2014).βCarboxysomal proteins assemble into highly organizedstructures in Nicotiana chloroplasts. The PlantJournal, 79(1), 1-12. Retrieved on March 30, 2015 fromhttp://onlinelibrary.wiley.com/doi/10.1111/tpj.12536/abstract

Lu, J., Brigham, C. J., Gai, C. S., & Sinskey, A. J. (2012).Studies on the production of branched-chain alcohols inengineered Ralstonia eutropha. Applied microbiologyand biotechnology, 96(1), 283-297. Retrieved on March30, 2015 from http://link.springer.com/article/10.1007/s00253-012-4320-9

Pohlmann,A.,Fricke,W.F.,Reinecke,F.,Kusian,B.,Liesegang,H., Cramm, R., & Bowien, B. (2006). Genome sequence ofthe bioplastic-producing “Knallgas” bacterium Ralstoniaeutropha H16. Nature biotechnology, 24(10), 1257-1262.Retrieved on March 30, 2015 from http://www.nature.com/nbt/journal/v24/n10/full/nbt1244.html

Price, G. D., & Badger, M. R. (1991). Evidence for the role ofcarboxysomes in the cyanobacterial CO2-concentratingmechanism. Canadian journal of botany,69(5), 963-973. Retrieved on March 30, 2015 from http://www.nrcresearchpress.com/doi/abs/10.1139/b91-124#.VRoatPnF8s8

Rae, B. D., Long, B. M., Whitehead, L. F., Förster, B.,Badger, M. R., & Price, G. D. (2013). CyanobacterialCarboxysomes: Microcompartments that FacilitateCO2 Fixation. Journal of molecular microbiology andbiotechnology,23(4-5), 300-307. Retrieved on March 30,2015 from researchgate.net

Sambrook J., Russell D.W. (2001) Molecular cloning - Alaboratory manual, 3rd edition. Gold Spring HarborLaboratory Press.

Simon, R., Priefer U., Pühler A. (1983). A broad host rangemobilization system for in vivo genetic engineering:transposon mutagenesis in gram negative bacteria. NatBiotechnol 1(9): 784-791. Retrieved on March 30, 2015from http://www.nature.com/nbt/journal/v1/n9/abs/nbt1183-784.html

Slater S, Houmiel KL, Tran M, Mitsky TA, Taylor NB, PadgetteSR, Gruys KJ (1998) Multiple β-ketothiolases mediatepoly(beta-hydroxyalkanoate) copolymer synthesis inRalstonia eutropha. J Bacteriol 180:1979-1987.

Yang, Y. H., Brigham, C. J., Budde, C. F., Boccazzi, P.,Willis, L. B., Hassan, M. A., ... & Sinskey, A. J. (2010).Optimization of growth media components forpolyhydroxyalkanoate (PHA) production from organicacids by Ralstonia eutropha. Applied microbiology andbiotechnology, 87(6), 2037-2045. Retrieved on March30, 2015 from http://link.springer.com/article/10.1007/s00253-010-2699-8

Yeates, T. O., Kerfeld, C. A., Heinhorst, S., Cannon, G. C., &Shively,J.M.(2008).Protein-basedorganellesinbacteria:carboxysomes and related microcompartments. NatureReviews Microbiology, 6(9), 681-691. Retrieved onMarch 30, 2015 from http://www.nature.com/nrmicro/journal/v6/n9/full/nrmicro1913.htm

York G.M., Lupberger J., Tian J., Lawrence A.G., Stubbe J.,Sinskey A.J. (2003) Ralstonia eutropha H16 encodes twoand possibly three intracellular poly[D-3-hydroxybutyrate]depolymerase genes. J Bacteriol 185:3788-3794.Retrieved on March 30, 2015 from http://jb.asm.org/content/185/13/3788.short▪

43


IntroductionThe decision to refinance mortgages is often

undermined, and unfortunately, misunderstood bymany homeowners. In particular, one overlookedconsideration is the closing costs associated withrefinancing. Modern real estate finance has 1developedan intuition for the decision, and subjects it to severalfactors, including: magnitude of closing costs, changein mortgage rate, borrowing costs, and the remainingtime of the mortgage. Interestingly enough, a decreasein mortgage rate or borrowing costs should not alwaysbe entertained with the decision to refinance. As we willsee, there exists a threshold for the time remaining onthe mortgage to validate the net present value of such adecision.

First, we develop a synthetic mortgage calculatorwith two main features. The first feature is very simple,and conveys to the homeowner the monthly paymentsassociated with a traditional mortgage. The secondfeature, however, provides the proportions of themortgage that are allocated towards the interest andprincipal payments. Upon completion of this mortgagecalculator, we will extend its features, and simulatescenarios of refinancing.

The remainder of the paper is organized as follows:an interpretation of the synthetic mortgage calculatorincludes the time-variant distribution of a monthlymortgage payment to its interest and principalcomponents will be followed by the discrete potentialmonthly savings available to a homeowner through1 In Collaboration with Morgan Stanley22 We thank Professor Peter J. Kempthorne1forhiscomments, revision,andguidance

refinancing. Here we will depict the conditions underwhich an option to refinance should be exercised.Next, we will simulate the growth and breakeven in netpresent value of a repayment option. We will see that theinitial costs are eventually overtaken by the savings indecreased monthly interest payments. The completionof the primal purpose will lead to a discussion ofimplications on contract duration, bond pricing, MBShedging as well as short-term interest rates.

Synthetic Mortgage Calculator

Here we present a synthetic mortgage calculator.Given a rate x, a term of N years, a notional of Bdollars, we will determine the monthly payment.Further, we will determine the portion of the paymentthat goes towards paying the principal and the portionof the payment that goes towards paying the interest.

The effective annual interest rate is therefore:

From the monthly payment, M, the interest expense,Ei, is equivalent to:

Where Bi is the outstanding balance. It follows thatthe remainder of the payment goes towards paying theprincipal, Ep:

On Mortgages and RefinancingKhizar Qureshi1, Cheng Su21Student Contributor, Class of 2015, Department of Mathematics, Massachusetts Institute of Technology (MIT),Cambridge, MA 02139, USA2Sloan School of Management, Massachusetts Institute of Technology (MIT), Cambridge, MA 02139, USA

In general, homeowners refinance in response to a decrease in interest rates, as their borrowingcosts are lowered. However, it is worth investigating the effects of refinancing after takingthe underlying costs into consideration. Here we develop a synthetic mortgage calculator thatsufficiently accounts for such costs and the implications on new monthly payments. To confirmthe accuracy of the calculator, we simulate the effects of refinancing over 15 and 30-year periods.We then model the effects of refinancing as risk to the issuer of the mortgage, as there is negativeduration associated with shifts in the interest rate. Furthermore, we investigate the effects on theswap market as well as the Treasury bond market. We model stochastic interest rates using theVasicek model.

44


The proportions, Xi and Xp are:

Refinancing Cash Flows with Prepayments

Let S = the monthly prepayment rate for n months.Also suppose we have the following variables in monthn without prepayment risk: Balance, Bn, Interest, In, andPrincipal, Pn. The total monthly payment in month n issimply:

with a scheduled principal portion of the monthlypayment equivalent to:

and an interest portion equivalent to:

and unscheduled principal payment in month nequivalent to:

and a remaining balance at the end of the monthequivalent to:

Through refinancing, the interest savings over time issimply:

It follows that for longer maturities, the accrued interestsavings on a mortgage are higher.

Savings

We now consider the option of refinancing such amortgage. Namely, given mortgages A and B, and arefinancing cost C, should mortgage A be refinancedto mortgage B for immediate cost C? And if it should,what is the time required to break even? We will see thatsuch a decision is dependent on factors such as: currentinterest rate, new interest rate, closing costs, and thetime planned to remain in the home.

Given a new interest rate, Xnew, the monthly payment is:

It follows that the monthly savings, Si, is simply:

The mortgage should only be refinanced if the netpresent value of doing so is positive. That is:

Figure 1. The figure above illustrates the portionsof the monthly payment that are paid towards interestand principal for a 15-year mortgage with a rate of 3.29percent. Notice that the portion of the payment allo-cated towards principal is far greater than that allocatedtowards interest.

Figure 2. The figure above illustrates the portions of the monthlypayment that are paid towards interest and principal for a 30-yearmortgage with a rate of 4.03 percent. Notice that initially, the por-tion of the payment allocated towards the interest is greater thanthat allocated towards the principal (unlike the 15-year mortgage).However, as time progresses, the payments are weighted towardsthe principal component.

45

Reports Volume 29, Spring 2015MURJTherefore, the number of months required for theinterest and principal mortgage insurance savings, S toexceed closing costs, C are:

However, there exist a few situations in whichrefinancing should not be done, even with loweredinterest rates:

• Only a short term structure for the loan remains• Prepayment penalties exceed expected savings• Mortgage rates are expected to drop further

Simulations

To develop a practical understanding of refinancing,we simulate a specific example. Suppose there is a20-year outstanding mortgage with an initial loanamount of 100,000. Further, assume that there is astate of nature in which there is a 1 percent drop ininterest rates. The homeowner is tempted to borrow atthis lower interest rate, and prepay, but should be waryof closing costs. In particular, the homeowner shouldponder the following: Given the closing costs, andthe remaining time period on the mortgage, is the netpresent value of refinancing positive? For any rationalmonetary decision, this is a strict condition. Below isa figure that illustrates the results of the simulation.Notice that due to closing costs, the initial NPVt=0, isnegative. However, 15 months into the refinancingdecision, the NPV breaks even.

Consider the very same mortgage, but withan outstanding term of less than 15 months. It followsthat with the same closing costs and base mortgagerate, the net present value of refinancing will surely benegative. That is, t<15years, NPVT <0. This is becausethe payment structure is not allotted sufficient timeto break even the costs. In scenarios such as these, themortgage should never be refinanced.

Next, we generalize our simulations to illustratethe difference in activity for a 15-year and 30-yearmortgage. In particular, with and without closing costs,we try to determine the effects of refinancing on thepayment structure for two traditional contract lengths.First, we consider an unrealistic case in which rates arelowered, and no closing fees are imposed:

The simulations above aid in the development of anintuition for mortgage activity for the following states:

• Refinancing without closing costs (15-yearmortgage)

• Refinancing without closing costs (30-yearmortgage)

• Refinancing with closing costs, λ (15-yearmortgage)

• Refinancing with closing costs, λ (30-yearmortgage)

It is clear that closing costs, depending onmagnitudemay significantly reduce the perceived benefits ofrefinancing through means of increasing the effectivemonthly payments.

Figure 3. The figure above illustrates the number of months [15]required for the net present value to breakeven during refinancing.Suppose there is a 20-year (outstanding) mortgage with an initialloan of 100,000. Although there is a 1 percent drop in interestrates, there is a remaining balance of 85812, and closing costs areapproximately 1000. The new loan issued through refinancing issimply the sum of the remaining balance and the quoted closingcosts.

Figure 4. The graph above illustrates the reduction inmonthly coupon with refinancing. Here we assume zerocosts to refinance and present the effects in both a 15-yearand 30-year mortgage. Notice that because of the unreal-istic zero costs, the effective monthly payment is reducedsignificantly. However, as time progresses, the monthly pay-ment converges. This holds for both a 15-year and 30-yearmortgage.

46


Negative Duration Risk

In most mortgages borrowers are given prepaymentoptions, and are typically exercised when interest ratesfall. Homeowners are able to borrow at these lowerinterest rates, and in most cases, effectively reduce theirmonthly payments. However, this results in a negativeconvexity for Mortgage-Backed Securities (MBS). Asinterest rates fall, the duration for Mortgage-BackedSecurities actually shortens. Inversely, an increase ininterest rates discourages homeowners from borrowingand refinancing, resulting in a positive duration. Theconsequences of such an effect are quite significant, andmay be readily observed. The equation below modelsthe effect on duration, D, with a change in yield (interestrates), y (Myers, 2011).

Notice that with an increase in interest rates,duration typically decreases. However, in the caseof MBS, duration decreases due to prepayment andrefinancing effects.

As a response to the shortening duration, marketparticipants hedge with interest rate swaps, and optto receive the fixed leg, or equivalently, terminateany floating receivables. Moreover, such an effect canresult in buying pressure on the “swap bond”, whichin turn results in temporary decrease in the swapspread. The mechanism above is opposite of traditionalbond activity, justifying the compensation marketparticipants demand for negative duration risk. Thislevel of intuition may be valuable when judging creditmarket activity.

It is critical to not only understand the effects ofduration on the credit market, but also how interestrates vary over time. In particular, we explore theVasicek model, which correctly hypothesizes the

stochastic movement of interest rates. We then extendthe model parameters to bond pricing, which is ofinterest to market participants who seek to proliferatefrom interest rate movement. Finally, we return to theidea of hedging, and why the MBS swap market is soliquid.

Vasicek Model for Interest Rates

The Vasicek Model in a continuous-time term-structure model for short rates. It follows a stochasticmean reversion process as follows (Vasicek, 1977):

where rt is the short rate, κ the mean reversioncoefficient, and dW captures a Wiener Process(Vasicek, 1977).

Implication on Bond Pricing

The price of a T-year Zero Coupon Bond may bedetermined by:

where

Once κ is determined from the Vasicek model, itmay be used to determine the resulting bond price.It is readily observable that a larger value of κ, orequivalently, a faster mean reversion process implieslower bond prices. This increase in yield translates to anarrowing in swap spreads.

Implication on Short Rate

Consider a three-month T-bill with rate rt at time

Figure 5. The graph above illustrates the reduction inmonthly coupon with refinancing. Here we assume fixed, non-zero costs to refinance and present the effects in both a 15-yearand 30-year mortgage. In particular, to account for the costs,the 15-year and 30-year mortgage rates are given a multiplierλ =1.20. Effectively, this increases the capital commitmentto replicate closing costs. Notice that due to the newly intro-duced costs, the effective difference in monthly payments hasnarrowed significantly. It is also observable that the differencenarrows more slowly with the 30-year mortgage, than it doeswith the 15-year mortgage. and the quoted closing costs.

47


t. The vasicek model, in discrete form, implies that attime t+1, the T-bill will carry rate:

where σ is the one-month volatility, Δ a change in timeof 1/12 years, and ε a small shock such that ε ~N(0,1).Note that in practice, the short rate would follow amulti-factor model.

Implication on MBS Hedging

Consider a swap spread λ-T where λ is the swaprate for a mortgage-backed security. It follows thatwith in increase in short rate rate, the swap spreadwould decrease. This is attributable to two factors. Onestraightforward reason is that the bond yield wouldincrease in response to the increase in short rate.The second reason, however, is that with an increasein duration, market participants will choose to sellduration through means of paying the fixed leg of theswap (or equivalently, sell current receivables), whicheffectively lowers the swap rate. A decrease in the swaprate and an increase in bond yield result in drasticdecreases in the swap spread. Inversely, decreasesin interest rates result in large increases in the swapspread.

Conclusion

A decrease in interest rates provides incentive torefinance mortgages. However, as we have seen, animportant consideration is the savings on the mortgagetaking into account the costs for refinancing. Only ifthe 15-year or 30-year sum of the change in monthlypayments exceed the borrowing costs should themortgage be refinanced. In particular:

• Differential changes in the interest rates shouldnot be reasons to refinance.

• There exists a minimum time T>0 for theremaining life of the mortgage only under whichit is optimal to refinance

• Closing costs are almost always non-zero, so thepreceeding conditions are therefore strict.

Refinancing and prepayment also provide sourcesof risk to mortgage issuers in the form of negativeduration. Issuers face the risk of having the sum of thetime-weighted payments reduced. Compensation tocarry such risks are likely factored into the mortgagerate offered to the homeowner depending on his/herindividual propensity to refinance or issue prepayments.

Changes in the interest rate have implications on the

MBS market, treasury bond market, and swap market.A decrease in the interest rate can lower the prices ofmortgage-backed securities, decrease the treasury bondyield, and widen the swap spread. Inversely, increasesin interest rates can increase the price of MBS, increasethe treasury bond yield, and narrow the swap spread.Such movements can be modeled through marketparticipants response in buying/selling fixed legs ofMBS swaps.

Future Considerations

In future studies, we wish to test the feasibility ofthe Vasicek model for term structure. In particular,we wish to determine the effectiveness in determiningthe change in swap spread due to negative durationhedging. We suspect that the model is insufficientfor capturing all sources of market risk, and thatmacroeconomic, systemic and credit factors will benecessary. In modern practice, market participants usemodified versions of the Heath-Jarrow-Morton (HJM)model or the multi-factor Hull-White model. Thereis also the unexplored possibility of using a modifiedversion of the Cox-Ingersoll-Ross stochastic model forinterest rates. To test the effectiveness of the model,6-month, 12-month, 5-year, 15-year, and 30-year termstructures will be used. Both the credit markets andmortgage markets are fascinating, and it remains tobe seen the effects of efficient refinancing on the assetpricing of underlying instruments.

References

[1] Oldrich Vasicek, An Equilibrium Characterisation ofthe Term Structure,. Journal of Financial Economics,1977.

[2] Brealey, Richard A.; Myers, Stewart C.; Allen,Franklin, Principles of Corporate Finance New York,NY: McGraw-Hill Irwin, pp. 50-53

stochastic movement of interest rates. We then extendthe model parameters to bond pricing, which is ofinterest to market participants who seek to proliferatefrom interest rate movement. Finally, we return to theidea of hedging, and why the MBS swap market is soliquid.

Vasicek Model for Interest Rates

The Vasicek Model in a continuous-time term-structure model for short rates. It follows a stochasticmean reversion process as follows (Vasicek, 1977):

where rt is the short rate, κ the mean reversioncoefficient, and dW captures a Wiener Process(Vasicek, 1977).

Implication on Bond Pricing

The price of a T-year Zero Coupon Bond may bedetermined by:

where

Once κ is determined from the Vasicek model, itmay be used to determine the resulting bond price.It is readily observable that a larger value of κ, orequivalently, a faster mean reversion process implieslower bond prices. This increase in yield translates to anarrowing in swap spreads.

Implication on Short Rate

Consider a three-month T-bill with rate rt at time

48


MURJ 2015 CPW Research Poster CompetitionBy Pratheek Nagaraj

For the second year in a row, MURJ has hosted a competition for undergraduate studentsfrom across all majors to present their research to the wide MIT community. This interdisciplin-ary competition takes place during Campus Preview Weekend (CPW), a time where admittedhigh-school students, colloquially termed pre-frosh, visit campus as part of their decision mak-ing to enroll as undergrads at MIT. Because of this overlap the turnout to the competition isvaried and diverse with attendees ranging from pre-frosh and their family to graduate studentsand post-docs. This year the competition featured work from seven students across six differentmajors including Biological Engineering, Mathematics, and Brain and Cognitive Sciences.

Communicating research to diverse audiences is a critical skill that is important to theresearch process. MURJ believes that this competition provides an excellent platform for manystudents to develop and hone this important skill and so we encourage MIT undergraduatesdoing research to participate next year.

This event would not have been possible without the support of many. First, MURJ wouldlike to acknowledge the support from the Biology and Mathematics departments for generouslyproviding food and funding for awards respectively. Second, thank you to the judges who kindlydonated their time to engage with undergraduate research and give feedback to the presenters.And third, MURJ appreciates the support from the Baker Foundation for its assistance withequipment.

The MURJ team is excited to make this event even better for next year and look forwardto your participating or attending! Thank you to all the presenters who shared their excitingresearch with the MIT community. Congratulations to our winners, Nelson Hall ’16 – OverallWinner, Daphne Superville ‘17 – Academic Content Winner, and Kara Presbrey ’16 – PresentationWinner.

Volume 29, Spring 2015 MURJ Reports

MURJ poster session, CPW 2015. Photo: Pratheek Nagaraj

Documents

)URP 'HDQ )UHHPDQ $ORRNDWWKHODWHVW6FLHQFH1HZVmurj.mit.edu/issues/MURJ29_full.pdf · 3 Letters Volume 29, Spring 2014MURJ May2015 DearMITCommunity, IamdelightedandhonoredtointroducetheSpring2015issueofthe