vol. 34, no. 1, 2001 - ornl review v34n1 2001.pdf · curtis boles 19 curtis boles 27. number one,...

1

2 2 4 7 8

9

10

111212131416

17181920222426

272829

Editorial: Unraveling Complex BiologicalSystems

Systems Biology: New Views of Life

Genes and Proteins: A Primer

Complex Biological Systems in Mice

Gene Chip Engineers

Searching for Mouse Models of HumanDisorders

Mouse Models for the Human Disease of ChronicHereditary Tyrosinemia

Obesity-related Gene in Mouse Discoveredat ORNL

MicroCAT “Sees” Hidden Mouse Defects

Curing Cancer in Mice

Search for Signs of Inflammatory Disease

Surprises in the Mouse Genome

Protein Identification by Mass Spectrometry

Rapid Genetic Disease Screening PossibleUsing Laser Mass Spectrometry

Lab on a Chip Used for Protein Studies

The Mouse House: From Old to New

Human Genome Analyzed Using Supercomputer

Protein Prediction Tool Has Good Prospects

Microbe Probe: Studying Bacterial Genomes

SNS and Biological Research

Accessing Information on the Human GenomeProject

A Model Fish for Pollutant Studies

Controlling Carbon in Hybrid Pop lar Trees

Disease Detectives

Vol. 34, No. 1, 2001

24

4

20

New Biology: Covering All The Bases

COVER: Biologist Madhu Dharuses the ORNL-developed MicroCATto image fat deposits in a mouse. TheMicroCAT and microarray (shown on the cover)are among the many technologies used at ORNL (and describedin this issue) to understand what genes and proteins do to makeorganisms function well or poorly. Dhar photo by Curtis Boles; covercollage by Jane Parrott.

7Curtis Boles

19Curtis Boles

27

Number One, 2001 1

Frank Harris

Editor and writer—Carolyn KrauseAssistant editor—Deborah BarnesEditorial Board—Lee Riedinger (chair),

Fred Bertrand, Eli Greenbaum,Russ Knapp, Reinhold Mann,Stan Milora, Thomas Zacharia

Designer and illustrator—Jane ParrottPhotographer—Curtis BolesPhotographic enhancement—Gail

Sweeden

The Oak Ridge National LaboratoryReview is published three times a yearand distributed to employees and othersassociated with or interested in ORNL.

Editorial office address:Building 4500-North, M.S. 6266,Oak Ridge, TN 37831-2008

Telephone: (865) 574-7183;FAX: (865) 241-6776;Electronic mail: [email protected] Review Web address: www.ornl.gov/ORNLReview/

The Review is printed in the UnitedStates of America and is also availablefrom the National Technical InformationService, U.S. Department of Commerce,5285 Port Royal Road, Springfield, VA22161

Oak Ridge National Laboratory ismanaged by UT-Battelle, LLC, for theDepartment of Energy under contractDE-AC05-00OR22725

ISSN 0048-1262

Oak Ridge National Laboratory is amultiprogram, multipurpose laboratorythat conducts research in energyproduction and end-use technologies;biological and environmental scienceand technology; advanced materialssynthesis, processing, andcharacterization; and the physicalsciences, including neutron-basedscience and technology.

Unraveling Complex Biological Systems

A new age of biology is being ushered in with the sequencing of the human and other genomes. Armed with this information about

the order of DNA bases in genes and the chromosomes on whichthey sit, scientists can now initiate the steps required to understand processes thatoccur in the human body, down to the level of the molecule. This new informa-tion promises a rich harvest: drugs targeted to our own personal DNA, earlywarnings of diseases to which we are genetically predisposed, and a more pro-found understanding of human evolution.

Now that we know where different genes are located on the 23 pairs of human chromosomes,scientists have begun to focus on what these genes do. Which genes are silent in each organ? Whichgenes are turned on, or expressed, directing cells to produce proteins in specific shapes that determinetheir functions? What are the exact structures of specific proteins that cause the body to become ill orwork well? Can drugs be made to dock with these proteins, like two jigsaw-puzzle pieces fitted together,to block or enhance their action in the body, improving overall health?

In this postgenomic era of biology, scientists are learning about gene expression and the nature ofthe expressed proteins—understanding life’s processes at the molecular level. Such knowledge couldlead to therapeutic drugs targeted to each individual’s genome, ensuring their effectiveness.

Researchers at ORNL—where the function of messenger RNA, the chromosomal basis for sexdetermination in mammals, DNA repair processes, and several important mouse genes have been dis-covered—are taking an interdisciplinary approach to unraveling complex biological systems. This ap-proach also integrates studies of mouse mutations with studies of genes and proteins, using varioustechnologies—automated DNA sequencing, biomedical imaging, microarrays (gene chips), mass spec-trometry, neutron sources, and terascale supercomputers.

As described in this issue of the Review, we are determin-ing which genes are expressed in microbes, fish, and mice duringexposures to environmental toxins. We are searching for uniqueprotein signatures of various microbes. We have already compu-tationally analyzed the human and mouse genomes, to predict thestructure of genes and proteins and make educated guesses aboutprotein function. We have one of the world’s top-ranking groupsin the area of computational prediction of protein structure. Wehave identified mouse genes that play a strong role in geneticdiseases similar to maladies that afflict humans, such as cancer,obesity, and epilepsy. We are collaborating with researchers inthe Tennessee Mouse Genome Consortium to determine which of ORNL’s 60,000 mutant mice are excel-lent models of human genetic diseases that can be used to test the effectiveness of various therapies.

Our experimental research is focused on a variety of organisms: microbes, zebrafish, hybrid pop-lar trees, and mice. Our computational research is focused on microbial, mouse, and human genes andproteins. We are now taking advantage of ORNL’s leading mass spectrometry capabilities to study pro-teins expressed by these organisms under varying conditions. The success of much of this work hasdepended—and will continue to depend—on the strength of the collaborations involving scientists atORNL and elsewhere who represent many different disciplines.

ORNL’s goal is to be a center of excellence and a resource for our understanding of (1) complexbiological systems, from the molecular to the cellular and to the organismal level, with emphasis onhuman susceptibility to becoming ill from exposure to low levels of radiation and other environmentalagents, and (2) the interactions of organisms with the environment. Innovative ways to observe andunderstand the functioning of complex biological systems will be developed and applied through ex-panded partnerships, to meet the needs of the U.S. Department of Energy.

One proposed DOE program is the “Genomes to Life” initiative. Its goals include identifying“protein machines,” the multiprotein complexes that carry out the functions of living systems; character-izing functions of microbes in their natural environments; and developing computer modeling technolo-gies to determine how complex biological systems can serve DOE’s environmental and health missions.ORNL hopes to shed light on these complicated processes through its systems biology expertise andresources, including DOE’s new Center for Structural and Molecular Biology (which will take advan-tage of the combined expertise of resident and visiting researchers in mass spectrometry, computationalbiology, and neutron sciences) and DOE’s Center for Computational Sciences.

This special issue of the ORNL Review showcases ORNL’s achievements and capabilities notonly in understanding genes and proteins but also in bioengineering developments that could improvehealth care, such as the lab on a chip, cantilever devices, and the multifunctional biochip. We also havepioneered a way to cure cancer in mice that may have applications to humans. These areas of endeavorare likely to be hallmarks of science and technology in this century.

, Associate Laboratory Director for Biological and Environmental Sciences

Barry Berven and Dabney Johnson inspectthe hoods in the new biology

laboratory at ORNL.


Number One, 2001

Cur

tis B

oles

2 Oak Ridge National Laboratory REVIEW

Systems Biology:New Views of Life


The genomes of the human,mouse, fruit fly, a worm, aweed, and many microbes havebeen mapped and sequenced.

We now have the parts lists for these organisms.We are learning that many of the parts—the genesthat direct cell machinery to produce proteins—are related, from organism to organism. Re-searchers are now trying to figure out what theseparts do in relationship to each other (systemsbiology) and how they vary among species andindividuals within each species. Then research-ers can write the operating manuals. The rewardswill be great.

“One long-term goal of this research is todevelop targeted drugs that are effective for a spe-cific disease,” says Michelle Buchanan, director

of ORNL’s Chemical and Analytical SciencesDivision. “To design these therapeutic drugs, youneed detailed knowledge about the many molecu-lar-level processes that occur within a cell.” Ac-quiring such knowledge is not an easy task.

“We frequently hear about new humangenes that play a role in cancer and diseases ofthe heart, central nervous system, and other or-gans,” says Reinhold Mann, director of ORNL’sLife Sciences Division. “Some diseases can betraced to one altered DNA base pair in a particu-lar gene. However, genome characteristics orchanges that make some people more likely toget sick involve complex, intricately timed, andbalanced interactions among a variety of genesand other signals encoded in the genome. Ourcurrent state of knowledge of how the genome is

interpreted to provide the diversity of life is ex-tremely limited.”

“The next step,” says Buchanan, “is toidentify which genes turn on to make particularproteins. Then we must identify the protein com-plexes, or protein machines, in which proteinswork together in the cell to carry out specific rolesand help perform life’s most essential functions.These protein complexes are involved in signal-ing pathways that tell cells what to do and allowthem to communicate with each other.”

Characterizing the roles of protein ma-chines in cells is the objective of DOE’s “Ge-nomes to Life” initiative. This knowledge willhelp scientists predict how cells and their geneswill respond to changes in the environment, suchas exposure to a toxin.

ORNL scientists are conducting research in functional genomics—thestudy of genomes to determine the biological function of all the genesand their products—and proteomics—the study of the full set of proteins encoded by a genome.

Genes and Proteins: A Primer

Consider a living cell, the fundamental unit of life. Each human cell contains the entire human genome—some 35,000genes. But only some genes are expressed within a specific cell, resulting in the production of specific proteins. The genesthat turn on in a liver cell, for example, are different from the genes that are expressed in a brain cell.

Under the Human Genome Project funded by the Department of Energy and the National Institutes of Health, varioussequencing tools have been used to determine the order of the building blocks of genes. Genes are made of millions ofdeoxyribonucleic acid (DNA) molecules. A DNA molecule is constructed like a spiral staircase, or a double helix. The rails ofthe staircase are made of a backbone structure of phosphates and sugars, and the steps are pairs of four nitrogen-containing bases—adenine (A), cytosine (C), guanine (G), and thymine (T). Through hydrogen bonds the two rails of thestaircase are kept together, A and T pair together, and C and G are partners (see image below). For example, a sequenceof TACAT would bond specifically with a sequence of ATGTA.

A gene is made of a unique sequence of DNA bases; it is like a message containing a unique combination of letters. Thismessage is translated into information for protein production. A protein is a folded chain of amino acids in a specific order;up to 20 different amino acids exist. The message for each amino acid within a protein is dictated by a sequence of three DNAbases. If the key word in the message provided by the gene is misspelled—say, part of it is supposed to be CATTAG but insteadis spelled CATGAG—then it will have a point mutation (G substitutes for the T that should be there). As a result of this alteredDNA base, the gene may produce a protein that has an incorrect shape so it won’t dock withanother protein (e.g., a receptor), leading to a mistake in the resulting message. In otherwords, if the message in the gene is misspelled, the protein it encodes may bewrong and its function in the body may be changed, sometimes for the worse.

It was once believed that each gene codes for a single protein. However, experi-mental and computational evidence (partly obtained at ORNL) shows that manygenes produce an average of three different proteins and as many as ten proteinproducts. Genes have protein-coding regions (exons) interspersed with non-codingregions (introns). Through “alternative splicing,” a gene’s exons can be combined indifferent ways to make variants of the complete protein.

Number One, 2001 3


Some of this knowledge will be obtainedat DOE’s Center for Structural and MolecularBiology at ORNL. Scientists at this user centerdirected by Buchanan will obtain informationabout protein interactions through mass spec-trometry, computational biology, and small-angleneutron scattering (SANS). SANS will be con-ducted using the planned Bio-SANS instrumentthat is to be completed at ORNL’s High Flux Iso-tope Reactor in 2003. Various studies of proteinsand other biological materials are also planned.These studies will be conducted using biologicalinstruments at the Spallation Neutron Source, tobe completed in 2006 at ORNL.

Research by Oak Ridge biological scien-tists is aimed at learning which genes are ex-pressed when certain microrganisms are exposedto environmental toxins or radiation. DOE is in-terested in funding microbial research partly be-cause microbes can help remediate mixed wastesites by converting toxic metals from the solubleto the insoluble state to help keep them on-site.

In addition, there is great interest in learn-ing about the proteins that are expressed by mi-crobes under various conditions. ORNL’s expertsin mass spectrometry can identify proteins by de-termining their molecular weight and amino-acidsequence and comparing this information with thatin a protein database. In this way, they can findprotein signatures and then work backwards todetermine the gene sequence coding for that pro-tein, thus identifying the gene that was expressed,say, as a result of exposure to a pollutant.

A protein can have as many as 200 modi-fications in response to the actions of other pro-teins or environmental influences. These post-translational modifications (PTM), such as theaddition of a phosphate or carbohydrate to a pro-tein, can change the protein’s activity. For ex-ample, if a PTM is present on a regulatory pro-tein A, it becomes a misshapen key that no longerfits into Protein B, preventing it from turning ona downstream gene, possibly causing miscom-munication between cells. Mass spectrometry isan excellent tool for identifying proteins that havebeen modified.

ORNL researchers seek to understand com-plex biological systems at the organism as well asthe molecular and cellular levels, says Mann. Tounderstand how hormone-mimicking chemicalscan affect development, ORNL researchers arestudying gene and protein expression in see-through embryos of zebrafish. They are also try-ing to identify the genes that enable trees to pro-duce better wood products and fuels and store morecarbon from the air. To understand the functionsof genes in mammals, ORNL researchers are de-termining which genes are expressed in the skinof mice and which mouse genes in their mutantform cause maladies also found in humans, suchas polycystic kidney disease, obesity, chronic he-reditary tyrosinemia, and epilepsy. Researchers usemicroarrays (gene chips) and computational toolsfor these expression studies. For example, our ex-perimental researchers collaborate with our com-

Yun You examines the image of a mouse embryo magnified in an optical microscope.The image is captured by a digital camera and then transferred to a computer.

Brett Pate, enhanced by Gail Sweeden

putational biology experts, whomake sense out of gene ex-pression data using super-computers. They are practicingthe discipline of bioinformatics,the study of genetic and otherbiological information usingcomputational and statisticaltechniques.

“Knowing functions ofall genes in the genome, by it-self, will not lead to under-standing the processes of a liv-ing organism,” Mann says.“The reason is the biologicalsystem’s complexity. Expres-sion of genes can be regulatedin a virtually unlimited numberof ways, depending on locationin the body, time in the devel-opment of the organism, andenvironmental conditions andexposures.

“Certain protein complexes can bind tospecific locations in an organism’s genome,thereby controlling the expression of a genesometimes far away from these binding sites. Thenumber of these regulatory protein complexes isfinite, perhaps some 10,000, but taken togetherwith the number of genes and regulatory bindingsites in the genome, there is a combinatorial ex-plosion that works against any brute force ap-proach solely based on experimental research.That is why collaborations between experiment-ers and computational biologists are so impor-tant. They are a hallmark of biological investiga-tions at ORNL.”

Computational biology researchers inORNL’s Life Sciences Division have identifiedmany genes in bacterial, mouse, and human ge-nomes and have computationally analyzed thehuman genome using an ORNL-developedgene-finding computer program. They havewritten and used assembly programs and analy-sis tools to produce draft sequences of the 300million DNAbase pairs inchromosomes19, 16, and 5 forDOE’s Joint Ge-nome Institute(JGI). They havealso analyzed 25complete micro-bial genomes(52,000 genes)and many JGIdraft microbialgenomes (1000genes/day). Theyhave predictedthe structures ofproteins (100pro te ins /day)from amino-acid

sequences using an ORNL-developed, protein-threading computer program.

The section’s programmers have writtenalgorithms and developed other tools to make iteasier for biologists to use computers to findgenes and make sense out of the rising flood ofbiological data. These data are produced in stud-ies of biochemical pathways and processes, cel-lular and developmental processes, tissue and or-ganism physiology, and ecological processes andpopulations. Through ORNL’s user-friendly Ge-nome Channel Web site, its Genomic IntegratedSupercomputing Toolkit, and the IBM super-computer at DOE’s Center for ComputationalSciences, the international biology community, in-cluding pharmaceutical industry researchers andacademics, have easily obtained genetically mean-ingful interpretations of their DNA sequences andother data. ORNL’s Web site is popular in the bio-logical community (150,000 sessions per month).

Thanks to ORNL’s interdisciplinary ap-proach to complex biology using state-of-the-arttechnologies, we believe we have the right stuffto better understand thestuff of life.

ORNL researchers use various technologies to characterize DNA andproteins. This image shows the order of chemical bases in a strand ofDNA. The bases are labeled with dyes that fluoresce in different colorswhen exposed to laser light. The sequence was obtained by gelelectrophoresis in a PE Biosystems DNA-sequencing machine at ORNL.


Using genetic engineering, gene microarrays, and computationaltechnologies, ORNL researchers are deciphering genetic variations in the

skin that lead to increased risk of disease from environmental factors.

Complex BiologicalSystems in Mice


I n a car, when the radiator springs aleak, the car engine heats up dan-gerously. You stop the car, fill theradiator with coolant, and drive the

“sick” car to the repair shop to plug the leak orreplace the radiator. The leaking radiator is likean altered, or mutated, gene. By learning that adefective radiator can make the car dangerouslyhot, you find out that a properly functioning ra-diator keeps the car engine cool enough to en-sure normal operation.

Similarly, ORNL biologists “break” genesin mice to find out their normal roles in a healthymouse. “We induce mutations in mouse genesand study the resulting disease state in mice sowe can determine what the normal versions ofthe genes do in the body,” says Ed Michaud, asenior research biologist in ORNL’s Life SciencesDivision (LSD). One way that mutations are cre-ated is to expose male mice to ethylnitrosourea(ENU), a powerful chemical mutagen discoveredby ORNL’s Bill Russell in 1979 that alters a singlebase pair in a gene. Another way is to use recom-binant DNA technology or ENU to alter a genein embryonic stem cells later inserted into mouseembryos. Or an altered gene can be inserted intoa fertilized egg, which is implanted in a femalemouse and brought to birth as a “transgenic”mouse.

“Because genes operate in complex inter-acting networks, the mutation of one gene oftenresults in an alteration of other genes in the samenetwork,” Michaud says. “Therefore, we are in-terested not only in determining the functions ofindividual genes but also how these genes inter-act with each other and with the environment.”

Michaud and other ORNL biologists arestudying complex biological systems in mice incollaboration with ORNL researchers usingmicroarrays, mass spectrometers, andbioinformatics—the discipline in which largeamounts of data are sorted into intuitive data-bases, analyzed, and presented in an understand-able form. The first complex biological systemhe has focused on is the network of genes thataffect the development and functioning of theskin. In 1992 Michaud and fellow ORNL biolo-gists identified and cloned the mouse agouti gene,

which plays a role in the development of skinand hair pigment. He subsequently identified andcloned mutant forms of the agouti gene that causeobesity, diabetes, and skin cancer in the mouse.The mouse agouti gene has a counterpart in thehuman genome.

Mice and humans each have some 35,000genes. These genes are distributed among chro-mosomes, which are long strands of DNA packedin the nucleus of each cell whose job is to deter-mine and transmit hereditary characteristics. Thehuman has 23 pairs of chromosomes and themouse has 20 pairs of chromosomes.

“We are now interested in genes scatteredamong all chromosomes of the mouse genomethat affect skin,” Michaud says. “The skin is ahighly metabolic organ with the largest surfacearea in the body. The skin has many importantprotective and defensive functions, such as regu-lating water loss and body temperature and de-fending the body against chemical and biologi-cal agents in the environment. Because the skincomes into direct contact with the external envi-ronment, it is ideally suited for studying geneticand environmental interactions. Mutations affect-ing the skin are also easy to observe and to studythroughout the life of the mouse.

“We look at how disease affects animalsunder different conditions. We are interested indetermining how certain genetic mutations makeanimals more sensitive to environmental toxins.The Department of Energy is interested in theeffects of environmental exposures on mice be-cause mice and humans have a similar geneticmakeup, and the information from mice canbe used to better understand humanhealth risks.”

MOUSE MUTANTWITH SKIN DISEASE

Michaud and his col-leagues are studying anORNL mouse mutant that wasborn with a disease called epider-mal dysplasia. Mice with this diseaselose their hair, have thickened skin, and are

more susceptible to getting skin cancer, observedas tumors on the skin.

“We are interested in seeing which genesare altered in mice with epidermal dysplasia be-cause that information will then point to the nor-mal role that these genes play in the developmentand functioning of the skin,” Michaud says. “Likeall organs in the body, the health of the skin isdependent upon the well-orchestrated interactionsbetween complex networks of genes and exposureto numerous environmental variables.”

Michaud gives an example of a geneticnetwork. “Suppose that gene A regulates genesB, C, and D,” he says. “What happens to thesethree genes if A is knocked out or mutated?” It’sa little like making a hole in the oil pan of a carand then driving it. Eventually, the lack of lubri-cation of all the mechanical parts will cause thecar to grind to a halt. The car breaks down be-cause the motor seizes up, but the hole in the oilpan is the main problem. Determining these typesof interactions between genes and the environ-ment gives Michaud and colleagues a better un-derstanding of how the skin protects the body.

Michaud suspects that epidermal dyspla-sia in the ORNL mice is due partly to a mutatedtranscription factor gene, whose job is to regu-late the function of many other genes. To findout what genes this transcription factor is respon-sible for, he and his colleagues had to determinewhich genes are turned on normally when thetranscription factor gene is working. Many ofthese genes would likely have a known functionin the skin, such as DNA repair, cell growth, cell

A mouse withepidermal dysplasia,a skin disease..

Number One, 2001 5

Mitch Doktyz observes the Virtek microarray in a test run in preparation for gene chip production.

Christine Schar, ORNL-UT Genome Science and Technology Fellow, and Brent Harker check the alignment of theVirtek spotter. This state-of-the-art instrument can produce more than 100 gene chips a day. Additionally, it canplace more than 40,000 spots onto a standard microscope slide.


differentiation, and programmed celldeath. If the transcription factor isknocked out, the expression of someof those genes will be altered.

MICROARRAYS AND MICE:HOW A GENE CHIP WORKS

To obtain this information,mouse geneticists Brynn Jones, BemCuliat, and Ed Michaud turned toMitch Doktycz, Peter Hoyt, and theircolleagues in LSD who designmicroarrays, or gene chips, and usethem to perform analyses of geneexpression patterns (See followingarticle.) A microarray allows a com-parison between genes expressed bya normal organism and genes ex-pressed by a mutant organism or anorganism exposed to an environmen-tal toxin. When a gene is turned on,a specific DNA segment correspond-ing to this gene is copied into a mes-senger RNA (mRNA) molecule,which is chemically very similar to DNA. Thisshorter-lived RNA copy moves from the cell’snucleus to the cytoplasm where its code—its se-quence of DNA bases, or nucleotides—is trans-lated, causing specific amino acids to be strungtogether in a specific order to form proteins.

To determine which genes are being ex-pressed at any given time for an important cellu-lar activity, scientists collect the mRNA mol-ecules transcribed in a cell or tissue at that mo-ment. In the laboratory, those RNAmessages are reverse transcribed toform more stable complementaryDNA (cDNA) molecules. ThesecDNA samples are prepared fromtissues in which biologists want toexamine differences in gene expres-sion, such as in skin from mutantmice and normal mice. To detectchanges in gene expression, the twocDNA samples are labeled with fluo-rescent dyes (one for each sample)and then allowed to bind with theircomplementary DNA templates ona gene chip. The gene chips are glassmicroscope slides that are spottedwith DNA sequences from manyhundreds or thousands of differentgenes in an orderly array. The slidesare then exposed to laser light oftwo different wavelengths and theratio of fluorescent intensities ismeasured for each gene. If a geneon the chip is expressed at a highlevel in the skin of the mutantmouse compared with the normalmouse, the dye used for the mutant

sample will shine brighter than the dye used forthe normal sample, and this difference can bequantified.

Computers are used to keep track of infor-mation for each gene on the microarray. Scien-tists need to know many variables related to themicroarray experiments, including the location ofeach gene on the array, the DNA sequence andidentity of each gene, the mouse tissue that the

labeled samples came from, the hybridization con-ditions, the fluorescent ratios for each gene, andthe analysis of the data. At ORNL researchers inLSD’s Computational Biology Section—JaySnoddy, Denise Schmoyer, and Sergey Petrov—are writing the computer programs to handle thedata as part of the development of ORNL’sGenosensor Information Management System(GIMS).

Curtis Boles, enhanced by Gail Sweeden

Curtis Boles, enhanced by Gail Sweeden


This array of blue and pink dots shows the results of a gene chip experiment comparing the gene-expression patterns in mouse skin tissue obtained from a skin mutant and a normal mouse (wild-typecontrol mouse). ORNL’s bioinformatics group analyzed gene-expression ratios (ratios of genes turnedon in mutant mice and genes turned on in normal mice).


For the analysis of genes from normalmice and mouse models of skin and hair disease,Jones, Culiat, and Michaud designed microarrayscontaining about 500 mouse genes that were ar-rayed in triplicate on glass slides. These genechips were used to determine which genes in theskin of the mice with epidermal dysplasia havealtered expression.

As a result of this work, these ORNL in-vestigators made some important discoveries.“We found 30 different genes with altered ex-pression—that is, the levels of mRNAs producedby these genes were different in the mutant micefrom those in the normal mice,” Michaud says.“We found six altered keratin genes, which areresponsible for scaffolding in the skin. We alsofound four programmed cell death genes and twogenes involved in the handling of calcium thatwere altered in the mutant mouse. These genesare important in the normal development and re-newal of the skin.”

Several other genes had unknown func-tions, but the gene chip study indicates that thesegenes play a role in skin and hair production andfunction. More biology experiments are beingdone to verify that these genes have a functionrelated to skin and hair.

Michaud noted that the mice may now beused to determine the effects of various environ-mental agents on gene networks in the skin. “Theskin often protects us from low doses of envi-

ronmental agents such as ultraviolet radiation,microbes, and chemicals,” he says. “Mice withmutations in skin genes, or humans with naturalgenetic variation, may be at increased risk fromthese same exposures, and the gene chips can helpus to uncover the relevant genes.”

Ed Michaud observes theactivity of normal and mutant

mice in large beakers at theMouse House.

The structure of the proteins produced bythe expressed genes—both the normal and mu-tant ones—is determined by LSD’s Gerry Bunickusing X-ray crystallography. Doktycz’s group isalso planning to develop a protein microarray topinpoint protein-DNA interactions.

“In our study of complex systems biology,we are focusing on genetic variationin the skin and increased susceptibil-ity to disease from environmental ex-posures,” Michaud says, “but this ap-proach for determining which genesare interacting can be applied to thestudy of any organ, ranging from theheart to the pancreas to the nervoussystem and brain. For example, wehave a collaboration with RussKnapp, leader of the Nuclear Medi-cine Group in LSD, in which we areexamining gene expression patternsin diabetic mice treated with newinsulin-sensitizing drugs.”

An important goal of complexsystems biology is to understand thefunctions and interactions of the es-timated 35,000 human genes, ofwhich very few are known. Fortu-nately, in this post-genomic era, newanalytical tools, including mutagen-esis techniques, gene chips, andsupercomputers, are available to helpscientists find meaning in the roughdrafts of DNA sequences that havebeen completed for the mouse andhuman.Tom Cerniglio, enhanced by Gail Sweeden

Number One, 2001 7

At ORNL, microarraysare being made fasterand cheaper to studygene expression in cellsfrom mice, fish, andother organisms.

Gene Chip EngineersW

hat does a gene do in amouse, fish, or some otherorganism? One technol-ogy that allows biologists

to spy on gene activity is the microarray, a mi-croscope glass slide dotted with an orderly arrayof DNA sequences.

Mitch Doktycz, Peter Hoyt, and their col-leagues in the Life Sciences Division (LSD) spe-cialize in designing and using microarrays, orgene chips, to help determine which genes areexpressed as a result of specific diseases or ex-posures to environmental toxins. They also havedeveloped technologies to speed up and reducethe costs of preparing DNA probes for gene chips,as well as genetic material from mice, fish, andbacteria.

“We look at hundreds to thousands ofsamples of biological liquids simultaneously onour microarrays,” Doktycz says. “We are evalu-ating expressed messenger RNA (mRNA) iso-lated from various mouse tissues, including skin,liver, lung, brain, muscle, kidney, fat, heart, pan-creas, spleen, gut, and testes.

In collaboration with Ed Michaud’s groupin LSD (see previous article), Doktycz’s grouphas analyzed gene expression in samples col-lected from mice afflicted with skin disease tofigure out which genes are altered in the diseasedmice compared with equivalent genes in normalmice. A gene that is altered produces abnormallyhigh or low levels of mRNA, which eventuallyresults in altered levels of protein.

Doktycz is also collaborating withORNL’s Russ Knapp and Ed Michaud to deter-mine the genes that are altered in mouse modelsof obesity and diabetes. Specifically, these ex-periments use gene chips to determine how newanti-diabetic drugs treat the disease and affect ex-pression patterns of these genes.

Working with Mark Greeley of ORNL’sEnvironmental Sciences Division (ESD),Doktycz and his colleagues have developed a“zebrafish tox-chip microarray.” It is used to de-termine which genes are turned on in zebrafishembryos exposed to hormone-mimickingchemicals.

Besides applying microarrays to gene ex-pression and genome studies, these ORNL

researchers are developing more economicalways to prepare samples—extracting mRNAfrom cells for gene expression studies and at-taching various DNA probes to microarrays.

“When we complete development of au-tomated sample processing, hundreds of tissuesamples will be processed in an afternoon,”Doktycz says. “Using conventional techniques,it can take three to four days to analyze 12samples. With our method, 96 samples can beprepared in parallel in just a few hours. Thissample productivity is needed because we canmake more than 100 gene chips in less than aday. Currently, we can print several thousandDNA spots on each gene chip.”

Hoyt has developed an inexpensive,high-throughput liquid-handling method of ex-tracting mRNA from tissues. A snippet ofmouse skin or other tissue is broken up intocells by simultaneously homogenizing thesamples. After the cells are placed in amicrotiter plate, mRNA is isolated from themusing an automated procedure. Hoyt is nowbeta testing a new Packard Bioscience robotic

At ORNL, microarraysare being made fasterand cheaper to studygene expression in cellsfrom mice, fish, andother organisms.

instrument for “walk-away” automated pro-cessing of mRNA samples. The same instru-ment is used to prepare the complementaryDNA test samples placed on the gene chip(where they will pair with the matching mRNAsamples).

Doktyz has been making a mark in thefield of spotting technologies. He recentlyhelped devise a commercial inkjet technologyfor dispensing microscopic drops of biologicalfluids at high speeds, a technology that couldhasten the development of new therapeuticdrugs. He worked on this project withRheodyne, a California company that makeshigh-end valves, under a cooperative researchand development agreement. The resulting “hy-brid valve” is now produced commercially byInnovadyne Technologies, Inc., a Rheodynespin-off company.

Gene chips and related technologies arerevolutionizing biological studies. To help meetthe need for faster, better, and cheaper ways tospy on genes, Doktycz and his colleagues arebeing ingenious.

Peter Hoyt checks the operation of the Packard multiprobe liquid-handling system and mechanicalgripper. This state-of-the-art robotic tool is used to prepare samples (e.g., cloned DNA) for use onmicroarrays to analyze gene expression.

Cur

tis B

oles

Curtis Boles; enhanced by Jane Parrott


Searching forMouse Models ofHuman Disorders


The lowly mouse is highly re-garded as a key to solving anumber of medical mysteries.The mouse has long been

known to be genetically similar to the human.The recently obtained drafts of the human andmouse genomes suggest that the two genomesare 85% identical: The differences involve a fewhundred of the approximately 35,000 genes inboth organisms. From an economic standpoint,mice and rats are small and inexpensive to main-tain, so it is not surprising that they are used in90% of the research involving animals.

Many mice born with mutations in at leastone gene are good models for human diseases.For example, some mutant mice produced byformer ORNL geneticist Ray Popp have sickle-cell anemia. They are being studied as models ofthe human disease at Meharry Medical College,a historically black institution and a participantwith ORNL in the Tennessee Mouse GenomeConsortium (TMGC).

When a mouse is a model for a human dis-ease, different treatments can be tested on it. Treat-ments found to control or cure the disease in themouse could lead to the development of a therapythat works in humans with a similar disease.

Not all mouse models of human disease areperfect yet they may still be useful, according toDabney Johnson, head of the Mammalian Genet-ics and Genomics Section in ORNL’s Life Sci-ences Division. “Children with cystic fibrosis dieof lung problems, whereas mice with CF die ofintestinal blockages. Mice with the CF gene cansurvive if put on a liquid diet. Perhaps mice havea gene that makes a protein that enables them tocompensate for the lung disorder caused by theCF gene. If so, knowledge of the structure andfunction of this protein could be the key to devel-oping a drug that could benefit humans with CF.”

To determine whether ORNL’s mutantmice are good models for human diseases of thecentral nervous system (CNS), Johnson is col-laborating on projects with the partners of

TMGC. The TMGC recently received a $12.7million grant from the National Institute of Men-tal Health (NIMH) to create 25 to 50 new strainsof mutant mice that will be used to study neuro-logical disorders. The partners include the Uni-versity of Tennessee at Knoxville, UT-Memphis,St. Jude Children’s Research Hospital, VanderbiltUniversity, and Meharry Medical College. TheTMGC is interested in mutations leading to newgenetic information about neurological condi-tions ranging from Alzheimer’s and Parkinson’sdiseases to depression and addiction.

“We produce mutant mice by treatingmales with the powerful mutagen ENU (ethyl-nitrosourea), mating them with females with par-ticular genetic characteristics that help trace thenewly induced mutations, and screening theirdescendants for disorders in the brain and cen-tral nervous system,” Johnson says. “The screen-ing is done using several tests. For example, miceare placed on a spinning rod to see how well theycan maintain their balance there before fallingoff. Mice having certain mutations lack the co-ordination and balance of normal mice and fall

Mutant mice are tested by ORNL researchers and theircollaborators to determine if these mice have diseasessimilar to those that afflict humans. Therapies tried onmouse models could lead to new medical treatments.

Two researchers characterize normal andmutant genes for ORNL biologists. (a) Tse-YuanShen Lu, a research associate with ORNL’s Life

Sciences Division, loads cultures of bacterialcells containing mouse DNA in a Qiagen

BioRobot, which extracts DNA from the cellsinto a liquid. (b) Shen Lu operates a liquid

dispenser for simultaneously filling 96-wellplates with DNA. (c) Melissa York, a research

assistant with the University of Tennessee’sGraduate School of Genome Science and

Technology, places a 96-well plate in apolymerase chain reaction (PCR) machine for

making thousands of copies of each DNAsequence in the plate. (d) Using a pipette,Shen Lu loads running buffer into the PEBiosystems automated DNA-sequencing

machine. (e) York inserts into the machine aloading comb on which fluorescent DNA has

been absorbed. (f) The researchers show an imageof the sequence of DNA bases labeled with dyesthat fluoresce when exposed to laser light. The

sequence was obtained by gel electrophoresis inthe DNA-sequencing machine at ORNL.

(a)

(b)

(c)

(d)

(e)

(f)

All

phot

os b

y C

urtis

Bol

es; e

nhan

ced

by G

ail S

wee

den

Number One, 2001 9


off this rotor-rod more quickly. We can detect whether a mouse is de-pressed by observing its behavior in a swimming test. If it tends to floatrather than swim vigorously to try to get out of the water, we classify itas a depressed mouse.

“An activity test is used to determine if a mouse is underactive oroveractive as a result of a CNS mutation. In this test to gauge a mouse’sactivity in a box, a photobeam sensor counts the number of times perminute that a mouse interrupts light beams sent across the box.”

Johnson and her associates also use this test to measure a mouse’sanxiety level. Mice are, by nature, anxious creatures. “A normal mousestays near the wall where it is more protected rather than going to themiddle of the box where it would be out in the open and feel more ex-posed to predators,” she says. “A less anxious mouse, one that is calmerthan the normal mouse and, therefore, likely to have a CNS mutation,ventures forth into the open space.”

Johnson’s group also uses an array of tests to measure learningabilities and memory in mice to screen for CNS mutations. For example,the researchers administer a mild foot shock and play a sound at thesame time. A day later, when a normal mouse hears that sound, it willfreeze in fear that the unpleasant shock may occur again. Mice with cer-tain CNS disorders will ignore the sound and continue their activity.

As a part of that same test, a normal mouse returned to the samebox 24 hours later will recognize the box as the site of the shock andfreeze. However, a mouse with a CNS disorder will be just as active inthe box as it was 24 hours before, prior to the administration of the shock.The two parts of this test measure two different kinds of memory.

Another CNS test is the startle test. “A normal mouse will have ameasurable startle response when it hears a loud noise,” Johnson says.“It will flinch, and this action will be detected by a load sensor. But anabnormal mouse may not startle at all, perhaps because it is deaf. Or amouse with a CNS disorder could startle too much rather than simplyjump or flinch.”

The TMGC partners help ORNL screen mice for new mutationsand analyze confirmed mutations in more detail. One new mutation re-cently discovered by ORNL’s Eugene Rinchik causes the mouse bornwith it to have continuous seizures. Mice with this mutation have beensent to consortium researchers who then conduct studies to determine ifthe cause of the seizures is neurochemical or neurophysical. They willtry to determine if this “seizure” mouse is a good model for some formof human epilepsy.

At ORNL, researchers run automated analyses of blood and urinesamples from mice, measuring their white and red blood cell counts andhemoglobin concentrations. The dip stick urine test is used to measure

Mouse Models for the Human Disease ofChronic Hereditary Tyrosinemia␣

When a section of mouse chromosome 7 containing the coatcolor c gene is deleted by exposing mice to radiation, “albino”mice are born with a white, hairless coat. ORNL researchersfound that a nearby gene that is also knocked out by irradia-tion causes chronic hereditary tyrosinemia in mice, a diseasethat also afflicts humans. Patients with this disease can developa wide range of liver and kidney problems, as well as problemsaffecting the eye, the skin of the feet and hands, and the cen-tral nervous system.

Using the powerful mutagen ENU to alter a single DNA basepair in this gene, ORNL biologists recently produced two mousemodels that more closely mimic the gene mutation that causeschronic hereditary tyrosinemia in humans.

Normal mice and people metabolize tyrosine, an amino acidavailable in food, to make melanin, a type of pigment producedin large amounts by dark-skinned people. But people and micewith the disease lack a normally functioning protein (enzyme)to carry out one step of the tyrosine metabolism process, whichinvolves a series of enzymes. So, unless people with this diseaseare put on a special tyrosine-free diet, a substance that is notbroken down because of the absence of a normally function-ing enzyme will build up to toxic levels in the liver and kidneys,a fatal condition in mice.

By sequencing the same chromosomal region from both nor-mal and abnormal mice and comparing the sequences, ORNLbiologists identified new mouse models that carry mutations inthis enzyme in the tyrosine breakdown pathway. “Mice withthis disease die of poisoned livers,” says ORNL biologist DabneyJohnson. “A by-product of the botched metabolism process issuccinyl acetone, which accumulates in the liver and is excretedin urine where it serves as a diagnostic indicator of the disease.Because mice entirely lacking this enzyme die right away, weexposed male mice to ENU to produce mice with this enzymein crippled form, rather than entirely missing, so we would havea live mouse model of a disease that some humans have. In thisway, interested researchers could assess the effect of therapyon mice with chronic hereditary tyrosinemia.”

Using X-ray crystallography, ORNL’s Gerard Bunick, along withco-workers Joel Harp and David Timm, determined the struc-ture of the enzyme that is responsible for hereditary tyrosine-mia in the mouse. The mouse enzyme serves as an easily stud-ied model for the same disease in humans.

“I found that this enzyme folds into a three-dimensional shapethat has never been seen before,” says Bunick. “On the basisof our structural observations, we were able to propose howthe enzyme works, allowing us to identify the key amino acidsthat lead to dysfunction of the enzyme when mutated, result-ing in tyrosinemia. We also identified the location of severalsites of known mutation in the structure that could cause theprotein to fold into an incorrect three-dimensional shape, whichwould also cause dysfunction of the enzyme.”

A paper on this research has been accepted for publication inthe journal Structure. A paper on the new mouse models for thedisease has been published in the Proceedings of the NationalAcademy of Sciences. A detailed understanding of the enzymemay lead to a drug to treat hereditary tyrosinemia in humans.

Dabney Johnson watches a mutantmouse afflicted with “behavorialdespair” (left) make no effort to

escape the water in a tank while thenormal mouse swims rapidly in an

effort to get out of the waterduring this swim test.

Tom Cerniglio, enhanced by Gail Sweeden



Some mice born at ORNL have grown dangerously fat, even thoughthey have been on a low-fat diet since birth. Although they do notappear overweight, these mice have a mutated gene that plays a strongrole in causing obesity in the form of internal fat deposits that are haz-ardous to their health. The gene was discovered by Madhu Dhar,postdoctoral fellow from the University of Tennessee who works in thelaboratory of Life Sciences Division’s Dabney Johnson. Dhar’s researchwas funded by the National Institutes of Health, and the findings werepublished in a paper in a recent issue of Physiological Genomics.

“We have found that the normal mouse has a gene on chromosome 7that probably plays a role in the transport of fat from the blood into fatcells, where fat is stored as a source of energy to keep the body healthy,”Johnson says. “If a mutant form of this gene is inherited from the motherin certain genetic backgrounds, the offspring grow 35 to 50 percentfatter by middle age than does a normal mouse, even though they areeating food low in fat.”

Unlike some known mouse obesity genes that can act all alone to causeexcessive body fat, the ORNL researchers have shown that the chromo-some 7 gene must act in concert with other genes involved in maintain-

Obesity-related Gene in Mouse Discovered at ORNLing the body’s energy balance. If femalemice possessing this mutated gene aremated with males having different geneticbackgrounds, the offspring may not be-come obese, suggesting they have genesthat code for proteins that suppress fataccumulation. Humans and mice are ge-netically similar and produce similar proteins.In humans, chromosome 15 is similar to chromosome 7 in the mouse.

The ORNL group has been focusing on chromosome 7 in the mouse forsome time. In the 1950s ORNL geneticists Bill and Liane Russell irradi-ated mice and observed that some of their offspring who survived hada pinkish coat and pink eyes instead of normal gray-brown fur pigmen-tation and dark eyes. It was later determined that the radiation knockedout DNA from a coat-color gene, called the pink-eye marker (p), onmouse chromosome 7. Neighboring genes also were deleted or alteredin some irradiated mice, and the defects were passed on.

Thanks to the availability of improved technologies, ORNL researchershave defined the small region on mouse chromosome 7 that containsthe p gene and its neighbors. Using recombinant DNA techniques, theyidentified and characterized genes fromthis small region in the normal mouse.And in mutant mice they have identifiedbehavioral oddities, internal defects, anddisorders such as epilepsy and obesitycaused by the absence or alteration ofcertain neighboring genes on chromo-some 7. From this information on geneticmaterial gone awry, they can determinethe function of the normal genes in thischromosome region.

Johnson says that, like humans, mice de-posit fat in their bodies in different pat-terns that are genetically controlled. Thedistribution of fat deposits in mice withthe obesity-related gene has been ob-served in three dimensions by Mike Paulusand Shaun Gleason in ORNL’s Instrumen-tation and Controls Division (see articleon p. 11). They have imaged internal fatdeposits in mice using the MicroCAT miniature X-ray computerized to-mography system they developed. Using special software, they havedetermined the size and weight of those fat deposits.

Madhu Dhar shows how a mouse is placed in a MicroCAT X-ray computerized to-mography system. This device has been used to produce images showing the locationsof life-threatening fat deposits in the body of a mouse with the newly discoveredobesity-related gene. Mice with this gene have fat pads in known regions of the body.Thanks to biomedical imaging using the MicroCAT scanner, it is no longer necessaryto sacrifice fat mice and dissect them to measure the size and weight of fat pads.


MicroCAT images showing fatdeposits in a normal mouse anda mouse with the obesity gene onchromosome 7.

Mutant mice lacking the coordination andbalance of normal mice fall off the rotor-rodmore quickly.

Bre

tt P

ate

Cur

tis B

oles

; enh

ance

d by

Gai

l Sw

eede

n

for sugar concentration and excess protein. Thisinformation tells the researchers whether miceare anemic or diabetic or suffer from infections,leukemia, or blood-clotting problems typical ofhemophilia.

“Copies of our mouse mutants go to UTat Memphis, which screens mice for mutant genesthat cause addiction to alcohol and drugs such ascocaine,” Johnson says. “For example, our miceare given the two-bottle test in which one bottlecontains water and the other, alcohol. A mousewith a genetic predisposition for alcoholismmight drink from the alcohol bottle when thirsty,but a normal mouse drinks only from the water

bottle. A normal mouse injected with alcohol fallsoff the rotor-rod quickly, and it loses its inhibi-tions and shows less anxiety—like most peoplewho have too much to drink.”

Mice suspected of having CNS disorders,based on ORNL tests, are sent to the UT Mem-phis Health Sciences Center. The mice are sacri-ficed and their brains are sliced, stained, and stud-ied to determine if they have an abnormalanatomy. The eyes of these mice are also exam-ined to determine whether the retina, neural con-nections, and other components are formed nor-mally. Researchers also check eye samples forsigns of macular degeneration and other predic-tors of impaired vision.

For many people, mice can be a nuisance,but the results obtained from research usingmice could give victims of some diseases a newlease on life.

Number One, 2001 11

MicroCAT “Sees” HiddenMouse Defects


The white mouse sleeps on a nar-row wooden trough outside anew high-resolution X-raycomputed tomography (CT)

system at ORNL called a MicroCAT scanner. Theanesthetized mouse is gently inserted into the in-strument, where it is scanned for about 20 minutes.Mike Paulusand ShaunGleason, re-searchers inORNL’s In-strumentationand ControlsDivision whodeveloped theM i c r o C ATscanner, notethat the scansshow spots in-dicating the lo-cations of lungtumors of 1 to2 millimetersin diameter.The mouse isthen returnedto its cage. Themouse is scanned several times over the next fewdays to get a profile of tumor growth in its lungs.

The mouse has tumors because it has beeninjected with lung cancer cells. Steve Kennel, aresearcher in ORNL’s Life Sciences Division(LSD), creates mice with lung tumors to test howwell they respond to a special kind ofradioimmunotherapy. When Kennel decides it istime to treat the white mouse, he injects it with amonoclonal antibody tagged with radioactive bis-muth-213 (a decay product from ORNL’s stock-pile of uranium-233). This monoclonal antibodyis like a smart bomb because it homes in on thetumor’s blood vessels, where the radioactive bis-muth-213 parks. The alpha radiation from the ra-dioisotope destroys the nearby tumor tissue.

The mouse is scanned repeatedly over aperiod of days both during and after treatment.The MicroCAT pictures reveal that the tumorsare progressively disappearing, indicating that thetreatment works. (See following article.)

The MicroCAT scanner, which was devel-oped using internal funding from ORNL’s Labo-ratory Directed Research and Development Pro-

gram, has been used to provide three-dimen-sional images for other small-animal studies:

• Measurement of the size, weight, and distri-bution of fat deposits in mice found to havean obesity-related gene on chromosome 7 (incollaboration with Madhu Dhar, as describedin the article on p. 10);

• Imaging of tumors on the prostate glandsof mice produced by researchers at BaylorUniversity;

• Imaging of rat bones to measure their resis-tance to the passage of X rays and their bonewall thickness as an indication of bone massand amount of calcium in bone, as part of astudy of osteoporosis in animals;

• Imaging the uterus of a pregnant mouse re-peatedly over time to determine the numberand time of death of individual fetuses (as aresult of genetic disorders);

• Validation of ORNL veterinarian CharmaineFoltz’s body condition scoring system, inwhich she presses her thumb against thebody of each mouse to rank the animal’shealth.

A second-g e n e r a t i o nMicroCAT in-strument is be-ing used forhigh-throughputscreening of mu-tant mice forinternal defectsin ORNL’s Lab-oratory for Com-parative andF u n c t i o n a lG e n o m i c s(Mouse House).The device savestime and moneybecause biolo-gists can screenmice individu-ally for internalmutations in 7to 20 minutes,without sacrific-ing and dissect-ing the animals.

ORNL’s X-ray computed tomography system allows internaldefects and organ changes in small animals to be mapped.

To improve accuracy and speed in assessingmouse organs for defects and damage, Gleason hasdeveloped an automatic organ-recognition algo-rithm for CT images of the mouse. Gleason hasshown that in the CT scan, the software can zeroin on the mouse’s kidneys and analyze differencesin kidney texture that would show up in mice withpolycystic kidney disease. He has also used thealgorithm to calculate the approximate size of thelung and heart and evaluate the lung’s level of fluidor amount of scar tissue, as an indication of itshealth or disease state.

Paulus has also used the MicroCAT scan-ner for research collaborations that do not involvesmall animals. For example, he has worked withplant geneticist Gerald Tuskan of ORNL’s Envi-ronmental Sciences Division to image three-dimensional details of wood cells and cell wallthicknesses in samples of loblolly pine (seearticle on p. 28). Paulus is writing an algorithmthat will enable automatic measurement of the sizeand shape of wood cells.

Paulus and Gleason have developed a newinstrument that combines X-ray imaging withsingle-photon-emission-computed tomography(SPECT). This MicroCAT SPECT scanning in-strument could be used to image lung tumors inmice and detect radioactivity from the treatedtumors, to map their precise location.

With support from UT-Battelle, Paulus andGleason are on entrepreneurial leavetwo days a week to run their newcompany, which manufactures

MicroCAT scanners. Thecompany—ImTek, Inc.(www.imtekinc.com)—has sold and deliveredfive MicroCAT scannersand is filling orders forthree more for 2001.Purchasers includedrug discovery compa-nies, universities,and biotechnologyfirms.

ImTek’s first customer,Henry Lopez of Parke DavisResearch Laboratories in Alameda,California, prepares a mouse for mappinginternal defects in a commercial version of the ORNLMicroCAT scanner. Photo courtesy of ImTek, Inc.

Image of a tumor in the lung of amouse injected with lung cancer cells.This MicroCAT image correlatesreasonably well with histology data.The results were published byORNL’s Steve Kennel in the May2000 issue of Medical Physics.

Enh

ance

d by

Gai

l Sw

eede

n


CuringCancerin Mice


ORNL researchers haveshown that a radioisotope-bearing antibody cantarget the blood vessels oflung tumors in mice,destroying the tumors.

One injection can give a mousecancer but a second can cureit. That was one of the conclu-sions made from recent experi-

ments on mice conducted by Steve Kennel andSaed Mirzadeh, both of ORNL’s Life SciencesDivision.

Dozens of mice were injected with lungcancer or breast cancer cells. The injected cellslodged in the mouse lungs and grew there. Sev-eral of the mice were later anesthetized, and thetiny, solid tumors in the lungs were imaged by

high-resolution X-ray computed tomography,using the ORNL-developed MicroCAT scanner(see previous article).

The cancer-stricken mice were then in-jected with a special monoclonal antibody chemi-cally hitched to a radioisotope produced atORNL. In this protocol, the antibody targets theblood vessels of each tumor like a smart bomb;the antibody, which is a protein, docks with pro-teins found in lung blood vessel cells. The radia-tion from the parked radioisotope destroys thetumor cells around the blood vessel but, remark-ably, leaves the vessel intact. Using theMicroCAT scanner, ORNL’s Mike Paulus tookimages of the mice every few days; the imagesshowed the gradual disappearance of the tumors.”

“We demonstrated that this technique curesmice of lung tumors,” Kennel says. “When we candeliver the radioisotope to the blood vessel thatserves the tumor, we can kill the tumor. As a re-sult of this radioimmunotherapy, the life spans ofthe treated mice are extended dramatically.”

By contrast, the mice with implanted can-cer cells that were not treated died in 15 days.The mice that received too low a dose of radia-tion also eventually died of the lung tumors.

Interestingly, the mice that received higherradiation doses were cured of cancer, but died ear-lier than normal, healthy mice. “These mice livedmuch longer than the other mice in the experi-ment, but they eventually died of pulmonaryfibrosis,” Kennel says. “The reason is that theirlungs exhibited an inflammatory response asdebris-collecting white blood cells were recruitedto the lung to remove the damaged and dead cells.”

Greg Hurst checks a sample plate with100 wells containing various proteinfragments. The plate will be inserted

into the time-of-flight mass spectrom-eter shown at right. This instrument is

used to identify and characterizesample proteins.

You fall on your shoulder and tear some cartilage, causingbone to rub against bone. Your shoulder becomes inflamedand begins to hurt because cytokine, a small signal proteinsecreted by your immune system, has recruited white bloodcells to clean up the damage.

ORNL biologist Steve Kennel recently worked on a project tomeasure levels of a specific cytokine in samples from mice.The goal was to screen for mice that are likely to develop in-flammatory diseases similar to those in humans, such asarthritis, psoriasis, and inflammatory bowel disease.

In a study supported by the Laboratory Directed Research andDevelopment Program at ORNL, Kennel collaborated with GregHurst, a mass spectrometry expert in ORNL’s Chemical and Ana-lytical Sciences Division. They developed a technique that com-bines affinity chromatography with mass spectrometry to sepa-rate three specific cytokines out of the 100 or so differentcytokines in mouse blood or fluid extracted from mouse cells.

“I obtained antibodies specific to different cytokines andattached them to beads,” Kennel says. “These beadswere tossed into a soup of proteins to fish out specific

Other types of experiments in radio-immunotherapy have been performed for yearsin mice, but the success rate has been low. “Pre-vious approaches for radioimmunotherapy ofsolid tumors does not work using labeled anti-bodies that bind directly to tumor cells, becauseantibody stays in the blood and only a small frac-tion reaches the cells in the solid tumor,” Kennelsays. “Our approach has been to select or make aradiolabeled antibody that targets and parks inthe blood vessels in the solid tumor. The type ofradiation used is an alpha particle emitter thatkills every cell within 100 microns.”

The researchers found that a targeted an-tibody labeled with either radioactive bismuth-213 (213Bi) or astatine-211 was most effective.Bismuth-213 is an alpha emitter obtainedat ORNL; it is a decay product from theLaboratory’s stockpile of uranium-233 left overfrom its molten-salt reactor experiments in thelate 1960s. The astatine-211 used comes fromthe National Institutes of Health.

Recently, Kennel and a postdoctoral sci-entist, Sandra Davern, used a bacterial virus en-gineered to display antibody-like molecules. Thebacterial virus was injected into mice with lungtumors. The researchers found that some of thevirus stuck to tumor blood vessels. These engi-neered antibodies were separated by molecularbiology techniques from the virus and replicatedand amplified in bacteria for use inradioimmunotherapy experiments in mice.

These ORNL successes in eradicating lungtumors in mice could provide important clues abouthow to achieve effective human cancer cures.

cytokines for analysis by matrix-assisted laser desorption ion-ization mass spectrometry.

“We were able to identify each cytokine. But because wehad only small amounts of blood from a mouse, we wereunable to detect unusually high levels of our target protein—the tumor necrosis factor alpha cytokine. So, we could notbe sure these test mice were showingan inflammation response.”

Search for Signs of Inflammatory Disease

12

Cur

tis B

oles

;en

hanc

ed b

y G

ail S

wee

den

Number One, 2001 13

Surprises in the MouseGenome


L ike a mouse darting across thehearth and disappearing behindthe home entertainment center,the mouse genome can surprise

even the most seasoned geneticist. For EugeneRinchik of ORNL’s Life Sciences Division, the“unexpected discovery” has been a theme in hiscareer of producing and studying mutant mice.

“We find a lot of wonderful surprises in thisresearch,” he says. For example, in the early 1990s,Rinchik and ORNL scientist Bem Culiat werestudying a form of inherited cleft palate, a facialdeformity and birth defect, in mice. They foundthat newborns affected with cleft palate were miss-ing a certain neurotransmitter receptor gene. Byadding a rat gene that codes for this neurotransmit-ter receptor to a fertilized mouse egg lacking themouse gene, Culiat corrected the disorder, and theresulting mouse was born without a cleft palate.

“The ‘surprise’ was that a reasonable pre-diction for the function of this gene would not haveincluded effects on the palate during fetal devel-opment,” says Rinchik. “If we used only theknown biochemical function of this receptor as aguide, we would expect to find neurological dys-function, not cleft palate, as the primary defect.”

In another example of surprising findings,Rinchik frequently tells audiences about four dif-ferent genes in mice that encode proteins thatshare several structural characteristics. Compu-tational gene modelers have classified these pro-teins as cell-signaling molecules. In general, such

molecules instruct cells to divide, grow, step uptheir metabolism, or die, for example.

“Scientists have studied mice with muta-tions in each of these four genes,” says Rinchik.“One mutant gene results in a de-fective protein that causes mouseembryos to die in the uterus. An-other mutant causes mice to be bornwith cleft palates (for an apparentlydifferent reason than that discussedabove). The third mutant results ininflammatory disease in young ani-mals, and the fourth causes themouse to be born with slight carti-lage abnormalities that show up asshortened ears in an otherwisehealthy animal.

“The point is that althoughthese proteins belong to the samegeneral family of cell-signalingmolecules, they have different func-tions in the mouse. The evidencegained from mouse-breeding experi-ments improve understanding of whathappens at the level of the organismand, therefore, add value to computational predic-tions about the biochemical functions of genes inthe mouse genome.”

Rinchik and his colleagues continue to lookfor new dominant and recessive single-base genemutations in the descendants of mice exposed toethylnitrosourea (ENU). He was inspired to use

ENU by long-time ORNLgeneticists Liane and

Bill Russell, whopioneered its use forproducing mousemutants that couldbe models for hu-man disease, mak-ing ORNL a worldleader in this area.For example, re-cently, Rinchik waspleasantly surprisedto find a new mousemutation that could

shed light on ahuman disease.Some of hismutant mice

were found tohave seizures

continuously for a few weeks

until they died. These mice

may be models for the human disorder epilepsy.The mice are currently being characterized byLisa Webb, a graduate student working in DabneyJohnson’s group at ORNL.

“Now that the DNA sequence of the mousegenome is being completed by the public sector,we should be able to locate and identify a mutatedgene more rapidly,” Rinchik says. “This can bedone by comparing a DNA sequence with an al-tered base from a mutant mouse with the normalDNA sequence from the mouse genome map.”

Rinchik sees more collaborations in thefuture between ORNL mouse geneticists andhuman geneticists. There is already a model forsuch interactions.

“In 1993, in a collaboration with RobNicholls, a human geneticist who is now at theUniversity of Pennsylvania, we identified the hu-man version of the mouse pink-eye dilution gene,which leads to a pigmentation defect in mice,”Rinchik says. “Subsequently, human geneticistsfound mutations in this gene to be responsible foralbinism in black Africans. They are born with littlepigment and are light-skinned as a result.”

Culiat, meanwhile, is using newly avail-able molecular tools to explain a surprising find-ing by retired biologist Walderico Generoso. Hefound evidence suggesting that eggs from somefemale mouse strains can correct damage insperm from male mice exposed to a toxic chemi-cal, reducing the percentage of embryo deaths.By studying gene expression profiles, Culiathopes to determine whether these eggs repairdamaged DNA or have an extracellular filter thatlets in normal sperm and keeps damaged spermout. Indeed, the results could be surprising.

In the live organism, not all mouse and human genes havepredictable functions, and proteins with similar structurescan have different functions.

Bem Culiat examines two autoradiograms showing the expression patterns of600 genes in egg cells from twostrains of mice, C3H (left)and T (right). Genetic studieshave shown thatC3H femalesdo not readilytransmit chemi-cally inducedDNA damagefrom the fatherto the offspring,while T femalesvery easilytransmit DNAdamageincurredby thefather.

Curtis Boles; enhanced by Jane Parrott

In 1979 ORNL’s Bill Russell discovered that the chemical ENUcauses mutations in mice. He and his wife Liane pioneeredENU’s use for producing mouse mutants that could be modelsfor human disease.


Protein Identificationby Mass Spectrometry


Proteins are large, complex mol-ecules that carry out the tasks oflife. They direct our bodies’ ac-tivities, organize our thoughts,

and defend us against infection, keeping ushealthy. But in their mutant forms and as coatson disease-causing microbes, proteins can helpmake us ill and threaten our health.

Each protein is initially formed as a stringof amino acids whose identity and order are dic-tated by a gene according to the sequence of itsDNA bases. The gene’s instructions—carried bymessenger RNA—also call for this string to befolded into a three-dimensional molecule that hasan intricate shape, ranging from a saucer to adumbbell to a corkscrew.

The amino-acid composition and se-quence, as well as the molecular weight, of a pro-tein produced by a certain type of bacteria aredifferent from those of the protein forming thecoat of a particular virus. Such a coat enablesthis molecular terrorist to break through the pro-tective membrane of a cell and command it toproduce more virus. Each type of microbe pro-duces unique proteins, providing a characteristicprotein signature and allowing identification ofthe microorganism. Thus, if the signature pro-teins in anthrax spores and botulism toxins couldbe accurately detected, it would be possible toprovide an early warning about the proximity ofbiological warfare weapons.

The ability to identify proteins is also im-portant because it allows researchers to determinewhether an organism has a genetic disease. Agenetic disease is often caused by a mutant pro-tein, which has a composition slightly differentfrom that of the normal protein it replaces.

One of the most powerful tools for detect-ing and identifying proteins is mass spectrometry,a technique that has been improved and used fora variety of research projects for many years atORNL. A mass spectrometer sorts out chargedparticles according to their masses, allowinganalysis of the elemental composition of com-plex molecules. A mass spectrometer produces aspectrum consisting of peaks and valleys thatindicate the identity and number of different at-oms making up the molecule being analyzed.

The mass spectrometer is an ideal instru-ment for identifying amino acids—the buildingblocks of proteins—and determining the orderin which they are arranged. The mass difference,or distance in atomic mass units between thepeaks along the spectrum, allows each amino acid(e.g., alinine, arsenine, glycine, or lysine—fourof the 20 possible amino acids) to be identified.

NEW IONIZATION METHODSAID PROTEIN ANALYSIS

With the discovery inthe past decade of two pow-erful methods that can be usedto produce ions for analysis ina quadrupole ion trap massspectrometer, progress in bio-logical mass spectrometry hasbeen nothing short of revolu-tionary. Mass spectrometryhas become, in just a fewyears, an important tool forprotein identification, peptidesequencing, identification andlocation of post-translationalmodifications of proteins,analysis of modified deoxyri-bonucleic and ribonucleicacids (DNA and RNA), andmany other biological appli-cations. These advances havebecome possible only throughthe capability to form ionsfrom large biomoleculesand through the researchcommunity’s growing under-standing of the chemistry ofbiological ions.

During the mid-1980s,the quadrupole ion trap wasbeginning to emerge as a massanalyzer with interesting char-acteristics for tandem massspectrometry. The quadrupoleion trap operates on the prin-ciple that ions can be storedwithin an oscillating electricfield. With appropriately

shaped electrodes, an oscillating electric field(usually a quadrupole field or a variation thereof)can be created that stores ions in three dimen-sions. Furthermore, the amplitude (strength) ofthe electric field can be changed so that ions ofdifferent mass-to-charge ratios are ejected fromthe ion trap and into a detector. In this way, theion trap serves as a mass spectrometer.

In the latter part of the 1980s, the OrganicMass Spectrometry Group in ORNL’s Chemicaland Analytical Sciences Division (CASD)—oneof the leading groups in the world today in thisarea of research—began to study electrospray

ORNL researchers are improving mass spectrometry tools to speed up protein identificationand to screen for disease-causing proteins and bacteria.

Greg Hurst (center) and Jim Stephenson (right) watch Bob Hettichprepare to use a pipette to dilute a sample of bacterial proteinsin a vial. The sample will be placed in the Fourier transformion cyclotron resonance mass spectrometer(in foreground) for analysis.

Cur

tis B

oles

; enh

ance

d by

Gai

l Sw

eede

n

Number One, 2001 15


ionization. It is one of the important new ioniza-tion methods for mass spectrometry because itforms gaseous ions from polar and nonvolatilemolecules in solution, without the addition ofheat. Gary Van Berkel, Scott McLuckey, and GaryGlish, all of CASD, were the first to couple theelectrospray ionization technique with the iontrap mass spectrometer.

“In electrospray, a tiny drop of protein-containing solution is injected into a thin glassneedle,” says CASD’s Jim Stephenson. “Theneedle is held at a potential of several thousandvolts, providing energy and adding multiplecharges to the protein. As the solution leaves theneedle, it evaporates, forming a fine mist ofcharged droplets from which ions emerge. Theseions in a gaseous state are introduced to the massspectrometer.”

Another relatively new ionization methodused at ORNL to analyze biological samples ismatrix-assisted laser desorption (MALDI) massspectrometry. For protein analysis, a protein isdissolved in solution and added to a matrix. Themixture is placed on a probe tip, which is illumi-nated by an ultraviolet light beam from a nitro-gen laser. The laser beam desorbs the protein offthe surface of the probe tip. Protons are then trans-ferred between the protein and matrix, leavingthe protein as a negatively charged ion.

BIOLOGICAL APPLICATIONSOF MASS SPECTROMETRY

Since coming to ORNL in 1995 as apostdoctoral researcher, Stephenson, now a staffscientist in CASD’s Organic Mass SpectrometryGroup, has been participating in biological re-search using mass spectrometry. He has beenworking with other CASD researchers, particu-larly Keiji Asano, Doug Goeringer, Bob Hettich,Greg Hurst, Rose Ramsey, Gary Van Berkel, andMichelle Buchanan, now director of CASD.

“We use mass spectrometry to identify thecharacteristic protein signatures of various bac-teria and viruses, such as the tobacco mosaicvirus,” he says. “We extract proteins fromEscherichia coli bacteria cells and look for pro-teins unique to these organisms. We can identifythe signature proteins in anthrax spores andbotulism toxins.”

Using mass spectrometry,Stephenson and his ORNL colleaguesobtain two important pieces of infor-mation that allow them to identify pro-teins. First, they get the molecularweight of the protein. Second, by mea-suring the distances between thepeaks in the spectrum, theycan figure out which aminoacids are present and in whatorder to get the “sequencetag.” They also can deducethe amino-acid sequence bythe way the protein falls apartwhen charges and energy areadded to it.

“With this informa-tion, we can go to theprotein sequence databaseand identify the protein,”Stephenson says. “For ex-ample, we might find six to-bacco mosaic virus proteinsin the database that have thesame sequence tag but differ-ent molecular weights. Onlyone protein in the databasematches the molecularweight and sequence tag wefound.”

Stephenson and hiscolleagues have been analyz-ing the disease state of mu-tant mice in a project with

mouse geneticist Gene Rinchik ofORNL’s Life Sciences Division. Forthis experiment, Rinchik exposes malemice to the powerful chemical mutagenethylnitrosourea (ENU) and mates themwith female mice from the same strain.As a result, many different mutationsare produced in otherwise identical off-spring because ENU can alter a singlebase pair in a gene. Stephenson andRinchik are interested in using the massspectrometer to detect mutant proteinsthat result from these inherited geneticchanges. As a proof of principle, theyare initially looking for mice born witha mutant hemoglobin protein.

Red blood cells carry oxygen toall parts of the body by using a proteincalled hemoglobin. A small sample ofblood is easy to obtain from a largenumber of mice, so hemoglobin is a

good model to test whether or not the mass spec-trometer could find small inherited proteinchanges in the progeny of mice treated with theENU mutagen. Rinchik was able to obtain asample of bloodwith knownmutant hemo-

Jim Stephenson adjusts an electrospray ionization ion trap massspectrometer.

A colored micrograph of the phage MS2 virus, an experimentally determined structure ofan MS2 viral coat protein, and a mass spectrum showing the MS2 coat protein’s partialamino acid sequence are displayed here.

Cur

tis B

oles

; enh

ance

d by

Gai

l Sw

eede

n



globins from Ray Popp, retired from ORNL’s former Biology Division,and Stephenson found that the mass spectrometer could easily identify thechanges. “Now,” Stephenson says, “we can use mass spectrometry onblood samples from the offspring of mutagen-treated mice to detect newmutations. If we are successful in recognizing hemoglobin protein vari-ants, we can use the technique to identify new, inherited mutations in anyprotein, increasing the efficiency and reducing the cost of finding new in-herited variants in mice.”

The ORNL group also uses mass spectrometry to detect post-translational modification (PTM) proteins. In a cell, a protein can be modi-fied in different ways by other proteins or by exposure to a pollutant. Forexample, in a process called phosphorylation, a phosphokinase enzymecan attach a phosphate to a protein to activate it or remove a phosphate toinactivate the protein. Mass spectrometry can be used to confirm the pres-ence or absence of a phosphate in a protein.

In addition, Stephenson and his colleagues are using electrosprayionization mass spectrometry (ESI/MS) to help determine the three-dimensional structure of proteins. Using a cross-linking chemical of aknown length that attaches between two neighboring lysines in a polypep-tide chain, the group can measure the molecular distance between theseamino acids. This information is of value to the Computational ProteinStructure Group, led by Ying Xu, which is part of LSD’s ComputationalBiology Section. This group uses a protein-threading computer model topredict the structure of proteins. (See article starting on p. 20.)

SPEEDING UP PROTEIN IDENTIFICATION

In a project supported by internal funding from the LaboratoryDirected Research and Development Program, Stephenson and his col-leagues further developed ESI/MS so that it could analyze proteins muchfaster than the conventional method he describes below:

“Traditionally, we take a purified protein and break it into smallerpieces by digesting it with a proteolytic enzyme that selectively cleavesthe protein at specified amino acid sites. The resulting products from theproteolytic digestion are then separated on a liquid chromatography col-umn and then are transferred to the mass spectrometer directly. Sequencetags are then generated by adding energy to the protein pieces via colli-sions with helium atoms. These protein pieces fall apart into the indi-vidual amino acids of the protein. From these data we can figure out thesequence of the sequence tag and identify the protein by checking the se-quence tag and molecular weight against protein data in the database. Thisapproach takes about a day.”

By making improvements in the ESI/MS technique and eliminatingliquid chromatography, Stephenson and his associates could analyze a singleprotein in just a few minutes, not a day. “The problem then was how to useESI/MS to identify many different proteins in a complex mixture at once.You could present one protein at a time to the mass spectrometer afterseparating the protein mixture by liquid chromatography or gel electro-phoresis. But this approach is time consuming. So we designed an ion-ionreaction instrument to separate out a target protein or allow the massspectrometer to look at one protein at a time.”

Stephenson and Ben Cargile, a former graduate student at the Uni-versity of Tennessee, developed a protein identification algorithm basedon their discovery of how intact proteins fall apart when energy is added.This algorithm can be used to take sequence-specific data from intact pro-teins and identify them through a database search.

Speeding up protein identification and the collection of informationon the compositional differences between normal and mutant proteins andthe measurements of distances between protein building blocks could leadto the rapid development of more effective therapeutic drugs. ORNL’s ca-pabilities in combining computational analysis with data obtained byORNL’s Organic Mass Spectrometry Group could result in increasedprotection and improvement of human health.

In 1993, using their new ability to detect 35 to 110 basepairs in a single DNA strand, Winston Chen and his asso-ciates at ORNL were the first to demonstrate the use oflaser desorption mass spectrometry (LDMS) to detect amutant gene responsible for cystic fibrosis (CF). Their dem-onstration used clinical samples custom prepared by Drs.Karla Matteson and Lan-Yang Chang, both of the Uni-versity of Tennessee Medical Center (UTMC).

CF is an inherited, fatal disease in which mucus builduppromotes digestive disorders and bacterial infections inthe lungs. Because each person with CF is the child ofparents who both carry defective forms (alleles) of a par-ticular gene, there is interest in large-scale screening tolet people know their chances of having a child with CF.

“Our technique could be used to rapidly screen manypeople for a specific defect in a gene on chromosome 7that causes 70% of all CF cases,” Chen says. “The de-fect is the absence of three base pairs of DNA in bothalleles that control production of CFTR, a protein thatprevents mucus buildup. CF carriers have a single defec-tive allele that may be passed on to their offspring, andpeople born with CF have two defective alleles.

The ORNL group was the first to show that LDMS candiagnose a genetic disease via DNA analysis. The tech-nique, which is not yet used commercially, screens for CFin minutes, not hours, making it 10 times faster than con-ventional gel electrophoresis. Also, it does not use toxicchemicals or radioactive materials, which require costlymethods of disposal.

In their continuing collaboration with Drs. Matteson andNick Potter, also of UTMC, Chen and his colleagues dem-onstrated in 1996 that a single-base mutation (replace-ment of the right base with the wrong one) can be de-tected by LDMS.

More recently, he and Dr. Potter showed that LDMS candetect neurodegenerative diseases that result from dy-namic mutations, such as Huntingdon Disease. The nor-mal Huntingdon gene has 9 to 24 repeats of the GACsequence. The mutant gene causing Huntingdon diseasehas more than 25 repeats. The researchers showed thatLDMS can identify the HD gene because of its greatermolecular weight, resulting from the many additionalrepeats.

“Laser desorption mass spectrometry,” says Chen, “isemerging as a new tool for screening populations for vari-ous genetic diseases.”

Rapid Genetic Disease Screening PossibleUsing Laser Mass Spectrometry

Number One, 2001 17

Lab on a Chip Usedfor Protein Studies


17

Ten years ago, ORNL’s MikeRamsey built the first lab on achip. Now, improved versionsof this miniature chemistry lab

are being shipped all over the world. A toaster-sized, computerized device containing fourmatchbox-sized protein identification chips mod-eled after Ramsey’s invention is being mass-produced by Caliper Technologies, Inc., in Cali-fornia. It is being sold to biotechnology firms byAgilent Technologies.

“Proteins, like DNA, pose a massivechemical measurement problem,” Ramsey says.“But we have learned how to use a lab on a chipto measure molecular weights of proteins in muchsmaller samples and in much shorter times thanare required by conventional methods.”

Caliper Technologies is developing micro-chips for drug discovery. Such a device wouldhelp pharmaceutical firms rapidly identify com-pounds effective in inhibiting the activity ofdisease-causing proteins. “We believe that anautomated device with massively parallelmicrofluidic chips can work with sample volumesthat are 1/10,000th the volumes analyzed in con-ventional benchtop devices, at 10 to 100 timesthe speed or more,” Ramsey says.

The current drug discovery chip containsfour channels—thinner than human hair—thatconnect reservoirs, all of which are carved into arectangular glass plate, using microfabricationtechnologies. A disease-related enzyme is intro-duced into a chip channel. Because of pressuredifferences, the enzyme and a modified substrateflow through the channel network, mix, and re-act. The reaction product is fluorescent whenexposed to a laser beam. The amount of fluores-cence is a measure of the reaction rate.

When a test compound is introduced intothe chip through another channel, it typically re-acts with the enzyme, blocking out the substrateso less of the fluorescent product is produced ata time. The reduced fluorescence signal indicatesthe effectiveness of the test inhibitor compound.By introducing different test compounds to thedevice every 5 seconds, it is possible to rapidlycompare reaction rates to identify potentially ef-fective drugs. In a recent demonstration at Cali-per, nearly a million compounds were screenedwhile using less than 1 microgram of enzyme(usually a very valuable material).

In ORNL’s Chemical and Analytical Sci-ences Division, considerable research is beingconducted by Ramsey’s group on developingimproved lab-on-a-chip technologies for biologi-cal, environmental, forensic, and defense appli-cations. The lab on a chip has been honored byR&D magazine as one of the 40 top innovationsthat have come about since the magazine beganits R&D 100 competition in 1963. It also has beenrecognized by a panel of citizens as one of thetop 23 technologies developed using Departmentof Energy funding.

In 1998 at Caliper Technologies, RoseRamsey (Mike’s wife) and a colleague there firstdemonstrated that the lab on a chip could com-plete a two-dimensional (2D) separation of pep-tides in under 10 minutes. The 2D chip separa-tion uses two types of separations to resolve thepeptides. In one channel, they are separated bydifferences in migration speed and in anotherchannel by differences in peptide charge and sizein response to an electric field (capillary electro-phoresis). By contrast, it takes 24 to 48 hours todo this separation using conventional 2D gel elec-trophoresis. More recently, Stephen Jacobson,Chris Culbertson, and Norbert Gottschlich con-tributed to a newly designed 2D chip that workseven faster.

Mike, Rose, and Robert Foote of ORNLthen received funding from the Laboratory

D i r e c t e d

Research and DevelopmentProgram at ORNL to analyzeproteins by combining the labon a chip with an electrosprayionization time-of-flight massspectrometer (ESI/MS). Rose,who conceived the idea, andIulia Lazar, a post-doctoral fel-low at ORNL, showed that theprocedure could be used to rap-idly analyze hemoglobin, theprotein that makes blood cellsred and transports oxygen fromthe lungs to the body tissues.

In the chip, the hemo-globin from a drop of blood is reacted with theenzyme trypsin, which cleaves the blood pro-tein at the scattered sites of two amino acids.The fragments of the hemoglobin areelectrosprayed directly from the chip as ions intothe mass spectrometer.

“The pattern of the fragments gives a fin-gerprint for the protein that can be compared toamino-acid sequences in the protein database, al-lowing us to identify the protein,” Rose says. Sheused this technique to sequence over 70% of hu-man hemoglobin from a drop of blood. She dem-onstrated that the technique can rapidly distinguishsickle-cell hemoglobin, whose fingerprint is dif-ferent from that of normal hemoglobin. This workwas recently published in Analytical Chemistry.This technique could also be used to rapidly screenfor other hemoglobin variants.

The lab on a chip may be small but itspotential for advancing medical diagnosis andtreatment is quite large.

ORNL’s lab on a chip is being used commercially to identify proteinsand shows promise for drug discovery and disease screening.

Rose Ramsey checks the lab on a chip developed foruse with a time-of-flight mass spectrometer (right).Proteins can be separated and digested chemicallyon the chip. The protein fragments are “sprayed” asions from the chip into the mass spectrometer forfurther analysis and identification.

Curtis Boles; enhanced by Gail Sweeden


The Mouse House:From Old to New


In the spring of 2003, ORNL’s newMouse House will open, replacingthe old Mouse House that dates backto 1948. The new Mouse House

(officially called the Laboratory for Comparativeand Functional Genomics) will hold 60,000 mice.It will be located in ORNL’s new Marilyn LloydEnvironmental and Life Sciences Complex.

The mice in the old Mouse House, includ-ing those allowed to grow old there, will not bemoving to the new Mouse House. The mice in thenew Mouse House will in-deed be new. But the geneticheritage of many of the newlybred mice will be traceableback to the old Mouse House.

Using cryopreser-vation techniques, some ofwhich were pioneered byformer ORNL biologist PeterMazur and his colleagues,staff in ORNL’s Life SciencesDivision (LSD) are freezingand preserving eight-cellmouse embryos, sperm, andegg-containing ovaries. “Wealready have 450 stocks offrozen mouse embryos,” saysLSD’s Eugene Rinchik.“Genetic characteristics andother information on eachmouse stock have been en-tered into a computerizeddatabase (http://www.lsd.ornl.gov/htmouse). If neededfor new experiments, any ofthese embryos can be thawedout and implanted into newfemale mice brought to thenew Mouse House. The miceborn to the surrogate moth-ers will have the same mutantgenes as mice now in the oldMouse House.”

The current ORNL Mouse House is a valu-able research resource. Its colony of mice is ofspecial interest because they represent a varietyof mutations to genes, including those that causeobesity, diabetes, skin and stomach cancer, leu-kemia, cleft palate, polycystic kidney disease,

is associated with the later-onsetabnormality.”

“We are also working with ourpartners in the Tennessee Mouse Ge-nome Consortium to examine oldermice, which carry mutations induced atORNL, for late-onset diseases and dis-orders,” Rinchik says. “The diseases ofinterest include, but are not limited to,cancer, diabetes, obesity, memory prob-lems similar to Alzheimer’s disease,neurological problems similar toParkinson’s disease, and inner-ear orother neurological problems that causeloss of balance, lack of coordination, oreven deafness.

“We are trying to find out if we can rec-ognize mutations that cause effects in older ani-mals but not necessarily in younger ones. Mostnew inherited disorders are currently identifiedin younger animals, and we are trying to extendthat observation period to older mice carryingmutations. One goal is to determine if we canrecognize good models for old-age diseases inour mutant mice.”

There are various ways to produce muta-tions in mice. One way is to expose them to

radiation and chemicals, in-cluding a powerful mu-tagenic chemical calledethylnitrosourea (ENU),first used by Bill and LianeRussell at ORNL. A newerway is to use DNA recom-binant technology or ENUto alter genes in embryonicstem cells from mice.

In one of Rinchik’sapproaches to breeding mu-tant mice, numerous malesare exposed to ENU andmated with females that havea known mutation that re-sults in an easily identifiedabnormality (e.g., unusuallyhairy ears). Such “visiblemarkers” allow geneticists totrack the mutated chromo-somes through subsequentgenerations. Various tests arethen performed on the de-scendants of these mice, todetermine if they have reces-sive or dominant mutationsthat result in changes in be-havior, biochemistry, physi-ology, or anatomy, or thatrender them more suscep-tible to other diseases.

Using mice formedical research is an old idea, but the ben-efits of this kind of research to humans arelikely to improve as resulting research atORNL and elsewhere yields new discoveries.The new Mouse House will allow ORNL re-searchers to carry on the Laboratory’s long tra-dition of mouse genetics.

chronic hereditary tyrosinemia, neurological dys-functions, seizures, and a wide variety of birthdefects. All of these diseases and disorders aresimilar to human afflictions.

For research on the effects of aging, somemutant mice are being allowed to grow old. “Inour lab, we are putting four males and four fe-males of some of our mouse families on the shelfand letting them age to 18 months, which is theequivalent of 75 years in humans,” says DabneyJohnson, head of LSD’s Mammalian Genetics

and Genomics Section. “As these mice approachold age, we screen them for late-onset diseases.

“So far our tests of non-mutant older micein mazes show that they have less agility withage but that activity levels and memory are thesame as in younger mice. If we find abnormalolder mice, we will determine which gene

While some ORNL mice are allowed to grow old for studies ofaging, mutant mouse embryos are being frozen, awaiting birthafter the new Mouse House is built.

ORNL’s new $13.9 million Mouse House (inset)will be built by spring 2003 by Turner-Universalof Nashville. It will be a state-of-the-art animalresearch facility with a capacity for 60,000 mice.The Mouse House and the buildings in the front(to be constructed between 2007 and 2111) willbe located in the Marilyn Lloyd Environmentaland Life Sciences Complex.

Number One, 2001 19

Human GenomeAnalyzed UsingSupercomputer


The human genome has 100,000genes. One gene makes one pro-tein. Humans and bacteria haveentirely different genes.

These common beliefs were shattered ear-lier this year by the findings of the InternationalHuman Genome Sequencing Consortium, whichincludes the Department of Energy Joint GenomeInstitute (JGI) to which ORNL contributes com-putational analysis. On February 15, 2001, threedays after a major announcement, the consortiumpublished the paper “Initial Sequencing and Analy-sis of the Human Genome” in the journal Nature.The paper states that the human genome has “about30,000 to 40,000 protein-coding genes, only abouttwice as many as in worm or fly”; each gene codesfor an average of three proteins; and it is possiblethat hundreds of genes were transferred frombacteria to the human genome.

Ed Uberbacher, head of the ComputationalBiology Section in ORNL’s Life Sciences Divi-sion, was one of the hundreds of contributors tothis landmark paper. He and his ORNL col-leagues performed computational analysis andannotation of the DNA data produced by JGI touncover evidence of the existence of genes aboutwhich little or nothing was known.

Uberbacher and his colleagues also per-formed an analysis of the complete, publiclyavailable, human genome. The analysis, fundedby DOE, was performed by ORNL, Universityof Tennessee, and Universityof Pennsylvania

researchers using three computational methods, theGenBank database, and the IBM RS/6000 SPsupercomputer at DOE’s Center for ComputationalSciences at ORNL. One of the computationalmethods used was the latest version of the GeneRecognition and Analysis Internet Link (GRAIL),which was developed by Uberbacher and othersat ORNL in 1990 and rewritten as GrailEXP forparallel supercomputers.

“We have found experimental and compu-tational evidence for some 35,000 genes,” saysUberbacher. “We have also provided informationon how many genes are expressed in different tis-sues and organs of the body. For example, we de-termined that more than 20,000 genes are ex-pressed in the central nervous system. About 10%of all human genes are expressed only in the brain.”

The researchers found 728 cell-signalinggenes that tell cells when to divide and when togrow. They identified “zinc fingers”—regulatoryproteins that bind to DNA bases composing genesto turn them on or off. These cell-signaling genesand zinc fingers are unique to the human genome.

GrailEXP located almost 2600 genes ex-hibiting “alternative splicing”—the ability to pro-duce two or more proteins by combining thegene’s dispersed protein-coding regions (exons)in different ways.

Each human gene contains multiple ex-ons separated by noncoding regions called in-trons. Cellular machinery called a

spliceosome strips out all the in-trons and joins the exons to-gether. Sometimes cer-tain exons are skipped.

A computational analysis of the human genome by ORNLand UT researchers provides insights into what our genes do.

Ed Uberbacher shows a printout, spread over aconference table, that depicts human chromosome20’s genes as bars of different colors and lengths.In the background are genes from microbes.

19Number One, 2001

The IBM RS/6000 SP supercomputer at ORNLwas used for human genome analysis.

Cur

tis B

oles

Enhanced by Gail Sweeden

“We found a gene with 10 exons, but in dif-ferent human tissues different individual exons arenot read, so part of the code is left out that directsthe cell to make a protein,” Uberbacher says. “Thisgene could have 10 different protein products.”

Some of the genes are known, and detailedinformation on their sequences is found inGenBank. Other genes are less well character-ized but are similar to genes found in model or-ganisms, such as the mouse. Still other genes areinferred based on expressed sequence tags(ESTs). An EST is a unique stretch of DNAwithin a coding region of a gene that can be usedto identify full-length genes. ESTs were used incomputational predictions to locate additionalgenes and predict the makeup, structure, andfunction of the proteins they encode.

In addition to genes, the researchers foundmany DNA sequences that are repeated in the hu-man genome. This “junk” DNA may have a pur-pose: It lowers the probability that random mutationsin DNA strike the coding sections of important genes.“Although the human and mouse genome are aboutthe same size,” Uberbacher says, “we found longerstretches of repeated DNA sequences, making up40 to 48% of the human genome, separating clus-ters of genes like vast deserts between metro-

politan areas.”


Protein Prediction ToolHas Good Prospects

The international competition topredict the three-dimensional(3D) structures of 43 proteins,using computational tools, was

intense. Of the 123 groups competing in thefourth Critical Assessment of Techniques for Pro-tein Structure Prediction (CASP-4) competition,which was held from June through September2000, an ORNL group placed sixth, putting it inthe top 4%. In fact, ORNL placed ahead of allother Department of Energy national laborato-ries in the contest.

The actual structures of the 43 target pro-teins were determined experimentally by nuclearmagnetic resonance (NMR) spectroscopy andX-ray crystallography, and the data were unpub-lished at the time of the competition. “The com-putational groups were provided with the iden-tity and order of amino acids making up eachprotein and the length of the one-dimensionalamino-acid sequence,” says Ying Xu, leader ofthe Computational Protein Structure Group in theComputational Biology Section of ORNL’s LifeSciences Division. “From this information ourteam predicted protein structure.” Other teammembers were Dong Xu, Oakley Crawford, andPhil LoCascio.

Motivating this competition is the searchby biologists to discover the function of indi-vidual proteins that work together in “proteinmachines” to form an organism or keep it alive.They also want to understand how these func-tions are performed at the molecular level. Pro-teins often do their work by docking with an-other protein. Because the function of a proteinis related to its shape, it is essential to learn the3D structure of each protein. Using the details ofa protein’s shape, a chemical compound can becustom designed to fit precisely in the protein,like a hand in a glove, blocking or enhancing theprotein’s activity. In this way, a highly effectivedrug with no side effects could be created foreach individual.

“The demand for rapid protein structuredetermination will grow drastically because in-formation that could be used for rational drugdesign is becoming available rapidly,” Xu says.“Traditional experimental methods for determin-

is stored in the Protein Data Bank,” Xu says. “Tokeep up with the production rate at which pro-tein sequences are being generated by the genomeprojects, computational methods are clearlyneeded. Structure predictions have been made bythe conventional ab initio technique in which asupercomputer is used to predict how an amino-acid chain can fold itself into a final shape basedon first principles of physics and chemistry. Un-fortunately, it takes weeks to months to predictthe structure of even the smallest protein using thisapproach and the prediction reliability is poor.”

The ORNL group uses template-basedmethods of protein structure prediction. Thesemethods rely on experimentally determined 3Dstructures in the Protein Data Bank. The ORNLgroup uses PROSPECT to do “protein thread-ing,” in which a string of amino acids iscomputationally aligned along different proteintemplates—like an embroidery thread drawnthrough a printed design—to determine whichtemplate gives the best fit. In a perfect alignment,the amino-acid atoms are at their preferred low-est energy levels and are compatible with neigh-boring atoms and the protein’s environment.

ORNL ranks high in its ability to computationally predict protein structures. Thenext step is to speed up predictions to facilitate the search for effective drugs.

High-priced bows for birthday presents? No, these images show the actual structures ofseveral proteins and their predicted structures using PROSPECT software at ORNL.

ing protein structure may not be able to keep upwith the pace at which amino-acid sequences arebeing generated. Computational techniques, inconjunction with experimental methods, couldmore rapidly determine protein structures on agenome scale.”

For the CASP-4 competition, the ORNLresearchers used a computer package that theydeveloped and continue to improve. It is calledthe Protein Structure Prediction and EvaluationComputer Toolkit (PROSPECT) and is one ofonly a few dozen protein-threading computerprograms in the world.

“In the CASP competition, you get a 0 ifyou fail to identify the correct structural tem-plate,” Xu says. “You get a 4 if your alignmentbetween the target protein and the template is per-fect. You get scores of 1 to 3 depending on howclose you are to being correct. The scores areadded up for all 43 proteins. We recognized two-thirds of the correct templates, which is the mostamong all the competing teams, and one-third ofour alignments were off.”

Recently, the ORNL team attended aconference inA s i l o m a r ,Cal i fornia,and learnedhow otherteams did inCASP-4 com-pared withPROSPECT.

“Some10,000proteinstruc-t u r e sh a v ebeen de-terminedexperimen-tally out ofthe 100,000 orso proteins be-lieved to exist,and the information

Protein Prediction ToolHas Good Prospects

Number One, 2001 21


The ORNL group also uses homologymodeling to fine-tune the predicted structure. Inthis technique, if two amino-acid sequences aresimilar and one sequence has a known structure,researchers can use this information to help de-termine the structure of the unknown protein se-quence. By calculating the detailed forces be-tween atoms and adjusting the final predictedstructure to minimize the atoms’ energies, the re-searchers computationally tweak the predictedstructure of the target protein to make it ener-getically more favorable.

“It is believed that about 1000 unique pro-tein structural folds exist in nature and that manyproteins share each of these unique structuralfolds,” Xu says. “Some 600 unique protein struc-tures have been determined experimentally. Oncethe 1000 unique structural folds are determinedby NMR and X-ray crystallography, the rest ofthe 100,000 protein structures can be accuratelymodeled computationally.”

Xu’s group, which has four staff research-ers and two postdoctoral scientists, is involvedin the National Institutes of Health’s StructuralGenome Initiative, which is dedicated to findingthe structures of 100,000 human proteins. As partof this effort, NIH has funded seven pilot centersfor experimentally determining protein struc-tures. They include centers at DOE’s Argonne,Brookhaven, Lawrence Berkeley, and LosAlamos national laboratories. At Lawrence Ber-keley, David Eisenberg, a pioneer in proteinthreading, is trying to determine the structuresof proteins in the genome of the rod-shaped bac-terium that causes tuberculosis.

“A new trend in structure prediction is theincorporation of partial experimental data as con-straints in the computation process, to make struc-ture prediction closer in accuracy to the experi-mental structures,” Xu says. “PROSPECT is ide-ally suited for incorporating data from local re-

searchers and from these pilot centers. The datainclude measurements of distances betweenamino acids and information on which aminoacids tend to be found on the surface of a proteinand which don’t.”

Recently, the ORNL group modeled a pro-tein complex using PROSPECT and experimen-tal data provided by Cynthia Peterson, a Univer-sity of Tennessee researcher who has identifieda number of disulfide bonds between amino ac-ids in certain parts of the protein. PROSPECT isbeing used to incorporate experimental data pro-vided by Greg Hurst and Jim Stephenson ofORNL’s Organic Mass Spectrometry Group.They are using electrospray ionization ion trapmass spectrometry and a cross-linking chemicalto determine the distances between two aminoacids—both lysines—in a protein. Their initialstudies found that the lysines linked by thischemical of a known length are 4 angstroms apart.(See article starting on p. 14.)

Because NMR data provide distances be-tween amino acids in a protein, the ORNL groupwill gladly accept partial NMR data from the pi-lot centers, which otherwise will not be used be-cause the information is insufficient to determinea whole protein structure. “This amount of datais good enough to help PROSPECT reliably pre-dict a protein structure,” Xu says. He notes thatthe level of confidence, or uncertainty, in know-ing a protein structure with complete accuracy iswithin 1 to 1.5 angstroms for X-ray crystallogra-phy, within 2 to 2.5 angstroms for NMR, andwithin 4 angstroms for computer modeling.

The ORNL group is focused not only onpredicting protein structures more accurately butalso on doing it much faster than current compu-tational techniques allow. “We now use PROS-PECT and 20 other computational tools to deter-mine protein structures in a semi-automatic

fashion,” Xu says. “Using the IBM RS/6000 SPsupercomputer at DOE’s Center for Computa-tional Sciences at ORNL, we can now thread 100or more proteins a day against 2000 possible tem-plate structures. We are seeking funding to de-velop software to build an expert system and anautomated protein structure pipeline to run onthe IBM supercomputer. The expert system willmimic the human decision-making process to au-tomate the computational tools. Our goal is topredict about 100 protein structures a day.

“If the proposal is funded, our first projectwill be to predict the structure of proteins of theProchlorococcus marinus genome, a bacteriumwith about 1600 genes. We hope to show that wecan predict these protein structures.”

In the next three years, the ORNL groupexpects to do some computer simulation that willbe of interest to the pharmaceutical industry.Their work could enable more rapid design ofdrugs that are safe and effective.

“To do this, you have to know whether aligand, which is a group of molecules typical ofa new drug, will dock with a particular protein toinhibit or stimulate its activity,” Xu says. “Wewill be doing computer modeling to determinewhether and how a ligand binds with various pro-teins to cause a healing or harmful effect.”

PROSPECT is a copyrighted computerprogram. It is being used by over 20 academicorganizations, including MIT, Columbia Univer-sity, the University of Michigan, and the Univer-sity of Texas. Millennium Pharmaceuticals is in-terested in licensing the program. ORNL’s Tech-nology Transfer and Economic Development Di-rectorate seeks to license this computer toolkitfor commercial use because its recent successessuggest it has very good prospects.


Microbe Probe: StudyingBacterial Genomes

In 1956 a new microbe was discov-ered in a can of spoiled ground beefthought to have been sterilized byradiation. The bacterium was named

Deinococcus radiodurans because it can survivedoses of radiation thousands of times higher thanwould kill most organisms, including humans.The Department of Energy is interested in study-ing D. radiodurans because of its efficiency inrepairing its own radiation damage and becauseit can reduce, or add electrons to, uranium, iron,chromium, and technetium. Thus, the bacteriummight be a strong candidate for remediatingmixed wastes—combinations of radioactive ma-terials and toxic metals—found at DOE sites. Forexample, it could be used to convert uranium instorage ponds from a soluble to an insoluble formso that it sinks into the sediments instead of re-maining in water that might flow off-site.

ORNL researchers are interested inknowing more about the bacterium’s genes thatrepair radiation-induced damage to its DNA.Knowing how these DNA repair genes worktogether in a network could help scientists betterunderstand how living organisms evolved andhow microbes that may have lived on Mars orother extraterrestrial sites tolerated extremeradiation environments. This knowledge maylead to gene therapy in which DNA repair genesare inserted into the body to prevent radiation-induced cancer or to treat the disease. DNA re-pair genes discovered in D. radiodurans couldbe transferred to other bacteria that reduce toxicmetals, to make them better able to survive ra-diation as they treat mixed wastes.

Thanks to advanced technologies such asmicroarrays and mass spectrometry and internalfunding from the Laboratory Directed Researchand Development Program, six ORNL research-ers are improving the understanding ofD. radiodurans. One of their goals is to identifywhich genes in D. radiodurans are expressed dur-ing exposure to high levels of radiation, resultingin the production of DNA repair proteins. Anothergoal is to discover the regulatory genes that con-trol other genes involved in radiation resistance.The researchers are Jizhong Zhou, Bob Burlage,and postdoctoral scientists Dorothea Thompsonand Alex Beliaev, all in the Environmental Sci-ences Division (ESD); Bob Hettich of the Chemi-

cal and Analytical Sciences Division, and RandyHobbs of the Research Reactors Division.

To identify the D. radiodurans genes in-volved in DNA repair, Zhou and his ESD col-leagues use microarrays, also known as genechips. (See p. 5 for an explanation of how genechips work.) They will place the entire comple-ment of genes from D. radiodurans on the samegene chip and then use the chip to identify dif-ferences in global gene expression betweenD. radiodurans cells exposed to high radiationand those not so exposed. (The expressed genesproduce messenger RNAs that can be detected.)Genes expressed only in the irradiated bacteriaare likely to play specific roles in DNA repair.

The ORNL researchers will then developgenetic vectors for generating deletion mutants—bacteria from which key regulatory genes in-volved in radiation resistance are removed. Theseregulatory genes code for proteins that may acti-vate other genes involved in DNA repair. If themutant bacteria die when exposed to high radia-tion doses, then the results of this experimentsuggest that the deleted genes play a key role inradiation resistance.

To understand the cellular response of theD. radiodurans bacteria to radiation, it is neces-sary to identify protein expression profiles andprotein-DNA interactions. Zhou and his col-leagues will culture the bacteria under both non-irradiated and irradiated conditions. The bacte-ria cells then will be lysed and the proteins willbe extracted, purified, and finally characterizedby Hettich, using high-resolution electrospraymass spectrometry.

“The capability of this technique for ac-curate measurement of the protein masses, as wellas identification of sequence tags (short amino-acid sections from the protein), provides unpar-alleled information for unambiguous proteinidentification,” Hettich says. Such informationwill be compared to that obtained from moreconventional two-dimensional gel electrophore-sis. The data should provide insights about howthe microbe’s proteome—its total protein pro-file—is altered in response to DNA damage, pro-viding important details about gene activity.

“Knowing when and where a gene is ex-pressed often provides a strong clue to its bio-logical function,” Zhou says. “Gene expression

patterns and protein profiles in a cell can providedetailed information about its state.” The re-searchers’ goal is to identify pathways of inter-actions between DNA and proteins that result inDNA repair of radiation damage.

Another group of researchers interested inD. radiodurans is the Computational BiologySection of ORNL’s Life Sciences Division.Using gene-recognition algorithms and othertools, Frank Larimer and his colleagues analyzecompleted DNA sequences of microbes andcompare the identity and order of these DNAbases with known sequences in databases. Theyidentify the known genes contained in thesemicrobial genome sequences and predict theamino-acid sequences, molecular sizes, andpossible functions of the proteins these genesencode. They also predict the makeup and func-tions of unknown genes and proteins.

“We have studied the completed sequenceof D. radiodurans and can recognize many of itsDNA repair genes,” Larimer says. “We inferredthe function of their proteins based on their

ORNL researchers are using gene chips, mass spectrometry, and computationalanalysis to understand what microbe genes do.

Deinococcus radiodurans is a microbe of particularinterest to DOE because of its ability to thrive inradiation levels thousands of times higher thanthose that would kill most organisms, includinghumans. It also may prove useful in bioremediationof toxic waste. Shown here is a group of fourmicrobe cells. Courtesy of Uniformed ServicesUniversity of the Health Sciences.

Number One, 2001 23

Tom

Cer

nigl

io

degree of similarity with proteins in the databaseassociated with repair genes.

“The majority view is that there is a‘eureka’ DNA repair gene to explain whyD. radiodurans is so resistant to high levels ofradiation. I take the minority view that, althoughthe bacterium’s DNA repair genes look like therepair genes in other organisms, D. radioduransdoes the repair job more efficiently. It is extremelyadept at aligning and recombining its brokenchromosome parts in the proper order.”

D. radiodurans is one of the millions ofmicrobes that have been evolving on the earth overthe past 3.8 billion years. Many of the 1% of themicrobes known to humans survive and thrive inextremes of radiation, heat, cold, pressure, salin-ity, and acidity, often where no other life formscould exist. DOE is interested in identifying andharnessing the talents of some of these microbesfor cleaning up hazardous wastes, producing en-ergy (e.g., methane), and sequestering carbon.

Since its establishment in 1994, DOE’sMicrobial Genome Program has focused on de-termining the sequences of the genomes of selectedbacteria and other microbes that do not cause dis-ease. The ORNL group provides computationaland other bioinformatic analysis of microbial se-quences obtained by DOE’s Joint Genome Insti-tute (Lawrence Berkeley, Lawrence Livermore,and Los Alamos national laboratories) at their pro-duction facility in Walnut Creek, California.Larimer and his colleagues “annotate” these mi-crobial DNA sequences and others provided byacademic groups. That is, they add to the data-

base “biological footnotes” about thegenes, the coding and noncoding re-gions of the genes, and the possiblestructure and function of the proteinsencoded by individual genes. For eachmicrobe, the researchers translateroughly 2 billion DNA bases intomeaningful information.

Using the IBM RS/6000 SPsupercomputer at DOE’s Center forComputational Sciences at ORNL, thegroup runs GrailEXP, the version ofthe ORNL-developed Gene Recogni-tion and Analysis Internet Linkwritten for parallel supercomputers,along with other gene modeling pro-grams to determine the correct DNAsequences in the microbial genes thatcode for proteins. “Ninety percent ofbacterial DNA encodes protein com-pared with less than 2% for humans,”Larimer says. “Microbes are veryefficient this way. So we try to distin-guish between what portions of thebacterial sequence are genes and whatis a statistical anomaly.”

The typical bacterium has 2000genes (each microbial gene has about900 bases). The ORNL group tries tofind gene signatures in bacteria—unique combinations of genes that in-dicate each bacterium’s identity.

“What is startling is that we canassign functions for 50% of the genesin a microbial gene, which might be1000 genes. The other half of thegenes have no known function. Half of thesegenes are unique—we’ve never seen them be-fore. There is a lot of genetic biodiversity in mi-crobes. Microbes acquire sets of genes from eachother and throw away large blocks of genes theydon’t need as they evolve.”

To date, 50 microbial genes have beencompletely sequenced. Larimer’s group has beenannotating them and working with 200 ongoingmicrobe sequencing projects. DOE’s goal is tosequence 250 microbes that could be useful forits energy and environmental missions.

In addition to D. radiodurans, some ofthe bacteria being analyzed are particularly

fascinating to Larimer. One isProchlorococcus marinus, a blue-green

marine alga that is the most abun-dant organism on earth in number

and mass. These oceanalgae, whose in-dividual cells areeach less than1 micron in di-ameter, togetherfix more carbondioxide than allof the terrestrialb i o s p h e r e .Prochlorococcus


False-color image (opposite page) of a microarrayused to analyze messenger RNA levels inShewanella oneidensis MR-1 cells under differentgrowth conditions. Two-color fluorescence de-tection allows two different biological samples to beanalyzed simultaneously on a single array. Geneshighly expressed under aerobic growth conditionsare shown in green (Cy3), whereas genes expressedspecifically under anaerobic conditions in thepresence of iron [Fe(III)] appear as red (Cy5) spots.

Julia Stair watches the Cartesian Technologies gene-chip makergo through a test run. It is used for studies of gene expression inbacteria of environmental significance. Stair, a post-mastersresearcher with Oak Ridge Associated Universities, preparesmicroarrays for studies of D. radiodurans bacteria.

marinus is found 100 meters deep in the oceanyet it absorbs sufficient blue-green photons pen-etrating to that depth to get enough energy tofix carbon.

Another interesting microbe that fixes car-bon dioxide is Nitrosomonas europa, which getsits energy not from sunlight but from oxidizingammonia. Using ammonia fertilizer, it is respon-sible for putting nitrogen into the soil, making itfertile for plants.

Zhou is fascinated by gene-chip and massspectrometry studies of the metal-reducingbacterium Shewanella oneidensis. “We comparedgene expression patterns in Escherichia colibacteria and a mutant of Shewanella oneidensisthat we generated,” Zhou says. “We found thatthe functions of about 75% of their genes aresimilar. But there are differences. In E. coli onegene activates nitrate reduction, but the counter-part gene in mutant Shewanella oneidensi re-presses nitrate reduction.

“Using bioinformatics, scientists oftenassign gene functions based on the similarity ofsequences of two different bacteria. This ap-proach may not correctly define all gene func-tions. To get the complete picture, bioinformaticsmust be supplemented by experimental ap-proaches that analyze expression in microbes,using microarrays and mass spectrometry.”

Bob Hettich shows the new high-performanceFourier transform ion cyclotron resonance massspectrometerthat he usesfor biologicalresearch.

Cur

tis B

oles

Enhanced by Gail Sweeden


SNSand BiologicalResearch

SNSand BiologicalResearch

Some drugs that combat AIDS—the deadly disease caused by thehuman immunodeficiency virus(HIV) that kills or damages the

body’s immune system cells so the body can’tfight infections—have been discovered by acci-dent. For example, the protease inhibitors usedto treat AIDS were originally tested as medica-tions for reducing blood pressure. These drugsresemble pieces of the protein chain that the HIVenzyme, called protease, normally cuts. By gum-ming up the protease scissors, HIV protease in-hibitors prevent the protease from cutting longpolyprotein chains into the shorter structural pro-teins HIV needs to assemble a cocoon of protec-tion around its RNA genome. Although the viruscan still invade other cells, without the protec-tive cocoon, the virus genome is exposed to hostenzymes, which easily destroy the invadingviral RNA and prevent further replication.

Finding the right chemical structures tohalt the replication of HIV and then bringing thedrugs to market can take years and cost millionsof dollars. As a result, millions of people whoneed these medications to survive cannot affordthem. However, there is cause for hope. Recentwork in biological neutron crystallography andnew biological instruments planned for the Spal-lation Neutron Source (SNS) at ORNL couldprovide rapid insights into macromolecular struc-tures that could lead to safer, more effective, andless expensive drugs.

Determination of protein structures byneutron diffraction will be useful for drug de-sign. HIV protease, a target of AIDS drug de-signers, belongs to the general class of protein-digesting enzymes called aspartic proteases.These enzymes (including the stomach’s pepsin)are named for aspartic acid, which is present attheir active sites. A solvent molecule boundtightly to aspartate carboxyl groups is presumedto take part in the catalytic mechanism that en-ables the enzyme to break a protein chain. Thebest currently accepted mechanisms are largelybased on X-ray images of inhibitor structures,but the active-site hydrogen atoms cannot be de-finitively located by current X-ray analyses, leav-ing an incomplete picture of enzyme catalysis.

Jonathan Cooper of the University ofSouthampton in England and Dean Myles of theEuropean Molecular Biology Laboratory Outsta-tion in Grenoble, France, published a report in2000 on the neutron diffraction structure of thefungal aspartic protease endothiapepsin. This workrepresents the largest protein solved by neutrondiffraction methods to date. The success of neu-tron diffraction in determining the positions ofcatalytic hydrogens reveals a route to the devel-opment of more effective inhibitors to asparticproteases. It also suggests that this method couldplay a significant role in rational drug design.

Similar and better atomic-resolutionimages of biological substances are anticipatedfrom the SNS after its completion in 2006 whenit is operating at 2 megawatts. That’s because theSNS will have a much higher neutron flux; it willoffer 10 times the number of neutrons now avail-able at any existing neutron research facility. Theintrinsic time structure of the pulsed neutronsource makes SNS ideal for atomic-resolutionprotein crystallography studies.

“If the SNS has an instrument that can dohigh-resolution crystallography on protein crys-tals with 100-angstrom repeats,” says GerardBunick of ORNL’s Life Sciences Division (LSD),“we could contribute crucial molecular informa-tion that couldlead to more ef-fective drugsagainst HIV, forexample.

“X raysgive excellenthigh-resolutionimages of heavyatoms in pro-tein crystals, butneutrons also seelighter atomssuch as hydrogen,which makes uphalf of all the at-oms in proteins.Only one in 100proteins crystal-

lizes well enough to get the resolution needed tosee hydrogen atoms using the bright X rays fromsynchrotrons, according to a survey of structuresin the Protein Data Bank.”

Only about a dozen protein structures havebeen solved using neutron diffraction. But thatcould change with the opening in 2001 of the newProtein Crystal Station at the Los Alamos Neu-tron Scattering Center and the later operation ofthe SNS. Bunick’s colleague Leif Hanson fromthe University of Tennessee says that within 10years, 50 to 100 protein structures will be deter-mined annually using neutron diffraction. Neutrondiffraction studies of protein structures are alsobeing facilitated by the availability of improvedneutron detectors that speed up data collection andthe ability to grow larger crystals of proteins.

“It is now easier to grow larger protein crys-tals in space and on earth,” says Bunick, who set upcrystal-growing experiments that were run on theU.S. Space Shuttle Columbia in 1995 and on theRussian Mir space station in the late 1990s. Nearlyperfect crystals of nucleosomes were grown in themicrogravity environment of these vehicles orbit-ing around the earth. Nucleosomes are the buildingblocks of chromosomes; they each consist of a coreof histone proteins around which approximately twoturns of double-stranded DNA are wrapped.

Three world-class biological instruments are being designed for the Spallation NeutronSource. They will help biologists determine the atomic-level structure of proteins andother signaling compounds that allow cells to communicate and coordinate activitiesacross an organism. The research could lead to safer, more effective drugs.

ORNL’s Gerry Bunick and his colleagues plan to use the SNS to better understandinteractions between DNA and proteins and between proteins. Below are colorrenderings of a nucleosome core particle (DNA in red, proteins in green) and detailfrom an exposed surface of the protein core. The ribbon (right) depicts the super-position of symmetry-related proteins in the structure. The transparent surfaceshows basic and acidic domains in blue and red respectively. The image shows thatthe acidic domains are where the symmetry-related proteins are not identical.

Three world-class biological instruments are being designed for the Spallation NeutronSource. They will help biologists determine the atomic-level structure of proteins andother signaling compounds that allow cells to communicate and coordinate activitiesacross an organism. The research could lead to safer, more effective drugs.

Number One, 2001 25


“We have proposed to NASA to grow largeprotein crystals for neutron studies in themicrogravity environment of the InternationalSpace Station,” Bunick says. “And better waysto grow large protein crystals on earth have beendeveloped as a result of microgravity crystalgrowth research sponsored by NASA.”

Bunick, Hanson, and Joel Harp, all ofLSD; Chris Dealwis of the University of Ten-nessee; and Jinkui Zhao of the SNS organizedand co-hosted an SNS workshop December 18,2000, in Knoxville. They were pleased that theworkshop participants recommended that twoinstruments be designed to do high-resolutionprotein crystallographystudies using SNS neu-trons. The first instrumentwould be used for high-resolution protein crystal-lography on crystals withup to 100-angstrom (Å)repeating motifs. The sec-ond instrument, if funded,would be located in thelong-wavelength target sta-tion. It would be used tostudy protein complexeswith 200–250 Å repeatingmotifs in the crystals, otherlarge macromolecularcomplexes at lower resolu-tion, and biological mem-brane systems.

According to ThomMason, ORNL’s associatelaboratory director for theSNS, biologists are inter-acting with engineers todesign 3 of the 12 world-class scientific instruments planned for the SNS.These instruments will complement the capabili-ties of DOE’s Center for Structural and Molecu-lar Biology at ORNL. They will help biologistsdetermine the atomic-level structure of proteins,

amino acids, hormones, peptides, and other sig-naling compounds that allow cells to communi-cate with each other and coordinate their activi-ties across the organism.

Besides the protein crystallography instru-ment, biologists plan to use a liquids reflectome-ter to study changes in surfaces, interfaces, andlayered structures in biological materials. “It couldhelp scientists study how proteins affect the struc-ture of membranes in cells,” Mason says.

A small-angle neutron scattering (SANS)instrument is also planned. It will “see” a targetwith a length scale ranging from thousands of ang-stroms to less than an angstrom. “It’s a little likehaving a camera with a zoom lens,” Mason says.“You zoom in on the subject to see the small de-tails. Then you pull the zoom lens back to get the

big picture, but you can’t see the smallest details.With the SANS at the SNS, we will be able to zoomin on the protein’s details at the interatomic level,what we call large Q. Then we can pull back andsee the whole protein, what we call small Q. We

can span this widerange in a single ex-periment, some-thing that can’t bedone easily with X-ray crystallographyor existing neutroninstruments.”

Zhao, aneutron scientistwho does biologi-cal studies at theHigh Flux IsotopeReactor (HFIR),says that the bio-logical SANS in-struments that willbe installed at bothHFIR and the SNS

will be used to study proteins and other biologicalmolecules in their natural solution. Both instru-ments will be superbly suited to study the shape,conformational changes, and dynamics of proteins.“The difference will be that the HFIR SANS willlook at large length scales, whereas the SNS SANSwill look at a range of length scales simultaneously,such as protein-membrane interactions, with pro-teins on the large-length-scale side and membraneson the small-scale side.”

“To study the interactions of proteins andother biological molecules in solution, we mustuse contrast matching,” Zhao says. In this tech-nique, some hydrogen atoms in the sample arereplaced with heavier hydrogen, or deuterium,atoms. Deuterium and hydrogen atoms scatterneutrons very differently. Changing the ratios of

hydrogen and deuterium inthe solvent water mixturechanges the visibility of pro-teins to neutrons. It’s likeadding red dye to a glass ofwater containing red andyellow balls so that all yousee are the yellow balls.”

“Scientists mask outthe portion of the proteinthey are not interested in andmake highly visible the ac-tive part of the protein thatdoes interest them,” saysMason. “The ratios of hydro-gen and deuterium in watermixtures can be changed invarious samples to meet re-searcher needs.”

The SNS can also beused for inelastic neutron-scattering experiments forthe study of protein dynam-ics. Inelastic scattering ofneutrons results from an “in-

elastic collision” in which the total kinetic en-ergy of neutrons colliding with target atoms isnot the same after the collision as before.

“Measurements of neutron-scatteringenergies will provide information on the collec-tive motion of proteins, which can be correlatedwith protein function,” Zhao says. “For example,when a catalytic protein is active, its motion israndom. But when its motion becomes harmonicat low temperature, its catalytic function stops.”

Bunick and his colleagues could use anSNS protein crystallography instrument to helpthem better understand the origin of Rett Syn-drome (RS), a debilitating neurodevelopmentaldisorder that causes loss of speech and profoundmental retardation in young girls. Studies at theSNS might improve understanding of the struc-ture and interactions of native and mutant DNA-binding proteins, enabling the production ofdrugs to reverse the deleterious effects of thedefective protein.

When research comes to life at the SNS, wecan expect some early discoveries to jump-start thedesign of drugs to improve human life.

Artist’s conception (opposite page) of the SpallationNeutron Source. Rendering by John Jordan.

The two turns of DNA from the nucleosome core particle, separated at the dyadaxis for clarity, rendered as a CPK model. The colors depict relative mobility of theatoms, with blue and red being the least and most mobile respectively. Neutrondiffraction data from the SNS will help reveal how water molecules affect thestructure of DNA and its interactions with proteins.

Schematic of the protein crystallographyinstrument planned for the SNS.


Accessing Informationon the HumanGenome Project

DOE’s Human GenomeManagement InformationSystem at ORNLcommunicates HumanGenome Projectinformation to the public.


I f you want news on the new biol-ogy, whom should you call? Try thededicated ORNL group that main-tains the Human Genome Manage-

ment Information System (HGMIS) for theDepartment of Energy. This group issues a num-ber of publications (including Human GenomeNews, published about twice a year) and maintainsa popular Web site (www.ornl.gov/hgmis/). TheHGMIS folks have become an important source ofinformation on the human, animal, plant, and mi-crobial genomes for the news media, the trade press,biology researchers, and teachers and students.

The HGMIS is DOE’s educational out-reach arm of the Human Genome Project, whichis supported by both DOE and the National In-stitutes of Health. The HGMIS Group, led byBetty Mansfield of ORNL’s Life Sciences Divi-sion since 1989, is responsible for providing in-formation about the project, its progress, its ap-plications, and its impacts on society. In 1997,on behalf of the group, Mansfield received anExceptional Service Award for Exploring Ge-nomes, presented at the 50th Anniversary Sym-posium of DOE’s Biological and EnvironmentalResearch (BER) Program; she was recognized“as founding and managing editor of Human Ge-nome News and for outstanding success in com-municating scientific information to the U.S. andinternational communities about theDepartment’s BER Program.”

In 2000 the HGMIS Group received in-quiries from ABC’s Who Wants to be a Million-aire?” and NBC’s Jeopardy concerning the ac-curacy of answers to questions on human genet-ics and the human genome. The ORNL grouphas provided information and graphics to a num-ber of U.S. and foreign news outlets, includingCNN, NBC, ABC’s Good Morning America, andABC London.

The group also has provided answers toquestions from reporters from the print media,including the Boston Globe, Chicago Tribune, LosAngeles Times, Wall Street Journal (interactiveonline edition), Science Magazine, and WiredMagazine. Marissa Mills, a trained journalist,marketing expert, and Web content developer inthe HGMIS Group, has developed press kits andanswered numerous electronic mail questions fromthe media related to human genome news events.

Mills provided assistance to two represen-tatives of the Native American community seek-

ing information about tracing Native Americanancestry through DNA. She also has created andpresented programs on careers in genetics forwomen, African Americans, and residents ofrural Appalachian communities.

Denise Casey, a trained biologist and ver-satile writer in the HGMIS Group, was guesteditor of and wrote an article for the special genesand justice issue of Judicature, a magazine forjudges, published in November-December 1999.Recently, she and group member Judy Wyrickcreated a human chromosome landmarks poster,jointly sponsored by DOE and Qiagen, a com-pany supplying reagents for genome research.

Other members of the HGMIS Group areAnne Adamson, Laura Yust, and Web architectSheryl Martin. A creative lot, they provide valu-able editing, publication layout, and Web designexpertise for the group. In addition to being thegroup’s primary editor, text manager, and facili-tator, Adamson recently co-authored texts on theDOE Ethical, Legal, and Social Issues programand contributed to an invited article with programmanager Dan Drell on how genomics may affectthe outcome of medicine 20 years from now. Yust,a graduate student, does Web work, coordinatesmailing and outreach, and maintains the ex-t e n s i v edatabaseof 15,000newslet-ter sub-scribers.

Thegroup isp a r t i c -u l a r l yproud oftheir Website. TheH G M I SWeb sitestatisticsshow anav e rageof 2.6 mil-lion usersessionsper year;the aver-age length

of a user session is 12.5 minutes. The site has9 million text file hits per year, and many morehits when graphic files are counted. The site hasspecialized pages, including pages on the ben-efits of genome research; explorations into theethical, legal, and social issues surrounding theavailability of personal genetic data; and prim-ers that are widely used by researchers from otherdisciplines who wish to contribute to genomicsresearch, teachers and students, genetic counse-lors, and biotechnology company personnel.

The Web site also brings in 2100 ques-tions a year by electronic mail, including lettersfrom parents of children with genetic diseases.

The work has its surprises. “When I gavea presentation on the Human Genome Project toa group of doctors,” Mansfield says, “I was firstasked not about the medical implications butrather for advice on investing in biotech stocks.”

A biologist by training, Mansfield’s priorresearch included studies on changes in the pro-teins produced by abnormal red blood cells inleukemic mice after chemicals were added tomake these cells normal. “I was doing proteomicsresearch back when it wasn’t cool,” Mansfieldsays. Now, she heads a leading source of infor-mation on a field that is really hot.

Judy Wyrick (right) and Sheryl Martin examine the on-screen image of the cover of adocument entitled Genomes to Life, which the HGMIS group prepared for DOE. Thedraft document, available on the HGMIS Web site, describes the goals of the proposedprogram, including identifying multiprotein complexes that carry out the functions ofliving systems. See www.DOEGenomesToLife.org.

Cur

tis B

oles

; enh

ance

d by

Gai

l Sw

eede

n

Number One, 2001 27

A Model Fish forPollutant Studies

M ark Greeley admits the one-inch-long, brownish-grayzebrafish is a “rather drabaquarium fish.” But he says

the small striped fish “can be considered themouse of the fish world, because it is arguablythe best vertebrate model for development.”Greeley then explains how to make a zebrafishlook interesting: Turn it green.

Greeley, director of ORNL’s Aquatic Toxi-cology Laboratory in the Environmental SciencesDivision, is an ecotoxicologist who studies theeffects of contamination on fish and relates theeffects back to humans and to other wildlife, aswell. He is particularly interested in endocrine-disrupting chemicals, artificial and natural sub-stances that mimic or block the normal actionsof hormones such as estrogen and testosterone.Endocrine-disrupting chemicals, including suchcommon or well-known substances as orga-nochlorine pesticides, plasticizers, dioxins, andPCBs, can cause significant health problems inwildlife and humans.

The risks of endocrine-disrupting chemi-cals are most apparent in fish and wildlife.“Downstream of sewage treatment plants, fishoften have combined male and female sex organs

from exposure to estrogenic compounds in phar-maceutical waste and other endocrine-disruptingchemicals,” Greeley says. “Downstream of a pa-per and pulp mill, we have studied a fish popula-tion that consists entirely of males because ofexposure to natural plant sterols and other hor-mone mimics in the wood-processing by-products.Endocrine-disrupting chemicals have been im-plicated in skewed sex ratios in birds, develop-mental abnormalities in frogs and other amphib-ians, deformed sex organs and other reproduc-tive problems in alligators, and decreased fertil-ity in a variety of wildlife species.”

Endocrine-disrupting chemicals may alsohave subtle but no less significant health effectson humans. Although the evidence is less con-clusive, these chemicals have been implicated inthe development of human breast and testicularcancers, early onset of puberty in both boys andgirls, lowered sperm counts, and decreased fer-tility in the developed world.

Greeley is using the zebrafish as a modelto study the molecular toxicology of endocrine-disrupting chemicals. “The ultimate goal of thisresearch,” he says, “is to understand how thesetoxicants cause their adverse effects by discover-ing which genes are turned on and off and which

proteins areproduced oraltered as a re-sult of expo-sure.”

To ex-amine the be-havior of spe-cific genes asa result ofexposure toendoc r i ne -d i s r u p t i n gc h e m i c a l s ,Greeley andhis associ-ates—SuzanneGarnmeister,Kitty Mc-Cracken, andZamin Yang—use a jelly-fish gene thatcodes for agreen fluores-

cent protein (GFP). The bioreporter GFP gene isfirst genetically fused to a hormone-responsivegene, such as the vitellogenin gene, which regu-lates the formation of yolk proteins in zebrafishand other non-mammalian vertebrate organisms.The resulting gene construct is then microinjectedinto fertilized single-cell eggs and incorporatedinto the zebrafish genome. Whenever the nativehormone-responsive gene is expressed in thesetransgenic fish in response to natural hormonesor from exposure to endocrine-disrupting chemi-cals, the bioreporter gene construct also turns onand glows with the typical intense blue-greenfluorescent signature of the GFP gene. Becauseof the remarkable clarity of the zebrafish embryo,the researchers can determine exactly when andwhere the specific gene of interest is activated orturned off in response to toxicant exposure.

The simultaneous expression of poten-tially thousands of zebrafish genes during expo-sure to endocrine-disrupting chemicals is alsobeing examined using a zebrafish DNA “tox-chip” microarray currently being developed incollaboration with Mitch Doktycz and Peter Hoytof the Life Sciences Division (see the article ontheir work on p. 7). “Results with an early ver-sion of this chip,” Greeley says, “clearly demon-strate the enormous potential of this tool in toxi-cological research.”

Genes alone can’t tell the whole story.Proteins produced as a result of gene expressionactually mediate the toxic response.

“In collaboration with Brian Bradley of theUniversity of Maryland–Baltimore County, wehave demonstrated significant changes in proteinexpression in developing zebrafish following ex-posure to estrogen and estrogen-mimicking chemi-cals,” Greeley says. “Proteins normally producedduring early development were absent, but atypi-cal proteins were produced. We expect thezebrafish to be an excellent model for studyingthe functional relationship between gene and pro-tein expression in response to toxicant exposure.

“Because humans have many of the samegenes, what we learn about gene and protein ex-pression in zebrafish exposed to endocrine-disrupting chemicals and other toxicants willhelp us better understand what these chemicalscan do to humans.”

Zebrafish are becoming more fascinatingnow that they are giving scientists the green light tolearn more about how toxins affect our genes.

The zebrafish is a model organism for studying the effects ofendocrine-disrupting chemicals on gene and protein expression.

Suzanne Garnmeister examines a zebrafish embryo in a fluorescence microscope whileMark Greeley looks at screen images of zebrafish embryos. Embryos were exposed toestrogen after being microinjected with a gene construct combining the vitellogeningene with a bioreporter gene for the jellyfish green fluorescent protein.

A Model Fish forPollutant Studies

Cur

tis B

oles

; enh

ance

d by

Gai

l Sw

eede

n


Controlling Carbon inHybrid Poplar Trees


Using carbon dioxide from theatmosphere, as well as sun-light and water, hybrid poplartrees grow fast and tall, up to

12 feet per year. They also harbor a considerableamount of carbon in their stems, branches, leaves,and roots. Plant geneticists would like to designhybrid poplar trees that maximize the amount ofcarbon they store in their cell walls. These treescould then be used to more effectively sequester

carbon dioxide, a greenhouse gas, through in-creased carbon storage in their roots and, afterthe roots decay, in soil. Alternatively, when har-vested and digested microbially, these “designer”trees could offer an increased yield of commod-ity chemicals (e.g., polylactic acid, furfural, andacetic acid) and ethanol fuel.

In trees, carbon is “allocated” betweenaboveground stems, branches, and leaves andbelowground roots. It is “partitioned,” or divided,

among three types of plant cell-wallcomponents—cellulose, hemicellu-lose, and lignin. A plant could be de-signed to have an unusually high cel-lulose content above ground, if in-creased ethanol production is desired.In addition, if carbon sequestration isthe goal, its roots could be designedto have unusually high lignin content,which is resistant to degradation bymicrobes, increasing the residencetime of carbon in the soil.

“In five years, we hope to de-termine which genes control carbonallocation and partitioning in hybridpoplar trees,” says Gerald Tuskan, aplant geneticist in ORNL’s Environ-mental Sciences Division (ESD).“Our research indicates that carbonallocation is controlled by a smallnumber of regulatory genes, that sepa-rate genes controlling cell-wall chem-istry operate independently aboveground and below ground, and thatgenes controlling carbon allocationaffect carbon partitioning.”

Tuskan is working on a three-year project to enhance bioenergy con-version and carbon sequestration inwoody plants with his ESD colleaguesStan Wullschleger, Tim Tschaplinski,and Lee Gunter; Brian Davison of theChemical Technology Division; andseveral researchers from DOE’s Na-tional Renewable Energy Laboratory.The team is studying wood tissuesamples from some 300 hybrid pop-lars grown in Washington that are theprogeny of trees from Minnesota andOregon parents.

Tuskan and his colleagues aremapping the hybrid poplar genome byfinding genetic “markers” unique totrees that have a desirable trait, such

as higher-than-normal cellulose content aboveground. A marker is a known DNA sequence as-sociated with a particular gene or trait; in thisstudy, it consists of two unique, non-repeatingDNA sequences flanking simple sequence re-peats, such as GAGAGAGAGA. Some 150 mark-ers have been found so far; the project’s goal is400 markers.

“Each hybrid poplar tree has a unique ge-netic fingerprint,” Tuskan says. “We look for anassociation between markers unique to each treeand variations in the allocation and partitioningof carbon content. Once we find the marker thatcontrols the trait we are interested in, such as highlignin content in the roots, then we will try tolocate the genes responsible. Such genes couldbe used to design tree root systems that are highin lignin content.”

Tuskan is also interested in finding thegenes that control the size and thickness of a tree’scell walls, the substructure of wood that deter-mines its usefulness and commercial value. “It’sbecause of differences in cell sizes and wall thick-nesses that oak floors are stronger than pinefloors, maple furniture is more attractive thanaspen furniture, and white oak rather than redoak is used to make barrels to store wine,” hesays. “Cell dimensions also determine whether atree’s wood is suitable for combustion or pro-duction of paper or ethanol.”

Use of a light microscope or scanning elec-tron microscope to determine wood cell dimen-sions in samples from various trees is expensiveand time consuming. So, Tuskan sought helpfrom Mike Paulus of ORNL’s Instrumentationand Controls Division. Paulus is a co-developerof the high-resolution, X-ray-computed tomog-raphy system called a MicroCAT scanner. Al-though used mostly to image internal defects insmall animals, the MicroCAT scanner also of-fers a faster, better, and cheaper way to measurethe lengths and diameters of cell walls in wood.(See article on p. 11.)

“With the MicroCAT, we can get cell mea-surements from an intact block of wood, whereasfor microscope studies, we have to slice woodinto very small pieces,” Tuskan says. “With thelight microscope, we were getting 100-micronresolution, but with the modified MicroCAT, weget 10-micron resolution and may be able to getdown to a resolution of one to two microns. TheMicroCAT is a great tool for rapidly screeningfor wood-cell dimensions in the context of a largegenetic mapping study.”

ORNL scientistsare helping tosearch for genesthat could allow thecreation of trees thatstore more carbon andoffer higher-value products.

To produce a hybrid poplar tree, flowers from the female,Populus trichocharpa, are isolated and inoculated with pollenfrom the male, Populus deltoides. Hybrid progeny grow fromthe resulting seeds. The flowers and pollen from the hybridtrees can be crossed to produce descendants of the“grandparent” trees. The genomes of the grandparent treesand their progeny can be compared to a deck of cards that isshuffled and reshuffled for each tree. In the grandparents, themale has all black cards and the female has all red cards. Inthe next generation, the cards are half red and half black. Inthe third generation, the percentages of black and red cardsvary greatly for each tree, producing genetic diversity thatallows the linkage of genes to highly desirable traits.

29Number One, 2001


Disease Detectives

CANTILEVER DEVICES

Take a silicon chip as small as a grain of rice and carve outbarely visible diving-board–like projections at one edge.Coat these cantilevers with gold. Attach thiolated single-stranded DNA to the gold-coated cantilevers. Allow

single-stranded DNA of different sequences to come in contact with theDNA attached to the cantilevers. If a DNA sequence complements theDNA on a cantilever, they will bind together, or hybridize, to form double-stranded DNA. Thomas Thundat and Karolyn Hansen, both of ORNL’sLife Sciences Division, use this recipe to distinguish between numbers ofbase pairs in DNA sequences on cantilevers.

“We immobilized single strands of DNA containing 20 bases on aseries of cantilevers,” Thundat says. “The DNA bases on these cantileverspaired with 20, 15, 10, and 9 bases of single-stranded DNA introduced tothe cantilevers. We found that the cantilevers all bend because of the changesin surface tension as a result of DNA hybridization.” The more bases acantilever holds, the more it bends, changing the angle of deflection oflaser light bounced off the cantilever, as recorded in a detector.

Thundat believes that this technology could be used for DNA sequenc-ing and that the approach would cost less and take less time than conventionaltechniques because it would avoid the step of adding fluorescent dyes to labelthe DNA bases. This technology has been licensed to Graviton, Inc.

Thundat believes that cantilevers can be used to detect defective genesthat cause breast cancer, colorectal cancer, and cystic fibrosis. These mu-tant genes have one incorrect DNA base. ORNL experiments have shownthat a DNA sequence in a liquid sample will hybridize with a complemen-tary DNA sequence bound to a cantilever, even if the sample sequence hasone wrong base, or a mismatch.

“We found that a mismatch causes the cantilever to bend up insteadof down,” Thundat says. “This change in bending direction could be usedto detect defective genes that cause disease.”

ORNL researchers are developing two types of miniaturizeddevices for diagnosing diseases. These devices are based oncantilevers and biochips.

Thomas Thundat examines a cantilever beam (see drawing above) onwhich an antibody for prostate-specific antigen is attached. By bendingwhen PSA binds to the probe antibody molecule, this device detects earlysigns of prostate cancer in serum samples with unusually high sensitivity.

BIOCHIP DEVICES

By detecting a mutant breast cancer gene, a doctor can pre-dict that a patient will get breast cancer. By detecting acertain protein, a doctor can determine that the patienthas adult-onset diabetes. Someday, physicians will be able

to rapidly analyze both genes and proteins from a single drop of a patient’sblood, using a palm-size device. At least that’s the goal of Tuan Vo-Dinhand his co-workers David Stokes, Minoo Askari, and Guy Griffin, all ofthe Life Sciences Division, and Alan Wintenberg of ORNL’s Instrumenta-tion and Controls Division.

“We can do genomics and proteomics on a single platform using ourmultifunctional biochip,” says Vo-Dinh. “The biochip is being designed toprocess up to 100 samples in 30 minutes.” The multifunctional biochip is anadvanced version of the group’s DNA biochip, which contains only DNAprobes. This technology has been licensed to HealthSpex, Inc., in Oak Ridge.

To sample a patient’s blood for a DNA sequence that is a red flag fora genetic disease, the multifunctional biochip has a complementary DNAsequence to which this mutant sequence will bind. To sample for a specificdisease-related protein, the biochip has a probe (e.g., an antibody or proteinreceptor) that will bind with this particular protein.

ORNL experiments have shown that the biochip can detect the tuber-culosis bacterium, the HIV gene, a cancer suppressor gene, the anthrax bac-terium used in biological warfare, and Escherichia coli found in contami-nated food. Thus, the biochip could be used for medical diagnosis, defense,and food safety applications.

Small though they are, both the cantilever device and biochip couldmake a big contribution to health care.

The cantilever technology could also be used to detect prostate can-cer. ORNL researchers have immobilized on a cantilever the antibody forprostate-specific antigen (PSA), the chemical signal for the disease. AnORNL collaboration with the University of California’s Professor ArunMajumdar has shown that the cantilever bends when its antibody matchesPSA in serum samples supplied by Majumdar. The sensitivity of this tech-nology is 10 times higher than that of conventional techniques.

To make a multifunctional biochip work in the doctor’s office, a patient’s bloodsample would be processed to separate DNA sequences and proteins, which arethen labeled with a fluorescent dye. The chip would then be exposed to theseblood fragments. Disease-related DNA and proteins will bind to the chip’scomplementary DNA and antibody probes. A miniaturized laser diodeilluminates the array of sites, causing fluorescence at the sites of hybridization.The signals are collected and sent to a microprocessor chip, which stores theidentity of the probes at each location. By matching signal locations to theprobes known to be there, the microprocessor detects disease-related genes andproteins in the patient’s blood and relays the diagnosis to the physician.

Curtis Boles

PRSRT STDU.S. Postage

PAIDPermit No. 3

Oak Ridge, TN

OAK RIDGE NATIONAL LABORATORY REVIEWP.O. Box 2008, Oak Ridge, Tennessee 37831-6266

The experimentally determined crystal structure of the nucleosome core particle is shownabove in a color rendering. The protein is represented by ribbon models of the secondary-structure elements. The DNA is represented as a wire frame model, indicating base pairingbetween complementary strands. The image recently graced the cover of ActaCrystallographica Section D, Biological Crystallography (Volume 56, Part 12, December2000). The work is a result of research by the Life Sciences Division’s Gerry Bunick, UT’sJoel Harp and Bryant Hanson, and David Timm of the Indiana University School of Medi-cine. The group hopes to use the Spallation Neutron Source to better understand interac-tions between DNA and proteins and between proteins. (See article starting on p. 24.)

The three images at the right are predicted protein structures. (See article starting on p. 20.)

vol. 34, no. 1, 2001 - ornl review v34n1 2001.pdf · curtis boles 19 curtis boles 27. number one,...

Documents